diff --git a/florence-2-l_vis1024-lang2048_dior-v2_b2x16-100e-slurm-zero2/trainer_state.json b/florence-2-l_vis1024-lang2048_dior-v2_b2x16-100e-slurm-zero2/trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..237c351c08ba6e79ef520b402b608a4a4a173bc1
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dior-v2_b2x16-100e-slurm-zero2/trainer_state.json
@@ -0,0 +1,256942 @@
+{
+ "best_metric": null,
+ "best_model_checkpoint": null,
+ "epoch": 100.0,
+ "eval_steps": 200.0,
+ "global_step": 36700,
+ "is_hyper_param_search": false,
+ "is_local_process_zero": true,
+ "is_world_process_zero": true,
+ "log_history": [
+ {
+ "epoch": 0.0027247956403269754,
+ "grad_norm": 32.00284194946289,
+ "learning_rate": 1.8165304268846506e-08,
+ "loss": 6.748,
+ "step": 1
+ },
+ {
+ "epoch": 0.005449591280653951,
+ "grad_norm": 36.0566520690918,
+ "learning_rate": 3.633060853769301e-08,
+ "loss": 6.5977,
+ "step": 2
+ },
+ {
+ "epoch": 0.008174386920980926,
+ "grad_norm": 52.746524810791016,
+ "learning_rate": 5.449591280653951e-08,
+ "loss": 6.6777,
+ "step": 3
+ },
+ {
+ "epoch": 0.010899182561307902,
+ "grad_norm": 42.455482482910156,
+ "learning_rate": 7.266121707538602e-08,
+ "loss": 6.6211,
+ "step": 4
+ },
+ {
+ "epoch": 0.013623978201634877,
+ "grad_norm": 35.57960891723633,
+ "learning_rate": 9.082652134423253e-08,
+ "loss": 6.6641,
+ "step": 5
+ },
+ {
+ "epoch": 0.01634877384196185,
+ "grad_norm": 34.983863830566406,
+ "learning_rate": 1.0899182561307902e-07,
+ "loss": 6.7207,
+ "step": 6
+ },
+ {
+ "epoch": 0.01907356948228883,
+ "grad_norm": 36.89677429199219,
+ "learning_rate": 1.2715712988192554e-07,
+ "loss": 6.6953,
+ "step": 7
+ },
+ {
+ "epoch": 0.021798365122615803,
+ "grad_norm": 38.256492614746094,
+ "learning_rate": 1.4532243415077205e-07,
+ "loss": 6.6895,
+ "step": 8
+ },
+ {
+ "epoch": 0.02452316076294278,
+ "grad_norm": 31.679651260375977,
+ "learning_rate": 1.6348773841961852e-07,
+ "loss": 6.6367,
+ "step": 9
+ },
+ {
+ "epoch": 0.027247956403269755,
+ "grad_norm": 47.846778869628906,
+ "learning_rate": 1.8165304268846506e-07,
+ "loss": 6.5566,
+ "step": 10
+ },
+ {
+ "epoch": 0.02997275204359673,
+ "grad_norm": 34.29279327392578,
+ "learning_rate": 1.9981834695731156e-07,
+ "loss": 6.6719,
+ "step": 11
+ },
+ {
+ "epoch": 0.0326975476839237,
+ "grad_norm": 33.140960693359375,
+ "learning_rate": 2.1798365122615804e-07,
+ "loss": 6.9473,
+ "step": 12
+ },
+ {
+ "epoch": 0.035422343324250684,
+ "grad_norm": 38.61561965942383,
+ "learning_rate": 2.3614895549500455e-07,
+ "loss": 6.7852,
+ "step": 13
+ },
+ {
+ "epoch": 0.03814713896457766,
+ "grad_norm": 31.86699104309082,
+ "learning_rate": 2.543142597638511e-07,
+ "loss": 6.7344,
+ "step": 14
+ },
+ {
+ "epoch": 0.04087193460490463,
+ "grad_norm": 37.8862190246582,
+ "learning_rate": 2.7247956403269756e-07,
+ "loss": 6.6465,
+ "step": 15
+ },
+ {
+ "epoch": 0.043596730245231606,
+ "grad_norm": 31.138391494750977,
+ "learning_rate": 2.906448683015441e-07,
+ "loss": 6.582,
+ "step": 16
+ },
+ {
+ "epoch": 0.04632152588555858,
+ "grad_norm": 29.93649673461914,
+ "learning_rate": 3.0881017257039057e-07,
+ "loss": 6.5879,
+ "step": 17
+ },
+ {
+ "epoch": 0.04904632152588556,
+ "grad_norm": 35.62410354614258,
+ "learning_rate": 3.2697547683923705e-07,
+ "loss": 6.6543,
+ "step": 18
+ },
+ {
+ "epoch": 0.051771117166212535,
+ "grad_norm": 36.28412628173828,
+ "learning_rate": 3.4514078110808363e-07,
+ "loss": 6.5527,
+ "step": 19
+ },
+ {
+ "epoch": 0.05449591280653951,
+ "grad_norm": 40.86235046386719,
+ "learning_rate": 3.633060853769301e-07,
+ "loss": 6.8809,
+ "step": 20
+ },
+ {
+ "epoch": 0.05722070844686648,
+ "grad_norm": 35.33604049682617,
+ "learning_rate": 3.814713896457766e-07,
+ "loss": 6.9238,
+ "step": 21
+ },
+ {
+ "epoch": 0.05994550408719346,
+ "grad_norm": 32.57571029663086,
+ "learning_rate": 3.996366939146231e-07,
+ "loss": 6.6289,
+ "step": 22
+ },
+ {
+ "epoch": 0.06267029972752043,
+ "grad_norm": 46.752933502197266,
+ "learning_rate": 4.178019981834696e-07,
+ "loss": 6.627,
+ "step": 23
+ },
+ {
+ "epoch": 0.0653950953678474,
+ "grad_norm": 36.405391693115234,
+ "learning_rate": 4.359673024523161e-07,
+ "loss": 6.4492,
+ "step": 24
+ },
+ {
+ "epoch": 0.0681198910081744,
+ "grad_norm": 48.150001525878906,
+ "learning_rate": 4.541326067211626e-07,
+ "loss": 6.3262,
+ "step": 25
+ },
+ {
+ "epoch": 0.07084468664850137,
+ "grad_norm": 32.77824020385742,
+ "learning_rate": 4.722979109900091e-07,
+ "loss": 6.5488,
+ "step": 26
+ },
+ {
+ "epoch": 0.07356948228882834,
+ "grad_norm": 32.786346435546875,
+ "learning_rate": 4.904632152588557e-07,
+ "loss": 6.5781,
+ "step": 27
+ },
+ {
+ "epoch": 0.07629427792915532,
+ "grad_norm": 33.72233200073242,
+ "learning_rate": 5.086285195277022e-07,
+ "loss": 6.3965,
+ "step": 28
+ },
+ {
+ "epoch": 0.07901907356948229,
+ "grad_norm": 49.315731048583984,
+ "learning_rate": 5.267938237965486e-07,
+ "loss": 6.7324,
+ "step": 29
+ },
+ {
+ "epoch": 0.08174386920980926,
+ "grad_norm": 38.151973724365234,
+ "learning_rate": 5.449591280653951e-07,
+ "loss": 6.625,
+ "step": 30
+ },
+ {
+ "epoch": 0.08446866485013624,
+ "grad_norm": 41.54045486450195,
+ "learning_rate": 5.631244323342416e-07,
+ "loss": 6.7617,
+ "step": 31
+ },
+ {
+ "epoch": 0.08719346049046321,
+ "grad_norm": 33.26874923706055,
+ "learning_rate": 5.812897366030882e-07,
+ "loss": 6.5879,
+ "step": 32
+ },
+ {
+ "epoch": 0.08991825613079019,
+ "grad_norm": 34.39425277709961,
+ "learning_rate": 5.994550408719347e-07,
+ "loss": 6.207,
+ "step": 33
+ },
+ {
+ "epoch": 0.09264305177111716,
+ "grad_norm": 37.94644546508789,
+ "learning_rate": 6.176203451407811e-07,
+ "loss": 6.5391,
+ "step": 34
+ },
+ {
+ "epoch": 0.09536784741144415,
+ "grad_norm": 27.162321090698242,
+ "learning_rate": 6.357856494096276e-07,
+ "loss": 6.4746,
+ "step": 35
+ },
+ {
+ "epoch": 0.09809264305177112,
+ "grad_norm": 30.717914581298828,
+ "learning_rate": 6.539509536784741e-07,
+ "loss": 6.3262,
+ "step": 36
+ },
+ {
+ "epoch": 0.1008174386920981,
+ "grad_norm": 44.60021209716797,
+ "learning_rate": 6.721162579473206e-07,
+ "loss": 6.1602,
+ "step": 37
+ },
+ {
+ "epoch": 0.10354223433242507,
+ "grad_norm": 29.34564208984375,
+ "learning_rate": 6.902815622161673e-07,
+ "loss": 6.418,
+ "step": 38
+ },
+ {
+ "epoch": 0.10626702997275204,
+ "grad_norm": 29.27506446838379,
+ "learning_rate": 7.084468664850137e-07,
+ "loss": 6.377,
+ "step": 39
+ },
+ {
+ "epoch": 0.10899182561307902,
+ "grad_norm": 32.7628288269043,
+ "learning_rate": 7.266121707538602e-07,
+ "loss": 6.5215,
+ "step": 40
+ },
+ {
+ "epoch": 0.11171662125340599,
+ "grad_norm": 26.057865142822266,
+ "learning_rate": 7.447774750227067e-07,
+ "loss": 6.291,
+ "step": 41
+ },
+ {
+ "epoch": 0.11444141689373297,
+ "grad_norm": 40.639034271240234,
+ "learning_rate": 7.629427792915532e-07,
+ "loss": 6.2383,
+ "step": 42
+ },
+ {
+ "epoch": 0.11716621253405994,
+ "grad_norm": 36.71115493774414,
+ "learning_rate": 7.811080835603997e-07,
+ "loss": 6.209,
+ "step": 43
+ },
+ {
+ "epoch": 0.11989100817438691,
+ "grad_norm": 28.97784996032715,
+ "learning_rate": 7.992733878292462e-07,
+ "loss": 6.3652,
+ "step": 44
+ },
+ {
+ "epoch": 0.1226158038147139,
+ "grad_norm": 25.279220581054688,
+ "learning_rate": 8.174386920980927e-07,
+ "loss": 5.8691,
+ "step": 45
+ },
+ {
+ "epoch": 0.12534059945504086,
+ "grad_norm": 39.73528289794922,
+ "learning_rate": 8.356039963669392e-07,
+ "loss": 5.8438,
+ "step": 46
+ },
+ {
+ "epoch": 0.12806539509536785,
+ "grad_norm": 30.93120765686035,
+ "learning_rate": 8.537693006357857e-07,
+ "loss": 5.7188,
+ "step": 47
+ },
+ {
+ "epoch": 0.1307901907356948,
+ "grad_norm": 26.659875869750977,
+ "learning_rate": 8.719346049046322e-07,
+ "loss": 6.084,
+ "step": 48
+ },
+ {
+ "epoch": 0.1335149863760218,
+ "grad_norm": 24.064189910888672,
+ "learning_rate": 8.900999091734786e-07,
+ "loss": 5.8418,
+ "step": 49
+ },
+ {
+ "epoch": 0.1362397820163488,
+ "grad_norm": 30.40485382080078,
+ "learning_rate": 9.082652134423252e-07,
+ "loss": 5.6426,
+ "step": 50
+ },
+ {
+ "epoch": 0.13896457765667575,
+ "grad_norm": 32.37661361694336,
+ "learning_rate": 9.264305177111717e-07,
+ "loss": 5.6387,
+ "step": 51
+ },
+ {
+ "epoch": 0.14168937329700274,
+ "grad_norm": 36.3615608215332,
+ "learning_rate": 9.445958219800182e-07,
+ "loss": 5.6289,
+ "step": 52
+ },
+ {
+ "epoch": 0.1444141689373297,
+ "grad_norm": 24.966564178466797,
+ "learning_rate": 9.627611262488648e-07,
+ "loss": 5.7168,
+ "step": 53
+ },
+ {
+ "epoch": 0.14713896457765668,
+ "grad_norm": 25.391984939575195,
+ "learning_rate": 9.809264305177114e-07,
+ "loss": 5.6152,
+ "step": 54
+ },
+ {
+ "epoch": 0.14986376021798364,
+ "grad_norm": 24.642980575561523,
+ "learning_rate": 9.990917347865577e-07,
+ "loss": 5.6621,
+ "step": 55
+ },
+ {
+ "epoch": 0.15258855585831063,
+ "grad_norm": 29.580062866210938,
+ "learning_rate": 1.0172570390554043e-06,
+ "loss": 5.4453,
+ "step": 56
+ },
+ {
+ "epoch": 0.1553133514986376,
+ "grad_norm": 28.979787826538086,
+ "learning_rate": 1.0354223433242507e-06,
+ "loss": 5.5742,
+ "step": 57
+ },
+ {
+ "epoch": 0.15803814713896458,
+ "grad_norm": 35.8126220703125,
+ "learning_rate": 1.0535876475930973e-06,
+ "loss": 5.5918,
+ "step": 58
+ },
+ {
+ "epoch": 0.16076294277929154,
+ "grad_norm": 50.86345291137695,
+ "learning_rate": 1.0717529518619439e-06,
+ "loss": 5.541,
+ "step": 59
+ },
+ {
+ "epoch": 0.16348773841961853,
+ "grad_norm": 48.65726852416992,
+ "learning_rate": 1.0899182561307902e-06,
+ "loss": 5.4824,
+ "step": 60
+ },
+ {
+ "epoch": 0.16621253405994552,
+ "grad_norm": 34.94480895996094,
+ "learning_rate": 1.1080835603996368e-06,
+ "loss": 5.5,
+ "step": 61
+ },
+ {
+ "epoch": 0.16893732970027248,
+ "grad_norm": 25.906768798828125,
+ "learning_rate": 1.1262488646684832e-06,
+ "loss": 5.2842,
+ "step": 62
+ },
+ {
+ "epoch": 0.17166212534059946,
+ "grad_norm": 28.621320724487305,
+ "learning_rate": 1.1444141689373298e-06,
+ "loss": 5.082,
+ "step": 63
+ },
+ {
+ "epoch": 0.17438692098092642,
+ "grad_norm": 76.06676483154297,
+ "learning_rate": 1.1625794732061764e-06,
+ "loss": 5.0586,
+ "step": 64
+ },
+ {
+ "epoch": 0.1771117166212534,
+ "grad_norm": 29.4881591796875,
+ "learning_rate": 1.1807447774750227e-06,
+ "loss": 4.9453,
+ "step": 65
+ },
+ {
+ "epoch": 0.17983651226158037,
+ "grad_norm": 93.43070983886719,
+ "learning_rate": 1.1989100817438693e-06,
+ "loss": 5.1123,
+ "step": 66
+ },
+ {
+ "epoch": 0.18256130790190736,
+ "grad_norm": 35.595359802246094,
+ "learning_rate": 1.217075386012716e-06,
+ "loss": 4.9238,
+ "step": 67
+ },
+ {
+ "epoch": 0.18528610354223432,
+ "grad_norm": 31.6319522857666,
+ "learning_rate": 1.2352406902815623e-06,
+ "loss": 4.9932,
+ "step": 68
+ },
+ {
+ "epoch": 0.1880108991825613,
+ "grad_norm": 21.580936431884766,
+ "learning_rate": 1.2534059945504087e-06,
+ "loss": 4.6309,
+ "step": 69
+ },
+ {
+ "epoch": 0.1907356948228883,
+ "grad_norm": 28.571224212646484,
+ "learning_rate": 1.2715712988192552e-06,
+ "loss": 4.9658,
+ "step": 70
+ },
+ {
+ "epoch": 0.19346049046321526,
+ "grad_norm": 26.74237060546875,
+ "learning_rate": 1.2897366030881018e-06,
+ "loss": 4.6211,
+ "step": 71
+ },
+ {
+ "epoch": 0.19618528610354224,
+ "grad_norm": 23.6983642578125,
+ "learning_rate": 1.3079019073569482e-06,
+ "loss": 4.7422,
+ "step": 72
+ },
+ {
+ "epoch": 0.1989100817438692,
+ "grad_norm": 17.95784568786621,
+ "learning_rate": 1.3260672116257948e-06,
+ "loss": 4.5059,
+ "step": 73
+ },
+ {
+ "epoch": 0.2016348773841962,
+ "grad_norm": 18.964731216430664,
+ "learning_rate": 1.3442325158946412e-06,
+ "loss": 4.3389,
+ "step": 74
+ },
+ {
+ "epoch": 0.20435967302452315,
+ "grad_norm": 15.5343599319458,
+ "learning_rate": 1.362397820163488e-06,
+ "loss": 4.5576,
+ "step": 75
+ },
+ {
+ "epoch": 0.20708446866485014,
+ "grad_norm": 14.381880760192871,
+ "learning_rate": 1.3805631244323345e-06,
+ "loss": 4.6758,
+ "step": 76
+ },
+ {
+ "epoch": 0.2098092643051771,
+ "grad_norm": 17.940288543701172,
+ "learning_rate": 1.398728428701181e-06,
+ "loss": 4.5156,
+ "step": 77
+ },
+ {
+ "epoch": 0.2125340599455041,
+ "grad_norm": 19.840164184570312,
+ "learning_rate": 1.4168937329700275e-06,
+ "loss": 4.5225,
+ "step": 78
+ },
+ {
+ "epoch": 0.21525885558583105,
+ "grad_norm": 17.79146957397461,
+ "learning_rate": 1.4350590372388739e-06,
+ "loss": 4.5449,
+ "step": 79
+ },
+ {
+ "epoch": 0.21798365122615804,
+ "grad_norm": 14.19536018371582,
+ "learning_rate": 1.4532243415077205e-06,
+ "loss": 4.1006,
+ "step": 80
+ },
+ {
+ "epoch": 0.22070844686648503,
+ "grad_norm": 11.174647331237793,
+ "learning_rate": 1.471389645776567e-06,
+ "loss": 4.3047,
+ "step": 81
+ },
+ {
+ "epoch": 0.22343324250681199,
+ "grad_norm": 19.54248809814453,
+ "learning_rate": 1.4895549500454134e-06,
+ "loss": 4.6221,
+ "step": 82
+ },
+ {
+ "epoch": 0.22615803814713897,
+ "grad_norm": 16.178730010986328,
+ "learning_rate": 1.50772025431426e-06,
+ "loss": 4.3145,
+ "step": 83
+ },
+ {
+ "epoch": 0.22888283378746593,
+ "grad_norm": 14.43359088897705,
+ "learning_rate": 1.5258855585831064e-06,
+ "loss": 4.2471,
+ "step": 84
+ },
+ {
+ "epoch": 0.23160762942779292,
+ "grad_norm": 13.826620101928711,
+ "learning_rate": 1.544050862851953e-06,
+ "loss": 4.5771,
+ "step": 85
+ },
+ {
+ "epoch": 0.23433242506811988,
+ "grad_norm": 15.556352615356445,
+ "learning_rate": 1.5622161671207993e-06,
+ "loss": 4.376,
+ "step": 86
+ },
+ {
+ "epoch": 0.23705722070844687,
+ "grad_norm": 11.810469627380371,
+ "learning_rate": 1.580381471389646e-06,
+ "loss": 4.3232,
+ "step": 87
+ },
+ {
+ "epoch": 0.23978201634877383,
+ "grad_norm": 17.136137008666992,
+ "learning_rate": 1.5985467756584925e-06,
+ "loss": 4.3037,
+ "step": 88
+ },
+ {
+ "epoch": 0.24250681198910082,
+ "grad_norm": 10.769259452819824,
+ "learning_rate": 1.6167120799273389e-06,
+ "loss": 4.0938,
+ "step": 89
+ },
+ {
+ "epoch": 0.2452316076294278,
+ "grad_norm": 14.444097518920898,
+ "learning_rate": 1.6348773841961855e-06,
+ "loss": 4.1367,
+ "step": 90
+ },
+ {
+ "epoch": 0.24795640326975477,
+ "grad_norm": 15.972183227539062,
+ "learning_rate": 1.6530426884650318e-06,
+ "loss": 4.2217,
+ "step": 91
+ },
+ {
+ "epoch": 0.2506811989100817,
+ "grad_norm": 10.982805252075195,
+ "learning_rate": 1.6712079927338784e-06,
+ "loss": 4.2949,
+ "step": 92
+ },
+ {
+ "epoch": 0.25340599455040874,
+ "grad_norm": 14.811541557312012,
+ "learning_rate": 1.689373297002725e-06,
+ "loss": 4.0723,
+ "step": 93
+ },
+ {
+ "epoch": 0.2561307901907357,
+ "grad_norm": 16.5225830078125,
+ "learning_rate": 1.7075386012715714e-06,
+ "loss": 3.9609,
+ "step": 94
+ },
+ {
+ "epoch": 0.25885558583106266,
+ "grad_norm": 30.34742546081543,
+ "learning_rate": 1.725703905540418e-06,
+ "loss": 4.0859,
+ "step": 95
+ },
+ {
+ "epoch": 0.2615803814713896,
+ "grad_norm": 11.73492431640625,
+ "learning_rate": 1.7438692098092643e-06,
+ "loss": 4.1816,
+ "step": 96
+ },
+ {
+ "epoch": 0.26430517711171664,
+ "grad_norm": 15.404382705688477,
+ "learning_rate": 1.762034514078111e-06,
+ "loss": 3.8652,
+ "step": 97
+ },
+ {
+ "epoch": 0.2670299727520436,
+ "grad_norm": 18.77298355102539,
+ "learning_rate": 1.7801998183469573e-06,
+ "loss": 3.7158,
+ "step": 98
+ },
+ {
+ "epoch": 0.26975476839237056,
+ "grad_norm": 11.848769187927246,
+ "learning_rate": 1.7983651226158039e-06,
+ "loss": 3.6748,
+ "step": 99
+ },
+ {
+ "epoch": 0.2724795640326976,
+ "grad_norm": 12.720091819763184,
+ "learning_rate": 1.8165304268846505e-06,
+ "loss": 4.3242,
+ "step": 100
+ },
+ {
+ "epoch": 0.27520435967302453,
+ "grad_norm": 10.321704864501953,
+ "learning_rate": 1.8346957311534968e-06,
+ "loss": 4.0615,
+ "step": 101
+ },
+ {
+ "epoch": 0.2779291553133515,
+ "grad_norm": 10.003022193908691,
+ "learning_rate": 1.8528610354223434e-06,
+ "loss": 3.8604,
+ "step": 102
+ },
+ {
+ "epoch": 0.28065395095367845,
+ "grad_norm": 14.036397933959961,
+ "learning_rate": 1.8710263396911898e-06,
+ "loss": 3.9707,
+ "step": 103
+ },
+ {
+ "epoch": 0.28337874659400547,
+ "grad_norm": 11.570597648620605,
+ "learning_rate": 1.8891916439600364e-06,
+ "loss": 3.6162,
+ "step": 104
+ },
+ {
+ "epoch": 0.28610354223433243,
+ "grad_norm": 13.200033187866211,
+ "learning_rate": 1.907356948228883e-06,
+ "loss": 3.8457,
+ "step": 105
+ },
+ {
+ "epoch": 0.2888283378746594,
+ "grad_norm": 15.501534461975098,
+ "learning_rate": 1.9255222524977295e-06,
+ "loss": 3.7734,
+ "step": 106
+ },
+ {
+ "epoch": 0.29155313351498635,
+ "grad_norm": 10.10851764678955,
+ "learning_rate": 1.943687556766576e-06,
+ "loss": 3.9707,
+ "step": 107
+ },
+ {
+ "epoch": 0.29427792915531337,
+ "grad_norm": 11.347644805908203,
+ "learning_rate": 1.9618528610354227e-06,
+ "loss": 3.6494,
+ "step": 108
+ },
+ {
+ "epoch": 0.2970027247956403,
+ "grad_norm": 10.150016784667969,
+ "learning_rate": 1.9800181653042693e-06,
+ "loss": 3.8184,
+ "step": 109
+ },
+ {
+ "epoch": 0.2997275204359673,
+ "grad_norm": 13.109970092773438,
+ "learning_rate": 1.9981834695731155e-06,
+ "loss": 3.791,
+ "step": 110
+ },
+ {
+ "epoch": 0.3024523160762943,
+ "grad_norm": 9.569967269897461,
+ "learning_rate": 2.016348773841962e-06,
+ "loss": 3.8574,
+ "step": 111
+ },
+ {
+ "epoch": 0.30517711171662126,
+ "grad_norm": 15.210165977478027,
+ "learning_rate": 2.0345140781108086e-06,
+ "loss": 3.9502,
+ "step": 112
+ },
+ {
+ "epoch": 0.3079019073569482,
+ "grad_norm": 13.579336166381836,
+ "learning_rate": 2.0526793823796552e-06,
+ "loss": 3.4453,
+ "step": 113
+ },
+ {
+ "epoch": 0.3106267029972752,
+ "grad_norm": 11.369577407836914,
+ "learning_rate": 2.0708446866485014e-06,
+ "loss": 3.666,
+ "step": 114
+ },
+ {
+ "epoch": 0.3133514986376022,
+ "grad_norm": 9.641692161560059,
+ "learning_rate": 2.089009990917348e-06,
+ "loss": 4.0488,
+ "step": 115
+ },
+ {
+ "epoch": 0.31607629427792916,
+ "grad_norm": 8.64608097076416,
+ "learning_rate": 2.1071752951861945e-06,
+ "loss": 3.7656,
+ "step": 116
+ },
+ {
+ "epoch": 0.3188010899182561,
+ "grad_norm": 10.101052284240723,
+ "learning_rate": 2.125340599455041e-06,
+ "loss": 3.8604,
+ "step": 117
+ },
+ {
+ "epoch": 0.3215258855585831,
+ "grad_norm": 9.095230102539062,
+ "learning_rate": 2.1435059037238877e-06,
+ "loss": 3.4141,
+ "step": 118
+ },
+ {
+ "epoch": 0.3242506811989101,
+ "grad_norm": 9.79998779296875,
+ "learning_rate": 2.161671207992734e-06,
+ "loss": 3.6123,
+ "step": 119
+ },
+ {
+ "epoch": 0.32697547683923706,
+ "grad_norm": 10.831903457641602,
+ "learning_rate": 2.1798365122615805e-06,
+ "loss": 3.8965,
+ "step": 120
+ },
+ {
+ "epoch": 0.329700272479564,
+ "grad_norm": 18.380979537963867,
+ "learning_rate": 2.198001816530427e-06,
+ "loss": 3.7705,
+ "step": 121
+ },
+ {
+ "epoch": 0.33242506811989103,
+ "grad_norm": 11.21811580657959,
+ "learning_rate": 2.2161671207992736e-06,
+ "loss": 3.7812,
+ "step": 122
+ },
+ {
+ "epoch": 0.335149863760218,
+ "grad_norm": 9.772245407104492,
+ "learning_rate": 2.2343324250681202e-06,
+ "loss": 3.2754,
+ "step": 123
+ },
+ {
+ "epoch": 0.33787465940054495,
+ "grad_norm": 10.682083129882812,
+ "learning_rate": 2.2524977293369664e-06,
+ "loss": 3.7627,
+ "step": 124
+ },
+ {
+ "epoch": 0.3405994550408719,
+ "grad_norm": 11.625896453857422,
+ "learning_rate": 2.270663033605813e-06,
+ "loss": 3.6289,
+ "step": 125
+ },
+ {
+ "epoch": 0.34332425068119893,
+ "grad_norm": 10.377275466918945,
+ "learning_rate": 2.2888283378746596e-06,
+ "loss": 3.8574,
+ "step": 126
+ },
+ {
+ "epoch": 0.3460490463215259,
+ "grad_norm": 11.314308166503906,
+ "learning_rate": 2.306993642143506e-06,
+ "loss": 3.4912,
+ "step": 127
+ },
+ {
+ "epoch": 0.34877384196185285,
+ "grad_norm": 8.465714454650879,
+ "learning_rate": 2.3251589464123527e-06,
+ "loss": 3.751,
+ "step": 128
+ },
+ {
+ "epoch": 0.35149863760217986,
+ "grad_norm": 8.958711624145508,
+ "learning_rate": 2.343324250681199e-06,
+ "loss": 3.4004,
+ "step": 129
+ },
+ {
+ "epoch": 0.3542234332425068,
+ "grad_norm": 9.359971046447754,
+ "learning_rate": 2.3614895549500455e-06,
+ "loss": 3.4033,
+ "step": 130
+ },
+ {
+ "epoch": 0.3569482288828338,
+ "grad_norm": 14.603437423706055,
+ "learning_rate": 2.379654859218892e-06,
+ "loss": 3.8086,
+ "step": 131
+ },
+ {
+ "epoch": 0.35967302452316074,
+ "grad_norm": 9.116939544677734,
+ "learning_rate": 2.3978201634877386e-06,
+ "loss": 3.6582,
+ "step": 132
+ },
+ {
+ "epoch": 0.36239782016348776,
+ "grad_norm": 12.235628128051758,
+ "learning_rate": 2.415985467756585e-06,
+ "loss": 3.6934,
+ "step": 133
+ },
+ {
+ "epoch": 0.3651226158038147,
+ "grad_norm": 9.428214073181152,
+ "learning_rate": 2.434150772025432e-06,
+ "loss": 3.4297,
+ "step": 134
+ },
+ {
+ "epoch": 0.3678474114441417,
+ "grad_norm": 9.38826847076416,
+ "learning_rate": 2.4523160762942784e-06,
+ "loss": 3.6426,
+ "step": 135
+ },
+ {
+ "epoch": 0.37057220708446864,
+ "grad_norm": 10.444653511047363,
+ "learning_rate": 2.4704813805631246e-06,
+ "loss": 3.54,
+ "step": 136
+ },
+ {
+ "epoch": 0.37329700272479566,
+ "grad_norm": 8.682459831237793,
+ "learning_rate": 2.488646684831971e-06,
+ "loss": 3.5508,
+ "step": 137
+ },
+ {
+ "epoch": 0.3760217983651226,
+ "grad_norm": 9.243608474731445,
+ "learning_rate": 2.5068119891008173e-06,
+ "loss": 3.5635,
+ "step": 138
+ },
+ {
+ "epoch": 0.3787465940054496,
+ "grad_norm": 8.790285110473633,
+ "learning_rate": 2.5249772933696643e-06,
+ "loss": 3.4053,
+ "step": 139
+ },
+ {
+ "epoch": 0.3814713896457766,
+ "grad_norm": 11.250055313110352,
+ "learning_rate": 2.5431425976385105e-06,
+ "loss": 3.3555,
+ "step": 140
+ },
+ {
+ "epoch": 0.38419618528610355,
+ "grad_norm": 8.680200576782227,
+ "learning_rate": 2.561307901907357e-06,
+ "loss": 3.3975,
+ "step": 141
+ },
+ {
+ "epoch": 0.3869209809264305,
+ "grad_norm": 10.055103302001953,
+ "learning_rate": 2.5794732061762036e-06,
+ "loss": 3.498,
+ "step": 142
+ },
+ {
+ "epoch": 0.3896457765667575,
+ "grad_norm": 9.284025192260742,
+ "learning_rate": 2.5976385104450502e-06,
+ "loss": 3.4365,
+ "step": 143
+ },
+ {
+ "epoch": 0.3923705722070845,
+ "grad_norm": 8.481389999389648,
+ "learning_rate": 2.6158038147138964e-06,
+ "loss": 3.3184,
+ "step": 144
+ },
+ {
+ "epoch": 0.39509536784741145,
+ "grad_norm": 8.089268684387207,
+ "learning_rate": 2.6339691189827434e-06,
+ "loss": 3.3809,
+ "step": 145
+ },
+ {
+ "epoch": 0.3978201634877384,
+ "grad_norm": 11.318544387817383,
+ "learning_rate": 2.6521344232515896e-06,
+ "loss": 3.5811,
+ "step": 146
+ },
+ {
+ "epoch": 0.40054495912806537,
+ "grad_norm": 10.320113182067871,
+ "learning_rate": 2.670299727520436e-06,
+ "loss": 3.4785,
+ "step": 147
+ },
+ {
+ "epoch": 0.4032697547683924,
+ "grad_norm": 8.867465019226074,
+ "learning_rate": 2.6884650317892823e-06,
+ "loss": 3.5459,
+ "step": 148
+ },
+ {
+ "epoch": 0.40599455040871935,
+ "grad_norm": 9.647247314453125,
+ "learning_rate": 2.7066303360581293e-06,
+ "loss": 3.6279,
+ "step": 149
+ },
+ {
+ "epoch": 0.4087193460490463,
+ "grad_norm": 8.813453674316406,
+ "learning_rate": 2.724795640326976e-06,
+ "loss": 3.2432,
+ "step": 150
+ },
+ {
+ "epoch": 0.4114441416893733,
+ "grad_norm": 9.403165817260742,
+ "learning_rate": 2.742960944595822e-06,
+ "loss": 3.2559,
+ "step": 151
+ },
+ {
+ "epoch": 0.4141689373297003,
+ "grad_norm": 8.76225471496582,
+ "learning_rate": 2.761126248864669e-06,
+ "loss": 3.6279,
+ "step": 152
+ },
+ {
+ "epoch": 0.41689373297002724,
+ "grad_norm": 8.464751243591309,
+ "learning_rate": 2.7792915531335152e-06,
+ "loss": 3.332,
+ "step": 153
+ },
+ {
+ "epoch": 0.4196185286103542,
+ "grad_norm": 13.83745288848877,
+ "learning_rate": 2.797456857402362e-06,
+ "loss": 3.4092,
+ "step": 154
+ },
+ {
+ "epoch": 0.4223433242506812,
+ "grad_norm": 8.37870979309082,
+ "learning_rate": 2.815622161671208e-06,
+ "loss": 3.3809,
+ "step": 155
+ },
+ {
+ "epoch": 0.4250681198910082,
+ "grad_norm": 9.180736541748047,
+ "learning_rate": 2.833787465940055e-06,
+ "loss": 3.3125,
+ "step": 156
+ },
+ {
+ "epoch": 0.42779291553133514,
+ "grad_norm": 13.312505722045898,
+ "learning_rate": 2.851952770208901e-06,
+ "loss": 3.4941,
+ "step": 157
+ },
+ {
+ "epoch": 0.4305177111716621,
+ "grad_norm": 8.892487525939941,
+ "learning_rate": 2.8701180744777477e-06,
+ "loss": 3.4863,
+ "step": 158
+ },
+ {
+ "epoch": 0.4332425068119891,
+ "grad_norm": 9.126548767089844,
+ "learning_rate": 2.8882833787465943e-06,
+ "loss": 3.3066,
+ "step": 159
+ },
+ {
+ "epoch": 0.4359673024523161,
+ "grad_norm": 7.950293064117432,
+ "learning_rate": 2.906448683015441e-06,
+ "loss": 3.3311,
+ "step": 160
+ },
+ {
+ "epoch": 0.43869209809264303,
+ "grad_norm": 7.878985404968262,
+ "learning_rate": 2.924613987284287e-06,
+ "loss": 3.3828,
+ "step": 161
+ },
+ {
+ "epoch": 0.44141689373297005,
+ "grad_norm": 8.849200248718262,
+ "learning_rate": 2.942779291553134e-06,
+ "loss": 3.3115,
+ "step": 162
+ },
+ {
+ "epoch": 0.444141689373297,
+ "grad_norm": 7.619378089904785,
+ "learning_rate": 2.9609445958219802e-06,
+ "loss": 3.1143,
+ "step": 163
+ },
+ {
+ "epoch": 0.44686648501362397,
+ "grad_norm": 9.767971992492676,
+ "learning_rate": 2.979109900090827e-06,
+ "loss": 3.5225,
+ "step": 164
+ },
+ {
+ "epoch": 0.44959128065395093,
+ "grad_norm": 7.132298946380615,
+ "learning_rate": 2.997275204359673e-06,
+ "loss": 3.3223,
+ "step": 165
+ },
+ {
+ "epoch": 0.45231607629427795,
+ "grad_norm": 8.684154510498047,
+ "learning_rate": 3.01544050862852e-06,
+ "loss": 3.3379,
+ "step": 166
+ },
+ {
+ "epoch": 0.4550408719346049,
+ "grad_norm": 10.941788673400879,
+ "learning_rate": 3.033605812897366e-06,
+ "loss": 3.1924,
+ "step": 167
+ },
+ {
+ "epoch": 0.45776566757493187,
+ "grad_norm": 11.09579849243164,
+ "learning_rate": 3.0517711171662127e-06,
+ "loss": 3.2383,
+ "step": 168
+ },
+ {
+ "epoch": 0.4604904632152589,
+ "grad_norm": 8.943426132202148,
+ "learning_rate": 3.069936421435059e-06,
+ "loss": 3.3164,
+ "step": 169
+ },
+ {
+ "epoch": 0.46321525885558584,
+ "grad_norm": 8.510838508605957,
+ "learning_rate": 3.088101725703906e-06,
+ "loss": 3.498,
+ "step": 170
+ },
+ {
+ "epoch": 0.4659400544959128,
+ "grad_norm": 9.172945022583008,
+ "learning_rate": 3.106267029972752e-06,
+ "loss": 3.4883,
+ "step": 171
+ },
+ {
+ "epoch": 0.46866485013623976,
+ "grad_norm": 10.886733055114746,
+ "learning_rate": 3.1244323342415987e-06,
+ "loss": 3.4062,
+ "step": 172
+ },
+ {
+ "epoch": 0.4713896457765668,
+ "grad_norm": 7.259507179260254,
+ "learning_rate": 3.1425976385104452e-06,
+ "loss": 3.1191,
+ "step": 173
+ },
+ {
+ "epoch": 0.47411444141689374,
+ "grad_norm": 17.1456241607666,
+ "learning_rate": 3.160762942779292e-06,
+ "loss": 3.3936,
+ "step": 174
+ },
+ {
+ "epoch": 0.4768392370572207,
+ "grad_norm": 7.412659645080566,
+ "learning_rate": 3.178928247048138e-06,
+ "loss": 3.2783,
+ "step": 175
+ },
+ {
+ "epoch": 0.47956403269754766,
+ "grad_norm": 10.424003601074219,
+ "learning_rate": 3.197093551316985e-06,
+ "loss": 3.3506,
+ "step": 176
+ },
+ {
+ "epoch": 0.4822888283378747,
+ "grad_norm": 9.243874549865723,
+ "learning_rate": 3.215258855585831e-06,
+ "loss": 3.4668,
+ "step": 177
+ },
+ {
+ "epoch": 0.48501362397820164,
+ "grad_norm": 6.991046905517578,
+ "learning_rate": 3.2334241598546777e-06,
+ "loss": 3.2861,
+ "step": 178
+ },
+ {
+ "epoch": 0.4877384196185286,
+ "grad_norm": 8.809893608093262,
+ "learning_rate": 3.2515894641235247e-06,
+ "loss": 3.1914,
+ "step": 179
+ },
+ {
+ "epoch": 0.4904632152588556,
+ "grad_norm": 8.128570556640625,
+ "learning_rate": 3.269754768392371e-06,
+ "loss": 3.2861,
+ "step": 180
+ },
+ {
+ "epoch": 0.49318801089918257,
+ "grad_norm": 9.240766525268555,
+ "learning_rate": 3.2879200726612175e-06,
+ "loss": 3.376,
+ "step": 181
+ },
+ {
+ "epoch": 0.49591280653950953,
+ "grad_norm": 7.798635482788086,
+ "learning_rate": 3.3060853769300637e-06,
+ "loss": 3.1436,
+ "step": 182
+ },
+ {
+ "epoch": 0.4986376021798365,
+ "grad_norm": 7.1677422523498535,
+ "learning_rate": 3.3242506811989107e-06,
+ "loss": 3.1553,
+ "step": 183
+ },
+ {
+ "epoch": 0.5013623978201635,
+ "grad_norm": 7.5272393226623535,
+ "learning_rate": 3.342415985467757e-06,
+ "loss": 3.1514,
+ "step": 184
+ },
+ {
+ "epoch": 0.5040871934604905,
+ "grad_norm": 9.626727104187012,
+ "learning_rate": 3.3605812897366034e-06,
+ "loss": 3.4561,
+ "step": 185
+ },
+ {
+ "epoch": 0.5068119891008175,
+ "grad_norm": 7.911482334136963,
+ "learning_rate": 3.37874659400545e-06,
+ "loss": 3.1074,
+ "step": 186
+ },
+ {
+ "epoch": 0.5095367847411444,
+ "grad_norm": 7.70817756652832,
+ "learning_rate": 3.3969118982742966e-06,
+ "loss": 3.2988,
+ "step": 187
+ },
+ {
+ "epoch": 0.5122615803814714,
+ "grad_norm": 15.456562995910645,
+ "learning_rate": 3.4150772025431427e-06,
+ "loss": 3.2002,
+ "step": 188
+ },
+ {
+ "epoch": 0.5149863760217984,
+ "grad_norm": 9.196293830871582,
+ "learning_rate": 3.4332425068119898e-06,
+ "loss": 3.2686,
+ "step": 189
+ },
+ {
+ "epoch": 0.5177111716621253,
+ "grad_norm": 7.613571643829346,
+ "learning_rate": 3.451407811080836e-06,
+ "loss": 2.9551,
+ "step": 190
+ },
+ {
+ "epoch": 0.5204359673024523,
+ "grad_norm": 7.925637245178223,
+ "learning_rate": 3.4695731153496825e-06,
+ "loss": 3.3584,
+ "step": 191
+ },
+ {
+ "epoch": 0.5231607629427792,
+ "grad_norm": 8.302532196044922,
+ "learning_rate": 3.4877384196185287e-06,
+ "loss": 3.0947,
+ "step": 192
+ },
+ {
+ "epoch": 0.5258855585831063,
+ "grad_norm": 8.558791160583496,
+ "learning_rate": 3.5059037238873757e-06,
+ "loss": 3.209,
+ "step": 193
+ },
+ {
+ "epoch": 0.5286103542234333,
+ "grad_norm": 8.977067947387695,
+ "learning_rate": 3.524069028156222e-06,
+ "loss": 3.3271,
+ "step": 194
+ },
+ {
+ "epoch": 0.5313351498637602,
+ "grad_norm": 7.109421253204346,
+ "learning_rate": 3.5422343324250684e-06,
+ "loss": 3.1582,
+ "step": 195
+ },
+ {
+ "epoch": 0.5340599455040872,
+ "grad_norm": 9.760332107543945,
+ "learning_rate": 3.5603996366939146e-06,
+ "loss": 3.0283,
+ "step": 196
+ },
+ {
+ "epoch": 0.5367847411444142,
+ "grad_norm": 9.577591896057129,
+ "learning_rate": 3.5785649409627616e-06,
+ "loss": 3.0195,
+ "step": 197
+ },
+ {
+ "epoch": 0.5395095367847411,
+ "grad_norm": 9.317022323608398,
+ "learning_rate": 3.5967302452316077e-06,
+ "loss": 2.8823,
+ "step": 198
+ },
+ {
+ "epoch": 0.5422343324250681,
+ "grad_norm": 8.604791641235352,
+ "learning_rate": 3.6148955495004543e-06,
+ "loss": 3.1699,
+ "step": 199
+ },
+ {
+ "epoch": 0.5449591280653951,
+ "grad_norm": 9.316139221191406,
+ "learning_rate": 3.633060853769301e-06,
+ "loss": 3.1865,
+ "step": 200
+ },
+ {
+ "epoch": 0.547683923705722,
+ "grad_norm": 8.660322189331055,
+ "learning_rate": 3.6512261580381475e-06,
+ "loss": 2.8271,
+ "step": 201
+ },
+ {
+ "epoch": 0.5504087193460491,
+ "grad_norm": 8.291754722595215,
+ "learning_rate": 3.6693914623069937e-06,
+ "loss": 3.1182,
+ "step": 202
+ },
+ {
+ "epoch": 0.553133514986376,
+ "grad_norm": 10.406461715698242,
+ "learning_rate": 3.6875567665758407e-06,
+ "loss": 3.1953,
+ "step": 203
+ },
+ {
+ "epoch": 0.555858310626703,
+ "grad_norm": 7.478999614715576,
+ "learning_rate": 3.705722070844687e-06,
+ "loss": 3.0176,
+ "step": 204
+ },
+ {
+ "epoch": 0.55858310626703,
+ "grad_norm": 7.893258094787598,
+ "learning_rate": 3.7238873751135334e-06,
+ "loss": 3.1904,
+ "step": 205
+ },
+ {
+ "epoch": 0.5613079019073569,
+ "grad_norm": 5.913602352142334,
+ "learning_rate": 3.7420526793823796e-06,
+ "loss": 3.1963,
+ "step": 206
+ },
+ {
+ "epoch": 0.5640326975476839,
+ "grad_norm": 8.231317520141602,
+ "learning_rate": 3.7602179836512266e-06,
+ "loss": 3.1885,
+ "step": 207
+ },
+ {
+ "epoch": 0.5667574931880109,
+ "grad_norm": 8.352523803710938,
+ "learning_rate": 3.7783832879200728e-06,
+ "loss": 3.3184,
+ "step": 208
+ },
+ {
+ "epoch": 0.5694822888283378,
+ "grad_norm": 7.702081203460693,
+ "learning_rate": 3.7965485921889193e-06,
+ "loss": 3.082,
+ "step": 209
+ },
+ {
+ "epoch": 0.5722070844686649,
+ "grad_norm": 6.759937763214111,
+ "learning_rate": 3.814713896457766e-06,
+ "loss": 3.0303,
+ "step": 210
+ },
+ {
+ "epoch": 0.5749318801089919,
+ "grad_norm": 8.072578430175781,
+ "learning_rate": 3.832879200726612e-06,
+ "loss": 3.5156,
+ "step": 211
+ },
+ {
+ "epoch": 0.5776566757493188,
+ "grad_norm": 9.648780822753906,
+ "learning_rate": 3.851044504995459e-06,
+ "loss": 3.2378,
+ "step": 212
+ },
+ {
+ "epoch": 0.5803814713896458,
+ "grad_norm": 12.612576484680176,
+ "learning_rate": 3.869209809264305e-06,
+ "loss": 3.4102,
+ "step": 213
+ },
+ {
+ "epoch": 0.5831062670299727,
+ "grad_norm": 7.714903354644775,
+ "learning_rate": 3.887375113533152e-06,
+ "loss": 3.1768,
+ "step": 214
+ },
+ {
+ "epoch": 0.5858310626702997,
+ "grad_norm": 7.917700290679932,
+ "learning_rate": 3.9055404178019984e-06,
+ "loss": 2.8613,
+ "step": 215
+ },
+ {
+ "epoch": 0.5885558583106267,
+ "grad_norm": 9.318526268005371,
+ "learning_rate": 3.9237057220708454e-06,
+ "loss": 3.1611,
+ "step": 216
+ },
+ {
+ "epoch": 0.5912806539509536,
+ "grad_norm": 8.743786811828613,
+ "learning_rate": 3.941871026339692e-06,
+ "loss": 3.3691,
+ "step": 217
+ },
+ {
+ "epoch": 0.5940054495912807,
+ "grad_norm": 15.29521369934082,
+ "learning_rate": 3.960036330608539e-06,
+ "loss": 3.0684,
+ "step": 218
+ },
+ {
+ "epoch": 0.5967302452316077,
+ "grad_norm": 8.825235366821289,
+ "learning_rate": 3.978201634877385e-06,
+ "loss": 3.2012,
+ "step": 219
+ },
+ {
+ "epoch": 0.5994550408719346,
+ "grad_norm": 7.756191253662109,
+ "learning_rate": 3.996366939146231e-06,
+ "loss": 3.1689,
+ "step": 220
+ },
+ {
+ "epoch": 0.6021798365122616,
+ "grad_norm": 7.637423515319824,
+ "learning_rate": 4.014532243415077e-06,
+ "loss": 3.1587,
+ "step": 221
+ },
+ {
+ "epoch": 0.6049046321525886,
+ "grad_norm": 10.598748207092285,
+ "learning_rate": 4.032697547683924e-06,
+ "loss": 2.9141,
+ "step": 222
+ },
+ {
+ "epoch": 0.6076294277929155,
+ "grad_norm": 6.778491973876953,
+ "learning_rate": 4.05086285195277e-06,
+ "loss": 3.0977,
+ "step": 223
+ },
+ {
+ "epoch": 0.6103542234332425,
+ "grad_norm": 7.174191951751709,
+ "learning_rate": 4.069028156221617e-06,
+ "loss": 2.9404,
+ "step": 224
+ },
+ {
+ "epoch": 0.6130790190735694,
+ "grad_norm": 10.684597969055176,
+ "learning_rate": 4.0871934604904634e-06,
+ "loss": 3.0586,
+ "step": 225
+ },
+ {
+ "epoch": 0.6158038147138964,
+ "grad_norm": 8.050337791442871,
+ "learning_rate": 4.1053587647593104e-06,
+ "loss": 3.1855,
+ "step": 226
+ },
+ {
+ "epoch": 0.6185286103542235,
+ "grad_norm": 8.274129867553711,
+ "learning_rate": 4.123524069028157e-06,
+ "loss": 3.0088,
+ "step": 227
+ },
+ {
+ "epoch": 0.6212534059945504,
+ "grad_norm": 11.98044204711914,
+ "learning_rate": 4.141689373297003e-06,
+ "loss": 3.1982,
+ "step": 228
+ },
+ {
+ "epoch": 0.6239782016348774,
+ "grad_norm": 6.696853160858154,
+ "learning_rate": 4.15985467756585e-06,
+ "loss": 2.793,
+ "step": 229
+ },
+ {
+ "epoch": 0.6267029972752044,
+ "grad_norm": 9.46538257598877,
+ "learning_rate": 4.178019981834696e-06,
+ "loss": 2.9468,
+ "step": 230
+ },
+ {
+ "epoch": 0.6294277929155313,
+ "grad_norm": 9.33486557006836,
+ "learning_rate": 4.196185286103542e-06,
+ "loss": 3.1514,
+ "step": 231
+ },
+ {
+ "epoch": 0.6321525885558583,
+ "grad_norm": 7.860947608947754,
+ "learning_rate": 4.214350590372389e-06,
+ "loss": 3.1055,
+ "step": 232
+ },
+ {
+ "epoch": 0.6348773841961853,
+ "grad_norm": 11.648983001708984,
+ "learning_rate": 4.232515894641235e-06,
+ "loss": 3.1367,
+ "step": 233
+ },
+ {
+ "epoch": 0.6376021798365122,
+ "grad_norm": 7.646440505981445,
+ "learning_rate": 4.250681198910082e-06,
+ "loss": 3.1396,
+ "step": 234
+ },
+ {
+ "epoch": 0.6403269754768393,
+ "grad_norm": 8.1253662109375,
+ "learning_rate": 4.2688465031789284e-06,
+ "loss": 3.0723,
+ "step": 235
+ },
+ {
+ "epoch": 0.6430517711171662,
+ "grad_norm": 7.996518135070801,
+ "learning_rate": 4.2870118074477754e-06,
+ "loss": 3.1123,
+ "step": 236
+ },
+ {
+ "epoch": 0.6457765667574932,
+ "grad_norm": 7.63090181350708,
+ "learning_rate": 4.305177111716622e-06,
+ "loss": 3.0645,
+ "step": 237
+ },
+ {
+ "epoch": 0.6485013623978202,
+ "grad_norm": 8.04654312133789,
+ "learning_rate": 4.323342415985468e-06,
+ "loss": 3.1094,
+ "step": 238
+ },
+ {
+ "epoch": 0.6512261580381471,
+ "grad_norm": 7.059414386749268,
+ "learning_rate": 4.341507720254315e-06,
+ "loss": 2.8076,
+ "step": 239
+ },
+ {
+ "epoch": 0.6539509536784741,
+ "grad_norm": 9.941123008728027,
+ "learning_rate": 4.359673024523161e-06,
+ "loss": 3.0684,
+ "step": 240
+ },
+ {
+ "epoch": 0.6566757493188011,
+ "grad_norm": 7.5402512550354,
+ "learning_rate": 4.377838328792008e-06,
+ "loss": 2.7773,
+ "step": 241
+ },
+ {
+ "epoch": 0.659400544959128,
+ "grad_norm": 8.088776588439941,
+ "learning_rate": 4.396003633060854e-06,
+ "loss": 2.958,
+ "step": 242
+ },
+ {
+ "epoch": 0.662125340599455,
+ "grad_norm": 7.672940731048584,
+ "learning_rate": 4.414168937329701e-06,
+ "loss": 3.0044,
+ "step": 243
+ },
+ {
+ "epoch": 0.6648501362397821,
+ "grad_norm": 7.196352005004883,
+ "learning_rate": 4.432334241598547e-06,
+ "loss": 2.8838,
+ "step": 244
+ },
+ {
+ "epoch": 0.667574931880109,
+ "grad_norm": 7.73391056060791,
+ "learning_rate": 4.4504995458673934e-06,
+ "loss": 2.8579,
+ "step": 245
+ },
+ {
+ "epoch": 0.670299727520436,
+ "grad_norm": 8.748419761657715,
+ "learning_rate": 4.4686648501362404e-06,
+ "loss": 3.1045,
+ "step": 246
+ },
+ {
+ "epoch": 0.6730245231607629,
+ "grad_norm": 8.288137435913086,
+ "learning_rate": 4.486830154405087e-06,
+ "loss": 3.2236,
+ "step": 247
+ },
+ {
+ "epoch": 0.6757493188010899,
+ "grad_norm": 7.427812099456787,
+ "learning_rate": 4.504995458673933e-06,
+ "loss": 3.0244,
+ "step": 248
+ },
+ {
+ "epoch": 0.6784741144414169,
+ "grad_norm": 7.199627876281738,
+ "learning_rate": 4.52316076294278e-06,
+ "loss": 2.918,
+ "step": 249
+ },
+ {
+ "epoch": 0.6811989100817438,
+ "grad_norm": 8.961594581604004,
+ "learning_rate": 4.541326067211626e-06,
+ "loss": 2.8311,
+ "step": 250
+ },
+ {
+ "epoch": 0.6839237057220708,
+ "grad_norm": 7.496204853057861,
+ "learning_rate": 4.559491371480473e-06,
+ "loss": 2.8477,
+ "step": 251
+ },
+ {
+ "epoch": 0.6866485013623979,
+ "grad_norm": 6.2377705574035645,
+ "learning_rate": 4.577656675749319e-06,
+ "loss": 2.6333,
+ "step": 252
+ },
+ {
+ "epoch": 0.6893732970027248,
+ "grad_norm": 9.205436706542969,
+ "learning_rate": 4.595821980018166e-06,
+ "loss": 3.041,
+ "step": 253
+ },
+ {
+ "epoch": 0.6920980926430518,
+ "grad_norm": 8.152443885803223,
+ "learning_rate": 4.613987284287012e-06,
+ "loss": 3.0186,
+ "step": 254
+ },
+ {
+ "epoch": 0.6948228882833788,
+ "grad_norm": 9.457436561584473,
+ "learning_rate": 4.6321525885558584e-06,
+ "loss": 3.0024,
+ "step": 255
+ },
+ {
+ "epoch": 0.6975476839237057,
+ "grad_norm": 11.40542221069336,
+ "learning_rate": 4.6503178928247054e-06,
+ "loss": 2.8584,
+ "step": 256
+ },
+ {
+ "epoch": 0.7002724795640327,
+ "grad_norm": 8.246293067932129,
+ "learning_rate": 4.668483197093552e-06,
+ "loss": 2.8994,
+ "step": 257
+ },
+ {
+ "epoch": 0.7029972752043597,
+ "grad_norm": 9.196247100830078,
+ "learning_rate": 4.686648501362398e-06,
+ "loss": 3.0532,
+ "step": 258
+ },
+ {
+ "epoch": 0.7057220708446866,
+ "grad_norm": 10.090413093566895,
+ "learning_rate": 4.704813805631245e-06,
+ "loss": 2.7295,
+ "step": 259
+ },
+ {
+ "epoch": 0.7084468664850136,
+ "grad_norm": 9.046536445617676,
+ "learning_rate": 4.722979109900091e-06,
+ "loss": 2.8555,
+ "step": 260
+ },
+ {
+ "epoch": 0.7111716621253406,
+ "grad_norm": 9.540751457214355,
+ "learning_rate": 4.741144414168938e-06,
+ "loss": 3.0605,
+ "step": 261
+ },
+ {
+ "epoch": 0.7138964577656676,
+ "grad_norm": 8.165807723999023,
+ "learning_rate": 4.759309718437784e-06,
+ "loss": 3.2383,
+ "step": 262
+ },
+ {
+ "epoch": 0.7166212534059946,
+ "grad_norm": 7.435905456542969,
+ "learning_rate": 4.777475022706631e-06,
+ "loss": 3.0703,
+ "step": 263
+ },
+ {
+ "epoch": 0.7193460490463215,
+ "grad_norm": 9.493722915649414,
+ "learning_rate": 4.795640326975477e-06,
+ "loss": 2.8745,
+ "step": 264
+ },
+ {
+ "epoch": 0.7220708446866485,
+ "grad_norm": 8.346881866455078,
+ "learning_rate": 4.8138056312443234e-06,
+ "loss": 2.7549,
+ "step": 265
+ },
+ {
+ "epoch": 0.7247956403269755,
+ "grad_norm": 9.425370216369629,
+ "learning_rate": 4.83197093551317e-06,
+ "loss": 2.9614,
+ "step": 266
+ },
+ {
+ "epoch": 0.7275204359673024,
+ "grad_norm": 7.455288410186768,
+ "learning_rate": 4.850136239782017e-06,
+ "loss": 2.8506,
+ "step": 267
+ },
+ {
+ "epoch": 0.7302452316076294,
+ "grad_norm": 9.171257972717285,
+ "learning_rate": 4.868301544050864e-06,
+ "loss": 2.959,
+ "step": 268
+ },
+ {
+ "epoch": 0.7329700272479565,
+ "grad_norm": 8.975666999816895,
+ "learning_rate": 4.88646684831971e-06,
+ "loss": 3.0991,
+ "step": 269
+ },
+ {
+ "epoch": 0.7356948228882834,
+ "grad_norm": 7.8134355545043945,
+ "learning_rate": 4.904632152588557e-06,
+ "loss": 2.8994,
+ "step": 270
+ },
+ {
+ "epoch": 0.7384196185286104,
+ "grad_norm": 8.253534317016602,
+ "learning_rate": 4.922797456857403e-06,
+ "loss": 2.7666,
+ "step": 271
+ },
+ {
+ "epoch": 0.7411444141689373,
+ "grad_norm": 7.1971540451049805,
+ "learning_rate": 4.940962761126249e-06,
+ "loss": 2.7188,
+ "step": 272
+ },
+ {
+ "epoch": 0.7438692098092643,
+ "grad_norm": 9.199867248535156,
+ "learning_rate": 4.959128065395096e-06,
+ "loss": 2.8145,
+ "step": 273
+ },
+ {
+ "epoch": 0.7465940054495913,
+ "grad_norm": 8.347177505493164,
+ "learning_rate": 4.977293369663942e-06,
+ "loss": 2.7651,
+ "step": 274
+ },
+ {
+ "epoch": 0.7493188010899182,
+ "grad_norm": 8.86236572265625,
+ "learning_rate": 4.9954586739327884e-06,
+ "loss": 2.9307,
+ "step": 275
+ },
+ {
+ "epoch": 0.7520435967302452,
+ "grad_norm": 8.273186683654785,
+ "learning_rate": 5.013623978201635e-06,
+ "loss": 2.6265,
+ "step": 276
+ },
+ {
+ "epoch": 0.7547683923705722,
+ "grad_norm": 7.767592430114746,
+ "learning_rate": 5.031789282470482e-06,
+ "loss": 2.417,
+ "step": 277
+ },
+ {
+ "epoch": 0.7574931880108992,
+ "grad_norm": 9.021896362304688,
+ "learning_rate": 5.049954586739329e-06,
+ "loss": 2.7437,
+ "step": 278
+ },
+ {
+ "epoch": 0.7602179836512262,
+ "grad_norm": 9.514054298400879,
+ "learning_rate": 5.068119891008176e-06,
+ "loss": 2.9072,
+ "step": 279
+ },
+ {
+ "epoch": 0.7629427792915532,
+ "grad_norm": 8.357827186584473,
+ "learning_rate": 5.086285195277021e-06,
+ "loss": 2.8574,
+ "step": 280
+ },
+ {
+ "epoch": 0.7656675749318801,
+ "grad_norm": 7.845438003540039,
+ "learning_rate": 5.104450499545868e-06,
+ "loss": 2.6797,
+ "step": 281
+ },
+ {
+ "epoch": 0.7683923705722071,
+ "grad_norm": 7.989553928375244,
+ "learning_rate": 5.122615803814714e-06,
+ "loss": 2.7788,
+ "step": 282
+ },
+ {
+ "epoch": 0.771117166212534,
+ "grad_norm": 10.936755180358887,
+ "learning_rate": 5.140781108083561e-06,
+ "loss": 2.9204,
+ "step": 283
+ },
+ {
+ "epoch": 0.773841961852861,
+ "grad_norm": 7.221974849700928,
+ "learning_rate": 5.158946412352407e-06,
+ "loss": 2.8789,
+ "step": 284
+ },
+ {
+ "epoch": 0.776566757493188,
+ "grad_norm": 7.222783088684082,
+ "learning_rate": 5.1771117166212534e-06,
+ "loss": 2.6826,
+ "step": 285
+ },
+ {
+ "epoch": 0.779291553133515,
+ "grad_norm": 7.458542346954346,
+ "learning_rate": 5.1952770208901005e-06,
+ "loss": 2.8057,
+ "step": 286
+ },
+ {
+ "epoch": 0.782016348773842,
+ "grad_norm": 10.094307899475098,
+ "learning_rate": 5.2134423251589475e-06,
+ "loss": 2.9951,
+ "step": 287
+ },
+ {
+ "epoch": 0.784741144414169,
+ "grad_norm": 9.34561824798584,
+ "learning_rate": 5.231607629427793e-06,
+ "loss": 2.75,
+ "step": 288
+ },
+ {
+ "epoch": 0.7874659400544959,
+ "grad_norm": 9.144036293029785,
+ "learning_rate": 5.24977293369664e-06,
+ "loss": 2.8091,
+ "step": 289
+ },
+ {
+ "epoch": 0.7901907356948229,
+ "grad_norm": 9.393390655517578,
+ "learning_rate": 5.267938237965487e-06,
+ "loss": 2.8164,
+ "step": 290
+ },
+ {
+ "epoch": 0.7929155313351499,
+ "grad_norm": 7.645183086395264,
+ "learning_rate": 5.286103542234333e-06,
+ "loss": 2.9092,
+ "step": 291
+ },
+ {
+ "epoch": 0.7956403269754768,
+ "grad_norm": 9.455459594726562,
+ "learning_rate": 5.304268846503179e-06,
+ "loss": 2.9609,
+ "step": 292
+ },
+ {
+ "epoch": 0.7983651226158038,
+ "grad_norm": 8.59483814239502,
+ "learning_rate": 5.322434150772025e-06,
+ "loss": 2.9463,
+ "step": 293
+ },
+ {
+ "epoch": 0.8010899182561307,
+ "grad_norm": 10.400053024291992,
+ "learning_rate": 5.340599455040872e-06,
+ "loss": 2.8667,
+ "step": 294
+ },
+ {
+ "epoch": 0.8038147138964578,
+ "grad_norm": 24.318883895874023,
+ "learning_rate": 5.358764759309719e-06,
+ "loss": 2.5986,
+ "step": 295
+ },
+ {
+ "epoch": 0.8065395095367848,
+ "grad_norm": 9.522290229797363,
+ "learning_rate": 5.376930063578565e-06,
+ "loss": 2.6943,
+ "step": 296
+ },
+ {
+ "epoch": 0.8092643051771117,
+ "grad_norm": 9.051152229309082,
+ "learning_rate": 5.395095367847412e-06,
+ "loss": 2.5518,
+ "step": 297
+ },
+ {
+ "epoch": 0.8119891008174387,
+ "grad_norm": 8.724372863769531,
+ "learning_rate": 5.413260672116259e-06,
+ "loss": 2.7051,
+ "step": 298
+ },
+ {
+ "epoch": 0.8147138964577657,
+ "grad_norm": 10.865671157836914,
+ "learning_rate": 5.431425976385105e-06,
+ "loss": 2.6768,
+ "step": 299
+ },
+ {
+ "epoch": 0.8174386920980926,
+ "grad_norm": 10.312459945678711,
+ "learning_rate": 5.449591280653952e-06,
+ "loss": 3.0498,
+ "step": 300
+ },
+ {
+ "epoch": 0.8201634877384196,
+ "grad_norm": 8.938870429992676,
+ "learning_rate": 5.467756584922798e-06,
+ "loss": 2.8027,
+ "step": 301
+ },
+ {
+ "epoch": 0.8228882833787466,
+ "grad_norm": 8.697027206420898,
+ "learning_rate": 5.485921889191644e-06,
+ "loss": 2.9688,
+ "step": 302
+ },
+ {
+ "epoch": 0.8256130790190735,
+ "grad_norm": 8.935403823852539,
+ "learning_rate": 5.504087193460491e-06,
+ "loss": 2.812,
+ "step": 303
+ },
+ {
+ "epoch": 0.8283378746594006,
+ "grad_norm": 8.842799186706543,
+ "learning_rate": 5.522252497729338e-06,
+ "loss": 2.6313,
+ "step": 304
+ },
+ {
+ "epoch": 0.8310626702997275,
+ "grad_norm": 8.981131553649902,
+ "learning_rate": 5.5404178019981835e-06,
+ "loss": 2.6938,
+ "step": 305
+ },
+ {
+ "epoch": 0.8337874659400545,
+ "grad_norm": 24.608022689819336,
+ "learning_rate": 5.5585831062670305e-06,
+ "loss": 2.96,
+ "step": 306
+ },
+ {
+ "epoch": 0.8365122615803815,
+ "grad_norm": 9.207291603088379,
+ "learning_rate": 5.5767484105358775e-06,
+ "loss": 3.0874,
+ "step": 307
+ },
+ {
+ "epoch": 0.8392370572207084,
+ "grad_norm": 10.982826232910156,
+ "learning_rate": 5.594913714804724e-06,
+ "loss": 2.8076,
+ "step": 308
+ },
+ {
+ "epoch": 0.8419618528610354,
+ "grad_norm": 8.468833923339844,
+ "learning_rate": 5.61307901907357e-06,
+ "loss": 2.7256,
+ "step": 309
+ },
+ {
+ "epoch": 0.8446866485013624,
+ "grad_norm": 10.0444917678833,
+ "learning_rate": 5.631244323342416e-06,
+ "loss": 3.085,
+ "step": 310
+ },
+ {
+ "epoch": 0.8474114441416893,
+ "grad_norm": 10.920013427734375,
+ "learning_rate": 5.649409627611263e-06,
+ "loss": 3.0684,
+ "step": 311
+ },
+ {
+ "epoch": 0.8501362397820164,
+ "grad_norm": 8.665658950805664,
+ "learning_rate": 5.66757493188011e-06,
+ "loss": 2.9375,
+ "step": 312
+ },
+ {
+ "epoch": 0.8528610354223434,
+ "grad_norm": 7.940450668334961,
+ "learning_rate": 5.685740236148955e-06,
+ "loss": 2.6875,
+ "step": 313
+ },
+ {
+ "epoch": 0.8555858310626703,
+ "grad_norm": 8.870205879211426,
+ "learning_rate": 5.703905540417802e-06,
+ "loss": 2.8457,
+ "step": 314
+ },
+ {
+ "epoch": 0.8583106267029973,
+ "grad_norm": 10.253340721130371,
+ "learning_rate": 5.722070844686649e-06,
+ "loss": 2.5732,
+ "step": 315
+ },
+ {
+ "epoch": 0.8610354223433242,
+ "grad_norm": 8.680900573730469,
+ "learning_rate": 5.7402361489554955e-06,
+ "loss": 2.7539,
+ "step": 316
+ },
+ {
+ "epoch": 0.8637602179836512,
+ "grad_norm": 10.851576805114746,
+ "learning_rate": 5.758401453224342e-06,
+ "loss": 3.0244,
+ "step": 317
+ },
+ {
+ "epoch": 0.8664850136239782,
+ "grad_norm": 9.967367172241211,
+ "learning_rate": 5.776566757493189e-06,
+ "loss": 2.7314,
+ "step": 318
+ },
+ {
+ "epoch": 0.8692098092643051,
+ "grad_norm": 8.775062561035156,
+ "learning_rate": 5.794732061762035e-06,
+ "loss": 3.0479,
+ "step": 319
+ },
+ {
+ "epoch": 0.8719346049046321,
+ "grad_norm": 6.563827037811279,
+ "learning_rate": 5.812897366030882e-06,
+ "loss": 2.7212,
+ "step": 320
+ },
+ {
+ "epoch": 0.8746594005449592,
+ "grad_norm": 9.360514640808105,
+ "learning_rate": 5.831062670299727e-06,
+ "loss": 2.9551,
+ "step": 321
+ },
+ {
+ "epoch": 0.8773841961852861,
+ "grad_norm": 6.7770280838012695,
+ "learning_rate": 5.849227974568574e-06,
+ "loss": 2.7002,
+ "step": 322
+ },
+ {
+ "epoch": 0.8801089918256131,
+ "grad_norm": 7.507646083831787,
+ "learning_rate": 5.867393278837421e-06,
+ "loss": 2.5884,
+ "step": 323
+ },
+ {
+ "epoch": 0.8828337874659401,
+ "grad_norm": 15.701087951660156,
+ "learning_rate": 5.885558583106268e-06,
+ "loss": 2.7158,
+ "step": 324
+ },
+ {
+ "epoch": 0.885558583106267,
+ "grad_norm": 8.826519012451172,
+ "learning_rate": 5.9037238873751135e-06,
+ "loss": 2.7222,
+ "step": 325
+ },
+ {
+ "epoch": 0.888283378746594,
+ "grad_norm": 8.839725494384766,
+ "learning_rate": 5.9218891916439605e-06,
+ "loss": 2.8682,
+ "step": 326
+ },
+ {
+ "epoch": 0.8910081743869209,
+ "grad_norm": 11.04948902130127,
+ "learning_rate": 5.940054495912807e-06,
+ "loss": 2.7378,
+ "step": 327
+ },
+ {
+ "epoch": 0.8937329700272479,
+ "grad_norm": 9.170727729797363,
+ "learning_rate": 5.958219800181654e-06,
+ "loss": 2.582,
+ "step": 328
+ },
+ {
+ "epoch": 0.896457765667575,
+ "grad_norm": 11.354449272155762,
+ "learning_rate": 5.976385104450501e-06,
+ "loss": 2.8125,
+ "step": 329
+ },
+ {
+ "epoch": 0.8991825613079019,
+ "grad_norm": 7.740424633026123,
+ "learning_rate": 5.994550408719346e-06,
+ "loss": 2.6904,
+ "step": 330
+ },
+ {
+ "epoch": 0.9019073569482289,
+ "grad_norm": 9.528890609741211,
+ "learning_rate": 6.012715712988193e-06,
+ "loss": 2.6445,
+ "step": 331
+ },
+ {
+ "epoch": 0.9046321525885559,
+ "grad_norm": 17.546615600585938,
+ "learning_rate": 6.03088101725704e-06,
+ "loss": 2.6362,
+ "step": 332
+ },
+ {
+ "epoch": 0.9073569482288828,
+ "grad_norm": 10.235654830932617,
+ "learning_rate": 6.049046321525886e-06,
+ "loss": 2.8271,
+ "step": 333
+ },
+ {
+ "epoch": 0.9100817438692098,
+ "grad_norm": 8.256182670593262,
+ "learning_rate": 6.067211625794732e-06,
+ "loss": 2.4731,
+ "step": 334
+ },
+ {
+ "epoch": 0.9128065395095368,
+ "grad_norm": 9.551855087280273,
+ "learning_rate": 6.085376930063579e-06,
+ "loss": 2.7817,
+ "step": 335
+ },
+ {
+ "epoch": 0.9155313351498637,
+ "grad_norm": 8.541698455810547,
+ "learning_rate": 6.1035422343324255e-06,
+ "loss": 2.9336,
+ "step": 336
+ },
+ {
+ "epoch": 0.9182561307901907,
+ "grad_norm": 9.929803848266602,
+ "learning_rate": 6.1217075386012725e-06,
+ "loss": 2.8589,
+ "step": 337
+ },
+ {
+ "epoch": 0.9209809264305178,
+ "grad_norm": 11.165647506713867,
+ "learning_rate": 6.139872842870118e-06,
+ "loss": 2.5322,
+ "step": 338
+ },
+ {
+ "epoch": 0.9237057220708447,
+ "grad_norm": 11.198612213134766,
+ "learning_rate": 6.158038147138965e-06,
+ "loss": 2.7822,
+ "step": 339
+ },
+ {
+ "epoch": 0.9264305177111717,
+ "grad_norm": 6.83568811416626,
+ "learning_rate": 6.176203451407812e-06,
+ "loss": 2.5703,
+ "step": 340
+ },
+ {
+ "epoch": 0.9291553133514986,
+ "grad_norm": 9.651077270507812,
+ "learning_rate": 6.194368755676659e-06,
+ "loss": 2.7598,
+ "step": 341
+ },
+ {
+ "epoch": 0.9318801089918256,
+ "grad_norm": 9.206592559814453,
+ "learning_rate": 6.212534059945504e-06,
+ "loss": 2.6309,
+ "step": 342
+ },
+ {
+ "epoch": 0.9346049046321526,
+ "grad_norm": 11.100067138671875,
+ "learning_rate": 6.230699364214351e-06,
+ "loss": 2.8062,
+ "step": 343
+ },
+ {
+ "epoch": 0.9373297002724795,
+ "grad_norm": 8.567194938659668,
+ "learning_rate": 6.248864668483197e-06,
+ "loss": 2.6919,
+ "step": 344
+ },
+ {
+ "epoch": 0.9400544959128065,
+ "grad_norm": 7.0061869621276855,
+ "learning_rate": 6.267029972752044e-06,
+ "loss": 2.6772,
+ "step": 345
+ },
+ {
+ "epoch": 0.9427792915531336,
+ "grad_norm": 9.850142478942871,
+ "learning_rate": 6.2851952770208905e-06,
+ "loss": 2.7637,
+ "step": 346
+ },
+ {
+ "epoch": 0.9455040871934605,
+ "grad_norm": 9.991357803344727,
+ "learning_rate": 6.303360581289737e-06,
+ "loss": 2.5811,
+ "step": 347
+ },
+ {
+ "epoch": 0.9482288828337875,
+ "grad_norm": 8.562417984008789,
+ "learning_rate": 6.321525885558584e-06,
+ "loss": 2.8257,
+ "step": 348
+ },
+ {
+ "epoch": 0.9509536784741145,
+ "grad_norm": 7.703474521636963,
+ "learning_rate": 6.339691189827431e-06,
+ "loss": 2.7578,
+ "step": 349
+ },
+ {
+ "epoch": 0.9536784741144414,
+ "grad_norm": 8.564915657043457,
+ "learning_rate": 6.357856494096276e-06,
+ "loss": 2.7144,
+ "step": 350
+ },
+ {
+ "epoch": 0.9564032697547684,
+ "grad_norm": 11.23455810546875,
+ "learning_rate": 6.376021798365123e-06,
+ "loss": 2.8691,
+ "step": 351
+ },
+ {
+ "epoch": 0.9591280653950953,
+ "grad_norm": 7.569808483123779,
+ "learning_rate": 6.39418710263397e-06,
+ "loss": 2.6821,
+ "step": 352
+ },
+ {
+ "epoch": 0.9618528610354223,
+ "grad_norm": 16.45203971862793,
+ "learning_rate": 6.412352406902816e-06,
+ "loss": 2.8862,
+ "step": 353
+ },
+ {
+ "epoch": 0.9645776566757494,
+ "grad_norm": 7.936426162719727,
+ "learning_rate": 6.430517711171662e-06,
+ "loss": 2.7178,
+ "step": 354
+ },
+ {
+ "epoch": 0.9673024523160763,
+ "grad_norm": 9.06844711303711,
+ "learning_rate": 6.448683015440509e-06,
+ "loss": 2.5205,
+ "step": 355
+ },
+ {
+ "epoch": 0.9700272479564033,
+ "grad_norm": 8.587845802307129,
+ "learning_rate": 6.4668483197093555e-06,
+ "loss": 2.7998,
+ "step": 356
+ },
+ {
+ "epoch": 0.9727520435967303,
+ "grad_norm": 11.425804138183594,
+ "learning_rate": 6.4850136239782025e-06,
+ "loss": 3.0015,
+ "step": 357
+ },
+ {
+ "epoch": 0.9754768392370572,
+ "grad_norm": 10.663166999816895,
+ "learning_rate": 6.5031789282470495e-06,
+ "loss": 2.5513,
+ "step": 358
+ },
+ {
+ "epoch": 0.9782016348773842,
+ "grad_norm": 10.648030281066895,
+ "learning_rate": 6.521344232515895e-06,
+ "loss": 2.6055,
+ "step": 359
+ },
+ {
+ "epoch": 0.9809264305177112,
+ "grad_norm": 9.925614356994629,
+ "learning_rate": 6.539509536784742e-06,
+ "loss": 2.71,
+ "step": 360
+ },
+ {
+ "epoch": 0.9836512261580381,
+ "grad_norm": 10.22243595123291,
+ "learning_rate": 6.557674841053589e-06,
+ "loss": 2.3369,
+ "step": 361
+ },
+ {
+ "epoch": 0.9863760217983651,
+ "grad_norm": 9.037812232971191,
+ "learning_rate": 6.575840145322435e-06,
+ "loss": 2.7031,
+ "step": 362
+ },
+ {
+ "epoch": 0.989100817438692,
+ "grad_norm": 9.280600547790527,
+ "learning_rate": 6.594005449591281e-06,
+ "loss": 2.8506,
+ "step": 363
+ },
+ {
+ "epoch": 0.9918256130790191,
+ "grad_norm": 8.923712730407715,
+ "learning_rate": 6.612170753860127e-06,
+ "loss": 2.7163,
+ "step": 364
+ },
+ {
+ "epoch": 0.9945504087193461,
+ "grad_norm": 12.905025482177734,
+ "learning_rate": 6.630336058128974e-06,
+ "loss": 2.5332,
+ "step": 365
+ },
+ {
+ "epoch": 0.997275204359673,
+ "grad_norm": 8.330647468566895,
+ "learning_rate": 6.648501362397821e-06,
+ "loss": 2.4893,
+ "step": 366
+ },
+ {
+ "epoch": 1.0,
+ "grad_norm": 9.324748992919922,
+ "learning_rate": 6.666666666666667e-06,
+ "loss": 2.769,
+ "step": 367
+ },
+ {
+ "epoch": 1.002724795640327,
+ "grad_norm": 8.635687828063965,
+ "learning_rate": 6.684831970935514e-06,
+ "loss": 2.7192,
+ "step": 368
+ },
+ {
+ "epoch": 1.005449591280654,
+ "grad_norm": 8.160465240478516,
+ "learning_rate": 6.702997275204361e-06,
+ "loss": 2.6948,
+ "step": 369
+ },
+ {
+ "epoch": 1.008174386920981,
+ "grad_norm": 8.604668617248535,
+ "learning_rate": 6.721162579473207e-06,
+ "loss": 2.4541,
+ "step": 370
+ },
+ {
+ "epoch": 1.0108991825613078,
+ "grad_norm": 10.951014518737793,
+ "learning_rate": 6.739327883742053e-06,
+ "loss": 2.6392,
+ "step": 371
+ },
+ {
+ "epoch": 1.013623978201635,
+ "grad_norm": 9.221234321594238,
+ "learning_rate": 6.7574931880109e-06,
+ "loss": 2.7905,
+ "step": 372
+ },
+ {
+ "epoch": 1.0163487738419619,
+ "grad_norm": 9.153060913085938,
+ "learning_rate": 6.775658492279746e-06,
+ "loss": 2.5908,
+ "step": 373
+ },
+ {
+ "epoch": 1.0190735694822888,
+ "grad_norm": 10.996193885803223,
+ "learning_rate": 6.793823796548593e-06,
+ "loss": 2.5732,
+ "step": 374
+ },
+ {
+ "epoch": 1.021798365122616,
+ "grad_norm": 9.299703598022461,
+ "learning_rate": 6.8119891008174385e-06,
+ "loss": 2.6699,
+ "step": 375
+ },
+ {
+ "epoch": 1.0245231607629428,
+ "grad_norm": 11.238974571228027,
+ "learning_rate": 6.8301544050862855e-06,
+ "loss": 2.6392,
+ "step": 376
+ },
+ {
+ "epoch": 1.0272479564032697,
+ "grad_norm": 10.425594329833984,
+ "learning_rate": 6.8483197093551325e-06,
+ "loss": 2.7471,
+ "step": 377
+ },
+ {
+ "epoch": 1.0299727520435966,
+ "grad_norm": 7.944182872772217,
+ "learning_rate": 6.8664850136239795e-06,
+ "loss": 2.7969,
+ "step": 378
+ },
+ {
+ "epoch": 1.0326975476839237,
+ "grad_norm": 8.943190574645996,
+ "learning_rate": 6.884650317892825e-06,
+ "loss": 2.7134,
+ "step": 379
+ },
+ {
+ "epoch": 1.0354223433242506,
+ "grad_norm": 10.596640586853027,
+ "learning_rate": 6.902815622161672e-06,
+ "loss": 2.6519,
+ "step": 380
+ },
+ {
+ "epoch": 1.0381471389645776,
+ "grad_norm": 7.6247782707214355,
+ "learning_rate": 6.920980926430518e-06,
+ "loss": 2.4971,
+ "step": 381
+ },
+ {
+ "epoch": 1.0408719346049047,
+ "grad_norm": 10.39920425415039,
+ "learning_rate": 6.939146230699365e-06,
+ "loss": 2.7231,
+ "step": 382
+ },
+ {
+ "epoch": 1.0435967302452316,
+ "grad_norm": 10.303878784179688,
+ "learning_rate": 6.957311534968211e-06,
+ "loss": 2.5679,
+ "step": 383
+ },
+ {
+ "epoch": 1.0463215258855585,
+ "grad_norm": 8.717978477478027,
+ "learning_rate": 6.975476839237057e-06,
+ "loss": 2.6963,
+ "step": 384
+ },
+ {
+ "epoch": 1.0490463215258856,
+ "grad_norm": 8.62670612335205,
+ "learning_rate": 6.993642143505904e-06,
+ "loss": 2.4697,
+ "step": 385
+ },
+ {
+ "epoch": 1.0517711171662125,
+ "grad_norm": 9.171549797058105,
+ "learning_rate": 7.011807447774751e-06,
+ "loss": 2.582,
+ "step": 386
+ },
+ {
+ "epoch": 1.0544959128065394,
+ "grad_norm": 9.976481437683105,
+ "learning_rate": 7.0299727520435975e-06,
+ "loss": 2.5825,
+ "step": 387
+ },
+ {
+ "epoch": 1.0572207084468666,
+ "grad_norm": 10.577069282531738,
+ "learning_rate": 7.048138056312444e-06,
+ "loss": 2.4307,
+ "step": 388
+ },
+ {
+ "epoch": 1.0599455040871935,
+ "grad_norm": 10.367929458618164,
+ "learning_rate": 7.066303360581291e-06,
+ "loss": 2.4648,
+ "step": 389
+ },
+ {
+ "epoch": 1.0626702997275204,
+ "grad_norm": 9.94495964050293,
+ "learning_rate": 7.084468664850137e-06,
+ "loss": 2.48,
+ "step": 390
+ },
+ {
+ "epoch": 1.0653950953678475,
+ "grad_norm": 11.44611644744873,
+ "learning_rate": 7.102633969118984e-06,
+ "loss": 2.3208,
+ "step": 391
+ },
+ {
+ "epoch": 1.0681198910081744,
+ "grad_norm": 8.668347358703613,
+ "learning_rate": 7.120799273387829e-06,
+ "loss": 2.689,
+ "step": 392
+ },
+ {
+ "epoch": 1.0708446866485013,
+ "grad_norm": 10.10559368133545,
+ "learning_rate": 7.138964577656676e-06,
+ "loss": 2.4907,
+ "step": 393
+ },
+ {
+ "epoch": 1.0735694822888284,
+ "grad_norm": 12.406932830810547,
+ "learning_rate": 7.157129881925523e-06,
+ "loss": 2.6147,
+ "step": 394
+ },
+ {
+ "epoch": 1.0762942779291553,
+ "grad_norm": 8.750371932983398,
+ "learning_rate": 7.17529518619437e-06,
+ "loss": 2.3989,
+ "step": 395
+ },
+ {
+ "epoch": 1.0790190735694822,
+ "grad_norm": 9.494389533996582,
+ "learning_rate": 7.1934604904632155e-06,
+ "loss": 2.6235,
+ "step": 396
+ },
+ {
+ "epoch": 1.0817438692098094,
+ "grad_norm": 8.633389472961426,
+ "learning_rate": 7.2116257947320625e-06,
+ "loss": 2.501,
+ "step": 397
+ },
+ {
+ "epoch": 1.0844686648501363,
+ "grad_norm": 9.764907836914062,
+ "learning_rate": 7.229791099000909e-06,
+ "loss": 2.3696,
+ "step": 398
+ },
+ {
+ "epoch": 1.0871934604904632,
+ "grad_norm": 7.311062812805176,
+ "learning_rate": 7.247956403269756e-06,
+ "loss": 2.187,
+ "step": 399
+ },
+ {
+ "epoch": 1.0899182561307903,
+ "grad_norm": 12.52861213684082,
+ "learning_rate": 7.266121707538602e-06,
+ "loss": 2.5962,
+ "step": 400
+ },
+ {
+ "epoch": 1.0926430517711172,
+ "grad_norm": 18.11480140686035,
+ "learning_rate": 7.284287011807448e-06,
+ "loss": 2.564,
+ "step": 401
+ },
+ {
+ "epoch": 1.095367847411444,
+ "grad_norm": 12.477897644042969,
+ "learning_rate": 7.302452316076295e-06,
+ "loss": 2.3735,
+ "step": 402
+ },
+ {
+ "epoch": 1.0980926430517712,
+ "grad_norm": 35.82415771484375,
+ "learning_rate": 7.320617620345142e-06,
+ "loss": 2.606,
+ "step": 403
+ },
+ {
+ "epoch": 1.1008174386920981,
+ "grad_norm": 8.72639274597168,
+ "learning_rate": 7.338782924613987e-06,
+ "loss": 2.7837,
+ "step": 404
+ },
+ {
+ "epoch": 1.103542234332425,
+ "grad_norm": 9.957283973693848,
+ "learning_rate": 7.356948228882834e-06,
+ "loss": 2.6846,
+ "step": 405
+ },
+ {
+ "epoch": 1.106267029972752,
+ "grad_norm": 9.602156639099121,
+ "learning_rate": 7.375113533151681e-06,
+ "loss": 2.3789,
+ "step": 406
+ },
+ {
+ "epoch": 1.108991825613079,
+ "grad_norm": 8.083453178405762,
+ "learning_rate": 7.3932788374205275e-06,
+ "loss": 2.4199,
+ "step": 407
+ },
+ {
+ "epoch": 1.111716621253406,
+ "grad_norm": 9.620992660522461,
+ "learning_rate": 7.411444141689374e-06,
+ "loss": 2.397,
+ "step": 408
+ },
+ {
+ "epoch": 1.1144414168937329,
+ "grad_norm": 8.023629188537598,
+ "learning_rate": 7.42960944595822e-06,
+ "loss": 2.3564,
+ "step": 409
+ },
+ {
+ "epoch": 1.11716621253406,
+ "grad_norm": 6.683465003967285,
+ "learning_rate": 7.447774750227067e-06,
+ "loss": 2.2563,
+ "step": 410
+ },
+ {
+ "epoch": 1.119891008174387,
+ "grad_norm": 9.92617416381836,
+ "learning_rate": 7.465940054495914e-06,
+ "loss": 2.6377,
+ "step": 411
+ },
+ {
+ "epoch": 1.1226158038147138,
+ "grad_norm": 9.70993709564209,
+ "learning_rate": 7.484105358764759e-06,
+ "loss": 2.6538,
+ "step": 412
+ },
+ {
+ "epoch": 1.125340599455041,
+ "grad_norm": 11.575554847717285,
+ "learning_rate": 7.502270663033606e-06,
+ "loss": 2.3135,
+ "step": 413
+ },
+ {
+ "epoch": 1.1280653950953679,
+ "grad_norm": 8.525997161865234,
+ "learning_rate": 7.520435967302453e-06,
+ "loss": 2.5376,
+ "step": 414
+ },
+ {
+ "epoch": 1.1307901907356948,
+ "grad_norm": 7.669206619262695,
+ "learning_rate": 7.538601271571299e-06,
+ "loss": 2.2119,
+ "step": 415
+ },
+ {
+ "epoch": 1.1335149863760219,
+ "grad_norm": 10.054433822631836,
+ "learning_rate": 7.5567665758401455e-06,
+ "loss": 2.3853,
+ "step": 416
+ },
+ {
+ "epoch": 1.1362397820163488,
+ "grad_norm": 10.294475555419922,
+ "learning_rate": 7.5749318801089925e-06,
+ "loss": 2.7271,
+ "step": 417
+ },
+ {
+ "epoch": 1.1389645776566757,
+ "grad_norm": 9.375436782836914,
+ "learning_rate": 7.593097184377839e-06,
+ "loss": 2.5381,
+ "step": 418
+ },
+ {
+ "epoch": 1.1416893732970028,
+ "grad_norm": 10.102754592895508,
+ "learning_rate": 7.611262488646686e-06,
+ "loss": 2.5752,
+ "step": 419
+ },
+ {
+ "epoch": 1.1444141689373297,
+ "grad_norm": 12.486855506896973,
+ "learning_rate": 7.629427792915532e-06,
+ "loss": 2.5874,
+ "step": 420
+ },
+ {
+ "epoch": 1.1471389645776566,
+ "grad_norm": 10.254337310791016,
+ "learning_rate": 7.647593097184378e-06,
+ "loss": 2.1899,
+ "step": 421
+ },
+ {
+ "epoch": 1.1498637602179835,
+ "grad_norm": 8.257769584655762,
+ "learning_rate": 7.665758401453224e-06,
+ "loss": 2.2534,
+ "step": 422
+ },
+ {
+ "epoch": 1.1525885558583107,
+ "grad_norm": 11.025477409362793,
+ "learning_rate": 7.683923705722072e-06,
+ "loss": 2.4028,
+ "step": 423
+ },
+ {
+ "epoch": 1.1553133514986376,
+ "grad_norm": 9.719670295715332,
+ "learning_rate": 7.702089009990918e-06,
+ "loss": 2.4272,
+ "step": 424
+ },
+ {
+ "epoch": 1.1580381471389645,
+ "grad_norm": 7.976486682891846,
+ "learning_rate": 7.720254314259764e-06,
+ "loss": 2.3418,
+ "step": 425
+ },
+ {
+ "epoch": 1.1607629427792916,
+ "grad_norm": 7.968567848205566,
+ "learning_rate": 7.73841961852861e-06,
+ "loss": 2.4326,
+ "step": 426
+ },
+ {
+ "epoch": 1.1634877384196185,
+ "grad_norm": 11.125061988830566,
+ "learning_rate": 7.756584922797458e-06,
+ "loss": 2.6636,
+ "step": 427
+ },
+ {
+ "epoch": 1.1662125340599454,
+ "grad_norm": 10.837315559387207,
+ "learning_rate": 7.774750227066305e-06,
+ "loss": 2.6221,
+ "step": 428
+ },
+ {
+ "epoch": 1.1689373297002725,
+ "grad_norm": 10.937686920166016,
+ "learning_rate": 7.79291553133515e-06,
+ "loss": 2.4414,
+ "step": 429
+ },
+ {
+ "epoch": 1.1716621253405994,
+ "grad_norm": 12.558049201965332,
+ "learning_rate": 7.811080835603997e-06,
+ "loss": 2.3853,
+ "step": 430
+ },
+ {
+ "epoch": 1.1743869209809263,
+ "grad_norm": 11.79797077178955,
+ "learning_rate": 7.829246139872843e-06,
+ "loss": 2.9502,
+ "step": 431
+ },
+ {
+ "epoch": 1.1771117166212535,
+ "grad_norm": 8.001665115356445,
+ "learning_rate": 7.847411444141691e-06,
+ "loss": 2.4966,
+ "step": 432
+ },
+ {
+ "epoch": 1.1798365122615804,
+ "grad_norm": 7.563295841217041,
+ "learning_rate": 7.865576748410535e-06,
+ "loss": 2.1045,
+ "step": 433
+ },
+ {
+ "epoch": 1.1825613079019073,
+ "grad_norm": 9.281866073608398,
+ "learning_rate": 7.883742052679383e-06,
+ "loss": 2.3369,
+ "step": 434
+ },
+ {
+ "epoch": 1.1852861035422344,
+ "grad_norm": 7.775791168212891,
+ "learning_rate": 7.90190735694823e-06,
+ "loss": 2.3755,
+ "step": 435
+ },
+ {
+ "epoch": 1.1880108991825613,
+ "grad_norm": 7.233153820037842,
+ "learning_rate": 7.920072661217077e-06,
+ "loss": 2.4375,
+ "step": 436
+ },
+ {
+ "epoch": 1.1907356948228882,
+ "grad_norm": 8.859134674072266,
+ "learning_rate": 7.938237965485922e-06,
+ "loss": 2.4453,
+ "step": 437
+ },
+ {
+ "epoch": 1.1934604904632153,
+ "grad_norm": 8.654932975769043,
+ "learning_rate": 7.95640326975477e-06,
+ "loss": 2.4043,
+ "step": 438
+ },
+ {
+ "epoch": 1.1961852861035422,
+ "grad_norm": 11.30862045288086,
+ "learning_rate": 7.974568574023616e-06,
+ "loss": 2.4409,
+ "step": 439
+ },
+ {
+ "epoch": 1.1989100817438691,
+ "grad_norm": 10.622076034545898,
+ "learning_rate": 7.992733878292462e-06,
+ "loss": 2.3765,
+ "step": 440
+ },
+ {
+ "epoch": 1.2016348773841963,
+ "grad_norm": 7.9575395584106445,
+ "learning_rate": 8.010899182561308e-06,
+ "loss": 2.4302,
+ "step": 441
+ },
+ {
+ "epoch": 1.2043596730245232,
+ "grad_norm": 10.018165588378906,
+ "learning_rate": 8.029064486830154e-06,
+ "loss": 2.3691,
+ "step": 442
+ },
+ {
+ "epoch": 1.20708446866485,
+ "grad_norm": 9.929265022277832,
+ "learning_rate": 8.047229791099002e-06,
+ "loss": 2.8418,
+ "step": 443
+ },
+ {
+ "epoch": 1.2098092643051772,
+ "grad_norm": 9.285006523132324,
+ "learning_rate": 8.065395095367848e-06,
+ "loss": 2.4893,
+ "step": 444
+ },
+ {
+ "epoch": 1.2125340599455041,
+ "grad_norm": 10.58349323272705,
+ "learning_rate": 8.083560399636694e-06,
+ "loss": 2.5391,
+ "step": 445
+ },
+ {
+ "epoch": 1.215258855585831,
+ "grad_norm": 11.750399589538574,
+ "learning_rate": 8.10172570390554e-06,
+ "loss": 2.4263,
+ "step": 446
+ },
+ {
+ "epoch": 1.2179836512261581,
+ "grad_norm": 8.646162986755371,
+ "learning_rate": 8.119891008174388e-06,
+ "loss": 2.2773,
+ "step": 447
+ },
+ {
+ "epoch": 1.220708446866485,
+ "grad_norm": 9.85015869140625,
+ "learning_rate": 8.138056312443235e-06,
+ "loss": 2.439,
+ "step": 448
+ },
+ {
+ "epoch": 1.223433242506812,
+ "grad_norm": 7.7857184410095215,
+ "learning_rate": 8.15622161671208e-06,
+ "loss": 2.4531,
+ "step": 449
+ },
+ {
+ "epoch": 1.226158038147139,
+ "grad_norm": 16.41480255126953,
+ "learning_rate": 8.174386920980927e-06,
+ "loss": 2.4424,
+ "step": 450
+ },
+ {
+ "epoch": 1.228882833787466,
+ "grad_norm": 8.749961853027344,
+ "learning_rate": 8.192552225249773e-06,
+ "loss": 2.3906,
+ "step": 451
+ },
+ {
+ "epoch": 1.231607629427793,
+ "grad_norm": 12.330133438110352,
+ "learning_rate": 8.210717529518621e-06,
+ "loss": 2.4966,
+ "step": 452
+ },
+ {
+ "epoch": 1.2343324250681198,
+ "grad_norm": 8.449640274047852,
+ "learning_rate": 8.228882833787467e-06,
+ "loss": 2.6147,
+ "step": 453
+ },
+ {
+ "epoch": 1.237057220708447,
+ "grad_norm": 9.252738952636719,
+ "learning_rate": 8.247048138056313e-06,
+ "loss": 2.3921,
+ "step": 454
+ },
+ {
+ "epoch": 1.2397820163487738,
+ "grad_norm": 11.831225395202637,
+ "learning_rate": 8.26521344232516e-06,
+ "loss": 2.3799,
+ "step": 455
+ },
+ {
+ "epoch": 1.2425068119891007,
+ "grad_norm": 10.857969284057617,
+ "learning_rate": 8.283378746594006e-06,
+ "loss": 2.7793,
+ "step": 456
+ },
+ {
+ "epoch": 1.2452316076294279,
+ "grad_norm": 9.51559066772461,
+ "learning_rate": 8.301544050862853e-06,
+ "loss": 2.2891,
+ "step": 457
+ },
+ {
+ "epoch": 1.2479564032697548,
+ "grad_norm": 8.42534351348877,
+ "learning_rate": 8.3197093551317e-06,
+ "loss": 2.2935,
+ "step": 458
+ },
+ {
+ "epoch": 1.2506811989100817,
+ "grad_norm": 10.774682998657227,
+ "learning_rate": 8.337874659400546e-06,
+ "loss": 2.4561,
+ "step": 459
+ },
+ {
+ "epoch": 1.2534059945504088,
+ "grad_norm": 10.927517890930176,
+ "learning_rate": 8.356039963669392e-06,
+ "loss": 2.145,
+ "step": 460
+ },
+ {
+ "epoch": 1.2561307901907357,
+ "grad_norm": 14.117505073547363,
+ "learning_rate": 8.37420526793824e-06,
+ "loss": 2.5869,
+ "step": 461
+ },
+ {
+ "epoch": 1.2588555858310626,
+ "grad_norm": 10.275975227355957,
+ "learning_rate": 8.392370572207084e-06,
+ "loss": 2.5449,
+ "step": 462
+ },
+ {
+ "epoch": 1.2615803814713895,
+ "grad_norm": 16.405057907104492,
+ "learning_rate": 8.410535876475932e-06,
+ "loss": 2.4468,
+ "step": 463
+ },
+ {
+ "epoch": 1.2643051771117166,
+ "grad_norm": 8.686537742614746,
+ "learning_rate": 8.428701180744778e-06,
+ "loss": 2.3931,
+ "step": 464
+ },
+ {
+ "epoch": 1.2670299727520435,
+ "grad_norm": 12.020901679992676,
+ "learning_rate": 8.446866485013624e-06,
+ "loss": 2.3818,
+ "step": 465
+ },
+ {
+ "epoch": 1.2697547683923704,
+ "grad_norm": 7.913347244262695,
+ "learning_rate": 8.46503178928247e-06,
+ "loss": 2.3379,
+ "step": 466
+ },
+ {
+ "epoch": 1.2724795640326976,
+ "grad_norm": 9.544164657592773,
+ "learning_rate": 8.483197093551317e-06,
+ "loss": 2.4434,
+ "step": 467
+ },
+ {
+ "epoch": 1.2752043596730245,
+ "grad_norm": 10.495853424072266,
+ "learning_rate": 8.501362397820165e-06,
+ "loss": 2.3521,
+ "step": 468
+ },
+ {
+ "epoch": 1.2779291553133514,
+ "grad_norm": 12.450571060180664,
+ "learning_rate": 8.51952770208901e-06,
+ "loss": 2.4907,
+ "step": 469
+ },
+ {
+ "epoch": 1.2806539509536785,
+ "grad_norm": 9.063766479492188,
+ "learning_rate": 8.537693006357857e-06,
+ "loss": 2.2095,
+ "step": 470
+ },
+ {
+ "epoch": 1.2833787465940054,
+ "grad_norm": 9.945204734802246,
+ "learning_rate": 8.555858310626703e-06,
+ "loss": 2.519,
+ "step": 471
+ },
+ {
+ "epoch": 1.2861035422343323,
+ "grad_norm": 9.296440124511719,
+ "learning_rate": 8.574023614895551e-06,
+ "loss": 2.4395,
+ "step": 472
+ },
+ {
+ "epoch": 1.2888283378746594,
+ "grad_norm": 9.701530456542969,
+ "learning_rate": 8.592188919164397e-06,
+ "loss": 2.3594,
+ "step": 473
+ },
+ {
+ "epoch": 1.2915531335149864,
+ "grad_norm": 9.079253196716309,
+ "learning_rate": 8.610354223433243e-06,
+ "loss": 2.1807,
+ "step": 474
+ },
+ {
+ "epoch": 1.2942779291553133,
+ "grad_norm": 13.774413108825684,
+ "learning_rate": 8.62851952770209e-06,
+ "loss": 2.6201,
+ "step": 475
+ },
+ {
+ "epoch": 1.2970027247956404,
+ "grad_norm": 12.405047416687012,
+ "learning_rate": 8.646684831970936e-06,
+ "loss": 2.3179,
+ "step": 476
+ },
+ {
+ "epoch": 1.2997275204359673,
+ "grad_norm": 11.16991138458252,
+ "learning_rate": 8.664850136239783e-06,
+ "loss": 2.2148,
+ "step": 477
+ },
+ {
+ "epoch": 1.3024523160762942,
+ "grad_norm": 18.193309783935547,
+ "learning_rate": 8.68301544050863e-06,
+ "loss": 2.3101,
+ "step": 478
+ },
+ {
+ "epoch": 1.3051771117166213,
+ "grad_norm": 12.198078155517578,
+ "learning_rate": 8.701180744777476e-06,
+ "loss": 2.4712,
+ "step": 479
+ },
+ {
+ "epoch": 1.3079019073569482,
+ "grad_norm": 12.856112480163574,
+ "learning_rate": 8.719346049046322e-06,
+ "loss": 2.3184,
+ "step": 480
+ },
+ {
+ "epoch": 1.3106267029972751,
+ "grad_norm": 11.01259994506836,
+ "learning_rate": 8.73751135331517e-06,
+ "loss": 2.3794,
+ "step": 481
+ },
+ {
+ "epoch": 1.3133514986376023,
+ "grad_norm": 16.54364776611328,
+ "learning_rate": 8.755676657584016e-06,
+ "loss": 2.4897,
+ "step": 482
+ },
+ {
+ "epoch": 1.3160762942779292,
+ "grad_norm": 13.459123611450195,
+ "learning_rate": 8.773841961852862e-06,
+ "loss": 2.502,
+ "step": 483
+ },
+ {
+ "epoch": 1.318801089918256,
+ "grad_norm": 12.226052284240723,
+ "learning_rate": 8.792007266121708e-06,
+ "loss": 2.2661,
+ "step": 484
+ },
+ {
+ "epoch": 1.3215258855585832,
+ "grad_norm": 18.36394691467285,
+ "learning_rate": 8.810172570390554e-06,
+ "loss": 2.3315,
+ "step": 485
+ },
+ {
+ "epoch": 1.32425068119891,
+ "grad_norm": 9.599638938903809,
+ "learning_rate": 8.828337874659402e-06,
+ "loss": 2.5405,
+ "step": 486
+ },
+ {
+ "epoch": 1.326975476839237,
+ "grad_norm": 10.80005931854248,
+ "learning_rate": 8.846503178928247e-06,
+ "loss": 2.2783,
+ "step": 487
+ },
+ {
+ "epoch": 1.3297002724795641,
+ "grad_norm": 11.59123706817627,
+ "learning_rate": 8.864668483197095e-06,
+ "loss": 2.2837,
+ "step": 488
+ },
+ {
+ "epoch": 1.332425068119891,
+ "grad_norm": 9.895734786987305,
+ "learning_rate": 8.88283378746594e-06,
+ "loss": 2.1348,
+ "step": 489
+ },
+ {
+ "epoch": 1.335149863760218,
+ "grad_norm": 11.754762649536133,
+ "learning_rate": 8.900999091734787e-06,
+ "loss": 2.2568,
+ "step": 490
+ },
+ {
+ "epoch": 1.337874659400545,
+ "grad_norm": 10.934577941894531,
+ "learning_rate": 8.919164396003633e-06,
+ "loss": 2.458,
+ "step": 491
+ },
+ {
+ "epoch": 1.340599455040872,
+ "grad_norm": 14.818517684936523,
+ "learning_rate": 8.937329700272481e-06,
+ "loss": 2.5557,
+ "step": 492
+ },
+ {
+ "epoch": 1.3433242506811989,
+ "grad_norm": 13.137955665588379,
+ "learning_rate": 8.955495004541327e-06,
+ "loss": 2.4614,
+ "step": 493
+ },
+ {
+ "epoch": 1.346049046321526,
+ "grad_norm": 21.432985305786133,
+ "learning_rate": 8.973660308810173e-06,
+ "loss": 2.4131,
+ "step": 494
+ },
+ {
+ "epoch": 1.348773841961853,
+ "grad_norm": 11.346712112426758,
+ "learning_rate": 8.99182561307902e-06,
+ "loss": 2.1982,
+ "step": 495
+ },
+ {
+ "epoch": 1.3514986376021798,
+ "grad_norm": 10.7266206741333,
+ "learning_rate": 9.009990917347866e-06,
+ "loss": 2.2046,
+ "step": 496
+ },
+ {
+ "epoch": 1.354223433242507,
+ "grad_norm": 10.556185722351074,
+ "learning_rate": 9.028156221616713e-06,
+ "loss": 2.0132,
+ "step": 497
+ },
+ {
+ "epoch": 1.3569482288828338,
+ "grad_norm": 9.045343399047852,
+ "learning_rate": 9.04632152588556e-06,
+ "loss": 2.4561,
+ "step": 498
+ },
+ {
+ "epoch": 1.3596730245231607,
+ "grad_norm": 11.195216178894043,
+ "learning_rate": 9.064486830154406e-06,
+ "loss": 2.5508,
+ "step": 499
+ },
+ {
+ "epoch": 1.3623978201634879,
+ "grad_norm": 14.421299934387207,
+ "learning_rate": 9.082652134423252e-06,
+ "loss": 2.5752,
+ "step": 500
+ },
+ {
+ "epoch": 1.3651226158038148,
+ "grad_norm": 13.239008903503418,
+ "learning_rate": 9.1008174386921e-06,
+ "loss": 2.2876,
+ "step": 501
+ },
+ {
+ "epoch": 1.3678474114441417,
+ "grad_norm": 11.14577579498291,
+ "learning_rate": 9.118982742960946e-06,
+ "loss": 2.7393,
+ "step": 502
+ },
+ {
+ "epoch": 1.3705722070844686,
+ "grad_norm": 10.346534729003906,
+ "learning_rate": 9.137148047229792e-06,
+ "loss": 2.5107,
+ "step": 503
+ },
+ {
+ "epoch": 1.3732970027247957,
+ "grad_norm": 12.070633888244629,
+ "learning_rate": 9.155313351498638e-06,
+ "loss": 2.395,
+ "step": 504
+ },
+ {
+ "epoch": 1.3760217983651226,
+ "grad_norm": 10.990833282470703,
+ "learning_rate": 9.173478655767484e-06,
+ "loss": 2.2461,
+ "step": 505
+ },
+ {
+ "epoch": 1.3787465940054495,
+ "grad_norm": 51.80665969848633,
+ "learning_rate": 9.191643960036332e-06,
+ "loss": 2.2378,
+ "step": 506
+ },
+ {
+ "epoch": 1.3814713896457766,
+ "grad_norm": 15.648660659790039,
+ "learning_rate": 9.209809264305178e-06,
+ "loss": 2.2183,
+ "step": 507
+ },
+ {
+ "epoch": 1.3841961852861036,
+ "grad_norm": 22.547460556030273,
+ "learning_rate": 9.227974568574025e-06,
+ "loss": 2.439,
+ "step": 508
+ },
+ {
+ "epoch": 1.3869209809264305,
+ "grad_norm": 13.129714965820312,
+ "learning_rate": 9.24613987284287e-06,
+ "loss": 2.377,
+ "step": 509
+ },
+ {
+ "epoch": 1.3896457765667574,
+ "grad_norm": 35.93815231323242,
+ "learning_rate": 9.264305177111717e-06,
+ "loss": 2.1997,
+ "step": 510
+ },
+ {
+ "epoch": 1.3923705722070845,
+ "grad_norm": 11.66264533996582,
+ "learning_rate": 9.282470481380565e-06,
+ "loss": 2.4775,
+ "step": 511
+ },
+ {
+ "epoch": 1.3950953678474114,
+ "grad_norm": 9.23564338684082,
+ "learning_rate": 9.300635785649411e-06,
+ "loss": 2.2349,
+ "step": 512
+ },
+ {
+ "epoch": 1.3978201634877383,
+ "grad_norm": 13.097249984741211,
+ "learning_rate": 9.318801089918257e-06,
+ "loss": 2.5811,
+ "step": 513
+ },
+ {
+ "epoch": 1.4005449591280654,
+ "grad_norm": 9.676923751831055,
+ "learning_rate": 9.336966394187103e-06,
+ "loss": 2.3486,
+ "step": 514
+ },
+ {
+ "epoch": 1.4032697547683923,
+ "grad_norm": 12.437355995178223,
+ "learning_rate": 9.355131698455951e-06,
+ "loss": 2.3462,
+ "step": 515
+ },
+ {
+ "epoch": 1.4059945504087192,
+ "grad_norm": 10.053210258483887,
+ "learning_rate": 9.373297002724796e-06,
+ "loss": 2.1992,
+ "step": 516
+ },
+ {
+ "epoch": 1.4087193460490464,
+ "grad_norm": 17.33440589904785,
+ "learning_rate": 9.391462306993643e-06,
+ "loss": 2.75,
+ "step": 517
+ },
+ {
+ "epoch": 1.4114441416893733,
+ "grad_norm": 11.860308647155762,
+ "learning_rate": 9.40962761126249e-06,
+ "loss": 2.417,
+ "step": 518
+ },
+ {
+ "epoch": 1.4141689373297002,
+ "grad_norm": 13.52396011352539,
+ "learning_rate": 9.427792915531336e-06,
+ "loss": 2.395,
+ "step": 519
+ },
+ {
+ "epoch": 1.4168937329700273,
+ "grad_norm": 14.017454147338867,
+ "learning_rate": 9.445958219800182e-06,
+ "loss": 2.3774,
+ "step": 520
+ },
+ {
+ "epoch": 1.4196185286103542,
+ "grad_norm": 11.888250350952148,
+ "learning_rate": 9.464123524069028e-06,
+ "loss": 2.1763,
+ "step": 521
+ },
+ {
+ "epoch": 1.422343324250681,
+ "grad_norm": 13.097738265991211,
+ "learning_rate": 9.482288828337876e-06,
+ "loss": 2.624,
+ "step": 522
+ },
+ {
+ "epoch": 1.4250681198910082,
+ "grad_norm": 16.048067092895508,
+ "learning_rate": 9.500454132606722e-06,
+ "loss": 2.3042,
+ "step": 523
+ },
+ {
+ "epoch": 1.4277929155313351,
+ "grad_norm": 13.387825012207031,
+ "learning_rate": 9.518619436875568e-06,
+ "loss": 2.2876,
+ "step": 524
+ },
+ {
+ "epoch": 1.430517711171662,
+ "grad_norm": 10.035544395446777,
+ "learning_rate": 9.536784741144414e-06,
+ "loss": 2.3608,
+ "step": 525
+ },
+ {
+ "epoch": 1.4332425068119892,
+ "grad_norm": 36.69792556762695,
+ "learning_rate": 9.554950045413262e-06,
+ "loss": 2.2505,
+ "step": 526
+ },
+ {
+ "epoch": 1.435967302452316,
+ "grad_norm": 9.41537094116211,
+ "learning_rate": 9.573115349682108e-06,
+ "loss": 2.083,
+ "step": 527
+ },
+ {
+ "epoch": 1.438692098092643,
+ "grad_norm": 11.909334182739258,
+ "learning_rate": 9.591280653950955e-06,
+ "loss": 2.3599,
+ "step": 528
+ },
+ {
+ "epoch": 1.44141689373297,
+ "grad_norm": 20.67137908935547,
+ "learning_rate": 9.6094459582198e-06,
+ "loss": 2.6206,
+ "step": 529
+ },
+ {
+ "epoch": 1.444141689373297,
+ "grad_norm": 10.567476272583008,
+ "learning_rate": 9.627611262488647e-06,
+ "loss": 2.355,
+ "step": 530
+ },
+ {
+ "epoch": 1.446866485013624,
+ "grad_norm": 12.235589027404785,
+ "learning_rate": 9.645776566757495e-06,
+ "loss": 2.3032,
+ "step": 531
+ },
+ {
+ "epoch": 1.449591280653951,
+ "grad_norm": 13.119529724121094,
+ "learning_rate": 9.66394187102634e-06,
+ "loss": 2.123,
+ "step": 532
+ },
+ {
+ "epoch": 1.452316076294278,
+ "grad_norm": 14.756513595581055,
+ "learning_rate": 9.682107175295187e-06,
+ "loss": 2.5542,
+ "step": 533
+ },
+ {
+ "epoch": 1.4550408719346049,
+ "grad_norm": 12.973993301391602,
+ "learning_rate": 9.700272479564033e-06,
+ "loss": 2.2686,
+ "step": 534
+ },
+ {
+ "epoch": 1.457765667574932,
+ "grad_norm": 12.999614715576172,
+ "learning_rate": 9.718437783832881e-06,
+ "loss": 2.3071,
+ "step": 535
+ },
+ {
+ "epoch": 1.4604904632152589,
+ "grad_norm": 11.062445640563965,
+ "learning_rate": 9.736603088101727e-06,
+ "loss": 2.2007,
+ "step": 536
+ },
+ {
+ "epoch": 1.4632152588555858,
+ "grad_norm": 10.445008277893066,
+ "learning_rate": 9.754768392370573e-06,
+ "loss": 2.5083,
+ "step": 537
+ },
+ {
+ "epoch": 1.465940054495913,
+ "grad_norm": 14.608230590820312,
+ "learning_rate": 9.77293369663942e-06,
+ "loss": 2.4771,
+ "step": 538
+ },
+ {
+ "epoch": 1.4686648501362398,
+ "grad_norm": 10.847367286682129,
+ "learning_rate": 9.791099000908266e-06,
+ "loss": 2.5156,
+ "step": 539
+ },
+ {
+ "epoch": 1.4713896457765667,
+ "grad_norm": 11.703137397766113,
+ "learning_rate": 9.809264305177114e-06,
+ "loss": 2.2729,
+ "step": 540
+ },
+ {
+ "epoch": 1.4741144414168939,
+ "grad_norm": 10.178947448730469,
+ "learning_rate": 9.827429609445958e-06,
+ "loss": 2.1074,
+ "step": 541
+ },
+ {
+ "epoch": 1.4768392370572208,
+ "grad_norm": 9.362521171569824,
+ "learning_rate": 9.845594913714806e-06,
+ "loss": 2.2017,
+ "step": 542
+ },
+ {
+ "epoch": 1.4795640326975477,
+ "grad_norm": 12.863320350646973,
+ "learning_rate": 9.863760217983652e-06,
+ "loss": 2.0615,
+ "step": 543
+ },
+ {
+ "epoch": 1.4822888283378748,
+ "grad_norm": 7.886809349060059,
+ "learning_rate": 9.881925522252498e-06,
+ "loss": 2.3208,
+ "step": 544
+ },
+ {
+ "epoch": 1.4850136239782017,
+ "grad_norm": 15.057289123535156,
+ "learning_rate": 9.900090826521344e-06,
+ "loss": 2.2334,
+ "step": 545
+ },
+ {
+ "epoch": 1.4877384196185286,
+ "grad_norm": 12.064061164855957,
+ "learning_rate": 9.918256130790192e-06,
+ "loss": 2.313,
+ "step": 546
+ },
+ {
+ "epoch": 1.4904632152588557,
+ "grad_norm": 13.74412727355957,
+ "learning_rate": 9.936421435059038e-06,
+ "loss": 2.4053,
+ "step": 547
+ },
+ {
+ "epoch": 1.4931880108991826,
+ "grad_norm": 14.102124214172363,
+ "learning_rate": 9.954586739327885e-06,
+ "loss": 2.3916,
+ "step": 548
+ },
+ {
+ "epoch": 1.4959128065395095,
+ "grad_norm": 14.244315147399902,
+ "learning_rate": 9.97275204359673e-06,
+ "loss": 2.2231,
+ "step": 549
+ },
+ {
+ "epoch": 1.4986376021798364,
+ "grad_norm": 16.0863037109375,
+ "learning_rate": 9.990917347865577e-06,
+ "loss": 2.1812,
+ "step": 550
+ },
+ {
+ "epoch": 1.5013623978201633,
+ "grad_norm": 11.130419731140137,
+ "learning_rate": 1.0009082652134423e-05,
+ "loss": 2.2417,
+ "step": 551
+ },
+ {
+ "epoch": 1.5040871934604905,
+ "grad_norm": 17.573894500732422,
+ "learning_rate": 1.002724795640327e-05,
+ "loss": 2.3628,
+ "step": 552
+ },
+ {
+ "epoch": 1.5068119891008176,
+ "grad_norm": 11.738608360290527,
+ "learning_rate": 1.0045413260672117e-05,
+ "loss": 2.3623,
+ "step": 553
+ },
+ {
+ "epoch": 1.5095367847411443,
+ "grad_norm": 12.370352745056152,
+ "learning_rate": 1.0063578564940963e-05,
+ "loss": 2.5171,
+ "step": 554
+ },
+ {
+ "epoch": 1.5122615803814714,
+ "grad_norm": 13.910186767578125,
+ "learning_rate": 1.008174386920981e-05,
+ "loss": 2.5283,
+ "step": 555
+ },
+ {
+ "epoch": 1.5149863760217985,
+ "grad_norm": 15.552416801452637,
+ "learning_rate": 1.0099909173478657e-05,
+ "loss": 2.3638,
+ "step": 556
+ },
+ {
+ "epoch": 1.5177111716621252,
+ "grad_norm": 9.657610893249512,
+ "learning_rate": 1.0118074477747503e-05,
+ "loss": 2.3296,
+ "step": 557
+ },
+ {
+ "epoch": 1.5204359673024523,
+ "grad_norm": 12.736481666564941,
+ "learning_rate": 1.0136239782016351e-05,
+ "loss": 2.248,
+ "step": 558
+ },
+ {
+ "epoch": 1.5231607629427792,
+ "grad_norm": 10.977620124816895,
+ "learning_rate": 1.0154405086285196e-05,
+ "loss": 2.2339,
+ "step": 559
+ },
+ {
+ "epoch": 1.5258855585831061,
+ "grad_norm": 11.643376350402832,
+ "learning_rate": 1.0172570390554042e-05,
+ "loss": 2.3652,
+ "step": 560
+ },
+ {
+ "epoch": 1.5286103542234333,
+ "grad_norm": 14.136712074279785,
+ "learning_rate": 1.0190735694822888e-05,
+ "loss": 2.2314,
+ "step": 561
+ },
+ {
+ "epoch": 1.5313351498637602,
+ "grad_norm": 14.599430084228516,
+ "learning_rate": 1.0208900999091736e-05,
+ "loss": 2.2817,
+ "step": 562
+ },
+ {
+ "epoch": 1.534059945504087,
+ "grad_norm": 13.482375144958496,
+ "learning_rate": 1.0227066303360582e-05,
+ "loss": 2.3896,
+ "step": 563
+ },
+ {
+ "epoch": 1.5367847411444142,
+ "grad_norm": 14.57656478881836,
+ "learning_rate": 1.0245231607629428e-05,
+ "loss": 2.481,
+ "step": 564
+ },
+ {
+ "epoch": 1.5395095367847411,
+ "grad_norm": 14.707375526428223,
+ "learning_rate": 1.0263396911898276e-05,
+ "loss": 2.0498,
+ "step": 565
+ },
+ {
+ "epoch": 1.542234332425068,
+ "grad_norm": 14.464433670043945,
+ "learning_rate": 1.0281562216167122e-05,
+ "loss": 2.4604,
+ "step": 566
+ },
+ {
+ "epoch": 1.5449591280653951,
+ "grad_norm": 12.12136173248291,
+ "learning_rate": 1.0299727520435968e-05,
+ "loss": 2.5098,
+ "step": 567
+ },
+ {
+ "epoch": 1.547683923705722,
+ "grad_norm": 11.15420913696289,
+ "learning_rate": 1.0317892824704815e-05,
+ "loss": 2.1392,
+ "step": 568
+ },
+ {
+ "epoch": 1.550408719346049,
+ "grad_norm": 14.904620170593262,
+ "learning_rate": 1.033605812897366e-05,
+ "loss": 2.2271,
+ "step": 569
+ },
+ {
+ "epoch": 1.553133514986376,
+ "grad_norm": 15.235796928405762,
+ "learning_rate": 1.0354223433242507e-05,
+ "loss": 2.3794,
+ "step": 570
+ },
+ {
+ "epoch": 1.555858310626703,
+ "grad_norm": 18.094228744506836,
+ "learning_rate": 1.0372388737511355e-05,
+ "loss": 2.3442,
+ "step": 571
+ },
+ {
+ "epoch": 1.55858310626703,
+ "grad_norm": 14.804851531982422,
+ "learning_rate": 1.0390554041780201e-05,
+ "loss": 2.2095,
+ "step": 572
+ },
+ {
+ "epoch": 1.561307901907357,
+ "grad_norm": 15.990046501159668,
+ "learning_rate": 1.0408719346049047e-05,
+ "loss": 2.3252,
+ "step": 573
+ },
+ {
+ "epoch": 1.564032697547684,
+ "grad_norm": 11.997264862060547,
+ "learning_rate": 1.0426884650317895e-05,
+ "loss": 2.3647,
+ "step": 574
+ },
+ {
+ "epoch": 1.5667574931880108,
+ "grad_norm": 11.681777000427246,
+ "learning_rate": 1.0445049954586741e-05,
+ "loss": 2.228,
+ "step": 575
+ },
+ {
+ "epoch": 1.569482288828338,
+ "grad_norm": 12.185470581054688,
+ "learning_rate": 1.0463215258855586e-05,
+ "loss": 2.4229,
+ "step": 576
+ },
+ {
+ "epoch": 1.5722070844686649,
+ "grad_norm": 14.22901725769043,
+ "learning_rate": 1.0481380563124432e-05,
+ "loss": 2.3843,
+ "step": 577
+ },
+ {
+ "epoch": 1.5749318801089918,
+ "grad_norm": 20.202341079711914,
+ "learning_rate": 1.049954586739328e-05,
+ "loss": 2.2554,
+ "step": 578
+ },
+ {
+ "epoch": 1.577656675749319,
+ "grad_norm": 13.772683143615723,
+ "learning_rate": 1.0517711171662126e-05,
+ "loss": 2.4346,
+ "step": 579
+ },
+ {
+ "epoch": 1.5803814713896458,
+ "grad_norm": 45.03996276855469,
+ "learning_rate": 1.0535876475930974e-05,
+ "loss": 2.4277,
+ "step": 580
+ },
+ {
+ "epoch": 1.5831062670299727,
+ "grad_norm": 12.460752487182617,
+ "learning_rate": 1.055404178019982e-05,
+ "loss": 2.187,
+ "step": 581
+ },
+ {
+ "epoch": 1.5858310626702998,
+ "grad_norm": 10.265328407287598,
+ "learning_rate": 1.0572207084468666e-05,
+ "loss": 2.2729,
+ "step": 582
+ },
+ {
+ "epoch": 1.5885558583106267,
+ "grad_norm": 12.573481559753418,
+ "learning_rate": 1.0590372388737514e-05,
+ "loss": 2.1494,
+ "step": 583
+ },
+ {
+ "epoch": 1.5912806539509536,
+ "grad_norm": 11.093875885009766,
+ "learning_rate": 1.0608537693006358e-05,
+ "loss": 2.0825,
+ "step": 584
+ },
+ {
+ "epoch": 1.5940054495912808,
+ "grad_norm": 12.266392707824707,
+ "learning_rate": 1.0626702997275204e-05,
+ "loss": 2.1938,
+ "step": 585
+ },
+ {
+ "epoch": 1.5967302452316077,
+ "grad_norm": 13.413431167602539,
+ "learning_rate": 1.064486830154405e-05,
+ "loss": 2.4868,
+ "step": 586
+ },
+ {
+ "epoch": 1.5994550408719346,
+ "grad_norm": 18.872541427612305,
+ "learning_rate": 1.0663033605812898e-05,
+ "loss": 2.4077,
+ "step": 587
+ },
+ {
+ "epoch": 1.6021798365122617,
+ "grad_norm": 14.07229232788086,
+ "learning_rate": 1.0681198910081745e-05,
+ "loss": 2.3486,
+ "step": 588
+ },
+ {
+ "epoch": 1.6049046321525886,
+ "grad_norm": 11.084822654724121,
+ "learning_rate": 1.069936421435059e-05,
+ "loss": 2.1528,
+ "step": 589
+ },
+ {
+ "epoch": 1.6076294277929155,
+ "grad_norm": 15.526721000671387,
+ "learning_rate": 1.0717529518619439e-05,
+ "loss": 2.1299,
+ "step": 590
+ },
+ {
+ "epoch": 1.6103542234332426,
+ "grad_norm": 12.699135780334473,
+ "learning_rate": 1.0735694822888285e-05,
+ "loss": 1.9307,
+ "step": 591
+ },
+ {
+ "epoch": 1.6130790190735693,
+ "grad_norm": 14.46304702758789,
+ "learning_rate": 1.075386012715713e-05,
+ "loss": 2.3247,
+ "step": 592
+ },
+ {
+ "epoch": 1.6158038147138964,
+ "grad_norm": 16.0272159576416,
+ "learning_rate": 1.0772025431425977e-05,
+ "loss": 2.2378,
+ "step": 593
+ },
+ {
+ "epoch": 1.6185286103542236,
+ "grad_norm": 16.642398834228516,
+ "learning_rate": 1.0790190735694823e-05,
+ "loss": 2.2158,
+ "step": 594
+ },
+ {
+ "epoch": 1.6212534059945503,
+ "grad_norm": 9.624032020568848,
+ "learning_rate": 1.080835603996367e-05,
+ "loss": 1.9731,
+ "step": 595
+ },
+ {
+ "epoch": 1.6239782016348774,
+ "grad_norm": 20.392581939697266,
+ "learning_rate": 1.0826521344232517e-05,
+ "loss": 2.1934,
+ "step": 596
+ },
+ {
+ "epoch": 1.6267029972752045,
+ "grad_norm": 12.304001808166504,
+ "learning_rate": 1.0844686648501363e-05,
+ "loss": 2.1323,
+ "step": 597
+ },
+ {
+ "epoch": 1.6294277929155312,
+ "grad_norm": 15.955755233764648,
+ "learning_rate": 1.086285195277021e-05,
+ "loss": 2.1558,
+ "step": 598
+ },
+ {
+ "epoch": 1.6321525885558583,
+ "grad_norm": 12.658171653747559,
+ "learning_rate": 1.0881017257039057e-05,
+ "loss": 2.2095,
+ "step": 599
+ },
+ {
+ "epoch": 1.6348773841961854,
+ "grad_norm": 18.214834213256836,
+ "learning_rate": 1.0899182561307904e-05,
+ "loss": 2.479,
+ "step": 600
+ },
+ {
+ "epoch": 1.6376021798365121,
+ "grad_norm": 13.241796493530273,
+ "learning_rate": 1.0917347865576748e-05,
+ "loss": 2.3037,
+ "step": 601
+ },
+ {
+ "epoch": 1.6403269754768393,
+ "grad_norm": 19.26212501525879,
+ "learning_rate": 1.0935513169845596e-05,
+ "loss": 2.3218,
+ "step": 602
+ },
+ {
+ "epoch": 1.6430517711171662,
+ "grad_norm": 11.671584129333496,
+ "learning_rate": 1.0953678474114442e-05,
+ "loss": 2.5195,
+ "step": 603
+ },
+ {
+ "epoch": 1.645776566757493,
+ "grad_norm": 15.691133499145508,
+ "learning_rate": 1.0971843778383288e-05,
+ "loss": 2.3599,
+ "step": 604
+ },
+ {
+ "epoch": 1.6485013623978202,
+ "grad_norm": 12.634986877441406,
+ "learning_rate": 1.0990009082652136e-05,
+ "loss": 1.9111,
+ "step": 605
+ },
+ {
+ "epoch": 1.651226158038147,
+ "grad_norm": 17.2967529296875,
+ "learning_rate": 1.1008174386920982e-05,
+ "loss": 2.3628,
+ "step": 606
+ },
+ {
+ "epoch": 1.653950953678474,
+ "grad_norm": 14.55881118774414,
+ "learning_rate": 1.1026339691189828e-05,
+ "loss": 2.4385,
+ "step": 607
+ },
+ {
+ "epoch": 1.6566757493188011,
+ "grad_norm": 9.866549491882324,
+ "learning_rate": 1.1044504995458676e-05,
+ "loss": 2.126,
+ "step": 608
+ },
+ {
+ "epoch": 1.659400544959128,
+ "grad_norm": 22.167556762695312,
+ "learning_rate": 1.106267029972752e-05,
+ "loss": 2.1294,
+ "step": 609
+ },
+ {
+ "epoch": 1.662125340599455,
+ "grad_norm": 10.998841285705566,
+ "learning_rate": 1.1080835603996367e-05,
+ "loss": 2.2329,
+ "step": 610
+ },
+ {
+ "epoch": 1.664850136239782,
+ "grad_norm": 22.989713668823242,
+ "learning_rate": 1.1099000908265215e-05,
+ "loss": 2.0459,
+ "step": 611
+ },
+ {
+ "epoch": 1.667574931880109,
+ "grad_norm": 13.676278114318848,
+ "learning_rate": 1.1117166212534061e-05,
+ "loss": 2.166,
+ "step": 612
+ },
+ {
+ "epoch": 1.6702997275204359,
+ "grad_norm": 20.328584671020508,
+ "learning_rate": 1.1135331516802907e-05,
+ "loss": 2.4155,
+ "step": 613
+ },
+ {
+ "epoch": 1.673024523160763,
+ "grad_norm": 20.008729934692383,
+ "learning_rate": 1.1153496821071755e-05,
+ "loss": 2.1924,
+ "step": 614
+ },
+ {
+ "epoch": 1.67574931880109,
+ "grad_norm": 12.8439302444458,
+ "learning_rate": 1.1171662125340601e-05,
+ "loss": 2.0991,
+ "step": 615
+ },
+ {
+ "epoch": 1.6784741144414168,
+ "grad_norm": 18.991416931152344,
+ "learning_rate": 1.1189827429609447e-05,
+ "loss": 2.4731,
+ "step": 616
+ },
+ {
+ "epoch": 1.681198910081744,
+ "grad_norm": 16.508209228515625,
+ "learning_rate": 1.1207992733878292e-05,
+ "loss": 2.4209,
+ "step": 617
+ },
+ {
+ "epoch": 1.6839237057220708,
+ "grad_norm": 14.190963745117188,
+ "learning_rate": 1.122615803814714e-05,
+ "loss": 2.3379,
+ "step": 618
+ },
+ {
+ "epoch": 1.6866485013623977,
+ "grad_norm": 12.22602653503418,
+ "learning_rate": 1.1244323342415986e-05,
+ "loss": 2.3101,
+ "step": 619
+ },
+ {
+ "epoch": 1.6893732970027249,
+ "grad_norm": 15.474383354187012,
+ "learning_rate": 1.1262488646684832e-05,
+ "loss": 2.2393,
+ "step": 620
+ },
+ {
+ "epoch": 1.6920980926430518,
+ "grad_norm": 13.547887802124023,
+ "learning_rate": 1.128065395095368e-05,
+ "loss": 2.1821,
+ "step": 621
+ },
+ {
+ "epoch": 1.6948228882833787,
+ "grad_norm": 12.980840682983398,
+ "learning_rate": 1.1298819255222526e-05,
+ "loss": 2.2896,
+ "step": 622
+ },
+ {
+ "epoch": 1.6975476839237058,
+ "grad_norm": 19.693824768066406,
+ "learning_rate": 1.1316984559491374e-05,
+ "loss": 2.1743,
+ "step": 623
+ },
+ {
+ "epoch": 1.7002724795640327,
+ "grad_norm": 9.108651161193848,
+ "learning_rate": 1.133514986376022e-05,
+ "loss": 2.1987,
+ "step": 624
+ },
+ {
+ "epoch": 1.7029972752043596,
+ "grad_norm": 14.882322311401367,
+ "learning_rate": 1.1353315168029066e-05,
+ "loss": 2.2964,
+ "step": 625
+ },
+ {
+ "epoch": 1.7057220708446867,
+ "grad_norm": 20.610509872436523,
+ "learning_rate": 1.137148047229791e-05,
+ "loss": 2.3555,
+ "step": 626
+ },
+ {
+ "epoch": 1.7084468664850136,
+ "grad_norm": 15.04577922821045,
+ "learning_rate": 1.1389645776566758e-05,
+ "loss": 2.3574,
+ "step": 627
+ },
+ {
+ "epoch": 1.7111716621253406,
+ "grad_norm": 17.739614486694336,
+ "learning_rate": 1.1407811080835605e-05,
+ "loss": 2.2461,
+ "step": 628
+ },
+ {
+ "epoch": 1.7138964577656677,
+ "grad_norm": 14.051782608032227,
+ "learning_rate": 1.142597638510445e-05,
+ "loss": 2.2827,
+ "step": 629
+ },
+ {
+ "epoch": 1.7166212534059946,
+ "grad_norm": 15.0321626663208,
+ "learning_rate": 1.1444141689373299e-05,
+ "loss": 2.4253,
+ "step": 630
+ },
+ {
+ "epoch": 1.7193460490463215,
+ "grad_norm": 14.484668731689453,
+ "learning_rate": 1.1462306993642145e-05,
+ "loss": 1.9019,
+ "step": 631
+ },
+ {
+ "epoch": 1.7220708446866486,
+ "grad_norm": 13.901585578918457,
+ "learning_rate": 1.1480472297910991e-05,
+ "loss": 2.208,
+ "step": 632
+ },
+ {
+ "epoch": 1.7247956403269755,
+ "grad_norm": 13.731334686279297,
+ "learning_rate": 1.1498637602179839e-05,
+ "loss": 2.2085,
+ "step": 633
+ },
+ {
+ "epoch": 1.7275204359673024,
+ "grad_norm": 15.021971702575684,
+ "learning_rate": 1.1516802906448683e-05,
+ "loss": 2.1226,
+ "step": 634
+ },
+ {
+ "epoch": 1.7302452316076296,
+ "grad_norm": 13.952202796936035,
+ "learning_rate": 1.153496821071753e-05,
+ "loss": 2.3809,
+ "step": 635
+ },
+ {
+ "epoch": 1.7329700272479565,
+ "grad_norm": 12.969078063964844,
+ "learning_rate": 1.1553133514986377e-05,
+ "loss": 2.0508,
+ "step": 636
+ },
+ {
+ "epoch": 1.7356948228882834,
+ "grad_norm": 9.670950889587402,
+ "learning_rate": 1.1571298819255223e-05,
+ "loss": 2.0933,
+ "step": 637
+ },
+ {
+ "epoch": 1.7384196185286105,
+ "grad_norm": 13.490191459655762,
+ "learning_rate": 1.158946412352407e-05,
+ "loss": 2.165,
+ "step": 638
+ },
+ {
+ "epoch": 1.7411444141689372,
+ "grad_norm": 13.26264762878418,
+ "learning_rate": 1.1607629427792917e-05,
+ "loss": 2.4482,
+ "step": 639
+ },
+ {
+ "epoch": 1.7438692098092643,
+ "grad_norm": 17.870344161987305,
+ "learning_rate": 1.1625794732061764e-05,
+ "loss": 2.1069,
+ "step": 640
+ },
+ {
+ "epoch": 1.7465940054495914,
+ "grad_norm": 11.028536796569824,
+ "learning_rate": 1.164396003633061e-05,
+ "loss": 2.1587,
+ "step": 641
+ },
+ {
+ "epoch": 1.749318801089918,
+ "grad_norm": 18.716938018798828,
+ "learning_rate": 1.1662125340599454e-05,
+ "loss": 2.0713,
+ "step": 642
+ },
+ {
+ "epoch": 1.7520435967302452,
+ "grad_norm": 11.706306457519531,
+ "learning_rate": 1.1680290644868302e-05,
+ "loss": 2.3467,
+ "step": 643
+ },
+ {
+ "epoch": 1.7547683923705724,
+ "grad_norm": 16.77366828918457,
+ "learning_rate": 1.1698455949137148e-05,
+ "loss": 2.0928,
+ "step": 644
+ },
+ {
+ "epoch": 1.757493188010899,
+ "grad_norm": 13.903936386108398,
+ "learning_rate": 1.1716621253405996e-05,
+ "loss": 2.5405,
+ "step": 645
+ },
+ {
+ "epoch": 1.7602179836512262,
+ "grad_norm": 17.29177474975586,
+ "learning_rate": 1.1734786557674842e-05,
+ "loss": 2.1528,
+ "step": 646
+ },
+ {
+ "epoch": 1.7629427792915533,
+ "grad_norm": 12.351784706115723,
+ "learning_rate": 1.1752951861943688e-05,
+ "loss": 2.144,
+ "step": 647
+ },
+ {
+ "epoch": 1.76566757493188,
+ "grad_norm": 11.504490852355957,
+ "learning_rate": 1.1771117166212536e-05,
+ "loss": 2.2334,
+ "step": 648
+ },
+ {
+ "epoch": 1.768392370572207,
+ "grad_norm": 15.75640869140625,
+ "learning_rate": 1.1789282470481382e-05,
+ "loss": 2.0249,
+ "step": 649
+ },
+ {
+ "epoch": 1.771117166212534,
+ "grad_norm": 12.768902778625488,
+ "learning_rate": 1.1807447774750227e-05,
+ "loss": 2.2949,
+ "step": 650
+ },
+ {
+ "epoch": 1.773841961852861,
+ "grad_norm": 12.6156644821167,
+ "learning_rate": 1.1825613079019073e-05,
+ "loss": 2.1211,
+ "step": 651
+ },
+ {
+ "epoch": 1.776566757493188,
+ "grad_norm": 11.044657707214355,
+ "learning_rate": 1.1843778383287921e-05,
+ "loss": 2.1797,
+ "step": 652
+ },
+ {
+ "epoch": 1.779291553133515,
+ "grad_norm": 8.811762809753418,
+ "learning_rate": 1.1861943687556767e-05,
+ "loss": 2.1133,
+ "step": 653
+ },
+ {
+ "epoch": 1.7820163487738419,
+ "grad_norm": 9.669388771057129,
+ "learning_rate": 1.1880108991825613e-05,
+ "loss": 2.166,
+ "step": 654
+ },
+ {
+ "epoch": 1.784741144414169,
+ "grad_norm": 9.426198959350586,
+ "learning_rate": 1.1898274296094461e-05,
+ "loss": 2.0283,
+ "step": 655
+ },
+ {
+ "epoch": 1.7874659400544959,
+ "grad_norm": 12.840489387512207,
+ "learning_rate": 1.1916439600363307e-05,
+ "loss": 2.166,
+ "step": 656
+ },
+ {
+ "epoch": 1.7901907356948228,
+ "grad_norm": 9.510618209838867,
+ "learning_rate": 1.1934604904632155e-05,
+ "loss": 2.0986,
+ "step": 657
+ },
+ {
+ "epoch": 1.79291553133515,
+ "grad_norm": 13.508071899414062,
+ "learning_rate": 1.1952770208901001e-05,
+ "loss": 2.269,
+ "step": 658
+ },
+ {
+ "epoch": 1.7956403269754768,
+ "grad_norm": 13.299254417419434,
+ "learning_rate": 1.1970935513169846e-05,
+ "loss": 2.3555,
+ "step": 659
+ },
+ {
+ "epoch": 1.7983651226158037,
+ "grad_norm": 12.642251968383789,
+ "learning_rate": 1.1989100817438692e-05,
+ "loss": 2.1577,
+ "step": 660
+ },
+ {
+ "epoch": 1.8010899182561309,
+ "grad_norm": 11.772926330566406,
+ "learning_rate": 1.200726612170754e-05,
+ "loss": 2.0703,
+ "step": 661
+ },
+ {
+ "epoch": 1.8038147138964578,
+ "grad_norm": 8.32027530670166,
+ "learning_rate": 1.2025431425976386e-05,
+ "loss": 2.4272,
+ "step": 662
+ },
+ {
+ "epoch": 1.8065395095367847,
+ "grad_norm": 10.469939231872559,
+ "learning_rate": 1.2043596730245232e-05,
+ "loss": 2.2148,
+ "step": 663
+ },
+ {
+ "epoch": 1.8092643051771118,
+ "grad_norm": 10.922989845275879,
+ "learning_rate": 1.206176203451408e-05,
+ "loss": 2.1646,
+ "step": 664
+ },
+ {
+ "epoch": 1.8119891008174387,
+ "grad_norm": 10.295989990234375,
+ "learning_rate": 1.2079927338782926e-05,
+ "loss": 2.1812,
+ "step": 665
+ },
+ {
+ "epoch": 1.8147138964577656,
+ "grad_norm": 14.525970458984375,
+ "learning_rate": 1.2098092643051772e-05,
+ "loss": 2.2251,
+ "step": 666
+ },
+ {
+ "epoch": 1.8174386920980927,
+ "grad_norm": 15.493196487426758,
+ "learning_rate": 1.2116257947320618e-05,
+ "loss": 2.3325,
+ "step": 667
+ },
+ {
+ "epoch": 1.8201634877384196,
+ "grad_norm": 14.37890911102295,
+ "learning_rate": 1.2134423251589465e-05,
+ "loss": 2.4829,
+ "step": 668
+ },
+ {
+ "epoch": 1.8228882833787465,
+ "grad_norm": 10.197138786315918,
+ "learning_rate": 1.215258855585831e-05,
+ "loss": 2.229,
+ "step": 669
+ },
+ {
+ "epoch": 1.8256130790190737,
+ "grad_norm": 10.858696937561035,
+ "learning_rate": 1.2170753860127159e-05,
+ "loss": 2.2202,
+ "step": 670
+ },
+ {
+ "epoch": 1.8283378746594006,
+ "grad_norm": 13.920210838317871,
+ "learning_rate": 1.2188919164396005e-05,
+ "loss": 2.085,
+ "step": 671
+ },
+ {
+ "epoch": 1.8310626702997275,
+ "grad_norm": 13.829449653625488,
+ "learning_rate": 1.2207084468664851e-05,
+ "loss": 2.1123,
+ "step": 672
+ },
+ {
+ "epoch": 1.8337874659400546,
+ "grad_norm": 11.385647773742676,
+ "learning_rate": 1.2225249772933699e-05,
+ "loss": 2.2666,
+ "step": 673
+ },
+ {
+ "epoch": 1.8365122615803815,
+ "grad_norm": 8.457975387573242,
+ "learning_rate": 1.2243415077202545e-05,
+ "loss": 2.2793,
+ "step": 674
+ },
+ {
+ "epoch": 1.8392370572207084,
+ "grad_norm": 9.353926658630371,
+ "learning_rate": 1.226158038147139e-05,
+ "loss": 2.1978,
+ "step": 675
+ },
+ {
+ "epoch": 1.8419618528610355,
+ "grad_norm": 10.374506950378418,
+ "learning_rate": 1.2279745685740236e-05,
+ "loss": 1.9604,
+ "step": 676
+ },
+ {
+ "epoch": 1.8446866485013624,
+ "grad_norm": 8.6756591796875,
+ "learning_rate": 1.2297910990009083e-05,
+ "loss": 2.0547,
+ "step": 677
+ },
+ {
+ "epoch": 1.8474114441416893,
+ "grad_norm": 11.98441219329834,
+ "learning_rate": 1.231607629427793e-05,
+ "loss": 2.2666,
+ "step": 678
+ },
+ {
+ "epoch": 1.8501362397820165,
+ "grad_norm": 14.200121879577637,
+ "learning_rate": 1.2334241598546777e-05,
+ "loss": 2.2173,
+ "step": 679
+ },
+ {
+ "epoch": 1.8528610354223434,
+ "grad_norm": 9.850325584411621,
+ "learning_rate": 1.2352406902815624e-05,
+ "loss": 2.2759,
+ "step": 680
+ },
+ {
+ "epoch": 1.8555858310626703,
+ "grad_norm": 10.285653114318848,
+ "learning_rate": 1.237057220708447e-05,
+ "loss": 2.4673,
+ "step": 681
+ },
+ {
+ "epoch": 1.8583106267029974,
+ "grad_norm": 10.727814674377441,
+ "learning_rate": 1.2388737511353318e-05,
+ "loss": 2.1455,
+ "step": 682
+ },
+ {
+ "epoch": 1.861035422343324,
+ "grad_norm": 9.151063919067383,
+ "learning_rate": 1.2406902815622162e-05,
+ "loss": 1.9395,
+ "step": 683
+ },
+ {
+ "epoch": 1.8637602179836512,
+ "grad_norm": 10.544675827026367,
+ "learning_rate": 1.2425068119891008e-05,
+ "loss": 1.9038,
+ "step": 684
+ },
+ {
+ "epoch": 1.8664850136239783,
+ "grad_norm": 14.572624206542969,
+ "learning_rate": 1.2443233424159854e-05,
+ "loss": 2.1958,
+ "step": 685
+ },
+ {
+ "epoch": 1.869209809264305,
+ "grad_norm": 15.917181015014648,
+ "learning_rate": 1.2461398728428702e-05,
+ "loss": 2.1758,
+ "step": 686
+ },
+ {
+ "epoch": 1.8719346049046321,
+ "grad_norm": 10.748626708984375,
+ "learning_rate": 1.2479564032697548e-05,
+ "loss": 2.3003,
+ "step": 687
+ },
+ {
+ "epoch": 1.8746594005449593,
+ "grad_norm": 17.07168960571289,
+ "learning_rate": 1.2497729336966395e-05,
+ "loss": 2.0483,
+ "step": 688
+ },
+ {
+ "epoch": 1.877384196185286,
+ "grad_norm": 9.101595878601074,
+ "learning_rate": 1.2515894641235242e-05,
+ "loss": 2.1719,
+ "step": 689
+ },
+ {
+ "epoch": 1.880108991825613,
+ "grad_norm": 9.80299186706543,
+ "learning_rate": 1.2534059945504089e-05,
+ "loss": 2.0679,
+ "step": 690
+ },
+ {
+ "epoch": 1.8828337874659402,
+ "grad_norm": 13.277621269226074,
+ "learning_rate": 1.2552225249772936e-05,
+ "loss": 2.2515,
+ "step": 691
+ },
+ {
+ "epoch": 1.885558583106267,
+ "grad_norm": 9.144055366516113,
+ "learning_rate": 1.2570390554041781e-05,
+ "loss": 2.0342,
+ "step": 692
+ },
+ {
+ "epoch": 1.888283378746594,
+ "grad_norm": 9.615757942199707,
+ "learning_rate": 1.2588555858310627e-05,
+ "loss": 2.2036,
+ "step": 693
+ },
+ {
+ "epoch": 1.891008174386921,
+ "grad_norm": 9.385332107543945,
+ "learning_rate": 1.2606721162579473e-05,
+ "loss": 2.3276,
+ "step": 694
+ },
+ {
+ "epoch": 1.8937329700272478,
+ "grad_norm": 10.023481369018555,
+ "learning_rate": 1.2624886466848321e-05,
+ "loss": 2.0503,
+ "step": 695
+ },
+ {
+ "epoch": 1.896457765667575,
+ "grad_norm": 8.705049514770508,
+ "learning_rate": 1.2643051771117167e-05,
+ "loss": 2.1201,
+ "step": 696
+ },
+ {
+ "epoch": 1.8991825613079019,
+ "grad_norm": 12.702301979064941,
+ "learning_rate": 1.2661217075386013e-05,
+ "loss": 2.2017,
+ "step": 697
+ },
+ {
+ "epoch": 1.9019073569482288,
+ "grad_norm": 13.401511192321777,
+ "learning_rate": 1.2679382379654861e-05,
+ "loss": 2.3477,
+ "step": 698
+ },
+ {
+ "epoch": 1.904632152588556,
+ "grad_norm": 10.565689086914062,
+ "learning_rate": 1.2697547683923707e-05,
+ "loss": 2.3101,
+ "step": 699
+ },
+ {
+ "epoch": 1.9073569482288828,
+ "grad_norm": 11.252874374389648,
+ "learning_rate": 1.2715712988192552e-05,
+ "loss": 2.1992,
+ "step": 700
+ },
+ {
+ "epoch": 1.9100817438692097,
+ "grad_norm": 12.494627952575684,
+ "learning_rate": 1.27338782924614e-05,
+ "loss": 2.1108,
+ "step": 701
+ },
+ {
+ "epoch": 1.9128065395095368,
+ "grad_norm": 9.181783676147461,
+ "learning_rate": 1.2752043596730246e-05,
+ "loss": 1.9751,
+ "step": 702
+ },
+ {
+ "epoch": 1.9155313351498637,
+ "grad_norm": 10.77589225769043,
+ "learning_rate": 1.2770208900999092e-05,
+ "loss": 2.1479,
+ "step": 703
+ },
+ {
+ "epoch": 1.9182561307901906,
+ "grad_norm": 13.986553192138672,
+ "learning_rate": 1.278837420526794e-05,
+ "loss": 2.104,
+ "step": 704
+ },
+ {
+ "epoch": 1.9209809264305178,
+ "grad_norm": 11.042075157165527,
+ "learning_rate": 1.2806539509536786e-05,
+ "loss": 2.2495,
+ "step": 705
+ },
+ {
+ "epoch": 1.9237057220708447,
+ "grad_norm": 10.622434616088867,
+ "learning_rate": 1.2824704813805632e-05,
+ "loss": 2.2886,
+ "step": 706
+ },
+ {
+ "epoch": 1.9264305177111716,
+ "grad_norm": 10.570306777954102,
+ "learning_rate": 1.284287011807448e-05,
+ "loss": 2.1807,
+ "step": 707
+ },
+ {
+ "epoch": 1.9291553133514987,
+ "grad_norm": 7.8538079261779785,
+ "learning_rate": 1.2861035422343325e-05,
+ "loss": 1.8638,
+ "step": 708
+ },
+ {
+ "epoch": 1.9318801089918256,
+ "grad_norm": 10.962336540222168,
+ "learning_rate": 1.287920072661217e-05,
+ "loss": 2.0078,
+ "step": 709
+ },
+ {
+ "epoch": 1.9346049046321525,
+ "grad_norm": 17.356441497802734,
+ "learning_rate": 1.2897366030881019e-05,
+ "loss": 2.019,
+ "step": 710
+ },
+ {
+ "epoch": 1.9373297002724796,
+ "grad_norm": 9.167985916137695,
+ "learning_rate": 1.2915531335149865e-05,
+ "loss": 2.0562,
+ "step": 711
+ },
+ {
+ "epoch": 1.9400544959128065,
+ "grad_norm": 11.627812385559082,
+ "learning_rate": 1.2933696639418711e-05,
+ "loss": 1.958,
+ "step": 712
+ },
+ {
+ "epoch": 1.9427792915531334,
+ "grad_norm": 10.228023529052734,
+ "learning_rate": 1.2951861943687559e-05,
+ "loss": 2.4058,
+ "step": 713
+ },
+ {
+ "epoch": 1.9455040871934606,
+ "grad_norm": 11.829590797424316,
+ "learning_rate": 1.2970027247956405e-05,
+ "loss": 2.1006,
+ "step": 714
+ },
+ {
+ "epoch": 1.9482288828337875,
+ "grad_norm": 10.62134838104248,
+ "learning_rate": 1.2988192552225251e-05,
+ "loss": 2.1152,
+ "step": 715
+ },
+ {
+ "epoch": 1.9509536784741144,
+ "grad_norm": 8.511664390563965,
+ "learning_rate": 1.3006357856494099e-05,
+ "loss": 2.2051,
+ "step": 716
+ },
+ {
+ "epoch": 1.9536784741144415,
+ "grad_norm": 10.296326637268066,
+ "learning_rate": 1.3024523160762943e-05,
+ "loss": 2.0068,
+ "step": 717
+ },
+ {
+ "epoch": 1.9564032697547684,
+ "grad_norm": 12.63467788696289,
+ "learning_rate": 1.304268846503179e-05,
+ "loss": 2.0557,
+ "step": 718
+ },
+ {
+ "epoch": 1.9591280653950953,
+ "grad_norm": 10.057330131530762,
+ "learning_rate": 1.3060853769300636e-05,
+ "loss": 2.3008,
+ "step": 719
+ },
+ {
+ "epoch": 1.9618528610354224,
+ "grad_norm": 13.027626991271973,
+ "learning_rate": 1.3079019073569484e-05,
+ "loss": 2.2842,
+ "step": 720
+ },
+ {
+ "epoch": 1.9645776566757494,
+ "grad_norm": 12.011673927307129,
+ "learning_rate": 1.309718437783833e-05,
+ "loss": 2.2417,
+ "step": 721
+ },
+ {
+ "epoch": 1.9673024523160763,
+ "grad_norm": 10.065938949584961,
+ "learning_rate": 1.3115349682107178e-05,
+ "loss": 2.2275,
+ "step": 722
+ },
+ {
+ "epoch": 1.9700272479564034,
+ "grad_norm": 10.303174018859863,
+ "learning_rate": 1.3133514986376024e-05,
+ "loss": 2.0435,
+ "step": 723
+ },
+ {
+ "epoch": 1.9727520435967303,
+ "grad_norm": 12.654775619506836,
+ "learning_rate": 1.315168029064487e-05,
+ "loss": 2.0049,
+ "step": 724
+ },
+ {
+ "epoch": 1.9754768392370572,
+ "grad_norm": 11.94584846496582,
+ "learning_rate": 1.3169845594913714e-05,
+ "loss": 2.1641,
+ "step": 725
+ },
+ {
+ "epoch": 1.9782016348773843,
+ "grad_norm": 14.929556846618652,
+ "learning_rate": 1.3188010899182562e-05,
+ "loss": 2.1772,
+ "step": 726
+ },
+ {
+ "epoch": 1.9809264305177112,
+ "grad_norm": 10.804765701293945,
+ "learning_rate": 1.3206176203451408e-05,
+ "loss": 2.0195,
+ "step": 727
+ },
+ {
+ "epoch": 1.9836512261580381,
+ "grad_norm": 18.633773803710938,
+ "learning_rate": 1.3224341507720255e-05,
+ "loss": 2.1865,
+ "step": 728
+ },
+ {
+ "epoch": 1.9863760217983653,
+ "grad_norm": 15.437492370605469,
+ "learning_rate": 1.3242506811989102e-05,
+ "loss": 1.9932,
+ "step": 729
+ },
+ {
+ "epoch": 1.989100817438692,
+ "grad_norm": 11.284488677978516,
+ "learning_rate": 1.3260672116257949e-05,
+ "loss": 2.3086,
+ "step": 730
+ },
+ {
+ "epoch": 1.991825613079019,
+ "grad_norm": 13.088354110717773,
+ "learning_rate": 1.3278837420526795e-05,
+ "loss": 2.1987,
+ "step": 731
+ },
+ {
+ "epoch": 1.9945504087193462,
+ "grad_norm": 10.205222129821777,
+ "learning_rate": 1.3297002724795643e-05,
+ "loss": 2.2471,
+ "step": 732
+ },
+ {
+ "epoch": 1.9972752043596729,
+ "grad_norm": 28.108232498168945,
+ "learning_rate": 1.3315168029064487e-05,
+ "loss": 2.0815,
+ "step": 733
+ },
+ {
+ "epoch": 2.0,
+ "grad_norm": 12.676140785217285,
+ "learning_rate": 1.3333333333333333e-05,
+ "loss": 2.0586,
+ "step": 734
+ },
+ {
+ "epoch": 2.002724795640327,
+ "grad_norm": 10.758481979370117,
+ "learning_rate": 1.3351498637602181e-05,
+ "loss": 2.1411,
+ "step": 735
+ },
+ {
+ "epoch": 2.005449591280654,
+ "grad_norm": 13.262362480163574,
+ "learning_rate": 1.3369663941871027e-05,
+ "loss": 2.1104,
+ "step": 736
+ },
+ {
+ "epoch": 2.008174386920981,
+ "grad_norm": 10.408624649047852,
+ "learning_rate": 1.3387829246139873e-05,
+ "loss": 2.1372,
+ "step": 737
+ },
+ {
+ "epoch": 2.010899182561308,
+ "grad_norm": 9.706769943237305,
+ "learning_rate": 1.3405994550408721e-05,
+ "loss": 1.9229,
+ "step": 738
+ },
+ {
+ "epoch": 2.0136239782016347,
+ "grad_norm": 15.630605697631836,
+ "learning_rate": 1.3424159854677567e-05,
+ "loss": 2.1362,
+ "step": 739
+ },
+ {
+ "epoch": 2.016348773841962,
+ "grad_norm": 13.618534088134766,
+ "learning_rate": 1.3442325158946414e-05,
+ "loss": 1.9727,
+ "step": 740
+ },
+ {
+ "epoch": 2.019073569482289,
+ "grad_norm": 12.626867294311523,
+ "learning_rate": 1.3460490463215258e-05,
+ "loss": 2.0679,
+ "step": 741
+ },
+ {
+ "epoch": 2.0217983651226157,
+ "grad_norm": 12.509529113769531,
+ "learning_rate": 1.3478655767484106e-05,
+ "loss": 2.0132,
+ "step": 742
+ },
+ {
+ "epoch": 2.024523160762943,
+ "grad_norm": 10.605827331542969,
+ "learning_rate": 1.3496821071752952e-05,
+ "loss": 2.1968,
+ "step": 743
+ },
+ {
+ "epoch": 2.02724795640327,
+ "grad_norm": 16.233068466186523,
+ "learning_rate": 1.35149863760218e-05,
+ "loss": 1.958,
+ "step": 744
+ },
+ {
+ "epoch": 2.0299727520435966,
+ "grad_norm": 12.001585006713867,
+ "learning_rate": 1.3533151680290646e-05,
+ "loss": 1.9941,
+ "step": 745
+ },
+ {
+ "epoch": 2.0326975476839237,
+ "grad_norm": 14.421778678894043,
+ "learning_rate": 1.3551316984559492e-05,
+ "loss": 1.9937,
+ "step": 746
+ },
+ {
+ "epoch": 2.035422343324251,
+ "grad_norm": 11.729394912719727,
+ "learning_rate": 1.356948228882834e-05,
+ "loss": 1.9082,
+ "step": 747
+ },
+ {
+ "epoch": 2.0381471389645776,
+ "grad_norm": 24.5251522064209,
+ "learning_rate": 1.3587647593097186e-05,
+ "loss": 2.0391,
+ "step": 748
+ },
+ {
+ "epoch": 2.0408719346049047,
+ "grad_norm": 8.919933319091797,
+ "learning_rate": 1.3605812897366032e-05,
+ "loss": 2.231,
+ "step": 749
+ },
+ {
+ "epoch": 2.043596730245232,
+ "grad_norm": 13.006795883178711,
+ "learning_rate": 1.3623978201634877e-05,
+ "loss": 2.127,
+ "step": 750
+ },
+ {
+ "epoch": 2.0463215258855585,
+ "grad_norm": 11.751535415649414,
+ "learning_rate": 1.3642143505903725e-05,
+ "loss": 1.979,
+ "step": 751
+ },
+ {
+ "epoch": 2.0490463215258856,
+ "grad_norm": 10.314565658569336,
+ "learning_rate": 1.3660308810172571e-05,
+ "loss": 2.1353,
+ "step": 752
+ },
+ {
+ "epoch": 2.0517711171662127,
+ "grad_norm": 7.867048263549805,
+ "learning_rate": 1.3678474114441417e-05,
+ "loss": 1.9341,
+ "step": 753
+ },
+ {
+ "epoch": 2.0544959128065394,
+ "grad_norm": 9.992166519165039,
+ "learning_rate": 1.3696639418710265e-05,
+ "loss": 2.4731,
+ "step": 754
+ },
+ {
+ "epoch": 2.0572207084468666,
+ "grad_norm": 9.752199172973633,
+ "learning_rate": 1.3714804722979111e-05,
+ "loss": 2.0742,
+ "step": 755
+ },
+ {
+ "epoch": 2.0599455040871932,
+ "grad_norm": 9.495267868041992,
+ "learning_rate": 1.3732970027247959e-05,
+ "loss": 2.1504,
+ "step": 756
+ },
+ {
+ "epoch": 2.0626702997275204,
+ "grad_norm": 10.785100936889648,
+ "learning_rate": 1.3751135331516805e-05,
+ "loss": 2.1772,
+ "step": 757
+ },
+ {
+ "epoch": 2.0653950953678475,
+ "grad_norm": 10.663728713989258,
+ "learning_rate": 1.376930063578565e-05,
+ "loss": 2.0005,
+ "step": 758
+ },
+ {
+ "epoch": 2.068119891008174,
+ "grad_norm": 11.44454288482666,
+ "learning_rate": 1.3787465940054496e-05,
+ "loss": 2.0688,
+ "step": 759
+ },
+ {
+ "epoch": 2.0708446866485013,
+ "grad_norm": 8.725317001342773,
+ "learning_rate": 1.3805631244323344e-05,
+ "loss": 1.9136,
+ "step": 760
+ },
+ {
+ "epoch": 2.0735694822888284,
+ "grad_norm": 9.857766151428223,
+ "learning_rate": 1.382379654859219e-05,
+ "loss": 2.1128,
+ "step": 761
+ },
+ {
+ "epoch": 2.076294277929155,
+ "grad_norm": 11.957629203796387,
+ "learning_rate": 1.3841961852861036e-05,
+ "loss": 2.2358,
+ "step": 762
+ },
+ {
+ "epoch": 2.0790190735694822,
+ "grad_norm": 29.050384521484375,
+ "learning_rate": 1.3860127157129884e-05,
+ "loss": 2.0791,
+ "step": 763
+ },
+ {
+ "epoch": 2.0817438692098094,
+ "grad_norm": 7.446699142456055,
+ "learning_rate": 1.387829246139873e-05,
+ "loss": 2.103,
+ "step": 764
+ },
+ {
+ "epoch": 2.084468664850136,
+ "grad_norm": 10.462141990661621,
+ "learning_rate": 1.3896457765667576e-05,
+ "loss": 2.0068,
+ "step": 765
+ },
+ {
+ "epoch": 2.087193460490463,
+ "grad_norm": 12.072513580322266,
+ "learning_rate": 1.3914623069936422e-05,
+ "loss": 2.3853,
+ "step": 766
+ },
+ {
+ "epoch": 2.0899182561307903,
+ "grad_norm": 8.111771583557129,
+ "learning_rate": 1.3932788374205268e-05,
+ "loss": 2.2271,
+ "step": 767
+ },
+ {
+ "epoch": 2.092643051771117,
+ "grad_norm": 25.044139862060547,
+ "learning_rate": 1.3950953678474115e-05,
+ "loss": 2.1484,
+ "step": 768
+ },
+ {
+ "epoch": 2.095367847411444,
+ "grad_norm": 13.04293441772461,
+ "learning_rate": 1.3969118982742963e-05,
+ "loss": 1.936,
+ "step": 769
+ },
+ {
+ "epoch": 2.0980926430517712,
+ "grad_norm": 10.704666137695312,
+ "learning_rate": 1.3987284287011809e-05,
+ "loss": 2.0571,
+ "step": 770
+ },
+ {
+ "epoch": 2.100817438692098,
+ "grad_norm": 8.816221237182617,
+ "learning_rate": 1.4005449591280655e-05,
+ "loss": 1.9722,
+ "step": 771
+ },
+ {
+ "epoch": 2.103542234332425,
+ "grad_norm": 19.80216407775879,
+ "learning_rate": 1.4023614895549503e-05,
+ "loss": 2.166,
+ "step": 772
+ },
+ {
+ "epoch": 2.106267029972752,
+ "grad_norm": 13.969697952270508,
+ "learning_rate": 1.4041780199818349e-05,
+ "loss": 2.1973,
+ "step": 773
+ },
+ {
+ "epoch": 2.108991825613079,
+ "grad_norm": 10.191165924072266,
+ "learning_rate": 1.4059945504087195e-05,
+ "loss": 2.1147,
+ "step": 774
+ },
+ {
+ "epoch": 2.111716621253406,
+ "grad_norm": 9.635372161865234,
+ "learning_rate": 1.407811080835604e-05,
+ "loss": 2.1294,
+ "step": 775
+ },
+ {
+ "epoch": 2.114441416893733,
+ "grad_norm": 10.28138542175293,
+ "learning_rate": 1.4096276112624887e-05,
+ "loss": 1.9863,
+ "step": 776
+ },
+ {
+ "epoch": 2.11716621253406,
+ "grad_norm": 8.500015258789062,
+ "learning_rate": 1.4114441416893733e-05,
+ "loss": 2.0591,
+ "step": 777
+ },
+ {
+ "epoch": 2.119891008174387,
+ "grad_norm": 27.842395782470703,
+ "learning_rate": 1.4132606721162581e-05,
+ "loss": 2.0239,
+ "step": 778
+ },
+ {
+ "epoch": 2.122615803814714,
+ "grad_norm": 9.072884559631348,
+ "learning_rate": 1.4150772025431428e-05,
+ "loss": 2.2642,
+ "step": 779
+ },
+ {
+ "epoch": 2.1253405994550407,
+ "grad_norm": 10.902450561523438,
+ "learning_rate": 1.4168937329700274e-05,
+ "loss": 2.2373,
+ "step": 780
+ },
+ {
+ "epoch": 2.128065395095368,
+ "grad_norm": 9.623045921325684,
+ "learning_rate": 1.4187102633969122e-05,
+ "loss": 1.9893,
+ "step": 781
+ },
+ {
+ "epoch": 2.130790190735695,
+ "grad_norm": 8.478821754455566,
+ "learning_rate": 1.4205267938237968e-05,
+ "loss": 1.8882,
+ "step": 782
+ },
+ {
+ "epoch": 2.1335149863760217,
+ "grad_norm": 8.97408676147461,
+ "learning_rate": 1.4223433242506812e-05,
+ "loss": 2.0884,
+ "step": 783
+ },
+ {
+ "epoch": 2.136239782016349,
+ "grad_norm": 9.81511402130127,
+ "learning_rate": 1.4241598546775658e-05,
+ "loss": 2.0352,
+ "step": 784
+ },
+ {
+ "epoch": 2.138964577656676,
+ "grad_norm": 10.338080406188965,
+ "learning_rate": 1.4259763851044506e-05,
+ "loss": 2.103,
+ "step": 785
+ },
+ {
+ "epoch": 2.1416893732970026,
+ "grad_norm": 9.241307258605957,
+ "learning_rate": 1.4277929155313352e-05,
+ "loss": 2.3418,
+ "step": 786
+ },
+ {
+ "epoch": 2.1444141689373297,
+ "grad_norm": 10.9200439453125,
+ "learning_rate": 1.4296094459582198e-05,
+ "loss": 2.0688,
+ "step": 787
+ },
+ {
+ "epoch": 2.147138964577657,
+ "grad_norm": 10.084722518920898,
+ "learning_rate": 1.4314259763851046e-05,
+ "loss": 2.1616,
+ "step": 788
+ },
+ {
+ "epoch": 2.1498637602179835,
+ "grad_norm": 11.765569686889648,
+ "learning_rate": 1.4332425068119893e-05,
+ "loss": 2.0034,
+ "step": 789
+ },
+ {
+ "epoch": 2.1525885558583107,
+ "grad_norm": 9.805519104003906,
+ "learning_rate": 1.435059037238874e-05,
+ "loss": 2.2407,
+ "step": 790
+ },
+ {
+ "epoch": 2.155313351498638,
+ "grad_norm": 22.71735191345215,
+ "learning_rate": 1.4368755676657585e-05,
+ "loss": 1.9043,
+ "step": 791
+ },
+ {
+ "epoch": 2.1580381471389645,
+ "grad_norm": 10.466413497924805,
+ "learning_rate": 1.4386920980926431e-05,
+ "loss": 2.2832,
+ "step": 792
+ },
+ {
+ "epoch": 2.1607629427792916,
+ "grad_norm": 9.28375244140625,
+ "learning_rate": 1.4405086285195277e-05,
+ "loss": 2.0098,
+ "step": 793
+ },
+ {
+ "epoch": 2.1634877384196187,
+ "grad_norm": 10.605369567871094,
+ "learning_rate": 1.4423251589464125e-05,
+ "loss": 1.8726,
+ "step": 794
+ },
+ {
+ "epoch": 2.1662125340599454,
+ "grad_norm": 13.982156753540039,
+ "learning_rate": 1.4441416893732971e-05,
+ "loss": 1.9941,
+ "step": 795
+ },
+ {
+ "epoch": 2.1689373297002725,
+ "grad_norm": 11.7191162109375,
+ "learning_rate": 1.4459582198001817e-05,
+ "loss": 1.978,
+ "step": 796
+ },
+ {
+ "epoch": 2.1716621253405997,
+ "grad_norm": 12.619403839111328,
+ "learning_rate": 1.4477747502270665e-05,
+ "loss": 2.2007,
+ "step": 797
+ },
+ {
+ "epoch": 2.1743869209809263,
+ "grad_norm": 8.853255271911621,
+ "learning_rate": 1.4495912806539511e-05,
+ "loss": 2.3208,
+ "step": 798
+ },
+ {
+ "epoch": 2.1771117166212535,
+ "grad_norm": 10.517702102661133,
+ "learning_rate": 1.4514078110808356e-05,
+ "loss": 2.0161,
+ "step": 799
+ },
+ {
+ "epoch": 2.1798365122615806,
+ "grad_norm": 10.933089256286621,
+ "learning_rate": 1.4532243415077204e-05,
+ "loss": 2.02,
+ "step": 800
+ },
+ {
+ "epoch": 2.1825613079019073,
+ "grad_norm": 9.504537582397461,
+ "learning_rate": 1.455040871934605e-05,
+ "loss": 2.0874,
+ "step": 801
+ },
+ {
+ "epoch": 2.1852861035422344,
+ "grad_norm": 9.586892127990723,
+ "learning_rate": 1.4568574023614896e-05,
+ "loss": 2.0737,
+ "step": 802
+ },
+ {
+ "epoch": 2.1880108991825615,
+ "grad_norm": 9.852901458740234,
+ "learning_rate": 1.4586739327883744e-05,
+ "loss": 1.8584,
+ "step": 803
+ },
+ {
+ "epoch": 2.190735694822888,
+ "grad_norm": 11.841078758239746,
+ "learning_rate": 1.460490463215259e-05,
+ "loss": 2.2261,
+ "step": 804
+ },
+ {
+ "epoch": 2.1934604904632153,
+ "grad_norm": 12.164193153381348,
+ "learning_rate": 1.4623069936421436e-05,
+ "loss": 1.9775,
+ "step": 805
+ },
+ {
+ "epoch": 2.1961852861035425,
+ "grad_norm": 9.975220680236816,
+ "learning_rate": 1.4641235240690284e-05,
+ "loss": 1.9868,
+ "step": 806
+ },
+ {
+ "epoch": 2.198910081743869,
+ "grad_norm": 10.224533081054688,
+ "learning_rate": 1.465940054495913e-05,
+ "loss": 2.1182,
+ "step": 807
+ },
+ {
+ "epoch": 2.2016348773841963,
+ "grad_norm": 14.166773796081543,
+ "learning_rate": 1.4677565849227975e-05,
+ "loss": 2.1851,
+ "step": 808
+ },
+ {
+ "epoch": 2.204359673024523,
+ "grad_norm": 10.865340232849121,
+ "learning_rate": 1.469573115349682e-05,
+ "loss": 1.9478,
+ "step": 809
+ },
+ {
+ "epoch": 2.20708446866485,
+ "grad_norm": 10.101744651794434,
+ "learning_rate": 1.4713896457765669e-05,
+ "loss": 2.1523,
+ "step": 810
+ },
+ {
+ "epoch": 2.209809264305177,
+ "grad_norm": 9.703421592712402,
+ "learning_rate": 1.4732061762034515e-05,
+ "loss": 1.8423,
+ "step": 811
+ },
+ {
+ "epoch": 2.212534059945504,
+ "grad_norm": 17.368026733398438,
+ "learning_rate": 1.4750227066303363e-05,
+ "loss": 1.8379,
+ "step": 812
+ },
+ {
+ "epoch": 2.215258855585831,
+ "grad_norm": 10.578218460083008,
+ "learning_rate": 1.4768392370572209e-05,
+ "loss": 2.0811,
+ "step": 813
+ },
+ {
+ "epoch": 2.217983651226158,
+ "grad_norm": 10.287043571472168,
+ "learning_rate": 1.4786557674841055e-05,
+ "loss": 1.8467,
+ "step": 814
+ },
+ {
+ "epoch": 2.220708446866485,
+ "grad_norm": 11.887712478637695,
+ "learning_rate": 1.4804722979109903e-05,
+ "loss": 2.0981,
+ "step": 815
+ },
+ {
+ "epoch": 2.223433242506812,
+ "grad_norm": 12.52492618560791,
+ "learning_rate": 1.4822888283378747e-05,
+ "loss": 2.0322,
+ "step": 816
+ },
+ {
+ "epoch": 2.226158038147139,
+ "grad_norm": 11.523629188537598,
+ "learning_rate": 1.4841053587647594e-05,
+ "loss": 2.3169,
+ "step": 817
+ },
+ {
+ "epoch": 2.2288828337874658,
+ "grad_norm": 12.850665092468262,
+ "learning_rate": 1.485921889191644e-05,
+ "loss": 1.9731,
+ "step": 818
+ },
+ {
+ "epoch": 2.231607629427793,
+ "grad_norm": 11.984336853027344,
+ "learning_rate": 1.4877384196185288e-05,
+ "loss": 2.3115,
+ "step": 819
+ },
+ {
+ "epoch": 2.23433242506812,
+ "grad_norm": 12.645007133483887,
+ "learning_rate": 1.4895549500454134e-05,
+ "loss": 1.9512,
+ "step": 820
+ },
+ {
+ "epoch": 2.2370572207084467,
+ "grad_norm": 9.355470657348633,
+ "learning_rate": 1.4913714804722982e-05,
+ "loss": 2.0093,
+ "step": 821
+ },
+ {
+ "epoch": 2.239782016348774,
+ "grad_norm": 15.1129789352417,
+ "learning_rate": 1.4931880108991828e-05,
+ "loss": 1.9556,
+ "step": 822
+ },
+ {
+ "epoch": 2.242506811989101,
+ "grad_norm": 10.328067779541016,
+ "learning_rate": 1.4950045413260674e-05,
+ "loss": 2.0464,
+ "step": 823
+ },
+ {
+ "epoch": 2.2452316076294276,
+ "grad_norm": 13.31899642944336,
+ "learning_rate": 1.4968210717529518e-05,
+ "loss": 2.095,
+ "step": 824
+ },
+ {
+ "epoch": 2.2479564032697548,
+ "grad_norm": 11.033398628234863,
+ "learning_rate": 1.4986376021798366e-05,
+ "loss": 1.918,
+ "step": 825
+ },
+ {
+ "epoch": 2.250681198910082,
+ "grad_norm": 11.66713809967041,
+ "learning_rate": 1.5004541326067212e-05,
+ "loss": 2.1934,
+ "step": 826
+ },
+ {
+ "epoch": 2.2534059945504086,
+ "grad_norm": 10.394495010375977,
+ "learning_rate": 1.5022706630336059e-05,
+ "loss": 2.0571,
+ "step": 827
+ },
+ {
+ "epoch": 2.2561307901907357,
+ "grad_norm": 11.521025657653809,
+ "learning_rate": 1.5040871934604906e-05,
+ "loss": 1.9473,
+ "step": 828
+ },
+ {
+ "epoch": 2.258855585831063,
+ "grad_norm": 14.14743709564209,
+ "learning_rate": 1.5059037238873753e-05,
+ "loss": 2.0439,
+ "step": 829
+ },
+ {
+ "epoch": 2.2615803814713895,
+ "grad_norm": 13.373621940612793,
+ "learning_rate": 1.5077202543142599e-05,
+ "loss": 1.9175,
+ "step": 830
+ },
+ {
+ "epoch": 2.2643051771117166,
+ "grad_norm": 8.883220672607422,
+ "learning_rate": 1.5095367847411447e-05,
+ "loss": 2.1616,
+ "step": 831
+ },
+ {
+ "epoch": 2.2670299727520438,
+ "grad_norm": 8.825295448303223,
+ "learning_rate": 1.5113533151680291e-05,
+ "loss": 2.1201,
+ "step": 832
+ },
+ {
+ "epoch": 2.2697547683923704,
+ "grad_norm": 19.074798583984375,
+ "learning_rate": 1.5131698455949137e-05,
+ "loss": 2.0015,
+ "step": 833
+ },
+ {
+ "epoch": 2.2724795640326976,
+ "grad_norm": 10.256616592407227,
+ "learning_rate": 1.5149863760217985e-05,
+ "loss": 2.1499,
+ "step": 834
+ },
+ {
+ "epoch": 2.2752043596730247,
+ "grad_norm": 12.374433517456055,
+ "learning_rate": 1.5168029064486831e-05,
+ "loss": 1.7856,
+ "step": 835
+ },
+ {
+ "epoch": 2.2779291553133514,
+ "grad_norm": 8.114362716674805,
+ "learning_rate": 1.5186194368755677e-05,
+ "loss": 1.9888,
+ "step": 836
+ },
+ {
+ "epoch": 2.2806539509536785,
+ "grad_norm": 12.041817665100098,
+ "learning_rate": 1.5204359673024525e-05,
+ "loss": 2.2554,
+ "step": 837
+ },
+ {
+ "epoch": 2.2833787465940056,
+ "grad_norm": 10.486526489257812,
+ "learning_rate": 1.5222524977293371e-05,
+ "loss": 2.0122,
+ "step": 838
+ },
+ {
+ "epoch": 2.2861035422343323,
+ "grad_norm": 9.341402053833008,
+ "learning_rate": 1.5240690281562218e-05,
+ "loss": 2.0186,
+ "step": 839
+ },
+ {
+ "epoch": 2.2888283378746594,
+ "grad_norm": 12.514056205749512,
+ "learning_rate": 1.5258855585831064e-05,
+ "loss": 2.0552,
+ "step": 840
+ },
+ {
+ "epoch": 2.291553133514986,
+ "grad_norm": 13.270676612854004,
+ "learning_rate": 1.5277020890099908e-05,
+ "loss": 1.845,
+ "step": 841
+ },
+ {
+ "epoch": 2.2942779291553133,
+ "grad_norm": 9.514739036560059,
+ "learning_rate": 1.5295186194368756e-05,
+ "loss": 1.918,
+ "step": 842
+ },
+ {
+ "epoch": 2.2970027247956404,
+ "grad_norm": 10.181303024291992,
+ "learning_rate": 1.5313351498637604e-05,
+ "loss": 2.0029,
+ "step": 843
+ },
+ {
+ "epoch": 2.299727520435967,
+ "grad_norm": 11.720427513122559,
+ "learning_rate": 1.533151680290645e-05,
+ "loss": 2.2749,
+ "step": 844
+ },
+ {
+ "epoch": 2.302452316076294,
+ "grad_norm": 12.217656135559082,
+ "learning_rate": 1.5349682107175296e-05,
+ "loss": 2.3193,
+ "step": 845
+ },
+ {
+ "epoch": 2.3051771117166213,
+ "grad_norm": 10.445160865783691,
+ "learning_rate": 1.5367847411444144e-05,
+ "loss": 1.9619,
+ "step": 846
+ },
+ {
+ "epoch": 2.307901907356948,
+ "grad_norm": 9.500401496887207,
+ "learning_rate": 1.538601271571299e-05,
+ "loss": 1.9717,
+ "step": 847
+ },
+ {
+ "epoch": 2.310626702997275,
+ "grad_norm": 10.876279830932617,
+ "learning_rate": 1.5404178019981836e-05,
+ "loss": 1.9165,
+ "step": 848
+ },
+ {
+ "epoch": 2.3133514986376023,
+ "grad_norm": 18.93634796142578,
+ "learning_rate": 1.542234332425068e-05,
+ "loss": 2.1362,
+ "step": 849
+ },
+ {
+ "epoch": 2.316076294277929,
+ "grad_norm": 11.713539123535156,
+ "learning_rate": 1.544050862851953e-05,
+ "loss": 2.1494,
+ "step": 850
+ },
+ {
+ "epoch": 2.318801089918256,
+ "grad_norm": 12.628451347351074,
+ "learning_rate": 1.5458673932788377e-05,
+ "loss": 2.2114,
+ "step": 851
+ },
+ {
+ "epoch": 2.321525885558583,
+ "grad_norm": 10.59986686706543,
+ "learning_rate": 1.547683923705722e-05,
+ "loss": 2.0918,
+ "step": 852
+ },
+ {
+ "epoch": 2.32425068119891,
+ "grad_norm": 9.933114051818848,
+ "learning_rate": 1.549500454132607e-05,
+ "loss": 1.8984,
+ "step": 853
+ },
+ {
+ "epoch": 2.326975476839237,
+ "grad_norm": 12.893415451049805,
+ "learning_rate": 1.5513169845594917e-05,
+ "loss": 2.0239,
+ "step": 854
+ },
+ {
+ "epoch": 2.329700272479564,
+ "grad_norm": 12.467634201049805,
+ "learning_rate": 1.553133514986376e-05,
+ "loss": 1.9937,
+ "step": 855
+ },
+ {
+ "epoch": 2.332425068119891,
+ "grad_norm": 10.379561424255371,
+ "learning_rate": 1.554950045413261e-05,
+ "loss": 1.9731,
+ "step": 856
+ },
+ {
+ "epoch": 2.335149863760218,
+ "grad_norm": 12.198521614074707,
+ "learning_rate": 1.5567665758401454e-05,
+ "loss": 2.1494,
+ "step": 857
+ },
+ {
+ "epoch": 2.337874659400545,
+ "grad_norm": 12.067804336547852,
+ "learning_rate": 1.55858310626703e-05,
+ "loss": 2.0063,
+ "step": 858
+ },
+ {
+ "epoch": 2.3405994550408717,
+ "grad_norm": 11.9051513671875,
+ "learning_rate": 1.5603996366939146e-05,
+ "loss": 2.0791,
+ "step": 859
+ },
+ {
+ "epoch": 2.343324250681199,
+ "grad_norm": 16.913909912109375,
+ "learning_rate": 1.5622161671207994e-05,
+ "loss": 2.1777,
+ "step": 860
+ },
+ {
+ "epoch": 2.346049046321526,
+ "grad_norm": 12.158474922180176,
+ "learning_rate": 1.564032697547684e-05,
+ "loss": 2.0068,
+ "step": 861
+ },
+ {
+ "epoch": 2.3487738419618527,
+ "grad_norm": 8.058082580566406,
+ "learning_rate": 1.5658492279745686e-05,
+ "loss": 1.9526,
+ "step": 862
+ },
+ {
+ "epoch": 2.35149863760218,
+ "grad_norm": 11.490276336669922,
+ "learning_rate": 1.5676657584014534e-05,
+ "loss": 1.9834,
+ "step": 863
+ },
+ {
+ "epoch": 2.354223433242507,
+ "grad_norm": 10.264274597167969,
+ "learning_rate": 1.5694822888283382e-05,
+ "loss": 1.8203,
+ "step": 864
+ },
+ {
+ "epoch": 2.3569482288828336,
+ "grad_norm": 11.697925567626953,
+ "learning_rate": 1.5712988192552226e-05,
+ "loss": 1.9136,
+ "step": 865
+ },
+ {
+ "epoch": 2.3596730245231607,
+ "grad_norm": 10.974601745605469,
+ "learning_rate": 1.573115349682107e-05,
+ "loss": 1.9209,
+ "step": 866
+ },
+ {
+ "epoch": 2.362397820163488,
+ "grad_norm": 15.019807815551758,
+ "learning_rate": 1.574931880108992e-05,
+ "loss": 1.9561,
+ "step": 867
+ },
+ {
+ "epoch": 2.3651226158038146,
+ "grad_norm": 50.877174377441406,
+ "learning_rate": 1.5767484105358766e-05,
+ "loss": 2.29,
+ "step": 868
+ },
+ {
+ "epoch": 2.3678474114441417,
+ "grad_norm": 12.312771797180176,
+ "learning_rate": 1.578564940962761e-05,
+ "loss": 2.168,
+ "step": 869
+ },
+ {
+ "epoch": 2.370572207084469,
+ "grad_norm": 14.136480331420898,
+ "learning_rate": 1.580381471389646e-05,
+ "loss": 1.8037,
+ "step": 870
+ },
+ {
+ "epoch": 2.3732970027247955,
+ "grad_norm": 9.425222396850586,
+ "learning_rate": 1.5821980018165307e-05,
+ "loss": 2.0479,
+ "step": 871
+ },
+ {
+ "epoch": 2.3760217983651226,
+ "grad_norm": 10.186162948608398,
+ "learning_rate": 1.5840145322434154e-05,
+ "loss": 2.208,
+ "step": 872
+ },
+ {
+ "epoch": 2.3787465940054497,
+ "grad_norm": 12.471014976501465,
+ "learning_rate": 1.5858310626703e-05,
+ "loss": 2.1689,
+ "step": 873
+ },
+ {
+ "epoch": 2.3814713896457764,
+ "grad_norm": 13.84926700592041,
+ "learning_rate": 1.5876475930971843e-05,
+ "loss": 1.9946,
+ "step": 874
+ },
+ {
+ "epoch": 2.3841961852861036,
+ "grad_norm": 8.127616882324219,
+ "learning_rate": 1.589464123524069e-05,
+ "loss": 1.8049,
+ "step": 875
+ },
+ {
+ "epoch": 2.3869209809264307,
+ "grad_norm": 13.027189254760742,
+ "learning_rate": 1.591280653950954e-05,
+ "loss": 1.9307,
+ "step": 876
+ },
+ {
+ "epoch": 2.3896457765667574,
+ "grad_norm": 7.429171085357666,
+ "learning_rate": 1.5930971843778384e-05,
+ "loss": 1.9351,
+ "step": 877
+ },
+ {
+ "epoch": 2.3923705722070845,
+ "grad_norm": 8.571662902832031,
+ "learning_rate": 1.594913714804723e-05,
+ "loss": 1.937,
+ "step": 878
+ },
+ {
+ "epoch": 2.3950953678474116,
+ "grad_norm": 8.260794639587402,
+ "learning_rate": 1.596730245231608e-05,
+ "loss": 2.1812,
+ "step": 879
+ },
+ {
+ "epoch": 2.3978201634877383,
+ "grad_norm": 8.720732688903809,
+ "learning_rate": 1.5985467756584924e-05,
+ "loss": 2.104,
+ "step": 880
+ },
+ {
+ "epoch": 2.4005449591280654,
+ "grad_norm": 18.84224510192871,
+ "learning_rate": 1.600363306085377e-05,
+ "loss": 2.2051,
+ "step": 881
+ },
+ {
+ "epoch": 2.4032697547683926,
+ "grad_norm": 10.656434059143066,
+ "learning_rate": 1.6021798365122616e-05,
+ "loss": 2.0327,
+ "step": 882
+ },
+ {
+ "epoch": 2.4059945504087192,
+ "grad_norm": 21.36994743347168,
+ "learning_rate": 1.6039963669391464e-05,
+ "loss": 1.9419,
+ "step": 883
+ },
+ {
+ "epoch": 2.4087193460490464,
+ "grad_norm": 15.157690048217773,
+ "learning_rate": 1.605812897366031e-05,
+ "loss": 2.27,
+ "step": 884
+ },
+ {
+ "epoch": 2.4114441416893735,
+ "grad_norm": 14.788305282592773,
+ "learning_rate": 1.6076294277929156e-05,
+ "loss": 2.1211,
+ "step": 885
+ },
+ {
+ "epoch": 2.4141689373297,
+ "grad_norm": 9.01799201965332,
+ "learning_rate": 1.6094459582198004e-05,
+ "loss": 1.8618,
+ "step": 886
+ },
+ {
+ "epoch": 2.4168937329700273,
+ "grad_norm": 11.318547248840332,
+ "learning_rate": 1.611262488646685e-05,
+ "loss": 1.8384,
+ "step": 887
+ },
+ {
+ "epoch": 2.4196185286103544,
+ "grad_norm": 11.178836822509766,
+ "learning_rate": 1.6130790190735696e-05,
+ "loss": 1.873,
+ "step": 888
+ },
+ {
+ "epoch": 2.422343324250681,
+ "grad_norm": 10.019317626953125,
+ "learning_rate": 1.6148955495004544e-05,
+ "loss": 1.958,
+ "step": 889
+ },
+ {
+ "epoch": 2.4250681198910082,
+ "grad_norm": 11.550836563110352,
+ "learning_rate": 1.616712079927339e-05,
+ "loss": 2.0493,
+ "step": 890
+ },
+ {
+ "epoch": 2.4277929155313354,
+ "grad_norm": 13.8357572555542,
+ "learning_rate": 1.6185286103542233e-05,
+ "loss": 2.0205,
+ "step": 891
+ },
+ {
+ "epoch": 2.430517711171662,
+ "grad_norm": 9.950587272644043,
+ "learning_rate": 1.620345140781108e-05,
+ "loss": 2.2007,
+ "step": 892
+ },
+ {
+ "epoch": 2.433242506811989,
+ "grad_norm": 9.454350471496582,
+ "learning_rate": 1.622161671207993e-05,
+ "loss": 1.9946,
+ "step": 893
+ },
+ {
+ "epoch": 2.4359673024523163,
+ "grad_norm": 10.966967582702637,
+ "learning_rate": 1.6239782016348777e-05,
+ "loss": 1.9487,
+ "step": 894
+ },
+ {
+ "epoch": 2.438692098092643,
+ "grad_norm": 12.306376457214355,
+ "learning_rate": 1.625794732061762e-05,
+ "loss": 2.0146,
+ "step": 895
+ },
+ {
+ "epoch": 2.44141689373297,
+ "grad_norm": 14.904928207397461,
+ "learning_rate": 1.627611262488647e-05,
+ "loss": 1.9429,
+ "step": 896
+ },
+ {
+ "epoch": 2.4441416893732972,
+ "grad_norm": 15.005547523498535,
+ "learning_rate": 1.6294277929155317e-05,
+ "loss": 2.1309,
+ "step": 897
+ },
+ {
+ "epoch": 2.446866485013624,
+ "grad_norm": 8.949843406677246,
+ "learning_rate": 1.631244323342416e-05,
+ "loss": 1.9355,
+ "step": 898
+ },
+ {
+ "epoch": 2.449591280653951,
+ "grad_norm": 11.011309623718262,
+ "learning_rate": 1.6330608537693006e-05,
+ "loss": 2.145,
+ "step": 899
+ },
+ {
+ "epoch": 2.452316076294278,
+ "grad_norm": 29.216814041137695,
+ "learning_rate": 1.6348773841961854e-05,
+ "loss": 1.9756,
+ "step": 900
+ },
+ {
+ "epoch": 2.455040871934605,
+ "grad_norm": 24.92950439453125,
+ "learning_rate": 1.63669391462307e-05,
+ "loss": 2.0986,
+ "step": 901
+ },
+ {
+ "epoch": 2.457765667574932,
+ "grad_norm": 11.205491065979004,
+ "learning_rate": 1.6385104450499546e-05,
+ "loss": 1.9683,
+ "step": 902
+ },
+ {
+ "epoch": 2.460490463215259,
+ "grad_norm": 10.578439712524414,
+ "learning_rate": 1.6403269754768394e-05,
+ "loss": 1.9824,
+ "step": 903
+ },
+ {
+ "epoch": 2.463215258855586,
+ "grad_norm": 16.741554260253906,
+ "learning_rate": 1.6421435059037242e-05,
+ "loss": 2.147,
+ "step": 904
+ },
+ {
+ "epoch": 2.465940054495913,
+ "grad_norm": 15.855415344238281,
+ "learning_rate": 1.6439600363306086e-05,
+ "loss": 1.9873,
+ "step": 905
+ },
+ {
+ "epoch": 2.4686648501362396,
+ "grad_norm": 15.258964538574219,
+ "learning_rate": 1.6457765667574934e-05,
+ "loss": 2.0503,
+ "step": 906
+ },
+ {
+ "epoch": 2.4713896457765667,
+ "grad_norm": 15.970451354980469,
+ "learning_rate": 1.647593097184378e-05,
+ "loss": 2.1001,
+ "step": 907
+ },
+ {
+ "epoch": 2.474114441416894,
+ "grad_norm": 19.13205909729004,
+ "learning_rate": 1.6494096276112626e-05,
+ "loss": 2.4414,
+ "step": 908
+ },
+ {
+ "epoch": 2.4768392370572205,
+ "grad_norm": 21.848424911499023,
+ "learning_rate": 1.651226158038147e-05,
+ "loss": 2.3789,
+ "step": 909
+ },
+ {
+ "epoch": 2.4795640326975477,
+ "grad_norm": 9.960251808166504,
+ "learning_rate": 1.653042688465032e-05,
+ "loss": 1.8672,
+ "step": 910
+ },
+ {
+ "epoch": 2.482288828337875,
+ "grad_norm": 8.895149230957031,
+ "learning_rate": 1.6548592188919167e-05,
+ "loss": 1.7749,
+ "step": 911
+ },
+ {
+ "epoch": 2.4850136239782015,
+ "grad_norm": 25.562145233154297,
+ "learning_rate": 1.656675749318801e-05,
+ "loss": 2.2832,
+ "step": 912
+ },
+ {
+ "epoch": 2.4877384196185286,
+ "grad_norm": 12.549063682556152,
+ "learning_rate": 1.658492279745686e-05,
+ "loss": 2.1973,
+ "step": 913
+ },
+ {
+ "epoch": 2.4904632152588557,
+ "grad_norm": 12.085905075073242,
+ "learning_rate": 1.6603088101725707e-05,
+ "loss": 1.9858,
+ "step": 914
+ },
+ {
+ "epoch": 2.4931880108991824,
+ "grad_norm": 10.60605239868164,
+ "learning_rate": 1.662125340599455e-05,
+ "loss": 2.2271,
+ "step": 915
+ },
+ {
+ "epoch": 2.4959128065395095,
+ "grad_norm": 10.788206100463867,
+ "learning_rate": 1.66394187102634e-05,
+ "loss": 2.2041,
+ "step": 916
+ },
+ {
+ "epoch": 2.4986376021798367,
+ "grad_norm": 8.639183044433594,
+ "learning_rate": 1.6657584014532244e-05,
+ "loss": 1.9478,
+ "step": 917
+ },
+ {
+ "epoch": 2.5013623978201633,
+ "grad_norm": 11.193388938903809,
+ "learning_rate": 1.667574931880109e-05,
+ "loss": 2.0649,
+ "step": 918
+ },
+ {
+ "epoch": 2.5040871934604905,
+ "grad_norm": 9.984907150268555,
+ "learning_rate": 1.669391462306994e-05,
+ "loss": 1.8789,
+ "step": 919
+ },
+ {
+ "epoch": 2.5068119891008176,
+ "grad_norm": 8.873444557189941,
+ "learning_rate": 1.6712079927338784e-05,
+ "loss": 2.0762,
+ "step": 920
+ },
+ {
+ "epoch": 2.5095367847411443,
+ "grad_norm": 9.470831871032715,
+ "learning_rate": 1.673024523160763e-05,
+ "loss": 2.1216,
+ "step": 921
+ },
+ {
+ "epoch": 2.5122615803814714,
+ "grad_norm": 9.29857063293457,
+ "learning_rate": 1.674841053587648e-05,
+ "loss": 1.9434,
+ "step": 922
+ },
+ {
+ "epoch": 2.5149863760217985,
+ "grad_norm": 10.398829460144043,
+ "learning_rate": 1.6766575840145324e-05,
+ "loss": 1.8462,
+ "step": 923
+ },
+ {
+ "epoch": 2.517711171662125,
+ "grad_norm": 7.153753757476807,
+ "learning_rate": 1.678474114441417e-05,
+ "loss": 1.9458,
+ "step": 924
+ },
+ {
+ "epoch": 2.5204359673024523,
+ "grad_norm": 8.812695503234863,
+ "learning_rate": 1.6802906448683016e-05,
+ "loss": 1.8662,
+ "step": 925
+ },
+ {
+ "epoch": 2.523160762942779,
+ "grad_norm": 11.077635765075684,
+ "learning_rate": 1.6821071752951864e-05,
+ "loss": 2.0415,
+ "step": 926
+ },
+ {
+ "epoch": 2.525885558583106,
+ "grad_norm": 14.207860946655273,
+ "learning_rate": 1.683923705722071e-05,
+ "loss": 2.002,
+ "step": 927
+ },
+ {
+ "epoch": 2.5286103542234333,
+ "grad_norm": 17.286357879638672,
+ "learning_rate": 1.6857402361489556e-05,
+ "loss": 2.0718,
+ "step": 928
+ },
+ {
+ "epoch": 2.53133514986376,
+ "grad_norm": 10.753449440002441,
+ "learning_rate": 1.6875567665758404e-05,
+ "loss": 1.9053,
+ "step": 929
+ },
+ {
+ "epoch": 2.534059945504087,
+ "grad_norm": 11.433552742004395,
+ "learning_rate": 1.689373297002725e-05,
+ "loss": 2.1675,
+ "step": 930
+ },
+ {
+ "epoch": 2.536784741144414,
+ "grad_norm": 8.137096405029297,
+ "learning_rate": 1.6911898274296097e-05,
+ "loss": 1.9941,
+ "step": 931
+ },
+ {
+ "epoch": 2.539509536784741,
+ "grad_norm": 9.175704956054688,
+ "learning_rate": 1.693006357856494e-05,
+ "loss": 2.0288,
+ "step": 932
+ },
+ {
+ "epoch": 2.542234332425068,
+ "grad_norm": 12.608711242675781,
+ "learning_rate": 1.694822888283379e-05,
+ "loss": 2.0142,
+ "step": 933
+ },
+ {
+ "epoch": 2.544959128065395,
+ "grad_norm": 11.335437774658203,
+ "learning_rate": 1.6966394187102633e-05,
+ "loss": 2.2241,
+ "step": 934
+ },
+ {
+ "epoch": 2.547683923705722,
+ "grad_norm": 9.712677955627441,
+ "learning_rate": 1.698455949137148e-05,
+ "loss": 1.9741,
+ "step": 935
+ },
+ {
+ "epoch": 2.550408719346049,
+ "grad_norm": 9.112726211547852,
+ "learning_rate": 1.700272479564033e-05,
+ "loss": 2.0225,
+ "step": 936
+ },
+ {
+ "epoch": 2.553133514986376,
+ "grad_norm": 11.21507740020752,
+ "learning_rate": 1.7020890099909177e-05,
+ "loss": 1.8994,
+ "step": 937
+ },
+ {
+ "epoch": 2.5558583106267028,
+ "grad_norm": 14.425416946411133,
+ "learning_rate": 1.703905540417802e-05,
+ "loss": 2.2158,
+ "step": 938
+ },
+ {
+ "epoch": 2.55858310626703,
+ "grad_norm": 8.555197715759277,
+ "learning_rate": 1.705722070844687e-05,
+ "loss": 2.2285,
+ "step": 939
+ },
+ {
+ "epoch": 2.561307901907357,
+ "grad_norm": 8.4214448928833,
+ "learning_rate": 1.7075386012715714e-05,
+ "loss": 1.9604,
+ "step": 940
+ },
+ {
+ "epoch": 2.5640326975476837,
+ "grad_norm": 21.438478469848633,
+ "learning_rate": 1.709355131698456e-05,
+ "loss": 2.1079,
+ "step": 941
+ },
+ {
+ "epoch": 2.566757493188011,
+ "grad_norm": 8.835628509521484,
+ "learning_rate": 1.7111716621253406e-05,
+ "loss": 2.1118,
+ "step": 942
+ },
+ {
+ "epoch": 2.569482288828338,
+ "grad_norm": 9.615950584411621,
+ "learning_rate": 1.7129881925522254e-05,
+ "loss": 2.043,
+ "step": 943
+ },
+ {
+ "epoch": 2.5722070844686646,
+ "grad_norm": 15.055055618286133,
+ "learning_rate": 1.7148047229791102e-05,
+ "loss": 1.9893,
+ "step": 944
+ },
+ {
+ "epoch": 2.5749318801089918,
+ "grad_norm": 9.757176399230957,
+ "learning_rate": 1.7166212534059946e-05,
+ "loss": 1.9995,
+ "step": 945
+ },
+ {
+ "epoch": 2.577656675749319,
+ "grad_norm": 9.33747673034668,
+ "learning_rate": 1.7184377838328794e-05,
+ "loss": 1.7993,
+ "step": 946
+ },
+ {
+ "epoch": 2.5803814713896456,
+ "grad_norm": 14.770505905151367,
+ "learning_rate": 1.7202543142597642e-05,
+ "loss": 2.0537,
+ "step": 947
+ },
+ {
+ "epoch": 2.5831062670299727,
+ "grad_norm": 10.552059173583984,
+ "learning_rate": 1.7220708446866486e-05,
+ "loss": 1.7827,
+ "step": 948
+ },
+ {
+ "epoch": 2.5858310626703,
+ "grad_norm": 12.206165313720703,
+ "learning_rate": 1.723887375113533e-05,
+ "loss": 1.8933,
+ "step": 949
+ },
+ {
+ "epoch": 2.5885558583106265,
+ "grad_norm": 10.038641929626465,
+ "learning_rate": 1.725703905540418e-05,
+ "loss": 1.916,
+ "step": 950
+ },
+ {
+ "epoch": 2.5912806539509536,
+ "grad_norm": 11.65165901184082,
+ "learning_rate": 1.7275204359673027e-05,
+ "loss": 2.0186,
+ "step": 951
+ },
+ {
+ "epoch": 2.5940054495912808,
+ "grad_norm": 19.512056350708008,
+ "learning_rate": 1.729336966394187e-05,
+ "loss": 2.0605,
+ "step": 952
+ },
+ {
+ "epoch": 2.5967302452316074,
+ "grad_norm": 9.305648803710938,
+ "learning_rate": 1.731153496821072e-05,
+ "loss": 1.9097,
+ "step": 953
+ },
+ {
+ "epoch": 2.5994550408719346,
+ "grad_norm": 9.158435821533203,
+ "learning_rate": 1.7329700272479567e-05,
+ "loss": 2.1626,
+ "step": 954
+ },
+ {
+ "epoch": 2.6021798365122617,
+ "grad_norm": 12.203804969787598,
+ "learning_rate": 1.734786557674841e-05,
+ "loss": 2.1958,
+ "step": 955
+ },
+ {
+ "epoch": 2.6049046321525884,
+ "grad_norm": 16.611064910888672,
+ "learning_rate": 1.736603088101726e-05,
+ "loss": 2.3867,
+ "step": 956
+ },
+ {
+ "epoch": 2.6076294277929155,
+ "grad_norm": 11.887727737426758,
+ "learning_rate": 1.7384196185286104e-05,
+ "loss": 2.082,
+ "step": 957
+ },
+ {
+ "epoch": 2.6103542234332426,
+ "grad_norm": 9.6937837600708,
+ "learning_rate": 1.740236148955495e-05,
+ "loss": 1.8745,
+ "step": 958
+ },
+ {
+ "epoch": 2.6130790190735693,
+ "grad_norm": 15.233810424804688,
+ "learning_rate": 1.74205267938238e-05,
+ "loss": 1.9136,
+ "step": 959
+ },
+ {
+ "epoch": 2.6158038147138964,
+ "grad_norm": 11.430521965026855,
+ "learning_rate": 1.7438692098092644e-05,
+ "loss": 2.1313,
+ "step": 960
+ },
+ {
+ "epoch": 2.6185286103542236,
+ "grad_norm": 10.866662979125977,
+ "learning_rate": 1.745685740236149e-05,
+ "loss": 2.0684,
+ "step": 961
+ },
+ {
+ "epoch": 2.6212534059945503,
+ "grad_norm": 9.40779972076416,
+ "learning_rate": 1.747502270663034e-05,
+ "loss": 1.9028,
+ "step": 962
+ },
+ {
+ "epoch": 2.6239782016348774,
+ "grad_norm": 21.536073684692383,
+ "learning_rate": 1.7493188010899184e-05,
+ "loss": 1.9629,
+ "step": 963
+ },
+ {
+ "epoch": 2.6267029972752045,
+ "grad_norm": 11.291322708129883,
+ "learning_rate": 1.7511353315168032e-05,
+ "loss": 1.8857,
+ "step": 964
+ },
+ {
+ "epoch": 2.629427792915531,
+ "grad_norm": 11.789603233337402,
+ "learning_rate": 1.7529518619436876e-05,
+ "loss": 2.1143,
+ "step": 965
+ },
+ {
+ "epoch": 2.6321525885558583,
+ "grad_norm": 10.198453903198242,
+ "learning_rate": 1.7547683923705724e-05,
+ "loss": 1.9321,
+ "step": 966
+ },
+ {
+ "epoch": 2.6348773841961854,
+ "grad_norm": 10.767601013183594,
+ "learning_rate": 1.756584922797457e-05,
+ "loss": 2.0391,
+ "step": 967
+ },
+ {
+ "epoch": 2.637602179836512,
+ "grad_norm": 14.731465339660645,
+ "learning_rate": 1.7584014532243416e-05,
+ "loss": 2.0195,
+ "step": 968
+ },
+ {
+ "epoch": 2.6403269754768393,
+ "grad_norm": 9.702617645263672,
+ "learning_rate": 1.7602179836512264e-05,
+ "loss": 1.9238,
+ "step": 969
+ },
+ {
+ "epoch": 2.6430517711171664,
+ "grad_norm": 26.57699203491211,
+ "learning_rate": 1.762034514078111e-05,
+ "loss": 2.1631,
+ "step": 970
+ },
+ {
+ "epoch": 2.645776566757493,
+ "grad_norm": 18.070432662963867,
+ "learning_rate": 1.7638510445049957e-05,
+ "loss": 2.0752,
+ "step": 971
+ },
+ {
+ "epoch": 2.64850136239782,
+ "grad_norm": 10.054424285888672,
+ "learning_rate": 1.7656675749318804e-05,
+ "loss": 1.9517,
+ "step": 972
+ },
+ {
+ "epoch": 2.6512261580381473,
+ "grad_norm": 11.254864692687988,
+ "learning_rate": 1.767484105358765e-05,
+ "loss": 1.8672,
+ "step": 973
+ },
+ {
+ "epoch": 2.653950953678474,
+ "grad_norm": 11.593107223510742,
+ "learning_rate": 1.7693006357856493e-05,
+ "loss": 2.02,
+ "step": 974
+ },
+ {
+ "epoch": 2.656675749318801,
+ "grad_norm": 12.124439239501953,
+ "learning_rate": 1.771117166212534e-05,
+ "loss": 2.0215,
+ "step": 975
+ },
+ {
+ "epoch": 2.6594005449591283,
+ "grad_norm": 12.281224250793457,
+ "learning_rate": 1.772933696639419e-05,
+ "loss": 2.1313,
+ "step": 976
+ },
+ {
+ "epoch": 2.662125340599455,
+ "grad_norm": 10.418290138244629,
+ "learning_rate": 1.7747502270663034e-05,
+ "loss": 1.9077,
+ "step": 977
+ },
+ {
+ "epoch": 2.664850136239782,
+ "grad_norm": 11.25628662109375,
+ "learning_rate": 1.776566757493188e-05,
+ "loss": 1.8809,
+ "step": 978
+ },
+ {
+ "epoch": 2.667574931880109,
+ "grad_norm": 12.278833389282227,
+ "learning_rate": 1.778383287920073e-05,
+ "loss": 2.3364,
+ "step": 979
+ },
+ {
+ "epoch": 2.670299727520436,
+ "grad_norm": 9.75459098815918,
+ "learning_rate": 1.7801998183469574e-05,
+ "loss": 2.0024,
+ "step": 980
+ },
+ {
+ "epoch": 2.673024523160763,
+ "grad_norm": 9.556586265563965,
+ "learning_rate": 1.782016348773842e-05,
+ "loss": 1.9697,
+ "step": 981
+ },
+ {
+ "epoch": 2.67574931880109,
+ "grad_norm": 8.527470588684082,
+ "learning_rate": 1.7838328792007266e-05,
+ "loss": 1.9014,
+ "step": 982
+ },
+ {
+ "epoch": 2.678474114441417,
+ "grad_norm": 10.903023719787598,
+ "learning_rate": 1.7856494096276114e-05,
+ "loss": 1.9937,
+ "step": 983
+ },
+ {
+ "epoch": 2.681198910081744,
+ "grad_norm": 11.214929580688477,
+ "learning_rate": 1.7874659400544962e-05,
+ "loss": 2.2075,
+ "step": 984
+ },
+ {
+ "epoch": 2.683923705722071,
+ "grad_norm": 9.724494934082031,
+ "learning_rate": 1.7892824704813806e-05,
+ "loss": 1.811,
+ "step": 985
+ },
+ {
+ "epoch": 2.6866485013623977,
+ "grad_norm": 9.260843276977539,
+ "learning_rate": 1.7910990009082654e-05,
+ "loss": 1.9224,
+ "step": 986
+ },
+ {
+ "epoch": 2.689373297002725,
+ "grad_norm": 11.609862327575684,
+ "learning_rate": 1.7929155313351502e-05,
+ "loss": 2.1343,
+ "step": 987
+ },
+ {
+ "epoch": 2.692098092643052,
+ "grad_norm": 9.972542762756348,
+ "learning_rate": 1.7947320617620346e-05,
+ "loss": 1.9648,
+ "step": 988
+ },
+ {
+ "epoch": 2.6948228882833787,
+ "grad_norm": 8.748735427856445,
+ "learning_rate": 1.7965485921889194e-05,
+ "loss": 1.8608,
+ "step": 989
+ },
+ {
+ "epoch": 2.697547683923706,
+ "grad_norm": 10.71142864227295,
+ "learning_rate": 1.798365122615804e-05,
+ "loss": 1.938,
+ "step": 990
+ },
+ {
+ "epoch": 2.700272479564033,
+ "grad_norm": 9.90158462524414,
+ "learning_rate": 1.8001816530426887e-05,
+ "loss": 2.0342,
+ "step": 991
+ },
+ {
+ "epoch": 2.7029972752043596,
+ "grad_norm": 8.216854095458984,
+ "learning_rate": 1.801998183469573e-05,
+ "loss": 2.3257,
+ "step": 992
+ },
+ {
+ "epoch": 2.7057220708446867,
+ "grad_norm": 11.613279342651367,
+ "learning_rate": 1.803814713896458e-05,
+ "loss": 2.2539,
+ "step": 993
+ },
+ {
+ "epoch": 2.708446866485014,
+ "grad_norm": 9.236078262329102,
+ "learning_rate": 1.8056312443233427e-05,
+ "loss": 2.0449,
+ "step": 994
+ },
+ {
+ "epoch": 2.7111716621253406,
+ "grad_norm": 8.92188835144043,
+ "learning_rate": 1.807447774750227e-05,
+ "loss": 1.8311,
+ "step": 995
+ },
+ {
+ "epoch": 2.7138964577656677,
+ "grad_norm": 9.407740592956543,
+ "learning_rate": 1.809264305177112e-05,
+ "loss": 2.1436,
+ "step": 996
+ },
+ {
+ "epoch": 2.716621253405995,
+ "grad_norm": 8.75705337524414,
+ "learning_rate": 1.8110808356039967e-05,
+ "loss": 2.0613,
+ "step": 997
+ },
+ {
+ "epoch": 2.7193460490463215,
+ "grad_norm": 8.363595962524414,
+ "learning_rate": 1.812897366030881e-05,
+ "loss": 1.8091,
+ "step": 998
+ },
+ {
+ "epoch": 2.7220708446866486,
+ "grad_norm": 16.87444496154785,
+ "learning_rate": 1.8147138964577656e-05,
+ "loss": 2.1694,
+ "step": 999
+ },
+ {
+ "epoch": 2.7247956403269757,
+ "grad_norm": 9.609050750732422,
+ "learning_rate": 1.8165304268846504e-05,
+ "loss": 1.8794,
+ "step": 1000
+ },
+ {
+ "epoch": 2.7275204359673024,
+ "grad_norm": 11.181875228881836,
+ "learning_rate": 1.818346957311535e-05,
+ "loss": 2.0537,
+ "step": 1001
+ },
+ {
+ "epoch": 2.7302452316076296,
+ "grad_norm": 7.988255500793457,
+ "learning_rate": 1.82016348773842e-05,
+ "loss": 1.8022,
+ "step": 1002
+ },
+ {
+ "epoch": 2.7329700272479567,
+ "grad_norm": 8.757339477539062,
+ "learning_rate": 1.8219800181653044e-05,
+ "loss": 1.9097,
+ "step": 1003
+ },
+ {
+ "epoch": 2.7356948228882834,
+ "grad_norm": 10.019728660583496,
+ "learning_rate": 1.8237965485921892e-05,
+ "loss": 2.1758,
+ "step": 1004
+ },
+ {
+ "epoch": 2.7384196185286105,
+ "grad_norm": 11.55337142944336,
+ "learning_rate": 1.825613079019074e-05,
+ "loss": 2.0527,
+ "step": 1005
+ },
+ {
+ "epoch": 2.741144414168937,
+ "grad_norm": 8.637880325317383,
+ "learning_rate": 1.8274296094459584e-05,
+ "loss": 1.7627,
+ "step": 1006
+ },
+ {
+ "epoch": 2.7438692098092643,
+ "grad_norm": 10.108887672424316,
+ "learning_rate": 1.829246139872843e-05,
+ "loss": 1.9849,
+ "step": 1007
+ },
+ {
+ "epoch": 2.7465940054495914,
+ "grad_norm": 8.768725395202637,
+ "learning_rate": 1.8310626702997276e-05,
+ "loss": 1.8774,
+ "step": 1008
+ },
+ {
+ "epoch": 2.749318801089918,
+ "grad_norm": 8.873408317565918,
+ "learning_rate": 1.8328792007266124e-05,
+ "loss": 2.0957,
+ "step": 1009
+ },
+ {
+ "epoch": 2.7520435967302452,
+ "grad_norm": 10.233674049377441,
+ "learning_rate": 1.834695731153497e-05,
+ "loss": 2.125,
+ "step": 1010
+ },
+ {
+ "epoch": 2.7547683923705724,
+ "grad_norm": 9.679442405700684,
+ "learning_rate": 1.8365122615803817e-05,
+ "loss": 2.1631,
+ "step": 1011
+ },
+ {
+ "epoch": 2.757493188010899,
+ "grad_norm": 9.121768951416016,
+ "learning_rate": 1.8383287920072664e-05,
+ "loss": 2.0449,
+ "step": 1012
+ },
+ {
+ "epoch": 2.760217983651226,
+ "grad_norm": 10.679054260253906,
+ "learning_rate": 1.840145322434151e-05,
+ "loss": 1.8528,
+ "step": 1013
+ },
+ {
+ "epoch": 2.7629427792915533,
+ "grad_norm": 11.020075798034668,
+ "learning_rate": 1.8419618528610357e-05,
+ "loss": 2.0688,
+ "step": 1014
+ },
+ {
+ "epoch": 2.76566757493188,
+ "grad_norm": 8.46170425415039,
+ "learning_rate": 1.84377838328792e-05,
+ "loss": 1.9302,
+ "step": 1015
+ },
+ {
+ "epoch": 2.768392370572207,
+ "grad_norm": 14.452118873596191,
+ "learning_rate": 1.845594913714805e-05,
+ "loss": 2.2603,
+ "step": 1016
+ },
+ {
+ "epoch": 2.771117166212534,
+ "grad_norm": 10.289412498474121,
+ "learning_rate": 1.8474114441416894e-05,
+ "loss": 1.9512,
+ "step": 1017
+ },
+ {
+ "epoch": 2.773841961852861,
+ "grad_norm": 9.330587387084961,
+ "learning_rate": 1.849227974568574e-05,
+ "loss": 2.0229,
+ "step": 1018
+ },
+ {
+ "epoch": 2.776566757493188,
+ "grad_norm": 14.548686027526855,
+ "learning_rate": 1.851044504995459e-05,
+ "loss": 1.9165,
+ "step": 1019
+ },
+ {
+ "epoch": 2.7792915531335147,
+ "grad_norm": 11.193462371826172,
+ "learning_rate": 1.8528610354223434e-05,
+ "loss": 1.9673,
+ "step": 1020
+ },
+ {
+ "epoch": 2.782016348773842,
+ "grad_norm": 15.824004173278809,
+ "learning_rate": 1.854677565849228e-05,
+ "loss": 1.9995,
+ "step": 1021
+ },
+ {
+ "epoch": 2.784741144414169,
+ "grad_norm": 8.91195011138916,
+ "learning_rate": 1.856494096276113e-05,
+ "loss": 2.04,
+ "step": 1022
+ },
+ {
+ "epoch": 2.7874659400544957,
+ "grad_norm": 8.997485160827637,
+ "learning_rate": 1.8583106267029974e-05,
+ "loss": 1.8086,
+ "step": 1023
+ },
+ {
+ "epoch": 2.790190735694823,
+ "grad_norm": 13.889126777648926,
+ "learning_rate": 1.8601271571298822e-05,
+ "loss": 1.9194,
+ "step": 1024
+ },
+ {
+ "epoch": 2.79291553133515,
+ "grad_norm": 10.392932891845703,
+ "learning_rate": 1.8619436875567666e-05,
+ "loss": 1.9883,
+ "step": 1025
+ },
+ {
+ "epoch": 2.7956403269754766,
+ "grad_norm": 8.134140014648438,
+ "learning_rate": 1.8637602179836514e-05,
+ "loss": 1.8984,
+ "step": 1026
+ },
+ {
+ "epoch": 2.7983651226158037,
+ "grad_norm": 9.642759323120117,
+ "learning_rate": 1.8655767484105362e-05,
+ "loss": 1.9663,
+ "step": 1027
+ },
+ {
+ "epoch": 2.801089918256131,
+ "grad_norm": 11.782906532287598,
+ "learning_rate": 1.8673932788374206e-05,
+ "loss": 1.9062,
+ "step": 1028
+ },
+ {
+ "epoch": 2.8038147138964575,
+ "grad_norm": 10.656364440917969,
+ "learning_rate": 1.8692098092643054e-05,
+ "loss": 1.8872,
+ "step": 1029
+ },
+ {
+ "epoch": 2.8065395095367847,
+ "grad_norm": 10.768020629882812,
+ "learning_rate": 1.8710263396911902e-05,
+ "loss": 1.9004,
+ "step": 1030
+ },
+ {
+ "epoch": 2.809264305177112,
+ "grad_norm": 13.038018226623535,
+ "learning_rate": 1.8728428701180747e-05,
+ "loss": 1.9038,
+ "step": 1031
+ },
+ {
+ "epoch": 2.8119891008174385,
+ "grad_norm": 9.458123207092285,
+ "learning_rate": 1.874659400544959e-05,
+ "loss": 1.9724,
+ "step": 1032
+ },
+ {
+ "epoch": 2.8147138964577656,
+ "grad_norm": 11.42751407623291,
+ "learning_rate": 1.876475930971844e-05,
+ "loss": 2.0469,
+ "step": 1033
+ },
+ {
+ "epoch": 2.8174386920980927,
+ "grad_norm": 8.140854835510254,
+ "learning_rate": 1.8782924613987287e-05,
+ "loss": 1.8706,
+ "step": 1034
+ },
+ {
+ "epoch": 2.8201634877384194,
+ "grad_norm": 10.019620895385742,
+ "learning_rate": 1.880108991825613e-05,
+ "loss": 2.145,
+ "step": 1035
+ },
+ {
+ "epoch": 2.8228882833787465,
+ "grad_norm": 11.080718994140625,
+ "learning_rate": 1.881925522252498e-05,
+ "loss": 2.0464,
+ "step": 1036
+ },
+ {
+ "epoch": 2.8256130790190737,
+ "grad_norm": 9.135918617248535,
+ "learning_rate": 1.8837420526793827e-05,
+ "loss": 2.1357,
+ "step": 1037
+ },
+ {
+ "epoch": 2.8283378746594003,
+ "grad_norm": 8.98420524597168,
+ "learning_rate": 1.885558583106267e-05,
+ "loss": 2.0977,
+ "step": 1038
+ },
+ {
+ "epoch": 2.8310626702997275,
+ "grad_norm": 8.704095840454102,
+ "learning_rate": 1.8873751135331516e-05,
+ "loss": 1.8442,
+ "step": 1039
+ },
+ {
+ "epoch": 2.8337874659400546,
+ "grad_norm": 10.60055160522461,
+ "learning_rate": 1.8891916439600364e-05,
+ "loss": 1.9695,
+ "step": 1040
+ },
+ {
+ "epoch": 2.8365122615803813,
+ "grad_norm": 7.9067583084106445,
+ "learning_rate": 1.891008174386921e-05,
+ "loss": 1.8496,
+ "step": 1041
+ },
+ {
+ "epoch": 2.8392370572207084,
+ "grad_norm": 9.128850936889648,
+ "learning_rate": 1.8928247048138056e-05,
+ "loss": 1.77,
+ "step": 1042
+ },
+ {
+ "epoch": 2.8419618528610355,
+ "grad_norm": 12.708388328552246,
+ "learning_rate": 1.8946412352406904e-05,
+ "loss": 2.0347,
+ "step": 1043
+ },
+ {
+ "epoch": 2.844686648501362,
+ "grad_norm": 9.919172286987305,
+ "learning_rate": 1.8964577656675752e-05,
+ "loss": 2.0386,
+ "step": 1044
+ },
+ {
+ "epoch": 2.8474114441416893,
+ "grad_norm": 17.127466201782227,
+ "learning_rate": 1.8982742960944596e-05,
+ "loss": 2.0281,
+ "step": 1045
+ },
+ {
+ "epoch": 2.8501362397820165,
+ "grad_norm": 9.493831634521484,
+ "learning_rate": 1.9000908265213444e-05,
+ "loss": 1.7896,
+ "step": 1046
+ },
+ {
+ "epoch": 2.852861035422343,
+ "grad_norm": 10.712773323059082,
+ "learning_rate": 1.9019073569482292e-05,
+ "loss": 1.7253,
+ "step": 1047
+ },
+ {
+ "epoch": 2.8555858310626703,
+ "grad_norm": 8.9692964553833,
+ "learning_rate": 1.9037238873751136e-05,
+ "loss": 1.9565,
+ "step": 1048
+ },
+ {
+ "epoch": 2.8583106267029974,
+ "grad_norm": 10.709541320800781,
+ "learning_rate": 1.9055404178019984e-05,
+ "loss": 1.8325,
+ "step": 1049
+ },
+ {
+ "epoch": 2.861035422343324,
+ "grad_norm": 9.731253623962402,
+ "learning_rate": 1.907356948228883e-05,
+ "loss": 2.0957,
+ "step": 1050
+ },
+ {
+ "epoch": 2.863760217983651,
+ "grad_norm": 8.981160163879395,
+ "learning_rate": 1.9091734786557677e-05,
+ "loss": 1.9331,
+ "step": 1051
+ },
+ {
+ "epoch": 2.8664850136239783,
+ "grad_norm": 9.39439868927002,
+ "learning_rate": 1.9109900090826524e-05,
+ "loss": 1.8765,
+ "step": 1052
+ },
+ {
+ "epoch": 2.869209809264305,
+ "grad_norm": 12.431492805480957,
+ "learning_rate": 1.912806539509537e-05,
+ "loss": 2.104,
+ "step": 1053
+ },
+ {
+ "epoch": 2.871934604904632,
+ "grad_norm": 13.97305679321289,
+ "learning_rate": 1.9146230699364217e-05,
+ "loss": 1.9526,
+ "step": 1054
+ },
+ {
+ "epoch": 2.8746594005449593,
+ "grad_norm": 9.368614196777344,
+ "learning_rate": 1.9164396003633065e-05,
+ "loss": 2.2915,
+ "step": 1055
+ },
+ {
+ "epoch": 2.877384196185286,
+ "grad_norm": 8.295093536376953,
+ "learning_rate": 1.918256130790191e-05,
+ "loss": 1.8481,
+ "step": 1056
+ },
+ {
+ "epoch": 2.880108991825613,
+ "grad_norm": 7.692062854766846,
+ "learning_rate": 1.9200726612170754e-05,
+ "loss": 1.9199,
+ "step": 1057
+ },
+ {
+ "epoch": 2.88283378746594,
+ "grad_norm": 8.370375633239746,
+ "learning_rate": 1.92188919164396e-05,
+ "loss": 1.9951,
+ "step": 1058
+ },
+ {
+ "epoch": 2.885558583106267,
+ "grad_norm": 9.188200950622559,
+ "learning_rate": 1.923705722070845e-05,
+ "loss": 1.9009,
+ "step": 1059
+ },
+ {
+ "epoch": 2.888283378746594,
+ "grad_norm": 7.3844895362854,
+ "learning_rate": 1.9255222524977294e-05,
+ "loss": 1.9185,
+ "step": 1060
+ },
+ {
+ "epoch": 2.891008174386921,
+ "grad_norm": 8.950010299682617,
+ "learning_rate": 1.927338782924614e-05,
+ "loss": 2.0215,
+ "step": 1061
+ },
+ {
+ "epoch": 2.893732970027248,
+ "grad_norm": 58.49240493774414,
+ "learning_rate": 1.929155313351499e-05,
+ "loss": 2.0337,
+ "step": 1062
+ },
+ {
+ "epoch": 2.896457765667575,
+ "grad_norm": 7.721260070800781,
+ "learning_rate": 1.9309718437783834e-05,
+ "loss": 1.9102,
+ "step": 1063
+ },
+ {
+ "epoch": 2.899182561307902,
+ "grad_norm": 9.676554679870605,
+ "learning_rate": 1.932788374205268e-05,
+ "loss": 1.9795,
+ "step": 1064
+ },
+ {
+ "epoch": 2.9019073569482288,
+ "grad_norm": 8.46129322052002,
+ "learning_rate": 1.9346049046321526e-05,
+ "loss": 1.7671,
+ "step": 1065
+ },
+ {
+ "epoch": 2.904632152588556,
+ "grad_norm": 7.848261833190918,
+ "learning_rate": 1.9364214350590374e-05,
+ "loss": 2.0371,
+ "step": 1066
+ },
+ {
+ "epoch": 2.907356948228883,
+ "grad_norm": 9.203500747680664,
+ "learning_rate": 1.938237965485922e-05,
+ "loss": 1.8848,
+ "step": 1067
+ },
+ {
+ "epoch": 2.9100817438692097,
+ "grad_norm": 7.942507266998291,
+ "learning_rate": 1.9400544959128066e-05,
+ "loss": 1.8438,
+ "step": 1068
+ },
+ {
+ "epoch": 2.912806539509537,
+ "grad_norm": 7.777604579925537,
+ "learning_rate": 1.9418710263396914e-05,
+ "loss": 1.6978,
+ "step": 1069
+ },
+ {
+ "epoch": 2.915531335149864,
+ "grad_norm": 8.911617279052734,
+ "learning_rate": 1.9436875567665762e-05,
+ "loss": 1.8291,
+ "step": 1070
+ },
+ {
+ "epoch": 2.9182561307901906,
+ "grad_norm": 8.319960594177246,
+ "learning_rate": 1.9455040871934607e-05,
+ "loss": 1.9961,
+ "step": 1071
+ },
+ {
+ "epoch": 2.9209809264305178,
+ "grad_norm": 11.947103500366211,
+ "learning_rate": 1.9473206176203454e-05,
+ "loss": 1.9973,
+ "step": 1072
+ },
+ {
+ "epoch": 2.923705722070845,
+ "grad_norm": 9.745553016662598,
+ "learning_rate": 1.94913714804723e-05,
+ "loss": 2.1323,
+ "step": 1073
+ },
+ {
+ "epoch": 2.9264305177111716,
+ "grad_norm": 8.263842582702637,
+ "learning_rate": 1.9509536784741147e-05,
+ "loss": 1.8572,
+ "step": 1074
+ },
+ {
+ "epoch": 2.9291553133514987,
+ "grad_norm": 10.004130363464355,
+ "learning_rate": 1.952770208900999e-05,
+ "loss": 2.0366,
+ "step": 1075
+ },
+ {
+ "epoch": 2.931880108991826,
+ "grad_norm": 10.774187088012695,
+ "learning_rate": 1.954586739327884e-05,
+ "loss": 1.8882,
+ "step": 1076
+ },
+ {
+ "epoch": 2.9346049046321525,
+ "grad_norm": 9.680246353149414,
+ "learning_rate": 1.9564032697547687e-05,
+ "loss": 2.1313,
+ "step": 1077
+ },
+ {
+ "epoch": 2.9373297002724796,
+ "grad_norm": 10.95883560180664,
+ "learning_rate": 1.958219800181653e-05,
+ "loss": 1.936,
+ "step": 1078
+ },
+ {
+ "epoch": 2.9400544959128068,
+ "grad_norm": 8.73660659790039,
+ "learning_rate": 1.960036330608538e-05,
+ "loss": 1.7588,
+ "step": 1079
+ },
+ {
+ "epoch": 2.9427792915531334,
+ "grad_norm": 13.1550931930542,
+ "learning_rate": 1.9618528610354227e-05,
+ "loss": 1.9736,
+ "step": 1080
+ },
+ {
+ "epoch": 2.9455040871934606,
+ "grad_norm": 8.755561828613281,
+ "learning_rate": 1.963669391462307e-05,
+ "loss": 1.9453,
+ "step": 1081
+ },
+ {
+ "epoch": 2.9482288828337877,
+ "grad_norm": 9.163195610046387,
+ "learning_rate": 1.9654859218891916e-05,
+ "loss": 1.9976,
+ "step": 1082
+ },
+ {
+ "epoch": 2.9509536784741144,
+ "grad_norm": 10.460067749023438,
+ "learning_rate": 1.9673024523160764e-05,
+ "loss": 2.0288,
+ "step": 1083
+ },
+ {
+ "epoch": 2.9536784741144415,
+ "grad_norm": 10.246499061584473,
+ "learning_rate": 1.9691189827429612e-05,
+ "loss": 1.9331,
+ "step": 1084
+ },
+ {
+ "epoch": 2.9564032697547686,
+ "grad_norm": 9.747570037841797,
+ "learning_rate": 1.9709355131698456e-05,
+ "loss": 1.9927,
+ "step": 1085
+ },
+ {
+ "epoch": 2.9591280653950953,
+ "grad_norm": 7.422813415527344,
+ "learning_rate": 1.9727520435967304e-05,
+ "loss": 1.7336,
+ "step": 1086
+ },
+ {
+ "epoch": 2.9618528610354224,
+ "grad_norm": 8.913872718811035,
+ "learning_rate": 1.9745685740236152e-05,
+ "loss": 1.7925,
+ "step": 1087
+ },
+ {
+ "epoch": 2.9645776566757496,
+ "grad_norm": 8.548074722290039,
+ "learning_rate": 1.9763851044504996e-05,
+ "loss": 1.9819,
+ "step": 1088
+ },
+ {
+ "epoch": 2.9673024523160763,
+ "grad_norm": 8.673812866210938,
+ "learning_rate": 1.978201634877384e-05,
+ "loss": 1.957,
+ "step": 1089
+ },
+ {
+ "epoch": 2.9700272479564034,
+ "grad_norm": 8.26024055480957,
+ "learning_rate": 1.980018165304269e-05,
+ "loss": 2.1763,
+ "step": 1090
+ },
+ {
+ "epoch": 2.9727520435967305,
+ "grad_norm": 9.876177787780762,
+ "learning_rate": 1.9818346957311537e-05,
+ "loss": 1.8667,
+ "step": 1091
+ },
+ {
+ "epoch": 2.975476839237057,
+ "grad_norm": 9.82155704498291,
+ "learning_rate": 1.9836512261580384e-05,
+ "loss": 1.8447,
+ "step": 1092
+ },
+ {
+ "epoch": 2.9782016348773843,
+ "grad_norm": 8.930692672729492,
+ "learning_rate": 1.985467756584923e-05,
+ "loss": 2.2339,
+ "step": 1093
+ },
+ {
+ "epoch": 2.9809264305177114,
+ "grad_norm": 9.736824989318848,
+ "learning_rate": 1.9872842870118077e-05,
+ "loss": 1.9065,
+ "step": 1094
+ },
+ {
+ "epoch": 2.983651226158038,
+ "grad_norm": 8.123510360717773,
+ "learning_rate": 1.9891008174386925e-05,
+ "loss": 1.7866,
+ "step": 1095
+ },
+ {
+ "epoch": 2.9863760217983653,
+ "grad_norm": 10.167742729187012,
+ "learning_rate": 1.990917347865577e-05,
+ "loss": 1.8833,
+ "step": 1096
+ },
+ {
+ "epoch": 2.989100817438692,
+ "grad_norm": 12.495636940002441,
+ "learning_rate": 1.9927338782924614e-05,
+ "loss": 1.8958,
+ "step": 1097
+ },
+ {
+ "epoch": 2.991825613079019,
+ "grad_norm": 7.741083145141602,
+ "learning_rate": 1.994550408719346e-05,
+ "loss": 1.9985,
+ "step": 1098
+ },
+ {
+ "epoch": 2.994550408719346,
+ "grad_norm": 8.489039421081543,
+ "learning_rate": 1.996366939146231e-05,
+ "loss": 1.8335,
+ "step": 1099
+ },
+ {
+ "epoch": 2.997275204359673,
+ "grad_norm": 10.809964179992676,
+ "learning_rate": 1.9981834695731154e-05,
+ "loss": 2.1426,
+ "step": 1100
+ },
+ {
+ "epoch": 3.0,
+ "grad_norm": 7.268111228942871,
+ "learning_rate": 2e-05,
+ "loss": 1.9604,
+ "step": 1101
+ },
+ {
+ "epoch": 3.002724795640327,
+ "grad_norm": 9.166092872619629,
+ "learning_rate": 1.9999999961060162e-05,
+ "loss": 1.7344,
+ "step": 1102
+ },
+ {
+ "epoch": 3.005449591280654,
+ "grad_norm": 7.744848728179932,
+ "learning_rate": 1.9999999844240644e-05,
+ "loss": 2.1106,
+ "step": 1103
+ },
+ {
+ "epoch": 3.008174386920981,
+ "grad_norm": 9.175067901611328,
+ "learning_rate": 1.999999964954145e-05,
+ "loss": 1.8164,
+ "step": 1104
+ },
+ {
+ "epoch": 3.010899182561308,
+ "grad_norm": 8.082588195800781,
+ "learning_rate": 1.9999999376962583e-05,
+ "loss": 2.0142,
+ "step": 1105
+ },
+ {
+ "epoch": 3.0136239782016347,
+ "grad_norm": 7.535369873046875,
+ "learning_rate": 1.999999902650404e-05,
+ "loss": 1.8682,
+ "step": 1106
+ },
+ {
+ "epoch": 3.016348773841962,
+ "grad_norm": 11.99268627166748,
+ "learning_rate": 1.9999998598165827e-05,
+ "loss": 1.7817,
+ "step": 1107
+ },
+ {
+ "epoch": 3.019073569482289,
+ "grad_norm": 13.754222869873047,
+ "learning_rate": 1.9999998091947947e-05,
+ "loss": 2.0571,
+ "step": 1108
+ },
+ {
+ "epoch": 3.0217983651226157,
+ "grad_norm": 9.36998462677002,
+ "learning_rate": 1.9999997507850404e-05,
+ "loss": 1.7712,
+ "step": 1109
+ },
+ {
+ "epoch": 3.024523160762943,
+ "grad_norm": 15.121356964111328,
+ "learning_rate": 1.99999968458732e-05,
+ "loss": 1.751,
+ "step": 1110
+ },
+ {
+ "epoch": 3.02724795640327,
+ "grad_norm": 9.549973487854004,
+ "learning_rate": 1.9999996106016345e-05,
+ "loss": 1.9609,
+ "step": 1111
+ },
+ {
+ "epoch": 3.0299727520435966,
+ "grad_norm": 7.955581188201904,
+ "learning_rate": 1.999999528827984e-05,
+ "loss": 1.8018,
+ "step": 1112
+ },
+ {
+ "epoch": 3.0326975476839237,
+ "grad_norm": 9.091797828674316,
+ "learning_rate": 1.9999994392663697e-05,
+ "loss": 1.8252,
+ "step": 1113
+ },
+ {
+ "epoch": 3.035422343324251,
+ "grad_norm": 8.390670776367188,
+ "learning_rate": 1.9999993419167914e-05,
+ "loss": 1.8496,
+ "step": 1114
+ },
+ {
+ "epoch": 3.0381471389645776,
+ "grad_norm": 8.406384468078613,
+ "learning_rate": 1.999999236779251e-05,
+ "loss": 1.8772,
+ "step": 1115
+ },
+ {
+ "epoch": 3.0408719346049047,
+ "grad_norm": 9.405580520629883,
+ "learning_rate": 1.9999991238537482e-05,
+ "loss": 1.9546,
+ "step": 1116
+ },
+ {
+ "epoch": 3.043596730245232,
+ "grad_norm": 9.658448219299316,
+ "learning_rate": 1.999999003140285e-05,
+ "loss": 1.6912,
+ "step": 1117
+ },
+ {
+ "epoch": 3.0463215258855585,
+ "grad_norm": 7.176756858825684,
+ "learning_rate": 1.9999988746388613e-05,
+ "loss": 1.7378,
+ "step": 1118
+ },
+ {
+ "epoch": 3.0490463215258856,
+ "grad_norm": 107.75994110107422,
+ "learning_rate": 1.9999987383494786e-05,
+ "loss": 1.7202,
+ "step": 1119
+ },
+ {
+ "epoch": 3.0517711171662127,
+ "grad_norm": 9.298667907714844,
+ "learning_rate": 1.999998594272138e-05,
+ "loss": 1.7603,
+ "step": 1120
+ },
+ {
+ "epoch": 3.0544959128065394,
+ "grad_norm": 7.971722602844238,
+ "learning_rate": 1.9999984424068405e-05,
+ "loss": 1.6792,
+ "step": 1121
+ },
+ {
+ "epoch": 3.0572207084468666,
+ "grad_norm": 10.661787986755371,
+ "learning_rate": 1.9999982827535874e-05,
+ "loss": 2.0054,
+ "step": 1122
+ },
+ {
+ "epoch": 3.0599455040871932,
+ "grad_norm": 7.763070106506348,
+ "learning_rate": 1.9999981153123797e-05,
+ "loss": 1.7087,
+ "step": 1123
+ },
+ {
+ "epoch": 3.0626702997275204,
+ "grad_norm": 9.361419677734375,
+ "learning_rate": 1.999997940083219e-05,
+ "loss": 1.8989,
+ "step": 1124
+ },
+ {
+ "epoch": 3.0653950953678475,
+ "grad_norm": 6.877585411071777,
+ "learning_rate": 1.9999977570661064e-05,
+ "loss": 1.647,
+ "step": 1125
+ },
+ {
+ "epoch": 3.068119891008174,
+ "grad_norm": 11.645733833312988,
+ "learning_rate": 1.9999975662610435e-05,
+ "loss": 1.9216,
+ "step": 1126
+ },
+ {
+ "epoch": 3.0708446866485013,
+ "grad_norm": 8.425326347351074,
+ "learning_rate": 1.9999973676680318e-05,
+ "loss": 1.7939,
+ "step": 1127
+ },
+ {
+ "epoch": 3.0735694822888284,
+ "grad_norm": 7.795475482940674,
+ "learning_rate": 1.999997161287073e-05,
+ "loss": 1.835,
+ "step": 1128
+ },
+ {
+ "epoch": 3.076294277929155,
+ "grad_norm": 7.606486797332764,
+ "learning_rate": 1.999996947118168e-05,
+ "loss": 1.7539,
+ "step": 1129
+ },
+ {
+ "epoch": 3.0790190735694822,
+ "grad_norm": 9.56530475616455,
+ "learning_rate": 1.999996725161319e-05,
+ "loss": 1.8774,
+ "step": 1130
+ },
+ {
+ "epoch": 3.0817438692098094,
+ "grad_norm": 12.67878246307373,
+ "learning_rate": 1.999996495416528e-05,
+ "loss": 1.8452,
+ "step": 1131
+ },
+ {
+ "epoch": 3.084468664850136,
+ "grad_norm": 8.474047660827637,
+ "learning_rate": 1.9999962578837967e-05,
+ "loss": 1.8213,
+ "step": 1132
+ },
+ {
+ "epoch": 3.087193460490463,
+ "grad_norm": 9.535200119018555,
+ "learning_rate": 1.999996012563126e-05,
+ "loss": 1.7183,
+ "step": 1133
+ },
+ {
+ "epoch": 3.0899182561307903,
+ "grad_norm": 10.665349006652832,
+ "learning_rate": 1.9999957594545193e-05,
+ "loss": 1.855,
+ "step": 1134
+ },
+ {
+ "epoch": 3.092643051771117,
+ "grad_norm": 9.225936889648438,
+ "learning_rate": 1.9999954985579774e-05,
+ "loss": 1.824,
+ "step": 1135
+ },
+ {
+ "epoch": 3.095367847411444,
+ "grad_norm": 9.966315269470215,
+ "learning_rate": 1.999995229873503e-05,
+ "loss": 1.8521,
+ "step": 1136
+ },
+ {
+ "epoch": 3.0980926430517712,
+ "grad_norm": 10.129884719848633,
+ "learning_rate": 1.9999949534010975e-05,
+ "loss": 2.0864,
+ "step": 1137
+ },
+ {
+ "epoch": 3.100817438692098,
+ "grad_norm": 12.97407054901123,
+ "learning_rate": 1.9999946691407635e-05,
+ "loss": 1.623,
+ "step": 1138
+ },
+ {
+ "epoch": 3.103542234332425,
+ "grad_norm": 10.48910903930664,
+ "learning_rate": 1.9999943770925037e-05,
+ "loss": 1.7559,
+ "step": 1139
+ },
+ {
+ "epoch": 3.106267029972752,
+ "grad_norm": 7.656950950622559,
+ "learning_rate": 1.9999940772563194e-05,
+ "loss": 1.75,
+ "step": 1140
+ },
+ {
+ "epoch": 3.108991825613079,
+ "grad_norm": 8.480740547180176,
+ "learning_rate": 1.9999937696322136e-05,
+ "loss": 1.668,
+ "step": 1141
+ },
+ {
+ "epoch": 3.111716621253406,
+ "grad_norm": 10.221088409423828,
+ "learning_rate": 1.9999934542201888e-05,
+ "loss": 1.916,
+ "step": 1142
+ },
+ {
+ "epoch": 3.114441416893733,
+ "grad_norm": 13.228669166564941,
+ "learning_rate": 1.999993131020247e-05,
+ "loss": 1.9434,
+ "step": 1143
+ },
+ {
+ "epoch": 3.11716621253406,
+ "grad_norm": 17.25545883178711,
+ "learning_rate": 1.9999928000323903e-05,
+ "loss": 1.8149,
+ "step": 1144
+ },
+ {
+ "epoch": 3.119891008174387,
+ "grad_norm": 9.24235725402832,
+ "learning_rate": 1.9999924612566225e-05,
+ "loss": 1.936,
+ "step": 1145
+ },
+ {
+ "epoch": 3.122615803814714,
+ "grad_norm": 11.124603271484375,
+ "learning_rate": 1.9999921146929454e-05,
+ "loss": 1.6782,
+ "step": 1146
+ },
+ {
+ "epoch": 3.1253405994550407,
+ "grad_norm": 9.634218215942383,
+ "learning_rate": 1.9999917603413618e-05,
+ "loss": 1.8579,
+ "step": 1147
+ },
+ {
+ "epoch": 3.128065395095368,
+ "grad_norm": 7.349420070648193,
+ "learning_rate": 1.9999913982018747e-05,
+ "loss": 1.8623,
+ "step": 1148
+ },
+ {
+ "epoch": 3.130790190735695,
+ "grad_norm": 7.989165306091309,
+ "learning_rate": 1.9999910282744865e-05,
+ "loss": 1.821,
+ "step": 1149
+ },
+ {
+ "epoch": 3.1335149863760217,
+ "grad_norm": 16.752174377441406,
+ "learning_rate": 1.999990650559201e-05,
+ "loss": 1.8301,
+ "step": 1150
+ },
+ {
+ "epoch": 3.136239782016349,
+ "grad_norm": 9.370292663574219,
+ "learning_rate": 1.99999026505602e-05,
+ "loss": 1.8301,
+ "step": 1151
+ },
+ {
+ "epoch": 3.138964577656676,
+ "grad_norm": 8.419126510620117,
+ "learning_rate": 1.999989871764947e-05,
+ "loss": 1.6052,
+ "step": 1152
+ },
+ {
+ "epoch": 3.1416893732970026,
+ "grad_norm": 8.884900093078613,
+ "learning_rate": 1.9999894706859852e-05,
+ "loss": 1.937,
+ "step": 1153
+ },
+ {
+ "epoch": 3.1444141689373297,
+ "grad_norm": 18.284271240234375,
+ "learning_rate": 1.9999890618191375e-05,
+ "loss": 1.7983,
+ "step": 1154
+ },
+ {
+ "epoch": 3.147138964577657,
+ "grad_norm": 11.208566665649414,
+ "learning_rate": 1.999988645164407e-05,
+ "loss": 1.9102,
+ "step": 1155
+ },
+ {
+ "epoch": 3.1498637602179835,
+ "grad_norm": 9.461328506469727,
+ "learning_rate": 1.9999882207217976e-05,
+ "loss": 1.9668,
+ "step": 1156
+ },
+ {
+ "epoch": 3.1525885558583107,
+ "grad_norm": 8.512341499328613,
+ "learning_rate": 1.9999877884913116e-05,
+ "loss": 1.7991,
+ "step": 1157
+ },
+ {
+ "epoch": 3.155313351498638,
+ "grad_norm": 10.825557708740234,
+ "learning_rate": 1.9999873484729533e-05,
+ "loss": 1.9751,
+ "step": 1158
+ },
+ {
+ "epoch": 3.1580381471389645,
+ "grad_norm": 10.35736083984375,
+ "learning_rate": 1.9999869006667254e-05,
+ "loss": 1.7261,
+ "step": 1159
+ },
+ {
+ "epoch": 3.1607629427792916,
+ "grad_norm": 7.053883075714111,
+ "learning_rate": 1.999986445072632e-05,
+ "loss": 1.7021,
+ "step": 1160
+ },
+ {
+ "epoch": 3.1634877384196187,
+ "grad_norm": 8.23268985748291,
+ "learning_rate": 1.9999859816906763e-05,
+ "loss": 1.7603,
+ "step": 1161
+ },
+ {
+ "epoch": 3.1662125340599454,
+ "grad_norm": 12.892945289611816,
+ "learning_rate": 1.9999855105208616e-05,
+ "loss": 1.853,
+ "step": 1162
+ },
+ {
+ "epoch": 3.1689373297002725,
+ "grad_norm": 9.366336822509766,
+ "learning_rate": 1.9999850315631925e-05,
+ "loss": 1.9128,
+ "step": 1163
+ },
+ {
+ "epoch": 3.1716621253405997,
+ "grad_norm": 9.767892837524414,
+ "learning_rate": 1.9999845448176717e-05,
+ "loss": 1.873,
+ "step": 1164
+ },
+ {
+ "epoch": 3.1743869209809263,
+ "grad_norm": 8.574372291564941,
+ "learning_rate": 1.9999840502843034e-05,
+ "loss": 1.6907,
+ "step": 1165
+ },
+ {
+ "epoch": 3.1771117166212535,
+ "grad_norm": 10.088845252990723,
+ "learning_rate": 1.999983547963092e-05,
+ "loss": 1.8413,
+ "step": 1166
+ },
+ {
+ "epoch": 3.1798365122615806,
+ "grad_norm": 9.044718742370605,
+ "learning_rate": 1.9999830378540403e-05,
+ "loss": 1.7429,
+ "step": 1167
+ },
+ {
+ "epoch": 3.1825613079019073,
+ "grad_norm": 8.50458812713623,
+ "learning_rate": 1.9999825199571533e-05,
+ "loss": 1.7864,
+ "step": 1168
+ },
+ {
+ "epoch": 3.1852861035422344,
+ "grad_norm": 9.272790908813477,
+ "learning_rate": 1.9999819942724343e-05,
+ "loss": 1.9028,
+ "step": 1169
+ },
+ {
+ "epoch": 3.1880108991825615,
+ "grad_norm": 11.100831031799316,
+ "learning_rate": 1.9999814607998884e-05,
+ "loss": 1.9453,
+ "step": 1170
+ },
+ {
+ "epoch": 3.190735694822888,
+ "grad_norm": 10.659100532531738,
+ "learning_rate": 1.9999809195395182e-05,
+ "loss": 1.8159,
+ "step": 1171
+ },
+ {
+ "epoch": 3.1934604904632153,
+ "grad_norm": 14.258978843688965,
+ "learning_rate": 1.9999803704913293e-05,
+ "loss": 1.8989,
+ "step": 1172
+ },
+ {
+ "epoch": 3.1961852861035425,
+ "grad_norm": 12.50585651397705,
+ "learning_rate": 1.9999798136553253e-05,
+ "loss": 1.8374,
+ "step": 1173
+ },
+ {
+ "epoch": 3.198910081743869,
+ "grad_norm": 8.218421936035156,
+ "learning_rate": 1.9999792490315107e-05,
+ "loss": 2.0269,
+ "step": 1174
+ },
+ {
+ "epoch": 3.2016348773841963,
+ "grad_norm": 8.219926834106445,
+ "learning_rate": 1.9999786766198902e-05,
+ "loss": 1.6787,
+ "step": 1175
+ },
+ {
+ "epoch": 3.204359673024523,
+ "grad_norm": 6.338052749633789,
+ "learning_rate": 1.9999780964204675e-05,
+ "loss": 1.7798,
+ "step": 1176
+ },
+ {
+ "epoch": 3.20708446866485,
+ "grad_norm": 8.424579620361328,
+ "learning_rate": 1.9999775084332477e-05,
+ "loss": 1.8496,
+ "step": 1177
+ },
+ {
+ "epoch": 3.209809264305177,
+ "grad_norm": 15.640963554382324,
+ "learning_rate": 1.999976912658235e-05,
+ "loss": 1.981,
+ "step": 1178
+ },
+ {
+ "epoch": 3.212534059945504,
+ "grad_norm": 11.739542007446289,
+ "learning_rate": 1.999976309095435e-05,
+ "loss": 1.7754,
+ "step": 1179
+ },
+ {
+ "epoch": 3.215258855585831,
+ "grad_norm": 11.112152099609375,
+ "learning_rate": 1.999975697744851e-05,
+ "loss": 1.9946,
+ "step": 1180
+ },
+ {
+ "epoch": 3.217983651226158,
+ "grad_norm": 8.312841415405273,
+ "learning_rate": 1.999975078606489e-05,
+ "loss": 1.5332,
+ "step": 1181
+ },
+ {
+ "epoch": 3.220708446866485,
+ "grad_norm": 7.601229190826416,
+ "learning_rate": 1.9999744516803527e-05,
+ "loss": 2.0059,
+ "step": 1182
+ },
+ {
+ "epoch": 3.223433242506812,
+ "grad_norm": 11.420334815979004,
+ "learning_rate": 1.999973816966448e-05,
+ "loss": 2.085,
+ "step": 1183
+ },
+ {
+ "epoch": 3.226158038147139,
+ "grad_norm": 9.687848091125488,
+ "learning_rate": 1.999973174464779e-05,
+ "loss": 1.7739,
+ "step": 1184
+ },
+ {
+ "epoch": 3.2288828337874658,
+ "grad_norm": 9.021745681762695,
+ "learning_rate": 1.9999725241753516e-05,
+ "loss": 1.7266,
+ "step": 1185
+ },
+ {
+ "epoch": 3.231607629427793,
+ "grad_norm": 21.235626220703125,
+ "learning_rate": 1.9999718660981703e-05,
+ "loss": 1.9561,
+ "step": 1186
+ },
+ {
+ "epoch": 3.23433242506812,
+ "grad_norm": 8.093171119689941,
+ "learning_rate": 1.9999712002332397e-05,
+ "loss": 1.7197,
+ "step": 1187
+ },
+ {
+ "epoch": 3.2370572207084467,
+ "grad_norm": 10.616466522216797,
+ "learning_rate": 1.9999705265805662e-05,
+ "loss": 1.8701,
+ "step": 1188
+ },
+ {
+ "epoch": 3.239782016348774,
+ "grad_norm": 9.980297088623047,
+ "learning_rate": 1.9999698451401542e-05,
+ "loss": 1.6729,
+ "step": 1189
+ },
+ {
+ "epoch": 3.242506811989101,
+ "grad_norm": 9.78648853302002,
+ "learning_rate": 1.9999691559120092e-05,
+ "loss": 1.8533,
+ "step": 1190
+ },
+ {
+ "epoch": 3.2452316076294276,
+ "grad_norm": 10.780373573303223,
+ "learning_rate": 1.999968458896137e-05,
+ "loss": 1.7625,
+ "step": 1191
+ },
+ {
+ "epoch": 3.2479564032697548,
+ "grad_norm": 9.349573135375977,
+ "learning_rate": 1.9999677540925416e-05,
+ "loss": 1.6006,
+ "step": 1192
+ },
+ {
+ "epoch": 3.250681198910082,
+ "grad_norm": 7.98806095123291,
+ "learning_rate": 1.9999670415012306e-05,
+ "loss": 1.6597,
+ "step": 1193
+ },
+ {
+ "epoch": 3.2534059945504086,
+ "grad_norm": 9.595312118530273,
+ "learning_rate": 1.9999663211222078e-05,
+ "loss": 1.9124,
+ "step": 1194
+ },
+ {
+ "epoch": 3.2561307901907357,
+ "grad_norm": 9.717113494873047,
+ "learning_rate": 1.9999655929554798e-05,
+ "loss": 1.7925,
+ "step": 1195
+ },
+ {
+ "epoch": 3.258855585831063,
+ "grad_norm": 10.403453826904297,
+ "learning_rate": 1.999964857001052e-05,
+ "loss": 2.02,
+ "step": 1196
+ },
+ {
+ "epoch": 3.2615803814713895,
+ "grad_norm": 9.493276596069336,
+ "learning_rate": 1.99996411325893e-05,
+ "loss": 1.7822,
+ "step": 1197
+ },
+ {
+ "epoch": 3.2643051771117166,
+ "grad_norm": 9.495288848876953,
+ "learning_rate": 1.9999633617291196e-05,
+ "loss": 1.6096,
+ "step": 1198
+ },
+ {
+ "epoch": 3.2670299727520438,
+ "grad_norm": 10.892885208129883,
+ "learning_rate": 1.9999626024116267e-05,
+ "loss": 1.8101,
+ "step": 1199
+ },
+ {
+ "epoch": 3.2697547683923704,
+ "grad_norm": 9.678936004638672,
+ "learning_rate": 1.999961835306457e-05,
+ "loss": 1.854,
+ "step": 1200
+ },
+ {
+ "epoch": 3.2724795640326976,
+ "grad_norm": 6.917518138885498,
+ "learning_rate": 1.999961060413617e-05,
+ "loss": 1.8389,
+ "step": 1201
+ },
+ {
+ "epoch": 3.2752043596730247,
+ "grad_norm": 9.572186470031738,
+ "learning_rate": 1.9999602777331124e-05,
+ "loss": 1.9426,
+ "step": 1202
+ },
+ {
+ "epoch": 3.2779291553133514,
+ "grad_norm": 10.134210586547852,
+ "learning_rate": 1.9999594872649495e-05,
+ "loss": 1.6909,
+ "step": 1203
+ },
+ {
+ "epoch": 3.2806539509536785,
+ "grad_norm": 13.50228500366211,
+ "learning_rate": 1.999958689009134e-05,
+ "loss": 2.0244,
+ "step": 1204
+ },
+ {
+ "epoch": 3.2833787465940056,
+ "grad_norm": 8.332965850830078,
+ "learning_rate": 1.9999578829656724e-05,
+ "loss": 1.749,
+ "step": 1205
+ },
+ {
+ "epoch": 3.2861035422343323,
+ "grad_norm": 14.706710815429688,
+ "learning_rate": 1.9999570691345712e-05,
+ "loss": 1.8784,
+ "step": 1206
+ },
+ {
+ "epoch": 3.2888283378746594,
+ "grad_norm": 7.984215259552002,
+ "learning_rate": 1.9999562475158365e-05,
+ "loss": 1.7832,
+ "step": 1207
+ },
+ {
+ "epoch": 3.291553133514986,
+ "grad_norm": 7.736905097961426,
+ "learning_rate": 1.9999554181094748e-05,
+ "loss": 1.6206,
+ "step": 1208
+ },
+ {
+ "epoch": 3.2942779291553133,
+ "grad_norm": 10.505777359008789,
+ "learning_rate": 1.9999545809154925e-05,
+ "loss": 1.9771,
+ "step": 1209
+ },
+ {
+ "epoch": 3.2970027247956404,
+ "grad_norm": 11.714659690856934,
+ "learning_rate": 1.9999537359338957e-05,
+ "loss": 2.0229,
+ "step": 1210
+ },
+ {
+ "epoch": 3.299727520435967,
+ "grad_norm": 7.147460460662842,
+ "learning_rate": 1.999952883164692e-05,
+ "loss": 1.7773,
+ "step": 1211
+ },
+ {
+ "epoch": 3.302452316076294,
+ "grad_norm": 7.624472141265869,
+ "learning_rate": 1.9999520226078872e-05,
+ "loss": 2.0332,
+ "step": 1212
+ },
+ {
+ "epoch": 3.3051771117166213,
+ "grad_norm": 8.018260955810547,
+ "learning_rate": 1.999951154263488e-05,
+ "loss": 1.5349,
+ "step": 1213
+ },
+ {
+ "epoch": 3.307901907356948,
+ "grad_norm": 10.637834548950195,
+ "learning_rate": 1.9999502781315018e-05,
+ "loss": 2.0303,
+ "step": 1214
+ },
+ {
+ "epoch": 3.310626702997275,
+ "grad_norm": 8.472352027893066,
+ "learning_rate": 1.9999493942119346e-05,
+ "loss": 1.78,
+ "step": 1215
+ },
+ {
+ "epoch": 3.3133514986376023,
+ "grad_norm": 9.631685256958008,
+ "learning_rate": 1.999948502504794e-05,
+ "loss": 1.728,
+ "step": 1216
+ },
+ {
+ "epoch": 3.316076294277929,
+ "grad_norm": 9.096489906311035,
+ "learning_rate": 1.9999476030100862e-05,
+ "loss": 1.7917,
+ "step": 1217
+ },
+ {
+ "epoch": 3.318801089918256,
+ "grad_norm": 12.95285701751709,
+ "learning_rate": 1.999946695727819e-05,
+ "loss": 1.9072,
+ "step": 1218
+ },
+ {
+ "epoch": 3.321525885558583,
+ "grad_norm": 11.47706413269043,
+ "learning_rate": 1.999945780657999e-05,
+ "loss": 1.9102,
+ "step": 1219
+ },
+ {
+ "epoch": 3.32425068119891,
+ "grad_norm": 8.722806930541992,
+ "learning_rate": 1.9999448578006337e-05,
+ "loss": 1.7346,
+ "step": 1220
+ },
+ {
+ "epoch": 3.326975476839237,
+ "grad_norm": 8.225024223327637,
+ "learning_rate": 1.9999439271557296e-05,
+ "loss": 2.0669,
+ "step": 1221
+ },
+ {
+ "epoch": 3.329700272479564,
+ "grad_norm": 9.275016784667969,
+ "learning_rate": 1.9999429887232948e-05,
+ "loss": 1.7715,
+ "step": 1222
+ },
+ {
+ "epoch": 3.332425068119891,
+ "grad_norm": 10.683072090148926,
+ "learning_rate": 1.999942042503336e-05,
+ "loss": 1.7544,
+ "step": 1223
+ },
+ {
+ "epoch": 3.335149863760218,
+ "grad_norm": 9.817999839782715,
+ "learning_rate": 1.9999410884958608e-05,
+ "loss": 1.6128,
+ "step": 1224
+ },
+ {
+ "epoch": 3.337874659400545,
+ "grad_norm": 17.06625747680664,
+ "learning_rate": 1.9999401267008766e-05,
+ "loss": 1.7949,
+ "step": 1225
+ },
+ {
+ "epoch": 3.3405994550408717,
+ "grad_norm": 8.310747146606445,
+ "learning_rate": 1.9999391571183907e-05,
+ "loss": 1.8447,
+ "step": 1226
+ },
+ {
+ "epoch": 3.343324250681199,
+ "grad_norm": 8.697625160217285,
+ "learning_rate": 1.9999381797484113e-05,
+ "loss": 1.7803,
+ "step": 1227
+ },
+ {
+ "epoch": 3.346049046321526,
+ "grad_norm": 8.940540313720703,
+ "learning_rate": 1.999937194590945e-05,
+ "loss": 2.1494,
+ "step": 1228
+ },
+ {
+ "epoch": 3.3487738419618527,
+ "grad_norm": 8.396451950073242,
+ "learning_rate": 1.9999362016460005e-05,
+ "loss": 1.7114,
+ "step": 1229
+ },
+ {
+ "epoch": 3.35149863760218,
+ "grad_norm": 10.124202728271484,
+ "learning_rate": 1.9999352009135846e-05,
+ "loss": 1.9092,
+ "step": 1230
+ },
+ {
+ "epoch": 3.354223433242507,
+ "grad_norm": 7.952688217163086,
+ "learning_rate": 1.9999341923937056e-05,
+ "loss": 1.6101,
+ "step": 1231
+ },
+ {
+ "epoch": 3.3569482288828336,
+ "grad_norm": 15.077750205993652,
+ "learning_rate": 1.9999331760863715e-05,
+ "loss": 1.7925,
+ "step": 1232
+ },
+ {
+ "epoch": 3.3596730245231607,
+ "grad_norm": 25.50319480895996,
+ "learning_rate": 1.99993215199159e-05,
+ "loss": 1.8171,
+ "step": 1233
+ },
+ {
+ "epoch": 3.362397820163488,
+ "grad_norm": 9.526121139526367,
+ "learning_rate": 1.999931120109369e-05,
+ "loss": 1.9714,
+ "step": 1234
+ },
+ {
+ "epoch": 3.3651226158038146,
+ "grad_norm": 7.219566345214844,
+ "learning_rate": 1.9999300804397164e-05,
+ "loss": 1.8828,
+ "step": 1235
+ },
+ {
+ "epoch": 3.3678474114441417,
+ "grad_norm": 6.291162967681885,
+ "learning_rate": 1.9999290329826406e-05,
+ "loss": 1.5781,
+ "step": 1236
+ },
+ {
+ "epoch": 3.370572207084469,
+ "grad_norm": 6.604852676391602,
+ "learning_rate": 1.99992797773815e-05,
+ "loss": 1.7612,
+ "step": 1237
+ },
+ {
+ "epoch": 3.3732970027247955,
+ "grad_norm": 7.027307033538818,
+ "learning_rate": 1.999926914706252e-05,
+ "loss": 1.79,
+ "step": 1238
+ },
+ {
+ "epoch": 3.3760217983651226,
+ "grad_norm": 8.008625984191895,
+ "learning_rate": 1.9999258438869558e-05,
+ "loss": 1.9111,
+ "step": 1239
+ },
+ {
+ "epoch": 3.3787465940054497,
+ "grad_norm": 9.246003150939941,
+ "learning_rate": 1.999924765280269e-05,
+ "loss": 1.7434,
+ "step": 1240
+ },
+ {
+ "epoch": 3.3814713896457764,
+ "grad_norm": 9.255450248718262,
+ "learning_rate": 1.9999236788862006e-05,
+ "loss": 1.6729,
+ "step": 1241
+ },
+ {
+ "epoch": 3.3841961852861036,
+ "grad_norm": 11.025579452514648,
+ "learning_rate": 1.999922584704759e-05,
+ "loss": 1.8604,
+ "step": 1242
+ },
+ {
+ "epoch": 3.3869209809264307,
+ "grad_norm": 9.993318557739258,
+ "learning_rate": 1.9999214827359518e-05,
+ "loss": 2.2808,
+ "step": 1243
+ },
+ {
+ "epoch": 3.3896457765667574,
+ "grad_norm": 10.769620895385742,
+ "learning_rate": 1.9999203729797887e-05,
+ "loss": 1.875,
+ "step": 1244
+ },
+ {
+ "epoch": 3.3923705722070845,
+ "grad_norm": 8.073010444641113,
+ "learning_rate": 1.9999192554362777e-05,
+ "loss": 1.8884,
+ "step": 1245
+ },
+ {
+ "epoch": 3.3950953678474116,
+ "grad_norm": 7.63186502456665,
+ "learning_rate": 1.999918130105428e-05,
+ "loss": 1.6306,
+ "step": 1246
+ },
+ {
+ "epoch": 3.3978201634877383,
+ "grad_norm": 7.426112651824951,
+ "learning_rate": 1.999916996987248e-05,
+ "loss": 1.8506,
+ "step": 1247
+ },
+ {
+ "epoch": 3.4005449591280654,
+ "grad_norm": 8.368783950805664,
+ "learning_rate": 1.9999158560817463e-05,
+ "loss": 1.6318,
+ "step": 1248
+ },
+ {
+ "epoch": 3.4032697547683926,
+ "grad_norm": 12.027238845825195,
+ "learning_rate": 1.9999147073889325e-05,
+ "loss": 1.8999,
+ "step": 1249
+ },
+ {
+ "epoch": 3.4059945504087192,
+ "grad_norm": 8.650468826293945,
+ "learning_rate": 1.999913550908815e-05,
+ "loss": 2.144,
+ "step": 1250
+ },
+ {
+ "epoch": 3.4087193460490464,
+ "grad_norm": 6.946683406829834,
+ "learning_rate": 1.9999123866414026e-05,
+ "loss": 1.7266,
+ "step": 1251
+ },
+ {
+ "epoch": 3.4114441416893735,
+ "grad_norm": 10.529762268066406,
+ "learning_rate": 1.9999112145867055e-05,
+ "loss": 1.8828,
+ "step": 1252
+ },
+ {
+ "epoch": 3.4141689373297,
+ "grad_norm": 6.534323692321777,
+ "learning_rate": 1.9999100347447314e-05,
+ "loss": 1.7263,
+ "step": 1253
+ },
+ {
+ "epoch": 3.4168937329700273,
+ "grad_norm": 6.989877700805664,
+ "learning_rate": 1.9999088471154904e-05,
+ "loss": 1.9609,
+ "step": 1254
+ },
+ {
+ "epoch": 3.4196185286103544,
+ "grad_norm": 8.66665267944336,
+ "learning_rate": 1.999907651698991e-05,
+ "loss": 2.1255,
+ "step": 1255
+ },
+ {
+ "epoch": 3.422343324250681,
+ "grad_norm": 6.886997222900391,
+ "learning_rate": 1.999906448495244e-05,
+ "loss": 1.7437,
+ "step": 1256
+ },
+ {
+ "epoch": 3.4250681198910082,
+ "grad_norm": 7.171701908111572,
+ "learning_rate": 1.999905237504257e-05,
+ "loss": 1.646,
+ "step": 1257
+ },
+ {
+ "epoch": 3.4277929155313354,
+ "grad_norm": 13.07217025756836,
+ "learning_rate": 1.9999040187260403e-05,
+ "loss": 1.8066,
+ "step": 1258
+ },
+ {
+ "epoch": 3.430517711171662,
+ "grad_norm": 8.112521171569824,
+ "learning_rate": 1.9999027921606033e-05,
+ "loss": 1.7847,
+ "step": 1259
+ },
+ {
+ "epoch": 3.433242506811989,
+ "grad_norm": 8.04460620880127,
+ "learning_rate": 1.9999015578079553e-05,
+ "loss": 1.7898,
+ "step": 1260
+ },
+ {
+ "epoch": 3.4359673024523163,
+ "grad_norm": 8.872373580932617,
+ "learning_rate": 1.9999003156681066e-05,
+ "loss": 1.9102,
+ "step": 1261
+ },
+ {
+ "epoch": 3.438692098092643,
+ "grad_norm": 8.70421314239502,
+ "learning_rate": 1.999899065741066e-05,
+ "loss": 1.7324,
+ "step": 1262
+ },
+ {
+ "epoch": 3.44141689373297,
+ "grad_norm": 6.915367126464844,
+ "learning_rate": 1.999897808026844e-05,
+ "loss": 1.9417,
+ "step": 1263
+ },
+ {
+ "epoch": 3.4441416893732972,
+ "grad_norm": 9.468276977539062,
+ "learning_rate": 1.99989654252545e-05,
+ "loss": 1.8035,
+ "step": 1264
+ },
+ {
+ "epoch": 3.446866485013624,
+ "grad_norm": 6.98765754699707,
+ "learning_rate": 1.9998952692368937e-05,
+ "loss": 2.082,
+ "step": 1265
+ },
+ {
+ "epoch": 3.449591280653951,
+ "grad_norm": 7.00003719329834,
+ "learning_rate": 1.9998939881611857e-05,
+ "loss": 1.688,
+ "step": 1266
+ },
+ {
+ "epoch": 3.452316076294278,
+ "grad_norm": 9.563433647155762,
+ "learning_rate": 1.9998926992983348e-05,
+ "loss": 1.844,
+ "step": 1267
+ },
+ {
+ "epoch": 3.455040871934605,
+ "grad_norm": 10.383186340332031,
+ "learning_rate": 1.9998914026483526e-05,
+ "loss": 1.7056,
+ "step": 1268
+ },
+ {
+ "epoch": 3.457765667574932,
+ "grad_norm": 8.172235488891602,
+ "learning_rate": 1.9998900982112475e-05,
+ "loss": 1.8098,
+ "step": 1269
+ },
+ {
+ "epoch": 3.460490463215259,
+ "grad_norm": 13.764888763427734,
+ "learning_rate": 1.999888785987031e-05,
+ "loss": 2.1206,
+ "step": 1270
+ },
+ {
+ "epoch": 3.463215258855586,
+ "grad_norm": 10.939154624938965,
+ "learning_rate": 1.9998874659757125e-05,
+ "loss": 1.9473,
+ "step": 1271
+ },
+ {
+ "epoch": 3.465940054495913,
+ "grad_norm": 8.621336936950684,
+ "learning_rate": 1.999886138177303e-05,
+ "loss": 1.7461,
+ "step": 1272
+ },
+ {
+ "epoch": 3.4686648501362396,
+ "grad_norm": 11.209137916564941,
+ "learning_rate": 1.999884802591812e-05,
+ "loss": 1.5256,
+ "step": 1273
+ },
+ {
+ "epoch": 3.4713896457765667,
+ "grad_norm": 11.211379051208496,
+ "learning_rate": 1.999883459219251e-05,
+ "loss": 1.8311,
+ "step": 1274
+ },
+ {
+ "epoch": 3.474114441416894,
+ "grad_norm": 8.0631685256958,
+ "learning_rate": 1.999882108059629e-05,
+ "loss": 1.8853,
+ "step": 1275
+ },
+ {
+ "epoch": 3.4768392370572205,
+ "grad_norm": 8.133895874023438,
+ "learning_rate": 1.9998807491129576e-05,
+ "loss": 1.7358,
+ "step": 1276
+ },
+ {
+ "epoch": 3.4795640326975477,
+ "grad_norm": 9.907873153686523,
+ "learning_rate": 1.9998793823792474e-05,
+ "loss": 2.0115,
+ "step": 1277
+ },
+ {
+ "epoch": 3.482288828337875,
+ "grad_norm": 7.711964130401611,
+ "learning_rate": 1.9998780078585085e-05,
+ "loss": 1.9102,
+ "step": 1278
+ },
+ {
+ "epoch": 3.4850136239782015,
+ "grad_norm": 9.835423469543457,
+ "learning_rate": 1.999876625550752e-05,
+ "loss": 1.8042,
+ "step": 1279
+ },
+ {
+ "epoch": 3.4877384196185286,
+ "grad_norm": 9.073649406433105,
+ "learning_rate": 1.999875235455988e-05,
+ "loss": 1.6558,
+ "step": 1280
+ },
+ {
+ "epoch": 3.4904632152588557,
+ "grad_norm": 9.008769035339355,
+ "learning_rate": 1.9998738375742283e-05,
+ "loss": 1.8687,
+ "step": 1281
+ },
+ {
+ "epoch": 3.4931880108991824,
+ "grad_norm": 9.119298934936523,
+ "learning_rate": 1.9998724319054836e-05,
+ "loss": 2.0308,
+ "step": 1282
+ },
+ {
+ "epoch": 3.4959128065395095,
+ "grad_norm": 6.739688396453857,
+ "learning_rate": 1.999871018449764e-05,
+ "loss": 1.5837,
+ "step": 1283
+ },
+ {
+ "epoch": 3.4986376021798367,
+ "grad_norm": 6.520759105682373,
+ "learning_rate": 1.9998695972070814e-05,
+ "loss": 1.5688,
+ "step": 1284
+ },
+ {
+ "epoch": 3.5013623978201633,
+ "grad_norm": 7.264618873596191,
+ "learning_rate": 1.9998681681774465e-05,
+ "loss": 1.5786,
+ "step": 1285
+ },
+ {
+ "epoch": 3.5040871934604905,
+ "grad_norm": 7.286742210388184,
+ "learning_rate": 1.9998667313608706e-05,
+ "loss": 1.6384,
+ "step": 1286
+ },
+ {
+ "epoch": 3.5068119891008176,
+ "grad_norm": 7.8493828773498535,
+ "learning_rate": 1.9998652867573646e-05,
+ "loss": 1.6152,
+ "step": 1287
+ },
+ {
+ "epoch": 3.5095367847411443,
+ "grad_norm": 8.601771354675293,
+ "learning_rate": 1.99986383436694e-05,
+ "loss": 1.9187,
+ "step": 1288
+ },
+ {
+ "epoch": 3.5122615803814714,
+ "grad_norm": 10.515268325805664,
+ "learning_rate": 1.9998623741896082e-05,
+ "loss": 1.9033,
+ "step": 1289
+ },
+ {
+ "epoch": 3.5149863760217985,
+ "grad_norm": 8.543349266052246,
+ "learning_rate": 1.99986090622538e-05,
+ "loss": 2.1929,
+ "step": 1290
+ },
+ {
+ "epoch": 3.517711171662125,
+ "grad_norm": 7.427716255187988,
+ "learning_rate": 1.9998594304742677e-05,
+ "loss": 1.6658,
+ "step": 1291
+ },
+ {
+ "epoch": 3.5204359673024523,
+ "grad_norm": 7.886240482330322,
+ "learning_rate": 1.999857946936282e-05,
+ "loss": 1.7354,
+ "step": 1292
+ },
+ {
+ "epoch": 3.523160762942779,
+ "grad_norm": 9.57736587524414,
+ "learning_rate": 1.999856455611435e-05,
+ "loss": 1.6567,
+ "step": 1293
+ },
+ {
+ "epoch": 3.525885558583106,
+ "grad_norm": 7.3349995613098145,
+ "learning_rate": 1.999854956499738e-05,
+ "loss": 1.6038,
+ "step": 1294
+ },
+ {
+ "epoch": 3.5286103542234333,
+ "grad_norm": 6.272554874420166,
+ "learning_rate": 1.9998534496012026e-05,
+ "loss": 1.7808,
+ "step": 1295
+ },
+ {
+ "epoch": 3.53133514986376,
+ "grad_norm": 10.40566635131836,
+ "learning_rate": 1.999851934915841e-05,
+ "loss": 1.8052,
+ "step": 1296
+ },
+ {
+ "epoch": 3.534059945504087,
+ "grad_norm": 7.208871364593506,
+ "learning_rate": 1.9998504124436647e-05,
+ "loss": 1.9629,
+ "step": 1297
+ },
+ {
+ "epoch": 3.536784741144414,
+ "grad_norm": 6.684272766113281,
+ "learning_rate": 1.9998488821846855e-05,
+ "loss": 1.4717,
+ "step": 1298
+ },
+ {
+ "epoch": 3.539509536784741,
+ "grad_norm": 6.549276351928711,
+ "learning_rate": 1.999847344138915e-05,
+ "loss": 1.7974,
+ "step": 1299
+ },
+ {
+ "epoch": 3.542234332425068,
+ "grad_norm": 7.418403625488281,
+ "learning_rate": 1.999845798306366e-05,
+ "loss": 1.5657,
+ "step": 1300
+ },
+ {
+ "epoch": 3.544959128065395,
+ "grad_norm": 7.693958759307861,
+ "learning_rate": 1.9998442446870503e-05,
+ "loss": 1.7651,
+ "step": 1301
+ },
+ {
+ "epoch": 3.547683923705722,
+ "grad_norm": 7.910401821136475,
+ "learning_rate": 1.9998426832809796e-05,
+ "loss": 1.75,
+ "step": 1302
+ },
+ {
+ "epoch": 3.550408719346049,
+ "grad_norm": 6.9944868087768555,
+ "learning_rate": 1.9998411140881662e-05,
+ "loss": 1.8208,
+ "step": 1303
+ },
+ {
+ "epoch": 3.553133514986376,
+ "grad_norm": 7.55907678604126,
+ "learning_rate": 1.9998395371086223e-05,
+ "loss": 2.019,
+ "step": 1304
+ },
+ {
+ "epoch": 3.5558583106267028,
+ "grad_norm": 6.756530284881592,
+ "learning_rate": 1.9998379523423604e-05,
+ "loss": 1.6924,
+ "step": 1305
+ },
+ {
+ "epoch": 3.55858310626703,
+ "grad_norm": 7.895157337188721,
+ "learning_rate": 1.9998363597893927e-05,
+ "loss": 1.7705,
+ "step": 1306
+ },
+ {
+ "epoch": 3.561307901907357,
+ "grad_norm": 9.324334144592285,
+ "learning_rate": 1.9998347594497314e-05,
+ "loss": 1.9438,
+ "step": 1307
+ },
+ {
+ "epoch": 3.5640326975476837,
+ "grad_norm": 7.188936233520508,
+ "learning_rate": 1.9998331513233894e-05,
+ "loss": 1.947,
+ "step": 1308
+ },
+ {
+ "epoch": 3.566757493188011,
+ "grad_norm": 11.831025123596191,
+ "learning_rate": 1.9998315354103793e-05,
+ "loss": 1.7488,
+ "step": 1309
+ },
+ {
+ "epoch": 3.569482288828338,
+ "grad_norm": 14.046435356140137,
+ "learning_rate": 1.9998299117107128e-05,
+ "loss": 1.8594,
+ "step": 1310
+ },
+ {
+ "epoch": 3.5722070844686646,
+ "grad_norm": 13.019118309020996,
+ "learning_rate": 1.9998282802244032e-05,
+ "loss": 1.8584,
+ "step": 1311
+ },
+ {
+ "epoch": 3.5749318801089918,
+ "grad_norm": 6.714863300323486,
+ "learning_rate": 1.9998266409514634e-05,
+ "loss": 1.5896,
+ "step": 1312
+ },
+ {
+ "epoch": 3.577656675749319,
+ "grad_norm": 8.124160766601562,
+ "learning_rate": 1.9998249938919056e-05,
+ "loss": 1.6316,
+ "step": 1313
+ },
+ {
+ "epoch": 3.5803814713896456,
+ "grad_norm": 9.683789253234863,
+ "learning_rate": 1.9998233390457433e-05,
+ "loss": 1.8806,
+ "step": 1314
+ },
+ {
+ "epoch": 3.5831062670299727,
+ "grad_norm": 12.51480770111084,
+ "learning_rate": 1.9998216764129887e-05,
+ "loss": 2.0864,
+ "step": 1315
+ },
+ {
+ "epoch": 3.5858310626703,
+ "grad_norm": 7.977465629577637,
+ "learning_rate": 1.9998200059936554e-05,
+ "loss": 1.7783,
+ "step": 1316
+ },
+ {
+ "epoch": 3.5885558583106265,
+ "grad_norm": 9.067839622497559,
+ "learning_rate": 1.9998183277877555e-05,
+ "loss": 1.9639,
+ "step": 1317
+ },
+ {
+ "epoch": 3.5912806539509536,
+ "grad_norm": 8.058653831481934,
+ "learning_rate": 1.9998166417953034e-05,
+ "loss": 1.6387,
+ "step": 1318
+ },
+ {
+ "epoch": 3.5940054495912808,
+ "grad_norm": 7.699615955352783,
+ "learning_rate": 1.9998149480163108e-05,
+ "loss": 1.748,
+ "step": 1319
+ },
+ {
+ "epoch": 3.5967302452316074,
+ "grad_norm": 8.603288650512695,
+ "learning_rate": 1.999813246450792e-05,
+ "loss": 1.8372,
+ "step": 1320
+ },
+ {
+ "epoch": 3.5994550408719346,
+ "grad_norm": 9.643807411193848,
+ "learning_rate": 1.9998115370987595e-05,
+ "loss": 1.8862,
+ "step": 1321
+ },
+ {
+ "epoch": 3.6021798365122617,
+ "grad_norm": 7.772497653961182,
+ "learning_rate": 1.9998098199602273e-05,
+ "loss": 1.6187,
+ "step": 1322
+ },
+ {
+ "epoch": 3.6049046321525884,
+ "grad_norm": 8.318263053894043,
+ "learning_rate": 1.999808095035208e-05,
+ "loss": 1.7034,
+ "step": 1323
+ },
+ {
+ "epoch": 3.6076294277929155,
+ "grad_norm": 9.342367172241211,
+ "learning_rate": 1.9998063623237156e-05,
+ "loss": 1.9819,
+ "step": 1324
+ },
+ {
+ "epoch": 3.6103542234332426,
+ "grad_norm": 10.661358833312988,
+ "learning_rate": 1.9998046218257636e-05,
+ "loss": 1.7305,
+ "step": 1325
+ },
+ {
+ "epoch": 3.6130790190735693,
+ "grad_norm": 9.874344825744629,
+ "learning_rate": 1.9998028735413653e-05,
+ "loss": 1.8345,
+ "step": 1326
+ },
+ {
+ "epoch": 3.6158038147138964,
+ "grad_norm": 21.867250442504883,
+ "learning_rate": 1.9998011174705344e-05,
+ "loss": 1.5757,
+ "step": 1327
+ },
+ {
+ "epoch": 3.6185286103542236,
+ "grad_norm": 8.06169605255127,
+ "learning_rate": 1.9997993536132846e-05,
+ "loss": 1.7373,
+ "step": 1328
+ },
+ {
+ "epoch": 3.6212534059945503,
+ "grad_norm": 7.470646381378174,
+ "learning_rate": 1.9997975819696293e-05,
+ "loss": 1.7646,
+ "step": 1329
+ },
+ {
+ "epoch": 3.6239782016348774,
+ "grad_norm": 8.074699401855469,
+ "learning_rate": 1.9997958025395826e-05,
+ "loss": 1.6255,
+ "step": 1330
+ },
+ {
+ "epoch": 3.6267029972752045,
+ "grad_norm": 8.183103561401367,
+ "learning_rate": 1.999794015323159e-05,
+ "loss": 1.8416,
+ "step": 1331
+ },
+ {
+ "epoch": 3.629427792915531,
+ "grad_norm": 8.8412504196167,
+ "learning_rate": 1.9997922203203713e-05,
+ "loss": 1.8076,
+ "step": 1332
+ },
+ {
+ "epoch": 3.6321525885558583,
+ "grad_norm": 12.103985786437988,
+ "learning_rate": 1.999790417531234e-05,
+ "loss": 1.7451,
+ "step": 1333
+ },
+ {
+ "epoch": 3.6348773841961854,
+ "grad_norm": 7.650243759155273,
+ "learning_rate": 1.999788606955761e-05,
+ "loss": 1.6338,
+ "step": 1334
+ },
+ {
+ "epoch": 3.637602179836512,
+ "grad_norm": 7.686456203460693,
+ "learning_rate": 1.9997867885939668e-05,
+ "loss": 1.5525,
+ "step": 1335
+ },
+ {
+ "epoch": 3.6403269754768393,
+ "grad_norm": 9.113301277160645,
+ "learning_rate": 1.999784962445865e-05,
+ "loss": 2.189,
+ "step": 1336
+ },
+ {
+ "epoch": 3.6430517711171664,
+ "grad_norm": 7.84440803527832,
+ "learning_rate": 1.9997831285114704e-05,
+ "loss": 1.616,
+ "step": 1337
+ },
+ {
+ "epoch": 3.645776566757493,
+ "grad_norm": 9.606657981872559,
+ "learning_rate": 1.9997812867907966e-05,
+ "loss": 1.5798,
+ "step": 1338
+ },
+ {
+ "epoch": 3.64850136239782,
+ "grad_norm": 7.277917385101318,
+ "learning_rate": 1.9997794372838584e-05,
+ "loss": 1.7537,
+ "step": 1339
+ },
+ {
+ "epoch": 3.6512261580381473,
+ "grad_norm": 6.5008463859558105,
+ "learning_rate": 1.9997775799906706e-05,
+ "loss": 1.7476,
+ "step": 1340
+ },
+ {
+ "epoch": 3.653950953678474,
+ "grad_norm": 6.770064353942871,
+ "learning_rate": 1.9997757149112465e-05,
+ "loss": 1.6602,
+ "step": 1341
+ },
+ {
+ "epoch": 3.656675749318801,
+ "grad_norm": 6.9005446434021,
+ "learning_rate": 1.9997738420456015e-05,
+ "loss": 1.6152,
+ "step": 1342
+ },
+ {
+ "epoch": 3.6594005449591283,
+ "grad_norm": 8.411625862121582,
+ "learning_rate": 1.9997719613937505e-05,
+ "loss": 1.9312,
+ "step": 1343
+ },
+ {
+ "epoch": 3.662125340599455,
+ "grad_norm": 8.165867805480957,
+ "learning_rate": 1.9997700729557074e-05,
+ "loss": 1.7017,
+ "step": 1344
+ },
+ {
+ "epoch": 3.664850136239782,
+ "grad_norm": 6.334288120269775,
+ "learning_rate": 1.999768176731487e-05,
+ "loss": 1.728,
+ "step": 1345
+ },
+ {
+ "epoch": 3.667574931880109,
+ "grad_norm": 7.393435001373291,
+ "learning_rate": 1.999766272721104e-05,
+ "loss": 1.8401,
+ "step": 1346
+ },
+ {
+ "epoch": 3.670299727520436,
+ "grad_norm": 8.655657768249512,
+ "learning_rate": 1.9997643609245738e-05,
+ "loss": 1.6099,
+ "step": 1347
+ },
+ {
+ "epoch": 3.673024523160763,
+ "grad_norm": 6.5279364585876465,
+ "learning_rate": 1.9997624413419108e-05,
+ "loss": 1.6265,
+ "step": 1348
+ },
+ {
+ "epoch": 3.67574931880109,
+ "grad_norm": 11.264801025390625,
+ "learning_rate": 1.99976051397313e-05,
+ "loss": 1.7551,
+ "step": 1349
+ },
+ {
+ "epoch": 3.678474114441417,
+ "grad_norm": 6.59581995010376,
+ "learning_rate": 1.999758578818247e-05,
+ "loss": 1.644,
+ "step": 1350
+ },
+ {
+ "epoch": 3.681198910081744,
+ "grad_norm": 6.9846014976501465,
+ "learning_rate": 1.9997566358772755e-05,
+ "loss": 1.7244,
+ "step": 1351
+ },
+ {
+ "epoch": 3.683923705722071,
+ "grad_norm": 7.350223541259766,
+ "learning_rate": 1.9997546851502322e-05,
+ "loss": 1.625,
+ "step": 1352
+ },
+ {
+ "epoch": 3.6866485013623977,
+ "grad_norm": 7.893728733062744,
+ "learning_rate": 1.999752726637131e-05,
+ "loss": 1.7332,
+ "step": 1353
+ },
+ {
+ "epoch": 3.689373297002725,
+ "grad_norm": 7.071447372436523,
+ "learning_rate": 1.9997507603379883e-05,
+ "loss": 1.7056,
+ "step": 1354
+ },
+ {
+ "epoch": 3.692098092643052,
+ "grad_norm": 7.456488132476807,
+ "learning_rate": 1.9997487862528186e-05,
+ "loss": 1.8623,
+ "step": 1355
+ },
+ {
+ "epoch": 3.6948228882833787,
+ "grad_norm": 7.914857387542725,
+ "learning_rate": 1.9997468043816376e-05,
+ "loss": 1.5439,
+ "step": 1356
+ },
+ {
+ "epoch": 3.697547683923706,
+ "grad_norm": 9.582695960998535,
+ "learning_rate": 1.9997448147244607e-05,
+ "loss": 1.8726,
+ "step": 1357
+ },
+ {
+ "epoch": 3.700272479564033,
+ "grad_norm": 5.854775428771973,
+ "learning_rate": 1.999742817281303e-05,
+ "loss": 1.7258,
+ "step": 1358
+ },
+ {
+ "epoch": 3.7029972752043596,
+ "grad_norm": 7.519739627838135,
+ "learning_rate": 1.999740812052181e-05,
+ "loss": 1.6487,
+ "step": 1359
+ },
+ {
+ "epoch": 3.7057220708446867,
+ "grad_norm": 5.872730731964111,
+ "learning_rate": 1.9997387990371092e-05,
+ "loss": 1.6479,
+ "step": 1360
+ },
+ {
+ "epoch": 3.708446866485014,
+ "grad_norm": 7.478868007659912,
+ "learning_rate": 1.9997367782361042e-05,
+ "loss": 1.9268,
+ "step": 1361
+ },
+ {
+ "epoch": 3.7111716621253406,
+ "grad_norm": 8.448175430297852,
+ "learning_rate": 1.9997347496491812e-05,
+ "loss": 1.8159,
+ "step": 1362
+ },
+ {
+ "epoch": 3.7138964577656677,
+ "grad_norm": 6.642178058624268,
+ "learning_rate": 1.999732713276356e-05,
+ "loss": 1.6938,
+ "step": 1363
+ },
+ {
+ "epoch": 3.716621253405995,
+ "grad_norm": 6.929697036743164,
+ "learning_rate": 1.9997306691176448e-05,
+ "loss": 1.6719,
+ "step": 1364
+ },
+ {
+ "epoch": 3.7193460490463215,
+ "grad_norm": 7.296602249145508,
+ "learning_rate": 1.9997286171730632e-05,
+ "loss": 1.5879,
+ "step": 1365
+ },
+ {
+ "epoch": 3.7220708446866486,
+ "grad_norm": 6.695741653442383,
+ "learning_rate": 1.9997265574426276e-05,
+ "loss": 1.4265,
+ "step": 1366
+ },
+ {
+ "epoch": 3.7247956403269757,
+ "grad_norm": 8.240156173706055,
+ "learning_rate": 1.9997244899263536e-05,
+ "loss": 1.7512,
+ "step": 1367
+ },
+ {
+ "epoch": 3.7275204359673024,
+ "grad_norm": 7.536651611328125,
+ "learning_rate": 1.999722414624257e-05,
+ "loss": 1.8296,
+ "step": 1368
+ },
+ {
+ "epoch": 3.7302452316076296,
+ "grad_norm": 8.759150505065918,
+ "learning_rate": 1.999720331536355e-05,
+ "loss": 1.9648,
+ "step": 1369
+ },
+ {
+ "epoch": 3.7329700272479567,
+ "grad_norm": 8.0873384475708,
+ "learning_rate": 1.999718240662663e-05,
+ "loss": 1.7673,
+ "step": 1370
+ },
+ {
+ "epoch": 3.7356948228882834,
+ "grad_norm": 6.582300662994385,
+ "learning_rate": 1.9997161420031975e-05,
+ "loss": 1.6719,
+ "step": 1371
+ },
+ {
+ "epoch": 3.7384196185286105,
+ "grad_norm": 6.851629734039307,
+ "learning_rate": 1.999714035557975e-05,
+ "loss": 1.7725,
+ "step": 1372
+ },
+ {
+ "epoch": 3.741144414168937,
+ "grad_norm": 8.481106758117676,
+ "learning_rate": 1.9997119213270117e-05,
+ "loss": 1.8611,
+ "step": 1373
+ },
+ {
+ "epoch": 3.7438692098092643,
+ "grad_norm": 7.501500606536865,
+ "learning_rate": 1.999709799310324e-05,
+ "loss": 1.7134,
+ "step": 1374
+ },
+ {
+ "epoch": 3.7465940054495914,
+ "grad_norm": 8.318303108215332,
+ "learning_rate": 1.999707669507929e-05,
+ "loss": 1.7451,
+ "step": 1375
+ },
+ {
+ "epoch": 3.749318801089918,
+ "grad_norm": 8.081513404846191,
+ "learning_rate": 1.9997055319198422e-05,
+ "loss": 1.738,
+ "step": 1376
+ },
+ {
+ "epoch": 3.7520435967302452,
+ "grad_norm": 8.949186325073242,
+ "learning_rate": 1.9997033865460812e-05,
+ "loss": 1.5828,
+ "step": 1377
+ },
+ {
+ "epoch": 3.7547683923705724,
+ "grad_norm": 7.720224857330322,
+ "learning_rate": 1.9997012333866623e-05,
+ "loss": 1.6499,
+ "step": 1378
+ },
+ {
+ "epoch": 3.757493188010899,
+ "grad_norm": 6.31644344329834,
+ "learning_rate": 1.9996990724416022e-05,
+ "loss": 1.6658,
+ "step": 1379
+ },
+ {
+ "epoch": 3.760217983651226,
+ "grad_norm": 8.776480674743652,
+ "learning_rate": 1.9996969037109185e-05,
+ "loss": 1.7622,
+ "step": 1380
+ },
+ {
+ "epoch": 3.7629427792915533,
+ "grad_norm": 6.267784595489502,
+ "learning_rate": 1.999694727194627e-05,
+ "loss": 1.4507,
+ "step": 1381
+ },
+ {
+ "epoch": 3.76566757493188,
+ "grad_norm": 6.604657173156738,
+ "learning_rate": 1.9996925428927456e-05,
+ "loss": 1.8013,
+ "step": 1382
+ },
+ {
+ "epoch": 3.768392370572207,
+ "grad_norm": 6.861454486846924,
+ "learning_rate": 1.9996903508052903e-05,
+ "loss": 1.6631,
+ "step": 1383
+ },
+ {
+ "epoch": 3.771117166212534,
+ "grad_norm": 13.694665908813477,
+ "learning_rate": 1.999688150932279e-05,
+ "loss": 1.6819,
+ "step": 1384
+ },
+ {
+ "epoch": 3.773841961852861,
+ "grad_norm": 8.601831436157227,
+ "learning_rate": 1.9996859432737287e-05,
+ "loss": 1.8052,
+ "step": 1385
+ },
+ {
+ "epoch": 3.776566757493188,
+ "grad_norm": 9.151429176330566,
+ "learning_rate": 1.999683727829656e-05,
+ "loss": 1.6763,
+ "step": 1386
+ },
+ {
+ "epoch": 3.7792915531335147,
+ "grad_norm": 7.394098281860352,
+ "learning_rate": 1.999681504600079e-05,
+ "loss": 1.738,
+ "step": 1387
+ },
+ {
+ "epoch": 3.782016348773842,
+ "grad_norm": 7.859614372253418,
+ "learning_rate": 1.9996792735850145e-05,
+ "loss": 1.772,
+ "step": 1388
+ },
+ {
+ "epoch": 3.784741144414169,
+ "grad_norm": 8.550800323486328,
+ "learning_rate": 1.99967703478448e-05,
+ "loss": 1.626,
+ "step": 1389
+ },
+ {
+ "epoch": 3.7874659400544957,
+ "grad_norm": 6.659253120422363,
+ "learning_rate": 1.9996747881984928e-05,
+ "loss": 1.6821,
+ "step": 1390
+ },
+ {
+ "epoch": 3.790190735694823,
+ "grad_norm": 6.841320991516113,
+ "learning_rate": 1.9996725338270705e-05,
+ "loss": 1.772,
+ "step": 1391
+ },
+ {
+ "epoch": 3.79291553133515,
+ "grad_norm": 8.40398120880127,
+ "learning_rate": 1.999670271670231e-05,
+ "loss": 1.7151,
+ "step": 1392
+ },
+ {
+ "epoch": 3.7956403269754766,
+ "grad_norm": 6.894169807434082,
+ "learning_rate": 1.9996680017279916e-05,
+ "loss": 1.6277,
+ "step": 1393
+ },
+ {
+ "epoch": 3.7983651226158037,
+ "grad_norm": 6.456517696380615,
+ "learning_rate": 1.9996657240003692e-05,
+ "loss": 1.5464,
+ "step": 1394
+ },
+ {
+ "epoch": 3.801089918256131,
+ "grad_norm": 8.566692352294922,
+ "learning_rate": 1.9996634384873832e-05,
+ "loss": 1.5122,
+ "step": 1395
+ },
+ {
+ "epoch": 3.8038147138964575,
+ "grad_norm": 8.019652366638184,
+ "learning_rate": 1.9996611451890503e-05,
+ "loss": 1.6104,
+ "step": 1396
+ },
+ {
+ "epoch": 3.8065395095367847,
+ "grad_norm": 8.956162452697754,
+ "learning_rate": 1.9996588441053886e-05,
+ "loss": 1.8037,
+ "step": 1397
+ },
+ {
+ "epoch": 3.809264305177112,
+ "grad_norm": 7.350297927856445,
+ "learning_rate": 1.9996565352364156e-05,
+ "loss": 1.6694,
+ "step": 1398
+ },
+ {
+ "epoch": 3.8119891008174385,
+ "grad_norm": 12.742020606994629,
+ "learning_rate": 1.99965421858215e-05,
+ "loss": 1.8008,
+ "step": 1399
+ },
+ {
+ "epoch": 3.8147138964577656,
+ "grad_norm": 9.180830001831055,
+ "learning_rate": 1.9996518941426096e-05,
+ "loss": 1.7876,
+ "step": 1400
+ },
+ {
+ "epoch": 3.8174386920980927,
+ "grad_norm": 6.453849792480469,
+ "learning_rate": 1.9996495619178124e-05,
+ "loss": 1.6824,
+ "step": 1401
+ },
+ {
+ "epoch": 3.8201634877384194,
+ "grad_norm": 6.997298240661621,
+ "learning_rate": 1.999647221907776e-05,
+ "loss": 1.7915,
+ "step": 1402
+ },
+ {
+ "epoch": 3.8228882833787465,
+ "grad_norm": 7.191959381103516,
+ "learning_rate": 1.99964487411252e-05,
+ "loss": 1.7153,
+ "step": 1403
+ },
+ {
+ "epoch": 3.8256130790190737,
+ "grad_norm": 7.514167308807373,
+ "learning_rate": 1.9996425185320616e-05,
+ "loss": 1.6958,
+ "step": 1404
+ },
+ {
+ "epoch": 3.8283378746594003,
+ "grad_norm": 6.850180625915527,
+ "learning_rate": 1.9996401551664197e-05,
+ "loss": 1.5608,
+ "step": 1405
+ },
+ {
+ "epoch": 3.8310626702997275,
+ "grad_norm": 7.981141567230225,
+ "learning_rate": 1.999637784015612e-05,
+ "loss": 1.6543,
+ "step": 1406
+ },
+ {
+ "epoch": 3.8337874659400546,
+ "grad_norm": 8.2358980178833,
+ "learning_rate": 1.9996354050796577e-05,
+ "loss": 1.5713,
+ "step": 1407
+ },
+ {
+ "epoch": 3.8365122615803813,
+ "grad_norm": 7.5839104652404785,
+ "learning_rate": 1.9996330183585747e-05,
+ "loss": 1.6919,
+ "step": 1408
+ },
+ {
+ "epoch": 3.8392370572207084,
+ "grad_norm": 7.587404727935791,
+ "learning_rate": 1.999630623852382e-05,
+ "loss": 1.771,
+ "step": 1409
+ },
+ {
+ "epoch": 3.8419618528610355,
+ "grad_norm": 8.106740951538086,
+ "learning_rate": 1.9996282215610984e-05,
+ "loss": 1.4944,
+ "step": 1410
+ },
+ {
+ "epoch": 3.844686648501362,
+ "grad_norm": 37.11008834838867,
+ "learning_rate": 1.9996258114847427e-05,
+ "loss": 1.5405,
+ "step": 1411
+ },
+ {
+ "epoch": 3.8474114441416893,
+ "grad_norm": 8.352994918823242,
+ "learning_rate": 1.999623393623333e-05,
+ "loss": 1.8989,
+ "step": 1412
+ },
+ {
+ "epoch": 3.8501362397820165,
+ "grad_norm": 7.6126203536987305,
+ "learning_rate": 1.999620967976888e-05,
+ "loss": 1.6052,
+ "step": 1413
+ },
+ {
+ "epoch": 3.852861035422343,
+ "grad_norm": 9.373207092285156,
+ "learning_rate": 1.999618534545428e-05,
+ "loss": 1.5542,
+ "step": 1414
+ },
+ {
+ "epoch": 3.8555858310626703,
+ "grad_norm": 7.17564582824707,
+ "learning_rate": 1.9996160933289704e-05,
+ "loss": 1.79,
+ "step": 1415
+ },
+ {
+ "epoch": 3.8583106267029974,
+ "grad_norm": 7.9729695320129395,
+ "learning_rate": 1.999613644327535e-05,
+ "loss": 1.6101,
+ "step": 1416
+ },
+ {
+ "epoch": 3.861035422343324,
+ "grad_norm": 8.744307518005371,
+ "learning_rate": 1.9996111875411407e-05,
+ "loss": 1.7905,
+ "step": 1417
+ },
+ {
+ "epoch": 3.863760217983651,
+ "grad_norm": 6.597487926483154,
+ "learning_rate": 1.9996087229698064e-05,
+ "loss": 1.6567,
+ "step": 1418
+ },
+ {
+ "epoch": 3.8664850136239783,
+ "grad_norm": 8.053556442260742,
+ "learning_rate": 1.9996062506135518e-05,
+ "loss": 1.5027,
+ "step": 1419
+ },
+ {
+ "epoch": 3.869209809264305,
+ "grad_norm": 7.315485000610352,
+ "learning_rate": 1.999603770472396e-05,
+ "loss": 1.4714,
+ "step": 1420
+ },
+ {
+ "epoch": 3.871934604904632,
+ "grad_norm": 7.823679447174072,
+ "learning_rate": 1.999601282546358e-05,
+ "loss": 1.7585,
+ "step": 1421
+ },
+ {
+ "epoch": 3.8746594005449593,
+ "grad_norm": 8.394383430480957,
+ "learning_rate": 1.9995987868354577e-05,
+ "loss": 1.782,
+ "step": 1422
+ },
+ {
+ "epoch": 3.877384196185286,
+ "grad_norm": 11.648750305175781,
+ "learning_rate": 1.999596283339714e-05,
+ "loss": 1.8115,
+ "step": 1423
+ },
+ {
+ "epoch": 3.880108991825613,
+ "grad_norm": 38.12660217285156,
+ "learning_rate": 1.9995937720591464e-05,
+ "loss": 1.7202,
+ "step": 1424
+ },
+ {
+ "epoch": 3.88283378746594,
+ "grad_norm": 11.35765266418457,
+ "learning_rate": 1.999591252993775e-05,
+ "loss": 1.6924,
+ "step": 1425
+ },
+ {
+ "epoch": 3.885558583106267,
+ "grad_norm": 7.928035736083984,
+ "learning_rate": 1.9995887261436188e-05,
+ "loss": 1.7451,
+ "step": 1426
+ },
+ {
+ "epoch": 3.888283378746594,
+ "grad_norm": 10.286693572998047,
+ "learning_rate": 1.999586191508698e-05,
+ "loss": 1.6907,
+ "step": 1427
+ },
+ {
+ "epoch": 3.891008174386921,
+ "grad_norm": 11.02782917022705,
+ "learning_rate": 1.999583649089032e-05,
+ "loss": 1.8323,
+ "step": 1428
+ },
+ {
+ "epoch": 3.893732970027248,
+ "grad_norm": 8.212396621704102,
+ "learning_rate": 1.999581098884641e-05,
+ "loss": 1.6035,
+ "step": 1429
+ },
+ {
+ "epoch": 3.896457765667575,
+ "grad_norm": 7.471797466278076,
+ "learning_rate": 1.999578540895544e-05,
+ "loss": 1.7646,
+ "step": 1430
+ },
+ {
+ "epoch": 3.899182561307902,
+ "grad_norm": 13.852618217468262,
+ "learning_rate": 1.9995759751217617e-05,
+ "loss": 1.8142,
+ "step": 1431
+ },
+ {
+ "epoch": 3.9019073569482288,
+ "grad_norm": 11.58817195892334,
+ "learning_rate": 1.999573401563314e-05,
+ "loss": 1.6655,
+ "step": 1432
+ },
+ {
+ "epoch": 3.904632152588556,
+ "grad_norm": 14.474498748779297,
+ "learning_rate": 1.9995708202202212e-05,
+ "loss": 1.6169,
+ "step": 1433
+ },
+ {
+ "epoch": 3.907356948228883,
+ "grad_norm": 10.655352592468262,
+ "learning_rate": 1.9995682310925025e-05,
+ "loss": 1.8713,
+ "step": 1434
+ },
+ {
+ "epoch": 3.9100817438692097,
+ "grad_norm": 15.403002738952637,
+ "learning_rate": 1.9995656341801788e-05,
+ "loss": 1.6191,
+ "step": 1435
+ },
+ {
+ "epoch": 3.912806539509537,
+ "grad_norm": 12.528199195861816,
+ "learning_rate": 1.9995630294832698e-05,
+ "loss": 1.7893,
+ "step": 1436
+ },
+ {
+ "epoch": 3.915531335149864,
+ "grad_norm": 9.763728141784668,
+ "learning_rate": 1.9995604170017967e-05,
+ "loss": 1.5823,
+ "step": 1437
+ },
+ {
+ "epoch": 3.9182561307901906,
+ "grad_norm": 10.758548736572266,
+ "learning_rate": 1.9995577967357788e-05,
+ "loss": 1.8389,
+ "step": 1438
+ },
+ {
+ "epoch": 3.9209809264305178,
+ "grad_norm": 10.564231872558594,
+ "learning_rate": 1.999555168685237e-05,
+ "loss": 1.781,
+ "step": 1439
+ },
+ {
+ "epoch": 3.923705722070845,
+ "grad_norm": 9.315851211547852,
+ "learning_rate": 1.999552532850192e-05,
+ "loss": 1.8865,
+ "step": 1440
+ },
+ {
+ "epoch": 3.9264305177111716,
+ "grad_norm": 56.96832275390625,
+ "learning_rate": 1.9995498892306634e-05,
+ "loss": 1.6589,
+ "step": 1441
+ },
+ {
+ "epoch": 3.9291553133514987,
+ "grad_norm": 8.838242530822754,
+ "learning_rate": 1.999547237826673e-05,
+ "loss": 1.3882,
+ "step": 1442
+ },
+ {
+ "epoch": 3.931880108991826,
+ "grad_norm": 13.597760200500488,
+ "learning_rate": 1.9995445786382408e-05,
+ "loss": 1.7344,
+ "step": 1443
+ },
+ {
+ "epoch": 3.9346049046321525,
+ "grad_norm": 14.991342544555664,
+ "learning_rate": 1.9995419116653876e-05,
+ "loss": 1.6475,
+ "step": 1444
+ },
+ {
+ "epoch": 3.9373297002724796,
+ "grad_norm": 14.62321949005127,
+ "learning_rate": 1.999539236908134e-05,
+ "loss": 1.9482,
+ "step": 1445
+ },
+ {
+ "epoch": 3.9400544959128068,
+ "grad_norm": 14.784954071044922,
+ "learning_rate": 1.999536554366501e-05,
+ "loss": 1.9209,
+ "step": 1446
+ },
+ {
+ "epoch": 3.9427792915531334,
+ "grad_norm": 14.615570068359375,
+ "learning_rate": 1.9995338640405096e-05,
+ "loss": 1.75,
+ "step": 1447
+ },
+ {
+ "epoch": 3.9455040871934606,
+ "grad_norm": 39.841617584228516,
+ "learning_rate": 1.9995311659301804e-05,
+ "loss": 1.625,
+ "step": 1448
+ },
+ {
+ "epoch": 3.9482288828337877,
+ "grad_norm": 12.779487609863281,
+ "learning_rate": 1.9995284600355348e-05,
+ "loss": 1.7417,
+ "step": 1449
+ },
+ {
+ "epoch": 3.9509536784741144,
+ "grad_norm": 10.443445205688477,
+ "learning_rate": 1.9995257463565937e-05,
+ "loss": 1.4424,
+ "step": 1450
+ },
+ {
+ "epoch": 3.9536784741144415,
+ "grad_norm": 12.623815536499023,
+ "learning_rate": 1.9995230248933784e-05,
+ "loss": 1.6353,
+ "step": 1451
+ },
+ {
+ "epoch": 3.9564032697547686,
+ "grad_norm": 9.363016128540039,
+ "learning_rate": 1.99952029564591e-05,
+ "loss": 1.6353,
+ "step": 1452
+ },
+ {
+ "epoch": 3.9591280653950953,
+ "grad_norm": 14.185697555541992,
+ "learning_rate": 1.9995175586142096e-05,
+ "loss": 1.9553,
+ "step": 1453
+ },
+ {
+ "epoch": 3.9618528610354224,
+ "grad_norm": 11.08867073059082,
+ "learning_rate": 1.9995148137982984e-05,
+ "loss": 1.7131,
+ "step": 1454
+ },
+ {
+ "epoch": 3.9645776566757496,
+ "grad_norm": 11.309182167053223,
+ "learning_rate": 1.9995120611981984e-05,
+ "loss": 1.8123,
+ "step": 1455
+ },
+ {
+ "epoch": 3.9673024523160763,
+ "grad_norm": 8.553887367248535,
+ "learning_rate": 1.9995093008139304e-05,
+ "loss": 1.468,
+ "step": 1456
+ },
+ {
+ "epoch": 3.9700272479564034,
+ "grad_norm": 7.511438846588135,
+ "learning_rate": 1.999506532645516e-05,
+ "loss": 1.7515,
+ "step": 1457
+ },
+ {
+ "epoch": 3.9727520435967305,
+ "grad_norm": 9.916815757751465,
+ "learning_rate": 1.9995037566929772e-05,
+ "loss": 1.7085,
+ "step": 1458
+ },
+ {
+ "epoch": 3.975476839237057,
+ "grad_norm": 11.643606185913086,
+ "learning_rate": 1.999500972956335e-05,
+ "loss": 1.5674,
+ "step": 1459
+ },
+ {
+ "epoch": 3.9782016348773843,
+ "grad_norm": 13.842655181884766,
+ "learning_rate": 1.9994981814356116e-05,
+ "loss": 1.5847,
+ "step": 1460
+ },
+ {
+ "epoch": 3.9809264305177114,
+ "grad_norm": 8.34160327911377,
+ "learning_rate": 1.9994953821308284e-05,
+ "loss": 1.6272,
+ "step": 1461
+ },
+ {
+ "epoch": 3.983651226158038,
+ "grad_norm": 7.717387676239014,
+ "learning_rate": 1.9994925750420077e-05,
+ "loss": 1.8,
+ "step": 1462
+ },
+ {
+ "epoch": 3.9863760217983653,
+ "grad_norm": 8.668373107910156,
+ "learning_rate": 1.9994897601691706e-05,
+ "loss": 1.5916,
+ "step": 1463
+ },
+ {
+ "epoch": 3.989100817438692,
+ "grad_norm": 22.320735931396484,
+ "learning_rate": 1.9994869375123397e-05,
+ "loss": 1.5747,
+ "step": 1464
+ },
+ {
+ "epoch": 3.991825613079019,
+ "grad_norm": 13.898236274719238,
+ "learning_rate": 1.9994841070715365e-05,
+ "loss": 1.6787,
+ "step": 1465
+ },
+ {
+ "epoch": 3.994550408719346,
+ "grad_norm": 8.619722366333008,
+ "learning_rate": 1.999481268846783e-05,
+ "loss": 1.5227,
+ "step": 1466
+ },
+ {
+ "epoch": 3.997275204359673,
+ "grad_norm": 8.676521301269531,
+ "learning_rate": 1.9994784228381018e-05,
+ "loss": 1.7715,
+ "step": 1467
+ },
+ {
+ "epoch": 4.0,
+ "grad_norm": 8.700546264648438,
+ "learning_rate": 1.9994755690455154e-05,
+ "loss": 1.7297,
+ "step": 1468
+ },
+ {
+ "epoch": 4.002724795640327,
+ "grad_norm": 19.045896530151367,
+ "learning_rate": 1.9994727074690444e-05,
+ "loss": 1.6096,
+ "step": 1469
+ },
+ {
+ "epoch": 4.005449591280654,
+ "grad_norm": 22.447357177734375,
+ "learning_rate": 1.9994698381087128e-05,
+ "loss": 1.7842,
+ "step": 1470
+ },
+ {
+ "epoch": 4.008174386920981,
+ "grad_norm": 10.952781677246094,
+ "learning_rate": 1.999466960964542e-05,
+ "loss": 1.6987,
+ "step": 1471
+ },
+ {
+ "epoch": 4.010899182561308,
+ "grad_norm": 12.661768913269043,
+ "learning_rate": 1.9994640760365545e-05,
+ "loss": 1.7905,
+ "step": 1472
+ },
+ {
+ "epoch": 4.013623978201635,
+ "grad_norm": 9.215839385986328,
+ "learning_rate": 1.9994611833247733e-05,
+ "loss": 1.3618,
+ "step": 1473
+ },
+ {
+ "epoch": 4.016348773841962,
+ "grad_norm": 7.643862247467041,
+ "learning_rate": 1.9994582828292205e-05,
+ "loss": 1.47,
+ "step": 1474
+ },
+ {
+ "epoch": 4.0190735694822886,
+ "grad_norm": 12.22031307220459,
+ "learning_rate": 1.9994553745499187e-05,
+ "loss": 1.7747,
+ "step": 1475
+ },
+ {
+ "epoch": 4.021798365122616,
+ "grad_norm": 9.688786506652832,
+ "learning_rate": 1.9994524584868907e-05,
+ "loss": 1.6509,
+ "step": 1476
+ },
+ {
+ "epoch": 4.024523160762943,
+ "grad_norm": 8.996979713439941,
+ "learning_rate": 1.999449534640159e-05,
+ "loss": 1.6997,
+ "step": 1477
+ },
+ {
+ "epoch": 4.0272479564032695,
+ "grad_norm": 11.732244491577148,
+ "learning_rate": 1.9994466030097466e-05,
+ "loss": 1.5132,
+ "step": 1478
+ },
+ {
+ "epoch": 4.029972752043597,
+ "grad_norm": 8.356077194213867,
+ "learning_rate": 1.9994436635956763e-05,
+ "loss": 1.7354,
+ "step": 1479
+ },
+ {
+ "epoch": 4.032697547683924,
+ "grad_norm": 11.178964614868164,
+ "learning_rate": 1.999440716397971e-05,
+ "loss": 1.6614,
+ "step": 1480
+ },
+ {
+ "epoch": 4.03542234332425,
+ "grad_norm": 9.696405410766602,
+ "learning_rate": 1.999437761416653e-05,
+ "loss": 1.6611,
+ "step": 1481
+ },
+ {
+ "epoch": 4.038147138964578,
+ "grad_norm": 14.273547172546387,
+ "learning_rate": 1.9994347986517462e-05,
+ "loss": 1.7249,
+ "step": 1482
+ },
+ {
+ "epoch": 4.040871934604905,
+ "grad_norm": 11.215656280517578,
+ "learning_rate": 1.9994318281032733e-05,
+ "loss": 1.6819,
+ "step": 1483
+ },
+ {
+ "epoch": 4.043596730245231,
+ "grad_norm": 9.931842803955078,
+ "learning_rate": 1.9994288497712575e-05,
+ "loss": 2.0005,
+ "step": 1484
+ },
+ {
+ "epoch": 4.046321525885559,
+ "grad_norm": 7.003057479858398,
+ "learning_rate": 1.999425863655722e-05,
+ "loss": 1.6094,
+ "step": 1485
+ },
+ {
+ "epoch": 4.049046321525886,
+ "grad_norm": 11.406493186950684,
+ "learning_rate": 1.9994228697566905e-05,
+ "loss": 1.5625,
+ "step": 1486
+ },
+ {
+ "epoch": 4.051771117166212,
+ "grad_norm": 9.449951171875,
+ "learning_rate": 1.999419868074185e-05,
+ "loss": 1.6687,
+ "step": 1487
+ },
+ {
+ "epoch": 4.05449591280654,
+ "grad_norm": 8.297751426696777,
+ "learning_rate": 1.9994168586082305e-05,
+ "loss": 1.4631,
+ "step": 1488
+ },
+ {
+ "epoch": 4.0572207084468666,
+ "grad_norm": 7.9022908210754395,
+ "learning_rate": 1.999413841358849e-05,
+ "loss": 1.5723,
+ "step": 1489
+ },
+ {
+ "epoch": 4.059945504087193,
+ "grad_norm": 7.0970458984375,
+ "learning_rate": 1.999410816326065e-05,
+ "loss": 1.3599,
+ "step": 1490
+ },
+ {
+ "epoch": 4.062670299727521,
+ "grad_norm": 10.842693328857422,
+ "learning_rate": 1.999407783509902e-05,
+ "loss": 1.749,
+ "step": 1491
+ },
+ {
+ "epoch": 4.0653950953678475,
+ "grad_norm": 7.488298416137695,
+ "learning_rate": 1.9994047429103832e-05,
+ "loss": 1.4102,
+ "step": 1492
+ },
+ {
+ "epoch": 4.068119891008174,
+ "grad_norm": 8.624256134033203,
+ "learning_rate": 1.9994016945275324e-05,
+ "loss": 1.801,
+ "step": 1493
+ },
+ {
+ "epoch": 4.070844686648502,
+ "grad_norm": 8.191618919372559,
+ "learning_rate": 1.999398638361373e-05,
+ "loss": 1.5208,
+ "step": 1494
+ },
+ {
+ "epoch": 4.073569482288828,
+ "grad_norm": 8.972846031188965,
+ "learning_rate": 1.9993955744119297e-05,
+ "loss": 1.5217,
+ "step": 1495
+ },
+ {
+ "epoch": 4.076294277929155,
+ "grad_norm": 9.642993927001953,
+ "learning_rate": 1.9993925026792252e-05,
+ "loss": 1.5923,
+ "step": 1496
+ },
+ {
+ "epoch": 4.079019073569483,
+ "grad_norm": 17.76791763305664,
+ "learning_rate": 1.9993894231632847e-05,
+ "loss": 1.6638,
+ "step": 1497
+ },
+ {
+ "epoch": 4.081743869209809,
+ "grad_norm": 7.981385707855225,
+ "learning_rate": 1.9993863358641313e-05,
+ "loss": 1.6207,
+ "step": 1498
+ },
+ {
+ "epoch": 4.084468664850136,
+ "grad_norm": 9.830915451049805,
+ "learning_rate": 1.9993832407817893e-05,
+ "loss": 1.4187,
+ "step": 1499
+ },
+ {
+ "epoch": 4.087193460490464,
+ "grad_norm": 10.941875457763672,
+ "learning_rate": 1.999380137916283e-05,
+ "loss": 1.7397,
+ "step": 1500
+ },
+ {
+ "epoch": 4.08991825613079,
+ "grad_norm": 7.904561519622803,
+ "learning_rate": 1.9993770272676358e-05,
+ "loss": 1.7231,
+ "step": 1501
+ },
+ {
+ "epoch": 4.092643051771117,
+ "grad_norm": 10.005932807922363,
+ "learning_rate": 1.9993739088358728e-05,
+ "loss": 1.521,
+ "step": 1502
+ },
+ {
+ "epoch": 4.0953678474114446,
+ "grad_norm": 8.690088272094727,
+ "learning_rate": 1.999370782621018e-05,
+ "loss": 1.7207,
+ "step": 1503
+ },
+ {
+ "epoch": 4.098092643051771,
+ "grad_norm": 7.502809524536133,
+ "learning_rate": 1.9993676486230958e-05,
+ "loss": 1.6072,
+ "step": 1504
+ },
+ {
+ "epoch": 4.100817438692098,
+ "grad_norm": 9.120586395263672,
+ "learning_rate": 1.99936450684213e-05,
+ "loss": 1.5757,
+ "step": 1505
+ },
+ {
+ "epoch": 4.1035422343324255,
+ "grad_norm": 9.650954246520996,
+ "learning_rate": 1.9993613572781463e-05,
+ "loss": 1.428,
+ "step": 1506
+ },
+ {
+ "epoch": 4.106267029972752,
+ "grad_norm": 7.417420864105225,
+ "learning_rate": 1.9993581999311683e-05,
+ "loss": 1.5776,
+ "step": 1507
+ },
+ {
+ "epoch": 4.108991825613079,
+ "grad_norm": 7.51955509185791,
+ "learning_rate": 1.9993550348012202e-05,
+ "loss": 1.7192,
+ "step": 1508
+ },
+ {
+ "epoch": 4.111716621253406,
+ "grad_norm": 10.530597686767578,
+ "learning_rate": 1.9993518618883278e-05,
+ "loss": 1.4868,
+ "step": 1509
+ },
+ {
+ "epoch": 4.114441416893733,
+ "grad_norm": 11.161538124084473,
+ "learning_rate": 1.9993486811925152e-05,
+ "loss": 1.7754,
+ "step": 1510
+ },
+ {
+ "epoch": 4.11716621253406,
+ "grad_norm": 11.958690643310547,
+ "learning_rate": 1.9993454927138075e-05,
+ "loss": 1.5588,
+ "step": 1511
+ },
+ {
+ "epoch": 4.1198910081743865,
+ "grad_norm": 11.982670783996582,
+ "learning_rate": 1.999342296452229e-05,
+ "loss": 1.7695,
+ "step": 1512
+ },
+ {
+ "epoch": 4.122615803814714,
+ "grad_norm": 17.812929153442383,
+ "learning_rate": 1.9993390924078048e-05,
+ "loss": 1.4983,
+ "step": 1513
+ },
+ {
+ "epoch": 4.125340599455041,
+ "grad_norm": 9.266668319702148,
+ "learning_rate": 1.99933588058056e-05,
+ "loss": 1.4363,
+ "step": 1514
+ },
+ {
+ "epoch": 4.128065395095367,
+ "grad_norm": 9.505965232849121,
+ "learning_rate": 1.9993326609705193e-05,
+ "loss": 1.6885,
+ "step": 1515
+ },
+ {
+ "epoch": 4.130790190735695,
+ "grad_norm": 12.357368469238281,
+ "learning_rate": 1.999329433577708e-05,
+ "loss": 1.4883,
+ "step": 1516
+ },
+ {
+ "epoch": 4.133514986376022,
+ "grad_norm": 9.566493034362793,
+ "learning_rate": 1.9993261984021516e-05,
+ "loss": 1.6785,
+ "step": 1517
+ },
+ {
+ "epoch": 4.136239782016348,
+ "grad_norm": 15.662911415100098,
+ "learning_rate": 1.9993229554438748e-05,
+ "loss": 1.5293,
+ "step": 1518
+ },
+ {
+ "epoch": 4.138964577656676,
+ "grad_norm": 10.700569152832031,
+ "learning_rate": 1.999319704702903e-05,
+ "loss": 1.8071,
+ "step": 1519
+ },
+ {
+ "epoch": 4.141689373297003,
+ "grad_norm": 9.475998878479004,
+ "learning_rate": 1.9993164461792617e-05,
+ "loss": 1.3359,
+ "step": 1520
+ },
+ {
+ "epoch": 4.144414168937329,
+ "grad_norm": 8.790034294128418,
+ "learning_rate": 1.9993131798729757e-05,
+ "loss": 1.7146,
+ "step": 1521
+ },
+ {
+ "epoch": 4.147138964577657,
+ "grad_norm": 9.035723686218262,
+ "learning_rate": 1.999309905784071e-05,
+ "loss": 1.4546,
+ "step": 1522
+ },
+ {
+ "epoch": 4.1498637602179835,
+ "grad_norm": 14.3681640625,
+ "learning_rate": 1.9993066239125726e-05,
+ "loss": 1.4983,
+ "step": 1523
+ },
+ {
+ "epoch": 4.15258855585831,
+ "grad_norm": 8.432818412780762,
+ "learning_rate": 1.9993033342585068e-05,
+ "loss": 1.4534,
+ "step": 1524
+ },
+ {
+ "epoch": 4.155313351498638,
+ "grad_norm": 8.936095237731934,
+ "learning_rate": 1.9993000368218985e-05,
+ "loss": 1.7019,
+ "step": 1525
+ },
+ {
+ "epoch": 4.1580381471389645,
+ "grad_norm": 8.83277416229248,
+ "learning_rate": 1.9992967316027738e-05,
+ "loss": 1.6741,
+ "step": 1526
+ },
+ {
+ "epoch": 4.160762942779291,
+ "grad_norm": 10.384167671203613,
+ "learning_rate": 1.9992934186011582e-05,
+ "loss": 1.4677,
+ "step": 1527
+ },
+ {
+ "epoch": 4.163487738419619,
+ "grad_norm": 8.396014213562012,
+ "learning_rate": 1.999290097817078e-05,
+ "loss": 1.811,
+ "step": 1528
+ },
+ {
+ "epoch": 4.166212534059945,
+ "grad_norm": 8.714115142822266,
+ "learning_rate": 1.9992867692505582e-05,
+ "loss": 1.5088,
+ "step": 1529
+ },
+ {
+ "epoch": 4.168937329700272,
+ "grad_norm": 10.128386497497559,
+ "learning_rate": 1.9992834329016256e-05,
+ "loss": 1.6638,
+ "step": 1530
+ },
+ {
+ "epoch": 4.1716621253406,
+ "grad_norm": 9.742781639099121,
+ "learning_rate": 1.9992800887703057e-05,
+ "loss": 1.6484,
+ "step": 1531
+ },
+ {
+ "epoch": 4.174386920980926,
+ "grad_norm": 16.12179183959961,
+ "learning_rate": 1.9992767368566244e-05,
+ "loss": 1.5925,
+ "step": 1532
+ },
+ {
+ "epoch": 4.177111716621253,
+ "grad_norm": 8.934818267822266,
+ "learning_rate": 1.9992733771606082e-05,
+ "loss": 1.5288,
+ "step": 1533
+ },
+ {
+ "epoch": 4.179836512261581,
+ "grad_norm": 9.01435375213623,
+ "learning_rate": 1.999270009682283e-05,
+ "loss": 1.541,
+ "step": 1534
+ },
+ {
+ "epoch": 4.182561307901907,
+ "grad_norm": 14.064192771911621,
+ "learning_rate": 1.9992666344216754e-05,
+ "loss": 1.7253,
+ "step": 1535
+ },
+ {
+ "epoch": 4.185286103542234,
+ "grad_norm": 8.600523948669434,
+ "learning_rate": 1.999263251378811e-05,
+ "loss": 1.7197,
+ "step": 1536
+ },
+ {
+ "epoch": 4.1880108991825615,
+ "grad_norm": 6.7723236083984375,
+ "learning_rate": 1.999259860553717e-05,
+ "loss": 1.446,
+ "step": 1537
+ },
+ {
+ "epoch": 4.190735694822888,
+ "grad_norm": 10.094219207763672,
+ "learning_rate": 1.999256461946419e-05,
+ "loss": 1.4636,
+ "step": 1538
+ },
+ {
+ "epoch": 4.193460490463215,
+ "grad_norm": 10.928735733032227,
+ "learning_rate": 1.999253055556944e-05,
+ "loss": 1.6467,
+ "step": 1539
+ },
+ {
+ "epoch": 4.1961852861035425,
+ "grad_norm": 9.1437406539917,
+ "learning_rate": 1.999249641385318e-05,
+ "loss": 1.5391,
+ "step": 1540
+ },
+ {
+ "epoch": 4.198910081743869,
+ "grad_norm": 8.220349311828613,
+ "learning_rate": 1.9992462194315685e-05,
+ "loss": 1.5342,
+ "step": 1541
+ },
+ {
+ "epoch": 4.201634877384196,
+ "grad_norm": 10.445140838623047,
+ "learning_rate": 1.9992427896957214e-05,
+ "loss": 1.4663,
+ "step": 1542
+ },
+ {
+ "epoch": 4.204359673024523,
+ "grad_norm": 7.789694786071777,
+ "learning_rate": 1.9992393521778038e-05,
+ "loss": 1.5208,
+ "step": 1543
+ },
+ {
+ "epoch": 4.20708446866485,
+ "grad_norm": 8.026616096496582,
+ "learning_rate": 1.999235906877842e-05,
+ "loss": 1.5374,
+ "step": 1544
+ },
+ {
+ "epoch": 4.209809264305177,
+ "grad_norm": 25.26694679260254,
+ "learning_rate": 1.9992324537958628e-05,
+ "loss": 1.5847,
+ "step": 1545
+ },
+ {
+ "epoch": 4.212534059945504,
+ "grad_norm": 10.576409339904785,
+ "learning_rate": 1.9992289929318944e-05,
+ "loss": 1.5679,
+ "step": 1546
+ },
+ {
+ "epoch": 4.215258855585831,
+ "grad_norm": 9.362836837768555,
+ "learning_rate": 1.9992255242859618e-05,
+ "loss": 1.4597,
+ "step": 1547
+ },
+ {
+ "epoch": 4.217983651226158,
+ "grad_norm": 10.311944007873535,
+ "learning_rate": 1.9992220478580936e-05,
+ "loss": 1.7074,
+ "step": 1548
+ },
+ {
+ "epoch": 4.220708446866485,
+ "grad_norm": 8.956902503967285,
+ "learning_rate": 1.999218563648316e-05,
+ "loss": 1.5408,
+ "step": 1549
+ },
+ {
+ "epoch": 4.223433242506812,
+ "grad_norm": 8.682573318481445,
+ "learning_rate": 1.9992150716566564e-05,
+ "loss": 1.7444,
+ "step": 1550
+ },
+ {
+ "epoch": 4.226158038147139,
+ "grad_norm": 8.537198066711426,
+ "learning_rate": 1.9992115718831416e-05,
+ "loss": 1.4636,
+ "step": 1551
+ },
+ {
+ "epoch": 4.228882833787466,
+ "grad_norm": 11.102084159851074,
+ "learning_rate": 1.9992080643278e-05,
+ "loss": 1.5237,
+ "step": 1552
+ },
+ {
+ "epoch": 4.231607629427793,
+ "grad_norm": 10.040871620178223,
+ "learning_rate": 1.9992045489906578e-05,
+ "loss": 1.667,
+ "step": 1553
+ },
+ {
+ "epoch": 4.23433242506812,
+ "grad_norm": 6.165706634521484,
+ "learning_rate": 1.9992010258717423e-05,
+ "loss": 1.4426,
+ "step": 1554
+ },
+ {
+ "epoch": 4.237057220708447,
+ "grad_norm": 8.590490341186523,
+ "learning_rate": 1.9991974949710823e-05,
+ "loss": 1.8145,
+ "step": 1555
+ },
+ {
+ "epoch": 4.239782016348774,
+ "grad_norm": 8.802373886108398,
+ "learning_rate": 1.9991939562887035e-05,
+ "loss": 1.865,
+ "step": 1556
+ },
+ {
+ "epoch": 4.2425068119891005,
+ "grad_norm": 8.291229248046875,
+ "learning_rate": 1.9991904098246347e-05,
+ "loss": 1.5222,
+ "step": 1557
+ },
+ {
+ "epoch": 4.245231607629428,
+ "grad_norm": 8.882878303527832,
+ "learning_rate": 1.999186855578903e-05,
+ "loss": 1.5586,
+ "step": 1558
+ },
+ {
+ "epoch": 4.247956403269755,
+ "grad_norm": 7.198384761810303,
+ "learning_rate": 1.9991832935515366e-05,
+ "loss": 1.4697,
+ "step": 1559
+ },
+ {
+ "epoch": 4.2506811989100814,
+ "grad_norm": 10.37208366394043,
+ "learning_rate": 1.9991797237425625e-05,
+ "loss": 1.6831,
+ "step": 1560
+ },
+ {
+ "epoch": 4.253405994550409,
+ "grad_norm": 6.954916477203369,
+ "learning_rate": 1.9991761461520088e-05,
+ "loss": 1.5461,
+ "step": 1561
+ },
+ {
+ "epoch": 4.256130790190736,
+ "grad_norm": 10.92075252532959,
+ "learning_rate": 1.9991725607799037e-05,
+ "loss": 1.6638,
+ "step": 1562
+ },
+ {
+ "epoch": 4.258855585831062,
+ "grad_norm": 15.699225425720215,
+ "learning_rate": 1.999168967626275e-05,
+ "loss": 1.6201,
+ "step": 1563
+ },
+ {
+ "epoch": 4.26158038147139,
+ "grad_norm": 10.398561477661133,
+ "learning_rate": 1.99916536669115e-05,
+ "loss": 1.7385,
+ "step": 1564
+ },
+ {
+ "epoch": 4.264305177111717,
+ "grad_norm": 8.457077980041504,
+ "learning_rate": 1.999161757974558e-05,
+ "loss": 1.7009,
+ "step": 1565
+ },
+ {
+ "epoch": 4.267029972752043,
+ "grad_norm": 8.201227188110352,
+ "learning_rate": 1.9991581414765256e-05,
+ "loss": 1.5369,
+ "step": 1566
+ },
+ {
+ "epoch": 4.269754768392371,
+ "grad_norm": 8.4393310546875,
+ "learning_rate": 1.9991545171970822e-05,
+ "loss": 1.5798,
+ "step": 1567
+ },
+ {
+ "epoch": 4.272479564032698,
+ "grad_norm": 8.569188117980957,
+ "learning_rate": 1.9991508851362555e-05,
+ "loss": 1.5259,
+ "step": 1568
+ },
+ {
+ "epoch": 4.275204359673024,
+ "grad_norm": 8.788260459899902,
+ "learning_rate": 1.9991472452940738e-05,
+ "loss": 1.6353,
+ "step": 1569
+ },
+ {
+ "epoch": 4.277929155313352,
+ "grad_norm": 13.729315757751465,
+ "learning_rate": 1.9991435976705657e-05,
+ "loss": 1.6667,
+ "step": 1570
+ },
+ {
+ "epoch": 4.2806539509536785,
+ "grad_norm": 9.042226791381836,
+ "learning_rate": 1.9991399422657593e-05,
+ "loss": 1.7686,
+ "step": 1571
+ },
+ {
+ "epoch": 4.283378746594005,
+ "grad_norm": 44.30101013183594,
+ "learning_rate": 1.999136279079683e-05,
+ "loss": 1.7812,
+ "step": 1572
+ },
+ {
+ "epoch": 4.286103542234333,
+ "grad_norm": 8.55893325805664,
+ "learning_rate": 1.999132608112366e-05,
+ "loss": 1.5825,
+ "step": 1573
+ },
+ {
+ "epoch": 4.2888283378746594,
+ "grad_norm": 7.176781177520752,
+ "learning_rate": 1.999128929363836e-05,
+ "loss": 1.585,
+ "step": 1574
+ },
+ {
+ "epoch": 4.291553133514986,
+ "grad_norm": 11.204346656799316,
+ "learning_rate": 1.9991252428341222e-05,
+ "loss": 1.4973,
+ "step": 1575
+ },
+ {
+ "epoch": 4.294277929155314,
+ "grad_norm": 8.480810165405273,
+ "learning_rate": 1.999121548523253e-05,
+ "loss": 1.5833,
+ "step": 1576
+ },
+ {
+ "epoch": 4.29700272479564,
+ "grad_norm": 8.083670616149902,
+ "learning_rate": 1.9991178464312575e-05,
+ "loss": 1.6016,
+ "step": 1577
+ },
+ {
+ "epoch": 4.299727520435967,
+ "grad_norm": 10.553550720214844,
+ "learning_rate": 1.9991141365581644e-05,
+ "loss": 1.5754,
+ "step": 1578
+ },
+ {
+ "epoch": 4.302452316076295,
+ "grad_norm": 12.233283996582031,
+ "learning_rate": 1.9991104189040024e-05,
+ "loss": 1.5112,
+ "step": 1579
+ },
+ {
+ "epoch": 4.305177111716621,
+ "grad_norm": 11.310342788696289,
+ "learning_rate": 1.999106693468801e-05,
+ "loss": 1.5933,
+ "step": 1580
+ },
+ {
+ "epoch": 4.307901907356948,
+ "grad_norm": 9.73255729675293,
+ "learning_rate": 1.9991029602525884e-05,
+ "loss": 1.7429,
+ "step": 1581
+ },
+ {
+ "epoch": 4.310626702997276,
+ "grad_norm": 10.82450008392334,
+ "learning_rate": 1.999099219255394e-05,
+ "loss": 1.9783,
+ "step": 1582
+ },
+ {
+ "epoch": 4.313351498637602,
+ "grad_norm": 11.155922889709473,
+ "learning_rate": 1.9990954704772473e-05,
+ "loss": 1.6384,
+ "step": 1583
+ },
+ {
+ "epoch": 4.316076294277929,
+ "grad_norm": 12.833930969238281,
+ "learning_rate": 1.9990917139181772e-05,
+ "loss": 1.8367,
+ "step": 1584
+ },
+ {
+ "epoch": 4.3188010899182565,
+ "grad_norm": 12.300129890441895,
+ "learning_rate": 1.999087949578213e-05,
+ "loss": 1.4626,
+ "step": 1585
+ },
+ {
+ "epoch": 4.321525885558583,
+ "grad_norm": 9.975184440612793,
+ "learning_rate": 1.9990841774573843e-05,
+ "loss": 1.7156,
+ "step": 1586
+ },
+ {
+ "epoch": 4.32425068119891,
+ "grad_norm": 11.834527015686035,
+ "learning_rate": 1.9990803975557197e-05,
+ "loss": 1.8274,
+ "step": 1587
+ },
+ {
+ "epoch": 4.3269754768392374,
+ "grad_norm": 8.867939949035645,
+ "learning_rate": 1.9990766098732493e-05,
+ "loss": 1.5525,
+ "step": 1588
+ },
+ {
+ "epoch": 4.329700272479564,
+ "grad_norm": 8.73129940032959,
+ "learning_rate": 1.9990728144100025e-05,
+ "loss": 1.6191,
+ "step": 1589
+ },
+ {
+ "epoch": 4.332425068119891,
+ "grad_norm": 20.48993682861328,
+ "learning_rate": 1.999069011166009e-05,
+ "loss": 1.7534,
+ "step": 1590
+ },
+ {
+ "epoch": 4.335149863760218,
+ "grad_norm": 9.350762367248535,
+ "learning_rate": 1.9990652001412973e-05,
+ "loss": 1.8711,
+ "step": 1591
+ },
+ {
+ "epoch": 4.337874659400545,
+ "grad_norm": 9.669034957885742,
+ "learning_rate": 1.9990613813358987e-05,
+ "loss": 1.6609,
+ "step": 1592
+ },
+ {
+ "epoch": 4.340599455040872,
+ "grad_norm": 9.409421920776367,
+ "learning_rate": 1.999057554749842e-05,
+ "loss": 1.5535,
+ "step": 1593
+ },
+ {
+ "epoch": 4.343324250681199,
+ "grad_norm": 9.087482452392578,
+ "learning_rate": 1.9990537203831573e-05,
+ "loss": 1.7368,
+ "step": 1594
+ },
+ {
+ "epoch": 4.346049046321526,
+ "grad_norm": 10.740350723266602,
+ "learning_rate": 1.9990498782358745e-05,
+ "loss": 1.6572,
+ "step": 1595
+ },
+ {
+ "epoch": 4.348773841961853,
+ "grad_norm": 8.663341522216797,
+ "learning_rate": 1.9990460283080232e-05,
+ "loss": 1.7166,
+ "step": 1596
+ },
+ {
+ "epoch": 4.35149863760218,
+ "grad_norm": 7.9335808753967285,
+ "learning_rate": 1.9990421705996336e-05,
+ "loss": 1.5979,
+ "step": 1597
+ },
+ {
+ "epoch": 4.354223433242507,
+ "grad_norm": 11.205650329589844,
+ "learning_rate": 1.999038305110736e-05,
+ "loss": 2.0193,
+ "step": 1598
+ },
+ {
+ "epoch": 4.356948228882834,
+ "grad_norm": 12.5615873336792,
+ "learning_rate": 1.9990344318413598e-05,
+ "loss": 1.4092,
+ "step": 1599
+ },
+ {
+ "epoch": 4.359673024523161,
+ "grad_norm": 8.803854942321777,
+ "learning_rate": 1.999030550791536e-05,
+ "loss": 1.5735,
+ "step": 1600
+ },
+ {
+ "epoch": 4.362397820163488,
+ "grad_norm": 10.278597831726074,
+ "learning_rate": 1.999026661961294e-05,
+ "loss": 1.5647,
+ "step": 1601
+ },
+ {
+ "epoch": 4.3651226158038146,
+ "grad_norm": 8.949265480041504,
+ "learning_rate": 1.999022765350665e-05,
+ "loss": 1.5891,
+ "step": 1602
+ },
+ {
+ "epoch": 4.367847411444142,
+ "grad_norm": 9.420422554016113,
+ "learning_rate": 1.9990188609596786e-05,
+ "loss": 1.4153,
+ "step": 1603
+ },
+ {
+ "epoch": 4.370572207084469,
+ "grad_norm": 8.191904067993164,
+ "learning_rate": 1.9990149487883653e-05,
+ "loss": 1.8687,
+ "step": 1604
+ },
+ {
+ "epoch": 4.3732970027247955,
+ "grad_norm": 8.539397239685059,
+ "learning_rate": 1.999011028836756e-05,
+ "loss": 1.5171,
+ "step": 1605
+ },
+ {
+ "epoch": 4.376021798365123,
+ "grad_norm": 7.269594192504883,
+ "learning_rate": 1.9990071011048807e-05,
+ "loss": 1.6658,
+ "step": 1606
+ },
+ {
+ "epoch": 4.37874659400545,
+ "grad_norm": 9.769559860229492,
+ "learning_rate": 1.999003165592771e-05,
+ "loss": 1.7302,
+ "step": 1607
+ },
+ {
+ "epoch": 4.381471389645776,
+ "grad_norm": 7.7932844161987305,
+ "learning_rate": 1.998999222300456e-05,
+ "loss": 1.4153,
+ "step": 1608
+ },
+ {
+ "epoch": 4.384196185286104,
+ "grad_norm": 6.8717827796936035,
+ "learning_rate": 1.9989952712279674e-05,
+ "loss": 1.6401,
+ "step": 1609
+ },
+ {
+ "epoch": 4.386920980926431,
+ "grad_norm": 9.949843406677246,
+ "learning_rate": 1.998991312375336e-05,
+ "loss": 1.7163,
+ "step": 1610
+ },
+ {
+ "epoch": 4.389645776566757,
+ "grad_norm": 7.0845794677734375,
+ "learning_rate": 1.9989873457425922e-05,
+ "loss": 1.551,
+ "step": 1611
+ },
+ {
+ "epoch": 4.392370572207085,
+ "grad_norm": 10.45530891418457,
+ "learning_rate": 1.998983371329767e-05,
+ "loss": 1.7991,
+ "step": 1612
+ },
+ {
+ "epoch": 4.395095367847412,
+ "grad_norm": 6.3778815269470215,
+ "learning_rate": 1.9989793891368915e-05,
+ "loss": 1.4226,
+ "step": 1613
+ },
+ {
+ "epoch": 4.397820163487738,
+ "grad_norm": 8.620266914367676,
+ "learning_rate": 1.9989753991639968e-05,
+ "loss": 1.5359,
+ "step": 1614
+ },
+ {
+ "epoch": 4.400544959128065,
+ "grad_norm": 11.054869651794434,
+ "learning_rate": 1.998971401411114e-05,
+ "loss": 1.5781,
+ "step": 1615
+ },
+ {
+ "epoch": 4.4032697547683926,
+ "grad_norm": 7.2482171058654785,
+ "learning_rate": 1.9989673958782735e-05,
+ "loss": 1.5864,
+ "step": 1616
+ },
+ {
+ "epoch": 4.405994550408719,
+ "grad_norm": 9.136199951171875,
+ "learning_rate": 1.998963382565508e-05,
+ "loss": 1.5535,
+ "step": 1617
+ },
+ {
+ "epoch": 4.408719346049046,
+ "grad_norm": 10.343496322631836,
+ "learning_rate": 1.998959361472847e-05,
+ "loss": 1.5176,
+ "step": 1618
+ },
+ {
+ "epoch": 4.4114441416893735,
+ "grad_norm": 14.777742385864258,
+ "learning_rate": 1.9989553326003226e-05,
+ "loss": 1.7981,
+ "step": 1619
+ },
+ {
+ "epoch": 4.4141689373297,
+ "grad_norm": 9.127223014831543,
+ "learning_rate": 1.9989512959479666e-05,
+ "loss": 1.5691,
+ "step": 1620
+ },
+ {
+ "epoch": 4.416893732970027,
+ "grad_norm": 10.75251579284668,
+ "learning_rate": 1.9989472515158102e-05,
+ "loss": 1.9231,
+ "step": 1621
+ },
+ {
+ "epoch": 4.419618528610354,
+ "grad_norm": 12.898049354553223,
+ "learning_rate": 1.9989431993038842e-05,
+ "loss": 1.7014,
+ "step": 1622
+ },
+ {
+ "epoch": 4.422343324250681,
+ "grad_norm": 7.789234638214111,
+ "learning_rate": 1.998939139312221e-05,
+ "loss": 1.5867,
+ "step": 1623
+ },
+ {
+ "epoch": 4.425068119891008,
+ "grad_norm": 10.412339210510254,
+ "learning_rate": 1.9989350715408525e-05,
+ "loss": 1.5508,
+ "step": 1624
+ },
+ {
+ "epoch": 4.427792915531335,
+ "grad_norm": 8.015629768371582,
+ "learning_rate": 1.998930995989809e-05,
+ "loss": 1.5833,
+ "step": 1625
+ },
+ {
+ "epoch": 4.430517711171662,
+ "grad_norm": 6.748058795928955,
+ "learning_rate": 1.9989269126591236e-05,
+ "loss": 1.8262,
+ "step": 1626
+ },
+ {
+ "epoch": 4.433242506811989,
+ "grad_norm": 8.060413360595703,
+ "learning_rate": 1.9989228215488273e-05,
+ "loss": 1.3208,
+ "step": 1627
+ },
+ {
+ "epoch": 4.435967302452316,
+ "grad_norm": 13.267457008361816,
+ "learning_rate": 1.9989187226589522e-05,
+ "loss": 1.717,
+ "step": 1628
+ },
+ {
+ "epoch": 4.438692098092643,
+ "grad_norm": 7.762689590454102,
+ "learning_rate": 1.9989146159895305e-05,
+ "loss": 1.6211,
+ "step": 1629
+ },
+ {
+ "epoch": 4.44141689373297,
+ "grad_norm": 11.200394630432129,
+ "learning_rate": 1.9989105015405935e-05,
+ "loss": 1.6267,
+ "step": 1630
+ },
+ {
+ "epoch": 4.444141689373297,
+ "grad_norm": 8.566879272460938,
+ "learning_rate": 1.9989063793121736e-05,
+ "loss": 1.8394,
+ "step": 1631
+ },
+ {
+ "epoch": 4.446866485013624,
+ "grad_norm": 9.473567962646484,
+ "learning_rate": 1.9989022493043033e-05,
+ "loss": 1.5176,
+ "step": 1632
+ },
+ {
+ "epoch": 4.449591280653951,
+ "grad_norm": 10.344234466552734,
+ "learning_rate": 1.9988981115170144e-05,
+ "loss": 1.6987,
+ "step": 1633
+ },
+ {
+ "epoch": 4.452316076294278,
+ "grad_norm": 8.260364532470703,
+ "learning_rate": 1.998893965950339e-05,
+ "loss": 1.8455,
+ "step": 1634
+ },
+ {
+ "epoch": 4.455040871934605,
+ "grad_norm": 9.973236083984375,
+ "learning_rate": 1.99888981260431e-05,
+ "loss": 1.4463,
+ "step": 1635
+ },
+ {
+ "epoch": 4.4577656675749315,
+ "grad_norm": 7.468135833740234,
+ "learning_rate": 1.998885651478959e-05,
+ "loss": 1.5984,
+ "step": 1636
+ },
+ {
+ "epoch": 4.460490463215259,
+ "grad_norm": 7.310967922210693,
+ "learning_rate": 1.9988814825743183e-05,
+ "loss": 1.5193,
+ "step": 1637
+ },
+ {
+ "epoch": 4.463215258855586,
+ "grad_norm": 7.59950065612793,
+ "learning_rate": 1.9988773058904213e-05,
+ "loss": 1.5027,
+ "step": 1638
+ },
+ {
+ "epoch": 4.4659400544959125,
+ "grad_norm": 10.033221244812012,
+ "learning_rate": 1.9988731214272995e-05,
+ "loss": 1.8521,
+ "step": 1639
+ },
+ {
+ "epoch": 4.46866485013624,
+ "grad_norm": 7.905033111572266,
+ "learning_rate": 1.9988689291849865e-05,
+ "loss": 1.6111,
+ "step": 1640
+ },
+ {
+ "epoch": 4.471389645776567,
+ "grad_norm": 9.09703254699707,
+ "learning_rate": 1.9988647291635143e-05,
+ "loss": 1.7876,
+ "step": 1641
+ },
+ {
+ "epoch": 4.474114441416893,
+ "grad_norm": 9.600266456604004,
+ "learning_rate": 1.9988605213629157e-05,
+ "loss": 1.5457,
+ "step": 1642
+ },
+ {
+ "epoch": 4.476839237057221,
+ "grad_norm": 9.986776351928711,
+ "learning_rate": 1.9988563057832234e-05,
+ "loss": 1.605,
+ "step": 1643
+ },
+ {
+ "epoch": 4.479564032697548,
+ "grad_norm": 8.521038055419922,
+ "learning_rate": 1.9988520824244703e-05,
+ "loss": 1.5676,
+ "step": 1644
+ },
+ {
+ "epoch": 4.482288828337874,
+ "grad_norm": 11.27340030670166,
+ "learning_rate": 1.9988478512866893e-05,
+ "loss": 1.5403,
+ "step": 1645
+ },
+ {
+ "epoch": 4.485013623978202,
+ "grad_norm": 9.78267765045166,
+ "learning_rate": 1.9988436123699135e-05,
+ "loss": 1.3704,
+ "step": 1646
+ },
+ {
+ "epoch": 4.487738419618529,
+ "grad_norm": 7.427676200866699,
+ "learning_rate": 1.998839365674176e-05,
+ "loss": 1.7512,
+ "step": 1647
+ },
+ {
+ "epoch": 4.490463215258855,
+ "grad_norm": 8.687450408935547,
+ "learning_rate": 1.9988351111995095e-05,
+ "loss": 1.4592,
+ "step": 1648
+ },
+ {
+ "epoch": 4.493188010899183,
+ "grad_norm": 10.796483993530273,
+ "learning_rate": 1.998830848945947e-05,
+ "loss": 1.8359,
+ "step": 1649
+ },
+ {
+ "epoch": 4.4959128065395095,
+ "grad_norm": 8.370170593261719,
+ "learning_rate": 1.9988265789135224e-05,
+ "loss": 1.6335,
+ "step": 1650
+ },
+ {
+ "epoch": 4.498637602179836,
+ "grad_norm": 7.719264507293701,
+ "learning_rate": 1.9988223011022686e-05,
+ "loss": 1.5684,
+ "step": 1651
+ },
+ {
+ "epoch": 4.501362397820164,
+ "grad_norm": 10.483525276184082,
+ "learning_rate": 1.9988180155122187e-05,
+ "loss": 1.6345,
+ "step": 1652
+ },
+ {
+ "epoch": 4.5040871934604905,
+ "grad_norm": 8.12997055053711,
+ "learning_rate": 1.998813722143406e-05,
+ "loss": 1.7349,
+ "step": 1653
+ },
+ {
+ "epoch": 4.506811989100817,
+ "grad_norm": 8.512079238891602,
+ "learning_rate": 1.9988094209958644e-05,
+ "loss": 1.6807,
+ "step": 1654
+ },
+ {
+ "epoch": 4.509536784741145,
+ "grad_norm": 8.54195785522461,
+ "learning_rate": 1.998805112069627e-05,
+ "loss": 1.4023,
+ "step": 1655
+ },
+ {
+ "epoch": 4.512261580381471,
+ "grad_norm": 8.169804573059082,
+ "learning_rate": 1.9988007953647278e-05,
+ "loss": 1.3809,
+ "step": 1656
+ },
+ {
+ "epoch": 4.514986376021798,
+ "grad_norm": 8.23045825958252,
+ "learning_rate": 1.9987964708812e-05,
+ "loss": 1.6584,
+ "step": 1657
+ },
+ {
+ "epoch": 4.517711171662126,
+ "grad_norm": 11.504939079284668,
+ "learning_rate": 1.9987921386190777e-05,
+ "loss": 1.5562,
+ "step": 1658
+ },
+ {
+ "epoch": 4.520435967302452,
+ "grad_norm": 17.11711311340332,
+ "learning_rate": 1.998787798578394e-05,
+ "loss": 1.5947,
+ "step": 1659
+ },
+ {
+ "epoch": 4.523160762942779,
+ "grad_norm": 8.016472816467285,
+ "learning_rate": 1.9987834507591834e-05,
+ "loss": 1.6846,
+ "step": 1660
+ },
+ {
+ "epoch": 4.525885558583107,
+ "grad_norm": 7.977776050567627,
+ "learning_rate": 1.9987790951614792e-05,
+ "loss": 1.7544,
+ "step": 1661
+ },
+ {
+ "epoch": 4.528610354223433,
+ "grad_norm": 7.286856651306152,
+ "learning_rate": 1.9987747317853156e-05,
+ "loss": 1.5601,
+ "step": 1662
+ },
+ {
+ "epoch": 4.53133514986376,
+ "grad_norm": 8.33243465423584,
+ "learning_rate": 1.9987703606307268e-05,
+ "loss": 1.5698,
+ "step": 1663
+ },
+ {
+ "epoch": 4.5340599455040875,
+ "grad_norm": 10.420464515686035,
+ "learning_rate": 1.998765981697746e-05,
+ "loss": 1.7095,
+ "step": 1664
+ },
+ {
+ "epoch": 4.536784741144414,
+ "grad_norm": 9.191597938537598,
+ "learning_rate": 1.9987615949864085e-05,
+ "loss": 1.8093,
+ "step": 1665
+ },
+ {
+ "epoch": 4.539509536784741,
+ "grad_norm": 7.787792682647705,
+ "learning_rate": 1.9987572004967474e-05,
+ "loss": 1.6724,
+ "step": 1666
+ },
+ {
+ "epoch": 4.5422343324250685,
+ "grad_norm": 12.707786560058594,
+ "learning_rate": 1.9987527982287977e-05,
+ "loss": 1.5574,
+ "step": 1667
+ },
+ {
+ "epoch": 4.544959128065395,
+ "grad_norm": 6.869774341583252,
+ "learning_rate": 1.9987483881825928e-05,
+ "loss": 1.6035,
+ "step": 1668
+ },
+ {
+ "epoch": 4.547683923705722,
+ "grad_norm": 7.024541854858398,
+ "learning_rate": 1.998743970358168e-05,
+ "loss": 1.5132,
+ "step": 1669
+ },
+ {
+ "epoch": 4.550408719346049,
+ "grad_norm": 7.383622646331787,
+ "learning_rate": 1.9987395447555574e-05,
+ "loss": 1.3018,
+ "step": 1670
+ },
+ {
+ "epoch": 4.553133514986376,
+ "grad_norm": 9.555183410644531,
+ "learning_rate": 1.9987351113747953e-05,
+ "loss": 1.488,
+ "step": 1671
+ },
+ {
+ "epoch": 4.555858310626703,
+ "grad_norm": 7.561858654022217,
+ "learning_rate": 1.998730670215916e-05,
+ "loss": 1.4753,
+ "step": 1672
+ },
+ {
+ "epoch": 4.55858310626703,
+ "grad_norm": 10.767231941223145,
+ "learning_rate": 1.9987262212789547e-05,
+ "loss": 1.4021,
+ "step": 1673
+ },
+ {
+ "epoch": 4.561307901907357,
+ "grad_norm": 8.209709167480469,
+ "learning_rate": 1.998721764563946e-05,
+ "loss": 1.4421,
+ "step": 1674
+ },
+ {
+ "epoch": 4.564032697547684,
+ "grad_norm": 8.655367851257324,
+ "learning_rate": 1.9987173000709238e-05,
+ "loss": 1.7014,
+ "step": 1675
+ },
+ {
+ "epoch": 4.566757493188011,
+ "grad_norm": 6.918274402618408,
+ "learning_rate": 1.9987128277999236e-05,
+ "loss": 1.4946,
+ "step": 1676
+ },
+ {
+ "epoch": 4.569482288828338,
+ "grad_norm": 8.026473045349121,
+ "learning_rate": 1.99870834775098e-05,
+ "loss": 1.5791,
+ "step": 1677
+ },
+ {
+ "epoch": 4.572207084468665,
+ "grad_norm": 12.75072193145752,
+ "learning_rate": 1.998703859924128e-05,
+ "loss": 1.5779,
+ "step": 1678
+ },
+ {
+ "epoch": 4.574931880108992,
+ "grad_norm": 7.941193103790283,
+ "learning_rate": 1.9986993643194022e-05,
+ "loss": 1.5745,
+ "step": 1679
+ },
+ {
+ "epoch": 4.577656675749319,
+ "grad_norm": 8.674474716186523,
+ "learning_rate": 1.9986948609368385e-05,
+ "loss": 1.7532,
+ "step": 1680
+ },
+ {
+ "epoch": 4.580381471389646,
+ "grad_norm": 11.028712272644043,
+ "learning_rate": 1.998690349776471e-05,
+ "loss": 1.5718,
+ "step": 1681
+ },
+ {
+ "epoch": 4.583106267029972,
+ "grad_norm": 8.243412971496582,
+ "learning_rate": 1.998685830838335e-05,
+ "loss": 1.4973,
+ "step": 1682
+ },
+ {
+ "epoch": 4.5858310626703,
+ "grad_norm": 12.946802139282227,
+ "learning_rate": 1.9986813041224662e-05,
+ "loss": 1.5483,
+ "step": 1683
+ },
+ {
+ "epoch": 4.5885558583106265,
+ "grad_norm": 9.966217041015625,
+ "learning_rate": 1.9986767696288997e-05,
+ "loss": 1.7817,
+ "step": 1684
+ },
+ {
+ "epoch": 4.591280653950953,
+ "grad_norm": 9.317113876342773,
+ "learning_rate": 1.9986722273576703e-05,
+ "loss": 1.6619,
+ "step": 1685
+ },
+ {
+ "epoch": 4.594005449591281,
+ "grad_norm": 7.250195026397705,
+ "learning_rate": 1.9986676773088137e-05,
+ "loss": 1.426,
+ "step": 1686
+ },
+ {
+ "epoch": 4.5967302452316074,
+ "grad_norm": 9.530462265014648,
+ "learning_rate": 1.9986631194823657e-05,
+ "loss": 1.5642,
+ "step": 1687
+ },
+ {
+ "epoch": 4.599455040871934,
+ "grad_norm": 7.1351423263549805,
+ "learning_rate": 1.998658553878361e-05,
+ "loss": 1.5024,
+ "step": 1688
+ },
+ {
+ "epoch": 4.602179836512262,
+ "grad_norm": 6.7440338134765625,
+ "learning_rate": 1.998653980496836e-05,
+ "loss": 1.4683,
+ "step": 1689
+ },
+ {
+ "epoch": 4.604904632152588,
+ "grad_norm": 8.621153831481934,
+ "learning_rate": 1.998649399337826e-05,
+ "loss": 1.8215,
+ "step": 1690
+ },
+ {
+ "epoch": 4.607629427792915,
+ "grad_norm": 7.739974498748779,
+ "learning_rate": 1.9986448104013664e-05,
+ "loss": 1.3901,
+ "step": 1691
+ },
+ {
+ "epoch": 4.610354223433243,
+ "grad_norm": 7.003955841064453,
+ "learning_rate": 1.998640213687493e-05,
+ "loss": 1.519,
+ "step": 1692
+ },
+ {
+ "epoch": 4.613079019073569,
+ "grad_norm": 7.233630180358887,
+ "learning_rate": 1.9986356091962423e-05,
+ "loss": 1.5044,
+ "step": 1693
+ },
+ {
+ "epoch": 4.615803814713896,
+ "grad_norm": 7.16238260269165,
+ "learning_rate": 1.998630996927649e-05,
+ "loss": 1.7249,
+ "step": 1694
+ },
+ {
+ "epoch": 4.618528610354224,
+ "grad_norm": 8.690790176391602,
+ "learning_rate": 1.9986263768817502e-05,
+ "loss": 1.6011,
+ "step": 1695
+ },
+ {
+ "epoch": 4.62125340599455,
+ "grad_norm": 7.450441360473633,
+ "learning_rate": 1.998621749058581e-05,
+ "loss": 1.4695,
+ "step": 1696
+ },
+ {
+ "epoch": 4.623978201634877,
+ "grad_norm": 6.028726100921631,
+ "learning_rate": 1.9986171134581778e-05,
+ "loss": 1.5476,
+ "step": 1697
+ },
+ {
+ "epoch": 4.6267029972752045,
+ "grad_norm": 7.644749641418457,
+ "learning_rate": 1.9986124700805764e-05,
+ "loss": 1.4636,
+ "step": 1698
+ },
+ {
+ "epoch": 4.629427792915531,
+ "grad_norm": 7.60107421875,
+ "learning_rate": 1.998607818925814e-05,
+ "loss": 1.5432,
+ "step": 1699
+ },
+ {
+ "epoch": 4.632152588555858,
+ "grad_norm": 7.10397481918335,
+ "learning_rate": 1.9986031599939253e-05,
+ "loss": 1.5535,
+ "step": 1700
+ },
+ {
+ "epoch": 4.6348773841961854,
+ "grad_norm": 8.438920974731445,
+ "learning_rate": 1.9985984932849475e-05,
+ "loss": 1.3706,
+ "step": 1701
+ },
+ {
+ "epoch": 4.637602179836512,
+ "grad_norm": 15.496280670166016,
+ "learning_rate": 1.998593818798917e-05,
+ "loss": 1.738,
+ "step": 1702
+ },
+ {
+ "epoch": 4.640326975476839,
+ "grad_norm": 6.5038886070251465,
+ "learning_rate": 1.9985891365358697e-05,
+ "loss": 1.3423,
+ "step": 1703
+ },
+ {
+ "epoch": 4.643051771117166,
+ "grad_norm": 7.746236324310303,
+ "learning_rate": 1.9985844464958423e-05,
+ "loss": 1.8184,
+ "step": 1704
+ },
+ {
+ "epoch": 4.645776566757493,
+ "grad_norm": 7.3002119064331055,
+ "learning_rate": 1.998579748678872e-05,
+ "loss": 1.9292,
+ "step": 1705
+ },
+ {
+ "epoch": 4.64850136239782,
+ "grad_norm": 7.324510097503662,
+ "learning_rate": 1.998575043084994e-05,
+ "loss": 1.7493,
+ "step": 1706
+ },
+ {
+ "epoch": 4.651226158038147,
+ "grad_norm": 7.569739818572998,
+ "learning_rate": 1.998570329714246e-05,
+ "loss": 1.5522,
+ "step": 1707
+ },
+ {
+ "epoch": 4.653950953678474,
+ "grad_norm": 7.636623859405518,
+ "learning_rate": 1.9985656085666646e-05,
+ "loss": 1.6182,
+ "step": 1708
+ },
+ {
+ "epoch": 4.656675749318801,
+ "grad_norm": 7.1200761795043945,
+ "learning_rate": 1.9985608796422863e-05,
+ "loss": 1.3469,
+ "step": 1709
+ },
+ {
+ "epoch": 4.659400544959128,
+ "grad_norm": 8.993254661560059,
+ "learning_rate": 1.9985561429411478e-05,
+ "loss": 1.3491,
+ "step": 1710
+ },
+ {
+ "epoch": 4.662125340599455,
+ "grad_norm": 8.854870796203613,
+ "learning_rate": 1.998551398463286e-05,
+ "loss": 1.5115,
+ "step": 1711
+ },
+ {
+ "epoch": 4.664850136239782,
+ "grad_norm": 6.510181427001953,
+ "learning_rate": 1.9985466462087383e-05,
+ "loss": 1.5044,
+ "step": 1712
+ },
+ {
+ "epoch": 4.667574931880109,
+ "grad_norm": 12.196049690246582,
+ "learning_rate": 1.9985418861775416e-05,
+ "loss": 1.5388,
+ "step": 1713
+ },
+ {
+ "epoch": 4.670299727520436,
+ "grad_norm": 7.211997985839844,
+ "learning_rate": 1.9985371183697327e-05,
+ "loss": 1.6028,
+ "step": 1714
+ },
+ {
+ "epoch": 4.6730245231607626,
+ "grad_norm": 6.917252063751221,
+ "learning_rate": 1.9985323427853492e-05,
+ "loss": 1.6985,
+ "step": 1715
+ },
+ {
+ "epoch": 4.67574931880109,
+ "grad_norm": 6.779542922973633,
+ "learning_rate": 1.9985275594244277e-05,
+ "loss": 1.4832,
+ "step": 1716
+ },
+ {
+ "epoch": 4.678474114441417,
+ "grad_norm": 6.808323383331299,
+ "learning_rate": 1.9985227682870054e-05,
+ "loss": 1.27,
+ "step": 1717
+ },
+ {
+ "epoch": 4.6811989100817435,
+ "grad_norm": 10.249691009521484,
+ "learning_rate": 1.9985179693731204e-05,
+ "loss": 1.7119,
+ "step": 1718
+ },
+ {
+ "epoch": 4.683923705722071,
+ "grad_norm": 7.141353607177734,
+ "learning_rate": 1.9985131626828092e-05,
+ "loss": 1.4146,
+ "step": 1719
+ },
+ {
+ "epoch": 4.686648501362398,
+ "grad_norm": 8.66392993927002,
+ "learning_rate": 1.9985083482161098e-05,
+ "loss": 1.4209,
+ "step": 1720
+ },
+ {
+ "epoch": 4.689373297002724,
+ "grad_norm": 6.368734359741211,
+ "learning_rate": 1.9985035259730593e-05,
+ "loss": 1.6777,
+ "step": 1721
+ },
+ {
+ "epoch": 4.692098092643052,
+ "grad_norm": 8.035052299499512,
+ "learning_rate": 1.998498695953696e-05,
+ "loss": 1.7607,
+ "step": 1722
+ },
+ {
+ "epoch": 4.694822888283379,
+ "grad_norm": 13.738799095153809,
+ "learning_rate": 1.9984938581580566e-05,
+ "loss": 1.5417,
+ "step": 1723
+ },
+ {
+ "epoch": 4.697547683923705,
+ "grad_norm": 8.732844352722168,
+ "learning_rate": 1.998489012586179e-05,
+ "loss": 1.7795,
+ "step": 1724
+ },
+ {
+ "epoch": 4.700272479564033,
+ "grad_norm": 7.641689300537109,
+ "learning_rate": 1.9984841592381014e-05,
+ "loss": 1.4915,
+ "step": 1725
+ },
+ {
+ "epoch": 4.70299727520436,
+ "grad_norm": 6.789289951324463,
+ "learning_rate": 1.9984792981138616e-05,
+ "loss": 1.5554,
+ "step": 1726
+ },
+ {
+ "epoch": 4.705722070844686,
+ "grad_norm": 6.860891342163086,
+ "learning_rate": 1.9984744292134964e-05,
+ "loss": 1.614,
+ "step": 1727
+ },
+ {
+ "epoch": 4.708446866485014,
+ "grad_norm": 6.264583587646484,
+ "learning_rate": 1.998469552537045e-05,
+ "loss": 1.5942,
+ "step": 1728
+ },
+ {
+ "epoch": 4.7111716621253406,
+ "grad_norm": 7.513585090637207,
+ "learning_rate": 1.9984646680845446e-05,
+ "loss": 1.4795,
+ "step": 1729
+ },
+ {
+ "epoch": 4.713896457765667,
+ "grad_norm": 7.607376575469971,
+ "learning_rate": 1.998459775856034e-05,
+ "loss": 1.938,
+ "step": 1730
+ },
+ {
+ "epoch": 4.716621253405995,
+ "grad_norm": 5.915128707885742,
+ "learning_rate": 1.9984548758515503e-05,
+ "loss": 1.3201,
+ "step": 1731
+ },
+ {
+ "epoch": 4.7193460490463215,
+ "grad_norm": 6.803768157958984,
+ "learning_rate": 1.9984499680711322e-05,
+ "loss": 1.5959,
+ "step": 1732
+ },
+ {
+ "epoch": 4.722070844686648,
+ "grad_norm": 9.065163612365723,
+ "learning_rate": 1.9984450525148184e-05,
+ "loss": 1.5081,
+ "step": 1733
+ },
+ {
+ "epoch": 4.724795640326976,
+ "grad_norm": 6.951340675354004,
+ "learning_rate": 1.998440129182646e-05,
+ "loss": 1.5046,
+ "step": 1734
+ },
+ {
+ "epoch": 4.727520435967302,
+ "grad_norm": 6.557684421539307,
+ "learning_rate": 1.9984351980746545e-05,
+ "loss": 1.6035,
+ "step": 1735
+ },
+ {
+ "epoch": 4.730245231607629,
+ "grad_norm": 7.115765571594238,
+ "learning_rate": 1.9984302591908812e-05,
+ "loss": 1.8291,
+ "step": 1736
+ },
+ {
+ "epoch": 4.732970027247957,
+ "grad_norm": 8.59748363494873,
+ "learning_rate": 1.998425312531366e-05,
+ "loss": 1.6152,
+ "step": 1737
+ },
+ {
+ "epoch": 4.735694822888283,
+ "grad_norm": 8.62453842163086,
+ "learning_rate": 1.998420358096146e-05,
+ "loss": 1.699,
+ "step": 1738
+ },
+ {
+ "epoch": 4.73841961852861,
+ "grad_norm": 15.304737091064453,
+ "learning_rate": 1.9984153958852605e-05,
+ "loss": 1.6331,
+ "step": 1739
+ },
+ {
+ "epoch": 4.741144414168938,
+ "grad_norm": 7.161703586578369,
+ "learning_rate": 1.998410425898748e-05,
+ "loss": 1.48,
+ "step": 1740
+ },
+ {
+ "epoch": 4.743869209809264,
+ "grad_norm": 8.097888946533203,
+ "learning_rate": 1.9984054481366474e-05,
+ "loss": 1.5593,
+ "step": 1741
+ },
+ {
+ "epoch": 4.746594005449591,
+ "grad_norm": 8.044931411743164,
+ "learning_rate": 1.9984004625989974e-05,
+ "loss": 1.5856,
+ "step": 1742
+ },
+ {
+ "epoch": 4.7493188010899186,
+ "grad_norm": 5.199501991271973,
+ "learning_rate": 1.9983954692858363e-05,
+ "loss": 1.3999,
+ "step": 1743
+ },
+ {
+ "epoch": 4.752043596730245,
+ "grad_norm": 8.534812927246094,
+ "learning_rate": 1.9983904681972034e-05,
+ "loss": 1.5923,
+ "step": 1744
+ },
+ {
+ "epoch": 4.754768392370572,
+ "grad_norm": 7.132694244384766,
+ "learning_rate": 1.998385459333138e-05,
+ "loss": 1.55,
+ "step": 1745
+ },
+ {
+ "epoch": 4.7574931880108995,
+ "grad_norm": 7.248411655426025,
+ "learning_rate": 1.9983804426936787e-05,
+ "loss": 1.8682,
+ "step": 1746
+ },
+ {
+ "epoch": 4.760217983651226,
+ "grad_norm": 6.998132705688477,
+ "learning_rate": 1.9983754182788647e-05,
+ "loss": 1.6304,
+ "step": 1747
+ },
+ {
+ "epoch": 4.762942779291553,
+ "grad_norm": 6.49261999130249,
+ "learning_rate": 1.9983703860887355e-05,
+ "loss": 1.4314,
+ "step": 1748
+ },
+ {
+ "epoch": 4.76566757493188,
+ "grad_norm": 6.228968620300293,
+ "learning_rate": 1.998365346123329e-05,
+ "loss": 1.478,
+ "step": 1749
+ },
+ {
+ "epoch": 4.768392370572207,
+ "grad_norm": 7.053474426269531,
+ "learning_rate": 1.998360298382686e-05,
+ "loss": 1.7373,
+ "step": 1750
+ },
+ {
+ "epoch": 4.771117166212534,
+ "grad_norm": 6.739225387573242,
+ "learning_rate": 1.9983552428668448e-05,
+ "loss": 1.7947,
+ "step": 1751
+ },
+ {
+ "epoch": 4.773841961852861,
+ "grad_norm": 7.3383469581604,
+ "learning_rate": 1.998350179575845e-05,
+ "loss": 1.624,
+ "step": 1752
+ },
+ {
+ "epoch": 4.776566757493188,
+ "grad_norm": 9.675017356872559,
+ "learning_rate": 1.9983451085097262e-05,
+ "loss": 1.6443,
+ "step": 1753
+ },
+ {
+ "epoch": 4.779291553133515,
+ "grad_norm": 8.301816940307617,
+ "learning_rate": 1.998340029668528e-05,
+ "loss": 1.4907,
+ "step": 1754
+ },
+ {
+ "epoch": 4.782016348773842,
+ "grad_norm": 7.318321704864502,
+ "learning_rate": 1.9983349430522896e-05,
+ "loss": 1.3391,
+ "step": 1755
+ },
+ {
+ "epoch": 4.784741144414169,
+ "grad_norm": 6.588681221008301,
+ "learning_rate": 1.9983298486610508e-05,
+ "loss": 1.4258,
+ "step": 1756
+ },
+ {
+ "epoch": 4.787465940054496,
+ "grad_norm": 8.24355411529541,
+ "learning_rate": 1.9983247464948515e-05,
+ "loss": 1.563,
+ "step": 1757
+ },
+ {
+ "epoch": 4.790190735694823,
+ "grad_norm": 9.268683433532715,
+ "learning_rate": 1.9983196365537313e-05,
+ "loss": 1.4722,
+ "step": 1758
+ },
+ {
+ "epoch": 4.79291553133515,
+ "grad_norm": 14.490931510925293,
+ "learning_rate": 1.9983145188377297e-05,
+ "loss": 1.6125,
+ "step": 1759
+ },
+ {
+ "epoch": 4.795640326975477,
+ "grad_norm": 6.875260829925537,
+ "learning_rate": 1.9983093933468864e-05,
+ "loss": 1.3254,
+ "step": 1760
+ },
+ {
+ "epoch": 4.798365122615804,
+ "grad_norm": 9.080835342407227,
+ "learning_rate": 1.9983042600812423e-05,
+ "loss": 1.5542,
+ "step": 1761
+ },
+ {
+ "epoch": 4.801089918256131,
+ "grad_norm": 8.066596984863281,
+ "learning_rate": 1.9982991190408362e-05,
+ "loss": 1.3853,
+ "step": 1762
+ },
+ {
+ "epoch": 4.8038147138964575,
+ "grad_norm": 7.388097286224365,
+ "learning_rate": 1.9982939702257088e-05,
+ "loss": 1.5002,
+ "step": 1763
+ },
+ {
+ "epoch": 4.806539509536785,
+ "grad_norm": 8.93244457244873,
+ "learning_rate": 1.9982888136359e-05,
+ "loss": 1.8406,
+ "step": 1764
+ },
+ {
+ "epoch": 4.809264305177112,
+ "grad_norm": 7.477293491363525,
+ "learning_rate": 1.9982836492714505e-05,
+ "loss": 1.2197,
+ "step": 1765
+ },
+ {
+ "epoch": 4.8119891008174385,
+ "grad_norm": 6.864194869995117,
+ "learning_rate": 1.9982784771324e-05,
+ "loss": 1.4424,
+ "step": 1766
+ },
+ {
+ "epoch": 4.814713896457766,
+ "grad_norm": 6.041901111602783,
+ "learning_rate": 1.9982732972187884e-05,
+ "loss": 1.4983,
+ "step": 1767
+ },
+ {
+ "epoch": 4.817438692098093,
+ "grad_norm": 7.30835485458374,
+ "learning_rate": 1.998268109530657e-05,
+ "loss": 1.4241,
+ "step": 1768
+ },
+ {
+ "epoch": 4.820163487738419,
+ "grad_norm": 11.82546615600586,
+ "learning_rate": 1.9982629140680453e-05,
+ "loss": 1.5884,
+ "step": 1769
+ },
+ {
+ "epoch": 4.822888283378747,
+ "grad_norm": 6.453680992126465,
+ "learning_rate": 1.9982577108309945e-05,
+ "loss": 1.4915,
+ "step": 1770
+ },
+ {
+ "epoch": 4.825613079019074,
+ "grad_norm": 8.563255310058594,
+ "learning_rate": 1.9982524998195446e-05,
+ "loss": 1.6785,
+ "step": 1771
+ },
+ {
+ "epoch": 4.8283378746594,
+ "grad_norm": 7.708741664886475,
+ "learning_rate": 1.9982472810337363e-05,
+ "loss": 1.5515,
+ "step": 1772
+ },
+ {
+ "epoch": 4.831062670299728,
+ "grad_norm": 8.020515441894531,
+ "learning_rate": 1.9982420544736106e-05,
+ "loss": 1.7856,
+ "step": 1773
+ },
+ {
+ "epoch": 4.833787465940055,
+ "grad_norm": 8.902491569519043,
+ "learning_rate": 1.9982368201392074e-05,
+ "loss": 1.4004,
+ "step": 1774
+ },
+ {
+ "epoch": 4.836512261580381,
+ "grad_norm": 7.848010540008545,
+ "learning_rate": 1.998231578030568e-05,
+ "loss": 1.5056,
+ "step": 1775
+ },
+ {
+ "epoch": 4.839237057220709,
+ "grad_norm": 7.3329854011535645,
+ "learning_rate": 1.9982263281477338e-05,
+ "loss": 1.5613,
+ "step": 1776
+ },
+ {
+ "epoch": 4.8419618528610355,
+ "grad_norm": 9.430790901184082,
+ "learning_rate": 1.9982210704907446e-05,
+ "loss": 1.6243,
+ "step": 1777
+ },
+ {
+ "epoch": 4.844686648501362,
+ "grad_norm": 7.791731357574463,
+ "learning_rate": 1.998215805059642e-05,
+ "loss": 1.4758,
+ "step": 1778
+ },
+ {
+ "epoch": 4.84741144414169,
+ "grad_norm": 7.968623161315918,
+ "learning_rate": 1.998210531854467e-05,
+ "loss": 1.5867,
+ "step": 1779
+ },
+ {
+ "epoch": 4.8501362397820165,
+ "grad_norm": 5.81412410736084,
+ "learning_rate": 1.9982052508752605e-05,
+ "loss": 1.5918,
+ "step": 1780
+ },
+ {
+ "epoch": 4.852861035422343,
+ "grad_norm": 6.71942138671875,
+ "learning_rate": 1.998199962122063e-05,
+ "loss": 1.3552,
+ "step": 1781
+ },
+ {
+ "epoch": 4.855585831062671,
+ "grad_norm": 6.662970542907715,
+ "learning_rate": 1.998194665594917e-05,
+ "loss": 1.5269,
+ "step": 1782
+ },
+ {
+ "epoch": 4.858310626702997,
+ "grad_norm": 7.6843061447143555,
+ "learning_rate": 1.9981893612938624e-05,
+ "loss": 1.6807,
+ "step": 1783
+ },
+ {
+ "epoch": 4.861035422343324,
+ "grad_norm": 20.497112274169922,
+ "learning_rate": 1.998184049218942e-05,
+ "loss": 1.5715,
+ "step": 1784
+ },
+ {
+ "epoch": 4.863760217983652,
+ "grad_norm": 6.352015018463135,
+ "learning_rate": 1.9981787293701956e-05,
+ "loss": 1.429,
+ "step": 1785
+ },
+ {
+ "epoch": 4.866485013623978,
+ "grad_norm": 6.527281761169434,
+ "learning_rate": 1.998173401747666e-05,
+ "loss": 1.5354,
+ "step": 1786
+ },
+ {
+ "epoch": 4.869209809264305,
+ "grad_norm": 8.808930397033691,
+ "learning_rate": 1.998168066351394e-05,
+ "loss": 1.6792,
+ "step": 1787
+ },
+ {
+ "epoch": 4.871934604904633,
+ "grad_norm": 7.4636945724487305,
+ "learning_rate": 1.9981627231814207e-05,
+ "loss": 1.8101,
+ "step": 1788
+ },
+ {
+ "epoch": 4.874659400544959,
+ "grad_norm": 7.174690246582031,
+ "learning_rate": 1.9981573722377888e-05,
+ "loss": 1.5906,
+ "step": 1789
+ },
+ {
+ "epoch": 4.877384196185286,
+ "grad_norm": 7.273940563201904,
+ "learning_rate": 1.998152013520539e-05,
+ "loss": 1.5576,
+ "step": 1790
+ },
+ {
+ "epoch": 4.8801089918256135,
+ "grad_norm": 7.737430572509766,
+ "learning_rate": 1.9981466470297136e-05,
+ "loss": 1.7065,
+ "step": 1791
+ },
+ {
+ "epoch": 4.88283378746594,
+ "grad_norm": 8.75093936920166,
+ "learning_rate": 1.9981412727653544e-05,
+ "loss": 1.3315,
+ "step": 1792
+ },
+ {
+ "epoch": 4.885558583106267,
+ "grad_norm": 9.180643081665039,
+ "learning_rate": 1.998135890727503e-05,
+ "loss": 1.6284,
+ "step": 1793
+ },
+ {
+ "epoch": 4.8882833787465945,
+ "grad_norm": 9.501355171203613,
+ "learning_rate": 1.998130500916201e-05,
+ "loss": 1.6099,
+ "step": 1794
+ },
+ {
+ "epoch": 4.891008174386921,
+ "grad_norm": 7.564662456512451,
+ "learning_rate": 1.9981251033314914e-05,
+ "loss": 1.3591,
+ "step": 1795
+ },
+ {
+ "epoch": 4.893732970027248,
+ "grad_norm": 7.397062301635742,
+ "learning_rate": 1.9981196979734157e-05,
+ "loss": 1.668,
+ "step": 1796
+ },
+ {
+ "epoch": 4.896457765667575,
+ "grad_norm": 8.441000938415527,
+ "learning_rate": 1.9981142848420152e-05,
+ "loss": 1.5234,
+ "step": 1797
+ },
+ {
+ "epoch": 4.899182561307902,
+ "grad_norm": 8.225923538208008,
+ "learning_rate": 1.9981088639373335e-05,
+ "loss": 1.7961,
+ "step": 1798
+ },
+ {
+ "epoch": 4.901907356948229,
+ "grad_norm": 5.830610275268555,
+ "learning_rate": 1.9981034352594117e-05,
+ "loss": 1.4094,
+ "step": 1799
+ },
+ {
+ "epoch": 4.904632152588556,
+ "grad_norm": 6.624537944793701,
+ "learning_rate": 1.9980979988082923e-05,
+ "loss": 1.6052,
+ "step": 1800
+ },
+ {
+ "epoch": 4.907356948228883,
+ "grad_norm": 6.884772777557373,
+ "learning_rate": 1.9980925545840183e-05,
+ "loss": 1.5708,
+ "step": 1801
+ },
+ {
+ "epoch": 4.91008174386921,
+ "grad_norm": 6.872514724731445,
+ "learning_rate": 1.9980871025866314e-05,
+ "loss": 1.5615,
+ "step": 1802
+ },
+ {
+ "epoch": 4.912806539509537,
+ "grad_norm": 7.608044624328613,
+ "learning_rate": 1.998081642816174e-05,
+ "loss": 1.4961,
+ "step": 1803
+ },
+ {
+ "epoch": 4.915531335149864,
+ "grad_norm": 5.399438381195068,
+ "learning_rate": 1.998076175272689e-05,
+ "loss": 1.4631,
+ "step": 1804
+ },
+ {
+ "epoch": 4.918256130790191,
+ "grad_norm": 7.608586311340332,
+ "learning_rate": 1.998070699956219e-05,
+ "loss": 1.5435,
+ "step": 1805
+ },
+ {
+ "epoch": 4.920980926430518,
+ "grad_norm": 7.474642753601074,
+ "learning_rate": 1.9980652168668063e-05,
+ "loss": 1.7625,
+ "step": 1806
+ },
+ {
+ "epoch": 4.923705722070845,
+ "grad_norm": 7.582441329956055,
+ "learning_rate": 1.9980597260044944e-05,
+ "loss": 1.5417,
+ "step": 1807
+ },
+ {
+ "epoch": 4.926430517711172,
+ "grad_norm": 8.808675765991211,
+ "learning_rate": 1.998054227369325e-05,
+ "loss": 1.7822,
+ "step": 1808
+ },
+ {
+ "epoch": 4.929155313351498,
+ "grad_norm": 6.414207458496094,
+ "learning_rate": 1.998048720961341e-05,
+ "loss": 1.5444,
+ "step": 1809
+ },
+ {
+ "epoch": 4.931880108991826,
+ "grad_norm": 8.321244239807129,
+ "learning_rate": 1.9980432067805865e-05,
+ "loss": 1.5503,
+ "step": 1810
+ },
+ {
+ "epoch": 4.9346049046321525,
+ "grad_norm": 6.391635417938232,
+ "learning_rate": 1.9980376848271032e-05,
+ "loss": 1.5913,
+ "step": 1811
+ },
+ {
+ "epoch": 4.937329700272479,
+ "grad_norm": 6.656155586242676,
+ "learning_rate": 1.998032155100935e-05,
+ "loss": 1.4604,
+ "step": 1812
+ },
+ {
+ "epoch": 4.940054495912807,
+ "grad_norm": 6.772690773010254,
+ "learning_rate": 1.998026617602124e-05,
+ "loss": 1.6338,
+ "step": 1813
+ },
+ {
+ "epoch": 4.9427792915531334,
+ "grad_norm": 5.684878349304199,
+ "learning_rate": 1.998021072330714e-05,
+ "loss": 1.4419,
+ "step": 1814
+ },
+ {
+ "epoch": 4.94550408719346,
+ "grad_norm": 6.281477451324463,
+ "learning_rate": 1.9980155192867477e-05,
+ "loss": 1.7461,
+ "step": 1815
+ },
+ {
+ "epoch": 4.948228882833788,
+ "grad_norm": 9.203144073486328,
+ "learning_rate": 1.998009958470269e-05,
+ "loss": 1.5579,
+ "step": 1816
+ },
+ {
+ "epoch": 4.950953678474114,
+ "grad_norm": 9.395030975341797,
+ "learning_rate": 1.9980043898813206e-05,
+ "loss": 1.6514,
+ "step": 1817
+ },
+ {
+ "epoch": 4.953678474114441,
+ "grad_norm": 7.176726818084717,
+ "learning_rate": 1.9979988135199464e-05,
+ "loss": 1.6274,
+ "step": 1818
+ },
+ {
+ "epoch": 4.956403269754769,
+ "grad_norm": 9.602917671203613,
+ "learning_rate": 1.9979932293861895e-05,
+ "loss": 1.5317,
+ "step": 1819
+ },
+ {
+ "epoch": 4.959128065395095,
+ "grad_norm": 7.056041717529297,
+ "learning_rate": 1.9979876374800935e-05,
+ "loss": 1.4417,
+ "step": 1820
+ },
+ {
+ "epoch": 4.961852861035422,
+ "grad_norm": 6.735265254974365,
+ "learning_rate": 1.9979820378017017e-05,
+ "loss": 1.3628,
+ "step": 1821
+ },
+ {
+ "epoch": 4.96457765667575,
+ "grad_norm": 7.935208320617676,
+ "learning_rate": 1.997976430351058e-05,
+ "loss": 1.7375,
+ "step": 1822
+ },
+ {
+ "epoch": 4.967302452316076,
+ "grad_norm": 8.892919540405273,
+ "learning_rate": 1.997970815128206e-05,
+ "loss": 1.4785,
+ "step": 1823
+ },
+ {
+ "epoch": 4.970027247956403,
+ "grad_norm": 7.8730998039245605,
+ "learning_rate": 1.9979651921331893e-05,
+ "loss": 1.626,
+ "step": 1824
+ },
+ {
+ "epoch": 4.9727520435967305,
+ "grad_norm": 7.54627799987793,
+ "learning_rate": 1.997959561366052e-05,
+ "loss": 1.7141,
+ "step": 1825
+ },
+ {
+ "epoch": 4.975476839237057,
+ "grad_norm": 8.441864013671875,
+ "learning_rate": 1.9979539228268377e-05,
+ "loss": 1.7312,
+ "step": 1826
+ },
+ {
+ "epoch": 4.978201634877384,
+ "grad_norm": 6.060825824737549,
+ "learning_rate": 1.9979482765155905e-05,
+ "loss": 1.4448,
+ "step": 1827
+ },
+ {
+ "epoch": 4.9809264305177114,
+ "grad_norm": 6.302468299865723,
+ "learning_rate": 1.997942622432354e-05,
+ "loss": 1.3923,
+ "step": 1828
+ },
+ {
+ "epoch": 4.983651226158038,
+ "grad_norm": 4.885349750518799,
+ "learning_rate": 1.9979369605771722e-05,
+ "loss": 1.46,
+ "step": 1829
+ },
+ {
+ "epoch": 4.986376021798365,
+ "grad_norm": 6.704103469848633,
+ "learning_rate": 1.9979312909500902e-05,
+ "loss": 1.3721,
+ "step": 1830
+ },
+ {
+ "epoch": 4.989100817438692,
+ "grad_norm": 8.163519859313965,
+ "learning_rate": 1.997925613551151e-05,
+ "loss": 1.5491,
+ "step": 1831
+ },
+ {
+ "epoch": 4.991825613079019,
+ "grad_norm": 8.34795093536377,
+ "learning_rate": 1.9979199283803994e-05,
+ "loss": 1.5508,
+ "step": 1832
+ },
+ {
+ "epoch": 4.994550408719346,
+ "grad_norm": 6.509402751922607,
+ "learning_rate": 1.9979142354378793e-05,
+ "loss": 1.4331,
+ "step": 1833
+ },
+ {
+ "epoch": 4.997275204359673,
+ "grad_norm": 8.779300689697266,
+ "learning_rate": 1.9979085347236355e-05,
+ "loss": 1.5508,
+ "step": 1834
+ },
+ {
+ "epoch": 5.0,
+ "grad_norm": 6.185759544372559,
+ "learning_rate": 1.997902826237712e-05,
+ "loss": 1.4512,
+ "step": 1835
+ },
+ {
+ "epoch": 5.002724795640327,
+ "grad_norm": 6.25944709777832,
+ "learning_rate": 1.9978971099801532e-05,
+ "loss": 1.2344,
+ "step": 1836
+ },
+ {
+ "epoch": 5.005449591280654,
+ "grad_norm": 5.246951103210449,
+ "learning_rate": 1.997891385951004e-05,
+ "loss": 1.4939,
+ "step": 1837
+ },
+ {
+ "epoch": 5.008174386920981,
+ "grad_norm": 9.330697059631348,
+ "learning_rate": 1.997885654150309e-05,
+ "loss": 1.4509,
+ "step": 1838
+ },
+ {
+ "epoch": 5.010899182561308,
+ "grad_norm": 8.93309211730957,
+ "learning_rate": 1.9978799145781123e-05,
+ "loss": 1.7783,
+ "step": 1839
+ },
+ {
+ "epoch": 5.013623978201635,
+ "grad_norm": 7.102755069732666,
+ "learning_rate": 1.9978741672344594e-05,
+ "loss": 1.2507,
+ "step": 1840
+ },
+ {
+ "epoch": 5.016348773841962,
+ "grad_norm": 7.225795745849609,
+ "learning_rate": 1.9978684121193944e-05,
+ "loss": 1.4971,
+ "step": 1841
+ },
+ {
+ "epoch": 5.0190735694822886,
+ "grad_norm": 5.950686931610107,
+ "learning_rate": 1.997862649232962e-05,
+ "loss": 1.4897,
+ "step": 1842
+ },
+ {
+ "epoch": 5.021798365122616,
+ "grad_norm": 6.798070907592773,
+ "learning_rate": 1.9978568785752077e-05,
+ "loss": 1.4883,
+ "step": 1843
+ },
+ {
+ "epoch": 5.024523160762943,
+ "grad_norm": 7.06300163269043,
+ "learning_rate": 1.997851100146176e-05,
+ "loss": 1.5066,
+ "step": 1844
+ },
+ {
+ "epoch": 5.0272479564032695,
+ "grad_norm": 7.256128311157227,
+ "learning_rate": 1.9978453139459123e-05,
+ "loss": 1.3584,
+ "step": 1845
+ },
+ {
+ "epoch": 5.029972752043597,
+ "grad_norm": 5.9200849533081055,
+ "learning_rate": 1.997839519974461e-05,
+ "loss": 1.2444,
+ "step": 1846
+ },
+ {
+ "epoch": 5.032697547683924,
+ "grad_norm": 6.699158668518066,
+ "learning_rate": 1.9978337182318678e-05,
+ "loss": 1.3015,
+ "step": 1847
+ },
+ {
+ "epoch": 5.03542234332425,
+ "grad_norm": 9.171258926391602,
+ "learning_rate": 1.997827908718178e-05,
+ "loss": 1.4963,
+ "step": 1848
+ },
+ {
+ "epoch": 5.038147138964578,
+ "grad_norm": 7.329318046569824,
+ "learning_rate": 1.9978220914334363e-05,
+ "loss": 1.5491,
+ "step": 1849
+ },
+ {
+ "epoch": 5.040871934604905,
+ "grad_norm": 7.742669105529785,
+ "learning_rate": 1.9978162663776882e-05,
+ "loss": 1.4851,
+ "step": 1850
+ },
+ {
+ "epoch": 5.043596730245231,
+ "grad_norm": 6.855071544647217,
+ "learning_rate": 1.9978104335509796e-05,
+ "loss": 1.2888,
+ "step": 1851
+ },
+ {
+ "epoch": 5.046321525885559,
+ "grad_norm": 6.685558795928955,
+ "learning_rate": 1.997804592953355e-05,
+ "loss": 1.396,
+ "step": 1852
+ },
+ {
+ "epoch": 5.049046321525886,
+ "grad_norm": 6.880176067352295,
+ "learning_rate": 1.9977987445848604e-05,
+ "loss": 1.4087,
+ "step": 1853
+ },
+ {
+ "epoch": 5.051771117166212,
+ "grad_norm": 5.831528663635254,
+ "learning_rate": 1.9977928884455412e-05,
+ "loss": 1.137,
+ "step": 1854
+ },
+ {
+ "epoch": 5.05449591280654,
+ "grad_norm": 7.707138538360596,
+ "learning_rate": 1.9977870245354435e-05,
+ "loss": 1.699,
+ "step": 1855
+ },
+ {
+ "epoch": 5.0572207084468666,
+ "grad_norm": 7.300176620483398,
+ "learning_rate": 1.9977811528546124e-05,
+ "loss": 1.5381,
+ "step": 1856
+ },
+ {
+ "epoch": 5.059945504087193,
+ "grad_norm": 7.214634418487549,
+ "learning_rate": 1.9977752734030935e-05,
+ "loss": 1.5693,
+ "step": 1857
+ },
+ {
+ "epoch": 5.062670299727521,
+ "grad_norm": 10.572924613952637,
+ "learning_rate": 1.9977693861809333e-05,
+ "loss": 1.3801,
+ "step": 1858
+ },
+ {
+ "epoch": 5.0653950953678475,
+ "grad_norm": 7.471271514892578,
+ "learning_rate": 1.997763491188177e-05,
+ "loss": 1.356,
+ "step": 1859
+ },
+ {
+ "epoch": 5.068119891008174,
+ "grad_norm": 6.880827903747559,
+ "learning_rate": 1.9977575884248707e-05,
+ "loss": 1.2148,
+ "step": 1860
+ },
+ {
+ "epoch": 5.070844686648502,
+ "grad_norm": 10.070199012756348,
+ "learning_rate": 1.9977516778910607e-05,
+ "loss": 1.6284,
+ "step": 1861
+ },
+ {
+ "epoch": 5.073569482288828,
+ "grad_norm": 8.91168212890625,
+ "learning_rate": 1.9977457595867924e-05,
+ "loss": 1.585,
+ "step": 1862
+ },
+ {
+ "epoch": 5.076294277929155,
+ "grad_norm": 9.260234832763672,
+ "learning_rate": 1.9977398335121126e-05,
+ "loss": 1.6936,
+ "step": 1863
+ },
+ {
+ "epoch": 5.079019073569483,
+ "grad_norm": 5.806761264801025,
+ "learning_rate": 1.9977338996670665e-05,
+ "loss": 1.3259,
+ "step": 1864
+ },
+ {
+ "epoch": 5.081743869209809,
+ "grad_norm": 7.548891544342041,
+ "learning_rate": 1.9977279580517014e-05,
+ "loss": 1.5637,
+ "step": 1865
+ },
+ {
+ "epoch": 5.084468664850136,
+ "grad_norm": 7.342206954956055,
+ "learning_rate": 1.997722008666063e-05,
+ "loss": 1.4719,
+ "step": 1866
+ },
+ {
+ "epoch": 5.087193460490464,
+ "grad_norm": 7.5871148109436035,
+ "learning_rate": 1.9977160515101975e-05,
+ "loss": 1.2747,
+ "step": 1867
+ },
+ {
+ "epoch": 5.08991825613079,
+ "grad_norm": 7.762136936187744,
+ "learning_rate": 1.9977100865841516e-05,
+ "loss": 1.5693,
+ "step": 1868
+ },
+ {
+ "epoch": 5.092643051771117,
+ "grad_norm": 7.792172431945801,
+ "learning_rate": 1.997704113887972e-05,
+ "loss": 1.5973,
+ "step": 1869
+ },
+ {
+ "epoch": 5.0953678474114446,
+ "grad_norm": 9.21053695678711,
+ "learning_rate": 1.9976981334217046e-05,
+ "loss": 1.6135,
+ "step": 1870
+ },
+ {
+ "epoch": 5.098092643051771,
+ "grad_norm": 6.7395524978637695,
+ "learning_rate": 1.997692145185396e-05,
+ "loss": 1.4824,
+ "step": 1871
+ },
+ {
+ "epoch": 5.100817438692098,
+ "grad_norm": 8.646878242492676,
+ "learning_rate": 1.9976861491790936e-05,
+ "loss": 1.5859,
+ "step": 1872
+ },
+ {
+ "epoch": 5.1035422343324255,
+ "grad_norm": 8.821683883666992,
+ "learning_rate": 1.997680145402843e-05,
+ "loss": 1.5454,
+ "step": 1873
+ },
+ {
+ "epoch": 5.106267029972752,
+ "grad_norm": 8.019302368164062,
+ "learning_rate": 1.997674133856692e-05,
+ "loss": 1.5549,
+ "step": 1874
+ },
+ {
+ "epoch": 5.108991825613079,
+ "grad_norm": 7.5369133949279785,
+ "learning_rate": 1.997668114540687e-05,
+ "loss": 1.4929,
+ "step": 1875
+ },
+ {
+ "epoch": 5.111716621253406,
+ "grad_norm": 8.564929008483887,
+ "learning_rate": 1.997662087454874e-05,
+ "loss": 1.6216,
+ "step": 1876
+ },
+ {
+ "epoch": 5.114441416893733,
+ "grad_norm": 7.038465976715088,
+ "learning_rate": 1.9976560525993015e-05,
+ "loss": 1.2622,
+ "step": 1877
+ },
+ {
+ "epoch": 5.11716621253406,
+ "grad_norm": 7.362818717956543,
+ "learning_rate": 1.997650009974016e-05,
+ "loss": 1.2566,
+ "step": 1878
+ },
+ {
+ "epoch": 5.1198910081743865,
+ "grad_norm": 7.665921211242676,
+ "learning_rate": 1.997643959579064e-05,
+ "loss": 1.2551,
+ "step": 1879
+ },
+ {
+ "epoch": 5.122615803814714,
+ "grad_norm": 7.9219255447387695,
+ "learning_rate": 1.997637901414493e-05,
+ "loss": 1.4392,
+ "step": 1880
+ },
+ {
+ "epoch": 5.125340599455041,
+ "grad_norm": 6.21446418762207,
+ "learning_rate": 1.99763183548035e-05,
+ "loss": 1.3013,
+ "step": 1881
+ },
+ {
+ "epoch": 5.128065395095367,
+ "grad_norm": 6.279685974121094,
+ "learning_rate": 1.9976257617766824e-05,
+ "loss": 1.4744,
+ "step": 1882
+ },
+ {
+ "epoch": 5.130790190735695,
+ "grad_norm": 7.221911430358887,
+ "learning_rate": 1.9976196803035375e-05,
+ "loss": 1.3043,
+ "step": 1883
+ },
+ {
+ "epoch": 5.133514986376022,
+ "grad_norm": 7.0314249992370605,
+ "learning_rate": 1.9976135910609627e-05,
+ "loss": 1.5352,
+ "step": 1884
+ },
+ {
+ "epoch": 5.136239782016348,
+ "grad_norm": 9.1758451461792,
+ "learning_rate": 1.9976074940490052e-05,
+ "loss": 1.396,
+ "step": 1885
+ },
+ {
+ "epoch": 5.138964577656676,
+ "grad_norm": 9.04499626159668,
+ "learning_rate": 1.997601389267713e-05,
+ "loss": 1.4922,
+ "step": 1886
+ },
+ {
+ "epoch": 5.141689373297003,
+ "grad_norm": 9.007833480834961,
+ "learning_rate": 1.997595276717133e-05,
+ "loss": 1.574,
+ "step": 1887
+ },
+ {
+ "epoch": 5.144414168937329,
+ "grad_norm": 8.031352043151855,
+ "learning_rate": 1.9975891563973132e-05,
+ "loss": 1.4495,
+ "step": 1888
+ },
+ {
+ "epoch": 5.147138964577657,
+ "grad_norm": 7.575414180755615,
+ "learning_rate": 1.997583028308301e-05,
+ "loss": 1.438,
+ "step": 1889
+ },
+ {
+ "epoch": 5.1498637602179835,
+ "grad_norm": 6.2173590660095215,
+ "learning_rate": 1.9975768924501444e-05,
+ "loss": 1.3989,
+ "step": 1890
+ },
+ {
+ "epoch": 5.15258855585831,
+ "grad_norm": 8.020543098449707,
+ "learning_rate": 1.997570748822891e-05,
+ "loss": 1.5562,
+ "step": 1891
+ },
+ {
+ "epoch": 5.155313351498638,
+ "grad_norm": 7.866913795471191,
+ "learning_rate": 1.997564597426589e-05,
+ "loss": 1.4888,
+ "step": 1892
+ },
+ {
+ "epoch": 5.1580381471389645,
+ "grad_norm": 6.492123603820801,
+ "learning_rate": 1.9975584382612855e-05,
+ "loss": 1.4153,
+ "step": 1893
+ },
+ {
+ "epoch": 5.160762942779291,
+ "grad_norm": 7.195379257202148,
+ "learning_rate": 1.9975522713270296e-05,
+ "loss": 1.3247,
+ "step": 1894
+ },
+ {
+ "epoch": 5.163487738419619,
+ "grad_norm": 7.053745746612549,
+ "learning_rate": 1.997546096623868e-05,
+ "loss": 1.4705,
+ "step": 1895
+ },
+ {
+ "epoch": 5.166212534059945,
+ "grad_norm": 6.8382415771484375,
+ "learning_rate": 1.99753991415185e-05,
+ "loss": 1.616,
+ "step": 1896
+ },
+ {
+ "epoch": 5.168937329700272,
+ "grad_norm": 13.253124237060547,
+ "learning_rate": 1.997533723911023e-05,
+ "loss": 1.51,
+ "step": 1897
+ },
+ {
+ "epoch": 5.1716621253406,
+ "grad_norm": 8.265405654907227,
+ "learning_rate": 1.997527525901436e-05,
+ "loss": 1.5369,
+ "step": 1898
+ },
+ {
+ "epoch": 5.174386920980926,
+ "grad_norm": 7.4289164543151855,
+ "learning_rate": 1.9975213201231363e-05,
+ "loss": 1.4924,
+ "step": 1899
+ },
+ {
+ "epoch": 5.177111716621253,
+ "grad_norm": 6.071115016937256,
+ "learning_rate": 1.9975151065761728e-05,
+ "loss": 1.4685,
+ "step": 1900
+ },
+ {
+ "epoch": 5.179836512261581,
+ "grad_norm": 6.903639793395996,
+ "learning_rate": 1.9975088852605938e-05,
+ "loss": 1.4624,
+ "step": 1901
+ },
+ {
+ "epoch": 5.182561307901907,
+ "grad_norm": 10.362053871154785,
+ "learning_rate": 1.9975026561764474e-05,
+ "loss": 1.2532,
+ "step": 1902
+ },
+ {
+ "epoch": 5.185286103542234,
+ "grad_norm": 7.644742012023926,
+ "learning_rate": 1.9974964193237823e-05,
+ "loss": 1.5945,
+ "step": 1903
+ },
+ {
+ "epoch": 5.1880108991825615,
+ "grad_norm": 7.478208065032959,
+ "learning_rate": 1.9974901747026474e-05,
+ "loss": 1.2898,
+ "step": 1904
+ },
+ {
+ "epoch": 5.190735694822888,
+ "grad_norm": 9.466500282287598,
+ "learning_rate": 1.997483922313092e-05,
+ "loss": 1.4819,
+ "step": 1905
+ },
+ {
+ "epoch": 5.193460490463215,
+ "grad_norm": 7.108099937438965,
+ "learning_rate": 1.997477662155163e-05,
+ "loss": 1.4153,
+ "step": 1906
+ },
+ {
+ "epoch": 5.1961852861035425,
+ "grad_norm": 7.2346930503845215,
+ "learning_rate": 1.99747139422891e-05,
+ "loss": 1.4583,
+ "step": 1907
+ },
+ {
+ "epoch": 5.198910081743869,
+ "grad_norm": 7.050342082977295,
+ "learning_rate": 1.9974651185343822e-05,
+ "loss": 1.4951,
+ "step": 1908
+ },
+ {
+ "epoch": 5.201634877384196,
+ "grad_norm": 9.188331604003906,
+ "learning_rate": 1.997458835071628e-05,
+ "loss": 1.4048,
+ "step": 1909
+ },
+ {
+ "epoch": 5.204359673024523,
+ "grad_norm": 9.014885902404785,
+ "learning_rate": 1.9974525438406966e-05,
+ "loss": 1.667,
+ "step": 1910
+ },
+ {
+ "epoch": 5.20708446866485,
+ "grad_norm": 37.04357147216797,
+ "learning_rate": 1.9974462448416365e-05,
+ "loss": 1.5117,
+ "step": 1911
+ },
+ {
+ "epoch": 5.209809264305177,
+ "grad_norm": 11.221299171447754,
+ "learning_rate": 1.9974399380744976e-05,
+ "loss": 1.4817,
+ "step": 1912
+ },
+ {
+ "epoch": 5.212534059945504,
+ "grad_norm": 10.714330673217773,
+ "learning_rate": 1.9974336235393283e-05,
+ "loss": 1.405,
+ "step": 1913
+ },
+ {
+ "epoch": 5.215258855585831,
+ "grad_norm": 7.548624038696289,
+ "learning_rate": 1.9974273012361784e-05,
+ "loss": 1.5278,
+ "step": 1914
+ },
+ {
+ "epoch": 5.217983651226158,
+ "grad_norm": 14.866302490234375,
+ "learning_rate": 1.9974209711650964e-05,
+ "loss": 1.6294,
+ "step": 1915
+ },
+ {
+ "epoch": 5.220708446866485,
+ "grad_norm": 10.0299654006958,
+ "learning_rate": 1.9974146333261323e-05,
+ "loss": 1.5884,
+ "step": 1916
+ },
+ {
+ "epoch": 5.223433242506812,
+ "grad_norm": 8.88427448272705,
+ "learning_rate": 1.997408287719335e-05,
+ "loss": 1.5615,
+ "step": 1917
+ },
+ {
+ "epoch": 5.226158038147139,
+ "grad_norm": 9.259163856506348,
+ "learning_rate": 1.997401934344754e-05,
+ "loss": 1.47,
+ "step": 1918
+ },
+ {
+ "epoch": 5.228882833787466,
+ "grad_norm": 7.746896266937256,
+ "learning_rate": 1.997395573202439e-05,
+ "loss": 1.5205,
+ "step": 1919
+ },
+ {
+ "epoch": 5.231607629427793,
+ "grad_norm": 9.229618072509766,
+ "learning_rate": 1.997389204292439e-05,
+ "loss": 1.4465,
+ "step": 1920
+ },
+ {
+ "epoch": 5.23433242506812,
+ "grad_norm": 7.003979682922363,
+ "learning_rate": 1.9973828276148043e-05,
+ "loss": 1.1965,
+ "step": 1921
+ },
+ {
+ "epoch": 5.237057220708447,
+ "grad_norm": 7.926522731781006,
+ "learning_rate": 1.9973764431695844e-05,
+ "loss": 1.5271,
+ "step": 1922
+ },
+ {
+ "epoch": 5.239782016348774,
+ "grad_norm": 9.412335395812988,
+ "learning_rate": 1.9973700509568287e-05,
+ "loss": 1.6592,
+ "step": 1923
+ },
+ {
+ "epoch": 5.2425068119891005,
+ "grad_norm": 9.875425338745117,
+ "learning_rate": 1.997363650976587e-05,
+ "loss": 1.5463,
+ "step": 1924
+ },
+ {
+ "epoch": 5.245231607629428,
+ "grad_norm": 7.646869659423828,
+ "learning_rate": 1.9973572432289094e-05,
+ "loss": 1.5835,
+ "step": 1925
+ },
+ {
+ "epoch": 5.247956403269755,
+ "grad_norm": 7.144330024719238,
+ "learning_rate": 1.9973508277138457e-05,
+ "loss": 1.4333,
+ "step": 1926
+ },
+ {
+ "epoch": 5.2506811989100814,
+ "grad_norm": 8.541711807250977,
+ "learning_rate": 1.9973444044314458e-05,
+ "loss": 1.3984,
+ "step": 1927
+ },
+ {
+ "epoch": 5.253405994550409,
+ "grad_norm": 11.260079383850098,
+ "learning_rate": 1.99733797338176e-05,
+ "loss": 1.5291,
+ "step": 1928
+ },
+ {
+ "epoch": 5.256130790190736,
+ "grad_norm": 8.887072563171387,
+ "learning_rate": 1.9973315345648374e-05,
+ "loss": 1.6733,
+ "step": 1929
+ },
+ {
+ "epoch": 5.258855585831062,
+ "grad_norm": 6.9576029777526855,
+ "learning_rate": 1.9973250879807295e-05,
+ "loss": 1.4594,
+ "step": 1930
+ },
+ {
+ "epoch": 5.26158038147139,
+ "grad_norm": 9.713239669799805,
+ "learning_rate": 1.9973186336294857e-05,
+ "loss": 1.8354,
+ "step": 1931
+ },
+ {
+ "epoch": 5.264305177111717,
+ "grad_norm": 9.130321502685547,
+ "learning_rate": 1.9973121715111568e-05,
+ "loss": 1.4912,
+ "step": 1932
+ },
+ {
+ "epoch": 5.267029972752043,
+ "grad_norm": 6.530932426452637,
+ "learning_rate": 1.997305701625793e-05,
+ "loss": 1.4561,
+ "step": 1933
+ },
+ {
+ "epoch": 5.269754768392371,
+ "grad_norm": 8.118372917175293,
+ "learning_rate": 1.997299223973444e-05,
+ "loss": 1.3738,
+ "step": 1934
+ },
+ {
+ "epoch": 5.272479564032698,
+ "grad_norm": 8.884854316711426,
+ "learning_rate": 1.997292738554161e-05,
+ "loss": 1.5518,
+ "step": 1935
+ },
+ {
+ "epoch": 5.275204359673024,
+ "grad_norm": 8.189332008361816,
+ "learning_rate": 1.997286245367994e-05,
+ "loss": 1.6702,
+ "step": 1936
+ },
+ {
+ "epoch": 5.277929155313352,
+ "grad_norm": 8.00985050201416,
+ "learning_rate": 1.997279744414994e-05,
+ "loss": 1.4741,
+ "step": 1937
+ },
+ {
+ "epoch": 5.2806539509536785,
+ "grad_norm": 8.210562705993652,
+ "learning_rate": 1.9972732356952115e-05,
+ "loss": 1.6299,
+ "step": 1938
+ },
+ {
+ "epoch": 5.283378746594005,
+ "grad_norm": 7.862004280090332,
+ "learning_rate": 1.997266719208697e-05,
+ "loss": 1.2439,
+ "step": 1939
+ },
+ {
+ "epoch": 5.286103542234333,
+ "grad_norm": 7.7602128982543945,
+ "learning_rate": 1.9972601949555015e-05,
+ "loss": 1.8284,
+ "step": 1940
+ },
+ {
+ "epoch": 5.2888283378746594,
+ "grad_norm": 7.245870590209961,
+ "learning_rate": 1.9972536629356756e-05,
+ "loss": 1.7546,
+ "step": 1941
+ },
+ {
+ "epoch": 5.291553133514986,
+ "grad_norm": 6.997647762298584,
+ "learning_rate": 1.99724712314927e-05,
+ "loss": 1.6165,
+ "step": 1942
+ },
+ {
+ "epoch": 5.294277929155314,
+ "grad_norm": 6.528262615203857,
+ "learning_rate": 1.9972405755963362e-05,
+ "loss": 1.4805,
+ "step": 1943
+ },
+ {
+ "epoch": 5.29700272479564,
+ "grad_norm": 7.540473937988281,
+ "learning_rate": 1.997234020276925e-05,
+ "loss": 1.4551,
+ "step": 1944
+ },
+ {
+ "epoch": 5.299727520435967,
+ "grad_norm": 20.5256290435791,
+ "learning_rate": 1.9972274571910873e-05,
+ "loss": 1.3977,
+ "step": 1945
+ },
+ {
+ "epoch": 5.302452316076295,
+ "grad_norm": 8.877883911132812,
+ "learning_rate": 1.9972208863388744e-05,
+ "loss": 1.3507,
+ "step": 1946
+ },
+ {
+ "epoch": 5.305177111716621,
+ "grad_norm": 7.286627292633057,
+ "learning_rate": 1.997214307720337e-05,
+ "loss": 1.46,
+ "step": 1947
+ },
+ {
+ "epoch": 5.307901907356948,
+ "grad_norm": 7.170977592468262,
+ "learning_rate": 1.997207721335527e-05,
+ "loss": 1.5598,
+ "step": 1948
+ },
+ {
+ "epoch": 5.310626702997276,
+ "grad_norm": 6.340270042419434,
+ "learning_rate": 1.997201127184495e-05,
+ "loss": 1.5981,
+ "step": 1949
+ },
+ {
+ "epoch": 5.313351498637602,
+ "grad_norm": 7.464330196380615,
+ "learning_rate": 1.9971945252672927e-05,
+ "loss": 1.5732,
+ "step": 1950
+ },
+ {
+ "epoch": 5.316076294277929,
+ "grad_norm": 7.865177154541016,
+ "learning_rate": 1.9971879155839717e-05,
+ "loss": 1.459,
+ "step": 1951
+ },
+ {
+ "epoch": 5.3188010899182565,
+ "grad_norm": 5.812832832336426,
+ "learning_rate": 1.9971812981345834e-05,
+ "loss": 1.2888,
+ "step": 1952
+ },
+ {
+ "epoch": 5.321525885558583,
+ "grad_norm": 9.448551177978516,
+ "learning_rate": 1.9971746729191792e-05,
+ "loss": 1.707,
+ "step": 1953
+ },
+ {
+ "epoch": 5.32425068119891,
+ "grad_norm": 8.460504531860352,
+ "learning_rate": 1.9971680399378107e-05,
+ "loss": 1.3721,
+ "step": 1954
+ },
+ {
+ "epoch": 5.3269754768392374,
+ "grad_norm": 6.036039352416992,
+ "learning_rate": 1.9971613991905298e-05,
+ "loss": 1.1572,
+ "step": 1955
+ },
+ {
+ "epoch": 5.329700272479564,
+ "grad_norm": 6.652822494506836,
+ "learning_rate": 1.9971547506773875e-05,
+ "loss": 1.3206,
+ "step": 1956
+ },
+ {
+ "epoch": 5.332425068119891,
+ "grad_norm": 7.835741996765137,
+ "learning_rate": 1.9971480943984364e-05,
+ "loss": 1.4736,
+ "step": 1957
+ },
+ {
+ "epoch": 5.335149863760218,
+ "grad_norm": 8.656624794006348,
+ "learning_rate": 1.997141430353728e-05,
+ "loss": 1.3674,
+ "step": 1958
+ },
+ {
+ "epoch": 5.337874659400545,
+ "grad_norm": 8.774532318115234,
+ "learning_rate": 1.9971347585433145e-05,
+ "loss": 1.5103,
+ "step": 1959
+ },
+ {
+ "epoch": 5.340599455040872,
+ "grad_norm": 6.957152366638184,
+ "learning_rate": 1.9971280789672474e-05,
+ "loss": 1.2168,
+ "step": 1960
+ },
+ {
+ "epoch": 5.343324250681199,
+ "grad_norm": 6.78039026260376,
+ "learning_rate": 1.9971213916255788e-05,
+ "loss": 1.4299,
+ "step": 1961
+ },
+ {
+ "epoch": 5.346049046321526,
+ "grad_norm": 6.878546237945557,
+ "learning_rate": 1.997114696518361e-05,
+ "loss": 1.3533,
+ "step": 1962
+ },
+ {
+ "epoch": 5.348773841961853,
+ "grad_norm": 6.876976490020752,
+ "learning_rate": 1.997107993645646e-05,
+ "loss": 1.3473,
+ "step": 1963
+ },
+ {
+ "epoch": 5.35149863760218,
+ "grad_norm": 9.121848106384277,
+ "learning_rate": 1.997101283007486e-05,
+ "loss": 1.6475,
+ "step": 1964
+ },
+ {
+ "epoch": 5.354223433242507,
+ "grad_norm": 7.719760894775391,
+ "learning_rate": 1.9970945646039336e-05,
+ "loss": 1.6538,
+ "step": 1965
+ },
+ {
+ "epoch": 5.356948228882834,
+ "grad_norm": 7.375722885131836,
+ "learning_rate": 1.9970878384350404e-05,
+ "loss": 1.3501,
+ "step": 1966
+ },
+ {
+ "epoch": 5.359673024523161,
+ "grad_norm": 7.239620208740234,
+ "learning_rate": 1.9970811045008592e-05,
+ "loss": 1.3867,
+ "step": 1967
+ },
+ {
+ "epoch": 5.362397820163488,
+ "grad_norm": 7.577230453491211,
+ "learning_rate": 1.9970743628014428e-05,
+ "loss": 1.6536,
+ "step": 1968
+ },
+ {
+ "epoch": 5.3651226158038146,
+ "grad_norm": 6.134706974029541,
+ "learning_rate": 1.997067613336843e-05,
+ "loss": 1.4597,
+ "step": 1969
+ },
+ {
+ "epoch": 5.367847411444142,
+ "grad_norm": 7.231548309326172,
+ "learning_rate": 1.997060856107113e-05,
+ "loss": 1.6169,
+ "step": 1970
+ },
+ {
+ "epoch": 5.370572207084469,
+ "grad_norm": 8.130701065063477,
+ "learning_rate": 1.9970540911123052e-05,
+ "loss": 1.54,
+ "step": 1971
+ },
+ {
+ "epoch": 5.3732970027247955,
+ "grad_norm": 6.530113220214844,
+ "learning_rate": 1.997047318352472e-05,
+ "loss": 1.2562,
+ "step": 1972
+ },
+ {
+ "epoch": 5.376021798365123,
+ "grad_norm": 8.105958938598633,
+ "learning_rate": 1.9970405378276666e-05,
+ "loss": 1.4736,
+ "step": 1973
+ },
+ {
+ "epoch": 5.37874659400545,
+ "grad_norm": 8.039284706115723,
+ "learning_rate": 1.997033749537941e-05,
+ "loss": 1.6919,
+ "step": 1974
+ },
+ {
+ "epoch": 5.381471389645776,
+ "grad_norm": 7.621415138244629,
+ "learning_rate": 1.9970269534833492e-05,
+ "loss": 1.2859,
+ "step": 1975
+ },
+ {
+ "epoch": 5.384196185286104,
+ "grad_norm": 6.722444534301758,
+ "learning_rate": 1.9970201496639436e-05,
+ "loss": 1.5076,
+ "step": 1976
+ },
+ {
+ "epoch": 5.386920980926431,
+ "grad_norm": 6.524257183074951,
+ "learning_rate": 1.997013338079777e-05,
+ "loss": 1.2942,
+ "step": 1977
+ },
+ {
+ "epoch": 5.389645776566757,
+ "grad_norm": 8.051794052124023,
+ "learning_rate": 1.9970065187309025e-05,
+ "loss": 1.5122,
+ "step": 1978
+ },
+ {
+ "epoch": 5.392370572207085,
+ "grad_norm": 8.080615997314453,
+ "learning_rate": 1.9969996916173735e-05,
+ "loss": 1.7144,
+ "step": 1979
+ },
+ {
+ "epoch": 5.395095367847412,
+ "grad_norm": 9.678326606750488,
+ "learning_rate": 1.996992856739243e-05,
+ "loss": 1.5857,
+ "step": 1980
+ },
+ {
+ "epoch": 5.397820163487738,
+ "grad_norm": 8.182806968688965,
+ "learning_rate": 1.9969860140965643e-05,
+ "loss": 1.4128,
+ "step": 1981
+ },
+ {
+ "epoch": 5.400544959128065,
+ "grad_norm": 7.851433277130127,
+ "learning_rate": 1.9969791636893904e-05,
+ "loss": 1.4243,
+ "step": 1982
+ },
+ {
+ "epoch": 5.4032697547683926,
+ "grad_norm": 7.919556617736816,
+ "learning_rate": 1.9969723055177748e-05,
+ "loss": 1.5793,
+ "step": 1983
+ },
+ {
+ "epoch": 5.405994550408719,
+ "grad_norm": 7.096700191497803,
+ "learning_rate": 1.9969654395817714e-05,
+ "loss": 1.5078,
+ "step": 1984
+ },
+ {
+ "epoch": 5.408719346049046,
+ "grad_norm": 8.100212097167969,
+ "learning_rate": 1.9969585658814327e-05,
+ "loss": 1.6006,
+ "step": 1985
+ },
+ {
+ "epoch": 5.4114441416893735,
+ "grad_norm": 7.162127494812012,
+ "learning_rate": 1.9969516844168132e-05,
+ "loss": 1.3657,
+ "step": 1986
+ },
+ {
+ "epoch": 5.4141689373297,
+ "grad_norm": 10.330747604370117,
+ "learning_rate": 1.9969447951879655e-05,
+ "loss": 1.6113,
+ "step": 1987
+ },
+ {
+ "epoch": 5.416893732970027,
+ "grad_norm": 8.385213851928711,
+ "learning_rate": 1.9969378981949445e-05,
+ "loss": 1.4902,
+ "step": 1988
+ },
+ {
+ "epoch": 5.419618528610354,
+ "grad_norm": 8.500687599182129,
+ "learning_rate": 1.996930993437803e-05,
+ "loss": 1.3843,
+ "step": 1989
+ },
+ {
+ "epoch": 5.422343324250681,
+ "grad_norm": 8.683425903320312,
+ "learning_rate": 1.9969240809165948e-05,
+ "loss": 1.3828,
+ "step": 1990
+ },
+ {
+ "epoch": 5.425068119891008,
+ "grad_norm": 7.9052958488464355,
+ "learning_rate": 1.9969171606313742e-05,
+ "loss": 1.4419,
+ "step": 1991
+ },
+ {
+ "epoch": 5.427792915531335,
+ "grad_norm": 7.818244934082031,
+ "learning_rate": 1.9969102325821947e-05,
+ "loss": 1.6282,
+ "step": 1992
+ },
+ {
+ "epoch": 5.430517711171662,
+ "grad_norm": 7.670775890350342,
+ "learning_rate": 1.9969032967691106e-05,
+ "loss": 1.5955,
+ "step": 1993
+ },
+ {
+ "epoch": 5.433242506811989,
+ "grad_norm": 9.080238342285156,
+ "learning_rate": 1.9968963531921753e-05,
+ "loss": 1.553,
+ "step": 1994
+ },
+ {
+ "epoch": 5.435967302452316,
+ "grad_norm": 9.8353853225708,
+ "learning_rate": 1.9968894018514434e-05,
+ "loss": 1.4219,
+ "step": 1995
+ },
+ {
+ "epoch": 5.438692098092643,
+ "grad_norm": 7.02793550491333,
+ "learning_rate": 1.996882442746969e-05,
+ "loss": 1.3164,
+ "step": 1996
+ },
+ {
+ "epoch": 5.44141689373297,
+ "grad_norm": 6.971395492553711,
+ "learning_rate": 1.9968754758788064e-05,
+ "loss": 1.4983,
+ "step": 1997
+ },
+ {
+ "epoch": 5.444141689373297,
+ "grad_norm": 7.428837776184082,
+ "learning_rate": 1.9968685012470097e-05,
+ "loss": 1.5442,
+ "step": 1998
+ },
+ {
+ "epoch": 5.446866485013624,
+ "grad_norm": 7.762703895568848,
+ "learning_rate": 1.996861518851633e-05,
+ "loss": 1.6121,
+ "step": 1999
+ },
+ {
+ "epoch": 5.449591280653951,
+ "grad_norm": 6.860504627227783,
+ "learning_rate": 1.996854528692731e-05,
+ "loss": 1.4131,
+ "step": 2000
+ },
+ {
+ "epoch": 5.452316076294278,
+ "grad_norm": 9.326668739318848,
+ "learning_rate": 1.996847530770358e-05,
+ "loss": 1.2739,
+ "step": 2001
+ },
+ {
+ "epoch": 5.455040871934605,
+ "grad_norm": 8.93021011352539,
+ "learning_rate": 1.996840525084568e-05,
+ "loss": 1.6929,
+ "step": 2002
+ },
+ {
+ "epoch": 5.4577656675749315,
+ "grad_norm": 8.212053298950195,
+ "learning_rate": 1.996833511635417e-05,
+ "loss": 1.3025,
+ "step": 2003
+ },
+ {
+ "epoch": 5.460490463215259,
+ "grad_norm": 9.911699295043945,
+ "learning_rate": 1.996826490422958e-05,
+ "loss": 1.5183,
+ "step": 2004
+ },
+ {
+ "epoch": 5.463215258855586,
+ "grad_norm": 7.174154281616211,
+ "learning_rate": 1.9968194614472465e-05,
+ "loss": 1.2676,
+ "step": 2005
+ },
+ {
+ "epoch": 5.4659400544959125,
+ "grad_norm": 8.777009010314941,
+ "learning_rate": 1.9968124247083375e-05,
+ "loss": 1.2917,
+ "step": 2006
+ },
+ {
+ "epoch": 5.46866485013624,
+ "grad_norm": 7.669890880584717,
+ "learning_rate": 1.996805380206285e-05,
+ "loss": 1.4783,
+ "step": 2007
+ },
+ {
+ "epoch": 5.471389645776567,
+ "grad_norm": 8.551259994506836,
+ "learning_rate": 1.9967983279411442e-05,
+ "loss": 1.3237,
+ "step": 2008
+ },
+ {
+ "epoch": 5.474114441416893,
+ "grad_norm": 7.941195487976074,
+ "learning_rate": 1.9967912679129708e-05,
+ "loss": 1.5247,
+ "step": 2009
+ },
+ {
+ "epoch": 5.476839237057221,
+ "grad_norm": 8.978386878967285,
+ "learning_rate": 1.9967842001218183e-05,
+ "loss": 1.4792,
+ "step": 2010
+ },
+ {
+ "epoch": 5.479564032697548,
+ "grad_norm": 6.765439987182617,
+ "learning_rate": 1.9967771245677427e-05,
+ "loss": 1.707,
+ "step": 2011
+ },
+ {
+ "epoch": 5.482288828337874,
+ "grad_norm": 9.516766548156738,
+ "learning_rate": 1.996770041250799e-05,
+ "loss": 1.5715,
+ "step": 2012
+ },
+ {
+ "epoch": 5.485013623978202,
+ "grad_norm": 7.124636650085449,
+ "learning_rate": 1.9967629501710423e-05,
+ "loss": 1.3438,
+ "step": 2013
+ },
+ {
+ "epoch": 5.487738419618529,
+ "grad_norm": 20.968006134033203,
+ "learning_rate": 1.996755851328528e-05,
+ "loss": 1.5161,
+ "step": 2014
+ },
+ {
+ "epoch": 5.490463215258855,
+ "grad_norm": 7.957486152648926,
+ "learning_rate": 1.9967487447233114e-05,
+ "loss": 1.593,
+ "step": 2015
+ },
+ {
+ "epoch": 5.493188010899183,
+ "grad_norm": 5.967541217803955,
+ "learning_rate": 1.9967416303554472e-05,
+ "loss": 1.2258,
+ "step": 2016
+ },
+ {
+ "epoch": 5.4959128065395095,
+ "grad_norm": 7.26482629776001,
+ "learning_rate": 1.9967345082249912e-05,
+ "loss": 1.4187,
+ "step": 2017
+ },
+ {
+ "epoch": 5.498637602179836,
+ "grad_norm": 7.375432014465332,
+ "learning_rate": 1.9967273783319993e-05,
+ "loss": 1.3284,
+ "step": 2018
+ },
+ {
+ "epoch": 5.501362397820164,
+ "grad_norm": 11.6714506149292,
+ "learning_rate": 1.996720240676527e-05,
+ "loss": 1.4417,
+ "step": 2019
+ },
+ {
+ "epoch": 5.5040871934604905,
+ "grad_norm": 6.783967971801758,
+ "learning_rate": 1.9967130952586286e-05,
+ "loss": 1.5923,
+ "step": 2020
+ },
+ {
+ "epoch": 5.506811989100817,
+ "grad_norm": 10.09848690032959,
+ "learning_rate": 1.9967059420783612e-05,
+ "loss": 1.5264,
+ "step": 2021
+ },
+ {
+ "epoch": 5.509536784741145,
+ "grad_norm": 6.087915897369385,
+ "learning_rate": 1.9966987811357798e-05,
+ "loss": 1.4722,
+ "step": 2022
+ },
+ {
+ "epoch": 5.512261580381471,
+ "grad_norm": 5.5742998123168945,
+ "learning_rate": 1.9966916124309404e-05,
+ "loss": 1.3296,
+ "step": 2023
+ },
+ {
+ "epoch": 5.514986376021798,
+ "grad_norm": 8.72879695892334,
+ "learning_rate": 1.9966844359638992e-05,
+ "loss": 1.4121,
+ "step": 2024
+ },
+ {
+ "epoch": 5.517711171662126,
+ "grad_norm": 7.758880615234375,
+ "learning_rate": 1.9966772517347114e-05,
+ "loss": 1.3042,
+ "step": 2025
+ },
+ {
+ "epoch": 5.520435967302452,
+ "grad_norm": 7.967844486236572,
+ "learning_rate": 1.9966700597434333e-05,
+ "loss": 1.502,
+ "step": 2026
+ },
+ {
+ "epoch": 5.523160762942779,
+ "grad_norm": 6.834554195404053,
+ "learning_rate": 1.9966628599901208e-05,
+ "loss": 1.3772,
+ "step": 2027
+ },
+ {
+ "epoch": 5.525885558583107,
+ "grad_norm": 8.97197151184082,
+ "learning_rate": 1.99665565247483e-05,
+ "loss": 1.6299,
+ "step": 2028
+ },
+ {
+ "epoch": 5.528610354223433,
+ "grad_norm": 7.702461242675781,
+ "learning_rate": 1.9966484371976168e-05,
+ "loss": 1.7761,
+ "step": 2029
+ },
+ {
+ "epoch": 5.53133514986376,
+ "grad_norm": 12.832172393798828,
+ "learning_rate": 1.996641214158538e-05,
+ "loss": 1.6763,
+ "step": 2030
+ },
+ {
+ "epoch": 5.5340599455040875,
+ "grad_norm": 7.488429069519043,
+ "learning_rate": 1.9966339833576496e-05,
+ "loss": 1.4131,
+ "step": 2031
+ },
+ {
+ "epoch": 5.536784741144414,
+ "grad_norm": 7.440750598907471,
+ "learning_rate": 1.9966267447950076e-05,
+ "loss": 1.6399,
+ "step": 2032
+ },
+ {
+ "epoch": 5.539509536784741,
+ "grad_norm": 7.981937408447266,
+ "learning_rate": 1.9966194984706686e-05,
+ "loss": 1.6094,
+ "step": 2033
+ },
+ {
+ "epoch": 5.5422343324250685,
+ "grad_norm": 7.988561630249023,
+ "learning_rate": 1.996612244384689e-05,
+ "loss": 1.6282,
+ "step": 2034
+ },
+ {
+ "epoch": 5.544959128065395,
+ "grad_norm": 12.110140800476074,
+ "learning_rate": 1.9966049825371254e-05,
+ "loss": 1.6382,
+ "step": 2035
+ },
+ {
+ "epoch": 5.547683923705722,
+ "grad_norm": 6.685496807098389,
+ "learning_rate": 1.9965977129280344e-05,
+ "loss": 1.4629,
+ "step": 2036
+ },
+ {
+ "epoch": 5.550408719346049,
+ "grad_norm": 13.49795150756836,
+ "learning_rate": 1.9965904355574723e-05,
+ "loss": 1.5266,
+ "step": 2037
+ },
+ {
+ "epoch": 5.553133514986376,
+ "grad_norm": 13.989706993103027,
+ "learning_rate": 1.996583150425496e-05,
+ "loss": 1.8953,
+ "step": 2038
+ },
+ {
+ "epoch": 5.555858310626703,
+ "grad_norm": 7.643697261810303,
+ "learning_rate": 1.996575857532162e-05,
+ "loss": 1.5239,
+ "step": 2039
+ },
+ {
+ "epoch": 5.55858310626703,
+ "grad_norm": 8.651869773864746,
+ "learning_rate": 1.996568556877528e-05,
+ "loss": 1.4373,
+ "step": 2040
+ },
+ {
+ "epoch": 5.561307901907357,
+ "grad_norm": 6.885852813720703,
+ "learning_rate": 1.996561248461649e-05,
+ "loss": 1.4758,
+ "step": 2041
+ },
+ {
+ "epoch": 5.564032697547684,
+ "grad_norm": 7.442098140716553,
+ "learning_rate": 1.996553932284584e-05,
+ "loss": 1.4558,
+ "step": 2042
+ },
+ {
+ "epoch": 5.566757493188011,
+ "grad_norm": 6.795966148376465,
+ "learning_rate": 1.9965466083463886e-05,
+ "loss": 1.2732,
+ "step": 2043
+ },
+ {
+ "epoch": 5.569482288828338,
+ "grad_norm": 8.581652641296387,
+ "learning_rate": 1.9965392766471205e-05,
+ "loss": 1.708,
+ "step": 2044
+ },
+ {
+ "epoch": 5.572207084468665,
+ "grad_norm": 7.221965312957764,
+ "learning_rate": 1.9965319371868367e-05,
+ "loss": 1.323,
+ "step": 2045
+ },
+ {
+ "epoch": 5.574931880108992,
+ "grad_norm": 5.831789493560791,
+ "learning_rate": 1.996524589965594e-05,
+ "loss": 1.5112,
+ "step": 2046
+ },
+ {
+ "epoch": 5.577656675749319,
+ "grad_norm": 6.767343997955322,
+ "learning_rate": 1.99651723498345e-05,
+ "loss": 1.4597,
+ "step": 2047
+ },
+ {
+ "epoch": 5.580381471389646,
+ "grad_norm": 9.21025276184082,
+ "learning_rate": 1.9965098722404617e-05,
+ "loss": 1.4497,
+ "step": 2048
+ },
+ {
+ "epoch": 5.583106267029972,
+ "grad_norm": 7.779251575469971,
+ "learning_rate": 1.9965025017366867e-05,
+ "loss": 1.5552,
+ "step": 2049
+ },
+ {
+ "epoch": 5.5858310626703,
+ "grad_norm": 7.790543079376221,
+ "learning_rate": 1.9964951234721824e-05,
+ "loss": 1.5339,
+ "step": 2050
+ },
+ {
+ "epoch": 5.5885558583106265,
+ "grad_norm": 6.558294773101807,
+ "learning_rate": 1.996487737447006e-05,
+ "loss": 1.26,
+ "step": 2051
+ },
+ {
+ "epoch": 5.591280653950953,
+ "grad_norm": 9.148202896118164,
+ "learning_rate": 1.9964803436612152e-05,
+ "loss": 1.521,
+ "step": 2052
+ },
+ {
+ "epoch": 5.594005449591281,
+ "grad_norm": 8.626147270202637,
+ "learning_rate": 1.9964729421148676e-05,
+ "loss": 1.5474,
+ "step": 2053
+ },
+ {
+ "epoch": 5.5967302452316074,
+ "grad_norm": 10.415085792541504,
+ "learning_rate": 1.9964655328080207e-05,
+ "loss": 1.3,
+ "step": 2054
+ },
+ {
+ "epoch": 5.599455040871934,
+ "grad_norm": 7.9319634437561035,
+ "learning_rate": 1.9964581157407324e-05,
+ "loss": 1.231,
+ "step": 2055
+ },
+ {
+ "epoch": 5.602179836512262,
+ "grad_norm": 6.71903657913208,
+ "learning_rate": 1.99645069091306e-05,
+ "loss": 1.2703,
+ "step": 2056
+ },
+ {
+ "epoch": 5.604904632152588,
+ "grad_norm": 7.845226764678955,
+ "learning_rate": 1.996443258325062e-05,
+ "loss": 1.5466,
+ "step": 2057
+ },
+ {
+ "epoch": 5.607629427792915,
+ "grad_norm": 7.468419551849365,
+ "learning_rate": 1.996435817976796e-05,
+ "loss": 1.4768,
+ "step": 2058
+ },
+ {
+ "epoch": 5.610354223433243,
+ "grad_norm": 7.999795913696289,
+ "learning_rate": 1.99642836986832e-05,
+ "loss": 1.6245,
+ "step": 2059
+ },
+ {
+ "epoch": 5.613079019073569,
+ "grad_norm": 7.380253314971924,
+ "learning_rate": 1.9964209139996917e-05,
+ "loss": 1.4014,
+ "step": 2060
+ },
+ {
+ "epoch": 5.615803814713896,
+ "grad_norm": 6.440184593200684,
+ "learning_rate": 1.9964134503709695e-05,
+ "loss": 1.3333,
+ "step": 2061
+ },
+ {
+ "epoch": 5.618528610354224,
+ "grad_norm": 9.359847068786621,
+ "learning_rate": 1.9964059789822118e-05,
+ "loss": 1.791,
+ "step": 2062
+ },
+ {
+ "epoch": 5.62125340599455,
+ "grad_norm": 6.5328545570373535,
+ "learning_rate": 1.9963984998334757e-05,
+ "loss": 1.4441,
+ "step": 2063
+ },
+ {
+ "epoch": 5.623978201634877,
+ "grad_norm": 6.32890510559082,
+ "learning_rate": 1.9963910129248206e-05,
+ "loss": 1.4219,
+ "step": 2064
+ },
+ {
+ "epoch": 5.6267029972752045,
+ "grad_norm": 7.325685501098633,
+ "learning_rate": 1.9963835182563043e-05,
+ "loss": 1.5687,
+ "step": 2065
+ },
+ {
+ "epoch": 5.629427792915531,
+ "grad_norm": 6.925869941711426,
+ "learning_rate": 1.996376015827985e-05,
+ "loss": 1.2864,
+ "step": 2066
+ },
+ {
+ "epoch": 5.632152588555858,
+ "grad_norm": 5.9621357917785645,
+ "learning_rate": 1.9963685056399217e-05,
+ "loss": 1.3557,
+ "step": 2067
+ },
+ {
+ "epoch": 5.6348773841961854,
+ "grad_norm": 7.199607849121094,
+ "learning_rate": 1.9963609876921722e-05,
+ "loss": 1.5796,
+ "step": 2068
+ },
+ {
+ "epoch": 5.637602179836512,
+ "grad_norm": 8.235106468200684,
+ "learning_rate": 1.9963534619847956e-05,
+ "loss": 1.6292,
+ "step": 2069
+ },
+ {
+ "epoch": 5.640326975476839,
+ "grad_norm": 9.266824722290039,
+ "learning_rate": 1.9963459285178505e-05,
+ "loss": 1.5454,
+ "step": 2070
+ },
+ {
+ "epoch": 5.643051771117166,
+ "grad_norm": 8.1749906539917,
+ "learning_rate": 1.996338387291395e-05,
+ "loss": 1.3257,
+ "step": 2071
+ },
+ {
+ "epoch": 5.645776566757493,
+ "grad_norm": 8.913423538208008,
+ "learning_rate": 1.9963308383054883e-05,
+ "loss": 1.3979,
+ "step": 2072
+ },
+ {
+ "epoch": 5.64850136239782,
+ "grad_norm": 7.577320575714111,
+ "learning_rate": 1.9963232815601896e-05,
+ "loss": 1.488,
+ "step": 2073
+ },
+ {
+ "epoch": 5.651226158038147,
+ "grad_norm": 8.939695358276367,
+ "learning_rate": 1.9963157170555564e-05,
+ "loss": 1.4382,
+ "step": 2074
+ },
+ {
+ "epoch": 5.653950953678474,
+ "grad_norm": 6.884528160095215,
+ "learning_rate": 1.996308144791649e-05,
+ "loss": 1.4756,
+ "step": 2075
+ },
+ {
+ "epoch": 5.656675749318801,
+ "grad_norm": 8.004799842834473,
+ "learning_rate": 1.9963005647685256e-05,
+ "loss": 1.2793,
+ "step": 2076
+ },
+ {
+ "epoch": 5.659400544959128,
+ "grad_norm": 7.5881876945495605,
+ "learning_rate": 1.9962929769862456e-05,
+ "loss": 1.5342,
+ "step": 2077
+ },
+ {
+ "epoch": 5.662125340599455,
+ "grad_norm": 6.353926181793213,
+ "learning_rate": 1.996285381444868e-05,
+ "loss": 1.4609,
+ "step": 2078
+ },
+ {
+ "epoch": 5.664850136239782,
+ "grad_norm": 6.638195037841797,
+ "learning_rate": 1.9962777781444518e-05,
+ "loss": 1.6111,
+ "step": 2079
+ },
+ {
+ "epoch": 5.667574931880109,
+ "grad_norm": 7.294384002685547,
+ "learning_rate": 1.9962701670850566e-05,
+ "loss": 1.6304,
+ "step": 2080
+ },
+ {
+ "epoch": 5.670299727520436,
+ "grad_norm": 5.5814080238342285,
+ "learning_rate": 1.996262548266741e-05,
+ "loss": 1.3074,
+ "step": 2081
+ },
+ {
+ "epoch": 5.6730245231607626,
+ "grad_norm": 7.207130432128906,
+ "learning_rate": 1.996254921689565e-05,
+ "loss": 1.4958,
+ "step": 2082
+ },
+ {
+ "epoch": 5.67574931880109,
+ "grad_norm": 7.843624591827393,
+ "learning_rate": 1.9962472873535877e-05,
+ "loss": 1.6235,
+ "step": 2083
+ },
+ {
+ "epoch": 5.678474114441417,
+ "grad_norm": 8.740017890930176,
+ "learning_rate": 1.9962396452588684e-05,
+ "loss": 1.5449,
+ "step": 2084
+ },
+ {
+ "epoch": 5.6811989100817435,
+ "grad_norm": 11.211609840393066,
+ "learning_rate": 1.996231995405467e-05,
+ "loss": 1.2988,
+ "step": 2085
+ },
+ {
+ "epoch": 5.683923705722071,
+ "grad_norm": 6.239091873168945,
+ "learning_rate": 1.996224337793443e-05,
+ "loss": 1.1233,
+ "step": 2086
+ },
+ {
+ "epoch": 5.686648501362398,
+ "grad_norm": 7.952038288116455,
+ "learning_rate": 1.996216672422856e-05,
+ "loss": 1.5242,
+ "step": 2087
+ },
+ {
+ "epoch": 5.689373297002724,
+ "grad_norm": 7.918951988220215,
+ "learning_rate": 1.996208999293765e-05,
+ "loss": 1.4192,
+ "step": 2088
+ },
+ {
+ "epoch": 5.692098092643052,
+ "grad_norm": 6.894144058227539,
+ "learning_rate": 1.9962013184062308e-05,
+ "loss": 1.3533,
+ "step": 2089
+ },
+ {
+ "epoch": 5.694822888283379,
+ "grad_norm": 6.77570915222168,
+ "learning_rate": 1.996193629760313e-05,
+ "loss": 1.2773,
+ "step": 2090
+ },
+ {
+ "epoch": 5.697547683923705,
+ "grad_norm": 7.05912446975708,
+ "learning_rate": 1.9961859333560708e-05,
+ "loss": 1.5557,
+ "step": 2091
+ },
+ {
+ "epoch": 5.700272479564033,
+ "grad_norm": 8.893268585205078,
+ "learning_rate": 1.9961782291935648e-05,
+ "loss": 1.498,
+ "step": 2092
+ },
+ {
+ "epoch": 5.70299727520436,
+ "grad_norm": 8.7378511428833,
+ "learning_rate": 1.9961705172728547e-05,
+ "loss": 1.6077,
+ "step": 2093
+ },
+ {
+ "epoch": 5.705722070844686,
+ "grad_norm": 7.976889610290527,
+ "learning_rate": 1.996162797594001e-05,
+ "loss": 1.6121,
+ "step": 2094
+ },
+ {
+ "epoch": 5.708446866485014,
+ "grad_norm": 6.766124725341797,
+ "learning_rate": 1.9961550701570633e-05,
+ "loss": 1.1616,
+ "step": 2095
+ },
+ {
+ "epoch": 5.7111716621253406,
+ "grad_norm": 7.024877548217773,
+ "learning_rate": 1.996147334962102e-05,
+ "loss": 1.5349,
+ "step": 2096
+ },
+ {
+ "epoch": 5.713896457765667,
+ "grad_norm": 7.896556854248047,
+ "learning_rate": 1.9961395920091773e-05,
+ "loss": 1.4443,
+ "step": 2097
+ },
+ {
+ "epoch": 5.716621253405995,
+ "grad_norm": 6.779217720031738,
+ "learning_rate": 1.9961318412983494e-05,
+ "loss": 1.5242,
+ "step": 2098
+ },
+ {
+ "epoch": 5.7193460490463215,
+ "grad_norm": 7.419445991516113,
+ "learning_rate": 1.996124082829679e-05,
+ "loss": 1.6218,
+ "step": 2099
+ },
+ {
+ "epoch": 5.722070844686648,
+ "grad_norm": 7.950963973999023,
+ "learning_rate": 1.996116316603226e-05,
+ "loss": 1.7378,
+ "step": 2100
+ },
+ {
+ "epoch": 5.724795640326976,
+ "grad_norm": 7.401852130889893,
+ "learning_rate": 1.9961085426190516e-05,
+ "loss": 1.7026,
+ "step": 2101
+ },
+ {
+ "epoch": 5.727520435967302,
+ "grad_norm": 7.2434234619140625,
+ "learning_rate": 1.9961007608772157e-05,
+ "loss": 1.5139,
+ "step": 2102
+ },
+ {
+ "epoch": 5.730245231607629,
+ "grad_norm": 7.984586238861084,
+ "learning_rate": 1.996092971377779e-05,
+ "loss": 1.5962,
+ "step": 2103
+ },
+ {
+ "epoch": 5.732970027247957,
+ "grad_norm": 7.417317867279053,
+ "learning_rate": 1.996085174120803e-05,
+ "loss": 1.6494,
+ "step": 2104
+ },
+ {
+ "epoch": 5.735694822888283,
+ "grad_norm": 7.8913140296936035,
+ "learning_rate": 1.9960773691063468e-05,
+ "loss": 1.5664,
+ "step": 2105
+ },
+ {
+ "epoch": 5.73841961852861,
+ "grad_norm": 8.56258773803711,
+ "learning_rate": 1.9960695563344727e-05,
+ "loss": 1.5811,
+ "step": 2106
+ },
+ {
+ "epoch": 5.741144414168938,
+ "grad_norm": 7.104063987731934,
+ "learning_rate": 1.9960617358052404e-05,
+ "loss": 1.4961,
+ "step": 2107
+ },
+ {
+ "epoch": 5.743869209809264,
+ "grad_norm": 6.0080766677856445,
+ "learning_rate": 1.9960539075187116e-05,
+ "loss": 1.2708,
+ "step": 2108
+ },
+ {
+ "epoch": 5.746594005449591,
+ "grad_norm": 7.973326206207275,
+ "learning_rate": 1.996046071474947e-05,
+ "loss": 1.6682,
+ "step": 2109
+ },
+ {
+ "epoch": 5.7493188010899186,
+ "grad_norm": 8.521141052246094,
+ "learning_rate": 1.996038227674008e-05,
+ "loss": 1.3582,
+ "step": 2110
+ },
+ {
+ "epoch": 5.752043596730245,
+ "grad_norm": 6.58418083190918,
+ "learning_rate": 1.9960303761159552e-05,
+ "loss": 1.4167,
+ "step": 2111
+ },
+ {
+ "epoch": 5.754768392370572,
+ "grad_norm": 7.55171537399292,
+ "learning_rate": 1.9960225168008498e-05,
+ "loss": 1.2144,
+ "step": 2112
+ },
+ {
+ "epoch": 5.7574931880108995,
+ "grad_norm": 7.937721252441406,
+ "learning_rate": 1.9960146497287532e-05,
+ "loss": 1.6614,
+ "step": 2113
+ },
+ {
+ "epoch": 5.760217983651226,
+ "grad_norm": 8.056745529174805,
+ "learning_rate": 1.9960067748997264e-05,
+ "loss": 1.6538,
+ "step": 2114
+ },
+ {
+ "epoch": 5.762942779291553,
+ "grad_norm": 7.476974010467529,
+ "learning_rate": 1.995998892313831e-05,
+ "loss": 1.5037,
+ "step": 2115
+ },
+ {
+ "epoch": 5.76566757493188,
+ "grad_norm": 6.951935291290283,
+ "learning_rate": 1.9959910019711282e-05,
+ "loss": 1.4297,
+ "step": 2116
+ },
+ {
+ "epoch": 5.768392370572207,
+ "grad_norm": 9.600790977478027,
+ "learning_rate": 1.9959831038716798e-05,
+ "loss": 1.7073,
+ "step": 2117
+ },
+ {
+ "epoch": 5.771117166212534,
+ "grad_norm": 7.262009143829346,
+ "learning_rate": 1.9959751980155465e-05,
+ "loss": 1.5393,
+ "step": 2118
+ },
+ {
+ "epoch": 5.773841961852861,
+ "grad_norm": 7.151832103729248,
+ "learning_rate": 1.995967284402791e-05,
+ "loss": 1.6116,
+ "step": 2119
+ },
+ {
+ "epoch": 5.776566757493188,
+ "grad_norm": 7.455227375030518,
+ "learning_rate": 1.995959363033474e-05,
+ "loss": 1.4209,
+ "step": 2120
+ },
+ {
+ "epoch": 5.779291553133515,
+ "grad_norm": 7.272648811340332,
+ "learning_rate": 1.9959514339076575e-05,
+ "loss": 1.5452,
+ "step": 2121
+ },
+ {
+ "epoch": 5.782016348773842,
+ "grad_norm": 9.09617805480957,
+ "learning_rate": 1.9959434970254036e-05,
+ "loss": 1.5681,
+ "step": 2122
+ },
+ {
+ "epoch": 5.784741144414169,
+ "grad_norm": 6.741121292114258,
+ "learning_rate": 1.9959355523867734e-05,
+ "loss": 1.4624,
+ "step": 2123
+ },
+ {
+ "epoch": 5.787465940054496,
+ "grad_norm": 6.86889123916626,
+ "learning_rate": 1.9959275999918293e-05,
+ "loss": 1.5488,
+ "step": 2124
+ },
+ {
+ "epoch": 5.790190735694823,
+ "grad_norm": 6.119922161102295,
+ "learning_rate": 1.9959196398406334e-05,
+ "loss": 1.2942,
+ "step": 2125
+ },
+ {
+ "epoch": 5.79291553133515,
+ "grad_norm": 6.622120380401611,
+ "learning_rate": 1.9959116719332468e-05,
+ "loss": 1.3682,
+ "step": 2126
+ },
+ {
+ "epoch": 5.795640326975477,
+ "grad_norm": 6.444990634918213,
+ "learning_rate": 1.9959036962697327e-05,
+ "loss": 1.5864,
+ "step": 2127
+ },
+ {
+ "epoch": 5.798365122615804,
+ "grad_norm": 7.939360618591309,
+ "learning_rate": 1.9958957128501528e-05,
+ "loss": 1.4182,
+ "step": 2128
+ },
+ {
+ "epoch": 5.801089918256131,
+ "grad_norm": 6.958868026733398,
+ "learning_rate": 1.9958877216745685e-05,
+ "loss": 1.4609,
+ "step": 2129
+ },
+ {
+ "epoch": 5.8038147138964575,
+ "grad_norm": 11.643026351928711,
+ "learning_rate": 1.995879722743043e-05,
+ "loss": 1.5981,
+ "step": 2130
+ },
+ {
+ "epoch": 5.806539509536785,
+ "grad_norm": 6.2430877685546875,
+ "learning_rate": 1.9958717160556385e-05,
+ "loss": 1.3325,
+ "step": 2131
+ },
+ {
+ "epoch": 5.809264305177112,
+ "grad_norm": 7.235838413238525,
+ "learning_rate": 1.9958637016124167e-05,
+ "loss": 1.3545,
+ "step": 2132
+ },
+ {
+ "epoch": 5.8119891008174385,
+ "grad_norm": 6.6930975914001465,
+ "learning_rate": 1.9958556794134407e-05,
+ "loss": 1.6221,
+ "step": 2133
+ },
+ {
+ "epoch": 5.814713896457766,
+ "grad_norm": 7.792903900146484,
+ "learning_rate": 1.9958476494587725e-05,
+ "loss": 1.7886,
+ "step": 2134
+ },
+ {
+ "epoch": 5.817438692098093,
+ "grad_norm": 7.610497951507568,
+ "learning_rate": 1.995839611748475e-05,
+ "loss": 1.4221,
+ "step": 2135
+ },
+ {
+ "epoch": 5.820163487738419,
+ "grad_norm": 7.732573986053467,
+ "learning_rate": 1.995831566282611e-05,
+ "loss": 1.4275,
+ "step": 2136
+ },
+ {
+ "epoch": 5.822888283378747,
+ "grad_norm": 7.285153865814209,
+ "learning_rate": 1.9958235130612423e-05,
+ "loss": 1.5498,
+ "step": 2137
+ },
+ {
+ "epoch": 5.825613079019074,
+ "grad_norm": 9.44613265991211,
+ "learning_rate": 1.9958154520844326e-05,
+ "loss": 1.6519,
+ "step": 2138
+ },
+ {
+ "epoch": 5.8283378746594,
+ "grad_norm": 8.652655601501465,
+ "learning_rate": 1.9958073833522437e-05,
+ "loss": 1.5396,
+ "step": 2139
+ },
+ {
+ "epoch": 5.831062670299728,
+ "grad_norm": 6.473405361175537,
+ "learning_rate": 1.9957993068647394e-05,
+ "loss": 1.3296,
+ "step": 2140
+ },
+ {
+ "epoch": 5.833787465940055,
+ "grad_norm": 6.9404072761535645,
+ "learning_rate": 1.9957912226219817e-05,
+ "loss": 1.563,
+ "step": 2141
+ },
+ {
+ "epoch": 5.836512261580381,
+ "grad_norm": 9.054208755493164,
+ "learning_rate": 1.9957831306240343e-05,
+ "loss": 1.6614,
+ "step": 2142
+ },
+ {
+ "epoch": 5.839237057220709,
+ "grad_norm": 6.439208984375,
+ "learning_rate": 1.9957750308709598e-05,
+ "loss": 1.6079,
+ "step": 2143
+ },
+ {
+ "epoch": 5.8419618528610355,
+ "grad_norm": 9.228903770446777,
+ "learning_rate": 1.9957669233628216e-05,
+ "loss": 1.5405,
+ "step": 2144
+ },
+ {
+ "epoch": 5.844686648501362,
+ "grad_norm": 6.279675483703613,
+ "learning_rate": 1.9957588080996825e-05,
+ "loss": 1.5442,
+ "step": 2145
+ },
+ {
+ "epoch": 5.84741144414169,
+ "grad_norm": 6.690636157989502,
+ "learning_rate": 1.995750685081606e-05,
+ "loss": 1.4731,
+ "step": 2146
+ },
+ {
+ "epoch": 5.8501362397820165,
+ "grad_norm": 7.680854797363281,
+ "learning_rate": 1.995742554308655e-05,
+ "loss": 1.5649,
+ "step": 2147
+ },
+ {
+ "epoch": 5.852861035422343,
+ "grad_norm": 7.445345878601074,
+ "learning_rate": 1.995734415780893e-05,
+ "loss": 1.7627,
+ "step": 2148
+ },
+ {
+ "epoch": 5.855585831062671,
+ "grad_norm": 7.148237705230713,
+ "learning_rate": 1.9957262694983837e-05,
+ "loss": 1.5129,
+ "step": 2149
+ },
+ {
+ "epoch": 5.858310626702997,
+ "grad_norm": 7.847738265991211,
+ "learning_rate": 1.9957181154611903e-05,
+ "loss": 1.4824,
+ "step": 2150
+ },
+ {
+ "epoch": 5.861035422343324,
+ "grad_norm": 7.020155429840088,
+ "learning_rate": 1.995709953669376e-05,
+ "loss": 1.3098,
+ "step": 2151
+ },
+ {
+ "epoch": 5.863760217983652,
+ "grad_norm": 7.076805114746094,
+ "learning_rate": 1.9957017841230045e-05,
+ "loss": 1.6147,
+ "step": 2152
+ },
+ {
+ "epoch": 5.866485013623978,
+ "grad_norm": 6.8648905754089355,
+ "learning_rate": 1.9956936068221396e-05,
+ "loss": 1.3206,
+ "step": 2153
+ },
+ {
+ "epoch": 5.869209809264305,
+ "grad_norm": 7.3780012130737305,
+ "learning_rate": 1.995685421766845e-05,
+ "loss": 1.6167,
+ "step": 2154
+ },
+ {
+ "epoch": 5.871934604904633,
+ "grad_norm": 6.327786922454834,
+ "learning_rate": 1.9956772289571845e-05,
+ "loss": 1.3589,
+ "step": 2155
+ },
+ {
+ "epoch": 5.874659400544959,
+ "grad_norm": 6.464724063873291,
+ "learning_rate": 1.995669028393222e-05,
+ "loss": 1.48,
+ "step": 2156
+ },
+ {
+ "epoch": 5.877384196185286,
+ "grad_norm": 7.924178123474121,
+ "learning_rate": 1.995660820075021e-05,
+ "loss": 1.6685,
+ "step": 2157
+ },
+ {
+ "epoch": 5.8801089918256135,
+ "grad_norm": 7.038736820220947,
+ "learning_rate": 1.9956526040026453e-05,
+ "loss": 1.3684,
+ "step": 2158
+ },
+ {
+ "epoch": 5.88283378746594,
+ "grad_norm": 8.212275505065918,
+ "learning_rate": 1.9956443801761593e-05,
+ "loss": 1.3296,
+ "step": 2159
+ },
+ {
+ "epoch": 5.885558583106267,
+ "grad_norm": 7.147834777832031,
+ "learning_rate": 1.995636148595627e-05,
+ "loss": 1.4097,
+ "step": 2160
+ },
+ {
+ "epoch": 5.8882833787465945,
+ "grad_norm": 7.97268009185791,
+ "learning_rate": 1.9956279092611123e-05,
+ "loss": 1.7007,
+ "step": 2161
+ },
+ {
+ "epoch": 5.891008174386921,
+ "grad_norm": 6.550907135009766,
+ "learning_rate": 1.9956196621726794e-05,
+ "loss": 1.4331,
+ "step": 2162
+ },
+ {
+ "epoch": 5.893732970027248,
+ "grad_norm": 6.236309051513672,
+ "learning_rate": 1.995611407330393e-05,
+ "loss": 1.3484,
+ "step": 2163
+ },
+ {
+ "epoch": 5.896457765667575,
+ "grad_norm": 7.387538433074951,
+ "learning_rate": 1.9956031447343166e-05,
+ "loss": 1.4839,
+ "step": 2164
+ },
+ {
+ "epoch": 5.899182561307902,
+ "grad_norm": 7.860008716583252,
+ "learning_rate": 1.9955948743845152e-05,
+ "loss": 1.2744,
+ "step": 2165
+ },
+ {
+ "epoch": 5.901907356948229,
+ "grad_norm": 6.077544212341309,
+ "learning_rate": 1.995586596281053e-05,
+ "loss": 1.3765,
+ "step": 2166
+ },
+ {
+ "epoch": 5.904632152588556,
+ "grad_norm": 6.912436008453369,
+ "learning_rate": 1.9955783104239943e-05,
+ "loss": 1.4634,
+ "step": 2167
+ },
+ {
+ "epoch": 5.907356948228883,
+ "grad_norm": 7.646695613861084,
+ "learning_rate": 1.995570016813404e-05,
+ "loss": 1.4949,
+ "step": 2168
+ },
+ {
+ "epoch": 5.91008174386921,
+ "grad_norm": 6.595940113067627,
+ "learning_rate": 1.9955617154493462e-05,
+ "loss": 1.4551,
+ "step": 2169
+ },
+ {
+ "epoch": 5.912806539509537,
+ "grad_norm": 6.5655341148376465,
+ "learning_rate": 1.995553406331886e-05,
+ "loss": 1.4905,
+ "step": 2170
+ },
+ {
+ "epoch": 5.915531335149864,
+ "grad_norm": 6.112204074859619,
+ "learning_rate": 1.9955450894610875e-05,
+ "loss": 1.3696,
+ "step": 2171
+ },
+ {
+ "epoch": 5.918256130790191,
+ "grad_norm": 5.653378009796143,
+ "learning_rate": 1.995536764837016e-05,
+ "loss": 1.3813,
+ "step": 2172
+ },
+ {
+ "epoch": 5.920980926430518,
+ "grad_norm": 7.4013752937316895,
+ "learning_rate": 1.9955284324597365e-05,
+ "loss": 1.6377,
+ "step": 2173
+ },
+ {
+ "epoch": 5.923705722070845,
+ "grad_norm": 7.518268585205078,
+ "learning_rate": 1.9955200923293134e-05,
+ "loss": 1.3596,
+ "step": 2174
+ },
+ {
+ "epoch": 5.926430517711172,
+ "grad_norm": 7.140881061553955,
+ "learning_rate": 1.9955117444458118e-05,
+ "loss": 1.5474,
+ "step": 2175
+ },
+ {
+ "epoch": 5.929155313351498,
+ "grad_norm": 6.834468364715576,
+ "learning_rate": 1.9955033888092973e-05,
+ "loss": 1.5312,
+ "step": 2176
+ },
+ {
+ "epoch": 5.931880108991826,
+ "grad_norm": 7.536330699920654,
+ "learning_rate": 1.995495025419834e-05,
+ "loss": 1.5154,
+ "step": 2177
+ },
+ {
+ "epoch": 5.9346049046321525,
+ "grad_norm": 12.11955451965332,
+ "learning_rate": 1.9954866542774872e-05,
+ "loss": 1.6335,
+ "step": 2178
+ },
+ {
+ "epoch": 5.937329700272479,
+ "grad_norm": 5.52813196182251,
+ "learning_rate": 1.9954782753823227e-05,
+ "loss": 1.3606,
+ "step": 2179
+ },
+ {
+ "epoch": 5.940054495912807,
+ "grad_norm": 6.139300346374512,
+ "learning_rate": 1.9954698887344057e-05,
+ "loss": 1.4795,
+ "step": 2180
+ },
+ {
+ "epoch": 5.9427792915531334,
+ "grad_norm": 8.384958267211914,
+ "learning_rate": 1.995461494333801e-05,
+ "loss": 1.6245,
+ "step": 2181
+ },
+ {
+ "epoch": 5.94550408719346,
+ "grad_norm": 8.001988410949707,
+ "learning_rate": 1.995453092180574e-05,
+ "loss": 1.4329,
+ "step": 2182
+ },
+ {
+ "epoch": 5.948228882833788,
+ "grad_norm": 8.331911087036133,
+ "learning_rate": 1.9954446822747905e-05,
+ "loss": 1.5049,
+ "step": 2183
+ },
+ {
+ "epoch": 5.950953678474114,
+ "grad_norm": 6.180192470550537,
+ "learning_rate": 1.995436264616516e-05,
+ "loss": 1.5457,
+ "step": 2184
+ },
+ {
+ "epoch": 5.953678474114441,
+ "grad_norm": 6.414090156555176,
+ "learning_rate": 1.995427839205816e-05,
+ "loss": 1.5535,
+ "step": 2185
+ },
+ {
+ "epoch": 5.956403269754769,
+ "grad_norm": 6.469266891479492,
+ "learning_rate": 1.9954194060427557e-05,
+ "loss": 1.7083,
+ "step": 2186
+ },
+ {
+ "epoch": 5.959128065395095,
+ "grad_norm": 6.218597412109375,
+ "learning_rate": 1.9954109651274015e-05,
+ "loss": 1.467,
+ "step": 2187
+ },
+ {
+ "epoch": 5.961852861035422,
+ "grad_norm": 59.32062530517578,
+ "learning_rate": 1.9954025164598184e-05,
+ "loss": 1.4551,
+ "step": 2188
+ },
+ {
+ "epoch": 5.96457765667575,
+ "grad_norm": 6.187668800354004,
+ "learning_rate": 1.9953940600400726e-05,
+ "loss": 1.511,
+ "step": 2189
+ },
+ {
+ "epoch": 5.967302452316076,
+ "grad_norm": 7.052244186401367,
+ "learning_rate": 1.9953855958682297e-05,
+ "loss": 1.4541,
+ "step": 2190
+ },
+ {
+ "epoch": 5.970027247956403,
+ "grad_norm": 6.119859218597412,
+ "learning_rate": 1.9953771239443562e-05,
+ "loss": 1.3833,
+ "step": 2191
+ },
+ {
+ "epoch": 5.9727520435967305,
+ "grad_norm": 8.263998985290527,
+ "learning_rate": 1.9953686442685175e-05,
+ "loss": 1.3889,
+ "step": 2192
+ },
+ {
+ "epoch": 5.975476839237057,
+ "grad_norm": 8.343914985656738,
+ "learning_rate": 1.99536015684078e-05,
+ "loss": 1.436,
+ "step": 2193
+ },
+ {
+ "epoch": 5.978201634877384,
+ "grad_norm": 7.842564105987549,
+ "learning_rate": 1.9953516616612093e-05,
+ "loss": 1.3628,
+ "step": 2194
+ },
+ {
+ "epoch": 5.9809264305177114,
+ "grad_norm": 9.52951431274414,
+ "learning_rate": 1.9953431587298723e-05,
+ "loss": 1.7554,
+ "step": 2195
+ },
+ {
+ "epoch": 5.983651226158038,
+ "grad_norm": 6.144392967224121,
+ "learning_rate": 1.9953346480468347e-05,
+ "loss": 1.5908,
+ "step": 2196
+ },
+ {
+ "epoch": 5.986376021798365,
+ "grad_norm": 7.88716459274292,
+ "learning_rate": 1.9953261296121625e-05,
+ "loss": 1.6904,
+ "step": 2197
+ },
+ {
+ "epoch": 5.989100817438692,
+ "grad_norm": 7.453547477722168,
+ "learning_rate": 1.995317603425923e-05,
+ "loss": 1.6934,
+ "step": 2198
+ },
+ {
+ "epoch": 5.991825613079019,
+ "grad_norm": 9.736735343933105,
+ "learning_rate": 1.9953090694881815e-05,
+ "loss": 1.6877,
+ "step": 2199
+ },
+ {
+ "epoch": 5.994550408719346,
+ "grad_norm": 26.428295135498047,
+ "learning_rate": 1.9953005277990054e-05,
+ "loss": 1.6006,
+ "step": 2200
+ },
+ {
+ "epoch": 5.997275204359673,
+ "grad_norm": 7.595977306365967,
+ "learning_rate": 1.9952919783584608e-05,
+ "loss": 1.6646,
+ "step": 2201
+ },
+ {
+ "epoch": 6.0,
+ "grad_norm": 8.266937255859375,
+ "learning_rate": 1.995283421166614e-05,
+ "loss": 1.4033,
+ "step": 2202
+ },
+ {
+ "epoch": 6.002724795640327,
+ "grad_norm": 8.208110809326172,
+ "learning_rate": 1.9952748562235325e-05,
+ "loss": 1.2881,
+ "step": 2203
+ },
+ {
+ "epoch": 6.005449591280654,
+ "grad_norm": 8.129433631896973,
+ "learning_rate": 1.995266283529282e-05,
+ "loss": 1.4575,
+ "step": 2204
+ },
+ {
+ "epoch": 6.008174386920981,
+ "grad_norm": 9.764735221862793,
+ "learning_rate": 1.9952577030839297e-05,
+ "loss": 1.2126,
+ "step": 2205
+ },
+ {
+ "epoch": 6.010899182561308,
+ "grad_norm": 10.686315536499023,
+ "learning_rate": 1.9952491148875427e-05,
+ "loss": 1.3074,
+ "step": 2206
+ },
+ {
+ "epoch": 6.013623978201635,
+ "grad_norm": 8.018460273742676,
+ "learning_rate": 1.9952405189401872e-05,
+ "loss": 1.4231,
+ "step": 2207
+ },
+ {
+ "epoch": 6.016348773841962,
+ "grad_norm": 7.24132776260376,
+ "learning_rate": 1.9952319152419314e-05,
+ "loss": 1.396,
+ "step": 2208
+ },
+ {
+ "epoch": 6.0190735694822886,
+ "grad_norm": 9.536149024963379,
+ "learning_rate": 1.9952233037928406e-05,
+ "loss": 1.4585,
+ "step": 2209
+ },
+ {
+ "epoch": 6.021798365122616,
+ "grad_norm": 17.634485244750977,
+ "learning_rate": 1.9952146845929833e-05,
+ "loss": 1.3113,
+ "step": 2210
+ },
+ {
+ "epoch": 6.024523160762943,
+ "grad_norm": 12.354101181030273,
+ "learning_rate": 1.9952060576424256e-05,
+ "loss": 1.4004,
+ "step": 2211
+ },
+ {
+ "epoch": 6.0272479564032695,
+ "grad_norm": 9.01684284210205,
+ "learning_rate": 1.9951974229412355e-05,
+ "loss": 1.2961,
+ "step": 2212
+ },
+ {
+ "epoch": 6.029972752043597,
+ "grad_norm": 6.916833877563477,
+ "learning_rate": 1.99518878048948e-05,
+ "loss": 1.2466,
+ "step": 2213
+ },
+ {
+ "epoch": 6.032697547683924,
+ "grad_norm": 7.172754764556885,
+ "learning_rate": 1.995180130287226e-05,
+ "loss": 1.5039,
+ "step": 2214
+ },
+ {
+ "epoch": 6.03542234332425,
+ "grad_norm": 7.271836280822754,
+ "learning_rate": 1.995171472334541e-05,
+ "loss": 1.551,
+ "step": 2215
+ },
+ {
+ "epoch": 6.038147138964578,
+ "grad_norm": 12.699302673339844,
+ "learning_rate": 1.9951628066314928e-05,
+ "loss": 1.2898,
+ "step": 2216
+ },
+ {
+ "epoch": 6.040871934604905,
+ "grad_norm": 6.869850158691406,
+ "learning_rate": 1.995154133178149e-05,
+ "loss": 1.49,
+ "step": 2217
+ },
+ {
+ "epoch": 6.043596730245231,
+ "grad_norm": 9.432917594909668,
+ "learning_rate": 1.9951454519745767e-05,
+ "loss": 1.2981,
+ "step": 2218
+ },
+ {
+ "epoch": 6.046321525885559,
+ "grad_norm": 7.822467803955078,
+ "learning_rate": 1.9951367630208437e-05,
+ "loss": 1.3733,
+ "step": 2219
+ },
+ {
+ "epoch": 6.049046321525886,
+ "grad_norm": 7.4163923263549805,
+ "learning_rate": 1.9951280663170174e-05,
+ "loss": 1.3428,
+ "step": 2220
+ },
+ {
+ "epoch": 6.051771117166212,
+ "grad_norm": 10.42732048034668,
+ "learning_rate": 1.9951193618631655e-05,
+ "loss": 1.3306,
+ "step": 2221
+ },
+ {
+ "epoch": 6.05449591280654,
+ "grad_norm": 12.341014862060547,
+ "learning_rate": 1.9951106496593566e-05,
+ "loss": 1.5347,
+ "step": 2222
+ },
+ {
+ "epoch": 6.0572207084468666,
+ "grad_norm": 7.879974842071533,
+ "learning_rate": 1.995101929705658e-05,
+ "loss": 1.459,
+ "step": 2223
+ },
+ {
+ "epoch": 6.059945504087193,
+ "grad_norm": 12.8678560256958,
+ "learning_rate": 1.9950932020021372e-05,
+ "loss": 1.561,
+ "step": 2224
+ },
+ {
+ "epoch": 6.062670299727521,
+ "grad_norm": 11.594139099121094,
+ "learning_rate": 1.9950844665488626e-05,
+ "loss": 1.6082,
+ "step": 2225
+ },
+ {
+ "epoch": 6.0653950953678475,
+ "grad_norm": 9.652727127075195,
+ "learning_rate": 1.9950757233459022e-05,
+ "loss": 1.4487,
+ "step": 2226
+ },
+ {
+ "epoch": 6.068119891008174,
+ "grad_norm": 8.242108345031738,
+ "learning_rate": 1.9950669723933246e-05,
+ "loss": 1.4021,
+ "step": 2227
+ },
+ {
+ "epoch": 6.070844686648502,
+ "grad_norm": 8.821127891540527,
+ "learning_rate": 1.9950582136911972e-05,
+ "loss": 1.7139,
+ "step": 2228
+ },
+ {
+ "epoch": 6.073569482288828,
+ "grad_norm": 16.895620346069336,
+ "learning_rate": 1.995049447239588e-05,
+ "loss": 1.2474,
+ "step": 2229
+ },
+ {
+ "epoch": 6.076294277929155,
+ "grad_norm": 9.042691230773926,
+ "learning_rate": 1.9950406730385663e-05,
+ "loss": 1.5156,
+ "step": 2230
+ },
+ {
+ "epoch": 6.079019073569483,
+ "grad_norm": 9.889644622802734,
+ "learning_rate": 1.9950318910881993e-05,
+ "loss": 1.3511,
+ "step": 2231
+ },
+ {
+ "epoch": 6.081743869209809,
+ "grad_norm": 7.158227443695068,
+ "learning_rate": 1.9950231013885566e-05,
+ "loss": 1.415,
+ "step": 2232
+ },
+ {
+ "epoch": 6.084468664850136,
+ "grad_norm": 7.619608402252197,
+ "learning_rate": 1.9950143039397056e-05,
+ "loss": 1.3857,
+ "step": 2233
+ },
+ {
+ "epoch": 6.087193460490464,
+ "grad_norm": 7.267209529876709,
+ "learning_rate": 1.995005498741715e-05,
+ "loss": 1.1912,
+ "step": 2234
+ },
+ {
+ "epoch": 6.08991825613079,
+ "grad_norm": 7.907135009765625,
+ "learning_rate": 1.994996685794654e-05,
+ "loss": 1.4131,
+ "step": 2235
+ },
+ {
+ "epoch": 6.092643051771117,
+ "grad_norm": 9.729490280151367,
+ "learning_rate": 1.994987865098591e-05,
+ "loss": 1.4048,
+ "step": 2236
+ },
+ {
+ "epoch": 6.0953678474114446,
+ "grad_norm": 9.517294883728027,
+ "learning_rate": 1.9949790366535943e-05,
+ "loss": 1.3472,
+ "step": 2237
+ },
+ {
+ "epoch": 6.098092643051771,
+ "grad_norm": 10.050874710083008,
+ "learning_rate": 1.9949702004597327e-05,
+ "loss": 1.3152,
+ "step": 2238
+ },
+ {
+ "epoch": 6.100817438692098,
+ "grad_norm": 9.48713207244873,
+ "learning_rate": 1.9949613565170755e-05,
+ "loss": 1.4072,
+ "step": 2239
+ },
+ {
+ "epoch": 6.1035422343324255,
+ "grad_norm": 7.30112886428833,
+ "learning_rate": 1.994952504825691e-05,
+ "loss": 1.2908,
+ "step": 2240
+ },
+ {
+ "epoch": 6.106267029972752,
+ "grad_norm": 8.594221115112305,
+ "learning_rate": 1.9949436453856487e-05,
+ "loss": 1.5215,
+ "step": 2241
+ },
+ {
+ "epoch": 6.108991825613079,
+ "grad_norm": 10.833154678344727,
+ "learning_rate": 1.9949347781970172e-05,
+ "loss": 1.6592,
+ "step": 2242
+ },
+ {
+ "epoch": 6.111716621253406,
+ "grad_norm": 6.648782253265381,
+ "learning_rate": 1.9949259032598656e-05,
+ "loss": 1.1801,
+ "step": 2243
+ },
+ {
+ "epoch": 6.114441416893733,
+ "grad_norm": 7.6475958824157715,
+ "learning_rate": 1.994917020574263e-05,
+ "loss": 1.3877,
+ "step": 2244
+ },
+ {
+ "epoch": 6.11716621253406,
+ "grad_norm": 7.958492755889893,
+ "learning_rate": 1.994908130140279e-05,
+ "loss": 1.2544,
+ "step": 2245
+ },
+ {
+ "epoch": 6.1198910081743865,
+ "grad_norm": 6.293396472930908,
+ "learning_rate": 1.9948992319579822e-05,
+ "loss": 1.2732,
+ "step": 2246
+ },
+ {
+ "epoch": 6.122615803814714,
+ "grad_norm": 7.287437438964844,
+ "learning_rate": 1.9948903260274424e-05,
+ "loss": 1.2732,
+ "step": 2247
+ },
+ {
+ "epoch": 6.125340599455041,
+ "grad_norm": 7.425198078155518,
+ "learning_rate": 1.9948814123487286e-05,
+ "loss": 1.509,
+ "step": 2248
+ },
+ {
+ "epoch": 6.128065395095367,
+ "grad_norm": 16.254817962646484,
+ "learning_rate": 1.9948724909219104e-05,
+ "loss": 1.6379,
+ "step": 2249
+ },
+ {
+ "epoch": 6.130790190735695,
+ "grad_norm": 8.264799118041992,
+ "learning_rate": 1.9948635617470572e-05,
+ "loss": 1.1494,
+ "step": 2250
+ },
+ {
+ "epoch": 6.133514986376022,
+ "grad_norm": 9.320589065551758,
+ "learning_rate": 1.994854624824239e-05,
+ "loss": 1.522,
+ "step": 2251
+ },
+ {
+ "epoch": 6.136239782016348,
+ "grad_norm": 7.1113057136535645,
+ "learning_rate": 1.9948456801535247e-05,
+ "loss": 1.3127,
+ "step": 2252
+ },
+ {
+ "epoch": 6.138964577656676,
+ "grad_norm": 10.387081146240234,
+ "learning_rate": 1.994836727734984e-05,
+ "loss": 1.511,
+ "step": 2253
+ },
+ {
+ "epoch": 6.141689373297003,
+ "grad_norm": 8.192678451538086,
+ "learning_rate": 1.9948277675686873e-05,
+ "loss": 1.4673,
+ "step": 2254
+ },
+ {
+ "epoch": 6.144414168937329,
+ "grad_norm": 6.649348258972168,
+ "learning_rate": 1.9948187996547037e-05,
+ "loss": 1.5508,
+ "step": 2255
+ },
+ {
+ "epoch": 6.147138964577657,
+ "grad_norm": 15.370036125183105,
+ "learning_rate": 1.9948098239931035e-05,
+ "loss": 1.5483,
+ "step": 2256
+ },
+ {
+ "epoch": 6.1498637602179835,
+ "grad_norm": 25.892786026000977,
+ "learning_rate": 1.994800840583956e-05,
+ "loss": 1.447,
+ "step": 2257
+ },
+ {
+ "epoch": 6.15258855585831,
+ "grad_norm": 9.651581764221191,
+ "learning_rate": 1.9947918494273322e-05,
+ "loss": 1.397,
+ "step": 2258
+ },
+ {
+ "epoch": 6.155313351498638,
+ "grad_norm": 8.368578910827637,
+ "learning_rate": 1.994782850523301e-05,
+ "loss": 1.4485,
+ "step": 2259
+ },
+ {
+ "epoch": 6.1580381471389645,
+ "grad_norm": 7.645608425140381,
+ "learning_rate": 1.9947738438719332e-05,
+ "loss": 1.5557,
+ "step": 2260
+ },
+ {
+ "epoch": 6.160762942779291,
+ "grad_norm": 14.208962440490723,
+ "learning_rate": 1.9947648294732988e-05,
+ "loss": 1.6394,
+ "step": 2261
+ },
+ {
+ "epoch": 6.163487738419619,
+ "grad_norm": 9.030295372009277,
+ "learning_rate": 1.994755807327468e-05,
+ "loss": 1.4233,
+ "step": 2262
+ },
+ {
+ "epoch": 6.166212534059945,
+ "grad_norm": 8.869603157043457,
+ "learning_rate": 1.9947467774345108e-05,
+ "loss": 1.3494,
+ "step": 2263
+ },
+ {
+ "epoch": 6.168937329700272,
+ "grad_norm": 9.084758758544922,
+ "learning_rate": 1.9947377397944974e-05,
+ "loss": 1.3647,
+ "step": 2264
+ },
+ {
+ "epoch": 6.1716621253406,
+ "grad_norm": 7.110030174255371,
+ "learning_rate": 1.994728694407499e-05,
+ "loss": 1.2905,
+ "step": 2265
+ },
+ {
+ "epoch": 6.174386920980926,
+ "grad_norm": 10.713805198669434,
+ "learning_rate": 1.9947196412735855e-05,
+ "loss": 1.8457,
+ "step": 2266
+ },
+ {
+ "epoch": 6.177111716621253,
+ "grad_norm": 7.273036479949951,
+ "learning_rate": 1.9947105803928272e-05,
+ "loss": 1.3535,
+ "step": 2267
+ },
+ {
+ "epoch": 6.179836512261581,
+ "grad_norm": 7.770942211151123,
+ "learning_rate": 1.994701511765295e-05,
+ "loss": 1.2278,
+ "step": 2268
+ },
+ {
+ "epoch": 6.182561307901907,
+ "grad_norm": 7.533062934875488,
+ "learning_rate": 1.9946924353910595e-05,
+ "loss": 1.4065,
+ "step": 2269
+ },
+ {
+ "epoch": 6.185286103542234,
+ "grad_norm": 8.072220802307129,
+ "learning_rate": 1.994683351270191e-05,
+ "loss": 1.4268,
+ "step": 2270
+ },
+ {
+ "epoch": 6.1880108991825615,
+ "grad_norm": 8.256223678588867,
+ "learning_rate": 1.994674259402761e-05,
+ "loss": 1.5955,
+ "step": 2271
+ },
+ {
+ "epoch": 6.190735694822888,
+ "grad_norm": 8.5989990234375,
+ "learning_rate": 1.99466515978884e-05,
+ "loss": 1.3276,
+ "step": 2272
+ },
+ {
+ "epoch": 6.193460490463215,
+ "grad_norm": 8.990948677062988,
+ "learning_rate": 1.9946560524284985e-05,
+ "loss": 1.5356,
+ "step": 2273
+ },
+ {
+ "epoch": 6.1961852861035425,
+ "grad_norm": 8.66739559173584,
+ "learning_rate": 1.9946469373218076e-05,
+ "loss": 1.3496,
+ "step": 2274
+ },
+ {
+ "epoch": 6.198910081743869,
+ "grad_norm": 7.030218601226807,
+ "learning_rate": 1.994637814468838e-05,
+ "loss": 1.4331,
+ "step": 2275
+ },
+ {
+ "epoch": 6.201634877384196,
+ "grad_norm": 7.883135795593262,
+ "learning_rate": 1.994628683869662e-05,
+ "loss": 1.3325,
+ "step": 2276
+ },
+ {
+ "epoch": 6.204359673024523,
+ "grad_norm": 7.4221720695495605,
+ "learning_rate": 1.9946195455243494e-05,
+ "loss": 1.2993,
+ "step": 2277
+ },
+ {
+ "epoch": 6.20708446866485,
+ "grad_norm": 11.761012077331543,
+ "learning_rate": 1.9946103994329714e-05,
+ "loss": 1.5208,
+ "step": 2278
+ },
+ {
+ "epoch": 6.209809264305177,
+ "grad_norm": 11.263177871704102,
+ "learning_rate": 1.9946012455956004e-05,
+ "loss": 1.5994,
+ "step": 2279
+ },
+ {
+ "epoch": 6.212534059945504,
+ "grad_norm": 6.860296726226807,
+ "learning_rate": 1.9945920840123062e-05,
+ "loss": 1.3588,
+ "step": 2280
+ },
+ {
+ "epoch": 6.215258855585831,
+ "grad_norm": 6.807911396026611,
+ "learning_rate": 1.9945829146831614e-05,
+ "loss": 1.2678,
+ "step": 2281
+ },
+ {
+ "epoch": 6.217983651226158,
+ "grad_norm": 8.724958419799805,
+ "learning_rate": 1.9945737376082363e-05,
+ "loss": 1.5256,
+ "step": 2282
+ },
+ {
+ "epoch": 6.220708446866485,
+ "grad_norm": 9.21982192993164,
+ "learning_rate": 1.9945645527876033e-05,
+ "loss": 1.4141,
+ "step": 2283
+ },
+ {
+ "epoch": 6.223433242506812,
+ "grad_norm": 18.68528938293457,
+ "learning_rate": 1.9945553602213336e-05,
+ "loss": 1.436,
+ "step": 2284
+ },
+ {
+ "epoch": 6.226158038147139,
+ "grad_norm": 6.97844123840332,
+ "learning_rate": 1.994546159909499e-05,
+ "loss": 1.179,
+ "step": 2285
+ },
+ {
+ "epoch": 6.228882833787466,
+ "grad_norm": 7.6594953536987305,
+ "learning_rate": 1.9945369518521704e-05,
+ "loss": 1.6167,
+ "step": 2286
+ },
+ {
+ "epoch": 6.231607629427793,
+ "grad_norm": 7.278682708740234,
+ "learning_rate": 1.9945277360494204e-05,
+ "loss": 1.3269,
+ "step": 2287
+ },
+ {
+ "epoch": 6.23433242506812,
+ "grad_norm": 7.76326847076416,
+ "learning_rate": 1.99451851250132e-05,
+ "loss": 1.6023,
+ "step": 2288
+ },
+ {
+ "epoch": 6.237057220708447,
+ "grad_norm": 7.965860366821289,
+ "learning_rate": 1.994509281207942e-05,
+ "loss": 1.519,
+ "step": 2289
+ },
+ {
+ "epoch": 6.239782016348774,
+ "grad_norm": 6.67733097076416,
+ "learning_rate": 1.9945000421693574e-05,
+ "loss": 1.4731,
+ "step": 2290
+ },
+ {
+ "epoch": 6.2425068119891005,
+ "grad_norm": 6.553910732269287,
+ "learning_rate": 1.9944907953856384e-05,
+ "loss": 1.3484,
+ "step": 2291
+ },
+ {
+ "epoch": 6.245231607629428,
+ "grad_norm": 7.167792797088623,
+ "learning_rate": 1.9944815408568572e-05,
+ "loss": 1.3789,
+ "step": 2292
+ },
+ {
+ "epoch": 6.247956403269755,
+ "grad_norm": 41.22723388671875,
+ "learning_rate": 1.9944722785830855e-05,
+ "loss": 1.5237,
+ "step": 2293
+ },
+ {
+ "epoch": 6.2506811989100814,
+ "grad_norm": 8.987257957458496,
+ "learning_rate": 1.994463008564396e-05,
+ "loss": 1.4153,
+ "step": 2294
+ },
+ {
+ "epoch": 6.253405994550409,
+ "grad_norm": 7.701432228088379,
+ "learning_rate": 1.9944537308008606e-05,
+ "loss": 1.2646,
+ "step": 2295
+ },
+ {
+ "epoch": 6.256130790190736,
+ "grad_norm": 6.967992782592773,
+ "learning_rate": 1.9944444452925514e-05,
+ "loss": 1.2646,
+ "step": 2296
+ },
+ {
+ "epoch": 6.258855585831062,
+ "grad_norm": 14.145384788513184,
+ "learning_rate": 1.994435152039541e-05,
+ "loss": 1.4822,
+ "step": 2297
+ },
+ {
+ "epoch": 6.26158038147139,
+ "grad_norm": 7.09717321395874,
+ "learning_rate": 1.9944258510419018e-05,
+ "loss": 1.4299,
+ "step": 2298
+ },
+ {
+ "epoch": 6.264305177111717,
+ "grad_norm": 9.883954048156738,
+ "learning_rate": 1.9944165422997058e-05,
+ "loss": 1.4988,
+ "step": 2299
+ },
+ {
+ "epoch": 6.267029972752043,
+ "grad_norm": 7.450252056121826,
+ "learning_rate": 1.994407225813026e-05,
+ "loss": 1.3799,
+ "step": 2300
+ },
+ {
+ "epoch": 6.269754768392371,
+ "grad_norm": 10.96161937713623,
+ "learning_rate": 1.9943979015819343e-05,
+ "loss": 1.4767,
+ "step": 2301
+ },
+ {
+ "epoch": 6.272479564032698,
+ "grad_norm": 10.635453224182129,
+ "learning_rate": 1.994388569606504e-05,
+ "loss": 1.4475,
+ "step": 2302
+ },
+ {
+ "epoch": 6.275204359673024,
+ "grad_norm": 8.448601722717285,
+ "learning_rate": 1.9943792298868076e-05,
+ "loss": 1.5522,
+ "step": 2303
+ },
+ {
+ "epoch": 6.277929155313352,
+ "grad_norm": 10.190055847167969,
+ "learning_rate": 1.994369882422918e-05,
+ "loss": 1.5234,
+ "step": 2304
+ },
+ {
+ "epoch": 6.2806539509536785,
+ "grad_norm": 6.170011043548584,
+ "learning_rate": 1.9943605272149074e-05,
+ "loss": 1.2205,
+ "step": 2305
+ },
+ {
+ "epoch": 6.283378746594005,
+ "grad_norm": 7.211528778076172,
+ "learning_rate": 1.9943511642628488e-05,
+ "loss": 1.1318,
+ "step": 2306
+ },
+ {
+ "epoch": 6.286103542234333,
+ "grad_norm": 19.993593215942383,
+ "learning_rate": 1.9943417935668158e-05,
+ "loss": 1.5166,
+ "step": 2307
+ },
+ {
+ "epoch": 6.2888283378746594,
+ "grad_norm": 8.65777587890625,
+ "learning_rate": 1.9943324151268807e-05,
+ "loss": 1.5149,
+ "step": 2308
+ },
+ {
+ "epoch": 6.291553133514986,
+ "grad_norm": 8.570378303527832,
+ "learning_rate": 1.994323028943117e-05,
+ "loss": 1.4392,
+ "step": 2309
+ },
+ {
+ "epoch": 6.294277929155314,
+ "grad_norm": 8.419354438781738,
+ "learning_rate": 1.994313635015597e-05,
+ "loss": 1.5146,
+ "step": 2310
+ },
+ {
+ "epoch": 6.29700272479564,
+ "grad_norm": 8.05183219909668,
+ "learning_rate": 1.994304233344395e-05,
+ "loss": 1.4653,
+ "step": 2311
+ },
+ {
+ "epoch": 6.299727520435967,
+ "grad_norm": 7.633604049682617,
+ "learning_rate": 1.9942948239295833e-05,
+ "loss": 1.4565,
+ "step": 2312
+ },
+ {
+ "epoch": 6.302452316076295,
+ "grad_norm": 10.049834251403809,
+ "learning_rate": 1.9942854067712355e-05,
+ "loss": 1.2456,
+ "step": 2313
+ },
+ {
+ "epoch": 6.305177111716621,
+ "grad_norm": 7.7690653800964355,
+ "learning_rate": 1.994275981869425e-05,
+ "loss": 1.4863,
+ "step": 2314
+ },
+ {
+ "epoch": 6.307901907356948,
+ "grad_norm": 8.550948143005371,
+ "learning_rate": 1.9942665492242256e-05,
+ "loss": 1.4524,
+ "step": 2315
+ },
+ {
+ "epoch": 6.310626702997276,
+ "grad_norm": 8.089266777038574,
+ "learning_rate": 1.9942571088357095e-05,
+ "loss": 1.3684,
+ "step": 2316
+ },
+ {
+ "epoch": 6.313351498637602,
+ "grad_norm": 9.125869750976562,
+ "learning_rate": 1.994247660703952e-05,
+ "loss": 1.3267,
+ "step": 2317
+ },
+ {
+ "epoch": 6.316076294277929,
+ "grad_norm": 17.101762771606445,
+ "learning_rate": 1.9942382048290248e-05,
+ "loss": 1.5366,
+ "step": 2318
+ },
+ {
+ "epoch": 6.3188010899182565,
+ "grad_norm": 7.853949546813965,
+ "learning_rate": 1.994228741211003e-05,
+ "loss": 1.2781,
+ "step": 2319
+ },
+ {
+ "epoch": 6.321525885558583,
+ "grad_norm": 8.73260498046875,
+ "learning_rate": 1.9942192698499596e-05,
+ "loss": 1.6145,
+ "step": 2320
+ },
+ {
+ "epoch": 6.32425068119891,
+ "grad_norm": 8.895113945007324,
+ "learning_rate": 1.9942097907459687e-05,
+ "loss": 1.4651,
+ "step": 2321
+ },
+ {
+ "epoch": 6.3269754768392374,
+ "grad_norm": 18.743423461914062,
+ "learning_rate": 1.9942003038991035e-05,
+ "loss": 1.4757,
+ "step": 2322
+ },
+ {
+ "epoch": 6.329700272479564,
+ "grad_norm": 8.328478813171387,
+ "learning_rate": 1.9941908093094387e-05,
+ "loss": 1.2729,
+ "step": 2323
+ },
+ {
+ "epoch": 6.332425068119891,
+ "grad_norm": 7.33073091506958,
+ "learning_rate": 1.9941813069770478e-05,
+ "loss": 1.1643,
+ "step": 2324
+ },
+ {
+ "epoch": 6.335149863760218,
+ "grad_norm": 7.987620830535889,
+ "learning_rate": 1.9941717969020046e-05,
+ "loss": 1.5857,
+ "step": 2325
+ },
+ {
+ "epoch": 6.337874659400545,
+ "grad_norm": 9.618622779846191,
+ "learning_rate": 1.994162279084384e-05,
+ "loss": 1.3596,
+ "step": 2326
+ },
+ {
+ "epoch": 6.340599455040872,
+ "grad_norm": 10.125624656677246,
+ "learning_rate": 1.9941527535242592e-05,
+ "loss": 1.6409,
+ "step": 2327
+ },
+ {
+ "epoch": 6.343324250681199,
+ "grad_norm": 19.003952026367188,
+ "learning_rate": 1.9941432202217047e-05,
+ "loss": 1.4646,
+ "step": 2328
+ },
+ {
+ "epoch": 6.346049046321526,
+ "grad_norm": 9.63425064086914,
+ "learning_rate": 1.994133679176795e-05,
+ "loss": 1.3555,
+ "step": 2329
+ },
+ {
+ "epoch": 6.348773841961853,
+ "grad_norm": 7.4560346603393555,
+ "learning_rate": 1.994124130389604e-05,
+ "loss": 1.4397,
+ "step": 2330
+ },
+ {
+ "epoch": 6.35149863760218,
+ "grad_norm": 8.627058029174805,
+ "learning_rate": 1.9941145738602065e-05,
+ "loss": 1.2212,
+ "step": 2331
+ },
+ {
+ "epoch": 6.354223433242507,
+ "grad_norm": 8.751408576965332,
+ "learning_rate": 1.9941050095886763e-05,
+ "loss": 1.5532,
+ "step": 2332
+ },
+ {
+ "epoch": 6.356948228882834,
+ "grad_norm": 11.239310264587402,
+ "learning_rate": 1.9940954375750888e-05,
+ "loss": 1.5029,
+ "step": 2333
+ },
+ {
+ "epoch": 6.359673024523161,
+ "grad_norm": 9.95463752746582,
+ "learning_rate": 1.994085857819518e-05,
+ "loss": 1.5583,
+ "step": 2334
+ },
+ {
+ "epoch": 6.362397820163488,
+ "grad_norm": 7.851717948913574,
+ "learning_rate": 1.9940762703220383e-05,
+ "loss": 1.2769,
+ "step": 2335
+ },
+ {
+ "epoch": 6.3651226158038146,
+ "grad_norm": 7.805356502532959,
+ "learning_rate": 1.9940666750827247e-05,
+ "loss": 1.6638,
+ "step": 2336
+ },
+ {
+ "epoch": 6.367847411444142,
+ "grad_norm": 9.646978378295898,
+ "learning_rate": 1.9940570721016516e-05,
+ "loss": 1.2905,
+ "step": 2337
+ },
+ {
+ "epoch": 6.370572207084469,
+ "grad_norm": 7.88627815246582,
+ "learning_rate": 1.994047461378894e-05,
+ "loss": 1.3328,
+ "step": 2338
+ },
+ {
+ "epoch": 6.3732970027247955,
+ "grad_norm": 8.983908653259277,
+ "learning_rate": 1.994037842914527e-05,
+ "loss": 1.4492,
+ "step": 2339
+ },
+ {
+ "epoch": 6.376021798365123,
+ "grad_norm": 11.249740600585938,
+ "learning_rate": 1.9940282167086255e-05,
+ "loss": 1.3718,
+ "step": 2340
+ },
+ {
+ "epoch": 6.37874659400545,
+ "grad_norm": 7.740970134735107,
+ "learning_rate": 1.9940185827612638e-05,
+ "loss": 1.4724,
+ "step": 2341
+ },
+ {
+ "epoch": 6.381471389645776,
+ "grad_norm": 7.317564010620117,
+ "learning_rate": 1.9940089410725174e-05,
+ "loss": 1.3342,
+ "step": 2342
+ },
+ {
+ "epoch": 6.384196185286104,
+ "grad_norm": 8.715740203857422,
+ "learning_rate": 1.9939992916424614e-05,
+ "loss": 1.2908,
+ "step": 2343
+ },
+ {
+ "epoch": 6.386920980926431,
+ "grad_norm": 23.59552764892578,
+ "learning_rate": 1.9939896344711714e-05,
+ "loss": 1.4153,
+ "step": 2344
+ },
+ {
+ "epoch": 6.389645776566757,
+ "grad_norm": 7.656932830810547,
+ "learning_rate": 1.9939799695587213e-05,
+ "loss": 1.3672,
+ "step": 2345
+ },
+ {
+ "epoch": 6.392370572207085,
+ "grad_norm": 8.092866897583008,
+ "learning_rate": 1.9939702969051876e-05,
+ "loss": 1.3208,
+ "step": 2346
+ },
+ {
+ "epoch": 6.395095367847412,
+ "grad_norm": 9.64970588684082,
+ "learning_rate": 1.993960616510645e-05,
+ "loss": 1.427,
+ "step": 2347
+ },
+ {
+ "epoch": 6.397820163487738,
+ "grad_norm": 8.243301391601562,
+ "learning_rate": 1.9939509283751696e-05,
+ "loss": 1.208,
+ "step": 2348
+ },
+ {
+ "epoch": 6.400544959128065,
+ "grad_norm": 10.50653076171875,
+ "learning_rate": 1.993941232498836e-05,
+ "loss": 1.2107,
+ "step": 2349
+ },
+ {
+ "epoch": 6.4032697547683926,
+ "grad_norm": 11.515603065490723,
+ "learning_rate": 1.9939315288817203e-05,
+ "loss": 1.5923,
+ "step": 2350
+ },
+ {
+ "epoch": 6.405994550408719,
+ "grad_norm": 7.8392815589904785,
+ "learning_rate": 1.993921817523898e-05,
+ "loss": 1.3711,
+ "step": 2351
+ },
+ {
+ "epoch": 6.408719346049046,
+ "grad_norm": 7.378281116485596,
+ "learning_rate": 1.993912098425444e-05,
+ "loss": 1.1779,
+ "step": 2352
+ },
+ {
+ "epoch": 6.4114441416893735,
+ "grad_norm": 8.652050971984863,
+ "learning_rate": 1.993902371586435e-05,
+ "loss": 1.4629,
+ "step": 2353
+ },
+ {
+ "epoch": 6.4141689373297,
+ "grad_norm": 8.526771545410156,
+ "learning_rate": 1.993892637006946e-05,
+ "loss": 1.5588,
+ "step": 2354
+ },
+ {
+ "epoch": 6.416893732970027,
+ "grad_norm": 8.640652656555176,
+ "learning_rate": 1.993882894687053e-05,
+ "loss": 1.5608,
+ "step": 2355
+ },
+ {
+ "epoch": 6.419618528610354,
+ "grad_norm": 10.2577486038208,
+ "learning_rate": 1.993873144626833e-05,
+ "loss": 1.3018,
+ "step": 2356
+ },
+ {
+ "epoch": 6.422343324250681,
+ "grad_norm": 8.789144515991211,
+ "learning_rate": 1.9938633868263597e-05,
+ "loss": 1.4634,
+ "step": 2357
+ },
+ {
+ "epoch": 6.425068119891008,
+ "grad_norm": 7.703369140625,
+ "learning_rate": 1.993853621285711e-05,
+ "loss": 1.1882,
+ "step": 2358
+ },
+ {
+ "epoch": 6.427792915531335,
+ "grad_norm": 8.806507110595703,
+ "learning_rate": 1.993843848004962e-05,
+ "loss": 1.4905,
+ "step": 2359
+ },
+ {
+ "epoch": 6.430517711171662,
+ "grad_norm": 21.855043411254883,
+ "learning_rate": 1.9938340669841893e-05,
+ "loss": 1.3977,
+ "step": 2360
+ },
+ {
+ "epoch": 6.433242506811989,
+ "grad_norm": 7.3547139167785645,
+ "learning_rate": 1.993824278223469e-05,
+ "loss": 1.5046,
+ "step": 2361
+ },
+ {
+ "epoch": 6.435967302452316,
+ "grad_norm": 8.148509979248047,
+ "learning_rate": 1.993814481722877e-05,
+ "loss": 1.511,
+ "step": 2362
+ },
+ {
+ "epoch": 6.438692098092643,
+ "grad_norm": 6.897467613220215,
+ "learning_rate": 1.9938046774824897e-05,
+ "loss": 1.4138,
+ "step": 2363
+ },
+ {
+ "epoch": 6.44141689373297,
+ "grad_norm": 9.47750473022461,
+ "learning_rate": 1.9937948655023838e-05,
+ "loss": 1.5466,
+ "step": 2364
+ },
+ {
+ "epoch": 6.444141689373297,
+ "grad_norm": 8.061921119689941,
+ "learning_rate": 1.9937850457826354e-05,
+ "loss": 1.3315,
+ "step": 2365
+ },
+ {
+ "epoch": 6.446866485013624,
+ "grad_norm": 8.671846389770508,
+ "learning_rate": 1.993775218323321e-05,
+ "loss": 1.4353,
+ "step": 2366
+ },
+ {
+ "epoch": 6.449591280653951,
+ "grad_norm": 8.110825538635254,
+ "learning_rate": 1.993765383124517e-05,
+ "loss": 1.5913,
+ "step": 2367
+ },
+ {
+ "epoch": 6.452316076294278,
+ "grad_norm": 7.873919486999512,
+ "learning_rate": 1.9937555401863005e-05,
+ "loss": 1.373,
+ "step": 2368
+ },
+ {
+ "epoch": 6.455040871934605,
+ "grad_norm": 8.330255508422852,
+ "learning_rate": 1.9937456895087477e-05,
+ "loss": 1.4163,
+ "step": 2369
+ },
+ {
+ "epoch": 6.4577656675749315,
+ "grad_norm": 13.39629077911377,
+ "learning_rate": 1.9937358310919356e-05,
+ "loss": 1.4807,
+ "step": 2370
+ },
+ {
+ "epoch": 6.460490463215259,
+ "grad_norm": 8.64048957824707,
+ "learning_rate": 1.9937259649359406e-05,
+ "loss": 1.4216,
+ "step": 2371
+ },
+ {
+ "epoch": 6.463215258855586,
+ "grad_norm": 7.050153732299805,
+ "learning_rate": 1.9937160910408396e-05,
+ "loss": 1.6523,
+ "step": 2372
+ },
+ {
+ "epoch": 6.4659400544959125,
+ "grad_norm": 10.897089004516602,
+ "learning_rate": 1.99370620940671e-05,
+ "loss": 1.2563,
+ "step": 2373
+ },
+ {
+ "epoch": 6.46866485013624,
+ "grad_norm": 5.598983287811279,
+ "learning_rate": 1.9936963200336287e-05,
+ "loss": 1.2004,
+ "step": 2374
+ },
+ {
+ "epoch": 6.471389645776567,
+ "grad_norm": 7.257842063903809,
+ "learning_rate": 1.9936864229216718e-05,
+ "loss": 1.1863,
+ "step": 2375
+ },
+ {
+ "epoch": 6.474114441416893,
+ "grad_norm": 6.657573223114014,
+ "learning_rate": 1.9936765180709173e-05,
+ "loss": 1.2603,
+ "step": 2376
+ },
+ {
+ "epoch": 6.476839237057221,
+ "grad_norm": 9.118010520935059,
+ "learning_rate": 1.993666605481442e-05,
+ "loss": 1.3511,
+ "step": 2377
+ },
+ {
+ "epoch": 6.479564032697548,
+ "grad_norm": 7.0830078125,
+ "learning_rate": 1.9936566851533232e-05,
+ "loss": 1.312,
+ "step": 2378
+ },
+ {
+ "epoch": 6.482288828337874,
+ "grad_norm": 5.9527506828308105,
+ "learning_rate": 1.9936467570866384e-05,
+ "loss": 1.0928,
+ "step": 2379
+ },
+ {
+ "epoch": 6.485013623978202,
+ "grad_norm": 7.874990940093994,
+ "learning_rate": 1.9936368212814643e-05,
+ "loss": 1.7275,
+ "step": 2380
+ },
+ {
+ "epoch": 6.487738419618529,
+ "grad_norm": 6.341900825500488,
+ "learning_rate": 1.9936268777378785e-05,
+ "loss": 1.2456,
+ "step": 2381
+ },
+ {
+ "epoch": 6.490463215258855,
+ "grad_norm": 8.184723854064941,
+ "learning_rate": 1.993616926455959e-05,
+ "loss": 1.2834,
+ "step": 2382
+ },
+ {
+ "epoch": 6.493188010899183,
+ "grad_norm": 9.989303588867188,
+ "learning_rate": 1.9936069674357826e-05,
+ "loss": 1.3376,
+ "step": 2383
+ },
+ {
+ "epoch": 6.4959128065395095,
+ "grad_norm": 8.05588150024414,
+ "learning_rate": 1.993597000677427e-05,
+ "loss": 1.3701,
+ "step": 2384
+ },
+ {
+ "epoch": 6.498637602179836,
+ "grad_norm": 9.415688514709473,
+ "learning_rate": 1.9935870261809703e-05,
+ "loss": 1.4473,
+ "step": 2385
+ },
+ {
+ "epoch": 6.501362397820164,
+ "grad_norm": 6.630662441253662,
+ "learning_rate": 1.9935770439464896e-05,
+ "loss": 1.4065,
+ "step": 2386
+ },
+ {
+ "epoch": 6.5040871934604905,
+ "grad_norm": 7.8249287605285645,
+ "learning_rate": 1.9935670539740626e-05,
+ "loss": 1.3108,
+ "step": 2387
+ },
+ {
+ "epoch": 6.506811989100817,
+ "grad_norm": 12.083762168884277,
+ "learning_rate": 1.9935570562637675e-05,
+ "loss": 1.4197,
+ "step": 2388
+ },
+ {
+ "epoch": 6.509536784741145,
+ "grad_norm": 10.591339111328125,
+ "learning_rate": 1.993547050815682e-05,
+ "loss": 1.543,
+ "step": 2389
+ },
+ {
+ "epoch": 6.512261580381471,
+ "grad_norm": 9.105937957763672,
+ "learning_rate": 1.9935370376298842e-05,
+ "loss": 1.4675,
+ "step": 2390
+ },
+ {
+ "epoch": 6.514986376021798,
+ "grad_norm": 6.7126688957214355,
+ "learning_rate": 1.9935270167064523e-05,
+ "loss": 1.2876,
+ "step": 2391
+ },
+ {
+ "epoch": 6.517711171662126,
+ "grad_norm": 7.337221622467041,
+ "learning_rate": 1.9935169880454635e-05,
+ "loss": 1.3328,
+ "step": 2392
+ },
+ {
+ "epoch": 6.520435967302452,
+ "grad_norm": 6.913704872131348,
+ "learning_rate": 1.9935069516469964e-05,
+ "loss": 1.4011,
+ "step": 2393
+ },
+ {
+ "epoch": 6.523160762942779,
+ "grad_norm": 6.872998237609863,
+ "learning_rate": 1.993496907511129e-05,
+ "loss": 1.3965,
+ "step": 2394
+ },
+ {
+ "epoch": 6.525885558583107,
+ "grad_norm": 6.28589391708374,
+ "learning_rate": 1.99348685563794e-05,
+ "loss": 1.3176,
+ "step": 2395
+ },
+ {
+ "epoch": 6.528610354223433,
+ "grad_norm": 9.023761749267578,
+ "learning_rate": 1.993476796027507e-05,
+ "loss": 1.5098,
+ "step": 2396
+ },
+ {
+ "epoch": 6.53133514986376,
+ "grad_norm": 7.279783248901367,
+ "learning_rate": 1.993466728679909e-05,
+ "loss": 1.572,
+ "step": 2397
+ },
+ {
+ "epoch": 6.5340599455040875,
+ "grad_norm": 6.276538372039795,
+ "learning_rate": 1.993456653595224e-05,
+ "loss": 1.2197,
+ "step": 2398
+ },
+ {
+ "epoch": 6.536784741144414,
+ "grad_norm": 8.648844718933105,
+ "learning_rate": 1.9934465707735304e-05,
+ "loss": 1.5398,
+ "step": 2399
+ },
+ {
+ "epoch": 6.539509536784741,
+ "grad_norm": 8.035370826721191,
+ "learning_rate": 1.993436480214907e-05,
+ "loss": 1.3188,
+ "step": 2400
+ },
+ {
+ "epoch": 6.5422343324250685,
+ "grad_norm": 6.565829277038574,
+ "learning_rate": 1.993426381919432e-05,
+ "loss": 1.3395,
+ "step": 2401
+ },
+ {
+ "epoch": 6.544959128065395,
+ "grad_norm": 8.744397163391113,
+ "learning_rate": 1.993416275887185e-05,
+ "loss": 1.6406,
+ "step": 2402
+ },
+ {
+ "epoch": 6.547683923705722,
+ "grad_norm": 8.441126823425293,
+ "learning_rate": 1.9934061621182433e-05,
+ "loss": 1.4414,
+ "step": 2403
+ },
+ {
+ "epoch": 6.550408719346049,
+ "grad_norm": 6.554496765136719,
+ "learning_rate": 1.993396040612687e-05,
+ "loss": 1.3237,
+ "step": 2404
+ },
+ {
+ "epoch": 6.553133514986376,
+ "grad_norm": 6.534287452697754,
+ "learning_rate": 1.9933859113705937e-05,
+ "loss": 1.4065,
+ "step": 2405
+ },
+ {
+ "epoch": 6.555858310626703,
+ "grad_norm": 6.461065769195557,
+ "learning_rate": 1.9933757743920433e-05,
+ "loss": 1.4048,
+ "step": 2406
+ },
+ {
+ "epoch": 6.55858310626703,
+ "grad_norm": 7.475233554840088,
+ "learning_rate": 1.993365629677114e-05,
+ "loss": 1.4773,
+ "step": 2407
+ },
+ {
+ "epoch": 6.561307901907357,
+ "grad_norm": 7.022715091705322,
+ "learning_rate": 1.9933554772258853e-05,
+ "loss": 1.4546,
+ "step": 2408
+ },
+ {
+ "epoch": 6.564032697547684,
+ "grad_norm": 6.8610687255859375,
+ "learning_rate": 1.9933453170384363e-05,
+ "loss": 1.4185,
+ "step": 2409
+ },
+ {
+ "epoch": 6.566757493188011,
+ "grad_norm": 9.284634590148926,
+ "learning_rate": 1.993335149114846e-05,
+ "loss": 1.3384,
+ "step": 2410
+ },
+ {
+ "epoch": 6.569482288828338,
+ "grad_norm": 7.712599754333496,
+ "learning_rate": 1.9933249734551934e-05,
+ "loss": 1.4016,
+ "step": 2411
+ },
+ {
+ "epoch": 6.572207084468665,
+ "grad_norm": 7.09529972076416,
+ "learning_rate": 1.9933147900595576e-05,
+ "loss": 1.4326,
+ "step": 2412
+ },
+ {
+ "epoch": 6.574931880108992,
+ "grad_norm": 6.15859317779541,
+ "learning_rate": 1.9933045989280188e-05,
+ "loss": 1.3044,
+ "step": 2413
+ },
+ {
+ "epoch": 6.577656675749319,
+ "grad_norm": 6.8839945793151855,
+ "learning_rate": 1.993294400060655e-05,
+ "loss": 1.5354,
+ "step": 2414
+ },
+ {
+ "epoch": 6.580381471389646,
+ "grad_norm": 8.6067476272583,
+ "learning_rate": 1.9932841934575468e-05,
+ "loss": 1.6125,
+ "step": 2415
+ },
+ {
+ "epoch": 6.583106267029972,
+ "grad_norm": 8.191581726074219,
+ "learning_rate": 1.9932739791187733e-05,
+ "loss": 1.5308,
+ "step": 2416
+ },
+ {
+ "epoch": 6.5858310626703,
+ "grad_norm": 6.960817813873291,
+ "learning_rate": 1.9932637570444138e-05,
+ "loss": 1.2053,
+ "step": 2417
+ },
+ {
+ "epoch": 6.5885558583106265,
+ "grad_norm": 7.062411785125732,
+ "learning_rate": 1.9932535272345482e-05,
+ "loss": 1.4342,
+ "step": 2418
+ },
+ {
+ "epoch": 6.591280653950953,
+ "grad_norm": 6.600743293762207,
+ "learning_rate": 1.9932432896892563e-05,
+ "loss": 1.3677,
+ "step": 2419
+ },
+ {
+ "epoch": 6.594005449591281,
+ "grad_norm": 9.019222259521484,
+ "learning_rate": 1.9932330444086175e-05,
+ "loss": 1.3835,
+ "step": 2420
+ },
+ {
+ "epoch": 6.5967302452316074,
+ "grad_norm": 7.506911754608154,
+ "learning_rate": 1.993222791392712e-05,
+ "loss": 1.2214,
+ "step": 2421
+ },
+ {
+ "epoch": 6.599455040871934,
+ "grad_norm": 7.894431114196777,
+ "learning_rate": 1.9932125306416185e-05,
+ "loss": 1.4053,
+ "step": 2422
+ },
+ {
+ "epoch": 6.602179836512262,
+ "grad_norm": 6.900913715362549,
+ "learning_rate": 1.9932022621554185e-05,
+ "loss": 1.4836,
+ "step": 2423
+ },
+ {
+ "epoch": 6.604904632152588,
+ "grad_norm": 7.44419002532959,
+ "learning_rate": 1.993191985934191e-05,
+ "loss": 1.3921,
+ "step": 2424
+ },
+ {
+ "epoch": 6.607629427792915,
+ "grad_norm": 7.529562950134277,
+ "learning_rate": 1.9931817019780165e-05,
+ "loss": 1.5759,
+ "step": 2425
+ },
+ {
+ "epoch": 6.610354223433243,
+ "grad_norm": 7.220672607421875,
+ "learning_rate": 1.993171410286975e-05,
+ "loss": 1.335,
+ "step": 2426
+ },
+ {
+ "epoch": 6.613079019073569,
+ "grad_norm": 6.616175174713135,
+ "learning_rate": 1.9931611108611458e-05,
+ "loss": 1.4446,
+ "step": 2427
+ },
+ {
+ "epoch": 6.615803814713896,
+ "grad_norm": 6.439411640167236,
+ "learning_rate": 1.9931508037006106e-05,
+ "loss": 1.4312,
+ "step": 2428
+ },
+ {
+ "epoch": 6.618528610354224,
+ "grad_norm": 7.212050914764404,
+ "learning_rate": 1.9931404888054486e-05,
+ "loss": 1.5339,
+ "step": 2429
+ },
+ {
+ "epoch": 6.62125340599455,
+ "grad_norm": 6.550750732421875,
+ "learning_rate": 1.99313016617574e-05,
+ "loss": 1.1913,
+ "step": 2430
+ },
+ {
+ "epoch": 6.623978201634877,
+ "grad_norm": 8.173152923583984,
+ "learning_rate": 1.9931198358115665e-05,
+ "loss": 1.0569,
+ "step": 2431
+ },
+ {
+ "epoch": 6.6267029972752045,
+ "grad_norm": 6.673369407653809,
+ "learning_rate": 1.9931094977130072e-05,
+ "loss": 1.2876,
+ "step": 2432
+ },
+ {
+ "epoch": 6.629427792915531,
+ "grad_norm": 6.5792036056518555,
+ "learning_rate": 1.9930991518801434e-05,
+ "loss": 1.4368,
+ "step": 2433
+ },
+ {
+ "epoch": 6.632152588555858,
+ "grad_norm": 7.3828606605529785,
+ "learning_rate": 1.993088798313055e-05,
+ "loss": 1.429,
+ "step": 2434
+ },
+ {
+ "epoch": 6.6348773841961854,
+ "grad_norm": 6.92962646484375,
+ "learning_rate": 1.993078437011823e-05,
+ "loss": 1.4338,
+ "step": 2435
+ },
+ {
+ "epoch": 6.637602179836512,
+ "grad_norm": 6.220202445983887,
+ "learning_rate": 1.9930680679765283e-05,
+ "loss": 1.4431,
+ "step": 2436
+ },
+ {
+ "epoch": 6.640326975476839,
+ "grad_norm": 7.776976108551025,
+ "learning_rate": 1.9930576912072514e-05,
+ "loss": 1.5146,
+ "step": 2437
+ },
+ {
+ "epoch": 6.643051771117166,
+ "grad_norm": 7.726137161254883,
+ "learning_rate": 1.993047306704073e-05,
+ "loss": 1.2146,
+ "step": 2438
+ },
+ {
+ "epoch": 6.645776566757493,
+ "grad_norm": 7.748034477233887,
+ "learning_rate": 1.9930369144670744e-05,
+ "loss": 1.4182,
+ "step": 2439
+ },
+ {
+ "epoch": 6.64850136239782,
+ "grad_norm": 8.11764144897461,
+ "learning_rate": 1.993026514496336e-05,
+ "loss": 1.4199,
+ "step": 2440
+ },
+ {
+ "epoch": 6.651226158038147,
+ "grad_norm": 5.926012992858887,
+ "learning_rate": 1.9930161067919392e-05,
+ "loss": 1.4683,
+ "step": 2441
+ },
+ {
+ "epoch": 6.653950953678474,
+ "grad_norm": 6.206475734710693,
+ "learning_rate": 1.9930056913539647e-05,
+ "loss": 1.4199,
+ "step": 2442
+ },
+ {
+ "epoch": 6.656675749318801,
+ "grad_norm": 9.622681617736816,
+ "learning_rate": 1.9929952681824944e-05,
+ "loss": 1.5488,
+ "step": 2443
+ },
+ {
+ "epoch": 6.659400544959128,
+ "grad_norm": 6.621033668518066,
+ "learning_rate": 1.9929848372776085e-05,
+ "loss": 1.395,
+ "step": 2444
+ },
+ {
+ "epoch": 6.662125340599455,
+ "grad_norm": 8.755743026733398,
+ "learning_rate": 1.9929743986393886e-05,
+ "loss": 1.4692,
+ "step": 2445
+ },
+ {
+ "epoch": 6.664850136239782,
+ "grad_norm": 6.953009605407715,
+ "learning_rate": 1.9929639522679163e-05,
+ "loss": 1.3135,
+ "step": 2446
+ },
+ {
+ "epoch": 6.667574931880109,
+ "grad_norm": 6.75003719329834,
+ "learning_rate": 1.9929534981632726e-05,
+ "loss": 1.4324,
+ "step": 2447
+ },
+ {
+ "epoch": 6.670299727520436,
+ "grad_norm": 5.510983943939209,
+ "learning_rate": 1.9929430363255387e-05,
+ "loss": 1.136,
+ "step": 2448
+ },
+ {
+ "epoch": 6.6730245231607626,
+ "grad_norm": 8.291926383972168,
+ "learning_rate": 1.9929325667547968e-05,
+ "loss": 1.5469,
+ "step": 2449
+ },
+ {
+ "epoch": 6.67574931880109,
+ "grad_norm": 6.4640212059021,
+ "learning_rate": 1.9929220894511275e-05,
+ "loss": 1.3589,
+ "step": 2450
+ },
+ {
+ "epoch": 6.678474114441417,
+ "grad_norm": 8.107608795166016,
+ "learning_rate": 1.9929116044146137e-05,
+ "loss": 1.4546,
+ "step": 2451
+ },
+ {
+ "epoch": 6.6811989100817435,
+ "grad_norm": 8.111456871032715,
+ "learning_rate": 1.9929011116453358e-05,
+ "loss": 1.5234,
+ "step": 2452
+ },
+ {
+ "epoch": 6.683923705722071,
+ "grad_norm": 6.997490882873535,
+ "learning_rate": 1.9928906111433762e-05,
+ "loss": 1.2061,
+ "step": 2453
+ },
+ {
+ "epoch": 6.686648501362398,
+ "grad_norm": 7.325955867767334,
+ "learning_rate": 1.9928801029088162e-05,
+ "loss": 1.2029,
+ "step": 2454
+ },
+ {
+ "epoch": 6.689373297002724,
+ "grad_norm": 8.261399269104004,
+ "learning_rate": 1.9928695869417383e-05,
+ "loss": 1.5835,
+ "step": 2455
+ },
+ {
+ "epoch": 6.692098092643052,
+ "grad_norm": 6.652465343475342,
+ "learning_rate": 1.9928590632422236e-05,
+ "loss": 1.3281,
+ "step": 2456
+ },
+ {
+ "epoch": 6.694822888283379,
+ "grad_norm": 9.42170524597168,
+ "learning_rate": 1.9928485318103545e-05,
+ "loss": 1.3701,
+ "step": 2457
+ },
+ {
+ "epoch": 6.697547683923705,
+ "grad_norm": 8.449064254760742,
+ "learning_rate": 1.992837992646213e-05,
+ "loss": 1.7002,
+ "step": 2458
+ },
+ {
+ "epoch": 6.700272479564033,
+ "grad_norm": 9.318658828735352,
+ "learning_rate": 1.9928274457498818e-05,
+ "loss": 1.2947,
+ "step": 2459
+ },
+ {
+ "epoch": 6.70299727520436,
+ "grad_norm": 8.183136940002441,
+ "learning_rate": 1.9928168911214415e-05,
+ "loss": 1.3979,
+ "step": 2460
+ },
+ {
+ "epoch": 6.705722070844686,
+ "grad_norm": 7.631108283996582,
+ "learning_rate": 1.9928063287609758e-05,
+ "loss": 1.4382,
+ "step": 2461
+ },
+ {
+ "epoch": 6.708446866485014,
+ "grad_norm": 8.89854621887207,
+ "learning_rate": 1.9927957586685664e-05,
+ "loss": 1.4785,
+ "step": 2462
+ },
+ {
+ "epoch": 6.7111716621253406,
+ "grad_norm": 8.294106483459473,
+ "learning_rate": 1.992785180844295e-05,
+ "loss": 1.4382,
+ "step": 2463
+ },
+ {
+ "epoch": 6.713896457765667,
+ "grad_norm": 8.125578880310059,
+ "learning_rate": 1.9927745952882452e-05,
+ "loss": 1.2832,
+ "step": 2464
+ },
+ {
+ "epoch": 6.716621253405995,
+ "grad_norm": 8.050348281860352,
+ "learning_rate": 1.9927640020004985e-05,
+ "loss": 1.3547,
+ "step": 2465
+ },
+ {
+ "epoch": 6.7193460490463215,
+ "grad_norm": 9.000950813293457,
+ "learning_rate": 1.9927534009811377e-05,
+ "loss": 1.3628,
+ "step": 2466
+ },
+ {
+ "epoch": 6.722070844686648,
+ "grad_norm": 7.949063777923584,
+ "learning_rate": 1.9927427922302452e-05,
+ "loss": 1.4971,
+ "step": 2467
+ },
+ {
+ "epoch": 6.724795640326976,
+ "grad_norm": 7.387415885925293,
+ "learning_rate": 1.9927321757479043e-05,
+ "loss": 1.2711,
+ "step": 2468
+ },
+ {
+ "epoch": 6.727520435967302,
+ "grad_norm": 6.908481121063232,
+ "learning_rate": 1.992721551534197e-05,
+ "loss": 1.4038,
+ "step": 2469
+ },
+ {
+ "epoch": 6.730245231607629,
+ "grad_norm": 6.895254135131836,
+ "learning_rate": 1.9927109195892062e-05,
+ "loss": 1.1826,
+ "step": 2470
+ },
+ {
+ "epoch": 6.732970027247957,
+ "grad_norm": 7.875513076782227,
+ "learning_rate": 1.9927002799130144e-05,
+ "loss": 1.592,
+ "step": 2471
+ },
+ {
+ "epoch": 6.735694822888283,
+ "grad_norm": 6.7661566734313965,
+ "learning_rate": 1.992689632505705e-05,
+ "loss": 1.4575,
+ "step": 2472
+ },
+ {
+ "epoch": 6.73841961852861,
+ "grad_norm": 10.105669021606445,
+ "learning_rate": 1.9926789773673606e-05,
+ "loss": 1.5623,
+ "step": 2473
+ },
+ {
+ "epoch": 6.741144414168938,
+ "grad_norm": 10.847541809082031,
+ "learning_rate": 1.9926683144980643e-05,
+ "loss": 1.4927,
+ "step": 2474
+ },
+ {
+ "epoch": 6.743869209809264,
+ "grad_norm": 7.324737071990967,
+ "learning_rate": 1.992657643897899e-05,
+ "loss": 1.5974,
+ "step": 2475
+ },
+ {
+ "epoch": 6.746594005449591,
+ "grad_norm": 7.341214656829834,
+ "learning_rate": 1.992646965566948e-05,
+ "loss": 1.3914,
+ "step": 2476
+ },
+ {
+ "epoch": 6.7493188010899186,
+ "grad_norm": 8.01235580444336,
+ "learning_rate": 1.9926362795052947e-05,
+ "loss": 1.3079,
+ "step": 2477
+ },
+ {
+ "epoch": 6.752043596730245,
+ "grad_norm": 8.438486099243164,
+ "learning_rate": 1.9926255857130216e-05,
+ "loss": 1.5027,
+ "step": 2478
+ },
+ {
+ "epoch": 6.754768392370572,
+ "grad_norm": 6.07291316986084,
+ "learning_rate": 1.9926148841902123e-05,
+ "loss": 1.2405,
+ "step": 2479
+ },
+ {
+ "epoch": 6.7574931880108995,
+ "grad_norm": 7.5130462646484375,
+ "learning_rate": 1.9926041749369507e-05,
+ "loss": 1.6062,
+ "step": 2480
+ },
+ {
+ "epoch": 6.760217983651226,
+ "grad_norm": 28.45442771911621,
+ "learning_rate": 1.9925934579533195e-05,
+ "loss": 1.511,
+ "step": 2481
+ },
+ {
+ "epoch": 6.762942779291553,
+ "grad_norm": 6.216997146606445,
+ "learning_rate": 1.9925827332394023e-05,
+ "loss": 1.2495,
+ "step": 2482
+ },
+ {
+ "epoch": 6.76566757493188,
+ "grad_norm": 7.470963478088379,
+ "learning_rate": 1.9925720007952826e-05,
+ "loss": 1.3809,
+ "step": 2483
+ },
+ {
+ "epoch": 6.768392370572207,
+ "grad_norm": 6.0761919021606445,
+ "learning_rate": 1.9925612606210446e-05,
+ "loss": 1.261,
+ "step": 2484
+ },
+ {
+ "epoch": 6.771117166212534,
+ "grad_norm": 6.613674640655518,
+ "learning_rate": 1.992550512716771e-05,
+ "loss": 1.5662,
+ "step": 2485
+ },
+ {
+ "epoch": 6.773841961852861,
+ "grad_norm": 8.827567100524902,
+ "learning_rate": 1.9925397570825463e-05,
+ "loss": 1.4873,
+ "step": 2486
+ },
+ {
+ "epoch": 6.776566757493188,
+ "grad_norm": 7.785755157470703,
+ "learning_rate": 1.9925289937184534e-05,
+ "loss": 1.4443,
+ "step": 2487
+ },
+ {
+ "epoch": 6.779291553133515,
+ "grad_norm": 6.925635814666748,
+ "learning_rate": 1.992518222624577e-05,
+ "loss": 1.4456,
+ "step": 2488
+ },
+ {
+ "epoch": 6.782016348773842,
+ "grad_norm": 27.685195922851562,
+ "learning_rate": 1.992507443801001e-05,
+ "loss": 1.4456,
+ "step": 2489
+ },
+ {
+ "epoch": 6.784741144414169,
+ "grad_norm": 6.300507068634033,
+ "learning_rate": 1.9924966572478083e-05,
+ "loss": 1.3611,
+ "step": 2490
+ },
+ {
+ "epoch": 6.787465940054496,
+ "grad_norm": 7.346882343292236,
+ "learning_rate": 1.992485862965084e-05,
+ "loss": 1.2679,
+ "step": 2491
+ },
+ {
+ "epoch": 6.790190735694823,
+ "grad_norm": 7.450957775115967,
+ "learning_rate": 1.9924750609529116e-05,
+ "loss": 1.3137,
+ "step": 2492
+ },
+ {
+ "epoch": 6.79291553133515,
+ "grad_norm": 7.339519500732422,
+ "learning_rate": 1.992464251211375e-05,
+ "loss": 1.5105,
+ "step": 2493
+ },
+ {
+ "epoch": 6.795640326975477,
+ "grad_norm": 6.717392444610596,
+ "learning_rate": 1.992453433740559e-05,
+ "loss": 1.3586,
+ "step": 2494
+ },
+ {
+ "epoch": 6.798365122615804,
+ "grad_norm": 7.2443413734436035,
+ "learning_rate": 1.9924426085405478e-05,
+ "loss": 1.3564,
+ "step": 2495
+ },
+ {
+ "epoch": 6.801089918256131,
+ "grad_norm": 8.842680931091309,
+ "learning_rate": 1.9924317756114255e-05,
+ "loss": 1.3662,
+ "step": 2496
+ },
+ {
+ "epoch": 6.8038147138964575,
+ "grad_norm": 8.169709205627441,
+ "learning_rate": 1.9924209349532763e-05,
+ "loss": 1.3611,
+ "step": 2497
+ },
+ {
+ "epoch": 6.806539509536785,
+ "grad_norm": 6.333761692047119,
+ "learning_rate": 1.992410086566185e-05,
+ "loss": 1.3672,
+ "step": 2498
+ },
+ {
+ "epoch": 6.809264305177112,
+ "grad_norm": 6.677126407623291,
+ "learning_rate": 1.9923992304502357e-05,
+ "loss": 1.21,
+ "step": 2499
+ },
+ {
+ "epoch": 6.8119891008174385,
+ "grad_norm": 8.255160331726074,
+ "learning_rate": 1.9923883666055133e-05,
+ "loss": 1.5156,
+ "step": 2500
+ },
+ {
+ "epoch": 6.814713896457766,
+ "grad_norm": 6.572474479675293,
+ "learning_rate": 1.992377495032102e-05,
+ "loss": 1.4407,
+ "step": 2501
+ },
+ {
+ "epoch": 6.817438692098093,
+ "grad_norm": 8.083229064941406,
+ "learning_rate": 1.992366615730087e-05,
+ "loss": 1.4246,
+ "step": 2502
+ },
+ {
+ "epoch": 6.820163487738419,
+ "grad_norm": 6.874819755554199,
+ "learning_rate": 1.9923557286995524e-05,
+ "loss": 1.3096,
+ "step": 2503
+ },
+ {
+ "epoch": 6.822888283378747,
+ "grad_norm": 7.388861656188965,
+ "learning_rate": 1.9923448339405835e-05,
+ "loss": 1.2778,
+ "step": 2504
+ },
+ {
+ "epoch": 6.825613079019074,
+ "grad_norm": 7.943492889404297,
+ "learning_rate": 1.992333931453265e-05,
+ "loss": 1.4456,
+ "step": 2505
+ },
+ {
+ "epoch": 6.8283378746594,
+ "grad_norm": 6.515859603881836,
+ "learning_rate": 1.992323021237682e-05,
+ "loss": 1.3547,
+ "step": 2506
+ },
+ {
+ "epoch": 6.831062670299728,
+ "grad_norm": 7.426928520202637,
+ "learning_rate": 1.992312103293919e-05,
+ "loss": 1.4216,
+ "step": 2507
+ },
+ {
+ "epoch": 6.833787465940055,
+ "grad_norm": 7.113417625427246,
+ "learning_rate": 1.9923011776220612e-05,
+ "loss": 1.3721,
+ "step": 2508
+ },
+ {
+ "epoch": 6.836512261580381,
+ "grad_norm": 8.768729209899902,
+ "learning_rate": 1.992290244222194e-05,
+ "loss": 1.8093,
+ "step": 2509
+ },
+ {
+ "epoch": 6.839237057220709,
+ "grad_norm": 6.769860744476318,
+ "learning_rate": 1.992279303094402e-05,
+ "loss": 1.3611,
+ "step": 2510
+ },
+ {
+ "epoch": 6.8419618528610355,
+ "grad_norm": 6.736438274383545,
+ "learning_rate": 1.9922683542387707e-05,
+ "loss": 1.3521,
+ "step": 2511
+ },
+ {
+ "epoch": 6.844686648501362,
+ "grad_norm": 6.9644246101379395,
+ "learning_rate": 1.9922573976553857e-05,
+ "loss": 1.634,
+ "step": 2512
+ },
+ {
+ "epoch": 6.84741144414169,
+ "grad_norm": 7.978062629699707,
+ "learning_rate": 1.9922464333443318e-05,
+ "loss": 1.303,
+ "step": 2513
+ },
+ {
+ "epoch": 6.8501362397820165,
+ "grad_norm": 6.847949981689453,
+ "learning_rate": 1.9922354613056946e-05,
+ "loss": 1.3572,
+ "step": 2514
+ },
+ {
+ "epoch": 6.852861035422343,
+ "grad_norm": 6.631079196929932,
+ "learning_rate": 1.9922244815395597e-05,
+ "loss": 1.1931,
+ "step": 2515
+ },
+ {
+ "epoch": 6.855585831062671,
+ "grad_norm": 7.419920921325684,
+ "learning_rate": 1.9922134940460126e-05,
+ "loss": 1.3584,
+ "step": 2516
+ },
+ {
+ "epoch": 6.858310626702997,
+ "grad_norm": 6.412455081939697,
+ "learning_rate": 1.9922024988251384e-05,
+ "loss": 1.2366,
+ "step": 2517
+ },
+ {
+ "epoch": 6.861035422343324,
+ "grad_norm": 7.975281238555908,
+ "learning_rate": 1.9921914958770235e-05,
+ "loss": 1.3828,
+ "step": 2518
+ },
+ {
+ "epoch": 6.863760217983652,
+ "grad_norm": 8.20242691040039,
+ "learning_rate": 1.992180485201753e-05,
+ "loss": 1.3796,
+ "step": 2519
+ },
+ {
+ "epoch": 6.866485013623978,
+ "grad_norm": 7.767943859100342,
+ "learning_rate": 1.9921694667994126e-05,
+ "loss": 1.6025,
+ "step": 2520
+ },
+ {
+ "epoch": 6.869209809264305,
+ "grad_norm": 7.013779163360596,
+ "learning_rate": 1.9921584406700883e-05,
+ "loss": 1.905,
+ "step": 2521
+ },
+ {
+ "epoch": 6.871934604904633,
+ "grad_norm": 8.225257873535156,
+ "learning_rate": 1.9921474068138665e-05,
+ "loss": 1.3472,
+ "step": 2522
+ },
+ {
+ "epoch": 6.874659400544959,
+ "grad_norm": 6.98057222366333,
+ "learning_rate": 1.9921363652308324e-05,
+ "loss": 1.4053,
+ "step": 2523
+ },
+ {
+ "epoch": 6.877384196185286,
+ "grad_norm": 7.621967315673828,
+ "learning_rate": 1.9921253159210718e-05,
+ "loss": 1.4795,
+ "step": 2524
+ },
+ {
+ "epoch": 6.8801089918256135,
+ "grad_norm": 7.491020202636719,
+ "learning_rate": 1.9921142588846717e-05,
+ "loss": 1.5432,
+ "step": 2525
+ },
+ {
+ "epoch": 6.88283378746594,
+ "grad_norm": 7.078356742858887,
+ "learning_rate": 1.9921031941217173e-05,
+ "loss": 1.3303,
+ "step": 2526
+ },
+ {
+ "epoch": 6.885558583106267,
+ "grad_norm": 6.242447853088379,
+ "learning_rate": 1.9920921216322953e-05,
+ "loss": 1.1945,
+ "step": 2527
+ },
+ {
+ "epoch": 6.8882833787465945,
+ "grad_norm": 7.436166763305664,
+ "learning_rate": 1.992081041416492e-05,
+ "loss": 1.2896,
+ "step": 2528
+ },
+ {
+ "epoch": 6.891008174386921,
+ "grad_norm": 7.5910210609436035,
+ "learning_rate": 1.9920699534743932e-05,
+ "loss": 1.5586,
+ "step": 2529
+ },
+ {
+ "epoch": 6.893732970027248,
+ "grad_norm": 6.452704906463623,
+ "learning_rate": 1.9920588578060854e-05,
+ "loss": 1.3939,
+ "step": 2530
+ },
+ {
+ "epoch": 6.896457765667575,
+ "grad_norm": 5.932440757751465,
+ "learning_rate": 1.9920477544116554e-05,
+ "loss": 1.3723,
+ "step": 2531
+ },
+ {
+ "epoch": 6.899182561307902,
+ "grad_norm": 7.539181709289551,
+ "learning_rate": 1.9920366432911895e-05,
+ "loss": 1.6017,
+ "step": 2532
+ },
+ {
+ "epoch": 6.901907356948229,
+ "grad_norm": 15.600433349609375,
+ "learning_rate": 1.992025524444774e-05,
+ "loss": 1.4824,
+ "step": 2533
+ },
+ {
+ "epoch": 6.904632152588556,
+ "grad_norm": 8.792279243469238,
+ "learning_rate": 1.9920143978724958e-05,
+ "loss": 1.4893,
+ "step": 2534
+ },
+ {
+ "epoch": 6.907356948228883,
+ "grad_norm": 8.643928527832031,
+ "learning_rate": 1.992003263574441e-05,
+ "loss": 1.5005,
+ "step": 2535
+ },
+ {
+ "epoch": 6.91008174386921,
+ "grad_norm": 7.131587505340576,
+ "learning_rate": 1.991992121550697e-05,
+ "loss": 1.4358,
+ "step": 2536
+ },
+ {
+ "epoch": 6.912806539509537,
+ "grad_norm": 6.637341022491455,
+ "learning_rate": 1.9919809718013505e-05,
+ "loss": 1.282,
+ "step": 2537
+ },
+ {
+ "epoch": 6.915531335149864,
+ "grad_norm": 6.289490222930908,
+ "learning_rate": 1.991969814326488e-05,
+ "loss": 1.4126,
+ "step": 2538
+ },
+ {
+ "epoch": 6.918256130790191,
+ "grad_norm": 7.545890808105469,
+ "learning_rate": 1.9919586491261962e-05,
+ "loss": 1.4014,
+ "step": 2539
+ },
+ {
+ "epoch": 6.920980926430518,
+ "grad_norm": 7.596290111541748,
+ "learning_rate": 1.9919474762005625e-05,
+ "loss": 1.5613,
+ "step": 2540
+ },
+ {
+ "epoch": 6.923705722070845,
+ "grad_norm": 6.706675052642822,
+ "learning_rate": 1.991936295549674e-05,
+ "loss": 1.626,
+ "step": 2541
+ },
+ {
+ "epoch": 6.926430517711172,
+ "grad_norm": 7.789161682128906,
+ "learning_rate": 1.9919251071736176e-05,
+ "loss": 1.5339,
+ "step": 2542
+ },
+ {
+ "epoch": 6.929155313351498,
+ "grad_norm": 7.960029125213623,
+ "learning_rate": 1.9919139110724798e-05,
+ "loss": 1.4414,
+ "step": 2543
+ },
+ {
+ "epoch": 6.931880108991826,
+ "grad_norm": 10.37203311920166,
+ "learning_rate": 1.991902707246349e-05,
+ "loss": 1.342,
+ "step": 2544
+ },
+ {
+ "epoch": 6.9346049046321525,
+ "grad_norm": 9.642186164855957,
+ "learning_rate": 1.9918914956953116e-05,
+ "loss": 1.3645,
+ "step": 2545
+ },
+ {
+ "epoch": 6.937329700272479,
+ "grad_norm": 9.908488273620605,
+ "learning_rate": 1.991880276419455e-05,
+ "loss": 1.3721,
+ "step": 2546
+ },
+ {
+ "epoch": 6.940054495912807,
+ "grad_norm": 6.77915096282959,
+ "learning_rate": 1.991869049418867e-05,
+ "loss": 1.3875,
+ "step": 2547
+ },
+ {
+ "epoch": 6.9427792915531334,
+ "grad_norm": 20.526159286499023,
+ "learning_rate": 1.9918578146936346e-05,
+ "loss": 1.5962,
+ "step": 2548
+ },
+ {
+ "epoch": 6.94550408719346,
+ "grad_norm": 9.632704734802246,
+ "learning_rate": 1.9918465722438456e-05,
+ "loss": 1.5083,
+ "step": 2549
+ },
+ {
+ "epoch": 6.948228882833788,
+ "grad_norm": 6.912860870361328,
+ "learning_rate": 1.9918353220695875e-05,
+ "loss": 1.4641,
+ "step": 2550
+ },
+ {
+ "epoch": 6.950953678474114,
+ "grad_norm": 7.2022013664245605,
+ "learning_rate": 1.9918240641709475e-05,
+ "loss": 1.4419,
+ "step": 2551
+ },
+ {
+ "epoch": 6.953678474114441,
+ "grad_norm": 7.831475257873535,
+ "learning_rate": 1.991812798548014e-05,
+ "loss": 1.5632,
+ "step": 2552
+ },
+ {
+ "epoch": 6.956403269754769,
+ "grad_norm": 6.887185096740723,
+ "learning_rate": 1.991801525200874e-05,
+ "loss": 1.3933,
+ "step": 2553
+ },
+ {
+ "epoch": 6.959128065395095,
+ "grad_norm": 7.002577781677246,
+ "learning_rate": 1.9917902441296156e-05,
+ "loss": 1.4451,
+ "step": 2554
+ },
+ {
+ "epoch": 6.961852861035422,
+ "grad_norm": 8.499896049499512,
+ "learning_rate": 1.991778955334327e-05,
+ "loss": 1.3269,
+ "step": 2555
+ },
+ {
+ "epoch": 6.96457765667575,
+ "grad_norm": 24.235363006591797,
+ "learning_rate": 1.991767658815096e-05,
+ "loss": 1.5703,
+ "step": 2556
+ },
+ {
+ "epoch": 6.967302452316076,
+ "grad_norm": 11.380416870117188,
+ "learning_rate": 1.9917563545720098e-05,
+ "loss": 1.6172,
+ "step": 2557
+ },
+ {
+ "epoch": 6.970027247956403,
+ "grad_norm": 11.243563652038574,
+ "learning_rate": 1.9917450426051572e-05,
+ "loss": 1.427,
+ "step": 2558
+ },
+ {
+ "epoch": 6.9727520435967305,
+ "grad_norm": 6.485527515411377,
+ "learning_rate": 1.9917337229146266e-05,
+ "loss": 1.3167,
+ "step": 2559
+ },
+ {
+ "epoch": 6.975476839237057,
+ "grad_norm": 7.107969760894775,
+ "learning_rate": 1.9917223955005053e-05,
+ "loss": 1.2766,
+ "step": 2560
+ },
+ {
+ "epoch": 6.978201634877384,
+ "grad_norm": 8.056703567504883,
+ "learning_rate": 1.9917110603628818e-05,
+ "loss": 1.405,
+ "step": 2561
+ },
+ {
+ "epoch": 6.9809264305177114,
+ "grad_norm": 8.373993873596191,
+ "learning_rate": 1.9916997175018445e-05,
+ "loss": 1.2739,
+ "step": 2562
+ },
+ {
+ "epoch": 6.983651226158038,
+ "grad_norm": 6.725265979766846,
+ "learning_rate": 1.9916883669174818e-05,
+ "loss": 1.311,
+ "step": 2563
+ },
+ {
+ "epoch": 6.986376021798365,
+ "grad_norm": 7.789998531341553,
+ "learning_rate": 1.991677008609882e-05,
+ "loss": 1.509,
+ "step": 2564
+ },
+ {
+ "epoch": 6.989100817438692,
+ "grad_norm": 9.457737922668457,
+ "learning_rate": 1.9916656425791338e-05,
+ "loss": 1.179,
+ "step": 2565
+ },
+ {
+ "epoch": 6.991825613079019,
+ "grad_norm": 10.736144065856934,
+ "learning_rate": 1.9916542688253253e-05,
+ "loss": 1.417,
+ "step": 2566
+ },
+ {
+ "epoch": 6.994550408719346,
+ "grad_norm": 9.309311866760254,
+ "learning_rate": 1.9916428873485452e-05,
+ "loss": 1.5552,
+ "step": 2567
+ },
+ {
+ "epoch": 6.997275204359673,
+ "grad_norm": 9.60059642791748,
+ "learning_rate": 1.9916314981488822e-05,
+ "loss": 1.3225,
+ "step": 2568
+ },
+ {
+ "epoch": 7.0,
+ "grad_norm": 8.055014610290527,
+ "learning_rate": 1.9916201012264255e-05,
+ "loss": 1.2693,
+ "step": 2569
+ },
+ {
+ "epoch": 7.002724795640327,
+ "grad_norm": 8.72629451751709,
+ "learning_rate": 1.9916086965812627e-05,
+ "loss": 1.4043,
+ "step": 2570
+ },
+ {
+ "epoch": 7.005449591280654,
+ "grad_norm": 13.681690216064453,
+ "learning_rate": 1.9915972842134837e-05,
+ "loss": 1.4795,
+ "step": 2571
+ },
+ {
+ "epoch": 7.008174386920981,
+ "grad_norm": 8.961599349975586,
+ "learning_rate": 1.991585864123177e-05,
+ "loss": 1.4006,
+ "step": 2572
+ },
+ {
+ "epoch": 7.010899182561308,
+ "grad_norm": 6.990932464599609,
+ "learning_rate": 1.991574436310431e-05,
+ "loss": 1.4353,
+ "step": 2573
+ },
+ {
+ "epoch": 7.013623978201635,
+ "grad_norm": 8.562370300292969,
+ "learning_rate": 1.991563000775336e-05,
+ "loss": 1.488,
+ "step": 2574
+ },
+ {
+ "epoch": 7.016348773841962,
+ "grad_norm": 8.870628356933594,
+ "learning_rate": 1.9915515575179793e-05,
+ "loss": 1.2719,
+ "step": 2575
+ },
+ {
+ "epoch": 7.0190735694822886,
+ "grad_norm": 8.464811325073242,
+ "learning_rate": 1.9915401065384518e-05,
+ "loss": 1.4137,
+ "step": 2576
+ },
+ {
+ "epoch": 7.021798365122616,
+ "grad_norm": 6.561202049255371,
+ "learning_rate": 1.9915286478368414e-05,
+ "loss": 1.4417,
+ "step": 2577
+ },
+ {
+ "epoch": 7.024523160762943,
+ "grad_norm": 10.164609909057617,
+ "learning_rate": 1.9915171814132378e-05,
+ "loss": 1.2061,
+ "step": 2578
+ },
+ {
+ "epoch": 7.0272479564032695,
+ "grad_norm": 18.185876846313477,
+ "learning_rate": 1.9915057072677305e-05,
+ "loss": 1.4175,
+ "step": 2579
+ },
+ {
+ "epoch": 7.029972752043597,
+ "grad_norm": 8.86538314819336,
+ "learning_rate": 1.9914942254004086e-05,
+ "loss": 1.448,
+ "step": 2580
+ },
+ {
+ "epoch": 7.032697547683924,
+ "grad_norm": 9.0018892288208,
+ "learning_rate": 1.9914827358113616e-05,
+ "loss": 1.2047,
+ "step": 2581
+ },
+ {
+ "epoch": 7.03542234332425,
+ "grad_norm": 8.659799575805664,
+ "learning_rate": 1.991471238500679e-05,
+ "loss": 1.3889,
+ "step": 2582
+ },
+ {
+ "epoch": 7.038147138964578,
+ "grad_norm": 7.920725345611572,
+ "learning_rate": 1.9914597334684502e-05,
+ "loss": 1.4626,
+ "step": 2583
+ },
+ {
+ "epoch": 7.040871934604905,
+ "grad_norm": 7.034748554229736,
+ "learning_rate": 1.9914482207147648e-05,
+ "loss": 1.324,
+ "step": 2584
+ },
+ {
+ "epoch": 7.043596730245231,
+ "grad_norm": 8.431821823120117,
+ "learning_rate": 1.9914367002397127e-05,
+ "loss": 1.4033,
+ "step": 2585
+ },
+ {
+ "epoch": 7.046321525885559,
+ "grad_norm": 9.534443855285645,
+ "learning_rate": 1.9914251720433837e-05,
+ "loss": 1.2295,
+ "step": 2586
+ },
+ {
+ "epoch": 7.049046321525886,
+ "grad_norm": 11.702399253845215,
+ "learning_rate": 1.9914136361258673e-05,
+ "loss": 1.5476,
+ "step": 2587
+ },
+ {
+ "epoch": 7.051771117166212,
+ "grad_norm": 9.313124656677246,
+ "learning_rate": 1.991402092487253e-05,
+ "loss": 1.4163,
+ "step": 2588
+ },
+ {
+ "epoch": 7.05449591280654,
+ "grad_norm": 7.37802267074585,
+ "learning_rate": 1.991390541127632e-05,
+ "loss": 1.3716,
+ "step": 2589
+ },
+ {
+ "epoch": 7.0572207084468666,
+ "grad_norm": 9.159991264343262,
+ "learning_rate": 1.9913789820470925e-05,
+ "loss": 1.5237,
+ "step": 2590
+ },
+ {
+ "epoch": 7.059945504087193,
+ "grad_norm": 7.437915325164795,
+ "learning_rate": 1.9913674152457254e-05,
+ "loss": 1.2534,
+ "step": 2591
+ },
+ {
+ "epoch": 7.062670299727521,
+ "grad_norm": 7.337875843048096,
+ "learning_rate": 1.991355840723621e-05,
+ "loss": 1.3533,
+ "step": 2592
+ },
+ {
+ "epoch": 7.0653950953678475,
+ "grad_norm": 7.345780372619629,
+ "learning_rate": 1.9913442584808694e-05,
+ "loss": 1.2837,
+ "step": 2593
+ },
+ {
+ "epoch": 7.068119891008174,
+ "grad_norm": 11.815799713134766,
+ "learning_rate": 1.9913326685175606e-05,
+ "loss": 1.2227,
+ "step": 2594
+ },
+ {
+ "epoch": 7.070844686648502,
+ "grad_norm": 8.777254104614258,
+ "learning_rate": 1.991321070833785e-05,
+ "loss": 1.231,
+ "step": 2595
+ },
+ {
+ "epoch": 7.073569482288828,
+ "grad_norm": 11.258925437927246,
+ "learning_rate": 1.9913094654296324e-05,
+ "loss": 1.2454,
+ "step": 2596
+ },
+ {
+ "epoch": 7.076294277929155,
+ "grad_norm": 15.257661819458008,
+ "learning_rate": 1.9912978523051936e-05,
+ "loss": 1.2188,
+ "step": 2597
+ },
+ {
+ "epoch": 7.079019073569483,
+ "grad_norm": 8.178746223449707,
+ "learning_rate": 1.9912862314605594e-05,
+ "loss": 1.2598,
+ "step": 2598
+ },
+ {
+ "epoch": 7.081743869209809,
+ "grad_norm": 7.409693241119385,
+ "learning_rate": 1.9912746028958195e-05,
+ "loss": 1.2517,
+ "step": 2599
+ },
+ {
+ "epoch": 7.084468664850136,
+ "grad_norm": 8.080255508422852,
+ "learning_rate": 1.9912629666110656e-05,
+ "loss": 1.3474,
+ "step": 2600
+ },
+ {
+ "epoch": 7.087193460490464,
+ "grad_norm": 7.083448886871338,
+ "learning_rate": 1.9912513226063866e-05,
+ "loss": 1.1367,
+ "step": 2601
+ },
+ {
+ "epoch": 7.08991825613079,
+ "grad_norm": 9.915042877197266,
+ "learning_rate": 1.9912396708818748e-05,
+ "loss": 1.3496,
+ "step": 2602
+ },
+ {
+ "epoch": 7.092643051771117,
+ "grad_norm": 9.506633758544922,
+ "learning_rate": 1.9912280114376205e-05,
+ "loss": 1.5576,
+ "step": 2603
+ },
+ {
+ "epoch": 7.0953678474114446,
+ "grad_norm": 8.341775894165039,
+ "learning_rate": 1.9912163442737143e-05,
+ "loss": 1.1616,
+ "step": 2604
+ },
+ {
+ "epoch": 7.098092643051771,
+ "grad_norm": 8.883849143981934,
+ "learning_rate": 1.9912046693902466e-05,
+ "loss": 1.4207,
+ "step": 2605
+ },
+ {
+ "epoch": 7.100817438692098,
+ "grad_norm": 9.84878921508789,
+ "learning_rate": 1.9911929867873093e-05,
+ "loss": 1.4424,
+ "step": 2606
+ },
+ {
+ "epoch": 7.1035422343324255,
+ "grad_norm": 7.159688472747803,
+ "learning_rate": 1.991181296464993e-05,
+ "loss": 1.1067,
+ "step": 2607
+ },
+ {
+ "epoch": 7.106267029972752,
+ "grad_norm": 8.226995468139648,
+ "learning_rate": 1.9911695984233885e-05,
+ "loss": 1.2118,
+ "step": 2608
+ },
+ {
+ "epoch": 7.108991825613079,
+ "grad_norm": 7.753889083862305,
+ "learning_rate": 1.991157892662587e-05,
+ "loss": 1.376,
+ "step": 2609
+ },
+ {
+ "epoch": 7.111716621253406,
+ "grad_norm": 10.06631851196289,
+ "learning_rate": 1.99114617918268e-05,
+ "loss": 1.3923,
+ "step": 2610
+ },
+ {
+ "epoch": 7.114441416893733,
+ "grad_norm": 10.797825813293457,
+ "learning_rate": 1.991134457983758e-05,
+ "loss": 1.4121,
+ "step": 2611
+ },
+ {
+ "epoch": 7.11716621253406,
+ "grad_norm": 7.541214942932129,
+ "learning_rate": 1.991122729065913e-05,
+ "loss": 1.3525,
+ "step": 2612
+ },
+ {
+ "epoch": 7.1198910081743865,
+ "grad_norm": 9.477609634399414,
+ "learning_rate": 1.9911109924292362e-05,
+ "loss": 1.1582,
+ "step": 2613
+ },
+ {
+ "epoch": 7.122615803814714,
+ "grad_norm": 9.118778228759766,
+ "learning_rate": 1.991099248073819e-05,
+ "loss": 1.1956,
+ "step": 2614
+ },
+ {
+ "epoch": 7.125340599455041,
+ "grad_norm": 6.8957953453063965,
+ "learning_rate": 1.991087495999752e-05,
+ "loss": 1.3535,
+ "step": 2615
+ },
+ {
+ "epoch": 7.128065395095367,
+ "grad_norm": 9.191266059875488,
+ "learning_rate": 1.9910757362071282e-05,
+ "loss": 1.1387,
+ "step": 2616
+ },
+ {
+ "epoch": 7.130790190735695,
+ "grad_norm": 7.089980602264404,
+ "learning_rate": 1.9910639686960382e-05,
+ "loss": 1.1562,
+ "step": 2617
+ },
+ {
+ "epoch": 7.133514986376022,
+ "grad_norm": 9.796401977539062,
+ "learning_rate": 1.991052193466574e-05,
+ "loss": 1.364,
+ "step": 2618
+ },
+ {
+ "epoch": 7.136239782016348,
+ "grad_norm": 8.165899276733398,
+ "learning_rate": 1.991040410518827e-05,
+ "loss": 1.2588,
+ "step": 2619
+ },
+ {
+ "epoch": 7.138964577656676,
+ "grad_norm": 10.613434791564941,
+ "learning_rate": 1.9910286198528897e-05,
+ "loss": 1.4243,
+ "step": 2620
+ },
+ {
+ "epoch": 7.141689373297003,
+ "grad_norm": 7.741814613342285,
+ "learning_rate": 1.9910168214688527e-05,
+ "loss": 1.2632,
+ "step": 2621
+ },
+ {
+ "epoch": 7.144414168937329,
+ "grad_norm": 7.31572151184082,
+ "learning_rate": 1.991005015366809e-05,
+ "loss": 1.1633,
+ "step": 2622
+ },
+ {
+ "epoch": 7.147138964577657,
+ "grad_norm": 8.135992050170898,
+ "learning_rate": 1.9909932015468504e-05,
+ "loss": 1.3506,
+ "step": 2623
+ },
+ {
+ "epoch": 7.1498637602179835,
+ "grad_norm": 7.060415267944336,
+ "learning_rate": 1.9909813800090682e-05,
+ "loss": 1.2905,
+ "step": 2624
+ },
+ {
+ "epoch": 7.15258855585831,
+ "grad_norm": 9.621841430664062,
+ "learning_rate": 1.9909695507535552e-05,
+ "loss": 1.4009,
+ "step": 2625
+ },
+ {
+ "epoch": 7.155313351498638,
+ "grad_norm": 7.952399730682373,
+ "learning_rate": 1.9909577137804028e-05,
+ "loss": 1.2603,
+ "step": 2626
+ },
+ {
+ "epoch": 7.1580381471389645,
+ "grad_norm": 5.800620079040527,
+ "learning_rate": 1.9909458690897038e-05,
+ "loss": 1.0779,
+ "step": 2627
+ },
+ {
+ "epoch": 7.160762942779291,
+ "grad_norm": 9.602071762084961,
+ "learning_rate": 1.9909340166815504e-05,
+ "loss": 1.3928,
+ "step": 2628
+ },
+ {
+ "epoch": 7.163487738419619,
+ "grad_norm": 12.935136795043945,
+ "learning_rate": 1.9909221565560348e-05,
+ "loss": 1.2561,
+ "step": 2629
+ },
+ {
+ "epoch": 7.166212534059945,
+ "grad_norm": 21.123899459838867,
+ "learning_rate": 1.9909102887132494e-05,
+ "loss": 1.3269,
+ "step": 2630
+ },
+ {
+ "epoch": 7.168937329700272,
+ "grad_norm": 11.29782485961914,
+ "learning_rate": 1.9908984131532865e-05,
+ "loss": 1.2053,
+ "step": 2631
+ },
+ {
+ "epoch": 7.1716621253406,
+ "grad_norm": 7.305057525634766,
+ "learning_rate": 1.9908865298762384e-05,
+ "loss": 1.5251,
+ "step": 2632
+ },
+ {
+ "epoch": 7.174386920980926,
+ "grad_norm": 8.624639511108398,
+ "learning_rate": 1.990874638882198e-05,
+ "loss": 1.2002,
+ "step": 2633
+ },
+ {
+ "epoch": 7.177111716621253,
+ "grad_norm": 7.460022449493408,
+ "learning_rate": 1.990862740171258e-05,
+ "loss": 1.2141,
+ "step": 2634
+ },
+ {
+ "epoch": 7.179836512261581,
+ "grad_norm": 9.15855598449707,
+ "learning_rate": 1.990850833743511e-05,
+ "loss": 1.3542,
+ "step": 2635
+ },
+ {
+ "epoch": 7.182561307901907,
+ "grad_norm": 7.166216850280762,
+ "learning_rate": 1.9908389195990492e-05,
+ "loss": 1.3315,
+ "step": 2636
+ },
+ {
+ "epoch": 7.185286103542234,
+ "grad_norm": 7.993762969970703,
+ "learning_rate": 1.9908269977379658e-05,
+ "loss": 1.2646,
+ "step": 2637
+ },
+ {
+ "epoch": 7.1880108991825615,
+ "grad_norm": 7.316840648651123,
+ "learning_rate": 1.990815068160354e-05,
+ "loss": 1.2129,
+ "step": 2638
+ },
+ {
+ "epoch": 7.190735694822888,
+ "grad_norm": 7.4955244064331055,
+ "learning_rate": 1.990803130866306e-05,
+ "loss": 1.2378,
+ "step": 2639
+ },
+ {
+ "epoch": 7.193460490463215,
+ "grad_norm": 7.234216690063477,
+ "learning_rate": 1.990791185855915e-05,
+ "loss": 1.3328,
+ "step": 2640
+ },
+ {
+ "epoch": 7.1961852861035425,
+ "grad_norm": 8.3865327835083,
+ "learning_rate": 1.9907792331292747e-05,
+ "loss": 1.2744,
+ "step": 2641
+ },
+ {
+ "epoch": 7.198910081743869,
+ "grad_norm": 7.540198802947998,
+ "learning_rate": 1.9907672726864772e-05,
+ "loss": 1.3259,
+ "step": 2642
+ },
+ {
+ "epoch": 7.201634877384196,
+ "grad_norm": 9.075204849243164,
+ "learning_rate": 1.9907553045276158e-05,
+ "loss": 1.4135,
+ "step": 2643
+ },
+ {
+ "epoch": 7.204359673024523,
+ "grad_norm": 7.82284688949585,
+ "learning_rate": 1.9907433286527843e-05,
+ "loss": 1.3226,
+ "step": 2644
+ },
+ {
+ "epoch": 7.20708446866485,
+ "grad_norm": 7.4608941078186035,
+ "learning_rate": 1.990731345062076e-05,
+ "loss": 1.553,
+ "step": 2645
+ },
+ {
+ "epoch": 7.209809264305177,
+ "grad_norm": 9.801565170288086,
+ "learning_rate": 1.9907193537555832e-05,
+ "loss": 1.498,
+ "step": 2646
+ },
+ {
+ "epoch": 7.212534059945504,
+ "grad_norm": 8.238839149475098,
+ "learning_rate": 1.9907073547334002e-05,
+ "loss": 1.1868,
+ "step": 2647
+ },
+ {
+ "epoch": 7.215258855585831,
+ "grad_norm": 6.529207229614258,
+ "learning_rate": 1.9906953479956206e-05,
+ "loss": 1.1726,
+ "step": 2648
+ },
+ {
+ "epoch": 7.217983651226158,
+ "grad_norm": 9.57068920135498,
+ "learning_rate": 1.990683333542337e-05,
+ "loss": 1.3892,
+ "step": 2649
+ },
+ {
+ "epoch": 7.220708446866485,
+ "grad_norm": 6.971983432769775,
+ "learning_rate": 1.990671311373644e-05,
+ "loss": 1.3267,
+ "step": 2650
+ },
+ {
+ "epoch": 7.223433242506812,
+ "grad_norm": 37.79975891113281,
+ "learning_rate": 1.9906592814896343e-05,
+ "loss": 1.5474,
+ "step": 2651
+ },
+ {
+ "epoch": 7.226158038147139,
+ "grad_norm": 6.882627487182617,
+ "learning_rate": 1.990647243890402e-05,
+ "loss": 1.2913,
+ "step": 2652
+ },
+ {
+ "epoch": 7.228882833787466,
+ "grad_norm": 9.166817665100098,
+ "learning_rate": 1.990635198576041e-05,
+ "loss": 1.3096,
+ "step": 2653
+ },
+ {
+ "epoch": 7.231607629427793,
+ "grad_norm": 8.919292449951172,
+ "learning_rate": 1.990623145546645e-05,
+ "loss": 1.3481,
+ "step": 2654
+ },
+ {
+ "epoch": 7.23433242506812,
+ "grad_norm": 10.063273429870605,
+ "learning_rate": 1.9906110848023074e-05,
+ "loss": 1.3494,
+ "step": 2655
+ },
+ {
+ "epoch": 7.237057220708447,
+ "grad_norm": 6.623021125793457,
+ "learning_rate": 1.990599016343123e-05,
+ "loss": 1.2644,
+ "step": 2656
+ },
+ {
+ "epoch": 7.239782016348774,
+ "grad_norm": 7.145871162414551,
+ "learning_rate": 1.9905869401691854e-05,
+ "loss": 1.3201,
+ "step": 2657
+ },
+ {
+ "epoch": 7.2425068119891005,
+ "grad_norm": 6.786539077758789,
+ "learning_rate": 1.990574856280588e-05,
+ "loss": 1.3357,
+ "step": 2658
+ },
+ {
+ "epoch": 7.245231607629428,
+ "grad_norm": 8.135835647583008,
+ "learning_rate": 1.9905627646774262e-05,
+ "loss": 1.4216,
+ "step": 2659
+ },
+ {
+ "epoch": 7.247956403269755,
+ "grad_norm": 9.786505699157715,
+ "learning_rate": 1.990550665359793e-05,
+ "loss": 1.4321,
+ "step": 2660
+ },
+ {
+ "epoch": 7.2506811989100814,
+ "grad_norm": 6.786805629730225,
+ "learning_rate": 1.9905385583277834e-05,
+ "loss": 1.4629,
+ "step": 2661
+ },
+ {
+ "epoch": 7.253405994550409,
+ "grad_norm": 8.286147117614746,
+ "learning_rate": 1.9905264435814913e-05,
+ "loss": 1.24,
+ "step": 2662
+ },
+ {
+ "epoch": 7.256130790190736,
+ "grad_norm": 7.399831295013428,
+ "learning_rate": 1.990514321121011e-05,
+ "loss": 1.2522,
+ "step": 2663
+ },
+ {
+ "epoch": 7.258855585831062,
+ "grad_norm": 7.541548728942871,
+ "learning_rate": 1.9905021909464372e-05,
+ "loss": 1.3198,
+ "step": 2664
+ },
+ {
+ "epoch": 7.26158038147139,
+ "grad_norm": 7.2452263832092285,
+ "learning_rate": 1.9904900530578643e-05,
+ "loss": 1.291,
+ "step": 2665
+ },
+ {
+ "epoch": 7.264305177111717,
+ "grad_norm": 8.113701820373535,
+ "learning_rate": 1.9904779074553864e-05,
+ "loss": 1.3452,
+ "step": 2666
+ },
+ {
+ "epoch": 7.267029972752043,
+ "grad_norm": 7.9984917640686035,
+ "learning_rate": 1.990465754139099e-05,
+ "loss": 1.4257,
+ "step": 2667
+ },
+ {
+ "epoch": 7.269754768392371,
+ "grad_norm": 8.617762565612793,
+ "learning_rate": 1.9904535931090957e-05,
+ "loss": 1.438,
+ "step": 2668
+ },
+ {
+ "epoch": 7.272479564032698,
+ "grad_norm": 8.943818092346191,
+ "learning_rate": 1.9904414243654716e-05,
+ "loss": 1.3358,
+ "step": 2669
+ },
+ {
+ "epoch": 7.275204359673024,
+ "grad_norm": 8.454083442687988,
+ "learning_rate": 1.9904292479083217e-05,
+ "loss": 1.3064,
+ "step": 2670
+ },
+ {
+ "epoch": 7.277929155313352,
+ "grad_norm": 7.77817440032959,
+ "learning_rate": 1.9904170637377408e-05,
+ "loss": 1.4434,
+ "step": 2671
+ },
+ {
+ "epoch": 7.2806539509536785,
+ "grad_norm": 8.732303619384766,
+ "learning_rate": 1.9904048718538238e-05,
+ "loss": 1.2681,
+ "step": 2672
+ },
+ {
+ "epoch": 7.283378746594005,
+ "grad_norm": 8.765336036682129,
+ "learning_rate": 1.9903926722566655e-05,
+ "loss": 1.2432,
+ "step": 2673
+ },
+ {
+ "epoch": 7.286103542234333,
+ "grad_norm": 8.141328811645508,
+ "learning_rate": 1.990380464946361e-05,
+ "loss": 1.1755,
+ "step": 2674
+ },
+ {
+ "epoch": 7.2888283378746594,
+ "grad_norm": 8.56711196899414,
+ "learning_rate": 1.990368249923005e-05,
+ "loss": 1.5942,
+ "step": 2675
+ },
+ {
+ "epoch": 7.291553133514986,
+ "grad_norm": 9.373531341552734,
+ "learning_rate": 1.990356027186693e-05,
+ "loss": 1.2061,
+ "step": 2676
+ },
+ {
+ "epoch": 7.294277929155314,
+ "grad_norm": 6.667110443115234,
+ "learning_rate": 1.9903437967375202e-05,
+ "loss": 1.1996,
+ "step": 2677
+ },
+ {
+ "epoch": 7.29700272479564,
+ "grad_norm": 7.602336406707764,
+ "learning_rate": 1.990331558575582e-05,
+ "loss": 1.3154,
+ "step": 2678
+ },
+ {
+ "epoch": 7.299727520435967,
+ "grad_norm": 8.987732887268066,
+ "learning_rate": 1.990319312700973e-05,
+ "loss": 1.3518,
+ "step": 2679
+ },
+ {
+ "epoch": 7.302452316076295,
+ "grad_norm": 7.405623435974121,
+ "learning_rate": 1.9903070591137898e-05,
+ "loss": 0.9977,
+ "step": 2680
+ },
+ {
+ "epoch": 7.305177111716621,
+ "grad_norm": 6.405440807342529,
+ "learning_rate": 1.9902947978141265e-05,
+ "loss": 1.2524,
+ "step": 2681
+ },
+ {
+ "epoch": 7.307901907356948,
+ "grad_norm": 5.988929748535156,
+ "learning_rate": 1.9902825288020794e-05,
+ "loss": 1.3276,
+ "step": 2682
+ },
+ {
+ "epoch": 7.310626702997276,
+ "grad_norm": 10.550492286682129,
+ "learning_rate": 1.990270252077744e-05,
+ "loss": 1.5786,
+ "step": 2683
+ },
+ {
+ "epoch": 7.313351498637602,
+ "grad_norm": 9.189135551452637,
+ "learning_rate": 1.9902579676412154e-05,
+ "loss": 1.3159,
+ "step": 2684
+ },
+ {
+ "epoch": 7.316076294277929,
+ "grad_norm": 6.353959560394287,
+ "learning_rate": 1.9902456754925898e-05,
+ "loss": 1.2183,
+ "step": 2685
+ },
+ {
+ "epoch": 7.3188010899182565,
+ "grad_norm": 8.8511381149292,
+ "learning_rate": 1.990233375631963e-05,
+ "loss": 1.3748,
+ "step": 2686
+ },
+ {
+ "epoch": 7.321525885558583,
+ "grad_norm": 8.13774299621582,
+ "learning_rate": 1.99022106805943e-05,
+ "loss": 1.2363,
+ "step": 2687
+ },
+ {
+ "epoch": 7.32425068119891,
+ "grad_norm": 8.307180404663086,
+ "learning_rate": 1.9902087527750876e-05,
+ "loss": 1.3706,
+ "step": 2688
+ },
+ {
+ "epoch": 7.3269754768392374,
+ "grad_norm": 7.619308948516846,
+ "learning_rate": 1.990196429779031e-05,
+ "loss": 1.3523,
+ "step": 2689
+ },
+ {
+ "epoch": 7.329700272479564,
+ "grad_norm": 7.28941011428833,
+ "learning_rate": 1.9901840990713566e-05,
+ "loss": 1.2434,
+ "step": 2690
+ },
+ {
+ "epoch": 7.332425068119891,
+ "grad_norm": 8.371277809143066,
+ "learning_rate": 1.9901717606521607e-05,
+ "loss": 1.4146,
+ "step": 2691
+ },
+ {
+ "epoch": 7.335149863760218,
+ "grad_norm": 8.823556900024414,
+ "learning_rate": 1.9901594145215382e-05,
+ "loss": 1.4583,
+ "step": 2692
+ },
+ {
+ "epoch": 7.337874659400545,
+ "grad_norm": 8.13704776763916,
+ "learning_rate": 1.9901470606795867e-05,
+ "loss": 1.3215,
+ "step": 2693
+ },
+ {
+ "epoch": 7.340599455040872,
+ "grad_norm": 8.459822654724121,
+ "learning_rate": 1.9901346991264018e-05,
+ "loss": 1.3286,
+ "step": 2694
+ },
+ {
+ "epoch": 7.343324250681199,
+ "grad_norm": 8.95654582977295,
+ "learning_rate": 1.9901223298620793e-05,
+ "loss": 1.2622,
+ "step": 2695
+ },
+ {
+ "epoch": 7.346049046321526,
+ "grad_norm": 8.602657318115234,
+ "learning_rate": 1.9901099528867164e-05,
+ "loss": 1.4883,
+ "step": 2696
+ },
+ {
+ "epoch": 7.348773841961853,
+ "grad_norm": 7.879006862640381,
+ "learning_rate": 1.9900975682004088e-05,
+ "loss": 1.3687,
+ "step": 2697
+ },
+ {
+ "epoch": 7.35149863760218,
+ "grad_norm": 7.693984508514404,
+ "learning_rate": 1.990085175803253e-05,
+ "loss": 1.3337,
+ "step": 2698
+ },
+ {
+ "epoch": 7.354223433242507,
+ "grad_norm": 7.662132263183594,
+ "learning_rate": 1.990072775695346e-05,
+ "loss": 1.2798,
+ "step": 2699
+ },
+ {
+ "epoch": 7.356948228882834,
+ "grad_norm": 6.233168125152588,
+ "learning_rate": 1.990060367876784e-05,
+ "loss": 1.3894,
+ "step": 2700
+ },
+ {
+ "epoch": 7.359673024523161,
+ "grad_norm": 7.656670093536377,
+ "learning_rate": 1.9900479523476637e-05,
+ "loss": 1.2703,
+ "step": 2701
+ },
+ {
+ "epoch": 7.362397820163488,
+ "grad_norm": 8.63752555847168,
+ "learning_rate": 1.9900355291080816e-05,
+ "loss": 1.3215,
+ "step": 2702
+ },
+ {
+ "epoch": 7.3651226158038146,
+ "grad_norm": 8.558443069458008,
+ "learning_rate": 1.990023098158135e-05,
+ "loss": 1.3491,
+ "step": 2703
+ },
+ {
+ "epoch": 7.367847411444142,
+ "grad_norm": 11.786442756652832,
+ "learning_rate": 1.9900106594979204e-05,
+ "loss": 1.3748,
+ "step": 2704
+ },
+ {
+ "epoch": 7.370572207084469,
+ "grad_norm": 8.644862174987793,
+ "learning_rate": 1.9899982131275345e-05,
+ "loss": 1.311,
+ "step": 2705
+ },
+ {
+ "epoch": 7.3732970027247955,
+ "grad_norm": 10.715784072875977,
+ "learning_rate": 1.9899857590470745e-05,
+ "loss": 1.3445,
+ "step": 2706
+ },
+ {
+ "epoch": 7.376021798365123,
+ "grad_norm": 5.951897621154785,
+ "learning_rate": 1.989973297256637e-05,
+ "loss": 1.3218,
+ "step": 2707
+ },
+ {
+ "epoch": 7.37874659400545,
+ "grad_norm": 7.871620178222656,
+ "learning_rate": 1.9899608277563197e-05,
+ "loss": 1.1782,
+ "step": 2708
+ },
+ {
+ "epoch": 7.381471389645776,
+ "grad_norm": 7.311661720275879,
+ "learning_rate": 1.989948350546219e-05,
+ "loss": 1.1316,
+ "step": 2709
+ },
+ {
+ "epoch": 7.384196185286104,
+ "grad_norm": 8.803568840026855,
+ "learning_rate": 1.9899358656264325e-05,
+ "loss": 1.2803,
+ "step": 2710
+ },
+ {
+ "epoch": 7.386920980926431,
+ "grad_norm": 8.371910095214844,
+ "learning_rate": 1.9899233729970576e-05,
+ "loss": 1.1799,
+ "step": 2711
+ },
+ {
+ "epoch": 7.389645776566757,
+ "grad_norm": 8.22465705871582,
+ "learning_rate": 1.9899108726581907e-05,
+ "loss": 1.2141,
+ "step": 2712
+ },
+ {
+ "epoch": 7.392370572207085,
+ "grad_norm": 6.310781002044678,
+ "learning_rate": 1.9898983646099304e-05,
+ "loss": 1.1555,
+ "step": 2713
+ },
+ {
+ "epoch": 7.395095367847412,
+ "grad_norm": 6.839968681335449,
+ "learning_rate": 1.989885848852373e-05,
+ "loss": 1.4275,
+ "step": 2714
+ },
+ {
+ "epoch": 7.397820163487738,
+ "grad_norm": 7.015069484710693,
+ "learning_rate": 1.989873325385617e-05,
+ "loss": 1.2966,
+ "step": 2715
+ },
+ {
+ "epoch": 7.400544959128065,
+ "grad_norm": 7.386134624481201,
+ "learning_rate": 1.989860794209759e-05,
+ "loss": 1.3499,
+ "step": 2716
+ },
+ {
+ "epoch": 7.4032697547683926,
+ "grad_norm": 7.467782974243164,
+ "learning_rate": 1.989848255324897e-05,
+ "loss": 1.3809,
+ "step": 2717
+ },
+ {
+ "epoch": 7.405994550408719,
+ "grad_norm": 8.890801429748535,
+ "learning_rate": 1.989835708731129e-05,
+ "loss": 1.2385,
+ "step": 2718
+ },
+ {
+ "epoch": 7.408719346049046,
+ "grad_norm": 8.179885864257812,
+ "learning_rate": 1.9898231544285523e-05,
+ "loss": 1.3928,
+ "step": 2719
+ },
+ {
+ "epoch": 7.4114441416893735,
+ "grad_norm": 10.72572135925293,
+ "learning_rate": 1.9898105924172647e-05,
+ "loss": 1.3096,
+ "step": 2720
+ },
+ {
+ "epoch": 7.4141689373297,
+ "grad_norm": 5.725125312805176,
+ "learning_rate": 1.989798022697364e-05,
+ "loss": 1.0693,
+ "step": 2721
+ },
+ {
+ "epoch": 7.416893732970027,
+ "grad_norm": 8.071654319763184,
+ "learning_rate": 1.989785445268948e-05,
+ "loss": 1.2739,
+ "step": 2722
+ },
+ {
+ "epoch": 7.419618528610354,
+ "grad_norm": 13.280393600463867,
+ "learning_rate": 1.9897728601321148e-05,
+ "loss": 1.447,
+ "step": 2723
+ },
+ {
+ "epoch": 7.422343324250681,
+ "grad_norm": 7.438470840454102,
+ "learning_rate": 1.9897602672869624e-05,
+ "loss": 1.292,
+ "step": 2724
+ },
+ {
+ "epoch": 7.425068119891008,
+ "grad_norm": 8.170390129089355,
+ "learning_rate": 1.989747666733589e-05,
+ "loss": 1.2505,
+ "step": 2725
+ },
+ {
+ "epoch": 7.427792915531335,
+ "grad_norm": 11.986186981201172,
+ "learning_rate": 1.9897350584720928e-05,
+ "loss": 1.3345,
+ "step": 2726
+ },
+ {
+ "epoch": 7.430517711171662,
+ "grad_norm": 12.195749282836914,
+ "learning_rate": 1.9897224425025716e-05,
+ "loss": 1.2157,
+ "step": 2727
+ },
+ {
+ "epoch": 7.433242506811989,
+ "grad_norm": 11.467879295349121,
+ "learning_rate": 1.9897098188251244e-05,
+ "loss": 1.1802,
+ "step": 2728
+ },
+ {
+ "epoch": 7.435967302452316,
+ "grad_norm": 10.521123886108398,
+ "learning_rate": 1.9896971874398485e-05,
+ "loss": 1.369,
+ "step": 2729
+ },
+ {
+ "epoch": 7.438692098092643,
+ "grad_norm": 77.02208709716797,
+ "learning_rate": 1.9896845483468424e-05,
+ "loss": 1.4336,
+ "step": 2730
+ },
+ {
+ "epoch": 7.44141689373297,
+ "grad_norm": 10.476792335510254,
+ "learning_rate": 1.9896719015462055e-05,
+ "loss": 1.2937,
+ "step": 2731
+ },
+ {
+ "epoch": 7.444141689373297,
+ "grad_norm": 9.210094451904297,
+ "learning_rate": 1.9896592470380354e-05,
+ "loss": 1.3645,
+ "step": 2732
+ },
+ {
+ "epoch": 7.446866485013624,
+ "grad_norm": 8.787375450134277,
+ "learning_rate": 1.9896465848224312e-05,
+ "loss": 1.3481,
+ "step": 2733
+ },
+ {
+ "epoch": 7.449591280653951,
+ "grad_norm": 18.40317726135254,
+ "learning_rate": 1.9896339148994906e-05,
+ "loss": 1.345,
+ "step": 2734
+ },
+ {
+ "epoch": 7.452316076294278,
+ "grad_norm": 12.705636978149414,
+ "learning_rate": 1.9896212372693135e-05,
+ "loss": 1.3352,
+ "step": 2735
+ },
+ {
+ "epoch": 7.455040871934605,
+ "grad_norm": 10.528398513793945,
+ "learning_rate": 1.9896085519319982e-05,
+ "loss": 1.4685,
+ "step": 2736
+ },
+ {
+ "epoch": 7.4577656675749315,
+ "grad_norm": 9.275914192199707,
+ "learning_rate": 1.989595858887643e-05,
+ "loss": 1.4587,
+ "step": 2737
+ },
+ {
+ "epoch": 7.460490463215259,
+ "grad_norm": 12.22417163848877,
+ "learning_rate": 1.989583158136347e-05,
+ "loss": 1.3215,
+ "step": 2738
+ },
+ {
+ "epoch": 7.463215258855586,
+ "grad_norm": 9.307333946228027,
+ "learning_rate": 1.9895704496782097e-05,
+ "loss": 1.6091,
+ "step": 2739
+ },
+ {
+ "epoch": 7.4659400544959125,
+ "grad_norm": 10.014742851257324,
+ "learning_rate": 1.989557733513329e-05,
+ "loss": 1.2314,
+ "step": 2740
+ },
+ {
+ "epoch": 7.46866485013624,
+ "grad_norm": 11.895537376403809,
+ "learning_rate": 1.989545009641805e-05,
+ "loss": 1.2214,
+ "step": 2741
+ },
+ {
+ "epoch": 7.471389645776567,
+ "grad_norm": 16.69625473022461,
+ "learning_rate": 1.9895322780637358e-05,
+ "loss": 1.3777,
+ "step": 2742
+ },
+ {
+ "epoch": 7.474114441416893,
+ "grad_norm": 8.64697265625,
+ "learning_rate": 1.9895195387792214e-05,
+ "loss": 1.246,
+ "step": 2743
+ },
+ {
+ "epoch": 7.476839237057221,
+ "grad_norm": 14.263720512390137,
+ "learning_rate": 1.989506791788361e-05,
+ "loss": 1.4434,
+ "step": 2744
+ },
+ {
+ "epoch": 7.479564032697548,
+ "grad_norm": 7.612471580505371,
+ "learning_rate": 1.989494037091253e-05,
+ "loss": 1.3774,
+ "step": 2745
+ },
+ {
+ "epoch": 7.482288828337874,
+ "grad_norm": 9.99272632598877,
+ "learning_rate": 1.9894812746879976e-05,
+ "loss": 1.3682,
+ "step": 2746
+ },
+ {
+ "epoch": 7.485013623978202,
+ "grad_norm": 12.111319541931152,
+ "learning_rate": 1.9894685045786935e-05,
+ "loss": 1.4227,
+ "step": 2747
+ },
+ {
+ "epoch": 7.487738419618529,
+ "grad_norm": 8.02752685546875,
+ "learning_rate": 1.989455726763441e-05,
+ "loss": 1.3975,
+ "step": 2748
+ },
+ {
+ "epoch": 7.490463215258855,
+ "grad_norm": 15.913735389709473,
+ "learning_rate": 1.9894429412423388e-05,
+ "loss": 1.2395,
+ "step": 2749
+ },
+ {
+ "epoch": 7.493188010899183,
+ "grad_norm": 8.949234008789062,
+ "learning_rate": 1.9894301480154873e-05,
+ "loss": 1.4807,
+ "step": 2750
+ },
+ {
+ "epoch": 7.4959128065395095,
+ "grad_norm": 10.25567626953125,
+ "learning_rate": 1.9894173470829853e-05,
+ "loss": 1.2905,
+ "step": 2751
+ },
+ {
+ "epoch": 7.498637602179836,
+ "grad_norm": 5.8087873458862305,
+ "learning_rate": 1.9894045384449328e-05,
+ "loss": 1.1296,
+ "step": 2752
+ },
+ {
+ "epoch": 7.501362397820164,
+ "grad_norm": 7.848937034606934,
+ "learning_rate": 1.9893917221014294e-05,
+ "loss": 1.3398,
+ "step": 2753
+ },
+ {
+ "epoch": 7.5040871934604905,
+ "grad_norm": 8.157247543334961,
+ "learning_rate": 1.9893788980525756e-05,
+ "loss": 1.313,
+ "step": 2754
+ },
+ {
+ "epoch": 7.506811989100817,
+ "grad_norm": 10.420892715454102,
+ "learning_rate": 1.9893660662984705e-05,
+ "loss": 1.375,
+ "step": 2755
+ },
+ {
+ "epoch": 7.509536784741145,
+ "grad_norm": 7.761693477630615,
+ "learning_rate": 1.9893532268392143e-05,
+ "loss": 1.1527,
+ "step": 2756
+ },
+ {
+ "epoch": 7.512261580381471,
+ "grad_norm": 6.901098251342773,
+ "learning_rate": 1.989340379674907e-05,
+ "loss": 1.5154,
+ "step": 2757
+ },
+ {
+ "epoch": 7.514986376021798,
+ "grad_norm": 10.140838623046875,
+ "learning_rate": 1.989327524805649e-05,
+ "loss": 1.4189,
+ "step": 2758
+ },
+ {
+ "epoch": 7.517711171662126,
+ "grad_norm": 9.050856590270996,
+ "learning_rate": 1.9893146622315398e-05,
+ "loss": 1.4526,
+ "step": 2759
+ },
+ {
+ "epoch": 7.520435967302452,
+ "grad_norm": 8.08426284790039,
+ "learning_rate": 1.9893017919526798e-05,
+ "loss": 1.4043,
+ "step": 2760
+ },
+ {
+ "epoch": 7.523160762942779,
+ "grad_norm": 11.233299255371094,
+ "learning_rate": 1.9892889139691695e-05,
+ "loss": 1.2881,
+ "step": 2761
+ },
+ {
+ "epoch": 7.525885558583107,
+ "grad_norm": 7.793832302093506,
+ "learning_rate": 1.9892760282811085e-05,
+ "loss": 1.6763,
+ "step": 2762
+ },
+ {
+ "epoch": 7.528610354223433,
+ "grad_norm": 8.406953811645508,
+ "learning_rate": 1.989263134888598e-05,
+ "loss": 1.1794,
+ "step": 2763
+ },
+ {
+ "epoch": 7.53133514986376,
+ "grad_norm": 9.268343925476074,
+ "learning_rate": 1.9892502337917382e-05,
+ "loss": 1.4292,
+ "step": 2764
+ },
+ {
+ "epoch": 7.5340599455040875,
+ "grad_norm": 9.231379508972168,
+ "learning_rate": 1.9892373249906292e-05,
+ "loss": 1.2036,
+ "step": 2765
+ },
+ {
+ "epoch": 7.536784741144414,
+ "grad_norm": 12.321741104125977,
+ "learning_rate": 1.989224408485372e-05,
+ "loss": 1.5674,
+ "step": 2766
+ },
+ {
+ "epoch": 7.539509536784741,
+ "grad_norm": 7.858902454376221,
+ "learning_rate": 1.9892114842760666e-05,
+ "loss": 1.3157,
+ "step": 2767
+ },
+ {
+ "epoch": 7.5422343324250685,
+ "grad_norm": 7.1961870193481445,
+ "learning_rate": 1.9891985523628142e-05,
+ "loss": 1.5784,
+ "step": 2768
+ },
+ {
+ "epoch": 7.544959128065395,
+ "grad_norm": 9.360504150390625,
+ "learning_rate": 1.9891856127457152e-05,
+ "loss": 1.2747,
+ "step": 2769
+ },
+ {
+ "epoch": 7.547683923705722,
+ "grad_norm": 6.838207244873047,
+ "learning_rate": 1.9891726654248705e-05,
+ "loss": 1.3323,
+ "step": 2770
+ },
+ {
+ "epoch": 7.550408719346049,
+ "grad_norm": 8.240283966064453,
+ "learning_rate": 1.9891597104003807e-05,
+ "loss": 1.4299,
+ "step": 2771
+ },
+ {
+ "epoch": 7.553133514986376,
+ "grad_norm": 8.082183837890625,
+ "learning_rate": 1.9891467476723472e-05,
+ "loss": 1.3545,
+ "step": 2772
+ },
+ {
+ "epoch": 7.555858310626703,
+ "grad_norm": 6.930272579193115,
+ "learning_rate": 1.9891337772408706e-05,
+ "loss": 1.2576,
+ "step": 2773
+ },
+ {
+ "epoch": 7.55858310626703,
+ "grad_norm": 7.374207496643066,
+ "learning_rate": 1.9891207991060525e-05,
+ "loss": 1.3247,
+ "step": 2774
+ },
+ {
+ "epoch": 7.561307901907357,
+ "grad_norm": 8.299930572509766,
+ "learning_rate": 1.989107813267993e-05,
+ "loss": 1.3523,
+ "step": 2775
+ },
+ {
+ "epoch": 7.564032697547684,
+ "grad_norm": 7.999646186828613,
+ "learning_rate": 1.9890948197267933e-05,
+ "loss": 1.312,
+ "step": 2776
+ },
+ {
+ "epoch": 7.566757493188011,
+ "grad_norm": 7.514304161071777,
+ "learning_rate": 1.9890818184825552e-05,
+ "loss": 1.2522,
+ "step": 2777
+ },
+ {
+ "epoch": 7.569482288828338,
+ "grad_norm": 7.257941722869873,
+ "learning_rate": 1.9890688095353802e-05,
+ "loss": 1.2993,
+ "step": 2778
+ },
+ {
+ "epoch": 7.572207084468665,
+ "grad_norm": 15.037428855895996,
+ "learning_rate": 1.9890557928853685e-05,
+ "loss": 1.2651,
+ "step": 2779
+ },
+ {
+ "epoch": 7.574931880108992,
+ "grad_norm": 7.599269866943359,
+ "learning_rate": 1.9890427685326225e-05,
+ "loss": 1.5518,
+ "step": 2780
+ },
+ {
+ "epoch": 7.577656675749319,
+ "grad_norm": 15.512941360473633,
+ "learning_rate": 1.989029736477243e-05,
+ "loss": 1.4465,
+ "step": 2781
+ },
+ {
+ "epoch": 7.580381471389646,
+ "grad_norm": 8.45249080657959,
+ "learning_rate": 1.989016696719332e-05,
+ "loss": 1.2639,
+ "step": 2782
+ },
+ {
+ "epoch": 7.583106267029972,
+ "grad_norm": 7.9410014152526855,
+ "learning_rate": 1.989003649258991e-05,
+ "loss": 1.33,
+ "step": 2783
+ },
+ {
+ "epoch": 7.5858310626703,
+ "grad_norm": 7.711286544799805,
+ "learning_rate": 1.988990594096321e-05,
+ "loss": 1.3835,
+ "step": 2784
+ },
+ {
+ "epoch": 7.5885558583106265,
+ "grad_norm": 7.25645112991333,
+ "learning_rate": 1.988977531231424e-05,
+ "loss": 1.2898,
+ "step": 2785
+ },
+ {
+ "epoch": 7.591280653950953,
+ "grad_norm": 14.64525032043457,
+ "learning_rate": 1.988964460664402e-05,
+ "loss": 1.4407,
+ "step": 2786
+ },
+ {
+ "epoch": 7.594005449591281,
+ "grad_norm": 7.487483024597168,
+ "learning_rate": 1.9889513823953568e-05,
+ "loss": 1.3835,
+ "step": 2787
+ },
+ {
+ "epoch": 7.5967302452316074,
+ "grad_norm": 7.937008380889893,
+ "learning_rate": 1.9889382964243896e-05,
+ "loss": 1.3516,
+ "step": 2788
+ },
+ {
+ "epoch": 7.599455040871934,
+ "grad_norm": 7.496700286865234,
+ "learning_rate": 1.9889252027516033e-05,
+ "loss": 1.3242,
+ "step": 2789
+ },
+ {
+ "epoch": 7.602179836512262,
+ "grad_norm": 7.9101104736328125,
+ "learning_rate": 1.988912101377099e-05,
+ "loss": 1.3184,
+ "step": 2790
+ },
+ {
+ "epoch": 7.604904632152588,
+ "grad_norm": 7.429482936859131,
+ "learning_rate": 1.9888989923009792e-05,
+ "loss": 1.4116,
+ "step": 2791
+ },
+ {
+ "epoch": 7.607629427792915,
+ "grad_norm": 7.869822025299072,
+ "learning_rate": 1.988885875523346e-05,
+ "loss": 1.4871,
+ "step": 2792
+ },
+ {
+ "epoch": 7.610354223433243,
+ "grad_norm": 6.766198635101318,
+ "learning_rate": 1.988872751044301e-05,
+ "loss": 1.1609,
+ "step": 2793
+ },
+ {
+ "epoch": 7.613079019073569,
+ "grad_norm": 7.963929653167725,
+ "learning_rate": 1.9888596188639476e-05,
+ "loss": 1.4077,
+ "step": 2794
+ },
+ {
+ "epoch": 7.615803814713896,
+ "grad_norm": 11.447989463806152,
+ "learning_rate": 1.9888464789823867e-05,
+ "loss": 1.3979,
+ "step": 2795
+ },
+ {
+ "epoch": 7.618528610354224,
+ "grad_norm": 6.677192211151123,
+ "learning_rate": 1.9888333313997212e-05,
+ "loss": 1.3188,
+ "step": 2796
+ },
+ {
+ "epoch": 7.62125340599455,
+ "grad_norm": 6.883086681365967,
+ "learning_rate": 1.988820176116054e-05,
+ "loss": 1.3765,
+ "step": 2797
+ },
+ {
+ "epoch": 7.623978201634877,
+ "grad_norm": 7.704653263092041,
+ "learning_rate": 1.988807013131487e-05,
+ "loss": 1.4355,
+ "step": 2798
+ },
+ {
+ "epoch": 7.6267029972752045,
+ "grad_norm": 7.397902488708496,
+ "learning_rate": 1.9887938424461224e-05,
+ "loss": 1.0964,
+ "step": 2799
+ },
+ {
+ "epoch": 7.629427792915531,
+ "grad_norm": 8.43236255645752,
+ "learning_rate": 1.9887806640600633e-05,
+ "loss": 1.3298,
+ "step": 2800
+ },
+ {
+ "epoch": 7.632152588555858,
+ "grad_norm": 8.536269187927246,
+ "learning_rate": 1.9887674779734125e-05,
+ "loss": 1.2266,
+ "step": 2801
+ },
+ {
+ "epoch": 7.6348773841961854,
+ "grad_norm": 7.96507453918457,
+ "learning_rate": 1.9887542841862723e-05,
+ "loss": 1.5627,
+ "step": 2802
+ },
+ {
+ "epoch": 7.637602179836512,
+ "grad_norm": 6.571836471557617,
+ "learning_rate": 1.9887410826987455e-05,
+ "loss": 1.2734,
+ "step": 2803
+ },
+ {
+ "epoch": 7.640326975476839,
+ "grad_norm": 7.011343955993652,
+ "learning_rate": 1.988727873510935e-05,
+ "loss": 1.24,
+ "step": 2804
+ },
+ {
+ "epoch": 7.643051771117166,
+ "grad_norm": 6.604487419128418,
+ "learning_rate": 1.9887146566229437e-05,
+ "loss": 1.4058,
+ "step": 2805
+ },
+ {
+ "epoch": 7.645776566757493,
+ "grad_norm": 5.616354465484619,
+ "learning_rate": 1.988701432034874e-05,
+ "loss": 1.0657,
+ "step": 2806
+ },
+ {
+ "epoch": 7.64850136239782,
+ "grad_norm": 10.035872459411621,
+ "learning_rate": 1.9886881997468298e-05,
+ "loss": 1.3433,
+ "step": 2807
+ },
+ {
+ "epoch": 7.651226158038147,
+ "grad_norm": 7.262260437011719,
+ "learning_rate": 1.988674959758914e-05,
+ "loss": 1.3374,
+ "step": 2808
+ },
+ {
+ "epoch": 7.653950953678474,
+ "grad_norm": 8.633551597595215,
+ "learning_rate": 1.988661712071229e-05,
+ "loss": 1.3906,
+ "step": 2809
+ },
+ {
+ "epoch": 7.656675749318801,
+ "grad_norm": 7.220731258392334,
+ "learning_rate": 1.9886484566838785e-05,
+ "loss": 1.4402,
+ "step": 2810
+ },
+ {
+ "epoch": 7.659400544959128,
+ "grad_norm": 9.968592643737793,
+ "learning_rate": 1.9886351935969658e-05,
+ "loss": 1.3552,
+ "step": 2811
+ },
+ {
+ "epoch": 7.662125340599455,
+ "grad_norm": 8.029172897338867,
+ "learning_rate": 1.9886219228105937e-05,
+ "loss": 1.4636,
+ "step": 2812
+ },
+ {
+ "epoch": 7.664850136239782,
+ "grad_norm": 6.421130657196045,
+ "learning_rate": 1.9886086443248663e-05,
+ "loss": 1.3391,
+ "step": 2813
+ },
+ {
+ "epoch": 7.667574931880109,
+ "grad_norm": 9.19809627532959,
+ "learning_rate": 1.9885953581398862e-05,
+ "loss": 1.3179,
+ "step": 2814
+ },
+ {
+ "epoch": 7.670299727520436,
+ "grad_norm": 12.610840797424316,
+ "learning_rate": 1.9885820642557574e-05,
+ "loss": 1.3894,
+ "step": 2815
+ },
+ {
+ "epoch": 7.6730245231607626,
+ "grad_norm": 5.442038059234619,
+ "learning_rate": 1.9885687626725833e-05,
+ "loss": 1.2891,
+ "step": 2816
+ },
+ {
+ "epoch": 7.67574931880109,
+ "grad_norm": 8.97093677520752,
+ "learning_rate": 1.9885554533904673e-05,
+ "loss": 1.3113,
+ "step": 2817
+ },
+ {
+ "epoch": 7.678474114441417,
+ "grad_norm": 8.05801010131836,
+ "learning_rate": 1.988542136409514e-05,
+ "loss": 1.3469,
+ "step": 2818
+ },
+ {
+ "epoch": 7.6811989100817435,
+ "grad_norm": 5.120886325836182,
+ "learning_rate": 1.9885288117298256e-05,
+ "loss": 1.1122,
+ "step": 2819
+ },
+ {
+ "epoch": 7.683923705722071,
+ "grad_norm": 9.530280113220215,
+ "learning_rate": 1.9885154793515066e-05,
+ "loss": 1.3269,
+ "step": 2820
+ },
+ {
+ "epoch": 7.686648501362398,
+ "grad_norm": 5.885916233062744,
+ "learning_rate": 1.9885021392746614e-05,
+ "loss": 1.0955,
+ "step": 2821
+ },
+ {
+ "epoch": 7.689373297002724,
+ "grad_norm": 8.865469932556152,
+ "learning_rate": 1.9884887914993932e-05,
+ "loss": 1.3652,
+ "step": 2822
+ },
+ {
+ "epoch": 7.692098092643052,
+ "grad_norm": 7.625950813293457,
+ "learning_rate": 1.9884754360258058e-05,
+ "loss": 1.5759,
+ "step": 2823
+ },
+ {
+ "epoch": 7.694822888283379,
+ "grad_norm": 6.429784774780273,
+ "learning_rate": 1.9884620728540035e-05,
+ "loss": 1.2266,
+ "step": 2824
+ },
+ {
+ "epoch": 7.697547683923705,
+ "grad_norm": 7.330230712890625,
+ "learning_rate": 1.9884487019840907e-05,
+ "loss": 1.379,
+ "step": 2825
+ },
+ {
+ "epoch": 7.700272479564033,
+ "grad_norm": 7.753212928771973,
+ "learning_rate": 1.9884353234161713e-05,
+ "loss": 1.3203,
+ "step": 2826
+ },
+ {
+ "epoch": 7.70299727520436,
+ "grad_norm": 6.241684913635254,
+ "learning_rate": 1.988421937150349e-05,
+ "loss": 1.1987,
+ "step": 2827
+ },
+ {
+ "epoch": 7.705722070844686,
+ "grad_norm": 30.249324798583984,
+ "learning_rate": 1.9884085431867287e-05,
+ "loss": 1.2871,
+ "step": 2828
+ },
+ {
+ "epoch": 7.708446866485014,
+ "grad_norm": 8.727538108825684,
+ "learning_rate": 1.9883951415254146e-05,
+ "loss": 1.3181,
+ "step": 2829
+ },
+ {
+ "epoch": 7.7111716621253406,
+ "grad_norm": 7.481237888336182,
+ "learning_rate": 1.988381732166511e-05,
+ "loss": 1.2246,
+ "step": 2830
+ },
+ {
+ "epoch": 7.713896457765667,
+ "grad_norm": 6.977359771728516,
+ "learning_rate": 1.988368315110122e-05,
+ "loss": 1.2551,
+ "step": 2831
+ },
+ {
+ "epoch": 7.716621253405995,
+ "grad_norm": 11.25141716003418,
+ "learning_rate": 1.9883548903563525e-05,
+ "loss": 1.3442,
+ "step": 2832
+ },
+ {
+ "epoch": 7.7193460490463215,
+ "grad_norm": 10.669550895690918,
+ "learning_rate": 1.988341457905307e-05,
+ "loss": 1.3367,
+ "step": 2833
+ },
+ {
+ "epoch": 7.722070844686648,
+ "grad_norm": 10.289504051208496,
+ "learning_rate": 1.9883280177570898e-05,
+ "loss": 1.6221,
+ "step": 2834
+ },
+ {
+ "epoch": 7.724795640326976,
+ "grad_norm": 7.420779705047607,
+ "learning_rate": 1.988314569911806e-05,
+ "loss": 1.2251,
+ "step": 2835
+ },
+ {
+ "epoch": 7.727520435967302,
+ "grad_norm": 6.875089645385742,
+ "learning_rate": 1.98830111436956e-05,
+ "loss": 1.3071,
+ "step": 2836
+ },
+ {
+ "epoch": 7.730245231607629,
+ "grad_norm": 8.199589729309082,
+ "learning_rate": 1.988287651130457e-05,
+ "loss": 1.6548,
+ "step": 2837
+ },
+ {
+ "epoch": 7.732970027247957,
+ "grad_norm": 14.022754669189453,
+ "learning_rate": 1.9882741801946014e-05,
+ "loss": 1.3164,
+ "step": 2838
+ },
+ {
+ "epoch": 7.735694822888283,
+ "grad_norm": 9.068694114685059,
+ "learning_rate": 1.988260701562098e-05,
+ "loss": 1.5554,
+ "step": 2839
+ },
+ {
+ "epoch": 7.73841961852861,
+ "grad_norm": 8.005928993225098,
+ "learning_rate": 1.9882472152330522e-05,
+ "loss": 1.3862,
+ "step": 2840
+ },
+ {
+ "epoch": 7.741144414168938,
+ "grad_norm": 6.632816314697266,
+ "learning_rate": 1.9882337212075693e-05,
+ "loss": 1.2141,
+ "step": 2841
+ },
+ {
+ "epoch": 7.743869209809264,
+ "grad_norm": 9.634324073791504,
+ "learning_rate": 1.9882202194857537e-05,
+ "loss": 1.4153,
+ "step": 2842
+ },
+ {
+ "epoch": 7.746594005449591,
+ "grad_norm": 8.650710105895996,
+ "learning_rate": 1.988206710067711e-05,
+ "loss": 1.3193,
+ "step": 2843
+ },
+ {
+ "epoch": 7.7493188010899186,
+ "grad_norm": 6.179229736328125,
+ "learning_rate": 1.9881931929535462e-05,
+ "loss": 1.2681,
+ "step": 2844
+ },
+ {
+ "epoch": 7.752043596730245,
+ "grad_norm": 13.732995986938477,
+ "learning_rate": 1.988179668143364e-05,
+ "loss": 1.4353,
+ "step": 2845
+ },
+ {
+ "epoch": 7.754768392370572,
+ "grad_norm": 24.489669799804688,
+ "learning_rate": 1.988166135637271e-05,
+ "loss": 1.4338,
+ "step": 2846
+ },
+ {
+ "epoch": 7.7574931880108995,
+ "grad_norm": 8.121292114257812,
+ "learning_rate": 1.988152595435372e-05,
+ "loss": 1.6597,
+ "step": 2847
+ },
+ {
+ "epoch": 7.760217983651226,
+ "grad_norm": 10.086445808410645,
+ "learning_rate": 1.988139047537772e-05,
+ "loss": 1.395,
+ "step": 2848
+ },
+ {
+ "epoch": 7.762942779291553,
+ "grad_norm": 7.253067970275879,
+ "learning_rate": 1.9881254919445773e-05,
+ "loss": 1.4075,
+ "step": 2849
+ },
+ {
+ "epoch": 7.76566757493188,
+ "grad_norm": 7.581152439117432,
+ "learning_rate": 1.9881119286558927e-05,
+ "loss": 1.3621,
+ "step": 2850
+ },
+ {
+ "epoch": 7.768392370572207,
+ "grad_norm": 7.197292804718018,
+ "learning_rate": 1.9880983576718247e-05,
+ "loss": 1.4368,
+ "step": 2851
+ },
+ {
+ "epoch": 7.771117166212534,
+ "grad_norm": 6.693124294281006,
+ "learning_rate": 1.988084778992478e-05,
+ "loss": 1.1824,
+ "step": 2852
+ },
+ {
+ "epoch": 7.773841961852861,
+ "grad_norm": 8.53558349609375,
+ "learning_rate": 1.988071192617959e-05,
+ "loss": 1.5376,
+ "step": 2853
+ },
+ {
+ "epoch": 7.776566757493188,
+ "grad_norm": 9.388649940490723,
+ "learning_rate": 1.9880575985483734e-05,
+ "loss": 1.5547,
+ "step": 2854
+ },
+ {
+ "epoch": 7.779291553133515,
+ "grad_norm": 7.268526554107666,
+ "learning_rate": 1.988043996783827e-05,
+ "loss": 1.1875,
+ "step": 2855
+ },
+ {
+ "epoch": 7.782016348773842,
+ "grad_norm": 8.347193717956543,
+ "learning_rate": 1.9880303873244258e-05,
+ "loss": 1.5359,
+ "step": 2856
+ },
+ {
+ "epoch": 7.784741144414169,
+ "grad_norm": 8.753171920776367,
+ "learning_rate": 1.9880167701702757e-05,
+ "loss": 1.4231,
+ "step": 2857
+ },
+ {
+ "epoch": 7.787465940054496,
+ "grad_norm": 8.244034767150879,
+ "learning_rate": 1.988003145321483e-05,
+ "loss": 1.5846,
+ "step": 2858
+ },
+ {
+ "epoch": 7.790190735694823,
+ "grad_norm": 6.692843914031982,
+ "learning_rate": 1.9879895127781533e-05,
+ "loss": 1.2046,
+ "step": 2859
+ },
+ {
+ "epoch": 7.79291553133515,
+ "grad_norm": 6.613217830657959,
+ "learning_rate": 1.987975872540393e-05,
+ "loss": 1.438,
+ "step": 2860
+ },
+ {
+ "epoch": 7.795640326975477,
+ "grad_norm": 6.128908634185791,
+ "learning_rate": 1.9879622246083087e-05,
+ "loss": 1.1636,
+ "step": 2861
+ },
+ {
+ "epoch": 7.798365122615804,
+ "grad_norm": 6.825357913970947,
+ "learning_rate": 1.987948568982006e-05,
+ "loss": 1.4036,
+ "step": 2862
+ },
+ {
+ "epoch": 7.801089918256131,
+ "grad_norm": 7.084530830383301,
+ "learning_rate": 1.987934905661592e-05,
+ "loss": 1.2964,
+ "step": 2863
+ },
+ {
+ "epoch": 7.8038147138964575,
+ "grad_norm": 7.33847713470459,
+ "learning_rate": 1.9879212346471726e-05,
+ "loss": 1.5413,
+ "step": 2864
+ },
+ {
+ "epoch": 7.806539509536785,
+ "grad_norm": 8.678008079528809,
+ "learning_rate": 1.9879075559388547e-05,
+ "loss": 1.3032,
+ "step": 2865
+ },
+ {
+ "epoch": 7.809264305177112,
+ "grad_norm": 8.071329116821289,
+ "learning_rate": 1.9878938695367444e-05,
+ "loss": 1.2622,
+ "step": 2866
+ },
+ {
+ "epoch": 7.8119891008174385,
+ "grad_norm": 6.420267105102539,
+ "learning_rate": 1.987880175440948e-05,
+ "loss": 1.187,
+ "step": 2867
+ },
+ {
+ "epoch": 7.814713896457766,
+ "grad_norm": 7.070373058319092,
+ "learning_rate": 1.9878664736515732e-05,
+ "loss": 1.4448,
+ "step": 2868
+ },
+ {
+ "epoch": 7.817438692098093,
+ "grad_norm": 6.533475399017334,
+ "learning_rate": 1.9878527641687255e-05,
+ "loss": 1.3079,
+ "step": 2869
+ },
+ {
+ "epoch": 7.820163487738419,
+ "grad_norm": 9.800215721130371,
+ "learning_rate": 1.987839046992513e-05,
+ "loss": 1.3243,
+ "step": 2870
+ },
+ {
+ "epoch": 7.822888283378747,
+ "grad_norm": 12.426836967468262,
+ "learning_rate": 1.9878253221230414e-05,
+ "loss": 1.6714,
+ "step": 2871
+ },
+ {
+ "epoch": 7.825613079019074,
+ "grad_norm": 7.1756367683410645,
+ "learning_rate": 1.9878115895604178e-05,
+ "loss": 1.2466,
+ "step": 2872
+ },
+ {
+ "epoch": 7.8283378746594,
+ "grad_norm": 8.547040939331055,
+ "learning_rate": 1.9877978493047496e-05,
+ "loss": 1.1387,
+ "step": 2873
+ },
+ {
+ "epoch": 7.831062670299728,
+ "grad_norm": 6.291370868682861,
+ "learning_rate": 1.987784101356143e-05,
+ "loss": 1.3657,
+ "step": 2874
+ },
+ {
+ "epoch": 7.833787465940055,
+ "grad_norm": 6.968871116638184,
+ "learning_rate": 1.9877703457147062e-05,
+ "loss": 1.3291,
+ "step": 2875
+ },
+ {
+ "epoch": 7.836512261580381,
+ "grad_norm": 7.6342668533325195,
+ "learning_rate": 1.987756582380546e-05,
+ "loss": 1.3236,
+ "step": 2876
+ },
+ {
+ "epoch": 7.839237057220709,
+ "grad_norm": 20.787864685058594,
+ "learning_rate": 1.9877428113537686e-05,
+ "loss": 1.4573,
+ "step": 2877
+ },
+ {
+ "epoch": 7.8419618528610355,
+ "grad_norm": 11.896600723266602,
+ "learning_rate": 1.9877290326344826e-05,
+ "loss": 1.366,
+ "step": 2878
+ },
+ {
+ "epoch": 7.844686648501362,
+ "grad_norm": 8.875763893127441,
+ "learning_rate": 1.987715246222794e-05,
+ "loss": 1.4983,
+ "step": 2879
+ },
+ {
+ "epoch": 7.84741144414169,
+ "grad_norm": 24.83260726928711,
+ "learning_rate": 1.9877014521188115e-05,
+ "loss": 1.4463,
+ "step": 2880
+ },
+ {
+ "epoch": 7.8501362397820165,
+ "grad_norm": 10.134485244750977,
+ "learning_rate": 1.9876876503226418e-05,
+ "loss": 1.384,
+ "step": 2881
+ },
+ {
+ "epoch": 7.852861035422343,
+ "grad_norm": 7.006833076477051,
+ "learning_rate": 1.987673840834392e-05,
+ "loss": 1.2415,
+ "step": 2882
+ },
+ {
+ "epoch": 7.855585831062671,
+ "grad_norm": 8.083553314208984,
+ "learning_rate": 1.9876600236541705e-05,
+ "loss": 1.2932,
+ "step": 2883
+ },
+ {
+ "epoch": 7.858310626702997,
+ "grad_norm": 8.806413650512695,
+ "learning_rate": 1.9876461987820847e-05,
+ "loss": 1.4778,
+ "step": 2884
+ },
+ {
+ "epoch": 7.861035422343324,
+ "grad_norm": 5.902005195617676,
+ "learning_rate": 1.9876323662182417e-05,
+ "loss": 1.2515,
+ "step": 2885
+ },
+ {
+ "epoch": 7.863760217983652,
+ "grad_norm": 8.51309585571289,
+ "learning_rate": 1.9876185259627498e-05,
+ "loss": 1.5347,
+ "step": 2886
+ },
+ {
+ "epoch": 7.866485013623978,
+ "grad_norm": 6.45297384262085,
+ "learning_rate": 1.987604678015716e-05,
+ "loss": 1.3887,
+ "step": 2887
+ },
+ {
+ "epoch": 7.869209809264305,
+ "grad_norm": 10.144701957702637,
+ "learning_rate": 1.9875908223772497e-05,
+ "loss": 1.4268,
+ "step": 2888
+ },
+ {
+ "epoch": 7.871934604904633,
+ "grad_norm": 6.433525562286377,
+ "learning_rate": 1.9875769590474574e-05,
+ "loss": 1.1702,
+ "step": 2889
+ },
+ {
+ "epoch": 7.874659400544959,
+ "grad_norm": 7.332644462585449,
+ "learning_rate": 1.9875630880264474e-05,
+ "loss": 1.6685,
+ "step": 2890
+ },
+ {
+ "epoch": 7.877384196185286,
+ "grad_norm": 9.220657348632812,
+ "learning_rate": 1.987549209314328e-05,
+ "loss": 1.3511,
+ "step": 2891
+ },
+ {
+ "epoch": 7.8801089918256135,
+ "grad_norm": 10.015419006347656,
+ "learning_rate": 1.9875353229112076e-05,
+ "loss": 1.4231,
+ "step": 2892
+ },
+ {
+ "epoch": 7.88283378746594,
+ "grad_norm": 8.237641334533691,
+ "learning_rate": 1.9875214288171932e-05,
+ "loss": 1.1953,
+ "step": 2893
+ },
+ {
+ "epoch": 7.885558583106267,
+ "grad_norm": 7.453221797943115,
+ "learning_rate": 1.9875075270323942e-05,
+ "loss": 1.3694,
+ "step": 2894
+ },
+ {
+ "epoch": 7.8882833787465945,
+ "grad_norm": 8.641449928283691,
+ "learning_rate": 1.9874936175569178e-05,
+ "loss": 1.4167,
+ "step": 2895
+ },
+ {
+ "epoch": 7.891008174386921,
+ "grad_norm": 10.859975814819336,
+ "learning_rate": 1.9874797003908735e-05,
+ "loss": 1.4412,
+ "step": 2896
+ },
+ {
+ "epoch": 7.893732970027248,
+ "grad_norm": 6.654394149780273,
+ "learning_rate": 1.9874657755343688e-05,
+ "loss": 1.3091,
+ "step": 2897
+ },
+ {
+ "epoch": 7.896457765667575,
+ "grad_norm": 12.730084419250488,
+ "learning_rate": 1.9874518429875126e-05,
+ "loss": 1.334,
+ "step": 2898
+ },
+ {
+ "epoch": 7.899182561307902,
+ "grad_norm": 6.202551364898682,
+ "learning_rate": 1.987437902750413e-05,
+ "loss": 1.1465,
+ "step": 2899
+ },
+ {
+ "epoch": 7.901907356948229,
+ "grad_norm": 15.460163116455078,
+ "learning_rate": 1.9874239548231786e-05,
+ "loss": 1.2944,
+ "step": 2900
+ },
+ {
+ "epoch": 7.904632152588556,
+ "grad_norm": 8.967248916625977,
+ "learning_rate": 1.9874099992059185e-05,
+ "loss": 1.6355,
+ "step": 2901
+ },
+ {
+ "epoch": 7.907356948228883,
+ "grad_norm": 7.932471752166748,
+ "learning_rate": 1.9873960358987414e-05,
+ "loss": 1.2759,
+ "step": 2902
+ },
+ {
+ "epoch": 7.91008174386921,
+ "grad_norm": 6.513246536254883,
+ "learning_rate": 1.9873820649017553e-05,
+ "loss": 1.334,
+ "step": 2903
+ },
+ {
+ "epoch": 7.912806539509537,
+ "grad_norm": 6.791923522949219,
+ "learning_rate": 1.9873680862150694e-05,
+ "loss": 1.3655,
+ "step": 2904
+ },
+ {
+ "epoch": 7.915531335149864,
+ "grad_norm": 12.07205867767334,
+ "learning_rate": 1.9873540998387928e-05,
+ "loss": 1.3091,
+ "step": 2905
+ },
+ {
+ "epoch": 7.918256130790191,
+ "grad_norm": 9.107305526733398,
+ "learning_rate": 1.9873401057730345e-05,
+ "loss": 1.2517,
+ "step": 2906
+ },
+ {
+ "epoch": 7.920980926430518,
+ "grad_norm": 12.882670402526855,
+ "learning_rate": 1.987326104017903e-05,
+ "loss": 1.3999,
+ "step": 2907
+ },
+ {
+ "epoch": 7.923705722070845,
+ "grad_norm": 43.50045394897461,
+ "learning_rate": 1.9873120945735073e-05,
+ "loss": 1.7263,
+ "step": 2908
+ },
+ {
+ "epoch": 7.926430517711172,
+ "grad_norm": 7.436546802520752,
+ "learning_rate": 1.9872980774399573e-05,
+ "loss": 1.333,
+ "step": 2909
+ },
+ {
+ "epoch": 7.929155313351498,
+ "grad_norm": 6.836589336395264,
+ "learning_rate": 1.987284052617361e-05,
+ "loss": 1.1501,
+ "step": 2910
+ },
+ {
+ "epoch": 7.931880108991826,
+ "grad_norm": 13.878046035766602,
+ "learning_rate": 1.9872700201058287e-05,
+ "loss": 1.3772,
+ "step": 2911
+ },
+ {
+ "epoch": 7.9346049046321525,
+ "grad_norm": 8.241393089294434,
+ "learning_rate": 1.9872559799054695e-05,
+ "loss": 1.4651,
+ "step": 2912
+ },
+ {
+ "epoch": 7.937329700272479,
+ "grad_norm": 8.951183319091797,
+ "learning_rate": 1.9872419320163924e-05,
+ "loss": 1.4827,
+ "step": 2913
+ },
+ {
+ "epoch": 7.940054495912807,
+ "grad_norm": 15.275472640991211,
+ "learning_rate": 1.9872278764387064e-05,
+ "loss": 1.2002,
+ "step": 2914
+ },
+ {
+ "epoch": 7.9427792915531334,
+ "grad_norm": 9.016220092773438,
+ "learning_rate": 1.987213813172522e-05,
+ "loss": 1.3218,
+ "step": 2915
+ },
+ {
+ "epoch": 7.94550408719346,
+ "grad_norm": 7.204616546630859,
+ "learning_rate": 1.987199742217948e-05,
+ "loss": 1.25,
+ "step": 2916
+ },
+ {
+ "epoch": 7.948228882833788,
+ "grad_norm": 6.835882663726807,
+ "learning_rate": 1.987185663575094e-05,
+ "loss": 1.2004,
+ "step": 2917
+ },
+ {
+ "epoch": 7.950953678474114,
+ "grad_norm": 13.520551681518555,
+ "learning_rate": 1.98717157724407e-05,
+ "loss": 1.2512,
+ "step": 2918
+ },
+ {
+ "epoch": 7.953678474114441,
+ "grad_norm": 7.039772987365723,
+ "learning_rate": 1.9871574832249858e-05,
+ "loss": 1.1721,
+ "step": 2919
+ },
+ {
+ "epoch": 7.956403269754769,
+ "grad_norm": 9.12216567993164,
+ "learning_rate": 1.9871433815179505e-05,
+ "loss": 1.2219,
+ "step": 2920
+ },
+ {
+ "epoch": 7.959128065395095,
+ "grad_norm": 7.779405117034912,
+ "learning_rate": 1.9871292721230745e-05,
+ "loss": 1.1235,
+ "step": 2921
+ },
+ {
+ "epoch": 7.961852861035422,
+ "grad_norm": 6.2565016746521,
+ "learning_rate": 1.9871151550404675e-05,
+ "loss": 1.3604,
+ "step": 2922
+ },
+ {
+ "epoch": 7.96457765667575,
+ "grad_norm": 14.671350479125977,
+ "learning_rate": 1.9871010302702393e-05,
+ "loss": 1.4851,
+ "step": 2923
+ },
+ {
+ "epoch": 7.967302452316076,
+ "grad_norm": 8.147441864013672,
+ "learning_rate": 1.9870868978125003e-05,
+ "loss": 1.4617,
+ "step": 2924
+ },
+ {
+ "epoch": 7.970027247956403,
+ "grad_norm": 6.774234771728516,
+ "learning_rate": 1.98707275766736e-05,
+ "loss": 1.2744,
+ "step": 2925
+ },
+ {
+ "epoch": 7.9727520435967305,
+ "grad_norm": 6.3735127449035645,
+ "learning_rate": 1.9870586098349288e-05,
+ "loss": 1.3147,
+ "step": 2926
+ },
+ {
+ "epoch": 7.975476839237057,
+ "grad_norm": 5.340735912322998,
+ "learning_rate": 1.987044454315317e-05,
+ "loss": 1.3677,
+ "step": 2927
+ },
+ {
+ "epoch": 7.978201634877384,
+ "grad_norm": 6.6280083656311035,
+ "learning_rate": 1.987030291108635e-05,
+ "loss": 1.4351,
+ "step": 2928
+ },
+ {
+ "epoch": 7.9809264305177114,
+ "grad_norm": 6.613189220428467,
+ "learning_rate": 1.987016120214993e-05,
+ "loss": 1.4902,
+ "step": 2929
+ },
+ {
+ "epoch": 7.983651226158038,
+ "grad_norm": 7.308218002319336,
+ "learning_rate": 1.9870019416345008e-05,
+ "loss": 1.5923,
+ "step": 2930
+ },
+ {
+ "epoch": 7.986376021798365,
+ "grad_norm": 6.508357524871826,
+ "learning_rate": 1.9869877553672693e-05,
+ "loss": 1.2554,
+ "step": 2931
+ },
+ {
+ "epoch": 7.989100817438692,
+ "grad_norm": 18.815410614013672,
+ "learning_rate": 1.9869735614134092e-05,
+ "loss": 1.54,
+ "step": 2932
+ },
+ {
+ "epoch": 7.991825613079019,
+ "grad_norm": 10.382793426513672,
+ "learning_rate": 1.9869593597730305e-05,
+ "loss": 1.2286,
+ "step": 2933
+ },
+ {
+ "epoch": 7.994550408719346,
+ "grad_norm": 6.455565929412842,
+ "learning_rate": 1.986945150446244e-05,
+ "loss": 1.071,
+ "step": 2934
+ },
+ {
+ "epoch": 7.997275204359673,
+ "grad_norm": 7.475588321685791,
+ "learning_rate": 1.9869309334331606e-05,
+ "loss": 1.6025,
+ "step": 2935
+ },
+ {
+ "epoch": 8.0,
+ "grad_norm": 8.586991310119629,
+ "learning_rate": 1.9869167087338908e-05,
+ "loss": 1.4749,
+ "step": 2936
+ },
+ {
+ "epoch": 8.002724795640328,
+ "grad_norm": 5.768120288848877,
+ "learning_rate": 1.9869024763485454e-05,
+ "loss": 1.3406,
+ "step": 2937
+ },
+ {
+ "epoch": 8.005449591280653,
+ "grad_norm": 7.407169818878174,
+ "learning_rate": 1.9868882362772356e-05,
+ "loss": 1.2592,
+ "step": 2938
+ },
+ {
+ "epoch": 8.008174386920981,
+ "grad_norm": 5.705663681030273,
+ "learning_rate": 1.9868739885200718e-05,
+ "loss": 1.5059,
+ "step": 2939
+ },
+ {
+ "epoch": 8.010899182561309,
+ "grad_norm": 7.680881023406982,
+ "learning_rate": 1.9868597330771648e-05,
+ "loss": 1.4463,
+ "step": 2940
+ },
+ {
+ "epoch": 8.013623978201634,
+ "grad_norm": 5.735972881317139,
+ "learning_rate": 1.986845469948626e-05,
+ "loss": 1.2146,
+ "step": 2941
+ },
+ {
+ "epoch": 8.016348773841962,
+ "grad_norm": 7.844153881072998,
+ "learning_rate": 1.986831199134567e-05,
+ "loss": 1.1738,
+ "step": 2942
+ },
+ {
+ "epoch": 8.01907356948229,
+ "grad_norm": 6.359085559844971,
+ "learning_rate": 1.986816920635098e-05,
+ "loss": 1.2883,
+ "step": 2943
+ },
+ {
+ "epoch": 8.021798365122615,
+ "grad_norm": 6.8043341636657715,
+ "learning_rate": 1.9868026344503307e-05,
+ "loss": 1.1653,
+ "step": 2944
+ },
+ {
+ "epoch": 8.024523160762943,
+ "grad_norm": 7.508440017700195,
+ "learning_rate": 1.986788340580376e-05,
+ "loss": 1.0073,
+ "step": 2945
+ },
+ {
+ "epoch": 8.02724795640327,
+ "grad_norm": 6.768466472625732,
+ "learning_rate": 1.9867740390253458e-05,
+ "loss": 1.2103,
+ "step": 2946
+ },
+ {
+ "epoch": 8.029972752043596,
+ "grad_norm": 7.07720422744751,
+ "learning_rate": 1.986759729785351e-05,
+ "loss": 1.156,
+ "step": 2947
+ },
+ {
+ "epoch": 8.032697547683924,
+ "grad_norm": 6.73090124130249,
+ "learning_rate": 1.9867454128605033e-05,
+ "loss": 1.3374,
+ "step": 2948
+ },
+ {
+ "epoch": 8.035422343324251,
+ "grad_norm": 6.404378414154053,
+ "learning_rate": 1.9867310882509136e-05,
+ "loss": 1.1804,
+ "step": 2949
+ },
+ {
+ "epoch": 8.038147138964577,
+ "grad_norm": 6.880995273590088,
+ "learning_rate": 1.9867167559566944e-05,
+ "loss": 1.188,
+ "step": 2950
+ },
+ {
+ "epoch": 8.040871934604905,
+ "grad_norm": 5.8374924659729,
+ "learning_rate": 1.986702415977957e-05,
+ "loss": 1.1125,
+ "step": 2951
+ },
+ {
+ "epoch": 8.043596730245232,
+ "grad_norm": 5.849694728851318,
+ "learning_rate": 1.9866880683148128e-05,
+ "loss": 1.1589,
+ "step": 2952
+ },
+ {
+ "epoch": 8.046321525885558,
+ "grad_norm": 6.668493270874023,
+ "learning_rate": 1.9866737129673733e-05,
+ "loss": 1.1934,
+ "step": 2953
+ },
+ {
+ "epoch": 8.049046321525886,
+ "grad_norm": 10.446045875549316,
+ "learning_rate": 1.9866593499357508e-05,
+ "loss": 1.4453,
+ "step": 2954
+ },
+ {
+ "epoch": 8.051771117166213,
+ "grad_norm": 7.018033981323242,
+ "learning_rate": 1.9866449792200572e-05,
+ "loss": 1.2622,
+ "step": 2955
+ },
+ {
+ "epoch": 8.054495912806539,
+ "grad_norm": 7.068787574768066,
+ "learning_rate": 1.9866306008204042e-05,
+ "loss": 1.1117,
+ "step": 2956
+ },
+ {
+ "epoch": 8.057220708446867,
+ "grad_norm": 6.050174236297607,
+ "learning_rate": 1.986616214736904e-05,
+ "loss": 1.4575,
+ "step": 2957
+ },
+ {
+ "epoch": 8.059945504087194,
+ "grad_norm": 7.226597309112549,
+ "learning_rate": 1.9866018209696686e-05,
+ "loss": 1.6963,
+ "step": 2958
+ },
+ {
+ "epoch": 8.06267029972752,
+ "grad_norm": 7.827121734619141,
+ "learning_rate": 1.9865874195188095e-05,
+ "loss": 1.3118,
+ "step": 2959
+ },
+ {
+ "epoch": 8.065395095367847,
+ "grad_norm": 7.329259872436523,
+ "learning_rate": 1.9865730103844397e-05,
+ "loss": 1.3479,
+ "step": 2960
+ },
+ {
+ "epoch": 8.068119891008175,
+ "grad_norm": 9.836698532104492,
+ "learning_rate": 1.9865585935666706e-05,
+ "loss": 1.2512,
+ "step": 2961
+ },
+ {
+ "epoch": 8.0708446866485,
+ "grad_norm": 5.768904209136963,
+ "learning_rate": 1.9865441690656154e-05,
+ "loss": 1.0239,
+ "step": 2962
+ },
+ {
+ "epoch": 8.073569482288828,
+ "grad_norm": 6.831222057342529,
+ "learning_rate": 1.986529736881386e-05,
+ "loss": 1.1401,
+ "step": 2963
+ },
+ {
+ "epoch": 8.076294277929156,
+ "grad_norm": 10.426448822021484,
+ "learning_rate": 1.9865152970140946e-05,
+ "loss": 1.4255,
+ "step": 2964
+ },
+ {
+ "epoch": 8.079019073569482,
+ "grad_norm": 6.713279724121094,
+ "learning_rate": 1.9865008494638536e-05,
+ "loss": 1.3181,
+ "step": 2965
+ },
+ {
+ "epoch": 8.08174386920981,
+ "grad_norm": 7.251582622528076,
+ "learning_rate": 1.986486394230776e-05,
+ "loss": 1.1899,
+ "step": 2966
+ },
+ {
+ "epoch": 8.084468664850137,
+ "grad_norm": 6.9533538818359375,
+ "learning_rate": 1.986471931314974e-05,
+ "loss": 1.3518,
+ "step": 2967
+ },
+ {
+ "epoch": 8.087193460490463,
+ "grad_norm": 7.211970806121826,
+ "learning_rate": 1.9864574607165602e-05,
+ "loss": 1.3586,
+ "step": 2968
+ },
+ {
+ "epoch": 8.08991825613079,
+ "grad_norm": 7.899008750915527,
+ "learning_rate": 1.9864429824356478e-05,
+ "loss": 1.1912,
+ "step": 2969
+ },
+ {
+ "epoch": 8.092643051771118,
+ "grad_norm": 7.199990272521973,
+ "learning_rate": 1.9864284964723488e-05,
+ "loss": 1.25,
+ "step": 2970
+ },
+ {
+ "epoch": 8.095367847411444,
+ "grad_norm": 7.356903076171875,
+ "learning_rate": 1.9864140028267767e-05,
+ "loss": 1.5278,
+ "step": 2971
+ },
+ {
+ "epoch": 8.098092643051771,
+ "grad_norm": 6.798772811889648,
+ "learning_rate": 1.986399501499044e-05,
+ "loss": 1.2404,
+ "step": 2972
+ },
+ {
+ "epoch": 8.100817438692099,
+ "grad_norm": 6.381214141845703,
+ "learning_rate": 1.986384992489264e-05,
+ "loss": 1.1482,
+ "step": 2973
+ },
+ {
+ "epoch": 8.103542234332425,
+ "grad_norm": 6.320764541625977,
+ "learning_rate": 1.9863704757975492e-05,
+ "loss": 1.1101,
+ "step": 2974
+ },
+ {
+ "epoch": 8.106267029972752,
+ "grad_norm": 6.5356526374816895,
+ "learning_rate": 1.9863559514240128e-05,
+ "loss": 1.2888,
+ "step": 2975
+ },
+ {
+ "epoch": 8.10899182561308,
+ "grad_norm": 8.064299583435059,
+ "learning_rate": 1.986341419368768e-05,
+ "loss": 1.2512,
+ "step": 2976
+ },
+ {
+ "epoch": 8.111716621253406,
+ "grad_norm": 5.662990093231201,
+ "learning_rate": 1.9863268796319278e-05,
+ "loss": 1.1467,
+ "step": 2977
+ },
+ {
+ "epoch": 8.114441416893733,
+ "grad_norm": 8.11336898803711,
+ "learning_rate": 1.9863123322136058e-05,
+ "loss": 1.2795,
+ "step": 2978
+ },
+ {
+ "epoch": 8.11716621253406,
+ "grad_norm": 7.131218433380127,
+ "learning_rate": 1.9862977771139154e-05,
+ "loss": 1.2556,
+ "step": 2979
+ },
+ {
+ "epoch": 8.119891008174386,
+ "grad_norm": 8.011634826660156,
+ "learning_rate": 1.9862832143329694e-05,
+ "loss": 1.2859,
+ "step": 2980
+ },
+ {
+ "epoch": 8.122615803814714,
+ "grad_norm": 8.358835220336914,
+ "learning_rate": 1.9862686438708813e-05,
+ "loss": 1.1409,
+ "step": 2981
+ },
+ {
+ "epoch": 8.125340599455042,
+ "grad_norm": 8.611673355102539,
+ "learning_rate": 1.986254065727765e-05,
+ "loss": 1.088,
+ "step": 2982
+ },
+ {
+ "epoch": 8.128065395095367,
+ "grad_norm": 8.097973823547363,
+ "learning_rate": 1.9862394799037338e-05,
+ "loss": 1.2695,
+ "step": 2983
+ },
+ {
+ "epoch": 8.130790190735695,
+ "grad_norm": 9.139933586120605,
+ "learning_rate": 1.986224886398901e-05,
+ "loss": 1.3474,
+ "step": 2984
+ },
+ {
+ "epoch": 8.133514986376023,
+ "grad_norm": 8.094538688659668,
+ "learning_rate": 1.986210285213381e-05,
+ "loss": 1.2483,
+ "step": 2985
+ },
+ {
+ "epoch": 8.136239782016348,
+ "grad_norm": 6.038456916809082,
+ "learning_rate": 1.9861956763472866e-05,
+ "loss": 1.2332,
+ "step": 2986
+ },
+ {
+ "epoch": 8.138964577656676,
+ "grad_norm": 7.886776924133301,
+ "learning_rate": 1.986181059800732e-05,
+ "loss": 1.3191,
+ "step": 2987
+ },
+ {
+ "epoch": 8.141689373297003,
+ "grad_norm": 7.248572826385498,
+ "learning_rate": 1.986166435573831e-05,
+ "loss": 1.1704,
+ "step": 2988
+ },
+ {
+ "epoch": 8.14441416893733,
+ "grad_norm": 7.355432987213135,
+ "learning_rate": 1.9861518036666978e-05,
+ "loss": 1.3875,
+ "step": 2989
+ },
+ {
+ "epoch": 8.147138964577657,
+ "grad_norm": 6.016048431396484,
+ "learning_rate": 1.986137164079446e-05,
+ "loss": 1.1089,
+ "step": 2990
+ },
+ {
+ "epoch": 8.149863760217984,
+ "grad_norm": 5.9033966064453125,
+ "learning_rate": 1.98612251681219e-05,
+ "loss": 1.218,
+ "step": 2991
+ },
+ {
+ "epoch": 8.15258855585831,
+ "grad_norm": 6.114132881164551,
+ "learning_rate": 1.986107861865043e-05,
+ "loss": 1.252,
+ "step": 2992
+ },
+ {
+ "epoch": 8.155313351498638,
+ "grad_norm": 8.173006057739258,
+ "learning_rate": 1.9860931992381202e-05,
+ "loss": 1.1711,
+ "step": 2993
+ },
+ {
+ "epoch": 8.158038147138965,
+ "grad_norm": 6.308248996734619,
+ "learning_rate": 1.9860785289315352e-05,
+ "loss": 1.0472,
+ "step": 2994
+ },
+ {
+ "epoch": 8.160762942779291,
+ "grad_norm": 9.244794845581055,
+ "learning_rate": 1.986063850945402e-05,
+ "loss": 1.2808,
+ "step": 2995
+ },
+ {
+ "epoch": 8.163487738419619,
+ "grad_norm": 8.935945510864258,
+ "learning_rate": 1.9860491652798358e-05,
+ "loss": 1.3135,
+ "step": 2996
+ },
+ {
+ "epoch": 8.166212534059946,
+ "grad_norm": 8.755207061767578,
+ "learning_rate": 1.98603447193495e-05,
+ "loss": 1.3064,
+ "step": 2997
+ },
+ {
+ "epoch": 8.168937329700272,
+ "grad_norm": 6.95626974105835,
+ "learning_rate": 1.9860197709108596e-05,
+ "loss": 1.209,
+ "step": 2998
+ },
+ {
+ "epoch": 8.1716621253406,
+ "grad_norm": 6.643039703369141,
+ "learning_rate": 1.986005062207679e-05,
+ "loss": 1.1719,
+ "step": 2999
+ },
+ {
+ "epoch": 8.174386920980927,
+ "grad_norm": 7.07211446762085,
+ "learning_rate": 1.9859903458255227e-05,
+ "loss": 1.3521,
+ "step": 3000
+ },
+ {
+ "epoch": 8.177111716621253,
+ "grad_norm": 7.402812957763672,
+ "learning_rate": 1.9859756217645054e-05,
+ "loss": 1.1826,
+ "step": 3001
+ },
+ {
+ "epoch": 8.17983651226158,
+ "grad_norm": 8.55255126953125,
+ "learning_rate": 1.9859608900247417e-05,
+ "loss": 1.2236,
+ "step": 3002
+ },
+ {
+ "epoch": 8.182561307901908,
+ "grad_norm": 8.00808048248291,
+ "learning_rate": 1.985946150606346e-05,
+ "loss": 1.2607,
+ "step": 3003
+ },
+ {
+ "epoch": 8.185286103542234,
+ "grad_norm": 6.262120723724365,
+ "learning_rate": 1.985931403509434e-05,
+ "loss": 1.2266,
+ "step": 3004
+ },
+ {
+ "epoch": 8.188010899182562,
+ "grad_norm": 6.980482578277588,
+ "learning_rate": 1.9859166487341194e-05,
+ "loss": 1.2206,
+ "step": 3005
+ },
+ {
+ "epoch": 8.190735694822889,
+ "grad_norm": 6.359993934631348,
+ "learning_rate": 1.985901886280518e-05,
+ "loss": 1.073,
+ "step": 3006
+ },
+ {
+ "epoch": 8.193460490463215,
+ "grad_norm": 8.098401069641113,
+ "learning_rate": 1.985887116148744e-05,
+ "loss": 1.3821,
+ "step": 3007
+ },
+ {
+ "epoch": 8.196185286103542,
+ "grad_norm": 6.095962047576904,
+ "learning_rate": 1.9858723383389134e-05,
+ "loss": 1.0132,
+ "step": 3008
+ },
+ {
+ "epoch": 8.19891008174387,
+ "grad_norm": 10.081521987915039,
+ "learning_rate": 1.9858575528511406e-05,
+ "loss": 1.2812,
+ "step": 3009
+ },
+ {
+ "epoch": 8.201634877384196,
+ "grad_norm": 7.564338207244873,
+ "learning_rate": 1.9858427596855407e-05,
+ "loss": 1.3091,
+ "step": 3010
+ },
+ {
+ "epoch": 8.204359673024523,
+ "grad_norm": 7.901644229888916,
+ "learning_rate": 1.9858279588422294e-05,
+ "loss": 1.1772,
+ "step": 3011
+ },
+ {
+ "epoch": 8.207084468664851,
+ "grad_norm": 7.364603519439697,
+ "learning_rate": 1.9858131503213214e-05,
+ "loss": 1.3472,
+ "step": 3012
+ },
+ {
+ "epoch": 8.209809264305177,
+ "grad_norm": 34.25517654418945,
+ "learning_rate": 1.9857983341229326e-05,
+ "loss": 1.4805,
+ "step": 3013
+ },
+ {
+ "epoch": 8.212534059945504,
+ "grad_norm": 6.624706268310547,
+ "learning_rate": 1.985783510247178e-05,
+ "loss": 1.2981,
+ "step": 3014
+ },
+ {
+ "epoch": 8.215258855585832,
+ "grad_norm": 8.078665733337402,
+ "learning_rate": 1.9857686786941733e-05,
+ "loss": 1.2886,
+ "step": 3015
+ },
+ {
+ "epoch": 8.217983651226158,
+ "grad_norm": 8.067582130432129,
+ "learning_rate": 1.9857538394640336e-05,
+ "loss": 1.1934,
+ "step": 3016
+ },
+ {
+ "epoch": 8.220708446866485,
+ "grad_norm": 7.0521697998046875,
+ "learning_rate": 1.9857389925568745e-05,
+ "loss": 1.3357,
+ "step": 3017
+ },
+ {
+ "epoch": 8.223433242506813,
+ "grad_norm": 8.329730033874512,
+ "learning_rate": 1.9857241379728122e-05,
+ "loss": 1.2659,
+ "step": 3018
+ },
+ {
+ "epoch": 8.226158038147139,
+ "grad_norm": 7.6741509437561035,
+ "learning_rate": 1.9857092757119622e-05,
+ "loss": 1.2891,
+ "step": 3019
+ },
+ {
+ "epoch": 8.228882833787466,
+ "grad_norm": 8.11025333404541,
+ "learning_rate": 1.9856944057744398e-05,
+ "loss": 1.3245,
+ "step": 3020
+ },
+ {
+ "epoch": 8.231607629427794,
+ "grad_norm": 6.626626968383789,
+ "learning_rate": 1.985679528160361e-05,
+ "loss": 1.2791,
+ "step": 3021
+ },
+ {
+ "epoch": 8.23433242506812,
+ "grad_norm": 6.376816272735596,
+ "learning_rate": 1.985664642869842e-05,
+ "loss": 1.1323,
+ "step": 3022
+ },
+ {
+ "epoch": 8.237057220708447,
+ "grad_norm": 6.568450927734375,
+ "learning_rate": 1.9856497499029987e-05,
+ "loss": 1.1384,
+ "step": 3023
+ },
+ {
+ "epoch": 8.239782016348773,
+ "grad_norm": 8.474916458129883,
+ "learning_rate": 1.9856348492599464e-05,
+ "loss": 1.2402,
+ "step": 3024
+ },
+ {
+ "epoch": 8.2425068119891,
+ "grad_norm": 7.144291877746582,
+ "learning_rate": 1.9856199409408017e-05,
+ "loss": 1.0488,
+ "step": 3025
+ },
+ {
+ "epoch": 8.245231607629428,
+ "grad_norm": 8.898880004882812,
+ "learning_rate": 1.9856050249456808e-05,
+ "loss": 1.1816,
+ "step": 3026
+ },
+ {
+ "epoch": 8.247956403269754,
+ "grad_norm": 8.124098777770996,
+ "learning_rate": 1.9855901012746996e-05,
+ "loss": 1.4033,
+ "step": 3027
+ },
+ {
+ "epoch": 8.250681198910081,
+ "grad_norm": 7.890181064605713,
+ "learning_rate": 1.9855751699279745e-05,
+ "loss": 1.3364,
+ "step": 3028
+ },
+ {
+ "epoch": 8.253405994550409,
+ "grad_norm": 8.375751495361328,
+ "learning_rate": 1.9855602309056216e-05,
+ "loss": 1.2546,
+ "step": 3029
+ },
+ {
+ "epoch": 8.256130790190735,
+ "grad_norm": 6.659442901611328,
+ "learning_rate": 1.9855452842077573e-05,
+ "loss": 1.4692,
+ "step": 3030
+ },
+ {
+ "epoch": 8.258855585831062,
+ "grad_norm": 7.482369422912598,
+ "learning_rate": 1.9855303298344985e-05,
+ "loss": 1.1014,
+ "step": 3031
+ },
+ {
+ "epoch": 8.26158038147139,
+ "grad_norm": 6.484124183654785,
+ "learning_rate": 1.9855153677859604e-05,
+ "loss": 1.3389,
+ "step": 3032
+ },
+ {
+ "epoch": 8.264305177111716,
+ "grad_norm": 7.5051798820495605,
+ "learning_rate": 1.985500398062261e-05,
+ "loss": 1.1819,
+ "step": 3033
+ },
+ {
+ "epoch": 8.267029972752043,
+ "grad_norm": 6.459608554840088,
+ "learning_rate": 1.985485420663516e-05,
+ "loss": 1.2311,
+ "step": 3034
+ },
+ {
+ "epoch": 8.269754768392371,
+ "grad_norm": 7.402273654937744,
+ "learning_rate": 1.985470435589842e-05,
+ "loss": 1.1426,
+ "step": 3035
+ },
+ {
+ "epoch": 8.272479564032697,
+ "grad_norm": 6.756234645843506,
+ "learning_rate": 1.9854554428413563e-05,
+ "loss": 1.2069,
+ "step": 3036
+ },
+ {
+ "epoch": 8.275204359673024,
+ "grad_norm": 6.9964189529418945,
+ "learning_rate": 1.9854404424181752e-05,
+ "loss": 1.1497,
+ "step": 3037
+ },
+ {
+ "epoch": 8.277929155313352,
+ "grad_norm": 8.75035285949707,
+ "learning_rate": 1.9854254343204155e-05,
+ "loss": 1.1406,
+ "step": 3038
+ },
+ {
+ "epoch": 8.280653950953678,
+ "grad_norm": 6.984670162200928,
+ "learning_rate": 1.9854104185481944e-05,
+ "loss": 1.1865,
+ "step": 3039
+ },
+ {
+ "epoch": 8.283378746594005,
+ "grad_norm": 9.734557151794434,
+ "learning_rate": 1.985395395101629e-05,
+ "loss": 1.2793,
+ "step": 3040
+ },
+ {
+ "epoch": 8.286103542234333,
+ "grad_norm": 7.413298606872559,
+ "learning_rate": 1.9853803639808357e-05,
+ "loss": 1.175,
+ "step": 3041
+ },
+ {
+ "epoch": 8.288828337874659,
+ "grad_norm": 8.552112579345703,
+ "learning_rate": 1.9853653251859315e-05,
+ "loss": 1.2483,
+ "step": 3042
+ },
+ {
+ "epoch": 8.291553133514986,
+ "grad_norm": 7.219217300415039,
+ "learning_rate": 1.9853502787170345e-05,
+ "loss": 1.2419,
+ "step": 3043
+ },
+ {
+ "epoch": 8.294277929155314,
+ "grad_norm": 6.75034236907959,
+ "learning_rate": 1.985335224574261e-05,
+ "loss": 1.208,
+ "step": 3044
+ },
+ {
+ "epoch": 8.29700272479564,
+ "grad_norm": 27.292917251586914,
+ "learning_rate": 1.9853201627577283e-05,
+ "loss": 1.1841,
+ "step": 3045
+ },
+ {
+ "epoch": 8.299727520435967,
+ "grad_norm": 12.455243110656738,
+ "learning_rate": 1.985305093267554e-05,
+ "loss": 1.301,
+ "step": 3046
+ },
+ {
+ "epoch": 8.302452316076295,
+ "grad_norm": 6.868085861206055,
+ "learning_rate": 1.9852900161038552e-05,
+ "loss": 1.208,
+ "step": 3047
+ },
+ {
+ "epoch": 8.30517711171662,
+ "grad_norm": 5.608100891113281,
+ "learning_rate": 1.9852749312667496e-05,
+ "loss": 1.1272,
+ "step": 3048
+ },
+ {
+ "epoch": 8.307901907356948,
+ "grad_norm": 7.861515998840332,
+ "learning_rate": 1.9852598387563545e-05,
+ "loss": 1.1997,
+ "step": 3049
+ },
+ {
+ "epoch": 8.310626702997276,
+ "grad_norm": 6.082080841064453,
+ "learning_rate": 1.9852447385727876e-05,
+ "loss": 1.168,
+ "step": 3050
+ },
+ {
+ "epoch": 8.313351498637601,
+ "grad_norm": 11.103292465209961,
+ "learning_rate": 1.985229630716166e-05,
+ "loss": 1.3933,
+ "step": 3051
+ },
+ {
+ "epoch": 8.316076294277929,
+ "grad_norm": 8.117807388305664,
+ "learning_rate": 1.985214515186608e-05,
+ "loss": 1.5276,
+ "step": 3052
+ },
+ {
+ "epoch": 8.318801089918257,
+ "grad_norm": 6.325285911560059,
+ "learning_rate": 1.9851993919842314e-05,
+ "loss": 1.2639,
+ "step": 3053
+ },
+ {
+ "epoch": 8.321525885558582,
+ "grad_norm": 7.542845726013184,
+ "learning_rate": 1.9851842611091534e-05,
+ "loss": 1.3381,
+ "step": 3054
+ },
+ {
+ "epoch": 8.32425068119891,
+ "grad_norm": 6.883162975311279,
+ "learning_rate": 1.985169122561492e-05,
+ "loss": 1.302,
+ "step": 3055
+ },
+ {
+ "epoch": 8.326975476839237,
+ "grad_norm": 10.602054595947266,
+ "learning_rate": 1.9851539763413653e-05,
+ "loss": 1.2539,
+ "step": 3056
+ },
+ {
+ "epoch": 8.329700272479563,
+ "grad_norm": 6.517338275909424,
+ "learning_rate": 1.985138822448891e-05,
+ "loss": 1.236,
+ "step": 3057
+ },
+ {
+ "epoch": 8.33242506811989,
+ "grad_norm": 5.632065773010254,
+ "learning_rate": 1.9851236608841874e-05,
+ "loss": 1.1675,
+ "step": 3058
+ },
+ {
+ "epoch": 8.335149863760218,
+ "grad_norm": 6.56447172164917,
+ "learning_rate": 1.985108491647372e-05,
+ "loss": 1.2244,
+ "step": 3059
+ },
+ {
+ "epoch": 8.337874659400544,
+ "grad_norm": 6.19110631942749,
+ "learning_rate": 1.9850933147385637e-05,
+ "loss": 1.1288,
+ "step": 3060
+ },
+ {
+ "epoch": 8.340599455040872,
+ "grad_norm": 6.876585960388184,
+ "learning_rate": 1.9850781301578808e-05,
+ "loss": 1.231,
+ "step": 3061
+ },
+ {
+ "epoch": 8.3433242506812,
+ "grad_norm": 7.70264196395874,
+ "learning_rate": 1.985062937905441e-05,
+ "loss": 1.3328,
+ "step": 3062
+ },
+ {
+ "epoch": 8.346049046321525,
+ "grad_norm": 10.664886474609375,
+ "learning_rate": 1.9850477379813622e-05,
+ "loss": 1.364,
+ "step": 3063
+ },
+ {
+ "epoch": 8.348773841961853,
+ "grad_norm": 6.265231609344482,
+ "learning_rate": 1.9850325303857634e-05,
+ "loss": 1.3152,
+ "step": 3064
+ },
+ {
+ "epoch": 8.35149863760218,
+ "grad_norm": 5.94236946105957,
+ "learning_rate": 1.9850173151187634e-05,
+ "loss": 1.2632,
+ "step": 3065
+ },
+ {
+ "epoch": 8.354223433242506,
+ "grad_norm": 19.339860916137695,
+ "learning_rate": 1.9850020921804797e-05,
+ "loss": 1.4517,
+ "step": 3066
+ },
+ {
+ "epoch": 8.356948228882834,
+ "grad_norm": 7.919244766235352,
+ "learning_rate": 1.9849868615710316e-05,
+ "loss": 1.2981,
+ "step": 3067
+ },
+ {
+ "epoch": 8.359673024523161,
+ "grad_norm": 5.552083969116211,
+ "learning_rate": 1.9849716232905378e-05,
+ "loss": 1.1255,
+ "step": 3068
+ },
+ {
+ "epoch": 8.362397820163487,
+ "grad_norm": 11.505508422851562,
+ "learning_rate": 1.9849563773391163e-05,
+ "loss": 1.2644,
+ "step": 3069
+ },
+ {
+ "epoch": 8.365122615803815,
+ "grad_norm": 8.644599914550781,
+ "learning_rate": 1.9849411237168866e-05,
+ "loss": 1.3274,
+ "step": 3070
+ },
+ {
+ "epoch": 8.367847411444142,
+ "grad_norm": 7.751000881195068,
+ "learning_rate": 1.9849258624239672e-05,
+ "loss": 1.3992,
+ "step": 3071
+ },
+ {
+ "epoch": 8.370572207084468,
+ "grad_norm": 16.31162452697754,
+ "learning_rate": 1.9849105934604766e-05,
+ "loss": 1.3015,
+ "step": 3072
+ },
+ {
+ "epoch": 8.373297002724795,
+ "grad_norm": 8.71719741821289,
+ "learning_rate": 1.9848953168265342e-05,
+ "loss": 1.2463,
+ "step": 3073
+ },
+ {
+ "epoch": 8.376021798365123,
+ "grad_norm": 6.5024566650390625,
+ "learning_rate": 1.9848800325222585e-05,
+ "loss": 1.3672,
+ "step": 3074
+ },
+ {
+ "epoch": 8.378746594005449,
+ "grad_norm": 7.686312675476074,
+ "learning_rate": 1.984864740547769e-05,
+ "loss": 1.417,
+ "step": 3075
+ },
+ {
+ "epoch": 8.381471389645776,
+ "grad_norm": 6.17648458480835,
+ "learning_rate": 1.9848494409031843e-05,
+ "loss": 0.9602,
+ "step": 3076
+ },
+ {
+ "epoch": 8.384196185286104,
+ "grad_norm": 7.479003429412842,
+ "learning_rate": 1.9848341335886243e-05,
+ "loss": 1.2825,
+ "step": 3077
+ },
+ {
+ "epoch": 8.38692098092643,
+ "grad_norm": 7.820697784423828,
+ "learning_rate": 1.9848188186042077e-05,
+ "loss": 1.1973,
+ "step": 3078
+ },
+ {
+ "epoch": 8.389645776566757,
+ "grad_norm": 13.820751190185547,
+ "learning_rate": 1.9848034959500535e-05,
+ "loss": 1.13,
+ "step": 3079
+ },
+ {
+ "epoch": 8.392370572207085,
+ "grad_norm": 7.552674293518066,
+ "learning_rate": 1.9847881656262816e-05,
+ "loss": 1.3979,
+ "step": 3080
+ },
+ {
+ "epoch": 8.39509536784741,
+ "grad_norm": 6.972393989562988,
+ "learning_rate": 1.9847728276330113e-05,
+ "loss": 1.1543,
+ "step": 3081
+ },
+ {
+ "epoch": 8.397820163487738,
+ "grad_norm": 9.59457778930664,
+ "learning_rate": 1.9847574819703617e-05,
+ "loss": 1.3203,
+ "step": 3082
+ },
+ {
+ "epoch": 8.400544959128066,
+ "grad_norm": 6.74641227722168,
+ "learning_rate": 1.9847421286384526e-05,
+ "loss": 1.1758,
+ "step": 3083
+ },
+ {
+ "epoch": 8.403269754768392,
+ "grad_norm": 8.357032775878906,
+ "learning_rate": 1.9847267676374035e-05,
+ "loss": 1.1687,
+ "step": 3084
+ },
+ {
+ "epoch": 8.40599455040872,
+ "grad_norm": 7.799906253814697,
+ "learning_rate": 1.984711398967334e-05,
+ "loss": 1.2007,
+ "step": 3085
+ },
+ {
+ "epoch": 8.408719346049047,
+ "grad_norm": 6.938974380493164,
+ "learning_rate": 1.9846960226283638e-05,
+ "loss": 1.3242,
+ "step": 3086
+ },
+ {
+ "epoch": 8.411444141689373,
+ "grad_norm": 9.076458930969238,
+ "learning_rate": 1.9846806386206127e-05,
+ "loss": 1.4954,
+ "step": 3087
+ },
+ {
+ "epoch": 8.4141689373297,
+ "grad_norm": 8.82581615447998,
+ "learning_rate": 1.9846652469442003e-05,
+ "loss": 1.4407,
+ "step": 3088
+ },
+ {
+ "epoch": 8.416893732970028,
+ "grad_norm": 6.532219409942627,
+ "learning_rate": 1.9846498475992467e-05,
+ "loss": 1.1755,
+ "step": 3089
+ },
+ {
+ "epoch": 8.419618528610354,
+ "grad_norm": 15.95340347290039,
+ "learning_rate": 1.984634440585872e-05,
+ "loss": 1.1929,
+ "step": 3090
+ },
+ {
+ "epoch": 8.422343324250681,
+ "grad_norm": 6.894241809844971,
+ "learning_rate": 1.9846190259041956e-05,
+ "loss": 1.394,
+ "step": 3091
+ },
+ {
+ "epoch": 8.425068119891009,
+ "grad_norm": 7.552821159362793,
+ "learning_rate": 1.9846036035543384e-05,
+ "loss": 1.3638,
+ "step": 3092
+ },
+ {
+ "epoch": 8.427792915531334,
+ "grad_norm": 7.358844757080078,
+ "learning_rate": 1.98458817353642e-05,
+ "loss": 1.2852,
+ "step": 3093
+ },
+ {
+ "epoch": 8.430517711171662,
+ "grad_norm": 6.742053985595703,
+ "learning_rate": 1.9845727358505603e-05,
+ "loss": 1.2185,
+ "step": 3094
+ },
+ {
+ "epoch": 8.43324250681199,
+ "grad_norm": 7.090906620025635,
+ "learning_rate": 1.9845572904968797e-05,
+ "loss": 1.3276,
+ "step": 3095
+ },
+ {
+ "epoch": 8.435967302452315,
+ "grad_norm": 7.579506874084473,
+ "learning_rate": 1.984541837475499e-05,
+ "loss": 1.344,
+ "step": 3096
+ },
+ {
+ "epoch": 8.438692098092643,
+ "grad_norm": 7.888333320617676,
+ "learning_rate": 1.984526376786538e-05,
+ "loss": 1.4078,
+ "step": 3097
+ },
+ {
+ "epoch": 8.44141689373297,
+ "grad_norm": 7.560827732086182,
+ "learning_rate": 1.9845109084301173e-05,
+ "loss": 1.332,
+ "step": 3098
+ },
+ {
+ "epoch": 8.444141689373296,
+ "grad_norm": 7.30055046081543,
+ "learning_rate": 1.984495432406357e-05,
+ "loss": 1.3423,
+ "step": 3099
+ },
+ {
+ "epoch": 8.446866485013624,
+ "grad_norm": 7.555819988250732,
+ "learning_rate": 1.9844799487153782e-05,
+ "loss": 1.3101,
+ "step": 3100
+ },
+ {
+ "epoch": 8.449591280653951,
+ "grad_norm": 6.0318684577941895,
+ "learning_rate": 1.9844644573573016e-05,
+ "loss": 1.2766,
+ "step": 3101
+ },
+ {
+ "epoch": 8.452316076294277,
+ "grad_norm": 12.336150169372559,
+ "learning_rate": 1.984448958332247e-05,
+ "loss": 1.2471,
+ "step": 3102
+ },
+ {
+ "epoch": 8.455040871934605,
+ "grad_norm": 6.012147903442383,
+ "learning_rate": 1.9844334516403356e-05,
+ "loss": 1.1469,
+ "step": 3103
+ },
+ {
+ "epoch": 8.457765667574932,
+ "grad_norm": 7.038894176483154,
+ "learning_rate": 1.9844179372816884e-05,
+ "loss": 1.3403,
+ "step": 3104
+ },
+ {
+ "epoch": 8.460490463215258,
+ "grad_norm": 122.76052856445312,
+ "learning_rate": 1.984402415256426e-05,
+ "loss": 1.4971,
+ "step": 3105
+ },
+ {
+ "epoch": 8.463215258855586,
+ "grad_norm": 5.568105220794678,
+ "learning_rate": 1.984386885564669e-05,
+ "loss": 1.1191,
+ "step": 3106
+ },
+ {
+ "epoch": 8.465940054495913,
+ "grad_norm": 8.142412185668945,
+ "learning_rate": 1.984371348206539e-05,
+ "loss": 1.364,
+ "step": 3107
+ },
+ {
+ "epoch": 8.46866485013624,
+ "grad_norm": 7.994696140289307,
+ "learning_rate": 1.9843558031821563e-05,
+ "loss": 1.4546,
+ "step": 3108
+ },
+ {
+ "epoch": 8.471389645776567,
+ "grad_norm": 6.918599605560303,
+ "learning_rate": 1.9843402504916424e-05,
+ "loss": 1.5549,
+ "step": 3109
+ },
+ {
+ "epoch": 8.474114441416894,
+ "grad_norm": 7.517693996429443,
+ "learning_rate": 1.9843246901351184e-05,
+ "loss": 1.207,
+ "step": 3110
+ },
+ {
+ "epoch": 8.47683923705722,
+ "grad_norm": 14.629093170166016,
+ "learning_rate": 1.9843091221127053e-05,
+ "loss": 1.4976,
+ "step": 3111
+ },
+ {
+ "epoch": 8.479564032697548,
+ "grad_norm": 11.482966423034668,
+ "learning_rate": 1.9842935464245245e-05,
+ "loss": 1.4558,
+ "step": 3112
+ },
+ {
+ "epoch": 8.482288828337875,
+ "grad_norm": 6.96058464050293,
+ "learning_rate": 1.9842779630706972e-05,
+ "loss": 1.2905,
+ "step": 3113
+ },
+ {
+ "epoch": 8.485013623978201,
+ "grad_norm": 10.963068962097168,
+ "learning_rate": 1.9842623720513447e-05,
+ "loss": 1.5632,
+ "step": 3114
+ },
+ {
+ "epoch": 8.487738419618529,
+ "grad_norm": 7.508437156677246,
+ "learning_rate": 1.9842467733665886e-05,
+ "loss": 1.2925,
+ "step": 3115
+ },
+ {
+ "epoch": 8.490463215258856,
+ "grad_norm": 8.231701850891113,
+ "learning_rate": 1.9842311670165503e-05,
+ "loss": 1.2908,
+ "step": 3116
+ },
+ {
+ "epoch": 8.493188010899182,
+ "grad_norm": 8.733600616455078,
+ "learning_rate": 1.9842155530013514e-05,
+ "loss": 1.448,
+ "step": 3117
+ },
+ {
+ "epoch": 8.49591280653951,
+ "grad_norm": 8.473097801208496,
+ "learning_rate": 1.9841999313211135e-05,
+ "loss": 1.4788,
+ "step": 3118
+ },
+ {
+ "epoch": 8.498637602179837,
+ "grad_norm": 11.48928451538086,
+ "learning_rate": 1.9841843019759583e-05,
+ "loss": 1.4165,
+ "step": 3119
+ },
+ {
+ "epoch": 8.501362397820163,
+ "grad_norm": 8.182844161987305,
+ "learning_rate": 1.984168664966007e-05,
+ "loss": 1.3257,
+ "step": 3120
+ },
+ {
+ "epoch": 8.50408719346049,
+ "grad_norm": 7.303596496582031,
+ "learning_rate": 1.9841530202913822e-05,
+ "loss": 1.2642,
+ "step": 3121
+ },
+ {
+ "epoch": 8.506811989100818,
+ "grad_norm": 9.317054748535156,
+ "learning_rate": 1.984137367952205e-05,
+ "loss": 1.3142,
+ "step": 3122
+ },
+ {
+ "epoch": 8.509536784741144,
+ "grad_norm": 10.313058853149414,
+ "learning_rate": 1.984121707948598e-05,
+ "loss": 1.5654,
+ "step": 3123
+ },
+ {
+ "epoch": 8.512261580381471,
+ "grad_norm": 8.488879203796387,
+ "learning_rate": 1.9841060402806826e-05,
+ "loss": 1.3746,
+ "step": 3124
+ },
+ {
+ "epoch": 8.514986376021799,
+ "grad_norm": 8.488832473754883,
+ "learning_rate": 1.9840903649485812e-05,
+ "loss": 1.4536,
+ "step": 3125
+ },
+ {
+ "epoch": 8.517711171662125,
+ "grad_norm": 7.687344074249268,
+ "learning_rate": 1.9840746819524157e-05,
+ "loss": 1.24,
+ "step": 3126
+ },
+ {
+ "epoch": 8.520435967302452,
+ "grad_norm": 7.279966354370117,
+ "learning_rate": 1.9840589912923082e-05,
+ "loss": 1.2598,
+ "step": 3127
+ },
+ {
+ "epoch": 8.52316076294278,
+ "grad_norm": 8.035114288330078,
+ "learning_rate": 1.984043292968381e-05,
+ "loss": 1.1858,
+ "step": 3128
+ },
+ {
+ "epoch": 8.525885558583106,
+ "grad_norm": 13.695732116699219,
+ "learning_rate": 1.984027586980756e-05,
+ "loss": 1.3462,
+ "step": 3129
+ },
+ {
+ "epoch": 8.528610354223433,
+ "grad_norm": 7.5380377769470215,
+ "learning_rate": 1.984011873329556e-05,
+ "loss": 1.0784,
+ "step": 3130
+ },
+ {
+ "epoch": 8.53133514986376,
+ "grad_norm": 7.911507606506348,
+ "learning_rate": 1.9839961520149035e-05,
+ "loss": 1.3416,
+ "step": 3131
+ },
+ {
+ "epoch": 8.534059945504087,
+ "grad_norm": 9.214513778686523,
+ "learning_rate": 1.9839804230369204e-05,
+ "loss": 1.2861,
+ "step": 3132
+ },
+ {
+ "epoch": 8.536784741144414,
+ "grad_norm": 7.603087425231934,
+ "learning_rate": 1.9839646863957294e-05,
+ "loss": 1.2405,
+ "step": 3133
+ },
+ {
+ "epoch": 8.539509536784742,
+ "grad_norm": 6.235926151275635,
+ "learning_rate": 1.9839489420914533e-05,
+ "loss": 1.2102,
+ "step": 3134
+ },
+ {
+ "epoch": 8.542234332425068,
+ "grad_norm": 8.465983390808105,
+ "learning_rate": 1.9839331901242144e-05,
+ "loss": 1.3782,
+ "step": 3135
+ },
+ {
+ "epoch": 8.544959128065395,
+ "grad_norm": 9.529025077819824,
+ "learning_rate": 1.983917430494135e-05,
+ "loss": 1.302,
+ "step": 3136
+ },
+ {
+ "epoch": 8.547683923705723,
+ "grad_norm": 7.514167785644531,
+ "learning_rate": 1.9839016632013388e-05,
+ "loss": 1.4319,
+ "step": 3137
+ },
+ {
+ "epoch": 8.550408719346049,
+ "grad_norm": 11.003640174865723,
+ "learning_rate": 1.9838858882459483e-05,
+ "loss": 1.457,
+ "step": 3138
+ },
+ {
+ "epoch": 8.553133514986376,
+ "grad_norm": 7.572007656097412,
+ "learning_rate": 1.9838701056280857e-05,
+ "loss": 1.1895,
+ "step": 3139
+ },
+ {
+ "epoch": 8.555858310626704,
+ "grad_norm": 7.431583881378174,
+ "learning_rate": 1.9838543153478746e-05,
+ "loss": 1.2991,
+ "step": 3140
+ },
+ {
+ "epoch": 8.55858310626703,
+ "grad_norm": 8.193229675292969,
+ "learning_rate": 1.9838385174054376e-05,
+ "loss": 1.2798,
+ "step": 3141
+ },
+ {
+ "epoch": 8.561307901907357,
+ "grad_norm": 8.877386093139648,
+ "learning_rate": 1.9838227118008982e-05,
+ "loss": 1.5627,
+ "step": 3142
+ },
+ {
+ "epoch": 8.564032697547685,
+ "grad_norm": 5.953948974609375,
+ "learning_rate": 1.983806898534379e-05,
+ "loss": 1.2419,
+ "step": 3143
+ },
+ {
+ "epoch": 8.56675749318801,
+ "grad_norm": 7.014137268066406,
+ "learning_rate": 1.9837910776060036e-05,
+ "loss": 1.3438,
+ "step": 3144
+ },
+ {
+ "epoch": 8.569482288828338,
+ "grad_norm": 6.45278787612915,
+ "learning_rate": 1.9837752490158946e-05,
+ "loss": 1.2441,
+ "step": 3145
+ },
+ {
+ "epoch": 8.572207084468666,
+ "grad_norm": 7.292781829833984,
+ "learning_rate": 1.9837594127641758e-05,
+ "loss": 1.4128,
+ "step": 3146
+ },
+ {
+ "epoch": 8.574931880108991,
+ "grad_norm": 6.353744983673096,
+ "learning_rate": 1.9837435688509704e-05,
+ "loss": 1.2181,
+ "step": 3147
+ },
+ {
+ "epoch": 8.577656675749319,
+ "grad_norm": 7.322988033294678,
+ "learning_rate": 1.9837277172764016e-05,
+ "loss": 1.4976,
+ "step": 3148
+ },
+ {
+ "epoch": 8.580381471389646,
+ "grad_norm": 6.034753799438477,
+ "learning_rate": 1.9837118580405928e-05,
+ "loss": 1.1519,
+ "step": 3149
+ },
+ {
+ "epoch": 8.583106267029972,
+ "grad_norm": 7.617753982543945,
+ "learning_rate": 1.9836959911436682e-05,
+ "loss": 1.4844,
+ "step": 3150
+ },
+ {
+ "epoch": 8.5858310626703,
+ "grad_norm": 7.455654144287109,
+ "learning_rate": 1.9836801165857507e-05,
+ "loss": 1.2834,
+ "step": 3151
+ },
+ {
+ "epoch": 8.588555858310627,
+ "grad_norm": 6.627842426300049,
+ "learning_rate": 1.983664234366964e-05,
+ "loss": 1.2705,
+ "step": 3152
+ },
+ {
+ "epoch": 8.591280653950953,
+ "grad_norm": 10.12028980255127,
+ "learning_rate": 1.9836483444874317e-05,
+ "loss": 1.1738,
+ "step": 3153
+ },
+ {
+ "epoch": 8.59400544959128,
+ "grad_norm": 6.676175117492676,
+ "learning_rate": 1.9836324469472782e-05,
+ "loss": 1.2788,
+ "step": 3154
+ },
+ {
+ "epoch": 8.596730245231608,
+ "grad_norm": 7.006638050079346,
+ "learning_rate": 1.9836165417466266e-05,
+ "loss": 1.1604,
+ "step": 3155
+ },
+ {
+ "epoch": 8.599455040871934,
+ "grad_norm": 7.232136249542236,
+ "learning_rate": 1.983600628885601e-05,
+ "loss": 1.2286,
+ "step": 3156
+ },
+ {
+ "epoch": 8.602179836512262,
+ "grad_norm": 8.915482521057129,
+ "learning_rate": 1.9835847083643256e-05,
+ "loss": 1.5327,
+ "step": 3157
+ },
+ {
+ "epoch": 8.60490463215259,
+ "grad_norm": 10.728601455688477,
+ "learning_rate": 1.983568780182924e-05,
+ "loss": 1.1997,
+ "step": 3158
+ },
+ {
+ "epoch": 8.607629427792915,
+ "grad_norm": 8.589773178100586,
+ "learning_rate": 1.9835528443415205e-05,
+ "loss": 1.3025,
+ "step": 3159
+ },
+ {
+ "epoch": 8.610354223433243,
+ "grad_norm": 6.627230644226074,
+ "learning_rate": 1.983536900840239e-05,
+ "loss": 1.3269,
+ "step": 3160
+ },
+ {
+ "epoch": 8.61307901907357,
+ "grad_norm": 10.845163345336914,
+ "learning_rate": 1.9835209496792038e-05,
+ "loss": 1.3457,
+ "step": 3161
+ },
+ {
+ "epoch": 8.615803814713896,
+ "grad_norm": 8.399524688720703,
+ "learning_rate": 1.9835049908585388e-05,
+ "loss": 1.1249,
+ "step": 3162
+ },
+ {
+ "epoch": 8.618528610354224,
+ "grad_norm": 6.071731090545654,
+ "learning_rate": 1.983489024378369e-05,
+ "loss": 1.1001,
+ "step": 3163
+ },
+ {
+ "epoch": 8.621253405994551,
+ "grad_norm": 5.896820068359375,
+ "learning_rate": 1.983473050238818e-05,
+ "loss": 1.0281,
+ "step": 3164
+ },
+ {
+ "epoch": 8.623978201634877,
+ "grad_norm": 6.548626899719238,
+ "learning_rate": 1.983457068440011e-05,
+ "loss": 1.1426,
+ "step": 3165
+ },
+ {
+ "epoch": 8.626702997275205,
+ "grad_norm": 9.089279174804688,
+ "learning_rate": 1.983441078982072e-05,
+ "loss": 1.2891,
+ "step": 3166
+ },
+ {
+ "epoch": 8.629427792915532,
+ "grad_norm": 5.758304595947266,
+ "learning_rate": 1.983425081865125e-05,
+ "loss": 1.2773,
+ "step": 3167
+ },
+ {
+ "epoch": 8.632152588555858,
+ "grad_norm": 6.587614059448242,
+ "learning_rate": 1.9834090770892952e-05,
+ "loss": 1.394,
+ "step": 3168
+ },
+ {
+ "epoch": 8.634877384196185,
+ "grad_norm": 6.4414143562316895,
+ "learning_rate": 1.9833930646547074e-05,
+ "loss": 1.1208,
+ "step": 3169
+ },
+ {
+ "epoch": 8.637602179836513,
+ "grad_norm": 8.07323169708252,
+ "learning_rate": 1.983377044561486e-05,
+ "loss": 1.4404,
+ "step": 3170
+ },
+ {
+ "epoch": 8.640326975476839,
+ "grad_norm": 7.165570259094238,
+ "learning_rate": 1.9833610168097558e-05,
+ "loss": 1.2147,
+ "step": 3171
+ },
+ {
+ "epoch": 8.643051771117166,
+ "grad_norm": 7.543591022491455,
+ "learning_rate": 1.9833449813996416e-05,
+ "loss": 1.4175,
+ "step": 3172
+ },
+ {
+ "epoch": 8.645776566757494,
+ "grad_norm": 6.776225566864014,
+ "learning_rate": 1.9833289383312687e-05,
+ "loss": 1.1775,
+ "step": 3173
+ },
+ {
+ "epoch": 8.64850136239782,
+ "grad_norm": 7.795539855957031,
+ "learning_rate": 1.9833128876047612e-05,
+ "loss": 1.3931,
+ "step": 3174
+ },
+ {
+ "epoch": 8.651226158038147,
+ "grad_norm": 5.68392276763916,
+ "learning_rate": 1.9832968292202447e-05,
+ "loss": 1.2319,
+ "step": 3175
+ },
+ {
+ "epoch": 8.653950953678475,
+ "grad_norm": 7.501158714294434,
+ "learning_rate": 1.9832807631778443e-05,
+ "loss": 1.1423,
+ "step": 3176
+ },
+ {
+ "epoch": 8.6566757493188,
+ "grad_norm": 21.968408584594727,
+ "learning_rate": 1.9832646894776847e-05,
+ "loss": 1.3254,
+ "step": 3177
+ },
+ {
+ "epoch": 8.659400544959128,
+ "grad_norm": 7.666440963745117,
+ "learning_rate": 1.9832486081198915e-05,
+ "loss": 1.2635,
+ "step": 3178
+ },
+ {
+ "epoch": 8.662125340599456,
+ "grad_norm": 5.603022575378418,
+ "learning_rate": 1.9832325191045898e-05,
+ "loss": 1.1052,
+ "step": 3179
+ },
+ {
+ "epoch": 8.664850136239782,
+ "grad_norm": 5.907281398773193,
+ "learning_rate": 1.983216422431905e-05,
+ "loss": 1.1873,
+ "step": 3180
+ },
+ {
+ "epoch": 8.66757493188011,
+ "grad_norm": 7.887642860412598,
+ "learning_rate": 1.9832003181019622e-05,
+ "loss": 1.1956,
+ "step": 3181
+ },
+ {
+ "epoch": 8.670299727520437,
+ "grad_norm": 7.220061779022217,
+ "learning_rate": 1.983184206114887e-05,
+ "loss": 1.2854,
+ "step": 3182
+ },
+ {
+ "epoch": 8.673024523160763,
+ "grad_norm": 8.437344551086426,
+ "learning_rate": 1.983168086470805e-05,
+ "loss": 1.2124,
+ "step": 3183
+ },
+ {
+ "epoch": 8.67574931880109,
+ "grad_norm": 8.07899284362793,
+ "learning_rate": 1.9831519591698413e-05,
+ "loss": 1.292,
+ "step": 3184
+ },
+ {
+ "epoch": 8.678474114441418,
+ "grad_norm": 9.709419250488281,
+ "learning_rate": 1.983135824212122e-05,
+ "loss": 1.4258,
+ "step": 3185
+ },
+ {
+ "epoch": 8.681198910081743,
+ "grad_norm": 6.9092698097229,
+ "learning_rate": 1.9831196815977724e-05,
+ "loss": 1.3843,
+ "step": 3186
+ },
+ {
+ "epoch": 8.683923705722071,
+ "grad_norm": 7.14180326461792,
+ "learning_rate": 1.9831035313269185e-05,
+ "loss": 1.4507,
+ "step": 3187
+ },
+ {
+ "epoch": 8.686648501362399,
+ "grad_norm": 10.511031150817871,
+ "learning_rate": 1.9830873733996857e-05,
+ "loss": 1.4756,
+ "step": 3188
+ },
+ {
+ "epoch": 8.689373297002724,
+ "grad_norm": 7.794114112854004,
+ "learning_rate": 1.9830712078162e-05,
+ "loss": 1.3135,
+ "step": 3189
+ },
+ {
+ "epoch": 8.692098092643052,
+ "grad_norm": 7.074607849121094,
+ "learning_rate": 1.983055034576588e-05,
+ "loss": 1.3947,
+ "step": 3190
+ },
+ {
+ "epoch": 8.69482288828338,
+ "grad_norm": 7.243511199951172,
+ "learning_rate": 1.9830388536809746e-05,
+ "loss": 1.2673,
+ "step": 3191
+ },
+ {
+ "epoch": 8.697547683923705,
+ "grad_norm": 8.690658569335938,
+ "learning_rate": 1.983022665129486e-05,
+ "loss": 1.5276,
+ "step": 3192
+ },
+ {
+ "epoch": 8.700272479564033,
+ "grad_norm": 8.475358009338379,
+ "learning_rate": 1.983006468922249e-05,
+ "loss": 1.3611,
+ "step": 3193
+ },
+ {
+ "epoch": 8.70299727520436,
+ "grad_norm": 7.035951137542725,
+ "learning_rate": 1.982990265059389e-05,
+ "loss": 1.2065,
+ "step": 3194
+ },
+ {
+ "epoch": 8.705722070844686,
+ "grad_norm": 13.182921409606934,
+ "learning_rate": 1.9829740535410326e-05,
+ "loss": 1.3032,
+ "step": 3195
+ },
+ {
+ "epoch": 8.708446866485014,
+ "grad_norm": 20.13959503173828,
+ "learning_rate": 1.9829578343673058e-05,
+ "loss": 1.2526,
+ "step": 3196
+ },
+ {
+ "epoch": 8.711171662125341,
+ "grad_norm": 7.165153980255127,
+ "learning_rate": 1.982941607538335e-05,
+ "loss": 1.5789,
+ "step": 3197
+ },
+ {
+ "epoch": 8.713896457765667,
+ "grad_norm": 8.415545463562012,
+ "learning_rate": 1.9829253730542466e-05,
+ "loss": 1.1951,
+ "step": 3198
+ },
+ {
+ "epoch": 8.716621253405995,
+ "grad_norm": 6.981612205505371,
+ "learning_rate": 1.9829091309151672e-05,
+ "loss": 1.2297,
+ "step": 3199
+ },
+ {
+ "epoch": 8.719346049046322,
+ "grad_norm": 7.90787410736084,
+ "learning_rate": 1.9828928811212228e-05,
+ "loss": 1.1694,
+ "step": 3200
+ },
+ {
+ "epoch": 8.722070844686648,
+ "grad_norm": 10.680697441101074,
+ "learning_rate": 1.9828766236725406e-05,
+ "loss": 1.2593,
+ "step": 3201
+ },
+ {
+ "epoch": 8.724795640326976,
+ "grad_norm": 7.614038467407227,
+ "learning_rate": 1.9828603585692468e-05,
+ "loss": 1.0659,
+ "step": 3202
+ },
+ {
+ "epoch": 8.727520435967303,
+ "grad_norm": 9.44372844696045,
+ "learning_rate": 1.982844085811468e-05,
+ "loss": 1.3779,
+ "step": 3203
+ },
+ {
+ "epoch": 8.730245231607629,
+ "grad_norm": 13.917326927185059,
+ "learning_rate": 1.9828278053993313e-05,
+ "loss": 1.1577,
+ "step": 3204
+ },
+ {
+ "epoch": 8.732970027247957,
+ "grad_norm": 9.828734397888184,
+ "learning_rate": 1.982811517332963e-05,
+ "loss": 1.1846,
+ "step": 3205
+ },
+ {
+ "epoch": 8.735694822888284,
+ "grad_norm": 8.257280349731445,
+ "learning_rate": 1.9827952216124908e-05,
+ "loss": 1.4081,
+ "step": 3206
+ },
+ {
+ "epoch": 8.73841961852861,
+ "grad_norm": 7.818528652191162,
+ "learning_rate": 1.9827789182380407e-05,
+ "loss": 1.1483,
+ "step": 3207
+ },
+ {
+ "epoch": 8.741144414168938,
+ "grad_norm": 11.127863883972168,
+ "learning_rate": 1.9827626072097395e-05,
+ "loss": 1.2131,
+ "step": 3208
+ },
+ {
+ "epoch": 8.743869209809265,
+ "grad_norm": 8.062250137329102,
+ "learning_rate": 1.9827462885277155e-05,
+ "loss": 1.3787,
+ "step": 3209
+ },
+ {
+ "epoch": 8.746594005449591,
+ "grad_norm": 8.683839797973633,
+ "learning_rate": 1.9827299621920948e-05,
+ "loss": 1.3284,
+ "step": 3210
+ },
+ {
+ "epoch": 8.749318801089919,
+ "grad_norm": 8.792764663696289,
+ "learning_rate": 1.982713628203005e-05,
+ "loss": 1.416,
+ "step": 3211
+ },
+ {
+ "epoch": 8.752043596730246,
+ "grad_norm": 6.705663681030273,
+ "learning_rate": 1.9826972865605725e-05,
+ "loss": 1.334,
+ "step": 3212
+ },
+ {
+ "epoch": 8.754768392370572,
+ "grad_norm": 6.464195728302002,
+ "learning_rate": 1.9826809372649258e-05,
+ "loss": 1.3606,
+ "step": 3213
+ },
+ {
+ "epoch": 8.7574931880109,
+ "grad_norm": 13.502424240112305,
+ "learning_rate": 1.982664580316191e-05,
+ "loss": 1.3445,
+ "step": 3214
+ },
+ {
+ "epoch": 8.760217983651227,
+ "grad_norm": 16.271743774414062,
+ "learning_rate": 1.9826482157144962e-05,
+ "loss": 1.3135,
+ "step": 3215
+ },
+ {
+ "epoch": 8.762942779291553,
+ "grad_norm": 7.448805809020996,
+ "learning_rate": 1.9826318434599693e-05,
+ "loss": 1.3044,
+ "step": 3216
+ },
+ {
+ "epoch": 8.76566757493188,
+ "grad_norm": 7.994974613189697,
+ "learning_rate": 1.9826154635527368e-05,
+ "loss": 1.4043,
+ "step": 3217
+ },
+ {
+ "epoch": 8.768392370572208,
+ "grad_norm": 6.819921016693115,
+ "learning_rate": 1.982599075992927e-05,
+ "loss": 1.436,
+ "step": 3218
+ },
+ {
+ "epoch": 8.771117166212534,
+ "grad_norm": 8.157017707824707,
+ "learning_rate": 1.9825826807806665e-05,
+ "loss": 1.2849,
+ "step": 3219
+ },
+ {
+ "epoch": 8.773841961852861,
+ "grad_norm": 13.353950500488281,
+ "learning_rate": 1.9825662779160843e-05,
+ "loss": 1.2495,
+ "step": 3220
+ },
+ {
+ "epoch": 8.776566757493189,
+ "grad_norm": 8.773139953613281,
+ "learning_rate": 1.9825498673993074e-05,
+ "loss": 1.3298,
+ "step": 3221
+ },
+ {
+ "epoch": 8.779291553133515,
+ "grad_norm": 7.86522102355957,
+ "learning_rate": 1.9825334492304637e-05,
+ "loss": 1.2429,
+ "step": 3222
+ },
+ {
+ "epoch": 8.782016348773842,
+ "grad_norm": 7.383118152618408,
+ "learning_rate": 1.9825170234096812e-05,
+ "loss": 1.2466,
+ "step": 3223
+ },
+ {
+ "epoch": 8.78474114441417,
+ "grad_norm": 10.245564460754395,
+ "learning_rate": 1.9825005899370877e-05,
+ "loss": 1.5474,
+ "step": 3224
+ },
+ {
+ "epoch": 8.787465940054496,
+ "grad_norm": 7.073742389678955,
+ "learning_rate": 1.982484148812811e-05,
+ "loss": 1.3696,
+ "step": 3225
+ },
+ {
+ "epoch": 8.790190735694823,
+ "grad_norm": 8.140562057495117,
+ "learning_rate": 1.9824677000369798e-05,
+ "loss": 1.3745,
+ "step": 3226
+ },
+ {
+ "epoch": 8.79291553133515,
+ "grad_norm": 7.103660583496094,
+ "learning_rate": 1.9824512436097216e-05,
+ "loss": 1.3787,
+ "step": 3227
+ },
+ {
+ "epoch": 8.795640326975477,
+ "grad_norm": 11.56889533996582,
+ "learning_rate": 1.9824347795311646e-05,
+ "loss": 1.1506,
+ "step": 3228
+ },
+ {
+ "epoch": 8.798365122615804,
+ "grad_norm": 6.064287185668945,
+ "learning_rate": 1.9824183078014373e-05,
+ "loss": 1.3218,
+ "step": 3229
+ },
+ {
+ "epoch": 8.80108991825613,
+ "grad_norm": 7.481529235839844,
+ "learning_rate": 1.9824018284206677e-05,
+ "loss": 1.2439,
+ "step": 3230
+ },
+ {
+ "epoch": 8.803814713896458,
+ "grad_norm": 7.189199924468994,
+ "learning_rate": 1.9823853413889846e-05,
+ "loss": 1.2642,
+ "step": 3231
+ },
+ {
+ "epoch": 8.806539509536785,
+ "grad_norm": 7.7545576095581055,
+ "learning_rate": 1.982368846706516e-05,
+ "loss": 1.2446,
+ "step": 3232
+ },
+ {
+ "epoch": 8.809264305177111,
+ "grad_norm": 8.48669719696045,
+ "learning_rate": 1.98235234437339e-05,
+ "loss": 1.3872,
+ "step": 3233
+ },
+ {
+ "epoch": 8.811989100817438,
+ "grad_norm": 7.038283824920654,
+ "learning_rate": 1.982335834389736e-05,
+ "loss": 1.2585,
+ "step": 3234
+ },
+ {
+ "epoch": 8.814713896457766,
+ "grad_norm": 6.595608711242676,
+ "learning_rate": 1.982319316755682e-05,
+ "loss": 1.4927,
+ "step": 3235
+ },
+ {
+ "epoch": 8.817438692098092,
+ "grad_norm": 6.130034446716309,
+ "learning_rate": 1.9823027914713567e-05,
+ "loss": 1.2283,
+ "step": 3236
+ },
+ {
+ "epoch": 8.82016348773842,
+ "grad_norm": 5.694086074829102,
+ "learning_rate": 1.982286258536889e-05,
+ "loss": 1.3115,
+ "step": 3237
+ },
+ {
+ "epoch": 8.822888283378747,
+ "grad_norm": 8.048131942749023,
+ "learning_rate": 1.9822697179524074e-05,
+ "loss": 1.5,
+ "step": 3238
+ },
+ {
+ "epoch": 8.825613079019073,
+ "grad_norm": 8.297684669494629,
+ "learning_rate": 1.9822531697180412e-05,
+ "loss": 1.2681,
+ "step": 3239
+ },
+ {
+ "epoch": 8.8283378746594,
+ "grad_norm": 11.500406265258789,
+ "learning_rate": 1.9822366138339185e-05,
+ "loss": 1.3494,
+ "step": 3240
+ },
+ {
+ "epoch": 8.831062670299728,
+ "grad_norm": 6.304725170135498,
+ "learning_rate": 1.982220050300169e-05,
+ "loss": 1.324,
+ "step": 3241
+ },
+ {
+ "epoch": 8.833787465940054,
+ "grad_norm": 10.080437660217285,
+ "learning_rate": 1.982203479116921e-05,
+ "loss": 1.3937,
+ "step": 3242
+ },
+ {
+ "epoch": 8.836512261580381,
+ "grad_norm": 6.559786796569824,
+ "learning_rate": 1.982186900284304e-05,
+ "loss": 1.3735,
+ "step": 3243
+ },
+ {
+ "epoch": 8.839237057220709,
+ "grad_norm": 6.773789882659912,
+ "learning_rate": 1.982170313802447e-05,
+ "loss": 1.3733,
+ "step": 3244
+ },
+ {
+ "epoch": 8.841961852861035,
+ "grad_norm": 7.3282976150512695,
+ "learning_rate": 1.9821537196714797e-05,
+ "loss": 1.1898,
+ "step": 3245
+ },
+ {
+ "epoch": 8.844686648501362,
+ "grad_norm": 9.572917938232422,
+ "learning_rate": 1.9821371178915307e-05,
+ "loss": 1.4387,
+ "step": 3246
+ },
+ {
+ "epoch": 8.84741144414169,
+ "grad_norm": 11.082111358642578,
+ "learning_rate": 1.9821205084627292e-05,
+ "loss": 1.2761,
+ "step": 3247
+ },
+ {
+ "epoch": 8.850136239782016,
+ "grad_norm": 6.091163635253906,
+ "learning_rate": 1.982103891385205e-05,
+ "loss": 1.1528,
+ "step": 3248
+ },
+ {
+ "epoch": 8.852861035422343,
+ "grad_norm": 7.704655170440674,
+ "learning_rate": 1.982087266659087e-05,
+ "loss": 1.1941,
+ "step": 3249
+ },
+ {
+ "epoch": 8.85558583106267,
+ "grad_norm": 7.301915168762207,
+ "learning_rate": 1.982070634284505e-05,
+ "loss": 1.2214,
+ "step": 3250
+ },
+ {
+ "epoch": 8.858310626702997,
+ "grad_norm": 10.17814826965332,
+ "learning_rate": 1.9820539942615883e-05,
+ "loss": 1.2229,
+ "step": 3251
+ },
+ {
+ "epoch": 8.861035422343324,
+ "grad_norm": 9.381146430969238,
+ "learning_rate": 1.9820373465904672e-05,
+ "loss": 1.1143,
+ "step": 3252
+ },
+ {
+ "epoch": 8.863760217983652,
+ "grad_norm": 8.097393035888672,
+ "learning_rate": 1.982020691271271e-05,
+ "loss": 1.344,
+ "step": 3253
+ },
+ {
+ "epoch": 8.866485013623977,
+ "grad_norm": 11.498970985412598,
+ "learning_rate": 1.9820040283041287e-05,
+ "loss": 1.0969,
+ "step": 3254
+ },
+ {
+ "epoch": 8.869209809264305,
+ "grad_norm": 9.005964279174805,
+ "learning_rate": 1.9819873576891707e-05,
+ "loss": 1.3206,
+ "step": 3255
+ },
+ {
+ "epoch": 8.871934604904633,
+ "grad_norm": 10.238490104675293,
+ "learning_rate": 1.981970679426527e-05,
+ "loss": 1.3083,
+ "step": 3256
+ },
+ {
+ "epoch": 8.874659400544958,
+ "grad_norm": 10.27416706085205,
+ "learning_rate": 1.9819539935163272e-05,
+ "loss": 1.5782,
+ "step": 3257
+ },
+ {
+ "epoch": 8.877384196185286,
+ "grad_norm": 7.6527814865112305,
+ "learning_rate": 1.9819372999587015e-05,
+ "loss": 1.457,
+ "step": 3258
+ },
+ {
+ "epoch": 8.880108991825614,
+ "grad_norm": 7.866795063018799,
+ "learning_rate": 1.9819205987537792e-05,
+ "loss": 1.2756,
+ "step": 3259
+ },
+ {
+ "epoch": 8.88283378746594,
+ "grad_norm": 8.902711868286133,
+ "learning_rate": 1.9819038899016913e-05,
+ "loss": 1.1931,
+ "step": 3260
+ },
+ {
+ "epoch": 8.885558583106267,
+ "grad_norm": 7.637500762939453,
+ "learning_rate": 1.9818871734025672e-05,
+ "loss": 1.0181,
+ "step": 3261
+ },
+ {
+ "epoch": 8.888283378746594,
+ "grad_norm": 7.880782604217529,
+ "learning_rate": 1.9818704492565378e-05,
+ "loss": 1.3274,
+ "step": 3262
+ },
+ {
+ "epoch": 8.89100817438692,
+ "grad_norm": 7.11788272857666,
+ "learning_rate": 1.9818537174637328e-05,
+ "loss": 1.126,
+ "step": 3263
+ },
+ {
+ "epoch": 8.893732970027248,
+ "grad_norm": 10.272889137268066,
+ "learning_rate": 1.9818369780242822e-05,
+ "loss": 1.3323,
+ "step": 3264
+ },
+ {
+ "epoch": 8.896457765667575,
+ "grad_norm": 6.814765930175781,
+ "learning_rate": 1.9818202309383173e-05,
+ "loss": 1.1445,
+ "step": 3265
+ },
+ {
+ "epoch": 8.899182561307901,
+ "grad_norm": 8.707869529724121,
+ "learning_rate": 1.9818034762059678e-05,
+ "loss": 1.4993,
+ "step": 3266
+ },
+ {
+ "epoch": 8.901907356948229,
+ "grad_norm": 6.796337127685547,
+ "learning_rate": 1.9817867138273645e-05,
+ "loss": 1.4026,
+ "step": 3267
+ },
+ {
+ "epoch": 8.904632152588556,
+ "grad_norm": 7.0194830894470215,
+ "learning_rate": 1.981769943802638e-05,
+ "loss": 1.2152,
+ "step": 3268
+ },
+ {
+ "epoch": 8.907356948228882,
+ "grad_norm": 7.964925765991211,
+ "learning_rate": 1.981753166131919e-05,
+ "loss": 1.1226,
+ "step": 3269
+ },
+ {
+ "epoch": 8.91008174386921,
+ "grad_norm": 8.967559814453125,
+ "learning_rate": 1.9817363808153372e-05,
+ "loss": 1.4028,
+ "step": 3270
+ },
+ {
+ "epoch": 8.912806539509537,
+ "grad_norm": 8.808220863342285,
+ "learning_rate": 1.9817195878530247e-05,
+ "loss": 1.4417,
+ "step": 3271
+ },
+ {
+ "epoch": 8.915531335149863,
+ "grad_norm": 8.03864574432373,
+ "learning_rate": 1.9817027872451112e-05,
+ "loss": 1.1934,
+ "step": 3272
+ },
+ {
+ "epoch": 8.91825613079019,
+ "grad_norm": 7.733057975769043,
+ "learning_rate": 1.9816859789917283e-05,
+ "loss": 1.4937,
+ "step": 3273
+ },
+ {
+ "epoch": 8.920980926430518,
+ "grad_norm": 7.16773796081543,
+ "learning_rate": 1.9816691630930067e-05,
+ "loss": 1.1904,
+ "step": 3274
+ },
+ {
+ "epoch": 8.923705722070844,
+ "grad_norm": 10.570013046264648,
+ "learning_rate": 1.981652339549077e-05,
+ "loss": 1.4353,
+ "step": 3275
+ },
+ {
+ "epoch": 8.926430517711172,
+ "grad_norm": 7.202054500579834,
+ "learning_rate": 1.9816355083600704e-05,
+ "loss": 1.1926,
+ "step": 3276
+ },
+ {
+ "epoch": 8.9291553133515,
+ "grad_norm": 9.775311470031738,
+ "learning_rate": 1.9816186695261184e-05,
+ "loss": 1.4546,
+ "step": 3277
+ },
+ {
+ "epoch": 8.931880108991825,
+ "grad_norm": 7.07136869430542,
+ "learning_rate": 1.981601823047352e-05,
+ "loss": 1.1936,
+ "step": 3278
+ },
+ {
+ "epoch": 8.934604904632153,
+ "grad_norm": 15.822300910949707,
+ "learning_rate": 1.9815849689239015e-05,
+ "loss": 1.1809,
+ "step": 3279
+ },
+ {
+ "epoch": 8.93732970027248,
+ "grad_norm": 8.788463592529297,
+ "learning_rate": 1.9815681071558992e-05,
+ "loss": 1.2156,
+ "step": 3280
+ },
+ {
+ "epoch": 8.940054495912806,
+ "grad_norm": 22.279029846191406,
+ "learning_rate": 1.9815512377434757e-05,
+ "loss": 1.3289,
+ "step": 3281
+ },
+ {
+ "epoch": 8.942779291553133,
+ "grad_norm": 7.904595375061035,
+ "learning_rate": 1.9815343606867632e-05,
+ "loss": 1.3184,
+ "step": 3282
+ },
+ {
+ "epoch": 8.945504087193461,
+ "grad_norm": 7.921975135803223,
+ "learning_rate": 1.9815174759858927e-05,
+ "loss": 1.3777,
+ "step": 3283
+ },
+ {
+ "epoch": 8.948228882833787,
+ "grad_norm": 5.569732189178467,
+ "learning_rate": 1.9815005836409954e-05,
+ "loss": 1.0359,
+ "step": 3284
+ },
+ {
+ "epoch": 8.950953678474114,
+ "grad_norm": 8.362802505493164,
+ "learning_rate": 1.9814836836522035e-05,
+ "loss": 1.3842,
+ "step": 3285
+ },
+ {
+ "epoch": 8.953678474114442,
+ "grad_norm": 5.604005336761475,
+ "learning_rate": 1.981466776019648e-05,
+ "loss": 1.062,
+ "step": 3286
+ },
+ {
+ "epoch": 8.956403269754768,
+ "grad_norm": 7.380295276641846,
+ "learning_rate": 1.9814498607434607e-05,
+ "loss": 1.1053,
+ "step": 3287
+ },
+ {
+ "epoch": 8.959128065395095,
+ "grad_norm": 7.996654033660889,
+ "learning_rate": 1.9814329378237738e-05,
+ "loss": 1.1082,
+ "step": 3288
+ },
+ {
+ "epoch": 8.961852861035423,
+ "grad_norm": 6.769784450531006,
+ "learning_rate": 1.9814160072607188e-05,
+ "loss": 1.2859,
+ "step": 3289
+ },
+ {
+ "epoch": 8.964577656675749,
+ "grad_norm": 8.257189750671387,
+ "learning_rate": 1.981399069054427e-05,
+ "loss": 1.3376,
+ "step": 3290
+ },
+ {
+ "epoch": 8.967302452316076,
+ "grad_norm": 8.391393661499023,
+ "learning_rate": 1.981382123205031e-05,
+ "loss": 1.27,
+ "step": 3291
+ },
+ {
+ "epoch": 8.970027247956404,
+ "grad_norm": 9.7191801071167,
+ "learning_rate": 1.981365169712663e-05,
+ "loss": 1.5459,
+ "step": 3292
+ },
+ {
+ "epoch": 8.97275204359673,
+ "grad_norm": 8.911846160888672,
+ "learning_rate": 1.9813482085774548e-05,
+ "loss": 1.3499,
+ "step": 3293
+ },
+ {
+ "epoch": 8.975476839237057,
+ "grad_norm": 7.771843910217285,
+ "learning_rate": 1.9813312397995376e-05,
+ "loss": 1.3723,
+ "step": 3294
+ },
+ {
+ "epoch": 8.978201634877385,
+ "grad_norm": 5.928864002227783,
+ "learning_rate": 1.9813142633790448e-05,
+ "loss": 1.2073,
+ "step": 3295
+ },
+ {
+ "epoch": 8.98092643051771,
+ "grad_norm": 8.504435539245605,
+ "learning_rate": 1.9812972793161077e-05,
+ "loss": 1.2151,
+ "step": 3296
+ },
+ {
+ "epoch": 8.983651226158038,
+ "grad_norm": 11.318719863891602,
+ "learning_rate": 1.9812802876108593e-05,
+ "loss": 1.2388,
+ "step": 3297
+ },
+ {
+ "epoch": 8.986376021798366,
+ "grad_norm": 6.212708473205566,
+ "learning_rate": 1.9812632882634316e-05,
+ "loss": 1.0239,
+ "step": 3298
+ },
+ {
+ "epoch": 8.989100817438691,
+ "grad_norm": 6.98320198059082,
+ "learning_rate": 1.981246281273957e-05,
+ "loss": 1.1826,
+ "step": 3299
+ },
+ {
+ "epoch": 8.991825613079019,
+ "grad_norm": 13.135302543640137,
+ "learning_rate": 1.981229266642568e-05,
+ "loss": 1.4192,
+ "step": 3300
+ },
+ {
+ "epoch": 8.994550408719347,
+ "grad_norm": 6.720589637756348,
+ "learning_rate": 1.9812122443693967e-05,
+ "loss": 1.1868,
+ "step": 3301
+ },
+ {
+ "epoch": 8.997275204359672,
+ "grad_norm": 9.981111526489258,
+ "learning_rate": 1.9811952144545764e-05,
+ "loss": 1.4614,
+ "step": 3302
+ },
+ {
+ "epoch": 9.0,
+ "grad_norm": 8.446125984191895,
+ "learning_rate": 1.9811781768982392e-05,
+ "loss": 1.4148,
+ "step": 3303
+ },
+ {
+ "epoch": 9.002724795640328,
+ "grad_norm": 8.478903770446777,
+ "learning_rate": 1.981161131700518e-05,
+ "loss": 1.3093,
+ "step": 3304
+ },
+ {
+ "epoch": 9.005449591280653,
+ "grad_norm": 11.67760944366455,
+ "learning_rate": 1.9811440788615454e-05,
+ "loss": 1.125,
+ "step": 3305
+ },
+ {
+ "epoch": 9.008174386920981,
+ "grad_norm": 5.653506755828857,
+ "learning_rate": 1.981127018381454e-05,
+ "loss": 1.1561,
+ "step": 3306
+ },
+ {
+ "epoch": 9.010899182561309,
+ "grad_norm": 7.869010925292969,
+ "learning_rate": 1.9811099502603775e-05,
+ "loss": 1.0967,
+ "step": 3307
+ },
+ {
+ "epoch": 9.013623978201634,
+ "grad_norm": 7.7242960929870605,
+ "learning_rate": 1.981092874498448e-05,
+ "loss": 1.376,
+ "step": 3308
+ },
+ {
+ "epoch": 9.016348773841962,
+ "grad_norm": 6.81362247467041,
+ "learning_rate": 1.981075791095799e-05,
+ "loss": 1.1853,
+ "step": 3309
+ },
+ {
+ "epoch": 9.01907356948229,
+ "grad_norm": 7.402996063232422,
+ "learning_rate": 1.9810587000525632e-05,
+ "loss": 1.2783,
+ "step": 3310
+ },
+ {
+ "epoch": 9.021798365122615,
+ "grad_norm": 6.188375473022461,
+ "learning_rate": 1.9810416013688737e-05,
+ "loss": 1.1956,
+ "step": 3311
+ },
+ {
+ "epoch": 9.024523160762943,
+ "grad_norm": 6.837196350097656,
+ "learning_rate": 1.9810244950448636e-05,
+ "loss": 1.0897,
+ "step": 3312
+ },
+ {
+ "epoch": 9.02724795640327,
+ "grad_norm": 8.616416931152344,
+ "learning_rate": 1.9810073810806668e-05,
+ "loss": 1.0444,
+ "step": 3313
+ },
+ {
+ "epoch": 9.029972752043596,
+ "grad_norm": 7.880061149597168,
+ "learning_rate": 1.9809902594764157e-05,
+ "loss": 1.2117,
+ "step": 3314
+ },
+ {
+ "epoch": 9.032697547683924,
+ "grad_norm": 7.442047119140625,
+ "learning_rate": 1.9809731302322443e-05,
+ "loss": 1.178,
+ "step": 3315
+ },
+ {
+ "epoch": 9.035422343324251,
+ "grad_norm": 5.953456401824951,
+ "learning_rate": 1.9809559933482853e-05,
+ "loss": 1.1047,
+ "step": 3316
+ },
+ {
+ "epoch": 9.038147138964577,
+ "grad_norm": 5.830977916717529,
+ "learning_rate": 1.980938848824673e-05,
+ "loss": 1.1013,
+ "step": 3317
+ },
+ {
+ "epoch": 9.040871934604905,
+ "grad_norm": 9.867141723632812,
+ "learning_rate": 1.9809216966615406e-05,
+ "loss": 1.1038,
+ "step": 3318
+ },
+ {
+ "epoch": 9.043596730245232,
+ "grad_norm": 7.219592094421387,
+ "learning_rate": 1.9809045368590215e-05,
+ "loss": 0.9771,
+ "step": 3319
+ },
+ {
+ "epoch": 9.046321525885558,
+ "grad_norm": 8.31104850769043,
+ "learning_rate": 1.980887369417249e-05,
+ "loss": 1.373,
+ "step": 3320
+ },
+ {
+ "epoch": 9.049046321525886,
+ "grad_norm": 7.118813514709473,
+ "learning_rate": 1.9808701943363577e-05,
+ "loss": 1.0984,
+ "step": 3321
+ },
+ {
+ "epoch": 9.051771117166213,
+ "grad_norm": 11.474752426147461,
+ "learning_rate": 1.980853011616481e-05,
+ "loss": 1.2233,
+ "step": 3322
+ },
+ {
+ "epoch": 9.054495912806539,
+ "grad_norm": 12.81163215637207,
+ "learning_rate": 1.9808358212577524e-05,
+ "loss": 1.186,
+ "step": 3323
+ },
+ {
+ "epoch": 9.057220708446867,
+ "grad_norm": 7.416503429412842,
+ "learning_rate": 1.980818623260306e-05,
+ "loss": 1.1885,
+ "step": 3324
+ },
+ {
+ "epoch": 9.059945504087194,
+ "grad_norm": 8.78193473815918,
+ "learning_rate": 1.9808014176242754e-05,
+ "loss": 1.2747,
+ "step": 3325
+ },
+ {
+ "epoch": 9.06267029972752,
+ "grad_norm": 6.7608723640441895,
+ "learning_rate": 1.9807842043497953e-05,
+ "loss": 0.9774,
+ "step": 3326
+ },
+ {
+ "epoch": 9.065395095367847,
+ "grad_norm": 8.016523361206055,
+ "learning_rate": 1.9807669834369996e-05,
+ "loss": 1.4215,
+ "step": 3327
+ },
+ {
+ "epoch": 9.068119891008175,
+ "grad_norm": 8.774405479431152,
+ "learning_rate": 1.9807497548860218e-05,
+ "loss": 1.324,
+ "step": 3328
+ },
+ {
+ "epoch": 9.0708446866485,
+ "grad_norm": 6.416987419128418,
+ "learning_rate": 1.9807325186969964e-05,
+ "loss": 0.9692,
+ "step": 3329
+ },
+ {
+ "epoch": 9.073569482288828,
+ "grad_norm": 7.593839645385742,
+ "learning_rate": 1.980715274870058e-05,
+ "loss": 1.0758,
+ "step": 3330
+ },
+ {
+ "epoch": 9.076294277929156,
+ "grad_norm": 10.965270042419434,
+ "learning_rate": 1.9806980234053405e-05,
+ "loss": 1.4077,
+ "step": 3331
+ },
+ {
+ "epoch": 9.079019073569482,
+ "grad_norm": 6.837986946105957,
+ "learning_rate": 1.9806807643029786e-05,
+ "loss": 1.3087,
+ "step": 3332
+ },
+ {
+ "epoch": 9.08174386920981,
+ "grad_norm": 6.921714782714844,
+ "learning_rate": 1.9806634975631062e-05,
+ "loss": 0.9661,
+ "step": 3333
+ },
+ {
+ "epoch": 9.084468664850137,
+ "grad_norm": 7.6175761222839355,
+ "learning_rate": 1.980646223185858e-05,
+ "loss": 1.3503,
+ "step": 3334
+ },
+ {
+ "epoch": 9.087193460490463,
+ "grad_norm": 7.371955394744873,
+ "learning_rate": 1.9806289411713686e-05,
+ "loss": 1.2271,
+ "step": 3335
+ },
+ {
+ "epoch": 9.08991825613079,
+ "grad_norm": 20.435834884643555,
+ "learning_rate": 1.9806116515197728e-05,
+ "loss": 1.2555,
+ "step": 3336
+ },
+ {
+ "epoch": 9.092643051771118,
+ "grad_norm": 11.705950736999512,
+ "learning_rate": 1.980594354231205e-05,
+ "loss": 1.1538,
+ "step": 3337
+ },
+ {
+ "epoch": 9.095367847411444,
+ "grad_norm": 7.436139106750488,
+ "learning_rate": 1.9805770493057994e-05,
+ "loss": 1.427,
+ "step": 3338
+ },
+ {
+ "epoch": 9.098092643051771,
+ "grad_norm": 6.631337642669678,
+ "learning_rate": 1.980559736743692e-05,
+ "loss": 1.1345,
+ "step": 3339
+ },
+ {
+ "epoch": 9.100817438692099,
+ "grad_norm": 9.533553123474121,
+ "learning_rate": 1.9805424165450164e-05,
+ "loss": 1.1541,
+ "step": 3340
+ },
+ {
+ "epoch": 9.103542234332425,
+ "grad_norm": 7.453282833099365,
+ "learning_rate": 1.980525088709908e-05,
+ "loss": 1.1127,
+ "step": 3341
+ },
+ {
+ "epoch": 9.106267029972752,
+ "grad_norm": 7.381934642791748,
+ "learning_rate": 1.9805077532385024e-05,
+ "loss": 1.5269,
+ "step": 3342
+ },
+ {
+ "epoch": 9.10899182561308,
+ "grad_norm": 9.067350387573242,
+ "learning_rate": 1.9804904101309333e-05,
+ "loss": 1.2083,
+ "step": 3343
+ },
+ {
+ "epoch": 9.111716621253406,
+ "grad_norm": 7.1999101638793945,
+ "learning_rate": 1.980473059387337e-05,
+ "loss": 1.2966,
+ "step": 3344
+ },
+ {
+ "epoch": 9.114441416893733,
+ "grad_norm": 6.43306303024292,
+ "learning_rate": 1.9804557010078475e-05,
+ "loss": 1.2288,
+ "step": 3345
+ },
+ {
+ "epoch": 9.11716621253406,
+ "grad_norm": 6.836722373962402,
+ "learning_rate": 1.9804383349926007e-05,
+ "loss": 1.0789,
+ "step": 3346
+ },
+ {
+ "epoch": 9.119891008174386,
+ "grad_norm": 6.860068321228027,
+ "learning_rate": 1.980420961341732e-05,
+ "loss": 1.2922,
+ "step": 3347
+ },
+ {
+ "epoch": 9.122615803814714,
+ "grad_norm": 9.097952842712402,
+ "learning_rate": 1.9804035800553764e-05,
+ "loss": 1.3193,
+ "step": 3348
+ },
+ {
+ "epoch": 9.125340599455042,
+ "grad_norm": 9.991456031799316,
+ "learning_rate": 1.980386191133669e-05,
+ "loss": 1.1228,
+ "step": 3349
+ },
+ {
+ "epoch": 9.128065395095367,
+ "grad_norm": 9.366521835327148,
+ "learning_rate": 1.9803687945767457e-05,
+ "loss": 0.9895,
+ "step": 3350
+ },
+ {
+ "epoch": 9.130790190735695,
+ "grad_norm": 10.031807899475098,
+ "learning_rate": 1.9803513903847418e-05,
+ "loss": 1.0912,
+ "step": 3351
+ },
+ {
+ "epoch": 9.133514986376023,
+ "grad_norm": 7.012123107910156,
+ "learning_rate": 1.9803339785577926e-05,
+ "loss": 1.1423,
+ "step": 3352
+ },
+ {
+ "epoch": 9.136239782016348,
+ "grad_norm": 7.033047676086426,
+ "learning_rate": 1.980316559096034e-05,
+ "loss": 1.3169,
+ "step": 3353
+ },
+ {
+ "epoch": 9.138964577656676,
+ "grad_norm": 6.142458438873291,
+ "learning_rate": 1.9802991319996015e-05,
+ "loss": 1.4497,
+ "step": 3354
+ },
+ {
+ "epoch": 9.141689373297003,
+ "grad_norm": 10.970399856567383,
+ "learning_rate": 1.980281697268631e-05,
+ "loss": 1.1792,
+ "step": 3355
+ },
+ {
+ "epoch": 9.14441416893733,
+ "grad_norm": 6.632962226867676,
+ "learning_rate": 1.9802642549032585e-05,
+ "loss": 0.9985,
+ "step": 3356
+ },
+ {
+ "epoch": 9.147138964577657,
+ "grad_norm": 9.675651550292969,
+ "learning_rate": 1.9802468049036192e-05,
+ "loss": 1.2371,
+ "step": 3357
+ },
+ {
+ "epoch": 9.149863760217984,
+ "grad_norm": 7.614232540130615,
+ "learning_rate": 1.9802293472698496e-05,
+ "loss": 1.1033,
+ "step": 3358
+ },
+ {
+ "epoch": 9.15258855585831,
+ "grad_norm": 10.562257766723633,
+ "learning_rate": 1.980211882002085e-05,
+ "loss": 1.4919,
+ "step": 3359
+ },
+ {
+ "epoch": 9.155313351498638,
+ "grad_norm": 7.72969388961792,
+ "learning_rate": 1.980194409100462e-05,
+ "loss": 1.1777,
+ "step": 3360
+ },
+ {
+ "epoch": 9.158038147138965,
+ "grad_norm": 8.452287673950195,
+ "learning_rate": 1.9801769285651168e-05,
+ "loss": 1.3047,
+ "step": 3361
+ },
+ {
+ "epoch": 9.160762942779291,
+ "grad_norm": 7.715450286865234,
+ "learning_rate": 1.9801594403961844e-05,
+ "loss": 1.3958,
+ "step": 3362
+ },
+ {
+ "epoch": 9.163487738419619,
+ "grad_norm": 7.842536449432373,
+ "learning_rate": 1.9801419445938027e-05,
+ "loss": 1.1233,
+ "step": 3363
+ },
+ {
+ "epoch": 9.166212534059946,
+ "grad_norm": 7.468984603881836,
+ "learning_rate": 1.9801244411581068e-05,
+ "loss": 1.3408,
+ "step": 3364
+ },
+ {
+ "epoch": 9.168937329700272,
+ "grad_norm": 6.169687747955322,
+ "learning_rate": 1.980106930089233e-05,
+ "loss": 1.1736,
+ "step": 3365
+ },
+ {
+ "epoch": 9.1716621253406,
+ "grad_norm": 5.873082160949707,
+ "learning_rate": 1.980089411387318e-05,
+ "loss": 1.3286,
+ "step": 3366
+ },
+ {
+ "epoch": 9.174386920980927,
+ "grad_norm": 6.32707405090332,
+ "learning_rate": 1.9800718850524987e-05,
+ "loss": 1.3552,
+ "step": 3367
+ },
+ {
+ "epoch": 9.177111716621253,
+ "grad_norm": 6.8150763511657715,
+ "learning_rate": 1.9800543510849107e-05,
+ "loss": 1.2468,
+ "step": 3368
+ },
+ {
+ "epoch": 9.17983651226158,
+ "grad_norm": 6.803704261779785,
+ "learning_rate": 1.980036809484691e-05,
+ "loss": 1.05,
+ "step": 3369
+ },
+ {
+ "epoch": 9.182561307901908,
+ "grad_norm": 29.067962646484375,
+ "learning_rate": 1.9800192602519762e-05,
+ "loss": 1.2849,
+ "step": 3370
+ },
+ {
+ "epoch": 9.185286103542234,
+ "grad_norm": 9.103719711303711,
+ "learning_rate": 1.9800017033869033e-05,
+ "loss": 1.1963,
+ "step": 3371
+ },
+ {
+ "epoch": 9.188010899182562,
+ "grad_norm": 7.1451029777526855,
+ "learning_rate": 1.979984138889608e-05,
+ "loss": 1.1132,
+ "step": 3372
+ },
+ {
+ "epoch": 9.190735694822889,
+ "grad_norm": 6.8827223777771,
+ "learning_rate": 1.9799665667602282e-05,
+ "loss": 1.3799,
+ "step": 3373
+ },
+ {
+ "epoch": 9.193460490463215,
+ "grad_norm": 11.939201354980469,
+ "learning_rate": 1.9799489869989003e-05,
+ "loss": 1.2014,
+ "step": 3374
+ },
+ {
+ "epoch": 9.196185286103542,
+ "grad_norm": 12.468143463134766,
+ "learning_rate": 1.979931399605761e-05,
+ "loss": 1.2307,
+ "step": 3375
+ },
+ {
+ "epoch": 9.19891008174387,
+ "grad_norm": 7.666806221008301,
+ "learning_rate": 1.9799138045809482e-05,
+ "loss": 1.2786,
+ "step": 3376
+ },
+ {
+ "epoch": 9.201634877384196,
+ "grad_norm": 8.011543273925781,
+ "learning_rate": 1.979896201924597e-05,
+ "loss": 1.2234,
+ "step": 3377
+ },
+ {
+ "epoch": 9.204359673024523,
+ "grad_norm": 10.653948783874512,
+ "learning_rate": 1.979878591636847e-05,
+ "loss": 1.3811,
+ "step": 3378
+ },
+ {
+ "epoch": 9.207084468664851,
+ "grad_norm": 6.664824962615967,
+ "learning_rate": 1.9798609737178332e-05,
+ "loss": 1.1646,
+ "step": 3379
+ },
+ {
+ "epoch": 9.209809264305177,
+ "grad_norm": 8.63076114654541,
+ "learning_rate": 1.979843348167694e-05,
+ "loss": 1.3516,
+ "step": 3380
+ },
+ {
+ "epoch": 9.212534059945504,
+ "grad_norm": 14.606650352478027,
+ "learning_rate": 1.9798257149865664e-05,
+ "loss": 1.0874,
+ "step": 3381
+ },
+ {
+ "epoch": 9.215258855585832,
+ "grad_norm": 7.923644542694092,
+ "learning_rate": 1.9798080741745876e-05,
+ "loss": 1.3149,
+ "step": 3382
+ },
+ {
+ "epoch": 9.217983651226158,
+ "grad_norm": 10.9000825881958,
+ "learning_rate": 1.979790425731895e-05,
+ "loss": 1.3816,
+ "step": 3383
+ },
+ {
+ "epoch": 9.220708446866485,
+ "grad_norm": 8.52334213256836,
+ "learning_rate": 1.979772769658626e-05,
+ "loss": 1.3682,
+ "step": 3384
+ },
+ {
+ "epoch": 9.223433242506813,
+ "grad_norm": 7.846671104431152,
+ "learning_rate": 1.9797551059549182e-05,
+ "loss": 1.1868,
+ "step": 3385
+ },
+ {
+ "epoch": 9.226158038147139,
+ "grad_norm": 8.460245132446289,
+ "learning_rate": 1.979737434620909e-05,
+ "loss": 1.1506,
+ "step": 3386
+ },
+ {
+ "epoch": 9.228882833787466,
+ "grad_norm": 7.3510918617248535,
+ "learning_rate": 1.9797197556567368e-05,
+ "loss": 1.2625,
+ "step": 3387
+ },
+ {
+ "epoch": 9.231607629427794,
+ "grad_norm": 11.694857597351074,
+ "learning_rate": 1.9797020690625385e-05,
+ "loss": 1.2441,
+ "step": 3388
+ },
+ {
+ "epoch": 9.23433242506812,
+ "grad_norm": 7.855964183807373,
+ "learning_rate": 1.9796843748384516e-05,
+ "loss": 1.2532,
+ "step": 3389
+ },
+ {
+ "epoch": 9.237057220708447,
+ "grad_norm": 7.452142715454102,
+ "learning_rate": 1.979666672984615e-05,
+ "loss": 1.3174,
+ "step": 3390
+ },
+ {
+ "epoch": 9.239782016348773,
+ "grad_norm": 8.646413803100586,
+ "learning_rate": 1.979648963501165e-05,
+ "loss": 1.3127,
+ "step": 3391
+ },
+ {
+ "epoch": 9.2425068119891,
+ "grad_norm": 9.066518783569336,
+ "learning_rate": 1.9796312463882413e-05,
+ "loss": 1.1433,
+ "step": 3392
+ },
+ {
+ "epoch": 9.245231607629428,
+ "grad_norm": 7.602927207946777,
+ "learning_rate": 1.9796135216459806e-05,
+ "loss": 1.1301,
+ "step": 3393
+ },
+ {
+ "epoch": 9.247956403269754,
+ "grad_norm": 8.04531192779541,
+ "learning_rate": 1.9795957892745212e-05,
+ "loss": 1.1848,
+ "step": 3394
+ },
+ {
+ "epoch": 9.250681198910081,
+ "grad_norm": 8.439618110656738,
+ "learning_rate": 1.9795780492740014e-05,
+ "loss": 1.2974,
+ "step": 3395
+ },
+ {
+ "epoch": 9.253405994550409,
+ "grad_norm": 14.157349586486816,
+ "learning_rate": 1.979560301644559e-05,
+ "loss": 1.1624,
+ "step": 3396
+ },
+ {
+ "epoch": 9.256130790190735,
+ "grad_norm": 9.555171012878418,
+ "learning_rate": 1.9795425463863328e-05,
+ "loss": 1.2209,
+ "step": 3397
+ },
+ {
+ "epoch": 9.258855585831062,
+ "grad_norm": 8.269161224365234,
+ "learning_rate": 1.979524783499461e-05,
+ "loss": 1.4299,
+ "step": 3398
+ },
+ {
+ "epoch": 9.26158038147139,
+ "grad_norm": 7.183748722076416,
+ "learning_rate": 1.9795070129840814e-05,
+ "loss": 1.1445,
+ "step": 3399
+ },
+ {
+ "epoch": 9.264305177111716,
+ "grad_norm": 6.645070552825928,
+ "learning_rate": 1.9794892348403327e-05,
+ "loss": 1.1426,
+ "step": 3400
+ },
+ {
+ "epoch": 9.267029972752043,
+ "grad_norm": 8.494816780090332,
+ "learning_rate": 1.9794714490683534e-05,
+ "loss": 1.137,
+ "step": 3401
+ },
+ {
+ "epoch": 9.269754768392371,
+ "grad_norm": 6.0630340576171875,
+ "learning_rate": 1.979453655668282e-05,
+ "loss": 0.9487,
+ "step": 3402
+ },
+ {
+ "epoch": 9.272479564032697,
+ "grad_norm": 5.827259540557861,
+ "learning_rate": 1.979435854640257e-05,
+ "loss": 1.0625,
+ "step": 3403
+ },
+ {
+ "epoch": 9.275204359673024,
+ "grad_norm": 6.8729567527771,
+ "learning_rate": 1.979418045984417e-05,
+ "loss": 1.1641,
+ "step": 3404
+ },
+ {
+ "epoch": 9.277929155313352,
+ "grad_norm": 7.308586120605469,
+ "learning_rate": 1.9794002297009012e-05,
+ "loss": 1.3342,
+ "step": 3405
+ },
+ {
+ "epoch": 9.280653950953678,
+ "grad_norm": 7.857737064361572,
+ "learning_rate": 1.979382405789848e-05,
+ "loss": 1.2334,
+ "step": 3406
+ },
+ {
+ "epoch": 9.283378746594005,
+ "grad_norm": 7.057602882385254,
+ "learning_rate": 1.9793645742513953e-05,
+ "loss": 1.2434,
+ "step": 3407
+ },
+ {
+ "epoch": 9.286103542234333,
+ "grad_norm": 7.977108001708984,
+ "learning_rate": 1.9793467350856838e-05,
+ "loss": 1.2488,
+ "step": 3408
+ },
+ {
+ "epoch": 9.288828337874659,
+ "grad_norm": 8.51467227935791,
+ "learning_rate": 1.979328888292851e-05,
+ "loss": 1.4236,
+ "step": 3409
+ },
+ {
+ "epoch": 9.291553133514986,
+ "grad_norm": 6.123144149780273,
+ "learning_rate": 1.9793110338730362e-05,
+ "loss": 1.1069,
+ "step": 3410
+ },
+ {
+ "epoch": 9.294277929155314,
+ "grad_norm": 8.04094409942627,
+ "learning_rate": 1.9792931718263786e-05,
+ "loss": 1.0613,
+ "step": 3411
+ },
+ {
+ "epoch": 9.29700272479564,
+ "grad_norm": 6.682175636291504,
+ "learning_rate": 1.9792753021530174e-05,
+ "loss": 1.0771,
+ "step": 3412
+ },
+ {
+ "epoch": 9.299727520435967,
+ "grad_norm": 7.542644023895264,
+ "learning_rate": 1.9792574248530915e-05,
+ "loss": 1.3638,
+ "step": 3413
+ },
+ {
+ "epoch": 9.302452316076295,
+ "grad_norm": 6.093067169189453,
+ "learning_rate": 1.9792395399267404e-05,
+ "loss": 0.9479,
+ "step": 3414
+ },
+ {
+ "epoch": 9.30517711171662,
+ "grad_norm": 6.844335079193115,
+ "learning_rate": 1.9792216473741035e-05,
+ "loss": 1.2109,
+ "step": 3415
+ },
+ {
+ "epoch": 9.307901907356948,
+ "grad_norm": 7.803158283233643,
+ "learning_rate": 1.9792037471953196e-05,
+ "loss": 1.3918,
+ "step": 3416
+ },
+ {
+ "epoch": 9.310626702997276,
+ "grad_norm": 6.820953845977783,
+ "learning_rate": 1.9791858393905287e-05,
+ "loss": 1.5508,
+ "step": 3417
+ },
+ {
+ "epoch": 9.313351498637601,
+ "grad_norm": 7.07657527923584,
+ "learning_rate": 1.97916792395987e-05,
+ "loss": 1.3322,
+ "step": 3418
+ },
+ {
+ "epoch": 9.316076294277929,
+ "grad_norm": 8.23022747039795,
+ "learning_rate": 1.979150000903483e-05,
+ "loss": 1.2996,
+ "step": 3419
+ },
+ {
+ "epoch": 9.318801089918257,
+ "grad_norm": 5.375296115875244,
+ "learning_rate": 1.979132070221507e-05,
+ "loss": 0.9929,
+ "step": 3420
+ },
+ {
+ "epoch": 9.321525885558582,
+ "grad_norm": 10.229243278503418,
+ "learning_rate": 1.979114131914082e-05,
+ "loss": 1.3218,
+ "step": 3421
+ },
+ {
+ "epoch": 9.32425068119891,
+ "grad_norm": 6.986461162567139,
+ "learning_rate": 1.979096185981348e-05,
+ "loss": 1.0975,
+ "step": 3422
+ },
+ {
+ "epoch": 9.326975476839237,
+ "grad_norm": 8.433475494384766,
+ "learning_rate": 1.9790782324234442e-05,
+ "loss": 1.1729,
+ "step": 3423
+ },
+ {
+ "epoch": 9.329700272479563,
+ "grad_norm": 6.107909202575684,
+ "learning_rate": 1.9790602712405106e-05,
+ "loss": 1.0486,
+ "step": 3424
+ },
+ {
+ "epoch": 9.33242506811989,
+ "grad_norm": 6.445185661315918,
+ "learning_rate": 1.9790423024326873e-05,
+ "loss": 1.1262,
+ "step": 3425
+ },
+ {
+ "epoch": 9.335149863760218,
+ "grad_norm": 7.288705348968506,
+ "learning_rate": 1.979024326000114e-05,
+ "loss": 1.197,
+ "step": 3426
+ },
+ {
+ "epoch": 9.337874659400544,
+ "grad_norm": 5.79853630065918,
+ "learning_rate": 1.9790063419429307e-05,
+ "loss": 1.187,
+ "step": 3427
+ },
+ {
+ "epoch": 9.340599455040872,
+ "grad_norm": 6.320503234863281,
+ "learning_rate": 1.9789883502612778e-05,
+ "loss": 1.1582,
+ "step": 3428
+ },
+ {
+ "epoch": 9.3433242506812,
+ "grad_norm": 7.147209644317627,
+ "learning_rate": 1.9789703509552947e-05,
+ "loss": 1.2231,
+ "step": 3429
+ },
+ {
+ "epoch": 9.346049046321525,
+ "grad_norm": 8.704099655151367,
+ "learning_rate": 1.9789523440251223e-05,
+ "loss": 1.2361,
+ "step": 3430
+ },
+ {
+ "epoch": 9.348773841961853,
+ "grad_norm": 8.322648048400879,
+ "learning_rate": 1.9789343294709e-05,
+ "loss": 1.0696,
+ "step": 3431
+ },
+ {
+ "epoch": 9.35149863760218,
+ "grad_norm": 6.949365615844727,
+ "learning_rate": 1.9789163072927696e-05,
+ "loss": 1.6848,
+ "step": 3432
+ },
+ {
+ "epoch": 9.354223433242506,
+ "grad_norm": 5.86214017868042,
+ "learning_rate": 1.9788982774908695e-05,
+ "loss": 1.1476,
+ "step": 3433
+ },
+ {
+ "epoch": 9.356948228882834,
+ "grad_norm": 7.259187698364258,
+ "learning_rate": 1.9788802400653416e-05,
+ "loss": 1.1331,
+ "step": 3434
+ },
+ {
+ "epoch": 9.359673024523161,
+ "grad_norm": 6.996931076049805,
+ "learning_rate": 1.9788621950163264e-05,
+ "loss": 1.2363,
+ "step": 3435
+ },
+ {
+ "epoch": 9.362397820163487,
+ "grad_norm": 6.897806644439697,
+ "learning_rate": 1.978844142343963e-05,
+ "loss": 1.2161,
+ "step": 3436
+ },
+ {
+ "epoch": 9.365122615803815,
+ "grad_norm": 5.657095432281494,
+ "learning_rate": 1.9788260820483935e-05,
+ "loss": 1.0845,
+ "step": 3437
+ },
+ {
+ "epoch": 9.367847411444142,
+ "grad_norm": 5.8765058517456055,
+ "learning_rate": 1.9788080141297577e-05,
+ "loss": 0.8861,
+ "step": 3438
+ },
+ {
+ "epoch": 9.370572207084468,
+ "grad_norm": 7.122038841247559,
+ "learning_rate": 1.978789938588197e-05,
+ "loss": 1.4258,
+ "step": 3439
+ },
+ {
+ "epoch": 9.373297002724795,
+ "grad_norm": 7.160912036895752,
+ "learning_rate": 1.9787718554238516e-05,
+ "loss": 1.3088,
+ "step": 3440
+ },
+ {
+ "epoch": 9.376021798365123,
+ "grad_norm": 6.8669562339782715,
+ "learning_rate": 1.978753764636862e-05,
+ "loss": 1.1094,
+ "step": 3441
+ },
+ {
+ "epoch": 9.378746594005449,
+ "grad_norm": 6.353593826293945,
+ "learning_rate": 1.9787356662273702e-05,
+ "loss": 1.0625,
+ "step": 3442
+ },
+ {
+ "epoch": 9.381471389645776,
+ "grad_norm": 5.858900547027588,
+ "learning_rate": 1.9787175601955163e-05,
+ "loss": 1.1562,
+ "step": 3443
+ },
+ {
+ "epoch": 9.384196185286104,
+ "grad_norm": 7.564014911651611,
+ "learning_rate": 1.9786994465414415e-05,
+ "loss": 1.2832,
+ "step": 3444
+ },
+ {
+ "epoch": 9.38692098092643,
+ "grad_norm": 6.015104293823242,
+ "learning_rate": 1.9786813252652866e-05,
+ "loss": 0.9606,
+ "step": 3445
+ },
+ {
+ "epoch": 9.389645776566757,
+ "grad_norm": 6.736959934234619,
+ "learning_rate": 1.9786631963671935e-05,
+ "loss": 1.3354,
+ "step": 3446
+ },
+ {
+ "epoch": 9.392370572207085,
+ "grad_norm": 7.265168190002441,
+ "learning_rate": 1.978645059847303e-05,
+ "loss": 0.9147,
+ "step": 3447
+ },
+ {
+ "epoch": 9.39509536784741,
+ "grad_norm": 5.480492115020752,
+ "learning_rate": 1.978626915705756e-05,
+ "loss": 1.0781,
+ "step": 3448
+ },
+ {
+ "epoch": 9.397820163487738,
+ "grad_norm": 6.731616497039795,
+ "learning_rate": 1.9786087639426942e-05,
+ "loss": 1.5063,
+ "step": 3449
+ },
+ {
+ "epoch": 9.400544959128066,
+ "grad_norm": 7.381555080413818,
+ "learning_rate": 1.978590604558259e-05,
+ "loss": 1.2437,
+ "step": 3450
+ },
+ {
+ "epoch": 9.403269754768392,
+ "grad_norm": 6.2519612312316895,
+ "learning_rate": 1.9785724375525913e-05,
+ "loss": 1.1776,
+ "step": 3451
+ },
+ {
+ "epoch": 9.40599455040872,
+ "grad_norm": 6.502010345458984,
+ "learning_rate": 1.978554262925833e-05,
+ "loss": 1.1321,
+ "step": 3452
+ },
+ {
+ "epoch": 9.408719346049047,
+ "grad_norm": 7.483091354370117,
+ "learning_rate": 1.9785360806781256e-05,
+ "loss": 1.3423,
+ "step": 3453
+ },
+ {
+ "epoch": 9.411444141689373,
+ "grad_norm": 6.490332126617432,
+ "learning_rate": 1.978517890809611e-05,
+ "loss": 1.2178,
+ "step": 3454
+ },
+ {
+ "epoch": 9.4141689373297,
+ "grad_norm": 12.169720649719238,
+ "learning_rate": 1.9784996933204303e-05,
+ "loss": 1.4697,
+ "step": 3455
+ },
+ {
+ "epoch": 9.416893732970028,
+ "grad_norm": 6.885982036590576,
+ "learning_rate": 1.9784814882107256e-05,
+ "loss": 1.1953,
+ "step": 3456
+ },
+ {
+ "epoch": 9.419618528610354,
+ "grad_norm": 5.7435126304626465,
+ "learning_rate": 1.9784632754806385e-05,
+ "loss": 1.1442,
+ "step": 3457
+ },
+ {
+ "epoch": 9.422343324250681,
+ "grad_norm": 6.668470859527588,
+ "learning_rate": 1.978445055130311e-05,
+ "loss": 1.1389,
+ "step": 3458
+ },
+ {
+ "epoch": 9.425068119891009,
+ "grad_norm": 34.67387771606445,
+ "learning_rate": 1.9784268271598847e-05,
+ "loss": 0.9355,
+ "step": 3459
+ },
+ {
+ "epoch": 9.427792915531334,
+ "grad_norm": 6.119838714599609,
+ "learning_rate": 1.978408591569502e-05,
+ "loss": 1.1517,
+ "step": 3460
+ },
+ {
+ "epoch": 9.430517711171662,
+ "grad_norm": 8.593868255615234,
+ "learning_rate": 1.9783903483593045e-05,
+ "loss": 1.0516,
+ "step": 3461
+ },
+ {
+ "epoch": 9.43324250681199,
+ "grad_norm": 7.0536651611328125,
+ "learning_rate": 1.9783720975294345e-05,
+ "loss": 1.1307,
+ "step": 3462
+ },
+ {
+ "epoch": 9.435967302452315,
+ "grad_norm": 5.961587429046631,
+ "learning_rate": 1.978353839080034e-05,
+ "loss": 1.2378,
+ "step": 3463
+ },
+ {
+ "epoch": 9.438692098092643,
+ "grad_norm": 8.953097343444824,
+ "learning_rate": 1.9783355730112454e-05,
+ "loss": 1.352,
+ "step": 3464
+ },
+ {
+ "epoch": 9.44141689373297,
+ "grad_norm": 9.958894729614258,
+ "learning_rate": 1.978317299323211e-05,
+ "loss": 1.2385,
+ "step": 3465
+ },
+ {
+ "epoch": 9.444141689373296,
+ "grad_norm": 7.406508445739746,
+ "learning_rate": 1.9782990180160726e-05,
+ "loss": 1.3145,
+ "step": 3466
+ },
+ {
+ "epoch": 9.446866485013624,
+ "grad_norm": 30.128145217895508,
+ "learning_rate": 1.9782807290899732e-05,
+ "loss": 1.1198,
+ "step": 3467
+ },
+ {
+ "epoch": 9.449591280653951,
+ "grad_norm": 6.777979850769043,
+ "learning_rate": 1.978262432545055e-05,
+ "loss": 1.1243,
+ "step": 3468
+ },
+ {
+ "epoch": 9.452316076294277,
+ "grad_norm": 7.659389972686768,
+ "learning_rate": 1.9782441283814606e-05,
+ "loss": 1.1938,
+ "step": 3469
+ },
+ {
+ "epoch": 9.455040871934605,
+ "grad_norm": 7.506655216217041,
+ "learning_rate": 1.978225816599332e-05,
+ "loss": 1.2649,
+ "step": 3470
+ },
+ {
+ "epoch": 9.457765667574932,
+ "grad_norm": 6.156186580657959,
+ "learning_rate": 1.9782074971988122e-05,
+ "loss": 1.0808,
+ "step": 3471
+ },
+ {
+ "epoch": 9.460490463215258,
+ "grad_norm": 7.348114013671875,
+ "learning_rate": 1.978189170180044e-05,
+ "loss": 1.166,
+ "step": 3472
+ },
+ {
+ "epoch": 9.463215258855586,
+ "grad_norm": 7.6004557609558105,
+ "learning_rate": 1.9781708355431704e-05,
+ "loss": 1.2249,
+ "step": 3473
+ },
+ {
+ "epoch": 9.465940054495913,
+ "grad_norm": 7.202709674835205,
+ "learning_rate": 1.9781524932883334e-05,
+ "loss": 1.0432,
+ "step": 3474
+ },
+ {
+ "epoch": 9.46866485013624,
+ "grad_norm": 7.769924640655518,
+ "learning_rate": 1.978134143415676e-05,
+ "loss": 1.0596,
+ "step": 3475
+ },
+ {
+ "epoch": 9.471389645776567,
+ "grad_norm": 6.260732173919678,
+ "learning_rate": 1.9781157859253422e-05,
+ "loss": 1.1287,
+ "step": 3476
+ },
+ {
+ "epoch": 9.474114441416894,
+ "grad_norm": 15.13672161102295,
+ "learning_rate": 1.9780974208174737e-05,
+ "loss": 1.2578,
+ "step": 3477
+ },
+ {
+ "epoch": 9.47683923705722,
+ "grad_norm": 6.467770099639893,
+ "learning_rate": 1.978079048092214e-05,
+ "loss": 1.1738,
+ "step": 3478
+ },
+ {
+ "epoch": 9.479564032697548,
+ "grad_norm": 12.26695442199707,
+ "learning_rate": 1.978060667749706e-05,
+ "loss": 1.1919,
+ "step": 3479
+ },
+ {
+ "epoch": 9.482288828337875,
+ "grad_norm": 7.249942302703857,
+ "learning_rate": 1.9780422797900935e-05,
+ "loss": 1.1445,
+ "step": 3480
+ },
+ {
+ "epoch": 9.485013623978201,
+ "grad_norm": 7.353272914886475,
+ "learning_rate": 1.9780238842135184e-05,
+ "loss": 1.1394,
+ "step": 3481
+ },
+ {
+ "epoch": 9.487738419618529,
+ "grad_norm": 8.5175199508667,
+ "learning_rate": 1.9780054810201252e-05,
+ "loss": 1.3413,
+ "step": 3482
+ },
+ {
+ "epoch": 9.490463215258856,
+ "grad_norm": 8.30860710144043,
+ "learning_rate": 1.9779870702100566e-05,
+ "loss": 1.1482,
+ "step": 3483
+ },
+ {
+ "epoch": 9.493188010899182,
+ "grad_norm": 6.665664196014404,
+ "learning_rate": 1.9779686517834566e-05,
+ "loss": 1.1045,
+ "step": 3484
+ },
+ {
+ "epoch": 9.49591280653951,
+ "grad_norm": 6.544793128967285,
+ "learning_rate": 1.977950225740468e-05,
+ "loss": 0.9917,
+ "step": 3485
+ },
+ {
+ "epoch": 9.498637602179837,
+ "grad_norm": 6.271844863891602,
+ "learning_rate": 1.9779317920812345e-05,
+ "loss": 1.1927,
+ "step": 3486
+ },
+ {
+ "epoch": 9.501362397820163,
+ "grad_norm": 6.559974193572998,
+ "learning_rate": 1.9779133508058997e-05,
+ "loss": 1.1489,
+ "step": 3487
+ },
+ {
+ "epoch": 9.50408719346049,
+ "grad_norm": 8.81902027130127,
+ "learning_rate": 1.977894901914607e-05,
+ "loss": 1.1729,
+ "step": 3488
+ },
+ {
+ "epoch": 9.506811989100818,
+ "grad_norm": 7.8292622566223145,
+ "learning_rate": 1.9778764454075006e-05,
+ "loss": 1.5208,
+ "step": 3489
+ },
+ {
+ "epoch": 9.509536784741144,
+ "grad_norm": 7.179342269897461,
+ "learning_rate": 1.9778579812847236e-05,
+ "loss": 1.2505,
+ "step": 3490
+ },
+ {
+ "epoch": 9.512261580381471,
+ "grad_norm": 5.537903308868408,
+ "learning_rate": 1.9778395095464204e-05,
+ "loss": 1.0701,
+ "step": 3491
+ },
+ {
+ "epoch": 9.514986376021799,
+ "grad_norm": 8.181119918823242,
+ "learning_rate": 1.9778210301927344e-05,
+ "loss": 1.2039,
+ "step": 3492
+ },
+ {
+ "epoch": 9.517711171662125,
+ "grad_norm": 11.096929550170898,
+ "learning_rate": 1.97780254322381e-05,
+ "loss": 1.1716,
+ "step": 3493
+ },
+ {
+ "epoch": 9.520435967302452,
+ "grad_norm": 7.409578800201416,
+ "learning_rate": 1.9777840486397904e-05,
+ "loss": 1.1899,
+ "step": 3494
+ },
+ {
+ "epoch": 9.52316076294278,
+ "grad_norm": 9.095075607299805,
+ "learning_rate": 1.9777655464408204e-05,
+ "loss": 1.3606,
+ "step": 3495
+ },
+ {
+ "epoch": 9.525885558583106,
+ "grad_norm": 6.521109104156494,
+ "learning_rate": 1.9777470366270436e-05,
+ "loss": 0.9761,
+ "step": 3496
+ },
+ {
+ "epoch": 9.528610354223433,
+ "grad_norm": 8.804996490478516,
+ "learning_rate": 1.9777285191986045e-05,
+ "loss": 1.3833,
+ "step": 3497
+ },
+ {
+ "epoch": 9.53133514986376,
+ "grad_norm": 8.817750930786133,
+ "learning_rate": 1.977709994155647e-05,
+ "loss": 1.1655,
+ "step": 3498
+ },
+ {
+ "epoch": 9.534059945504087,
+ "grad_norm": 7.285440444946289,
+ "learning_rate": 1.9776914614983157e-05,
+ "loss": 1.3995,
+ "step": 3499
+ },
+ {
+ "epoch": 9.536784741144414,
+ "grad_norm": 7.140861988067627,
+ "learning_rate": 1.9776729212267547e-05,
+ "loss": 1.3005,
+ "step": 3500
+ },
+ {
+ "epoch": 9.539509536784742,
+ "grad_norm": 7.087491512298584,
+ "learning_rate": 1.9776543733411084e-05,
+ "loss": 1.348,
+ "step": 3501
+ },
+ {
+ "epoch": 9.542234332425068,
+ "grad_norm": 10.124818801879883,
+ "learning_rate": 1.9776358178415213e-05,
+ "loss": 1.2375,
+ "step": 3502
+ },
+ {
+ "epoch": 9.544959128065395,
+ "grad_norm": 5.689785957336426,
+ "learning_rate": 1.977617254728138e-05,
+ "loss": 1.1404,
+ "step": 3503
+ },
+ {
+ "epoch": 9.547683923705723,
+ "grad_norm": 6.573259353637695,
+ "learning_rate": 1.977598684001103e-05,
+ "loss": 1.1503,
+ "step": 3504
+ },
+ {
+ "epoch": 9.550408719346049,
+ "grad_norm": 6.923213481903076,
+ "learning_rate": 1.977580105660561e-05,
+ "loss": 1.1055,
+ "step": 3505
+ },
+ {
+ "epoch": 9.553133514986376,
+ "grad_norm": 8.691814422607422,
+ "learning_rate": 1.977561519706656e-05,
+ "loss": 1.0996,
+ "step": 3506
+ },
+ {
+ "epoch": 9.555858310626704,
+ "grad_norm": 9.01028060913086,
+ "learning_rate": 1.9775429261395343e-05,
+ "loss": 1.2131,
+ "step": 3507
+ },
+ {
+ "epoch": 9.55858310626703,
+ "grad_norm": 6.702378273010254,
+ "learning_rate": 1.977524324959339e-05,
+ "loss": 0.9832,
+ "step": 3508
+ },
+ {
+ "epoch": 9.561307901907357,
+ "grad_norm": 7.559467792510986,
+ "learning_rate": 1.9775057161662163e-05,
+ "loss": 1.3809,
+ "step": 3509
+ },
+ {
+ "epoch": 9.564032697547685,
+ "grad_norm": 9.303908348083496,
+ "learning_rate": 1.97748709976031e-05,
+ "loss": 1.3486,
+ "step": 3510
+ },
+ {
+ "epoch": 9.56675749318801,
+ "grad_norm": 8.965313911437988,
+ "learning_rate": 1.977468475741766e-05,
+ "loss": 1.2686,
+ "step": 3511
+ },
+ {
+ "epoch": 9.569482288828338,
+ "grad_norm": 8.924571990966797,
+ "learning_rate": 1.9774498441107288e-05,
+ "loss": 1.3508,
+ "step": 3512
+ },
+ {
+ "epoch": 9.572207084468666,
+ "grad_norm": 9.311145782470703,
+ "learning_rate": 1.9774312048673438e-05,
+ "loss": 0.986,
+ "step": 3513
+ },
+ {
+ "epoch": 9.574931880108991,
+ "grad_norm": 6.401684284210205,
+ "learning_rate": 1.9774125580117557e-05,
+ "loss": 1.1914,
+ "step": 3514
+ },
+ {
+ "epoch": 9.577656675749319,
+ "grad_norm": 7.010136127471924,
+ "learning_rate": 1.9773939035441106e-05,
+ "loss": 1.1519,
+ "step": 3515
+ },
+ {
+ "epoch": 9.580381471389646,
+ "grad_norm": 9.271431922912598,
+ "learning_rate": 1.9773752414645527e-05,
+ "loss": 1.2739,
+ "step": 3516
+ },
+ {
+ "epoch": 9.583106267029972,
+ "grad_norm": 6.545601844787598,
+ "learning_rate": 1.977356571773228e-05,
+ "loss": 1.2871,
+ "step": 3517
+ },
+ {
+ "epoch": 9.5858310626703,
+ "grad_norm": 7.4237542152404785,
+ "learning_rate": 1.9773378944702818e-05,
+ "loss": 1.1187,
+ "step": 3518
+ },
+ {
+ "epoch": 9.588555858310627,
+ "grad_norm": 7.374046802520752,
+ "learning_rate": 1.97731920955586e-05,
+ "loss": 1.1042,
+ "step": 3519
+ },
+ {
+ "epoch": 9.591280653950953,
+ "grad_norm": 8.895092010498047,
+ "learning_rate": 1.9773005170301072e-05,
+ "loss": 1.2362,
+ "step": 3520
+ },
+ {
+ "epoch": 9.59400544959128,
+ "grad_norm": 6.175750732421875,
+ "learning_rate": 1.9772818168931693e-05,
+ "loss": 1.1533,
+ "step": 3521
+ },
+ {
+ "epoch": 9.596730245231608,
+ "grad_norm": 7.412048816680908,
+ "learning_rate": 1.9772631091451924e-05,
+ "loss": 1.1963,
+ "step": 3522
+ },
+ {
+ "epoch": 9.599455040871934,
+ "grad_norm": 6.138168811798096,
+ "learning_rate": 1.9772443937863216e-05,
+ "loss": 1.1604,
+ "step": 3523
+ },
+ {
+ "epoch": 9.602179836512262,
+ "grad_norm": 12.526262283325195,
+ "learning_rate": 1.977225670816703e-05,
+ "loss": 1.2441,
+ "step": 3524
+ },
+ {
+ "epoch": 9.60490463215259,
+ "grad_norm": 6.384923934936523,
+ "learning_rate": 1.9772069402364823e-05,
+ "loss": 1.1567,
+ "step": 3525
+ },
+ {
+ "epoch": 9.607629427792915,
+ "grad_norm": 10.761649131774902,
+ "learning_rate": 1.9771882020458055e-05,
+ "loss": 1.2649,
+ "step": 3526
+ },
+ {
+ "epoch": 9.610354223433243,
+ "grad_norm": 7.394378662109375,
+ "learning_rate": 1.9771694562448186e-05,
+ "loss": 1.4949,
+ "step": 3527
+ },
+ {
+ "epoch": 9.61307901907357,
+ "grad_norm": 12.486881256103516,
+ "learning_rate": 1.9771507028336672e-05,
+ "loss": 1.0554,
+ "step": 3528
+ },
+ {
+ "epoch": 9.615803814713896,
+ "grad_norm": 8.686569213867188,
+ "learning_rate": 1.9771319418124974e-05,
+ "loss": 1.1133,
+ "step": 3529
+ },
+ {
+ "epoch": 9.618528610354224,
+ "grad_norm": 5.500112056732178,
+ "learning_rate": 1.9771131731814558e-05,
+ "loss": 1.0767,
+ "step": 3530
+ },
+ {
+ "epoch": 9.621253405994551,
+ "grad_norm": 6.697922229766846,
+ "learning_rate": 1.9770943969406882e-05,
+ "loss": 1.2058,
+ "step": 3531
+ },
+ {
+ "epoch": 9.623978201634877,
+ "grad_norm": 6.648760795593262,
+ "learning_rate": 1.977075613090341e-05,
+ "loss": 1.1871,
+ "step": 3532
+ },
+ {
+ "epoch": 9.626702997275205,
+ "grad_norm": 7.214866638183594,
+ "learning_rate": 1.97705682163056e-05,
+ "loss": 1.0613,
+ "step": 3533
+ },
+ {
+ "epoch": 9.629427792915532,
+ "grad_norm": 6.812121391296387,
+ "learning_rate": 1.977038022561492e-05,
+ "loss": 1.1538,
+ "step": 3534
+ },
+ {
+ "epoch": 9.632152588555858,
+ "grad_norm": 6.176699638366699,
+ "learning_rate": 1.9770192158832837e-05,
+ "loss": 1.1719,
+ "step": 3535
+ },
+ {
+ "epoch": 9.634877384196185,
+ "grad_norm": 5.683945655822754,
+ "learning_rate": 1.9770004015960808e-05,
+ "loss": 1.1552,
+ "step": 3536
+ },
+ {
+ "epoch": 9.637602179836513,
+ "grad_norm": 40.2102165222168,
+ "learning_rate": 1.9769815797000303e-05,
+ "loss": 1.0234,
+ "step": 3537
+ },
+ {
+ "epoch": 9.640326975476839,
+ "grad_norm": 10.819061279296875,
+ "learning_rate": 1.9769627501952787e-05,
+ "loss": 1.1046,
+ "step": 3538
+ },
+ {
+ "epoch": 9.643051771117166,
+ "grad_norm": 7.18360710144043,
+ "learning_rate": 1.9769439130819727e-05,
+ "loss": 1.1526,
+ "step": 3539
+ },
+ {
+ "epoch": 9.645776566757494,
+ "grad_norm": 8.572402954101562,
+ "learning_rate": 1.976925068360259e-05,
+ "loss": 1.1509,
+ "step": 3540
+ },
+ {
+ "epoch": 9.64850136239782,
+ "grad_norm": 5.53800630569458,
+ "learning_rate": 1.976906216030284e-05,
+ "loss": 1.1333,
+ "step": 3541
+ },
+ {
+ "epoch": 9.651226158038147,
+ "grad_norm": 6.445679664611816,
+ "learning_rate": 1.9768873560921954e-05,
+ "loss": 1.175,
+ "step": 3542
+ },
+ {
+ "epoch": 9.653950953678475,
+ "grad_norm": 7.860922813415527,
+ "learning_rate": 1.9768684885461394e-05,
+ "loss": 1.2639,
+ "step": 3543
+ },
+ {
+ "epoch": 9.6566757493188,
+ "grad_norm": 7.3368706703186035,
+ "learning_rate": 1.9768496133922626e-05,
+ "loss": 1.2769,
+ "step": 3544
+ },
+ {
+ "epoch": 9.659400544959128,
+ "grad_norm": 5.389446258544922,
+ "learning_rate": 1.9768307306307125e-05,
+ "loss": 1.1731,
+ "step": 3545
+ },
+ {
+ "epoch": 9.662125340599456,
+ "grad_norm": 6.950031280517578,
+ "learning_rate": 1.9768118402616367e-05,
+ "loss": 1.2444,
+ "step": 3546
+ },
+ {
+ "epoch": 9.664850136239782,
+ "grad_norm": 5.864431381225586,
+ "learning_rate": 1.9767929422851814e-05,
+ "loss": 1.162,
+ "step": 3547
+ },
+ {
+ "epoch": 9.66757493188011,
+ "grad_norm": 7.1974287033081055,
+ "learning_rate": 1.9767740367014937e-05,
+ "loss": 1.083,
+ "step": 3548
+ },
+ {
+ "epoch": 9.670299727520437,
+ "grad_norm": 8.36790657043457,
+ "learning_rate": 1.9767551235107218e-05,
+ "loss": 1.2568,
+ "step": 3549
+ },
+ {
+ "epoch": 9.673024523160763,
+ "grad_norm": 8.978728294372559,
+ "learning_rate": 1.9767362027130122e-05,
+ "loss": 1.0459,
+ "step": 3550
+ },
+ {
+ "epoch": 9.67574931880109,
+ "grad_norm": 7.60194206237793,
+ "learning_rate": 1.9767172743085124e-05,
+ "loss": 1.385,
+ "step": 3551
+ },
+ {
+ "epoch": 9.678474114441418,
+ "grad_norm": 6.440871238708496,
+ "learning_rate": 1.97669833829737e-05,
+ "loss": 1.0022,
+ "step": 3552
+ },
+ {
+ "epoch": 9.681198910081743,
+ "grad_norm": 5.703713893890381,
+ "learning_rate": 1.9766793946797324e-05,
+ "loss": 1.1082,
+ "step": 3553
+ },
+ {
+ "epoch": 9.683923705722071,
+ "grad_norm": 7.382491111755371,
+ "learning_rate": 1.976660443455747e-05,
+ "loss": 1.229,
+ "step": 3554
+ },
+ {
+ "epoch": 9.686648501362399,
+ "grad_norm": 8.599220275878906,
+ "learning_rate": 1.9766414846255614e-05,
+ "loss": 1.2101,
+ "step": 3555
+ },
+ {
+ "epoch": 9.689373297002724,
+ "grad_norm": 8.160773277282715,
+ "learning_rate": 1.9766225181893236e-05,
+ "loss": 1.2644,
+ "step": 3556
+ },
+ {
+ "epoch": 9.692098092643052,
+ "grad_norm": 7.109304904937744,
+ "learning_rate": 1.9766035441471813e-05,
+ "loss": 1.15,
+ "step": 3557
+ },
+ {
+ "epoch": 9.69482288828338,
+ "grad_norm": 7.16922664642334,
+ "learning_rate": 1.9765845624992814e-05,
+ "loss": 1.1599,
+ "step": 3558
+ },
+ {
+ "epoch": 9.697547683923705,
+ "grad_norm": 6.349972724914551,
+ "learning_rate": 1.9765655732457728e-05,
+ "loss": 1.2817,
+ "step": 3559
+ },
+ {
+ "epoch": 9.700272479564033,
+ "grad_norm": 7.150129318237305,
+ "learning_rate": 1.9765465763868027e-05,
+ "loss": 1.156,
+ "step": 3560
+ },
+ {
+ "epoch": 9.70299727520436,
+ "grad_norm": 5.980002403259277,
+ "learning_rate": 1.9765275719225196e-05,
+ "loss": 1.1072,
+ "step": 3561
+ },
+ {
+ "epoch": 9.705722070844686,
+ "grad_norm": 8.498557090759277,
+ "learning_rate": 1.976508559853071e-05,
+ "loss": 1.3396,
+ "step": 3562
+ },
+ {
+ "epoch": 9.708446866485014,
+ "grad_norm": 7.218272686004639,
+ "learning_rate": 1.9764895401786054e-05,
+ "loss": 1.3105,
+ "step": 3563
+ },
+ {
+ "epoch": 9.711171662125341,
+ "grad_norm": 10.41379451751709,
+ "learning_rate": 1.9764705128992704e-05,
+ "loss": 1.2356,
+ "step": 3564
+ },
+ {
+ "epoch": 9.713896457765667,
+ "grad_norm": 7.281015872955322,
+ "learning_rate": 1.9764514780152146e-05,
+ "loss": 1.2336,
+ "step": 3565
+ },
+ {
+ "epoch": 9.716621253405995,
+ "grad_norm": 10.16819953918457,
+ "learning_rate": 1.976432435526586e-05,
+ "loss": 1.1453,
+ "step": 3566
+ },
+ {
+ "epoch": 9.719346049046322,
+ "grad_norm": 6.634669780731201,
+ "learning_rate": 1.976413385433533e-05,
+ "loss": 1.1531,
+ "step": 3567
+ },
+ {
+ "epoch": 9.722070844686648,
+ "grad_norm": 5.786578178405762,
+ "learning_rate": 1.976394327736204e-05,
+ "loss": 1.1741,
+ "step": 3568
+ },
+ {
+ "epoch": 9.724795640326976,
+ "grad_norm": 13.40177059173584,
+ "learning_rate": 1.9763752624347475e-05,
+ "loss": 1.071,
+ "step": 3569
+ },
+ {
+ "epoch": 9.727520435967303,
+ "grad_norm": 7.472966194152832,
+ "learning_rate": 1.976356189529312e-05,
+ "loss": 1.04,
+ "step": 3570
+ },
+ {
+ "epoch": 9.730245231607629,
+ "grad_norm": 13.058670997619629,
+ "learning_rate": 1.9763371090200456e-05,
+ "loss": 1.1873,
+ "step": 3571
+ },
+ {
+ "epoch": 9.732970027247957,
+ "grad_norm": 8.645931243896484,
+ "learning_rate": 1.9763180209070977e-05,
+ "loss": 1.1304,
+ "step": 3572
+ },
+ {
+ "epoch": 9.735694822888284,
+ "grad_norm": 7.86329460144043,
+ "learning_rate": 1.976298925190616e-05,
+ "loss": 1.1063,
+ "step": 3573
+ },
+ {
+ "epoch": 9.73841961852861,
+ "grad_norm": 7.650310516357422,
+ "learning_rate": 1.9762798218707496e-05,
+ "loss": 1.0797,
+ "step": 3574
+ },
+ {
+ "epoch": 9.741144414168938,
+ "grad_norm": 8.478599548339844,
+ "learning_rate": 1.976260710947648e-05,
+ "loss": 1.3689,
+ "step": 3575
+ },
+ {
+ "epoch": 9.743869209809265,
+ "grad_norm": 7.0309576988220215,
+ "learning_rate": 1.976241592421459e-05,
+ "loss": 1.2605,
+ "step": 3576
+ },
+ {
+ "epoch": 9.746594005449591,
+ "grad_norm": 7.315009117126465,
+ "learning_rate": 1.976222466292332e-05,
+ "loss": 1.2374,
+ "step": 3577
+ },
+ {
+ "epoch": 9.749318801089919,
+ "grad_norm": 6.303478240966797,
+ "learning_rate": 1.9762033325604156e-05,
+ "loss": 1.1309,
+ "step": 3578
+ },
+ {
+ "epoch": 9.752043596730246,
+ "grad_norm": 7.549428462982178,
+ "learning_rate": 1.9761841912258594e-05,
+ "loss": 1.1725,
+ "step": 3579
+ },
+ {
+ "epoch": 9.754768392370572,
+ "grad_norm": 6.831321716308594,
+ "learning_rate": 1.9761650422888118e-05,
+ "loss": 1.121,
+ "step": 3580
+ },
+ {
+ "epoch": 9.7574931880109,
+ "grad_norm": 6.770758628845215,
+ "learning_rate": 1.9761458857494223e-05,
+ "loss": 1.231,
+ "step": 3581
+ },
+ {
+ "epoch": 9.760217983651227,
+ "grad_norm": 35.611228942871094,
+ "learning_rate": 1.9761267216078403e-05,
+ "loss": 1.1985,
+ "step": 3582
+ },
+ {
+ "epoch": 9.762942779291553,
+ "grad_norm": 8.895747184753418,
+ "learning_rate": 1.9761075498642144e-05,
+ "loss": 1.2332,
+ "step": 3583
+ },
+ {
+ "epoch": 9.76566757493188,
+ "grad_norm": 13.806024551391602,
+ "learning_rate": 1.976088370518695e-05,
+ "loss": 1.1577,
+ "step": 3584
+ },
+ {
+ "epoch": 9.768392370572208,
+ "grad_norm": 6.128435134887695,
+ "learning_rate": 1.9760691835714304e-05,
+ "loss": 1.2043,
+ "step": 3585
+ },
+ {
+ "epoch": 9.771117166212534,
+ "grad_norm": 8.102156639099121,
+ "learning_rate": 1.9760499890225704e-05,
+ "loss": 1.0107,
+ "step": 3586
+ },
+ {
+ "epoch": 9.773841961852861,
+ "grad_norm": 8.435470581054688,
+ "learning_rate": 1.9760307868722646e-05,
+ "loss": 1.1458,
+ "step": 3587
+ },
+ {
+ "epoch": 9.776566757493189,
+ "grad_norm": 8.517647743225098,
+ "learning_rate": 1.9760115771206624e-05,
+ "loss": 1.0186,
+ "step": 3588
+ },
+ {
+ "epoch": 9.779291553133515,
+ "grad_norm": 8.50827407836914,
+ "learning_rate": 1.9759923597679133e-05,
+ "loss": 1.1276,
+ "step": 3589
+ },
+ {
+ "epoch": 9.782016348773842,
+ "grad_norm": 8.346512794494629,
+ "learning_rate": 1.9759731348141673e-05,
+ "loss": 1.2394,
+ "step": 3590
+ },
+ {
+ "epoch": 9.78474114441417,
+ "grad_norm": 25.931093215942383,
+ "learning_rate": 1.975953902259574e-05,
+ "loss": 1.238,
+ "step": 3591
+ },
+ {
+ "epoch": 9.787465940054496,
+ "grad_norm": 6.544132709503174,
+ "learning_rate": 1.9759346621042827e-05,
+ "loss": 1.2471,
+ "step": 3592
+ },
+ {
+ "epoch": 9.790190735694823,
+ "grad_norm": 9.871321678161621,
+ "learning_rate": 1.9759154143484443e-05,
+ "loss": 1.2117,
+ "step": 3593
+ },
+ {
+ "epoch": 9.79291553133515,
+ "grad_norm": 7.717954158782959,
+ "learning_rate": 1.9758961589922078e-05,
+ "loss": 1.0868,
+ "step": 3594
+ },
+ {
+ "epoch": 9.795640326975477,
+ "grad_norm": 9.126582145690918,
+ "learning_rate": 1.975876896035723e-05,
+ "loss": 1.3408,
+ "step": 3595
+ },
+ {
+ "epoch": 9.798365122615804,
+ "grad_norm": 6.443045139312744,
+ "learning_rate": 1.975857625479141e-05,
+ "loss": 1.1343,
+ "step": 3596
+ },
+ {
+ "epoch": 9.80108991825613,
+ "grad_norm": 7.2004780769348145,
+ "learning_rate": 1.9758383473226107e-05,
+ "loss": 1.2589,
+ "step": 3597
+ },
+ {
+ "epoch": 9.803814713896458,
+ "grad_norm": 6.678862571716309,
+ "learning_rate": 1.9758190615662827e-05,
+ "loss": 1.0693,
+ "step": 3598
+ },
+ {
+ "epoch": 9.806539509536785,
+ "grad_norm": 9.836601257324219,
+ "learning_rate": 1.9757997682103077e-05,
+ "loss": 1.067,
+ "step": 3599
+ },
+ {
+ "epoch": 9.809264305177111,
+ "grad_norm": 7.376387596130371,
+ "learning_rate": 1.975780467254835e-05,
+ "loss": 1.1204,
+ "step": 3600
+ },
+ {
+ "epoch": 9.811989100817438,
+ "grad_norm": 95.54757690429688,
+ "learning_rate": 1.9757611587000164e-05,
+ "loss": 1.1755,
+ "step": 3601
+ },
+ {
+ "epoch": 9.814713896457766,
+ "grad_norm": 7.1106438636779785,
+ "learning_rate": 1.9757418425460005e-05,
+ "loss": 1.3003,
+ "step": 3602
+ },
+ {
+ "epoch": 9.817438692098092,
+ "grad_norm": 10.430929183959961,
+ "learning_rate": 1.9757225187929387e-05,
+ "loss": 1.3706,
+ "step": 3603
+ },
+ {
+ "epoch": 9.82016348773842,
+ "grad_norm": 8.28550910949707,
+ "learning_rate": 1.9757031874409813e-05,
+ "loss": 1.4526,
+ "step": 3604
+ },
+ {
+ "epoch": 9.822888283378747,
+ "grad_norm": 10.364866256713867,
+ "learning_rate": 1.975683848490279e-05,
+ "loss": 1.3821,
+ "step": 3605
+ },
+ {
+ "epoch": 9.825613079019073,
+ "grad_norm": 8.871850967407227,
+ "learning_rate": 1.975664501940982e-05,
+ "loss": 1.2964,
+ "step": 3606
+ },
+ {
+ "epoch": 9.8283378746594,
+ "grad_norm": 9.014577865600586,
+ "learning_rate": 1.9756451477932414e-05,
+ "loss": 1.2812,
+ "step": 3607
+ },
+ {
+ "epoch": 9.831062670299728,
+ "grad_norm": 6.911698818206787,
+ "learning_rate": 1.975625786047208e-05,
+ "loss": 1.0063,
+ "step": 3608
+ },
+ {
+ "epoch": 9.833787465940054,
+ "grad_norm": 6.666546821594238,
+ "learning_rate": 1.9756064167030324e-05,
+ "loss": 1.1276,
+ "step": 3609
+ },
+ {
+ "epoch": 9.836512261580381,
+ "grad_norm": 22.350099563598633,
+ "learning_rate": 1.975587039760865e-05,
+ "loss": 1.1252,
+ "step": 3610
+ },
+ {
+ "epoch": 9.839237057220709,
+ "grad_norm": 10.239272117614746,
+ "learning_rate": 1.9755676552208577e-05,
+ "loss": 1.1505,
+ "step": 3611
+ },
+ {
+ "epoch": 9.841961852861035,
+ "grad_norm": 9.355154991149902,
+ "learning_rate": 1.97554826308316e-05,
+ "loss": 1.2861,
+ "step": 3612
+ },
+ {
+ "epoch": 9.844686648501362,
+ "grad_norm": 7.087742805480957,
+ "learning_rate": 1.9755288633479245e-05,
+ "loss": 1.3713,
+ "step": 3613
+ },
+ {
+ "epoch": 9.84741144414169,
+ "grad_norm": 8.15880298614502,
+ "learning_rate": 1.975509456015302e-05,
+ "loss": 1.3054,
+ "step": 3614
+ },
+ {
+ "epoch": 9.850136239782016,
+ "grad_norm": 8.04164981842041,
+ "learning_rate": 1.9754900410854425e-05,
+ "loss": 1.1138,
+ "step": 3615
+ },
+ {
+ "epoch": 9.852861035422343,
+ "grad_norm": 11.656779289245605,
+ "learning_rate": 1.9754706185584982e-05,
+ "loss": 1.1033,
+ "step": 3616
+ },
+ {
+ "epoch": 9.85558583106267,
+ "grad_norm": 8.02888011932373,
+ "learning_rate": 1.97545118843462e-05,
+ "loss": 1.4253,
+ "step": 3617
+ },
+ {
+ "epoch": 9.858310626702997,
+ "grad_norm": 10.126209259033203,
+ "learning_rate": 1.9754317507139596e-05,
+ "loss": 1.3455,
+ "step": 3618
+ },
+ {
+ "epoch": 9.861035422343324,
+ "grad_norm": 11.216506004333496,
+ "learning_rate": 1.9754123053966682e-05,
+ "loss": 1.3361,
+ "step": 3619
+ },
+ {
+ "epoch": 9.863760217983652,
+ "grad_norm": 8.01203441619873,
+ "learning_rate": 1.9753928524828967e-05,
+ "loss": 1.2095,
+ "step": 3620
+ },
+ {
+ "epoch": 9.866485013623977,
+ "grad_norm": 7.592010021209717,
+ "learning_rate": 1.975373391972797e-05,
+ "loss": 1.2737,
+ "step": 3621
+ },
+ {
+ "epoch": 9.869209809264305,
+ "grad_norm": 6.2159857749938965,
+ "learning_rate": 1.9753539238665214e-05,
+ "loss": 1.0996,
+ "step": 3622
+ },
+ {
+ "epoch": 9.871934604904633,
+ "grad_norm": 7.8470988273620605,
+ "learning_rate": 1.9753344481642205e-05,
+ "loss": 1.0559,
+ "step": 3623
+ },
+ {
+ "epoch": 9.874659400544958,
+ "grad_norm": 8.86409854888916,
+ "learning_rate": 1.9753149648660463e-05,
+ "loss": 1.1125,
+ "step": 3624
+ },
+ {
+ "epoch": 9.877384196185286,
+ "grad_norm": 10.926403045654297,
+ "learning_rate": 1.9752954739721506e-05,
+ "loss": 1.1628,
+ "step": 3625
+ },
+ {
+ "epoch": 9.880108991825614,
+ "grad_norm": 8.127362251281738,
+ "learning_rate": 1.9752759754826848e-05,
+ "loss": 1.5154,
+ "step": 3626
+ },
+ {
+ "epoch": 9.88283378746594,
+ "grad_norm": 8.736924171447754,
+ "learning_rate": 1.9752564693978017e-05,
+ "loss": 1.3979,
+ "step": 3627
+ },
+ {
+ "epoch": 9.885558583106267,
+ "grad_norm": 7.401130676269531,
+ "learning_rate": 1.975236955717652e-05,
+ "loss": 1.0859,
+ "step": 3628
+ },
+ {
+ "epoch": 9.888283378746594,
+ "grad_norm": 3481.053466796875,
+ "learning_rate": 1.9752174344423887e-05,
+ "loss": 1.4897,
+ "step": 3629
+ },
+ {
+ "epoch": 9.89100817438692,
+ "grad_norm": 12.448040962219238,
+ "learning_rate": 1.975197905572163e-05,
+ "loss": 1.3945,
+ "step": 3630
+ },
+ {
+ "epoch": 9.893732970027248,
+ "grad_norm": 11.228221893310547,
+ "learning_rate": 1.9751783691071278e-05,
+ "loss": 1.3674,
+ "step": 3631
+ },
+ {
+ "epoch": 9.896457765667575,
+ "grad_norm": 11.56096076965332,
+ "learning_rate": 1.9751588250474347e-05,
+ "loss": 1.1636,
+ "step": 3632
+ },
+ {
+ "epoch": 9.899182561307901,
+ "grad_norm": 20.25699806213379,
+ "learning_rate": 1.975139273393236e-05,
+ "loss": 1.3604,
+ "step": 3633
+ },
+ {
+ "epoch": 9.901907356948229,
+ "grad_norm": 24.01834487915039,
+ "learning_rate": 1.975119714144684e-05,
+ "loss": 1.5735,
+ "step": 3634
+ },
+ {
+ "epoch": 9.904632152588556,
+ "grad_norm": 31.29701042175293,
+ "learning_rate": 1.9751001473019312e-05,
+ "loss": 1.8545,
+ "step": 3635
+ },
+ {
+ "epoch": 9.907356948228882,
+ "grad_norm": 19.145599365234375,
+ "learning_rate": 1.97508057286513e-05,
+ "loss": 1.4712,
+ "step": 3636
+ },
+ {
+ "epoch": 9.91008174386921,
+ "grad_norm": 18.91514778137207,
+ "learning_rate": 1.975060990834432e-05,
+ "loss": 1.8567,
+ "step": 3637
+ },
+ {
+ "epoch": 9.912806539509537,
+ "grad_norm": 22.237411499023438,
+ "learning_rate": 1.975041401209991e-05,
+ "loss": 1.8042,
+ "step": 3638
+ },
+ {
+ "epoch": 9.915531335149863,
+ "grad_norm": 18.719974517822266,
+ "learning_rate": 1.9750218039919587e-05,
+ "loss": 1.5715,
+ "step": 3639
+ },
+ {
+ "epoch": 9.91825613079019,
+ "grad_norm": 18.660114288330078,
+ "learning_rate": 1.975002199180488e-05,
+ "loss": 1.7932,
+ "step": 3640
+ },
+ {
+ "epoch": 9.920980926430518,
+ "grad_norm": 13.645734786987305,
+ "learning_rate": 1.9749825867757316e-05,
+ "loss": 1.4287,
+ "step": 3641
+ },
+ {
+ "epoch": 9.923705722070844,
+ "grad_norm": 18.809452056884766,
+ "learning_rate": 1.974962966777842e-05,
+ "loss": 1.8418,
+ "step": 3642
+ },
+ {
+ "epoch": 9.926430517711172,
+ "grad_norm": 17.151277542114258,
+ "learning_rate": 1.9749433391869725e-05,
+ "loss": 1.6538,
+ "step": 3643
+ },
+ {
+ "epoch": 9.9291553133515,
+ "grad_norm": 15.19141674041748,
+ "learning_rate": 1.9749237040032758e-05,
+ "loss": 1.4712,
+ "step": 3644
+ },
+ {
+ "epoch": 9.931880108991825,
+ "grad_norm": 15.529809951782227,
+ "learning_rate": 1.974904061226904e-05,
+ "loss": 1.5632,
+ "step": 3645
+ },
+ {
+ "epoch": 9.934604904632153,
+ "grad_norm": 13.362323760986328,
+ "learning_rate": 1.9748844108580113e-05,
+ "loss": 1.5332,
+ "step": 3646
+ },
+ {
+ "epoch": 9.93732970027248,
+ "grad_norm": 15.601862907409668,
+ "learning_rate": 1.9748647528967503e-05,
+ "loss": 1.7017,
+ "step": 3647
+ },
+ {
+ "epoch": 9.940054495912806,
+ "grad_norm": 10.198585510253906,
+ "learning_rate": 1.9748450873432737e-05,
+ "loss": 1.1887,
+ "step": 3648
+ },
+ {
+ "epoch": 9.942779291553133,
+ "grad_norm": 12.278666496276855,
+ "learning_rate": 1.974825414197735e-05,
+ "loss": 1.5374,
+ "step": 3649
+ },
+ {
+ "epoch": 9.945504087193461,
+ "grad_norm": 17.62723159790039,
+ "learning_rate": 1.9748057334602873e-05,
+ "loss": 1.1973,
+ "step": 3650
+ },
+ {
+ "epoch": 9.948228882833787,
+ "grad_norm": 10.020537376403809,
+ "learning_rate": 1.9747860451310838e-05,
+ "loss": 1.4282,
+ "step": 3651
+ },
+ {
+ "epoch": 9.950953678474114,
+ "grad_norm": 103.85107421875,
+ "learning_rate": 1.974766349210278e-05,
+ "loss": 1.3,
+ "step": 3652
+ },
+ {
+ "epoch": 9.953678474114442,
+ "grad_norm": 11.39072036743164,
+ "learning_rate": 1.9747466456980232e-05,
+ "loss": 1.2866,
+ "step": 3653
+ },
+ {
+ "epoch": 9.956403269754768,
+ "grad_norm": 16.906389236450195,
+ "learning_rate": 1.974726934594473e-05,
+ "loss": 1.624,
+ "step": 3654
+ },
+ {
+ "epoch": 9.959128065395095,
+ "grad_norm": 25.58646011352539,
+ "learning_rate": 1.9747072158997808e-05,
+ "loss": 1.6707,
+ "step": 3655
+ },
+ {
+ "epoch": 9.961852861035423,
+ "grad_norm": 15.000348091125488,
+ "learning_rate": 1.9746874896141005e-05,
+ "loss": 1.5266,
+ "step": 3656
+ },
+ {
+ "epoch": 9.964577656675749,
+ "grad_norm": 19.849376678466797,
+ "learning_rate": 1.9746677557375847e-05,
+ "loss": 1.4927,
+ "step": 3657
+ },
+ {
+ "epoch": 9.967302452316076,
+ "grad_norm": 25.91085433959961,
+ "learning_rate": 1.9746480142703883e-05,
+ "loss": 1.488,
+ "step": 3658
+ },
+ {
+ "epoch": 9.970027247956404,
+ "grad_norm": 16.064367294311523,
+ "learning_rate": 1.9746282652126646e-05,
+ "loss": 1.3398,
+ "step": 3659
+ },
+ {
+ "epoch": 9.97275204359673,
+ "grad_norm": 14.867619514465332,
+ "learning_rate": 1.974608508564567e-05,
+ "loss": 1.4592,
+ "step": 3660
+ },
+ {
+ "epoch": 9.975476839237057,
+ "grad_norm": 11.469196319580078,
+ "learning_rate": 1.97458874432625e-05,
+ "loss": 1.3716,
+ "step": 3661
+ },
+ {
+ "epoch": 9.978201634877385,
+ "grad_norm": 13.643699645996094,
+ "learning_rate": 1.974568972497867e-05,
+ "loss": 1.6055,
+ "step": 3662
+ },
+ {
+ "epoch": 9.98092643051771,
+ "grad_norm": 15.681665420532227,
+ "learning_rate": 1.9745491930795725e-05,
+ "loss": 1.4609,
+ "step": 3663
+ },
+ {
+ "epoch": 9.983651226158038,
+ "grad_norm": 16.954286575317383,
+ "learning_rate": 1.97452940607152e-05,
+ "loss": 1.4668,
+ "step": 3664
+ },
+ {
+ "epoch": 9.986376021798366,
+ "grad_norm": 12.205987930297852,
+ "learning_rate": 1.9745096114738637e-05,
+ "loss": 1.6926,
+ "step": 3665
+ },
+ {
+ "epoch": 9.989100817438691,
+ "grad_norm": 19.05699348449707,
+ "learning_rate": 1.9744898092867584e-05,
+ "loss": 1.4529,
+ "step": 3666
+ },
+ {
+ "epoch": 9.991825613079019,
+ "grad_norm": 14.771956443786621,
+ "learning_rate": 1.9744699995103574e-05,
+ "loss": 1.2874,
+ "step": 3667
+ },
+ {
+ "epoch": 9.994550408719347,
+ "grad_norm": 14.152657508850098,
+ "learning_rate": 1.9744501821448154e-05,
+ "loss": 1.6135,
+ "step": 3668
+ },
+ {
+ "epoch": 9.997275204359672,
+ "grad_norm": 15.363713264465332,
+ "learning_rate": 1.974430357190287e-05,
+ "loss": 1.3921,
+ "step": 3669
+ },
+ {
+ "epoch": 10.0,
+ "grad_norm": 9.591265678405762,
+ "learning_rate": 1.9744105246469264e-05,
+ "loss": 1.5527,
+ "step": 3670
+ },
+ {
+ "epoch": 10.002724795640328,
+ "grad_norm": 12.253864288330078,
+ "learning_rate": 1.9743906845148877e-05,
+ "loss": 1.3445,
+ "step": 3671
+ },
+ {
+ "epoch": 10.005449591280653,
+ "grad_norm": 12.19582748413086,
+ "learning_rate": 1.9743708367943256e-05,
+ "loss": 1.2744,
+ "step": 3672
+ },
+ {
+ "epoch": 10.008174386920981,
+ "grad_norm": 12.801325798034668,
+ "learning_rate": 1.974350981485395e-05,
+ "loss": 1.3838,
+ "step": 3673
+ },
+ {
+ "epoch": 10.010899182561309,
+ "grad_norm": 11.139613151550293,
+ "learning_rate": 1.9743311185882503e-05,
+ "loss": 1.3638,
+ "step": 3674
+ },
+ {
+ "epoch": 10.013623978201634,
+ "grad_norm": 10.15425968170166,
+ "learning_rate": 1.9743112481030465e-05,
+ "loss": 1.3989,
+ "step": 3675
+ },
+ {
+ "epoch": 10.016348773841962,
+ "grad_norm": 11.830443382263184,
+ "learning_rate": 1.974291370029938e-05,
+ "loss": 1.3801,
+ "step": 3676
+ },
+ {
+ "epoch": 10.01907356948229,
+ "grad_norm": 9.02843952178955,
+ "learning_rate": 1.974271484369079e-05,
+ "loss": 1.3232,
+ "step": 3677
+ },
+ {
+ "epoch": 10.021798365122615,
+ "grad_norm": 12.154871940612793,
+ "learning_rate": 1.9742515911206258e-05,
+ "loss": 1.3154,
+ "step": 3678
+ },
+ {
+ "epoch": 10.024523160762943,
+ "grad_norm": 13.292705535888672,
+ "learning_rate": 1.9742316902847322e-05,
+ "loss": 1.3484,
+ "step": 3679
+ },
+ {
+ "epoch": 10.02724795640327,
+ "grad_norm": 9.92895793914795,
+ "learning_rate": 1.9742117818615536e-05,
+ "loss": 1.2634,
+ "step": 3680
+ },
+ {
+ "epoch": 10.029972752043596,
+ "grad_norm": 11.013121604919434,
+ "learning_rate": 1.974191865851245e-05,
+ "loss": 1.1926,
+ "step": 3681
+ },
+ {
+ "epoch": 10.032697547683924,
+ "grad_norm": 8.834386825561523,
+ "learning_rate": 1.9741719422539614e-05,
+ "loss": 1.161,
+ "step": 3682
+ },
+ {
+ "epoch": 10.035422343324251,
+ "grad_norm": 12.060046195983887,
+ "learning_rate": 1.9741520110698582e-05,
+ "loss": 1.1837,
+ "step": 3683
+ },
+ {
+ "epoch": 10.038147138964577,
+ "grad_norm": 67.2748794555664,
+ "learning_rate": 1.9741320722990903e-05,
+ "loss": 1.3889,
+ "step": 3684
+ },
+ {
+ "epoch": 10.040871934604905,
+ "grad_norm": 9.751359939575195,
+ "learning_rate": 1.9741121259418135e-05,
+ "loss": 1.2412,
+ "step": 3685
+ },
+ {
+ "epoch": 10.043596730245232,
+ "grad_norm": 11.913490295410156,
+ "learning_rate": 1.9740921719981827e-05,
+ "loss": 1.3435,
+ "step": 3686
+ },
+ {
+ "epoch": 10.046321525885558,
+ "grad_norm": 10.595745086669922,
+ "learning_rate": 1.9740722104683534e-05,
+ "loss": 1.3184,
+ "step": 3687
+ },
+ {
+ "epoch": 10.049046321525886,
+ "grad_norm": 14.093160629272461,
+ "learning_rate": 1.974052241352481e-05,
+ "loss": 1.2961,
+ "step": 3688
+ },
+ {
+ "epoch": 10.051771117166213,
+ "grad_norm": 10.86572265625,
+ "learning_rate": 1.9740322646507213e-05,
+ "loss": 1.219,
+ "step": 3689
+ },
+ {
+ "epoch": 10.054495912806539,
+ "grad_norm": 11.271374702453613,
+ "learning_rate": 1.9740122803632295e-05,
+ "loss": 1.2532,
+ "step": 3690
+ },
+ {
+ "epoch": 10.057220708446867,
+ "grad_norm": 11.470671653747559,
+ "learning_rate": 1.9739922884901612e-05,
+ "loss": 1.2129,
+ "step": 3691
+ },
+ {
+ "epoch": 10.059945504087194,
+ "grad_norm": 9.367589950561523,
+ "learning_rate": 1.9739722890316728e-05,
+ "loss": 1.1118,
+ "step": 3692
+ },
+ {
+ "epoch": 10.06267029972752,
+ "grad_norm": 10.246399879455566,
+ "learning_rate": 1.9739522819879192e-05,
+ "loss": 1.2852,
+ "step": 3693
+ },
+ {
+ "epoch": 10.065395095367847,
+ "grad_norm": 11.071337699890137,
+ "learning_rate": 1.9739322673590567e-05,
+ "loss": 1.197,
+ "step": 3694
+ },
+ {
+ "epoch": 10.068119891008175,
+ "grad_norm": 10.514875411987305,
+ "learning_rate": 1.9739122451452412e-05,
+ "loss": 1.2434,
+ "step": 3695
+ },
+ {
+ "epoch": 10.0708446866485,
+ "grad_norm": 12.366474151611328,
+ "learning_rate": 1.9738922153466282e-05,
+ "loss": 1.4819,
+ "step": 3696
+ },
+ {
+ "epoch": 10.073569482288828,
+ "grad_norm": 7.928983688354492,
+ "learning_rate": 1.973872177963374e-05,
+ "loss": 1.1301,
+ "step": 3697
+ },
+ {
+ "epoch": 10.076294277929156,
+ "grad_norm": 8.069535255432129,
+ "learning_rate": 1.9738521329956347e-05,
+ "loss": 1.0312,
+ "step": 3698
+ },
+ {
+ "epoch": 10.079019073569482,
+ "grad_norm": 17.79938316345215,
+ "learning_rate": 1.9738320804435662e-05,
+ "loss": 1.4001,
+ "step": 3699
+ },
+ {
+ "epoch": 10.08174386920981,
+ "grad_norm": 8.293427467346191,
+ "learning_rate": 1.973812020307325e-05,
+ "loss": 1.1453,
+ "step": 3700
+ },
+ {
+ "epoch": 10.084468664850137,
+ "grad_norm": 11.034486770629883,
+ "learning_rate": 1.973791952587067e-05,
+ "loss": 1.6006,
+ "step": 3701
+ },
+ {
+ "epoch": 10.087193460490463,
+ "grad_norm": 7.1742963790893555,
+ "learning_rate": 1.9737718772829485e-05,
+ "loss": 1.1616,
+ "step": 3702
+ },
+ {
+ "epoch": 10.08991825613079,
+ "grad_norm": 14.742067337036133,
+ "learning_rate": 1.9737517943951262e-05,
+ "loss": 1.125,
+ "step": 3703
+ },
+ {
+ "epoch": 10.092643051771118,
+ "grad_norm": 9.339225769042969,
+ "learning_rate": 1.9737317039237563e-05,
+ "loss": 1.2705,
+ "step": 3704
+ },
+ {
+ "epoch": 10.095367847411444,
+ "grad_norm": 10.711777687072754,
+ "learning_rate": 1.973711605868995e-05,
+ "loss": 1.3081,
+ "step": 3705
+ },
+ {
+ "epoch": 10.098092643051771,
+ "grad_norm": 8.772830963134766,
+ "learning_rate": 1.973691500230999e-05,
+ "loss": 1.364,
+ "step": 3706
+ },
+ {
+ "epoch": 10.100817438692099,
+ "grad_norm": 8.609234809875488,
+ "learning_rate": 1.9736713870099253e-05,
+ "loss": 1.0784,
+ "step": 3707
+ },
+ {
+ "epoch": 10.103542234332425,
+ "grad_norm": 7.832282543182373,
+ "learning_rate": 1.97365126620593e-05,
+ "loss": 1.2267,
+ "step": 3708
+ },
+ {
+ "epoch": 10.106267029972752,
+ "grad_norm": 17.29844856262207,
+ "learning_rate": 1.97363113781917e-05,
+ "loss": 1.1885,
+ "step": 3709
+ },
+ {
+ "epoch": 10.10899182561308,
+ "grad_norm": 9.92955207824707,
+ "learning_rate": 1.9736110018498017e-05,
+ "loss": 1.1882,
+ "step": 3710
+ },
+ {
+ "epoch": 10.111716621253406,
+ "grad_norm": 10.351883888244629,
+ "learning_rate": 1.973590858297983e-05,
+ "loss": 1.2561,
+ "step": 3711
+ },
+ {
+ "epoch": 10.114441416893733,
+ "grad_norm": 8.720454216003418,
+ "learning_rate": 1.9735707071638693e-05,
+ "loss": 1.0909,
+ "step": 3712
+ },
+ {
+ "epoch": 10.11716621253406,
+ "grad_norm": 11.84240436553955,
+ "learning_rate": 1.9735505484476186e-05,
+ "loss": 1.1296,
+ "step": 3713
+ },
+ {
+ "epoch": 10.119891008174386,
+ "grad_norm": 8.756014823913574,
+ "learning_rate": 1.9735303821493876e-05,
+ "loss": 1.1089,
+ "step": 3714
+ },
+ {
+ "epoch": 10.122615803814714,
+ "grad_norm": 9.413190841674805,
+ "learning_rate": 1.9735102082693332e-05,
+ "loss": 1.2961,
+ "step": 3715
+ },
+ {
+ "epoch": 10.125340599455042,
+ "grad_norm": 12.350908279418945,
+ "learning_rate": 1.9734900268076126e-05,
+ "loss": 1.2798,
+ "step": 3716
+ },
+ {
+ "epoch": 10.128065395095367,
+ "grad_norm": 10.52960205078125,
+ "learning_rate": 1.973469837764383e-05,
+ "loss": 1.2402,
+ "step": 3717
+ },
+ {
+ "epoch": 10.130790190735695,
+ "grad_norm": 9.146392822265625,
+ "learning_rate": 1.9734496411398015e-05,
+ "loss": 1.1663,
+ "step": 3718
+ },
+ {
+ "epoch": 10.133514986376023,
+ "grad_norm": 7.834981441497803,
+ "learning_rate": 1.973429436934026e-05,
+ "loss": 1.2649,
+ "step": 3719
+ },
+ {
+ "epoch": 10.136239782016348,
+ "grad_norm": 8.250489234924316,
+ "learning_rate": 1.973409225147213e-05,
+ "loss": 1.0134,
+ "step": 3720
+ },
+ {
+ "epoch": 10.138964577656676,
+ "grad_norm": 7.558762550354004,
+ "learning_rate": 1.9733890057795203e-05,
+ "loss": 1.0573,
+ "step": 3721
+ },
+ {
+ "epoch": 10.141689373297003,
+ "grad_norm": 9.883357048034668,
+ "learning_rate": 1.9733687788311055e-05,
+ "loss": 1.2424,
+ "step": 3722
+ },
+ {
+ "epoch": 10.14441416893733,
+ "grad_norm": 7.904563903808594,
+ "learning_rate": 1.973348544302126e-05,
+ "loss": 1.3923,
+ "step": 3723
+ },
+ {
+ "epoch": 10.147138964577657,
+ "grad_norm": 11.666852951049805,
+ "learning_rate": 1.973328302192739e-05,
+ "loss": 1.4563,
+ "step": 3724
+ },
+ {
+ "epoch": 10.149863760217984,
+ "grad_norm": 8.474404335021973,
+ "learning_rate": 1.973308052503103e-05,
+ "loss": 1.1587,
+ "step": 3725
+ },
+ {
+ "epoch": 10.15258855585831,
+ "grad_norm": 8.203660011291504,
+ "learning_rate": 1.9732877952333748e-05,
+ "loss": 1.2905,
+ "step": 3726
+ },
+ {
+ "epoch": 10.155313351498638,
+ "grad_norm": 10.832432746887207,
+ "learning_rate": 1.973267530383713e-05,
+ "loss": 1.3057,
+ "step": 3727
+ },
+ {
+ "epoch": 10.158038147138965,
+ "grad_norm": 7.435604095458984,
+ "learning_rate": 1.9732472579542746e-05,
+ "loss": 1.2734,
+ "step": 3728
+ },
+ {
+ "epoch": 10.160762942779291,
+ "grad_norm": 9.250262260437012,
+ "learning_rate": 1.973226977945218e-05,
+ "loss": 1.2043,
+ "step": 3729
+ },
+ {
+ "epoch": 10.163487738419619,
+ "grad_norm": 10.409645080566406,
+ "learning_rate": 1.973206690356701e-05,
+ "loss": 1.1812,
+ "step": 3730
+ },
+ {
+ "epoch": 10.166212534059946,
+ "grad_norm": 12.442349433898926,
+ "learning_rate": 1.9731863951888818e-05,
+ "loss": 1.4661,
+ "step": 3731
+ },
+ {
+ "epoch": 10.168937329700272,
+ "grad_norm": 15.570923805236816,
+ "learning_rate": 1.9731660924419182e-05,
+ "loss": 1.3845,
+ "step": 3732
+ },
+ {
+ "epoch": 10.1716621253406,
+ "grad_norm": 12.21881103515625,
+ "learning_rate": 1.9731457821159684e-05,
+ "loss": 1.0405,
+ "step": 3733
+ },
+ {
+ "epoch": 10.174386920980927,
+ "grad_norm": 17.417682647705078,
+ "learning_rate": 1.9731254642111902e-05,
+ "loss": 1.199,
+ "step": 3734
+ },
+ {
+ "epoch": 10.177111716621253,
+ "grad_norm": 11.305848121643066,
+ "learning_rate": 1.9731051387277427e-05,
+ "loss": 1.0884,
+ "step": 3735
+ },
+ {
+ "epoch": 10.17983651226158,
+ "grad_norm": 6.832694053649902,
+ "learning_rate": 1.9730848056657836e-05,
+ "loss": 1.2855,
+ "step": 3736
+ },
+ {
+ "epoch": 10.182561307901908,
+ "grad_norm": 9.419797897338867,
+ "learning_rate": 1.9730644650254713e-05,
+ "loss": 1.2144,
+ "step": 3737
+ },
+ {
+ "epoch": 10.185286103542234,
+ "grad_norm": 10.504746437072754,
+ "learning_rate": 1.973044116806964e-05,
+ "loss": 1.1311,
+ "step": 3738
+ },
+ {
+ "epoch": 10.188010899182562,
+ "grad_norm": 7.877094745635986,
+ "learning_rate": 1.9730237610104206e-05,
+ "loss": 1.0391,
+ "step": 3739
+ },
+ {
+ "epoch": 10.190735694822889,
+ "grad_norm": 7.2681884765625,
+ "learning_rate": 1.9730033976359997e-05,
+ "loss": 1.1045,
+ "step": 3740
+ },
+ {
+ "epoch": 10.193460490463215,
+ "grad_norm": 7.762524604797363,
+ "learning_rate": 1.9729830266838595e-05,
+ "loss": 1.2251,
+ "step": 3741
+ },
+ {
+ "epoch": 10.196185286103542,
+ "grad_norm": 7.2067365646362305,
+ "learning_rate": 1.9729626481541588e-05,
+ "loss": 1.2778,
+ "step": 3742
+ },
+ {
+ "epoch": 10.19891008174387,
+ "grad_norm": 7.826615810394287,
+ "learning_rate": 1.9729422620470565e-05,
+ "loss": 1.369,
+ "step": 3743
+ },
+ {
+ "epoch": 10.201634877384196,
+ "grad_norm": 12.013307571411133,
+ "learning_rate": 1.972921868362711e-05,
+ "loss": 1.1669,
+ "step": 3744
+ },
+ {
+ "epoch": 10.204359673024523,
+ "grad_norm": 8.024679183959961,
+ "learning_rate": 1.972901467101281e-05,
+ "loss": 1.0647,
+ "step": 3745
+ },
+ {
+ "epoch": 10.207084468664851,
+ "grad_norm": 9.573027610778809,
+ "learning_rate": 1.972881058262926e-05,
+ "loss": 1.1094,
+ "step": 3746
+ },
+ {
+ "epoch": 10.209809264305177,
+ "grad_norm": 8.3164644241333,
+ "learning_rate": 1.9728606418478046e-05,
+ "loss": 1.3103,
+ "step": 3747
+ },
+ {
+ "epoch": 10.212534059945504,
+ "grad_norm": 7.2291083335876465,
+ "learning_rate": 1.9728402178560757e-05,
+ "loss": 1.0315,
+ "step": 3748
+ },
+ {
+ "epoch": 10.215258855585832,
+ "grad_norm": 9.45434284210205,
+ "learning_rate": 1.972819786287899e-05,
+ "loss": 1.0261,
+ "step": 3749
+ },
+ {
+ "epoch": 10.217983651226158,
+ "grad_norm": 8.173750877380371,
+ "learning_rate": 1.9727993471434326e-05,
+ "loss": 1.219,
+ "step": 3750
+ },
+ {
+ "epoch": 10.220708446866485,
+ "grad_norm": 6.997307300567627,
+ "learning_rate": 1.9727789004228362e-05,
+ "loss": 1.1887,
+ "step": 3751
+ },
+ {
+ "epoch": 10.223433242506813,
+ "grad_norm": 8.035279273986816,
+ "learning_rate": 1.972758446126269e-05,
+ "loss": 1.3751,
+ "step": 3752
+ },
+ {
+ "epoch": 10.226158038147139,
+ "grad_norm": 7.729323387145996,
+ "learning_rate": 1.972737984253891e-05,
+ "loss": 1.2571,
+ "step": 3753
+ },
+ {
+ "epoch": 10.228882833787466,
+ "grad_norm": 6.181939125061035,
+ "learning_rate": 1.9727175148058603e-05,
+ "loss": 1.0676,
+ "step": 3754
+ },
+ {
+ "epoch": 10.231607629427794,
+ "grad_norm": 9.303966522216797,
+ "learning_rate": 1.9726970377823367e-05,
+ "loss": 1.0984,
+ "step": 3755
+ },
+ {
+ "epoch": 10.23433242506812,
+ "grad_norm": 10.263012886047363,
+ "learning_rate": 1.9726765531834803e-05,
+ "loss": 1.3699,
+ "step": 3756
+ },
+ {
+ "epoch": 10.237057220708447,
+ "grad_norm": 9.619956970214844,
+ "learning_rate": 1.9726560610094497e-05,
+ "loss": 1.0916,
+ "step": 3757
+ },
+ {
+ "epoch": 10.239782016348773,
+ "grad_norm": 7.459681510925293,
+ "learning_rate": 1.9726355612604058e-05,
+ "loss": 1.1418,
+ "step": 3758
+ },
+ {
+ "epoch": 10.2425068119891,
+ "grad_norm": 7.82033634185791,
+ "learning_rate": 1.972615053936507e-05,
+ "loss": 1.2256,
+ "step": 3759
+ },
+ {
+ "epoch": 10.245231607629428,
+ "grad_norm": 8.450642585754395,
+ "learning_rate": 1.9725945390379135e-05,
+ "loss": 1.1772,
+ "step": 3760
+ },
+ {
+ "epoch": 10.247956403269754,
+ "grad_norm": 12.190950393676758,
+ "learning_rate": 1.972574016564785e-05,
+ "loss": 1.3452,
+ "step": 3761
+ },
+ {
+ "epoch": 10.250681198910081,
+ "grad_norm": 11.40127944946289,
+ "learning_rate": 1.9725534865172814e-05,
+ "loss": 1.5155,
+ "step": 3762
+ },
+ {
+ "epoch": 10.253405994550409,
+ "grad_norm": 13.994333267211914,
+ "learning_rate": 1.9725329488955625e-05,
+ "loss": 1.2532,
+ "step": 3763
+ },
+ {
+ "epoch": 10.256130790190735,
+ "grad_norm": 9.886280059814453,
+ "learning_rate": 1.972512403699788e-05,
+ "loss": 1.0574,
+ "step": 3764
+ },
+ {
+ "epoch": 10.258855585831062,
+ "grad_norm": 13.502911567687988,
+ "learning_rate": 1.9724918509301186e-05,
+ "loss": 1.2803,
+ "step": 3765
+ },
+ {
+ "epoch": 10.26158038147139,
+ "grad_norm": 34.35505294799805,
+ "learning_rate": 1.972471290586714e-05,
+ "loss": 1.3633,
+ "step": 3766
+ },
+ {
+ "epoch": 10.264305177111716,
+ "grad_norm": 8.205253601074219,
+ "learning_rate": 1.9724507226697343e-05,
+ "loss": 1.0892,
+ "step": 3767
+ },
+ {
+ "epoch": 10.267029972752043,
+ "grad_norm": 7.548518657684326,
+ "learning_rate": 1.9724301471793394e-05,
+ "loss": 0.9908,
+ "step": 3768
+ },
+ {
+ "epoch": 10.269754768392371,
+ "grad_norm": 12.83620834350586,
+ "learning_rate": 1.9724095641156903e-05,
+ "loss": 1.033,
+ "step": 3769
+ },
+ {
+ "epoch": 10.272479564032697,
+ "grad_norm": 13.945982933044434,
+ "learning_rate": 1.972388973478946e-05,
+ "loss": 1.2754,
+ "step": 3770
+ },
+ {
+ "epoch": 10.275204359673024,
+ "grad_norm": 15.30763053894043,
+ "learning_rate": 1.9723683752692683e-05,
+ "loss": 1.457,
+ "step": 3771
+ },
+ {
+ "epoch": 10.277929155313352,
+ "grad_norm": 7.554656982421875,
+ "learning_rate": 1.9723477694868166e-05,
+ "loss": 1.0938,
+ "step": 3772
+ },
+ {
+ "epoch": 10.280653950953678,
+ "grad_norm": 9.48392391204834,
+ "learning_rate": 1.972327156131752e-05,
+ "loss": 1.3281,
+ "step": 3773
+ },
+ {
+ "epoch": 10.283378746594005,
+ "grad_norm": 9.544161796569824,
+ "learning_rate": 1.972306535204235e-05,
+ "loss": 1.489,
+ "step": 3774
+ },
+ {
+ "epoch": 10.286103542234333,
+ "grad_norm": 8.200508117675781,
+ "learning_rate": 1.972285906704426e-05,
+ "loss": 1.1425,
+ "step": 3775
+ },
+ {
+ "epoch": 10.288828337874659,
+ "grad_norm": 10.27327823638916,
+ "learning_rate": 1.9722652706324853e-05,
+ "loss": 1.3013,
+ "step": 3776
+ },
+ {
+ "epoch": 10.291553133514986,
+ "grad_norm": 8.085951805114746,
+ "learning_rate": 1.972244626988574e-05,
+ "loss": 1.2046,
+ "step": 3777
+ },
+ {
+ "epoch": 10.294277929155314,
+ "grad_norm": 8.203631401062012,
+ "learning_rate": 1.972223975772853e-05,
+ "loss": 1.2805,
+ "step": 3778
+ },
+ {
+ "epoch": 10.29700272479564,
+ "grad_norm": 9.850308418273926,
+ "learning_rate": 1.9722033169854828e-05,
+ "loss": 1.2058,
+ "step": 3779
+ },
+ {
+ "epoch": 10.299727520435967,
+ "grad_norm": 8.00572395324707,
+ "learning_rate": 1.972182650626625e-05,
+ "loss": 1.2004,
+ "step": 3780
+ },
+ {
+ "epoch": 10.302452316076295,
+ "grad_norm": 13.008173942565918,
+ "learning_rate": 1.9721619766964393e-05,
+ "loss": 1.145,
+ "step": 3781
+ },
+ {
+ "epoch": 10.30517711171662,
+ "grad_norm": 7.926570892333984,
+ "learning_rate": 1.9721412951950876e-05,
+ "loss": 1.1924,
+ "step": 3782
+ },
+ {
+ "epoch": 10.307901907356948,
+ "grad_norm": 7.959515571594238,
+ "learning_rate": 1.9721206061227312e-05,
+ "loss": 1.0864,
+ "step": 3783
+ },
+ {
+ "epoch": 10.310626702997276,
+ "grad_norm": 9.349390983581543,
+ "learning_rate": 1.9720999094795303e-05,
+ "loss": 1.3108,
+ "step": 3784
+ },
+ {
+ "epoch": 10.313351498637601,
+ "grad_norm": 6.111342906951904,
+ "learning_rate": 1.9720792052656468e-05,
+ "loss": 1.1169,
+ "step": 3785
+ },
+ {
+ "epoch": 10.316076294277929,
+ "grad_norm": 12.590240478515625,
+ "learning_rate": 1.9720584934812417e-05,
+ "loss": 1.2366,
+ "step": 3786
+ },
+ {
+ "epoch": 10.318801089918257,
+ "grad_norm": 7.094612121582031,
+ "learning_rate": 1.9720377741264762e-05,
+ "loss": 1.1785,
+ "step": 3787
+ },
+ {
+ "epoch": 10.321525885558582,
+ "grad_norm": 10.494339942932129,
+ "learning_rate": 1.972017047201512e-05,
+ "loss": 1.3018,
+ "step": 3788
+ },
+ {
+ "epoch": 10.32425068119891,
+ "grad_norm": 7.665545463562012,
+ "learning_rate": 1.9719963127065106e-05,
+ "loss": 1.1545,
+ "step": 3789
+ },
+ {
+ "epoch": 10.326975476839237,
+ "grad_norm": 15.356550216674805,
+ "learning_rate": 1.9719755706416327e-05,
+ "loss": 1.2166,
+ "step": 3790
+ },
+ {
+ "epoch": 10.329700272479563,
+ "grad_norm": 9.577674865722656,
+ "learning_rate": 1.9719548210070406e-05,
+ "loss": 1.0068,
+ "step": 3791
+ },
+ {
+ "epoch": 10.33242506811989,
+ "grad_norm": 8.080168724060059,
+ "learning_rate": 1.9719340638028952e-05,
+ "loss": 1.1154,
+ "step": 3792
+ },
+ {
+ "epoch": 10.335149863760218,
+ "grad_norm": 7.914016246795654,
+ "learning_rate": 1.971913299029359e-05,
+ "loss": 1.1511,
+ "step": 3793
+ },
+ {
+ "epoch": 10.337874659400544,
+ "grad_norm": 10.83732795715332,
+ "learning_rate": 1.9718925266865933e-05,
+ "loss": 1.0859,
+ "step": 3794
+ },
+ {
+ "epoch": 10.340599455040872,
+ "grad_norm": 7.65169620513916,
+ "learning_rate": 1.97187174677476e-05,
+ "loss": 1.2258,
+ "step": 3795
+ },
+ {
+ "epoch": 10.3433242506812,
+ "grad_norm": 9.187763214111328,
+ "learning_rate": 1.9718509592940206e-05,
+ "loss": 1.509,
+ "step": 3796
+ },
+ {
+ "epoch": 10.346049046321525,
+ "grad_norm": 7.308699607849121,
+ "learning_rate": 1.971830164244537e-05,
+ "loss": 1.4558,
+ "step": 3797
+ },
+ {
+ "epoch": 10.348773841961853,
+ "grad_norm": 11.15339183807373,
+ "learning_rate": 1.9718093616264717e-05,
+ "loss": 1.3391,
+ "step": 3798
+ },
+ {
+ "epoch": 10.35149863760218,
+ "grad_norm": 7.759003162384033,
+ "learning_rate": 1.9717885514399865e-05,
+ "loss": 1.189,
+ "step": 3799
+ },
+ {
+ "epoch": 10.354223433242506,
+ "grad_norm": 7.761136531829834,
+ "learning_rate": 1.971767733685243e-05,
+ "loss": 1.1958,
+ "step": 3800
+ },
+ {
+ "epoch": 10.356948228882834,
+ "grad_norm": 8.449786186218262,
+ "learning_rate": 1.9717469083624037e-05,
+ "loss": 1.0537,
+ "step": 3801
+ },
+ {
+ "epoch": 10.359673024523161,
+ "grad_norm": 10.144342422485352,
+ "learning_rate": 1.9717260754716306e-05,
+ "loss": 1.1143,
+ "step": 3802
+ },
+ {
+ "epoch": 10.362397820163487,
+ "grad_norm": 10.833005905151367,
+ "learning_rate": 1.9717052350130864e-05,
+ "loss": 1.2683,
+ "step": 3803
+ },
+ {
+ "epoch": 10.365122615803815,
+ "grad_norm": 8.77840805053711,
+ "learning_rate": 1.971684386986933e-05,
+ "loss": 1.1941,
+ "step": 3804
+ },
+ {
+ "epoch": 10.367847411444142,
+ "grad_norm": 9.213385581970215,
+ "learning_rate": 1.9716635313933334e-05,
+ "loss": 1.0796,
+ "step": 3805
+ },
+ {
+ "epoch": 10.370572207084468,
+ "grad_norm": 8.934281349182129,
+ "learning_rate": 1.9716426682324488e-05,
+ "loss": 1.0317,
+ "step": 3806
+ },
+ {
+ "epoch": 10.373297002724795,
+ "grad_norm": 9.853482246398926,
+ "learning_rate": 1.9716217975044426e-05,
+ "loss": 1.0713,
+ "step": 3807
+ },
+ {
+ "epoch": 10.376021798365123,
+ "grad_norm": 8.844365119934082,
+ "learning_rate": 1.971600919209477e-05,
+ "loss": 1.1797,
+ "step": 3808
+ },
+ {
+ "epoch": 10.378746594005449,
+ "grad_norm": 9.398346900939941,
+ "learning_rate": 1.9715800333477148e-05,
+ "loss": 1.163,
+ "step": 3809
+ },
+ {
+ "epoch": 10.381471389645776,
+ "grad_norm": 8.325582504272461,
+ "learning_rate": 1.9715591399193184e-05,
+ "loss": 1.1298,
+ "step": 3810
+ },
+ {
+ "epoch": 10.384196185286104,
+ "grad_norm": 8.730047225952148,
+ "learning_rate": 1.971538238924451e-05,
+ "loss": 1.3547,
+ "step": 3811
+ },
+ {
+ "epoch": 10.38692098092643,
+ "grad_norm": 7.728232383728027,
+ "learning_rate": 1.971517330363275e-05,
+ "loss": 1.1624,
+ "step": 3812
+ },
+ {
+ "epoch": 10.389645776566757,
+ "grad_norm": 8.164695739746094,
+ "learning_rate": 1.971496414235953e-05,
+ "loss": 1.2239,
+ "step": 3813
+ },
+ {
+ "epoch": 10.392370572207085,
+ "grad_norm": 10.444753646850586,
+ "learning_rate": 1.9714754905426484e-05,
+ "loss": 1.0493,
+ "step": 3814
+ },
+ {
+ "epoch": 10.39509536784741,
+ "grad_norm": 8.868157386779785,
+ "learning_rate": 1.971454559283524e-05,
+ "loss": 1.2437,
+ "step": 3815
+ },
+ {
+ "epoch": 10.397820163487738,
+ "grad_norm": 8.108026504516602,
+ "learning_rate": 1.971433620458743e-05,
+ "loss": 1.0942,
+ "step": 3816
+ },
+ {
+ "epoch": 10.400544959128066,
+ "grad_norm": 7.994935035705566,
+ "learning_rate": 1.9714126740684677e-05,
+ "loss": 1.0762,
+ "step": 3817
+ },
+ {
+ "epoch": 10.403269754768392,
+ "grad_norm": 8.728080749511719,
+ "learning_rate": 1.971391720112862e-05,
+ "loss": 1.155,
+ "step": 3818
+ },
+ {
+ "epoch": 10.40599455040872,
+ "grad_norm": 9.615818977355957,
+ "learning_rate": 1.971370758592089e-05,
+ "loss": 1.3136,
+ "step": 3819
+ },
+ {
+ "epoch": 10.408719346049047,
+ "grad_norm": 8.452714920043945,
+ "learning_rate": 1.9713497895063116e-05,
+ "loss": 1.3281,
+ "step": 3820
+ },
+ {
+ "epoch": 10.411444141689373,
+ "grad_norm": 9.467696189880371,
+ "learning_rate": 1.9713288128556932e-05,
+ "loss": 1.1877,
+ "step": 3821
+ },
+ {
+ "epoch": 10.4141689373297,
+ "grad_norm": 13.691346168518066,
+ "learning_rate": 1.9713078286403973e-05,
+ "loss": 1.2886,
+ "step": 3822
+ },
+ {
+ "epoch": 10.416893732970028,
+ "grad_norm": 8.293333053588867,
+ "learning_rate": 1.9712868368605877e-05,
+ "loss": 1.1771,
+ "step": 3823
+ },
+ {
+ "epoch": 10.419618528610354,
+ "grad_norm": 9.708534240722656,
+ "learning_rate": 1.9712658375164272e-05,
+ "loss": 1.1155,
+ "step": 3824
+ },
+ {
+ "epoch": 10.422343324250681,
+ "grad_norm": 8.688053131103516,
+ "learning_rate": 1.9712448306080793e-05,
+ "loss": 1.1792,
+ "step": 3825
+ },
+ {
+ "epoch": 10.425068119891009,
+ "grad_norm": 7.905061721801758,
+ "learning_rate": 1.9712238161357084e-05,
+ "loss": 1.0693,
+ "step": 3826
+ },
+ {
+ "epoch": 10.427792915531334,
+ "grad_norm": 11.181714057922363,
+ "learning_rate": 1.9712027940994777e-05,
+ "loss": 1.4639,
+ "step": 3827
+ },
+ {
+ "epoch": 10.430517711171662,
+ "grad_norm": 11.698137283325195,
+ "learning_rate": 1.9711817644995504e-05,
+ "loss": 1.189,
+ "step": 3828
+ },
+ {
+ "epoch": 10.43324250681199,
+ "grad_norm": 6.352274417877197,
+ "learning_rate": 1.9711607273360912e-05,
+ "loss": 0.9788,
+ "step": 3829
+ },
+ {
+ "epoch": 10.435967302452315,
+ "grad_norm": 10.760963439941406,
+ "learning_rate": 1.9711396826092634e-05,
+ "loss": 1.1357,
+ "step": 3830
+ },
+ {
+ "epoch": 10.438692098092643,
+ "grad_norm": 7.958438396453857,
+ "learning_rate": 1.971118630319231e-05,
+ "loss": 1.1244,
+ "step": 3831
+ },
+ {
+ "epoch": 10.44141689373297,
+ "grad_norm": 11.469450950622559,
+ "learning_rate": 1.971097570466158e-05,
+ "loss": 1.2788,
+ "step": 3832
+ },
+ {
+ "epoch": 10.444141689373296,
+ "grad_norm": 6.867679595947266,
+ "learning_rate": 1.9710765030502085e-05,
+ "loss": 1.1975,
+ "step": 3833
+ },
+ {
+ "epoch": 10.446866485013624,
+ "grad_norm": 13.968586921691895,
+ "learning_rate": 1.971055428071546e-05,
+ "loss": 1.3179,
+ "step": 3834
+ },
+ {
+ "epoch": 10.449591280653951,
+ "grad_norm": 8.896160125732422,
+ "learning_rate": 1.9710343455303354e-05,
+ "loss": 1.3503,
+ "step": 3835
+ },
+ {
+ "epoch": 10.452316076294277,
+ "grad_norm": 8.06062126159668,
+ "learning_rate": 1.9710132554267404e-05,
+ "loss": 1.2216,
+ "step": 3836
+ },
+ {
+ "epoch": 10.455040871934605,
+ "grad_norm": 8.38322639465332,
+ "learning_rate": 1.9709921577609256e-05,
+ "loss": 1.135,
+ "step": 3837
+ },
+ {
+ "epoch": 10.457765667574932,
+ "grad_norm": 7.959988594055176,
+ "learning_rate": 1.9709710525330548e-05,
+ "loss": 1.155,
+ "step": 3838
+ },
+ {
+ "epoch": 10.460490463215258,
+ "grad_norm": 6.774838924407959,
+ "learning_rate": 1.9709499397432933e-05,
+ "loss": 1.042,
+ "step": 3839
+ },
+ {
+ "epoch": 10.463215258855586,
+ "grad_norm": 8.353154182434082,
+ "learning_rate": 1.9709288193918044e-05,
+ "loss": 1.2507,
+ "step": 3840
+ },
+ {
+ "epoch": 10.465940054495913,
+ "grad_norm": 8.687185287475586,
+ "learning_rate": 1.9709076914787535e-05,
+ "loss": 1.3516,
+ "step": 3841
+ },
+ {
+ "epoch": 10.46866485013624,
+ "grad_norm": 7.83223819732666,
+ "learning_rate": 1.9708865560043044e-05,
+ "loss": 1.2035,
+ "step": 3842
+ },
+ {
+ "epoch": 10.471389645776567,
+ "grad_norm": 9.42057991027832,
+ "learning_rate": 1.9708654129686224e-05,
+ "loss": 1.2656,
+ "step": 3843
+ },
+ {
+ "epoch": 10.474114441416894,
+ "grad_norm": 9.055431365966797,
+ "learning_rate": 1.9708442623718715e-05,
+ "loss": 1.2795,
+ "step": 3844
+ },
+ {
+ "epoch": 10.47683923705722,
+ "grad_norm": 6.744367599487305,
+ "learning_rate": 1.9708231042142166e-05,
+ "loss": 1.0947,
+ "step": 3845
+ },
+ {
+ "epoch": 10.479564032697548,
+ "grad_norm": 7.077953338623047,
+ "learning_rate": 1.970801938495823e-05,
+ "loss": 1.2078,
+ "step": 3846
+ },
+ {
+ "epoch": 10.482288828337875,
+ "grad_norm": 7.5632219314575195,
+ "learning_rate": 1.970780765216855e-05,
+ "loss": 1.1743,
+ "step": 3847
+ },
+ {
+ "epoch": 10.485013623978201,
+ "grad_norm": 8.792860984802246,
+ "learning_rate": 1.9707595843774775e-05,
+ "loss": 1.2559,
+ "step": 3848
+ },
+ {
+ "epoch": 10.487738419618529,
+ "grad_norm": 8.674450874328613,
+ "learning_rate": 1.970738395977856e-05,
+ "loss": 1.3398,
+ "step": 3849
+ },
+ {
+ "epoch": 10.490463215258856,
+ "grad_norm": 9.62348747253418,
+ "learning_rate": 1.9707172000181546e-05,
+ "loss": 0.9988,
+ "step": 3850
+ },
+ {
+ "epoch": 10.493188010899182,
+ "grad_norm": 9.784550666809082,
+ "learning_rate": 1.9706959964985393e-05,
+ "loss": 1.2141,
+ "step": 3851
+ },
+ {
+ "epoch": 10.49591280653951,
+ "grad_norm": 7.044509410858154,
+ "learning_rate": 1.9706747854191745e-05,
+ "loss": 1.0884,
+ "step": 3852
+ },
+ {
+ "epoch": 10.498637602179837,
+ "grad_norm": 6.42462158203125,
+ "learning_rate": 1.970653566780226e-05,
+ "loss": 1.1487,
+ "step": 3853
+ },
+ {
+ "epoch": 10.501362397820163,
+ "grad_norm": 7.424637794494629,
+ "learning_rate": 1.9706323405818585e-05,
+ "loss": 1.0198,
+ "step": 3854
+ },
+ {
+ "epoch": 10.50408719346049,
+ "grad_norm": 9.120077133178711,
+ "learning_rate": 1.970611106824238e-05,
+ "loss": 1.1648,
+ "step": 3855
+ },
+ {
+ "epoch": 10.506811989100818,
+ "grad_norm": 12.282148361206055,
+ "learning_rate": 1.970589865507529e-05,
+ "loss": 1.29,
+ "step": 3856
+ },
+ {
+ "epoch": 10.509536784741144,
+ "grad_norm": 9.207600593566895,
+ "learning_rate": 1.9705686166318973e-05,
+ "loss": 1.0338,
+ "step": 3857
+ },
+ {
+ "epoch": 10.512261580381471,
+ "grad_norm": 12.616487503051758,
+ "learning_rate": 1.9705473601975086e-05,
+ "loss": 1.248,
+ "step": 3858
+ },
+ {
+ "epoch": 10.514986376021799,
+ "grad_norm": 8.448892593383789,
+ "learning_rate": 1.9705260962045286e-05,
+ "loss": 0.9753,
+ "step": 3859
+ },
+ {
+ "epoch": 10.517711171662125,
+ "grad_norm": 7.18719482421875,
+ "learning_rate": 1.9705048246531223e-05,
+ "loss": 1.1677,
+ "step": 3860
+ },
+ {
+ "epoch": 10.520435967302452,
+ "grad_norm": 7.5851969718933105,
+ "learning_rate": 1.9704835455434556e-05,
+ "loss": 1.1337,
+ "step": 3861
+ },
+ {
+ "epoch": 10.52316076294278,
+ "grad_norm": 7.456258296966553,
+ "learning_rate": 1.9704622588756946e-05,
+ "loss": 1.0076,
+ "step": 3862
+ },
+ {
+ "epoch": 10.525885558583106,
+ "grad_norm": 12.191134452819824,
+ "learning_rate": 1.9704409646500044e-05,
+ "loss": 1.1499,
+ "step": 3863
+ },
+ {
+ "epoch": 10.528610354223433,
+ "grad_norm": 8.46213436126709,
+ "learning_rate": 1.970419662866551e-05,
+ "loss": 1.2076,
+ "step": 3864
+ },
+ {
+ "epoch": 10.53133514986376,
+ "grad_norm": 7.285557746887207,
+ "learning_rate": 1.9703983535255013e-05,
+ "loss": 1.3665,
+ "step": 3865
+ },
+ {
+ "epoch": 10.534059945504087,
+ "grad_norm": 7.149416446685791,
+ "learning_rate": 1.97037703662702e-05,
+ "loss": 1.1638,
+ "step": 3866
+ },
+ {
+ "epoch": 10.536784741144414,
+ "grad_norm": 6.993155002593994,
+ "learning_rate": 1.9703557121712737e-05,
+ "loss": 1.0959,
+ "step": 3867
+ },
+ {
+ "epoch": 10.539509536784742,
+ "grad_norm": 11.321159362792969,
+ "learning_rate": 1.9703343801584282e-05,
+ "loss": 1.1399,
+ "step": 3868
+ },
+ {
+ "epoch": 10.542234332425068,
+ "grad_norm": 11.617790222167969,
+ "learning_rate": 1.9703130405886496e-05,
+ "loss": 1.1311,
+ "step": 3869
+ },
+ {
+ "epoch": 10.544959128065395,
+ "grad_norm": 7.791107177734375,
+ "learning_rate": 1.9702916934621044e-05,
+ "loss": 1.1624,
+ "step": 3870
+ },
+ {
+ "epoch": 10.547683923705723,
+ "grad_norm": 8.269343376159668,
+ "learning_rate": 1.970270338778959e-05,
+ "loss": 1.1655,
+ "step": 3871
+ },
+ {
+ "epoch": 10.550408719346049,
+ "grad_norm": 7.9192728996276855,
+ "learning_rate": 1.970248976539379e-05,
+ "loss": 1.322,
+ "step": 3872
+ },
+ {
+ "epoch": 10.553133514986376,
+ "grad_norm": 7.300230503082275,
+ "learning_rate": 1.9702276067435316e-05,
+ "loss": 1.1975,
+ "step": 3873
+ },
+ {
+ "epoch": 10.555858310626704,
+ "grad_norm": 8.565752983093262,
+ "learning_rate": 1.9702062293915827e-05,
+ "loss": 1.2988,
+ "step": 3874
+ },
+ {
+ "epoch": 10.55858310626703,
+ "grad_norm": 7.208471775054932,
+ "learning_rate": 1.9701848444836987e-05,
+ "loss": 1.2122,
+ "step": 3875
+ },
+ {
+ "epoch": 10.561307901907357,
+ "grad_norm": 7.772277355194092,
+ "learning_rate": 1.9701634520200468e-05,
+ "loss": 1.3394,
+ "step": 3876
+ },
+ {
+ "epoch": 10.564032697547685,
+ "grad_norm": 8.325145721435547,
+ "learning_rate": 1.9701420520007928e-05,
+ "loss": 1.2415,
+ "step": 3877
+ },
+ {
+ "epoch": 10.56675749318801,
+ "grad_norm": 7.305802822113037,
+ "learning_rate": 1.970120644426104e-05,
+ "loss": 1.1433,
+ "step": 3878
+ },
+ {
+ "epoch": 10.569482288828338,
+ "grad_norm": 9.247827529907227,
+ "learning_rate": 1.9700992292961464e-05,
+ "loss": 1.3889,
+ "step": 3879
+ },
+ {
+ "epoch": 10.572207084468666,
+ "grad_norm": 12.324711799621582,
+ "learning_rate": 1.9700778066110873e-05,
+ "loss": 1.2961,
+ "step": 3880
+ },
+ {
+ "epoch": 10.574931880108991,
+ "grad_norm": 9.259279251098633,
+ "learning_rate": 1.9700563763710937e-05,
+ "loss": 1.3311,
+ "step": 3881
+ },
+ {
+ "epoch": 10.577656675749319,
+ "grad_norm": 16.73259735107422,
+ "learning_rate": 1.9700349385763323e-05,
+ "loss": 1.1592,
+ "step": 3882
+ },
+ {
+ "epoch": 10.580381471389646,
+ "grad_norm": 10.001047134399414,
+ "learning_rate": 1.9700134932269698e-05,
+ "loss": 0.9969,
+ "step": 3883
+ },
+ {
+ "epoch": 10.583106267029972,
+ "grad_norm": 8.598341941833496,
+ "learning_rate": 1.9699920403231734e-05,
+ "loss": 0.9746,
+ "step": 3884
+ },
+ {
+ "epoch": 10.5858310626703,
+ "grad_norm": 8.761030197143555,
+ "learning_rate": 1.9699705798651104e-05,
+ "loss": 1.2654,
+ "step": 3885
+ },
+ {
+ "epoch": 10.588555858310627,
+ "grad_norm": 7.775053977966309,
+ "learning_rate": 1.9699491118529476e-05,
+ "loss": 1.0703,
+ "step": 3886
+ },
+ {
+ "epoch": 10.591280653950953,
+ "grad_norm": 7.942005157470703,
+ "learning_rate": 1.9699276362868526e-05,
+ "loss": 1.1943,
+ "step": 3887
+ },
+ {
+ "epoch": 10.59400544959128,
+ "grad_norm": 8.420134544372559,
+ "learning_rate": 1.9699061531669922e-05,
+ "loss": 1.183,
+ "step": 3888
+ },
+ {
+ "epoch": 10.596730245231608,
+ "grad_norm": 7.983354568481445,
+ "learning_rate": 1.969884662493534e-05,
+ "loss": 1.1479,
+ "step": 3889
+ },
+ {
+ "epoch": 10.599455040871934,
+ "grad_norm": 7.870924949645996,
+ "learning_rate": 1.969863164266645e-05,
+ "loss": 1.1466,
+ "step": 3890
+ },
+ {
+ "epoch": 10.602179836512262,
+ "grad_norm": 7.549853801727295,
+ "learning_rate": 1.9698416584864932e-05,
+ "loss": 1.3181,
+ "step": 3891
+ },
+ {
+ "epoch": 10.60490463215259,
+ "grad_norm": 8.590463638305664,
+ "learning_rate": 1.9698201451532458e-05,
+ "loss": 1.1636,
+ "step": 3892
+ },
+ {
+ "epoch": 10.607629427792915,
+ "grad_norm": 8.007684707641602,
+ "learning_rate": 1.9697986242670703e-05,
+ "loss": 1.0754,
+ "step": 3893
+ },
+ {
+ "epoch": 10.610354223433243,
+ "grad_norm": 9.13226318359375,
+ "learning_rate": 1.969777095828134e-05,
+ "loss": 1.3057,
+ "step": 3894
+ },
+ {
+ "epoch": 10.61307901907357,
+ "grad_norm": 7.670153617858887,
+ "learning_rate": 1.969755559836605e-05,
+ "loss": 0.9391,
+ "step": 3895
+ },
+ {
+ "epoch": 10.615803814713896,
+ "grad_norm": 7.889237403869629,
+ "learning_rate": 1.969734016292651e-05,
+ "loss": 1.1409,
+ "step": 3896
+ },
+ {
+ "epoch": 10.618528610354224,
+ "grad_norm": 10.103259086608887,
+ "learning_rate": 1.9697124651964397e-05,
+ "loss": 1.1982,
+ "step": 3897
+ },
+ {
+ "epoch": 10.621253405994551,
+ "grad_norm": 13.44520092010498,
+ "learning_rate": 1.969690906548139e-05,
+ "loss": 1.2229,
+ "step": 3898
+ },
+ {
+ "epoch": 10.623978201634877,
+ "grad_norm": 7.624771595001221,
+ "learning_rate": 1.9696693403479165e-05,
+ "loss": 1.3313,
+ "step": 3899
+ },
+ {
+ "epoch": 10.626702997275205,
+ "grad_norm": 12.588001251220703,
+ "learning_rate": 1.9696477665959405e-05,
+ "loss": 1.1904,
+ "step": 3900
+ },
+ {
+ "epoch": 10.629427792915532,
+ "grad_norm": 8.43225383758545,
+ "learning_rate": 1.969626185292379e-05,
+ "loss": 1.137,
+ "step": 3901
+ },
+ {
+ "epoch": 10.632152588555858,
+ "grad_norm": 10.251015663146973,
+ "learning_rate": 1.9696045964374e-05,
+ "loss": 1.1285,
+ "step": 3902
+ },
+ {
+ "epoch": 10.634877384196185,
+ "grad_norm": 9.039337158203125,
+ "learning_rate": 1.9695830000311716e-05,
+ "loss": 1.0513,
+ "step": 3903
+ },
+ {
+ "epoch": 10.637602179836513,
+ "grad_norm": 10.205893516540527,
+ "learning_rate": 1.9695613960738617e-05,
+ "loss": 1.2751,
+ "step": 3904
+ },
+ {
+ "epoch": 10.640326975476839,
+ "grad_norm": 21.56682586669922,
+ "learning_rate": 1.9695397845656393e-05,
+ "loss": 0.9766,
+ "step": 3905
+ },
+ {
+ "epoch": 10.643051771117166,
+ "grad_norm": 9.040979385375977,
+ "learning_rate": 1.9695181655066718e-05,
+ "loss": 1.1393,
+ "step": 3906
+ },
+ {
+ "epoch": 10.645776566757494,
+ "grad_norm": 7.8287248611450195,
+ "learning_rate": 1.9694965388971283e-05,
+ "loss": 1.2354,
+ "step": 3907
+ },
+ {
+ "epoch": 10.64850136239782,
+ "grad_norm": 8.137161254882812,
+ "learning_rate": 1.969474904737177e-05,
+ "loss": 1.1077,
+ "step": 3908
+ },
+ {
+ "epoch": 10.651226158038147,
+ "grad_norm": 6.744873523712158,
+ "learning_rate": 1.9694532630269864e-05,
+ "loss": 1.0603,
+ "step": 3909
+ },
+ {
+ "epoch": 10.653950953678475,
+ "grad_norm": 7.790252685546875,
+ "learning_rate": 1.9694316137667248e-05,
+ "loss": 1.2073,
+ "step": 3910
+ },
+ {
+ "epoch": 10.6566757493188,
+ "grad_norm": 7.70151424407959,
+ "learning_rate": 1.969409956956561e-05,
+ "loss": 1.1245,
+ "step": 3911
+ },
+ {
+ "epoch": 10.659400544959128,
+ "grad_norm": 7.561593055725098,
+ "learning_rate": 1.969388292596664e-05,
+ "loss": 1.005,
+ "step": 3912
+ },
+ {
+ "epoch": 10.662125340599456,
+ "grad_norm": 6.765565872192383,
+ "learning_rate": 1.9693666206872017e-05,
+ "loss": 1.0077,
+ "step": 3913
+ },
+ {
+ "epoch": 10.664850136239782,
+ "grad_norm": 7.8531718254089355,
+ "learning_rate": 1.9693449412283435e-05,
+ "loss": 1.1377,
+ "step": 3914
+ },
+ {
+ "epoch": 10.66757493188011,
+ "grad_norm": 9.242159843444824,
+ "learning_rate": 1.9693232542202582e-05,
+ "loss": 1.2229,
+ "step": 3915
+ },
+ {
+ "epoch": 10.670299727520437,
+ "grad_norm": 8.110725402832031,
+ "learning_rate": 1.9693015596631143e-05,
+ "loss": 1.1121,
+ "step": 3916
+ },
+ {
+ "epoch": 10.673024523160763,
+ "grad_norm": 7.560335636138916,
+ "learning_rate": 1.9692798575570814e-05,
+ "loss": 1.2334,
+ "step": 3917
+ },
+ {
+ "epoch": 10.67574931880109,
+ "grad_norm": 6.967021942138672,
+ "learning_rate": 1.969258147902328e-05,
+ "loss": 1.292,
+ "step": 3918
+ },
+ {
+ "epoch": 10.678474114441418,
+ "grad_norm": 8.248456954956055,
+ "learning_rate": 1.9692364306990235e-05,
+ "loss": 1.1367,
+ "step": 3919
+ },
+ {
+ "epoch": 10.681198910081743,
+ "grad_norm": 8.610984802246094,
+ "learning_rate": 1.9692147059473367e-05,
+ "loss": 1.0854,
+ "step": 3920
+ },
+ {
+ "epoch": 10.683923705722071,
+ "grad_norm": 10.09537124633789,
+ "learning_rate": 1.969192973647437e-05,
+ "loss": 1.1533,
+ "step": 3921
+ },
+ {
+ "epoch": 10.686648501362399,
+ "grad_norm": 8.375680923461914,
+ "learning_rate": 1.9691712337994938e-05,
+ "loss": 1.3547,
+ "step": 3922
+ },
+ {
+ "epoch": 10.689373297002724,
+ "grad_norm": 7.340131759643555,
+ "learning_rate": 1.969149486403676e-05,
+ "loss": 1.1682,
+ "step": 3923
+ },
+ {
+ "epoch": 10.692098092643052,
+ "grad_norm": 7.926560878753662,
+ "learning_rate": 1.9691277314601534e-05,
+ "loss": 1.1702,
+ "step": 3924
+ },
+ {
+ "epoch": 10.69482288828338,
+ "grad_norm": 7.351539611816406,
+ "learning_rate": 1.969105968969095e-05,
+ "loss": 1.2164,
+ "step": 3925
+ },
+ {
+ "epoch": 10.697547683923705,
+ "grad_norm": 8.158242225646973,
+ "learning_rate": 1.9690841989306703e-05,
+ "loss": 1.198,
+ "step": 3926
+ },
+ {
+ "epoch": 10.700272479564033,
+ "grad_norm": 7.358476161956787,
+ "learning_rate": 1.9690624213450493e-05,
+ "loss": 1.1428,
+ "step": 3927
+ },
+ {
+ "epoch": 10.70299727520436,
+ "grad_norm": 8.12960147857666,
+ "learning_rate": 1.9690406362124014e-05,
+ "loss": 1.0957,
+ "step": 3928
+ },
+ {
+ "epoch": 10.705722070844686,
+ "grad_norm": 9.938840866088867,
+ "learning_rate": 1.9690188435328962e-05,
+ "loss": 1.3049,
+ "step": 3929
+ },
+ {
+ "epoch": 10.708446866485014,
+ "grad_norm": 7.457942962646484,
+ "learning_rate": 1.9689970433067035e-05,
+ "loss": 1.1602,
+ "step": 3930
+ },
+ {
+ "epoch": 10.711171662125341,
+ "grad_norm": 9.33325481414795,
+ "learning_rate": 1.9689752355339928e-05,
+ "loss": 1.3438,
+ "step": 3931
+ },
+ {
+ "epoch": 10.713896457765667,
+ "grad_norm": 9.003604888916016,
+ "learning_rate": 1.9689534202149344e-05,
+ "loss": 1.1539,
+ "step": 3932
+ },
+ {
+ "epoch": 10.716621253405995,
+ "grad_norm": 8.176101684570312,
+ "learning_rate": 1.9689315973496978e-05,
+ "loss": 1.2275,
+ "step": 3933
+ },
+ {
+ "epoch": 10.719346049046322,
+ "grad_norm": 7.784386157989502,
+ "learning_rate": 1.9689097669384533e-05,
+ "loss": 1.0452,
+ "step": 3934
+ },
+ {
+ "epoch": 10.722070844686648,
+ "grad_norm": 18.423673629760742,
+ "learning_rate": 1.9688879289813705e-05,
+ "loss": 1.0115,
+ "step": 3935
+ },
+ {
+ "epoch": 10.724795640326976,
+ "grad_norm": 8.284371376037598,
+ "learning_rate": 1.96886608347862e-05,
+ "loss": 1.2068,
+ "step": 3936
+ },
+ {
+ "epoch": 10.727520435967303,
+ "grad_norm": 17.474014282226562,
+ "learning_rate": 1.9688442304303713e-05,
+ "loss": 1.2393,
+ "step": 3937
+ },
+ {
+ "epoch": 10.730245231607629,
+ "grad_norm": 7.8281707763671875,
+ "learning_rate": 1.968822369836795e-05,
+ "loss": 1.1611,
+ "step": 3938
+ },
+ {
+ "epoch": 10.732970027247957,
+ "grad_norm": 7.7236833572387695,
+ "learning_rate": 1.968800501698061e-05,
+ "loss": 1.3236,
+ "step": 3939
+ },
+ {
+ "epoch": 10.735694822888284,
+ "grad_norm": 10.31396484375,
+ "learning_rate": 1.96877862601434e-05,
+ "loss": 1.0922,
+ "step": 3940
+ },
+ {
+ "epoch": 10.73841961852861,
+ "grad_norm": 7.801764488220215,
+ "learning_rate": 1.9687567427858026e-05,
+ "loss": 1.0758,
+ "step": 3941
+ },
+ {
+ "epoch": 10.741144414168938,
+ "grad_norm": 9.994705200195312,
+ "learning_rate": 1.9687348520126187e-05,
+ "loss": 1.2751,
+ "step": 3942
+ },
+ {
+ "epoch": 10.743869209809265,
+ "grad_norm": 9.861577033996582,
+ "learning_rate": 1.9687129536949587e-05,
+ "loss": 1.2371,
+ "step": 3943
+ },
+ {
+ "epoch": 10.746594005449591,
+ "grad_norm": 9.011484146118164,
+ "learning_rate": 1.9686910478329938e-05,
+ "loss": 1.1123,
+ "step": 3944
+ },
+ {
+ "epoch": 10.749318801089919,
+ "grad_norm": 5.850427150726318,
+ "learning_rate": 1.968669134426894e-05,
+ "loss": 1.0098,
+ "step": 3945
+ },
+ {
+ "epoch": 10.752043596730246,
+ "grad_norm": 9.292623519897461,
+ "learning_rate": 1.96864721347683e-05,
+ "loss": 1.2416,
+ "step": 3946
+ },
+ {
+ "epoch": 10.754768392370572,
+ "grad_norm": 8.86750316619873,
+ "learning_rate": 1.968625284982973e-05,
+ "loss": 1.0756,
+ "step": 3947
+ },
+ {
+ "epoch": 10.7574931880109,
+ "grad_norm": 10.271933555603027,
+ "learning_rate": 1.9686033489454934e-05,
+ "loss": 1.408,
+ "step": 3948
+ },
+ {
+ "epoch": 10.760217983651227,
+ "grad_norm": 10.96593189239502,
+ "learning_rate": 1.9685814053645618e-05,
+ "loss": 1.2534,
+ "step": 3949
+ },
+ {
+ "epoch": 10.762942779291553,
+ "grad_norm": 7.65649938583374,
+ "learning_rate": 1.9685594542403494e-05,
+ "loss": 1.1747,
+ "step": 3950
+ },
+ {
+ "epoch": 10.76566757493188,
+ "grad_norm": 9.098298072814941,
+ "learning_rate": 1.9685374955730275e-05,
+ "loss": 1.1104,
+ "step": 3951
+ },
+ {
+ "epoch": 10.768392370572208,
+ "grad_norm": 11.066080093383789,
+ "learning_rate": 1.9685155293627664e-05,
+ "loss": 1.2205,
+ "step": 3952
+ },
+ {
+ "epoch": 10.771117166212534,
+ "grad_norm": 6.907124996185303,
+ "learning_rate": 1.9684935556097378e-05,
+ "loss": 1.0427,
+ "step": 3953
+ },
+ {
+ "epoch": 10.773841961852861,
+ "grad_norm": 9.716444969177246,
+ "learning_rate": 1.9684715743141126e-05,
+ "loss": 1.176,
+ "step": 3954
+ },
+ {
+ "epoch": 10.776566757493189,
+ "grad_norm": 7.205817222595215,
+ "learning_rate": 1.9684495854760616e-05,
+ "loss": 1.2249,
+ "step": 3955
+ },
+ {
+ "epoch": 10.779291553133515,
+ "grad_norm": 7.544517517089844,
+ "learning_rate": 1.9684275890957564e-05,
+ "loss": 1.0177,
+ "step": 3956
+ },
+ {
+ "epoch": 10.782016348773842,
+ "grad_norm": 9.040143966674805,
+ "learning_rate": 1.968405585173369e-05,
+ "loss": 1.4138,
+ "step": 3957
+ },
+ {
+ "epoch": 10.78474114441417,
+ "grad_norm": 8.383528709411621,
+ "learning_rate": 1.968383573709069e-05,
+ "loss": 1.1372,
+ "step": 3958
+ },
+ {
+ "epoch": 10.787465940054496,
+ "grad_norm": 7.467051029205322,
+ "learning_rate": 1.9683615547030296e-05,
+ "loss": 1.0881,
+ "step": 3959
+ },
+ {
+ "epoch": 10.790190735694823,
+ "grad_norm": 8.10969352722168,
+ "learning_rate": 1.9683395281554217e-05,
+ "loss": 1.0349,
+ "step": 3960
+ },
+ {
+ "epoch": 10.79291553133515,
+ "grad_norm": 8.394343376159668,
+ "learning_rate": 1.9683174940664163e-05,
+ "loss": 1.1704,
+ "step": 3961
+ },
+ {
+ "epoch": 10.795640326975477,
+ "grad_norm": 8.338258743286133,
+ "learning_rate": 1.9682954524361853e-05,
+ "loss": 1.1282,
+ "step": 3962
+ },
+ {
+ "epoch": 10.798365122615804,
+ "grad_norm": 7.719311237335205,
+ "learning_rate": 1.968273403264901e-05,
+ "loss": 1.1184,
+ "step": 3963
+ },
+ {
+ "epoch": 10.80108991825613,
+ "grad_norm": 7.421672344207764,
+ "learning_rate": 1.9682513465527345e-05,
+ "loss": 1.2175,
+ "step": 3964
+ },
+ {
+ "epoch": 10.803814713896458,
+ "grad_norm": 9.28896427154541,
+ "learning_rate": 1.9682292822998575e-05,
+ "loss": 1.1619,
+ "step": 3965
+ },
+ {
+ "epoch": 10.806539509536785,
+ "grad_norm": 9.778258323669434,
+ "learning_rate": 1.968207210506442e-05,
+ "loss": 1.4382,
+ "step": 3966
+ },
+ {
+ "epoch": 10.809264305177111,
+ "grad_norm": 10.83773422241211,
+ "learning_rate": 1.96818513117266e-05,
+ "loss": 1.2141,
+ "step": 3967
+ },
+ {
+ "epoch": 10.811989100817438,
+ "grad_norm": 8.503150939941406,
+ "learning_rate": 1.968163044298683e-05,
+ "loss": 1.2845,
+ "step": 3968
+ },
+ {
+ "epoch": 10.814713896457766,
+ "grad_norm": 8.739851951599121,
+ "learning_rate": 1.968140949884684e-05,
+ "loss": 1.1726,
+ "step": 3969
+ },
+ {
+ "epoch": 10.817438692098092,
+ "grad_norm": 16.882143020629883,
+ "learning_rate": 1.9681188479308343e-05,
+ "loss": 1.0481,
+ "step": 3970
+ },
+ {
+ "epoch": 10.82016348773842,
+ "grad_norm": 8.620797157287598,
+ "learning_rate": 1.9680967384373058e-05,
+ "loss": 1.2856,
+ "step": 3971
+ },
+ {
+ "epoch": 10.822888283378747,
+ "grad_norm": 9.403024673461914,
+ "learning_rate": 1.9680746214042714e-05,
+ "loss": 1.1787,
+ "step": 3972
+ },
+ {
+ "epoch": 10.825613079019073,
+ "grad_norm": 7.95114803314209,
+ "learning_rate": 1.9680524968319027e-05,
+ "loss": 1.1379,
+ "step": 3973
+ },
+ {
+ "epoch": 10.8283378746594,
+ "grad_norm": 7.485889434814453,
+ "learning_rate": 1.9680303647203723e-05,
+ "loss": 1.1501,
+ "step": 3974
+ },
+ {
+ "epoch": 10.831062670299728,
+ "grad_norm": 7.693819046020508,
+ "learning_rate": 1.968008225069853e-05,
+ "loss": 1.1084,
+ "step": 3975
+ },
+ {
+ "epoch": 10.833787465940054,
+ "grad_norm": 7.251269340515137,
+ "learning_rate": 1.9679860778805167e-05,
+ "loss": 1.2551,
+ "step": 3976
+ },
+ {
+ "epoch": 10.836512261580381,
+ "grad_norm": 19.532175064086914,
+ "learning_rate": 1.9679639231525357e-05,
+ "loss": 1.1777,
+ "step": 3977
+ },
+ {
+ "epoch": 10.839237057220709,
+ "grad_norm": 14.64571762084961,
+ "learning_rate": 1.9679417608860828e-05,
+ "loss": 1.0488,
+ "step": 3978
+ },
+ {
+ "epoch": 10.841961852861035,
+ "grad_norm": 7.412012100219727,
+ "learning_rate": 1.9679195910813308e-05,
+ "loss": 1.0646,
+ "step": 3979
+ },
+ {
+ "epoch": 10.844686648501362,
+ "grad_norm": 8.485784530639648,
+ "learning_rate": 1.9678974137384524e-05,
+ "loss": 1.0432,
+ "step": 3980
+ },
+ {
+ "epoch": 10.84741144414169,
+ "grad_norm": 7.922987937927246,
+ "learning_rate": 1.9678752288576198e-05,
+ "loss": 1.2507,
+ "step": 3981
+ },
+ {
+ "epoch": 10.850136239782016,
+ "grad_norm": 7.283071994781494,
+ "learning_rate": 1.967853036439006e-05,
+ "loss": 1.0754,
+ "step": 3982
+ },
+ {
+ "epoch": 10.852861035422343,
+ "grad_norm": 9.855236053466797,
+ "learning_rate": 1.967830836482784e-05,
+ "loss": 1.0935,
+ "step": 3983
+ },
+ {
+ "epoch": 10.85558583106267,
+ "grad_norm": 7.983922958374023,
+ "learning_rate": 1.9678086289891265e-05,
+ "loss": 1.0237,
+ "step": 3984
+ },
+ {
+ "epoch": 10.858310626702997,
+ "grad_norm": 11.663695335388184,
+ "learning_rate": 1.9677864139582066e-05,
+ "loss": 1.4836,
+ "step": 3985
+ },
+ {
+ "epoch": 10.861035422343324,
+ "grad_norm": 6.82507848739624,
+ "learning_rate": 1.9677641913901975e-05,
+ "loss": 1.0743,
+ "step": 3986
+ },
+ {
+ "epoch": 10.863760217983652,
+ "grad_norm": 6.450811386108398,
+ "learning_rate": 1.9677419612852716e-05,
+ "loss": 1.0247,
+ "step": 3987
+ },
+ {
+ "epoch": 10.866485013623977,
+ "grad_norm": 7.571865081787109,
+ "learning_rate": 1.9677197236436027e-05,
+ "loss": 1.2283,
+ "step": 3988
+ },
+ {
+ "epoch": 10.869209809264305,
+ "grad_norm": 8.740765571594238,
+ "learning_rate": 1.967697478465364e-05,
+ "loss": 1.1443,
+ "step": 3989
+ },
+ {
+ "epoch": 10.871934604904633,
+ "grad_norm": 8.630934715270996,
+ "learning_rate": 1.9676752257507282e-05,
+ "loss": 1.0513,
+ "step": 3990
+ },
+ {
+ "epoch": 10.874659400544958,
+ "grad_norm": 9.44717025756836,
+ "learning_rate": 1.967652965499869e-05,
+ "loss": 1.1893,
+ "step": 3991
+ },
+ {
+ "epoch": 10.877384196185286,
+ "grad_norm": 8.772364616394043,
+ "learning_rate": 1.9676306977129595e-05,
+ "loss": 1.1353,
+ "step": 3992
+ },
+ {
+ "epoch": 10.880108991825614,
+ "grad_norm": 6.646779537200928,
+ "learning_rate": 1.967608422390173e-05,
+ "loss": 1.004,
+ "step": 3993
+ },
+ {
+ "epoch": 10.88283378746594,
+ "grad_norm": 12.866748809814453,
+ "learning_rate": 1.967586139531684e-05,
+ "loss": 1.3582,
+ "step": 3994
+ },
+ {
+ "epoch": 10.885558583106267,
+ "grad_norm": 7.846841335296631,
+ "learning_rate": 1.9675638491376646e-05,
+ "loss": 1.2777,
+ "step": 3995
+ },
+ {
+ "epoch": 10.888283378746594,
+ "grad_norm": 10.780961036682129,
+ "learning_rate": 1.9675415512082895e-05,
+ "loss": 1.2417,
+ "step": 3996
+ },
+ {
+ "epoch": 10.89100817438692,
+ "grad_norm": 8.2462797164917,
+ "learning_rate": 1.9675192457437316e-05,
+ "loss": 1.2009,
+ "step": 3997
+ },
+ {
+ "epoch": 10.893732970027248,
+ "grad_norm": 6.949042797088623,
+ "learning_rate": 1.9674969327441652e-05,
+ "loss": 1.1777,
+ "step": 3998
+ },
+ {
+ "epoch": 10.896457765667575,
+ "grad_norm": 9.260941505432129,
+ "learning_rate": 1.9674746122097637e-05,
+ "loss": 1.2566,
+ "step": 3999
+ },
+ {
+ "epoch": 10.899182561307901,
+ "grad_norm": 8.459162712097168,
+ "learning_rate": 1.967452284140701e-05,
+ "loss": 1.063,
+ "step": 4000
+ },
+ {
+ "epoch": 10.901907356948229,
+ "grad_norm": 7.8846116065979,
+ "learning_rate": 1.967429948537151e-05,
+ "loss": 1.0645,
+ "step": 4001
+ },
+ {
+ "epoch": 10.904632152588556,
+ "grad_norm": 7.580723762512207,
+ "learning_rate": 1.967407605399288e-05,
+ "loss": 1.1626,
+ "step": 4002
+ },
+ {
+ "epoch": 10.907356948228882,
+ "grad_norm": 7.422186851501465,
+ "learning_rate": 1.967385254727286e-05,
+ "loss": 1.1195,
+ "step": 4003
+ },
+ {
+ "epoch": 10.91008174386921,
+ "grad_norm": 6.743069648742676,
+ "learning_rate": 1.9673628965213177e-05,
+ "loss": 1.0981,
+ "step": 4004
+ },
+ {
+ "epoch": 10.912806539509537,
+ "grad_norm": 9.216634750366211,
+ "learning_rate": 1.9673405307815592e-05,
+ "loss": 1.0603,
+ "step": 4005
+ },
+ {
+ "epoch": 10.915531335149863,
+ "grad_norm": 7.595468997955322,
+ "learning_rate": 1.9673181575081837e-05,
+ "loss": 1.0574,
+ "step": 4006
+ },
+ {
+ "epoch": 10.91825613079019,
+ "grad_norm": 8.560811042785645,
+ "learning_rate": 1.9672957767013652e-05,
+ "loss": 1.3442,
+ "step": 4007
+ },
+ {
+ "epoch": 10.920980926430518,
+ "grad_norm": 9.088674545288086,
+ "learning_rate": 1.9672733883612786e-05,
+ "loss": 1.0693,
+ "step": 4008
+ },
+ {
+ "epoch": 10.923705722070844,
+ "grad_norm": 9.471978187561035,
+ "learning_rate": 1.9672509924880977e-05,
+ "loss": 1.106,
+ "step": 4009
+ },
+ {
+ "epoch": 10.926430517711172,
+ "grad_norm": 7.067034721374512,
+ "learning_rate": 1.9672285890819974e-05,
+ "loss": 1.1902,
+ "step": 4010
+ },
+ {
+ "epoch": 10.9291553133515,
+ "grad_norm": 7.774810791015625,
+ "learning_rate": 1.967206178143152e-05,
+ "loss": 1.0181,
+ "step": 4011
+ },
+ {
+ "epoch": 10.931880108991825,
+ "grad_norm": 8.161649703979492,
+ "learning_rate": 1.967183759671736e-05,
+ "loss": 1.1851,
+ "step": 4012
+ },
+ {
+ "epoch": 10.934604904632153,
+ "grad_norm": 7.094213008880615,
+ "learning_rate": 1.9671613336679237e-05,
+ "loss": 1.0645,
+ "step": 4013
+ },
+ {
+ "epoch": 10.93732970027248,
+ "grad_norm": 9.804855346679688,
+ "learning_rate": 1.9671389001318904e-05,
+ "loss": 1.2017,
+ "step": 4014
+ },
+ {
+ "epoch": 10.940054495912806,
+ "grad_norm": 7.829232215881348,
+ "learning_rate": 1.96711645906381e-05,
+ "loss": 1.0676,
+ "step": 4015
+ },
+ {
+ "epoch": 10.942779291553133,
+ "grad_norm": 11.222532272338867,
+ "learning_rate": 1.9670940104638582e-05,
+ "loss": 1.397,
+ "step": 4016
+ },
+ {
+ "epoch": 10.945504087193461,
+ "grad_norm": 9.05797290802002,
+ "learning_rate": 1.9670715543322093e-05,
+ "loss": 1.22,
+ "step": 4017
+ },
+ {
+ "epoch": 10.948228882833787,
+ "grad_norm": 6.172121047973633,
+ "learning_rate": 1.9670490906690386e-05,
+ "loss": 0.9,
+ "step": 4018
+ },
+ {
+ "epoch": 10.950953678474114,
+ "grad_norm": 6.866868019104004,
+ "learning_rate": 1.96702661947452e-05,
+ "loss": 1.0474,
+ "step": 4019
+ },
+ {
+ "epoch": 10.953678474114442,
+ "grad_norm": 9.9358549118042,
+ "learning_rate": 1.9670041407488298e-05,
+ "loss": 1.3047,
+ "step": 4020
+ },
+ {
+ "epoch": 10.956403269754768,
+ "grad_norm": 15.360501289367676,
+ "learning_rate": 1.966981654492142e-05,
+ "loss": 1.1531,
+ "step": 4021
+ },
+ {
+ "epoch": 10.959128065395095,
+ "grad_norm": 8.18236255645752,
+ "learning_rate": 1.9669591607046325e-05,
+ "loss": 1.438,
+ "step": 4022
+ },
+ {
+ "epoch": 10.961852861035423,
+ "grad_norm": 6.540932655334473,
+ "learning_rate": 1.9669366593864763e-05,
+ "loss": 1.0188,
+ "step": 4023
+ },
+ {
+ "epoch": 10.964577656675749,
+ "grad_norm": 9.190437316894531,
+ "learning_rate": 1.966914150537848e-05,
+ "loss": 1.2109,
+ "step": 4024
+ },
+ {
+ "epoch": 10.967302452316076,
+ "grad_norm": 6.407480239868164,
+ "learning_rate": 1.9668916341589242e-05,
+ "loss": 1.0989,
+ "step": 4025
+ },
+ {
+ "epoch": 10.970027247956404,
+ "grad_norm": 9.326363563537598,
+ "learning_rate": 1.966869110249879e-05,
+ "loss": 1.0818,
+ "step": 4026
+ },
+ {
+ "epoch": 10.97275204359673,
+ "grad_norm": 7.70641565322876,
+ "learning_rate": 1.966846578810888e-05,
+ "loss": 1.1974,
+ "step": 4027
+ },
+ {
+ "epoch": 10.975476839237057,
+ "grad_norm": 9.702550888061523,
+ "learning_rate": 1.9668240398421275e-05,
+ "loss": 1.197,
+ "step": 4028
+ },
+ {
+ "epoch": 10.978201634877385,
+ "grad_norm": 6.666703224182129,
+ "learning_rate": 1.966801493343772e-05,
+ "loss": 1.052,
+ "step": 4029
+ },
+ {
+ "epoch": 10.98092643051771,
+ "grad_norm": 7.76691198348999,
+ "learning_rate": 1.966778939315998e-05,
+ "loss": 1.2256,
+ "step": 4030
+ },
+ {
+ "epoch": 10.983651226158038,
+ "grad_norm": 9.64727783203125,
+ "learning_rate": 1.9667563777589805e-05,
+ "loss": 1.0485,
+ "step": 4031
+ },
+ {
+ "epoch": 10.986376021798366,
+ "grad_norm": 9.86203384399414,
+ "learning_rate": 1.9667338086728955e-05,
+ "loss": 1.3101,
+ "step": 4032
+ },
+ {
+ "epoch": 10.989100817438691,
+ "grad_norm": 7.948212146759033,
+ "learning_rate": 1.966711232057919e-05,
+ "loss": 1.2087,
+ "step": 4033
+ },
+ {
+ "epoch": 10.991825613079019,
+ "grad_norm": 7.725721836090088,
+ "learning_rate": 1.966688647914226e-05,
+ "loss": 1.0043,
+ "step": 4034
+ },
+ {
+ "epoch": 10.994550408719347,
+ "grad_norm": 9.455055236816406,
+ "learning_rate": 1.9666660562419936e-05,
+ "loss": 1.3184,
+ "step": 4035
+ },
+ {
+ "epoch": 10.997275204359672,
+ "grad_norm": 8.816896438598633,
+ "learning_rate": 1.9666434570413964e-05,
+ "loss": 1.2233,
+ "step": 4036
+ },
+ {
+ "epoch": 11.0,
+ "grad_norm": 11.72046947479248,
+ "learning_rate": 1.9666208503126115e-05,
+ "loss": 1.043,
+ "step": 4037
+ },
+ {
+ "epoch": 11.002724795640328,
+ "grad_norm": 6.440786361694336,
+ "learning_rate": 1.9665982360558143e-05,
+ "loss": 1.1156,
+ "step": 4038
+ },
+ {
+ "epoch": 11.005449591280653,
+ "grad_norm": 9.08065128326416,
+ "learning_rate": 1.966575614271181e-05,
+ "loss": 1.0593,
+ "step": 4039
+ },
+ {
+ "epoch": 11.008174386920981,
+ "grad_norm": 7.377621650695801,
+ "learning_rate": 1.9665529849588882e-05,
+ "loss": 1.1218,
+ "step": 4040
+ },
+ {
+ "epoch": 11.010899182561309,
+ "grad_norm": 18.99961280822754,
+ "learning_rate": 1.966530348119112e-05,
+ "loss": 0.9414,
+ "step": 4041
+ },
+ {
+ "epoch": 11.013623978201634,
+ "grad_norm": 5.680116653442383,
+ "learning_rate": 1.9665077037520283e-05,
+ "loss": 0.9194,
+ "step": 4042
+ },
+ {
+ "epoch": 11.016348773841962,
+ "grad_norm": 7.087353706359863,
+ "learning_rate": 1.966485051857814e-05,
+ "loss": 1.0015,
+ "step": 4043
+ },
+ {
+ "epoch": 11.01907356948229,
+ "grad_norm": 7.583617210388184,
+ "learning_rate": 1.9664623924366446e-05,
+ "loss": 1.2432,
+ "step": 4044
+ },
+ {
+ "epoch": 11.021798365122615,
+ "grad_norm": 7.95800256729126,
+ "learning_rate": 1.9664397254886975e-05,
+ "loss": 1.2769,
+ "step": 4045
+ },
+ {
+ "epoch": 11.024523160762943,
+ "grad_norm": 7.22228479385376,
+ "learning_rate": 1.9664170510141495e-05,
+ "loss": 1.14,
+ "step": 4046
+ },
+ {
+ "epoch": 11.02724795640327,
+ "grad_norm": 7.350344657897949,
+ "learning_rate": 1.966394369013176e-05,
+ "loss": 1.0356,
+ "step": 4047
+ },
+ {
+ "epoch": 11.029972752043596,
+ "grad_norm": 8.845749855041504,
+ "learning_rate": 1.9663716794859543e-05,
+ "loss": 1.0864,
+ "step": 4048
+ },
+ {
+ "epoch": 11.032697547683924,
+ "grad_norm": 5.7134552001953125,
+ "learning_rate": 1.9663489824326612e-05,
+ "loss": 0.9784,
+ "step": 4049
+ },
+ {
+ "epoch": 11.035422343324251,
+ "grad_norm": 10.388118743896484,
+ "learning_rate": 1.966326277853473e-05,
+ "loss": 1.1653,
+ "step": 4050
+ },
+ {
+ "epoch": 11.038147138964577,
+ "grad_norm": 9.788719177246094,
+ "learning_rate": 1.9663035657485673e-05,
+ "loss": 1.0022,
+ "step": 4051
+ },
+ {
+ "epoch": 11.040871934604905,
+ "grad_norm": 6.892394065856934,
+ "learning_rate": 1.9662808461181206e-05,
+ "loss": 0.9648,
+ "step": 4052
+ },
+ {
+ "epoch": 11.043596730245232,
+ "grad_norm": 7.546081066131592,
+ "learning_rate": 1.9662581189623093e-05,
+ "loss": 0.9475,
+ "step": 4053
+ },
+ {
+ "epoch": 11.046321525885558,
+ "grad_norm": 7.647053241729736,
+ "learning_rate": 1.9662353842813112e-05,
+ "loss": 1.1475,
+ "step": 4054
+ },
+ {
+ "epoch": 11.049046321525886,
+ "grad_norm": 7.968468189239502,
+ "learning_rate": 1.9662126420753026e-05,
+ "loss": 1.0687,
+ "step": 4055
+ },
+ {
+ "epoch": 11.051771117166213,
+ "grad_norm": 6.872508525848389,
+ "learning_rate": 1.9661898923444616e-05,
+ "loss": 0.9222,
+ "step": 4056
+ },
+ {
+ "epoch": 11.054495912806539,
+ "grad_norm": 6.17036771774292,
+ "learning_rate": 1.9661671350889645e-05,
+ "loss": 0.8577,
+ "step": 4057
+ },
+ {
+ "epoch": 11.057220708446867,
+ "grad_norm": 7.6626434326171875,
+ "learning_rate": 1.9661443703089888e-05,
+ "loss": 0.8716,
+ "step": 4058
+ },
+ {
+ "epoch": 11.059945504087194,
+ "grad_norm": 10.566581726074219,
+ "learning_rate": 1.966121598004712e-05,
+ "loss": 1.1716,
+ "step": 4059
+ },
+ {
+ "epoch": 11.06267029972752,
+ "grad_norm": 8.921685218811035,
+ "learning_rate": 1.9660988181763114e-05,
+ "loss": 1.0935,
+ "step": 4060
+ },
+ {
+ "epoch": 11.065395095367847,
+ "grad_norm": 9.120361328125,
+ "learning_rate": 1.966076030823964e-05,
+ "loss": 0.925,
+ "step": 4061
+ },
+ {
+ "epoch": 11.068119891008175,
+ "grad_norm": 9.817638397216797,
+ "learning_rate": 1.9660532359478476e-05,
+ "loss": 1.0197,
+ "step": 4062
+ },
+ {
+ "epoch": 11.0708446866485,
+ "grad_norm": 8.099934577941895,
+ "learning_rate": 1.9660304335481398e-05,
+ "loss": 1.0557,
+ "step": 4063
+ },
+ {
+ "epoch": 11.073569482288828,
+ "grad_norm": 10.00467300415039,
+ "learning_rate": 1.966007623625018e-05,
+ "loss": 1.0203,
+ "step": 4064
+ },
+ {
+ "epoch": 11.076294277929156,
+ "grad_norm": 8.336186408996582,
+ "learning_rate": 1.96598480617866e-05,
+ "loss": 1.0862,
+ "step": 4065
+ },
+ {
+ "epoch": 11.079019073569482,
+ "grad_norm": 8.720233917236328,
+ "learning_rate": 1.9659619812092434e-05,
+ "loss": 1.007,
+ "step": 4066
+ },
+ {
+ "epoch": 11.08174386920981,
+ "grad_norm": 7.31000280380249,
+ "learning_rate": 1.965939148716946e-05,
+ "loss": 0.8894,
+ "step": 4067
+ },
+ {
+ "epoch": 11.084468664850137,
+ "grad_norm": 9.147107124328613,
+ "learning_rate": 1.9659163087019454e-05,
+ "loss": 0.9534,
+ "step": 4068
+ },
+ {
+ "epoch": 11.087193460490463,
+ "grad_norm": 6.494155406951904,
+ "learning_rate": 1.96589346116442e-05,
+ "loss": 0.8719,
+ "step": 4069
+ },
+ {
+ "epoch": 11.08991825613079,
+ "grad_norm": 9.749225616455078,
+ "learning_rate": 1.9658706061045473e-05,
+ "loss": 1.0601,
+ "step": 4070
+ },
+ {
+ "epoch": 11.092643051771118,
+ "grad_norm": 8.545897483825684,
+ "learning_rate": 1.9658477435225053e-05,
+ "loss": 1.0754,
+ "step": 4071
+ },
+ {
+ "epoch": 11.095367847411444,
+ "grad_norm": 6.812800884246826,
+ "learning_rate": 1.9658248734184725e-05,
+ "loss": 0.9545,
+ "step": 4072
+ },
+ {
+ "epoch": 11.098092643051771,
+ "grad_norm": 6.0325927734375,
+ "learning_rate": 1.9658019957926263e-05,
+ "loss": 0.8706,
+ "step": 4073
+ },
+ {
+ "epoch": 11.100817438692099,
+ "grad_norm": 9.972147941589355,
+ "learning_rate": 1.9657791106451454e-05,
+ "loss": 1.1357,
+ "step": 4074
+ },
+ {
+ "epoch": 11.103542234332425,
+ "grad_norm": 11.526996612548828,
+ "learning_rate": 1.9657562179762078e-05,
+ "loss": 1.123,
+ "step": 4075
+ },
+ {
+ "epoch": 11.106267029972752,
+ "grad_norm": 10.470778465270996,
+ "learning_rate": 1.965733317785992e-05,
+ "loss": 1.4003,
+ "step": 4076
+ },
+ {
+ "epoch": 11.10899182561308,
+ "grad_norm": 41.16800308227539,
+ "learning_rate": 1.965710410074676e-05,
+ "loss": 1.1692,
+ "step": 4077
+ },
+ {
+ "epoch": 11.111716621253406,
+ "grad_norm": 6.447212219238281,
+ "learning_rate": 1.9656874948424384e-05,
+ "loss": 0.9832,
+ "step": 4078
+ },
+ {
+ "epoch": 11.114441416893733,
+ "grad_norm": 7.034350872039795,
+ "learning_rate": 1.9656645720894577e-05,
+ "loss": 1.0288,
+ "step": 4079
+ },
+ {
+ "epoch": 11.11716621253406,
+ "grad_norm": 13.16606616973877,
+ "learning_rate": 1.9656416418159127e-05,
+ "loss": 0.9401,
+ "step": 4080
+ },
+ {
+ "epoch": 11.119891008174386,
+ "grad_norm": 8.847200393676758,
+ "learning_rate": 1.9656187040219812e-05,
+ "loss": 1.1851,
+ "step": 4081
+ },
+ {
+ "epoch": 11.122615803814714,
+ "grad_norm": 9.639972686767578,
+ "learning_rate": 1.9655957587078426e-05,
+ "loss": 1.0765,
+ "step": 4082
+ },
+ {
+ "epoch": 11.125340599455042,
+ "grad_norm": 6.945981502532959,
+ "learning_rate": 1.9655728058736753e-05,
+ "loss": 1.061,
+ "step": 4083
+ },
+ {
+ "epoch": 11.128065395095367,
+ "grad_norm": 8.499841690063477,
+ "learning_rate": 1.9655498455196576e-05,
+ "loss": 0.999,
+ "step": 4084
+ },
+ {
+ "epoch": 11.130790190735695,
+ "grad_norm": 8.03131103515625,
+ "learning_rate": 1.9655268776459694e-05,
+ "loss": 1.1819,
+ "step": 4085
+ },
+ {
+ "epoch": 11.133514986376023,
+ "grad_norm": 8.216924667358398,
+ "learning_rate": 1.965503902252789e-05,
+ "loss": 1.0276,
+ "step": 4086
+ },
+ {
+ "epoch": 11.136239782016348,
+ "grad_norm": 8.959362983703613,
+ "learning_rate": 1.965480919340295e-05,
+ "loss": 1.02,
+ "step": 4087
+ },
+ {
+ "epoch": 11.138964577656676,
+ "grad_norm": 8.515168190002441,
+ "learning_rate": 1.9654579289086665e-05,
+ "loss": 0.9724,
+ "step": 4088
+ },
+ {
+ "epoch": 11.141689373297003,
+ "grad_norm": 8.281926155090332,
+ "learning_rate": 1.965434930958083e-05,
+ "loss": 0.8901,
+ "step": 4089
+ },
+ {
+ "epoch": 11.14441416893733,
+ "grad_norm": 7.039126396179199,
+ "learning_rate": 1.965411925488723e-05,
+ "loss": 1.0291,
+ "step": 4090
+ },
+ {
+ "epoch": 11.147138964577657,
+ "grad_norm": 8.549878120422363,
+ "learning_rate": 1.9653889125007665e-05,
+ "loss": 0.9561,
+ "step": 4091
+ },
+ {
+ "epoch": 11.149863760217984,
+ "grad_norm": 5.93820333480835,
+ "learning_rate": 1.965365891994392e-05,
+ "loss": 0.7604,
+ "step": 4092
+ },
+ {
+ "epoch": 11.15258855585831,
+ "grad_norm": 7.722904682159424,
+ "learning_rate": 1.9653428639697788e-05,
+ "loss": 1.0602,
+ "step": 4093
+ },
+ {
+ "epoch": 11.155313351498638,
+ "grad_norm": 7.734607696533203,
+ "learning_rate": 1.9653198284271065e-05,
+ "loss": 1.1741,
+ "step": 4094
+ },
+ {
+ "epoch": 11.158038147138965,
+ "grad_norm": 8.349475860595703,
+ "learning_rate": 1.9652967853665547e-05,
+ "loss": 0.8684,
+ "step": 4095
+ },
+ {
+ "epoch": 11.160762942779291,
+ "grad_norm": 6.856114387512207,
+ "learning_rate": 1.9652737347883025e-05,
+ "loss": 0.9692,
+ "step": 4096
+ },
+ {
+ "epoch": 11.163487738419619,
+ "grad_norm": 7.792014122009277,
+ "learning_rate": 1.9652506766925296e-05,
+ "loss": 0.9426,
+ "step": 4097
+ },
+ {
+ "epoch": 11.166212534059946,
+ "grad_norm": 6.886579513549805,
+ "learning_rate": 1.965227611079415e-05,
+ "loss": 0.8748,
+ "step": 4098
+ },
+ {
+ "epoch": 11.168937329700272,
+ "grad_norm": 7.984208106994629,
+ "learning_rate": 1.9652045379491395e-05,
+ "loss": 1.1666,
+ "step": 4099
+ },
+ {
+ "epoch": 11.1716621253406,
+ "grad_norm": 12.934913635253906,
+ "learning_rate": 1.9651814573018818e-05,
+ "loss": 1.0282,
+ "step": 4100
+ },
+ {
+ "epoch": 11.174386920980927,
+ "grad_norm": 7.4530253410339355,
+ "learning_rate": 1.965158369137822e-05,
+ "loss": 1.2424,
+ "step": 4101
+ },
+ {
+ "epoch": 11.177111716621253,
+ "grad_norm": 8.283652305603027,
+ "learning_rate": 1.96513527345714e-05,
+ "loss": 1.1567,
+ "step": 4102
+ },
+ {
+ "epoch": 11.17983651226158,
+ "grad_norm": 6.190092086791992,
+ "learning_rate": 1.9651121702600155e-05,
+ "loss": 1.0322,
+ "step": 4103
+ },
+ {
+ "epoch": 11.182561307901908,
+ "grad_norm": 9.445829391479492,
+ "learning_rate": 1.9650890595466284e-05,
+ "loss": 0.9178,
+ "step": 4104
+ },
+ {
+ "epoch": 11.185286103542234,
+ "grad_norm": 8.530731201171875,
+ "learning_rate": 1.965065941317159e-05,
+ "loss": 1.0237,
+ "step": 4105
+ },
+ {
+ "epoch": 11.188010899182562,
+ "grad_norm": 6.681700229644775,
+ "learning_rate": 1.9650428155717867e-05,
+ "loss": 1.1252,
+ "step": 4106
+ },
+ {
+ "epoch": 11.190735694822889,
+ "grad_norm": 9.164388656616211,
+ "learning_rate": 1.965019682310692e-05,
+ "loss": 1.2725,
+ "step": 4107
+ },
+ {
+ "epoch": 11.193460490463215,
+ "grad_norm": 13.078254699707031,
+ "learning_rate": 1.9649965415340553e-05,
+ "loss": 1.0651,
+ "step": 4108
+ },
+ {
+ "epoch": 11.196185286103542,
+ "grad_norm": 6.8021721839904785,
+ "learning_rate": 1.9649733932420566e-05,
+ "loss": 1.2031,
+ "step": 4109
+ },
+ {
+ "epoch": 11.19891008174387,
+ "grad_norm": 8.907111167907715,
+ "learning_rate": 1.9649502374348763e-05,
+ "loss": 0.962,
+ "step": 4110
+ },
+ {
+ "epoch": 11.201634877384196,
+ "grad_norm": 12.938401222229004,
+ "learning_rate": 1.9649270741126944e-05,
+ "loss": 1.2463,
+ "step": 4111
+ },
+ {
+ "epoch": 11.204359673024523,
+ "grad_norm": 7.75813627243042,
+ "learning_rate": 1.9649039032756913e-05,
+ "loss": 1.1289,
+ "step": 4112
+ },
+ {
+ "epoch": 11.207084468664851,
+ "grad_norm": 8.361738204956055,
+ "learning_rate": 1.964880724924048e-05,
+ "loss": 1.1453,
+ "step": 4113
+ },
+ {
+ "epoch": 11.209809264305177,
+ "grad_norm": 7.279664516448975,
+ "learning_rate": 1.9648575390579447e-05,
+ "loss": 1.2229,
+ "step": 4114
+ },
+ {
+ "epoch": 11.212534059945504,
+ "grad_norm": 7.184321880340576,
+ "learning_rate": 1.964834345677562e-05,
+ "loss": 1.3884,
+ "step": 4115
+ },
+ {
+ "epoch": 11.215258855585832,
+ "grad_norm": 7.697027683258057,
+ "learning_rate": 1.96481114478308e-05,
+ "loss": 0.9807,
+ "step": 4116
+ },
+ {
+ "epoch": 11.217983651226158,
+ "grad_norm": 7.138815402984619,
+ "learning_rate": 1.96478793637468e-05,
+ "loss": 1.1998,
+ "step": 4117
+ },
+ {
+ "epoch": 11.220708446866485,
+ "grad_norm": 8.731977462768555,
+ "learning_rate": 1.964764720452543e-05,
+ "loss": 1.0516,
+ "step": 4118
+ },
+ {
+ "epoch": 11.223433242506813,
+ "grad_norm": 6.829278945922852,
+ "learning_rate": 1.964741497016849e-05,
+ "loss": 1.1632,
+ "step": 4119
+ },
+ {
+ "epoch": 11.226158038147139,
+ "grad_norm": 6.991519927978516,
+ "learning_rate": 1.9647182660677796e-05,
+ "loss": 0.8855,
+ "step": 4120
+ },
+ {
+ "epoch": 11.228882833787466,
+ "grad_norm": 6.657474517822266,
+ "learning_rate": 1.9646950276055153e-05,
+ "loss": 1.0239,
+ "step": 4121
+ },
+ {
+ "epoch": 11.231607629427794,
+ "grad_norm": 7.161469459533691,
+ "learning_rate": 1.964671781630237e-05,
+ "loss": 0.9976,
+ "step": 4122
+ },
+ {
+ "epoch": 11.23433242506812,
+ "grad_norm": 9.29500675201416,
+ "learning_rate": 1.964648528142126e-05,
+ "loss": 1.1016,
+ "step": 4123
+ },
+ {
+ "epoch": 11.237057220708447,
+ "grad_norm": 6.218743801116943,
+ "learning_rate": 1.9646252671413633e-05,
+ "loss": 0.87,
+ "step": 4124
+ },
+ {
+ "epoch": 11.239782016348773,
+ "grad_norm": 7.202930927276611,
+ "learning_rate": 1.96460199862813e-05,
+ "loss": 1.0542,
+ "step": 4125
+ },
+ {
+ "epoch": 11.2425068119891,
+ "grad_norm": 10.74658203125,
+ "learning_rate": 1.9645787226026075e-05,
+ "loss": 1.0764,
+ "step": 4126
+ },
+ {
+ "epoch": 11.245231607629428,
+ "grad_norm": 9.060245513916016,
+ "learning_rate": 1.964555439064977e-05,
+ "loss": 1.0808,
+ "step": 4127
+ },
+ {
+ "epoch": 11.247956403269754,
+ "grad_norm": 8.77108383178711,
+ "learning_rate": 1.96453214801542e-05,
+ "loss": 1.1189,
+ "step": 4128
+ },
+ {
+ "epoch": 11.250681198910081,
+ "grad_norm": 8.342645645141602,
+ "learning_rate": 1.9645088494541172e-05,
+ "loss": 1.1555,
+ "step": 4129
+ },
+ {
+ "epoch": 11.253405994550409,
+ "grad_norm": 9.413081169128418,
+ "learning_rate": 1.9644855433812507e-05,
+ "loss": 1.285,
+ "step": 4130
+ },
+ {
+ "epoch": 11.256130790190735,
+ "grad_norm": 7.048334121704102,
+ "learning_rate": 1.9644622297970018e-05,
+ "loss": 1.0137,
+ "step": 4131
+ },
+ {
+ "epoch": 11.258855585831062,
+ "grad_norm": 12.10335636138916,
+ "learning_rate": 1.964438908701552e-05,
+ "loss": 1.1187,
+ "step": 4132
+ },
+ {
+ "epoch": 11.26158038147139,
+ "grad_norm": 8.42898178100586,
+ "learning_rate": 1.9644155800950832e-05,
+ "loss": 1.1182,
+ "step": 4133
+ },
+ {
+ "epoch": 11.264305177111716,
+ "grad_norm": 6.698251724243164,
+ "learning_rate": 1.964392243977777e-05,
+ "loss": 1.1055,
+ "step": 4134
+ },
+ {
+ "epoch": 11.267029972752043,
+ "grad_norm": 9.098014831542969,
+ "learning_rate": 1.9643689003498147e-05,
+ "loss": 1.1422,
+ "step": 4135
+ },
+ {
+ "epoch": 11.269754768392371,
+ "grad_norm": 7.705501079559326,
+ "learning_rate": 1.964345549211379e-05,
+ "loss": 1.0651,
+ "step": 4136
+ },
+ {
+ "epoch": 11.272479564032697,
+ "grad_norm": 7.653578281402588,
+ "learning_rate": 1.9643221905626505e-05,
+ "loss": 0.9446,
+ "step": 4137
+ },
+ {
+ "epoch": 11.275204359673024,
+ "grad_norm": 8.788771629333496,
+ "learning_rate": 1.9642988244038122e-05,
+ "loss": 1.0808,
+ "step": 4138
+ },
+ {
+ "epoch": 11.277929155313352,
+ "grad_norm": 8.508222579956055,
+ "learning_rate": 1.9642754507350457e-05,
+ "loss": 1.0811,
+ "step": 4139
+ },
+ {
+ "epoch": 11.280653950953678,
+ "grad_norm": 7.484086036682129,
+ "learning_rate": 1.9642520695565328e-05,
+ "loss": 1.0334,
+ "step": 4140
+ },
+ {
+ "epoch": 11.283378746594005,
+ "grad_norm": 7.382587909698486,
+ "learning_rate": 1.9642286808684562e-05,
+ "loss": 1.2607,
+ "step": 4141
+ },
+ {
+ "epoch": 11.286103542234333,
+ "grad_norm": 6.845112323760986,
+ "learning_rate": 1.9642052846709975e-05,
+ "loss": 1.0083,
+ "step": 4142
+ },
+ {
+ "epoch": 11.288828337874659,
+ "grad_norm": 9.257438659667969,
+ "learning_rate": 1.964181880964339e-05,
+ "loss": 1.2485,
+ "step": 4143
+ },
+ {
+ "epoch": 11.291553133514986,
+ "grad_norm": 10.066730499267578,
+ "learning_rate": 1.9641584697486632e-05,
+ "loss": 1.0308,
+ "step": 4144
+ },
+ {
+ "epoch": 11.294277929155314,
+ "grad_norm": 10.6140775680542,
+ "learning_rate": 1.964135051024152e-05,
+ "loss": 1.2114,
+ "step": 4145
+ },
+ {
+ "epoch": 11.29700272479564,
+ "grad_norm": 6.769592761993408,
+ "learning_rate": 1.964111624790988e-05,
+ "loss": 1.0986,
+ "step": 4146
+ },
+ {
+ "epoch": 11.299727520435967,
+ "grad_norm": 8.386119842529297,
+ "learning_rate": 1.964088191049354e-05,
+ "loss": 1.1345,
+ "step": 4147
+ },
+ {
+ "epoch": 11.302452316076295,
+ "grad_norm": 5.869018077850342,
+ "learning_rate": 1.964064749799432e-05,
+ "loss": 1.0402,
+ "step": 4148
+ },
+ {
+ "epoch": 11.30517711171662,
+ "grad_norm": 6.0534539222717285,
+ "learning_rate": 1.9640413010414048e-05,
+ "loss": 0.9066,
+ "step": 4149
+ },
+ {
+ "epoch": 11.307901907356948,
+ "grad_norm": 6.562158584594727,
+ "learning_rate": 1.9640178447754554e-05,
+ "loss": 0.8562,
+ "step": 4150
+ },
+ {
+ "epoch": 11.310626702997276,
+ "grad_norm": 7.962159633636475,
+ "learning_rate": 1.9639943810017654e-05,
+ "loss": 1.1052,
+ "step": 4151
+ },
+ {
+ "epoch": 11.313351498637601,
+ "grad_norm": 7.923190116882324,
+ "learning_rate": 1.9639709097205185e-05,
+ "loss": 1.1571,
+ "step": 4152
+ },
+ {
+ "epoch": 11.316076294277929,
+ "grad_norm": 7.5282182693481445,
+ "learning_rate": 1.963947430931897e-05,
+ "loss": 1.2361,
+ "step": 4153
+ },
+ {
+ "epoch": 11.318801089918257,
+ "grad_norm": 18.881086349487305,
+ "learning_rate": 1.963923944636084e-05,
+ "loss": 1.1514,
+ "step": 4154
+ },
+ {
+ "epoch": 11.321525885558582,
+ "grad_norm": 7.609150409698486,
+ "learning_rate": 1.9639004508332627e-05,
+ "loss": 0.9963,
+ "step": 4155
+ },
+ {
+ "epoch": 11.32425068119891,
+ "grad_norm": 7.38583517074585,
+ "learning_rate": 1.9638769495236154e-05,
+ "loss": 1.1138,
+ "step": 4156
+ },
+ {
+ "epoch": 11.326975476839237,
+ "grad_norm": 6.654665470123291,
+ "learning_rate": 1.963853440707325e-05,
+ "loss": 1.21,
+ "step": 4157
+ },
+ {
+ "epoch": 11.329700272479563,
+ "grad_norm": 7.605174541473389,
+ "learning_rate": 1.9638299243845758e-05,
+ "loss": 1.1177,
+ "step": 4158
+ },
+ {
+ "epoch": 11.33242506811989,
+ "grad_norm": 8.497819900512695,
+ "learning_rate": 1.96380640055555e-05,
+ "loss": 0.9911,
+ "step": 4159
+ },
+ {
+ "epoch": 11.335149863760218,
+ "grad_norm": 8.258249282836914,
+ "learning_rate": 1.9637828692204307e-05,
+ "loss": 0.9546,
+ "step": 4160
+ },
+ {
+ "epoch": 11.337874659400544,
+ "grad_norm": 7.528987407684326,
+ "learning_rate": 1.9637593303794015e-05,
+ "loss": 1.2327,
+ "step": 4161
+ },
+ {
+ "epoch": 11.340599455040872,
+ "grad_norm": 6.577043533325195,
+ "learning_rate": 1.9637357840326457e-05,
+ "loss": 0.9558,
+ "step": 4162
+ },
+ {
+ "epoch": 11.3433242506812,
+ "grad_norm": 25.01720428466797,
+ "learning_rate": 1.9637122301803466e-05,
+ "loss": 0.9707,
+ "step": 4163
+ },
+ {
+ "epoch": 11.346049046321525,
+ "grad_norm": 7.703424453735352,
+ "learning_rate": 1.963688668822688e-05,
+ "loss": 0.8982,
+ "step": 4164
+ },
+ {
+ "epoch": 11.348773841961853,
+ "grad_norm": 8.75214958190918,
+ "learning_rate": 1.9636650999598527e-05,
+ "loss": 1.0118,
+ "step": 4165
+ },
+ {
+ "epoch": 11.35149863760218,
+ "grad_norm": 6.6170501708984375,
+ "learning_rate": 1.9636415235920246e-05,
+ "loss": 1.0789,
+ "step": 4166
+ },
+ {
+ "epoch": 11.354223433242506,
+ "grad_norm": 7.466395854949951,
+ "learning_rate": 1.9636179397193874e-05,
+ "loss": 1.179,
+ "step": 4167
+ },
+ {
+ "epoch": 11.356948228882834,
+ "grad_norm": 6.15164041519165,
+ "learning_rate": 1.9635943483421246e-05,
+ "loss": 1.0601,
+ "step": 4168
+ },
+ {
+ "epoch": 11.359673024523161,
+ "grad_norm": 7.586962699890137,
+ "learning_rate": 1.9635707494604203e-05,
+ "loss": 0.9155,
+ "step": 4169
+ },
+ {
+ "epoch": 11.362397820163487,
+ "grad_norm": 8.211374282836914,
+ "learning_rate": 1.9635471430744578e-05,
+ "loss": 1.0671,
+ "step": 4170
+ },
+ {
+ "epoch": 11.365122615803815,
+ "grad_norm": 6.054488182067871,
+ "learning_rate": 1.9635235291844213e-05,
+ "loss": 1.116,
+ "step": 4171
+ },
+ {
+ "epoch": 11.367847411444142,
+ "grad_norm": 9.64517593383789,
+ "learning_rate": 1.9634999077904945e-05,
+ "loss": 1.1882,
+ "step": 4172
+ },
+ {
+ "epoch": 11.370572207084468,
+ "grad_norm": 6.744877338409424,
+ "learning_rate": 1.9634762788928615e-05,
+ "loss": 1.1185,
+ "step": 4173
+ },
+ {
+ "epoch": 11.373297002724795,
+ "grad_norm": 7.894425868988037,
+ "learning_rate": 1.963452642491706e-05,
+ "loss": 0.9531,
+ "step": 4174
+ },
+ {
+ "epoch": 11.376021798365123,
+ "grad_norm": 7.983966827392578,
+ "learning_rate": 1.9634289985872126e-05,
+ "loss": 1.0096,
+ "step": 4175
+ },
+ {
+ "epoch": 11.378746594005449,
+ "grad_norm": 7.138466835021973,
+ "learning_rate": 1.963405347179565e-05,
+ "loss": 1.1764,
+ "step": 4176
+ },
+ {
+ "epoch": 11.381471389645776,
+ "grad_norm": 6.448553085327148,
+ "learning_rate": 1.963381688268948e-05,
+ "loss": 0.9452,
+ "step": 4177
+ },
+ {
+ "epoch": 11.384196185286104,
+ "grad_norm": 8.217049598693848,
+ "learning_rate": 1.963358021855545e-05,
+ "loss": 1.1299,
+ "step": 4178
+ },
+ {
+ "epoch": 11.38692098092643,
+ "grad_norm": 7.03897762298584,
+ "learning_rate": 1.9633343479395405e-05,
+ "loss": 1.1641,
+ "step": 4179
+ },
+ {
+ "epoch": 11.389645776566757,
+ "grad_norm": 7.485842227935791,
+ "learning_rate": 1.9633106665211196e-05,
+ "loss": 1.0092,
+ "step": 4180
+ },
+ {
+ "epoch": 11.392370572207085,
+ "grad_norm": 7.845271110534668,
+ "learning_rate": 1.963286977600466e-05,
+ "loss": 1.151,
+ "step": 4181
+ },
+ {
+ "epoch": 11.39509536784741,
+ "grad_norm": 7.305263519287109,
+ "learning_rate": 1.9632632811777644e-05,
+ "loss": 1.3477,
+ "step": 4182
+ },
+ {
+ "epoch": 11.397820163487738,
+ "grad_norm": 7.606513500213623,
+ "learning_rate": 1.9632395772531993e-05,
+ "loss": 1.1609,
+ "step": 4183
+ },
+ {
+ "epoch": 11.400544959128066,
+ "grad_norm": 8.449793815612793,
+ "learning_rate": 1.9632158658269554e-05,
+ "loss": 1.1174,
+ "step": 4184
+ },
+ {
+ "epoch": 11.403269754768392,
+ "grad_norm": 8.331226348876953,
+ "learning_rate": 1.9631921468992173e-05,
+ "loss": 0.9598,
+ "step": 4185
+ },
+ {
+ "epoch": 11.40599455040872,
+ "grad_norm": 7.322021961212158,
+ "learning_rate": 1.96316842047017e-05,
+ "loss": 1.0874,
+ "step": 4186
+ },
+ {
+ "epoch": 11.408719346049047,
+ "grad_norm": 6.666060447692871,
+ "learning_rate": 1.963144686539998e-05,
+ "loss": 0.9089,
+ "step": 4187
+ },
+ {
+ "epoch": 11.411444141689373,
+ "grad_norm": 14.103044509887695,
+ "learning_rate": 1.963120945108886e-05,
+ "loss": 0.9418,
+ "step": 4188
+ },
+ {
+ "epoch": 11.4141689373297,
+ "grad_norm": 8.893685340881348,
+ "learning_rate": 1.9630971961770194e-05,
+ "loss": 1.08,
+ "step": 4189
+ },
+ {
+ "epoch": 11.416893732970028,
+ "grad_norm": 6.879014492034912,
+ "learning_rate": 1.9630734397445824e-05,
+ "loss": 1.1863,
+ "step": 4190
+ },
+ {
+ "epoch": 11.419618528610354,
+ "grad_norm": 9.25615119934082,
+ "learning_rate": 1.963049675811761e-05,
+ "loss": 1.1143,
+ "step": 4191
+ },
+ {
+ "epoch": 11.422343324250681,
+ "grad_norm": 7.276339530944824,
+ "learning_rate": 1.963025904378739e-05,
+ "loss": 0.9471,
+ "step": 4192
+ },
+ {
+ "epoch": 11.425068119891009,
+ "grad_norm": 7.350114822387695,
+ "learning_rate": 1.9630021254457026e-05,
+ "loss": 1.1475,
+ "step": 4193
+ },
+ {
+ "epoch": 11.427792915531334,
+ "grad_norm": 9.32851505279541,
+ "learning_rate": 1.9629783390128366e-05,
+ "loss": 1.3047,
+ "step": 4194
+ },
+ {
+ "epoch": 11.430517711171662,
+ "grad_norm": 8.629213333129883,
+ "learning_rate": 1.9629545450803266e-05,
+ "loss": 1.1572,
+ "step": 4195
+ },
+ {
+ "epoch": 11.43324250681199,
+ "grad_norm": 6.012125015258789,
+ "learning_rate": 1.9629307436483573e-05,
+ "loss": 0.9198,
+ "step": 4196
+ },
+ {
+ "epoch": 11.435967302452315,
+ "grad_norm": 6.589315414428711,
+ "learning_rate": 1.9629069347171144e-05,
+ "loss": 1.3062,
+ "step": 4197
+ },
+ {
+ "epoch": 11.438692098092643,
+ "grad_norm": 6.8728108406066895,
+ "learning_rate": 1.9628831182867833e-05,
+ "loss": 1.1765,
+ "step": 4198
+ },
+ {
+ "epoch": 11.44141689373297,
+ "grad_norm": 6.035057544708252,
+ "learning_rate": 1.9628592943575494e-05,
+ "loss": 1.0212,
+ "step": 4199
+ },
+ {
+ "epoch": 11.444141689373296,
+ "grad_norm": 6.884873867034912,
+ "learning_rate": 1.9628354629295983e-05,
+ "loss": 1.1331,
+ "step": 4200
+ },
+ {
+ "epoch": 11.446866485013624,
+ "grad_norm": 6.8663177490234375,
+ "learning_rate": 1.962811624003116e-05,
+ "loss": 1.116,
+ "step": 4201
+ },
+ {
+ "epoch": 11.449591280653951,
+ "grad_norm": 6.953235149383545,
+ "learning_rate": 1.9627877775782873e-05,
+ "loss": 1.0096,
+ "step": 4202
+ },
+ {
+ "epoch": 11.452316076294277,
+ "grad_norm": 7.713603973388672,
+ "learning_rate": 1.9627639236552983e-05,
+ "loss": 1.1882,
+ "step": 4203
+ },
+ {
+ "epoch": 11.455040871934605,
+ "grad_norm": 7.972574710845947,
+ "learning_rate": 1.9627400622343354e-05,
+ "loss": 1.1194,
+ "step": 4204
+ },
+ {
+ "epoch": 11.457765667574932,
+ "grad_norm": 8.037703514099121,
+ "learning_rate": 1.9627161933155833e-05,
+ "loss": 1.1724,
+ "step": 4205
+ },
+ {
+ "epoch": 11.460490463215258,
+ "grad_norm": 7.492434501647949,
+ "learning_rate": 1.962692316899229e-05,
+ "loss": 0.9735,
+ "step": 4206
+ },
+ {
+ "epoch": 11.463215258855586,
+ "grad_norm": 8.838690757751465,
+ "learning_rate": 1.9626684329854575e-05,
+ "loss": 0.9778,
+ "step": 4207
+ },
+ {
+ "epoch": 11.465940054495913,
+ "grad_norm": 7.17144250869751,
+ "learning_rate": 1.9626445415744552e-05,
+ "loss": 0.8794,
+ "step": 4208
+ },
+ {
+ "epoch": 11.46866485013624,
+ "grad_norm": 6.494210720062256,
+ "learning_rate": 1.9626206426664084e-05,
+ "loss": 0.9868,
+ "step": 4209
+ },
+ {
+ "epoch": 11.471389645776567,
+ "grad_norm": 6.2275390625,
+ "learning_rate": 1.962596736261503e-05,
+ "loss": 0.9924,
+ "step": 4210
+ },
+ {
+ "epoch": 11.474114441416894,
+ "grad_norm": 8.345746994018555,
+ "learning_rate": 1.9625728223599253e-05,
+ "loss": 1.129,
+ "step": 4211
+ },
+ {
+ "epoch": 11.47683923705722,
+ "grad_norm": 7.962212085723877,
+ "learning_rate": 1.962548900961861e-05,
+ "loss": 0.9542,
+ "step": 4212
+ },
+ {
+ "epoch": 11.479564032697548,
+ "grad_norm": 6.464784622192383,
+ "learning_rate": 1.9625249720674975e-05,
+ "loss": 1.2203,
+ "step": 4213
+ },
+ {
+ "epoch": 11.482288828337875,
+ "grad_norm": 7.042607307434082,
+ "learning_rate": 1.96250103567702e-05,
+ "loss": 1.2759,
+ "step": 4214
+ },
+ {
+ "epoch": 11.485013623978201,
+ "grad_norm": 10.316446304321289,
+ "learning_rate": 1.9624770917906156e-05,
+ "loss": 0.8961,
+ "step": 4215
+ },
+ {
+ "epoch": 11.487738419618529,
+ "grad_norm": 7.282829284667969,
+ "learning_rate": 1.9624531404084704e-05,
+ "loss": 1.1144,
+ "step": 4216
+ },
+ {
+ "epoch": 11.490463215258856,
+ "grad_norm": 9.247491836547852,
+ "learning_rate": 1.962429181530771e-05,
+ "loss": 1.1284,
+ "step": 4217
+ },
+ {
+ "epoch": 11.493188010899182,
+ "grad_norm": 6.852824687957764,
+ "learning_rate": 1.9624052151577047e-05,
+ "loss": 0.8964,
+ "step": 4218
+ },
+ {
+ "epoch": 11.49591280653951,
+ "grad_norm": 7.450499534606934,
+ "learning_rate": 1.962381241289457e-05,
+ "loss": 1.2124,
+ "step": 4219
+ },
+ {
+ "epoch": 11.498637602179837,
+ "grad_norm": 10.606607437133789,
+ "learning_rate": 1.9623572599262156e-05,
+ "loss": 1.1562,
+ "step": 4220
+ },
+ {
+ "epoch": 11.501362397820163,
+ "grad_norm": 8.954209327697754,
+ "learning_rate": 1.9623332710681666e-05,
+ "loss": 1.0,
+ "step": 4221
+ },
+ {
+ "epoch": 11.50408719346049,
+ "grad_norm": 6.404054164886475,
+ "learning_rate": 1.962309274715497e-05,
+ "loss": 0.9752,
+ "step": 4222
+ },
+ {
+ "epoch": 11.506811989100818,
+ "grad_norm": 7.652896881103516,
+ "learning_rate": 1.962285270868394e-05,
+ "loss": 1.1272,
+ "step": 4223
+ },
+ {
+ "epoch": 11.509536784741144,
+ "grad_norm": 7.61530065536499,
+ "learning_rate": 1.962261259527044e-05,
+ "loss": 1.0515,
+ "step": 4224
+ },
+ {
+ "epoch": 11.512261580381471,
+ "grad_norm": 8.351517677307129,
+ "learning_rate": 1.9622372406916343e-05,
+ "loss": 0.9734,
+ "step": 4225
+ },
+ {
+ "epoch": 11.514986376021799,
+ "grad_norm": 7.562206268310547,
+ "learning_rate": 1.9622132143623524e-05,
+ "loss": 1.0869,
+ "step": 4226
+ },
+ {
+ "epoch": 11.517711171662125,
+ "grad_norm": 8.741758346557617,
+ "learning_rate": 1.9621891805393846e-05,
+ "loss": 1.0742,
+ "step": 4227
+ },
+ {
+ "epoch": 11.520435967302452,
+ "grad_norm": 6.766845703125,
+ "learning_rate": 1.9621651392229182e-05,
+ "loss": 1.2001,
+ "step": 4228
+ },
+ {
+ "epoch": 11.52316076294278,
+ "grad_norm": 6.957767009735107,
+ "learning_rate": 1.9621410904131407e-05,
+ "loss": 1.1196,
+ "step": 4229
+ },
+ {
+ "epoch": 11.525885558583106,
+ "grad_norm": 7.6866350173950195,
+ "learning_rate": 1.9621170341102396e-05,
+ "loss": 1.26,
+ "step": 4230
+ },
+ {
+ "epoch": 11.528610354223433,
+ "grad_norm": 9.070810317993164,
+ "learning_rate": 1.962092970314402e-05,
+ "loss": 1.1958,
+ "step": 4231
+ },
+ {
+ "epoch": 11.53133514986376,
+ "grad_norm": 6.253953456878662,
+ "learning_rate": 1.9620688990258152e-05,
+ "loss": 1.1799,
+ "step": 4232
+ },
+ {
+ "epoch": 11.534059945504087,
+ "grad_norm": 7.1140594482421875,
+ "learning_rate": 1.962044820244667e-05,
+ "loss": 0.9934,
+ "step": 4233
+ },
+ {
+ "epoch": 11.536784741144414,
+ "grad_norm": 25.949594497680664,
+ "learning_rate": 1.9620207339711442e-05,
+ "loss": 0.981,
+ "step": 4234
+ },
+ {
+ "epoch": 11.539509536784742,
+ "grad_norm": 6.960265636444092,
+ "learning_rate": 1.9619966402054355e-05,
+ "loss": 1.1719,
+ "step": 4235
+ },
+ {
+ "epoch": 11.542234332425068,
+ "grad_norm": 8.840120315551758,
+ "learning_rate": 1.9619725389477275e-05,
+ "loss": 1.0884,
+ "step": 4236
+ },
+ {
+ "epoch": 11.544959128065395,
+ "grad_norm": 6.605489253997803,
+ "learning_rate": 1.9619484301982086e-05,
+ "loss": 1.0811,
+ "step": 4237
+ },
+ {
+ "epoch": 11.547683923705723,
+ "grad_norm": 5.886390686035156,
+ "learning_rate": 1.961924313957066e-05,
+ "loss": 1.213,
+ "step": 4238
+ },
+ {
+ "epoch": 11.550408719346049,
+ "grad_norm": 8.48935604095459,
+ "learning_rate": 1.961900190224488e-05,
+ "loss": 1.2811,
+ "step": 4239
+ },
+ {
+ "epoch": 11.553133514986376,
+ "grad_norm": 10.251641273498535,
+ "learning_rate": 1.9618760590006623e-05,
+ "loss": 0.9816,
+ "step": 4240
+ },
+ {
+ "epoch": 11.555858310626704,
+ "grad_norm": 7.327285289764404,
+ "learning_rate": 1.9618519202857768e-05,
+ "loss": 1.2625,
+ "step": 4241
+ },
+ {
+ "epoch": 11.55858310626703,
+ "grad_norm": 6.994567394256592,
+ "learning_rate": 1.9618277740800195e-05,
+ "loss": 0.9119,
+ "step": 4242
+ },
+ {
+ "epoch": 11.561307901907357,
+ "grad_norm": 7.749634265899658,
+ "learning_rate": 1.961803620383578e-05,
+ "loss": 1.1088,
+ "step": 4243
+ },
+ {
+ "epoch": 11.564032697547685,
+ "grad_norm": 6.253412246704102,
+ "learning_rate": 1.9617794591966418e-05,
+ "loss": 0.9851,
+ "step": 4244
+ },
+ {
+ "epoch": 11.56675749318801,
+ "grad_norm": 7.237356662750244,
+ "learning_rate": 1.9617552905193972e-05,
+ "loss": 1.2299,
+ "step": 4245
+ },
+ {
+ "epoch": 11.569482288828338,
+ "grad_norm": 6.221981048583984,
+ "learning_rate": 1.961731114352034e-05,
+ "loss": 1.1929,
+ "step": 4246
+ },
+ {
+ "epoch": 11.572207084468666,
+ "grad_norm": 7.852269649505615,
+ "learning_rate": 1.9617069306947394e-05,
+ "loss": 1.0908,
+ "step": 4247
+ },
+ {
+ "epoch": 11.574931880108991,
+ "grad_norm": 6.650592803955078,
+ "learning_rate": 1.9616827395477026e-05,
+ "loss": 0.9221,
+ "step": 4248
+ },
+ {
+ "epoch": 11.577656675749319,
+ "grad_norm": 11.2562255859375,
+ "learning_rate": 1.961658540911111e-05,
+ "loss": 1.166,
+ "step": 4249
+ },
+ {
+ "epoch": 11.580381471389646,
+ "grad_norm": 8.52875804901123,
+ "learning_rate": 1.961634334785154e-05,
+ "loss": 1.1914,
+ "step": 4250
+ },
+ {
+ "epoch": 11.583106267029972,
+ "grad_norm": 7.987171649932861,
+ "learning_rate": 1.9616101211700198e-05,
+ "loss": 1.027,
+ "step": 4251
+ },
+ {
+ "epoch": 11.5858310626703,
+ "grad_norm": 7.115975379943848,
+ "learning_rate": 1.961585900065897e-05,
+ "loss": 1.1348,
+ "step": 4252
+ },
+ {
+ "epoch": 11.588555858310627,
+ "grad_norm": 7.772615432739258,
+ "learning_rate": 1.961561671472974e-05,
+ "loss": 1.1953,
+ "step": 4253
+ },
+ {
+ "epoch": 11.591280653950953,
+ "grad_norm": 7.0153326988220215,
+ "learning_rate": 1.9615374353914395e-05,
+ "loss": 1.2195,
+ "step": 4254
+ },
+ {
+ "epoch": 11.59400544959128,
+ "grad_norm": 7.245928764343262,
+ "learning_rate": 1.9615131918214824e-05,
+ "loss": 1.2666,
+ "step": 4255
+ },
+ {
+ "epoch": 11.596730245231608,
+ "grad_norm": 8.815357208251953,
+ "learning_rate": 1.961488940763292e-05,
+ "loss": 1.0857,
+ "step": 4256
+ },
+ {
+ "epoch": 11.599455040871934,
+ "grad_norm": 9.163494110107422,
+ "learning_rate": 1.961464682217056e-05,
+ "loss": 1.1953,
+ "step": 4257
+ },
+ {
+ "epoch": 11.602179836512262,
+ "grad_norm": 9.221513748168945,
+ "learning_rate": 1.9614404161829643e-05,
+ "loss": 1.1794,
+ "step": 4258
+ },
+ {
+ "epoch": 11.60490463215259,
+ "grad_norm": 7.49953031539917,
+ "learning_rate": 1.9614161426612056e-05,
+ "loss": 1.0183,
+ "step": 4259
+ },
+ {
+ "epoch": 11.607629427792915,
+ "grad_norm": 10.897595405578613,
+ "learning_rate": 1.9613918616519688e-05,
+ "loss": 1.0793,
+ "step": 4260
+ },
+ {
+ "epoch": 11.610354223433243,
+ "grad_norm": 11.387365341186523,
+ "learning_rate": 1.9613675731554435e-05,
+ "loss": 1.0389,
+ "step": 4261
+ },
+ {
+ "epoch": 11.61307901907357,
+ "grad_norm": 10.990483283996582,
+ "learning_rate": 1.961343277171818e-05,
+ "loss": 1.1741,
+ "step": 4262
+ },
+ {
+ "epoch": 11.615803814713896,
+ "grad_norm": 7.207435131072998,
+ "learning_rate": 1.9613189737012824e-05,
+ "loss": 0.9458,
+ "step": 4263
+ },
+ {
+ "epoch": 11.618528610354224,
+ "grad_norm": 6.2447309494018555,
+ "learning_rate": 1.9612946627440254e-05,
+ "loss": 1.0291,
+ "step": 4264
+ },
+ {
+ "epoch": 11.621253405994551,
+ "grad_norm": 10.74925422668457,
+ "learning_rate": 1.9612703443002366e-05,
+ "loss": 1.2764,
+ "step": 4265
+ },
+ {
+ "epoch": 11.623978201634877,
+ "grad_norm": 10.540519714355469,
+ "learning_rate": 1.9612460183701053e-05,
+ "loss": 0.9601,
+ "step": 4266
+ },
+ {
+ "epoch": 11.626702997275205,
+ "grad_norm": 6.874648571014404,
+ "learning_rate": 1.9612216849538205e-05,
+ "loss": 0.9436,
+ "step": 4267
+ },
+ {
+ "epoch": 11.629427792915532,
+ "grad_norm": 9.658646583557129,
+ "learning_rate": 1.9611973440515724e-05,
+ "loss": 1.1855,
+ "step": 4268
+ },
+ {
+ "epoch": 11.632152588555858,
+ "grad_norm": 7.239814281463623,
+ "learning_rate": 1.9611729956635507e-05,
+ "loss": 0.9449,
+ "step": 4269
+ },
+ {
+ "epoch": 11.634877384196185,
+ "grad_norm": 9.3568115234375,
+ "learning_rate": 1.9611486397899445e-05,
+ "loss": 1.1553,
+ "step": 4270
+ },
+ {
+ "epoch": 11.637602179836513,
+ "grad_norm": 6.999940395355225,
+ "learning_rate": 1.9611242764309434e-05,
+ "loss": 1.2444,
+ "step": 4271
+ },
+ {
+ "epoch": 11.640326975476839,
+ "grad_norm": 7.168976306915283,
+ "learning_rate": 1.9610999055867376e-05,
+ "loss": 1.0953,
+ "step": 4272
+ },
+ {
+ "epoch": 11.643051771117166,
+ "grad_norm": 6.548735618591309,
+ "learning_rate": 1.9610755272575167e-05,
+ "loss": 1.1986,
+ "step": 4273
+ },
+ {
+ "epoch": 11.645776566757494,
+ "grad_norm": 10.056276321411133,
+ "learning_rate": 1.96105114144347e-05,
+ "loss": 0.9729,
+ "step": 4274
+ },
+ {
+ "epoch": 11.64850136239782,
+ "grad_norm": 7.083216667175293,
+ "learning_rate": 1.9610267481447883e-05,
+ "loss": 1.1777,
+ "step": 4275
+ },
+ {
+ "epoch": 11.651226158038147,
+ "grad_norm": 5.7767863273620605,
+ "learning_rate": 1.9610023473616616e-05,
+ "loss": 0.9896,
+ "step": 4276
+ },
+ {
+ "epoch": 11.653950953678475,
+ "grad_norm": 7.62014102935791,
+ "learning_rate": 1.960977939094279e-05,
+ "loss": 1.2051,
+ "step": 4277
+ },
+ {
+ "epoch": 11.6566757493188,
+ "grad_norm": 8.748889923095703,
+ "learning_rate": 1.9609535233428313e-05,
+ "loss": 1.1377,
+ "step": 4278
+ },
+ {
+ "epoch": 11.659400544959128,
+ "grad_norm": 8.948205947875977,
+ "learning_rate": 1.9609291001075082e-05,
+ "loss": 1.2507,
+ "step": 4279
+ },
+ {
+ "epoch": 11.662125340599456,
+ "grad_norm": 9.956958770751953,
+ "learning_rate": 1.9609046693885007e-05,
+ "loss": 1.1399,
+ "step": 4280
+ },
+ {
+ "epoch": 11.664850136239782,
+ "grad_norm": 6.880049705505371,
+ "learning_rate": 1.9608802311859984e-05,
+ "loss": 1.1814,
+ "step": 4281
+ },
+ {
+ "epoch": 11.66757493188011,
+ "grad_norm": 11.572159767150879,
+ "learning_rate": 1.9608557855001918e-05,
+ "loss": 1.1292,
+ "step": 4282
+ },
+ {
+ "epoch": 11.670299727520437,
+ "grad_norm": 6.775529384613037,
+ "learning_rate": 1.9608313323312712e-05,
+ "loss": 1.1223,
+ "step": 4283
+ },
+ {
+ "epoch": 11.673024523160763,
+ "grad_norm": 9.93600082397461,
+ "learning_rate": 1.960806871679427e-05,
+ "loss": 1.0342,
+ "step": 4284
+ },
+ {
+ "epoch": 11.67574931880109,
+ "grad_norm": 6.776366233825684,
+ "learning_rate": 1.96078240354485e-05,
+ "loss": 0.9503,
+ "step": 4285
+ },
+ {
+ "epoch": 11.678474114441418,
+ "grad_norm": 6.5114054679870605,
+ "learning_rate": 1.9607579279277304e-05,
+ "loss": 0.9771,
+ "step": 4286
+ },
+ {
+ "epoch": 11.681198910081743,
+ "grad_norm": 7.798119068145752,
+ "learning_rate": 1.960733444828259e-05,
+ "loss": 1.2966,
+ "step": 4287
+ },
+ {
+ "epoch": 11.683923705722071,
+ "grad_norm": 6.28831148147583,
+ "learning_rate": 1.9607089542466266e-05,
+ "loss": 1.0802,
+ "step": 4288
+ },
+ {
+ "epoch": 11.686648501362399,
+ "grad_norm": 7.590810298919678,
+ "learning_rate": 1.960684456183024e-05,
+ "loss": 1.1726,
+ "step": 4289
+ },
+ {
+ "epoch": 11.689373297002724,
+ "grad_norm": 6.139025688171387,
+ "learning_rate": 1.9606599506376415e-05,
+ "loss": 0.9763,
+ "step": 4290
+ },
+ {
+ "epoch": 11.692098092643052,
+ "grad_norm": 6.597779750823975,
+ "learning_rate": 1.96063543761067e-05,
+ "loss": 1.0352,
+ "step": 4291
+ },
+ {
+ "epoch": 11.69482288828338,
+ "grad_norm": 7.056490421295166,
+ "learning_rate": 1.960610917102301e-05,
+ "loss": 1.1475,
+ "step": 4292
+ },
+ {
+ "epoch": 11.697547683923705,
+ "grad_norm": 7.991150856018066,
+ "learning_rate": 1.9605863891127247e-05,
+ "loss": 1.0476,
+ "step": 4293
+ },
+ {
+ "epoch": 11.700272479564033,
+ "grad_norm": 6.341061115264893,
+ "learning_rate": 1.960561853642133e-05,
+ "loss": 1.0273,
+ "step": 4294
+ },
+ {
+ "epoch": 11.70299727520436,
+ "grad_norm": 6.297184467315674,
+ "learning_rate": 1.960537310690716e-05,
+ "loss": 0.8853,
+ "step": 4295
+ },
+ {
+ "epoch": 11.705722070844686,
+ "grad_norm": 7.434573650360107,
+ "learning_rate": 1.9605127602586656e-05,
+ "loss": 1.142,
+ "step": 4296
+ },
+ {
+ "epoch": 11.708446866485014,
+ "grad_norm": 8.553709983825684,
+ "learning_rate": 1.9604882023461725e-05,
+ "loss": 1.217,
+ "step": 4297
+ },
+ {
+ "epoch": 11.711171662125341,
+ "grad_norm": 7.114347457885742,
+ "learning_rate": 1.9604636369534285e-05,
+ "loss": 1.052,
+ "step": 4298
+ },
+ {
+ "epoch": 11.713896457765667,
+ "grad_norm": 8.586565971374512,
+ "learning_rate": 1.9604390640806246e-05,
+ "loss": 1.1025,
+ "step": 4299
+ },
+ {
+ "epoch": 11.716621253405995,
+ "grad_norm": 6.76774787902832,
+ "learning_rate": 1.9604144837279523e-05,
+ "loss": 1.0201,
+ "step": 4300
+ },
+ {
+ "epoch": 11.719346049046322,
+ "grad_norm": 11.73022174835205,
+ "learning_rate": 1.9603898958956024e-05,
+ "loss": 1.0005,
+ "step": 4301
+ },
+ {
+ "epoch": 11.722070844686648,
+ "grad_norm": 7.182959079742432,
+ "learning_rate": 1.960365300583767e-05,
+ "loss": 0.9047,
+ "step": 4302
+ },
+ {
+ "epoch": 11.724795640326976,
+ "grad_norm": 7.677026271820068,
+ "learning_rate": 1.960340697792638e-05,
+ "loss": 1.037,
+ "step": 4303
+ },
+ {
+ "epoch": 11.727520435967303,
+ "grad_norm": 6.79896354675293,
+ "learning_rate": 1.9603160875224067e-05,
+ "loss": 1.2573,
+ "step": 4304
+ },
+ {
+ "epoch": 11.730245231607629,
+ "grad_norm": 8.253938674926758,
+ "learning_rate": 1.960291469773264e-05,
+ "loss": 1.1111,
+ "step": 4305
+ },
+ {
+ "epoch": 11.732970027247957,
+ "grad_norm": 13.566267013549805,
+ "learning_rate": 1.9602668445454024e-05,
+ "loss": 1.0074,
+ "step": 4306
+ },
+ {
+ "epoch": 11.735694822888284,
+ "grad_norm": 6.529451847076416,
+ "learning_rate": 1.9602422118390138e-05,
+ "loss": 1.1641,
+ "step": 4307
+ },
+ {
+ "epoch": 11.73841961852861,
+ "grad_norm": 6.8001179695129395,
+ "learning_rate": 1.9602175716542895e-05,
+ "loss": 0.9596,
+ "step": 4308
+ },
+ {
+ "epoch": 11.741144414168938,
+ "grad_norm": 6.728170871734619,
+ "learning_rate": 1.9601929239914217e-05,
+ "loss": 1.1416,
+ "step": 4309
+ },
+ {
+ "epoch": 11.743869209809265,
+ "grad_norm": 6.7742509841918945,
+ "learning_rate": 1.960168268850602e-05,
+ "loss": 1.0455,
+ "step": 4310
+ },
+ {
+ "epoch": 11.746594005449591,
+ "grad_norm": 8.66490364074707,
+ "learning_rate": 1.9601436062320232e-05,
+ "loss": 1.1746,
+ "step": 4311
+ },
+ {
+ "epoch": 11.749318801089919,
+ "grad_norm": 6.436126708984375,
+ "learning_rate": 1.960118936135877e-05,
+ "loss": 1.05,
+ "step": 4312
+ },
+ {
+ "epoch": 11.752043596730246,
+ "grad_norm": 10.69386100769043,
+ "learning_rate": 1.960094258562355e-05,
+ "loss": 1.2749,
+ "step": 4313
+ },
+ {
+ "epoch": 11.754768392370572,
+ "grad_norm": 7.94373893737793,
+ "learning_rate": 1.9600695735116503e-05,
+ "loss": 1.0645,
+ "step": 4314
+ },
+ {
+ "epoch": 11.7574931880109,
+ "grad_norm": 8.081480979919434,
+ "learning_rate": 1.9600448809839542e-05,
+ "loss": 1.2273,
+ "step": 4315
+ },
+ {
+ "epoch": 11.760217983651227,
+ "grad_norm": 9.528131484985352,
+ "learning_rate": 1.9600201809794596e-05,
+ "loss": 1.1052,
+ "step": 4316
+ },
+ {
+ "epoch": 11.762942779291553,
+ "grad_norm": 5.941060543060303,
+ "learning_rate": 1.9599954734983586e-05,
+ "loss": 1.1348,
+ "step": 4317
+ },
+ {
+ "epoch": 11.76566757493188,
+ "grad_norm": 7.506882667541504,
+ "learning_rate": 1.9599707585408438e-05,
+ "loss": 1.1407,
+ "step": 4318
+ },
+ {
+ "epoch": 11.768392370572208,
+ "grad_norm": 7.964119911193848,
+ "learning_rate": 1.959946036107108e-05,
+ "loss": 1.218,
+ "step": 4319
+ },
+ {
+ "epoch": 11.771117166212534,
+ "grad_norm": 6.277833461761475,
+ "learning_rate": 1.959921306197343e-05,
+ "loss": 1.0358,
+ "step": 4320
+ },
+ {
+ "epoch": 11.773841961852861,
+ "grad_norm": 7.132114887237549,
+ "learning_rate": 1.959896568811742e-05,
+ "loss": 1.0387,
+ "step": 4321
+ },
+ {
+ "epoch": 11.776566757493189,
+ "grad_norm": 10.367997169494629,
+ "learning_rate": 1.959871823950497e-05,
+ "loss": 1.0356,
+ "step": 4322
+ },
+ {
+ "epoch": 11.779291553133515,
+ "grad_norm": 7.236018180847168,
+ "learning_rate": 1.9598470716138015e-05,
+ "loss": 1.2275,
+ "step": 4323
+ },
+ {
+ "epoch": 11.782016348773842,
+ "grad_norm": 6.842081069946289,
+ "learning_rate": 1.9598223118018477e-05,
+ "loss": 1.1635,
+ "step": 4324
+ },
+ {
+ "epoch": 11.78474114441417,
+ "grad_norm": 6.063444137573242,
+ "learning_rate": 1.959797544514829e-05,
+ "loss": 1.0771,
+ "step": 4325
+ },
+ {
+ "epoch": 11.787465940054496,
+ "grad_norm": 7.95065450668335,
+ "learning_rate": 1.9597727697529373e-05,
+ "loss": 1.0715,
+ "step": 4326
+ },
+ {
+ "epoch": 11.790190735694823,
+ "grad_norm": 8.508203506469727,
+ "learning_rate": 1.9597479875163666e-05,
+ "loss": 1.0999,
+ "step": 4327
+ },
+ {
+ "epoch": 11.79291553133515,
+ "grad_norm": 6.932796001434326,
+ "learning_rate": 1.9597231978053095e-05,
+ "loss": 0.936,
+ "step": 4328
+ },
+ {
+ "epoch": 11.795640326975477,
+ "grad_norm": 7.1728105545043945,
+ "learning_rate": 1.9596984006199585e-05,
+ "loss": 1.2188,
+ "step": 4329
+ },
+ {
+ "epoch": 11.798365122615804,
+ "grad_norm": 7.652414321899414,
+ "learning_rate": 1.9596735959605077e-05,
+ "loss": 1.111,
+ "step": 4330
+ },
+ {
+ "epoch": 11.80108991825613,
+ "grad_norm": 6.13052225112915,
+ "learning_rate": 1.95964878382715e-05,
+ "loss": 1.004,
+ "step": 4331
+ },
+ {
+ "epoch": 11.803814713896458,
+ "grad_norm": 7.247925758361816,
+ "learning_rate": 1.9596239642200784e-05,
+ "loss": 1.3035,
+ "step": 4332
+ },
+ {
+ "epoch": 11.806539509536785,
+ "grad_norm": 7.495167255401611,
+ "learning_rate": 1.959599137139486e-05,
+ "loss": 1.0316,
+ "step": 4333
+ },
+ {
+ "epoch": 11.809264305177111,
+ "grad_norm": 7.758265018463135,
+ "learning_rate": 1.959574302585566e-05,
+ "loss": 1.0986,
+ "step": 4334
+ },
+ {
+ "epoch": 11.811989100817438,
+ "grad_norm": 7.082614421844482,
+ "learning_rate": 1.9595494605585127e-05,
+ "loss": 1.0696,
+ "step": 4335
+ },
+ {
+ "epoch": 11.814713896457766,
+ "grad_norm": 8.044011116027832,
+ "learning_rate": 1.9595246110585194e-05,
+ "loss": 1.1687,
+ "step": 4336
+ },
+ {
+ "epoch": 11.817438692098092,
+ "grad_norm": 7.690464973449707,
+ "learning_rate": 1.959499754085779e-05,
+ "loss": 1.1138,
+ "step": 4337
+ },
+ {
+ "epoch": 11.82016348773842,
+ "grad_norm": 6.13950777053833,
+ "learning_rate": 1.9594748896404853e-05,
+ "loss": 1.0403,
+ "step": 4338
+ },
+ {
+ "epoch": 11.822888283378747,
+ "grad_norm": 9.340838432312012,
+ "learning_rate": 1.9594500177228322e-05,
+ "loss": 1.1777,
+ "step": 4339
+ },
+ {
+ "epoch": 11.825613079019073,
+ "grad_norm": 9.100065231323242,
+ "learning_rate": 1.959425138333013e-05,
+ "loss": 1.3145,
+ "step": 4340
+ },
+ {
+ "epoch": 11.8283378746594,
+ "grad_norm": 6.088210105895996,
+ "learning_rate": 1.959400251471222e-05,
+ "loss": 1.2097,
+ "step": 4341
+ },
+ {
+ "epoch": 11.831062670299728,
+ "grad_norm": 7.304007053375244,
+ "learning_rate": 1.9593753571376524e-05,
+ "loss": 1.0623,
+ "step": 4342
+ },
+ {
+ "epoch": 11.833787465940054,
+ "grad_norm": 7.297685623168945,
+ "learning_rate": 1.9593504553324987e-05,
+ "loss": 1.0051,
+ "step": 4343
+ },
+ {
+ "epoch": 11.836512261580381,
+ "grad_norm": 6.891335964202881,
+ "learning_rate": 1.9593255460559543e-05,
+ "loss": 1.0629,
+ "step": 4344
+ },
+ {
+ "epoch": 11.839237057220709,
+ "grad_norm": 8.22743034362793,
+ "learning_rate": 1.9593006293082134e-05,
+ "loss": 1.2067,
+ "step": 4345
+ },
+ {
+ "epoch": 11.841961852861035,
+ "grad_norm": 7.124205112457275,
+ "learning_rate": 1.9592757050894705e-05,
+ "loss": 0.99,
+ "step": 4346
+ },
+ {
+ "epoch": 11.844686648501362,
+ "grad_norm": 6.263728618621826,
+ "learning_rate": 1.9592507733999187e-05,
+ "loss": 1.0009,
+ "step": 4347
+ },
+ {
+ "epoch": 11.84741144414169,
+ "grad_norm": 7.248764991760254,
+ "learning_rate": 1.959225834239753e-05,
+ "loss": 1.1062,
+ "step": 4348
+ },
+ {
+ "epoch": 11.850136239782016,
+ "grad_norm": 7.125821113586426,
+ "learning_rate": 1.9592008876091676e-05,
+ "loss": 1.2828,
+ "step": 4349
+ },
+ {
+ "epoch": 11.852861035422343,
+ "grad_norm": 7.002594470977783,
+ "learning_rate": 1.9591759335083563e-05,
+ "loss": 1.0288,
+ "step": 4350
+ },
+ {
+ "epoch": 11.85558583106267,
+ "grad_norm": 7.160062789916992,
+ "learning_rate": 1.9591509719375136e-05,
+ "loss": 1.1199,
+ "step": 4351
+ },
+ {
+ "epoch": 11.858310626702997,
+ "grad_norm": 7.415883541107178,
+ "learning_rate": 1.9591260028968344e-05,
+ "loss": 1.2295,
+ "step": 4352
+ },
+ {
+ "epoch": 11.861035422343324,
+ "grad_norm": 6.625397682189941,
+ "learning_rate": 1.9591010263865125e-05,
+ "loss": 0.8923,
+ "step": 4353
+ },
+ {
+ "epoch": 11.863760217983652,
+ "grad_norm": 7.215281963348389,
+ "learning_rate": 1.9590760424067424e-05,
+ "loss": 1.1598,
+ "step": 4354
+ },
+ {
+ "epoch": 11.866485013623977,
+ "grad_norm": 8.622038841247559,
+ "learning_rate": 1.9590510509577195e-05,
+ "loss": 1.2485,
+ "step": 4355
+ },
+ {
+ "epoch": 11.869209809264305,
+ "grad_norm": 6.935097694396973,
+ "learning_rate": 1.9590260520396377e-05,
+ "loss": 1.1013,
+ "step": 4356
+ },
+ {
+ "epoch": 11.871934604904633,
+ "grad_norm": 7.819272041320801,
+ "learning_rate": 1.9590010456526917e-05,
+ "loss": 1.2095,
+ "step": 4357
+ },
+ {
+ "epoch": 11.874659400544958,
+ "grad_norm": 7.4594902992248535,
+ "learning_rate": 1.9589760317970765e-05,
+ "loss": 0.9735,
+ "step": 4358
+ },
+ {
+ "epoch": 11.877384196185286,
+ "grad_norm": 6.736591339111328,
+ "learning_rate": 1.958951010472987e-05,
+ "loss": 0.9434,
+ "step": 4359
+ },
+ {
+ "epoch": 11.880108991825614,
+ "grad_norm": 6.660336017608643,
+ "learning_rate": 1.9589259816806177e-05,
+ "loss": 1.3135,
+ "step": 4360
+ },
+ {
+ "epoch": 11.88283378746594,
+ "grad_norm": 6.942047119140625,
+ "learning_rate": 1.9589009454201634e-05,
+ "loss": 1.1146,
+ "step": 4361
+ },
+ {
+ "epoch": 11.885558583106267,
+ "grad_norm": 7.565939426422119,
+ "learning_rate": 1.9588759016918198e-05,
+ "loss": 0.8447,
+ "step": 4362
+ },
+ {
+ "epoch": 11.888283378746594,
+ "grad_norm": 6.850754261016846,
+ "learning_rate": 1.9588508504957816e-05,
+ "loss": 1.1766,
+ "step": 4363
+ },
+ {
+ "epoch": 11.89100817438692,
+ "grad_norm": 7.702634334564209,
+ "learning_rate": 1.958825791832244e-05,
+ "loss": 1.1648,
+ "step": 4364
+ },
+ {
+ "epoch": 11.893732970027248,
+ "grad_norm": 8.181907653808594,
+ "learning_rate": 1.9588007257014013e-05,
+ "loss": 1.2133,
+ "step": 4365
+ },
+ {
+ "epoch": 11.896457765667575,
+ "grad_norm": 6.545398235321045,
+ "learning_rate": 1.95877565210345e-05,
+ "loss": 0.9912,
+ "step": 4366
+ },
+ {
+ "epoch": 11.899182561307901,
+ "grad_norm": 10.96051025390625,
+ "learning_rate": 1.9587505710385847e-05,
+ "loss": 1.1298,
+ "step": 4367
+ },
+ {
+ "epoch": 11.901907356948229,
+ "grad_norm": 6.98739767074585,
+ "learning_rate": 1.958725482507001e-05,
+ "loss": 1.0903,
+ "step": 4368
+ },
+ {
+ "epoch": 11.904632152588556,
+ "grad_norm": 8.152664184570312,
+ "learning_rate": 1.9587003865088937e-05,
+ "loss": 1.0969,
+ "step": 4369
+ },
+ {
+ "epoch": 11.907356948228882,
+ "grad_norm": 9.344841957092285,
+ "learning_rate": 1.9586752830444585e-05,
+ "loss": 1.1953,
+ "step": 4370
+ },
+ {
+ "epoch": 11.91008174386921,
+ "grad_norm": 6.637526035308838,
+ "learning_rate": 1.9586501721138914e-05,
+ "loss": 1.0686,
+ "step": 4371
+ },
+ {
+ "epoch": 11.912806539509537,
+ "grad_norm": 6.210618495941162,
+ "learning_rate": 1.9586250537173878e-05,
+ "loss": 1.0723,
+ "step": 4372
+ },
+ {
+ "epoch": 11.915531335149863,
+ "grad_norm": 6.431886672973633,
+ "learning_rate": 1.9585999278551426e-05,
+ "loss": 1.0621,
+ "step": 4373
+ },
+ {
+ "epoch": 11.91825613079019,
+ "grad_norm": 9.248215675354004,
+ "learning_rate": 1.9585747945273526e-05,
+ "loss": 1.301,
+ "step": 4374
+ },
+ {
+ "epoch": 11.920980926430518,
+ "grad_norm": 6.088803291320801,
+ "learning_rate": 1.9585496537342123e-05,
+ "loss": 1.0945,
+ "step": 4375
+ },
+ {
+ "epoch": 11.923705722070844,
+ "grad_norm": 5.909554958343506,
+ "learning_rate": 1.958524505475919e-05,
+ "loss": 0.9685,
+ "step": 4376
+ },
+ {
+ "epoch": 11.926430517711172,
+ "grad_norm": 6.614158630371094,
+ "learning_rate": 1.958499349752667e-05,
+ "loss": 1.0278,
+ "step": 4377
+ },
+ {
+ "epoch": 11.9291553133515,
+ "grad_norm": 6.517280578613281,
+ "learning_rate": 1.958474186564653e-05,
+ "loss": 1.1936,
+ "step": 4378
+ },
+ {
+ "epoch": 11.931880108991825,
+ "grad_norm": 7.640071392059326,
+ "learning_rate": 1.958449015912073e-05,
+ "loss": 1.3459,
+ "step": 4379
+ },
+ {
+ "epoch": 11.934604904632153,
+ "grad_norm": 7.099886417388916,
+ "learning_rate": 1.958423837795123e-05,
+ "loss": 1.0105,
+ "step": 4380
+ },
+ {
+ "epoch": 11.93732970027248,
+ "grad_norm": 7.545454978942871,
+ "learning_rate": 1.958398652213999e-05,
+ "loss": 1.0212,
+ "step": 4381
+ },
+ {
+ "epoch": 11.940054495912806,
+ "grad_norm": 7.9395432472229,
+ "learning_rate": 1.9583734591688966e-05,
+ "loss": 1.0776,
+ "step": 4382
+ },
+ {
+ "epoch": 11.942779291553133,
+ "grad_norm": 6.184682369232178,
+ "learning_rate": 1.9583482586600132e-05,
+ "loss": 0.9714,
+ "step": 4383
+ },
+ {
+ "epoch": 11.945504087193461,
+ "grad_norm": 7.863597869873047,
+ "learning_rate": 1.958323050687544e-05,
+ "loss": 1.2097,
+ "step": 4384
+ },
+ {
+ "epoch": 11.948228882833787,
+ "grad_norm": 7.603104114532471,
+ "learning_rate": 1.9582978352516863e-05,
+ "loss": 1.067,
+ "step": 4385
+ },
+ {
+ "epoch": 11.950953678474114,
+ "grad_norm": 9.516350746154785,
+ "learning_rate": 1.958272612352635e-05,
+ "loss": 1.0347,
+ "step": 4386
+ },
+ {
+ "epoch": 11.953678474114442,
+ "grad_norm": 8.684057235717773,
+ "learning_rate": 1.9582473819905884e-05,
+ "loss": 1.0845,
+ "step": 4387
+ },
+ {
+ "epoch": 11.956403269754768,
+ "grad_norm": 7.133975505828857,
+ "learning_rate": 1.9582221441657413e-05,
+ "loss": 1.1387,
+ "step": 4388
+ },
+ {
+ "epoch": 11.959128065395095,
+ "grad_norm": 28.829124450683594,
+ "learning_rate": 1.9581968988782914e-05,
+ "loss": 1.1113,
+ "step": 4389
+ },
+ {
+ "epoch": 11.961852861035423,
+ "grad_norm": 9.247411727905273,
+ "learning_rate": 1.9581716461284344e-05,
+ "loss": 1.1771,
+ "step": 4390
+ },
+ {
+ "epoch": 11.964577656675749,
+ "grad_norm": 12.302692413330078,
+ "learning_rate": 1.958146385916368e-05,
+ "loss": 1.0413,
+ "step": 4391
+ },
+ {
+ "epoch": 11.967302452316076,
+ "grad_norm": 8.387167930603027,
+ "learning_rate": 1.9581211182422882e-05,
+ "loss": 0.9473,
+ "step": 4392
+ },
+ {
+ "epoch": 11.970027247956404,
+ "grad_norm": 6.993246555328369,
+ "learning_rate": 1.958095843106392e-05,
+ "loss": 0.9446,
+ "step": 4393
+ },
+ {
+ "epoch": 11.97275204359673,
+ "grad_norm": 8.019044876098633,
+ "learning_rate": 1.958070560508876e-05,
+ "loss": 1.0288,
+ "step": 4394
+ },
+ {
+ "epoch": 11.975476839237057,
+ "grad_norm": 7.4735612869262695,
+ "learning_rate": 1.9580452704499375e-05,
+ "loss": 1.0364,
+ "step": 4395
+ },
+ {
+ "epoch": 11.978201634877385,
+ "grad_norm": 6.469635963439941,
+ "learning_rate": 1.9580199729297736e-05,
+ "loss": 1.1414,
+ "step": 4396
+ },
+ {
+ "epoch": 11.98092643051771,
+ "grad_norm": 7.731234550476074,
+ "learning_rate": 1.9579946679485803e-05,
+ "loss": 1.1611,
+ "step": 4397
+ },
+ {
+ "epoch": 11.983651226158038,
+ "grad_norm": 8.404643058776855,
+ "learning_rate": 1.9579693555065558e-05,
+ "loss": 1.0288,
+ "step": 4398
+ },
+ {
+ "epoch": 11.986376021798366,
+ "grad_norm": 7.689642429351807,
+ "learning_rate": 1.9579440356038966e-05,
+ "loss": 1.4324,
+ "step": 4399
+ },
+ {
+ "epoch": 11.989100817438691,
+ "grad_norm": 10.964292526245117,
+ "learning_rate": 1.9579187082408003e-05,
+ "loss": 1.14,
+ "step": 4400
+ },
+ {
+ "epoch": 11.991825613079019,
+ "grad_norm": 6.971282958984375,
+ "learning_rate": 1.957893373417464e-05,
+ "loss": 1.1802,
+ "step": 4401
+ },
+ {
+ "epoch": 11.994550408719347,
+ "grad_norm": 5.91721248626709,
+ "learning_rate": 1.9578680311340846e-05,
+ "loss": 1.0325,
+ "step": 4402
+ },
+ {
+ "epoch": 11.997275204359672,
+ "grad_norm": 7.110546588897705,
+ "learning_rate": 1.95784268139086e-05,
+ "loss": 1.1353,
+ "step": 4403
+ },
+ {
+ "epoch": 12.0,
+ "grad_norm": 7.302107334136963,
+ "learning_rate": 1.957817324187987e-05,
+ "loss": 1.0115,
+ "step": 4404
+ },
+ {
+ "epoch": 12.002724795640328,
+ "grad_norm": 6.36070442199707,
+ "learning_rate": 1.9577919595256642e-05,
+ "loss": 0.8485,
+ "step": 4405
+ },
+ {
+ "epoch": 12.005449591280653,
+ "grad_norm": 6.086374759674072,
+ "learning_rate": 1.9577665874040882e-05,
+ "loss": 0.9215,
+ "step": 4406
+ },
+ {
+ "epoch": 12.008174386920981,
+ "grad_norm": 6.267107009887695,
+ "learning_rate": 1.9577412078234565e-05,
+ "loss": 0.8892,
+ "step": 4407
+ },
+ {
+ "epoch": 12.010899182561309,
+ "grad_norm": 5.649913787841797,
+ "learning_rate": 1.9577158207839674e-05,
+ "loss": 0.9069,
+ "step": 4408
+ },
+ {
+ "epoch": 12.013623978201634,
+ "grad_norm": 5.798284530639648,
+ "learning_rate": 1.9576904262858183e-05,
+ "loss": 0.9878,
+ "step": 4409
+ },
+ {
+ "epoch": 12.016348773841962,
+ "grad_norm": 5.548786640167236,
+ "learning_rate": 1.957665024329207e-05,
+ "loss": 0.9946,
+ "step": 4410
+ },
+ {
+ "epoch": 12.01907356948229,
+ "grad_norm": 5.765269756317139,
+ "learning_rate": 1.9576396149143312e-05,
+ "loss": 0.8904,
+ "step": 4411
+ },
+ {
+ "epoch": 12.021798365122615,
+ "grad_norm": 7.161705493927002,
+ "learning_rate": 1.9576141980413893e-05,
+ "loss": 1.1051,
+ "step": 4412
+ },
+ {
+ "epoch": 12.024523160762943,
+ "grad_norm": 7.32899284362793,
+ "learning_rate": 1.957588773710578e-05,
+ "loss": 1.1105,
+ "step": 4413
+ },
+ {
+ "epoch": 12.02724795640327,
+ "grad_norm": 6.176748752593994,
+ "learning_rate": 1.957563341922097e-05,
+ "loss": 0.9674,
+ "step": 4414
+ },
+ {
+ "epoch": 12.029972752043596,
+ "grad_norm": 6.612304210662842,
+ "learning_rate": 1.957537902676143e-05,
+ "loss": 1.0909,
+ "step": 4415
+ },
+ {
+ "epoch": 12.032697547683924,
+ "grad_norm": 6.329996585845947,
+ "learning_rate": 1.9575124559729147e-05,
+ "loss": 0.8394,
+ "step": 4416
+ },
+ {
+ "epoch": 12.035422343324251,
+ "grad_norm": 6.215456008911133,
+ "learning_rate": 1.9574870018126108e-05,
+ "loss": 1.1157,
+ "step": 4417
+ },
+ {
+ "epoch": 12.038147138964577,
+ "grad_norm": 6.929882049560547,
+ "learning_rate": 1.9574615401954283e-05,
+ "loss": 1.0488,
+ "step": 4418
+ },
+ {
+ "epoch": 12.040871934604905,
+ "grad_norm": 6.7328572273254395,
+ "learning_rate": 1.957436071121566e-05,
+ "loss": 0.8989,
+ "step": 4419
+ },
+ {
+ "epoch": 12.043596730245232,
+ "grad_norm": 6.993234634399414,
+ "learning_rate": 1.957410594591223e-05,
+ "loss": 0.9229,
+ "step": 4420
+ },
+ {
+ "epoch": 12.046321525885558,
+ "grad_norm": 7.745464324951172,
+ "learning_rate": 1.9573851106045965e-05,
+ "loss": 1.0842,
+ "step": 4421
+ },
+ {
+ "epoch": 12.049046321525886,
+ "grad_norm": 6.4803547859191895,
+ "learning_rate": 1.957359619161886e-05,
+ "loss": 1.0957,
+ "step": 4422
+ },
+ {
+ "epoch": 12.051771117166213,
+ "grad_norm": 6.850926876068115,
+ "learning_rate": 1.9573341202632897e-05,
+ "loss": 0.9598,
+ "step": 4423
+ },
+ {
+ "epoch": 12.054495912806539,
+ "grad_norm": 7.264588832855225,
+ "learning_rate": 1.9573086139090056e-05,
+ "loss": 0.9924,
+ "step": 4424
+ },
+ {
+ "epoch": 12.057220708446867,
+ "grad_norm": 6.775201797485352,
+ "learning_rate": 1.957283100099233e-05,
+ "loss": 0.859,
+ "step": 4425
+ },
+ {
+ "epoch": 12.059945504087194,
+ "grad_norm": 6.906174182891846,
+ "learning_rate": 1.9572575788341707e-05,
+ "loss": 1.0768,
+ "step": 4426
+ },
+ {
+ "epoch": 12.06267029972752,
+ "grad_norm": 8.009795188903809,
+ "learning_rate": 1.957232050114017e-05,
+ "loss": 0.9979,
+ "step": 4427
+ },
+ {
+ "epoch": 12.065395095367847,
+ "grad_norm": 6.672793388366699,
+ "learning_rate": 1.957206513938971e-05,
+ "loss": 1.1185,
+ "step": 4428
+ },
+ {
+ "epoch": 12.068119891008175,
+ "grad_norm": 7.434624671936035,
+ "learning_rate": 1.957180970309231e-05,
+ "loss": 1.05,
+ "step": 4429
+ },
+ {
+ "epoch": 12.0708446866485,
+ "grad_norm": 8.33358097076416,
+ "learning_rate": 1.957155419224997e-05,
+ "loss": 1.3,
+ "step": 4430
+ },
+ {
+ "epoch": 12.073569482288828,
+ "grad_norm": 8.042304039001465,
+ "learning_rate": 1.957129860686467e-05,
+ "loss": 0.9375,
+ "step": 4431
+ },
+ {
+ "epoch": 12.076294277929156,
+ "grad_norm": 7.082531452178955,
+ "learning_rate": 1.9571042946938408e-05,
+ "loss": 1.0282,
+ "step": 4432
+ },
+ {
+ "epoch": 12.079019073569482,
+ "grad_norm": 7.858602523803711,
+ "learning_rate": 1.957078721247317e-05,
+ "loss": 1.1204,
+ "step": 4433
+ },
+ {
+ "epoch": 12.08174386920981,
+ "grad_norm": 9.217330932617188,
+ "learning_rate": 1.957053140347095e-05,
+ "loss": 0.8784,
+ "step": 4434
+ },
+ {
+ "epoch": 12.084468664850137,
+ "grad_norm": 7.062445640563965,
+ "learning_rate": 1.957027551993374e-05,
+ "loss": 0.9315,
+ "step": 4435
+ },
+ {
+ "epoch": 12.087193460490463,
+ "grad_norm": 5.703321933746338,
+ "learning_rate": 1.957001956186353e-05,
+ "loss": 1.0282,
+ "step": 4436
+ },
+ {
+ "epoch": 12.08991825613079,
+ "grad_norm": 6.803126811981201,
+ "learning_rate": 1.9569763529262315e-05,
+ "loss": 1.0366,
+ "step": 4437
+ },
+ {
+ "epoch": 12.092643051771118,
+ "grad_norm": 7.230068683624268,
+ "learning_rate": 1.9569507422132092e-05,
+ "loss": 1.0393,
+ "step": 4438
+ },
+ {
+ "epoch": 12.095367847411444,
+ "grad_norm": 8.00177001953125,
+ "learning_rate": 1.956925124047485e-05,
+ "loss": 1.0786,
+ "step": 4439
+ },
+ {
+ "epoch": 12.098092643051771,
+ "grad_norm": 6.813270568847656,
+ "learning_rate": 1.9568994984292592e-05,
+ "loss": 1.0187,
+ "step": 4440
+ },
+ {
+ "epoch": 12.100817438692099,
+ "grad_norm": 14.719013214111328,
+ "learning_rate": 1.9568738653587304e-05,
+ "loss": 0.9973,
+ "step": 4441
+ },
+ {
+ "epoch": 12.103542234332425,
+ "grad_norm": 5.707857608795166,
+ "learning_rate": 1.9568482248360992e-05,
+ "loss": 0.8575,
+ "step": 4442
+ },
+ {
+ "epoch": 12.106267029972752,
+ "grad_norm": 7.245703220367432,
+ "learning_rate": 1.956822576861565e-05,
+ "loss": 0.9373,
+ "step": 4443
+ },
+ {
+ "epoch": 12.10899182561308,
+ "grad_norm": 6.485202312469482,
+ "learning_rate": 1.9567969214353266e-05,
+ "loss": 1.012,
+ "step": 4444
+ },
+ {
+ "epoch": 12.111716621253406,
+ "grad_norm": 6.57015323638916,
+ "learning_rate": 1.956771258557585e-05,
+ "loss": 0.9738,
+ "step": 4445
+ },
+ {
+ "epoch": 12.114441416893733,
+ "grad_norm": 7.286382675170898,
+ "learning_rate": 1.9567455882285397e-05,
+ "loss": 1.1201,
+ "step": 4446
+ },
+ {
+ "epoch": 12.11716621253406,
+ "grad_norm": 5.3262619972229,
+ "learning_rate": 1.95671991044839e-05,
+ "loss": 0.8875,
+ "step": 4447
+ },
+ {
+ "epoch": 12.119891008174386,
+ "grad_norm": 7.8376970291137695,
+ "learning_rate": 1.9566942252173373e-05,
+ "loss": 1.0769,
+ "step": 4448
+ },
+ {
+ "epoch": 12.122615803814714,
+ "grad_norm": 8.119297981262207,
+ "learning_rate": 1.9566685325355803e-05,
+ "loss": 1.142,
+ "step": 4449
+ },
+ {
+ "epoch": 12.125340599455042,
+ "grad_norm": 6.434397220611572,
+ "learning_rate": 1.9566428324033192e-05,
+ "loss": 0.9238,
+ "step": 4450
+ },
+ {
+ "epoch": 12.128065395095367,
+ "grad_norm": 7.574224472045898,
+ "learning_rate": 1.9566171248207552e-05,
+ "loss": 1.0465,
+ "step": 4451
+ },
+ {
+ "epoch": 12.130790190735695,
+ "grad_norm": 7.3289289474487305,
+ "learning_rate": 1.9565914097880874e-05,
+ "loss": 1.0586,
+ "step": 4452
+ },
+ {
+ "epoch": 12.133514986376023,
+ "grad_norm": 7.254357814788818,
+ "learning_rate": 1.956565687305517e-05,
+ "loss": 1.1259,
+ "step": 4453
+ },
+ {
+ "epoch": 12.136239782016348,
+ "grad_norm": 5.836382865905762,
+ "learning_rate": 1.9565399573732432e-05,
+ "loss": 1.0131,
+ "step": 4454
+ },
+ {
+ "epoch": 12.138964577656676,
+ "grad_norm": 6.244225025177002,
+ "learning_rate": 1.956514219991467e-05,
+ "loss": 0.9343,
+ "step": 4455
+ },
+ {
+ "epoch": 12.141689373297003,
+ "grad_norm": 9.760071754455566,
+ "learning_rate": 1.956488475160389e-05,
+ "loss": 1.1011,
+ "step": 4456
+ },
+ {
+ "epoch": 12.14441416893733,
+ "grad_norm": 5.95624303817749,
+ "learning_rate": 1.9564627228802098e-05,
+ "loss": 1.0164,
+ "step": 4457
+ },
+ {
+ "epoch": 12.147138964577657,
+ "grad_norm": 6.010317325592041,
+ "learning_rate": 1.9564369631511296e-05,
+ "loss": 1.1836,
+ "step": 4458
+ },
+ {
+ "epoch": 12.149863760217984,
+ "grad_norm": 6.346007823944092,
+ "learning_rate": 1.956411195973349e-05,
+ "loss": 0.9626,
+ "step": 4459
+ },
+ {
+ "epoch": 12.15258855585831,
+ "grad_norm": 6.7293219566345215,
+ "learning_rate": 1.956385421347069e-05,
+ "loss": 0.9597,
+ "step": 4460
+ },
+ {
+ "epoch": 12.155313351498638,
+ "grad_norm": 6.554243564605713,
+ "learning_rate": 1.9563596392724897e-05,
+ "loss": 1.1527,
+ "step": 4461
+ },
+ {
+ "epoch": 12.158038147138965,
+ "grad_norm": 7.238015174865723,
+ "learning_rate": 1.9563338497498127e-05,
+ "loss": 1.1536,
+ "step": 4462
+ },
+ {
+ "epoch": 12.160762942779291,
+ "grad_norm": 7.1554856300354,
+ "learning_rate": 1.9563080527792383e-05,
+ "loss": 1.0044,
+ "step": 4463
+ },
+ {
+ "epoch": 12.163487738419619,
+ "grad_norm": 8.222060203552246,
+ "learning_rate": 1.9562822483609675e-05,
+ "loss": 0.9294,
+ "step": 4464
+ },
+ {
+ "epoch": 12.166212534059946,
+ "grad_norm": 7.817305088043213,
+ "learning_rate": 1.9562564364952012e-05,
+ "loss": 1.1396,
+ "step": 4465
+ },
+ {
+ "epoch": 12.168937329700272,
+ "grad_norm": 6.5493059158325195,
+ "learning_rate": 1.956230617182141e-05,
+ "loss": 0.8979,
+ "step": 4466
+ },
+ {
+ "epoch": 12.1716621253406,
+ "grad_norm": 7.5641937255859375,
+ "learning_rate": 1.956204790421987e-05,
+ "loss": 1.0679,
+ "step": 4467
+ },
+ {
+ "epoch": 12.174386920980927,
+ "grad_norm": 6.7236809730529785,
+ "learning_rate": 1.956178956214941e-05,
+ "loss": 1.0852,
+ "step": 4468
+ },
+ {
+ "epoch": 12.177111716621253,
+ "grad_norm": 5.441164016723633,
+ "learning_rate": 1.9561531145612042e-05,
+ "loss": 0.8318,
+ "step": 4469
+ },
+ {
+ "epoch": 12.17983651226158,
+ "grad_norm": 6.103809356689453,
+ "learning_rate": 1.956127265460978e-05,
+ "loss": 0.9293,
+ "step": 4470
+ },
+ {
+ "epoch": 12.182561307901908,
+ "grad_norm": 6.256866455078125,
+ "learning_rate": 1.9561014089144634e-05,
+ "loss": 1.1145,
+ "step": 4471
+ },
+ {
+ "epoch": 12.185286103542234,
+ "grad_norm": 6.847880840301514,
+ "learning_rate": 1.9560755449218612e-05,
+ "loss": 0.884,
+ "step": 4472
+ },
+ {
+ "epoch": 12.188010899182562,
+ "grad_norm": 8.559353828430176,
+ "learning_rate": 1.956049673483374e-05,
+ "loss": 0.9395,
+ "step": 4473
+ },
+ {
+ "epoch": 12.190735694822889,
+ "grad_norm": 5.659292221069336,
+ "learning_rate": 1.956023794599202e-05,
+ "loss": 1.0708,
+ "step": 4474
+ },
+ {
+ "epoch": 12.193460490463215,
+ "grad_norm": 6.5723066329956055,
+ "learning_rate": 1.9559979082695483e-05,
+ "loss": 1.1318,
+ "step": 4475
+ },
+ {
+ "epoch": 12.196185286103542,
+ "grad_norm": 8.922590255737305,
+ "learning_rate": 1.955972014494613e-05,
+ "loss": 1.1111,
+ "step": 4476
+ },
+ {
+ "epoch": 12.19891008174387,
+ "grad_norm": 6.509793281555176,
+ "learning_rate": 1.955946113274599e-05,
+ "loss": 0.9377,
+ "step": 4477
+ },
+ {
+ "epoch": 12.201634877384196,
+ "grad_norm": 5.996234893798828,
+ "learning_rate": 1.955920204609707e-05,
+ "loss": 1.0582,
+ "step": 4478
+ },
+ {
+ "epoch": 12.204359673024523,
+ "grad_norm": 6.371071815490723,
+ "learning_rate": 1.9558942885001394e-05,
+ "loss": 1.121,
+ "step": 4479
+ },
+ {
+ "epoch": 12.207084468664851,
+ "grad_norm": 11.321345329284668,
+ "learning_rate": 1.9558683649460977e-05,
+ "loss": 1.1001,
+ "step": 4480
+ },
+ {
+ "epoch": 12.209809264305177,
+ "grad_norm": 8.580024719238281,
+ "learning_rate": 1.955842433947784e-05,
+ "loss": 0.9493,
+ "step": 4481
+ },
+ {
+ "epoch": 12.212534059945504,
+ "grad_norm": 6.603857040405273,
+ "learning_rate": 1.9558164955054005e-05,
+ "loss": 0.976,
+ "step": 4482
+ },
+ {
+ "epoch": 12.215258855585832,
+ "grad_norm": 6.731605529785156,
+ "learning_rate": 1.9557905496191485e-05,
+ "loss": 0.9015,
+ "step": 4483
+ },
+ {
+ "epoch": 12.217983651226158,
+ "grad_norm": 6.754188537597656,
+ "learning_rate": 1.9557645962892306e-05,
+ "loss": 0.8676,
+ "step": 4484
+ },
+ {
+ "epoch": 12.220708446866485,
+ "grad_norm": 8.059829711914062,
+ "learning_rate": 1.9557386355158487e-05,
+ "loss": 0.895,
+ "step": 4485
+ },
+ {
+ "epoch": 12.223433242506813,
+ "grad_norm": 5.666400909423828,
+ "learning_rate": 1.955712667299205e-05,
+ "loss": 1.0367,
+ "step": 4486
+ },
+ {
+ "epoch": 12.226158038147139,
+ "grad_norm": 6.189394950866699,
+ "learning_rate": 1.9556866916395025e-05,
+ "loss": 1.1599,
+ "step": 4487
+ },
+ {
+ "epoch": 12.228882833787466,
+ "grad_norm": 6.235940933227539,
+ "learning_rate": 1.9556607085369423e-05,
+ "loss": 0.8811,
+ "step": 4488
+ },
+ {
+ "epoch": 12.231607629427794,
+ "grad_norm": 6.477928638458252,
+ "learning_rate": 1.955634717991727e-05,
+ "loss": 0.9636,
+ "step": 4489
+ },
+ {
+ "epoch": 12.23433242506812,
+ "grad_norm": 7.549765110015869,
+ "learning_rate": 1.9556087200040595e-05,
+ "loss": 1.002,
+ "step": 4490
+ },
+ {
+ "epoch": 12.237057220708447,
+ "grad_norm": 19.572811126708984,
+ "learning_rate": 1.9555827145741418e-05,
+ "loss": 0.9313,
+ "step": 4491
+ },
+ {
+ "epoch": 12.239782016348773,
+ "grad_norm": 6.787742614746094,
+ "learning_rate": 1.955556701702177e-05,
+ "loss": 0.8984,
+ "step": 4492
+ },
+ {
+ "epoch": 12.2425068119891,
+ "grad_norm": 7.650527477264404,
+ "learning_rate": 1.955530681388367e-05,
+ "loss": 0.9174,
+ "step": 4493
+ },
+ {
+ "epoch": 12.245231607629428,
+ "grad_norm": 10.568982124328613,
+ "learning_rate": 1.9555046536329154e-05,
+ "loss": 0.9803,
+ "step": 4494
+ },
+ {
+ "epoch": 12.247956403269754,
+ "grad_norm": 6.881139278411865,
+ "learning_rate": 1.955478618436024e-05,
+ "loss": 1.0298,
+ "step": 4495
+ },
+ {
+ "epoch": 12.250681198910081,
+ "grad_norm": 7.805494785308838,
+ "learning_rate": 1.9554525757978958e-05,
+ "loss": 1.08,
+ "step": 4496
+ },
+ {
+ "epoch": 12.253405994550409,
+ "grad_norm": 6.392650127410889,
+ "learning_rate": 1.9554265257187336e-05,
+ "loss": 0.9188,
+ "step": 4497
+ },
+ {
+ "epoch": 12.256130790190735,
+ "grad_norm": 8.11082935333252,
+ "learning_rate": 1.955400468198741e-05,
+ "loss": 0.9257,
+ "step": 4498
+ },
+ {
+ "epoch": 12.258855585831062,
+ "grad_norm": 7.071857452392578,
+ "learning_rate": 1.9553744032381198e-05,
+ "loss": 1.0542,
+ "step": 4499
+ },
+ {
+ "epoch": 12.26158038147139,
+ "grad_norm": 5.893857002258301,
+ "learning_rate": 1.9553483308370732e-05,
+ "loss": 0.9825,
+ "step": 4500
+ },
+ {
+ "epoch": 12.264305177111716,
+ "grad_norm": 6.294366836547852,
+ "learning_rate": 1.955322250995805e-05,
+ "loss": 0.986,
+ "step": 4501
+ },
+ {
+ "epoch": 12.267029972752043,
+ "grad_norm": 6.73477840423584,
+ "learning_rate": 1.9552961637145177e-05,
+ "loss": 1.0017,
+ "step": 4502
+ },
+ {
+ "epoch": 12.269754768392371,
+ "grad_norm": 6.7410712242126465,
+ "learning_rate": 1.9552700689934146e-05,
+ "loss": 1.0667,
+ "step": 4503
+ },
+ {
+ "epoch": 12.272479564032697,
+ "grad_norm": 9.9473876953125,
+ "learning_rate": 1.9552439668326993e-05,
+ "loss": 0.9885,
+ "step": 4504
+ },
+ {
+ "epoch": 12.275204359673024,
+ "grad_norm": 6.462554931640625,
+ "learning_rate": 1.955217857232574e-05,
+ "loss": 1.0182,
+ "step": 4505
+ },
+ {
+ "epoch": 12.277929155313352,
+ "grad_norm": 7.129510879516602,
+ "learning_rate": 1.9551917401932436e-05,
+ "loss": 1.0662,
+ "step": 4506
+ },
+ {
+ "epoch": 12.280653950953678,
+ "grad_norm": 7.244655132293701,
+ "learning_rate": 1.95516561571491e-05,
+ "loss": 0.9302,
+ "step": 4507
+ },
+ {
+ "epoch": 12.283378746594005,
+ "grad_norm": 7.241304874420166,
+ "learning_rate": 1.9551394837977776e-05,
+ "loss": 1.1545,
+ "step": 4508
+ },
+ {
+ "epoch": 12.286103542234333,
+ "grad_norm": 7.047787189483643,
+ "learning_rate": 1.9551133444420495e-05,
+ "loss": 1.0198,
+ "step": 4509
+ },
+ {
+ "epoch": 12.288828337874659,
+ "grad_norm": 6.893231391906738,
+ "learning_rate": 1.9550871976479297e-05,
+ "loss": 0.9543,
+ "step": 4510
+ },
+ {
+ "epoch": 12.291553133514986,
+ "grad_norm": 5.670594692230225,
+ "learning_rate": 1.9550610434156213e-05,
+ "loss": 0.8799,
+ "step": 4511
+ },
+ {
+ "epoch": 12.294277929155314,
+ "grad_norm": 6.841028690338135,
+ "learning_rate": 1.9550348817453286e-05,
+ "loss": 1.1211,
+ "step": 4512
+ },
+ {
+ "epoch": 12.29700272479564,
+ "grad_norm": 7.723793983459473,
+ "learning_rate": 1.9550087126372544e-05,
+ "loss": 0.8868,
+ "step": 4513
+ },
+ {
+ "epoch": 12.299727520435967,
+ "grad_norm": 6.6285319328308105,
+ "learning_rate": 1.9549825360916035e-05,
+ "loss": 0.8978,
+ "step": 4514
+ },
+ {
+ "epoch": 12.302452316076295,
+ "grad_norm": 7.060239791870117,
+ "learning_rate": 1.954956352108579e-05,
+ "loss": 0.8463,
+ "step": 4515
+ },
+ {
+ "epoch": 12.30517711171662,
+ "grad_norm": 6.2819976806640625,
+ "learning_rate": 1.9549301606883855e-05,
+ "loss": 1.0291,
+ "step": 4516
+ },
+ {
+ "epoch": 12.307901907356948,
+ "grad_norm": 7.163010120391846,
+ "learning_rate": 1.9549039618312268e-05,
+ "loss": 1.0298,
+ "step": 4517
+ },
+ {
+ "epoch": 12.310626702997276,
+ "grad_norm": 6.741421222686768,
+ "learning_rate": 1.9548777555373065e-05,
+ "loss": 1.0171,
+ "step": 4518
+ },
+ {
+ "epoch": 12.313351498637601,
+ "grad_norm": 6.7361884117126465,
+ "learning_rate": 1.954851541806829e-05,
+ "loss": 1.0184,
+ "step": 4519
+ },
+ {
+ "epoch": 12.316076294277929,
+ "grad_norm": 4.9358954429626465,
+ "learning_rate": 1.9548253206399983e-05,
+ "loss": 0.9579,
+ "step": 4520
+ },
+ {
+ "epoch": 12.318801089918257,
+ "grad_norm": 7.675539970397949,
+ "learning_rate": 1.954799092037019e-05,
+ "loss": 1.1696,
+ "step": 4521
+ },
+ {
+ "epoch": 12.321525885558582,
+ "grad_norm": 7.025536060333252,
+ "learning_rate": 1.9547728559980948e-05,
+ "loss": 1.0291,
+ "step": 4522
+ },
+ {
+ "epoch": 12.32425068119891,
+ "grad_norm": 6.843891620635986,
+ "learning_rate": 1.9547466125234305e-05,
+ "loss": 0.921,
+ "step": 4523
+ },
+ {
+ "epoch": 12.326975476839237,
+ "grad_norm": 6.536768436431885,
+ "learning_rate": 1.95472036161323e-05,
+ "loss": 1.0422,
+ "step": 4524
+ },
+ {
+ "epoch": 12.329700272479563,
+ "grad_norm": 7.380743503570557,
+ "learning_rate": 1.9546941032676986e-05,
+ "loss": 1.3315,
+ "step": 4525
+ },
+ {
+ "epoch": 12.33242506811989,
+ "grad_norm": 6.818024158477783,
+ "learning_rate": 1.95466783748704e-05,
+ "loss": 1.1113,
+ "step": 4526
+ },
+ {
+ "epoch": 12.335149863760218,
+ "grad_norm": 7.1066975593566895,
+ "learning_rate": 1.954641564271459e-05,
+ "loss": 0.9219,
+ "step": 4527
+ },
+ {
+ "epoch": 12.337874659400544,
+ "grad_norm": 7.016533851623535,
+ "learning_rate": 1.95461528362116e-05,
+ "loss": 1.1138,
+ "step": 4528
+ },
+ {
+ "epoch": 12.340599455040872,
+ "grad_norm": 7.665073394775391,
+ "learning_rate": 1.9545889955363483e-05,
+ "loss": 1.1604,
+ "step": 4529
+ },
+ {
+ "epoch": 12.3433242506812,
+ "grad_norm": 7.746526718139648,
+ "learning_rate": 1.954562700017228e-05,
+ "loss": 1.1758,
+ "step": 4530
+ },
+ {
+ "epoch": 12.346049046321525,
+ "grad_norm": 8.191317558288574,
+ "learning_rate": 1.954536397064004e-05,
+ "loss": 1.1571,
+ "step": 4531
+ },
+ {
+ "epoch": 12.348773841961853,
+ "grad_norm": 8.292984008789062,
+ "learning_rate": 1.954510086676882e-05,
+ "loss": 0.9645,
+ "step": 4532
+ },
+ {
+ "epoch": 12.35149863760218,
+ "grad_norm": 6.618183135986328,
+ "learning_rate": 1.9544837688560653e-05,
+ "loss": 0.8956,
+ "step": 4533
+ },
+ {
+ "epoch": 12.354223433242506,
+ "grad_norm": 6.356087684631348,
+ "learning_rate": 1.95445744360176e-05,
+ "loss": 1.2876,
+ "step": 4534
+ },
+ {
+ "epoch": 12.356948228882834,
+ "grad_norm": 6.87322473526001,
+ "learning_rate": 1.954431110914171e-05,
+ "loss": 1.0398,
+ "step": 4535
+ },
+ {
+ "epoch": 12.359673024523161,
+ "grad_norm": 6.9940996170043945,
+ "learning_rate": 1.9544047707935026e-05,
+ "loss": 1.1428,
+ "step": 4536
+ },
+ {
+ "epoch": 12.362397820163487,
+ "grad_norm": 6.57802677154541,
+ "learning_rate": 1.954378423239961e-05,
+ "loss": 0.9673,
+ "step": 4537
+ },
+ {
+ "epoch": 12.365122615803815,
+ "grad_norm": 6.282308101654053,
+ "learning_rate": 1.9543520682537514e-05,
+ "loss": 1.0536,
+ "step": 4538
+ },
+ {
+ "epoch": 12.367847411444142,
+ "grad_norm": 6.919661521911621,
+ "learning_rate": 1.954325705835078e-05,
+ "loss": 1.186,
+ "step": 4539
+ },
+ {
+ "epoch": 12.370572207084468,
+ "grad_norm": 6.682716369628906,
+ "learning_rate": 1.954299335984147e-05,
+ "loss": 0.9131,
+ "step": 4540
+ },
+ {
+ "epoch": 12.373297002724795,
+ "grad_norm": 7.2920074462890625,
+ "learning_rate": 1.9542729587011635e-05,
+ "loss": 1.0038,
+ "step": 4541
+ },
+ {
+ "epoch": 12.376021798365123,
+ "grad_norm": 6.9396514892578125,
+ "learning_rate": 1.954246573986333e-05,
+ "loss": 0.8965,
+ "step": 4542
+ },
+ {
+ "epoch": 12.378746594005449,
+ "grad_norm": 6.466540336608887,
+ "learning_rate": 1.9542201818398606e-05,
+ "loss": 1.0679,
+ "step": 4543
+ },
+ {
+ "epoch": 12.381471389645776,
+ "grad_norm": 6.6070966720581055,
+ "learning_rate": 1.9541937822619524e-05,
+ "loss": 0.9301,
+ "step": 4544
+ },
+ {
+ "epoch": 12.384196185286104,
+ "grad_norm": 7.690656661987305,
+ "learning_rate": 1.9541673752528135e-05,
+ "loss": 1.0297,
+ "step": 4545
+ },
+ {
+ "epoch": 12.38692098092643,
+ "grad_norm": 6.483881950378418,
+ "learning_rate": 1.95414096081265e-05,
+ "loss": 1.0634,
+ "step": 4546
+ },
+ {
+ "epoch": 12.389645776566757,
+ "grad_norm": 10.867969512939453,
+ "learning_rate": 1.954114538941667e-05,
+ "loss": 0.873,
+ "step": 4547
+ },
+ {
+ "epoch": 12.392370572207085,
+ "grad_norm": 8.498041152954102,
+ "learning_rate": 1.954088109640071e-05,
+ "loss": 1.0586,
+ "step": 4548
+ },
+ {
+ "epoch": 12.39509536784741,
+ "grad_norm": 6.8494486808776855,
+ "learning_rate": 1.9540616729080676e-05,
+ "loss": 1.1958,
+ "step": 4549
+ },
+ {
+ "epoch": 12.397820163487738,
+ "grad_norm": 8.04880142211914,
+ "learning_rate": 1.9540352287458624e-05,
+ "loss": 1.0947,
+ "step": 4550
+ },
+ {
+ "epoch": 12.400544959128066,
+ "grad_norm": 8.042343139648438,
+ "learning_rate": 1.9540087771536617e-05,
+ "loss": 1.0767,
+ "step": 4551
+ },
+ {
+ "epoch": 12.403269754768392,
+ "grad_norm": 8.339240074157715,
+ "learning_rate": 1.953982318131671e-05,
+ "loss": 0.996,
+ "step": 4552
+ },
+ {
+ "epoch": 12.40599455040872,
+ "grad_norm": 6.718893051147461,
+ "learning_rate": 1.9539558516800968e-05,
+ "loss": 0.9237,
+ "step": 4553
+ },
+ {
+ "epoch": 12.408719346049047,
+ "grad_norm": 7.775845050811768,
+ "learning_rate": 1.9539293777991456e-05,
+ "loss": 1.0598,
+ "step": 4554
+ },
+ {
+ "epoch": 12.411444141689373,
+ "grad_norm": 8.123579025268555,
+ "learning_rate": 1.9539028964890227e-05,
+ "loss": 1.0967,
+ "step": 4555
+ },
+ {
+ "epoch": 12.4141689373297,
+ "grad_norm": 7.44754695892334,
+ "learning_rate": 1.953876407749935e-05,
+ "loss": 1.0677,
+ "step": 4556
+ },
+ {
+ "epoch": 12.416893732970028,
+ "grad_norm": 6.368758201599121,
+ "learning_rate": 1.9538499115820883e-05,
+ "loss": 1.0291,
+ "step": 4557
+ },
+ {
+ "epoch": 12.419618528610354,
+ "grad_norm": 11.399771690368652,
+ "learning_rate": 1.9538234079856892e-05,
+ "loss": 1.2906,
+ "step": 4558
+ },
+ {
+ "epoch": 12.422343324250681,
+ "grad_norm": 7.041598320007324,
+ "learning_rate": 1.9537968969609444e-05,
+ "loss": 0.9615,
+ "step": 4559
+ },
+ {
+ "epoch": 12.425068119891009,
+ "grad_norm": 6.146443843841553,
+ "learning_rate": 1.9537703785080598e-05,
+ "loss": 0.8502,
+ "step": 4560
+ },
+ {
+ "epoch": 12.427792915531334,
+ "grad_norm": 7.814451694488525,
+ "learning_rate": 1.953743852627242e-05,
+ "loss": 1.0658,
+ "step": 4561
+ },
+ {
+ "epoch": 12.430517711171662,
+ "grad_norm": 6.779821872711182,
+ "learning_rate": 1.953717319318698e-05,
+ "loss": 0.9967,
+ "step": 4562
+ },
+ {
+ "epoch": 12.43324250681199,
+ "grad_norm": 6.233975410461426,
+ "learning_rate": 1.9536907785826345e-05,
+ "loss": 1.0165,
+ "step": 4563
+ },
+ {
+ "epoch": 12.435967302452315,
+ "grad_norm": 7.926974296569824,
+ "learning_rate": 1.9536642304192574e-05,
+ "loss": 1.2899,
+ "step": 4564
+ },
+ {
+ "epoch": 12.438692098092643,
+ "grad_norm": 5.928084373474121,
+ "learning_rate": 1.9536376748287743e-05,
+ "loss": 0.9335,
+ "step": 4565
+ },
+ {
+ "epoch": 12.44141689373297,
+ "grad_norm": 5.872074604034424,
+ "learning_rate": 1.9536111118113917e-05,
+ "loss": 0.9841,
+ "step": 4566
+ },
+ {
+ "epoch": 12.444141689373296,
+ "grad_norm": 7.07937479019165,
+ "learning_rate": 1.953584541367316e-05,
+ "loss": 1.1121,
+ "step": 4567
+ },
+ {
+ "epoch": 12.446866485013624,
+ "grad_norm": 11.801812171936035,
+ "learning_rate": 1.9535579634967548e-05,
+ "loss": 0.838,
+ "step": 4568
+ },
+ {
+ "epoch": 12.449591280653951,
+ "grad_norm": 9.145610809326172,
+ "learning_rate": 1.9535313781999155e-05,
+ "loss": 0.988,
+ "step": 4569
+ },
+ {
+ "epoch": 12.452316076294277,
+ "grad_norm": 6.585156440734863,
+ "learning_rate": 1.9535047854770038e-05,
+ "loss": 1.0248,
+ "step": 4570
+ },
+ {
+ "epoch": 12.455040871934605,
+ "grad_norm": 7.698279857635498,
+ "learning_rate": 1.9534781853282278e-05,
+ "loss": 1.0365,
+ "step": 4571
+ },
+ {
+ "epoch": 12.457765667574932,
+ "grad_norm": 9.520374298095703,
+ "learning_rate": 1.953451577753794e-05,
+ "loss": 1.1616,
+ "step": 4572
+ },
+ {
+ "epoch": 12.460490463215258,
+ "grad_norm": 7.511384010314941,
+ "learning_rate": 1.9534249627539105e-05,
+ "loss": 0.9479,
+ "step": 4573
+ },
+ {
+ "epoch": 12.463215258855586,
+ "grad_norm": 8.412039756774902,
+ "learning_rate": 1.953398340328784e-05,
+ "loss": 0.8976,
+ "step": 4574
+ },
+ {
+ "epoch": 12.465940054495913,
+ "grad_norm": 7.360005855560303,
+ "learning_rate": 1.9533717104786218e-05,
+ "loss": 0.9281,
+ "step": 4575
+ },
+ {
+ "epoch": 12.46866485013624,
+ "grad_norm": 8.331911087036133,
+ "learning_rate": 1.953345073203631e-05,
+ "loss": 1.1377,
+ "step": 4576
+ },
+ {
+ "epoch": 12.471389645776567,
+ "grad_norm": 7.162471771240234,
+ "learning_rate": 1.95331842850402e-05,
+ "loss": 1.1339,
+ "step": 4577
+ },
+ {
+ "epoch": 12.474114441416894,
+ "grad_norm": 7.478108882904053,
+ "learning_rate": 1.9532917763799957e-05,
+ "loss": 0.9885,
+ "step": 4578
+ },
+ {
+ "epoch": 12.47683923705722,
+ "grad_norm": 6.804027080535889,
+ "learning_rate": 1.9532651168317657e-05,
+ "loss": 0.99,
+ "step": 4579
+ },
+ {
+ "epoch": 12.479564032697548,
+ "grad_norm": 6.420581817626953,
+ "learning_rate": 1.9532384498595376e-05,
+ "loss": 1.0907,
+ "step": 4580
+ },
+ {
+ "epoch": 12.482288828337875,
+ "grad_norm": 10.799995422363281,
+ "learning_rate": 1.953211775463519e-05,
+ "loss": 1.2024,
+ "step": 4581
+ },
+ {
+ "epoch": 12.485013623978201,
+ "grad_norm": 7.723841667175293,
+ "learning_rate": 1.9531850936439177e-05,
+ "loss": 1.1011,
+ "step": 4582
+ },
+ {
+ "epoch": 12.487738419618529,
+ "grad_norm": 6.291035175323486,
+ "learning_rate": 1.953158404400942e-05,
+ "loss": 1.0339,
+ "step": 4583
+ },
+ {
+ "epoch": 12.490463215258856,
+ "grad_norm": 8.12030029296875,
+ "learning_rate": 1.953131707734799e-05,
+ "loss": 0.9492,
+ "step": 4584
+ },
+ {
+ "epoch": 12.493188010899182,
+ "grad_norm": 7.230988025665283,
+ "learning_rate": 1.9531050036456972e-05,
+ "loss": 1.1072,
+ "step": 4585
+ },
+ {
+ "epoch": 12.49591280653951,
+ "grad_norm": 6.638936996459961,
+ "learning_rate": 1.9530782921338442e-05,
+ "loss": 0.9879,
+ "step": 4586
+ },
+ {
+ "epoch": 12.498637602179837,
+ "grad_norm": 8.36536693572998,
+ "learning_rate": 1.953051573199448e-05,
+ "loss": 0.8993,
+ "step": 4587
+ },
+ {
+ "epoch": 12.501362397820163,
+ "grad_norm": 7.621925354003906,
+ "learning_rate": 1.9530248468427166e-05,
+ "loss": 1.1056,
+ "step": 4588
+ },
+ {
+ "epoch": 12.50408719346049,
+ "grad_norm": 7.584161758422852,
+ "learning_rate": 1.9529981130638585e-05,
+ "loss": 1.0157,
+ "step": 4589
+ },
+ {
+ "epoch": 12.506811989100818,
+ "grad_norm": 6.6787333488464355,
+ "learning_rate": 1.952971371863082e-05,
+ "loss": 1.1035,
+ "step": 4590
+ },
+ {
+ "epoch": 12.509536784741144,
+ "grad_norm": 7.749690055847168,
+ "learning_rate": 1.952944623240595e-05,
+ "loss": 0.9297,
+ "step": 4591
+ },
+ {
+ "epoch": 12.512261580381471,
+ "grad_norm": 7.556591987609863,
+ "learning_rate": 1.952917867196606e-05,
+ "loss": 1.1523,
+ "step": 4592
+ },
+ {
+ "epoch": 12.514986376021799,
+ "grad_norm": 5.751350402832031,
+ "learning_rate": 1.9528911037313233e-05,
+ "loss": 0.9449,
+ "step": 4593
+ },
+ {
+ "epoch": 12.517711171662125,
+ "grad_norm": 6.182998180389404,
+ "learning_rate": 1.9528643328449554e-05,
+ "loss": 0.9425,
+ "step": 4594
+ },
+ {
+ "epoch": 12.520435967302452,
+ "grad_norm": 6.5363640785217285,
+ "learning_rate": 1.9528375545377107e-05,
+ "loss": 0.873,
+ "step": 4595
+ },
+ {
+ "epoch": 12.52316076294278,
+ "grad_norm": 8.259293556213379,
+ "learning_rate": 1.9528107688097978e-05,
+ "loss": 0.809,
+ "step": 4596
+ },
+ {
+ "epoch": 12.525885558583106,
+ "grad_norm": 7.481297016143799,
+ "learning_rate": 1.9527839756614253e-05,
+ "loss": 0.9757,
+ "step": 4597
+ },
+ {
+ "epoch": 12.528610354223433,
+ "grad_norm": 7.600178241729736,
+ "learning_rate": 1.952757175092802e-05,
+ "loss": 1.0627,
+ "step": 4598
+ },
+ {
+ "epoch": 12.53133514986376,
+ "grad_norm": 6.583541393280029,
+ "learning_rate": 1.952730367104136e-05,
+ "loss": 0.8335,
+ "step": 4599
+ },
+ {
+ "epoch": 12.534059945504087,
+ "grad_norm": 7.15864896774292,
+ "learning_rate": 1.9527035516956368e-05,
+ "loss": 1.1448,
+ "step": 4600
+ },
+ {
+ "epoch": 12.536784741144414,
+ "grad_norm": 6.342986106872559,
+ "learning_rate": 1.9526767288675134e-05,
+ "loss": 0.8878,
+ "step": 4601
+ },
+ {
+ "epoch": 12.539509536784742,
+ "grad_norm": 7.345736503601074,
+ "learning_rate": 1.952649898619974e-05,
+ "loss": 1.0642,
+ "step": 4602
+ },
+ {
+ "epoch": 12.542234332425068,
+ "grad_norm": 7.477541446685791,
+ "learning_rate": 1.952623060953228e-05,
+ "loss": 1.0187,
+ "step": 4603
+ },
+ {
+ "epoch": 12.544959128065395,
+ "grad_norm": 7.651790142059326,
+ "learning_rate": 1.952596215867484e-05,
+ "loss": 0.868,
+ "step": 4604
+ },
+ {
+ "epoch": 12.547683923705723,
+ "grad_norm": 8.378753662109375,
+ "learning_rate": 1.9525693633629514e-05,
+ "loss": 1.1136,
+ "step": 4605
+ },
+ {
+ "epoch": 12.550408719346049,
+ "grad_norm": 7.696621417999268,
+ "learning_rate": 1.9525425034398394e-05,
+ "loss": 1.0427,
+ "step": 4606
+ },
+ {
+ "epoch": 12.553133514986376,
+ "grad_norm": 6.484177112579346,
+ "learning_rate": 1.952515636098357e-05,
+ "loss": 0.8851,
+ "step": 4607
+ },
+ {
+ "epoch": 12.555858310626704,
+ "grad_norm": 7.780947685241699,
+ "learning_rate": 1.9524887613387133e-05,
+ "loss": 0.9663,
+ "step": 4608
+ },
+ {
+ "epoch": 12.55858310626703,
+ "grad_norm": 7.85284423828125,
+ "learning_rate": 1.9524618791611177e-05,
+ "loss": 0.9666,
+ "step": 4609
+ },
+ {
+ "epoch": 12.561307901907357,
+ "grad_norm": 8.799872398376465,
+ "learning_rate": 1.9524349895657798e-05,
+ "loss": 1.0056,
+ "step": 4610
+ },
+ {
+ "epoch": 12.564032697547685,
+ "grad_norm": 7.083038806915283,
+ "learning_rate": 1.9524080925529093e-05,
+ "loss": 1.1653,
+ "step": 4611
+ },
+ {
+ "epoch": 12.56675749318801,
+ "grad_norm": 9.19692611694336,
+ "learning_rate": 1.9523811881227147e-05,
+ "loss": 1.061,
+ "step": 4612
+ },
+ {
+ "epoch": 12.569482288828338,
+ "grad_norm": 7.265958309173584,
+ "learning_rate": 1.952354276275406e-05,
+ "loss": 0.8646,
+ "step": 4613
+ },
+ {
+ "epoch": 12.572207084468666,
+ "grad_norm": 8.999086380004883,
+ "learning_rate": 1.952327357011193e-05,
+ "loss": 1.0077,
+ "step": 4614
+ },
+ {
+ "epoch": 12.574931880108991,
+ "grad_norm": 9.533391952514648,
+ "learning_rate": 1.9523004303302854e-05,
+ "loss": 1.3274,
+ "step": 4615
+ },
+ {
+ "epoch": 12.577656675749319,
+ "grad_norm": 7.212647914886475,
+ "learning_rate": 1.9522734962328926e-05,
+ "loss": 1.1779,
+ "step": 4616
+ },
+ {
+ "epoch": 12.580381471389646,
+ "grad_norm": 7.341479301452637,
+ "learning_rate": 1.952246554719224e-05,
+ "loss": 1.0322,
+ "step": 4617
+ },
+ {
+ "epoch": 12.583106267029972,
+ "grad_norm": 7.917439937591553,
+ "learning_rate": 1.9522196057894905e-05,
+ "loss": 0.7914,
+ "step": 4618
+ },
+ {
+ "epoch": 12.5858310626703,
+ "grad_norm": 5.908268451690674,
+ "learning_rate": 1.952192649443901e-05,
+ "loss": 0.817,
+ "step": 4619
+ },
+ {
+ "epoch": 12.588555858310627,
+ "grad_norm": 5.751713275909424,
+ "learning_rate": 1.952165685682666e-05,
+ "loss": 1.0133,
+ "step": 4620
+ },
+ {
+ "epoch": 12.591280653950953,
+ "grad_norm": 9.762153625488281,
+ "learning_rate": 1.9521387145059952e-05,
+ "loss": 1.0336,
+ "step": 4621
+ },
+ {
+ "epoch": 12.59400544959128,
+ "grad_norm": 6.750847339630127,
+ "learning_rate": 1.9521117359140985e-05,
+ "loss": 1.1378,
+ "step": 4622
+ },
+ {
+ "epoch": 12.596730245231608,
+ "grad_norm": 6.888245105743408,
+ "learning_rate": 1.952084749907187e-05,
+ "loss": 1.0416,
+ "step": 4623
+ },
+ {
+ "epoch": 12.599455040871934,
+ "grad_norm": 6.726238250732422,
+ "learning_rate": 1.9520577564854696e-05,
+ "loss": 1.0657,
+ "step": 4624
+ },
+ {
+ "epoch": 12.602179836512262,
+ "grad_norm": 6.599887847900391,
+ "learning_rate": 1.952030755649157e-05,
+ "loss": 0.9473,
+ "step": 4625
+ },
+ {
+ "epoch": 12.60490463215259,
+ "grad_norm": 7.329834461212158,
+ "learning_rate": 1.952003747398459e-05,
+ "loss": 0.9773,
+ "step": 4626
+ },
+ {
+ "epoch": 12.607629427792915,
+ "grad_norm": 7.259126663208008,
+ "learning_rate": 1.9519767317335875e-05,
+ "loss": 1.0894,
+ "step": 4627
+ },
+ {
+ "epoch": 12.610354223433243,
+ "grad_norm": 7.521123886108398,
+ "learning_rate": 1.951949708654751e-05,
+ "loss": 1.1246,
+ "step": 4628
+ },
+ {
+ "epoch": 12.61307901907357,
+ "grad_norm": 9.167497634887695,
+ "learning_rate": 1.9519226781621613e-05,
+ "loss": 1.0057,
+ "step": 4629
+ },
+ {
+ "epoch": 12.615803814713896,
+ "grad_norm": 7.006461143493652,
+ "learning_rate": 1.9518956402560282e-05,
+ "loss": 0.9153,
+ "step": 4630
+ },
+ {
+ "epoch": 12.618528610354224,
+ "grad_norm": 7.442938804626465,
+ "learning_rate": 1.9518685949365626e-05,
+ "loss": 1.002,
+ "step": 4631
+ },
+ {
+ "epoch": 12.621253405994551,
+ "grad_norm": 6.07548713684082,
+ "learning_rate": 1.951841542203975e-05,
+ "loss": 0.9915,
+ "step": 4632
+ },
+ {
+ "epoch": 12.623978201634877,
+ "grad_norm": 5.731812477111816,
+ "learning_rate": 1.9518144820584763e-05,
+ "loss": 0.7992,
+ "step": 4633
+ },
+ {
+ "epoch": 12.626702997275205,
+ "grad_norm": 6.99486780166626,
+ "learning_rate": 1.9517874145002768e-05,
+ "loss": 0.9199,
+ "step": 4634
+ },
+ {
+ "epoch": 12.629427792915532,
+ "grad_norm": 6.985575199127197,
+ "learning_rate": 1.9517603395295873e-05,
+ "loss": 0.9974,
+ "step": 4635
+ },
+ {
+ "epoch": 12.632152588555858,
+ "grad_norm": 7.08441686630249,
+ "learning_rate": 1.951733257146619e-05,
+ "loss": 1.0208,
+ "step": 4636
+ },
+ {
+ "epoch": 12.634877384196185,
+ "grad_norm": 6.927825927734375,
+ "learning_rate": 1.9517061673515832e-05,
+ "loss": 0.9764,
+ "step": 4637
+ },
+ {
+ "epoch": 12.637602179836513,
+ "grad_norm": 7.524370193481445,
+ "learning_rate": 1.95167907014469e-05,
+ "loss": 0.8669,
+ "step": 4638
+ },
+ {
+ "epoch": 12.640326975476839,
+ "grad_norm": 7.8481268882751465,
+ "learning_rate": 1.951651965526151e-05,
+ "loss": 0.9241,
+ "step": 4639
+ },
+ {
+ "epoch": 12.643051771117166,
+ "grad_norm": 6.821708679199219,
+ "learning_rate": 1.9516248534961773e-05,
+ "loss": 1.0132,
+ "step": 4640
+ },
+ {
+ "epoch": 12.645776566757494,
+ "grad_norm": 6.4938507080078125,
+ "learning_rate": 1.9515977340549794e-05,
+ "loss": 0.9412,
+ "step": 4641
+ },
+ {
+ "epoch": 12.64850136239782,
+ "grad_norm": 7.857921600341797,
+ "learning_rate": 1.9515706072027694e-05,
+ "loss": 1.1722,
+ "step": 4642
+ },
+ {
+ "epoch": 12.651226158038147,
+ "grad_norm": 8.843948364257812,
+ "learning_rate": 1.9515434729397577e-05,
+ "loss": 0.9553,
+ "step": 4643
+ },
+ {
+ "epoch": 12.653950953678475,
+ "grad_norm": 7.717622756958008,
+ "learning_rate": 1.9515163312661563e-05,
+ "loss": 1.1135,
+ "step": 4644
+ },
+ {
+ "epoch": 12.6566757493188,
+ "grad_norm": 6.566433429718018,
+ "learning_rate": 1.9514891821821768e-05,
+ "loss": 0.9659,
+ "step": 4645
+ },
+ {
+ "epoch": 12.659400544959128,
+ "grad_norm": 6.236119747161865,
+ "learning_rate": 1.9514620256880294e-05,
+ "loss": 1.2078,
+ "step": 4646
+ },
+ {
+ "epoch": 12.662125340599456,
+ "grad_norm": 6.825615406036377,
+ "learning_rate": 1.9514348617839268e-05,
+ "loss": 1.0314,
+ "step": 4647
+ },
+ {
+ "epoch": 12.664850136239782,
+ "grad_norm": 7.122828006744385,
+ "learning_rate": 1.95140769047008e-05,
+ "loss": 1.0638,
+ "step": 4648
+ },
+ {
+ "epoch": 12.66757493188011,
+ "grad_norm": 6.480970859527588,
+ "learning_rate": 1.9513805117467007e-05,
+ "loss": 0.9857,
+ "step": 4649
+ },
+ {
+ "epoch": 12.670299727520437,
+ "grad_norm": 7.273794651031494,
+ "learning_rate": 1.9513533256140004e-05,
+ "loss": 0.8658,
+ "step": 4650
+ },
+ {
+ "epoch": 12.673024523160763,
+ "grad_norm": 7.660243988037109,
+ "learning_rate": 1.9513261320721915e-05,
+ "loss": 0.9846,
+ "step": 4651
+ },
+ {
+ "epoch": 12.67574931880109,
+ "grad_norm": 9.342142105102539,
+ "learning_rate": 1.951298931121485e-05,
+ "loss": 1.0084,
+ "step": 4652
+ },
+ {
+ "epoch": 12.678474114441418,
+ "grad_norm": 6.851946830749512,
+ "learning_rate": 1.951271722762093e-05,
+ "loss": 1.1948,
+ "step": 4653
+ },
+ {
+ "epoch": 12.681198910081743,
+ "grad_norm": 8.495182037353516,
+ "learning_rate": 1.9512445069942275e-05,
+ "loss": 1.0303,
+ "step": 4654
+ },
+ {
+ "epoch": 12.683923705722071,
+ "grad_norm": 6.7690324783325195,
+ "learning_rate": 1.9512172838181005e-05,
+ "loss": 0.9442,
+ "step": 4655
+ },
+ {
+ "epoch": 12.686648501362399,
+ "grad_norm": 9.511191368103027,
+ "learning_rate": 1.9511900532339237e-05,
+ "loss": 1.015,
+ "step": 4656
+ },
+ {
+ "epoch": 12.689373297002724,
+ "grad_norm": 7.090142726898193,
+ "learning_rate": 1.9511628152419098e-05,
+ "loss": 0.9379,
+ "step": 4657
+ },
+ {
+ "epoch": 12.692098092643052,
+ "grad_norm": 8.02755069732666,
+ "learning_rate": 1.9511355698422702e-05,
+ "loss": 0.9957,
+ "step": 4658
+ },
+ {
+ "epoch": 12.69482288828338,
+ "grad_norm": 9.077164649963379,
+ "learning_rate": 1.9511083170352173e-05,
+ "loss": 1.3088,
+ "step": 4659
+ },
+ {
+ "epoch": 12.697547683923705,
+ "grad_norm": 6.869210243225098,
+ "learning_rate": 1.9510810568209634e-05,
+ "loss": 0.792,
+ "step": 4660
+ },
+ {
+ "epoch": 12.700272479564033,
+ "grad_norm": 7.030060768127441,
+ "learning_rate": 1.9510537891997214e-05,
+ "loss": 0.9194,
+ "step": 4661
+ },
+ {
+ "epoch": 12.70299727520436,
+ "grad_norm": 7.466184139251709,
+ "learning_rate": 1.9510265141717024e-05,
+ "loss": 1.0447,
+ "step": 4662
+ },
+ {
+ "epoch": 12.705722070844686,
+ "grad_norm": 7.0918965339660645,
+ "learning_rate": 1.9509992317371196e-05,
+ "loss": 1.0659,
+ "step": 4663
+ },
+ {
+ "epoch": 12.708446866485014,
+ "grad_norm": 5.006387233734131,
+ "learning_rate": 1.9509719418961858e-05,
+ "loss": 0.874,
+ "step": 4664
+ },
+ {
+ "epoch": 12.711171662125341,
+ "grad_norm": 7.380950927734375,
+ "learning_rate": 1.9509446446491127e-05,
+ "loss": 0.8652,
+ "step": 4665
+ },
+ {
+ "epoch": 12.713896457765667,
+ "grad_norm": 6.832798004150391,
+ "learning_rate": 1.9509173399961137e-05,
+ "loss": 1.1395,
+ "step": 4666
+ },
+ {
+ "epoch": 12.716621253405995,
+ "grad_norm": 10.570134162902832,
+ "learning_rate": 1.950890027937401e-05,
+ "loss": 0.9299,
+ "step": 4667
+ },
+ {
+ "epoch": 12.719346049046322,
+ "grad_norm": 6.010441303253174,
+ "learning_rate": 1.9508627084731872e-05,
+ "loss": 0.9895,
+ "step": 4668
+ },
+ {
+ "epoch": 12.722070844686648,
+ "grad_norm": 6.997371196746826,
+ "learning_rate": 1.9508353816036853e-05,
+ "loss": 0.9886,
+ "step": 4669
+ },
+ {
+ "epoch": 12.724795640326976,
+ "grad_norm": 6.087446689605713,
+ "learning_rate": 1.950808047329108e-05,
+ "loss": 1.0167,
+ "step": 4670
+ },
+ {
+ "epoch": 12.727520435967303,
+ "grad_norm": 7.723606586456299,
+ "learning_rate": 1.9507807056496683e-05,
+ "loss": 0.9147,
+ "step": 4671
+ },
+ {
+ "epoch": 12.730245231607629,
+ "grad_norm": 8.377662658691406,
+ "learning_rate": 1.950753356565579e-05,
+ "loss": 1.0024,
+ "step": 4672
+ },
+ {
+ "epoch": 12.732970027247957,
+ "grad_norm": 7.017062187194824,
+ "learning_rate": 1.950726000077053e-05,
+ "loss": 1.3787,
+ "step": 4673
+ },
+ {
+ "epoch": 12.735694822888284,
+ "grad_norm": 6.854485511779785,
+ "learning_rate": 1.950698636184304e-05,
+ "loss": 1.1415,
+ "step": 4674
+ },
+ {
+ "epoch": 12.73841961852861,
+ "grad_norm": 7.179381370544434,
+ "learning_rate": 1.9506712648875444e-05,
+ "loss": 1.135,
+ "step": 4675
+ },
+ {
+ "epoch": 12.741144414168938,
+ "grad_norm": 6.765000343322754,
+ "learning_rate": 1.9506438861869875e-05,
+ "loss": 1.0637,
+ "step": 4676
+ },
+ {
+ "epoch": 12.743869209809265,
+ "grad_norm": 14.968036651611328,
+ "learning_rate": 1.9506165000828467e-05,
+ "loss": 0.8574,
+ "step": 4677
+ },
+ {
+ "epoch": 12.746594005449591,
+ "grad_norm": 6.567181587219238,
+ "learning_rate": 1.950589106575335e-05,
+ "loss": 1.0569,
+ "step": 4678
+ },
+ {
+ "epoch": 12.749318801089919,
+ "grad_norm": 8.405080795288086,
+ "learning_rate": 1.9505617056646658e-05,
+ "loss": 1.1519,
+ "step": 4679
+ },
+ {
+ "epoch": 12.752043596730246,
+ "grad_norm": 6.418135643005371,
+ "learning_rate": 1.950534297351053e-05,
+ "loss": 0.9833,
+ "step": 4680
+ },
+ {
+ "epoch": 12.754768392370572,
+ "grad_norm": 5.3978095054626465,
+ "learning_rate": 1.9505068816347095e-05,
+ "loss": 1.0806,
+ "step": 4681
+ },
+ {
+ "epoch": 12.7574931880109,
+ "grad_norm": 8.308835983276367,
+ "learning_rate": 1.9504794585158492e-05,
+ "loss": 0.8489,
+ "step": 4682
+ },
+ {
+ "epoch": 12.760217983651227,
+ "grad_norm": 9.189422607421875,
+ "learning_rate": 1.9504520279946853e-05,
+ "loss": 0.9236,
+ "step": 4683
+ },
+ {
+ "epoch": 12.762942779291553,
+ "grad_norm": 7.211663246154785,
+ "learning_rate": 1.9504245900714313e-05,
+ "loss": 1.0786,
+ "step": 4684
+ },
+ {
+ "epoch": 12.76566757493188,
+ "grad_norm": 5.411872863769531,
+ "learning_rate": 1.9503971447463016e-05,
+ "loss": 0.954,
+ "step": 4685
+ },
+ {
+ "epoch": 12.768392370572208,
+ "grad_norm": 7.783181667327881,
+ "learning_rate": 1.9503696920195096e-05,
+ "loss": 0.99,
+ "step": 4686
+ },
+ {
+ "epoch": 12.771117166212534,
+ "grad_norm": 6.2306437492370605,
+ "learning_rate": 1.950342231891269e-05,
+ "loss": 1.0286,
+ "step": 4687
+ },
+ {
+ "epoch": 12.773841961852861,
+ "grad_norm": 6.862693786621094,
+ "learning_rate": 1.9503147643617932e-05,
+ "loss": 0.9723,
+ "step": 4688
+ },
+ {
+ "epoch": 12.776566757493189,
+ "grad_norm": 8.049107551574707,
+ "learning_rate": 1.950287289431297e-05,
+ "loss": 1.1086,
+ "step": 4689
+ },
+ {
+ "epoch": 12.779291553133515,
+ "grad_norm": 7.628965377807617,
+ "learning_rate": 1.950259807099994e-05,
+ "loss": 0.8241,
+ "step": 4690
+ },
+ {
+ "epoch": 12.782016348773842,
+ "grad_norm": 7.4149627685546875,
+ "learning_rate": 1.9502323173680983e-05,
+ "loss": 0.9404,
+ "step": 4691
+ },
+ {
+ "epoch": 12.78474114441417,
+ "grad_norm": 6.816451072692871,
+ "learning_rate": 1.9502048202358236e-05,
+ "loss": 0.8033,
+ "step": 4692
+ },
+ {
+ "epoch": 12.787465940054496,
+ "grad_norm": 8.501646041870117,
+ "learning_rate": 1.9501773157033847e-05,
+ "loss": 0.8751,
+ "step": 4693
+ },
+ {
+ "epoch": 12.790190735694823,
+ "grad_norm": 6.936130523681641,
+ "learning_rate": 1.9501498037709948e-05,
+ "loss": 0.7798,
+ "step": 4694
+ },
+ {
+ "epoch": 12.79291553133515,
+ "grad_norm": 7.984213829040527,
+ "learning_rate": 1.9501222844388696e-05,
+ "loss": 0.9736,
+ "step": 4695
+ },
+ {
+ "epoch": 12.795640326975477,
+ "grad_norm": 6.261298179626465,
+ "learning_rate": 1.950094757707222e-05,
+ "loss": 1.0789,
+ "step": 4696
+ },
+ {
+ "epoch": 12.798365122615804,
+ "grad_norm": 7.020403861999512,
+ "learning_rate": 1.950067223576267e-05,
+ "loss": 1.0449,
+ "step": 4697
+ },
+ {
+ "epoch": 12.80108991825613,
+ "grad_norm": 6.254876136779785,
+ "learning_rate": 1.9500396820462194e-05,
+ "loss": 0.9968,
+ "step": 4698
+ },
+ {
+ "epoch": 12.803814713896458,
+ "grad_norm": 7.256025314331055,
+ "learning_rate": 1.9500121331172936e-05,
+ "loss": 0.8909,
+ "step": 4699
+ },
+ {
+ "epoch": 12.806539509536785,
+ "grad_norm": 9.703126907348633,
+ "learning_rate": 1.9499845767897033e-05,
+ "loss": 1.2249,
+ "step": 4700
+ },
+ {
+ "epoch": 12.809264305177111,
+ "grad_norm": 6.98081636428833,
+ "learning_rate": 1.9499570130636638e-05,
+ "loss": 0.8525,
+ "step": 4701
+ },
+ {
+ "epoch": 12.811989100817438,
+ "grad_norm": 7.822127342224121,
+ "learning_rate": 1.9499294419393898e-05,
+ "loss": 0.9948,
+ "step": 4702
+ },
+ {
+ "epoch": 12.814713896457766,
+ "grad_norm": 6.485651969909668,
+ "learning_rate": 1.949901863417096e-05,
+ "loss": 0.9397,
+ "step": 4703
+ },
+ {
+ "epoch": 12.817438692098092,
+ "grad_norm": 6.382424354553223,
+ "learning_rate": 1.9498742774969968e-05,
+ "loss": 0.8655,
+ "step": 4704
+ },
+ {
+ "epoch": 12.82016348773842,
+ "grad_norm": 7.07916259765625,
+ "learning_rate": 1.9498466841793073e-05,
+ "loss": 1.3171,
+ "step": 4705
+ },
+ {
+ "epoch": 12.822888283378747,
+ "grad_norm": 6.161258220672607,
+ "learning_rate": 1.9498190834642426e-05,
+ "loss": 0.8739,
+ "step": 4706
+ },
+ {
+ "epoch": 12.825613079019073,
+ "grad_norm": 7.156404972076416,
+ "learning_rate": 1.9497914753520175e-05,
+ "loss": 1.126,
+ "step": 4707
+ },
+ {
+ "epoch": 12.8283378746594,
+ "grad_norm": 7.689642429351807,
+ "learning_rate": 1.9497638598428464e-05,
+ "loss": 0.9779,
+ "step": 4708
+ },
+ {
+ "epoch": 12.831062670299728,
+ "grad_norm": 8.414322853088379,
+ "learning_rate": 1.9497362369369456e-05,
+ "loss": 0.9517,
+ "step": 4709
+ },
+ {
+ "epoch": 12.833787465940054,
+ "grad_norm": 7.499088764190674,
+ "learning_rate": 1.949708606634529e-05,
+ "loss": 1.2542,
+ "step": 4710
+ },
+ {
+ "epoch": 12.836512261580381,
+ "grad_norm": 6.543384075164795,
+ "learning_rate": 1.9496809689358127e-05,
+ "loss": 0.9731,
+ "step": 4711
+ },
+ {
+ "epoch": 12.839237057220709,
+ "grad_norm": 6.061197280883789,
+ "learning_rate": 1.9496533238410117e-05,
+ "loss": 0.9868,
+ "step": 4712
+ },
+ {
+ "epoch": 12.841961852861035,
+ "grad_norm": 6.966991901397705,
+ "learning_rate": 1.949625671350341e-05,
+ "loss": 1.0381,
+ "step": 4713
+ },
+ {
+ "epoch": 12.844686648501362,
+ "grad_norm": 7.603959083557129,
+ "learning_rate": 1.949598011464016e-05,
+ "loss": 1.1046,
+ "step": 4714
+ },
+ {
+ "epoch": 12.84741144414169,
+ "grad_norm": 6.7787933349609375,
+ "learning_rate": 1.949570344182252e-05,
+ "loss": 1.0684,
+ "step": 4715
+ },
+ {
+ "epoch": 12.850136239782016,
+ "grad_norm": 5.998163223266602,
+ "learning_rate": 1.9495426695052652e-05,
+ "loss": 1.0054,
+ "step": 4716
+ },
+ {
+ "epoch": 12.852861035422343,
+ "grad_norm": 6.986926078796387,
+ "learning_rate": 1.9495149874332707e-05,
+ "loss": 0.8761,
+ "step": 4717
+ },
+ {
+ "epoch": 12.85558583106267,
+ "grad_norm": 6.750517845153809,
+ "learning_rate": 1.9494872979664838e-05,
+ "loss": 0.9871,
+ "step": 4718
+ },
+ {
+ "epoch": 12.858310626702997,
+ "grad_norm": 6.614873886108398,
+ "learning_rate": 1.9494596011051205e-05,
+ "loss": 0.8934,
+ "step": 4719
+ },
+ {
+ "epoch": 12.861035422343324,
+ "grad_norm": 7.81092643737793,
+ "learning_rate": 1.9494318968493965e-05,
+ "loss": 0.9445,
+ "step": 4720
+ },
+ {
+ "epoch": 12.863760217983652,
+ "grad_norm": 7.475430965423584,
+ "learning_rate": 1.949404185199527e-05,
+ "loss": 0.9026,
+ "step": 4721
+ },
+ {
+ "epoch": 12.866485013623977,
+ "grad_norm": 5.888453483581543,
+ "learning_rate": 1.9493764661557287e-05,
+ "loss": 0.9146,
+ "step": 4722
+ },
+ {
+ "epoch": 12.869209809264305,
+ "grad_norm": 6.414893627166748,
+ "learning_rate": 1.949348739718217e-05,
+ "loss": 1.1019,
+ "step": 4723
+ },
+ {
+ "epoch": 12.871934604904633,
+ "grad_norm": 7.385310173034668,
+ "learning_rate": 1.9493210058872075e-05,
+ "loss": 1.3055,
+ "step": 4724
+ },
+ {
+ "epoch": 12.874659400544958,
+ "grad_norm": 6.49179744720459,
+ "learning_rate": 1.949293264662917e-05,
+ "loss": 0.8679,
+ "step": 4725
+ },
+ {
+ "epoch": 12.877384196185286,
+ "grad_norm": 6.547841548919678,
+ "learning_rate": 1.9492655160455607e-05,
+ "loss": 0.873,
+ "step": 4726
+ },
+ {
+ "epoch": 12.880108991825614,
+ "grad_norm": 7.425289154052734,
+ "learning_rate": 1.9492377600353555e-05,
+ "loss": 0.9902,
+ "step": 4727
+ },
+ {
+ "epoch": 12.88283378746594,
+ "grad_norm": 8.017386436462402,
+ "learning_rate": 1.9492099966325168e-05,
+ "loss": 0.9097,
+ "step": 4728
+ },
+ {
+ "epoch": 12.885558583106267,
+ "grad_norm": 7.3176140785217285,
+ "learning_rate": 1.9491822258372614e-05,
+ "loss": 1.0709,
+ "step": 4729
+ },
+ {
+ "epoch": 12.888283378746594,
+ "grad_norm": 7.717101097106934,
+ "learning_rate": 1.9491544476498054e-05,
+ "loss": 1.05,
+ "step": 4730
+ },
+ {
+ "epoch": 12.89100817438692,
+ "grad_norm": 6.921051979064941,
+ "learning_rate": 1.949126662070365e-05,
+ "loss": 0.938,
+ "step": 4731
+ },
+ {
+ "epoch": 12.893732970027248,
+ "grad_norm": 6.796942710876465,
+ "learning_rate": 1.949098869099157e-05,
+ "loss": 1.0991,
+ "step": 4732
+ },
+ {
+ "epoch": 12.896457765667575,
+ "grad_norm": 6.868369102478027,
+ "learning_rate": 1.949071068736397e-05,
+ "loss": 1.0172,
+ "step": 4733
+ },
+ {
+ "epoch": 12.899182561307901,
+ "grad_norm": 6.994481086730957,
+ "learning_rate": 1.9490432609823027e-05,
+ "loss": 1.0178,
+ "step": 4734
+ },
+ {
+ "epoch": 12.901907356948229,
+ "grad_norm": 6.877704620361328,
+ "learning_rate": 1.94901544583709e-05,
+ "loss": 1.0311,
+ "step": 4735
+ },
+ {
+ "epoch": 12.904632152588556,
+ "grad_norm": 7.139933109283447,
+ "learning_rate": 1.9489876233009747e-05,
+ "loss": 1.187,
+ "step": 4736
+ },
+ {
+ "epoch": 12.907356948228882,
+ "grad_norm": 6.465014934539795,
+ "learning_rate": 1.9489597933741754e-05,
+ "loss": 0.9951,
+ "step": 4737
+ },
+ {
+ "epoch": 12.91008174386921,
+ "grad_norm": 7.957826137542725,
+ "learning_rate": 1.948931956056907e-05,
+ "loss": 1.0706,
+ "step": 4738
+ },
+ {
+ "epoch": 12.912806539509537,
+ "grad_norm": 6.403878211975098,
+ "learning_rate": 1.9489041113493875e-05,
+ "loss": 1.0649,
+ "step": 4739
+ },
+ {
+ "epoch": 12.915531335149863,
+ "grad_norm": 6.253860950469971,
+ "learning_rate": 1.9488762592518332e-05,
+ "loss": 0.9141,
+ "step": 4740
+ },
+ {
+ "epoch": 12.91825613079019,
+ "grad_norm": 6.656635761260986,
+ "learning_rate": 1.9488483997644613e-05,
+ "loss": 0.9078,
+ "step": 4741
+ },
+ {
+ "epoch": 12.920980926430518,
+ "grad_norm": 6.91517972946167,
+ "learning_rate": 1.948820532887488e-05,
+ "loss": 1.0459,
+ "step": 4742
+ },
+ {
+ "epoch": 12.923705722070844,
+ "grad_norm": 7.206698894500732,
+ "learning_rate": 1.9487926586211314e-05,
+ "loss": 0.9363,
+ "step": 4743
+ },
+ {
+ "epoch": 12.926430517711172,
+ "grad_norm": 5.817498207092285,
+ "learning_rate": 1.9487647769656082e-05,
+ "loss": 1.1816,
+ "step": 4744
+ },
+ {
+ "epoch": 12.9291553133515,
+ "grad_norm": 7.4043707847595215,
+ "learning_rate": 1.948736887921135e-05,
+ "loss": 1.025,
+ "step": 4745
+ },
+ {
+ "epoch": 12.931880108991825,
+ "grad_norm": 7.161343574523926,
+ "learning_rate": 1.94870899148793e-05,
+ "loss": 1.1104,
+ "step": 4746
+ },
+ {
+ "epoch": 12.934604904632153,
+ "grad_norm": 98.30435943603516,
+ "learning_rate": 1.9486810876662097e-05,
+ "loss": 0.9248,
+ "step": 4747
+ },
+ {
+ "epoch": 12.93732970027248,
+ "grad_norm": 7.17219352722168,
+ "learning_rate": 1.9486531764561915e-05,
+ "loss": 1.0955,
+ "step": 4748
+ },
+ {
+ "epoch": 12.940054495912806,
+ "grad_norm": 7.5102152824401855,
+ "learning_rate": 1.948625257858093e-05,
+ "loss": 0.9966,
+ "step": 4749
+ },
+ {
+ "epoch": 12.942779291553133,
+ "grad_norm": 7.709928035736084,
+ "learning_rate": 1.9485973318721316e-05,
+ "loss": 1.03,
+ "step": 4750
+ },
+ {
+ "epoch": 12.945504087193461,
+ "grad_norm": 6.321061611175537,
+ "learning_rate": 1.9485693984985246e-05,
+ "loss": 0.9944,
+ "step": 4751
+ },
+ {
+ "epoch": 12.948228882833787,
+ "grad_norm": 11.503897666931152,
+ "learning_rate": 1.9485414577374896e-05,
+ "loss": 0.8779,
+ "step": 4752
+ },
+ {
+ "epoch": 12.950953678474114,
+ "grad_norm": 8.628007888793945,
+ "learning_rate": 1.9485135095892444e-05,
+ "loss": 1.0074,
+ "step": 4753
+ },
+ {
+ "epoch": 12.953678474114442,
+ "grad_norm": 7.065698623657227,
+ "learning_rate": 1.9484855540540063e-05,
+ "loss": 0.9706,
+ "step": 4754
+ },
+ {
+ "epoch": 12.956403269754768,
+ "grad_norm": 9.52556037902832,
+ "learning_rate": 1.9484575911319934e-05,
+ "loss": 1.083,
+ "step": 4755
+ },
+ {
+ "epoch": 12.959128065395095,
+ "grad_norm": 8.182491302490234,
+ "learning_rate": 1.9484296208234232e-05,
+ "loss": 1.2278,
+ "step": 4756
+ },
+ {
+ "epoch": 12.961852861035423,
+ "grad_norm": 9.490069389343262,
+ "learning_rate": 1.9484016431285137e-05,
+ "loss": 1.0276,
+ "step": 4757
+ },
+ {
+ "epoch": 12.964577656675749,
+ "grad_norm": 9.629108428955078,
+ "learning_rate": 1.948373658047483e-05,
+ "loss": 0.9907,
+ "step": 4758
+ },
+ {
+ "epoch": 12.967302452316076,
+ "grad_norm": 8.475467681884766,
+ "learning_rate": 1.9483456655805485e-05,
+ "loss": 1.1726,
+ "step": 4759
+ },
+ {
+ "epoch": 12.970027247956404,
+ "grad_norm": 9.778740882873535,
+ "learning_rate": 1.9483176657279282e-05,
+ "loss": 1.2473,
+ "step": 4760
+ },
+ {
+ "epoch": 12.97275204359673,
+ "grad_norm": 7.9676432609558105,
+ "learning_rate": 1.948289658489841e-05,
+ "loss": 1.0681,
+ "step": 4761
+ },
+ {
+ "epoch": 12.975476839237057,
+ "grad_norm": 11.81965446472168,
+ "learning_rate": 1.9482616438665037e-05,
+ "loss": 1.2407,
+ "step": 4762
+ },
+ {
+ "epoch": 12.978201634877385,
+ "grad_norm": 10.506232261657715,
+ "learning_rate": 1.9482336218581358e-05,
+ "loss": 1.193,
+ "step": 4763
+ },
+ {
+ "epoch": 12.98092643051771,
+ "grad_norm": 8.289572715759277,
+ "learning_rate": 1.9482055924649552e-05,
+ "loss": 1.1055,
+ "step": 4764
+ },
+ {
+ "epoch": 12.983651226158038,
+ "grad_norm": 6.4512152671813965,
+ "learning_rate": 1.9481775556871795e-05,
+ "loss": 1.0283,
+ "step": 4765
+ },
+ {
+ "epoch": 12.986376021798366,
+ "grad_norm": 7.259413719177246,
+ "learning_rate": 1.9481495115250277e-05,
+ "loss": 0.9795,
+ "step": 4766
+ },
+ {
+ "epoch": 12.989100817438691,
+ "grad_norm": 11.75411605834961,
+ "learning_rate": 1.9481214599787178e-05,
+ "loss": 1.0775,
+ "step": 4767
+ },
+ {
+ "epoch": 12.991825613079019,
+ "grad_norm": 7.595376491546631,
+ "learning_rate": 1.948093401048469e-05,
+ "loss": 1.0232,
+ "step": 4768
+ },
+ {
+ "epoch": 12.994550408719347,
+ "grad_norm": 8.230860710144043,
+ "learning_rate": 1.9480653347344992e-05,
+ "loss": 1.1561,
+ "step": 4769
+ },
+ {
+ "epoch": 12.997275204359672,
+ "grad_norm": 9.399571418762207,
+ "learning_rate": 1.9480372610370265e-05,
+ "loss": 0.8331,
+ "step": 4770
+ },
+ {
+ "epoch": 13.0,
+ "grad_norm": 9.44799518585205,
+ "learning_rate": 1.9480091799562706e-05,
+ "loss": 0.9374,
+ "step": 4771
+ },
+ {
+ "epoch": 13.002724795640328,
+ "grad_norm": 8.244596481323242,
+ "learning_rate": 1.9479810914924498e-05,
+ "loss": 0.7963,
+ "step": 4772
+ },
+ {
+ "epoch": 13.005449591280653,
+ "grad_norm": 19.829164505004883,
+ "learning_rate": 1.9479529956457825e-05,
+ "loss": 0.912,
+ "step": 4773
+ },
+ {
+ "epoch": 13.008174386920981,
+ "grad_norm": 7.833473205566406,
+ "learning_rate": 1.9479248924164878e-05,
+ "loss": 1.0054,
+ "step": 4774
+ },
+ {
+ "epoch": 13.010899182561309,
+ "grad_norm": 9.74729061126709,
+ "learning_rate": 1.9478967818047846e-05,
+ "loss": 0.948,
+ "step": 4775
+ },
+ {
+ "epoch": 13.013623978201634,
+ "grad_norm": 7.318088054656982,
+ "learning_rate": 1.947868663810892e-05,
+ "loss": 0.863,
+ "step": 4776
+ },
+ {
+ "epoch": 13.016348773841962,
+ "grad_norm": 10.633477210998535,
+ "learning_rate": 1.9478405384350283e-05,
+ "loss": 0.8845,
+ "step": 4777
+ },
+ {
+ "epoch": 13.01907356948229,
+ "grad_norm": 6.630046367645264,
+ "learning_rate": 1.9478124056774133e-05,
+ "loss": 0.8929,
+ "step": 4778
+ },
+ {
+ "epoch": 13.021798365122615,
+ "grad_norm": 8.609410285949707,
+ "learning_rate": 1.9477842655382656e-05,
+ "loss": 0.9517,
+ "step": 4779
+ },
+ {
+ "epoch": 13.024523160762943,
+ "grad_norm": 19.68306541442871,
+ "learning_rate": 1.9477561180178048e-05,
+ "loss": 1.0657,
+ "step": 4780
+ },
+ {
+ "epoch": 13.02724795640327,
+ "grad_norm": 7.244594573974609,
+ "learning_rate": 1.9477279631162494e-05,
+ "loss": 0.8766,
+ "step": 4781
+ },
+ {
+ "epoch": 13.029972752043596,
+ "grad_norm": 8.661189079284668,
+ "learning_rate": 1.9476998008338197e-05,
+ "loss": 0.8771,
+ "step": 4782
+ },
+ {
+ "epoch": 13.032697547683924,
+ "grad_norm": 10.965076446533203,
+ "learning_rate": 1.9476716311707342e-05,
+ "loss": 1.1462,
+ "step": 4783
+ },
+ {
+ "epoch": 13.035422343324251,
+ "grad_norm": 20.85923957824707,
+ "learning_rate": 1.9476434541272127e-05,
+ "loss": 0.9294,
+ "step": 4784
+ },
+ {
+ "epoch": 13.038147138964577,
+ "grad_norm": 8.536860466003418,
+ "learning_rate": 1.9476152697034743e-05,
+ "loss": 0.9277,
+ "step": 4785
+ },
+ {
+ "epoch": 13.040871934604905,
+ "grad_norm": 9.493371963500977,
+ "learning_rate": 1.9475870778997386e-05,
+ "loss": 0.978,
+ "step": 4786
+ },
+ {
+ "epoch": 13.043596730245232,
+ "grad_norm": 6.416027545928955,
+ "learning_rate": 1.947558878716225e-05,
+ "loss": 0.99,
+ "step": 4787
+ },
+ {
+ "epoch": 13.046321525885558,
+ "grad_norm": 8.1671781539917,
+ "learning_rate": 1.947530672153154e-05,
+ "loss": 1.0199,
+ "step": 4788
+ },
+ {
+ "epoch": 13.049046321525886,
+ "grad_norm": 7.836258888244629,
+ "learning_rate": 1.9475024582107443e-05,
+ "loss": 1.027,
+ "step": 4789
+ },
+ {
+ "epoch": 13.051771117166213,
+ "grad_norm": 8.518571853637695,
+ "learning_rate": 1.9474742368892162e-05,
+ "loss": 0.9796,
+ "step": 4790
+ },
+ {
+ "epoch": 13.054495912806539,
+ "grad_norm": 6.77135705947876,
+ "learning_rate": 1.9474460081887888e-05,
+ "loss": 1.0402,
+ "step": 4791
+ },
+ {
+ "epoch": 13.057220708446867,
+ "grad_norm": 7.403370380401611,
+ "learning_rate": 1.9474177721096825e-05,
+ "loss": 0.9595,
+ "step": 4792
+ },
+ {
+ "epoch": 13.059945504087194,
+ "grad_norm": 13.965553283691406,
+ "learning_rate": 1.9473895286521173e-05,
+ "loss": 0.8463,
+ "step": 4793
+ },
+ {
+ "epoch": 13.06267029972752,
+ "grad_norm": 6.673770904541016,
+ "learning_rate": 1.9473612778163126e-05,
+ "loss": 0.8306,
+ "step": 4794
+ },
+ {
+ "epoch": 13.065395095367847,
+ "grad_norm": 6.716458320617676,
+ "learning_rate": 1.9473330196024892e-05,
+ "loss": 0.9636,
+ "step": 4795
+ },
+ {
+ "epoch": 13.068119891008175,
+ "grad_norm": 9.071009635925293,
+ "learning_rate": 1.947304754010866e-05,
+ "loss": 0.8844,
+ "step": 4796
+ },
+ {
+ "epoch": 13.0708446866485,
+ "grad_norm": 8.002140045166016,
+ "learning_rate": 1.9472764810416647e-05,
+ "loss": 0.8833,
+ "step": 4797
+ },
+ {
+ "epoch": 13.073569482288828,
+ "grad_norm": 6.863930702209473,
+ "learning_rate": 1.9472482006951045e-05,
+ "loss": 0.8179,
+ "step": 4798
+ },
+ {
+ "epoch": 13.076294277929156,
+ "grad_norm": 9.530876159667969,
+ "learning_rate": 1.9472199129714053e-05,
+ "loss": 0.8154,
+ "step": 4799
+ },
+ {
+ "epoch": 13.079019073569482,
+ "grad_norm": 7.177833557128906,
+ "learning_rate": 1.947191617870788e-05,
+ "loss": 0.9362,
+ "step": 4800
+ },
+ {
+ "epoch": 13.08174386920981,
+ "grad_norm": 9.34472370147705,
+ "learning_rate": 1.9471633153934735e-05,
+ "loss": 0.8334,
+ "step": 4801
+ },
+ {
+ "epoch": 13.084468664850137,
+ "grad_norm": 8.261250495910645,
+ "learning_rate": 1.9471350055396813e-05,
+ "loss": 0.8359,
+ "step": 4802
+ },
+ {
+ "epoch": 13.087193460490463,
+ "grad_norm": 6.949221134185791,
+ "learning_rate": 1.9471066883096318e-05,
+ "loss": 0.9225,
+ "step": 4803
+ },
+ {
+ "epoch": 13.08991825613079,
+ "grad_norm": 8.423523902893066,
+ "learning_rate": 1.947078363703546e-05,
+ "loss": 0.9609,
+ "step": 4804
+ },
+ {
+ "epoch": 13.092643051771118,
+ "grad_norm": 10.226131439208984,
+ "learning_rate": 1.9470500317216447e-05,
+ "loss": 0.9315,
+ "step": 4805
+ },
+ {
+ "epoch": 13.095367847411444,
+ "grad_norm": 6.992786407470703,
+ "learning_rate": 1.947021692364148e-05,
+ "loss": 0.8628,
+ "step": 4806
+ },
+ {
+ "epoch": 13.098092643051771,
+ "grad_norm": 7.6111159324646,
+ "learning_rate": 1.9469933456312768e-05,
+ "loss": 0.9386,
+ "step": 4807
+ },
+ {
+ "epoch": 13.100817438692099,
+ "grad_norm": 6.257993698120117,
+ "learning_rate": 1.946964991523252e-05,
+ "loss": 0.9611,
+ "step": 4808
+ },
+ {
+ "epoch": 13.103542234332425,
+ "grad_norm": 12.221894264221191,
+ "learning_rate": 1.9469366300402943e-05,
+ "loss": 0.7778,
+ "step": 4809
+ },
+ {
+ "epoch": 13.106267029972752,
+ "grad_norm": 7.064905643463135,
+ "learning_rate": 1.9469082611826247e-05,
+ "loss": 0.9811,
+ "step": 4810
+ },
+ {
+ "epoch": 13.10899182561308,
+ "grad_norm": 8.026780128479004,
+ "learning_rate": 1.9468798849504637e-05,
+ "loss": 0.9771,
+ "step": 4811
+ },
+ {
+ "epoch": 13.111716621253406,
+ "grad_norm": 8.455852508544922,
+ "learning_rate": 1.946851501344033e-05,
+ "loss": 0.9896,
+ "step": 4812
+ },
+ {
+ "epoch": 13.114441416893733,
+ "grad_norm": 6.430542469024658,
+ "learning_rate": 1.946823110363553e-05,
+ "loss": 0.8081,
+ "step": 4813
+ },
+ {
+ "epoch": 13.11716621253406,
+ "grad_norm": 8.00733757019043,
+ "learning_rate": 1.9467947120092456e-05,
+ "loss": 1.0397,
+ "step": 4814
+ },
+ {
+ "epoch": 13.119891008174386,
+ "grad_norm": 8.320761680603027,
+ "learning_rate": 1.946766306281331e-05,
+ "loss": 1.1177,
+ "step": 4815
+ },
+ {
+ "epoch": 13.122615803814714,
+ "grad_norm": 7.327835559844971,
+ "learning_rate": 1.9467378931800307e-05,
+ "loss": 1.0721,
+ "step": 4816
+ },
+ {
+ "epoch": 13.125340599455042,
+ "grad_norm": 7.292349338531494,
+ "learning_rate": 1.9467094727055666e-05,
+ "loss": 0.8448,
+ "step": 4817
+ },
+ {
+ "epoch": 13.128065395095367,
+ "grad_norm": 7.833173751831055,
+ "learning_rate": 1.9466810448581593e-05,
+ "loss": 1.1456,
+ "step": 4818
+ },
+ {
+ "epoch": 13.130790190735695,
+ "grad_norm": 7.434957504272461,
+ "learning_rate": 1.9466526096380306e-05,
+ "loss": 0.9426,
+ "step": 4819
+ },
+ {
+ "epoch": 13.133514986376023,
+ "grad_norm": 7.383454322814941,
+ "learning_rate": 1.946624167045402e-05,
+ "loss": 0.9487,
+ "step": 4820
+ },
+ {
+ "epoch": 13.136239782016348,
+ "grad_norm": 6.504461765289307,
+ "learning_rate": 1.9465957170804945e-05,
+ "loss": 0.7384,
+ "step": 4821
+ },
+ {
+ "epoch": 13.138964577656676,
+ "grad_norm": 7.401673316955566,
+ "learning_rate": 1.9465672597435305e-05,
+ "loss": 0.8757,
+ "step": 4822
+ },
+ {
+ "epoch": 13.141689373297003,
+ "grad_norm": 7.7824506759643555,
+ "learning_rate": 1.946538795034731e-05,
+ "loss": 0.9836,
+ "step": 4823
+ },
+ {
+ "epoch": 13.14441416893733,
+ "grad_norm": 6.3840556144714355,
+ "learning_rate": 1.9465103229543177e-05,
+ "loss": 0.7789,
+ "step": 4824
+ },
+ {
+ "epoch": 13.147138964577657,
+ "grad_norm": 7.301362991333008,
+ "learning_rate": 1.9464818435025123e-05,
+ "loss": 0.9507,
+ "step": 4825
+ },
+ {
+ "epoch": 13.149863760217984,
+ "grad_norm": 8.0537748336792,
+ "learning_rate": 1.946453356679537e-05,
+ "loss": 1.2024,
+ "step": 4826
+ },
+ {
+ "epoch": 13.15258855585831,
+ "grad_norm": 6.651594638824463,
+ "learning_rate": 1.946424862485613e-05,
+ "loss": 0.9006,
+ "step": 4827
+ },
+ {
+ "epoch": 13.155313351498638,
+ "grad_norm": 8.839438438415527,
+ "learning_rate": 1.9463963609209634e-05,
+ "loss": 0.8715,
+ "step": 4828
+ },
+ {
+ "epoch": 13.158038147138965,
+ "grad_norm": 7.252025604248047,
+ "learning_rate": 1.946367851985809e-05,
+ "loss": 0.9246,
+ "step": 4829
+ },
+ {
+ "epoch": 13.160762942779291,
+ "grad_norm": 7.809706687927246,
+ "learning_rate": 1.946339335680372e-05,
+ "loss": 1.0353,
+ "step": 4830
+ },
+ {
+ "epoch": 13.163487738419619,
+ "grad_norm": 7.083587169647217,
+ "learning_rate": 1.946310812004875e-05,
+ "loss": 0.9889,
+ "step": 4831
+ },
+ {
+ "epoch": 13.166212534059946,
+ "grad_norm": 6.538549900054932,
+ "learning_rate": 1.94628228095954e-05,
+ "loss": 0.8677,
+ "step": 4832
+ },
+ {
+ "epoch": 13.168937329700272,
+ "grad_norm": 8.11311149597168,
+ "learning_rate": 1.9462537425445886e-05,
+ "loss": 1.0907,
+ "step": 4833
+ },
+ {
+ "epoch": 13.1716621253406,
+ "grad_norm": 6.37860107421875,
+ "learning_rate": 1.946225196760244e-05,
+ "loss": 0.9545,
+ "step": 4834
+ },
+ {
+ "epoch": 13.174386920980927,
+ "grad_norm": 8.381884574890137,
+ "learning_rate": 1.9461966436067277e-05,
+ "loss": 1.0531,
+ "step": 4835
+ },
+ {
+ "epoch": 13.177111716621253,
+ "grad_norm": 6.1375555992126465,
+ "learning_rate": 1.9461680830842623e-05,
+ "loss": 0.9749,
+ "step": 4836
+ },
+ {
+ "epoch": 13.17983651226158,
+ "grad_norm": 7.943714618682861,
+ "learning_rate": 1.9461395151930707e-05,
+ "loss": 1.0591,
+ "step": 4837
+ },
+ {
+ "epoch": 13.182561307901908,
+ "grad_norm": 8.022335052490234,
+ "learning_rate": 1.9461109399333748e-05,
+ "loss": 1.175,
+ "step": 4838
+ },
+ {
+ "epoch": 13.185286103542234,
+ "grad_norm": 6.417407989501953,
+ "learning_rate": 1.9460823573053972e-05,
+ "loss": 1.0693,
+ "step": 4839
+ },
+ {
+ "epoch": 13.188010899182562,
+ "grad_norm": 6.654033660888672,
+ "learning_rate": 1.946053767309361e-05,
+ "loss": 0.9177,
+ "step": 4840
+ },
+ {
+ "epoch": 13.190735694822889,
+ "grad_norm": 7.856115341186523,
+ "learning_rate": 1.946025169945488e-05,
+ "loss": 0.8889,
+ "step": 4841
+ },
+ {
+ "epoch": 13.193460490463215,
+ "grad_norm": 8.370399475097656,
+ "learning_rate": 1.945996565214002e-05,
+ "loss": 0.9121,
+ "step": 4842
+ },
+ {
+ "epoch": 13.196185286103542,
+ "grad_norm": 7.593133449554443,
+ "learning_rate": 1.9459679531151246e-05,
+ "loss": 0.8115,
+ "step": 4843
+ },
+ {
+ "epoch": 13.19891008174387,
+ "grad_norm": 6.9923505783081055,
+ "learning_rate": 1.9459393336490797e-05,
+ "loss": 0.8719,
+ "step": 4844
+ },
+ {
+ "epoch": 13.201634877384196,
+ "grad_norm": 8.118725776672363,
+ "learning_rate": 1.9459107068160898e-05,
+ "loss": 0.926,
+ "step": 4845
+ },
+ {
+ "epoch": 13.204359673024523,
+ "grad_norm": 7.490218639373779,
+ "learning_rate": 1.9458820726163774e-05,
+ "loss": 0.9304,
+ "step": 4846
+ },
+ {
+ "epoch": 13.207084468664851,
+ "grad_norm": 6.650937557220459,
+ "learning_rate": 1.945853431050166e-05,
+ "loss": 0.8439,
+ "step": 4847
+ },
+ {
+ "epoch": 13.209809264305177,
+ "grad_norm": 7.192839622497559,
+ "learning_rate": 1.9458247821176787e-05,
+ "loss": 0.8254,
+ "step": 4848
+ },
+ {
+ "epoch": 13.212534059945504,
+ "grad_norm": 7.078209400177002,
+ "learning_rate": 1.945796125819138e-05,
+ "loss": 0.9397,
+ "step": 4849
+ },
+ {
+ "epoch": 13.215258855585832,
+ "grad_norm": 6.355413436889648,
+ "learning_rate": 1.9457674621547678e-05,
+ "loss": 0.9648,
+ "step": 4850
+ },
+ {
+ "epoch": 13.217983651226158,
+ "grad_norm": 9.742717742919922,
+ "learning_rate": 1.945738791124791e-05,
+ "loss": 0.9922,
+ "step": 4851
+ },
+ {
+ "epoch": 13.220708446866485,
+ "grad_norm": 6.192273139953613,
+ "learning_rate": 1.945710112729431e-05,
+ "loss": 0.8534,
+ "step": 4852
+ },
+ {
+ "epoch": 13.223433242506813,
+ "grad_norm": 9.603716850280762,
+ "learning_rate": 1.9456814269689112e-05,
+ "loss": 1.1655,
+ "step": 4853
+ },
+ {
+ "epoch": 13.226158038147139,
+ "grad_norm": 7.578845977783203,
+ "learning_rate": 1.9456527338434545e-05,
+ "loss": 0.926,
+ "step": 4854
+ },
+ {
+ "epoch": 13.228882833787466,
+ "grad_norm": 8.070075988769531,
+ "learning_rate": 1.945624033353285e-05,
+ "loss": 1.1516,
+ "step": 4855
+ },
+ {
+ "epoch": 13.231607629427794,
+ "grad_norm": 7.977468013763428,
+ "learning_rate": 1.945595325498626e-05,
+ "loss": 1.0048,
+ "step": 4856
+ },
+ {
+ "epoch": 13.23433242506812,
+ "grad_norm": 9.181925773620605,
+ "learning_rate": 1.9455666102797007e-05,
+ "loss": 1.012,
+ "step": 4857
+ },
+ {
+ "epoch": 13.237057220708447,
+ "grad_norm": 6.72907018661499,
+ "learning_rate": 1.9455378876967335e-05,
+ "loss": 0.8461,
+ "step": 4858
+ },
+ {
+ "epoch": 13.239782016348773,
+ "grad_norm": 6.958976745605469,
+ "learning_rate": 1.9455091577499476e-05,
+ "loss": 0.9999,
+ "step": 4859
+ },
+ {
+ "epoch": 13.2425068119891,
+ "grad_norm": 8.195622444152832,
+ "learning_rate": 1.9454804204395667e-05,
+ "loss": 0.8752,
+ "step": 4860
+ },
+ {
+ "epoch": 13.245231607629428,
+ "grad_norm": 8.399968147277832,
+ "learning_rate": 1.9454516757658142e-05,
+ "loss": 0.9711,
+ "step": 4861
+ },
+ {
+ "epoch": 13.247956403269754,
+ "grad_norm": 14.523238182067871,
+ "learning_rate": 1.9454229237289152e-05,
+ "loss": 0.9007,
+ "step": 4862
+ },
+ {
+ "epoch": 13.250681198910081,
+ "grad_norm": 9.995874404907227,
+ "learning_rate": 1.9453941643290928e-05,
+ "loss": 0.9795,
+ "step": 4863
+ },
+ {
+ "epoch": 13.253405994550409,
+ "grad_norm": 8.373580932617188,
+ "learning_rate": 1.945365397566571e-05,
+ "loss": 0.8952,
+ "step": 4864
+ },
+ {
+ "epoch": 13.256130790190735,
+ "grad_norm": 9.095903396606445,
+ "learning_rate": 1.9453366234415737e-05,
+ "loss": 1.0,
+ "step": 4865
+ },
+ {
+ "epoch": 13.258855585831062,
+ "grad_norm": 6.816840171813965,
+ "learning_rate": 1.9453078419543257e-05,
+ "loss": 0.7858,
+ "step": 4866
+ },
+ {
+ "epoch": 13.26158038147139,
+ "grad_norm": 10.036340713500977,
+ "learning_rate": 1.9452790531050502e-05,
+ "loss": 0.9304,
+ "step": 4867
+ },
+ {
+ "epoch": 13.264305177111716,
+ "grad_norm": 7.289087295532227,
+ "learning_rate": 1.945250256893972e-05,
+ "loss": 0.8828,
+ "step": 4868
+ },
+ {
+ "epoch": 13.267029972752043,
+ "grad_norm": 7.56739616394043,
+ "learning_rate": 1.945221453321315e-05,
+ "loss": 1.095,
+ "step": 4869
+ },
+ {
+ "epoch": 13.269754768392371,
+ "grad_norm": 7.3365559577941895,
+ "learning_rate": 1.945192642387304e-05,
+ "loss": 1.0979,
+ "step": 4870
+ },
+ {
+ "epoch": 13.272479564032697,
+ "grad_norm": 7.831923007965088,
+ "learning_rate": 1.945163824092163e-05,
+ "loss": 1.0586,
+ "step": 4871
+ },
+ {
+ "epoch": 13.275204359673024,
+ "grad_norm": 7.7605366706848145,
+ "learning_rate": 1.9451349984361168e-05,
+ "loss": 0.9161,
+ "step": 4872
+ },
+ {
+ "epoch": 13.277929155313352,
+ "grad_norm": 7.798645496368408,
+ "learning_rate": 1.9451061654193893e-05,
+ "loss": 0.8824,
+ "step": 4873
+ },
+ {
+ "epoch": 13.280653950953678,
+ "grad_norm": 7.644713878631592,
+ "learning_rate": 1.9450773250422058e-05,
+ "loss": 0.8862,
+ "step": 4874
+ },
+ {
+ "epoch": 13.283378746594005,
+ "grad_norm": 6.493762969970703,
+ "learning_rate": 1.9450484773047905e-05,
+ "loss": 0.7802,
+ "step": 4875
+ },
+ {
+ "epoch": 13.286103542234333,
+ "grad_norm": 6.979416370391846,
+ "learning_rate": 1.9450196222073676e-05,
+ "loss": 0.8601,
+ "step": 4876
+ },
+ {
+ "epoch": 13.288828337874659,
+ "grad_norm": 6.995616436004639,
+ "learning_rate": 1.9449907597501626e-05,
+ "loss": 1.1354,
+ "step": 4877
+ },
+ {
+ "epoch": 13.291553133514986,
+ "grad_norm": 7.996072769165039,
+ "learning_rate": 1.9449618899334e-05,
+ "loss": 0.7102,
+ "step": 4878
+ },
+ {
+ "epoch": 13.294277929155314,
+ "grad_norm": 6.949777126312256,
+ "learning_rate": 1.9449330127573044e-05,
+ "loss": 0.9196,
+ "step": 4879
+ },
+ {
+ "epoch": 13.29700272479564,
+ "grad_norm": 6.334749698638916,
+ "learning_rate": 1.9449041282221012e-05,
+ "loss": 0.9225,
+ "step": 4880
+ },
+ {
+ "epoch": 13.299727520435967,
+ "grad_norm": 10.872946739196777,
+ "learning_rate": 1.9448752363280147e-05,
+ "loss": 1.1615,
+ "step": 4881
+ },
+ {
+ "epoch": 13.302452316076295,
+ "grad_norm": 7.165452480316162,
+ "learning_rate": 1.9448463370752705e-05,
+ "loss": 0.7859,
+ "step": 4882
+ },
+ {
+ "epoch": 13.30517711171662,
+ "grad_norm": 5.3282151222229,
+ "learning_rate": 1.9448174304640936e-05,
+ "loss": 0.7449,
+ "step": 4883
+ },
+ {
+ "epoch": 13.307901907356948,
+ "grad_norm": 10.083754539489746,
+ "learning_rate": 1.944788516494709e-05,
+ "loss": 1.0245,
+ "step": 4884
+ },
+ {
+ "epoch": 13.310626702997276,
+ "grad_norm": 8.380343437194824,
+ "learning_rate": 1.9447595951673415e-05,
+ "loss": 0.8978,
+ "step": 4885
+ },
+ {
+ "epoch": 13.313351498637601,
+ "grad_norm": 12.231260299682617,
+ "learning_rate": 1.9447306664822166e-05,
+ "loss": 1.0125,
+ "step": 4886
+ },
+ {
+ "epoch": 13.316076294277929,
+ "grad_norm": 7.740874767303467,
+ "learning_rate": 1.94470173043956e-05,
+ "loss": 0.9863,
+ "step": 4887
+ },
+ {
+ "epoch": 13.318801089918257,
+ "grad_norm": 7.022454738616943,
+ "learning_rate": 1.9446727870395964e-05,
+ "loss": 0.8018,
+ "step": 4888
+ },
+ {
+ "epoch": 13.321525885558582,
+ "grad_norm": 8.127674102783203,
+ "learning_rate": 1.944643836282552e-05,
+ "loss": 0.9465,
+ "step": 4889
+ },
+ {
+ "epoch": 13.32425068119891,
+ "grad_norm": 7.249255180358887,
+ "learning_rate": 1.9446148781686515e-05,
+ "loss": 0.9192,
+ "step": 4890
+ },
+ {
+ "epoch": 13.326975476839237,
+ "grad_norm": 6.11199426651001,
+ "learning_rate": 1.9445859126981207e-05,
+ "loss": 0.7665,
+ "step": 4891
+ },
+ {
+ "epoch": 13.329700272479563,
+ "grad_norm": 8.588767051696777,
+ "learning_rate": 1.9445569398711853e-05,
+ "loss": 0.9841,
+ "step": 4892
+ },
+ {
+ "epoch": 13.33242506811989,
+ "grad_norm": 6.676856517791748,
+ "learning_rate": 1.944527959688071e-05,
+ "loss": 1.0819,
+ "step": 4893
+ },
+ {
+ "epoch": 13.335149863760218,
+ "grad_norm": 36.00337600708008,
+ "learning_rate": 1.9444989721490034e-05,
+ "loss": 1.0759,
+ "step": 4894
+ },
+ {
+ "epoch": 13.337874659400544,
+ "grad_norm": 7.513136863708496,
+ "learning_rate": 1.9444699772542076e-05,
+ "loss": 0.8727,
+ "step": 4895
+ },
+ {
+ "epoch": 13.340599455040872,
+ "grad_norm": 7.7315239906311035,
+ "learning_rate": 1.9444409750039107e-05,
+ "loss": 0.7689,
+ "step": 4896
+ },
+ {
+ "epoch": 13.3433242506812,
+ "grad_norm": 5.967472553253174,
+ "learning_rate": 1.9444119653983376e-05,
+ "loss": 1.0309,
+ "step": 4897
+ },
+ {
+ "epoch": 13.346049046321525,
+ "grad_norm": 8.213624000549316,
+ "learning_rate": 1.9443829484377147e-05,
+ "loss": 1.042,
+ "step": 4898
+ },
+ {
+ "epoch": 13.348773841961853,
+ "grad_norm": 7.630632400512695,
+ "learning_rate": 1.9443539241222672e-05,
+ "loss": 0.9795,
+ "step": 4899
+ },
+ {
+ "epoch": 13.35149863760218,
+ "grad_norm": 6.520260810852051,
+ "learning_rate": 1.9443248924522223e-05,
+ "loss": 0.7528,
+ "step": 4900
+ },
+ {
+ "epoch": 13.354223433242506,
+ "grad_norm": 7.042547702789307,
+ "learning_rate": 1.9442958534278054e-05,
+ "loss": 1.0549,
+ "step": 4901
+ },
+ {
+ "epoch": 13.356948228882834,
+ "grad_norm": 7.763444423675537,
+ "learning_rate": 1.944266807049243e-05,
+ "loss": 0.8856,
+ "step": 4902
+ },
+ {
+ "epoch": 13.359673024523161,
+ "grad_norm": 8.002070426940918,
+ "learning_rate": 1.944237753316761e-05,
+ "loss": 0.8573,
+ "step": 4903
+ },
+ {
+ "epoch": 13.362397820163487,
+ "grad_norm": 10.079537391662598,
+ "learning_rate": 1.944208692230585e-05,
+ "loss": 0.9453,
+ "step": 4904
+ },
+ {
+ "epoch": 13.365122615803815,
+ "grad_norm": 7.365504741668701,
+ "learning_rate": 1.944179623790943e-05,
+ "loss": 0.8053,
+ "step": 4905
+ },
+ {
+ "epoch": 13.367847411444142,
+ "grad_norm": 7.551240921020508,
+ "learning_rate": 1.9441505479980602e-05,
+ "loss": 0.8964,
+ "step": 4906
+ },
+ {
+ "epoch": 13.370572207084468,
+ "grad_norm": 8.915811538696289,
+ "learning_rate": 1.9441214648521633e-05,
+ "loss": 1.0874,
+ "step": 4907
+ },
+ {
+ "epoch": 13.373297002724795,
+ "grad_norm": 7.377119541168213,
+ "learning_rate": 1.944092374353479e-05,
+ "loss": 0.901,
+ "step": 4908
+ },
+ {
+ "epoch": 13.376021798365123,
+ "grad_norm": 8.097033500671387,
+ "learning_rate": 1.9440632765022332e-05,
+ "loss": 0.8466,
+ "step": 4909
+ },
+ {
+ "epoch": 13.378746594005449,
+ "grad_norm": 8.775738716125488,
+ "learning_rate": 1.9440341712986534e-05,
+ "loss": 0.9343,
+ "step": 4910
+ },
+ {
+ "epoch": 13.381471389645776,
+ "grad_norm": 6.575735569000244,
+ "learning_rate": 1.944005058742966e-05,
+ "loss": 0.9802,
+ "step": 4911
+ },
+ {
+ "epoch": 13.384196185286104,
+ "grad_norm": 6.155643463134766,
+ "learning_rate": 1.9439759388353972e-05,
+ "loss": 0.8528,
+ "step": 4912
+ },
+ {
+ "epoch": 13.38692098092643,
+ "grad_norm": 7.085944175720215,
+ "learning_rate": 1.9439468115761745e-05,
+ "loss": 1.0486,
+ "step": 4913
+ },
+ {
+ "epoch": 13.389645776566757,
+ "grad_norm": 6.510679244995117,
+ "learning_rate": 1.943917676965524e-05,
+ "loss": 0.9318,
+ "step": 4914
+ },
+ {
+ "epoch": 13.392370572207085,
+ "grad_norm": 6.82840633392334,
+ "learning_rate": 1.9438885350036735e-05,
+ "loss": 1.1751,
+ "step": 4915
+ },
+ {
+ "epoch": 13.39509536784741,
+ "grad_norm": 6.939377307891846,
+ "learning_rate": 1.943859385690849e-05,
+ "loss": 0.8862,
+ "step": 4916
+ },
+ {
+ "epoch": 13.397820163487738,
+ "grad_norm": 7.082571983337402,
+ "learning_rate": 1.9438302290272782e-05,
+ "loss": 0.8796,
+ "step": 4917
+ },
+ {
+ "epoch": 13.400544959128066,
+ "grad_norm": 6.676981449127197,
+ "learning_rate": 1.943801065013188e-05,
+ "loss": 1.2188,
+ "step": 4918
+ },
+ {
+ "epoch": 13.403269754768392,
+ "grad_norm": 6.111840724945068,
+ "learning_rate": 1.9437718936488054e-05,
+ "loss": 0.9233,
+ "step": 4919
+ },
+ {
+ "epoch": 13.40599455040872,
+ "grad_norm": 7.194887161254883,
+ "learning_rate": 1.9437427149343576e-05,
+ "loss": 0.905,
+ "step": 4920
+ },
+ {
+ "epoch": 13.408719346049047,
+ "grad_norm": 8.22214412689209,
+ "learning_rate": 1.943713528870072e-05,
+ "loss": 0.9601,
+ "step": 4921
+ },
+ {
+ "epoch": 13.411444141689373,
+ "grad_norm": 6.435676574707031,
+ "learning_rate": 1.943684335456176e-05,
+ "loss": 0.9977,
+ "step": 4922
+ },
+ {
+ "epoch": 13.4141689373297,
+ "grad_norm": 8.388131141662598,
+ "learning_rate": 1.9436551346928963e-05,
+ "loss": 0.9851,
+ "step": 4923
+ },
+ {
+ "epoch": 13.416893732970028,
+ "grad_norm": 6.375876426696777,
+ "learning_rate": 1.943625926580461e-05,
+ "loss": 0.847,
+ "step": 4924
+ },
+ {
+ "epoch": 13.419618528610354,
+ "grad_norm": 8.57419490814209,
+ "learning_rate": 1.9435967111190975e-05,
+ "loss": 0.9354,
+ "step": 4925
+ },
+ {
+ "epoch": 13.422343324250681,
+ "grad_norm": 6.817331790924072,
+ "learning_rate": 1.9435674883090333e-05,
+ "loss": 1.0522,
+ "step": 4926
+ },
+ {
+ "epoch": 13.425068119891009,
+ "grad_norm": 8.383650779724121,
+ "learning_rate": 1.9435382581504955e-05,
+ "loss": 0.9231,
+ "step": 4927
+ },
+ {
+ "epoch": 13.427792915531334,
+ "grad_norm": 6.774690628051758,
+ "learning_rate": 1.9435090206437124e-05,
+ "loss": 0.8994,
+ "step": 4928
+ },
+ {
+ "epoch": 13.430517711171662,
+ "grad_norm": 7.372954368591309,
+ "learning_rate": 1.9434797757889114e-05,
+ "loss": 0.8389,
+ "step": 4929
+ },
+ {
+ "epoch": 13.43324250681199,
+ "grad_norm": 7.065223217010498,
+ "learning_rate": 1.94345052358632e-05,
+ "loss": 1.1116,
+ "step": 4930
+ },
+ {
+ "epoch": 13.435967302452315,
+ "grad_norm": 9.452176094055176,
+ "learning_rate": 1.9434212640361662e-05,
+ "loss": 0.8865,
+ "step": 4931
+ },
+ {
+ "epoch": 13.438692098092643,
+ "grad_norm": 6.2181925773620605,
+ "learning_rate": 1.9433919971386785e-05,
+ "loss": 0.7875,
+ "step": 4932
+ },
+ {
+ "epoch": 13.44141689373297,
+ "grad_norm": 7.458146572113037,
+ "learning_rate": 1.943362722894084e-05,
+ "loss": 1.1216,
+ "step": 4933
+ },
+ {
+ "epoch": 13.444141689373296,
+ "grad_norm": 7.24932336807251,
+ "learning_rate": 1.943333441302611e-05,
+ "loss": 0.8511,
+ "step": 4934
+ },
+ {
+ "epoch": 13.446866485013624,
+ "grad_norm": 5.325828552246094,
+ "learning_rate": 1.9433041523644873e-05,
+ "loss": 0.7935,
+ "step": 4935
+ },
+ {
+ "epoch": 13.449591280653951,
+ "grad_norm": 8.474371910095215,
+ "learning_rate": 1.9432748560799416e-05,
+ "loss": 1.111,
+ "step": 4936
+ },
+ {
+ "epoch": 13.452316076294277,
+ "grad_norm": 6.679025650024414,
+ "learning_rate": 1.9432455524492014e-05,
+ "loss": 0.7539,
+ "step": 4937
+ },
+ {
+ "epoch": 13.455040871934605,
+ "grad_norm": 6.754443168640137,
+ "learning_rate": 1.9432162414724952e-05,
+ "loss": 0.8634,
+ "step": 4938
+ },
+ {
+ "epoch": 13.457765667574932,
+ "grad_norm": 8.68956184387207,
+ "learning_rate": 1.9431869231500516e-05,
+ "loss": 0.9755,
+ "step": 4939
+ },
+ {
+ "epoch": 13.460490463215258,
+ "grad_norm": 7.326023101806641,
+ "learning_rate": 1.9431575974820983e-05,
+ "loss": 0.9406,
+ "step": 4940
+ },
+ {
+ "epoch": 13.463215258855586,
+ "grad_norm": 6.7393479347229,
+ "learning_rate": 1.943128264468864e-05,
+ "loss": 0.8517,
+ "step": 4941
+ },
+ {
+ "epoch": 13.465940054495913,
+ "grad_norm": 7.56080436706543,
+ "learning_rate": 1.943098924110577e-05,
+ "loss": 1.0422,
+ "step": 4942
+ },
+ {
+ "epoch": 13.46866485013624,
+ "grad_norm": 7.912402629852295,
+ "learning_rate": 1.9430695764074662e-05,
+ "loss": 0.9664,
+ "step": 4943
+ },
+ {
+ "epoch": 13.471389645776567,
+ "grad_norm": 7.679460525512695,
+ "learning_rate": 1.9430402213597598e-05,
+ "loss": 0.8623,
+ "step": 4944
+ },
+ {
+ "epoch": 13.474114441416894,
+ "grad_norm": 6.885580539703369,
+ "learning_rate": 1.9430108589676868e-05,
+ "loss": 1.1494,
+ "step": 4945
+ },
+ {
+ "epoch": 13.47683923705722,
+ "grad_norm": 7.96921968460083,
+ "learning_rate": 1.9429814892314752e-05,
+ "loss": 1.0903,
+ "step": 4946
+ },
+ {
+ "epoch": 13.479564032697548,
+ "grad_norm": 6.113422870635986,
+ "learning_rate": 1.9429521121513542e-05,
+ "loss": 0.8741,
+ "step": 4947
+ },
+ {
+ "epoch": 13.482288828337875,
+ "grad_norm": 8.063274383544922,
+ "learning_rate": 1.9429227277275525e-05,
+ "loss": 0.8678,
+ "step": 4948
+ },
+ {
+ "epoch": 13.485013623978201,
+ "grad_norm": 205.6051025390625,
+ "learning_rate": 1.942893335960299e-05,
+ "loss": 0.9664,
+ "step": 4949
+ },
+ {
+ "epoch": 13.487738419618529,
+ "grad_norm": 6.834926605224609,
+ "learning_rate": 1.9428639368498223e-05,
+ "loss": 1.0803,
+ "step": 4950
+ },
+ {
+ "epoch": 13.490463215258856,
+ "grad_norm": 8.659931182861328,
+ "learning_rate": 1.9428345303963517e-05,
+ "loss": 0.9171,
+ "step": 4951
+ },
+ {
+ "epoch": 13.493188010899182,
+ "grad_norm": 9.128621101379395,
+ "learning_rate": 1.9428051166001162e-05,
+ "loss": 0.9993,
+ "step": 4952
+ },
+ {
+ "epoch": 13.49591280653951,
+ "grad_norm": 7.697845458984375,
+ "learning_rate": 1.942775695461345e-05,
+ "loss": 0.95,
+ "step": 4953
+ },
+ {
+ "epoch": 13.498637602179837,
+ "grad_norm": 11.67182445526123,
+ "learning_rate": 1.942746266980267e-05,
+ "loss": 1.1415,
+ "step": 4954
+ },
+ {
+ "epoch": 13.501362397820163,
+ "grad_norm": 13.387124061584473,
+ "learning_rate": 1.942716831157111e-05,
+ "loss": 0.8293,
+ "step": 4955
+ },
+ {
+ "epoch": 13.50408719346049,
+ "grad_norm": 13.74110221862793,
+ "learning_rate": 1.942687387992107e-05,
+ "loss": 1.1968,
+ "step": 4956
+ },
+ {
+ "epoch": 13.506811989100818,
+ "grad_norm": 13.10708999633789,
+ "learning_rate": 1.9426579374854838e-05,
+ "loss": 1.1931,
+ "step": 4957
+ },
+ {
+ "epoch": 13.509536784741144,
+ "grad_norm": 13.886868476867676,
+ "learning_rate": 1.9426284796374712e-05,
+ "loss": 1.3069,
+ "step": 4958
+ },
+ {
+ "epoch": 13.512261580381471,
+ "grad_norm": 24.858642578125,
+ "learning_rate": 1.942599014448298e-05,
+ "loss": 1.4338,
+ "step": 4959
+ },
+ {
+ "epoch": 13.514986376021799,
+ "grad_norm": 10.993972778320312,
+ "learning_rate": 1.942569541918194e-05,
+ "loss": 0.9342,
+ "step": 4960
+ },
+ {
+ "epoch": 13.517711171662125,
+ "grad_norm": 8.235885620117188,
+ "learning_rate": 1.9425400620473893e-05,
+ "loss": 1.0774,
+ "step": 4961
+ },
+ {
+ "epoch": 13.520435967302452,
+ "grad_norm": 10.39051628112793,
+ "learning_rate": 1.9425105748361122e-05,
+ "loss": 0.9329,
+ "step": 4962
+ },
+ {
+ "epoch": 13.52316076294278,
+ "grad_norm": 12.096440315246582,
+ "learning_rate": 1.9424810802845937e-05,
+ "loss": 1.0715,
+ "step": 4963
+ },
+ {
+ "epoch": 13.525885558583106,
+ "grad_norm": 8.995593070983887,
+ "learning_rate": 1.9424515783930624e-05,
+ "loss": 1.0829,
+ "step": 4964
+ },
+ {
+ "epoch": 13.528610354223433,
+ "grad_norm": 31.661426544189453,
+ "learning_rate": 1.942422069161749e-05,
+ "loss": 1.0729,
+ "step": 4965
+ },
+ {
+ "epoch": 13.53133514986376,
+ "grad_norm": 9.70279598236084,
+ "learning_rate": 1.9423925525908826e-05,
+ "loss": 1.0879,
+ "step": 4966
+ },
+ {
+ "epoch": 13.534059945504087,
+ "grad_norm": 9.995427131652832,
+ "learning_rate": 1.9423630286806928e-05,
+ "loss": 1.0925,
+ "step": 4967
+ },
+ {
+ "epoch": 13.536784741144414,
+ "grad_norm": 8.869283676147461,
+ "learning_rate": 1.942333497431411e-05,
+ "loss": 1.1863,
+ "step": 4968
+ },
+ {
+ "epoch": 13.539509536784742,
+ "grad_norm": 9.1719331741333,
+ "learning_rate": 1.9423039588432657e-05,
+ "loss": 1.0502,
+ "step": 4969
+ },
+ {
+ "epoch": 13.542234332425068,
+ "grad_norm": 9.887907981872559,
+ "learning_rate": 1.9422744129164877e-05,
+ "loss": 1.1663,
+ "step": 4970
+ },
+ {
+ "epoch": 13.544959128065395,
+ "grad_norm": 11.405031204223633,
+ "learning_rate": 1.9422448596513067e-05,
+ "loss": 1.1328,
+ "step": 4971
+ },
+ {
+ "epoch": 13.547683923705723,
+ "grad_norm": 8.26784896850586,
+ "learning_rate": 1.9422152990479533e-05,
+ "loss": 1.189,
+ "step": 4972
+ },
+ {
+ "epoch": 13.550408719346049,
+ "grad_norm": 10.563349723815918,
+ "learning_rate": 1.9421857311066575e-05,
+ "loss": 1.1589,
+ "step": 4973
+ },
+ {
+ "epoch": 13.553133514986376,
+ "grad_norm": 9.405486106872559,
+ "learning_rate": 1.9421561558276495e-05,
+ "loss": 1.0366,
+ "step": 4974
+ },
+ {
+ "epoch": 13.555858310626704,
+ "grad_norm": 8.012311935424805,
+ "learning_rate": 1.942126573211159e-05,
+ "loss": 0.9626,
+ "step": 4975
+ },
+ {
+ "epoch": 13.55858310626703,
+ "grad_norm": 16.19670867919922,
+ "learning_rate": 1.942096983257418e-05,
+ "loss": 1.1593,
+ "step": 4976
+ },
+ {
+ "epoch": 13.561307901907357,
+ "grad_norm": 8.918001174926758,
+ "learning_rate": 1.9420673859666557e-05,
+ "loss": 1.1481,
+ "step": 4977
+ },
+ {
+ "epoch": 13.564032697547685,
+ "grad_norm": 9.493700981140137,
+ "learning_rate": 1.942037781339103e-05,
+ "loss": 1.2657,
+ "step": 4978
+ },
+ {
+ "epoch": 13.56675749318801,
+ "grad_norm": 8.065284729003906,
+ "learning_rate": 1.9420081693749903e-05,
+ "loss": 0.8613,
+ "step": 4979
+ },
+ {
+ "epoch": 13.569482288828338,
+ "grad_norm": 8.27554702758789,
+ "learning_rate": 1.941978550074548e-05,
+ "loss": 0.9911,
+ "step": 4980
+ },
+ {
+ "epoch": 13.572207084468666,
+ "grad_norm": 9.209878921508789,
+ "learning_rate": 1.9419489234380077e-05,
+ "loss": 0.9402,
+ "step": 4981
+ },
+ {
+ "epoch": 13.574931880108991,
+ "grad_norm": 7.841694355010986,
+ "learning_rate": 1.941919289465599e-05,
+ "loss": 1.0109,
+ "step": 4982
+ },
+ {
+ "epoch": 13.577656675749319,
+ "grad_norm": 9.493952751159668,
+ "learning_rate": 1.9418896481575534e-05,
+ "loss": 1.0122,
+ "step": 4983
+ },
+ {
+ "epoch": 13.580381471389646,
+ "grad_norm": 9.248424530029297,
+ "learning_rate": 1.9418599995141013e-05,
+ "loss": 0.9558,
+ "step": 4984
+ },
+ {
+ "epoch": 13.583106267029972,
+ "grad_norm": 7.403234004974365,
+ "learning_rate": 1.941830343535474e-05,
+ "loss": 1.0684,
+ "step": 4985
+ },
+ {
+ "epoch": 13.5858310626703,
+ "grad_norm": 11.699258804321289,
+ "learning_rate": 1.9418006802219022e-05,
+ "loss": 1.1045,
+ "step": 4986
+ },
+ {
+ "epoch": 13.588555858310627,
+ "grad_norm": 7.7209038734436035,
+ "learning_rate": 1.9417710095736173e-05,
+ "loss": 0.9996,
+ "step": 4987
+ },
+ {
+ "epoch": 13.591280653950953,
+ "grad_norm": 6.901909351348877,
+ "learning_rate": 1.94174133159085e-05,
+ "loss": 0.8037,
+ "step": 4988
+ },
+ {
+ "epoch": 13.59400544959128,
+ "grad_norm": 9.153884887695312,
+ "learning_rate": 1.9417116462738312e-05,
+ "loss": 1.1987,
+ "step": 4989
+ },
+ {
+ "epoch": 13.596730245231608,
+ "grad_norm": 10.095368385314941,
+ "learning_rate": 1.9416819536227928e-05,
+ "loss": 1.1111,
+ "step": 4990
+ },
+ {
+ "epoch": 13.599455040871934,
+ "grad_norm": 8.798355102539062,
+ "learning_rate": 1.9416522536379653e-05,
+ "loss": 0.9673,
+ "step": 4991
+ },
+ {
+ "epoch": 13.602179836512262,
+ "grad_norm": 8.561857223510742,
+ "learning_rate": 1.9416225463195807e-05,
+ "loss": 1.199,
+ "step": 4992
+ },
+ {
+ "epoch": 13.60490463215259,
+ "grad_norm": 7.14668607711792,
+ "learning_rate": 1.94159283166787e-05,
+ "loss": 0.9043,
+ "step": 4993
+ },
+ {
+ "epoch": 13.607629427792915,
+ "grad_norm": 7.5243239402771,
+ "learning_rate": 1.9415631096830644e-05,
+ "loss": 1.041,
+ "step": 4994
+ },
+ {
+ "epoch": 13.610354223433243,
+ "grad_norm": 10.2526216506958,
+ "learning_rate": 1.9415333803653955e-05,
+ "loss": 0.9493,
+ "step": 4995
+ },
+ {
+ "epoch": 13.61307901907357,
+ "grad_norm": 10.328581809997559,
+ "learning_rate": 1.941503643715095e-05,
+ "loss": 0.9935,
+ "step": 4996
+ },
+ {
+ "epoch": 13.615803814713896,
+ "grad_norm": 8.774404525756836,
+ "learning_rate": 1.9414738997323947e-05,
+ "loss": 1.0056,
+ "step": 4997
+ },
+ {
+ "epoch": 13.618528610354224,
+ "grad_norm": 7.80064058303833,
+ "learning_rate": 1.9414441484175258e-05,
+ "loss": 0.9949,
+ "step": 4998
+ },
+ {
+ "epoch": 13.621253405994551,
+ "grad_norm": 9.98238468170166,
+ "learning_rate": 1.9414143897707204e-05,
+ "loss": 1.0182,
+ "step": 4999
+ },
+ {
+ "epoch": 13.623978201634877,
+ "grad_norm": 7.018301010131836,
+ "learning_rate": 1.9413846237922098e-05,
+ "loss": 1.1149,
+ "step": 5000
+ },
+ {
+ "epoch": 13.626702997275205,
+ "grad_norm": 9.805243492126465,
+ "learning_rate": 1.941354850482226e-05,
+ "loss": 1.1243,
+ "step": 5001
+ },
+ {
+ "epoch": 13.629427792915532,
+ "grad_norm": 9.409788131713867,
+ "learning_rate": 1.941325069841001e-05,
+ "loss": 1.2126,
+ "step": 5002
+ },
+ {
+ "epoch": 13.632152588555858,
+ "grad_norm": 8.863076210021973,
+ "learning_rate": 1.9412952818687665e-05,
+ "loss": 1.0991,
+ "step": 5003
+ },
+ {
+ "epoch": 13.634877384196185,
+ "grad_norm": 7.9986701011657715,
+ "learning_rate": 1.941265486565755e-05,
+ "loss": 1.2161,
+ "step": 5004
+ },
+ {
+ "epoch": 13.637602179836513,
+ "grad_norm": 6.932985305786133,
+ "learning_rate": 1.941235683932198e-05,
+ "loss": 1.0854,
+ "step": 5005
+ },
+ {
+ "epoch": 13.640326975476839,
+ "grad_norm": 7.162502765655518,
+ "learning_rate": 1.9412058739683276e-05,
+ "loss": 1.2744,
+ "step": 5006
+ },
+ {
+ "epoch": 13.643051771117166,
+ "grad_norm": 8.016437530517578,
+ "learning_rate": 1.9411760566743765e-05,
+ "loss": 1.0729,
+ "step": 5007
+ },
+ {
+ "epoch": 13.645776566757494,
+ "grad_norm": 8.695459365844727,
+ "learning_rate": 1.941146232050576e-05,
+ "loss": 1.1378,
+ "step": 5008
+ },
+ {
+ "epoch": 13.64850136239782,
+ "grad_norm": 11.653491020202637,
+ "learning_rate": 1.9411164000971597e-05,
+ "loss": 1.1511,
+ "step": 5009
+ },
+ {
+ "epoch": 13.651226158038147,
+ "grad_norm": 8.149260520935059,
+ "learning_rate": 1.9410865608143586e-05,
+ "loss": 1.0286,
+ "step": 5010
+ },
+ {
+ "epoch": 13.653950953678475,
+ "grad_norm": 5.772851943969727,
+ "learning_rate": 1.941056714202406e-05,
+ "loss": 1.056,
+ "step": 5011
+ },
+ {
+ "epoch": 13.6566757493188,
+ "grad_norm": 12.314532279968262,
+ "learning_rate": 1.9410268602615338e-05,
+ "loss": 1.0375,
+ "step": 5012
+ },
+ {
+ "epoch": 13.659400544959128,
+ "grad_norm": 6.898438453674316,
+ "learning_rate": 1.9409969989919746e-05,
+ "loss": 1.0171,
+ "step": 5013
+ },
+ {
+ "epoch": 13.662125340599456,
+ "grad_norm": 8.139533996582031,
+ "learning_rate": 1.940967130393961e-05,
+ "loss": 1.0424,
+ "step": 5014
+ },
+ {
+ "epoch": 13.664850136239782,
+ "grad_norm": 18.7755069732666,
+ "learning_rate": 1.940937254467726e-05,
+ "loss": 0.9956,
+ "step": 5015
+ },
+ {
+ "epoch": 13.66757493188011,
+ "grad_norm": 8.913077354431152,
+ "learning_rate": 1.940907371213502e-05,
+ "loss": 0.8772,
+ "step": 5016
+ },
+ {
+ "epoch": 13.670299727520437,
+ "grad_norm": 9.922028541564941,
+ "learning_rate": 1.9408774806315215e-05,
+ "loss": 0.9487,
+ "step": 5017
+ },
+ {
+ "epoch": 13.673024523160763,
+ "grad_norm": 7.402695655822754,
+ "learning_rate": 1.9408475827220173e-05,
+ "loss": 0.999,
+ "step": 5018
+ },
+ {
+ "epoch": 13.67574931880109,
+ "grad_norm": 7.302099227905273,
+ "learning_rate": 1.9408176774852224e-05,
+ "loss": 0.8208,
+ "step": 5019
+ },
+ {
+ "epoch": 13.678474114441418,
+ "grad_norm": 5.386303424835205,
+ "learning_rate": 1.9407877649213696e-05,
+ "loss": 0.8572,
+ "step": 5020
+ },
+ {
+ "epoch": 13.681198910081743,
+ "grad_norm": 6.705547332763672,
+ "learning_rate": 1.9407578450306926e-05,
+ "loss": 0.95,
+ "step": 5021
+ },
+ {
+ "epoch": 13.683923705722071,
+ "grad_norm": 8.195853233337402,
+ "learning_rate": 1.9407279178134233e-05,
+ "loss": 0.749,
+ "step": 5022
+ },
+ {
+ "epoch": 13.686648501362399,
+ "grad_norm": 7.185318946838379,
+ "learning_rate": 1.9406979832697953e-05,
+ "loss": 1.1508,
+ "step": 5023
+ },
+ {
+ "epoch": 13.689373297002724,
+ "grad_norm": 8.673990249633789,
+ "learning_rate": 1.9406680414000418e-05,
+ "loss": 1.0205,
+ "step": 5024
+ },
+ {
+ "epoch": 13.692098092643052,
+ "grad_norm": 7.482603073120117,
+ "learning_rate": 1.9406380922043955e-05,
+ "loss": 0.9009,
+ "step": 5025
+ },
+ {
+ "epoch": 13.69482288828338,
+ "grad_norm": 7.246166229248047,
+ "learning_rate": 1.9406081356830906e-05,
+ "loss": 0.8674,
+ "step": 5026
+ },
+ {
+ "epoch": 13.697547683923705,
+ "grad_norm": 7.740346908569336,
+ "learning_rate": 1.9405781718363595e-05,
+ "loss": 0.9142,
+ "step": 5027
+ },
+ {
+ "epoch": 13.700272479564033,
+ "grad_norm": 8.320613861083984,
+ "learning_rate": 1.940548200664436e-05,
+ "loss": 1.0043,
+ "step": 5028
+ },
+ {
+ "epoch": 13.70299727520436,
+ "grad_norm": 8.0038423538208,
+ "learning_rate": 1.9405182221675535e-05,
+ "loss": 1.0093,
+ "step": 5029
+ },
+ {
+ "epoch": 13.705722070844686,
+ "grad_norm": 7.58731746673584,
+ "learning_rate": 1.940488236345945e-05,
+ "loss": 0.9141,
+ "step": 5030
+ },
+ {
+ "epoch": 13.708446866485014,
+ "grad_norm": 8.680437088012695,
+ "learning_rate": 1.9404582431998446e-05,
+ "loss": 1.042,
+ "step": 5031
+ },
+ {
+ "epoch": 13.711171662125341,
+ "grad_norm": 7.866394996643066,
+ "learning_rate": 1.9404282427294854e-05,
+ "loss": 0.9747,
+ "step": 5032
+ },
+ {
+ "epoch": 13.713896457765667,
+ "grad_norm": 6.994747161865234,
+ "learning_rate": 1.9403982349351014e-05,
+ "loss": 0.9153,
+ "step": 5033
+ },
+ {
+ "epoch": 13.716621253405995,
+ "grad_norm": 7.649085998535156,
+ "learning_rate": 1.9403682198169266e-05,
+ "loss": 1.0215,
+ "step": 5034
+ },
+ {
+ "epoch": 13.719346049046322,
+ "grad_norm": 7.346132278442383,
+ "learning_rate": 1.940338197375194e-05,
+ "loss": 0.9293,
+ "step": 5035
+ },
+ {
+ "epoch": 13.722070844686648,
+ "grad_norm": 8.720535278320312,
+ "learning_rate": 1.940308167610138e-05,
+ "loss": 1.1384,
+ "step": 5036
+ },
+ {
+ "epoch": 13.724795640326976,
+ "grad_norm": 7.834371089935303,
+ "learning_rate": 1.940278130521992e-05,
+ "loss": 0.9084,
+ "step": 5037
+ },
+ {
+ "epoch": 13.727520435967303,
+ "grad_norm": 7.634432315826416,
+ "learning_rate": 1.94024808611099e-05,
+ "loss": 1.0111,
+ "step": 5038
+ },
+ {
+ "epoch": 13.730245231607629,
+ "grad_norm": 23.647619247436523,
+ "learning_rate": 1.9402180343773665e-05,
+ "loss": 0.9861,
+ "step": 5039
+ },
+ {
+ "epoch": 13.732970027247957,
+ "grad_norm": 8.106048583984375,
+ "learning_rate": 1.9401879753213556e-05,
+ "loss": 1.1377,
+ "step": 5040
+ },
+ {
+ "epoch": 13.735694822888284,
+ "grad_norm": 8.670890808105469,
+ "learning_rate": 1.9401579089431904e-05,
+ "loss": 0.682,
+ "step": 5041
+ },
+ {
+ "epoch": 13.73841961852861,
+ "grad_norm": 8.890100479125977,
+ "learning_rate": 1.940127835243106e-05,
+ "loss": 0.9855,
+ "step": 5042
+ },
+ {
+ "epoch": 13.741144414168938,
+ "grad_norm": 7.181202411651611,
+ "learning_rate": 1.940097754221336e-05,
+ "loss": 1.343,
+ "step": 5043
+ },
+ {
+ "epoch": 13.743869209809265,
+ "grad_norm": 10.443450927734375,
+ "learning_rate": 1.940067665878115e-05,
+ "loss": 1.2084,
+ "step": 5044
+ },
+ {
+ "epoch": 13.746594005449591,
+ "grad_norm": 7.208229064941406,
+ "learning_rate": 1.9400375702136777e-05,
+ "loss": 0.7529,
+ "step": 5045
+ },
+ {
+ "epoch": 13.749318801089919,
+ "grad_norm": 6.497509479522705,
+ "learning_rate": 1.9400074672282576e-05,
+ "loss": 1.0563,
+ "step": 5046
+ },
+ {
+ "epoch": 13.752043596730246,
+ "grad_norm": 8.703438758850098,
+ "learning_rate": 1.9399773569220896e-05,
+ "loss": 0.9897,
+ "step": 5047
+ },
+ {
+ "epoch": 13.754768392370572,
+ "grad_norm": 5.790050029754639,
+ "learning_rate": 1.939947239295408e-05,
+ "loss": 0.9197,
+ "step": 5048
+ },
+ {
+ "epoch": 13.7574931880109,
+ "grad_norm": 7.058571815490723,
+ "learning_rate": 1.939917114348448e-05,
+ "loss": 0.9258,
+ "step": 5049
+ },
+ {
+ "epoch": 13.760217983651227,
+ "grad_norm": 6.935174942016602,
+ "learning_rate": 1.939886982081444e-05,
+ "loss": 0.9951,
+ "step": 5050
+ },
+ {
+ "epoch": 13.762942779291553,
+ "grad_norm": 8.018680572509766,
+ "learning_rate": 1.93985684249463e-05,
+ "loss": 0.9423,
+ "step": 5051
+ },
+ {
+ "epoch": 13.76566757493188,
+ "grad_norm": 6.092556953430176,
+ "learning_rate": 1.939826695588241e-05,
+ "loss": 1.0454,
+ "step": 5052
+ },
+ {
+ "epoch": 13.768392370572208,
+ "grad_norm": 8.12808609008789,
+ "learning_rate": 1.939796541362512e-05,
+ "loss": 1.0618,
+ "step": 5053
+ },
+ {
+ "epoch": 13.771117166212534,
+ "grad_norm": 7.610034465789795,
+ "learning_rate": 1.9397663798176783e-05,
+ "loss": 0.9559,
+ "step": 5054
+ },
+ {
+ "epoch": 13.773841961852861,
+ "grad_norm": 7.591068744659424,
+ "learning_rate": 1.939736210953974e-05,
+ "loss": 0.9476,
+ "step": 5055
+ },
+ {
+ "epoch": 13.776566757493189,
+ "grad_norm": 7.808644771575928,
+ "learning_rate": 1.939706034771634e-05,
+ "loss": 0.9628,
+ "step": 5056
+ },
+ {
+ "epoch": 13.779291553133515,
+ "grad_norm": 6.579038143157959,
+ "learning_rate": 1.9396758512708943e-05,
+ "loss": 0.9534,
+ "step": 5057
+ },
+ {
+ "epoch": 13.782016348773842,
+ "grad_norm": 8.343643188476562,
+ "learning_rate": 1.9396456604519885e-05,
+ "loss": 1.0653,
+ "step": 5058
+ },
+ {
+ "epoch": 13.78474114441417,
+ "grad_norm": 6.355676651000977,
+ "learning_rate": 1.939615462315153e-05,
+ "loss": 0.9265,
+ "step": 5059
+ },
+ {
+ "epoch": 13.787465940054496,
+ "grad_norm": 5.702615737915039,
+ "learning_rate": 1.9395852568606225e-05,
+ "loss": 0.7544,
+ "step": 5060
+ },
+ {
+ "epoch": 13.790190735694823,
+ "grad_norm": 6.784058570861816,
+ "learning_rate": 1.9395550440886325e-05,
+ "loss": 1.1382,
+ "step": 5061
+ },
+ {
+ "epoch": 13.79291553133515,
+ "grad_norm": 7.309321403503418,
+ "learning_rate": 1.9395248239994177e-05,
+ "loss": 0.9602,
+ "step": 5062
+ },
+ {
+ "epoch": 13.795640326975477,
+ "grad_norm": 7.84268045425415,
+ "learning_rate": 1.939494596593214e-05,
+ "loss": 0.8876,
+ "step": 5063
+ },
+ {
+ "epoch": 13.798365122615804,
+ "grad_norm": 7.967621326446533,
+ "learning_rate": 1.939464361870256e-05,
+ "loss": 0.8743,
+ "step": 5064
+ },
+ {
+ "epoch": 13.80108991825613,
+ "grad_norm": 7.326691150665283,
+ "learning_rate": 1.9394341198307808e-05,
+ "loss": 0.8545,
+ "step": 5065
+ },
+ {
+ "epoch": 13.803814713896458,
+ "grad_norm": 16.698392868041992,
+ "learning_rate": 1.9394038704750225e-05,
+ "loss": 1.1228,
+ "step": 5066
+ },
+ {
+ "epoch": 13.806539509536785,
+ "grad_norm": 7.768336296081543,
+ "learning_rate": 1.939373613803217e-05,
+ "loss": 1.0032,
+ "step": 5067
+ },
+ {
+ "epoch": 13.809264305177111,
+ "grad_norm": 7.1871562004089355,
+ "learning_rate": 1.9393433498155997e-05,
+ "loss": 0.813,
+ "step": 5068
+ },
+ {
+ "epoch": 13.811989100817438,
+ "grad_norm": 6.594654083251953,
+ "learning_rate": 1.9393130785124073e-05,
+ "loss": 1.1415,
+ "step": 5069
+ },
+ {
+ "epoch": 13.814713896457766,
+ "grad_norm": 7.397449970245361,
+ "learning_rate": 1.9392827998938744e-05,
+ "loss": 0.9208,
+ "step": 5070
+ },
+ {
+ "epoch": 13.817438692098092,
+ "grad_norm": 6.035793304443359,
+ "learning_rate": 1.9392525139602373e-05,
+ "loss": 0.7709,
+ "step": 5071
+ },
+ {
+ "epoch": 13.82016348773842,
+ "grad_norm": 8.236284255981445,
+ "learning_rate": 1.939222220711732e-05,
+ "loss": 1.0828,
+ "step": 5072
+ },
+ {
+ "epoch": 13.822888283378747,
+ "grad_norm": 6.910146713256836,
+ "learning_rate": 1.9391919201485945e-05,
+ "loss": 1.0883,
+ "step": 5073
+ },
+ {
+ "epoch": 13.825613079019073,
+ "grad_norm": 7.125742435455322,
+ "learning_rate": 1.93916161227106e-05,
+ "loss": 0.8333,
+ "step": 5074
+ },
+ {
+ "epoch": 13.8283378746594,
+ "grad_norm": 6.1564531326293945,
+ "learning_rate": 1.9391312970793654e-05,
+ "loss": 0.8605,
+ "step": 5075
+ },
+ {
+ "epoch": 13.831062670299728,
+ "grad_norm": 6.9136528968811035,
+ "learning_rate": 1.9391009745737465e-05,
+ "loss": 0.9281,
+ "step": 5076
+ },
+ {
+ "epoch": 13.833787465940054,
+ "grad_norm": 6.259978771209717,
+ "learning_rate": 1.9390706447544393e-05,
+ "loss": 0.8185,
+ "step": 5077
+ },
+ {
+ "epoch": 13.836512261580381,
+ "grad_norm": 7.583415508270264,
+ "learning_rate": 1.9390403076216805e-05,
+ "loss": 1.0083,
+ "step": 5078
+ },
+ {
+ "epoch": 13.839237057220709,
+ "grad_norm": 6.612548828125,
+ "learning_rate": 1.9390099631757057e-05,
+ "loss": 1.1534,
+ "step": 5079
+ },
+ {
+ "epoch": 13.841961852861035,
+ "grad_norm": 6.207594394683838,
+ "learning_rate": 1.938979611416752e-05,
+ "loss": 0.8405,
+ "step": 5080
+ },
+ {
+ "epoch": 13.844686648501362,
+ "grad_norm": 10.629512786865234,
+ "learning_rate": 1.9389492523450545e-05,
+ "loss": 0.8699,
+ "step": 5081
+ },
+ {
+ "epoch": 13.84741144414169,
+ "grad_norm": 8.121286392211914,
+ "learning_rate": 1.938918885960851e-05,
+ "loss": 0.8459,
+ "step": 5082
+ },
+ {
+ "epoch": 13.850136239782016,
+ "grad_norm": 13.624798774719238,
+ "learning_rate": 1.9388885122643773e-05,
+ "loss": 0.949,
+ "step": 5083
+ },
+ {
+ "epoch": 13.852861035422343,
+ "grad_norm": 21.77162742614746,
+ "learning_rate": 1.9388581312558702e-05,
+ "loss": 1.0023,
+ "step": 5084
+ },
+ {
+ "epoch": 13.85558583106267,
+ "grad_norm": 6.888341903686523,
+ "learning_rate": 1.938827742935566e-05,
+ "loss": 1.0459,
+ "step": 5085
+ },
+ {
+ "epoch": 13.858310626702997,
+ "grad_norm": 14.308305740356445,
+ "learning_rate": 1.9387973473037016e-05,
+ "loss": 0.9685,
+ "step": 5086
+ },
+ {
+ "epoch": 13.861035422343324,
+ "grad_norm": 6.8888421058654785,
+ "learning_rate": 1.9387669443605137e-05,
+ "loss": 0.8759,
+ "step": 5087
+ },
+ {
+ "epoch": 13.863760217983652,
+ "grad_norm": 7.493553161621094,
+ "learning_rate": 1.938736534106239e-05,
+ "loss": 0.9979,
+ "step": 5088
+ },
+ {
+ "epoch": 13.866485013623977,
+ "grad_norm": 8.65969181060791,
+ "learning_rate": 1.938706116541114e-05,
+ "loss": 1.0867,
+ "step": 5089
+ },
+ {
+ "epoch": 13.869209809264305,
+ "grad_norm": 7.55653715133667,
+ "learning_rate": 1.9386756916653765e-05,
+ "loss": 0.95,
+ "step": 5090
+ },
+ {
+ "epoch": 13.871934604904633,
+ "grad_norm": 7.130828380584717,
+ "learning_rate": 1.938645259479263e-05,
+ "loss": 1.1249,
+ "step": 5091
+ },
+ {
+ "epoch": 13.874659400544958,
+ "grad_norm": 8.540849685668945,
+ "learning_rate": 1.9386148199830104e-05,
+ "loss": 1.0305,
+ "step": 5092
+ },
+ {
+ "epoch": 13.877384196185286,
+ "grad_norm": 7.238102436065674,
+ "learning_rate": 1.9385843731768556e-05,
+ "loss": 0.8386,
+ "step": 5093
+ },
+ {
+ "epoch": 13.880108991825614,
+ "grad_norm": 8.209512710571289,
+ "learning_rate": 1.9385539190610358e-05,
+ "loss": 1.0289,
+ "step": 5094
+ },
+ {
+ "epoch": 13.88283378746594,
+ "grad_norm": 7.684393882751465,
+ "learning_rate": 1.9385234576357885e-05,
+ "loss": 0.959,
+ "step": 5095
+ },
+ {
+ "epoch": 13.885558583106267,
+ "grad_norm": 8.239577293395996,
+ "learning_rate": 1.9384929889013507e-05,
+ "loss": 0.9572,
+ "step": 5096
+ },
+ {
+ "epoch": 13.888283378746594,
+ "grad_norm": 8.164934158325195,
+ "learning_rate": 1.9384625128579597e-05,
+ "loss": 0.933,
+ "step": 5097
+ },
+ {
+ "epoch": 13.89100817438692,
+ "grad_norm": 6.69886589050293,
+ "learning_rate": 1.9384320295058528e-05,
+ "loss": 1.123,
+ "step": 5098
+ },
+ {
+ "epoch": 13.893732970027248,
+ "grad_norm": 7.207015037536621,
+ "learning_rate": 1.9384015388452674e-05,
+ "loss": 1.0684,
+ "step": 5099
+ },
+ {
+ "epoch": 13.896457765667575,
+ "grad_norm": 8.002460479736328,
+ "learning_rate": 1.9383710408764412e-05,
+ "loss": 0.9329,
+ "step": 5100
+ },
+ {
+ "epoch": 13.899182561307901,
+ "grad_norm": 9.501195907592773,
+ "learning_rate": 1.938340535599611e-05,
+ "loss": 0.8939,
+ "step": 5101
+ },
+ {
+ "epoch": 13.901907356948229,
+ "grad_norm": 7.880015850067139,
+ "learning_rate": 1.938310023015016e-05,
+ "loss": 0.9442,
+ "step": 5102
+ },
+ {
+ "epoch": 13.904632152588556,
+ "grad_norm": 6.862239837646484,
+ "learning_rate": 1.9382795031228915e-05,
+ "loss": 1.0992,
+ "step": 5103
+ },
+ {
+ "epoch": 13.907356948228882,
+ "grad_norm": 7.977880477905273,
+ "learning_rate": 1.9382489759234773e-05,
+ "loss": 1.0347,
+ "step": 5104
+ },
+ {
+ "epoch": 13.91008174386921,
+ "grad_norm": 10.709538459777832,
+ "learning_rate": 1.93821844141701e-05,
+ "loss": 1.077,
+ "step": 5105
+ },
+ {
+ "epoch": 13.912806539509537,
+ "grad_norm": 6.473933219909668,
+ "learning_rate": 1.9381878996037276e-05,
+ "loss": 1.126,
+ "step": 5106
+ },
+ {
+ "epoch": 13.915531335149863,
+ "grad_norm": 7.625763893127441,
+ "learning_rate": 1.938157350483868e-05,
+ "loss": 0.9342,
+ "step": 5107
+ },
+ {
+ "epoch": 13.91825613079019,
+ "grad_norm": 8.736607551574707,
+ "learning_rate": 1.9381267940576694e-05,
+ "loss": 0.9774,
+ "step": 5108
+ },
+ {
+ "epoch": 13.920980926430518,
+ "grad_norm": 9.25937271118164,
+ "learning_rate": 1.9380962303253695e-05,
+ "loss": 1.1155,
+ "step": 5109
+ },
+ {
+ "epoch": 13.923705722070844,
+ "grad_norm": 11.903998374938965,
+ "learning_rate": 1.938065659287206e-05,
+ "loss": 1.2911,
+ "step": 5110
+ },
+ {
+ "epoch": 13.926430517711172,
+ "grad_norm": 8.088606834411621,
+ "learning_rate": 1.9380350809434176e-05,
+ "loss": 1.1135,
+ "step": 5111
+ },
+ {
+ "epoch": 13.9291553133515,
+ "grad_norm": 7.063750743865967,
+ "learning_rate": 1.938004495294242e-05,
+ "loss": 1.0657,
+ "step": 5112
+ },
+ {
+ "epoch": 13.931880108991825,
+ "grad_norm": 6.140817165374756,
+ "learning_rate": 1.9379739023399177e-05,
+ "loss": 0.9557,
+ "step": 5113
+ },
+ {
+ "epoch": 13.934604904632153,
+ "grad_norm": 8.313331604003906,
+ "learning_rate": 1.937943302080683e-05,
+ "loss": 1.0566,
+ "step": 5114
+ },
+ {
+ "epoch": 13.93732970027248,
+ "grad_norm": 8.463521003723145,
+ "learning_rate": 1.9379126945167755e-05,
+ "loss": 1.0652,
+ "step": 5115
+ },
+ {
+ "epoch": 13.940054495912806,
+ "grad_norm": 6.157297611236572,
+ "learning_rate": 1.9378820796484348e-05,
+ "loss": 0.9067,
+ "step": 5116
+ },
+ {
+ "epoch": 13.942779291553133,
+ "grad_norm": 7.8188557624816895,
+ "learning_rate": 1.9378514574758983e-05,
+ "loss": 0.9333,
+ "step": 5117
+ },
+ {
+ "epoch": 13.945504087193461,
+ "grad_norm": 7.52417516708374,
+ "learning_rate": 1.937820827999405e-05,
+ "loss": 0.7734,
+ "step": 5118
+ },
+ {
+ "epoch": 13.948228882833787,
+ "grad_norm": 7.877678394317627,
+ "learning_rate": 1.937790191219193e-05,
+ "loss": 1.0818,
+ "step": 5119
+ },
+ {
+ "epoch": 13.950953678474114,
+ "grad_norm": 7.296222686767578,
+ "learning_rate": 1.9377595471355016e-05,
+ "loss": 0.9546,
+ "step": 5120
+ },
+ {
+ "epoch": 13.953678474114442,
+ "grad_norm": 6.739790439605713,
+ "learning_rate": 1.9377288957485687e-05,
+ "loss": 1.2065,
+ "step": 5121
+ },
+ {
+ "epoch": 13.956403269754768,
+ "grad_norm": 6.1138081550598145,
+ "learning_rate": 1.9376982370586335e-05,
+ "loss": 0.7732,
+ "step": 5122
+ },
+ {
+ "epoch": 13.959128065395095,
+ "grad_norm": 8.651626586914062,
+ "learning_rate": 1.9376675710659344e-05,
+ "loss": 0.9863,
+ "step": 5123
+ },
+ {
+ "epoch": 13.961852861035423,
+ "grad_norm": 7.864006996154785,
+ "learning_rate": 1.9376368977707107e-05,
+ "loss": 1.093,
+ "step": 5124
+ },
+ {
+ "epoch": 13.964577656675749,
+ "grad_norm": 6.719333171844482,
+ "learning_rate": 1.9376062171732005e-05,
+ "loss": 0.8527,
+ "step": 5125
+ },
+ {
+ "epoch": 13.967302452316076,
+ "grad_norm": 8.420954704284668,
+ "learning_rate": 1.937575529273644e-05,
+ "loss": 1.1621,
+ "step": 5126
+ },
+ {
+ "epoch": 13.970027247956404,
+ "grad_norm": 6.830046653747559,
+ "learning_rate": 1.937544834072279e-05,
+ "loss": 0.9778,
+ "step": 5127
+ },
+ {
+ "epoch": 13.97275204359673,
+ "grad_norm": 8.303977012634277,
+ "learning_rate": 1.937514131569345e-05,
+ "loss": 0.9546,
+ "step": 5128
+ },
+ {
+ "epoch": 13.975476839237057,
+ "grad_norm": 9.30400276184082,
+ "learning_rate": 1.9374834217650814e-05,
+ "loss": 1.1865,
+ "step": 5129
+ },
+ {
+ "epoch": 13.978201634877385,
+ "grad_norm": 6.830667972564697,
+ "learning_rate": 1.9374527046597267e-05,
+ "loss": 0.9004,
+ "step": 5130
+ },
+ {
+ "epoch": 13.98092643051771,
+ "grad_norm": 7.258408069610596,
+ "learning_rate": 1.9374219802535207e-05,
+ "loss": 1.1782,
+ "step": 5131
+ },
+ {
+ "epoch": 13.983651226158038,
+ "grad_norm": 13.136831283569336,
+ "learning_rate": 1.9373912485467027e-05,
+ "loss": 1.0781,
+ "step": 5132
+ },
+ {
+ "epoch": 13.986376021798366,
+ "grad_norm": 8.29191780090332,
+ "learning_rate": 1.9373605095395114e-05,
+ "loss": 0.9336,
+ "step": 5133
+ },
+ {
+ "epoch": 13.989100817438691,
+ "grad_norm": 5.848191261291504,
+ "learning_rate": 1.937329763232187e-05,
+ "loss": 1.0808,
+ "step": 5134
+ },
+ {
+ "epoch": 13.991825613079019,
+ "grad_norm": 7.506320476531982,
+ "learning_rate": 1.9372990096249684e-05,
+ "loss": 1.014,
+ "step": 5135
+ },
+ {
+ "epoch": 13.994550408719347,
+ "grad_norm": 7.855730056762695,
+ "learning_rate": 1.9372682487180953e-05,
+ "loss": 0.9414,
+ "step": 5136
+ },
+ {
+ "epoch": 13.997275204359672,
+ "grad_norm": 7.058894634246826,
+ "learning_rate": 1.937237480511807e-05,
+ "loss": 0.8606,
+ "step": 5137
+ },
+ {
+ "epoch": 14.0,
+ "grad_norm": 5.880127429962158,
+ "learning_rate": 1.937206705006344e-05,
+ "loss": 0.8962,
+ "step": 5138
+ },
+ {
+ "epoch": 14.002724795640328,
+ "grad_norm": 7.4724249839782715,
+ "learning_rate": 1.9371759222019446e-05,
+ "loss": 1.0181,
+ "step": 5139
+ },
+ {
+ "epoch": 14.005449591280653,
+ "grad_norm": 6.089542865753174,
+ "learning_rate": 1.93714513209885e-05,
+ "loss": 1.0171,
+ "step": 5140
+ },
+ {
+ "epoch": 14.008174386920981,
+ "grad_norm": 8.010563850402832,
+ "learning_rate": 1.937114334697299e-05,
+ "loss": 0.9187,
+ "step": 5141
+ },
+ {
+ "epoch": 14.010899182561309,
+ "grad_norm": 9.726513862609863,
+ "learning_rate": 1.9370835299975312e-05,
+ "loss": 0.7335,
+ "step": 5142
+ },
+ {
+ "epoch": 14.013623978201634,
+ "grad_norm": 7.325217247009277,
+ "learning_rate": 1.9370527179997877e-05,
+ "loss": 0.8984,
+ "step": 5143
+ },
+ {
+ "epoch": 14.016348773841962,
+ "grad_norm": 6.243043422698975,
+ "learning_rate": 1.937021898704307e-05,
+ "loss": 0.8103,
+ "step": 5144
+ },
+ {
+ "epoch": 14.01907356948229,
+ "grad_norm": 6.776004791259766,
+ "learning_rate": 1.9369910721113307e-05,
+ "loss": 0.8518,
+ "step": 5145
+ },
+ {
+ "epoch": 14.021798365122615,
+ "grad_norm": 12.149592399597168,
+ "learning_rate": 1.936960238221098e-05,
+ "loss": 0.8827,
+ "step": 5146
+ },
+ {
+ "epoch": 14.024523160762943,
+ "grad_norm": 7.283549785614014,
+ "learning_rate": 1.936929397033849e-05,
+ "loss": 0.8717,
+ "step": 5147
+ },
+ {
+ "epoch": 14.02724795640327,
+ "grad_norm": 7.665163993835449,
+ "learning_rate": 1.936898548549824e-05,
+ "loss": 0.9991,
+ "step": 5148
+ },
+ {
+ "epoch": 14.029972752043596,
+ "grad_norm": 5.9212846755981445,
+ "learning_rate": 1.9368676927692633e-05,
+ "loss": 0.7067,
+ "step": 5149
+ },
+ {
+ "epoch": 14.032697547683924,
+ "grad_norm": 7.160329818725586,
+ "learning_rate": 1.936836829692407e-05,
+ "loss": 0.8361,
+ "step": 5150
+ },
+ {
+ "epoch": 14.035422343324251,
+ "grad_norm": 7.785150051116943,
+ "learning_rate": 1.9368059593194957e-05,
+ "loss": 0.7406,
+ "step": 5151
+ },
+ {
+ "epoch": 14.038147138964577,
+ "grad_norm": 5.683224678039551,
+ "learning_rate": 1.9367750816507697e-05,
+ "loss": 0.7941,
+ "step": 5152
+ },
+ {
+ "epoch": 14.040871934604905,
+ "grad_norm": 5.95977258682251,
+ "learning_rate": 1.9367441966864696e-05,
+ "loss": 0.9783,
+ "step": 5153
+ },
+ {
+ "epoch": 14.043596730245232,
+ "grad_norm": 5.52392053604126,
+ "learning_rate": 1.9367133044268357e-05,
+ "loss": 0.9491,
+ "step": 5154
+ },
+ {
+ "epoch": 14.046321525885558,
+ "grad_norm": 6.46225643157959,
+ "learning_rate": 1.9366824048721092e-05,
+ "loss": 0.7584,
+ "step": 5155
+ },
+ {
+ "epoch": 14.049046321525886,
+ "grad_norm": 7.566909313201904,
+ "learning_rate": 1.9366514980225297e-05,
+ "loss": 0.7461,
+ "step": 5156
+ },
+ {
+ "epoch": 14.051771117166213,
+ "grad_norm": 6.46348237991333,
+ "learning_rate": 1.9366205838783386e-05,
+ "loss": 0.7073,
+ "step": 5157
+ },
+ {
+ "epoch": 14.054495912806539,
+ "grad_norm": 7.458186149597168,
+ "learning_rate": 1.9365896624397767e-05,
+ "loss": 0.9371,
+ "step": 5158
+ },
+ {
+ "epoch": 14.057220708446867,
+ "grad_norm": 12.747864723205566,
+ "learning_rate": 1.9365587337070847e-05,
+ "loss": 0.9507,
+ "step": 5159
+ },
+ {
+ "epoch": 14.059945504087194,
+ "grad_norm": 7.096256256103516,
+ "learning_rate": 1.936527797680503e-05,
+ "loss": 0.8441,
+ "step": 5160
+ },
+ {
+ "epoch": 14.06267029972752,
+ "grad_norm": 7.402813911437988,
+ "learning_rate": 1.9364968543602734e-05,
+ "loss": 1.0602,
+ "step": 5161
+ },
+ {
+ "epoch": 14.065395095367847,
+ "grad_norm": 7.614022731781006,
+ "learning_rate": 1.9364659037466363e-05,
+ "loss": 0.924,
+ "step": 5162
+ },
+ {
+ "epoch": 14.068119891008175,
+ "grad_norm": 6.135637283325195,
+ "learning_rate": 1.936434945839833e-05,
+ "loss": 0.7841,
+ "step": 5163
+ },
+ {
+ "epoch": 14.0708446866485,
+ "grad_norm": 6.767189979553223,
+ "learning_rate": 1.9364039806401044e-05,
+ "loss": 0.8767,
+ "step": 5164
+ },
+ {
+ "epoch": 14.073569482288828,
+ "grad_norm": 7.126509189605713,
+ "learning_rate": 1.9363730081476917e-05,
+ "loss": 0.874,
+ "step": 5165
+ },
+ {
+ "epoch": 14.076294277929156,
+ "grad_norm": 7.417111396789551,
+ "learning_rate": 1.9363420283628362e-05,
+ "loss": 0.8395,
+ "step": 5166
+ },
+ {
+ "epoch": 14.079019073569482,
+ "grad_norm": 6.799693584442139,
+ "learning_rate": 1.9363110412857792e-05,
+ "loss": 0.9878,
+ "step": 5167
+ },
+ {
+ "epoch": 14.08174386920981,
+ "grad_norm": 6.034433364868164,
+ "learning_rate": 1.9362800469167616e-05,
+ "loss": 0.843,
+ "step": 5168
+ },
+ {
+ "epoch": 14.084468664850137,
+ "grad_norm": 6.7286810874938965,
+ "learning_rate": 1.9362490452560253e-05,
+ "loss": 0.9755,
+ "step": 5169
+ },
+ {
+ "epoch": 14.087193460490463,
+ "grad_norm": 6.4515862464904785,
+ "learning_rate": 1.936218036303812e-05,
+ "loss": 0.9814,
+ "step": 5170
+ },
+ {
+ "epoch": 14.08991825613079,
+ "grad_norm": 6.4896650314331055,
+ "learning_rate": 1.9361870200603622e-05,
+ "loss": 0.7784,
+ "step": 5171
+ },
+ {
+ "epoch": 14.092643051771118,
+ "grad_norm": 7.703859329223633,
+ "learning_rate": 1.9361559965259183e-05,
+ "loss": 0.8644,
+ "step": 5172
+ },
+ {
+ "epoch": 14.095367847411444,
+ "grad_norm": 9.590167999267578,
+ "learning_rate": 1.9361249657007212e-05,
+ "loss": 0.953,
+ "step": 5173
+ },
+ {
+ "epoch": 14.098092643051771,
+ "grad_norm": 6.623793601989746,
+ "learning_rate": 1.9360939275850136e-05,
+ "loss": 0.8085,
+ "step": 5174
+ },
+ {
+ "epoch": 14.100817438692099,
+ "grad_norm": 7.032140731811523,
+ "learning_rate": 1.9360628821790365e-05,
+ "loss": 0.8754,
+ "step": 5175
+ },
+ {
+ "epoch": 14.103542234332425,
+ "grad_norm": 6.530693054199219,
+ "learning_rate": 1.9360318294830316e-05,
+ "loss": 0.856,
+ "step": 5176
+ },
+ {
+ "epoch": 14.106267029972752,
+ "grad_norm": 7.487061500549316,
+ "learning_rate": 1.936000769497241e-05,
+ "loss": 0.8568,
+ "step": 5177
+ },
+ {
+ "epoch": 14.10899182561308,
+ "grad_norm": 7.303623676300049,
+ "learning_rate": 1.9359697022219065e-05,
+ "loss": 0.8862,
+ "step": 5178
+ },
+ {
+ "epoch": 14.111716621253406,
+ "grad_norm": 5.943865776062012,
+ "learning_rate": 1.93593862765727e-05,
+ "loss": 0.736,
+ "step": 5179
+ },
+ {
+ "epoch": 14.114441416893733,
+ "grad_norm": 5.827780723571777,
+ "learning_rate": 1.9359075458035737e-05,
+ "loss": 0.8232,
+ "step": 5180
+ },
+ {
+ "epoch": 14.11716621253406,
+ "grad_norm": 6.7490997314453125,
+ "learning_rate": 1.9358764566610598e-05,
+ "loss": 0.8878,
+ "step": 5181
+ },
+ {
+ "epoch": 14.119891008174386,
+ "grad_norm": 8.479043006896973,
+ "learning_rate": 1.9358453602299697e-05,
+ "loss": 0.8594,
+ "step": 5182
+ },
+ {
+ "epoch": 14.122615803814714,
+ "grad_norm": 7.4174089431762695,
+ "learning_rate": 1.9358142565105463e-05,
+ "loss": 1.0638,
+ "step": 5183
+ },
+ {
+ "epoch": 14.125340599455042,
+ "grad_norm": 7.801029682159424,
+ "learning_rate": 1.9357831455030315e-05,
+ "loss": 0.8523,
+ "step": 5184
+ },
+ {
+ "epoch": 14.128065395095367,
+ "grad_norm": 6.827777862548828,
+ "learning_rate": 1.9357520272076676e-05,
+ "loss": 0.7872,
+ "step": 5185
+ },
+ {
+ "epoch": 14.130790190735695,
+ "grad_norm": 6.49449348449707,
+ "learning_rate": 1.9357209016246972e-05,
+ "loss": 0.7926,
+ "step": 5186
+ },
+ {
+ "epoch": 14.133514986376023,
+ "grad_norm": 6.809727191925049,
+ "learning_rate": 1.9356897687543626e-05,
+ "loss": 0.9047,
+ "step": 5187
+ },
+ {
+ "epoch": 14.136239782016348,
+ "grad_norm": 8.636801719665527,
+ "learning_rate": 1.9356586285969062e-05,
+ "loss": 0.7983,
+ "step": 5188
+ },
+ {
+ "epoch": 14.138964577656676,
+ "grad_norm": 6.544518947601318,
+ "learning_rate": 1.9356274811525707e-05,
+ "loss": 0.8096,
+ "step": 5189
+ },
+ {
+ "epoch": 14.141689373297003,
+ "grad_norm": 6.552819728851318,
+ "learning_rate": 1.935596326421598e-05,
+ "loss": 0.8727,
+ "step": 5190
+ },
+ {
+ "epoch": 14.14441416893733,
+ "grad_norm": 6.428091049194336,
+ "learning_rate": 1.9355651644042312e-05,
+ "loss": 1.115,
+ "step": 5191
+ },
+ {
+ "epoch": 14.147138964577657,
+ "grad_norm": 7.542853832244873,
+ "learning_rate": 1.9355339951007133e-05,
+ "loss": 1.0242,
+ "step": 5192
+ },
+ {
+ "epoch": 14.149863760217984,
+ "grad_norm": 5.700021743774414,
+ "learning_rate": 1.935502818511287e-05,
+ "loss": 0.9309,
+ "step": 5193
+ },
+ {
+ "epoch": 14.15258855585831,
+ "grad_norm": 5.6223835945129395,
+ "learning_rate": 1.9354716346361947e-05,
+ "loss": 0.6384,
+ "step": 5194
+ },
+ {
+ "epoch": 14.155313351498638,
+ "grad_norm": 5.987936496734619,
+ "learning_rate": 1.935440443475679e-05,
+ "loss": 0.7936,
+ "step": 5195
+ },
+ {
+ "epoch": 14.158038147138965,
+ "grad_norm": 7.295841217041016,
+ "learning_rate": 1.9354092450299835e-05,
+ "loss": 1.1113,
+ "step": 5196
+ },
+ {
+ "epoch": 14.160762942779291,
+ "grad_norm": 5.969254016876221,
+ "learning_rate": 1.9353780392993512e-05,
+ "loss": 0.825,
+ "step": 5197
+ },
+ {
+ "epoch": 14.163487738419619,
+ "grad_norm": 6.158127784729004,
+ "learning_rate": 1.9353468262840242e-05,
+ "loss": 0.8519,
+ "step": 5198
+ },
+ {
+ "epoch": 14.166212534059946,
+ "grad_norm": 6.879106044769287,
+ "learning_rate": 1.9353156059842467e-05,
+ "loss": 0.9048,
+ "step": 5199
+ },
+ {
+ "epoch": 14.168937329700272,
+ "grad_norm": 6.491672515869141,
+ "learning_rate": 1.9352843784002613e-05,
+ "loss": 0.8423,
+ "step": 5200
+ },
+ {
+ "epoch": 14.1716621253406,
+ "grad_norm": 6.876067638397217,
+ "learning_rate": 1.9352531435323113e-05,
+ "loss": 0.728,
+ "step": 5201
+ },
+ {
+ "epoch": 14.174386920980927,
+ "grad_norm": 6.470943450927734,
+ "learning_rate": 1.9352219013806396e-05,
+ "loss": 0.8417,
+ "step": 5202
+ },
+ {
+ "epoch": 14.177111716621253,
+ "grad_norm": 7.748780250549316,
+ "learning_rate": 1.93519065194549e-05,
+ "loss": 0.896,
+ "step": 5203
+ },
+ {
+ "epoch": 14.17983651226158,
+ "grad_norm": 11.352484703063965,
+ "learning_rate": 1.935159395227106e-05,
+ "loss": 0.8801,
+ "step": 5204
+ },
+ {
+ "epoch": 14.182561307901908,
+ "grad_norm": 10.383734703063965,
+ "learning_rate": 1.9351281312257304e-05,
+ "loss": 0.8064,
+ "step": 5205
+ },
+ {
+ "epoch": 14.185286103542234,
+ "grad_norm": 7.815541744232178,
+ "learning_rate": 1.9350968599416073e-05,
+ "loss": 1.1243,
+ "step": 5206
+ },
+ {
+ "epoch": 14.188010899182562,
+ "grad_norm": 7.227829933166504,
+ "learning_rate": 1.9350655813749795e-05,
+ "loss": 0.849,
+ "step": 5207
+ },
+ {
+ "epoch": 14.190735694822889,
+ "grad_norm": 6.127223014831543,
+ "learning_rate": 1.9350342955260913e-05,
+ "loss": 0.7124,
+ "step": 5208
+ },
+ {
+ "epoch": 14.193460490463215,
+ "grad_norm": 7.1068949699401855,
+ "learning_rate": 1.935003002395186e-05,
+ "loss": 0.8885,
+ "step": 5209
+ },
+ {
+ "epoch": 14.196185286103542,
+ "grad_norm": 7.92221736907959,
+ "learning_rate": 1.9349717019825076e-05,
+ "loss": 0.989,
+ "step": 5210
+ },
+ {
+ "epoch": 14.19891008174387,
+ "grad_norm": 6.467373847961426,
+ "learning_rate": 1.9349403942882995e-05,
+ "loss": 0.8878,
+ "step": 5211
+ },
+ {
+ "epoch": 14.201634877384196,
+ "grad_norm": 6.83078670501709,
+ "learning_rate": 1.934909079312806e-05,
+ "loss": 0.9343,
+ "step": 5212
+ },
+ {
+ "epoch": 14.204359673024523,
+ "grad_norm": 6.337018013000488,
+ "learning_rate": 1.9348777570562705e-05,
+ "loss": 1.0657,
+ "step": 5213
+ },
+ {
+ "epoch": 14.207084468664851,
+ "grad_norm": 6.715846538543701,
+ "learning_rate": 1.934846427518937e-05,
+ "loss": 0.8439,
+ "step": 5214
+ },
+ {
+ "epoch": 14.209809264305177,
+ "grad_norm": 7.647075653076172,
+ "learning_rate": 1.9348150907010495e-05,
+ "loss": 0.901,
+ "step": 5215
+ },
+ {
+ "epoch": 14.212534059945504,
+ "grad_norm": 7.013647079467773,
+ "learning_rate": 1.9347837466028524e-05,
+ "loss": 0.8993,
+ "step": 5216
+ },
+ {
+ "epoch": 14.215258855585832,
+ "grad_norm": 7.757715225219727,
+ "learning_rate": 1.9347523952245898e-05,
+ "loss": 0.9277,
+ "step": 5217
+ },
+ {
+ "epoch": 14.217983651226158,
+ "grad_norm": 7.30740213394165,
+ "learning_rate": 1.9347210365665052e-05,
+ "loss": 0.7382,
+ "step": 5218
+ },
+ {
+ "epoch": 14.220708446866485,
+ "grad_norm": 7.803001880645752,
+ "learning_rate": 1.9346896706288433e-05,
+ "loss": 1.0208,
+ "step": 5219
+ },
+ {
+ "epoch": 14.223433242506813,
+ "grad_norm": 6.9735636711120605,
+ "learning_rate": 1.9346582974118484e-05,
+ "loss": 1.0549,
+ "step": 5220
+ },
+ {
+ "epoch": 14.226158038147139,
+ "grad_norm": 6.851255893707275,
+ "learning_rate": 1.934626916915765e-05,
+ "loss": 0.9052,
+ "step": 5221
+ },
+ {
+ "epoch": 14.228882833787466,
+ "grad_norm": 7.544332981109619,
+ "learning_rate": 1.9345955291408368e-05,
+ "loss": 0.9615,
+ "step": 5222
+ },
+ {
+ "epoch": 14.231607629427794,
+ "grad_norm": 7.781844139099121,
+ "learning_rate": 1.934564134087309e-05,
+ "loss": 1.0048,
+ "step": 5223
+ },
+ {
+ "epoch": 14.23433242506812,
+ "grad_norm": 5.873414993286133,
+ "learning_rate": 1.9345327317554258e-05,
+ "loss": 0.754,
+ "step": 5224
+ },
+ {
+ "epoch": 14.237057220708447,
+ "grad_norm": 6.835072040557861,
+ "learning_rate": 1.934501322145432e-05,
+ "loss": 0.7856,
+ "step": 5225
+ },
+ {
+ "epoch": 14.239782016348773,
+ "grad_norm": 6.703495502471924,
+ "learning_rate": 1.9344699052575713e-05,
+ "loss": 0.8,
+ "step": 5226
+ },
+ {
+ "epoch": 14.2425068119891,
+ "grad_norm": 8.074563980102539,
+ "learning_rate": 1.9344384810920894e-05,
+ "loss": 0.8605,
+ "step": 5227
+ },
+ {
+ "epoch": 14.245231607629428,
+ "grad_norm": 6.096956253051758,
+ "learning_rate": 1.9344070496492307e-05,
+ "loss": 0.863,
+ "step": 5228
+ },
+ {
+ "epoch": 14.247956403269754,
+ "grad_norm": 7.368281364440918,
+ "learning_rate": 1.93437561092924e-05,
+ "loss": 0.7742,
+ "step": 5229
+ },
+ {
+ "epoch": 14.250681198910081,
+ "grad_norm": 12.391910552978516,
+ "learning_rate": 1.9343441649323623e-05,
+ "loss": 0.9924,
+ "step": 5230
+ },
+ {
+ "epoch": 14.253405994550409,
+ "grad_norm": 7.035614013671875,
+ "learning_rate": 1.9343127116588422e-05,
+ "loss": 0.8539,
+ "step": 5231
+ },
+ {
+ "epoch": 14.256130790190735,
+ "grad_norm": 8.192720413208008,
+ "learning_rate": 1.9342812511089245e-05,
+ "loss": 0.8695,
+ "step": 5232
+ },
+ {
+ "epoch": 14.258855585831062,
+ "grad_norm": 7.5290045738220215,
+ "learning_rate": 1.9342497832828545e-05,
+ "loss": 1.0352,
+ "step": 5233
+ },
+ {
+ "epoch": 14.26158038147139,
+ "grad_norm": 7.3202009201049805,
+ "learning_rate": 1.9342183081808776e-05,
+ "loss": 0.7771,
+ "step": 5234
+ },
+ {
+ "epoch": 14.264305177111716,
+ "grad_norm": 9.09864330291748,
+ "learning_rate": 1.9341868258032383e-05,
+ "loss": 0.9066,
+ "step": 5235
+ },
+ {
+ "epoch": 14.267029972752043,
+ "grad_norm": 8.371217727661133,
+ "learning_rate": 1.9341553361501823e-05,
+ "loss": 0.8461,
+ "step": 5236
+ },
+ {
+ "epoch": 14.269754768392371,
+ "grad_norm": 6.800161361694336,
+ "learning_rate": 1.9341238392219546e-05,
+ "loss": 0.7509,
+ "step": 5237
+ },
+ {
+ "epoch": 14.272479564032697,
+ "grad_norm": 8.149608612060547,
+ "learning_rate": 1.9340923350188004e-05,
+ "loss": 0.8162,
+ "step": 5238
+ },
+ {
+ "epoch": 14.275204359673024,
+ "grad_norm": 8.0971097946167,
+ "learning_rate": 1.934060823540965e-05,
+ "loss": 0.8064,
+ "step": 5239
+ },
+ {
+ "epoch": 14.277929155313352,
+ "grad_norm": 7.733577728271484,
+ "learning_rate": 1.9340293047886943e-05,
+ "loss": 0.9824,
+ "step": 5240
+ },
+ {
+ "epoch": 14.280653950953678,
+ "grad_norm": 6.68411922454834,
+ "learning_rate": 1.9339977787622332e-05,
+ "loss": 0.8723,
+ "step": 5241
+ },
+ {
+ "epoch": 14.283378746594005,
+ "grad_norm": 5.8032450675964355,
+ "learning_rate": 1.9339662454618277e-05,
+ "loss": 0.8684,
+ "step": 5242
+ },
+ {
+ "epoch": 14.286103542234333,
+ "grad_norm": 7.260004997253418,
+ "learning_rate": 1.9339347048877228e-05,
+ "loss": 0.7178,
+ "step": 5243
+ },
+ {
+ "epoch": 14.288828337874659,
+ "grad_norm": 7.4702467918396,
+ "learning_rate": 1.9339031570401647e-05,
+ "loss": 0.9866,
+ "step": 5244
+ },
+ {
+ "epoch": 14.291553133514986,
+ "grad_norm": 8.059065818786621,
+ "learning_rate": 1.933871601919399e-05,
+ "loss": 1.4344,
+ "step": 5245
+ },
+ {
+ "epoch": 14.294277929155314,
+ "grad_norm": 8.49100399017334,
+ "learning_rate": 1.9338400395256708e-05,
+ "loss": 1.0409,
+ "step": 5246
+ },
+ {
+ "epoch": 14.29700272479564,
+ "grad_norm": 7.083617210388184,
+ "learning_rate": 1.9338084698592266e-05,
+ "loss": 0.9543,
+ "step": 5247
+ },
+ {
+ "epoch": 14.299727520435967,
+ "grad_norm": 7.114346027374268,
+ "learning_rate": 1.9337768929203124e-05,
+ "loss": 0.7516,
+ "step": 5248
+ },
+ {
+ "epoch": 14.302452316076295,
+ "grad_norm": 5.945745944976807,
+ "learning_rate": 1.933745308709174e-05,
+ "loss": 0.8759,
+ "step": 5249
+ },
+ {
+ "epoch": 14.30517711171662,
+ "grad_norm": 9.192170143127441,
+ "learning_rate": 1.9337137172260568e-05,
+ "loss": 1.0043,
+ "step": 5250
+ },
+ {
+ "epoch": 14.307901907356948,
+ "grad_norm": 6.73648738861084,
+ "learning_rate": 1.933682118471207e-05,
+ "loss": 0.8384,
+ "step": 5251
+ },
+ {
+ "epoch": 14.310626702997276,
+ "grad_norm": 7.346246242523193,
+ "learning_rate": 1.9336505124448714e-05,
+ "loss": 0.9014,
+ "step": 5252
+ },
+ {
+ "epoch": 14.313351498637601,
+ "grad_norm": 8.15287971496582,
+ "learning_rate": 1.9336188991472956e-05,
+ "loss": 0.8849,
+ "step": 5253
+ },
+ {
+ "epoch": 14.316076294277929,
+ "grad_norm": 6.414109706878662,
+ "learning_rate": 1.9335872785787257e-05,
+ "loss": 0.8748,
+ "step": 5254
+ },
+ {
+ "epoch": 14.318801089918257,
+ "grad_norm": 7.597804546356201,
+ "learning_rate": 1.933555650739408e-05,
+ "loss": 1.0706,
+ "step": 5255
+ },
+ {
+ "epoch": 14.321525885558582,
+ "grad_norm": 7.46559476852417,
+ "learning_rate": 1.933524015629589e-05,
+ "loss": 1.0236,
+ "step": 5256
+ },
+ {
+ "epoch": 14.32425068119891,
+ "grad_norm": 7.691039562225342,
+ "learning_rate": 1.9334923732495153e-05,
+ "loss": 1.0046,
+ "step": 5257
+ },
+ {
+ "epoch": 14.326975476839237,
+ "grad_norm": 8.448208808898926,
+ "learning_rate": 1.9334607235994328e-05,
+ "loss": 0.9407,
+ "step": 5258
+ },
+ {
+ "epoch": 14.329700272479563,
+ "grad_norm": 6.055136203765869,
+ "learning_rate": 1.933429066679588e-05,
+ "loss": 0.8368,
+ "step": 5259
+ },
+ {
+ "epoch": 14.33242506811989,
+ "grad_norm": 7.799905776977539,
+ "learning_rate": 1.9333974024902278e-05,
+ "loss": 0.8485,
+ "step": 5260
+ },
+ {
+ "epoch": 14.335149863760218,
+ "grad_norm": 7.75402307510376,
+ "learning_rate": 1.933365731031599e-05,
+ "loss": 0.9211,
+ "step": 5261
+ },
+ {
+ "epoch": 14.337874659400544,
+ "grad_norm": 6.094085693359375,
+ "learning_rate": 1.9333340523039477e-05,
+ "loss": 0.9147,
+ "step": 5262
+ },
+ {
+ "epoch": 14.340599455040872,
+ "grad_norm": 6.7752180099487305,
+ "learning_rate": 1.9333023663075208e-05,
+ "loss": 0.7588,
+ "step": 5263
+ },
+ {
+ "epoch": 14.3433242506812,
+ "grad_norm": 7.056731224060059,
+ "learning_rate": 1.9332706730425654e-05,
+ "loss": 0.9907,
+ "step": 5264
+ },
+ {
+ "epoch": 14.346049046321525,
+ "grad_norm": 6.896046161651611,
+ "learning_rate": 1.9332389725093276e-05,
+ "loss": 0.9024,
+ "step": 5265
+ },
+ {
+ "epoch": 14.348773841961853,
+ "grad_norm": 6.126890659332275,
+ "learning_rate": 1.9332072647080548e-05,
+ "loss": 0.7236,
+ "step": 5266
+ },
+ {
+ "epoch": 14.35149863760218,
+ "grad_norm": 6.809377193450928,
+ "learning_rate": 1.933175549638994e-05,
+ "loss": 0.7505,
+ "step": 5267
+ },
+ {
+ "epoch": 14.354223433242506,
+ "grad_norm": 7.401408672332764,
+ "learning_rate": 1.933143827302392e-05,
+ "loss": 0.8175,
+ "step": 5268
+ },
+ {
+ "epoch": 14.356948228882834,
+ "grad_norm": 7.856297969818115,
+ "learning_rate": 1.9331120976984962e-05,
+ "loss": 0.804,
+ "step": 5269
+ },
+ {
+ "epoch": 14.359673024523161,
+ "grad_norm": 9.381133079528809,
+ "learning_rate": 1.933080360827553e-05,
+ "loss": 0.9163,
+ "step": 5270
+ },
+ {
+ "epoch": 14.362397820163487,
+ "grad_norm": 6.617667198181152,
+ "learning_rate": 1.93304861668981e-05,
+ "loss": 0.847,
+ "step": 5271
+ },
+ {
+ "epoch": 14.365122615803815,
+ "grad_norm": 5.891904354095459,
+ "learning_rate": 1.933016865285515e-05,
+ "loss": 0.9486,
+ "step": 5272
+ },
+ {
+ "epoch": 14.367847411444142,
+ "grad_norm": 6.372308731079102,
+ "learning_rate": 1.9329851066149138e-05,
+ "loss": 0.882,
+ "step": 5273
+ },
+ {
+ "epoch": 14.370572207084468,
+ "grad_norm": 10.835223197937012,
+ "learning_rate": 1.932953340678255e-05,
+ "loss": 0.9453,
+ "step": 5274
+ },
+ {
+ "epoch": 14.373297002724795,
+ "grad_norm": 7.374895095825195,
+ "learning_rate": 1.9329215674757856e-05,
+ "loss": 0.795,
+ "step": 5275
+ },
+ {
+ "epoch": 14.376021798365123,
+ "grad_norm": 6.009027004241943,
+ "learning_rate": 1.932889787007753e-05,
+ "loss": 0.7967,
+ "step": 5276
+ },
+ {
+ "epoch": 14.378746594005449,
+ "grad_norm": 6.169981002807617,
+ "learning_rate": 1.932857999274405e-05,
+ "loss": 0.8424,
+ "step": 5277
+ },
+ {
+ "epoch": 14.381471389645776,
+ "grad_norm": 7.074678421020508,
+ "learning_rate": 1.932826204275989e-05,
+ "loss": 0.7505,
+ "step": 5278
+ },
+ {
+ "epoch": 14.384196185286104,
+ "grad_norm": 7.290410041809082,
+ "learning_rate": 1.932794402012752e-05,
+ "loss": 0.896,
+ "step": 5279
+ },
+ {
+ "epoch": 14.38692098092643,
+ "grad_norm": 7.822486400604248,
+ "learning_rate": 1.932762592484943e-05,
+ "loss": 1.0909,
+ "step": 5280
+ },
+ {
+ "epoch": 14.389645776566757,
+ "grad_norm": 6.871390342712402,
+ "learning_rate": 1.9327307756928082e-05,
+ "loss": 0.9905,
+ "step": 5281
+ },
+ {
+ "epoch": 14.392370572207085,
+ "grad_norm": 8.101964950561523,
+ "learning_rate": 1.9326989516365965e-05,
+ "loss": 1.0891,
+ "step": 5282
+ },
+ {
+ "epoch": 14.39509536784741,
+ "grad_norm": 6.259859561920166,
+ "learning_rate": 1.9326671203165555e-05,
+ "loss": 0.8612,
+ "step": 5283
+ },
+ {
+ "epoch": 14.397820163487738,
+ "grad_norm": 13.807276725769043,
+ "learning_rate": 1.9326352817329325e-05,
+ "loss": 1.0312,
+ "step": 5284
+ },
+ {
+ "epoch": 14.400544959128066,
+ "grad_norm": 6.838610649108887,
+ "learning_rate": 1.9326034358859763e-05,
+ "loss": 0.8702,
+ "step": 5285
+ },
+ {
+ "epoch": 14.403269754768392,
+ "grad_norm": 6.238323211669922,
+ "learning_rate": 1.9325715827759347e-05,
+ "loss": 0.9851,
+ "step": 5286
+ },
+ {
+ "epoch": 14.40599455040872,
+ "grad_norm": 8.663369178771973,
+ "learning_rate": 1.9325397224030556e-05,
+ "loss": 0.8279,
+ "step": 5287
+ },
+ {
+ "epoch": 14.408719346049047,
+ "grad_norm": 6.031283378601074,
+ "learning_rate": 1.932507854767587e-05,
+ "loss": 0.9431,
+ "step": 5288
+ },
+ {
+ "epoch": 14.411444141689373,
+ "grad_norm": 6.534623622894287,
+ "learning_rate": 1.932475979869777e-05,
+ "loss": 0.9147,
+ "step": 5289
+ },
+ {
+ "epoch": 14.4141689373297,
+ "grad_norm": 7.298571586608887,
+ "learning_rate": 1.9324440977098743e-05,
+ "loss": 1.0977,
+ "step": 5290
+ },
+ {
+ "epoch": 14.416893732970028,
+ "grad_norm": 6.757731914520264,
+ "learning_rate": 1.9324122082881273e-05,
+ "loss": 0.8654,
+ "step": 5291
+ },
+ {
+ "epoch": 14.419618528610354,
+ "grad_norm": 6.29318380355835,
+ "learning_rate": 1.9323803116047835e-05,
+ "loss": 0.834,
+ "step": 5292
+ },
+ {
+ "epoch": 14.422343324250681,
+ "grad_norm": 8.42442512512207,
+ "learning_rate": 1.932348407660092e-05,
+ "loss": 1.0122,
+ "step": 5293
+ },
+ {
+ "epoch": 14.425068119891009,
+ "grad_norm": 6.719336986541748,
+ "learning_rate": 1.932316496454301e-05,
+ "loss": 0.7056,
+ "step": 5294
+ },
+ {
+ "epoch": 14.427792915531334,
+ "grad_norm": 7.850996017456055,
+ "learning_rate": 1.9322845779876594e-05,
+ "loss": 0.9302,
+ "step": 5295
+ },
+ {
+ "epoch": 14.430517711171662,
+ "grad_norm": 9.901152610778809,
+ "learning_rate": 1.9322526522604158e-05,
+ "loss": 0.8333,
+ "step": 5296
+ },
+ {
+ "epoch": 14.43324250681199,
+ "grad_norm": 7.915676593780518,
+ "learning_rate": 1.932220719272818e-05,
+ "loss": 0.8522,
+ "step": 5297
+ },
+ {
+ "epoch": 14.435967302452315,
+ "grad_norm": 7.401548862457275,
+ "learning_rate": 1.932188779025115e-05,
+ "loss": 0.9188,
+ "step": 5298
+ },
+ {
+ "epoch": 14.438692098092643,
+ "grad_norm": 7.331689834594727,
+ "learning_rate": 1.9321568315175566e-05,
+ "loss": 0.8483,
+ "step": 5299
+ },
+ {
+ "epoch": 14.44141689373297,
+ "grad_norm": 7.091240882873535,
+ "learning_rate": 1.93212487675039e-05,
+ "loss": 0.8898,
+ "step": 5300
+ },
+ {
+ "epoch": 14.444141689373296,
+ "grad_norm": 6.826676368713379,
+ "learning_rate": 1.9320929147238656e-05,
+ "loss": 0.7787,
+ "step": 5301
+ },
+ {
+ "epoch": 14.446866485013624,
+ "grad_norm": 6.954625129699707,
+ "learning_rate": 1.9320609454382314e-05,
+ "loss": 0.8833,
+ "step": 5302
+ },
+ {
+ "epoch": 14.449591280653951,
+ "grad_norm": 8.827156066894531,
+ "learning_rate": 1.932028968893736e-05,
+ "loss": 0.8628,
+ "step": 5303
+ },
+ {
+ "epoch": 14.452316076294277,
+ "grad_norm": 6.41854190826416,
+ "learning_rate": 1.9319969850906293e-05,
+ "loss": 0.8109,
+ "step": 5304
+ },
+ {
+ "epoch": 14.455040871934605,
+ "grad_norm": 7.307374954223633,
+ "learning_rate": 1.9319649940291605e-05,
+ "loss": 0.8356,
+ "step": 5305
+ },
+ {
+ "epoch": 14.457765667574932,
+ "grad_norm": 7.88008975982666,
+ "learning_rate": 1.9319329957095777e-05,
+ "loss": 1.2778,
+ "step": 5306
+ },
+ {
+ "epoch": 14.460490463215258,
+ "grad_norm": 5.459295749664307,
+ "learning_rate": 1.9319009901321312e-05,
+ "loss": 0.7239,
+ "step": 5307
+ },
+ {
+ "epoch": 14.463215258855586,
+ "grad_norm": 7.055986404418945,
+ "learning_rate": 1.9318689772970695e-05,
+ "loss": 0.8497,
+ "step": 5308
+ },
+ {
+ "epoch": 14.465940054495913,
+ "grad_norm": 6.684093475341797,
+ "learning_rate": 1.9318369572046425e-05,
+ "loss": 0.8337,
+ "step": 5309
+ },
+ {
+ "epoch": 14.46866485013624,
+ "grad_norm": 7.021854400634766,
+ "learning_rate": 1.931804929855099e-05,
+ "loss": 1.1449,
+ "step": 5310
+ },
+ {
+ "epoch": 14.471389645776567,
+ "grad_norm": 6.605978012084961,
+ "learning_rate": 1.931772895248689e-05,
+ "loss": 0.8884,
+ "step": 5311
+ },
+ {
+ "epoch": 14.474114441416894,
+ "grad_norm": 9.01488971710205,
+ "learning_rate": 1.931740853385661e-05,
+ "loss": 0.85,
+ "step": 5312
+ },
+ {
+ "epoch": 14.47683923705722,
+ "grad_norm": 6.907607078552246,
+ "learning_rate": 1.9317088042662656e-05,
+ "loss": 1.0256,
+ "step": 5313
+ },
+ {
+ "epoch": 14.479564032697548,
+ "grad_norm": 7.363582134246826,
+ "learning_rate": 1.931676747890752e-05,
+ "loss": 0.8414,
+ "step": 5314
+ },
+ {
+ "epoch": 14.482288828337875,
+ "grad_norm": 6.794772148132324,
+ "learning_rate": 1.93164468425937e-05,
+ "loss": 1.022,
+ "step": 5315
+ },
+ {
+ "epoch": 14.485013623978201,
+ "grad_norm": 7.065379619598389,
+ "learning_rate": 1.9316126133723694e-05,
+ "loss": 0.8495,
+ "step": 5316
+ },
+ {
+ "epoch": 14.487738419618529,
+ "grad_norm": 8.827138900756836,
+ "learning_rate": 1.9315805352299996e-05,
+ "loss": 0.7312,
+ "step": 5317
+ },
+ {
+ "epoch": 14.490463215258856,
+ "grad_norm": 6.604440212249756,
+ "learning_rate": 1.9315484498325103e-05,
+ "loss": 0.7959,
+ "step": 5318
+ },
+ {
+ "epoch": 14.493188010899182,
+ "grad_norm": 6.601905822753906,
+ "learning_rate": 1.9315163571801517e-05,
+ "loss": 0.8354,
+ "step": 5319
+ },
+ {
+ "epoch": 14.49591280653951,
+ "grad_norm": 7.269133567810059,
+ "learning_rate": 1.9314842572731737e-05,
+ "loss": 0.9791,
+ "step": 5320
+ },
+ {
+ "epoch": 14.498637602179837,
+ "grad_norm": 6.492870330810547,
+ "learning_rate": 1.9314521501118265e-05,
+ "loss": 0.8372,
+ "step": 5321
+ },
+ {
+ "epoch": 14.501362397820163,
+ "grad_norm": 6.616516590118408,
+ "learning_rate": 1.93142003569636e-05,
+ "loss": 0.9645,
+ "step": 5322
+ },
+ {
+ "epoch": 14.50408719346049,
+ "grad_norm": 5.674969673156738,
+ "learning_rate": 1.931387914027024e-05,
+ "loss": 0.8497,
+ "step": 5323
+ },
+ {
+ "epoch": 14.506811989100818,
+ "grad_norm": 6.296539306640625,
+ "learning_rate": 1.931355785104069e-05,
+ "loss": 0.825,
+ "step": 5324
+ },
+ {
+ "epoch": 14.509536784741144,
+ "grad_norm": 20.166385650634766,
+ "learning_rate": 1.931323648927745e-05,
+ "loss": 0.9806,
+ "step": 5325
+ },
+ {
+ "epoch": 14.512261580381471,
+ "grad_norm": 6.7371439933776855,
+ "learning_rate": 1.9312915054983023e-05,
+ "loss": 0.8998,
+ "step": 5326
+ },
+ {
+ "epoch": 14.514986376021799,
+ "grad_norm": 7.497361660003662,
+ "learning_rate": 1.9312593548159916e-05,
+ "loss": 0.9914,
+ "step": 5327
+ },
+ {
+ "epoch": 14.517711171662125,
+ "grad_norm": 8.213802337646484,
+ "learning_rate": 1.931227196881063e-05,
+ "loss": 0.8677,
+ "step": 5328
+ },
+ {
+ "epoch": 14.520435967302452,
+ "grad_norm": 7.080141067504883,
+ "learning_rate": 1.9311950316937666e-05,
+ "loss": 0.7308,
+ "step": 5329
+ },
+ {
+ "epoch": 14.52316076294278,
+ "grad_norm": 9.280269622802734,
+ "learning_rate": 1.9311628592543538e-05,
+ "loss": 0.9652,
+ "step": 5330
+ },
+ {
+ "epoch": 14.525885558583106,
+ "grad_norm": 6.849862098693848,
+ "learning_rate": 1.9311306795630745e-05,
+ "loss": 1.0118,
+ "step": 5331
+ },
+ {
+ "epoch": 14.528610354223433,
+ "grad_norm": 6.310232639312744,
+ "learning_rate": 1.931098492620179e-05,
+ "loss": 0.8436,
+ "step": 5332
+ },
+ {
+ "epoch": 14.53133514986376,
+ "grad_norm": 8.924739837646484,
+ "learning_rate": 1.9310662984259187e-05,
+ "loss": 0.9111,
+ "step": 5333
+ },
+ {
+ "epoch": 14.534059945504087,
+ "grad_norm": 7.729876518249512,
+ "learning_rate": 1.931034096980544e-05,
+ "loss": 0.6868,
+ "step": 5334
+ },
+ {
+ "epoch": 14.536784741144414,
+ "grad_norm": 5.489418983459473,
+ "learning_rate": 1.931001888284306e-05,
+ "loss": 0.6805,
+ "step": 5335
+ },
+ {
+ "epoch": 14.539509536784742,
+ "grad_norm": 7.440989017486572,
+ "learning_rate": 1.9309696723374548e-05,
+ "loss": 0.9846,
+ "step": 5336
+ },
+ {
+ "epoch": 14.542234332425068,
+ "grad_norm": 6.7116475105285645,
+ "learning_rate": 1.930937449140242e-05,
+ "loss": 0.8572,
+ "step": 5337
+ },
+ {
+ "epoch": 14.544959128065395,
+ "grad_norm": 7.820311546325684,
+ "learning_rate": 1.9309052186929184e-05,
+ "loss": 0.9161,
+ "step": 5338
+ },
+ {
+ "epoch": 14.547683923705723,
+ "grad_norm": 6.997105121612549,
+ "learning_rate": 1.930872980995735e-05,
+ "loss": 0.7952,
+ "step": 5339
+ },
+ {
+ "epoch": 14.550408719346049,
+ "grad_norm": 10.520129203796387,
+ "learning_rate": 1.930840736048942e-05,
+ "loss": 0.7191,
+ "step": 5340
+ },
+ {
+ "epoch": 14.553133514986376,
+ "grad_norm": 6.770661354064941,
+ "learning_rate": 1.9308084838527922e-05,
+ "loss": 0.8959,
+ "step": 5341
+ },
+ {
+ "epoch": 14.555858310626704,
+ "grad_norm": 6.75016450881958,
+ "learning_rate": 1.9307762244075357e-05,
+ "loss": 1.054,
+ "step": 5342
+ },
+ {
+ "epoch": 14.55858310626703,
+ "grad_norm": 8.341164588928223,
+ "learning_rate": 1.9307439577134238e-05,
+ "loss": 1.0342,
+ "step": 5343
+ },
+ {
+ "epoch": 14.561307901907357,
+ "grad_norm": 7.20265007019043,
+ "learning_rate": 1.9307116837707078e-05,
+ "loss": 1.1128,
+ "step": 5344
+ },
+ {
+ "epoch": 14.564032697547685,
+ "grad_norm": 7.443593502044678,
+ "learning_rate": 1.9306794025796395e-05,
+ "loss": 0.9138,
+ "step": 5345
+ },
+ {
+ "epoch": 14.56675749318801,
+ "grad_norm": 17.75119400024414,
+ "learning_rate": 1.9306471141404696e-05,
+ "loss": 0.7974,
+ "step": 5346
+ },
+ {
+ "epoch": 14.569482288828338,
+ "grad_norm": 5.7164835929870605,
+ "learning_rate": 1.93061481845345e-05,
+ "loss": 0.9545,
+ "step": 5347
+ },
+ {
+ "epoch": 14.572207084468666,
+ "grad_norm": 9.015913009643555,
+ "learning_rate": 1.9305825155188324e-05,
+ "loss": 0.8982,
+ "step": 5348
+ },
+ {
+ "epoch": 14.574931880108991,
+ "grad_norm": 6.98476505279541,
+ "learning_rate": 1.930550205336868e-05,
+ "loss": 0.8571,
+ "step": 5349
+ },
+ {
+ "epoch": 14.577656675749319,
+ "grad_norm": 7.362364292144775,
+ "learning_rate": 1.9305178879078083e-05,
+ "loss": 0.8394,
+ "step": 5350
+ },
+ {
+ "epoch": 14.580381471389646,
+ "grad_norm": 7.211151123046875,
+ "learning_rate": 1.9304855632319052e-05,
+ "loss": 0.9617,
+ "step": 5351
+ },
+ {
+ "epoch": 14.583106267029972,
+ "grad_norm": 8.036325454711914,
+ "learning_rate": 1.930453231309411e-05,
+ "loss": 0.8229,
+ "step": 5352
+ },
+ {
+ "epoch": 14.5858310626703,
+ "grad_norm": 8.503474235534668,
+ "learning_rate": 1.9304208921405766e-05,
+ "loss": 0.9106,
+ "step": 5353
+ },
+ {
+ "epoch": 14.588555858310627,
+ "grad_norm": 6.906664848327637,
+ "learning_rate": 1.9303885457256543e-05,
+ "loss": 0.7217,
+ "step": 5354
+ },
+ {
+ "epoch": 14.591280653950953,
+ "grad_norm": 6.982332229614258,
+ "learning_rate": 1.930356192064896e-05,
+ "loss": 0.7982,
+ "step": 5355
+ },
+ {
+ "epoch": 14.59400544959128,
+ "grad_norm": 7.324875831604004,
+ "learning_rate": 1.9303238311585534e-05,
+ "loss": 0.9591,
+ "step": 5356
+ },
+ {
+ "epoch": 14.596730245231608,
+ "grad_norm": 8.100831985473633,
+ "learning_rate": 1.930291463006879e-05,
+ "loss": 0.8888,
+ "step": 5357
+ },
+ {
+ "epoch": 14.599455040871934,
+ "grad_norm": 6.132966041564941,
+ "learning_rate": 1.9302590876101248e-05,
+ "loss": 0.8403,
+ "step": 5358
+ },
+ {
+ "epoch": 14.602179836512262,
+ "grad_norm": 8.492993354797363,
+ "learning_rate": 1.9302267049685424e-05,
+ "loss": 0.889,
+ "step": 5359
+ },
+ {
+ "epoch": 14.60490463215259,
+ "grad_norm": 7.630669116973877,
+ "learning_rate": 1.9301943150823845e-05,
+ "loss": 0.9507,
+ "step": 5360
+ },
+ {
+ "epoch": 14.607629427792915,
+ "grad_norm": 8.701238632202148,
+ "learning_rate": 1.930161917951903e-05,
+ "loss": 0.8563,
+ "step": 5361
+ },
+ {
+ "epoch": 14.610354223433243,
+ "grad_norm": 6.044345855712891,
+ "learning_rate": 1.9301295135773507e-05,
+ "loss": 0.9822,
+ "step": 5362
+ },
+ {
+ "epoch": 14.61307901907357,
+ "grad_norm": 6.527432441711426,
+ "learning_rate": 1.9300971019589796e-05,
+ "loss": 0.8234,
+ "step": 5363
+ },
+ {
+ "epoch": 14.615803814713896,
+ "grad_norm": 6.889712333679199,
+ "learning_rate": 1.9300646830970423e-05,
+ "loss": 0.9165,
+ "step": 5364
+ },
+ {
+ "epoch": 14.618528610354224,
+ "grad_norm": 7.296000003814697,
+ "learning_rate": 1.930032256991791e-05,
+ "loss": 1.1511,
+ "step": 5365
+ },
+ {
+ "epoch": 14.621253405994551,
+ "grad_norm": 7.893062591552734,
+ "learning_rate": 1.9299998236434786e-05,
+ "loss": 0.9232,
+ "step": 5366
+ },
+ {
+ "epoch": 14.623978201634877,
+ "grad_norm": 7.144567966461182,
+ "learning_rate": 1.9299673830523574e-05,
+ "loss": 1.0099,
+ "step": 5367
+ },
+ {
+ "epoch": 14.626702997275205,
+ "grad_norm": 11.318686485290527,
+ "learning_rate": 1.9299349352186803e-05,
+ "loss": 1.0311,
+ "step": 5368
+ },
+ {
+ "epoch": 14.629427792915532,
+ "grad_norm": 8.052887916564941,
+ "learning_rate": 1.9299024801426994e-05,
+ "loss": 1.0177,
+ "step": 5369
+ },
+ {
+ "epoch": 14.632152588555858,
+ "grad_norm": 5.795243740081787,
+ "learning_rate": 1.9298700178246685e-05,
+ "loss": 0.848,
+ "step": 5370
+ },
+ {
+ "epoch": 14.634877384196185,
+ "grad_norm": 7.630644798278809,
+ "learning_rate": 1.9298375482648396e-05,
+ "loss": 0.8417,
+ "step": 5371
+ },
+ {
+ "epoch": 14.637602179836513,
+ "grad_norm": 6.593286991119385,
+ "learning_rate": 1.9298050714634654e-05,
+ "loss": 0.9425,
+ "step": 5372
+ },
+ {
+ "epoch": 14.640326975476839,
+ "grad_norm": 8.323579788208008,
+ "learning_rate": 1.9297725874208e-05,
+ "loss": 0.9565,
+ "step": 5373
+ },
+ {
+ "epoch": 14.643051771117166,
+ "grad_norm": 6.37605619430542,
+ "learning_rate": 1.9297400961370953e-05,
+ "loss": 0.7975,
+ "step": 5374
+ },
+ {
+ "epoch": 14.645776566757494,
+ "grad_norm": 5.983356475830078,
+ "learning_rate": 1.9297075976126044e-05,
+ "loss": 0.8888,
+ "step": 5375
+ },
+ {
+ "epoch": 14.64850136239782,
+ "grad_norm": 8.301172256469727,
+ "learning_rate": 1.9296750918475807e-05,
+ "loss": 0.9546,
+ "step": 5376
+ },
+ {
+ "epoch": 14.651226158038147,
+ "grad_norm": 8.783949851989746,
+ "learning_rate": 1.9296425788422776e-05,
+ "loss": 0.8198,
+ "step": 5377
+ },
+ {
+ "epoch": 14.653950953678475,
+ "grad_norm": 7.575946807861328,
+ "learning_rate": 1.9296100585969478e-05,
+ "loss": 0.7024,
+ "step": 5378
+ },
+ {
+ "epoch": 14.6566757493188,
+ "grad_norm": 7.132654666900635,
+ "learning_rate": 1.9295775311118445e-05,
+ "loss": 1.0837,
+ "step": 5379
+ },
+ {
+ "epoch": 14.659400544959128,
+ "grad_norm": 7.104681968688965,
+ "learning_rate": 1.9295449963872217e-05,
+ "loss": 0.969,
+ "step": 5380
+ },
+ {
+ "epoch": 14.662125340599456,
+ "grad_norm": 6.632146835327148,
+ "learning_rate": 1.9295124544233322e-05,
+ "loss": 0.8992,
+ "step": 5381
+ },
+ {
+ "epoch": 14.664850136239782,
+ "grad_norm": 8.473021507263184,
+ "learning_rate": 1.92947990522043e-05,
+ "loss": 0.9257,
+ "step": 5382
+ },
+ {
+ "epoch": 14.66757493188011,
+ "grad_norm": 5.589237689971924,
+ "learning_rate": 1.929447348778768e-05,
+ "loss": 1.0437,
+ "step": 5383
+ },
+ {
+ "epoch": 14.670299727520437,
+ "grad_norm": 5.86954402923584,
+ "learning_rate": 1.9294147850985997e-05,
+ "loss": 0.7059,
+ "step": 5384
+ },
+ {
+ "epoch": 14.673024523160763,
+ "grad_norm": 8.508489608764648,
+ "learning_rate": 1.929382214180179e-05,
+ "loss": 0.7839,
+ "step": 5385
+ },
+ {
+ "epoch": 14.67574931880109,
+ "grad_norm": 6.583826065063477,
+ "learning_rate": 1.92934963602376e-05,
+ "loss": 0.8043,
+ "step": 5386
+ },
+ {
+ "epoch": 14.678474114441418,
+ "grad_norm": 7.76545524597168,
+ "learning_rate": 1.9293170506295955e-05,
+ "loss": 0.959,
+ "step": 5387
+ },
+ {
+ "epoch": 14.681198910081743,
+ "grad_norm": 7.465504169464111,
+ "learning_rate": 1.92928445799794e-05,
+ "loss": 0.8566,
+ "step": 5388
+ },
+ {
+ "epoch": 14.683923705722071,
+ "grad_norm": 6.743685245513916,
+ "learning_rate": 1.929251858129047e-05,
+ "loss": 0.954,
+ "step": 5389
+ },
+ {
+ "epoch": 14.686648501362399,
+ "grad_norm": 9.027506828308105,
+ "learning_rate": 1.92921925102317e-05,
+ "loss": 1.0117,
+ "step": 5390
+ },
+ {
+ "epoch": 14.689373297002724,
+ "grad_norm": 6.821288585662842,
+ "learning_rate": 1.929186636680564e-05,
+ "loss": 1.0209,
+ "step": 5391
+ },
+ {
+ "epoch": 14.692098092643052,
+ "grad_norm": 6.616784572601318,
+ "learning_rate": 1.929154015101482e-05,
+ "loss": 0.9808,
+ "step": 5392
+ },
+ {
+ "epoch": 14.69482288828338,
+ "grad_norm": 6.947346210479736,
+ "learning_rate": 1.9291213862861785e-05,
+ "loss": 0.8458,
+ "step": 5393
+ },
+ {
+ "epoch": 14.697547683923705,
+ "grad_norm": 8.364912986755371,
+ "learning_rate": 1.9290887502349078e-05,
+ "loss": 0.9807,
+ "step": 5394
+ },
+ {
+ "epoch": 14.700272479564033,
+ "grad_norm": 7.661632537841797,
+ "learning_rate": 1.9290561069479237e-05,
+ "loss": 0.8877,
+ "step": 5395
+ },
+ {
+ "epoch": 14.70299727520436,
+ "grad_norm": 8.79617977142334,
+ "learning_rate": 1.9290234564254803e-05,
+ "loss": 0.9801,
+ "step": 5396
+ },
+ {
+ "epoch": 14.705722070844686,
+ "grad_norm": 6.967199802398682,
+ "learning_rate": 1.928990798667832e-05,
+ "loss": 0.924,
+ "step": 5397
+ },
+ {
+ "epoch": 14.708446866485014,
+ "grad_norm": 8.028648376464844,
+ "learning_rate": 1.9289581336752338e-05,
+ "loss": 0.9121,
+ "step": 5398
+ },
+ {
+ "epoch": 14.711171662125341,
+ "grad_norm": 5.9009504318237305,
+ "learning_rate": 1.928925461447939e-05,
+ "loss": 0.8898,
+ "step": 5399
+ },
+ {
+ "epoch": 14.713896457765667,
+ "grad_norm": 7.983279705047607,
+ "learning_rate": 1.9288927819862033e-05,
+ "loss": 0.9337,
+ "step": 5400
+ },
+ {
+ "epoch": 14.716621253405995,
+ "grad_norm": 7.8363356590271,
+ "learning_rate": 1.92886009529028e-05,
+ "loss": 0.8182,
+ "step": 5401
+ },
+ {
+ "epoch": 14.719346049046322,
+ "grad_norm": 6.975879669189453,
+ "learning_rate": 1.9288274013604242e-05,
+ "loss": 0.9218,
+ "step": 5402
+ },
+ {
+ "epoch": 14.722070844686648,
+ "grad_norm": 6.726172924041748,
+ "learning_rate": 1.928794700196891e-05,
+ "loss": 1.0679,
+ "step": 5403
+ },
+ {
+ "epoch": 14.724795640326976,
+ "grad_norm": 6.088404178619385,
+ "learning_rate": 1.9287619917999336e-05,
+ "loss": 0.9387,
+ "step": 5404
+ },
+ {
+ "epoch": 14.727520435967303,
+ "grad_norm": 7.172768592834473,
+ "learning_rate": 1.9287292761698083e-05,
+ "loss": 0.8754,
+ "step": 5405
+ },
+ {
+ "epoch": 14.730245231607629,
+ "grad_norm": 7.780706882476807,
+ "learning_rate": 1.9286965533067694e-05,
+ "loss": 0.8988,
+ "step": 5406
+ },
+ {
+ "epoch": 14.732970027247957,
+ "grad_norm": 6.14064884185791,
+ "learning_rate": 1.9286638232110714e-05,
+ "loss": 1.0183,
+ "step": 5407
+ },
+ {
+ "epoch": 14.735694822888284,
+ "grad_norm": 7.362756252288818,
+ "learning_rate": 1.9286310858829696e-05,
+ "loss": 1.0092,
+ "step": 5408
+ },
+ {
+ "epoch": 14.73841961852861,
+ "grad_norm": 6.246393203735352,
+ "learning_rate": 1.9285983413227186e-05,
+ "loss": 0.6294,
+ "step": 5409
+ },
+ {
+ "epoch": 14.741144414168938,
+ "grad_norm": 7.6853437423706055,
+ "learning_rate": 1.928565589530574e-05,
+ "loss": 0.7958,
+ "step": 5410
+ },
+ {
+ "epoch": 14.743869209809265,
+ "grad_norm": 14.611734390258789,
+ "learning_rate": 1.9285328305067898e-05,
+ "loss": 1.0488,
+ "step": 5411
+ },
+ {
+ "epoch": 14.746594005449591,
+ "grad_norm": 7.215236663818359,
+ "learning_rate": 1.928500064251622e-05,
+ "loss": 1.0889,
+ "step": 5412
+ },
+ {
+ "epoch": 14.749318801089919,
+ "grad_norm": 6.529226779937744,
+ "learning_rate": 1.9284672907653258e-05,
+ "loss": 0.8896,
+ "step": 5413
+ },
+ {
+ "epoch": 14.752043596730246,
+ "grad_norm": 8.142340660095215,
+ "learning_rate": 1.928434510048156e-05,
+ "loss": 0.9536,
+ "step": 5414
+ },
+ {
+ "epoch": 14.754768392370572,
+ "grad_norm": 7.175665855407715,
+ "learning_rate": 1.9284017221003686e-05,
+ "loss": 0.938,
+ "step": 5415
+ },
+ {
+ "epoch": 14.7574931880109,
+ "grad_norm": 7.5522308349609375,
+ "learning_rate": 1.928368926922218e-05,
+ "loss": 0.9656,
+ "step": 5416
+ },
+ {
+ "epoch": 14.760217983651227,
+ "grad_norm": 8.586769104003906,
+ "learning_rate": 1.92833612451396e-05,
+ "loss": 0.8143,
+ "step": 5417
+ },
+ {
+ "epoch": 14.762942779291553,
+ "grad_norm": 7.689342498779297,
+ "learning_rate": 1.9283033148758506e-05,
+ "loss": 0.8588,
+ "step": 5418
+ },
+ {
+ "epoch": 14.76566757493188,
+ "grad_norm": 8.352788925170898,
+ "learning_rate": 1.9282704980081446e-05,
+ "loss": 0.9508,
+ "step": 5419
+ },
+ {
+ "epoch": 14.768392370572208,
+ "grad_norm": 7.808650970458984,
+ "learning_rate": 1.928237673911098e-05,
+ "loss": 0.8856,
+ "step": 5420
+ },
+ {
+ "epoch": 14.771117166212534,
+ "grad_norm": 7.95379114151001,
+ "learning_rate": 1.928204842584966e-05,
+ "loss": 0.8269,
+ "step": 5421
+ },
+ {
+ "epoch": 14.773841961852861,
+ "grad_norm": 7.725977897644043,
+ "learning_rate": 1.928172004030005e-05,
+ "loss": 0.9272,
+ "step": 5422
+ },
+ {
+ "epoch": 14.776566757493189,
+ "grad_norm": 8.180289268493652,
+ "learning_rate": 1.9281391582464697e-05,
+ "loss": 1.0446,
+ "step": 5423
+ },
+ {
+ "epoch": 14.779291553133515,
+ "grad_norm": 7.836215496063232,
+ "learning_rate": 1.9281063052346168e-05,
+ "loss": 0.7751,
+ "step": 5424
+ },
+ {
+ "epoch": 14.782016348773842,
+ "grad_norm": 6.195650100708008,
+ "learning_rate": 1.928073444994702e-05,
+ "loss": 0.8496,
+ "step": 5425
+ },
+ {
+ "epoch": 14.78474114441417,
+ "grad_norm": 5.684749603271484,
+ "learning_rate": 1.928040577526981e-05,
+ "loss": 0.8086,
+ "step": 5426
+ },
+ {
+ "epoch": 14.787465940054496,
+ "grad_norm": 6.26707649230957,
+ "learning_rate": 1.9280077028317096e-05,
+ "loss": 0.9204,
+ "step": 5427
+ },
+ {
+ "epoch": 14.790190735694823,
+ "grad_norm": 8.142394065856934,
+ "learning_rate": 1.927974820909144e-05,
+ "loss": 0.981,
+ "step": 5428
+ },
+ {
+ "epoch": 14.79291553133515,
+ "grad_norm": 6.452527046203613,
+ "learning_rate": 1.927941931759541e-05,
+ "loss": 0.9333,
+ "step": 5429
+ },
+ {
+ "epoch": 14.795640326975477,
+ "grad_norm": 6.213247299194336,
+ "learning_rate": 1.9279090353831557e-05,
+ "loss": 0.8317,
+ "step": 5430
+ },
+ {
+ "epoch": 14.798365122615804,
+ "grad_norm": 7.343374252319336,
+ "learning_rate": 1.9278761317802448e-05,
+ "loss": 0.985,
+ "step": 5431
+ },
+ {
+ "epoch": 14.80108991825613,
+ "grad_norm": 7.769726753234863,
+ "learning_rate": 1.9278432209510644e-05,
+ "loss": 0.9292,
+ "step": 5432
+ },
+ {
+ "epoch": 14.803814713896458,
+ "grad_norm": 7.146502494812012,
+ "learning_rate": 1.9278103028958707e-05,
+ "loss": 0.8694,
+ "step": 5433
+ },
+ {
+ "epoch": 14.806539509536785,
+ "grad_norm": 7.114870071411133,
+ "learning_rate": 1.9277773776149208e-05,
+ "loss": 0.8503,
+ "step": 5434
+ },
+ {
+ "epoch": 14.809264305177111,
+ "grad_norm": 7.9964094161987305,
+ "learning_rate": 1.92774444510847e-05,
+ "loss": 0.9578,
+ "step": 5435
+ },
+ {
+ "epoch": 14.811989100817438,
+ "grad_norm": 7.162196159362793,
+ "learning_rate": 1.927711505376775e-05,
+ "loss": 0.9363,
+ "step": 5436
+ },
+ {
+ "epoch": 14.814713896457766,
+ "grad_norm": 6.731115341186523,
+ "learning_rate": 1.9276785584200935e-05,
+ "loss": 0.8127,
+ "step": 5437
+ },
+ {
+ "epoch": 14.817438692098092,
+ "grad_norm": 8.819793701171875,
+ "learning_rate": 1.9276456042386806e-05,
+ "loss": 0.8938,
+ "step": 5438
+ },
+ {
+ "epoch": 14.82016348773842,
+ "grad_norm": 6.954529762268066,
+ "learning_rate": 1.927612642832794e-05,
+ "loss": 0.9351,
+ "step": 5439
+ },
+ {
+ "epoch": 14.822888283378747,
+ "grad_norm": 6.728962421417236,
+ "learning_rate": 1.9275796742026898e-05,
+ "loss": 0.8748,
+ "step": 5440
+ },
+ {
+ "epoch": 14.825613079019073,
+ "grad_norm": 6.933099269866943,
+ "learning_rate": 1.9275466983486247e-05,
+ "loss": 0.9622,
+ "step": 5441
+ },
+ {
+ "epoch": 14.8283378746594,
+ "grad_norm": 5.7251386642456055,
+ "learning_rate": 1.9275137152708564e-05,
+ "loss": 0.7306,
+ "step": 5442
+ },
+ {
+ "epoch": 14.831062670299728,
+ "grad_norm": 6.516031265258789,
+ "learning_rate": 1.9274807249696405e-05,
+ "loss": 0.8115,
+ "step": 5443
+ },
+ {
+ "epoch": 14.833787465940054,
+ "grad_norm": 7.265990734100342,
+ "learning_rate": 1.927447727445235e-05,
+ "loss": 0.8241,
+ "step": 5444
+ },
+ {
+ "epoch": 14.836512261580381,
+ "grad_norm": 6.916747093200684,
+ "learning_rate": 1.9274147226978962e-05,
+ "loss": 0.7212,
+ "step": 5445
+ },
+ {
+ "epoch": 14.839237057220709,
+ "grad_norm": 7.8474345207214355,
+ "learning_rate": 1.9273817107278815e-05,
+ "loss": 0.8588,
+ "step": 5446
+ },
+ {
+ "epoch": 14.841961852861035,
+ "grad_norm": 6.769237995147705,
+ "learning_rate": 1.927348691535448e-05,
+ "loss": 0.8295,
+ "step": 5447
+ },
+ {
+ "epoch": 14.844686648501362,
+ "grad_norm": 10.177266120910645,
+ "learning_rate": 1.9273156651208525e-05,
+ "loss": 1.0973,
+ "step": 5448
+ },
+ {
+ "epoch": 14.84741144414169,
+ "grad_norm": 7.6884565353393555,
+ "learning_rate": 1.9272826314843526e-05,
+ "loss": 0.8113,
+ "step": 5449
+ },
+ {
+ "epoch": 14.850136239782016,
+ "grad_norm": 6.548815727233887,
+ "learning_rate": 1.9272495906262053e-05,
+ "loss": 0.7988,
+ "step": 5450
+ },
+ {
+ "epoch": 14.852861035422343,
+ "grad_norm": 7.85021448135376,
+ "learning_rate": 1.927216542546668e-05,
+ "loss": 0.9486,
+ "step": 5451
+ },
+ {
+ "epoch": 14.85558583106267,
+ "grad_norm": 6.15355920791626,
+ "learning_rate": 1.9271834872459982e-05,
+ "loss": 0.8695,
+ "step": 5452
+ },
+ {
+ "epoch": 14.858310626702997,
+ "grad_norm": 7.169907093048096,
+ "learning_rate": 1.9271504247244534e-05,
+ "loss": 1.0061,
+ "step": 5453
+ },
+ {
+ "epoch": 14.861035422343324,
+ "grad_norm": 6.482378005981445,
+ "learning_rate": 1.927117354982291e-05,
+ "loss": 0.9452,
+ "step": 5454
+ },
+ {
+ "epoch": 14.863760217983652,
+ "grad_norm": 6.496779441833496,
+ "learning_rate": 1.927084278019768e-05,
+ "loss": 0.8666,
+ "step": 5455
+ },
+ {
+ "epoch": 14.866485013623977,
+ "grad_norm": 5.943950176239014,
+ "learning_rate": 1.9270511938371425e-05,
+ "loss": 0.9652,
+ "step": 5456
+ },
+ {
+ "epoch": 14.869209809264305,
+ "grad_norm": 5.247099876403809,
+ "learning_rate": 1.9270181024346727e-05,
+ "loss": 0.8453,
+ "step": 5457
+ },
+ {
+ "epoch": 14.871934604904633,
+ "grad_norm": 7.753555774688721,
+ "learning_rate": 1.9269850038126152e-05,
+ "loss": 0.7598,
+ "step": 5458
+ },
+ {
+ "epoch": 14.874659400544958,
+ "grad_norm": 6.149864196777344,
+ "learning_rate": 1.9269518979712285e-05,
+ "loss": 0.9895,
+ "step": 5459
+ },
+ {
+ "epoch": 14.877384196185286,
+ "grad_norm": 10.685378074645996,
+ "learning_rate": 1.9269187849107703e-05,
+ "loss": 0.9651,
+ "step": 5460
+ },
+ {
+ "epoch": 14.880108991825614,
+ "grad_norm": 6.582197666168213,
+ "learning_rate": 1.9268856646314983e-05,
+ "loss": 0.965,
+ "step": 5461
+ },
+ {
+ "epoch": 14.88283378746594,
+ "grad_norm": 6.377091407775879,
+ "learning_rate": 1.926852537133671e-05,
+ "loss": 1.0217,
+ "step": 5462
+ },
+ {
+ "epoch": 14.885558583106267,
+ "grad_norm": 7.425685405731201,
+ "learning_rate": 1.926819402417546e-05,
+ "loss": 0.9824,
+ "step": 5463
+ },
+ {
+ "epoch": 14.888283378746594,
+ "grad_norm": 7.376522541046143,
+ "learning_rate": 1.9267862604833808e-05,
+ "loss": 0.8955,
+ "step": 5464
+ },
+ {
+ "epoch": 14.89100817438692,
+ "grad_norm": 6.524284839630127,
+ "learning_rate": 1.9267531113314343e-05,
+ "loss": 1.0575,
+ "step": 5465
+ },
+ {
+ "epoch": 14.893732970027248,
+ "grad_norm": 8.5330171585083,
+ "learning_rate": 1.9267199549619643e-05,
+ "loss": 0.8433,
+ "step": 5466
+ },
+ {
+ "epoch": 14.896457765667575,
+ "grad_norm": 7.424759387969971,
+ "learning_rate": 1.926686791375229e-05,
+ "loss": 0.8295,
+ "step": 5467
+ },
+ {
+ "epoch": 14.899182561307901,
+ "grad_norm": 7.9398016929626465,
+ "learning_rate": 1.926653620571487e-05,
+ "loss": 1.0312,
+ "step": 5468
+ },
+ {
+ "epoch": 14.901907356948229,
+ "grad_norm": 6.4702982902526855,
+ "learning_rate": 1.9266204425509962e-05,
+ "loss": 0.964,
+ "step": 5469
+ },
+ {
+ "epoch": 14.904632152588556,
+ "grad_norm": 6.550892353057861,
+ "learning_rate": 1.9265872573140156e-05,
+ "loss": 0.7937,
+ "step": 5470
+ },
+ {
+ "epoch": 14.907356948228882,
+ "grad_norm": 8.299272537231445,
+ "learning_rate": 1.926554064860803e-05,
+ "loss": 0.9628,
+ "step": 5471
+ },
+ {
+ "epoch": 14.91008174386921,
+ "grad_norm": 6.110050678253174,
+ "learning_rate": 1.9265208651916172e-05,
+ "loss": 0.8537,
+ "step": 5472
+ },
+ {
+ "epoch": 14.912806539509537,
+ "grad_norm": 8.679011344909668,
+ "learning_rate": 1.9264876583067167e-05,
+ "loss": 1.0769,
+ "step": 5473
+ },
+ {
+ "epoch": 14.915531335149863,
+ "grad_norm": 6.639733791351318,
+ "learning_rate": 1.92645444420636e-05,
+ "loss": 0.913,
+ "step": 5474
+ },
+ {
+ "epoch": 14.91825613079019,
+ "grad_norm": 8.241230010986328,
+ "learning_rate": 1.9264212228908063e-05,
+ "loss": 0.8138,
+ "step": 5475
+ },
+ {
+ "epoch": 14.920980926430518,
+ "grad_norm": 7.203669548034668,
+ "learning_rate": 1.9263879943603136e-05,
+ "loss": 0.8297,
+ "step": 5476
+ },
+ {
+ "epoch": 14.923705722070844,
+ "grad_norm": 5.9377288818359375,
+ "learning_rate": 1.9263547586151412e-05,
+ "loss": 0.7825,
+ "step": 5477
+ },
+ {
+ "epoch": 14.926430517711172,
+ "grad_norm": 7.974017143249512,
+ "learning_rate": 1.9263215156555477e-05,
+ "loss": 0.9575,
+ "step": 5478
+ },
+ {
+ "epoch": 14.9291553133515,
+ "grad_norm": 10.160615921020508,
+ "learning_rate": 1.9262882654817918e-05,
+ "loss": 0.9752,
+ "step": 5479
+ },
+ {
+ "epoch": 14.931880108991825,
+ "grad_norm": 7.268916606903076,
+ "learning_rate": 1.926255008094133e-05,
+ "loss": 0.8473,
+ "step": 5480
+ },
+ {
+ "epoch": 14.934604904632153,
+ "grad_norm": 6.624278545379639,
+ "learning_rate": 1.92622174349283e-05,
+ "loss": 0.806,
+ "step": 5481
+ },
+ {
+ "epoch": 14.93732970027248,
+ "grad_norm": 7.365627765655518,
+ "learning_rate": 1.9261884716781417e-05,
+ "loss": 0.9154,
+ "step": 5482
+ },
+ {
+ "epoch": 14.940054495912806,
+ "grad_norm": 7.26140022277832,
+ "learning_rate": 1.9261551926503277e-05,
+ "loss": 0.8812,
+ "step": 5483
+ },
+ {
+ "epoch": 14.942779291553133,
+ "grad_norm": 9.271730422973633,
+ "learning_rate": 1.9261219064096467e-05,
+ "loss": 0.9291,
+ "step": 5484
+ },
+ {
+ "epoch": 14.945504087193461,
+ "grad_norm": 6.636730670928955,
+ "learning_rate": 1.926088612956358e-05,
+ "loss": 1.0101,
+ "step": 5485
+ },
+ {
+ "epoch": 14.948228882833787,
+ "grad_norm": 6.2664570808410645,
+ "learning_rate": 1.926055312290721e-05,
+ "loss": 1.0187,
+ "step": 5486
+ },
+ {
+ "epoch": 14.950953678474114,
+ "grad_norm": 6.636969566345215,
+ "learning_rate": 1.9260220044129953e-05,
+ "loss": 0.7567,
+ "step": 5487
+ },
+ {
+ "epoch": 14.953678474114442,
+ "grad_norm": 10.052999496459961,
+ "learning_rate": 1.9259886893234397e-05,
+ "loss": 0.9513,
+ "step": 5488
+ },
+ {
+ "epoch": 14.956403269754768,
+ "grad_norm": 6.946488857269287,
+ "learning_rate": 1.925955367022314e-05,
+ "loss": 0.9397,
+ "step": 5489
+ },
+ {
+ "epoch": 14.959128065395095,
+ "grad_norm": 6.447600364685059,
+ "learning_rate": 1.925922037509878e-05,
+ "loss": 0.9103,
+ "step": 5490
+ },
+ {
+ "epoch": 14.961852861035423,
+ "grad_norm": 7.716054916381836,
+ "learning_rate": 1.925888700786391e-05,
+ "loss": 0.8329,
+ "step": 5491
+ },
+ {
+ "epoch": 14.964577656675749,
+ "grad_norm": 7.812199592590332,
+ "learning_rate": 1.925855356852112e-05,
+ "loss": 0.9076,
+ "step": 5492
+ },
+ {
+ "epoch": 14.967302452316076,
+ "grad_norm": 7.085150241851807,
+ "learning_rate": 1.925822005707302e-05,
+ "loss": 0.9966,
+ "step": 5493
+ },
+ {
+ "epoch": 14.970027247956404,
+ "grad_norm": 13.153969764709473,
+ "learning_rate": 1.9257886473522194e-05,
+ "loss": 1.0276,
+ "step": 5494
+ },
+ {
+ "epoch": 14.97275204359673,
+ "grad_norm": 6.313168525695801,
+ "learning_rate": 1.925755281787125e-05,
+ "loss": 0.9982,
+ "step": 5495
+ },
+ {
+ "epoch": 14.975476839237057,
+ "grad_norm": 5.819637775421143,
+ "learning_rate": 1.9257219090122782e-05,
+ "loss": 0.8372,
+ "step": 5496
+ },
+ {
+ "epoch": 14.978201634877385,
+ "grad_norm": 7.591944217681885,
+ "learning_rate": 1.925688529027939e-05,
+ "loss": 1.1781,
+ "step": 5497
+ },
+ {
+ "epoch": 14.98092643051771,
+ "grad_norm": 6.398579120635986,
+ "learning_rate": 1.925655141834367e-05,
+ "loss": 1.0542,
+ "step": 5498
+ },
+ {
+ "epoch": 14.983651226158038,
+ "grad_norm": 6.718228340148926,
+ "learning_rate": 1.9256217474318227e-05,
+ "loss": 1.0505,
+ "step": 5499
+ },
+ {
+ "epoch": 14.986376021798366,
+ "grad_norm": 7.58510160446167,
+ "learning_rate": 1.9255883458205664e-05,
+ "loss": 0.72,
+ "step": 5500
+ },
+ {
+ "epoch": 14.989100817438691,
+ "grad_norm": 5.954460620880127,
+ "learning_rate": 1.9255549370008575e-05,
+ "loss": 0.835,
+ "step": 5501
+ },
+ {
+ "epoch": 14.991825613079019,
+ "grad_norm": 6.462665557861328,
+ "learning_rate": 1.925521520972957e-05,
+ "loss": 1.0078,
+ "step": 5502
+ },
+ {
+ "epoch": 14.994550408719347,
+ "grad_norm": 22.162961959838867,
+ "learning_rate": 1.925488097737124e-05,
+ "loss": 0.8738,
+ "step": 5503
+ },
+ {
+ "epoch": 14.997275204359672,
+ "grad_norm": 7.239997863769531,
+ "learning_rate": 1.92545466729362e-05,
+ "loss": 0.9418,
+ "step": 5504
+ },
+ {
+ "epoch": 15.0,
+ "grad_norm": 7.567640781402588,
+ "learning_rate": 1.9254212296427043e-05,
+ "loss": 0.9392,
+ "step": 5505
+ },
+ {
+ "epoch": 15.002724795640328,
+ "grad_norm": 5.573485851287842,
+ "learning_rate": 1.925387784784638e-05,
+ "loss": 0.6578,
+ "step": 5506
+ },
+ {
+ "epoch": 15.005449591280653,
+ "grad_norm": 5.552582740783691,
+ "learning_rate": 1.925354332719682e-05,
+ "loss": 0.619,
+ "step": 5507
+ },
+ {
+ "epoch": 15.008174386920981,
+ "grad_norm": 6.500991344451904,
+ "learning_rate": 1.9253208734480953e-05,
+ "loss": 0.7969,
+ "step": 5508
+ },
+ {
+ "epoch": 15.010899182561309,
+ "grad_norm": 6.627384662628174,
+ "learning_rate": 1.9252874069701405e-05,
+ "loss": 0.7496,
+ "step": 5509
+ },
+ {
+ "epoch": 15.013623978201634,
+ "grad_norm": 7.461287021636963,
+ "learning_rate": 1.9252539332860762e-05,
+ "loss": 1.09,
+ "step": 5510
+ },
+ {
+ "epoch": 15.016348773841962,
+ "grad_norm": 6.71066427230835,
+ "learning_rate": 1.9252204523961645e-05,
+ "loss": 0.7671,
+ "step": 5511
+ },
+ {
+ "epoch": 15.01907356948229,
+ "grad_norm": 6.83350133895874,
+ "learning_rate": 1.9251869643006657e-05,
+ "loss": 0.9976,
+ "step": 5512
+ },
+ {
+ "epoch": 15.021798365122615,
+ "grad_norm": 5.333657741546631,
+ "learning_rate": 1.9251534689998406e-05,
+ "loss": 0.9679,
+ "step": 5513
+ },
+ {
+ "epoch": 15.024523160762943,
+ "grad_norm": 5.950364112854004,
+ "learning_rate": 1.92511996649395e-05,
+ "loss": 0.7102,
+ "step": 5514
+ },
+ {
+ "epoch": 15.02724795640327,
+ "grad_norm": 7.221141815185547,
+ "learning_rate": 1.925086456783255e-05,
+ "loss": 0.8989,
+ "step": 5515
+ },
+ {
+ "epoch": 15.029972752043596,
+ "grad_norm": 7.413654804229736,
+ "learning_rate": 1.925052939868016e-05,
+ "loss": 0.778,
+ "step": 5516
+ },
+ {
+ "epoch": 15.032697547683924,
+ "grad_norm": 5.464250564575195,
+ "learning_rate": 1.925019415748495e-05,
+ "loss": 0.8337,
+ "step": 5517
+ },
+ {
+ "epoch": 15.035422343324251,
+ "grad_norm": 6.192759037017822,
+ "learning_rate": 1.924985884424952e-05,
+ "loss": 0.7405,
+ "step": 5518
+ },
+ {
+ "epoch": 15.038147138964577,
+ "grad_norm": 5.740165710449219,
+ "learning_rate": 1.9249523458976494e-05,
+ "loss": 0.7142,
+ "step": 5519
+ },
+ {
+ "epoch": 15.040871934604905,
+ "grad_norm": 7.2649383544921875,
+ "learning_rate": 1.9249188001668474e-05,
+ "loss": 0.7133,
+ "step": 5520
+ },
+ {
+ "epoch": 15.043596730245232,
+ "grad_norm": 7.132024765014648,
+ "learning_rate": 1.9248852472328075e-05,
+ "loss": 0.7566,
+ "step": 5521
+ },
+ {
+ "epoch": 15.046321525885558,
+ "grad_norm": 8.164922714233398,
+ "learning_rate": 1.924851687095791e-05,
+ "loss": 0.9955,
+ "step": 5522
+ },
+ {
+ "epoch": 15.049046321525886,
+ "grad_norm": 5.916840076446533,
+ "learning_rate": 1.9248181197560594e-05,
+ "loss": 0.7535,
+ "step": 5523
+ },
+ {
+ "epoch": 15.051771117166213,
+ "grad_norm": 7.332503795623779,
+ "learning_rate": 1.924784545213874e-05,
+ "loss": 0.7048,
+ "step": 5524
+ },
+ {
+ "epoch": 15.054495912806539,
+ "grad_norm": 7.632011890411377,
+ "learning_rate": 1.9247509634694963e-05,
+ "loss": 0.7832,
+ "step": 5525
+ },
+ {
+ "epoch": 15.057220708446867,
+ "grad_norm": 6.37957239151001,
+ "learning_rate": 1.924717374523188e-05,
+ "loss": 0.9103,
+ "step": 5526
+ },
+ {
+ "epoch": 15.059945504087194,
+ "grad_norm": 6.516311168670654,
+ "learning_rate": 1.9246837783752106e-05,
+ "loss": 0.8928,
+ "step": 5527
+ },
+ {
+ "epoch": 15.06267029972752,
+ "grad_norm": 6.260121822357178,
+ "learning_rate": 1.9246501750258256e-05,
+ "loss": 0.6807,
+ "step": 5528
+ },
+ {
+ "epoch": 15.065395095367847,
+ "grad_norm": 5.888677597045898,
+ "learning_rate": 1.9246165644752944e-05,
+ "loss": 0.733,
+ "step": 5529
+ },
+ {
+ "epoch": 15.068119891008175,
+ "grad_norm": 6.397183418273926,
+ "learning_rate": 1.92458294672388e-05,
+ "loss": 0.7579,
+ "step": 5530
+ },
+ {
+ "epoch": 15.0708446866485,
+ "grad_norm": 6.0291056632995605,
+ "learning_rate": 1.9245493217718426e-05,
+ "loss": 0.731,
+ "step": 5531
+ },
+ {
+ "epoch": 15.073569482288828,
+ "grad_norm": 5.952199935913086,
+ "learning_rate": 1.924515689619445e-05,
+ "loss": 0.7798,
+ "step": 5532
+ },
+ {
+ "epoch": 15.076294277929156,
+ "grad_norm": 6.658409118652344,
+ "learning_rate": 1.9244820502669494e-05,
+ "loss": 1.0248,
+ "step": 5533
+ },
+ {
+ "epoch": 15.079019073569482,
+ "grad_norm": 7.342145919799805,
+ "learning_rate": 1.924448403714617e-05,
+ "loss": 0.7891,
+ "step": 5534
+ },
+ {
+ "epoch": 15.08174386920981,
+ "grad_norm": 8.366787910461426,
+ "learning_rate": 1.9244147499627103e-05,
+ "loss": 0.7726,
+ "step": 5535
+ },
+ {
+ "epoch": 15.084468664850137,
+ "grad_norm": 7.446083068847656,
+ "learning_rate": 1.924381089011491e-05,
+ "loss": 0.8088,
+ "step": 5536
+ },
+ {
+ "epoch": 15.087193460490463,
+ "grad_norm": 6.046850681304932,
+ "learning_rate": 1.924347420861222e-05,
+ "loss": 0.7563,
+ "step": 5537
+ },
+ {
+ "epoch": 15.08991825613079,
+ "grad_norm": 7.401241302490234,
+ "learning_rate": 1.9243137455121648e-05,
+ "loss": 0.7976,
+ "step": 5538
+ },
+ {
+ "epoch": 15.092643051771118,
+ "grad_norm": 6.168306350708008,
+ "learning_rate": 1.9242800629645815e-05,
+ "loss": 0.7372,
+ "step": 5539
+ },
+ {
+ "epoch": 15.095367847411444,
+ "grad_norm": 6.912512302398682,
+ "learning_rate": 1.9242463732187355e-05,
+ "loss": 0.7833,
+ "step": 5540
+ },
+ {
+ "epoch": 15.098092643051771,
+ "grad_norm": 7.530439853668213,
+ "learning_rate": 1.924212676274888e-05,
+ "loss": 0.8145,
+ "step": 5541
+ },
+ {
+ "epoch": 15.100817438692099,
+ "grad_norm": 6.5108137130737305,
+ "learning_rate": 1.9241789721333023e-05,
+ "loss": 0.8685,
+ "step": 5542
+ },
+ {
+ "epoch": 15.103542234332425,
+ "grad_norm": 7.094371318817139,
+ "learning_rate": 1.9241452607942405e-05,
+ "loss": 0.8308,
+ "step": 5543
+ },
+ {
+ "epoch": 15.106267029972752,
+ "grad_norm": 7.522477626800537,
+ "learning_rate": 1.9241115422579647e-05,
+ "loss": 0.8992,
+ "step": 5544
+ },
+ {
+ "epoch": 15.10899182561308,
+ "grad_norm": 6.854132175445557,
+ "learning_rate": 1.9240778165247384e-05,
+ "loss": 0.8732,
+ "step": 5545
+ },
+ {
+ "epoch": 15.111716621253406,
+ "grad_norm": 5.568113803863525,
+ "learning_rate": 1.9240440835948236e-05,
+ "loss": 0.7975,
+ "step": 5546
+ },
+ {
+ "epoch": 15.114441416893733,
+ "grad_norm": 8.514400482177734,
+ "learning_rate": 1.924010343468483e-05,
+ "loss": 0.7588,
+ "step": 5547
+ },
+ {
+ "epoch": 15.11716621253406,
+ "grad_norm": 11.083649635314941,
+ "learning_rate": 1.92397659614598e-05,
+ "loss": 0.8364,
+ "step": 5548
+ },
+ {
+ "epoch": 15.119891008174386,
+ "grad_norm": 6.1065802574157715,
+ "learning_rate": 1.923942841627577e-05,
+ "loss": 0.8857,
+ "step": 5549
+ },
+ {
+ "epoch": 15.122615803814714,
+ "grad_norm": 5.683555603027344,
+ "learning_rate": 1.9239090799135365e-05,
+ "loss": 0.8895,
+ "step": 5550
+ },
+ {
+ "epoch": 15.125340599455042,
+ "grad_norm": 6.905640602111816,
+ "learning_rate": 1.923875311004122e-05,
+ "loss": 0.7614,
+ "step": 5551
+ },
+ {
+ "epoch": 15.128065395095367,
+ "grad_norm": 6.712569236755371,
+ "learning_rate": 1.9238415348995965e-05,
+ "loss": 0.8079,
+ "step": 5552
+ },
+ {
+ "epoch": 15.130790190735695,
+ "grad_norm": 7.048198699951172,
+ "learning_rate": 1.9238077516002228e-05,
+ "loss": 0.9469,
+ "step": 5553
+ },
+ {
+ "epoch": 15.133514986376023,
+ "grad_norm": 6.2088398933410645,
+ "learning_rate": 1.9237739611062637e-05,
+ "loss": 0.8414,
+ "step": 5554
+ },
+ {
+ "epoch": 15.136239782016348,
+ "grad_norm": 9.634664535522461,
+ "learning_rate": 1.923740163417983e-05,
+ "loss": 0.8689,
+ "step": 5555
+ },
+ {
+ "epoch": 15.138964577656676,
+ "grad_norm": 6.104285717010498,
+ "learning_rate": 1.9237063585356438e-05,
+ "loss": 0.7422,
+ "step": 5556
+ },
+ {
+ "epoch": 15.141689373297003,
+ "grad_norm": 6.216026782989502,
+ "learning_rate": 1.923672546459509e-05,
+ "loss": 0.8453,
+ "step": 5557
+ },
+ {
+ "epoch": 15.14441416893733,
+ "grad_norm": 6.2625532150268555,
+ "learning_rate": 1.9236387271898422e-05,
+ "loss": 0.728,
+ "step": 5558
+ },
+ {
+ "epoch": 15.147138964577657,
+ "grad_norm": 7.047122001647949,
+ "learning_rate": 1.9236049007269065e-05,
+ "loss": 0.7814,
+ "step": 5559
+ },
+ {
+ "epoch": 15.149863760217984,
+ "grad_norm": 6.299657344818115,
+ "learning_rate": 1.923571067070966e-05,
+ "loss": 0.8104,
+ "step": 5560
+ },
+ {
+ "epoch": 15.15258855585831,
+ "grad_norm": 8.870927810668945,
+ "learning_rate": 1.9235372262222836e-05,
+ "loss": 0.9333,
+ "step": 5561
+ },
+ {
+ "epoch": 15.155313351498638,
+ "grad_norm": 6.846146583557129,
+ "learning_rate": 1.923503378181123e-05,
+ "loss": 0.8345,
+ "step": 5562
+ },
+ {
+ "epoch": 15.158038147138965,
+ "grad_norm": 8.76327896118164,
+ "learning_rate": 1.9234695229477475e-05,
+ "loss": 0.7478,
+ "step": 5563
+ },
+ {
+ "epoch": 15.160762942779291,
+ "grad_norm": 6.585755825042725,
+ "learning_rate": 1.9234356605224214e-05,
+ "loss": 0.9306,
+ "step": 5564
+ },
+ {
+ "epoch": 15.163487738419619,
+ "grad_norm": 6.831243515014648,
+ "learning_rate": 1.923401790905408e-05,
+ "loss": 0.8928,
+ "step": 5565
+ },
+ {
+ "epoch": 15.166212534059946,
+ "grad_norm": 8.085003852844238,
+ "learning_rate": 1.9233679140969712e-05,
+ "loss": 0.9055,
+ "step": 5566
+ },
+ {
+ "epoch": 15.168937329700272,
+ "grad_norm": 7.749589443206787,
+ "learning_rate": 1.9233340300973748e-05,
+ "loss": 0.8322,
+ "step": 5567
+ },
+ {
+ "epoch": 15.1716621253406,
+ "grad_norm": 6.532967567443848,
+ "learning_rate": 1.923300138906883e-05,
+ "loss": 0.6995,
+ "step": 5568
+ },
+ {
+ "epoch": 15.174386920980927,
+ "grad_norm": 6.59461784362793,
+ "learning_rate": 1.9232662405257587e-05,
+ "loss": 0.8593,
+ "step": 5569
+ },
+ {
+ "epoch": 15.177111716621253,
+ "grad_norm": 5.474265098571777,
+ "learning_rate": 1.9232323349542672e-05,
+ "loss": 0.812,
+ "step": 5570
+ },
+ {
+ "epoch": 15.17983651226158,
+ "grad_norm": 8.02682113647461,
+ "learning_rate": 1.923198422192672e-05,
+ "loss": 0.6027,
+ "step": 5571
+ },
+ {
+ "epoch": 15.182561307901908,
+ "grad_norm": 6.994975566864014,
+ "learning_rate": 1.9231645022412367e-05,
+ "loss": 1.0195,
+ "step": 5572
+ },
+ {
+ "epoch": 15.185286103542234,
+ "grad_norm": 5.893030166625977,
+ "learning_rate": 1.9231305751002264e-05,
+ "loss": 0.9056,
+ "step": 5573
+ },
+ {
+ "epoch": 15.188010899182562,
+ "grad_norm": 6.167283535003662,
+ "learning_rate": 1.9230966407699048e-05,
+ "loss": 0.8457,
+ "step": 5574
+ },
+ {
+ "epoch": 15.190735694822889,
+ "grad_norm": 6.520694255828857,
+ "learning_rate": 1.9230626992505363e-05,
+ "loss": 0.8534,
+ "step": 5575
+ },
+ {
+ "epoch": 15.193460490463215,
+ "grad_norm": 6.628677845001221,
+ "learning_rate": 1.9230287505423854e-05,
+ "loss": 0.7101,
+ "step": 5576
+ },
+ {
+ "epoch": 15.196185286103542,
+ "grad_norm": 6.3892412185668945,
+ "learning_rate": 1.922994794645716e-05,
+ "loss": 0.8259,
+ "step": 5577
+ },
+ {
+ "epoch": 15.19891008174387,
+ "grad_norm": 6.8276472091674805,
+ "learning_rate": 1.9229608315607927e-05,
+ "loss": 0.7754,
+ "step": 5578
+ },
+ {
+ "epoch": 15.201634877384196,
+ "grad_norm": 5.964027404785156,
+ "learning_rate": 1.9229268612878805e-05,
+ "loss": 0.9369,
+ "step": 5579
+ },
+ {
+ "epoch": 15.204359673024523,
+ "grad_norm": 7.117208957672119,
+ "learning_rate": 1.9228928838272436e-05,
+ "loss": 0.7117,
+ "step": 5580
+ },
+ {
+ "epoch": 15.207084468664851,
+ "grad_norm": 7.495050430297852,
+ "learning_rate": 1.9228588991791462e-05,
+ "loss": 0.9524,
+ "step": 5581
+ },
+ {
+ "epoch": 15.209809264305177,
+ "grad_norm": 6.356069087982178,
+ "learning_rate": 1.922824907343854e-05,
+ "loss": 0.7513,
+ "step": 5582
+ },
+ {
+ "epoch": 15.212534059945504,
+ "grad_norm": 7.591334819793701,
+ "learning_rate": 1.9227909083216305e-05,
+ "loss": 0.8827,
+ "step": 5583
+ },
+ {
+ "epoch": 15.215258855585832,
+ "grad_norm": 8.609272956848145,
+ "learning_rate": 1.9227569021127418e-05,
+ "loss": 0.761,
+ "step": 5584
+ },
+ {
+ "epoch": 15.217983651226158,
+ "grad_norm": 8.00248908996582,
+ "learning_rate": 1.9227228887174517e-05,
+ "loss": 0.8268,
+ "step": 5585
+ },
+ {
+ "epoch": 15.220708446866485,
+ "grad_norm": 7.06692361831665,
+ "learning_rate": 1.9226888681360253e-05,
+ "loss": 0.9744,
+ "step": 5586
+ },
+ {
+ "epoch": 15.223433242506813,
+ "grad_norm": 6.692086696624756,
+ "learning_rate": 1.9226548403687277e-05,
+ "loss": 0.8196,
+ "step": 5587
+ },
+ {
+ "epoch": 15.226158038147139,
+ "grad_norm": 8.890912055969238,
+ "learning_rate": 1.922620805415824e-05,
+ "loss": 0.9852,
+ "step": 5588
+ },
+ {
+ "epoch": 15.228882833787466,
+ "grad_norm": 7.410240173339844,
+ "learning_rate": 1.9225867632775792e-05,
+ "loss": 0.8002,
+ "step": 5589
+ },
+ {
+ "epoch": 15.231607629427794,
+ "grad_norm": 6.333847522735596,
+ "learning_rate": 1.922552713954258e-05,
+ "loss": 0.9514,
+ "step": 5590
+ },
+ {
+ "epoch": 15.23433242506812,
+ "grad_norm": 5.733343601226807,
+ "learning_rate": 1.9225186574461267e-05,
+ "loss": 0.7769,
+ "step": 5591
+ },
+ {
+ "epoch": 15.237057220708447,
+ "grad_norm": 6.6877851486206055,
+ "learning_rate": 1.9224845937534493e-05,
+ "loss": 0.7332,
+ "step": 5592
+ },
+ {
+ "epoch": 15.239782016348773,
+ "grad_norm": 5.560786247253418,
+ "learning_rate": 1.9224505228764916e-05,
+ "loss": 0.9098,
+ "step": 5593
+ },
+ {
+ "epoch": 15.2425068119891,
+ "grad_norm": 6.476012229919434,
+ "learning_rate": 1.9224164448155188e-05,
+ "loss": 0.7627,
+ "step": 5594
+ },
+ {
+ "epoch": 15.245231607629428,
+ "grad_norm": 6.700287818908691,
+ "learning_rate": 1.9223823595707965e-05,
+ "loss": 0.7937,
+ "step": 5595
+ },
+ {
+ "epoch": 15.247956403269754,
+ "grad_norm": 7.33428430557251,
+ "learning_rate": 1.9223482671425904e-05,
+ "loss": 0.8184,
+ "step": 5596
+ },
+ {
+ "epoch": 15.250681198910081,
+ "grad_norm": 6.963656425476074,
+ "learning_rate": 1.9223141675311655e-05,
+ "loss": 0.7018,
+ "step": 5597
+ },
+ {
+ "epoch": 15.253405994550409,
+ "grad_norm": 6.33238410949707,
+ "learning_rate": 1.9222800607367877e-05,
+ "loss": 0.8357,
+ "step": 5598
+ },
+ {
+ "epoch": 15.256130790190735,
+ "grad_norm": 6.005176067352295,
+ "learning_rate": 1.9222459467597222e-05,
+ "loss": 0.8777,
+ "step": 5599
+ },
+ {
+ "epoch": 15.258855585831062,
+ "grad_norm": 7.983567714691162,
+ "learning_rate": 1.9222118256002348e-05,
+ "loss": 0.8131,
+ "step": 5600
+ },
+ {
+ "epoch": 15.26158038147139,
+ "grad_norm": 7.443377494812012,
+ "learning_rate": 1.9221776972585917e-05,
+ "loss": 0.8812,
+ "step": 5601
+ },
+ {
+ "epoch": 15.264305177111716,
+ "grad_norm": 6.164748668670654,
+ "learning_rate": 1.9221435617350583e-05,
+ "loss": 0.8237,
+ "step": 5602
+ },
+ {
+ "epoch": 15.267029972752043,
+ "grad_norm": 6.913761615753174,
+ "learning_rate": 1.9221094190299004e-05,
+ "loss": 0.8751,
+ "step": 5603
+ },
+ {
+ "epoch": 15.269754768392371,
+ "grad_norm": 6.493129253387451,
+ "learning_rate": 1.922075269143384e-05,
+ "loss": 0.6121,
+ "step": 5604
+ },
+ {
+ "epoch": 15.272479564032697,
+ "grad_norm": 6.518794059753418,
+ "learning_rate": 1.9220411120757753e-05,
+ "loss": 0.6835,
+ "step": 5605
+ },
+ {
+ "epoch": 15.275204359673024,
+ "grad_norm": 8.418124198913574,
+ "learning_rate": 1.92200694782734e-05,
+ "loss": 0.8562,
+ "step": 5606
+ },
+ {
+ "epoch": 15.277929155313352,
+ "grad_norm": 7.660461902618408,
+ "learning_rate": 1.9219727763983443e-05,
+ "loss": 0.7721,
+ "step": 5607
+ },
+ {
+ "epoch": 15.280653950953678,
+ "grad_norm": 5.481871128082275,
+ "learning_rate": 1.9219385977890542e-05,
+ "loss": 0.7341,
+ "step": 5608
+ },
+ {
+ "epoch": 15.283378746594005,
+ "grad_norm": 6.16661262512207,
+ "learning_rate": 1.9219044119997362e-05,
+ "loss": 1.0398,
+ "step": 5609
+ },
+ {
+ "epoch": 15.286103542234333,
+ "grad_norm": 6.785894870758057,
+ "learning_rate": 1.9218702190306563e-05,
+ "loss": 0.6848,
+ "step": 5610
+ },
+ {
+ "epoch": 15.288828337874659,
+ "grad_norm": 6.799986839294434,
+ "learning_rate": 1.9218360188820804e-05,
+ "loss": 0.8405,
+ "step": 5611
+ },
+ {
+ "epoch": 15.291553133514986,
+ "grad_norm": 5.747994899749756,
+ "learning_rate": 1.9218018115542755e-05,
+ "loss": 0.792,
+ "step": 5612
+ },
+ {
+ "epoch": 15.294277929155314,
+ "grad_norm": 6.383426189422607,
+ "learning_rate": 1.9217675970475076e-05,
+ "loss": 0.837,
+ "step": 5613
+ },
+ {
+ "epoch": 15.29700272479564,
+ "grad_norm": 5.863010406494141,
+ "learning_rate": 1.9217333753620436e-05,
+ "loss": 0.8414,
+ "step": 5614
+ },
+ {
+ "epoch": 15.299727520435967,
+ "grad_norm": 7.50922155380249,
+ "learning_rate": 1.9216991464981496e-05,
+ "loss": 0.8745,
+ "step": 5615
+ },
+ {
+ "epoch": 15.302452316076295,
+ "grad_norm": 6.41884708404541,
+ "learning_rate": 1.9216649104560924e-05,
+ "loss": 0.957,
+ "step": 5616
+ },
+ {
+ "epoch": 15.30517711171662,
+ "grad_norm": 7.427330493927002,
+ "learning_rate": 1.9216306672361385e-05,
+ "loss": 0.6696,
+ "step": 5617
+ },
+ {
+ "epoch": 15.307901907356948,
+ "grad_norm": 6.378982067108154,
+ "learning_rate": 1.9215964168385547e-05,
+ "loss": 0.859,
+ "step": 5618
+ },
+ {
+ "epoch": 15.310626702997276,
+ "grad_norm": 5.954253673553467,
+ "learning_rate": 1.9215621592636075e-05,
+ "loss": 0.8569,
+ "step": 5619
+ },
+ {
+ "epoch": 15.313351498637601,
+ "grad_norm": 5.169866561889648,
+ "learning_rate": 1.9215278945115638e-05,
+ "loss": 0.9056,
+ "step": 5620
+ },
+ {
+ "epoch": 15.316076294277929,
+ "grad_norm": 7.552786350250244,
+ "learning_rate": 1.9214936225826905e-05,
+ "loss": 0.8129,
+ "step": 5621
+ },
+ {
+ "epoch": 15.318801089918257,
+ "grad_norm": 5.989914894104004,
+ "learning_rate": 1.9214593434772547e-05,
+ "loss": 0.7206,
+ "step": 5622
+ },
+ {
+ "epoch": 15.321525885558582,
+ "grad_norm": 7.527641296386719,
+ "learning_rate": 1.9214250571955233e-05,
+ "loss": 0.9288,
+ "step": 5623
+ },
+ {
+ "epoch": 15.32425068119891,
+ "grad_norm": 6.529603481292725,
+ "learning_rate": 1.921390763737763e-05,
+ "loss": 0.6135,
+ "step": 5624
+ },
+ {
+ "epoch": 15.326975476839237,
+ "grad_norm": 6.202674388885498,
+ "learning_rate": 1.9213564631042412e-05,
+ "loss": 0.8602,
+ "step": 5625
+ },
+ {
+ "epoch": 15.329700272479563,
+ "grad_norm": 6.001922130584717,
+ "learning_rate": 1.9213221552952248e-05,
+ "loss": 0.8313,
+ "step": 5626
+ },
+ {
+ "epoch": 15.33242506811989,
+ "grad_norm": 7.060513496398926,
+ "learning_rate": 1.921287840310981e-05,
+ "loss": 0.7209,
+ "step": 5627
+ },
+ {
+ "epoch": 15.335149863760218,
+ "grad_norm": 6.601349353790283,
+ "learning_rate": 1.9212535181517773e-05,
+ "loss": 0.7668,
+ "step": 5628
+ },
+ {
+ "epoch": 15.337874659400544,
+ "grad_norm": 9.648773193359375,
+ "learning_rate": 1.9212191888178807e-05,
+ "loss": 0.8069,
+ "step": 5629
+ },
+ {
+ "epoch": 15.340599455040872,
+ "grad_norm": 8.913548469543457,
+ "learning_rate": 1.9211848523095594e-05,
+ "loss": 0.795,
+ "step": 5630
+ },
+ {
+ "epoch": 15.3433242506812,
+ "grad_norm": 5.8466901779174805,
+ "learning_rate": 1.9211505086270795e-05,
+ "loss": 0.7971,
+ "step": 5631
+ },
+ {
+ "epoch": 15.346049046321525,
+ "grad_norm": 7.43646764755249,
+ "learning_rate": 1.9211161577707092e-05,
+ "loss": 0.7909,
+ "step": 5632
+ },
+ {
+ "epoch": 15.348773841961853,
+ "grad_norm": 7.00393009185791,
+ "learning_rate": 1.9210817997407156e-05,
+ "loss": 0.8466,
+ "step": 5633
+ },
+ {
+ "epoch": 15.35149863760218,
+ "grad_norm": 7.019863605499268,
+ "learning_rate": 1.921047434537367e-05,
+ "loss": 0.8304,
+ "step": 5634
+ },
+ {
+ "epoch": 15.354223433242506,
+ "grad_norm": 6.0329270362854,
+ "learning_rate": 1.9210130621609307e-05,
+ "loss": 0.847,
+ "step": 5635
+ },
+ {
+ "epoch": 15.356948228882834,
+ "grad_norm": 7.036632537841797,
+ "learning_rate": 1.9209786826116738e-05,
+ "loss": 0.7914,
+ "step": 5636
+ },
+ {
+ "epoch": 15.359673024523161,
+ "grad_norm": 6.853795528411865,
+ "learning_rate": 1.920944295889865e-05,
+ "loss": 0.7611,
+ "step": 5637
+ },
+ {
+ "epoch": 15.362397820163487,
+ "grad_norm": 6.251479625701904,
+ "learning_rate": 1.9209099019957715e-05,
+ "loss": 0.9102,
+ "step": 5638
+ },
+ {
+ "epoch": 15.365122615803815,
+ "grad_norm": 7.765817165374756,
+ "learning_rate": 1.9208755009296615e-05,
+ "loss": 0.7671,
+ "step": 5639
+ },
+ {
+ "epoch": 15.367847411444142,
+ "grad_norm": 6.9009270668029785,
+ "learning_rate": 1.9208410926918025e-05,
+ "loss": 0.8103,
+ "step": 5640
+ },
+ {
+ "epoch": 15.370572207084468,
+ "grad_norm": 6.744650363922119,
+ "learning_rate": 1.9208066772824628e-05,
+ "loss": 0.6534,
+ "step": 5641
+ },
+ {
+ "epoch": 15.373297002724795,
+ "grad_norm": 6.634628772735596,
+ "learning_rate": 1.9207722547019104e-05,
+ "loss": 0.8688,
+ "step": 5642
+ },
+ {
+ "epoch": 15.376021798365123,
+ "grad_norm": 7.7573394775390625,
+ "learning_rate": 1.9207378249504134e-05,
+ "loss": 0.7495,
+ "step": 5643
+ },
+ {
+ "epoch": 15.378746594005449,
+ "grad_norm": 8.130145072937012,
+ "learning_rate": 1.9207033880282397e-05,
+ "loss": 0.7903,
+ "step": 5644
+ },
+ {
+ "epoch": 15.381471389645776,
+ "grad_norm": 7.984460353851318,
+ "learning_rate": 1.9206689439356576e-05,
+ "loss": 0.8011,
+ "step": 5645
+ },
+ {
+ "epoch": 15.384196185286104,
+ "grad_norm": 7.084766387939453,
+ "learning_rate": 1.9206344926729355e-05,
+ "loss": 0.8411,
+ "step": 5646
+ },
+ {
+ "epoch": 15.38692098092643,
+ "grad_norm": 8.859514236450195,
+ "learning_rate": 1.9206000342403415e-05,
+ "loss": 0.8171,
+ "step": 5647
+ },
+ {
+ "epoch": 15.389645776566757,
+ "grad_norm": 6.8072829246521,
+ "learning_rate": 1.920565568638144e-05,
+ "loss": 0.9768,
+ "step": 5648
+ },
+ {
+ "epoch": 15.392370572207085,
+ "grad_norm": 6.604030609130859,
+ "learning_rate": 1.9205310958666118e-05,
+ "loss": 0.8571,
+ "step": 5649
+ },
+ {
+ "epoch": 15.39509536784741,
+ "grad_norm": 10.732048034667969,
+ "learning_rate": 1.9204966159260128e-05,
+ "loss": 0.8324,
+ "step": 5650
+ },
+ {
+ "epoch": 15.397820163487738,
+ "grad_norm": 5.953035831451416,
+ "learning_rate": 1.9204621288166158e-05,
+ "loss": 0.707,
+ "step": 5651
+ },
+ {
+ "epoch": 15.400544959128066,
+ "grad_norm": 5.503138542175293,
+ "learning_rate": 1.9204276345386893e-05,
+ "loss": 0.8394,
+ "step": 5652
+ },
+ {
+ "epoch": 15.403269754768392,
+ "grad_norm": 6.248448371887207,
+ "learning_rate": 1.9203931330925024e-05,
+ "loss": 0.8706,
+ "step": 5653
+ },
+ {
+ "epoch": 15.40599455040872,
+ "grad_norm": 6.415957450866699,
+ "learning_rate": 1.920358624478323e-05,
+ "loss": 0.9706,
+ "step": 5654
+ },
+ {
+ "epoch": 15.408719346049047,
+ "grad_norm": 5.665203094482422,
+ "learning_rate": 1.9203241086964203e-05,
+ "loss": 0.7603,
+ "step": 5655
+ },
+ {
+ "epoch": 15.411444141689373,
+ "grad_norm": 7.178077220916748,
+ "learning_rate": 1.920289585747063e-05,
+ "loss": 0.9242,
+ "step": 5656
+ },
+ {
+ "epoch": 15.4141689373297,
+ "grad_norm": 8.517338752746582,
+ "learning_rate": 1.9202550556305203e-05,
+ "loss": 0.7488,
+ "step": 5657
+ },
+ {
+ "epoch": 15.416893732970028,
+ "grad_norm": 7.164817810058594,
+ "learning_rate": 1.9202205183470607e-05,
+ "loss": 0.8044,
+ "step": 5658
+ },
+ {
+ "epoch": 15.419618528610354,
+ "grad_norm": 8.74505615234375,
+ "learning_rate": 1.9201859738969532e-05,
+ "loss": 0.8199,
+ "step": 5659
+ },
+ {
+ "epoch": 15.422343324250681,
+ "grad_norm": 6.898816108703613,
+ "learning_rate": 1.9201514222804672e-05,
+ "loss": 0.8788,
+ "step": 5660
+ },
+ {
+ "epoch": 15.425068119891009,
+ "grad_norm": 8.024271011352539,
+ "learning_rate": 1.9201168634978714e-05,
+ "loss": 0.985,
+ "step": 5661
+ },
+ {
+ "epoch": 15.427792915531334,
+ "grad_norm": 6.4318389892578125,
+ "learning_rate": 1.920082297549435e-05,
+ "loss": 0.8899,
+ "step": 5662
+ },
+ {
+ "epoch": 15.430517711171662,
+ "grad_norm": 7.104996204376221,
+ "learning_rate": 1.9200477244354274e-05,
+ "loss": 1.0854,
+ "step": 5663
+ },
+ {
+ "epoch": 15.43324250681199,
+ "grad_norm": 6.242406368255615,
+ "learning_rate": 1.920013144156118e-05,
+ "loss": 0.8374,
+ "step": 5664
+ },
+ {
+ "epoch": 15.435967302452315,
+ "grad_norm": 7.952263355255127,
+ "learning_rate": 1.9199785567117755e-05,
+ "loss": 0.9474,
+ "step": 5665
+ },
+ {
+ "epoch": 15.438692098092643,
+ "grad_norm": 7.347584247589111,
+ "learning_rate": 1.9199439621026696e-05,
+ "loss": 0.8922,
+ "step": 5666
+ },
+ {
+ "epoch": 15.44141689373297,
+ "grad_norm": 9.530574798583984,
+ "learning_rate": 1.9199093603290697e-05,
+ "loss": 0.9352,
+ "step": 5667
+ },
+ {
+ "epoch": 15.444141689373296,
+ "grad_norm": 6.900759220123291,
+ "learning_rate": 1.9198747513912455e-05,
+ "loss": 0.7925,
+ "step": 5668
+ },
+ {
+ "epoch": 15.446866485013624,
+ "grad_norm": 8.794690132141113,
+ "learning_rate": 1.9198401352894663e-05,
+ "loss": 0.6682,
+ "step": 5669
+ },
+ {
+ "epoch": 15.449591280653951,
+ "grad_norm": 6.300912857055664,
+ "learning_rate": 1.9198055120240012e-05,
+ "loss": 0.8361,
+ "step": 5670
+ },
+ {
+ "epoch": 15.452316076294277,
+ "grad_norm": 6.173508644104004,
+ "learning_rate": 1.919770881595121e-05,
+ "loss": 0.7848,
+ "step": 5671
+ },
+ {
+ "epoch": 15.455040871934605,
+ "grad_norm": 6.643235206604004,
+ "learning_rate": 1.9197362440030946e-05,
+ "loss": 0.8022,
+ "step": 5672
+ },
+ {
+ "epoch": 15.457765667574932,
+ "grad_norm": 7.634005069732666,
+ "learning_rate": 1.9197015992481918e-05,
+ "loss": 0.8588,
+ "step": 5673
+ },
+ {
+ "epoch": 15.460490463215258,
+ "grad_norm": 6.389156818389893,
+ "learning_rate": 1.919666947330683e-05,
+ "loss": 0.7095,
+ "step": 5674
+ },
+ {
+ "epoch": 15.463215258855586,
+ "grad_norm": 7.521469593048096,
+ "learning_rate": 1.9196322882508372e-05,
+ "loss": 0.7125,
+ "step": 5675
+ },
+ {
+ "epoch": 15.465940054495913,
+ "grad_norm": 6.636131763458252,
+ "learning_rate": 1.919597622008925e-05,
+ "loss": 0.9688,
+ "step": 5676
+ },
+ {
+ "epoch": 15.46866485013624,
+ "grad_norm": 7.681743144989014,
+ "learning_rate": 1.919562948605216e-05,
+ "loss": 0.8436,
+ "step": 5677
+ },
+ {
+ "epoch": 15.471389645776567,
+ "grad_norm": 7.934110641479492,
+ "learning_rate": 1.9195282680399804e-05,
+ "loss": 0.8882,
+ "step": 5678
+ },
+ {
+ "epoch": 15.474114441416894,
+ "grad_norm": 6.334452152252197,
+ "learning_rate": 1.919493580313488e-05,
+ "loss": 0.8168,
+ "step": 5679
+ },
+ {
+ "epoch": 15.47683923705722,
+ "grad_norm": 7.946335315704346,
+ "learning_rate": 1.919458885426009e-05,
+ "loss": 0.8015,
+ "step": 5680
+ },
+ {
+ "epoch": 15.479564032697548,
+ "grad_norm": 8.06879997253418,
+ "learning_rate": 1.9194241833778148e-05,
+ "loss": 1.0266,
+ "step": 5681
+ },
+ {
+ "epoch": 15.482288828337875,
+ "grad_norm": 7.608769416809082,
+ "learning_rate": 1.9193894741691735e-05,
+ "loss": 0.794,
+ "step": 5682
+ },
+ {
+ "epoch": 15.485013623978201,
+ "grad_norm": 6.1882853507995605,
+ "learning_rate": 1.9193547578003574e-05,
+ "loss": 0.8613,
+ "step": 5683
+ },
+ {
+ "epoch": 15.487738419618529,
+ "grad_norm": 6.742659568786621,
+ "learning_rate": 1.9193200342716355e-05,
+ "loss": 0.7448,
+ "step": 5684
+ },
+ {
+ "epoch": 15.490463215258856,
+ "grad_norm": 6.453981876373291,
+ "learning_rate": 1.919285303583279e-05,
+ "loss": 0.9304,
+ "step": 5685
+ },
+ {
+ "epoch": 15.493188010899182,
+ "grad_norm": 6.880506992340088,
+ "learning_rate": 1.919250565735558e-05,
+ "loss": 0.7859,
+ "step": 5686
+ },
+ {
+ "epoch": 15.49591280653951,
+ "grad_norm": 10.189485549926758,
+ "learning_rate": 1.9192158207287435e-05,
+ "loss": 0.7957,
+ "step": 5687
+ },
+ {
+ "epoch": 15.498637602179837,
+ "grad_norm": 7.095956802368164,
+ "learning_rate": 1.9191810685631055e-05,
+ "loss": 0.9421,
+ "step": 5688
+ },
+ {
+ "epoch": 15.501362397820163,
+ "grad_norm": 6.398034572601318,
+ "learning_rate": 1.919146309238915e-05,
+ "loss": 0.8729,
+ "step": 5689
+ },
+ {
+ "epoch": 15.50408719346049,
+ "grad_norm": 6.859742164611816,
+ "learning_rate": 1.9191115427564425e-05,
+ "loss": 0.7594,
+ "step": 5690
+ },
+ {
+ "epoch": 15.506811989100818,
+ "grad_norm": 7.503488540649414,
+ "learning_rate": 1.9190767691159592e-05,
+ "loss": 0.8644,
+ "step": 5691
+ },
+ {
+ "epoch": 15.509536784741144,
+ "grad_norm": 6.094329357147217,
+ "learning_rate": 1.9190419883177355e-05,
+ "loss": 0.754,
+ "step": 5692
+ },
+ {
+ "epoch": 15.512261580381471,
+ "grad_norm": 8.201502799987793,
+ "learning_rate": 1.919007200362042e-05,
+ "loss": 0.8658,
+ "step": 5693
+ },
+ {
+ "epoch": 15.514986376021799,
+ "grad_norm": 6.7607526779174805,
+ "learning_rate": 1.9189724052491502e-05,
+ "loss": 0.9075,
+ "step": 5694
+ },
+ {
+ "epoch": 15.517711171662125,
+ "grad_norm": 6.579628944396973,
+ "learning_rate": 1.918937602979331e-05,
+ "loss": 0.8699,
+ "step": 5695
+ },
+ {
+ "epoch": 15.520435967302452,
+ "grad_norm": 6.731390476226807,
+ "learning_rate": 1.9189027935528553e-05,
+ "loss": 0.7721,
+ "step": 5696
+ },
+ {
+ "epoch": 15.52316076294278,
+ "grad_norm": 8.79029655456543,
+ "learning_rate": 1.9188679769699944e-05,
+ "loss": 0.7908,
+ "step": 5697
+ },
+ {
+ "epoch": 15.525885558583106,
+ "grad_norm": 6.0726213455200195,
+ "learning_rate": 1.918833153231019e-05,
+ "loss": 0.7179,
+ "step": 5698
+ },
+ {
+ "epoch": 15.528610354223433,
+ "grad_norm": 6.778722763061523,
+ "learning_rate": 1.9187983223362006e-05,
+ "loss": 0.8701,
+ "step": 5699
+ },
+ {
+ "epoch": 15.53133514986376,
+ "grad_norm": 7.016279220581055,
+ "learning_rate": 1.9187634842858104e-05,
+ "loss": 0.7414,
+ "step": 5700
+ },
+ {
+ "epoch": 15.534059945504087,
+ "grad_norm": 7.78290319442749,
+ "learning_rate": 1.91872863908012e-05,
+ "loss": 0.7498,
+ "step": 5701
+ },
+ {
+ "epoch": 15.536784741144414,
+ "grad_norm": 6.577802658081055,
+ "learning_rate": 1.9186937867194e-05,
+ "loss": 0.8737,
+ "step": 5702
+ },
+ {
+ "epoch": 15.539509536784742,
+ "grad_norm": 6.689164161682129,
+ "learning_rate": 1.9186589272039227e-05,
+ "loss": 0.9777,
+ "step": 5703
+ },
+ {
+ "epoch": 15.542234332425068,
+ "grad_norm": 8.187846183776855,
+ "learning_rate": 1.9186240605339596e-05,
+ "loss": 0.8928,
+ "step": 5704
+ },
+ {
+ "epoch": 15.544959128065395,
+ "grad_norm": 7.3906636238098145,
+ "learning_rate": 1.9185891867097813e-05,
+ "loss": 0.7247,
+ "step": 5705
+ },
+ {
+ "epoch": 15.547683923705723,
+ "grad_norm": 7.529016971588135,
+ "learning_rate": 1.9185543057316604e-05,
+ "loss": 0.7626,
+ "step": 5706
+ },
+ {
+ "epoch": 15.550408719346049,
+ "grad_norm": 7.673582077026367,
+ "learning_rate": 1.9185194175998682e-05,
+ "loss": 0.8665,
+ "step": 5707
+ },
+ {
+ "epoch": 15.553133514986376,
+ "grad_norm": 5.945589065551758,
+ "learning_rate": 1.9184845223146757e-05,
+ "loss": 0.9484,
+ "step": 5708
+ },
+ {
+ "epoch": 15.555858310626704,
+ "grad_norm": 6.922646522521973,
+ "learning_rate": 1.9184496198763557e-05,
+ "loss": 0.9078,
+ "step": 5709
+ },
+ {
+ "epoch": 15.55858310626703,
+ "grad_norm": 6.4051642417907715,
+ "learning_rate": 1.9184147102851796e-05,
+ "loss": 0.8036,
+ "step": 5710
+ },
+ {
+ "epoch": 15.561307901907357,
+ "grad_norm": 4.865403175354004,
+ "learning_rate": 1.9183797935414194e-05,
+ "loss": 0.6754,
+ "step": 5711
+ },
+ {
+ "epoch": 15.564032697547685,
+ "grad_norm": 5.191896438598633,
+ "learning_rate": 1.918344869645347e-05,
+ "loss": 0.6707,
+ "step": 5712
+ },
+ {
+ "epoch": 15.56675749318801,
+ "grad_norm": 7.448020935058594,
+ "learning_rate": 1.918309938597234e-05,
+ "loss": 0.9403,
+ "step": 5713
+ },
+ {
+ "epoch": 15.569482288828338,
+ "grad_norm": 9.264307975769043,
+ "learning_rate": 1.918275000397353e-05,
+ "loss": 0.9089,
+ "step": 5714
+ },
+ {
+ "epoch": 15.572207084468666,
+ "grad_norm": 6.0142502784729,
+ "learning_rate": 1.9182400550459758e-05,
+ "loss": 0.8688,
+ "step": 5715
+ },
+ {
+ "epoch": 15.574931880108991,
+ "grad_norm": 6.379959583282471,
+ "learning_rate": 1.9182051025433743e-05,
+ "loss": 0.8478,
+ "step": 5716
+ },
+ {
+ "epoch": 15.577656675749319,
+ "grad_norm": 5.886523246765137,
+ "learning_rate": 1.9181701428898215e-05,
+ "loss": 0.9939,
+ "step": 5717
+ },
+ {
+ "epoch": 15.580381471389646,
+ "grad_norm": 9.920634269714355,
+ "learning_rate": 1.9181351760855888e-05,
+ "loss": 1.0432,
+ "step": 5718
+ },
+ {
+ "epoch": 15.583106267029972,
+ "grad_norm": 5.495692729949951,
+ "learning_rate": 1.918100202130949e-05,
+ "loss": 0.84,
+ "step": 5719
+ },
+ {
+ "epoch": 15.5858310626703,
+ "grad_norm": 5.769711017608643,
+ "learning_rate": 1.9180652210261743e-05,
+ "loss": 0.7039,
+ "step": 5720
+ },
+ {
+ "epoch": 15.588555858310627,
+ "grad_norm": 7.993760585784912,
+ "learning_rate": 1.9180302327715376e-05,
+ "loss": 1.0555,
+ "step": 5721
+ },
+ {
+ "epoch": 15.591280653950953,
+ "grad_norm": 6.700951099395752,
+ "learning_rate": 1.9179952373673106e-05,
+ "loss": 0.9059,
+ "step": 5722
+ },
+ {
+ "epoch": 15.59400544959128,
+ "grad_norm": 6.907508373260498,
+ "learning_rate": 1.9179602348137664e-05,
+ "loss": 0.7939,
+ "step": 5723
+ },
+ {
+ "epoch": 15.596730245231608,
+ "grad_norm": 5.9784417152404785,
+ "learning_rate": 1.9179252251111774e-05,
+ "loss": 0.8207,
+ "step": 5724
+ },
+ {
+ "epoch": 15.599455040871934,
+ "grad_norm": 6.103968620300293,
+ "learning_rate": 1.917890208259816e-05,
+ "loss": 0.96,
+ "step": 5725
+ },
+ {
+ "epoch": 15.602179836512262,
+ "grad_norm": 6.2187066078186035,
+ "learning_rate": 1.917855184259956e-05,
+ "loss": 0.8641,
+ "step": 5726
+ },
+ {
+ "epoch": 15.60490463215259,
+ "grad_norm": 7.2202277183532715,
+ "learning_rate": 1.9178201531118686e-05,
+ "loss": 0.7076,
+ "step": 5727
+ },
+ {
+ "epoch": 15.607629427792915,
+ "grad_norm": 6.238146781921387,
+ "learning_rate": 1.9177851148158274e-05,
+ "loss": 0.7742,
+ "step": 5728
+ },
+ {
+ "epoch": 15.610354223433243,
+ "grad_norm": 6.914647102355957,
+ "learning_rate": 1.917750069372106e-05,
+ "loss": 0.8019,
+ "step": 5729
+ },
+ {
+ "epoch": 15.61307901907357,
+ "grad_norm": 7.91395378112793,
+ "learning_rate": 1.9177150167809756e-05,
+ "loss": 1.0576,
+ "step": 5730
+ },
+ {
+ "epoch": 15.615803814713896,
+ "grad_norm": 6.992804050445557,
+ "learning_rate": 1.9176799570427108e-05,
+ "loss": 0.8164,
+ "step": 5731
+ },
+ {
+ "epoch": 15.618528610354224,
+ "grad_norm": 7.9754252433776855,
+ "learning_rate": 1.9176448901575838e-05,
+ "loss": 0.7488,
+ "step": 5732
+ },
+ {
+ "epoch": 15.621253405994551,
+ "grad_norm": 11.75976848602295,
+ "learning_rate": 1.9176098161258684e-05,
+ "loss": 0.6258,
+ "step": 5733
+ },
+ {
+ "epoch": 15.623978201634877,
+ "grad_norm": 6.831855773925781,
+ "learning_rate": 1.9175747349478366e-05,
+ "loss": 0.7266,
+ "step": 5734
+ },
+ {
+ "epoch": 15.626702997275205,
+ "grad_norm": 6.168244361877441,
+ "learning_rate": 1.917539646623763e-05,
+ "loss": 0.7218,
+ "step": 5735
+ },
+ {
+ "epoch": 15.629427792915532,
+ "grad_norm": 6.760199069976807,
+ "learning_rate": 1.9175045511539197e-05,
+ "loss": 0.6874,
+ "step": 5736
+ },
+ {
+ "epoch": 15.632152588555858,
+ "grad_norm": 6.421217441558838,
+ "learning_rate": 1.9174694485385805e-05,
+ "loss": 0.9561,
+ "step": 5737
+ },
+ {
+ "epoch": 15.634877384196185,
+ "grad_norm": 6.8177947998046875,
+ "learning_rate": 1.917434338778019e-05,
+ "loss": 0.85,
+ "step": 5738
+ },
+ {
+ "epoch": 15.637602179836513,
+ "grad_norm": 19.962932586669922,
+ "learning_rate": 1.917399221872508e-05,
+ "loss": 0.6639,
+ "step": 5739
+ },
+ {
+ "epoch": 15.640326975476839,
+ "grad_norm": 7.379504203796387,
+ "learning_rate": 1.9173640978223214e-05,
+ "loss": 0.8411,
+ "step": 5740
+ },
+ {
+ "epoch": 15.643051771117166,
+ "grad_norm": 6.093384265899658,
+ "learning_rate": 1.917328966627733e-05,
+ "loss": 0.8464,
+ "step": 5741
+ },
+ {
+ "epoch": 15.645776566757494,
+ "grad_norm": 8.604357719421387,
+ "learning_rate": 1.9172938282890164e-05,
+ "loss": 0.8209,
+ "step": 5742
+ },
+ {
+ "epoch": 15.64850136239782,
+ "grad_norm": 7.351316928863525,
+ "learning_rate": 1.9172586828064446e-05,
+ "loss": 0.8317,
+ "step": 5743
+ },
+ {
+ "epoch": 15.651226158038147,
+ "grad_norm": 6.172524929046631,
+ "learning_rate": 1.9172235301802916e-05,
+ "loss": 0.8116,
+ "step": 5744
+ },
+ {
+ "epoch": 15.653950953678475,
+ "grad_norm": 6.425074577331543,
+ "learning_rate": 1.9171883704108316e-05,
+ "loss": 0.9739,
+ "step": 5745
+ },
+ {
+ "epoch": 15.6566757493188,
+ "grad_norm": 5.519540786743164,
+ "learning_rate": 1.9171532034983377e-05,
+ "loss": 0.7811,
+ "step": 5746
+ },
+ {
+ "epoch": 15.659400544959128,
+ "grad_norm": 5.83468770980835,
+ "learning_rate": 1.9171180294430847e-05,
+ "loss": 0.8641,
+ "step": 5747
+ },
+ {
+ "epoch": 15.662125340599456,
+ "grad_norm": 7.776528358459473,
+ "learning_rate": 1.9170828482453456e-05,
+ "loss": 0.7771,
+ "step": 5748
+ },
+ {
+ "epoch": 15.664850136239782,
+ "grad_norm": 6.93515682220459,
+ "learning_rate": 1.9170476599053952e-05,
+ "loss": 1.0522,
+ "step": 5749
+ },
+ {
+ "epoch": 15.66757493188011,
+ "grad_norm": 6.827121734619141,
+ "learning_rate": 1.9170124644235066e-05,
+ "loss": 1.0356,
+ "step": 5750
+ },
+ {
+ "epoch": 15.670299727520437,
+ "grad_norm": 7.102541446685791,
+ "learning_rate": 1.9169772617999545e-05,
+ "loss": 0.9261,
+ "step": 5751
+ },
+ {
+ "epoch": 15.673024523160763,
+ "grad_norm": 7.596420764923096,
+ "learning_rate": 1.9169420520350135e-05,
+ "loss": 0.8351,
+ "step": 5752
+ },
+ {
+ "epoch": 15.67574931880109,
+ "grad_norm": 6.9390482902526855,
+ "learning_rate": 1.916906835128957e-05,
+ "loss": 0.7264,
+ "step": 5753
+ },
+ {
+ "epoch": 15.678474114441418,
+ "grad_norm": 6.5754008293151855,
+ "learning_rate": 1.9168716110820592e-05,
+ "loss": 0.8666,
+ "step": 5754
+ },
+ {
+ "epoch": 15.681198910081743,
+ "grad_norm": 6.613892555236816,
+ "learning_rate": 1.9168363798945955e-05,
+ "loss": 0.9362,
+ "step": 5755
+ },
+ {
+ "epoch": 15.683923705722071,
+ "grad_norm": 6.825206279754639,
+ "learning_rate": 1.9168011415668392e-05,
+ "loss": 0.6989,
+ "step": 5756
+ },
+ {
+ "epoch": 15.686648501362399,
+ "grad_norm": 6.706659317016602,
+ "learning_rate": 1.916765896099065e-05,
+ "loss": 0.9755,
+ "step": 5757
+ },
+ {
+ "epoch": 15.689373297002724,
+ "grad_norm": 6.210824489593506,
+ "learning_rate": 1.9167306434915475e-05,
+ "loss": 0.8313,
+ "step": 5758
+ },
+ {
+ "epoch": 15.692098092643052,
+ "grad_norm": 5.774549961090088,
+ "learning_rate": 1.9166953837445618e-05,
+ "loss": 0.7968,
+ "step": 5759
+ },
+ {
+ "epoch": 15.69482288828338,
+ "grad_norm": 6.366158962249756,
+ "learning_rate": 1.9166601168583815e-05,
+ "loss": 1.0294,
+ "step": 5760
+ },
+ {
+ "epoch": 15.697547683923705,
+ "grad_norm": 9.248703002929688,
+ "learning_rate": 1.9166248428332816e-05,
+ "loss": 0.8993,
+ "step": 5761
+ },
+ {
+ "epoch": 15.700272479564033,
+ "grad_norm": 6.259676456451416,
+ "learning_rate": 1.9165895616695373e-05,
+ "loss": 0.799,
+ "step": 5762
+ },
+ {
+ "epoch": 15.70299727520436,
+ "grad_norm": 6.421290397644043,
+ "learning_rate": 1.9165542733674226e-05,
+ "loss": 0.9453,
+ "step": 5763
+ },
+ {
+ "epoch": 15.705722070844686,
+ "grad_norm": 5.737821102142334,
+ "learning_rate": 1.916518977927213e-05,
+ "loss": 0.8583,
+ "step": 5764
+ },
+ {
+ "epoch": 15.708446866485014,
+ "grad_norm": 6.452734470367432,
+ "learning_rate": 1.9164836753491834e-05,
+ "loss": 0.9154,
+ "step": 5765
+ },
+ {
+ "epoch": 15.711171662125341,
+ "grad_norm": 5.841000556945801,
+ "learning_rate": 1.916448365633608e-05,
+ "loss": 0.9117,
+ "step": 5766
+ },
+ {
+ "epoch": 15.713896457765667,
+ "grad_norm": 6.851568698883057,
+ "learning_rate": 1.9164130487807623e-05,
+ "loss": 0.7635,
+ "step": 5767
+ },
+ {
+ "epoch": 15.716621253405995,
+ "grad_norm": 6.752496242523193,
+ "learning_rate": 1.916377724790921e-05,
+ "loss": 1.0164,
+ "step": 5768
+ },
+ {
+ "epoch": 15.719346049046322,
+ "grad_norm": 6.905272960662842,
+ "learning_rate": 1.9163423936643598e-05,
+ "loss": 1.0006,
+ "step": 5769
+ },
+ {
+ "epoch": 15.722070844686648,
+ "grad_norm": 6.870218276977539,
+ "learning_rate": 1.9163070554013535e-05,
+ "loss": 0.7601,
+ "step": 5770
+ },
+ {
+ "epoch": 15.724795640326976,
+ "grad_norm": 7.269658088684082,
+ "learning_rate": 1.9162717100021772e-05,
+ "loss": 0.7849,
+ "step": 5771
+ },
+ {
+ "epoch": 15.727520435967303,
+ "grad_norm": 7.203489303588867,
+ "learning_rate": 1.9162363574671062e-05,
+ "loss": 0.7517,
+ "step": 5772
+ },
+ {
+ "epoch": 15.730245231607629,
+ "grad_norm": 7.973775863647461,
+ "learning_rate": 1.9162009977964165e-05,
+ "loss": 0.8706,
+ "step": 5773
+ },
+ {
+ "epoch": 15.732970027247957,
+ "grad_norm": 7.286405563354492,
+ "learning_rate": 1.9161656309903825e-05,
+ "loss": 0.8734,
+ "step": 5774
+ },
+ {
+ "epoch": 15.735694822888284,
+ "grad_norm": 6.231945037841797,
+ "learning_rate": 1.91613025704928e-05,
+ "loss": 0.834,
+ "step": 5775
+ },
+ {
+ "epoch": 15.73841961852861,
+ "grad_norm": 6.39176607131958,
+ "learning_rate": 1.916094875973385e-05,
+ "loss": 0.943,
+ "step": 5776
+ },
+ {
+ "epoch": 15.741144414168938,
+ "grad_norm": 7.048614501953125,
+ "learning_rate": 1.916059487762972e-05,
+ "loss": 0.9695,
+ "step": 5777
+ },
+ {
+ "epoch": 15.743869209809265,
+ "grad_norm": 6.226118087768555,
+ "learning_rate": 1.9160240924183174e-05,
+ "loss": 0.9288,
+ "step": 5778
+ },
+ {
+ "epoch": 15.746594005449591,
+ "grad_norm": 6.214672565460205,
+ "learning_rate": 1.9159886899396966e-05,
+ "loss": 0.8083,
+ "step": 5779
+ },
+ {
+ "epoch": 15.749318801089919,
+ "grad_norm": 8.135629653930664,
+ "learning_rate": 1.9159532803273856e-05,
+ "loss": 0.6698,
+ "step": 5780
+ },
+ {
+ "epoch": 15.752043596730246,
+ "grad_norm": 8.025202751159668,
+ "learning_rate": 1.91591786358166e-05,
+ "loss": 0.9285,
+ "step": 5781
+ },
+ {
+ "epoch": 15.754768392370572,
+ "grad_norm": 6.19317102432251,
+ "learning_rate": 1.9158824397027954e-05,
+ "loss": 0.8495,
+ "step": 5782
+ },
+ {
+ "epoch": 15.7574931880109,
+ "grad_norm": 6.036993503570557,
+ "learning_rate": 1.9158470086910677e-05,
+ "loss": 0.8765,
+ "step": 5783
+ },
+ {
+ "epoch": 15.760217983651227,
+ "grad_norm": 7.619156837463379,
+ "learning_rate": 1.9158115705467528e-05,
+ "loss": 0.8176,
+ "step": 5784
+ },
+ {
+ "epoch": 15.762942779291553,
+ "grad_norm": 6.406201362609863,
+ "learning_rate": 1.9157761252701273e-05,
+ "loss": 0.7764,
+ "step": 5785
+ },
+ {
+ "epoch": 15.76566757493188,
+ "grad_norm": 6.501182556152344,
+ "learning_rate": 1.915740672861467e-05,
+ "loss": 0.9984,
+ "step": 5786
+ },
+ {
+ "epoch": 15.768392370572208,
+ "grad_norm": 7.164403915405273,
+ "learning_rate": 1.915705213321047e-05,
+ "loss": 0.8083,
+ "step": 5787
+ },
+ {
+ "epoch": 15.771117166212534,
+ "grad_norm": 6.710945129394531,
+ "learning_rate": 1.9156697466491447e-05,
+ "loss": 0.8049,
+ "step": 5788
+ },
+ {
+ "epoch": 15.773841961852861,
+ "grad_norm": 6.996535301208496,
+ "learning_rate": 1.915634272846036e-05,
+ "loss": 0.838,
+ "step": 5789
+ },
+ {
+ "epoch": 15.776566757493189,
+ "grad_norm": 9.282404899597168,
+ "learning_rate": 1.9155987919119973e-05,
+ "loss": 0.9637,
+ "step": 5790
+ },
+ {
+ "epoch": 15.779291553133515,
+ "grad_norm": 7.332329750061035,
+ "learning_rate": 1.915563303847304e-05,
+ "loss": 0.8324,
+ "step": 5791
+ },
+ {
+ "epoch": 15.782016348773842,
+ "grad_norm": 7.8522257804870605,
+ "learning_rate": 1.9155278086522334e-05,
+ "loss": 0.8036,
+ "step": 5792
+ },
+ {
+ "epoch": 15.78474114441417,
+ "grad_norm": 7.629213333129883,
+ "learning_rate": 1.915492306327062e-05,
+ "loss": 0.9206,
+ "step": 5793
+ },
+ {
+ "epoch": 15.787465940054496,
+ "grad_norm": 6.015303611755371,
+ "learning_rate": 1.9154567968720656e-05,
+ "loss": 0.6599,
+ "step": 5794
+ },
+ {
+ "epoch": 15.790190735694823,
+ "grad_norm": 10.811718940734863,
+ "learning_rate": 1.9154212802875215e-05,
+ "loss": 0.8517,
+ "step": 5795
+ },
+ {
+ "epoch": 15.79291553133515,
+ "grad_norm": 11.415766716003418,
+ "learning_rate": 1.9153857565737058e-05,
+ "loss": 0.7698,
+ "step": 5796
+ },
+ {
+ "epoch": 15.795640326975477,
+ "grad_norm": 7.590892314910889,
+ "learning_rate": 1.9153502257308952e-05,
+ "loss": 0.8029,
+ "step": 5797
+ },
+ {
+ "epoch": 15.798365122615804,
+ "grad_norm": 4.9959845542907715,
+ "learning_rate": 1.9153146877593666e-05,
+ "loss": 1.0223,
+ "step": 5798
+ },
+ {
+ "epoch": 15.80108991825613,
+ "grad_norm": 5.715133190155029,
+ "learning_rate": 1.9152791426593964e-05,
+ "loss": 0.672,
+ "step": 5799
+ },
+ {
+ "epoch": 15.803814713896458,
+ "grad_norm": 6.843607425689697,
+ "learning_rate": 1.9152435904312617e-05,
+ "loss": 0.7843,
+ "step": 5800
+ },
+ {
+ "epoch": 15.806539509536785,
+ "grad_norm": 7.882494926452637,
+ "learning_rate": 1.9152080310752398e-05,
+ "loss": 0.9387,
+ "step": 5801
+ },
+ {
+ "epoch": 15.809264305177111,
+ "grad_norm": 6.637618064880371,
+ "learning_rate": 1.9151724645916067e-05,
+ "loss": 0.7522,
+ "step": 5802
+ },
+ {
+ "epoch": 15.811989100817438,
+ "grad_norm": 7.111424446105957,
+ "learning_rate": 1.9151368909806403e-05,
+ "loss": 0.9568,
+ "step": 5803
+ },
+ {
+ "epoch": 15.814713896457766,
+ "grad_norm": 5.895694732666016,
+ "learning_rate": 1.9151013102426174e-05,
+ "loss": 0.8414,
+ "step": 5804
+ },
+ {
+ "epoch": 15.817438692098092,
+ "grad_norm": 7.2356648445129395,
+ "learning_rate": 1.9150657223778144e-05,
+ "loss": 0.9486,
+ "step": 5805
+ },
+ {
+ "epoch": 15.82016348773842,
+ "grad_norm": 7.7642388343811035,
+ "learning_rate": 1.9150301273865093e-05,
+ "loss": 1.0056,
+ "step": 5806
+ },
+ {
+ "epoch": 15.822888283378747,
+ "grad_norm": 7.225085735321045,
+ "learning_rate": 1.914994525268979e-05,
+ "loss": 0.7854,
+ "step": 5807
+ },
+ {
+ "epoch": 15.825613079019073,
+ "grad_norm": 5.893949508666992,
+ "learning_rate": 1.914958916025501e-05,
+ "loss": 0.7853,
+ "step": 5808
+ },
+ {
+ "epoch": 15.8283378746594,
+ "grad_norm": 6.320692539215088,
+ "learning_rate": 1.914923299656352e-05,
+ "loss": 0.9812,
+ "step": 5809
+ },
+ {
+ "epoch": 15.831062670299728,
+ "grad_norm": 6.904065132141113,
+ "learning_rate": 1.91488767616181e-05,
+ "loss": 0.6854,
+ "step": 5810
+ },
+ {
+ "epoch": 15.833787465940054,
+ "grad_norm": 7.7707390785217285,
+ "learning_rate": 1.914852045542152e-05,
+ "loss": 0.9803,
+ "step": 5811
+ },
+ {
+ "epoch": 15.836512261580381,
+ "grad_norm": 8.346186637878418,
+ "learning_rate": 1.914816407797656e-05,
+ "loss": 0.971,
+ "step": 5812
+ },
+ {
+ "epoch": 15.839237057220709,
+ "grad_norm": 6.175897121429443,
+ "learning_rate": 1.9147807629285993e-05,
+ "loss": 0.7538,
+ "step": 5813
+ },
+ {
+ "epoch": 15.841961852861035,
+ "grad_norm": 6.433761119842529,
+ "learning_rate": 1.9147451109352592e-05,
+ "loss": 0.6831,
+ "step": 5814
+ },
+ {
+ "epoch": 15.844686648501362,
+ "grad_norm": 8.20675277709961,
+ "learning_rate": 1.9147094518179136e-05,
+ "loss": 0.8671,
+ "step": 5815
+ },
+ {
+ "epoch": 15.84741144414169,
+ "grad_norm": 6.3148722648620605,
+ "learning_rate": 1.9146737855768407e-05,
+ "loss": 0.8359,
+ "step": 5816
+ },
+ {
+ "epoch": 15.850136239782016,
+ "grad_norm": 7.4485321044921875,
+ "learning_rate": 1.9146381122123176e-05,
+ "loss": 0.8394,
+ "step": 5817
+ },
+ {
+ "epoch": 15.852861035422343,
+ "grad_norm": 7.943326950073242,
+ "learning_rate": 1.9146024317246217e-05,
+ "loss": 0.7914,
+ "step": 5818
+ },
+ {
+ "epoch": 15.85558583106267,
+ "grad_norm": 6.290095806121826,
+ "learning_rate": 1.914566744114032e-05,
+ "loss": 0.6837,
+ "step": 5819
+ },
+ {
+ "epoch": 15.858310626702997,
+ "grad_norm": 6.775827407836914,
+ "learning_rate": 1.9145310493808264e-05,
+ "loss": 0.6675,
+ "step": 5820
+ },
+ {
+ "epoch": 15.861035422343324,
+ "grad_norm": 6.625891208648682,
+ "learning_rate": 1.9144953475252818e-05,
+ "loss": 0.7229,
+ "step": 5821
+ },
+ {
+ "epoch": 15.863760217983652,
+ "grad_norm": 7.4435811042785645,
+ "learning_rate": 1.914459638547677e-05,
+ "loss": 1.0129,
+ "step": 5822
+ },
+ {
+ "epoch": 15.866485013623977,
+ "grad_norm": 7.048725128173828,
+ "learning_rate": 1.91442392244829e-05,
+ "loss": 0.7068,
+ "step": 5823
+ },
+ {
+ "epoch": 15.869209809264305,
+ "grad_norm": 6.099376201629639,
+ "learning_rate": 1.9143881992273988e-05,
+ "loss": 0.8258,
+ "step": 5824
+ },
+ {
+ "epoch": 15.871934604904633,
+ "grad_norm": 7.386951446533203,
+ "learning_rate": 1.914352468885282e-05,
+ "loss": 0.8246,
+ "step": 5825
+ },
+ {
+ "epoch": 15.874659400544958,
+ "grad_norm": 7.296832084655762,
+ "learning_rate": 1.9143167314222172e-05,
+ "loss": 0.7937,
+ "step": 5826
+ },
+ {
+ "epoch": 15.877384196185286,
+ "grad_norm": 7.06572961807251,
+ "learning_rate": 1.9142809868384836e-05,
+ "loss": 0.8932,
+ "step": 5827
+ },
+ {
+ "epoch": 15.880108991825614,
+ "grad_norm": 5.549808025360107,
+ "learning_rate": 1.9142452351343585e-05,
+ "loss": 0.9347,
+ "step": 5828
+ },
+ {
+ "epoch": 15.88283378746594,
+ "grad_norm": 7.106066703796387,
+ "learning_rate": 1.9142094763101216e-05,
+ "loss": 1.0645,
+ "step": 5829
+ },
+ {
+ "epoch": 15.885558583106267,
+ "grad_norm": 7.044342041015625,
+ "learning_rate": 1.9141737103660503e-05,
+ "loss": 0.9331,
+ "step": 5830
+ },
+ {
+ "epoch": 15.888283378746594,
+ "grad_norm": 5.535400867462158,
+ "learning_rate": 1.9141379373024235e-05,
+ "loss": 0.7642,
+ "step": 5831
+ },
+ {
+ "epoch": 15.89100817438692,
+ "grad_norm": 7.408315658569336,
+ "learning_rate": 1.9141021571195204e-05,
+ "loss": 0.7627,
+ "step": 5832
+ },
+ {
+ "epoch": 15.893732970027248,
+ "grad_norm": 7.5653395652771,
+ "learning_rate": 1.9140663698176183e-05,
+ "loss": 0.9247,
+ "step": 5833
+ },
+ {
+ "epoch": 15.896457765667575,
+ "grad_norm": 6.435869216918945,
+ "learning_rate": 1.9140305753969977e-05,
+ "loss": 0.8177,
+ "step": 5834
+ },
+ {
+ "epoch": 15.899182561307901,
+ "grad_norm": 5.750683307647705,
+ "learning_rate": 1.913994773857936e-05,
+ "loss": 1.0347,
+ "step": 5835
+ },
+ {
+ "epoch": 15.901907356948229,
+ "grad_norm": 7.024409770965576,
+ "learning_rate": 1.913958965200712e-05,
+ "loss": 0.9099,
+ "step": 5836
+ },
+ {
+ "epoch": 15.904632152588556,
+ "grad_norm": 6.379297733306885,
+ "learning_rate": 1.9139231494256054e-05,
+ "loss": 0.946,
+ "step": 5837
+ },
+ {
+ "epoch": 15.907356948228882,
+ "grad_norm": 8.775691986083984,
+ "learning_rate": 1.9138873265328944e-05,
+ "loss": 0.6967,
+ "step": 5838
+ },
+ {
+ "epoch": 15.91008174386921,
+ "grad_norm": 6.851693153381348,
+ "learning_rate": 1.9138514965228587e-05,
+ "loss": 0.9529,
+ "step": 5839
+ },
+ {
+ "epoch": 15.912806539509537,
+ "grad_norm": 6.384395599365234,
+ "learning_rate": 1.9138156593957767e-05,
+ "loss": 0.8943,
+ "step": 5840
+ },
+ {
+ "epoch": 15.915531335149863,
+ "grad_norm": 7.3799591064453125,
+ "learning_rate": 1.913779815151928e-05,
+ "loss": 0.7863,
+ "step": 5841
+ },
+ {
+ "epoch": 15.91825613079019,
+ "grad_norm": 5.607107639312744,
+ "learning_rate": 1.9137439637915912e-05,
+ "loss": 0.7784,
+ "step": 5842
+ },
+ {
+ "epoch": 15.920980926430518,
+ "grad_norm": 8.555621147155762,
+ "learning_rate": 1.9137081053150465e-05,
+ "loss": 0.8181,
+ "step": 5843
+ },
+ {
+ "epoch": 15.923705722070844,
+ "grad_norm": 6.665183067321777,
+ "learning_rate": 1.913672239722572e-05,
+ "loss": 0.8885,
+ "step": 5844
+ },
+ {
+ "epoch": 15.926430517711172,
+ "grad_norm": 6.748533725738525,
+ "learning_rate": 1.9136363670144472e-05,
+ "loss": 0.8528,
+ "step": 5845
+ },
+ {
+ "epoch": 15.9291553133515,
+ "grad_norm": 6.989460468292236,
+ "learning_rate": 1.913600487190952e-05,
+ "loss": 0.7178,
+ "step": 5846
+ },
+ {
+ "epoch": 15.931880108991825,
+ "grad_norm": 8.220732688903809,
+ "learning_rate": 1.913564600252366e-05,
+ "loss": 0.6922,
+ "step": 5847
+ },
+ {
+ "epoch": 15.934604904632153,
+ "grad_norm": 6.726073265075684,
+ "learning_rate": 1.9135287061989677e-05,
+ "loss": 0.9069,
+ "step": 5848
+ },
+ {
+ "epoch": 15.93732970027248,
+ "grad_norm": 5.286011695861816,
+ "learning_rate": 1.9134928050310374e-05,
+ "loss": 0.8733,
+ "step": 5849
+ },
+ {
+ "epoch": 15.940054495912806,
+ "grad_norm": 6.190932273864746,
+ "learning_rate": 1.9134568967488548e-05,
+ "loss": 0.736,
+ "step": 5850
+ },
+ {
+ "epoch": 15.942779291553133,
+ "grad_norm": 7.049252510070801,
+ "learning_rate": 1.9134209813526992e-05,
+ "loss": 0.8224,
+ "step": 5851
+ },
+ {
+ "epoch": 15.945504087193461,
+ "grad_norm": 5.984988689422607,
+ "learning_rate": 1.9133850588428503e-05,
+ "loss": 0.8229,
+ "step": 5852
+ },
+ {
+ "epoch": 15.948228882833787,
+ "grad_norm": 5.805099964141846,
+ "learning_rate": 1.9133491292195878e-05,
+ "loss": 0.8472,
+ "step": 5853
+ },
+ {
+ "epoch": 15.950953678474114,
+ "grad_norm": 7.318448066711426,
+ "learning_rate": 1.9133131924831917e-05,
+ "loss": 0.8337,
+ "step": 5854
+ },
+ {
+ "epoch": 15.953678474114442,
+ "grad_norm": 9.317394256591797,
+ "learning_rate": 1.913277248633942e-05,
+ "loss": 0.7816,
+ "step": 5855
+ },
+ {
+ "epoch": 15.956403269754768,
+ "grad_norm": 5.69371223449707,
+ "learning_rate": 1.9132412976721185e-05,
+ "loss": 0.92,
+ "step": 5856
+ },
+ {
+ "epoch": 15.959128065395095,
+ "grad_norm": 7.1621904373168945,
+ "learning_rate": 1.9132053395980013e-05,
+ "loss": 0.9084,
+ "step": 5857
+ },
+ {
+ "epoch": 15.961852861035423,
+ "grad_norm": 6.71399450302124,
+ "learning_rate": 1.91316937441187e-05,
+ "loss": 0.7166,
+ "step": 5858
+ },
+ {
+ "epoch": 15.964577656675749,
+ "grad_norm": 5.74859094619751,
+ "learning_rate": 1.9131334021140052e-05,
+ "loss": 0.7606,
+ "step": 5859
+ },
+ {
+ "epoch": 15.967302452316076,
+ "grad_norm": 6.332076549530029,
+ "learning_rate": 1.913097422704687e-05,
+ "loss": 0.8687,
+ "step": 5860
+ },
+ {
+ "epoch": 15.970027247956404,
+ "grad_norm": 5.929424285888672,
+ "learning_rate": 1.913061436184195e-05,
+ "loss": 0.9536,
+ "step": 5861
+ },
+ {
+ "epoch": 15.97275204359673,
+ "grad_norm": 7.00347900390625,
+ "learning_rate": 1.9130254425528102e-05,
+ "loss": 0.7335,
+ "step": 5862
+ },
+ {
+ "epoch": 15.975476839237057,
+ "grad_norm": 6.681676387786865,
+ "learning_rate": 1.912989441810813e-05,
+ "loss": 0.7062,
+ "step": 5863
+ },
+ {
+ "epoch": 15.978201634877385,
+ "grad_norm": 6.001529693603516,
+ "learning_rate": 1.912953433958483e-05,
+ "loss": 0.8563,
+ "step": 5864
+ },
+ {
+ "epoch": 15.98092643051771,
+ "grad_norm": 6.9956278800964355,
+ "learning_rate": 1.9129174189961008e-05,
+ "loss": 0.9441,
+ "step": 5865
+ },
+ {
+ "epoch": 15.983651226158038,
+ "grad_norm": 6.675997734069824,
+ "learning_rate": 1.9128813969239477e-05,
+ "loss": 0.8309,
+ "step": 5866
+ },
+ {
+ "epoch": 15.986376021798366,
+ "grad_norm": 7.460393905639648,
+ "learning_rate": 1.9128453677423032e-05,
+ "loss": 0.957,
+ "step": 5867
+ },
+ {
+ "epoch": 15.989100817438691,
+ "grad_norm": 6.839108467102051,
+ "learning_rate": 1.9128093314514485e-05,
+ "loss": 0.8009,
+ "step": 5868
+ },
+ {
+ "epoch": 15.991825613079019,
+ "grad_norm": 6.485276699066162,
+ "learning_rate": 1.912773288051664e-05,
+ "loss": 0.651,
+ "step": 5869
+ },
+ {
+ "epoch": 15.994550408719347,
+ "grad_norm": 8.217398643493652,
+ "learning_rate": 1.912737237543231e-05,
+ "loss": 0.8849,
+ "step": 5870
+ },
+ {
+ "epoch": 15.997275204359672,
+ "grad_norm": 7.335553169250488,
+ "learning_rate": 1.9127011799264292e-05,
+ "loss": 0.8571,
+ "step": 5871
+ },
+ {
+ "epoch": 16.0,
+ "grad_norm": 6.4670820236206055,
+ "learning_rate": 1.9126651152015404e-05,
+ "loss": 0.6832,
+ "step": 5872
+ },
+ {
+ "epoch": 16.002724795640326,
+ "grad_norm": 6.90968132019043,
+ "learning_rate": 1.9126290433688448e-05,
+ "loss": 0.8047,
+ "step": 5873
+ },
+ {
+ "epoch": 16.005449591280655,
+ "grad_norm": 7.128833293914795,
+ "learning_rate": 1.9125929644286235e-05,
+ "loss": 0.7675,
+ "step": 5874
+ },
+ {
+ "epoch": 16.00817438692098,
+ "grad_norm": 6.856328964233398,
+ "learning_rate": 1.9125568783811578e-05,
+ "loss": 0.6809,
+ "step": 5875
+ },
+ {
+ "epoch": 16.010899182561307,
+ "grad_norm": 5.821172714233398,
+ "learning_rate": 1.9125207852267286e-05,
+ "loss": 0.6826,
+ "step": 5876
+ },
+ {
+ "epoch": 16.013623978201636,
+ "grad_norm": 6.329996109008789,
+ "learning_rate": 1.9124846849656164e-05,
+ "loss": 0.7458,
+ "step": 5877
+ },
+ {
+ "epoch": 16.016348773841962,
+ "grad_norm": 6.113714694976807,
+ "learning_rate": 1.9124485775981034e-05,
+ "loss": 0.7715,
+ "step": 5878
+ },
+ {
+ "epoch": 16.019073569482288,
+ "grad_norm": 6.206613540649414,
+ "learning_rate": 1.9124124631244698e-05,
+ "loss": 0.7045,
+ "step": 5879
+ },
+ {
+ "epoch": 16.021798365122617,
+ "grad_norm": 6.224862098693848,
+ "learning_rate": 1.9123763415449972e-05,
+ "loss": 0.8536,
+ "step": 5880
+ },
+ {
+ "epoch": 16.024523160762943,
+ "grad_norm": 6.136050701141357,
+ "learning_rate": 1.9123402128599675e-05,
+ "loss": 0.6921,
+ "step": 5881
+ },
+ {
+ "epoch": 16.02724795640327,
+ "grad_norm": 6.716785430908203,
+ "learning_rate": 1.9123040770696616e-05,
+ "loss": 0.7418,
+ "step": 5882
+ },
+ {
+ "epoch": 16.029972752043598,
+ "grad_norm": 6.387701034545898,
+ "learning_rate": 1.9122679341743606e-05,
+ "loss": 0.6746,
+ "step": 5883
+ },
+ {
+ "epoch": 16.032697547683924,
+ "grad_norm": 7.007884979248047,
+ "learning_rate": 1.912231784174346e-05,
+ "loss": 0.8228,
+ "step": 5884
+ },
+ {
+ "epoch": 16.03542234332425,
+ "grad_norm": 7.064517974853516,
+ "learning_rate": 1.9121956270699005e-05,
+ "loss": 0.7455,
+ "step": 5885
+ },
+ {
+ "epoch": 16.03814713896458,
+ "grad_norm": 8.262537956237793,
+ "learning_rate": 1.9121594628613044e-05,
+ "loss": 0.6998,
+ "step": 5886
+ },
+ {
+ "epoch": 16.040871934604905,
+ "grad_norm": 6.61881685256958,
+ "learning_rate": 1.91212329154884e-05,
+ "loss": 0.7762,
+ "step": 5887
+ },
+ {
+ "epoch": 16.04359673024523,
+ "grad_norm": 5.5625081062316895,
+ "learning_rate": 1.912087113132788e-05,
+ "loss": 0.6599,
+ "step": 5888
+ },
+ {
+ "epoch": 16.04632152588556,
+ "grad_norm": 7.218498229980469,
+ "learning_rate": 1.9120509276134315e-05,
+ "loss": 0.8407,
+ "step": 5889
+ },
+ {
+ "epoch": 16.049046321525886,
+ "grad_norm": 6.475902557373047,
+ "learning_rate": 1.9120147349910515e-05,
+ "loss": 0.8175,
+ "step": 5890
+ },
+ {
+ "epoch": 16.05177111716621,
+ "grad_norm": 6.492495059967041,
+ "learning_rate": 1.9119785352659303e-05,
+ "loss": 0.718,
+ "step": 5891
+ },
+ {
+ "epoch": 16.05449591280654,
+ "grad_norm": 6.20407772064209,
+ "learning_rate": 1.9119423284383494e-05,
+ "loss": 0.9092,
+ "step": 5892
+ },
+ {
+ "epoch": 16.057220708446867,
+ "grad_norm": 6.733720779418945,
+ "learning_rate": 1.9119061145085913e-05,
+ "loss": 0.6893,
+ "step": 5893
+ },
+ {
+ "epoch": 16.059945504087192,
+ "grad_norm": 5.840706825256348,
+ "learning_rate": 1.9118698934769374e-05,
+ "loss": 0.7776,
+ "step": 5894
+ },
+ {
+ "epoch": 16.06267029972752,
+ "grad_norm": 6.546199798583984,
+ "learning_rate": 1.91183366534367e-05,
+ "loss": 0.9122,
+ "step": 5895
+ },
+ {
+ "epoch": 16.065395095367847,
+ "grad_norm": 6.108469009399414,
+ "learning_rate": 1.911797430109072e-05,
+ "loss": 0.7568,
+ "step": 5896
+ },
+ {
+ "epoch": 16.068119891008173,
+ "grad_norm": 6.564669132232666,
+ "learning_rate": 1.9117611877734244e-05,
+ "loss": 0.7784,
+ "step": 5897
+ },
+ {
+ "epoch": 16.070844686648503,
+ "grad_norm": 6.777740955352783,
+ "learning_rate": 1.9117249383370105e-05,
+ "loss": 0.6998,
+ "step": 5898
+ },
+ {
+ "epoch": 16.07356948228883,
+ "grad_norm": 7.013524532318115,
+ "learning_rate": 1.9116886818001118e-05,
+ "loss": 0.8253,
+ "step": 5899
+ },
+ {
+ "epoch": 16.076294277929154,
+ "grad_norm": 7.3758978843688965,
+ "learning_rate": 1.9116524181630106e-05,
+ "loss": 0.9147,
+ "step": 5900
+ },
+ {
+ "epoch": 16.079019073569484,
+ "grad_norm": 5.975619792938232,
+ "learning_rate": 1.9116161474259902e-05,
+ "loss": 0.8406,
+ "step": 5901
+ },
+ {
+ "epoch": 16.08174386920981,
+ "grad_norm": 6.557789325714111,
+ "learning_rate": 1.9115798695893327e-05,
+ "loss": 0.7585,
+ "step": 5902
+ },
+ {
+ "epoch": 16.084468664850135,
+ "grad_norm": 7.081116676330566,
+ "learning_rate": 1.9115435846533204e-05,
+ "loss": 0.7911,
+ "step": 5903
+ },
+ {
+ "epoch": 16.087193460490465,
+ "grad_norm": 7.696207046508789,
+ "learning_rate": 1.9115072926182358e-05,
+ "loss": 0.6847,
+ "step": 5904
+ },
+ {
+ "epoch": 16.08991825613079,
+ "grad_norm": 6.8146586418151855,
+ "learning_rate": 1.911470993484362e-05,
+ "loss": 0.6952,
+ "step": 5905
+ },
+ {
+ "epoch": 16.092643051771116,
+ "grad_norm": 5.633756160736084,
+ "learning_rate": 1.911434687251981e-05,
+ "loss": 0.6611,
+ "step": 5906
+ },
+ {
+ "epoch": 16.095367847411445,
+ "grad_norm": 6.456246852874756,
+ "learning_rate": 1.9113983739213765e-05,
+ "loss": 0.7195,
+ "step": 5907
+ },
+ {
+ "epoch": 16.09809264305177,
+ "grad_norm": 6.737067222595215,
+ "learning_rate": 1.9113620534928307e-05,
+ "loss": 0.7029,
+ "step": 5908
+ },
+ {
+ "epoch": 16.100817438692097,
+ "grad_norm": 7.3578948974609375,
+ "learning_rate": 1.9113257259666262e-05,
+ "loss": 0.8208,
+ "step": 5909
+ },
+ {
+ "epoch": 16.103542234332426,
+ "grad_norm": 6.300441265106201,
+ "learning_rate": 1.9112893913430466e-05,
+ "loss": 0.7772,
+ "step": 5910
+ },
+ {
+ "epoch": 16.106267029972752,
+ "grad_norm": 5.543631553649902,
+ "learning_rate": 1.9112530496223748e-05,
+ "loss": 0.8356,
+ "step": 5911
+ },
+ {
+ "epoch": 16.108991825613078,
+ "grad_norm": 6.640744209289551,
+ "learning_rate": 1.911216700804893e-05,
+ "loss": 0.6301,
+ "step": 5912
+ },
+ {
+ "epoch": 16.111716621253407,
+ "grad_norm": 5.794637203216553,
+ "learning_rate": 1.911180344890885e-05,
+ "loss": 0.6647,
+ "step": 5913
+ },
+ {
+ "epoch": 16.114441416893733,
+ "grad_norm": 6.540340900421143,
+ "learning_rate": 1.9111439818806338e-05,
+ "loss": 0.7202,
+ "step": 5914
+ },
+ {
+ "epoch": 16.11716621253406,
+ "grad_norm": 6.564042091369629,
+ "learning_rate": 1.911107611774423e-05,
+ "loss": 0.5585,
+ "step": 5915
+ },
+ {
+ "epoch": 16.11989100817439,
+ "grad_norm": 6.179925918579102,
+ "learning_rate": 1.911071234572535e-05,
+ "loss": 0.7448,
+ "step": 5916
+ },
+ {
+ "epoch": 16.122615803814714,
+ "grad_norm": 6.678143501281738,
+ "learning_rate": 1.9110348502752537e-05,
+ "loss": 0.7947,
+ "step": 5917
+ },
+ {
+ "epoch": 16.12534059945504,
+ "grad_norm": 6.12836217880249,
+ "learning_rate": 1.9109984588828622e-05,
+ "loss": 0.6464,
+ "step": 5918
+ },
+ {
+ "epoch": 16.12806539509537,
+ "grad_norm": 6.328923225402832,
+ "learning_rate": 1.9109620603956442e-05,
+ "loss": 0.7947,
+ "step": 5919
+ },
+ {
+ "epoch": 16.130790190735695,
+ "grad_norm": 5.1486287117004395,
+ "learning_rate": 1.9109256548138832e-05,
+ "loss": 0.7683,
+ "step": 5920
+ },
+ {
+ "epoch": 16.13351498637602,
+ "grad_norm": 5.606860160827637,
+ "learning_rate": 1.910889242137862e-05,
+ "loss": 0.8109,
+ "step": 5921
+ },
+ {
+ "epoch": 16.13623978201635,
+ "grad_norm": 6.558614730834961,
+ "learning_rate": 1.9108528223678652e-05,
+ "loss": 0.7444,
+ "step": 5922
+ },
+ {
+ "epoch": 16.138964577656676,
+ "grad_norm": 7.097546100616455,
+ "learning_rate": 1.9108163955041754e-05,
+ "loss": 0.6744,
+ "step": 5923
+ },
+ {
+ "epoch": 16.141689373297,
+ "grad_norm": 5.7642059326171875,
+ "learning_rate": 1.9107799615470773e-05,
+ "loss": 0.7715,
+ "step": 5924
+ },
+ {
+ "epoch": 16.14441416893733,
+ "grad_norm": 5.838220119476318,
+ "learning_rate": 1.9107435204968538e-05,
+ "loss": 0.7586,
+ "step": 5925
+ },
+ {
+ "epoch": 16.147138964577657,
+ "grad_norm": 6.609309673309326,
+ "learning_rate": 1.910707072353789e-05,
+ "loss": 0.8116,
+ "step": 5926
+ },
+ {
+ "epoch": 16.149863760217983,
+ "grad_norm": 5.974928379058838,
+ "learning_rate": 1.9106706171181673e-05,
+ "loss": 0.7491,
+ "step": 5927
+ },
+ {
+ "epoch": 16.152588555858312,
+ "grad_norm": 8.344934463500977,
+ "learning_rate": 1.910634154790272e-05,
+ "loss": 0.8746,
+ "step": 5928
+ },
+ {
+ "epoch": 16.155313351498638,
+ "grad_norm": 6.7588300704956055,
+ "learning_rate": 1.9105976853703875e-05,
+ "loss": 0.8109,
+ "step": 5929
+ },
+ {
+ "epoch": 16.158038147138964,
+ "grad_norm": 6.443511962890625,
+ "learning_rate": 1.910561208858797e-05,
+ "loss": 0.801,
+ "step": 5930
+ },
+ {
+ "epoch": 16.160762942779293,
+ "grad_norm": 5.911397457122803,
+ "learning_rate": 1.910524725255785e-05,
+ "loss": 0.6277,
+ "step": 5931
+ },
+ {
+ "epoch": 16.16348773841962,
+ "grad_norm": 7.966153621673584,
+ "learning_rate": 1.9104882345616363e-05,
+ "loss": 0.7987,
+ "step": 5932
+ },
+ {
+ "epoch": 16.166212534059945,
+ "grad_norm": 5.962772846221924,
+ "learning_rate": 1.9104517367766344e-05,
+ "loss": 0.7925,
+ "step": 5933
+ },
+ {
+ "epoch": 16.168937329700274,
+ "grad_norm": 6.967324733734131,
+ "learning_rate": 1.9104152319010634e-05,
+ "loss": 0.8164,
+ "step": 5934
+ },
+ {
+ "epoch": 16.1716621253406,
+ "grad_norm": 7.45419979095459,
+ "learning_rate": 1.9103787199352077e-05,
+ "loss": 0.6599,
+ "step": 5935
+ },
+ {
+ "epoch": 16.174386920980925,
+ "grad_norm": 6.625638484954834,
+ "learning_rate": 1.9103422008793524e-05,
+ "loss": 0.8076,
+ "step": 5936
+ },
+ {
+ "epoch": 16.177111716621255,
+ "grad_norm": 5.833662986755371,
+ "learning_rate": 1.9103056747337813e-05,
+ "loss": 0.8146,
+ "step": 5937
+ },
+ {
+ "epoch": 16.17983651226158,
+ "grad_norm": 6.794573783874512,
+ "learning_rate": 1.9102691414987785e-05,
+ "loss": 0.8016,
+ "step": 5938
+ },
+ {
+ "epoch": 16.182561307901906,
+ "grad_norm": 6.238320350646973,
+ "learning_rate": 1.910232601174629e-05,
+ "loss": 0.7515,
+ "step": 5939
+ },
+ {
+ "epoch": 16.185286103542236,
+ "grad_norm": 8.153271675109863,
+ "learning_rate": 1.9101960537616174e-05,
+ "loss": 0.8971,
+ "step": 5940
+ },
+ {
+ "epoch": 16.18801089918256,
+ "grad_norm": 6.311692714691162,
+ "learning_rate": 1.9101594992600283e-05,
+ "loss": 0.9589,
+ "step": 5941
+ },
+ {
+ "epoch": 16.190735694822887,
+ "grad_norm": 8.563075065612793,
+ "learning_rate": 1.9101229376701463e-05,
+ "loss": 0.7478,
+ "step": 5942
+ },
+ {
+ "epoch": 16.193460490463217,
+ "grad_norm": 6.834074020385742,
+ "learning_rate": 1.910086368992256e-05,
+ "loss": 0.7039,
+ "step": 5943
+ },
+ {
+ "epoch": 16.196185286103542,
+ "grad_norm": 7.05785608291626,
+ "learning_rate": 1.9100497932266423e-05,
+ "loss": 0.743,
+ "step": 5944
+ },
+ {
+ "epoch": 16.19891008174387,
+ "grad_norm": 6.088530540466309,
+ "learning_rate": 1.9100132103735906e-05,
+ "loss": 0.6478,
+ "step": 5945
+ },
+ {
+ "epoch": 16.201634877384198,
+ "grad_norm": 6.519425392150879,
+ "learning_rate": 1.9099766204333847e-05,
+ "loss": 0.9271,
+ "step": 5946
+ },
+ {
+ "epoch": 16.204359673024523,
+ "grad_norm": 7.9030327796936035,
+ "learning_rate": 1.9099400234063107e-05,
+ "loss": 0.8002,
+ "step": 5947
+ },
+ {
+ "epoch": 16.20708446866485,
+ "grad_norm": 7.0459442138671875,
+ "learning_rate": 1.9099034192926527e-05,
+ "loss": 0.724,
+ "step": 5948
+ },
+ {
+ "epoch": 16.20980926430518,
+ "grad_norm": 6.174426078796387,
+ "learning_rate": 1.909866808092696e-05,
+ "loss": 0.8162,
+ "step": 5949
+ },
+ {
+ "epoch": 16.212534059945504,
+ "grad_norm": 5.904473781585693,
+ "learning_rate": 1.9098301898067264e-05,
+ "loss": 0.6277,
+ "step": 5950
+ },
+ {
+ "epoch": 16.21525885558583,
+ "grad_norm": 6.9940996170043945,
+ "learning_rate": 1.9097935644350284e-05,
+ "loss": 0.7338,
+ "step": 5951
+ },
+ {
+ "epoch": 16.21798365122616,
+ "grad_norm": 6.569117069244385,
+ "learning_rate": 1.909756931977887e-05,
+ "loss": 0.7844,
+ "step": 5952
+ },
+ {
+ "epoch": 16.220708446866485,
+ "grad_norm": 6.031002998352051,
+ "learning_rate": 1.9097202924355884e-05,
+ "loss": 0.8755,
+ "step": 5953
+ },
+ {
+ "epoch": 16.22343324250681,
+ "grad_norm": 8.042561531066895,
+ "learning_rate": 1.909683645808417e-05,
+ "loss": 0.8591,
+ "step": 5954
+ },
+ {
+ "epoch": 16.22615803814714,
+ "grad_norm": 7.074176788330078,
+ "learning_rate": 1.9096469920966592e-05,
+ "loss": 0.7994,
+ "step": 5955
+ },
+ {
+ "epoch": 16.228882833787466,
+ "grad_norm": 5.777893543243408,
+ "learning_rate": 1.9096103313005992e-05,
+ "loss": 0.7089,
+ "step": 5956
+ },
+ {
+ "epoch": 16.231607629427792,
+ "grad_norm": 7.541585922241211,
+ "learning_rate": 1.909573663420524e-05,
+ "loss": 0.6281,
+ "step": 5957
+ },
+ {
+ "epoch": 16.23433242506812,
+ "grad_norm": 5.780476093292236,
+ "learning_rate": 1.9095369884567178e-05,
+ "loss": 0.8124,
+ "step": 5958
+ },
+ {
+ "epoch": 16.237057220708447,
+ "grad_norm": 6.806787014007568,
+ "learning_rate": 1.909500306409467e-05,
+ "loss": 0.7896,
+ "step": 5959
+ },
+ {
+ "epoch": 16.239782016348773,
+ "grad_norm": 6.394571304321289,
+ "learning_rate": 1.909463617279057e-05,
+ "loss": 0.5951,
+ "step": 5960
+ },
+ {
+ "epoch": 16.242506811989102,
+ "grad_norm": 11.314346313476562,
+ "learning_rate": 1.9094269210657737e-05,
+ "loss": 0.9081,
+ "step": 5961
+ },
+ {
+ "epoch": 16.245231607629428,
+ "grad_norm": 7.0076680183410645,
+ "learning_rate": 1.9093902177699028e-05,
+ "loss": 0.8959,
+ "step": 5962
+ },
+ {
+ "epoch": 16.247956403269754,
+ "grad_norm": 6.294768810272217,
+ "learning_rate": 1.9093535073917304e-05,
+ "loss": 0.842,
+ "step": 5963
+ },
+ {
+ "epoch": 16.250681198910083,
+ "grad_norm": 6.9344162940979,
+ "learning_rate": 1.9093167899315417e-05,
+ "loss": 0.8876,
+ "step": 5964
+ },
+ {
+ "epoch": 16.25340599455041,
+ "grad_norm": 6.383873462677002,
+ "learning_rate": 1.9092800653896236e-05,
+ "loss": 0.6699,
+ "step": 5965
+ },
+ {
+ "epoch": 16.256130790190735,
+ "grad_norm": 6.577739238739014,
+ "learning_rate": 1.9092433337662614e-05,
+ "loss": 0.7014,
+ "step": 5966
+ },
+ {
+ "epoch": 16.258855585831064,
+ "grad_norm": 5.984896183013916,
+ "learning_rate": 1.909206595061741e-05,
+ "loss": 0.6426,
+ "step": 5967
+ },
+ {
+ "epoch": 16.26158038147139,
+ "grad_norm": 8.970513343811035,
+ "learning_rate": 1.9091698492763494e-05,
+ "loss": 0.9004,
+ "step": 5968
+ },
+ {
+ "epoch": 16.264305177111716,
+ "grad_norm": 6.716660499572754,
+ "learning_rate": 1.909133096410372e-05,
+ "loss": 0.7777,
+ "step": 5969
+ },
+ {
+ "epoch": 16.267029972752045,
+ "grad_norm": 6.018154621124268,
+ "learning_rate": 1.9090963364640954e-05,
+ "loss": 0.7413,
+ "step": 5970
+ },
+ {
+ "epoch": 16.26975476839237,
+ "grad_norm": 6.185939788818359,
+ "learning_rate": 1.909059569437806e-05,
+ "loss": 0.7689,
+ "step": 5971
+ },
+ {
+ "epoch": 16.272479564032697,
+ "grad_norm": 6.875397682189941,
+ "learning_rate": 1.9090227953317896e-05,
+ "loss": 0.8215,
+ "step": 5972
+ },
+ {
+ "epoch": 16.275204359673026,
+ "grad_norm": 7.186788082122803,
+ "learning_rate": 1.908986014146333e-05,
+ "loss": 0.74,
+ "step": 5973
+ },
+ {
+ "epoch": 16.277929155313352,
+ "grad_norm": 6.562900066375732,
+ "learning_rate": 1.9089492258817225e-05,
+ "loss": 0.7438,
+ "step": 5974
+ },
+ {
+ "epoch": 16.280653950953678,
+ "grad_norm": 6.587159633636475,
+ "learning_rate": 1.908912430538245e-05,
+ "loss": 0.719,
+ "step": 5975
+ },
+ {
+ "epoch": 16.283378746594007,
+ "grad_norm": 6.337550640106201,
+ "learning_rate": 1.9088756281161866e-05,
+ "loss": 0.6321,
+ "step": 5976
+ },
+ {
+ "epoch": 16.286103542234333,
+ "grad_norm": 6.916007041931152,
+ "learning_rate": 1.908838818615834e-05,
+ "loss": 0.7924,
+ "step": 5977
+ },
+ {
+ "epoch": 16.28882833787466,
+ "grad_norm": 7.297861576080322,
+ "learning_rate": 1.908802002037474e-05,
+ "loss": 0.7641,
+ "step": 5978
+ },
+ {
+ "epoch": 16.291553133514988,
+ "grad_norm": 6.680345058441162,
+ "learning_rate": 1.908765178381393e-05,
+ "loss": 0.6245,
+ "step": 5979
+ },
+ {
+ "epoch": 16.294277929155314,
+ "grad_norm": 9.59532356262207,
+ "learning_rate": 1.9087283476478784e-05,
+ "loss": 0.8411,
+ "step": 5980
+ },
+ {
+ "epoch": 16.29700272479564,
+ "grad_norm": 7.451826572418213,
+ "learning_rate": 1.9086915098372166e-05,
+ "loss": 0.8873,
+ "step": 5981
+ },
+ {
+ "epoch": 16.29972752043597,
+ "grad_norm": 6.1218438148498535,
+ "learning_rate": 1.9086546649496943e-05,
+ "loss": 0.6066,
+ "step": 5982
+ },
+ {
+ "epoch": 16.302452316076295,
+ "grad_norm": 8.233154296875,
+ "learning_rate": 1.908617812985599e-05,
+ "loss": 0.7994,
+ "step": 5983
+ },
+ {
+ "epoch": 16.30517711171662,
+ "grad_norm": 6.490713596343994,
+ "learning_rate": 1.9085809539452174e-05,
+ "loss": 0.7521,
+ "step": 5984
+ },
+ {
+ "epoch": 16.30790190735695,
+ "grad_norm": 7.527616500854492,
+ "learning_rate": 1.9085440878288363e-05,
+ "loss": 0.7051,
+ "step": 5985
+ },
+ {
+ "epoch": 16.310626702997276,
+ "grad_norm": 8.30439281463623,
+ "learning_rate": 1.9085072146367434e-05,
+ "loss": 0.8616,
+ "step": 5986
+ },
+ {
+ "epoch": 16.3133514986376,
+ "grad_norm": 5.695662498474121,
+ "learning_rate": 1.9084703343692253e-05,
+ "loss": 0.769,
+ "step": 5987
+ },
+ {
+ "epoch": 16.31607629427793,
+ "grad_norm": 6.327755928039551,
+ "learning_rate": 1.90843344702657e-05,
+ "loss": 0.6487,
+ "step": 5988
+ },
+ {
+ "epoch": 16.318801089918257,
+ "grad_norm": 7.536658763885498,
+ "learning_rate": 1.9083965526090635e-05,
+ "loss": 0.8772,
+ "step": 5989
+ },
+ {
+ "epoch": 16.321525885558582,
+ "grad_norm": 6.162464141845703,
+ "learning_rate": 1.908359651116994e-05,
+ "loss": 1.0596,
+ "step": 5990
+ },
+ {
+ "epoch": 16.32425068119891,
+ "grad_norm": 5.7072978019714355,
+ "learning_rate": 1.908322742550649e-05,
+ "loss": 0.6167,
+ "step": 5991
+ },
+ {
+ "epoch": 16.326975476839237,
+ "grad_norm": 7.876220226287842,
+ "learning_rate": 1.9082858269103162e-05,
+ "loss": 0.8759,
+ "step": 5992
+ },
+ {
+ "epoch": 16.329700272479563,
+ "grad_norm": 6.899759292602539,
+ "learning_rate": 1.908248904196282e-05,
+ "loss": 0.7682,
+ "step": 5993
+ },
+ {
+ "epoch": 16.332425068119893,
+ "grad_norm": 6.547876358032227,
+ "learning_rate": 1.9082119744088348e-05,
+ "loss": 0.8817,
+ "step": 5994
+ },
+ {
+ "epoch": 16.33514986376022,
+ "grad_norm": 6.491705417633057,
+ "learning_rate": 1.9081750375482618e-05,
+ "loss": 0.6852,
+ "step": 5995
+ },
+ {
+ "epoch": 16.337874659400544,
+ "grad_norm": 8.751527786254883,
+ "learning_rate": 1.908138093614851e-05,
+ "loss": 0.8324,
+ "step": 5996
+ },
+ {
+ "epoch": 16.340599455040874,
+ "grad_norm": 6.358096599578857,
+ "learning_rate": 1.90810114260889e-05,
+ "loss": 0.7438,
+ "step": 5997
+ },
+ {
+ "epoch": 16.3433242506812,
+ "grad_norm": 5.616379261016846,
+ "learning_rate": 1.908064184530666e-05,
+ "loss": 0.6867,
+ "step": 5998
+ },
+ {
+ "epoch": 16.346049046321525,
+ "grad_norm": 5.86985969543457,
+ "learning_rate": 1.908027219380468e-05,
+ "loss": 0.6598,
+ "step": 5999
+ },
+ {
+ "epoch": 16.348773841961854,
+ "grad_norm": 6.388766765594482,
+ "learning_rate": 1.907990247158583e-05,
+ "loss": 0.6981,
+ "step": 6000
+ },
+ {
+ "epoch": 16.35149863760218,
+ "grad_norm": 7.1630659103393555,
+ "learning_rate": 1.907953267865299e-05,
+ "loss": 0.7817,
+ "step": 6001
+ },
+ {
+ "epoch": 16.354223433242506,
+ "grad_norm": 7.11736536026001,
+ "learning_rate": 1.9079162815009045e-05,
+ "loss": 0.8551,
+ "step": 6002
+ },
+ {
+ "epoch": 16.356948228882835,
+ "grad_norm": 6.205382347106934,
+ "learning_rate": 1.9078792880656868e-05,
+ "loss": 0.6581,
+ "step": 6003
+ },
+ {
+ "epoch": 16.35967302452316,
+ "grad_norm": 5.568708419799805,
+ "learning_rate": 1.907842287559935e-05,
+ "loss": 0.7789,
+ "step": 6004
+ },
+ {
+ "epoch": 16.362397820163487,
+ "grad_norm": 6.953674793243408,
+ "learning_rate": 1.9078052799839363e-05,
+ "loss": 0.6958,
+ "step": 6005
+ },
+ {
+ "epoch": 16.365122615803816,
+ "grad_norm": 7.995293140411377,
+ "learning_rate": 1.9077682653379795e-05,
+ "loss": 0.7677,
+ "step": 6006
+ },
+ {
+ "epoch": 16.367847411444142,
+ "grad_norm": 7.480947494506836,
+ "learning_rate": 1.9077312436223524e-05,
+ "loss": 0.6947,
+ "step": 6007
+ },
+ {
+ "epoch": 16.370572207084468,
+ "grad_norm": 6.758172512054443,
+ "learning_rate": 1.907694214837344e-05,
+ "loss": 0.8152,
+ "step": 6008
+ },
+ {
+ "epoch": 16.373297002724797,
+ "grad_norm": 8.033045768737793,
+ "learning_rate": 1.9076571789832416e-05,
+ "loss": 0.8137,
+ "step": 6009
+ },
+ {
+ "epoch": 16.376021798365123,
+ "grad_norm": 7.9241533279418945,
+ "learning_rate": 1.9076201360603347e-05,
+ "loss": 0.717,
+ "step": 6010
+ },
+ {
+ "epoch": 16.37874659400545,
+ "grad_norm": 6.523370265960693,
+ "learning_rate": 1.9075830860689114e-05,
+ "loss": 0.8529,
+ "step": 6011
+ },
+ {
+ "epoch": 16.381471389645778,
+ "grad_norm": 6.41068696975708,
+ "learning_rate": 1.9075460290092604e-05,
+ "loss": 0.8228,
+ "step": 6012
+ },
+ {
+ "epoch": 16.384196185286104,
+ "grad_norm": 7.027105808258057,
+ "learning_rate": 1.9075089648816697e-05,
+ "loss": 0.9005,
+ "step": 6013
+ },
+ {
+ "epoch": 16.38692098092643,
+ "grad_norm": 6.447543621063232,
+ "learning_rate": 1.907471893686429e-05,
+ "loss": 0.6479,
+ "step": 6014
+ },
+ {
+ "epoch": 16.38964577656676,
+ "grad_norm": 6.032678127288818,
+ "learning_rate": 1.907434815423826e-05,
+ "loss": 0.7245,
+ "step": 6015
+ },
+ {
+ "epoch": 16.392370572207085,
+ "grad_norm": 6.0707688331604,
+ "learning_rate": 1.9073977300941497e-05,
+ "loss": 0.6539,
+ "step": 6016
+ },
+ {
+ "epoch": 16.39509536784741,
+ "grad_norm": 6.174981117248535,
+ "learning_rate": 1.9073606376976895e-05,
+ "loss": 0.8652,
+ "step": 6017
+ },
+ {
+ "epoch": 16.39782016348774,
+ "grad_norm": 7.820618152618408,
+ "learning_rate": 1.9073235382347336e-05,
+ "loss": 0.8679,
+ "step": 6018
+ },
+ {
+ "epoch": 16.400544959128066,
+ "grad_norm": 6.377321243286133,
+ "learning_rate": 1.907286431705571e-05,
+ "loss": 0.6606,
+ "step": 6019
+ },
+ {
+ "epoch": 16.40326975476839,
+ "grad_norm": 5.8169379234313965,
+ "learning_rate": 1.9072493181104916e-05,
+ "loss": 0.757,
+ "step": 6020
+ },
+ {
+ "epoch": 16.40599455040872,
+ "grad_norm": 6.786158084869385,
+ "learning_rate": 1.907212197449783e-05,
+ "loss": 0.745,
+ "step": 6021
+ },
+ {
+ "epoch": 16.408719346049047,
+ "grad_norm": 8.320942878723145,
+ "learning_rate": 1.907175069723735e-05,
+ "loss": 0.795,
+ "step": 6022
+ },
+ {
+ "epoch": 16.411444141689373,
+ "grad_norm": 5.9972405433654785,
+ "learning_rate": 1.9071379349326373e-05,
+ "loss": 0.8138,
+ "step": 6023
+ },
+ {
+ "epoch": 16.414168937329702,
+ "grad_norm": 6.2193684577941895,
+ "learning_rate": 1.9071007930767782e-05,
+ "loss": 0.7446,
+ "step": 6024
+ },
+ {
+ "epoch": 16.416893732970028,
+ "grad_norm": 8.4401273727417,
+ "learning_rate": 1.9070636441564474e-05,
+ "loss": 1.1315,
+ "step": 6025
+ },
+ {
+ "epoch": 16.419618528610354,
+ "grad_norm": 6.036195755004883,
+ "learning_rate": 1.907026488171934e-05,
+ "loss": 0.6182,
+ "step": 6026
+ },
+ {
+ "epoch": 16.422343324250683,
+ "grad_norm": 6.588200569152832,
+ "learning_rate": 1.9069893251235277e-05,
+ "loss": 0.7491,
+ "step": 6027
+ },
+ {
+ "epoch": 16.42506811989101,
+ "grad_norm": 7.059725284576416,
+ "learning_rate": 1.9069521550115175e-05,
+ "loss": 0.8099,
+ "step": 6028
+ },
+ {
+ "epoch": 16.427792915531334,
+ "grad_norm": 7.952618598937988,
+ "learning_rate": 1.9069149778361934e-05,
+ "loss": 0.7188,
+ "step": 6029
+ },
+ {
+ "epoch": 16.430517711171664,
+ "grad_norm": 5.816744804382324,
+ "learning_rate": 1.9068777935978444e-05,
+ "loss": 0.7584,
+ "step": 6030
+ },
+ {
+ "epoch": 16.43324250681199,
+ "grad_norm": 6.990025997161865,
+ "learning_rate": 1.9068406022967605e-05,
+ "loss": 0.8468,
+ "step": 6031
+ },
+ {
+ "epoch": 16.435967302452315,
+ "grad_norm": 7.275354385375977,
+ "learning_rate": 1.906803403933231e-05,
+ "loss": 1.0437,
+ "step": 6032
+ },
+ {
+ "epoch": 16.438692098092645,
+ "grad_norm": 7.226301670074463,
+ "learning_rate": 1.906766198507546e-05,
+ "loss": 0.8419,
+ "step": 6033
+ },
+ {
+ "epoch": 16.44141689373297,
+ "grad_norm": 6.441589832305908,
+ "learning_rate": 1.9067289860199945e-05,
+ "loss": 0.8918,
+ "step": 6034
+ },
+ {
+ "epoch": 16.444141689373296,
+ "grad_norm": 6.7550554275512695,
+ "learning_rate": 1.9066917664708674e-05,
+ "loss": 0.7467,
+ "step": 6035
+ },
+ {
+ "epoch": 16.446866485013626,
+ "grad_norm": 6.511507034301758,
+ "learning_rate": 1.9066545398604538e-05,
+ "loss": 0.7428,
+ "step": 6036
+ },
+ {
+ "epoch": 16.44959128065395,
+ "grad_norm": 5.9525909423828125,
+ "learning_rate": 1.9066173061890438e-05,
+ "loss": 0.7898,
+ "step": 6037
+ },
+ {
+ "epoch": 16.452316076294277,
+ "grad_norm": 12.588763236999512,
+ "learning_rate": 1.9065800654569277e-05,
+ "loss": 0.6769,
+ "step": 6038
+ },
+ {
+ "epoch": 16.455040871934607,
+ "grad_norm": 6.5953688621521,
+ "learning_rate": 1.906542817664395e-05,
+ "loss": 0.6399,
+ "step": 6039
+ },
+ {
+ "epoch": 16.457765667574932,
+ "grad_norm": 7.763669490814209,
+ "learning_rate": 1.9065055628117358e-05,
+ "loss": 0.7659,
+ "step": 6040
+ },
+ {
+ "epoch": 16.460490463215258,
+ "grad_norm": 6.939055442810059,
+ "learning_rate": 1.9064683008992407e-05,
+ "loss": 0.6677,
+ "step": 6041
+ },
+ {
+ "epoch": 16.463215258855588,
+ "grad_norm": 7.445368766784668,
+ "learning_rate": 1.9064310319271997e-05,
+ "loss": 0.8044,
+ "step": 6042
+ },
+ {
+ "epoch": 16.465940054495913,
+ "grad_norm": 6.867668628692627,
+ "learning_rate": 1.906393755895903e-05,
+ "loss": 0.7483,
+ "step": 6043
+ },
+ {
+ "epoch": 16.46866485013624,
+ "grad_norm": 6.8181867599487305,
+ "learning_rate": 1.906356472805641e-05,
+ "loss": 0.9464,
+ "step": 6044
+ },
+ {
+ "epoch": 16.47138964577657,
+ "grad_norm": 7.57024621963501,
+ "learning_rate": 1.9063191826567035e-05,
+ "loss": 0.6796,
+ "step": 6045
+ },
+ {
+ "epoch": 16.474114441416894,
+ "grad_norm": 7.079565525054932,
+ "learning_rate": 1.9062818854493816e-05,
+ "loss": 0.7321,
+ "step": 6046
+ },
+ {
+ "epoch": 16.47683923705722,
+ "grad_norm": 7.091064929962158,
+ "learning_rate": 1.9062445811839655e-05,
+ "loss": 0.7213,
+ "step": 6047
+ },
+ {
+ "epoch": 16.479564032697546,
+ "grad_norm": 7.10012674331665,
+ "learning_rate": 1.9062072698607457e-05,
+ "loss": 0.7441,
+ "step": 6048
+ },
+ {
+ "epoch": 16.482288828337875,
+ "grad_norm": 5.7275543212890625,
+ "learning_rate": 1.9061699514800128e-05,
+ "loss": 0.6647,
+ "step": 6049
+ },
+ {
+ "epoch": 16.4850136239782,
+ "grad_norm": 7.66431188583374,
+ "learning_rate": 1.906132626042058e-05,
+ "loss": 0.7612,
+ "step": 6050
+ },
+ {
+ "epoch": 16.48773841961853,
+ "grad_norm": 7.276266098022461,
+ "learning_rate": 1.906095293547171e-05,
+ "loss": 0.6694,
+ "step": 6051
+ },
+ {
+ "epoch": 16.490463215258856,
+ "grad_norm": 7.760282516479492,
+ "learning_rate": 1.906057953995643e-05,
+ "loss": 0.7504,
+ "step": 6052
+ },
+ {
+ "epoch": 16.493188010899182,
+ "grad_norm": 7.155226230621338,
+ "learning_rate": 1.9060206073877655e-05,
+ "loss": 0.8178,
+ "step": 6053
+ },
+ {
+ "epoch": 16.495912806539508,
+ "grad_norm": 7.091317653656006,
+ "learning_rate": 1.9059832537238278e-05,
+ "loss": 0.7032,
+ "step": 6054
+ },
+ {
+ "epoch": 16.498637602179837,
+ "grad_norm": 6.443547248840332,
+ "learning_rate": 1.9059458930041222e-05,
+ "loss": 0.7234,
+ "step": 6055
+ },
+ {
+ "epoch": 16.501362397820163,
+ "grad_norm": 7.410799980163574,
+ "learning_rate": 1.9059085252289388e-05,
+ "loss": 0.652,
+ "step": 6056
+ },
+ {
+ "epoch": 16.504087193460492,
+ "grad_norm": 6.442809581756592,
+ "learning_rate": 1.9058711503985692e-05,
+ "loss": 0.7501,
+ "step": 6057
+ },
+ {
+ "epoch": 16.506811989100818,
+ "grad_norm": 6.18709135055542,
+ "learning_rate": 1.9058337685133043e-05,
+ "loss": 0.7489,
+ "step": 6058
+ },
+ {
+ "epoch": 16.509536784741144,
+ "grad_norm": 7.789896488189697,
+ "learning_rate": 1.905796379573435e-05,
+ "loss": 1.124,
+ "step": 6059
+ },
+ {
+ "epoch": 16.51226158038147,
+ "grad_norm": 7.391740322113037,
+ "learning_rate": 1.9057589835792524e-05,
+ "loss": 0.9508,
+ "step": 6060
+ },
+ {
+ "epoch": 16.5149863760218,
+ "grad_norm": 5.804466724395752,
+ "learning_rate": 1.9057215805310485e-05,
+ "loss": 0.7594,
+ "step": 6061
+ },
+ {
+ "epoch": 16.517711171662125,
+ "grad_norm": 7.261319637298584,
+ "learning_rate": 1.9056841704291137e-05,
+ "loss": 0.8425,
+ "step": 6062
+ },
+ {
+ "epoch": 16.520435967302454,
+ "grad_norm": 5.971974849700928,
+ "learning_rate": 1.90564675327374e-05,
+ "loss": 0.7501,
+ "step": 6063
+ },
+ {
+ "epoch": 16.52316076294278,
+ "grad_norm": 7.483309745788574,
+ "learning_rate": 1.9056093290652186e-05,
+ "loss": 0.6384,
+ "step": 6064
+ },
+ {
+ "epoch": 16.525885558583106,
+ "grad_norm": 6.24228048324585,
+ "learning_rate": 1.9055718978038404e-05,
+ "loss": 0.7867,
+ "step": 6065
+ },
+ {
+ "epoch": 16.52861035422343,
+ "grad_norm": 6.908617973327637,
+ "learning_rate": 1.9055344594898978e-05,
+ "loss": 0.7889,
+ "step": 6066
+ },
+ {
+ "epoch": 16.53133514986376,
+ "grad_norm": 8.063348770141602,
+ "learning_rate": 1.905497014123682e-05,
+ "loss": 0.7058,
+ "step": 6067
+ },
+ {
+ "epoch": 16.534059945504087,
+ "grad_norm": 6.614887237548828,
+ "learning_rate": 1.9054595617054844e-05,
+ "loss": 0.8152,
+ "step": 6068
+ },
+ {
+ "epoch": 16.536784741144416,
+ "grad_norm": 7.029272556304932,
+ "learning_rate": 1.905422102235597e-05,
+ "loss": 0.8845,
+ "step": 6069
+ },
+ {
+ "epoch": 16.539509536784742,
+ "grad_norm": 5.82067346572876,
+ "learning_rate": 1.9053846357143114e-05,
+ "loss": 0.7056,
+ "step": 6070
+ },
+ {
+ "epoch": 16.542234332425068,
+ "grad_norm": 7.62905216217041,
+ "learning_rate": 1.9053471621419193e-05,
+ "loss": 0.9445,
+ "step": 6071
+ },
+ {
+ "epoch": 16.544959128065393,
+ "grad_norm": 7.2660088539123535,
+ "learning_rate": 1.9053096815187125e-05,
+ "loss": 0.9979,
+ "step": 6072
+ },
+ {
+ "epoch": 16.547683923705723,
+ "grad_norm": 12.6041841506958,
+ "learning_rate": 1.9052721938449834e-05,
+ "loss": 0.9946,
+ "step": 6073
+ },
+ {
+ "epoch": 16.55040871934605,
+ "grad_norm": 6.338008880615234,
+ "learning_rate": 1.9052346991210237e-05,
+ "loss": 0.7008,
+ "step": 6074
+ },
+ {
+ "epoch": 16.553133514986374,
+ "grad_norm": 6.75604248046875,
+ "learning_rate": 1.905197197347125e-05,
+ "loss": 0.8501,
+ "step": 6075
+ },
+ {
+ "epoch": 16.555858310626704,
+ "grad_norm": 12.002813339233398,
+ "learning_rate": 1.9051596885235795e-05,
+ "loss": 0.8337,
+ "step": 6076
+ },
+ {
+ "epoch": 16.55858310626703,
+ "grad_norm": 6.3340229988098145,
+ "learning_rate": 1.9051221726506795e-05,
+ "loss": 0.8169,
+ "step": 6077
+ },
+ {
+ "epoch": 16.561307901907355,
+ "grad_norm": 6.599599361419678,
+ "learning_rate": 1.9050846497287175e-05,
+ "loss": 0.7798,
+ "step": 6078
+ },
+ {
+ "epoch": 16.564032697547685,
+ "grad_norm": 7.155139923095703,
+ "learning_rate": 1.905047119757985e-05,
+ "loss": 0.9264,
+ "step": 6079
+ },
+ {
+ "epoch": 16.56675749318801,
+ "grad_norm": 7.238029956817627,
+ "learning_rate": 1.9050095827387748e-05,
+ "loss": 0.9443,
+ "step": 6080
+ },
+ {
+ "epoch": 16.569482288828336,
+ "grad_norm": 7.0447916984558105,
+ "learning_rate": 1.9049720386713796e-05,
+ "loss": 0.7963,
+ "step": 6081
+ },
+ {
+ "epoch": 16.572207084468666,
+ "grad_norm": 9.390327453613281,
+ "learning_rate": 1.9049344875560906e-05,
+ "loss": 0.6975,
+ "step": 6082
+ },
+ {
+ "epoch": 16.57493188010899,
+ "grad_norm": 6.9206719398498535,
+ "learning_rate": 1.9048969293932012e-05,
+ "loss": 1.0018,
+ "step": 6083
+ },
+ {
+ "epoch": 16.577656675749317,
+ "grad_norm": 5.9548211097717285,
+ "learning_rate": 1.9048593641830034e-05,
+ "loss": 0.717,
+ "step": 6084
+ },
+ {
+ "epoch": 16.580381471389646,
+ "grad_norm": 7.115381240844727,
+ "learning_rate": 1.9048217919257907e-05,
+ "loss": 1.1006,
+ "step": 6085
+ },
+ {
+ "epoch": 16.583106267029972,
+ "grad_norm": 6.8633928298950195,
+ "learning_rate": 1.9047842126218544e-05,
+ "loss": 0.687,
+ "step": 6086
+ },
+ {
+ "epoch": 16.585831062670298,
+ "grad_norm": 7.407203197479248,
+ "learning_rate": 1.904746626271488e-05,
+ "loss": 0.6962,
+ "step": 6087
+ },
+ {
+ "epoch": 16.588555858310627,
+ "grad_norm": 6.302432537078857,
+ "learning_rate": 1.9047090328749837e-05,
+ "loss": 0.6949,
+ "step": 6088
+ },
+ {
+ "epoch": 16.591280653950953,
+ "grad_norm": 7.083056926727295,
+ "learning_rate": 1.9046714324326347e-05,
+ "loss": 0.8442,
+ "step": 6089
+ },
+ {
+ "epoch": 16.59400544959128,
+ "grad_norm": 6.815189361572266,
+ "learning_rate": 1.9046338249447335e-05,
+ "loss": 0.8414,
+ "step": 6090
+ },
+ {
+ "epoch": 16.59673024523161,
+ "grad_norm": 7.5467753410339355,
+ "learning_rate": 1.9045962104115738e-05,
+ "loss": 0.808,
+ "step": 6091
+ },
+ {
+ "epoch": 16.599455040871934,
+ "grad_norm": 7.471316814422607,
+ "learning_rate": 1.9045585888334475e-05,
+ "loss": 0.7245,
+ "step": 6092
+ },
+ {
+ "epoch": 16.60217983651226,
+ "grad_norm": 8.062198638916016,
+ "learning_rate": 1.904520960210648e-05,
+ "loss": 0.7976,
+ "step": 6093
+ },
+ {
+ "epoch": 16.60490463215259,
+ "grad_norm": 25.604543685913086,
+ "learning_rate": 1.9044833245434686e-05,
+ "loss": 0.7249,
+ "step": 6094
+ },
+ {
+ "epoch": 16.607629427792915,
+ "grad_norm": 7.371628284454346,
+ "learning_rate": 1.904445681832202e-05,
+ "loss": 0.9144,
+ "step": 6095
+ },
+ {
+ "epoch": 16.61035422343324,
+ "grad_norm": 6.009514331817627,
+ "learning_rate": 1.9044080320771416e-05,
+ "loss": 0.863,
+ "step": 6096
+ },
+ {
+ "epoch": 16.61307901907357,
+ "grad_norm": 5.105881214141846,
+ "learning_rate": 1.9043703752785804e-05,
+ "loss": 0.7183,
+ "step": 6097
+ },
+ {
+ "epoch": 16.615803814713896,
+ "grad_norm": 5.2238264083862305,
+ "learning_rate": 1.9043327114368118e-05,
+ "loss": 0.6768,
+ "step": 6098
+ },
+ {
+ "epoch": 16.618528610354222,
+ "grad_norm": 6.056881904602051,
+ "learning_rate": 1.9042950405521296e-05,
+ "loss": 0.79,
+ "step": 6099
+ },
+ {
+ "epoch": 16.62125340599455,
+ "grad_norm": 6.8914594650268555,
+ "learning_rate": 1.9042573626248266e-05,
+ "loss": 0.8563,
+ "step": 6100
+ },
+ {
+ "epoch": 16.623978201634877,
+ "grad_norm": 6.7143731117248535,
+ "learning_rate": 1.904219677655196e-05,
+ "loss": 0.7421,
+ "step": 6101
+ },
+ {
+ "epoch": 16.626702997275203,
+ "grad_norm": 5.886285305023193,
+ "learning_rate": 1.9041819856435318e-05,
+ "loss": 0.9108,
+ "step": 6102
+ },
+ {
+ "epoch": 16.629427792915532,
+ "grad_norm": 6.537633895874023,
+ "learning_rate": 1.9041442865901276e-05,
+ "loss": 0.8192,
+ "step": 6103
+ },
+ {
+ "epoch": 16.632152588555858,
+ "grad_norm": 8.000080108642578,
+ "learning_rate": 1.9041065804952764e-05,
+ "loss": 0.853,
+ "step": 6104
+ },
+ {
+ "epoch": 16.634877384196184,
+ "grad_norm": 7.60627555847168,
+ "learning_rate": 1.9040688673592725e-05,
+ "loss": 0.8622,
+ "step": 6105
+ },
+ {
+ "epoch": 16.637602179836513,
+ "grad_norm": 5.841114044189453,
+ "learning_rate": 1.9040311471824092e-05,
+ "loss": 0.8433,
+ "step": 6106
+ },
+ {
+ "epoch": 16.64032697547684,
+ "grad_norm": 6.566752910614014,
+ "learning_rate": 1.903993419964981e-05,
+ "loss": 0.8085,
+ "step": 6107
+ },
+ {
+ "epoch": 16.643051771117165,
+ "grad_norm": 6.407968044281006,
+ "learning_rate": 1.9039556857072808e-05,
+ "loss": 0.7419,
+ "step": 6108
+ },
+ {
+ "epoch": 16.645776566757494,
+ "grad_norm": 6.520883083343506,
+ "learning_rate": 1.9039179444096025e-05,
+ "loss": 1.0001,
+ "step": 6109
+ },
+ {
+ "epoch": 16.64850136239782,
+ "grad_norm": 6.325121879577637,
+ "learning_rate": 1.9038801960722405e-05,
+ "loss": 0.8993,
+ "step": 6110
+ },
+ {
+ "epoch": 16.651226158038146,
+ "grad_norm": 6.965338706970215,
+ "learning_rate": 1.903842440695489e-05,
+ "loss": 0.7339,
+ "step": 6111
+ },
+ {
+ "epoch": 16.653950953678475,
+ "grad_norm": 7.198559761047363,
+ "learning_rate": 1.903804678279641e-05,
+ "loss": 0.8905,
+ "step": 6112
+ },
+ {
+ "epoch": 16.6566757493188,
+ "grad_norm": 5.85977840423584,
+ "learning_rate": 1.903766908824992e-05,
+ "loss": 0.7698,
+ "step": 6113
+ },
+ {
+ "epoch": 16.659400544959126,
+ "grad_norm": 5.2927985191345215,
+ "learning_rate": 1.9037291323318348e-05,
+ "loss": 0.7847,
+ "step": 6114
+ },
+ {
+ "epoch": 16.662125340599456,
+ "grad_norm": 7.242940902709961,
+ "learning_rate": 1.9036913488004646e-05,
+ "loss": 0.6659,
+ "step": 6115
+ },
+ {
+ "epoch": 16.66485013623978,
+ "grad_norm": 5.561131000518799,
+ "learning_rate": 1.903653558231175e-05,
+ "loss": 0.8729,
+ "step": 6116
+ },
+ {
+ "epoch": 16.667574931880107,
+ "grad_norm": 7.044807434082031,
+ "learning_rate": 1.9036157606242603e-05,
+ "loss": 0.7496,
+ "step": 6117
+ },
+ {
+ "epoch": 16.670299727520437,
+ "grad_norm": 6.627017974853516,
+ "learning_rate": 1.903577955980016e-05,
+ "loss": 0.8147,
+ "step": 6118
+ },
+ {
+ "epoch": 16.673024523160763,
+ "grad_norm": 9.127052307128906,
+ "learning_rate": 1.903540144298735e-05,
+ "loss": 0.876,
+ "step": 6119
+ },
+ {
+ "epoch": 16.67574931880109,
+ "grad_norm": 7.582970142364502,
+ "learning_rate": 1.9035023255807124e-05,
+ "loss": 0.7174,
+ "step": 6120
+ },
+ {
+ "epoch": 16.678474114441418,
+ "grad_norm": 6.642630577087402,
+ "learning_rate": 1.903464499826243e-05,
+ "loss": 0.7786,
+ "step": 6121
+ },
+ {
+ "epoch": 16.681198910081743,
+ "grad_norm": 5.713918685913086,
+ "learning_rate": 1.903426667035621e-05,
+ "loss": 0.6425,
+ "step": 6122
+ },
+ {
+ "epoch": 16.68392370572207,
+ "grad_norm": 6.467620849609375,
+ "learning_rate": 1.9033888272091416e-05,
+ "loss": 0.7826,
+ "step": 6123
+ },
+ {
+ "epoch": 16.6866485013624,
+ "grad_norm": 7.108287811279297,
+ "learning_rate": 1.9033509803470986e-05,
+ "loss": 1.033,
+ "step": 6124
+ },
+ {
+ "epoch": 16.689373297002724,
+ "grad_norm": 6.360099792480469,
+ "learning_rate": 1.9033131264497876e-05,
+ "loss": 0.8988,
+ "step": 6125
+ },
+ {
+ "epoch": 16.69209809264305,
+ "grad_norm": 11.044900894165039,
+ "learning_rate": 1.9032752655175025e-05,
+ "loss": 0.8567,
+ "step": 6126
+ },
+ {
+ "epoch": 16.69482288828338,
+ "grad_norm": 6.069056034088135,
+ "learning_rate": 1.9032373975505393e-05,
+ "loss": 0.8324,
+ "step": 6127
+ },
+ {
+ "epoch": 16.697547683923705,
+ "grad_norm": 6.552725315093994,
+ "learning_rate": 1.903199522549192e-05,
+ "loss": 0.7156,
+ "step": 6128
+ },
+ {
+ "epoch": 16.70027247956403,
+ "grad_norm": 7.365549087524414,
+ "learning_rate": 1.9031616405137563e-05,
+ "loss": 0.8718,
+ "step": 6129
+ },
+ {
+ "epoch": 16.70299727520436,
+ "grad_norm": 7.178698539733887,
+ "learning_rate": 1.9031237514445264e-05,
+ "loss": 0.7133,
+ "step": 6130
+ },
+ {
+ "epoch": 16.705722070844686,
+ "grad_norm": 6.295802593231201,
+ "learning_rate": 1.9030858553417978e-05,
+ "loss": 0.7004,
+ "step": 6131
+ },
+ {
+ "epoch": 16.708446866485012,
+ "grad_norm": 7.233146667480469,
+ "learning_rate": 1.9030479522058656e-05,
+ "loss": 0.88,
+ "step": 6132
+ },
+ {
+ "epoch": 16.71117166212534,
+ "grad_norm": 6.640514850616455,
+ "learning_rate": 1.9030100420370254e-05,
+ "loss": 0.6899,
+ "step": 6133
+ },
+ {
+ "epoch": 16.713896457765667,
+ "grad_norm": 6.011962890625,
+ "learning_rate": 1.9029721248355718e-05,
+ "loss": 0.7134,
+ "step": 6134
+ },
+ {
+ "epoch": 16.716621253405993,
+ "grad_norm": 7.030774116516113,
+ "learning_rate": 1.9029342006018002e-05,
+ "loss": 0.7238,
+ "step": 6135
+ },
+ {
+ "epoch": 16.719346049046322,
+ "grad_norm": 8.276910781860352,
+ "learning_rate": 1.9028962693360062e-05,
+ "loss": 0.8547,
+ "step": 6136
+ },
+ {
+ "epoch": 16.722070844686648,
+ "grad_norm": 6.005258083343506,
+ "learning_rate": 1.9028583310384852e-05,
+ "loss": 0.6494,
+ "step": 6137
+ },
+ {
+ "epoch": 16.724795640326974,
+ "grad_norm": 7.169383525848389,
+ "learning_rate": 1.9028203857095326e-05,
+ "loss": 0.9922,
+ "step": 6138
+ },
+ {
+ "epoch": 16.727520435967303,
+ "grad_norm": 6.39536190032959,
+ "learning_rate": 1.902782433349444e-05,
+ "loss": 0.7281,
+ "step": 6139
+ },
+ {
+ "epoch": 16.73024523160763,
+ "grad_norm": 7.415670871734619,
+ "learning_rate": 1.9027444739585145e-05,
+ "loss": 0.6993,
+ "step": 6140
+ },
+ {
+ "epoch": 16.732970027247955,
+ "grad_norm": 6.376102447509766,
+ "learning_rate": 1.9027065075370405e-05,
+ "loss": 0.6367,
+ "step": 6141
+ },
+ {
+ "epoch": 16.735694822888284,
+ "grad_norm": 7.423565864562988,
+ "learning_rate": 1.9026685340853167e-05,
+ "loss": 0.6837,
+ "step": 6142
+ },
+ {
+ "epoch": 16.73841961852861,
+ "grad_norm": 6.810519218444824,
+ "learning_rate": 1.90263055360364e-05,
+ "loss": 0.8342,
+ "step": 6143
+ },
+ {
+ "epoch": 16.741144414168936,
+ "grad_norm": 6.527738094329834,
+ "learning_rate": 1.9025925660923053e-05,
+ "loss": 0.8506,
+ "step": 6144
+ },
+ {
+ "epoch": 16.743869209809265,
+ "grad_norm": 6.673910617828369,
+ "learning_rate": 1.902554571551609e-05,
+ "loss": 0.6898,
+ "step": 6145
+ },
+ {
+ "epoch": 16.74659400544959,
+ "grad_norm": 6.677182197570801,
+ "learning_rate": 1.902516569981846e-05,
+ "loss": 0.7877,
+ "step": 6146
+ },
+ {
+ "epoch": 16.749318801089917,
+ "grad_norm": 6.553023338317871,
+ "learning_rate": 1.902478561383314e-05,
+ "loss": 0.7085,
+ "step": 6147
+ },
+ {
+ "epoch": 16.752043596730246,
+ "grad_norm": 6.94491720199585,
+ "learning_rate": 1.9024405457563072e-05,
+ "loss": 0.7949,
+ "step": 6148
+ },
+ {
+ "epoch": 16.754768392370572,
+ "grad_norm": 8.255630493164062,
+ "learning_rate": 1.902402523101123e-05,
+ "loss": 1.0306,
+ "step": 6149
+ },
+ {
+ "epoch": 16.757493188010898,
+ "grad_norm": 5.6130523681640625,
+ "learning_rate": 1.9023644934180568e-05,
+ "loss": 0.7332,
+ "step": 6150
+ },
+ {
+ "epoch": 16.760217983651227,
+ "grad_norm": 7.453123092651367,
+ "learning_rate": 1.902326456707405e-05,
+ "loss": 0.7909,
+ "step": 6151
+ },
+ {
+ "epoch": 16.762942779291553,
+ "grad_norm": 6.357696533203125,
+ "learning_rate": 1.902288412969464e-05,
+ "loss": 0.6957,
+ "step": 6152
+ },
+ {
+ "epoch": 16.76566757493188,
+ "grad_norm": 6.241941928863525,
+ "learning_rate": 1.9022503622045297e-05,
+ "loss": 0.8075,
+ "step": 6153
+ },
+ {
+ "epoch": 16.768392370572208,
+ "grad_norm": 6.278064250946045,
+ "learning_rate": 1.9022123044128985e-05,
+ "loss": 0.8077,
+ "step": 6154
+ },
+ {
+ "epoch": 16.771117166212534,
+ "grad_norm": 6.858665943145752,
+ "learning_rate": 1.9021742395948668e-05,
+ "loss": 0.787,
+ "step": 6155
+ },
+ {
+ "epoch": 16.77384196185286,
+ "grad_norm": 7.443262577056885,
+ "learning_rate": 1.9021361677507316e-05,
+ "loss": 0.9387,
+ "step": 6156
+ },
+ {
+ "epoch": 16.77656675749319,
+ "grad_norm": 6.349826335906982,
+ "learning_rate": 1.902098088880789e-05,
+ "loss": 0.6034,
+ "step": 6157
+ },
+ {
+ "epoch": 16.779291553133515,
+ "grad_norm": 6.898726940155029,
+ "learning_rate": 1.902060002985335e-05,
+ "loss": 0.7655,
+ "step": 6158
+ },
+ {
+ "epoch": 16.78201634877384,
+ "grad_norm": 6.217432022094727,
+ "learning_rate": 1.9020219100646667e-05,
+ "loss": 0.7142,
+ "step": 6159
+ },
+ {
+ "epoch": 16.78474114441417,
+ "grad_norm": 7.797333240509033,
+ "learning_rate": 1.9019838101190812e-05,
+ "loss": 1.0063,
+ "step": 6160
+ },
+ {
+ "epoch": 16.787465940054496,
+ "grad_norm": 6.136629581451416,
+ "learning_rate": 1.9019457031488747e-05,
+ "loss": 0.8528,
+ "step": 6161
+ },
+ {
+ "epoch": 16.79019073569482,
+ "grad_norm": 6.793035984039307,
+ "learning_rate": 1.901907589154344e-05,
+ "loss": 0.5966,
+ "step": 6162
+ },
+ {
+ "epoch": 16.79291553133515,
+ "grad_norm": 6.63559627532959,
+ "learning_rate": 1.9018694681357865e-05,
+ "loss": 0.9628,
+ "step": 6163
+ },
+ {
+ "epoch": 16.795640326975477,
+ "grad_norm": 6.415094375610352,
+ "learning_rate": 1.901831340093498e-05,
+ "loss": 0.7163,
+ "step": 6164
+ },
+ {
+ "epoch": 16.798365122615802,
+ "grad_norm": 5.825899124145508,
+ "learning_rate": 1.9017932050277762e-05,
+ "loss": 0.7753,
+ "step": 6165
+ },
+ {
+ "epoch": 16.80108991825613,
+ "grad_norm": 6.3780598640441895,
+ "learning_rate": 1.901755062938918e-05,
+ "loss": 0.7355,
+ "step": 6166
+ },
+ {
+ "epoch": 16.803814713896458,
+ "grad_norm": 5.1487274169921875,
+ "learning_rate": 1.9017169138272203e-05,
+ "loss": 0.7248,
+ "step": 6167
+ },
+ {
+ "epoch": 16.806539509536783,
+ "grad_norm": 6.605749607086182,
+ "learning_rate": 1.9016787576929805e-05,
+ "loss": 0.7603,
+ "step": 6168
+ },
+ {
+ "epoch": 16.809264305177113,
+ "grad_norm": 15.651237487792969,
+ "learning_rate": 1.901640594536495e-05,
+ "loss": 0.7714,
+ "step": 6169
+ },
+ {
+ "epoch": 16.81198910081744,
+ "grad_norm": 8.62767505645752,
+ "learning_rate": 1.9016024243580626e-05,
+ "loss": 0.7141,
+ "step": 6170
+ },
+ {
+ "epoch": 16.814713896457764,
+ "grad_norm": 6.662627696990967,
+ "learning_rate": 1.9015642471579786e-05,
+ "loss": 0.8783,
+ "step": 6171
+ },
+ {
+ "epoch": 16.817438692098094,
+ "grad_norm": 6.703714847564697,
+ "learning_rate": 1.9015260629365417e-05,
+ "loss": 0.8884,
+ "step": 6172
+ },
+ {
+ "epoch": 16.82016348773842,
+ "grad_norm": 6.115324020385742,
+ "learning_rate": 1.9014878716940487e-05,
+ "loss": 0.7794,
+ "step": 6173
+ },
+ {
+ "epoch": 16.822888283378745,
+ "grad_norm": 6.070549964904785,
+ "learning_rate": 1.9014496734307974e-05,
+ "loss": 0.6642,
+ "step": 6174
+ },
+ {
+ "epoch": 16.825613079019075,
+ "grad_norm": 8.101480484008789,
+ "learning_rate": 1.9014114681470847e-05,
+ "loss": 0.9772,
+ "step": 6175
+ },
+ {
+ "epoch": 16.8283378746594,
+ "grad_norm": 8.611977577209473,
+ "learning_rate": 1.9013732558432086e-05,
+ "loss": 0.887,
+ "step": 6176
+ },
+ {
+ "epoch": 16.831062670299726,
+ "grad_norm": 6.284693717956543,
+ "learning_rate": 1.901335036519467e-05,
+ "loss": 0.743,
+ "step": 6177
+ },
+ {
+ "epoch": 16.833787465940055,
+ "grad_norm": 6.176537036895752,
+ "learning_rate": 1.9012968101761567e-05,
+ "loss": 0.7262,
+ "step": 6178
+ },
+ {
+ "epoch": 16.83651226158038,
+ "grad_norm": 8.002274513244629,
+ "learning_rate": 1.901258576813576e-05,
+ "loss": 0.9172,
+ "step": 6179
+ },
+ {
+ "epoch": 16.839237057220707,
+ "grad_norm": 7.610520839691162,
+ "learning_rate": 1.9012203364320223e-05,
+ "loss": 0.7518,
+ "step": 6180
+ },
+ {
+ "epoch": 16.841961852861036,
+ "grad_norm": 6.118548393249512,
+ "learning_rate": 1.901182089031794e-05,
+ "loss": 0.7421,
+ "step": 6181
+ },
+ {
+ "epoch": 16.844686648501362,
+ "grad_norm": 6.358120918273926,
+ "learning_rate": 1.9011438346131882e-05,
+ "loss": 0.7026,
+ "step": 6182
+ },
+ {
+ "epoch": 16.847411444141688,
+ "grad_norm": 6.791859149932861,
+ "learning_rate": 1.9011055731765036e-05,
+ "loss": 0.6853,
+ "step": 6183
+ },
+ {
+ "epoch": 16.850136239782017,
+ "grad_norm": 7.1138434410095215,
+ "learning_rate": 1.9010673047220373e-05,
+ "loss": 0.9576,
+ "step": 6184
+ },
+ {
+ "epoch": 16.852861035422343,
+ "grad_norm": 8.003449440002441,
+ "learning_rate": 1.9010290292500883e-05,
+ "loss": 0.8339,
+ "step": 6185
+ },
+ {
+ "epoch": 16.85558583106267,
+ "grad_norm": 6.595926761627197,
+ "learning_rate": 1.9009907467609544e-05,
+ "loss": 0.719,
+ "step": 6186
+ },
+ {
+ "epoch": 16.858310626703,
+ "grad_norm": 6.365407943725586,
+ "learning_rate": 1.900952457254933e-05,
+ "loss": 0.8488,
+ "step": 6187
+ },
+ {
+ "epoch": 16.861035422343324,
+ "grad_norm": 6.022199630737305,
+ "learning_rate": 1.900914160732323e-05,
+ "loss": 0.8788,
+ "step": 6188
+ },
+ {
+ "epoch": 16.86376021798365,
+ "grad_norm": 6.689509391784668,
+ "learning_rate": 1.900875857193423e-05,
+ "loss": 0.7406,
+ "step": 6189
+ },
+ {
+ "epoch": 16.86648501362398,
+ "grad_norm": 6.8711838722229,
+ "learning_rate": 1.90083754663853e-05,
+ "loss": 0.8289,
+ "step": 6190
+ },
+ {
+ "epoch": 16.869209809264305,
+ "grad_norm": 6.442479610443115,
+ "learning_rate": 1.9007992290679437e-05,
+ "loss": 0.7645,
+ "step": 6191
+ },
+ {
+ "epoch": 16.87193460490463,
+ "grad_norm": 7.6858696937561035,
+ "learning_rate": 1.900760904481962e-05,
+ "loss": 0.6952,
+ "step": 6192
+ },
+ {
+ "epoch": 16.87465940054496,
+ "grad_norm": 6.231497287750244,
+ "learning_rate": 1.900722572880883e-05,
+ "loss": 0.7335,
+ "step": 6193
+ },
+ {
+ "epoch": 16.877384196185286,
+ "grad_norm": 7.495328426361084,
+ "learning_rate": 1.900684234265006e-05,
+ "loss": 0.7901,
+ "step": 6194
+ },
+ {
+ "epoch": 16.88010899182561,
+ "grad_norm": 8.808419227600098,
+ "learning_rate": 1.900645888634629e-05,
+ "loss": 0.869,
+ "step": 6195
+ },
+ {
+ "epoch": 16.88283378746594,
+ "grad_norm": 6.027729511260986,
+ "learning_rate": 1.9006075359900507e-05,
+ "loss": 0.6429,
+ "step": 6196
+ },
+ {
+ "epoch": 16.885558583106267,
+ "grad_norm": 7.605374813079834,
+ "learning_rate": 1.9005691763315703e-05,
+ "loss": 0.7036,
+ "step": 6197
+ },
+ {
+ "epoch": 16.888283378746593,
+ "grad_norm": 8.086008071899414,
+ "learning_rate": 1.9005308096594856e-05,
+ "loss": 0.7173,
+ "step": 6198
+ },
+ {
+ "epoch": 16.891008174386922,
+ "grad_norm": 7.0862321853637695,
+ "learning_rate": 1.9004924359740965e-05,
+ "loss": 0.8044,
+ "step": 6199
+ },
+ {
+ "epoch": 16.893732970027248,
+ "grad_norm": 6.6795549392700195,
+ "learning_rate": 1.9004540552757008e-05,
+ "loss": 0.7637,
+ "step": 6200
+ },
+ {
+ "epoch": 16.896457765667574,
+ "grad_norm": 7.395474433898926,
+ "learning_rate": 1.900415667564598e-05,
+ "loss": 0.8144,
+ "step": 6201
+ },
+ {
+ "epoch": 16.899182561307903,
+ "grad_norm": 5.923164367675781,
+ "learning_rate": 1.900377272841087e-05,
+ "loss": 0.6559,
+ "step": 6202
+ },
+ {
+ "epoch": 16.90190735694823,
+ "grad_norm": 7.671173095703125,
+ "learning_rate": 1.900338871105467e-05,
+ "loss": 0.9193,
+ "step": 6203
+ },
+ {
+ "epoch": 16.904632152588555,
+ "grad_norm": 8.8159761428833,
+ "learning_rate": 1.9003004623580365e-05,
+ "loss": 0.86,
+ "step": 6204
+ },
+ {
+ "epoch": 16.907356948228884,
+ "grad_norm": 30.316810607910156,
+ "learning_rate": 1.9002620465990952e-05,
+ "loss": 0.7484,
+ "step": 6205
+ },
+ {
+ "epoch": 16.91008174386921,
+ "grad_norm": 7.444633960723877,
+ "learning_rate": 1.900223623828942e-05,
+ "loss": 0.8221,
+ "step": 6206
+ },
+ {
+ "epoch": 16.912806539509535,
+ "grad_norm": 6.948957920074463,
+ "learning_rate": 1.9001851940478765e-05,
+ "loss": 0.7515,
+ "step": 6207
+ },
+ {
+ "epoch": 16.915531335149865,
+ "grad_norm": 5.956913948059082,
+ "learning_rate": 1.9001467572561974e-05,
+ "loss": 0.8501,
+ "step": 6208
+ },
+ {
+ "epoch": 16.91825613079019,
+ "grad_norm": 6.309549808502197,
+ "learning_rate": 1.900108313454204e-05,
+ "loss": 0.735,
+ "step": 6209
+ },
+ {
+ "epoch": 16.920980926430516,
+ "grad_norm": 7.597446441650391,
+ "learning_rate": 1.9000698626421966e-05,
+ "loss": 0.755,
+ "step": 6210
+ },
+ {
+ "epoch": 16.923705722070846,
+ "grad_norm": 6.638579368591309,
+ "learning_rate": 1.9000314048204736e-05,
+ "loss": 0.7739,
+ "step": 6211
+ },
+ {
+ "epoch": 16.92643051771117,
+ "grad_norm": 7.528500556945801,
+ "learning_rate": 1.899992939989335e-05,
+ "loss": 0.9274,
+ "step": 6212
+ },
+ {
+ "epoch": 16.929155313351497,
+ "grad_norm": 6.948042392730713,
+ "learning_rate": 1.8999544681490807e-05,
+ "loss": 0.7637,
+ "step": 6213
+ },
+ {
+ "epoch": 16.931880108991827,
+ "grad_norm": 6.696169853210449,
+ "learning_rate": 1.89991598930001e-05,
+ "loss": 0.749,
+ "step": 6214
+ },
+ {
+ "epoch": 16.934604904632153,
+ "grad_norm": 6.307385444641113,
+ "learning_rate": 1.8998775034424222e-05,
+ "loss": 0.9828,
+ "step": 6215
+ },
+ {
+ "epoch": 16.93732970027248,
+ "grad_norm": 6.623286247253418,
+ "learning_rate": 1.8998390105766175e-05,
+ "loss": 0.7594,
+ "step": 6216
+ },
+ {
+ "epoch": 16.940054495912808,
+ "grad_norm": 7.397196292877197,
+ "learning_rate": 1.8998005107028956e-05,
+ "loss": 0.6481,
+ "step": 6217
+ },
+ {
+ "epoch": 16.942779291553133,
+ "grad_norm": 7.211545944213867,
+ "learning_rate": 1.8997620038215562e-05,
+ "loss": 0.7859,
+ "step": 6218
+ },
+ {
+ "epoch": 16.94550408719346,
+ "grad_norm": 7.971397399902344,
+ "learning_rate": 1.8997234899328997e-05,
+ "loss": 0.7963,
+ "step": 6219
+ },
+ {
+ "epoch": 16.94822888283379,
+ "grad_norm": 7.020237445831299,
+ "learning_rate": 1.899684969037225e-05,
+ "loss": 0.7078,
+ "step": 6220
+ },
+ {
+ "epoch": 16.950953678474114,
+ "grad_norm": 8.155545234680176,
+ "learning_rate": 1.899646441134833e-05,
+ "loss": 0.7478,
+ "step": 6221
+ },
+ {
+ "epoch": 16.95367847411444,
+ "grad_norm": 7.1253533363342285,
+ "learning_rate": 1.8996079062260234e-05,
+ "loss": 0.9309,
+ "step": 6222
+ },
+ {
+ "epoch": 16.95640326975477,
+ "grad_norm": 6.485459804534912,
+ "learning_rate": 1.8995693643110964e-05,
+ "loss": 0.8203,
+ "step": 6223
+ },
+ {
+ "epoch": 16.959128065395095,
+ "grad_norm": 5.499973297119141,
+ "learning_rate": 1.899530815390352e-05,
+ "loss": 0.6416,
+ "step": 6224
+ },
+ {
+ "epoch": 16.96185286103542,
+ "grad_norm": 6.858002662658691,
+ "learning_rate": 1.8994922594640907e-05,
+ "loss": 0.7297,
+ "step": 6225
+ },
+ {
+ "epoch": 16.96457765667575,
+ "grad_norm": 6.953660488128662,
+ "learning_rate": 1.899453696532613e-05,
+ "loss": 0.9147,
+ "step": 6226
+ },
+ {
+ "epoch": 16.967302452316076,
+ "grad_norm": 8.102987289428711,
+ "learning_rate": 1.8994151265962184e-05,
+ "loss": 0.9213,
+ "step": 6227
+ },
+ {
+ "epoch": 16.970027247956402,
+ "grad_norm": 6.692060470581055,
+ "learning_rate": 1.8993765496552078e-05,
+ "loss": 0.816,
+ "step": 6228
+ },
+ {
+ "epoch": 16.97275204359673,
+ "grad_norm": 7.124213695526123,
+ "learning_rate": 1.8993379657098815e-05,
+ "loss": 0.6643,
+ "step": 6229
+ },
+ {
+ "epoch": 16.975476839237057,
+ "grad_norm": 8.641468048095703,
+ "learning_rate": 1.89929937476054e-05,
+ "loss": 0.8864,
+ "step": 6230
+ },
+ {
+ "epoch": 16.978201634877383,
+ "grad_norm": 7.933237552642822,
+ "learning_rate": 1.899260776807484e-05,
+ "loss": 1.0063,
+ "step": 6231
+ },
+ {
+ "epoch": 16.980926430517712,
+ "grad_norm": 6.9891438484191895,
+ "learning_rate": 1.8992221718510143e-05,
+ "loss": 0.6268,
+ "step": 6232
+ },
+ {
+ "epoch": 16.983651226158038,
+ "grad_norm": 6.6686530113220215,
+ "learning_rate": 1.8991835598914313e-05,
+ "loss": 0.8322,
+ "step": 6233
+ },
+ {
+ "epoch": 16.986376021798364,
+ "grad_norm": 7.1252760887146,
+ "learning_rate": 1.8991449409290355e-05,
+ "loss": 0.8416,
+ "step": 6234
+ },
+ {
+ "epoch": 16.989100817438693,
+ "grad_norm": 6.719127178192139,
+ "learning_rate": 1.8991063149641276e-05,
+ "loss": 0.9313,
+ "step": 6235
+ },
+ {
+ "epoch": 16.99182561307902,
+ "grad_norm": 8.646967887878418,
+ "learning_rate": 1.8990676819970087e-05,
+ "loss": 0.876,
+ "step": 6236
+ },
+ {
+ "epoch": 16.994550408719345,
+ "grad_norm": 6.132476806640625,
+ "learning_rate": 1.89902904202798e-05,
+ "loss": 0.6207,
+ "step": 6237
+ },
+ {
+ "epoch": 16.997275204359674,
+ "grad_norm": 8.386879920959473,
+ "learning_rate": 1.8989903950573422e-05,
+ "loss": 0.8141,
+ "step": 6238
+ },
+ {
+ "epoch": 17.0,
+ "grad_norm": 7.309690952301025,
+ "learning_rate": 1.8989517410853956e-05,
+ "loss": 0.6632,
+ "step": 6239
+ },
+ {
+ "epoch": 17.002724795640326,
+ "grad_norm": 7.900304317474365,
+ "learning_rate": 1.898913080112442e-05,
+ "loss": 0.8077,
+ "step": 6240
+ },
+ {
+ "epoch": 17.005449591280655,
+ "grad_norm": 7.038301467895508,
+ "learning_rate": 1.898874412138782e-05,
+ "loss": 0.7733,
+ "step": 6241
+ },
+ {
+ "epoch": 17.00817438692098,
+ "grad_norm": 8.787518501281738,
+ "learning_rate": 1.8988357371647173e-05,
+ "loss": 0.6498,
+ "step": 6242
+ },
+ {
+ "epoch": 17.010899182561307,
+ "grad_norm": 6.3860015869140625,
+ "learning_rate": 1.898797055190549e-05,
+ "loss": 0.6042,
+ "step": 6243
+ },
+ {
+ "epoch": 17.013623978201636,
+ "grad_norm": 7.07371711730957,
+ "learning_rate": 1.8987583662165775e-05,
+ "loss": 0.656,
+ "step": 6244
+ },
+ {
+ "epoch": 17.016348773841962,
+ "grad_norm": 5.861824989318848,
+ "learning_rate": 1.8987196702431053e-05,
+ "loss": 0.6637,
+ "step": 6245
+ },
+ {
+ "epoch": 17.019073569482288,
+ "grad_norm": 6.316766738891602,
+ "learning_rate": 1.8986809672704328e-05,
+ "loss": 0.5622,
+ "step": 6246
+ },
+ {
+ "epoch": 17.021798365122617,
+ "grad_norm": 6.155854225158691,
+ "learning_rate": 1.8986422572988626e-05,
+ "loss": 0.7516,
+ "step": 6247
+ },
+ {
+ "epoch": 17.024523160762943,
+ "grad_norm": 7.228900909423828,
+ "learning_rate": 1.8986035403286948e-05,
+ "loss": 0.8264,
+ "step": 6248
+ },
+ {
+ "epoch": 17.02724795640327,
+ "grad_norm": 6.894256114959717,
+ "learning_rate": 1.8985648163602318e-05,
+ "loss": 0.7422,
+ "step": 6249
+ },
+ {
+ "epoch": 17.029972752043598,
+ "grad_norm": 9.867069244384766,
+ "learning_rate": 1.898526085393775e-05,
+ "loss": 0.8016,
+ "step": 6250
+ },
+ {
+ "epoch": 17.032697547683924,
+ "grad_norm": 7.057434558868408,
+ "learning_rate": 1.8984873474296256e-05,
+ "loss": 0.5826,
+ "step": 6251
+ },
+ {
+ "epoch": 17.03542234332425,
+ "grad_norm": 6.610015392303467,
+ "learning_rate": 1.898448602468086e-05,
+ "loss": 0.8226,
+ "step": 6252
+ },
+ {
+ "epoch": 17.03814713896458,
+ "grad_norm": 6.233609676361084,
+ "learning_rate": 1.8984098505094574e-05,
+ "loss": 0.5338,
+ "step": 6253
+ },
+ {
+ "epoch": 17.040871934604905,
+ "grad_norm": 7.139139652252197,
+ "learning_rate": 1.898371091554042e-05,
+ "loss": 0.6487,
+ "step": 6254
+ },
+ {
+ "epoch": 17.04359673024523,
+ "grad_norm": 5.406883239746094,
+ "learning_rate": 1.8983323256021412e-05,
+ "loss": 0.666,
+ "step": 6255
+ },
+ {
+ "epoch": 17.04632152588556,
+ "grad_norm": 7.8883161544799805,
+ "learning_rate": 1.8982935526540572e-05,
+ "loss": 0.7443,
+ "step": 6256
+ },
+ {
+ "epoch": 17.049046321525886,
+ "grad_norm": 6.907222747802734,
+ "learning_rate": 1.898254772710092e-05,
+ "loss": 0.6919,
+ "step": 6257
+ },
+ {
+ "epoch": 17.05177111716621,
+ "grad_norm": 6.390595436096191,
+ "learning_rate": 1.8982159857705475e-05,
+ "loss": 0.6058,
+ "step": 6258
+ },
+ {
+ "epoch": 17.05449591280654,
+ "grad_norm": 6.6581854820251465,
+ "learning_rate": 1.8981771918357257e-05,
+ "loss": 0.7394,
+ "step": 6259
+ },
+ {
+ "epoch": 17.057220708446867,
+ "grad_norm": 5.99835205078125,
+ "learning_rate": 1.898138390905929e-05,
+ "loss": 0.9042,
+ "step": 6260
+ },
+ {
+ "epoch": 17.059945504087192,
+ "grad_norm": 6.984434604644775,
+ "learning_rate": 1.8980995829814593e-05,
+ "loss": 0.8203,
+ "step": 6261
+ },
+ {
+ "epoch": 17.06267029972752,
+ "grad_norm": 6.42523193359375,
+ "learning_rate": 1.8980607680626194e-05,
+ "loss": 0.5149,
+ "step": 6262
+ },
+ {
+ "epoch": 17.065395095367847,
+ "grad_norm": 7.250277996063232,
+ "learning_rate": 1.8980219461497106e-05,
+ "loss": 0.5605,
+ "step": 6263
+ },
+ {
+ "epoch": 17.068119891008173,
+ "grad_norm": 6.645427703857422,
+ "learning_rate": 1.8979831172430364e-05,
+ "loss": 0.721,
+ "step": 6264
+ },
+ {
+ "epoch": 17.070844686648503,
+ "grad_norm": 5.983193874359131,
+ "learning_rate": 1.897944281342898e-05,
+ "loss": 0.5955,
+ "step": 6265
+ },
+ {
+ "epoch": 17.07356948228883,
+ "grad_norm": 8.255365371704102,
+ "learning_rate": 1.8979054384495987e-05,
+ "loss": 0.6471,
+ "step": 6266
+ },
+ {
+ "epoch": 17.076294277929154,
+ "grad_norm": 6.856241703033447,
+ "learning_rate": 1.8978665885634406e-05,
+ "loss": 0.7079,
+ "step": 6267
+ },
+ {
+ "epoch": 17.079019073569484,
+ "grad_norm": 7.275993347167969,
+ "learning_rate": 1.8978277316847265e-05,
+ "loss": 0.6599,
+ "step": 6268
+ },
+ {
+ "epoch": 17.08174386920981,
+ "grad_norm": 5.8820977210998535,
+ "learning_rate": 1.897788867813759e-05,
+ "loss": 0.7008,
+ "step": 6269
+ },
+ {
+ "epoch": 17.084468664850135,
+ "grad_norm": 7.619244575500488,
+ "learning_rate": 1.8977499969508405e-05,
+ "loss": 0.8054,
+ "step": 6270
+ },
+ {
+ "epoch": 17.087193460490465,
+ "grad_norm": 5.152529716491699,
+ "learning_rate": 1.897711119096274e-05,
+ "loss": 0.6021,
+ "step": 6271
+ },
+ {
+ "epoch": 17.08991825613079,
+ "grad_norm": 7.614264488220215,
+ "learning_rate": 1.897672234250362e-05,
+ "loss": 0.7166,
+ "step": 6272
+ },
+ {
+ "epoch": 17.092643051771116,
+ "grad_norm": 5.804299831390381,
+ "learning_rate": 1.897633342413408e-05,
+ "loss": 0.5257,
+ "step": 6273
+ },
+ {
+ "epoch": 17.095367847411445,
+ "grad_norm": 6.922432899475098,
+ "learning_rate": 1.897594443585714e-05,
+ "loss": 0.7531,
+ "step": 6274
+ },
+ {
+ "epoch": 17.09809264305177,
+ "grad_norm": 5.398029804229736,
+ "learning_rate": 1.8975555377675835e-05,
+ "loss": 0.6086,
+ "step": 6275
+ },
+ {
+ "epoch": 17.100817438692097,
+ "grad_norm": 7.279467582702637,
+ "learning_rate": 1.8975166249593194e-05,
+ "loss": 0.5972,
+ "step": 6276
+ },
+ {
+ "epoch": 17.103542234332426,
+ "grad_norm": 7.466846942901611,
+ "learning_rate": 1.897477705161225e-05,
+ "loss": 0.66,
+ "step": 6277
+ },
+ {
+ "epoch": 17.106267029972752,
+ "grad_norm": 9.215846061706543,
+ "learning_rate": 1.8974387783736022e-05,
+ "loss": 0.751,
+ "step": 6278
+ },
+ {
+ "epoch": 17.108991825613078,
+ "grad_norm": 5.291007041931152,
+ "learning_rate": 1.897399844596756e-05,
+ "loss": 0.8178,
+ "step": 6279
+ },
+ {
+ "epoch": 17.111716621253407,
+ "grad_norm": 6.434169292449951,
+ "learning_rate": 1.8973609038309883e-05,
+ "loss": 0.6153,
+ "step": 6280
+ },
+ {
+ "epoch": 17.114441416893733,
+ "grad_norm": 7.005979537963867,
+ "learning_rate": 1.897321956076603e-05,
+ "loss": 0.6875,
+ "step": 6281
+ },
+ {
+ "epoch": 17.11716621253406,
+ "grad_norm": 6.124795913696289,
+ "learning_rate": 1.897283001333903e-05,
+ "loss": 0.642,
+ "step": 6282
+ },
+ {
+ "epoch": 17.11989100817439,
+ "grad_norm": 6.8803815841674805,
+ "learning_rate": 1.8972440396031916e-05,
+ "loss": 0.7649,
+ "step": 6283
+ },
+ {
+ "epoch": 17.122615803814714,
+ "grad_norm": 6.2144856452941895,
+ "learning_rate": 1.897205070884773e-05,
+ "loss": 0.7095,
+ "step": 6284
+ },
+ {
+ "epoch": 17.12534059945504,
+ "grad_norm": 8.787986755371094,
+ "learning_rate": 1.89716609517895e-05,
+ "loss": 0.8822,
+ "step": 6285
+ },
+ {
+ "epoch": 17.12806539509537,
+ "grad_norm": 7.370906352996826,
+ "learning_rate": 1.8971271124860263e-05,
+ "loss": 0.8054,
+ "step": 6286
+ },
+ {
+ "epoch": 17.130790190735695,
+ "grad_norm": 6.905709743499756,
+ "learning_rate": 1.8970881228063055e-05,
+ "loss": 0.5979,
+ "step": 6287
+ },
+ {
+ "epoch": 17.13351498637602,
+ "grad_norm": 9.528563499450684,
+ "learning_rate": 1.8970491261400913e-05,
+ "loss": 0.745,
+ "step": 6288
+ },
+ {
+ "epoch": 17.13623978201635,
+ "grad_norm": 5.9495415687561035,
+ "learning_rate": 1.8970101224876872e-05,
+ "loss": 0.6755,
+ "step": 6289
+ },
+ {
+ "epoch": 17.138964577656676,
+ "grad_norm": 9.190239906311035,
+ "learning_rate": 1.8969711118493976e-05,
+ "loss": 0.8062,
+ "step": 6290
+ },
+ {
+ "epoch": 17.141689373297,
+ "grad_norm": 7.346141815185547,
+ "learning_rate": 1.8969320942255252e-05,
+ "loss": 0.7225,
+ "step": 6291
+ },
+ {
+ "epoch": 17.14441416893733,
+ "grad_norm": 5.524956226348877,
+ "learning_rate": 1.896893069616375e-05,
+ "loss": 0.6184,
+ "step": 6292
+ },
+ {
+ "epoch": 17.147138964577657,
+ "grad_norm": 7.064297676086426,
+ "learning_rate": 1.89685403802225e-05,
+ "loss": 0.9586,
+ "step": 6293
+ },
+ {
+ "epoch": 17.149863760217983,
+ "grad_norm": 5.855534553527832,
+ "learning_rate": 1.8968149994434552e-05,
+ "loss": 0.6437,
+ "step": 6294
+ },
+ {
+ "epoch": 17.152588555858312,
+ "grad_norm": 7.246648788452148,
+ "learning_rate": 1.896775953880294e-05,
+ "loss": 0.7099,
+ "step": 6295
+ },
+ {
+ "epoch": 17.155313351498638,
+ "grad_norm": 6.906251430511475,
+ "learning_rate": 1.8967369013330702e-05,
+ "loss": 0.6466,
+ "step": 6296
+ },
+ {
+ "epoch": 17.158038147138964,
+ "grad_norm": 6.563923358917236,
+ "learning_rate": 1.8966978418020883e-05,
+ "loss": 0.7247,
+ "step": 6297
+ },
+ {
+ "epoch": 17.160762942779293,
+ "grad_norm": 5.128292083740234,
+ "learning_rate": 1.8966587752876524e-05,
+ "loss": 0.7286,
+ "step": 6298
+ },
+ {
+ "epoch": 17.16348773841962,
+ "grad_norm": 7.34670877456665,
+ "learning_rate": 1.8966197017900672e-05,
+ "loss": 0.7284,
+ "step": 6299
+ },
+ {
+ "epoch": 17.166212534059945,
+ "grad_norm": 6.791901111602783,
+ "learning_rate": 1.8965806213096363e-05,
+ "loss": 0.6202,
+ "step": 6300
+ },
+ {
+ "epoch": 17.168937329700274,
+ "grad_norm": 6.966001033782959,
+ "learning_rate": 1.8965415338466645e-05,
+ "loss": 0.7554,
+ "step": 6301
+ },
+ {
+ "epoch": 17.1716621253406,
+ "grad_norm": 5.787513256072998,
+ "learning_rate": 1.896502439401456e-05,
+ "loss": 0.6716,
+ "step": 6302
+ },
+ {
+ "epoch": 17.174386920980925,
+ "grad_norm": 6.597262382507324,
+ "learning_rate": 1.8964633379743154e-05,
+ "loss": 0.7257,
+ "step": 6303
+ },
+ {
+ "epoch": 17.177111716621255,
+ "grad_norm": 5.432859420776367,
+ "learning_rate": 1.8964242295655473e-05,
+ "loss": 0.5633,
+ "step": 6304
+ },
+ {
+ "epoch": 17.17983651226158,
+ "grad_norm": 7.394481182098389,
+ "learning_rate": 1.8963851141754558e-05,
+ "loss": 0.9386,
+ "step": 6305
+ },
+ {
+ "epoch": 17.182561307901906,
+ "grad_norm": 7.5224175453186035,
+ "learning_rate": 1.8963459918043463e-05,
+ "loss": 0.7823,
+ "step": 6306
+ },
+ {
+ "epoch": 17.185286103542236,
+ "grad_norm": 18.190393447875977,
+ "learning_rate": 1.896306862452523e-05,
+ "loss": 0.6426,
+ "step": 6307
+ },
+ {
+ "epoch": 17.18801089918256,
+ "grad_norm": 6.948112964630127,
+ "learning_rate": 1.8962677261202905e-05,
+ "loss": 0.8157,
+ "step": 6308
+ },
+ {
+ "epoch": 17.190735694822887,
+ "grad_norm": 5.916109085083008,
+ "learning_rate": 1.896228582807954e-05,
+ "loss": 0.6531,
+ "step": 6309
+ },
+ {
+ "epoch": 17.193460490463217,
+ "grad_norm": 9.558351516723633,
+ "learning_rate": 1.896189432515818e-05,
+ "loss": 0.7236,
+ "step": 6310
+ },
+ {
+ "epoch": 17.196185286103542,
+ "grad_norm": 6.427715301513672,
+ "learning_rate": 1.8961502752441875e-05,
+ "loss": 0.7009,
+ "step": 6311
+ },
+ {
+ "epoch": 17.19891008174387,
+ "grad_norm": 6.248469829559326,
+ "learning_rate": 1.8961111109933673e-05,
+ "loss": 0.8118,
+ "step": 6312
+ },
+ {
+ "epoch": 17.201634877384198,
+ "grad_norm": 6.928211212158203,
+ "learning_rate": 1.896071939763663e-05,
+ "loss": 0.7572,
+ "step": 6313
+ },
+ {
+ "epoch": 17.204359673024523,
+ "grad_norm": 6.250397682189941,
+ "learning_rate": 1.8960327615553794e-05,
+ "loss": 0.7011,
+ "step": 6314
+ },
+ {
+ "epoch": 17.20708446866485,
+ "grad_norm": 6.1008124351501465,
+ "learning_rate": 1.8959935763688212e-05,
+ "loss": 0.8258,
+ "step": 6315
+ },
+ {
+ "epoch": 17.20980926430518,
+ "grad_norm": 7.243523597717285,
+ "learning_rate": 1.895954384204294e-05,
+ "loss": 0.6493,
+ "step": 6316
+ },
+ {
+ "epoch": 17.212534059945504,
+ "grad_norm": 7.7331743240356445,
+ "learning_rate": 1.8959151850621025e-05,
+ "loss": 0.7296,
+ "step": 6317
+ },
+ {
+ "epoch": 17.21525885558583,
+ "grad_norm": 5.623905181884766,
+ "learning_rate": 1.895875978942553e-05,
+ "loss": 0.677,
+ "step": 6318
+ },
+ {
+ "epoch": 17.21798365122616,
+ "grad_norm": 5.60208797454834,
+ "learning_rate": 1.8958367658459494e-05,
+ "loss": 0.6311,
+ "step": 6319
+ },
+ {
+ "epoch": 17.220708446866485,
+ "grad_norm": 5.9532036781311035,
+ "learning_rate": 1.8957975457725985e-05,
+ "loss": 0.6965,
+ "step": 6320
+ },
+ {
+ "epoch": 17.22343324250681,
+ "grad_norm": 9.193092346191406,
+ "learning_rate": 1.895758318722805e-05,
+ "loss": 0.7743,
+ "step": 6321
+ },
+ {
+ "epoch": 17.22615803814714,
+ "grad_norm": 8.151643753051758,
+ "learning_rate": 1.8957190846968747e-05,
+ "loss": 0.7368,
+ "step": 6322
+ },
+ {
+ "epoch": 17.228882833787466,
+ "grad_norm": 7.296803951263428,
+ "learning_rate": 1.895679843695113e-05,
+ "loss": 0.7054,
+ "step": 6323
+ },
+ {
+ "epoch": 17.231607629427792,
+ "grad_norm": 7.099100112915039,
+ "learning_rate": 1.8956405957178253e-05,
+ "loss": 0.7827,
+ "step": 6324
+ },
+ {
+ "epoch": 17.23433242506812,
+ "grad_norm": 7.496584415435791,
+ "learning_rate": 1.8956013407653175e-05,
+ "loss": 0.577,
+ "step": 6325
+ },
+ {
+ "epoch": 17.237057220708447,
+ "grad_norm": 6.990920543670654,
+ "learning_rate": 1.895562078837895e-05,
+ "loss": 0.7979,
+ "step": 6326
+ },
+ {
+ "epoch": 17.239782016348773,
+ "grad_norm": 6.5025835037231445,
+ "learning_rate": 1.8955228099358646e-05,
+ "loss": 0.7057,
+ "step": 6327
+ },
+ {
+ "epoch": 17.242506811989102,
+ "grad_norm": 6.824868202209473,
+ "learning_rate": 1.8954835340595305e-05,
+ "loss": 0.7205,
+ "step": 6328
+ },
+ {
+ "epoch": 17.245231607629428,
+ "grad_norm": 6.487971305847168,
+ "learning_rate": 1.8954442512092e-05,
+ "loss": 0.6322,
+ "step": 6329
+ },
+ {
+ "epoch": 17.247956403269754,
+ "grad_norm": 6.571447849273682,
+ "learning_rate": 1.8954049613851783e-05,
+ "loss": 0.9177,
+ "step": 6330
+ },
+ {
+ "epoch": 17.250681198910083,
+ "grad_norm": 9.61840534210205,
+ "learning_rate": 1.895365664587772e-05,
+ "loss": 0.8303,
+ "step": 6331
+ },
+ {
+ "epoch": 17.25340599455041,
+ "grad_norm": 7.252959251403809,
+ "learning_rate": 1.8953263608172864e-05,
+ "loss": 0.7188,
+ "step": 6332
+ },
+ {
+ "epoch": 17.256130790190735,
+ "grad_norm": 5.826846599578857,
+ "learning_rate": 1.895287050074028e-05,
+ "loss": 0.7489,
+ "step": 6333
+ },
+ {
+ "epoch": 17.258855585831064,
+ "grad_norm": 19.090900421142578,
+ "learning_rate": 1.8952477323583024e-05,
+ "loss": 0.6481,
+ "step": 6334
+ },
+ {
+ "epoch": 17.26158038147139,
+ "grad_norm": 5.683408260345459,
+ "learning_rate": 1.895208407670417e-05,
+ "loss": 0.6774,
+ "step": 6335
+ },
+ {
+ "epoch": 17.264305177111716,
+ "grad_norm": 7.670852184295654,
+ "learning_rate": 1.895169076010677e-05,
+ "loss": 0.7694,
+ "step": 6336
+ },
+ {
+ "epoch": 17.267029972752045,
+ "grad_norm": 7.345246315002441,
+ "learning_rate": 1.895129737379389e-05,
+ "loss": 0.8094,
+ "step": 6337
+ },
+ {
+ "epoch": 17.26975476839237,
+ "grad_norm": 8.049223899841309,
+ "learning_rate": 1.8950903917768594e-05,
+ "loss": 0.7202,
+ "step": 6338
+ },
+ {
+ "epoch": 17.272479564032697,
+ "grad_norm": 7.275326728820801,
+ "learning_rate": 1.8950510392033945e-05,
+ "loss": 0.8127,
+ "step": 6339
+ },
+ {
+ "epoch": 17.275204359673026,
+ "grad_norm": 8.11840534210205,
+ "learning_rate": 1.8950116796593013e-05,
+ "loss": 0.6619,
+ "step": 6340
+ },
+ {
+ "epoch": 17.277929155313352,
+ "grad_norm": 6.708151340484619,
+ "learning_rate": 1.894972313144886e-05,
+ "loss": 0.7583,
+ "step": 6341
+ },
+ {
+ "epoch": 17.280653950953678,
+ "grad_norm": 6.415965557098389,
+ "learning_rate": 1.8949329396604544e-05,
+ "loss": 0.6183,
+ "step": 6342
+ },
+ {
+ "epoch": 17.283378746594007,
+ "grad_norm": 6.682216644287109,
+ "learning_rate": 1.8948935592063145e-05,
+ "loss": 0.6939,
+ "step": 6343
+ },
+ {
+ "epoch": 17.286103542234333,
+ "grad_norm": 6.1094889640808105,
+ "learning_rate": 1.8948541717827723e-05,
+ "loss": 0.8011,
+ "step": 6344
+ },
+ {
+ "epoch": 17.28882833787466,
+ "grad_norm": 6.307723045349121,
+ "learning_rate": 1.8948147773901343e-05,
+ "loss": 0.7853,
+ "step": 6345
+ },
+ {
+ "epoch": 17.291553133514988,
+ "grad_norm": 6.287580966949463,
+ "learning_rate": 1.894775376028708e-05,
+ "loss": 0.6055,
+ "step": 6346
+ },
+ {
+ "epoch": 17.294277929155314,
+ "grad_norm": 5.958017826080322,
+ "learning_rate": 1.8947359676988e-05,
+ "loss": 0.5553,
+ "step": 6347
+ },
+ {
+ "epoch": 17.29700272479564,
+ "grad_norm": 9.289746284484863,
+ "learning_rate": 1.8946965524007166e-05,
+ "loss": 0.6244,
+ "step": 6348
+ },
+ {
+ "epoch": 17.29972752043597,
+ "grad_norm": 6.596959114074707,
+ "learning_rate": 1.8946571301347654e-05,
+ "loss": 0.6786,
+ "step": 6349
+ },
+ {
+ "epoch": 17.302452316076295,
+ "grad_norm": 6.253579139709473,
+ "learning_rate": 1.894617700901253e-05,
+ "loss": 0.7845,
+ "step": 6350
+ },
+ {
+ "epoch": 17.30517711171662,
+ "grad_norm": 5.801887512207031,
+ "learning_rate": 1.894578264700487e-05,
+ "loss": 0.623,
+ "step": 6351
+ },
+ {
+ "epoch": 17.30790190735695,
+ "grad_norm": 7.860487461090088,
+ "learning_rate": 1.8945388215327746e-05,
+ "loss": 0.7064,
+ "step": 6352
+ },
+ {
+ "epoch": 17.310626702997276,
+ "grad_norm": 6.10924768447876,
+ "learning_rate": 1.8944993713984224e-05,
+ "loss": 0.9369,
+ "step": 6353
+ },
+ {
+ "epoch": 17.3133514986376,
+ "grad_norm": 5.688379287719727,
+ "learning_rate": 1.8944599142977378e-05,
+ "loss": 0.6447,
+ "step": 6354
+ },
+ {
+ "epoch": 17.31607629427793,
+ "grad_norm": 7.069969654083252,
+ "learning_rate": 1.8944204502310282e-05,
+ "loss": 0.7485,
+ "step": 6355
+ },
+ {
+ "epoch": 17.318801089918257,
+ "grad_norm": 7.075206756591797,
+ "learning_rate": 1.894380979198601e-05,
+ "loss": 0.7811,
+ "step": 6356
+ },
+ {
+ "epoch": 17.321525885558582,
+ "grad_norm": 6.056546688079834,
+ "learning_rate": 1.8943415012007634e-05,
+ "loss": 0.6851,
+ "step": 6357
+ },
+ {
+ "epoch": 17.32425068119891,
+ "grad_norm": 5.592612266540527,
+ "learning_rate": 1.8943020162378228e-05,
+ "loss": 0.5781,
+ "step": 6358
+ },
+ {
+ "epoch": 17.326975476839237,
+ "grad_norm": 6.410040378570557,
+ "learning_rate": 1.8942625243100874e-05,
+ "loss": 0.7051,
+ "step": 6359
+ },
+ {
+ "epoch": 17.329700272479563,
+ "grad_norm": 5.837237358093262,
+ "learning_rate": 1.8942230254178638e-05,
+ "loss": 0.6859,
+ "step": 6360
+ },
+ {
+ "epoch": 17.332425068119893,
+ "grad_norm": 6.765936851501465,
+ "learning_rate": 1.8941835195614603e-05,
+ "loss": 0.7742,
+ "step": 6361
+ },
+ {
+ "epoch": 17.33514986376022,
+ "grad_norm": 7.134358882904053,
+ "learning_rate": 1.894144006741184e-05,
+ "loss": 0.7793,
+ "step": 6362
+ },
+ {
+ "epoch": 17.337874659400544,
+ "grad_norm": 7.1174798011779785,
+ "learning_rate": 1.8941044869573433e-05,
+ "loss": 0.7377,
+ "step": 6363
+ },
+ {
+ "epoch": 17.340599455040874,
+ "grad_norm": 6.469444751739502,
+ "learning_rate": 1.8940649602102454e-05,
+ "loss": 0.5914,
+ "step": 6364
+ },
+ {
+ "epoch": 17.3433242506812,
+ "grad_norm": 7.302258491516113,
+ "learning_rate": 1.8940254265001986e-05,
+ "loss": 0.6718,
+ "step": 6365
+ },
+ {
+ "epoch": 17.346049046321525,
+ "grad_norm": 6.6016364097595215,
+ "learning_rate": 1.8939858858275104e-05,
+ "loss": 0.9785,
+ "step": 6366
+ },
+ {
+ "epoch": 17.348773841961854,
+ "grad_norm": 5.968013763427734,
+ "learning_rate": 1.893946338192489e-05,
+ "loss": 0.8253,
+ "step": 6367
+ },
+ {
+ "epoch": 17.35149863760218,
+ "grad_norm": 5.980764865875244,
+ "learning_rate": 1.8939067835954423e-05,
+ "loss": 0.6736,
+ "step": 6368
+ },
+ {
+ "epoch": 17.354223433242506,
+ "grad_norm": 11.60478687286377,
+ "learning_rate": 1.8938672220366784e-05,
+ "loss": 0.6237,
+ "step": 6369
+ },
+ {
+ "epoch": 17.356948228882835,
+ "grad_norm": 5.897181510925293,
+ "learning_rate": 1.893827653516505e-05,
+ "loss": 0.7073,
+ "step": 6370
+ },
+ {
+ "epoch": 17.35967302452316,
+ "grad_norm": 6.693094730377197,
+ "learning_rate": 1.8937880780352307e-05,
+ "loss": 1.1505,
+ "step": 6371
+ },
+ {
+ "epoch": 17.362397820163487,
+ "grad_norm": 6.291951656341553,
+ "learning_rate": 1.893748495593164e-05,
+ "loss": 0.6931,
+ "step": 6372
+ },
+ {
+ "epoch": 17.365122615803816,
+ "grad_norm": 6.088106632232666,
+ "learning_rate": 1.8937089061906123e-05,
+ "loss": 0.7579,
+ "step": 6373
+ },
+ {
+ "epoch": 17.367847411444142,
+ "grad_norm": 6.92391300201416,
+ "learning_rate": 1.8936693098278846e-05,
+ "loss": 0.8954,
+ "step": 6374
+ },
+ {
+ "epoch": 17.370572207084468,
+ "grad_norm": 7.604726314544678,
+ "learning_rate": 1.893629706505289e-05,
+ "loss": 0.7224,
+ "step": 6375
+ },
+ {
+ "epoch": 17.373297002724797,
+ "grad_norm": 5.662083148956299,
+ "learning_rate": 1.8935900962231342e-05,
+ "loss": 0.7203,
+ "step": 6376
+ },
+ {
+ "epoch": 17.376021798365123,
+ "grad_norm": 6.79353666305542,
+ "learning_rate": 1.8935504789817284e-05,
+ "loss": 0.6809,
+ "step": 6377
+ },
+ {
+ "epoch": 17.37874659400545,
+ "grad_norm": 5.929373264312744,
+ "learning_rate": 1.89351085478138e-05,
+ "loss": 0.6534,
+ "step": 6378
+ },
+ {
+ "epoch": 17.381471389645778,
+ "grad_norm": 7.969925403594971,
+ "learning_rate": 1.8934712236223984e-05,
+ "loss": 0.5735,
+ "step": 6379
+ },
+ {
+ "epoch": 17.384196185286104,
+ "grad_norm": 6.933822154998779,
+ "learning_rate": 1.893431585505091e-05,
+ "loss": 0.6334,
+ "step": 6380
+ },
+ {
+ "epoch": 17.38692098092643,
+ "grad_norm": 6.938984394073486,
+ "learning_rate": 1.8933919404297676e-05,
+ "loss": 0.7036,
+ "step": 6381
+ },
+ {
+ "epoch": 17.38964577656676,
+ "grad_norm": 5.739816188812256,
+ "learning_rate": 1.8933522883967363e-05,
+ "loss": 0.6583,
+ "step": 6382
+ },
+ {
+ "epoch": 17.392370572207085,
+ "grad_norm": 6.179049968719482,
+ "learning_rate": 1.893312629406306e-05,
+ "loss": 0.7606,
+ "step": 6383
+ },
+ {
+ "epoch": 17.39509536784741,
+ "grad_norm": 7.696129322052002,
+ "learning_rate": 1.8932729634587863e-05,
+ "loss": 0.767,
+ "step": 6384
+ },
+ {
+ "epoch": 17.39782016348774,
+ "grad_norm": 6.63026237487793,
+ "learning_rate": 1.893233290554485e-05,
+ "loss": 0.7032,
+ "step": 6385
+ },
+ {
+ "epoch": 17.400544959128066,
+ "grad_norm": 7.381507873535156,
+ "learning_rate": 1.8931936106937116e-05,
+ "loss": 0.7191,
+ "step": 6386
+ },
+ {
+ "epoch": 17.40326975476839,
+ "grad_norm": 5.834630489349365,
+ "learning_rate": 1.8931539238767753e-05,
+ "loss": 0.6741,
+ "step": 6387
+ },
+ {
+ "epoch": 17.40599455040872,
+ "grad_norm": 6.68784761428833,
+ "learning_rate": 1.893114230103985e-05,
+ "loss": 0.6779,
+ "step": 6388
+ },
+ {
+ "epoch": 17.408719346049047,
+ "grad_norm": 7.498769760131836,
+ "learning_rate": 1.8930745293756496e-05,
+ "loss": 0.7709,
+ "step": 6389
+ },
+ {
+ "epoch": 17.411444141689373,
+ "grad_norm": 6.854574680328369,
+ "learning_rate": 1.8930348216920786e-05,
+ "loss": 0.6721,
+ "step": 6390
+ },
+ {
+ "epoch": 17.414168937329702,
+ "grad_norm": 6.426905632019043,
+ "learning_rate": 1.8929951070535814e-05,
+ "loss": 0.8548,
+ "step": 6391
+ },
+ {
+ "epoch": 17.416893732970028,
+ "grad_norm": 5.987253189086914,
+ "learning_rate": 1.8929553854604667e-05,
+ "loss": 0.7737,
+ "step": 6392
+ },
+ {
+ "epoch": 17.419618528610354,
+ "grad_norm": 8.715110778808594,
+ "learning_rate": 1.892915656913044e-05,
+ "loss": 0.6545,
+ "step": 6393
+ },
+ {
+ "epoch": 17.422343324250683,
+ "grad_norm": 6.16935396194458,
+ "learning_rate": 1.8928759214116236e-05,
+ "loss": 0.8419,
+ "step": 6394
+ },
+ {
+ "epoch": 17.42506811989101,
+ "grad_norm": 7.999082565307617,
+ "learning_rate": 1.8928361789565136e-05,
+ "loss": 0.7374,
+ "step": 6395
+ },
+ {
+ "epoch": 17.427792915531334,
+ "grad_norm": 7.618422985076904,
+ "learning_rate": 1.8927964295480244e-05,
+ "loss": 0.6821,
+ "step": 6396
+ },
+ {
+ "epoch": 17.430517711171664,
+ "grad_norm": 7.5985541343688965,
+ "learning_rate": 1.8927566731864658e-05,
+ "loss": 0.863,
+ "step": 6397
+ },
+ {
+ "epoch": 17.43324250681199,
+ "grad_norm": 6.130251407623291,
+ "learning_rate": 1.8927169098721462e-05,
+ "loss": 0.8207,
+ "step": 6398
+ },
+ {
+ "epoch": 17.435967302452315,
+ "grad_norm": 6.389687538146973,
+ "learning_rate": 1.892677139605377e-05,
+ "loss": 0.8141,
+ "step": 6399
+ },
+ {
+ "epoch": 17.438692098092645,
+ "grad_norm": 6.378605365753174,
+ "learning_rate": 1.892637362386466e-05,
+ "loss": 0.7578,
+ "step": 6400
+ },
+ {
+ "epoch": 17.44141689373297,
+ "grad_norm": 7.127614974975586,
+ "learning_rate": 1.8925975782157243e-05,
+ "loss": 0.7412,
+ "step": 6401
+ },
+ {
+ "epoch": 17.444141689373296,
+ "grad_norm": 5.518735408782959,
+ "learning_rate": 1.8925577870934612e-05,
+ "loss": 0.5993,
+ "step": 6402
+ },
+ {
+ "epoch": 17.446866485013626,
+ "grad_norm": 7.124680042266846,
+ "learning_rate": 1.8925179890199873e-05,
+ "loss": 0.703,
+ "step": 6403
+ },
+ {
+ "epoch": 17.44959128065395,
+ "grad_norm": 6.9113030433654785,
+ "learning_rate": 1.8924781839956116e-05,
+ "loss": 0.9032,
+ "step": 6404
+ },
+ {
+ "epoch": 17.452316076294277,
+ "grad_norm": 6.565731525421143,
+ "learning_rate": 1.8924383720206445e-05,
+ "loss": 0.7692,
+ "step": 6405
+ },
+ {
+ "epoch": 17.455040871934607,
+ "grad_norm": 10.349663734436035,
+ "learning_rate": 1.8923985530953963e-05,
+ "loss": 0.7841,
+ "step": 6406
+ },
+ {
+ "epoch": 17.457765667574932,
+ "grad_norm": 7.195166110992432,
+ "learning_rate": 1.892358727220177e-05,
+ "loss": 0.7512,
+ "step": 6407
+ },
+ {
+ "epoch": 17.460490463215258,
+ "grad_norm": 14.380609512329102,
+ "learning_rate": 1.892318894395296e-05,
+ "loss": 0.8475,
+ "step": 6408
+ },
+ {
+ "epoch": 17.463215258855588,
+ "grad_norm": 5.904404163360596,
+ "learning_rate": 1.892279054621065e-05,
+ "loss": 0.6354,
+ "step": 6409
+ },
+ {
+ "epoch": 17.465940054495913,
+ "grad_norm": 6.776883125305176,
+ "learning_rate": 1.892239207897793e-05,
+ "loss": 0.7165,
+ "step": 6410
+ },
+ {
+ "epoch": 17.46866485013624,
+ "grad_norm": 7.655741214752197,
+ "learning_rate": 1.892199354225791e-05,
+ "loss": 0.7864,
+ "step": 6411
+ },
+ {
+ "epoch": 17.47138964577657,
+ "grad_norm": 6.556282997131348,
+ "learning_rate": 1.8921594936053686e-05,
+ "loss": 0.8648,
+ "step": 6412
+ },
+ {
+ "epoch": 17.474114441416894,
+ "grad_norm": 6.528144836425781,
+ "learning_rate": 1.8921196260368374e-05,
+ "loss": 0.6112,
+ "step": 6413
+ },
+ {
+ "epoch": 17.47683923705722,
+ "grad_norm": 6.337281227111816,
+ "learning_rate": 1.892079751520507e-05,
+ "loss": 0.6017,
+ "step": 6414
+ },
+ {
+ "epoch": 17.479564032697546,
+ "grad_norm": 6.495326995849609,
+ "learning_rate": 1.8920398700566884e-05,
+ "loss": 0.62,
+ "step": 6415
+ },
+ {
+ "epoch": 17.482288828337875,
+ "grad_norm": 6.80182409286499,
+ "learning_rate": 1.8919999816456917e-05,
+ "loss": 0.7388,
+ "step": 6416
+ },
+ {
+ "epoch": 17.4850136239782,
+ "grad_norm": 6.490679740905762,
+ "learning_rate": 1.8919600862878282e-05,
+ "loss": 0.6786,
+ "step": 6417
+ },
+ {
+ "epoch": 17.48773841961853,
+ "grad_norm": 6.740110397338867,
+ "learning_rate": 1.8919201839834078e-05,
+ "loss": 0.8469,
+ "step": 6418
+ },
+ {
+ "epoch": 17.490463215258856,
+ "grad_norm": 6.411724090576172,
+ "learning_rate": 1.891880274732742e-05,
+ "loss": 0.6962,
+ "step": 6419
+ },
+ {
+ "epoch": 17.493188010899182,
+ "grad_norm": 5.537814140319824,
+ "learning_rate": 1.891840358536141e-05,
+ "loss": 0.5286,
+ "step": 6420
+ },
+ {
+ "epoch": 17.495912806539508,
+ "grad_norm": 7.260157108306885,
+ "learning_rate": 1.8918004353939166e-05,
+ "loss": 0.666,
+ "step": 6421
+ },
+ {
+ "epoch": 17.498637602179837,
+ "grad_norm": 8.216387748718262,
+ "learning_rate": 1.891760505306379e-05,
+ "loss": 0.7596,
+ "step": 6422
+ },
+ {
+ "epoch": 17.501362397820163,
+ "grad_norm": 7.431641101837158,
+ "learning_rate": 1.8917205682738387e-05,
+ "loss": 0.7251,
+ "step": 6423
+ },
+ {
+ "epoch": 17.504087193460492,
+ "grad_norm": 6.253106117248535,
+ "learning_rate": 1.891680624296608e-05,
+ "loss": 0.712,
+ "step": 6424
+ },
+ {
+ "epoch": 17.506811989100818,
+ "grad_norm": 7.209530353546143,
+ "learning_rate": 1.891640673374997e-05,
+ "loss": 0.7337,
+ "step": 6425
+ },
+ {
+ "epoch": 17.509536784741144,
+ "grad_norm": 6.101980209350586,
+ "learning_rate": 1.8916007155093168e-05,
+ "loss": 0.692,
+ "step": 6426
+ },
+ {
+ "epoch": 17.51226158038147,
+ "grad_norm": 6.481486797332764,
+ "learning_rate": 1.8915607506998796e-05,
+ "loss": 0.9257,
+ "step": 6427
+ },
+ {
+ "epoch": 17.5149863760218,
+ "grad_norm": 6.46837043762207,
+ "learning_rate": 1.8915207789469954e-05,
+ "loss": 0.7322,
+ "step": 6428
+ },
+ {
+ "epoch": 17.517711171662125,
+ "grad_norm": 7.046135902404785,
+ "learning_rate": 1.891480800250976e-05,
+ "loss": 0.6371,
+ "step": 6429
+ },
+ {
+ "epoch": 17.520435967302454,
+ "grad_norm": 6.586500644683838,
+ "learning_rate": 1.8914408146121333e-05,
+ "loss": 0.7146,
+ "step": 6430
+ },
+ {
+ "epoch": 17.52316076294278,
+ "grad_norm": 6.093857288360596,
+ "learning_rate": 1.891400822030778e-05,
+ "loss": 0.9069,
+ "step": 6431
+ },
+ {
+ "epoch": 17.525885558583106,
+ "grad_norm": 6.9554619789123535,
+ "learning_rate": 1.891360822507222e-05,
+ "loss": 0.7208,
+ "step": 6432
+ },
+ {
+ "epoch": 17.52861035422343,
+ "grad_norm": 5.9927144050598145,
+ "learning_rate": 1.8913208160417763e-05,
+ "loss": 0.74,
+ "step": 6433
+ },
+ {
+ "epoch": 17.53133514986376,
+ "grad_norm": 6.99214506149292,
+ "learning_rate": 1.8912808026347527e-05,
+ "loss": 0.7452,
+ "step": 6434
+ },
+ {
+ "epoch": 17.534059945504087,
+ "grad_norm": 5.745810031890869,
+ "learning_rate": 1.8912407822864632e-05,
+ "loss": 0.6077,
+ "step": 6435
+ },
+ {
+ "epoch": 17.536784741144416,
+ "grad_norm": 7.139151096343994,
+ "learning_rate": 1.891200754997219e-05,
+ "loss": 0.8361,
+ "step": 6436
+ },
+ {
+ "epoch": 17.539509536784742,
+ "grad_norm": 5.1481499671936035,
+ "learning_rate": 1.8911607207673318e-05,
+ "loss": 0.5491,
+ "step": 6437
+ },
+ {
+ "epoch": 17.542234332425068,
+ "grad_norm": 7.99806022644043,
+ "learning_rate": 1.891120679597114e-05,
+ "loss": 0.809,
+ "step": 6438
+ },
+ {
+ "epoch": 17.544959128065393,
+ "grad_norm": 7.969474792480469,
+ "learning_rate": 1.891080631486877e-05,
+ "loss": 0.7459,
+ "step": 6439
+ },
+ {
+ "epoch": 17.547683923705723,
+ "grad_norm": 6.42384147644043,
+ "learning_rate": 1.8910405764369326e-05,
+ "loss": 0.8567,
+ "step": 6440
+ },
+ {
+ "epoch": 17.55040871934605,
+ "grad_norm": 7.031632423400879,
+ "learning_rate": 1.891000514447593e-05,
+ "loss": 0.6246,
+ "step": 6441
+ },
+ {
+ "epoch": 17.553133514986374,
+ "grad_norm": 7.860013008117676,
+ "learning_rate": 1.8909604455191698e-05,
+ "loss": 0.5764,
+ "step": 6442
+ },
+ {
+ "epoch": 17.555858310626704,
+ "grad_norm": 5.868042945861816,
+ "learning_rate": 1.8909203696519753e-05,
+ "loss": 0.6936,
+ "step": 6443
+ },
+ {
+ "epoch": 17.55858310626703,
+ "grad_norm": 7.228658676147461,
+ "learning_rate": 1.890880286846322e-05,
+ "loss": 0.7672,
+ "step": 6444
+ },
+ {
+ "epoch": 17.561307901907355,
+ "grad_norm": 6.019719123840332,
+ "learning_rate": 1.8908401971025214e-05,
+ "loss": 0.6075,
+ "step": 6445
+ },
+ {
+ "epoch": 17.564032697547685,
+ "grad_norm": 6.074864387512207,
+ "learning_rate": 1.890800100420886e-05,
+ "loss": 0.6664,
+ "step": 6446
+ },
+ {
+ "epoch": 17.56675749318801,
+ "grad_norm": 6.469318866729736,
+ "learning_rate": 1.8907599968017284e-05,
+ "loss": 0.7177,
+ "step": 6447
+ },
+ {
+ "epoch": 17.569482288828336,
+ "grad_norm": 6.0272603034973145,
+ "learning_rate": 1.8907198862453604e-05,
+ "loss": 0.91,
+ "step": 6448
+ },
+ {
+ "epoch": 17.572207084468666,
+ "grad_norm": 6.914638042449951,
+ "learning_rate": 1.8906797687520944e-05,
+ "loss": 0.7159,
+ "step": 6449
+ },
+ {
+ "epoch": 17.57493188010899,
+ "grad_norm": 5.88789701461792,
+ "learning_rate": 1.8906396443222436e-05,
+ "loss": 0.7061,
+ "step": 6450
+ },
+ {
+ "epoch": 17.577656675749317,
+ "grad_norm": 7.105113506317139,
+ "learning_rate": 1.8905995129561195e-05,
+ "loss": 0.8513,
+ "step": 6451
+ },
+ {
+ "epoch": 17.580381471389646,
+ "grad_norm": 6.635855674743652,
+ "learning_rate": 1.890559374654035e-05,
+ "loss": 0.7407,
+ "step": 6452
+ },
+ {
+ "epoch": 17.583106267029972,
+ "grad_norm": 6.744505882263184,
+ "learning_rate": 1.8905192294163028e-05,
+ "loss": 0.6859,
+ "step": 6453
+ },
+ {
+ "epoch": 17.585831062670298,
+ "grad_norm": 6.9313554763793945,
+ "learning_rate": 1.8904790772432355e-05,
+ "loss": 0.8226,
+ "step": 6454
+ },
+ {
+ "epoch": 17.588555858310627,
+ "grad_norm": 7.417338848114014,
+ "learning_rate": 1.890438918135146e-05,
+ "loss": 0.8438,
+ "step": 6455
+ },
+ {
+ "epoch": 17.591280653950953,
+ "grad_norm": 6.574675559997559,
+ "learning_rate": 1.8903987520923465e-05,
+ "loss": 0.7068,
+ "step": 6456
+ },
+ {
+ "epoch": 17.59400544959128,
+ "grad_norm": 6.491776943206787,
+ "learning_rate": 1.8903585791151506e-05,
+ "loss": 0.829,
+ "step": 6457
+ },
+ {
+ "epoch": 17.59673024523161,
+ "grad_norm": 7.703031063079834,
+ "learning_rate": 1.8903183992038705e-05,
+ "loss": 0.6719,
+ "step": 6458
+ },
+ {
+ "epoch": 17.599455040871934,
+ "grad_norm": 7.192594528198242,
+ "learning_rate": 1.8902782123588194e-05,
+ "loss": 0.8514,
+ "step": 6459
+ },
+ {
+ "epoch": 17.60217983651226,
+ "grad_norm": 5.557867527008057,
+ "learning_rate": 1.8902380185803103e-05,
+ "loss": 0.6461,
+ "step": 6460
+ },
+ {
+ "epoch": 17.60490463215259,
+ "grad_norm": 5.802346706390381,
+ "learning_rate": 1.8901978178686563e-05,
+ "loss": 0.7916,
+ "step": 6461
+ },
+ {
+ "epoch": 17.607629427792915,
+ "grad_norm": 6.89634370803833,
+ "learning_rate": 1.8901576102241698e-05,
+ "loss": 0.8473,
+ "step": 6462
+ },
+ {
+ "epoch": 17.61035422343324,
+ "grad_norm": 7.122852802276611,
+ "learning_rate": 1.890117395647165e-05,
+ "loss": 0.7355,
+ "step": 6463
+ },
+ {
+ "epoch": 17.61307901907357,
+ "grad_norm": 6.810441493988037,
+ "learning_rate": 1.8900771741379544e-05,
+ "loss": 0.7378,
+ "step": 6464
+ },
+ {
+ "epoch": 17.615803814713896,
+ "grad_norm": 7.492290496826172,
+ "learning_rate": 1.8900369456968513e-05,
+ "loss": 0.774,
+ "step": 6465
+ },
+ {
+ "epoch": 17.618528610354222,
+ "grad_norm": 7.233572006225586,
+ "learning_rate": 1.8899967103241693e-05,
+ "loss": 0.6384,
+ "step": 6466
+ },
+ {
+ "epoch": 17.62125340599455,
+ "grad_norm": 6.744292259216309,
+ "learning_rate": 1.8899564680202215e-05,
+ "loss": 0.8895,
+ "step": 6467
+ },
+ {
+ "epoch": 17.623978201634877,
+ "grad_norm": 6.629166126251221,
+ "learning_rate": 1.8899162187853215e-05,
+ "loss": 0.7294,
+ "step": 6468
+ },
+ {
+ "epoch": 17.626702997275203,
+ "grad_norm": 5.930874347686768,
+ "learning_rate": 1.889875962619782e-05,
+ "loss": 0.701,
+ "step": 6469
+ },
+ {
+ "epoch": 17.629427792915532,
+ "grad_norm": 6.553250789642334,
+ "learning_rate": 1.889835699523918e-05,
+ "loss": 0.7316,
+ "step": 6470
+ },
+ {
+ "epoch": 17.632152588555858,
+ "grad_norm": 7.031999588012695,
+ "learning_rate": 1.889795429498042e-05,
+ "loss": 0.7273,
+ "step": 6471
+ },
+ {
+ "epoch": 17.634877384196184,
+ "grad_norm": 7.214503288269043,
+ "learning_rate": 1.8897551525424674e-05,
+ "loss": 0.7576,
+ "step": 6472
+ },
+ {
+ "epoch": 17.637602179836513,
+ "grad_norm": 5.924612522125244,
+ "learning_rate": 1.889714868657509e-05,
+ "loss": 0.7648,
+ "step": 6473
+ },
+ {
+ "epoch": 17.64032697547684,
+ "grad_norm": 6.036075115203857,
+ "learning_rate": 1.8896745778434793e-05,
+ "loss": 0.6593,
+ "step": 6474
+ },
+ {
+ "epoch": 17.643051771117165,
+ "grad_norm": 5.628846168518066,
+ "learning_rate": 1.8896342801006927e-05,
+ "loss": 0.6788,
+ "step": 6475
+ },
+ {
+ "epoch": 17.645776566757494,
+ "grad_norm": 6.95073127746582,
+ "learning_rate": 1.8895939754294627e-05,
+ "loss": 0.661,
+ "step": 6476
+ },
+ {
+ "epoch": 17.64850136239782,
+ "grad_norm": 5.747248649597168,
+ "learning_rate": 1.889553663830104e-05,
+ "loss": 0.8115,
+ "step": 6477
+ },
+ {
+ "epoch": 17.651226158038146,
+ "grad_norm": 6.813165187835693,
+ "learning_rate": 1.88951334530293e-05,
+ "loss": 0.913,
+ "step": 6478
+ },
+ {
+ "epoch": 17.653950953678475,
+ "grad_norm": 6.562873840332031,
+ "learning_rate": 1.8894730198482543e-05,
+ "loss": 0.8793,
+ "step": 6479
+ },
+ {
+ "epoch": 17.6566757493188,
+ "grad_norm": 6.354843616485596,
+ "learning_rate": 1.889432687466391e-05,
+ "loss": 0.6229,
+ "step": 6480
+ },
+ {
+ "epoch": 17.659400544959126,
+ "grad_norm": 6.961399555206299,
+ "learning_rate": 1.8893923481576555e-05,
+ "loss": 0.655,
+ "step": 6481
+ },
+ {
+ "epoch": 17.662125340599456,
+ "grad_norm": 6.117333889007568,
+ "learning_rate": 1.8893520019223605e-05,
+ "loss": 0.6793,
+ "step": 6482
+ },
+ {
+ "epoch": 17.66485013623978,
+ "grad_norm": 5.8801398277282715,
+ "learning_rate": 1.889311648760821e-05,
+ "loss": 0.9008,
+ "step": 6483
+ },
+ {
+ "epoch": 17.667574931880107,
+ "grad_norm": 5.5251030921936035,
+ "learning_rate": 1.8892712886733506e-05,
+ "loss": 0.9375,
+ "step": 6484
+ },
+ {
+ "epoch": 17.670299727520437,
+ "grad_norm": 6.593271255493164,
+ "learning_rate": 1.8892309216602642e-05,
+ "loss": 0.7653,
+ "step": 6485
+ },
+ {
+ "epoch": 17.673024523160763,
+ "grad_norm": 10.704645156860352,
+ "learning_rate": 1.889190547721876e-05,
+ "loss": 0.8214,
+ "step": 6486
+ },
+ {
+ "epoch": 17.67574931880109,
+ "grad_norm": 7.75901985168457,
+ "learning_rate": 1.8891501668585006e-05,
+ "loss": 0.7301,
+ "step": 6487
+ },
+ {
+ "epoch": 17.678474114441418,
+ "grad_norm": 6.14428186416626,
+ "learning_rate": 1.8891097790704522e-05,
+ "loss": 0.797,
+ "step": 6488
+ },
+ {
+ "epoch": 17.681198910081743,
+ "grad_norm": 6.840229511260986,
+ "learning_rate": 1.8890693843580454e-05,
+ "loss": 0.6873,
+ "step": 6489
+ },
+ {
+ "epoch": 17.68392370572207,
+ "grad_norm": 7.053150177001953,
+ "learning_rate": 1.8890289827215952e-05,
+ "loss": 0.8159,
+ "step": 6490
+ },
+ {
+ "epoch": 17.6866485013624,
+ "grad_norm": 7.654336452484131,
+ "learning_rate": 1.8889885741614153e-05,
+ "loss": 0.8462,
+ "step": 6491
+ },
+ {
+ "epoch": 17.689373297002724,
+ "grad_norm": 7.769216060638428,
+ "learning_rate": 1.8889481586778212e-05,
+ "loss": 0.7961,
+ "step": 6492
+ },
+ {
+ "epoch": 17.69209809264305,
+ "grad_norm": 5.878815174102783,
+ "learning_rate": 1.8889077362711277e-05,
+ "loss": 0.5836,
+ "step": 6493
+ },
+ {
+ "epoch": 17.69482288828338,
+ "grad_norm": 6.14359188079834,
+ "learning_rate": 1.888867306941649e-05,
+ "loss": 0.7972,
+ "step": 6494
+ },
+ {
+ "epoch": 17.697547683923705,
+ "grad_norm": 7.654476642608643,
+ "learning_rate": 1.8888268706897007e-05,
+ "loss": 0.8374,
+ "step": 6495
+ },
+ {
+ "epoch": 17.70027247956403,
+ "grad_norm": 6.5504865646362305,
+ "learning_rate": 1.888786427515597e-05,
+ "loss": 0.6888,
+ "step": 6496
+ },
+ {
+ "epoch": 17.70299727520436,
+ "grad_norm": 7.856649875640869,
+ "learning_rate": 1.8887459774196536e-05,
+ "loss": 0.9766,
+ "step": 6497
+ },
+ {
+ "epoch": 17.705722070844686,
+ "grad_norm": 7.253120422363281,
+ "learning_rate": 1.888705520402185e-05,
+ "loss": 0.6275,
+ "step": 6498
+ },
+ {
+ "epoch": 17.708446866485012,
+ "grad_norm": 6.891391277313232,
+ "learning_rate": 1.888665056463506e-05,
+ "loss": 0.8591,
+ "step": 6499
+ },
+ {
+ "epoch": 17.71117166212534,
+ "grad_norm": 6.038326740264893,
+ "learning_rate": 1.888624585603933e-05,
+ "loss": 0.7386,
+ "step": 6500
+ },
+ {
+ "epoch": 17.713896457765667,
+ "grad_norm": 7.223166465759277,
+ "learning_rate": 1.888584107823779e-05,
+ "loss": 0.7108,
+ "step": 6501
+ },
+ {
+ "epoch": 17.716621253405993,
+ "grad_norm": 6.119983196258545,
+ "learning_rate": 1.8885436231233617e-05,
+ "loss": 0.8743,
+ "step": 6502
+ },
+ {
+ "epoch": 17.719346049046322,
+ "grad_norm": 6.001791477203369,
+ "learning_rate": 1.888503131502995e-05,
+ "loss": 0.7081,
+ "step": 6503
+ },
+ {
+ "epoch": 17.722070844686648,
+ "grad_norm": 5.916830539703369,
+ "learning_rate": 1.8884626329629942e-05,
+ "loss": 1.0344,
+ "step": 6504
+ },
+ {
+ "epoch": 17.724795640326974,
+ "grad_norm": 6.2743120193481445,
+ "learning_rate": 1.8884221275036753e-05,
+ "loss": 0.5215,
+ "step": 6505
+ },
+ {
+ "epoch": 17.727520435967303,
+ "grad_norm": 5.770493030548096,
+ "learning_rate": 1.8883816151253532e-05,
+ "loss": 0.7489,
+ "step": 6506
+ },
+ {
+ "epoch": 17.73024523160763,
+ "grad_norm": 5.62264347076416,
+ "learning_rate": 1.888341095828344e-05,
+ "loss": 0.7727,
+ "step": 6507
+ },
+ {
+ "epoch": 17.732970027247955,
+ "grad_norm": 7.803295612335205,
+ "learning_rate": 1.888300569612963e-05,
+ "loss": 0.7489,
+ "step": 6508
+ },
+ {
+ "epoch": 17.735694822888284,
+ "grad_norm": 6.395860195159912,
+ "learning_rate": 1.8882600364795255e-05,
+ "loss": 0.9067,
+ "step": 6509
+ },
+ {
+ "epoch": 17.73841961852861,
+ "grad_norm": 6.857898712158203,
+ "learning_rate": 1.8882194964283473e-05,
+ "loss": 0.6097,
+ "step": 6510
+ },
+ {
+ "epoch": 17.741144414168936,
+ "grad_norm": 6.514215469360352,
+ "learning_rate": 1.888178949459745e-05,
+ "loss": 0.5797,
+ "step": 6511
+ },
+ {
+ "epoch": 17.743869209809265,
+ "grad_norm": 8.13459300994873,
+ "learning_rate": 1.8881383955740324e-05,
+ "loss": 0.7632,
+ "step": 6512
+ },
+ {
+ "epoch": 17.74659400544959,
+ "grad_norm": 6.359437465667725,
+ "learning_rate": 1.8880978347715272e-05,
+ "loss": 0.605,
+ "step": 6513
+ },
+ {
+ "epoch": 17.749318801089917,
+ "grad_norm": 7.551234245300293,
+ "learning_rate": 1.888057267052545e-05,
+ "loss": 0.7133,
+ "step": 6514
+ },
+ {
+ "epoch": 17.752043596730246,
+ "grad_norm": 6.648248672485352,
+ "learning_rate": 1.888016692417401e-05,
+ "loss": 0.7493,
+ "step": 6515
+ },
+ {
+ "epoch": 17.754768392370572,
+ "grad_norm": 8.08946418762207,
+ "learning_rate": 1.8879761108664114e-05,
+ "loss": 0.6606,
+ "step": 6516
+ },
+ {
+ "epoch": 17.757493188010898,
+ "grad_norm": 7.416854381561279,
+ "learning_rate": 1.8879355223998928e-05,
+ "loss": 0.6082,
+ "step": 6517
+ },
+ {
+ "epoch": 17.760217983651227,
+ "grad_norm": 5.944798469543457,
+ "learning_rate": 1.8878949270181606e-05,
+ "loss": 0.7454,
+ "step": 6518
+ },
+ {
+ "epoch": 17.762942779291553,
+ "grad_norm": 7.338103771209717,
+ "learning_rate": 1.8878543247215314e-05,
+ "loss": 0.8257,
+ "step": 6519
+ },
+ {
+ "epoch": 17.76566757493188,
+ "grad_norm": 6.079237461090088,
+ "learning_rate": 1.8878137155103214e-05,
+ "loss": 0.5762,
+ "step": 6520
+ },
+ {
+ "epoch": 17.768392370572208,
+ "grad_norm": 6.191689491271973,
+ "learning_rate": 1.887773099384847e-05,
+ "loss": 0.6746,
+ "step": 6521
+ },
+ {
+ "epoch": 17.771117166212534,
+ "grad_norm": 6.357407093048096,
+ "learning_rate": 1.887732476345424e-05,
+ "loss": 0.923,
+ "step": 6522
+ },
+ {
+ "epoch": 17.77384196185286,
+ "grad_norm": 7.965139389038086,
+ "learning_rate": 1.887691846392369e-05,
+ "loss": 0.6908,
+ "step": 6523
+ },
+ {
+ "epoch": 17.77656675749319,
+ "grad_norm": 6.2291741371154785,
+ "learning_rate": 1.8876512095259986e-05,
+ "loss": 0.6198,
+ "step": 6524
+ },
+ {
+ "epoch": 17.779291553133515,
+ "grad_norm": 6.5397257804870605,
+ "learning_rate": 1.8876105657466287e-05,
+ "loss": 0.6146,
+ "step": 6525
+ },
+ {
+ "epoch": 17.78201634877384,
+ "grad_norm": 7.39561653137207,
+ "learning_rate": 1.8875699150545768e-05,
+ "loss": 0.7688,
+ "step": 6526
+ },
+ {
+ "epoch": 17.78474114441417,
+ "grad_norm": 6.516382694244385,
+ "learning_rate": 1.887529257450159e-05,
+ "loss": 0.7908,
+ "step": 6527
+ },
+ {
+ "epoch": 17.787465940054496,
+ "grad_norm": 6.62148380279541,
+ "learning_rate": 1.8874885929336916e-05,
+ "loss": 0.749,
+ "step": 6528
+ },
+ {
+ "epoch": 17.79019073569482,
+ "grad_norm": 6.113258361816406,
+ "learning_rate": 1.8874479215054916e-05,
+ "loss": 0.5889,
+ "step": 6529
+ },
+ {
+ "epoch": 17.79291553133515,
+ "grad_norm": 6.964879512786865,
+ "learning_rate": 1.887407243165876e-05,
+ "loss": 0.7944,
+ "step": 6530
+ },
+ {
+ "epoch": 17.795640326975477,
+ "grad_norm": 5.306907653808594,
+ "learning_rate": 1.887366557915161e-05,
+ "loss": 0.9553,
+ "step": 6531
+ },
+ {
+ "epoch": 17.798365122615802,
+ "grad_norm": 6.924889087677002,
+ "learning_rate": 1.8873258657536643e-05,
+ "loss": 0.5768,
+ "step": 6532
+ },
+ {
+ "epoch": 17.80108991825613,
+ "grad_norm": 7.970389366149902,
+ "learning_rate": 1.8872851666817017e-05,
+ "loss": 0.7852,
+ "step": 6533
+ },
+ {
+ "epoch": 17.803814713896458,
+ "grad_norm": 7.1712164878845215,
+ "learning_rate": 1.8872444606995915e-05,
+ "loss": 0.7311,
+ "step": 6534
+ },
+ {
+ "epoch": 17.806539509536783,
+ "grad_norm": 5.594740390777588,
+ "learning_rate": 1.8872037478076494e-05,
+ "loss": 0.8818,
+ "step": 6535
+ },
+ {
+ "epoch": 17.809264305177113,
+ "grad_norm": 6.725142478942871,
+ "learning_rate": 1.8871630280061933e-05,
+ "loss": 0.7449,
+ "step": 6536
+ },
+ {
+ "epoch": 17.81198910081744,
+ "grad_norm": 6.123635768890381,
+ "learning_rate": 1.8871223012955402e-05,
+ "loss": 0.5738,
+ "step": 6537
+ },
+ {
+ "epoch": 17.814713896457764,
+ "grad_norm": 6.182042598724365,
+ "learning_rate": 1.887081567676007e-05,
+ "loss": 0.5588,
+ "step": 6538
+ },
+ {
+ "epoch": 17.817438692098094,
+ "grad_norm": 7.886967658996582,
+ "learning_rate": 1.887040827147911e-05,
+ "loss": 0.8085,
+ "step": 6539
+ },
+ {
+ "epoch": 17.82016348773842,
+ "grad_norm": 7.686052322387695,
+ "learning_rate": 1.88700007971157e-05,
+ "loss": 0.6774,
+ "step": 6540
+ },
+ {
+ "epoch": 17.822888283378745,
+ "grad_norm": 8.700678825378418,
+ "learning_rate": 1.8869593253673005e-05,
+ "loss": 0.6626,
+ "step": 6541
+ },
+ {
+ "epoch": 17.825613079019075,
+ "grad_norm": 8.628596305847168,
+ "learning_rate": 1.8869185641154204e-05,
+ "loss": 0.7817,
+ "step": 6542
+ },
+ {
+ "epoch": 17.8283378746594,
+ "grad_norm": 6.66777229309082,
+ "learning_rate": 1.8868777959562472e-05,
+ "loss": 0.9121,
+ "step": 6543
+ },
+ {
+ "epoch": 17.831062670299726,
+ "grad_norm": 7.844986438751221,
+ "learning_rate": 1.886837020890098e-05,
+ "loss": 0.7449,
+ "step": 6544
+ },
+ {
+ "epoch": 17.833787465940055,
+ "grad_norm": 6.333627700805664,
+ "learning_rate": 1.8867962389172912e-05,
+ "loss": 0.6987,
+ "step": 6545
+ },
+ {
+ "epoch": 17.83651226158038,
+ "grad_norm": 6.033593654632568,
+ "learning_rate": 1.8867554500381433e-05,
+ "loss": 0.6743,
+ "step": 6546
+ },
+ {
+ "epoch": 17.839237057220707,
+ "grad_norm": 7.763118743896484,
+ "learning_rate": 1.886714654252973e-05,
+ "loss": 0.4988,
+ "step": 6547
+ },
+ {
+ "epoch": 17.841961852861036,
+ "grad_norm": 5.281829833984375,
+ "learning_rate": 1.886673851562097e-05,
+ "loss": 0.6134,
+ "step": 6548
+ },
+ {
+ "epoch": 17.844686648501362,
+ "grad_norm": 7.660382270812988,
+ "learning_rate": 1.8866330419658338e-05,
+ "loss": 0.8232,
+ "step": 6549
+ },
+ {
+ "epoch": 17.847411444141688,
+ "grad_norm": 6.185505390167236,
+ "learning_rate": 1.8865922254645008e-05,
+ "loss": 0.8203,
+ "step": 6550
+ },
+ {
+ "epoch": 17.850136239782017,
+ "grad_norm": 6.84859037399292,
+ "learning_rate": 1.8865514020584164e-05,
+ "loss": 0.8129,
+ "step": 6551
+ },
+ {
+ "epoch": 17.852861035422343,
+ "grad_norm": 5.90626335144043,
+ "learning_rate": 1.886510571747898e-05,
+ "loss": 0.8241,
+ "step": 6552
+ },
+ {
+ "epoch": 17.85558583106267,
+ "grad_norm": 6.755615234375,
+ "learning_rate": 1.886469734533264e-05,
+ "loss": 0.5665,
+ "step": 6553
+ },
+ {
+ "epoch": 17.858310626703,
+ "grad_norm": 6.548464775085449,
+ "learning_rate": 1.886428890414832e-05,
+ "loss": 0.6842,
+ "step": 6554
+ },
+ {
+ "epoch": 17.861035422343324,
+ "grad_norm": 14.509011268615723,
+ "learning_rate": 1.8863880393929202e-05,
+ "loss": 0.658,
+ "step": 6555
+ },
+ {
+ "epoch": 17.86376021798365,
+ "grad_norm": 10.149523735046387,
+ "learning_rate": 1.8863471814678475e-05,
+ "loss": 0.7458,
+ "step": 6556
+ },
+ {
+ "epoch": 17.86648501362398,
+ "grad_norm": 9.879592895507812,
+ "learning_rate": 1.886306316639931e-05,
+ "loss": 0.7416,
+ "step": 6557
+ },
+ {
+ "epoch": 17.869209809264305,
+ "grad_norm": 7.519158363342285,
+ "learning_rate": 1.8862654449094892e-05,
+ "loss": 0.707,
+ "step": 6558
+ },
+ {
+ "epoch": 17.87193460490463,
+ "grad_norm": 6.413763999938965,
+ "learning_rate": 1.886224566276841e-05,
+ "loss": 0.5567,
+ "step": 6559
+ },
+ {
+ "epoch": 17.87465940054496,
+ "grad_norm": 6.773137092590332,
+ "learning_rate": 1.8861836807423045e-05,
+ "loss": 0.6744,
+ "step": 6560
+ },
+ {
+ "epoch": 17.877384196185286,
+ "grad_norm": 6.8968963623046875,
+ "learning_rate": 1.8861427883061977e-05,
+ "loss": 0.7505,
+ "step": 6561
+ },
+ {
+ "epoch": 17.88010899182561,
+ "grad_norm": 6.040518760681152,
+ "learning_rate": 1.8861018889688395e-05,
+ "loss": 0.5392,
+ "step": 6562
+ },
+ {
+ "epoch": 17.88283378746594,
+ "grad_norm": 7.514745235443115,
+ "learning_rate": 1.8860609827305487e-05,
+ "loss": 0.8495,
+ "step": 6563
+ },
+ {
+ "epoch": 17.885558583106267,
+ "grad_norm": 7.569820404052734,
+ "learning_rate": 1.886020069591643e-05,
+ "loss": 0.8439,
+ "step": 6564
+ },
+ {
+ "epoch": 17.888283378746593,
+ "grad_norm": 6.754626274108887,
+ "learning_rate": 1.885979149552442e-05,
+ "loss": 0.814,
+ "step": 6565
+ },
+ {
+ "epoch": 17.891008174386922,
+ "grad_norm": 6.4310221672058105,
+ "learning_rate": 1.8859382226132635e-05,
+ "loss": 0.7659,
+ "step": 6566
+ },
+ {
+ "epoch": 17.893732970027248,
+ "grad_norm": 8.358661651611328,
+ "learning_rate": 1.8858972887744268e-05,
+ "loss": 0.7772,
+ "step": 6567
+ },
+ {
+ "epoch": 17.896457765667574,
+ "grad_norm": 6.831119537353516,
+ "learning_rate": 1.8858563480362503e-05,
+ "loss": 0.7501,
+ "step": 6568
+ },
+ {
+ "epoch": 17.899182561307903,
+ "grad_norm": 6.5615363121032715,
+ "learning_rate": 1.885815400399053e-05,
+ "loss": 0.6976,
+ "step": 6569
+ },
+ {
+ "epoch": 17.90190735694823,
+ "grad_norm": 6.125516891479492,
+ "learning_rate": 1.885774445863154e-05,
+ "loss": 0.7427,
+ "step": 6570
+ },
+ {
+ "epoch": 17.904632152588555,
+ "grad_norm": 6.838286399841309,
+ "learning_rate": 1.8857334844288723e-05,
+ "loss": 0.741,
+ "step": 6571
+ },
+ {
+ "epoch": 17.907356948228884,
+ "grad_norm": 5.889449596405029,
+ "learning_rate": 1.8856925160965267e-05,
+ "loss": 0.7582,
+ "step": 6572
+ },
+ {
+ "epoch": 17.91008174386921,
+ "grad_norm": 7.025384426116943,
+ "learning_rate": 1.8856515408664363e-05,
+ "loss": 0.7183,
+ "step": 6573
+ },
+ {
+ "epoch": 17.912806539509535,
+ "grad_norm": 7.2751545906066895,
+ "learning_rate": 1.88561055873892e-05,
+ "loss": 0.8398,
+ "step": 6574
+ },
+ {
+ "epoch": 17.915531335149865,
+ "grad_norm": 6.409760475158691,
+ "learning_rate": 1.8855695697142972e-05,
+ "loss": 0.8127,
+ "step": 6575
+ },
+ {
+ "epoch": 17.91825613079019,
+ "grad_norm": 7.040431499481201,
+ "learning_rate": 1.8855285737928874e-05,
+ "loss": 0.937,
+ "step": 6576
+ },
+ {
+ "epoch": 17.920980926430516,
+ "grad_norm": 5.643787860870361,
+ "learning_rate": 1.8854875709750092e-05,
+ "loss": 0.7179,
+ "step": 6577
+ },
+ {
+ "epoch": 17.923705722070846,
+ "grad_norm": 5.96733283996582,
+ "learning_rate": 1.8854465612609828e-05,
+ "loss": 0.7644,
+ "step": 6578
+ },
+ {
+ "epoch": 17.92643051771117,
+ "grad_norm": 7.184720039367676,
+ "learning_rate": 1.8854055446511267e-05,
+ "loss": 0.5235,
+ "step": 6579
+ },
+ {
+ "epoch": 17.929155313351497,
+ "grad_norm": 6.1996169090271,
+ "learning_rate": 1.8853645211457606e-05,
+ "loss": 0.7499,
+ "step": 6580
+ },
+ {
+ "epoch": 17.931880108991827,
+ "grad_norm": 5.501131057739258,
+ "learning_rate": 1.8853234907452042e-05,
+ "loss": 0.8622,
+ "step": 6581
+ },
+ {
+ "epoch": 17.934604904632153,
+ "grad_norm": 6.992715358734131,
+ "learning_rate": 1.8852824534497772e-05,
+ "loss": 0.7018,
+ "step": 6582
+ },
+ {
+ "epoch": 17.93732970027248,
+ "grad_norm": 10.29998779296875,
+ "learning_rate": 1.8852414092597985e-05,
+ "loss": 0.628,
+ "step": 6583
+ },
+ {
+ "epoch": 17.940054495912808,
+ "grad_norm": 6.426336765289307,
+ "learning_rate": 1.8852003581755886e-05,
+ "loss": 0.7931,
+ "step": 6584
+ },
+ {
+ "epoch": 17.942779291553133,
+ "grad_norm": 6.999417304992676,
+ "learning_rate": 1.8851593001974663e-05,
+ "loss": 0.611,
+ "step": 6585
+ },
+ {
+ "epoch": 17.94550408719346,
+ "grad_norm": 6.1762166023254395,
+ "learning_rate": 1.8851182353257524e-05,
+ "loss": 0.8945,
+ "step": 6586
+ },
+ {
+ "epoch": 17.94822888283379,
+ "grad_norm": 6.045926094055176,
+ "learning_rate": 1.885077163560766e-05,
+ "loss": 0.5651,
+ "step": 6587
+ },
+ {
+ "epoch": 17.950953678474114,
+ "grad_norm": 9.832141876220703,
+ "learning_rate": 1.885036084902827e-05,
+ "loss": 0.7238,
+ "step": 6588
+ },
+ {
+ "epoch": 17.95367847411444,
+ "grad_norm": 7.618719577789307,
+ "learning_rate": 1.8849949993522557e-05,
+ "loss": 0.7789,
+ "step": 6589
+ },
+ {
+ "epoch": 17.95640326975477,
+ "grad_norm": 6.402445316314697,
+ "learning_rate": 1.8849539069093717e-05,
+ "loss": 0.7103,
+ "step": 6590
+ },
+ {
+ "epoch": 17.959128065395095,
+ "grad_norm": 5.970659255981445,
+ "learning_rate": 1.884912807574495e-05,
+ "loss": 0.7883,
+ "step": 6591
+ },
+ {
+ "epoch": 17.96185286103542,
+ "grad_norm": 6.528757095336914,
+ "learning_rate": 1.884871701347946e-05,
+ "loss": 0.7683,
+ "step": 6592
+ },
+ {
+ "epoch": 17.96457765667575,
+ "grad_norm": 7.1998186111450195,
+ "learning_rate": 1.884830588230045e-05,
+ "loss": 0.7384,
+ "step": 6593
+ },
+ {
+ "epoch": 17.967302452316076,
+ "grad_norm": 5.9098429679870605,
+ "learning_rate": 1.8847894682211113e-05,
+ "loss": 0.7875,
+ "step": 6594
+ },
+ {
+ "epoch": 17.970027247956402,
+ "grad_norm": 6.804289817810059,
+ "learning_rate": 1.884748341321466e-05,
+ "loss": 0.7573,
+ "step": 6595
+ },
+ {
+ "epoch": 17.97275204359673,
+ "grad_norm": 7.005715847015381,
+ "learning_rate": 1.8847072075314288e-05,
+ "loss": 0.7312,
+ "step": 6596
+ },
+ {
+ "epoch": 17.975476839237057,
+ "grad_norm": 5.764132499694824,
+ "learning_rate": 1.8846660668513206e-05,
+ "loss": 0.6249,
+ "step": 6597
+ },
+ {
+ "epoch": 17.978201634877383,
+ "grad_norm": 7.86998176574707,
+ "learning_rate": 1.8846249192814613e-05,
+ "loss": 0.7675,
+ "step": 6598
+ },
+ {
+ "epoch": 17.980926430517712,
+ "grad_norm": 6.480493068695068,
+ "learning_rate": 1.884583764822172e-05,
+ "loss": 0.6157,
+ "step": 6599
+ },
+ {
+ "epoch": 17.983651226158038,
+ "grad_norm": 7.017398834228516,
+ "learning_rate": 1.8845426034737724e-05,
+ "loss": 0.6484,
+ "step": 6600
+ },
+ {
+ "epoch": 17.986376021798364,
+ "grad_norm": 7.206132411956787,
+ "learning_rate": 1.8845014352365836e-05,
+ "loss": 0.6311,
+ "step": 6601
+ },
+ {
+ "epoch": 17.989100817438693,
+ "grad_norm": 7.665340900421143,
+ "learning_rate": 1.8844602601109265e-05,
+ "loss": 0.7877,
+ "step": 6602
+ },
+ {
+ "epoch": 17.99182561307902,
+ "grad_norm": 7.438289642333984,
+ "learning_rate": 1.884419078097121e-05,
+ "loss": 0.7581,
+ "step": 6603
+ },
+ {
+ "epoch": 17.994550408719345,
+ "grad_norm": 5.898067951202393,
+ "learning_rate": 1.8843778891954883e-05,
+ "loss": 0.6417,
+ "step": 6604
+ },
+ {
+ "epoch": 17.997275204359674,
+ "grad_norm": 7.426933288574219,
+ "learning_rate": 1.884336693406349e-05,
+ "loss": 0.8036,
+ "step": 6605
+ },
+ {
+ "epoch": 18.0,
+ "grad_norm": 7.811229228973389,
+ "learning_rate": 1.8842954907300236e-05,
+ "loss": 0.9031,
+ "step": 6606
+ },
+ {
+ "epoch": 18.002724795640326,
+ "grad_norm": 6.531326770782471,
+ "learning_rate": 1.884254281166834e-05,
+ "loss": 0.6298,
+ "step": 6607
+ },
+ {
+ "epoch": 18.005449591280655,
+ "grad_norm": 6.042975902557373,
+ "learning_rate": 1.8842130647171003e-05,
+ "loss": 0.5524,
+ "step": 6608
+ },
+ {
+ "epoch": 18.00817438692098,
+ "grad_norm": 8.079333305358887,
+ "learning_rate": 1.8841718413811433e-05,
+ "loss": 0.7737,
+ "step": 6609
+ },
+ {
+ "epoch": 18.010899182561307,
+ "grad_norm": 6.4736809730529785,
+ "learning_rate": 1.884130611159285e-05,
+ "loss": 0.7003,
+ "step": 6610
+ },
+ {
+ "epoch": 18.013623978201636,
+ "grad_norm": 6.99594783782959,
+ "learning_rate": 1.8840893740518458e-05,
+ "loss": 0.6761,
+ "step": 6611
+ },
+ {
+ "epoch": 18.016348773841962,
+ "grad_norm": 6.746195316314697,
+ "learning_rate": 1.8840481300591473e-05,
+ "loss": 0.5439,
+ "step": 6612
+ },
+ {
+ "epoch": 18.019073569482288,
+ "grad_norm": 5.2232136726379395,
+ "learning_rate": 1.88400687918151e-05,
+ "loss": 0.9193,
+ "step": 6613
+ },
+ {
+ "epoch": 18.021798365122617,
+ "grad_norm": 6.719673156738281,
+ "learning_rate": 1.8839656214192557e-05,
+ "loss": 0.5704,
+ "step": 6614
+ },
+ {
+ "epoch": 18.024523160762943,
+ "grad_norm": 6.533470630645752,
+ "learning_rate": 1.8839243567727053e-05,
+ "loss": 0.7123,
+ "step": 6615
+ },
+ {
+ "epoch": 18.02724795640327,
+ "grad_norm": 6.208836555480957,
+ "learning_rate": 1.8838830852421807e-05,
+ "loss": 0.6174,
+ "step": 6616
+ },
+ {
+ "epoch": 18.029972752043598,
+ "grad_norm": 5.640427589416504,
+ "learning_rate": 1.8838418068280032e-05,
+ "loss": 0.6174,
+ "step": 6617
+ },
+ {
+ "epoch": 18.032697547683924,
+ "grad_norm": 6.88861083984375,
+ "learning_rate": 1.883800521530494e-05,
+ "loss": 0.6836,
+ "step": 6618
+ },
+ {
+ "epoch": 18.03542234332425,
+ "grad_norm": 7.165849208831787,
+ "learning_rate": 1.8837592293499747e-05,
+ "loss": 0.7393,
+ "step": 6619
+ },
+ {
+ "epoch": 18.03814713896458,
+ "grad_norm": 6.708693981170654,
+ "learning_rate": 1.883717930286767e-05,
+ "loss": 0.6758,
+ "step": 6620
+ },
+ {
+ "epoch": 18.040871934604905,
+ "grad_norm": 7.733780384063721,
+ "learning_rate": 1.8836766243411932e-05,
+ "loss": 0.691,
+ "step": 6621
+ },
+ {
+ "epoch": 18.04359673024523,
+ "grad_norm": 5.742239475250244,
+ "learning_rate": 1.8836353115135735e-05,
+ "loss": 0.5908,
+ "step": 6622
+ },
+ {
+ "epoch": 18.04632152588556,
+ "grad_norm": 6.73600959777832,
+ "learning_rate": 1.8835939918042308e-05,
+ "loss": 0.6534,
+ "step": 6623
+ },
+ {
+ "epoch": 18.049046321525886,
+ "grad_norm": 6.1416730880737305,
+ "learning_rate": 1.883552665213486e-05,
+ "loss": 0.6978,
+ "step": 6624
+ },
+ {
+ "epoch": 18.05177111716621,
+ "grad_norm": 6.578459739685059,
+ "learning_rate": 1.883511331741662e-05,
+ "loss": 0.6556,
+ "step": 6625
+ },
+ {
+ "epoch": 18.05449591280654,
+ "grad_norm": 5.637022972106934,
+ "learning_rate": 1.88346999138908e-05,
+ "loss": 0.6133,
+ "step": 6626
+ },
+ {
+ "epoch": 18.057220708446867,
+ "grad_norm": 6.380706310272217,
+ "learning_rate": 1.8834286441560622e-05,
+ "loss": 0.6968,
+ "step": 6627
+ },
+ {
+ "epoch": 18.059945504087192,
+ "grad_norm": 6.247730731964111,
+ "learning_rate": 1.8833872900429307e-05,
+ "loss": 0.8238,
+ "step": 6628
+ },
+ {
+ "epoch": 18.06267029972752,
+ "grad_norm": 5.7287702560424805,
+ "learning_rate": 1.8833459290500073e-05,
+ "loss": 0.5812,
+ "step": 6629
+ },
+ {
+ "epoch": 18.065395095367847,
+ "grad_norm": 7.37725830078125,
+ "learning_rate": 1.8833045611776143e-05,
+ "loss": 0.7969,
+ "step": 6630
+ },
+ {
+ "epoch": 18.068119891008173,
+ "grad_norm": 6.631471633911133,
+ "learning_rate": 1.883263186426073e-05,
+ "loss": 0.6853,
+ "step": 6631
+ },
+ {
+ "epoch": 18.070844686648503,
+ "grad_norm": 6.1050639152526855,
+ "learning_rate": 1.8832218047957072e-05,
+ "loss": 0.6076,
+ "step": 6632
+ },
+ {
+ "epoch": 18.07356948228883,
+ "grad_norm": 6.923410892486572,
+ "learning_rate": 1.883180416286839e-05,
+ "loss": 0.7183,
+ "step": 6633
+ },
+ {
+ "epoch": 18.076294277929154,
+ "grad_norm": 7.432962894439697,
+ "learning_rate": 1.8831390208997892e-05,
+ "loss": 0.5341,
+ "step": 6634
+ },
+ {
+ "epoch": 18.079019073569484,
+ "grad_norm": 6.332204818725586,
+ "learning_rate": 1.883097618634881e-05,
+ "loss": 0.577,
+ "step": 6635
+ },
+ {
+ "epoch": 18.08174386920981,
+ "grad_norm": 5.646083831787109,
+ "learning_rate": 1.8830562094924375e-05,
+ "loss": 0.6365,
+ "step": 6636
+ },
+ {
+ "epoch": 18.084468664850135,
+ "grad_norm": 6.705163955688477,
+ "learning_rate": 1.8830147934727806e-05,
+ "loss": 0.6218,
+ "step": 6637
+ },
+ {
+ "epoch": 18.087193460490465,
+ "grad_norm": 7.536165237426758,
+ "learning_rate": 1.882973370576233e-05,
+ "loss": 0.7391,
+ "step": 6638
+ },
+ {
+ "epoch": 18.08991825613079,
+ "grad_norm": 6.288671016693115,
+ "learning_rate": 1.882931940803117e-05,
+ "loss": 0.7083,
+ "step": 6639
+ },
+ {
+ "epoch": 18.092643051771116,
+ "grad_norm": 5.878170967102051,
+ "learning_rate": 1.8828905041537556e-05,
+ "loss": 0.6467,
+ "step": 6640
+ },
+ {
+ "epoch": 18.095367847411445,
+ "grad_norm": 6.804635524749756,
+ "learning_rate": 1.882849060628471e-05,
+ "loss": 0.6221,
+ "step": 6641
+ },
+ {
+ "epoch": 18.09809264305177,
+ "grad_norm": 6.529529094696045,
+ "learning_rate": 1.882807610227587e-05,
+ "loss": 0.7373,
+ "step": 6642
+ },
+ {
+ "epoch": 18.100817438692097,
+ "grad_norm": 7.237727642059326,
+ "learning_rate": 1.882766152951425e-05,
+ "loss": 0.7006,
+ "step": 6643
+ },
+ {
+ "epoch": 18.103542234332426,
+ "grad_norm": 6.628530025482178,
+ "learning_rate": 1.882724688800309e-05,
+ "loss": 0.7601,
+ "step": 6644
+ },
+ {
+ "epoch": 18.106267029972752,
+ "grad_norm": 6.423205852508545,
+ "learning_rate": 1.8826832177745617e-05,
+ "loss": 0.7941,
+ "step": 6645
+ },
+ {
+ "epoch": 18.108991825613078,
+ "grad_norm": 5.305749416351318,
+ "learning_rate": 1.882641739874506e-05,
+ "loss": 0.7411,
+ "step": 6646
+ },
+ {
+ "epoch": 18.111716621253407,
+ "grad_norm": 7.452727794647217,
+ "learning_rate": 1.882600255100464e-05,
+ "loss": 0.6425,
+ "step": 6647
+ },
+ {
+ "epoch": 18.114441416893733,
+ "grad_norm": 6.141794204711914,
+ "learning_rate": 1.8825587634527603e-05,
+ "loss": 0.717,
+ "step": 6648
+ },
+ {
+ "epoch": 18.11716621253406,
+ "grad_norm": 6.0949296951293945,
+ "learning_rate": 1.8825172649317172e-05,
+ "loss": 0.7033,
+ "step": 6649
+ },
+ {
+ "epoch": 18.11989100817439,
+ "grad_norm": 6.7433247566223145,
+ "learning_rate": 1.882475759537658e-05,
+ "loss": 0.7313,
+ "step": 6650
+ },
+ {
+ "epoch": 18.122615803814714,
+ "grad_norm": 5.8499298095703125,
+ "learning_rate": 1.882434247270906e-05,
+ "loss": 0.693,
+ "step": 6651
+ },
+ {
+ "epoch": 18.12534059945504,
+ "grad_norm": 5.276959419250488,
+ "learning_rate": 1.8823927281317848e-05,
+ "loss": 0.7653,
+ "step": 6652
+ },
+ {
+ "epoch": 18.12806539509537,
+ "grad_norm": 5.192943572998047,
+ "learning_rate": 1.882351202120617e-05,
+ "loss": 0.5162,
+ "step": 6653
+ },
+ {
+ "epoch": 18.130790190735695,
+ "grad_norm": 5.950687408447266,
+ "learning_rate": 1.882309669237727e-05,
+ "loss": 0.6361,
+ "step": 6654
+ },
+ {
+ "epoch": 18.13351498637602,
+ "grad_norm": 5.997944355010986,
+ "learning_rate": 1.882268129483437e-05,
+ "loss": 0.5464,
+ "step": 6655
+ },
+ {
+ "epoch": 18.13623978201635,
+ "grad_norm": 6.4330596923828125,
+ "learning_rate": 1.8822265828580716e-05,
+ "loss": 0.6426,
+ "step": 6656
+ },
+ {
+ "epoch": 18.138964577656676,
+ "grad_norm": 5.502564430236816,
+ "learning_rate": 1.8821850293619536e-05,
+ "loss": 0.7952,
+ "step": 6657
+ },
+ {
+ "epoch": 18.141689373297,
+ "grad_norm": 6.618991374969482,
+ "learning_rate": 1.8821434689954074e-05,
+ "loss": 0.6271,
+ "step": 6658
+ },
+ {
+ "epoch": 18.14441416893733,
+ "grad_norm": 6.477290153503418,
+ "learning_rate": 1.8821019017587562e-05,
+ "loss": 0.6677,
+ "step": 6659
+ },
+ {
+ "epoch": 18.147138964577657,
+ "grad_norm": 6.304305553436279,
+ "learning_rate": 1.882060327652324e-05,
+ "loss": 0.7767,
+ "step": 6660
+ },
+ {
+ "epoch": 18.149863760217983,
+ "grad_norm": 8.49791431427002,
+ "learning_rate": 1.8820187466764337e-05,
+ "loss": 0.6599,
+ "step": 6661
+ },
+ {
+ "epoch": 18.152588555858312,
+ "grad_norm": 9.429057121276855,
+ "learning_rate": 1.88197715883141e-05,
+ "loss": 0.6262,
+ "step": 6662
+ },
+ {
+ "epoch": 18.155313351498638,
+ "grad_norm": 7.741325378417969,
+ "learning_rate": 1.8819355641175767e-05,
+ "loss": 0.6376,
+ "step": 6663
+ },
+ {
+ "epoch": 18.158038147138964,
+ "grad_norm": 6.712416172027588,
+ "learning_rate": 1.8818939625352577e-05,
+ "loss": 0.5126,
+ "step": 6664
+ },
+ {
+ "epoch": 18.160762942779293,
+ "grad_norm": 5.416841506958008,
+ "learning_rate": 1.8818523540847768e-05,
+ "loss": 0.5785,
+ "step": 6665
+ },
+ {
+ "epoch": 18.16348773841962,
+ "grad_norm": 6.969899654388428,
+ "learning_rate": 1.8818107387664582e-05,
+ "loss": 0.5327,
+ "step": 6666
+ },
+ {
+ "epoch": 18.166212534059945,
+ "grad_norm": 6.303111553192139,
+ "learning_rate": 1.881769116580626e-05,
+ "loss": 0.6647,
+ "step": 6667
+ },
+ {
+ "epoch": 18.168937329700274,
+ "grad_norm": 6.621161460876465,
+ "learning_rate": 1.881727487527604e-05,
+ "loss": 0.7457,
+ "step": 6668
+ },
+ {
+ "epoch": 18.1716621253406,
+ "grad_norm": 6.522156238555908,
+ "learning_rate": 1.881685851607717e-05,
+ "loss": 0.661,
+ "step": 6669
+ },
+ {
+ "epoch": 18.174386920980925,
+ "grad_norm": 7.653639316558838,
+ "learning_rate": 1.881644208821289e-05,
+ "loss": 0.515,
+ "step": 6670
+ },
+ {
+ "epoch": 18.177111716621255,
+ "grad_norm": 9.39794921875,
+ "learning_rate": 1.881602559168644e-05,
+ "loss": 0.8864,
+ "step": 6671
+ },
+ {
+ "epoch": 18.17983651226158,
+ "grad_norm": 6.138011932373047,
+ "learning_rate": 1.881560902650107e-05,
+ "loss": 0.6152,
+ "step": 6672
+ },
+ {
+ "epoch": 18.182561307901906,
+ "grad_norm": 6.383389949798584,
+ "learning_rate": 1.881519239266002e-05,
+ "loss": 0.838,
+ "step": 6673
+ },
+ {
+ "epoch": 18.185286103542236,
+ "grad_norm": 6.732948303222656,
+ "learning_rate": 1.8814775690166535e-05,
+ "loss": 0.6424,
+ "step": 6674
+ },
+ {
+ "epoch": 18.18801089918256,
+ "grad_norm": 5.938075065612793,
+ "learning_rate": 1.8814358919023855e-05,
+ "loss": 0.5884,
+ "step": 6675
+ },
+ {
+ "epoch": 18.190735694822887,
+ "grad_norm": 6.654816627502441,
+ "learning_rate": 1.8813942079235237e-05,
+ "loss": 0.5179,
+ "step": 6676
+ },
+ {
+ "epoch": 18.193460490463217,
+ "grad_norm": 7.520196914672852,
+ "learning_rate": 1.881352517080392e-05,
+ "loss": 0.792,
+ "step": 6677
+ },
+ {
+ "epoch": 18.196185286103542,
+ "grad_norm": 6.491469383239746,
+ "learning_rate": 1.8813108193733155e-05,
+ "loss": 0.6345,
+ "step": 6678
+ },
+ {
+ "epoch": 18.19891008174387,
+ "grad_norm": 6.971287727355957,
+ "learning_rate": 1.8812691148026186e-05,
+ "loss": 0.5911,
+ "step": 6679
+ },
+ {
+ "epoch": 18.201634877384198,
+ "grad_norm": 6.27214241027832,
+ "learning_rate": 1.881227403368626e-05,
+ "loss": 0.6115,
+ "step": 6680
+ },
+ {
+ "epoch": 18.204359673024523,
+ "grad_norm": 6.773125648498535,
+ "learning_rate": 1.881185685071663e-05,
+ "loss": 0.6463,
+ "step": 6681
+ },
+ {
+ "epoch": 18.20708446866485,
+ "grad_norm": 7.108669757843018,
+ "learning_rate": 1.8811439599120543e-05,
+ "loss": 0.7583,
+ "step": 6682
+ },
+ {
+ "epoch": 18.20980926430518,
+ "grad_norm": 5.235254764556885,
+ "learning_rate": 1.8811022278901244e-05,
+ "loss": 0.76,
+ "step": 6683
+ },
+ {
+ "epoch": 18.212534059945504,
+ "grad_norm": 6.39813232421875,
+ "learning_rate": 1.881060489006199e-05,
+ "loss": 0.8232,
+ "step": 6684
+ },
+ {
+ "epoch": 18.21525885558583,
+ "grad_norm": 6.405544757843018,
+ "learning_rate": 1.881018743260603e-05,
+ "loss": 0.5749,
+ "step": 6685
+ },
+ {
+ "epoch": 18.21798365122616,
+ "grad_norm": 5.674301624298096,
+ "learning_rate": 1.880976990653661e-05,
+ "loss": 0.7519,
+ "step": 6686
+ },
+ {
+ "epoch": 18.220708446866485,
+ "grad_norm": 39.612674713134766,
+ "learning_rate": 1.8809352311856987e-05,
+ "loss": 0.5761,
+ "step": 6687
+ },
+ {
+ "epoch": 18.22343324250681,
+ "grad_norm": 6.625097751617432,
+ "learning_rate": 1.8808934648570412e-05,
+ "loss": 0.5001,
+ "step": 6688
+ },
+ {
+ "epoch": 18.22615803814714,
+ "grad_norm": 5.841471195220947,
+ "learning_rate": 1.8808516916680138e-05,
+ "loss": 0.7864,
+ "step": 6689
+ },
+ {
+ "epoch": 18.228882833787466,
+ "grad_norm": 6.493096351623535,
+ "learning_rate": 1.880809911618942e-05,
+ "loss": 0.7527,
+ "step": 6690
+ },
+ {
+ "epoch": 18.231607629427792,
+ "grad_norm": 6.844666957855225,
+ "learning_rate": 1.8807681247101507e-05,
+ "loss": 0.7268,
+ "step": 6691
+ },
+ {
+ "epoch": 18.23433242506812,
+ "grad_norm": 7.9295973777771,
+ "learning_rate": 1.8807263309419656e-05,
+ "loss": 0.5621,
+ "step": 6692
+ },
+ {
+ "epoch": 18.237057220708447,
+ "grad_norm": 6.03066349029541,
+ "learning_rate": 1.8806845303147123e-05,
+ "loss": 0.7612,
+ "step": 6693
+ },
+ {
+ "epoch": 18.239782016348773,
+ "grad_norm": 6.784713268280029,
+ "learning_rate": 1.880642722828716e-05,
+ "loss": 0.8254,
+ "step": 6694
+ },
+ {
+ "epoch": 18.242506811989102,
+ "grad_norm": 7.496746063232422,
+ "learning_rate": 1.8806009084843025e-05,
+ "loss": 0.6436,
+ "step": 6695
+ },
+ {
+ "epoch": 18.245231607629428,
+ "grad_norm": 8.940818786621094,
+ "learning_rate": 1.8805590872817976e-05,
+ "loss": 0.7382,
+ "step": 6696
+ },
+ {
+ "epoch": 18.247956403269754,
+ "grad_norm": 6.5962605476379395,
+ "learning_rate": 1.880517259221527e-05,
+ "loss": 0.5312,
+ "step": 6697
+ },
+ {
+ "epoch": 18.250681198910083,
+ "grad_norm": 6.473339080810547,
+ "learning_rate": 1.880475424303816e-05,
+ "loss": 0.5739,
+ "step": 6698
+ },
+ {
+ "epoch": 18.25340599455041,
+ "grad_norm": 6.677170753479004,
+ "learning_rate": 1.880433582528991e-05,
+ "loss": 0.6602,
+ "step": 6699
+ },
+ {
+ "epoch": 18.256130790190735,
+ "grad_norm": 7.117041110992432,
+ "learning_rate": 1.8803917338973778e-05,
+ "loss": 0.6115,
+ "step": 6700
+ },
+ {
+ "epoch": 18.258855585831064,
+ "grad_norm": 6.8219404220581055,
+ "learning_rate": 1.8803498784093014e-05,
+ "loss": 0.824,
+ "step": 6701
+ },
+ {
+ "epoch": 18.26158038147139,
+ "grad_norm": 5.341588020324707,
+ "learning_rate": 1.880308016065089e-05,
+ "loss": 0.6737,
+ "step": 6702
+ },
+ {
+ "epoch": 18.264305177111716,
+ "grad_norm": 5.815537929534912,
+ "learning_rate": 1.880266146865066e-05,
+ "loss": 0.5534,
+ "step": 6703
+ },
+ {
+ "epoch": 18.267029972752045,
+ "grad_norm": 7.702908039093018,
+ "learning_rate": 1.8802242708095586e-05,
+ "loss": 0.663,
+ "step": 6704
+ },
+ {
+ "epoch": 18.26975476839237,
+ "grad_norm": 7.0402398109436035,
+ "learning_rate": 1.880182387898893e-05,
+ "loss": 0.7654,
+ "step": 6705
+ },
+ {
+ "epoch": 18.272479564032697,
+ "grad_norm": 8.090280532836914,
+ "learning_rate": 1.880140498133395e-05,
+ "loss": 0.8198,
+ "step": 6706
+ },
+ {
+ "epoch": 18.275204359673026,
+ "grad_norm": 6.805511474609375,
+ "learning_rate": 1.8800986015133914e-05,
+ "loss": 0.7954,
+ "step": 6707
+ },
+ {
+ "epoch": 18.277929155313352,
+ "grad_norm": 5.948941230773926,
+ "learning_rate": 1.8800566980392082e-05,
+ "loss": 0.6927,
+ "step": 6708
+ },
+ {
+ "epoch": 18.280653950953678,
+ "grad_norm": 6.420788288116455,
+ "learning_rate": 1.8800147877111716e-05,
+ "loss": 0.5272,
+ "step": 6709
+ },
+ {
+ "epoch": 18.283378746594007,
+ "grad_norm": 6.486981391906738,
+ "learning_rate": 1.8799728705296084e-05,
+ "loss": 0.6008,
+ "step": 6710
+ },
+ {
+ "epoch": 18.286103542234333,
+ "grad_norm": 5.717773914337158,
+ "learning_rate": 1.8799309464948446e-05,
+ "loss": 0.7407,
+ "step": 6711
+ },
+ {
+ "epoch": 18.28882833787466,
+ "grad_norm": 6.138091087341309,
+ "learning_rate": 1.8798890156072068e-05,
+ "loss": 0.5967,
+ "step": 6712
+ },
+ {
+ "epoch": 18.291553133514988,
+ "grad_norm": 9.062820434570312,
+ "learning_rate": 1.879847077867022e-05,
+ "loss": 0.6991,
+ "step": 6713
+ },
+ {
+ "epoch": 18.294277929155314,
+ "grad_norm": 5.894587993621826,
+ "learning_rate": 1.879805133274616e-05,
+ "loss": 0.6011,
+ "step": 6714
+ },
+ {
+ "epoch": 18.29700272479564,
+ "grad_norm": 12.40303897857666,
+ "learning_rate": 1.8797631818303164e-05,
+ "loss": 0.7303,
+ "step": 6715
+ },
+ {
+ "epoch": 18.29972752043597,
+ "grad_norm": 7.138026237487793,
+ "learning_rate": 1.8797212235344492e-05,
+ "loss": 0.6902,
+ "step": 6716
+ },
+ {
+ "epoch": 18.302452316076295,
+ "grad_norm": 11.192110061645508,
+ "learning_rate": 1.8796792583873418e-05,
+ "loss": 0.8864,
+ "step": 6717
+ },
+ {
+ "epoch": 18.30517711171662,
+ "grad_norm": 6.647374629974365,
+ "learning_rate": 1.87963728638932e-05,
+ "loss": 0.6692,
+ "step": 6718
+ },
+ {
+ "epoch": 18.30790190735695,
+ "grad_norm": 6.773301601409912,
+ "learning_rate": 1.8795953075407117e-05,
+ "loss": 0.6723,
+ "step": 6719
+ },
+ {
+ "epoch": 18.310626702997276,
+ "grad_norm": 8.09667682647705,
+ "learning_rate": 1.8795533218418437e-05,
+ "loss": 0.6707,
+ "step": 6720
+ },
+ {
+ "epoch": 18.3133514986376,
+ "grad_norm": 9.299240112304688,
+ "learning_rate": 1.8795113292930425e-05,
+ "loss": 0.7009,
+ "step": 6721
+ },
+ {
+ "epoch": 18.31607629427793,
+ "grad_norm": 6.4720611572265625,
+ "learning_rate": 1.8794693298946352e-05,
+ "loss": 0.7379,
+ "step": 6722
+ },
+ {
+ "epoch": 18.318801089918257,
+ "grad_norm": 7.565492630004883,
+ "learning_rate": 1.8794273236469495e-05,
+ "loss": 0.6927,
+ "step": 6723
+ },
+ {
+ "epoch": 18.321525885558582,
+ "grad_norm": 7.442156791687012,
+ "learning_rate": 1.8793853105503118e-05,
+ "loss": 0.7285,
+ "step": 6724
+ },
+ {
+ "epoch": 18.32425068119891,
+ "grad_norm": 5.594560623168945,
+ "learning_rate": 1.8793432906050496e-05,
+ "loss": 0.6952,
+ "step": 6725
+ },
+ {
+ "epoch": 18.326975476839237,
+ "grad_norm": 5.855583667755127,
+ "learning_rate": 1.87930126381149e-05,
+ "loss": 0.5842,
+ "step": 6726
+ },
+ {
+ "epoch": 18.329700272479563,
+ "grad_norm": 7.538547992706299,
+ "learning_rate": 1.879259230169961e-05,
+ "loss": 0.624,
+ "step": 6727
+ },
+ {
+ "epoch": 18.332425068119893,
+ "grad_norm": 6.938891887664795,
+ "learning_rate": 1.8792171896807886e-05,
+ "loss": 0.7976,
+ "step": 6728
+ },
+ {
+ "epoch": 18.33514986376022,
+ "grad_norm": 6.316887855529785,
+ "learning_rate": 1.879175142344302e-05,
+ "loss": 0.7184,
+ "step": 6729
+ },
+ {
+ "epoch": 18.337874659400544,
+ "grad_norm": 6.728006839752197,
+ "learning_rate": 1.8791330881608266e-05,
+ "loss": 0.6302,
+ "step": 6730
+ },
+ {
+ "epoch": 18.340599455040874,
+ "grad_norm": 6.75941276550293,
+ "learning_rate": 1.879091027130692e-05,
+ "loss": 0.595,
+ "step": 6731
+ },
+ {
+ "epoch": 18.3433242506812,
+ "grad_norm": 6.749118328094482,
+ "learning_rate": 1.8790489592542243e-05,
+ "loss": 0.9406,
+ "step": 6732
+ },
+ {
+ "epoch": 18.346049046321525,
+ "grad_norm": 7.029022216796875,
+ "learning_rate": 1.8790068845317516e-05,
+ "loss": 0.7247,
+ "step": 6733
+ },
+ {
+ "epoch": 18.348773841961854,
+ "grad_norm": 8.143996238708496,
+ "learning_rate": 1.8789648029636014e-05,
+ "loss": 0.6085,
+ "step": 6734
+ },
+ {
+ "epoch": 18.35149863760218,
+ "grad_norm": 6.151351451873779,
+ "learning_rate": 1.8789227145501023e-05,
+ "loss": 0.5411,
+ "step": 6735
+ },
+ {
+ "epoch": 18.354223433242506,
+ "grad_norm": 6.702400207519531,
+ "learning_rate": 1.8788806192915804e-05,
+ "loss": 0.6144,
+ "step": 6736
+ },
+ {
+ "epoch": 18.356948228882835,
+ "grad_norm": 6.716687202453613,
+ "learning_rate": 1.8788385171883653e-05,
+ "loss": 0.7966,
+ "step": 6737
+ },
+ {
+ "epoch": 18.35967302452316,
+ "grad_norm": 6.035650730133057,
+ "learning_rate": 1.878796408240784e-05,
+ "loss": 0.5903,
+ "step": 6738
+ },
+ {
+ "epoch": 18.362397820163487,
+ "grad_norm": 7.766988277435303,
+ "learning_rate": 1.878754292449164e-05,
+ "loss": 0.7619,
+ "step": 6739
+ },
+ {
+ "epoch": 18.365122615803816,
+ "grad_norm": 6.451503753662109,
+ "learning_rate": 1.8787121698138344e-05,
+ "loss": 0.649,
+ "step": 6740
+ },
+ {
+ "epoch": 18.367847411444142,
+ "grad_norm": 7.602853298187256,
+ "learning_rate": 1.8786700403351226e-05,
+ "loss": 0.8048,
+ "step": 6741
+ },
+ {
+ "epoch": 18.370572207084468,
+ "grad_norm": 7.25816011428833,
+ "learning_rate": 1.878627904013357e-05,
+ "loss": 0.7247,
+ "step": 6742
+ },
+ {
+ "epoch": 18.373297002724797,
+ "grad_norm": 7.321986675262451,
+ "learning_rate": 1.878585760848865e-05,
+ "loss": 0.5671,
+ "step": 6743
+ },
+ {
+ "epoch": 18.376021798365123,
+ "grad_norm": 8.390989303588867,
+ "learning_rate": 1.878543610841976e-05,
+ "loss": 0.7078,
+ "step": 6744
+ },
+ {
+ "epoch": 18.37874659400545,
+ "grad_norm": 7.434986591339111,
+ "learning_rate": 1.878501453993017e-05,
+ "loss": 0.7037,
+ "step": 6745
+ },
+ {
+ "epoch": 18.381471389645778,
+ "grad_norm": 6.367483139038086,
+ "learning_rate": 1.8784592903023175e-05,
+ "loss": 0.713,
+ "step": 6746
+ },
+ {
+ "epoch": 18.384196185286104,
+ "grad_norm": 6.508609294891357,
+ "learning_rate": 1.8784171197702053e-05,
+ "loss": 0.6319,
+ "step": 6747
+ },
+ {
+ "epoch": 18.38692098092643,
+ "grad_norm": 7.209470272064209,
+ "learning_rate": 1.8783749423970085e-05,
+ "loss": 0.6519,
+ "step": 6748
+ },
+ {
+ "epoch": 18.38964577656676,
+ "grad_norm": 6.297450542449951,
+ "learning_rate": 1.878332758183056e-05,
+ "loss": 0.7872,
+ "step": 6749
+ },
+ {
+ "epoch": 18.392370572207085,
+ "grad_norm": 6.257380485534668,
+ "learning_rate": 1.8782905671286763e-05,
+ "loss": 0.7465,
+ "step": 6750
+ },
+ {
+ "epoch": 18.39509536784741,
+ "grad_norm": 7.1928863525390625,
+ "learning_rate": 1.878248369234198e-05,
+ "loss": 0.7355,
+ "step": 6751
+ },
+ {
+ "epoch": 18.39782016348774,
+ "grad_norm": 6.287572383880615,
+ "learning_rate": 1.8782061644999497e-05,
+ "loss": 0.7584,
+ "step": 6752
+ },
+ {
+ "epoch": 18.400544959128066,
+ "grad_norm": 7.312450408935547,
+ "learning_rate": 1.8781639529262598e-05,
+ "loss": 0.5624,
+ "step": 6753
+ },
+ {
+ "epoch": 18.40326975476839,
+ "grad_norm": 8.292914390563965,
+ "learning_rate": 1.8781217345134574e-05,
+ "loss": 0.7487,
+ "step": 6754
+ },
+ {
+ "epoch": 18.40599455040872,
+ "grad_norm": 8.438583374023438,
+ "learning_rate": 1.8780795092618717e-05,
+ "loss": 0.6614,
+ "step": 6755
+ },
+ {
+ "epoch": 18.408719346049047,
+ "grad_norm": 6.502033710479736,
+ "learning_rate": 1.8780372771718302e-05,
+ "loss": 0.7734,
+ "step": 6756
+ },
+ {
+ "epoch": 18.411444141689373,
+ "grad_norm": 9.18928050994873,
+ "learning_rate": 1.8779950382436632e-05,
+ "loss": 0.646,
+ "step": 6757
+ },
+ {
+ "epoch": 18.414168937329702,
+ "grad_norm": 7.516249656677246,
+ "learning_rate": 1.8779527924776988e-05,
+ "loss": 0.759,
+ "step": 6758
+ },
+ {
+ "epoch": 18.416893732970028,
+ "grad_norm": 6.375486850738525,
+ "learning_rate": 1.877910539874267e-05,
+ "loss": 0.6379,
+ "step": 6759
+ },
+ {
+ "epoch": 18.419618528610354,
+ "grad_norm": 6.705495834350586,
+ "learning_rate": 1.877868280433695e-05,
+ "loss": 0.7108,
+ "step": 6760
+ },
+ {
+ "epoch": 18.422343324250683,
+ "grad_norm": 6.322361469268799,
+ "learning_rate": 1.877826014156314e-05,
+ "loss": 0.7731,
+ "step": 6761
+ },
+ {
+ "epoch": 18.42506811989101,
+ "grad_norm": 5.349233150482178,
+ "learning_rate": 1.877783741042452e-05,
+ "loss": 0.7321,
+ "step": 6762
+ },
+ {
+ "epoch": 18.427792915531334,
+ "grad_norm": 6.05946683883667,
+ "learning_rate": 1.8777414610924386e-05,
+ "loss": 0.5829,
+ "step": 6763
+ },
+ {
+ "epoch": 18.430517711171664,
+ "grad_norm": 5.77848482131958,
+ "learning_rate": 1.8776991743066025e-05,
+ "loss": 0.7382,
+ "step": 6764
+ },
+ {
+ "epoch": 18.43324250681199,
+ "grad_norm": 21.503324508666992,
+ "learning_rate": 1.8776568806852738e-05,
+ "loss": 0.6625,
+ "step": 6765
+ },
+ {
+ "epoch": 18.435967302452315,
+ "grad_norm": 5.688875675201416,
+ "learning_rate": 1.8776145802287816e-05,
+ "loss": 0.6244,
+ "step": 6766
+ },
+ {
+ "epoch": 18.438692098092645,
+ "grad_norm": 7.668795585632324,
+ "learning_rate": 1.8775722729374552e-05,
+ "loss": 0.7024,
+ "step": 6767
+ },
+ {
+ "epoch": 18.44141689373297,
+ "grad_norm": 6.336337089538574,
+ "learning_rate": 1.8775299588116243e-05,
+ "loss": 0.5566,
+ "step": 6768
+ },
+ {
+ "epoch": 18.444141689373296,
+ "grad_norm": 5.8938374519348145,
+ "learning_rate": 1.877487637851618e-05,
+ "loss": 0.6287,
+ "step": 6769
+ },
+ {
+ "epoch": 18.446866485013626,
+ "grad_norm": 5.710447311401367,
+ "learning_rate": 1.8774453100577664e-05,
+ "loss": 0.9163,
+ "step": 6770
+ },
+ {
+ "epoch": 18.44959128065395,
+ "grad_norm": 6.963895797729492,
+ "learning_rate": 1.8774029754303995e-05,
+ "loss": 0.6748,
+ "step": 6771
+ },
+ {
+ "epoch": 18.452316076294277,
+ "grad_norm": 5.276904582977295,
+ "learning_rate": 1.877360633969846e-05,
+ "loss": 0.655,
+ "step": 6772
+ },
+ {
+ "epoch": 18.455040871934607,
+ "grad_norm": 6.160674571990967,
+ "learning_rate": 1.877318285676436e-05,
+ "loss": 0.5867,
+ "step": 6773
+ },
+ {
+ "epoch": 18.457765667574932,
+ "grad_norm": 5.826469898223877,
+ "learning_rate": 1.8772759305504996e-05,
+ "loss": 0.6397,
+ "step": 6774
+ },
+ {
+ "epoch": 18.460490463215258,
+ "grad_norm": 5.571272850036621,
+ "learning_rate": 1.877233568592366e-05,
+ "loss": 0.7468,
+ "step": 6775
+ },
+ {
+ "epoch": 18.463215258855588,
+ "grad_norm": 6.32149600982666,
+ "learning_rate": 1.877191199802366e-05,
+ "loss": 0.7749,
+ "step": 6776
+ },
+ {
+ "epoch": 18.465940054495913,
+ "grad_norm": 5.181055068969727,
+ "learning_rate": 1.8771488241808295e-05,
+ "loss": 0.5796,
+ "step": 6777
+ },
+ {
+ "epoch": 18.46866485013624,
+ "grad_norm": 5.2100348472595215,
+ "learning_rate": 1.8771064417280856e-05,
+ "loss": 0.7129,
+ "step": 6778
+ },
+ {
+ "epoch": 18.47138964577657,
+ "grad_norm": 7.008127212524414,
+ "learning_rate": 1.8770640524444653e-05,
+ "loss": 0.7389,
+ "step": 6779
+ },
+ {
+ "epoch": 18.474114441416894,
+ "grad_norm": 5.9197163581848145,
+ "learning_rate": 1.8770216563302984e-05,
+ "loss": 0.5883,
+ "step": 6780
+ },
+ {
+ "epoch": 18.47683923705722,
+ "grad_norm": 7.229288578033447,
+ "learning_rate": 1.876979253385915e-05,
+ "loss": 0.5946,
+ "step": 6781
+ },
+ {
+ "epoch": 18.479564032697546,
+ "grad_norm": 6.109094619750977,
+ "learning_rate": 1.8769368436116454e-05,
+ "loss": 0.6171,
+ "step": 6782
+ },
+ {
+ "epoch": 18.482288828337875,
+ "grad_norm": 6.929074287414551,
+ "learning_rate": 1.87689442700782e-05,
+ "loss": 0.7766,
+ "step": 6783
+ },
+ {
+ "epoch": 18.4850136239782,
+ "grad_norm": 6.9800262451171875,
+ "learning_rate": 1.876852003574769e-05,
+ "loss": 0.6738,
+ "step": 6784
+ },
+ {
+ "epoch": 18.48773841961853,
+ "grad_norm": 6.498555660247803,
+ "learning_rate": 1.8768095733128226e-05,
+ "loss": 0.7582,
+ "step": 6785
+ },
+ {
+ "epoch": 18.490463215258856,
+ "grad_norm": 6.23679256439209,
+ "learning_rate": 1.8767671362223117e-05,
+ "loss": 0.6245,
+ "step": 6786
+ },
+ {
+ "epoch": 18.493188010899182,
+ "grad_norm": 6.432779788970947,
+ "learning_rate": 1.8767246923035663e-05,
+ "loss": 0.6516,
+ "step": 6787
+ },
+ {
+ "epoch": 18.495912806539508,
+ "grad_norm": 5.853081226348877,
+ "learning_rate": 1.8766822415569175e-05,
+ "loss": 0.6453,
+ "step": 6788
+ },
+ {
+ "epoch": 18.498637602179837,
+ "grad_norm": 6.303401470184326,
+ "learning_rate": 1.8766397839826952e-05,
+ "loss": 0.7274,
+ "step": 6789
+ },
+ {
+ "epoch": 18.501362397820163,
+ "grad_norm": 6.066772937774658,
+ "learning_rate": 1.876597319581231e-05,
+ "loss": 0.6401,
+ "step": 6790
+ },
+ {
+ "epoch": 18.504087193460492,
+ "grad_norm": 5.871001243591309,
+ "learning_rate": 1.8765548483528548e-05,
+ "loss": 0.721,
+ "step": 6791
+ },
+ {
+ "epoch": 18.506811989100818,
+ "grad_norm": 6.707411289215088,
+ "learning_rate": 1.876512370297898e-05,
+ "loss": 0.6425,
+ "step": 6792
+ },
+ {
+ "epoch": 18.509536784741144,
+ "grad_norm": 5.8320817947387695,
+ "learning_rate": 1.8764698854166904e-05,
+ "loss": 0.7137,
+ "step": 6793
+ },
+ {
+ "epoch": 18.51226158038147,
+ "grad_norm": 5.645232677459717,
+ "learning_rate": 1.8764273937095642e-05,
+ "loss": 0.6823,
+ "step": 6794
+ },
+ {
+ "epoch": 18.5149863760218,
+ "grad_norm": 6.042266368865967,
+ "learning_rate": 1.876384895176849e-05,
+ "loss": 0.6348,
+ "step": 6795
+ },
+ {
+ "epoch": 18.517711171662125,
+ "grad_norm": 6.833873748779297,
+ "learning_rate": 1.876342389818877e-05,
+ "loss": 0.5905,
+ "step": 6796
+ },
+ {
+ "epoch": 18.520435967302454,
+ "grad_norm": 7.458170413970947,
+ "learning_rate": 1.8762998776359785e-05,
+ "loss": 0.6431,
+ "step": 6797
+ },
+ {
+ "epoch": 18.52316076294278,
+ "grad_norm": 6.905101776123047,
+ "learning_rate": 1.8762573586284847e-05,
+ "loss": 0.8477,
+ "step": 6798
+ },
+ {
+ "epoch": 18.525885558583106,
+ "grad_norm": 6.287354946136475,
+ "learning_rate": 1.876214832796727e-05,
+ "loss": 0.5556,
+ "step": 6799
+ },
+ {
+ "epoch": 18.52861035422343,
+ "grad_norm": 5.875155925750732,
+ "learning_rate": 1.876172300141036e-05,
+ "loss": 0.6425,
+ "step": 6800
+ },
+ {
+ "epoch": 18.53133514986376,
+ "grad_norm": 7.368239402770996,
+ "learning_rate": 1.8761297606617436e-05,
+ "loss": 0.6041,
+ "step": 6801
+ },
+ {
+ "epoch": 18.534059945504087,
+ "grad_norm": 5.609370231628418,
+ "learning_rate": 1.876087214359181e-05,
+ "loss": 0.6917,
+ "step": 6802
+ },
+ {
+ "epoch": 18.536784741144416,
+ "grad_norm": 5.980183124542236,
+ "learning_rate": 1.876044661233679e-05,
+ "loss": 0.7563,
+ "step": 6803
+ },
+ {
+ "epoch": 18.539509536784742,
+ "grad_norm": 6.50679874420166,
+ "learning_rate": 1.8760021012855694e-05,
+ "loss": 0.8522,
+ "step": 6804
+ },
+ {
+ "epoch": 18.542234332425068,
+ "grad_norm": 6.937196731567383,
+ "learning_rate": 1.875959534515184e-05,
+ "loss": 0.7105,
+ "step": 6805
+ },
+ {
+ "epoch": 18.544959128065393,
+ "grad_norm": 5.926344394683838,
+ "learning_rate": 1.8759169609228536e-05,
+ "loss": 0.6889,
+ "step": 6806
+ },
+ {
+ "epoch": 18.547683923705723,
+ "grad_norm": 6.6368021965026855,
+ "learning_rate": 1.8758743805089104e-05,
+ "loss": 0.5592,
+ "step": 6807
+ },
+ {
+ "epoch": 18.55040871934605,
+ "grad_norm": 6.585180759429932,
+ "learning_rate": 1.8758317932736855e-05,
+ "loss": 0.6586,
+ "step": 6808
+ },
+ {
+ "epoch": 18.553133514986374,
+ "grad_norm": 6.058976173400879,
+ "learning_rate": 1.8757891992175106e-05,
+ "loss": 0.7345,
+ "step": 6809
+ },
+ {
+ "epoch": 18.555858310626704,
+ "grad_norm": 6.126471042633057,
+ "learning_rate": 1.875746598340718e-05,
+ "loss": 0.6908,
+ "step": 6810
+ },
+ {
+ "epoch": 18.55858310626703,
+ "grad_norm": 7.108178615570068,
+ "learning_rate": 1.8757039906436388e-05,
+ "loss": 0.6349,
+ "step": 6811
+ },
+ {
+ "epoch": 18.561307901907355,
+ "grad_norm": 5.744268894195557,
+ "learning_rate": 1.8756613761266052e-05,
+ "loss": 0.6946,
+ "step": 6812
+ },
+ {
+ "epoch": 18.564032697547685,
+ "grad_norm": 6.942866325378418,
+ "learning_rate": 1.875618754789949e-05,
+ "loss": 0.7565,
+ "step": 6813
+ },
+ {
+ "epoch": 18.56675749318801,
+ "grad_norm": 6.471741199493408,
+ "learning_rate": 1.8755761266340018e-05,
+ "loss": 0.6633,
+ "step": 6814
+ },
+ {
+ "epoch": 18.569482288828336,
+ "grad_norm": 5.42349100112915,
+ "learning_rate": 1.8755334916590964e-05,
+ "loss": 0.71,
+ "step": 6815
+ },
+ {
+ "epoch": 18.572207084468666,
+ "grad_norm": 6.5400872230529785,
+ "learning_rate": 1.875490849865564e-05,
+ "loss": 0.7877,
+ "step": 6816
+ },
+ {
+ "epoch": 18.57493188010899,
+ "grad_norm": 6.019377708435059,
+ "learning_rate": 1.8754482012537374e-05,
+ "loss": 0.6616,
+ "step": 6817
+ },
+ {
+ "epoch": 18.577656675749317,
+ "grad_norm": 9.954610824584961,
+ "learning_rate": 1.875405545823948e-05,
+ "loss": 0.4915,
+ "step": 6818
+ },
+ {
+ "epoch": 18.580381471389646,
+ "grad_norm": 6.453253746032715,
+ "learning_rate": 1.8753628835765284e-05,
+ "loss": 0.6388,
+ "step": 6819
+ },
+ {
+ "epoch": 18.583106267029972,
+ "grad_norm": 6.318390369415283,
+ "learning_rate": 1.875320214511811e-05,
+ "loss": 0.6584,
+ "step": 6820
+ },
+ {
+ "epoch": 18.585831062670298,
+ "grad_norm": 7.15533971786499,
+ "learning_rate": 1.875277538630128e-05,
+ "loss": 0.5543,
+ "step": 6821
+ },
+ {
+ "epoch": 18.588555858310627,
+ "grad_norm": 7.1196746826171875,
+ "learning_rate": 1.8752348559318116e-05,
+ "loss": 0.5983,
+ "step": 6822
+ },
+ {
+ "epoch": 18.591280653950953,
+ "grad_norm": 6.184885501861572,
+ "learning_rate": 1.8751921664171944e-05,
+ "loss": 0.4814,
+ "step": 6823
+ },
+ {
+ "epoch": 18.59400544959128,
+ "grad_norm": 6.742511749267578,
+ "learning_rate": 1.8751494700866088e-05,
+ "loss": 0.5856,
+ "step": 6824
+ },
+ {
+ "epoch": 18.59673024523161,
+ "grad_norm": 5.961170673370361,
+ "learning_rate": 1.875106766940387e-05,
+ "loss": 0.8816,
+ "step": 6825
+ },
+ {
+ "epoch": 18.599455040871934,
+ "grad_norm": 6.5132737159729,
+ "learning_rate": 1.8750640569788624e-05,
+ "loss": 0.6939,
+ "step": 6826
+ },
+ {
+ "epoch": 18.60217983651226,
+ "grad_norm": 6.672764778137207,
+ "learning_rate": 1.8750213402023664e-05,
+ "loss": 0.6628,
+ "step": 6827
+ },
+ {
+ "epoch": 18.60490463215259,
+ "grad_norm": 6.2762298583984375,
+ "learning_rate": 1.874978616611233e-05,
+ "loss": 0.7495,
+ "step": 6828
+ },
+ {
+ "epoch": 18.607629427792915,
+ "grad_norm": 6.011996746063232,
+ "learning_rate": 1.874935886205794e-05,
+ "loss": 0.6588,
+ "step": 6829
+ },
+ {
+ "epoch": 18.61035422343324,
+ "grad_norm": 6.632678508758545,
+ "learning_rate": 1.8748931489863823e-05,
+ "loss": 0.7246,
+ "step": 6830
+ },
+ {
+ "epoch": 18.61307901907357,
+ "grad_norm": 6.407528877258301,
+ "learning_rate": 1.874850404953331e-05,
+ "loss": 0.6694,
+ "step": 6831
+ },
+ {
+ "epoch": 18.615803814713896,
+ "grad_norm": 6.929873943328857,
+ "learning_rate": 1.8748076541069734e-05,
+ "loss": 0.7327,
+ "step": 6832
+ },
+ {
+ "epoch": 18.618528610354222,
+ "grad_norm": 5.946615695953369,
+ "learning_rate": 1.874764896447641e-05,
+ "loss": 0.7109,
+ "step": 6833
+ },
+ {
+ "epoch": 18.62125340599455,
+ "grad_norm": 6.4113969802856445,
+ "learning_rate": 1.8747221319756686e-05,
+ "loss": 0.7073,
+ "step": 6834
+ },
+ {
+ "epoch": 18.623978201634877,
+ "grad_norm": 6.269152641296387,
+ "learning_rate": 1.874679360691388e-05,
+ "loss": 0.7769,
+ "step": 6835
+ },
+ {
+ "epoch": 18.626702997275203,
+ "grad_norm": 7.122358798980713,
+ "learning_rate": 1.874636582595133e-05,
+ "loss": 0.7488,
+ "step": 6836
+ },
+ {
+ "epoch": 18.629427792915532,
+ "grad_norm": 6.47542142868042,
+ "learning_rate": 1.874593797687236e-05,
+ "loss": 0.5358,
+ "step": 6837
+ },
+ {
+ "epoch": 18.632152588555858,
+ "grad_norm": 7.2293477058410645,
+ "learning_rate": 1.874551005968031e-05,
+ "loss": 0.6221,
+ "step": 6838
+ },
+ {
+ "epoch": 18.634877384196184,
+ "grad_norm": 6.802626609802246,
+ "learning_rate": 1.8745082074378507e-05,
+ "loss": 0.7543,
+ "step": 6839
+ },
+ {
+ "epoch": 18.637602179836513,
+ "grad_norm": 7.149992942810059,
+ "learning_rate": 1.8744654020970284e-05,
+ "loss": 0.7138,
+ "step": 6840
+ },
+ {
+ "epoch": 18.64032697547684,
+ "grad_norm": 6.205387592315674,
+ "learning_rate": 1.8744225899458982e-05,
+ "loss": 0.4858,
+ "step": 6841
+ },
+ {
+ "epoch": 18.643051771117165,
+ "grad_norm": 5.974452018737793,
+ "learning_rate": 1.874379770984793e-05,
+ "loss": 0.5787,
+ "step": 6842
+ },
+ {
+ "epoch": 18.645776566757494,
+ "grad_norm": 6.029594898223877,
+ "learning_rate": 1.8743369452140457e-05,
+ "loss": 0.7482,
+ "step": 6843
+ },
+ {
+ "epoch": 18.64850136239782,
+ "grad_norm": 6.565267562866211,
+ "learning_rate": 1.874294112633991e-05,
+ "loss": 0.6791,
+ "step": 6844
+ },
+ {
+ "epoch": 18.651226158038146,
+ "grad_norm": 5.796443462371826,
+ "learning_rate": 1.874251273244962e-05,
+ "loss": 0.6821,
+ "step": 6845
+ },
+ {
+ "epoch": 18.653950953678475,
+ "grad_norm": 5.9635491371154785,
+ "learning_rate": 1.874208427047292e-05,
+ "loss": 0.7996,
+ "step": 6846
+ },
+ {
+ "epoch": 18.6566757493188,
+ "grad_norm": 8.253134727478027,
+ "learning_rate": 1.8741655740413143e-05,
+ "loss": 0.6998,
+ "step": 6847
+ },
+ {
+ "epoch": 18.659400544959126,
+ "grad_norm": 5.050978660583496,
+ "learning_rate": 1.874122714227364e-05,
+ "loss": 0.8492,
+ "step": 6848
+ },
+ {
+ "epoch": 18.662125340599456,
+ "grad_norm": 5.963735103607178,
+ "learning_rate": 1.8740798476057737e-05,
+ "loss": 0.7432,
+ "step": 6849
+ },
+ {
+ "epoch": 18.66485013623978,
+ "grad_norm": 6.004485130310059,
+ "learning_rate": 1.874036974176878e-05,
+ "loss": 0.5386,
+ "step": 6850
+ },
+ {
+ "epoch": 18.667574931880107,
+ "grad_norm": 5.695806980133057,
+ "learning_rate": 1.8739940939410102e-05,
+ "loss": 0.6125,
+ "step": 6851
+ },
+ {
+ "epoch": 18.670299727520437,
+ "grad_norm": 7.048089027404785,
+ "learning_rate": 1.8739512068985052e-05,
+ "loss": 0.7284,
+ "step": 6852
+ },
+ {
+ "epoch": 18.673024523160763,
+ "grad_norm": 6.8418354988098145,
+ "learning_rate": 1.8739083130496957e-05,
+ "loss": 0.6649,
+ "step": 6853
+ },
+ {
+ "epoch": 18.67574931880109,
+ "grad_norm": 7.313706874847412,
+ "learning_rate": 1.8738654123949165e-05,
+ "loss": 0.7244,
+ "step": 6854
+ },
+ {
+ "epoch": 18.678474114441418,
+ "grad_norm": 7.697445869445801,
+ "learning_rate": 1.8738225049345017e-05,
+ "loss": 0.5612,
+ "step": 6855
+ },
+ {
+ "epoch": 18.681198910081743,
+ "grad_norm": 5.536866188049316,
+ "learning_rate": 1.873779590668785e-05,
+ "loss": 0.6003,
+ "step": 6856
+ },
+ {
+ "epoch": 18.68392370572207,
+ "grad_norm": 8.122591018676758,
+ "learning_rate": 1.873736669598101e-05,
+ "loss": 0.6216,
+ "step": 6857
+ },
+ {
+ "epoch": 18.6866485013624,
+ "grad_norm": 6.950551986694336,
+ "learning_rate": 1.8736937417227845e-05,
+ "loss": 0.8317,
+ "step": 6858
+ },
+ {
+ "epoch": 18.689373297002724,
+ "grad_norm": 7.391556262969971,
+ "learning_rate": 1.873650807043169e-05,
+ "loss": 0.5065,
+ "step": 6859
+ },
+ {
+ "epoch": 18.69209809264305,
+ "grad_norm": 7.136149883270264,
+ "learning_rate": 1.873607865559589e-05,
+ "loss": 0.9524,
+ "step": 6860
+ },
+ {
+ "epoch": 18.69482288828338,
+ "grad_norm": 15.553674697875977,
+ "learning_rate": 1.8735649172723792e-05,
+ "loss": 0.6761,
+ "step": 6861
+ },
+ {
+ "epoch": 18.697547683923705,
+ "grad_norm": 7.299941539764404,
+ "learning_rate": 1.8735219621818737e-05,
+ "loss": 0.6375,
+ "step": 6862
+ },
+ {
+ "epoch": 18.70027247956403,
+ "grad_norm": 7.038736343383789,
+ "learning_rate": 1.8734790002884074e-05,
+ "loss": 0.5403,
+ "step": 6863
+ },
+ {
+ "epoch": 18.70299727520436,
+ "grad_norm": 7.11916971206665,
+ "learning_rate": 1.8734360315923146e-05,
+ "loss": 0.6473,
+ "step": 6864
+ },
+ {
+ "epoch": 18.705722070844686,
+ "grad_norm": 7.282226085662842,
+ "learning_rate": 1.8733930560939305e-05,
+ "loss": 0.7379,
+ "step": 6865
+ },
+ {
+ "epoch": 18.708446866485012,
+ "grad_norm": 7.591634273529053,
+ "learning_rate": 1.873350073793589e-05,
+ "loss": 0.8335,
+ "step": 6866
+ },
+ {
+ "epoch": 18.71117166212534,
+ "grad_norm": 7.3172287940979,
+ "learning_rate": 1.8733070846916253e-05,
+ "loss": 0.6442,
+ "step": 6867
+ },
+ {
+ "epoch": 18.713896457765667,
+ "grad_norm": 6.631596088409424,
+ "learning_rate": 1.8732640887883744e-05,
+ "loss": 0.7559,
+ "step": 6868
+ },
+ {
+ "epoch": 18.716621253405993,
+ "grad_norm": 6.091032028198242,
+ "learning_rate": 1.8732210860841706e-05,
+ "loss": 0.7959,
+ "step": 6869
+ },
+ {
+ "epoch": 18.719346049046322,
+ "grad_norm": 7.1278276443481445,
+ "learning_rate": 1.8731780765793493e-05,
+ "loss": 0.6948,
+ "step": 6870
+ },
+ {
+ "epoch": 18.722070844686648,
+ "grad_norm": 5.938060283660889,
+ "learning_rate": 1.8731350602742453e-05,
+ "loss": 0.7999,
+ "step": 6871
+ },
+ {
+ "epoch": 18.724795640326974,
+ "grad_norm": 5.775383472442627,
+ "learning_rate": 1.8730920371691934e-05,
+ "loss": 0.8232,
+ "step": 6872
+ },
+ {
+ "epoch": 18.727520435967303,
+ "grad_norm": 7.16690731048584,
+ "learning_rate": 1.8730490072645285e-05,
+ "loss": 0.6418,
+ "step": 6873
+ },
+ {
+ "epoch": 18.73024523160763,
+ "grad_norm": 6.339574337005615,
+ "learning_rate": 1.8730059705605862e-05,
+ "loss": 0.6198,
+ "step": 6874
+ },
+ {
+ "epoch": 18.732970027247955,
+ "grad_norm": 6.414369583129883,
+ "learning_rate": 1.8729629270577018e-05,
+ "loss": 0.6204,
+ "step": 6875
+ },
+ {
+ "epoch": 18.735694822888284,
+ "grad_norm": 7.326970100402832,
+ "learning_rate": 1.8729198767562102e-05,
+ "loss": 0.6168,
+ "step": 6876
+ },
+ {
+ "epoch": 18.73841961852861,
+ "grad_norm": 6.697366237640381,
+ "learning_rate": 1.8728768196564467e-05,
+ "loss": 0.6282,
+ "step": 6877
+ },
+ {
+ "epoch": 18.741144414168936,
+ "grad_norm": 7.021429538726807,
+ "learning_rate": 1.8728337557587464e-05,
+ "loss": 0.7621,
+ "step": 6878
+ },
+ {
+ "epoch": 18.743869209809265,
+ "grad_norm": 5.894412994384766,
+ "learning_rate": 1.872790685063445e-05,
+ "loss": 0.67,
+ "step": 6879
+ },
+ {
+ "epoch": 18.74659400544959,
+ "grad_norm": 6.807458877563477,
+ "learning_rate": 1.8727476075708778e-05,
+ "loss": 0.7531,
+ "step": 6880
+ },
+ {
+ "epoch": 18.749318801089917,
+ "grad_norm": 8.268558502197266,
+ "learning_rate": 1.8727045232813804e-05,
+ "loss": 0.5853,
+ "step": 6881
+ },
+ {
+ "epoch": 18.752043596730246,
+ "grad_norm": 5.9530229568481445,
+ "learning_rate": 1.872661432195288e-05,
+ "loss": 0.6077,
+ "step": 6882
+ },
+ {
+ "epoch": 18.754768392370572,
+ "grad_norm": 6.597161769866943,
+ "learning_rate": 1.8726183343129368e-05,
+ "loss": 0.7512,
+ "step": 6883
+ },
+ {
+ "epoch": 18.757493188010898,
+ "grad_norm": 6.530778408050537,
+ "learning_rate": 1.872575229634662e-05,
+ "loss": 0.8463,
+ "step": 6884
+ },
+ {
+ "epoch": 18.760217983651227,
+ "grad_norm": 6.659308910369873,
+ "learning_rate": 1.8725321181607993e-05,
+ "loss": 0.6763,
+ "step": 6885
+ },
+ {
+ "epoch": 18.762942779291553,
+ "grad_norm": 6.8689703941345215,
+ "learning_rate": 1.8724889998916845e-05,
+ "loss": 0.8646,
+ "step": 6886
+ },
+ {
+ "epoch": 18.76566757493188,
+ "grad_norm": 7.582526206970215,
+ "learning_rate": 1.8724458748276534e-05,
+ "loss": 0.7147,
+ "step": 6887
+ },
+ {
+ "epoch": 18.768392370572208,
+ "grad_norm": 5.014669418334961,
+ "learning_rate": 1.8724027429690422e-05,
+ "loss": 0.5937,
+ "step": 6888
+ },
+ {
+ "epoch": 18.771117166212534,
+ "grad_norm": 6.410392761230469,
+ "learning_rate": 1.8723596043161865e-05,
+ "loss": 0.6257,
+ "step": 6889
+ },
+ {
+ "epoch": 18.77384196185286,
+ "grad_norm": 6.192749977111816,
+ "learning_rate": 1.872316458869422e-05,
+ "loss": 0.5997,
+ "step": 6890
+ },
+ {
+ "epoch": 18.77656675749319,
+ "grad_norm": 5.985950946807861,
+ "learning_rate": 1.872273306629085e-05,
+ "loss": 0.7275,
+ "step": 6891
+ },
+ {
+ "epoch": 18.779291553133515,
+ "grad_norm": 9.104208946228027,
+ "learning_rate": 1.872230147595512e-05,
+ "loss": 0.6115,
+ "step": 6892
+ },
+ {
+ "epoch": 18.78201634877384,
+ "grad_norm": 6.131310939788818,
+ "learning_rate": 1.872186981769038e-05,
+ "loss": 0.5777,
+ "step": 6893
+ },
+ {
+ "epoch": 18.78474114441417,
+ "grad_norm": 6.669586658477783,
+ "learning_rate": 1.8721438091500003e-05,
+ "loss": 0.7468,
+ "step": 6894
+ },
+ {
+ "epoch": 18.787465940054496,
+ "grad_norm": 6.523062705993652,
+ "learning_rate": 1.8721006297387348e-05,
+ "loss": 0.9042,
+ "step": 6895
+ },
+ {
+ "epoch": 18.79019073569482,
+ "grad_norm": 7.425229072570801,
+ "learning_rate": 1.8720574435355776e-05,
+ "loss": 0.6797,
+ "step": 6896
+ },
+ {
+ "epoch": 18.79291553133515,
+ "grad_norm": 5.965880393981934,
+ "learning_rate": 1.8720142505408648e-05,
+ "loss": 0.7054,
+ "step": 6897
+ },
+ {
+ "epoch": 18.795640326975477,
+ "grad_norm": 5.94712495803833,
+ "learning_rate": 1.8719710507549332e-05,
+ "loss": 0.6583,
+ "step": 6898
+ },
+ {
+ "epoch": 18.798365122615802,
+ "grad_norm": 7.191242694854736,
+ "learning_rate": 1.8719278441781193e-05,
+ "loss": 0.8247,
+ "step": 6899
+ },
+ {
+ "epoch": 18.80108991825613,
+ "grad_norm": 6.155765056610107,
+ "learning_rate": 1.871884630810759e-05,
+ "loss": 0.6423,
+ "step": 6900
+ },
+ {
+ "epoch": 18.803814713896458,
+ "grad_norm": 5.80852746963501,
+ "learning_rate": 1.8718414106531898e-05,
+ "loss": 0.6028,
+ "step": 6901
+ },
+ {
+ "epoch": 18.806539509536783,
+ "grad_norm": 6.5575337409973145,
+ "learning_rate": 1.8717981837057475e-05,
+ "loss": 0.7118,
+ "step": 6902
+ },
+ {
+ "epoch": 18.809264305177113,
+ "grad_norm": 5.630735397338867,
+ "learning_rate": 1.871754949968769e-05,
+ "loss": 0.678,
+ "step": 6903
+ },
+ {
+ "epoch": 18.81198910081744,
+ "grad_norm": 6.404414176940918,
+ "learning_rate": 1.871711709442591e-05,
+ "loss": 0.6389,
+ "step": 6904
+ },
+ {
+ "epoch": 18.814713896457764,
+ "grad_norm": 7.043416976928711,
+ "learning_rate": 1.87166846212755e-05,
+ "loss": 0.7106,
+ "step": 6905
+ },
+ {
+ "epoch": 18.817438692098094,
+ "grad_norm": 6.194410800933838,
+ "learning_rate": 1.8716252080239835e-05,
+ "loss": 0.6476,
+ "step": 6906
+ },
+ {
+ "epoch": 18.82016348773842,
+ "grad_norm": 6.904215335845947,
+ "learning_rate": 1.8715819471322278e-05,
+ "loss": 0.8463,
+ "step": 6907
+ },
+ {
+ "epoch": 18.822888283378745,
+ "grad_norm": 7.967529296875,
+ "learning_rate": 1.87153867945262e-05,
+ "loss": 0.7247,
+ "step": 6908
+ },
+ {
+ "epoch": 18.825613079019075,
+ "grad_norm": 7.148397922515869,
+ "learning_rate": 1.871495404985497e-05,
+ "loss": 0.6653,
+ "step": 6909
+ },
+ {
+ "epoch": 18.8283378746594,
+ "grad_norm": 5.634891986846924,
+ "learning_rate": 1.8714521237311956e-05,
+ "loss": 0.6337,
+ "step": 6910
+ },
+ {
+ "epoch": 18.831062670299726,
+ "grad_norm": 5.634641170501709,
+ "learning_rate": 1.8714088356900536e-05,
+ "loss": 0.7933,
+ "step": 6911
+ },
+ {
+ "epoch": 18.833787465940055,
+ "grad_norm": 7.6832780838012695,
+ "learning_rate": 1.8713655408624073e-05,
+ "loss": 0.7045,
+ "step": 6912
+ },
+ {
+ "epoch": 18.83651226158038,
+ "grad_norm": 7.616636276245117,
+ "learning_rate": 1.8713222392485943e-05,
+ "loss": 0.6187,
+ "step": 6913
+ },
+ {
+ "epoch": 18.839237057220707,
+ "grad_norm": 6.521064758300781,
+ "learning_rate": 1.8712789308489516e-05,
+ "loss": 0.6803,
+ "step": 6914
+ },
+ {
+ "epoch": 18.841961852861036,
+ "grad_norm": 6.8637919425964355,
+ "learning_rate": 1.8712356156638164e-05,
+ "loss": 0.6851,
+ "step": 6915
+ },
+ {
+ "epoch": 18.844686648501362,
+ "grad_norm": 6.525694370269775,
+ "learning_rate": 1.871192293693527e-05,
+ "loss": 0.7714,
+ "step": 6916
+ },
+ {
+ "epoch": 18.847411444141688,
+ "grad_norm": 7.482306003570557,
+ "learning_rate": 1.8711489649384194e-05,
+ "loss": 0.6254,
+ "step": 6917
+ },
+ {
+ "epoch": 18.850136239782017,
+ "grad_norm": 6.487865447998047,
+ "learning_rate": 1.8711056293988318e-05,
+ "loss": 0.7349,
+ "step": 6918
+ },
+ {
+ "epoch": 18.852861035422343,
+ "grad_norm": 6.945767402648926,
+ "learning_rate": 1.871062287075102e-05,
+ "loss": 0.5544,
+ "step": 6919
+ },
+ {
+ "epoch": 18.85558583106267,
+ "grad_norm": 7.753935813903809,
+ "learning_rate": 1.8710189379675667e-05,
+ "loss": 0.7583,
+ "step": 6920
+ },
+ {
+ "epoch": 18.858310626703,
+ "grad_norm": 8.071741104125977,
+ "learning_rate": 1.870975582076564e-05,
+ "loss": 0.6706,
+ "step": 6921
+ },
+ {
+ "epoch": 18.861035422343324,
+ "grad_norm": 6.120944499969482,
+ "learning_rate": 1.870932219402432e-05,
+ "loss": 0.7295,
+ "step": 6922
+ },
+ {
+ "epoch": 18.86376021798365,
+ "grad_norm": 6.6758623123168945,
+ "learning_rate": 1.8708888499455075e-05,
+ "loss": 0.6644,
+ "step": 6923
+ },
+ {
+ "epoch": 18.86648501362398,
+ "grad_norm": 6.314319610595703,
+ "learning_rate": 1.8708454737061284e-05,
+ "loss": 0.6884,
+ "step": 6924
+ },
+ {
+ "epoch": 18.869209809264305,
+ "grad_norm": 6.211917877197266,
+ "learning_rate": 1.870802090684633e-05,
+ "loss": 0.549,
+ "step": 6925
+ },
+ {
+ "epoch": 18.87193460490463,
+ "grad_norm": 7.109409809112549,
+ "learning_rate": 1.8707587008813593e-05,
+ "loss": 0.6519,
+ "step": 6926
+ },
+ {
+ "epoch": 18.87465940054496,
+ "grad_norm": 6.58275032043457,
+ "learning_rate": 1.8707153042966445e-05,
+ "loss": 0.7092,
+ "step": 6927
+ },
+ {
+ "epoch": 18.877384196185286,
+ "grad_norm": 7.7323408126831055,
+ "learning_rate": 1.870671900930827e-05,
+ "loss": 0.8113,
+ "step": 6928
+ },
+ {
+ "epoch": 18.88010899182561,
+ "grad_norm": 6.093190670013428,
+ "learning_rate": 1.8706284907842446e-05,
+ "loss": 0.7377,
+ "step": 6929
+ },
+ {
+ "epoch": 18.88283378746594,
+ "grad_norm": 7.627821922302246,
+ "learning_rate": 1.8705850738572357e-05,
+ "loss": 0.5571,
+ "step": 6930
+ },
+ {
+ "epoch": 18.885558583106267,
+ "grad_norm": 8.167842864990234,
+ "learning_rate": 1.8705416501501383e-05,
+ "loss": 0.5842,
+ "step": 6931
+ },
+ {
+ "epoch": 18.888283378746593,
+ "grad_norm": 5.880207061767578,
+ "learning_rate": 1.8704982196632904e-05,
+ "loss": 0.5385,
+ "step": 6932
+ },
+ {
+ "epoch": 18.891008174386922,
+ "grad_norm": 8.070178985595703,
+ "learning_rate": 1.8704547823970306e-05,
+ "loss": 0.7018,
+ "step": 6933
+ },
+ {
+ "epoch": 18.893732970027248,
+ "grad_norm": 7.765368938446045,
+ "learning_rate": 1.8704113383516967e-05,
+ "loss": 0.6935,
+ "step": 6934
+ },
+ {
+ "epoch": 18.896457765667574,
+ "grad_norm": 6.145782947540283,
+ "learning_rate": 1.8703678875276277e-05,
+ "loss": 0.6048,
+ "step": 6935
+ },
+ {
+ "epoch": 18.899182561307903,
+ "grad_norm": 8.724090576171875,
+ "learning_rate": 1.8703244299251612e-05,
+ "loss": 0.6954,
+ "step": 6936
+ },
+ {
+ "epoch": 18.90190735694823,
+ "grad_norm": 8.859442710876465,
+ "learning_rate": 1.8702809655446364e-05,
+ "loss": 0.88,
+ "step": 6937
+ },
+ {
+ "epoch": 18.904632152588555,
+ "grad_norm": 10.67265510559082,
+ "learning_rate": 1.8702374943863912e-05,
+ "loss": 0.6643,
+ "step": 6938
+ },
+ {
+ "epoch": 18.907356948228884,
+ "grad_norm": 7.071560382843018,
+ "learning_rate": 1.8701940164507646e-05,
+ "loss": 0.8211,
+ "step": 6939
+ },
+ {
+ "epoch": 18.91008174386921,
+ "grad_norm": 5.7907304763793945,
+ "learning_rate": 1.870150531738095e-05,
+ "loss": 0.6726,
+ "step": 6940
+ },
+ {
+ "epoch": 18.912806539509535,
+ "grad_norm": 6.529600620269775,
+ "learning_rate": 1.8701070402487208e-05,
+ "loss": 0.7419,
+ "step": 6941
+ },
+ {
+ "epoch": 18.915531335149865,
+ "grad_norm": 9.500993728637695,
+ "learning_rate": 1.8700635419829808e-05,
+ "loss": 0.8208,
+ "step": 6942
+ },
+ {
+ "epoch": 18.91825613079019,
+ "grad_norm": 7.0881733894348145,
+ "learning_rate": 1.8700200369412143e-05,
+ "loss": 0.8892,
+ "step": 6943
+ },
+ {
+ "epoch": 18.920980926430516,
+ "grad_norm": 7.655481815338135,
+ "learning_rate": 1.86997652512376e-05,
+ "loss": 0.6145,
+ "step": 6944
+ },
+ {
+ "epoch": 18.923705722070846,
+ "grad_norm": 6.836276054382324,
+ "learning_rate": 1.869933006530956e-05,
+ "loss": 0.6143,
+ "step": 6945
+ },
+ {
+ "epoch": 18.92643051771117,
+ "grad_norm": 6.524843692779541,
+ "learning_rate": 1.869889481163142e-05,
+ "loss": 0.6563,
+ "step": 6946
+ },
+ {
+ "epoch": 18.929155313351497,
+ "grad_norm": 10.126830101013184,
+ "learning_rate": 1.8698459490206568e-05,
+ "loss": 0.7102,
+ "step": 6947
+ },
+ {
+ "epoch": 18.931880108991827,
+ "grad_norm": 6.692455768585205,
+ "learning_rate": 1.8698024101038395e-05,
+ "loss": 0.5805,
+ "step": 6948
+ },
+ {
+ "epoch": 18.934604904632153,
+ "grad_norm": 6.551772117614746,
+ "learning_rate": 1.8697588644130283e-05,
+ "loss": 0.7601,
+ "step": 6949
+ },
+ {
+ "epoch": 18.93732970027248,
+ "grad_norm": 7.799405097961426,
+ "learning_rate": 1.8697153119485638e-05,
+ "loss": 0.8154,
+ "step": 6950
+ },
+ {
+ "epoch": 18.940054495912808,
+ "grad_norm": 6.703145980834961,
+ "learning_rate": 1.869671752710784e-05,
+ "loss": 0.8209,
+ "step": 6951
+ },
+ {
+ "epoch": 18.942779291553133,
+ "grad_norm": 7.078989028930664,
+ "learning_rate": 1.869628186700029e-05,
+ "loss": 0.657,
+ "step": 6952
+ },
+ {
+ "epoch": 18.94550408719346,
+ "grad_norm": 7.225976467132568,
+ "learning_rate": 1.8695846139166372e-05,
+ "loss": 0.7543,
+ "step": 6953
+ },
+ {
+ "epoch": 18.94822888283379,
+ "grad_norm": 6.803812503814697,
+ "learning_rate": 1.8695410343609487e-05,
+ "loss": 0.625,
+ "step": 6954
+ },
+ {
+ "epoch": 18.950953678474114,
+ "grad_norm": 7.01662015914917,
+ "learning_rate": 1.8694974480333027e-05,
+ "loss": 0.674,
+ "step": 6955
+ },
+ {
+ "epoch": 18.95367847411444,
+ "grad_norm": 7.047749996185303,
+ "learning_rate": 1.8694538549340385e-05,
+ "loss": 0.5236,
+ "step": 6956
+ },
+ {
+ "epoch": 18.95640326975477,
+ "grad_norm": 6.538590431213379,
+ "learning_rate": 1.8694102550634953e-05,
+ "loss": 0.7423,
+ "step": 6957
+ },
+ {
+ "epoch": 18.959128065395095,
+ "grad_norm": 7.644791603088379,
+ "learning_rate": 1.8693666484220134e-05,
+ "loss": 0.8859,
+ "step": 6958
+ },
+ {
+ "epoch": 18.96185286103542,
+ "grad_norm": 5.7876715660095215,
+ "learning_rate": 1.8693230350099318e-05,
+ "loss": 0.5617,
+ "step": 6959
+ },
+ {
+ "epoch": 18.96457765667575,
+ "grad_norm": 6.724064826965332,
+ "learning_rate": 1.8692794148275904e-05,
+ "loss": 0.6246,
+ "step": 6960
+ },
+ {
+ "epoch": 18.967302452316076,
+ "grad_norm": 7.128213405609131,
+ "learning_rate": 1.869235787875329e-05,
+ "loss": 0.584,
+ "step": 6961
+ },
+ {
+ "epoch": 18.970027247956402,
+ "grad_norm": 7.456297397613525,
+ "learning_rate": 1.8691921541534876e-05,
+ "loss": 0.6891,
+ "step": 6962
+ },
+ {
+ "epoch": 18.97275204359673,
+ "grad_norm": 8.220142364501953,
+ "learning_rate": 1.869148513662405e-05,
+ "loss": 0.7355,
+ "step": 6963
+ },
+ {
+ "epoch": 18.975476839237057,
+ "grad_norm": 6.242100238800049,
+ "learning_rate": 1.869104866402422e-05,
+ "loss": 0.7303,
+ "step": 6964
+ },
+ {
+ "epoch": 18.978201634877383,
+ "grad_norm": 11.183067321777344,
+ "learning_rate": 1.8690612123738784e-05,
+ "loss": 0.6638,
+ "step": 6965
+ },
+ {
+ "epoch": 18.980926430517712,
+ "grad_norm": 6.605279922485352,
+ "learning_rate": 1.869017551577114e-05,
+ "loss": 0.6329,
+ "step": 6966
+ },
+ {
+ "epoch": 18.983651226158038,
+ "grad_norm": 7.174684047698975,
+ "learning_rate": 1.8689738840124688e-05,
+ "loss": 0.5535,
+ "step": 6967
+ },
+ {
+ "epoch": 18.986376021798364,
+ "grad_norm": 7.133973121643066,
+ "learning_rate": 1.868930209680283e-05,
+ "loss": 0.5927,
+ "step": 6968
+ },
+ {
+ "epoch": 18.989100817438693,
+ "grad_norm": 7.753392219543457,
+ "learning_rate": 1.8688865285808968e-05,
+ "loss": 0.9243,
+ "step": 6969
+ },
+ {
+ "epoch": 18.99182561307902,
+ "grad_norm": 6.787723064422607,
+ "learning_rate": 1.8688428407146504e-05,
+ "loss": 0.6735,
+ "step": 6970
+ },
+ {
+ "epoch": 18.994550408719345,
+ "grad_norm": 6.459990501403809,
+ "learning_rate": 1.8687991460818836e-05,
+ "loss": 0.6132,
+ "step": 6971
+ },
+ {
+ "epoch": 18.997275204359674,
+ "grad_norm": 6.484928131103516,
+ "learning_rate": 1.868755444682937e-05,
+ "loss": 0.5814,
+ "step": 6972
+ },
+ {
+ "epoch": 19.0,
+ "grad_norm": 5.527313232421875,
+ "learning_rate": 1.8687117365181514e-05,
+ "loss": 0.6041,
+ "step": 6973
+ },
+ {
+ "epoch": 19.002724795640326,
+ "grad_norm": 6.904544353485107,
+ "learning_rate": 1.8686680215878663e-05,
+ "loss": 0.6171,
+ "step": 6974
+ },
+ {
+ "epoch": 19.005449591280655,
+ "grad_norm": 7.0227251052856445,
+ "learning_rate": 1.8686242998924226e-05,
+ "loss": 0.5375,
+ "step": 6975
+ },
+ {
+ "epoch": 19.00817438692098,
+ "grad_norm": 8.04870319366455,
+ "learning_rate": 1.868580571432161e-05,
+ "loss": 0.7644,
+ "step": 6976
+ },
+ {
+ "epoch": 19.010899182561307,
+ "grad_norm": 5.930017948150635,
+ "learning_rate": 1.868536836207422e-05,
+ "loss": 0.6938,
+ "step": 6977
+ },
+ {
+ "epoch": 19.013623978201636,
+ "grad_norm": 5.56489372253418,
+ "learning_rate": 1.8684930942185457e-05,
+ "loss": 0.5797,
+ "step": 6978
+ },
+ {
+ "epoch": 19.016348773841962,
+ "grad_norm": 7.7298431396484375,
+ "learning_rate": 1.8684493454658736e-05,
+ "loss": 0.5956,
+ "step": 6979
+ },
+ {
+ "epoch": 19.019073569482288,
+ "grad_norm": 7.636287212371826,
+ "learning_rate": 1.8684055899497456e-05,
+ "loss": 0.6725,
+ "step": 6980
+ },
+ {
+ "epoch": 19.021798365122617,
+ "grad_norm": 6.970780372619629,
+ "learning_rate": 1.8683618276705028e-05,
+ "loss": 0.6855,
+ "step": 6981
+ },
+ {
+ "epoch": 19.024523160762943,
+ "grad_norm": 7.067695617675781,
+ "learning_rate": 1.8683180586284862e-05,
+ "loss": 0.6902,
+ "step": 6982
+ },
+ {
+ "epoch": 19.02724795640327,
+ "grad_norm": 5.289407730102539,
+ "learning_rate": 1.8682742828240364e-05,
+ "loss": 0.667,
+ "step": 6983
+ },
+ {
+ "epoch": 19.029972752043598,
+ "grad_norm": 6.511101722717285,
+ "learning_rate": 1.8682305002574942e-05,
+ "loss": 0.8411,
+ "step": 6984
+ },
+ {
+ "epoch": 19.032697547683924,
+ "grad_norm": 6.793363571166992,
+ "learning_rate": 1.8681867109292012e-05,
+ "loss": 0.5453,
+ "step": 6985
+ },
+ {
+ "epoch": 19.03542234332425,
+ "grad_norm": 7.129678726196289,
+ "learning_rate": 1.868142914839498e-05,
+ "loss": 0.6685,
+ "step": 6986
+ },
+ {
+ "epoch": 19.03814713896458,
+ "grad_norm": 5.837164878845215,
+ "learning_rate": 1.8680991119887256e-05,
+ "loss": 0.6599,
+ "step": 6987
+ },
+ {
+ "epoch": 19.040871934604905,
+ "grad_norm": 6.888824939727783,
+ "learning_rate": 1.868055302377225e-05,
+ "loss": 0.6969,
+ "step": 6988
+ },
+ {
+ "epoch": 19.04359673024523,
+ "grad_norm": 8.182014465332031,
+ "learning_rate": 1.8680114860053384e-05,
+ "loss": 0.5854,
+ "step": 6989
+ },
+ {
+ "epoch": 19.04632152588556,
+ "grad_norm": 8.802042961120605,
+ "learning_rate": 1.867967662873406e-05,
+ "loss": 0.708,
+ "step": 6990
+ },
+ {
+ "epoch": 19.049046321525886,
+ "grad_norm": 6.499661922454834,
+ "learning_rate": 1.867923832981769e-05,
+ "loss": 0.6157,
+ "step": 6991
+ },
+ {
+ "epoch": 19.05177111716621,
+ "grad_norm": 7.046994686126709,
+ "learning_rate": 1.8678799963307693e-05,
+ "loss": 0.5677,
+ "step": 6992
+ },
+ {
+ "epoch": 19.05449591280654,
+ "grad_norm": 6.15169095993042,
+ "learning_rate": 1.867836152920748e-05,
+ "loss": 0.5699,
+ "step": 6993
+ },
+ {
+ "epoch": 19.057220708446867,
+ "grad_norm": 6.4529008865356445,
+ "learning_rate": 1.867792302752047e-05,
+ "loss": 0.6541,
+ "step": 6994
+ },
+ {
+ "epoch": 19.059945504087192,
+ "grad_norm": 6.4599385261535645,
+ "learning_rate": 1.867748445825007e-05,
+ "loss": 0.5459,
+ "step": 6995
+ },
+ {
+ "epoch": 19.06267029972752,
+ "grad_norm": 6.972471714019775,
+ "learning_rate": 1.8677045821399705e-05,
+ "loss": 0.5957,
+ "step": 6996
+ },
+ {
+ "epoch": 19.065395095367847,
+ "grad_norm": 9.830824851989746,
+ "learning_rate": 1.8676607116972786e-05,
+ "loss": 0.6025,
+ "step": 6997
+ },
+ {
+ "epoch": 19.068119891008173,
+ "grad_norm": 7.8514909744262695,
+ "learning_rate": 1.8676168344972726e-05,
+ "loss": 0.741,
+ "step": 6998
+ },
+ {
+ "epoch": 19.070844686648503,
+ "grad_norm": 7.270137786865234,
+ "learning_rate": 1.867572950540295e-05,
+ "loss": 0.6702,
+ "step": 6999
+ },
+ {
+ "epoch": 19.07356948228883,
+ "grad_norm": 7.0346150398254395,
+ "learning_rate": 1.8675290598266872e-05,
+ "loss": 0.5615,
+ "step": 7000
+ },
+ {
+ "epoch": 19.076294277929154,
+ "grad_norm": 7.0730156898498535,
+ "learning_rate": 1.8674851623567908e-05,
+ "loss": 0.6435,
+ "step": 7001
+ },
+ {
+ "epoch": 19.079019073569484,
+ "grad_norm": 7.088490009307861,
+ "learning_rate": 1.8674412581309477e-05,
+ "loss": 0.7203,
+ "step": 7002
+ },
+ {
+ "epoch": 19.08174386920981,
+ "grad_norm": 6.703186511993408,
+ "learning_rate": 1.8673973471495005e-05,
+ "loss": 0.6553,
+ "step": 7003
+ },
+ {
+ "epoch": 19.084468664850135,
+ "grad_norm": 7.210083961486816,
+ "learning_rate": 1.8673534294127905e-05,
+ "loss": 0.7843,
+ "step": 7004
+ },
+ {
+ "epoch": 19.087193460490465,
+ "grad_norm": 6.837465286254883,
+ "learning_rate": 1.8673095049211597e-05,
+ "loss": 0.5693,
+ "step": 7005
+ },
+ {
+ "epoch": 19.08991825613079,
+ "grad_norm": 8.2601900100708,
+ "learning_rate": 1.8672655736749507e-05,
+ "loss": 0.7108,
+ "step": 7006
+ },
+ {
+ "epoch": 19.092643051771116,
+ "grad_norm": 8.59414005279541,
+ "learning_rate": 1.867221635674505e-05,
+ "loss": 0.5912,
+ "step": 7007
+ },
+ {
+ "epoch": 19.095367847411445,
+ "grad_norm": 6.336438179016113,
+ "learning_rate": 1.867177690920165e-05,
+ "loss": 0.6964,
+ "step": 7008
+ },
+ {
+ "epoch": 19.09809264305177,
+ "grad_norm": 5.364658832550049,
+ "learning_rate": 1.8671337394122733e-05,
+ "loss": 0.4557,
+ "step": 7009
+ },
+ {
+ "epoch": 19.100817438692097,
+ "grad_norm": 6.105134963989258,
+ "learning_rate": 1.867089781151172e-05,
+ "loss": 0.574,
+ "step": 7010
+ },
+ {
+ "epoch": 19.103542234332426,
+ "grad_norm": 10.349735260009766,
+ "learning_rate": 1.8670458161372033e-05,
+ "loss": 0.7322,
+ "step": 7011
+ },
+ {
+ "epoch": 19.106267029972752,
+ "grad_norm": 7.866164684295654,
+ "learning_rate": 1.8670018443707098e-05,
+ "loss": 0.5683,
+ "step": 7012
+ },
+ {
+ "epoch": 19.108991825613078,
+ "grad_norm": 5.974966526031494,
+ "learning_rate": 1.866957865852034e-05,
+ "loss": 0.5554,
+ "step": 7013
+ },
+ {
+ "epoch": 19.111716621253407,
+ "grad_norm": 6.067837715148926,
+ "learning_rate": 1.866913880581518e-05,
+ "loss": 0.6356,
+ "step": 7014
+ },
+ {
+ "epoch": 19.114441416893733,
+ "grad_norm": 6.938493251800537,
+ "learning_rate": 1.8668698885595045e-05,
+ "loss": 0.6473,
+ "step": 7015
+ },
+ {
+ "epoch": 19.11716621253406,
+ "grad_norm": 5.8853068351745605,
+ "learning_rate": 1.866825889786336e-05,
+ "loss": 0.6384,
+ "step": 7016
+ },
+ {
+ "epoch": 19.11989100817439,
+ "grad_norm": 5.926418304443359,
+ "learning_rate": 1.866781884262356e-05,
+ "loss": 0.4824,
+ "step": 7017
+ },
+ {
+ "epoch": 19.122615803814714,
+ "grad_norm": 13.150179862976074,
+ "learning_rate": 1.866737871987906e-05,
+ "loss": 0.6407,
+ "step": 7018
+ },
+ {
+ "epoch": 19.12534059945504,
+ "grad_norm": 7.374396800994873,
+ "learning_rate": 1.86669385296333e-05,
+ "loss": 0.6325,
+ "step": 7019
+ },
+ {
+ "epoch": 19.12806539509537,
+ "grad_norm": 6.097074508666992,
+ "learning_rate": 1.8666498271889698e-05,
+ "loss": 0.4865,
+ "step": 7020
+ },
+ {
+ "epoch": 19.130790190735695,
+ "grad_norm": 6.99901008605957,
+ "learning_rate": 1.8666057946651682e-05,
+ "loss": 0.5145,
+ "step": 7021
+ },
+ {
+ "epoch": 19.13351498637602,
+ "grad_norm": 5.961134433746338,
+ "learning_rate": 1.866561755392269e-05,
+ "loss": 0.6768,
+ "step": 7022
+ },
+ {
+ "epoch": 19.13623978201635,
+ "grad_norm": 6.325680732727051,
+ "learning_rate": 1.866517709370615e-05,
+ "loss": 0.5093,
+ "step": 7023
+ },
+ {
+ "epoch": 19.138964577656676,
+ "grad_norm": 6.764310836791992,
+ "learning_rate": 1.8664736566005487e-05,
+ "loss": 0.5704,
+ "step": 7024
+ },
+ {
+ "epoch": 19.141689373297,
+ "grad_norm": 6.860869884490967,
+ "learning_rate": 1.8664295970824136e-05,
+ "loss": 0.5425,
+ "step": 7025
+ },
+ {
+ "epoch": 19.14441416893733,
+ "grad_norm": 7.356738090515137,
+ "learning_rate": 1.8663855308165524e-05,
+ "loss": 0.82,
+ "step": 7026
+ },
+ {
+ "epoch": 19.147138964577657,
+ "grad_norm": 7.302488803863525,
+ "learning_rate": 1.8663414578033084e-05,
+ "loss": 0.6849,
+ "step": 7027
+ },
+ {
+ "epoch": 19.149863760217983,
+ "grad_norm": 8.088655471801758,
+ "learning_rate": 1.8662973780430256e-05,
+ "loss": 0.7511,
+ "step": 7028
+ },
+ {
+ "epoch": 19.152588555858312,
+ "grad_norm": 7.170904636383057,
+ "learning_rate": 1.8662532915360466e-05,
+ "loss": 0.5874,
+ "step": 7029
+ },
+ {
+ "epoch": 19.155313351498638,
+ "grad_norm": 7.379308223724365,
+ "learning_rate": 1.8662091982827142e-05,
+ "loss": 0.5641,
+ "step": 7030
+ },
+ {
+ "epoch": 19.158038147138964,
+ "grad_norm": 8.560761451721191,
+ "learning_rate": 1.8661650982833733e-05,
+ "loss": 0.5329,
+ "step": 7031
+ },
+ {
+ "epoch": 19.160762942779293,
+ "grad_norm": 6.991546630859375,
+ "learning_rate": 1.866120991538366e-05,
+ "loss": 0.6995,
+ "step": 7032
+ },
+ {
+ "epoch": 19.16348773841962,
+ "grad_norm": 11.478714942932129,
+ "learning_rate": 1.8660768780480364e-05,
+ "loss": 0.4547,
+ "step": 7033
+ },
+ {
+ "epoch": 19.166212534059945,
+ "grad_norm": 8.419322967529297,
+ "learning_rate": 1.866032757812728e-05,
+ "loss": 0.5837,
+ "step": 7034
+ },
+ {
+ "epoch": 19.168937329700274,
+ "grad_norm": 7.4215617179870605,
+ "learning_rate": 1.8659886308327842e-05,
+ "loss": 0.4934,
+ "step": 7035
+ },
+ {
+ "epoch": 19.1716621253406,
+ "grad_norm": 7.458568096160889,
+ "learning_rate": 1.865944497108549e-05,
+ "loss": 0.6769,
+ "step": 7036
+ },
+ {
+ "epoch": 19.174386920980925,
+ "grad_norm": 7.43142032623291,
+ "learning_rate": 1.8659003566403658e-05,
+ "loss": 0.629,
+ "step": 7037
+ },
+ {
+ "epoch": 19.177111716621255,
+ "grad_norm": 6.053524494171143,
+ "learning_rate": 1.8658562094285786e-05,
+ "loss": 0.538,
+ "step": 7038
+ },
+ {
+ "epoch": 19.17983651226158,
+ "grad_norm": 56.13513946533203,
+ "learning_rate": 1.865812055473531e-05,
+ "loss": 0.5398,
+ "step": 7039
+ },
+ {
+ "epoch": 19.182561307901906,
+ "grad_norm": 5.919869899749756,
+ "learning_rate": 1.865767894775567e-05,
+ "loss": 0.7458,
+ "step": 7040
+ },
+ {
+ "epoch": 19.185286103542236,
+ "grad_norm": 5.897231101989746,
+ "learning_rate": 1.8657237273350305e-05,
+ "loss": 0.7715,
+ "step": 7041
+ },
+ {
+ "epoch": 19.18801089918256,
+ "grad_norm": 8.5324068069458,
+ "learning_rate": 1.8656795531522655e-05,
+ "loss": 0.4376,
+ "step": 7042
+ },
+ {
+ "epoch": 19.190735694822887,
+ "grad_norm": 6.6526103019714355,
+ "learning_rate": 1.865635372227616e-05,
+ "loss": 0.7852,
+ "step": 7043
+ },
+ {
+ "epoch": 19.193460490463217,
+ "grad_norm": 7.942925930023193,
+ "learning_rate": 1.865591184561426e-05,
+ "loss": 0.6008,
+ "step": 7044
+ },
+ {
+ "epoch": 19.196185286103542,
+ "grad_norm": 7.121811866760254,
+ "learning_rate": 1.8655469901540398e-05,
+ "loss": 0.6053,
+ "step": 7045
+ },
+ {
+ "epoch": 19.19891008174387,
+ "grad_norm": 6.535566806793213,
+ "learning_rate": 1.8655027890058014e-05,
+ "loss": 0.5255,
+ "step": 7046
+ },
+ {
+ "epoch": 19.201634877384198,
+ "grad_norm": 6.859278678894043,
+ "learning_rate": 1.8654585811170553e-05,
+ "loss": 0.5909,
+ "step": 7047
+ },
+ {
+ "epoch": 19.204359673024523,
+ "grad_norm": 13.454601287841797,
+ "learning_rate": 1.8654143664881455e-05,
+ "loss": 0.731,
+ "step": 7048
+ },
+ {
+ "epoch": 19.20708446866485,
+ "grad_norm": 8.128264427185059,
+ "learning_rate": 1.8653701451194166e-05,
+ "loss": 0.6651,
+ "step": 7049
+ },
+ {
+ "epoch": 19.20980926430518,
+ "grad_norm": 8.000060081481934,
+ "learning_rate": 1.8653259170112128e-05,
+ "loss": 0.6178,
+ "step": 7050
+ },
+ {
+ "epoch": 19.212534059945504,
+ "grad_norm": 9.531864166259766,
+ "learning_rate": 1.8652816821638783e-05,
+ "loss": 0.6243,
+ "step": 7051
+ },
+ {
+ "epoch": 19.21525885558583,
+ "grad_norm": 8.216516494750977,
+ "learning_rate": 1.8652374405777584e-05,
+ "loss": 0.6959,
+ "step": 7052
+ },
+ {
+ "epoch": 19.21798365122616,
+ "grad_norm": 8.569267272949219,
+ "learning_rate": 1.865193192253197e-05,
+ "loss": 0.6239,
+ "step": 7053
+ },
+ {
+ "epoch": 19.220708446866485,
+ "grad_norm": 6.105313301086426,
+ "learning_rate": 1.8651489371905386e-05,
+ "loss": 0.4784,
+ "step": 7054
+ },
+ {
+ "epoch": 19.22343324250681,
+ "grad_norm": 7.686140060424805,
+ "learning_rate": 1.8651046753901286e-05,
+ "loss": 0.6089,
+ "step": 7055
+ },
+ {
+ "epoch": 19.22615803814714,
+ "grad_norm": 7.090569972991943,
+ "learning_rate": 1.8650604068523107e-05,
+ "loss": 0.7593,
+ "step": 7056
+ },
+ {
+ "epoch": 19.228882833787466,
+ "grad_norm": 7.714623928070068,
+ "learning_rate": 1.8650161315774306e-05,
+ "loss": 0.6705,
+ "step": 7057
+ },
+ {
+ "epoch": 19.231607629427792,
+ "grad_norm": 8.148009300231934,
+ "learning_rate": 1.8649718495658325e-05,
+ "loss": 0.7399,
+ "step": 7058
+ },
+ {
+ "epoch": 19.23433242506812,
+ "grad_norm": 7.166345596313477,
+ "learning_rate": 1.8649275608178617e-05,
+ "loss": 0.7351,
+ "step": 7059
+ },
+ {
+ "epoch": 19.237057220708447,
+ "grad_norm": 7.013566970825195,
+ "learning_rate": 1.8648832653338628e-05,
+ "loss": 0.7634,
+ "step": 7060
+ },
+ {
+ "epoch": 19.239782016348773,
+ "grad_norm": 6.877170085906982,
+ "learning_rate": 1.864838963114181e-05,
+ "loss": 0.5928,
+ "step": 7061
+ },
+ {
+ "epoch": 19.242506811989102,
+ "grad_norm": 6.479413032531738,
+ "learning_rate": 1.8647946541591607e-05,
+ "loss": 0.7411,
+ "step": 7062
+ },
+ {
+ "epoch": 19.245231607629428,
+ "grad_norm": 6.917880058288574,
+ "learning_rate": 1.8647503384691477e-05,
+ "loss": 0.5546,
+ "step": 7063
+ },
+ {
+ "epoch": 19.247956403269754,
+ "grad_norm": 6.818930149078369,
+ "learning_rate": 1.864706016044487e-05,
+ "loss": 0.564,
+ "step": 7064
+ },
+ {
+ "epoch": 19.250681198910083,
+ "grad_norm": 8.061506271362305,
+ "learning_rate": 1.8646616868855236e-05,
+ "loss": 0.5805,
+ "step": 7065
+ },
+ {
+ "epoch": 19.25340599455041,
+ "grad_norm": 7.993408679962158,
+ "learning_rate": 1.864617350992603e-05,
+ "loss": 0.581,
+ "step": 7066
+ },
+ {
+ "epoch": 19.256130790190735,
+ "grad_norm": 7.914290428161621,
+ "learning_rate": 1.86457300836607e-05,
+ "loss": 0.6129,
+ "step": 7067
+ },
+ {
+ "epoch": 19.258855585831064,
+ "grad_norm": 13.053926467895508,
+ "learning_rate": 1.8645286590062705e-05,
+ "loss": 0.5899,
+ "step": 7068
+ },
+ {
+ "epoch": 19.26158038147139,
+ "grad_norm": 7.743175983428955,
+ "learning_rate": 1.8644843029135496e-05,
+ "loss": 0.5933,
+ "step": 7069
+ },
+ {
+ "epoch": 19.264305177111716,
+ "grad_norm": 8.46878719329834,
+ "learning_rate": 1.8644399400882525e-05,
+ "loss": 0.5908,
+ "step": 7070
+ },
+ {
+ "epoch": 19.267029972752045,
+ "grad_norm": 6.0180768966674805,
+ "learning_rate": 1.8643955705307252e-05,
+ "loss": 0.495,
+ "step": 7071
+ },
+ {
+ "epoch": 19.26975476839237,
+ "grad_norm": 15.613967895507812,
+ "learning_rate": 1.864351194241313e-05,
+ "loss": 0.701,
+ "step": 7072
+ },
+ {
+ "epoch": 19.272479564032697,
+ "grad_norm": 6.134161472320557,
+ "learning_rate": 1.8643068112203612e-05,
+ "loss": 0.5889,
+ "step": 7073
+ },
+ {
+ "epoch": 19.275204359673026,
+ "grad_norm": 7.331305503845215,
+ "learning_rate": 1.864262421468216e-05,
+ "loss": 0.6984,
+ "step": 7074
+ },
+ {
+ "epoch": 19.277929155313352,
+ "grad_norm": 8.162590980529785,
+ "learning_rate": 1.864218024985223e-05,
+ "loss": 0.6718,
+ "step": 7075
+ },
+ {
+ "epoch": 19.280653950953678,
+ "grad_norm": 13.397427558898926,
+ "learning_rate": 1.864173621771728e-05,
+ "loss": 0.5046,
+ "step": 7076
+ },
+ {
+ "epoch": 19.283378746594007,
+ "grad_norm": 7.846876621246338,
+ "learning_rate": 1.8641292118280764e-05,
+ "loss": 0.6979,
+ "step": 7077
+ },
+ {
+ "epoch": 19.286103542234333,
+ "grad_norm": 7.323906898498535,
+ "learning_rate": 1.8640847951546145e-05,
+ "loss": 0.6589,
+ "step": 7078
+ },
+ {
+ "epoch": 19.28882833787466,
+ "grad_norm": 7.374205589294434,
+ "learning_rate": 1.8640403717516877e-05,
+ "loss": 0.6165,
+ "step": 7079
+ },
+ {
+ "epoch": 19.291553133514988,
+ "grad_norm": 6.858243465423584,
+ "learning_rate": 1.8639959416196424e-05,
+ "loss": 0.5265,
+ "step": 7080
+ },
+ {
+ "epoch": 19.294277929155314,
+ "grad_norm": 6.692946910858154,
+ "learning_rate": 1.8639515047588247e-05,
+ "loss": 0.5421,
+ "step": 7081
+ },
+ {
+ "epoch": 19.29700272479564,
+ "grad_norm": 8.978104591369629,
+ "learning_rate": 1.86390706116958e-05,
+ "loss": 0.6698,
+ "step": 7082
+ },
+ {
+ "epoch": 19.29972752043597,
+ "grad_norm": 9.767349243164062,
+ "learning_rate": 1.863862610852255e-05,
+ "loss": 0.7329,
+ "step": 7083
+ },
+ {
+ "epoch": 19.302452316076295,
+ "grad_norm": 6.10684061050415,
+ "learning_rate": 1.8638181538071966e-05,
+ "loss": 0.5245,
+ "step": 7084
+ },
+ {
+ "epoch": 19.30517711171662,
+ "grad_norm": 8.089315414428711,
+ "learning_rate": 1.8637736900347496e-05,
+ "loss": 0.719,
+ "step": 7085
+ },
+ {
+ "epoch": 19.30790190735695,
+ "grad_norm": 7.747014045715332,
+ "learning_rate": 1.863729219535261e-05,
+ "loss": 0.8131,
+ "step": 7086
+ },
+ {
+ "epoch": 19.310626702997276,
+ "grad_norm": 5.898434638977051,
+ "learning_rate": 1.8636847423090768e-05,
+ "loss": 0.5937,
+ "step": 7087
+ },
+ {
+ "epoch": 19.3133514986376,
+ "grad_norm": 7.791942119598389,
+ "learning_rate": 1.863640258356544e-05,
+ "loss": 0.6298,
+ "step": 7088
+ },
+ {
+ "epoch": 19.31607629427793,
+ "grad_norm": 10.149212837219238,
+ "learning_rate": 1.8635957676780085e-05,
+ "loss": 0.6246,
+ "step": 7089
+ },
+ {
+ "epoch": 19.318801089918257,
+ "grad_norm": 8.726940155029297,
+ "learning_rate": 1.8635512702738168e-05,
+ "loss": 0.786,
+ "step": 7090
+ },
+ {
+ "epoch": 19.321525885558582,
+ "grad_norm": 6.904002666473389,
+ "learning_rate": 1.8635067661443157e-05,
+ "loss": 0.6941,
+ "step": 7091
+ },
+ {
+ "epoch": 19.32425068119891,
+ "grad_norm": 5.512078285217285,
+ "learning_rate": 1.8634622552898516e-05,
+ "loss": 0.9141,
+ "step": 7092
+ },
+ {
+ "epoch": 19.326975476839237,
+ "grad_norm": 7.560044765472412,
+ "learning_rate": 1.8634177377107717e-05,
+ "loss": 0.6692,
+ "step": 7093
+ },
+ {
+ "epoch": 19.329700272479563,
+ "grad_norm": 6.593295097351074,
+ "learning_rate": 1.863373213407422e-05,
+ "loss": 0.5018,
+ "step": 7094
+ },
+ {
+ "epoch": 19.332425068119893,
+ "grad_norm": 6.250489711761475,
+ "learning_rate": 1.8633286823801495e-05,
+ "loss": 0.6336,
+ "step": 7095
+ },
+ {
+ "epoch": 19.33514986376022,
+ "grad_norm": 7.044358253479004,
+ "learning_rate": 1.8632841446293008e-05,
+ "loss": 0.6442,
+ "step": 7096
+ },
+ {
+ "epoch": 19.337874659400544,
+ "grad_norm": 7.508694171905518,
+ "learning_rate": 1.8632396001552232e-05,
+ "loss": 0.5732,
+ "step": 7097
+ },
+ {
+ "epoch": 19.340599455040874,
+ "grad_norm": 7.475589275360107,
+ "learning_rate": 1.863195048958263e-05,
+ "loss": 0.7311,
+ "step": 7098
+ },
+ {
+ "epoch": 19.3433242506812,
+ "grad_norm": 7.98246955871582,
+ "learning_rate": 1.863150491038768e-05,
+ "loss": 0.6296,
+ "step": 7099
+ },
+ {
+ "epoch": 19.346049046321525,
+ "grad_norm": 7.1813883781433105,
+ "learning_rate": 1.863105926397085e-05,
+ "loss": 0.6597,
+ "step": 7100
+ },
+ {
+ "epoch": 19.348773841961854,
+ "grad_norm": 14.236019134521484,
+ "learning_rate": 1.86306135503356e-05,
+ "loss": 0.5901,
+ "step": 7101
+ },
+ {
+ "epoch": 19.35149863760218,
+ "grad_norm": 6.5680060386657715,
+ "learning_rate": 1.8630167769485417e-05,
+ "loss": 0.6921,
+ "step": 7102
+ },
+ {
+ "epoch": 19.354223433242506,
+ "grad_norm": 7.416837215423584,
+ "learning_rate": 1.862972192142376e-05,
+ "loss": 0.6556,
+ "step": 7103
+ },
+ {
+ "epoch": 19.356948228882835,
+ "grad_norm": 7.153679370880127,
+ "learning_rate": 1.862927600615411e-05,
+ "loss": 0.5911,
+ "step": 7104
+ },
+ {
+ "epoch": 19.35967302452316,
+ "grad_norm": 7.799759864807129,
+ "learning_rate": 1.862883002367993e-05,
+ "loss": 0.5865,
+ "step": 7105
+ },
+ {
+ "epoch": 19.362397820163487,
+ "grad_norm": 7.206714153289795,
+ "learning_rate": 1.8628383974004707e-05,
+ "loss": 0.7263,
+ "step": 7106
+ },
+ {
+ "epoch": 19.365122615803816,
+ "grad_norm": 7.409224987030029,
+ "learning_rate": 1.8627937857131904e-05,
+ "loss": 0.6926,
+ "step": 7107
+ },
+ {
+ "epoch": 19.367847411444142,
+ "grad_norm": 9.9913911819458,
+ "learning_rate": 1.8627491673065e-05,
+ "loss": 0.6636,
+ "step": 7108
+ },
+ {
+ "epoch": 19.370572207084468,
+ "grad_norm": 6.525021076202393,
+ "learning_rate": 1.8627045421807466e-05,
+ "loss": 0.6033,
+ "step": 7109
+ },
+ {
+ "epoch": 19.373297002724797,
+ "grad_norm": 8.829724311828613,
+ "learning_rate": 1.8626599103362783e-05,
+ "loss": 0.5574,
+ "step": 7110
+ },
+ {
+ "epoch": 19.376021798365123,
+ "grad_norm": 6.547801494598389,
+ "learning_rate": 1.862615271773442e-05,
+ "loss": 0.5632,
+ "step": 7111
+ },
+ {
+ "epoch": 19.37874659400545,
+ "grad_norm": 5.6741557121276855,
+ "learning_rate": 1.862570626492586e-05,
+ "loss": 0.4953,
+ "step": 7112
+ },
+ {
+ "epoch": 19.381471389645778,
+ "grad_norm": 5.066644191741943,
+ "learning_rate": 1.8625259744940578e-05,
+ "loss": 0.5991,
+ "step": 7113
+ },
+ {
+ "epoch": 19.384196185286104,
+ "grad_norm": 6.425942897796631,
+ "learning_rate": 1.862481315778205e-05,
+ "loss": 0.634,
+ "step": 7114
+ },
+ {
+ "epoch": 19.38692098092643,
+ "grad_norm": 6.192266464233398,
+ "learning_rate": 1.8624366503453752e-05,
+ "loss": 0.7917,
+ "step": 7115
+ },
+ {
+ "epoch": 19.38964577656676,
+ "grad_norm": 5.9544677734375,
+ "learning_rate": 1.862391978195917e-05,
+ "loss": 0.5447,
+ "step": 7116
+ },
+ {
+ "epoch": 19.392370572207085,
+ "grad_norm": 6.1069016456604,
+ "learning_rate": 1.8623472993301774e-05,
+ "loss": 0.537,
+ "step": 7117
+ },
+ {
+ "epoch": 19.39509536784741,
+ "grad_norm": 6.635929107666016,
+ "learning_rate": 1.8623026137485047e-05,
+ "loss": 0.7072,
+ "step": 7118
+ },
+ {
+ "epoch": 19.39782016348774,
+ "grad_norm": 6.074080944061279,
+ "learning_rate": 1.8622579214512476e-05,
+ "loss": 0.6587,
+ "step": 7119
+ },
+ {
+ "epoch": 19.400544959128066,
+ "grad_norm": 9.254389762878418,
+ "learning_rate": 1.862213222438753e-05,
+ "loss": 0.7334,
+ "step": 7120
+ },
+ {
+ "epoch": 19.40326975476839,
+ "grad_norm": 7.0395355224609375,
+ "learning_rate": 1.86216851671137e-05,
+ "loss": 0.6082,
+ "step": 7121
+ },
+ {
+ "epoch": 19.40599455040872,
+ "grad_norm": 5.30159854888916,
+ "learning_rate": 1.862123804269446e-05,
+ "loss": 0.6794,
+ "step": 7122
+ },
+ {
+ "epoch": 19.408719346049047,
+ "grad_norm": 6.109806537628174,
+ "learning_rate": 1.8620790851133294e-05,
+ "loss": 0.8315,
+ "step": 7123
+ },
+ {
+ "epoch": 19.411444141689373,
+ "grad_norm": 12.378950119018555,
+ "learning_rate": 1.862034359243369e-05,
+ "loss": 0.6949,
+ "step": 7124
+ },
+ {
+ "epoch": 19.414168937329702,
+ "grad_norm": 7.202772617340088,
+ "learning_rate": 1.8619896266599128e-05,
+ "loss": 0.6462,
+ "step": 7125
+ },
+ {
+ "epoch": 19.416893732970028,
+ "grad_norm": 7.736551284790039,
+ "learning_rate": 1.861944887363309e-05,
+ "loss": 0.5554,
+ "step": 7126
+ },
+ {
+ "epoch": 19.419618528610354,
+ "grad_norm": 7.718057632446289,
+ "learning_rate": 1.861900141353906e-05,
+ "loss": 0.5559,
+ "step": 7127
+ },
+ {
+ "epoch": 19.422343324250683,
+ "grad_norm": 7.58015251159668,
+ "learning_rate": 1.8618553886320527e-05,
+ "loss": 0.6467,
+ "step": 7128
+ },
+ {
+ "epoch": 19.42506811989101,
+ "grad_norm": 6.910373687744141,
+ "learning_rate": 1.861810629198097e-05,
+ "loss": 0.7523,
+ "step": 7129
+ },
+ {
+ "epoch": 19.427792915531334,
+ "grad_norm": 5.550629138946533,
+ "learning_rate": 1.8617658630523884e-05,
+ "loss": 0.8076,
+ "step": 7130
+ },
+ {
+ "epoch": 19.430517711171664,
+ "grad_norm": 7.823291778564453,
+ "learning_rate": 1.8617210901952747e-05,
+ "loss": 0.6736,
+ "step": 7131
+ },
+ {
+ "epoch": 19.43324250681199,
+ "grad_norm": 6.508481979370117,
+ "learning_rate": 1.861676310627105e-05,
+ "loss": 0.519,
+ "step": 7132
+ },
+ {
+ "epoch": 19.435967302452315,
+ "grad_norm": 6.689467906951904,
+ "learning_rate": 1.8616315243482278e-05,
+ "loss": 0.5737,
+ "step": 7133
+ },
+ {
+ "epoch": 19.438692098092645,
+ "grad_norm": 5.774669170379639,
+ "learning_rate": 1.861586731358992e-05,
+ "loss": 0.6881,
+ "step": 7134
+ },
+ {
+ "epoch": 19.44141689373297,
+ "grad_norm": 6.4125075340271,
+ "learning_rate": 1.8615419316597466e-05,
+ "loss": 0.7319,
+ "step": 7135
+ },
+ {
+ "epoch": 19.444141689373296,
+ "grad_norm": 6.2294921875,
+ "learning_rate": 1.8614971252508404e-05,
+ "loss": 0.567,
+ "step": 7136
+ },
+ {
+ "epoch": 19.446866485013626,
+ "grad_norm": 6.171770095825195,
+ "learning_rate": 1.861452312132622e-05,
+ "loss": 0.5265,
+ "step": 7137
+ },
+ {
+ "epoch": 19.44959128065395,
+ "grad_norm": 8.403230667114258,
+ "learning_rate": 1.861407492305441e-05,
+ "loss": 0.6724,
+ "step": 7138
+ },
+ {
+ "epoch": 19.452316076294277,
+ "grad_norm": 6.022368907928467,
+ "learning_rate": 1.861362665769646e-05,
+ "loss": 0.5043,
+ "step": 7139
+ },
+ {
+ "epoch": 19.455040871934607,
+ "grad_norm": 6.366517543792725,
+ "learning_rate": 1.8613178325255864e-05,
+ "loss": 0.6111,
+ "step": 7140
+ },
+ {
+ "epoch": 19.457765667574932,
+ "grad_norm": 5.627801418304443,
+ "learning_rate": 1.861272992573611e-05,
+ "loss": 0.6529,
+ "step": 7141
+ },
+ {
+ "epoch": 19.460490463215258,
+ "grad_norm": 6.524795055389404,
+ "learning_rate": 1.8612281459140697e-05,
+ "loss": 0.7396,
+ "step": 7142
+ },
+ {
+ "epoch": 19.463215258855588,
+ "grad_norm": 7.048426151275635,
+ "learning_rate": 1.861183292547311e-05,
+ "loss": 0.5731,
+ "step": 7143
+ },
+ {
+ "epoch": 19.465940054495913,
+ "grad_norm": 7.072524547576904,
+ "learning_rate": 1.8611384324736846e-05,
+ "loss": 0.7032,
+ "step": 7144
+ },
+ {
+ "epoch": 19.46866485013624,
+ "grad_norm": 13.199339866638184,
+ "learning_rate": 1.8610935656935397e-05,
+ "loss": 0.5778,
+ "step": 7145
+ },
+ {
+ "epoch": 19.47138964577657,
+ "grad_norm": 7.156103134155273,
+ "learning_rate": 1.8610486922072263e-05,
+ "loss": 0.7258,
+ "step": 7146
+ },
+ {
+ "epoch": 19.474114441416894,
+ "grad_norm": 14.056463241577148,
+ "learning_rate": 1.861003812015093e-05,
+ "loss": 0.6205,
+ "step": 7147
+ },
+ {
+ "epoch": 19.47683923705722,
+ "grad_norm": 6.980556964874268,
+ "learning_rate": 1.8609589251174897e-05,
+ "loss": 0.5867,
+ "step": 7148
+ },
+ {
+ "epoch": 19.479564032697546,
+ "grad_norm": 7.333771705627441,
+ "learning_rate": 1.860914031514766e-05,
+ "loss": 0.7377,
+ "step": 7149
+ },
+ {
+ "epoch": 19.482288828337875,
+ "grad_norm": 6.965285778045654,
+ "learning_rate": 1.8608691312072714e-05,
+ "loss": 0.5198,
+ "step": 7150
+ },
+ {
+ "epoch": 19.4850136239782,
+ "grad_norm": 6.68442964553833,
+ "learning_rate": 1.8608242241953563e-05,
+ "loss": 0.6316,
+ "step": 7151
+ },
+ {
+ "epoch": 19.48773841961853,
+ "grad_norm": 6.524171352386475,
+ "learning_rate": 1.8607793104793696e-05,
+ "loss": 0.7114,
+ "step": 7152
+ },
+ {
+ "epoch": 19.490463215258856,
+ "grad_norm": 7.786107540130615,
+ "learning_rate": 1.8607343900596612e-05,
+ "loss": 0.5728,
+ "step": 7153
+ },
+ {
+ "epoch": 19.493188010899182,
+ "grad_norm": 8.028909683227539,
+ "learning_rate": 1.860689462936581e-05,
+ "loss": 0.7377,
+ "step": 7154
+ },
+ {
+ "epoch": 19.495912806539508,
+ "grad_norm": 6.5673747062683105,
+ "learning_rate": 1.860644529110479e-05,
+ "loss": 0.7299,
+ "step": 7155
+ },
+ {
+ "epoch": 19.498637602179837,
+ "grad_norm": 9.590473175048828,
+ "learning_rate": 1.860599588581705e-05,
+ "loss": 0.7009,
+ "step": 7156
+ },
+ {
+ "epoch": 19.501362397820163,
+ "grad_norm": 5.793774604797363,
+ "learning_rate": 1.8605546413506092e-05,
+ "loss": 0.5983,
+ "step": 7157
+ },
+ {
+ "epoch": 19.504087193460492,
+ "grad_norm": 7.473902702331543,
+ "learning_rate": 1.8605096874175418e-05,
+ "loss": 0.5812,
+ "step": 7158
+ },
+ {
+ "epoch": 19.506811989100818,
+ "grad_norm": 8.201847076416016,
+ "learning_rate": 1.860464726782852e-05,
+ "loss": 0.6611,
+ "step": 7159
+ },
+ {
+ "epoch": 19.509536784741144,
+ "grad_norm": 6.505906105041504,
+ "learning_rate": 1.860419759446891e-05,
+ "loss": 0.6234,
+ "step": 7160
+ },
+ {
+ "epoch": 19.51226158038147,
+ "grad_norm": 7.290673732757568,
+ "learning_rate": 1.860374785410009e-05,
+ "loss": 0.6139,
+ "step": 7161
+ },
+ {
+ "epoch": 19.5149863760218,
+ "grad_norm": 6.844306945800781,
+ "learning_rate": 1.8603298046725556e-05,
+ "loss": 0.5355,
+ "step": 7162
+ },
+ {
+ "epoch": 19.517711171662125,
+ "grad_norm": 6.990944862365723,
+ "learning_rate": 1.860284817234881e-05,
+ "loss": 0.8591,
+ "step": 7163
+ },
+ {
+ "epoch": 19.520435967302454,
+ "grad_norm": 5.967893600463867,
+ "learning_rate": 1.8602398230973363e-05,
+ "loss": 0.6672,
+ "step": 7164
+ },
+ {
+ "epoch": 19.52316076294278,
+ "grad_norm": 5.820087909698486,
+ "learning_rate": 1.8601948222602715e-05,
+ "loss": 0.5549,
+ "step": 7165
+ },
+ {
+ "epoch": 19.525885558583106,
+ "grad_norm": 27.065914154052734,
+ "learning_rate": 1.860149814724037e-05,
+ "loss": 0.6985,
+ "step": 7166
+ },
+ {
+ "epoch": 19.52861035422343,
+ "grad_norm": 6.7624945640563965,
+ "learning_rate": 1.8601048004889835e-05,
+ "loss": 0.7301,
+ "step": 7167
+ },
+ {
+ "epoch": 19.53133514986376,
+ "grad_norm": 5.747246265411377,
+ "learning_rate": 1.8600597795554613e-05,
+ "loss": 0.7587,
+ "step": 7168
+ },
+ {
+ "epoch": 19.534059945504087,
+ "grad_norm": 5.961423873901367,
+ "learning_rate": 1.8600147519238217e-05,
+ "loss": 0.4536,
+ "step": 7169
+ },
+ {
+ "epoch": 19.536784741144416,
+ "grad_norm": 7.202899932861328,
+ "learning_rate": 1.8599697175944145e-05,
+ "loss": 0.5563,
+ "step": 7170
+ },
+ {
+ "epoch": 19.539509536784742,
+ "grad_norm": 6.756170272827148,
+ "learning_rate": 1.859924676567591e-05,
+ "loss": 0.6687,
+ "step": 7171
+ },
+ {
+ "epoch": 19.542234332425068,
+ "grad_norm": 5.981444358825684,
+ "learning_rate": 1.8598796288437017e-05,
+ "loss": 0.8013,
+ "step": 7172
+ },
+ {
+ "epoch": 19.544959128065393,
+ "grad_norm": 7.157764911651611,
+ "learning_rate": 1.8598345744230975e-05,
+ "loss": 0.5178,
+ "step": 7173
+ },
+ {
+ "epoch": 19.547683923705723,
+ "grad_norm": 6.694261074066162,
+ "learning_rate": 1.8597895133061293e-05,
+ "loss": 0.5019,
+ "step": 7174
+ },
+ {
+ "epoch": 19.55040871934605,
+ "grad_norm": 7.24384069442749,
+ "learning_rate": 1.859744445493148e-05,
+ "loss": 0.7969,
+ "step": 7175
+ },
+ {
+ "epoch": 19.553133514986374,
+ "grad_norm": 6.642652988433838,
+ "learning_rate": 1.8596993709845047e-05,
+ "loss": 0.6328,
+ "step": 7176
+ },
+ {
+ "epoch": 19.555858310626704,
+ "grad_norm": 7.291753768920898,
+ "learning_rate": 1.8596542897805504e-05,
+ "loss": 0.6659,
+ "step": 7177
+ },
+ {
+ "epoch": 19.55858310626703,
+ "grad_norm": 6.5086798667907715,
+ "learning_rate": 1.8596092018816364e-05,
+ "loss": 0.5974,
+ "step": 7178
+ },
+ {
+ "epoch": 19.561307901907355,
+ "grad_norm": 7.8980393409729,
+ "learning_rate": 1.859564107288113e-05,
+ "loss": 0.5888,
+ "step": 7179
+ },
+ {
+ "epoch": 19.564032697547685,
+ "grad_norm": 5.9287238121032715,
+ "learning_rate": 1.8595190060003327e-05,
+ "loss": 0.6462,
+ "step": 7180
+ },
+ {
+ "epoch": 19.56675749318801,
+ "grad_norm": 6.828423976898193,
+ "learning_rate": 1.8594738980186456e-05,
+ "loss": 0.583,
+ "step": 7181
+ },
+ {
+ "epoch": 19.569482288828336,
+ "grad_norm": 6.432905197143555,
+ "learning_rate": 1.8594287833434033e-05,
+ "loss": 0.5305,
+ "step": 7182
+ },
+ {
+ "epoch": 19.572207084468666,
+ "grad_norm": 161.7238311767578,
+ "learning_rate": 1.8593836619749576e-05,
+ "loss": 0.7422,
+ "step": 7183
+ },
+ {
+ "epoch": 19.57493188010899,
+ "grad_norm": 7.707223892211914,
+ "learning_rate": 1.8593385339136596e-05,
+ "loss": 0.7953,
+ "step": 7184
+ },
+ {
+ "epoch": 19.577656675749317,
+ "grad_norm": 5.989534854888916,
+ "learning_rate": 1.8592933991598605e-05,
+ "loss": 0.6459,
+ "step": 7185
+ },
+ {
+ "epoch": 19.580381471389646,
+ "grad_norm": 7.815117835998535,
+ "learning_rate": 1.8592482577139125e-05,
+ "loss": 0.6329,
+ "step": 7186
+ },
+ {
+ "epoch": 19.583106267029972,
+ "grad_norm": 8.599873542785645,
+ "learning_rate": 1.8592031095761665e-05,
+ "loss": 0.5523,
+ "step": 7187
+ },
+ {
+ "epoch": 19.585831062670298,
+ "grad_norm": 6.529746055603027,
+ "learning_rate": 1.8591579547469743e-05,
+ "loss": 0.5743,
+ "step": 7188
+ },
+ {
+ "epoch": 19.588555858310627,
+ "grad_norm": 7.773322105407715,
+ "learning_rate": 1.8591127932266874e-05,
+ "loss": 0.6161,
+ "step": 7189
+ },
+ {
+ "epoch": 19.591280653950953,
+ "grad_norm": 6.594780445098877,
+ "learning_rate": 1.859067625015658e-05,
+ "loss": 0.7932,
+ "step": 7190
+ },
+ {
+ "epoch": 19.59400544959128,
+ "grad_norm": 8.915833473205566,
+ "learning_rate": 1.8590224501142377e-05,
+ "loss": 0.7393,
+ "step": 7191
+ },
+ {
+ "epoch": 19.59673024523161,
+ "grad_norm": 7.231869697570801,
+ "learning_rate": 1.8589772685227777e-05,
+ "loss": 0.8992,
+ "step": 7192
+ },
+ {
+ "epoch": 19.599455040871934,
+ "grad_norm": 9.902299880981445,
+ "learning_rate": 1.858932080241631e-05,
+ "loss": 0.5167,
+ "step": 7193
+ },
+ {
+ "epoch": 19.60217983651226,
+ "grad_norm": 8.73497486114502,
+ "learning_rate": 1.8588868852711487e-05,
+ "loss": 0.6732,
+ "step": 7194
+ },
+ {
+ "epoch": 19.60490463215259,
+ "grad_norm": 8.076318740844727,
+ "learning_rate": 1.8588416836116824e-05,
+ "loss": 0.8286,
+ "step": 7195
+ },
+ {
+ "epoch": 19.607629427792915,
+ "grad_norm": 7.265976428985596,
+ "learning_rate": 1.8587964752635856e-05,
+ "loss": 0.5906,
+ "step": 7196
+ },
+ {
+ "epoch": 19.61035422343324,
+ "grad_norm": 7.615754127502441,
+ "learning_rate": 1.8587512602272088e-05,
+ "loss": 0.7413,
+ "step": 7197
+ },
+ {
+ "epoch": 19.61307901907357,
+ "grad_norm": 7.679302215576172,
+ "learning_rate": 1.8587060385029052e-05,
+ "loss": 0.7584,
+ "step": 7198
+ },
+ {
+ "epoch": 19.615803814713896,
+ "grad_norm": 7.445467948913574,
+ "learning_rate": 1.8586608100910265e-05,
+ "loss": 0.702,
+ "step": 7199
+ },
+ {
+ "epoch": 19.618528610354222,
+ "grad_norm": 6.944090843200684,
+ "learning_rate": 1.858615574991925e-05,
+ "loss": 0.7039,
+ "step": 7200
+ },
+ {
+ "epoch": 19.62125340599455,
+ "grad_norm": 25.596141815185547,
+ "learning_rate": 1.858570333205953e-05,
+ "loss": 0.8243,
+ "step": 7201
+ },
+ {
+ "epoch": 19.623978201634877,
+ "grad_norm": 8.27812671661377,
+ "learning_rate": 1.858525084733463e-05,
+ "loss": 0.5999,
+ "step": 7202
+ },
+ {
+ "epoch": 19.626702997275203,
+ "grad_norm": 6.74351692199707,
+ "learning_rate": 1.8584798295748072e-05,
+ "loss": 0.7444,
+ "step": 7203
+ },
+ {
+ "epoch": 19.629427792915532,
+ "grad_norm": 6.864198207855225,
+ "learning_rate": 1.8584345677303385e-05,
+ "loss": 0.6627,
+ "step": 7204
+ },
+ {
+ "epoch": 19.632152588555858,
+ "grad_norm": 7.106120586395264,
+ "learning_rate": 1.8583892992004084e-05,
+ "loss": 0.6129,
+ "step": 7205
+ },
+ {
+ "epoch": 19.634877384196184,
+ "grad_norm": 7.282804489135742,
+ "learning_rate": 1.8583440239853705e-05,
+ "loss": 0.6901,
+ "step": 7206
+ },
+ {
+ "epoch": 19.637602179836513,
+ "grad_norm": 6.740818500518799,
+ "learning_rate": 1.8582987420855766e-05,
+ "loss": 0.7731,
+ "step": 7207
+ },
+ {
+ "epoch": 19.64032697547684,
+ "grad_norm": 16.960086822509766,
+ "learning_rate": 1.8582534535013802e-05,
+ "loss": 0.7325,
+ "step": 7208
+ },
+ {
+ "epoch": 19.643051771117165,
+ "grad_norm": 8.8674898147583,
+ "learning_rate": 1.858208158233133e-05,
+ "loss": 0.6349,
+ "step": 7209
+ },
+ {
+ "epoch": 19.645776566757494,
+ "grad_norm": 6.304328441619873,
+ "learning_rate": 1.8581628562811887e-05,
+ "loss": 0.7913,
+ "step": 7210
+ },
+ {
+ "epoch": 19.64850136239782,
+ "grad_norm": 8.251081466674805,
+ "learning_rate": 1.8581175476458993e-05,
+ "loss": 0.8625,
+ "step": 7211
+ },
+ {
+ "epoch": 19.651226158038146,
+ "grad_norm": 7.601869583129883,
+ "learning_rate": 1.8580722323276186e-05,
+ "loss": 0.7343,
+ "step": 7212
+ },
+ {
+ "epoch": 19.653950953678475,
+ "grad_norm": 8.6659574508667,
+ "learning_rate": 1.8580269103266985e-05,
+ "loss": 0.5933,
+ "step": 7213
+ },
+ {
+ "epoch": 19.6566757493188,
+ "grad_norm": 7.345037460327148,
+ "learning_rate": 1.8579815816434928e-05,
+ "loss": 0.7733,
+ "step": 7214
+ },
+ {
+ "epoch": 19.659400544959126,
+ "grad_norm": 7.696609973907471,
+ "learning_rate": 1.857936246278354e-05,
+ "loss": 0.5823,
+ "step": 7215
+ },
+ {
+ "epoch": 19.662125340599456,
+ "grad_norm": 7.0348639488220215,
+ "learning_rate": 1.857890904231635e-05,
+ "loss": 0.7399,
+ "step": 7216
+ },
+ {
+ "epoch": 19.66485013623978,
+ "grad_norm": 8.785232543945312,
+ "learning_rate": 1.8578455555036898e-05,
+ "loss": 0.8542,
+ "step": 7217
+ },
+ {
+ "epoch": 19.667574931880107,
+ "grad_norm": 6.973820209503174,
+ "learning_rate": 1.8578002000948707e-05,
+ "loss": 0.796,
+ "step": 7218
+ },
+ {
+ "epoch": 19.670299727520437,
+ "grad_norm": 7.065167427062988,
+ "learning_rate": 1.8577548380055312e-05,
+ "loss": 0.6429,
+ "step": 7219
+ },
+ {
+ "epoch": 19.673024523160763,
+ "grad_norm": 6.5639238357543945,
+ "learning_rate": 1.8577094692360248e-05,
+ "loss": 0.6954,
+ "step": 7220
+ },
+ {
+ "epoch": 19.67574931880109,
+ "grad_norm": 8.30419921875,
+ "learning_rate": 1.8576640937867047e-05,
+ "loss": 0.6342,
+ "step": 7221
+ },
+ {
+ "epoch": 19.678474114441418,
+ "grad_norm": 6.616217136383057,
+ "learning_rate": 1.857618711657924e-05,
+ "loss": 0.5658,
+ "step": 7222
+ },
+ {
+ "epoch": 19.681198910081743,
+ "grad_norm": 10.555207252502441,
+ "learning_rate": 1.8575733228500364e-05,
+ "loss": 0.7848,
+ "step": 7223
+ },
+ {
+ "epoch": 19.68392370572207,
+ "grad_norm": 7.156619071960449,
+ "learning_rate": 1.8575279273633957e-05,
+ "loss": 0.6635,
+ "step": 7224
+ },
+ {
+ "epoch": 19.6866485013624,
+ "grad_norm": 6.334125995635986,
+ "learning_rate": 1.857482525198355e-05,
+ "loss": 0.7209,
+ "step": 7225
+ },
+ {
+ "epoch": 19.689373297002724,
+ "grad_norm": 7.727910041809082,
+ "learning_rate": 1.8574371163552677e-05,
+ "loss": 0.6125,
+ "step": 7226
+ },
+ {
+ "epoch": 19.69209809264305,
+ "grad_norm": 6.3382110595703125,
+ "learning_rate": 1.857391700834488e-05,
+ "loss": 0.658,
+ "step": 7227
+ },
+ {
+ "epoch": 19.69482288828338,
+ "grad_norm": 7.151101589202881,
+ "learning_rate": 1.857346278636369e-05,
+ "loss": 0.6647,
+ "step": 7228
+ },
+ {
+ "epoch": 19.697547683923705,
+ "grad_norm": 7.8673906326293945,
+ "learning_rate": 1.857300849761265e-05,
+ "loss": 0.7291,
+ "step": 7229
+ },
+ {
+ "epoch": 19.70027247956403,
+ "grad_norm": 6.770479679107666,
+ "learning_rate": 1.8572554142095296e-05,
+ "loss": 0.6501,
+ "step": 7230
+ },
+ {
+ "epoch": 19.70299727520436,
+ "grad_norm": 6.884183406829834,
+ "learning_rate": 1.8572099719815167e-05,
+ "loss": 0.6016,
+ "step": 7231
+ },
+ {
+ "epoch": 19.705722070844686,
+ "grad_norm": 6.333080768585205,
+ "learning_rate": 1.85716452307758e-05,
+ "loss": 0.7103,
+ "step": 7232
+ },
+ {
+ "epoch": 19.708446866485012,
+ "grad_norm": 6.11884069442749,
+ "learning_rate": 1.8571190674980734e-05,
+ "loss": 0.7423,
+ "step": 7233
+ },
+ {
+ "epoch": 19.71117166212534,
+ "grad_norm": 6.871713638305664,
+ "learning_rate": 1.8570736052433515e-05,
+ "loss": 0.5898,
+ "step": 7234
+ },
+ {
+ "epoch": 19.713896457765667,
+ "grad_norm": 7.70427942276001,
+ "learning_rate": 1.8570281363137677e-05,
+ "loss": 0.6857,
+ "step": 7235
+ },
+ {
+ "epoch": 19.716621253405993,
+ "grad_norm": 5.892122268676758,
+ "learning_rate": 1.8569826607096764e-05,
+ "loss": 0.5967,
+ "step": 7236
+ },
+ {
+ "epoch": 19.719346049046322,
+ "grad_norm": 6.9117431640625,
+ "learning_rate": 1.8569371784314318e-05,
+ "loss": 0.6092,
+ "step": 7237
+ },
+ {
+ "epoch": 19.722070844686648,
+ "grad_norm": 8.622511863708496,
+ "learning_rate": 1.856891689479388e-05,
+ "loss": 0.7522,
+ "step": 7238
+ },
+ {
+ "epoch": 19.724795640326974,
+ "grad_norm": 6.664440631866455,
+ "learning_rate": 1.8568461938538994e-05,
+ "loss": 0.6016,
+ "step": 7239
+ },
+ {
+ "epoch": 19.727520435967303,
+ "grad_norm": 6.671781063079834,
+ "learning_rate": 1.8568006915553202e-05,
+ "loss": 0.5802,
+ "step": 7240
+ },
+ {
+ "epoch": 19.73024523160763,
+ "grad_norm": 6.033954620361328,
+ "learning_rate": 1.856755182584004e-05,
+ "loss": 0.5421,
+ "step": 7241
+ },
+ {
+ "epoch": 19.732970027247955,
+ "grad_norm": 6.388408660888672,
+ "learning_rate": 1.856709666940307e-05,
+ "loss": 0.7227,
+ "step": 7242
+ },
+ {
+ "epoch": 19.735694822888284,
+ "grad_norm": 6.138409614562988,
+ "learning_rate": 1.8566641446245825e-05,
+ "loss": 0.749,
+ "step": 7243
+ },
+ {
+ "epoch": 19.73841961852861,
+ "grad_norm": 11.597804069519043,
+ "learning_rate": 1.856618615637185e-05,
+ "loss": 0.6938,
+ "step": 7244
+ },
+ {
+ "epoch": 19.741144414168936,
+ "grad_norm": 6.512270450592041,
+ "learning_rate": 1.8565730799784693e-05,
+ "loss": 0.5875,
+ "step": 7245
+ },
+ {
+ "epoch": 19.743869209809265,
+ "grad_norm": 11.96609115600586,
+ "learning_rate": 1.85652753764879e-05,
+ "loss": 0.5868,
+ "step": 7246
+ },
+ {
+ "epoch": 19.74659400544959,
+ "grad_norm": 5.910905361175537,
+ "learning_rate": 1.856481988648502e-05,
+ "loss": 0.6007,
+ "step": 7247
+ },
+ {
+ "epoch": 19.749318801089917,
+ "grad_norm": 10.077803611755371,
+ "learning_rate": 1.85643643297796e-05,
+ "loss": 0.7849,
+ "step": 7248
+ },
+ {
+ "epoch": 19.752043596730246,
+ "grad_norm": 6.34931755065918,
+ "learning_rate": 1.8563908706375183e-05,
+ "loss": 0.719,
+ "step": 7249
+ },
+ {
+ "epoch": 19.754768392370572,
+ "grad_norm": 6.884245872497559,
+ "learning_rate": 1.8563453016275322e-05,
+ "loss": 0.6873,
+ "step": 7250
+ },
+ {
+ "epoch": 19.757493188010898,
+ "grad_norm": 5.955710411071777,
+ "learning_rate": 1.8562997259483563e-05,
+ "loss": 0.7332,
+ "step": 7251
+ },
+ {
+ "epoch": 19.760217983651227,
+ "grad_norm": 7.162776947021484,
+ "learning_rate": 1.856254143600346e-05,
+ "loss": 0.8262,
+ "step": 7252
+ },
+ {
+ "epoch": 19.762942779291553,
+ "grad_norm": 6.761709213256836,
+ "learning_rate": 1.8562085545838557e-05,
+ "loss": 0.7379,
+ "step": 7253
+ },
+ {
+ "epoch": 19.76566757493188,
+ "grad_norm": 6.938498497009277,
+ "learning_rate": 1.8561629588992405e-05,
+ "loss": 0.5635,
+ "step": 7254
+ },
+ {
+ "epoch": 19.768392370572208,
+ "grad_norm": 5.810408115386963,
+ "learning_rate": 1.856117356546856e-05,
+ "loss": 0.616,
+ "step": 7255
+ },
+ {
+ "epoch": 19.771117166212534,
+ "grad_norm": 10.42446517944336,
+ "learning_rate": 1.8560717475270573e-05,
+ "loss": 0.6256,
+ "step": 7256
+ },
+ {
+ "epoch": 19.77384196185286,
+ "grad_norm": 6.122094631195068,
+ "learning_rate": 1.856026131840199e-05,
+ "loss": 0.5939,
+ "step": 7257
+ },
+ {
+ "epoch": 19.77656675749319,
+ "grad_norm": 6.047171592712402,
+ "learning_rate": 1.855980509486637e-05,
+ "loss": 0.6394,
+ "step": 7258
+ },
+ {
+ "epoch": 19.779291553133515,
+ "grad_norm": 6.6370415687561035,
+ "learning_rate": 1.8559348804667265e-05,
+ "loss": 0.5835,
+ "step": 7259
+ },
+ {
+ "epoch": 19.78201634877384,
+ "grad_norm": 5.400707721710205,
+ "learning_rate": 1.8558892447808224e-05,
+ "loss": 0.7173,
+ "step": 7260
+ },
+ {
+ "epoch": 19.78474114441417,
+ "grad_norm": 7.478552341461182,
+ "learning_rate": 1.8558436024292803e-05,
+ "loss": 0.7388,
+ "step": 7261
+ },
+ {
+ "epoch": 19.787465940054496,
+ "grad_norm": 6.656715393066406,
+ "learning_rate": 1.855797953412456e-05,
+ "loss": 0.6008,
+ "step": 7262
+ },
+ {
+ "epoch": 19.79019073569482,
+ "grad_norm": 7.756550312042236,
+ "learning_rate": 1.8557522977307045e-05,
+ "loss": 0.6945,
+ "step": 7263
+ },
+ {
+ "epoch": 19.79291553133515,
+ "grad_norm": 7.284280776977539,
+ "learning_rate": 1.8557066353843822e-05,
+ "loss": 0.5712,
+ "step": 7264
+ },
+ {
+ "epoch": 19.795640326975477,
+ "grad_norm": 6.931217193603516,
+ "learning_rate": 1.8556609663738435e-05,
+ "loss": 0.7531,
+ "step": 7265
+ },
+ {
+ "epoch": 19.798365122615802,
+ "grad_norm": 6.6884942054748535,
+ "learning_rate": 1.8556152906994453e-05,
+ "loss": 0.7483,
+ "step": 7266
+ },
+ {
+ "epoch": 19.80108991825613,
+ "grad_norm": 6.651729106903076,
+ "learning_rate": 1.8555696083615425e-05,
+ "loss": 0.6664,
+ "step": 7267
+ },
+ {
+ "epoch": 19.803814713896458,
+ "grad_norm": 6.5602030754089355,
+ "learning_rate": 1.8555239193604913e-05,
+ "loss": 0.6028,
+ "step": 7268
+ },
+ {
+ "epoch": 19.806539509536783,
+ "grad_norm": 7.153371334075928,
+ "learning_rate": 1.855478223696647e-05,
+ "loss": 0.6362,
+ "step": 7269
+ },
+ {
+ "epoch": 19.809264305177113,
+ "grad_norm": 6.693792343139648,
+ "learning_rate": 1.8554325213703663e-05,
+ "loss": 0.6558,
+ "step": 7270
+ },
+ {
+ "epoch": 19.81198910081744,
+ "grad_norm": 6.659425258636475,
+ "learning_rate": 1.8553868123820043e-05,
+ "loss": 0.6407,
+ "step": 7271
+ },
+ {
+ "epoch": 19.814713896457764,
+ "grad_norm": 5.8461527824401855,
+ "learning_rate": 1.8553410967319173e-05,
+ "loss": 0.8785,
+ "step": 7272
+ },
+ {
+ "epoch": 19.817438692098094,
+ "grad_norm": 5.463052272796631,
+ "learning_rate": 1.8552953744204613e-05,
+ "loss": 0.6613,
+ "step": 7273
+ },
+ {
+ "epoch": 19.82016348773842,
+ "grad_norm": 6.810503959655762,
+ "learning_rate": 1.855249645447993e-05,
+ "loss": 0.6901,
+ "step": 7274
+ },
+ {
+ "epoch": 19.822888283378745,
+ "grad_norm": 6.054370403289795,
+ "learning_rate": 1.8552039098148677e-05,
+ "loss": 0.6302,
+ "step": 7275
+ },
+ {
+ "epoch": 19.825613079019075,
+ "grad_norm": 7.32468318939209,
+ "learning_rate": 1.8551581675214416e-05,
+ "loss": 0.6813,
+ "step": 7276
+ },
+ {
+ "epoch": 19.8283378746594,
+ "grad_norm": 7.992151737213135,
+ "learning_rate": 1.8551124185680716e-05,
+ "loss": 0.5047,
+ "step": 7277
+ },
+ {
+ "epoch": 19.831062670299726,
+ "grad_norm": 7.378367900848389,
+ "learning_rate": 1.855066662955113e-05,
+ "loss": 0.7171,
+ "step": 7278
+ },
+ {
+ "epoch": 19.833787465940055,
+ "grad_norm": 6.132101058959961,
+ "learning_rate": 1.8550209006829236e-05,
+ "loss": 0.9371,
+ "step": 7279
+ },
+ {
+ "epoch": 19.83651226158038,
+ "grad_norm": 6.623978614807129,
+ "learning_rate": 1.8549751317518583e-05,
+ "loss": 0.7372,
+ "step": 7280
+ },
+ {
+ "epoch": 19.839237057220707,
+ "grad_norm": 5.973078727722168,
+ "learning_rate": 1.8549293561622742e-05,
+ "loss": 0.6409,
+ "step": 7281
+ },
+ {
+ "epoch": 19.841961852861036,
+ "grad_norm": 6.439085960388184,
+ "learning_rate": 1.854883573914528e-05,
+ "loss": 0.5786,
+ "step": 7282
+ },
+ {
+ "epoch": 19.844686648501362,
+ "grad_norm": 6.63276481628418,
+ "learning_rate": 1.8548377850089764e-05,
+ "loss": 0.5807,
+ "step": 7283
+ },
+ {
+ "epoch": 19.847411444141688,
+ "grad_norm": 44.11280822753906,
+ "learning_rate": 1.8547919894459753e-05,
+ "loss": 0.6481,
+ "step": 7284
+ },
+ {
+ "epoch": 19.850136239782017,
+ "grad_norm": 7.010636806488037,
+ "learning_rate": 1.8547461872258816e-05,
+ "loss": 0.6009,
+ "step": 7285
+ },
+ {
+ "epoch": 19.852861035422343,
+ "grad_norm": 7.010998725891113,
+ "learning_rate": 1.854700378349052e-05,
+ "loss": 0.8474,
+ "step": 7286
+ },
+ {
+ "epoch": 19.85558583106267,
+ "grad_norm": 7.728806972503662,
+ "learning_rate": 1.854654562815844e-05,
+ "loss": 0.7013,
+ "step": 7287
+ },
+ {
+ "epoch": 19.858310626703,
+ "grad_norm": 7.5362229347229,
+ "learning_rate": 1.854608740626613e-05,
+ "loss": 0.8662,
+ "step": 7288
+ },
+ {
+ "epoch": 19.861035422343324,
+ "grad_norm": 5.720480918884277,
+ "learning_rate": 1.854562911781717e-05,
+ "loss": 0.61,
+ "step": 7289
+ },
+ {
+ "epoch": 19.86376021798365,
+ "grad_norm": 7.116855144500732,
+ "learning_rate": 1.8545170762815122e-05,
+ "loss": 0.7177,
+ "step": 7290
+ },
+ {
+ "epoch": 19.86648501362398,
+ "grad_norm": 7.555663585662842,
+ "learning_rate": 1.8544712341263566e-05,
+ "loss": 0.728,
+ "step": 7291
+ },
+ {
+ "epoch": 19.869209809264305,
+ "grad_norm": 10.37500286102295,
+ "learning_rate": 1.854425385316606e-05,
+ "loss": 0.8479,
+ "step": 7292
+ },
+ {
+ "epoch": 19.87193460490463,
+ "grad_norm": 5.959344863891602,
+ "learning_rate": 1.854379529852618e-05,
+ "loss": 0.7393,
+ "step": 7293
+ },
+ {
+ "epoch": 19.87465940054496,
+ "grad_norm": 7.234870433807373,
+ "learning_rate": 1.85433366773475e-05,
+ "loss": 0.6945,
+ "step": 7294
+ },
+ {
+ "epoch": 19.877384196185286,
+ "grad_norm": 7.19219970703125,
+ "learning_rate": 1.854287798963359e-05,
+ "loss": 0.6594,
+ "step": 7295
+ },
+ {
+ "epoch": 19.88010899182561,
+ "grad_norm": 6.701728820800781,
+ "learning_rate": 1.8542419235388016e-05,
+ "loss": 0.6553,
+ "step": 7296
+ },
+ {
+ "epoch": 19.88283378746594,
+ "grad_norm": 8.00748348236084,
+ "learning_rate": 1.854196041461436e-05,
+ "loss": 0.8209,
+ "step": 7297
+ },
+ {
+ "epoch": 19.885558583106267,
+ "grad_norm": 6.522715091705322,
+ "learning_rate": 1.854150152731619e-05,
+ "loss": 0.5846,
+ "step": 7298
+ },
+ {
+ "epoch": 19.888283378746593,
+ "grad_norm": 5.627171516418457,
+ "learning_rate": 1.854104257349708e-05,
+ "loss": 0.6824,
+ "step": 7299
+ },
+ {
+ "epoch": 19.891008174386922,
+ "grad_norm": 7.171675205230713,
+ "learning_rate": 1.8540583553160606e-05,
+ "loss": 0.7336,
+ "step": 7300
+ },
+ {
+ "epoch": 19.893732970027248,
+ "grad_norm": 6.7835798263549805,
+ "learning_rate": 1.8540124466310345e-05,
+ "loss": 0.6887,
+ "step": 7301
+ },
+ {
+ "epoch": 19.896457765667574,
+ "grad_norm": 6.552082061767578,
+ "learning_rate": 1.8539665312949864e-05,
+ "loss": 0.8053,
+ "step": 7302
+ },
+ {
+ "epoch": 19.899182561307903,
+ "grad_norm": 286.9576416015625,
+ "learning_rate": 1.8539206093082748e-05,
+ "loss": 0.6068,
+ "step": 7303
+ },
+ {
+ "epoch": 19.90190735694823,
+ "grad_norm": 9.889516830444336,
+ "learning_rate": 1.8538746806712568e-05,
+ "loss": 0.5937,
+ "step": 7304
+ },
+ {
+ "epoch": 19.904632152588555,
+ "grad_norm": 7.597204685211182,
+ "learning_rate": 1.8538287453842906e-05,
+ "loss": 0.7594,
+ "step": 7305
+ },
+ {
+ "epoch": 19.907356948228884,
+ "grad_norm": 8.210258483886719,
+ "learning_rate": 1.853782803447733e-05,
+ "loss": 0.681,
+ "step": 7306
+ },
+ {
+ "epoch": 19.91008174386921,
+ "grad_norm": 11.477255821228027,
+ "learning_rate": 1.853736854861943e-05,
+ "loss": 0.8431,
+ "step": 7307
+ },
+ {
+ "epoch": 19.912806539509535,
+ "grad_norm": 9.768145561218262,
+ "learning_rate": 1.8536908996272777e-05,
+ "loss": 0.8397,
+ "step": 7308
+ },
+ {
+ "epoch": 19.915531335149865,
+ "grad_norm": 14.854004859924316,
+ "learning_rate": 1.853644937744095e-05,
+ "loss": 0.7717,
+ "step": 7309
+ },
+ {
+ "epoch": 19.91825613079019,
+ "grad_norm": 11.967172622680664,
+ "learning_rate": 1.853598969212753e-05,
+ "loss": 0.8673,
+ "step": 7310
+ },
+ {
+ "epoch": 19.920980926430516,
+ "grad_norm": 11.247687339782715,
+ "learning_rate": 1.8535529940336098e-05,
+ "loss": 0.7777,
+ "step": 7311
+ },
+ {
+ "epoch": 19.923705722070846,
+ "grad_norm": 10.99918270111084,
+ "learning_rate": 1.8535070122070237e-05,
+ "loss": 0.7543,
+ "step": 7312
+ },
+ {
+ "epoch": 19.92643051771117,
+ "grad_norm": 16.548458099365234,
+ "learning_rate": 1.853461023733352e-05,
+ "loss": 0.7911,
+ "step": 7313
+ },
+ {
+ "epoch": 19.929155313351497,
+ "grad_norm": 16.924104690551758,
+ "learning_rate": 1.853415028612953e-05,
+ "loss": 0.8726,
+ "step": 7314
+ },
+ {
+ "epoch": 19.931880108991827,
+ "grad_norm": 11.881562232971191,
+ "learning_rate": 1.8533690268461862e-05,
+ "loss": 0.8284,
+ "step": 7315
+ },
+ {
+ "epoch": 19.934604904632153,
+ "grad_norm": 12.334859848022461,
+ "learning_rate": 1.853323018433408e-05,
+ "loss": 0.9352,
+ "step": 7316
+ },
+ {
+ "epoch": 19.93732970027248,
+ "grad_norm": 11.414156913757324,
+ "learning_rate": 1.853277003374978e-05,
+ "loss": 0.87,
+ "step": 7317
+ },
+ {
+ "epoch": 19.940054495912808,
+ "grad_norm": 13.206284523010254,
+ "learning_rate": 1.8532309816712542e-05,
+ "loss": 0.7273,
+ "step": 7318
+ },
+ {
+ "epoch": 19.942779291553133,
+ "grad_norm": 13.142497062683105,
+ "learning_rate": 1.8531849533225948e-05,
+ "loss": 0.9993,
+ "step": 7319
+ },
+ {
+ "epoch": 19.94550408719346,
+ "grad_norm": 12.856634140014648,
+ "learning_rate": 1.8531389183293583e-05,
+ "loss": 0.8236,
+ "step": 7320
+ },
+ {
+ "epoch": 19.94822888283379,
+ "grad_norm": 10.547633171081543,
+ "learning_rate": 1.853092876691904e-05,
+ "loss": 1.0232,
+ "step": 7321
+ },
+ {
+ "epoch": 19.950953678474114,
+ "grad_norm": 9.29246997833252,
+ "learning_rate": 1.853046828410589e-05,
+ "loss": 0.8094,
+ "step": 7322
+ },
+ {
+ "epoch": 19.95367847411444,
+ "grad_norm": 8.399765968322754,
+ "learning_rate": 1.8530007734857734e-05,
+ "loss": 0.6976,
+ "step": 7323
+ },
+ {
+ "epoch": 19.95640326975477,
+ "grad_norm": 13.324989318847656,
+ "learning_rate": 1.852954711917815e-05,
+ "loss": 0.7832,
+ "step": 7324
+ },
+ {
+ "epoch": 19.959128065395095,
+ "grad_norm": 9.017197608947754,
+ "learning_rate": 1.8529086437070726e-05,
+ "loss": 0.8633,
+ "step": 7325
+ },
+ {
+ "epoch": 19.96185286103542,
+ "grad_norm": 8.983302116394043,
+ "learning_rate": 1.8528625688539055e-05,
+ "loss": 0.8477,
+ "step": 7326
+ },
+ {
+ "epoch": 19.96457765667575,
+ "grad_norm": 8.390277862548828,
+ "learning_rate": 1.8528164873586717e-05,
+ "loss": 0.8243,
+ "step": 7327
+ },
+ {
+ "epoch": 19.967302452316076,
+ "grad_norm": 18.7827091217041,
+ "learning_rate": 1.852770399221731e-05,
+ "loss": 0.905,
+ "step": 7328
+ },
+ {
+ "epoch": 19.970027247956402,
+ "grad_norm": 9.379305839538574,
+ "learning_rate": 1.8527243044434415e-05,
+ "loss": 0.7304,
+ "step": 7329
+ },
+ {
+ "epoch": 19.97275204359673,
+ "grad_norm": 7.8202738761901855,
+ "learning_rate": 1.852678203024163e-05,
+ "loss": 0.8698,
+ "step": 7330
+ },
+ {
+ "epoch": 19.975476839237057,
+ "grad_norm": 9.599730491638184,
+ "learning_rate": 1.8526320949642536e-05,
+ "loss": 0.8987,
+ "step": 7331
+ },
+ {
+ "epoch": 19.978201634877383,
+ "grad_norm": 8.737628936767578,
+ "learning_rate": 1.8525859802640733e-05,
+ "loss": 0.6768,
+ "step": 7332
+ },
+ {
+ "epoch": 19.980926430517712,
+ "grad_norm": 8.598307609558105,
+ "learning_rate": 1.8525398589239803e-05,
+ "loss": 0.8574,
+ "step": 7333
+ },
+ {
+ "epoch": 19.983651226158038,
+ "grad_norm": 11.687674522399902,
+ "learning_rate": 1.8524937309443346e-05,
+ "loss": 0.7122,
+ "step": 7334
+ },
+ {
+ "epoch": 19.986376021798364,
+ "grad_norm": 10.34839916229248,
+ "learning_rate": 1.8524475963254955e-05,
+ "loss": 0.8394,
+ "step": 7335
+ },
+ {
+ "epoch": 19.989100817438693,
+ "grad_norm": 7.692890167236328,
+ "learning_rate": 1.8524014550678212e-05,
+ "loss": 0.6886,
+ "step": 7336
+ },
+ {
+ "epoch": 19.99182561307902,
+ "grad_norm": 8.730321884155273,
+ "learning_rate": 1.8523553071716726e-05,
+ "loss": 0.6675,
+ "step": 7337
+ },
+ {
+ "epoch": 19.994550408719345,
+ "grad_norm": 9.423654556274414,
+ "learning_rate": 1.852309152637408e-05,
+ "loss": 0.8239,
+ "step": 7338
+ },
+ {
+ "epoch": 19.997275204359674,
+ "grad_norm": 10.066039085388184,
+ "learning_rate": 1.852262991465387e-05,
+ "loss": 0.808,
+ "step": 7339
+ },
+ {
+ "epoch": 20.0,
+ "grad_norm": 6.156219959259033,
+ "learning_rate": 1.8522168236559693e-05,
+ "loss": 0.5946,
+ "step": 7340
+ },
+ {
+ "epoch": 20.002724795640326,
+ "grad_norm": 9.319891929626465,
+ "learning_rate": 1.8521706492095147e-05,
+ "loss": 0.8275,
+ "step": 7341
+ },
+ {
+ "epoch": 20.005449591280655,
+ "grad_norm": 8.304445266723633,
+ "learning_rate": 1.852124468126382e-05,
+ "loss": 0.7894,
+ "step": 7342
+ },
+ {
+ "epoch": 20.00817438692098,
+ "grad_norm": 9.10930061340332,
+ "learning_rate": 1.8520782804069317e-05,
+ "loss": 0.571,
+ "step": 7343
+ },
+ {
+ "epoch": 20.010899182561307,
+ "grad_norm": 7.984755992889404,
+ "learning_rate": 1.8520320860515233e-05,
+ "loss": 0.6566,
+ "step": 7344
+ },
+ {
+ "epoch": 20.013623978201636,
+ "grad_norm": 7.1055588722229,
+ "learning_rate": 1.851985885060516e-05,
+ "loss": 0.6488,
+ "step": 7345
+ },
+ {
+ "epoch": 20.016348773841962,
+ "grad_norm": 9.93221664428711,
+ "learning_rate": 1.8519396774342707e-05,
+ "loss": 0.6732,
+ "step": 7346
+ },
+ {
+ "epoch": 20.019073569482288,
+ "grad_norm": 7.15091609954834,
+ "learning_rate": 1.8518934631731463e-05,
+ "loss": 0.776,
+ "step": 7347
+ },
+ {
+ "epoch": 20.021798365122617,
+ "grad_norm": 8.36777114868164,
+ "learning_rate": 1.8518472422775027e-05,
+ "loss": 0.5845,
+ "step": 7348
+ },
+ {
+ "epoch": 20.024523160762943,
+ "grad_norm": 6.9597039222717285,
+ "learning_rate": 1.8518010147477007e-05,
+ "loss": 0.6794,
+ "step": 7349
+ },
+ {
+ "epoch": 20.02724795640327,
+ "grad_norm": 6.552559852600098,
+ "learning_rate": 1.8517547805840995e-05,
+ "loss": 0.894,
+ "step": 7350
+ },
+ {
+ "epoch": 20.029972752043598,
+ "grad_norm": 5.9363789558410645,
+ "learning_rate": 1.8517085397870594e-05,
+ "loss": 0.6094,
+ "step": 7351
+ },
+ {
+ "epoch": 20.032697547683924,
+ "grad_norm": 8.362524032592773,
+ "learning_rate": 1.8516622923569406e-05,
+ "loss": 0.6586,
+ "step": 7352
+ },
+ {
+ "epoch": 20.03542234332425,
+ "grad_norm": 7.386313438415527,
+ "learning_rate": 1.8516160382941035e-05,
+ "loss": 0.7511,
+ "step": 7353
+ },
+ {
+ "epoch": 20.03814713896458,
+ "grad_norm": 7.5082902908325195,
+ "learning_rate": 1.851569777598908e-05,
+ "loss": 0.5984,
+ "step": 7354
+ },
+ {
+ "epoch": 20.040871934604905,
+ "grad_norm": 7.392009735107422,
+ "learning_rate": 1.8515235102717145e-05,
+ "loss": 0.5814,
+ "step": 7355
+ },
+ {
+ "epoch": 20.04359673024523,
+ "grad_norm": 7.026287078857422,
+ "learning_rate": 1.8514772363128833e-05,
+ "loss": 0.7058,
+ "step": 7356
+ },
+ {
+ "epoch": 20.04632152588556,
+ "grad_norm": 7.165449619293213,
+ "learning_rate": 1.851430955722775e-05,
+ "loss": 0.5007,
+ "step": 7357
+ },
+ {
+ "epoch": 20.049046321525886,
+ "grad_norm": 8.432865142822266,
+ "learning_rate": 1.8513846685017498e-05,
+ "loss": 0.6886,
+ "step": 7358
+ },
+ {
+ "epoch": 20.05177111716621,
+ "grad_norm": 8.437450408935547,
+ "learning_rate": 1.8513383746501678e-05,
+ "loss": 0.7836,
+ "step": 7359
+ },
+ {
+ "epoch": 20.05449591280654,
+ "grad_norm": 6.300036907196045,
+ "learning_rate": 1.85129207416839e-05,
+ "loss": 0.6027,
+ "step": 7360
+ },
+ {
+ "epoch": 20.057220708446867,
+ "grad_norm": 7.2547149658203125,
+ "learning_rate": 1.8512457670567773e-05,
+ "loss": 0.6791,
+ "step": 7361
+ },
+ {
+ "epoch": 20.059945504087192,
+ "grad_norm": 7.099516868591309,
+ "learning_rate": 1.8511994533156898e-05,
+ "loss": 0.7173,
+ "step": 7362
+ },
+ {
+ "epoch": 20.06267029972752,
+ "grad_norm": 8.449621200561523,
+ "learning_rate": 1.8511531329454883e-05,
+ "loss": 0.6081,
+ "step": 7363
+ },
+ {
+ "epoch": 20.065395095367847,
+ "grad_norm": 6.550309181213379,
+ "learning_rate": 1.8511068059465334e-05,
+ "loss": 0.6102,
+ "step": 7364
+ },
+ {
+ "epoch": 20.068119891008173,
+ "grad_norm": 8.018385887145996,
+ "learning_rate": 1.8510604723191865e-05,
+ "loss": 0.6094,
+ "step": 7365
+ },
+ {
+ "epoch": 20.070844686648503,
+ "grad_norm": 6.489816188812256,
+ "learning_rate": 1.8510141320638078e-05,
+ "loss": 0.61,
+ "step": 7366
+ },
+ {
+ "epoch": 20.07356948228883,
+ "grad_norm": 5.6420698165893555,
+ "learning_rate": 1.8509677851807584e-05,
+ "loss": 0.6945,
+ "step": 7367
+ },
+ {
+ "epoch": 20.076294277929154,
+ "grad_norm": 7.573943614959717,
+ "learning_rate": 1.8509214316703993e-05,
+ "loss": 0.7859,
+ "step": 7368
+ },
+ {
+ "epoch": 20.079019073569484,
+ "grad_norm": 8.612110137939453,
+ "learning_rate": 1.8508750715330913e-05,
+ "loss": 0.5538,
+ "step": 7369
+ },
+ {
+ "epoch": 20.08174386920981,
+ "grad_norm": 8.293294906616211,
+ "learning_rate": 1.850828704769196e-05,
+ "loss": 0.6544,
+ "step": 7370
+ },
+ {
+ "epoch": 20.084468664850135,
+ "grad_norm": 10.102967262268066,
+ "learning_rate": 1.850782331379074e-05,
+ "loss": 0.6469,
+ "step": 7371
+ },
+ {
+ "epoch": 20.087193460490465,
+ "grad_norm": 6.807920455932617,
+ "learning_rate": 1.8507359513630865e-05,
+ "loss": 0.5916,
+ "step": 7372
+ },
+ {
+ "epoch": 20.08991825613079,
+ "grad_norm": 7.522378444671631,
+ "learning_rate": 1.850689564721595e-05,
+ "loss": 0.6233,
+ "step": 7373
+ },
+ {
+ "epoch": 20.092643051771116,
+ "grad_norm": 7.564815998077393,
+ "learning_rate": 1.8506431714549606e-05,
+ "loss": 0.7704,
+ "step": 7374
+ },
+ {
+ "epoch": 20.095367847411445,
+ "grad_norm": 7.773350238800049,
+ "learning_rate": 1.850596771563544e-05,
+ "loss": 0.6683,
+ "step": 7375
+ },
+ {
+ "epoch": 20.09809264305177,
+ "grad_norm": 7.9525251388549805,
+ "learning_rate": 1.8505503650477074e-05,
+ "loss": 0.688,
+ "step": 7376
+ },
+ {
+ "epoch": 20.100817438692097,
+ "grad_norm": 6.03648042678833,
+ "learning_rate": 1.850503951907812e-05,
+ "loss": 0.7654,
+ "step": 7377
+ },
+ {
+ "epoch": 20.103542234332426,
+ "grad_norm": 8.273544311523438,
+ "learning_rate": 1.8504575321442194e-05,
+ "loss": 0.6237,
+ "step": 7378
+ },
+ {
+ "epoch": 20.106267029972752,
+ "grad_norm": 6.682157039642334,
+ "learning_rate": 1.8504111057572905e-05,
+ "loss": 0.6143,
+ "step": 7379
+ },
+ {
+ "epoch": 20.108991825613078,
+ "grad_norm": 8.839594841003418,
+ "learning_rate": 1.8503646727473877e-05,
+ "loss": 0.6196,
+ "step": 7380
+ },
+ {
+ "epoch": 20.111716621253407,
+ "grad_norm": 6.570153713226318,
+ "learning_rate": 1.850318233114872e-05,
+ "loss": 0.4948,
+ "step": 7381
+ },
+ {
+ "epoch": 20.114441416893733,
+ "grad_norm": 6.142697811126709,
+ "learning_rate": 1.8502717868601054e-05,
+ "loss": 0.5244,
+ "step": 7382
+ },
+ {
+ "epoch": 20.11716621253406,
+ "grad_norm": 7.722724437713623,
+ "learning_rate": 1.850225333983449e-05,
+ "loss": 0.6463,
+ "step": 7383
+ },
+ {
+ "epoch": 20.11989100817439,
+ "grad_norm": 12.669514656066895,
+ "learning_rate": 1.8501788744852652e-05,
+ "loss": 0.6289,
+ "step": 7384
+ },
+ {
+ "epoch": 20.122615803814714,
+ "grad_norm": 6.122628688812256,
+ "learning_rate": 1.850132408365916e-05,
+ "loss": 0.491,
+ "step": 7385
+ },
+ {
+ "epoch": 20.12534059945504,
+ "grad_norm": 6.307292461395264,
+ "learning_rate": 1.850085935625763e-05,
+ "loss": 0.6992,
+ "step": 7386
+ },
+ {
+ "epoch": 20.12806539509537,
+ "grad_norm": 7.345569133758545,
+ "learning_rate": 1.850039456265168e-05,
+ "loss": 0.72,
+ "step": 7387
+ },
+ {
+ "epoch": 20.130790190735695,
+ "grad_norm": 8.138498306274414,
+ "learning_rate": 1.8499929702844932e-05,
+ "loss": 0.6212,
+ "step": 7388
+ },
+ {
+ "epoch": 20.13351498637602,
+ "grad_norm": 7.170979022979736,
+ "learning_rate": 1.8499464776841e-05,
+ "loss": 0.6086,
+ "step": 7389
+ },
+ {
+ "epoch": 20.13623978201635,
+ "grad_norm": 7.619695663452148,
+ "learning_rate": 1.8498999784643517e-05,
+ "loss": 0.6611,
+ "step": 7390
+ },
+ {
+ "epoch": 20.138964577656676,
+ "grad_norm": 5.397146224975586,
+ "learning_rate": 1.8498534726256094e-05,
+ "loss": 0.6439,
+ "step": 7391
+ },
+ {
+ "epoch": 20.141689373297,
+ "grad_norm": 6.029022216796875,
+ "learning_rate": 1.8498069601682353e-05,
+ "loss": 0.6318,
+ "step": 7392
+ },
+ {
+ "epoch": 20.14441416893733,
+ "grad_norm": 5.875804901123047,
+ "learning_rate": 1.8497604410925923e-05,
+ "loss": 0.6078,
+ "step": 7393
+ },
+ {
+ "epoch": 20.147138964577657,
+ "grad_norm": 6.126670837402344,
+ "learning_rate": 1.849713915399042e-05,
+ "loss": 0.6259,
+ "step": 7394
+ },
+ {
+ "epoch": 20.149863760217983,
+ "grad_norm": 6.460773944854736,
+ "learning_rate": 1.8496673830879475e-05,
+ "loss": 0.5081,
+ "step": 7395
+ },
+ {
+ "epoch": 20.152588555858312,
+ "grad_norm": 7.232714653015137,
+ "learning_rate": 1.8496208441596706e-05,
+ "loss": 0.6108,
+ "step": 7396
+ },
+ {
+ "epoch": 20.155313351498638,
+ "grad_norm": 6.821713924407959,
+ "learning_rate": 1.8495742986145737e-05,
+ "loss": 0.5094,
+ "step": 7397
+ },
+ {
+ "epoch": 20.158038147138964,
+ "grad_norm": 7.970686435699463,
+ "learning_rate": 1.8495277464530198e-05,
+ "loss": 0.5845,
+ "step": 7398
+ },
+ {
+ "epoch": 20.160762942779293,
+ "grad_norm": 6.9570817947387695,
+ "learning_rate": 1.8494811876753712e-05,
+ "loss": 0.5705,
+ "step": 7399
+ },
+ {
+ "epoch": 20.16348773841962,
+ "grad_norm": 10.322071075439453,
+ "learning_rate": 1.84943462228199e-05,
+ "loss": 0.5936,
+ "step": 7400
+ },
+ {
+ "epoch": 20.166212534059945,
+ "grad_norm": 7.291942596435547,
+ "learning_rate": 1.84938805027324e-05,
+ "loss": 0.6395,
+ "step": 7401
+ },
+ {
+ "epoch": 20.168937329700274,
+ "grad_norm": 8.336548805236816,
+ "learning_rate": 1.8493414716494826e-05,
+ "loss": 0.5478,
+ "step": 7402
+ },
+ {
+ "epoch": 20.1716621253406,
+ "grad_norm": 7.6306891441345215,
+ "learning_rate": 1.849294886411081e-05,
+ "loss": 0.559,
+ "step": 7403
+ },
+ {
+ "epoch": 20.174386920980925,
+ "grad_norm": 6.883423328399658,
+ "learning_rate": 1.849248294558399e-05,
+ "loss": 0.5792,
+ "step": 7404
+ },
+ {
+ "epoch": 20.177111716621255,
+ "grad_norm": 14.581059455871582,
+ "learning_rate": 1.849201696091798e-05,
+ "loss": 0.5329,
+ "step": 7405
+ },
+ {
+ "epoch": 20.17983651226158,
+ "grad_norm": 31.70160484313965,
+ "learning_rate": 1.8491550910116415e-05,
+ "loss": 0.6996,
+ "step": 7406
+ },
+ {
+ "epoch": 20.182561307901906,
+ "grad_norm": 5.6532111167907715,
+ "learning_rate": 1.8491084793182925e-05,
+ "loss": 0.6974,
+ "step": 7407
+ },
+ {
+ "epoch": 20.185286103542236,
+ "grad_norm": 7.053765773773193,
+ "learning_rate": 1.8490618610121144e-05,
+ "loss": 0.4703,
+ "step": 7408
+ },
+ {
+ "epoch": 20.18801089918256,
+ "grad_norm": 6.918330669403076,
+ "learning_rate": 1.8490152360934695e-05,
+ "loss": 0.524,
+ "step": 7409
+ },
+ {
+ "epoch": 20.190735694822887,
+ "grad_norm": 5.365565776824951,
+ "learning_rate": 1.848968604562721e-05,
+ "loss": 0.6478,
+ "step": 7410
+ },
+ {
+ "epoch": 20.193460490463217,
+ "grad_norm": 7.114992141723633,
+ "learning_rate": 1.848921966420233e-05,
+ "loss": 0.5775,
+ "step": 7411
+ },
+ {
+ "epoch": 20.196185286103542,
+ "grad_norm": 7.654002666473389,
+ "learning_rate": 1.8488753216663675e-05,
+ "loss": 0.6323,
+ "step": 7412
+ },
+ {
+ "epoch": 20.19891008174387,
+ "grad_norm": 7.621139049530029,
+ "learning_rate": 1.848828670301489e-05,
+ "loss": 0.6014,
+ "step": 7413
+ },
+ {
+ "epoch": 20.201634877384198,
+ "grad_norm": 7.0192975997924805,
+ "learning_rate": 1.8487820123259595e-05,
+ "loss": 0.5389,
+ "step": 7414
+ },
+ {
+ "epoch": 20.204359673024523,
+ "grad_norm": 7.479555130004883,
+ "learning_rate": 1.848735347740143e-05,
+ "loss": 0.6378,
+ "step": 7415
+ },
+ {
+ "epoch": 20.20708446866485,
+ "grad_norm": 7.342499256134033,
+ "learning_rate": 1.848688676544403e-05,
+ "loss": 0.7016,
+ "step": 7416
+ },
+ {
+ "epoch": 20.20980926430518,
+ "grad_norm": 8.390324592590332,
+ "learning_rate": 1.848641998739103e-05,
+ "loss": 0.6399,
+ "step": 7417
+ },
+ {
+ "epoch": 20.212534059945504,
+ "grad_norm": 7.939183235168457,
+ "learning_rate": 1.8485953143246065e-05,
+ "loss": 0.5072,
+ "step": 7418
+ },
+ {
+ "epoch": 20.21525885558583,
+ "grad_norm": 6.3030595779418945,
+ "learning_rate": 1.848548623301277e-05,
+ "loss": 0.5161,
+ "step": 7419
+ },
+ {
+ "epoch": 20.21798365122616,
+ "grad_norm": 6.817437648773193,
+ "learning_rate": 1.848501925669478e-05,
+ "loss": 0.5904,
+ "step": 7420
+ },
+ {
+ "epoch": 20.220708446866485,
+ "grad_norm": 7.666940212249756,
+ "learning_rate": 1.8484552214295735e-05,
+ "loss": 0.5455,
+ "step": 7421
+ },
+ {
+ "epoch": 20.22343324250681,
+ "grad_norm": 6.85036039352417,
+ "learning_rate": 1.8484085105819267e-05,
+ "loss": 0.6516,
+ "step": 7422
+ },
+ {
+ "epoch": 20.22615803814714,
+ "grad_norm": 9.209121704101562,
+ "learning_rate": 1.8483617931269022e-05,
+ "loss": 0.5706,
+ "step": 7423
+ },
+ {
+ "epoch": 20.228882833787466,
+ "grad_norm": 6.071787357330322,
+ "learning_rate": 1.848315069064863e-05,
+ "loss": 0.565,
+ "step": 7424
+ },
+ {
+ "epoch": 20.231607629427792,
+ "grad_norm": 7.05095100402832,
+ "learning_rate": 1.8482683383961734e-05,
+ "loss": 0.4854,
+ "step": 7425
+ },
+ {
+ "epoch": 20.23433242506812,
+ "grad_norm": 7.849099159240723,
+ "learning_rate": 1.8482216011211976e-05,
+ "loss": 0.593,
+ "step": 7426
+ },
+ {
+ "epoch": 20.237057220708447,
+ "grad_norm": 7.368453025817871,
+ "learning_rate": 1.8481748572402986e-05,
+ "loss": 0.6931,
+ "step": 7427
+ },
+ {
+ "epoch": 20.239782016348773,
+ "grad_norm": 7.857269763946533,
+ "learning_rate": 1.8481281067538416e-05,
+ "loss": 0.5957,
+ "step": 7428
+ },
+ {
+ "epoch": 20.242506811989102,
+ "grad_norm": 7.017577171325684,
+ "learning_rate": 1.8480813496621902e-05,
+ "loss": 0.6511,
+ "step": 7429
+ },
+ {
+ "epoch": 20.245231607629428,
+ "grad_norm": 6.279410362243652,
+ "learning_rate": 1.8480345859657085e-05,
+ "loss": 0.6472,
+ "step": 7430
+ },
+ {
+ "epoch": 20.247956403269754,
+ "grad_norm": 6.576488971710205,
+ "learning_rate": 1.8479878156647606e-05,
+ "loss": 0.5498,
+ "step": 7431
+ },
+ {
+ "epoch": 20.250681198910083,
+ "grad_norm": 6.631252288818359,
+ "learning_rate": 1.847941038759711e-05,
+ "loss": 0.6842,
+ "step": 7432
+ },
+ {
+ "epoch": 20.25340599455041,
+ "grad_norm": 6.34262228012085,
+ "learning_rate": 1.847894255250924e-05,
+ "loss": 0.5776,
+ "step": 7433
+ },
+ {
+ "epoch": 20.256130790190735,
+ "grad_norm": 7.548246383666992,
+ "learning_rate": 1.8478474651387633e-05,
+ "loss": 0.6116,
+ "step": 7434
+ },
+ {
+ "epoch": 20.258855585831064,
+ "grad_norm": 7.438286304473877,
+ "learning_rate": 1.8478006684235944e-05,
+ "loss": 0.6305,
+ "step": 7435
+ },
+ {
+ "epoch": 20.26158038147139,
+ "grad_norm": 5.9378981590271,
+ "learning_rate": 1.8477538651057812e-05,
+ "loss": 0.4507,
+ "step": 7436
+ },
+ {
+ "epoch": 20.264305177111716,
+ "grad_norm": 7.063061714172363,
+ "learning_rate": 1.847707055185688e-05,
+ "loss": 0.6778,
+ "step": 7437
+ },
+ {
+ "epoch": 20.267029972752045,
+ "grad_norm": 6.514150142669678,
+ "learning_rate": 1.8476602386636795e-05,
+ "loss": 0.5745,
+ "step": 7438
+ },
+ {
+ "epoch": 20.26975476839237,
+ "grad_norm": 7.343629360198975,
+ "learning_rate": 1.8476134155401202e-05,
+ "loss": 0.4822,
+ "step": 7439
+ },
+ {
+ "epoch": 20.272479564032697,
+ "grad_norm": 6.159067153930664,
+ "learning_rate": 1.8475665858153752e-05,
+ "loss": 0.5482,
+ "step": 7440
+ },
+ {
+ "epoch": 20.275204359673026,
+ "grad_norm": 6.265310764312744,
+ "learning_rate": 1.847519749489809e-05,
+ "loss": 0.5963,
+ "step": 7441
+ },
+ {
+ "epoch": 20.277929155313352,
+ "grad_norm": 5.768505573272705,
+ "learning_rate": 1.8474729065637858e-05,
+ "loss": 0.6273,
+ "step": 7442
+ },
+ {
+ "epoch": 20.280653950953678,
+ "grad_norm": 7.319711685180664,
+ "learning_rate": 1.847426057037671e-05,
+ "loss": 0.6104,
+ "step": 7443
+ },
+ {
+ "epoch": 20.283378746594007,
+ "grad_norm": 7.702725410461426,
+ "learning_rate": 1.8473792009118297e-05,
+ "loss": 0.595,
+ "step": 7444
+ },
+ {
+ "epoch": 20.286103542234333,
+ "grad_norm": 5.889584064483643,
+ "learning_rate": 1.8473323381866262e-05,
+ "loss": 0.6074,
+ "step": 7445
+ },
+ {
+ "epoch": 20.28882833787466,
+ "grad_norm": 6.7178568840026855,
+ "learning_rate": 1.8472854688624255e-05,
+ "loss": 0.5411,
+ "step": 7446
+ },
+ {
+ "epoch": 20.291553133514988,
+ "grad_norm": 9.406813621520996,
+ "learning_rate": 1.847238592939593e-05,
+ "loss": 0.6263,
+ "step": 7447
+ },
+ {
+ "epoch": 20.294277929155314,
+ "grad_norm": 6.751778602600098,
+ "learning_rate": 1.847191710418494e-05,
+ "loss": 0.603,
+ "step": 7448
+ },
+ {
+ "epoch": 20.29700272479564,
+ "grad_norm": 7.132476806640625,
+ "learning_rate": 1.8471448212994927e-05,
+ "loss": 0.6511,
+ "step": 7449
+ },
+ {
+ "epoch": 20.29972752043597,
+ "grad_norm": 7.657885551452637,
+ "learning_rate": 1.847097925582955e-05,
+ "loss": 0.697,
+ "step": 7450
+ },
+ {
+ "epoch": 20.302452316076295,
+ "grad_norm": 37.21819305419922,
+ "learning_rate": 1.847051023269246e-05,
+ "loss": 0.5516,
+ "step": 7451
+ },
+ {
+ "epoch": 20.30517711171662,
+ "grad_norm": 7.033862590789795,
+ "learning_rate": 1.8470041143587307e-05,
+ "loss": 0.7551,
+ "step": 7452
+ },
+ {
+ "epoch": 20.30790190735695,
+ "grad_norm": 6.545276641845703,
+ "learning_rate": 1.846957198851775e-05,
+ "loss": 0.6677,
+ "step": 7453
+ },
+ {
+ "epoch": 20.310626702997276,
+ "grad_norm": 9.14720630645752,
+ "learning_rate": 1.8469102767487434e-05,
+ "loss": 0.6114,
+ "step": 7454
+ },
+ {
+ "epoch": 20.3133514986376,
+ "grad_norm": 9.293963432312012,
+ "learning_rate": 1.8468633480500024e-05,
+ "loss": 0.7546,
+ "step": 7455
+ },
+ {
+ "epoch": 20.31607629427793,
+ "grad_norm": 6.091339588165283,
+ "learning_rate": 1.8468164127559162e-05,
+ "loss": 0.5432,
+ "step": 7456
+ },
+ {
+ "epoch": 20.318801089918257,
+ "grad_norm": 6.849093914031982,
+ "learning_rate": 1.8467694708668516e-05,
+ "loss": 0.7089,
+ "step": 7457
+ },
+ {
+ "epoch": 20.321525885558582,
+ "grad_norm": 6.173511505126953,
+ "learning_rate": 1.8467225223831733e-05,
+ "loss": 0.6547,
+ "step": 7458
+ },
+ {
+ "epoch": 20.32425068119891,
+ "grad_norm": 12.58679485321045,
+ "learning_rate": 1.8466755673052472e-05,
+ "loss": 0.6376,
+ "step": 7459
+ },
+ {
+ "epoch": 20.326975476839237,
+ "grad_norm": 7.1735758781433105,
+ "learning_rate": 1.8466286056334394e-05,
+ "loss": 0.7063,
+ "step": 7460
+ },
+ {
+ "epoch": 20.329700272479563,
+ "grad_norm": 7.652830123901367,
+ "learning_rate": 1.846581637368115e-05,
+ "loss": 0.5396,
+ "step": 7461
+ },
+ {
+ "epoch": 20.332425068119893,
+ "grad_norm": 6.747897148132324,
+ "learning_rate": 1.84653466250964e-05,
+ "loss": 0.6112,
+ "step": 7462
+ },
+ {
+ "epoch": 20.33514986376022,
+ "grad_norm": 6.812202453613281,
+ "learning_rate": 1.8464876810583806e-05,
+ "loss": 0.6353,
+ "step": 7463
+ },
+ {
+ "epoch": 20.337874659400544,
+ "grad_norm": 6.634808540344238,
+ "learning_rate": 1.846440693014702e-05,
+ "loss": 0.7388,
+ "step": 7464
+ },
+ {
+ "epoch": 20.340599455040874,
+ "grad_norm": 6.901304721832275,
+ "learning_rate": 1.8463936983789706e-05,
+ "loss": 0.6673,
+ "step": 7465
+ },
+ {
+ "epoch": 20.3433242506812,
+ "grad_norm": 7.048333644866943,
+ "learning_rate": 1.8463466971515525e-05,
+ "loss": 0.6387,
+ "step": 7466
+ },
+ {
+ "epoch": 20.346049046321525,
+ "grad_norm": 5.139136791229248,
+ "learning_rate": 1.8462996893328134e-05,
+ "loss": 0.6248,
+ "step": 7467
+ },
+ {
+ "epoch": 20.348773841961854,
+ "grad_norm": 8.944092750549316,
+ "learning_rate": 1.84625267492312e-05,
+ "loss": 0.7455,
+ "step": 7468
+ },
+ {
+ "epoch": 20.35149863760218,
+ "grad_norm": 6.550175666809082,
+ "learning_rate": 1.8462056539228378e-05,
+ "loss": 0.6135,
+ "step": 7469
+ },
+ {
+ "epoch": 20.354223433242506,
+ "grad_norm": 6.428384304046631,
+ "learning_rate": 1.8461586263323327e-05,
+ "loss": 0.7493,
+ "step": 7470
+ },
+ {
+ "epoch": 20.356948228882835,
+ "grad_norm": 6.537440299987793,
+ "learning_rate": 1.846111592151972e-05,
+ "loss": 0.4592,
+ "step": 7471
+ },
+ {
+ "epoch": 20.35967302452316,
+ "grad_norm": 8.14555835723877,
+ "learning_rate": 1.846064551382121e-05,
+ "loss": 0.8044,
+ "step": 7472
+ },
+ {
+ "epoch": 20.362397820163487,
+ "grad_norm": 7.384801387786865,
+ "learning_rate": 1.8460175040231468e-05,
+ "loss": 0.7201,
+ "step": 7473
+ },
+ {
+ "epoch": 20.365122615803816,
+ "grad_norm": 5.848476409912109,
+ "learning_rate": 1.8459704500754155e-05,
+ "loss": 0.5886,
+ "step": 7474
+ },
+ {
+ "epoch": 20.367847411444142,
+ "grad_norm": 6.191623210906982,
+ "learning_rate": 1.8459233895392933e-05,
+ "loss": 0.6442,
+ "step": 7475
+ },
+ {
+ "epoch": 20.370572207084468,
+ "grad_norm": 5.926096439361572,
+ "learning_rate": 1.8458763224151476e-05,
+ "loss": 0.8867,
+ "step": 7476
+ },
+ {
+ "epoch": 20.373297002724797,
+ "grad_norm": 7.30550479888916,
+ "learning_rate": 1.845829248703344e-05,
+ "loss": 0.6277,
+ "step": 7477
+ },
+ {
+ "epoch": 20.376021798365123,
+ "grad_norm": 7.423349380493164,
+ "learning_rate": 1.8457821684042492e-05,
+ "loss": 0.6702,
+ "step": 7478
+ },
+ {
+ "epoch": 20.37874659400545,
+ "grad_norm": 6.720351696014404,
+ "learning_rate": 1.8457350815182303e-05,
+ "loss": 0.5789,
+ "step": 7479
+ },
+ {
+ "epoch": 20.381471389645778,
+ "grad_norm": 5.1237969398498535,
+ "learning_rate": 1.8456879880456537e-05,
+ "loss": 0.6006,
+ "step": 7480
+ },
+ {
+ "epoch": 20.384196185286104,
+ "grad_norm": 7.060522556304932,
+ "learning_rate": 1.8456408879868864e-05,
+ "loss": 0.7378,
+ "step": 7481
+ },
+ {
+ "epoch": 20.38692098092643,
+ "grad_norm": 6.870255470275879,
+ "learning_rate": 1.845593781342295e-05,
+ "loss": 0.4823,
+ "step": 7482
+ },
+ {
+ "epoch": 20.38964577656676,
+ "grad_norm": 7.552305698394775,
+ "learning_rate": 1.8455466681122463e-05,
+ "loss": 0.6057,
+ "step": 7483
+ },
+ {
+ "epoch": 20.392370572207085,
+ "grad_norm": 5.855003833770752,
+ "learning_rate": 1.8454995482971077e-05,
+ "loss": 0.7471,
+ "step": 7484
+ },
+ {
+ "epoch": 20.39509536784741,
+ "grad_norm": 6.74839973449707,
+ "learning_rate": 1.8454524218972457e-05,
+ "loss": 0.6536,
+ "step": 7485
+ },
+ {
+ "epoch": 20.39782016348774,
+ "grad_norm": 7.653990268707275,
+ "learning_rate": 1.8454052889130273e-05,
+ "loss": 0.6367,
+ "step": 7486
+ },
+ {
+ "epoch": 20.400544959128066,
+ "grad_norm": 6.498241424560547,
+ "learning_rate": 1.84535814934482e-05,
+ "loss": 0.5942,
+ "step": 7487
+ },
+ {
+ "epoch": 20.40326975476839,
+ "grad_norm": 6.963813304901123,
+ "learning_rate": 1.84531100319299e-05,
+ "loss": 0.776,
+ "step": 7488
+ },
+ {
+ "epoch": 20.40599455040872,
+ "grad_norm": 6.597719669342041,
+ "learning_rate": 1.8452638504579056e-05,
+ "loss": 0.5381,
+ "step": 7489
+ },
+ {
+ "epoch": 20.408719346049047,
+ "grad_norm": 6.298782825469971,
+ "learning_rate": 1.8452166911399336e-05,
+ "loss": 0.5416,
+ "step": 7490
+ },
+ {
+ "epoch": 20.411444141689373,
+ "grad_norm": 22.0516414642334,
+ "learning_rate": 1.8451695252394413e-05,
+ "loss": 0.7064,
+ "step": 7491
+ },
+ {
+ "epoch": 20.414168937329702,
+ "grad_norm": 6.775282382965088,
+ "learning_rate": 1.8451223527567956e-05,
+ "loss": 0.7319,
+ "step": 7492
+ },
+ {
+ "epoch": 20.416893732970028,
+ "grad_norm": 6.802198886871338,
+ "learning_rate": 1.8450751736923644e-05,
+ "loss": 0.6585,
+ "step": 7493
+ },
+ {
+ "epoch": 20.419618528610354,
+ "grad_norm": 7.133429050445557,
+ "learning_rate": 1.845027988046515e-05,
+ "loss": 0.6628,
+ "step": 7494
+ },
+ {
+ "epoch": 20.422343324250683,
+ "grad_norm": 6.561207294464111,
+ "learning_rate": 1.844980795819615e-05,
+ "loss": 0.6565,
+ "step": 7495
+ },
+ {
+ "epoch": 20.42506811989101,
+ "grad_norm": 5.924463272094727,
+ "learning_rate": 1.844933597012031e-05,
+ "loss": 0.6222,
+ "step": 7496
+ },
+ {
+ "epoch": 20.427792915531334,
+ "grad_norm": 6.8265533447265625,
+ "learning_rate": 1.8448863916241324e-05,
+ "loss": 0.5431,
+ "step": 7497
+ },
+ {
+ "epoch": 20.430517711171664,
+ "grad_norm": 6.490199089050293,
+ "learning_rate": 1.844839179656285e-05,
+ "loss": 0.5709,
+ "step": 7498
+ },
+ {
+ "epoch": 20.43324250681199,
+ "grad_norm": 6.1342453956604,
+ "learning_rate": 1.8447919611088574e-05,
+ "loss": 0.6835,
+ "step": 7499
+ },
+ {
+ "epoch": 20.435967302452315,
+ "grad_norm": 8.439270973205566,
+ "learning_rate": 1.8447447359822172e-05,
+ "loss": 0.5485,
+ "step": 7500
+ },
+ {
+ "epoch": 20.438692098092645,
+ "grad_norm": 8.250201225280762,
+ "learning_rate": 1.844697504276732e-05,
+ "loss": 0.7146,
+ "step": 7501
+ },
+ {
+ "epoch": 20.44141689373297,
+ "grad_norm": 6.788997173309326,
+ "learning_rate": 1.8446502659927702e-05,
+ "loss": 0.6049,
+ "step": 7502
+ },
+ {
+ "epoch": 20.444141689373296,
+ "grad_norm": 6.106504917144775,
+ "learning_rate": 1.8446030211306993e-05,
+ "loss": 0.4502,
+ "step": 7503
+ },
+ {
+ "epoch": 20.446866485013626,
+ "grad_norm": 8.516481399536133,
+ "learning_rate": 1.8445557696908874e-05,
+ "loss": 0.662,
+ "step": 7504
+ },
+ {
+ "epoch": 20.44959128065395,
+ "grad_norm": 12.115754127502441,
+ "learning_rate": 1.844508511673702e-05,
+ "loss": 0.619,
+ "step": 7505
+ },
+ {
+ "epoch": 20.452316076294277,
+ "grad_norm": 7.15468168258667,
+ "learning_rate": 1.8444612470795114e-05,
+ "loss": 0.754,
+ "step": 7506
+ },
+ {
+ "epoch": 20.455040871934607,
+ "grad_norm": 6.674200057983398,
+ "learning_rate": 1.844413975908684e-05,
+ "loss": 0.5682,
+ "step": 7507
+ },
+ {
+ "epoch": 20.457765667574932,
+ "grad_norm": 5.785441875457764,
+ "learning_rate": 1.844366698161588e-05,
+ "loss": 0.4798,
+ "step": 7508
+ },
+ {
+ "epoch": 20.460490463215258,
+ "grad_norm": 6.9585700035095215,
+ "learning_rate": 1.844319413838591e-05,
+ "loss": 0.6736,
+ "step": 7509
+ },
+ {
+ "epoch": 20.463215258855588,
+ "grad_norm": 9.167404174804688,
+ "learning_rate": 1.8442721229400615e-05,
+ "loss": 0.5681,
+ "step": 7510
+ },
+ {
+ "epoch": 20.465940054495913,
+ "grad_norm": 8.051729202270508,
+ "learning_rate": 1.844224825466368e-05,
+ "loss": 0.6514,
+ "step": 7511
+ },
+ {
+ "epoch": 20.46866485013624,
+ "grad_norm": 7.068601608276367,
+ "learning_rate": 1.8441775214178784e-05,
+ "loss": 0.584,
+ "step": 7512
+ },
+ {
+ "epoch": 20.47138964577657,
+ "grad_norm": 7.211991310119629,
+ "learning_rate": 1.844130210794962e-05,
+ "loss": 0.6573,
+ "step": 7513
+ },
+ {
+ "epoch": 20.474114441416894,
+ "grad_norm": 6.11018705368042,
+ "learning_rate": 1.8440828935979865e-05,
+ "loss": 0.5897,
+ "step": 7514
+ },
+ {
+ "epoch": 20.47683923705722,
+ "grad_norm": 6.356911659240723,
+ "learning_rate": 1.8440355698273203e-05,
+ "loss": 0.5967,
+ "step": 7515
+ },
+ {
+ "epoch": 20.479564032697546,
+ "grad_norm": 6.632688045501709,
+ "learning_rate": 1.843988239483333e-05,
+ "loss": 0.507,
+ "step": 7516
+ },
+ {
+ "epoch": 20.482288828337875,
+ "grad_norm": 5.640937805175781,
+ "learning_rate": 1.8439409025663915e-05,
+ "loss": 0.6627,
+ "step": 7517
+ },
+ {
+ "epoch": 20.4850136239782,
+ "grad_norm": 6.745912551879883,
+ "learning_rate": 1.8438935590768658e-05,
+ "loss": 0.7139,
+ "step": 7518
+ },
+ {
+ "epoch": 20.48773841961853,
+ "grad_norm": 6.27691125869751,
+ "learning_rate": 1.8438462090151247e-05,
+ "loss": 0.671,
+ "step": 7519
+ },
+ {
+ "epoch": 20.490463215258856,
+ "grad_norm": 9.154484748840332,
+ "learning_rate": 1.843798852381536e-05,
+ "loss": 0.5403,
+ "step": 7520
+ },
+ {
+ "epoch": 20.493188010899182,
+ "grad_norm": 6.6805739402771,
+ "learning_rate": 1.843751489176469e-05,
+ "loss": 0.5941,
+ "step": 7521
+ },
+ {
+ "epoch": 20.495912806539508,
+ "grad_norm": 7.109960556030273,
+ "learning_rate": 1.8437041194002924e-05,
+ "loss": 0.4933,
+ "step": 7522
+ },
+ {
+ "epoch": 20.498637602179837,
+ "grad_norm": 7.604310512542725,
+ "learning_rate": 1.8436567430533757e-05,
+ "loss": 0.6859,
+ "step": 7523
+ },
+ {
+ "epoch": 20.501362397820163,
+ "grad_norm": 7.562671184539795,
+ "learning_rate": 1.8436093601360872e-05,
+ "loss": 0.8177,
+ "step": 7524
+ },
+ {
+ "epoch": 20.504087193460492,
+ "grad_norm": 6.853848457336426,
+ "learning_rate": 1.843561970648796e-05,
+ "loss": 0.6115,
+ "step": 7525
+ },
+ {
+ "epoch": 20.506811989100818,
+ "grad_norm": 7.657264709472656,
+ "learning_rate": 1.8435145745918713e-05,
+ "loss": 0.5498,
+ "step": 7526
+ },
+ {
+ "epoch": 20.509536784741144,
+ "grad_norm": 7.27643346786499,
+ "learning_rate": 1.8434671719656826e-05,
+ "loss": 0.5503,
+ "step": 7527
+ },
+ {
+ "epoch": 20.51226158038147,
+ "grad_norm": 6.822559356689453,
+ "learning_rate": 1.843419762770599e-05,
+ "loss": 0.8564,
+ "step": 7528
+ },
+ {
+ "epoch": 20.5149863760218,
+ "grad_norm": 6.0041022300720215,
+ "learning_rate": 1.8433723470069888e-05,
+ "loss": 0.6589,
+ "step": 7529
+ },
+ {
+ "epoch": 20.517711171662125,
+ "grad_norm": 6.933685779571533,
+ "learning_rate": 1.843324924675222e-05,
+ "loss": 0.6263,
+ "step": 7530
+ },
+ {
+ "epoch": 20.520435967302454,
+ "grad_norm": 6.376912593841553,
+ "learning_rate": 1.8432774957756677e-05,
+ "loss": 0.6069,
+ "step": 7531
+ },
+ {
+ "epoch": 20.52316076294278,
+ "grad_norm": 6.533065319061279,
+ "learning_rate": 1.8432300603086957e-05,
+ "loss": 0.7103,
+ "step": 7532
+ },
+ {
+ "epoch": 20.525885558583106,
+ "grad_norm": 6.853610515594482,
+ "learning_rate": 1.843182618274675e-05,
+ "loss": 0.6749,
+ "step": 7533
+ },
+ {
+ "epoch": 20.52861035422343,
+ "grad_norm": 6.506772041320801,
+ "learning_rate": 1.8431351696739755e-05,
+ "loss": 0.6256,
+ "step": 7534
+ },
+ {
+ "epoch": 20.53133514986376,
+ "grad_norm": 5.860067844390869,
+ "learning_rate": 1.843087714506966e-05,
+ "loss": 0.6755,
+ "step": 7535
+ },
+ {
+ "epoch": 20.534059945504087,
+ "grad_norm": 6.773581027984619,
+ "learning_rate": 1.843040252774017e-05,
+ "loss": 0.4974,
+ "step": 7536
+ },
+ {
+ "epoch": 20.536784741144416,
+ "grad_norm": 6.082549095153809,
+ "learning_rate": 1.8429927844754972e-05,
+ "loss": 0.7556,
+ "step": 7537
+ },
+ {
+ "epoch": 20.539509536784742,
+ "grad_norm": 6.2713847160339355,
+ "learning_rate": 1.842945309611777e-05,
+ "loss": 0.6793,
+ "step": 7538
+ },
+ {
+ "epoch": 20.542234332425068,
+ "grad_norm": 8.360901832580566,
+ "learning_rate": 1.8428978281832255e-05,
+ "loss": 0.5924,
+ "step": 7539
+ },
+ {
+ "epoch": 20.544959128065393,
+ "grad_norm": 6.261889457702637,
+ "learning_rate": 1.842850340190213e-05,
+ "loss": 0.6514,
+ "step": 7540
+ },
+ {
+ "epoch": 20.547683923705723,
+ "grad_norm": 6.3065948486328125,
+ "learning_rate": 1.8428028456331093e-05,
+ "loss": 0.5103,
+ "step": 7541
+ },
+ {
+ "epoch": 20.55040871934605,
+ "grad_norm": 6.293213367462158,
+ "learning_rate": 1.8427553445122842e-05,
+ "loss": 0.5329,
+ "step": 7542
+ },
+ {
+ "epoch": 20.553133514986374,
+ "grad_norm": 6.569939613342285,
+ "learning_rate": 1.8427078368281077e-05,
+ "loss": 0.6885,
+ "step": 7543
+ },
+ {
+ "epoch": 20.555858310626704,
+ "grad_norm": 5.970045566558838,
+ "learning_rate": 1.8426603225809493e-05,
+ "loss": 0.7238,
+ "step": 7544
+ },
+ {
+ "epoch": 20.55858310626703,
+ "grad_norm": 6.3127641677856445,
+ "learning_rate": 1.8426128017711797e-05,
+ "loss": 0.7026,
+ "step": 7545
+ },
+ {
+ "epoch": 20.561307901907355,
+ "grad_norm": 7.052011489868164,
+ "learning_rate": 1.8425652743991688e-05,
+ "loss": 0.6165,
+ "step": 7546
+ },
+ {
+ "epoch": 20.564032697547685,
+ "grad_norm": 6.319211959838867,
+ "learning_rate": 1.8425177404652865e-05,
+ "loss": 0.6046,
+ "step": 7547
+ },
+ {
+ "epoch": 20.56675749318801,
+ "grad_norm": 6.494566917419434,
+ "learning_rate": 1.842470199969903e-05,
+ "loss": 0.5013,
+ "step": 7548
+ },
+ {
+ "epoch": 20.569482288828336,
+ "grad_norm": 5.645216464996338,
+ "learning_rate": 1.842422652913389e-05,
+ "loss": 0.5153,
+ "step": 7549
+ },
+ {
+ "epoch": 20.572207084468666,
+ "grad_norm": 6.763337135314941,
+ "learning_rate": 1.8423750992961143e-05,
+ "loss": 0.631,
+ "step": 7550
+ },
+ {
+ "epoch": 20.57493188010899,
+ "grad_norm": 7.134343147277832,
+ "learning_rate": 1.8423275391184496e-05,
+ "loss": 0.6459,
+ "step": 7551
+ },
+ {
+ "epoch": 20.577656675749317,
+ "grad_norm": 6.727360725402832,
+ "learning_rate": 1.842279972380765e-05,
+ "loss": 0.5684,
+ "step": 7552
+ },
+ {
+ "epoch": 20.580381471389646,
+ "grad_norm": 5.283220291137695,
+ "learning_rate": 1.842232399083431e-05,
+ "loss": 0.6039,
+ "step": 7553
+ },
+ {
+ "epoch": 20.583106267029972,
+ "grad_norm": 9.500081062316895,
+ "learning_rate": 1.8421848192268187e-05,
+ "loss": 0.5726,
+ "step": 7554
+ },
+ {
+ "epoch": 20.585831062670298,
+ "grad_norm": 6.482804775238037,
+ "learning_rate": 1.8421372328112974e-05,
+ "loss": 0.5626,
+ "step": 7555
+ },
+ {
+ "epoch": 20.588555858310627,
+ "grad_norm": 6.145748138427734,
+ "learning_rate": 1.842089639837239e-05,
+ "loss": 0.6482,
+ "step": 7556
+ },
+ {
+ "epoch": 20.591280653950953,
+ "grad_norm": 5.464261531829834,
+ "learning_rate": 1.8420420403050134e-05,
+ "loss": 0.5834,
+ "step": 7557
+ },
+ {
+ "epoch": 20.59400544959128,
+ "grad_norm": 5.811768054962158,
+ "learning_rate": 1.8419944342149913e-05,
+ "loss": 0.7224,
+ "step": 7558
+ },
+ {
+ "epoch": 20.59673024523161,
+ "grad_norm": 9.587906837463379,
+ "learning_rate": 1.8419468215675437e-05,
+ "loss": 0.7137,
+ "step": 7559
+ },
+ {
+ "epoch": 20.599455040871934,
+ "grad_norm": 7.1863298416137695,
+ "learning_rate": 1.8418992023630416e-05,
+ "loss": 0.6338,
+ "step": 7560
+ },
+ {
+ "epoch": 20.60217983651226,
+ "grad_norm": 5.723667621612549,
+ "learning_rate": 1.8418515766018553e-05,
+ "loss": 0.5102,
+ "step": 7561
+ },
+ {
+ "epoch": 20.60490463215259,
+ "grad_norm": 5.703042030334473,
+ "learning_rate": 1.8418039442843562e-05,
+ "loss": 0.6384,
+ "step": 7562
+ },
+ {
+ "epoch": 20.607629427792915,
+ "grad_norm": 7.204758167266846,
+ "learning_rate": 1.8417563054109153e-05,
+ "loss": 0.6119,
+ "step": 7563
+ },
+ {
+ "epoch": 20.61035422343324,
+ "grad_norm": 5.964805603027344,
+ "learning_rate": 1.8417086599819028e-05,
+ "loss": 0.6236,
+ "step": 7564
+ },
+ {
+ "epoch": 20.61307901907357,
+ "grad_norm": 7.582167625427246,
+ "learning_rate": 1.841661007997691e-05,
+ "loss": 0.6096,
+ "step": 7565
+ },
+ {
+ "epoch": 20.615803814713896,
+ "grad_norm": 5.905071258544922,
+ "learning_rate": 1.84161334945865e-05,
+ "loss": 0.5629,
+ "step": 7566
+ },
+ {
+ "epoch": 20.618528610354222,
+ "grad_norm": 5.809918403625488,
+ "learning_rate": 1.8415656843651514e-05,
+ "loss": 0.5155,
+ "step": 7567
+ },
+ {
+ "epoch": 20.62125340599455,
+ "grad_norm": 6.695687294006348,
+ "learning_rate": 1.841518012717566e-05,
+ "loss": 0.6443,
+ "step": 7568
+ },
+ {
+ "epoch": 20.623978201634877,
+ "grad_norm": 5.845952033996582,
+ "learning_rate": 1.8414703345162656e-05,
+ "loss": 0.6078,
+ "step": 7569
+ },
+ {
+ "epoch": 20.626702997275203,
+ "grad_norm": 7.953639030456543,
+ "learning_rate": 1.8414226497616215e-05,
+ "loss": 0.5724,
+ "step": 7570
+ },
+ {
+ "epoch": 20.629427792915532,
+ "grad_norm": 8.782962799072266,
+ "learning_rate": 1.841374958454005e-05,
+ "loss": 0.7505,
+ "step": 7571
+ },
+ {
+ "epoch": 20.632152588555858,
+ "grad_norm": 7.775026798248291,
+ "learning_rate": 1.8413272605937872e-05,
+ "loss": 0.7001,
+ "step": 7572
+ },
+ {
+ "epoch": 20.634877384196184,
+ "grad_norm": 7.159183502197266,
+ "learning_rate": 1.8412795561813397e-05,
+ "loss": 0.5953,
+ "step": 7573
+ },
+ {
+ "epoch": 20.637602179836513,
+ "grad_norm": 6.495147228240967,
+ "learning_rate": 1.8412318452170344e-05,
+ "loss": 0.5932,
+ "step": 7574
+ },
+ {
+ "epoch": 20.64032697547684,
+ "grad_norm": 6.1667022705078125,
+ "learning_rate": 1.8411841277012423e-05,
+ "loss": 0.6603,
+ "step": 7575
+ },
+ {
+ "epoch": 20.643051771117165,
+ "grad_norm": 6.496042251586914,
+ "learning_rate": 1.841136403634335e-05,
+ "loss": 0.6452,
+ "step": 7576
+ },
+ {
+ "epoch": 20.645776566757494,
+ "grad_norm": 6.172253131866455,
+ "learning_rate": 1.8410886730166848e-05,
+ "loss": 0.6108,
+ "step": 7577
+ },
+ {
+ "epoch": 20.64850136239782,
+ "grad_norm": 6.109203815460205,
+ "learning_rate": 1.841040935848663e-05,
+ "loss": 0.5656,
+ "step": 7578
+ },
+ {
+ "epoch": 20.651226158038146,
+ "grad_norm": 5.7887420654296875,
+ "learning_rate": 1.840993192130642e-05,
+ "loss": 0.5042,
+ "step": 7579
+ },
+ {
+ "epoch": 20.653950953678475,
+ "grad_norm": 6.025203704833984,
+ "learning_rate": 1.8409454418629925e-05,
+ "loss": 0.4666,
+ "step": 7580
+ },
+ {
+ "epoch": 20.6566757493188,
+ "grad_norm": 6.290037631988525,
+ "learning_rate": 1.840897685046087e-05,
+ "loss": 0.693,
+ "step": 7581
+ },
+ {
+ "epoch": 20.659400544959126,
+ "grad_norm": 51.80162048339844,
+ "learning_rate": 1.8408499216802974e-05,
+ "loss": 0.5951,
+ "step": 7582
+ },
+ {
+ "epoch": 20.662125340599456,
+ "grad_norm": 7.848277568817139,
+ "learning_rate": 1.840802151765996e-05,
+ "loss": 0.4905,
+ "step": 7583
+ },
+ {
+ "epoch": 20.66485013623978,
+ "grad_norm": 7.89337158203125,
+ "learning_rate": 1.840754375303554e-05,
+ "loss": 0.6813,
+ "step": 7584
+ },
+ {
+ "epoch": 20.667574931880107,
+ "grad_norm": 5.717499732971191,
+ "learning_rate": 1.840706592293344e-05,
+ "loss": 0.5997,
+ "step": 7585
+ },
+ {
+ "epoch": 20.670299727520437,
+ "grad_norm": 6.814132213592529,
+ "learning_rate": 1.8406588027357383e-05,
+ "loss": 0.7931,
+ "step": 7586
+ },
+ {
+ "epoch": 20.673024523160763,
+ "grad_norm": 5.948505401611328,
+ "learning_rate": 1.8406110066311093e-05,
+ "loss": 0.6998,
+ "step": 7587
+ },
+ {
+ "epoch": 20.67574931880109,
+ "grad_norm": 6.887718200683594,
+ "learning_rate": 1.8405632039798283e-05,
+ "loss": 0.6923,
+ "step": 7588
+ },
+ {
+ "epoch": 20.678474114441418,
+ "grad_norm": 7.694522857666016,
+ "learning_rate": 1.840515394782268e-05,
+ "loss": 0.6591,
+ "step": 7589
+ },
+ {
+ "epoch": 20.681198910081743,
+ "grad_norm": 7.5265374183654785,
+ "learning_rate": 1.8404675790388017e-05,
+ "loss": 0.8051,
+ "step": 7590
+ },
+ {
+ "epoch": 20.68392370572207,
+ "grad_norm": 7.004940986633301,
+ "learning_rate": 1.8404197567498e-05,
+ "loss": 0.8016,
+ "step": 7591
+ },
+ {
+ "epoch": 20.6866485013624,
+ "grad_norm": 7.95519495010376,
+ "learning_rate": 1.840371927915637e-05,
+ "loss": 0.6243,
+ "step": 7592
+ },
+ {
+ "epoch": 20.689373297002724,
+ "grad_norm": 6.651413440704346,
+ "learning_rate": 1.8403240925366842e-05,
+ "loss": 0.5908,
+ "step": 7593
+ },
+ {
+ "epoch": 20.69209809264305,
+ "grad_norm": 6.6448750495910645,
+ "learning_rate": 1.8402762506133145e-05,
+ "loss": 0.5802,
+ "step": 7594
+ },
+ {
+ "epoch": 20.69482288828338,
+ "grad_norm": 7.074429035186768,
+ "learning_rate": 1.8402284021459e-05,
+ "loss": 0.5922,
+ "step": 7595
+ },
+ {
+ "epoch": 20.697547683923705,
+ "grad_norm": 7.175753116607666,
+ "learning_rate": 1.840180547134814e-05,
+ "loss": 0.6481,
+ "step": 7596
+ },
+ {
+ "epoch": 20.70027247956403,
+ "grad_norm": 7.3734846115112305,
+ "learning_rate": 1.8401326855804293e-05,
+ "loss": 0.5608,
+ "step": 7597
+ },
+ {
+ "epoch": 20.70299727520436,
+ "grad_norm": 7.359796047210693,
+ "learning_rate": 1.8400848174831178e-05,
+ "loss": 0.5191,
+ "step": 7598
+ },
+ {
+ "epoch": 20.705722070844686,
+ "grad_norm": 6.520831108093262,
+ "learning_rate": 1.840036942843253e-05,
+ "loss": 0.7526,
+ "step": 7599
+ },
+ {
+ "epoch": 20.708446866485012,
+ "grad_norm": 7.0437331199646,
+ "learning_rate": 1.8399890616612073e-05,
+ "loss": 0.7015,
+ "step": 7600
+ },
+ {
+ "epoch": 20.71117166212534,
+ "grad_norm": 6.182680606842041,
+ "learning_rate": 1.839941173937354e-05,
+ "loss": 0.5869,
+ "step": 7601
+ },
+ {
+ "epoch": 20.713896457765667,
+ "grad_norm": 6.433438777923584,
+ "learning_rate": 1.839893279672066e-05,
+ "loss": 0.7183,
+ "step": 7602
+ },
+ {
+ "epoch": 20.716621253405993,
+ "grad_norm": 7.26737117767334,
+ "learning_rate": 1.839845378865716e-05,
+ "loss": 0.7616,
+ "step": 7603
+ },
+ {
+ "epoch": 20.719346049046322,
+ "grad_norm": 7.0249128341674805,
+ "learning_rate": 1.839797471518677e-05,
+ "loss": 0.8538,
+ "step": 7604
+ },
+ {
+ "epoch": 20.722070844686648,
+ "grad_norm": 8.416068077087402,
+ "learning_rate": 1.8397495576313226e-05,
+ "loss": 0.6364,
+ "step": 7605
+ },
+ {
+ "epoch": 20.724795640326974,
+ "grad_norm": 8.353381156921387,
+ "learning_rate": 1.8397016372040256e-05,
+ "loss": 0.5898,
+ "step": 7606
+ },
+ {
+ "epoch": 20.727520435967303,
+ "grad_norm": 6.08835506439209,
+ "learning_rate": 1.8396537102371593e-05,
+ "loss": 0.5312,
+ "step": 7607
+ },
+ {
+ "epoch": 20.73024523160763,
+ "grad_norm": 9.962550163269043,
+ "learning_rate": 1.839605776731097e-05,
+ "loss": 0.6469,
+ "step": 7608
+ },
+ {
+ "epoch": 20.732970027247955,
+ "grad_norm": 8.20128059387207,
+ "learning_rate": 1.839557836686212e-05,
+ "loss": 0.6311,
+ "step": 7609
+ },
+ {
+ "epoch": 20.735694822888284,
+ "grad_norm": 6.333535194396973,
+ "learning_rate": 1.8395098901028774e-05,
+ "loss": 0.5685,
+ "step": 7610
+ },
+ {
+ "epoch": 20.73841961852861,
+ "grad_norm": 7.139937877655029,
+ "learning_rate": 1.8394619369814666e-05,
+ "loss": 0.7208,
+ "step": 7611
+ },
+ {
+ "epoch": 20.741144414168936,
+ "grad_norm": 7.08389949798584,
+ "learning_rate": 1.8394139773223537e-05,
+ "loss": 0.659,
+ "step": 7612
+ },
+ {
+ "epoch": 20.743869209809265,
+ "grad_norm": 7.838805198669434,
+ "learning_rate": 1.8393660111259116e-05,
+ "loss": 0.6711,
+ "step": 7613
+ },
+ {
+ "epoch": 20.74659400544959,
+ "grad_norm": 7.755255222320557,
+ "learning_rate": 1.839318038392514e-05,
+ "loss": 0.7861,
+ "step": 7614
+ },
+ {
+ "epoch": 20.749318801089917,
+ "grad_norm": 6.045393466949463,
+ "learning_rate": 1.8392700591225346e-05,
+ "loss": 0.5959,
+ "step": 7615
+ },
+ {
+ "epoch": 20.752043596730246,
+ "grad_norm": 6.506926536560059,
+ "learning_rate": 1.8392220733163467e-05,
+ "loss": 0.5284,
+ "step": 7616
+ },
+ {
+ "epoch": 20.754768392370572,
+ "grad_norm": 5.4934773445129395,
+ "learning_rate": 1.8391740809743245e-05,
+ "loss": 0.6431,
+ "step": 7617
+ },
+ {
+ "epoch": 20.757493188010898,
+ "grad_norm": 8.959447860717773,
+ "learning_rate": 1.8391260820968417e-05,
+ "loss": 0.662,
+ "step": 7618
+ },
+ {
+ "epoch": 20.760217983651227,
+ "grad_norm": 6.669636249542236,
+ "learning_rate": 1.8390780766842717e-05,
+ "loss": 0.6351,
+ "step": 7619
+ },
+ {
+ "epoch": 20.762942779291553,
+ "grad_norm": 7.6507568359375,
+ "learning_rate": 1.839030064736989e-05,
+ "loss": 0.6418,
+ "step": 7620
+ },
+ {
+ "epoch": 20.76566757493188,
+ "grad_norm": 5.285655975341797,
+ "learning_rate": 1.838982046255367e-05,
+ "loss": 0.4709,
+ "step": 7621
+ },
+ {
+ "epoch": 20.768392370572208,
+ "grad_norm": 7.123846530914307,
+ "learning_rate": 1.8389340212397797e-05,
+ "loss": 0.512,
+ "step": 7622
+ },
+ {
+ "epoch": 20.771117166212534,
+ "grad_norm": 6.314147472381592,
+ "learning_rate": 1.8388859896906014e-05,
+ "loss": 0.5146,
+ "step": 7623
+ },
+ {
+ "epoch": 20.77384196185286,
+ "grad_norm": 5.560674667358398,
+ "learning_rate": 1.838837951608206e-05,
+ "loss": 0.6073,
+ "step": 7624
+ },
+ {
+ "epoch": 20.77656675749319,
+ "grad_norm": 7.63792610168457,
+ "learning_rate": 1.8387899069929673e-05,
+ "loss": 0.6653,
+ "step": 7625
+ },
+ {
+ "epoch": 20.779291553133515,
+ "grad_norm": 8.802462577819824,
+ "learning_rate": 1.8387418558452602e-05,
+ "loss": 0.887,
+ "step": 7626
+ },
+ {
+ "epoch": 20.78201634877384,
+ "grad_norm": 6.802992820739746,
+ "learning_rate": 1.8386937981654583e-05,
+ "loss": 0.5366,
+ "step": 7627
+ },
+ {
+ "epoch": 20.78474114441417,
+ "grad_norm": 7.283810615539551,
+ "learning_rate": 1.8386457339539364e-05,
+ "loss": 0.5237,
+ "step": 7628
+ },
+ {
+ "epoch": 20.787465940054496,
+ "grad_norm": 6.706197738647461,
+ "learning_rate": 1.8385976632110684e-05,
+ "loss": 0.5291,
+ "step": 7629
+ },
+ {
+ "epoch": 20.79019073569482,
+ "grad_norm": 6.273044586181641,
+ "learning_rate": 1.8385495859372288e-05,
+ "loss": 0.8496,
+ "step": 7630
+ },
+ {
+ "epoch": 20.79291553133515,
+ "grad_norm": 6.862782955169678,
+ "learning_rate": 1.8385015021327916e-05,
+ "loss": 0.7897,
+ "step": 7631
+ },
+ {
+ "epoch": 20.795640326975477,
+ "grad_norm": 6.6968793869018555,
+ "learning_rate": 1.838453411798132e-05,
+ "loss": 0.5916,
+ "step": 7632
+ },
+ {
+ "epoch": 20.798365122615802,
+ "grad_norm": 6.524605751037598,
+ "learning_rate": 1.8384053149336243e-05,
+ "loss": 0.4972,
+ "step": 7633
+ },
+ {
+ "epoch": 20.80108991825613,
+ "grad_norm": 6.316053867340088,
+ "learning_rate": 1.838357211539643e-05,
+ "loss": 0.6474,
+ "step": 7634
+ },
+ {
+ "epoch": 20.803814713896458,
+ "grad_norm": 6.819964408874512,
+ "learning_rate": 1.8383091016165626e-05,
+ "loss": 0.5823,
+ "step": 7635
+ },
+ {
+ "epoch": 20.806539509536783,
+ "grad_norm": 5.857781887054443,
+ "learning_rate": 1.838260985164758e-05,
+ "loss": 0.7029,
+ "step": 7636
+ },
+ {
+ "epoch": 20.809264305177113,
+ "grad_norm": 5.885871887207031,
+ "learning_rate": 1.8382128621846034e-05,
+ "loss": 0.8182,
+ "step": 7637
+ },
+ {
+ "epoch": 20.81198910081744,
+ "grad_norm": 5.709894180297852,
+ "learning_rate": 1.8381647326764744e-05,
+ "loss": 0.4644,
+ "step": 7638
+ },
+ {
+ "epoch": 20.814713896457764,
+ "grad_norm": 6.17929220199585,
+ "learning_rate": 1.8381165966407456e-05,
+ "loss": 0.5627,
+ "step": 7639
+ },
+ {
+ "epoch": 20.817438692098094,
+ "grad_norm": 5.6450419425964355,
+ "learning_rate": 1.8380684540777914e-05,
+ "loss": 0.4742,
+ "step": 7640
+ },
+ {
+ "epoch": 20.82016348773842,
+ "grad_norm": 6.4967360496521,
+ "learning_rate": 1.8380203049879872e-05,
+ "loss": 0.6175,
+ "step": 7641
+ },
+ {
+ "epoch": 20.822888283378745,
+ "grad_norm": 5.43927526473999,
+ "learning_rate": 1.8379721493717077e-05,
+ "loss": 0.5505,
+ "step": 7642
+ },
+ {
+ "epoch": 20.825613079019075,
+ "grad_norm": 6.280770301818848,
+ "learning_rate": 1.837923987229328e-05,
+ "loss": 0.6435,
+ "step": 7643
+ },
+ {
+ "epoch": 20.8283378746594,
+ "grad_norm": 6.45099401473999,
+ "learning_rate": 1.8378758185612237e-05,
+ "loss": 0.6295,
+ "step": 7644
+ },
+ {
+ "epoch": 20.831062670299726,
+ "grad_norm": 6.84363317489624,
+ "learning_rate": 1.8378276433677692e-05,
+ "loss": 0.5424,
+ "step": 7645
+ },
+ {
+ "epoch": 20.833787465940055,
+ "grad_norm": 6.212651252746582,
+ "learning_rate": 1.83777946164934e-05,
+ "loss": 0.5677,
+ "step": 7646
+ },
+ {
+ "epoch": 20.83651226158038,
+ "grad_norm": 5.381026744842529,
+ "learning_rate": 1.8377312734063113e-05,
+ "loss": 0.5352,
+ "step": 7647
+ },
+ {
+ "epoch": 20.839237057220707,
+ "grad_norm": 7.903063774108887,
+ "learning_rate": 1.8376830786390584e-05,
+ "loss": 0.5009,
+ "step": 7648
+ },
+ {
+ "epoch": 20.841961852861036,
+ "grad_norm": 5.986403942108154,
+ "learning_rate": 1.8376348773479568e-05,
+ "loss": 0.6136,
+ "step": 7649
+ },
+ {
+ "epoch": 20.844686648501362,
+ "grad_norm": 6.812338352203369,
+ "learning_rate": 1.8375866695333814e-05,
+ "loss": 0.7622,
+ "step": 7650
+ },
+ {
+ "epoch": 20.847411444141688,
+ "grad_norm": 5.377939224243164,
+ "learning_rate": 1.8375384551957084e-05,
+ "loss": 0.6833,
+ "step": 7651
+ },
+ {
+ "epoch": 20.850136239782017,
+ "grad_norm": 7.31725549697876,
+ "learning_rate": 1.8374902343353124e-05,
+ "loss": 0.4638,
+ "step": 7652
+ },
+ {
+ "epoch": 20.852861035422343,
+ "grad_norm": 5.683324813842773,
+ "learning_rate": 1.8374420069525697e-05,
+ "loss": 0.5302,
+ "step": 7653
+ },
+ {
+ "epoch": 20.85558583106267,
+ "grad_norm": 6.659883975982666,
+ "learning_rate": 1.8373937730478556e-05,
+ "loss": 0.5646,
+ "step": 7654
+ },
+ {
+ "epoch": 20.858310626703,
+ "grad_norm": 7.245208740234375,
+ "learning_rate": 1.8373455326215458e-05,
+ "loss": 0.6982,
+ "step": 7655
+ },
+ {
+ "epoch": 20.861035422343324,
+ "grad_norm": 6.123354434967041,
+ "learning_rate": 1.8372972856740157e-05,
+ "loss": 0.6162,
+ "step": 7656
+ },
+ {
+ "epoch": 20.86376021798365,
+ "grad_norm": 6.468822479248047,
+ "learning_rate": 1.8372490322056414e-05,
+ "loss": 0.6285,
+ "step": 7657
+ },
+ {
+ "epoch": 20.86648501362398,
+ "grad_norm": 6.18245267868042,
+ "learning_rate": 1.8372007722167987e-05,
+ "loss": 0.6196,
+ "step": 7658
+ },
+ {
+ "epoch": 20.869209809264305,
+ "grad_norm": 5.157792091369629,
+ "learning_rate": 1.8371525057078633e-05,
+ "loss": 0.5688,
+ "step": 7659
+ },
+ {
+ "epoch": 20.87193460490463,
+ "grad_norm": 8.294964790344238,
+ "learning_rate": 1.837104232679211e-05,
+ "loss": 0.6482,
+ "step": 7660
+ },
+ {
+ "epoch": 20.87465940054496,
+ "grad_norm": 5.91681432723999,
+ "learning_rate": 1.837055953131218e-05,
+ "loss": 0.7697,
+ "step": 7661
+ },
+ {
+ "epoch": 20.877384196185286,
+ "grad_norm": 5.8500213623046875,
+ "learning_rate": 1.83700766706426e-05,
+ "loss": 0.7121,
+ "step": 7662
+ },
+ {
+ "epoch": 20.88010899182561,
+ "grad_norm": 6.316312789916992,
+ "learning_rate": 1.8369593744787135e-05,
+ "loss": 0.5889,
+ "step": 7663
+ },
+ {
+ "epoch": 20.88283378746594,
+ "grad_norm": 7.16827917098999,
+ "learning_rate": 1.836911075374954e-05,
+ "loss": 0.5862,
+ "step": 7664
+ },
+ {
+ "epoch": 20.885558583106267,
+ "grad_norm": 6.436217308044434,
+ "learning_rate": 1.8368627697533583e-05,
+ "loss": 0.6395,
+ "step": 7665
+ },
+ {
+ "epoch": 20.888283378746593,
+ "grad_norm": 8.296398162841797,
+ "learning_rate": 1.836814457614302e-05,
+ "loss": 0.742,
+ "step": 7666
+ },
+ {
+ "epoch": 20.891008174386922,
+ "grad_norm": 7.2222161293029785,
+ "learning_rate": 1.836766138958162e-05,
+ "loss": 0.5807,
+ "step": 7667
+ },
+ {
+ "epoch": 20.893732970027248,
+ "grad_norm": 7.2359514236450195,
+ "learning_rate": 1.836717813785314e-05,
+ "loss": 0.5182,
+ "step": 7668
+ },
+ {
+ "epoch": 20.896457765667574,
+ "grad_norm": 6.271487712860107,
+ "learning_rate": 1.8366694820961346e-05,
+ "loss": 0.7699,
+ "step": 7669
+ },
+ {
+ "epoch": 20.899182561307903,
+ "grad_norm": 6.447906494140625,
+ "learning_rate": 1.8366211438910004e-05,
+ "loss": 0.6846,
+ "step": 7670
+ },
+ {
+ "epoch": 20.90190735694823,
+ "grad_norm": 6.554131984710693,
+ "learning_rate": 1.8365727991702874e-05,
+ "loss": 0.5198,
+ "step": 7671
+ },
+ {
+ "epoch": 20.904632152588555,
+ "grad_norm": 6.649665355682373,
+ "learning_rate": 1.8365244479343723e-05,
+ "loss": 0.6749,
+ "step": 7672
+ },
+ {
+ "epoch": 20.907356948228884,
+ "grad_norm": 5.913463115692139,
+ "learning_rate": 1.836476090183632e-05,
+ "loss": 0.6299,
+ "step": 7673
+ },
+ {
+ "epoch": 20.91008174386921,
+ "grad_norm": 6.969850063323975,
+ "learning_rate": 1.8364277259184428e-05,
+ "loss": 0.5596,
+ "step": 7674
+ },
+ {
+ "epoch": 20.912806539509535,
+ "grad_norm": 6.029188632965088,
+ "learning_rate": 1.8363793551391814e-05,
+ "loss": 0.6958,
+ "step": 7675
+ },
+ {
+ "epoch": 20.915531335149865,
+ "grad_norm": 5.917076110839844,
+ "learning_rate": 1.836330977846224e-05,
+ "loss": 0.4263,
+ "step": 7676
+ },
+ {
+ "epoch": 20.91825613079019,
+ "grad_norm": 6.526070594787598,
+ "learning_rate": 1.8362825940399483e-05,
+ "loss": 0.5039,
+ "step": 7677
+ },
+ {
+ "epoch": 20.920980926430516,
+ "grad_norm": 6.059049129486084,
+ "learning_rate": 1.8362342037207308e-05,
+ "loss": 0.7803,
+ "step": 7678
+ },
+ {
+ "epoch": 20.923705722070846,
+ "grad_norm": 8.233510971069336,
+ "learning_rate": 1.836185806888948e-05,
+ "loss": 0.5881,
+ "step": 7679
+ },
+ {
+ "epoch": 20.92643051771117,
+ "grad_norm": 6.215970993041992,
+ "learning_rate": 1.836137403544977e-05,
+ "loss": 0.6128,
+ "step": 7680
+ },
+ {
+ "epoch": 20.929155313351497,
+ "grad_norm": 7.390205383300781,
+ "learning_rate": 1.836088993689195e-05,
+ "loss": 0.4973,
+ "step": 7681
+ },
+ {
+ "epoch": 20.931880108991827,
+ "grad_norm": 6.155555725097656,
+ "learning_rate": 1.836040577321979e-05,
+ "loss": 0.6104,
+ "step": 7682
+ },
+ {
+ "epoch": 20.934604904632153,
+ "grad_norm": 6.417907238006592,
+ "learning_rate": 1.8359921544437055e-05,
+ "loss": 0.6124,
+ "step": 7683
+ },
+ {
+ "epoch": 20.93732970027248,
+ "grad_norm": 7.302303791046143,
+ "learning_rate": 1.835943725054752e-05,
+ "loss": 0.6134,
+ "step": 7684
+ },
+ {
+ "epoch": 20.940054495912808,
+ "grad_norm": 7.016286373138428,
+ "learning_rate": 1.835895289155496e-05,
+ "loss": 0.6446,
+ "step": 7685
+ },
+ {
+ "epoch": 20.942779291553133,
+ "grad_norm": 7.20880651473999,
+ "learning_rate": 1.835846846746314e-05,
+ "loss": 0.6113,
+ "step": 7686
+ },
+ {
+ "epoch": 20.94550408719346,
+ "grad_norm": 7.32924222946167,
+ "learning_rate": 1.835798397827584e-05,
+ "loss": 0.4851,
+ "step": 7687
+ },
+ {
+ "epoch": 20.94822888283379,
+ "grad_norm": 7.647403240203857,
+ "learning_rate": 1.835749942399683e-05,
+ "loss": 0.5799,
+ "step": 7688
+ },
+ {
+ "epoch": 20.950953678474114,
+ "grad_norm": 7.558221817016602,
+ "learning_rate": 1.835701480462988e-05,
+ "loss": 0.5892,
+ "step": 7689
+ },
+ {
+ "epoch": 20.95367847411444,
+ "grad_norm": 5.413008213043213,
+ "learning_rate": 1.8356530120178767e-05,
+ "loss": 0.6805,
+ "step": 7690
+ },
+ {
+ "epoch": 20.95640326975477,
+ "grad_norm": 6.129105567932129,
+ "learning_rate": 1.835604537064727e-05,
+ "loss": 0.687,
+ "step": 7691
+ },
+ {
+ "epoch": 20.959128065395095,
+ "grad_norm": 6.3839192390441895,
+ "learning_rate": 1.8355560556039164e-05,
+ "loss": 0.5175,
+ "step": 7692
+ },
+ {
+ "epoch": 20.96185286103542,
+ "grad_norm": 5.301498889923096,
+ "learning_rate": 1.8355075676358215e-05,
+ "loss": 0.7051,
+ "step": 7693
+ },
+ {
+ "epoch": 20.96457765667575,
+ "grad_norm": 6.997338771820068,
+ "learning_rate": 1.835459073160821e-05,
+ "loss": 0.5053,
+ "step": 7694
+ },
+ {
+ "epoch": 20.967302452316076,
+ "grad_norm": 6.255156993865967,
+ "learning_rate": 1.8354105721792917e-05,
+ "loss": 0.6621,
+ "step": 7695
+ },
+ {
+ "epoch": 20.970027247956402,
+ "grad_norm": 6.175135135650635,
+ "learning_rate": 1.8353620646916116e-05,
+ "loss": 0.6339,
+ "step": 7696
+ },
+ {
+ "epoch": 20.97275204359673,
+ "grad_norm": 6.051612854003906,
+ "learning_rate": 1.835313550698159e-05,
+ "loss": 0.6694,
+ "step": 7697
+ },
+ {
+ "epoch": 20.975476839237057,
+ "grad_norm": 6.311227321624756,
+ "learning_rate": 1.8352650301993112e-05,
+ "loss": 0.4738,
+ "step": 7698
+ },
+ {
+ "epoch": 20.978201634877383,
+ "grad_norm": 5.6289801597595215,
+ "learning_rate": 1.8352165031954468e-05,
+ "loss": 0.5416,
+ "step": 7699
+ },
+ {
+ "epoch": 20.980926430517712,
+ "grad_norm": 6.637048244476318,
+ "learning_rate": 1.8351679696869427e-05,
+ "loss": 0.6134,
+ "step": 7700
+ },
+ {
+ "epoch": 20.983651226158038,
+ "grad_norm": 6.527094841003418,
+ "learning_rate": 1.835119429674177e-05,
+ "loss": 0.5776,
+ "step": 7701
+ },
+ {
+ "epoch": 20.986376021798364,
+ "grad_norm": 7.028120994567871,
+ "learning_rate": 1.8350708831575286e-05,
+ "loss": 0.697,
+ "step": 7702
+ },
+ {
+ "epoch": 20.989100817438693,
+ "grad_norm": 7.419378757476807,
+ "learning_rate": 1.835022330137375e-05,
+ "loss": 0.5464,
+ "step": 7703
+ },
+ {
+ "epoch": 20.99182561307902,
+ "grad_norm": 7.397516250610352,
+ "learning_rate": 1.834973770614094e-05,
+ "loss": 0.8333,
+ "step": 7704
+ },
+ {
+ "epoch": 20.994550408719345,
+ "grad_norm": 5.747107028961182,
+ "learning_rate": 1.834925204588064e-05,
+ "loss": 0.4767,
+ "step": 7705
+ },
+ {
+ "epoch": 20.997275204359674,
+ "grad_norm": 6.229519367218018,
+ "learning_rate": 1.8348766320596638e-05,
+ "loss": 0.7128,
+ "step": 7706
+ },
+ {
+ "epoch": 21.0,
+ "grad_norm": 5.274659156799316,
+ "learning_rate": 1.8348280530292712e-05,
+ "loss": 0.5777,
+ "step": 7707
+ },
+ {
+ "epoch": 21.002724795640326,
+ "grad_norm": 7.5977911949157715,
+ "learning_rate": 1.834779467497265e-05,
+ "loss": 0.4589,
+ "step": 7708
+ },
+ {
+ "epoch": 21.005449591280655,
+ "grad_norm": 6.649643421173096,
+ "learning_rate": 1.834730875464022e-05,
+ "loss": 0.5156,
+ "step": 7709
+ },
+ {
+ "epoch": 21.00817438692098,
+ "grad_norm": 6.523000240325928,
+ "learning_rate": 1.834682276929923e-05,
+ "loss": 0.6686,
+ "step": 7710
+ },
+ {
+ "epoch": 21.010899182561307,
+ "grad_norm": 6.095635414123535,
+ "learning_rate": 1.8346336718953447e-05,
+ "loss": 0.4986,
+ "step": 7711
+ },
+ {
+ "epoch": 21.013623978201636,
+ "grad_norm": 5.78445291519165,
+ "learning_rate": 1.8345850603606666e-05,
+ "loss": 0.6182,
+ "step": 7712
+ },
+ {
+ "epoch": 21.016348773841962,
+ "grad_norm": 6.265995979309082,
+ "learning_rate": 1.8345364423262666e-05,
+ "loss": 0.4601,
+ "step": 7713
+ },
+ {
+ "epoch": 21.019073569482288,
+ "grad_norm": 5.090563774108887,
+ "learning_rate": 1.8344878177925238e-05,
+ "loss": 0.4989,
+ "step": 7714
+ },
+ {
+ "epoch": 21.021798365122617,
+ "grad_norm": 5.9531168937683105,
+ "learning_rate": 1.8344391867598164e-05,
+ "loss": 0.4763,
+ "step": 7715
+ },
+ {
+ "epoch": 21.024523160762943,
+ "grad_norm": 6.875079154968262,
+ "learning_rate": 1.8343905492285238e-05,
+ "loss": 0.5976,
+ "step": 7716
+ },
+ {
+ "epoch": 21.02724795640327,
+ "grad_norm": 6.213831901550293,
+ "learning_rate": 1.8343419051990245e-05,
+ "loss": 0.6141,
+ "step": 7717
+ },
+ {
+ "epoch": 21.029972752043598,
+ "grad_norm": 5.5872721672058105,
+ "learning_rate": 1.8342932546716973e-05,
+ "loss": 0.491,
+ "step": 7718
+ },
+ {
+ "epoch": 21.032697547683924,
+ "grad_norm": 4.839221477508545,
+ "learning_rate": 1.8342445976469206e-05,
+ "loss": 0.524,
+ "step": 7719
+ },
+ {
+ "epoch": 21.03542234332425,
+ "grad_norm": 5.3887810707092285,
+ "learning_rate": 1.8341959341250743e-05,
+ "loss": 0.4903,
+ "step": 7720
+ },
+ {
+ "epoch": 21.03814713896458,
+ "grad_norm": 6.382885456085205,
+ "learning_rate": 1.834147264106537e-05,
+ "loss": 0.4085,
+ "step": 7721
+ },
+ {
+ "epoch": 21.040871934604905,
+ "grad_norm": 5.466611385345459,
+ "learning_rate": 1.8340985875916874e-05,
+ "loss": 0.5096,
+ "step": 7722
+ },
+ {
+ "epoch": 21.04359673024523,
+ "grad_norm": 7.100013732910156,
+ "learning_rate": 1.834049904580905e-05,
+ "loss": 0.7031,
+ "step": 7723
+ },
+ {
+ "epoch": 21.04632152588556,
+ "grad_norm": 6.264891147613525,
+ "learning_rate": 1.8340012150745685e-05,
+ "loss": 0.4255,
+ "step": 7724
+ },
+ {
+ "epoch": 21.049046321525886,
+ "grad_norm": 7.548360347747803,
+ "learning_rate": 1.8339525190730577e-05,
+ "loss": 0.6315,
+ "step": 7725
+ },
+ {
+ "epoch": 21.05177111716621,
+ "grad_norm": 5.948115348815918,
+ "learning_rate": 1.833903816576751e-05,
+ "loss": 0.4656,
+ "step": 7726
+ },
+ {
+ "epoch": 21.05449591280654,
+ "grad_norm": 18.790987014770508,
+ "learning_rate": 1.8338551075860288e-05,
+ "loss": 0.483,
+ "step": 7727
+ },
+ {
+ "epoch": 21.057220708446867,
+ "grad_norm": 5.478869438171387,
+ "learning_rate": 1.83380639210127e-05,
+ "loss": 0.572,
+ "step": 7728
+ },
+ {
+ "epoch": 21.059945504087192,
+ "grad_norm": 5.297567844390869,
+ "learning_rate": 1.833757670122853e-05,
+ "loss": 0.4523,
+ "step": 7729
+ },
+ {
+ "epoch": 21.06267029972752,
+ "grad_norm": 6.019900321960449,
+ "learning_rate": 1.833708941651159e-05,
+ "loss": 0.6673,
+ "step": 7730
+ },
+ {
+ "epoch": 21.065395095367847,
+ "grad_norm": 7.422511100769043,
+ "learning_rate": 1.833660206686566e-05,
+ "loss": 0.6491,
+ "step": 7731
+ },
+ {
+ "epoch": 21.068119891008173,
+ "grad_norm": 5.882957935333252,
+ "learning_rate": 1.8336114652294547e-05,
+ "loss": 0.4778,
+ "step": 7732
+ },
+ {
+ "epoch": 21.070844686648503,
+ "grad_norm": 5.664695739746094,
+ "learning_rate": 1.8335627172802034e-05,
+ "loss": 0.5631,
+ "step": 7733
+ },
+ {
+ "epoch": 21.07356948228883,
+ "grad_norm": 5.96216344833374,
+ "learning_rate": 1.833513962839193e-05,
+ "loss": 0.4543,
+ "step": 7734
+ },
+ {
+ "epoch": 21.076294277929154,
+ "grad_norm": 7.169559478759766,
+ "learning_rate": 1.8334652019068024e-05,
+ "loss": 0.4842,
+ "step": 7735
+ },
+ {
+ "epoch": 21.079019073569484,
+ "grad_norm": 6.014753818511963,
+ "learning_rate": 1.833416434483412e-05,
+ "loss": 0.5577,
+ "step": 7736
+ },
+ {
+ "epoch": 21.08174386920981,
+ "grad_norm": 5.780186653137207,
+ "learning_rate": 1.833367660569401e-05,
+ "loss": 0.607,
+ "step": 7737
+ },
+ {
+ "epoch": 21.084468664850135,
+ "grad_norm": 5.843094348907471,
+ "learning_rate": 1.8333188801651497e-05,
+ "loss": 0.6078,
+ "step": 7738
+ },
+ {
+ "epoch": 21.087193460490465,
+ "grad_norm": 5.458644866943359,
+ "learning_rate": 1.8332700932710375e-05,
+ "loss": 0.4523,
+ "step": 7739
+ },
+ {
+ "epoch": 21.08991825613079,
+ "grad_norm": 9.096043586730957,
+ "learning_rate": 1.8332212998874445e-05,
+ "loss": 0.4999,
+ "step": 7740
+ },
+ {
+ "epoch": 21.092643051771116,
+ "grad_norm": 5.916035175323486,
+ "learning_rate": 1.833172500014751e-05,
+ "loss": 0.4811,
+ "step": 7741
+ },
+ {
+ "epoch": 21.095367847411445,
+ "grad_norm": 6.376270771026611,
+ "learning_rate": 1.8331236936533372e-05,
+ "loss": 0.4533,
+ "step": 7742
+ },
+ {
+ "epoch": 21.09809264305177,
+ "grad_norm": 6.347781181335449,
+ "learning_rate": 1.8330748808035827e-05,
+ "loss": 0.5187,
+ "step": 7743
+ },
+ {
+ "epoch": 21.100817438692097,
+ "grad_norm": 6.587207317352295,
+ "learning_rate": 1.8330260614658674e-05,
+ "loss": 0.5446,
+ "step": 7744
+ },
+ {
+ "epoch": 21.103542234332426,
+ "grad_norm": 5.754343509674072,
+ "learning_rate": 1.8329772356405723e-05,
+ "loss": 0.7166,
+ "step": 7745
+ },
+ {
+ "epoch": 21.106267029972752,
+ "grad_norm": 6.725513458251953,
+ "learning_rate": 1.8329284033280775e-05,
+ "loss": 0.3613,
+ "step": 7746
+ },
+ {
+ "epoch": 21.108991825613078,
+ "grad_norm": 7.814524173736572,
+ "learning_rate": 1.832879564528763e-05,
+ "loss": 0.6393,
+ "step": 7747
+ },
+ {
+ "epoch": 21.111716621253407,
+ "grad_norm": 6.372974872589111,
+ "learning_rate": 1.832830719243009e-05,
+ "loss": 0.5073,
+ "step": 7748
+ },
+ {
+ "epoch": 21.114441416893733,
+ "grad_norm": 6.214714050292969,
+ "learning_rate": 1.8327818674711965e-05,
+ "loss": 0.4735,
+ "step": 7749
+ },
+ {
+ "epoch": 21.11716621253406,
+ "grad_norm": 7.009729862213135,
+ "learning_rate": 1.832733009213705e-05,
+ "loss": 0.5699,
+ "step": 7750
+ },
+ {
+ "epoch": 21.11989100817439,
+ "grad_norm": 6.717540740966797,
+ "learning_rate": 1.832684144470916e-05,
+ "loss": 0.6153,
+ "step": 7751
+ },
+ {
+ "epoch": 21.122615803814714,
+ "grad_norm": 4.514061450958252,
+ "learning_rate": 1.8326352732432103e-05,
+ "loss": 0.4754,
+ "step": 7752
+ },
+ {
+ "epoch": 21.12534059945504,
+ "grad_norm": 5.914026260375977,
+ "learning_rate": 1.8325863955309673e-05,
+ "loss": 0.5533,
+ "step": 7753
+ },
+ {
+ "epoch": 21.12806539509537,
+ "grad_norm": 5.987513065338135,
+ "learning_rate": 1.8325375113345684e-05,
+ "loss": 0.5094,
+ "step": 7754
+ },
+ {
+ "epoch": 21.130790190735695,
+ "grad_norm": 8.338969230651855,
+ "learning_rate": 1.832488620654394e-05,
+ "loss": 0.5895,
+ "step": 7755
+ },
+ {
+ "epoch": 21.13351498637602,
+ "grad_norm": 6.171380996704102,
+ "learning_rate": 1.8324397234908252e-05,
+ "loss": 0.6176,
+ "step": 7756
+ },
+ {
+ "epoch": 21.13623978201635,
+ "grad_norm": 7.388641357421875,
+ "learning_rate": 1.8323908198442423e-05,
+ "loss": 0.5243,
+ "step": 7757
+ },
+ {
+ "epoch": 21.138964577656676,
+ "grad_norm": 6.47926664352417,
+ "learning_rate": 1.8323419097150264e-05,
+ "loss": 0.6151,
+ "step": 7758
+ },
+ {
+ "epoch": 21.141689373297,
+ "grad_norm": 6.845445156097412,
+ "learning_rate": 1.832292993103559e-05,
+ "loss": 0.5165,
+ "step": 7759
+ },
+ {
+ "epoch": 21.14441416893733,
+ "grad_norm": 5.223355770111084,
+ "learning_rate": 1.8322440700102203e-05,
+ "loss": 0.6291,
+ "step": 7760
+ },
+ {
+ "epoch": 21.147138964577657,
+ "grad_norm": 7.6689453125,
+ "learning_rate": 1.8321951404353916e-05,
+ "loss": 0.4983,
+ "step": 7761
+ },
+ {
+ "epoch": 21.149863760217983,
+ "grad_norm": 5.808589935302734,
+ "learning_rate": 1.8321462043794538e-05,
+ "loss": 0.8104,
+ "step": 7762
+ },
+ {
+ "epoch": 21.152588555858312,
+ "grad_norm": 5.829456329345703,
+ "learning_rate": 1.8320972618427885e-05,
+ "loss": 0.5044,
+ "step": 7763
+ },
+ {
+ "epoch": 21.155313351498638,
+ "grad_norm": 6.281116485595703,
+ "learning_rate": 1.832048312825776e-05,
+ "loss": 0.6157,
+ "step": 7764
+ },
+ {
+ "epoch": 21.158038147138964,
+ "grad_norm": 6.115004539489746,
+ "learning_rate": 1.8319993573287984e-05,
+ "loss": 0.7912,
+ "step": 7765
+ },
+ {
+ "epoch": 21.160762942779293,
+ "grad_norm": 8.314201354980469,
+ "learning_rate": 1.8319503953522368e-05,
+ "loss": 0.3992,
+ "step": 7766
+ },
+ {
+ "epoch": 21.16348773841962,
+ "grad_norm": 5.1933064460754395,
+ "learning_rate": 1.8319014268964718e-05,
+ "loss": 0.468,
+ "step": 7767
+ },
+ {
+ "epoch": 21.166212534059945,
+ "grad_norm": 5.636848449707031,
+ "learning_rate": 1.8318524519618853e-05,
+ "loss": 0.4379,
+ "step": 7768
+ },
+ {
+ "epoch": 21.168937329700274,
+ "grad_norm": 14.576801300048828,
+ "learning_rate": 1.831803470548859e-05,
+ "loss": 0.4233,
+ "step": 7769
+ },
+ {
+ "epoch": 21.1716621253406,
+ "grad_norm": 5.899982452392578,
+ "learning_rate": 1.8317544826577738e-05,
+ "loss": 0.6118,
+ "step": 7770
+ },
+ {
+ "epoch": 21.174386920980925,
+ "grad_norm": 6.482058048248291,
+ "learning_rate": 1.8317054882890114e-05,
+ "loss": 0.5085,
+ "step": 7771
+ },
+ {
+ "epoch": 21.177111716621255,
+ "grad_norm": 5.534772872924805,
+ "learning_rate": 1.8316564874429535e-05,
+ "loss": 0.5203,
+ "step": 7772
+ },
+ {
+ "epoch": 21.17983651226158,
+ "grad_norm": 5.479641914367676,
+ "learning_rate": 1.831607480119982e-05,
+ "loss": 0.4387,
+ "step": 7773
+ },
+ {
+ "epoch": 21.182561307901906,
+ "grad_norm": 6.422123432159424,
+ "learning_rate": 1.8315584663204776e-05,
+ "loss": 0.5268,
+ "step": 7774
+ },
+ {
+ "epoch": 21.185286103542236,
+ "grad_norm": 7.454172134399414,
+ "learning_rate": 1.8315094460448232e-05,
+ "loss": 0.5029,
+ "step": 7775
+ },
+ {
+ "epoch": 21.18801089918256,
+ "grad_norm": 5.704333305358887,
+ "learning_rate": 1.8314604192933997e-05,
+ "loss": 0.4736,
+ "step": 7776
+ },
+ {
+ "epoch": 21.190735694822887,
+ "grad_norm": 6.720086574554443,
+ "learning_rate": 1.8314113860665892e-05,
+ "loss": 0.6872,
+ "step": 7777
+ },
+ {
+ "epoch": 21.193460490463217,
+ "grad_norm": 6.68890380859375,
+ "learning_rate": 1.8313623463647736e-05,
+ "loss": 0.5096,
+ "step": 7778
+ },
+ {
+ "epoch": 21.196185286103542,
+ "grad_norm": 6.201164722442627,
+ "learning_rate": 1.831313300188335e-05,
+ "loss": 0.5785,
+ "step": 7779
+ },
+ {
+ "epoch": 21.19891008174387,
+ "grad_norm": 6.292909622192383,
+ "learning_rate": 1.8312642475376547e-05,
+ "loss": 0.5184,
+ "step": 7780
+ },
+ {
+ "epoch": 21.201634877384198,
+ "grad_norm": 6.464934825897217,
+ "learning_rate": 1.8312151884131157e-05,
+ "loss": 0.6072,
+ "step": 7781
+ },
+ {
+ "epoch": 21.204359673024523,
+ "grad_norm": 5.851186752319336,
+ "learning_rate": 1.8311661228150994e-05,
+ "loss": 0.3823,
+ "step": 7782
+ },
+ {
+ "epoch": 21.20708446866485,
+ "grad_norm": 5.6403069496154785,
+ "learning_rate": 1.8311170507439878e-05,
+ "loss": 0.5648,
+ "step": 7783
+ },
+ {
+ "epoch": 21.20980926430518,
+ "grad_norm": 5.377665042877197,
+ "learning_rate": 1.8310679722001638e-05,
+ "loss": 0.4908,
+ "step": 7784
+ },
+ {
+ "epoch": 21.212534059945504,
+ "grad_norm": 6.480130195617676,
+ "learning_rate": 1.831018887184009e-05,
+ "loss": 0.6585,
+ "step": 7785
+ },
+ {
+ "epoch": 21.21525885558583,
+ "grad_norm": 5.4656548500061035,
+ "learning_rate": 1.8309697956959057e-05,
+ "loss": 0.4617,
+ "step": 7786
+ },
+ {
+ "epoch": 21.21798365122616,
+ "grad_norm": 6.185543537139893,
+ "learning_rate": 1.8309206977362367e-05,
+ "loss": 0.7892,
+ "step": 7787
+ },
+ {
+ "epoch": 21.220708446866485,
+ "grad_norm": 6.2226033210754395,
+ "learning_rate": 1.8308715933053836e-05,
+ "loss": 0.3924,
+ "step": 7788
+ },
+ {
+ "epoch": 21.22343324250681,
+ "grad_norm": 6.522709846496582,
+ "learning_rate": 1.8308224824037295e-05,
+ "loss": 0.5439,
+ "step": 7789
+ },
+ {
+ "epoch": 21.22615803814714,
+ "grad_norm": 7.3227338790893555,
+ "learning_rate": 1.8307733650316567e-05,
+ "loss": 0.4915,
+ "step": 7790
+ },
+ {
+ "epoch": 21.228882833787466,
+ "grad_norm": 5.745213508605957,
+ "learning_rate": 1.8307242411895478e-05,
+ "loss": 0.7288,
+ "step": 7791
+ },
+ {
+ "epoch": 21.231607629427792,
+ "grad_norm": 5.827697277069092,
+ "learning_rate": 1.8306751108777853e-05,
+ "loss": 0.4708,
+ "step": 7792
+ },
+ {
+ "epoch": 21.23433242506812,
+ "grad_norm": 8.773041725158691,
+ "learning_rate": 1.8306259740967515e-05,
+ "loss": 0.4817,
+ "step": 7793
+ },
+ {
+ "epoch": 21.237057220708447,
+ "grad_norm": 7.096762180328369,
+ "learning_rate": 1.8305768308468294e-05,
+ "loss": 0.5035,
+ "step": 7794
+ },
+ {
+ "epoch": 21.239782016348773,
+ "grad_norm": 6.2928547859191895,
+ "learning_rate": 1.8305276811284013e-05,
+ "loss": 0.4556,
+ "step": 7795
+ },
+ {
+ "epoch": 21.242506811989102,
+ "grad_norm": 5.519431114196777,
+ "learning_rate": 1.8304785249418508e-05,
+ "loss": 0.5415,
+ "step": 7796
+ },
+ {
+ "epoch": 21.245231607629428,
+ "grad_norm": 5.863287925720215,
+ "learning_rate": 1.8304293622875603e-05,
+ "loss": 0.4595,
+ "step": 7797
+ },
+ {
+ "epoch": 21.247956403269754,
+ "grad_norm": 7.069830894470215,
+ "learning_rate": 1.8303801931659127e-05,
+ "loss": 0.7639,
+ "step": 7798
+ },
+ {
+ "epoch": 21.250681198910083,
+ "grad_norm": 6.369099140167236,
+ "learning_rate": 1.8303310175772905e-05,
+ "loss": 0.5285,
+ "step": 7799
+ },
+ {
+ "epoch": 21.25340599455041,
+ "grad_norm": 6.115304470062256,
+ "learning_rate": 1.8302818355220776e-05,
+ "loss": 0.441,
+ "step": 7800
+ },
+ {
+ "epoch": 21.256130790190735,
+ "grad_norm": 5.4862470626831055,
+ "learning_rate": 1.8302326470006562e-05,
+ "loss": 0.6801,
+ "step": 7801
+ },
+ {
+ "epoch": 21.258855585831064,
+ "grad_norm": 4.712049961090088,
+ "learning_rate": 1.8301834520134098e-05,
+ "loss": 0.5965,
+ "step": 7802
+ },
+ {
+ "epoch": 21.26158038147139,
+ "grad_norm": 5.418254375457764,
+ "learning_rate": 1.830134250560721e-05,
+ "loss": 0.4472,
+ "step": 7803
+ },
+ {
+ "epoch": 21.264305177111716,
+ "grad_norm": 5.803359508514404,
+ "learning_rate": 1.8300850426429738e-05,
+ "loss": 0.5047,
+ "step": 7804
+ },
+ {
+ "epoch": 21.267029972752045,
+ "grad_norm": 4.8905205726623535,
+ "learning_rate": 1.830035828260551e-05,
+ "loss": 0.3875,
+ "step": 7805
+ },
+ {
+ "epoch": 21.26975476839237,
+ "grad_norm": 5.2387189865112305,
+ "learning_rate": 1.8299866074138356e-05,
+ "loss": 0.4391,
+ "step": 7806
+ },
+ {
+ "epoch": 21.272479564032697,
+ "grad_norm": 5.852396488189697,
+ "learning_rate": 1.829937380103211e-05,
+ "loss": 0.463,
+ "step": 7807
+ },
+ {
+ "epoch": 21.275204359673026,
+ "grad_norm": 5.72178316116333,
+ "learning_rate": 1.8298881463290613e-05,
+ "loss": 0.6542,
+ "step": 7808
+ },
+ {
+ "epoch": 21.277929155313352,
+ "grad_norm": 5.515141487121582,
+ "learning_rate": 1.8298389060917696e-05,
+ "loss": 0.4547,
+ "step": 7809
+ },
+ {
+ "epoch": 21.280653950953678,
+ "grad_norm": 5.703946590423584,
+ "learning_rate": 1.8297896593917187e-05,
+ "loss": 0.5289,
+ "step": 7810
+ },
+ {
+ "epoch": 21.283378746594007,
+ "grad_norm": 5.898927211761475,
+ "learning_rate": 1.8297404062292927e-05,
+ "loss": 0.4128,
+ "step": 7811
+ },
+ {
+ "epoch": 21.286103542234333,
+ "grad_norm": 6.135253429412842,
+ "learning_rate": 1.8296911466048754e-05,
+ "loss": 0.5403,
+ "step": 7812
+ },
+ {
+ "epoch": 21.28882833787466,
+ "grad_norm": 4.973138809204102,
+ "learning_rate": 1.82964188051885e-05,
+ "loss": 0.5304,
+ "step": 7813
+ },
+ {
+ "epoch": 21.291553133514988,
+ "grad_norm": 6.326437950134277,
+ "learning_rate": 1.8295926079716002e-05,
+ "loss": 0.5547,
+ "step": 7814
+ },
+ {
+ "epoch": 21.294277929155314,
+ "grad_norm": 6.260693073272705,
+ "learning_rate": 1.82954332896351e-05,
+ "loss": 0.5434,
+ "step": 7815
+ },
+ {
+ "epoch": 21.29700272479564,
+ "grad_norm": 6.2184624671936035,
+ "learning_rate": 1.829494043494963e-05,
+ "loss": 0.5605,
+ "step": 7816
+ },
+ {
+ "epoch": 21.29972752043597,
+ "grad_norm": 7.582458019256592,
+ "learning_rate": 1.829444751566343e-05,
+ "loss": 0.5309,
+ "step": 7817
+ },
+ {
+ "epoch": 21.302452316076295,
+ "grad_norm": 6.090841293334961,
+ "learning_rate": 1.829395453178034e-05,
+ "loss": 0.455,
+ "step": 7818
+ },
+ {
+ "epoch": 21.30517711171662,
+ "grad_norm": 7.296097755432129,
+ "learning_rate": 1.8293461483304203e-05,
+ "loss": 0.6533,
+ "step": 7819
+ },
+ {
+ "epoch": 21.30790190735695,
+ "grad_norm": 6.912149429321289,
+ "learning_rate": 1.8292968370238852e-05,
+ "loss": 0.4343,
+ "step": 7820
+ },
+ {
+ "epoch": 21.310626702997276,
+ "grad_norm": 5.292464256286621,
+ "learning_rate": 1.829247519258813e-05,
+ "loss": 0.5023,
+ "step": 7821
+ },
+ {
+ "epoch": 21.3133514986376,
+ "grad_norm": 5.647106170654297,
+ "learning_rate": 1.8291981950355878e-05,
+ "loss": 0.556,
+ "step": 7822
+ },
+ {
+ "epoch": 21.31607629427793,
+ "grad_norm": 6.542977333068848,
+ "learning_rate": 1.829148864354594e-05,
+ "loss": 0.4246,
+ "step": 7823
+ },
+ {
+ "epoch": 21.318801089918257,
+ "grad_norm": 6.5619072914123535,
+ "learning_rate": 1.829099527216215e-05,
+ "loss": 0.4771,
+ "step": 7824
+ },
+ {
+ "epoch": 21.321525885558582,
+ "grad_norm": 6.477159023284912,
+ "learning_rate": 1.8290501836208357e-05,
+ "loss": 0.5739,
+ "step": 7825
+ },
+ {
+ "epoch": 21.32425068119891,
+ "grad_norm": 6.243288516998291,
+ "learning_rate": 1.8290008335688404e-05,
+ "loss": 0.5125,
+ "step": 7826
+ },
+ {
+ "epoch": 21.326975476839237,
+ "grad_norm": 5.888049125671387,
+ "learning_rate": 1.8289514770606133e-05,
+ "loss": 0.5372,
+ "step": 7827
+ },
+ {
+ "epoch": 21.329700272479563,
+ "grad_norm": 5.28791618347168,
+ "learning_rate": 1.828902114096539e-05,
+ "loss": 0.5761,
+ "step": 7828
+ },
+ {
+ "epoch": 21.332425068119893,
+ "grad_norm": 6.69451379776001,
+ "learning_rate": 1.8288527446770015e-05,
+ "loss": 0.4189,
+ "step": 7829
+ },
+ {
+ "epoch": 21.33514986376022,
+ "grad_norm": 4.869025230407715,
+ "learning_rate": 1.8288033688023855e-05,
+ "loss": 0.4818,
+ "step": 7830
+ },
+ {
+ "epoch": 21.337874659400544,
+ "grad_norm": 6.417300224304199,
+ "learning_rate": 1.8287539864730756e-05,
+ "loss": 0.4928,
+ "step": 7831
+ },
+ {
+ "epoch": 21.340599455040874,
+ "grad_norm": 6.442546844482422,
+ "learning_rate": 1.8287045976894563e-05,
+ "loss": 0.5411,
+ "step": 7832
+ },
+ {
+ "epoch": 21.3433242506812,
+ "grad_norm": 5.993861675262451,
+ "learning_rate": 1.828655202451912e-05,
+ "loss": 0.5687,
+ "step": 7833
+ },
+ {
+ "epoch": 21.346049046321525,
+ "grad_norm": 5.399357795715332,
+ "learning_rate": 1.828605800760828e-05,
+ "loss": 0.4589,
+ "step": 7834
+ },
+ {
+ "epoch": 21.348773841961854,
+ "grad_norm": 7.255837440490723,
+ "learning_rate": 1.8285563926165884e-05,
+ "loss": 0.6049,
+ "step": 7835
+ },
+ {
+ "epoch": 21.35149863760218,
+ "grad_norm": 5.856992244720459,
+ "learning_rate": 1.8285069780195787e-05,
+ "loss": 0.5095,
+ "step": 7836
+ },
+ {
+ "epoch": 21.354223433242506,
+ "grad_norm": 6.892065048217773,
+ "learning_rate": 1.828457556970183e-05,
+ "loss": 0.6315,
+ "step": 7837
+ },
+ {
+ "epoch": 21.356948228882835,
+ "grad_norm": 6.408921718597412,
+ "learning_rate": 1.8284081294687866e-05,
+ "loss": 0.5124,
+ "step": 7838
+ },
+ {
+ "epoch": 21.35967302452316,
+ "grad_norm": 5.412475109100342,
+ "learning_rate": 1.8283586955157744e-05,
+ "loss": 0.4677,
+ "step": 7839
+ },
+ {
+ "epoch": 21.362397820163487,
+ "grad_norm": 6.798328876495361,
+ "learning_rate": 1.8283092551115312e-05,
+ "loss": 0.4752,
+ "step": 7840
+ },
+ {
+ "epoch": 21.365122615803816,
+ "grad_norm": 7.840207099914551,
+ "learning_rate": 1.8282598082564423e-05,
+ "loss": 0.516,
+ "step": 7841
+ },
+ {
+ "epoch": 21.367847411444142,
+ "grad_norm": 5.971236228942871,
+ "learning_rate": 1.828210354950893e-05,
+ "loss": 0.4903,
+ "step": 7842
+ },
+ {
+ "epoch": 21.370572207084468,
+ "grad_norm": 5.827205181121826,
+ "learning_rate": 1.8281608951952675e-05,
+ "loss": 0.5263,
+ "step": 7843
+ },
+ {
+ "epoch": 21.373297002724797,
+ "grad_norm": 5.700268745422363,
+ "learning_rate": 1.8281114289899515e-05,
+ "loss": 0.6471,
+ "step": 7844
+ },
+ {
+ "epoch": 21.376021798365123,
+ "grad_norm": 5.5344133377075195,
+ "learning_rate": 1.8280619563353307e-05,
+ "loss": 0.5042,
+ "step": 7845
+ },
+ {
+ "epoch": 21.37874659400545,
+ "grad_norm": 6.332845211029053,
+ "learning_rate": 1.8280124772317903e-05,
+ "loss": 0.5621,
+ "step": 7846
+ },
+ {
+ "epoch": 21.381471389645778,
+ "grad_norm": 6.224171161651611,
+ "learning_rate": 1.827962991679715e-05,
+ "loss": 0.6599,
+ "step": 7847
+ },
+ {
+ "epoch": 21.384196185286104,
+ "grad_norm": 6.13968563079834,
+ "learning_rate": 1.8279134996794904e-05,
+ "loss": 0.4896,
+ "step": 7848
+ },
+ {
+ "epoch": 21.38692098092643,
+ "grad_norm": 6.897185325622559,
+ "learning_rate": 1.8278640012315026e-05,
+ "loss": 0.6511,
+ "step": 7849
+ },
+ {
+ "epoch": 21.38964577656676,
+ "grad_norm": 5.575146675109863,
+ "learning_rate": 1.8278144963361363e-05,
+ "loss": 0.7006,
+ "step": 7850
+ },
+ {
+ "epoch": 21.392370572207085,
+ "grad_norm": 5.910147190093994,
+ "learning_rate": 1.8277649849937776e-05,
+ "loss": 0.4107,
+ "step": 7851
+ },
+ {
+ "epoch": 21.39509536784741,
+ "grad_norm": 7.528545379638672,
+ "learning_rate": 1.8277154672048117e-05,
+ "loss": 0.6575,
+ "step": 7852
+ },
+ {
+ "epoch": 21.39782016348774,
+ "grad_norm": 5.373162269592285,
+ "learning_rate": 1.827665942969624e-05,
+ "loss": 0.6416,
+ "step": 7853
+ },
+ {
+ "epoch": 21.400544959128066,
+ "grad_norm": 5.650115489959717,
+ "learning_rate": 1.827616412288601e-05,
+ "loss": 0.5191,
+ "step": 7854
+ },
+ {
+ "epoch": 21.40326975476839,
+ "grad_norm": 5.738048076629639,
+ "learning_rate": 1.827566875162128e-05,
+ "loss": 0.5342,
+ "step": 7855
+ },
+ {
+ "epoch": 21.40599455040872,
+ "grad_norm": 6.33491849899292,
+ "learning_rate": 1.827517331590591e-05,
+ "loss": 0.4532,
+ "step": 7856
+ },
+ {
+ "epoch": 21.408719346049047,
+ "grad_norm": 7.074545860290527,
+ "learning_rate": 1.827467781574375e-05,
+ "loss": 0.5522,
+ "step": 7857
+ },
+ {
+ "epoch": 21.411444141689373,
+ "grad_norm": 5.793247222900391,
+ "learning_rate": 1.827418225113867e-05,
+ "loss": 0.5139,
+ "step": 7858
+ },
+ {
+ "epoch": 21.414168937329702,
+ "grad_norm": 5.254024982452393,
+ "learning_rate": 1.8273686622094528e-05,
+ "loss": 0.734,
+ "step": 7859
+ },
+ {
+ "epoch": 21.416893732970028,
+ "grad_norm": 5.920283317565918,
+ "learning_rate": 1.8273190928615175e-05,
+ "loss": 0.5147,
+ "step": 7860
+ },
+ {
+ "epoch": 21.419618528610354,
+ "grad_norm": 6.276407718658447,
+ "learning_rate": 1.827269517070448e-05,
+ "loss": 0.5869,
+ "step": 7861
+ },
+ {
+ "epoch": 21.422343324250683,
+ "grad_norm": 5.496118545532227,
+ "learning_rate": 1.82721993483663e-05,
+ "loss": 0.5742,
+ "step": 7862
+ },
+ {
+ "epoch": 21.42506811989101,
+ "grad_norm": 5.566563606262207,
+ "learning_rate": 1.82717034616045e-05,
+ "loss": 0.7227,
+ "step": 7863
+ },
+ {
+ "epoch": 21.427792915531334,
+ "grad_norm": 6.3988823890686035,
+ "learning_rate": 1.8271207510422936e-05,
+ "loss": 0.4668,
+ "step": 7864
+ },
+ {
+ "epoch": 21.430517711171664,
+ "grad_norm": 6.209607124328613,
+ "learning_rate": 1.827071149482548e-05,
+ "loss": 0.6602,
+ "step": 7865
+ },
+ {
+ "epoch": 21.43324250681199,
+ "grad_norm": 5.166659355163574,
+ "learning_rate": 1.8270215414815983e-05,
+ "loss": 0.6253,
+ "step": 7866
+ },
+ {
+ "epoch": 21.435967302452315,
+ "grad_norm": 6.424147605895996,
+ "learning_rate": 1.826971927039832e-05,
+ "loss": 0.5088,
+ "step": 7867
+ },
+ {
+ "epoch": 21.438692098092645,
+ "grad_norm": 4.956645488739014,
+ "learning_rate": 1.8269223061576345e-05,
+ "loss": 0.3929,
+ "step": 7868
+ },
+ {
+ "epoch": 21.44141689373297,
+ "grad_norm": 5.738945484161377,
+ "learning_rate": 1.8268726788353928e-05,
+ "loss": 0.4877,
+ "step": 7869
+ },
+ {
+ "epoch": 21.444141689373296,
+ "grad_norm": 6.023827075958252,
+ "learning_rate": 1.8268230450734933e-05,
+ "loss": 0.541,
+ "step": 7870
+ },
+ {
+ "epoch": 21.446866485013626,
+ "grad_norm": 5.881688117980957,
+ "learning_rate": 1.8267734048723227e-05,
+ "loss": 0.5731,
+ "step": 7871
+ },
+ {
+ "epoch": 21.44959128065395,
+ "grad_norm": 6.343794822692871,
+ "learning_rate": 1.826723758232267e-05,
+ "loss": 0.5553,
+ "step": 7872
+ },
+ {
+ "epoch": 21.452316076294277,
+ "grad_norm": 6.356383800506592,
+ "learning_rate": 1.8266741051537138e-05,
+ "loss": 0.6263,
+ "step": 7873
+ },
+ {
+ "epoch": 21.455040871934607,
+ "grad_norm": 6.712540149688721,
+ "learning_rate": 1.826624445637049e-05,
+ "loss": 0.5055,
+ "step": 7874
+ },
+ {
+ "epoch": 21.457765667574932,
+ "grad_norm": 5.928344249725342,
+ "learning_rate": 1.82657477968266e-05,
+ "loss": 0.6287,
+ "step": 7875
+ },
+ {
+ "epoch": 21.460490463215258,
+ "grad_norm": 5.936520576477051,
+ "learning_rate": 1.8265251072909325e-05,
+ "loss": 0.5344,
+ "step": 7876
+ },
+ {
+ "epoch": 21.463215258855588,
+ "grad_norm": 5.9764604568481445,
+ "learning_rate": 1.8264754284622544e-05,
+ "loss": 0.5848,
+ "step": 7877
+ },
+ {
+ "epoch": 21.465940054495913,
+ "grad_norm": 5.367731094360352,
+ "learning_rate": 1.8264257431970126e-05,
+ "loss": 0.4961,
+ "step": 7878
+ },
+ {
+ "epoch": 21.46866485013624,
+ "grad_norm": 6.694281101226807,
+ "learning_rate": 1.8263760514955934e-05,
+ "loss": 0.4861,
+ "step": 7879
+ },
+ {
+ "epoch": 21.47138964577657,
+ "grad_norm": 7.277377605438232,
+ "learning_rate": 1.826326353358384e-05,
+ "loss": 0.4783,
+ "step": 7880
+ },
+ {
+ "epoch": 21.474114441416894,
+ "grad_norm": 5.65437650680542,
+ "learning_rate": 1.826276648785772e-05,
+ "loss": 0.5977,
+ "step": 7881
+ },
+ {
+ "epoch": 21.47683923705722,
+ "grad_norm": 5.217709541320801,
+ "learning_rate": 1.8262269377781433e-05,
+ "loss": 0.5851,
+ "step": 7882
+ },
+ {
+ "epoch": 21.479564032697546,
+ "grad_norm": 4.7698187828063965,
+ "learning_rate": 1.8261772203358864e-05,
+ "loss": 0.5237,
+ "step": 7883
+ },
+ {
+ "epoch": 21.482288828337875,
+ "grad_norm": 6.1915507316589355,
+ "learning_rate": 1.826127496459388e-05,
+ "loss": 0.7609,
+ "step": 7884
+ },
+ {
+ "epoch": 21.4850136239782,
+ "grad_norm": 5.520107746124268,
+ "learning_rate": 1.8260777661490346e-05,
+ "loss": 0.6243,
+ "step": 7885
+ },
+ {
+ "epoch": 21.48773841961853,
+ "grad_norm": 5.61135196685791,
+ "learning_rate": 1.8260280294052147e-05,
+ "loss": 0.4222,
+ "step": 7886
+ },
+ {
+ "epoch": 21.490463215258856,
+ "grad_norm": 5.8247480392456055,
+ "learning_rate": 1.825978286228315e-05,
+ "loss": 0.4221,
+ "step": 7887
+ },
+ {
+ "epoch": 21.493188010899182,
+ "grad_norm": 6.079692363739014,
+ "learning_rate": 1.825928536618723e-05,
+ "loss": 0.6467,
+ "step": 7888
+ },
+ {
+ "epoch": 21.495912806539508,
+ "grad_norm": 5.328104496002197,
+ "learning_rate": 1.825878780576826e-05,
+ "loss": 0.5922,
+ "step": 7889
+ },
+ {
+ "epoch": 21.498637602179837,
+ "grad_norm": 6.257778644561768,
+ "learning_rate": 1.8258290181030118e-05,
+ "loss": 0.5129,
+ "step": 7890
+ },
+ {
+ "epoch": 21.501362397820163,
+ "grad_norm": 6.034976959228516,
+ "learning_rate": 1.8257792491976676e-05,
+ "loss": 0.4925,
+ "step": 7891
+ },
+ {
+ "epoch": 21.504087193460492,
+ "grad_norm": 5.124576091766357,
+ "learning_rate": 1.8257294738611814e-05,
+ "loss": 0.4985,
+ "step": 7892
+ },
+ {
+ "epoch": 21.506811989100818,
+ "grad_norm": 5.940557479858398,
+ "learning_rate": 1.8256796920939405e-05,
+ "loss": 0.4103,
+ "step": 7893
+ },
+ {
+ "epoch": 21.509536784741144,
+ "grad_norm": 5.921528339385986,
+ "learning_rate": 1.8256299038963326e-05,
+ "loss": 0.5347,
+ "step": 7894
+ },
+ {
+ "epoch": 21.51226158038147,
+ "grad_norm": 7.089655876159668,
+ "learning_rate": 1.8255801092687456e-05,
+ "loss": 0.4396,
+ "step": 7895
+ },
+ {
+ "epoch": 21.5149863760218,
+ "grad_norm": 6.032310485839844,
+ "learning_rate": 1.8255303082115676e-05,
+ "loss": 0.4786,
+ "step": 7896
+ },
+ {
+ "epoch": 21.517711171662125,
+ "grad_norm": 5.6254963874816895,
+ "learning_rate": 1.825480500725186e-05,
+ "loss": 0.6237,
+ "step": 7897
+ },
+ {
+ "epoch": 21.520435967302454,
+ "grad_norm": 7.059391021728516,
+ "learning_rate": 1.8254306868099885e-05,
+ "loss": 0.7599,
+ "step": 7898
+ },
+ {
+ "epoch": 21.52316076294278,
+ "grad_norm": 5.5015387535095215,
+ "learning_rate": 1.8253808664663636e-05,
+ "loss": 0.3917,
+ "step": 7899
+ },
+ {
+ "epoch": 21.525885558583106,
+ "grad_norm": 7.42938232421875,
+ "learning_rate": 1.8253310396946992e-05,
+ "loss": 0.5424,
+ "step": 7900
+ },
+ {
+ "epoch": 21.52861035422343,
+ "grad_norm": 6.3859734535217285,
+ "learning_rate": 1.8252812064953832e-05,
+ "loss": 0.4698,
+ "step": 7901
+ },
+ {
+ "epoch": 21.53133514986376,
+ "grad_norm": 6.684717655181885,
+ "learning_rate": 1.8252313668688036e-05,
+ "loss": 0.5241,
+ "step": 7902
+ },
+ {
+ "epoch": 21.534059945504087,
+ "grad_norm": 5.322765350341797,
+ "learning_rate": 1.825181520815349e-05,
+ "loss": 0.5039,
+ "step": 7903
+ },
+ {
+ "epoch": 21.536784741144416,
+ "grad_norm": 7.488748073577881,
+ "learning_rate": 1.825131668335407e-05,
+ "loss": 0.4409,
+ "step": 7904
+ },
+ {
+ "epoch": 21.539509536784742,
+ "grad_norm": 5.466688632965088,
+ "learning_rate": 1.825081809429366e-05,
+ "loss": 0.4077,
+ "step": 7905
+ },
+ {
+ "epoch": 21.542234332425068,
+ "grad_norm": 8.959193229675293,
+ "learning_rate": 1.8250319440976146e-05,
+ "loss": 0.4377,
+ "step": 7906
+ },
+ {
+ "epoch": 21.544959128065393,
+ "grad_norm": 5.4138031005859375,
+ "learning_rate": 1.824982072340541e-05,
+ "loss": 0.5129,
+ "step": 7907
+ },
+ {
+ "epoch": 21.547683923705723,
+ "grad_norm": 6.554802894592285,
+ "learning_rate": 1.8249321941585333e-05,
+ "loss": 0.429,
+ "step": 7908
+ },
+ {
+ "epoch": 21.55040871934605,
+ "grad_norm": 6.162092208862305,
+ "learning_rate": 1.8248823095519804e-05,
+ "loss": 0.6288,
+ "step": 7909
+ },
+ {
+ "epoch": 21.553133514986374,
+ "grad_norm": 6.126169681549072,
+ "learning_rate": 1.824832418521271e-05,
+ "loss": 0.6216,
+ "step": 7910
+ },
+ {
+ "epoch": 21.555858310626704,
+ "grad_norm": 6.796255111694336,
+ "learning_rate": 1.8247825210667926e-05,
+ "loss": 0.5835,
+ "step": 7911
+ },
+ {
+ "epoch": 21.55858310626703,
+ "grad_norm": 6.341454029083252,
+ "learning_rate": 1.8247326171889346e-05,
+ "loss": 0.5371,
+ "step": 7912
+ },
+ {
+ "epoch": 21.561307901907355,
+ "grad_norm": 6.530711650848389,
+ "learning_rate": 1.824682706888086e-05,
+ "loss": 0.5344,
+ "step": 7913
+ },
+ {
+ "epoch": 21.564032697547685,
+ "grad_norm": 5.348074913024902,
+ "learning_rate": 1.8246327901646344e-05,
+ "loss": 0.6793,
+ "step": 7914
+ },
+ {
+ "epoch": 21.56675749318801,
+ "grad_norm": 5.473548412322998,
+ "learning_rate": 1.8245828670189697e-05,
+ "loss": 0.4391,
+ "step": 7915
+ },
+ {
+ "epoch": 21.569482288828336,
+ "grad_norm": 5.46661376953125,
+ "learning_rate": 1.8245329374514803e-05,
+ "loss": 0.5068,
+ "step": 7916
+ },
+ {
+ "epoch": 21.572207084468666,
+ "grad_norm": 7.211655139923096,
+ "learning_rate": 1.8244830014625546e-05,
+ "loss": 0.54,
+ "step": 7917
+ },
+ {
+ "epoch": 21.57493188010899,
+ "grad_norm": 5.700046539306641,
+ "learning_rate": 1.8244330590525817e-05,
+ "loss": 0.4625,
+ "step": 7918
+ },
+ {
+ "epoch": 21.577656675749317,
+ "grad_norm": 7.159949779510498,
+ "learning_rate": 1.8243831102219507e-05,
+ "loss": 0.6549,
+ "step": 7919
+ },
+ {
+ "epoch": 21.580381471389646,
+ "grad_norm": 9.861040115356445,
+ "learning_rate": 1.8243331549710505e-05,
+ "loss": 0.4073,
+ "step": 7920
+ },
+ {
+ "epoch": 21.583106267029972,
+ "grad_norm": 6.237105846405029,
+ "learning_rate": 1.8242831933002704e-05,
+ "loss": 0.4108,
+ "step": 7921
+ },
+ {
+ "epoch": 21.585831062670298,
+ "grad_norm": 6.439182281494141,
+ "learning_rate": 1.8242332252099994e-05,
+ "loss": 0.6625,
+ "step": 7922
+ },
+ {
+ "epoch": 21.588555858310627,
+ "grad_norm": 5.436136722564697,
+ "learning_rate": 1.8241832507006265e-05,
+ "loss": 0.5723,
+ "step": 7923
+ },
+ {
+ "epoch": 21.591280653950953,
+ "grad_norm": 5.730956554412842,
+ "learning_rate": 1.824133269772541e-05,
+ "loss": 0.6364,
+ "step": 7924
+ },
+ {
+ "epoch": 21.59400544959128,
+ "grad_norm": 9.899507522583008,
+ "learning_rate": 1.824083282426132e-05,
+ "loss": 0.492,
+ "step": 7925
+ },
+ {
+ "epoch": 21.59673024523161,
+ "grad_norm": 5.837891101837158,
+ "learning_rate": 1.824033288661789e-05,
+ "loss": 0.5613,
+ "step": 7926
+ },
+ {
+ "epoch": 21.599455040871934,
+ "grad_norm": 8.180339813232422,
+ "learning_rate": 1.823983288479901e-05,
+ "loss": 0.4421,
+ "step": 7927
+ },
+ {
+ "epoch": 21.60217983651226,
+ "grad_norm": 6.170220375061035,
+ "learning_rate": 1.823933281880858e-05,
+ "loss": 0.5661,
+ "step": 7928
+ },
+ {
+ "epoch": 21.60490463215259,
+ "grad_norm": 8.234917640686035,
+ "learning_rate": 1.823883268865049e-05,
+ "loss": 0.525,
+ "step": 7929
+ },
+ {
+ "epoch": 21.607629427792915,
+ "grad_norm": 5.9672064781188965,
+ "learning_rate": 1.823833249432864e-05,
+ "loss": 0.5446,
+ "step": 7930
+ },
+ {
+ "epoch": 21.61035422343324,
+ "grad_norm": 6.023001194000244,
+ "learning_rate": 1.8237832235846916e-05,
+ "loss": 0.5753,
+ "step": 7931
+ },
+ {
+ "epoch": 21.61307901907357,
+ "grad_norm": 6.428154468536377,
+ "learning_rate": 1.8237331913209223e-05,
+ "loss": 0.5445,
+ "step": 7932
+ },
+ {
+ "epoch": 21.615803814713896,
+ "grad_norm": 5.483814239501953,
+ "learning_rate": 1.8236831526419453e-05,
+ "loss": 0.376,
+ "step": 7933
+ },
+ {
+ "epoch": 21.618528610354222,
+ "grad_norm": 7.613330841064453,
+ "learning_rate": 1.8236331075481503e-05,
+ "loss": 0.434,
+ "step": 7934
+ },
+ {
+ "epoch": 21.62125340599455,
+ "grad_norm": 7.4714674949646,
+ "learning_rate": 1.8235830560399273e-05,
+ "loss": 0.4666,
+ "step": 7935
+ },
+ {
+ "epoch": 21.623978201634877,
+ "grad_norm": 6.239795207977295,
+ "learning_rate": 1.823532998117666e-05,
+ "loss": 0.5266,
+ "step": 7936
+ },
+ {
+ "epoch": 21.626702997275203,
+ "grad_norm": 5.838024139404297,
+ "learning_rate": 1.8234829337817562e-05,
+ "loss": 0.5607,
+ "step": 7937
+ },
+ {
+ "epoch": 21.629427792915532,
+ "grad_norm": 7.246597766876221,
+ "learning_rate": 1.8234328630325876e-05,
+ "loss": 0.6503,
+ "step": 7938
+ },
+ {
+ "epoch": 21.632152588555858,
+ "grad_norm": 5.666714191436768,
+ "learning_rate": 1.8233827858705507e-05,
+ "loss": 0.6861,
+ "step": 7939
+ },
+ {
+ "epoch": 21.634877384196184,
+ "grad_norm": 7.080030918121338,
+ "learning_rate": 1.823332702296035e-05,
+ "loss": 0.5517,
+ "step": 7940
+ },
+ {
+ "epoch": 21.637602179836513,
+ "grad_norm": 18.99826431274414,
+ "learning_rate": 1.8232826123094308e-05,
+ "loss": 0.444,
+ "step": 7941
+ },
+ {
+ "epoch": 21.64032697547684,
+ "grad_norm": 6.171910762786865,
+ "learning_rate": 1.823232515911128e-05,
+ "loss": 0.6218,
+ "step": 7942
+ },
+ {
+ "epoch": 21.643051771117165,
+ "grad_norm": 5.043895244598389,
+ "learning_rate": 1.8231824131015166e-05,
+ "loss": 0.3918,
+ "step": 7943
+ },
+ {
+ "epoch": 21.645776566757494,
+ "grad_norm": 5.823441028594971,
+ "learning_rate": 1.823132303880988e-05,
+ "loss": 0.5177,
+ "step": 7944
+ },
+ {
+ "epoch": 21.64850136239782,
+ "grad_norm": 6.351459503173828,
+ "learning_rate": 1.8230821882499307e-05,
+ "loss": 0.6653,
+ "step": 7945
+ },
+ {
+ "epoch": 21.651226158038146,
+ "grad_norm": 6.540627956390381,
+ "learning_rate": 1.823032066208736e-05,
+ "loss": 0.558,
+ "step": 7946
+ },
+ {
+ "epoch": 21.653950953678475,
+ "grad_norm": 7.862635135650635,
+ "learning_rate": 1.8229819377577938e-05,
+ "loss": 0.4922,
+ "step": 7947
+ },
+ {
+ "epoch": 21.6566757493188,
+ "grad_norm": 5.918650150299072,
+ "learning_rate": 1.822931802897495e-05,
+ "loss": 0.4669,
+ "step": 7948
+ },
+ {
+ "epoch": 21.659400544959126,
+ "grad_norm": 8.157974243164062,
+ "learning_rate": 1.8228816616282298e-05,
+ "loss": 0.5812,
+ "step": 7949
+ },
+ {
+ "epoch": 21.662125340599456,
+ "grad_norm": 6.290282726287842,
+ "learning_rate": 1.822831513950389e-05,
+ "loss": 0.5709,
+ "step": 7950
+ },
+ {
+ "epoch": 21.66485013623978,
+ "grad_norm": 6.061375617980957,
+ "learning_rate": 1.8227813598643623e-05,
+ "loss": 0.427,
+ "step": 7951
+ },
+ {
+ "epoch": 21.667574931880107,
+ "grad_norm": 6.90337610244751,
+ "learning_rate": 1.822731199370541e-05,
+ "loss": 0.5262,
+ "step": 7952
+ },
+ {
+ "epoch": 21.670299727520437,
+ "grad_norm": 5.399316310882568,
+ "learning_rate": 1.822681032469316e-05,
+ "loss": 0.5081,
+ "step": 7953
+ },
+ {
+ "epoch": 21.673024523160763,
+ "grad_norm": 5.918412208557129,
+ "learning_rate": 1.8226308591610772e-05,
+ "loss": 0.4744,
+ "step": 7954
+ },
+ {
+ "epoch": 21.67574931880109,
+ "grad_norm": 6.98594856262207,
+ "learning_rate": 1.8225806794462156e-05,
+ "loss": 0.6912,
+ "step": 7955
+ },
+ {
+ "epoch": 21.678474114441418,
+ "grad_norm": 7.935492515563965,
+ "learning_rate": 1.8225304933251226e-05,
+ "loss": 0.6049,
+ "step": 7956
+ },
+ {
+ "epoch": 21.681198910081743,
+ "grad_norm": 5.814154148101807,
+ "learning_rate": 1.8224803007981883e-05,
+ "loss": 0.461,
+ "step": 7957
+ },
+ {
+ "epoch": 21.68392370572207,
+ "grad_norm": 5.474069595336914,
+ "learning_rate": 1.8224301018658042e-05,
+ "loss": 0.4174,
+ "step": 7958
+ },
+ {
+ "epoch": 21.6866485013624,
+ "grad_norm": 6.181677341461182,
+ "learning_rate": 1.8223798965283606e-05,
+ "loss": 0.603,
+ "step": 7959
+ },
+ {
+ "epoch": 21.689373297002724,
+ "grad_norm": 6.438231945037842,
+ "learning_rate": 1.8223296847862492e-05,
+ "loss": 0.5592,
+ "step": 7960
+ },
+ {
+ "epoch": 21.69209809264305,
+ "grad_norm": 5.378051280975342,
+ "learning_rate": 1.8222794666398602e-05,
+ "loss": 0.5544,
+ "step": 7961
+ },
+ {
+ "epoch": 21.69482288828338,
+ "grad_norm": 6.118982791900635,
+ "learning_rate": 1.8222292420895857e-05,
+ "loss": 0.5759,
+ "step": 7962
+ },
+ {
+ "epoch": 21.697547683923705,
+ "grad_norm": 6.013671398162842,
+ "learning_rate": 1.8221790111358156e-05,
+ "loss": 0.4791,
+ "step": 7963
+ },
+ {
+ "epoch": 21.70027247956403,
+ "grad_norm": 4.885951995849609,
+ "learning_rate": 1.8221287737789423e-05,
+ "loss": 0.5774,
+ "step": 7964
+ },
+ {
+ "epoch": 21.70299727520436,
+ "grad_norm": 5.8371968269348145,
+ "learning_rate": 1.8220785300193565e-05,
+ "loss": 0.5955,
+ "step": 7965
+ },
+ {
+ "epoch": 21.705722070844686,
+ "grad_norm": 5.985840797424316,
+ "learning_rate": 1.8220282798574497e-05,
+ "loss": 0.6238,
+ "step": 7966
+ },
+ {
+ "epoch": 21.708446866485012,
+ "grad_norm": 5.392521381378174,
+ "learning_rate": 1.821978023293613e-05,
+ "loss": 0.4698,
+ "step": 7967
+ },
+ {
+ "epoch": 21.71117166212534,
+ "grad_norm": 5.635871887207031,
+ "learning_rate": 1.8219277603282378e-05,
+ "loss": 0.6533,
+ "step": 7968
+ },
+ {
+ "epoch": 21.713896457765667,
+ "grad_norm": 7.626245975494385,
+ "learning_rate": 1.8218774909617157e-05,
+ "loss": 0.6814,
+ "step": 7969
+ },
+ {
+ "epoch": 21.716621253405993,
+ "grad_norm": 6.3506035804748535,
+ "learning_rate": 1.8218272151944383e-05,
+ "loss": 0.6121,
+ "step": 7970
+ },
+ {
+ "epoch": 21.719346049046322,
+ "grad_norm": 5.254416465759277,
+ "learning_rate": 1.821776933026797e-05,
+ "loss": 0.561,
+ "step": 7971
+ },
+ {
+ "epoch": 21.722070844686648,
+ "grad_norm": 6.696042537689209,
+ "learning_rate": 1.8217266444591835e-05,
+ "loss": 0.4442,
+ "step": 7972
+ },
+ {
+ "epoch": 21.724795640326974,
+ "grad_norm": 6.128903865814209,
+ "learning_rate": 1.8216763494919888e-05,
+ "loss": 0.6187,
+ "step": 7973
+ },
+ {
+ "epoch": 21.727520435967303,
+ "grad_norm": 5.460041522979736,
+ "learning_rate": 1.8216260481256058e-05,
+ "loss": 0.4824,
+ "step": 7974
+ },
+ {
+ "epoch": 21.73024523160763,
+ "grad_norm": 5.770877361297607,
+ "learning_rate": 1.8215757403604252e-05,
+ "loss": 0.5107,
+ "step": 7975
+ },
+ {
+ "epoch": 21.732970027247955,
+ "grad_norm": 5.82627534866333,
+ "learning_rate": 1.821525426196839e-05,
+ "loss": 0.7224,
+ "step": 7976
+ },
+ {
+ "epoch": 21.735694822888284,
+ "grad_norm": 5.801827430725098,
+ "learning_rate": 1.8214751056352397e-05,
+ "loss": 0.461,
+ "step": 7977
+ },
+ {
+ "epoch": 21.73841961852861,
+ "grad_norm": 5.308289527893066,
+ "learning_rate": 1.8214247786760184e-05,
+ "loss": 0.4703,
+ "step": 7978
+ },
+ {
+ "epoch": 21.741144414168936,
+ "grad_norm": 6.970522403717041,
+ "learning_rate": 1.8213744453195676e-05,
+ "loss": 0.7163,
+ "step": 7979
+ },
+ {
+ "epoch": 21.743869209809265,
+ "grad_norm": 5.97580623626709,
+ "learning_rate": 1.8213241055662786e-05,
+ "loss": 0.5487,
+ "step": 7980
+ },
+ {
+ "epoch": 21.74659400544959,
+ "grad_norm": 5.600148677825928,
+ "learning_rate": 1.8212737594165444e-05,
+ "loss": 0.5074,
+ "step": 7981
+ },
+ {
+ "epoch": 21.749318801089917,
+ "grad_norm": 5.716439247131348,
+ "learning_rate": 1.8212234068707563e-05,
+ "loss": 0.5457,
+ "step": 7982
+ },
+ {
+ "epoch": 21.752043596730246,
+ "grad_norm": 5.431583881378174,
+ "learning_rate": 1.8211730479293067e-05,
+ "loss": 0.4861,
+ "step": 7983
+ },
+ {
+ "epoch": 21.754768392370572,
+ "grad_norm": 5.838796138763428,
+ "learning_rate": 1.8211226825925883e-05,
+ "loss": 0.4687,
+ "step": 7984
+ },
+ {
+ "epoch": 21.757493188010898,
+ "grad_norm": 6.609949111938477,
+ "learning_rate": 1.8210723108609923e-05,
+ "loss": 0.5385,
+ "step": 7985
+ },
+ {
+ "epoch": 21.760217983651227,
+ "grad_norm": 6.743381500244141,
+ "learning_rate": 1.8210219327349122e-05,
+ "loss": 0.5498,
+ "step": 7986
+ },
+ {
+ "epoch": 21.762942779291553,
+ "grad_norm": 5.8138933181762695,
+ "learning_rate": 1.8209715482147393e-05,
+ "loss": 0.4129,
+ "step": 7987
+ },
+ {
+ "epoch": 21.76566757493188,
+ "grad_norm": 5.7851643562316895,
+ "learning_rate": 1.8209211573008663e-05,
+ "loss": 0.6584,
+ "step": 7988
+ },
+ {
+ "epoch": 21.768392370572208,
+ "grad_norm": 6.5312957763671875,
+ "learning_rate": 1.8208707599936858e-05,
+ "loss": 0.6279,
+ "step": 7989
+ },
+ {
+ "epoch": 21.771117166212534,
+ "grad_norm": 7.346785545349121,
+ "learning_rate": 1.8208203562935902e-05,
+ "loss": 0.5481,
+ "step": 7990
+ },
+ {
+ "epoch": 21.77384196185286,
+ "grad_norm": 6.24441385269165,
+ "learning_rate": 1.8207699462009722e-05,
+ "loss": 0.48,
+ "step": 7991
+ },
+ {
+ "epoch": 21.77656675749319,
+ "grad_norm": 5.952081203460693,
+ "learning_rate": 1.8207195297162245e-05,
+ "loss": 0.4801,
+ "step": 7992
+ },
+ {
+ "epoch": 21.779291553133515,
+ "grad_norm": 6.025998115539551,
+ "learning_rate": 1.820669106839739e-05,
+ "loss": 0.5834,
+ "step": 7993
+ },
+ {
+ "epoch": 21.78201634877384,
+ "grad_norm": 5.039198875427246,
+ "learning_rate": 1.8206186775719096e-05,
+ "loss": 0.4333,
+ "step": 7994
+ },
+ {
+ "epoch": 21.78474114441417,
+ "grad_norm": 6.9681715965271,
+ "learning_rate": 1.8205682419131277e-05,
+ "loss": 0.5359,
+ "step": 7995
+ },
+ {
+ "epoch": 21.787465940054496,
+ "grad_norm": 6.894341468811035,
+ "learning_rate": 1.8205177998637872e-05,
+ "loss": 0.6056,
+ "step": 7996
+ },
+ {
+ "epoch": 21.79019073569482,
+ "grad_norm": 8.482925415039062,
+ "learning_rate": 1.8204673514242804e-05,
+ "loss": 0.6261,
+ "step": 7997
+ },
+ {
+ "epoch": 21.79291553133515,
+ "grad_norm": 6.680730819702148,
+ "learning_rate": 1.820416896595e-05,
+ "loss": 0.5401,
+ "step": 7998
+ },
+ {
+ "epoch": 21.795640326975477,
+ "grad_norm": 6.907870769500732,
+ "learning_rate": 1.8203664353763393e-05,
+ "loss": 0.449,
+ "step": 7999
+ },
+ {
+ "epoch": 21.798365122615802,
+ "grad_norm": 5.91664457321167,
+ "learning_rate": 1.8203159677686914e-05,
+ "loss": 0.4884,
+ "step": 8000
+ },
+ {
+ "epoch": 21.80108991825613,
+ "grad_norm": 5.6211256980896,
+ "learning_rate": 1.820265493772449e-05,
+ "loss": 0.4304,
+ "step": 8001
+ },
+ {
+ "epoch": 21.803814713896458,
+ "grad_norm": 8.808941841125488,
+ "learning_rate": 1.8202150133880052e-05,
+ "loss": 0.5236,
+ "step": 8002
+ },
+ {
+ "epoch": 21.806539509536783,
+ "grad_norm": 11.842449188232422,
+ "learning_rate": 1.8201645266157534e-05,
+ "loss": 0.5942,
+ "step": 8003
+ },
+ {
+ "epoch": 21.809264305177113,
+ "grad_norm": 5.853658199310303,
+ "learning_rate": 1.8201140334560867e-05,
+ "loss": 0.6423,
+ "step": 8004
+ },
+ {
+ "epoch": 21.81198910081744,
+ "grad_norm": 7.32819128036499,
+ "learning_rate": 1.8200635339093983e-05,
+ "loss": 0.4752,
+ "step": 8005
+ },
+ {
+ "epoch": 21.814713896457764,
+ "grad_norm": 17.632057189941406,
+ "learning_rate": 1.8200130279760814e-05,
+ "loss": 0.6003,
+ "step": 8006
+ },
+ {
+ "epoch": 21.817438692098094,
+ "grad_norm": 31.791088104248047,
+ "learning_rate": 1.8199625156565295e-05,
+ "loss": 0.6349,
+ "step": 8007
+ },
+ {
+ "epoch": 21.82016348773842,
+ "grad_norm": 7.382207870483398,
+ "learning_rate": 1.8199119969511362e-05,
+ "loss": 0.4429,
+ "step": 8008
+ },
+ {
+ "epoch": 21.822888283378745,
+ "grad_norm": 5.737339019775391,
+ "learning_rate": 1.8198614718602943e-05,
+ "loss": 0.545,
+ "step": 8009
+ },
+ {
+ "epoch": 21.825613079019075,
+ "grad_norm": 6.656451225280762,
+ "learning_rate": 1.8198109403843977e-05,
+ "loss": 0.5056,
+ "step": 8010
+ },
+ {
+ "epoch": 21.8283378746594,
+ "grad_norm": 6.518035411834717,
+ "learning_rate": 1.8197604025238398e-05,
+ "loss": 0.6136,
+ "step": 8011
+ },
+ {
+ "epoch": 21.831062670299726,
+ "grad_norm": 10.158177375793457,
+ "learning_rate": 1.8197098582790144e-05,
+ "loss": 0.5067,
+ "step": 8012
+ },
+ {
+ "epoch": 21.833787465940055,
+ "grad_norm": 5.7067551612854,
+ "learning_rate": 1.819659307650315e-05,
+ "loss": 0.5745,
+ "step": 8013
+ },
+ {
+ "epoch": 21.83651226158038,
+ "grad_norm": 6.970973968505859,
+ "learning_rate": 1.8196087506381356e-05,
+ "loss": 0.619,
+ "step": 8014
+ },
+ {
+ "epoch": 21.839237057220707,
+ "grad_norm": 8.677037239074707,
+ "learning_rate": 1.8195581872428693e-05,
+ "loss": 0.5645,
+ "step": 8015
+ },
+ {
+ "epoch": 21.841961852861036,
+ "grad_norm": 5.472105979919434,
+ "learning_rate": 1.81950761746491e-05,
+ "loss": 0.5778,
+ "step": 8016
+ },
+ {
+ "epoch": 21.844686648501362,
+ "grad_norm": 7.211095809936523,
+ "learning_rate": 1.819457041304652e-05,
+ "loss": 0.6204,
+ "step": 8017
+ },
+ {
+ "epoch": 21.847411444141688,
+ "grad_norm": 10.444666862487793,
+ "learning_rate": 1.819406458762489e-05,
+ "loss": 0.5694,
+ "step": 8018
+ },
+ {
+ "epoch": 21.850136239782017,
+ "grad_norm": 6.8338422775268555,
+ "learning_rate": 1.8193558698388148e-05,
+ "loss": 0.5174,
+ "step": 8019
+ },
+ {
+ "epoch": 21.852861035422343,
+ "grad_norm": 6.663456916809082,
+ "learning_rate": 1.819305274534023e-05,
+ "loss": 0.5885,
+ "step": 8020
+ },
+ {
+ "epoch": 21.85558583106267,
+ "grad_norm": 11.01484203338623,
+ "learning_rate": 1.8192546728485087e-05,
+ "loss": 0.491,
+ "step": 8021
+ },
+ {
+ "epoch": 21.858310626703,
+ "grad_norm": 9.548139572143555,
+ "learning_rate": 1.8192040647826652e-05,
+ "loss": 0.7937,
+ "step": 8022
+ },
+ {
+ "epoch": 21.861035422343324,
+ "grad_norm": 6.685322284698486,
+ "learning_rate": 1.819153450336887e-05,
+ "loss": 0.4647,
+ "step": 8023
+ },
+ {
+ "epoch": 21.86376021798365,
+ "grad_norm": 362.7572937011719,
+ "learning_rate": 1.819102829511568e-05,
+ "loss": 0.4812,
+ "step": 8024
+ },
+ {
+ "epoch": 21.86648501362398,
+ "grad_norm": 8.741470336914062,
+ "learning_rate": 1.819052202307102e-05,
+ "loss": 0.5931,
+ "step": 8025
+ },
+ {
+ "epoch": 21.869209809264305,
+ "grad_norm": 41.90818405151367,
+ "learning_rate": 1.8190015687238848e-05,
+ "loss": 0.6832,
+ "step": 8026
+ },
+ {
+ "epoch": 21.87193460490463,
+ "grad_norm": 8.217859268188477,
+ "learning_rate": 1.818950928762309e-05,
+ "loss": 0.5731,
+ "step": 8027
+ },
+ {
+ "epoch": 21.87465940054496,
+ "grad_norm": 9.2962007522583,
+ "learning_rate": 1.81890028242277e-05,
+ "loss": 0.6142,
+ "step": 8028
+ },
+ {
+ "epoch": 21.877384196185286,
+ "grad_norm": 23.523733139038086,
+ "learning_rate": 1.8188496297056616e-05,
+ "loss": 0.6073,
+ "step": 8029
+ },
+ {
+ "epoch": 21.88010899182561,
+ "grad_norm": 12.070014953613281,
+ "learning_rate": 1.818798970611379e-05,
+ "loss": 0.7061,
+ "step": 8030
+ },
+ {
+ "epoch": 21.88283378746594,
+ "grad_norm": 38.599422454833984,
+ "learning_rate": 1.8187483051403162e-05,
+ "loss": 0.8501,
+ "step": 8031
+ },
+ {
+ "epoch": 21.885558583106267,
+ "grad_norm": 13.429490089416504,
+ "learning_rate": 1.818697633292868e-05,
+ "loss": 0.5769,
+ "step": 8032
+ },
+ {
+ "epoch": 21.888283378746593,
+ "grad_norm": 10.110140800476074,
+ "learning_rate": 1.818646955069429e-05,
+ "loss": 0.7906,
+ "step": 8033
+ },
+ {
+ "epoch": 21.891008174386922,
+ "grad_norm": 16.479171752929688,
+ "learning_rate": 1.8185962704703942e-05,
+ "loss": 0.6018,
+ "step": 8034
+ },
+ {
+ "epoch": 21.893732970027248,
+ "grad_norm": 9.962614059448242,
+ "learning_rate": 1.8185455794961574e-05,
+ "loss": 0.8174,
+ "step": 8035
+ },
+ {
+ "epoch": 21.896457765667574,
+ "grad_norm": 13.182805061340332,
+ "learning_rate": 1.8184948821471144e-05,
+ "loss": 0.6968,
+ "step": 8036
+ },
+ {
+ "epoch": 21.899182561307903,
+ "grad_norm": 12.504403114318848,
+ "learning_rate": 1.8184441784236595e-05,
+ "loss": 0.8348,
+ "step": 8037
+ },
+ {
+ "epoch": 21.90190735694823,
+ "grad_norm": 8.403986930847168,
+ "learning_rate": 1.8183934683261875e-05,
+ "loss": 0.7549,
+ "step": 8038
+ },
+ {
+ "epoch": 21.904632152588555,
+ "grad_norm": 8.992491722106934,
+ "learning_rate": 1.8183427518550934e-05,
+ "loss": 0.6244,
+ "step": 8039
+ },
+ {
+ "epoch": 21.907356948228884,
+ "grad_norm": 8.725276947021484,
+ "learning_rate": 1.8182920290107727e-05,
+ "loss": 0.8142,
+ "step": 8040
+ },
+ {
+ "epoch": 21.91008174386921,
+ "grad_norm": 7.707148551940918,
+ "learning_rate": 1.8182412997936197e-05,
+ "loss": 0.8951,
+ "step": 8041
+ },
+ {
+ "epoch": 21.912806539509535,
+ "grad_norm": 9.601200103759766,
+ "learning_rate": 1.8181905642040298e-05,
+ "loss": 0.5831,
+ "step": 8042
+ },
+ {
+ "epoch": 21.915531335149865,
+ "grad_norm": 8.56784725189209,
+ "learning_rate": 1.8181398222423984e-05,
+ "loss": 0.6381,
+ "step": 8043
+ },
+ {
+ "epoch": 21.91825613079019,
+ "grad_norm": 11.084944725036621,
+ "learning_rate": 1.8180890739091205e-05,
+ "loss": 0.5733,
+ "step": 8044
+ },
+ {
+ "epoch": 21.920980926430516,
+ "grad_norm": 8.981050491333008,
+ "learning_rate": 1.8180383192045906e-05,
+ "loss": 0.6132,
+ "step": 8045
+ },
+ {
+ "epoch": 21.923705722070846,
+ "grad_norm": 9.40047550201416,
+ "learning_rate": 1.817987558129205e-05,
+ "loss": 0.7256,
+ "step": 8046
+ },
+ {
+ "epoch": 21.92643051771117,
+ "grad_norm": 10.190051078796387,
+ "learning_rate": 1.8179367906833586e-05,
+ "loss": 0.5851,
+ "step": 8047
+ },
+ {
+ "epoch": 21.929155313351497,
+ "grad_norm": 8.13205337524414,
+ "learning_rate": 1.817886016867447e-05,
+ "loss": 0.5619,
+ "step": 8048
+ },
+ {
+ "epoch": 21.931880108991827,
+ "grad_norm": 7.345907211303711,
+ "learning_rate": 1.817835236681865e-05,
+ "loss": 0.6339,
+ "step": 8049
+ },
+ {
+ "epoch": 21.934604904632153,
+ "grad_norm": 8.420438766479492,
+ "learning_rate": 1.8177844501270088e-05,
+ "loss": 0.696,
+ "step": 8050
+ },
+ {
+ "epoch": 21.93732970027248,
+ "grad_norm": 8.262900352478027,
+ "learning_rate": 1.817733657203273e-05,
+ "loss": 0.7177,
+ "step": 8051
+ },
+ {
+ "epoch": 21.940054495912808,
+ "grad_norm": 7.63875150680542,
+ "learning_rate": 1.817682857911055e-05,
+ "loss": 0.734,
+ "step": 8052
+ },
+ {
+ "epoch": 21.942779291553133,
+ "grad_norm": 10.72281265258789,
+ "learning_rate": 1.8176320522507485e-05,
+ "loss": 0.5347,
+ "step": 8053
+ },
+ {
+ "epoch": 21.94550408719346,
+ "grad_norm": 6.902315139770508,
+ "learning_rate": 1.81758124022275e-05,
+ "loss": 0.5462,
+ "step": 8054
+ },
+ {
+ "epoch": 21.94822888283379,
+ "grad_norm": 6.360598564147949,
+ "learning_rate": 1.817530421827455e-05,
+ "loss": 0.772,
+ "step": 8055
+ },
+ {
+ "epoch": 21.950953678474114,
+ "grad_norm": 6.316249370574951,
+ "learning_rate": 1.8174795970652592e-05,
+ "loss": 0.6562,
+ "step": 8056
+ },
+ {
+ "epoch": 21.95367847411444,
+ "grad_norm": 5.5031890869140625,
+ "learning_rate": 1.817428765936559e-05,
+ "loss": 0.7445,
+ "step": 8057
+ },
+ {
+ "epoch": 21.95640326975477,
+ "grad_norm": 8.260133743286133,
+ "learning_rate": 1.8173779284417497e-05,
+ "loss": 0.8109,
+ "step": 8058
+ },
+ {
+ "epoch": 21.959128065395095,
+ "grad_norm": 7.3662614822387695,
+ "learning_rate": 1.8173270845812273e-05,
+ "loss": 0.5605,
+ "step": 8059
+ },
+ {
+ "epoch": 21.96185286103542,
+ "grad_norm": 9.07149600982666,
+ "learning_rate": 1.8172762343553883e-05,
+ "loss": 0.6533,
+ "step": 8060
+ },
+ {
+ "epoch": 21.96457765667575,
+ "grad_norm": 7.246875762939453,
+ "learning_rate": 1.817225377764628e-05,
+ "loss": 0.5771,
+ "step": 8061
+ },
+ {
+ "epoch": 21.967302452316076,
+ "grad_norm": 6.986541748046875,
+ "learning_rate": 1.8171745148093425e-05,
+ "loss": 0.5493,
+ "step": 8062
+ },
+ {
+ "epoch": 21.970027247956402,
+ "grad_norm": 8.198639869689941,
+ "learning_rate": 1.8171236454899283e-05,
+ "loss": 0.6404,
+ "step": 8063
+ },
+ {
+ "epoch": 21.97275204359673,
+ "grad_norm": 5.832398414611816,
+ "learning_rate": 1.8170727698067817e-05,
+ "loss": 0.7172,
+ "step": 8064
+ },
+ {
+ "epoch": 21.975476839237057,
+ "grad_norm": 8.525985717773438,
+ "learning_rate": 1.8170218877602987e-05,
+ "loss": 0.6661,
+ "step": 8065
+ },
+ {
+ "epoch": 21.978201634877383,
+ "grad_norm": 6.787590026855469,
+ "learning_rate": 1.816970999350875e-05,
+ "loss": 0.573,
+ "step": 8066
+ },
+ {
+ "epoch": 21.980926430517712,
+ "grad_norm": 6.952709674835205,
+ "learning_rate": 1.816920104578908e-05,
+ "loss": 0.6409,
+ "step": 8067
+ },
+ {
+ "epoch": 21.983651226158038,
+ "grad_norm": 7.8918046951293945,
+ "learning_rate": 1.816869203444793e-05,
+ "loss": 0.7479,
+ "step": 8068
+ },
+ {
+ "epoch": 21.986376021798364,
+ "grad_norm": 7.470007419586182,
+ "learning_rate": 1.8168182959489277e-05,
+ "loss": 0.5551,
+ "step": 8069
+ },
+ {
+ "epoch": 21.989100817438693,
+ "grad_norm": 6.126591205596924,
+ "learning_rate": 1.8167673820917073e-05,
+ "loss": 0.6218,
+ "step": 8070
+ },
+ {
+ "epoch": 21.99182561307902,
+ "grad_norm": 6.596359729766846,
+ "learning_rate": 1.816716461873529e-05,
+ "loss": 0.6057,
+ "step": 8071
+ },
+ {
+ "epoch": 21.994550408719345,
+ "grad_norm": 6.980469703674316,
+ "learning_rate": 1.816665535294789e-05,
+ "loss": 0.5852,
+ "step": 8072
+ },
+ {
+ "epoch": 21.997275204359674,
+ "grad_norm": 7.250155925750732,
+ "learning_rate": 1.816614602355884e-05,
+ "loss": 0.6021,
+ "step": 8073
+ },
+ {
+ "epoch": 22.0,
+ "grad_norm": 7.7300190925598145,
+ "learning_rate": 1.816563663057211e-05,
+ "loss": 0.5472,
+ "step": 8074
+ },
+ {
+ "epoch": 22.002724795640326,
+ "grad_norm": 5.230376243591309,
+ "learning_rate": 1.8165127173991667e-05,
+ "loss": 0.561,
+ "step": 8075
+ },
+ {
+ "epoch": 22.005449591280655,
+ "grad_norm": 6.372800350189209,
+ "learning_rate": 1.816461765382147e-05,
+ "loss": 0.6713,
+ "step": 8076
+ },
+ {
+ "epoch": 22.00817438692098,
+ "grad_norm": 6.171205043792725,
+ "learning_rate": 1.8164108070065498e-05,
+ "loss": 0.3545,
+ "step": 8077
+ },
+ {
+ "epoch": 22.010899182561307,
+ "grad_norm": 7.058981418609619,
+ "learning_rate": 1.8163598422727714e-05,
+ "loss": 0.6122,
+ "step": 8078
+ },
+ {
+ "epoch": 22.013623978201636,
+ "grad_norm": 5.800168514251709,
+ "learning_rate": 1.816308871181209e-05,
+ "loss": 0.366,
+ "step": 8079
+ },
+ {
+ "epoch": 22.016348773841962,
+ "grad_norm": 6.280452728271484,
+ "learning_rate": 1.816257893732259e-05,
+ "loss": 0.6224,
+ "step": 8080
+ },
+ {
+ "epoch": 22.019073569482288,
+ "grad_norm": 6.074062824249268,
+ "learning_rate": 1.816206909926319e-05,
+ "loss": 0.4435,
+ "step": 8081
+ },
+ {
+ "epoch": 22.021798365122617,
+ "grad_norm": 6.218075752258301,
+ "learning_rate": 1.816155919763786e-05,
+ "loss": 0.7267,
+ "step": 8082
+ },
+ {
+ "epoch": 22.024523160762943,
+ "grad_norm": 6.1108245849609375,
+ "learning_rate": 1.8161049232450566e-05,
+ "loss": 0.5352,
+ "step": 8083
+ },
+ {
+ "epoch": 22.02724795640327,
+ "grad_norm": 5.95872688293457,
+ "learning_rate": 1.8160539203705285e-05,
+ "loss": 0.4501,
+ "step": 8084
+ },
+ {
+ "epoch": 22.029972752043598,
+ "grad_norm": 7.183713912963867,
+ "learning_rate": 1.8160029111405986e-05,
+ "loss": 0.6617,
+ "step": 8085
+ },
+ {
+ "epoch": 22.032697547683924,
+ "grad_norm": 6.023658752441406,
+ "learning_rate": 1.815951895555665e-05,
+ "loss": 0.4887,
+ "step": 8086
+ },
+ {
+ "epoch": 22.03542234332425,
+ "grad_norm": 7.4763712882995605,
+ "learning_rate": 1.8159008736161236e-05,
+ "loss": 0.5405,
+ "step": 8087
+ },
+ {
+ "epoch": 22.03814713896458,
+ "grad_norm": 5.9674811363220215,
+ "learning_rate": 1.815849845322373e-05,
+ "loss": 0.446,
+ "step": 8088
+ },
+ {
+ "epoch": 22.040871934604905,
+ "grad_norm": 6.368412017822266,
+ "learning_rate": 1.815798810674809e-05,
+ "loss": 0.3873,
+ "step": 8089
+ },
+ {
+ "epoch": 22.04359673024523,
+ "grad_norm": 5.885277271270752,
+ "learning_rate": 1.8157477696738312e-05,
+ "loss": 0.6035,
+ "step": 8090
+ },
+ {
+ "epoch": 22.04632152588556,
+ "grad_norm": 6.044500827789307,
+ "learning_rate": 1.8156967223198358e-05,
+ "loss": 0.4785,
+ "step": 8091
+ },
+ {
+ "epoch": 22.049046321525886,
+ "grad_norm": 8.22576904296875,
+ "learning_rate": 1.81564566861322e-05,
+ "loss": 0.5209,
+ "step": 8092
+ },
+ {
+ "epoch": 22.05177111716621,
+ "grad_norm": 6.92186164855957,
+ "learning_rate": 1.8155946085543825e-05,
+ "loss": 0.466,
+ "step": 8093
+ },
+ {
+ "epoch": 22.05449591280654,
+ "grad_norm": 6.227148056030273,
+ "learning_rate": 1.8155435421437202e-05,
+ "loss": 0.5255,
+ "step": 8094
+ },
+ {
+ "epoch": 22.057220708446867,
+ "grad_norm": 13.529058456420898,
+ "learning_rate": 1.8154924693816312e-05,
+ "loss": 0.5854,
+ "step": 8095
+ },
+ {
+ "epoch": 22.059945504087192,
+ "grad_norm": 6.8699798583984375,
+ "learning_rate": 1.8154413902685132e-05,
+ "loss": 0.4753,
+ "step": 8096
+ },
+ {
+ "epoch": 22.06267029972752,
+ "grad_norm": 5.445714950561523,
+ "learning_rate": 1.8153903048047636e-05,
+ "loss": 0.3486,
+ "step": 8097
+ },
+ {
+ "epoch": 22.065395095367847,
+ "grad_norm": 7.411291122436523,
+ "learning_rate": 1.815339212990781e-05,
+ "loss": 0.4642,
+ "step": 8098
+ },
+ {
+ "epoch": 22.068119891008173,
+ "grad_norm": 8.115220069885254,
+ "learning_rate": 1.8152881148269626e-05,
+ "loss": 0.5812,
+ "step": 8099
+ },
+ {
+ "epoch": 22.070844686648503,
+ "grad_norm": 7.643735408782959,
+ "learning_rate": 1.8152370103137063e-05,
+ "loss": 0.4488,
+ "step": 8100
+ },
+ {
+ "epoch": 22.07356948228883,
+ "grad_norm": 6.086093425750732,
+ "learning_rate": 1.815185899451411e-05,
+ "loss": 0.7108,
+ "step": 8101
+ },
+ {
+ "epoch": 22.076294277929154,
+ "grad_norm": 6.676787853240967,
+ "learning_rate": 1.8151347822404735e-05,
+ "loss": 0.4533,
+ "step": 8102
+ },
+ {
+ "epoch": 22.079019073569484,
+ "grad_norm": 14.559338569641113,
+ "learning_rate": 1.8150836586812927e-05,
+ "loss": 0.5242,
+ "step": 8103
+ },
+ {
+ "epoch": 22.08174386920981,
+ "grad_norm": 6.38640832901001,
+ "learning_rate": 1.815032528774267e-05,
+ "loss": 0.7144,
+ "step": 8104
+ },
+ {
+ "epoch": 22.084468664850135,
+ "grad_norm": 7.68470573425293,
+ "learning_rate": 1.814981392519794e-05,
+ "loss": 0.4543,
+ "step": 8105
+ },
+ {
+ "epoch": 22.087193460490465,
+ "grad_norm": 6.324126720428467,
+ "learning_rate": 1.814930249918272e-05,
+ "loss": 0.5295,
+ "step": 8106
+ },
+ {
+ "epoch": 22.08991825613079,
+ "grad_norm": 6.492694854736328,
+ "learning_rate": 1.8148791009700994e-05,
+ "loss": 0.5103,
+ "step": 8107
+ },
+ {
+ "epoch": 22.092643051771116,
+ "grad_norm": 7.684177875518799,
+ "learning_rate": 1.814827945675675e-05,
+ "loss": 0.4639,
+ "step": 8108
+ },
+ {
+ "epoch": 22.095367847411445,
+ "grad_norm": 9.622164726257324,
+ "learning_rate": 1.8147767840353966e-05,
+ "loss": 0.6135,
+ "step": 8109
+ },
+ {
+ "epoch": 22.09809264305177,
+ "grad_norm": 6.612240314483643,
+ "learning_rate": 1.8147256160496627e-05,
+ "loss": 0.474,
+ "step": 8110
+ },
+ {
+ "epoch": 22.100817438692097,
+ "grad_norm": 6.611939907073975,
+ "learning_rate": 1.814674441718872e-05,
+ "loss": 0.4176,
+ "step": 8111
+ },
+ {
+ "epoch": 22.103542234332426,
+ "grad_norm": 7.75114107131958,
+ "learning_rate": 1.814623261043423e-05,
+ "loss": 0.4915,
+ "step": 8112
+ },
+ {
+ "epoch": 22.106267029972752,
+ "grad_norm": 8.26905345916748,
+ "learning_rate": 1.8145720740237142e-05,
+ "loss": 0.5902,
+ "step": 8113
+ },
+ {
+ "epoch": 22.108991825613078,
+ "grad_norm": 6.504248142242432,
+ "learning_rate": 1.814520880660144e-05,
+ "loss": 0.4838,
+ "step": 8114
+ },
+ {
+ "epoch": 22.111716621253407,
+ "grad_norm": 8.124812126159668,
+ "learning_rate": 1.8144696809531117e-05,
+ "loss": 0.4534,
+ "step": 8115
+ },
+ {
+ "epoch": 22.114441416893733,
+ "grad_norm": 6.631236553192139,
+ "learning_rate": 1.814418474903016e-05,
+ "loss": 0.427,
+ "step": 8116
+ },
+ {
+ "epoch": 22.11716621253406,
+ "grad_norm": 7.432625770568848,
+ "learning_rate": 1.814367262510255e-05,
+ "loss": 0.5916,
+ "step": 8117
+ },
+ {
+ "epoch": 22.11989100817439,
+ "grad_norm": 7.1134257316589355,
+ "learning_rate": 1.8143160437752282e-05,
+ "loss": 0.4964,
+ "step": 8118
+ },
+ {
+ "epoch": 22.122615803814714,
+ "grad_norm": 6.335321426391602,
+ "learning_rate": 1.8142648186983342e-05,
+ "loss": 0.4457,
+ "step": 8119
+ },
+ {
+ "epoch": 22.12534059945504,
+ "grad_norm": 5.772454738616943,
+ "learning_rate": 1.814213587279972e-05,
+ "loss": 0.3415,
+ "step": 8120
+ },
+ {
+ "epoch": 22.12806539509537,
+ "grad_norm": 5.893414497375488,
+ "learning_rate": 1.8141623495205407e-05,
+ "loss": 0.4753,
+ "step": 8121
+ },
+ {
+ "epoch": 22.130790190735695,
+ "grad_norm": 7.034484386444092,
+ "learning_rate": 1.814111105420439e-05,
+ "loss": 0.588,
+ "step": 8122
+ },
+ {
+ "epoch": 22.13351498637602,
+ "grad_norm": 7.8339362144470215,
+ "learning_rate": 1.8140598549800665e-05,
+ "loss": 0.5364,
+ "step": 8123
+ },
+ {
+ "epoch": 22.13623978201635,
+ "grad_norm": 7.399385929107666,
+ "learning_rate": 1.8140085981998217e-05,
+ "loss": 0.744,
+ "step": 8124
+ },
+ {
+ "epoch": 22.138964577656676,
+ "grad_norm": 5.760389804840088,
+ "learning_rate": 1.8139573350801044e-05,
+ "loss": 0.5178,
+ "step": 8125
+ },
+ {
+ "epoch": 22.141689373297,
+ "grad_norm": 6.724281311035156,
+ "learning_rate": 1.8139060656213136e-05,
+ "loss": 0.5708,
+ "step": 8126
+ },
+ {
+ "epoch": 22.14441416893733,
+ "grad_norm": 6.031400680541992,
+ "learning_rate": 1.8138547898238482e-05,
+ "loss": 0.7349,
+ "step": 8127
+ },
+ {
+ "epoch": 22.147138964577657,
+ "grad_norm": 5.938206672668457,
+ "learning_rate": 1.813803507688108e-05,
+ "loss": 0.5341,
+ "step": 8128
+ },
+ {
+ "epoch": 22.149863760217983,
+ "grad_norm": 5.989213943481445,
+ "learning_rate": 1.813752219214493e-05,
+ "loss": 0.434,
+ "step": 8129
+ },
+ {
+ "epoch": 22.152588555858312,
+ "grad_norm": 6.611841201782227,
+ "learning_rate": 1.813700924403401e-05,
+ "loss": 0.7073,
+ "step": 8130
+ },
+ {
+ "epoch": 22.155313351498638,
+ "grad_norm": 4.457615852355957,
+ "learning_rate": 1.8136496232552327e-05,
+ "loss": 0.3464,
+ "step": 8131
+ },
+ {
+ "epoch": 22.158038147138964,
+ "grad_norm": 5.3777241706848145,
+ "learning_rate": 1.8135983157703875e-05,
+ "loss": 0.4323,
+ "step": 8132
+ },
+ {
+ "epoch": 22.160762942779293,
+ "grad_norm": 5.754644393920898,
+ "learning_rate": 1.8135470019492645e-05,
+ "loss": 0.3658,
+ "step": 8133
+ },
+ {
+ "epoch": 22.16348773841962,
+ "grad_norm": 6.732206344604492,
+ "learning_rate": 1.813495681792264e-05,
+ "loss": 0.3533,
+ "step": 8134
+ },
+ {
+ "epoch": 22.166212534059945,
+ "grad_norm": 5.607558250427246,
+ "learning_rate": 1.8134443552997848e-05,
+ "loss": 0.3959,
+ "step": 8135
+ },
+ {
+ "epoch": 22.168937329700274,
+ "grad_norm": 5.39448356628418,
+ "learning_rate": 1.8133930224722274e-05,
+ "loss": 0.4844,
+ "step": 8136
+ },
+ {
+ "epoch": 22.1716621253406,
+ "grad_norm": 6.56884241104126,
+ "learning_rate": 1.813341683309991e-05,
+ "loss": 0.4407,
+ "step": 8137
+ },
+ {
+ "epoch": 22.174386920980925,
+ "grad_norm": 8.589064598083496,
+ "learning_rate": 1.8132903378134764e-05,
+ "loss": 0.4876,
+ "step": 8138
+ },
+ {
+ "epoch": 22.177111716621255,
+ "grad_norm": 4.917667388916016,
+ "learning_rate": 1.8132389859830825e-05,
+ "loss": 0.5688,
+ "step": 8139
+ },
+ {
+ "epoch": 22.17983651226158,
+ "grad_norm": 5.67755651473999,
+ "learning_rate": 1.8131876278192097e-05,
+ "loss": 0.4678,
+ "step": 8140
+ },
+ {
+ "epoch": 22.182561307901906,
+ "grad_norm": 6.088091850280762,
+ "learning_rate": 1.8131362633222577e-05,
+ "loss": 0.5945,
+ "step": 8141
+ },
+ {
+ "epoch": 22.185286103542236,
+ "grad_norm": 6.500578880310059,
+ "learning_rate": 1.8130848924926264e-05,
+ "loss": 0.5139,
+ "step": 8142
+ },
+ {
+ "epoch": 22.18801089918256,
+ "grad_norm": 6.906770706176758,
+ "learning_rate": 1.8130335153307167e-05,
+ "loss": 0.4307,
+ "step": 8143
+ },
+ {
+ "epoch": 22.190735694822887,
+ "grad_norm": 5.142308712005615,
+ "learning_rate": 1.8129821318369278e-05,
+ "loss": 0.3801,
+ "step": 8144
+ },
+ {
+ "epoch": 22.193460490463217,
+ "grad_norm": 8.87263298034668,
+ "learning_rate": 1.81293074201166e-05,
+ "loss": 0.4671,
+ "step": 8145
+ },
+ {
+ "epoch": 22.196185286103542,
+ "grad_norm": 5.367218494415283,
+ "learning_rate": 1.812879345855314e-05,
+ "loss": 0.4236,
+ "step": 8146
+ },
+ {
+ "epoch": 22.19891008174387,
+ "grad_norm": 6.132066249847412,
+ "learning_rate": 1.8128279433682898e-05,
+ "loss": 0.5242,
+ "step": 8147
+ },
+ {
+ "epoch": 22.201634877384198,
+ "grad_norm": 6.4848952293396,
+ "learning_rate": 1.812776534550988e-05,
+ "loss": 0.5467,
+ "step": 8148
+ },
+ {
+ "epoch": 22.204359673024523,
+ "grad_norm": 5.534457683563232,
+ "learning_rate": 1.812725119403808e-05,
+ "loss": 0.5464,
+ "step": 8149
+ },
+ {
+ "epoch": 22.20708446866485,
+ "grad_norm": 6.114357948303223,
+ "learning_rate": 1.8126736979271516e-05,
+ "loss": 0.5015,
+ "step": 8150
+ },
+ {
+ "epoch": 22.20980926430518,
+ "grad_norm": 5.375624656677246,
+ "learning_rate": 1.8126222701214185e-05,
+ "loss": 0.5756,
+ "step": 8151
+ },
+ {
+ "epoch": 22.212534059945504,
+ "grad_norm": 5.585423946380615,
+ "learning_rate": 1.812570835987009e-05,
+ "loss": 0.5959,
+ "step": 8152
+ },
+ {
+ "epoch": 22.21525885558583,
+ "grad_norm": 6.273038387298584,
+ "learning_rate": 1.8125193955243244e-05,
+ "loss": 0.5856,
+ "step": 8153
+ },
+ {
+ "epoch": 22.21798365122616,
+ "grad_norm": 5.636498928070068,
+ "learning_rate": 1.8124679487337647e-05,
+ "loss": 0.5247,
+ "step": 8154
+ },
+ {
+ "epoch": 22.220708446866485,
+ "grad_norm": 5.462798118591309,
+ "learning_rate": 1.8124164956157306e-05,
+ "loss": 0.3804,
+ "step": 8155
+ },
+ {
+ "epoch": 22.22343324250681,
+ "grad_norm": 6.219504356384277,
+ "learning_rate": 1.8123650361706234e-05,
+ "loss": 0.3325,
+ "step": 8156
+ },
+ {
+ "epoch": 22.22615803814714,
+ "grad_norm": 6.102568626403809,
+ "learning_rate": 1.812313570398843e-05,
+ "loss": 0.6346,
+ "step": 8157
+ },
+ {
+ "epoch": 22.228882833787466,
+ "grad_norm": 5.644361972808838,
+ "learning_rate": 1.8122620983007908e-05,
+ "loss": 0.5297,
+ "step": 8158
+ },
+ {
+ "epoch": 22.231607629427792,
+ "grad_norm": 7.132186412811279,
+ "learning_rate": 1.8122106198768675e-05,
+ "loss": 0.5273,
+ "step": 8159
+ },
+ {
+ "epoch": 22.23433242506812,
+ "grad_norm": 6.802036762237549,
+ "learning_rate": 1.8121591351274743e-05,
+ "loss": 0.5205,
+ "step": 8160
+ },
+ {
+ "epoch": 22.237057220708447,
+ "grad_norm": 6.223399639129639,
+ "learning_rate": 1.8121076440530116e-05,
+ "loss": 0.5031,
+ "step": 8161
+ },
+ {
+ "epoch": 22.239782016348773,
+ "grad_norm": 17.52618980407715,
+ "learning_rate": 1.8120561466538806e-05,
+ "loss": 0.4538,
+ "step": 8162
+ },
+ {
+ "epoch": 22.242506811989102,
+ "grad_norm": 6.468094348907471,
+ "learning_rate": 1.8120046429304828e-05,
+ "loss": 0.3783,
+ "step": 8163
+ },
+ {
+ "epoch": 22.245231607629428,
+ "grad_norm": 6.594408988952637,
+ "learning_rate": 1.8119531328832186e-05,
+ "loss": 0.55,
+ "step": 8164
+ },
+ {
+ "epoch": 22.247956403269754,
+ "grad_norm": 7.806664943695068,
+ "learning_rate": 1.81190161651249e-05,
+ "loss": 0.4379,
+ "step": 8165
+ },
+ {
+ "epoch": 22.250681198910083,
+ "grad_norm": 5.836969375610352,
+ "learning_rate": 1.811850093818697e-05,
+ "loss": 0.4152,
+ "step": 8166
+ },
+ {
+ "epoch": 22.25340599455041,
+ "grad_norm": 5.012983322143555,
+ "learning_rate": 1.811798564802242e-05,
+ "loss": 0.605,
+ "step": 8167
+ },
+ {
+ "epoch": 22.256130790190735,
+ "grad_norm": 6.593403339385986,
+ "learning_rate": 1.8117470294635263e-05,
+ "loss": 0.4485,
+ "step": 8168
+ },
+ {
+ "epoch": 22.258855585831064,
+ "grad_norm": 5.899615287780762,
+ "learning_rate": 1.8116954878029505e-05,
+ "loss": 0.5354,
+ "step": 8169
+ },
+ {
+ "epoch": 22.26158038147139,
+ "grad_norm": 6.373673915863037,
+ "learning_rate": 1.8116439398209164e-05,
+ "loss": 0.4139,
+ "step": 8170
+ },
+ {
+ "epoch": 22.264305177111716,
+ "grad_norm": 6.67788553237915,
+ "learning_rate": 1.8115923855178253e-05,
+ "loss": 0.4377,
+ "step": 8171
+ },
+ {
+ "epoch": 22.267029972752045,
+ "grad_norm": 5.453061103820801,
+ "learning_rate": 1.8115408248940792e-05,
+ "loss": 0.4043,
+ "step": 8172
+ },
+ {
+ "epoch": 22.26975476839237,
+ "grad_norm": 5.577812194824219,
+ "learning_rate": 1.8114892579500792e-05,
+ "loss": 0.4975,
+ "step": 8173
+ },
+ {
+ "epoch": 22.272479564032697,
+ "grad_norm": 21.541675567626953,
+ "learning_rate": 1.811437684686227e-05,
+ "loss": 0.4288,
+ "step": 8174
+ },
+ {
+ "epoch": 22.275204359673026,
+ "grad_norm": 6.749484062194824,
+ "learning_rate": 1.8113861051029235e-05,
+ "loss": 0.5382,
+ "step": 8175
+ },
+ {
+ "epoch": 22.277929155313352,
+ "grad_norm": 5.651033401489258,
+ "learning_rate": 1.8113345192005718e-05,
+ "loss": 0.4503,
+ "step": 8176
+ },
+ {
+ "epoch": 22.280653950953678,
+ "grad_norm": 7.403512477874756,
+ "learning_rate": 1.8112829269795728e-05,
+ "loss": 0.5171,
+ "step": 8177
+ },
+ {
+ "epoch": 22.283378746594007,
+ "grad_norm": 8.63598918914795,
+ "learning_rate": 1.8112313284403287e-05,
+ "loss": 0.4468,
+ "step": 8178
+ },
+ {
+ "epoch": 22.286103542234333,
+ "grad_norm": 5.703848361968994,
+ "learning_rate": 1.8111797235832407e-05,
+ "loss": 0.4041,
+ "step": 8179
+ },
+ {
+ "epoch": 22.28882833787466,
+ "grad_norm": 7.1442108154296875,
+ "learning_rate": 1.8111281124087113e-05,
+ "loss": 0.7516,
+ "step": 8180
+ },
+ {
+ "epoch": 22.291553133514988,
+ "grad_norm": 7.106650352478027,
+ "learning_rate": 1.8110764949171423e-05,
+ "loss": 0.5577,
+ "step": 8181
+ },
+ {
+ "epoch": 22.294277929155314,
+ "grad_norm": 6.567619323730469,
+ "learning_rate": 1.811024871108936e-05,
+ "loss": 0.4413,
+ "step": 8182
+ },
+ {
+ "epoch": 22.29700272479564,
+ "grad_norm": 7.693620204925537,
+ "learning_rate": 1.8109732409844936e-05,
+ "loss": 0.6775,
+ "step": 8183
+ },
+ {
+ "epoch": 22.29972752043597,
+ "grad_norm": 5.038509368896484,
+ "learning_rate": 1.810921604544218e-05,
+ "loss": 0.5187,
+ "step": 8184
+ },
+ {
+ "epoch": 22.302452316076295,
+ "grad_norm": 5.689952850341797,
+ "learning_rate": 1.810869961788511e-05,
+ "loss": 0.4651,
+ "step": 8185
+ },
+ {
+ "epoch": 22.30517711171662,
+ "grad_norm": 5.655453205108643,
+ "learning_rate": 1.810818312717775e-05,
+ "loss": 0.3825,
+ "step": 8186
+ },
+ {
+ "epoch": 22.30790190735695,
+ "grad_norm": 7.581567764282227,
+ "learning_rate": 1.8107666573324113e-05,
+ "loss": 0.7438,
+ "step": 8187
+ },
+ {
+ "epoch": 22.310626702997276,
+ "grad_norm": 6.298895835876465,
+ "learning_rate": 1.810714995632824e-05,
+ "loss": 0.5629,
+ "step": 8188
+ },
+ {
+ "epoch": 22.3133514986376,
+ "grad_norm": 6.161152362823486,
+ "learning_rate": 1.8106633276194138e-05,
+ "loss": 0.4097,
+ "step": 8189
+ },
+ {
+ "epoch": 22.31607629427793,
+ "grad_norm": 7.818055629730225,
+ "learning_rate": 1.8106116532925835e-05,
+ "loss": 0.5541,
+ "step": 8190
+ },
+ {
+ "epoch": 22.318801089918257,
+ "grad_norm": 5.405686378479004,
+ "learning_rate": 1.8105599726527364e-05,
+ "loss": 0.5745,
+ "step": 8191
+ },
+ {
+ "epoch": 22.321525885558582,
+ "grad_norm": 7.048875331878662,
+ "learning_rate": 1.810508285700274e-05,
+ "loss": 0.4564,
+ "step": 8192
+ },
+ {
+ "epoch": 22.32425068119891,
+ "grad_norm": 6.550835609436035,
+ "learning_rate": 1.8104565924355987e-05,
+ "loss": 0.3931,
+ "step": 8193
+ },
+ {
+ "epoch": 22.326975476839237,
+ "grad_norm": 7.92356014251709,
+ "learning_rate": 1.810404892859114e-05,
+ "loss": 0.5366,
+ "step": 8194
+ },
+ {
+ "epoch": 22.329700272479563,
+ "grad_norm": 7.970541477203369,
+ "learning_rate": 1.8103531869712222e-05,
+ "loss": 0.4872,
+ "step": 8195
+ },
+ {
+ "epoch": 22.332425068119893,
+ "grad_norm": 7.237485885620117,
+ "learning_rate": 1.8103014747723258e-05,
+ "loss": 0.5832,
+ "step": 8196
+ },
+ {
+ "epoch": 22.33514986376022,
+ "grad_norm": 8.232694625854492,
+ "learning_rate": 1.8102497562628272e-05,
+ "loss": 0.5523,
+ "step": 8197
+ },
+ {
+ "epoch": 22.337874659400544,
+ "grad_norm": 17.874187469482422,
+ "learning_rate": 1.81019803144313e-05,
+ "loss": 0.5077,
+ "step": 8198
+ },
+ {
+ "epoch": 22.340599455040874,
+ "grad_norm": 5.849801540374756,
+ "learning_rate": 1.8101463003136365e-05,
+ "loss": 0.4286,
+ "step": 8199
+ },
+ {
+ "epoch": 22.3433242506812,
+ "grad_norm": 6.564208984375,
+ "learning_rate": 1.81009456287475e-05,
+ "loss": 0.4924,
+ "step": 8200
+ },
+ {
+ "epoch": 22.346049046321525,
+ "grad_norm": 5.699175834655762,
+ "learning_rate": 1.8100428191268724e-05,
+ "loss": 0.4433,
+ "step": 8201
+ },
+ {
+ "epoch": 22.348773841961854,
+ "grad_norm": 4.99918794631958,
+ "learning_rate": 1.809991069070408e-05,
+ "loss": 0.7114,
+ "step": 8202
+ },
+ {
+ "epoch": 22.35149863760218,
+ "grad_norm": 6.502413272857666,
+ "learning_rate": 1.809939312705759e-05,
+ "loss": 0.4559,
+ "step": 8203
+ },
+ {
+ "epoch": 22.354223433242506,
+ "grad_norm": 5.570622444152832,
+ "learning_rate": 1.8098875500333287e-05,
+ "loss": 0.7119,
+ "step": 8204
+ },
+ {
+ "epoch": 22.356948228882835,
+ "grad_norm": 6.257114410400391,
+ "learning_rate": 1.8098357810535204e-05,
+ "loss": 0.4516,
+ "step": 8205
+ },
+ {
+ "epoch": 22.35967302452316,
+ "grad_norm": 7.7145514488220215,
+ "learning_rate": 1.8097840057667366e-05,
+ "loss": 0.5105,
+ "step": 8206
+ },
+ {
+ "epoch": 22.362397820163487,
+ "grad_norm": 6.42124605178833,
+ "learning_rate": 1.8097322241733815e-05,
+ "loss": 0.3595,
+ "step": 8207
+ },
+ {
+ "epoch": 22.365122615803816,
+ "grad_norm": 7.352561950683594,
+ "learning_rate": 1.8096804362738577e-05,
+ "loss": 0.7229,
+ "step": 8208
+ },
+ {
+ "epoch": 22.367847411444142,
+ "grad_norm": 6.0485076904296875,
+ "learning_rate": 1.809628642068569e-05,
+ "loss": 0.5649,
+ "step": 8209
+ },
+ {
+ "epoch": 22.370572207084468,
+ "grad_norm": 6.0878586769104,
+ "learning_rate": 1.809576841557918e-05,
+ "loss": 0.4623,
+ "step": 8210
+ },
+ {
+ "epoch": 22.373297002724797,
+ "grad_norm": 5.611634254455566,
+ "learning_rate": 1.809525034742309e-05,
+ "loss": 0.6194,
+ "step": 8211
+ },
+ {
+ "epoch": 22.376021798365123,
+ "grad_norm": 6.603185653686523,
+ "learning_rate": 1.809473221622145e-05,
+ "loss": 0.5157,
+ "step": 8212
+ },
+ {
+ "epoch": 22.37874659400545,
+ "grad_norm": 5.519434452056885,
+ "learning_rate": 1.8094214021978295e-05,
+ "loss": 0.4111,
+ "step": 8213
+ },
+ {
+ "epoch": 22.381471389645778,
+ "grad_norm": 6.368724346160889,
+ "learning_rate": 1.8093695764697664e-05,
+ "loss": 0.474,
+ "step": 8214
+ },
+ {
+ "epoch": 22.384196185286104,
+ "grad_norm": 5.602844715118408,
+ "learning_rate": 1.8093177444383586e-05,
+ "loss": 0.6957,
+ "step": 8215
+ },
+ {
+ "epoch": 22.38692098092643,
+ "grad_norm": 5.553603649139404,
+ "learning_rate": 1.809265906104011e-05,
+ "loss": 0.6661,
+ "step": 8216
+ },
+ {
+ "epoch": 22.38964577656676,
+ "grad_norm": 6.849163055419922,
+ "learning_rate": 1.809214061467126e-05,
+ "loss": 0.4873,
+ "step": 8217
+ },
+ {
+ "epoch": 22.392370572207085,
+ "grad_norm": 6.851098537445068,
+ "learning_rate": 1.809162210528108e-05,
+ "loss": 0.5494,
+ "step": 8218
+ },
+ {
+ "epoch": 22.39509536784741,
+ "grad_norm": 6.5125861167907715,
+ "learning_rate": 1.809110353287361e-05,
+ "loss": 0.5214,
+ "step": 8219
+ },
+ {
+ "epoch": 22.39782016348774,
+ "grad_norm": 5.459994316101074,
+ "learning_rate": 1.8090584897452884e-05,
+ "loss": 0.4953,
+ "step": 8220
+ },
+ {
+ "epoch": 22.400544959128066,
+ "grad_norm": 6.080134391784668,
+ "learning_rate": 1.8090066199022946e-05,
+ "loss": 0.4996,
+ "step": 8221
+ },
+ {
+ "epoch": 22.40326975476839,
+ "grad_norm": 5.984493732452393,
+ "learning_rate": 1.8089547437587828e-05,
+ "loss": 0.4946,
+ "step": 8222
+ },
+ {
+ "epoch": 22.40599455040872,
+ "grad_norm": 6.213628768920898,
+ "learning_rate": 1.8089028613151578e-05,
+ "loss": 0.3862,
+ "step": 8223
+ },
+ {
+ "epoch": 22.408719346049047,
+ "grad_norm": 7.038131237030029,
+ "learning_rate": 1.808850972571823e-05,
+ "loss": 0.4329,
+ "step": 8224
+ },
+ {
+ "epoch": 22.411444141689373,
+ "grad_norm": 6.532097339630127,
+ "learning_rate": 1.8087990775291835e-05,
+ "loss": 0.5966,
+ "step": 8225
+ },
+ {
+ "epoch": 22.414168937329702,
+ "grad_norm": 5.885113716125488,
+ "learning_rate": 1.8087471761876425e-05,
+ "loss": 0.7897,
+ "step": 8226
+ },
+ {
+ "epoch": 22.416893732970028,
+ "grad_norm": 5.528205394744873,
+ "learning_rate": 1.8086952685476046e-05,
+ "loss": 0.4689,
+ "step": 8227
+ },
+ {
+ "epoch": 22.419618528610354,
+ "grad_norm": 6.1295084953308105,
+ "learning_rate": 1.808643354609474e-05,
+ "loss": 0.3901,
+ "step": 8228
+ },
+ {
+ "epoch": 22.422343324250683,
+ "grad_norm": 6.992239952087402,
+ "learning_rate": 1.8085914343736545e-05,
+ "loss": 0.5576,
+ "step": 8229
+ },
+ {
+ "epoch": 22.42506811989101,
+ "grad_norm": 6.2595624923706055,
+ "learning_rate": 1.8085395078405514e-05,
+ "loss": 0.4391,
+ "step": 8230
+ },
+ {
+ "epoch": 22.427792915531334,
+ "grad_norm": 179.20140075683594,
+ "learning_rate": 1.8084875750105685e-05,
+ "loss": 0.4467,
+ "step": 8231
+ },
+ {
+ "epoch": 22.430517711171664,
+ "grad_norm": 7.221296787261963,
+ "learning_rate": 1.8084356358841106e-05,
+ "loss": 0.5215,
+ "step": 8232
+ },
+ {
+ "epoch": 22.43324250681199,
+ "grad_norm": 7.443523406982422,
+ "learning_rate": 1.8083836904615815e-05,
+ "loss": 0.3937,
+ "step": 8233
+ },
+ {
+ "epoch": 22.435967302452315,
+ "grad_norm": 7.525096893310547,
+ "learning_rate": 1.8083317387433866e-05,
+ "loss": 0.3868,
+ "step": 8234
+ },
+ {
+ "epoch": 22.438692098092645,
+ "grad_norm": 10.935585975646973,
+ "learning_rate": 1.80827978072993e-05,
+ "loss": 0.6929,
+ "step": 8235
+ },
+ {
+ "epoch": 22.44141689373297,
+ "grad_norm": 8.14183521270752,
+ "learning_rate": 1.8082278164216166e-05,
+ "loss": 0.7593,
+ "step": 8236
+ },
+ {
+ "epoch": 22.444141689373296,
+ "grad_norm": 9.951396942138672,
+ "learning_rate": 1.808175845818851e-05,
+ "loss": 0.733,
+ "step": 8237
+ },
+ {
+ "epoch": 22.446866485013626,
+ "grad_norm": 11.301628112792969,
+ "learning_rate": 1.808123868922038e-05,
+ "loss": 0.5373,
+ "step": 8238
+ },
+ {
+ "epoch": 22.44959128065395,
+ "grad_norm": 13.336760520935059,
+ "learning_rate": 1.808071885731582e-05,
+ "loss": 0.715,
+ "step": 8239
+ },
+ {
+ "epoch": 22.452316076294277,
+ "grad_norm": 8.107128143310547,
+ "learning_rate": 1.8080198962478884e-05,
+ "loss": 0.5331,
+ "step": 8240
+ },
+ {
+ "epoch": 22.455040871934607,
+ "grad_norm": 8.255146980285645,
+ "learning_rate": 1.8079679004713618e-05,
+ "loss": 0.6222,
+ "step": 8241
+ },
+ {
+ "epoch": 22.457765667574932,
+ "grad_norm": 9.312801361083984,
+ "learning_rate": 1.8079158984024073e-05,
+ "loss": 0.5579,
+ "step": 8242
+ },
+ {
+ "epoch": 22.460490463215258,
+ "grad_norm": 13.008007049560547,
+ "learning_rate": 1.8078638900414292e-05,
+ "loss": 0.8069,
+ "step": 8243
+ },
+ {
+ "epoch": 22.463215258855588,
+ "grad_norm": 13.58984088897705,
+ "learning_rate": 1.8078118753888337e-05,
+ "loss": 0.5294,
+ "step": 8244
+ },
+ {
+ "epoch": 22.465940054495913,
+ "grad_norm": 8.426814079284668,
+ "learning_rate": 1.8077598544450253e-05,
+ "loss": 0.5095,
+ "step": 8245
+ },
+ {
+ "epoch": 22.46866485013624,
+ "grad_norm": 8.120780944824219,
+ "learning_rate": 1.8077078272104093e-05,
+ "loss": 0.6186,
+ "step": 8246
+ },
+ {
+ "epoch": 22.47138964577657,
+ "grad_norm": 8.36736011505127,
+ "learning_rate": 1.8076557936853905e-05,
+ "loss": 0.5867,
+ "step": 8247
+ },
+ {
+ "epoch": 22.474114441416894,
+ "grad_norm": 7.957752227783203,
+ "learning_rate": 1.807603753870374e-05,
+ "loss": 0.5263,
+ "step": 8248
+ },
+ {
+ "epoch": 22.47683923705722,
+ "grad_norm": 10.599550247192383,
+ "learning_rate": 1.807551707765766e-05,
+ "loss": 0.7098,
+ "step": 8249
+ },
+ {
+ "epoch": 22.479564032697546,
+ "grad_norm": 8.781487464904785,
+ "learning_rate": 1.807499655371971e-05,
+ "loss": 0.4863,
+ "step": 8250
+ },
+ {
+ "epoch": 22.482288828337875,
+ "grad_norm": 10.426389694213867,
+ "learning_rate": 1.807447596689395e-05,
+ "loss": 0.5001,
+ "step": 8251
+ },
+ {
+ "epoch": 22.4850136239782,
+ "grad_norm": 11.386076927185059,
+ "learning_rate": 1.807395531718443e-05,
+ "loss": 0.463,
+ "step": 8252
+ },
+ {
+ "epoch": 22.48773841961853,
+ "grad_norm": 9.302346229553223,
+ "learning_rate": 1.8073434604595203e-05,
+ "loss": 0.6196,
+ "step": 8253
+ },
+ {
+ "epoch": 22.490463215258856,
+ "grad_norm": 9.05832290649414,
+ "learning_rate": 1.8072913829130326e-05,
+ "loss": 0.6016,
+ "step": 8254
+ },
+ {
+ "epoch": 22.493188010899182,
+ "grad_norm": 7.459694862365723,
+ "learning_rate": 1.8072392990793864e-05,
+ "loss": 0.5218,
+ "step": 8255
+ },
+ {
+ "epoch": 22.495912806539508,
+ "grad_norm": 9.514650344848633,
+ "learning_rate": 1.8071872089589857e-05,
+ "loss": 0.5082,
+ "step": 8256
+ },
+ {
+ "epoch": 22.498637602179837,
+ "grad_norm": 8.406590461730957,
+ "learning_rate": 1.8071351125522372e-05,
+ "loss": 0.5548,
+ "step": 8257
+ },
+ {
+ "epoch": 22.501362397820163,
+ "grad_norm": 7.891927719116211,
+ "learning_rate": 1.8070830098595467e-05,
+ "loss": 0.508,
+ "step": 8258
+ },
+ {
+ "epoch": 22.504087193460492,
+ "grad_norm": 7.701966285705566,
+ "learning_rate": 1.8070309008813195e-05,
+ "loss": 0.7043,
+ "step": 8259
+ },
+ {
+ "epoch": 22.506811989100818,
+ "grad_norm": 7.538219451904297,
+ "learning_rate": 1.8069787856179618e-05,
+ "loss": 0.6061,
+ "step": 8260
+ },
+ {
+ "epoch": 22.509536784741144,
+ "grad_norm": 8.044837951660156,
+ "learning_rate": 1.806926664069879e-05,
+ "loss": 0.5685,
+ "step": 8261
+ },
+ {
+ "epoch": 22.51226158038147,
+ "grad_norm": 7.891442775726318,
+ "learning_rate": 1.8068745362374774e-05,
+ "loss": 0.5588,
+ "step": 8262
+ },
+ {
+ "epoch": 22.5149863760218,
+ "grad_norm": 7.312035083770752,
+ "learning_rate": 1.806822402121163e-05,
+ "loss": 0.6605,
+ "step": 8263
+ },
+ {
+ "epoch": 22.517711171662125,
+ "grad_norm": 7.770387172698975,
+ "learning_rate": 1.8067702617213415e-05,
+ "loss": 0.5394,
+ "step": 8264
+ },
+ {
+ "epoch": 22.520435967302454,
+ "grad_norm": 6.839579105377197,
+ "learning_rate": 1.8067181150384195e-05,
+ "loss": 0.5626,
+ "step": 8265
+ },
+ {
+ "epoch": 22.52316076294278,
+ "grad_norm": 7.967758655548096,
+ "learning_rate": 1.8066659620728023e-05,
+ "loss": 0.6058,
+ "step": 8266
+ },
+ {
+ "epoch": 22.525885558583106,
+ "grad_norm": 8.885454177856445,
+ "learning_rate": 1.806613802824897e-05,
+ "loss": 0.5887,
+ "step": 8267
+ },
+ {
+ "epoch": 22.52861035422343,
+ "grad_norm": 7.154987812042236,
+ "learning_rate": 1.8065616372951093e-05,
+ "loss": 0.577,
+ "step": 8268
+ },
+ {
+ "epoch": 22.53133514986376,
+ "grad_norm": 7.35183572769165,
+ "learning_rate": 1.8065094654838452e-05,
+ "loss": 0.6163,
+ "step": 8269
+ },
+ {
+ "epoch": 22.534059945504087,
+ "grad_norm": 6.463513374328613,
+ "learning_rate": 1.8064572873915117e-05,
+ "loss": 0.493,
+ "step": 8270
+ },
+ {
+ "epoch": 22.536784741144416,
+ "grad_norm": 8.285663604736328,
+ "learning_rate": 1.8064051030185145e-05,
+ "loss": 0.8926,
+ "step": 8271
+ },
+ {
+ "epoch": 22.539509536784742,
+ "grad_norm": 7.231849670410156,
+ "learning_rate": 1.8063529123652604e-05,
+ "loss": 0.533,
+ "step": 8272
+ },
+ {
+ "epoch": 22.542234332425068,
+ "grad_norm": 7.035409927368164,
+ "learning_rate": 1.8063007154321558e-05,
+ "loss": 0.5084,
+ "step": 8273
+ },
+ {
+ "epoch": 22.544959128065393,
+ "grad_norm": 8.199568748474121,
+ "learning_rate": 1.8062485122196074e-05,
+ "loss": 0.657,
+ "step": 8274
+ },
+ {
+ "epoch": 22.547683923705723,
+ "grad_norm": 7.243514060974121,
+ "learning_rate": 1.806196302728021e-05,
+ "loss": 0.6238,
+ "step": 8275
+ },
+ {
+ "epoch": 22.55040871934605,
+ "grad_norm": 6.895151615142822,
+ "learning_rate": 1.8061440869578043e-05,
+ "loss": 0.541,
+ "step": 8276
+ },
+ {
+ "epoch": 22.553133514986374,
+ "grad_norm": 8.077116966247559,
+ "learning_rate": 1.8060918649093634e-05,
+ "loss": 0.5078,
+ "step": 8277
+ },
+ {
+ "epoch": 22.555858310626704,
+ "grad_norm": 7.23660945892334,
+ "learning_rate": 1.8060396365831046e-05,
+ "loss": 0.532,
+ "step": 8278
+ },
+ {
+ "epoch": 22.55858310626703,
+ "grad_norm": 9.758158683776855,
+ "learning_rate": 1.8059874019794352e-05,
+ "loss": 0.5045,
+ "step": 8279
+ },
+ {
+ "epoch": 22.561307901907355,
+ "grad_norm": 8.4933500289917,
+ "learning_rate": 1.8059351610987617e-05,
+ "loss": 0.7649,
+ "step": 8280
+ },
+ {
+ "epoch": 22.564032697547685,
+ "grad_norm": 7.6448445320129395,
+ "learning_rate": 1.8058829139414915e-05,
+ "loss": 0.4773,
+ "step": 8281
+ },
+ {
+ "epoch": 22.56675749318801,
+ "grad_norm": 7.042953014373779,
+ "learning_rate": 1.805830660508031e-05,
+ "loss": 0.5839,
+ "step": 8282
+ },
+ {
+ "epoch": 22.569482288828336,
+ "grad_norm": 6.717342853546143,
+ "learning_rate": 1.805778400798787e-05,
+ "loss": 0.5673,
+ "step": 8283
+ },
+ {
+ "epoch": 22.572207084468666,
+ "grad_norm": 6.983149528503418,
+ "learning_rate": 1.805726134814167e-05,
+ "loss": 0.6405,
+ "step": 8284
+ },
+ {
+ "epoch": 22.57493188010899,
+ "grad_norm": 6.395967483520508,
+ "learning_rate": 1.8056738625545778e-05,
+ "loss": 0.5899,
+ "step": 8285
+ },
+ {
+ "epoch": 22.577656675749317,
+ "grad_norm": 7.932030200958252,
+ "learning_rate": 1.8056215840204262e-05,
+ "loss": 0.5816,
+ "step": 8286
+ },
+ {
+ "epoch": 22.580381471389646,
+ "grad_norm": 7.092746734619141,
+ "learning_rate": 1.80556929921212e-05,
+ "loss": 0.573,
+ "step": 8287
+ },
+ {
+ "epoch": 22.583106267029972,
+ "grad_norm": 6.6379218101501465,
+ "learning_rate": 1.8055170081300656e-05,
+ "loss": 0.3869,
+ "step": 8288
+ },
+ {
+ "epoch": 22.585831062670298,
+ "grad_norm": 7.532810211181641,
+ "learning_rate": 1.805464710774671e-05,
+ "loss": 0.6583,
+ "step": 8289
+ },
+ {
+ "epoch": 22.588555858310627,
+ "grad_norm": 6.313208103179932,
+ "learning_rate": 1.805412407146343e-05,
+ "loss": 0.532,
+ "step": 8290
+ },
+ {
+ "epoch": 22.591280653950953,
+ "grad_norm": 5.882076740264893,
+ "learning_rate": 1.805360097245489e-05,
+ "loss": 0.4964,
+ "step": 8291
+ },
+ {
+ "epoch": 22.59400544959128,
+ "grad_norm": 7.728720188140869,
+ "learning_rate": 1.8053077810725167e-05,
+ "loss": 0.5496,
+ "step": 8292
+ },
+ {
+ "epoch": 22.59673024523161,
+ "grad_norm": 7.4095072746276855,
+ "learning_rate": 1.8052554586278335e-05,
+ "loss": 0.5129,
+ "step": 8293
+ },
+ {
+ "epoch": 22.599455040871934,
+ "grad_norm": 8.038764953613281,
+ "learning_rate": 1.8052031299118465e-05,
+ "loss": 0.5312,
+ "step": 8294
+ },
+ {
+ "epoch": 22.60217983651226,
+ "grad_norm": 7.683620929718018,
+ "learning_rate": 1.805150794924963e-05,
+ "loss": 0.6431,
+ "step": 8295
+ },
+ {
+ "epoch": 22.60490463215259,
+ "grad_norm": 6.565217971801758,
+ "learning_rate": 1.8050984536675918e-05,
+ "loss": 0.5713,
+ "step": 8296
+ },
+ {
+ "epoch": 22.607629427792915,
+ "grad_norm": 7.774906158447266,
+ "learning_rate": 1.8050461061401394e-05,
+ "loss": 0.6,
+ "step": 8297
+ },
+ {
+ "epoch": 22.61035422343324,
+ "grad_norm": 7.082777500152588,
+ "learning_rate": 1.8049937523430138e-05,
+ "loss": 0.517,
+ "step": 8298
+ },
+ {
+ "epoch": 22.61307901907357,
+ "grad_norm": 7.010653972625732,
+ "learning_rate": 1.8049413922766227e-05,
+ "loss": 0.6608,
+ "step": 8299
+ },
+ {
+ "epoch": 22.615803814713896,
+ "grad_norm": 7.130710601806641,
+ "learning_rate": 1.804889025941374e-05,
+ "loss": 0.4719,
+ "step": 8300
+ },
+ {
+ "epoch": 22.618528610354222,
+ "grad_norm": 6.865055561065674,
+ "learning_rate": 1.8048366533376754e-05,
+ "loss": 0.4596,
+ "step": 8301
+ },
+ {
+ "epoch": 22.62125340599455,
+ "grad_norm": 7.552041053771973,
+ "learning_rate": 1.804784274465935e-05,
+ "loss": 0.7009,
+ "step": 8302
+ },
+ {
+ "epoch": 22.623978201634877,
+ "grad_norm": 6.218852519989014,
+ "learning_rate": 1.8047318893265606e-05,
+ "loss": 0.4914,
+ "step": 8303
+ },
+ {
+ "epoch": 22.626702997275203,
+ "grad_norm": 7.711087703704834,
+ "learning_rate": 1.80467949791996e-05,
+ "loss": 0.6098,
+ "step": 8304
+ },
+ {
+ "epoch": 22.629427792915532,
+ "grad_norm": 5.993838787078857,
+ "learning_rate": 1.8046271002465416e-05,
+ "loss": 0.6401,
+ "step": 8305
+ },
+ {
+ "epoch": 22.632152588555858,
+ "grad_norm": 6.546947479248047,
+ "learning_rate": 1.804574696306713e-05,
+ "loss": 0.5213,
+ "step": 8306
+ },
+ {
+ "epoch": 22.634877384196184,
+ "grad_norm": 6.3088884353637695,
+ "learning_rate": 1.8045222861008826e-05,
+ "loss": 0.4095,
+ "step": 8307
+ },
+ {
+ "epoch": 22.637602179836513,
+ "grad_norm": 5.701735019683838,
+ "learning_rate": 1.8044698696294586e-05,
+ "loss": 0.4384,
+ "step": 8308
+ },
+ {
+ "epoch": 22.64032697547684,
+ "grad_norm": 5.987392902374268,
+ "learning_rate": 1.804417446892849e-05,
+ "loss": 0.4576,
+ "step": 8309
+ },
+ {
+ "epoch": 22.643051771117165,
+ "grad_norm": 7.573123455047607,
+ "learning_rate": 1.8043650178914624e-05,
+ "loss": 0.639,
+ "step": 8310
+ },
+ {
+ "epoch": 22.645776566757494,
+ "grad_norm": 9.374713897705078,
+ "learning_rate": 1.8043125826257067e-05,
+ "loss": 0.5944,
+ "step": 8311
+ },
+ {
+ "epoch": 22.64850136239782,
+ "grad_norm": 6.850228309631348,
+ "learning_rate": 1.8042601410959907e-05,
+ "loss": 0.7288,
+ "step": 8312
+ },
+ {
+ "epoch": 22.651226158038146,
+ "grad_norm": 6.960007667541504,
+ "learning_rate": 1.8042076933027228e-05,
+ "loss": 0.5217,
+ "step": 8313
+ },
+ {
+ "epoch": 22.653950953678475,
+ "grad_norm": 6.281553268432617,
+ "learning_rate": 1.804155239246311e-05,
+ "loss": 0.5616,
+ "step": 8314
+ },
+ {
+ "epoch": 22.6566757493188,
+ "grad_norm": 6.733818054199219,
+ "learning_rate": 1.8041027789271642e-05,
+ "loss": 0.4652,
+ "step": 8315
+ },
+ {
+ "epoch": 22.659400544959126,
+ "grad_norm": 5.808813571929932,
+ "learning_rate": 1.8040503123456905e-05,
+ "loss": 0.6815,
+ "step": 8316
+ },
+ {
+ "epoch": 22.662125340599456,
+ "grad_norm": 7.375040054321289,
+ "learning_rate": 1.8039978395022993e-05,
+ "loss": 0.6473,
+ "step": 8317
+ },
+ {
+ "epoch": 22.66485013623978,
+ "grad_norm": 6.292185306549072,
+ "learning_rate": 1.8039453603973984e-05,
+ "loss": 0.5232,
+ "step": 8318
+ },
+ {
+ "epoch": 22.667574931880107,
+ "grad_norm": 6.994907855987549,
+ "learning_rate": 1.803892875031397e-05,
+ "loss": 0.5811,
+ "step": 8319
+ },
+ {
+ "epoch": 22.670299727520437,
+ "grad_norm": 6.422686576843262,
+ "learning_rate": 1.803840383404704e-05,
+ "loss": 0.6116,
+ "step": 8320
+ },
+ {
+ "epoch": 22.673024523160763,
+ "grad_norm": 6.523843765258789,
+ "learning_rate": 1.8037878855177276e-05,
+ "loss": 0.8618,
+ "step": 8321
+ },
+ {
+ "epoch": 22.67574931880109,
+ "grad_norm": 5.881405353546143,
+ "learning_rate": 1.8037353813708773e-05,
+ "loss": 0.5918,
+ "step": 8322
+ },
+ {
+ "epoch": 22.678474114441418,
+ "grad_norm": 5.949334144592285,
+ "learning_rate": 1.8036828709645616e-05,
+ "loss": 0.4604,
+ "step": 8323
+ },
+ {
+ "epoch": 22.681198910081743,
+ "grad_norm": 5.819462299346924,
+ "learning_rate": 1.8036303542991898e-05,
+ "loss": 0.4788,
+ "step": 8324
+ },
+ {
+ "epoch": 22.68392370572207,
+ "grad_norm": 5.578276634216309,
+ "learning_rate": 1.8035778313751704e-05,
+ "loss": 0.5748,
+ "step": 8325
+ },
+ {
+ "epoch": 22.6866485013624,
+ "grad_norm": 7.244131565093994,
+ "learning_rate": 1.803525302192913e-05,
+ "loss": 0.6348,
+ "step": 8326
+ },
+ {
+ "epoch": 22.689373297002724,
+ "grad_norm": 5.688049793243408,
+ "learning_rate": 1.803472766752826e-05,
+ "loss": 0.5269,
+ "step": 8327
+ },
+ {
+ "epoch": 22.69209809264305,
+ "grad_norm": 6.169025421142578,
+ "learning_rate": 1.8034202250553193e-05,
+ "loss": 0.5794,
+ "step": 8328
+ },
+ {
+ "epoch": 22.69482288828338,
+ "grad_norm": 6.552395343780518,
+ "learning_rate": 1.8033676771008014e-05,
+ "loss": 0.594,
+ "step": 8329
+ },
+ {
+ "epoch": 22.697547683923705,
+ "grad_norm": 7.823551177978516,
+ "learning_rate": 1.803315122889682e-05,
+ "loss": 0.457,
+ "step": 8330
+ },
+ {
+ "epoch": 22.70027247956403,
+ "grad_norm": 6.949804306030273,
+ "learning_rate": 1.8032625624223708e-05,
+ "loss": 0.6165,
+ "step": 8331
+ },
+ {
+ "epoch": 22.70299727520436,
+ "grad_norm": 7.676548004150391,
+ "learning_rate": 1.803209995699276e-05,
+ "loss": 0.4224,
+ "step": 8332
+ },
+ {
+ "epoch": 22.705722070844686,
+ "grad_norm": 5.827585220336914,
+ "learning_rate": 1.8031574227208078e-05,
+ "loss": 0.3958,
+ "step": 8333
+ },
+ {
+ "epoch": 22.708446866485012,
+ "grad_norm": 6.929993152618408,
+ "learning_rate": 1.8031048434873754e-05,
+ "loss": 0.6976,
+ "step": 8334
+ },
+ {
+ "epoch": 22.71117166212534,
+ "grad_norm": 6.310476303100586,
+ "learning_rate": 1.8030522579993884e-05,
+ "loss": 0.4771,
+ "step": 8335
+ },
+ {
+ "epoch": 22.713896457765667,
+ "grad_norm": 7.199219703674316,
+ "learning_rate": 1.802999666257256e-05,
+ "loss": 0.4531,
+ "step": 8336
+ },
+ {
+ "epoch": 22.716621253405993,
+ "grad_norm": 7.139401912689209,
+ "learning_rate": 1.8029470682613886e-05,
+ "loss": 0.6378,
+ "step": 8337
+ },
+ {
+ "epoch": 22.719346049046322,
+ "grad_norm": 6.2556610107421875,
+ "learning_rate": 1.8028944640121948e-05,
+ "loss": 0.5284,
+ "step": 8338
+ },
+ {
+ "epoch": 22.722070844686648,
+ "grad_norm": 6.484303951263428,
+ "learning_rate": 1.8028418535100848e-05,
+ "loss": 0.5298,
+ "step": 8339
+ },
+ {
+ "epoch": 22.724795640326974,
+ "grad_norm": 8.00982666015625,
+ "learning_rate": 1.8027892367554686e-05,
+ "loss": 0.4662,
+ "step": 8340
+ },
+ {
+ "epoch": 22.727520435967303,
+ "grad_norm": 9.428058624267578,
+ "learning_rate": 1.8027366137487555e-05,
+ "loss": 0.5162,
+ "step": 8341
+ },
+ {
+ "epoch": 22.73024523160763,
+ "grad_norm": 5.954753875732422,
+ "learning_rate": 1.8026839844903552e-05,
+ "loss": 0.4494,
+ "step": 8342
+ },
+ {
+ "epoch": 22.732970027247955,
+ "grad_norm": 6.732738971710205,
+ "learning_rate": 1.8026313489806783e-05,
+ "loss": 0.4738,
+ "step": 8343
+ },
+ {
+ "epoch": 22.735694822888284,
+ "grad_norm": 7.650108814239502,
+ "learning_rate": 1.802578707220134e-05,
+ "loss": 0.4681,
+ "step": 8344
+ },
+ {
+ "epoch": 22.73841961852861,
+ "grad_norm": 5.561438083648682,
+ "learning_rate": 1.8025260592091326e-05,
+ "loss": 0.6804,
+ "step": 8345
+ },
+ {
+ "epoch": 22.741144414168936,
+ "grad_norm": 6.2208685874938965,
+ "learning_rate": 1.8024734049480843e-05,
+ "loss": 0.5287,
+ "step": 8346
+ },
+ {
+ "epoch": 22.743869209809265,
+ "grad_norm": 6.659306049346924,
+ "learning_rate": 1.8024207444373988e-05,
+ "loss": 0.5878,
+ "step": 8347
+ },
+ {
+ "epoch": 22.74659400544959,
+ "grad_norm": 5.9798688888549805,
+ "learning_rate": 1.802368077677486e-05,
+ "loss": 0.46,
+ "step": 8348
+ },
+ {
+ "epoch": 22.749318801089917,
+ "grad_norm": 8.776793479919434,
+ "learning_rate": 1.802315404668757e-05,
+ "loss": 0.6202,
+ "step": 8349
+ },
+ {
+ "epoch": 22.752043596730246,
+ "grad_norm": 8.000130653381348,
+ "learning_rate": 1.802262725411621e-05,
+ "loss": 0.5592,
+ "step": 8350
+ },
+ {
+ "epoch": 22.754768392370572,
+ "grad_norm": 7.564798355102539,
+ "learning_rate": 1.802210039906489e-05,
+ "loss": 0.5742,
+ "step": 8351
+ },
+ {
+ "epoch": 22.757493188010898,
+ "grad_norm": 5.547773838043213,
+ "learning_rate": 1.8021573481537712e-05,
+ "loss": 0.64,
+ "step": 8352
+ },
+ {
+ "epoch": 22.760217983651227,
+ "grad_norm": 17.832874298095703,
+ "learning_rate": 1.8021046501538774e-05,
+ "loss": 0.5108,
+ "step": 8353
+ },
+ {
+ "epoch": 22.762942779291553,
+ "grad_norm": 5.810003757476807,
+ "learning_rate": 1.8020519459072186e-05,
+ "loss": 0.4972,
+ "step": 8354
+ },
+ {
+ "epoch": 22.76566757493188,
+ "grad_norm": 7.060512065887451,
+ "learning_rate": 1.801999235414205e-05,
+ "loss": 0.6761,
+ "step": 8355
+ },
+ {
+ "epoch": 22.768392370572208,
+ "grad_norm": 6.724385738372803,
+ "learning_rate": 1.801946518675247e-05,
+ "loss": 0.5965,
+ "step": 8356
+ },
+ {
+ "epoch": 22.771117166212534,
+ "grad_norm": 6.460507392883301,
+ "learning_rate": 1.8018937956907556e-05,
+ "loss": 0.5924,
+ "step": 8357
+ },
+ {
+ "epoch": 22.77384196185286,
+ "grad_norm": 5.325287818908691,
+ "learning_rate": 1.8018410664611413e-05,
+ "loss": 0.6364,
+ "step": 8358
+ },
+ {
+ "epoch": 22.77656675749319,
+ "grad_norm": 5.724000930786133,
+ "learning_rate": 1.801788330986814e-05,
+ "loss": 0.6128,
+ "step": 8359
+ },
+ {
+ "epoch": 22.779291553133515,
+ "grad_norm": 6.648040771484375,
+ "learning_rate": 1.8017355892681855e-05,
+ "loss": 0.4785,
+ "step": 8360
+ },
+ {
+ "epoch": 22.78201634877384,
+ "grad_norm": 7.602153778076172,
+ "learning_rate": 1.8016828413056657e-05,
+ "loss": 0.428,
+ "step": 8361
+ },
+ {
+ "epoch": 22.78474114441417,
+ "grad_norm": 5.764381408691406,
+ "learning_rate": 1.8016300870996657e-05,
+ "loss": 0.5455,
+ "step": 8362
+ },
+ {
+ "epoch": 22.787465940054496,
+ "grad_norm": 5.9340410232543945,
+ "learning_rate": 1.8015773266505968e-05,
+ "loss": 0.5135,
+ "step": 8363
+ },
+ {
+ "epoch": 22.79019073569482,
+ "grad_norm": 5.69059944152832,
+ "learning_rate": 1.801524559958869e-05,
+ "loss": 0.5939,
+ "step": 8364
+ },
+ {
+ "epoch": 22.79291553133515,
+ "grad_norm": 6.490854263305664,
+ "learning_rate": 1.801471787024894e-05,
+ "loss": 0.5928,
+ "step": 8365
+ },
+ {
+ "epoch": 22.795640326975477,
+ "grad_norm": 7.228536128997803,
+ "learning_rate": 1.8014190078490827e-05,
+ "loss": 0.6299,
+ "step": 8366
+ },
+ {
+ "epoch": 22.798365122615802,
+ "grad_norm": 5.796403408050537,
+ "learning_rate": 1.8013662224318454e-05,
+ "loss": 0.4331,
+ "step": 8367
+ },
+ {
+ "epoch": 22.80108991825613,
+ "grad_norm": 6.551239967346191,
+ "learning_rate": 1.801313430773594e-05,
+ "loss": 0.5572,
+ "step": 8368
+ },
+ {
+ "epoch": 22.803814713896458,
+ "grad_norm": 8.039823532104492,
+ "learning_rate": 1.8012606328747396e-05,
+ "loss": 0.5786,
+ "step": 8369
+ },
+ {
+ "epoch": 22.806539509536783,
+ "grad_norm": 6.096043109893799,
+ "learning_rate": 1.801207828735693e-05,
+ "loss": 0.4696,
+ "step": 8370
+ },
+ {
+ "epoch": 22.809264305177113,
+ "grad_norm": 6.935184478759766,
+ "learning_rate": 1.801155018356866e-05,
+ "loss": 0.7114,
+ "step": 8371
+ },
+ {
+ "epoch": 22.81198910081744,
+ "grad_norm": 6.293839931488037,
+ "learning_rate": 1.801102201738669e-05,
+ "loss": 0.592,
+ "step": 8372
+ },
+ {
+ "epoch": 22.814713896457764,
+ "grad_norm": 5.9083757400512695,
+ "learning_rate": 1.801049378881514e-05,
+ "loss": 0.6953,
+ "step": 8373
+ },
+ {
+ "epoch": 22.817438692098094,
+ "grad_norm": 7.007228851318359,
+ "learning_rate": 1.8009965497858122e-05,
+ "loss": 0.6986,
+ "step": 8374
+ },
+ {
+ "epoch": 22.82016348773842,
+ "grad_norm": 5.931309223175049,
+ "learning_rate": 1.8009437144519754e-05,
+ "loss": 0.5048,
+ "step": 8375
+ },
+ {
+ "epoch": 22.822888283378745,
+ "grad_norm": 8.787540435791016,
+ "learning_rate": 1.800890872880414e-05,
+ "loss": 0.5601,
+ "step": 8376
+ },
+ {
+ "epoch": 22.825613079019075,
+ "grad_norm": 6.320806980133057,
+ "learning_rate": 1.8008380250715407e-05,
+ "loss": 0.5872,
+ "step": 8377
+ },
+ {
+ "epoch": 22.8283378746594,
+ "grad_norm": 6.035807132720947,
+ "learning_rate": 1.8007851710257665e-05,
+ "loss": 0.5808,
+ "step": 8378
+ },
+ {
+ "epoch": 22.831062670299726,
+ "grad_norm": 6.12576150894165,
+ "learning_rate": 1.8007323107435036e-05,
+ "loss": 0.5607,
+ "step": 8379
+ },
+ {
+ "epoch": 22.833787465940055,
+ "grad_norm": 6.336933612823486,
+ "learning_rate": 1.8006794442251628e-05,
+ "loss": 0.4505,
+ "step": 8380
+ },
+ {
+ "epoch": 22.83651226158038,
+ "grad_norm": 6.15171480178833,
+ "learning_rate": 1.8006265714711563e-05,
+ "loss": 0.5301,
+ "step": 8381
+ },
+ {
+ "epoch": 22.839237057220707,
+ "grad_norm": 5.886575222015381,
+ "learning_rate": 1.800573692481896e-05,
+ "loss": 0.5269,
+ "step": 8382
+ },
+ {
+ "epoch": 22.841961852861036,
+ "grad_norm": 6.686758995056152,
+ "learning_rate": 1.8005208072577934e-05,
+ "loss": 0.5745,
+ "step": 8383
+ },
+ {
+ "epoch": 22.844686648501362,
+ "grad_norm": 6.302023410797119,
+ "learning_rate": 1.8004679157992603e-05,
+ "loss": 0.5023,
+ "step": 8384
+ },
+ {
+ "epoch": 22.847411444141688,
+ "grad_norm": 5.289888381958008,
+ "learning_rate": 1.8004150181067092e-05,
+ "loss": 0.3955,
+ "step": 8385
+ },
+ {
+ "epoch": 22.850136239782017,
+ "grad_norm": 6.396733283996582,
+ "learning_rate": 1.8003621141805515e-05,
+ "loss": 0.678,
+ "step": 8386
+ },
+ {
+ "epoch": 22.852861035422343,
+ "grad_norm": 6.023980140686035,
+ "learning_rate": 1.8003092040211997e-05,
+ "loss": 0.5513,
+ "step": 8387
+ },
+ {
+ "epoch": 22.85558583106267,
+ "grad_norm": 8.331377983093262,
+ "learning_rate": 1.8002562876290653e-05,
+ "loss": 0.5518,
+ "step": 8388
+ },
+ {
+ "epoch": 22.858310626703,
+ "grad_norm": 8.36719799041748,
+ "learning_rate": 1.8002033650045608e-05,
+ "loss": 0.6078,
+ "step": 8389
+ },
+ {
+ "epoch": 22.861035422343324,
+ "grad_norm": 5.00295877456665,
+ "learning_rate": 1.800150436148098e-05,
+ "loss": 0.6407,
+ "step": 8390
+ },
+ {
+ "epoch": 22.86376021798365,
+ "grad_norm": 6.039684295654297,
+ "learning_rate": 1.8000975010600896e-05,
+ "loss": 0.4854,
+ "step": 8391
+ },
+ {
+ "epoch": 22.86648501362398,
+ "grad_norm": 7.02407693862915,
+ "learning_rate": 1.8000445597409476e-05,
+ "loss": 0.5603,
+ "step": 8392
+ },
+ {
+ "epoch": 22.869209809264305,
+ "grad_norm": 6.4965925216674805,
+ "learning_rate": 1.7999916121910844e-05,
+ "loss": 0.6829,
+ "step": 8393
+ },
+ {
+ "epoch": 22.87193460490463,
+ "grad_norm": 5.936263561248779,
+ "learning_rate": 1.799938658410912e-05,
+ "loss": 0.4778,
+ "step": 8394
+ },
+ {
+ "epoch": 22.87465940054496,
+ "grad_norm": 6.786654472351074,
+ "learning_rate": 1.7998856984008435e-05,
+ "loss": 0.4206,
+ "step": 8395
+ },
+ {
+ "epoch": 22.877384196185286,
+ "grad_norm": 5.670699596405029,
+ "learning_rate": 1.799832732161291e-05,
+ "loss": 0.5476,
+ "step": 8396
+ },
+ {
+ "epoch": 22.88010899182561,
+ "grad_norm": 7.2859392166137695,
+ "learning_rate": 1.7997797596926663e-05,
+ "loss": 0.5105,
+ "step": 8397
+ },
+ {
+ "epoch": 22.88283378746594,
+ "grad_norm": 7.786726951599121,
+ "learning_rate": 1.799726780995383e-05,
+ "loss": 0.4612,
+ "step": 8398
+ },
+ {
+ "epoch": 22.885558583106267,
+ "grad_norm": 5.828619956970215,
+ "learning_rate": 1.799673796069853e-05,
+ "loss": 0.4897,
+ "step": 8399
+ },
+ {
+ "epoch": 22.888283378746593,
+ "grad_norm": 6.947359085083008,
+ "learning_rate": 1.7996208049164893e-05,
+ "loss": 0.6434,
+ "step": 8400
+ },
+ {
+ "epoch": 22.891008174386922,
+ "grad_norm": 6.970804214477539,
+ "learning_rate": 1.7995678075357046e-05,
+ "loss": 0.7255,
+ "step": 8401
+ },
+ {
+ "epoch": 22.893732970027248,
+ "grad_norm": 8.23088550567627,
+ "learning_rate": 1.7995148039279114e-05,
+ "loss": 0.4154,
+ "step": 8402
+ },
+ {
+ "epoch": 22.896457765667574,
+ "grad_norm": 7.036977767944336,
+ "learning_rate": 1.799461794093523e-05,
+ "loss": 0.5821,
+ "step": 8403
+ },
+ {
+ "epoch": 22.899182561307903,
+ "grad_norm": 6.535486221313477,
+ "learning_rate": 1.7994087780329515e-05,
+ "loss": 0.5435,
+ "step": 8404
+ },
+ {
+ "epoch": 22.90190735694823,
+ "grad_norm": 7.891038417816162,
+ "learning_rate": 1.79935575574661e-05,
+ "loss": 0.4809,
+ "step": 8405
+ },
+ {
+ "epoch": 22.904632152588555,
+ "grad_norm": 7.139584541320801,
+ "learning_rate": 1.7993027272349122e-05,
+ "loss": 0.5738,
+ "step": 8406
+ },
+ {
+ "epoch": 22.907356948228884,
+ "grad_norm": 7.118578910827637,
+ "learning_rate": 1.79924969249827e-05,
+ "loss": 0.465,
+ "step": 8407
+ },
+ {
+ "epoch": 22.91008174386921,
+ "grad_norm": 5.711153507232666,
+ "learning_rate": 1.799196651537097e-05,
+ "loss": 0.5887,
+ "step": 8408
+ },
+ {
+ "epoch": 22.912806539509535,
+ "grad_norm": 5.376428604125977,
+ "learning_rate": 1.7991436043518063e-05,
+ "loss": 0.5421,
+ "step": 8409
+ },
+ {
+ "epoch": 22.915531335149865,
+ "grad_norm": 6.1056718826293945,
+ "learning_rate": 1.7990905509428113e-05,
+ "loss": 0.6903,
+ "step": 8410
+ },
+ {
+ "epoch": 22.91825613079019,
+ "grad_norm": 8.252373695373535,
+ "learning_rate": 1.7990374913105243e-05,
+ "loss": 0.551,
+ "step": 8411
+ },
+ {
+ "epoch": 22.920980926430516,
+ "grad_norm": 7.100516319274902,
+ "learning_rate": 1.798984425455359e-05,
+ "loss": 0.552,
+ "step": 8412
+ },
+ {
+ "epoch": 22.923705722070846,
+ "grad_norm": 5.6428985595703125,
+ "learning_rate": 1.798931353377729e-05,
+ "loss": 0.4525,
+ "step": 8413
+ },
+ {
+ "epoch": 22.92643051771117,
+ "grad_norm": 9.856022834777832,
+ "learning_rate": 1.7988782750780474e-05,
+ "loss": 0.5928,
+ "step": 8414
+ },
+ {
+ "epoch": 22.929155313351497,
+ "grad_norm": 5.398067951202393,
+ "learning_rate": 1.7988251905567272e-05,
+ "loss": 0.3996,
+ "step": 8415
+ },
+ {
+ "epoch": 22.931880108991827,
+ "grad_norm": 7.281002998352051,
+ "learning_rate": 1.798772099814182e-05,
+ "loss": 0.402,
+ "step": 8416
+ },
+ {
+ "epoch": 22.934604904632153,
+ "grad_norm": 6.432676792144775,
+ "learning_rate": 1.7987190028508258e-05,
+ "loss": 0.6481,
+ "step": 8417
+ },
+ {
+ "epoch": 22.93732970027248,
+ "grad_norm": 7.606274127960205,
+ "learning_rate": 1.7986658996670716e-05,
+ "loss": 0.4348,
+ "step": 8418
+ },
+ {
+ "epoch": 22.940054495912808,
+ "grad_norm": 6.298891067504883,
+ "learning_rate": 1.798612790263333e-05,
+ "loss": 0.5649,
+ "step": 8419
+ },
+ {
+ "epoch": 22.942779291553133,
+ "grad_norm": 5.7977776527404785,
+ "learning_rate": 1.7985596746400234e-05,
+ "loss": 0.5222,
+ "step": 8420
+ },
+ {
+ "epoch": 22.94550408719346,
+ "grad_norm": 5.932393550872803,
+ "learning_rate": 1.7985065527975574e-05,
+ "loss": 0.6252,
+ "step": 8421
+ },
+ {
+ "epoch": 22.94822888283379,
+ "grad_norm": 6.647258758544922,
+ "learning_rate": 1.798453424736348e-05,
+ "loss": 0.5156,
+ "step": 8422
+ },
+ {
+ "epoch": 22.950953678474114,
+ "grad_norm": 6.904192924499512,
+ "learning_rate": 1.7984002904568085e-05,
+ "loss": 0.5181,
+ "step": 8423
+ },
+ {
+ "epoch": 22.95367847411444,
+ "grad_norm": 6.2996931076049805,
+ "learning_rate": 1.7983471499593536e-05,
+ "loss": 0.5929,
+ "step": 8424
+ },
+ {
+ "epoch": 22.95640326975477,
+ "grad_norm": 6.929685115814209,
+ "learning_rate": 1.7982940032443967e-05,
+ "loss": 0.5581,
+ "step": 8425
+ },
+ {
+ "epoch": 22.959128065395095,
+ "grad_norm": 5.348496913909912,
+ "learning_rate": 1.7982408503123518e-05,
+ "loss": 0.4612,
+ "step": 8426
+ },
+ {
+ "epoch": 22.96185286103542,
+ "grad_norm": 5.900527477264404,
+ "learning_rate": 1.7981876911636325e-05,
+ "loss": 0.504,
+ "step": 8427
+ },
+ {
+ "epoch": 22.96457765667575,
+ "grad_norm": 6.055607795715332,
+ "learning_rate": 1.7981345257986537e-05,
+ "loss": 0.5537,
+ "step": 8428
+ },
+ {
+ "epoch": 22.967302452316076,
+ "grad_norm": 6.2587456703186035,
+ "learning_rate": 1.798081354217829e-05,
+ "loss": 0.5735,
+ "step": 8429
+ },
+ {
+ "epoch": 22.970027247956402,
+ "grad_norm": 5.79225492477417,
+ "learning_rate": 1.7980281764215716e-05,
+ "loss": 0.5574,
+ "step": 8430
+ },
+ {
+ "epoch": 22.97275204359673,
+ "grad_norm": 7.906903266906738,
+ "learning_rate": 1.797974992410297e-05,
+ "loss": 0.7552,
+ "step": 8431
+ },
+ {
+ "epoch": 22.975476839237057,
+ "grad_norm": 6.491408348083496,
+ "learning_rate": 1.797921802184419e-05,
+ "loss": 0.4931,
+ "step": 8432
+ },
+ {
+ "epoch": 22.978201634877383,
+ "grad_norm": 6.00089693069458,
+ "learning_rate": 1.7978686057443514e-05,
+ "loss": 0.5513,
+ "step": 8433
+ },
+ {
+ "epoch": 22.980926430517712,
+ "grad_norm": 10.86459732055664,
+ "learning_rate": 1.797815403090509e-05,
+ "loss": 0.5257,
+ "step": 8434
+ },
+ {
+ "epoch": 22.983651226158038,
+ "grad_norm": 6.612439155578613,
+ "learning_rate": 1.7977621942233057e-05,
+ "loss": 0.7601,
+ "step": 8435
+ },
+ {
+ "epoch": 22.986376021798364,
+ "grad_norm": 6.744450092315674,
+ "learning_rate": 1.797708979143156e-05,
+ "loss": 0.6495,
+ "step": 8436
+ },
+ {
+ "epoch": 22.989100817438693,
+ "grad_norm": 5.4789299964904785,
+ "learning_rate": 1.797655757850475e-05,
+ "loss": 0.5673,
+ "step": 8437
+ },
+ {
+ "epoch": 22.99182561307902,
+ "grad_norm": 7.1541032791137695,
+ "learning_rate": 1.797602530345676e-05,
+ "loss": 0.4822,
+ "step": 8438
+ },
+ {
+ "epoch": 22.994550408719345,
+ "grad_norm": 8.174612998962402,
+ "learning_rate": 1.7975492966291743e-05,
+ "loss": 0.6094,
+ "step": 8439
+ },
+ {
+ "epoch": 22.997275204359674,
+ "grad_norm": 7.121578216552734,
+ "learning_rate": 1.7974960567013844e-05,
+ "loss": 0.5688,
+ "step": 8440
+ },
+ {
+ "epoch": 23.0,
+ "grad_norm": 4.902931213378906,
+ "learning_rate": 1.797442810562721e-05,
+ "loss": 0.3458,
+ "step": 8441
+ },
+ {
+ "epoch": 23.002724795640326,
+ "grad_norm": 4.319786071777344,
+ "learning_rate": 1.7973895582135984e-05,
+ "loss": 0.4641,
+ "step": 8442
+ },
+ {
+ "epoch": 23.005449591280655,
+ "grad_norm": 6.006614685058594,
+ "learning_rate": 1.7973362996544318e-05,
+ "loss": 0.3728,
+ "step": 8443
+ },
+ {
+ "epoch": 23.00817438692098,
+ "grad_norm": 5.994820594787598,
+ "learning_rate": 1.7972830348856358e-05,
+ "loss": 0.4102,
+ "step": 8444
+ },
+ {
+ "epoch": 23.010899182561307,
+ "grad_norm": 6.196841716766357,
+ "learning_rate": 1.797229763907625e-05,
+ "loss": 0.4537,
+ "step": 8445
+ },
+ {
+ "epoch": 23.013623978201636,
+ "grad_norm": 5.151231288909912,
+ "learning_rate": 1.797176486720815e-05,
+ "loss": 0.5126,
+ "step": 8446
+ },
+ {
+ "epoch": 23.016348773841962,
+ "grad_norm": 6.709209442138672,
+ "learning_rate": 1.7971232033256197e-05,
+ "loss": 0.3504,
+ "step": 8447
+ },
+ {
+ "epoch": 23.019073569482288,
+ "grad_norm": 5.3690290451049805,
+ "learning_rate": 1.7970699137224543e-05,
+ "loss": 0.5491,
+ "step": 8448
+ },
+ {
+ "epoch": 23.021798365122617,
+ "grad_norm": 6.460380554199219,
+ "learning_rate": 1.7970166179117344e-05,
+ "loss": 0.6218,
+ "step": 8449
+ },
+ {
+ "epoch": 23.024523160762943,
+ "grad_norm": 6.977889537811279,
+ "learning_rate": 1.7969633158938745e-05,
+ "loss": 0.486,
+ "step": 8450
+ },
+ {
+ "epoch": 23.02724795640327,
+ "grad_norm": 5.438969612121582,
+ "learning_rate": 1.79691000766929e-05,
+ "loss": 0.4887,
+ "step": 8451
+ },
+ {
+ "epoch": 23.029972752043598,
+ "grad_norm": 4.784780502319336,
+ "learning_rate": 1.7968566932383968e-05,
+ "loss": 0.4118,
+ "step": 8452
+ },
+ {
+ "epoch": 23.032697547683924,
+ "grad_norm": 6.972580432891846,
+ "learning_rate": 1.7968033726016085e-05,
+ "loss": 0.4331,
+ "step": 8453
+ },
+ {
+ "epoch": 23.03542234332425,
+ "grad_norm": 6.082568645477295,
+ "learning_rate": 1.7967500457593412e-05,
+ "loss": 0.5016,
+ "step": 8454
+ },
+ {
+ "epoch": 23.03814713896458,
+ "grad_norm": 5.4815754890441895,
+ "learning_rate": 1.79669671271201e-05,
+ "loss": 0.5278,
+ "step": 8455
+ },
+ {
+ "epoch": 23.040871934604905,
+ "grad_norm": 5.149569988250732,
+ "learning_rate": 1.796643373460031e-05,
+ "loss": 0.434,
+ "step": 8456
+ },
+ {
+ "epoch": 23.04359673024523,
+ "grad_norm": 6.47538948059082,
+ "learning_rate": 1.796590028003819e-05,
+ "loss": 0.4598,
+ "step": 8457
+ },
+ {
+ "epoch": 23.04632152588556,
+ "grad_norm": 5.371490955352783,
+ "learning_rate": 1.7965366763437893e-05,
+ "loss": 0.5008,
+ "step": 8458
+ },
+ {
+ "epoch": 23.049046321525886,
+ "grad_norm": 6.860915184020996,
+ "learning_rate": 1.7964833184803576e-05,
+ "loss": 0.4267,
+ "step": 8459
+ },
+ {
+ "epoch": 23.05177111716621,
+ "grad_norm": 4.436588764190674,
+ "learning_rate": 1.7964299544139394e-05,
+ "loss": 0.4087,
+ "step": 8460
+ },
+ {
+ "epoch": 23.05449591280654,
+ "grad_norm": 5.581408977508545,
+ "learning_rate": 1.7963765841449506e-05,
+ "loss": 0.4594,
+ "step": 8461
+ },
+ {
+ "epoch": 23.057220708446867,
+ "grad_norm": 6.348690032958984,
+ "learning_rate": 1.7963232076738063e-05,
+ "loss": 0.4368,
+ "step": 8462
+ },
+ {
+ "epoch": 23.059945504087192,
+ "grad_norm": 6.7468085289001465,
+ "learning_rate": 1.7962698250009226e-05,
+ "loss": 0.4341,
+ "step": 8463
+ },
+ {
+ "epoch": 23.06267029972752,
+ "grad_norm": 5.046350955963135,
+ "learning_rate": 1.7962164361267152e-05,
+ "loss": 0.3951,
+ "step": 8464
+ },
+ {
+ "epoch": 23.065395095367847,
+ "grad_norm": 6.5964035987854,
+ "learning_rate": 1.7961630410515998e-05,
+ "loss": 0.4029,
+ "step": 8465
+ },
+ {
+ "epoch": 23.068119891008173,
+ "grad_norm": 5.733188629150391,
+ "learning_rate": 1.7961096397759922e-05,
+ "loss": 0.4281,
+ "step": 8466
+ },
+ {
+ "epoch": 23.070844686648503,
+ "grad_norm": 5.402604579925537,
+ "learning_rate": 1.7960562323003087e-05,
+ "loss": 0.3829,
+ "step": 8467
+ },
+ {
+ "epoch": 23.07356948228883,
+ "grad_norm": 6.682921886444092,
+ "learning_rate": 1.7960028186249644e-05,
+ "loss": 0.5636,
+ "step": 8468
+ },
+ {
+ "epoch": 23.076294277929154,
+ "grad_norm": 5.853186130523682,
+ "learning_rate": 1.7959493987503757e-05,
+ "loss": 0.3913,
+ "step": 8469
+ },
+ {
+ "epoch": 23.079019073569484,
+ "grad_norm": 5.574013710021973,
+ "learning_rate": 1.795895972676959e-05,
+ "loss": 0.4354,
+ "step": 8470
+ },
+ {
+ "epoch": 23.08174386920981,
+ "grad_norm": 6.798804759979248,
+ "learning_rate": 1.79584254040513e-05,
+ "loss": 0.3772,
+ "step": 8471
+ },
+ {
+ "epoch": 23.084468664850135,
+ "grad_norm": 6.968006610870361,
+ "learning_rate": 1.7957891019353048e-05,
+ "loss": 0.4962,
+ "step": 8472
+ },
+ {
+ "epoch": 23.087193460490465,
+ "grad_norm": 5.490119457244873,
+ "learning_rate": 1.7957356572678998e-05,
+ "loss": 0.3618,
+ "step": 8473
+ },
+ {
+ "epoch": 23.08991825613079,
+ "grad_norm": 7.63191556930542,
+ "learning_rate": 1.7956822064033313e-05,
+ "loss": 0.4698,
+ "step": 8474
+ },
+ {
+ "epoch": 23.092643051771116,
+ "grad_norm": 6.752242565155029,
+ "learning_rate": 1.7956287493420152e-05,
+ "loss": 0.5204,
+ "step": 8475
+ },
+ {
+ "epoch": 23.095367847411445,
+ "grad_norm": 5.709225177764893,
+ "learning_rate": 1.7955752860843678e-05,
+ "loss": 0.4392,
+ "step": 8476
+ },
+ {
+ "epoch": 23.09809264305177,
+ "grad_norm": 6.952545166015625,
+ "learning_rate": 1.795521816630806e-05,
+ "loss": 0.4969,
+ "step": 8477
+ },
+ {
+ "epoch": 23.100817438692097,
+ "grad_norm": 5.980533123016357,
+ "learning_rate": 1.7954683409817457e-05,
+ "loss": 0.3753,
+ "step": 8478
+ },
+ {
+ "epoch": 23.103542234332426,
+ "grad_norm": 5.720138072967529,
+ "learning_rate": 1.7954148591376038e-05,
+ "loss": 0.5308,
+ "step": 8479
+ },
+ {
+ "epoch": 23.106267029972752,
+ "grad_norm": 5.523719310760498,
+ "learning_rate": 1.7953613710987966e-05,
+ "loss": 0.386,
+ "step": 8480
+ },
+ {
+ "epoch": 23.108991825613078,
+ "grad_norm": 5.55640983581543,
+ "learning_rate": 1.7953078768657404e-05,
+ "loss": 0.4268,
+ "step": 8481
+ },
+ {
+ "epoch": 23.111716621253407,
+ "grad_norm": 5.8524250984191895,
+ "learning_rate": 1.795254376438852e-05,
+ "loss": 0.4151,
+ "step": 8482
+ },
+ {
+ "epoch": 23.114441416893733,
+ "grad_norm": 5.890065670013428,
+ "learning_rate": 1.7952008698185483e-05,
+ "loss": 0.5322,
+ "step": 8483
+ },
+ {
+ "epoch": 23.11716621253406,
+ "grad_norm": 5.814637660980225,
+ "learning_rate": 1.7951473570052457e-05,
+ "loss": 0.5635,
+ "step": 8484
+ },
+ {
+ "epoch": 23.11989100817439,
+ "grad_norm": 6.418423652648926,
+ "learning_rate": 1.795093837999361e-05,
+ "loss": 0.3712,
+ "step": 8485
+ },
+ {
+ "epoch": 23.122615803814714,
+ "grad_norm": 5.217944622039795,
+ "learning_rate": 1.795040312801311e-05,
+ "loss": 0.371,
+ "step": 8486
+ },
+ {
+ "epoch": 23.12534059945504,
+ "grad_norm": 5.430063247680664,
+ "learning_rate": 1.794986781411513e-05,
+ "loss": 0.4781,
+ "step": 8487
+ },
+ {
+ "epoch": 23.12806539509537,
+ "grad_norm": 5.238339424133301,
+ "learning_rate": 1.7949332438303834e-05,
+ "loss": 0.3886,
+ "step": 8488
+ },
+ {
+ "epoch": 23.130790190735695,
+ "grad_norm": 5.204761028289795,
+ "learning_rate": 1.7948797000583394e-05,
+ "loss": 0.4719,
+ "step": 8489
+ },
+ {
+ "epoch": 23.13351498637602,
+ "grad_norm": 5.8582587242126465,
+ "learning_rate": 1.7948261500957974e-05,
+ "loss": 0.4106,
+ "step": 8490
+ },
+ {
+ "epoch": 23.13623978201635,
+ "grad_norm": 6.394131660461426,
+ "learning_rate": 1.7947725939431756e-05,
+ "loss": 0.3871,
+ "step": 8491
+ },
+ {
+ "epoch": 23.138964577656676,
+ "grad_norm": 6.012871742248535,
+ "learning_rate": 1.7947190316008896e-05,
+ "loss": 0.4056,
+ "step": 8492
+ },
+ {
+ "epoch": 23.141689373297,
+ "grad_norm": 6.837770462036133,
+ "learning_rate": 1.7946654630693583e-05,
+ "loss": 0.517,
+ "step": 8493
+ },
+ {
+ "epoch": 23.14441416893733,
+ "grad_norm": 5.639486312866211,
+ "learning_rate": 1.7946118883489974e-05,
+ "loss": 0.5459,
+ "step": 8494
+ },
+ {
+ "epoch": 23.147138964577657,
+ "grad_norm": 6.541543483734131,
+ "learning_rate": 1.7945583074402246e-05,
+ "loss": 0.3818,
+ "step": 8495
+ },
+ {
+ "epoch": 23.149863760217983,
+ "grad_norm": 5.396148681640625,
+ "learning_rate": 1.7945047203434573e-05,
+ "loss": 0.4304,
+ "step": 8496
+ },
+ {
+ "epoch": 23.152588555858312,
+ "grad_norm": 5.918567657470703,
+ "learning_rate": 1.794451127059113e-05,
+ "loss": 0.5516,
+ "step": 8497
+ },
+ {
+ "epoch": 23.155313351498638,
+ "grad_norm": 5.68643045425415,
+ "learning_rate": 1.7943975275876088e-05,
+ "loss": 0.4506,
+ "step": 8498
+ },
+ {
+ "epoch": 23.158038147138964,
+ "grad_norm": 6.296571731567383,
+ "learning_rate": 1.7943439219293623e-05,
+ "loss": 0.5034,
+ "step": 8499
+ },
+ {
+ "epoch": 23.160762942779293,
+ "grad_norm": 9.516465187072754,
+ "learning_rate": 1.794290310084791e-05,
+ "loss": 0.3637,
+ "step": 8500
+ },
+ {
+ "epoch": 23.16348773841962,
+ "grad_norm": 5.491303443908691,
+ "learning_rate": 1.794236692054312e-05,
+ "loss": 0.3928,
+ "step": 8501
+ },
+ {
+ "epoch": 23.166212534059945,
+ "grad_norm": 5.504067420959473,
+ "learning_rate": 1.7941830678383433e-05,
+ "loss": 0.6379,
+ "step": 8502
+ },
+ {
+ "epoch": 23.168937329700274,
+ "grad_norm": 5.039755821228027,
+ "learning_rate": 1.794129437437303e-05,
+ "loss": 0.461,
+ "step": 8503
+ },
+ {
+ "epoch": 23.1716621253406,
+ "grad_norm": 6.771104335784912,
+ "learning_rate": 1.7940758008516074e-05,
+ "loss": 0.4759,
+ "step": 8504
+ },
+ {
+ "epoch": 23.174386920980925,
+ "grad_norm": 5.8978705406188965,
+ "learning_rate": 1.7940221580816755e-05,
+ "loss": 0.4367,
+ "step": 8505
+ },
+ {
+ "epoch": 23.177111716621255,
+ "grad_norm": 5.212526798248291,
+ "learning_rate": 1.7939685091279244e-05,
+ "loss": 0.3959,
+ "step": 8506
+ },
+ {
+ "epoch": 23.17983651226158,
+ "grad_norm": 6.610459327697754,
+ "learning_rate": 1.7939148539907722e-05,
+ "loss": 0.4451,
+ "step": 8507
+ },
+ {
+ "epoch": 23.182561307901906,
+ "grad_norm": 10.938274383544922,
+ "learning_rate": 1.7938611926706365e-05,
+ "loss": 0.4306,
+ "step": 8508
+ },
+ {
+ "epoch": 23.185286103542236,
+ "grad_norm": 6.370612621307373,
+ "learning_rate": 1.7938075251679356e-05,
+ "loss": 0.4931,
+ "step": 8509
+ },
+ {
+ "epoch": 23.18801089918256,
+ "grad_norm": 5.807062149047852,
+ "learning_rate": 1.793753851483087e-05,
+ "loss": 0.3302,
+ "step": 8510
+ },
+ {
+ "epoch": 23.190735694822887,
+ "grad_norm": 5.843219757080078,
+ "learning_rate": 1.793700171616509e-05,
+ "loss": 0.4984,
+ "step": 8511
+ },
+ {
+ "epoch": 23.193460490463217,
+ "grad_norm": 5.29647970199585,
+ "learning_rate": 1.7936464855686197e-05,
+ "loss": 0.5327,
+ "step": 8512
+ },
+ {
+ "epoch": 23.196185286103542,
+ "grad_norm": 6.996339797973633,
+ "learning_rate": 1.7935927933398374e-05,
+ "loss": 0.4402,
+ "step": 8513
+ },
+ {
+ "epoch": 23.19891008174387,
+ "grad_norm": 5.6097211837768555,
+ "learning_rate": 1.7935390949305796e-05,
+ "loss": 0.6355,
+ "step": 8514
+ },
+ {
+ "epoch": 23.201634877384198,
+ "grad_norm": 5.417934417724609,
+ "learning_rate": 1.7934853903412648e-05,
+ "loss": 0.4609,
+ "step": 8515
+ },
+ {
+ "epoch": 23.204359673024523,
+ "grad_norm": 6.048050880432129,
+ "learning_rate": 1.7934316795723116e-05,
+ "loss": 0.4696,
+ "step": 8516
+ },
+ {
+ "epoch": 23.20708446866485,
+ "grad_norm": 6.537152290344238,
+ "learning_rate": 1.793377962624138e-05,
+ "loss": 0.4638,
+ "step": 8517
+ },
+ {
+ "epoch": 23.20980926430518,
+ "grad_norm": 5.470832824707031,
+ "learning_rate": 1.793324239497162e-05,
+ "loss": 0.356,
+ "step": 8518
+ },
+ {
+ "epoch": 23.212534059945504,
+ "grad_norm": 5.512134552001953,
+ "learning_rate": 1.7932705101918025e-05,
+ "loss": 0.5834,
+ "step": 8519
+ },
+ {
+ "epoch": 23.21525885558583,
+ "grad_norm": 6.58221960067749,
+ "learning_rate": 1.793216774708478e-05,
+ "loss": 0.3472,
+ "step": 8520
+ },
+ {
+ "epoch": 23.21798365122616,
+ "grad_norm": 6.857476711273193,
+ "learning_rate": 1.7931630330476066e-05,
+ "loss": 0.5113,
+ "step": 8521
+ },
+ {
+ "epoch": 23.220708446866485,
+ "grad_norm": 9.61906909942627,
+ "learning_rate": 1.793109285209607e-05,
+ "loss": 0.414,
+ "step": 8522
+ },
+ {
+ "epoch": 23.22343324250681,
+ "grad_norm": 5.1094255447387695,
+ "learning_rate": 1.793055531194898e-05,
+ "loss": 0.4451,
+ "step": 8523
+ },
+ {
+ "epoch": 23.22615803814714,
+ "grad_norm": 4.868687152862549,
+ "learning_rate": 1.793001771003898e-05,
+ "loss": 0.4315,
+ "step": 8524
+ },
+ {
+ "epoch": 23.228882833787466,
+ "grad_norm": 8.179791450500488,
+ "learning_rate": 1.7929480046370257e-05,
+ "loss": 0.5508,
+ "step": 8525
+ },
+ {
+ "epoch": 23.231607629427792,
+ "grad_norm": 6.513568878173828,
+ "learning_rate": 1.7928942320947e-05,
+ "loss": 0.6143,
+ "step": 8526
+ },
+ {
+ "epoch": 23.23433242506812,
+ "grad_norm": 6.177130699157715,
+ "learning_rate": 1.7928404533773394e-05,
+ "loss": 0.5115,
+ "step": 8527
+ },
+ {
+ "epoch": 23.237057220708447,
+ "grad_norm": 7.569170951843262,
+ "learning_rate": 1.7927866684853628e-05,
+ "loss": 0.4139,
+ "step": 8528
+ },
+ {
+ "epoch": 23.239782016348773,
+ "grad_norm": 6.225154876708984,
+ "learning_rate": 1.792732877419189e-05,
+ "loss": 0.4716,
+ "step": 8529
+ },
+ {
+ "epoch": 23.242506811989102,
+ "grad_norm": 6.720955848693848,
+ "learning_rate": 1.7926790801792374e-05,
+ "loss": 0.518,
+ "step": 8530
+ },
+ {
+ "epoch": 23.245231607629428,
+ "grad_norm": 5.427977561950684,
+ "learning_rate": 1.7926252767659265e-05,
+ "loss": 0.514,
+ "step": 8531
+ },
+ {
+ "epoch": 23.247956403269754,
+ "grad_norm": 6.594433784484863,
+ "learning_rate": 1.7925714671796754e-05,
+ "loss": 0.3976,
+ "step": 8532
+ },
+ {
+ "epoch": 23.250681198910083,
+ "grad_norm": 6.112623691558838,
+ "learning_rate": 1.7925176514209033e-05,
+ "loss": 0.5648,
+ "step": 8533
+ },
+ {
+ "epoch": 23.25340599455041,
+ "grad_norm": 5.751542091369629,
+ "learning_rate": 1.7924638294900292e-05,
+ "loss": 0.5437,
+ "step": 8534
+ },
+ {
+ "epoch": 23.256130790190735,
+ "grad_norm": 5.23914098739624,
+ "learning_rate": 1.7924100013874725e-05,
+ "loss": 0.6124,
+ "step": 8535
+ },
+ {
+ "epoch": 23.258855585831064,
+ "grad_norm": 6.711471080780029,
+ "learning_rate": 1.792356167113652e-05,
+ "loss": 0.4122,
+ "step": 8536
+ },
+ {
+ "epoch": 23.26158038147139,
+ "grad_norm": 5.451192855834961,
+ "learning_rate": 1.792302326668987e-05,
+ "loss": 0.3857,
+ "step": 8537
+ },
+ {
+ "epoch": 23.264305177111716,
+ "grad_norm": 6.537862777709961,
+ "learning_rate": 1.7922484800538973e-05,
+ "loss": 0.3556,
+ "step": 8538
+ },
+ {
+ "epoch": 23.267029972752045,
+ "grad_norm": 6.42893123626709,
+ "learning_rate": 1.7921946272688018e-05,
+ "loss": 0.4828,
+ "step": 8539
+ },
+ {
+ "epoch": 23.26975476839237,
+ "grad_norm": 5.245488166809082,
+ "learning_rate": 1.79214076831412e-05,
+ "loss": 0.3732,
+ "step": 8540
+ },
+ {
+ "epoch": 23.272479564032697,
+ "grad_norm": 5.6094584465026855,
+ "learning_rate": 1.7920869031902715e-05,
+ "loss": 0.4851,
+ "step": 8541
+ },
+ {
+ "epoch": 23.275204359673026,
+ "grad_norm": 5.0258660316467285,
+ "learning_rate": 1.792033031897676e-05,
+ "loss": 0.3192,
+ "step": 8542
+ },
+ {
+ "epoch": 23.277929155313352,
+ "grad_norm": 5.8088884353637695,
+ "learning_rate": 1.791979154436752e-05,
+ "loss": 0.5436,
+ "step": 8543
+ },
+ {
+ "epoch": 23.280653950953678,
+ "grad_norm": 5.712719440460205,
+ "learning_rate": 1.7919252708079203e-05,
+ "loss": 0.4938,
+ "step": 8544
+ },
+ {
+ "epoch": 23.283378746594007,
+ "grad_norm": 5.650703430175781,
+ "learning_rate": 1.7918713810116e-05,
+ "loss": 0.5297,
+ "step": 8545
+ },
+ {
+ "epoch": 23.286103542234333,
+ "grad_norm": 5.181207180023193,
+ "learning_rate": 1.791817485048211e-05,
+ "loss": 0.4509,
+ "step": 8546
+ },
+ {
+ "epoch": 23.28882833787466,
+ "grad_norm": 5.207966327667236,
+ "learning_rate": 1.7917635829181726e-05,
+ "loss": 0.3967,
+ "step": 8547
+ },
+ {
+ "epoch": 23.291553133514988,
+ "grad_norm": 5.931425094604492,
+ "learning_rate": 1.7917096746219053e-05,
+ "loss": 0.3824,
+ "step": 8548
+ },
+ {
+ "epoch": 23.294277929155314,
+ "grad_norm": 5.336034297943115,
+ "learning_rate": 1.7916557601598283e-05,
+ "loss": 0.5178,
+ "step": 8549
+ },
+ {
+ "epoch": 23.29700272479564,
+ "grad_norm": 6.400340557098389,
+ "learning_rate": 1.7916018395323614e-05,
+ "loss": 0.3469,
+ "step": 8550
+ },
+ {
+ "epoch": 23.29972752043597,
+ "grad_norm": 6.1717753410339355,
+ "learning_rate": 1.7915479127399255e-05,
+ "loss": 0.4749,
+ "step": 8551
+ },
+ {
+ "epoch": 23.302452316076295,
+ "grad_norm": 5.726232528686523,
+ "learning_rate": 1.7914939797829393e-05,
+ "loss": 0.7012,
+ "step": 8552
+ },
+ {
+ "epoch": 23.30517711171662,
+ "grad_norm": 5.63735294342041,
+ "learning_rate": 1.791440040661824e-05,
+ "loss": 0.4526,
+ "step": 8553
+ },
+ {
+ "epoch": 23.30790190735695,
+ "grad_norm": 6.525546550750732,
+ "learning_rate": 1.7913860953769987e-05,
+ "loss": 0.3878,
+ "step": 8554
+ },
+ {
+ "epoch": 23.310626702997276,
+ "grad_norm": 6.347872734069824,
+ "learning_rate": 1.791332143928884e-05,
+ "loss": 0.4606,
+ "step": 8555
+ },
+ {
+ "epoch": 23.3133514986376,
+ "grad_norm": 5.721081733703613,
+ "learning_rate": 1.7912781863179005e-05,
+ "loss": 0.4918,
+ "step": 8556
+ },
+ {
+ "epoch": 23.31607629427793,
+ "grad_norm": 5.198385238647461,
+ "learning_rate": 1.7912242225444676e-05,
+ "loss": 0.5143,
+ "step": 8557
+ },
+ {
+ "epoch": 23.318801089918257,
+ "grad_norm": 5.652178764343262,
+ "learning_rate": 1.791170252609006e-05,
+ "loss": 0.4419,
+ "step": 8558
+ },
+ {
+ "epoch": 23.321525885558582,
+ "grad_norm": 6.685900688171387,
+ "learning_rate": 1.7911162765119357e-05,
+ "loss": 0.4797,
+ "step": 8559
+ },
+ {
+ "epoch": 23.32425068119891,
+ "grad_norm": 6.986310005187988,
+ "learning_rate": 1.7910622942536777e-05,
+ "loss": 0.5007,
+ "step": 8560
+ },
+ {
+ "epoch": 23.326975476839237,
+ "grad_norm": 4.961942195892334,
+ "learning_rate": 1.7910083058346516e-05,
+ "loss": 0.6426,
+ "step": 8561
+ },
+ {
+ "epoch": 23.329700272479563,
+ "grad_norm": 4.514638423919678,
+ "learning_rate": 1.790954311255279e-05,
+ "loss": 0.3539,
+ "step": 8562
+ },
+ {
+ "epoch": 23.332425068119893,
+ "grad_norm": 6.418118953704834,
+ "learning_rate": 1.790900310515979e-05,
+ "loss": 0.5201,
+ "step": 8563
+ },
+ {
+ "epoch": 23.33514986376022,
+ "grad_norm": 5.593393325805664,
+ "learning_rate": 1.790846303617173e-05,
+ "loss": 0.5619,
+ "step": 8564
+ },
+ {
+ "epoch": 23.337874659400544,
+ "grad_norm": 5.699489593505859,
+ "learning_rate": 1.7907922905592813e-05,
+ "loss": 0.4404,
+ "step": 8565
+ },
+ {
+ "epoch": 23.340599455040874,
+ "grad_norm": 5.737560749053955,
+ "learning_rate": 1.790738271342725e-05,
+ "loss": 0.4419,
+ "step": 8566
+ },
+ {
+ "epoch": 23.3433242506812,
+ "grad_norm": 6.059821605682373,
+ "learning_rate": 1.7906842459679242e-05,
+ "loss": 0.5351,
+ "step": 8567
+ },
+ {
+ "epoch": 23.346049046321525,
+ "grad_norm": 8.688183784484863,
+ "learning_rate": 1.7906302144353002e-05,
+ "loss": 0.4892,
+ "step": 8568
+ },
+ {
+ "epoch": 23.348773841961854,
+ "grad_norm": 5.029569625854492,
+ "learning_rate": 1.7905761767452735e-05,
+ "loss": 0.455,
+ "step": 8569
+ },
+ {
+ "epoch": 23.35149863760218,
+ "grad_norm": 6.375662803649902,
+ "learning_rate": 1.7905221328982647e-05,
+ "loss": 0.5518,
+ "step": 8570
+ },
+ {
+ "epoch": 23.354223433242506,
+ "grad_norm": 5.912632465362549,
+ "learning_rate": 1.790468082894695e-05,
+ "loss": 0.493,
+ "step": 8571
+ },
+ {
+ "epoch": 23.356948228882835,
+ "grad_norm": 4.947330951690674,
+ "learning_rate": 1.7904140267349856e-05,
+ "loss": 0.3582,
+ "step": 8572
+ },
+ {
+ "epoch": 23.35967302452316,
+ "grad_norm": 4.714034080505371,
+ "learning_rate": 1.7903599644195573e-05,
+ "loss": 0.4768,
+ "step": 8573
+ },
+ {
+ "epoch": 23.362397820163487,
+ "grad_norm": 5.655596733093262,
+ "learning_rate": 1.7903058959488307e-05,
+ "loss": 0.4032,
+ "step": 8574
+ },
+ {
+ "epoch": 23.365122615803816,
+ "grad_norm": 5.277097702026367,
+ "learning_rate": 1.7902518213232273e-05,
+ "loss": 0.4848,
+ "step": 8575
+ },
+ {
+ "epoch": 23.367847411444142,
+ "grad_norm": 5.259819507598877,
+ "learning_rate": 1.7901977405431683e-05,
+ "loss": 0.4923,
+ "step": 8576
+ },
+ {
+ "epoch": 23.370572207084468,
+ "grad_norm": 6.728734970092773,
+ "learning_rate": 1.7901436536090747e-05,
+ "loss": 0.4941,
+ "step": 8577
+ },
+ {
+ "epoch": 23.373297002724797,
+ "grad_norm": 7.218616008758545,
+ "learning_rate": 1.7900895605213673e-05,
+ "loss": 0.4794,
+ "step": 8578
+ },
+ {
+ "epoch": 23.376021798365123,
+ "grad_norm": 6.109345436096191,
+ "learning_rate": 1.7900354612804685e-05,
+ "loss": 0.5389,
+ "step": 8579
+ },
+ {
+ "epoch": 23.37874659400545,
+ "grad_norm": 6.026385307312012,
+ "learning_rate": 1.7899813558867985e-05,
+ "loss": 0.5732,
+ "step": 8580
+ },
+ {
+ "epoch": 23.381471389645778,
+ "grad_norm": 5.888758182525635,
+ "learning_rate": 1.7899272443407792e-05,
+ "loss": 0.4734,
+ "step": 8581
+ },
+ {
+ "epoch": 23.384196185286104,
+ "grad_norm": 6.326576232910156,
+ "learning_rate": 1.789873126642832e-05,
+ "loss": 0.6313,
+ "step": 8582
+ },
+ {
+ "epoch": 23.38692098092643,
+ "grad_norm": 4.561391830444336,
+ "learning_rate": 1.7898190027933785e-05,
+ "loss": 0.3617,
+ "step": 8583
+ },
+ {
+ "epoch": 23.38964577656676,
+ "grad_norm": 6.6524176597595215,
+ "learning_rate": 1.78976487279284e-05,
+ "loss": 0.4192,
+ "step": 8584
+ },
+ {
+ "epoch": 23.392370572207085,
+ "grad_norm": 6.1215996742248535,
+ "learning_rate": 1.7897107366416378e-05,
+ "loss": 0.5343,
+ "step": 8585
+ },
+ {
+ "epoch": 23.39509536784741,
+ "grad_norm": 8.58316421508789,
+ "learning_rate": 1.789656594340194e-05,
+ "loss": 0.4664,
+ "step": 8586
+ },
+ {
+ "epoch": 23.39782016348774,
+ "grad_norm": 5.488253593444824,
+ "learning_rate": 1.78960244588893e-05,
+ "loss": 0.3581,
+ "step": 8587
+ },
+ {
+ "epoch": 23.400544959128066,
+ "grad_norm": 6.52658748626709,
+ "learning_rate": 1.7895482912882673e-05,
+ "loss": 0.4294,
+ "step": 8588
+ },
+ {
+ "epoch": 23.40326975476839,
+ "grad_norm": 4.935141563415527,
+ "learning_rate": 1.7894941305386283e-05,
+ "loss": 0.6022,
+ "step": 8589
+ },
+ {
+ "epoch": 23.40599455040872,
+ "grad_norm": 6.0698723793029785,
+ "learning_rate": 1.7894399636404344e-05,
+ "loss": 0.397,
+ "step": 8590
+ },
+ {
+ "epoch": 23.408719346049047,
+ "grad_norm": 5.632525444030762,
+ "learning_rate": 1.7893857905941074e-05,
+ "loss": 0.4578,
+ "step": 8591
+ },
+ {
+ "epoch": 23.411444141689373,
+ "grad_norm": 5.183593273162842,
+ "learning_rate": 1.789331611400069e-05,
+ "loss": 0.5265,
+ "step": 8592
+ },
+ {
+ "epoch": 23.414168937329702,
+ "grad_norm": 5.359197616577148,
+ "learning_rate": 1.789277426058742e-05,
+ "loss": 0.3425,
+ "step": 8593
+ },
+ {
+ "epoch": 23.416893732970028,
+ "grad_norm": 6.475907802581787,
+ "learning_rate": 1.7892232345705475e-05,
+ "loss": 0.5065,
+ "step": 8594
+ },
+ {
+ "epoch": 23.419618528610354,
+ "grad_norm": 5.7222747802734375,
+ "learning_rate": 1.7891690369359077e-05,
+ "loss": 0.4578,
+ "step": 8595
+ },
+ {
+ "epoch": 23.422343324250683,
+ "grad_norm": 6.62075662612915,
+ "learning_rate": 1.789114833155245e-05,
+ "loss": 0.5696,
+ "step": 8596
+ },
+ {
+ "epoch": 23.42506811989101,
+ "grad_norm": 6.1349053382873535,
+ "learning_rate": 1.7890606232289815e-05,
+ "loss": 0.5133,
+ "step": 8597
+ },
+ {
+ "epoch": 23.427792915531334,
+ "grad_norm": 5.929188251495361,
+ "learning_rate": 1.789006407157539e-05,
+ "loss": 0.5255,
+ "step": 8598
+ },
+ {
+ "epoch": 23.430517711171664,
+ "grad_norm": 5.4338483810424805,
+ "learning_rate": 1.78895218494134e-05,
+ "loss": 0.439,
+ "step": 8599
+ },
+ {
+ "epoch": 23.43324250681199,
+ "grad_norm": 5.5654377937316895,
+ "learning_rate": 1.788897956580807e-05,
+ "loss": 0.3967,
+ "step": 8600
+ },
+ {
+ "epoch": 23.435967302452315,
+ "grad_norm": 6.25413703918457,
+ "learning_rate": 1.7888437220763618e-05,
+ "loss": 0.4529,
+ "step": 8601
+ },
+ {
+ "epoch": 23.438692098092645,
+ "grad_norm": 5.945039749145508,
+ "learning_rate": 1.7887894814284274e-05,
+ "loss": 0.3684,
+ "step": 8602
+ },
+ {
+ "epoch": 23.44141689373297,
+ "grad_norm": 6.790313720703125,
+ "learning_rate": 1.7887352346374255e-05,
+ "loss": 0.4329,
+ "step": 8603
+ },
+ {
+ "epoch": 23.444141689373296,
+ "grad_norm": 6.135364055633545,
+ "learning_rate": 1.7886809817037797e-05,
+ "loss": 0.4982,
+ "step": 8604
+ },
+ {
+ "epoch": 23.446866485013626,
+ "grad_norm": 6.314703464508057,
+ "learning_rate": 1.7886267226279114e-05,
+ "loss": 0.4789,
+ "step": 8605
+ },
+ {
+ "epoch": 23.44959128065395,
+ "grad_norm": 5.485182762145996,
+ "learning_rate": 1.788572457410243e-05,
+ "loss": 0.3937,
+ "step": 8606
+ },
+ {
+ "epoch": 23.452316076294277,
+ "grad_norm": 8.23728084564209,
+ "learning_rate": 1.7885181860511983e-05,
+ "loss": 0.4464,
+ "step": 8607
+ },
+ {
+ "epoch": 23.455040871934607,
+ "grad_norm": 5.5518693923950195,
+ "learning_rate": 1.7884639085511996e-05,
+ "loss": 0.4333,
+ "step": 8608
+ },
+ {
+ "epoch": 23.457765667574932,
+ "grad_norm": 6.086019992828369,
+ "learning_rate": 1.7884096249106687e-05,
+ "loss": 0.47,
+ "step": 8609
+ },
+ {
+ "epoch": 23.460490463215258,
+ "grad_norm": 6.261456489562988,
+ "learning_rate": 1.7883553351300292e-05,
+ "loss": 0.4594,
+ "step": 8610
+ },
+ {
+ "epoch": 23.463215258855588,
+ "grad_norm": 5.2257513999938965,
+ "learning_rate": 1.788301039209704e-05,
+ "loss": 0.4789,
+ "step": 8611
+ },
+ {
+ "epoch": 23.465940054495913,
+ "grad_norm": 5.454006195068359,
+ "learning_rate": 1.7882467371501156e-05,
+ "loss": 0.3983,
+ "step": 8612
+ },
+ {
+ "epoch": 23.46866485013624,
+ "grad_norm": 5.600664138793945,
+ "learning_rate": 1.788192428951687e-05,
+ "loss": 0.4689,
+ "step": 8613
+ },
+ {
+ "epoch": 23.47138964577657,
+ "grad_norm": 5.640192031860352,
+ "learning_rate": 1.788138114614841e-05,
+ "loss": 0.4866,
+ "step": 8614
+ },
+ {
+ "epoch": 23.474114441416894,
+ "grad_norm": 6.099294662475586,
+ "learning_rate": 1.7880837941400005e-05,
+ "loss": 0.483,
+ "step": 8615
+ },
+ {
+ "epoch": 23.47683923705722,
+ "grad_norm": 5.4578094482421875,
+ "learning_rate": 1.788029467527589e-05,
+ "loss": 0.5018,
+ "step": 8616
+ },
+ {
+ "epoch": 23.479564032697546,
+ "grad_norm": 5.151687145233154,
+ "learning_rate": 1.7879751347780297e-05,
+ "loss": 0.3944,
+ "step": 8617
+ },
+ {
+ "epoch": 23.482288828337875,
+ "grad_norm": 5.99576473236084,
+ "learning_rate": 1.7879207958917447e-05,
+ "loss": 0.3472,
+ "step": 8618
+ },
+ {
+ "epoch": 23.4850136239782,
+ "grad_norm": 6.2755913734436035,
+ "learning_rate": 1.7878664508691586e-05,
+ "loss": 0.4242,
+ "step": 8619
+ },
+ {
+ "epoch": 23.48773841961853,
+ "grad_norm": 5.501990795135498,
+ "learning_rate": 1.7878120997106937e-05,
+ "loss": 0.3637,
+ "step": 8620
+ },
+ {
+ "epoch": 23.490463215258856,
+ "grad_norm": 5.846157073974609,
+ "learning_rate": 1.7877577424167734e-05,
+ "loss": 0.3648,
+ "step": 8621
+ },
+ {
+ "epoch": 23.493188010899182,
+ "grad_norm": 5.956545352935791,
+ "learning_rate": 1.7877033789878212e-05,
+ "loss": 0.5757,
+ "step": 8622
+ },
+ {
+ "epoch": 23.495912806539508,
+ "grad_norm": 5.604905128479004,
+ "learning_rate": 1.7876490094242604e-05,
+ "loss": 0.5396,
+ "step": 8623
+ },
+ {
+ "epoch": 23.498637602179837,
+ "grad_norm": 5.4336724281311035,
+ "learning_rate": 1.7875946337265143e-05,
+ "loss": 0.4486,
+ "step": 8624
+ },
+ {
+ "epoch": 23.501362397820163,
+ "grad_norm": 6.402991771697998,
+ "learning_rate": 1.787540251895007e-05,
+ "loss": 0.6067,
+ "step": 8625
+ },
+ {
+ "epoch": 23.504087193460492,
+ "grad_norm": 5.886371612548828,
+ "learning_rate": 1.7874858639301616e-05,
+ "loss": 0.4803,
+ "step": 8626
+ },
+ {
+ "epoch": 23.506811989100818,
+ "grad_norm": 6.539076328277588,
+ "learning_rate": 1.7874314698324012e-05,
+ "loss": 0.5058,
+ "step": 8627
+ },
+ {
+ "epoch": 23.509536784741144,
+ "grad_norm": 6.142359256744385,
+ "learning_rate": 1.78737706960215e-05,
+ "loss": 0.5498,
+ "step": 8628
+ },
+ {
+ "epoch": 23.51226158038147,
+ "grad_norm": 6.924804210662842,
+ "learning_rate": 1.7873226632398317e-05,
+ "loss": 0.5751,
+ "step": 8629
+ },
+ {
+ "epoch": 23.5149863760218,
+ "grad_norm": 6.1959452629089355,
+ "learning_rate": 1.7872682507458696e-05,
+ "loss": 0.5513,
+ "step": 8630
+ },
+ {
+ "epoch": 23.517711171662125,
+ "grad_norm": 5.745352268218994,
+ "learning_rate": 1.7872138321206876e-05,
+ "loss": 0.4954,
+ "step": 8631
+ },
+ {
+ "epoch": 23.520435967302454,
+ "grad_norm": 5.731938362121582,
+ "learning_rate": 1.78715940736471e-05,
+ "loss": 0.4283,
+ "step": 8632
+ },
+ {
+ "epoch": 23.52316076294278,
+ "grad_norm": 5.788858890533447,
+ "learning_rate": 1.7871049764783598e-05,
+ "loss": 0.4741,
+ "step": 8633
+ },
+ {
+ "epoch": 23.525885558583106,
+ "grad_norm": 5.21207857131958,
+ "learning_rate": 1.7870505394620617e-05,
+ "loss": 0.472,
+ "step": 8634
+ },
+ {
+ "epoch": 23.52861035422343,
+ "grad_norm": 4.9614644050598145,
+ "learning_rate": 1.7869960963162395e-05,
+ "loss": 0.3557,
+ "step": 8635
+ },
+ {
+ "epoch": 23.53133514986376,
+ "grad_norm": 5.98054838180542,
+ "learning_rate": 1.7869416470413167e-05,
+ "loss": 0.4821,
+ "step": 8636
+ },
+ {
+ "epoch": 23.534059945504087,
+ "grad_norm": 6.522042751312256,
+ "learning_rate": 1.7868871916377178e-05,
+ "loss": 0.5333,
+ "step": 8637
+ },
+ {
+ "epoch": 23.536784741144416,
+ "grad_norm": 5.415054798126221,
+ "learning_rate": 1.7868327301058668e-05,
+ "loss": 0.4015,
+ "step": 8638
+ },
+ {
+ "epoch": 23.539509536784742,
+ "grad_norm": 6.223445892333984,
+ "learning_rate": 1.7867782624461877e-05,
+ "loss": 0.3592,
+ "step": 8639
+ },
+ {
+ "epoch": 23.542234332425068,
+ "grad_norm": 5.5288920402526855,
+ "learning_rate": 1.786723788659105e-05,
+ "loss": 0.5221,
+ "step": 8640
+ },
+ {
+ "epoch": 23.544959128065393,
+ "grad_norm": 5.534078598022461,
+ "learning_rate": 1.7866693087450426e-05,
+ "loss": 0.3898,
+ "step": 8641
+ },
+ {
+ "epoch": 23.547683923705723,
+ "grad_norm": 5.140962600708008,
+ "learning_rate": 1.786614822704425e-05,
+ "loss": 0.4922,
+ "step": 8642
+ },
+ {
+ "epoch": 23.55040871934605,
+ "grad_norm": 4.935930252075195,
+ "learning_rate": 1.7865603305376765e-05,
+ "loss": 0.5097,
+ "step": 8643
+ },
+ {
+ "epoch": 23.553133514986374,
+ "grad_norm": 5.087876796722412,
+ "learning_rate": 1.7865058322452214e-05,
+ "loss": 0.4529,
+ "step": 8644
+ },
+ {
+ "epoch": 23.555858310626704,
+ "grad_norm": 5.783227920532227,
+ "learning_rate": 1.7864513278274843e-05,
+ "loss": 0.5394,
+ "step": 8645
+ },
+ {
+ "epoch": 23.55858310626703,
+ "grad_norm": 5.756707191467285,
+ "learning_rate": 1.7863968172848897e-05,
+ "loss": 0.3569,
+ "step": 8646
+ },
+ {
+ "epoch": 23.561307901907355,
+ "grad_norm": 4.840692520141602,
+ "learning_rate": 1.7863423006178616e-05,
+ "loss": 0.4358,
+ "step": 8647
+ },
+ {
+ "epoch": 23.564032697547685,
+ "grad_norm": 5.690715789794922,
+ "learning_rate": 1.786287777826825e-05,
+ "loss": 0.4267,
+ "step": 8648
+ },
+ {
+ "epoch": 23.56675749318801,
+ "grad_norm": 5.443446159362793,
+ "learning_rate": 1.786233248912205e-05,
+ "loss": 0.3448,
+ "step": 8649
+ },
+ {
+ "epoch": 23.569482288828336,
+ "grad_norm": 6.244948387145996,
+ "learning_rate": 1.7861787138744254e-05,
+ "loss": 0.5072,
+ "step": 8650
+ },
+ {
+ "epoch": 23.572207084468666,
+ "grad_norm": 5.7256693840026855,
+ "learning_rate": 1.7861241727139113e-05,
+ "loss": 0.5837,
+ "step": 8651
+ },
+ {
+ "epoch": 23.57493188010899,
+ "grad_norm": 6.121038436889648,
+ "learning_rate": 1.7860696254310874e-05,
+ "loss": 0.4305,
+ "step": 8652
+ },
+ {
+ "epoch": 23.577656675749317,
+ "grad_norm": 6.847372531890869,
+ "learning_rate": 1.7860150720263786e-05,
+ "loss": 0.3922,
+ "step": 8653
+ },
+ {
+ "epoch": 23.580381471389646,
+ "grad_norm": 6.23608922958374,
+ "learning_rate": 1.7859605125002098e-05,
+ "loss": 0.3429,
+ "step": 8654
+ },
+ {
+ "epoch": 23.583106267029972,
+ "grad_norm": 8.2210693359375,
+ "learning_rate": 1.7859059468530055e-05,
+ "loss": 0.3873,
+ "step": 8655
+ },
+ {
+ "epoch": 23.585831062670298,
+ "grad_norm": 6.68287992477417,
+ "learning_rate": 1.7858513750851913e-05,
+ "loss": 0.5414,
+ "step": 8656
+ },
+ {
+ "epoch": 23.588555858310627,
+ "grad_norm": 6.5625386238098145,
+ "learning_rate": 1.785796797197192e-05,
+ "loss": 0.4076,
+ "step": 8657
+ },
+ {
+ "epoch": 23.591280653950953,
+ "grad_norm": 5.698672771453857,
+ "learning_rate": 1.7857422131894322e-05,
+ "loss": 0.5437,
+ "step": 8658
+ },
+ {
+ "epoch": 23.59400544959128,
+ "grad_norm": 5.085134029388428,
+ "learning_rate": 1.7856876230623374e-05,
+ "loss": 0.3978,
+ "step": 8659
+ },
+ {
+ "epoch": 23.59673024523161,
+ "grad_norm": 5.725650310516357,
+ "learning_rate": 1.785633026816333e-05,
+ "loss": 0.472,
+ "step": 8660
+ },
+ {
+ "epoch": 23.599455040871934,
+ "grad_norm": 5.658379554748535,
+ "learning_rate": 1.7855784244518434e-05,
+ "loss": 0.3611,
+ "step": 8661
+ },
+ {
+ "epoch": 23.60217983651226,
+ "grad_norm": 4.903540134429932,
+ "learning_rate": 1.7855238159692943e-05,
+ "loss": 0.3466,
+ "step": 8662
+ },
+ {
+ "epoch": 23.60490463215259,
+ "grad_norm": 5.588335990905762,
+ "learning_rate": 1.7854692013691114e-05,
+ "loss": 0.4419,
+ "step": 8663
+ },
+ {
+ "epoch": 23.607629427792915,
+ "grad_norm": 7.128772735595703,
+ "learning_rate": 1.7854145806517195e-05,
+ "loss": 0.4494,
+ "step": 8664
+ },
+ {
+ "epoch": 23.61035422343324,
+ "grad_norm": 5.376194477081299,
+ "learning_rate": 1.7853599538175443e-05,
+ "loss": 0.4866,
+ "step": 8665
+ },
+ {
+ "epoch": 23.61307901907357,
+ "grad_norm": 5.624248504638672,
+ "learning_rate": 1.785305320867011e-05,
+ "loss": 0.3914,
+ "step": 8666
+ },
+ {
+ "epoch": 23.615803814713896,
+ "grad_norm": 5.858697414398193,
+ "learning_rate": 1.785250681800545e-05,
+ "loss": 0.3456,
+ "step": 8667
+ },
+ {
+ "epoch": 23.618528610354222,
+ "grad_norm": 4.45041561126709,
+ "learning_rate": 1.7851960366185718e-05,
+ "loss": 0.4218,
+ "step": 8668
+ },
+ {
+ "epoch": 23.62125340599455,
+ "grad_norm": 6.466994285583496,
+ "learning_rate": 1.7851413853215175e-05,
+ "loss": 0.4608,
+ "step": 8669
+ },
+ {
+ "epoch": 23.623978201634877,
+ "grad_norm": 7.6471266746521,
+ "learning_rate": 1.785086727909807e-05,
+ "loss": 0.5364,
+ "step": 8670
+ },
+ {
+ "epoch": 23.626702997275203,
+ "grad_norm": 5.692680358886719,
+ "learning_rate": 1.7850320643838666e-05,
+ "loss": 0.4006,
+ "step": 8671
+ },
+ {
+ "epoch": 23.629427792915532,
+ "grad_norm": 5.619343280792236,
+ "learning_rate": 1.7849773947441215e-05,
+ "loss": 0.4422,
+ "step": 8672
+ },
+ {
+ "epoch": 23.632152588555858,
+ "grad_norm": 8.094730377197266,
+ "learning_rate": 1.784922718990998e-05,
+ "loss": 0.3941,
+ "step": 8673
+ },
+ {
+ "epoch": 23.634877384196184,
+ "grad_norm": 6.297821521759033,
+ "learning_rate": 1.7848680371249214e-05,
+ "loss": 0.4475,
+ "step": 8674
+ },
+ {
+ "epoch": 23.637602179836513,
+ "grad_norm": 6.446152210235596,
+ "learning_rate": 1.7848133491463176e-05,
+ "loss": 0.5477,
+ "step": 8675
+ },
+ {
+ "epoch": 23.64032697547684,
+ "grad_norm": 4.971529006958008,
+ "learning_rate": 1.7847586550556132e-05,
+ "loss": 0.3863,
+ "step": 8676
+ },
+ {
+ "epoch": 23.643051771117165,
+ "grad_norm": 6.761627197265625,
+ "learning_rate": 1.7847039548532333e-05,
+ "loss": 0.4704,
+ "step": 8677
+ },
+ {
+ "epoch": 23.645776566757494,
+ "grad_norm": 5.539989471435547,
+ "learning_rate": 1.7846492485396046e-05,
+ "loss": 0.4274,
+ "step": 8678
+ },
+ {
+ "epoch": 23.64850136239782,
+ "grad_norm": 6.167756080627441,
+ "learning_rate": 1.7845945361151526e-05,
+ "loss": 0.5418,
+ "step": 8679
+ },
+ {
+ "epoch": 23.651226158038146,
+ "grad_norm": 5.476174354553223,
+ "learning_rate": 1.7845398175803035e-05,
+ "loss": 0.5606,
+ "step": 8680
+ },
+ {
+ "epoch": 23.653950953678475,
+ "grad_norm": 6.187638282775879,
+ "learning_rate": 1.7844850929354835e-05,
+ "loss": 0.496,
+ "step": 8681
+ },
+ {
+ "epoch": 23.6566757493188,
+ "grad_norm": 5.924915790557861,
+ "learning_rate": 1.784430362181119e-05,
+ "loss": 0.5053,
+ "step": 8682
+ },
+ {
+ "epoch": 23.659400544959126,
+ "grad_norm": 5.7016825675964355,
+ "learning_rate": 1.7843756253176362e-05,
+ "loss": 0.3604,
+ "step": 8683
+ },
+ {
+ "epoch": 23.662125340599456,
+ "grad_norm": 4.99558162689209,
+ "learning_rate": 1.784320882345461e-05,
+ "loss": 0.4903,
+ "step": 8684
+ },
+ {
+ "epoch": 23.66485013623978,
+ "grad_norm": 4.536379814147949,
+ "learning_rate": 1.7842661332650202e-05,
+ "loss": 0.658,
+ "step": 8685
+ },
+ {
+ "epoch": 23.667574931880107,
+ "grad_norm": 5.127689838409424,
+ "learning_rate": 1.78421137807674e-05,
+ "loss": 0.3423,
+ "step": 8686
+ },
+ {
+ "epoch": 23.670299727520437,
+ "grad_norm": 5.500685691833496,
+ "learning_rate": 1.7841566167810467e-05,
+ "loss": 0.3532,
+ "step": 8687
+ },
+ {
+ "epoch": 23.673024523160763,
+ "grad_norm": 5.793558120727539,
+ "learning_rate": 1.784101849378367e-05,
+ "loss": 0.4622,
+ "step": 8688
+ },
+ {
+ "epoch": 23.67574931880109,
+ "grad_norm": 5.604568958282471,
+ "learning_rate": 1.7840470758691274e-05,
+ "loss": 0.4931,
+ "step": 8689
+ },
+ {
+ "epoch": 23.678474114441418,
+ "grad_norm": 5.239307403564453,
+ "learning_rate": 1.7839922962537543e-05,
+ "loss": 0.4184,
+ "step": 8690
+ },
+ {
+ "epoch": 23.681198910081743,
+ "grad_norm": 7.843378067016602,
+ "learning_rate": 1.7839375105326743e-05,
+ "loss": 0.6271,
+ "step": 8691
+ },
+ {
+ "epoch": 23.68392370572207,
+ "grad_norm": 5.199438571929932,
+ "learning_rate": 1.7838827187063144e-05,
+ "loss": 0.4644,
+ "step": 8692
+ },
+ {
+ "epoch": 23.6866485013624,
+ "grad_norm": 5.25718355178833,
+ "learning_rate": 1.7838279207751013e-05,
+ "loss": 0.4468,
+ "step": 8693
+ },
+ {
+ "epoch": 23.689373297002724,
+ "grad_norm": 5.313921928405762,
+ "learning_rate": 1.7837731167394614e-05,
+ "loss": 0.6082,
+ "step": 8694
+ },
+ {
+ "epoch": 23.69209809264305,
+ "grad_norm": 4.857893943786621,
+ "learning_rate": 1.7837183065998215e-05,
+ "loss": 0.4332,
+ "step": 8695
+ },
+ {
+ "epoch": 23.69482288828338,
+ "grad_norm": 5.487356662750244,
+ "learning_rate": 1.783663490356609e-05,
+ "loss": 0.7845,
+ "step": 8696
+ },
+ {
+ "epoch": 23.697547683923705,
+ "grad_norm": 7.278802871704102,
+ "learning_rate": 1.78360866801025e-05,
+ "loss": 0.6034,
+ "step": 8697
+ },
+ {
+ "epoch": 23.70027247956403,
+ "grad_norm": 6.77960205078125,
+ "learning_rate": 1.783553839561172e-05,
+ "loss": 0.3517,
+ "step": 8698
+ },
+ {
+ "epoch": 23.70299727520436,
+ "grad_norm": 5.676662921905518,
+ "learning_rate": 1.7834990050098025e-05,
+ "loss": 0.5861,
+ "step": 8699
+ },
+ {
+ "epoch": 23.705722070844686,
+ "grad_norm": 5.575546741485596,
+ "learning_rate": 1.7834441643565674e-05,
+ "loss": 0.4057,
+ "step": 8700
+ },
+ {
+ "epoch": 23.708446866485012,
+ "grad_norm": 5.581966400146484,
+ "learning_rate": 1.7833893176018945e-05,
+ "loss": 0.4954,
+ "step": 8701
+ },
+ {
+ "epoch": 23.71117166212534,
+ "grad_norm": 5.427151203155518,
+ "learning_rate": 1.7833344647462106e-05,
+ "loss": 0.4133,
+ "step": 8702
+ },
+ {
+ "epoch": 23.713896457765667,
+ "grad_norm": 5.988466739654541,
+ "learning_rate": 1.7832796057899434e-05,
+ "loss": 0.3986,
+ "step": 8703
+ },
+ {
+ "epoch": 23.716621253405993,
+ "grad_norm": 5.697509765625,
+ "learning_rate": 1.7832247407335197e-05,
+ "loss": 0.4852,
+ "step": 8704
+ },
+ {
+ "epoch": 23.719346049046322,
+ "grad_norm": 5.190816402435303,
+ "learning_rate": 1.783169869577367e-05,
+ "loss": 0.4592,
+ "step": 8705
+ },
+ {
+ "epoch": 23.722070844686648,
+ "grad_norm": 4.988455772399902,
+ "learning_rate": 1.7831149923219126e-05,
+ "loss": 0.5578,
+ "step": 8706
+ },
+ {
+ "epoch": 23.724795640326974,
+ "grad_norm": 5.666781902313232,
+ "learning_rate": 1.783060108967584e-05,
+ "loss": 0.486,
+ "step": 8707
+ },
+ {
+ "epoch": 23.727520435967303,
+ "grad_norm": 7.143799781799316,
+ "learning_rate": 1.7830052195148083e-05,
+ "loss": 0.6008,
+ "step": 8708
+ },
+ {
+ "epoch": 23.73024523160763,
+ "grad_norm": 5.031253814697266,
+ "learning_rate": 1.782950323964013e-05,
+ "loss": 0.3386,
+ "step": 8709
+ },
+ {
+ "epoch": 23.732970027247955,
+ "grad_norm": 6.096299648284912,
+ "learning_rate": 1.7828954223156257e-05,
+ "loss": 0.4055,
+ "step": 8710
+ },
+ {
+ "epoch": 23.735694822888284,
+ "grad_norm": 6.645317077636719,
+ "learning_rate": 1.7828405145700743e-05,
+ "loss": 0.4108,
+ "step": 8711
+ },
+ {
+ "epoch": 23.73841961852861,
+ "grad_norm": 5.982500076293945,
+ "learning_rate": 1.7827856007277865e-05,
+ "loss": 0.4243,
+ "step": 8712
+ },
+ {
+ "epoch": 23.741144414168936,
+ "grad_norm": 5.8503522872924805,
+ "learning_rate": 1.782730680789189e-05,
+ "loss": 0.5005,
+ "step": 8713
+ },
+ {
+ "epoch": 23.743869209809265,
+ "grad_norm": 6.175079822540283,
+ "learning_rate": 1.78267575475471e-05,
+ "loss": 0.502,
+ "step": 8714
+ },
+ {
+ "epoch": 23.74659400544959,
+ "grad_norm": 6.095036029815674,
+ "learning_rate": 1.7826208226247783e-05,
+ "loss": 0.454,
+ "step": 8715
+ },
+ {
+ "epoch": 23.749318801089917,
+ "grad_norm": 5.226436614990234,
+ "learning_rate": 1.78256588439982e-05,
+ "loss": 0.3421,
+ "step": 8716
+ },
+ {
+ "epoch": 23.752043596730246,
+ "grad_norm": 7.435934066772461,
+ "learning_rate": 1.7825109400802643e-05,
+ "loss": 0.6492,
+ "step": 8717
+ },
+ {
+ "epoch": 23.754768392370572,
+ "grad_norm": 6.659297943115234,
+ "learning_rate": 1.7824559896665385e-05,
+ "loss": 0.4353,
+ "step": 8718
+ },
+ {
+ "epoch": 23.757493188010898,
+ "grad_norm": 7.91219425201416,
+ "learning_rate": 1.7824010331590703e-05,
+ "loss": 0.3616,
+ "step": 8719
+ },
+ {
+ "epoch": 23.760217983651227,
+ "grad_norm": 5.757700443267822,
+ "learning_rate": 1.7823460705582884e-05,
+ "loss": 0.4744,
+ "step": 8720
+ },
+ {
+ "epoch": 23.762942779291553,
+ "grad_norm": 5.67860746383667,
+ "learning_rate": 1.7822911018646202e-05,
+ "loss": 0.2822,
+ "step": 8721
+ },
+ {
+ "epoch": 23.76566757493188,
+ "grad_norm": 5.831265449523926,
+ "learning_rate": 1.782236127078494e-05,
+ "loss": 0.444,
+ "step": 8722
+ },
+ {
+ "epoch": 23.768392370572208,
+ "grad_norm": 5.632410526275635,
+ "learning_rate": 1.7821811462003387e-05,
+ "loss": 0.44,
+ "step": 8723
+ },
+ {
+ "epoch": 23.771117166212534,
+ "grad_norm": 5.530257225036621,
+ "learning_rate": 1.782126159230581e-05,
+ "loss": 0.407,
+ "step": 8724
+ },
+ {
+ "epoch": 23.77384196185286,
+ "grad_norm": 5.894223213195801,
+ "learning_rate": 1.7820711661696504e-05,
+ "loss": 0.4835,
+ "step": 8725
+ },
+ {
+ "epoch": 23.77656675749319,
+ "grad_norm": 5.536808013916016,
+ "learning_rate": 1.7820161670179748e-05,
+ "loss": 0.4334,
+ "step": 8726
+ },
+ {
+ "epoch": 23.779291553133515,
+ "grad_norm": 6.751597881317139,
+ "learning_rate": 1.7819611617759826e-05,
+ "loss": 0.4511,
+ "step": 8727
+ },
+ {
+ "epoch": 23.78201634877384,
+ "grad_norm": 5.278200149536133,
+ "learning_rate": 1.7819061504441016e-05,
+ "loss": 0.3793,
+ "step": 8728
+ },
+ {
+ "epoch": 23.78474114441417,
+ "grad_norm": 5.744964122772217,
+ "learning_rate": 1.7818511330227604e-05,
+ "loss": 0.5089,
+ "step": 8729
+ },
+ {
+ "epoch": 23.787465940054496,
+ "grad_norm": 5.106729507446289,
+ "learning_rate": 1.7817961095123885e-05,
+ "loss": 0.4807,
+ "step": 8730
+ },
+ {
+ "epoch": 23.79019073569482,
+ "grad_norm": 7.729740142822266,
+ "learning_rate": 1.7817410799134133e-05,
+ "loss": 0.6862,
+ "step": 8731
+ },
+ {
+ "epoch": 23.79291553133515,
+ "grad_norm": 5.469242572784424,
+ "learning_rate": 1.7816860442262637e-05,
+ "loss": 0.4816,
+ "step": 8732
+ },
+ {
+ "epoch": 23.795640326975477,
+ "grad_norm": 6.5888261795043945,
+ "learning_rate": 1.781631002451368e-05,
+ "loss": 0.5001,
+ "step": 8733
+ },
+ {
+ "epoch": 23.798365122615802,
+ "grad_norm": 6.336596488952637,
+ "learning_rate": 1.7815759545891556e-05,
+ "loss": 0.4156,
+ "step": 8734
+ },
+ {
+ "epoch": 23.80108991825613,
+ "grad_norm": 5.422203540802002,
+ "learning_rate": 1.781520900640055e-05,
+ "loss": 0.3983,
+ "step": 8735
+ },
+ {
+ "epoch": 23.803814713896458,
+ "grad_norm": 5.620798587799072,
+ "learning_rate": 1.7814658406044948e-05,
+ "loss": 0.5104,
+ "step": 8736
+ },
+ {
+ "epoch": 23.806539509536783,
+ "grad_norm": 5.943844318389893,
+ "learning_rate": 1.7814107744829036e-05,
+ "loss": 0.4641,
+ "step": 8737
+ },
+ {
+ "epoch": 23.809264305177113,
+ "grad_norm": 6.198447227478027,
+ "learning_rate": 1.7813557022757103e-05,
+ "loss": 0.5462,
+ "step": 8738
+ },
+ {
+ "epoch": 23.81198910081744,
+ "grad_norm": 6.402525901794434,
+ "learning_rate": 1.781300623983344e-05,
+ "loss": 0.4212,
+ "step": 8739
+ },
+ {
+ "epoch": 23.814713896457764,
+ "grad_norm": 5.113837242126465,
+ "learning_rate": 1.7812455396062336e-05,
+ "loss": 0.4707,
+ "step": 8740
+ },
+ {
+ "epoch": 23.817438692098094,
+ "grad_norm": 4.764026641845703,
+ "learning_rate": 1.781190449144808e-05,
+ "loss": 0.4246,
+ "step": 8741
+ },
+ {
+ "epoch": 23.82016348773842,
+ "grad_norm": 5.452980041503906,
+ "learning_rate": 1.7811353525994967e-05,
+ "loss": 0.4612,
+ "step": 8742
+ },
+ {
+ "epoch": 23.822888283378745,
+ "grad_norm": 5.345305442810059,
+ "learning_rate": 1.781080249970728e-05,
+ "loss": 0.5027,
+ "step": 8743
+ },
+ {
+ "epoch": 23.825613079019075,
+ "grad_norm": 6.897902011871338,
+ "learning_rate": 1.7810251412589315e-05,
+ "loss": 0.4415,
+ "step": 8744
+ },
+ {
+ "epoch": 23.8283378746594,
+ "grad_norm": 5.209244728088379,
+ "learning_rate": 1.7809700264645365e-05,
+ "loss": 0.5408,
+ "step": 8745
+ },
+ {
+ "epoch": 23.831062670299726,
+ "grad_norm": 5.518956661224365,
+ "learning_rate": 1.780914905587972e-05,
+ "loss": 0.4904,
+ "step": 8746
+ },
+ {
+ "epoch": 23.833787465940055,
+ "grad_norm": 6.317090034484863,
+ "learning_rate": 1.7808597786296673e-05,
+ "loss": 0.3757,
+ "step": 8747
+ },
+ {
+ "epoch": 23.83651226158038,
+ "grad_norm": 5.908511638641357,
+ "learning_rate": 1.7808046455900517e-05,
+ "loss": 0.4675,
+ "step": 8748
+ },
+ {
+ "epoch": 23.839237057220707,
+ "grad_norm": 5.465956687927246,
+ "learning_rate": 1.780749506469555e-05,
+ "loss": 0.5063,
+ "step": 8749
+ },
+ {
+ "epoch": 23.841961852861036,
+ "grad_norm": 6.86112117767334,
+ "learning_rate": 1.7806943612686058e-05,
+ "loss": 0.5051,
+ "step": 8750
+ },
+ {
+ "epoch": 23.844686648501362,
+ "grad_norm": 5.167027950286865,
+ "learning_rate": 1.7806392099876343e-05,
+ "loss": 0.4423,
+ "step": 8751
+ },
+ {
+ "epoch": 23.847411444141688,
+ "grad_norm": 7.360213756561279,
+ "learning_rate": 1.78058405262707e-05,
+ "loss": 0.5656,
+ "step": 8752
+ },
+ {
+ "epoch": 23.850136239782017,
+ "grad_norm": 4.88274621963501,
+ "learning_rate": 1.780528889187342e-05,
+ "loss": 0.5758,
+ "step": 8753
+ },
+ {
+ "epoch": 23.852861035422343,
+ "grad_norm": 4.911193370819092,
+ "learning_rate": 1.78047371966888e-05,
+ "loss": 0.4261,
+ "step": 8754
+ },
+ {
+ "epoch": 23.85558583106267,
+ "grad_norm": 5.769683837890625,
+ "learning_rate": 1.7804185440721136e-05,
+ "loss": 0.3593,
+ "step": 8755
+ },
+ {
+ "epoch": 23.858310626703,
+ "grad_norm": 4.997826099395752,
+ "learning_rate": 1.7803633623974732e-05,
+ "loss": 0.4402,
+ "step": 8756
+ },
+ {
+ "epoch": 23.861035422343324,
+ "grad_norm": 5.453546047210693,
+ "learning_rate": 1.7803081746453877e-05,
+ "loss": 0.3427,
+ "step": 8757
+ },
+ {
+ "epoch": 23.86376021798365,
+ "grad_norm": 5.951051235198975,
+ "learning_rate": 1.7802529808162877e-05,
+ "loss": 0.441,
+ "step": 8758
+ },
+ {
+ "epoch": 23.86648501362398,
+ "grad_norm": 9.82780933380127,
+ "learning_rate": 1.7801977809106024e-05,
+ "loss": 0.6013,
+ "step": 8759
+ },
+ {
+ "epoch": 23.869209809264305,
+ "grad_norm": 6.326391220092773,
+ "learning_rate": 1.7801425749287616e-05,
+ "loss": 0.524,
+ "step": 8760
+ },
+ {
+ "epoch": 23.87193460490463,
+ "grad_norm": 4.835983753204346,
+ "learning_rate": 1.7800873628711957e-05,
+ "loss": 0.5193,
+ "step": 8761
+ },
+ {
+ "epoch": 23.87465940054496,
+ "grad_norm": 6.248645782470703,
+ "learning_rate": 1.7800321447383345e-05,
+ "loss": 0.711,
+ "step": 8762
+ },
+ {
+ "epoch": 23.877384196185286,
+ "grad_norm": 5.30921745300293,
+ "learning_rate": 1.7799769205306082e-05,
+ "loss": 0.6008,
+ "step": 8763
+ },
+ {
+ "epoch": 23.88010899182561,
+ "grad_norm": 8.549253463745117,
+ "learning_rate": 1.7799216902484465e-05,
+ "loss": 0.4142,
+ "step": 8764
+ },
+ {
+ "epoch": 23.88283378746594,
+ "grad_norm": 5.522342205047607,
+ "learning_rate": 1.77986645389228e-05,
+ "loss": 0.3609,
+ "step": 8765
+ },
+ {
+ "epoch": 23.885558583106267,
+ "grad_norm": 5.234336853027344,
+ "learning_rate": 1.7798112114625385e-05,
+ "loss": 0.6215,
+ "step": 8766
+ },
+ {
+ "epoch": 23.888283378746593,
+ "grad_norm": 5.513620376586914,
+ "learning_rate": 1.7797559629596525e-05,
+ "loss": 0.3922,
+ "step": 8767
+ },
+ {
+ "epoch": 23.891008174386922,
+ "grad_norm": 6.012533664703369,
+ "learning_rate": 1.7797007083840522e-05,
+ "loss": 0.5626,
+ "step": 8768
+ },
+ {
+ "epoch": 23.893732970027248,
+ "grad_norm": 5.3587141036987305,
+ "learning_rate": 1.7796454477361676e-05,
+ "loss": 0.5891,
+ "step": 8769
+ },
+ {
+ "epoch": 23.896457765667574,
+ "grad_norm": 6.47929573059082,
+ "learning_rate": 1.7795901810164296e-05,
+ "loss": 0.3947,
+ "step": 8770
+ },
+ {
+ "epoch": 23.899182561307903,
+ "grad_norm": 5.516554355621338,
+ "learning_rate": 1.7795349082252686e-05,
+ "loss": 0.4537,
+ "step": 8771
+ },
+ {
+ "epoch": 23.90190735694823,
+ "grad_norm": 6.426537036895752,
+ "learning_rate": 1.7794796293631144e-05,
+ "loss": 0.4794,
+ "step": 8772
+ },
+ {
+ "epoch": 23.904632152588555,
+ "grad_norm": 6.000232219696045,
+ "learning_rate": 1.779424344430398e-05,
+ "loss": 0.4282,
+ "step": 8773
+ },
+ {
+ "epoch": 23.907356948228884,
+ "grad_norm": 5.953726768493652,
+ "learning_rate": 1.77936905342755e-05,
+ "loss": 0.4272,
+ "step": 8774
+ },
+ {
+ "epoch": 23.91008174386921,
+ "grad_norm": 4.6150994300842285,
+ "learning_rate": 1.7793137563550006e-05,
+ "loss": 0.4537,
+ "step": 8775
+ },
+ {
+ "epoch": 23.912806539509535,
+ "grad_norm": 9.116332054138184,
+ "learning_rate": 1.7792584532131813e-05,
+ "loss": 0.4112,
+ "step": 8776
+ },
+ {
+ "epoch": 23.915531335149865,
+ "grad_norm": 5.997339248657227,
+ "learning_rate": 1.779203144002522e-05,
+ "loss": 0.5474,
+ "step": 8777
+ },
+ {
+ "epoch": 23.91825613079019,
+ "grad_norm": 6.094305992126465,
+ "learning_rate": 1.7791478287234534e-05,
+ "loss": 0.4313,
+ "step": 8778
+ },
+ {
+ "epoch": 23.920980926430516,
+ "grad_norm": 5.7967143058776855,
+ "learning_rate": 1.779092507376407e-05,
+ "loss": 0.473,
+ "step": 8779
+ },
+ {
+ "epoch": 23.923705722070846,
+ "grad_norm": 5.653780460357666,
+ "learning_rate": 1.779037179961813e-05,
+ "loss": 0.5735,
+ "step": 8780
+ },
+ {
+ "epoch": 23.92643051771117,
+ "grad_norm": 6.708689212799072,
+ "learning_rate": 1.7789818464801025e-05,
+ "loss": 0.5555,
+ "step": 8781
+ },
+ {
+ "epoch": 23.929155313351497,
+ "grad_norm": 5.507002830505371,
+ "learning_rate": 1.7789265069317067e-05,
+ "loss": 0.4243,
+ "step": 8782
+ },
+ {
+ "epoch": 23.931880108991827,
+ "grad_norm": 6.126735687255859,
+ "learning_rate": 1.778871161317056e-05,
+ "loss": 0.4241,
+ "step": 8783
+ },
+ {
+ "epoch": 23.934604904632153,
+ "grad_norm": 6.6175127029418945,
+ "learning_rate": 1.778815809636582e-05,
+ "loss": 0.4376,
+ "step": 8784
+ },
+ {
+ "epoch": 23.93732970027248,
+ "grad_norm": 5.631956577301025,
+ "learning_rate": 1.7787604518907152e-05,
+ "loss": 0.7006,
+ "step": 8785
+ },
+ {
+ "epoch": 23.940054495912808,
+ "grad_norm": 6.06506872177124,
+ "learning_rate": 1.7787050880798877e-05,
+ "loss": 0.4412,
+ "step": 8786
+ },
+ {
+ "epoch": 23.942779291553133,
+ "grad_norm": 5.805269718170166,
+ "learning_rate": 1.7786497182045295e-05,
+ "loss": 0.4713,
+ "step": 8787
+ },
+ {
+ "epoch": 23.94550408719346,
+ "grad_norm": 7.573705196380615,
+ "learning_rate": 1.7785943422650724e-05,
+ "loss": 0.3913,
+ "step": 8788
+ },
+ {
+ "epoch": 23.94822888283379,
+ "grad_norm": 6.195974826812744,
+ "learning_rate": 1.7785389602619477e-05,
+ "loss": 0.3799,
+ "step": 8789
+ },
+ {
+ "epoch": 23.950953678474114,
+ "grad_norm": 4.9317851066589355,
+ "learning_rate": 1.7784835721955866e-05,
+ "loss": 0.4276,
+ "step": 8790
+ },
+ {
+ "epoch": 23.95367847411444,
+ "grad_norm": 9.827341079711914,
+ "learning_rate": 1.778428178066421e-05,
+ "loss": 0.5682,
+ "step": 8791
+ },
+ {
+ "epoch": 23.95640326975477,
+ "grad_norm": 5.874907493591309,
+ "learning_rate": 1.778372777874881e-05,
+ "loss": 0.416,
+ "step": 8792
+ },
+ {
+ "epoch": 23.959128065395095,
+ "grad_norm": 4.897655487060547,
+ "learning_rate": 1.7783173716213996e-05,
+ "loss": 0.4318,
+ "step": 8793
+ },
+ {
+ "epoch": 23.96185286103542,
+ "grad_norm": 6.086601734161377,
+ "learning_rate": 1.778261959306407e-05,
+ "loss": 0.5375,
+ "step": 8794
+ },
+ {
+ "epoch": 23.96457765667575,
+ "grad_norm": 6.121303558349609,
+ "learning_rate": 1.7782065409303354e-05,
+ "loss": 0.4693,
+ "step": 8795
+ },
+ {
+ "epoch": 23.967302452316076,
+ "grad_norm": 5.466486930847168,
+ "learning_rate": 1.7781511164936166e-05,
+ "loss": 0.4069,
+ "step": 8796
+ },
+ {
+ "epoch": 23.970027247956402,
+ "grad_norm": 5.532132148742676,
+ "learning_rate": 1.7780956859966816e-05,
+ "loss": 0.3607,
+ "step": 8797
+ },
+ {
+ "epoch": 23.97275204359673,
+ "grad_norm": 5.8270697593688965,
+ "learning_rate": 1.7780402494399625e-05,
+ "loss": 0.4694,
+ "step": 8798
+ },
+ {
+ "epoch": 23.975476839237057,
+ "grad_norm": 5.881699562072754,
+ "learning_rate": 1.777984806823891e-05,
+ "loss": 0.7293,
+ "step": 8799
+ },
+ {
+ "epoch": 23.978201634877383,
+ "grad_norm": 5.9827880859375,
+ "learning_rate": 1.777929358148899e-05,
+ "loss": 0.3498,
+ "step": 8800
+ },
+ {
+ "epoch": 23.980926430517712,
+ "grad_norm": 5.319509983062744,
+ "learning_rate": 1.777873903415418e-05,
+ "loss": 0.4287,
+ "step": 8801
+ },
+ {
+ "epoch": 23.983651226158038,
+ "grad_norm": 5.021702289581299,
+ "learning_rate": 1.7778184426238805e-05,
+ "loss": 0.5154,
+ "step": 8802
+ },
+ {
+ "epoch": 23.986376021798364,
+ "grad_norm": 4.681661128997803,
+ "learning_rate": 1.7777629757747177e-05,
+ "loss": 0.5037,
+ "step": 8803
+ },
+ {
+ "epoch": 23.989100817438693,
+ "grad_norm": 5.3313751220703125,
+ "learning_rate": 1.7777075028683617e-05,
+ "loss": 0.3304,
+ "step": 8804
+ },
+ {
+ "epoch": 23.99182561307902,
+ "grad_norm": 9.241132736206055,
+ "learning_rate": 1.7776520239052447e-05,
+ "loss": 0.4781,
+ "step": 8805
+ },
+ {
+ "epoch": 23.994550408719345,
+ "grad_norm": 4.792355060577393,
+ "learning_rate": 1.777596538885799e-05,
+ "loss": 0.5277,
+ "step": 8806
+ },
+ {
+ "epoch": 23.997275204359674,
+ "grad_norm": 5.029292106628418,
+ "learning_rate": 1.7775410478104566e-05,
+ "loss": 0.5368,
+ "step": 8807
+ },
+ {
+ "epoch": 24.0,
+ "grad_norm": 5.854164123535156,
+ "learning_rate": 1.7774855506796497e-05,
+ "loss": 0.4556,
+ "step": 8808
+ },
+ {
+ "epoch": 24.002724795640326,
+ "grad_norm": 6.256642818450928,
+ "learning_rate": 1.77743004749381e-05,
+ "loss": 0.5337,
+ "step": 8809
+ },
+ {
+ "epoch": 24.005449591280655,
+ "grad_norm": 5.631007671356201,
+ "learning_rate": 1.77737453825337e-05,
+ "loss": 0.3318,
+ "step": 8810
+ },
+ {
+ "epoch": 24.00817438692098,
+ "grad_norm": 4.717926025390625,
+ "learning_rate": 1.7773190229587624e-05,
+ "loss": 0.459,
+ "step": 8811
+ },
+ {
+ "epoch": 24.010899182561307,
+ "grad_norm": 5.650022506713867,
+ "learning_rate": 1.777263501610419e-05,
+ "loss": 0.3588,
+ "step": 8812
+ },
+ {
+ "epoch": 24.013623978201636,
+ "grad_norm": 5.280185222625732,
+ "learning_rate": 1.7772079742087728e-05,
+ "loss": 0.4258,
+ "step": 8813
+ },
+ {
+ "epoch": 24.016348773841962,
+ "grad_norm": 5.713511943817139,
+ "learning_rate": 1.7771524407542557e-05,
+ "loss": 0.3028,
+ "step": 8814
+ },
+ {
+ "epoch": 24.019073569482288,
+ "grad_norm": 5.299055576324463,
+ "learning_rate": 1.7770969012473002e-05,
+ "loss": 0.3254,
+ "step": 8815
+ },
+ {
+ "epoch": 24.021798365122617,
+ "grad_norm": 4.9886064529418945,
+ "learning_rate": 1.777041355688339e-05,
+ "loss": 0.3936,
+ "step": 8816
+ },
+ {
+ "epoch": 24.024523160762943,
+ "grad_norm": 5.939755439758301,
+ "learning_rate": 1.7769858040778052e-05,
+ "loss": 0.3662,
+ "step": 8817
+ },
+ {
+ "epoch": 24.02724795640327,
+ "grad_norm": 4.548478603363037,
+ "learning_rate": 1.7769302464161307e-05,
+ "loss": 0.4057,
+ "step": 8818
+ },
+ {
+ "epoch": 24.029972752043598,
+ "grad_norm": 5.712402820587158,
+ "learning_rate": 1.7768746827037485e-05,
+ "loss": 0.5594,
+ "step": 8819
+ },
+ {
+ "epoch": 24.032697547683924,
+ "grad_norm": 5.726139068603516,
+ "learning_rate": 1.7768191129410913e-05,
+ "loss": 0.3629,
+ "step": 8820
+ },
+ {
+ "epoch": 24.03542234332425,
+ "grad_norm": 5.00911808013916,
+ "learning_rate": 1.776763537128592e-05,
+ "loss": 0.4583,
+ "step": 8821
+ },
+ {
+ "epoch": 24.03814713896458,
+ "grad_norm": 5.051666736602783,
+ "learning_rate": 1.7767079552666827e-05,
+ "loss": 0.4148,
+ "step": 8822
+ },
+ {
+ "epoch": 24.040871934604905,
+ "grad_norm": 5.41791296005249,
+ "learning_rate": 1.7766523673557973e-05,
+ "loss": 0.4785,
+ "step": 8823
+ },
+ {
+ "epoch": 24.04359673024523,
+ "grad_norm": 5.533160209655762,
+ "learning_rate": 1.7765967733963683e-05,
+ "loss": 0.4858,
+ "step": 8824
+ },
+ {
+ "epoch": 24.04632152588556,
+ "grad_norm": 5.127633094787598,
+ "learning_rate": 1.7765411733888286e-05,
+ "loss": 0.33,
+ "step": 8825
+ },
+ {
+ "epoch": 24.049046321525886,
+ "grad_norm": 5.57954740524292,
+ "learning_rate": 1.776485567333611e-05,
+ "loss": 0.5783,
+ "step": 8826
+ },
+ {
+ "epoch": 24.05177111716621,
+ "grad_norm": 5.173678874969482,
+ "learning_rate": 1.776429955231149e-05,
+ "loss": 0.3175,
+ "step": 8827
+ },
+ {
+ "epoch": 24.05449591280654,
+ "grad_norm": 4.745274543762207,
+ "learning_rate": 1.7763743370818754e-05,
+ "loss": 0.3666,
+ "step": 8828
+ },
+ {
+ "epoch": 24.057220708446867,
+ "grad_norm": 5.103257656097412,
+ "learning_rate": 1.7763187128862237e-05,
+ "loss": 0.3499,
+ "step": 8829
+ },
+ {
+ "epoch": 24.059945504087192,
+ "grad_norm": 4.750213146209717,
+ "learning_rate": 1.7762630826446266e-05,
+ "loss": 0.4097,
+ "step": 8830
+ },
+ {
+ "epoch": 24.06267029972752,
+ "grad_norm": 5.353237628936768,
+ "learning_rate": 1.7762074463575175e-05,
+ "loss": 0.3795,
+ "step": 8831
+ },
+ {
+ "epoch": 24.065395095367847,
+ "grad_norm": 5.203042507171631,
+ "learning_rate": 1.77615180402533e-05,
+ "loss": 0.3544,
+ "step": 8832
+ },
+ {
+ "epoch": 24.068119891008173,
+ "grad_norm": 4.6664509773254395,
+ "learning_rate": 1.776096155648497e-05,
+ "loss": 0.3633,
+ "step": 8833
+ },
+ {
+ "epoch": 24.070844686648503,
+ "grad_norm": 5.730208396911621,
+ "learning_rate": 1.7760405012274524e-05,
+ "loss": 0.5496,
+ "step": 8834
+ },
+ {
+ "epoch": 24.07356948228883,
+ "grad_norm": 4.194002151489258,
+ "learning_rate": 1.775984840762629e-05,
+ "loss": 0.4095,
+ "step": 8835
+ },
+ {
+ "epoch": 24.076294277929154,
+ "grad_norm": 4.474147796630859,
+ "learning_rate": 1.775929174254461e-05,
+ "loss": 0.4277,
+ "step": 8836
+ },
+ {
+ "epoch": 24.079019073569484,
+ "grad_norm": 6.178847789764404,
+ "learning_rate": 1.7758735017033815e-05,
+ "loss": 0.3637,
+ "step": 8837
+ },
+ {
+ "epoch": 24.08174386920981,
+ "grad_norm": 5.144662380218506,
+ "learning_rate": 1.775817823109824e-05,
+ "loss": 0.3756,
+ "step": 8838
+ },
+ {
+ "epoch": 24.084468664850135,
+ "grad_norm": 5.001590728759766,
+ "learning_rate": 1.7757621384742224e-05,
+ "loss": 0.3719,
+ "step": 8839
+ },
+ {
+ "epoch": 24.087193460490465,
+ "grad_norm": 5.672701358795166,
+ "learning_rate": 1.7757064477970102e-05,
+ "loss": 0.3835,
+ "step": 8840
+ },
+ {
+ "epoch": 24.08991825613079,
+ "grad_norm": 5.297117233276367,
+ "learning_rate": 1.775650751078621e-05,
+ "loss": 0.3966,
+ "step": 8841
+ },
+ {
+ "epoch": 24.092643051771116,
+ "grad_norm": 6.071321487426758,
+ "learning_rate": 1.775595048319489e-05,
+ "loss": 0.2974,
+ "step": 8842
+ },
+ {
+ "epoch": 24.095367847411445,
+ "grad_norm": 5.990397930145264,
+ "learning_rate": 1.7755393395200474e-05,
+ "loss": 0.3459,
+ "step": 8843
+ },
+ {
+ "epoch": 24.09809264305177,
+ "grad_norm": 5.174765110015869,
+ "learning_rate": 1.7754836246807308e-05,
+ "loss": 0.3426,
+ "step": 8844
+ },
+ {
+ "epoch": 24.100817438692097,
+ "grad_norm": 5.264282703399658,
+ "learning_rate": 1.775427903801972e-05,
+ "loss": 0.6276,
+ "step": 8845
+ },
+ {
+ "epoch": 24.103542234332426,
+ "grad_norm": 6.398179531097412,
+ "learning_rate": 1.7753721768842062e-05,
+ "loss": 0.5856,
+ "step": 8846
+ },
+ {
+ "epoch": 24.106267029972752,
+ "grad_norm": 4.774051666259766,
+ "learning_rate": 1.775316443927867e-05,
+ "loss": 0.4257,
+ "step": 8847
+ },
+ {
+ "epoch": 24.108991825613078,
+ "grad_norm": 6.434476852416992,
+ "learning_rate": 1.7752607049333878e-05,
+ "loss": 0.4412,
+ "step": 8848
+ },
+ {
+ "epoch": 24.111716621253407,
+ "grad_norm": 6.50593900680542,
+ "learning_rate": 1.7752049599012037e-05,
+ "loss": 0.4909,
+ "step": 8849
+ },
+ {
+ "epoch": 24.114441416893733,
+ "grad_norm": 5.552506923675537,
+ "learning_rate": 1.775149208831748e-05,
+ "loss": 0.4393,
+ "step": 8850
+ },
+ {
+ "epoch": 24.11716621253406,
+ "grad_norm": 5.510268211364746,
+ "learning_rate": 1.775093451725455e-05,
+ "loss": 0.2821,
+ "step": 8851
+ },
+ {
+ "epoch": 24.11989100817439,
+ "grad_norm": 5.341363430023193,
+ "learning_rate": 1.775037688582759e-05,
+ "loss": 0.3547,
+ "step": 8852
+ },
+ {
+ "epoch": 24.122615803814714,
+ "grad_norm": 6.147729396820068,
+ "learning_rate": 1.7749819194040953e-05,
+ "loss": 0.4425,
+ "step": 8853
+ },
+ {
+ "epoch": 24.12534059945504,
+ "grad_norm": 4.967210292816162,
+ "learning_rate": 1.774926144189897e-05,
+ "loss": 0.316,
+ "step": 8854
+ },
+ {
+ "epoch": 24.12806539509537,
+ "grad_norm": 6.360644340515137,
+ "learning_rate": 1.7748703629405985e-05,
+ "loss": 0.4238,
+ "step": 8855
+ },
+ {
+ "epoch": 24.130790190735695,
+ "grad_norm": 5.806127548217773,
+ "learning_rate": 1.7748145756566346e-05,
+ "loss": 0.45,
+ "step": 8856
+ },
+ {
+ "epoch": 24.13351498637602,
+ "grad_norm": 5.187007427215576,
+ "learning_rate": 1.7747587823384398e-05,
+ "loss": 0.4183,
+ "step": 8857
+ },
+ {
+ "epoch": 24.13623978201635,
+ "grad_norm": 5.072420120239258,
+ "learning_rate": 1.7747029829864485e-05,
+ "loss": 0.3031,
+ "step": 8858
+ },
+ {
+ "epoch": 24.138964577656676,
+ "grad_norm": 6.479490756988525,
+ "learning_rate": 1.7746471776010953e-05,
+ "loss": 0.4295,
+ "step": 8859
+ },
+ {
+ "epoch": 24.141689373297,
+ "grad_norm": 4.955697536468506,
+ "learning_rate": 1.7745913661828148e-05,
+ "loss": 0.5367,
+ "step": 8860
+ },
+ {
+ "epoch": 24.14441416893733,
+ "grad_norm": 5.075957775115967,
+ "learning_rate": 1.7745355487320418e-05,
+ "loss": 0.3658,
+ "step": 8861
+ },
+ {
+ "epoch": 24.147138964577657,
+ "grad_norm": 5.295302391052246,
+ "learning_rate": 1.774479725249211e-05,
+ "loss": 0.2811,
+ "step": 8862
+ },
+ {
+ "epoch": 24.149863760217983,
+ "grad_norm": 5.116213798522949,
+ "learning_rate": 1.774423895734757e-05,
+ "loss": 0.3062,
+ "step": 8863
+ },
+ {
+ "epoch": 24.152588555858312,
+ "grad_norm": 5.51216459274292,
+ "learning_rate": 1.7743680601891146e-05,
+ "loss": 0.3251,
+ "step": 8864
+ },
+ {
+ "epoch": 24.155313351498638,
+ "grad_norm": 4.585820198059082,
+ "learning_rate": 1.7743122186127186e-05,
+ "loss": 0.3418,
+ "step": 8865
+ },
+ {
+ "epoch": 24.158038147138964,
+ "grad_norm": 5.851943016052246,
+ "learning_rate": 1.7742563710060036e-05,
+ "loss": 0.5404,
+ "step": 8866
+ },
+ {
+ "epoch": 24.160762942779293,
+ "grad_norm": 4.86922025680542,
+ "learning_rate": 1.774200517369405e-05,
+ "loss": 0.3346,
+ "step": 8867
+ },
+ {
+ "epoch": 24.16348773841962,
+ "grad_norm": 5.692673683166504,
+ "learning_rate": 1.7741446577033584e-05,
+ "loss": 0.5404,
+ "step": 8868
+ },
+ {
+ "epoch": 24.166212534059945,
+ "grad_norm": 5.4698638916015625,
+ "learning_rate": 1.7740887920082975e-05,
+ "loss": 0.4047,
+ "step": 8869
+ },
+ {
+ "epoch": 24.168937329700274,
+ "grad_norm": 5.149357318878174,
+ "learning_rate": 1.774032920284658e-05,
+ "loss": 0.3208,
+ "step": 8870
+ },
+ {
+ "epoch": 24.1716621253406,
+ "grad_norm": 4.9086127281188965,
+ "learning_rate": 1.7739770425328755e-05,
+ "loss": 0.3737,
+ "step": 8871
+ },
+ {
+ "epoch": 24.174386920980925,
+ "grad_norm": 5.200900554656982,
+ "learning_rate": 1.773921158753384e-05,
+ "loss": 0.3432,
+ "step": 8872
+ },
+ {
+ "epoch": 24.177111716621255,
+ "grad_norm": 5.483757019042969,
+ "learning_rate": 1.77386526894662e-05,
+ "loss": 0.2609,
+ "step": 8873
+ },
+ {
+ "epoch": 24.17983651226158,
+ "grad_norm": 5.298186779022217,
+ "learning_rate": 1.773809373113018e-05,
+ "loss": 0.4041,
+ "step": 8874
+ },
+ {
+ "epoch": 24.182561307901906,
+ "grad_norm": 5.345581531524658,
+ "learning_rate": 1.7737534712530133e-05,
+ "loss": 0.2676,
+ "step": 8875
+ },
+ {
+ "epoch": 24.185286103542236,
+ "grad_norm": 4.5108747482299805,
+ "learning_rate": 1.7736975633670418e-05,
+ "loss": 0.4271,
+ "step": 8876
+ },
+ {
+ "epoch": 24.18801089918256,
+ "grad_norm": 4.346174240112305,
+ "learning_rate": 1.7736416494555384e-05,
+ "loss": 0.4348,
+ "step": 8877
+ },
+ {
+ "epoch": 24.190735694822887,
+ "grad_norm": 6.506301403045654,
+ "learning_rate": 1.773585729518939e-05,
+ "loss": 0.3784,
+ "step": 8878
+ },
+ {
+ "epoch": 24.193460490463217,
+ "grad_norm": 5.314094543457031,
+ "learning_rate": 1.7735298035576784e-05,
+ "loss": 0.403,
+ "step": 8879
+ },
+ {
+ "epoch": 24.196185286103542,
+ "grad_norm": 4.769566059112549,
+ "learning_rate": 1.7734738715721928e-05,
+ "loss": 0.3085,
+ "step": 8880
+ },
+ {
+ "epoch": 24.19891008174387,
+ "grad_norm": 5.417290210723877,
+ "learning_rate": 1.7734179335629175e-05,
+ "loss": 0.4977,
+ "step": 8881
+ },
+ {
+ "epoch": 24.201634877384198,
+ "grad_norm": 4.615503311157227,
+ "learning_rate": 1.7733619895302886e-05,
+ "loss": 0.3212,
+ "step": 8882
+ },
+ {
+ "epoch": 24.204359673024523,
+ "grad_norm": 5.541677951812744,
+ "learning_rate": 1.773306039474741e-05,
+ "loss": 0.3639,
+ "step": 8883
+ },
+ {
+ "epoch": 24.20708446866485,
+ "grad_norm": 4.97153902053833,
+ "learning_rate": 1.7732500833967112e-05,
+ "loss": 0.3338,
+ "step": 8884
+ },
+ {
+ "epoch": 24.20980926430518,
+ "grad_norm": 6.520368576049805,
+ "learning_rate": 1.773194121296634e-05,
+ "loss": 0.3954,
+ "step": 8885
+ },
+ {
+ "epoch": 24.212534059945504,
+ "grad_norm": 6.227540969848633,
+ "learning_rate": 1.7731381531749465e-05,
+ "loss": 0.4011,
+ "step": 8886
+ },
+ {
+ "epoch": 24.21525885558583,
+ "grad_norm": 5.718157768249512,
+ "learning_rate": 1.7730821790320836e-05,
+ "loss": 0.5,
+ "step": 8887
+ },
+ {
+ "epoch": 24.21798365122616,
+ "grad_norm": 5.290030002593994,
+ "learning_rate": 1.7730261988684817e-05,
+ "loss": 0.2841,
+ "step": 8888
+ },
+ {
+ "epoch": 24.220708446866485,
+ "grad_norm": 5.775897979736328,
+ "learning_rate": 1.772970212684577e-05,
+ "loss": 0.2852,
+ "step": 8889
+ },
+ {
+ "epoch": 24.22343324250681,
+ "grad_norm": 4.58942174911499,
+ "learning_rate": 1.7729142204808044e-05,
+ "loss": 0.6225,
+ "step": 8890
+ },
+ {
+ "epoch": 24.22615803814714,
+ "grad_norm": 5.221415042877197,
+ "learning_rate": 1.7728582222576012e-05,
+ "loss": 0.4288,
+ "step": 8891
+ },
+ {
+ "epoch": 24.228882833787466,
+ "grad_norm": 6.372081279754639,
+ "learning_rate": 1.7728022180154032e-05,
+ "loss": 0.3987,
+ "step": 8892
+ },
+ {
+ "epoch": 24.231607629427792,
+ "grad_norm": 4.113396644592285,
+ "learning_rate": 1.772746207754646e-05,
+ "loss": 0.3642,
+ "step": 8893
+ },
+ {
+ "epoch": 24.23433242506812,
+ "grad_norm": 5.77240514755249,
+ "learning_rate": 1.7726901914757664e-05,
+ "loss": 0.4073,
+ "step": 8894
+ },
+ {
+ "epoch": 24.237057220708447,
+ "grad_norm": 6.256892681121826,
+ "learning_rate": 1.7726341691792007e-05,
+ "loss": 0.3926,
+ "step": 8895
+ },
+ {
+ "epoch": 24.239782016348773,
+ "grad_norm": 5.544714450836182,
+ "learning_rate": 1.7725781408653844e-05,
+ "loss": 0.2644,
+ "step": 8896
+ },
+ {
+ "epoch": 24.242506811989102,
+ "grad_norm": 5.274021148681641,
+ "learning_rate": 1.772522106534755e-05,
+ "loss": 0.33,
+ "step": 8897
+ },
+ {
+ "epoch": 24.245231607629428,
+ "grad_norm": 5.773635387420654,
+ "learning_rate": 1.7724660661877478e-05,
+ "loss": 0.3185,
+ "step": 8898
+ },
+ {
+ "epoch": 24.247956403269754,
+ "grad_norm": 6.1496710777282715,
+ "learning_rate": 1.7724100198248e-05,
+ "loss": 0.4512,
+ "step": 8899
+ },
+ {
+ "epoch": 24.250681198910083,
+ "grad_norm": 5.14358377456665,
+ "learning_rate": 1.7723539674463473e-05,
+ "loss": 0.5068,
+ "step": 8900
+ },
+ {
+ "epoch": 24.25340599455041,
+ "grad_norm": 7.764667987823486,
+ "learning_rate": 1.7722979090528275e-05,
+ "loss": 0.5191,
+ "step": 8901
+ },
+ {
+ "epoch": 24.256130790190735,
+ "grad_norm": 5.219480037689209,
+ "learning_rate": 1.7722418446446764e-05,
+ "loss": 0.4523,
+ "step": 8902
+ },
+ {
+ "epoch": 24.258855585831064,
+ "grad_norm": 5.546377182006836,
+ "learning_rate": 1.7721857742223302e-05,
+ "loss": 0.3621,
+ "step": 8903
+ },
+ {
+ "epoch": 24.26158038147139,
+ "grad_norm": 5.009552001953125,
+ "learning_rate": 1.7721296977862263e-05,
+ "loss": 0.4973,
+ "step": 8904
+ },
+ {
+ "epoch": 24.264305177111716,
+ "grad_norm": 5.480803489685059,
+ "learning_rate": 1.772073615336801e-05,
+ "loss": 0.4568,
+ "step": 8905
+ },
+ {
+ "epoch": 24.267029972752045,
+ "grad_norm": 5.276950359344482,
+ "learning_rate": 1.7720175268744914e-05,
+ "loss": 0.4216,
+ "step": 8906
+ },
+ {
+ "epoch": 24.26975476839237,
+ "grad_norm": 4.7104172706604,
+ "learning_rate": 1.771961432399734e-05,
+ "loss": 0.3276,
+ "step": 8907
+ },
+ {
+ "epoch": 24.272479564032697,
+ "grad_norm": 5.371553897857666,
+ "learning_rate": 1.771905331912966e-05,
+ "loss": 0.3318,
+ "step": 8908
+ },
+ {
+ "epoch": 24.275204359673026,
+ "grad_norm": 5.592592239379883,
+ "learning_rate": 1.771849225414624e-05,
+ "loss": 0.3965,
+ "step": 8909
+ },
+ {
+ "epoch": 24.277929155313352,
+ "grad_norm": 5.06408166885376,
+ "learning_rate": 1.7717931129051453e-05,
+ "loss": 0.4497,
+ "step": 8910
+ },
+ {
+ "epoch": 24.280653950953678,
+ "grad_norm": 4.991852760314941,
+ "learning_rate": 1.7717369943849666e-05,
+ "loss": 0.3788,
+ "step": 8911
+ },
+ {
+ "epoch": 24.283378746594007,
+ "grad_norm": 5.195321083068848,
+ "learning_rate": 1.771680869854525e-05,
+ "loss": 0.5079,
+ "step": 8912
+ },
+ {
+ "epoch": 24.286103542234333,
+ "grad_norm": 4.654787540435791,
+ "learning_rate": 1.7716247393142574e-05,
+ "loss": 0.6041,
+ "step": 8913
+ },
+ {
+ "epoch": 24.28882833787466,
+ "grad_norm": 7.494384288787842,
+ "learning_rate": 1.7715686027646012e-05,
+ "loss": 0.4799,
+ "step": 8914
+ },
+ {
+ "epoch": 24.291553133514988,
+ "grad_norm": 4.861104965209961,
+ "learning_rate": 1.7715124602059938e-05,
+ "loss": 0.3271,
+ "step": 8915
+ },
+ {
+ "epoch": 24.294277929155314,
+ "grad_norm": 6.424059867858887,
+ "learning_rate": 1.7714563116388717e-05,
+ "loss": 0.3539,
+ "step": 8916
+ },
+ {
+ "epoch": 24.29700272479564,
+ "grad_norm": 4.84353494644165,
+ "learning_rate": 1.771400157063673e-05,
+ "loss": 0.37,
+ "step": 8917
+ },
+ {
+ "epoch": 24.29972752043597,
+ "grad_norm": 5.129514694213867,
+ "learning_rate": 1.771343996480835e-05,
+ "loss": 0.2927,
+ "step": 8918
+ },
+ {
+ "epoch": 24.302452316076295,
+ "grad_norm": 4.686768531799316,
+ "learning_rate": 1.7712878298907947e-05,
+ "loss": 0.3882,
+ "step": 8919
+ },
+ {
+ "epoch": 24.30517711171662,
+ "grad_norm": 5.026096343994141,
+ "learning_rate": 1.7712316572939893e-05,
+ "loss": 0.4384,
+ "step": 8920
+ },
+ {
+ "epoch": 24.30790190735695,
+ "grad_norm": 6.732193946838379,
+ "learning_rate": 1.7711754786908566e-05,
+ "loss": 0.5249,
+ "step": 8921
+ },
+ {
+ "epoch": 24.310626702997276,
+ "grad_norm": 6.742700576782227,
+ "learning_rate": 1.771119294081834e-05,
+ "loss": 0.3958,
+ "step": 8922
+ },
+ {
+ "epoch": 24.3133514986376,
+ "grad_norm": 6.46943473815918,
+ "learning_rate": 1.7710631034673596e-05,
+ "loss": 0.4866,
+ "step": 8923
+ },
+ {
+ "epoch": 24.31607629427793,
+ "grad_norm": 7.243430137634277,
+ "learning_rate": 1.77100690684787e-05,
+ "loss": 0.4479,
+ "step": 8924
+ },
+ {
+ "epoch": 24.318801089918257,
+ "grad_norm": 6.871385097503662,
+ "learning_rate": 1.770950704223804e-05,
+ "loss": 0.3524,
+ "step": 8925
+ },
+ {
+ "epoch": 24.321525885558582,
+ "grad_norm": 5.186017990112305,
+ "learning_rate": 1.7708944955955983e-05,
+ "loss": 0.3222,
+ "step": 8926
+ },
+ {
+ "epoch": 24.32425068119891,
+ "grad_norm": 4.682200908660889,
+ "learning_rate": 1.7708382809636916e-05,
+ "loss": 0.4547,
+ "step": 8927
+ },
+ {
+ "epoch": 24.326975476839237,
+ "grad_norm": 5.612353801727295,
+ "learning_rate": 1.7707820603285205e-05,
+ "loss": 0.3444,
+ "step": 8928
+ },
+ {
+ "epoch": 24.329700272479563,
+ "grad_norm": 5.764529228210449,
+ "learning_rate": 1.770725833690524e-05,
+ "loss": 0.4335,
+ "step": 8929
+ },
+ {
+ "epoch": 24.332425068119893,
+ "grad_norm": 5.217144966125488,
+ "learning_rate": 1.7706696010501393e-05,
+ "loss": 0.4211,
+ "step": 8930
+ },
+ {
+ "epoch": 24.33514986376022,
+ "grad_norm": 5.640509128570557,
+ "learning_rate": 1.7706133624078046e-05,
+ "loss": 0.4271,
+ "step": 8931
+ },
+ {
+ "epoch": 24.337874659400544,
+ "grad_norm": 5.5102691650390625,
+ "learning_rate": 1.7705571177639576e-05,
+ "loss": 0.3714,
+ "step": 8932
+ },
+ {
+ "epoch": 24.340599455040874,
+ "grad_norm": 5.13256311416626,
+ "learning_rate": 1.770500867119037e-05,
+ "loss": 0.5094,
+ "step": 8933
+ },
+ {
+ "epoch": 24.3433242506812,
+ "grad_norm": 6.2727484703063965,
+ "learning_rate": 1.7704446104734803e-05,
+ "loss": 0.393,
+ "step": 8934
+ },
+ {
+ "epoch": 24.346049046321525,
+ "grad_norm": 5.514694690704346,
+ "learning_rate": 1.7703883478277254e-05,
+ "loss": 0.365,
+ "step": 8935
+ },
+ {
+ "epoch": 24.348773841961854,
+ "grad_norm": 6.159389019012451,
+ "learning_rate": 1.7703320791822115e-05,
+ "loss": 0.4022,
+ "step": 8936
+ },
+ {
+ "epoch": 24.35149863760218,
+ "grad_norm": 5.744335174560547,
+ "learning_rate": 1.7702758045373757e-05,
+ "loss": 0.3369,
+ "step": 8937
+ },
+ {
+ "epoch": 24.354223433242506,
+ "grad_norm": 4.810899257659912,
+ "learning_rate": 1.7702195238936567e-05,
+ "loss": 0.3327,
+ "step": 8938
+ },
+ {
+ "epoch": 24.356948228882835,
+ "grad_norm": 5.09812068939209,
+ "learning_rate": 1.7701632372514934e-05,
+ "loss": 0.3568,
+ "step": 8939
+ },
+ {
+ "epoch": 24.35967302452316,
+ "grad_norm": 6.613391876220703,
+ "learning_rate": 1.770106944611323e-05,
+ "loss": 0.37,
+ "step": 8940
+ },
+ {
+ "epoch": 24.362397820163487,
+ "grad_norm": 5.673793792724609,
+ "learning_rate": 1.7700506459735844e-05,
+ "loss": 0.4018,
+ "step": 8941
+ },
+ {
+ "epoch": 24.365122615803816,
+ "grad_norm": 6.146055698394775,
+ "learning_rate": 1.7699943413387166e-05,
+ "loss": 0.4182,
+ "step": 8942
+ },
+ {
+ "epoch": 24.367847411444142,
+ "grad_norm": 6.327613353729248,
+ "learning_rate": 1.7699380307071575e-05,
+ "loss": 0.4341,
+ "step": 8943
+ },
+ {
+ "epoch": 24.370572207084468,
+ "grad_norm": 5.499996662139893,
+ "learning_rate": 1.7698817140793458e-05,
+ "loss": 0.3416,
+ "step": 8944
+ },
+ {
+ "epoch": 24.373297002724797,
+ "grad_norm": 4.968870162963867,
+ "learning_rate": 1.76982539145572e-05,
+ "loss": 0.6296,
+ "step": 8945
+ },
+ {
+ "epoch": 24.376021798365123,
+ "grad_norm": 10.559367179870605,
+ "learning_rate": 1.7697690628367188e-05,
+ "loss": 0.3947,
+ "step": 8946
+ },
+ {
+ "epoch": 24.37874659400545,
+ "grad_norm": 5.834281921386719,
+ "learning_rate": 1.769712728222781e-05,
+ "loss": 0.476,
+ "step": 8947
+ },
+ {
+ "epoch": 24.381471389645778,
+ "grad_norm": 5.55001974105835,
+ "learning_rate": 1.7696563876143455e-05,
+ "loss": 0.4725,
+ "step": 8948
+ },
+ {
+ "epoch": 24.384196185286104,
+ "grad_norm": 5.402935028076172,
+ "learning_rate": 1.7696000410118504e-05,
+ "loss": 0.3608,
+ "step": 8949
+ },
+ {
+ "epoch": 24.38692098092643,
+ "grad_norm": 5.089282035827637,
+ "learning_rate": 1.7695436884157353e-05,
+ "loss": 0.3655,
+ "step": 8950
+ },
+ {
+ "epoch": 24.38964577656676,
+ "grad_norm": 5.299681186676025,
+ "learning_rate": 1.7694873298264385e-05,
+ "loss": 0.4435,
+ "step": 8951
+ },
+ {
+ "epoch": 24.392370572207085,
+ "grad_norm": 5.367883205413818,
+ "learning_rate": 1.769430965244399e-05,
+ "loss": 0.4352,
+ "step": 8952
+ },
+ {
+ "epoch": 24.39509536784741,
+ "grad_norm": 5.406439304351807,
+ "learning_rate": 1.7693745946700566e-05,
+ "loss": 0.3882,
+ "step": 8953
+ },
+ {
+ "epoch": 24.39782016348774,
+ "grad_norm": 5.950461387634277,
+ "learning_rate": 1.769318218103849e-05,
+ "loss": 0.3944,
+ "step": 8954
+ },
+ {
+ "epoch": 24.400544959128066,
+ "grad_norm": 5.762792110443115,
+ "learning_rate": 1.7692618355462157e-05,
+ "loss": 0.3252,
+ "step": 8955
+ },
+ {
+ "epoch": 24.40326975476839,
+ "grad_norm": 5.195339202880859,
+ "learning_rate": 1.7692054469975964e-05,
+ "loss": 0.3009,
+ "step": 8956
+ },
+ {
+ "epoch": 24.40599455040872,
+ "grad_norm": 6.964745044708252,
+ "learning_rate": 1.7691490524584297e-05,
+ "loss": 0.5466,
+ "step": 8957
+ },
+ {
+ "epoch": 24.408719346049047,
+ "grad_norm": 7.1578192710876465,
+ "learning_rate": 1.7690926519291548e-05,
+ "loss": 0.5915,
+ "step": 8958
+ },
+ {
+ "epoch": 24.411444141689373,
+ "grad_norm": 5.116558074951172,
+ "learning_rate": 1.7690362454102114e-05,
+ "loss": 0.381,
+ "step": 8959
+ },
+ {
+ "epoch": 24.414168937329702,
+ "grad_norm": 5.784854412078857,
+ "learning_rate": 1.7689798329020386e-05,
+ "loss": 0.4079,
+ "step": 8960
+ },
+ {
+ "epoch": 24.416893732970028,
+ "grad_norm": 6.028428554534912,
+ "learning_rate": 1.768923414405075e-05,
+ "loss": 0.3804,
+ "step": 8961
+ },
+ {
+ "epoch": 24.419618528610354,
+ "grad_norm": 5.990976333618164,
+ "learning_rate": 1.768866989919761e-05,
+ "loss": 0.3787,
+ "step": 8962
+ },
+ {
+ "epoch": 24.422343324250683,
+ "grad_norm": 4.815041542053223,
+ "learning_rate": 1.7688105594465356e-05,
+ "loss": 0.4916,
+ "step": 8963
+ },
+ {
+ "epoch": 24.42506811989101,
+ "grad_norm": 4.815680503845215,
+ "learning_rate": 1.768754122985838e-05,
+ "loss": 0.3873,
+ "step": 8964
+ },
+ {
+ "epoch": 24.427792915531334,
+ "grad_norm": 6.647727966308594,
+ "learning_rate": 1.7686976805381084e-05,
+ "loss": 0.4224,
+ "step": 8965
+ },
+ {
+ "epoch": 24.430517711171664,
+ "grad_norm": 5.048221588134766,
+ "learning_rate": 1.7686412321037857e-05,
+ "loss": 0.3478,
+ "step": 8966
+ },
+ {
+ "epoch": 24.43324250681199,
+ "grad_norm": 4.7357177734375,
+ "learning_rate": 1.76858477768331e-05,
+ "loss": 0.4428,
+ "step": 8967
+ },
+ {
+ "epoch": 24.435967302452315,
+ "grad_norm": 31.133148193359375,
+ "learning_rate": 1.768528317277121e-05,
+ "loss": 0.2885,
+ "step": 8968
+ },
+ {
+ "epoch": 24.438692098092645,
+ "grad_norm": 5.4990363121032715,
+ "learning_rate": 1.768471850885658e-05,
+ "loss": 0.3177,
+ "step": 8969
+ },
+ {
+ "epoch": 24.44141689373297,
+ "grad_norm": 4.886033058166504,
+ "learning_rate": 1.768415378509361e-05,
+ "loss": 0.5025,
+ "step": 8970
+ },
+ {
+ "epoch": 24.444141689373296,
+ "grad_norm": 4.8222503662109375,
+ "learning_rate": 1.7683589001486696e-05,
+ "loss": 0.5475,
+ "step": 8971
+ },
+ {
+ "epoch": 24.446866485013626,
+ "grad_norm": 4.9131059646606445,
+ "learning_rate": 1.768302415804024e-05,
+ "loss": 0.3961,
+ "step": 8972
+ },
+ {
+ "epoch": 24.44959128065395,
+ "grad_norm": 6.185993671417236,
+ "learning_rate": 1.768245925475864e-05,
+ "loss": 0.4988,
+ "step": 8973
+ },
+ {
+ "epoch": 24.452316076294277,
+ "grad_norm": 5.193041801452637,
+ "learning_rate": 1.7681894291646293e-05,
+ "loss": 0.269,
+ "step": 8974
+ },
+ {
+ "epoch": 24.455040871934607,
+ "grad_norm": 5.556548595428467,
+ "learning_rate": 1.7681329268707598e-05,
+ "loss": 0.3699,
+ "step": 8975
+ },
+ {
+ "epoch": 24.457765667574932,
+ "grad_norm": 4.790220737457275,
+ "learning_rate": 1.768076418594696e-05,
+ "loss": 0.4822,
+ "step": 8976
+ },
+ {
+ "epoch": 24.460490463215258,
+ "grad_norm": 5.592531204223633,
+ "learning_rate": 1.768019904336878e-05,
+ "loss": 0.4026,
+ "step": 8977
+ },
+ {
+ "epoch": 24.463215258855588,
+ "grad_norm": 4.974039554595947,
+ "learning_rate": 1.7679633840977454e-05,
+ "loss": 0.3463,
+ "step": 8978
+ },
+ {
+ "epoch": 24.465940054495913,
+ "grad_norm": 8.408161163330078,
+ "learning_rate": 1.7679068578777387e-05,
+ "loss": 0.4794,
+ "step": 8979
+ },
+ {
+ "epoch": 24.46866485013624,
+ "grad_norm": 5.098791599273682,
+ "learning_rate": 1.7678503256772985e-05,
+ "loss": 0.438,
+ "step": 8980
+ },
+ {
+ "epoch": 24.47138964577657,
+ "grad_norm": 6.452724933624268,
+ "learning_rate": 1.7677937874968646e-05,
+ "loss": 0.3255,
+ "step": 8981
+ },
+ {
+ "epoch": 24.474114441416894,
+ "grad_norm": 6.2594780921936035,
+ "learning_rate": 1.767737243336877e-05,
+ "loss": 0.527,
+ "step": 8982
+ },
+ {
+ "epoch": 24.47683923705722,
+ "grad_norm": 13.065546035766602,
+ "learning_rate": 1.7676806931977765e-05,
+ "loss": 0.4032,
+ "step": 8983
+ },
+ {
+ "epoch": 24.479564032697546,
+ "grad_norm": 6.82124662399292,
+ "learning_rate": 1.7676241370800037e-05,
+ "loss": 0.4402,
+ "step": 8984
+ },
+ {
+ "epoch": 24.482288828337875,
+ "grad_norm": 5.752935409545898,
+ "learning_rate": 1.7675675749839988e-05,
+ "loss": 0.3996,
+ "step": 8985
+ },
+ {
+ "epoch": 24.4850136239782,
+ "grad_norm": 4.6033148765563965,
+ "learning_rate": 1.7675110069102022e-05,
+ "loss": 0.4032,
+ "step": 8986
+ },
+ {
+ "epoch": 24.48773841961853,
+ "grad_norm": 5.816371917724609,
+ "learning_rate": 1.7674544328590547e-05,
+ "loss": 0.3384,
+ "step": 8987
+ },
+ {
+ "epoch": 24.490463215258856,
+ "grad_norm": 5.218268394470215,
+ "learning_rate": 1.7673978528309964e-05,
+ "loss": 0.4401,
+ "step": 8988
+ },
+ {
+ "epoch": 24.493188010899182,
+ "grad_norm": 5.222329139709473,
+ "learning_rate": 1.7673412668264687e-05,
+ "loss": 0.4529,
+ "step": 8989
+ },
+ {
+ "epoch": 24.495912806539508,
+ "grad_norm": 5.612181186676025,
+ "learning_rate": 1.767284674845912e-05,
+ "loss": 0.4014,
+ "step": 8990
+ },
+ {
+ "epoch": 24.498637602179837,
+ "grad_norm": 4.7710490226745605,
+ "learning_rate": 1.767228076889767e-05,
+ "loss": 0.3485,
+ "step": 8991
+ },
+ {
+ "epoch": 24.501362397820163,
+ "grad_norm": 5.991434574127197,
+ "learning_rate": 1.767171472958474e-05,
+ "loss": 0.4785,
+ "step": 8992
+ },
+ {
+ "epoch": 24.504087193460492,
+ "grad_norm": 5.506546497344971,
+ "learning_rate": 1.7671148630524745e-05,
+ "loss": 0.3617,
+ "step": 8993
+ },
+ {
+ "epoch": 24.506811989100818,
+ "grad_norm": 21.61460304260254,
+ "learning_rate": 1.767058247172209e-05,
+ "loss": 0.4691,
+ "step": 8994
+ },
+ {
+ "epoch": 24.509536784741144,
+ "grad_norm": 6.0051069259643555,
+ "learning_rate": 1.767001625318119e-05,
+ "loss": 0.4208,
+ "step": 8995
+ },
+ {
+ "epoch": 24.51226158038147,
+ "grad_norm": 7.637426376342773,
+ "learning_rate": 1.7669449974906446e-05,
+ "loss": 0.4875,
+ "step": 8996
+ },
+ {
+ "epoch": 24.5149863760218,
+ "grad_norm": 4.497837543487549,
+ "learning_rate": 1.766888363690227e-05,
+ "loss": 0.5005,
+ "step": 8997
+ },
+ {
+ "epoch": 24.517711171662125,
+ "grad_norm": 4.956736087799072,
+ "learning_rate": 1.766831723917308e-05,
+ "loss": 0.4423,
+ "step": 8998
+ },
+ {
+ "epoch": 24.520435967302454,
+ "grad_norm": 5.219514846801758,
+ "learning_rate": 1.7667750781723282e-05,
+ "loss": 0.386,
+ "step": 8999
+ },
+ {
+ "epoch": 24.52316076294278,
+ "grad_norm": 4.769038200378418,
+ "learning_rate": 1.7667184264557286e-05,
+ "loss": 0.5372,
+ "step": 9000
+ },
+ {
+ "epoch": 24.525885558583106,
+ "grad_norm": 5.445932865142822,
+ "learning_rate": 1.7666617687679505e-05,
+ "loss": 0.3532,
+ "step": 9001
+ },
+ {
+ "epoch": 24.52861035422343,
+ "grad_norm": 5.409191131591797,
+ "learning_rate": 1.766605105109435e-05,
+ "loss": 0.4931,
+ "step": 9002
+ },
+ {
+ "epoch": 24.53133514986376,
+ "grad_norm": 5.564929962158203,
+ "learning_rate": 1.7665484354806244e-05,
+ "loss": 0.5821,
+ "step": 9003
+ },
+ {
+ "epoch": 24.534059945504087,
+ "grad_norm": 5.380222320556641,
+ "learning_rate": 1.766491759881959e-05,
+ "loss": 0.3791,
+ "step": 9004
+ },
+ {
+ "epoch": 24.536784741144416,
+ "grad_norm": 6.1016974449157715,
+ "learning_rate": 1.76643507831388e-05,
+ "loss": 0.3829,
+ "step": 9005
+ },
+ {
+ "epoch": 24.539509536784742,
+ "grad_norm": 5.443460464477539,
+ "learning_rate": 1.7663783907768298e-05,
+ "loss": 0.5637,
+ "step": 9006
+ },
+ {
+ "epoch": 24.542234332425068,
+ "grad_norm": 5.161444664001465,
+ "learning_rate": 1.766321697271249e-05,
+ "loss": 0.3953,
+ "step": 9007
+ },
+ {
+ "epoch": 24.544959128065393,
+ "grad_norm": 6.895048141479492,
+ "learning_rate": 1.7662649977975794e-05,
+ "loss": 0.3322,
+ "step": 9008
+ },
+ {
+ "epoch": 24.547683923705723,
+ "grad_norm": 5.2697224617004395,
+ "learning_rate": 1.766208292356263e-05,
+ "loss": 0.4196,
+ "step": 9009
+ },
+ {
+ "epoch": 24.55040871934605,
+ "grad_norm": 5.544307231903076,
+ "learning_rate": 1.766151580947741e-05,
+ "loss": 0.5376,
+ "step": 9010
+ },
+ {
+ "epoch": 24.553133514986374,
+ "grad_norm": 5.619485378265381,
+ "learning_rate": 1.766094863572455e-05,
+ "loss": 0.4865,
+ "step": 9011
+ },
+ {
+ "epoch": 24.555858310626704,
+ "grad_norm": 5.183757305145264,
+ "learning_rate": 1.7660381402308465e-05,
+ "loss": 0.3813,
+ "step": 9012
+ },
+ {
+ "epoch": 24.55858310626703,
+ "grad_norm": 6.168740272521973,
+ "learning_rate": 1.765981410923358e-05,
+ "loss": 0.384,
+ "step": 9013
+ },
+ {
+ "epoch": 24.561307901907355,
+ "grad_norm": 5.621607303619385,
+ "learning_rate": 1.7659246756504313e-05,
+ "loss": 0.3374,
+ "step": 9014
+ },
+ {
+ "epoch": 24.564032697547685,
+ "grad_norm": 6.364660739898682,
+ "learning_rate": 1.7658679344125075e-05,
+ "loss": 0.2946,
+ "step": 9015
+ },
+ {
+ "epoch": 24.56675749318801,
+ "grad_norm": 5.154107570648193,
+ "learning_rate": 1.7658111872100287e-05,
+ "loss": 0.2488,
+ "step": 9016
+ },
+ {
+ "epoch": 24.569482288828336,
+ "grad_norm": 5.104134559631348,
+ "learning_rate": 1.7657544340434375e-05,
+ "loss": 0.4076,
+ "step": 9017
+ },
+ {
+ "epoch": 24.572207084468666,
+ "grad_norm": 5.270888805389404,
+ "learning_rate": 1.7656976749131753e-05,
+ "loss": 0.3459,
+ "step": 9018
+ },
+ {
+ "epoch": 24.57493188010899,
+ "grad_norm": 5.874686241149902,
+ "learning_rate": 1.765640909819684e-05,
+ "loss": 0.3478,
+ "step": 9019
+ },
+ {
+ "epoch": 24.577656675749317,
+ "grad_norm": 5.84893798828125,
+ "learning_rate": 1.7655841387634057e-05,
+ "loss": 0.7636,
+ "step": 9020
+ },
+ {
+ "epoch": 24.580381471389646,
+ "grad_norm": 5.9361748695373535,
+ "learning_rate": 1.765527361744783e-05,
+ "loss": 0.5214,
+ "step": 9021
+ },
+ {
+ "epoch": 24.583106267029972,
+ "grad_norm": 5.555140495300293,
+ "learning_rate": 1.765470578764258e-05,
+ "loss": 0.4042,
+ "step": 9022
+ },
+ {
+ "epoch": 24.585831062670298,
+ "grad_norm": 6.253485202789307,
+ "learning_rate": 1.7654137898222728e-05,
+ "loss": 0.5389,
+ "step": 9023
+ },
+ {
+ "epoch": 24.588555858310627,
+ "grad_norm": 5.046559810638428,
+ "learning_rate": 1.7653569949192698e-05,
+ "loss": 0.2973,
+ "step": 9024
+ },
+ {
+ "epoch": 24.591280653950953,
+ "grad_norm": 5.4978790283203125,
+ "learning_rate": 1.765300194055691e-05,
+ "loss": 0.4683,
+ "step": 9025
+ },
+ {
+ "epoch": 24.59400544959128,
+ "grad_norm": 5.5893120765686035,
+ "learning_rate": 1.7652433872319785e-05,
+ "loss": 0.4358,
+ "step": 9026
+ },
+ {
+ "epoch": 24.59673024523161,
+ "grad_norm": 5.007052421569824,
+ "learning_rate": 1.7651865744485755e-05,
+ "loss": 0.3845,
+ "step": 9027
+ },
+ {
+ "epoch": 24.599455040871934,
+ "grad_norm": 5.514000415802002,
+ "learning_rate": 1.765129755705924e-05,
+ "loss": 0.4879,
+ "step": 9028
+ },
+ {
+ "epoch": 24.60217983651226,
+ "grad_norm": 8.804703712463379,
+ "learning_rate": 1.7650729310044667e-05,
+ "loss": 0.472,
+ "step": 9029
+ },
+ {
+ "epoch": 24.60490463215259,
+ "grad_norm": 5.5313801765441895,
+ "learning_rate": 1.765016100344646e-05,
+ "loss": 0.4963,
+ "step": 9030
+ },
+ {
+ "epoch": 24.607629427792915,
+ "grad_norm": 6.36633825302124,
+ "learning_rate": 1.7649592637269045e-05,
+ "loss": 0.4374,
+ "step": 9031
+ },
+ {
+ "epoch": 24.61035422343324,
+ "grad_norm": 5.716895580291748,
+ "learning_rate": 1.7649024211516848e-05,
+ "loss": 0.3771,
+ "step": 9032
+ },
+ {
+ "epoch": 24.61307901907357,
+ "grad_norm": 5.609775543212891,
+ "learning_rate": 1.76484557261943e-05,
+ "loss": 0.3676,
+ "step": 9033
+ },
+ {
+ "epoch": 24.615803814713896,
+ "grad_norm": 5.65683126449585,
+ "learning_rate": 1.7647887181305822e-05,
+ "loss": 0.456,
+ "step": 9034
+ },
+ {
+ "epoch": 24.618528610354222,
+ "grad_norm": 5.488038063049316,
+ "learning_rate": 1.7647318576855843e-05,
+ "loss": 0.4781,
+ "step": 9035
+ },
+ {
+ "epoch": 24.62125340599455,
+ "grad_norm": 5.370746612548828,
+ "learning_rate": 1.7646749912848795e-05,
+ "loss": 0.3965,
+ "step": 9036
+ },
+ {
+ "epoch": 24.623978201634877,
+ "grad_norm": 5.420413017272949,
+ "learning_rate": 1.7646181189289108e-05,
+ "loss": 0.3974,
+ "step": 9037
+ },
+ {
+ "epoch": 24.626702997275203,
+ "grad_norm": 6.932467937469482,
+ "learning_rate": 1.7645612406181204e-05,
+ "loss": 0.4082,
+ "step": 9038
+ },
+ {
+ "epoch": 24.629427792915532,
+ "grad_norm": 4.7583770751953125,
+ "learning_rate": 1.7645043563529518e-05,
+ "loss": 0.3118,
+ "step": 9039
+ },
+ {
+ "epoch": 24.632152588555858,
+ "grad_norm": 6.439189910888672,
+ "learning_rate": 1.764447466133848e-05,
+ "loss": 0.5584,
+ "step": 9040
+ },
+ {
+ "epoch": 24.634877384196184,
+ "grad_norm": 5.009640216827393,
+ "learning_rate": 1.7643905699612513e-05,
+ "loss": 0.3463,
+ "step": 9041
+ },
+ {
+ "epoch": 24.637602179836513,
+ "grad_norm": 4.691561698913574,
+ "learning_rate": 1.7643336678356063e-05,
+ "loss": 0.4746,
+ "step": 9042
+ },
+ {
+ "epoch": 24.64032697547684,
+ "grad_norm": 5.361378192901611,
+ "learning_rate": 1.7642767597573548e-05,
+ "loss": 0.3744,
+ "step": 9043
+ },
+ {
+ "epoch": 24.643051771117165,
+ "grad_norm": 5.248488426208496,
+ "learning_rate": 1.7642198457269405e-05,
+ "loss": 0.2822,
+ "step": 9044
+ },
+ {
+ "epoch": 24.645776566757494,
+ "grad_norm": 6.459550380706787,
+ "learning_rate": 1.7641629257448068e-05,
+ "loss": 0.3806,
+ "step": 9045
+ },
+ {
+ "epoch": 24.64850136239782,
+ "grad_norm": 6.972738742828369,
+ "learning_rate": 1.7641059998113966e-05,
+ "loss": 0.424,
+ "step": 9046
+ },
+ {
+ "epoch": 24.651226158038146,
+ "grad_norm": 5.950128078460693,
+ "learning_rate": 1.7640490679271537e-05,
+ "loss": 0.5011,
+ "step": 9047
+ },
+ {
+ "epoch": 24.653950953678475,
+ "grad_norm": 6.4490647315979,
+ "learning_rate": 1.7639921300925213e-05,
+ "loss": 0.4512,
+ "step": 9048
+ },
+ {
+ "epoch": 24.6566757493188,
+ "grad_norm": 7.613819122314453,
+ "learning_rate": 1.7639351863079426e-05,
+ "loss": 0.4508,
+ "step": 9049
+ },
+ {
+ "epoch": 24.659400544959126,
+ "grad_norm": 6.61224889755249,
+ "learning_rate": 1.7638782365738613e-05,
+ "loss": 0.4662,
+ "step": 9050
+ },
+ {
+ "epoch": 24.662125340599456,
+ "grad_norm": 6.3704423904418945,
+ "learning_rate": 1.763821280890721e-05,
+ "loss": 0.6129,
+ "step": 9051
+ },
+ {
+ "epoch": 24.66485013623978,
+ "grad_norm": 6.0949835777282715,
+ "learning_rate": 1.763764319258965e-05,
+ "loss": 0.3687,
+ "step": 9052
+ },
+ {
+ "epoch": 24.667574931880107,
+ "grad_norm": 5.711676597595215,
+ "learning_rate": 1.7637073516790373e-05,
+ "loss": 0.3834,
+ "step": 9053
+ },
+ {
+ "epoch": 24.670299727520437,
+ "grad_norm": 5.027652263641357,
+ "learning_rate": 1.7636503781513808e-05,
+ "loss": 0.4698,
+ "step": 9054
+ },
+ {
+ "epoch": 24.673024523160763,
+ "grad_norm": 5.252974033355713,
+ "learning_rate": 1.7635933986764403e-05,
+ "loss": 0.3558,
+ "step": 9055
+ },
+ {
+ "epoch": 24.67574931880109,
+ "grad_norm": 6.636815547943115,
+ "learning_rate": 1.7635364132546587e-05,
+ "loss": 0.2989,
+ "step": 9056
+ },
+ {
+ "epoch": 24.678474114441418,
+ "grad_norm": 4.620126247406006,
+ "learning_rate": 1.7634794218864802e-05,
+ "loss": 0.5907,
+ "step": 9057
+ },
+ {
+ "epoch": 24.681198910081743,
+ "grad_norm": 7.7050580978393555,
+ "learning_rate": 1.7634224245723482e-05,
+ "loss": 0.4868,
+ "step": 9058
+ },
+ {
+ "epoch": 24.68392370572207,
+ "grad_norm": 5.498633861541748,
+ "learning_rate": 1.7633654213127074e-05,
+ "loss": 0.378,
+ "step": 9059
+ },
+ {
+ "epoch": 24.6866485013624,
+ "grad_norm": 9.557458877563477,
+ "learning_rate": 1.763308412108001e-05,
+ "loss": 0.5641,
+ "step": 9060
+ },
+ {
+ "epoch": 24.689373297002724,
+ "grad_norm": 5.794239044189453,
+ "learning_rate": 1.7632513969586736e-05,
+ "loss": 0.3808,
+ "step": 9061
+ },
+ {
+ "epoch": 24.69209809264305,
+ "grad_norm": 6.800891876220703,
+ "learning_rate": 1.7631943758651687e-05,
+ "loss": 0.421,
+ "step": 9062
+ },
+ {
+ "epoch": 24.69482288828338,
+ "grad_norm": 5.532435894012451,
+ "learning_rate": 1.7631373488279303e-05,
+ "loss": 0.5308,
+ "step": 9063
+ },
+ {
+ "epoch": 24.697547683923705,
+ "grad_norm": 5.051397323608398,
+ "learning_rate": 1.763080315847403e-05,
+ "loss": 0.4367,
+ "step": 9064
+ },
+ {
+ "epoch": 24.70027247956403,
+ "grad_norm": 5.802183151245117,
+ "learning_rate": 1.7630232769240307e-05,
+ "loss": 0.3415,
+ "step": 9065
+ },
+ {
+ "epoch": 24.70299727520436,
+ "grad_norm": 6.128408432006836,
+ "learning_rate": 1.7629662320582578e-05,
+ "loss": 0.5529,
+ "step": 9066
+ },
+ {
+ "epoch": 24.705722070844686,
+ "grad_norm": 5.6915435791015625,
+ "learning_rate": 1.7629091812505285e-05,
+ "loss": 0.5169,
+ "step": 9067
+ },
+ {
+ "epoch": 24.708446866485012,
+ "grad_norm": 10.840140342712402,
+ "learning_rate": 1.762852124501287e-05,
+ "loss": 0.4152,
+ "step": 9068
+ },
+ {
+ "epoch": 24.71117166212534,
+ "grad_norm": 6.416767597198486,
+ "learning_rate": 1.7627950618109775e-05,
+ "loss": 0.4197,
+ "step": 9069
+ },
+ {
+ "epoch": 24.713896457765667,
+ "grad_norm": 6.069099426269531,
+ "learning_rate": 1.7627379931800446e-05,
+ "loss": 0.3725,
+ "step": 9070
+ },
+ {
+ "epoch": 24.716621253405993,
+ "grad_norm": 5.843745708465576,
+ "learning_rate": 1.762680918608933e-05,
+ "loss": 0.484,
+ "step": 9071
+ },
+ {
+ "epoch": 24.719346049046322,
+ "grad_norm": 6.50100564956665,
+ "learning_rate": 1.7626238380980866e-05,
+ "loss": 0.489,
+ "step": 9072
+ },
+ {
+ "epoch": 24.722070844686648,
+ "grad_norm": 4.916182518005371,
+ "learning_rate": 1.7625667516479508e-05,
+ "loss": 0.464,
+ "step": 9073
+ },
+ {
+ "epoch": 24.724795640326974,
+ "grad_norm": 5.325462818145752,
+ "learning_rate": 1.7625096592589692e-05,
+ "loss": 0.4665,
+ "step": 9074
+ },
+ {
+ "epoch": 24.727520435967303,
+ "grad_norm": 6.646984577178955,
+ "learning_rate": 1.7624525609315873e-05,
+ "loss": 0.4526,
+ "step": 9075
+ },
+ {
+ "epoch": 24.73024523160763,
+ "grad_norm": 5.167802810668945,
+ "learning_rate": 1.762395456666249e-05,
+ "loss": 0.4169,
+ "step": 9076
+ },
+ {
+ "epoch": 24.732970027247955,
+ "grad_norm": 6.207061767578125,
+ "learning_rate": 1.7623383464633996e-05,
+ "loss": 0.483,
+ "step": 9077
+ },
+ {
+ "epoch": 24.735694822888284,
+ "grad_norm": 6.9759016036987305,
+ "learning_rate": 1.7622812303234837e-05,
+ "loss": 0.5053,
+ "step": 9078
+ },
+ {
+ "epoch": 24.73841961852861,
+ "grad_norm": 5.6943254470825195,
+ "learning_rate": 1.7622241082469463e-05,
+ "loss": 0.5726,
+ "step": 9079
+ },
+ {
+ "epoch": 24.741144414168936,
+ "grad_norm": 6.048657417297363,
+ "learning_rate": 1.762166980234232e-05,
+ "loss": 0.475,
+ "step": 9080
+ },
+ {
+ "epoch": 24.743869209809265,
+ "grad_norm": 5.4786810874938965,
+ "learning_rate": 1.762109846285786e-05,
+ "loss": 0.6344,
+ "step": 9081
+ },
+ {
+ "epoch": 24.74659400544959,
+ "grad_norm": 7.059934616088867,
+ "learning_rate": 1.7620527064020524e-05,
+ "loss": 0.4774,
+ "step": 9082
+ },
+ {
+ "epoch": 24.749318801089917,
+ "grad_norm": 5.626350402832031,
+ "learning_rate": 1.761995560583477e-05,
+ "loss": 0.4844,
+ "step": 9083
+ },
+ {
+ "epoch": 24.752043596730246,
+ "grad_norm": 6.07266902923584,
+ "learning_rate": 1.761938408830505e-05,
+ "loss": 0.4931,
+ "step": 9084
+ },
+ {
+ "epoch": 24.754768392370572,
+ "grad_norm": 5.9608988761901855,
+ "learning_rate": 1.761881251143581e-05,
+ "loss": 0.5255,
+ "step": 9085
+ },
+ {
+ "epoch": 24.757493188010898,
+ "grad_norm": 4.983918190002441,
+ "learning_rate": 1.7618240875231504e-05,
+ "loss": 0.386,
+ "step": 9086
+ },
+ {
+ "epoch": 24.760217983651227,
+ "grad_norm": 5.4183855056762695,
+ "learning_rate": 1.7617669179696584e-05,
+ "loss": 0.4023,
+ "step": 9087
+ },
+ {
+ "epoch": 24.762942779291553,
+ "grad_norm": 6.751500129699707,
+ "learning_rate": 1.76170974248355e-05,
+ "loss": 0.4073,
+ "step": 9088
+ },
+ {
+ "epoch": 24.76566757493188,
+ "grad_norm": 5.476382255554199,
+ "learning_rate": 1.761652561065271e-05,
+ "loss": 0.3553,
+ "step": 9089
+ },
+ {
+ "epoch": 24.768392370572208,
+ "grad_norm": 4.853948593139648,
+ "learning_rate": 1.7615953737152662e-05,
+ "loss": 0.4897,
+ "step": 9090
+ },
+ {
+ "epoch": 24.771117166212534,
+ "grad_norm": 6.10638427734375,
+ "learning_rate": 1.7615381804339812e-05,
+ "loss": 0.3481,
+ "step": 9091
+ },
+ {
+ "epoch": 24.77384196185286,
+ "grad_norm": 7.427316188812256,
+ "learning_rate": 1.761480981221861e-05,
+ "loss": 0.5103,
+ "step": 9092
+ },
+ {
+ "epoch": 24.77656675749319,
+ "grad_norm": 5.1420674324035645,
+ "learning_rate": 1.7614237760793517e-05,
+ "loss": 0.387,
+ "step": 9093
+ },
+ {
+ "epoch": 24.779291553133515,
+ "grad_norm": 5.7164740562438965,
+ "learning_rate": 1.7613665650068986e-05,
+ "loss": 0.5341,
+ "step": 9094
+ },
+ {
+ "epoch": 24.78201634877384,
+ "grad_norm": 5.298192024230957,
+ "learning_rate": 1.7613093480049472e-05,
+ "loss": 0.3653,
+ "step": 9095
+ },
+ {
+ "epoch": 24.78474114441417,
+ "grad_norm": 5.666665077209473,
+ "learning_rate": 1.7612521250739427e-05,
+ "loss": 0.2879,
+ "step": 9096
+ },
+ {
+ "epoch": 24.787465940054496,
+ "grad_norm": 7.55239200592041,
+ "learning_rate": 1.7611948962143317e-05,
+ "loss": 0.5129,
+ "step": 9097
+ },
+ {
+ "epoch": 24.79019073569482,
+ "grad_norm": 4.915546417236328,
+ "learning_rate": 1.7611376614265586e-05,
+ "loss": 0.4421,
+ "step": 9098
+ },
+ {
+ "epoch": 24.79291553133515,
+ "grad_norm": 5.833803176879883,
+ "learning_rate": 1.7610804207110708e-05,
+ "loss": 0.425,
+ "step": 9099
+ },
+ {
+ "epoch": 24.795640326975477,
+ "grad_norm": 6.972508907318115,
+ "learning_rate": 1.7610231740683127e-05,
+ "loss": 0.6023,
+ "step": 9100
+ },
+ {
+ "epoch": 24.798365122615802,
+ "grad_norm": 5.834273338317871,
+ "learning_rate": 1.7609659214987306e-05,
+ "loss": 0.3774,
+ "step": 9101
+ },
+ {
+ "epoch": 24.80108991825613,
+ "grad_norm": 5.774014949798584,
+ "learning_rate": 1.7609086630027703e-05,
+ "loss": 0.4208,
+ "step": 9102
+ },
+ {
+ "epoch": 24.803814713896458,
+ "grad_norm": 7.034475326538086,
+ "learning_rate": 1.760851398580878e-05,
+ "loss": 0.5127,
+ "step": 9103
+ },
+ {
+ "epoch": 24.806539509536783,
+ "grad_norm": 7.968355655670166,
+ "learning_rate": 1.7607941282334994e-05,
+ "loss": 0.4521,
+ "step": 9104
+ },
+ {
+ "epoch": 24.809264305177113,
+ "grad_norm": 6.236171245574951,
+ "learning_rate": 1.760736851961081e-05,
+ "loss": 0.515,
+ "step": 9105
+ },
+ {
+ "epoch": 24.81198910081744,
+ "grad_norm": 7.1209540367126465,
+ "learning_rate": 1.760679569764068e-05,
+ "loss": 0.5348,
+ "step": 9106
+ },
+ {
+ "epoch": 24.814713896457764,
+ "grad_norm": 6.401973247528076,
+ "learning_rate": 1.7606222816429072e-05,
+ "loss": 0.3845,
+ "step": 9107
+ },
+ {
+ "epoch": 24.817438692098094,
+ "grad_norm": 5.021841049194336,
+ "learning_rate": 1.7605649875980444e-05,
+ "loss": 0.3688,
+ "step": 9108
+ },
+ {
+ "epoch": 24.82016348773842,
+ "grad_norm": 5.27953577041626,
+ "learning_rate": 1.760507687629926e-05,
+ "loss": 0.3665,
+ "step": 9109
+ },
+ {
+ "epoch": 24.822888283378745,
+ "grad_norm": 4.782501697540283,
+ "learning_rate": 1.7604503817389984e-05,
+ "loss": 0.4363,
+ "step": 9110
+ },
+ {
+ "epoch": 24.825613079019075,
+ "grad_norm": 5.284876346588135,
+ "learning_rate": 1.7603930699257077e-05,
+ "loss": 0.428,
+ "step": 9111
+ },
+ {
+ "epoch": 24.8283378746594,
+ "grad_norm": 6.523740291595459,
+ "learning_rate": 1.7603357521905002e-05,
+ "loss": 0.5459,
+ "step": 9112
+ },
+ {
+ "epoch": 24.831062670299726,
+ "grad_norm": 6.100449562072754,
+ "learning_rate": 1.760278428533822e-05,
+ "loss": 0.427,
+ "step": 9113
+ },
+ {
+ "epoch": 24.833787465940055,
+ "grad_norm": 5.359560012817383,
+ "learning_rate": 1.7602210989561202e-05,
+ "loss": 0.4301,
+ "step": 9114
+ },
+ {
+ "epoch": 24.83651226158038,
+ "grad_norm": 6.18666410446167,
+ "learning_rate": 1.760163763457841e-05,
+ "loss": 0.3946,
+ "step": 9115
+ },
+ {
+ "epoch": 24.839237057220707,
+ "grad_norm": 5.748347282409668,
+ "learning_rate": 1.7601064220394305e-05,
+ "loss": 0.3292,
+ "step": 9116
+ },
+ {
+ "epoch": 24.841961852861036,
+ "grad_norm": 6.263171672821045,
+ "learning_rate": 1.760049074701336e-05,
+ "loss": 0.2596,
+ "step": 9117
+ },
+ {
+ "epoch": 24.844686648501362,
+ "grad_norm": 5.487074375152588,
+ "learning_rate": 1.7599917214440035e-05,
+ "loss": 0.4161,
+ "step": 9118
+ },
+ {
+ "epoch": 24.847411444141688,
+ "grad_norm": 4.927175998687744,
+ "learning_rate": 1.7599343622678802e-05,
+ "loss": 0.3557,
+ "step": 9119
+ },
+ {
+ "epoch": 24.850136239782017,
+ "grad_norm": 6.068423271179199,
+ "learning_rate": 1.7598769971734127e-05,
+ "loss": 0.5449,
+ "step": 9120
+ },
+ {
+ "epoch": 24.852861035422343,
+ "grad_norm": 6.664170265197754,
+ "learning_rate": 1.7598196261610476e-05,
+ "loss": 0.4018,
+ "step": 9121
+ },
+ {
+ "epoch": 24.85558583106267,
+ "grad_norm": 5.204150676727295,
+ "learning_rate": 1.7597622492312313e-05,
+ "loss": 0.3827,
+ "step": 9122
+ },
+ {
+ "epoch": 24.858310626703,
+ "grad_norm": 4.71235990524292,
+ "learning_rate": 1.759704866384411e-05,
+ "loss": 0.2475,
+ "step": 9123
+ },
+ {
+ "epoch": 24.861035422343324,
+ "grad_norm": 4.915319919586182,
+ "learning_rate": 1.759647477621034e-05,
+ "loss": 0.4243,
+ "step": 9124
+ },
+ {
+ "epoch": 24.86376021798365,
+ "grad_norm": 5.743703842163086,
+ "learning_rate": 1.759590082941547e-05,
+ "loss": 0.5135,
+ "step": 9125
+ },
+ {
+ "epoch": 24.86648501362398,
+ "grad_norm": 5.761299133300781,
+ "learning_rate": 1.7595326823463963e-05,
+ "loss": 0.329,
+ "step": 9126
+ },
+ {
+ "epoch": 24.869209809264305,
+ "grad_norm": 4.884932041168213,
+ "learning_rate": 1.75947527583603e-05,
+ "loss": 0.2815,
+ "step": 9127
+ },
+ {
+ "epoch": 24.87193460490463,
+ "grad_norm": 6.46493673324585,
+ "learning_rate": 1.7594178634108944e-05,
+ "loss": 0.4443,
+ "step": 9128
+ },
+ {
+ "epoch": 24.87465940054496,
+ "grad_norm": 5.638288497924805,
+ "learning_rate": 1.759360445071437e-05,
+ "loss": 0.4998,
+ "step": 9129
+ },
+ {
+ "epoch": 24.877384196185286,
+ "grad_norm": 5.071884632110596,
+ "learning_rate": 1.7593030208181052e-05,
+ "loss": 0.5054,
+ "step": 9130
+ },
+ {
+ "epoch": 24.88010899182561,
+ "grad_norm": 5.4245100021362305,
+ "learning_rate": 1.7592455906513456e-05,
+ "loss": 0.5549,
+ "step": 9131
+ },
+ {
+ "epoch": 24.88283378746594,
+ "grad_norm": 6.483921527862549,
+ "learning_rate": 1.7591881545716058e-05,
+ "loss": 0.4008,
+ "step": 9132
+ },
+ {
+ "epoch": 24.885558583106267,
+ "grad_norm": 4.965951919555664,
+ "learning_rate": 1.7591307125793332e-05,
+ "loss": 0.3546,
+ "step": 9133
+ },
+ {
+ "epoch": 24.888283378746593,
+ "grad_norm": 5.43743896484375,
+ "learning_rate": 1.759073264674975e-05,
+ "loss": 0.5785,
+ "step": 9134
+ },
+ {
+ "epoch": 24.891008174386922,
+ "grad_norm": 5.9398956298828125,
+ "learning_rate": 1.7590158108589785e-05,
+ "loss": 0.4575,
+ "step": 9135
+ },
+ {
+ "epoch": 24.893732970027248,
+ "grad_norm": 5.995389938354492,
+ "learning_rate": 1.7589583511317915e-05,
+ "loss": 0.5029,
+ "step": 9136
+ },
+ {
+ "epoch": 24.896457765667574,
+ "grad_norm": 5.372235298156738,
+ "learning_rate": 1.7589008854938613e-05,
+ "loss": 0.3608,
+ "step": 9137
+ },
+ {
+ "epoch": 24.899182561307903,
+ "grad_norm": 5.936992645263672,
+ "learning_rate": 1.7588434139456354e-05,
+ "loss": 0.4138,
+ "step": 9138
+ },
+ {
+ "epoch": 24.90190735694823,
+ "grad_norm": 5.600959300994873,
+ "learning_rate": 1.7587859364875613e-05,
+ "loss": 0.5074,
+ "step": 9139
+ },
+ {
+ "epoch": 24.904632152588555,
+ "grad_norm": 5.758541107177734,
+ "learning_rate": 1.7587284531200866e-05,
+ "loss": 0.3127,
+ "step": 9140
+ },
+ {
+ "epoch": 24.907356948228884,
+ "grad_norm": 5.339210510253906,
+ "learning_rate": 1.7586709638436593e-05,
+ "loss": 0.5391,
+ "step": 9141
+ },
+ {
+ "epoch": 24.91008174386921,
+ "grad_norm": 6.411759376525879,
+ "learning_rate": 1.758613468658727e-05,
+ "loss": 0.3963,
+ "step": 9142
+ },
+ {
+ "epoch": 24.912806539509535,
+ "grad_norm": 6.826527118682861,
+ "learning_rate": 1.7585559675657377e-05,
+ "loss": 0.4825,
+ "step": 9143
+ },
+ {
+ "epoch": 24.915531335149865,
+ "grad_norm": 6.3697381019592285,
+ "learning_rate": 1.7584984605651385e-05,
+ "loss": 0.3922,
+ "step": 9144
+ },
+ {
+ "epoch": 24.91825613079019,
+ "grad_norm": 6.317148685455322,
+ "learning_rate": 1.758440947657378e-05,
+ "loss": 0.4296,
+ "step": 9145
+ },
+ {
+ "epoch": 24.920980926430516,
+ "grad_norm": 7.26331901550293,
+ "learning_rate": 1.7583834288429032e-05,
+ "loss": 0.5178,
+ "step": 9146
+ },
+ {
+ "epoch": 24.923705722070846,
+ "grad_norm": 6.611044883728027,
+ "learning_rate": 1.7583259041221634e-05,
+ "loss": 0.3286,
+ "step": 9147
+ },
+ {
+ "epoch": 24.92643051771117,
+ "grad_norm": 5.8957133293151855,
+ "learning_rate": 1.7582683734956055e-05,
+ "loss": 0.363,
+ "step": 9148
+ },
+ {
+ "epoch": 24.929155313351497,
+ "grad_norm": 5.980958461761475,
+ "learning_rate": 1.758210836963678e-05,
+ "loss": 0.4628,
+ "step": 9149
+ },
+ {
+ "epoch": 24.931880108991827,
+ "grad_norm": 4.979247093200684,
+ "learning_rate": 1.758153294526829e-05,
+ "loss": 0.3523,
+ "step": 9150
+ },
+ {
+ "epoch": 24.934604904632153,
+ "grad_norm": 5.881022930145264,
+ "learning_rate": 1.7580957461855064e-05,
+ "loss": 0.5098,
+ "step": 9151
+ },
+ {
+ "epoch": 24.93732970027248,
+ "grad_norm": 5.594860553741455,
+ "learning_rate": 1.7580381919401586e-05,
+ "loss": 0.4715,
+ "step": 9152
+ },
+ {
+ "epoch": 24.940054495912808,
+ "grad_norm": 6.043241500854492,
+ "learning_rate": 1.7579806317912336e-05,
+ "loss": 0.3055,
+ "step": 9153
+ },
+ {
+ "epoch": 24.942779291553133,
+ "grad_norm": 5.018749713897705,
+ "learning_rate": 1.7579230657391803e-05,
+ "loss": 0.3304,
+ "step": 9154
+ },
+ {
+ "epoch": 24.94550408719346,
+ "grad_norm": 6.202396392822266,
+ "learning_rate": 1.757865493784446e-05,
+ "loss": 0.4147,
+ "step": 9155
+ },
+ {
+ "epoch": 24.94822888283379,
+ "grad_norm": 9.449963569641113,
+ "learning_rate": 1.7578079159274798e-05,
+ "loss": 0.4481,
+ "step": 9156
+ },
+ {
+ "epoch": 24.950953678474114,
+ "grad_norm": 7.409104347229004,
+ "learning_rate": 1.75775033216873e-05,
+ "loss": 0.4634,
+ "step": 9157
+ },
+ {
+ "epoch": 24.95367847411444,
+ "grad_norm": 4.017806053161621,
+ "learning_rate": 1.757692742508645e-05,
+ "loss": 0.3206,
+ "step": 9158
+ },
+ {
+ "epoch": 24.95640326975477,
+ "grad_norm": 6.900577068328857,
+ "learning_rate": 1.7576351469476735e-05,
+ "loss": 0.423,
+ "step": 9159
+ },
+ {
+ "epoch": 24.959128065395095,
+ "grad_norm": 6.460623264312744,
+ "learning_rate": 1.7575775454862636e-05,
+ "loss": 0.4009,
+ "step": 9160
+ },
+ {
+ "epoch": 24.96185286103542,
+ "grad_norm": 6.528751850128174,
+ "learning_rate": 1.7575199381248643e-05,
+ "loss": 0.4219,
+ "step": 9161
+ },
+ {
+ "epoch": 24.96457765667575,
+ "grad_norm": 6.1053996086120605,
+ "learning_rate": 1.757462324863924e-05,
+ "loss": 0.3516,
+ "step": 9162
+ },
+ {
+ "epoch": 24.967302452316076,
+ "grad_norm": 5.956650733947754,
+ "learning_rate": 1.7574047057038914e-05,
+ "loss": 0.4384,
+ "step": 9163
+ },
+ {
+ "epoch": 24.970027247956402,
+ "grad_norm": 5.902890682220459,
+ "learning_rate": 1.7573470806452156e-05,
+ "loss": 0.4669,
+ "step": 9164
+ },
+ {
+ "epoch": 24.97275204359673,
+ "grad_norm": 5.7701544761657715,
+ "learning_rate": 1.7572894496883448e-05,
+ "loss": 0.4149,
+ "step": 9165
+ },
+ {
+ "epoch": 24.975476839237057,
+ "grad_norm": 6.656943321228027,
+ "learning_rate": 1.7572318128337286e-05,
+ "loss": 0.3624,
+ "step": 9166
+ },
+ {
+ "epoch": 24.978201634877383,
+ "grad_norm": 6.860302448272705,
+ "learning_rate": 1.757174170081815e-05,
+ "loss": 0.5593,
+ "step": 9167
+ },
+ {
+ "epoch": 24.980926430517712,
+ "grad_norm": 5.622548580169678,
+ "learning_rate": 1.7571165214330533e-05,
+ "loss": 0.3493,
+ "step": 9168
+ },
+ {
+ "epoch": 24.983651226158038,
+ "grad_norm": 6.355636119842529,
+ "learning_rate": 1.7570588668878925e-05,
+ "loss": 0.6282,
+ "step": 9169
+ },
+ {
+ "epoch": 24.986376021798364,
+ "grad_norm": 5.640994548797607,
+ "learning_rate": 1.757001206446782e-05,
+ "loss": 0.465,
+ "step": 9170
+ },
+ {
+ "epoch": 24.989100817438693,
+ "grad_norm": 421.69952392578125,
+ "learning_rate": 1.7569435401101702e-05,
+ "loss": 0.3846,
+ "step": 9171
+ },
+ {
+ "epoch": 24.99182561307902,
+ "grad_norm": 5.581060886383057,
+ "learning_rate": 1.7568858678785067e-05,
+ "loss": 0.502,
+ "step": 9172
+ },
+ {
+ "epoch": 24.994550408719345,
+ "grad_norm": 6.82476282119751,
+ "learning_rate": 1.75682818975224e-05,
+ "loss": 0.4932,
+ "step": 9173
+ },
+ {
+ "epoch": 24.997275204359674,
+ "grad_norm": 7.723385334014893,
+ "learning_rate": 1.75677050573182e-05,
+ "loss": 0.4769,
+ "step": 9174
+ },
+ {
+ "epoch": 25.0,
+ "grad_norm": 7.597902774810791,
+ "learning_rate": 1.7567128158176955e-05,
+ "loss": 0.6104,
+ "step": 9175
+ },
+ {
+ "epoch": 25.002724795640326,
+ "grad_norm": 9.60449504852295,
+ "learning_rate": 1.756655120010316e-05,
+ "loss": 0.5697,
+ "step": 9176
+ },
+ {
+ "epoch": 25.005449591280655,
+ "grad_norm": 9.60594367980957,
+ "learning_rate": 1.756597418310131e-05,
+ "loss": 0.4288,
+ "step": 9177
+ },
+ {
+ "epoch": 25.00817438692098,
+ "grad_norm": 12.010396957397461,
+ "learning_rate": 1.7565397107175893e-05,
+ "loss": 0.6053,
+ "step": 9178
+ },
+ {
+ "epoch": 25.010899182561307,
+ "grad_norm": 12.338308334350586,
+ "learning_rate": 1.7564819972331408e-05,
+ "loss": 0.5698,
+ "step": 9179
+ },
+ {
+ "epoch": 25.013623978201636,
+ "grad_norm": 9.454119682312012,
+ "learning_rate": 1.7564242778572348e-05,
+ "loss": 0.624,
+ "step": 9180
+ },
+ {
+ "epoch": 25.016348773841962,
+ "grad_norm": 10.310940742492676,
+ "learning_rate": 1.756366552590321e-05,
+ "loss": 0.4464,
+ "step": 9181
+ },
+ {
+ "epoch": 25.019073569482288,
+ "grad_norm": 13.576168060302734,
+ "learning_rate": 1.756308821432849e-05,
+ "loss": 0.5324,
+ "step": 9182
+ },
+ {
+ "epoch": 25.021798365122617,
+ "grad_norm": 11.421992301940918,
+ "learning_rate": 1.7562510843852678e-05,
+ "loss": 0.546,
+ "step": 9183
+ },
+ {
+ "epoch": 25.024523160762943,
+ "grad_norm": 19.36672592163086,
+ "learning_rate": 1.7561933414480278e-05,
+ "loss": 0.5693,
+ "step": 9184
+ },
+ {
+ "epoch": 25.02724795640327,
+ "grad_norm": 11.04211711883545,
+ "learning_rate": 1.7561355926215784e-05,
+ "loss": 0.5866,
+ "step": 9185
+ },
+ {
+ "epoch": 25.029972752043598,
+ "grad_norm": 9.017316818237305,
+ "learning_rate": 1.7560778379063693e-05,
+ "loss": 0.5226,
+ "step": 9186
+ },
+ {
+ "epoch": 25.032697547683924,
+ "grad_norm": 9.203912734985352,
+ "learning_rate": 1.75602007730285e-05,
+ "loss": 0.5224,
+ "step": 9187
+ },
+ {
+ "epoch": 25.03542234332425,
+ "grad_norm": 8.069424629211426,
+ "learning_rate": 1.7559623108114713e-05,
+ "loss": 0.4943,
+ "step": 9188
+ },
+ {
+ "epoch": 25.03814713896458,
+ "grad_norm": 9.20228099822998,
+ "learning_rate": 1.755904538432682e-05,
+ "loss": 0.6708,
+ "step": 9189
+ },
+ {
+ "epoch": 25.040871934604905,
+ "grad_norm": 7.994393348693848,
+ "learning_rate": 1.7558467601669326e-05,
+ "loss": 0.6422,
+ "step": 9190
+ },
+ {
+ "epoch": 25.04359673024523,
+ "grad_norm": 7.7300567626953125,
+ "learning_rate": 1.755788976014673e-05,
+ "loss": 0.4911,
+ "step": 9191
+ },
+ {
+ "epoch": 25.04632152588556,
+ "grad_norm": 9.254310607910156,
+ "learning_rate": 1.755731185976353e-05,
+ "loss": 0.4701,
+ "step": 9192
+ },
+ {
+ "epoch": 25.049046321525886,
+ "grad_norm": 7.002472877502441,
+ "learning_rate": 1.755673390052423e-05,
+ "loss": 0.3925,
+ "step": 9193
+ },
+ {
+ "epoch": 25.05177111716621,
+ "grad_norm": 8.597843170166016,
+ "learning_rate": 1.755615588243333e-05,
+ "loss": 0.4078,
+ "step": 9194
+ },
+ {
+ "epoch": 25.05449591280654,
+ "grad_norm": 8.937694549560547,
+ "learning_rate": 1.755557780549533e-05,
+ "loss": 0.4384,
+ "step": 9195
+ },
+ {
+ "epoch": 25.057220708446867,
+ "grad_norm": 7.242659091949463,
+ "learning_rate": 1.7554999669714736e-05,
+ "loss": 0.5779,
+ "step": 9196
+ },
+ {
+ "epoch": 25.059945504087192,
+ "grad_norm": 6.1539387702941895,
+ "learning_rate": 1.7554421475096047e-05,
+ "loss": 0.4781,
+ "step": 9197
+ },
+ {
+ "epoch": 25.06267029972752,
+ "grad_norm": 7.02477502822876,
+ "learning_rate": 1.7553843221643764e-05,
+ "loss": 0.4529,
+ "step": 9198
+ },
+ {
+ "epoch": 25.065395095367847,
+ "grad_norm": 7.648223876953125,
+ "learning_rate": 1.7553264909362397e-05,
+ "loss": 0.5692,
+ "step": 9199
+ },
+ {
+ "epoch": 25.068119891008173,
+ "grad_norm": 7.227846145629883,
+ "learning_rate": 1.755268653825644e-05,
+ "loss": 0.593,
+ "step": 9200
+ },
+ {
+ "epoch": 25.070844686648503,
+ "grad_norm": 11.580743789672852,
+ "learning_rate": 1.7552108108330407e-05,
+ "loss": 0.484,
+ "step": 9201
+ },
+ {
+ "epoch": 25.07356948228883,
+ "grad_norm": 10.157035827636719,
+ "learning_rate": 1.75515296195888e-05,
+ "loss": 0.4254,
+ "step": 9202
+ },
+ {
+ "epoch": 25.076294277929154,
+ "grad_norm": 7.166711330413818,
+ "learning_rate": 1.755095107203612e-05,
+ "loss": 0.6195,
+ "step": 9203
+ },
+ {
+ "epoch": 25.079019073569484,
+ "grad_norm": 8.808980941772461,
+ "learning_rate": 1.755037246567688e-05,
+ "loss": 0.5246,
+ "step": 9204
+ },
+ {
+ "epoch": 25.08174386920981,
+ "grad_norm": 12.502464294433594,
+ "learning_rate": 1.754979380051558e-05,
+ "loss": 0.4339,
+ "step": 9205
+ },
+ {
+ "epoch": 25.084468664850135,
+ "grad_norm": 7.2137980461120605,
+ "learning_rate": 1.754921507655673e-05,
+ "loss": 0.5745,
+ "step": 9206
+ },
+ {
+ "epoch": 25.087193460490465,
+ "grad_norm": 7.667659282684326,
+ "learning_rate": 1.7548636293804838e-05,
+ "loss": 0.5934,
+ "step": 9207
+ },
+ {
+ "epoch": 25.08991825613079,
+ "grad_norm": 6.594106197357178,
+ "learning_rate": 1.7548057452264404e-05,
+ "loss": 0.5295,
+ "step": 9208
+ },
+ {
+ "epoch": 25.092643051771116,
+ "grad_norm": 8.067695617675781,
+ "learning_rate": 1.7547478551939944e-05,
+ "loss": 0.447,
+ "step": 9209
+ },
+ {
+ "epoch": 25.095367847411445,
+ "grad_norm": 29.305702209472656,
+ "learning_rate": 1.7546899592835968e-05,
+ "loss": 0.4026,
+ "step": 9210
+ },
+ {
+ "epoch": 25.09809264305177,
+ "grad_norm": 6.708586692810059,
+ "learning_rate": 1.7546320574956976e-05,
+ "loss": 0.463,
+ "step": 9211
+ },
+ {
+ "epoch": 25.100817438692097,
+ "grad_norm": 7.65814208984375,
+ "learning_rate": 1.7545741498307483e-05,
+ "loss": 0.4276,
+ "step": 9212
+ },
+ {
+ "epoch": 25.103542234332426,
+ "grad_norm": 6.719625949859619,
+ "learning_rate": 1.7545162362891997e-05,
+ "loss": 0.381,
+ "step": 9213
+ },
+ {
+ "epoch": 25.106267029972752,
+ "grad_norm": 8.442155838012695,
+ "learning_rate": 1.754458316871503e-05,
+ "loss": 0.4301,
+ "step": 9214
+ },
+ {
+ "epoch": 25.108991825613078,
+ "grad_norm": 7.2818603515625,
+ "learning_rate": 1.7544003915781097e-05,
+ "loss": 0.4919,
+ "step": 9215
+ },
+ {
+ "epoch": 25.111716621253407,
+ "grad_norm": 8.89192008972168,
+ "learning_rate": 1.7543424604094697e-05,
+ "loss": 0.4326,
+ "step": 9216
+ },
+ {
+ "epoch": 25.114441416893733,
+ "grad_norm": 7.811793804168701,
+ "learning_rate": 1.754284523366035e-05,
+ "loss": 0.5294,
+ "step": 9217
+ },
+ {
+ "epoch": 25.11716621253406,
+ "grad_norm": 7.885205268859863,
+ "learning_rate": 1.7542265804482573e-05,
+ "loss": 0.6017,
+ "step": 9218
+ },
+ {
+ "epoch": 25.11989100817439,
+ "grad_norm": 6.005732536315918,
+ "learning_rate": 1.754168631656587e-05,
+ "loss": 0.4796,
+ "step": 9219
+ },
+ {
+ "epoch": 25.122615803814714,
+ "grad_norm": 7.2632269859313965,
+ "learning_rate": 1.754110676991475e-05,
+ "loss": 0.473,
+ "step": 9220
+ },
+ {
+ "epoch": 25.12534059945504,
+ "grad_norm": 5.173118591308594,
+ "learning_rate": 1.7540527164533744e-05,
+ "loss": 0.5084,
+ "step": 9221
+ },
+ {
+ "epoch": 25.12806539509537,
+ "grad_norm": 5.963922023773193,
+ "learning_rate": 1.7539947500427346e-05,
+ "loss": 0.4399,
+ "step": 9222
+ },
+ {
+ "epoch": 25.130790190735695,
+ "grad_norm": 6.863542556762695,
+ "learning_rate": 1.7539367777600088e-05,
+ "loss": 0.4054,
+ "step": 9223
+ },
+ {
+ "epoch": 25.13351498637602,
+ "grad_norm": 5.678715229034424,
+ "learning_rate": 1.753878799605647e-05,
+ "loss": 0.4289,
+ "step": 9224
+ },
+ {
+ "epoch": 25.13623978201635,
+ "grad_norm": 7.171862602233887,
+ "learning_rate": 1.7538208155801014e-05,
+ "loss": 0.459,
+ "step": 9225
+ },
+ {
+ "epoch": 25.138964577656676,
+ "grad_norm": 7.243014335632324,
+ "learning_rate": 1.7537628256838238e-05,
+ "loss": 0.4376,
+ "step": 9226
+ },
+ {
+ "epoch": 25.141689373297,
+ "grad_norm": 7.341374397277832,
+ "learning_rate": 1.7537048299172657e-05,
+ "loss": 0.4075,
+ "step": 9227
+ },
+ {
+ "epoch": 25.14441416893733,
+ "grad_norm": 6.940046310424805,
+ "learning_rate": 1.7536468282808787e-05,
+ "loss": 0.4278,
+ "step": 9228
+ },
+ {
+ "epoch": 25.147138964577657,
+ "grad_norm": 8.078675270080566,
+ "learning_rate": 1.753588820775114e-05,
+ "loss": 0.4744,
+ "step": 9229
+ },
+ {
+ "epoch": 25.149863760217983,
+ "grad_norm": 6.404881954193115,
+ "learning_rate": 1.7535308074004242e-05,
+ "loss": 0.4152,
+ "step": 9230
+ },
+ {
+ "epoch": 25.152588555858312,
+ "grad_norm": 6.236560344696045,
+ "learning_rate": 1.7534727881572603e-05,
+ "loss": 0.458,
+ "step": 9231
+ },
+ {
+ "epoch": 25.155313351498638,
+ "grad_norm": 6.243899822235107,
+ "learning_rate": 1.753414763046075e-05,
+ "loss": 0.4051,
+ "step": 9232
+ },
+ {
+ "epoch": 25.158038147138964,
+ "grad_norm": 6.456615924835205,
+ "learning_rate": 1.7533567320673197e-05,
+ "loss": 0.4629,
+ "step": 9233
+ },
+ {
+ "epoch": 25.160762942779293,
+ "grad_norm": 6.56041955947876,
+ "learning_rate": 1.7532986952214458e-05,
+ "loss": 0.4037,
+ "step": 9234
+ },
+ {
+ "epoch": 25.16348773841962,
+ "grad_norm": 7.606236934661865,
+ "learning_rate": 1.7532406525089065e-05,
+ "loss": 0.3604,
+ "step": 9235
+ },
+ {
+ "epoch": 25.166212534059945,
+ "grad_norm": 6.660259246826172,
+ "learning_rate": 1.7531826039301534e-05,
+ "loss": 0.4528,
+ "step": 9236
+ },
+ {
+ "epoch": 25.168937329700274,
+ "grad_norm": 7.934656620025635,
+ "learning_rate": 1.7531245494856383e-05,
+ "loss": 0.5214,
+ "step": 9237
+ },
+ {
+ "epoch": 25.1716621253406,
+ "grad_norm": 8.484705924987793,
+ "learning_rate": 1.753066489175813e-05,
+ "loss": 0.6052,
+ "step": 9238
+ },
+ {
+ "epoch": 25.174386920980925,
+ "grad_norm": 6.998429298400879,
+ "learning_rate": 1.7530084230011305e-05,
+ "loss": 0.4489,
+ "step": 9239
+ },
+ {
+ "epoch": 25.177111716621255,
+ "grad_norm": 7.183679103851318,
+ "learning_rate": 1.7529503509620426e-05,
+ "loss": 0.3872,
+ "step": 9240
+ },
+ {
+ "epoch": 25.17983651226158,
+ "grad_norm": 7.253841876983643,
+ "learning_rate": 1.7528922730590014e-05,
+ "loss": 0.3907,
+ "step": 9241
+ },
+ {
+ "epoch": 25.182561307901906,
+ "grad_norm": 6.764344215393066,
+ "learning_rate": 1.7528341892924598e-05,
+ "loss": 0.409,
+ "step": 9242
+ },
+ {
+ "epoch": 25.185286103542236,
+ "grad_norm": 6.599880695343018,
+ "learning_rate": 1.7527760996628695e-05,
+ "loss": 0.5244,
+ "step": 9243
+ },
+ {
+ "epoch": 25.18801089918256,
+ "grad_norm": 6.207182884216309,
+ "learning_rate": 1.7527180041706834e-05,
+ "loss": 0.4219,
+ "step": 9244
+ },
+ {
+ "epoch": 25.190735694822887,
+ "grad_norm": 6.098278999328613,
+ "learning_rate": 1.7526599028163534e-05,
+ "loss": 0.5291,
+ "step": 9245
+ },
+ {
+ "epoch": 25.193460490463217,
+ "grad_norm": 9.025604248046875,
+ "learning_rate": 1.7526017956003324e-05,
+ "loss": 0.4296,
+ "step": 9246
+ },
+ {
+ "epoch": 25.196185286103542,
+ "grad_norm": 6.238251209259033,
+ "learning_rate": 1.7525436825230728e-05,
+ "loss": 0.2836,
+ "step": 9247
+ },
+ {
+ "epoch": 25.19891008174387,
+ "grad_norm": 5.595916271209717,
+ "learning_rate": 1.7524855635850272e-05,
+ "loss": 0.3926,
+ "step": 9248
+ },
+ {
+ "epoch": 25.201634877384198,
+ "grad_norm": 7.216546535491943,
+ "learning_rate": 1.7524274387866483e-05,
+ "loss": 0.5383,
+ "step": 9249
+ },
+ {
+ "epoch": 25.204359673024523,
+ "grad_norm": 6.588919162750244,
+ "learning_rate": 1.752369308128389e-05,
+ "loss": 0.3379,
+ "step": 9250
+ },
+ {
+ "epoch": 25.20708446866485,
+ "grad_norm": 5.629561901092529,
+ "learning_rate": 1.7523111716107016e-05,
+ "loss": 0.3605,
+ "step": 9251
+ },
+ {
+ "epoch": 25.20980926430518,
+ "grad_norm": 6.839292526245117,
+ "learning_rate": 1.7522530292340387e-05,
+ "loss": 0.4694,
+ "step": 9252
+ },
+ {
+ "epoch": 25.212534059945504,
+ "grad_norm": 5.325597286224365,
+ "learning_rate": 1.7521948809988538e-05,
+ "loss": 0.4357,
+ "step": 9253
+ },
+ {
+ "epoch": 25.21525885558583,
+ "grad_norm": 7.831445217132568,
+ "learning_rate": 1.752136726905599e-05,
+ "loss": 0.439,
+ "step": 9254
+ },
+ {
+ "epoch": 25.21798365122616,
+ "grad_norm": 6.591104507446289,
+ "learning_rate": 1.752078566954728e-05,
+ "loss": 0.3864,
+ "step": 9255
+ },
+ {
+ "epoch": 25.220708446866485,
+ "grad_norm": 6.260085582733154,
+ "learning_rate": 1.752020401146693e-05,
+ "loss": 0.4812,
+ "step": 9256
+ },
+ {
+ "epoch": 25.22343324250681,
+ "grad_norm": 5.982943534851074,
+ "learning_rate": 1.7519622294819474e-05,
+ "loss": 0.3713,
+ "step": 9257
+ },
+ {
+ "epoch": 25.22615803814714,
+ "grad_norm": 7.15734338760376,
+ "learning_rate": 1.7519040519609443e-05,
+ "loss": 0.4473,
+ "step": 9258
+ },
+ {
+ "epoch": 25.228882833787466,
+ "grad_norm": 5.3927106857299805,
+ "learning_rate": 1.751845868584136e-05,
+ "loss": 0.3001,
+ "step": 9259
+ },
+ {
+ "epoch": 25.231607629427792,
+ "grad_norm": 5.456971168518066,
+ "learning_rate": 1.7517876793519772e-05,
+ "loss": 0.5421,
+ "step": 9260
+ },
+ {
+ "epoch": 25.23433242506812,
+ "grad_norm": 5.600273609161377,
+ "learning_rate": 1.75172948426492e-05,
+ "loss": 0.381,
+ "step": 9261
+ },
+ {
+ "epoch": 25.237057220708447,
+ "grad_norm": 5.7097320556640625,
+ "learning_rate": 1.751671283323417e-05,
+ "loss": 0.4266,
+ "step": 9262
+ },
+ {
+ "epoch": 25.239782016348773,
+ "grad_norm": 5.386323928833008,
+ "learning_rate": 1.751613076527923e-05,
+ "loss": 0.4469,
+ "step": 9263
+ },
+ {
+ "epoch": 25.242506811989102,
+ "grad_norm": 5.671041488647461,
+ "learning_rate": 1.7515548638788904e-05,
+ "loss": 0.4424,
+ "step": 9264
+ },
+ {
+ "epoch": 25.245231607629428,
+ "grad_norm": 6.289502143859863,
+ "learning_rate": 1.7514966453767725e-05,
+ "loss": 0.3772,
+ "step": 9265
+ },
+ {
+ "epoch": 25.247956403269754,
+ "grad_norm": 7.657263278961182,
+ "learning_rate": 1.751438421022023e-05,
+ "loss": 0.6123,
+ "step": 9266
+ },
+ {
+ "epoch": 25.250681198910083,
+ "grad_norm": 18.1718692779541,
+ "learning_rate": 1.7513801908150953e-05,
+ "loss": 0.497,
+ "step": 9267
+ },
+ {
+ "epoch": 25.25340599455041,
+ "grad_norm": 7.065165996551514,
+ "learning_rate": 1.7513219547564427e-05,
+ "loss": 0.3799,
+ "step": 9268
+ },
+ {
+ "epoch": 25.256130790190735,
+ "grad_norm": 7.403648853302002,
+ "learning_rate": 1.7512637128465195e-05,
+ "loss": 0.4446,
+ "step": 9269
+ },
+ {
+ "epoch": 25.258855585831064,
+ "grad_norm": 5.60500955581665,
+ "learning_rate": 1.751205465085778e-05,
+ "loss": 0.4512,
+ "step": 9270
+ },
+ {
+ "epoch": 25.26158038147139,
+ "grad_norm": 5.742873191833496,
+ "learning_rate": 1.7511472114746727e-05,
+ "loss": 0.4292,
+ "step": 9271
+ },
+ {
+ "epoch": 25.264305177111716,
+ "grad_norm": 8.447866439819336,
+ "learning_rate": 1.7510889520136572e-05,
+ "loss": 0.3664,
+ "step": 9272
+ },
+ {
+ "epoch": 25.267029972752045,
+ "grad_norm": 14.001521110534668,
+ "learning_rate": 1.751030686703185e-05,
+ "loss": 0.3725,
+ "step": 9273
+ },
+ {
+ "epoch": 25.26975476839237,
+ "grad_norm": 5.547813415527344,
+ "learning_rate": 1.7509724155437104e-05,
+ "loss": 0.4427,
+ "step": 9274
+ },
+ {
+ "epoch": 25.272479564032697,
+ "grad_norm": 6.379539966583252,
+ "learning_rate": 1.750914138535686e-05,
+ "loss": 0.3105,
+ "step": 9275
+ },
+ {
+ "epoch": 25.275204359673026,
+ "grad_norm": 7.451393127441406,
+ "learning_rate": 1.750855855679567e-05,
+ "loss": 0.3867,
+ "step": 9276
+ },
+ {
+ "epoch": 25.277929155313352,
+ "grad_norm": 6.188321590423584,
+ "learning_rate": 1.750797566975807e-05,
+ "loss": 0.4244,
+ "step": 9277
+ },
+ {
+ "epoch": 25.280653950953678,
+ "grad_norm": 6.606024265289307,
+ "learning_rate": 1.7507392724248592e-05,
+ "loss": 0.4526,
+ "step": 9278
+ },
+ {
+ "epoch": 25.283378746594007,
+ "grad_norm": 7.5595173835754395,
+ "learning_rate": 1.750680972027178e-05,
+ "loss": 0.4405,
+ "step": 9279
+ },
+ {
+ "epoch": 25.286103542234333,
+ "grad_norm": 4.500346660614014,
+ "learning_rate": 1.750622665783218e-05,
+ "loss": 0.4827,
+ "step": 9280
+ },
+ {
+ "epoch": 25.28882833787466,
+ "grad_norm": 7.574316501617432,
+ "learning_rate": 1.7505643536934326e-05,
+ "loss": 0.5822,
+ "step": 9281
+ },
+ {
+ "epoch": 25.291553133514988,
+ "grad_norm": 6.954140663146973,
+ "learning_rate": 1.750506035758276e-05,
+ "loss": 0.4285,
+ "step": 9282
+ },
+ {
+ "epoch": 25.294277929155314,
+ "grad_norm": 9.19731616973877,
+ "learning_rate": 1.750447711978203e-05,
+ "loss": 0.4156,
+ "step": 9283
+ },
+ {
+ "epoch": 25.29700272479564,
+ "grad_norm": 4.962310314178467,
+ "learning_rate": 1.7503893823536675e-05,
+ "loss": 0.3965,
+ "step": 9284
+ },
+ {
+ "epoch": 25.29972752043597,
+ "grad_norm": 7.17042875289917,
+ "learning_rate": 1.7503310468851232e-05,
+ "loss": 0.468,
+ "step": 9285
+ },
+ {
+ "epoch": 25.302452316076295,
+ "grad_norm": 6.362421989440918,
+ "learning_rate": 1.750272705573025e-05,
+ "loss": 0.302,
+ "step": 9286
+ },
+ {
+ "epoch": 25.30517711171662,
+ "grad_norm": 5.822933197021484,
+ "learning_rate": 1.750214358417827e-05,
+ "loss": 0.367,
+ "step": 9287
+ },
+ {
+ "epoch": 25.30790190735695,
+ "grad_norm": 8.94520378112793,
+ "learning_rate": 1.750156005419984e-05,
+ "loss": 0.4989,
+ "step": 9288
+ },
+ {
+ "epoch": 25.310626702997276,
+ "grad_norm": 5.624583721160889,
+ "learning_rate": 1.7500976465799496e-05,
+ "loss": 0.7115,
+ "step": 9289
+ },
+ {
+ "epoch": 25.3133514986376,
+ "grad_norm": 11.277554512023926,
+ "learning_rate": 1.7500392818981792e-05,
+ "loss": 0.3924,
+ "step": 9290
+ },
+ {
+ "epoch": 25.31607629427793,
+ "grad_norm": 5.809271335601807,
+ "learning_rate": 1.749980911375127e-05,
+ "loss": 0.3959,
+ "step": 9291
+ },
+ {
+ "epoch": 25.318801089918257,
+ "grad_norm": 5.607304096221924,
+ "learning_rate": 1.7499225350112475e-05,
+ "loss": 0.3732,
+ "step": 9292
+ },
+ {
+ "epoch": 25.321525885558582,
+ "grad_norm": 6.443030834197998,
+ "learning_rate": 1.7498641528069956e-05,
+ "loss": 0.4068,
+ "step": 9293
+ },
+ {
+ "epoch": 25.32425068119891,
+ "grad_norm": 6.422985553741455,
+ "learning_rate": 1.749805764762826e-05,
+ "loss": 0.4725,
+ "step": 9294
+ },
+ {
+ "epoch": 25.326975476839237,
+ "grad_norm": 5.0748491287231445,
+ "learning_rate": 1.749747370879193e-05,
+ "loss": 0.4275,
+ "step": 9295
+ },
+ {
+ "epoch": 25.329700272479563,
+ "grad_norm": 6.2115325927734375,
+ "learning_rate": 1.7496889711565516e-05,
+ "loss": 0.4953,
+ "step": 9296
+ },
+ {
+ "epoch": 25.332425068119893,
+ "grad_norm": 6.497741222381592,
+ "learning_rate": 1.7496305655953567e-05,
+ "loss": 0.3317,
+ "step": 9297
+ },
+ {
+ "epoch": 25.33514986376022,
+ "grad_norm": 7.866917610168457,
+ "learning_rate": 1.749572154196063e-05,
+ "loss": 0.4189,
+ "step": 9298
+ },
+ {
+ "epoch": 25.337874659400544,
+ "grad_norm": 5.386141777038574,
+ "learning_rate": 1.749513736959126e-05,
+ "loss": 0.5776,
+ "step": 9299
+ },
+ {
+ "epoch": 25.340599455040874,
+ "grad_norm": 7.379188060760498,
+ "learning_rate": 1.7494553138849993e-05,
+ "loss": 0.3466,
+ "step": 9300
+ },
+ {
+ "epoch": 25.3433242506812,
+ "grad_norm": 6.158961296081543,
+ "learning_rate": 1.749396884974139e-05,
+ "loss": 0.5743,
+ "step": 9301
+ },
+ {
+ "epoch": 25.346049046321525,
+ "grad_norm": 7.3553948402404785,
+ "learning_rate": 1.7493384502270003e-05,
+ "loss": 0.5161,
+ "step": 9302
+ },
+ {
+ "epoch": 25.348773841961854,
+ "grad_norm": 7.207698345184326,
+ "learning_rate": 1.7492800096440376e-05,
+ "loss": 0.3448,
+ "step": 9303
+ },
+ {
+ "epoch": 25.35149863760218,
+ "grad_norm": 5.6903157234191895,
+ "learning_rate": 1.749221563225706e-05,
+ "loss": 0.4448,
+ "step": 9304
+ },
+ {
+ "epoch": 25.354223433242506,
+ "grad_norm": 6.536190032958984,
+ "learning_rate": 1.7491631109724613e-05,
+ "loss": 0.5129,
+ "step": 9305
+ },
+ {
+ "epoch": 25.356948228882835,
+ "grad_norm": 5.492410182952881,
+ "learning_rate": 1.7491046528847584e-05,
+ "loss": 0.4062,
+ "step": 9306
+ },
+ {
+ "epoch": 25.35967302452316,
+ "grad_norm": 6.7478556632995605,
+ "learning_rate": 1.7490461889630523e-05,
+ "loss": 0.4094,
+ "step": 9307
+ },
+ {
+ "epoch": 25.362397820163487,
+ "grad_norm": 12.042680740356445,
+ "learning_rate": 1.748987719207799e-05,
+ "loss": 0.3854,
+ "step": 9308
+ },
+ {
+ "epoch": 25.365122615803816,
+ "grad_norm": 6.677364349365234,
+ "learning_rate": 1.748929243619453e-05,
+ "loss": 0.5107,
+ "step": 9309
+ },
+ {
+ "epoch": 25.367847411444142,
+ "grad_norm": 6.435739040374756,
+ "learning_rate": 1.7488707621984702e-05,
+ "loss": 0.3479,
+ "step": 9310
+ },
+ {
+ "epoch": 25.370572207084468,
+ "grad_norm": 5.602840423583984,
+ "learning_rate": 1.7488122749453067e-05,
+ "loss": 0.4135,
+ "step": 9311
+ },
+ {
+ "epoch": 25.373297002724797,
+ "grad_norm": 5.576169967651367,
+ "learning_rate": 1.7487537818604166e-05,
+ "loss": 0.4619,
+ "step": 9312
+ },
+ {
+ "epoch": 25.376021798365123,
+ "grad_norm": 6.137109279632568,
+ "learning_rate": 1.7486952829442562e-05,
+ "loss": 0.4504,
+ "step": 9313
+ },
+ {
+ "epoch": 25.37874659400545,
+ "grad_norm": 6.180505752563477,
+ "learning_rate": 1.7486367781972813e-05,
+ "loss": 0.5272,
+ "step": 9314
+ },
+ {
+ "epoch": 25.381471389645778,
+ "grad_norm": 7.081398010253906,
+ "learning_rate": 1.7485782676199475e-05,
+ "loss": 0.4865,
+ "step": 9315
+ },
+ {
+ "epoch": 25.384196185286104,
+ "grad_norm": 7.056877136230469,
+ "learning_rate": 1.7485197512127096e-05,
+ "loss": 0.317,
+ "step": 9316
+ },
+ {
+ "epoch": 25.38692098092643,
+ "grad_norm": 9.547191619873047,
+ "learning_rate": 1.7484612289760242e-05,
+ "loss": 0.3692,
+ "step": 9317
+ },
+ {
+ "epoch": 25.38964577656676,
+ "grad_norm": 5.58565616607666,
+ "learning_rate": 1.748402700910347e-05,
+ "loss": 0.4453,
+ "step": 9318
+ },
+ {
+ "epoch": 25.392370572207085,
+ "grad_norm": 9.23723316192627,
+ "learning_rate": 1.7483441670161336e-05,
+ "loss": 0.3096,
+ "step": 9319
+ },
+ {
+ "epoch": 25.39509536784741,
+ "grad_norm": 6.172657012939453,
+ "learning_rate": 1.7482856272938398e-05,
+ "loss": 0.3919,
+ "step": 9320
+ },
+ {
+ "epoch": 25.39782016348774,
+ "grad_norm": 8.690281867980957,
+ "learning_rate": 1.7482270817439217e-05,
+ "loss": 0.6666,
+ "step": 9321
+ },
+ {
+ "epoch": 25.400544959128066,
+ "grad_norm": 6.293361186981201,
+ "learning_rate": 1.7481685303668353e-05,
+ "loss": 0.417,
+ "step": 9322
+ },
+ {
+ "epoch": 25.40326975476839,
+ "grad_norm": 6.252566814422607,
+ "learning_rate": 1.7481099731630365e-05,
+ "loss": 0.4028,
+ "step": 9323
+ },
+ {
+ "epoch": 25.40599455040872,
+ "grad_norm": 5.9662017822265625,
+ "learning_rate": 1.7480514101329808e-05,
+ "loss": 0.4409,
+ "step": 9324
+ },
+ {
+ "epoch": 25.408719346049047,
+ "grad_norm": 5.690085411071777,
+ "learning_rate": 1.7479928412771253e-05,
+ "loss": 0.4428,
+ "step": 9325
+ },
+ {
+ "epoch": 25.411444141689373,
+ "grad_norm": 5.979255199432373,
+ "learning_rate": 1.7479342665959254e-05,
+ "loss": 0.3799,
+ "step": 9326
+ },
+ {
+ "epoch": 25.414168937329702,
+ "grad_norm": 5.331563949584961,
+ "learning_rate": 1.7478756860898378e-05,
+ "loss": 0.4063,
+ "step": 9327
+ },
+ {
+ "epoch": 25.416893732970028,
+ "grad_norm": 5.551055908203125,
+ "learning_rate": 1.747817099759318e-05,
+ "loss": 0.3455,
+ "step": 9328
+ },
+ {
+ "epoch": 25.419618528610354,
+ "grad_norm": 7.793542861938477,
+ "learning_rate": 1.747758507604823e-05,
+ "loss": 0.4305,
+ "step": 9329
+ },
+ {
+ "epoch": 25.422343324250683,
+ "grad_norm": 7.135063171386719,
+ "learning_rate": 1.7476999096268086e-05,
+ "loss": 0.2703,
+ "step": 9330
+ },
+ {
+ "epoch": 25.42506811989101,
+ "grad_norm": 6.508609294891357,
+ "learning_rate": 1.7476413058257313e-05,
+ "loss": 0.4518,
+ "step": 9331
+ },
+ {
+ "epoch": 25.427792915531334,
+ "grad_norm": 5.494690895080566,
+ "learning_rate": 1.7475826962020476e-05,
+ "loss": 0.4485,
+ "step": 9332
+ },
+ {
+ "epoch": 25.430517711171664,
+ "grad_norm": 6.774230003356934,
+ "learning_rate": 1.747524080756214e-05,
+ "loss": 0.5831,
+ "step": 9333
+ },
+ {
+ "epoch": 25.43324250681199,
+ "grad_norm": 5.717245101928711,
+ "learning_rate": 1.7474654594886873e-05,
+ "loss": 0.3007,
+ "step": 9334
+ },
+ {
+ "epoch": 25.435967302452315,
+ "grad_norm": 5.920994758605957,
+ "learning_rate": 1.7474068323999232e-05,
+ "loss": 0.3448,
+ "step": 9335
+ },
+ {
+ "epoch": 25.438692098092645,
+ "grad_norm": 5.4181623458862305,
+ "learning_rate": 1.7473481994903788e-05,
+ "loss": 0.3691,
+ "step": 9336
+ },
+ {
+ "epoch": 25.44141689373297,
+ "grad_norm": 6.997366905212402,
+ "learning_rate": 1.747289560760511e-05,
+ "loss": 0.4197,
+ "step": 9337
+ },
+ {
+ "epoch": 25.444141689373296,
+ "grad_norm": 14.668591499328613,
+ "learning_rate": 1.7472309162107757e-05,
+ "loss": 0.3518,
+ "step": 9338
+ },
+ {
+ "epoch": 25.446866485013626,
+ "grad_norm": 5.282878875732422,
+ "learning_rate": 1.7471722658416302e-05,
+ "loss": 0.2891,
+ "step": 9339
+ },
+ {
+ "epoch": 25.44959128065395,
+ "grad_norm": 5.814706802368164,
+ "learning_rate": 1.7471136096535316e-05,
+ "loss": 0.5331,
+ "step": 9340
+ },
+ {
+ "epoch": 25.452316076294277,
+ "grad_norm": 6.879452228546143,
+ "learning_rate": 1.7470549476469357e-05,
+ "loss": 0.4375,
+ "step": 9341
+ },
+ {
+ "epoch": 25.455040871934607,
+ "grad_norm": 6.617806911468506,
+ "learning_rate": 1.7469962798223e-05,
+ "loss": 0.5512,
+ "step": 9342
+ },
+ {
+ "epoch": 25.457765667574932,
+ "grad_norm": 5.68919038772583,
+ "learning_rate": 1.7469376061800815e-05,
+ "loss": 0.3652,
+ "step": 9343
+ },
+ {
+ "epoch": 25.460490463215258,
+ "grad_norm": 6.381850719451904,
+ "learning_rate": 1.746878926720737e-05,
+ "loss": 0.4889,
+ "step": 9344
+ },
+ {
+ "epoch": 25.463215258855588,
+ "grad_norm": 6.160511493682861,
+ "learning_rate": 1.746820241444723e-05,
+ "loss": 0.3833,
+ "step": 9345
+ },
+ {
+ "epoch": 25.465940054495913,
+ "grad_norm": 6.978784084320068,
+ "learning_rate": 1.7467615503524973e-05,
+ "loss": 0.4503,
+ "step": 9346
+ },
+ {
+ "epoch": 25.46866485013624,
+ "grad_norm": 6.207093715667725,
+ "learning_rate": 1.7467028534445166e-05,
+ "loss": 0.3815,
+ "step": 9347
+ },
+ {
+ "epoch": 25.47138964577657,
+ "grad_norm": 4.6039628982543945,
+ "learning_rate": 1.7466441507212383e-05,
+ "loss": 0.2935,
+ "step": 9348
+ },
+ {
+ "epoch": 25.474114441416894,
+ "grad_norm": 5.727050304412842,
+ "learning_rate": 1.7465854421831192e-05,
+ "loss": 0.4077,
+ "step": 9349
+ },
+ {
+ "epoch": 25.47683923705722,
+ "grad_norm": 7.455406188964844,
+ "learning_rate": 1.746526727830617e-05,
+ "loss": 0.4187,
+ "step": 9350
+ },
+ {
+ "epoch": 25.479564032697546,
+ "grad_norm": 6.522113800048828,
+ "learning_rate": 1.7464680076641883e-05,
+ "loss": 0.475,
+ "step": 9351
+ },
+ {
+ "epoch": 25.482288828337875,
+ "grad_norm": 5.544370174407959,
+ "learning_rate": 1.746409281684291e-05,
+ "loss": 0.4795,
+ "step": 9352
+ },
+ {
+ "epoch": 25.4850136239782,
+ "grad_norm": 5.678996562957764,
+ "learning_rate": 1.746350549891382e-05,
+ "loss": 0.3318,
+ "step": 9353
+ },
+ {
+ "epoch": 25.48773841961853,
+ "grad_norm": 5.848020076751709,
+ "learning_rate": 1.7462918122859193e-05,
+ "loss": 0.4023,
+ "step": 9354
+ },
+ {
+ "epoch": 25.490463215258856,
+ "grad_norm": 6.031424045562744,
+ "learning_rate": 1.7462330688683596e-05,
+ "loss": 0.4885,
+ "step": 9355
+ },
+ {
+ "epoch": 25.493188010899182,
+ "grad_norm": 4.1672892570495605,
+ "learning_rate": 1.7461743196391613e-05,
+ "loss": 0.6375,
+ "step": 9356
+ },
+ {
+ "epoch": 25.495912806539508,
+ "grad_norm": 5.704432487487793,
+ "learning_rate": 1.746115564598781e-05,
+ "loss": 0.356,
+ "step": 9357
+ },
+ {
+ "epoch": 25.498637602179837,
+ "grad_norm": 5.651536464691162,
+ "learning_rate": 1.746056803747677e-05,
+ "loss": 0.4055,
+ "step": 9358
+ },
+ {
+ "epoch": 25.501362397820163,
+ "grad_norm": 9.146065711975098,
+ "learning_rate": 1.7459980370863063e-05,
+ "loss": 0.5352,
+ "step": 9359
+ },
+ {
+ "epoch": 25.504087193460492,
+ "grad_norm": 4.399184226989746,
+ "learning_rate": 1.7459392646151266e-05,
+ "loss": 0.4147,
+ "step": 9360
+ },
+ {
+ "epoch": 25.506811989100818,
+ "grad_norm": 5.44091796875,
+ "learning_rate": 1.7458804863345966e-05,
+ "loss": 0.3826,
+ "step": 9361
+ },
+ {
+ "epoch": 25.509536784741144,
+ "grad_norm": 6.111367702484131,
+ "learning_rate": 1.745821702245173e-05,
+ "loss": 0.3342,
+ "step": 9362
+ },
+ {
+ "epoch": 25.51226158038147,
+ "grad_norm": 6.707502841949463,
+ "learning_rate": 1.7457629123473144e-05,
+ "loss": 0.3901,
+ "step": 9363
+ },
+ {
+ "epoch": 25.5149863760218,
+ "grad_norm": 7.246630668640137,
+ "learning_rate": 1.745704116641478e-05,
+ "loss": 0.4018,
+ "step": 9364
+ },
+ {
+ "epoch": 25.517711171662125,
+ "grad_norm": 6.3893561363220215,
+ "learning_rate": 1.7456453151281215e-05,
+ "loss": 0.3481,
+ "step": 9365
+ },
+ {
+ "epoch": 25.520435967302454,
+ "grad_norm": 5.618650913238525,
+ "learning_rate": 1.7455865078077038e-05,
+ "loss": 0.4632,
+ "step": 9366
+ },
+ {
+ "epoch": 25.52316076294278,
+ "grad_norm": 5.5011820793151855,
+ "learning_rate": 1.745527694680682e-05,
+ "loss": 0.4782,
+ "step": 9367
+ },
+ {
+ "epoch": 25.525885558583106,
+ "grad_norm": 5.666864395141602,
+ "learning_rate": 1.745468875747515e-05,
+ "loss": 0.3772,
+ "step": 9368
+ },
+ {
+ "epoch": 25.52861035422343,
+ "grad_norm": 6.579024791717529,
+ "learning_rate": 1.74541005100866e-05,
+ "loss": 0.4249,
+ "step": 9369
+ },
+ {
+ "epoch": 25.53133514986376,
+ "grad_norm": 6.811323642730713,
+ "learning_rate": 1.7453512204645756e-05,
+ "loss": 0.4998,
+ "step": 9370
+ },
+ {
+ "epoch": 25.534059945504087,
+ "grad_norm": 5.246577739715576,
+ "learning_rate": 1.7452923841157198e-05,
+ "loss": 0.3117,
+ "step": 9371
+ },
+ {
+ "epoch": 25.536784741144416,
+ "grad_norm": 5.61734676361084,
+ "learning_rate": 1.7452335419625514e-05,
+ "loss": 0.4222,
+ "step": 9372
+ },
+ {
+ "epoch": 25.539509536784742,
+ "grad_norm": 5.823202133178711,
+ "learning_rate": 1.7451746940055278e-05,
+ "loss": 0.3732,
+ "step": 9373
+ },
+ {
+ "epoch": 25.542234332425068,
+ "grad_norm": 5.525468826293945,
+ "learning_rate": 1.7451158402451078e-05,
+ "loss": 0.3474,
+ "step": 9374
+ },
+ {
+ "epoch": 25.544959128065393,
+ "grad_norm": 7.083858966827393,
+ "learning_rate": 1.7450569806817494e-05,
+ "loss": 0.4326,
+ "step": 9375
+ },
+ {
+ "epoch": 25.547683923705723,
+ "grad_norm": 7.645444393157959,
+ "learning_rate": 1.744998115315911e-05,
+ "loss": 0.5322,
+ "step": 9376
+ },
+ {
+ "epoch": 25.55040871934605,
+ "grad_norm": 5.670146465301514,
+ "learning_rate": 1.744939244148052e-05,
+ "loss": 0.3339,
+ "step": 9377
+ },
+ {
+ "epoch": 25.553133514986374,
+ "grad_norm": 6.320120334625244,
+ "learning_rate": 1.7448803671786292e-05,
+ "loss": 0.5113,
+ "step": 9378
+ },
+ {
+ "epoch": 25.555858310626704,
+ "grad_norm": 5.7277960777282715,
+ "learning_rate": 1.7448214844081026e-05,
+ "loss": 0.4041,
+ "step": 9379
+ },
+ {
+ "epoch": 25.55858310626703,
+ "grad_norm": 8.52052116394043,
+ "learning_rate": 1.7447625958369303e-05,
+ "loss": 0.3917,
+ "step": 9380
+ },
+ {
+ "epoch": 25.561307901907355,
+ "grad_norm": 5.95053243637085,
+ "learning_rate": 1.744703701465571e-05,
+ "loss": 0.3638,
+ "step": 9381
+ },
+ {
+ "epoch": 25.564032697547685,
+ "grad_norm": 5.211405277252197,
+ "learning_rate": 1.744644801294483e-05,
+ "loss": 0.4354,
+ "step": 9382
+ },
+ {
+ "epoch": 25.56675749318801,
+ "grad_norm": 6.374545574188232,
+ "learning_rate": 1.744585895324125e-05,
+ "loss": 0.3098,
+ "step": 9383
+ },
+ {
+ "epoch": 25.569482288828336,
+ "grad_norm": 4.946403980255127,
+ "learning_rate": 1.7445269835549565e-05,
+ "loss": 0.5058,
+ "step": 9384
+ },
+ {
+ "epoch": 25.572207084468666,
+ "grad_norm": 6.0899529457092285,
+ "learning_rate": 1.7444680659874354e-05,
+ "loss": 0.3073,
+ "step": 9385
+ },
+ {
+ "epoch": 25.57493188010899,
+ "grad_norm": 6.078880786895752,
+ "learning_rate": 1.744409142622021e-05,
+ "loss": 0.375,
+ "step": 9386
+ },
+ {
+ "epoch": 25.577656675749317,
+ "grad_norm": 6.38592004776001,
+ "learning_rate": 1.7443502134591725e-05,
+ "loss": 0.3476,
+ "step": 9387
+ },
+ {
+ "epoch": 25.580381471389646,
+ "grad_norm": 5.988442420959473,
+ "learning_rate": 1.744291278499348e-05,
+ "loss": 0.4496,
+ "step": 9388
+ },
+ {
+ "epoch": 25.583106267029972,
+ "grad_norm": 5.1993021965026855,
+ "learning_rate": 1.744232337743008e-05,
+ "loss": 0.4206,
+ "step": 9389
+ },
+ {
+ "epoch": 25.585831062670298,
+ "grad_norm": 5.966709136962891,
+ "learning_rate": 1.7441733911906092e-05,
+ "loss": 0.3679,
+ "step": 9390
+ },
+ {
+ "epoch": 25.588555858310627,
+ "grad_norm": 5.602008819580078,
+ "learning_rate": 1.7441144388426126e-05,
+ "loss": 0.3961,
+ "step": 9391
+ },
+ {
+ "epoch": 25.591280653950953,
+ "grad_norm": 8.632800102233887,
+ "learning_rate": 1.7440554806994768e-05,
+ "loss": 0.3805,
+ "step": 9392
+ },
+ {
+ "epoch": 25.59400544959128,
+ "grad_norm": 6.797333240509033,
+ "learning_rate": 1.743996516761661e-05,
+ "loss": 0.3655,
+ "step": 9393
+ },
+ {
+ "epoch": 25.59673024523161,
+ "grad_norm": 17.15534019470215,
+ "learning_rate": 1.743937547029624e-05,
+ "loss": 0.5673,
+ "step": 9394
+ },
+ {
+ "epoch": 25.599455040871934,
+ "grad_norm": 5.834554195404053,
+ "learning_rate": 1.7438785715038253e-05,
+ "loss": 0.3483,
+ "step": 9395
+ },
+ {
+ "epoch": 25.60217983651226,
+ "grad_norm": 6.06937313079834,
+ "learning_rate": 1.7438195901847243e-05,
+ "loss": 0.4414,
+ "step": 9396
+ },
+ {
+ "epoch": 25.60490463215259,
+ "grad_norm": 6.17116641998291,
+ "learning_rate": 1.7437606030727804e-05,
+ "loss": 0.4452,
+ "step": 9397
+ },
+ {
+ "epoch": 25.607629427792915,
+ "grad_norm": 4.926543712615967,
+ "learning_rate": 1.7437016101684526e-05,
+ "loss": 0.3515,
+ "step": 9398
+ },
+ {
+ "epoch": 25.61035422343324,
+ "grad_norm": 5.402313709259033,
+ "learning_rate": 1.7436426114722006e-05,
+ "loss": 0.4082,
+ "step": 9399
+ },
+ {
+ "epoch": 25.61307901907357,
+ "grad_norm": 6.385349273681641,
+ "learning_rate": 1.7435836069844838e-05,
+ "loss": 0.4534,
+ "step": 9400
+ },
+ {
+ "epoch": 25.615803814713896,
+ "grad_norm": 8.790868759155273,
+ "learning_rate": 1.7435245967057623e-05,
+ "loss": 0.6467,
+ "step": 9401
+ },
+ {
+ "epoch": 25.618528610354222,
+ "grad_norm": 5.587525367736816,
+ "learning_rate": 1.7434655806364945e-05,
+ "loss": 0.3253,
+ "step": 9402
+ },
+ {
+ "epoch": 25.62125340599455,
+ "grad_norm": 6.404189109802246,
+ "learning_rate": 1.7434065587771413e-05,
+ "loss": 0.6073,
+ "step": 9403
+ },
+ {
+ "epoch": 25.623978201634877,
+ "grad_norm": 6.260900020599365,
+ "learning_rate": 1.7433475311281616e-05,
+ "loss": 0.4753,
+ "step": 9404
+ },
+ {
+ "epoch": 25.626702997275203,
+ "grad_norm": 5.485964775085449,
+ "learning_rate": 1.7432884976900147e-05,
+ "loss": 0.3904,
+ "step": 9405
+ },
+ {
+ "epoch": 25.629427792915532,
+ "grad_norm": 5.611900806427002,
+ "learning_rate": 1.7432294584631616e-05,
+ "loss": 0.3928,
+ "step": 9406
+ },
+ {
+ "epoch": 25.632152588555858,
+ "grad_norm": 6.608458042144775,
+ "learning_rate": 1.7431704134480608e-05,
+ "loss": 0.4416,
+ "step": 9407
+ },
+ {
+ "epoch": 25.634877384196184,
+ "grad_norm": 5.258899688720703,
+ "learning_rate": 1.7431113626451733e-05,
+ "loss": 0.299,
+ "step": 9408
+ },
+ {
+ "epoch": 25.637602179836513,
+ "grad_norm": 6.611827850341797,
+ "learning_rate": 1.7430523060549586e-05,
+ "loss": 0.3568,
+ "step": 9409
+ },
+ {
+ "epoch": 25.64032697547684,
+ "grad_norm": 6.170421600341797,
+ "learning_rate": 1.7429932436778755e-05,
+ "loss": 0.4912,
+ "step": 9410
+ },
+ {
+ "epoch": 25.643051771117165,
+ "grad_norm": 5.806344985961914,
+ "learning_rate": 1.7429341755143857e-05,
+ "loss": 0.3315,
+ "step": 9411
+ },
+ {
+ "epoch": 25.645776566757494,
+ "grad_norm": 5.893872261047363,
+ "learning_rate": 1.7428751015649483e-05,
+ "loss": 0.5998,
+ "step": 9412
+ },
+ {
+ "epoch": 25.64850136239782,
+ "grad_norm": 6.419220447540283,
+ "learning_rate": 1.7428160218300236e-05,
+ "loss": 0.4603,
+ "step": 9413
+ },
+ {
+ "epoch": 25.651226158038146,
+ "grad_norm": 4.974557399749756,
+ "learning_rate": 1.7427569363100714e-05,
+ "loss": 0.432,
+ "step": 9414
+ },
+ {
+ "epoch": 25.653950953678475,
+ "grad_norm": 4.8727850914001465,
+ "learning_rate": 1.742697845005552e-05,
+ "loss": 0.4541,
+ "step": 9415
+ },
+ {
+ "epoch": 25.6566757493188,
+ "grad_norm": 5.570352554321289,
+ "learning_rate": 1.7426387479169256e-05,
+ "loss": 0.3606,
+ "step": 9416
+ },
+ {
+ "epoch": 25.659400544959126,
+ "grad_norm": 8.528862953186035,
+ "learning_rate": 1.7425796450446532e-05,
+ "loss": 0.411,
+ "step": 9417
+ },
+ {
+ "epoch": 25.662125340599456,
+ "grad_norm": 5.671764850616455,
+ "learning_rate": 1.7425205363891938e-05,
+ "loss": 0.4756,
+ "step": 9418
+ },
+ {
+ "epoch": 25.66485013623978,
+ "grad_norm": 5.788801670074463,
+ "learning_rate": 1.7424614219510087e-05,
+ "loss": 0.6254,
+ "step": 9419
+ },
+ {
+ "epoch": 25.667574931880107,
+ "grad_norm": 5.057888031005859,
+ "learning_rate": 1.742402301730558e-05,
+ "loss": 0.4305,
+ "step": 9420
+ },
+ {
+ "epoch": 25.670299727520437,
+ "grad_norm": 4.472919940948486,
+ "learning_rate": 1.7423431757283018e-05,
+ "loss": 0.3138,
+ "step": 9421
+ },
+ {
+ "epoch": 25.673024523160763,
+ "grad_norm": 6.632477760314941,
+ "learning_rate": 1.742284043944701e-05,
+ "loss": 0.3751,
+ "step": 9422
+ },
+ {
+ "epoch": 25.67574931880109,
+ "grad_norm": 5.274984359741211,
+ "learning_rate": 1.742224906380216e-05,
+ "loss": 0.303,
+ "step": 9423
+ },
+ {
+ "epoch": 25.678474114441418,
+ "grad_norm": 31.579347610473633,
+ "learning_rate": 1.7421657630353068e-05,
+ "loss": 0.3925,
+ "step": 9424
+ },
+ {
+ "epoch": 25.681198910081743,
+ "grad_norm": 5.024258136749268,
+ "learning_rate": 1.742106613910435e-05,
+ "loss": 0.406,
+ "step": 9425
+ },
+ {
+ "epoch": 25.68392370572207,
+ "grad_norm": 6.168879508972168,
+ "learning_rate": 1.7420474590060606e-05,
+ "loss": 0.3506,
+ "step": 9426
+ },
+ {
+ "epoch": 25.6866485013624,
+ "grad_norm": 5.0484724044799805,
+ "learning_rate": 1.7419882983226446e-05,
+ "loss": 0.4326,
+ "step": 9427
+ },
+ {
+ "epoch": 25.689373297002724,
+ "grad_norm": 5.8042755126953125,
+ "learning_rate": 1.7419291318606476e-05,
+ "loss": 0.4946,
+ "step": 9428
+ },
+ {
+ "epoch": 25.69209809264305,
+ "grad_norm": 5.766669750213623,
+ "learning_rate": 1.7418699596205303e-05,
+ "loss": 0.3895,
+ "step": 9429
+ },
+ {
+ "epoch": 25.69482288828338,
+ "grad_norm": 5.623227119445801,
+ "learning_rate": 1.7418107816027538e-05,
+ "loss": 0.3716,
+ "step": 9430
+ },
+ {
+ "epoch": 25.697547683923705,
+ "grad_norm": 4.734564781188965,
+ "learning_rate": 1.741751597807779e-05,
+ "loss": 0.4386,
+ "step": 9431
+ },
+ {
+ "epoch": 25.70027247956403,
+ "grad_norm": 5.973719120025635,
+ "learning_rate": 1.741692408236066e-05,
+ "loss": 0.3394,
+ "step": 9432
+ },
+ {
+ "epoch": 25.70299727520436,
+ "grad_norm": 4.706031322479248,
+ "learning_rate": 1.7416332128880768e-05,
+ "loss": 0.3977,
+ "step": 9433
+ },
+ {
+ "epoch": 25.705722070844686,
+ "grad_norm": 6.022923469543457,
+ "learning_rate": 1.7415740117642718e-05,
+ "loss": 0.44,
+ "step": 9434
+ },
+ {
+ "epoch": 25.708446866485012,
+ "grad_norm": 5.815680027008057,
+ "learning_rate": 1.7415148048651126e-05,
+ "loss": 0.362,
+ "step": 9435
+ },
+ {
+ "epoch": 25.71117166212534,
+ "grad_norm": 5.911482334136963,
+ "learning_rate": 1.7414555921910597e-05,
+ "loss": 0.5844,
+ "step": 9436
+ },
+ {
+ "epoch": 25.713896457765667,
+ "grad_norm": 5.9180426597595215,
+ "learning_rate": 1.7413963737425747e-05,
+ "loss": 0.3764,
+ "step": 9437
+ },
+ {
+ "epoch": 25.716621253405993,
+ "grad_norm": 5.470936298370361,
+ "learning_rate": 1.7413371495201183e-05,
+ "loss": 0.3371,
+ "step": 9438
+ },
+ {
+ "epoch": 25.719346049046322,
+ "grad_norm": 6.977021217346191,
+ "learning_rate": 1.7412779195241524e-05,
+ "loss": 0.8017,
+ "step": 9439
+ },
+ {
+ "epoch": 25.722070844686648,
+ "grad_norm": 5.997401237487793,
+ "learning_rate": 1.741218683755138e-05,
+ "loss": 0.3568,
+ "step": 9440
+ },
+ {
+ "epoch": 25.724795640326974,
+ "grad_norm": 6.173062801361084,
+ "learning_rate": 1.7411594422135357e-05,
+ "loss": 0.3252,
+ "step": 9441
+ },
+ {
+ "epoch": 25.727520435967303,
+ "grad_norm": 5.288327217102051,
+ "learning_rate": 1.741100194899808e-05,
+ "loss": 0.454,
+ "step": 9442
+ },
+ {
+ "epoch": 25.73024523160763,
+ "grad_norm": 6.00217866897583,
+ "learning_rate": 1.741040941814416e-05,
+ "loss": 0.4152,
+ "step": 9443
+ },
+ {
+ "epoch": 25.732970027247955,
+ "grad_norm": 6.386344909667969,
+ "learning_rate": 1.7409816829578205e-05,
+ "loss": 0.5486,
+ "step": 9444
+ },
+ {
+ "epoch": 25.735694822888284,
+ "grad_norm": 5.891502380371094,
+ "learning_rate": 1.7409224183304836e-05,
+ "loss": 0.3581,
+ "step": 9445
+ },
+ {
+ "epoch": 25.73841961852861,
+ "grad_norm": 6.704747200012207,
+ "learning_rate": 1.740863147932867e-05,
+ "loss": 0.3387,
+ "step": 9446
+ },
+ {
+ "epoch": 25.741144414168936,
+ "grad_norm": 5.751468181610107,
+ "learning_rate": 1.740803871765432e-05,
+ "loss": 0.5156,
+ "step": 9447
+ },
+ {
+ "epoch": 25.743869209809265,
+ "grad_norm": 7.555211544036865,
+ "learning_rate": 1.7407445898286405e-05,
+ "loss": 0.3697,
+ "step": 9448
+ },
+ {
+ "epoch": 25.74659400544959,
+ "grad_norm": 5.714015483856201,
+ "learning_rate": 1.7406853021229535e-05,
+ "loss": 0.5378,
+ "step": 9449
+ },
+ {
+ "epoch": 25.749318801089917,
+ "grad_norm": 5.962901592254639,
+ "learning_rate": 1.7406260086488336e-05,
+ "loss": 0.4113,
+ "step": 9450
+ },
+ {
+ "epoch": 25.752043596730246,
+ "grad_norm": 5.170448303222656,
+ "learning_rate": 1.740566709406742e-05,
+ "loss": 0.5316,
+ "step": 9451
+ },
+ {
+ "epoch": 25.754768392370572,
+ "grad_norm": 5.6077399253845215,
+ "learning_rate": 1.7405074043971407e-05,
+ "loss": 0.5199,
+ "step": 9452
+ },
+ {
+ "epoch": 25.757493188010898,
+ "grad_norm": 6.204434871673584,
+ "learning_rate": 1.7404480936204914e-05,
+ "loss": 0.5255,
+ "step": 9453
+ },
+ {
+ "epoch": 25.760217983651227,
+ "grad_norm": 6.042555332183838,
+ "learning_rate": 1.7403887770772564e-05,
+ "loss": 0.3425,
+ "step": 9454
+ },
+ {
+ "epoch": 25.762942779291553,
+ "grad_norm": 5.335410118103027,
+ "learning_rate": 1.7403294547678975e-05,
+ "loss": 0.5235,
+ "step": 9455
+ },
+ {
+ "epoch": 25.76566757493188,
+ "grad_norm": 5.661258220672607,
+ "learning_rate": 1.7402701266928767e-05,
+ "loss": 0.3767,
+ "step": 9456
+ },
+ {
+ "epoch": 25.768392370572208,
+ "grad_norm": 5.9232354164123535,
+ "learning_rate": 1.7402107928526556e-05,
+ "loss": 0.3528,
+ "step": 9457
+ },
+ {
+ "epoch": 25.771117166212534,
+ "grad_norm": 5.13308572769165,
+ "learning_rate": 1.740151453247697e-05,
+ "loss": 0.327,
+ "step": 9458
+ },
+ {
+ "epoch": 25.77384196185286,
+ "grad_norm": 5.199887275695801,
+ "learning_rate": 1.7400921078784624e-05,
+ "loss": 0.4482,
+ "step": 9459
+ },
+ {
+ "epoch": 25.77656675749319,
+ "grad_norm": 5.926246643066406,
+ "learning_rate": 1.7400327567454145e-05,
+ "loss": 0.3155,
+ "step": 9460
+ },
+ {
+ "epoch": 25.779291553133515,
+ "grad_norm": 4.858626842498779,
+ "learning_rate": 1.7399733998490152e-05,
+ "loss": 0.4713,
+ "step": 9461
+ },
+ {
+ "epoch": 25.78201634877384,
+ "grad_norm": 6.024007797241211,
+ "learning_rate": 1.739914037189727e-05,
+ "loss": 0.4859,
+ "step": 9462
+ },
+ {
+ "epoch": 25.78474114441417,
+ "grad_norm": 4.670284271240234,
+ "learning_rate": 1.739854668768012e-05,
+ "loss": 0.6274,
+ "step": 9463
+ },
+ {
+ "epoch": 25.787465940054496,
+ "grad_norm": 6.490248680114746,
+ "learning_rate": 1.7397952945843325e-05,
+ "loss": 0.4872,
+ "step": 9464
+ },
+ {
+ "epoch": 25.79019073569482,
+ "grad_norm": 5.181295871734619,
+ "learning_rate": 1.7397359146391513e-05,
+ "loss": 0.4471,
+ "step": 9465
+ },
+ {
+ "epoch": 25.79291553133515,
+ "grad_norm": 5.302846431732178,
+ "learning_rate": 1.7396765289329308e-05,
+ "loss": 0.4028,
+ "step": 9466
+ },
+ {
+ "epoch": 25.795640326975477,
+ "grad_norm": 5.8060784339904785,
+ "learning_rate": 1.739617137466133e-05,
+ "loss": 0.4236,
+ "step": 9467
+ },
+ {
+ "epoch": 25.798365122615802,
+ "grad_norm": 5.637409210205078,
+ "learning_rate": 1.7395577402392208e-05,
+ "loss": 0.4417,
+ "step": 9468
+ },
+ {
+ "epoch": 25.80108991825613,
+ "grad_norm": 5.857173919677734,
+ "learning_rate": 1.739498337252657e-05,
+ "loss": 0.4736,
+ "step": 9469
+ },
+ {
+ "epoch": 25.803814713896458,
+ "grad_norm": 6.2213592529296875,
+ "learning_rate": 1.7394389285069034e-05,
+ "loss": 0.6266,
+ "step": 9470
+ },
+ {
+ "epoch": 25.806539509536783,
+ "grad_norm": 8.066509246826172,
+ "learning_rate": 1.7393795140024235e-05,
+ "loss": 0.6044,
+ "step": 9471
+ },
+ {
+ "epoch": 25.809264305177113,
+ "grad_norm": 5.900491714477539,
+ "learning_rate": 1.7393200937396795e-05,
+ "loss": 0.4434,
+ "step": 9472
+ },
+ {
+ "epoch": 25.81198910081744,
+ "grad_norm": 5.15139102935791,
+ "learning_rate": 1.7392606677191345e-05,
+ "loss": 0.3926,
+ "step": 9473
+ },
+ {
+ "epoch": 25.814713896457764,
+ "grad_norm": 6.155881881713867,
+ "learning_rate": 1.7392012359412517e-05,
+ "loss": 0.3125,
+ "step": 9474
+ },
+ {
+ "epoch": 25.817438692098094,
+ "grad_norm": 6.071492671966553,
+ "learning_rate": 1.7391417984064932e-05,
+ "loss": 0.5966,
+ "step": 9475
+ },
+ {
+ "epoch": 25.82016348773842,
+ "grad_norm": 5.230311870574951,
+ "learning_rate": 1.7390823551153218e-05,
+ "loss": 0.3412,
+ "step": 9476
+ },
+ {
+ "epoch": 25.822888283378745,
+ "grad_norm": 5.747670650482178,
+ "learning_rate": 1.739022906068201e-05,
+ "loss": 0.463,
+ "step": 9477
+ },
+ {
+ "epoch": 25.825613079019075,
+ "grad_norm": 5.632741928100586,
+ "learning_rate": 1.738963451265594e-05,
+ "loss": 0.4456,
+ "step": 9478
+ },
+ {
+ "epoch": 25.8283378746594,
+ "grad_norm": 6.210419178009033,
+ "learning_rate": 1.738903990707963e-05,
+ "loss": 0.3859,
+ "step": 9479
+ },
+ {
+ "epoch": 25.831062670299726,
+ "grad_norm": 8.28834342956543,
+ "learning_rate": 1.7388445243957717e-05,
+ "loss": 0.3901,
+ "step": 9480
+ },
+ {
+ "epoch": 25.833787465940055,
+ "grad_norm": 5.508493900299072,
+ "learning_rate": 1.7387850523294827e-05,
+ "loss": 0.2978,
+ "step": 9481
+ },
+ {
+ "epoch": 25.83651226158038,
+ "grad_norm": 7.931019306182861,
+ "learning_rate": 1.7387255745095597e-05,
+ "loss": 0.2922,
+ "step": 9482
+ },
+ {
+ "epoch": 25.839237057220707,
+ "grad_norm": 5.028665542602539,
+ "learning_rate": 1.738666090936466e-05,
+ "loss": 0.3886,
+ "step": 9483
+ },
+ {
+ "epoch": 25.841961852861036,
+ "grad_norm": 4.650830268859863,
+ "learning_rate": 1.738606601610664e-05,
+ "loss": 0.3497,
+ "step": 9484
+ },
+ {
+ "epoch": 25.844686648501362,
+ "grad_norm": 4.689539909362793,
+ "learning_rate": 1.738547106532618e-05,
+ "loss": 0.2878,
+ "step": 9485
+ },
+ {
+ "epoch": 25.847411444141688,
+ "grad_norm": 9.65332317352295,
+ "learning_rate": 1.7384876057027907e-05,
+ "loss": 0.4294,
+ "step": 9486
+ },
+ {
+ "epoch": 25.850136239782017,
+ "grad_norm": 5.933709144592285,
+ "learning_rate": 1.7384280991216455e-05,
+ "loss": 0.3036,
+ "step": 9487
+ },
+ {
+ "epoch": 25.852861035422343,
+ "grad_norm": 5.7842302322387695,
+ "learning_rate": 1.7383685867896463e-05,
+ "loss": 0.5367,
+ "step": 9488
+ },
+ {
+ "epoch": 25.85558583106267,
+ "grad_norm": 4.849140167236328,
+ "learning_rate": 1.7383090687072563e-05,
+ "loss": 0.317,
+ "step": 9489
+ },
+ {
+ "epoch": 25.858310626703,
+ "grad_norm": 5.407693386077881,
+ "learning_rate": 1.738249544874939e-05,
+ "loss": 0.3636,
+ "step": 9490
+ },
+ {
+ "epoch": 25.861035422343324,
+ "grad_norm": 5.958927631378174,
+ "learning_rate": 1.7381900152931582e-05,
+ "loss": 0.5835,
+ "step": 9491
+ },
+ {
+ "epoch": 25.86376021798365,
+ "grad_norm": 6.185679912567139,
+ "learning_rate": 1.738130479962377e-05,
+ "loss": 0.3127,
+ "step": 9492
+ },
+ {
+ "epoch": 25.86648501362398,
+ "grad_norm": 7.101397514343262,
+ "learning_rate": 1.7380709388830595e-05,
+ "loss": 0.4488,
+ "step": 9493
+ },
+ {
+ "epoch": 25.869209809264305,
+ "grad_norm": 6.953850746154785,
+ "learning_rate": 1.738011392055669e-05,
+ "loss": 0.3703,
+ "step": 9494
+ },
+ {
+ "epoch": 25.87193460490463,
+ "grad_norm": 5.2850542068481445,
+ "learning_rate": 1.7379518394806703e-05,
+ "loss": 0.4586,
+ "step": 9495
+ },
+ {
+ "epoch": 25.87465940054496,
+ "grad_norm": 5.190978527069092,
+ "learning_rate": 1.7378922811585256e-05,
+ "loss": 0.6218,
+ "step": 9496
+ },
+ {
+ "epoch": 25.877384196185286,
+ "grad_norm": 6.343438148498535,
+ "learning_rate": 1.7378327170897e-05,
+ "loss": 0.4213,
+ "step": 9497
+ },
+ {
+ "epoch": 25.88010899182561,
+ "grad_norm": 4.018118381500244,
+ "learning_rate": 1.737773147274657e-05,
+ "loss": 0.2906,
+ "step": 9498
+ },
+ {
+ "epoch": 25.88283378746594,
+ "grad_norm": 7.9682745933532715,
+ "learning_rate": 1.73771357171386e-05,
+ "loss": 0.7003,
+ "step": 9499
+ },
+ {
+ "epoch": 25.885558583106267,
+ "grad_norm": 5.834094047546387,
+ "learning_rate": 1.7376539904077742e-05,
+ "loss": 0.4599,
+ "step": 9500
+ },
+ {
+ "epoch": 25.888283378746593,
+ "grad_norm": 6.32597541809082,
+ "learning_rate": 1.7375944033568625e-05,
+ "loss": 0.452,
+ "step": 9501
+ },
+ {
+ "epoch": 25.891008174386922,
+ "grad_norm": 5.8352837562561035,
+ "learning_rate": 1.737534810561589e-05,
+ "loss": 0.4821,
+ "step": 9502
+ },
+ {
+ "epoch": 25.893732970027248,
+ "grad_norm": 8.509065628051758,
+ "learning_rate": 1.7374752120224184e-05,
+ "loss": 0.5315,
+ "step": 9503
+ },
+ {
+ "epoch": 25.896457765667574,
+ "grad_norm": 5.6432037353515625,
+ "learning_rate": 1.7374156077398144e-05,
+ "loss": 0.5052,
+ "step": 9504
+ },
+ {
+ "epoch": 25.899182561307903,
+ "grad_norm": 6.692959785461426,
+ "learning_rate": 1.7373559977142416e-05,
+ "loss": 0.392,
+ "step": 9505
+ },
+ {
+ "epoch": 25.90190735694823,
+ "grad_norm": 5.042212009429932,
+ "learning_rate": 1.7372963819461642e-05,
+ "loss": 0.5379,
+ "step": 9506
+ },
+ {
+ "epoch": 25.904632152588555,
+ "grad_norm": 5.635720729827881,
+ "learning_rate": 1.737236760436046e-05,
+ "loss": 0.4675,
+ "step": 9507
+ },
+ {
+ "epoch": 25.907356948228884,
+ "grad_norm": 7.361470699310303,
+ "learning_rate": 1.7371771331843517e-05,
+ "loss": 0.4144,
+ "step": 9508
+ },
+ {
+ "epoch": 25.91008174386921,
+ "grad_norm": 6.393855094909668,
+ "learning_rate": 1.7371175001915455e-05,
+ "loss": 0.468,
+ "step": 9509
+ },
+ {
+ "epoch": 25.912806539509535,
+ "grad_norm": 6.087363243103027,
+ "learning_rate": 1.7370578614580922e-05,
+ "loss": 0.4271,
+ "step": 9510
+ },
+ {
+ "epoch": 25.915531335149865,
+ "grad_norm": 6.660183429718018,
+ "learning_rate": 1.736998216984456e-05,
+ "loss": 0.3884,
+ "step": 9511
+ },
+ {
+ "epoch": 25.91825613079019,
+ "grad_norm": 6.591582775115967,
+ "learning_rate": 1.7369385667711013e-05,
+ "loss": 0.5068,
+ "step": 9512
+ },
+ {
+ "epoch": 25.920980926430516,
+ "grad_norm": 4.747461795806885,
+ "learning_rate": 1.7368789108184927e-05,
+ "loss": 0.3432,
+ "step": 9513
+ },
+ {
+ "epoch": 25.923705722070846,
+ "grad_norm": 5.667545795440674,
+ "learning_rate": 1.7368192491270946e-05,
+ "loss": 0.3923,
+ "step": 9514
+ },
+ {
+ "epoch": 25.92643051771117,
+ "grad_norm": 6.585363864898682,
+ "learning_rate": 1.7367595816973722e-05,
+ "loss": 0.4223,
+ "step": 9515
+ },
+ {
+ "epoch": 25.929155313351497,
+ "grad_norm": 5.761697769165039,
+ "learning_rate": 1.73669990852979e-05,
+ "loss": 0.3336,
+ "step": 9516
+ },
+ {
+ "epoch": 25.931880108991827,
+ "grad_norm": 6.663090705871582,
+ "learning_rate": 1.7366402296248126e-05,
+ "loss": 0.472,
+ "step": 9517
+ },
+ {
+ "epoch": 25.934604904632153,
+ "grad_norm": 4.756765365600586,
+ "learning_rate": 1.7365805449829047e-05,
+ "loss": 0.5712,
+ "step": 9518
+ },
+ {
+ "epoch": 25.93732970027248,
+ "grad_norm": 5.738847255706787,
+ "learning_rate": 1.7365208546045315e-05,
+ "loss": 0.3881,
+ "step": 9519
+ },
+ {
+ "epoch": 25.940054495912808,
+ "grad_norm": 5.557242393493652,
+ "learning_rate": 1.736461158490157e-05,
+ "loss": 0.4065,
+ "step": 9520
+ },
+ {
+ "epoch": 25.942779291553133,
+ "grad_norm": 11.829328536987305,
+ "learning_rate": 1.7364014566402472e-05,
+ "loss": 0.4108,
+ "step": 9521
+ },
+ {
+ "epoch": 25.94550408719346,
+ "grad_norm": 5.53961706161499,
+ "learning_rate": 1.7363417490552664e-05,
+ "loss": 0.3463,
+ "step": 9522
+ },
+ {
+ "epoch": 25.94822888283379,
+ "grad_norm": 4.436404705047607,
+ "learning_rate": 1.7362820357356794e-05,
+ "loss": 0.3343,
+ "step": 9523
+ },
+ {
+ "epoch": 25.950953678474114,
+ "grad_norm": 5.3378705978393555,
+ "learning_rate": 1.736222316681952e-05,
+ "loss": 0.4808,
+ "step": 9524
+ },
+ {
+ "epoch": 25.95367847411444,
+ "grad_norm": 5.352426528930664,
+ "learning_rate": 1.736162591894549e-05,
+ "loss": 0.5475,
+ "step": 9525
+ },
+ {
+ "epoch": 25.95640326975477,
+ "grad_norm": 5.803027629852295,
+ "learning_rate": 1.7361028613739354e-05,
+ "loss": 0.3848,
+ "step": 9526
+ },
+ {
+ "epoch": 25.959128065395095,
+ "grad_norm": 5.39794921875,
+ "learning_rate": 1.7360431251205763e-05,
+ "loss": 0.4656,
+ "step": 9527
+ },
+ {
+ "epoch": 25.96185286103542,
+ "grad_norm": 5.063699245452881,
+ "learning_rate": 1.735983383134937e-05,
+ "loss": 0.5359,
+ "step": 9528
+ },
+ {
+ "epoch": 25.96457765667575,
+ "grad_norm": 4.978878021240234,
+ "learning_rate": 1.7359236354174826e-05,
+ "loss": 0.4773,
+ "step": 9529
+ },
+ {
+ "epoch": 25.967302452316076,
+ "grad_norm": 5.358898639678955,
+ "learning_rate": 1.7358638819686787e-05,
+ "loss": 0.3793,
+ "step": 9530
+ },
+ {
+ "epoch": 25.970027247956402,
+ "grad_norm": 4.96940279006958,
+ "learning_rate": 1.735804122788991e-05,
+ "loss": 0.4826,
+ "step": 9531
+ },
+ {
+ "epoch": 25.97275204359673,
+ "grad_norm": 4.9680633544921875,
+ "learning_rate": 1.735744357878884e-05,
+ "loss": 0.4406,
+ "step": 9532
+ },
+ {
+ "epoch": 25.975476839237057,
+ "grad_norm": 5.316905498504639,
+ "learning_rate": 1.7356845872388238e-05,
+ "loss": 0.6782,
+ "step": 9533
+ },
+ {
+ "epoch": 25.978201634877383,
+ "grad_norm": 6.721235752105713,
+ "learning_rate": 1.7356248108692758e-05,
+ "loss": 0.3407,
+ "step": 9534
+ },
+ {
+ "epoch": 25.980926430517712,
+ "grad_norm": 5.7405900955200195,
+ "learning_rate": 1.735565028770705e-05,
+ "loss": 0.504,
+ "step": 9535
+ },
+ {
+ "epoch": 25.983651226158038,
+ "grad_norm": 5.763979434967041,
+ "learning_rate": 1.735505240943578e-05,
+ "loss": 0.481,
+ "step": 9536
+ },
+ {
+ "epoch": 25.986376021798364,
+ "grad_norm": 5.437042713165283,
+ "learning_rate": 1.7354454473883596e-05,
+ "loss": 0.375,
+ "step": 9537
+ },
+ {
+ "epoch": 25.989100817438693,
+ "grad_norm": 5.930602073669434,
+ "learning_rate": 1.7353856481055157e-05,
+ "loss": 0.527,
+ "step": 9538
+ },
+ {
+ "epoch": 25.99182561307902,
+ "grad_norm": 9.323917388916016,
+ "learning_rate": 1.735325843095512e-05,
+ "loss": 0.3232,
+ "step": 9539
+ },
+ {
+ "epoch": 25.994550408719345,
+ "grad_norm": 5.943012237548828,
+ "learning_rate": 1.7352660323588146e-05,
+ "loss": 0.5914,
+ "step": 9540
+ },
+ {
+ "epoch": 25.997275204359674,
+ "grad_norm": 4.763924598693848,
+ "learning_rate": 1.7352062158958887e-05,
+ "loss": 0.3195,
+ "step": 9541
+ },
+ {
+ "epoch": 26.0,
+ "grad_norm": 5.877326011657715,
+ "learning_rate": 1.7351463937072008e-05,
+ "loss": 0.4057,
+ "step": 9542
+ },
+ {
+ "epoch": 26.002724795640326,
+ "grad_norm": 5.893112659454346,
+ "learning_rate": 1.735086565793216e-05,
+ "loss": 0.3365,
+ "step": 9543
+ },
+ {
+ "epoch": 26.005449591280655,
+ "grad_norm": 5.752626419067383,
+ "learning_rate": 1.735026732154401e-05,
+ "loss": 0.4935,
+ "step": 9544
+ },
+ {
+ "epoch": 26.00817438692098,
+ "grad_norm": 5.411423206329346,
+ "learning_rate": 1.734966892791221e-05,
+ "loss": 0.3409,
+ "step": 9545
+ },
+ {
+ "epoch": 26.010899182561307,
+ "grad_norm": 5.57521390914917,
+ "learning_rate": 1.734907047704143e-05,
+ "loss": 0.4855,
+ "step": 9546
+ },
+ {
+ "epoch": 26.013623978201636,
+ "grad_norm": 3.7991559505462646,
+ "learning_rate": 1.7348471968936325e-05,
+ "loss": 0.3265,
+ "step": 9547
+ },
+ {
+ "epoch": 26.016348773841962,
+ "grad_norm": 4.873567581176758,
+ "learning_rate": 1.7347873403601557e-05,
+ "loss": 0.3478,
+ "step": 9548
+ },
+ {
+ "epoch": 26.019073569482288,
+ "grad_norm": 4.438267707824707,
+ "learning_rate": 1.7347274781041784e-05,
+ "loss": 0.2972,
+ "step": 9549
+ },
+ {
+ "epoch": 26.021798365122617,
+ "grad_norm": 4.858517646789551,
+ "learning_rate": 1.7346676101261676e-05,
+ "loss": 0.2517,
+ "step": 9550
+ },
+ {
+ "epoch": 26.024523160762943,
+ "grad_norm": 6.016982555389404,
+ "learning_rate": 1.734607736426589e-05,
+ "loss": 0.3412,
+ "step": 9551
+ },
+ {
+ "epoch": 26.02724795640327,
+ "grad_norm": 4.886334419250488,
+ "learning_rate": 1.7345478570059084e-05,
+ "loss": 0.3893,
+ "step": 9552
+ },
+ {
+ "epoch": 26.029972752043598,
+ "grad_norm": 4.867129325866699,
+ "learning_rate": 1.7344879718645934e-05,
+ "loss": 0.2368,
+ "step": 9553
+ },
+ {
+ "epoch": 26.032697547683924,
+ "grad_norm": 4.811803340911865,
+ "learning_rate": 1.7344280810031097e-05,
+ "loss": 0.4498,
+ "step": 9554
+ },
+ {
+ "epoch": 26.03542234332425,
+ "grad_norm": 5.041390419006348,
+ "learning_rate": 1.7343681844219232e-05,
+ "loss": 0.4561,
+ "step": 9555
+ },
+ {
+ "epoch": 26.03814713896458,
+ "grad_norm": 4.898367404937744,
+ "learning_rate": 1.7343082821215015e-05,
+ "loss": 0.3274,
+ "step": 9556
+ },
+ {
+ "epoch": 26.040871934604905,
+ "grad_norm": 5.4627580642700195,
+ "learning_rate": 1.73424837410231e-05,
+ "loss": 0.2953,
+ "step": 9557
+ },
+ {
+ "epoch": 26.04359673024523,
+ "grad_norm": 4.572257995605469,
+ "learning_rate": 1.7341884603648164e-05,
+ "loss": 0.3908,
+ "step": 9558
+ },
+ {
+ "epoch": 26.04632152588556,
+ "grad_norm": 4.909732818603516,
+ "learning_rate": 1.7341285409094862e-05,
+ "loss": 0.4703,
+ "step": 9559
+ },
+ {
+ "epoch": 26.049046321525886,
+ "grad_norm": 5.502349376678467,
+ "learning_rate": 1.7340686157367866e-05,
+ "loss": 0.2724,
+ "step": 9560
+ },
+ {
+ "epoch": 26.05177111716621,
+ "grad_norm": 6.280913829803467,
+ "learning_rate": 1.7340086848471848e-05,
+ "loss": 0.3684,
+ "step": 9561
+ },
+ {
+ "epoch": 26.05449591280654,
+ "grad_norm": 4.866971015930176,
+ "learning_rate": 1.7339487482411464e-05,
+ "loss": 0.3865,
+ "step": 9562
+ },
+ {
+ "epoch": 26.057220708446867,
+ "grad_norm": 5.4211883544921875,
+ "learning_rate": 1.733888805919139e-05,
+ "loss": 0.4964,
+ "step": 9563
+ },
+ {
+ "epoch": 26.059945504087192,
+ "grad_norm": 6.292038440704346,
+ "learning_rate": 1.7338288578816292e-05,
+ "loss": 0.423,
+ "step": 9564
+ },
+ {
+ "epoch": 26.06267029972752,
+ "grad_norm": 4.854907989501953,
+ "learning_rate": 1.733768904129084e-05,
+ "loss": 0.334,
+ "step": 9565
+ },
+ {
+ "epoch": 26.065395095367847,
+ "grad_norm": 5.8626909255981445,
+ "learning_rate": 1.7337089446619697e-05,
+ "loss": 0.4172,
+ "step": 9566
+ },
+ {
+ "epoch": 26.068119891008173,
+ "grad_norm": 4.90035343170166,
+ "learning_rate": 1.733648979480754e-05,
+ "loss": 0.3714,
+ "step": 9567
+ },
+ {
+ "epoch": 26.070844686648503,
+ "grad_norm": 4.515876293182373,
+ "learning_rate": 1.733589008585904e-05,
+ "loss": 0.362,
+ "step": 9568
+ },
+ {
+ "epoch": 26.07356948228883,
+ "grad_norm": 4.712894916534424,
+ "learning_rate": 1.733529031977886e-05,
+ "loss": 0.1912,
+ "step": 9569
+ },
+ {
+ "epoch": 26.076294277929154,
+ "grad_norm": 4.807603359222412,
+ "learning_rate": 1.7334690496571677e-05,
+ "loss": 0.2482,
+ "step": 9570
+ },
+ {
+ "epoch": 26.079019073569484,
+ "grad_norm": 6.552379131317139,
+ "learning_rate": 1.733409061624216e-05,
+ "loss": 0.3473,
+ "step": 9571
+ },
+ {
+ "epoch": 26.08174386920981,
+ "grad_norm": 5.273391246795654,
+ "learning_rate": 1.733349067879498e-05,
+ "loss": 0.4319,
+ "step": 9572
+ },
+ {
+ "epoch": 26.084468664850135,
+ "grad_norm": 5.220705509185791,
+ "learning_rate": 1.7332890684234815e-05,
+ "loss": 0.3952,
+ "step": 9573
+ },
+ {
+ "epoch": 26.087193460490465,
+ "grad_norm": 4.729302883148193,
+ "learning_rate": 1.7332290632566325e-05,
+ "loss": 0.2991,
+ "step": 9574
+ },
+ {
+ "epoch": 26.08991825613079,
+ "grad_norm": 4.819477081298828,
+ "learning_rate": 1.73316905237942e-05,
+ "loss": 0.3031,
+ "step": 9575
+ },
+ {
+ "epoch": 26.092643051771116,
+ "grad_norm": 5.66650390625,
+ "learning_rate": 1.7331090357923103e-05,
+ "loss": 0.3136,
+ "step": 9576
+ },
+ {
+ "epoch": 26.095367847411445,
+ "grad_norm": 5.878567695617676,
+ "learning_rate": 1.7330490134957707e-05,
+ "loss": 0.3119,
+ "step": 9577
+ },
+ {
+ "epoch": 26.09809264305177,
+ "grad_norm": 5.958669662475586,
+ "learning_rate": 1.732988985490269e-05,
+ "loss": 0.3584,
+ "step": 9578
+ },
+ {
+ "epoch": 26.100817438692097,
+ "grad_norm": 5.324524402618408,
+ "learning_rate": 1.7329289517762733e-05,
+ "loss": 0.3109,
+ "step": 9579
+ },
+ {
+ "epoch": 26.103542234332426,
+ "grad_norm": 4.692817211151123,
+ "learning_rate": 1.73286891235425e-05,
+ "loss": 0.3107,
+ "step": 9580
+ },
+ {
+ "epoch": 26.106267029972752,
+ "grad_norm": 5.552659034729004,
+ "learning_rate": 1.7328088672246674e-05,
+ "loss": 0.2321,
+ "step": 9581
+ },
+ {
+ "epoch": 26.108991825613078,
+ "grad_norm": 4.943843841552734,
+ "learning_rate": 1.7327488163879923e-05,
+ "loss": 0.3538,
+ "step": 9582
+ },
+ {
+ "epoch": 26.111716621253407,
+ "grad_norm": 5.669398784637451,
+ "learning_rate": 1.7326887598446937e-05,
+ "loss": 0.346,
+ "step": 9583
+ },
+ {
+ "epoch": 26.114441416893733,
+ "grad_norm": 6.3304314613342285,
+ "learning_rate": 1.7326286975952387e-05,
+ "loss": 0.3071,
+ "step": 9584
+ },
+ {
+ "epoch": 26.11716621253406,
+ "grad_norm": 3.9886975288391113,
+ "learning_rate": 1.7325686296400946e-05,
+ "loss": 0.3317,
+ "step": 9585
+ },
+ {
+ "epoch": 26.11989100817439,
+ "grad_norm": 5.046403408050537,
+ "learning_rate": 1.7325085559797295e-05,
+ "loss": 0.3106,
+ "step": 9586
+ },
+ {
+ "epoch": 26.122615803814714,
+ "grad_norm": 6.530880928039551,
+ "learning_rate": 1.7324484766146118e-05,
+ "loss": 0.3808,
+ "step": 9587
+ },
+ {
+ "epoch": 26.12534059945504,
+ "grad_norm": 5.2441582679748535,
+ "learning_rate": 1.7323883915452087e-05,
+ "loss": 0.3418,
+ "step": 9588
+ },
+ {
+ "epoch": 26.12806539509537,
+ "grad_norm": 4.432911396026611,
+ "learning_rate": 1.7323283007719883e-05,
+ "loss": 0.361,
+ "step": 9589
+ },
+ {
+ "epoch": 26.130790190735695,
+ "grad_norm": 4.430814743041992,
+ "learning_rate": 1.7322682042954186e-05,
+ "loss": 0.3676,
+ "step": 9590
+ },
+ {
+ "epoch": 26.13351498637602,
+ "grad_norm": 6.1370673179626465,
+ "learning_rate": 1.732208102115968e-05,
+ "loss": 0.5799,
+ "step": 9591
+ },
+ {
+ "epoch": 26.13623978201635,
+ "grad_norm": 4.432404518127441,
+ "learning_rate": 1.732147994234104e-05,
+ "loss": 0.3183,
+ "step": 9592
+ },
+ {
+ "epoch": 26.138964577656676,
+ "grad_norm": 5.902565002441406,
+ "learning_rate": 1.732087880650295e-05,
+ "loss": 0.3636,
+ "step": 9593
+ },
+ {
+ "epoch": 26.141689373297,
+ "grad_norm": 5.079205513000488,
+ "learning_rate": 1.7320277613650093e-05,
+ "loss": 0.3626,
+ "step": 9594
+ },
+ {
+ "epoch": 26.14441416893733,
+ "grad_norm": 4.50781774520874,
+ "learning_rate": 1.7319676363787146e-05,
+ "loss": 0.4388,
+ "step": 9595
+ },
+ {
+ "epoch": 26.147138964577657,
+ "grad_norm": 4.463652610778809,
+ "learning_rate": 1.7319075056918798e-05,
+ "loss": 0.2926,
+ "step": 9596
+ },
+ {
+ "epoch": 26.149863760217983,
+ "grad_norm": 5.024562835693359,
+ "learning_rate": 1.731847369304973e-05,
+ "loss": 0.2872,
+ "step": 9597
+ },
+ {
+ "epoch": 26.152588555858312,
+ "grad_norm": 5.042383193969727,
+ "learning_rate": 1.7317872272184624e-05,
+ "loss": 0.3237,
+ "step": 9598
+ },
+ {
+ "epoch": 26.155313351498638,
+ "grad_norm": 4.830783367156982,
+ "learning_rate": 1.7317270794328164e-05,
+ "loss": 0.3375,
+ "step": 9599
+ },
+ {
+ "epoch": 26.158038147138964,
+ "grad_norm": 9.866787910461426,
+ "learning_rate": 1.731666925948503e-05,
+ "loss": 0.5447,
+ "step": 9600
+ },
+ {
+ "epoch": 26.160762942779293,
+ "grad_norm": 5.429697036743164,
+ "learning_rate": 1.731606766765992e-05,
+ "loss": 0.3155,
+ "step": 9601
+ },
+ {
+ "epoch": 26.16348773841962,
+ "grad_norm": 5.121406078338623,
+ "learning_rate": 1.7315466018857505e-05,
+ "loss": 0.3381,
+ "step": 9602
+ },
+ {
+ "epoch": 26.166212534059945,
+ "grad_norm": 4.530823707580566,
+ "learning_rate": 1.7314864313082477e-05,
+ "loss": 0.342,
+ "step": 9603
+ },
+ {
+ "epoch": 26.168937329700274,
+ "grad_norm": 5.143324851989746,
+ "learning_rate": 1.731426255033952e-05,
+ "loss": 0.3627,
+ "step": 9604
+ },
+ {
+ "epoch": 26.1716621253406,
+ "grad_norm": 5.0236358642578125,
+ "learning_rate": 1.7313660730633324e-05,
+ "loss": 0.3,
+ "step": 9605
+ },
+ {
+ "epoch": 26.174386920980925,
+ "grad_norm": 4.806825160980225,
+ "learning_rate": 1.7313058853968572e-05,
+ "loss": 0.4811,
+ "step": 9606
+ },
+ {
+ "epoch": 26.177111716621255,
+ "grad_norm": 6.0061936378479,
+ "learning_rate": 1.731245692034995e-05,
+ "loss": 0.2744,
+ "step": 9607
+ },
+ {
+ "epoch": 26.17983651226158,
+ "grad_norm": 5.067643642425537,
+ "learning_rate": 1.7311854929782154e-05,
+ "loss": 0.3163,
+ "step": 9608
+ },
+ {
+ "epoch": 26.182561307901906,
+ "grad_norm": 5.221210479736328,
+ "learning_rate": 1.7311252882269864e-05,
+ "loss": 0.3683,
+ "step": 9609
+ },
+ {
+ "epoch": 26.185286103542236,
+ "grad_norm": 5.807234764099121,
+ "learning_rate": 1.7310650777817775e-05,
+ "loss": 0.4321,
+ "step": 9610
+ },
+ {
+ "epoch": 26.18801089918256,
+ "grad_norm": 4.861908435821533,
+ "learning_rate": 1.7310048616430572e-05,
+ "loss": 0.4933,
+ "step": 9611
+ },
+ {
+ "epoch": 26.190735694822887,
+ "grad_norm": 4.579342365264893,
+ "learning_rate": 1.7309446398112942e-05,
+ "loss": 0.3686,
+ "step": 9612
+ },
+ {
+ "epoch": 26.193460490463217,
+ "grad_norm": 5.444657802581787,
+ "learning_rate": 1.7308844122869583e-05,
+ "loss": 0.5516,
+ "step": 9613
+ },
+ {
+ "epoch": 26.196185286103542,
+ "grad_norm": 4.814473628997803,
+ "learning_rate": 1.7308241790705182e-05,
+ "loss": 0.5398,
+ "step": 9614
+ },
+ {
+ "epoch": 26.19891008174387,
+ "grad_norm": 5.0465497970581055,
+ "learning_rate": 1.7307639401624426e-05,
+ "loss": 0.329,
+ "step": 9615
+ },
+ {
+ "epoch": 26.201634877384198,
+ "grad_norm": 5.05996561050415,
+ "learning_rate": 1.730703695563201e-05,
+ "loss": 0.3214,
+ "step": 9616
+ },
+ {
+ "epoch": 26.204359673024523,
+ "grad_norm": 6.593718528747559,
+ "learning_rate": 1.7306434452732626e-05,
+ "loss": 0.6557,
+ "step": 9617
+ },
+ {
+ "epoch": 26.20708446866485,
+ "grad_norm": 4.793399333953857,
+ "learning_rate": 1.7305831892930967e-05,
+ "loss": 0.2751,
+ "step": 9618
+ },
+ {
+ "epoch": 26.20980926430518,
+ "grad_norm": 4.8659515380859375,
+ "learning_rate": 1.7305229276231723e-05,
+ "loss": 0.3298,
+ "step": 9619
+ },
+ {
+ "epoch": 26.212534059945504,
+ "grad_norm": 3.9265530109405518,
+ "learning_rate": 1.7304626602639586e-05,
+ "loss": 0.4164,
+ "step": 9620
+ },
+ {
+ "epoch": 26.21525885558583,
+ "grad_norm": 4.719543933868408,
+ "learning_rate": 1.7304023872159253e-05,
+ "loss": 0.2626,
+ "step": 9621
+ },
+ {
+ "epoch": 26.21798365122616,
+ "grad_norm": 5.600584983825684,
+ "learning_rate": 1.7303421084795423e-05,
+ "loss": 0.3089,
+ "step": 9622
+ },
+ {
+ "epoch": 26.220708446866485,
+ "grad_norm": 4.869002342224121,
+ "learning_rate": 1.7302818240552775e-05,
+ "loss": 0.3534,
+ "step": 9623
+ },
+ {
+ "epoch": 26.22343324250681,
+ "grad_norm": 5.236456871032715,
+ "learning_rate": 1.7302215339436022e-05,
+ "loss": 0.3117,
+ "step": 9624
+ },
+ {
+ "epoch": 26.22615803814714,
+ "grad_norm": 5.731258392333984,
+ "learning_rate": 1.7301612381449846e-05,
+ "loss": 0.3016,
+ "step": 9625
+ },
+ {
+ "epoch": 26.228882833787466,
+ "grad_norm": 5.627731800079346,
+ "learning_rate": 1.730100936659895e-05,
+ "loss": 0.5138,
+ "step": 9626
+ },
+ {
+ "epoch": 26.231607629427792,
+ "grad_norm": 4.44386625289917,
+ "learning_rate": 1.7300406294888027e-05,
+ "loss": 0.4635,
+ "step": 9627
+ },
+ {
+ "epoch": 26.23433242506812,
+ "grad_norm": 4.983173847198486,
+ "learning_rate": 1.7299803166321778e-05,
+ "loss": 0.4725,
+ "step": 9628
+ },
+ {
+ "epoch": 26.237057220708447,
+ "grad_norm": 4.8770270347595215,
+ "learning_rate": 1.729919998090489e-05,
+ "loss": 0.3995,
+ "step": 9629
+ },
+ {
+ "epoch": 26.239782016348773,
+ "grad_norm": 5.287332057952881,
+ "learning_rate": 1.7298596738642074e-05,
+ "loss": 0.3204,
+ "step": 9630
+ },
+ {
+ "epoch": 26.242506811989102,
+ "grad_norm": 4.327709197998047,
+ "learning_rate": 1.7297993439538017e-05,
+ "loss": 0.2646,
+ "step": 9631
+ },
+ {
+ "epoch": 26.245231607629428,
+ "grad_norm": 4.718850612640381,
+ "learning_rate": 1.7297390083597426e-05,
+ "loss": 0.3932,
+ "step": 9632
+ },
+ {
+ "epoch": 26.247956403269754,
+ "grad_norm": 5.663214683532715,
+ "learning_rate": 1.7296786670824993e-05,
+ "loss": 0.433,
+ "step": 9633
+ },
+ {
+ "epoch": 26.250681198910083,
+ "grad_norm": 5.663807392120361,
+ "learning_rate": 1.7296183201225422e-05,
+ "loss": 0.2924,
+ "step": 9634
+ },
+ {
+ "epoch": 26.25340599455041,
+ "grad_norm": 5.24136209487915,
+ "learning_rate": 1.729557967480341e-05,
+ "loss": 0.3217,
+ "step": 9635
+ },
+ {
+ "epoch": 26.256130790190735,
+ "grad_norm": 5.15399169921875,
+ "learning_rate": 1.7294976091563663e-05,
+ "loss": 0.3188,
+ "step": 9636
+ },
+ {
+ "epoch": 26.258855585831064,
+ "grad_norm": 5.671777725219727,
+ "learning_rate": 1.729437245151087e-05,
+ "loss": 0.2683,
+ "step": 9637
+ },
+ {
+ "epoch": 26.26158038147139,
+ "grad_norm": 5.825791358947754,
+ "learning_rate": 1.7293768754649743e-05,
+ "loss": 0.291,
+ "step": 9638
+ },
+ {
+ "epoch": 26.264305177111716,
+ "grad_norm": 4.8174519538879395,
+ "learning_rate": 1.7293165000984977e-05,
+ "loss": 0.4143,
+ "step": 9639
+ },
+ {
+ "epoch": 26.267029972752045,
+ "grad_norm": 5.217013835906982,
+ "learning_rate": 1.729256119052128e-05,
+ "loss": 0.3682,
+ "step": 9640
+ },
+ {
+ "epoch": 26.26975476839237,
+ "grad_norm": 4.773427486419678,
+ "learning_rate": 1.729195732326335e-05,
+ "loss": 0.2862,
+ "step": 9641
+ },
+ {
+ "epoch": 26.272479564032697,
+ "grad_norm": 6.086861610412598,
+ "learning_rate": 1.729135339921589e-05,
+ "loss": 0.4637,
+ "step": 9642
+ },
+ {
+ "epoch": 26.275204359673026,
+ "grad_norm": 6.256682395935059,
+ "learning_rate": 1.7290749418383604e-05,
+ "loss": 0.3887,
+ "step": 9643
+ },
+ {
+ "epoch": 26.277929155313352,
+ "grad_norm": 5.457511901855469,
+ "learning_rate": 1.72901453807712e-05,
+ "loss": 0.4073,
+ "step": 9644
+ },
+ {
+ "epoch": 26.280653950953678,
+ "grad_norm": 4.575545787811279,
+ "learning_rate": 1.7289541286383374e-05,
+ "loss": 0.3368,
+ "step": 9645
+ },
+ {
+ "epoch": 26.283378746594007,
+ "grad_norm": 4.568928241729736,
+ "learning_rate": 1.7288937135224833e-05,
+ "loss": 0.2791,
+ "step": 9646
+ },
+ {
+ "epoch": 26.286103542234333,
+ "grad_norm": 5.563901424407959,
+ "learning_rate": 1.7288332927300292e-05,
+ "loss": 0.3126,
+ "step": 9647
+ },
+ {
+ "epoch": 26.28882833787466,
+ "grad_norm": 4.98119592666626,
+ "learning_rate": 1.7287728662614444e-05,
+ "loss": 0.3748,
+ "step": 9648
+ },
+ {
+ "epoch": 26.291553133514988,
+ "grad_norm": 4.727115154266357,
+ "learning_rate": 1.7287124341171998e-05,
+ "loss": 0.3696,
+ "step": 9649
+ },
+ {
+ "epoch": 26.294277929155314,
+ "grad_norm": 5.128929138183594,
+ "learning_rate": 1.7286519962977663e-05,
+ "loss": 0.317,
+ "step": 9650
+ },
+ {
+ "epoch": 26.29700272479564,
+ "grad_norm": 4.632268905639648,
+ "learning_rate": 1.728591552803615e-05,
+ "loss": 0.3581,
+ "step": 9651
+ },
+ {
+ "epoch": 26.29972752043597,
+ "grad_norm": 5.6300272941589355,
+ "learning_rate": 1.7285311036352156e-05,
+ "loss": 0.4615,
+ "step": 9652
+ },
+ {
+ "epoch": 26.302452316076295,
+ "grad_norm": 6.2477707862854,
+ "learning_rate": 1.7284706487930396e-05,
+ "loss": 0.4671,
+ "step": 9653
+ },
+ {
+ "epoch": 26.30517711171662,
+ "grad_norm": 5.91253137588501,
+ "learning_rate": 1.7284101882775575e-05,
+ "loss": 0.4229,
+ "step": 9654
+ },
+ {
+ "epoch": 26.30790190735695,
+ "grad_norm": 4.456441879272461,
+ "learning_rate": 1.7283497220892406e-05,
+ "loss": 0.2735,
+ "step": 9655
+ },
+ {
+ "epoch": 26.310626702997276,
+ "grad_norm": 4.945108413696289,
+ "learning_rate": 1.728289250228559e-05,
+ "loss": 0.3104,
+ "step": 9656
+ },
+ {
+ "epoch": 26.3133514986376,
+ "grad_norm": 4.695904731750488,
+ "learning_rate": 1.7282287726959847e-05,
+ "loss": 0.4269,
+ "step": 9657
+ },
+ {
+ "epoch": 26.31607629427793,
+ "grad_norm": 5.637450218200684,
+ "learning_rate": 1.728168289491988e-05,
+ "loss": 0.3641,
+ "step": 9658
+ },
+ {
+ "epoch": 26.318801089918257,
+ "grad_norm": 4.775197505950928,
+ "learning_rate": 1.72810780061704e-05,
+ "loss": 0.2301,
+ "step": 9659
+ },
+ {
+ "epoch": 26.321525885558582,
+ "grad_norm": 4.860262870788574,
+ "learning_rate": 1.728047306071612e-05,
+ "loss": 0.4309,
+ "step": 9660
+ },
+ {
+ "epoch": 26.32425068119891,
+ "grad_norm": 5.189157485961914,
+ "learning_rate": 1.727986805856175e-05,
+ "loss": 0.2606,
+ "step": 9661
+ },
+ {
+ "epoch": 26.326975476839237,
+ "grad_norm": 4.8771562576293945,
+ "learning_rate": 1.7279262999712e-05,
+ "loss": 0.2667,
+ "step": 9662
+ },
+ {
+ "epoch": 26.329700272479563,
+ "grad_norm": 9.391637802124023,
+ "learning_rate": 1.7278657884171585e-05,
+ "loss": 0.4034,
+ "step": 9663
+ },
+ {
+ "epoch": 26.332425068119893,
+ "grad_norm": 5.105279922485352,
+ "learning_rate": 1.7278052711945218e-05,
+ "loss": 0.2936,
+ "step": 9664
+ },
+ {
+ "epoch": 26.33514986376022,
+ "grad_norm": 5.621169090270996,
+ "learning_rate": 1.7277447483037608e-05,
+ "loss": 0.3354,
+ "step": 9665
+ },
+ {
+ "epoch": 26.337874659400544,
+ "grad_norm": 5.3450751304626465,
+ "learning_rate": 1.7276842197453475e-05,
+ "loss": 0.3693,
+ "step": 9666
+ },
+ {
+ "epoch": 26.340599455040874,
+ "grad_norm": 5.028762340545654,
+ "learning_rate": 1.7276236855197526e-05,
+ "loss": 0.3631,
+ "step": 9667
+ },
+ {
+ "epoch": 26.3433242506812,
+ "grad_norm": 5.640786170959473,
+ "learning_rate": 1.7275631456274476e-05,
+ "loss": 0.3529,
+ "step": 9668
+ },
+ {
+ "epoch": 26.346049046321525,
+ "grad_norm": 5.816191673278809,
+ "learning_rate": 1.727502600068905e-05,
+ "loss": 0.3417,
+ "step": 9669
+ },
+ {
+ "epoch": 26.348773841961854,
+ "grad_norm": 4.972781181335449,
+ "learning_rate": 1.727442048844595e-05,
+ "loss": 0.6865,
+ "step": 9670
+ },
+ {
+ "epoch": 26.35149863760218,
+ "grad_norm": 6.147048473358154,
+ "learning_rate": 1.72738149195499e-05,
+ "loss": 0.5561,
+ "step": 9671
+ },
+ {
+ "epoch": 26.354223433242506,
+ "grad_norm": 5.191177845001221,
+ "learning_rate": 1.727320929400561e-05,
+ "loss": 0.3355,
+ "step": 9672
+ },
+ {
+ "epoch": 26.356948228882835,
+ "grad_norm": 4.288514137268066,
+ "learning_rate": 1.7272603611817803e-05,
+ "loss": 0.4132,
+ "step": 9673
+ },
+ {
+ "epoch": 26.35967302452316,
+ "grad_norm": 5.099119663238525,
+ "learning_rate": 1.7271997872991192e-05,
+ "loss": 0.3051,
+ "step": 9674
+ },
+ {
+ "epoch": 26.362397820163487,
+ "grad_norm": 4.855091094970703,
+ "learning_rate": 1.7271392077530498e-05,
+ "loss": 0.367,
+ "step": 9675
+ },
+ {
+ "epoch": 26.365122615803816,
+ "grad_norm": 6.036309242248535,
+ "learning_rate": 1.7270786225440434e-05,
+ "loss": 0.3267,
+ "step": 9676
+ },
+ {
+ "epoch": 26.367847411444142,
+ "grad_norm": 5.4419169425964355,
+ "learning_rate": 1.727018031672572e-05,
+ "loss": 0.3506,
+ "step": 9677
+ },
+ {
+ "epoch": 26.370572207084468,
+ "grad_norm": 5.218389987945557,
+ "learning_rate": 1.726957435139108e-05,
+ "loss": 0.2875,
+ "step": 9678
+ },
+ {
+ "epoch": 26.373297002724797,
+ "grad_norm": 4.6730499267578125,
+ "learning_rate": 1.7268968329441226e-05,
+ "loss": 0.2636,
+ "step": 9679
+ },
+ {
+ "epoch": 26.376021798365123,
+ "grad_norm": 5.683510780334473,
+ "learning_rate": 1.7268362250880878e-05,
+ "loss": 0.4951,
+ "step": 9680
+ },
+ {
+ "epoch": 26.37874659400545,
+ "grad_norm": 5.406607151031494,
+ "learning_rate": 1.726775611571476e-05,
+ "loss": 0.2713,
+ "step": 9681
+ },
+ {
+ "epoch": 26.381471389645778,
+ "grad_norm": 5.254608631134033,
+ "learning_rate": 1.7267149923947598e-05,
+ "loss": 0.3688,
+ "step": 9682
+ },
+ {
+ "epoch": 26.384196185286104,
+ "grad_norm": 5.4616594314575195,
+ "learning_rate": 1.72665436755841e-05,
+ "loss": 0.3474,
+ "step": 9683
+ },
+ {
+ "epoch": 26.38692098092643,
+ "grad_norm": 5.289910316467285,
+ "learning_rate": 1.7265937370628995e-05,
+ "loss": 0.3303,
+ "step": 9684
+ },
+ {
+ "epoch": 26.38964577656676,
+ "grad_norm": 5.264852523803711,
+ "learning_rate": 1.7265331009087004e-05,
+ "loss": 0.3346,
+ "step": 9685
+ },
+ {
+ "epoch": 26.392370572207085,
+ "grad_norm": 4.701746940612793,
+ "learning_rate": 1.7264724590962846e-05,
+ "loss": 0.2961,
+ "step": 9686
+ },
+ {
+ "epoch": 26.39509536784741,
+ "grad_norm": 5.42265510559082,
+ "learning_rate": 1.726411811626125e-05,
+ "loss": 0.3327,
+ "step": 9687
+ },
+ {
+ "epoch": 26.39782016348774,
+ "grad_norm": 5.79636812210083,
+ "learning_rate": 1.7263511584986935e-05,
+ "loss": 0.4022,
+ "step": 9688
+ },
+ {
+ "epoch": 26.400544959128066,
+ "grad_norm": 4.072159290313721,
+ "learning_rate": 1.7262904997144626e-05,
+ "loss": 0.289,
+ "step": 9689
+ },
+ {
+ "epoch": 26.40326975476839,
+ "grad_norm": 5.272220611572266,
+ "learning_rate": 1.7262298352739048e-05,
+ "loss": 0.3423,
+ "step": 9690
+ },
+ {
+ "epoch": 26.40599455040872,
+ "grad_norm": 5.655509948730469,
+ "learning_rate": 1.7261691651774922e-05,
+ "loss": 0.4893,
+ "step": 9691
+ },
+ {
+ "epoch": 26.408719346049047,
+ "grad_norm": 4.917480945587158,
+ "learning_rate": 1.7261084894256975e-05,
+ "loss": 0.4194,
+ "step": 9692
+ },
+ {
+ "epoch": 26.411444141689373,
+ "grad_norm": 4.470883369445801,
+ "learning_rate": 1.7260478080189932e-05,
+ "loss": 0.4492,
+ "step": 9693
+ },
+ {
+ "epoch": 26.414168937329702,
+ "grad_norm": 4.619857311248779,
+ "learning_rate": 1.725987120957852e-05,
+ "loss": 0.5266,
+ "step": 9694
+ },
+ {
+ "epoch": 26.416893732970028,
+ "grad_norm": 5.772851943969727,
+ "learning_rate": 1.7259264282427467e-05,
+ "loss": 0.3348,
+ "step": 9695
+ },
+ {
+ "epoch": 26.419618528610354,
+ "grad_norm": 5.385730743408203,
+ "learning_rate": 1.7258657298741494e-05,
+ "loss": 0.2747,
+ "step": 9696
+ },
+ {
+ "epoch": 26.422343324250683,
+ "grad_norm": 4.798292636871338,
+ "learning_rate": 1.7258050258525335e-05,
+ "loss": 0.3125,
+ "step": 9697
+ },
+ {
+ "epoch": 26.42506811989101,
+ "grad_norm": 9.545672416687012,
+ "learning_rate": 1.7257443161783712e-05,
+ "loss": 0.2602,
+ "step": 9698
+ },
+ {
+ "epoch": 26.427792915531334,
+ "grad_norm": 5.125973701477051,
+ "learning_rate": 1.7256836008521354e-05,
+ "loss": 0.4936,
+ "step": 9699
+ },
+ {
+ "epoch": 26.430517711171664,
+ "grad_norm": 5.370091438293457,
+ "learning_rate": 1.725622879874299e-05,
+ "loss": 0.251,
+ "step": 9700
+ },
+ {
+ "epoch": 26.43324250681199,
+ "grad_norm": 4.601172924041748,
+ "learning_rate": 1.7255621532453356e-05,
+ "loss": 0.4319,
+ "step": 9701
+ },
+ {
+ "epoch": 26.435967302452315,
+ "grad_norm": 4.70193338394165,
+ "learning_rate": 1.725501420965717e-05,
+ "loss": 0.2766,
+ "step": 9702
+ },
+ {
+ "epoch": 26.438692098092645,
+ "grad_norm": 5.12919807434082,
+ "learning_rate": 1.7254406830359166e-05,
+ "loss": 0.3498,
+ "step": 9703
+ },
+ {
+ "epoch": 26.44141689373297,
+ "grad_norm": 5.7979536056518555,
+ "learning_rate": 1.725379939456408e-05,
+ "loss": 0.3133,
+ "step": 9704
+ },
+ {
+ "epoch": 26.444141689373296,
+ "grad_norm": 4.484868049621582,
+ "learning_rate": 1.7253191902276634e-05,
+ "loss": 0.3361,
+ "step": 9705
+ },
+ {
+ "epoch": 26.446866485013626,
+ "grad_norm": 5.575591564178467,
+ "learning_rate": 1.725258435350156e-05,
+ "loss": 0.446,
+ "step": 9706
+ },
+ {
+ "epoch": 26.44959128065395,
+ "grad_norm": 5.204803466796875,
+ "learning_rate": 1.7251976748243597e-05,
+ "loss": 0.4214,
+ "step": 9707
+ },
+ {
+ "epoch": 26.452316076294277,
+ "grad_norm": 5.442605018615723,
+ "learning_rate": 1.7251369086507472e-05,
+ "loss": 0.4561,
+ "step": 9708
+ },
+ {
+ "epoch": 26.455040871934607,
+ "grad_norm": 6.704582214355469,
+ "learning_rate": 1.725076136829792e-05,
+ "loss": 0.4162,
+ "step": 9709
+ },
+ {
+ "epoch": 26.457765667574932,
+ "grad_norm": 4.999823570251465,
+ "learning_rate": 1.725015359361967e-05,
+ "loss": 0.4992,
+ "step": 9710
+ },
+ {
+ "epoch": 26.460490463215258,
+ "grad_norm": 5.126186847686768,
+ "learning_rate": 1.7249545762477454e-05,
+ "loss": 0.3583,
+ "step": 9711
+ },
+ {
+ "epoch": 26.463215258855588,
+ "grad_norm": 4.544831275939941,
+ "learning_rate": 1.7248937874876013e-05,
+ "loss": 0.3402,
+ "step": 9712
+ },
+ {
+ "epoch": 26.465940054495913,
+ "grad_norm": 5.425344944000244,
+ "learning_rate": 1.7248329930820076e-05,
+ "loss": 0.2102,
+ "step": 9713
+ },
+ {
+ "epoch": 26.46866485013624,
+ "grad_norm": 4.952349662780762,
+ "learning_rate": 1.7247721930314377e-05,
+ "loss": 0.3184,
+ "step": 9714
+ },
+ {
+ "epoch": 26.47138964577657,
+ "grad_norm": 5.955500602722168,
+ "learning_rate": 1.7247113873363656e-05,
+ "loss": 0.3907,
+ "step": 9715
+ },
+ {
+ "epoch": 26.474114441416894,
+ "grad_norm": 5.46176290512085,
+ "learning_rate": 1.7246505759972643e-05,
+ "loss": 0.3689,
+ "step": 9716
+ },
+ {
+ "epoch": 26.47683923705722,
+ "grad_norm": 6.092106342315674,
+ "learning_rate": 1.7245897590146078e-05,
+ "loss": 0.418,
+ "step": 9717
+ },
+ {
+ "epoch": 26.479564032697546,
+ "grad_norm": 5.202536582946777,
+ "learning_rate": 1.7245289363888694e-05,
+ "loss": 0.3428,
+ "step": 9718
+ },
+ {
+ "epoch": 26.482288828337875,
+ "grad_norm": 4.810392379760742,
+ "learning_rate": 1.7244681081205227e-05,
+ "loss": 0.2923,
+ "step": 9719
+ },
+ {
+ "epoch": 26.4850136239782,
+ "grad_norm": 4.7935638427734375,
+ "learning_rate": 1.7244072742100422e-05,
+ "loss": 0.2703,
+ "step": 9720
+ },
+ {
+ "epoch": 26.48773841961853,
+ "grad_norm": 5.649646759033203,
+ "learning_rate": 1.724346434657901e-05,
+ "loss": 0.4126,
+ "step": 9721
+ },
+ {
+ "epoch": 26.490463215258856,
+ "grad_norm": 5.113173961639404,
+ "learning_rate": 1.724285589464573e-05,
+ "loss": 0.2601,
+ "step": 9722
+ },
+ {
+ "epoch": 26.493188010899182,
+ "grad_norm": 4.989759922027588,
+ "learning_rate": 1.7242247386305324e-05,
+ "loss": 0.2895,
+ "step": 9723
+ },
+ {
+ "epoch": 26.495912806539508,
+ "grad_norm": 6.241610050201416,
+ "learning_rate": 1.7241638821562526e-05,
+ "loss": 0.3012,
+ "step": 9724
+ },
+ {
+ "epoch": 26.498637602179837,
+ "grad_norm": 4.621860504150391,
+ "learning_rate": 1.7241030200422078e-05,
+ "loss": 0.3381,
+ "step": 9725
+ },
+ {
+ "epoch": 26.501362397820163,
+ "grad_norm": 4.580186367034912,
+ "learning_rate": 1.724042152288872e-05,
+ "loss": 0.3771,
+ "step": 9726
+ },
+ {
+ "epoch": 26.504087193460492,
+ "grad_norm": 4.8551344871521,
+ "learning_rate": 1.7239812788967195e-05,
+ "loss": 0.2782,
+ "step": 9727
+ },
+ {
+ "epoch": 26.506811989100818,
+ "grad_norm": 4.79332160949707,
+ "learning_rate": 1.7239203998662234e-05,
+ "loss": 0.3402,
+ "step": 9728
+ },
+ {
+ "epoch": 26.509536784741144,
+ "grad_norm": 6.020784854888916,
+ "learning_rate": 1.7238595151978593e-05,
+ "loss": 0.2958,
+ "step": 9729
+ },
+ {
+ "epoch": 26.51226158038147,
+ "grad_norm": 5.243679523468018,
+ "learning_rate": 1.7237986248921003e-05,
+ "loss": 0.2872,
+ "step": 9730
+ },
+ {
+ "epoch": 26.5149863760218,
+ "grad_norm": 4.846377372741699,
+ "learning_rate": 1.7237377289494207e-05,
+ "loss": 0.3764,
+ "step": 9731
+ },
+ {
+ "epoch": 26.517711171662125,
+ "grad_norm": 5.08028507232666,
+ "learning_rate": 1.7236768273702952e-05,
+ "loss": 0.3066,
+ "step": 9732
+ },
+ {
+ "epoch": 26.520435967302454,
+ "grad_norm": 5.443795680999756,
+ "learning_rate": 1.7236159201551978e-05,
+ "loss": 0.3092,
+ "step": 9733
+ },
+ {
+ "epoch": 26.52316076294278,
+ "grad_norm": 4.086330413818359,
+ "learning_rate": 1.723555007304603e-05,
+ "loss": 0.2693,
+ "step": 9734
+ },
+ {
+ "epoch": 26.525885558583106,
+ "grad_norm": 5.191492557525635,
+ "learning_rate": 1.7234940888189847e-05,
+ "loss": 0.2877,
+ "step": 9735
+ },
+ {
+ "epoch": 26.52861035422343,
+ "grad_norm": 6.07861328125,
+ "learning_rate": 1.7234331646988183e-05,
+ "loss": 0.337,
+ "step": 9736
+ },
+ {
+ "epoch": 26.53133514986376,
+ "grad_norm": 5.139482498168945,
+ "learning_rate": 1.7233722349445774e-05,
+ "loss": 0.3201,
+ "step": 9737
+ },
+ {
+ "epoch": 26.534059945504087,
+ "grad_norm": 6.162802696228027,
+ "learning_rate": 1.7233112995567366e-05,
+ "loss": 0.3492,
+ "step": 9738
+ },
+ {
+ "epoch": 26.536784741144416,
+ "grad_norm": 4.841139793395996,
+ "learning_rate": 1.723250358535771e-05,
+ "loss": 0.337,
+ "step": 9739
+ },
+ {
+ "epoch": 26.539509536784742,
+ "grad_norm": 5.1576995849609375,
+ "learning_rate": 1.723189411882155e-05,
+ "loss": 0.4422,
+ "step": 9740
+ },
+ {
+ "epoch": 26.542234332425068,
+ "grad_norm": 4.565083980560303,
+ "learning_rate": 1.723128459596363e-05,
+ "loss": 0.3947,
+ "step": 9741
+ },
+ {
+ "epoch": 26.544959128065393,
+ "grad_norm": 4.596365928649902,
+ "learning_rate": 1.7230675016788697e-05,
+ "loss": 0.3687,
+ "step": 9742
+ },
+ {
+ "epoch": 26.547683923705723,
+ "grad_norm": 7.19928503036499,
+ "learning_rate": 1.72300653813015e-05,
+ "loss": 0.3517,
+ "step": 9743
+ },
+ {
+ "epoch": 26.55040871934605,
+ "grad_norm": 5.094775676727295,
+ "learning_rate": 1.7229455689506786e-05,
+ "loss": 0.4207,
+ "step": 9744
+ },
+ {
+ "epoch": 26.553133514986374,
+ "grad_norm": 5.421530723571777,
+ "learning_rate": 1.7228845941409304e-05,
+ "loss": 0.4356,
+ "step": 9745
+ },
+ {
+ "epoch": 26.555858310626704,
+ "grad_norm": 4.442439079284668,
+ "learning_rate": 1.7228236137013804e-05,
+ "loss": 0.3468,
+ "step": 9746
+ },
+ {
+ "epoch": 26.55858310626703,
+ "grad_norm": 4.649865627288818,
+ "learning_rate": 1.7227626276325033e-05,
+ "loss": 0.2286,
+ "step": 9747
+ },
+ {
+ "epoch": 26.561307901907355,
+ "grad_norm": 5.123966693878174,
+ "learning_rate": 1.722701635934774e-05,
+ "loss": 0.3573,
+ "step": 9748
+ },
+ {
+ "epoch": 26.564032697547685,
+ "grad_norm": 5.303154945373535,
+ "learning_rate": 1.7226406386086675e-05,
+ "loss": 0.5486,
+ "step": 9749
+ },
+ {
+ "epoch": 26.56675749318801,
+ "grad_norm": 4.781341552734375,
+ "learning_rate": 1.7225796356546592e-05,
+ "loss": 0.4092,
+ "step": 9750
+ },
+ {
+ "epoch": 26.569482288828336,
+ "grad_norm": 5.870811462402344,
+ "learning_rate": 1.7225186270732242e-05,
+ "loss": 0.3941,
+ "step": 9751
+ },
+ {
+ "epoch": 26.572207084468666,
+ "grad_norm": 4.273779392242432,
+ "learning_rate": 1.722457612864837e-05,
+ "loss": 0.2782,
+ "step": 9752
+ },
+ {
+ "epoch": 26.57493188010899,
+ "grad_norm": 4.752996444702148,
+ "learning_rate": 1.7223965930299732e-05,
+ "loss": 0.3726,
+ "step": 9753
+ },
+ {
+ "epoch": 26.577656675749317,
+ "grad_norm": 4.609536170959473,
+ "learning_rate": 1.722335567569108e-05,
+ "loss": 0.6096,
+ "step": 9754
+ },
+ {
+ "epoch": 26.580381471389646,
+ "grad_norm": 6.84182596206665,
+ "learning_rate": 1.7222745364827167e-05,
+ "loss": 0.2984,
+ "step": 9755
+ },
+ {
+ "epoch": 26.583106267029972,
+ "grad_norm": 5.491619110107422,
+ "learning_rate": 1.7222134997712748e-05,
+ "loss": 0.4432,
+ "step": 9756
+ },
+ {
+ "epoch": 26.585831062670298,
+ "grad_norm": 4.899320602416992,
+ "learning_rate": 1.722152457435257e-05,
+ "loss": 0.4076,
+ "step": 9757
+ },
+ {
+ "epoch": 26.588555858310627,
+ "grad_norm": 6.034642696380615,
+ "learning_rate": 1.722091409475139e-05,
+ "loss": 0.4374,
+ "step": 9758
+ },
+ {
+ "epoch": 26.591280653950953,
+ "grad_norm": 6.01954984664917,
+ "learning_rate": 1.722030355891397e-05,
+ "loss": 0.3447,
+ "step": 9759
+ },
+ {
+ "epoch": 26.59400544959128,
+ "grad_norm": 4.995231628417969,
+ "learning_rate": 1.721969296684505e-05,
+ "loss": 0.3151,
+ "step": 9760
+ },
+ {
+ "epoch": 26.59673024523161,
+ "grad_norm": 4.955073356628418,
+ "learning_rate": 1.72190823185494e-05,
+ "loss": 0.2901,
+ "step": 9761
+ },
+ {
+ "epoch": 26.599455040871934,
+ "grad_norm": 4.507307052612305,
+ "learning_rate": 1.721847161403177e-05,
+ "loss": 0.5973,
+ "step": 9762
+ },
+ {
+ "epoch": 26.60217983651226,
+ "grad_norm": 4.576388359069824,
+ "learning_rate": 1.721786085329691e-05,
+ "loss": 0.254,
+ "step": 9763
+ },
+ {
+ "epoch": 26.60490463215259,
+ "grad_norm": 4.959639072418213,
+ "learning_rate": 1.7217250036349584e-05,
+ "loss": 0.3894,
+ "step": 9764
+ },
+ {
+ "epoch": 26.607629427792915,
+ "grad_norm": 4.852689266204834,
+ "learning_rate": 1.721663916319455e-05,
+ "loss": 0.2772,
+ "step": 9765
+ },
+ {
+ "epoch": 26.61035422343324,
+ "grad_norm": 5.047691822052002,
+ "learning_rate": 1.721602823383656e-05,
+ "loss": 0.2727,
+ "step": 9766
+ },
+ {
+ "epoch": 26.61307901907357,
+ "grad_norm": 4.992856979370117,
+ "learning_rate": 1.7215417248280374e-05,
+ "loss": 0.528,
+ "step": 9767
+ },
+ {
+ "epoch": 26.615803814713896,
+ "grad_norm": 5.258423328399658,
+ "learning_rate": 1.721480620653075e-05,
+ "loss": 0.3567,
+ "step": 9768
+ },
+ {
+ "epoch": 26.618528610354222,
+ "grad_norm": 5.421665668487549,
+ "learning_rate": 1.7214195108592448e-05,
+ "loss": 0.4393,
+ "step": 9769
+ },
+ {
+ "epoch": 26.62125340599455,
+ "grad_norm": 8.156648635864258,
+ "learning_rate": 1.7213583954470228e-05,
+ "loss": 0.4498,
+ "step": 9770
+ },
+ {
+ "epoch": 26.623978201634877,
+ "grad_norm": 5.990218639373779,
+ "learning_rate": 1.721297274416885e-05,
+ "loss": 0.3347,
+ "step": 9771
+ },
+ {
+ "epoch": 26.626702997275203,
+ "grad_norm": 5.673974990844727,
+ "learning_rate": 1.7212361477693068e-05,
+ "loss": 0.316,
+ "step": 9772
+ },
+ {
+ "epoch": 26.629427792915532,
+ "grad_norm": 6.174537181854248,
+ "learning_rate": 1.721175015504765e-05,
+ "loss": 0.2661,
+ "step": 9773
+ },
+ {
+ "epoch": 26.632152588555858,
+ "grad_norm": 4.834343910217285,
+ "learning_rate": 1.721113877623735e-05,
+ "loss": 0.3207,
+ "step": 9774
+ },
+ {
+ "epoch": 26.634877384196184,
+ "grad_norm": 6.918548583984375,
+ "learning_rate": 1.7210527341266936e-05,
+ "loss": 0.4369,
+ "step": 9775
+ },
+ {
+ "epoch": 26.637602179836513,
+ "grad_norm": 5.097619533538818,
+ "learning_rate": 1.720991585014117e-05,
+ "loss": 0.2963,
+ "step": 9776
+ },
+ {
+ "epoch": 26.64032697547684,
+ "grad_norm": 6.452464580535889,
+ "learning_rate": 1.7209304302864808e-05,
+ "loss": 0.3073,
+ "step": 9777
+ },
+ {
+ "epoch": 26.643051771117165,
+ "grad_norm": 5.1045989990234375,
+ "learning_rate": 1.7208692699442618e-05,
+ "loss": 0.2293,
+ "step": 9778
+ },
+ {
+ "epoch": 26.645776566757494,
+ "grad_norm": 5.079566955566406,
+ "learning_rate": 1.7208081039879355e-05,
+ "loss": 0.3772,
+ "step": 9779
+ },
+ {
+ "epoch": 26.64850136239782,
+ "grad_norm": 5.077652454376221,
+ "learning_rate": 1.7207469324179796e-05,
+ "loss": 0.3343,
+ "step": 9780
+ },
+ {
+ "epoch": 26.651226158038146,
+ "grad_norm": 4.961864948272705,
+ "learning_rate": 1.7206857552348696e-05,
+ "loss": 0.4345,
+ "step": 9781
+ },
+ {
+ "epoch": 26.653950953678475,
+ "grad_norm": 4.143431186676025,
+ "learning_rate": 1.720624572439082e-05,
+ "loss": 0.2865,
+ "step": 9782
+ },
+ {
+ "epoch": 26.6566757493188,
+ "grad_norm": 5.189888000488281,
+ "learning_rate": 1.7205633840310935e-05,
+ "loss": 0.4075,
+ "step": 9783
+ },
+ {
+ "epoch": 26.659400544959126,
+ "grad_norm": 4.926806449890137,
+ "learning_rate": 1.7205021900113808e-05,
+ "loss": 0.4387,
+ "step": 9784
+ },
+ {
+ "epoch": 26.662125340599456,
+ "grad_norm": 5.521529674530029,
+ "learning_rate": 1.72044099038042e-05,
+ "loss": 0.3987,
+ "step": 9785
+ },
+ {
+ "epoch": 26.66485013623978,
+ "grad_norm": 5.059208393096924,
+ "learning_rate": 1.7203797851386878e-05,
+ "loss": 0.3296,
+ "step": 9786
+ },
+ {
+ "epoch": 26.667574931880107,
+ "grad_norm": 154.3404998779297,
+ "learning_rate": 1.720318574286661e-05,
+ "loss": 0.3446,
+ "step": 9787
+ },
+ {
+ "epoch": 26.670299727520437,
+ "grad_norm": 4.742673873901367,
+ "learning_rate": 1.7202573578248166e-05,
+ "loss": 0.3175,
+ "step": 9788
+ },
+ {
+ "epoch": 26.673024523160763,
+ "grad_norm": 5.342245578765869,
+ "learning_rate": 1.7201961357536313e-05,
+ "loss": 0.3294,
+ "step": 9789
+ },
+ {
+ "epoch": 26.67574931880109,
+ "grad_norm": 5.849545478820801,
+ "learning_rate": 1.720134908073581e-05,
+ "loss": 0.3277,
+ "step": 9790
+ },
+ {
+ "epoch": 26.678474114441418,
+ "grad_norm": 135.12876892089844,
+ "learning_rate": 1.720073674785144e-05,
+ "loss": 0.3453,
+ "step": 9791
+ },
+ {
+ "epoch": 26.681198910081743,
+ "grad_norm": 7.306275844573975,
+ "learning_rate": 1.720012435888796e-05,
+ "loss": 0.3771,
+ "step": 9792
+ },
+ {
+ "epoch": 26.68392370572207,
+ "grad_norm": 9.141103744506836,
+ "learning_rate": 1.7199511913850142e-05,
+ "loss": 0.4292,
+ "step": 9793
+ },
+ {
+ "epoch": 26.6866485013624,
+ "grad_norm": 6.8838372230529785,
+ "learning_rate": 1.7198899412742756e-05,
+ "loss": 0.4407,
+ "step": 9794
+ },
+ {
+ "epoch": 26.689373297002724,
+ "grad_norm": 9.202927589416504,
+ "learning_rate": 1.7198286855570575e-05,
+ "loss": 0.5496,
+ "step": 9795
+ },
+ {
+ "epoch": 26.69209809264305,
+ "grad_norm": 10.833867073059082,
+ "learning_rate": 1.719767424233837e-05,
+ "loss": 0.6208,
+ "step": 9796
+ },
+ {
+ "epoch": 26.69482288828338,
+ "grad_norm": 10.93331241607666,
+ "learning_rate": 1.719706157305091e-05,
+ "loss": 0.5602,
+ "step": 9797
+ },
+ {
+ "epoch": 26.697547683923705,
+ "grad_norm": 24.29605484008789,
+ "learning_rate": 1.719644884771296e-05,
+ "loss": 0.3899,
+ "step": 9798
+ },
+ {
+ "epoch": 26.70027247956403,
+ "grad_norm": 8.936422348022461,
+ "learning_rate": 1.7195836066329304e-05,
+ "loss": 0.5727,
+ "step": 9799
+ },
+ {
+ "epoch": 26.70299727520436,
+ "grad_norm": 9.637221336364746,
+ "learning_rate": 1.7195223228904705e-05,
+ "loss": 0.4326,
+ "step": 9800
+ },
+ {
+ "epoch": 26.705722070844686,
+ "grad_norm": 7.862738609313965,
+ "learning_rate": 1.719461033544394e-05,
+ "loss": 0.3947,
+ "step": 9801
+ },
+ {
+ "epoch": 26.708446866485012,
+ "grad_norm": 8.379895210266113,
+ "learning_rate": 1.719399738595178e-05,
+ "loss": 0.5593,
+ "step": 9802
+ },
+ {
+ "epoch": 26.71117166212534,
+ "grad_norm": 9.003290176391602,
+ "learning_rate": 1.7193384380433e-05,
+ "loss": 0.4718,
+ "step": 9803
+ },
+ {
+ "epoch": 26.713896457765667,
+ "grad_norm": 9.4240140914917,
+ "learning_rate": 1.7192771318892377e-05,
+ "loss": 0.4684,
+ "step": 9804
+ },
+ {
+ "epoch": 26.716621253405993,
+ "grad_norm": 7.688920497894287,
+ "learning_rate": 1.719215820133468e-05,
+ "loss": 0.4122,
+ "step": 9805
+ },
+ {
+ "epoch": 26.719346049046322,
+ "grad_norm": 13.624361991882324,
+ "learning_rate": 1.7191545027764686e-05,
+ "loss": 0.4911,
+ "step": 9806
+ },
+ {
+ "epoch": 26.722070844686648,
+ "grad_norm": 11.531464576721191,
+ "learning_rate": 1.7190931798187173e-05,
+ "loss": 0.6,
+ "step": 9807
+ },
+ {
+ "epoch": 26.724795640326974,
+ "grad_norm": 10.174246788024902,
+ "learning_rate": 1.719031851260691e-05,
+ "loss": 0.5358,
+ "step": 9808
+ },
+ {
+ "epoch": 26.727520435967303,
+ "grad_norm": 21.457069396972656,
+ "learning_rate": 1.7189705171028684e-05,
+ "loss": 0.4449,
+ "step": 9809
+ },
+ {
+ "epoch": 26.73024523160763,
+ "grad_norm": 9.617992401123047,
+ "learning_rate": 1.7189091773457263e-05,
+ "loss": 0.4902,
+ "step": 9810
+ },
+ {
+ "epoch": 26.732970027247955,
+ "grad_norm": 10.918708801269531,
+ "learning_rate": 1.7188478319897425e-05,
+ "loss": 0.5883,
+ "step": 9811
+ },
+ {
+ "epoch": 26.735694822888284,
+ "grad_norm": 8.089766502380371,
+ "learning_rate": 1.718786481035395e-05,
+ "loss": 0.4916,
+ "step": 9812
+ },
+ {
+ "epoch": 26.73841961852861,
+ "grad_norm": 7.103590965270996,
+ "learning_rate": 1.7187251244831617e-05,
+ "loss": 0.4775,
+ "step": 9813
+ },
+ {
+ "epoch": 26.741144414168936,
+ "grad_norm": 7.786047458648682,
+ "learning_rate": 1.7186637623335202e-05,
+ "loss": 0.4981,
+ "step": 9814
+ },
+ {
+ "epoch": 26.743869209809265,
+ "grad_norm": 6.762415409088135,
+ "learning_rate": 1.7186023945869482e-05,
+ "loss": 0.4829,
+ "step": 9815
+ },
+ {
+ "epoch": 26.74659400544959,
+ "grad_norm": 6.786586284637451,
+ "learning_rate": 1.718541021243924e-05,
+ "loss": 0.3913,
+ "step": 9816
+ },
+ {
+ "epoch": 26.749318801089917,
+ "grad_norm": 7.2916059494018555,
+ "learning_rate": 1.7184796423049258e-05,
+ "loss": 0.4487,
+ "step": 9817
+ },
+ {
+ "epoch": 26.752043596730246,
+ "grad_norm": 10.812594413757324,
+ "learning_rate": 1.7184182577704308e-05,
+ "loss": 0.5911,
+ "step": 9818
+ },
+ {
+ "epoch": 26.754768392370572,
+ "grad_norm": 5.950011730194092,
+ "learning_rate": 1.7183568676409177e-05,
+ "loss": 0.4564,
+ "step": 9819
+ },
+ {
+ "epoch": 26.757493188010898,
+ "grad_norm": 11.04636287689209,
+ "learning_rate": 1.7182954719168644e-05,
+ "loss": 0.4611,
+ "step": 9820
+ },
+ {
+ "epoch": 26.760217983651227,
+ "grad_norm": 7.629820823669434,
+ "learning_rate": 1.7182340705987493e-05,
+ "loss": 0.4377,
+ "step": 9821
+ },
+ {
+ "epoch": 26.762942779291553,
+ "grad_norm": 6.659231185913086,
+ "learning_rate": 1.7181726636870503e-05,
+ "loss": 0.4788,
+ "step": 9822
+ },
+ {
+ "epoch": 26.76566757493188,
+ "grad_norm": 7.286716938018799,
+ "learning_rate": 1.7181112511822454e-05,
+ "loss": 0.4264,
+ "step": 9823
+ },
+ {
+ "epoch": 26.768392370572208,
+ "grad_norm": 6.088977813720703,
+ "learning_rate": 1.7180498330848136e-05,
+ "loss": 0.6226,
+ "step": 9824
+ },
+ {
+ "epoch": 26.771117166212534,
+ "grad_norm": 7.253275394439697,
+ "learning_rate": 1.7179884093952323e-05,
+ "loss": 0.3702,
+ "step": 9825
+ },
+ {
+ "epoch": 26.77384196185286,
+ "grad_norm": 7.373318195343018,
+ "learning_rate": 1.717926980113981e-05,
+ "loss": 0.3958,
+ "step": 9826
+ },
+ {
+ "epoch": 26.77656675749319,
+ "grad_norm": 6.174760341644287,
+ "learning_rate": 1.717865545241537e-05,
+ "loss": 0.4348,
+ "step": 9827
+ },
+ {
+ "epoch": 26.779291553133515,
+ "grad_norm": 5.556024074554443,
+ "learning_rate": 1.7178041047783794e-05,
+ "loss": 0.4401,
+ "step": 9828
+ },
+ {
+ "epoch": 26.78201634877384,
+ "grad_norm": 7.6754374504089355,
+ "learning_rate": 1.7177426587249867e-05,
+ "loss": 0.4122,
+ "step": 9829
+ },
+ {
+ "epoch": 26.78474114441417,
+ "grad_norm": 6.726299285888672,
+ "learning_rate": 1.717681207081837e-05,
+ "loss": 0.3368,
+ "step": 9830
+ },
+ {
+ "epoch": 26.787465940054496,
+ "grad_norm": 5.833086967468262,
+ "learning_rate": 1.717619749849409e-05,
+ "loss": 0.4646,
+ "step": 9831
+ },
+ {
+ "epoch": 26.79019073569482,
+ "grad_norm": 6.659196376800537,
+ "learning_rate": 1.717558287028182e-05,
+ "loss": 0.524,
+ "step": 9832
+ },
+ {
+ "epoch": 26.79291553133515,
+ "grad_norm": 6.900911808013916,
+ "learning_rate": 1.717496818618634e-05,
+ "loss": 0.4461,
+ "step": 9833
+ },
+ {
+ "epoch": 26.795640326975477,
+ "grad_norm": 6.478199005126953,
+ "learning_rate": 1.7174353446212438e-05,
+ "loss": 0.343,
+ "step": 9834
+ },
+ {
+ "epoch": 26.798365122615802,
+ "grad_norm": 7.359694957733154,
+ "learning_rate": 1.71737386503649e-05,
+ "loss": 0.4348,
+ "step": 9835
+ },
+ {
+ "epoch": 26.80108991825613,
+ "grad_norm": 8.83171272277832,
+ "learning_rate": 1.7173123798648517e-05,
+ "loss": 0.5356,
+ "step": 9836
+ },
+ {
+ "epoch": 26.803814713896458,
+ "grad_norm": 6.10063362121582,
+ "learning_rate": 1.717250889106808e-05,
+ "loss": 0.3399,
+ "step": 9837
+ },
+ {
+ "epoch": 26.806539509536783,
+ "grad_norm": 6.662333011627197,
+ "learning_rate": 1.7171893927628374e-05,
+ "loss": 0.5104,
+ "step": 9838
+ },
+ {
+ "epoch": 26.809264305177113,
+ "grad_norm": 7.070944309234619,
+ "learning_rate": 1.7171278908334188e-05,
+ "loss": 0.5454,
+ "step": 9839
+ },
+ {
+ "epoch": 26.81198910081744,
+ "grad_norm": 8.581802368164062,
+ "learning_rate": 1.717066383319031e-05,
+ "loss": 0.4466,
+ "step": 9840
+ },
+ {
+ "epoch": 26.814713896457764,
+ "grad_norm": 11.248456001281738,
+ "learning_rate": 1.7170048702201536e-05,
+ "loss": 0.6162,
+ "step": 9841
+ },
+ {
+ "epoch": 26.817438692098094,
+ "grad_norm": 5.493569850921631,
+ "learning_rate": 1.7169433515372652e-05,
+ "loss": 0.4262,
+ "step": 9842
+ },
+ {
+ "epoch": 26.82016348773842,
+ "grad_norm": 6.751622200012207,
+ "learning_rate": 1.716881827270845e-05,
+ "loss": 0.3821,
+ "step": 9843
+ },
+ {
+ "epoch": 26.822888283378745,
+ "grad_norm": 5.713374614715576,
+ "learning_rate": 1.7168202974213723e-05,
+ "loss": 0.4545,
+ "step": 9844
+ },
+ {
+ "epoch": 26.825613079019075,
+ "grad_norm": 5.525545120239258,
+ "learning_rate": 1.7167587619893266e-05,
+ "loss": 0.3255,
+ "step": 9845
+ },
+ {
+ "epoch": 26.8283378746594,
+ "grad_norm": 8.607173919677734,
+ "learning_rate": 1.7166972209751863e-05,
+ "loss": 0.5545,
+ "step": 9846
+ },
+ {
+ "epoch": 26.831062670299726,
+ "grad_norm": 5.944239139556885,
+ "learning_rate": 1.716635674379431e-05,
+ "loss": 0.4122,
+ "step": 9847
+ },
+ {
+ "epoch": 26.833787465940055,
+ "grad_norm": 6.472640037536621,
+ "learning_rate": 1.7165741222025404e-05,
+ "loss": 0.368,
+ "step": 9848
+ },
+ {
+ "epoch": 26.83651226158038,
+ "grad_norm": 6.018770694732666,
+ "learning_rate": 1.7165125644449935e-05,
+ "loss": 0.3604,
+ "step": 9849
+ },
+ {
+ "epoch": 26.839237057220707,
+ "grad_norm": 5.767523765563965,
+ "learning_rate": 1.7164510011072698e-05,
+ "loss": 0.3447,
+ "step": 9850
+ },
+ {
+ "epoch": 26.841961852861036,
+ "grad_norm": 6.949464797973633,
+ "learning_rate": 1.7163894321898484e-05,
+ "loss": 0.3396,
+ "step": 9851
+ },
+ {
+ "epoch": 26.844686648501362,
+ "grad_norm": 7.876722812652588,
+ "learning_rate": 1.7163278576932096e-05,
+ "loss": 0.5427,
+ "step": 9852
+ },
+ {
+ "epoch": 26.847411444141688,
+ "grad_norm": 6.1353759765625,
+ "learning_rate": 1.7162662776178326e-05,
+ "loss": 0.3046,
+ "step": 9853
+ },
+ {
+ "epoch": 26.850136239782017,
+ "grad_norm": 5.048327445983887,
+ "learning_rate": 1.7162046919641965e-05,
+ "loss": 0.4042,
+ "step": 9854
+ },
+ {
+ "epoch": 26.852861035422343,
+ "grad_norm": 5.276171684265137,
+ "learning_rate": 1.7161431007327817e-05,
+ "loss": 0.2927,
+ "step": 9855
+ },
+ {
+ "epoch": 26.85558583106267,
+ "grad_norm": 6.662628173828125,
+ "learning_rate": 1.7160815039240673e-05,
+ "loss": 0.357,
+ "step": 9856
+ },
+ {
+ "epoch": 26.858310626703,
+ "grad_norm": 6.133718967437744,
+ "learning_rate": 1.716019901538533e-05,
+ "loss": 0.5529,
+ "step": 9857
+ },
+ {
+ "epoch": 26.861035422343324,
+ "grad_norm": 6.5695576667785645,
+ "learning_rate": 1.715958293576659e-05,
+ "loss": 0.559,
+ "step": 9858
+ },
+ {
+ "epoch": 26.86376021798365,
+ "grad_norm": 5.893548488616943,
+ "learning_rate": 1.7158966800389248e-05,
+ "loss": 0.4602,
+ "step": 9859
+ },
+ {
+ "epoch": 26.86648501362398,
+ "grad_norm": 6.913093090057373,
+ "learning_rate": 1.7158350609258102e-05,
+ "loss": 0.4647,
+ "step": 9860
+ },
+ {
+ "epoch": 26.869209809264305,
+ "grad_norm": 5.447824001312256,
+ "learning_rate": 1.7157734362377954e-05,
+ "loss": 0.4779,
+ "step": 9861
+ },
+ {
+ "epoch": 26.87193460490463,
+ "grad_norm": 4.7569098472595215,
+ "learning_rate": 1.71571180597536e-05,
+ "loss": 0.4321,
+ "step": 9862
+ },
+ {
+ "epoch": 26.87465940054496,
+ "grad_norm": 6.814856052398682,
+ "learning_rate": 1.715650170138984e-05,
+ "loss": 0.4279,
+ "step": 9863
+ },
+ {
+ "epoch": 26.877384196185286,
+ "grad_norm": 5.9102559089660645,
+ "learning_rate": 1.7155885287291477e-05,
+ "loss": 0.4078,
+ "step": 9864
+ },
+ {
+ "epoch": 26.88010899182561,
+ "grad_norm": 6.099296569824219,
+ "learning_rate": 1.715526881746331e-05,
+ "loss": 0.3528,
+ "step": 9865
+ },
+ {
+ "epoch": 26.88283378746594,
+ "grad_norm": 5.032665252685547,
+ "learning_rate": 1.7154652291910134e-05,
+ "loss": 0.3933,
+ "step": 9866
+ },
+ {
+ "epoch": 26.885558583106267,
+ "grad_norm": 5.6574835777282715,
+ "learning_rate": 1.7154035710636764e-05,
+ "loss": 0.3091,
+ "step": 9867
+ },
+ {
+ "epoch": 26.888283378746593,
+ "grad_norm": 6.175564765930176,
+ "learning_rate": 1.715341907364799e-05,
+ "loss": 0.4059,
+ "step": 9868
+ },
+ {
+ "epoch": 26.891008174386922,
+ "grad_norm": 6.385797500610352,
+ "learning_rate": 1.7152802380948618e-05,
+ "loss": 0.3319,
+ "step": 9869
+ },
+ {
+ "epoch": 26.893732970027248,
+ "grad_norm": 5.189164638519287,
+ "learning_rate": 1.7152185632543453e-05,
+ "loss": 0.3748,
+ "step": 9870
+ },
+ {
+ "epoch": 26.896457765667574,
+ "grad_norm": 6.051988124847412,
+ "learning_rate": 1.7151568828437294e-05,
+ "loss": 0.3259,
+ "step": 9871
+ },
+ {
+ "epoch": 26.899182561307903,
+ "grad_norm": 9.091777801513672,
+ "learning_rate": 1.7150951968634947e-05,
+ "loss": 0.4872,
+ "step": 9872
+ },
+ {
+ "epoch": 26.90190735694823,
+ "grad_norm": 5.828537940979004,
+ "learning_rate": 1.715033505314122e-05,
+ "loss": 0.3523,
+ "step": 9873
+ },
+ {
+ "epoch": 26.904632152588555,
+ "grad_norm": 5.783144950866699,
+ "learning_rate": 1.7149718081960912e-05,
+ "loss": 0.4985,
+ "step": 9874
+ },
+ {
+ "epoch": 26.907356948228884,
+ "grad_norm": 5.220401763916016,
+ "learning_rate": 1.714910105509883e-05,
+ "loss": 0.3933,
+ "step": 9875
+ },
+ {
+ "epoch": 26.91008174386921,
+ "grad_norm": 6.512011528015137,
+ "learning_rate": 1.7148483972559778e-05,
+ "loss": 0.3698,
+ "step": 9876
+ },
+ {
+ "epoch": 26.912806539509535,
+ "grad_norm": 5.555382251739502,
+ "learning_rate": 1.7147866834348567e-05,
+ "loss": 0.5542,
+ "step": 9877
+ },
+ {
+ "epoch": 26.915531335149865,
+ "grad_norm": 5.456238269805908,
+ "learning_rate": 1.7147249640469995e-05,
+ "loss": 0.5287,
+ "step": 9878
+ },
+ {
+ "epoch": 26.91825613079019,
+ "grad_norm": 5.542177677154541,
+ "learning_rate": 1.7146632390928874e-05,
+ "loss": 0.3628,
+ "step": 9879
+ },
+ {
+ "epoch": 26.920980926430516,
+ "grad_norm": 5.5648627281188965,
+ "learning_rate": 1.7146015085730006e-05,
+ "loss": 0.332,
+ "step": 9880
+ },
+ {
+ "epoch": 26.923705722070846,
+ "grad_norm": 6.356338024139404,
+ "learning_rate": 1.7145397724878208e-05,
+ "loss": 0.4455,
+ "step": 9881
+ },
+ {
+ "epoch": 26.92643051771117,
+ "grad_norm": 5.476496696472168,
+ "learning_rate": 1.7144780308378283e-05,
+ "loss": 0.2975,
+ "step": 9882
+ },
+ {
+ "epoch": 26.929155313351497,
+ "grad_norm": 6.191989898681641,
+ "learning_rate": 1.7144162836235036e-05,
+ "loss": 0.4046,
+ "step": 9883
+ },
+ {
+ "epoch": 26.931880108991827,
+ "grad_norm": 7.239715576171875,
+ "learning_rate": 1.7143545308453274e-05,
+ "loss": 0.5102,
+ "step": 9884
+ },
+ {
+ "epoch": 26.934604904632153,
+ "grad_norm": 5.949123859405518,
+ "learning_rate": 1.714292772503782e-05,
+ "loss": 0.3895,
+ "step": 9885
+ },
+ {
+ "epoch": 26.93732970027248,
+ "grad_norm": 9.60324478149414,
+ "learning_rate": 1.714231008599347e-05,
+ "loss": 0.418,
+ "step": 9886
+ },
+ {
+ "epoch": 26.940054495912808,
+ "grad_norm": 6.960704803466797,
+ "learning_rate": 1.7141692391325038e-05,
+ "loss": 0.4978,
+ "step": 9887
+ },
+ {
+ "epoch": 26.942779291553133,
+ "grad_norm": 5.399795055389404,
+ "learning_rate": 1.714107464103734e-05,
+ "loss": 0.4772,
+ "step": 9888
+ },
+ {
+ "epoch": 26.94550408719346,
+ "grad_norm": 6.030450820922852,
+ "learning_rate": 1.714045683513518e-05,
+ "loss": 0.3835,
+ "step": 9889
+ },
+ {
+ "epoch": 26.94822888283379,
+ "grad_norm": 5.271973133087158,
+ "learning_rate": 1.7139838973623372e-05,
+ "loss": 0.3285,
+ "step": 9890
+ },
+ {
+ "epoch": 26.950953678474114,
+ "grad_norm": 7.310559272766113,
+ "learning_rate": 1.7139221056506728e-05,
+ "loss": 0.3952,
+ "step": 9891
+ },
+ {
+ "epoch": 26.95367847411444,
+ "grad_norm": 5.800814628601074,
+ "learning_rate": 1.7138603083790058e-05,
+ "loss": 0.3549,
+ "step": 9892
+ },
+ {
+ "epoch": 26.95640326975477,
+ "grad_norm": 5.85589599609375,
+ "learning_rate": 1.713798505547818e-05,
+ "loss": 0.3412,
+ "step": 9893
+ },
+ {
+ "epoch": 26.959128065395095,
+ "grad_norm": 6.218149185180664,
+ "learning_rate": 1.7137366971575905e-05,
+ "loss": 0.4436,
+ "step": 9894
+ },
+ {
+ "epoch": 26.96185286103542,
+ "grad_norm": 5.7112202644348145,
+ "learning_rate": 1.7136748832088043e-05,
+ "loss": 0.3037,
+ "step": 9895
+ },
+ {
+ "epoch": 26.96457765667575,
+ "grad_norm": 5.4224138259887695,
+ "learning_rate": 1.713613063701941e-05,
+ "loss": 0.3983,
+ "step": 9896
+ },
+ {
+ "epoch": 26.967302452316076,
+ "grad_norm": 5.269515514373779,
+ "learning_rate": 1.7135512386374823e-05,
+ "loss": 0.6104,
+ "step": 9897
+ },
+ {
+ "epoch": 26.970027247956402,
+ "grad_norm": 5.6309709548950195,
+ "learning_rate": 1.7134894080159096e-05,
+ "loss": 0.5439,
+ "step": 9898
+ },
+ {
+ "epoch": 26.97275204359673,
+ "grad_norm": 5.971057415008545,
+ "learning_rate": 1.7134275718377043e-05,
+ "loss": 0.3646,
+ "step": 9899
+ },
+ {
+ "epoch": 26.975476839237057,
+ "grad_norm": 5.0091376304626465,
+ "learning_rate": 1.713365730103348e-05,
+ "loss": 0.5311,
+ "step": 9900
+ },
+ {
+ "epoch": 26.978201634877383,
+ "grad_norm": 6.561567783355713,
+ "learning_rate": 1.7133038828133224e-05,
+ "loss": 0.3407,
+ "step": 9901
+ },
+ {
+ "epoch": 26.980926430517712,
+ "grad_norm": 6.122241973876953,
+ "learning_rate": 1.713242029968109e-05,
+ "loss": 0.4955,
+ "step": 9902
+ },
+ {
+ "epoch": 26.983651226158038,
+ "grad_norm": 5.375112056732178,
+ "learning_rate": 1.71318017156819e-05,
+ "loss": 0.3343,
+ "step": 9903
+ },
+ {
+ "epoch": 26.986376021798364,
+ "grad_norm": 5.568964004516602,
+ "learning_rate": 1.7131183076140462e-05,
+ "loss": 0.5515,
+ "step": 9904
+ },
+ {
+ "epoch": 26.989100817438693,
+ "grad_norm": 5.834478855133057,
+ "learning_rate": 1.7130564381061603e-05,
+ "loss": 0.3448,
+ "step": 9905
+ },
+ {
+ "epoch": 26.99182561307902,
+ "grad_norm": 7.519752502441406,
+ "learning_rate": 1.7129945630450136e-05,
+ "loss": 0.326,
+ "step": 9906
+ },
+ {
+ "epoch": 26.994550408719345,
+ "grad_norm": 5.384719371795654,
+ "learning_rate": 1.7129326824310882e-05,
+ "loss": 0.5793,
+ "step": 9907
+ },
+ {
+ "epoch": 26.997275204359674,
+ "grad_norm": 4.788144588470459,
+ "learning_rate": 1.7128707962648662e-05,
+ "loss": 0.287,
+ "step": 9908
+ },
+ {
+ "epoch": 27.0,
+ "grad_norm": 5.651982307434082,
+ "learning_rate": 1.7128089045468294e-05,
+ "loss": 0.3273,
+ "step": 9909
+ },
+ {
+ "epoch": 27.002724795640326,
+ "grad_norm": 6.532353401184082,
+ "learning_rate": 1.71274700727746e-05,
+ "loss": 0.437,
+ "step": 9910
+ },
+ {
+ "epoch": 27.005449591280655,
+ "grad_norm": 5.968925952911377,
+ "learning_rate": 1.7126851044572395e-05,
+ "loss": 0.4691,
+ "step": 9911
+ },
+ {
+ "epoch": 27.00817438692098,
+ "grad_norm": 5.829761028289795,
+ "learning_rate": 1.7126231960866505e-05,
+ "loss": 0.3541,
+ "step": 9912
+ },
+ {
+ "epoch": 27.010899182561307,
+ "grad_norm": 5.907785892486572,
+ "learning_rate": 1.7125612821661746e-05,
+ "loss": 0.3253,
+ "step": 9913
+ },
+ {
+ "epoch": 27.013623978201636,
+ "grad_norm": 4.717392921447754,
+ "learning_rate": 1.7124993626962946e-05,
+ "loss": 0.3831,
+ "step": 9914
+ },
+ {
+ "epoch": 27.016348773841962,
+ "grad_norm": 4.384026527404785,
+ "learning_rate": 1.7124374376774926e-05,
+ "loss": 0.528,
+ "step": 9915
+ },
+ {
+ "epoch": 27.019073569482288,
+ "grad_norm": 5.901668548583984,
+ "learning_rate": 1.712375507110251e-05,
+ "loss": 0.506,
+ "step": 9916
+ },
+ {
+ "epoch": 27.021798365122617,
+ "grad_norm": 5.34744119644165,
+ "learning_rate": 1.7123135709950515e-05,
+ "loss": 0.3662,
+ "step": 9917
+ },
+ {
+ "epoch": 27.024523160762943,
+ "grad_norm": 5.025190830230713,
+ "learning_rate": 1.7122516293323767e-05,
+ "loss": 0.3506,
+ "step": 9918
+ },
+ {
+ "epoch": 27.02724795640327,
+ "grad_norm": 4.95759391784668,
+ "learning_rate": 1.7121896821227094e-05,
+ "loss": 0.2999,
+ "step": 9919
+ },
+ {
+ "epoch": 27.029972752043598,
+ "grad_norm": 5.441646099090576,
+ "learning_rate": 1.712127729366532e-05,
+ "loss": 0.3341,
+ "step": 9920
+ },
+ {
+ "epoch": 27.032697547683924,
+ "grad_norm": 5.002205848693848,
+ "learning_rate": 1.7120657710643264e-05,
+ "loss": 0.4816,
+ "step": 9921
+ },
+ {
+ "epoch": 27.03542234332425,
+ "grad_norm": 7.033825874328613,
+ "learning_rate": 1.7120038072165756e-05,
+ "loss": 0.3462,
+ "step": 9922
+ },
+ {
+ "epoch": 27.03814713896458,
+ "grad_norm": 4.98377799987793,
+ "learning_rate": 1.7119418378237623e-05,
+ "loss": 0.416,
+ "step": 9923
+ },
+ {
+ "epoch": 27.040871934604905,
+ "grad_norm": 5.869990825653076,
+ "learning_rate": 1.7118798628863685e-05,
+ "loss": 0.4081,
+ "step": 9924
+ },
+ {
+ "epoch": 27.04359673024523,
+ "grad_norm": 6.792593955993652,
+ "learning_rate": 1.7118178824048774e-05,
+ "loss": 0.4229,
+ "step": 9925
+ },
+ {
+ "epoch": 27.04632152588556,
+ "grad_norm": 4.7826433181762695,
+ "learning_rate": 1.7117558963797715e-05,
+ "loss": 0.2914,
+ "step": 9926
+ },
+ {
+ "epoch": 27.049046321525886,
+ "grad_norm": 4.066246509552002,
+ "learning_rate": 1.7116939048115336e-05,
+ "loss": 0.2584,
+ "step": 9927
+ },
+ {
+ "epoch": 27.05177111716621,
+ "grad_norm": 5.365878105163574,
+ "learning_rate": 1.711631907700647e-05,
+ "loss": 0.2938,
+ "step": 9928
+ },
+ {
+ "epoch": 27.05449591280654,
+ "grad_norm": 9.358704566955566,
+ "learning_rate": 1.711569905047593e-05,
+ "loss": 0.5488,
+ "step": 9929
+ },
+ {
+ "epoch": 27.057220708446867,
+ "grad_norm": 5.152641296386719,
+ "learning_rate": 1.7115078968528563e-05,
+ "loss": 0.4046,
+ "step": 9930
+ },
+ {
+ "epoch": 27.059945504087192,
+ "grad_norm": 6.016860008239746,
+ "learning_rate": 1.7114458831169184e-05,
+ "loss": 0.3737,
+ "step": 9931
+ },
+ {
+ "epoch": 27.06267029972752,
+ "grad_norm": 5.843413829803467,
+ "learning_rate": 1.7113838638402633e-05,
+ "loss": 0.3139,
+ "step": 9932
+ },
+ {
+ "epoch": 27.065395095367847,
+ "grad_norm": 4.6083083152771,
+ "learning_rate": 1.7113218390233734e-05,
+ "loss": 0.4467,
+ "step": 9933
+ },
+ {
+ "epoch": 27.068119891008173,
+ "grad_norm": 6.329434394836426,
+ "learning_rate": 1.711259808666732e-05,
+ "loss": 0.3575,
+ "step": 9934
+ },
+ {
+ "epoch": 27.070844686648503,
+ "grad_norm": 5.806003570556641,
+ "learning_rate": 1.711197772770822e-05,
+ "loss": 0.2532,
+ "step": 9935
+ },
+ {
+ "epoch": 27.07356948228883,
+ "grad_norm": 6.547754287719727,
+ "learning_rate": 1.7111357313361264e-05,
+ "loss": 0.2673,
+ "step": 9936
+ },
+ {
+ "epoch": 27.076294277929154,
+ "grad_norm": 4.18019437789917,
+ "learning_rate": 1.7110736843631294e-05,
+ "loss": 0.4905,
+ "step": 9937
+ },
+ {
+ "epoch": 27.079019073569484,
+ "grad_norm": 4.6429829597473145,
+ "learning_rate": 1.7110116318523126e-05,
+ "loss": 0.3816,
+ "step": 9938
+ },
+ {
+ "epoch": 27.08174386920981,
+ "grad_norm": 8.305445671081543,
+ "learning_rate": 1.7109495738041605e-05,
+ "loss": 0.3471,
+ "step": 9939
+ },
+ {
+ "epoch": 27.084468664850135,
+ "grad_norm": 8.329751968383789,
+ "learning_rate": 1.7108875102191557e-05,
+ "loss": 0.2662,
+ "step": 9940
+ },
+ {
+ "epoch": 27.087193460490465,
+ "grad_norm": 5.953464984893799,
+ "learning_rate": 1.7108254410977822e-05,
+ "loss": 0.3427,
+ "step": 9941
+ },
+ {
+ "epoch": 27.08991825613079,
+ "grad_norm": 5.11924409866333,
+ "learning_rate": 1.710763366440523e-05,
+ "loss": 0.2774,
+ "step": 9942
+ },
+ {
+ "epoch": 27.092643051771116,
+ "grad_norm": 8.909274101257324,
+ "learning_rate": 1.7107012862478614e-05,
+ "loss": 0.2374,
+ "step": 9943
+ },
+ {
+ "epoch": 27.095367847411445,
+ "grad_norm": 4.669586181640625,
+ "learning_rate": 1.710639200520281e-05,
+ "loss": 0.2718,
+ "step": 9944
+ },
+ {
+ "epoch": 27.09809264305177,
+ "grad_norm": 5.3440093994140625,
+ "learning_rate": 1.7105771092582655e-05,
+ "loss": 0.3531,
+ "step": 9945
+ },
+ {
+ "epoch": 27.100817438692097,
+ "grad_norm": 5.6332688331604,
+ "learning_rate": 1.7105150124622985e-05,
+ "loss": 0.461,
+ "step": 9946
+ },
+ {
+ "epoch": 27.103542234332426,
+ "grad_norm": 5.569468975067139,
+ "learning_rate": 1.7104529101328632e-05,
+ "loss": 0.4098,
+ "step": 9947
+ },
+ {
+ "epoch": 27.106267029972752,
+ "grad_norm": 5.5289154052734375,
+ "learning_rate": 1.7103908022704436e-05,
+ "loss": 0.3031,
+ "step": 9948
+ },
+ {
+ "epoch": 27.108991825613078,
+ "grad_norm": 5.1774396896362305,
+ "learning_rate": 1.7103286888755237e-05,
+ "loss": 0.3968,
+ "step": 9949
+ },
+ {
+ "epoch": 27.111716621253407,
+ "grad_norm": 5.872607231140137,
+ "learning_rate": 1.710266569948586e-05,
+ "loss": 0.3927,
+ "step": 9950
+ },
+ {
+ "epoch": 27.114441416893733,
+ "grad_norm": 3.9197275638580322,
+ "learning_rate": 1.7102044454901155e-05,
+ "loss": 0.3073,
+ "step": 9951
+ },
+ {
+ "epoch": 27.11716621253406,
+ "grad_norm": 4.936652183532715,
+ "learning_rate": 1.7101423155005957e-05,
+ "loss": 0.3452,
+ "step": 9952
+ },
+ {
+ "epoch": 27.11989100817439,
+ "grad_norm": 4.71969747543335,
+ "learning_rate": 1.7100801799805104e-05,
+ "loss": 0.4089,
+ "step": 9953
+ },
+ {
+ "epoch": 27.122615803814714,
+ "grad_norm": 5.414344787597656,
+ "learning_rate": 1.7100180389303435e-05,
+ "loss": 0.2846,
+ "step": 9954
+ },
+ {
+ "epoch": 27.12534059945504,
+ "grad_norm": 5.840177059173584,
+ "learning_rate": 1.7099558923505792e-05,
+ "loss": 0.3899,
+ "step": 9955
+ },
+ {
+ "epoch": 27.12806539509537,
+ "grad_norm": 3.9657328128814697,
+ "learning_rate": 1.709893740241701e-05,
+ "loss": 0.2659,
+ "step": 9956
+ },
+ {
+ "epoch": 27.130790190735695,
+ "grad_norm": 4.62798547744751,
+ "learning_rate": 1.7098315826041932e-05,
+ "loss": 0.2752,
+ "step": 9957
+ },
+ {
+ "epoch": 27.13351498637602,
+ "grad_norm": 5.268348217010498,
+ "learning_rate": 1.7097694194385395e-05,
+ "loss": 0.4811,
+ "step": 9958
+ },
+ {
+ "epoch": 27.13623978201635,
+ "grad_norm": 4.997903347015381,
+ "learning_rate": 1.709707250745225e-05,
+ "loss": 0.2761,
+ "step": 9959
+ },
+ {
+ "epoch": 27.138964577656676,
+ "grad_norm": 5.543661594390869,
+ "learning_rate": 1.7096450765247333e-05,
+ "loss": 0.4148,
+ "step": 9960
+ },
+ {
+ "epoch": 27.141689373297,
+ "grad_norm": 4.802408218383789,
+ "learning_rate": 1.7095828967775485e-05,
+ "loss": 0.2964,
+ "step": 9961
+ },
+ {
+ "epoch": 27.14441416893733,
+ "grad_norm": 5.131856918334961,
+ "learning_rate": 1.7095207115041546e-05,
+ "loss": 0.37,
+ "step": 9962
+ },
+ {
+ "epoch": 27.147138964577657,
+ "grad_norm": 6.031793117523193,
+ "learning_rate": 1.7094585207050363e-05,
+ "loss": 0.4574,
+ "step": 9963
+ },
+ {
+ "epoch": 27.149863760217983,
+ "grad_norm": 4.540724754333496,
+ "learning_rate": 1.709396324380678e-05,
+ "loss": 0.2865,
+ "step": 9964
+ },
+ {
+ "epoch": 27.152588555858312,
+ "grad_norm": 4.215776443481445,
+ "learning_rate": 1.709334122531564e-05,
+ "loss": 0.4566,
+ "step": 9965
+ },
+ {
+ "epoch": 27.155313351498638,
+ "grad_norm": 5.023232460021973,
+ "learning_rate": 1.7092719151581786e-05,
+ "loss": 0.3592,
+ "step": 9966
+ },
+ {
+ "epoch": 27.158038147138964,
+ "grad_norm": 4.531785488128662,
+ "learning_rate": 1.7092097022610064e-05,
+ "loss": 0.2422,
+ "step": 9967
+ },
+ {
+ "epoch": 27.160762942779293,
+ "grad_norm": 5.287015438079834,
+ "learning_rate": 1.709147483840532e-05,
+ "loss": 0.3608,
+ "step": 9968
+ },
+ {
+ "epoch": 27.16348773841962,
+ "grad_norm": 4.872899532318115,
+ "learning_rate": 1.7090852598972398e-05,
+ "loss": 0.3399,
+ "step": 9969
+ },
+ {
+ "epoch": 27.166212534059945,
+ "grad_norm": 3.8876864910125732,
+ "learning_rate": 1.7090230304316142e-05,
+ "loss": 0.3239,
+ "step": 9970
+ },
+ {
+ "epoch": 27.168937329700274,
+ "grad_norm": 5.806251525878906,
+ "learning_rate": 1.7089607954441403e-05,
+ "loss": 0.3427,
+ "step": 9971
+ },
+ {
+ "epoch": 27.1716621253406,
+ "grad_norm": 3.985478162765503,
+ "learning_rate": 1.7088985549353024e-05,
+ "loss": 0.2488,
+ "step": 9972
+ },
+ {
+ "epoch": 27.174386920980925,
+ "grad_norm": 5.465435028076172,
+ "learning_rate": 1.7088363089055854e-05,
+ "loss": 0.3214,
+ "step": 9973
+ },
+ {
+ "epoch": 27.177111716621255,
+ "grad_norm": 5.451912879943848,
+ "learning_rate": 1.708774057355474e-05,
+ "loss": 0.5836,
+ "step": 9974
+ },
+ {
+ "epoch": 27.17983651226158,
+ "grad_norm": 3.8994202613830566,
+ "learning_rate": 1.708711800285453e-05,
+ "loss": 0.2742,
+ "step": 9975
+ },
+ {
+ "epoch": 27.182561307901906,
+ "grad_norm": 5.447590351104736,
+ "learning_rate": 1.7086495376960073e-05,
+ "loss": 0.4318,
+ "step": 9976
+ },
+ {
+ "epoch": 27.185286103542236,
+ "grad_norm": 6.573159217834473,
+ "learning_rate": 1.708587269587622e-05,
+ "loss": 0.2899,
+ "step": 9977
+ },
+ {
+ "epoch": 27.18801089918256,
+ "grad_norm": 5.6273627281188965,
+ "learning_rate": 1.7085249959607818e-05,
+ "loss": 0.4948,
+ "step": 9978
+ },
+ {
+ "epoch": 27.190735694822887,
+ "grad_norm": 5.153083801269531,
+ "learning_rate": 1.7084627168159716e-05,
+ "loss": 0.5888,
+ "step": 9979
+ },
+ {
+ "epoch": 27.193460490463217,
+ "grad_norm": 6.318135738372803,
+ "learning_rate": 1.708400432153677e-05,
+ "loss": 0.3296,
+ "step": 9980
+ },
+ {
+ "epoch": 27.196185286103542,
+ "grad_norm": 4.645571708679199,
+ "learning_rate": 1.7083381419743817e-05,
+ "loss": 0.2847,
+ "step": 9981
+ },
+ {
+ "epoch": 27.19891008174387,
+ "grad_norm": 5.035417079925537,
+ "learning_rate": 1.7082758462785728e-05,
+ "loss": 0.4161,
+ "step": 9982
+ },
+ {
+ "epoch": 27.201634877384198,
+ "grad_norm": 4.9795026779174805,
+ "learning_rate": 1.7082135450667335e-05,
+ "loss": 0.4096,
+ "step": 9983
+ },
+ {
+ "epoch": 27.204359673024523,
+ "grad_norm": 5.218931198120117,
+ "learning_rate": 1.7081512383393507e-05,
+ "loss": 0.3114,
+ "step": 9984
+ },
+ {
+ "epoch": 27.20708446866485,
+ "grad_norm": 5.069077968597412,
+ "learning_rate": 1.7080889260969083e-05,
+ "loss": 0.4161,
+ "step": 9985
+ },
+ {
+ "epoch": 27.20980926430518,
+ "grad_norm": 5.229343891143799,
+ "learning_rate": 1.708026608339892e-05,
+ "loss": 0.2799,
+ "step": 9986
+ },
+ {
+ "epoch": 27.212534059945504,
+ "grad_norm": 5.315517902374268,
+ "learning_rate": 1.7079642850687877e-05,
+ "loss": 0.4067,
+ "step": 9987
+ },
+ {
+ "epoch": 27.21525885558583,
+ "grad_norm": 6.2057204246521,
+ "learning_rate": 1.7079019562840805e-05,
+ "loss": 0.4612,
+ "step": 9988
+ },
+ {
+ "epoch": 27.21798365122616,
+ "grad_norm": 4.244253635406494,
+ "learning_rate": 1.7078396219862555e-05,
+ "loss": 0.4487,
+ "step": 9989
+ },
+ {
+ "epoch": 27.220708446866485,
+ "grad_norm": 4.930257797241211,
+ "learning_rate": 1.707777282175798e-05,
+ "loss": 0.3609,
+ "step": 9990
+ },
+ {
+ "epoch": 27.22343324250681,
+ "grad_norm": 5.732973575592041,
+ "learning_rate": 1.707714936853194e-05,
+ "loss": 0.319,
+ "step": 9991
+ },
+ {
+ "epoch": 27.22615803814714,
+ "grad_norm": 4.790039539337158,
+ "learning_rate": 1.7076525860189287e-05,
+ "loss": 0.448,
+ "step": 9992
+ },
+ {
+ "epoch": 27.228882833787466,
+ "grad_norm": 5.01780366897583,
+ "learning_rate": 1.707590229673488e-05,
+ "loss": 0.2178,
+ "step": 9993
+ },
+ {
+ "epoch": 27.231607629427792,
+ "grad_norm": 5.023081302642822,
+ "learning_rate": 1.7075278678173574e-05,
+ "loss": 0.2717,
+ "step": 9994
+ },
+ {
+ "epoch": 27.23433242506812,
+ "grad_norm": 5.330355167388916,
+ "learning_rate": 1.7074655004510225e-05,
+ "loss": 0.2859,
+ "step": 9995
+ },
+ {
+ "epoch": 27.237057220708447,
+ "grad_norm": 5.243969917297363,
+ "learning_rate": 1.7074031275749686e-05,
+ "loss": 0.4065,
+ "step": 9996
+ },
+ {
+ "epoch": 27.239782016348773,
+ "grad_norm": 6.028010368347168,
+ "learning_rate": 1.7073407491896824e-05,
+ "loss": 0.354,
+ "step": 9997
+ },
+ {
+ "epoch": 27.242506811989102,
+ "grad_norm": 5.086205959320068,
+ "learning_rate": 1.7072783652956493e-05,
+ "loss": 0.3713,
+ "step": 9998
+ },
+ {
+ "epoch": 27.245231607629428,
+ "grad_norm": 4.849929332733154,
+ "learning_rate": 1.707215975893355e-05,
+ "loss": 0.3431,
+ "step": 9999
+ },
+ {
+ "epoch": 27.247956403269754,
+ "grad_norm": 5.67385721206665,
+ "learning_rate": 1.7071535809832852e-05,
+ "loss": 0.3865,
+ "step": 10000
+ },
+ {
+ "epoch": 27.250681198910083,
+ "grad_norm": 6.044569492340088,
+ "learning_rate": 1.707091180565926e-05,
+ "loss": 0.267,
+ "step": 10001
+ },
+ {
+ "epoch": 27.25340599455041,
+ "grad_norm": 6.110007286071777,
+ "learning_rate": 1.7070287746417636e-05,
+ "loss": 0.3218,
+ "step": 10002
+ },
+ {
+ "epoch": 27.256130790190735,
+ "grad_norm": 6.86362886428833,
+ "learning_rate": 1.706966363211284e-05,
+ "loss": 0.3984,
+ "step": 10003
+ },
+ {
+ "epoch": 27.258855585831064,
+ "grad_norm": 5.410529136657715,
+ "learning_rate": 1.7069039462749734e-05,
+ "loss": 0.4008,
+ "step": 10004
+ },
+ {
+ "epoch": 27.26158038147139,
+ "grad_norm": 15.645142555236816,
+ "learning_rate": 1.706841523833317e-05,
+ "loss": 0.4326,
+ "step": 10005
+ },
+ {
+ "epoch": 27.264305177111716,
+ "grad_norm": 14.088157653808594,
+ "learning_rate": 1.7067790958868017e-05,
+ "loss": 0.4409,
+ "step": 10006
+ },
+ {
+ "epoch": 27.267029972752045,
+ "grad_norm": 5.084339141845703,
+ "learning_rate": 1.7067166624359138e-05,
+ "loss": 0.2227,
+ "step": 10007
+ },
+ {
+ "epoch": 27.26975476839237,
+ "grad_norm": 5.2032060623168945,
+ "learning_rate": 1.706654223481139e-05,
+ "loss": 0.3451,
+ "step": 10008
+ },
+ {
+ "epoch": 27.272479564032697,
+ "grad_norm": 5.205188274383545,
+ "learning_rate": 1.706591779022964e-05,
+ "loss": 0.4133,
+ "step": 10009
+ },
+ {
+ "epoch": 27.275204359673026,
+ "grad_norm": 4.178635597229004,
+ "learning_rate": 1.706529329061875e-05,
+ "loss": 0.3994,
+ "step": 10010
+ },
+ {
+ "epoch": 27.277929155313352,
+ "grad_norm": 6.104649543762207,
+ "learning_rate": 1.706466873598358e-05,
+ "loss": 0.2564,
+ "step": 10011
+ },
+ {
+ "epoch": 27.280653950953678,
+ "grad_norm": 4.22910737991333,
+ "learning_rate": 1.7064044126329e-05,
+ "loss": 0.2565,
+ "step": 10012
+ },
+ {
+ "epoch": 27.283378746594007,
+ "grad_norm": 5.521817207336426,
+ "learning_rate": 1.706341946165987e-05,
+ "loss": 0.3365,
+ "step": 10013
+ },
+ {
+ "epoch": 27.286103542234333,
+ "grad_norm": 4.440884113311768,
+ "learning_rate": 1.706279474198106e-05,
+ "loss": 0.2472,
+ "step": 10014
+ },
+ {
+ "epoch": 27.28882833787466,
+ "grad_norm": 5.5350260734558105,
+ "learning_rate": 1.7062169967297432e-05,
+ "loss": 0.2972,
+ "step": 10015
+ },
+ {
+ "epoch": 27.291553133514988,
+ "grad_norm": 5.036799430847168,
+ "learning_rate": 1.7061545137613847e-05,
+ "loss": 0.5827,
+ "step": 10016
+ },
+ {
+ "epoch": 27.294277929155314,
+ "grad_norm": 4.988478183746338,
+ "learning_rate": 1.706092025293518e-05,
+ "loss": 0.2178,
+ "step": 10017
+ },
+ {
+ "epoch": 27.29700272479564,
+ "grad_norm": 6.175372123718262,
+ "learning_rate": 1.7060295313266292e-05,
+ "loss": 0.3397,
+ "step": 10018
+ },
+ {
+ "epoch": 27.29972752043597,
+ "grad_norm": 5.248056888580322,
+ "learning_rate": 1.7059670318612053e-05,
+ "loss": 0.2004,
+ "step": 10019
+ },
+ {
+ "epoch": 27.302452316076295,
+ "grad_norm": 5.2282867431640625,
+ "learning_rate": 1.7059045268977325e-05,
+ "loss": 0.29,
+ "step": 10020
+ },
+ {
+ "epoch": 27.30517711171662,
+ "grad_norm": 4.7777204513549805,
+ "learning_rate": 1.705842016436698e-05,
+ "loss": 0.3834,
+ "step": 10021
+ },
+ {
+ "epoch": 27.30790190735695,
+ "grad_norm": 4.574029445648193,
+ "learning_rate": 1.705779500478589e-05,
+ "loss": 0.4147,
+ "step": 10022
+ },
+ {
+ "epoch": 27.310626702997276,
+ "grad_norm": 5.310520172119141,
+ "learning_rate": 1.7057169790238916e-05,
+ "loss": 0.3155,
+ "step": 10023
+ },
+ {
+ "epoch": 27.3133514986376,
+ "grad_norm": 6.366519451141357,
+ "learning_rate": 1.705654452073093e-05,
+ "loss": 0.4001,
+ "step": 10024
+ },
+ {
+ "epoch": 27.31607629427793,
+ "grad_norm": 5.041665554046631,
+ "learning_rate": 1.7055919196266806e-05,
+ "loss": 0.3954,
+ "step": 10025
+ },
+ {
+ "epoch": 27.318801089918257,
+ "grad_norm": 6.168102264404297,
+ "learning_rate": 1.705529381685141e-05,
+ "loss": 0.3109,
+ "step": 10026
+ },
+ {
+ "epoch": 27.321525885558582,
+ "grad_norm": 5.068386554718018,
+ "learning_rate": 1.705466838248961e-05,
+ "loss": 0.3686,
+ "step": 10027
+ },
+ {
+ "epoch": 27.32425068119891,
+ "grad_norm": 4.801697731018066,
+ "learning_rate": 1.705404289318628e-05,
+ "loss": 0.3047,
+ "step": 10028
+ },
+ {
+ "epoch": 27.326975476839237,
+ "grad_norm": 5.894521713256836,
+ "learning_rate": 1.7053417348946293e-05,
+ "loss": 0.3841,
+ "step": 10029
+ },
+ {
+ "epoch": 27.329700272479563,
+ "grad_norm": 4.669535160064697,
+ "learning_rate": 1.705279174977452e-05,
+ "loss": 0.287,
+ "step": 10030
+ },
+ {
+ "epoch": 27.332425068119893,
+ "grad_norm": 4.643848896026611,
+ "learning_rate": 1.7052166095675827e-05,
+ "loss": 0.4261,
+ "step": 10031
+ },
+ {
+ "epoch": 27.33514986376022,
+ "grad_norm": 4.92600679397583,
+ "learning_rate": 1.7051540386655094e-05,
+ "loss": 0.2913,
+ "step": 10032
+ },
+ {
+ "epoch": 27.337874659400544,
+ "grad_norm": 7.221554279327393,
+ "learning_rate": 1.705091462271719e-05,
+ "loss": 0.4932,
+ "step": 10033
+ },
+ {
+ "epoch": 27.340599455040874,
+ "grad_norm": 5.536395072937012,
+ "learning_rate": 1.7050288803866992e-05,
+ "loss": 0.4837,
+ "step": 10034
+ },
+ {
+ "epoch": 27.3433242506812,
+ "grad_norm": 5.581630706787109,
+ "learning_rate": 1.7049662930109373e-05,
+ "loss": 0.4682,
+ "step": 10035
+ },
+ {
+ "epoch": 27.346049046321525,
+ "grad_norm": 5.505551338195801,
+ "learning_rate": 1.7049037001449202e-05,
+ "loss": 0.355,
+ "step": 10036
+ },
+ {
+ "epoch": 27.348773841961854,
+ "grad_norm": 5.493169784545898,
+ "learning_rate": 1.704841101789136e-05,
+ "loss": 0.3004,
+ "step": 10037
+ },
+ {
+ "epoch": 27.35149863760218,
+ "grad_norm": 4.911767959594727,
+ "learning_rate": 1.704778497944072e-05,
+ "loss": 0.2961,
+ "step": 10038
+ },
+ {
+ "epoch": 27.354223433242506,
+ "grad_norm": 5.603453159332275,
+ "learning_rate": 1.704715888610216e-05,
+ "loss": 0.2933,
+ "step": 10039
+ },
+ {
+ "epoch": 27.356948228882835,
+ "grad_norm": 4.990756988525391,
+ "learning_rate": 1.704653273788055e-05,
+ "loss": 0.5237,
+ "step": 10040
+ },
+ {
+ "epoch": 27.35967302452316,
+ "grad_norm": 5.234003067016602,
+ "learning_rate": 1.704590653478077e-05,
+ "loss": 0.4423,
+ "step": 10041
+ },
+ {
+ "epoch": 27.362397820163487,
+ "grad_norm": 6.7180047035217285,
+ "learning_rate": 1.7045280276807697e-05,
+ "loss": 0.4488,
+ "step": 10042
+ },
+ {
+ "epoch": 27.365122615803816,
+ "grad_norm": 5.32593297958374,
+ "learning_rate": 1.7044653963966206e-05,
+ "loss": 0.5555,
+ "step": 10043
+ },
+ {
+ "epoch": 27.367847411444142,
+ "grad_norm": 9.9306001663208,
+ "learning_rate": 1.7044027596261178e-05,
+ "loss": 0.3595,
+ "step": 10044
+ },
+ {
+ "epoch": 27.370572207084468,
+ "grad_norm": 5.757988929748535,
+ "learning_rate": 1.704340117369749e-05,
+ "loss": 0.2549,
+ "step": 10045
+ },
+ {
+ "epoch": 27.373297002724797,
+ "grad_norm": 5.238148212432861,
+ "learning_rate": 1.7042774696280023e-05,
+ "loss": 0.644,
+ "step": 10046
+ },
+ {
+ "epoch": 27.376021798365123,
+ "grad_norm": 5.699423313140869,
+ "learning_rate": 1.704214816401365e-05,
+ "loss": 0.4741,
+ "step": 10047
+ },
+ {
+ "epoch": 27.37874659400545,
+ "grad_norm": 4.643774032592773,
+ "learning_rate": 1.7041521576903254e-05,
+ "loss": 0.3265,
+ "step": 10048
+ },
+ {
+ "epoch": 27.381471389645778,
+ "grad_norm": 4.8292412757873535,
+ "learning_rate": 1.7040894934953715e-05,
+ "loss": 0.4151,
+ "step": 10049
+ },
+ {
+ "epoch": 27.384196185286104,
+ "grad_norm": 4.763095855712891,
+ "learning_rate": 1.7040268238169916e-05,
+ "loss": 0.3074,
+ "step": 10050
+ },
+ {
+ "epoch": 27.38692098092643,
+ "grad_norm": 5.387937545776367,
+ "learning_rate": 1.7039641486556732e-05,
+ "loss": 0.2842,
+ "step": 10051
+ },
+ {
+ "epoch": 27.38964577656676,
+ "grad_norm": 5.783563137054443,
+ "learning_rate": 1.7039014680119046e-05,
+ "loss": 0.3756,
+ "step": 10052
+ },
+ {
+ "epoch": 27.392370572207085,
+ "grad_norm": 5.080140590667725,
+ "learning_rate": 1.703838781886174e-05,
+ "loss": 0.2933,
+ "step": 10053
+ },
+ {
+ "epoch": 27.39509536784741,
+ "grad_norm": 5.309555530548096,
+ "learning_rate": 1.7037760902789696e-05,
+ "loss": 0.4599,
+ "step": 10054
+ },
+ {
+ "epoch": 27.39782016348774,
+ "grad_norm": 4.236416816711426,
+ "learning_rate": 1.7037133931907797e-05,
+ "loss": 0.3279,
+ "step": 10055
+ },
+ {
+ "epoch": 27.400544959128066,
+ "grad_norm": 5.168328285217285,
+ "learning_rate": 1.7036506906220928e-05,
+ "loss": 0.2792,
+ "step": 10056
+ },
+ {
+ "epoch": 27.40326975476839,
+ "grad_norm": 5.925442218780518,
+ "learning_rate": 1.7035879825733965e-05,
+ "loss": 0.3369,
+ "step": 10057
+ },
+ {
+ "epoch": 27.40599455040872,
+ "grad_norm": 5.654895782470703,
+ "learning_rate": 1.70352526904518e-05,
+ "loss": 0.3733,
+ "step": 10058
+ },
+ {
+ "epoch": 27.408719346049047,
+ "grad_norm": 6.028935432434082,
+ "learning_rate": 1.7034625500379312e-05,
+ "loss": 0.2626,
+ "step": 10059
+ },
+ {
+ "epoch": 27.411444141689373,
+ "grad_norm": 4.383516788482666,
+ "learning_rate": 1.7033998255521388e-05,
+ "loss": 0.5562,
+ "step": 10060
+ },
+ {
+ "epoch": 27.414168937329702,
+ "grad_norm": 4.988570213317871,
+ "learning_rate": 1.703337095588291e-05,
+ "loss": 0.3397,
+ "step": 10061
+ },
+ {
+ "epoch": 27.416893732970028,
+ "grad_norm": 4.905068874359131,
+ "learning_rate": 1.703274360146877e-05,
+ "loss": 0.4252,
+ "step": 10062
+ },
+ {
+ "epoch": 27.419618528610354,
+ "grad_norm": 5.765669822692871,
+ "learning_rate": 1.7032116192283844e-05,
+ "loss": 0.3731,
+ "step": 10063
+ },
+ {
+ "epoch": 27.422343324250683,
+ "grad_norm": 5.245814323425293,
+ "learning_rate": 1.7031488728333025e-05,
+ "loss": 0.2612,
+ "step": 10064
+ },
+ {
+ "epoch": 27.42506811989101,
+ "grad_norm": 4.8864030838012695,
+ "learning_rate": 1.70308612096212e-05,
+ "loss": 0.3139,
+ "step": 10065
+ },
+ {
+ "epoch": 27.427792915531334,
+ "grad_norm": 5.1695942878723145,
+ "learning_rate": 1.703023363615325e-05,
+ "loss": 0.4958,
+ "step": 10066
+ },
+ {
+ "epoch": 27.430517711171664,
+ "grad_norm": 5.44508171081543,
+ "learning_rate": 1.7029606007934073e-05,
+ "loss": 0.2925,
+ "step": 10067
+ },
+ {
+ "epoch": 27.43324250681199,
+ "grad_norm": 5.140865802764893,
+ "learning_rate": 1.7028978324968543e-05,
+ "loss": 0.4796,
+ "step": 10068
+ },
+ {
+ "epoch": 27.435967302452315,
+ "grad_norm": 5.728974342346191,
+ "learning_rate": 1.702835058726156e-05,
+ "loss": 0.332,
+ "step": 10069
+ },
+ {
+ "epoch": 27.438692098092645,
+ "grad_norm": 4.789264678955078,
+ "learning_rate": 1.7027722794818008e-05,
+ "loss": 0.2394,
+ "step": 10070
+ },
+ {
+ "epoch": 27.44141689373297,
+ "grad_norm": 4.577560901641846,
+ "learning_rate": 1.702709494764278e-05,
+ "loss": 0.2845,
+ "step": 10071
+ },
+ {
+ "epoch": 27.444141689373296,
+ "grad_norm": 6.5032477378845215,
+ "learning_rate": 1.7026467045740762e-05,
+ "loss": 0.4089,
+ "step": 10072
+ },
+ {
+ "epoch": 27.446866485013626,
+ "grad_norm": 4.876463413238525,
+ "learning_rate": 1.7025839089116843e-05,
+ "loss": 0.2899,
+ "step": 10073
+ },
+ {
+ "epoch": 27.44959128065395,
+ "grad_norm": 4.325783729553223,
+ "learning_rate": 1.7025211077775915e-05,
+ "loss": 0.2409,
+ "step": 10074
+ },
+ {
+ "epoch": 27.452316076294277,
+ "grad_norm": 5.269494533538818,
+ "learning_rate": 1.702458301172287e-05,
+ "loss": 0.3281,
+ "step": 10075
+ },
+ {
+ "epoch": 27.455040871934607,
+ "grad_norm": 4.481354236602783,
+ "learning_rate": 1.70239548909626e-05,
+ "loss": 0.2202,
+ "step": 10076
+ },
+ {
+ "epoch": 27.457765667574932,
+ "grad_norm": 5.786200046539307,
+ "learning_rate": 1.7023326715499994e-05,
+ "loss": 0.3376,
+ "step": 10077
+ },
+ {
+ "epoch": 27.460490463215258,
+ "grad_norm": 5.412074089050293,
+ "learning_rate": 1.7022698485339945e-05,
+ "loss": 0.4366,
+ "step": 10078
+ },
+ {
+ "epoch": 27.463215258855588,
+ "grad_norm": 4.260880470275879,
+ "learning_rate": 1.7022070200487344e-05,
+ "loss": 0.2972,
+ "step": 10079
+ },
+ {
+ "epoch": 27.465940054495913,
+ "grad_norm": 5.933574676513672,
+ "learning_rate": 1.702144186094709e-05,
+ "loss": 0.3106,
+ "step": 10080
+ },
+ {
+ "epoch": 27.46866485013624,
+ "grad_norm": 5.4546308517456055,
+ "learning_rate": 1.7020813466724072e-05,
+ "loss": 0.2892,
+ "step": 10081
+ },
+ {
+ "epoch": 27.47138964577657,
+ "grad_norm": 4.915536403656006,
+ "learning_rate": 1.7020185017823185e-05,
+ "loss": 0.2462,
+ "step": 10082
+ },
+ {
+ "epoch": 27.474114441416894,
+ "grad_norm": 7.712029933929443,
+ "learning_rate": 1.7019556514249323e-05,
+ "loss": 0.3048,
+ "step": 10083
+ },
+ {
+ "epoch": 27.47683923705722,
+ "grad_norm": 5.038845062255859,
+ "learning_rate": 1.701892795600738e-05,
+ "loss": 0.2901,
+ "step": 10084
+ },
+ {
+ "epoch": 27.479564032697546,
+ "grad_norm": 4.911445617675781,
+ "learning_rate": 1.7018299343102252e-05,
+ "loss": 0.2652,
+ "step": 10085
+ },
+ {
+ "epoch": 27.482288828337875,
+ "grad_norm": 5.114807605743408,
+ "learning_rate": 1.7017670675538834e-05,
+ "loss": 0.3145,
+ "step": 10086
+ },
+ {
+ "epoch": 27.4850136239782,
+ "grad_norm": 7.453251361846924,
+ "learning_rate": 1.7017041953322025e-05,
+ "loss": 0.2471,
+ "step": 10087
+ },
+ {
+ "epoch": 27.48773841961853,
+ "grad_norm": 4.8490986824035645,
+ "learning_rate": 1.7016413176456717e-05,
+ "loss": 0.3456,
+ "step": 10088
+ },
+ {
+ "epoch": 27.490463215258856,
+ "grad_norm": 4.292917251586914,
+ "learning_rate": 1.701578434494781e-05,
+ "loss": 0.2744,
+ "step": 10089
+ },
+ {
+ "epoch": 27.493188010899182,
+ "grad_norm": 4.817429542541504,
+ "learning_rate": 1.70151554588002e-05,
+ "loss": 0.3901,
+ "step": 10090
+ },
+ {
+ "epoch": 27.495912806539508,
+ "grad_norm": 4.526362419128418,
+ "learning_rate": 1.7014526518018787e-05,
+ "loss": 0.2747,
+ "step": 10091
+ },
+ {
+ "epoch": 27.498637602179837,
+ "grad_norm": 4.509451389312744,
+ "learning_rate": 1.7013897522608464e-05,
+ "loss": 0.3055,
+ "step": 10092
+ },
+ {
+ "epoch": 27.501362397820163,
+ "grad_norm": 6.948440074920654,
+ "learning_rate": 1.7013268472574133e-05,
+ "loss": 0.4374,
+ "step": 10093
+ },
+ {
+ "epoch": 27.504087193460492,
+ "grad_norm": 4.874791622161865,
+ "learning_rate": 1.7012639367920694e-05,
+ "loss": 0.2955,
+ "step": 10094
+ },
+ {
+ "epoch": 27.506811989100818,
+ "grad_norm": 5.496635913848877,
+ "learning_rate": 1.7012010208653044e-05,
+ "loss": 0.3595,
+ "step": 10095
+ },
+ {
+ "epoch": 27.509536784741144,
+ "grad_norm": 5.246387481689453,
+ "learning_rate": 1.7011380994776086e-05,
+ "loss": 0.3528,
+ "step": 10096
+ },
+ {
+ "epoch": 27.51226158038147,
+ "grad_norm": 4.93645715713501,
+ "learning_rate": 1.701075172629472e-05,
+ "loss": 0.3052,
+ "step": 10097
+ },
+ {
+ "epoch": 27.5149863760218,
+ "grad_norm": 5.026438236236572,
+ "learning_rate": 1.701012240321384e-05,
+ "loss": 0.2834,
+ "step": 10098
+ },
+ {
+ "epoch": 27.517711171662125,
+ "grad_norm": 6.180301189422607,
+ "learning_rate": 1.7009493025538356e-05,
+ "loss": 0.3714,
+ "step": 10099
+ },
+ {
+ "epoch": 27.520435967302454,
+ "grad_norm": 5.171062469482422,
+ "learning_rate": 1.7008863593273167e-05,
+ "loss": 0.2974,
+ "step": 10100
+ },
+ {
+ "epoch": 27.52316076294278,
+ "grad_norm": 4.884304046630859,
+ "learning_rate": 1.7008234106423175e-05,
+ "loss": 0.3545,
+ "step": 10101
+ },
+ {
+ "epoch": 27.525885558583106,
+ "grad_norm": 6.168331146240234,
+ "learning_rate": 1.7007604564993277e-05,
+ "loss": 0.3508,
+ "step": 10102
+ },
+ {
+ "epoch": 27.52861035422343,
+ "grad_norm": 6.211284637451172,
+ "learning_rate": 1.7006974968988382e-05,
+ "loss": 0.4481,
+ "step": 10103
+ },
+ {
+ "epoch": 27.53133514986376,
+ "grad_norm": 5.325413703918457,
+ "learning_rate": 1.700634531841339e-05,
+ "loss": 0.3448,
+ "step": 10104
+ },
+ {
+ "epoch": 27.534059945504087,
+ "grad_norm": 5.873477935791016,
+ "learning_rate": 1.7005715613273206e-05,
+ "loss": 0.2296,
+ "step": 10105
+ },
+ {
+ "epoch": 27.536784741144416,
+ "grad_norm": 4.89999532699585,
+ "learning_rate": 1.7005085853572736e-05,
+ "loss": 0.3757,
+ "step": 10106
+ },
+ {
+ "epoch": 27.539509536784742,
+ "grad_norm": 3.9019815921783447,
+ "learning_rate": 1.7004456039316883e-05,
+ "loss": 0.2444,
+ "step": 10107
+ },
+ {
+ "epoch": 27.542234332425068,
+ "grad_norm": 4.57985258102417,
+ "learning_rate": 1.700382617051055e-05,
+ "loss": 0.3573,
+ "step": 10108
+ },
+ {
+ "epoch": 27.544959128065393,
+ "grad_norm": 5.33076810836792,
+ "learning_rate": 1.700319624715865e-05,
+ "loss": 0.2482,
+ "step": 10109
+ },
+ {
+ "epoch": 27.547683923705723,
+ "grad_norm": 4.6094794273376465,
+ "learning_rate": 1.7002566269266078e-05,
+ "loss": 0.2002,
+ "step": 10110
+ },
+ {
+ "epoch": 27.55040871934605,
+ "grad_norm": 6.188515663146973,
+ "learning_rate": 1.7001936236837743e-05,
+ "loss": 0.4299,
+ "step": 10111
+ },
+ {
+ "epoch": 27.553133514986374,
+ "grad_norm": 11.234764099121094,
+ "learning_rate": 1.700130614987856e-05,
+ "loss": 0.3802,
+ "step": 10112
+ },
+ {
+ "epoch": 27.555858310626704,
+ "grad_norm": 5.274439334869385,
+ "learning_rate": 1.700067600839342e-05,
+ "loss": 0.2372,
+ "step": 10113
+ },
+ {
+ "epoch": 27.55858310626703,
+ "grad_norm": 4.657918930053711,
+ "learning_rate": 1.7000045812387253e-05,
+ "loss": 0.2866,
+ "step": 10114
+ },
+ {
+ "epoch": 27.561307901907355,
+ "grad_norm": 5.632442474365234,
+ "learning_rate": 1.6999415561864946e-05,
+ "loss": 0.3322,
+ "step": 10115
+ },
+ {
+ "epoch": 27.564032697547685,
+ "grad_norm": 5.518496036529541,
+ "learning_rate": 1.699878525683142e-05,
+ "loss": 0.4108,
+ "step": 10116
+ },
+ {
+ "epoch": 27.56675749318801,
+ "grad_norm": 6.064355850219727,
+ "learning_rate": 1.699815489729158e-05,
+ "loss": 0.306,
+ "step": 10117
+ },
+ {
+ "epoch": 27.569482288828336,
+ "grad_norm": 5.1227545738220215,
+ "learning_rate": 1.699752448325033e-05,
+ "loss": 0.3076,
+ "step": 10118
+ },
+ {
+ "epoch": 27.572207084468666,
+ "grad_norm": 4.863996982574463,
+ "learning_rate": 1.699689401471259e-05,
+ "loss": 0.331,
+ "step": 10119
+ },
+ {
+ "epoch": 27.57493188010899,
+ "grad_norm": 4.6978440284729,
+ "learning_rate": 1.6996263491683263e-05,
+ "loss": 0.3618,
+ "step": 10120
+ },
+ {
+ "epoch": 27.577656675749317,
+ "grad_norm": 4.251979351043701,
+ "learning_rate": 1.699563291416726e-05,
+ "loss": 0.2686,
+ "step": 10121
+ },
+ {
+ "epoch": 27.580381471389646,
+ "grad_norm": 5.008840560913086,
+ "learning_rate": 1.699500228216949e-05,
+ "loss": 0.3066,
+ "step": 10122
+ },
+ {
+ "epoch": 27.583106267029972,
+ "grad_norm": 5.299466133117676,
+ "learning_rate": 1.6994371595694875e-05,
+ "loss": 0.3311,
+ "step": 10123
+ },
+ {
+ "epoch": 27.585831062670298,
+ "grad_norm": 4.630898475646973,
+ "learning_rate": 1.6993740854748315e-05,
+ "loss": 0.3452,
+ "step": 10124
+ },
+ {
+ "epoch": 27.588555858310627,
+ "grad_norm": 5.981410026550293,
+ "learning_rate": 1.6993110059334725e-05,
+ "loss": 0.369,
+ "step": 10125
+ },
+ {
+ "epoch": 27.591280653950953,
+ "grad_norm": 5.396410942077637,
+ "learning_rate": 1.699247920945902e-05,
+ "loss": 0.595,
+ "step": 10126
+ },
+ {
+ "epoch": 27.59400544959128,
+ "grad_norm": 34.171974182128906,
+ "learning_rate": 1.6991848305126113e-05,
+ "loss": 0.2452,
+ "step": 10127
+ },
+ {
+ "epoch": 27.59673024523161,
+ "grad_norm": 5.18298864364624,
+ "learning_rate": 1.6991217346340917e-05,
+ "loss": 0.2806,
+ "step": 10128
+ },
+ {
+ "epoch": 27.599455040871934,
+ "grad_norm": 4.584865570068359,
+ "learning_rate": 1.6990586333108342e-05,
+ "loss": 0.3193,
+ "step": 10129
+ },
+ {
+ "epoch": 27.60217983651226,
+ "grad_norm": 5.4814629554748535,
+ "learning_rate": 1.698995526543331e-05,
+ "loss": 0.3326,
+ "step": 10130
+ },
+ {
+ "epoch": 27.60490463215259,
+ "grad_norm": 5.602553844451904,
+ "learning_rate": 1.698932414332073e-05,
+ "loss": 0.4404,
+ "step": 10131
+ },
+ {
+ "epoch": 27.607629427792915,
+ "grad_norm": 4.443238258361816,
+ "learning_rate": 1.6988692966775515e-05,
+ "loss": 0.2841,
+ "step": 10132
+ },
+ {
+ "epoch": 27.61035422343324,
+ "grad_norm": 4.786772727966309,
+ "learning_rate": 1.6988061735802587e-05,
+ "loss": 0.3412,
+ "step": 10133
+ },
+ {
+ "epoch": 27.61307901907357,
+ "grad_norm": 6.136547565460205,
+ "learning_rate": 1.6987430450406857e-05,
+ "loss": 0.3142,
+ "step": 10134
+ },
+ {
+ "epoch": 27.615803814713896,
+ "grad_norm": 4.482685565948486,
+ "learning_rate": 1.698679911059325e-05,
+ "loss": 0.4558,
+ "step": 10135
+ },
+ {
+ "epoch": 27.618528610354222,
+ "grad_norm": 4.903445243835449,
+ "learning_rate": 1.698616771636667e-05,
+ "loss": 0.3307,
+ "step": 10136
+ },
+ {
+ "epoch": 27.62125340599455,
+ "grad_norm": 4.520987510681152,
+ "learning_rate": 1.698553626773204e-05,
+ "loss": 0.3405,
+ "step": 10137
+ },
+ {
+ "epoch": 27.623978201634877,
+ "grad_norm": 5.721822261810303,
+ "learning_rate": 1.6984904764694282e-05,
+ "loss": 0.3094,
+ "step": 10138
+ },
+ {
+ "epoch": 27.626702997275203,
+ "grad_norm": 4.86265754699707,
+ "learning_rate": 1.6984273207258312e-05,
+ "loss": 0.3151,
+ "step": 10139
+ },
+ {
+ "epoch": 27.629427792915532,
+ "grad_norm": 4.779309272766113,
+ "learning_rate": 1.698364159542904e-05,
+ "loss": 0.2783,
+ "step": 10140
+ },
+ {
+ "epoch": 27.632152588555858,
+ "grad_norm": 5.118013381958008,
+ "learning_rate": 1.6983009929211397e-05,
+ "loss": 0.2587,
+ "step": 10141
+ },
+ {
+ "epoch": 27.634877384196184,
+ "grad_norm": 5.287686824798584,
+ "learning_rate": 1.6982378208610298e-05,
+ "loss": 0.3038,
+ "step": 10142
+ },
+ {
+ "epoch": 27.637602179836513,
+ "grad_norm": 4.768444061279297,
+ "learning_rate": 1.698174643363066e-05,
+ "loss": 0.3416,
+ "step": 10143
+ },
+ {
+ "epoch": 27.64032697547684,
+ "grad_norm": 5.341958045959473,
+ "learning_rate": 1.698111460427741e-05,
+ "loss": 0.3171,
+ "step": 10144
+ },
+ {
+ "epoch": 27.643051771117165,
+ "grad_norm": 5.178884983062744,
+ "learning_rate": 1.6980482720555457e-05,
+ "loss": 0.3542,
+ "step": 10145
+ },
+ {
+ "epoch": 27.645776566757494,
+ "grad_norm": 5.123482704162598,
+ "learning_rate": 1.6979850782469734e-05,
+ "loss": 0.2336,
+ "step": 10146
+ },
+ {
+ "epoch": 27.64850136239782,
+ "grad_norm": 5.512136459350586,
+ "learning_rate": 1.6979218790025157e-05,
+ "loss": 0.3117,
+ "step": 10147
+ },
+ {
+ "epoch": 27.651226158038146,
+ "grad_norm": 4.252807140350342,
+ "learning_rate": 1.6978586743226646e-05,
+ "loss": 0.3023,
+ "step": 10148
+ },
+ {
+ "epoch": 27.653950953678475,
+ "grad_norm": 4.892229080200195,
+ "learning_rate": 1.697795464207913e-05,
+ "loss": 0.2248,
+ "step": 10149
+ },
+ {
+ "epoch": 27.6566757493188,
+ "grad_norm": 5.369317054748535,
+ "learning_rate": 1.6977322486587527e-05,
+ "loss": 0.3015,
+ "step": 10150
+ },
+ {
+ "epoch": 27.659400544959126,
+ "grad_norm": 5.784381866455078,
+ "learning_rate": 1.697669027675676e-05,
+ "loss": 0.3367,
+ "step": 10151
+ },
+ {
+ "epoch": 27.662125340599456,
+ "grad_norm": 5.423375129699707,
+ "learning_rate": 1.6976058012591755e-05,
+ "loss": 0.3097,
+ "step": 10152
+ },
+ {
+ "epoch": 27.66485013623978,
+ "grad_norm": 5.225542068481445,
+ "learning_rate": 1.697542569409743e-05,
+ "loss": 0.3458,
+ "step": 10153
+ },
+ {
+ "epoch": 27.667574931880107,
+ "grad_norm": 5.198846340179443,
+ "learning_rate": 1.697479332127872e-05,
+ "loss": 0.5257,
+ "step": 10154
+ },
+ {
+ "epoch": 27.670299727520437,
+ "grad_norm": 5.464980602264404,
+ "learning_rate": 1.697416089414054e-05,
+ "loss": 0.4137,
+ "step": 10155
+ },
+ {
+ "epoch": 27.673024523160763,
+ "grad_norm": 5.154241561889648,
+ "learning_rate": 1.6973528412687824e-05,
+ "loss": 0.4446,
+ "step": 10156
+ },
+ {
+ "epoch": 27.67574931880109,
+ "grad_norm": 5.381682395935059,
+ "learning_rate": 1.697289587692549e-05,
+ "loss": 0.3635,
+ "step": 10157
+ },
+ {
+ "epoch": 27.678474114441418,
+ "grad_norm": 7.300772190093994,
+ "learning_rate": 1.6972263286858467e-05,
+ "loss": 0.2684,
+ "step": 10158
+ },
+ {
+ "epoch": 27.681198910081743,
+ "grad_norm": 5.311357498168945,
+ "learning_rate": 1.6971630642491683e-05,
+ "loss": 0.267,
+ "step": 10159
+ },
+ {
+ "epoch": 27.68392370572207,
+ "grad_norm": 4.842854976654053,
+ "learning_rate": 1.6970997943830062e-05,
+ "loss": 0.221,
+ "step": 10160
+ },
+ {
+ "epoch": 27.6866485013624,
+ "grad_norm": 5.6701483726501465,
+ "learning_rate": 1.6970365190878536e-05,
+ "loss": 0.2942,
+ "step": 10161
+ },
+ {
+ "epoch": 27.689373297002724,
+ "grad_norm": 4.846139907836914,
+ "learning_rate": 1.696973238364203e-05,
+ "loss": 0.2726,
+ "step": 10162
+ },
+ {
+ "epoch": 27.69209809264305,
+ "grad_norm": 5.639392852783203,
+ "learning_rate": 1.6969099522125475e-05,
+ "loss": 0.407,
+ "step": 10163
+ },
+ {
+ "epoch": 27.69482288828338,
+ "grad_norm": 4.361480712890625,
+ "learning_rate": 1.6968466606333792e-05,
+ "loss": 0.2668,
+ "step": 10164
+ },
+ {
+ "epoch": 27.697547683923705,
+ "grad_norm": 5.761633396148682,
+ "learning_rate": 1.6967833636271916e-05,
+ "loss": 0.3477,
+ "step": 10165
+ },
+ {
+ "epoch": 27.70027247956403,
+ "grad_norm": 4.269412994384766,
+ "learning_rate": 1.6967200611944776e-05,
+ "loss": 0.2573,
+ "step": 10166
+ },
+ {
+ "epoch": 27.70299727520436,
+ "grad_norm": 4.878856658935547,
+ "learning_rate": 1.69665675333573e-05,
+ "loss": 0.351,
+ "step": 10167
+ },
+ {
+ "epoch": 27.705722070844686,
+ "grad_norm": 5.025777816772461,
+ "learning_rate": 1.696593440051443e-05,
+ "loss": 0.4036,
+ "step": 10168
+ },
+ {
+ "epoch": 27.708446866485012,
+ "grad_norm": 4.94193172454834,
+ "learning_rate": 1.6965301213421075e-05,
+ "loss": 0.505,
+ "step": 10169
+ },
+ {
+ "epoch": 27.71117166212534,
+ "grad_norm": 6.366814613342285,
+ "learning_rate": 1.6964667972082187e-05,
+ "loss": 0.3371,
+ "step": 10170
+ },
+ {
+ "epoch": 27.713896457765667,
+ "grad_norm": 4.604916095733643,
+ "learning_rate": 1.6964034676502686e-05,
+ "loss": 0.3215,
+ "step": 10171
+ },
+ {
+ "epoch": 27.716621253405993,
+ "grad_norm": 6.2079925537109375,
+ "learning_rate": 1.6963401326687506e-05,
+ "loss": 0.3253,
+ "step": 10172
+ },
+ {
+ "epoch": 27.719346049046322,
+ "grad_norm": 5.239664554595947,
+ "learning_rate": 1.696276792264158e-05,
+ "loss": 0.3401,
+ "step": 10173
+ },
+ {
+ "epoch": 27.722070844686648,
+ "grad_norm": 4.3239545822143555,
+ "learning_rate": 1.6962134464369846e-05,
+ "loss": 0.2795,
+ "step": 10174
+ },
+ {
+ "epoch": 27.724795640326974,
+ "grad_norm": 4.265858173370361,
+ "learning_rate": 1.696150095187723e-05,
+ "loss": 0.2855,
+ "step": 10175
+ },
+ {
+ "epoch": 27.727520435967303,
+ "grad_norm": 3.9349844455718994,
+ "learning_rate": 1.6960867385168668e-05,
+ "loss": 0.2179,
+ "step": 10176
+ },
+ {
+ "epoch": 27.73024523160763,
+ "grad_norm": 4.0783843994140625,
+ "learning_rate": 1.6960233764249095e-05,
+ "loss": 0.3398,
+ "step": 10177
+ },
+ {
+ "epoch": 27.732970027247955,
+ "grad_norm": 5.081615447998047,
+ "learning_rate": 1.6959600089123447e-05,
+ "loss": 0.3744,
+ "step": 10178
+ },
+ {
+ "epoch": 27.735694822888284,
+ "grad_norm": 5.580626010894775,
+ "learning_rate": 1.6958966359796657e-05,
+ "loss": 0.2958,
+ "step": 10179
+ },
+ {
+ "epoch": 27.73841961852861,
+ "grad_norm": 4.565976619720459,
+ "learning_rate": 1.695833257627366e-05,
+ "loss": 0.4792,
+ "step": 10180
+ },
+ {
+ "epoch": 27.741144414168936,
+ "grad_norm": 4.693761825561523,
+ "learning_rate": 1.695769873855939e-05,
+ "loss": 0.2962,
+ "step": 10181
+ },
+ {
+ "epoch": 27.743869209809265,
+ "grad_norm": 4.896193981170654,
+ "learning_rate": 1.695706484665879e-05,
+ "loss": 0.4898,
+ "step": 10182
+ },
+ {
+ "epoch": 27.74659400544959,
+ "grad_norm": 5.1602983474731445,
+ "learning_rate": 1.695643090057679e-05,
+ "loss": 0.3251,
+ "step": 10183
+ },
+ {
+ "epoch": 27.749318801089917,
+ "grad_norm": 4.677697658538818,
+ "learning_rate": 1.6955796900318337e-05,
+ "loss": 0.273,
+ "step": 10184
+ },
+ {
+ "epoch": 27.752043596730246,
+ "grad_norm": 5.536619663238525,
+ "learning_rate": 1.6955162845888354e-05,
+ "loss": 0.326,
+ "step": 10185
+ },
+ {
+ "epoch": 27.754768392370572,
+ "grad_norm": 4.491034507751465,
+ "learning_rate": 1.6954528737291787e-05,
+ "loss": 0.1935,
+ "step": 10186
+ },
+ {
+ "epoch": 27.757493188010898,
+ "grad_norm": 5.130890369415283,
+ "learning_rate": 1.695389457453358e-05,
+ "loss": 0.2246,
+ "step": 10187
+ },
+ {
+ "epoch": 27.760217983651227,
+ "grad_norm": 5.151829719543457,
+ "learning_rate": 1.6953260357618662e-05,
+ "loss": 0.2625,
+ "step": 10188
+ },
+ {
+ "epoch": 27.762942779291553,
+ "grad_norm": 4.85291862487793,
+ "learning_rate": 1.6952626086551977e-05,
+ "loss": 0.2613,
+ "step": 10189
+ },
+ {
+ "epoch": 27.76566757493188,
+ "grad_norm": 9.296930313110352,
+ "learning_rate": 1.6951991761338463e-05,
+ "loss": 0.3222,
+ "step": 10190
+ },
+ {
+ "epoch": 27.768392370572208,
+ "grad_norm": 5.643762588500977,
+ "learning_rate": 1.695135738198306e-05,
+ "loss": 0.332,
+ "step": 10191
+ },
+ {
+ "epoch": 27.771117166212534,
+ "grad_norm": 5.350772857666016,
+ "learning_rate": 1.6950722948490713e-05,
+ "loss": 0.3538,
+ "step": 10192
+ },
+ {
+ "epoch": 27.77384196185286,
+ "grad_norm": 4.740004062652588,
+ "learning_rate": 1.6950088460866357e-05,
+ "loss": 0.2922,
+ "step": 10193
+ },
+ {
+ "epoch": 27.77656675749319,
+ "grad_norm": 6.163573265075684,
+ "learning_rate": 1.694945391911493e-05,
+ "loss": 0.2526,
+ "step": 10194
+ },
+ {
+ "epoch": 27.779291553133515,
+ "grad_norm": 5.353569030761719,
+ "learning_rate": 1.694881932324139e-05,
+ "loss": 0.2899,
+ "step": 10195
+ },
+ {
+ "epoch": 27.78201634877384,
+ "grad_norm": 5.403836250305176,
+ "learning_rate": 1.6948184673250665e-05,
+ "loss": 0.3015,
+ "step": 10196
+ },
+ {
+ "epoch": 27.78474114441417,
+ "grad_norm": 5.395393371582031,
+ "learning_rate": 1.69475499691477e-05,
+ "loss": 0.3483,
+ "step": 10197
+ },
+ {
+ "epoch": 27.787465940054496,
+ "grad_norm": 5.1931376457214355,
+ "learning_rate": 1.694691521093744e-05,
+ "loss": 0.2612,
+ "step": 10198
+ },
+ {
+ "epoch": 27.79019073569482,
+ "grad_norm": 6.095047473907471,
+ "learning_rate": 1.6946280398624828e-05,
+ "loss": 0.3311,
+ "step": 10199
+ },
+ {
+ "epoch": 27.79291553133515,
+ "grad_norm": 4.96342134475708,
+ "learning_rate": 1.6945645532214805e-05,
+ "loss": 0.3921,
+ "step": 10200
+ },
+ {
+ "epoch": 27.795640326975477,
+ "grad_norm": 4.840270042419434,
+ "learning_rate": 1.694501061171232e-05,
+ "loss": 0.2799,
+ "step": 10201
+ },
+ {
+ "epoch": 27.798365122615802,
+ "grad_norm": 5.995936870574951,
+ "learning_rate": 1.6944375637122318e-05,
+ "loss": 0.341,
+ "step": 10202
+ },
+ {
+ "epoch": 27.80108991825613,
+ "grad_norm": 5.229803085327148,
+ "learning_rate": 1.694374060844974e-05,
+ "loss": 0.3386,
+ "step": 10203
+ },
+ {
+ "epoch": 27.803814713896458,
+ "grad_norm": 5.566660404205322,
+ "learning_rate": 1.6943105525699534e-05,
+ "loss": 0.3999,
+ "step": 10204
+ },
+ {
+ "epoch": 27.806539509536783,
+ "grad_norm": 5.201536178588867,
+ "learning_rate": 1.6942470388876645e-05,
+ "loss": 0.4227,
+ "step": 10205
+ },
+ {
+ "epoch": 27.809264305177113,
+ "grad_norm": 5.207836151123047,
+ "learning_rate": 1.6941835197986023e-05,
+ "loss": 0.3451,
+ "step": 10206
+ },
+ {
+ "epoch": 27.81198910081744,
+ "grad_norm": 6.146342754364014,
+ "learning_rate": 1.694119995303261e-05,
+ "loss": 0.3842,
+ "step": 10207
+ },
+ {
+ "epoch": 27.814713896457764,
+ "grad_norm": 5.106663703918457,
+ "learning_rate": 1.6940564654021355e-05,
+ "loss": 0.3528,
+ "step": 10208
+ },
+ {
+ "epoch": 27.817438692098094,
+ "grad_norm": 5.420879364013672,
+ "learning_rate": 1.6939929300957206e-05,
+ "loss": 0.3481,
+ "step": 10209
+ },
+ {
+ "epoch": 27.82016348773842,
+ "grad_norm": 5.758074760437012,
+ "learning_rate": 1.693929389384511e-05,
+ "loss": 0.2897,
+ "step": 10210
+ },
+ {
+ "epoch": 27.822888283378745,
+ "grad_norm": 4.140540599822998,
+ "learning_rate": 1.6938658432690016e-05,
+ "loss": 0.2436,
+ "step": 10211
+ },
+ {
+ "epoch": 27.825613079019075,
+ "grad_norm": 4.473278999328613,
+ "learning_rate": 1.6938022917496876e-05,
+ "loss": 0.2672,
+ "step": 10212
+ },
+ {
+ "epoch": 27.8283378746594,
+ "grad_norm": 4.347257614135742,
+ "learning_rate": 1.6937387348270636e-05,
+ "loss": 0.2085,
+ "step": 10213
+ },
+ {
+ "epoch": 27.831062670299726,
+ "grad_norm": 4.400454521179199,
+ "learning_rate": 1.6936751725016248e-05,
+ "loss": 0.3526,
+ "step": 10214
+ },
+ {
+ "epoch": 27.833787465940055,
+ "grad_norm": 4.264406204223633,
+ "learning_rate": 1.6936116047738657e-05,
+ "loss": 0.3657,
+ "step": 10215
+ },
+ {
+ "epoch": 27.83651226158038,
+ "grad_norm": 4.644855976104736,
+ "learning_rate": 1.6935480316442818e-05,
+ "loss": 0.2845,
+ "step": 10216
+ },
+ {
+ "epoch": 27.839237057220707,
+ "grad_norm": 5.094082355499268,
+ "learning_rate": 1.6934844531133686e-05,
+ "loss": 0.2867,
+ "step": 10217
+ },
+ {
+ "epoch": 27.841961852861036,
+ "grad_norm": 4.733489036560059,
+ "learning_rate": 1.6934208691816203e-05,
+ "loss": 0.2544,
+ "step": 10218
+ },
+ {
+ "epoch": 27.844686648501362,
+ "grad_norm": 5.06028938293457,
+ "learning_rate": 1.6933572798495327e-05,
+ "loss": 0.2559,
+ "step": 10219
+ },
+ {
+ "epoch": 27.847411444141688,
+ "grad_norm": 5.4426188468933105,
+ "learning_rate": 1.6932936851176013e-05,
+ "loss": 0.4555,
+ "step": 10220
+ },
+ {
+ "epoch": 27.850136239782017,
+ "grad_norm": 4.860383033752441,
+ "learning_rate": 1.6932300849863207e-05,
+ "loss": 0.2724,
+ "step": 10221
+ },
+ {
+ "epoch": 27.852861035422343,
+ "grad_norm": 5.712159633636475,
+ "learning_rate": 1.6931664794561862e-05,
+ "loss": 0.3504,
+ "step": 10222
+ },
+ {
+ "epoch": 27.85558583106267,
+ "grad_norm": 5.541167259216309,
+ "learning_rate": 1.693102868527694e-05,
+ "loss": 0.4378,
+ "step": 10223
+ },
+ {
+ "epoch": 27.858310626703,
+ "grad_norm": 4.544509410858154,
+ "learning_rate": 1.693039252201339e-05,
+ "loss": 0.2323,
+ "step": 10224
+ },
+ {
+ "epoch": 27.861035422343324,
+ "grad_norm": 4.7020158767700195,
+ "learning_rate": 1.692975630477616e-05,
+ "loss": 0.5797,
+ "step": 10225
+ },
+ {
+ "epoch": 27.86376021798365,
+ "grad_norm": 8.22085952758789,
+ "learning_rate": 1.6929120033570218e-05,
+ "loss": 0.3912,
+ "step": 10226
+ },
+ {
+ "epoch": 27.86648501362398,
+ "grad_norm": 5.784850120544434,
+ "learning_rate": 1.6928483708400504e-05,
+ "loss": 0.2781,
+ "step": 10227
+ },
+ {
+ "epoch": 27.869209809264305,
+ "grad_norm": 4.1860551834106445,
+ "learning_rate": 1.692784732927199e-05,
+ "loss": 0.3097,
+ "step": 10228
+ },
+ {
+ "epoch": 27.87193460490463,
+ "grad_norm": 5.328650951385498,
+ "learning_rate": 1.692721089618962e-05,
+ "loss": 0.2977,
+ "step": 10229
+ },
+ {
+ "epoch": 27.87465940054496,
+ "grad_norm": 6.5369977951049805,
+ "learning_rate": 1.6926574409158354e-05,
+ "loss": 0.3914,
+ "step": 10230
+ },
+ {
+ "epoch": 27.877384196185286,
+ "grad_norm": 4.641357421875,
+ "learning_rate": 1.692593786818315e-05,
+ "loss": 0.3465,
+ "step": 10231
+ },
+ {
+ "epoch": 27.88010899182561,
+ "grad_norm": 6.705341339111328,
+ "learning_rate": 1.6925301273268965e-05,
+ "loss": 0.3076,
+ "step": 10232
+ },
+ {
+ "epoch": 27.88283378746594,
+ "grad_norm": 6.148820877075195,
+ "learning_rate": 1.6924664624420758e-05,
+ "loss": 0.3047,
+ "step": 10233
+ },
+ {
+ "epoch": 27.885558583106267,
+ "grad_norm": 5.407176971435547,
+ "learning_rate": 1.6924027921643485e-05,
+ "loss": 0.2842,
+ "step": 10234
+ },
+ {
+ "epoch": 27.888283378746593,
+ "grad_norm": 7.05771541595459,
+ "learning_rate": 1.6923391164942106e-05,
+ "loss": 0.4727,
+ "step": 10235
+ },
+ {
+ "epoch": 27.891008174386922,
+ "grad_norm": 5.3796916007995605,
+ "learning_rate": 1.6922754354321577e-05,
+ "loss": 0.5219,
+ "step": 10236
+ },
+ {
+ "epoch": 27.893732970027248,
+ "grad_norm": 4.517141819000244,
+ "learning_rate": 1.6922117489786863e-05,
+ "loss": 0.3971,
+ "step": 10237
+ },
+ {
+ "epoch": 27.896457765667574,
+ "grad_norm": 5.803973197937012,
+ "learning_rate": 1.6921480571342916e-05,
+ "loss": 0.3858,
+ "step": 10238
+ },
+ {
+ "epoch": 27.899182561307903,
+ "grad_norm": 5.765970706939697,
+ "learning_rate": 1.6920843598994707e-05,
+ "loss": 0.2553,
+ "step": 10239
+ },
+ {
+ "epoch": 27.90190735694823,
+ "grad_norm": 7.121570110321045,
+ "learning_rate": 1.6920206572747186e-05,
+ "loss": 0.3111,
+ "step": 10240
+ },
+ {
+ "epoch": 27.904632152588555,
+ "grad_norm": 4.6284589767456055,
+ "learning_rate": 1.691956949260532e-05,
+ "loss": 0.2787,
+ "step": 10241
+ },
+ {
+ "epoch": 27.907356948228884,
+ "grad_norm": 4.878785610198975,
+ "learning_rate": 1.691893235857407e-05,
+ "loss": 0.2762,
+ "step": 10242
+ },
+ {
+ "epoch": 27.91008174386921,
+ "grad_norm": 4.893333911895752,
+ "learning_rate": 1.69182951706584e-05,
+ "loss": 0.3965,
+ "step": 10243
+ },
+ {
+ "epoch": 27.912806539509535,
+ "grad_norm": 5.686673164367676,
+ "learning_rate": 1.6917657928863267e-05,
+ "loss": 0.2924,
+ "step": 10244
+ },
+ {
+ "epoch": 27.915531335149865,
+ "grad_norm": 4.324870586395264,
+ "learning_rate": 1.691702063319364e-05,
+ "loss": 0.2816,
+ "step": 10245
+ },
+ {
+ "epoch": 27.91825613079019,
+ "grad_norm": 5.113224506378174,
+ "learning_rate": 1.6916383283654478e-05,
+ "loss": 0.457,
+ "step": 10246
+ },
+ {
+ "epoch": 27.920980926430516,
+ "grad_norm": 5.34783935546875,
+ "learning_rate": 1.691574588025074e-05,
+ "loss": 0.3546,
+ "step": 10247
+ },
+ {
+ "epoch": 27.923705722070846,
+ "grad_norm": 5.102035045623779,
+ "learning_rate": 1.6915108422987402e-05,
+ "loss": 0.3601,
+ "step": 10248
+ },
+ {
+ "epoch": 27.92643051771117,
+ "grad_norm": 5.083774089813232,
+ "learning_rate": 1.691447091186942e-05,
+ "loss": 0.3471,
+ "step": 10249
+ },
+ {
+ "epoch": 27.929155313351497,
+ "grad_norm": 5.293685436248779,
+ "learning_rate": 1.6913833346901758e-05,
+ "loss": 0.4251,
+ "step": 10250
+ },
+ {
+ "epoch": 27.931880108991827,
+ "grad_norm": 4.96642541885376,
+ "learning_rate": 1.6913195728089388e-05,
+ "loss": 0.3449,
+ "step": 10251
+ },
+ {
+ "epoch": 27.934604904632153,
+ "grad_norm": 8.112710952758789,
+ "learning_rate": 1.6912558055437273e-05,
+ "loss": 0.2983,
+ "step": 10252
+ },
+ {
+ "epoch": 27.93732970027248,
+ "grad_norm": 19.07801628112793,
+ "learning_rate": 1.6911920328950376e-05,
+ "loss": 0.3798,
+ "step": 10253
+ },
+ {
+ "epoch": 27.940054495912808,
+ "grad_norm": 5.400221824645996,
+ "learning_rate": 1.6911282548633664e-05,
+ "loss": 0.5408,
+ "step": 10254
+ },
+ {
+ "epoch": 27.942779291553133,
+ "grad_norm": 5.097553730010986,
+ "learning_rate": 1.691064471449211e-05,
+ "loss": 0.2258,
+ "step": 10255
+ },
+ {
+ "epoch": 27.94550408719346,
+ "grad_norm": 5.449814796447754,
+ "learning_rate": 1.691000682653068e-05,
+ "loss": 0.3215,
+ "step": 10256
+ },
+ {
+ "epoch": 27.94822888283379,
+ "grad_norm": 5.01417875289917,
+ "learning_rate": 1.690936888475433e-05,
+ "loss": 0.2682,
+ "step": 10257
+ },
+ {
+ "epoch": 27.950953678474114,
+ "grad_norm": 4.919515609741211,
+ "learning_rate": 1.6908730889168047e-05,
+ "loss": 0.2378,
+ "step": 10258
+ },
+ {
+ "epoch": 27.95367847411444,
+ "grad_norm": 6.381681442260742,
+ "learning_rate": 1.690809283977678e-05,
+ "loss": 0.4083,
+ "step": 10259
+ },
+ {
+ "epoch": 27.95640326975477,
+ "grad_norm": 4.422179698944092,
+ "learning_rate": 1.6907454736585517e-05,
+ "loss": 0.3703,
+ "step": 10260
+ },
+ {
+ "epoch": 27.959128065395095,
+ "grad_norm": 4.35486364364624,
+ "learning_rate": 1.6906816579599217e-05,
+ "loss": 0.4108,
+ "step": 10261
+ },
+ {
+ "epoch": 27.96185286103542,
+ "grad_norm": 4.714322566986084,
+ "learning_rate": 1.690617836882285e-05,
+ "loss": 0.2712,
+ "step": 10262
+ },
+ {
+ "epoch": 27.96457765667575,
+ "grad_norm": 4.96153450012207,
+ "learning_rate": 1.690554010426139e-05,
+ "loss": 0.3662,
+ "step": 10263
+ },
+ {
+ "epoch": 27.967302452316076,
+ "grad_norm": 5.23045539855957,
+ "learning_rate": 1.6904901785919806e-05,
+ "loss": 0.4942,
+ "step": 10264
+ },
+ {
+ "epoch": 27.970027247956402,
+ "grad_norm": 4.072609901428223,
+ "learning_rate": 1.6904263413803068e-05,
+ "loss": 0.236,
+ "step": 10265
+ },
+ {
+ "epoch": 27.97275204359673,
+ "grad_norm": 4.597963809967041,
+ "learning_rate": 1.6903624987916147e-05,
+ "loss": 0.3163,
+ "step": 10266
+ },
+ {
+ "epoch": 27.975476839237057,
+ "grad_norm": 4.55546236038208,
+ "learning_rate": 1.690298650826402e-05,
+ "loss": 0.3786,
+ "step": 10267
+ },
+ {
+ "epoch": 27.978201634877383,
+ "grad_norm": 6.247783660888672,
+ "learning_rate": 1.6902347974851653e-05,
+ "loss": 0.3364,
+ "step": 10268
+ },
+ {
+ "epoch": 27.980926430517712,
+ "grad_norm": 4.579067230224609,
+ "learning_rate": 1.6901709387684025e-05,
+ "loss": 0.257,
+ "step": 10269
+ },
+ {
+ "epoch": 27.983651226158038,
+ "grad_norm": 4.850199222564697,
+ "learning_rate": 1.6901070746766108e-05,
+ "loss": 0.3183,
+ "step": 10270
+ },
+ {
+ "epoch": 27.986376021798364,
+ "grad_norm": 4.970810890197754,
+ "learning_rate": 1.690043205210287e-05,
+ "loss": 0.3058,
+ "step": 10271
+ },
+ {
+ "epoch": 27.989100817438693,
+ "grad_norm": 6.197716236114502,
+ "learning_rate": 1.689979330369929e-05,
+ "loss": 0.3031,
+ "step": 10272
+ },
+ {
+ "epoch": 27.99182561307902,
+ "grad_norm": 5.586562156677246,
+ "learning_rate": 1.6899154501560344e-05,
+ "loss": 0.2379,
+ "step": 10273
+ },
+ {
+ "epoch": 27.994550408719345,
+ "grad_norm": 4.794958114624023,
+ "learning_rate": 1.6898515645691002e-05,
+ "loss": 0.205,
+ "step": 10274
+ },
+ {
+ "epoch": 27.997275204359674,
+ "grad_norm": 4.432571887969971,
+ "learning_rate": 1.6897876736096247e-05,
+ "loss": 0.3431,
+ "step": 10275
+ },
+ {
+ "epoch": 28.0,
+ "grad_norm": 4.240803241729736,
+ "learning_rate": 1.6897237772781046e-05,
+ "loss": 0.4248,
+ "step": 10276
+ },
+ {
+ "epoch": 28.002724795640326,
+ "grad_norm": 5.93013858795166,
+ "learning_rate": 1.6896598755750376e-05,
+ "loss": 0.4361,
+ "step": 10277
+ },
+ {
+ "epoch": 28.005449591280655,
+ "grad_norm": 4.435560703277588,
+ "learning_rate": 1.689595968500922e-05,
+ "loss": 0.2372,
+ "step": 10278
+ },
+ {
+ "epoch": 28.00817438692098,
+ "grad_norm": 4.387332439422607,
+ "learning_rate": 1.6895320560562554e-05,
+ "loss": 0.3088,
+ "step": 10279
+ },
+ {
+ "epoch": 28.010899182561307,
+ "grad_norm": 8.539623260498047,
+ "learning_rate": 1.6894681382415353e-05,
+ "loss": 0.2962,
+ "step": 10280
+ },
+ {
+ "epoch": 28.013623978201636,
+ "grad_norm": 5.029465675354004,
+ "learning_rate": 1.6894042150572594e-05,
+ "loss": 0.2423,
+ "step": 10281
+ },
+ {
+ "epoch": 28.016348773841962,
+ "grad_norm": 6.758065223693848,
+ "learning_rate": 1.6893402865039257e-05,
+ "loss": 0.3262,
+ "step": 10282
+ },
+ {
+ "epoch": 28.019073569482288,
+ "grad_norm": 4.85852575302124,
+ "learning_rate": 1.689276352582032e-05,
+ "loss": 0.3302,
+ "step": 10283
+ },
+ {
+ "epoch": 28.021798365122617,
+ "grad_norm": 5.148909091949463,
+ "learning_rate": 1.6892124132920763e-05,
+ "loss": 0.2087,
+ "step": 10284
+ },
+ {
+ "epoch": 28.024523160762943,
+ "grad_norm": 4.370366096496582,
+ "learning_rate": 1.6891484686345565e-05,
+ "loss": 0.357,
+ "step": 10285
+ },
+ {
+ "epoch": 28.02724795640327,
+ "grad_norm": 4.710180282592773,
+ "learning_rate": 1.6890845186099705e-05,
+ "loss": 0.529,
+ "step": 10286
+ },
+ {
+ "epoch": 28.029972752043598,
+ "grad_norm": 5.0709004402160645,
+ "learning_rate": 1.6890205632188166e-05,
+ "loss": 0.2461,
+ "step": 10287
+ },
+ {
+ "epoch": 28.032697547683924,
+ "grad_norm": 5.952876567840576,
+ "learning_rate": 1.6889566024615928e-05,
+ "loss": 0.2424,
+ "step": 10288
+ },
+ {
+ "epoch": 28.03542234332425,
+ "grad_norm": 4.099681854248047,
+ "learning_rate": 1.688892636338797e-05,
+ "loss": 0.1684,
+ "step": 10289
+ },
+ {
+ "epoch": 28.03814713896458,
+ "grad_norm": 4.158721923828125,
+ "learning_rate": 1.6888286648509272e-05,
+ "loss": 0.2359,
+ "step": 10290
+ },
+ {
+ "epoch": 28.040871934604905,
+ "grad_norm": 4.780111312866211,
+ "learning_rate": 1.6887646879984826e-05,
+ "loss": 0.2687,
+ "step": 10291
+ },
+ {
+ "epoch": 28.04359673024523,
+ "grad_norm": 4.982387065887451,
+ "learning_rate": 1.6887007057819606e-05,
+ "loss": 0.3914,
+ "step": 10292
+ },
+ {
+ "epoch": 28.04632152588556,
+ "grad_norm": 4.861568450927734,
+ "learning_rate": 1.6886367182018593e-05,
+ "loss": 0.3649,
+ "step": 10293
+ },
+ {
+ "epoch": 28.049046321525886,
+ "grad_norm": 4.5405683517456055,
+ "learning_rate": 1.6885727252586774e-05,
+ "loss": 0.3019,
+ "step": 10294
+ },
+ {
+ "epoch": 28.05177111716621,
+ "grad_norm": 4.119233131408691,
+ "learning_rate": 1.6885087269529136e-05,
+ "loss": 0.5659,
+ "step": 10295
+ },
+ {
+ "epoch": 28.05449591280654,
+ "grad_norm": 5.361600875854492,
+ "learning_rate": 1.6884447232850657e-05,
+ "loss": 0.4324,
+ "step": 10296
+ },
+ {
+ "epoch": 28.057220708446867,
+ "grad_norm": 4.881106853485107,
+ "learning_rate": 1.688380714255633e-05,
+ "loss": 0.4273,
+ "step": 10297
+ },
+ {
+ "epoch": 28.059945504087192,
+ "grad_norm": 5.0557050704956055,
+ "learning_rate": 1.688316699865113e-05,
+ "loss": 0.2545,
+ "step": 10298
+ },
+ {
+ "epoch": 28.06267029972752,
+ "grad_norm": 4.747816562652588,
+ "learning_rate": 1.6882526801140046e-05,
+ "loss": 0.2592,
+ "step": 10299
+ },
+ {
+ "epoch": 28.065395095367847,
+ "grad_norm": 5.393918514251709,
+ "learning_rate": 1.6881886550028068e-05,
+ "loss": 0.1927,
+ "step": 10300
+ },
+ {
+ "epoch": 28.068119891008173,
+ "grad_norm": 4.8607611656188965,
+ "learning_rate": 1.688124624532018e-05,
+ "loss": 0.2247,
+ "step": 10301
+ },
+ {
+ "epoch": 28.070844686648503,
+ "grad_norm": 5.269425392150879,
+ "learning_rate": 1.6880605887021362e-05,
+ "loss": 0.2506,
+ "step": 10302
+ },
+ {
+ "epoch": 28.07356948228883,
+ "grad_norm": 5.2715888023376465,
+ "learning_rate": 1.687996547513661e-05,
+ "loss": 0.3665,
+ "step": 10303
+ },
+ {
+ "epoch": 28.076294277929154,
+ "grad_norm": 5.59796667098999,
+ "learning_rate": 1.6879325009670912e-05,
+ "loss": 0.3442,
+ "step": 10304
+ },
+ {
+ "epoch": 28.079019073569484,
+ "grad_norm": 5.015444755554199,
+ "learning_rate": 1.6878684490629247e-05,
+ "loss": 0.3254,
+ "step": 10305
+ },
+ {
+ "epoch": 28.08174386920981,
+ "grad_norm": 4.151947021484375,
+ "learning_rate": 1.687804391801661e-05,
+ "loss": 0.3462,
+ "step": 10306
+ },
+ {
+ "epoch": 28.084468664850135,
+ "grad_norm": 4.794727802276611,
+ "learning_rate": 1.687740329183799e-05,
+ "loss": 0.207,
+ "step": 10307
+ },
+ {
+ "epoch": 28.087193460490465,
+ "grad_norm": 4.612616062164307,
+ "learning_rate": 1.687676261209837e-05,
+ "loss": 0.2843,
+ "step": 10308
+ },
+ {
+ "epoch": 28.08991825613079,
+ "grad_norm": 4.6198272705078125,
+ "learning_rate": 1.6876121878802748e-05,
+ "loss": 0.2848,
+ "step": 10309
+ },
+ {
+ "epoch": 28.092643051771116,
+ "grad_norm": 4.764092922210693,
+ "learning_rate": 1.687548109195611e-05,
+ "loss": 0.3821,
+ "step": 10310
+ },
+ {
+ "epoch": 28.095367847411445,
+ "grad_norm": 5.082345008850098,
+ "learning_rate": 1.6874840251563448e-05,
+ "loss": 0.3875,
+ "step": 10311
+ },
+ {
+ "epoch": 28.09809264305177,
+ "grad_norm": 4.5211358070373535,
+ "learning_rate": 1.6874199357629747e-05,
+ "loss": 0.3862,
+ "step": 10312
+ },
+ {
+ "epoch": 28.100817438692097,
+ "grad_norm": 4.83707857131958,
+ "learning_rate": 1.6873558410160006e-05,
+ "loss": 0.198,
+ "step": 10313
+ },
+ {
+ "epoch": 28.103542234332426,
+ "grad_norm": 3.788132429122925,
+ "learning_rate": 1.6872917409159213e-05,
+ "loss": 0.1468,
+ "step": 10314
+ },
+ {
+ "epoch": 28.106267029972752,
+ "grad_norm": 4.512640476226807,
+ "learning_rate": 1.6872276354632358e-05,
+ "loss": 0.295,
+ "step": 10315
+ },
+ {
+ "epoch": 28.108991825613078,
+ "grad_norm": 5.213919639587402,
+ "learning_rate": 1.687163524658444e-05,
+ "loss": 0.2534,
+ "step": 10316
+ },
+ {
+ "epoch": 28.111716621253407,
+ "grad_norm": 4.434065341949463,
+ "learning_rate": 1.6870994085020446e-05,
+ "loss": 0.2752,
+ "step": 10317
+ },
+ {
+ "epoch": 28.114441416893733,
+ "grad_norm": 5.06138801574707,
+ "learning_rate": 1.687035286994537e-05,
+ "loss": 0.2795,
+ "step": 10318
+ },
+ {
+ "epoch": 28.11716621253406,
+ "grad_norm": 7.1856889724731445,
+ "learning_rate": 1.6869711601364206e-05,
+ "loss": 0.2723,
+ "step": 10319
+ },
+ {
+ "epoch": 28.11989100817439,
+ "grad_norm": 4.303959846496582,
+ "learning_rate": 1.686907027928195e-05,
+ "loss": 0.2639,
+ "step": 10320
+ },
+ {
+ "epoch": 28.122615803814714,
+ "grad_norm": 7.749054908752441,
+ "learning_rate": 1.6868428903703598e-05,
+ "loss": 0.2714,
+ "step": 10321
+ },
+ {
+ "epoch": 28.12534059945504,
+ "grad_norm": 4.785575866699219,
+ "learning_rate": 1.6867787474634142e-05,
+ "loss": 0.3404,
+ "step": 10322
+ },
+ {
+ "epoch": 28.12806539509537,
+ "grad_norm": 4.583596706390381,
+ "learning_rate": 1.6867145992078577e-05,
+ "loss": 0.2892,
+ "step": 10323
+ },
+ {
+ "epoch": 28.130790190735695,
+ "grad_norm": 5.075932502746582,
+ "learning_rate": 1.68665044560419e-05,
+ "loss": 0.2543,
+ "step": 10324
+ },
+ {
+ "epoch": 28.13351498637602,
+ "grad_norm": 5.077108383178711,
+ "learning_rate": 1.686586286652911e-05,
+ "loss": 0.2419,
+ "step": 10325
+ },
+ {
+ "epoch": 28.13623978201635,
+ "grad_norm": 4.243706226348877,
+ "learning_rate": 1.6865221223545198e-05,
+ "loss": 0.2569,
+ "step": 10326
+ },
+ {
+ "epoch": 28.138964577656676,
+ "grad_norm": 4.452584266662598,
+ "learning_rate": 1.6864579527095163e-05,
+ "loss": 0.2774,
+ "step": 10327
+ },
+ {
+ "epoch": 28.141689373297,
+ "grad_norm": 4.444798946380615,
+ "learning_rate": 1.6863937777184006e-05,
+ "loss": 0.2855,
+ "step": 10328
+ },
+ {
+ "epoch": 28.14441416893733,
+ "grad_norm": 4.348587512969971,
+ "learning_rate": 1.686329597381672e-05,
+ "loss": 0.2851,
+ "step": 10329
+ },
+ {
+ "epoch": 28.147138964577657,
+ "grad_norm": 5.770857334136963,
+ "learning_rate": 1.686265411699831e-05,
+ "loss": 0.2923,
+ "step": 10330
+ },
+ {
+ "epoch": 28.149863760217983,
+ "grad_norm": 5.559133052825928,
+ "learning_rate": 1.6862012206733768e-05,
+ "loss": 0.4936,
+ "step": 10331
+ },
+ {
+ "epoch": 28.152588555858312,
+ "grad_norm": 4.57008171081543,
+ "learning_rate": 1.6861370243028097e-05,
+ "loss": 0.4223,
+ "step": 10332
+ },
+ {
+ "epoch": 28.155313351498638,
+ "grad_norm": 4.433175086975098,
+ "learning_rate": 1.6860728225886295e-05,
+ "loss": 0.2509,
+ "step": 10333
+ },
+ {
+ "epoch": 28.158038147138964,
+ "grad_norm": 4.190781116485596,
+ "learning_rate": 1.6860086155313364e-05,
+ "loss": 0.3195,
+ "step": 10334
+ },
+ {
+ "epoch": 28.160762942779293,
+ "grad_norm": 4.5344109535217285,
+ "learning_rate": 1.6859444031314297e-05,
+ "loss": 0.276,
+ "step": 10335
+ },
+ {
+ "epoch": 28.16348773841962,
+ "grad_norm": 4.638936519622803,
+ "learning_rate": 1.6858801853894105e-05,
+ "loss": 0.2019,
+ "step": 10336
+ },
+ {
+ "epoch": 28.166212534059945,
+ "grad_norm": 3.7201638221740723,
+ "learning_rate": 1.6858159623057787e-05,
+ "loss": 0.305,
+ "step": 10337
+ },
+ {
+ "epoch": 28.168937329700274,
+ "grad_norm": 5.98661470413208,
+ "learning_rate": 1.685751733881034e-05,
+ "loss": 0.2444,
+ "step": 10338
+ },
+ {
+ "epoch": 28.1716621253406,
+ "grad_norm": 4.521297931671143,
+ "learning_rate": 1.685687500115677e-05,
+ "loss": 0.4398,
+ "step": 10339
+ },
+ {
+ "epoch": 28.174386920980925,
+ "grad_norm": 4.747360706329346,
+ "learning_rate": 1.6856232610102074e-05,
+ "loss": 0.3007,
+ "step": 10340
+ },
+ {
+ "epoch": 28.177111716621255,
+ "grad_norm": 4.401281833648682,
+ "learning_rate": 1.6855590165651263e-05,
+ "loss": 0.388,
+ "step": 10341
+ },
+ {
+ "epoch": 28.17983651226158,
+ "grad_norm": 3.9401938915252686,
+ "learning_rate": 1.6854947667809333e-05,
+ "loss": 0.2491,
+ "step": 10342
+ },
+ {
+ "epoch": 28.182561307901906,
+ "grad_norm": 5.443638801574707,
+ "learning_rate": 1.6854305116581294e-05,
+ "loss": 0.2999,
+ "step": 10343
+ },
+ {
+ "epoch": 28.185286103542236,
+ "grad_norm": 4.871983528137207,
+ "learning_rate": 1.685366251197215e-05,
+ "loss": 0.4414,
+ "step": 10344
+ },
+ {
+ "epoch": 28.18801089918256,
+ "grad_norm": 4.485688209533691,
+ "learning_rate": 1.68530198539869e-05,
+ "loss": 0.2489,
+ "step": 10345
+ },
+ {
+ "epoch": 28.190735694822887,
+ "grad_norm": 4.270843982696533,
+ "learning_rate": 1.685237714263055e-05,
+ "loss": 0.5172,
+ "step": 10346
+ },
+ {
+ "epoch": 28.193460490463217,
+ "grad_norm": 5.275914669036865,
+ "learning_rate": 1.6851734377908106e-05,
+ "loss": 0.221,
+ "step": 10347
+ },
+ {
+ "epoch": 28.196185286103542,
+ "grad_norm": 5.253921985626221,
+ "learning_rate": 1.685109155982458e-05,
+ "loss": 0.3562,
+ "step": 10348
+ },
+ {
+ "epoch": 28.19891008174387,
+ "grad_norm": 4.804921627044678,
+ "learning_rate": 1.6850448688384972e-05,
+ "loss": 0.3721,
+ "step": 10349
+ },
+ {
+ "epoch": 28.201634877384198,
+ "grad_norm": 4.144891262054443,
+ "learning_rate": 1.6849805763594293e-05,
+ "loss": 0.3538,
+ "step": 10350
+ },
+ {
+ "epoch": 28.204359673024523,
+ "grad_norm": 4.628454208374023,
+ "learning_rate": 1.684916278545754e-05,
+ "loss": 0.2185,
+ "step": 10351
+ },
+ {
+ "epoch": 28.20708446866485,
+ "grad_norm": 5.1118950843811035,
+ "learning_rate": 1.6848519753979732e-05,
+ "loss": 0.2051,
+ "step": 10352
+ },
+ {
+ "epoch": 28.20980926430518,
+ "grad_norm": 4.7121477127075195,
+ "learning_rate": 1.6847876669165874e-05,
+ "loss": 0.3796,
+ "step": 10353
+ },
+ {
+ "epoch": 28.212534059945504,
+ "grad_norm": 5.12462043762207,
+ "learning_rate": 1.684723353102097e-05,
+ "loss": 0.2867,
+ "step": 10354
+ },
+ {
+ "epoch": 28.21525885558583,
+ "grad_norm": 4.812641143798828,
+ "learning_rate": 1.684659033955003e-05,
+ "loss": 0.2995,
+ "step": 10355
+ },
+ {
+ "epoch": 28.21798365122616,
+ "grad_norm": 4.512703895568848,
+ "learning_rate": 1.684594709475807e-05,
+ "loss": 0.3385,
+ "step": 10356
+ },
+ {
+ "epoch": 28.220708446866485,
+ "grad_norm": 5.507938385009766,
+ "learning_rate": 1.6845303796650094e-05,
+ "loss": 0.2086,
+ "step": 10357
+ },
+ {
+ "epoch": 28.22343324250681,
+ "grad_norm": 5.700564384460449,
+ "learning_rate": 1.684466044523111e-05,
+ "loss": 0.3046,
+ "step": 10358
+ },
+ {
+ "epoch": 28.22615803814714,
+ "grad_norm": 5.919508934020996,
+ "learning_rate": 1.6844017040506132e-05,
+ "loss": 0.397,
+ "step": 10359
+ },
+ {
+ "epoch": 28.228882833787466,
+ "grad_norm": 5.222629070281982,
+ "learning_rate": 1.6843373582480168e-05,
+ "loss": 0.2272,
+ "step": 10360
+ },
+ {
+ "epoch": 28.231607629427792,
+ "grad_norm": 4.30925178527832,
+ "learning_rate": 1.684273007115823e-05,
+ "loss": 0.2285,
+ "step": 10361
+ },
+ {
+ "epoch": 28.23433242506812,
+ "grad_norm": 4.468297004699707,
+ "learning_rate": 1.6842086506545334e-05,
+ "loss": 0.1852,
+ "step": 10362
+ },
+ {
+ "epoch": 28.237057220708447,
+ "grad_norm": 5.32032585144043,
+ "learning_rate": 1.6841442888646483e-05,
+ "loss": 0.2327,
+ "step": 10363
+ },
+ {
+ "epoch": 28.239782016348773,
+ "grad_norm": 4.914029598236084,
+ "learning_rate": 1.68407992174667e-05,
+ "loss": 0.2635,
+ "step": 10364
+ },
+ {
+ "epoch": 28.242506811989102,
+ "grad_norm": 8.559861183166504,
+ "learning_rate": 1.684015549301099e-05,
+ "loss": 0.2819,
+ "step": 10365
+ },
+ {
+ "epoch": 28.245231607629428,
+ "grad_norm": 5.324853897094727,
+ "learning_rate": 1.6839511715284372e-05,
+ "loss": 0.3201,
+ "step": 10366
+ },
+ {
+ "epoch": 28.247956403269754,
+ "grad_norm": 5.456117153167725,
+ "learning_rate": 1.6838867884291855e-05,
+ "loss": 0.2419,
+ "step": 10367
+ },
+ {
+ "epoch": 28.250681198910083,
+ "grad_norm": 4.296640396118164,
+ "learning_rate": 1.6838224000038455e-05,
+ "loss": 0.3814,
+ "step": 10368
+ },
+ {
+ "epoch": 28.25340599455041,
+ "grad_norm": 4.549238204956055,
+ "learning_rate": 1.683758006252919e-05,
+ "loss": 0.4298,
+ "step": 10369
+ },
+ {
+ "epoch": 28.256130790190735,
+ "grad_norm": 5.322376251220703,
+ "learning_rate": 1.6836936071769064e-05,
+ "loss": 0.2871,
+ "step": 10370
+ },
+ {
+ "epoch": 28.258855585831064,
+ "grad_norm": 4.609232425689697,
+ "learning_rate": 1.6836292027763102e-05,
+ "loss": 0.2358,
+ "step": 10371
+ },
+ {
+ "epoch": 28.26158038147139,
+ "grad_norm": 4.7855095863342285,
+ "learning_rate": 1.6835647930516322e-05,
+ "loss": 0.2758,
+ "step": 10372
+ },
+ {
+ "epoch": 28.264305177111716,
+ "grad_norm": 3.9289863109588623,
+ "learning_rate": 1.6835003780033738e-05,
+ "loss": 0.2841,
+ "step": 10373
+ },
+ {
+ "epoch": 28.267029972752045,
+ "grad_norm": 7.139620304107666,
+ "learning_rate": 1.6834359576320357e-05,
+ "loss": 0.2546,
+ "step": 10374
+ },
+ {
+ "epoch": 28.26975476839237,
+ "grad_norm": 5.4900221824646,
+ "learning_rate": 1.683371531938121e-05,
+ "loss": 0.3024,
+ "step": 10375
+ },
+ {
+ "epoch": 28.272479564032697,
+ "grad_norm": 5.049400806427002,
+ "learning_rate": 1.6833071009221306e-05,
+ "loss": 0.3163,
+ "step": 10376
+ },
+ {
+ "epoch": 28.275204359673026,
+ "grad_norm": 4.4455742835998535,
+ "learning_rate": 1.683242664584566e-05,
+ "loss": 0.3474,
+ "step": 10377
+ },
+ {
+ "epoch": 28.277929155313352,
+ "grad_norm": 5.076432704925537,
+ "learning_rate": 1.68317822292593e-05,
+ "loss": 0.2225,
+ "step": 10378
+ },
+ {
+ "epoch": 28.280653950953678,
+ "grad_norm": 5.2520222663879395,
+ "learning_rate": 1.683113775946724e-05,
+ "loss": 0.4432,
+ "step": 10379
+ },
+ {
+ "epoch": 28.283378746594007,
+ "grad_norm": 3.670811176300049,
+ "learning_rate": 1.6830493236474497e-05,
+ "loss": 0.1448,
+ "step": 10380
+ },
+ {
+ "epoch": 28.286103542234333,
+ "grad_norm": 4.692698001861572,
+ "learning_rate": 1.6829848660286093e-05,
+ "loss": 0.4561,
+ "step": 10381
+ },
+ {
+ "epoch": 28.28882833787466,
+ "grad_norm": 4.6405744552612305,
+ "learning_rate": 1.6829204030907045e-05,
+ "loss": 0.2995,
+ "step": 10382
+ },
+ {
+ "epoch": 28.291553133514988,
+ "grad_norm": 4.4055256843566895,
+ "learning_rate": 1.6828559348342382e-05,
+ "loss": 0.2367,
+ "step": 10383
+ },
+ {
+ "epoch": 28.294277929155314,
+ "grad_norm": 4.753334045410156,
+ "learning_rate": 1.6827914612597115e-05,
+ "loss": 0.1514,
+ "step": 10384
+ },
+ {
+ "epoch": 28.29700272479564,
+ "grad_norm": 4.87755823135376,
+ "learning_rate": 1.682726982367627e-05,
+ "loss": 0.2197,
+ "step": 10385
+ },
+ {
+ "epoch": 28.29972752043597,
+ "grad_norm": 8.6826753616333,
+ "learning_rate": 1.6826624981584868e-05,
+ "loss": 0.4924,
+ "step": 10386
+ },
+ {
+ "epoch": 28.302452316076295,
+ "grad_norm": 4.6673688888549805,
+ "learning_rate": 1.6825980086327927e-05,
+ "loss": 0.2021,
+ "step": 10387
+ },
+ {
+ "epoch": 28.30517711171662,
+ "grad_norm": 4.698017597198486,
+ "learning_rate": 1.6825335137910474e-05,
+ "loss": 0.2188,
+ "step": 10388
+ },
+ {
+ "epoch": 28.30790190735695,
+ "grad_norm": 5.459916591644287,
+ "learning_rate": 1.6824690136337533e-05,
+ "loss": 0.3697,
+ "step": 10389
+ },
+ {
+ "epoch": 28.310626702997276,
+ "grad_norm": 5.019954681396484,
+ "learning_rate": 1.682404508161412e-05,
+ "loss": 0.4133,
+ "step": 10390
+ },
+ {
+ "epoch": 28.3133514986376,
+ "grad_norm": 4.582798480987549,
+ "learning_rate": 1.682339997374527e-05,
+ "loss": 0.3987,
+ "step": 10391
+ },
+ {
+ "epoch": 28.31607629427793,
+ "grad_norm": 4.564081192016602,
+ "learning_rate": 1.6822754812735995e-05,
+ "loss": 0.2337,
+ "step": 10392
+ },
+ {
+ "epoch": 28.318801089918257,
+ "grad_norm": 4.07550048828125,
+ "learning_rate": 1.6822109598591327e-05,
+ "loss": 0.2335,
+ "step": 10393
+ },
+ {
+ "epoch": 28.321525885558582,
+ "grad_norm": 4.7736992835998535,
+ "learning_rate": 1.682146433131629e-05,
+ "loss": 0.2211,
+ "step": 10394
+ },
+ {
+ "epoch": 28.32425068119891,
+ "grad_norm": 4.518836498260498,
+ "learning_rate": 1.682081901091591e-05,
+ "loss": 0.2532,
+ "step": 10395
+ },
+ {
+ "epoch": 28.326975476839237,
+ "grad_norm": 4.928576469421387,
+ "learning_rate": 1.6820173637395208e-05,
+ "loss": 0.3347,
+ "step": 10396
+ },
+ {
+ "epoch": 28.329700272479563,
+ "grad_norm": 21.999847412109375,
+ "learning_rate": 1.6819528210759216e-05,
+ "loss": 0.3116,
+ "step": 10397
+ },
+ {
+ "epoch": 28.332425068119893,
+ "grad_norm": 4.116878986358643,
+ "learning_rate": 1.6818882731012954e-05,
+ "loss": 0.4288,
+ "step": 10398
+ },
+ {
+ "epoch": 28.33514986376022,
+ "grad_norm": 4.256111145019531,
+ "learning_rate": 1.6818237198161456e-05,
+ "loss": 0.4787,
+ "step": 10399
+ },
+ {
+ "epoch": 28.337874659400544,
+ "grad_norm": 4.541457653045654,
+ "learning_rate": 1.6817591612209744e-05,
+ "loss": 0.2936,
+ "step": 10400
+ },
+ {
+ "epoch": 28.340599455040874,
+ "grad_norm": 4.576930046081543,
+ "learning_rate": 1.681694597316285e-05,
+ "loss": 0.2494,
+ "step": 10401
+ },
+ {
+ "epoch": 28.3433242506812,
+ "grad_norm": 4.4591875076293945,
+ "learning_rate": 1.6816300281025796e-05,
+ "loss": 0.3944,
+ "step": 10402
+ },
+ {
+ "epoch": 28.346049046321525,
+ "grad_norm": 5.372528553009033,
+ "learning_rate": 1.681565453580362e-05,
+ "loss": 0.3109,
+ "step": 10403
+ },
+ {
+ "epoch": 28.348773841961854,
+ "grad_norm": 4.129212856292725,
+ "learning_rate": 1.681500873750134e-05,
+ "loss": 0.3197,
+ "step": 10404
+ },
+ {
+ "epoch": 28.35149863760218,
+ "grad_norm": 4.656406402587891,
+ "learning_rate": 1.6814362886123994e-05,
+ "loss": 0.2466,
+ "step": 10405
+ },
+ {
+ "epoch": 28.354223433242506,
+ "grad_norm": 3.94122052192688,
+ "learning_rate": 1.6813716981676612e-05,
+ "loss": 0.2822,
+ "step": 10406
+ },
+ {
+ "epoch": 28.356948228882835,
+ "grad_norm": 5.580506324768066,
+ "learning_rate": 1.681307102416422e-05,
+ "loss": 0.2155,
+ "step": 10407
+ },
+ {
+ "epoch": 28.35967302452316,
+ "grad_norm": 4.3633856773376465,
+ "learning_rate": 1.681242501359185e-05,
+ "loss": 0.2256,
+ "step": 10408
+ },
+ {
+ "epoch": 28.362397820163487,
+ "grad_norm": 3.8029916286468506,
+ "learning_rate": 1.681177894996453e-05,
+ "loss": 0.2824,
+ "step": 10409
+ },
+ {
+ "epoch": 28.365122615803816,
+ "grad_norm": 5.2708587646484375,
+ "learning_rate": 1.6811132833287296e-05,
+ "loss": 0.2801,
+ "step": 10410
+ },
+ {
+ "epoch": 28.367847411444142,
+ "grad_norm": 4.72566556930542,
+ "learning_rate": 1.6810486663565183e-05,
+ "loss": 0.2152,
+ "step": 10411
+ },
+ {
+ "epoch": 28.370572207084468,
+ "grad_norm": 5.475715160369873,
+ "learning_rate": 1.680984044080322e-05,
+ "loss": 0.4638,
+ "step": 10412
+ },
+ {
+ "epoch": 28.373297002724797,
+ "grad_norm": 13.203269004821777,
+ "learning_rate": 1.680919416500643e-05,
+ "loss": 0.246,
+ "step": 10413
+ },
+ {
+ "epoch": 28.376021798365123,
+ "grad_norm": 5.970368385314941,
+ "learning_rate": 1.6808547836179862e-05,
+ "loss": 0.3158,
+ "step": 10414
+ },
+ {
+ "epoch": 28.37874659400545,
+ "grad_norm": 5.112679481506348,
+ "learning_rate": 1.680790145432854e-05,
+ "loss": 0.3093,
+ "step": 10415
+ },
+ {
+ "epoch": 28.381471389645778,
+ "grad_norm": 4.852305889129639,
+ "learning_rate": 1.6807255019457503e-05,
+ "loss": 0.186,
+ "step": 10416
+ },
+ {
+ "epoch": 28.384196185286104,
+ "grad_norm": 5.56129789352417,
+ "learning_rate": 1.680660853157178e-05,
+ "loss": 0.3488,
+ "step": 10417
+ },
+ {
+ "epoch": 28.38692098092643,
+ "grad_norm": 4.688525676727295,
+ "learning_rate": 1.680596199067641e-05,
+ "loss": 0.3678,
+ "step": 10418
+ },
+ {
+ "epoch": 28.38964577656676,
+ "grad_norm": 4.384783744812012,
+ "learning_rate": 1.6805315396776427e-05,
+ "loss": 0.279,
+ "step": 10419
+ },
+ {
+ "epoch": 28.392370572207085,
+ "grad_norm": 5.139349460601807,
+ "learning_rate": 1.6804668749876867e-05,
+ "loss": 0.291,
+ "step": 10420
+ },
+ {
+ "epoch": 28.39509536784741,
+ "grad_norm": 6.445595741271973,
+ "learning_rate": 1.6804022049982768e-05,
+ "loss": 0.3852,
+ "step": 10421
+ },
+ {
+ "epoch": 28.39782016348774,
+ "grad_norm": 5.166479587554932,
+ "learning_rate": 1.6803375297099162e-05,
+ "loss": 0.284,
+ "step": 10422
+ },
+ {
+ "epoch": 28.400544959128066,
+ "grad_norm": 5.164698123931885,
+ "learning_rate": 1.6802728491231083e-05,
+ "loss": 0.2101,
+ "step": 10423
+ },
+ {
+ "epoch": 28.40326975476839,
+ "grad_norm": 5.02501106262207,
+ "learning_rate": 1.680208163238358e-05,
+ "loss": 0.3263,
+ "step": 10424
+ },
+ {
+ "epoch": 28.40599455040872,
+ "grad_norm": 8.500020980834961,
+ "learning_rate": 1.6801434720561683e-05,
+ "loss": 0.3816,
+ "step": 10425
+ },
+ {
+ "epoch": 28.408719346049047,
+ "grad_norm": 4.521326541900635,
+ "learning_rate": 1.680078775577043e-05,
+ "loss": 0.299,
+ "step": 10426
+ },
+ {
+ "epoch": 28.411444141689373,
+ "grad_norm": 5.140456676483154,
+ "learning_rate": 1.680014073801486e-05,
+ "loss": 0.3279,
+ "step": 10427
+ },
+ {
+ "epoch": 28.414168937329702,
+ "grad_norm": 7.356240272521973,
+ "learning_rate": 1.6799493667300018e-05,
+ "loss": 0.2819,
+ "step": 10428
+ },
+ {
+ "epoch": 28.416893732970028,
+ "grad_norm": 5.9566545486450195,
+ "learning_rate": 1.6798846543630934e-05,
+ "loss": 0.2881,
+ "step": 10429
+ },
+ {
+ "epoch": 28.419618528610354,
+ "grad_norm": 4.500622272491455,
+ "learning_rate": 1.6798199367012652e-05,
+ "loss": 0.266,
+ "step": 10430
+ },
+ {
+ "epoch": 28.422343324250683,
+ "grad_norm": 4.68209171295166,
+ "learning_rate": 1.6797552137450214e-05,
+ "loss": 0.3055,
+ "step": 10431
+ },
+ {
+ "epoch": 28.42506811989101,
+ "grad_norm": 6.539977073669434,
+ "learning_rate": 1.6796904854948656e-05,
+ "loss": 0.265,
+ "step": 10432
+ },
+ {
+ "epoch": 28.427792915531334,
+ "grad_norm": 4.196413040161133,
+ "learning_rate": 1.6796257519513024e-05,
+ "loss": 0.3267,
+ "step": 10433
+ },
+ {
+ "epoch": 28.430517711171664,
+ "grad_norm": 5.390618801116943,
+ "learning_rate": 1.6795610131148357e-05,
+ "loss": 0.4259,
+ "step": 10434
+ },
+ {
+ "epoch": 28.43324250681199,
+ "grad_norm": 4.998218059539795,
+ "learning_rate": 1.6794962689859696e-05,
+ "loss": 0.2974,
+ "step": 10435
+ },
+ {
+ "epoch": 28.435967302452315,
+ "grad_norm": 5.010826587677002,
+ "learning_rate": 1.6794315195652085e-05,
+ "loss": 0.2844,
+ "step": 10436
+ },
+ {
+ "epoch": 28.438692098092645,
+ "grad_norm": 4.973742485046387,
+ "learning_rate": 1.6793667648530567e-05,
+ "loss": 0.4095,
+ "step": 10437
+ },
+ {
+ "epoch": 28.44141689373297,
+ "grad_norm": 5.154592990875244,
+ "learning_rate": 1.6793020048500183e-05,
+ "loss": 0.2725,
+ "step": 10438
+ },
+ {
+ "epoch": 28.444141689373296,
+ "grad_norm": 5.760514736175537,
+ "learning_rate": 1.6792372395565978e-05,
+ "loss": 0.5192,
+ "step": 10439
+ },
+ {
+ "epoch": 28.446866485013626,
+ "grad_norm": 4.185200214385986,
+ "learning_rate": 1.6791724689732992e-05,
+ "loss": 0.5423,
+ "step": 10440
+ },
+ {
+ "epoch": 28.44959128065395,
+ "grad_norm": 5.270141124725342,
+ "learning_rate": 1.6791076931006276e-05,
+ "loss": 0.2482,
+ "step": 10441
+ },
+ {
+ "epoch": 28.452316076294277,
+ "grad_norm": 4.442587375640869,
+ "learning_rate": 1.679042911939087e-05,
+ "loss": 0.1888,
+ "step": 10442
+ },
+ {
+ "epoch": 28.455040871934607,
+ "grad_norm": 4.7379889488220215,
+ "learning_rate": 1.6789781254891822e-05,
+ "loss": 0.3118,
+ "step": 10443
+ },
+ {
+ "epoch": 28.457765667574932,
+ "grad_norm": 5.281705856323242,
+ "learning_rate": 1.6789133337514178e-05,
+ "loss": 0.3739,
+ "step": 10444
+ },
+ {
+ "epoch": 28.460490463215258,
+ "grad_norm": 4.412702560424805,
+ "learning_rate": 1.6788485367262975e-05,
+ "loss": 0.4869,
+ "step": 10445
+ },
+ {
+ "epoch": 28.463215258855588,
+ "grad_norm": 4.959930419921875,
+ "learning_rate": 1.6787837344143273e-05,
+ "loss": 0.347,
+ "step": 10446
+ },
+ {
+ "epoch": 28.465940054495913,
+ "grad_norm": 5.195252895355225,
+ "learning_rate": 1.678718926816011e-05,
+ "loss": 0.4127,
+ "step": 10447
+ },
+ {
+ "epoch": 28.46866485013624,
+ "grad_norm": 4.799053192138672,
+ "learning_rate": 1.6786541139318537e-05,
+ "loss": 0.2281,
+ "step": 10448
+ },
+ {
+ "epoch": 28.47138964577657,
+ "grad_norm": 4.857353210449219,
+ "learning_rate": 1.6785892957623598e-05,
+ "loss": 0.3602,
+ "step": 10449
+ },
+ {
+ "epoch": 28.474114441416894,
+ "grad_norm": 4.618188858032227,
+ "learning_rate": 1.678524472308034e-05,
+ "loss": 0.2344,
+ "step": 10450
+ },
+ {
+ "epoch": 28.47683923705722,
+ "grad_norm": 4.763349533081055,
+ "learning_rate": 1.678459643569382e-05,
+ "loss": 0.3824,
+ "step": 10451
+ },
+ {
+ "epoch": 28.479564032697546,
+ "grad_norm": 3.376983404159546,
+ "learning_rate": 1.6783948095469082e-05,
+ "loss": 0.2058,
+ "step": 10452
+ },
+ {
+ "epoch": 28.482288828337875,
+ "grad_norm": 4.78580379486084,
+ "learning_rate": 1.678329970241117e-05,
+ "loss": 0.2483,
+ "step": 10453
+ },
+ {
+ "epoch": 28.4850136239782,
+ "grad_norm": 4.42450475692749,
+ "learning_rate": 1.678265125652514e-05,
+ "loss": 0.369,
+ "step": 10454
+ },
+ {
+ "epoch": 28.48773841961853,
+ "grad_norm": 4.797671794891357,
+ "learning_rate": 1.6782002757816043e-05,
+ "loss": 0.2957,
+ "step": 10455
+ },
+ {
+ "epoch": 28.490463215258856,
+ "grad_norm": 5.316877365112305,
+ "learning_rate": 1.6781354206288924e-05,
+ "loss": 0.326,
+ "step": 10456
+ },
+ {
+ "epoch": 28.493188010899182,
+ "grad_norm": 5.498935699462891,
+ "learning_rate": 1.6780705601948833e-05,
+ "loss": 0.2272,
+ "step": 10457
+ },
+ {
+ "epoch": 28.495912806539508,
+ "grad_norm": 4.436408519744873,
+ "learning_rate": 1.6780056944800833e-05,
+ "loss": 0.2681,
+ "step": 10458
+ },
+ {
+ "epoch": 28.498637602179837,
+ "grad_norm": 4.863140106201172,
+ "learning_rate": 1.6779408234849964e-05,
+ "loss": 0.3317,
+ "step": 10459
+ },
+ {
+ "epoch": 28.501362397820163,
+ "grad_norm": 5.636782646179199,
+ "learning_rate": 1.677875947210128e-05,
+ "loss": 0.3084,
+ "step": 10460
+ },
+ {
+ "epoch": 28.504087193460492,
+ "grad_norm": 4.273879528045654,
+ "learning_rate": 1.6778110656559837e-05,
+ "loss": 0.2432,
+ "step": 10461
+ },
+ {
+ "epoch": 28.506811989100818,
+ "grad_norm": 5.533421039581299,
+ "learning_rate": 1.677746178823069e-05,
+ "loss": 0.3857,
+ "step": 10462
+ },
+ {
+ "epoch": 28.509536784741144,
+ "grad_norm": 6.676644802093506,
+ "learning_rate": 1.6776812867118885e-05,
+ "loss": 0.4023,
+ "step": 10463
+ },
+ {
+ "epoch": 28.51226158038147,
+ "grad_norm": 5.201764106750488,
+ "learning_rate": 1.6776163893229476e-05,
+ "loss": 0.2684,
+ "step": 10464
+ },
+ {
+ "epoch": 28.5149863760218,
+ "grad_norm": 4.76231575012207,
+ "learning_rate": 1.6775514866567527e-05,
+ "loss": 0.46,
+ "step": 10465
+ },
+ {
+ "epoch": 28.517711171662125,
+ "grad_norm": 4.810202598571777,
+ "learning_rate": 1.677486578713808e-05,
+ "loss": 0.2329,
+ "step": 10466
+ },
+ {
+ "epoch": 28.520435967302454,
+ "grad_norm": 11.341081619262695,
+ "learning_rate": 1.6774216654946203e-05,
+ "loss": 0.2551,
+ "step": 10467
+ },
+ {
+ "epoch": 28.52316076294278,
+ "grad_norm": 5.281602382659912,
+ "learning_rate": 1.677356746999694e-05,
+ "loss": 0.3408,
+ "step": 10468
+ },
+ {
+ "epoch": 28.525885558583106,
+ "grad_norm": 4.721526622772217,
+ "learning_rate": 1.6772918232295353e-05,
+ "loss": 0.2898,
+ "step": 10469
+ },
+ {
+ "epoch": 28.52861035422343,
+ "grad_norm": 5.6954569816589355,
+ "learning_rate": 1.6772268941846496e-05,
+ "loss": 0.2209,
+ "step": 10470
+ },
+ {
+ "epoch": 28.53133514986376,
+ "grad_norm": 4.510961055755615,
+ "learning_rate": 1.6771619598655432e-05,
+ "loss": 0.234,
+ "step": 10471
+ },
+ {
+ "epoch": 28.534059945504087,
+ "grad_norm": 4.935581207275391,
+ "learning_rate": 1.6770970202727206e-05,
+ "loss": 0.3896,
+ "step": 10472
+ },
+ {
+ "epoch": 28.536784741144416,
+ "grad_norm": 4.643245697021484,
+ "learning_rate": 1.6770320754066886e-05,
+ "loss": 0.4205,
+ "step": 10473
+ },
+ {
+ "epoch": 28.539509536784742,
+ "grad_norm": 5.127284526824951,
+ "learning_rate": 1.676967125267952e-05,
+ "loss": 0.243,
+ "step": 10474
+ },
+ {
+ "epoch": 28.542234332425068,
+ "grad_norm": 4.840200424194336,
+ "learning_rate": 1.6769021698570178e-05,
+ "loss": 0.3224,
+ "step": 10475
+ },
+ {
+ "epoch": 28.544959128065393,
+ "grad_norm": 4.4068756103515625,
+ "learning_rate": 1.676837209174391e-05,
+ "loss": 0.2469,
+ "step": 10476
+ },
+ {
+ "epoch": 28.547683923705723,
+ "grad_norm": 4.923526287078857,
+ "learning_rate": 1.676772243220578e-05,
+ "loss": 0.3113,
+ "step": 10477
+ },
+ {
+ "epoch": 28.55040871934605,
+ "grad_norm": 5.020417213439941,
+ "learning_rate": 1.6767072719960845e-05,
+ "loss": 0.2791,
+ "step": 10478
+ },
+ {
+ "epoch": 28.553133514986374,
+ "grad_norm": 4.625273704528809,
+ "learning_rate": 1.6766422955014164e-05,
+ "loss": 0.2287,
+ "step": 10479
+ },
+ {
+ "epoch": 28.555858310626704,
+ "grad_norm": 4.492345809936523,
+ "learning_rate": 1.67657731373708e-05,
+ "loss": 0.4156,
+ "step": 10480
+ },
+ {
+ "epoch": 28.55858310626703,
+ "grad_norm": 5.633136749267578,
+ "learning_rate": 1.6765123267035813e-05,
+ "loss": 0.3589,
+ "step": 10481
+ },
+ {
+ "epoch": 28.561307901907355,
+ "grad_norm": 5.001399517059326,
+ "learning_rate": 1.6764473344014264e-05,
+ "loss": 0.2437,
+ "step": 10482
+ },
+ {
+ "epoch": 28.564032697547685,
+ "grad_norm": 3.947585344314575,
+ "learning_rate": 1.6763823368311213e-05,
+ "loss": 0.2255,
+ "step": 10483
+ },
+ {
+ "epoch": 28.56675749318801,
+ "grad_norm": 4.353891849517822,
+ "learning_rate": 1.6763173339931718e-05,
+ "loss": 0.1896,
+ "step": 10484
+ },
+ {
+ "epoch": 28.569482288828336,
+ "grad_norm": 4.702773094177246,
+ "learning_rate": 1.6762523258880855e-05,
+ "loss": 0.2815,
+ "step": 10485
+ },
+ {
+ "epoch": 28.572207084468666,
+ "grad_norm": 4.608099937438965,
+ "learning_rate": 1.6761873125163676e-05,
+ "loss": 0.278,
+ "step": 10486
+ },
+ {
+ "epoch": 28.57493188010899,
+ "grad_norm": 5.119320392608643,
+ "learning_rate": 1.6761222938785245e-05,
+ "loss": 0.2375,
+ "step": 10487
+ },
+ {
+ "epoch": 28.577656675749317,
+ "grad_norm": 6.019540309906006,
+ "learning_rate": 1.676057269975063e-05,
+ "loss": 0.3755,
+ "step": 10488
+ },
+ {
+ "epoch": 28.580381471389646,
+ "grad_norm": 5.056300163269043,
+ "learning_rate": 1.675992240806489e-05,
+ "loss": 0.2027,
+ "step": 10489
+ },
+ {
+ "epoch": 28.583106267029972,
+ "grad_norm": 6.11469030380249,
+ "learning_rate": 1.6759272063733094e-05,
+ "loss": 0.4217,
+ "step": 10490
+ },
+ {
+ "epoch": 28.585831062670298,
+ "grad_norm": 4.313836097717285,
+ "learning_rate": 1.6758621666760304e-05,
+ "loss": 0.2628,
+ "step": 10491
+ },
+ {
+ "epoch": 28.588555858310627,
+ "grad_norm": 4.314679145812988,
+ "learning_rate": 1.675797121715159e-05,
+ "loss": 0.2185,
+ "step": 10492
+ },
+ {
+ "epoch": 28.591280653950953,
+ "grad_norm": 4.360710144042969,
+ "learning_rate": 1.6757320714912007e-05,
+ "loss": 0.2466,
+ "step": 10493
+ },
+ {
+ "epoch": 28.59400544959128,
+ "grad_norm": 5.254417419433594,
+ "learning_rate": 1.6756670160046628e-05,
+ "loss": 0.2401,
+ "step": 10494
+ },
+ {
+ "epoch": 28.59673024523161,
+ "grad_norm": 5.535014629364014,
+ "learning_rate": 1.6756019552560522e-05,
+ "loss": 0.3847,
+ "step": 10495
+ },
+ {
+ "epoch": 28.599455040871934,
+ "grad_norm": 5.323489189147949,
+ "learning_rate": 1.6755368892458753e-05,
+ "loss": 0.2849,
+ "step": 10496
+ },
+ {
+ "epoch": 28.60217983651226,
+ "grad_norm": 4.9937591552734375,
+ "learning_rate": 1.675471817974639e-05,
+ "loss": 0.4564,
+ "step": 10497
+ },
+ {
+ "epoch": 28.60490463215259,
+ "grad_norm": 4.710936069488525,
+ "learning_rate": 1.6754067414428497e-05,
+ "loss": 0.4174,
+ "step": 10498
+ },
+ {
+ "epoch": 28.607629427792915,
+ "grad_norm": 4.587750434875488,
+ "learning_rate": 1.6753416596510144e-05,
+ "loss": 0.2281,
+ "step": 10499
+ },
+ {
+ "epoch": 28.61035422343324,
+ "grad_norm": 4.083234786987305,
+ "learning_rate": 1.6752765725996402e-05,
+ "loss": 0.2737,
+ "step": 10500
+ },
+ {
+ "epoch": 28.61307901907357,
+ "grad_norm": 4.63611364364624,
+ "learning_rate": 1.6752114802892332e-05,
+ "loss": 0.2749,
+ "step": 10501
+ },
+ {
+ "epoch": 28.615803814713896,
+ "grad_norm": 4.7624382972717285,
+ "learning_rate": 1.6751463827203018e-05,
+ "loss": 0.256,
+ "step": 10502
+ },
+ {
+ "epoch": 28.618528610354222,
+ "grad_norm": 6.9603753089904785,
+ "learning_rate": 1.6750812798933512e-05,
+ "loss": 0.416,
+ "step": 10503
+ },
+ {
+ "epoch": 28.62125340599455,
+ "grad_norm": 4.672041416168213,
+ "learning_rate": 1.6750161718088902e-05,
+ "loss": 0.2497,
+ "step": 10504
+ },
+ {
+ "epoch": 28.623978201634877,
+ "grad_norm": 4.666864395141602,
+ "learning_rate": 1.6749510584674243e-05,
+ "loss": 0.2072,
+ "step": 10505
+ },
+ {
+ "epoch": 28.626702997275203,
+ "grad_norm": 3.7327029705047607,
+ "learning_rate": 1.6748859398694616e-05,
+ "loss": 0.1759,
+ "step": 10506
+ },
+ {
+ "epoch": 28.629427792915532,
+ "grad_norm": 5.486154079437256,
+ "learning_rate": 1.674820816015509e-05,
+ "loss": 0.2673,
+ "step": 10507
+ },
+ {
+ "epoch": 28.632152588555858,
+ "grad_norm": 5.7112555503845215,
+ "learning_rate": 1.6747556869060735e-05,
+ "loss": 0.3034,
+ "step": 10508
+ },
+ {
+ "epoch": 28.634877384196184,
+ "grad_norm": 4.248378276824951,
+ "learning_rate": 1.6746905525416625e-05,
+ "loss": 0.2141,
+ "step": 10509
+ },
+ {
+ "epoch": 28.637602179836513,
+ "grad_norm": 5.350340366363525,
+ "learning_rate": 1.674625412922783e-05,
+ "loss": 0.3573,
+ "step": 10510
+ },
+ {
+ "epoch": 28.64032697547684,
+ "grad_norm": 8.455004692077637,
+ "learning_rate": 1.6745602680499426e-05,
+ "loss": 0.2968,
+ "step": 10511
+ },
+ {
+ "epoch": 28.643051771117165,
+ "grad_norm": 4.932601451873779,
+ "learning_rate": 1.6744951179236487e-05,
+ "loss": 0.28,
+ "step": 10512
+ },
+ {
+ "epoch": 28.645776566757494,
+ "grad_norm": 4.384659767150879,
+ "learning_rate": 1.6744299625444082e-05,
+ "loss": 0.2578,
+ "step": 10513
+ },
+ {
+ "epoch": 28.64850136239782,
+ "grad_norm": 6.013816833496094,
+ "learning_rate": 1.674364801912729e-05,
+ "loss": 0.2885,
+ "step": 10514
+ },
+ {
+ "epoch": 28.651226158038146,
+ "grad_norm": 4.646938800811768,
+ "learning_rate": 1.6742996360291184e-05,
+ "loss": 0.3225,
+ "step": 10515
+ },
+ {
+ "epoch": 28.653950953678475,
+ "grad_norm": 4.439748287200928,
+ "learning_rate": 1.674234464894084e-05,
+ "loss": 0.2218,
+ "step": 10516
+ },
+ {
+ "epoch": 28.6566757493188,
+ "grad_norm": 3.8608551025390625,
+ "learning_rate": 1.6741692885081333e-05,
+ "loss": 0.3274,
+ "step": 10517
+ },
+ {
+ "epoch": 28.659400544959126,
+ "grad_norm": 4.895925998687744,
+ "learning_rate": 1.6741041068717737e-05,
+ "loss": 0.3772,
+ "step": 10518
+ },
+ {
+ "epoch": 28.662125340599456,
+ "grad_norm": 7.44224214553833,
+ "learning_rate": 1.674038919985513e-05,
+ "loss": 0.2444,
+ "step": 10519
+ },
+ {
+ "epoch": 28.66485013623978,
+ "grad_norm": 4.4791646003723145,
+ "learning_rate": 1.6739737278498595e-05,
+ "loss": 0.2964,
+ "step": 10520
+ },
+ {
+ "epoch": 28.667574931880107,
+ "grad_norm": 4.428526401519775,
+ "learning_rate": 1.6739085304653197e-05,
+ "loss": 0.3305,
+ "step": 10521
+ },
+ {
+ "epoch": 28.670299727520437,
+ "grad_norm": 6.100392818450928,
+ "learning_rate": 1.673843327832402e-05,
+ "loss": 0.3739,
+ "step": 10522
+ },
+ {
+ "epoch": 28.673024523160763,
+ "grad_norm": 4.601495265960693,
+ "learning_rate": 1.6737781199516143e-05,
+ "loss": 0.3124,
+ "step": 10523
+ },
+ {
+ "epoch": 28.67574931880109,
+ "grad_norm": 4.7969441413879395,
+ "learning_rate": 1.6737129068234642e-05,
+ "loss": 0.3861,
+ "step": 10524
+ },
+ {
+ "epoch": 28.678474114441418,
+ "grad_norm": 6.164493560791016,
+ "learning_rate": 1.67364768844846e-05,
+ "loss": 0.3911,
+ "step": 10525
+ },
+ {
+ "epoch": 28.681198910081743,
+ "grad_norm": 4.500364780426025,
+ "learning_rate": 1.6735824648271085e-05,
+ "loss": 0.2213,
+ "step": 10526
+ },
+ {
+ "epoch": 28.68392370572207,
+ "grad_norm": 8.306814193725586,
+ "learning_rate": 1.6735172359599193e-05,
+ "loss": 0.4521,
+ "step": 10527
+ },
+ {
+ "epoch": 28.6866485013624,
+ "grad_norm": 5.113101482391357,
+ "learning_rate": 1.673452001847399e-05,
+ "loss": 0.3369,
+ "step": 10528
+ },
+ {
+ "epoch": 28.689373297002724,
+ "grad_norm": 5.124936580657959,
+ "learning_rate": 1.6733867624900565e-05,
+ "loss": 0.4462,
+ "step": 10529
+ },
+ {
+ "epoch": 28.69209809264305,
+ "grad_norm": 6.268643856048584,
+ "learning_rate": 1.6733215178883996e-05,
+ "loss": 0.2527,
+ "step": 10530
+ },
+ {
+ "epoch": 28.69482288828338,
+ "grad_norm": 5.718371391296387,
+ "learning_rate": 1.6732562680429358e-05,
+ "loss": 0.2327,
+ "step": 10531
+ },
+ {
+ "epoch": 28.697547683923705,
+ "grad_norm": 4.619215965270996,
+ "learning_rate": 1.6731910129541745e-05,
+ "loss": 0.3264,
+ "step": 10532
+ },
+ {
+ "epoch": 28.70027247956403,
+ "grad_norm": 4.640407562255859,
+ "learning_rate": 1.673125752622623e-05,
+ "loss": 0.3553,
+ "step": 10533
+ },
+ {
+ "epoch": 28.70299727520436,
+ "grad_norm": 5.534787654876709,
+ "learning_rate": 1.67306048704879e-05,
+ "loss": 0.3087,
+ "step": 10534
+ },
+ {
+ "epoch": 28.705722070844686,
+ "grad_norm": 5.21694278717041,
+ "learning_rate": 1.672995216233183e-05,
+ "loss": 0.2909,
+ "step": 10535
+ },
+ {
+ "epoch": 28.708446866485012,
+ "grad_norm": 4.744660377502441,
+ "learning_rate": 1.6729299401763116e-05,
+ "loss": 0.2735,
+ "step": 10536
+ },
+ {
+ "epoch": 28.71117166212534,
+ "grad_norm": 4.789371013641357,
+ "learning_rate": 1.672864658878683e-05,
+ "loss": 0.2603,
+ "step": 10537
+ },
+ {
+ "epoch": 28.713896457765667,
+ "grad_norm": 4.311310768127441,
+ "learning_rate": 1.6727993723408066e-05,
+ "loss": 0.246,
+ "step": 10538
+ },
+ {
+ "epoch": 28.716621253405993,
+ "grad_norm": 5.790502548217773,
+ "learning_rate": 1.67273408056319e-05,
+ "loss": 0.2204,
+ "step": 10539
+ },
+ {
+ "epoch": 28.719346049046322,
+ "grad_norm": 5.3421478271484375,
+ "learning_rate": 1.672668783546342e-05,
+ "loss": 0.3974,
+ "step": 10540
+ },
+ {
+ "epoch": 28.722070844686648,
+ "grad_norm": 14.68706226348877,
+ "learning_rate": 1.6726034812907713e-05,
+ "loss": 0.2956,
+ "step": 10541
+ },
+ {
+ "epoch": 28.724795640326974,
+ "grad_norm": 4.422563552856445,
+ "learning_rate": 1.672538173796986e-05,
+ "loss": 0.2564,
+ "step": 10542
+ },
+ {
+ "epoch": 28.727520435967303,
+ "grad_norm": 5.566542148590088,
+ "learning_rate": 1.6724728610654955e-05,
+ "loss": 0.2345,
+ "step": 10543
+ },
+ {
+ "epoch": 28.73024523160763,
+ "grad_norm": 5.301816940307617,
+ "learning_rate": 1.672407543096808e-05,
+ "loss": 0.2881,
+ "step": 10544
+ },
+ {
+ "epoch": 28.732970027247955,
+ "grad_norm": 4.646319389343262,
+ "learning_rate": 1.6723422198914317e-05,
+ "loss": 0.2803,
+ "step": 10545
+ },
+ {
+ "epoch": 28.735694822888284,
+ "grad_norm": 5.499476432800293,
+ "learning_rate": 1.6722768914498763e-05,
+ "loss": 0.1901,
+ "step": 10546
+ },
+ {
+ "epoch": 28.73841961852861,
+ "grad_norm": 6.309245586395264,
+ "learning_rate": 1.6722115577726497e-05,
+ "loss": 0.4004,
+ "step": 10547
+ },
+ {
+ "epoch": 28.741144414168936,
+ "grad_norm": 10.687594413757324,
+ "learning_rate": 1.6721462188602618e-05,
+ "loss": 0.237,
+ "step": 10548
+ },
+ {
+ "epoch": 28.743869209809265,
+ "grad_norm": 5.62407112121582,
+ "learning_rate": 1.6720808747132204e-05,
+ "loss": 0.2073,
+ "step": 10549
+ },
+ {
+ "epoch": 28.74659400544959,
+ "grad_norm": 5.183177947998047,
+ "learning_rate": 1.6720155253320348e-05,
+ "loss": 0.2858,
+ "step": 10550
+ },
+ {
+ "epoch": 28.749318801089917,
+ "grad_norm": 4.2622904777526855,
+ "learning_rate": 1.6719501707172135e-05,
+ "loss": 0.1773,
+ "step": 10551
+ },
+ {
+ "epoch": 28.752043596730246,
+ "grad_norm": 6.124391078948975,
+ "learning_rate": 1.6718848108692665e-05,
+ "loss": 0.394,
+ "step": 10552
+ },
+ {
+ "epoch": 28.754768392370572,
+ "grad_norm": 5.7481689453125,
+ "learning_rate": 1.671819445788702e-05,
+ "loss": 0.3386,
+ "step": 10553
+ },
+ {
+ "epoch": 28.757493188010898,
+ "grad_norm": 4.912421226501465,
+ "learning_rate": 1.671754075476029e-05,
+ "loss": 0.4714,
+ "step": 10554
+ },
+ {
+ "epoch": 28.760217983651227,
+ "grad_norm": 4.777478218078613,
+ "learning_rate": 1.6716886999317573e-05,
+ "loss": 0.2572,
+ "step": 10555
+ },
+ {
+ "epoch": 28.762942779291553,
+ "grad_norm": 4.630034923553467,
+ "learning_rate": 1.6716233191563956e-05,
+ "loss": 0.226,
+ "step": 10556
+ },
+ {
+ "epoch": 28.76566757493188,
+ "grad_norm": 5.568329811096191,
+ "learning_rate": 1.671557933150453e-05,
+ "loss": 0.3525,
+ "step": 10557
+ },
+ {
+ "epoch": 28.768392370572208,
+ "grad_norm": 5.665154933929443,
+ "learning_rate": 1.6714925419144384e-05,
+ "loss": 0.3292,
+ "step": 10558
+ },
+ {
+ "epoch": 28.771117166212534,
+ "grad_norm": 5.361033916473389,
+ "learning_rate": 1.671427145448862e-05,
+ "loss": 0.2169,
+ "step": 10559
+ },
+ {
+ "epoch": 28.77384196185286,
+ "grad_norm": 5.1427717208862305,
+ "learning_rate": 1.6713617437542328e-05,
+ "loss": 0.2284,
+ "step": 10560
+ },
+ {
+ "epoch": 28.77656675749319,
+ "grad_norm": 4.310766696929932,
+ "learning_rate": 1.6712963368310596e-05,
+ "loss": 0.5115,
+ "step": 10561
+ },
+ {
+ "epoch": 28.779291553133515,
+ "grad_norm": 5.340289115905762,
+ "learning_rate": 1.671230924679852e-05,
+ "loss": 0.4457,
+ "step": 10562
+ },
+ {
+ "epoch": 28.78201634877384,
+ "grad_norm": 5.705671310424805,
+ "learning_rate": 1.6711655073011196e-05,
+ "loss": 0.2885,
+ "step": 10563
+ },
+ {
+ "epoch": 28.78474114441417,
+ "grad_norm": 5.048360824584961,
+ "learning_rate": 1.6711000846953722e-05,
+ "loss": 0.2961,
+ "step": 10564
+ },
+ {
+ "epoch": 28.787465940054496,
+ "grad_norm": 4.9710798263549805,
+ "learning_rate": 1.671034656863119e-05,
+ "loss": 0.5142,
+ "step": 10565
+ },
+ {
+ "epoch": 28.79019073569482,
+ "grad_norm": 4.9344401359558105,
+ "learning_rate": 1.6709692238048688e-05,
+ "loss": 0.2443,
+ "step": 10566
+ },
+ {
+ "epoch": 28.79291553133515,
+ "grad_norm": 5.780189514160156,
+ "learning_rate": 1.670903785521132e-05,
+ "loss": 0.3118,
+ "step": 10567
+ },
+ {
+ "epoch": 28.795640326975477,
+ "grad_norm": 5.2917399406433105,
+ "learning_rate": 1.670838342012419e-05,
+ "loss": 0.237,
+ "step": 10568
+ },
+ {
+ "epoch": 28.798365122615802,
+ "grad_norm": 5.248785495758057,
+ "learning_rate": 1.6707728932792376e-05,
+ "loss": 0.3246,
+ "step": 10569
+ },
+ {
+ "epoch": 28.80108991825613,
+ "grad_norm": 4.708765506744385,
+ "learning_rate": 1.670707439322099e-05,
+ "loss": 0.2927,
+ "step": 10570
+ },
+ {
+ "epoch": 28.803814713896458,
+ "grad_norm": 4.216134071350098,
+ "learning_rate": 1.6706419801415122e-05,
+ "loss": 0.4384,
+ "step": 10571
+ },
+ {
+ "epoch": 28.806539509536783,
+ "grad_norm": 3.788907766342163,
+ "learning_rate": 1.6705765157379876e-05,
+ "loss": 0.2007,
+ "step": 10572
+ },
+ {
+ "epoch": 28.809264305177113,
+ "grad_norm": 4.67361307144165,
+ "learning_rate": 1.6705110461120343e-05,
+ "loss": 0.3329,
+ "step": 10573
+ },
+ {
+ "epoch": 28.81198910081744,
+ "grad_norm": 4.807560443878174,
+ "learning_rate": 1.670445571264163e-05,
+ "loss": 0.2787,
+ "step": 10574
+ },
+ {
+ "epoch": 28.814713896457764,
+ "grad_norm": 4.988160610198975,
+ "learning_rate": 1.6703800911948825e-05,
+ "loss": 0.3736,
+ "step": 10575
+ },
+ {
+ "epoch": 28.817438692098094,
+ "grad_norm": 4.511654853820801,
+ "learning_rate": 1.6703146059047038e-05,
+ "loss": 0.2881,
+ "step": 10576
+ },
+ {
+ "epoch": 28.82016348773842,
+ "grad_norm": 5.224606037139893,
+ "learning_rate": 1.6702491153941367e-05,
+ "loss": 0.3093,
+ "step": 10577
+ },
+ {
+ "epoch": 28.822888283378745,
+ "grad_norm": 5.154632091522217,
+ "learning_rate": 1.6701836196636908e-05,
+ "loss": 0.3241,
+ "step": 10578
+ },
+ {
+ "epoch": 28.825613079019075,
+ "grad_norm": 7.736479759216309,
+ "learning_rate": 1.6701181187138767e-05,
+ "loss": 0.341,
+ "step": 10579
+ },
+ {
+ "epoch": 28.8283378746594,
+ "grad_norm": 4.560948371887207,
+ "learning_rate": 1.670052612545204e-05,
+ "loss": 0.2939,
+ "step": 10580
+ },
+ {
+ "epoch": 28.831062670299726,
+ "grad_norm": 5.038032054901123,
+ "learning_rate": 1.669987101158183e-05,
+ "loss": 0.3272,
+ "step": 10581
+ },
+ {
+ "epoch": 28.833787465940055,
+ "grad_norm": 4.690295219421387,
+ "learning_rate": 1.6699215845533244e-05,
+ "loss": 0.3158,
+ "step": 10582
+ },
+ {
+ "epoch": 28.83651226158038,
+ "grad_norm": 4.487279891967773,
+ "learning_rate": 1.6698560627311378e-05,
+ "loss": 0.2952,
+ "step": 10583
+ },
+ {
+ "epoch": 28.839237057220707,
+ "grad_norm": 5.106103420257568,
+ "learning_rate": 1.669790535692134e-05,
+ "loss": 0.2833,
+ "step": 10584
+ },
+ {
+ "epoch": 28.841961852861036,
+ "grad_norm": 4.50995397567749,
+ "learning_rate": 1.6697250034368226e-05,
+ "loss": 0.24,
+ "step": 10585
+ },
+ {
+ "epoch": 28.844686648501362,
+ "grad_norm": 4.830857276916504,
+ "learning_rate": 1.669659465965715e-05,
+ "loss": 0.2844,
+ "step": 10586
+ },
+ {
+ "epoch": 28.847411444141688,
+ "grad_norm": 5.037544250488281,
+ "learning_rate": 1.6695939232793206e-05,
+ "loss": 0.3429,
+ "step": 10587
+ },
+ {
+ "epoch": 28.850136239782017,
+ "grad_norm": 5.134025573730469,
+ "learning_rate": 1.66952837537815e-05,
+ "loss": 0.4196,
+ "step": 10588
+ },
+ {
+ "epoch": 28.852861035422343,
+ "grad_norm": 5.968637943267822,
+ "learning_rate": 1.6694628222627148e-05,
+ "loss": 0.3781,
+ "step": 10589
+ },
+ {
+ "epoch": 28.85558583106267,
+ "grad_norm": 5.027859687805176,
+ "learning_rate": 1.669397263933524e-05,
+ "loss": 0.3043,
+ "step": 10590
+ },
+ {
+ "epoch": 28.858310626703,
+ "grad_norm": 6.814409255981445,
+ "learning_rate": 1.6693317003910895e-05,
+ "loss": 0.44,
+ "step": 10591
+ },
+ {
+ "epoch": 28.861035422343324,
+ "grad_norm": 4.105199813842773,
+ "learning_rate": 1.6692661316359206e-05,
+ "loss": 0.2928,
+ "step": 10592
+ },
+ {
+ "epoch": 28.86376021798365,
+ "grad_norm": 4.699253082275391,
+ "learning_rate": 1.6692005576685288e-05,
+ "loss": 0.2207,
+ "step": 10593
+ },
+ {
+ "epoch": 28.86648501362398,
+ "grad_norm": 4.516366004943848,
+ "learning_rate": 1.669134978489425e-05,
+ "loss": 0.3271,
+ "step": 10594
+ },
+ {
+ "epoch": 28.869209809264305,
+ "grad_norm": 4.87831449508667,
+ "learning_rate": 1.669069394099119e-05,
+ "loss": 0.298,
+ "step": 10595
+ },
+ {
+ "epoch": 28.87193460490463,
+ "grad_norm": 4.645625591278076,
+ "learning_rate": 1.6690038044981222e-05,
+ "loss": 0.3162,
+ "step": 10596
+ },
+ {
+ "epoch": 28.87465940054496,
+ "grad_norm": 4.088955402374268,
+ "learning_rate": 1.6689382096869457e-05,
+ "loss": 0.3296,
+ "step": 10597
+ },
+ {
+ "epoch": 28.877384196185286,
+ "grad_norm": 4.882492542266846,
+ "learning_rate": 1.6688726096660994e-05,
+ "loss": 0.302,
+ "step": 10598
+ },
+ {
+ "epoch": 28.88010899182561,
+ "grad_norm": 5.136576175689697,
+ "learning_rate": 1.668807004436095e-05,
+ "loss": 0.2752,
+ "step": 10599
+ },
+ {
+ "epoch": 28.88283378746594,
+ "grad_norm": 5.104892253875732,
+ "learning_rate": 1.668741393997443e-05,
+ "loss": 0.5425,
+ "step": 10600
+ },
+ {
+ "epoch": 28.885558583106267,
+ "grad_norm": 5.060689926147461,
+ "learning_rate": 1.668675778350655e-05,
+ "loss": 0.2498,
+ "step": 10601
+ },
+ {
+ "epoch": 28.888283378746593,
+ "grad_norm": 4.7725958824157715,
+ "learning_rate": 1.6686101574962413e-05,
+ "loss": 0.2534,
+ "step": 10602
+ },
+ {
+ "epoch": 28.891008174386922,
+ "grad_norm": 5.465051174163818,
+ "learning_rate": 1.668544531434713e-05,
+ "loss": 0.3332,
+ "step": 10603
+ },
+ {
+ "epoch": 28.893732970027248,
+ "grad_norm": 5.188478946685791,
+ "learning_rate": 1.668478900166582e-05,
+ "loss": 0.3741,
+ "step": 10604
+ },
+ {
+ "epoch": 28.896457765667574,
+ "grad_norm": 5.617560386657715,
+ "learning_rate": 1.668413263692358e-05,
+ "loss": 0.3208,
+ "step": 10605
+ },
+ {
+ "epoch": 28.899182561307903,
+ "grad_norm": 4.634690761566162,
+ "learning_rate": 1.6683476220125535e-05,
+ "loss": 0.3303,
+ "step": 10606
+ },
+ {
+ "epoch": 28.90190735694823,
+ "grad_norm": 5.558864593505859,
+ "learning_rate": 1.668281975127679e-05,
+ "loss": 0.3262,
+ "step": 10607
+ },
+ {
+ "epoch": 28.904632152588555,
+ "grad_norm": 7.3302106857299805,
+ "learning_rate": 1.6682163230382465e-05,
+ "loss": 0.3019,
+ "step": 10608
+ },
+ {
+ "epoch": 28.907356948228884,
+ "grad_norm": 4.747617721557617,
+ "learning_rate": 1.6681506657447664e-05,
+ "loss": 0.2784,
+ "step": 10609
+ },
+ {
+ "epoch": 28.91008174386921,
+ "grad_norm": 4.667511940002441,
+ "learning_rate": 1.6680850032477504e-05,
+ "loss": 0.2985,
+ "step": 10610
+ },
+ {
+ "epoch": 28.912806539509535,
+ "grad_norm": 4.121743679046631,
+ "learning_rate": 1.66801933554771e-05,
+ "loss": 0.2234,
+ "step": 10611
+ },
+ {
+ "epoch": 28.915531335149865,
+ "grad_norm": 4.304055690765381,
+ "learning_rate": 1.6679536626451565e-05,
+ "loss": 0.273,
+ "step": 10612
+ },
+ {
+ "epoch": 28.91825613079019,
+ "grad_norm": 4.6662211418151855,
+ "learning_rate": 1.6678879845406014e-05,
+ "loss": 0.366,
+ "step": 10613
+ },
+ {
+ "epoch": 28.920980926430516,
+ "grad_norm": 4.341454029083252,
+ "learning_rate": 1.667822301234556e-05,
+ "loss": 0.1501,
+ "step": 10614
+ },
+ {
+ "epoch": 28.923705722070846,
+ "grad_norm": 4.764432907104492,
+ "learning_rate": 1.667756612727532e-05,
+ "loss": 0.4558,
+ "step": 10615
+ },
+ {
+ "epoch": 28.92643051771117,
+ "grad_norm": 4.411827087402344,
+ "learning_rate": 1.6676909190200416e-05,
+ "loss": 0.2117,
+ "step": 10616
+ },
+ {
+ "epoch": 28.929155313351497,
+ "grad_norm": 5.328108310699463,
+ "learning_rate": 1.667625220112595e-05,
+ "loss": 0.3712,
+ "step": 10617
+ },
+ {
+ "epoch": 28.931880108991827,
+ "grad_norm": 5.160487651824951,
+ "learning_rate": 1.6675595160057053e-05,
+ "loss": 0.3834,
+ "step": 10618
+ },
+ {
+ "epoch": 28.934604904632153,
+ "grad_norm": 6.197288990020752,
+ "learning_rate": 1.667493806699883e-05,
+ "loss": 0.4705,
+ "step": 10619
+ },
+ {
+ "epoch": 28.93732970027248,
+ "grad_norm": 4.933932781219482,
+ "learning_rate": 1.667428092195641e-05,
+ "loss": 0.212,
+ "step": 10620
+ },
+ {
+ "epoch": 28.940054495912808,
+ "grad_norm": 5.835660457611084,
+ "learning_rate": 1.6673623724934898e-05,
+ "loss": 0.3557,
+ "step": 10621
+ },
+ {
+ "epoch": 28.942779291553133,
+ "grad_norm": 4.243099212646484,
+ "learning_rate": 1.6672966475939426e-05,
+ "loss": 0.3196,
+ "step": 10622
+ },
+ {
+ "epoch": 28.94550408719346,
+ "grad_norm": 4.951198101043701,
+ "learning_rate": 1.66723091749751e-05,
+ "loss": 0.2316,
+ "step": 10623
+ },
+ {
+ "epoch": 28.94822888283379,
+ "grad_norm": 5.056251049041748,
+ "learning_rate": 1.6671651822047052e-05,
+ "loss": 0.2323,
+ "step": 10624
+ },
+ {
+ "epoch": 28.950953678474114,
+ "grad_norm": 5.313762187957764,
+ "learning_rate": 1.667099441716039e-05,
+ "loss": 0.3093,
+ "step": 10625
+ },
+ {
+ "epoch": 28.95367847411444,
+ "grad_norm": 4.952607154846191,
+ "learning_rate": 1.6670336960320237e-05,
+ "loss": 0.3232,
+ "step": 10626
+ },
+ {
+ "epoch": 28.95640326975477,
+ "grad_norm": 5.180213451385498,
+ "learning_rate": 1.6669679451531716e-05,
+ "loss": 0.2301,
+ "step": 10627
+ },
+ {
+ "epoch": 28.959128065395095,
+ "grad_norm": 4.52487850189209,
+ "learning_rate": 1.666902189079995e-05,
+ "loss": 0.2957,
+ "step": 10628
+ },
+ {
+ "epoch": 28.96185286103542,
+ "grad_norm": 4.240546703338623,
+ "learning_rate": 1.6668364278130047e-05,
+ "loss": 0.2468,
+ "step": 10629
+ },
+ {
+ "epoch": 28.96457765667575,
+ "grad_norm": 4.550522327423096,
+ "learning_rate": 1.6667706613527142e-05,
+ "loss": 0.356,
+ "step": 10630
+ },
+ {
+ "epoch": 28.967302452316076,
+ "grad_norm": 11.549551010131836,
+ "learning_rate": 1.6667048896996357e-05,
+ "loss": 0.2748,
+ "step": 10631
+ },
+ {
+ "epoch": 28.970027247956402,
+ "grad_norm": 4.7759881019592285,
+ "learning_rate": 1.6666391128542802e-05,
+ "loss": 0.2743,
+ "step": 10632
+ },
+ {
+ "epoch": 28.97275204359673,
+ "grad_norm": 6.148135185241699,
+ "learning_rate": 1.6665733308171615e-05,
+ "loss": 0.3392,
+ "step": 10633
+ },
+ {
+ "epoch": 28.975476839237057,
+ "grad_norm": 4.2947096824646,
+ "learning_rate": 1.6665075435887907e-05,
+ "loss": 0.3931,
+ "step": 10634
+ },
+ {
+ "epoch": 28.978201634877383,
+ "grad_norm": 4.679409503936768,
+ "learning_rate": 1.6664417511696807e-05,
+ "loss": 0.2484,
+ "step": 10635
+ },
+ {
+ "epoch": 28.980926430517712,
+ "grad_norm": 5.737270832061768,
+ "learning_rate": 1.6663759535603438e-05,
+ "loss": 0.27,
+ "step": 10636
+ },
+ {
+ "epoch": 28.983651226158038,
+ "grad_norm": 4.431032657623291,
+ "learning_rate": 1.6663101507612926e-05,
+ "loss": 0.2521,
+ "step": 10637
+ },
+ {
+ "epoch": 28.986376021798364,
+ "grad_norm": 4.933211803436279,
+ "learning_rate": 1.666244342773039e-05,
+ "loss": 0.2679,
+ "step": 10638
+ },
+ {
+ "epoch": 28.989100817438693,
+ "grad_norm": 5.457597255706787,
+ "learning_rate": 1.666178529596096e-05,
+ "loss": 0.38,
+ "step": 10639
+ },
+ {
+ "epoch": 28.99182561307902,
+ "grad_norm": 5.526261329650879,
+ "learning_rate": 1.666112711230976e-05,
+ "loss": 0.2724,
+ "step": 10640
+ },
+ {
+ "epoch": 28.994550408719345,
+ "grad_norm": 4.967635154724121,
+ "learning_rate": 1.666046887678192e-05,
+ "loss": 0.5058,
+ "step": 10641
+ },
+ {
+ "epoch": 28.997275204359674,
+ "grad_norm": 6.201898097991943,
+ "learning_rate": 1.665981058938256e-05,
+ "loss": 0.4775,
+ "step": 10642
+ },
+ {
+ "epoch": 29.0,
+ "grad_norm": 5.573678493499756,
+ "learning_rate": 1.665915225011681e-05,
+ "loss": 0.3667,
+ "step": 10643
+ },
+ {
+ "epoch": 29.002724795640326,
+ "grad_norm": 4.409809589385986,
+ "learning_rate": 1.66584938589898e-05,
+ "loss": 0.3792,
+ "step": 10644
+ },
+ {
+ "epoch": 29.005449591280655,
+ "grad_norm": 4.357861518859863,
+ "learning_rate": 1.665783541600665e-05,
+ "loss": 0.1969,
+ "step": 10645
+ },
+ {
+ "epoch": 29.00817438692098,
+ "grad_norm": 4.6477203369140625,
+ "learning_rate": 1.6657176921172492e-05,
+ "loss": 0.385,
+ "step": 10646
+ },
+ {
+ "epoch": 29.010899182561307,
+ "grad_norm": 4.680109024047852,
+ "learning_rate": 1.6656518374492458e-05,
+ "loss": 0.2393,
+ "step": 10647
+ },
+ {
+ "epoch": 29.013623978201636,
+ "grad_norm": 4.647577285766602,
+ "learning_rate": 1.665585977597167e-05,
+ "loss": 0.2409,
+ "step": 10648
+ },
+ {
+ "epoch": 29.016348773841962,
+ "grad_norm": 4.5102996826171875,
+ "learning_rate": 1.6655201125615263e-05,
+ "loss": 0.3117,
+ "step": 10649
+ },
+ {
+ "epoch": 29.019073569482288,
+ "grad_norm": 5.007046699523926,
+ "learning_rate": 1.665454242342836e-05,
+ "loss": 0.3336,
+ "step": 10650
+ },
+ {
+ "epoch": 29.021798365122617,
+ "grad_norm": 4.354625225067139,
+ "learning_rate": 1.6653883669416098e-05,
+ "loss": 0.4073,
+ "step": 10651
+ },
+ {
+ "epoch": 29.024523160762943,
+ "grad_norm": 4.620046615600586,
+ "learning_rate": 1.6653224863583602e-05,
+ "loss": 0.3358,
+ "step": 10652
+ },
+ {
+ "epoch": 29.02724795640327,
+ "grad_norm": 3.8112614154815674,
+ "learning_rate": 1.6652566005936004e-05,
+ "loss": 0.2183,
+ "step": 10653
+ },
+ {
+ "epoch": 29.029972752043598,
+ "grad_norm": 4.3881049156188965,
+ "learning_rate": 1.6651907096478436e-05,
+ "loss": 0.217,
+ "step": 10654
+ },
+ {
+ "epoch": 29.032697547683924,
+ "grad_norm": 5.141610145568848,
+ "learning_rate": 1.6651248135216032e-05,
+ "loss": 0.2437,
+ "step": 10655
+ },
+ {
+ "epoch": 29.03542234332425,
+ "grad_norm": 6.407204627990723,
+ "learning_rate": 1.665058912215392e-05,
+ "loss": 0.2294,
+ "step": 10656
+ },
+ {
+ "epoch": 29.03814713896458,
+ "grad_norm": 4.202515125274658,
+ "learning_rate": 1.6649930057297235e-05,
+ "loss": 0.1755,
+ "step": 10657
+ },
+ {
+ "epoch": 29.040871934604905,
+ "grad_norm": 6.4071221351623535,
+ "learning_rate": 1.664927094065111e-05,
+ "loss": 0.2688,
+ "step": 10658
+ },
+ {
+ "epoch": 29.04359673024523,
+ "grad_norm": 5.92539644241333,
+ "learning_rate": 1.6648611772220675e-05,
+ "loss": 0.218,
+ "step": 10659
+ },
+ {
+ "epoch": 29.04632152588556,
+ "grad_norm": 5.513072490692139,
+ "learning_rate": 1.664795255201107e-05,
+ "loss": 0.3838,
+ "step": 10660
+ },
+ {
+ "epoch": 29.049046321525886,
+ "grad_norm": 4.837181091308594,
+ "learning_rate": 1.664729328002742e-05,
+ "loss": 0.2659,
+ "step": 10661
+ },
+ {
+ "epoch": 29.05177111716621,
+ "grad_norm": 3.9357354640960693,
+ "learning_rate": 1.6646633956274862e-05,
+ "loss": 0.5664,
+ "step": 10662
+ },
+ {
+ "epoch": 29.05449591280654,
+ "grad_norm": 4.515839576721191,
+ "learning_rate": 1.6645974580758533e-05,
+ "loss": 0.212,
+ "step": 10663
+ },
+ {
+ "epoch": 29.057220708446867,
+ "grad_norm": 4.504817962646484,
+ "learning_rate": 1.6645315153483572e-05,
+ "loss": 0.4771,
+ "step": 10664
+ },
+ {
+ "epoch": 29.059945504087192,
+ "grad_norm": 5.760124206542969,
+ "learning_rate": 1.6644655674455107e-05,
+ "loss": 0.2305,
+ "step": 10665
+ },
+ {
+ "epoch": 29.06267029972752,
+ "grad_norm": 4.174933433532715,
+ "learning_rate": 1.664399614367828e-05,
+ "loss": 0.3594,
+ "step": 10666
+ },
+ {
+ "epoch": 29.065395095367847,
+ "grad_norm": 4.4070916175842285,
+ "learning_rate": 1.6643336561158222e-05,
+ "loss": 0.2462,
+ "step": 10667
+ },
+ {
+ "epoch": 29.068119891008173,
+ "grad_norm": 4.928384780883789,
+ "learning_rate": 1.6642676926900077e-05,
+ "loss": 0.284,
+ "step": 10668
+ },
+ {
+ "epoch": 29.070844686648503,
+ "grad_norm": 4.624230861663818,
+ "learning_rate": 1.6642017240908975e-05,
+ "loss": 0.1959,
+ "step": 10669
+ },
+ {
+ "epoch": 29.07356948228883,
+ "grad_norm": 4.612250328063965,
+ "learning_rate": 1.6641357503190055e-05,
+ "loss": 0.2986,
+ "step": 10670
+ },
+ {
+ "epoch": 29.076294277929154,
+ "grad_norm": 5.5595927238464355,
+ "learning_rate": 1.664069771374846e-05,
+ "loss": 0.2434,
+ "step": 10671
+ },
+ {
+ "epoch": 29.079019073569484,
+ "grad_norm": 3.848118543624878,
+ "learning_rate": 1.664003787258932e-05,
+ "loss": 0.157,
+ "step": 10672
+ },
+ {
+ "epoch": 29.08174386920981,
+ "grad_norm": 5.645239353179932,
+ "learning_rate": 1.6639377979717784e-05,
+ "loss": 0.2822,
+ "step": 10673
+ },
+ {
+ "epoch": 29.084468664850135,
+ "grad_norm": 4.704919338226318,
+ "learning_rate": 1.6638718035138985e-05,
+ "loss": 0.2761,
+ "step": 10674
+ },
+ {
+ "epoch": 29.087193460490465,
+ "grad_norm": 4.7690749168396,
+ "learning_rate": 1.663805803885806e-05,
+ "loss": 0.3558,
+ "step": 10675
+ },
+ {
+ "epoch": 29.08991825613079,
+ "grad_norm": 6.144473075866699,
+ "learning_rate": 1.6637397990880158e-05,
+ "loss": 0.2498,
+ "step": 10676
+ },
+ {
+ "epoch": 29.092643051771116,
+ "grad_norm": 4.6573262214660645,
+ "learning_rate": 1.6636737891210412e-05,
+ "loss": 0.2613,
+ "step": 10677
+ },
+ {
+ "epoch": 29.095367847411445,
+ "grad_norm": 4.908087730407715,
+ "learning_rate": 1.663607773985396e-05,
+ "loss": 0.2978,
+ "step": 10678
+ },
+ {
+ "epoch": 29.09809264305177,
+ "grad_norm": 7.011667251586914,
+ "learning_rate": 1.6635417536815953e-05,
+ "loss": 0.3942,
+ "step": 10679
+ },
+ {
+ "epoch": 29.100817438692097,
+ "grad_norm": 4.621870994567871,
+ "learning_rate": 1.663475728210153e-05,
+ "loss": 0.3307,
+ "step": 10680
+ },
+ {
+ "epoch": 29.103542234332426,
+ "grad_norm": 5.011868476867676,
+ "learning_rate": 1.6634096975715825e-05,
+ "loss": 0.3797,
+ "step": 10681
+ },
+ {
+ "epoch": 29.106267029972752,
+ "grad_norm": 5.321270942687988,
+ "learning_rate": 1.6633436617663988e-05,
+ "loss": 0.2447,
+ "step": 10682
+ },
+ {
+ "epoch": 29.108991825613078,
+ "grad_norm": 4.666157245635986,
+ "learning_rate": 1.6632776207951165e-05,
+ "loss": 0.2391,
+ "step": 10683
+ },
+ {
+ "epoch": 29.111716621253407,
+ "grad_norm": 4.329449653625488,
+ "learning_rate": 1.663211574658249e-05,
+ "loss": 0.2512,
+ "step": 10684
+ },
+ {
+ "epoch": 29.114441416893733,
+ "grad_norm": 4.796196937561035,
+ "learning_rate": 1.6631455233563113e-05,
+ "loss": 0.2652,
+ "step": 10685
+ },
+ {
+ "epoch": 29.11716621253406,
+ "grad_norm": 4.516221046447754,
+ "learning_rate": 1.6630794668898175e-05,
+ "loss": 0.2433,
+ "step": 10686
+ },
+ {
+ "epoch": 29.11989100817439,
+ "grad_norm": 5.4271392822265625,
+ "learning_rate": 1.6630134052592823e-05,
+ "loss": 0.3355,
+ "step": 10687
+ },
+ {
+ "epoch": 29.122615803814714,
+ "grad_norm": 4.4666619300842285,
+ "learning_rate": 1.6629473384652197e-05,
+ "loss": 0.2941,
+ "step": 10688
+ },
+ {
+ "epoch": 29.12534059945504,
+ "grad_norm": 5.648801803588867,
+ "learning_rate": 1.6628812665081446e-05,
+ "loss": 0.2358,
+ "step": 10689
+ },
+ {
+ "epoch": 29.12806539509537,
+ "grad_norm": 5.1695356369018555,
+ "learning_rate": 1.6628151893885715e-05,
+ "loss": 0.3057,
+ "step": 10690
+ },
+ {
+ "epoch": 29.130790190735695,
+ "grad_norm": 4.525134563446045,
+ "learning_rate": 1.6627491071070152e-05,
+ "loss": 0.2269,
+ "step": 10691
+ },
+ {
+ "epoch": 29.13351498637602,
+ "grad_norm": 6.250156402587891,
+ "learning_rate": 1.66268301966399e-05,
+ "loss": 0.2892,
+ "step": 10692
+ },
+ {
+ "epoch": 29.13623978201635,
+ "grad_norm": 4.478301048278809,
+ "learning_rate": 1.6626169270600113e-05,
+ "loss": 0.4275,
+ "step": 10693
+ },
+ {
+ "epoch": 29.138964577656676,
+ "grad_norm": 4.5793280601501465,
+ "learning_rate": 1.662550829295593e-05,
+ "loss": 0.2605,
+ "step": 10694
+ },
+ {
+ "epoch": 29.141689373297,
+ "grad_norm": 4.693602561950684,
+ "learning_rate": 1.66248472637125e-05,
+ "loss": 0.3068,
+ "step": 10695
+ },
+ {
+ "epoch": 29.14441416893733,
+ "grad_norm": 5.430422306060791,
+ "learning_rate": 1.6624186182874977e-05,
+ "loss": 0.2267,
+ "step": 10696
+ },
+ {
+ "epoch": 29.147138964577657,
+ "grad_norm": 4.910060882568359,
+ "learning_rate": 1.66235250504485e-05,
+ "loss": 0.2775,
+ "step": 10697
+ },
+ {
+ "epoch": 29.149863760217983,
+ "grad_norm": 4.249905109405518,
+ "learning_rate": 1.6622863866438223e-05,
+ "loss": 0.2487,
+ "step": 10698
+ },
+ {
+ "epoch": 29.152588555858312,
+ "grad_norm": 4.378042221069336,
+ "learning_rate": 1.66222026308493e-05,
+ "loss": 0.3288,
+ "step": 10699
+ },
+ {
+ "epoch": 29.155313351498638,
+ "grad_norm": 4.013815879821777,
+ "learning_rate": 1.662154134368687e-05,
+ "loss": 0.1836,
+ "step": 10700
+ },
+ {
+ "epoch": 29.158038147138964,
+ "grad_norm": 5.305545330047607,
+ "learning_rate": 1.6620880004956095e-05,
+ "loss": 0.1853,
+ "step": 10701
+ },
+ {
+ "epoch": 29.160762942779293,
+ "grad_norm": 5.167050361633301,
+ "learning_rate": 1.662021861466212e-05,
+ "loss": 0.3472,
+ "step": 10702
+ },
+ {
+ "epoch": 29.16348773841962,
+ "grad_norm": 4.497769355773926,
+ "learning_rate": 1.661955717281009e-05,
+ "loss": 0.3297,
+ "step": 10703
+ },
+ {
+ "epoch": 29.166212534059945,
+ "grad_norm": 4.396137714385986,
+ "learning_rate": 1.6618895679405165e-05,
+ "loss": 0.2783,
+ "step": 10704
+ },
+ {
+ "epoch": 29.168937329700274,
+ "grad_norm": 6.528575420379639,
+ "learning_rate": 1.6618234134452493e-05,
+ "loss": 0.2083,
+ "step": 10705
+ },
+ {
+ "epoch": 29.1716621253406,
+ "grad_norm": 4.273553371429443,
+ "learning_rate": 1.661757253795723e-05,
+ "loss": 0.274,
+ "step": 10706
+ },
+ {
+ "epoch": 29.174386920980925,
+ "grad_norm": 4.475641250610352,
+ "learning_rate": 1.661691088992452e-05,
+ "loss": 0.25,
+ "step": 10707
+ },
+ {
+ "epoch": 29.177111716621255,
+ "grad_norm": 4.643335342407227,
+ "learning_rate": 1.661624919035952e-05,
+ "loss": 0.216,
+ "step": 10708
+ },
+ {
+ "epoch": 29.17983651226158,
+ "grad_norm": 5.861490726470947,
+ "learning_rate": 1.661558743926739e-05,
+ "loss": 0.3279,
+ "step": 10709
+ },
+ {
+ "epoch": 29.182561307901906,
+ "grad_norm": 5.081967830657959,
+ "learning_rate": 1.6614925636653274e-05,
+ "loss": 0.3434,
+ "step": 10710
+ },
+ {
+ "epoch": 29.185286103542236,
+ "grad_norm": 4.124058723449707,
+ "learning_rate": 1.6614263782522332e-05,
+ "loss": 0.3114,
+ "step": 10711
+ },
+ {
+ "epoch": 29.18801089918256,
+ "grad_norm": 4.159549236297607,
+ "learning_rate": 1.6613601876879714e-05,
+ "loss": 0.3418,
+ "step": 10712
+ },
+ {
+ "epoch": 29.190735694822887,
+ "grad_norm": 4.578546047210693,
+ "learning_rate": 1.6612939919730578e-05,
+ "loss": 0.3005,
+ "step": 10713
+ },
+ {
+ "epoch": 29.193460490463217,
+ "grad_norm": 5.799752712249756,
+ "learning_rate": 1.661227791108008e-05,
+ "loss": 0.2864,
+ "step": 10714
+ },
+ {
+ "epoch": 29.196185286103542,
+ "grad_norm": 4.110982418060303,
+ "learning_rate": 1.6611615850933374e-05,
+ "loss": 0.2217,
+ "step": 10715
+ },
+ {
+ "epoch": 29.19891008174387,
+ "grad_norm": 6.603341579437256,
+ "learning_rate": 1.6610953739295617e-05,
+ "loss": 0.2925,
+ "step": 10716
+ },
+ {
+ "epoch": 29.201634877384198,
+ "grad_norm": 12.347016334533691,
+ "learning_rate": 1.6610291576171965e-05,
+ "loss": 0.2748,
+ "step": 10717
+ },
+ {
+ "epoch": 29.204359673024523,
+ "grad_norm": 4.486969947814941,
+ "learning_rate": 1.6609629361567578e-05,
+ "loss": 0.1712,
+ "step": 10718
+ },
+ {
+ "epoch": 29.20708446866485,
+ "grad_norm": 4.564652442932129,
+ "learning_rate": 1.6608967095487606e-05,
+ "loss": 0.3242,
+ "step": 10719
+ },
+ {
+ "epoch": 29.20980926430518,
+ "grad_norm": 3.900678873062134,
+ "learning_rate": 1.660830477793721e-05,
+ "loss": 0.3123,
+ "step": 10720
+ },
+ {
+ "epoch": 29.212534059945504,
+ "grad_norm": 4.413665294647217,
+ "learning_rate": 1.6607642408921552e-05,
+ "loss": 0.2185,
+ "step": 10721
+ },
+ {
+ "epoch": 29.21525885558583,
+ "grad_norm": 13.350144386291504,
+ "learning_rate": 1.6606979988445787e-05,
+ "loss": 0.2085,
+ "step": 10722
+ },
+ {
+ "epoch": 29.21798365122616,
+ "grad_norm": 5.76389217376709,
+ "learning_rate": 1.6606317516515074e-05,
+ "loss": 0.3489,
+ "step": 10723
+ },
+ {
+ "epoch": 29.220708446866485,
+ "grad_norm": 3.5523202419281006,
+ "learning_rate": 1.660565499313457e-05,
+ "loss": 0.1707,
+ "step": 10724
+ },
+ {
+ "epoch": 29.22343324250681,
+ "grad_norm": 5.432359218597412,
+ "learning_rate": 1.6604992418309443e-05,
+ "loss": 0.1822,
+ "step": 10725
+ },
+ {
+ "epoch": 29.22615803814714,
+ "grad_norm": 5.578645706176758,
+ "learning_rate": 1.6604329792044845e-05,
+ "loss": 0.2352,
+ "step": 10726
+ },
+ {
+ "epoch": 29.228882833787466,
+ "grad_norm": 4.744897842407227,
+ "learning_rate": 1.660366711434594e-05,
+ "loss": 0.1956,
+ "step": 10727
+ },
+ {
+ "epoch": 29.231607629427792,
+ "grad_norm": 4.719455242156982,
+ "learning_rate": 1.6603004385217887e-05,
+ "loss": 0.2151,
+ "step": 10728
+ },
+ {
+ "epoch": 29.23433242506812,
+ "grad_norm": 4.050261974334717,
+ "learning_rate": 1.6602341604665848e-05,
+ "loss": 0.3548,
+ "step": 10729
+ },
+ {
+ "epoch": 29.237057220708447,
+ "grad_norm": 4.452539920806885,
+ "learning_rate": 1.6601678772694985e-05,
+ "loss": 0.3203,
+ "step": 10730
+ },
+ {
+ "epoch": 29.239782016348773,
+ "grad_norm": 5.257819652557373,
+ "learning_rate": 1.6601015889310462e-05,
+ "loss": 0.3104,
+ "step": 10731
+ },
+ {
+ "epoch": 29.242506811989102,
+ "grad_norm": 4.579367637634277,
+ "learning_rate": 1.6600352954517437e-05,
+ "loss": 0.3036,
+ "step": 10732
+ },
+ {
+ "epoch": 29.245231607629428,
+ "grad_norm": 4.187880516052246,
+ "learning_rate": 1.659968996832108e-05,
+ "loss": 0.2945,
+ "step": 10733
+ },
+ {
+ "epoch": 29.247956403269754,
+ "grad_norm": 14.792920112609863,
+ "learning_rate": 1.6599026930726544e-05,
+ "loss": 0.2251,
+ "step": 10734
+ },
+ {
+ "epoch": 29.250681198910083,
+ "grad_norm": 17.46099853515625,
+ "learning_rate": 1.6598363841739004e-05,
+ "loss": 0.2473,
+ "step": 10735
+ },
+ {
+ "epoch": 29.25340599455041,
+ "grad_norm": 6.892497539520264,
+ "learning_rate": 1.6597700701363615e-05,
+ "loss": 0.2846,
+ "step": 10736
+ },
+ {
+ "epoch": 29.256130790190735,
+ "grad_norm": 4.994281768798828,
+ "learning_rate": 1.6597037509605546e-05,
+ "loss": 0.3468,
+ "step": 10737
+ },
+ {
+ "epoch": 29.258855585831064,
+ "grad_norm": 3.9868505001068115,
+ "learning_rate": 1.6596374266469963e-05,
+ "loss": 0.2867,
+ "step": 10738
+ },
+ {
+ "epoch": 29.26158038147139,
+ "grad_norm": 6.189479827880859,
+ "learning_rate": 1.6595710971962028e-05,
+ "loss": 0.3435,
+ "step": 10739
+ },
+ {
+ "epoch": 29.264305177111716,
+ "grad_norm": 4.441372871398926,
+ "learning_rate": 1.6595047626086908e-05,
+ "loss": 0.188,
+ "step": 10740
+ },
+ {
+ "epoch": 29.267029972752045,
+ "grad_norm": 5.144440174102783,
+ "learning_rate": 1.659438422884977e-05,
+ "loss": 0.4009,
+ "step": 10741
+ },
+ {
+ "epoch": 29.26975476839237,
+ "grad_norm": 4.555542945861816,
+ "learning_rate": 1.6593720780255778e-05,
+ "loss": 0.1919,
+ "step": 10742
+ },
+ {
+ "epoch": 29.272479564032697,
+ "grad_norm": 4.989017009735107,
+ "learning_rate": 1.6593057280310105e-05,
+ "loss": 0.2068,
+ "step": 10743
+ },
+ {
+ "epoch": 29.275204359673026,
+ "grad_norm": 4.1951141357421875,
+ "learning_rate": 1.659239372901791e-05,
+ "loss": 0.2275,
+ "step": 10744
+ },
+ {
+ "epoch": 29.277929155313352,
+ "grad_norm": 8.888466835021973,
+ "learning_rate": 1.659173012638437e-05,
+ "loss": 0.2766,
+ "step": 10745
+ },
+ {
+ "epoch": 29.280653950953678,
+ "grad_norm": 6.274106502532959,
+ "learning_rate": 1.6591066472414647e-05,
+ "loss": 0.3132,
+ "step": 10746
+ },
+ {
+ "epoch": 29.283378746594007,
+ "grad_norm": 5.166018962860107,
+ "learning_rate": 1.6590402767113903e-05,
+ "loss": 0.4216,
+ "step": 10747
+ },
+ {
+ "epoch": 29.286103542234333,
+ "grad_norm": 4.076252460479736,
+ "learning_rate": 1.6589739010487323e-05,
+ "loss": 0.3588,
+ "step": 10748
+ },
+ {
+ "epoch": 29.28882833787466,
+ "grad_norm": 8.13626766204834,
+ "learning_rate": 1.6589075202540064e-05,
+ "loss": 0.2346,
+ "step": 10749
+ },
+ {
+ "epoch": 29.291553133514988,
+ "grad_norm": 4.500086307525635,
+ "learning_rate": 1.65884113432773e-05,
+ "loss": 0.2599,
+ "step": 10750
+ },
+ {
+ "epoch": 29.294277929155314,
+ "grad_norm": 4.3430562019348145,
+ "learning_rate": 1.65877474327042e-05,
+ "loss": 0.2262,
+ "step": 10751
+ },
+ {
+ "epoch": 29.29700272479564,
+ "grad_norm": 7.059872150421143,
+ "learning_rate": 1.6587083470825933e-05,
+ "loss": 0.2176,
+ "step": 10752
+ },
+ {
+ "epoch": 29.29972752043597,
+ "grad_norm": 4.143370628356934,
+ "learning_rate": 1.6586419457647673e-05,
+ "loss": 0.2562,
+ "step": 10753
+ },
+ {
+ "epoch": 29.302452316076295,
+ "grad_norm": 4.943517208099365,
+ "learning_rate": 1.658575539317459e-05,
+ "loss": 0.2763,
+ "step": 10754
+ },
+ {
+ "epoch": 29.30517711171662,
+ "grad_norm": 4.132272720336914,
+ "learning_rate": 1.658509127741186e-05,
+ "loss": 0.2157,
+ "step": 10755
+ },
+ {
+ "epoch": 29.30790190735695,
+ "grad_norm": 4.736098289489746,
+ "learning_rate": 1.658442711036465e-05,
+ "loss": 0.337,
+ "step": 10756
+ },
+ {
+ "epoch": 29.310626702997276,
+ "grad_norm": 4.164719104766846,
+ "learning_rate": 1.658376289203813e-05,
+ "loss": 0.4118,
+ "step": 10757
+ },
+ {
+ "epoch": 29.3133514986376,
+ "grad_norm": 7.756408214569092,
+ "learning_rate": 1.6583098622437477e-05,
+ "loss": 0.3152,
+ "step": 10758
+ },
+ {
+ "epoch": 29.31607629427793,
+ "grad_norm": 5.1158671379089355,
+ "learning_rate": 1.6582434301567865e-05,
+ "loss": 0.3603,
+ "step": 10759
+ },
+ {
+ "epoch": 29.318801089918257,
+ "grad_norm": 4.905673980712891,
+ "learning_rate": 1.6581769929434467e-05,
+ "loss": 0.2895,
+ "step": 10760
+ },
+ {
+ "epoch": 29.321525885558582,
+ "grad_norm": 5.973021030426025,
+ "learning_rate": 1.6581105506042456e-05,
+ "loss": 0.2995,
+ "step": 10761
+ },
+ {
+ "epoch": 29.32425068119891,
+ "grad_norm": 4.100625038146973,
+ "learning_rate": 1.6580441031397008e-05,
+ "loss": 0.2262,
+ "step": 10762
+ },
+ {
+ "epoch": 29.326975476839237,
+ "grad_norm": 4.039339542388916,
+ "learning_rate": 1.6579776505503294e-05,
+ "loss": 0.209,
+ "step": 10763
+ },
+ {
+ "epoch": 29.329700272479563,
+ "grad_norm": 4.909232139587402,
+ "learning_rate": 1.65791119283665e-05,
+ "loss": 0.1891,
+ "step": 10764
+ },
+ {
+ "epoch": 29.332425068119893,
+ "grad_norm": 5.693903923034668,
+ "learning_rate": 1.6578447299991788e-05,
+ "loss": 0.2485,
+ "step": 10765
+ },
+ {
+ "epoch": 29.33514986376022,
+ "grad_norm": 4.744733810424805,
+ "learning_rate": 1.657778262038434e-05,
+ "loss": 0.2384,
+ "step": 10766
+ },
+ {
+ "epoch": 29.337874659400544,
+ "grad_norm": 5.42053747177124,
+ "learning_rate": 1.657711788954933e-05,
+ "loss": 0.3431,
+ "step": 10767
+ },
+ {
+ "epoch": 29.340599455040874,
+ "grad_norm": 5.004268169403076,
+ "learning_rate": 1.6576453107491945e-05,
+ "loss": 0.2023,
+ "step": 10768
+ },
+ {
+ "epoch": 29.3433242506812,
+ "grad_norm": 4.92340612411499,
+ "learning_rate": 1.6575788274217347e-05,
+ "loss": 0.2095,
+ "step": 10769
+ },
+ {
+ "epoch": 29.346049046321525,
+ "grad_norm": 4.904883861541748,
+ "learning_rate": 1.6575123389730728e-05,
+ "loss": 0.2415,
+ "step": 10770
+ },
+ {
+ "epoch": 29.348773841961854,
+ "grad_norm": 4.246276378631592,
+ "learning_rate": 1.6574458454037256e-05,
+ "loss": 0.2574,
+ "step": 10771
+ },
+ {
+ "epoch": 29.35149863760218,
+ "grad_norm": 4.460490703582764,
+ "learning_rate": 1.6573793467142114e-05,
+ "loss": 0.3445,
+ "step": 10772
+ },
+ {
+ "epoch": 29.354223433242506,
+ "grad_norm": 4.299620628356934,
+ "learning_rate": 1.657312842905048e-05,
+ "loss": 0.3438,
+ "step": 10773
+ },
+ {
+ "epoch": 29.356948228882835,
+ "grad_norm": 4.617226600646973,
+ "learning_rate": 1.6572463339767537e-05,
+ "loss": 0.2054,
+ "step": 10774
+ },
+ {
+ "epoch": 29.35967302452316,
+ "grad_norm": 4.526733875274658,
+ "learning_rate": 1.6571798199298457e-05,
+ "loss": 0.2741,
+ "step": 10775
+ },
+ {
+ "epoch": 29.362397820163487,
+ "grad_norm": 18.28742218017578,
+ "learning_rate": 1.6571133007648424e-05,
+ "loss": 0.3811,
+ "step": 10776
+ },
+ {
+ "epoch": 29.365122615803816,
+ "grad_norm": 5.967721462249756,
+ "learning_rate": 1.657046776482262e-05,
+ "loss": 0.3652,
+ "step": 10777
+ },
+ {
+ "epoch": 29.367847411444142,
+ "grad_norm": 3.941518545150757,
+ "learning_rate": 1.6569802470826226e-05,
+ "loss": 0.4325,
+ "step": 10778
+ },
+ {
+ "epoch": 29.370572207084468,
+ "grad_norm": 4.822251319885254,
+ "learning_rate": 1.6569137125664417e-05,
+ "loss": 0.3002,
+ "step": 10779
+ },
+ {
+ "epoch": 29.373297002724797,
+ "grad_norm": 4.902225017547607,
+ "learning_rate": 1.6568471729342384e-05,
+ "loss": 0.2798,
+ "step": 10780
+ },
+ {
+ "epoch": 29.376021798365123,
+ "grad_norm": 4.67047119140625,
+ "learning_rate": 1.6567806281865305e-05,
+ "loss": 0.2912,
+ "step": 10781
+ },
+ {
+ "epoch": 29.37874659400545,
+ "grad_norm": 24.605117797851562,
+ "learning_rate": 1.6567140783238358e-05,
+ "loss": 0.3201,
+ "step": 10782
+ },
+ {
+ "epoch": 29.381471389645778,
+ "grad_norm": 5.496262073516846,
+ "learning_rate": 1.6566475233466732e-05,
+ "loss": 0.2899,
+ "step": 10783
+ },
+ {
+ "epoch": 29.384196185286104,
+ "grad_norm": 4.705506324768066,
+ "learning_rate": 1.6565809632555607e-05,
+ "loss": 0.3082,
+ "step": 10784
+ },
+ {
+ "epoch": 29.38692098092643,
+ "grad_norm": 4.5692243576049805,
+ "learning_rate": 1.656514398051017e-05,
+ "loss": 0.3472,
+ "step": 10785
+ },
+ {
+ "epoch": 29.38964577656676,
+ "grad_norm": 5.057797908782959,
+ "learning_rate": 1.65644782773356e-05,
+ "loss": 0.2461,
+ "step": 10786
+ },
+ {
+ "epoch": 29.392370572207085,
+ "grad_norm": 6.020145416259766,
+ "learning_rate": 1.6563812523037087e-05,
+ "loss": 0.2049,
+ "step": 10787
+ },
+ {
+ "epoch": 29.39509536784741,
+ "grad_norm": 5.057622909545898,
+ "learning_rate": 1.6563146717619814e-05,
+ "loss": 0.2621,
+ "step": 10788
+ },
+ {
+ "epoch": 29.39782016348774,
+ "grad_norm": 4.980249404907227,
+ "learning_rate": 1.6562480861088966e-05,
+ "loss": 0.325,
+ "step": 10789
+ },
+ {
+ "epoch": 29.400544959128066,
+ "grad_norm": 4.552181243896484,
+ "learning_rate": 1.6561814953449728e-05,
+ "loss": 0.269,
+ "step": 10790
+ },
+ {
+ "epoch": 29.40326975476839,
+ "grad_norm": 6.426363945007324,
+ "learning_rate": 1.6561148994707285e-05,
+ "loss": 0.2704,
+ "step": 10791
+ },
+ {
+ "epoch": 29.40599455040872,
+ "grad_norm": 5.5157036781311035,
+ "learning_rate": 1.6560482984866823e-05,
+ "loss": 0.387,
+ "step": 10792
+ },
+ {
+ "epoch": 29.408719346049047,
+ "grad_norm": 5.275238037109375,
+ "learning_rate": 1.655981692393353e-05,
+ "loss": 0.3319,
+ "step": 10793
+ },
+ {
+ "epoch": 29.411444141689373,
+ "grad_norm": 4.8232741355896,
+ "learning_rate": 1.65591508119126e-05,
+ "loss": 0.5414,
+ "step": 10794
+ },
+ {
+ "epoch": 29.414168937329702,
+ "grad_norm": 4.8101582527160645,
+ "learning_rate": 1.655848464880921e-05,
+ "loss": 0.21,
+ "step": 10795
+ },
+ {
+ "epoch": 29.416893732970028,
+ "grad_norm": 4.364476203918457,
+ "learning_rate": 1.6557818434628555e-05,
+ "loss": 0.2118,
+ "step": 10796
+ },
+ {
+ "epoch": 29.419618528610354,
+ "grad_norm": 4.960760593414307,
+ "learning_rate": 1.655715216937582e-05,
+ "loss": 0.3261,
+ "step": 10797
+ },
+ {
+ "epoch": 29.422343324250683,
+ "grad_norm": 4.365078926086426,
+ "learning_rate": 1.6556485853056194e-05,
+ "loss": 0.2915,
+ "step": 10798
+ },
+ {
+ "epoch": 29.42506811989101,
+ "grad_norm": 5.93223237991333,
+ "learning_rate": 1.6555819485674865e-05,
+ "loss": 0.3684,
+ "step": 10799
+ },
+ {
+ "epoch": 29.427792915531334,
+ "grad_norm": 5.588153839111328,
+ "learning_rate": 1.655515306723703e-05,
+ "loss": 0.2351,
+ "step": 10800
+ },
+ {
+ "epoch": 29.430517711171664,
+ "grad_norm": 3.7630069255828857,
+ "learning_rate": 1.655448659774787e-05,
+ "loss": 0.2778,
+ "step": 10801
+ },
+ {
+ "epoch": 29.43324250681199,
+ "grad_norm": 5.143671989440918,
+ "learning_rate": 1.6553820077212585e-05,
+ "loss": 0.2849,
+ "step": 10802
+ },
+ {
+ "epoch": 29.435967302452315,
+ "grad_norm": 5.096506595611572,
+ "learning_rate": 1.6553153505636356e-05,
+ "loss": 0.2982,
+ "step": 10803
+ },
+ {
+ "epoch": 29.438692098092645,
+ "grad_norm": 4.355581283569336,
+ "learning_rate": 1.6552486883024376e-05,
+ "loss": 0.1787,
+ "step": 10804
+ },
+ {
+ "epoch": 29.44141689373297,
+ "grad_norm": 4.73683500289917,
+ "learning_rate": 1.6551820209381843e-05,
+ "loss": 0.2645,
+ "step": 10805
+ },
+ {
+ "epoch": 29.444141689373296,
+ "grad_norm": 5.104336261749268,
+ "learning_rate": 1.6551153484713943e-05,
+ "loss": 0.3414,
+ "step": 10806
+ },
+ {
+ "epoch": 29.446866485013626,
+ "grad_norm": 4.441556453704834,
+ "learning_rate": 1.655048670902587e-05,
+ "loss": 0.2029,
+ "step": 10807
+ },
+ {
+ "epoch": 29.44959128065395,
+ "grad_norm": 4.959859848022461,
+ "learning_rate": 1.654981988232282e-05,
+ "loss": 0.2789,
+ "step": 10808
+ },
+ {
+ "epoch": 29.452316076294277,
+ "grad_norm": 5.695981502532959,
+ "learning_rate": 1.6549153004609982e-05,
+ "loss": 0.2983,
+ "step": 10809
+ },
+ {
+ "epoch": 29.455040871934607,
+ "grad_norm": 5.461793422698975,
+ "learning_rate": 1.654848607589255e-05,
+ "loss": 0.4085,
+ "step": 10810
+ },
+ {
+ "epoch": 29.457765667574932,
+ "grad_norm": 4.517482757568359,
+ "learning_rate": 1.6547819096175722e-05,
+ "loss": 0.3353,
+ "step": 10811
+ },
+ {
+ "epoch": 29.460490463215258,
+ "grad_norm": 4.293356418609619,
+ "learning_rate": 1.6547152065464688e-05,
+ "loss": 0.2989,
+ "step": 10812
+ },
+ {
+ "epoch": 29.463215258855588,
+ "grad_norm": 4.285608768463135,
+ "learning_rate": 1.6546484983764646e-05,
+ "loss": 0.2658,
+ "step": 10813
+ },
+ {
+ "epoch": 29.465940054495913,
+ "grad_norm": 4.760890960693359,
+ "learning_rate": 1.6545817851080785e-05,
+ "loss": 0.2917,
+ "step": 10814
+ },
+ {
+ "epoch": 29.46866485013624,
+ "grad_norm": 5.727476119995117,
+ "learning_rate": 1.654515066741831e-05,
+ "loss": 0.2195,
+ "step": 10815
+ },
+ {
+ "epoch": 29.47138964577657,
+ "grad_norm": 4.85028600692749,
+ "learning_rate": 1.654448343278241e-05,
+ "loss": 0.316,
+ "step": 10816
+ },
+ {
+ "epoch": 29.474114441416894,
+ "grad_norm": 4.234750270843506,
+ "learning_rate": 1.6543816147178283e-05,
+ "loss": 0.2168,
+ "step": 10817
+ },
+ {
+ "epoch": 29.47683923705722,
+ "grad_norm": 6.2509613037109375,
+ "learning_rate": 1.654314881061113e-05,
+ "loss": 0.3495,
+ "step": 10818
+ },
+ {
+ "epoch": 29.479564032697546,
+ "grad_norm": 3.9587743282318115,
+ "learning_rate": 1.6542481423086145e-05,
+ "loss": 0.2639,
+ "step": 10819
+ },
+ {
+ "epoch": 29.482288828337875,
+ "grad_norm": 4.945070266723633,
+ "learning_rate": 1.654181398460852e-05,
+ "loss": 0.3029,
+ "step": 10820
+ },
+ {
+ "epoch": 29.4850136239782,
+ "grad_norm": 4.671682357788086,
+ "learning_rate": 1.6541146495183463e-05,
+ "loss": 0.3047,
+ "step": 10821
+ },
+ {
+ "epoch": 29.48773841961853,
+ "grad_norm": 6.540012359619141,
+ "learning_rate": 1.6540478954816164e-05,
+ "loss": 0.2895,
+ "step": 10822
+ },
+ {
+ "epoch": 29.490463215258856,
+ "grad_norm": 4.346771240234375,
+ "learning_rate": 1.6539811363511827e-05,
+ "loss": 0.1969,
+ "step": 10823
+ },
+ {
+ "epoch": 29.493188010899182,
+ "grad_norm": 4.781223773956299,
+ "learning_rate": 1.653914372127565e-05,
+ "loss": 0.2468,
+ "step": 10824
+ },
+ {
+ "epoch": 29.495912806539508,
+ "grad_norm": 4.751381874084473,
+ "learning_rate": 1.6538476028112832e-05,
+ "loss": 0.1876,
+ "step": 10825
+ },
+ {
+ "epoch": 29.498637602179837,
+ "grad_norm": 4.843021869659424,
+ "learning_rate": 1.653780828402857e-05,
+ "loss": 0.2167,
+ "step": 10826
+ },
+ {
+ "epoch": 29.501362397820163,
+ "grad_norm": 4.4665727615356445,
+ "learning_rate": 1.6537140489028073e-05,
+ "loss": 0.3499,
+ "step": 10827
+ },
+ {
+ "epoch": 29.504087193460492,
+ "grad_norm": 4.956500053405762,
+ "learning_rate": 1.653647264311653e-05,
+ "loss": 0.3194,
+ "step": 10828
+ },
+ {
+ "epoch": 29.506811989100818,
+ "grad_norm": 4.1348419189453125,
+ "learning_rate": 1.6535804746299154e-05,
+ "loss": 0.216,
+ "step": 10829
+ },
+ {
+ "epoch": 29.509536784741144,
+ "grad_norm": 6.316023349761963,
+ "learning_rate": 1.653513679858114e-05,
+ "loss": 0.2608,
+ "step": 10830
+ },
+ {
+ "epoch": 29.51226158038147,
+ "grad_norm": 4.912325382232666,
+ "learning_rate": 1.6534468799967692e-05,
+ "loss": 0.5074,
+ "step": 10831
+ },
+ {
+ "epoch": 29.5149863760218,
+ "grad_norm": 3.8118858337402344,
+ "learning_rate": 1.6533800750464006e-05,
+ "loss": 0.2006,
+ "step": 10832
+ },
+ {
+ "epoch": 29.517711171662125,
+ "grad_norm": 4.429504871368408,
+ "learning_rate": 1.6533132650075293e-05,
+ "loss": 0.2346,
+ "step": 10833
+ },
+ {
+ "epoch": 29.520435967302454,
+ "grad_norm": 4.567898273468018,
+ "learning_rate": 1.6532464498806755e-05,
+ "loss": 0.1652,
+ "step": 10834
+ },
+ {
+ "epoch": 29.52316076294278,
+ "grad_norm": 5.131767749786377,
+ "learning_rate": 1.6531796296663595e-05,
+ "loss": 0.2082,
+ "step": 10835
+ },
+ {
+ "epoch": 29.525885558583106,
+ "grad_norm": 10.329245567321777,
+ "learning_rate": 1.653112804365101e-05,
+ "loss": 0.3271,
+ "step": 10836
+ },
+ {
+ "epoch": 29.52861035422343,
+ "grad_norm": 4.262543201446533,
+ "learning_rate": 1.653045973977421e-05,
+ "loss": 0.2377,
+ "step": 10837
+ },
+ {
+ "epoch": 29.53133514986376,
+ "grad_norm": 4.615146160125732,
+ "learning_rate": 1.6529791385038404e-05,
+ "loss": 0.2096,
+ "step": 10838
+ },
+ {
+ "epoch": 29.534059945504087,
+ "grad_norm": 10.172688484191895,
+ "learning_rate": 1.6529122979448792e-05,
+ "loss": 0.3014,
+ "step": 10839
+ },
+ {
+ "epoch": 29.536784741144416,
+ "grad_norm": 9.128286361694336,
+ "learning_rate": 1.6528454523010583e-05,
+ "loss": 0.3341,
+ "step": 10840
+ },
+ {
+ "epoch": 29.539509536784742,
+ "grad_norm": 6.206465244293213,
+ "learning_rate": 1.6527786015728975e-05,
+ "loss": 0.3353,
+ "step": 10841
+ },
+ {
+ "epoch": 29.542234332425068,
+ "grad_norm": 4.353102207183838,
+ "learning_rate": 1.652711745760918e-05,
+ "loss": 0.2152,
+ "step": 10842
+ },
+ {
+ "epoch": 29.544959128065393,
+ "grad_norm": 4.565274715423584,
+ "learning_rate": 1.6526448848656407e-05,
+ "loss": 0.3101,
+ "step": 10843
+ },
+ {
+ "epoch": 29.547683923705723,
+ "grad_norm": 4.343389511108398,
+ "learning_rate": 1.652578018887586e-05,
+ "loss": 0.2545,
+ "step": 10844
+ },
+ {
+ "epoch": 29.55040871934605,
+ "grad_norm": 5.270857334136963,
+ "learning_rate": 1.6525111478272748e-05,
+ "loss": 0.207,
+ "step": 10845
+ },
+ {
+ "epoch": 29.553133514986374,
+ "grad_norm": 5.012444496154785,
+ "learning_rate": 1.6524442716852273e-05,
+ "loss": 0.3066,
+ "step": 10846
+ },
+ {
+ "epoch": 29.555858310626704,
+ "grad_norm": 5.0461249351501465,
+ "learning_rate": 1.6523773904619653e-05,
+ "loss": 0.2975,
+ "step": 10847
+ },
+ {
+ "epoch": 29.55858310626703,
+ "grad_norm": 4.437694072723389,
+ "learning_rate": 1.652310504158009e-05,
+ "loss": 0.241,
+ "step": 10848
+ },
+ {
+ "epoch": 29.561307901907355,
+ "grad_norm": 4.978779315948486,
+ "learning_rate": 1.6522436127738797e-05,
+ "loss": 0.2234,
+ "step": 10849
+ },
+ {
+ "epoch": 29.564032697547685,
+ "grad_norm": 8.885309219360352,
+ "learning_rate": 1.652176716310098e-05,
+ "loss": 0.3618,
+ "step": 10850
+ },
+ {
+ "epoch": 29.56675749318801,
+ "grad_norm": 4.834117889404297,
+ "learning_rate": 1.6521098147671847e-05,
+ "loss": 0.2395,
+ "step": 10851
+ },
+ {
+ "epoch": 29.569482288828336,
+ "grad_norm": 4.303838729858398,
+ "learning_rate": 1.6520429081456615e-05,
+ "loss": 0.2274,
+ "step": 10852
+ },
+ {
+ "epoch": 29.572207084468666,
+ "grad_norm": 4.8089094161987305,
+ "learning_rate": 1.6519759964460488e-05,
+ "loss": 0.4058,
+ "step": 10853
+ },
+ {
+ "epoch": 29.57493188010899,
+ "grad_norm": 5.078211784362793,
+ "learning_rate": 1.6519090796688685e-05,
+ "loss": 0.2469,
+ "step": 10854
+ },
+ {
+ "epoch": 29.577656675749317,
+ "grad_norm": 5.3373703956604,
+ "learning_rate": 1.651842157814641e-05,
+ "loss": 0.189,
+ "step": 10855
+ },
+ {
+ "epoch": 29.580381471389646,
+ "grad_norm": 4.547861576080322,
+ "learning_rate": 1.6517752308838877e-05,
+ "loss": 0.2478,
+ "step": 10856
+ },
+ {
+ "epoch": 29.583106267029972,
+ "grad_norm": 5.150286674499512,
+ "learning_rate": 1.65170829887713e-05,
+ "loss": 0.2939,
+ "step": 10857
+ },
+ {
+ "epoch": 29.585831062670298,
+ "grad_norm": 4.09398078918457,
+ "learning_rate": 1.651641361794889e-05,
+ "loss": 0.4836,
+ "step": 10858
+ },
+ {
+ "epoch": 29.588555858310627,
+ "grad_norm": 5.406062602996826,
+ "learning_rate": 1.6515744196376863e-05,
+ "loss": 0.2708,
+ "step": 10859
+ },
+ {
+ "epoch": 29.591280653950953,
+ "grad_norm": 4.650078296661377,
+ "learning_rate": 1.6515074724060425e-05,
+ "loss": 0.3926,
+ "step": 10860
+ },
+ {
+ "epoch": 29.59400544959128,
+ "grad_norm": 6.712960243225098,
+ "learning_rate": 1.6514405201004797e-05,
+ "loss": 0.2678,
+ "step": 10861
+ },
+ {
+ "epoch": 29.59673024523161,
+ "grad_norm": 4.830481052398682,
+ "learning_rate": 1.6513735627215192e-05,
+ "loss": 0.2303,
+ "step": 10862
+ },
+ {
+ "epoch": 29.599455040871934,
+ "grad_norm": 4.260829448699951,
+ "learning_rate": 1.651306600269682e-05,
+ "loss": 0.2648,
+ "step": 10863
+ },
+ {
+ "epoch": 29.60217983651226,
+ "grad_norm": 4.795943260192871,
+ "learning_rate": 1.65123963274549e-05,
+ "loss": 0.2664,
+ "step": 10864
+ },
+ {
+ "epoch": 29.60490463215259,
+ "grad_norm": 4.832176208496094,
+ "learning_rate": 1.6511726601494655e-05,
+ "loss": 0.4001,
+ "step": 10865
+ },
+ {
+ "epoch": 29.607629427792915,
+ "grad_norm": 4.2651262283325195,
+ "learning_rate": 1.6511056824821285e-05,
+ "loss": 0.2118,
+ "step": 10866
+ },
+ {
+ "epoch": 29.61035422343324,
+ "grad_norm": 3.875981092453003,
+ "learning_rate": 1.6510386997440017e-05,
+ "loss": 0.51,
+ "step": 10867
+ },
+ {
+ "epoch": 29.61307901907357,
+ "grad_norm": 5.185285568237305,
+ "learning_rate": 1.6509717119356065e-05,
+ "loss": 0.3201,
+ "step": 10868
+ },
+ {
+ "epoch": 29.615803814713896,
+ "grad_norm": 4.620301246643066,
+ "learning_rate": 1.6509047190574643e-05,
+ "loss": 0.2995,
+ "step": 10869
+ },
+ {
+ "epoch": 29.618528610354222,
+ "grad_norm": 4.5280938148498535,
+ "learning_rate": 1.650837721110097e-05,
+ "loss": 0.3524,
+ "step": 10870
+ },
+ {
+ "epoch": 29.62125340599455,
+ "grad_norm": 4.368135452270508,
+ "learning_rate": 1.6507707180940268e-05,
+ "loss": 0.3073,
+ "step": 10871
+ },
+ {
+ "epoch": 29.623978201634877,
+ "grad_norm": 5.049985885620117,
+ "learning_rate": 1.650703710009775e-05,
+ "loss": 0.208,
+ "step": 10872
+ },
+ {
+ "epoch": 29.626702997275203,
+ "grad_norm": 4.46707010269165,
+ "learning_rate": 1.6506366968578637e-05,
+ "loss": 0.3071,
+ "step": 10873
+ },
+ {
+ "epoch": 29.629427792915532,
+ "grad_norm": 4.344300746917725,
+ "learning_rate": 1.650569678638815e-05,
+ "loss": 0.4377,
+ "step": 10874
+ },
+ {
+ "epoch": 29.632152588555858,
+ "grad_norm": 4.491318702697754,
+ "learning_rate": 1.65050265535315e-05,
+ "loss": 0.3672,
+ "step": 10875
+ },
+ {
+ "epoch": 29.634877384196184,
+ "grad_norm": 4.696650505065918,
+ "learning_rate": 1.650435627001392e-05,
+ "loss": 0.2516,
+ "step": 10876
+ },
+ {
+ "epoch": 29.637602179836513,
+ "grad_norm": 4.889502048492432,
+ "learning_rate": 1.650368593584062e-05,
+ "loss": 0.223,
+ "step": 10877
+ },
+ {
+ "epoch": 29.64032697547684,
+ "grad_norm": 4.592113971710205,
+ "learning_rate": 1.650301555101682e-05,
+ "loss": 0.2641,
+ "step": 10878
+ },
+ {
+ "epoch": 29.643051771117165,
+ "grad_norm": 4.808929443359375,
+ "learning_rate": 1.6502345115547745e-05,
+ "loss": 0.214,
+ "step": 10879
+ },
+ {
+ "epoch": 29.645776566757494,
+ "grad_norm": 4.147544860839844,
+ "learning_rate": 1.6501674629438617e-05,
+ "loss": 0.2735,
+ "step": 10880
+ },
+ {
+ "epoch": 29.64850136239782,
+ "grad_norm": 4.4309563636779785,
+ "learning_rate": 1.6501004092694655e-05,
+ "loss": 0.2259,
+ "step": 10881
+ },
+ {
+ "epoch": 29.651226158038146,
+ "grad_norm": 5.618682384490967,
+ "learning_rate": 1.6500333505321085e-05,
+ "loss": 0.2525,
+ "step": 10882
+ },
+ {
+ "epoch": 29.653950953678475,
+ "grad_norm": 3.787888288497925,
+ "learning_rate": 1.6499662867323127e-05,
+ "loss": 0.2073,
+ "step": 10883
+ },
+ {
+ "epoch": 29.6566757493188,
+ "grad_norm": 4.621107578277588,
+ "learning_rate": 1.6498992178706002e-05,
+ "loss": 0.2034,
+ "step": 10884
+ },
+ {
+ "epoch": 29.659400544959126,
+ "grad_norm": 4.824036121368408,
+ "learning_rate": 1.6498321439474935e-05,
+ "loss": 0.3792,
+ "step": 10885
+ },
+ {
+ "epoch": 29.662125340599456,
+ "grad_norm": 5.101800918579102,
+ "learning_rate": 1.649765064963515e-05,
+ "loss": 0.2964,
+ "step": 10886
+ },
+ {
+ "epoch": 29.66485013623978,
+ "grad_norm": 5.154964923858643,
+ "learning_rate": 1.649697980919187e-05,
+ "loss": 0.4454,
+ "step": 10887
+ },
+ {
+ "epoch": 29.667574931880107,
+ "grad_norm": 4.345524311065674,
+ "learning_rate": 1.6496308918150323e-05,
+ "loss": 0.1986,
+ "step": 10888
+ },
+ {
+ "epoch": 29.670299727520437,
+ "grad_norm": 5.0136799812316895,
+ "learning_rate": 1.649563797651573e-05,
+ "loss": 0.1993,
+ "step": 10889
+ },
+ {
+ "epoch": 29.673024523160763,
+ "grad_norm": 4.827890396118164,
+ "learning_rate": 1.6494966984293316e-05,
+ "loss": 0.2124,
+ "step": 10890
+ },
+ {
+ "epoch": 29.67574931880109,
+ "grad_norm": 4.360700607299805,
+ "learning_rate": 1.6494295941488307e-05,
+ "loss": 0.2128,
+ "step": 10891
+ },
+ {
+ "epoch": 29.678474114441418,
+ "grad_norm": 5.309598922729492,
+ "learning_rate": 1.6493624848105934e-05,
+ "loss": 0.3205,
+ "step": 10892
+ },
+ {
+ "epoch": 29.681198910081743,
+ "grad_norm": 5.3124098777771,
+ "learning_rate": 1.649295370415142e-05,
+ "loss": 0.287,
+ "step": 10893
+ },
+ {
+ "epoch": 29.68392370572207,
+ "grad_norm": 4.666045188903809,
+ "learning_rate": 1.649228250962999e-05,
+ "loss": 0.2488,
+ "step": 10894
+ },
+ {
+ "epoch": 29.6866485013624,
+ "grad_norm": 5.44038724899292,
+ "learning_rate": 1.6491611264546868e-05,
+ "loss": 0.3532,
+ "step": 10895
+ },
+ {
+ "epoch": 29.689373297002724,
+ "grad_norm": 5.58378791809082,
+ "learning_rate": 1.6490939968907295e-05,
+ "loss": 0.2774,
+ "step": 10896
+ },
+ {
+ "epoch": 29.69209809264305,
+ "grad_norm": 5.122231483459473,
+ "learning_rate": 1.6490268622716484e-05,
+ "loss": 0.3623,
+ "step": 10897
+ },
+ {
+ "epoch": 29.69482288828338,
+ "grad_norm": 4.856808662414551,
+ "learning_rate": 1.6489597225979673e-05,
+ "loss": 0.1459,
+ "step": 10898
+ },
+ {
+ "epoch": 29.697547683923705,
+ "grad_norm": 4.066020965576172,
+ "learning_rate": 1.6488925778702084e-05,
+ "loss": 0.2045,
+ "step": 10899
+ },
+ {
+ "epoch": 29.70027247956403,
+ "grad_norm": 4.915604114532471,
+ "learning_rate": 1.6488254280888956e-05,
+ "loss": 0.4653,
+ "step": 10900
+ },
+ {
+ "epoch": 29.70299727520436,
+ "grad_norm": 4.934942245483398,
+ "learning_rate": 1.6487582732545506e-05,
+ "loss": 0.4388,
+ "step": 10901
+ },
+ {
+ "epoch": 29.705722070844686,
+ "grad_norm": 4.696750164031982,
+ "learning_rate": 1.6486911133676974e-05,
+ "loss": 0.307,
+ "step": 10902
+ },
+ {
+ "epoch": 29.708446866485012,
+ "grad_norm": 5.118165969848633,
+ "learning_rate": 1.6486239484288585e-05,
+ "loss": 0.378,
+ "step": 10903
+ },
+ {
+ "epoch": 29.71117166212534,
+ "grad_norm": 4.784078121185303,
+ "learning_rate": 1.6485567784385574e-05,
+ "loss": 0.2379,
+ "step": 10904
+ },
+ {
+ "epoch": 29.713896457765667,
+ "grad_norm": 4.532581806182861,
+ "learning_rate": 1.6484896033973168e-05,
+ "loss": 0.3362,
+ "step": 10905
+ },
+ {
+ "epoch": 29.716621253405993,
+ "grad_norm": 5.064365863800049,
+ "learning_rate": 1.6484224233056602e-05,
+ "loss": 0.2409,
+ "step": 10906
+ },
+ {
+ "epoch": 29.719346049046322,
+ "grad_norm": 4.933046817779541,
+ "learning_rate": 1.6483552381641107e-05,
+ "loss": 0.4019,
+ "step": 10907
+ },
+ {
+ "epoch": 29.722070844686648,
+ "grad_norm": 4.909958362579346,
+ "learning_rate": 1.6482880479731912e-05,
+ "loss": 0.3451,
+ "step": 10908
+ },
+ {
+ "epoch": 29.724795640326974,
+ "grad_norm": 4.545212268829346,
+ "learning_rate": 1.6482208527334257e-05,
+ "loss": 0.4113,
+ "step": 10909
+ },
+ {
+ "epoch": 29.727520435967303,
+ "grad_norm": 4.793064117431641,
+ "learning_rate": 1.6481536524453365e-05,
+ "loss": 0.2095,
+ "step": 10910
+ },
+ {
+ "epoch": 29.73024523160763,
+ "grad_norm": 4.07961368560791,
+ "learning_rate": 1.648086447109448e-05,
+ "loss": 0.1686,
+ "step": 10911
+ },
+ {
+ "epoch": 29.732970027247955,
+ "grad_norm": 4.605406284332275,
+ "learning_rate": 1.6480192367262827e-05,
+ "loss": 0.2839,
+ "step": 10912
+ },
+ {
+ "epoch": 29.735694822888284,
+ "grad_norm": 4.269509792327881,
+ "learning_rate": 1.6479520212963644e-05,
+ "loss": 0.2049,
+ "step": 10913
+ },
+ {
+ "epoch": 29.73841961852861,
+ "grad_norm": 4.191793918609619,
+ "learning_rate": 1.647884800820217e-05,
+ "loss": 0.2047,
+ "step": 10914
+ },
+ {
+ "epoch": 29.741144414168936,
+ "grad_norm": 5.726789951324463,
+ "learning_rate": 1.6478175752983632e-05,
+ "loss": 0.271,
+ "step": 10915
+ },
+ {
+ "epoch": 29.743869209809265,
+ "grad_norm": 4.235219478607178,
+ "learning_rate": 1.6477503447313276e-05,
+ "loss": 0.2215,
+ "step": 10916
+ },
+ {
+ "epoch": 29.74659400544959,
+ "grad_norm": 4.840961456298828,
+ "learning_rate": 1.6476831091196327e-05,
+ "loss": 0.2323,
+ "step": 10917
+ },
+ {
+ "epoch": 29.749318801089917,
+ "grad_norm": 4.3625311851501465,
+ "learning_rate": 1.6476158684638027e-05,
+ "loss": 0.4826,
+ "step": 10918
+ },
+ {
+ "epoch": 29.752043596730246,
+ "grad_norm": 4.573596954345703,
+ "learning_rate": 1.647548622764361e-05,
+ "loss": 0.248,
+ "step": 10919
+ },
+ {
+ "epoch": 29.754768392370572,
+ "grad_norm": 4.976794719696045,
+ "learning_rate": 1.6474813720218318e-05,
+ "loss": 0.2856,
+ "step": 10920
+ },
+ {
+ "epoch": 29.757493188010898,
+ "grad_norm": 4.313014984130859,
+ "learning_rate": 1.6474141162367384e-05,
+ "loss": 0.2551,
+ "step": 10921
+ },
+ {
+ "epoch": 29.760217983651227,
+ "grad_norm": 4.3383259773254395,
+ "learning_rate": 1.6473468554096045e-05,
+ "loss": 0.4107,
+ "step": 10922
+ },
+ {
+ "epoch": 29.762942779291553,
+ "grad_norm": 6.740191459655762,
+ "learning_rate": 1.6472795895409542e-05,
+ "loss": 0.1985,
+ "step": 10923
+ },
+ {
+ "epoch": 29.76566757493188,
+ "grad_norm": 3.902717113494873,
+ "learning_rate": 1.6472123186313114e-05,
+ "loss": 0.3195,
+ "step": 10924
+ },
+ {
+ "epoch": 29.768392370572208,
+ "grad_norm": 4.902856349945068,
+ "learning_rate": 1.6471450426811997e-05,
+ "loss": 0.2958,
+ "step": 10925
+ },
+ {
+ "epoch": 29.771117166212534,
+ "grad_norm": 6.227838039398193,
+ "learning_rate": 1.6470777616911435e-05,
+ "loss": 0.2136,
+ "step": 10926
+ },
+ {
+ "epoch": 29.77384196185286,
+ "grad_norm": 5.234126567840576,
+ "learning_rate": 1.6470104756616664e-05,
+ "loss": 0.291,
+ "step": 10927
+ },
+ {
+ "epoch": 29.77656675749319,
+ "grad_norm": 4.357918739318848,
+ "learning_rate": 1.6469431845932923e-05,
+ "loss": 0.3565,
+ "step": 10928
+ },
+ {
+ "epoch": 29.779291553133515,
+ "grad_norm": 4.866206169128418,
+ "learning_rate": 1.646875888486546e-05,
+ "loss": 0.2296,
+ "step": 10929
+ },
+ {
+ "epoch": 29.78201634877384,
+ "grad_norm": 5.232158184051514,
+ "learning_rate": 1.6468085873419502e-05,
+ "loss": 0.548,
+ "step": 10930
+ },
+ {
+ "epoch": 29.78474114441417,
+ "grad_norm": 5.201484680175781,
+ "learning_rate": 1.646741281160031e-05,
+ "loss": 0.1939,
+ "step": 10931
+ },
+ {
+ "epoch": 29.787465940054496,
+ "grad_norm": 3.8252975940704346,
+ "learning_rate": 1.646673969941311e-05,
+ "loss": 0.1698,
+ "step": 10932
+ },
+ {
+ "epoch": 29.79019073569482,
+ "grad_norm": 4.307361125946045,
+ "learning_rate": 1.646606653686315e-05,
+ "loss": 0.3103,
+ "step": 10933
+ },
+ {
+ "epoch": 29.79291553133515,
+ "grad_norm": 3.9691457748413086,
+ "learning_rate": 1.6465393323955673e-05,
+ "loss": 0.2468,
+ "step": 10934
+ },
+ {
+ "epoch": 29.795640326975477,
+ "grad_norm": 3.7958550453186035,
+ "learning_rate": 1.6464720060695918e-05,
+ "loss": 0.2316,
+ "step": 10935
+ },
+ {
+ "epoch": 29.798365122615802,
+ "grad_norm": 4.869331359863281,
+ "learning_rate": 1.6464046747089134e-05,
+ "loss": 0.3005,
+ "step": 10936
+ },
+ {
+ "epoch": 29.80108991825613,
+ "grad_norm": 6.487374305725098,
+ "learning_rate": 1.6463373383140563e-05,
+ "loss": 0.2779,
+ "step": 10937
+ },
+ {
+ "epoch": 29.803814713896458,
+ "grad_norm": 5.162075042724609,
+ "learning_rate": 1.6462699968855446e-05,
+ "loss": 0.3152,
+ "step": 10938
+ },
+ {
+ "epoch": 29.806539509536783,
+ "grad_norm": 4.885021686553955,
+ "learning_rate": 1.6462026504239026e-05,
+ "loss": 0.3055,
+ "step": 10939
+ },
+ {
+ "epoch": 29.809264305177113,
+ "grad_norm": 4.717816352844238,
+ "learning_rate": 1.6461352989296556e-05,
+ "loss": 0.28,
+ "step": 10940
+ },
+ {
+ "epoch": 29.81198910081744,
+ "grad_norm": 4.196158409118652,
+ "learning_rate": 1.6460679424033278e-05,
+ "loss": 0.1885,
+ "step": 10941
+ },
+ {
+ "epoch": 29.814713896457764,
+ "grad_norm": 6.0366034507751465,
+ "learning_rate": 1.6460005808454434e-05,
+ "loss": 0.2653,
+ "step": 10942
+ },
+ {
+ "epoch": 29.817438692098094,
+ "grad_norm": 4.4392876625061035,
+ "learning_rate": 1.6459332142565275e-05,
+ "loss": 0.2404,
+ "step": 10943
+ },
+ {
+ "epoch": 29.82016348773842,
+ "grad_norm": 5.144449710845947,
+ "learning_rate": 1.6458658426371043e-05,
+ "loss": 0.1861,
+ "step": 10944
+ },
+ {
+ "epoch": 29.822888283378745,
+ "grad_norm": 4.61625862121582,
+ "learning_rate": 1.645798465987699e-05,
+ "loss": 0.4422,
+ "step": 10945
+ },
+ {
+ "epoch": 29.825613079019075,
+ "grad_norm": 4.891772747039795,
+ "learning_rate": 1.6457310843088355e-05,
+ "loss": 0.1965,
+ "step": 10946
+ },
+ {
+ "epoch": 29.8283378746594,
+ "grad_norm": 4.031517028808594,
+ "learning_rate": 1.6456636976010396e-05,
+ "loss": 0.1787,
+ "step": 10947
+ },
+ {
+ "epoch": 29.831062670299726,
+ "grad_norm": 5.364249229431152,
+ "learning_rate": 1.6455963058648353e-05,
+ "loss": 0.417,
+ "step": 10948
+ },
+ {
+ "epoch": 29.833787465940055,
+ "grad_norm": 4.958280086517334,
+ "learning_rate": 1.6455289091007478e-05,
+ "loss": 0.1873,
+ "step": 10949
+ },
+ {
+ "epoch": 29.83651226158038,
+ "grad_norm": 4.575549602508545,
+ "learning_rate": 1.645461507309302e-05,
+ "loss": 0.2489,
+ "step": 10950
+ },
+ {
+ "epoch": 29.839237057220707,
+ "grad_norm": 5.588534832000732,
+ "learning_rate": 1.6453941004910227e-05,
+ "loss": 0.2332,
+ "step": 10951
+ },
+ {
+ "epoch": 29.841961852861036,
+ "grad_norm": 6.5234456062316895,
+ "learning_rate": 1.645326688646435e-05,
+ "loss": 0.4153,
+ "step": 10952
+ },
+ {
+ "epoch": 29.844686648501362,
+ "grad_norm": 5.854434490203857,
+ "learning_rate": 1.645259271776064e-05,
+ "loss": 0.3094,
+ "step": 10953
+ },
+ {
+ "epoch": 29.847411444141688,
+ "grad_norm": 6.151221752166748,
+ "learning_rate": 1.6451918498804343e-05,
+ "loss": 0.4009,
+ "step": 10954
+ },
+ {
+ "epoch": 29.850136239782017,
+ "grad_norm": 4.691900253295898,
+ "learning_rate": 1.645124422960071e-05,
+ "loss": 0.2706,
+ "step": 10955
+ },
+ {
+ "epoch": 29.852861035422343,
+ "grad_norm": 4.210886001586914,
+ "learning_rate": 1.6450569910154998e-05,
+ "loss": 0.2441,
+ "step": 10956
+ },
+ {
+ "epoch": 29.85558583106267,
+ "grad_norm": 5.123546600341797,
+ "learning_rate": 1.6449895540472455e-05,
+ "loss": 0.3376,
+ "step": 10957
+ },
+ {
+ "epoch": 29.858310626703,
+ "grad_norm": 5.5067458152771,
+ "learning_rate": 1.6449221120558334e-05,
+ "loss": 0.5457,
+ "step": 10958
+ },
+ {
+ "epoch": 29.861035422343324,
+ "grad_norm": 4.140815734863281,
+ "learning_rate": 1.6448546650417883e-05,
+ "loss": 0.1817,
+ "step": 10959
+ },
+ {
+ "epoch": 29.86376021798365,
+ "grad_norm": 4.559505462646484,
+ "learning_rate": 1.644787213005636e-05,
+ "loss": 0.4387,
+ "step": 10960
+ },
+ {
+ "epoch": 29.86648501362398,
+ "grad_norm": 4.8837199211120605,
+ "learning_rate": 1.6447197559479017e-05,
+ "loss": 0.3649,
+ "step": 10961
+ },
+ {
+ "epoch": 29.869209809264305,
+ "grad_norm": 5.871948719024658,
+ "learning_rate": 1.6446522938691107e-05,
+ "loss": 0.298,
+ "step": 10962
+ },
+ {
+ "epoch": 29.87193460490463,
+ "grad_norm": 4.6947832107543945,
+ "learning_rate": 1.6445848267697884e-05,
+ "loss": 0.4626,
+ "step": 10963
+ },
+ {
+ "epoch": 29.87465940054496,
+ "grad_norm": 4.719560146331787,
+ "learning_rate": 1.6445173546504603e-05,
+ "loss": 0.4795,
+ "step": 10964
+ },
+ {
+ "epoch": 29.877384196185286,
+ "grad_norm": 4.491894721984863,
+ "learning_rate": 1.6444498775116512e-05,
+ "loss": 0.2895,
+ "step": 10965
+ },
+ {
+ "epoch": 29.88010899182561,
+ "grad_norm": 4.1889472007751465,
+ "learning_rate": 1.6443823953538876e-05,
+ "loss": 0.2604,
+ "step": 10966
+ },
+ {
+ "epoch": 29.88283378746594,
+ "grad_norm": 5.344935417175293,
+ "learning_rate": 1.644314908177695e-05,
+ "loss": 0.3338,
+ "step": 10967
+ },
+ {
+ "epoch": 29.885558583106267,
+ "grad_norm": 4.173104763031006,
+ "learning_rate": 1.6442474159835984e-05,
+ "loss": 0.3718,
+ "step": 10968
+ },
+ {
+ "epoch": 29.888283378746593,
+ "grad_norm": 7.1409125328063965,
+ "learning_rate": 1.6441799187721237e-05,
+ "loss": 0.3983,
+ "step": 10969
+ },
+ {
+ "epoch": 29.891008174386922,
+ "grad_norm": 4.850971221923828,
+ "learning_rate": 1.6441124165437962e-05,
+ "loss": 0.2128,
+ "step": 10970
+ },
+ {
+ "epoch": 29.893732970027248,
+ "grad_norm": 4.199065685272217,
+ "learning_rate": 1.6440449092991424e-05,
+ "loss": 0.2654,
+ "step": 10971
+ },
+ {
+ "epoch": 29.896457765667574,
+ "grad_norm": 5.507880687713623,
+ "learning_rate": 1.643977397038687e-05,
+ "loss": 0.4085,
+ "step": 10972
+ },
+ {
+ "epoch": 29.899182561307903,
+ "grad_norm": 4.76394510269165,
+ "learning_rate": 1.6439098797629565e-05,
+ "loss": 0.2625,
+ "step": 10973
+ },
+ {
+ "epoch": 29.90190735694823,
+ "grad_norm": 4.549790859222412,
+ "learning_rate": 1.6438423574724768e-05,
+ "loss": 0.3164,
+ "step": 10974
+ },
+ {
+ "epoch": 29.904632152588555,
+ "grad_norm": 4.0453596115112305,
+ "learning_rate": 1.6437748301677738e-05,
+ "loss": 0.3408,
+ "step": 10975
+ },
+ {
+ "epoch": 29.907356948228884,
+ "grad_norm": 5.008920669555664,
+ "learning_rate": 1.643707297849373e-05,
+ "loss": 0.3096,
+ "step": 10976
+ },
+ {
+ "epoch": 29.91008174386921,
+ "grad_norm": 4.2852654457092285,
+ "learning_rate": 1.6436397605178e-05,
+ "loss": 0.2948,
+ "step": 10977
+ },
+ {
+ "epoch": 29.912806539509535,
+ "grad_norm": 4.597377300262451,
+ "learning_rate": 1.6435722181735817e-05,
+ "loss": 0.3118,
+ "step": 10978
+ },
+ {
+ "epoch": 29.915531335149865,
+ "grad_norm": 4.921483993530273,
+ "learning_rate": 1.6435046708172436e-05,
+ "loss": 0.2455,
+ "step": 10979
+ },
+ {
+ "epoch": 29.91825613079019,
+ "grad_norm": 4.161590576171875,
+ "learning_rate": 1.6434371184493117e-05,
+ "loss": 0.2235,
+ "step": 10980
+ },
+ {
+ "epoch": 29.920980926430516,
+ "grad_norm": 4.8175835609436035,
+ "learning_rate": 1.6433695610703123e-05,
+ "loss": 0.2417,
+ "step": 10981
+ },
+ {
+ "epoch": 29.923705722070846,
+ "grad_norm": 12.910106658935547,
+ "learning_rate": 1.6433019986807714e-05,
+ "loss": 0.3274,
+ "step": 10982
+ },
+ {
+ "epoch": 29.92643051771117,
+ "grad_norm": 5.099645137786865,
+ "learning_rate": 1.6432344312812153e-05,
+ "loss": 0.2939,
+ "step": 10983
+ },
+ {
+ "epoch": 29.929155313351497,
+ "grad_norm": 4.716278076171875,
+ "learning_rate": 1.64316685887217e-05,
+ "loss": 0.1621,
+ "step": 10984
+ },
+ {
+ "epoch": 29.931880108991827,
+ "grad_norm": 5.618128776550293,
+ "learning_rate": 1.6430992814541625e-05,
+ "loss": 0.4978,
+ "step": 10985
+ },
+ {
+ "epoch": 29.934604904632153,
+ "grad_norm": 7.898351669311523,
+ "learning_rate": 1.643031699027718e-05,
+ "loss": 0.3338,
+ "step": 10986
+ },
+ {
+ "epoch": 29.93732970027248,
+ "grad_norm": 4.548819541931152,
+ "learning_rate": 1.6429641115933635e-05,
+ "loss": 0.3203,
+ "step": 10987
+ },
+ {
+ "epoch": 29.940054495912808,
+ "grad_norm": 4.439125061035156,
+ "learning_rate": 1.642896519151625e-05,
+ "loss": 0.1708,
+ "step": 10988
+ },
+ {
+ "epoch": 29.942779291553133,
+ "grad_norm": 5.09724760055542,
+ "learning_rate": 1.642828921703029e-05,
+ "loss": 0.2367,
+ "step": 10989
+ },
+ {
+ "epoch": 29.94550408719346,
+ "grad_norm": 4.500386714935303,
+ "learning_rate": 1.6427613192481025e-05,
+ "loss": 0.2348,
+ "step": 10990
+ },
+ {
+ "epoch": 29.94822888283379,
+ "grad_norm": 5.305752754211426,
+ "learning_rate": 1.6426937117873713e-05,
+ "loss": 0.2196,
+ "step": 10991
+ },
+ {
+ "epoch": 29.950953678474114,
+ "grad_norm": 5.08002233505249,
+ "learning_rate": 1.642626099321362e-05,
+ "loss": 0.1789,
+ "step": 10992
+ },
+ {
+ "epoch": 29.95367847411444,
+ "grad_norm": 5.507477283477783,
+ "learning_rate": 1.6425584818506016e-05,
+ "loss": 0.2201,
+ "step": 10993
+ },
+ {
+ "epoch": 29.95640326975477,
+ "grad_norm": 4.864372253417969,
+ "learning_rate": 1.6424908593756162e-05,
+ "loss": 0.2166,
+ "step": 10994
+ },
+ {
+ "epoch": 29.959128065395095,
+ "grad_norm": 4.9740400314331055,
+ "learning_rate": 1.6424232318969327e-05,
+ "loss": 0.243,
+ "step": 10995
+ },
+ {
+ "epoch": 29.96185286103542,
+ "grad_norm": 4.4197845458984375,
+ "learning_rate": 1.6423555994150776e-05,
+ "loss": 0.3486,
+ "step": 10996
+ },
+ {
+ "epoch": 29.96457765667575,
+ "grad_norm": 4.772177696228027,
+ "learning_rate": 1.6422879619305782e-05,
+ "loss": 0.3012,
+ "step": 10997
+ },
+ {
+ "epoch": 29.967302452316076,
+ "grad_norm": 5.650397300720215,
+ "learning_rate": 1.6422203194439606e-05,
+ "loss": 0.2028,
+ "step": 10998
+ },
+ {
+ "epoch": 29.970027247956402,
+ "grad_norm": 4.762296199798584,
+ "learning_rate": 1.6421526719557514e-05,
+ "loss": 0.3828,
+ "step": 10999
+ },
+ {
+ "epoch": 29.97275204359673,
+ "grad_norm": 5.297033786773682,
+ "learning_rate": 1.6420850194664787e-05,
+ "loss": 0.2325,
+ "step": 11000
+ },
+ {
+ "epoch": 29.975476839237057,
+ "grad_norm": 4.966675281524658,
+ "learning_rate": 1.642017361976668e-05,
+ "loss": 0.3055,
+ "step": 11001
+ },
+ {
+ "epoch": 29.978201634877383,
+ "grad_norm": 5.283063888549805,
+ "learning_rate": 1.6419496994868467e-05,
+ "loss": 0.238,
+ "step": 11002
+ },
+ {
+ "epoch": 29.980926430517712,
+ "grad_norm": 5.294807434082031,
+ "learning_rate": 1.6418820319975418e-05,
+ "loss": 0.3213,
+ "step": 11003
+ },
+ {
+ "epoch": 29.983651226158038,
+ "grad_norm": 3.9922287464141846,
+ "learning_rate": 1.6418143595092805e-05,
+ "loss": 0.1878,
+ "step": 11004
+ },
+ {
+ "epoch": 29.986376021798364,
+ "grad_norm": 5.325415134429932,
+ "learning_rate": 1.6417466820225896e-05,
+ "loss": 0.3322,
+ "step": 11005
+ },
+ {
+ "epoch": 29.989100817438693,
+ "grad_norm": 5.334811210632324,
+ "learning_rate": 1.641678999537996e-05,
+ "loss": 0.2488,
+ "step": 11006
+ },
+ {
+ "epoch": 29.99182561307902,
+ "grad_norm": 4.67766809463501,
+ "learning_rate": 1.6416113120560275e-05,
+ "loss": 0.2149,
+ "step": 11007
+ },
+ {
+ "epoch": 29.994550408719345,
+ "grad_norm": 4.912754535675049,
+ "learning_rate": 1.64154361957721e-05,
+ "loss": 0.2256,
+ "step": 11008
+ },
+ {
+ "epoch": 29.997275204359674,
+ "grad_norm": 4.87392520904541,
+ "learning_rate": 1.641475922102072e-05,
+ "loss": 0.3028,
+ "step": 11009
+ },
+ {
+ "epoch": 30.0,
+ "grad_norm": 4.759665489196777,
+ "learning_rate": 1.6414082196311402e-05,
+ "loss": 0.332,
+ "step": 11010
+ },
+ {
+ "epoch": 30.002724795640326,
+ "grad_norm": 5.001436233520508,
+ "learning_rate": 1.6413405121649416e-05,
+ "loss": 0.1913,
+ "step": 11011
+ },
+ {
+ "epoch": 30.005449591280655,
+ "grad_norm": 4.336000442504883,
+ "learning_rate": 1.6412727997040036e-05,
+ "loss": 0.1958,
+ "step": 11012
+ },
+ {
+ "epoch": 30.00817438692098,
+ "grad_norm": 4.662959098815918,
+ "learning_rate": 1.641205082248854e-05,
+ "loss": 0.3465,
+ "step": 11013
+ },
+ {
+ "epoch": 30.010899182561307,
+ "grad_norm": 6.1139960289001465,
+ "learning_rate": 1.64113735980002e-05,
+ "loss": 0.2678,
+ "step": 11014
+ },
+ {
+ "epoch": 30.013623978201636,
+ "grad_norm": 4.931779861450195,
+ "learning_rate": 1.641069632358028e-05,
+ "loss": 0.2933,
+ "step": 11015
+ },
+ {
+ "epoch": 30.016348773841962,
+ "grad_norm": 5.162302494049072,
+ "learning_rate": 1.641001899923407e-05,
+ "loss": 0.1544,
+ "step": 11016
+ },
+ {
+ "epoch": 30.019073569482288,
+ "grad_norm": 3.903607130050659,
+ "learning_rate": 1.640934162496684e-05,
+ "loss": 0.2163,
+ "step": 11017
+ },
+ {
+ "epoch": 30.021798365122617,
+ "grad_norm": 3.9897124767303467,
+ "learning_rate": 1.640866420078386e-05,
+ "loss": 0.2524,
+ "step": 11018
+ },
+ {
+ "epoch": 30.024523160762943,
+ "grad_norm": 4.3666534423828125,
+ "learning_rate": 1.640798672669041e-05,
+ "loss": 0.1591,
+ "step": 11019
+ },
+ {
+ "epoch": 30.02724795640327,
+ "grad_norm": 4.784689903259277,
+ "learning_rate": 1.6407309202691765e-05,
+ "loss": 0.3047,
+ "step": 11020
+ },
+ {
+ "epoch": 30.029972752043598,
+ "grad_norm": 3.8604257106781006,
+ "learning_rate": 1.6406631628793205e-05,
+ "loss": 0.2073,
+ "step": 11021
+ },
+ {
+ "epoch": 30.032697547683924,
+ "grad_norm": 4.093293190002441,
+ "learning_rate": 1.6405954005e-05,
+ "loss": 0.2255,
+ "step": 11022
+ },
+ {
+ "epoch": 30.03542234332425,
+ "grad_norm": 5.009720325469971,
+ "learning_rate": 1.6405276331317437e-05,
+ "loss": 0.2585,
+ "step": 11023
+ },
+ {
+ "epoch": 30.03814713896458,
+ "grad_norm": 3.437756299972534,
+ "learning_rate": 1.6404598607750785e-05,
+ "loss": 0.1514,
+ "step": 11024
+ },
+ {
+ "epoch": 30.040871934604905,
+ "grad_norm": 4.701558589935303,
+ "learning_rate": 1.6403920834305324e-05,
+ "loss": 0.1808,
+ "step": 11025
+ },
+ {
+ "epoch": 30.04359673024523,
+ "grad_norm": 4.2602763175964355,
+ "learning_rate": 1.6403243010986332e-05,
+ "loss": 0.1976,
+ "step": 11026
+ },
+ {
+ "epoch": 30.04632152588556,
+ "grad_norm": 4.851477146148682,
+ "learning_rate": 1.6402565137799096e-05,
+ "loss": 0.1788,
+ "step": 11027
+ },
+ {
+ "epoch": 30.049046321525886,
+ "grad_norm": 7.80436897277832,
+ "learning_rate": 1.6401887214748887e-05,
+ "loss": 0.2559,
+ "step": 11028
+ },
+ {
+ "epoch": 30.05177111716621,
+ "grad_norm": 4.645834922790527,
+ "learning_rate": 1.6401209241840984e-05,
+ "loss": 0.1741,
+ "step": 11029
+ },
+ {
+ "epoch": 30.05449591280654,
+ "grad_norm": 6.314472198486328,
+ "learning_rate": 1.640053121908067e-05,
+ "loss": 0.2666,
+ "step": 11030
+ },
+ {
+ "epoch": 30.057220708446867,
+ "grad_norm": 4.307480335235596,
+ "learning_rate": 1.6399853146473223e-05,
+ "loss": 0.179,
+ "step": 11031
+ },
+ {
+ "epoch": 30.059945504087192,
+ "grad_norm": 6.031696319580078,
+ "learning_rate": 1.6399175024023927e-05,
+ "loss": 0.2493,
+ "step": 11032
+ },
+ {
+ "epoch": 30.06267029972752,
+ "grad_norm": 4.831297397613525,
+ "learning_rate": 1.6398496851738064e-05,
+ "loss": 0.1774,
+ "step": 11033
+ },
+ {
+ "epoch": 30.065395095367847,
+ "grad_norm": 5.611319541931152,
+ "learning_rate": 1.639781862962091e-05,
+ "loss": 0.1674,
+ "step": 11034
+ },
+ {
+ "epoch": 30.068119891008173,
+ "grad_norm": 5.340045928955078,
+ "learning_rate": 1.6397140357677753e-05,
+ "loss": 0.3239,
+ "step": 11035
+ },
+ {
+ "epoch": 30.070844686648503,
+ "grad_norm": 4.730235576629639,
+ "learning_rate": 1.6396462035913872e-05,
+ "loss": 0.23,
+ "step": 11036
+ },
+ {
+ "epoch": 30.07356948228883,
+ "grad_norm": 5.3859686851501465,
+ "learning_rate": 1.639578366433455e-05,
+ "loss": 0.3794,
+ "step": 11037
+ },
+ {
+ "epoch": 30.076294277929154,
+ "grad_norm": 6.614158630371094,
+ "learning_rate": 1.639510524294507e-05,
+ "loss": 0.2022,
+ "step": 11038
+ },
+ {
+ "epoch": 30.079019073569484,
+ "grad_norm": 4.539222240447998,
+ "learning_rate": 1.639442677175072e-05,
+ "loss": 0.2212,
+ "step": 11039
+ },
+ {
+ "epoch": 30.08174386920981,
+ "grad_norm": 4.861076831817627,
+ "learning_rate": 1.639374825075678e-05,
+ "loss": 0.3153,
+ "step": 11040
+ },
+ {
+ "epoch": 30.084468664850135,
+ "grad_norm": 5.070900917053223,
+ "learning_rate": 1.639306967996853e-05,
+ "loss": 0.3164,
+ "step": 11041
+ },
+ {
+ "epoch": 30.087193460490465,
+ "grad_norm": 3.919741153717041,
+ "learning_rate": 1.6392391059391258e-05,
+ "loss": 0.263,
+ "step": 11042
+ },
+ {
+ "epoch": 30.08991825613079,
+ "grad_norm": 4.097604274749756,
+ "learning_rate": 1.6391712389030253e-05,
+ "loss": 0.2602,
+ "step": 11043
+ },
+ {
+ "epoch": 30.092643051771116,
+ "grad_norm": 3.631805658340454,
+ "learning_rate": 1.63910336688908e-05,
+ "loss": 0.1587,
+ "step": 11044
+ },
+ {
+ "epoch": 30.095367847411445,
+ "grad_norm": 4.222127914428711,
+ "learning_rate": 1.639035489897818e-05,
+ "loss": 0.2666,
+ "step": 11045
+ },
+ {
+ "epoch": 30.09809264305177,
+ "grad_norm": 4.2098708152771,
+ "learning_rate": 1.6389676079297684e-05,
+ "loss": 0.2211,
+ "step": 11046
+ },
+ {
+ "epoch": 30.100817438692097,
+ "grad_norm": 6.670740127563477,
+ "learning_rate": 1.6388997209854595e-05,
+ "loss": 0.4906,
+ "step": 11047
+ },
+ {
+ "epoch": 30.103542234332426,
+ "grad_norm": 5.401679992675781,
+ "learning_rate": 1.63883182906542e-05,
+ "loss": 0.2229,
+ "step": 11048
+ },
+ {
+ "epoch": 30.106267029972752,
+ "grad_norm": 4.557952880859375,
+ "learning_rate": 1.6387639321701787e-05,
+ "loss": 0.2657,
+ "step": 11049
+ },
+ {
+ "epoch": 30.108991825613078,
+ "grad_norm": 4.163926124572754,
+ "learning_rate": 1.6386960303002647e-05,
+ "loss": 0.211,
+ "step": 11050
+ },
+ {
+ "epoch": 30.111716621253407,
+ "grad_norm": 4.83394193649292,
+ "learning_rate": 1.6386281234562065e-05,
+ "loss": 0.1846,
+ "step": 11051
+ },
+ {
+ "epoch": 30.114441416893733,
+ "grad_norm": 4.4056525230407715,
+ "learning_rate": 1.638560211638533e-05,
+ "loss": 0.2092,
+ "step": 11052
+ },
+ {
+ "epoch": 30.11716621253406,
+ "grad_norm": 3.874083995819092,
+ "learning_rate": 1.638492294847773e-05,
+ "loss": 0.2568,
+ "step": 11053
+ },
+ {
+ "epoch": 30.11989100817439,
+ "grad_norm": 4.195558547973633,
+ "learning_rate": 1.6384243730844556e-05,
+ "loss": 0.3187,
+ "step": 11054
+ },
+ {
+ "epoch": 30.122615803814714,
+ "grad_norm": 4.753927230834961,
+ "learning_rate": 1.63835644634911e-05,
+ "loss": 0.1863,
+ "step": 11055
+ },
+ {
+ "epoch": 30.12534059945504,
+ "grad_norm": 4.611453056335449,
+ "learning_rate": 1.6382885146422646e-05,
+ "loss": 0.2798,
+ "step": 11056
+ },
+ {
+ "epoch": 30.12806539509537,
+ "grad_norm": 4.653456687927246,
+ "learning_rate": 1.6382205779644487e-05,
+ "loss": 0.2402,
+ "step": 11057
+ },
+ {
+ "epoch": 30.130790190735695,
+ "grad_norm": 4.697696208953857,
+ "learning_rate": 1.638152636316192e-05,
+ "loss": 0.2448,
+ "step": 11058
+ },
+ {
+ "epoch": 30.13351498637602,
+ "grad_norm": 5.341663360595703,
+ "learning_rate": 1.638084689698023e-05,
+ "loss": 0.2923,
+ "step": 11059
+ },
+ {
+ "epoch": 30.13623978201635,
+ "grad_norm": 4.408714294433594,
+ "learning_rate": 1.6380167381104706e-05,
+ "loss": 0.281,
+ "step": 11060
+ },
+ {
+ "epoch": 30.138964577656676,
+ "grad_norm": 5.193309783935547,
+ "learning_rate": 1.6379487815540648e-05,
+ "loss": 0.2034,
+ "step": 11061
+ },
+ {
+ "epoch": 30.141689373297,
+ "grad_norm": 5.271580696105957,
+ "learning_rate": 1.6378808200293343e-05,
+ "loss": 0.3746,
+ "step": 11062
+ },
+ {
+ "epoch": 30.14441416893733,
+ "grad_norm": 4.708368301391602,
+ "learning_rate": 1.6378128535368084e-05,
+ "loss": 0.2205,
+ "step": 11063
+ },
+ {
+ "epoch": 30.147138964577657,
+ "grad_norm": 4.096297740936279,
+ "learning_rate": 1.6377448820770164e-05,
+ "loss": 0.1713,
+ "step": 11064
+ },
+ {
+ "epoch": 30.149863760217983,
+ "grad_norm": 4.438827991485596,
+ "learning_rate": 1.637676905650488e-05,
+ "loss": 0.186,
+ "step": 11065
+ },
+ {
+ "epoch": 30.152588555858312,
+ "grad_norm": 4.9791154861450195,
+ "learning_rate": 1.637608924257752e-05,
+ "loss": 0.3202,
+ "step": 11066
+ },
+ {
+ "epoch": 30.155313351498638,
+ "grad_norm": 5.891482830047607,
+ "learning_rate": 1.6375409378993387e-05,
+ "loss": 0.2246,
+ "step": 11067
+ },
+ {
+ "epoch": 30.158038147138964,
+ "grad_norm": 33.69089126586914,
+ "learning_rate": 1.637472946575777e-05,
+ "loss": 0.4628,
+ "step": 11068
+ },
+ {
+ "epoch": 30.160762942779293,
+ "grad_norm": 4.404466152191162,
+ "learning_rate": 1.6374049502875963e-05,
+ "loss": 0.2863,
+ "step": 11069
+ },
+ {
+ "epoch": 30.16348773841962,
+ "grad_norm": 6.054387092590332,
+ "learning_rate": 1.6373369490353265e-05,
+ "loss": 0.3682,
+ "step": 11070
+ },
+ {
+ "epoch": 30.166212534059945,
+ "grad_norm": 4.2704548835754395,
+ "learning_rate": 1.637268942819497e-05,
+ "loss": 0.2263,
+ "step": 11071
+ },
+ {
+ "epoch": 30.168937329700274,
+ "grad_norm": 4.712883949279785,
+ "learning_rate": 1.6372009316406376e-05,
+ "loss": 0.2667,
+ "step": 11072
+ },
+ {
+ "epoch": 30.1716621253406,
+ "grad_norm": 4.4753499031066895,
+ "learning_rate": 1.6371329154992775e-05,
+ "loss": 0.404,
+ "step": 11073
+ },
+ {
+ "epoch": 30.174386920980925,
+ "grad_norm": 5.137465953826904,
+ "learning_rate": 1.6370648943959473e-05,
+ "loss": 0.2366,
+ "step": 11074
+ },
+ {
+ "epoch": 30.177111716621255,
+ "grad_norm": 5.345340728759766,
+ "learning_rate": 1.6369968683311758e-05,
+ "loss": 0.433,
+ "step": 11075
+ },
+ {
+ "epoch": 30.17983651226158,
+ "grad_norm": 4.506862640380859,
+ "learning_rate": 1.6369288373054936e-05,
+ "loss": 0.4458,
+ "step": 11076
+ },
+ {
+ "epoch": 30.182561307901906,
+ "grad_norm": 4.623584270477295,
+ "learning_rate": 1.6368608013194296e-05,
+ "loss": 0.367,
+ "step": 11077
+ },
+ {
+ "epoch": 30.185286103542236,
+ "grad_norm": 5.97453498840332,
+ "learning_rate": 1.6367927603735144e-05,
+ "loss": 0.2979,
+ "step": 11078
+ },
+ {
+ "epoch": 30.18801089918256,
+ "grad_norm": 4.154163360595703,
+ "learning_rate": 1.636724714468278e-05,
+ "loss": 0.2118,
+ "step": 11079
+ },
+ {
+ "epoch": 30.190735694822887,
+ "grad_norm": 4.802804470062256,
+ "learning_rate": 1.6366566636042495e-05,
+ "loss": 0.3049,
+ "step": 11080
+ },
+ {
+ "epoch": 30.193460490463217,
+ "grad_norm": 5.274569988250732,
+ "learning_rate": 1.6365886077819597e-05,
+ "loss": 0.384,
+ "step": 11081
+ },
+ {
+ "epoch": 30.196185286103542,
+ "grad_norm": 5.44789457321167,
+ "learning_rate": 1.6365205470019383e-05,
+ "loss": 0.4189,
+ "step": 11082
+ },
+ {
+ "epoch": 30.19891008174387,
+ "grad_norm": 4.411463737487793,
+ "learning_rate": 1.6364524812647155e-05,
+ "loss": 0.3123,
+ "step": 11083
+ },
+ {
+ "epoch": 30.201634877384198,
+ "grad_norm": 4.093128204345703,
+ "learning_rate": 1.636384410570821e-05,
+ "loss": 0.2962,
+ "step": 11084
+ },
+ {
+ "epoch": 30.204359673024523,
+ "grad_norm": 4.298640727996826,
+ "learning_rate": 1.636316334920785e-05,
+ "loss": 0.2897,
+ "step": 11085
+ },
+ {
+ "epoch": 30.20708446866485,
+ "grad_norm": 4.591917991638184,
+ "learning_rate": 1.6362482543151382e-05,
+ "loss": 0.2518,
+ "step": 11086
+ },
+ {
+ "epoch": 30.20980926430518,
+ "grad_norm": 4.587109565734863,
+ "learning_rate": 1.6361801687544103e-05,
+ "loss": 0.2465,
+ "step": 11087
+ },
+ {
+ "epoch": 30.212534059945504,
+ "grad_norm": 4.755136489868164,
+ "learning_rate": 1.636112078239132e-05,
+ "loss": 0.2012,
+ "step": 11088
+ },
+ {
+ "epoch": 30.21525885558583,
+ "grad_norm": 3.7357871532440186,
+ "learning_rate": 1.636043982769833e-05,
+ "loss": 0.3669,
+ "step": 11089
+ },
+ {
+ "epoch": 30.21798365122616,
+ "grad_norm": 5.404349327087402,
+ "learning_rate": 1.635975882347044e-05,
+ "loss": 0.3434,
+ "step": 11090
+ },
+ {
+ "epoch": 30.220708446866485,
+ "grad_norm": 3.797605037689209,
+ "learning_rate": 1.6359077769712955e-05,
+ "loss": 0.1622,
+ "step": 11091
+ },
+ {
+ "epoch": 30.22343324250681,
+ "grad_norm": 4.463130950927734,
+ "learning_rate": 1.6358396666431176e-05,
+ "loss": 0.2743,
+ "step": 11092
+ },
+ {
+ "epoch": 30.22615803814714,
+ "grad_norm": 4.495753288269043,
+ "learning_rate": 1.6357715513630407e-05,
+ "loss": 0.1918,
+ "step": 11093
+ },
+ {
+ "epoch": 30.228882833787466,
+ "grad_norm": 3.635967254638672,
+ "learning_rate": 1.6357034311315954e-05,
+ "loss": 0.1514,
+ "step": 11094
+ },
+ {
+ "epoch": 30.231607629427792,
+ "grad_norm": 9.035719871520996,
+ "learning_rate": 1.635635305949312e-05,
+ "loss": 0.2926,
+ "step": 11095
+ },
+ {
+ "epoch": 30.23433242506812,
+ "grad_norm": 4.195995330810547,
+ "learning_rate": 1.6355671758167218e-05,
+ "loss": 0.2297,
+ "step": 11096
+ },
+ {
+ "epoch": 30.237057220708447,
+ "grad_norm": 6.496492862701416,
+ "learning_rate": 1.635499040734355e-05,
+ "loss": 0.2204,
+ "step": 11097
+ },
+ {
+ "epoch": 30.239782016348773,
+ "grad_norm": 5.197145462036133,
+ "learning_rate": 1.6354309007027417e-05,
+ "loss": 0.2449,
+ "step": 11098
+ },
+ {
+ "epoch": 30.242506811989102,
+ "grad_norm": 4.88724422454834,
+ "learning_rate": 1.635362755722413e-05,
+ "loss": 0.3269,
+ "step": 11099
+ },
+ {
+ "epoch": 30.245231607629428,
+ "grad_norm": 5.146007537841797,
+ "learning_rate": 1.6352946057938998e-05,
+ "loss": 0.2155,
+ "step": 11100
+ },
+ {
+ "epoch": 30.247956403269754,
+ "grad_norm": 6.393765449523926,
+ "learning_rate": 1.635226450917733e-05,
+ "loss": 0.1685,
+ "step": 11101
+ },
+ {
+ "epoch": 30.250681198910083,
+ "grad_norm": 4.552482604980469,
+ "learning_rate": 1.635158291094442e-05,
+ "loss": 0.2662,
+ "step": 11102
+ },
+ {
+ "epoch": 30.25340599455041,
+ "grad_norm": 4.136221408843994,
+ "learning_rate": 1.6350901263245594e-05,
+ "loss": 0.1879,
+ "step": 11103
+ },
+ {
+ "epoch": 30.256130790190735,
+ "grad_norm": 5.0724968910217285,
+ "learning_rate": 1.6350219566086156e-05,
+ "loss": 0.3266,
+ "step": 11104
+ },
+ {
+ "epoch": 30.258855585831064,
+ "grad_norm": 4.358459949493408,
+ "learning_rate": 1.6349537819471408e-05,
+ "loss": 0.3071,
+ "step": 11105
+ },
+ {
+ "epoch": 30.26158038147139,
+ "grad_norm": 5.012463092803955,
+ "learning_rate": 1.6348856023406666e-05,
+ "loss": 0.2729,
+ "step": 11106
+ },
+ {
+ "epoch": 30.264305177111716,
+ "grad_norm": 5.757782459259033,
+ "learning_rate": 1.6348174177897237e-05,
+ "loss": 0.3199,
+ "step": 11107
+ },
+ {
+ "epoch": 30.267029972752045,
+ "grad_norm": 4.2073163986206055,
+ "learning_rate": 1.634749228294843e-05,
+ "loss": 0.3328,
+ "step": 11108
+ },
+ {
+ "epoch": 30.26975476839237,
+ "grad_norm": 4.076428413391113,
+ "learning_rate": 1.634681033856556e-05,
+ "loss": 0.2555,
+ "step": 11109
+ },
+ {
+ "epoch": 30.272479564032697,
+ "grad_norm": 4.141939640045166,
+ "learning_rate": 1.6346128344753932e-05,
+ "loss": 0.282,
+ "step": 11110
+ },
+ {
+ "epoch": 30.275204359673026,
+ "grad_norm": 5.177359580993652,
+ "learning_rate": 1.6345446301518864e-05,
+ "loss": 0.2167,
+ "step": 11111
+ },
+ {
+ "epoch": 30.277929155313352,
+ "grad_norm": 4.151187419891357,
+ "learning_rate": 1.634476420886566e-05,
+ "loss": 0.3533,
+ "step": 11112
+ },
+ {
+ "epoch": 30.280653950953678,
+ "grad_norm": 3.9282584190368652,
+ "learning_rate": 1.634408206679964e-05,
+ "loss": 0.1964,
+ "step": 11113
+ },
+ {
+ "epoch": 30.283378746594007,
+ "grad_norm": 3.579700469970703,
+ "learning_rate": 1.6343399875326115e-05,
+ "loss": 0.2759,
+ "step": 11114
+ },
+ {
+ "epoch": 30.286103542234333,
+ "grad_norm": 4.597183704376221,
+ "learning_rate": 1.6342717634450393e-05,
+ "loss": 0.2432,
+ "step": 11115
+ },
+ {
+ "epoch": 30.28882833787466,
+ "grad_norm": 4.727047443389893,
+ "learning_rate": 1.634203534417779e-05,
+ "loss": 0.3857,
+ "step": 11116
+ },
+ {
+ "epoch": 30.291553133514988,
+ "grad_norm": 5.664674282073975,
+ "learning_rate": 1.6341353004513622e-05,
+ "loss": 0.4093,
+ "step": 11117
+ },
+ {
+ "epoch": 30.294277929155314,
+ "grad_norm": 4.2413859367370605,
+ "learning_rate": 1.63406706154632e-05,
+ "loss": 0.4193,
+ "step": 11118
+ },
+ {
+ "epoch": 30.29700272479564,
+ "grad_norm": 5.492681980133057,
+ "learning_rate": 1.633998817703184e-05,
+ "loss": 0.2528,
+ "step": 11119
+ },
+ {
+ "epoch": 30.29972752043597,
+ "grad_norm": 3.88256573677063,
+ "learning_rate": 1.6339305689224853e-05,
+ "loss": 0.2589,
+ "step": 11120
+ },
+ {
+ "epoch": 30.302452316076295,
+ "grad_norm": 4.069121837615967,
+ "learning_rate": 1.633862315204756e-05,
+ "loss": 0.2524,
+ "step": 11121
+ },
+ {
+ "epoch": 30.30517711171662,
+ "grad_norm": 5.239743709564209,
+ "learning_rate": 1.6337940565505275e-05,
+ "loss": 0.2442,
+ "step": 11122
+ },
+ {
+ "epoch": 30.30790190735695,
+ "grad_norm": 5.316188812255859,
+ "learning_rate": 1.633725792960331e-05,
+ "loss": 0.3052,
+ "step": 11123
+ },
+ {
+ "epoch": 30.310626702997276,
+ "grad_norm": 5.093254566192627,
+ "learning_rate": 1.6336575244346983e-05,
+ "loss": 0.2306,
+ "step": 11124
+ },
+ {
+ "epoch": 30.3133514986376,
+ "grad_norm": 4.788089752197266,
+ "learning_rate": 1.6335892509741618e-05,
+ "loss": 0.2638,
+ "step": 11125
+ },
+ {
+ "epoch": 30.31607629427793,
+ "grad_norm": 3.945237874984741,
+ "learning_rate": 1.633520972579252e-05,
+ "loss": 0.2151,
+ "step": 11126
+ },
+ {
+ "epoch": 30.318801089918257,
+ "grad_norm": 4.2455830574035645,
+ "learning_rate": 1.6334526892505015e-05,
+ "loss": 0.1703,
+ "step": 11127
+ },
+ {
+ "epoch": 30.321525885558582,
+ "grad_norm": 4.70673131942749,
+ "learning_rate": 1.633384400988442e-05,
+ "loss": 0.4934,
+ "step": 11128
+ },
+ {
+ "epoch": 30.32425068119891,
+ "grad_norm": 5.045173645019531,
+ "learning_rate": 1.633316107793605e-05,
+ "loss": 0.2346,
+ "step": 11129
+ },
+ {
+ "epoch": 30.326975476839237,
+ "grad_norm": 4.360701084136963,
+ "learning_rate": 1.6332478096665222e-05,
+ "loss": 0.2492,
+ "step": 11130
+ },
+ {
+ "epoch": 30.329700272479563,
+ "grad_norm": 4.257909297943115,
+ "learning_rate": 1.6331795066077262e-05,
+ "loss": 0.1692,
+ "step": 11131
+ },
+ {
+ "epoch": 30.332425068119893,
+ "grad_norm": 4.236592769622803,
+ "learning_rate": 1.633111198617749e-05,
+ "loss": 0.2047,
+ "step": 11132
+ },
+ {
+ "epoch": 30.33514986376022,
+ "grad_norm": 4.943687438964844,
+ "learning_rate": 1.6330428856971217e-05,
+ "loss": 0.1967,
+ "step": 11133
+ },
+ {
+ "epoch": 30.337874659400544,
+ "grad_norm": 4.278594970703125,
+ "learning_rate": 1.6329745678463765e-05,
+ "loss": 0.2325,
+ "step": 11134
+ },
+ {
+ "epoch": 30.340599455040874,
+ "grad_norm": 3.944190263748169,
+ "learning_rate": 1.6329062450660463e-05,
+ "loss": 0.242,
+ "step": 11135
+ },
+ {
+ "epoch": 30.3433242506812,
+ "grad_norm": 5.026414394378662,
+ "learning_rate": 1.6328379173566623e-05,
+ "loss": 0.2312,
+ "step": 11136
+ },
+ {
+ "epoch": 30.346049046321525,
+ "grad_norm": 3.940580368041992,
+ "learning_rate": 1.632769584718757e-05,
+ "loss": 0.2464,
+ "step": 11137
+ },
+ {
+ "epoch": 30.348773841961854,
+ "grad_norm": 3.6633148193359375,
+ "learning_rate": 1.6327012471528626e-05,
+ "loss": 0.1426,
+ "step": 11138
+ },
+ {
+ "epoch": 30.35149863760218,
+ "grad_norm": 5.079965591430664,
+ "learning_rate": 1.6326329046595112e-05,
+ "loss": 0.3267,
+ "step": 11139
+ },
+ {
+ "epoch": 30.354223433242506,
+ "grad_norm": 3.7113659381866455,
+ "learning_rate": 1.632564557239235e-05,
+ "loss": 0.1722,
+ "step": 11140
+ },
+ {
+ "epoch": 30.356948228882835,
+ "grad_norm": 5.152378082275391,
+ "learning_rate": 1.6324962048925666e-05,
+ "loss": 0.1334,
+ "step": 11141
+ },
+ {
+ "epoch": 30.35967302452316,
+ "grad_norm": 4.890884876251221,
+ "learning_rate": 1.632427847620038e-05,
+ "loss": 0.2916,
+ "step": 11142
+ },
+ {
+ "epoch": 30.362397820163487,
+ "grad_norm": 4.852251052856445,
+ "learning_rate": 1.632359485422182e-05,
+ "loss": 0.2273,
+ "step": 11143
+ },
+ {
+ "epoch": 30.365122615803816,
+ "grad_norm": 5.2370805740356445,
+ "learning_rate": 1.63229111829953e-05,
+ "loss": 0.3145,
+ "step": 11144
+ },
+ {
+ "epoch": 30.367847411444142,
+ "grad_norm": 5.232222557067871,
+ "learning_rate": 1.6322227462526157e-05,
+ "loss": 0.2473,
+ "step": 11145
+ },
+ {
+ "epoch": 30.370572207084468,
+ "grad_norm": 4.746659755706787,
+ "learning_rate": 1.6321543692819707e-05,
+ "loss": 0.1694,
+ "step": 11146
+ },
+ {
+ "epoch": 30.373297002724797,
+ "grad_norm": 4.979283809661865,
+ "learning_rate": 1.6320859873881282e-05,
+ "loss": 0.3317,
+ "step": 11147
+ },
+ {
+ "epoch": 30.376021798365123,
+ "grad_norm": 5.762821197509766,
+ "learning_rate": 1.63201760057162e-05,
+ "loss": 0.2942,
+ "step": 11148
+ },
+ {
+ "epoch": 30.37874659400545,
+ "grad_norm": 5.119614124298096,
+ "learning_rate": 1.6319492088329792e-05,
+ "loss": 0.3111,
+ "step": 11149
+ },
+ {
+ "epoch": 30.381471389645778,
+ "grad_norm": 4.400900363922119,
+ "learning_rate": 1.631880812172738e-05,
+ "loss": 0.209,
+ "step": 11150
+ },
+ {
+ "epoch": 30.384196185286104,
+ "grad_norm": 19.228811264038086,
+ "learning_rate": 1.63181241059143e-05,
+ "loss": 0.1884,
+ "step": 11151
+ },
+ {
+ "epoch": 30.38692098092643,
+ "grad_norm": 5.919473171234131,
+ "learning_rate": 1.631744004089587e-05,
+ "loss": 0.2309,
+ "step": 11152
+ },
+ {
+ "epoch": 30.38964577656676,
+ "grad_norm": 6.414313793182373,
+ "learning_rate": 1.6316755926677415e-05,
+ "loss": 0.3009,
+ "step": 11153
+ },
+ {
+ "epoch": 30.392370572207085,
+ "grad_norm": 4.064546585083008,
+ "learning_rate": 1.6316071763264277e-05,
+ "loss": 0.3392,
+ "step": 11154
+ },
+ {
+ "epoch": 30.39509536784741,
+ "grad_norm": 4.356159210205078,
+ "learning_rate": 1.631538755066177e-05,
+ "loss": 0.3518,
+ "step": 11155
+ },
+ {
+ "epoch": 30.39782016348774,
+ "grad_norm": 5.704617023468018,
+ "learning_rate": 1.631470328887523e-05,
+ "loss": 0.2728,
+ "step": 11156
+ },
+ {
+ "epoch": 30.400544959128066,
+ "grad_norm": 3.902219295501709,
+ "learning_rate": 1.6314018977909982e-05,
+ "loss": 0.2178,
+ "step": 11157
+ },
+ {
+ "epoch": 30.40326975476839,
+ "grad_norm": 4.5928754806518555,
+ "learning_rate": 1.631333461777136e-05,
+ "loss": 0.2579,
+ "step": 11158
+ },
+ {
+ "epoch": 30.40599455040872,
+ "grad_norm": 4.3343400955200195,
+ "learning_rate": 1.631265020846469e-05,
+ "loss": 0.2286,
+ "step": 11159
+ },
+ {
+ "epoch": 30.408719346049047,
+ "grad_norm": 4.689605712890625,
+ "learning_rate": 1.63119657499953e-05,
+ "loss": 0.2177,
+ "step": 11160
+ },
+ {
+ "epoch": 30.411444141689373,
+ "grad_norm": 4.551525592803955,
+ "learning_rate": 1.631128124236853e-05,
+ "loss": 0.2467,
+ "step": 11161
+ },
+ {
+ "epoch": 30.414168937329702,
+ "grad_norm": 4.264888763427734,
+ "learning_rate": 1.6310596685589703e-05,
+ "loss": 0.2791,
+ "step": 11162
+ },
+ {
+ "epoch": 30.416893732970028,
+ "grad_norm": 8.772961616516113,
+ "learning_rate": 1.6309912079664147e-05,
+ "loss": 0.2806,
+ "step": 11163
+ },
+ {
+ "epoch": 30.419618528610354,
+ "grad_norm": 4.722085952758789,
+ "learning_rate": 1.6309227424597207e-05,
+ "loss": 0.2987,
+ "step": 11164
+ },
+ {
+ "epoch": 30.422343324250683,
+ "grad_norm": 4.473195552825928,
+ "learning_rate": 1.63085427203942e-05,
+ "loss": 0.34,
+ "step": 11165
+ },
+ {
+ "epoch": 30.42506811989101,
+ "grad_norm": 5.241474628448486,
+ "learning_rate": 1.6307857967060466e-05,
+ "loss": 0.3197,
+ "step": 11166
+ },
+ {
+ "epoch": 30.427792915531334,
+ "grad_norm": 4.745359897613525,
+ "learning_rate": 1.630717316460134e-05,
+ "loss": 0.1971,
+ "step": 11167
+ },
+ {
+ "epoch": 30.430517711171664,
+ "grad_norm": 4.64052152633667,
+ "learning_rate": 1.630648831302215e-05,
+ "loss": 0.2424,
+ "step": 11168
+ },
+ {
+ "epoch": 30.43324250681199,
+ "grad_norm": 4.817426681518555,
+ "learning_rate": 1.6305803412328236e-05,
+ "loss": 0.2726,
+ "step": 11169
+ },
+ {
+ "epoch": 30.435967302452315,
+ "grad_norm": 4.1605448722839355,
+ "learning_rate": 1.6305118462524926e-05,
+ "loss": 0.2171,
+ "step": 11170
+ },
+ {
+ "epoch": 30.438692098092645,
+ "grad_norm": 4.609991073608398,
+ "learning_rate": 1.6304433463617554e-05,
+ "loss": 0.2585,
+ "step": 11171
+ },
+ {
+ "epoch": 30.44141689373297,
+ "grad_norm": 5.25722074508667,
+ "learning_rate": 1.630374841561146e-05,
+ "loss": 0.3395,
+ "step": 11172
+ },
+ {
+ "epoch": 30.444141689373296,
+ "grad_norm": 4.443602085113525,
+ "learning_rate": 1.6303063318511973e-05,
+ "loss": 0.192,
+ "step": 11173
+ },
+ {
+ "epoch": 30.446866485013626,
+ "grad_norm": 10.03481674194336,
+ "learning_rate": 1.6302378172324436e-05,
+ "loss": 0.2863,
+ "step": 11174
+ },
+ {
+ "epoch": 30.44959128065395,
+ "grad_norm": 4.34266471862793,
+ "learning_rate": 1.6301692977054177e-05,
+ "loss": 0.1791,
+ "step": 11175
+ },
+ {
+ "epoch": 30.452316076294277,
+ "grad_norm": 5.087733745574951,
+ "learning_rate": 1.630100773270654e-05,
+ "loss": 0.2586,
+ "step": 11176
+ },
+ {
+ "epoch": 30.455040871934607,
+ "grad_norm": 7.118537425994873,
+ "learning_rate": 1.6300322439286852e-05,
+ "loss": 0.2843,
+ "step": 11177
+ },
+ {
+ "epoch": 30.457765667574932,
+ "grad_norm": 4.797135829925537,
+ "learning_rate": 1.629963709680046e-05,
+ "loss": 0.1949,
+ "step": 11178
+ },
+ {
+ "epoch": 30.460490463215258,
+ "grad_norm": 5.405027389526367,
+ "learning_rate": 1.6298951705252697e-05,
+ "loss": 0.2666,
+ "step": 11179
+ },
+ {
+ "epoch": 30.463215258855588,
+ "grad_norm": 4.738603591918945,
+ "learning_rate": 1.6298266264648898e-05,
+ "loss": 0.3318,
+ "step": 11180
+ },
+ {
+ "epoch": 30.465940054495913,
+ "grad_norm": 6.066418647766113,
+ "learning_rate": 1.6297580774994404e-05,
+ "loss": 0.395,
+ "step": 11181
+ },
+ {
+ "epoch": 30.46866485013624,
+ "grad_norm": 5.694511413574219,
+ "learning_rate": 1.6296895236294558e-05,
+ "loss": 0.2209,
+ "step": 11182
+ },
+ {
+ "epoch": 30.47138964577657,
+ "grad_norm": 4.796864032745361,
+ "learning_rate": 1.629620964855469e-05,
+ "loss": 0.4514,
+ "step": 11183
+ },
+ {
+ "epoch": 30.474114441416894,
+ "grad_norm": 4.333698749542236,
+ "learning_rate": 1.6295524011780147e-05,
+ "loss": 0.4033,
+ "step": 11184
+ },
+ {
+ "epoch": 30.47683923705722,
+ "grad_norm": 5.5399980545043945,
+ "learning_rate": 1.629483832597626e-05,
+ "loss": 0.2153,
+ "step": 11185
+ },
+ {
+ "epoch": 30.479564032697546,
+ "grad_norm": 4.4958977699279785,
+ "learning_rate": 1.6294152591148386e-05,
+ "loss": 0.2729,
+ "step": 11186
+ },
+ {
+ "epoch": 30.482288828337875,
+ "grad_norm": 5.1053009033203125,
+ "learning_rate": 1.6293466807301843e-05,
+ "loss": 0.2032,
+ "step": 11187
+ },
+ {
+ "epoch": 30.4850136239782,
+ "grad_norm": 6.287261486053467,
+ "learning_rate": 1.629278097444199e-05,
+ "loss": 0.1922,
+ "step": 11188
+ },
+ {
+ "epoch": 30.48773841961853,
+ "grad_norm": 4.804606914520264,
+ "learning_rate": 1.6292095092574154e-05,
+ "loss": 0.2607,
+ "step": 11189
+ },
+ {
+ "epoch": 30.490463215258856,
+ "grad_norm": 5.342662334442139,
+ "learning_rate": 1.629140916170369e-05,
+ "loss": 0.2954,
+ "step": 11190
+ },
+ {
+ "epoch": 30.493188010899182,
+ "grad_norm": 4.909309387207031,
+ "learning_rate": 1.6290723181835933e-05,
+ "loss": 0.3791,
+ "step": 11191
+ },
+ {
+ "epoch": 30.495912806539508,
+ "grad_norm": 4.572601795196533,
+ "learning_rate": 1.6290037152976227e-05,
+ "loss": 0.3054,
+ "step": 11192
+ },
+ {
+ "epoch": 30.498637602179837,
+ "grad_norm": 5.421943187713623,
+ "learning_rate": 1.628935107512991e-05,
+ "loss": 0.2596,
+ "step": 11193
+ },
+ {
+ "epoch": 30.501362397820163,
+ "grad_norm": 5.107943534851074,
+ "learning_rate": 1.6288664948302333e-05,
+ "loss": 0.1923,
+ "step": 11194
+ },
+ {
+ "epoch": 30.504087193460492,
+ "grad_norm": 3.984907388687134,
+ "learning_rate": 1.6287978772498836e-05,
+ "loss": 0.1695,
+ "step": 11195
+ },
+ {
+ "epoch": 30.506811989100818,
+ "grad_norm": 5.0493388175964355,
+ "learning_rate": 1.628729254772476e-05,
+ "loss": 0.3748,
+ "step": 11196
+ },
+ {
+ "epoch": 30.509536784741144,
+ "grad_norm": 4.9664459228515625,
+ "learning_rate": 1.6286606273985452e-05,
+ "loss": 0.189,
+ "step": 11197
+ },
+ {
+ "epoch": 30.51226158038147,
+ "grad_norm": 4.305378437042236,
+ "learning_rate": 1.628591995128626e-05,
+ "loss": 0.2411,
+ "step": 11198
+ },
+ {
+ "epoch": 30.5149863760218,
+ "grad_norm": 4.317323207855225,
+ "learning_rate": 1.6285233579632525e-05,
+ "loss": 0.39,
+ "step": 11199
+ },
+ {
+ "epoch": 30.517711171662125,
+ "grad_norm": 4.901630878448486,
+ "learning_rate": 1.628454715902959e-05,
+ "loss": 0.2656,
+ "step": 11200
+ },
+ {
+ "epoch": 30.520435967302454,
+ "grad_norm": 4.839759826660156,
+ "learning_rate": 1.628386068948281e-05,
+ "loss": 0.2834,
+ "step": 11201
+ },
+ {
+ "epoch": 30.52316076294278,
+ "grad_norm": 4.8471198081970215,
+ "learning_rate": 1.6283174170997516e-05,
+ "loss": 0.2296,
+ "step": 11202
+ },
+ {
+ "epoch": 30.525885558583106,
+ "grad_norm": 4.5352044105529785,
+ "learning_rate": 1.6282487603579073e-05,
+ "loss": 0.3103,
+ "step": 11203
+ },
+ {
+ "epoch": 30.52861035422343,
+ "grad_norm": 4.2862114906311035,
+ "learning_rate": 1.6281800987232815e-05,
+ "loss": 0.2169,
+ "step": 11204
+ },
+ {
+ "epoch": 30.53133514986376,
+ "grad_norm": 5.282539367675781,
+ "learning_rate": 1.628111432196409e-05,
+ "loss": 0.25,
+ "step": 11205
+ },
+ {
+ "epoch": 30.534059945504087,
+ "grad_norm": 8.682198524475098,
+ "learning_rate": 1.628042760777825e-05,
+ "loss": 0.1931,
+ "step": 11206
+ },
+ {
+ "epoch": 30.536784741144416,
+ "grad_norm": 4.524226665496826,
+ "learning_rate": 1.6279740844680647e-05,
+ "loss": 0.2017,
+ "step": 11207
+ },
+ {
+ "epoch": 30.539509536784742,
+ "grad_norm": 4.9775824546813965,
+ "learning_rate": 1.627905403267662e-05,
+ "loss": 0.2219,
+ "step": 11208
+ },
+ {
+ "epoch": 30.542234332425068,
+ "grad_norm": 4.215161323547363,
+ "learning_rate": 1.6278367171771522e-05,
+ "loss": 0.1926,
+ "step": 11209
+ },
+ {
+ "epoch": 30.544959128065393,
+ "grad_norm": 5.261019706726074,
+ "learning_rate": 1.6277680261970706e-05,
+ "loss": 0.3343,
+ "step": 11210
+ },
+ {
+ "epoch": 30.547683923705723,
+ "grad_norm": 4.678142070770264,
+ "learning_rate": 1.6276993303279514e-05,
+ "loss": 0.2954,
+ "step": 11211
+ },
+ {
+ "epoch": 30.55040871934605,
+ "grad_norm": 4.842193126678467,
+ "learning_rate": 1.6276306295703307e-05,
+ "loss": 0.2254,
+ "step": 11212
+ },
+ {
+ "epoch": 30.553133514986374,
+ "grad_norm": 4.701113224029541,
+ "learning_rate": 1.6275619239247423e-05,
+ "loss": 0.2187,
+ "step": 11213
+ },
+ {
+ "epoch": 30.555858310626704,
+ "grad_norm": 5.1930460929870605,
+ "learning_rate": 1.6274932133917217e-05,
+ "loss": 0.2975,
+ "step": 11214
+ },
+ {
+ "epoch": 30.55858310626703,
+ "grad_norm": 4.311717987060547,
+ "learning_rate": 1.6274244979718044e-05,
+ "loss": 0.2475,
+ "step": 11215
+ },
+ {
+ "epoch": 30.561307901907355,
+ "grad_norm": 4.714442253112793,
+ "learning_rate": 1.627355777665525e-05,
+ "loss": 0.2389,
+ "step": 11216
+ },
+ {
+ "epoch": 30.564032697547685,
+ "grad_norm": 4.605578899383545,
+ "learning_rate": 1.6272870524734193e-05,
+ "loss": 0.1841,
+ "step": 11217
+ },
+ {
+ "epoch": 30.56675749318801,
+ "grad_norm": 4.381612777709961,
+ "learning_rate": 1.6272183223960224e-05,
+ "loss": 0.2561,
+ "step": 11218
+ },
+ {
+ "epoch": 30.569482288828336,
+ "grad_norm": 7.074498176574707,
+ "learning_rate": 1.6271495874338692e-05,
+ "loss": 0.2103,
+ "step": 11219
+ },
+ {
+ "epoch": 30.572207084468666,
+ "grad_norm": 4.864585876464844,
+ "learning_rate": 1.6270808475874953e-05,
+ "loss": 0.3053,
+ "step": 11220
+ },
+ {
+ "epoch": 30.57493188010899,
+ "grad_norm": 4.339353561401367,
+ "learning_rate": 1.6270121028574358e-05,
+ "loss": 0.3358,
+ "step": 11221
+ },
+ {
+ "epoch": 30.577656675749317,
+ "grad_norm": 5.2776947021484375,
+ "learning_rate": 1.6269433532442262e-05,
+ "loss": 0.263,
+ "step": 11222
+ },
+ {
+ "epoch": 30.580381471389646,
+ "grad_norm": 5.282337188720703,
+ "learning_rate": 1.626874598748402e-05,
+ "loss": 0.2361,
+ "step": 11223
+ },
+ {
+ "epoch": 30.583106267029972,
+ "grad_norm": 3.960113286972046,
+ "learning_rate": 1.6268058393704986e-05,
+ "loss": 0.1295,
+ "step": 11224
+ },
+ {
+ "epoch": 30.585831062670298,
+ "grad_norm": 4.236714839935303,
+ "learning_rate": 1.6267370751110516e-05,
+ "loss": 0.2178,
+ "step": 11225
+ },
+ {
+ "epoch": 30.588555858310627,
+ "grad_norm": 4.524733543395996,
+ "learning_rate": 1.626668305970596e-05,
+ "loss": 0.2102,
+ "step": 11226
+ },
+ {
+ "epoch": 30.591280653950953,
+ "grad_norm": 4.649762153625488,
+ "learning_rate": 1.626599531949668e-05,
+ "loss": 0.4159,
+ "step": 11227
+ },
+ {
+ "epoch": 30.59400544959128,
+ "grad_norm": 4.562127590179443,
+ "learning_rate": 1.626530753048803e-05,
+ "loss": 0.2526,
+ "step": 11228
+ },
+ {
+ "epoch": 30.59673024523161,
+ "grad_norm": 3.30283522605896,
+ "learning_rate": 1.626461969268537e-05,
+ "loss": 0.2555,
+ "step": 11229
+ },
+ {
+ "epoch": 30.599455040871934,
+ "grad_norm": 5.1068549156188965,
+ "learning_rate": 1.6263931806094053e-05,
+ "loss": 0.2988,
+ "step": 11230
+ },
+ {
+ "epoch": 30.60217983651226,
+ "grad_norm": 5.821933746337891,
+ "learning_rate": 1.6263243870719436e-05,
+ "loss": 0.2396,
+ "step": 11231
+ },
+ {
+ "epoch": 30.60490463215259,
+ "grad_norm": 4.927003383636475,
+ "learning_rate": 1.6262555886566874e-05,
+ "loss": 0.1935,
+ "step": 11232
+ },
+ {
+ "epoch": 30.607629427792915,
+ "grad_norm": 4.659766674041748,
+ "learning_rate": 1.626186785364173e-05,
+ "loss": 0.3157,
+ "step": 11233
+ },
+ {
+ "epoch": 30.61035422343324,
+ "grad_norm": 4.377122402191162,
+ "learning_rate": 1.6261179771949364e-05,
+ "loss": 0.2778,
+ "step": 11234
+ },
+ {
+ "epoch": 30.61307901907357,
+ "grad_norm": 4.551229476928711,
+ "learning_rate": 1.6260491641495127e-05,
+ "loss": 0.4281,
+ "step": 11235
+ },
+ {
+ "epoch": 30.615803814713896,
+ "grad_norm": 7.83804988861084,
+ "learning_rate": 1.6259803462284383e-05,
+ "loss": 0.3576,
+ "step": 11236
+ },
+ {
+ "epoch": 30.618528610354222,
+ "grad_norm": 4.2204766273498535,
+ "learning_rate": 1.6259115234322492e-05,
+ "loss": 0.1695,
+ "step": 11237
+ },
+ {
+ "epoch": 30.62125340599455,
+ "grad_norm": 4.474100112915039,
+ "learning_rate": 1.6258426957614814e-05,
+ "loss": 0.2426,
+ "step": 11238
+ },
+ {
+ "epoch": 30.623978201634877,
+ "grad_norm": 4.316636562347412,
+ "learning_rate": 1.625773863216671e-05,
+ "loss": 0.3057,
+ "step": 11239
+ },
+ {
+ "epoch": 30.626702997275203,
+ "grad_norm": 5.035547256469727,
+ "learning_rate": 1.6257050257983536e-05,
+ "loss": 0.2776,
+ "step": 11240
+ },
+ {
+ "epoch": 30.629427792915532,
+ "grad_norm": 5.1060566902160645,
+ "learning_rate": 1.625636183507066e-05,
+ "loss": 0.3109,
+ "step": 11241
+ },
+ {
+ "epoch": 30.632152588555858,
+ "grad_norm": 4.196608066558838,
+ "learning_rate": 1.6255673363433432e-05,
+ "loss": 0.3662,
+ "step": 11242
+ },
+ {
+ "epoch": 30.634877384196184,
+ "grad_norm": 4.147790908813477,
+ "learning_rate": 1.6254984843077228e-05,
+ "loss": 0.3089,
+ "step": 11243
+ },
+ {
+ "epoch": 30.637602179836513,
+ "grad_norm": 3.92214035987854,
+ "learning_rate": 1.62542962740074e-05,
+ "loss": 0.2053,
+ "step": 11244
+ },
+ {
+ "epoch": 30.64032697547684,
+ "grad_norm": 4.128028392791748,
+ "learning_rate": 1.625360765622931e-05,
+ "loss": 0.3975,
+ "step": 11245
+ },
+ {
+ "epoch": 30.643051771117165,
+ "grad_norm": 5.641008377075195,
+ "learning_rate": 1.6252918989748333e-05,
+ "loss": 0.3725,
+ "step": 11246
+ },
+ {
+ "epoch": 30.645776566757494,
+ "grad_norm": 5.541108131408691,
+ "learning_rate": 1.625223027456982e-05,
+ "loss": 0.2817,
+ "step": 11247
+ },
+ {
+ "epoch": 30.64850136239782,
+ "grad_norm": 4.827045440673828,
+ "learning_rate": 1.625154151069914e-05,
+ "loss": 0.3329,
+ "step": 11248
+ },
+ {
+ "epoch": 30.651226158038146,
+ "grad_norm": 3.8411483764648438,
+ "learning_rate": 1.6250852698141654e-05,
+ "loss": 0.1839,
+ "step": 11249
+ },
+ {
+ "epoch": 30.653950953678475,
+ "grad_norm": 4.752475261688232,
+ "learning_rate": 1.625016383690273e-05,
+ "loss": 0.2137,
+ "step": 11250
+ },
+ {
+ "epoch": 30.6566757493188,
+ "grad_norm": 4.610032558441162,
+ "learning_rate": 1.624947492698773e-05,
+ "loss": 0.3487,
+ "step": 11251
+ },
+ {
+ "epoch": 30.659400544959126,
+ "grad_norm": 5.163581848144531,
+ "learning_rate": 1.624878596840202e-05,
+ "loss": 0.2803,
+ "step": 11252
+ },
+ {
+ "epoch": 30.662125340599456,
+ "grad_norm": 5.5058393478393555,
+ "learning_rate": 1.624809696115097e-05,
+ "loss": 0.2325,
+ "step": 11253
+ },
+ {
+ "epoch": 30.66485013623978,
+ "grad_norm": 3.7788517475128174,
+ "learning_rate": 1.624740790523994e-05,
+ "loss": 0.4203,
+ "step": 11254
+ },
+ {
+ "epoch": 30.667574931880107,
+ "grad_norm": 4.700873851776123,
+ "learning_rate": 1.6246718800674296e-05,
+ "loss": 0.265,
+ "step": 11255
+ },
+ {
+ "epoch": 30.670299727520437,
+ "grad_norm": 4.356136322021484,
+ "learning_rate": 1.624602964745941e-05,
+ "loss": 0.2702,
+ "step": 11256
+ },
+ {
+ "epoch": 30.673024523160763,
+ "grad_norm": 4.121758937835693,
+ "learning_rate": 1.6245340445600643e-05,
+ "loss": 0.1542,
+ "step": 11257
+ },
+ {
+ "epoch": 30.67574931880109,
+ "grad_norm": 5.455111026763916,
+ "learning_rate": 1.6244651195103367e-05,
+ "loss": 0.3417,
+ "step": 11258
+ },
+ {
+ "epoch": 30.678474114441418,
+ "grad_norm": 6.180759906768799,
+ "learning_rate": 1.6243961895972948e-05,
+ "loss": 0.3301,
+ "step": 11259
+ },
+ {
+ "epoch": 30.681198910081743,
+ "grad_norm": 4.6043548583984375,
+ "learning_rate": 1.6243272548214758e-05,
+ "loss": 0.1307,
+ "step": 11260
+ },
+ {
+ "epoch": 30.68392370572207,
+ "grad_norm": 4.298545837402344,
+ "learning_rate": 1.624258315183416e-05,
+ "loss": 0.3944,
+ "step": 11261
+ },
+ {
+ "epoch": 30.6866485013624,
+ "grad_norm": 4.98190450668335,
+ "learning_rate": 1.6241893706836525e-05,
+ "loss": 0.4952,
+ "step": 11262
+ },
+ {
+ "epoch": 30.689373297002724,
+ "grad_norm": 4.119872570037842,
+ "learning_rate": 1.6241204213227223e-05,
+ "loss": 0.2665,
+ "step": 11263
+ },
+ {
+ "epoch": 30.69209809264305,
+ "grad_norm": 4.332726001739502,
+ "learning_rate": 1.6240514671011623e-05,
+ "loss": 0.2366,
+ "step": 11264
+ },
+ {
+ "epoch": 30.69482288828338,
+ "grad_norm": 4.951672077178955,
+ "learning_rate": 1.6239825080195096e-05,
+ "loss": 0.3089,
+ "step": 11265
+ },
+ {
+ "epoch": 30.697547683923705,
+ "grad_norm": 4.8824028968811035,
+ "learning_rate": 1.623913544078301e-05,
+ "loss": 0.3115,
+ "step": 11266
+ },
+ {
+ "epoch": 30.70027247956403,
+ "grad_norm": 4.581954479217529,
+ "learning_rate": 1.6238445752780742e-05,
+ "loss": 0.2643,
+ "step": 11267
+ },
+ {
+ "epoch": 30.70299727520436,
+ "grad_norm": 4.374332904815674,
+ "learning_rate": 1.623775601619366e-05,
+ "loss": 0.2783,
+ "step": 11268
+ },
+ {
+ "epoch": 30.705722070844686,
+ "grad_norm": 5.503300189971924,
+ "learning_rate": 1.623706623102713e-05,
+ "loss": 0.2707,
+ "step": 11269
+ },
+ {
+ "epoch": 30.708446866485012,
+ "grad_norm": 4.837397575378418,
+ "learning_rate": 1.623637639728653e-05,
+ "loss": 0.2881,
+ "step": 11270
+ },
+ {
+ "epoch": 30.71117166212534,
+ "grad_norm": 4.72066068649292,
+ "learning_rate": 1.623568651497723e-05,
+ "loss": 0.3072,
+ "step": 11271
+ },
+ {
+ "epoch": 30.713896457765667,
+ "grad_norm": 4.468803882598877,
+ "learning_rate": 1.623499658410461e-05,
+ "loss": 0.1504,
+ "step": 11272
+ },
+ {
+ "epoch": 30.716621253405993,
+ "grad_norm": 7.393949508666992,
+ "learning_rate": 1.6234306604674034e-05,
+ "loss": 0.2109,
+ "step": 11273
+ },
+ {
+ "epoch": 30.719346049046322,
+ "grad_norm": 5.861691474914551,
+ "learning_rate": 1.6233616576690876e-05,
+ "loss": 0.3026,
+ "step": 11274
+ },
+ {
+ "epoch": 30.722070844686648,
+ "grad_norm": 4.427733421325684,
+ "learning_rate": 1.6232926500160514e-05,
+ "loss": 0.4192,
+ "step": 11275
+ },
+ {
+ "epoch": 30.724795640326974,
+ "grad_norm": 4.146992206573486,
+ "learning_rate": 1.6232236375088322e-05,
+ "loss": 0.3429,
+ "step": 11276
+ },
+ {
+ "epoch": 30.727520435967303,
+ "grad_norm": 5.602183818817139,
+ "learning_rate": 1.6231546201479676e-05,
+ "loss": 0.3837,
+ "step": 11277
+ },
+ {
+ "epoch": 30.73024523160763,
+ "grad_norm": 4.2915778160095215,
+ "learning_rate": 1.623085597933995e-05,
+ "loss": 0.2582,
+ "step": 11278
+ },
+ {
+ "epoch": 30.732970027247955,
+ "grad_norm": 4.450540065765381,
+ "learning_rate": 1.6230165708674513e-05,
+ "loss": 0.2926,
+ "step": 11279
+ },
+ {
+ "epoch": 30.735694822888284,
+ "grad_norm": 4.728268623352051,
+ "learning_rate": 1.622947538948875e-05,
+ "loss": 0.3314,
+ "step": 11280
+ },
+ {
+ "epoch": 30.73841961852861,
+ "grad_norm": 4.299027442932129,
+ "learning_rate": 1.622878502178803e-05,
+ "loss": 0.3269,
+ "step": 11281
+ },
+ {
+ "epoch": 30.741144414168936,
+ "grad_norm": 3.85858154296875,
+ "learning_rate": 1.6228094605577734e-05,
+ "loss": 0.2463,
+ "step": 11282
+ },
+ {
+ "epoch": 30.743869209809265,
+ "grad_norm": 5.698709964752197,
+ "learning_rate": 1.6227404140863233e-05,
+ "loss": 0.2108,
+ "step": 11283
+ },
+ {
+ "epoch": 30.74659400544959,
+ "grad_norm": 3.806607484817505,
+ "learning_rate": 1.6226713627649914e-05,
+ "loss": 0.1714,
+ "step": 11284
+ },
+ {
+ "epoch": 30.749318801089917,
+ "grad_norm": 5.195178508758545,
+ "learning_rate": 1.622602306594315e-05,
+ "loss": 0.2538,
+ "step": 11285
+ },
+ {
+ "epoch": 30.752043596730246,
+ "grad_norm": 5.6360578536987305,
+ "learning_rate": 1.622533245574832e-05,
+ "loss": 0.2234,
+ "step": 11286
+ },
+ {
+ "epoch": 30.754768392370572,
+ "grad_norm": 4.190375328063965,
+ "learning_rate": 1.62246417970708e-05,
+ "loss": 0.1669,
+ "step": 11287
+ },
+ {
+ "epoch": 30.757493188010898,
+ "grad_norm": 4.265194416046143,
+ "learning_rate": 1.6223951089915968e-05,
+ "loss": 0.2747,
+ "step": 11288
+ },
+ {
+ "epoch": 30.760217983651227,
+ "grad_norm": 4.723963260650635,
+ "learning_rate": 1.6223260334289208e-05,
+ "loss": 0.3575,
+ "step": 11289
+ },
+ {
+ "epoch": 30.762942779291553,
+ "grad_norm": 5.299036979675293,
+ "learning_rate": 1.6222569530195894e-05,
+ "loss": 0.3556,
+ "step": 11290
+ },
+ {
+ "epoch": 30.76566757493188,
+ "grad_norm": 4.612747669219971,
+ "learning_rate": 1.622187867764141e-05,
+ "loss": 0.3401,
+ "step": 11291
+ },
+ {
+ "epoch": 30.768392370572208,
+ "grad_norm": 13.278213500976562,
+ "learning_rate": 1.6221187776631134e-05,
+ "loss": 0.2108,
+ "step": 11292
+ },
+ {
+ "epoch": 30.771117166212534,
+ "grad_norm": 4.459288120269775,
+ "learning_rate": 1.622049682717045e-05,
+ "loss": 0.2577,
+ "step": 11293
+ },
+ {
+ "epoch": 30.77384196185286,
+ "grad_norm": 4.341718673706055,
+ "learning_rate": 1.6219805829264738e-05,
+ "loss": 0.1836,
+ "step": 11294
+ },
+ {
+ "epoch": 30.77656675749319,
+ "grad_norm": 4.311419486999512,
+ "learning_rate": 1.621911478291937e-05,
+ "loss": 0.3585,
+ "step": 11295
+ },
+ {
+ "epoch": 30.779291553133515,
+ "grad_norm": 4.629698753356934,
+ "learning_rate": 1.6218423688139748e-05,
+ "loss": 0.2234,
+ "step": 11296
+ },
+ {
+ "epoch": 30.78201634877384,
+ "grad_norm": 4.519668102264404,
+ "learning_rate": 1.6217732544931237e-05,
+ "loss": 0.1829,
+ "step": 11297
+ },
+ {
+ "epoch": 30.78474114441417,
+ "grad_norm": 5.189295291900635,
+ "learning_rate": 1.6217041353299224e-05,
+ "loss": 0.2753,
+ "step": 11298
+ },
+ {
+ "epoch": 30.787465940054496,
+ "grad_norm": 4.184066295623779,
+ "learning_rate": 1.6216350113249094e-05,
+ "loss": 0.2791,
+ "step": 11299
+ },
+ {
+ "epoch": 30.79019073569482,
+ "grad_norm": 4.4450788497924805,
+ "learning_rate": 1.6215658824786226e-05,
+ "loss": 0.2473,
+ "step": 11300
+ },
+ {
+ "epoch": 30.79291553133515,
+ "grad_norm": 4.40746545791626,
+ "learning_rate": 1.6214967487916012e-05,
+ "loss": 0.1877,
+ "step": 11301
+ },
+ {
+ "epoch": 30.795640326975477,
+ "grad_norm": 4.493723392486572,
+ "learning_rate": 1.6214276102643828e-05,
+ "loss": 0.2032,
+ "step": 11302
+ },
+ {
+ "epoch": 30.798365122615802,
+ "grad_norm": 4.038294792175293,
+ "learning_rate": 1.6213584668975064e-05,
+ "loss": 0.1921,
+ "step": 11303
+ },
+ {
+ "epoch": 30.80108991825613,
+ "grad_norm": 3.832259178161621,
+ "learning_rate": 1.6212893186915105e-05,
+ "loss": 0.1886,
+ "step": 11304
+ },
+ {
+ "epoch": 30.803814713896458,
+ "grad_norm": 4.343914031982422,
+ "learning_rate": 1.6212201656469328e-05,
+ "loss": 0.4265,
+ "step": 11305
+ },
+ {
+ "epoch": 30.806539509536783,
+ "grad_norm": 5.041415691375732,
+ "learning_rate": 1.621151007764313e-05,
+ "loss": 0.2022,
+ "step": 11306
+ },
+ {
+ "epoch": 30.809264305177113,
+ "grad_norm": 5.410445690155029,
+ "learning_rate": 1.621081845044189e-05,
+ "loss": 0.1829,
+ "step": 11307
+ },
+ {
+ "epoch": 30.81198910081744,
+ "grad_norm": 136.3157196044922,
+ "learning_rate": 1.6210126774870992e-05,
+ "loss": 0.2757,
+ "step": 11308
+ },
+ {
+ "epoch": 30.814713896457764,
+ "grad_norm": 4.570010185241699,
+ "learning_rate": 1.620943505093583e-05,
+ "loss": 0.2284,
+ "step": 11309
+ },
+ {
+ "epoch": 30.817438692098094,
+ "grad_norm": 5.480228424072266,
+ "learning_rate": 1.620874327864179e-05,
+ "loss": 0.2266,
+ "step": 11310
+ },
+ {
+ "epoch": 30.82016348773842,
+ "grad_norm": 4.866992950439453,
+ "learning_rate": 1.620805145799425e-05,
+ "loss": 0.3292,
+ "step": 11311
+ },
+ {
+ "epoch": 30.822888283378745,
+ "grad_norm": 6.822628021240234,
+ "learning_rate": 1.6207359588998608e-05,
+ "loss": 0.3196,
+ "step": 11312
+ },
+ {
+ "epoch": 30.825613079019075,
+ "grad_norm": 4.831575393676758,
+ "learning_rate": 1.620666767166025e-05,
+ "loss": 0.3292,
+ "step": 11313
+ },
+ {
+ "epoch": 30.8283378746594,
+ "grad_norm": 6.605419158935547,
+ "learning_rate": 1.6205975705984565e-05,
+ "loss": 0.2454,
+ "step": 11314
+ },
+ {
+ "epoch": 30.831062670299726,
+ "grad_norm": 7.327605724334717,
+ "learning_rate": 1.620528369197694e-05,
+ "loss": 0.2229,
+ "step": 11315
+ },
+ {
+ "epoch": 30.833787465940055,
+ "grad_norm": 5.607766628265381,
+ "learning_rate": 1.6204591629642767e-05,
+ "loss": 0.2592,
+ "step": 11316
+ },
+ {
+ "epoch": 30.83651226158038,
+ "grad_norm": 6.251765727996826,
+ "learning_rate": 1.620389951898743e-05,
+ "loss": 0.2474,
+ "step": 11317
+ },
+ {
+ "epoch": 30.839237057220707,
+ "grad_norm": 5.385140419006348,
+ "learning_rate": 1.6203207360016328e-05,
+ "loss": 0.3005,
+ "step": 11318
+ },
+ {
+ "epoch": 30.841961852861036,
+ "grad_norm": 6.696568012237549,
+ "learning_rate": 1.6202515152734844e-05,
+ "loss": 0.3288,
+ "step": 11319
+ },
+ {
+ "epoch": 30.844686648501362,
+ "grad_norm": 5.201862335205078,
+ "learning_rate": 1.620182289714837e-05,
+ "loss": 0.276,
+ "step": 11320
+ },
+ {
+ "epoch": 30.847411444141688,
+ "grad_norm": 8.074922561645508,
+ "learning_rate": 1.62011305932623e-05,
+ "loss": 0.4071,
+ "step": 11321
+ },
+ {
+ "epoch": 30.850136239782017,
+ "grad_norm": 6.197356224060059,
+ "learning_rate": 1.6200438241082032e-05,
+ "loss": 0.2867,
+ "step": 11322
+ },
+ {
+ "epoch": 30.852861035422343,
+ "grad_norm": 5.983175754547119,
+ "learning_rate": 1.6199745840612942e-05,
+ "loss": 0.3126,
+ "step": 11323
+ },
+ {
+ "epoch": 30.85558583106267,
+ "grad_norm": 5.219690322875977,
+ "learning_rate": 1.619905339186043e-05,
+ "loss": 0.472,
+ "step": 11324
+ },
+ {
+ "epoch": 30.858310626703,
+ "grad_norm": 6.298698902130127,
+ "learning_rate": 1.6198360894829893e-05,
+ "loss": 0.3904,
+ "step": 11325
+ },
+ {
+ "epoch": 30.861035422343324,
+ "grad_norm": 7.4075775146484375,
+ "learning_rate": 1.619766834952672e-05,
+ "loss": 0.4625,
+ "step": 11326
+ },
+ {
+ "epoch": 30.86376021798365,
+ "grad_norm": 7.002719879150391,
+ "learning_rate": 1.6196975755956308e-05,
+ "loss": 0.2858,
+ "step": 11327
+ },
+ {
+ "epoch": 30.86648501362398,
+ "grad_norm": 5.470277786254883,
+ "learning_rate": 1.6196283114124045e-05,
+ "loss": 0.4653,
+ "step": 11328
+ },
+ {
+ "epoch": 30.869209809264305,
+ "grad_norm": 5.387900352478027,
+ "learning_rate": 1.619559042403533e-05,
+ "loss": 0.3419,
+ "step": 11329
+ },
+ {
+ "epoch": 30.87193460490463,
+ "grad_norm": 5.5453009605407715,
+ "learning_rate": 1.6194897685695558e-05,
+ "loss": 0.3119,
+ "step": 11330
+ },
+ {
+ "epoch": 30.87465940054496,
+ "grad_norm": 6.223090171813965,
+ "learning_rate": 1.6194204899110118e-05,
+ "loss": 0.3216,
+ "step": 11331
+ },
+ {
+ "epoch": 30.877384196185286,
+ "grad_norm": 6.402840614318848,
+ "learning_rate": 1.619351206428441e-05,
+ "loss": 0.3142,
+ "step": 11332
+ },
+ {
+ "epoch": 30.88010899182561,
+ "grad_norm": 5.013248920440674,
+ "learning_rate": 1.6192819181223835e-05,
+ "loss": 0.2729,
+ "step": 11333
+ },
+ {
+ "epoch": 30.88283378746594,
+ "grad_norm": 5.531585693359375,
+ "learning_rate": 1.6192126249933776e-05,
+ "loss": 0.7619,
+ "step": 11334
+ },
+ {
+ "epoch": 30.885558583106267,
+ "grad_norm": 6.204798221588135,
+ "learning_rate": 1.619143327041964e-05,
+ "loss": 0.3621,
+ "step": 11335
+ },
+ {
+ "epoch": 30.888283378746593,
+ "grad_norm": 5.826322555541992,
+ "learning_rate": 1.619074024268682e-05,
+ "loss": 0.3726,
+ "step": 11336
+ },
+ {
+ "epoch": 30.891008174386922,
+ "grad_norm": 5.286293029785156,
+ "learning_rate": 1.619004716674072e-05,
+ "loss": 0.3716,
+ "step": 11337
+ },
+ {
+ "epoch": 30.893732970027248,
+ "grad_norm": 6.445289611816406,
+ "learning_rate": 1.6189354042586728e-05,
+ "loss": 0.2957,
+ "step": 11338
+ },
+ {
+ "epoch": 30.896457765667574,
+ "grad_norm": 6.290192127227783,
+ "learning_rate": 1.6188660870230246e-05,
+ "loss": 0.2527,
+ "step": 11339
+ },
+ {
+ "epoch": 30.899182561307903,
+ "grad_norm": 5.1497483253479,
+ "learning_rate": 1.6187967649676674e-05,
+ "loss": 0.2993,
+ "step": 11340
+ },
+ {
+ "epoch": 30.90190735694823,
+ "grad_norm": 6.877128601074219,
+ "learning_rate": 1.6187274380931403e-05,
+ "loss": 0.2647,
+ "step": 11341
+ },
+ {
+ "epoch": 30.904632152588555,
+ "grad_norm": 4.928369522094727,
+ "learning_rate": 1.6186581063999844e-05,
+ "loss": 0.2753,
+ "step": 11342
+ },
+ {
+ "epoch": 30.907356948228884,
+ "grad_norm": 4.909536838531494,
+ "learning_rate": 1.6185887698887388e-05,
+ "loss": 0.3789,
+ "step": 11343
+ },
+ {
+ "epoch": 30.91008174386921,
+ "grad_norm": 5.300933361053467,
+ "learning_rate": 1.618519428559944e-05,
+ "loss": 0.208,
+ "step": 11344
+ },
+ {
+ "epoch": 30.912806539509535,
+ "grad_norm": 5.597089767456055,
+ "learning_rate": 1.6184500824141395e-05,
+ "loss": 0.3685,
+ "step": 11345
+ },
+ {
+ "epoch": 30.915531335149865,
+ "grad_norm": 4.353820323944092,
+ "learning_rate": 1.618380731451866e-05,
+ "loss": 0.3501,
+ "step": 11346
+ },
+ {
+ "epoch": 30.91825613079019,
+ "grad_norm": 4.740364074707031,
+ "learning_rate": 1.618311375673663e-05,
+ "loss": 0.145,
+ "step": 11347
+ },
+ {
+ "epoch": 30.920980926430516,
+ "grad_norm": 4.447443008422852,
+ "learning_rate": 1.6182420150800712e-05,
+ "loss": 0.3236,
+ "step": 11348
+ },
+ {
+ "epoch": 30.923705722070846,
+ "grad_norm": 4.58844518661499,
+ "learning_rate": 1.6181726496716304e-05,
+ "loss": 0.3837,
+ "step": 11349
+ },
+ {
+ "epoch": 30.92643051771117,
+ "grad_norm": 7.504674911499023,
+ "learning_rate": 1.6181032794488807e-05,
+ "loss": 0.2802,
+ "step": 11350
+ },
+ {
+ "epoch": 30.929155313351497,
+ "grad_norm": 8.294772148132324,
+ "learning_rate": 1.6180339044123625e-05,
+ "loss": 0.2956,
+ "step": 11351
+ },
+ {
+ "epoch": 30.931880108991827,
+ "grad_norm": 6.02821159362793,
+ "learning_rate": 1.6179645245626165e-05,
+ "loss": 0.2144,
+ "step": 11352
+ },
+ {
+ "epoch": 30.934604904632153,
+ "grad_norm": 5.937182426452637,
+ "learning_rate": 1.617895139900182e-05,
+ "loss": 0.2849,
+ "step": 11353
+ },
+ {
+ "epoch": 30.93732970027248,
+ "grad_norm": 4.785285949707031,
+ "learning_rate": 1.6178257504256006e-05,
+ "loss": 0.2508,
+ "step": 11354
+ },
+ {
+ "epoch": 30.940054495912808,
+ "grad_norm": 5.239635467529297,
+ "learning_rate": 1.6177563561394123e-05,
+ "loss": 0.3521,
+ "step": 11355
+ },
+ {
+ "epoch": 30.942779291553133,
+ "grad_norm": 5.135106563568115,
+ "learning_rate": 1.617686957042157e-05,
+ "loss": 0.3251,
+ "step": 11356
+ },
+ {
+ "epoch": 30.94550408719346,
+ "grad_norm": 5.3350138664245605,
+ "learning_rate": 1.617617553134376e-05,
+ "loss": 0.3301,
+ "step": 11357
+ },
+ {
+ "epoch": 30.94822888283379,
+ "grad_norm": 5.487081050872803,
+ "learning_rate": 1.617548144416609e-05,
+ "loss": 0.2332,
+ "step": 11358
+ },
+ {
+ "epoch": 30.950953678474114,
+ "grad_norm": 7.403356552124023,
+ "learning_rate": 1.617478730889397e-05,
+ "loss": 0.2541,
+ "step": 11359
+ },
+ {
+ "epoch": 30.95367847411444,
+ "grad_norm": 5.930181980133057,
+ "learning_rate": 1.6174093125532803e-05,
+ "loss": 0.4568,
+ "step": 11360
+ },
+ {
+ "epoch": 30.95640326975477,
+ "grad_norm": 4.403568744659424,
+ "learning_rate": 1.6173398894088e-05,
+ "loss": 0.4112,
+ "step": 11361
+ },
+ {
+ "epoch": 30.959128065395095,
+ "grad_norm": 4.668586254119873,
+ "learning_rate": 1.617270461456496e-05,
+ "loss": 0.2518,
+ "step": 11362
+ },
+ {
+ "epoch": 30.96185286103542,
+ "grad_norm": 5.322455883026123,
+ "learning_rate": 1.6172010286969105e-05,
+ "loss": 0.2281,
+ "step": 11363
+ },
+ {
+ "epoch": 30.96457765667575,
+ "grad_norm": 4.989904880523682,
+ "learning_rate": 1.6171315911305824e-05,
+ "loss": 0.3988,
+ "step": 11364
+ },
+ {
+ "epoch": 30.967302452316076,
+ "grad_norm": 5.904867172241211,
+ "learning_rate": 1.6170621487580538e-05,
+ "loss": 0.2437,
+ "step": 11365
+ },
+ {
+ "epoch": 30.970027247956402,
+ "grad_norm": 5.335436820983887,
+ "learning_rate": 1.616992701579865e-05,
+ "loss": 0.3287,
+ "step": 11366
+ },
+ {
+ "epoch": 30.97275204359673,
+ "grad_norm": 5.206510543823242,
+ "learning_rate": 1.616923249596557e-05,
+ "loss": 0.3575,
+ "step": 11367
+ },
+ {
+ "epoch": 30.975476839237057,
+ "grad_norm": 4.946302890777588,
+ "learning_rate": 1.6168537928086703e-05,
+ "loss": 0.3457,
+ "step": 11368
+ },
+ {
+ "epoch": 30.978201634877383,
+ "grad_norm": 5.078906059265137,
+ "learning_rate": 1.616784331216746e-05,
+ "loss": 0.3448,
+ "step": 11369
+ },
+ {
+ "epoch": 30.980926430517712,
+ "grad_norm": 6.125652313232422,
+ "learning_rate": 1.6167148648213255e-05,
+ "loss": 0.4592,
+ "step": 11370
+ },
+ {
+ "epoch": 30.983651226158038,
+ "grad_norm": 4.638002395629883,
+ "learning_rate": 1.616645393622949e-05,
+ "loss": 0.2685,
+ "step": 11371
+ },
+ {
+ "epoch": 30.986376021798364,
+ "grad_norm": 6.201627731323242,
+ "learning_rate": 1.6165759176221588e-05,
+ "loss": 0.217,
+ "step": 11372
+ },
+ {
+ "epoch": 30.989100817438693,
+ "grad_norm": 5.399773597717285,
+ "learning_rate": 1.6165064368194946e-05,
+ "loss": 0.3414,
+ "step": 11373
+ },
+ {
+ "epoch": 30.99182561307902,
+ "grad_norm": 4.372704029083252,
+ "learning_rate": 1.6164369512154984e-05,
+ "loss": 0.2769,
+ "step": 11374
+ },
+ {
+ "epoch": 30.994550408719345,
+ "grad_norm": 5.3142805099487305,
+ "learning_rate": 1.616367460810711e-05,
+ "loss": 0.4076,
+ "step": 11375
+ },
+ {
+ "epoch": 30.997275204359674,
+ "grad_norm": 4.854094982147217,
+ "learning_rate": 1.6162979656056735e-05,
+ "loss": 0.3071,
+ "step": 11376
+ },
+ {
+ "epoch": 31.0,
+ "grad_norm": 5.472402095794678,
+ "learning_rate": 1.6162284656009276e-05,
+ "loss": 0.2083,
+ "step": 11377
+ },
+ {
+ "epoch": 31.002724795640326,
+ "grad_norm": 4.712527275085449,
+ "learning_rate": 1.6161589607970137e-05,
+ "loss": 0.3185,
+ "step": 11378
+ },
+ {
+ "epoch": 31.005449591280655,
+ "grad_norm": 4.062997341156006,
+ "learning_rate": 1.616089451194474e-05,
+ "loss": 0.1706,
+ "step": 11379
+ },
+ {
+ "epoch": 31.00817438692098,
+ "grad_norm": 4.075993061065674,
+ "learning_rate": 1.6160199367938494e-05,
+ "loss": 0.2155,
+ "step": 11380
+ },
+ {
+ "epoch": 31.010899182561307,
+ "grad_norm": 6.037343978881836,
+ "learning_rate": 1.6159504175956815e-05,
+ "loss": 0.2727,
+ "step": 11381
+ },
+ {
+ "epoch": 31.013623978201636,
+ "grad_norm": 3.7347922325134277,
+ "learning_rate": 1.6158808936005116e-05,
+ "loss": 0.2558,
+ "step": 11382
+ },
+ {
+ "epoch": 31.016348773841962,
+ "grad_norm": 5.555375576019287,
+ "learning_rate": 1.615811364808881e-05,
+ "loss": 0.2346,
+ "step": 11383
+ },
+ {
+ "epoch": 31.019073569482288,
+ "grad_norm": 4.666495323181152,
+ "learning_rate": 1.6157418312213313e-05,
+ "loss": 0.3551,
+ "step": 11384
+ },
+ {
+ "epoch": 31.021798365122617,
+ "grad_norm": 4.517202377319336,
+ "learning_rate": 1.6156722928384042e-05,
+ "loss": 0.2845,
+ "step": 11385
+ },
+ {
+ "epoch": 31.024523160762943,
+ "grad_norm": 5.630231857299805,
+ "learning_rate": 1.6156027496606408e-05,
+ "loss": 0.2446,
+ "step": 11386
+ },
+ {
+ "epoch": 31.02724795640327,
+ "grad_norm": 4.168909072875977,
+ "learning_rate": 1.615533201688583e-05,
+ "loss": 0.154,
+ "step": 11387
+ },
+ {
+ "epoch": 31.029972752043598,
+ "grad_norm": 3.717923641204834,
+ "learning_rate": 1.6154636489227726e-05,
+ "loss": 0.304,
+ "step": 11388
+ },
+ {
+ "epoch": 31.032697547683924,
+ "grad_norm": 19.28887176513672,
+ "learning_rate": 1.615394091363751e-05,
+ "loss": 0.2336,
+ "step": 11389
+ },
+ {
+ "epoch": 31.03542234332425,
+ "grad_norm": 3.7430031299591064,
+ "learning_rate": 1.6153245290120602e-05,
+ "loss": 0.1975,
+ "step": 11390
+ },
+ {
+ "epoch": 31.03814713896458,
+ "grad_norm": 4.259280681610107,
+ "learning_rate": 1.6152549618682418e-05,
+ "loss": 0.2369,
+ "step": 11391
+ },
+ {
+ "epoch": 31.040871934604905,
+ "grad_norm": 4.017198085784912,
+ "learning_rate": 1.6151853899328377e-05,
+ "loss": 0.2667,
+ "step": 11392
+ },
+ {
+ "epoch": 31.04359673024523,
+ "grad_norm": 4.986457347869873,
+ "learning_rate": 1.6151158132063892e-05,
+ "loss": 0.2699,
+ "step": 11393
+ },
+ {
+ "epoch": 31.04632152588556,
+ "grad_norm": 5.874536037445068,
+ "learning_rate": 1.615046231689439e-05,
+ "loss": 0.3324,
+ "step": 11394
+ },
+ {
+ "epoch": 31.049046321525886,
+ "grad_norm": 4.71151065826416,
+ "learning_rate": 1.614976645382528e-05,
+ "loss": 0.2176,
+ "step": 11395
+ },
+ {
+ "epoch": 31.05177111716621,
+ "grad_norm": 5.637543201446533,
+ "learning_rate": 1.614907054286199e-05,
+ "loss": 0.2582,
+ "step": 11396
+ },
+ {
+ "epoch": 31.05449591280654,
+ "grad_norm": 4.47944974899292,
+ "learning_rate": 1.6148374584009934e-05,
+ "loss": 0.1496,
+ "step": 11397
+ },
+ {
+ "epoch": 31.057220708446867,
+ "grad_norm": 4.1507439613342285,
+ "learning_rate": 1.6147678577274537e-05,
+ "loss": 0.2368,
+ "step": 11398
+ },
+ {
+ "epoch": 31.059945504087192,
+ "grad_norm": 7.070873260498047,
+ "learning_rate": 1.6146982522661216e-05,
+ "loss": 0.2764,
+ "step": 11399
+ },
+ {
+ "epoch": 31.06267029972752,
+ "grad_norm": 3.9157798290252686,
+ "learning_rate": 1.6146286420175393e-05,
+ "loss": 0.1767,
+ "step": 11400
+ },
+ {
+ "epoch": 31.065395095367847,
+ "grad_norm": 4.812366008758545,
+ "learning_rate": 1.6145590269822487e-05,
+ "loss": 0.1882,
+ "step": 11401
+ },
+ {
+ "epoch": 31.068119891008173,
+ "grad_norm": 4.851911544799805,
+ "learning_rate": 1.6144894071607927e-05,
+ "loss": 0.32,
+ "step": 11402
+ },
+ {
+ "epoch": 31.070844686648503,
+ "grad_norm": 4.932345390319824,
+ "learning_rate": 1.6144197825537127e-05,
+ "loss": 0.3204,
+ "step": 11403
+ },
+ {
+ "epoch": 31.07356948228883,
+ "grad_norm": 4.042575836181641,
+ "learning_rate": 1.614350153161551e-05,
+ "loss": 0.2766,
+ "step": 11404
+ },
+ {
+ "epoch": 31.076294277929154,
+ "grad_norm": 4.819994926452637,
+ "learning_rate": 1.6142805189848504e-05,
+ "loss": 0.278,
+ "step": 11405
+ },
+ {
+ "epoch": 31.079019073569484,
+ "grad_norm": 5.318707466125488,
+ "learning_rate": 1.6142108800241527e-05,
+ "loss": 0.24,
+ "step": 11406
+ },
+ {
+ "epoch": 31.08174386920981,
+ "grad_norm": 4.22999382019043,
+ "learning_rate": 1.6141412362800007e-05,
+ "loss": 0.1485,
+ "step": 11407
+ },
+ {
+ "epoch": 31.084468664850135,
+ "grad_norm": 3.905897378921509,
+ "learning_rate": 1.6140715877529363e-05,
+ "loss": 0.1438,
+ "step": 11408
+ },
+ {
+ "epoch": 31.087193460490465,
+ "grad_norm": 3.5796613693237305,
+ "learning_rate": 1.614001934443502e-05,
+ "loss": 0.1509,
+ "step": 11409
+ },
+ {
+ "epoch": 31.08991825613079,
+ "grad_norm": 4.831905364990234,
+ "learning_rate": 1.613932276352241e-05,
+ "loss": 0.3198,
+ "step": 11410
+ },
+ {
+ "epoch": 31.092643051771116,
+ "grad_norm": 4.570455551147461,
+ "learning_rate": 1.6138626134796945e-05,
+ "loss": 0.2338,
+ "step": 11411
+ },
+ {
+ "epoch": 31.095367847411445,
+ "grad_norm": 4.029191017150879,
+ "learning_rate": 1.613792945826406e-05,
+ "loss": 0.3222,
+ "step": 11412
+ },
+ {
+ "epoch": 31.09809264305177,
+ "grad_norm": 4.348748683929443,
+ "learning_rate": 1.6137232733929176e-05,
+ "loss": 0.1839,
+ "step": 11413
+ },
+ {
+ "epoch": 31.100817438692097,
+ "grad_norm": 5.662988662719727,
+ "learning_rate": 1.6136535961797722e-05,
+ "loss": 0.4391,
+ "step": 11414
+ },
+ {
+ "epoch": 31.103542234332426,
+ "grad_norm": 3.533381700515747,
+ "learning_rate": 1.6135839141875127e-05,
+ "loss": 0.1579,
+ "step": 11415
+ },
+ {
+ "epoch": 31.106267029972752,
+ "grad_norm": 4.816019535064697,
+ "learning_rate": 1.613514227416681e-05,
+ "loss": 0.1701,
+ "step": 11416
+ },
+ {
+ "epoch": 31.108991825613078,
+ "grad_norm": 6.02421760559082,
+ "learning_rate": 1.6134445358678205e-05,
+ "loss": 0.1724,
+ "step": 11417
+ },
+ {
+ "epoch": 31.111716621253407,
+ "grad_norm": 4.71406888961792,
+ "learning_rate": 1.613374839541474e-05,
+ "loss": 0.2155,
+ "step": 11418
+ },
+ {
+ "epoch": 31.114441416893733,
+ "grad_norm": 4.834839344024658,
+ "learning_rate": 1.6133051384381833e-05,
+ "loss": 0.165,
+ "step": 11419
+ },
+ {
+ "epoch": 31.11716621253406,
+ "grad_norm": 5.023985862731934,
+ "learning_rate": 1.613235432558492e-05,
+ "loss": 0.1798,
+ "step": 11420
+ },
+ {
+ "epoch": 31.11989100817439,
+ "grad_norm": 7.6469950675964355,
+ "learning_rate": 1.6131657219029428e-05,
+ "loss": 0.3309,
+ "step": 11421
+ },
+ {
+ "epoch": 31.122615803814714,
+ "grad_norm": 5.366358757019043,
+ "learning_rate": 1.613096006472079e-05,
+ "loss": 0.2157,
+ "step": 11422
+ },
+ {
+ "epoch": 31.12534059945504,
+ "grad_norm": 4.35495662689209,
+ "learning_rate": 1.6130262862664432e-05,
+ "loss": 0.1895,
+ "step": 11423
+ },
+ {
+ "epoch": 31.12806539509537,
+ "grad_norm": 4.938991546630859,
+ "learning_rate": 1.6129565612865786e-05,
+ "loss": 0.3177,
+ "step": 11424
+ },
+ {
+ "epoch": 31.130790190735695,
+ "grad_norm": 4.36767578125,
+ "learning_rate": 1.6128868315330278e-05,
+ "loss": 0.2235,
+ "step": 11425
+ },
+ {
+ "epoch": 31.13351498637602,
+ "grad_norm": 4.894557952880859,
+ "learning_rate": 1.612817097006334e-05,
+ "loss": 0.3121,
+ "step": 11426
+ },
+ {
+ "epoch": 31.13623978201635,
+ "grad_norm": 4.4109392166137695,
+ "learning_rate": 1.6127473577070408e-05,
+ "loss": 0.2777,
+ "step": 11427
+ },
+ {
+ "epoch": 31.138964577656676,
+ "grad_norm": 4.843830108642578,
+ "learning_rate": 1.6126776136356906e-05,
+ "loss": 0.1909,
+ "step": 11428
+ },
+ {
+ "epoch": 31.141689373297,
+ "grad_norm": 4.859058856964111,
+ "learning_rate": 1.6126078647928266e-05,
+ "loss": 0.1955,
+ "step": 11429
+ },
+ {
+ "epoch": 31.14441416893733,
+ "grad_norm": 3.826395034790039,
+ "learning_rate": 1.6125381111789926e-05,
+ "loss": 0.212,
+ "step": 11430
+ },
+ {
+ "epoch": 31.147138964577657,
+ "grad_norm": 3.964515447616577,
+ "learning_rate": 1.6124683527947317e-05,
+ "loss": 0.2245,
+ "step": 11431
+ },
+ {
+ "epoch": 31.149863760217983,
+ "grad_norm": 4.04124116897583,
+ "learning_rate": 1.6123985896405863e-05,
+ "loss": 0.2358,
+ "step": 11432
+ },
+ {
+ "epoch": 31.152588555858312,
+ "grad_norm": 4.4868645668029785,
+ "learning_rate": 1.612328821717101e-05,
+ "loss": 0.3045,
+ "step": 11433
+ },
+ {
+ "epoch": 31.155313351498638,
+ "grad_norm": 4.742684841156006,
+ "learning_rate": 1.6122590490248185e-05,
+ "loss": 0.4073,
+ "step": 11434
+ },
+ {
+ "epoch": 31.158038147138964,
+ "grad_norm": 3.9538116455078125,
+ "learning_rate": 1.612189271564282e-05,
+ "loss": 0.2529,
+ "step": 11435
+ },
+ {
+ "epoch": 31.160762942779293,
+ "grad_norm": 4.0842604637146,
+ "learning_rate": 1.612119489336035e-05,
+ "loss": 0.3056,
+ "step": 11436
+ },
+ {
+ "epoch": 31.16348773841962,
+ "grad_norm": 3.946467161178589,
+ "learning_rate": 1.6120497023406215e-05,
+ "loss": 0.5802,
+ "step": 11437
+ },
+ {
+ "epoch": 31.166212534059945,
+ "grad_norm": 4.255927085876465,
+ "learning_rate": 1.6119799105785845e-05,
+ "loss": 0.1664,
+ "step": 11438
+ },
+ {
+ "epoch": 31.168937329700274,
+ "grad_norm": 5.011704444885254,
+ "learning_rate": 1.6119101140504676e-05,
+ "loss": 0.1825,
+ "step": 11439
+ },
+ {
+ "epoch": 31.1716621253406,
+ "grad_norm": 5.269890308380127,
+ "learning_rate": 1.6118403127568142e-05,
+ "loss": 0.3388,
+ "step": 11440
+ },
+ {
+ "epoch": 31.174386920980925,
+ "grad_norm": 3.7455332279205322,
+ "learning_rate": 1.6117705066981685e-05,
+ "loss": 0.2822,
+ "step": 11441
+ },
+ {
+ "epoch": 31.177111716621255,
+ "grad_norm": 4.504450798034668,
+ "learning_rate": 1.611700695875074e-05,
+ "loss": 0.2075,
+ "step": 11442
+ },
+ {
+ "epoch": 31.17983651226158,
+ "grad_norm": 4.467066764831543,
+ "learning_rate": 1.6116308802880737e-05,
+ "loss": 0.3093,
+ "step": 11443
+ },
+ {
+ "epoch": 31.182561307901906,
+ "grad_norm": 4.1285576820373535,
+ "learning_rate": 1.6115610599377118e-05,
+ "loss": 0.3925,
+ "step": 11444
+ },
+ {
+ "epoch": 31.185286103542236,
+ "grad_norm": 4.307610034942627,
+ "learning_rate": 1.6114912348245324e-05,
+ "loss": 0.2209,
+ "step": 11445
+ },
+ {
+ "epoch": 31.18801089918256,
+ "grad_norm": 4.79456090927124,
+ "learning_rate": 1.6114214049490787e-05,
+ "loss": 0.2282,
+ "step": 11446
+ },
+ {
+ "epoch": 31.190735694822887,
+ "grad_norm": 6.074712753295898,
+ "learning_rate": 1.6113515703118948e-05,
+ "loss": 0.2699,
+ "step": 11447
+ },
+ {
+ "epoch": 31.193460490463217,
+ "grad_norm": 4.74254846572876,
+ "learning_rate": 1.6112817309135244e-05,
+ "loss": 0.2828,
+ "step": 11448
+ },
+ {
+ "epoch": 31.196185286103542,
+ "grad_norm": 3.646820068359375,
+ "learning_rate": 1.6112118867545118e-05,
+ "loss": 0.4224,
+ "step": 11449
+ },
+ {
+ "epoch": 31.19891008174387,
+ "grad_norm": 4.100376129150391,
+ "learning_rate": 1.6111420378354003e-05,
+ "loss": 0.198,
+ "step": 11450
+ },
+ {
+ "epoch": 31.201634877384198,
+ "grad_norm": 10.02737045288086,
+ "learning_rate": 1.611072184156735e-05,
+ "loss": 0.4276,
+ "step": 11451
+ },
+ {
+ "epoch": 31.204359673024523,
+ "grad_norm": 3.8945183753967285,
+ "learning_rate": 1.6110023257190587e-05,
+ "loss": 0.2895,
+ "step": 11452
+ },
+ {
+ "epoch": 31.20708446866485,
+ "grad_norm": 4.325787544250488,
+ "learning_rate": 1.6109324625229158e-05,
+ "loss": 0.2627,
+ "step": 11453
+ },
+ {
+ "epoch": 31.20980926430518,
+ "grad_norm": 3.900214910507202,
+ "learning_rate": 1.610862594568851e-05,
+ "loss": 0.1796,
+ "step": 11454
+ },
+ {
+ "epoch": 31.212534059945504,
+ "grad_norm": 4.219604969024658,
+ "learning_rate": 1.6107927218574076e-05,
+ "loss": 0.1748,
+ "step": 11455
+ },
+ {
+ "epoch": 31.21525885558583,
+ "grad_norm": 7.001033782958984,
+ "learning_rate": 1.6107228443891303e-05,
+ "loss": 0.2332,
+ "step": 11456
+ },
+ {
+ "epoch": 31.21798365122616,
+ "grad_norm": 4.007717132568359,
+ "learning_rate": 1.610652962164563e-05,
+ "loss": 0.2358,
+ "step": 11457
+ },
+ {
+ "epoch": 31.220708446866485,
+ "grad_norm": 4.408379077911377,
+ "learning_rate": 1.6105830751842503e-05,
+ "loss": 0.2043,
+ "step": 11458
+ },
+ {
+ "epoch": 31.22343324250681,
+ "grad_norm": 3.9459965229034424,
+ "learning_rate": 1.6105131834487364e-05,
+ "loss": 0.198,
+ "step": 11459
+ },
+ {
+ "epoch": 31.22615803814714,
+ "grad_norm": 5.934104919433594,
+ "learning_rate": 1.6104432869585652e-05,
+ "loss": 0.3366,
+ "step": 11460
+ },
+ {
+ "epoch": 31.228882833787466,
+ "grad_norm": 5.301226615905762,
+ "learning_rate": 1.610373385714281e-05,
+ "loss": 0.2834,
+ "step": 11461
+ },
+ {
+ "epoch": 31.231607629427792,
+ "grad_norm": 4.3575310707092285,
+ "learning_rate": 1.6103034797164294e-05,
+ "loss": 0.17,
+ "step": 11462
+ },
+ {
+ "epoch": 31.23433242506812,
+ "grad_norm": 3.7450308799743652,
+ "learning_rate": 1.6102335689655533e-05,
+ "loss": 0.2277,
+ "step": 11463
+ },
+ {
+ "epoch": 31.237057220708447,
+ "grad_norm": 4.563078880310059,
+ "learning_rate": 1.610163653462198e-05,
+ "loss": 0.3363,
+ "step": 11464
+ },
+ {
+ "epoch": 31.239782016348773,
+ "grad_norm": 6.064635276794434,
+ "learning_rate": 1.6100937332069074e-05,
+ "loss": 0.2309,
+ "step": 11465
+ },
+ {
+ "epoch": 31.242506811989102,
+ "grad_norm": 4.769646644592285,
+ "learning_rate": 1.610023808200227e-05,
+ "loss": 0.3802,
+ "step": 11466
+ },
+ {
+ "epoch": 31.245231607629428,
+ "grad_norm": 4.042790412902832,
+ "learning_rate": 1.6099538784427005e-05,
+ "loss": 0.2391,
+ "step": 11467
+ },
+ {
+ "epoch": 31.247956403269754,
+ "grad_norm": 3.88063907623291,
+ "learning_rate": 1.609883943934873e-05,
+ "loss": 0.2753,
+ "step": 11468
+ },
+ {
+ "epoch": 31.250681198910083,
+ "grad_norm": 4.29428768157959,
+ "learning_rate": 1.609814004677289e-05,
+ "loss": 0.3249,
+ "step": 11469
+ },
+ {
+ "epoch": 31.25340599455041,
+ "grad_norm": 3.790071964263916,
+ "learning_rate": 1.6097440606704928e-05,
+ "loss": 0.2963,
+ "step": 11470
+ },
+ {
+ "epoch": 31.256130790190735,
+ "grad_norm": 4.7611823081970215,
+ "learning_rate": 1.60967411191503e-05,
+ "loss": 0.2596,
+ "step": 11471
+ },
+ {
+ "epoch": 31.258855585831064,
+ "grad_norm": 5.00996732711792,
+ "learning_rate": 1.6096041584114444e-05,
+ "loss": 0.2829,
+ "step": 11472
+ },
+ {
+ "epoch": 31.26158038147139,
+ "grad_norm": 4.826964855194092,
+ "learning_rate": 1.6095342001602813e-05,
+ "loss": 0.3364,
+ "step": 11473
+ },
+ {
+ "epoch": 31.264305177111716,
+ "grad_norm": 7.619364261627197,
+ "learning_rate": 1.6094642371620857e-05,
+ "loss": 0.161,
+ "step": 11474
+ },
+ {
+ "epoch": 31.267029972752045,
+ "grad_norm": 4.745420455932617,
+ "learning_rate": 1.6093942694174024e-05,
+ "loss": 0.2063,
+ "step": 11475
+ },
+ {
+ "epoch": 31.26975476839237,
+ "grad_norm": 4.049158573150635,
+ "learning_rate": 1.609324296926776e-05,
+ "loss": 0.2183,
+ "step": 11476
+ },
+ {
+ "epoch": 31.272479564032697,
+ "grad_norm": 4.864970684051514,
+ "learning_rate": 1.609254319690751e-05,
+ "loss": 0.2984,
+ "step": 11477
+ },
+ {
+ "epoch": 31.275204359673026,
+ "grad_norm": 4.308829307556152,
+ "learning_rate": 1.6091843377098734e-05,
+ "loss": 0.3523,
+ "step": 11478
+ },
+ {
+ "epoch": 31.277929155313352,
+ "grad_norm": 5.484410285949707,
+ "learning_rate": 1.609114350984688e-05,
+ "loss": 0.2946,
+ "step": 11479
+ },
+ {
+ "epoch": 31.280653950953678,
+ "grad_norm": 4.875092506408691,
+ "learning_rate": 1.6090443595157396e-05,
+ "loss": 0.4252,
+ "step": 11480
+ },
+ {
+ "epoch": 31.283378746594007,
+ "grad_norm": 4.515564441680908,
+ "learning_rate": 1.6089743633035732e-05,
+ "loss": 0.1919,
+ "step": 11481
+ },
+ {
+ "epoch": 31.286103542234333,
+ "grad_norm": 4.89269495010376,
+ "learning_rate": 1.6089043623487338e-05,
+ "loss": 0.295,
+ "step": 11482
+ },
+ {
+ "epoch": 31.28882833787466,
+ "grad_norm": 4.44584846496582,
+ "learning_rate": 1.608834356651767e-05,
+ "loss": 0.2442,
+ "step": 11483
+ },
+ {
+ "epoch": 31.291553133514988,
+ "grad_norm": 5.614002227783203,
+ "learning_rate": 1.608764346213218e-05,
+ "loss": 0.3117,
+ "step": 11484
+ },
+ {
+ "epoch": 31.294277929155314,
+ "grad_norm": 3.893573045730591,
+ "learning_rate": 1.6086943310336315e-05,
+ "loss": 0.1607,
+ "step": 11485
+ },
+ {
+ "epoch": 31.29700272479564,
+ "grad_norm": 4.50229024887085,
+ "learning_rate": 1.6086243111135537e-05,
+ "loss": 0.1688,
+ "step": 11486
+ },
+ {
+ "epoch": 31.29972752043597,
+ "grad_norm": 7.000305652618408,
+ "learning_rate": 1.608554286453529e-05,
+ "loss": 0.3512,
+ "step": 11487
+ },
+ {
+ "epoch": 31.302452316076295,
+ "grad_norm": 4.555546283721924,
+ "learning_rate": 1.608484257054103e-05,
+ "loss": 0.4617,
+ "step": 11488
+ },
+ {
+ "epoch": 31.30517711171662,
+ "grad_norm": 3.9883921146392822,
+ "learning_rate": 1.6084142229158214e-05,
+ "loss": 0.2874,
+ "step": 11489
+ },
+ {
+ "epoch": 31.30790190735695,
+ "grad_norm": 3.378063678741455,
+ "learning_rate": 1.608344184039229e-05,
+ "loss": 0.3012,
+ "step": 11490
+ },
+ {
+ "epoch": 31.310626702997276,
+ "grad_norm": 3.895366907119751,
+ "learning_rate": 1.6082741404248724e-05,
+ "loss": 0.2872,
+ "step": 11491
+ },
+ {
+ "epoch": 31.3133514986376,
+ "grad_norm": 4.337034702301025,
+ "learning_rate": 1.608204092073296e-05,
+ "loss": 0.1324,
+ "step": 11492
+ },
+ {
+ "epoch": 31.31607629427793,
+ "grad_norm": 5.544063091278076,
+ "learning_rate": 1.608134038985046e-05,
+ "loss": 0.2427,
+ "step": 11493
+ },
+ {
+ "epoch": 31.318801089918257,
+ "grad_norm": 4.2934064865112305,
+ "learning_rate": 1.608063981160667e-05,
+ "loss": 0.2644,
+ "step": 11494
+ },
+ {
+ "epoch": 31.321525885558582,
+ "grad_norm": 4.784576892852783,
+ "learning_rate": 1.6079939186007058e-05,
+ "loss": 0.3129,
+ "step": 11495
+ },
+ {
+ "epoch": 31.32425068119891,
+ "grad_norm": 5.337213039398193,
+ "learning_rate": 1.6079238513057073e-05,
+ "loss": 0.295,
+ "step": 11496
+ },
+ {
+ "epoch": 31.326975476839237,
+ "grad_norm": 4.322379112243652,
+ "learning_rate": 1.6078537792762174e-05,
+ "loss": 0.2125,
+ "step": 11497
+ },
+ {
+ "epoch": 31.329700272479563,
+ "grad_norm": 4.01151180267334,
+ "learning_rate": 1.6077837025127817e-05,
+ "loss": 0.3427,
+ "step": 11498
+ },
+ {
+ "epoch": 31.332425068119893,
+ "grad_norm": 4.300271511077881,
+ "learning_rate": 1.6077136210159466e-05,
+ "loss": 0.2953,
+ "step": 11499
+ },
+ {
+ "epoch": 31.33514986376022,
+ "grad_norm": 4.4044718742370605,
+ "learning_rate": 1.607643534786257e-05,
+ "loss": 0.2679,
+ "step": 11500
+ },
+ {
+ "epoch": 31.337874659400544,
+ "grad_norm": 4.419496536254883,
+ "learning_rate": 1.607573443824259e-05,
+ "loss": 0.3378,
+ "step": 11501
+ },
+ {
+ "epoch": 31.340599455040874,
+ "grad_norm": 4.303692817687988,
+ "learning_rate": 1.6075033481304993e-05,
+ "loss": 0.1945,
+ "step": 11502
+ },
+ {
+ "epoch": 31.3433242506812,
+ "grad_norm": 5.003098487854004,
+ "learning_rate": 1.6074332477055224e-05,
+ "loss": 0.2236,
+ "step": 11503
+ },
+ {
+ "epoch": 31.346049046321525,
+ "grad_norm": 3.669764757156372,
+ "learning_rate": 1.607363142549875e-05,
+ "loss": 0.2431,
+ "step": 11504
+ },
+ {
+ "epoch": 31.348773841961854,
+ "grad_norm": 4.343540191650391,
+ "learning_rate": 1.607293032664103e-05,
+ "loss": 0.3119,
+ "step": 11505
+ },
+ {
+ "epoch": 31.35149863760218,
+ "grad_norm": 4.798116207122803,
+ "learning_rate": 1.6072229180487523e-05,
+ "loss": 0.3385,
+ "step": 11506
+ },
+ {
+ "epoch": 31.354223433242506,
+ "grad_norm": 3.9332942962646484,
+ "learning_rate": 1.6071527987043693e-05,
+ "loss": 0.3155,
+ "step": 11507
+ },
+ {
+ "epoch": 31.356948228882835,
+ "grad_norm": 5.111228942871094,
+ "learning_rate": 1.6070826746314997e-05,
+ "loss": 0.2511,
+ "step": 11508
+ },
+ {
+ "epoch": 31.35967302452316,
+ "grad_norm": 4.554476261138916,
+ "learning_rate": 1.60701254583069e-05,
+ "loss": 0.3461,
+ "step": 11509
+ },
+ {
+ "epoch": 31.362397820163487,
+ "grad_norm": 5.083421230316162,
+ "learning_rate": 1.606942412302486e-05,
+ "loss": 0.2374,
+ "step": 11510
+ },
+ {
+ "epoch": 31.365122615803816,
+ "grad_norm": 3.815915822982788,
+ "learning_rate": 1.606872274047434e-05,
+ "loss": 0.286,
+ "step": 11511
+ },
+ {
+ "epoch": 31.367847411444142,
+ "grad_norm": 4.38032865524292,
+ "learning_rate": 1.6068021310660806e-05,
+ "loss": 0.2727,
+ "step": 11512
+ },
+ {
+ "epoch": 31.370572207084468,
+ "grad_norm": 4.268787860870361,
+ "learning_rate": 1.606731983358971e-05,
+ "loss": 0.1777,
+ "step": 11513
+ },
+ {
+ "epoch": 31.373297002724797,
+ "grad_norm": 5.121934413909912,
+ "learning_rate": 1.6066618309266527e-05,
+ "loss": 0.2123,
+ "step": 11514
+ },
+ {
+ "epoch": 31.376021798365123,
+ "grad_norm": 4.086214542388916,
+ "learning_rate": 1.6065916737696713e-05,
+ "loss": 0.3876,
+ "step": 11515
+ },
+ {
+ "epoch": 31.37874659400545,
+ "grad_norm": 5.671654224395752,
+ "learning_rate": 1.6065215118885737e-05,
+ "loss": 0.2936,
+ "step": 11516
+ },
+ {
+ "epoch": 31.381471389645778,
+ "grad_norm": 4.368946552276611,
+ "learning_rate": 1.6064513452839064e-05,
+ "loss": 0.1999,
+ "step": 11517
+ },
+ {
+ "epoch": 31.384196185286104,
+ "grad_norm": 5.032603740692139,
+ "learning_rate": 1.6063811739562153e-05,
+ "loss": 0.2476,
+ "step": 11518
+ },
+ {
+ "epoch": 31.38692098092643,
+ "grad_norm": 4.454460144042969,
+ "learning_rate": 1.606310997906047e-05,
+ "loss": 0.3927,
+ "step": 11519
+ },
+ {
+ "epoch": 31.38964577656676,
+ "grad_norm": 4.474380016326904,
+ "learning_rate": 1.606240817133948e-05,
+ "loss": 0.3343,
+ "step": 11520
+ },
+ {
+ "epoch": 31.392370572207085,
+ "grad_norm": 6.054551601409912,
+ "learning_rate": 1.6061706316404653e-05,
+ "loss": 0.2656,
+ "step": 11521
+ },
+ {
+ "epoch": 31.39509536784741,
+ "grad_norm": 4.331265449523926,
+ "learning_rate": 1.606100441426145e-05,
+ "loss": 0.2658,
+ "step": 11522
+ },
+ {
+ "epoch": 31.39782016348774,
+ "grad_norm": 4.406498908996582,
+ "learning_rate": 1.6060302464915346e-05,
+ "loss": 0.1566,
+ "step": 11523
+ },
+ {
+ "epoch": 31.400544959128066,
+ "grad_norm": 4.551695346832275,
+ "learning_rate": 1.6059600468371792e-05,
+ "loss": 0.2177,
+ "step": 11524
+ },
+ {
+ "epoch": 31.40326975476839,
+ "grad_norm": 3.914877414703369,
+ "learning_rate": 1.605889842463627e-05,
+ "loss": 0.2104,
+ "step": 11525
+ },
+ {
+ "epoch": 31.40599455040872,
+ "grad_norm": 4.125168800354004,
+ "learning_rate": 1.605819633371424e-05,
+ "loss": 0.1915,
+ "step": 11526
+ },
+ {
+ "epoch": 31.408719346049047,
+ "grad_norm": 5.768260478973389,
+ "learning_rate": 1.6057494195611177e-05,
+ "loss": 0.2021,
+ "step": 11527
+ },
+ {
+ "epoch": 31.411444141689373,
+ "grad_norm": 4.384498119354248,
+ "learning_rate": 1.605679201033254e-05,
+ "loss": 0.3053,
+ "step": 11528
+ },
+ {
+ "epoch": 31.414168937329702,
+ "grad_norm": 4.151440143585205,
+ "learning_rate": 1.6056089777883803e-05,
+ "loss": 0.2688,
+ "step": 11529
+ },
+ {
+ "epoch": 31.416893732970028,
+ "grad_norm": 4.120724201202393,
+ "learning_rate": 1.6055387498270435e-05,
+ "loss": 0.2949,
+ "step": 11530
+ },
+ {
+ "epoch": 31.419618528610354,
+ "grad_norm": 4.627317428588867,
+ "learning_rate": 1.60546851714979e-05,
+ "loss": 0.2314,
+ "step": 11531
+ },
+ {
+ "epoch": 31.422343324250683,
+ "grad_norm": 5.494587421417236,
+ "learning_rate": 1.6053982797571675e-05,
+ "loss": 0.2159,
+ "step": 11532
+ },
+ {
+ "epoch": 31.42506811989101,
+ "grad_norm": 3.9748806953430176,
+ "learning_rate": 1.6053280376497224e-05,
+ "loss": 0.1582,
+ "step": 11533
+ },
+ {
+ "epoch": 31.427792915531334,
+ "grad_norm": 4.441152572631836,
+ "learning_rate": 1.6052577908280022e-05,
+ "loss": 0.1968,
+ "step": 11534
+ },
+ {
+ "epoch": 31.430517711171664,
+ "grad_norm": 5.845150470733643,
+ "learning_rate": 1.605187539292554e-05,
+ "loss": 0.3256,
+ "step": 11535
+ },
+ {
+ "epoch": 31.43324250681199,
+ "grad_norm": 4.960531234741211,
+ "learning_rate": 1.6051172830439243e-05,
+ "loss": 0.2299,
+ "step": 11536
+ },
+ {
+ "epoch": 31.435967302452315,
+ "grad_norm": 5.931319713592529,
+ "learning_rate": 1.605047022082661e-05,
+ "loss": 0.2437,
+ "step": 11537
+ },
+ {
+ "epoch": 31.438692098092645,
+ "grad_norm": 5.465138912200928,
+ "learning_rate": 1.6049767564093107e-05,
+ "loss": 0.2257,
+ "step": 11538
+ },
+ {
+ "epoch": 31.44141689373297,
+ "grad_norm": 4.643120765686035,
+ "learning_rate": 1.604906486024421e-05,
+ "loss": 0.2901,
+ "step": 11539
+ },
+ {
+ "epoch": 31.444141689373296,
+ "grad_norm": 4.661689281463623,
+ "learning_rate": 1.6048362109285393e-05,
+ "loss": 0.2005,
+ "step": 11540
+ },
+ {
+ "epoch": 31.446866485013626,
+ "grad_norm": 4.769650459289551,
+ "learning_rate": 1.6047659311222124e-05,
+ "loss": 0.2271,
+ "step": 11541
+ },
+ {
+ "epoch": 31.44959128065395,
+ "grad_norm": 4.115447998046875,
+ "learning_rate": 1.604695646605988e-05,
+ "loss": 0.1885,
+ "step": 11542
+ },
+ {
+ "epoch": 31.452316076294277,
+ "grad_norm": 4.910819053649902,
+ "learning_rate": 1.6046253573804134e-05,
+ "loss": 0.2144,
+ "step": 11543
+ },
+ {
+ "epoch": 31.455040871934607,
+ "grad_norm": 3.9851105213165283,
+ "learning_rate": 1.6045550634460357e-05,
+ "loss": 0.2347,
+ "step": 11544
+ },
+ {
+ "epoch": 31.457765667574932,
+ "grad_norm": 4.107816219329834,
+ "learning_rate": 1.6044847648034026e-05,
+ "loss": 0.1886,
+ "step": 11545
+ },
+ {
+ "epoch": 31.460490463215258,
+ "grad_norm": 4.628753662109375,
+ "learning_rate": 1.604414461453062e-05,
+ "loss": 0.2072,
+ "step": 11546
+ },
+ {
+ "epoch": 31.463215258855588,
+ "grad_norm": 5.137694358825684,
+ "learning_rate": 1.604344153395561e-05,
+ "loss": 0.2443,
+ "step": 11547
+ },
+ {
+ "epoch": 31.465940054495913,
+ "grad_norm": 4.381475448608398,
+ "learning_rate": 1.604273840631447e-05,
+ "loss": 0.3046,
+ "step": 11548
+ },
+ {
+ "epoch": 31.46866485013624,
+ "grad_norm": 4.657466888427734,
+ "learning_rate": 1.6042035231612678e-05,
+ "loss": 0.2663,
+ "step": 11549
+ },
+ {
+ "epoch": 31.47138964577657,
+ "grad_norm": 4.931101322174072,
+ "learning_rate": 1.604133200985571e-05,
+ "loss": 0.3563,
+ "step": 11550
+ },
+ {
+ "epoch": 31.474114441416894,
+ "grad_norm": 7.437098503112793,
+ "learning_rate": 1.6040628741049044e-05,
+ "loss": 0.2439,
+ "step": 11551
+ },
+ {
+ "epoch": 31.47683923705722,
+ "grad_norm": 4.643455982208252,
+ "learning_rate": 1.6039925425198156e-05,
+ "loss": 0.2637,
+ "step": 11552
+ },
+ {
+ "epoch": 31.479564032697546,
+ "grad_norm": 3.6204020977020264,
+ "learning_rate": 1.603922206230852e-05,
+ "loss": 0.1902,
+ "step": 11553
+ },
+ {
+ "epoch": 31.482288828337875,
+ "grad_norm": 3.540100574493408,
+ "learning_rate": 1.603851865238562e-05,
+ "loss": 0.2625,
+ "step": 11554
+ },
+ {
+ "epoch": 31.4850136239782,
+ "grad_norm": 3.4193572998046875,
+ "learning_rate": 1.6037815195434928e-05,
+ "loss": 0.2042,
+ "step": 11555
+ },
+ {
+ "epoch": 31.48773841961853,
+ "grad_norm": 3.7349586486816406,
+ "learning_rate": 1.6037111691461932e-05,
+ "loss": 0.1494,
+ "step": 11556
+ },
+ {
+ "epoch": 31.490463215258856,
+ "grad_norm": 4.672253608703613,
+ "learning_rate": 1.6036408140472096e-05,
+ "loss": 0.2073,
+ "step": 11557
+ },
+ {
+ "epoch": 31.493188010899182,
+ "grad_norm": 4.930914878845215,
+ "learning_rate": 1.6035704542470915e-05,
+ "loss": 0.2233,
+ "step": 11558
+ },
+ {
+ "epoch": 31.495912806539508,
+ "grad_norm": 3.591963768005371,
+ "learning_rate": 1.6035000897463857e-05,
+ "loss": 0.2394,
+ "step": 11559
+ },
+ {
+ "epoch": 31.498637602179837,
+ "grad_norm": 4.462998390197754,
+ "learning_rate": 1.603429720545641e-05,
+ "loss": 0.3595,
+ "step": 11560
+ },
+ {
+ "epoch": 31.501362397820163,
+ "grad_norm": 4.174803733825684,
+ "learning_rate": 1.6033593466454043e-05,
+ "loss": 0.1728,
+ "step": 11561
+ },
+ {
+ "epoch": 31.504087193460492,
+ "grad_norm": 4.960818767547607,
+ "learning_rate": 1.603288968046225e-05,
+ "loss": 0.2405,
+ "step": 11562
+ },
+ {
+ "epoch": 31.506811989100818,
+ "grad_norm": 3.9373693466186523,
+ "learning_rate": 1.6032185847486504e-05,
+ "loss": 0.2136,
+ "step": 11563
+ },
+ {
+ "epoch": 31.509536784741144,
+ "grad_norm": 4.842967510223389,
+ "learning_rate": 1.6031481967532293e-05,
+ "loss": 0.3646,
+ "step": 11564
+ },
+ {
+ "epoch": 31.51226158038147,
+ "grad_norm": 4.269507884979248,
+ "learning_rate": 1.603077804060509e-05,
+ "loss": 0.2903,
+ "step": 11565
+ },
+ {
+ "epoch": 31.5149863760218,
+ "grad_norm": 5.098799705505371,
+ "learning_rate": 1.6030074066710387e-05,
+ "loss": 0.2866,
+ "step": 11566
+ },
+ {
+ "epoch": 31.517711171662125,
+ "grad_norm": 5.630182266235352,
+ "learning_rate": 1.6029370045853655e-05,
+ "loss": 0.2358,
+ "step": 11567
+ },
+ {
+ "epoch": 31.520435967302454,
+ "grad_norm": 4.64793062210083,
+ "learning_rate": 1.6028665978040388e-05,
+ "loss": 0.2192,
+ "step": 11568
+ },
+ {
+ "epoch": 31.52316076294278,
+ "grad_norm": 4.718940734863281,
+ "learning_rate": 1.602796186327606e-05,
+ "loss": 0.1987,
+ "step": 11569
+ },
+ {
+ "epoch": 31.525885558583106,
+ "grad_norm": 3.6881494522094727,
+ "learning_rate": 1.6027257701566162e-05,
+ "loss": 0.2498,
+ "step": 11570
+ },
+ {
+ "epoch": 31.52861035422343,
+ "grad_norm": 4.366286277770996,
+ "learning_rate": 1.6026553492916175e-05,
+ "loss": 0.1834,
+ "step": 11571
+ },
+ {
+ "epoch": 31.53133514986376,
+ "grad_norm": 5.800507545471191,
+ "learning_rate": 1.6025849237331583e-05,
+ "loss": 0.3428,
+ "step": 11572
+ },
+ {
+ "epoch": 31.534059945504087,
+ "grad_norm": 4.531710624694824,
+ "learning_rate": 1.602514493481787e-05,
+ "loss": 0.1364,
+ "step": 11573
+ },
+ {
+ "epoch": 31.536784741144416,
+ "grad_norm": 4.452208995819092,
+ "learning_rate": 1.6024440585380524e-05,
+ "loss": 0.2816,
+ "step": 11574
+ },
+ {
+ "epoch": 31.539509536784742,
+ "grad_norm": 4.73623514175415,
+ "learning_rate": 1.6023736189025027e-05,
+ "loss": 0.2255,
+ "step": 11575
+ },
+ {
+ "epoch": 31.542234332425068,
+ "grad_norm": 4.788729667663574,
+ "learning_rate": 1.6023031745756864e-05,
+ "loss": 0.2252,
+ "step": 11576
+ },
+ {
+ "epoch": 31.544959128065393,
+ "grad_norm": 4.884683609008789,
+ "learning_rate": 1.602232725558153e-05,
+ "loss": 0.4,
+ "step": 11577
+ },
+ {
+ "epoch": 31.547683923705723,
+ "grad_norm": 3.9115471839904785,
+ "learning_rate": 1.6021622718504503e-05,
+ "loss": 0.2424,
+ "step": 11578
+ },
+ {
+ "epoch": 31.55040871934605,
+ "grad_norm": 4.187444686889648,
+ "learning_rate": 1.602091813453127e-05,
+ "loss": 0.2551,
+ "step": 11579
+ },
+ {
+ "epoch": 31.553133514986374,
+ "grad_norm": 4.591029644012451,
+ "learning_rate": 1.602021350366732e-05,
+ "loss": 0.2949,
+ "step": 11580
+ },
+ {
+ "epoch": 31.555858310626704,
+ "grad_norm": 3.8972012996673584,
+ "learning_rate": 1.6019508825918144e-05,
+ "loss": 0.2248,
+ "step": 11581
+ },
+ {
+ "epoch": 31.55858310626703,
+ "grad_norm": 4.88706111907959,
+ "learning_rate": 1.6018804101289224e-05,
+ "loss": 0.4373,
+ "step": 11582
+ },
+ {
+ "epoch": 31.561307901907355,
+ "grad_norm": 7.922754287719727,
+ "learning_rate": 1.6018099329786053e-05,
+ "loss": 0.3043,
+ "step": 11583
+ },
+ {
+ "epoch": 31.564032697547685,
+ "grad_norm": 5.608799457550049,
+ "learning_rate": 1.6017394511414116e-05,
+ "loss": 0.27,
+ "step": 11584
+ },
+ {
+ "epoch": 31.56675749318801,
+ "grad_norm": 5.115306854248047,
+ "learning_rate": 1.6016689646178906e-05,
+ "loss": 0.2272,
+ "step": 11585
+ },
+ {
+ "epoch": 31.569482288828336,
+ "grad_norm": 4.570239067077637,
+ "learning_rate": 1.601598473408591e-05,
+ "loss": 0.3192,
+ "step": 11586
+ },
+ {
+ "epoch": 31.572207084468666,
+ "grad_norm": 4.194909572601318,
+ "learning_rate": 1.601527977514062e-05,
+ "loss": 0.246,
+ "step": 11587
+ },
+ {
+ "epoch": 31.57493188010899,
+ "grad_norm": 4.0992960929870605,
+ "learning_rate": 1.6014574769348526e-05,
+ "loss": 0.1839,
+ "step": 11588
+ },
+ {
+ "epoch": 31.577656675749317,
+ "grad_norm": 5.588586330413818,
+ "learning_rate": 1.6013869716715113e-05,
+ "loss": 0.3079,
+ "step": 11589
+ },
+ {
+ "epoch": 31.580381471389646,
+ "grad_norm": 4.3024091720581055,
+ "learning_rate": 1.601316461724588e-05,
+ "loss": 0.3631,
+ "step": 11590
+ },
+ {
+ "epoch": 31.583106267029972,
+ "grad_norm": 6.555357933044434,
+ "learning_rate": 1.6012459470946315e-05,
+ "loss": 0.2823,
+ "step": 11591
+ },
+ {
+ "epoch": 31.585831062670298,
+ "grad_norm": 5.622933387756348,
+ "learning_rate": 1.6011754277821906e-05,
+ "loss": 0.19,
+ "step": 11592
+ },
+ {
+ "epoch": 31.588555858310627,
+ "grad_norm": 4.179620265960693,
+ "learning_rate": 1.6011049037878145e-05,
+ "loss": 0.1564,
+ "step": 11593
+ },
+ {
+ "epoch": 31.591280653950953,
+ "grad_norm": 4.993957996368408,
+ "learning_rate": 1.6010343751120533e-05,
+ "loss": 0.2928,
+ "step": 11594
+ },
+ {
+ "epoch": 31.59400544959128,
+ "grad_norm": 3.7739384174346924,
+ "learning_rate": 1.6009638417554555e-05,
+ "loss": 0.2168,
+ "step": 11595
+ },
+ {
+ "epoch": 31.59673024523161,
+ "grad_norm": 4.137946605682373,
+ "learning_rate": 1.600893303718571e-05,
+ "loss": 0.2915,
+ "step": 11596
+ },
+ {
+ "epoch": 31.599455040871934,
+ "grad_norm": 4.26958703994751,
+ "learning_rate": 1.6008227610019485e-05,
+ "loss": 0.1727,
+ "step": 11597
+ },
+ {
+ "epoch": 31.60217983651226,
+ "grad_norm": 4.374048709869385,
+ "learning_rate": 1.6007522136061378e-05,
+ "loss": 0.1965,
+ "step": 11598
+ },
+ {
+ "epoch": 31.60490463215259,
+ "grad_norm": 4.956295013427734,
+ "learning_rate": 1.600681661531688e-05,
+ "loss": 0.2403,
+ "step": 11599
+ },
+ {
+ "epoch": 31.607629427792915,
+ "grad_norm": 4.513427257537842,
+ "learning_rate": 1.600611104779149e-05,
+ "loss": 0.2065,
+ "step": 11600
+ },
+ {
+ "epoch": 31.61035422343324,
+ "grad_norm": 4.654882431030273,
+ "learning_rate": 1.6005405433490694e-05,
+ "loss": 0.2258,
+ "step": 11601
+ },
+ {
+ "epoch": 31.61307901907357,
+ "grad_norm": 3.589621067047119,
+ "learning_rate": 1.600469977242e-05,
+ "loss": 0.361,
+ "step": 11602
+ },
+ {
+ "epoch": 31.615803814713896,
+ "grad_norm": 3.9495604038238525,
+ "learning_rate": 1.6003994064584895e-05,
+ "loss": 0.2173,
+ "step": 11603
+ },
+ {
+ "epoch": 31.618528610354222,
+ "grad_norm": 4.4352707862854,
+ "learning_rate": 1.600328830999088e-05,
+ "loss": 0.2304,
+ "step": 11604
+ },
+ {
+ "epoch": 31.62125340599455,
+ "grad_norm": 4.347018241882324,
+ "learning_rate": 1.6002582508643446e-05,
+ "loss": 0.2344,
+ "step": 11605
+ },
+ {
+ "epoch": 31.623978201634877,
+ "grad_norm": 4.929572105407715,
+ "learning_rate": 1.6001876660548098e-05,
+ "loss": 0.1629,
+ "step": 11606
+ },
+ {
+ "epoch": 31.626702997275203,
+ "grad_norm": 4.83356237411499,
+ "learning_rate": 1.6001170765710323e-05,
+ "loss": 0.3774,
+ "step": 11607
+ },
+ {
+ "epoch": 31.629427792915532,
+ "grad_norm": 3.4194416999816895,
+ "learning_rate": 1.6000464824135623e-05,
+ "loss": 0.1735,
+ "step": 11608
+ },
+ {
+ "epoch": 31.632152588555858,
+ "grad_norm": 5.059170246124268,
+ "learning_rate": 1.5999758835829497e-05,
+ "loss": 0.3324,
+ "step": 11609
+ },
+ {
+ "epoch": 31.634877384196184,
+ "grad_norm": 4.523209095001221,
+ "learning_rate": 1.599905280079744e-05,
+ "loss": 0.2711,
+ "step": 11610
+ },
+ {
+ "epoch": 31.637602179836513,
+ "grad_norm": 4.861470699310303,
+ "learning_rate": 1.5998346719044956e-05,
+ "loss": 0.25,
+ "step": 11611
+ },
+ {
+ "epoch": 31.64032697547684,
+ "grad_norm": 5.213073253631592,
+ "learning_rate": 1.5997640590577544e-05,
+ "loss": 0.2751,
+ "step": 11612
+ },
+ {
+ "epoch": 31.643051771117165,
+ "grad_norm": 3.9593191146850586,
+ "learning_rate": 1.5996934415400694e-05,
+ "loss": 0.2852,
+ "step": 11613
+ },
+ {
+ "epoch": 31.645776566757494,
+ "grad_norm": 5.1739702224731445,
+ "learning_rate": 1.5996228193519914e-05,
+ "loss": 0.2483,
+ "step": 11614
+ },
+ {
+ "epoch": 31.64850136239782,
+ "grad_norm": 4.652565002441406,
+ "learning_rate": 1.5995521924940704e-05,
+ "loss": 0.2316,
+ "step": 11615
+ },
+ {
+ "epoch": 31.651226158038146,
+ "grad_norm": 4.835082530975342,
+ "learning_rate": 1.5994815609668558e-05,
+ "loss": 0.4183,
+ "step": 11616
+ },
+ {
+ "epoch": 31.653950953678475,
+ "grad_norm": 4.898375511169434,
+ "learning_rate": 1.5994109247708983e-05,
+ "loss": 0.4653,
+ "step": 11617
+ },
+ {
+ "epoch": 31.6566757493188,
+ "grad_norm": 4.712100505828857,
+ "learning_rate": 1.5993402839067477e-05,
+ "loss": 0.2416,
+ "step": 11618
+ },
+ {
+ "epoch": 31.659400544959126,
+ "grad_norm": 5.625110149383545,
+ "learning_rate": 1.5992696383749543e-05,
+ "loss": 0.26,
+ "step": 11619
+ },
+ {
+ "epoch": 31.662125340599456,
+ "grad_norm": 4.367286205291748,
+ "learning_rate": 1.5991989881760687e-05,
+ "loss": 0.2606,
+ "step": 11620
+ },
+ {
+ "epoch": 31.66485013623978,
+ "grad_norm": 3.750417709350586,
+ "learning_rate": 1.59912833331064e-05,
+ "loss": 0.4348,
+ "step": 11621
+ },
+ {
+ "epoch": 31.667574931880107,
+ "grad_norm": 6.12209939956665,
+ "learning_rate": 1.5990576737792192e-05,
+ "loss": 0.2078,
+ "step": 11622
+ },
+ {
+ "epoch": 31.670299727520437,
+ "grad_norm": 4.55435037612915,
+ "learning_rate": 1.598987009582357e-05,
+ "loss": 0.2383,
+ "step": 11623
+ },
+ {
+ "epoch": 31.673024523160763,
+ "grad_norm": 4.647312641143799,
+ "learning_rate": 1.598916340720603e-05,
+ "loss": 0.2885,
+ "step": 11624
+ },
+ {
+ "epoch": 31.67574931880109,
+ "grad_norm": 5.479722023010254,
+ "learning_rate": 1.5988456671945078e-05,
+ "loss": 0.2362,
+ "step": 11625
+ },
+ {
+ "epoch": 31.678474114441418,
+ "grad_norm": 3.9234468936920166,
+ "learning_rate": 1.598774989004622e-05,
+ "loss": 0.3254,
+ "step": 11626
+ },
+ {
+ "epoch": 31.681198910081743,
+ "grad_norm": 4.091623306274414,
+ "learning_rate": 1.5987043061514953e-05,
+ "loss": 0.2492,
+ "step": 11627
+ },
+ {
+ "epoch": 31.68392370572207,
+ "grad_norm": 3.8270792961120605,
+ "learning_rate": 1.598633618635679e-05,
+ "loss": 0.3706,
+ "step": 11628
+ },
+ {
+ "epoch": 31.6866485013624,
+ "grad_norm": 5.170799255371094,
+ "learning_rate": 1.5985629264577237e-05,
+ "loss": 0.3988,
+ "step": 11629
+ },
+ {
+ "epoch": 31.689373297002724,
+ "grad_norm": 5.549046039581299,
+ "learning_rate": 1.5984922296181794e-05,
+ "loss": 0.3701,
+ "step": 11630
+ },
+ {
+ "epoch": 31.69209809264305,
+ "grad_norm": 5.570286273956299,
+ "learning_rate": 1.5984215281175964e-05,
+ "loss": 0.2318,
+ "step": 11631
+ },
+ {
+ "epoch": 31.69482288828338,
+ "grad_norm": 4.8129563331604,
+ "learning_rate": 1.5983508219565264e-05,
+ "loss": 0.1758,
+ "step": 11632
+ },
+ {
+ "epoch": 31.697547683923705,
+ "grad_norm": 3.9059598445892334,
+ "learning_rate": 1.5982801111355195e-05,
+ "loss": 0.23,
+ "step": 11633
+ },
+ {
+ "epoch": 31.70027247956403,
+ "grad_norm": 5.10753059387207,
+ "learning_rate": 1.598209395655126e-05,
+ "loss": 0.3435,
+ "step": 11634
+ },
+ {
+ "epoch": 31.70299727520436,
+ "grad_norm": 4.024787902832031,
+ "learning_rate": 1.598138675515897e-05,
+ "loss": 0.3341,
+ "step": 11635
+ },
+ {
+ "epoch": 31.705722070844686,
+ "grad_norm": 4.410848617553711,
+ "learning_rate": 1.5980679507183834e-05,
+ "loss": 0.2935,
+ "step": 11636
+ },
+ {
+ "epoch": 31.708446866485012,
+ "grad_norm": 5.057963848114014,
+ "learning_rate": 1.5979972212631358e-05,
+ "loss": 0.2401,
+ "step": 11637
+ },
+ {
+ "epoch": 31.71117166212534,
+ "grad_norm": 3.9551103115081787,
+ "learning_rate": 1.597926487150705e-05,
+ "loss": 0.2663,
+ "step": 11638
+ },
+ {
+ "epoch": 31.713896457765667,
+ "grad_norm": 4.115054130554199,
+ "learning_rate": 1.5978557483816423e-05,
+ "loss": 0.1882,
+ "step": 11639
+ },
+ {
+ "epoch": 31.716621253405993,
+ "grad_norm": 4.622337818145752,
+ "learning_rate": 1.5977850049564978e-05,
+ "loss": 0.3255,
+ "step": 11640
+ },
+ {
+ "epoch": 31.719346049046322,
+ "grad_norm": 4.634757995605469,
+ "learning_rate": 1.597714256875823e-05,
+ "loss": 0.2126,
+ "step": 11641
+ },
+ {
+ "epoch": 31.722070844686648,
+ "grad_norm": 4.702320098876953,
+ "learning_rate": 1.597643504140169e-05,
+ "loss": 0.2298,
+ "step": 11642
+ },
+ {
+ "epoch": 31.724795640326974,
+ "grad_norm": 4.638733863830566,
+ "learning_rate": 1.597572746750087e-05,
+ "loss": 0.2215,
+ "step": 11643
+ },
+ {
+ "epoch": 31.727520435967303,
+ "grad_norm": 4.194699764251709,
+ "learning_rate": 1.597501984706127e-05,
+ "loss": 0.2585,
+ "step": 11644
+ },
+ {
+ "epoch": 31.73024523160763,
+ "grad_norm": 5.10373067855835,
+ "learning_rate": 1.5974312180088407e-05,
+ "loss": 0.3501,
+ "step": 11645
+ },
+ {
+ "epoch": 31.732970027247955,
+ "grad_norm": 4.947055339813232,
+ "learning_rate": 1.5973604466587797e-05,
+ "loss": 0.2554,
+ "step": 11646
+ },
+ {
+ "epoch": 31.735694822888284,
+ "grad_norm": 3.9786081314086914,
+ "learning_rate": 1.597289670656495e-05,
+ "loss": 0.2357,
+ "step": 11647
+ },
+ {
+ "epoch": 31.73841961852861,
+ "grad_norm": 4.638571739196777,
+ "learning_rate": 1.5972188900025368e-05,
+ "loss": 0.2565,
+ "step": 11648
+ },
+ {
+ "epoch": 31.741144414168936,
+ "grad_norm": 3.59051251411438,
+ "learning_rate": 1.5971481046974576e-05,
+ "loss": 0.1296,
+ "step": 11649
+ },
+ {
+ "epoch": 31.743869209809265,
+ "grad_norm": 4.225552558898926,
+ "learning_rate": 1.5970773147418084e-05,
+ "loss": 0.2158,
+ "step": 11650
+ },
+ {
+ "epoch": 31.74659400544959,
+ "grad_norm": 4.156633377075195,
+ "learning_rate": 1.59700652013614e-05,
+ "loss": 0.2623,
+ "step": 11651
+ },
+ {
+ "epoch": 31.749318801089917,
+ "grad_norm": 5.363273620605469,
+ "learning_rate": 1.5969357208810043e-05,
+ "loss": 0.2107,
+ "step": 11652
+ },
+ {
+ "epoch": 31.752043596730246,
+ "grad_norm": 4.333967208862305,
+ "learning_rate": 1.5968649169769518e-05,
+ "loss": 0.2848,
+ "step": 11653
+ },
+ {
+ "epoch": 31.754768392370572,
+ "grad_norm": 4.447865962982178,
+ "learning_rate": 1.596794108424535e-05,
+ "loss": 0.1937,
+ "step": 11654
+ },
+ {
+ "epoch": 31.757493188010898,
+ "grad_norm": 4.73881721496582,
+ "learning_rate": 1.5967232952243048e-05,
+ "loss": 0.237,
+ "step": 11655
+ },
+ {
+ "epoch": 31.760217983651227,
+ "grad_norm": 4.096786022186279,
+ "learning_rate": 1.5966524773768127e-05,
+ "loss": 0.3057,
+ "step": 11656
+ },
+ {
+ "epoch": 31.762942779291553,
+ "grad_norm": 4.382179260253906,
+ "learning_rate": 1.5965816548826104e-05,
+ "loss": 0.213,
+ "step": 11657
+ },
+ {
+ "epoch": 31.76566757493188,
+ "grad_norm": 4.561371326446533,
+ "learning_rate": 1.5965108277422493e-05,
+ "loss": 0.2115,
+ "step": 11658
+ },
+ {
+ "epoch": 31.768392370572208,
+ "grad_norm": 4.440755367279053,
+ "learning_rate": 1.5964399959562812e-05,
+ "loss": 0.4111,
+ "step": 11659
+ },
+ {
+ "epoch": 31.771117166212534,
+ "grad_norm": 3.6808364391326904,
+ "learning_rate": 1.5963691595252576e-05,
+ "loss": 0.1834,
+ "step": 11660
+ },
+ {
+ "epoch": 31.77384196185286,
+ "grad_norm": 4.110843181610107,
+ "learning_rate": 1.5962983184497304e-05,
+ "loss": 0.2476,
+ "step": 11661
+ },
+ {
+ "epoch": 31.77656675749319,
+ "grad_norm": 4.533279895782471,
+ "learning_rate": 1.5962274727302506e-05,
+ "loss": 0.2476,
+ "step": 11662
+ },
+ {
+ "epoch": 31.779291553133515,
+ "grad_norm": 4.052728652954102,
+ "learning_rate": 1.5961566223673707e-05,
+ "loss": 0.2515,
+ "step": 11663
+ },
+ {
+ "epoch": 31.78201634877384,
+ "grad_norm": 4.542088031768799,
+ "learning_rate": 1.5960857673616423e-05,
+ "loss": 0.2107,
+ "step": 11664
+ },
+ {
+ "epoch": 31.78474114441417,
+ "grad_norm": 5.049928188323975,
+ "learning_rate": 1.5960149077136167e-05,
+ "loss": 0.1825,
+ "step": 11665
+ },
+ {
+ "epoch": 31.787465940054496,
+ "grad_norm": 4.265908718109131,
+ "learning_rate": 1.5959440434238465e-05,
+ "loss": 0.17,
+ "step": 11666
+ },
+ {
+ "epoch": 31.79019073569482,
+ "grad_norm": 4.9208455085754395,
+ "learning_rate": 1.5958731744928836e-05,
+ "loss": 0.1871,
+ "step": 11667
+ },
+ {
+ "epoch": 31.79291553133515,
+ "grad_norm": 4.513894081115723,
+ "learning_rate": 1.595802300921279e-05,
+ "loss": 0.341,
+ "step": 11668
+ },
+ {
+ "epoch": 31.795640326975477,
+ "grad_norm": 5.851859092712402,
+ "learning_rate": 1.5957314227095854e-05,
+ "loss": 0.2943,
+ "step": 11669
+ },
+ {
+ "epoch": 31.798365122615802,
+ "grad_norm": 3.8742423057556152,
+ "learning_rate": 1.5956605398583547e-05,
+ "loss": 0.2898,
+ "step": 11670
+ },
+ {
+ "epoch": 31.80108991825613,
+ "grad_norm": 3.971703290939331,
+ "learning_rate": 1.5955896523681392e-05,
+ "loss": 0.4166,
+ "step": 11671
+ },
+ {
+ "epoch": 31.803814713896458,
+ "grad_norm": 3.972158670425415,
+ "learning_rate": 1.5955187602394904e-05,
+ "loss": 0.1985,
+ "step": 11672
+ },
+ {
+ "epoch": 31.806539509536783,
+ "grad_norm": 3.8879880905151367,
+ "learning_rate": 1.5954478634729605e-05,
+ "loss": 0.3996,
+ "step": 11673
+ },
+ {
+ "epoch": 31.809264305177113,
+ "grad_norm": 4.473315238952637,
+ "learning_rate": 1.595376962069102e-05,
+ "loss": 0.2999,
+ "step": 11674
+ },
+ {
+ "epoch": 31.81198910081744,
+ "grad_norm": 4.950345993041992,
+ "learning_rate": 1.5953060560284666e-05,
+ "loss": 0.2257,
+ "step": 11675
+ },
+ {
+ "epoch": 31.814713896457764,
+ "grad_norm": 4.746415138244629,
+ "learning_rate": 1.5952351453516076e-05,
+ "loss": 0.1972,
+ "step": 11676
+ },
+ {
+ "epoch": 31.817438692098094,
+ "grad_norm": 4.748571872711182,
+ "learning_rate": 1.5951642300390755e-05,
+ "loss": 0.3399,
+ "step": 11677
+ },
+ {
+ "epoch": 31.82016348773842,
+ "grad_norm": 4.113661766052246,
+ "learning_rate": 1.5950933100914242e-05,
+ "loss": 0.1722,
+ "step": 11678
+ },
+ {
+ "epoch": 31.822888283378745,
+ "grad_norm": 4.2120771408081055,
+ "learning_rate": 1.5950223855092047e-05,
+ "loss": 0.3008,
+ "step": 11679
+ },
+ {
+ "epoch": 31.825613079019075,
+ "grad_norm": 4.297011375427246,
+ "learning_rate": 1.5949514562929706e-05,
+ "loss": 0.3976,
+ "step": 11680
+ },
+ {
+ "epoch": 31.8283378746594,
+ "grad_norm": 4.737847805023193,
+ "learning_rate": 1.5948805224432736e-05,
+ "loss": 0.234,
+ "step": 11681
+ },
+ {
+ "epoch": 31.831062670299726,
+ "grad_norm": 3.9679458141326904,
+ "learning_rate": 1.5948095839606657e-05,
+ "loss": 0.2011,
+ "step": 11682
+ },
+ {
+ "epoch": 31.833787465940055,
+ "grad_norm": 3.8341026306152344,
+ "learning_rate": 1.5947386408457005e-05,
+ "loss": 0.2053,
+ "step": 11683
+ },
+ {
+ "epoch": 31.83651226158038,
+ "grad_norm": 5.402500152587891,
+ "learning_rate": 1.5946676930989296e-05,
+ "loss": 0.2012,
+ "step": 11684
+ },
+ {
+ "epoch": 31.839237057220707,
+ "grad_norm": 4.49010705947876,
+ "learning_rate": 1.5945967407209062e-05,
+ "loss": 0.1647,
+ "step": 11685
+ },
+ {
+ "epoch": 31.841961852861036,
+ "grad_norm": 4.335521697998047,
+ "learning_rate": 1.5945257837121825e-05,
+ "loss": 0.1767,
+ "step": 11686
+ },
+ {
+ "epoch": 31.844686648501362,
+ "grad_norm": 4.82133150100708,
+ "learning_rate": 1.5944548220733107e-05,
+ "loss": 0.1909,
+ "step": 11687
+ },
+ {
+ "epoch": 31.847411444141688,
+ "grad_norm": 4.206963062286377,
+ "learning_rate": 1.5943838558048442e-05,
+ "loss": 0.1876,
+ "step": 11688
+ },
+ {
+ "epoch": 31.850136239782017,
+ "grad_norm": 4.530407428741455,
+ "learning_rate": 1.594312884907335e-05,
+ "loss": 0.2422,
+ "step": 11689
+ },
+ {
+ "epoch": 31.852861035422343,
+ "grad_norm": 4.30127477645874,
+ "learning_rate": 1.5942419093813365e-05,
+ "loss": 0.276,
+ "step": 11690
+ },
+ {
+ "epoch": 31.85558583106267,
+ "grad_norm": 5.597133159637451,
+ "learning_rate": 1.5941709292274012e-05,
+ "loss": 0.2039,
+ "step": 11691
+ },
+ {
+ "epoch": 31.858310626703,
+ "grad_norm": 5.340417861938477,
+ "learning_rate": 1.594099944446082e-05,
+ "loss": 0.278,
+ "step": 11692
+ },
+ {
+ "epoch": 31.861035422343324,
+ "grad_norm": 3.8648040294647217,
+ "learning_rate": 1.5940289550379312e-05,
+ "loss": 0.1894,
+ "step": 11693
+ },
+ {
+ "epoch": 31.86376021798365,
+ "grad_norm": 4.176829814910889,
+ "learning_rate": 1.5939579610035017e-05,
+ "loss": 0.3237,
+ "step": 11694
+ },
+ {
+ "epoch": 31.86648501362398,
+ "grad_norm": 5.330592632293701,
+ "learning_rate": 1.5938869623433474e-05,
+ "loss": 0.1717,
+ "step": 11695
+ },
+ {
+ "epoch": 31.869209809264305,
+ "grad_norm": 4.333252429962158,
+ "learning_rate": 1.59381595905802e-05,
+ "loss": 0.1821,
+ "step": 11696
+ },
+ {
+ "epoch": 31.87193460490463,
+ "grad_norm": 4.333609104156494,
+ "learning_rate": 1.5937449511480735e-05,
+ "loss": 0.279,
+ "step": 11697
+ },
+ {
+ "epoch": 31.87465940054496,
+ "grad_norm": 4.498837947845459,
+ "learning_rate": 1.5936739386140602e-05,
+ "loss": 0.3011,
+ "step": 11698
+ },
+ {
+ "epoch": 31.877384196185286,
+ "grad_norm": 4.509099006652832,
+ "learning_rate": 1.593602921456533e-05,
+ "loss": 0.2246,
+ "step": 11699
+ },
+ {
+ "epoch": 31.88010899182561,
+ "grad_norm": 5.692264556884766,
+ "learning_rate": 1.593531899676046e-05,
+ "loss": 0.1859,
+ "step": 11700
+ },
+ {
+ "epoch": 31.88283378746594,
+ "grad_norm": 4.23037052154541,
+ "learning_rate": 1.5934608732731516e-05,
+ "loss": 0.1425,
+ "step": 11701
+ },
+ {
+ "epoch": 31.885558583106267,
+ "grad_norm": 3.9947612285614014,
+ "learning_rate": 1.5933898422484026e-05,
+ "loss": 0.3832,
+ "step": 11702
+ },
+ {
+ "epoch": 31.888283378746593,
+ "grad_norm": 4.082513809204102,
+ "learning_rate": 1.593318806602353e-05,
+ "loss": 0.2457,
+ "step": 11703
+ },
+ {
+ "epoch": 31.891008174386922,
+ "grad_norm": 4.7094268798828125,
+ "learning_rate": 1.5932477663355554e-05,
+ "loss": 0.1789,
+ "step": 11704
+ },
+ {
+ "epoch": 31.893732970027248,
+ "grad_norm": 4.052562236785889,
+ "learning_rate": 1.593176721448563e-05,
+ "loss": 0.1707,
+ "step": 11705
+ },
+ {
+ "epoch": 31.896457765667574,
+ "grad_norm": 4.823851108551025,
+ "learning_rate": 1.59310567194193e-05,
+ "loss": 0.2258,
+ "step": 11706
+ },
+ {
+ "epoch": 31.899182561307903,
+ "grad_norm": 4.566215515136719,
+ "learning_rate": 1.593034617816209e-05,
+ "loss": 0.3638,
+ "step": 11707
+ },
+ {
+ "epoch": 31.90190735694823,
+ "grad_norm": 4.5977396965026855,
+ "learning_rate": 1.592963559071953e-05,
+ "loss": 0.199,
+ "step": 11708
+ },
+ {
+ "epoch": 31.904632152588555,
+ "grad_norm": 4.148005962371826,
+ "learning_rate": 1.5928924957097166e-05,
+ "loss": 0.1793,
+ "step": 11709
+ },
+ {
+ "epoch": 31.907356948228884,
+ "grad_norm": 5.769320964813232,
+ "learning_rate": 1.5928214277300517e-05,
+ "loss": 0.3122,
+ "step": 11710
+ },
+ {
+ "epoch": 31.91008174386921,
+ "grad_norm": 4.046979904174805,
+ "learning_rate": 1.5927503551335132e-05,
+ "loss": 0.1918,
+ "step": 11711
+ },
+ {
+ "epoch": 31.912806539509535,
+ "grad_norm": 4.552387237548828,
+ "learning_rate": 1.592679277920654e-05,
+ "loss": 0.2814,
+ "step": 11712
+ },
+ {
+ "epoch": 31.915531335149865,
+ "grad_norm": 4.431561470031738,
+ "learning_rate": 1.5926081960920276e-05,
+ "loss": 0.2562,
+ "step": 11713
+ },
+ {
+ "epoch": 31.91825613079019,
+ "grad_norm": 4.828364372253418,
+ "learning_rate": 1.5925371096481873e-05,
+ "loss": 0.314,
+ "step": 11714
+ },
+ {
+ "epoch": 31.920980926430516,
+ "grad_norm": 4.633449554443359,
+ "learning_rate": 1.592466018589687e-05,
+ "loss": 0.1848,
+ "step": 11715
+ },
+ {
+ "epoch": 31.923705722070846,
+ "grad_norm": 3.908780813217163,
+ "learning_rate": 1.592394922917081e-05,
+ "loss": 0.2734,
+ "step": 11716
+ },
+ {
+ "epoch": 31.92643051771117,
+ "grad_norm": 4.972283363342285,
+ "learning_rate": 1.592323822630922e-05,
+ "loss": 0.2374,
+ "step": 11717
+ },
+ {
+ "epoch": 31.929155313351497,
+ "grad_norm": 3.9196274280548096,
+ "learning_rate": 1.5922527177317646e-05,
+ "loss": 0.2333,
+ "step": 11718
+ },
+ {
+ "epoch": 31.931880108991827,
+ "grad_norm": 4.429347991943359,
+ "learning_rate": 1.5921816082201614e-05,
+ "loss": 0.2496,
+ "step": 11719
+ },
+ {
+ "epoch": 31.934604904632153,
+ "grad_norm": 3.86582612991333,
+ "learning_rate": 1.5921104940966675e-05,
+ "loss": 0.4768,
+ "step": 11720
+ },
+ {
+ "epoch": 31.93732970027248,
+ "grad_norm": 4.592306137084961,
+ "learning_rate": 1.592039375361836e-05,
+ "loss": 0.2618,
+ "step": 11721
+ },
+ {
+ "epoch": 31.940054495912808,
+ "grad_norm": 4.4390950202941895,
+ "learning_rate": 1.5919682520162208e-05,
+ "loss": 0.2925,
+ "step": 11722
+ },
+ {
+ "epoch": 31.942779291553133,
+ "grad_norm": 5.075019359588623,
+ "learning_rate": 1.591897124060376e-05,
+ "loss": 0.4381,
+ "step": 11723
+ },
+ {
+ "epoch": 31.94550408719346,
+ "grad_norm": 4.240757942199707,
+ "learning_rate": 1.591825991494855e-05,
+ "loss": 0.1771,
+ "step": 11724
+ },
+ {
+ "epoch": 31.94822888283379,
+ "grad_norm": 3.71997332572937,
+ "learning_rate": 1.5917548543202127e-05,
+ "loss": 0.2722,
+ "step": 11725
+ },
+ {
+ "epoch": 31.950953678474114,
+ "grad_norm": 5.110382556915283,
+ "learning_rate": 1.5916837125370025e-05,
+ "loss": 0.3023,
+ "step": 11726
+ },
+ {
+ "epoch": 31.95367847411444,
+ "grad_norm": 4.657098293304443,
+ "learning_rate": 1.5916125661457786e-05,
+ "loss": 0.2133,
+ "step": 11727
+ },
+ {
+ "epoch": 31.95640326975477,
+ "grad_norm": 4.721577167510986,
+ "learning_rate": 1.5915414151470953e-05,
+ "loss": 0.312,
+ "step": 11728
+ },
+ {
+ "epoch": 31.959128065395095,
+ "grad_norm": 4.303000450134277,
+ "learning_rate": 1.5914702595415062e-05,
+ "loss": 0.1837,
+ "step": 11729
+ },
+ {
+ "epoch": 31.96185286103542,
+ "grad_norm": 4.2332329750061035,
+ "learning_rate": 1.5913990993295655e-05,
+ "loss": 0.2847,
+ "step": 11730
+ },
+ {
+ "epoch": 31.96457765667575,
+ "grad_norm": 5.139564514160156,
+ "learning_rate": 1.5913279345118284e-05,
+ "loss": 0.2994,
+ "step": 11731
+ },
+ {
+ "epoch": 31.967302452316076,
+ "grad_norm": 4.840035915374756,
+ "learning_rate": 1.5912567650888476e-05,
+ "loss": 0.218,
+ "step": 11732
+ },
+ {
+ "epoch": 31.970027247956402,
+ "grad_norm": 5.449479579925537,
+ "learning_rate": 1.5911855910611785e-05,
+ "loss": 0.3442,
+ "step": 11733
+ },
+ {
+ "epoch": 31.97275204359673,
+ "grad_norm": 4.539453029632568,
+ "learning_rate": 1.591114412429375e-05,
+ "loss": 0.2863,
+ "step": 11734
+ },
+ {
+ "epoch": 31.975476839237057,
+ "grad_norm": 4.0494303703308105,
+ "learning_rate": 1.5910432291939917e-05,
+ "loss": 0.2295,
+ "step": 11735
+ },
+ {
+ "epoch": 31.978201634877383,
+ "grad_norm": 3.9810187816619873,
+ "learning_rate": 1.5909720413555826e-05,
+ "loss": 0.2716,
+ "step": 11736
+ },
+ {
+ "epoch": 31.980926430517712,
+ "grad_norm": 4.005823612213135,
+ "learning_rate": 1.590900848914702e-05,
+ "loss": 0.1851,
+ "step": 11737
+ },
+ {
+ "epoch": 31.983651226158038,
+ "grad_norm": 4.420792579650879,
+ "learning_rate": 1.590829651871905e-05,
+ "loss": 0.3125,
+ "step": 11738
+ },
+ {
+ "epoch": 31.986376021798364,
+ "grad_norm": 4.724678993225098,
+ "learning_rate": 1.5907584502277456e-05,
+ "loss": 0.23,
+ "step": 11739
+ },
+ {
+ "epoch": 31.989100817438693,
+ "grad_norm": 3.9495747089385986,
+ "learning_rate": 1.5906872439827782e-05,
+ "loss": 0.3834,
+ "step": 11740
+ },
+ {
+ "epoch": 31.99182561307902,
+ "grad_norm": 3.9503324031829834,
+ "learning_rate": 1.5906160331375577e-05,
+ "loss": 0.2865,
+ "step": 11741
+ },
+ {
+ "epoch": 31.994550408719345,
+ "grad_norm": 4.538914680480957,
+ "learning_rate": 1.5905448176926385e-05,
+ "loss": 0.1679,
+ "step": 11742
+ },
+ {
+ "epoch": 31.997275204359674,
+ "grad_norm": 5.24288272857666,
+ "learning_rate": 1.5904735976485754e-05,
+ "loss": 0.4558,
+ "step": 11743
+ },
+ {
+ "epoch": 32.0,
+ "grad_norm": 4.197296142578125,
+ "learning_rate": 1.5904023730059227e-05,
+ "loss": 0.4063,
+ "step": 11744
+ },
+ {
+ "epoch": 32.002724795640326,
+ "grad_norm": 3.45536208152771,
+ "learning_rate": 1.5903311437652357e-05,
+ "loss": 0.1888,
+ "step": 11745
+ },
+ {
+ "epoch": 32.00544959128065,
+ "grad_norm": 3.74198579788208,
+ "learning_rate": 1.5902599099270686e-05,
+ "loss": 0.2341,
+ "step": 11746
+ },
+ {
+ "epoch": 32.00817438692098,
+ "grad_norm": 5.219640731811523,
+ "learning_rate": 1.5901886714919758e-05,
+ "loss": 0.1414,
+ "step": 11747
+ },
+ {
+ "epoch": 32.01089918256131,
+ "grad_norm": 3.9816787242889404,
+ "learning_rate": 1.5901174284605132e-05,
+ "loss": 0.1774,
+ "step": 11748
+ },
+ {
+ "epoch": 32.013623978201636,
+ "grad_norm": 3.5285401344299316,
+ "learning_rate": 1.590046180833235e-05,
+ "loss": 0.192,
+ "step": 11749
+ },
+ {
+ "epoch": 32.01634877384196,
+ "grad_norm": 5.728002071380615,
+ "learning_rate": 1.5899749286106957e-05,
+ "loss": 0.3308,
+ "step": 11750
+ },
+ {
+ "epoch": 32.01907356948229,
+ "grad_norm": 3.7930960655212402,
+ "learning_rate": 1.5899036717934514e-05,
+ "loss": 0.2564,
+ "step": 11751
+ },
+ {
+ "epoch": 32.02179836512261,
+ "grad_norm": 4.146435737609863,
+ "learning_rate": 1.5898324103820556e-05,
+ "loss": 0.1678,
+ "step": 11752
+ },
+ {
+ "epoch": 32.02452316076294,
+ "grad_norm": 3.84220552444458,
+ "learning_rate": 1.5897611443770643e-05,
+ "loss": 0.1657,
+ "step": 11753
+ },
+ {
+ "epoch": 32.02724795640327,
+ "grad_norm": 4.228939056396484,
+ "learning_rate": 1.5896898737790323e-05,
+ "loss": 0.3627,
+ "step": 11754
+ },
+ {
+ "epoch": 32.0299727520436,
+ "grad_norm": 3.7248566150665283,
+ "learning_rate": 1.589618598588514e-05,
+ "loss": 0.2543,
+ "step": 11755
+ },
+ {
+ "epoch": 32.032697547683924,
+ "grad_norm": 4.011236190795898,
+ "learning_rate": 1.5895473188060654e-05,
+ "loss": 0.2009,
+ "step": 11756
+ },
+ {
+ "epoch": 32.03542234332425,
+ "grad_norm": 5.055806636810303,
+ "learning_rate": 1.5894760344322416e-05,
+ "loss": 0.1898,
+ "step": 11757
+ },
+ {
+ "epoch": 32.038147138964575,
+ "grad_norm": 3.926011085510254,
+ "learning_rate": 1.589404745467597e-05,
+ "loss": 0.2006,
+ "step": 11758
+ },
+ {
+ "epoch": 32.0408719346049,
+ "grad_norm": 3.693216562271118,
+ "learning_rate": 1.589333451912687e-05,
+ "loss": 0.4944,
+ "step": 11759
+ },
+ {
+ "epoch": 32.043596730245234,
+ "grad_norm": 3.8130877017974854,
+ "learning_rate": 1.5892621537680674e-05,
+ "loss": 0.1271,
+ "step": 11760
+ },
+ {
+ "epoch": 32.04632152588556,
+ "grad_norm": 4.764291286468506,
+ "learning_rate": 1.5891908510342932e-05,
+ "loss": 0.2296,
+ "step": 11761
+ },
+ {
+ "epoch": 32.049046321525886,
+ "grad_norm": 4.21867036819458,
+ "learning_rate": 1.5891195437119193e-05,
+ "loss": 0.3195,
+ "step": 11762
+ },
+ {
+ "epoch": 32.05177111716621,
+ "grad_norm": 6.698536396026611,
+ "learning_rate": 1.5890482318015017e-05,
+ "loss": 0.2473,
+ "step": 11763
+ },
+ {
+ "epoch": 32.05449591280654,
+ "grad_norm": 4.510526657104492,
+ "learning_rate": 1.5889769153035953e-05,
+ "loss": 0.2501,
+ "step": 11764
+ },
+ {
+ "epoch": 32.05722070844686,
+ "grad_norm": 3.731873035430908,
+ "learning_rate": 1.588905594218755e-05,
+ "loss": 0.1737,
+ "step": 11765
+ },
+ {
+ "epoch": 32.059945504087196,
+ "grad_norm": 3.388423442840576,
+ "learning_rate": 1.588834268547538e-05,
+ "loss": 0.1407,
+ "step": 11766
+ },
+ {
+ "epoch": 32.06267029972752,
+ "grad_norm": 3.800349473953247,
+ "learning_rate": 1.588762938290498e-05,
+ "loss": 0.3675,
+ "step": 11767
+ },
+ {
+ "epoch": 32.06539509536785,
+ "grad_norm": 4.096426486968994,
+ "learning_rate": 1.5886916034481914e-05,
+ "loss": 0.3716,
+ "step": 11768
+ },
+ {
+ "epoch": 32.06811989100817,
+ "grad_norm": 4.4045257568359375,
+ "learning_rate": 1.5886202640211733e-05,
+ "loss": 0.358,
+ "step": 11769
+ },
+ {
+ "epoch": 32.0708446866485,
+ "grad_norm": 3.6338589191436768,
+ "learning_rate": 1.5885489200099996e-05,
+ "loss": 0.1491,
+ "step": 11770
+ },
+ {
+ "epoch": 32.073569482288825,
+ "grad_norm": 4.586032867431641,
+ "learning_rate": 1.588477571415226e-05,
+ "loss": 0.2237,
+ "step": 11771
+ },
+ {
+ "epoch": 32.07629427792916,
+ "grad_norm": 3.8058464527130127,
+ "learning_rate": 1.5884062182374083e-05,
+ "loss": 0.1748,
+ "step": 11772
+ },
+ {
+ "epoch": 32.079019073569484,
+ "grad_norm": 3.732161521911621,
+ "learning_rate": 1.5883348604771013e-05,
+ "loss": 0.1178,
+ "step": 11773
+ },
+ {
+ "epoch": 32.08174386920981,
+ "grad_norm": 4.252373218536377,
+ "learning_rate": 1.5882634981348614e-05,
+ "loss": 0.2257,
+ "step": 11774
+ },
+ {
+ "epoch": 32.084468664850135,
+ "grad_norm": 3.74894642829895,
+ "learning_rate": 1.5881921312112446e-05,
+ "loss": 0.2345,
+ "step": 11775
+ },
+ {
+ "epoch": 32.08719346049046,
+ "grad_norm": 4.000766277313232,
+ "learning_rate": 1.5881207597068065e-05,
+ "loss": 0.2174,
+ "step": 11776
+ },
+ {
+ "epoch": 32.08991825613079,
+ "grad_norm": 4.284023761749268,
+ "learning_rate": 1.5880493836221025e-05,
+ "loss": 0.2185,
+ "step": 11777
+ },
+ {
+ "epoch": 32.09264305177112,
+ "grad_norm": 5.40669059753418,
+ "learning_rate": 1.587978002957689e-05,
+ "loss": 0.2708,
+ "step": 11778
+ },
+ {
+ "epoch": 32.095367847411445,
+ "grad_norm": 5.876282691955566,
+ "learning_rate": 1.587906617714122e-05,
+ "loss": 0.1758,
+ "step": 11779
+ },
+ {
+ "epoch": 32.09809264305177,
+ "grad_norm": 4.33402681350708,
+ "learning_rate": 1.587835227891957e-05,
+ "loss": 0.1969,
+ "step": 11780
+ },
+ {
+ "epoch": 32.1008174386921,
+ "grad_norm": 5.4814772605896,
+ "learning_rate": 1.58776383349175e-05,
+ "loss": 0.219,
+ "step": 11781
+ },
+ {
+ "epoch": 32.10354223433242,
+ "grad_norm": 4.114503860473633,
+ "learning_rate": 1.587692434514057e-05,
+ "loss": 0.1899,
+ "step": 11782
+ },
+ {
+ "epoch": 32.10626702997275,
+ "grad_norm": 3.756746530532837,
+ "learning_rate": 1.5876210309594347e-05,
+ "loss": 0.1373,
+ "step": 11783
+ },
+ {
+ "epoch": 32.10899182561308,
+ "grad_norm": 4.335127353668213,
+ "learning_rate": 1.5875496228284385e-05,
+ "loss": 0.3302,
+ "step": 11784
+ },
+ {
+ "epoch": 32.11171662125341,
+ "grad_norm": 4.094764709472656,
+ "learning_rate": 1.5874782101216246e-05,
+ "loss": 0.3154,
+ "step": 11785
+ },
+ {
+ "epoch": 32.11444141689373,
+ "grad_norm": 5.365327835083008,
+ "learning_rate": 1.5874067928395497e-05,
+ "loss": 0.2288,
+ "step": 11786
+ },
+ {
+ "epoch": 32.11716621253406,
+ "grad_norm": 4.152536869049072,
+ "learning_rate": 1.5873353709827695e-05,
+ "loss": 0.2609,
+ "step": 11787
+ },
+ {
+ "epoch": 32.119891008174385,
+ "grad_norm": 3.746070623397827,
+ "learning_rate": 1.58726394455184e-05,
+ "loss": 0.1287,
+ "step": 11788
+ },
+ {
+ "epoch": 32.12261580381471,
+ "grad_norm": 4.155667781829834,
+ "learning_rate": 1.587192513547318e-05,
+ "loss": 0.1786,
+ "step": 11789
+ },
+ {
+ "epoch": 32.12534059945504,
+ "grad_norm": 3.896745443344116,
+ "learning_rate": 1.5871210779697598e-05,
+ "loss": 0.2856,
+ "step": 11790
+ },
+ {
+ "epoch": 32.12806539509537,
+ "grad_norm": 4.5425543785095215,
+ "learning_rate": 1.587049637819721e-05,
+ "loss": 0.3343,
+ "step": 11791
+ },
+ {
+ "epoch": 32.130790190735695,
+ "grad_norm": 4.446254253387451,
+ "learning_rate": 1.586978193097759e-05,
+ "loss": 0.204,
+ "step": 11792
+ },
+ {
+ "epoch": 32.13351498637602,
+ "grad_norm": 3.6008646488189697,
+ "learning_rate": 1.5869067438044298e-05,
+ "loss": 0.1566,
+ "step": 11793
+ },
+ {
+ "epoch": 32.13623978201635,
+ "grad_norm": 3.534156322479248,
+ "learning_rate": 1.5868352899402895e-05,
+ "loss": 0.3783,
+ "step": 11794
+ },
+ {
+ "epoch": 32.13896457765667,
+ "grad_norm": 4.154545307159424,
+ "learning_rate": 1.5867638315058947e-05,
+ "loss": 0.3836,
+ "step": 11795
+ },
+ {
+ "epoch": 32.141689373297005,
+ "grad_norm": 3.838250160217285,
+ "learning_rate": 1.5866923685018026e-05,
+ "loss": 0.3572,
+ "step": 11796
+ },
+ {
+ "epoch": 32.14441416893733,
+ "grad_norm": 3.3384101390838623,
+ "learning_rate": 1.586620900928569e-05,
+ "loss": 0.171,
+ "step": 11797
+ },
+ {
+ "epoch": 32.14713896457766,
+ "grad_norm": 3.8671445846557617,
+ "learning_rate": 1.5865494287867504e-05,
+ "loss": 0.1742,
+ "step": 11798
+ },
+ {
+ "epoch": 32.14986376021798,
+ "grad_norm": 3.582954168319702,
+ "learning_rate": 1.586477952076904e-05,
+ "loss": 0.2198,
+ "step": 11799
+ },
+ {
+ "epoch": 32.15258855585831,
+ "grad_norm": 4.3708720207214355,
+ "learning_rate": 1.586406470799586e-05,
+ "loss": 0.3787,
+ "step": 11800
+ },
+ {
+ "epoch": 32.155313351498634,
+ "grad_norm": 3.950578212738037,
+ "learning_rate": 1.5863349849553533e-05,
+ "loss": 0.2229,
+ "step": 11801
+ },
+ {
+ "epoch": 32.15803814713897,
+ "grad_norm": 4.314383029937744,
+ "learning_rate": 1.5862634945447628e-05,
+ "loss": 0.2554,
+ "step": 11802
+ },
+ {
+ "epoch": 32.16076294277929,
+ "grad_norm": 3.9565558433532715,
+ "learning_rate": 1.5861919995683713e-05,
+ "loss": 0.2365,
+ "step": 11803
+ },
+ {
+ "epoch": 32.16348773841962,
+ "grad_norm": 4.115336894989014,
+ "learning_rate": 1.586120500026735e-05,
+ "loss": 0.1837,
+ "step": 11804
+ },
+ {
+ "epoch": 32.166212534059945,
+ "grad_norm": 5.295952320098877,
+ "learning_rate": 1.586048995920411e-05,
+ "loss": 0.1927,
+ "step": 11805
+ },
+ {
+ "epoch": 32.16893732970027,
+ "grad_norm": 3.1100692749023438,
+ "learning_rate": 1.5859774872499564e-05,
+ "loss": 0.1423,
+ "step": 11806
+ },
+ {
+ "epoch": 32.171662125340596,
+ "grad_norm": 4.09987211227417,
+ "learning_rate": 1.5859059740159283e-05,
+ "loss": 0.4083,
+ "step": 11807
+ },
+ {
+ "epoch": 32.17438692098093,
+ "grad_norm": 4.280137062072754,
+ "learning_rate": 1.585834456218883e-05,
+ "loss": 0.1449,
+ "step": 11808
+ },
+ {
+ "epoch": 32.177111716621255,
+ "grad_norm": 3.605107307434082,
+ "learning_rate": 1.585762933859378e-05,
+ "loss": 0.2554,
+ "step": 11809
+ },
+ {
+ "epoch": 32.17983651226158,
+ "grad_norm": 4.483892917633057,
+ "learning_rate": 1.58569140693797e-05,
+ "loss": 0.2092,
+ "step": 11810
+ },
+ {
+ "epoch": 32.182561307901906,
+ "grad_norm": 4.803561210632324,
+ "learning_rate": 1.585619875455216e-05,
+ "loss": 0.1534,
+ "step": 11811
+ },
+ {
+ "epoch": 32.18528610354223,
+ "grad_norm": 4.548089027404785,
+ "learning_rate": 1.5855483394116734e-05,
+ "loss": 0.2329,
+ "step": 11812
+ },
+ {
+ "epoch": 32.18801089918256,
+ "grad_norm": 4.273107051849365,
+ "learning_rate": 1.5854767988078992e-05,
+ "loss": 0.1776,
+ "step": 11813
+ },
+ {
+ "epoch": 32.19073569482289,
+ "grad_norm": 3.5236780643463135,
+ "learning_rate": 1.5854052536444505e-05,
+ "loss": 0.319,
+ "step": 11814
+ },
+ {
+ "epoch": 32.19346049046322,
+ "grad_norm": 3.644312620162964,
+ "learning_rate": 1.5853337039218846e-05,
+ "loss": 0.219,
+ "step": 11815
+ },
+ {
+ "epoch": 32.19618528610354,
+ "grad_norm": 4.620031356811523,
+ "learning_rate": 1.5852621496407584e-05,
+ "loss": 0.2329,
+ "step": 11816
+ },
+ {
+ "epoch": 32.19891008174387,
+ "grad_norm": 4.385442733764648,
+ "learning_rate": 1.5851905908016297e-05,
+ "loss": 0.2607,
+ "step": 11817
+ },
+ {
+ "epoch": 32.201634877384194,
+ "grad_norm": 3.9581923484802246,
+ "learning_rate": 1.5851190274050555e-05,
+ "loss": 0.1759,
+ "step": 11818
+ },
+ {
+ "epoch": 32.20435967302452,
+ "grad_norm": 4.515355110168457,
+ "learning_rate": 1.585047459451593e-05,
+ "loss": 0.2582,
+ "step": 11819
+ },
+ {
+ "epoch": 32.20708446866485,
+ "grad_norm": 3.5642290115356445,
+ "learning_rate": 1.5849758869418002e-05,
+ "loss": 0.1526,
+ "step": 11820
+ },
+ {
+ "epoch": 32.20980926430518,
+ "grad_norm": 4.132370471954346,
+ "learning_rate": 1.5849043098762335e-05,
+ "loss": 0.1728,
+ "step": 11821
+ },
+ {
+ "epoch": 32.212534059945504,
+ "grad_norm": 4.262805461883545,
+ "learning_rate": 1.584832728255451e-05,
+ "loss": 0.2791,
+ "step": 11822
+ },
+ {
+ "epoch": 32.21525885558583,
+ "grad_norm": 4.852242469787598,
+ "learning_rate": 1.58476114208001e-05,
+ "loss": 0.1863,
+ "step": 11823
+ },
+ {
+ "epoch": 32.217983651226156,
+ "grad_norm": 3.651420831680298,
+ "learning_rate": 1.584689551350468e-05,
+ "loss": 0.1741,
+ "step": 11824
+ },
+ {
+ "epoch": 32.22070844686648,
+ "grad_norm": 4.481435298919678,
+ "learning_rate": 1.5846179560673825e-05,
+ "loss": 0.1889,
+ "step": 11825
+ },
+ {
+ "epoch": 32.223433242506815,
+ "grad_norm": 3.8741614818573,
+ "learning_rate": 1.5845463562313113e-05,
+ "loss": 0.3054,
+ "step": 11826
+ },
+ {
+ "epoch": 32.22615803814714,
+ "grad_norm": 4.206030368804932,
+ "learning_rate": 1.5844747518428124e-05,
+ "loss": 0.2865,
+ "step": 11827
+ },
+ {
+ "epoch": 32.228882833787466,
+ "grad_norm": 4.113611698150635,
+ "learning_rate": 1.584403142902442e-05,
+ "loss": 0.3031,
+ "step": 11828
+ },
+ {
+ "epoch": 32.23160762942779,
+ "grad_norm": 4.322437763214111,
+ "learning_rate": 1.5843315294107595e-05,
+ "loss": 0.1344,
+ "step": 11829
+ },
+ {
+ "epoch": 32.23433242506812,
+ "grad_norm": 4.424440860748291,
+ "learning_rate": 1.5842599113683216e-05,
+ "loss": 0.2596,
+ "step": 11830
+ },
+ {
+ "epoch": 32.237057220708444,
+ "grad_norm": 4.068488597869873,
+ "learning_rate": 1.5841882887756864e-05,
+ "loss": 0.2388,
+ "step": 11831
+ },
+ {
+ "epoch": 32.23978201634878,
+ "grad_norm": 3.8762624263763428,
+ "learning_rate": 1.5841166616334112e-05,
+ "loss": 0.3146,
+ "step": 11832
+ },
+ {
+ "epoch": 32.2425068119891,
+ "grad_norm": 3.861346483230591,
+ "learning_rate": 1.5840450299420548e-05,
+ "loss": 0.2552,
+ "step": 11833
+ },
+ {
+ "epoch": 32.24523160762943,
+ "grad_norm": 3.6858370304107666,
+ "learning_rate": 1.5839733937021738e-05,
+ "loss": 0.141,
+ "step": 11834
+ },
+ {
+ "epoch": 32.247956403269754,
+ "grad_norm": 4.316443920135498,
+ "learning_rate": 1.5839017529143276e-05,
+ "loss": 0.2308,
+ "step": 11835
+ },
+ {
+ "epoch": 32.25068119891008,
+ "grad_norm": 4.266711711883545,
+ "learning_rate": 1.5838301075790726e-05,
+ "loss": 0.1839,
+ "step": 11836
+ },
+ {
+ "epoch": 32.253405994550405,
+ "grad_norm": 4.042697429656982,
+ "learning_rate": 1.583758457696968e-05,
+ "loss": 0.2148,
+ "step": 11837
+ },
+ {
+ "epoch": 32.25613079019074,
+ "grad_norm": 4.316853046417236,
+ "learning_rate": 1.5836868032685714e-05,
+ "loss": 0.2342,
+ "step": 11838
+ },
+ {
+ "epoch": 32.258855585831064,
+ "grad_norm": 4.251140594482422,
+ "learning_rate": 1.5836151442944406e-05,
+ "loss": 0.1679,
+ "step": 11839
+ },
+ {
+ "epoch": 32.26158038147139,
+ "grad_norm": 4.031492710113525,
+ "learning_rate": 1.5835434807751337e-05,
+ "loss": 0.1927,
+ "step": 11840
+ },
+ {
+ "epoch": 32.264305177111716,
+ "grad_norm": 3.909283399581909,
+ "learning_rate": 1.583471812711209e-05,
+ "loss": 0.2339,
+ "step": 11841
+ },
+ {
+ "epoch": 32.26702997275204,
+ "grad_norm": 5.109302997589111,
+ "learning_rate": 1.583400140103225e-05,
+ "loss": 0.198,
+ "step": 11842
+ },
+ {
+ "epoch": 32.26975476839237,
+ "grad_norm": 4.239581108093262,
+ "learning_rate": 1.583328462951739e-05,
+ "loss": 0.3108,
+ "step": 11843
+ },
+ {
+ "epoch": 32.2724795640327,
+ "grad_norm": 3.958503484725952,
+ "learning_rate": 1.5832567812573097e-05,
+ "loss": 0.1562,
+ "step": 11844
+ },
+ {
+ "epoch": 32.275204359673026,
+ "grad_norm": 3.575629949569702,
+ "learning_rate": 1.5831850950204955e-05,
+ "loss": 0.1378,
+ "step": 11845
+ },
+ {
+ "epoch": 32.27792915531335,
+ "grad_norm": 3.512599229812622,
+ "learning_rate": 1.5831134042418547e-05,
+ "loss": 0.243,
+ "step": 11846
+ },
+ {
+ "epoch": 32.28065395095368,
+ "grad_norm": 4.051735877990723,
+ "learning_rate": 1.5830417089219454e-05,
+ "loss": 0.1587,
+ "step": 11847
+ },
+ {
+ "epoch": 32.283378746594,
+ "grad_norm": 5.114348411560059,
+ "learning_rate": 1.582970009061326e-05,
+ "loss": 0.2288,
+ "step": 11848
+ },
+ {
+ "epoch": 32.28610354223433,
+ "grad_norm": 3.2694883346557617,
+ "learning_rate": 1.582898304660555e-05,
+ "loss": 0.1055,
+ "step": 11849
+ },
+ {
+ "epoch": 32.28882833787466,
+ "grad_norm": 3.80755352973938,
+ "learning_rate": 1.5828265957201904e-05,
+ "loss": 0.181,
+ "step": 11850
+ },
+ {
+ "epoch": 32.29155313351499,
+ "grad_norm": 4.48306131362915,
+ "learning_rate": 1.5827548822407912e-05,
+ "loss": 0.2744,
+ "step": 11851
+ },
+ {
+ "epoch": 32.294277929155314,
+ "grad_norm": 3.6559784412384033,
+ "learning_rate": 1.582683164222916e-05,
+ "loss": 0.1877,
+ "step": 11852
+ },
+ {
+ "epoch": 32.29700272479564,
+ "grad_norm": 3.7460670471191406,
+ "learning_rate": 1.582611441667123e-05,
+ "loss": 0.197,
+ "step": 11853
+ },
+ {
+ "epoch": 32.299727520435965,
+ "grad_norm": 4.4384331703186035,
+ "learning_rate": 1.5825397145739704e-05,
+ "loss": 0.2103,
+ "step": 11854
+ },
+ {
+ "epoch": 32.30245231607629,
+ "grad_norm": 4.464642524719238,
+ "learning_rate": 1.582467982944018e-05,
+ "loss": 0.2041,
+ "step": 11855
+ },
+ {
+ "epoch": 32.305177111716624,
+ "grad_norm": 4.258744239807129,
+ "learning_rate": 1.582396246777823e-05,
+ "loss": 0.3758,
+ "step": 11856
+ },
+ {
+ "epoch": 32.30790190735695,
+ "grad_norm": 3.949324607849121,
+ "learning_rate": 1.582324506075945e-05,
+ "loss": 0.2975,
+ "step": 11857
+ },
+ {
+ "epoch": 32.310626702997276,
+ "grad_norm": 4.918122291564941,
+ "learning_rate": 1.5822527608389428e-05,
+ "loss": 0.1497,
+ "step": 11858
+ },
+ {
+ "epoch": 32.3133514986376,
+ "grad_norm": 3.958353281021118,
+ "learning_rate": 1.5821810110673744e-05,
+ "loss": 0.1275,
+ "step": 11859
+ },
+ {
+ "epoch": 32.31607629427793,
+ "grad_norm": 4.233720302581787,
+ "learning_rate": 1.5821092567617993e-05,
+ "loss": 0.2652,
+ "step": 11860
+ },
+ {
+ "epoch": 32.31880108991825,
+ "grad_norm": 4.846134185791016,
+ "learning_rate": 1.582037497922776e-05,
+ "loss": 0.3228,
+ "step": 11861
+ },
+ {
+ "epoch": 32.321525885558586,
+ "grad_norm": 4.263634204864502,
+ "learning_rate": 1.581965734550863e-05,
+ "loss": 0.167,
+ "step": 11862
+ },
+ {
+ "epoch": 32.32425068119891,
+ "grad_norm": 4.853062152862549,
+ "learning_rate": 1.5818939666466203e-05,
+ "loss": 0.209,
+ "step": 11863
+ },
+ {
+ "epoch": 32.32697547683924,
+ "grad_norm": 4.385557174682617,
+ "learning_rate": 1.5818221942106058e-05,
+ "loss": 0.1784,
+ "step": 11864
+ },
+ {
+ "epoch": 32.32970027247956,
+ "grad_norm": 4.336406707763672,
+ "learning_rate": 1.581750417243379e-05,
+ "loss": 0.2261,
+ "step": 11865
+ },
+ {
+ "epoch": 32.33242506811989,
+ "grad_norm": 4.348932266235352,
+ "learning_rate": 1.5816786357454984e-05,
+ "loss": 0.1792,
+ "step": 11866
+ },
+ {
+ "epoch": 32.335149863760215,
+ "grad_norm": 3.720475196838379,
+ "learning_rate": 1.5816068497175233e-05,
+ "loss": 0.1235,
+ "step": 11867
+ },
+ {
+ "epoch": 32.33787465940055,
+ "grad_norm": 4.469127178192139,
+ "learning_rate": 1.5815350591600124e-05,
+ "loss": 0.2426,
+ "step": 11868
+ },
+ {
+ "epoch": 32.34059945504087,
+ "grad_norm": 3.719444990158081,
+ "learning_rate": 1.5814632640735258e-05,
+ "loss": 0.2317,
+ "step": 11869
+ },
+ {
+ "epoch": 32.3433242506812,
+ "grad_norm": 4.647012710571289,
+ "learning_rate": 1.5813914644586218e-05,
+ "loss": 0.1444,
+ "step": 11870
+ },
+ {
+ "epoch": 32.346049046321525,
+ "grad_norm": 4.522265911102295,
+ "learning_rate": 1.5813196603158594e-05,
+ "loss": 0.2307,
+ "step": 11871
+ },
+ {
+ "epoch": 32.34877384196185,
+ "grad_norm": 4.397472381591797,
+ "learning_rate": 1.581247851645799e-05,
+ "loss": 0.226,
+ "step": 11872
+ },
+ {
+ "epoch": 32.35149863760218,
+ "grad_norm": 5.177058696746826,
+ "learning_rate": 1.5811760384489983e-05,
+ "loss": 0.1538,
+ "step": 11873
+ },
+ {
+ "epoch": 32.35422343324251,
+ "grad_norm": 3.8141393661499023,
+ "learning_rate": 1.581104220726018e-05,
+ "loss": 0.2436,
+ "step": 11874
+ },
+ {
+ "epoch": 32.356948228882835,
+ "grad_norm": 4.722202777862549,
+ "learning_rate": 1.581032398477416e-05,
+ "loss": 0.2492,
+ "step": 11875
+ },
+ {
+ "epoch": 32.35967302452316,
+ "grad_norm": 3.748182773590088,
+ "learning_rate": 1.580960571703753e-05,
+ "loss": 0.2167,
+ "step": 11876
+ },
+ {
+ "epoch": 32.36239782016349,
+ "grad_norm": 4.5502800941467285,
+ "learning_rate": 1.5808887404055876e-05,
+ "loss": 0.1851,
+ "step": 11877
+ },
+ {
+ "epoch": 32.36512261580381,
+ "grad_norm": 4.2080979347229,
+ "learning_rate": 1.580816904583479e-05,
+ "loss": 0.2446,
+ "step": 11878
+ },
+ {
+ "epoch": 32.36784741144414,
+ "grad_norm": 4.016653060913086,
+ "learning_rate": 1.5807450642379874e-05,
+ "loss": 0.1463,
+ "step": 11879
+ },
+ {
+ "epoch": 32.37057220708447,
+ "grad_norm": 4.4161376953125,
+ "learning_rate": 1.580673219369672e-05,
+ "loss": 0.1893,
+ "step": 11880
+ },
+ {
+ "epoch": 32.3732970027248,
+ "grad_norm": 4.444279193878174,
+ "learning_rate": 1.580601369979092e-05,
+ "loss": 0.2563,
+ "step": 11881
+ },
+ {
+ "epoch": 32.37602179836512,
+ "grad_norm": 3.8709843158721924,
+ "learning_rate": 1.5805295160668075e-05,
+ "loss": 0.2676,
+ "step": 11882
+ },
+ {
+ "epoch": 32.37874659400545,
+ "grad_norm": 3.9414005279541016,
+ "learning_rate": 1.5804576576333776e-05,
+ "loss": 0.2343,
+ "step": 11883
+ },
+ {
+ "epoch": 32.381471389645775,
+ "grad_norm": 3.839449167251587,
+ "learning_rate": 1.580385794679362e-05,
+ "loss": 0.3509,
+ "step": 11884
+ },
+ {
+ "epoch": 32.3841961852861,
+ "grad_norm": 4.587550640106201,
+ "learning_rate": 1.580313927205321e-05,
+ "loss": 0.2859,
+ "step": 11885
+ },
+ {
+ "epoch": 32.38692098092643,
+ "grad_norm": 4.208675384521484,
+ "learning_rate": 1.5802420552118133e-05,
+ "loss": 0.1727,
+ "step": 11886
+ },
+ {
+ "epoch": 32.38964577656676,
+ "grad_norm": 4.262181282043457,
+ "learning_rate": 1.5801701786993992e-05,
+ "loss": 0.4262,
+ "step": 11887
+ },
+ {
+ "epoch": 32.392370572207085,
+ "grad_norm": 4.35349702835083,
+ "learning_rate": 1.5800982976686387e-05,
+ "loss": 0.1957,
+ "step": 11888
+ },
+ {
+ "epoch": 32.39509536784741,
+ "grad_norm": 3.7446377277374268,
+ "learning_rate": 1.580026412120091e-05,
+ "loss": 0.2728,
+ "step": 11889
+ },
+ {
+ "epoch": 32.39782016348774,
+ "grad_norm": 3.999094009399414,
+ "learning_rate": 1.5799545220543168e-05,
+ "loss": 0.1842,
+ "step": 11890
+ },
+ {
+ "epoch": 32.40054495912806,
+ "grad_norm": 3.866485834121704,
+ "learning_rate": 1.579882627471875e-05,
+ "loss": 0.1678,
+ "step": 11891
+ },
+ {
+ "epoch": 32.403269754768395,
+ "grad_norm": 5.520777225494385,
+ "learning_rate": 1.5798107283733265e-05,
+ "loss": 0.1747,
+ "step": 11892
+ },
+ {
+ "epoch": 32.40599455040872,
+ "grad_norm": 4.352555274963379,
+ "learning_rate": 1.5797388247592304e-05,
+ "loss": 0.24,
+ "step": 11893
+ },
+ {
+ "epoch": 32.40871934604905,
+ "grad_norm": 4.524620056152344,
+ "learning_rate": 1.5796669166301468e-05,
+ "loss": 0.3588,
+ "step": 11894
+ },
+ {
+ "epoch": 32.41144414168937,
+ "grad_norm": 4.369819164276123,
+ "learning_rate": 1.579595003986636e-05,
+ "loss": 0.1808,
+ "step": 11895
+ },
+ {
+ "epoch": 32.4141689373297,
+ "grad_norm": 3.911511182785034,
+ "learning_rate": 1.5795230868292576e-05,
+ "loss": 0.2937,
+ "step": 11896
+ },
+ {
+ "epoch": 32.416893732970024,
+ "grad_norm": 3.41284441947937,
+ "learning_rate": 1.5794511651585725e-05,
+ "loss": 0.1972,
+ "step": 11897
+ },
+ {
+ "epoch": 32.41961852861036,
+ "grad_norm": 4.028480052947998,
+ "learning_rate": 1.5793792389751402e-05,
+ "loss": 0.1828,
+ "step": 11898
+ },
+ {
+ "epoch": 32.42234332425068,
+ "grad_norm": 3.373119592666626,
+ "learning_rate": 1.5793073082795213e-05,
+ "loss": 0.2001,
+ "step": 11899
+ },
+ {
+ "epoch": 32.42506811989101,
+ "grad_norm": 4.44292688369751,
+ "learning_rate": 1.5792353730722757e-05,
+ "loss": 0.2027,
+ "step": 11900
+ },
+ {
+ "epoch": 32.427792915531334,
+ "grad_norm": 10.215131759643555,
+ "learning_rate": 1.5791634333539632e-05,
+ "loss": 0.2759,
+ "step": 11901
+ },
+ {
+ "epoch": 32.43051771117166,
+ "grad_norm": 4.755099296569824,
+ "learning_rate": 1.5790914891251448e-05,
+ "loss": 0.3314,
+ "step": 11902
+ },
+ {
+ "epoch": 32.433242506811986,
+ "grad_norm": 3.617554187774658,
+ "learning_rate": 1.5790195403863808e-05,
+ "loss": 0.2519,
+ "step": 11903
+ },
+ {
+ "epoch": 32.43596730245232,
+ "grad_norm": 3.870229959487915,
+ "learning_rate": 1.578947587138231e-05,
+ "loss": 0.1697,
+ "step": 11904
+ },
+ {
+ "epoch": 32.438692098092645,
+ "grad_norm": 3.858883857727051,
+ "learning_rate": 1.578875629381256e-05,
+ "loss": 0.1848,
+ "step": 11905
+ },
+ {
+ "epoch": 32.44141689373297,
+ "grad_norm": 7.615336894989014,
+ "learning_rate": 1.5788036671160163e-05,
+ "loss": 0.1749,
+ "step": 11906
+ },
+ {
+ "epoch": 32.444141689373296,
+ "grad_norm": 3.6982510089874268,
+ "learning_rate": 1.578731700343072e-05,
+ "loss": 0.2533,
+ "step": 11907
+ },
+ {
+ "epoch": 32.44686648501362,
+ "grad_norm": 3.3248324394226074,
+ "learning_rate": 1.5786597290629843e-05,
+ "loss": 0.1275,
+ "step": 11908
+ },
+ {
+ "epoch": 32.44959128065395,
+ "grad_norm": 4.114260196685791,
+ "learning_rate": 1.5785877532763132e-05,
+ "loss": 0.2322,
+ "step": 11909
+ },
+ {
+ "epoch": 32.45231607629428,
+ "grad_norm": 3.8239903450012207,
+ "learning_rate": 1.578515772983619e-05,
+ "loss": 0.1879,
+ "step": 11910
+ },
+ {
+ "epoch": 32.45504087193461,
+ "grad_norm": 4.313100814819336,
+ "learning_rate": 1.5784437881854626e-05,
+ "loss": 0.2636,
+ "step": 11911
+ },
+ {
+ "epoch": 32.45776566757493,
+ "grad_norm": 4.970459461212158,
+ "learning_rate": 1.578371798882405e-05,
+ "loss": 0.2558,
+ "step": 11912
+ },
+ {
+ "epoch": 32.46049046321526,
+ "grad_norm": 3.6384544372558594,
+ "learning_rate": 1.578299805075006e-05,
+ "loss": 0.2986,
+ "step": 11913
+ },
+ {
+ "epoch": 32.463215258855584,
+ "grad_norm": 4.295863628387451,
+ "learning_rate": 1.5782278067638267e-05,
+ "loss": 0.2667,
+ "step": 11914
+ },
+ {
+ "epoch": 32.46594005449591,
+ "grad_norm": 4.037357807159424,
+ "learning_rate": 1.5781558039494283e-05,
+ "loss": 0.143,
+ "step": 11915
+ },
+ {
+ "epoch": 32.46866485013624,
+ "grad_norm": 5.539492130279541,
+ "learning_rate": 1.578083796632371e-05,
+ "loss": 0.2262,
+ "step": 11916
+ },
+ {
+ "epoch": 32.47138964577657,
+ "grad_norm": 3.6543030738830566,
+ "learning_rate": 1.5780117848132154e-05,
+ "loss": 0.2702,
+ "step": 11917
+ },
+ {
+ "epoch": 32.474114441416894,
+ "grad_norm": 4.003937721252441,
+ "learning_rate": 1.5779397684925226e-05,
+ "loss": 0.1731,
+ "step": 11918
+ },
+ {
+ "epoch": 32.47683923705722,
+ "grad_norm": 3.9434401988983154,
+ "learning_rate": 1.5778677476708536e-05,
+ "loss": 0.1375,
+ "step": 11919
+ },
+ {
+ "epoch": 32.479564032697546,
+ "grad_norm": 3.9064178466796875,
+ "learning_rate": 1.577795722348769e-05,
+ "loss": 0.1412,
+ "step": 11920
+ },
+ {
+ "epoch": 32.48228882833787,
+ "grad_norm": 4.8439621925354,
+ "learning_rate": 1.5777236925268306e-05,
+ "loss": 0.2498,
+ "step": 11921
+ },
+ {
+ "epoch": 32.485013623978205,
+ "grad_norm": 4.430334091186523,
+ "learning_rate": 1.577651658205598e-05,
+ "loss": 0.3785,
+ "step": 11922
+ },
+ {
+ "epoch": 32.48773841961853,
+ "grad_norm": 3.8473896980285645,
+ "learning_rate": 1.5775796193856332e-05,
+ "loss": 0.1343,
+ "step": 11923
+ },
+ {
+ "epoch": 32.490463215258856,
+ "grad_norm": 3.6651551723480225,
+ "learning_rate": 1.5775075760674966e-05,
+ "loss": 0.1705,
+ "step": 11924
+ },
+ {
+ "epoch": 32.49318801089918,
+ "grad_norm": 4.143070220947266,
+ "learning_rate": 1.57743552825175e-05,
+ "loss": 0.2112,
+ "step": 11925
+ },
+ {
+ "epoch": 32.49591280653951,
+ "grad_norm": 5.3032073974609375,
+ "learning_rate": 1.5773634759389537e-05,
+ "loss": 0.1547,
+ "step": 11926
+ },
+ {
+ "epoch": 32.49863760217983,
+ "grad_norm": 3.834362506866455,
+ "learning_rate": 1.5772914191296697e-05,
+ "loss": 0.1821,
+ "step": 11927
+ },
+ {
+ "epoch": 32.50136239782017,
+ "grad_norm": 4.580528259277344,
+ "learning_rate": 1.577219357824458e-05,
+ "loss": 0.2096,
+ "step": 11928
+ },
+ {
+ "epoch": 32.50408719346049,
+ "grad_norm": 3.5919129848480225,
+ "learning_rate": 1.577147292023881e-05,
+ "loss": 0.3219,
+ "step": 11929
+ },
+ {
+ "epoch": 32.50681198910082,
+ "grad_norm": 3.5160129070281982,
+ "learning_rate": 1.5770752217284995e-05,
+ "loss": 0.2515,
+ "step": 11930
+ },
+ {
+ "epoch": 32.509536784741144,
+ "grad_norm": 3.8159353733062744,
+ "learning_rate": 1.5770031469388747e-05,
+ "loss": 0.3258,
+ "step": 11931
+ },
+ {
+ "epoch": 32.51226158038147,
+ "grad_norm": 4.179819107055664,
+ "learning_rate": 1.576931067655568e-05,
+ "loss": 0.1945,
+ "step": 11932
+ },
+ {
+ "epoch": 32.514986376021795,
+ "grad_norm": 3.8886542320251465,
+ "learning_rate": 1.5768589838791405e-05,
+ "loss": 0.1226,
+ "step": 11933
+ },
+ {
+ "epoch": 32.51771117166213,
+ "grad_norm": 4.560035705566406,
+ "learning_rate": 1.576786895610154e-05,
+ "loss": 0.2587,
+ "step": 11934
+ },
+ {
+ "epoch": 32.520435967302454,
+ "grad_norm": 4.033864974975586,
+ "learning_rate": 1.5767148028491696e-05,
+ "loss": 0.1746,
+ "step": 11935
+ },
+ {
+ "epoch": 32.52316076294278,
+ "grad_norm": 4.027731418609619,
+ "learning_rate": 1.5766427055967488e-05,
+ "loss": 0.1506,
+ "step": 11936
+ },
+ {
+ "epoch": 32.525885558583106,
+ "grad_norm": 4.393682956695557,
+ "learning_rate": 1.5765706038534534e-05,
+ "loss": 0.2973,
+ "step": 11937
+ },
+ {
+ "epoch": 32.52861035422343,
+ "grad_norm": 4.9071431159973145,
+ "learning_rate": 1.5764984976198447e-05,
+ "loss": 0.2886,
+ "step": 11938
+ },
+ {
+ "epoch": 32.53133514986376,
+ "grad_norm": 4.5589704513549805,
+ "learning_rate": 1.576426386896484e-05,
+ "loss": 0.2505,
+ "step": 11939
+ },
+ {
+ "epoch": 32.53405994550409,
+ "grad_norm": 3.668513536453247,
+ "learning_rate": 1.5763542716839332e-05,
+ "loss": 0.1562,
+ "step": 11940
+ },
+ {
+ "epoch": 32.536784741144416,
+ "grad_norm": 4.60407018661499,
+ "learning_rate": 1.576282151982754e-05,
+ "loss": 0.1784,
+ "step": 11941
+ },
+ {
+ "epoch": 32.53950953678474,
+ "grad_norm": 4.403726100921631,
+ "learning_rate": 1.576210027793508e-05,
+ "loss": 0.3616,
+ "step": 11942
+ },
+ {
+ "epoch": 32.54223433242507,
+ "grad_norm": 5.027153015136719,
+ "learning_rate": 1.5761378991167568e-05,
+ "loss": 0.3064,
+ "step": 11943
+ },
+ {
+ "epoch": 32.54495912806539,
+ "grad_norm": 5.124536991119385,
+ "learning_rate": 1.5760657659530622e-05,
+ "loss": 0.2676,
+ "step": 11944
+ },
+ {
+ "epoch": 32.54768392370572,
+ "grad_norm": 3.6511473655700684,
+ "learning_rate": 1.575993628302986e-05,
+ "loss": 0.2157,
+ "step": 11945
+ },
+ {
+ "epoch": 32.55040871934605,
+ "grad_norm": 4.727812767028809,
+ "learning_rate": 1.5759214861670894e-05,
+ "loss": 0.2208,
+ "step": 11946
+ },
+ {
+ "epoch": 32.55313351498638,
+ "grad_norm": 4.6300578117370605,
+ "learning_rate": 1.5758493395459353e-05,
+ "loss": 0.2439,
+ "step": 11947
+ },
+ {
+ "epoch": 32.555858310626704,
+ "grad_norm": 3.5169692039489746,
+ "learning_rate": 1.5757771884400853e-05,
+ "loss": 0.2143,
+ "step": 11948
+ },
+ {
+ "epoch": 32.55858310626703,
+ "grad_norm": 4.504608631134033,
+ "learning_rate": 1.5757050328501005e-05,
+ "loss": 0.273,
+ "step": 11949
+ },
+ {
+ "epoch": 32.561307901907355,
+ "grad_norm": 4.257322311401367,
+ "learning_rate": 1.5756328727765443e-05,
+ "loss": 0.2117,
+ "step": 11950
+ },
+ {
+ "epoch": 32.56403269754768,
+ "grad_norm": 4.164383888244629,
+ "learning_rate": 1.575560708219977e-05,
+ "loss": 0.2467,
+ "step": 11951
+ },
+ {
+ "epoch": 32.566757493188014,
+ "grad_norm": 5.031408309936523,
+ "learning_rate": 1.5754885391809618e-05,
+ "loss": 0.3217,
+ "step": 11952
+ },
+ {
+ "epoch": 32.56948228882834,
+ "grad_norm": 4.636307716369629,
+ "learning_rate": 1.5754163656600602e-05,
+ "loss": 0.1929,
+ "step": 11953
+ },
+ {
+ "epoch": 32.572207084468666,
+ "grad_norm": 4.676330089569092,
+ "learning_rate": 1.5753441876578347e-05,
+ "loss": 0.2653,
+ "step": 11954
+ },
+ {
+ "epoch": 32.57493188010899,
+ "grad_norm": 6.663448333740234,
+ "learning_rate": 1.575272005174847e-05,
+ "loss": 0.2716,
+ "step": 11955
+ },
+ {
+ "epoch": 32.57765667574932,
+ "grad_norm": 4.6932759284973145,
+ "learning_rate": 1.5751998182116595e-05,
+ "loss": 0.2754,
+ "step": 11956
+ },
+ {
+ "epoch": 32.58038147138964,
+ "grad_norm": 3.189453601837158,
+ "learning_rate": 1.575127626768834e-05,
+ "loss": 0.2317,
+ "step": 11957
+ },
+ {
+ "epoch": 32.583106267029976,
+ "grad_norm": 3.8600680828094482,
+ "learning_rate": 1.5750554308469337e-05,
+ "loss": 0.1548,
+ "step": 11958
+ },
+ {
+ "epoch": 32.5858310626703,
+ "grad_norm": 3.9114480018615723,
+ "learning_rate": 1.5749832304465197e-05,
+ "loss": 0.1971,
+ "step": 11959
+ },
+ {
+ "epoch": 32.58855585831063,
+ "grad_norm": 4.248748302459717,
+ "learning_rate": 1.574911025568155e-05,
+ "loss": 0.1831,
+ "step": 11960
+ },
+ {
+ "epoch": 32.59128065395095,
+ "grad_norm": 4.589288711547852,
+ "learning_rate": 1.5748388162124017e-05,
+ "loss": 0.2047,
+ "step": 11961
+ },
+ {
+ "epoch": 32.59400544959128,
+ "grad_norm": 5.132267951965332,
+ "learning_rate": 1.5747666023798226e-05,
+ "loss": 0.2029,
+ "step": 11962
+ },
+ {
+ "epoch": 32.596730245231605,
+ "grad_norm": 3.855661392211914,
+ "learning_rate": 1.5746943840709786e-05,
+ "loss": 0.1293,
+ "step": 11963
+ },
+ {
+ "epoch": 32.59945504087194,
+ "grad_norm": 3.7804558277130127,
+ "learning_rate": 1.5746221612864342e-05,
+ "loss": 0.1946,
+ "step": 11964
+ },
+ {
+ "epoch": 32.60217983651226,
+ "grad_norm": 3.9552547931671143,
+ "learning_rate": 1.5745499340267508e-05,
+ "loss": 0.3301,
+ "step": 11965
+ },
+ {
+ "epoch": 32.60490463215259,
+ "grad_norm": 4.611697673797607,
+ "learning_rate": 1.574477702292491e-05,
+ "loss": 0.2897,
+ "step": 11966
+ },
+ {
+ "epoch": 32.607629427792915,
+ "grad_norm": 3.7894444465637207,
+ "learning_rate": 1.574405466084217e-05,
+ "loss": 0.1392,
+ "step": 11967
+ },
+ {
+ "epoch": 32.61035422343324,
+ "grad_norm": 3.6013283729553223,
+ "learning_rate": 1.574333225402492e-05,
+ "loss": 0.1543,
+ "step": 11968
+ },
+ {
+ "epoch": 32.61307901907357,
+ "grad_norm": 3.7538623809814453,
+ "learning_rate": 1.5742609802478782e-05,
+ "loss": 0.1154,
+ "step": 11969
+ },
+ {
+ "epoch": 32.6158038147139,
+ "grad_norm": 4.621149063110352,
+ "learning_rate": 1.5741887306209383e-05,
+ "loss": 0.3123,
+ "step": 11970
+ },
+ {
+ "epoch": 32.618528610354225,
+ "grad_norm": 4.133847236633301,
+ "learning_rate": 1.574116476522235e-05,
+ "loss": 0.2058,
+ "step": 11971
+ },
+ {
+ "epoch": 32.62125340599455,
+ "grad_norm": 4.260160446166992,
+ "learning_rate": 1.5740442179523315e-05,
+ "loss": 0.2249,
+ "step": 11972
+ },
+ {
+ "epoch": 32.62397820163488,
+ "grad_norm": 3.4492599964141846,
+ "learning_rate": 1.5739719549117897e-05,
+ "loss": 0.2303,
+ "step": 11973
+ },
+ {
+ "epoch": 32.6267029972752,
+ "grad_norm": 4.579864025115967,
+ "learning_rate": 1.5738996874011728e-05,
+ "loss": 0.2199,
+ "step": 11974
+ },
+ {
+ "epoch": 32.62942779291553,
+ "grad_norm": 3.4036171436309814,
+ "learning_rate": 1.573827415421044e-05,
+ "loss": 0.2488,
+ "step": 11975
+ },
+ {
+ "epoch": 32.63215258855586,
+ "grad_norm": 4.294132709503174,
+ "learning_rate": 1.5737551389719655e-05,
+ "loss": 0.2491,
+ "step": 11976
+ },
+ {
+ "epoch": 32.63487738419619,
+ "grad_norm": 3.5100276470184326,
+ "learning_rate": 1.5736828580545003e-05,
+ "loss": 0.1481,
+ "step": 11977
+ },
+ {
+ "epoch": 32.63760217983651,
+ "grad_norm": 4.357295989990234,
+ "learning_rate": 1.5736105726692117e-05,
+ "loss": 0.1775,
+ "step": 11978
+ },
+ {
+ "epoch": 32.64032697547684,
+ "grad_norm": 3.748542070388794,
+ "learning_rate": 1.5735382828166623e-05,
+ "loss": 0.3969,
+ "step": 11979
+ },
+ {
+ "epoch": 32.643051771117165,
+ "grad_norm": 3.988121747970581,
+ "learning_rate": 1.573465988497415e-05,
+ "loss": 0.3205,
+ "step": 11980
+ },
+ {
+ "epoch": 32.64577656675749,
+ "grad_norm": 4.998047351837158,
+ "learning_rate": 1.5733936897120333e-05,
+ "loss": 0.2098,
+ "step": 11981
+ },
+ {
+ "epoch": 32.64850136239782,
+ "grad_norm": 5.146492004394531,
+ "learning_rate": 1.57332138646108e-05,
+ "loss": 0.1949,
+ "step": 11982
+ },
+ {
+ "epoch": 32.65122615803815,
+ "grad_norm": 3.724456787109375,
+ "learning_rate": 1.573249078745118e-05,
+ "loss": 0.2709,
+ "step": 11983
+ },
+ {
+ "epoch": 32.653950953678475,
+ "grad_norm": 4.431726932525635,
+ "learning_rate": 1.573176766564711e-05,
+ "loss": 0.2313,
+ "step": 11984
+ },
+ {
+ "epoch": 32.6566757493188,
+ "grad_norm": 4.149693489074707,
+ "learning_rate": 1.5731044499204213e-05,
+ "loss": 0.2468,
+ "step": 11985
+ },
+ {
+ "epoch": 32.65940054495913,
+ "grad_norm": 3.6718156337738037,
+ "learning_rate": 1.5730321288128128e-05,
+ "loss": 0.1298,
+ "step": 11986
+ },
+ {
+ "epoch": 32.66212534059945,
+ "grad_norm": 4.619180679321289,
+ "learning_rate": 1.5729598032424486e-05,
+ "loss": 0.2724,
+ "step": 11987
+ },
+ {
+ "epoch": 32.664850136239785,
+ "grad_norm": 3.586552143096924,
+ "learning_rate": 1.572887473209892e-05,
+ "loss": 0.1764,
+ "step": 11988
+ },
+ {
+ "epoch": 32.66757493188011,
+ "grad_norm": 5.2164483070373535,
+ "learning_rate": 1.572815138715706e-05,
+ "loss": 0.3042,
+ "step": 11989
+ },
+ {
+ "epoch": 32.67029972752044,
+ "grad_norm": 3.3664491176605225,
+ "learning_rate": 1.5727427997604545e-05,
+ "loss": 0.141,
+ "step": 11990
+ },
+ {
+ "epoch": 32.67302452316076,
+ "grad_norm": 4.4897332191467285,
+ "learning_rate": 1.5726704563446998e-05,
+ "loss": 0.1569,
+ "step": 11991
+ },
+ {
+ "epoch": 32.67574931880109,
+ "grad_norm": 4.519774436950684,
+ "learning_rate": 1.5725981084690067e-05,
+ "loss": 0.1632,
+ "step": 11992
+ },
+ {
+ "epoch": 32.678474114441414,
+ "grad_norm": 3.773897171020508,
+ "learning_rate": 1.5725257561339375e-05,
+ "loss": 0.2744,
+ "step": 11993
+ },
+ {
+ "epoch": 32.68119891008175,
+ "grad_norm": 3.876598596572876,
+ "learning_rate": 1.5724533993400564e-05,
+ "loss": 0.1693,
+ "step": 11994
+ },
+ {
+ "epoch": 32.68392370572207,
+ "grad_norm": 4.510540008544922,
+ "learning_rate": 1.572381038087927e-05,
+ "loss": 0.3618,
+ "step": 11995
+ },
+ {
+ "epoch": 32.6866485013624,
+ "grad_norm": 8.503240585327148,
+ "learning_rate": 1.5723086723781116e-05,
+ "loss": 0.268,
+ "step": 11996
+ },
+ {
+ "epoch": 32.689373297002724,
+ "grad_norm": 3.4333555698394775,
+ "learning_rate": 1.572236302211175e-05,
+ "loss": 0.1238,
+ "step": 11997
+ },
+ {
+ "epoch": 32.69209809264305,
+ "grad_norm": 4.48297119140625,
+ "learning_rate": 1.5721639275876807e-05,
+ "loss": 0.1878,
+ "step": 11998
+ },
+ {
+ "epoch": 32.694822888283376,
+ "grad_norm": 4.360779762268066,
+ "learning_rate": 1.5720915485081923e-05,
+ "loss": 0.1499,
+ "step": 11999
+ },
+ {
+ "epoch": 32.69754768392371,
+ "grad_norm": 4.889499664306641,
+ "learning_rate": 1.572019164973273e-05,
+ "loss": 0.2421,
+ "step": 12000
+ },
+ {
+ "epoch": 32.700272479564035,
+ "grad_norm": 3.514737844467163,
+ "learning_rate": 1.571946776983487e-05,
+ "loss": 0.1429,
+ "step": 12001
+ },
+ {
+ "epoch": 32.70299727520436,
+ "grad_norm": 3.8689756393432617,
+ "learning_rate": 1.571874384539398e-05,
+ "loss": 0.1987,
+ "step": 12002
+ },
+ {
+ "epoch": 32.705722070844686,
+ "grad_norm": 3.891524314880371,
+ "learning_rate": 1.5718019876415696e-05,
+ "loss": 0.3281,
+ "step": 12003
+ },
+ {
+ "epoch": 32.70844686648501,
+ "grad_norm": 3.8140971660614014,
+ "learning_rate": 1.571729586290566e-05,
+ "loss": 0.2283,
+ "step": 12004
+ },
+ {
+ "epoch": 32.71117166212534,
+ "grad_norm": 4.107378959655762,
+ "learning_rate": 1.5716571804869502e-05,
+ "loss": 0.1984,
+ "step": 12005
+ },
+ {
+ "epoch": 32.71389645776567,
+ "grad_norm": 8.490388870239258,
+ "learning_rate": 1.571584770231287e-05,
+ "loss": 0.2947,
+ "step": 12006
+ },
+ {
+ "epoch": 32.716621253406,
+ "grad_norm": 10.44328498840332,
+ "learning_rate": 1.5715123555241397e-05,
+ "loss": 0.2524,
+ "step": 12007
+ },
+ {
+ "epoch": 32.71934604904632,
+ "grad_norm": 3.989579677581787,
+ "learning_rate": 1.571439936366073e-05,
+ "loss": 0.2901,
+ "step": 12008
+ },
+ {
+ "epoch": 32.72207084468665,
+ "grad_norm": 4.183882236480713,
+ "learning_rate": 1.5713675127576504e-05,
+ "loss": 0.3547,
+ "step": 12009
+ },
+ {
+ "epoch": 32.724795640326974,
+ "grad_norm": 4.206899642944336,
+ "learning_rate": 1.571295084699436e-05,
+ "loss": 0.2711,
+ "step": 12010
+ },
+ {
+ "epoch": 32.7275204359673,
+ "grad_norm": 5.188673496246338,
+ "learning_rate": 1.5712226521919936e-05,
+ "loss": 0.2624,
+ "step": 12011
+ },
+ {
+ "epoch": 32.73024523160763,
+ "grad_norm": 6.960591793060303,
+ "learning_rate": 1.5711502152358878e-05,
+ "loss": 0.1719,
+ "step": 12012
+ },
+ {
+ "epoch": 32.73297002724796,
+ "grad_norm": 3.590029239654541,
+ "learning_rate": 1.5710777738316827e-05,
+ "loss": 0.1808,
+ "step": 12013
+ },
+ {
+ "epoch": 32.735694822888284,
+ "grad_norm": 4.6145920753479,
+ "learning_rate": 1.5710053279799416e-05,
+ "loss": 0.2544,
+ "step": 12014
+ },
+ {
+ "epoch": 32.73841961852861,
+ "grad_norm": 5.035254001617432,
+ "learning_rate": 1.5709328776812302e-05,
+ "loss": 0.2393,
+ "step": 12015
+ },
+ {
+ "epoch": 32.741144414168936,
+ "grad_norm": 5.673764228820801,
+ "learning_rate": 1.5708604229361115e-05,
+ "loss": 0.2311,
+ "step": 12016
+ },
+ {
+ "epoch": 32.74386920980926,
+ "grad_norm": 4.867123603820801,
+ "learning_rate": 1.5707879637451502e-05,
+ "loss": 0.2614,
+ "step": 12017
+ },
+ {
+ "epoch": 32.746594005449595,
+ "grad_norm": 4.365541934967041,
+ "learning_rate": 1.5707155001089106e-05,
+ "loss": 0.2863,
+ "step": 12018
+ },
+ {
+ "epoch": 32.74931880108992,
+ "grad_norm": 4.399753570556641,
+ "learning_rate": 1.570643032027957e-05,
+ "loss": 0.1817,
+ "step": 12019
+ },
+ {
+ "epoch": 32.752043596730246,
+ "grad_norm": 4.294652938842773,
+ "learning_rate": 1.570570559502854e-05,
+ "loss": 0.209,
+ "step": 12020
+ },
+ {
+ "epoch": 32.75476839237057,
+ "grad_norm": 4.378859996795654,
+ "learning_rate": 1.570498082534166e-05,
+ "loss": 0.3065,
+ "step": 12021
+ },
+ {
+ "epoch": 32.7574931880109,
+ "grad_norm": 4.858389854431152,
+ "learning_rate": 1.570425601122457e-05,
+ "loss": 0.2378,
+ "step": 12022
+ },
+ {
+ "epoch": 32.76021798365122,
+ "grad_norm": 4.137174129486084,
+ "learning_rate": 1.5703531152682918e-05,
+ "loss": 0.3088,
+ "step": 12023
+ },
+ {
+ "epoch": 32.762942779291556,
+ "grad_norm": 4.594271659851074,
+ "learning_rate": 1.570280624972235e-05,
+ "loss": 0.2674,
+ "step": 12024
+ },
+ {
+ "epoch": 32.76566757493188,
+ "grad_norm": 5.481046676635742,
+ "learning_rate": 1.570208130234851e-05,
+ "loss": 0.1948,
+ "step": 12025
+ },
+ {
+ "epoch": 32.76839237057221,
+ "grad_norm": 4.0230913162231445,
+ "learning_rate": 1.5701356310567044e-05,
+ "loss": 0.1467,
+ "step": 12026
+ },
+ {
+ "epoch": 32.771117166212534,
+ "grad_norm": 4.950342655181885,
+ "learning_rate": 1.57006312743836e-05,
+ "loss": 0.3604,
+ "step": 12027
+ },
+ {
+ "epoch": 32.77384196185286,
+ "grad_norm": 4.129463195800781,
+ "learning_rate": 1.569990619380382e-05,
+ "loss": 0.1376,
+ "step": 12028
+ },
+ {
+ "epoch": 32.776566757493185,
+ "grad_norm": 4.379932403564453,
+ "learning_rate": 1.5699181068833355e-05,
+ "loss": 0.1822,
+ "step": 12029
+ },
+ {
+ "epoch": 32.77929155313352,
+ "grad_norm": 4.8274688720703125,
+ "learning_rate": 1.5698455899477852e-05,
+ "loss": 0.2048,
+ "step": 12030
+ },
+ {
+ "epoch": 32.782016348773844,
+ "grad_norm": 4.060813903808594,
+ "learning_rate": 1.569773068574296e-05,
+ "loss": 0.1572,
+ "step": 12031
+ },
+ {
+ "epoch": 32.78474114441417,
+ "grad_norm": 4.8249006271362305,
+ "learning_rate": 1.5697005427634324e-05,
+ "loss": 0.2049,
+ "step": 12032
+ },
+ {
+ "epoch": 32.787465940054496,
+ "grad_norm": 4.721830368041992,
+ "learning_rate": 1.569628012515759e-05,
+ "loss": 0.2775,
+ "step": 12033
+ },
+ {
+ "epoch": 32.79019073569482,
+ "grad_norm": 9.899148941040039,
+ "learning_rate": 1.5695554778318413e-05,
+ "loss": 0.3561,
+ "step": 12034
+ },
+ {
+ "epoch": 32.79291553133515,
+ "grad_norm": 10.07511043548584,
+ "learning_rate": 1.5694829387122436e-05,
+ "loss": 0.1438,
+ "step": 12035
+ },
+ {
+ "epoch": 32.79564032697548,
+ "grad_norm": 5.518694877624512,
+ "learning_rate": 1.5694103951575313e-05,
+ "loss": 0.2902,
+ "step": 12036
+ },
+ {
+ "epoch": 32.798365122615806,
+ "grad_norm": 5.212947368621826,
+ "learning_rate": 1.5693378471682688e-05,
+ "loss": 0.1902,
+ "step": 12037
+ },
+ {
+ "epoch": 32.80108991825613,
+ "grad_norm": 4.274571895599365,
+ "learning_rate": 1.569265294745022e-05,
+ "loss": 0.3263,
+ "step": 12038
+ },
+ {
+ "epoch": 32.80381471389646,
+ "grad_norm": 6.353625297546387,
+ "learning_rate": 1.5691927378883555e-05,
+ "loss": 0.2645,
+ "step": 12039
+ },
+ {
+ "epoch": 32.80653950953678,
+ "grad_norm": 4.669061660766602,
+ "learning_rate": 1.5691201765988336e-05,
+ "loss": 0.2642,
+ "step": 12040
+ },
+ {
+ "epoch": 32.80926430517711,
+ "grad_norm": 5.159091949462891,
+ "learning_rate": 1.5690476108770224e-05,
+ "loss": 0.3665,
+ "step": 12041
+ },
+ {
+ "epoch": 32.81198910081744,
+ "grad_norm": 4.507942199707031,
+ "learning_rate": 1.568975040723487e-05,
+ "loss": 0.2001,
+ "step": 12042
+ },
+ {
+ "epoch": 32.81471389645777,
+ "grad_norm": 5.624659538269043,
+ "learning_rate": 1.568902466138792e-05,
+ "loss": 0.2633,
+ "step": 12043
+ },
+ {
+ "epoch": 32.817438692098094,
+ "grad_norm": 4.6467719078063965,
+ "learning_rate": 1.5688298871235028e-05,
+ "loss": 0.2694,
+ "step": 12044
+ },
+ {
+ "epoch": 32.82016348773842,
+ "grad_norm": 4.703892230987549,
+ "learning_rate": 1.568757303678185e-05,
+ "loss": 0.2039,
+ "step": 12045
+ },
+ {
+ "epoch": 32.822888283378745,
+ "grad_norm": 4.373074054718018,
+ "learning_rate": 1.5686847158034036e-05,
+ "loss": 0.2403,
+ "step": 12046
+ },
+ {
+ "epoch": 32.82561307901907,
+ "grad_norm": 4.848725318908691,
+ "learning_rate": 1.568612123499724e-05,
+ "loss": 0.2662,
+ "step": 12047
+ },
+ {
+ "epoch": 32.828337874659404,
+ "grad_norm": 4.005350589752197,
+ "learning_rate": 1.5685395267677113e-05,
+ "loss": 0.2477,
+ "step": 12048
+ },
+ {
+ "epoch": 32.83106267029973,
+ "grad_norm": 5.415238857269287,
+ "learning_rate": 1.568466925607931e-05,
+ "loss": 0.2851,
+ "step": 12049
+ },
+ {
+ "epoch": 32.833787465940055,
+ "grad_norm": 5.936225891113281,
+ "learning_rate": 1.5683943200209486e-05,
+ "loss": 0.1929,
+ "step": 12050
+ },
+ {
+ "epoch": 32.83651226158038,
+ "grad_norm": 4.823857307434082,
+ "learning_rate": 1.56832171000733e-05,
+ "loss": 0.2573,
+ "step": 12051
+ },
+ {
+ "epoch": 32.83923705722071,
+ "grad_norm": 4.324443817138672,
+ "learning_rate": 1.5682490955676395e-05,
+ "loss": 0.1877,
+ "step": 12052
+ },
+ {
+ "epoch": 32.84196185286103,
+ "grad_norm": 4.190413951873779,
+ "learning_rate": 1.5681764767024435e-05,
+ "loss": 0.2517,
+ "step": 12053
+ },
+ {
+ "epoch": 32.844686648501366,
+ "grad_norm": 3.9925661087036133,
+ "learning_rate": 1.5681038534123078e-05,
+ "loss": 0.3053,
+ "step": 12054
+ },
+ {
+ "epoch": 32.84741144414169,
+ "grad_norm": 4.44297981262207,
+ "learning_rate": 1.5680312256977974e-05,
+ "loss": 0.185,
+ "step": 12055
+ },
+ {
+ "epoch": 32.85013623978202,
+ "grad_norm": 4.5818681716918945,
+ "learning_rate": 1.5679585935594782e-05,
+ "loss": 0.3441,
+ "step": 12056
+ },
+ {
+ "epoch": 32.85286103542234,
+ "grad_norm": 3.9314095973968506,
+ "learning_rate": 1.5678859569979154e-05,
+ "loss": 0.1693,
+ "step": 12057
+ },
+ {
+ "epoch": 32.85558583106267,
+ "grad_norm": 4.226455211639404,
+ "learning_rate": 1.5678133160136754e-05,
+ "loss": 0.2572,
+ "step": 12058
+ },
+ {
+ "epoch": 32.858310626702995,
+ "grad_norm": 4.827029705047607,
+ "learning_rate": 1.5677406706073234e-05,
+ "loss": 0.2334,
+ "step": 12059
+ },
+ {
+ "epoch": 32.86103542234333,
+ "grad_norm": 4.032408714294434,
+ "learning_rate": 1.5676680207794252e-05,
+ "loss": 0.2501,
+ "step": 12060
+ },
+ {
+ "epoch": 32.86376021798365,
+ "grad_norm": 3.9352893829345703,
+ "learning_rate": 1.567595366530547e-05,
+ "loss": 0.4682,
+ "step": 12061
+ },
+ {
+ "epoch": 32.86648501362398,
+ "grad_norm": 4.284492015838623,
+ "learning_rate": 1.567522707861254e-05,
+ "loss": 0.4099,
+ "step": 12062
+ },
+ {
+ "epoch": 32.869209809264305,
+ "grad_norm": 5.576635360717773,
+ "learning_rate": 1.567450044772113e-05,
+ "loss": 0.2771,
+ "step": 12063
+ },
+ {
+ "epoch": 32.87193460490463,
+ "grad_norm": 4.8764729499816895,
+ "learning_rate": 1.5673773772636887e-05,
+ "loss": 0.2259,
+ "step": 12064
+ },
+ {
+ "epoch": 32.87465940054496,
+ "grad_norm": 3.631629228591919,
+ "learning_rate": 1.567304705336548e-05,
+ "loss": 0.1293,
+ "step": 12065
+ },
+ {
+ "epoch": 32.87738419618529,
+ "grad_norm": 4.416547775268555,
+ "learning_rate": 1.567232028991256e-05,
+ "loss": 0.2981,
+ "step": 12066
+ },
+ {
+ "epoch": 32.880108991825615,
+ "grad_norm": 4.148009300231934,
+ "learning_rate": 1.5671593482283797e-05,
+ "loss": 0.2098,
+ "step": 12067
+ },
+ {
+ "epoch": 32.88283378746594,
+ "grad_norm": 4.413739204406738,
+ "learning_rate": 1.5670866630484847e-05,
+ "loss": 0.2974,
+ "step": 12068
+ },
+ {
+ "epoch": 32.88555858310627,
+ "grad_norm": 4.637975215911865,
+ "learning_rate": 1.567013973452137e-05,
+ "loss": 0.2052,
+ "step": 12069
+ },
+ {
+ "epoch": 32.88828337874659,
+ "grad_norm": 4.219438076019287,
+ "learning_rate": 1.5669412794399027e-05,
+ "loss": 0.3179,
+ "step": 12070
+ },
+ {
+ "epoch": 32.89100817438692,
+ "grad_norm": 3.7017269134521484,
+ "learning_rate": 1.5668685810123477e-05,
+ "loss": 0.2923,
+ "step": 12071
+ },
+ {
+ "epoch": 32.89373297002725,
+ "grad_norm": 5.937873363494873,
+ "learning_rate": 1.5667958781700386e-05,
+ "loss": 0.3006,
+ "step": 12072
+ },
+ {
+ "epoch": 32.89645776566758,
+ "grad_norm": 4.130650520324707,
+ "learning_rate": 1.5667231709135414e-05,
+ "loss": 0.2703,
+ "step": 12073
+ },
+ {
+ "epoch": 32.8991825613079,
+ "grad_norm": 5.860306262969971,
+ "learning_rate": 1.5666504592434228e-05,
+ "loss": 0.1943,
+ "step": 12074
+ },
+ {
+ "epoch": 32.90190735694823,
+ "grad_norm": 4.51800012588501,
+ "learning_rate": 1.566577743160248e-05,
+ "loss": 0.2339,
+ "step": 12075
+ },
+ {
+ "epoch": 32.904632152588555,
+ "grad_norm": 6.462704181671143,
+ "learning_rate": 1.566505022664584e-05,
+ "loss": 0.2243,
+ "step": 12076
+ },
+ {
+ "epoch": 32.90735694822888,
+ "grad_norm": 4.28448486328125,
+ "learning_rate": 1.5664322977569976e-05,
+ "loss": 0.2813,
+ "step": 12077
+ },
+ {
+ "epoch": 32.91008174386921,
+ "grad_norm": 3.6238622665405273,
+ "learning_rate": 1.5663595684380544e-05,
+ "loss": 0.1544,
+ "step": 12078
+ },
+ {
+ "epoch": 32.91280653950954,
+ "grad_norm": 6.024811744689941,
+ "learning_rate": 1.5662868347083213e-05,
+ "loss": 0.28,
+ "step": 12079
+ },
+ {
+ "epoch": 32.915531335149865,
+ "grad_norm": 4.390356063842773,
+ "learning_rate": 1.5662140965683645e-05,
+ "loss": 0.337,
+ "step": 12080
+ },
+ {
+ "epoch": 32.91825613079019,
+ "grad_norm": 3.7591423988342285,
+ "learning_rate": 1.5661413540187504e-05,
+ "loss": 0.268,
+ "step": 12081
+ },
+ {
+ "epoch": 32.920980926430516,
+ "grad_norm": 3.4033737182617188,
+ "learning_rate": 1.566068607060046e-05,
+ "loss": 0.2239,
+ "step": 12082
+ },
+ {
+ "epoch": 32.92370572207084,
+ "grad_norm": 7.092580318450928,
+ "learning_rate": 1.565995855692817e-05,
+ "loss": 0.1917,
+ "step": 12083
+ },
+ {
+ "epoch": 32.926430517711175,
+ "grad_norm": 4.271806716918945,
+ "learning_rate": 1.5659230999176307e-05,
+ "loss": 0.1876,
+ "step": 12084
+ },
+ {
+ "epoch": 32.9291553133515,
+ "grad_norm": 3.870826005935669,
+ "learning_rate": 1.5658503397350536e-05,
+ "loss": 0.2863,
+ "step": 12085
+ },
+ {
+ "epoch": 32.93188010899183,
+ "grad_norm": 5.3072733879089355,
+ "learning_rate": 1.5657775751456524e-05,
+ "loss": 0.2198,
+ "step": 12086
+ },
+ {
+ "epoch": 32.93460490463215,
+ "grad_norm": 3.607224225997925,
+ "learning_rate": 1.5657048061499932e-05,
+ "loss": 0.2188,
+ "step": 12087
+ },
+ {
+ "epoch": 32.93732970027248,
+ "grad_norm": 4.132144451141357,
+ "learning_rate": 1.5656320327486433e-05,
+ "loss": 0.2327,
+ "step": 12088
+ },
+ {
+ "epoch": 32.940054495912804,
+ "grad_norm": 4.271884918212891,
+ "learning_rate": 1.5655592549421695e-05,
+ "loss": 0.3441,
+ "step": 12089
+ },
+ {
+ "epoch": 32.94277929155314,
+ "grad_norm": 3.6347427368164062,
+ "learning_rate": 1.565486472731139e-05,
+ "loss": 0.2748,
+ "step": 12090
+ },
+ {
+ "epoch": 32.94550408719346,
+ "grad_norm": 4.650032043457031,
+ "learning_rate": 1.5654136861161174e-05,
+ "loss": 0.3074,
+ "step": 12091
+ },
+ {
+ "epoch": 32.94822888283379,
+ "grad_norm": 4.391534805297852,
+ "learning_rate": 1.5653408950976717e-05,
+ "loss": 0.2554,
+ "step": 12092
+ },
+ {
+ "epoch": 32.950953678474114,
+ "grad_norm": 8.478348731994629,
+ "learning_rate": 1.5652680996763702e-05,
+ "loss": 0.2575,
+ "step": 12093
+ },
+ {
+ "epoch": 32.95367847411444,
+ "grad_norm": 4.116427421569824,
+ "learning_rate": 1.5651952998527786e-05,
+ "loss": 0.245,
+ "step": 12094
+ },
+ {
+ "epoch": 32.956403269754766,
+ "grad_norm": 4.085764408111572,
+ "learning_rate": 1.5651224956274643e-05,
+ "loss": 0.1886,
+ "step": 12095
+ },
+ {
+ "epoch": 32.95912806539509,
+ "grad_norm": 4.858429431915283,
+ "learning_rate": 1.565049687000994e-05,
+ "loss": 0.2343,
+ "step": 12096
+ },
+ {
+ "epoch": 32.961852861035425,
+ "grad_norm": 4.981671333312988,
+ "learning_rate": 1.5649768739739348e-05,
+ "loss": 0.3289,
+ "step": 12097
+ },
+ {
+ "epoch": 32.96457765667575,
+ "grad_norm": 4.482907772064209,
+ "learning_rate": 1.564904056546854e-05,
+ "loss": 0.2026,
+ "step": 12098
+ },
+ {
+ "epoch": 32.967302452316076,
+ "grad_norm": 4.4313740730285645,
+ "learning_rate": 1.564831234720319e-05,
+ "loss": 0.2125,
+ "step": 12099
+ },
+ {
+ "epoch": 32.9700272479564,
+ "grad_norm": 4.870089054107666,
+ "learning_rate": 1.564758408494896e-05,
+ "loss": 0.2978,
+ "step": 12100
+ },
+ {
+ "epoch": 32.97275204359673,
+ "grad_norm": 3.7818498611450195,
+ "learning_rate": 1.564685577871153e-05,
+ "loss": 0.2449,
+ "step": 12101
+ },
+ {
+ "epoch": 32.97547683923706,
+ "grad_norm": 5.005337238311768,
+ "learning_rate": 1.564612742849657e-05,
+ "loss": 0.2852,
+ "step": 12102
+ },
+ {
+ "epoch": 32.97820163487739,
+ "grad_norm": 3.9331626892089844,
+ "learning_rate": 1.564539903430975e-05,
+ "loss": 0.1745,
+ "step": 12103
+ },
+ {
+ "epoch": 32.98092643051771,
+ "grad_norm": 4.643917083740234,
+ "learning_rate": 1.564467059615674e-05,
+ "loss": 0.2474,
+ "step": 12104
+ },
+ {
+ "epoch": 32.98365122615804,
+ "grad_norm": 3.8314225673675537,
+ "learning_rate": 1.564394211404322e-05,
+ "loss": 0.1681,
+ "step": 12105
+ },
+ {
+ "epoch": 32.986376021798364,
+ "grad_norm": 3.9555909633636475,
+ "learning_rate": 1.564321358797486e-05,
+ "loss": 0.2622,
+ "step": 12106
+ },
+ {
+ "epoch": 32.98910081743869,
+ "grad_norm": 5.422633171081543,
+ "learning_rate": 1.564248501795734e-05,
+ "loss": 0.3849,
+ "step": 12107
+ },
+ {
+ "epoch": 32.991825613079016,
+ "grad_norm": 4.5062174797058105,
+ "learning_rate": 1.564175640399632e-05,
+ "loss": 0.2201,
+ "step": 12108
+ },
+ {
+ "epoch": 32.99455040871935,
+ "grad_norm": 6.3272247314453125,
+ "learning_rate": 1.5641027746097484e-05,
+ "loss": 0.5173,
+ "step": 12109
+ },
+ {
+ "epoch": 32.997275204359674,
+ "grad_norm": 5.264074325561523,
+ "learning_rate": 1.564029904426651e-05,
+ "loss": 0.3011,
+ "step": 12110
+ },
+ {
+ "epoch": 33.0,
+ "grad_norm": 3.451688051223755,
+ "learning_rate": 1.5639570298509067e-05,
+ "loss": 0.1148,
+ "step": 12111
+ },
+ {
+ "epoch": 33.002724795640326,
+ "grad_norm": 4.479669570922852,
+ "learning_rate": 1.563884150883083e-05,
+ "loss": 0.4402,
+ "step": 12112
+ },
+ {
+ "epoch": 33.00544959128065,
+ "grad_norm": 4.256744861602783,
+ "learning_rate": 1.563811267523748e-05,
+ "loss": 0.1618,
+ "step": 12113
+ },
+ {
+ "epoch": 33.00817438692098,
+ "grad_norm": 4.396951198577881,
+ "learning_rate": 1.5637383797734685e-05,
+ "loss": 0.3342,
+ "step": 12114
+ },
+ {
+ "epoch": 33.01089918256131,
+ "grad_norm": 4.253628253936768,
+ "learning_rate": 1.5636654876328134e-05,
+ "loss": 0.2566,
+ "step": 12115
+ },
+ {
+ "epoch": 33.013623978201636,
+ "grad_norm": 3.688251495361328,
+ "learning_rate": 1.563592591102349e-05,
+ "loss": 0.2681,
+ "step": 12116
+ },
+ {
+ "epoch": 33.01634877384196,
+ "grad_norm": 3.733710765838623,
+ "learning_rate": 1.5635196901826436e-05,
+ "loss": 0.5062,
+ "step": 12117
+ },
+ {
+ "epoch": 33.01907356948229,
+ "grad_norm": 4.305470943450928,
+ "learning_rate": 1.563446784874265e-05,
+ "loss": 0.1993,
+ "step": 12118
+ },
+ {
+ "epoch": 33.02179836512261,
+ "grad_norm": 4.978392124176025,
+ "learning_rate": 1.5633738751777814e-05,
+ "loss": 0.2044,
+ "step": 12119
+ },
+ {
+ "epoch": 33.02452316076294,
+ "grad_norm": 4.464511394500732,
+ "learning_rate": 1.56330096109376e-05,
+ "loss": 0.2119,
+ "step": 12120
+ },
+ {
+ "epoch": 33.02724795640327,
+ "grad_norm": 4.609066963195801,
+ "learning_rate": 1.5632280426227685e-05,
+ "loss": 0.4451,
+ "step": 12121
+ },
+ {
+ "epoch": 33.0299727520436,
+ "grad_norm": 3.5475664138793945,
+ "learning_rate": 1.5631551197653756e-05,
+ "loss": 0.2088,
+ "step": 12122
+ },
+ {
+ "epoch": 33.032697547683924,
+ "grad_norm": 3.199610948562622,
+ "learning_rate": 1.5630821925221488e-05,
+ "loss": 0.1312,
+ "step": 12123
+ },
+ {
+ "epoch": 33.03542234332425,
+ "grad_norm": 4.308503150939941,
+ "learning_rate": 1.563009260893656e-05,
+ "loss": 0.2318,
+ "step": 12124
+ },
+ {
+ "epoch": 33.038147138964575,
+ "grad_norm": 4.5556960105896,
+ "learning_rate": 1.5629363248804644e-05,
+ "loss": 0.1222,
+ "step": 12125
+ },
+ {
+ "epoch": 33.0408719346049,
+ "grad_norm": 4.171080589294434,
+ "learning_rate": 1.5628633844831436e-05,
+ "loss": 0.2122,
+ "step": 12126
+ },
+ {
+ "epoch": 33.043596730245234,
+ "grad_norm": 3.8739919662475586,
+ "learning_rate": 1.5627904397022605e-05,
+ "loss": 0.2936,
+ "step": 12127
+ },
+ {
+ "epoch": 33.04632152588556,
+ "grad_norm": 3.8731093406677246,
+ "learning_rate": 1.5627174905383838e-05,
+ "loss": 0.2058,
+ "step": 12128
+ },
+ {
+ "epoch": 33.049046321525886,
+ "grad_norm": 3.8274571895599365,
+ "learning_rate": 1.5626445369920816e-05,
+ "loss": 0.1362,
+ "step": 12129
+ },
+ {
+ "epoch": 33.05177111716621,
+ "grad_norm": 4.06612491607666,
+ "learning_rate": 1.5625715790639214e-05,
+ "loss": 0.1688,
+ "step": 12130
+ },
+ {
+ "epoch": 33.05449591280654,
+ "grad_norm": 3.317263126373291,
+ "learning_rate": 1.562498616754472e-05,
+ "loss": 0.2244,
+ "step": 12131
+ },
+ {
+ "epoch": 33.05722070844686,
+ "grad_norm": 3.8727192878723145,
+ "learning_rate": 1.5624256500643016e-05,
+ "loss": 0.2242,
+ "step": 12132
+ },
+ {
+ "epoch": 33.059945504087196,
+ "grad_norm": 4.40749979019165,
+ "learning_rate": 1.5623526789939784e-05,
+ "loss": 0.2649,
+ "step": 12133
+ },
+ {
+ "epoch": 33.06267029972752,
+ "grad_norm": 3.649848699569702,
+ "learning_rate": 1.5622797035440703e-05,
+ "loss": 0.2804,
+ "step": 12134
+ },
+ {
+ "epoch": 33.06539509536785,
+ "grad_norm": 4.372573375701904,
+ "learning_rate": 1.5622067237151465e-05,
+ "loss": 0.1511,
+ "step": 12135
+ },
+ {
+ "epoch": 33.06811989100817,
+ "grad_norm": 3.5299594402313232,
+ "learning_rate": 1.5621337395077742e-05,
+ "loss": 0.124,
+ "step": 12136
+ },
+ {
+ "epoch": 33.0708446866485,
+ "grad_norm": 3.305419921875,
+ "learning_rate": 1.5620607509225227e-05,
+ "loss": 0.3063,
+ "step": 12137
+ },
+ {
+ "epoch": 33.073569482288825,
+ "grad_norm": 3.8031184673309326,
+ "learning_rate": 1.5619877579599604e-05,
+ "loss": 0.2105,
+ "step": 12138
+ },
+ {
+ "epoch": 33.07629427792916,
+ "grad_norm": 4.468491077423096,
+ "learning_rate": 1.5619147606206552e-05,
+ "loss": 0.2431,
+ "step": 12139
+ },
+ {
+ "epoch": 33.079019073569484,
+ "grad_norm": 4.582039833068848,
+ "learning_rate": 1.561841758905176e-05,
+ "loss": 0.2618,
+ "step": 12140
+ },
+ {
+ "epoch": 33.08174386920981,
+ "grad_norm": 4.48699426651001,
+ "learning_rate": 1.5617687528140915e-05,
+ "loss": 0.2312,
+ "step": 12141
+ },
+ {
+ "epoch": 33.084468664850135,
+ "grad_norm": 2.9460766315460205,
+ "learning_rate": 1.5616957423479698e-05,
+ "loss": 0.1521,
+ "step": 12142
+ },
+ {
+ "epoch": 33.08719346049046,
+ "grad_norm": 3.607347011566162,
+ "learning_rate": 1.5616227275073796e-05,
+ "loss": 0.2362,
+ "step": 12143
+ },
+ {
+ "epoch": 33.08991825613079,
+ "grad_norm": 3.748016595840454,
+ "learning_rate": 1.56154970829289e-05,
+ "loss": 0.3024,
+ "step": 12144
+ },
+ {
+ "epoch": 33.09264305177112,
+ "grad_norm": 5.744332313537598,
+ "learning_rate": 1.561476684705069e-05,
+ "loss": 0.1448,
+ "step": 12145
+ },
+ {
+ "epoch": 33.095367847411445,
+ "grad_norm": 4.611015319824219,
+ "learning_rate": 1.561403656744486e-05,
+ "loss": 0.1479,
+ "step": 12146
+ },
+ {
+ "epoch": 33.09809264305177,
+ "grad_norm": 3.7908775806427,
+ "learning_rate": 1.5613306244117092e-05,
+ "loss": 0.2832,
+ "step": 12147
+ },
+ {
+ "epoch": 33.1008174386921,
+ "grad_norm": 4.208542823791504,
+ "learning_rate": 1.5612575877073078e-05,
+ "loss": 0.2249,
+ "step": 12148
+ },
+ {
+ "epoch": 33.10354223433242,
+ "grad_norm": 3.6432583332061768,
+ "learning_rate": 1.56118454663185e-05,
+ "loss": 0.2591,
+ "step": 12149
+ },
+ {
+ "epoch": 33.10626702997275,
+ "grad_norm": 3.5415537357330322,
+ "learning_rate": 1.5611115011859052e-05,
+ "loss": 0.164,
+ "step": 12150
+ },
+ {
+ "epoch": 33.10899182561308,
+ "grad_norm": 3.3571653366088867,
+ "learning_rate": 1.5610384513700422e-05,
+ "loss": 0.1897,
+ "step": 12151
+ },
+ {
+ "epoch": 33.11171662125341,
+ "grad_norm": 14.435155868530273,
+ "learning_rate": 1.5609653971848296e-05,
+ "loss": 0.1639,
+ "step": 12152
+ },
+ {
+ "epoch": 33.11444141689373,
+ "grad_norm": 4.71366024017334,
+ "learning_rate": 1.5608923386308365e-05,
+ "loss": 0.1504,
+ "step": 12153
+ },
+ {
+ "epoch": 33.11716621253406,
+ "grad_norm": 3.797769546508789,
+ "learning_rate": 1.560819275708632e-05,
+ "loss": 0.1973,
+ "step": 12154
+ },
+ {
+ "epoch": 33.119891008174385,
+ "grad_norm": 3.753394842147827,
+ "learning_rate": 1.5607462084187853e-05,
+ "loss": 0.1126,
+ "step": 12155
+ },
+ {
+ "epoch": 33.12261580381471,
+ "grad_norm": 4.530348300933838,
+ "learning_rate": 1.5606731367618647e-05,
+ "loss": 0.1785,
+ "step": 12156
+ },
+ {
+ "epoch": 33.12534059945504,
+ "grad_norm": 4.079743385314941,
+ "learning_rate": 1.5606000607384402e-05,
+ "loss": 0.2255,
+ "step": 12157
+ },
+ {
+ "epoch": 33.12806539509537,
+ "grad_norm": 4.506271839141846,
+ "learning_rate": 1.56052698034908e-05,
+ "loss": 0.2195,
+ "step": 12158
+ },
+ {
+ "epoch": 33.130790190735695,
+ "grad_norm": 3.3269481658935547,
+ "learning_rate": 1.5604538955943542e-05,
+ "loss": 0.1823,
+ "step": 12159
+ },
+ {
+ "epoch": 33.13351498637602,
+ "grad_norm": 4.420558929443359,
+ "learning_rate": 1.5603808064748313e-05,
+ "loss": 0.2434,
+ "step": 12160
+ },
+ {
+ "epoch": 33.13623978201635,
+ "grad_norm": 4.619765758514404,
+ "learning_rate": 1.5603077129910805e-05,
+ "loss": 0.1713,
+ "step": 12161
+ },
+ {
+ "epoch": 33.13896457765667,
+ "grad_norm": 3.6859776973724365,
+ "learning_rate": 1.5602346151436715e-05,
+ "loss": 0.1559,
+ "step": 12162
+ },
+ {
+ "epoch": 33.141689373297005,
+ "grad_norm": 3.6655168533325195,
+ "learning_rate": 1.5601615129331734e-05,
+ "loss": 0.2625,
+ "step": 12163
+ },
+ {
+ "epoch": 33.14441416893733,
+ "grad_norm": 4.480698585510254,
+ "learning_rate": 1.5600884063601554e-05,
+ "loss": 0.2052,
+ "step": 12164
+ },
+ {
+ "epoch": 33.14713896457766,
+ "grad_norm": 7.679446697235107,
+ "learning_rate": 1.560015295425187e-05,
+ "loss": 0.2228,
+ "step": 12165
+ },
+ {
+ "epoch": 33.14986376021798,
+ "grad_norm": 3.4671857357025146,
+ "learning_rate": 1.5599421801288373e-05,
+ "loss": 0.2139,
+ "step": 12166
+ },
+ {
+ "epoch": 33.15258855585831,
+ "grad_norm": 3.1641478538513184,
+ "learning_rate": 1.5598690604716763e-05,
+ "loss": 0.1939,
+ "step": 12167
+ },
+ {
+ "epoch": 33.155313351498634,
+ "grad_norm": 3.7767386436462402,
+ "learning_rate": 1.5597959364542726e-05,
+ "loss": 0.2185,
+ "step": 12168
+ },
+ {
+ "epoch": 33.15803814713897,
+ "grad_norm": 3.768197774887085,
+ "learning_rate": 1.559722808077196e-05,
+ "loss": 0.2285,
+ "step": 12169
+ },
+ {
+ "epoch": 33.16076294277929,
+ "grad_norm": 3.4886081218719482,
+ "learning_rate": 1.559649675341017e-05,
+ "loss": 0.2074,
+ "step": 12170
+ },
+ {
+ "epoch": 33.16348773841962,
+ "grad_norm": 3.594101905822754,
+ "learning_rate": 1.5595765382463035e-05,
+ "loss": 0.1466,
+ "step": 12171
+ },
+ {
+ "epoch": 33.166212534059945,
+ "grad_norm": 3.5610952377319336,
+ "learning_rate": 1.5595033967936264e-05,
+ "loss": 0.2788,
+ "step": 12172
+ },
+ {
+ "epoch": 33.16893732970027,
+ "grad_norm": 4.203703880310059,
+ "learning_rate": 1.5594302509835548e-05,
+ "loss": 0.3355,
+ "step": 12173
+ },
+ {
+ "epoch": 33.171662125340596,
+ "grad_norm": 4.241708278656006,
+ "learning_rate": 1.5593571008166587e-05,
+ "loss": 0.1806,
+ "step": 12174
+ },
+ {
+ "epoch": 33.17438692098093,
+ "grad_norm": 4.827541351318359,
+ "learning_rate": 1.559283946293507e-05,
+ "loss": 0.1634,
+ "step": 12175
+ },
+ {
+ "epoch": 33.177111716621255,
+ "grad_norm": 3.6416127681732178,
+ "learning_rate": 1.55921078741467e-05,
+ "loss": 0.14,
+ "step": 12176
+ },
+ {
+ "epoch": 33.17983651226158,
+ "grad_norm": 3.799471378326416,
+ "learning_rate": 1.5591376241807177e-05,
+ "loss": 0.2225,
+ "step": 12177
+ },
+ {
+ "epoch": 33.182561307901906,
+ "grad_norm": 3.974888801574707,
+ "learning_rate": 1.559064456592219e-05,
+ "loss": 0.2027,
+ "step": 12178
+ },
+ {
+ "epoch": 33.18528610354223,
+ "grad_norm": 4.755573272705078,
+ "learning_rate": 1.5589912846497447e-05,
+ "loss": 0.2166,
+ "step": 12179
+ },
+ {
+ "epoch": 33.18801089918256,
+ "grad_norm": 5.127408504486084,
+ "learning_rate": 1.5589181083538646e-05,
+ "loss": 0.316,
+ "step": 12180
+ },
+ {
+ "epoch": 33.19073569482289,
+ "grad_norm": 5.5539774894714355,
+ "learning_rate": 1.5588449277051476e-05,
+ "loss": 0.2688,
+ "step": 12181
+ },
+ {
+ "epoch": 33.19346049046322,
+ "grad_norm": 4.1330413818359375,
+ "learning_rate": 1.5587717427041645e-05,
+ "loss": 0.2343,
+ "step": 12182
+ },
+ {
+ "epoch": 33.19618528610354,
+ "grad_norm": 3.878577709197998,
+ "learning_rate": 1.5586985533514853e-05,
+ "loss": 0.1725,
+ "step": 12183
+ },
+ {
+ "epoch": 33.19891008174387,
+ "grad_norm": 7.5865864753723145,
+ "learning_rate": 1.5586253596476798e-05,
+ "loss": 0.1999,
+ "step": 12184
+ },
+ {
+ "epoch": 33.201634877384194,
+ "grad_norm": 3.5923099517822266,
+ "learning_rate": 1.5585521615933176e-05,
+ "loss": 0.2019,
+ "step": 12185
+ },
+ {
+ "epoch": 33.20435967302452,
+ "grad_norm": 4.147721290588379,
+ "learning_rate": 1.5584789591889693e-05,
+ "loss": 0.2747,
+ "step": 12186
+ },
+ {
+ "epoch": 33.20708446866485,
+ "grad_norm": 3.8481078147888184,
+ "learning_rate": 1.5584057524352047e-05,
+ "loss": 0.1259,
+ "step": 12187
+ },
+ {
+ "epoch": 33.20980926430518,
+ "grad_norm": 4.217566967010498,
+ "learning_rate": 1.5583325413325944e-05,
+ "loss": 0.3122,
+ "step": 12188
+ },
+ {
+ "epoch": 33.212534059945504,
+ "grad_norm": 4.201993465423584,
+ "learning_rate": 1.558259325881708e-05,
+ "loss": 0.5162,
+ "step": 12189
+ },
+ {
+ "epoch": 33.21525885558583,
+ "grad_norm": 4.1498823165893555,
+ "learning_rate": 1.558186106083116e-05,
+ "loss": 0.3337,
+ "step": 12190
+ },
+ {
+ "epoch": 33.217983651226156,
+ "grad_norm": 3.9883008003234863,
+ "learning_rate": 1.5581128819373885e-05,
+ "loss": 0.1768,
+ "step": 12191
+ },
+ {
+ "epoch": 33.22070844686648,
+ "grad_norm": 3.318418264389038,
+ "learning_rate": 1.558039653445096e-05,
+ "loss": 0.1894,
+ "step": 12192
+ },
+ {
+ "epoch": 33.223433242506815,
+ "grad_norm": 3.7223050594329834,
+ "learning_rate": 1.5579664206068087e-05,
+ "loss": 0.2216,
+ "step": 12193
+ },
+ {
+ "epoch": 33.22615803814714,
+ "grad_norm": 3.5537753105163574,
+ "learning_rate": 1.5578931834230968e-05,
+ "loss": 0.1722,
+ "step": 12194
+ },
+ {
+ "epoch": 33.228882833787466,
+ "grad_norm": 3.9813754558563232,
+ "learning_rate": 1.5578199418945304e-05,
+ "loss": 0.2582,
+ "step": 12195
+ },
+ {
+ "epoch": 33.23160762942779,
+ "grad_norm": 4.3961262702941895,
+ "learning_rate": 1.5577466960216805e-05,
+ "loss": 0.1547,
+ "step": 12196
+ },
+ {
+ "epoch": 33.23433242506812,
+ "grad_norm": 3.874664545059204,
+ "learning_rate": 1.557673445805117e-05,
+ "loss": 0.1768,
+ "step": 12197
+ },
+ {
+ "epoch": 33.237057220708444,
+ "grad_norm": 4.537845134735107,
+ "learning_rate": 1.5576001912454114e-05,
+ "loss": 0.2575,
+ "step": 12198
+ },
+ {
+ "epoch": 33.23978201634878,
+ "grad_norm": 2.9135775566101074,
+ "learning_rate": 1.5575269323431327e-05,
+ "loss": 0.2682,
+ "step": 12199
+ },
+ {
+ "epoch": 33.2425068119891,
+ "grad_norm": 4.031450271606445,
+ "learning_rate": 1.5574536690988526e-05,
+ "loss": 0.2194,
+ "step": 12200
+ },
+ {
+ "epoch": 33.24523160762943,
+ "grad_norm": 4.323424339294434,
+ "learning_rate": 1.5573804015131414e-05,
+ "loss": 0.2161,
+ "step": 12201
+ },
+ {
+ "epoch": 33.247956403269754,
+ "grad_norm": 3.8000926971435547,
+ "learning_rate": 1.5573071295865694e-05,
+ "loss": 0.1292,
+ "step": 12202
+ },
+ {
+ "epoch": 33.25068119891008,
+ "grad_norm": 5.519829750061035,
+ "learning_rate": 1.557233853319707e-05,
+ "loss": 0.1538,
+ "step": 12203
+ },
+ {
+ "epoch": 33.253405994550405,
+ "grad_norm": 4.231913089752197,
+ "learning_rate": 1.5571605727131255e-05,
+ "loss": 0.2475,
+ "step": 12204
+ },
+ {
+ "epoch": 33.25613079019074,
+ "grad_norm": 4.246346473693848,
+ "learning_rate": 1.5570872877673958e-05,
+ "loss": 0.1671,
+ "step": 12205
+ },
+ {
+ "epoch": 33.258855585831064,
+ "grad_norm": 3.611086368560791,
+ "learning_rate": 1.5570139984830878e-05,
+ "loss": 0.1232,
+ "step": 12206
+ },
+ {
+ "epoch": 33.26158038147139,
+ "grad_norm": 4.119976997375488,
+ "learning_rate": 1.556940704860773e-05,
+ "loss": 0.1931,
+ "step": 12207
+ },
+ {
+ "epoch": 33.264305177111716,
+ "grad_norm": 3.543471574783325,
+ "learning_rate": 1.5568674069010218e-05,
+ "loss": 0.3155,
+ "step": 12208
+ },
+ {
+ "epoch": 33.26702997275204,
+ "grad_norm": 3.72078800201416,
+ "learning_rate": 1.5567941046044053e-05,
+ "loss": 0.2122,
+ "step": 12209
+ },
+ {
+ "epoch": 33.26975476839237,
+ "grad_norm": 3.5586156845092773,
+ "learning_rate": 1.556720797971494e-05,
+ "loss": 0.1398,
+ "step": 12210
+ },
+ {
+ "epoch": 33.2724795640327,
+ "grad_norm": 3.5826845169067383,
+ "learning_rate": 1.556647487002859e-05,
+ "loss": 0.1493,
+ "step": 12211
+ },
+ {
+ "epoch": 33.275204359673026,
+ "grad_norm": 4.178651809692383,
+ "learning_rate": 1.5565741716990714e-05,
+ "loss": 0.3506,
+ "step": 12212
+ },
+ {
+ "epoch": 33.27792915531335,
+ "grad_norm": 4.168033123016357,
+ "learning_rate": 1.556500852060702e-05,
+ "loss": 0.2438,
+ "step": 12213
+ },
+ {
+ "epoch": 33.28065395095368,
+ "grad_norm": 6.558807373046875,
+ "learning_rate": 1.556427528088322e-05,
+ "loss": 0.2411,
+ "step": 12214
+ },
+ {
+ "epoch": 33.283378746594,
+ "grad_norm": 4.1515984535217285,
+ "learning_rate": 1.5563541997825025e-05,
+ "loss": 0.3406,
+ "step": 12215
+ },
+ {
+ "epoch": 33.28610354223433,
+ "grad_norm": 4.438435077667236,
+ "learning_rate": 1.556280867143814e-05,
+ "loss": 0.2237,
+ "step": 12216
+ },
+ {
+ "epoch": 33.28882833787466,
+ "grad_norm": 3.6069626808166504,
+ "learning_rate": 1.556207530172828e-05,
+ "loss": 0.198,
+ "step": 12217
+ },
+ {
+ "epoch": 33.29155313351499,
+ "grad_norm": 3.741222620010376,
+ "learning_rate": 1.556134188870116e-05,
+ "loss": 0.1407,
+ "step": 12218
+ },
+ {
+ "epoch": 33.294277929155314,
+ "grad_norm": 5.258026599884033,
+ "learning_rate": 1.5560608432362486e-05,
+ "loss": 0.1588,
+ "step": 12219
+ },
+ {
+ "epoch": 33.29700272479564,
+ "grad_norm": 3.973278045654297,
+ "learning_rate": 1.5559874932717972e-05,
+ "loss": 0.3113,
+ "step": 12220
+ },
+ {
+ "epoch": 33.299727520435965,
+ "grad_norm": 4.778507709503174,
+ "learning_rate": 1.5559141389773334e-05,
+ "loss": 0.1682,
+ "step": 12221
+ },
+ {
+ "epoch": 33.30245231607629,
+ "grad_norm": 3.970388412475586,
+ "learning_rate": 1.555840780353428e-05,
+ "loss": 0.1918,
+ "step": 12222
+ },
+ {
+ "epoch": 33.305177111716624,
+ "grad_norm": 4.370414733886719,
+ "learning_rate": 1.5557674174006525e-05,
+ "loss": 0.1369,
+ "step": 12223
+ },
+ {
+ "epoch": 33.30790190735695,
+ "grad_norm": 3.4287562370300293,
+ "learning_rate": 1.5556940501195784e-05,
+ "loss": 0.1353,
+ "step": 12224
+ },
+ {
+ "epoch": 33.310626702997276,
+ "grad_norm": 3.6426544189453125,
+ "learning_rate": 1.5556206785107766e-05,
+ "loss": 0.1676,
+ "step": 12225
+ },
+ {
+ "epoch": 33.3133514986376,
+ "grad_norm": 4.486706256866455,
+ "learning_rate": 1.555547302574819e-05,
+ "loss": 0.2093,
+ "step": 12226
+ },
+ {
+ "epoch": 33.31607629427793,
+ "grad_norm": 5.129892826080322,
+ "learning_rate": 1.5554739223122768e-05,
+ "loss": 0.1961,
+ "step": 12227
+ },
+ {
+ "epoch": 33.31880108991825,
+ "grad_norm": 4.488393783569336,
+ "learning_rate": 1.5554005377237217e-05,
+ "loss": 0.1827,
+ "step": 12228
+ },
+ {
+ "epoch": 33.321525885558586,
+ "grad_norm": 3.6833770275115967,
+ "learning_rate": 1.5553271488097247e-05,
+ "loss": 0.1615,
+ "step": 12229
+ },
+ {
+ "epoch": 33.32425068119891,
+ "grad_norm": 3.8918232917785645,
+ "learning_rate": 1.5552537555708584e-05,
+ "loss": 0.2614,
+ "step": 12230
+ },
+ {
+ "epoch": 33.32697547683924,
+ "grad_norm": 3.777308225631714,
+ "learning_rate": 1.5551803580076935e-05,
+ "loss": 0.0991,
+ "step": 12231
+ },
+ {
+ "epoch": 33.32970027247956,
+ "grad_norm": 3.491166353225708,
+ "learning_rate": 1.5551069561208018e-05,
+ "loss": 0.1297,
+ "step": 12232
+ },
+ {
+ "epoch": 33.33242506811989,
+ "grad_norm": 4.653164863586426,
+ "learning_rate": 1.555033549910755e-05,
+ "loss": 0.183,
+ "step": 12233
+ },
+ {
+ "epoch": 33.335149863760215,
+ "grad_norm": 4.101744651794434,
+ "learning_rate": 1.5549601393781244e-05,
+ "loss": 0.2946,
+ "step": 12234
+ },
+ {
+ "epoch": 33.33787465940055,
+ "grad_norm": 4.314894199371338,
+ "learning_rate": 1.5548867245234825e-05,
+ "loss": 0.2332,
+ "step": 12235
+ },
+ {
+ "epoch": 33.34059945504087,
+ "grad_norm": 3.9911906719207764,
+ "learning_rate": 1.5548133053474004e-05,
+ "loss": 0.1133,
+ "step": 12236
+ },
+ {
+ "epoch": 33.3433242506812,
+ "grad_norm": 4.810930252075195,
+ "learning_rate": 1.5547398818504502e-05,
+ "loss": 0.2382,
+ "step": 12237
+ },
+ {
+ "epoch": 33.346049046321525,
+ "grad_norm": 3.4910051822662354,
+ "learning_rate": 1.5546664540332037e-05,
+ "loss": 0.1045,
+ "step": 12238
+ },
+ {
+ "epoch": 33.34877384196185,
+ "grad_norm": 3.4904589653015137,
+ "learning_rate": 1.554593021896233e-05,
+ "loss": 0.1212,
+ "step": 12239
+ },
+ {
+ "epoch": 33.35149863760218,
+ "grad_norm": 4.434689998626709,
+ "learning_rate": 1.5545195854401095e-05,
+ "loss": 0.1794,
+ "step": 12240
+ },
+ {
+ "epoch": 33.35422343324251,
+ "grad_norm": 3.492683172225952,
+ "learning_rate": 1.5544461446654053e-05,
+ "loss": 0.097,
+ "step": 12241
+ },
+ {
+ "epoch": 33.356948228882835,
+ "grad_norm": 4.402559757232666,
+ "learning_rate": 1.554372699572692e-05,
+ "loss": 0.1881,
+ "step": 12242
+ },
+ {
+ "epoch": 33.35967302452316,
+ "grad_norm": 3.754403829574585,
+ "learning_rate": 1.5542992501625425e-05,
+ "loss": 0.1885,
+ "step": 12243
+ },
+ {
+ "epoch": 33.36239782016349,
+ "grad_norm": 3.997544527053833,
+ "learning_rate": 1.5542257964355282e-05,
+ "loss": 0.1965,
+ "step": 12244
+ },
+ {
+ "epoch": 33.36512261580381,
+ "grad_norm": 3.5068652629852295,
+ "learning_rate": 1.5541523383922206e-05,
+ "loss": 0.1973,
+ "step": 12245
+ },
+ {
+ "epoch": 33.36784741144414,
+ "grad_norm": 5.438575267791748,
+ "learning_rate": 1.554078876033193e-05,
+ "loss": 0.2884,
+ "step": 12246
+ },
+ {
+ "epoch": 33.37057220708447,
+ "grad_norm": 4.353632926940918,
+ "learning_rate": 1.5540054093590173e-05,
+ "loss": 0.2627,
+ "step": 12247
+ },
+ {
+ "epoch": 33.3732970027248,
+ "grad_norm": 4.308499813079834,
+ "learning_rate": 1.5539319383702646e-05,
+ "loss": 0.2195,
+ "step": 12248
+ },
+ {
+ "epoch": 33.37602179836512,
+ "grad_norm": 3.2354915142059326,
+ "learning_rate": 1.553858463067508e-05,
+ "loss": 0.1482,
+ "step": 12249
+ },
+ {
+ "epoch": 33.37874659400545,
+ "grad_norm": 3.5397026538848877,
+ "learning_rate": 1.5537849834513197e-05,
+ "loss": 0.1297,
+ "step": 12250
+ },
+ {
+ "epoch": 33.381471389645775,
+ "grad_norm": 4.178149700164795,
+ "learning_rate": 1.5537114995222716e-05,
+ "loss": 0.228,
+ "step": 12251
+ },
+ {
+ "epoch": 33.3841961852861,
+ "grad_norm": 3.679966926574707,
+ "learning_rate": 1.553638011280936e-05,
+ "loss": 0.2971,
+ "step": 12252
+ },
+ {
+ "epoch": 33.38692098092643,
+ "grad_norm": 3.5231542587280273,
+ "learning_rate": 1.553564518727886e-05,
+ "loss": 0.3609,
+ "step": 12253
+ },
+ {
+ "epoch": 33.38964577656676,
+ "grad_norm": 4.001139163970947,
+ "learning_rate": 1.553491021863693e-05,
+ "loss": 0.1567,
+ "step": 12254
+ },
+ {
+ "epoch": 33.392370572207085,
+ "grad_norm": 4.685465335845947,
+ "learning_rate": 1.553417520688929e-05,
+ "loss": 0.1775,
+ "step": 12255
+ },
+ {
+ "epoch": 33.39509536784741,
+ "grad_norm": 5.528614044189453,
+ "learning_rate": 1.553344015204168e-05,
+ "loss": 0.2257,
+ "step": 12256
+ },
+ {
+ "epoch": 33.39782016348774,
+ "grad_norm": 4.301812648773193,
+ "learning_rate": 1.5532705054099816e-05,
+ "loss": 0.235,
+ "step": 12257
+ },
+ {
+ "epoch": 33.40054495912806,
+ "grad_norm": 4.414159297943115,
+ "learning_rate": 1.5531969913069423e-05,
+ "loss": 0.1984,
+ "step": 12258
+ },
+ {
+ "epoch": 33.403269754768395,
+ "grad_norm": 5.015557765960693,
+ "learning_rate": 1.5531234728956227e-05,
+ "loss": 0.2532,
+ "step": 12259
+ },
+ {
+ "epoch": 33.40599455040872,
+ "grad_norm": 3.617192029953003,
+ "learning_rate": 1.553049950176595e-05,
+ "loss": 0.2791,
+ "step": 12260
+ },
+ {
+ "epoch": 33.40871934604905,
+ "grad_norm": 4.254549026489258,
+ "learning_rate": 1.552976423150432e-05,
+ "loss": 0.275,
+ "step": 12261
+ },
+ {
+ "epoch": 33.41144414168937,
+ "grad_norm": 3.6656060218811035,
+ "learning_rate": 1.5529028918177067e-05,
+ "loss": 0.3676,
+ "step": 12262
+ },
+ {
+ "epoch": 33.4141689373297,
+ "grad_norm": 3.7733817100524902,
+ "learning_rate": 1.5528293561789912e-05,
+ "loss": 0.2123,
+ "step": 12263
+ },
+ {
+ "epoch": 33.416893732970024,
+ "grad_norm": 4.857420921325684,
+ "learning_rate": 1.5527558162348588e-05,
+ "loss": 0.3754,
+ "step": 12264
+ },
+ {
+ "epoch": 33.41961852861036,
+ "grad_norm": 3.87345027923584,
+ "learning_rate": 1.5526822719858816e-05,
+ "loss": 0.209,
+ "step": 12265
+ },
+ {
+ "epoch": 33.42234332425068,
+ "grad_norm": 4.043323993682861,
+ "learning_rate": 1.5526087234326327e-05,
+ "loss": 0.2442,
+ "step": 12266
+ },
+ {
+ "epoch": 33.42506811989101,
+ "grad_norm": 3.6400339603424072,
+ "learning_rate": 1.5525351705756852e-05,
+ "loss": 0.2688,
+ "step": 12267
+ },
+ {
+ "epoch": 33.427792915531334,
+ "grad_norm": 3.880333662033081,
+ "learning_rate": 1.552461613415611e-05,
+ "loss": 0.4756,
+ "step": 12268
+ },
+ {
+ "epoch": 33.43051771117166,
+ "grad_norm": 3.4105639457702637,
+ "learning_rate": 1.5523880519529838e-05,
+ "loss": 0.4469,
+ "step": 12269
+ },
+ {
+ "epoch": 33.433242506811986,
+ "grad_norm": 3.8422341346740723,
+ "learning_rate": 1.5523144861883767e-05,
+ "loss": 0.2297,
+ "step": 12270
+ },
+ {
+ "epoch": 33.43596730245232,
+ "grad_norm": 3.9479172229766846,
+ "learning_rate": 1.5522409161223614e-05,
+ "loss": 0.2779,
+ "step": 12271
+ },
+ {
+ "epoch": 33.438692098092645,
+ "grad_norm": 3.327402353286743,
+ "learning_rate": 1.5521673417555117e-05,
+ "loss": 0.1458,
+ "step": 12272
+ },
+ {
+ "epoch": 33.44141689373297,
+ "grad_norm": 4.025568008422852,
+ "learning_rate": 1.552093763088401e-05,
+ "loss": 0.1682,
+ "step": 12273
+ },
+ {
+ "epoch": 33.444141689373296,
+ "grad_norm": 4.224785804748535,
+ "learning_rate": 1.5520201801216017e-05,
+ "loss": 0.2104,
+ "step": 12274
+ },
+ {
+ "epoch": 33.44686648501362,
+ "grad_norm": 4.02276611328125,
+ "learning_rate": 1.551946592855687e-05,
+ "loss": 0.3253,
+ "step": 12275
+ },
+ {
+ "epoch": 33.44959128065395,
+ "grad_norm": 3.8714821338653564,
+ "learning_rate": 1.55187300129123e-05,
+ "loss": 0.3851,
+ "step": 12276
+ },
+ {
+ "epoch": 33.45231607629428,
+ "grad_norm": 3.6774914264678955,
+ "learning_rate": 1.5517994054288035e-05,
+ "loss": 0.2269,
+ "step": 12277
+ },
+ {
+ "epoch": 33.45504087193461,
+ "grad_norm": 3.5672965049743652,
+ "learning_rate": 1.5517258052689813e-05,
+ "loss": 0.134,
+ "step": 12278
+ },
+ {
+ "epoch": 33.45776566757493,
+ "grad_norm": 4.7909111976623535,
+ "learning_rate": 1.5516522008123364e-05,
+ "loss": 0.189,
+ "step": 12279
+ },
+ {
+ "epoch": 33.46049046321526,
+ "grad_norm": 4.268580436706543,
+ "learning_rate": 1.551578592059442e-05,
+ "loss": 0.1878,
+ "step": 12280
+ },
+ {
+ "epoch": 33.463215258855584,
+ "grad_norm": 4.846640586853027,
+ "learning_rate": 1.5515049790108705e-05,
+ "loss": 0.1839,
+ "step": 12281
+ },
+ {
+ "epoch": 33.46594005449591,
+ "grad_norm": 3.9203596115112305,
+ "learning_rate": 1.5514313616671967e-05,
+ "loss": 0.2135,
+ "step": 12282
+ },
+ {
+ "epoch": 33.46866485013624,
+ "grad_norm": 3.8239309787750244,
+ "learning_rate": 1.5513577400289926e-05,
+ "loss": 0.165,
+ "step": 12283
+ },
+ {
+ "epoch": 33.47138964577657,
+ "grad_norm": 4.432106018066406,
+ "learning_rate": 1.551284114096833e-05,
+ "loss": 0.1654,
+ "step": 12284
+ },
+ {
+ "epoch": 33.474114441416894,
+ "grad_norm": 4.5181803703308105,
+ "learning_rate": 1.55121048387129e-05,
+ "loss": 0.4149,
+ "step": 12285
+ },
+ {
+ "epoch": 33.47683923705722,
+ "grad_norm": 3.045443058013916,
+ "learning_rate": 1.5511368493529375e-05,
+ "loss": 0.1683,
+ "step": 12286
+ },
+ {
+ "epoch": 33.479564032697546,
+ "grad_norm": 3.928257942199707,
+ "learning_rate": 1.551063210542349e-05,
+ "loss": 0.2708,
+ "step": 12287
+ },
+ {
+ "epoch": 33.48228882833787,
+ "grad_norm": 4.4394941329956055,
+ "learning_rate": 1.5509895674400977e-05,
+ "loss": 0.2778,
+ "step": 12288
+ },
+ {
+ "epoch": 33.485013623978205,
+ "grad_norm": 4.038187503814697,
+ "learning_rate": 1.5509159200467577e-05,
+ "loss": 0.1792,
+ "step": 12289
+ },
+ {
+ "epoch": 33.48773841961853,
+ "grad_norm": 3.4810585975646973,
+ "learning_rate": 1.5508422683629022e-05,
+ "loss": 0.1858,
+ "step": 12290
+ },
+ {
+ "epoch": 33.490463215258856,
+ "grad_norm": 4.398956775665283,
+ "learning_rate": 1.5507686123891046e-05,
+ "loss": 0.1903,
+ "step": 12291
+ },
+ {
+ "epoch": 33.49318801089918,
+ "grad_norm": 3.521336317062378,
+ "learning_rate": 1.5506949521259393e-05,
+ "loss": 0.1913,
+ "step": 12292
+ },
+ {
+ "epoch": 33.49591280653951,
+ "grad_norm": 4.378941535949707,
+ "learning_rate": 1.550621287573979e-05,
+ "loss": 0.384,
+ "step": 12293
+ },
+ {
+ "epoch": 33.49863760217983,
+ "grad_norm": 3.8069982528686523,
+ "learning_rate": 1.550547618733798e-05,
+ "loss": 0.3142,
+ "step": 12294
+ },
+ {
+ "epoch": 33.50136239782017,
+ "grad_norm": 4.185006141662598,
+ "learning_rate": 1.5504739456059694e-05,
+ "loss": 0.3477,
+ "step": 12295
+ },
+ {
+ "epoch": 33.50408719346049,
+ "grad_norm": 4.057317733764648,
+ "learning_rate": 1.550400268191068e-05,
+ "loss": 0.1465,
+ "step": 12296
+ },
+ {
+ "epoch": 33.50681198910082,
+ "grad_norm": 3.8958494663238525,
+ "learning_rate": 1.5503265864896668e-05,
+ "loss": 0.1689,
+ "step": 12297
+ },
+ {
+ "epoch": 33.509536784741144,
+ "grad_norm": 3.7731306552886963,
+ "learning_rate": 1.55025290050234e-05,
+ "loss": 0.2462,
+ "step": 12298
+ },
+ {
+ "epoch": 33.51226158038147,
+ "grad_norm": 3.612363576889038,
+ "learning_rate": 1.550179210229661e-05,
+ "loss": 0.3391,
+ "step": 12299
+ },
+ {
+ "epoch": 33.514986376021795,
+ "grad_norm": 4.337719440460205,
+ "learning_rate": 1.5501055156722044e-05,
+ "loss": 0.31,
+ "step": 12300
+ },
+ {
+ "epoch": 33.51771117166213,
+ "grad_norm": 4.172662734985352,
+ "learning_rate": 1.5500318168305434e-05,
+ "loss": 0.2839,
+ "step": 12301
+ },
+ {
+ "epoch": 33.520435967302454,
+ "grad_norm": 2.90749454498291,
+ "learning_rate": 1.5499581137052527e-05,
+ "loss": 0.3148,
+ "step": 12302
+ },
+ {
+ "epoch": 33.52316076294278,
+ "grad_norm": 3.603384017944336,
+ "learning_rate": 1.5498844062969055e-05,
+ "loss": 0.1531,
+ "step": 12303
+ },
+ {
+ "epoch": 33.525885558583106,
+ "grad_norm": 3.632977247238159,
+ "learning_rate": 1.5498106946060768e-05,
+ "loss": 0.2469,
+ "step": 12304
+ },
+ {
+ "epoch": 33.52861035422343,
+ "grad_norm": 4.333741664886475,
+ "learning_rate": 1.5497369786333398e-05,
+ "loss": 0.2023,
+ "step": 12305
+ },
+ {
+ "epoch": 33.53133514986376,
+ "grad_norm": 4.915125370025635,
+ "learning_rate": 1.5496632583792686e-05,
+ "loss": 0.2614,
+ "step": 12306
+ },
+ {
+ "epoch": 33.53405994550409,
+ "grad_norm": 3.6806869506835938,
+ "learning_rate": 1.549589533844438e-05,
+ "loss": 0.1748,
+ "step": 12307
+ },
+ {
+ "epoch": 33.536784741144416,
+ "grad_norm": 4.778502941131592,
+ "learning_rate": 1.5495158050294216e-05,
+ "loss": 0.2519,
+ "step": 12308
+ },
+ {
+ "epoch": 33.53950953678474,
+ "grad_norm": 4.250118255615234,
+ "learning_rate": 1.549442071934794e-05,
+ "loss": 0.1508,
+ "step": 12309
+ },
+ {
+ "epoch": 33.54223433242507,
+ "grad_norm": 4.211094856262207,
+ "learning_rate": 1.549368334561129e-05,
+ "loss": 0.1877,
+ "step": 12310
+ },
+ {
+ "epoch": 33.54495912806539,
+ "grad_norm": 3.4722750186920166,
+ "learning_rate": 1.5492945929090015e-05,
+ "loss": 0.209,
+ "step": 12311
+ },
+ {
+ "epoch": 33.54768392370572,
+ "grad_norm": 3.794459104537964,
+ "learning_rate": 1.549220846978985e-05,
+ "loss": 0.1787,
+ "step": 12312
+ },
+ {
+ "epoch": 33.55040871934605,
+ "grad_norm": 4.2351555824279785,
+ "learning_rate": 1.549147096771654e-05,
+ "loss": 0.194,
+ "step": 12313
+ },
+ {
+ "epoch": 33.55313351498638,
+ "grad_norm": 4.7593560218811035,
+ "learning_rate": 1.5490733422875836e-05,
+ "loss": 0.3366,
+ "step": 12314
+ },
+ {
+ "epoch": 33.555858310626704,
+ "grad_norm": 3.5925960540771484,
+ "learning_rate": 1.5489995835273476e-05,
+ "loss": 0.1538,
+ "step": 12315
+ },
+ {
+ "epoch": 33.55858310626703,
+ "grad_norm": 4.161036968231201,
+ "learning_rate": 1.5489258204915203e-05,
+ "loss": 0.3225,
+ "step": 12316
+ },
+ {
+ "epoch": 33.561307901907355,
+ "grad_norm": 3.8461785316467285,
+ "learning_rate": 1.5488520531806767e-05,
+ "loss": 0.1401,
+ "step": 12317
+ },
+ {
+ "epoch": 33.56403269754768,
+ "grad_norm": 5.158466339111328,
+ "learning_rate": 1.5487782815953908e-05,
+ "loss": 0.2735,
+ "step": 12318
+ },
+ {
+ "epoch": 33.566757493188014,
+ "grad_norm": 4.0418620109558105,
+ "learning_rate": 1.548704505736237e-05,
+ "loss": 0.2183,
+ "step": 12319
+ },
+ {
+ "epoch": 33.56948228882834,
+ "grad_norm": 3.7572193145751953,
+ "learning_rate": 1.5486307256037908e-05,
+ "loss": 0.3121,
+ "step": 12320
+ },
+ {
+ "epoch": 33.572207084468666,
+ "grad_norm": 3.657606363296509,
+ "learning_rate": 1.5485569411986253e-05,
+ "loss": 0.2205,
+ "step": 12321
+ },
+ {
+ "epoch": 33.57493188010899,
+ "grad_norm": 4.024258613586426,
+ "learning_rate": 1.5484831525213166e-05,
+ "loss": 0.3122,
+ "step": 12322
+ },
+ {
+ "epoch": 33.57765667574932,
+ "grad_norm": 4.280077934265137,
+ "learning_rate": 1.5484093595724388e-05,
+ "loss": 0.1682,
+ "step": 12323
+ },
+ {
+ "epoch": 33.58038147138964,
+ "grad_norm": 4.087749481201172,
+ "learning_rate": 1.548335562352566e-05,
+ "loss": 0.1683,
+ "step": 12324
+ },
+ {
+ "epoch": 33.583106267029976,
+ "grad_norm": 4.235592365264893,
+ "learning_rate": 1.5482617608622744e-05,
+ "loss": 0.1929,
+ "step": 12325
+ },
+ {
+ "epoch": 33.5858310626703,
+ "grad_norm": 4.780014991760254,
+ "learning_rate": 1.548187955102137e-05,
+ "loss": 0.2322,
+ "step": 12326
+ },
+ {
+ "epoch": 33.58855585831063,
+ "grad_norm": 4.197022438049316,
+ "learning_rate": 1.54811414507273e-05,
+ "loss": 0.1436,
+ "step": 12327
+ },
+ {
+ "epoch": 33.59128065395095,
+ "grad_norm": 4.4628214836120605,
+ "learning_rate": 1.5480403307746274e-05,
+ "loss": 0.1883,
+ "step": 12328
+ },
+ {
+ "epoch": 33.59400544959128,
+ "grad_norm": 3.5545246601104736,
+ "learning_rate": 1.5479665122084046e-05,
+ "loss": 0.2854,
+ "step": 12329
+ },
+ {
+ "epoch": 33.596730245231605,
+ "grad_norm": 3.4130759239196777,
+ "learning_rate": 1.547892689374636e-05,
+ "loss": 0.1533,
+ "step": 12330
+ },
+ {
+ "epoch": 33.59945504087194,
+ "grad_norm": 3.7907705307006836,
+ "learning_rate": 1.547818862273897e-05,
+ "loss": 0.1442,
+ "step": 12331
+ },
+ {
+ "epoch": 33.60217983651226,
+ "grad_norm": 3.981764793395996,
+ "learning_rate": 1.547745030906762e-05,
+ "loss": 0.1862,
+ "step": 12332
+ },
+ {
+ "epoch": 33.60490463215259,
+ "grad_norm": 3.6582791805267334,
+ "learning_rate": 1.5476711952738065e-05,
+ "loss": 0.2904,
+ "step": 12333
+ },
+ {
+ "epoch": 33.607629427792915,
+ "grad_norm": 4.69472599029541,
+ "learning_rate": 1.547597355375605e-05,
+ "loss": 0.2645,
+ "step": 12334
+ },
+ {
+ "epoch": 33.61035422343324,
+ "grad_norm": 4.697206497192383,
+ "learning_rate": 1.5475235112127334e-05,
+ "loss": 0.2648,
+ "step": 12335
+ },
+ {
+ "epoch": 33.61307901907357,
+ "grad_norm": 4.073110103607178,
+ "learning_rate": 1.547449662785766e-05,
+ "loss": 0.2169,
+ "step": 12336
+ },
+ {
+ "epoch": 33.6158038147139,
+ "grad_norm": 3.423541307449341,
+ "learning_rate": 1.5473758100952788e-05,
+ "loss": 0.1263,
+ "step": 12337
+ },
+ {
+ "epoch": 33.618528610354225,
+ "grad_norm": 4.684143543243408,
+ "learning_rate": 1.547301953141846e-05,
+ "loss": 0.1272,
+ "step": 12338
+ },
+ {
+ "epoch": 33.62125340599455,
+ "grad_norm": 3.8131532669067383,
+ "learning_rate": 1.5472280919260436e-05,
+ "loss": 0.1433,
+ "step": 12339
+ },
+ {
+ "epoch": 33.62397820163488,
+ "grad_norm": 5.071885108947754,
+ "learning_rate": 1.547154226448446e-05,
+ "loss": 0.2535,
+ "step": 12340
+ },
+ {
+ "epoch": 33.6267029972752,
+ "grad_norm": 3.6464240550994873,
+ "learning_rate": 1.5470803567096286e-05,
+ "loss": 0.1256,
+ "step": 12341
+ },
+ {
+ "epoch": 33.62942779291553,
+ "grad_norm": 6.879084587097168,
+ "learning_rate": 1.5470064827101677e-05,
+ "loss": 0.2536,
+ "step": 12342
+ },
+ {
+ "epoch": 33.63215258855586,
+ "grad_norm": 4.122193336486816,
+ "learning_rate": 1.5469326044506373e-05,
+ "loss": 0.202,
+ "step": 12343
+ },
+ {
+ "epoch": 33.63487738419619,
+ "grad_norm": 5.259288787841797,
+ "learning_rate": 1.5468587219316136e-05,
+ "loss": 0.3275,
+ "step": 12344
+ },
+ {
+ "epoch": 33.63760217983651,
+ "grad_norm": 4.6620564460754395,
+ "learning_rate": 1.5467848351536717e-05,
+ "loss": 0.393,
+ "step": 12345
+ },
+ {
+ "epoch": 33.64032697547684,
+ "grad_norm": 4.335585117340088,
+ "learning_rate": 1.5467109441173876e-05,
+ "loss": 0.2028,
+ "step": 12346
+ },
+ {
+ "epoch": 33.643051771117165,
+ "grad_norm": 4.067448616027832,
+ "learning_rate": 1.5466370488233358e-05,
+ "loss": 0.2369,
+ "step": 12347
+ },
+ {
+ "epoch": 33.64577656675749,
+ "grad_norm": 4.54278564453125,
+ "learning_rate": 1.5465631492720923e-05,
+ "loss": 0.185,
+ "step": 12348
+ },
+ {
+ "epoch": 33.64850136239782,
+ "grad_norm": 4.391808986663818,
+ "learning_rate": 1.546489245464233e-05,
+ "loss": 0.2029,
+ "step": 12349
+ },
+ {
+ "epoch": 33.65122615803815,
+ "grad_norm": 4.8760528564453125,
+ "learning_rate": 1.5464153374003323e-05,
+ "loss": 0.3941,
+ "step": 12350
+ },
+ {
+ "epoch": 33.653950953678475,
+ "grad_norm": 4.817796230316162,
+ "learning_rate": 1.546341425080967e-05,
+ "loss": 0.2983,
+ "step": 12351
+ },
+ {
+ "epoch": 33.6566757493188,
+ "grad_norm": 4.823096752166748,
+ "learning_rate": 1.546267508506712e-05,
+ "loss": 0.1842,
+ "step": 12352
+ },
+ {
+ "epoch": 33.65940054495913,
+ "grad_norm": 4.336400032043457,
+ "learning_rate": 1.5461935876781436e-05,
+ "loss": 0.2323,
+ "step": 12353
+ },
+ {
+ "epoch": 33.66212534059945,
+ "grad_norm": 3.6926486492156982,
+ "learning_rate": 1.546119662595837e-05,
+ "loss": 0.1983,
+ "step": 12354
+ },
+ {
+ "epoch": 33.664850136239785,
+ "grad_norm": 4.420987606048584,
+ "learning_rate": 1.546045733260368e-05,
+ "loss": 0.2033,
+ "step": 12355
+ },
+ {
+ "epoch": 33.66757493188011,
+ "grad_norm": 4.027066230773926,
+ "learning_rate": 1.5459717996723124e-05,
+ "loss": 0.2708,
+ "step": 12356
+ },
+ {
+ "epoch": 33.67029972752044,
+ "grad_norm": 4.226479530334473,
+ "learning_rate": 1.545897861832246e-05,
+ "loss": 0.2103,
+ "step": 12357
+ },
+ {
+ "epoch": 33.67302452316076,
+ "grad_norm": 3.5240867137908936,
+ "learning_rate": 1.5458239197407446e-05,
+ "loss": 0.1664,
+ "step": 12358
+ },
+ {
+ "epoch": 33.67574931880109,
+ "grad_norm": 4.353520393371582,
+ "learning_rate": 1.545749973398384e-05,
+ "loss": 0.3231,
+ "step": 12359
+ },
+ {
+ "epoch": 33.678474114441414,
+ "grad_norm": 3.9316420555114746,
+ "learning_rate": 1.5456760228057405e-05,
+ "loss": 0.1539,
+ "step": 12360
+ },
+ {
+ "epoch": 33.68119891008175,
+ "grad_norm": 4.141351699829102,
+ "learning_rate": 1.545602067963389e-05,
+ "loss": 0.1862,
+ "step": 12361
+ },
+ {
+ "epoch": 33.68392370572207,
+ "grad_norm": 6.668703556060791,
+ "learning_rate": 1.5455281088719067e-05,
+ "loss": 0.1787,
+ "step": 12362
+ },
+ {
+ "epoch": 33.6866485013624,
+ "grad_norm": 4.378644943237305,
+ "learning_rate": 1.5454541455318693e-05,
+ "loss": 0.1918,
+ "step": 12363
+ },
+ {
+ "epoch": 33.689373297002724,
+ "grad_norm": 3.8053319454193115,
+ "learning_rate": 1.545380177943852e-05,
+ "loss": 0.2383,
+ "step": 12364
+ },
+ {
+ "epoch": 33.69209809264305,
+ "grad_norm": 3.7894859313964844,
+ "learning_rate": 1.545306206108432e-05,
+ "loss": 0.1753,
+ "step": 12365
+ },
+ {
+ "epoch": 33.694822888283376,
+ "grad_norm": 3.840974807739258,
+ "learning_rate": 1.545232230026184e-05,
+ "loss": 0.1857,
+ "step": 12366
+ },
+ {
+ "epoch": 33.69754768392371,
+ "grad_norm": 3.442434072494507,
+ "learning_rate": 1.5451582496976856e-05,
+ "loss": 0.1595,
+ "step": 12367
+ },
+ {
+ "epoch": 33.700272479564035,
+ "grad_norm": 6.583978652954102,
+ "learning_rate": 1.5450842651235118e-05,
+ "loss": 0.2434,
+ "step": 12368
+ },
+ {
+ "epoch": 33.70299727520436,
+ "grad_norm": 4.144040584564209,
+ "learning_rate": 1.5450102763042395e-05,
+ "loss": 0.2112,
+ "step": 12369
+ },
+ {
+ "epoch": 33.705722070844686,
+ "grad_norm": 4.151094436645508,
+ "learning_rate": 1.5449362832404452e-05,
+ "loss": 0.1966,
+ "step": 12370
+ },
+ {
+ "epoch": 33.70844686648501,
+ "grad_norm": 3.304642677307129,
+ "learning_rate": 1.544862285932704e-05,
+ "loss": 0.1334,
+ "step": 12371
+ },
+ {
+ "epoch": 33.71117166212534,
+ "grad_norm": 5.228570938110352,
+ "learning_rate": 1.5447882843815932e-05,
+ "loss": 0.3366,
+ "step": 12372
+ },
+ {
+ "epoch": 33.71389645776567,
+ "grad_norm": 3.800921678543091,
+ "learning_rate": 1.544714278587689e-05,
+ "loss": 0.1672,
+ "step": 12373
+ },
+ {
+ "epoch": 33.716621253406,
+ "grad_norm": 3.9321811199188232,
+ "learning_rate": 1.544640268551567e-05,
+ "loss": 0.178,
+ "step": 12374
+ },
+ {
+ "epoch": 33.71934604904632,
+ "grad_norm": 4.742372035980225,
+ "learning_rate": 1.5445662542738044e-05,
+ "loss": 0.2566,
+ "step": 12375
+ },
+ {
+ "epoch": 33.72207084468665,
+ "grad_norm": 4.0693159103393555,
+ "learning_rate": 1.544492235754977e-05,
+ "loss": 0.1497,
+ "step": 12376
+ },
+ {
+ "epoch": 33.724795640326974,
+ "grad_norm": 3.727499008178711,
+ "learning_rate": 1.544418212995662e-05,
+ "loss": 0.225,
+ "step": 12377
+ },
+ {
+ "epoch": 33.7275204359673,
+ "grad_norm": 4.5937042236328125,
+ "learning_rate": 1.5443441859964353e-05,
+ "loss": 0.2596,
+ "step": 12378
+ },
+ {
+ "epoch": 33.73024523160763,
+ "grad_norm": 4.772783279418945,
+ "learning_rate": 1.5442701547578736e-05,
+ "loss": 0.1672,
+ "step": 12379
+ },
+ {
+ "epoch": 33.73297002724796,
+ "grad_norm": 4.779844284057617,
+ "learning_rate": 1.5441961192805538e-05,
+ "loss": 0.2212,
+ "step": 12380
+ },
+ {
+ "epoch": 33.735694822888284,
+ "grad_norm": 4.180213928222656,
+ "learning_rate": 1.5441220795650518e-05,
+ "loss": 0.1623,
+ "step": 12381
+ },
+ {
+ "epoch": 33.73841961852861,
+ "grad_norm": 3.6558289527893066,
+ "learning_rate": 1.5440480356119445e-05,
+ "loss": 0.1365,
+ "step": 12382
+ },
+ {
+ "epoch": 33.741144414168936,
+ "grad_norm": 4.4843292236328125,
+ "learning_rate": 1.5439739874218087e-05,
+ "loss": 0.3884,
+ "step": 12383
+ },
+ {
+ "epoch": 33.74386920980926,
+ "grad_norm": 3.9919939041137695,
+ "learning_rate": 1.543899934995221e-05,
+ "loss": 0.2513,
+ "step": 12384
+ },
+ {
+ "epoch": 33.746594005449595,
+ "grad_norm": 4.198332786560059,
+ "learning_rate": 1.543825878332758e-05,
+ "loss": 0.253,
+ "step": 12385
+ },
+ {
+ "epoch": 33.74931880108992,
+ "grad_norm": 3.919365406036377,
+ "learning_rate": 1.5437518174349967e-05,
+ "loss": 0.2056,
+ "step": 12386
+ },
+ {
+ "epoch": 33.752043596730246,
+ "grad_norm": 4.7017364501953125,
+ "learning_rate": 1.543677752302514e-05,
+ "loss": 0.2719,
+ "step": 12387
+ },
+ {
+ "epoch": 33.75476839237057,
+ "grad_norm": 3.513854742050171,
+ "learning_rate": 1.543603682935886e-05,
+ "loss": 0.3019,
+ "step": 12388
+ },
+ {
+ "epoch": 33.7574931880109,
+ "grad_norm": 3.860442876815796,
+ "learning_rate": 1.5435296093356902e-05,
+ "loss": 0.2068,
+ "step": 12389
+ },
+ {
+ "epoch": 33.76021798365122,
+ "grad_norm": 6.21521520614624,
+ "learning_rate": 1.5434555315025034e-05,
+ "loss": 0.1703,
+ "step": 12390
+ },
+ {
+ "epoch": 33.762942779291556,
+ "grad_norm": 3.9116103649139404,
+ "learning_rate": 1.543381449436902e-05,
+ "loss": 0.1922,
+ "step": 12391
+ },
+ {
+ "epoch": 33.76566757493188,
+ "grad_norm": 4.073206424713135,
+ "learning_rate": 1.5433073631394638e-05,
+ "loss": 0.1796,
+ "step": 12392
+ },
+ {
+ "epoch": 33.76839237057221,
+ "grad_norm": 4.485995292663574,
+ "learning_rate": 1.5432332726107653e-05,
+ "loss": 0.2348,
+ "step": 12393
+ },
+ {
+ "epoch": 33.771117166212534,
+ "grad_norm": 4.23948860168457,
+ "learning_rate": 1.5431591778513835e-05,
+ "loss": 0.237,
+ "step": 12394
+ },
+ {
+ "epoch": 33.77384196185286,
+ "grad_norm": 3.7431938648223877,
+ "learning_rate": 1.5430850788618955e-05,
+ "loss": 0.2174,
+ "step": 12395
+ },
+ {
+ "epoch": 33.776566757493185,
+ "grad_norm": 4.254263401031494,
+ "learning_rate": 1.5430109756428785e-05,
+ "loss": 0.2181,
+ "step": 12396
+ },
+ {
+ "epoch": 33.77929155313352,
+ "grad_norm": 4.097061634063721,
+ "learning_rate": 1.542936868194909e-05,
+ "loss": 0.1796,
+ "step": 12397
+ },
+ {
+ "epoch": 33.782016348773844,
+ "grad_norm": 3.576460123062134,
+ "learning_rate": 1.5428627565185652e-05,
+ "loss": 0.1584,
+ "step": 12398
+ },
+ {
+ "epoch": 33.78474114441417,
+ "grad_norm": 4.138657569885254,
+ "learning_rate": 1.5427886406144235e-05,
+ "loss": 0.2137,
+ "step": 12399
+ },
+ {
+ "epoch": 33.787465940054496,
+ "grad_norm": 3.5377743244171143,
+ "learning_rate": 1.5427145204830615e-05,
+ "loss": 0.2015,
+ "step": 12400
+ },
+ {
+ "epoch": 33.79019073569482,
+ "grad_norm": 3.905850410461426,
+ "learning_rate": 1.542640396125056e-05,
+ "loss": 0.1323,
+ "step": 12401
+ },
+ {
+ "epoch": 33.79291553133515,
+ "grad_norm": 4.795960426330566,
+ "learning_rate": 1.5425662675409845e-05,
+ "loss": 0.2211,
+ "step": 12402
+ },
+ {
+ "epoch": 33.79564032697548,
+ "grad_norm": 4.285989761352539,
+ "learning_rate": 1.5424921347314245e-05,
+ "loss": 0.1859,
+ "step": 12403
+ },
+ {
+ "epoch": 33.798365122615806,
+ "grad_norm": 4.007308483123779,
+ "learning_rate": 1.5424179976969536e-05,
+ "loss": 0.2729,
+ "step": 12404
+ },
+ {
+ "epoch": 33.80108991825613,
+ "grad_norm": 4.031655788421631,
+ "learning_rate": 1.5423438564381484e-05,
+ "loss": 0.2871,
+ "step": 12405
+ },
+ {
+ "epoch": 33.80381471389646,
+ "grad_norm": 3.7451324462890625,
+ "learning_rate": 1.5422697109555867e-05,
+ "loss": 0.2485,
+ "step": 12406
+ },
+ {
+ "epoch": 33.80653950953678,
+ "grad_norm": 4.351385116577148,
+ "learning_rate": 1.542195561249846e-05,
+ "loss": 0.1828,
+ "step": 12407
+ },
+ {
+ "epoch": 33.80926430517711,
+ "grad_norm": 4.365872859954834,
+ "learning_rate": 1.5421214073215037e-05,
+ "loss": 0.3054,
+ "step": 12408
+ },
+ {
+ "epoch": 33.81198910081744,
+ "grad_norm": 4.639953136444092,
+ "learning_rate": 1.5420472491711373e-05,
+ "loss": 0.1895,
+ "step": 12409
+ },
+ {
+ "epoch": 33.81471389645777,
+ "grad_norm": 4.596818447113037,
+ "learning_rate": 1.5419730867993245e-05,
+ "loss": 0.5234,
+ "step": 12410
+ },
+ {
+ "epoch": 33.817438692098094,
+ "grad_norm": 4.148452281951904,
+ "learning_rate": 1.5418989202066427e-05,
+ "loss": 0.1942,
+ "step": 12411
+ },
+ {
+ "epoch": 33.82016348773842,
+ "grad_norm": 4.3375349044799805,
+ "learning_rate": 1.5418247493936695e-05,
+ "loss": 0.1653,
+ "step": 12412
+ },
+ {
+ "epoch": 33.822888283378745,
+ "grad_norm": 3.488513946533203,
+ "learning_rate": 1.5417505743609828e-05,
+ "loss": 0.1566,
+ "step": 12413
+ },
+ {
+ "epoch": 33.82561307901907,
+ "grad_norm": 4.088630676269531,
+ "learning_rate": 1.54167639510916e-05,
+ "loss": 0.2447,
+ "step": 12414
+ },
+ {
+ "epoch": 33.828337874659404,
+ "grad_norm": 4.130410671234131,
+ "learning_rate": 1.5416022116387785e-05,
+ "loss": 0.2507,
+ "step": 12415
+ },
+ {
+ "epoch": 33.83106267029973,
+ "grad_norm": 4.6239542961120605,
+ "learning_rate": 1.541528023950417e-05,
+ "loss": 0.2611,
+ "step": 12416
+ },
+ {
+ "epoch": 33.833787465940055,
+ "grad_norm": 3.7303009033203125,
+ "learning_rate": 1.5414538320446523e-05,
+ "loss": 0.1703,
+ "step": 12417
+ },
+ {
+ "epoch": 33.83651226158038,
+ "grad_norm": 4.018074035644531,
+ "learning_rate": 1.5413796359220622e-05,
+ "loss": 0.3039,
+ "step": 12418
+ },
+ {
+ "epoch": 33.83923705722071,
+ "grad_norm": 3.5778536796569824,
+ "learning_rate": 1.5413054355832257e-05,
+ "loss": 0.1458,
+ "step": 12419
+ },
+ {
+ "epoch": 33.84196185286103,
+ "grad_norm": 3.7108561992645264,
+ "learning_rate": 1.541231231028719e-05,
+ "loss": 0.1875,
+ "step": 12420
+ },
+ {
+ "epoch": 33.844686648501366,
+ "grad_norm": 3.7494802474975586,
+ "learning_rate": 1.5411570222591217e-05,
+ "loss": 0.2172,
+ "step": 12421
+ },
+ {
+ "epoch": 33.84741144414169,
+ "grad_norm": 3.811206340789795,
+ "learning_rate": 1.5410828092750106e-05,
+ "loss": 0.1581,
+ "step": 12422
+ },
+ {
+ "epoch": 33.85013623978202,
+ "grad_norm": 3.7271111011505127,
+ "learning_rate": 1.541008592076964e-05,
+ "loss": 0.1893,
+ "step": 12423
+ },
+ {
+ "epoch": 33.85286103542234,
+ "grad_norm": 4.440730094909668,
+ "learning_rate": 1.5409343706655603e-05,
+ "loss": 0.2147,
+ "step": 12424
+ },
+ {
+ "epoch": 33.85558583106267,
+ "grad_norm": 4.289337158203125,
+ "learning_rate": 1.5408601450413767e-05,
+ "loss": 0.1359,
+ "step": 12425
+ },
+ {
+ "epoch": 33.858310626702995,
+ "grad_norm": 4.3795270919799805,
+ "learning_rate": 1.5407859152049915e-05,
+ "loss": 0.1237,
+ "step": 12426
+ },
+ {
+ "epoch": 33.86103542234333,
+ "grad_norm": 4.63032865524292,
+ "learning_rate": 1.5407116811569833e-05,
+ "loss": 0.3072,
+ "step": 12427
+ },
+ {
+ "epoch": 33.86376021798365,
+ "grad_norm": 4.250334739685059,
+ "learning_rate": 1.54063744289793e-05,
+ "loss": 0.1999,
+ "step": 12428
+ },
+ {
+ "epoch": 33.86648501362398,
+ "grad_norm": 3.7606618404388428,
+ "learning_rate": 1.5405632004284097e-05,
+ "loss": 0.1848,
+ "step": 12429
+ },
+ {
+ "epoch": 33.869209809264305,
+ "grad_norm": 4.012683391571045,
+ "learning_rate": 1.5404889537490004e-05,
+ "loss": 0.1404,
+ "step": 12430
+ },
+ {
+ "epoch": 33.87193460490463,
+ "grad_norm": 14.187862396240234,
+ "learning_rate": 1.5404147028602804e-05,
+ "loss": 0.2845,
+ "step": 12431
+ },
+ {
+ "epoch": 33.87465940054496,
+ "grad_norm": 4.943216323852539,
+ "learning_rate": 1.5403404477628286e-05,
+ "loss": 0.1133,
+ "step": 12432
+ },
+ {
+ "epoch": 33.87738419618529,
+ "grad_norm": 4.1964898109436035,
+ "learning_rate": 1.5402661884572222e-05,
+ "loss": 0.1604,
+ "step": 12433
+ },
+ {
+ "epoch": 33.880108991825615,
+ "grad_norm": 4.485938549041748,
+ "learning_rate": 1.5401919249440406e-05,
+ "loss": 0.2015,
+ "step": 12434
+ },
+ {
+ "epoch": 33.88283378746594,
+ "grad_norm": 3.7708847522735596,
+ "learning_rate": 1.5401176572238615e-05,
+ "loss": 0.1358,
+ "step": 12435
+ },
+ {
+ "epoch": 33.88555858310627,
+ "grad_norm": 3.5011823177337646,
+ "learning_rate": 1.5400433852972633e-05,
+ "loss": 0.225,
+ "step": 12436
+ },
+ {
+ "epoch": 33.88828337874659,
+ "grad_norm": 3.857724666595459,
+ "learning_rate": 1.5399691091648246e-05,
+ "loss": 0.2655,
+ "step": 12437
+ },
+ {
+ "epoch": 33.89100817438692,
+ "grad_norm": 4.839908599853516,
+ "learning_rate": 1.5398948288271242e-05,
+ "loss": 0.155,
+ "step": 12438
+ },
+ {
+ "epoch": 33.89373297002725,
+ "grad_norm": 4.099254131317139,
+ "learning_rate": 1.53982054428474e-05,
+ "loss": 0.1909,
+ "step": 12439
+ },
+ {
+ "epoch": 33.89645776566758,
+ "grad_norm": 3.835951089859009,
+ "learning_rate": 1.5397462555382508e-05,
+ "loss": 0.1558,
+ "step": 12440
+ },
+ {
+ "epoch": 33.8991825613079,
+ "grad_norm": 3.8467869758605957,
+ "learning_rate": 1.539671962588235e-05,
+ "loss": 0.235,
+ "step": 12441
+ },
+ {
+ "epoch": 33.90190735694823,
+ "grad_norm": 3.8629047870635986,
+ "learning_rate": 1.5395976654352715e-05,
+ "loss": 0.2114,
+ "step": 12442
+ },
+ {
+ "epoch": 33.904632152588555,
+ "grad_norm": 4.575008869171143,
+ "learning_rate": 1.539523364079939e-05,
+ "loss": 0.2168,
+ "step": 12443
+ },
+ {
+ "epoch": 33.90735694822888,
+ "grad_norm": 5.133179664611816,
+ "learning_rate": 1.5394490585228154e-05,
+ "loss": 0.1513,
+ "step": 12444
+ },
+ {
+ "epoch": 33.91008174386921,
+ "grad_norm": 3.948139190673828,
+ "learning_rate": 1.5393747487644803e-05,
+ "loss": 0.1702,
+ "step": 12445
+ },
+ {
+ "epoch": 33.91280653950954,
+ "grad_norm": 3.662663698196411,
+ "learning_rate": 1.539300434805512e-05,
+ "loss": 0.1786,
+ "step": 12446
+ },
+ {
+ "epoch": 33.915531335149865,
+ "grad_norm": 8.11739444732666,
+ "learning_rate": 1.5392261166464886e-05,
+ "loss": 0.1925,
+ "step": 12447
+ },
+ {
+ "epoch": 33.91825613079019,
+ "grad_norm": 4.183300495147705,
+ "learning_rate": 1.5391517942879905e-05,
+ "loss": 0.1815,
+ "step": 12448
+ },
+ {
+ "epoch": 33.920980926430516,
+ "grad_norm": 4.009114742279053,
+ "learning_rate": 1.539077467730595e-05,
+ "loss": 0.2298,
+ "step": 12449
+ },
+ {
+ "epoch": 33.92370572207084,
+ "grad_norm": 4.7915449142456055,
+ "learning_rate": 1.539003136974882e-05,
+ "loss": 0.2224,
+ "step": 12450
+ },
+ {
+ "epoch": 33.926430517711175,
+ "grad_norm": 3.8376431465148926,
+ "learning_rate": 1.538928802021429e-05,
+ "loss": 0.2306,
+ "step": 12451
+ },
+ {
+ "epoch": 33.9291553133515,
+ "grad_norm": 4.179988384246826,
+ "learning_rate": 1.5388544628708166e-05,
+ "loss": 0.2609,
+ "step": 12452
+ },
+ {
+ "epoch": 33.93188010899183,
+ "grad_norm": 3.8503451347351074,
+ "learning_rate": 1.5387801195236227e-05,
+ "loss": 0.1873,
+ "step": 12453
+ },
+ {
+ "epoch": 33.93460490463215,
+ "grad_norm": 4.586725234985352,
+ "learning_rate": 1.538705771980427e-05,
+ "loss": 0.2547,
+ "step": 12454
+ },
+ {
+ "epoch": 33.93732970027248,
+ "grad_norm": 4.888325214385986,
+ "learning_rate": 1.5386314202418076e-05,
+ "loss": 0.228,
+ "step": 12455
+ },
+ {
+ "epoch": 33.940054495912804,
+ "grad_norm": 3.1728692054748535,
+ "learning_rate": 1.538557064308344e-05,
+ "loss": 0.1177,
+ "step": 12456
+ },
+ {
+ "epoch": 33.94277929155314,
+ "grad_norm": 3.781783103942871,
+ "learning_rate": 1.5384827041806157e-05,
+ "loss": 0.2598,
+ "step": 12457
+ },
+ {
+ "epoch": 33.94550408719346,
+ "grad_norm": 3.7932097911834717,
+ "learning_rate": 1.5384083398592015e-05,
+ "loss": 0.2762,
+ "step": 12458
+ },
+ {
+ "epoch": 33.94822888283379,
+ "grad_norm": 4.893678188323975,
+ "learning_rate": 1.53833397134468e-05,
+ "loss": 0.1987,
+ "step": 12459
+ },
+ {
+ "epoch": 33.950953678474114,
+ "grad_norm": 3.998488426208496,
+ "learning_rate": 1.5382595986376312e-05,
+ "loss": 0.1876,
+ "step": 12460
+ },
+ {
+ "epoch": 33.95367847411444,
+ "grad_norm": 3.741912364959717,
+ "learning_rate": 1.538185221738634e-05,
+ "loss": 0.1724,
+ "step": 12461
+ },
+ {
+ "epoch": 33.956403269754766,
+ "grad_norm": 5.029571533203125,
+ "learning_rate": 1.538110840648267e-05,
+ "loss": 0.2587,
+ "step": 12462
+ },
+ {
+ "epoch": 33.95912806539509,
+ "grad_norm": 4.710949420928955,
+ "learning_rate": 1.5380364553671107e-05,
+ "loss": 0.2234,
+ "step": 12463
+ },
+ {
+ "epoch": 33.961852861035425,
+ "grad_norm": 3.569978713989258,
+ "learning_rate": 1.537962065895744e-05,
+ "loss": 0.1722,
+ "step": 12464
+ },
+ {
+ "epoch": 33.96457765667575,
+ "grad_norm": 5.035767555236816,
+ "learning_rate": 1.5378876722347455e-05,
+ "loss": 0.3922,
+ "step": 12465
+ },
+ {
+ "epoch": 33.967302452316076,
+ "grad_norm": 4.642547130584717,
+ "learning_rate": 1.5378132743846955e-05,
+ "loss": 0.1741,
+ "step": 12466
+ },
+ {
+ "epoch": 33.9700272479564,
+ "grad_norm": 12.071126937866211,
+ "learning_rate": 1.537738872346173e-05,
+ "loss": 0.1854,
+ "step": 12467
+ },
+ {
+ "epoch": 33.97275204359673,
+ "grad_norm": 5.060849666595459,
+ "learning_rate": 1.537664466119757e-05,
+ "loss": 0.2162,
+ "step": 12468
+ },
+ {
+ "epoch": 33.97547683923706,
+ "grad_norm": 4.250190734863281,
+ "learning_rate": 1.537590055706028e-05,
+ "loss": 0.2178,
+ "step": 12469
+ },
+ {
+ "epoch": 33.97820163487739,
+ "grad_norm": 4.278165340423584,
+ "learning_rate": 1.5375156411055644e-05,
+ "loss": 0.1904,
+ "step": 12470
+ },
+ {
+ "epoch": 33.98092643051771,
+ "grad_norm": 4.191485404968262,
+ "learning_rate": 1.5374412223189467e-05,
+ "loss": 0.1108,
+ "step": 12471
+ },
+ {
+ "epoch": 33.98365122615804,
+ "grad_norm": 4.481583118438721,
+ "learning_rate": 1.537366799346754e-05,
+ "loss": 0.3122,
+ "step": 12472
+ },
+ {
+ "epoch": 33.986376021798364,
+ "grad_norm": 4.4168219566345215,
+ "learning_rate": 1.537292372189566e-05,
+ "loss": 0.2341,
+ "step": 12473
+ },
+ {
+ "epoch": 33.98910081743869,
+ "grad_norm": 4.347963333129883,
+ "learning_rate": 1.5372179408479624e-05,
+ "loss": 0.1373,
+ "step": 12474
+ },
+ {
+ "epoch": 33.991825613079016,
+ "grad_norm": 4.9896721839904785,
+ "learning_rate": 1.5371435053225222e-05,
+ "loss": 0.2113,
+ "step": 12475
+ },
+ {
+ "epoch": 33.99455040871935,
+ "grad_norm": 3.8954334259033203,
+ "learning_rate": 1.5370690656138265e-05,
+ "loss": 0.2096,
+ "step": 12476
+ },
+ {
+ "epoch": 33.997275204359674,
+ "grad_norm": 4.1769232749938965,
+ "learning_rate": 1.5369946217224536e-05,
+ "loss": 0.1729,
+ "step": 12477
+ },
+ {
+ "epoch": 34.0,
+ "grad_norm": 5.055260181427002,
+ "learning_rate": 1.536920173648984e-05,
+ "loss": 0.3162,
+ "step": 12478
+ },
+ {
+ "epoch": 34.002724795640326,
+ "grad_norm": 3.198456048965454,
+ "learning_rate": 1.5368457213939973e-05,
+ "loss": 0.116,
+ "step": 12479
+ },
+ {
+ "epoch": 34.00544959128065,
+ "grad_norm": 4.011242389678955,
+ "learning_rate": 1.5367712649580736e-05,
+ "loss": 0.1857,
+ "step": 12480
+ },
+ {
+ "epoch": 34.00817438692098,
+ "grad_norm": 4.688368797302246,
+ "learning_rate": 1.5366968043417926e-05,
+ "loss": 0.1695,
+ "step": 12481
+ },
+ {
+ "epoch": 34.01089918256131,
+ "grad_norm": 3.932131290435791,
+ "learning_rate": 1.5366223395457337e-05,
+ "loss": 0.1707,
+ "step": 12482
+ },
+ {
+ "epoch": 34.013623978201636,
+ "grad_norm": 3.081482410430908,
+ "learning_rate": 1.536547870570478e-05,
+ "loss": 0.1703,
+ "step": 12483
+ },
+ {
+ "epoch": 34.01634877384196,
+ "grad_norm": 3.695115804672241,
+ "learning_rate": 1.5364733974166042e-05,
+ "loss": 0.2575,
+ "step": 12484
+ },
+ {
+ "epoch": 34.01907356948229,
+ "grad_norm": 4.686526298522949,
+ "learning_rate": 1.536398920084693e-05,
+ "loss": 0.2342,
+ "step": 12485
+ },
+ {
+ "epoch": 34.02179836512261,
+ "grad_norm": 4.516251564025879,
+ "learning_rate": 1.5363244385753244e-05,
+ "loss": 0.1466,
+ "step": 12486
+ },
+ {
+ "epoch": 34.02452316076294,
+ "grad_norm": 3.8072540760040283,
+ "learning_rate": 1.5362499528890782e-05,
+ "loss": 0.1192,
+ "step": 12487
+ },
+ {
+ "epoch": 34.02724795640327,
+ "grad_norm": 3.537353515625,
+ "learning_rate": 1.5361754630265347e-05,
+ "loss": 0.2,
+ "step": 12488
+ },
+ {
+ "epoch": 34.0299727520436,
+ "grad_norm": 3.1607940196990967,
+ "learning_rate": 1.5361009689882737e-05,
+ "loss": 0.0969,
+ "step": 12489
+ },
+ {
+ "epoch": 34.032697547683924,
+ "grad_norm": 4.175352573394775,
+ "learning_rate": 1.5360264707748758e-05,
+ "loss": 0.2824,
+ "step": 12490
+ },
+ {
+ "epoch": 34.03542234332425,
+ "grad_norm": 3.920475959777832,
+ "learning_rate": 1.5359519683869207e-05,
+ "loss": 0.1415,
+ "step": 12491
+ },
+ {
+ "epoch": 34.038147138964575,
+ "grad_norm": 3.954545021057129,
+ "learning_rate": 1.5358774618249895e-05,
+ "loss": 0.2449,
+ "step": 12492
+ },
+ {
+ "epoch": 34.0408719346049,
+ "grad_norm": 4.13594913482666,
+ "learning_rate": 1.5358029510896616e-05,
+ "loss": 0.2605,
+ "step": 12493
+ },
+ {
+ "epoch": 34.043596730245234,
+ "grad_norm": 3.6352548599243164,
+ "learning_rate": 1.5357284361815178e-05,
+ "loss": 0.1938,
+ "step": 12494
+ },
+ {
+ "epoch": 34.04632152588556,
+ "grad_norm": 3.6608757972717285,
+ "learning_rate": 1.535653917101138e-05,
+ "loss": 0.156,
+ "step": 12495
+ },
+ {
+ "epoch": 34.049046321525886,
+ "grad_norm": 4.087778091430664,
+ "learning_rate": 1.5355793938491025e-05,
+ "loss": 0.1299,
+ "step": 12496
+ },
+ {
+ "epoch": 34.05177111716621,
+ "grad_norm": 4.239498138427734,
+ "learning_rate": 1.535504866425992e-05,
+ "loss": 0.1684,
+ "step": 12497
+ },
+ {
+ "epoch": 34.05449591280654,
+ "grad_norm": 4.451730251312256,
+ "learning_rate": 1.5354303348323873e-05,
+ "loss": 0.1979,
+ "step": 12498
+ },
+ {
+ "epoch": 34.05722070844686,
+ "grad_norm": 8.924007415771484,
+ "learning_rate": 1.5353557990688675e-05,
+ "loss": 0.1334,
+ "step": 12499
+ },
+ {
+ "epoch": 34.059945504087196,
+ "grad_norm": 4.949765682220459,
+ "learning_rate": 1.5352812591360148e-05,
+ "loss": 0.2388,
+ "step": 12500
+ },
+ {
+ "epoch": 34.06267029972752,
+ "grad_norm": 4.262774467468262,
+ "learning_rate": 1.5352067150344084e-05,
+ "loss": 0.3221,
+ "step": 12501
+ },
+ {
+ "epoch": 34.06539509536785,
+ "grad_norm": 4.163339614868164,
+ "learning_rate": 1.5351321667646298e-05,
+ "loss": 0.1734,
+ "step": 12502
+ },
+ {
+ "epoch": 34.06811989100817,
+ "grad_norm": 4.292003631591797,
+ "learning_rate": 1.5350576143272587e-05,
+ "loss": 0.1678,
+ "step": 12503
+ },
+ {
+ "epoch": 34.0708446866485,
+ "grad_norm": 4.3787055015563965,
+ "learning_rate": 1.534983057722876e-05,
+ "loss": 0.1559,
+ "step": 12504
+ },
+ {
+ "epoch": 34.073569482288825,
+ "grad_norm": 3.808602809906006,
+ "learning_rate": 1.5349084969520628e-05,
+ "loss": 0.18,
+ "step": 12505
+ },
+ {
+ "epoch": 34.07629427792916,
+ "grad_norm": 5.020304203033447,
+ "learning_rate": 1.5348339320153992e-05,
+ "loss": 0.237,
+ "step": 12506
+ },
+ {
+ "epoch": 34.079019073569484,
+ "grad_norm": 3.621826171875,
+ "learning_rate": 1.5347593629134663e-05,
+ "loss": 0.1399,
+ "step": 12507
+ },
+ {
+ "epoch": 34.08174386920981,
+ "grad_norm": 3.9507155418395996,
+ "learning_rate": 1.534684789646845e-05,
+ "loss": 0.1882,
+ "step": 12508
+ },
+ {
+ "epoch": 34.084468664850135,
+ "grad_norm": 8.580351829528809,
+ "learning_rate": 1.5346102122161152e-05,
+ "loss": 0.2221,
+ "step": 12509
+ },
+ {
+ "epoch": 34.08719346049046,
+ "grad_norm": 6.1939616203308105,
+ "learning_rate": 1.5345356306218586e-05,
+ "loss": 0.2564,
+ "step": 12510
+ },
+ {
+ "epoch": 34.08991825613079,
+ "grad_norm": 4.0997700691223145,
+ "learning_rate": 1.5344610448646555e-05,
+ "loss": 0.1634,
+ "step": 12511
+ },
+ {
+ "epoch": 34.09264305177112,
+ "grad_norm": 5.027294158935547,
+ "learning_rate": 1.534386454945087e-05,
+ "loss": 0.272,
+ "step": 12512
+ },
+ {
+ "epoch": 34.095367847411445,
+ "grad_norm": 4.441617012023926,
+ "learning_rate": 1.5343118608637345e-05,
+ "loss": 0.2057,
+ "step": 12513
+ },
+ {
+ "epoch": 34.09809264305177,
+ "grad_norm": 4.461741924285889,
+ "learning_rate": 1.534237262621178e-05,
+ "loss": 0.2287,
+ "step": 12514
+ },
+ {
+ "epoch": 34.1008174386921,
+ "grad_norm": 3.8789846897125244,
+ "learning_rate": 1.5341626602179988e-05,
+ "loss": 0.201,
+ "step": 12515
+ },
+ {
+ "epoch": 34.10354223433242,
+ "grad_norm": 3.3091235160827637,
+ "learning_rate": 1.534088053654778e-05,
+ "loss": 0.2701,
+ "step": 12516
+ },
+ {
+ "epoch": 34.10626702997275,
+ "grad_norm": 3.4542226791381836,
+ "learning_rate": 1.534013442932097e-05,
+ "loss": 0.1929,
+ "step": 12517
+ },
+ {
+ "epoch": 34.10899182561308,
+ "grad_norm": 3.87680983543396,
+ "learning_rate": 1.533938828050536e-05,
+ "loss": 0.1738,
+ "step": 12518
+ },
+ {
+ "epoch": 34.11171662125341,
+ "grad_norm": 5.251006603240967,
+ "learning_rate": 1.533864209010677e-05,
+ "loss": 0.3465,
+ "step": 12519
+ },
+ {
+ "epoch": 34.11444141689373,
+ "grad_norm": 3.663482666015625,
+ "learning_rate": 1.5337895858131004e-05,
+ "loss": 0.1987,
+ "step": 12520
+ },
+ {
+ "epoch": 34.11716621253406,
+ "grad_norm": 3.5725269317626953,
+ "learning_rate": 1.5337149584583878e-05,
+ "loss": 0.3379,
+ "step": 12521
+ },
+ {
+ "epoch": 34.119891008174385,
+ "grad_norm": 4.119115352630615,
+ "learning_rate": 1.5336403269471202e-05,
+ "loss": 0.2662,
+ "step": 12522
+ },
+ {
+ "epoch": 34.12261580381471,
+ "grad_norm": 3.9197216033935547,
+ "learning_rate": 1.533565691279879e-05,
+ "loss": 0.1452,
+ "step": 12523
+ },
+ {
+ "epoch": 34.12534059945504,
+ "grad_norm": 3.410889148712158,
+ "learning_rate": 1.5334910514572453e-05,
+ "loss": 0.1355,
+ "step": 12524
+ },
+ {
+ "epoch": 34.12806539509537,
+ "grad_norm": 4.102633953094482,
+ "learning_rate": 1.5334164074798008e-05,
+ "loss": 0.2208,
+ "step": 12525
+ },
+ {
+ "epoch": 34.130790190735695,
+ "grad_norm": 6.443319797515869,
+ "learning_rate": 1.5333417593481262e-05,
+ "loss": 0.1829,
+ "step": 12526
+ },
+ {
+ "epoch": 34.13351498637602,
+ "grad_norm": 4.575308322906494,
+ "learning_rate": 1.533267107062803e-05,
+ "loss": 0.2931,
+ "step": 12527
+ },
+ {
+ "epoch": 34.13623978201635,
+ "grad_norm": 4.3238444328308105,
+ "learning_rate": 1.5331924506244132e-05,
+ "loss": 0.1434,
+ "step": 12528
+ },
+ {
+ "epoch": 34.13896457765667,
+ "grad_norm": 4.109264373779297,
+ "learning_rate": 1.5331177900335372e-05,
+ "loss": 0.159,
+ "step": 12529
+ },
+ {
+ "epoch": 34.141689373297005,
+ "grad_norm": 6.236820697784424,
+ "learning_rate": 1.5330431252907575e-05,
+ "loss": 0.1067,
+ "step": 12530
+ },
+ {
+ "epoch": 34.14441416893733,
+ "grad_norm": 3.6825740337371826,
+ "learning_rate": 1.5329684563966547e-05,
+ "loss": 0.2055,
+ "step": 12531
+ },
+ {
+ "epoch": 34.14713896457766,
+ "grad_norm": 4.415315628051758,
+ "learning_rate": 1.532893783351811e-05,
+ "loss": 0.2662,
+ "step": 12532
+ },
+ {
+ "epoch": 34.14986376021798,
+ "grad_norm": 3.594439744949341,
+ "learning_rate": 1.5328191061568076e-05,
+ "loss": 0.145,
+ "step": 12533
+ },
+ {
+ "epoch": 34.15258855585831,
+ "grad_norm": 10.084063529968262,
+ "learning_rate": 1.532744424812226e-05,
+ "loss": 0.1324,
+ "step": 12534
+ },
+ {
+ "epoch": 34.155313351498634,
+ "grad_norm": 3.973336696624756,
+ "learning_rate": 1.5326697393186482e-05,
+ "loss": 0.295,
+ "step": 12535
+ },
+ {
+ "epoch": 34.15803814713897,
+ "grad_norm": 5.704372882843018,
+ "learning_rate": 1.5325950496766557e-05,
+ "loss": 0.098,
+ "step": 12536
+ },
+ {
+ "epoch": 34.16076294277929,
+ "grad_norm": 3.9358956813812256,
+ "learning_rate": 1.53252035588683e-05,
+ "loss": 0.1727,
+ "step": 12537
+ },
+ {
+ "epoch": 34.16348773841962,
+ "grad_norm": 4.4942731857299805,
+ "learning_rate": 1.5324456579497525e-05,
+ "loss": 0.248,
+ "step": 12538
+ },
+ {
+ "epoch": 34.166212534059945,
+ "grad_norm": 3.9757235050201416,
+ "learning_rate": 1.5323709558660057e-05,
+ "loss": 0.1996,
+ "step": 12539
+ },
+ {
+ "epoch": 34.16893732970027,
+ "grad_norm": 3.4334442615509033,
+ "learning_rate": 1.5322962496361705e-05,
+ "loss": 0.3729,
+ "step": 12540
+ },
+ {
+ "epoch": 34.171662125340596,
+ "grad_norm": 3.7460391521453857,
+ "learning_rate": 1.53222153926083e-05,
+ "loss": 0.1449,
+ "step": 12541
+ },
+ {
+ "epoch": 34.17438692098093,
+ "grad_norm": 3.470555067062378,
+ "learning_rate": 1.532146824740565e-05,
+ "loss": 0.2098,
+ "step": 12542
+ },
+ {
+ "epoch": 34.177111716621255,
+ "grad_norm": 4.911859035491943,
+ "learning_rate": 1.532072106075958e-05,
+ "loss": 0.21,
+ "step": 12543
+ },
+ {
+ "epoch": 34.17983651226158,
+ "grad_norm": 4.185249328613281,
+ "learning_rate": 1.53199738326759e-05,
+ "loss": 0.2061,
+ "step": 12544
+ },
+ {
+ "epoch": 34.182561307901906,
+ "grad_norm": 4.487423419952393,
+ "learning_rate": 1.5319226563160438e-05,
+ "loss": 0.223,
+ "step": 12545
+ },
+ {
+ "epoch": 34.18528610354223,
+ "grad_norm": 3.7603137493133545,
+ "learning_rate": 1.531847925221901e-05,
+ "loss": 0.3064,
+ "step": 12546
+ },
+ {
+ "epoch": 34.18801089918256,
+ "grad_norm": 3.786120653152466,
+ "learning_rate": 1.5317731899857434e-05,
+ "loss": 0.184,
+ "step": 12547
+ },
+ {
+ "epoch": 34.19073569482289,
+ "grad_norm": 3.421546697616577,
+ "learning_rate": 1.5316984506081537e-05,
+ "loss": 0.1692,
+ "step": 12548
+ },
+ {
+ "epoch": 34.19346049046322,
+ "grad_norm": 3.9498181343078613,
+ "learning_rate": 1.5316237070897134e-05,
+ "loss": 0.14,
+ "step": 12549
+ },
+ {
+ "epoch": 34.19618528610354,
+ "grad_norm": 4.566798210144043,
+ "learning_rate": 1.5315489594310048e-05,
+ "loss": 0.1826,
+ "step": 12550
+ },
+ {
+ "epoch": 34.19891008174387,
+ "grad_norm": 3.644583225250244,
+ "learning_rate": 1.53147420763261e-05,
+ "loss": 0.4156,
+ "step": 12551
+ },
+ {
+ "epoch": 34.201634877384194,
+ "grad_norm": 6.18720006942749,
+ "learning_rate": 1.5313994516951112e-05,
+ "loss": 0.1711,
+ "step": 12552
+ },
+ {
+ "epoch": 34.20435967302452,
+ "grad_norm": 3.3769068717956543,
+ "learning_rate": 1.5313246916190907e-05,
+ "loss": 0.1811,
+ "step": 12553
+ },
+ {
+ "epoch": 34.20708446866485,
+ "grad_norm": 3.8051679134368896,
+ "learning_rate": 1.5312499274051303e-05,
+ "loss": 0.1361,
+ "step": 12554
+ },
+ {
+ "epoch": 34.20980926430518,
+ "grad_norm": 3.584310293197632,
+ "learning_rate": 1.531175159053813e-05,
+ "loss": 0.2214,
+ "step": 12555
+ },
+ {
+ "epoch": 34.212534059945504,
+ "grad_norm": 4.188620567321777,
+ "learning_rate": 1.5311003865657204e-05,
+ "loss": 0.1645,
+ "step": 12556
+ },
+ {
+ "epoch": 34.21525885558583,
+ "grad_norm": 4.450810432434082,
+ "learning_rate": 1.531025609941435e-05,
+ "loss": 0.1708,
+ "step": 12557
+ },
+ {
+ "epoch": 34.217983651226156,
+ "grad_norm": 2.710350513458252,
+ "learning_rate": 1.5309508291815395e-05,
+ "loss": 0.1223,
+ "step": 12558
+ },
+ {
+ "epoch": 34.22070844686648,
+ "grad_norm": 4.181339740753174,
+ "learning_rate": 1.530876044286616e-05,
+ "loss": 0.2155,
+ "step": 12559
+ },
+ {
+ "epoch": 34.223433242506815,
+ "grad_norm": 3.1210267543792725,
+ "learning_rate": 1.5308012552572466e-05,
+ "loss": 0.1521,
+ "step": 12560
+ },
+ {
+ "epoch": 34.22615803814714,
+ "grad_norm": 14.8631591796875,
+ "learning_rate": 1.5307264620940145e-05,
+ "loss": 0.173,
+ "step": 12561
+ },
+ {
+ "epoch": 34.228882833787466,
+ "grad_norm": 3.306499719619751,
+ "learning_rate": 1.5306516647975014e-05,
+ "loss": 0.1337,
+ "step": 12562
+ },
+ {
+ "epoch": 34.23160762942779,
+ "grad_norm": 3.772850513458252,
+ "learning_rate": 1.5305768633682906e-05,
+ "loss": 0.2173,
+ "step": 12563
+ },
+ {
+ "epoch": 34.23433242506812,
+ "grad_norm": 4.318735122680664,
+ "learning_rate": 1.530502057806964e-05,
+ "loss": 0.1411,
+ "step": 12564
+ },
+ {
+ "epoch": 34.237057220708444,
+ "grad_norm": 4.662812232971191,
+ "learning_rate": 1.5304272481141045e-05,
+ "loss": 0.1119,
+ "step": 12565
+ },
+ {
+ "epoch": 34.23978201634878,
+ "grad_norm": 3.875502109527588,
+ "learning_rate": 1.5303524342902948e-05,
+ "loss": 0.1459,
+ "step": 12566
+ },
+ {
+ "epoch": 34.2425068119891,
+ "grad_norm": 4.650404453277588,
+ "learning_rate": 1.5302776163361173e-05,
+ "loss": 0.1676,
+ "step": 12567
+ },
+ {
+ "epoch": 34.24523160762943,
+ "grad_norm": 4.080883502960205,
+ "learning_rate": 1.530202794252155e-05,
+ "loss": 0.2421,
+ "step": 12568
+ },
+ {
+ "epoch": 34.247956403269754,
+ "grad_norm": 3.7368338108062744,
+ "learning_rate": 1.5301279680389902e-05,
+ "loss": 0.2445,
+ "step": 12569
+ },
+ {
+ "epoch": 34.25068119891008,
+ "grad_norm": 3.651150703430176,
+ "learning_rate": 1.530053137697206e-05,
+ "loss": 0.1176,
+ "step": 12570
+ },
+ {
+ "epoch": 34.253405994550405,
+ "grad_norm": 4.309530735015869,
+ "learning_rate": 1.5299783032273848e-05,
+ "loss": 0.2971,
+ "step": 12571
+ },
+ {
+ "epoch": 34.25613079019074,
+ "grad_norm": 4.169337272644043,
+ "learning_rate": 1.5299034646301097e-05,
+ "loss": 0.1721,
+ "step": 12572
+ },
+ {
+ "epoch": 34.258855585831064,
+ "grad_norm": 14.111270904541016,
+ "learning_rate": 1.5298286219059634e-05,
+ "loss": 0.165,
+ "step": 12573
+ },
+ {
+ "epoch": 34.26158038147139,
+ "grad_norm": 3.6503474712371826,
+ "learning_rate": 1.5297537750555286e-05,
+ "loss": 0.244,
+ "step": 12574
+ },
+ {
+ "epoch": 34.264305177111716,
+ "grad_norm": 3.6323740482330322,
+ "learning_rate": 1.529678924079389e-05,
+ "loss": 0.1214,
+ "step": 12575
+ },
+ {
+ "epoch": 34.26702997275204,
+ "grad_norm": 3.412194013595581,
+ "learning_rate": 1.5296040689781267e-05,
+ "loss": 0.113,
+ "step": 12576
+ },
+ {
+ "epoch": 34.26975476839237,
+ "grad_norm": 3.409621477127075,
+ "learning_rate": 1.529529209752325e-05,
+ "loss": 0.1385,
+ "step": 12577
+ },
+ {
+ "epoch": 34.2724795640327,
+ "grad_norm": 4.03713846206665,
+ "learning_rate": 1.529454346402567e-05,
+ "loss": 0.2036,
+ "step": 12578
+ },
+ {
+ "epoch": 34.275204359673026,
+ "grad_norm": 3.750972270965576,
+ "learning_rate": 1.529379478929436e-05,
+ "loss": 0.3402,
+ "step": 12579
+ },
+ {
+ "epoch": 34.27792915531335,
+ "grad_norm": 5.637497425079346,
+ "learning_rate": 1.5293046073335137e-05,
+ "loss": 0.1995,
+ "step": 12580
+ },
+ {
+ "epoch": 34.28065395095368,
+ "grad_norm": 5.101443767547607,
+ "learning_rate": 1.529229731615385e-05,
+ "loss": 0.1806,
+ "step": 12581
+ },
+ {
+ "epoch": 34.283378746594,
+ "grad_norm": 4.728841781616211,
+ "learning_rate": 1.529154851775631e-05,
+ "loss": 0.1261,
+ "step": 12582
+ },
+ {
+ "epoch": 34.28610354223433,
+ "grad_norm": 5.042652130126953,
+ "learning_rate": 1.529079967814837e-05,
+ "loss": 0.2064,
+ "step": 12583
+ },
+ {
+ "epoch": 34.28882833787466,
+ "grad_norm": 3.8771603107452393,
+ "learning_rate": 1.529005079733585e-05,
+ "loss": 0.2054,
+ "step": 12584
+ },
+ {
+ "epoch": 34.29155313351499,
+ "grad_norm": 5.0003767013549805,
+ "learning_rate": 1.528930187532459e-05,
+ "loss": 0.2342,
+ "step": 12585
+ },
+ {
+ "epoch": 34.294277929155314,
+ "grad_norm": 4.707497596740723,
+ "learning_rate": 1.528855291212041e-05,
+ "loss": 0.253,
+ "step": 12586
+ },
+ {
+ "epoch": 34.29700272479564,
+ "grad_norm": 4.09159517288208,
+ "learning_rate": 1.5287803907729154e-05,
+ "loss": 0.2065,
+ "step": 12587
+ },
+ {
+ "epoch": 34.299727520435965,
+ "grad_norm": 4.3349809646606445,
+ "learning_rate": 1.5287054862156648e-05,
+ "loss": 0.2956,
+ "step": 12588
+ },
+ {
+ "epoch": 34.30245231607629,
+ "grad_norm": 7.449347496032715,
+ "learning_rate": 1.5286305775408732e-05,
+ "loss": 0.2752,
+ "step": 12589
+ },
+ {
+ "epoch": 34.305177111716624,
+ "grad_norm": 4.333142280578613,
+ "learning_rate": 1.5285556647491237e-05,
+ "loss": 0.194,
+ "step": 12590
+ },
+ {
+ "epoch": 34.30790190735695,
+ "grad_norm": 3.976889133453369,
+ "learning_rate": 1.5284807478409994e-05,
+ "loss": 0.1163,
+ "step": 12591
+ },
+ {
+ "epoch": 34.310626702997276,
+ "grad_norm": 4.640392780303955,
+ "learning_rate": 1.528405826817084e-05,
+ "loss": 0.1671,
+ "step": 12592
+ },
+ {
+ "epoch": 34.3133514986376,
+ "grad_norm": 3.893242835998535,
+ "learning_rate": 1.5283309016779615e-05,
+ "loss": 0.2715,
+ "step": 12593
+ },
+ {
+ "epoch": 34.31607629427793,
+ "grad_norm": 4.317715167999268,
+ "learning_rate": 1.5282559724242145e-05,
+ "loss": 0.1618,
+ "step": 12594
+ },
+ {
+ "epoch": 34.31880108991825,
+ "grad_norm": 3.8455636501312256,
+ "learning_rate": 1.5281810390564274e-05,
+ "loss": 0.1799,
+ "step": 12595
+ },
+ {
+ "epoch": 34.321525885558586,
+ "grad_norm": 5.478277206420898,
+ "learning_rate": 1.528106101575183e-05,
+ "loss": 0.2256,
+ "step": 12596
+ },
+ {
+ "epoch": 34.32425068119891,
+ "grad_norm": 4.111917018890381,
+ "learning_rate": 1.5280311599810654e-05,
+ "loss": 0.1329,
+ "step": 12597
+ },
+ {
+ "epoch": 34.32697547683924,
+ "grad_norm": 4.068848609924316,
+ "learning_rate": 1.527956214274658e-05,
+ "loss": 0.171,
+ "step": 12598
+ },
+ {
+ "epoch": 34.32970027247956,
+ "grad_norm": 5.5191650390625,
+ "learning_rate": 1.527881264456545e-05,
+ "loss": 0.1573,
+ "step": 12599
+ },
+ {
+ "epoch": 34.33242506811989,
+ "grad_norm": 4.122544765472412,
+ "learning_rate": 1.5278063105273094e-05,
+ "loss": 0.1905,
+ "step": 12600
+ },
+ {
+ "epoch": 34.335149863760215,
+ "grad_norm": 4.0446600914001465,
+ "learning_rate": 1.527731352487535e-05,
+ "loss": 0.1791,
+ "step": 12601
+ },
+ {
+ "epoch": 34.33787465940055,
+ "grad_norm": 4.741742134094238,
+ "learning_rate": 1.5276563903378064e-05,
+ "loss": 0.1846,
+ "step": 12602
+ },
+ {
+ "epoch": 34.34059945504087,
+ "grad_norm": 3.3646228313446045,
+ "learning_rate": 1.5275814240787064e-05,
+ "loss": 0.1729,
+ "step": 12603
+ },
+ {
+ "epoch": 34.3433242506812,
+ "grad_norm": 3.320540189743042,
+ "learning_rate": 1.527506453710819e-05,
+ "loss": 0.3739,
+ "step": 12604
+ },
+ {
+ "epoch": 34.346049046321525,
+ "grad_norm": 3.6020429134368896,
+ "learning_rate": 1.527431479234729e-05,
+ "loss": 0.2332,
+ "step": 12605
+ },
+ {
+ "epoch": 34.34877384196185,
+ "grad_norm": 4.07051944732666,
+ "learning_rate": 1.527356500651019e-05,
+ "loss": 0.3248,
+ "step": 12606
+ },
+ {
+ "epoch": 34.35149863760218,
+ "grad_norm": 3.802894353866577,
+ "learning_rate": 1.5272815179602743e-05,
+ "loss": 0.2896,
+ "step": 12607
+ },
+ {
+ "epoch": 34.35422343324251,
+ "grad_norm": 6.0409440994262695,
+ "learning_rate": 1.5272065311630773e-05,
+ "loss": 0.1996,
+ "step": 12608
+ },
+ {
+ "epoch": 34.356948228882835,
+ "grad_norm": 3.3933117389678955,
+ "learning_rate": 1.5271315402600132e-05,
+ "loss": 0.1207,
+ "step": 12609
+ },
+ {
+ "epoch": 34.35967302452316,
+ "grad_norm": 3.7280657291412354,
+ "learning_rate": 1.5270565452516656e-05,
+ "loss": 0.1775,
+ "step": 12610
+ },
+ {
+ "epoch": 34.36239782016349,
+ "grad_norm": 6.073631286621094,
+ "learning_rate": 1.5269815461386187e-05,
+ "loss": 0.1542,
+ "step": 12611
+ },
+ {
+ "epoch": 34.36512261580381,
+ "grad_norm": 4.037694454193115,
+ "learning_rate": 1.5269065429214563e-05,
+ "loss": 0.1361,
+ "step": 12612
+ },
+ {
+ "epoch": 34.36784741144414,
+ "grad_norm": 4.823380470275879,
+ "learning_rate": 1.5268315356007624e-05,
+ "loss": 0.1313,
+ "step": 12613
+ },
+ {
+ "epoch": 34.37057220708447,
+ "grad_norm": 3.2748751640319824,
+ "learning_rate": 1.526756524177122e-05,
+ "loss": 0.2138,
+ "step": 12614
+ },
+ {
+ "epoch": 34.3732970027248,
+ "grad_norm": 5.339231491088867,
+ "learning_rate": 1.5266815086511187e-05,
+ "loss": 0.2687,
+ "step": 12615
+ },
+ {
+ "epoch": 34.37602179836512,
+ "grad_norm": 5.644565105438232,
+ "learning_rate": 1.5266064890233363e-05,
+ "loss": 0.2548,
+ "step": 12616
+ },
+ {
+ "epoch": 34.37874659400545,
+ "grad_norm": 4.968973636627197,
+ "learning_rate": 1.52653146529436e-05,
+ "loss": 0.1416,
+ "step": 12617
+ },
+ {
+ "epoch": 34.381471389645775,
+ "grad_norm": 4.629998207092285,
+ "learning_rate": 1.5264564374647732e-05,
+ "loss": 0.3247,
+ "step": 12618
+ },
+ {
+ "epoch": 34.3841961852861,
+ "grad_norm": 3.6394248008728027,
+ "learning_rate": 1.526381405535161e-05,
+ "loss": 0.3274,
+ "step": 12619
+ },
+ {
+ "epoch": 34.38692098092643,
+ "grad_norm": 3.7731146812438965,
+ "learning_rate": 1.5263063695061072e-05,
+ "loss": 0.1993,
+ "step": 12620
+ },
+ {
+ "epoch": 34.38964577656676,
+ "grad_norm": 4.900196552276611,
+ "learning_rate": 1.526231329378196e-05,
+ "loss": 0.2648,
+ "step": 12621
+ },
+ {
+ "epoch": 34.392370572207085,
+ "grad_norm": 4.408911228179932,
+ "learning_rate": 1.5261562851520127e-05,
+ "loss": 0.2819,
+ "step": 12622
+ },
+ {
+ "epoch": 34.39509536784741,
+ "grad_norm": 4.449269771575928,
+ "learning_rate": 1.526081236828141e-05,
+ "loss": 0.1456,
+ "step": 12623
+ },
+ {
+ "epoch": 34.39782016348774,
+ "grad_norm": 3.9567222595214844,
+ "learning_rate": 1.5260061844071655e-05,
+ "loss": 0.1483,
+ "step": 12624
+ },
+ {
+ "epoch": 34.40054495912806,
+ "grad_norm": 4.529531478881836,
+ "learning_rate": 1.5259311278896706e-05,
+ "loss": 0.3099,
+ "step": 12625
+ },
+ {
+ "epoch": 34.403269754768395,
+ "grad_norm": 4.255303382873535,
+ "learning_rate": 1.525856067276241e-05,
+ "loss": 0.2217,
+ "step": 12626
+ },
+ {
+ "epoch": 34.40599455040872,
+ "grad_norm": 3.6326608657836914,
+ "learning_rate": 1.5257810025674613e-05,
+ "loss": 0.1392,
+ "step": 12627
+ },
+ {
+ "epoch": 34.40871934604905,
+ "grad_norm": 4.350475311279297,
+ "learning_rate": 1.5257059337639164e-05,
+ "loss": 0.1879,
+ "step": 12628
+ },
+ {
+ "epoch": 34.41144414168937,
+ "grad_norm": 3.692085027694702,
+ "learning_rate": 1.5256308608661904e-05,
+ "loss": 0.1259,
+ "step": 12629
+ },
+ {
+ "epoch": 34.4141689373297,
+ "grad_norm": 4.054514408111572,
+ "learning_rate": 1.5255557838748683e-05,
+ "loss": 0.1503,
+ "step": 12630
+ },
+ {
+ "epoch": 34.416893732970024,
+ "grad_norm": 3.4629414081573486,
+ "learning_rate": 1.5254807027905342e-05,
+ "loss": 0.1279,
+ "step": 12631
+ },
+ {
+ "epoch": 34.41961852861036,
+ "grad_norm": 4.080844402313232,
+ "learning_rate": 1.5254056176137742e-05,
+ "loss": 0.1768,
+ "step": 12632
+ },
+ {
+ "epoch": 34.42234332425068,
+ "grad_norm": 3.926633596420288,
+ "learning_rate": 1.5253305283451714e-05,
+ "loss": 0.2522,
+ "step": 12633
+ },
+ {
+ "epoch": 34.42506811989101,
+ "grad_norm": 4.453307151794434,
+ "learning_rate": 1.5252554349853114e-05,
+ "loss": 0.2162,
+ "step": 12634
+ },
+ {
+ "epoch": 34.427792915531334,
+ "grad_norm": 3.9146244525909424,
+ "learning_rate": 1.5251803375347792e-05,
+ "loss": 0.1542,
+ "step": 12635
+ },
+ {
+ "epoch": 34.43051771117166,
+ "grad_norm": 4.472159385681152,
+ "learning_rate": 1.5251052359941593e-05,
+ "loss": 0.3617,
+ "step": 12636
+ },
+ {
+ "epoch": 34.433242506811986,
+ "grad_norm": 3.7739126682281494,
+ "learning_rate": 1.5250301303640368e-05,
+ "loss": 0.1683,
+ "step": 12637
+ },
+ {
+ "epoch": 34.43596730245232,
+ "grad_norm": 4.374064922332764,
+ "learning_rate": 1.5249550206449968e-05,
+ "loss": 0.2868,
+ "step": 12638
+ },
+ {
+ "epoch": 34.438692098092645,
+ "grad_norm": 12.406790733337402,
+ "learning_rate": 1.5248799068376236e-05,
+ "loss": 0.3128,
+ "step": 12639
+ },
+ {
+ "epoch": 34.44141689373297,
+ "grad_norm": 4.499368190765381,
+ "learning_rate": 1.5248047889425026e-05,
+ "loss": 0.281,
+ "step": 12640
+ },
+ {
+ "epoch": 34.444141689373296,
+ "grad_norm": 3.74013090133667,
+ "learning_rate": 1.5247296669602189e-05,
+ "loss": 0.2336,
+ "step": 12641
+ },
+ {
+ "epoch": 34.44686648501362,
+ "grad_norm": 3.7118895053863525,
+ "learning_rate": 1.5246545408913575e-05,
+ "loss": 0.1346,
+ "step": 12642
+ },
+ {
+ "epoch": 34.44959128065395,
+ "grad_norm": 4.3404951095581055,
+ "learning_rate": 1.5245794107365034e-05,
+ "loss": 0.1747,
+ "step": 12643
+ },
+ {
+ "epoch": 34.45231607629428,
+ "grad_norm": 3.7151355743408203,
+ "learning_rate": 1.5245042764962416e-05,
+ "loss": 0.2351,
+ "step": 12644
+ },
+ {
+ "epoch": 34.45504087193461,
+ "grad_norm": 3.6569836139678955,
+ "learning_rate": 1.5244291381711574e-05,
+ "loss": 0.1561,
+ "step": 12645
+ },
+ {
+ "epoch": 34.45776566757493,
+ "grad_norm": 3.9268338680267334,
+ "learning_rate": 1.5243539957618363e-05,
+ "loss": 0.2826,
+ "step": 12646
+ },
+ {
+ "epoch": 34.46049046321526,
+ "grad_norm": 5.59399938583374,
+ "learning_rate": 1.5242788492688628e-05,
+ "loss": 0.36,
+ "step": 12647
+ },
+ {
+ "epoch": 34.463215258855584,
+ "grad_norm": 3.641062021255493,
+ "learning_rate": 1.5242036986928225e-05,
+ "loss": 0.1985,
+ "step": 12648
+ },
+ {
+ "epoch": 34.46594005449591,
+ "grad_norm": 4.140392303466797,
+ "learning_rate": 1.5241285440343008e-05,
+ "loss": 0.1777,
+ "step": 12649
+ },
+ {
+ "epoch": 34.46866485013624,
+ "grad_norm": 4.402441501617432,
+ "learning_rate": 1.524053385293883e-05,
+ "loss": 0.2275,
+ "step": 12650
+ },
+ {
+ "epoch": 34.47138964577657,
+ "grad_norm": 4.021166801452637,
+ "learning_rate": 1.5239782224721537e-05,
+ "loss": 0.2235,
+ "step": 12651
+ },
+ {
+ "epoch": 34.474114441416894,
+ "grad_norm": 6.489063739776611,
+ "learning_rate": 1.5239030555696995e-05,
+ "loss": 0.183,
+ "step": 12652
+ },
+ {
+ "epoch": 34.47683923705722,
+ "grad_norm": 4.218730449676514,
+ "learning_rate": 1.5238278845871048e-05,
+ "loss": 0.3355,
+ "step": 12653
+ },
+ {
+ "epoch": 34.479564032697546,
+ "grad_norm": 5.156269073486328,
+ "learning_rate": 1.5237527095249558e-05,
+ "loss": 0.1816,
+ "step": 12654
+ },
+ {
+ "epoch": 34.48228882833787,
+ "grad_norm": 4.21729040145874,
+ "learning_rate": 1.5236775303838371e-05,
+ "loss": 0.1741,
+ "step": 12655
+ },
+ {
+ "epoch": 34.485013623978205,
+ "grad_norm": 4.3573737144470215,
+ "learning_rate": 1.5236023471643352e-05,
+ "loss": 0.153,
+ "step": 12656
+ },
+ {
+ "epoch": 34.48773841961853,
+ "grad_norm": 3.594676971435547,
+ "learning_rate": 1.5235271598670345e-05,
+ "loss": 0.3212,
+ "step": 12657
+ },
+ {
+ "epoch": 34.490463215258856,
+ "grad_norm": 3.8681094646453857,
+ "learning_rate": 1.5234519684925215e-05,
+ "loss": 0.1807,
+ "step": 12658
+ },
+ {
+ "epoch": 34.49318801089918,
+ "grad_norm": 3.938227891921997,
+ "learning_rate": 1.5233767730413811e-05,
+ "loss": 0.1219,
+ "step": 12659
+ },
+ {
+ "epoch": 34.49591280653951,
+ "grad_norm": 3.7466773986816406,
+ "learning_rate": 1.5233015735141995e-05,
+ "loss": 0.1757,
+ "step": 12660
+ },
+ {
+ "epoch": 34.49863760217983,
+ "grad_norm": 15.543627738952637,
+ "learning_rate": 1.5232263699115616e-05,
+ "loss": 0.2554,
+ "step": 12661
+ },
+ {
+ "epoch": 34.50136239782017,
+ "grad_norm": 5.01516056060791,
+ "learning_rate": 1.5231511622340543e-05,
+ "loss": 0.1397,
+ "step": 12662
+ },
+ {
+ "epoch": 34.50408719346049,
+ "grad_norm": 3.941793203353882,
+ "learning_rate": 1.523075950482262e-05,
+ "loss": 0.1987,
+ "step": 12663
+ },
+ {
+ "epoch": 34.50681198910082,
+ "grad_norm": 3.9315476417541504,
+ "learning_rate": 1.5230007346567717e-05,
+ "loss": 0.1135,
+ "step": 12664
+ },
+ {
+ "epoch": 34.509536784741144,
+ "grad_norm": 3.6421163082122803,
+ "learning_rate": 1.5229255147581678e-05,
+ "loss": 0.4505,
+ "step": 12665
+ },
+ {
+ "epoch": 34.51226158038147,
+ "grad_norm": 5.010616302490234,
+ "learning_rate": 1.5228502907870373e-05,
+ "loss": 0.2266,
+ "step": 12666
+ },
+ {
+ "epoch": 34.514986376021795,
+ "grad_norm": 4.083823204040527,
+ "learning_rate": 1.5227750627439652e-05,
+ "loss": 0.2992,
+ "step": 12667
+ },
+ {
+ "epoch": 34.51771117166213,
+ "grad_norm": 3.294807195663452,
+ "learning_rate": 1.522699830629538e-05,
+ "loss": 0.2774,
+ "step": 12668
+ },
+ {
+ "epoch": 34.520435967302454,
+ "grad_norm": 3.7152440547943115,
+ "learning_rate": 1.522624594444341e-05,
+ "loss": 0.204,
+ "step": 12669
+ },
+ {
+ "epoch": 34.52316076294278,
+ "grad_norm": 4.492642879486084,
+ "learning_rate": 1.5225493541889609e-05,
+ "loss": 0.1308,
+ "step": 12670
+ },
+ {
+ "epoch": 34.525885558583106,
+ "grad_norm": 5.46452522277832,
+ "learning_rate": 1.522474109863983e-05,
+ "loss": 0.201,
+ "step": 12671
+ },
+ {
+ "epoch": 34.52861035422343,
+ "grad_norm": 4.101001739501953,
+ "learning_rate": 1.5223988614699939e-05,
+ "loss": 0.3002,
+ "step": 12672
+ },
+ {
+ "epoch": 34.53133514986376,
+ "grad_norm": 4.087790012359619,
+ "learning_rate": 1.5223236090075788e-05,
+ "loss": 0.1791,
+ "step": 12673
+ },
+ {
+ "epoch": 34.53405994550409,
+ "grad_norm": 3.758167028427124,
+ "learning_rate": 1.5222483524773246e-05,
+ "loss": 0.2958,
+ "step": 12674
+ },
+ {
+ "epoch": 34.536784741144416,
+ "grad_norm": 4.4005231857299805,
+ "learning_rate": 1.5221730918798168e-05,
+ "loss": 0.2562,
+ "step": 12675
+ },
+ {
+ "epoch": 34.53950953678474,
+ "grad_norm": 3.8436391353607178,
+ "learning_rate": 1.522097827215642e-05,
+ "loss": 0.166,
+ "step": 12676
+ },
+ {
+ "epoch": 34.54223433242507,
+ "grad_norm": 4.237051010131836,
+ "learning_rate": 1.5220225584853858e-05,
+ "loss": 0.2745,
+ "step": 12677
+ },
+ {
+ "epoch": 34.54495912806539,
+ "grad_norm": 22.09086799621582,
+ "learning_rate": 1.5219472856896348e-05,
+ "loss": 0.1518,
+ "step": 12678
+ },
+ {
+ "epoch": 34.54768392370572,
+ "grad_norm": 3.4759581089019775,
+ "learning_rate": 1.5218720088289753e-05,
+ "loss": 0.1561,
+ "step": 12679
+ },
+ {
+ "epoch": 34.55040871934605,
+ "grad_norm": 5.294992923736572,
+ "learning_rate": 1.5217967279039933e-05,
+ "loss": 0.3253,
+ "step": 12680
+ },
+ {
+ "epoch": 34.55313351498638,
+ "grad_norm": 4.6869330406188965,
+ "learning_rate": 1.5217214429152752e-05,
+ "loss": 0.1519,
+ "step": 12681
+ },
+ {
+ "epoch": 34.555858310626704,
+ "grad_norm": 4.041265964508057,
+ "learning_rate": 1.5216461538634076e-05,
+ "loss": 0.1818,
+ "step": 12682
+ },
+ {
+ "epoch": 34.55858310626703,
+ "grad_norm": 3.5531158447265625,
+ "learning_rate": 1.521570860748976e-05,
+ "loss": 0.2043,
+ "step": 12683
+ },
+ {
+ "epoch": 34.561307901907355,
+ "grad_norm": 3.737295627593994,
+ "learning_rate": 1.5214955635725677e-05,
+ "loss": 0.3943,
+ "step": 12684
+ },
+ {
+ "epoch": 34.56403269754768,
+ "grad_norm": 5.379068851470947,
+ "learning_rate": 1.5214202623347687e-05,
+ "loss": 0.1297,
+ "step": 12685
+ },
+ {
+ "epoch": 34.566757493188014,
+ "grad_norm": 5.940244197845459,
+ "learning_rate": 1.521344957036165e-05,
+ "loss": 0.1981,
+ "step": 12686
+ },
+ {
+ "epoch": 34.56948228882834,
+ "grad_norm": 4.261266708374023,
+ "learning_rate": 1.5212696476773441e-05,
+ "loss": 0.1466,
+ "step": 12687
+ },
+ {
+ "epoch": 34.572207084468666,
+ "grad_norm": 4.158520221710205,
+ "learning_rate": 1.5211943342588918e-05,
+ "loss": 0.2269,
+ "step": 12688
+ },
+ {
+ "epoch": 34.57493188010899,
+ "grad_norm": 3.9385745525360107,
+ "learning_rate": 1.521119016781395e-05,
+ "loss": 0.2436,
+ "step": 12689
+ },
+ {
+ "epoch": 34.57765667574932,
+ "grad_norm": 4.307703495025635,
+ "learning_rate": 1.5210436952454398e-05,
+ "loss": 0.2881,
+ "step": 12690
+ },
+ {
+ "epoch": 34.58038147138964,
+ "grad_norm": 4.697062015533447,
+ "learning_rate": 1.5209683696516129e-05,
+ "loss": 0.1178,
+ "step": 12691
+ },
+ {
+ "epoch": 34.583106267029976,
+ "grad_norm": 4.253344535827637,
+ "learning_rate": 1.5208930400005016e-05,
+ "loss": 0.1672,
+ "step": 12692
+ },
+ {
+ "epoch": 34.5858310626703,
+ "grad_norm": 7.025396823883057,
+ "learning_rate": 1.5208177062926916e-05,
+ "loss": 0.1564,
+ "step": 12693
+ },
+ {
+ "epoch": 34.58855585831063,
+ "grad_norm": 5.012394428253174,
+ "learning_rate": 1.5207423685287704e-05,
+ "loss": 0.2386,
+ "step": 12694
+ },
+ {
+ "epoch": 34.59128065395095,
+ "grad_norm": 3.4614877700805664,
+ "learning_rate": 1.5206670267093242e-05,
+ "loss": 0.1237,
+ "step": 12695
+ },
+ {
+ "epoch": 34.59400544959128,
+ "grad_norm": 4.61997652053833,
+ "learning_rate": 1.5205916808349401e-05,
+ "loss": 0.1621,
+ "step": 12696
+ },
+ {
+ "epoch": 34.596730245231605,
+ "grad_norm": 3.9014601707458496,
+ "learning_rate": 1.5205163309062048e-05,
+ "loss": 0.2467,
+ "step": 12697
+ },
+ {
+ "epoch": 34.59945504087194,
+ "grad_norm": 3.938969850540161,
+ "learning_rate": 1.5204409769237048e-05,
+ "loss": 0.1788,
+ "step": 12698
+ },
+ {
+ "epoch": 34.60217983651226,
+ "grad_norm": 3.5273020267486572,
+ "learning_rate": 1.5203656188880276e-05,
+ "loss": 0.181,
+ "step": 12699
+ },
+ {
+ "epoch": 34.60490463215259,
+ "grad_norm": 4.074251174926758,
+ "learning_rate": 1.5202902567997592e-05,
+ "loss": 0.2067,
+ "step": 12700
+ },
+ {
+ "epoch": 34.607629427792915,
+ "grad_norm": 5.160400390625,
+ "learning_rate": 1.5202148906594874e-05,
+ "loss": 0.226,
+ "step": 12701
+ },
+ {
+ "epoch": 34.61035422343324,
+ "grad_norm": 4.1027350425720215,
+ "learning_rate": 1.5201395204677986e-05,
+ "loss": 0.2766,
+ "step": 12702
+ },
+ {
+ "epoch": 34.61307901907357,
+ "grad_norm": 4.118257999420166,
+ "learning_rate": 1.5200641462252802e-05,
+ "loss": 0.1454,
+ "step": 12703
+ },
+ {
+ "epoch": 34.6158038147139,
+ "grad_norm": 4.168765544891357,
+ "learning_rate": 1.5199887679325185e-05,
+ "loss": 0.213,
+ "step": 12704
+ },
+ {
+ "epoch": 34.618528610354225,
+ "grad_norm": 4.441951274871826,
+ "learning_rate": 1.5199133855901013e-05,
+ "loss": 0.1948,
+ "step": 12705
+ },
+ {
+ "epoch": 34.62125340599455,
+ "grad_norm": 4.601809501647949,
+ "learning_rate": 1.5198379991986154e-05,
+ "loss": 0.2669,
+ "step": 12706
+ },
+ {
+ "epoch": 34.62397820163488,
+ "grad_norm": 4.380228519439697,
+ "learning_rate": 1.519762608758648e-05,
+ "loss": 0.2786,
+ "step": 12707
+ },
+ {
+ "epoch": 34.6267029972752,
+ "grad_norm": 4.594517707824707,
+ "learning_rate": 1.5196872142707859e-05,
+ "loss": 0.1499,
+ "step": 12708
+ },
+ {
+ "epoch": 34.62942779291553,
+ "grad_norm": 3.8537449836730957,
+ "learning_rate": 1.5196118157356168e-05,
+ "loss": 0.3418,
+ "step": 12709
+ },
+ {
+ "epoch": 34.63215258855586,
+ "grad_norm": 3.8615798950195312,
+ "learning_rate": 1.5195364131537273e-05,
+ "loss": 0.2072,
+ "step": 12710
+ },
+ {
+ "epoch": 34.63487738419619,
+ "grad_norm": 4.479331970214844,
+ "learning_rate": 1.5194610065257052e-05,
+ "loss": 0.27,
+ "step": 12711
+ },
+ {
+ "epoch": 34.63760217983651,
+ "grad_norm": 4.377658367156982,
+ "learning_rate": 1.519385595852137e-05,
+ "loss": 0.2395,
+ "step": 12712
+ },
+ {
+ "epoch": 34.64032697547684,
+ "grad_norm": 4.691817283630371,
+ "learning_rate": 1.519310181133611e-05,
+ "loss": 0.2637,
+ "step": 12713
+ },
+ {
+ "epoch": 34.643051771117165,
+ "grad_norm": 5.890909194946289,
+ "learning_rate": 1.5192347623707138e-05,
+ "loss": 0.3141,
+ "step": 12714
+ },
+ {
+ "epoch": 34.64577656675749,
+ "grad_norm": 4.201766014099121,
+ "learning_rate": 1.5191593395640331e-05,
+ "loss": 0.1819,
+ "step": 12715
+ },
+ {
+ "epoch": 34.64850136239782,
+ "grad_norm": 4.4146833419799805,
+ "learning_rate": 1.5190839127141562e-05,
+ "loss": 0.3867,
+ "step": 12716
+ },
+ {
+ "epoch": 34.65122615803815,
+ "grad_norm": 4.025171279907227,
+ "learning_rate": 1.5190084818216705e-05,
+ "loss": 0.2523,
+ "step": 12717
+ },
+ {
+ "epoch": 34.653950953678475,
+ "grad_norm": 4.280394554138184,
+ "learning_rate": 1.5189330468871633e-05,
+ "loss": 0.1358,
+ "step": 12718
+ },
+ {
+ "epoch": 34.6566757493188,
+ "grad_norm": 3.4281513690948486,
+ "learning_rate": 1.5188576079112225e-05,
+ "loss": 0.2076,
+ "step": 12719
+ },
+ {
+ "epoch": 34.65940054495913,
+ "grad_norm": 5.45773458480835,
+ "learning_rate": 1.5187821648944348e-05,
+ "loss": 0.1699,
+ "step": 12720
+ },
+ {
+ "epoch": 34.66212534059945,
+ "grad_norm": 3.9353418350219727,
+ "learning_rate": 1.5187067178373887e-05,
+ "loss": 0.2072,
+ "step": 12721
+ },
+ {
+ "epoch": 34.664850136239785,
+ "grad_norm": 4.314483642578125,
+ "learning_rate": 1.5186312667406712e-05,
+ "loss": 0.2222,
+ "step": 12722
+ },
+ {
+ "epoch": 34.66757493188011,
+ "grad_norm": 3.772320508956909,
+ "learning_rate": 1.5185558116048704e-05,
+ "loss": 0.134,
+ "step": 12723
+ },
+ {
+ "epoch": 34.67029972752044,
+ "grad_norm": 3.926166534423828,
+ "learning_rate": 1.5184803524305733e-05,
+ "loss": 0.1648,
+ "step": 12724
+ },
+ {
+ "epoch": 34.67302452316076,
+ "grad_norm": 4.154304027557373,
+ "learning_rate": 1.5184048892183683e-05,
+ "loss": 0.1572,
+ "step": 12725
+ },
+ {
+ "epoch": 34.67574931880109,
+ "grad_norm": 3.8988139629364014,
+ "learning_rate": 1.5183294219688423e-05,
+ "loss": 0.1607,
+ "step": 12726
+ },
+ {
+ "epoch": 34.678474114441414,
+ "grad_norm": 4.652369499206543,
+ "learning_rate": 1.5182539506825837e-05,
+ "loss": 0.2279,
+ "step": 12727
+ },
+ {
+ "epoch": 34.68119891008175,
+ "grad_norm": 4.210017681121826,
+ "learning_rate": 1.5181784753601797e-05,
+ "loss": 0.3312,
+ "step": 12728
+ },
+ {
+ "epoch": 34.68392370572207,
+ "grad_norm": 3.467942476272583,
+ "learning_rate": 1.5181029960022187e-05,
+ "loss": 0.2499,
+ "step": 12729
+ },
+ {
+ "epoch": 34.6866485013624,
+ "grad_norm": 3.9710886478424072,
+ "learning_rate": 1.518027512609288e-05,
+ "loss": 0.2196,
+ "step": 12730
+ },
+ {
+ "epoch": 34.689373297002724,
+ "grad_norm": 3.625730276107788,
+ "learning_rate": 1.5179520251819758e-05,
+ "loss": 0.1516,
+ "step": 12731
+ },
+ {
+ "epoch": 34.69209809264305,
+ "grad_norm": 4.209616661071777,
+ "learning_rate": 1.51787653372087e-05,
+ "loss": 0.2615,
+ "step": 12732
+ },
+ {
+ "epoch": 34.694822888283376,
+ "grad_norm": 3.9460699558258057,
+ "learning_rate": 1.5178010382265586e-05,
+ "loss": 0.1219,
+ "step": 12733
+ },
+ {
+ "epoch": 34.69754768392371,
+ "grad_norm": 4.4116926193237305,
+ "learning_rate": 1.517725538699629e-05,
+ "loss": 0.1944,
+ "step": 12734
+ },
+ {
+ "epoch": 34.700272479564035,
+ "grad_norm": 3.990994691848755,
+ "learning_rate": 1.51765003514067e-05,
+ "loss": 0.2125,
+ "step": 12735
+ },
+ {
+ "epoch": 34.70299727520436,
+ "grad_norm": 4.351999759674072,
+ "learning_rate": 1.5175745275502686e-05,
+ "loss": 0.1736,
+ "step": 12736
+ },
+ {
+ "epoch": 34.705722070844686,
+ "grad_norm": 4.3141770362854,
+ "learning_rate": 1.5174990159290137e-05,
+ "loss": 0.2467,
+ "step": 12737
+ },
+ {
+ "epoch": 34.70844686648501,
+ "grad_norm": 6.052056312561035,
+ "learning_rate": 1.5174235002774933e-05,
+ "loss": 0.2534,
+ "step": 12738
+ },
+ {
+ "epoch": 34.71117166212534,
+ "grad_norm": 22.77121353149414,
+ "learning_rate": 1.5173479805962954e-05,
+ "loss": 0.1615,
+ "step": 12739
+ },
+ {
+ "epoch": 34.71389645776567,
+ "grad_norm": 3.5521955490112305,
+ "learning_rate": 1.5172724568860075e-05,
+ "loss": 0.2127,
+ "step": 12740
+ },
+ {
+ "epoch": 34.716621253406,
+ "grad_norm": 10.866859436035156,
+ "learning_rate": 1.517196929147219e-05,
+ "loss": 0.2371,
+ "step": 12741
+ },
+ {
+ "epoch": 34.71934604904632,
+ "grad_norm": 11.723791122436523,
+ "learning_rate": 1.517121397380517e-05,
+ "loss": 0.2333,
+ "step": 12742
+ },
+ {
+ "epoch": 34.72207084468665,
+ "grad_norm": 5.191744804382324,
+ "learning_rate": 1.5170458615864906e-05,
+ "loss": 0.2818,
+ "step": 12743
+ },
+ {
+ "epoch": 34.724795640326974,
+ "grad_norm": 4.7205681800842285,
+ "learning_rate": 1.5169703217657274e-05,
+ "loss": 0.256,
+ "step": 12744
+ },
+ {
+ "epoch": 34.7275204359673,
+ "grad_norm": 4.57312536239624,
+ "learning_rate": 1.5168947779188163e-05,
+ "loss": 0.3879,
+ "step": 12745
+ },
+ {
+ "epoch": 34.73024523160763,
+ "grad_norm": 4.59470796585083,
+ "learning_rate": 1.5168192300463447e-05,
+ "loss": 0.2141,
+ "step": 12746
+ },
+ {
+ "epoch": 34.73297002724796,
+ "grad_norm": 3.461538791656494,
+ "learning_rate": 1.516743678148902e-05,
+ "loss": 0.3014,
+ "step": 12747
+ },
+ {
+ "epoch": 34.735694822888284,
+ "grad_norm": 5.382379531860352,
+ "learning_rate": 1.516668122227076e-05,
+ "loss": 0.2148,
+ "step": 12748
+ },
+ {
+ "epoch": 34.73841961852861,
+ "grad_norm": 3.87669038772583,
+ "learning_rate": 1.5165925622814554e-05,
+ "loss": 0.1451,
+ "step": 12749
+ },
+ {
+ "epoch": 34.741144414168936,
+ "grad_norm": 4.863086700439453,
+ "learning_rate": 1.5165169983126287e-05,
+ "loss": 0.1622,
+ "step": 12750
+ },
+ {
+ "epoch": 34.74386920980926,
+ "grad_norm": 4.54217529296875,
+ "learning_rate": 1.516441430321184e-05,
+ "loss": 0.1697,
+ "step": 12751
+ },
+ {
+ "epoch": 34.746594005449595,
+ "grad_norm": 3.9009664058685303,
+ "learning_rate": 1.5163658583077103e-05,
+ "loss": 0.1382,
+ "step": 12752
+ },
+ {
+ "epoch": 34.74931880108992,
+ "grad_norm": 5.528608798980713,
+ "learning_rate": 1.5162902822727956e-05,
+ "loss": 0.1712,
+ "step": 12753
+ },
+ {
+ "epoch": 34.752043596730246,
+ "grad_norm": 4.858880519866943,
+ "learning_rate": 1.5162147022170288e-05,
+ "loss": 0.2528,
+ "step": 12754
+ },
+ {
+ "epoch": 34.75476839237057,
+ "grad_norm": 4.3020920753479,
+ "learning_rate": 1.5161391181409987e-05,
+ "loss": 0.1658,
+ "step": 12755
+ },
+ {
+ "epoch": 34.7574931880109,
+ "grad_norm": 4.532046318054199,
+ "learning_rate": 1.5160635300452936e-05,
+ "loss": 0.1592,
+ "step": 12756
+ },
+ {
+ "epoch": 34.76021798365122,
+ "grad_norm": 6.457173824310303,
+ "learning_rate": 1.5159879379305028e-05,
+ "loss": 0.259,
+ "step": 12757
+ },
+ {
+ "epoch": 34.762942779291556,
+ "grad_norm": 5.784908771514893,
+ "learning_rate": 1.515912341797214e-05,
+ "loss": 0.1582,
+ "step": 12758
+ },
+ {
+ "epoch": 34.76566757493188,
+ "grad_norm": 4.073841571807861,
+ "learning_rate": 1.515836741646017e-05,
+ "loss": 0.152,
+ "step": 12759
+ },
+ {
+ "epoch": 34.76839237057221,
+ "grad_norm": 4.394538879394531,
+ "learning_rate": 1.5157611374774997e-05,
+ "loss": 0.2899,
+ "step": 12760
+ },
+ {
+ "epoch": 34.771117166212534,
+ "grad_norm": 3.573582887649536,
+ "learning_rate": 1.5156855292922512e-05,
+ "loss": 0.1582,
+ "step": 12761
+ },
+ {
+ "epoch": 34.77384196185286,
+ "grad_norm": 4.1153764724731445,
+ "learning_rate": 1.5156099170908605e-05,
+ "loss": 0.2782,
+ "step": 12762
+ },
+ {
+ "epoch": 34.776566757493185,
+ "grad_norm": 5.080151557922363,
+ "learning_rate": 1.5155343008739162e-05,
+ "loss": 0.1569,
+ "step": 12763
+ },
+ {
+ "epoch": 34.77929155313352,
+ "grad_norm": 4.848169803619385,
+ "learning_rate": 1.5154586806420078e-05,
+ "loss": 0.2072,
+ "step": 12764
+ },
+ {
+ "epoch": 34.782016348773844,
+ "grad_norm": 4.593221187591553,
+ "learning_rate": 1.5153830563957232e-05,
+ "loss": 0.338,
+ "step": 12765
+ },
+ {
+ "epoch": 34.78474114441417,
+ "grad_norm": 4.587562084197998,
+ "learning_rate": 1.5153074281356523e-05,
+ "loss": 0.1799,
+ "step": 12766
+ },
+ {
+ "epoch": 34.787465940054496,
+ "grad_norm": 4.335643768310547,
+ "learning_rate": 1.5152317958623836e-05,
+ "loss": 0.1549,
+ "step": 12767
+ },
+ {
+ "epoch": 34.79019073569482,
+ "grad_norm": 4.198123455047607,
+ "learning_rate": 1.5151561595765066e-05,
+ "loss": 0.142,
+ "step": 12768
+ },
+ {
+ "epoch": 34.79291553133515,
+ "grad_norm": 4.142108917236328,
+ "learning_rate": 1.5150805192786097e-05,
+ "loss": 0.1707,
+ "step": 12769
+ },
+ {
+ "epoch": 34.79564032697548,
+ "grad_norm": 6.094672679901123,
+ "learning_rate": 1.5150048749692822e-05,
+ "loss": 0.2169,
+ "step": 12770
+ },
+ {
+ "epoch": 34.798365122615806,
+ "grad_norm": 3.790233612060547,
+ "learning_rate": 1.5149292266491133e-05,
+ "loss": 0.1215,
+ "step": 12771
+ },
+ {
+ "epoch": 34.80108991825613,
+ "grad_norm": 3.7599799633026123,
+ "learning_rate": 1.5148535743186928e-05,
+ "loss": 0.3144,
+ "step": 12772
+ },
+ {
+ "epoch": 34.80381471389646,
+ "grad_norm": 19.627620697021484,
+ "learning_rate": 1.5147779179786084e-05,
+ "loss": 0.2732,
+ "step": 12773
+ },
+ {
+ "epoch": 34.80653950953678,
+ "grad_norm": 4.950121879577637,
+ "learning_rate": 1.5147022576294506e-05,
+ "loss": 0.3234,
+ "step": 12774
+ },
+ {
+ "epoch": 34.80926430517711,
+ "grad_norm": 4.043761253356934,
+ "learning_rate": 1.514626593271808e-05,
+ "loss": 0.1843,
+ "step": 12775
+ },
+ {
+ "epoch": 34.81198910081744,
+ "grad_norm": 5.774435997009277,
+ "learning_rate": 1.5145509249062702e-05,
+ "loss": 0.1332,
+ "step": 12776
+ },
+ {
+ "epoch": 34.81471389645777,
+ "grad_norm": 4.448856353759766,
+ "learning_rate": 1.5144752525334264e-05,
+ "loss": 0.2535,
+ "step": 12777
+ },
+ {
+ "epoch": 34.817438692098094,
+ "grad_norm": 5.205796241760254,
+ "learning_rate": 1.5143995761538657e-05,
+ "loss": 0.3186,
+ "step": 12778
+ },
+ {
+ "epoch": 34.82016348773842,
+ "grad_norm": 4.516385555267334,
+ "learning_rate": 1.5143238957681776e-05,
+ "loss": 0.1997,
+ "step": 12779
+ },
+ {
+ "epoch": 34.822888283378745,
+ "grad_norm": 4.446849822998047,
+ "learning_rate": 1.5142482113769519e-05,
+ "loss": 0.3358,
+ "step": 12780
+ },
+ {
+ "epoch": 34.82561307901907,
+ "grad_norm": 3.882852077484131,
+ "learning_rate": 1.5141725229807773e-05,
+ "loss": 0.2211,
+ "step": 12781
+ },
+ {
+ "epoch": 34.828337874659404,
+ "grad_norm": 3.629563808441162,
+ "learning_rate": 1.5140968305802438e-05,
+ "loss": 0.1335,
+ "step": 12782
+ },
+ {
+ "epoch": 34.83106267029973,
+ "grad_norm": 4.018277168273926,
+ "learning_rate": 1.5140211341759408e-05,
+ "loss": 0.1401,
+ "step": 12783
+ },
+ {
+ "epoch": 34.833787465940055,
+ "grad_norm": 4.1711745262146,
+ "learning_rate": 1.5139454337684577e-05,
+ "loss": 0.1823,
+ "step": 12784
+ },
+ {
+ "epoch": 34.83651226158038,
+ "grad_norm": 4.374452590942383,
+ "learning_rate": 1.5138697293583842e-05,
+ "loss": 0.1832,
+ "step": 12785
+ },
+ {
+ "epoch": 34.83923705722071,
+ "grad_norm": 5.284313678741455,
+ "learning_rate": 1.5137940209463095e-05,
+ "loss": 0.2207,
+ "step": 12786
+ },
+ {
+ "epoch": 34.84196185286103,
+ "grad_norm": 4.663943290710449,
+ "learning_rate": 1.5137183085328237e-05,
+ "loss": 0.1855,
+ "step": 12787
+ },
+ {
+ "epoch": 34.844686648501366,
+ "grad_norm": 5.102797985076904,
+ "learning_rate": 1.5136425921185162e-05,
+ "loss": 0.332,
+ "step": 12788
+ },
+ {
+ "epoch": 34.84741144414169,
+ "grad_norm": 14.628813743591309,
+ "learning_rate": 1.5135668717039767e-05,
+ "loss": 0.2207,
+ "step": 12789
+ },
+ {
+ "epoch": 34.85013623978202,
+ "grad_norm": 4.36284875869751,
+ "learning_rate": 1.513491147289795e-05,
+ "loss": 0.1591,
+ "step": 12790
+ },
+ {
+ "epoch": 34.85286103542234,
+ "grad_norm": 5.1945366859436035,
+ "learning_rate": 1.5134154188765608e-05,
+ "loss": 0.1539,
+ "step": 12791
+ },
+ {
+ "epoch": 34.85558583106267,
+ "grad_norm": 4.47944974899292,
+ "learning_rate": 1.513339686464864e-05,
+ "loss": 0.1563,
+ "step": 12792
+ },
+ {
+ "epoch": 34.858310626702995,
+ "grad_norm": 4.376889705657959,
+ "learning_rate": 1.513263950055294e-05,
+ "loss": 0.2331,
+ "step": 12793
+ },
+ {
+ "epoch": 34.86103542234333,
+ "grad_norm": 3.999389171600342,
+ "learning_rate": 1.5131882096484409e-05,
+ "loss": 0.1485,
+ "step": 12794
+ },
+ {
+ "epoch": 34.86376021798365,
+ "grad_norm": 4.724643230438232,
+ "learning_rate": 1.5131124652448943e-05,
+ "loss": 0.1865,
+ "step": 12795
+ },
+ {
+ "epoch": 34.86648501362398,
+ "grad_norm": 5.157161712646484,
+ "learning_rate": 1.5130367168452449e-05,
+ "loss": 0.2494,
+ "step": 12796
+ },
+ {
+ "epoch": 34.869209809264305,
+ "grad_norm": 3.9059741497039795,
+ "learning_rate": 1.5129609644500813e-05,
+ "loss": 0.2586,
+ "step": 12797
+ },
+ {
+ "epoch": 34.87193460490463,
+ "grad_norm": 4.243319988250732,
+ "learning_rate": 1.5128852080599949e-05,
+ "loss": 0.1804,
+ "step": 12798
+ },
+ {
+ "epoch": 34.87465940054496,
+ "grad_norm": 4.964089393615723,
+ "learning_rate": 1.5128094476755747e-05,
+ "loss": 0.2589,
+ "step": 12799
+ },
+ {
+ "epoch": 34.87738419618529,
+ "grad_norm": 5.099611759185791,
+ "learning_rate": 1.5127336832974111e-05,
+ "loss": 0.3148,
+ "step": 12800
+ },
+ {
+ "epoch": 34.880108991825615,
+ "grad_norm": 4.485393524169922,
+ "learning_rate": 1.512657914926094e-05,
+ "loss": 0.154,
+ "step": 12801
+ },
+ {
+ "epoch": 34.88283378746594,
+ "grad_norm": 3.9726200103759766,
+ "learning_rate": 1.5125821425622137e-05,
+ "loss": 0.2649,
+ "step": 12802
+ },
+ {
+ "epoch": 34.88555858310627,
+ "grad_norm": 4.088240146636963,
+ "learning_rate": 1.5125063662063602e-05,
+ "loss": 0.4252,
+ "step": 12803
+ },
+ {
+ "epoch": 34.88828337874659,
+ "grad_norm": 5.377208709716797,
+ "learning_rate": 1.5124305858591238e-05,
+ "loss": 0.2938,
+ "step": 12804
+ },
+ {
+ "epoch": 34.89100817438692,
+ "grad_norm": 5.132850170135498,
+ "learning_rate": 1.5123548015210941e-05,
+ "loss": 0.1404,
+ "step": 12805
+ },
+ {
+ "epoch": 34.89373297002725,
+ "grad_norm": 5.774279594421387,
+ "learning_rate": 1.5122790131928615e-05,
+ "loss": 0.1887,
+ "step": 12806
+ },
+ {
+ "epoch": 34.89645776566758,
+ "grad_norm": 3.956270694732666,
+ "learning_rate": 1.5122032208750168e-05,
+ "loss": 0.1735,
+ "step": 12807
+ },
+ {
+ "epoch": 34.8991825613079,
+ "grad_norm": 5.025761604309082,
+ "learning_rate": 1.51212742456815e-05,
+ "loss": 0.1685,
+ "step": 12808
+ },
+ {
+ "epoch": 34.90190735694823,
+ "grad_norm": 4.303150177001953,
+ "learning_rate": 1.5120516242728508e-05,
+ "loss": 0.2581,
+ "step": 12809
+ },
+ {
+ "epoch": 34.904632152588555,
+ "grad_norm": 8.459695816040039,
+ "learning_rate": 1.5119758199897106e-05,
+ "loss": 0.2835,
+ "step": 12810
+ },
+ {
+ "epoch": 34.90735694822888,
+ "grad_norm": 6.0611491203308105,
+ "learning_rate": 1.511900011719319e-05,
+ "loss": 0.2311,
+ "step": 12811
+ },
+ {
+ "epoch": 34.91008174386921,
+ "grad_norm": 3.946255683898926,
+ "learning_rate": 1.5118241994622666e-05,
+ "loss": 0.1763,
+ "step": 12812
+ },
+ {
+ "epoch": 34.91280653950954,
+ "grad_norm": 5.115240573883057,
+ "learning_rate": 1.5117483832191438e-05,
+ "loss": 0.164,
+ "step": 12813
+ },
+ {
+ "epoch": 34.915531335149865,
+ "grad_norm": 4.232547760009766,
+ "learning_rate": 1.5116725629905408e-05,
+ "loss": 0.3159,
+ "step": 12814
+ },
+ {
+ "epoch": 34.91825613079019,
+ "grad_norm": 3.759364366531372,
+ "learning_rate": 1.5115967387770484e-05,
+ "loss": 0.1098,
+ "step": 12815
+ },
+ {
+ "epoch": 34.920980926430516,
+ "grad_norm": 4.227931976318359,
+ "learning_rate": 1.5115209105792574e-05,
+ "loss": 0.1171,
+ "step": 12816
+ },
+ {
+ "epoch": 34.92370572207084,
+ "grad_norm": 4.266099452972412,
+ "learning_rate": 1.5114450783977579e-05,
+ "loss": 0.3556,
+ "step": 12817
+ },
+ {
+ "epoch": 34.926430517711175,
+ "grad_norm": 4.919320583343506,
+ "learning_rate": 1.5113692422331406e-05,
+ "loss": 0.3362,
+ "step": 12818
+ },
+ {
+ "epoch": 34.9291553133515,
+ "grad_norm": 3.7781293392181396,
+ "learning_rate": 1.511293402085996e-05,
+ "loss": 0.3219,
+ "step": 12819
+ },
+ {
+ "epoch": 34.93188010899183,
+ "grad_norm": 4.517542839050293,
+ "learning_rate": 1.5112175579569149e-05,
+ "loss": 0.1763,
+ "step": 12820
+ },
+ {
+ "epoch": 34.93460490463215,
+ "grad_norm": 5.254129409790039,
+ "learning_rate": 1.5111417098464879e-05,
+ "loss": 0.1642,
+ "step": 12821
+ },
+ {
+ "epoch": 34.93732970027248,
+ "grad_norm": 4.194706916809082,
+ "learning_rate": 1.5110658577553058e-05,
+ "loss": 0.3424,
+ "step": 12822
+ },
+ {
+ "epoch": 34.940054495912804,
+ "grad_norm": 4.224678993225098,
+ "learning_rate": 1.510990001683959e-05,
+ "loss": 0.1609,
+ "step": 12823
+ },
+ {
+ "epoch": 34.94277929155314,
+ "grad_norm": 4.6882171630859375,
+ "learning_rate": 1.5109141416330386e-05,
+ "loss": 0.2003,
+ "step": 12824
+ },
+ {
+ "epoch": 34.94550408719346,
+ "grad_norm": 4.523833751678467,
+ "learning_rate": 1.5108382776031354e-05,
+ "loss": 0.267,
+ "step": 12825
+ },
+ {
+ "epoch": 34.94822888283379,
+ "grad_norm": 3.887094020843506,
+ "learning_rate": 1.51076240959484e-05,
+ "loss": 0.1581,
+ "step": 12826
+ },
+ {
+ "epoch": 34.950953678474114,
+ "grad_norm": 3.709894895553589,
+ "learning_rate": 1.5106865376087437e-05,
+ "loss": 0.2332,
+ "step": 12827
+ },
+ {
+ "epoch": 34.95367847411444,
+ "grad_norm": 3.630887985229492,
+ "learning_rate": 1.5106106616454371e-05,
+ "loss": 0.1614,
+ "step": 12828
+ },
+ {
+ "epoch": 34.956403269754766,
+ "grad_norm": 18.661052703857422,
+ "learning_rate": 1.5105347817055106e-05,
+ "loss": 0.197,
+ "step": 12829
+ },
+ {
+ "epoch": 34.95912806539509,
+ "grad_norm": 4.740427494049072,
+ "learning_rate": 1.510458897789556e-05,
+ "loss": 0.284,
+ "step": 12830
+ },
+ {
+ "epoch": 34.961852861035425,
+ "grad_norm": 4.967202186584473,
+ "learning_rate": 1.5103830098981637e-05,
+ "loss": 0.229,
+ "step": 12831
+ },
+ {
+ "epoch": 34.96457765667575,
+ "grad_norm": 4.532550811767578,
+ "learning_rate": 1.5103071180319254e-05,
+ "loss": 0.1896,
+ "step": 12832
+ },
+ {
+ "epoch": 34.967302452316076,
+ "grad_norm": 4.629622459411621,
+ "learning_rate": 1.510231222191431e-05,
+ "loss": 0.1898,
+ "step": 12833
+ },
+ {
+ "epoch": 34.9700272479564,
+ "grad_norm": 4.466314315795898,
+ "learning_rate": 1.5101553223772728e-05,
+ "loss": 0.1367,
+ "step": 12834
+ },
+ {
+ "epoch": 34.97275204359673,
+ "grad_norm": 4.249719142913818,
+ "learning_rate": 1.5100794185900411e-05,
+ "loss": 0.1834,
+ "step": 12835
+ },
+ {
+ "epoch": 34.97547683923706,
+ "grad_norm": 4.874640464782715,
+ "learning_rate": 1.5100035108303278e-05,
+ "loss": 0.3472,
+ "step": 12836
+ },
+ {
+ "epoch": 34.97820163487739,
+ "grad_norm": 5.176187992095947,
+ "learning_rate": 1.509927599098723e-05,
+ "loss": 0.1275,
+ "step": 12837
+ },
+ {
+ "epoch": 34.98092643051771,
+ "grad_norm": 4.416406154632568,
+ "learning_rate": 1.5098516833958187e-05,
+ "loss": 0.1466,
+ "step": 12838
+ },
+ {
+ "epoch": 34.98365122615804,
+ "grad_norm": 5.283556938171387,
+ "learning_rate": 1.509775763722206e-05,
+ "loss": 0.2533,
+ "step": 12839
+ },
+ {
+ "epoch": 34.986376021798364,
+ "grad_norm": 3.8340160846710205,
+ "learning_rate": 1.5096998400784756e-05,
+ "loss": 0.3322,
+ "step": 12840
+ },
+ {
+ "epoch": 34.98910081743869,
+ "grad_norm": 4.773148536682129,
+ "learning_rate": 1.5096239124652197e-05,
+ "loss": 0.1449,
+ "step": 12841
+ },
+ {
+ "epoch": 34.991825613079016,
+ "grad_norm": 5.962275981903076,
+ "learning_rate": 1.5095479808830292e-05,
+ "loss": 0.5437,
+ "step": 12842
+ },
+ {
+ "epoch": 34.99455040871935,
+ "grad_norm": 5.279979705810547,
+ "learning_rate": 1.5094720453324952e-05,
+ "loss": 0.2117,
+ "step": 12843
+ },
+ {
+ "epoch": 34.997275204359674,
+ "grad_norm": 4.215738773345947,
+ "learning_rate": 1.5093961058142095e-05,
+ "loss": 0.287,
+ "step": 12844
+ },
+ {
+ "epoch": 35.0,
+ "grad_norm": 3.646427869796753,
+ "learning_rate": 1.5093201623287631e-05,
+ "loss": 0.2197,
+ "step": 12845
+ },
+ {
+ "epoch": 35.002724795640326,
+ "grad_norm": 4.128983020782471,
+ "learning_rate": 1.509244214876748e-05,
+ "loss": 0.2968,
+ "step": 12846
+ },
+ {
+ "epoch": 35.00544959128065,
+ "grad_norm": 3.246384859085083,
+ "learning_rate": 1.5091682634587547e-05,
+ "loss": 0.25,
+ "step": 12847
+ },
+ {
+ "epoch": 35.00817438692098,
+ "grad_norm": 4.90088415145874,
+ "learning_rate": 1.509092308075376e-05,
+ "loss": 0.2247,
+ "step": 12848
+ },
+ {
+ "epoch": 35.01089918256131,
+ "grad_norm": 4.031245231628418,
+ "learning_rate": 1.5090163487272027e-05,
+ "loss": 0.2035,
+ "step": 12849
+ },
+ {
+ "epoch": 35.013623978201636,
+ "grad_norm": 4.437304496765137,
+ "learning_rate": 1.5089403854148261e-05,
+ "loss": 0.221,
+ "step": 12850
+ },
+ {
+ "epoch": 35.01634877384196,
+ "grad_norm": 4.321059226989746,
+ "learning_rate": 1.5088644181388386e-05,
+ "loss": 0.1126,
+ "step": 12851
+ },
+ {
+ "epoch": 35.01907356948229,
+ "grad_norm": 4.411810398101807,
+ "learning_rate": 1.508788446899831e-05,
+ "loss": 0.2592,
+ "step": 12852
+ },
+ {
+ "epoch": 35.02179836512261,
+ "grad_norm": 3.637772560119629,
+ "learning_rate": 1.5087124716983956e-05,
+ "loss": 0.1244,
+ "step": 12853
+ },
+ {
+ "epoch": 35.02452316076294,
+ "grad_norm": 10.6876802444458,
+ "learning_rate": 1.5086364925351238e-05,
+ "loss": 0.1373,
+ "step": 12854
+ },
+ {
+ "epoch": 35.02724795640327,
+ "grad_norm": 12.887807846069336,
+ "learning_rate": 1.5085605094106072e-05,
+ "loss": 0.2658,
+ "step": 12855
+ },
+ {
+ "epoch": 35.0299727520436,
+ "grad_norm": 4.397464275360107,
+ "learning_rate": 1.5084845223254381e-05,
+ "loss": 0.1438,
+ "step": 12856
+ },
+ {
+ "epoch": 35.032697547683924,
+ "grad_norm": 3.920375347137451,
+ "learning_rate": 1.5084085312802075e-05,
+ "loss": 0.1133,
+ "step": 12857
+ },
+ {
+ "epoch": 35.03542234332425,
+ "grad_norm": 5.892287254333496,
+ "learning_rate": 1.5083325362755077e-05,
+ "loss": 0.1764,
+ "step": 12858
+ },
+ {
+ "epoch": 35.038147138964575,
+ "grad_norm": 4.2641777992248535,
+ "learning_rate": 1.5082565373119307e-05,
+ "loss": 0.1516,
+ "step": 12859
+ },
+ {
+ "epoch": 35.0408719346049,
+ "grad_norm": 4.047680377960205,
+ "learning_rate": 1.508180534390068e-05,
+ "loss": 0.2081,
+ "step": 12860
+ },
+ {
+ "epoch": 35.043596730245234,
+ "grad_norm": 4.77567195892334,
+ "learning_rate": 1.5081045275105117e-05,
+ "loss": 0.1916,
+ "step": 12861
+ },
+ {
+ "epoch": 35.04632152588556,
+ "grad_norm": 4.654408931732178,
+ "learning_rate": 1.5080285166738535e-05,
+ "loss": 0.276,
+ "step": 12862
+ },
+ {
+ "epoch": 35.049046321525886,
+ "grad_norm": 3.4873979091644287,
+ "learning_rate": 1.5079525018806856e-05,
+ "loss": 0.1258,
+ "step": 12863
+ },
+ {
+ "epoch": 35.05177111716621,
+ "grad_norm": 4.3788886070251465,
+ "learning_rate": 1.5078764831316002e-05,
+ "loss": 0.1456,
+ "step": 12864
+ },
+ {
+ "epoch": 35.05449591280654,
+ "grad_norm": 4.871580600738525,
+ "learning_rate": 1.507800460427189e-05,
+ "loss": 0.2455,
+ "step": 12865
+ },
+ {
+ "epoch": 35.05722070844686,
+ "grad_norm": 4.262026786804199,
+ "learning_rate": 1.5077244337680439e-05,
+ "loss": 0.1581,
+ "step": 12866
+ },
+ {
+ "epoch": 35.059945504087196,
+ "grad_norm": 4.3254313468933105,
+ "learning_rate": 1.5076484031547578e-05,
+ "loss": 0.4415,
+ "step": 12867
+ },
+ {
+ "epoch": 35.06267029972752,
+ "grad_norm": 4.155272960662842,
+ "learning_rate": 1.5075723685879218e-05,
+ "loss": 0.2035,
+ "step": 12868
+ },
+ {
+ "epoch": 35.06539509536785,
+ "grad_norm": 4.471890926361084,
+ "learning_rate": 1.5074963300681286e-05,
+ "loss": 0.1841,
+ "step": 12869
+ },
+ {
+ "epoch": 35.06811989100817,
+ "grad_norm": 3.5553901195526123,
+ "learning_rate": 1.5074202875959704e-05,
+ "loss": 0.1329,
+ "step": 12870
+ },
+ {
+ "epoch": 35.0708446866485,
+ "grad_norm": 3.843127965927124,
+ "learning_rate": 1.5073442411720397e-05,
+ "loss": 0.3156,
+ "step": 12871
+ },
+ {
+ "epoch": 35.073569482288825,
+ "grad_norm": 11.021859169006348,
+ "learning_rate": 1.5072681907969277e-05,
+ "loss": 0.2235,
+ "step": 12872
+ },
+ {
+ "epoch": 35.07629427792916,
+ "grad_norm": 4.334469318389893,
+ "learning_rate": 1.5071921364712278e-05,
+ "loss": 0.1765,
+ "step": 12873
+ },
+ {
+ "epoch": 35.079019073569484,
+ "grad_norm": 3.9780359268188477,
+ "learning_rate": 1.5071160781955316e-05,
+ "loss": 0.2417,
+ "step": 12874
+ },
+ {
+ "epoch": 35.08174386920981,
+ "grad_norm": 4.881362438201904,
+ "learning_rate": 1.5070400159704317e-05,
+ "loss": 0.2087,
+ "step": 12875
+ },
+ {
+ "epoch": 35.084468664850135,
+ "grad_norm": 5.543717384338379,
+ "learning_rate": 1.5069639497965205e-05,
+ "loss": 0.1568,
+ "step": 12876
+ },
+ {
+ "epoch": 35.08719346049046,
+ "grad_norm": 4.381588935852051,
+ "learning_rate": 1.5068878796743907e-05,
+ "loss": 0.1482,
+ "step": 12877
+ },
+ {
+ "epoch": 35.08991825613079,
+ "grad_norm": 19.586332321166992,
+ "learning_rate": 1.5068118056046342e-05,
+ "loss": 0.1761,
+ "step": 12878
+ },
+ {
+ "epoch": 35.09264305177112,
+ "grad_norm": 4.60221529006958,
+ "learning_rate": 1.5067357275878438e-05,
+ "loss": 0.1686,
+ "step": 12879
+ },
+ {
+ "epoch": 35.095367847411445,
+ "grad_norm": 6.6504621505737305,
+ "learning_rate": 1.5066596456246116e-05,
+ "loss": 0.1953,
+ "step": 12880
+ },
+ {
+ "epoch": 35.09809264305177,
+ "grad_norm": 4.999527931213379,
+ "learning_rate": 1.5065835597155305e-05,
+ "loss": 0.151,
+ "step": 12881
+ },
+ {
+ "epoch": 35.1008174386921,
+ "grad_norm": 5.010338306427002,
+ "learning_rate": 1.506507469861193e-05,
+ "loss": 0.1476,
+ "step": 12882
+ },
+ {
+ "epoch": 35.10354223433242,
+ "grad_norm": 4.020954132080078,
+ "learning_rate": 1.5064313760621913e-05,
+ "loss": 0.1267,
+ "step": 12883
+ },
+ {
+ "epoch": 35.10626702997275,
+ "grad_norm": 4.851685523986816,
+ "learning_rate": 1.5063552783191186e-05,
+ "loss": 0.2424,
+ "step": 12884
+ },
+ {
+ "epoch": 35.10899182561308,
+ "grad_norm": 6.055109977722168,
+ "learning_rate": 1.5062791766325676e-05,
+ "loss": 0.1541,
+ "step": 12885
+ },
+ {
+ "epoch": 35.11171662125341,
+ "grad_norm": 4.499035835266113,
+ "learning_rate": 1.5062030710031304e-05,
+ "loss": 0.139,
+ "step": 12886
+ },
+ {
+ "epoch": 35.11444141689373,
+ "grad_norm": 5.165180206298828,
+ "learning_rate": 1.5061269614314e-05,
+ "loss": 0.2301,
+ "step": 12887
+ },
+ {
+ "epoch": 35.11716621253406,
+ "grad_norm": 5.662968635559082,
+ "learning_rate": 1.506050847917969e-05,
+ "loss": 0.1796,
+ "step": 12888
+ },
+ {
+ "epoch": 35.119891008174385,
+ "grad_norm": 4.660906791687012,
+ "learning_rate": 1.5059747304634304e-05,
+ "loss": 0.2978,
+ "step": 12889
+ },
+ {
+ "epoch": 35.12261580381471,
+ "grad_norm": 6.050297260284424,
+ "learning_rate": 1.5058986090683767e-05,
+ "loss": 0.2516,
+ "step": 12890
+ },
+ {
+ "epoch": 35.12534059945504,
+ "grad_norm": 4.608145713806152,
+ "learning_rate": 1.5058224837334012e-05,
+ "loss": 0.2508,
+ "step": 12891
+ },
+ {
+ "epoch": 35.12806539509537,
+ "grad_norm": 5.7353363037109375,
+ "learning_rate": 1.505746354459096e-05,
+ "loss": 0.19,
+ "step": 12892
+ },
+ {
+ "epoch": 35.130790190735695,
+ "grad_norm": 3.989981174468994,
+ "learning_rate": 1.5056702212460547e-05,
+ "loss": 0.0889,
+ "step": 12893
+ },
+ {
+ "epoch": 35.13351498637602,
+ "grad_norm": 4.544556140899658,
+ "learning_rate": 1.5055940840948702e-05,
+ "loss": 0.1374,
+ "step": 12894
+ },
+ {
+ "epoch": 35.13623978201635,
+ "grad_norm": 5.330860137939453,
+ "learning_rate": 1.5055179430061354e-05,
+ "loss": 0.1622,
+ "step": 12895
+ },
+ {
+ "epoch": 35.13896457765667,
+ "grad_norm": 4.7821526527404785,
+ "learning_rate": 1.5054417979804426e-05,
+ "loss": 0.1457,
+ "step": 12896
+ },
+ {
+ "epoch": 35.141689373297005,
+ "grad_norm": 3.4477078914642334,
+ "learning_rate": 1.5053656490183856e-05,
+ "loss": 0.1409,
+ "step": 12897
+ },
+ {
+ "epoch": 35.14441416893733,
+ "grad_norm": 4.371578693389893,
+ "learning_rate": 1.5052894961205573e-05,
+ "loss": 0.1993,
+ "step": 12898
+ },
+ {
+ "epoch": 35.14713896457766,
+ "grad_norm": 4.482218265533447,
+ "learning_rate": 1.5052133392875505e-05,
+ "loss": 0.0875,
+ "step": 12899
+ },
+ {
+ "epoch": 35.14986376021798,
+ "grad_norm": 5.659038543701172,
+ "learning_rate": 1.5051371785199585e-05,
+ "loss": 0.2369,
+ "step": 12900
+ },
+ {
+ "epoch": 35.15258855585831,
+ "grad_norm": 5.808692932128906,
+ "learning_rate": 1.5050610138183746e-05,
+ "loss": 0.217,
+ "step": 12901
+ },
+ {
+ "epoch": 35.155313351498634,
+ "grad_norm": 3.681455612182617,
+ "learning_rate": 1.5049848451833912e-05,
+ "loss": 0.118,
+ "step": 12902
+ },
+ {
+ "epoch": 35.15803814713897,
+ "grad_norm": 4.412920951843262,
+ "learning_rate": 1.5049086726156028e-05,
+ "loss": 0.1597,
+ "step": 12903
+ },
+ {
+ "epoch": 35.16076294277929,
+ "grad_norm": 4.353061199188232,
+ "learning_rate": 1.5048324961156016e-05,
+ "loss": 0.1975,
+ "step": 12904
+ },
+ {
+ "epoch": 35.16348773841962,
+ "grad_norm": 3.8298871517181396,
+ "learning_rate": 1.5047563156839812e-05,
+ "loss": 0.1079,
+ "step": 12905
+ },
+ {
+ "epoch": 35.166212534059945,
+ "grad_norm": 4.335110187530518,
+ "learning_rate": 1.5046801313213349e-05,
+ "loss": 0.1526,
+ "step": 12906
+ },
+ {
+ "epoch": 35.16893732970027,
+ "grad_norm": 4.970110893249512,
+ "learning_rate": 1.5046039430282559e-05,
+ "loss": 0.3098,
+ "step": 12907
+ },
+ {
+ "epoch": 35.171662125340596,
+ "grad_norm": 4.288705825805664,
+ "learning_rate": 1.5045277508053376e-05,
+ "loss": 0.1339,
+ "step": 12908
+ },
+ {
+ "epoch": 35.17438692098093,
+ "grad_norm": 5.435235500335693,
+ "learning_rate": 1.5044515546531733e-05,
+ "loss": 0.2267,
+ "step": 12909
+ },
+ {
+ "epoch": 35.177111716621255,
+ "grad_norm": 5.425375938415527,
+ "learning_rate": 1.5043753545723568e-05,
+ "loss": 0.1719,
+ "step": 12910
+ },
+ {
+ "epoch": 35.17983651226158,
+ "grad_norm": 4.124241352081299,
+ "learning_rate": 1.5042991505634812e-05,
+ "loss": 0.2341,
+ "step": 12911
+ },
+ {
+ "epoch": 35.182561307901906,
+ "grad_norm": 3.540236711502075,
+ "learning_rate": 1.50422294262714e-05,
+ "loss": 0.1201,
+ "step": 12912
+ },
+ {
+ "epoch": 35.18528610354223,
+ "grad_norm": 3.895890951156616,
+ "learning_rate": 1.5041467307639268e-05,
+ "loss": 0.1448,
+ "step": 12913
+ },
+ {
+ "epoch": 35.18801089918256,
+ "grad_norm": 4.824029922485352,
+ "learning_rate": 1.504070514974435e-05,
+ "loss": 0.1508,
+ "step": 12914
+ },
+ {
+ "epoch": 35.19073569482289,
+ "grad_norm": 3.9976296424865723,
+ "learning_rate": 1.5039942952592585e-05,
+ "loss": 0.1294,
+ "step": 12915
+ },
+ {
+ "epoch": 35.19346049046322,
+ "grad_norm": 5.651499271392822,
+ "learning_rate": 1.5039180716189903e-05,
+ "loss": 0.1715,
+ "step": 12916
+ },
+ {
+ "epoch": 35.19618528610354,
+ "grad_norm": 3.3997914791107178,
+ "learning_rate": 1.5038418440542247e-05,
+ "loss": 0.1393,
+ "step": 12917
+ },
+ {
+ "epoch": 35.19891008174387,
+ "grad_norm": 4.434873104095459,
+ "learning_rate": 1.5037656125655543e-05,
+ "loss": 0.2672,
+ "step": 12918
+ },
+ {
+ "epoch": 35.201634877384194,
+ "grad_norm": 3.63848876953125,
+ "learning_rate": 1.5036893771535743e-05,
+ "loss": 0.3012,
+ "step": 12919
+ },
+ {
+ "epoch": 35.20435967302452,
+ "grad_norm": 3.9556684494018555,
+ "learning_rate": 1.5036131378188774e-05,
+ "loss": 0.2765,
+ "step": 12920
+ },
+ {
+ "epoch": 35.20708446866485,
+ "grad_norm": 3.9323573112487793,
+ "learning_rate": 1.5035368945620578e-05,
+ "loss": 0.2178,
+ "step": 12921
+ },
+ {
+ "epoch": 35.20980926430518,
+ "grad_norm": 3.9572694301605225,
+ "learning_rate": 1.503460647383709e-05,
+ "loss": 0.221,
+ "step": 12922
+ },
+ {
+ "epoch": 35.212534059945504,
+ "grad_norm": 5.448349475860596,
+ "learning_rate": 1.5033843962844246e-05,
+ "loss": 0.2061,
+ "step": 12923
+ },
+ {
+ "epoch": 35.21525885558583,
+ "grad_norm": 5.299439430236816,
+ "learning_rate": 1.503308141264799e-05,
+ "loss": 0.4421,
+ "step": 12924
+ },
+ {
+ "epoch": 35.217983651226156,
+ "grad_norm": 4.221270561218262,
+ "learning_rate": 1.5032318823254257e-05,
+ "loss": 0.1678,
+ "step": 12925
+ },
+ {
+ "epoch": 35.22070844686648,
+ "grad_norm": 4.930756092071533,
+ "learning_rate": 1.5031556194668988e-05,
+ "loss": 0.1634,
+ "step": 12926
+ },
+ {
+ "epoch": 35.223433242506815,
+ "grad_norm": 4.655137538909912,
+ "learning_rate": 1.5030793526898118e-05,
+ "loss": 0.1479,
+ "step": 12927
+ },
+ {
+ "epoch": 35.22615803814714,
+ "grad_norm": 4.059932231903076,
+ "learning_rate": 1.5030030819947592e-05,
+ "loss": 0.1759,
+ "step": 12928
+ },
+ {
+ "epoch": 35.228882833787466,
+ "grad_norm": 4.624604225158691,
+ "learning_rate": 1.5029268073823349e-05,
+ "loss": 0.3485,
+ "step": 12929
+ },
+ {
+ "epoch": 35.23160762942779,
+ "grad_norm": 4.681209087371826,
+ "learning_rate": 1.5028505288531325e-05,
+ "loss": 0.1633,
+ "step": 12930
+ },
+ {
+ "epoch": 35.23433242506812,
+ "grad_norm": 4.723489761352539,
+ "learning_rate": 1.5027742464077468e-05,
+ "loss": 0.1373,
+ "step": 12931
+ },
+ {
+ "epoch": 35.237057220708444,
+ "grad_norm": 5.052133083343506,
+ "learning_rate": 1.502697960046771e-05,
+ "loss": 0.3662,
+ "step": 12932
+ },
+ {
+ "epoch": 35.23978201634878,
+ "grad_norm": 4.3001790046691895,
+ "learning_rate": 1.5026216697708002e-05,
+ "loss": 0.2453,
+ "step": 12933
+ },
+ {
+ "epoch": 35.2425068119891,
+ "grad_norm": 3.9355995655059814,
+ "learning_rate": 1.5025453755804275e-05,
+ "loss": 0.1904,
+ "step": 12934
+ },
+ {
+ "epoch": 35.24523160762943,
+ "grad_norm": 4.463444709777832,
+ "learning_rate": 1.5024690774762478e-05,
+ "loss": 0.1297,
+ "step": 12935
+ },
+ {
+ "epoch": 35.247956403269754,
+ "grad_norm": 4.6066083908081055,
+ "learning_rate": 1.5023927754588551e-05,
+ "loss": 0.1106,
+ "step": 12936
+ },
+ {
+ "epoch": 35.25068119891008,
+ "grad_norm": 3.7600479125976562,
+ "learning_rate": 1.5023164695288435e-05,
+ "loss": 0.2813,
+ "step": 12937
+ },
+ {
+ "epoch": 35.253405994550405,
+ "grad_norm": 7.502614974975586,
+ "learning_rate": 1.502240159686808e-05,
+ "loss": 0.1782,
+ "step": 12938
+ },
+ {
+ "epoch": 35.25613079019074,
+ "grad_norm": 4.52699613571167,
+ "learning_rate": 1.5021638459333417e-05,
+ "loss": 0.254,
+ "step": 12939
+ },
+ {
+ "epoch": 35.258855585831064,
+ "grad_norm": 6.246139049530029,
+ "learning_rate": 1.5020875282690392e-05,
+ "loss": 0.1485,
+ "step": 12940
+ },
+ {
+ "epoch": 35.26158038147139,
+ "grad_norm": 3.873674154281616,
+ "learning_rate": 1.502011206694496e-05,
+ "loss": 0.1636,
+ "step": 12941
+ },
+ {
+ "epoch": 35.264305177111716,
+ "grad_norm": 4.497064113616943,
+ "learning_rate": 1.5019348812103052e-05,
+ "loss": 0.2366,
+ "step": 12942
+ },
+ {
+ "epoch": 35.26702997275204,
+ "grad_norm": 4.225985527038574,
+ "learning_rate": 1.501858551817062e-05,
+ "loss": 0.148,
+ "step": 12943
+ },
+ {
+ "epoch": 35.26975476839237,
+ "grad_norm": 3.748455047607422,
+ "learning_rate": 1.5017822185153602e-05,
+ "loss": 0.1779,
+ "step": 12944
+ },
+ {
+ "epoch": 35.2724795640327,
+ "grad_norm": 3.467377185821533,
+ "learning_rate": 1.5017058813057948e-05,
+ "loss": 0.2604,
+ "step": 12945
+ },
+ {
+ "epoch": 35.275204359673026,
+ "grad_norm": 3.814634084701538,
+ "learning_rate": 1.5016295401889602e-05,
+ "loss": 0.4154,
+ "step": 12946
+ },
+ {
+ "epoch": 35.27792915531335,
+ "grad_norm": 3.471400499343872,
+ "learning_rate": 1.501553195165451e-05,
+ "loss": 0.1225,
+ "step": 12947
+ },
+ {
+ "epoch": 35.28065395095368,
+ "grad_norm": 4.659462928771973,
+ "learning_rate": 1.5014768462358613e-05,
+ "loss": 0.1625,
+ "step": 12948
+ },
+ {
+ "epoch": 35.283378746594,
+ "grad_norm": 3.405113697052002,
+ "learning_rate": 1.5014004934007864e-05,
+ "loss": 0.2417,
+ "step": 12949
+ },
+ {
+ "epoch": 35.28610354223433,
+ "grad_norm": 5.922151565551758,
+ "learning_rate": 1.5013241366608202e-05,
+ "loss": 0.202,
+ "step": 12950
+ },
+ {
+ "epoch": 35.28882833787466,
+ "grad_norm": 3.901660442352295,
+ "learning_rate": 1.5012477760165581e-05,
+ "loss": 0.147,
+ "step": 12951
+ },
+ {
+ "epoch": 35.29155313351499,
+ "grad_norm": 7.899892330169678,
+ "learning_rate": 1.5011714114685942e-05,
+ "loss": 0.3378,
+ "step": 12952
+ },
+ {
+ "epoch": 35.294277929155314,
+ "grad_norm": 4.318886756896973,
+ "learning_rate": 1.5010950430175237e-05,
+ "loss": 0.1542,
+ "step": 12953
+ },
+ {
+ "epoch": 35.29700272479564,
+ "grad_norm": 4.3449201583862305,
+ "learning_rate": 1.5010186706639412e-05,
+ "loss": 0.1746,
+ "step": 12954
+ },
+ {
+ "epoch": 35.299727520435965,
+ "grad_norm": 6.395719528198242,
+ "learning_rate": 1.5009422944084411e-05,
+ "loss": 0.1693,
+ "step": 12955
+ },
+ {
+ "epoch": 35.30245231607629,
+ "grad_norm": 5.582629680633545,
+ "learning_rate": 1.500865914251619e-05,
+ "loss": 0.2959,
+ "step": 12956
+ },
+ {
+ "epoch": 35.305177111716624,
+ "grad_norm": 5.014416694641113,
+ "learning_rate": 1.500789530194069e-05,
+ "loss": 0.1983,
+ "step": 12957
+ },
+ {
+ "epoch": 35.30790190735695,
+ "grad_norm": 3.969390392303467,
+ "learning_rate": 1.5007131422363864e-05,
+ "loss": 0.2922,
+ "step": 12958
+ },
+ {
+ "epoch": 35.310626702997276,
+ "grad_norm": 3.7562599182128906,
+ "learning_rate": 1.5006367503791657e-05,
+ "loss": 0.1517,
+ "step": 12959
+ },
+ {
+ "epoch": 35.3133514986376,
+ "grad_norm": 3.8391759395599365,
+ "learning_rate": 1.5005603546230024e-05,
+ "loss": 0.1327,
+ "step": 12960
+ },
+ {
+ "epoch": 35.31607629427793,
+ "grad_norm": 4.1435017585754395,
+ "learning_rate": 1.5004839549684907e-05,
+ "loss": 0.0992,
+ "step": 12961
+ },
+ {
+ "epoch": 35.31880108991825,
+ "grad_norm": 4.357354164123535,
+ "learning_rate": 1.5004075514162264e-05,
+ "loss": 0.1792,
+ "step": 12962
+ },
+ {
+ "epoch": 35.321525885558586,
+ "grad_norm": 4.8638386726379395,
+ "learning_rate": 1.5003311439668041e-05,
+ "loss": 0.1904,
+ "step": 12963
+ },
+ {
+ "epoch": 35.32425068119891,
+ "grad_norm": 6.855134963989258,
+ "learning_rate": 1.5002547326208193e-05,
+ "loss": 0.1857,
+ "step": 12964
+ },
+ {
+ "epoch": 35.32697547683924,
+ "grad_norm": 3.4851222038269043,
+ "learning_rate": 1.5001783173788661e-05,
+ "loss": 0.1532,
+ "step": 12965
+ },
+ {
+ "epoch": 35.32970027247956,
+ "grad_norm": 4.445806503295898,
+ "learning_rate": 1.5001018982415407e-05,
+ "loss": 0.1298,
+ "step": 12966
+ },
+ {
+ "epoch": 35.33242506811989,
+ "grad_norm": 4.30107307434082,
+ "learning_rate": 1.5000254752094376e-05,
+ "loss": 0.166,
+ "step": 12967
+ },
+ {
+ "epoch": 35.335149863760215,
+ "grad_norm": 3.6918768882751465,
+ "learning_rate": 1.4999490482831526e-05,
+ "loss": 0.2088,
+ "step": 12968
+ },
+ {
+ "epoch": 35.33787465940055,
+ "grad_norm": 3.499185562133789,
+ "learning_rate": 1.49987261746328e-05,
+ "loss": 0.1366,
+ "step": 12969
+ },
+ {
+ "epoch": 35.34059945504087,
+ "grad_norm": 4.557539939880371,
+ "learning_rate": 1.4997961827504156e-05,
+ "loss": 0.411,
+ "step": 12970
+ },
+ {
+ "epoch": 35.3433242506812,
+ "grad_norm": 4.804594993591309,
+ "learning_rate": 1.499719744145155e-05,
+ "loss": 0.3562,
+ "step": 12971
+ },
+ {
+ "epoch": 35.346049046321525,
+ "grad_norm": 4.058318614959717,
+ "learning_rate": 1.499643301648093e-05,
+ "loss": 0.2807,
+ "step": 12972
+ },
+ {
+ "epoch": 35.34877384196185,
+ "grad_norm": 4.44939661026001,
+ "learning_rate": 1.4995668552598249e-05,
+ "loss": 0.2087,
+ "step": 12973
+ },
+ {
+ "epoch": 35.35149863760218,
+ "grad_norm": 4.338406085968018,
+ "learning_rate": 1.4994904049809463e-05,
+ "loss": 0.2563,
+ "step": 12974
+ },
+ {
+ "epoch": 35.35422343324251,
+ "grad_norm": 3.701251983642578,
+ "learning_rate": 1.4994139508120524e-05,
+ "loss": 0.1593,
+ "step": 12975
+ },
+ {
+ "epoch": 35.356948228882835,
+ "grad_norm": 4.0285844802856445,
+ "learning_rate": 1.499337492753739e-05,
+ "loss": 0.2462,
+ "step": 12976
+ },
+ {
+ "epoch": 35.35967302452316,
+ "grad_norm": 4.30666971206665,
+ "learning_rate": 1.4992610308066008e-05,
+ "loss": 0.2019,
+ "step": 12977
+ },
+ {
+ "epoch": 35.36239782016349,
+ "grad_norm": 4.7420549392700195,
+ "learning_rate": 1.499184564971234e-05,
+ "loss": 0.1793,
+ "step": 12978
+ },
+ {
+ "epoch": 35.36512261580381,
+ "grad_norm": 2.964585542678833,
+ "learning_rate": 1.4991080952482337e-05,
+ "loss": 0.1072,
+ "step": 12979
+ },
+ {
+ "epoch": 35.36784741144414,
+ "grad_norm": 5.0368499755859375,
+ "learning_rate": 1.4990316216381961e-05,
+ "loss": 0.1788,
+ "step": 12980
+ },
+ {
+ "epoch": 35.37057220708447,
+ "grad_norm": 4.200316905975342,
+ "learning_rate": 1.4989551441417159e-05,
+ "loss": 0.3448,
+ "step": 12981
+ },
+ {
+ "epoch": 35.3732970027248,
+ "grad_norm": 4.785541534423828,
+ "learning_rate": 1.4988786627593891e-05,
+ "loss": 0.2558,
+ "step": 12982
+ },
+ {
+ "epoch": 35.37602179836512,
+ "grad_norm": 4.06687068939209,
+ "learning_rate": 1.4988021774918116e-05,
+ "loss": 0.1752,
+ "step": 12983
+ },
+ {
+ "epoch": 35.37874659400545,
+ "grad_norm": 4.185251712799072,
+ "learning_rate": 1.4987256883395786e-05,
+ "loss": 0.1392,
+ "step": 12984
+ },
+ {
+ "epoch": 35.381471389645775,
+ "grad_norm": 4.375302791595459,
+ "learning_rate": 1.498649195303286e-05,
+ "loss": 0.2786,
+ "step": 12985
+ },
+ {
+ "epoch": 35.3841961852861,
+ "grad_norm": 3.6436538696289062,
+ "learning_rate": 1.4985726983835296e-05,
+ "loss": 0.151,
+ "step": 12986
+ },
+ {
+ "epoch": 35.38692098092643,
+ "grad_norm": 6.442212104797363,
+ "learning_rate": 1.498496197580905e-05,
+ "loss": 0.2409,
+ "step": 12987
+ },
+ {
+ "epoch": 35.38964577656676,
+ "grad_norm": 4.053532600402832,
+ "learning_rate": 1.498419692896008e-05,
+ "loss": 0.1638,
+ "step": 12988
+ },
+ {
+ "epoch": 35.392370572207085,
+ "grad_norm": 4.325973987579346,
+ "learning_rate": 1.4983431843294345e-05,
+ "loss": 0.0986,
+ "step": 12989
+ },
+ {
+ "epoch": 35.39509536784741,
+ "grad_norm": 4.880702018737793,
+ "learning_rate": 1.4982666718817804e-05,
+ "loss": 0.1382,
+ "step": 12990
+ },
+ {
+ "epoch": 35.39782016348774,
+ "grad_norm": 6.177012920379639,
+ "learning_rate": 1.4981901555536414e-05,
+ "loss": 0.3483,
+ "step": 12991
+ },
+ {
+ "epoch": 35.40054495912806,
+ "grad_norm": 5.096782684326172,
+ "learning_rate": 1.4981136353456135e-05,
+ "loss": 0.1831,
+ "step": 12992
+ },
+ {
+ "epoch": 35.403269754768395,
+ "grad_norm": 3.9485371112823486,
+ "learning_rate": 1.4980371112582927e-05,
+ "loss": 0.1433,
+ "step": 12993
+ },
+ {
+ "epoch": 35.40599455040872,
+ "grad_norm": 4.159912109375,
+ "learning_rate": 1.497960583292275e-05,
+ "loss": 0.1891,
+ "step": 12994
+ },
+ {
+ "epoch": 35.40871934604905,
+ "grad_norm": 4.713575839996338,
+ "learning_rate": 1.497884051448156e-05,
+ "loss": 0.1826,
+ "step": 12995
+ },
+ {
+ "epoch": 35.41144414168937,
+ "grad_norm": 4.425631046295166,
+ "learning_rate": 1.4978075157265324e-05,
+ "loss": 0.2449,
+ "step": 12996
+ },
+ {
+ "epoch": 35.4141689373297,
+ "grad_norm": 3.63668155670166,
+ "learning_rate": 1.4977309761279997e-05,
+ "loss": 0.1877,
+ "step": 12997
+ },
+ {
+ "epoch": 35.416893732970024,
+ "grad_norm": 4.844672679901123,
+ "learning_rate": 1.4976544326531542e-05,
+ "loss": 0.1658,
+ "step": 12998
+ },
+ {
+ "epoch": 35.41961852861036,
+ "grad_norm": 5.235696792602539,
+ "learning_rate": 1.497577885302592e-05,
+ "loss": 0.1373,
+ "step": 12999
+ },
+ {
+ "epoch": 35.42234332425068,
+ "grad_norm": 5.46209716796875,
+ "learning_rate": 1.4975013340769094e-05,
+ "loss": 0.1471,
+ "step": 13000
+ },
+ {
+ "epoch": 35.42506811989101,
+ "grad_norm": 6.566809177398682,
+ "learning_rate": 1.497424778976702e-05,
+ "loss": 0.1367,
+ "step": 13001
+ },
+ {
+ "epoch": 35.427792915531334,
+ "grad_norm": 4.545602798461914,
+ "learning_rate": 1.4973482200025669e-05,
+ "loss": 0.2527,
+ "step": 13002
+ },
+ {
+ "epoch": 35.43051771117166,
+ "grad_norm": 5.268848896026611,
+ "learning_rate": 1.4972716571550996e-05,
+ "loss": 0.2091,
+ "step": 13003
+ },
+ {
+ "epoch": 35.433242506811986,
+ "grad_norm": 4.332236289978027,
+ "learning_rate": 1.4971950904348966e-05,
+ "loss": 0.1612,
+ "step": 13004
+ },
+ {
+ "epoch": 35.43596730245232,
+ "grad_norm": 4.281850337982178,
+ "learning_rate": 1.4971185198425544e-05,
+ "loss": 0.2468,
+ "step": 13005
+ },
+ {
+ "epoch": 35.438692098092645,
+ "grad_norm": 4.628166675567627,
+ "learning_rate": 1.497041945378669e-05,
+ "loss": 0.3289,
+ "step": 13006
+ },
+ {
+ "epoch": 35.44141689373297,
+ "grad_norm": 4.808790683746338,
+ "learning_rate": 1.4969653670438372e-05,
+ "loss": 0.1709,
+ "step": 13007
+ },
+ {
+ "epoch": 35.444141689373296,
+ "grad_norm": 4.054821014404297,
+ "learning_rate": 1.496888784838655e-05,
+ "loss": 0.2759,
+ "step": 13008
+ },
+ {
+ "epoch": 35.44686648501362,
+ "grad_norm": 3.6904172897338867,
+ "learning_rate": 1.4968121987637188e-05,
+ "loss": 0.333,
+ "step": 13009
+ },
+ {
+ "epoch": 35.44959128065395,
+ "grad_norm": 5.441751480102539,
+ "learning_rate": 1.4967356088196252e-05,
+ "loss": 0.1878,
+ "step": 13010
+ },
+ {
+ "epoch": 35.45231607629428,
+ "grad_norm": 5.056656360626221,
+ "learning_rate": 1.496659015006971e-05,
+ "loss": 0.1189,
+ "step": 13011
+ },
+ {
+ "epoch": 35.45504087193461,
+ "grad_norm": 5.967219352722168,
+ "learning_rate": 1.4965824173263519e-05,
+ "loss": 0.1514,
+ "step": 13012
+ },
+ {
+ "epoch": 35.45776566757493,
+ "grad_norm": 4.734949588775635,
+ "learning_rate": 1.4965058157783651e-05,
+ "loss": 0.1641,
+ "step": 13013
+ },
+ {
+ "epoch": 35.46049046321526,
+ "grad_norm": 3.67097544670105,
+ "learning_rate": 1.4964292103636072e-05,
+ "loss": 0.1743,
+ "step": 13014
+ },
+ {
+ "epoch": 35.463215258855584,
+ "grad_norm": 4.714073657989502,
+ "learning_rate": 1.496352601082674e-05,
+ "loss": 0.2006,
+ "step": 13015
+ },
+ {
+ "epoch": 35.46594005449591,
+ "grad_norm": 4.3891801834106445,
+ "learning_rate": 1.4962759879361635e-05,
+ "loss": 0.159,
+ "step": 13016
+ },
+ {
+ "epoch": 35.46866485013624,
+ "grad_norm": 3.7784204483032227,
+ "learning_rate": 1.4961993709246709e-05,
+ "loss": 0.2493,
+ "step": 13017
+ },
+ {
+ "epoch": 35.47138964577657,
+ "grad_norm": 4.6050028800964355,
+ "learning_rate": 1.496122750048794e-05,
+ "loss": 0.1518,
+ "step": 13018
+ },
+ {
+ "epoch": 35.474114441416894,
+ "grad_norm": 4.518208026885986,
+ "learning_rate": 1.496046125309129e-05,
+ "loss": 0.2015,
+ "step": 13019
+ },
+ {
+ "epoch": 35.47683923705722,
+ "grad_norm": 3.8003554344177246,
+ "learning_rate": 1.4959694967062727e-05,
+ "loss": 0.283,
+ "step": 13020
+ },
+ {
+ "epoch": 35.479564032697546,
+ "grad_norm": 3.742164373397827,
+ "learning_rate": 1.4958928642408219e-05,
+ "loss": 0.1698,
+ "step": 13021
+ },
+ {
+ "epoch": 35.48228882833787,
+ "grad_norm": 4.267362117767334,
+ "learning_rate": 1.4958162279133734e-05,
+ "loss": 0.211,
+ "step": 13022
+ },
+ {
+ "epoch": 35.485013623978205,
+ "grad_norm": 143.3149871826172,
+ "learning_rate": 1.4957395877245242e-05,
+ "loss": 0.3494,
+ "step": 13023
+ },
+ {
+ "epoch": 35.48773841961853,
+ "grad_norm": 5.248658657073975,
+ "learning_rate": 1.4956629436748712e-05,
+ "loss": 0.1479,
+ "step": 13024
+ },
+ {
+ "epoch": 35.490463215258856,
+ "grad_norm": 4.090874671936035,
+ "learning_rate": 1.4955862957650108e-05,
+ "loss": 0.2726,
+ "step": 13025
+ },
+ {
+ "epoch": 35.49318801089918,
+ "grad_norm": 4.718405723571777,
+ "learning_rate": 1.4955096439955404e-05,
+ "loss": 0.2013,
+ "step": 13026
+ },
+ {
+ "epoch": 35.49591280653951,
+ "grad_norm": 4.434218406677246,
+ "learning_rate": 1.4954329883670569e-05,
+ "loss": 0.2979,
+ "step": 13027
+ },
+ {
+ "epoch": 35.49863760217983,
+ "grad_norm": 12.399559020996094,
+ "learning_rate": 1.4953563288801571e-05,
+ "loss": 0.2842,
+ "step": 13028
+ },
+ {
+ "epoch": 35.50136239782017,
+ "grad_norm": 6.77754020690918,
+ "learning_rate": 1.4952796655354383e-05,
+ "loss": 0.2796,
+ "step": 13029
+ },
+ {
+ "epoch": 35.50408719346049,
+ "grad_norm": 8.023210525512695,
+ "learning_rate": 1.4952029983334973e-05,
+ "loss": 0.2675,
+ "step": 13030
+ },
+ {
+ "epoch": 35.50681198910082,
+ "grad_norm": 7.539895534515381,
+ "learning_rate": 1.4951263272749313e-05,
+ "loss": 0.3327,
+ "step": 13031
+ },
+ {
+ "epoch": 35.509536784741144,
+ "grad_norm": 4.205755710601807,
+ "learning_rate": 1.4950496523603373e-05,
+ "loss": 0.1645,
+ "step": 13032
+ },
+ {
+ "epoch": 35.51226158038147,
+ "grad_norm": 6.905337810516357,
+ "learning_rate": 1.4949729735903127e-05,
+ "loss": 0.3237,
+ "step": 13033
+ },
+ {
+ "epoch": 35.514986376021795,
+ "grad_norm": 7.350479602813721,
+ "learning_rate": 1.4948962909654545e-05,
+ "loss": 0.2233,
+ "step": 13034
+ },
+ {
+ "epoch": 35.51771117166213,
+ "grad_norm": 6.934208393096924,
+ "learning_rate": 1.4948196044863599e-05,
+ "loss": 0.2503,
+ "step": 13035
+ },
+ {
+ "epoch": 35.520435967302454,
+ "grad_norm": 7.4420084953308105,
+ "learning_rate": 1.4947429141536262e-05,
+ "loss": 0.428,
+ "step": 13036
+ },
+ {
+ "epoch": 35.52316076294278,
+ "grad_norm": 7.076702117919922,
+ "learning_rate": 1.4946662199678503e-05,
+ "loss": 0.3726,
+ "step": 13037
+ },
+ {
+ "epoch": 35.525885558583106,
+ "grad_norm": 5.655088424682617,
+ "learning_rate": 1.49458952192963e-05,
+ "loss": 0.3107,
+ "step": 13038
+ },
+ {
+ "epoch": 35.52861035422343,
+ "grad_norm": 5.96241569519043,
+ "learning_rate": 1.4945128200395627e-05,
+ "loss": 0.3379,
+ "step": 13039
+ },
+ {
+ "epoch": 35.53133514986376,
+ "grad_norm": 6.668216228485107,
+ "learning_rate": 1.494436114298245e-05,
+ "loss": 0.2371,
+ "step": 13040
+ },
+ {
+ "epoch": 35.53405994550409,
+ "grad_norm": 5.729973793029785,
+ "learning_rate": 1.4943594047062752e-05,
+ "loss": 0.2365,
+ "step": 13041
+ },
+ {
+ "epoch": 35.536784741144416,
+ "grad_norm": 6.239397048950195,
+ "learning_rate": 1.4942826912642501e-05,
+ "loss": 0.2963,
+ "step": 13042
+ },
+ {
+ "epoch": 35.53950953678474,
+ "grad_norm": 13.716826438903809,
+ "learning_rate": 1.4942059739727674e-05,
+ "loss": 0.2714,
+ "step": 13043
+ },
+ {
+ "epoch": 35.54223433242507,
+ "grad_norm": 6.69462776184082,
+ "learning_rate": 1.4941292528324243e-05,
+ "loss": 0.3583,
+ "step": 13044
+ },
+ {
+ "epoch": 35.54495912806539,
+ "grad_norm": 7.304836273193359,
+ "learning_rate": 1.4940525278438185e-05,
+ "loss": 0.1919,
+ "step": 13045
+ },
+ {
+ "epoch": 35.54768392370572,
+ "grad_norm": 6.381016731262207,
+ "learning_rate": 1.4939757990075477e-05,
+ "loss": 0.2715,
+ "step": 13046
+ },
+ {
+ "epoch": 35.55040871934605,
+ "grad_norm": 4.59263277053833,
+ "learning_rate": 1.4938990663242088e-05,
+ "loss": 0.2507,
+ "step": 13047
+ },
+ {
+ "epoch": 35.55313351498638,
+ "grad_norm": 6.643576622009277,
+ "learning_rate": 1.4938223297944002e-05,
+ "loss": 0.3152,
+ "step": 13048
+ },
+ {
+ "epoch": 35.555858310626704,
+ "grad_norm": 6.813570976257324,
+ "learning_rate": 1.4937455894187194e-05,
+ "loss": 0.2436,
+ "step": 13049
+ },
+ {
+ "epoch": 35.55858310626703,
+ "grad_norm": 5.28364896774292,
+ "learning_rate": 1.4936688451977634e-05,
+ "loss": 0.1638,
+ "step": 13050
+ },
+ {
+ "epoch": 35.561307901907355,
+ "grad_norm": 5.779529094696045,
+ "learning_rate": 1.4935920971321307e-05,
+ "loss": 0.3256,
+ "step": 13051
+ },
+ {
+ "epoch": 35.56403269754768,
+ "grad_norm": 4.761242389678955,
+ "learning_rate": 1.4935153452224182e-05,
+ "loss": 0.2086,
+ "step": 13052
+ },
+ {
+ "epoch": 35.566757493188014,
+ "grad_norm": 6.375670909881592,
+ "learning_rate": 1.4934385894692247e-05,
+ "loss": 0.2838,
+ "step": 13053
+ },
+ {
+ "epoch": 35.56948228882834,
+ "grad_norm": 5.9322919845581055,
+ "learning_rate": 1.493361829873147e-05,
+ "loss": 0.2937,
+ "step": 13054
+ },
+ {
+ "epoch": 35.572207084468666,
+ "grad_norm": 5.634441375732422,
+ "learning_rate": 1.493285066434783e-05,
+ "loss": 0.3019,
+ "step": 13055
+ },
+ {
+ "epoch": 35.57493188010899,
+ "grad_norm": 8.522860527038574,
+ "learning_rate": 1.493208299154731e-05,
+ "loss": 0.2754,
+ "step": 13056
+ },
+ {
+ "epoch": 35.57765667574932,
+ "grad_norm": 4.747716426849365,
+ "learning_rate": 1.4931315280335888e-05,
+ "loss": 0.2119,
+ "step": 13057
+ },
+ {
+ "epoch": 35.58038147138964,
+ "grad_norm": 6.240516662597656,
+ "learning_rate": 1.4930547530719539e-05,
+ "loss": 0.1654,
+ "step": 13058
+ },
+ {
+ "epoch": 35.583106267029976,
+ "grad_norm": 5.059964179992676,
+ "learning_rate": 1.4929779742704246e-05,
+ "loss": 0.2612,
+ "step": 13059
+ },
+ {
+ "epoch": 35.5858310626703,
+ "grad_norm": 5.080691337585449,
+ "learning_rate": 1.4929011916295986e-05,
+ "loss": 0.1611,
+ "step": 13060
+ },
+ {
+ "epoch": 35.58855585831063,
+ "grad_norm": 5.747907638549805,
+ "learning_rate": 1.4928244051500742e-05,
+ "loss": 0.1696,
+ "step": 13061
+ },
+ {
+ "epoch": 35.59128065395095,
+ "grad_norm": 5.291440963745117,
+ "learning_rate": 1.4927476148324488e-05,
+ "loss": 0.2268,
+ "step": 13062
+ },
+ {
+ "epoch": 35.59400544959128,
+ "grad_norm": 6.02871561050415,
+ "learning_rate": 1.4926708206773213e-05,
+ "loss": 0.3525,
+ "step": 13063
+ },
+ {
+ "epoch": 35.596730245231605,
+ "grad_norm": 4.125657081604004,
+ "learning_rate": 1.4925940226852889e-05,
+ "loss": 0.2467,
+ "step": 13064
+ },
+ {
+ "epoch": 35.59945504087194,
+ "grad_norm": 5.131707668304443,
+ "learning_rate": 1.4925172208569502e-05,
+ "loss": 0.238,
+ "step": 13065
+ },
+ {
+ "epoch": 35.60217983651226,
+ "grad_norm": 7.76033878326416,
+ "learning_rate": 1.4924404151929033e-05,
+ "loss": 0.4094,
+ "step": 13066
+ },
+ {
+ "epoch": 35.60490463215259,
+ "grad_norm": 5.540083885192871,
+ "learning_rate": 1.4923636056937466e-05,
+ "loss": 0.1951,
+ "step": 13067
+ },
+ {
+ "epoch": 35.607629427792915,
+ "grad_norm": 8.395087242126465,
+ "learning_rate": 1.4922867923600774e-05,
+ "loss": 0.2464,
+ "step": 13068
+ },
+ {
+ "epoch": 35.61035422343324,
+ "grad_norm": 4.31557559967041,
+ "learning_rate": 1.492209975192495e-05,
+ "loss": 0.2076,
+ "step": 13069
+ },
+ {
+ "epoch": 35.61307901907357,
+ "grad_norm": 5.852571964263916,
+ "learning_rate": 1.492133154191597e-05,
+ "loss": 0.1904,
+ "step": 13070
+ },
+ {
+ "epoch": 35.6158038147139,
+ "grad_norm": 4.731469631195068,
+ "learning_rate": 1.4920563293579817e-05,
+ "loss": 0.3603,
+ "step": 13071
+ },
+ {
+ "epoch": 35.618528610354225,
+ "grad_norm": 4.541119575500488,
+ "learning_rate": 1.4919795006922475e-05,
+ "loss": 0.1538,
+ "step": 13072
+ },
+ {
+ "epoch": 35.62125340599455,
+ "grad_norm": 5.129117012023926,
+ "learning_rate": 1.491902668194993e-05,
+ "loss": 0.2325,
+ "step": 13073
+ },
+ {
+ "epoch": 35.62397820163488,
+ "grad_norm": 5.976446628570557,
+ "learning_rate": 1.491825831866816e-05,
+ "loss": 0.1588,
+ "step": 13074
+ },
+ {
+ "epoch": 35.6267029972752,
+ "grad_norm": 5.155496597290039,
+ "learning_rate": 1.4917489917083157e-05,
+ "loss": 0.208,
+ "step": 13075
+ },
+ {
+ "epoch": 35.62942779291553,
+ "grad_norm": 4.229480266571045,
+ "learning_rate": 1.49167214772009e-05,
+ "loss": 0.2575,
+ "step": 13076
+ },
+ {
+ "epoch": 35.63215258855586,
+ "grad_norm": 4.84398889541626,
+ "learning_rate": 1.4915952999027371e-05,
+ "loss": 0.256,
+ "step": 13077
+ },
+ {
+ "epoch": 35.63487738419619,
+ "grad_norm": 5.300017356872559,
+ "learning_rate": 1.4915184482568562e-05,
+ "loss": 0.1702,
+ "step": 13078
+ },
+ {
+ "epoch": 35.63760217983651,
+ "grad_norm": 4.25277853012085,
+ "learning_rate": 1.4914415927830455e-05,
+ "loss": 0.2213,
+ "step": 13079
+ },
+ {
+ "epoch": 35.64032697547684,
+ "grad_norm": 4.818574905395508,
+ "learning_rate": 1.491364733481903e-05,
+ "loss": 0.3513,
+ "step": 13080
+ },
+ {
+ "epoch": 35.643051771117165,
+ "grad_norm": 7.232956409454346,
+ "learning_rate": 1.4912878703540281e-05,
+ "loss": 0.1682,
+ "step": 13081
+ },
+ {
+ "epoch": 35.64577656675749,
+ "grad_norm": 4.405952453613281,
+ "learning_rate": 1.4912110034000189e-05,
+ "loss": 0.2632,
+ "step": 13082
+ },
+ {
+ "epoch": 35.64850136239782,
+ "grad_norm": 5.124270915985107,
+ "learning_rate": 1.4911341326204744e-05,
+ "loss": 0.2124,
+ "step": 13083
+ },
+ {
+ "epoch": 35.65122615803815,
+ "grad_norm": 4.954735279083252,
+ "learning_rate": 1.491057258015993e-05,
+ "loss": 0.1967,
+ "step": 13084
+ },
+ {
+ "epoch": 35.653950953678475,
+ "grad_norm": 4.650579929351807,
+ "learning_rate": 1.4909803795871738e-05,
+ "loss": 0.1299,
+ "step": 13085
+ },
+ {
+ "epoch": 35.6566757493188,
+ "grad_norm": 3.714184045791626,
+ "learning_rate": 1.4909034973346148e-05,
+ "loss": 0.2292,
+ "step": 13086
+ },
+ {
+ "epoch": 35.65940054495913,
+ "grad_norm": 5.20432186126709,
+ "learning_rate": 1.4908266112589154e-05,
+ "loss": 0.3738,
+ "step": 13087
+ },
+ {
+ "epoch": 35.66212534059945,
+ "grad_norm": 3.4993040561676025,
+ "learning_rate": 1.4907497213606738e-05,
+ "loss": 0.2418,
+ "step": 13088
+ },
+ {
+ "epoch": 35.664850136239785,
+ "grad_norm": 5.793128967285156,
+ "learning_rate": 1.4906728276404897e-05,
+ "loss": 0.2001,
+ "step": 13089
+ },
+ {
+ "epoch": 35.66757493188011,
+ "grad_norm": 6.007649898529053,
+ "learning_rate": 1.4905959300989607e-05,
+ "loss": 0.2869,
+ "step": 13090
+ },
+ {
+ "epoch": 35.67029972752044,
+ "grad_norm": 4.367301940917969,
+ "learning_rate": 1.4905190287366868e-05,
+ "loss": 0.1776,
+ "step": 13091
+ },
+ {
+ "epoch": 35.67302452316076,
+ "grad_norm": 6.266600131988525,
+ "learning_rate": 1.4904421235542666e-05,
+ "loss": 0.168,
+ "step": 13092
+ },
+ {
+ "epoch": 35.67574931880109,
+ "grad_norm": 5.083538055419922,
+ "learning_rate": 1.4903652145522989e-05,
+ "loss": 0.2018,
+ "step": 13093
+ },
+ {
+ "epoch": 35.678474114441414,
+ "grad_norm": 4.731904029846191,
+ "learning_rate": 1.4902883017313825e-05,
+ "loss": 0.2824,
+ "step": 13094
+ },
+ {
+ "epoch": 35.68119891008175,
+ "grad_norm": 4.35421895980835,
+ "learning_rate": 1.4902113850921169e-05,
+ "loss": 0.3289,
+ "step": 13095
+ },
+ {
+ "epoch": 35.68392370572207,
+ "grad_norm": 4.889714241027832,
+ "learning_rate": 1.4901344646351006e-05,
+ "loss": 0.1552,
+ "step": 13096
+ },
+ {
+ "epoch": 35.6866485013624,
+ "grad_norm": 4.448430061340332,
+ "learning_rate": 1.4900575403609328e-05,
+ "loss": 0.1708,
+ "step": 13097
+ },
+ {
+ "epoch": 35.689373297002724,
+ "grad_norm": 4.381519794464111,
+ "learning_rate": 1.4899806122702127e-05,
+ "loss": 0.2808,
+ "step": 13098
+ },
+ {
+ "epoch": 35.69209809264305,
+ "grad_norm": 3.85894513130188,
+ "learning_rate": 1.4899036803635393e-05,
+ "loss": 0.2752,
+ "step": 13099
+ },
+ {
+ "epoch": 35.694822888283376,
+ "grad_norm": 3.7592456340789795,
+ "learning_rate": 1.489826744641512e-05,
+ "loss": 0.1375,
+ "step": 13100
+ },
+ {
+ "epoch": 35.69754768392371,
+ "grad_norm": 4.325622081756592,
+ "learning_rate": 1.4897498051047297e-05,
+ "loss": 0.1454,
+ "step": 13101
+ },
+ {
+ "epoch": 35.700272479564035,
+ "grad_norm": 3.8246781826019287,
+ "learning_rate": 1.4896728617537914e-05,
+ "loss": 0.422,
+ "step": 13102
+ },
+ {
+ "epoch": 35.70299727520436,
+ "grad_norm": 3.7790422439575195,
+ "learning_rate": 1.489595914589297e-05,
+ "loss": 0.173,
+ "step": 13103
+ },
+ {
+ "epoch": 35.705722070844686,
+ "grad_norm": 4.811452388763428,
+ "learning_rate": 1.4895189636118453e-05,
+ "loss": 0.2352,
+ "step": 13104
+ },
+ {
+ "epoch": 35.70844686648501,
+ "grad_norm": 4.840193748474121,
+ "learning_rate": 1.4894420088220356e-05,
+ "loss": 0.1885,
+ "step": 13105
+ },
+ {
+ "epoch": 35.71117166212534,
+ "grad_norm": 5.039525508880615,
+ "learning_rate": 1.4893650502204673e-05,
+ "loss": 0.1738,
+ "step": 13106
+ },
+ {
+ "epoch": 35.71389645776567,
+ "grad_norm": 4.877858638763428,
+ "learning_rate": 1.4892880878077396e-05,
+ "loss": 0.4121,
+ "step": 13107
+ },
+ {
+ "epoch": 35.716621253406,
+ "grad_norm": 4.6762471199035645,
+ "learning_rate": 1.4892111215844523e-05,
+ "loss": 0.1779,
+ "step": 13108
+ },
+ {
+ "epoch": 35.71934604904632,
+ "grad_norm": 4.646894454956055,
+ "learning_rate": 1.489134151551204e-05,
+ "loss": 0.2887,
+ "step": 13109
+ },
+ {
+ "epoch": 35.72207084468665,
+ "grad_norm": 3.9621543884277344,
+ "learning_rate": 1.4890571777085954e-05,
+ "loss": 0.1311,
+ "step": 13110
+ },
+ {
+ "epoch": 35.724795640326974,
+ "grad_norm": 4.204470634460449,
+ "learning_rate": 1.4889802000572247e-05,
+ "loss": 0.2755,
+ "step": 13111
+ },
+ {
+ "epoch": 35.7275204359673,
+ "grad_norm": 4.183505535125732,
+ "learning_rate": 1.4889032185976922e-05,
+ "loss": 0.2119,
+ "step": 13112
+ },
+ {
+ "epoch": 35.73024523160763,
+ "grad_norm": 5.502959728240967,
+ "learning_rate": 1.4888262333305971e-05,
+ "loss": 0.2344,
+ "step": 13113
+ },
+ {
+ "epoch": 35.73297002724796,
+ "grad_norm": 4.546055793762207,
+ "learning_rate": 1.488749244256539e-05,
+ "loss": 0.2877,
+ "step": 13114
+ },
+ {
+ "epoch": 35.735694822888284,
+ "grad_norm": 4.891136169433594,
+ "learning_rate": 1.4886722513761176e-05,
+ "loss": 0.1855,
+ "step": 13115
+ },
+ {
+ "epoch": 35.73841961852861,
+ "grad_norm": 4.980343341827393,
+ "learning_rate": 1.4885952546899323e-05,
+ "loss": 0.2764,
+ "step": 13116
+ },
+ {
+ "epoch": 35.741144414168936,
+ "grad_norm": 7.542430877685547,
+ "learning_rate": 1.4885182541985827e-05,
+ "loss": 0.1898,
+ "step": 13117
+ },
+ {
+ "epoch": 35.74386920980926,
+ "grad_norm": 5.776932239532471,
+ "learning_rate": 1.4884412499026693e-05,
+ "loss": 0.2337,
+ "step": 13118
+ },
+ {
+ "epoch": 35.746594005449595,
+ "grad_norm": 4.72122859954834,
+ "learning_rate": 1.4883642418027906e-05,
+ "loss": 0.1821,
+ "step": 13119
+ },
+ {
+ "epoch": 35.74931880108992,
+ "grad_norm": 5.17772102355957,
+ "learning_rate": 1.4882872298995472e-05,
+ "loss": 0.1593,
+ "step": 13120
+ },
+ {
+ "epoch": 35.752043596730246,
+ "grad_norm": 4.028685092926025,
+ "learning_rate": 1.4882102141935381e-05,
+ "loss": 0.1483,
+ "step": 13121
+ },
+ {
+ "epoch": 35.75476839237057,
+ "grad_norm": 4.982350826263428,
+ "learning_rate": 1.4881331946853643e-05,
+ "loss": 0.2295,
+ "step": 13122
+ },
+ {
+ "epoch": 35.7574931880109,
+ "grad_norm": 5.0773396492004395,
+ "learning_rate": 1.4880561713756243e-05,
+ "loss": 0.2018,
+ "step": 13123
+ },
+ {
+ "epoch": 35.76021798365122,
+ "grad_norm": 3.9460132122039795,
+ "learning_rate": 1.4879791442649188e-05,
+ "loss": 0.1906,
+ "step": 13124
+ },
+ {
+ "epoch": 35.762942779291556,
+ "grad_norm": 5.051455497741699,
+ "learning_rate": 1.4879021133538471e-05,
+ "loss": 0.3445,
+ "step": 13125
+ },
+ {
+ "epoch": 35.76566757493188,
+ "grad_norm": 4.3634490966796875,
+ "learning_rate": 1.4878250786430099e-05,
+ "loss": 0.1542,
+ "step": 13126
+ },
+ {
+ "epoch": 35.76839237057221,
+ "grad_norm": 4.1861252784729,
+ "learning_rate": 1.4877480401330065e-05,
+ "loss": 0.3085,
+ "step": 13127
+ },
+ {
+ "epoch": 35.771117166212534,
+ "grad_norm": 4.37587308883667,
+ "learning_rate": 1.4876709978244372e-05,
+ "loss": 0.192,
+ "step": 13128
+ },
+ {
+ "epoch": 35.77384196185286,
+ "grad_norm": 5.53675651550293,
+ "learning_rate": 1.4875939517179016e-05,
+ "loss": 0.2951,
+ "step": 13129
+ },
+ {
+ "epoch": 35.776566757493185,
+ "grad_norm": 4.858280658721924,
+ "learning_rate": 1.4875169018140003e-05,
+ "loss": 0.1905,
+ "step": 13130
+ },
+ {
+ "epoch": 35.77929155313352,
+ "grad_norm": 4.2232465744018555,
+ "learning_rate": 1.4874398481133327e-05,
+ "loss": 0.2486,
+ "step": 13131
+ },
+ {
+ "epoch": 35.782016348773844,
+ "grad_norm": 4.716830253601074,
+ "learning_rate": 1.4873627906164997e-05,
+ "loss": 0.1655,
+ "step": 13132
+ },
+ {
+ "epoch": 35.78474114441417,
+ "grad_norm": 4.817296981811523,
+ "learning_rate": 1.4872857293241002e-05,
+ "loss": 0.1618,
+ "step": 13133
+ },
+ {
+ "epoch": 35.787465940054496,
+ "grad_norm": 3.969986915588379,
+ "learning_rate": 1.487208664236736e-05,
+ "loss": 0.1044,
+ "step": 13134
+ },
+ {
+ "epoch": 35.79019073569482,
+ "grad_norm": 4.522711753845215,
+ "learning_rate": 1.487131595355006e-05,
+ "loss": 0.1812,
+ "step": 13135
+ },
+ {
+ "epoch": 35.79291553133515,
+ "grad_norm": 4.263194561004639,
+ "learning_rate": 1.487054522679511e-05,
+ "loss": 0.1966,
+ "step": 13136
+ },
+ {
+ "epoch": 35.79564032697548,
+ "grad_norm": 5.172752857208252,
+ "learning_rate": 1.4869774462108506e-05,
+ "loss": 0.1498,
+ "step": 13137
+ },
+ {
+ "epoch": 35.798365122615806,
+ "grad_norm": 4.680929660797119,
+ "learning_rate": 1.4869003659496262e-05,
+ "loss": 0.2196,
+ "step": 13138
+ },
+ {
+ "epoch": 35.80108991825613,
+ "grad_norm": 4.241923809051514,
+ "learning_rate": 1.4868232818964368e-05,
+ "loss": 0.2355,
+ "step": 13139
+ },
+ {
+ "epoch": 35.80381471389646,
+ "grad_norm": 5.378446578979492,
+ "learning_rate": 1.4867461940518836e-05,
+ "loss": 0.1845,
+ "step": 13140
+ },
+ {
+ "epoch": 35.80653950953678,
+ "grad_norm": 5.020999908447266,
+ "learning_rate": 1.4866691024165665e-05,
+ "loss": 0.1831,
+ "step": 13141
+ },
+ {
+ "epoch": 35.80926430517711,
+ "grad_norm": 4.340857982635498,
+ "learning_rate": 1.4865920069910865e-05,
+ "loss": 0.3166,
+ "step": 13142
+ },
+ {
+ "epoch": 35.81198910081744,
+ "grad_norm": 4.293832778930664,
+ "learning_rate": 1.4865149077760435e-05,
+ "loss": 0.1904,
+ "step": 13143
+ },
+ {
+ "epoch": 35.81471389645777,
+ "grad_norm": 4.801344394683838,
+ "learning_rate": 1.486437804772038e-05,
+ "loss": 0.1771,
+ "step": 13144
+ },
+ {
+ "epoch": 35.817438692098094,
+ "grad_norm": 4.855064868927002,
+ "learning_rate": 1.4863606979796703e-05,
+ "loss": 0.2699,
+ "step": 13145
+ },
+ {
+ "epoch": 35.82016348773842,
+ "grad_norm": 4.0539655685424805,
+ "learning_rate": 1.4862835873995416e-05,
+ "loss": 0.1369,
+ "step": 13146
+ },
+ {
+ "epoch": 35.822888283378745,
+ "grad_norm": 5.3960676193237305,
+ "learning_rate": 1.4862064730322516e-05,
+ "loss": 0.1402,
+ "step": 13147
+ },
+ {
+ "epoch": 35.82561307901907,
+ "grad_norm": 4.771040439605713,
+ "learning_rate": 1.4861293548784016e-05,
+ "loss": 0.1386,
+ "step": 13148
+ },
+ {
+ "epoch": 35.828337874659404,
+ "grad_norm": 4.833395957946777,
+ "learning_rate": 1.4860522329385914e-05,
+ "loss": 0.1836,
+ "step": 13149
+ },
+ {
+ "epoch": 35.83106267029973,
+ "grad_norm": 7.365923881530762,
+ "learning_rate": 1.485975107213422e-05,
+ "loss": 0.1734,
+ "step": 13150
+ },
+ {
+ "epoch": 35.833787465940055,
+ "grad_norm": 3.7265877723693848,
+ "learning_rate": 1.4858979777034943e-05,
+ "loss": 0.1185,
+ "step": 13151
+ },
+ {
+ "epoch": 35.83651226158038,
+ "grad_norm": 4.751295566558838,
+ "learning_rate": 1.4858208444094092e-05,
+ "loss": 0.1926,
+ "step": 13152
+ },
+ {
+ "epoch": 35.83923705722071,
+ "grad_norm": 4.4549336433410645,
+ "learning_rate": 1.4857437073317665e-05,
+ "loss": 0.2475,
+ "step": 13153
+ },
+ {
+ "epoch": 35.84196185286103,
+ "grad_norm": 4.722554683685303,
+ "learning_rate": 1.4856665664711676e-05,
+ "loss": 0.2206,
+ "step": 13154
+ },
+ {
+ "epoch": 35.844686648501366,
+ "grad_norm": 4.200508117675781,
+ "learning_rate": 1.485589421828213e-05,
+ "loss": 0.3041,
+ "step": 13155
+ },
+ {
+ "epoch": 35.84741144414169,
+ "grad_norm": 4.402575969696045,
+ "learning_rate": 1.4855122734035037e-05,
+ "loss": 0.1971,
+ "step": 13156
+ },
+ {
+ "epoch": 35.85013623978202,
+ "grad_norm": 4.300631999969482,
+ "learning_rate": 1.4854351211976403e-05,
+ "loss": 0.2271,
+ "step": 13157
+ },
+ {
+ "epoch": 35.85286103542234,
+ "grad_norm": 4.441384315490723,
+ "learning_rate": 1.485357965211224e-05,
+ "loss": 0.3741,
+ "step": 13158
+ },
+ {
+ "epoch": 35.85558583106267,
+ "grad_norm": 3.9720427989959717,
+ "learning_rate": 1.4852808054448553e-05,
+ "loss": 0.2148,
+ "step": 13159
+ },
+ {
+ "epoch": 35.858310626702995,
+ "grad_norm": 3.7725727558135986,
+ "learning_rate": 1.4852036418991356e-05,
+ "loss": 0.1448,
+ "step": 13160
+ },
+ {
+ "epoch": 35.86103542234333,
+ "grad_norm": 4.2280354499816895,
+ "learning_rate": 1.4851264745746653e-05,
+ "loss": 0.1494,
+ "step": 13161
+ },
+ {
+ "epoch": 35.86376021798365,
+ "grad_norm": 6.143813133239746,
+ "learning_rate": 1.4850493034720457e-05,
+ "loss": 0.1492,
+ "step": 13162
+ },
+ {
+ "epoch": 35.86648501362398,
+ "grad_norm": 4.374630451202393,
+ "learning_rate": 1.4849721285918777e-05,
+ "loss": 0.1932,
+ "step": 13163
+ },
+ {
+ "epoch": 35.869209809264305,
+ "grad_norm": 4.536377429962158,
+ "learning_rate": 1.4848949499347625e-05,
+ "loss": 0.1666,
+ "step": 13164
+ },
+ {
+ "epoch": 35.87193460490463,
+ "grad_norm": 4.167899131774902,
+ "learning_rate": 1.4848177675013009e-05,
+ "loss": 0.2269,
+ "step": 13165
+ },
+ {
+ "epoch": 35.87465940054496,
+ "grad_norm": 4.527355194091797,
+ "learning_rate": 1.4847405812920943e-05,
+ "loss": 0.4542,
+ "step": 13166
+ },
+ {
+ "epoch": 35.87738419618529,
+ "grad_norm": 5.818422794342041,
+ "learning_rate": 1.4846633913077432e-05,
+ "loss": 0.1095,
+ "step": 13167
+ },
+ {
+ "epoch": 35.880108991825615,
+ "grad_norm": 4.600836277008057,
+ "learning_rate": 1.4845861975488495e-05,
+ "loss": 0.2381,
+ "step": 13168
+ },
+ {
+ "epoch": 35.88283378746594,
+ "grad_norm": 4.4786858558654785,
+ "learning_rate": 1.4845090000160142e-05,
+ "loss": 0.1858,
+ "step": 13169
+ },
+ {
+ "epoch": 35.88555858310627,
+ "grad_norm": 4.336361408233643,
+ "learning_rate": 1.4844317987098384e-05,
+ "loss": 0.1505,
+ "step": 13170
+ },
+ {
+ "epoch": 35.88828337874659,
+ "grad_norm": 4.189176082611084,
+ "learning_rate": 1.4843545936309233e-05,
+ "loss": 0.1754,
+ "step": 13171
+ },
+ {
+ "epoch": 35.89100817438692,
+ "grad_norm": 4.68109130859375,
+ "learning_rate": 1.4842773847798703e-05,
+ "loss": 0.2172,
+ "step": 13172
+ },
+ {
+ "epoch": 35.89373297002725,
+ "grad_norm": 5.0576395988464355,
+ "learning_rate": 1.4842001721572804e-05,
+ "loss": 0.1496,
+ "step": 13173
+ },
+ {
+ "epoch": 35.89645776566758,
+ "grad_norm": 5.786940097808838,
+ "learning_rate": 1.4841229557637553e-05,
+ "loss": 0.2011,
+ "step": 13174
+ },
+ {
+ "epoch": 35.8991825613079,
+ "grad_norm": 4.689223289489746,
+ "learning_rate": 1.4840457355998963e-05,
+ "loss": 0.2174,
+ "step": 13175
+ },
+ {
+ "epoch": 35.90190735694823,
+ "grad_norm": 4.017246723175049,
+ "learning_rate": 1.4839685116663042e-05,
+ "loss": 0.1422,
+ "step": 13176
+ },
+ {
+ "epoch": 35.904632152588555,
+ "grad_norm": 4.083903789520264,
+ "learning_rate": 1.4838912839635813e-05,
+ "loss": 0.0967,
+ "step": 13177
+ },
+ {
+ "epoch": 35.90735694822888,
+ "grad_norm": 5.791813850402832,
+ "learning_rate": 1.4838140524923286e-05,
+ "loss": 0.2299,
+ "step": 13178
+ },
+ {
+ "epoch": 35.91008174386921,
+ "grad_norm": 4.093670845031738,
+ "learning_rate": 1.4837368172531477e-05,
+ "loss": 0.2625,
+ "step": 13179
+ },
+ {
+ "epoch": 35.91280653950954,
+ "grad_norm": 3.459120750427246,
+ "learning_rate": 1.4836595782466401e-05,
+ "loss": 0.3665,
+ "step": 13180
+ },
+ {
+ "epoch": 35.915531335149865,
+ "grad_norm": 4.203023910522461,
+ "learning_rate": 1.4835823354734072e-05,
+ "loss": 0.2075,
+ "step": 13181
+ },
+ {
+ "epoch": 35.91825613079019,
+ "grad_norm": 6.916604518890381,
+ "learning_rate": 1.4835050889340506e-05,
+ "loss": 0.2465,
+ "step": 13182
+ },
+ {
+ "epoch": 35.920980926430516,
+ "grad_norm": 4.89990234375,
+ "learning_rate": 1.4834278386291718e-05,
+ "loss": 0.2136,
+ "step": 13183
+ },
+ {
+ "epoch": 35.92370572207084,
+ "grad_norm": 4.021000385284424,
+ "learning_rate": 1.4833505845593725e-05,
+ "loss": 0.2523,
+ "step": 13184
+ },
+ {
+ "epoch": 35.926430517711175,
+ "grad_norm": 3.8949482440948486,
+ "learning_rate": 1.4832733267252545e-05,
+ "loss": 0.1515,
+ "step": 13185
+ },
+ {
+ "epoch": 35.9291553133515,
+ "grad_norm": 3.7019901275634766,
+ "learning_rate": 1.4831960651274196e-05,
+ "loss": 0.1437,
+ "step": 13186
+ },
+ {
+ "epoch": 35.93188010899183,
+ "grad_norm": 5.708944320678711,
+ "learning_rate": 1.4831187997664691e-05,
+ "loss": 0.286,
+ "step": 13187
+ },
+ {
+ "epoch": 35.93460490463215,
+ "grad_norm": 4.221436500549316,
+ "learning_rate": 1.4830415306430051e-05,
+ "loss": 0.4161,
+ "step": 13188
+ },
+ {
+ "epoch": 35.93732970027248,
+ "grad_norm": 3.99018931388855,
+ "learning_rate": 1.4829642577576289e-05,
+ "loss": 0.185,
+ "step": 13189
+ },
+ {
+ "epoch": 35.940054495912804,
+ "grad_norm": 4.3635358810424805,
+ "learning_rate": 1.482886981110943e-05,
+ "loss": 0.1872,
+ "step": 13190
+ },
+ {
+ "epoch": 35.94277929155314,
+ "grad_norm": 4.48692512512207,
+ "learning_rate": 1.4828097007035484e-05,
+ "loss": 0.2321,
+ "step": 13191
+ },
+ {
+ "epoch": 35.94550408719346,
+ "grad_norm": 4.343438148498535,
+ "learning_rate": 1.4827324165360478e-05,
+ "loss": 0.2025,
+ "step": 13192
+ },
+ {
+ "epoch": 35.94822888283379,
+ "grad_norm": 5.758538246154785,
+ "learning_rate": 1.4826551286090425e-05,
+ "loss": 0.3954,
+ "step": 13193
+ },
+ {
+ "epoch": 35.950953678474114,
+ "grad_norm": 4.520304203033447,
+ "learning_rate": 1.4825778369231344e-05,
+ "loss": 0.1425,
+ "step": 13194
+ },
+ {
+ "epoch": 35.95367847411444,
+ "grad_norm": 4.578720569610596,
+ "learning_rate": 1.482500541478926e-05,
+ "loss": 0.3491,
+ "step": 13195
+ },
+ {
+ "epoch": 35.956403269754766,
+ "grad_norm": 4.06895637512207,
+ "learning_rate": 1.4824232422770185e-05,
+ "loss": 0.1882,
+ "step": 13196
+ },
+ {
+ "epoch": 35.95912806539509,
+ "grad_norm": 5.634603500366211,
+ "learning_rate": 1.4823459393180147e-05,
+ "loss": 0.1929,
+ "step": 13197
+ },
+ {
+ "epoch": 35.961852861035425,
+ "grad_norm": 4.259729385375977,
+ "learning_rate": 1.482268632602516e-05,
+ "loss": 0.275,
+ "step": 13198
+ },
+ {
+ "epoch": 35.96457765667575,
+ "grad_norm": 4.946630954742432,
+ "learning_rate": 1.4821913221311249e-05,
+ "loss": 0.2165,
+ "step": 13199
+ },
+ {
+ "epoch": 35.967302452316076,
+ "grad_norm": 4.880048751831055,
+ "learning_rate": 1.482114007904443e-05,
+ "loss": 0.2752,
+ "step": 13200
+ },
+ {
+ "epoch": 35.9700272479564,
+ "grad_norm": 8.350459098815918,
+ "learning_rate": 1.4820366899230729e-05,
+ "loss": 0.2155,
+ "step": 13201
+ },
+ {
+ "epoch": 35.97275204359673,
+ "grad_norm": 4.589763641357422,
+ "learning_rate": 1.4819593681876168e-05,
+ "loss": 0.2822,
+ "step": 13202
+ },
+ {
+ "epoch": 35.97547683923706,
+ "grad_norm": 4.318881988525391,
+ "learning_rate": 1.4818820426986763e-05,
+ "loss": 0.1741,
+ "step": 13203
+ },
+ {
+ "epoch": 35.97820163487739,
+ "grad_norm": 6.090561866760254,
+ "learning_rate": 1.481804713456854e-05,
+ "loss": 0.2234,
+ "step": 13204
+ },
+ {
+ "epoch": 35.98092643051771,
+ "grad_norm": 4.346580982208252,
+ "learning_rate": 1.4817273804627526e-05,
+ "loss": 0.2621,
+ "step": 13205
+ },
+ {
+ "epoch": 35.98365122615804,
+ "grad_norm": 4.7284345626831055,
+ "learning_rate": 1.4816500437169735e-05,
+ "loss": 0.3008,
+ "step": 13206
+ },
+ {
+ "epoch": 35.986376021798364,
+ "grad_norm": 5.506720542907715,
+ "learning_rate": 1.4815727032201192e-05,
+ "loss": 0.2406,
+ "step": 13207
+ },
+ {
+ "epoch": 35.98910081743869,
+ "grad_norm": 3.944866895675659,
+ "learning_rate": 1.4814953589727922e-05,
+ "loss": 0.1563,
+ "step": 13208
+ },
+ {
+ "epoch": 35.991825613079016,
+ "grad_norm": 5.0538530349731445,
+ "learning_rate": 1.4814180109755954e-05,
+ "loss": 0.2029,
+ "step": 13209
+ },
+ {
+ "epoch": 35.99455040871935,
+ "grad_norm": 4.684808731079102,
+ "learning_rate": 1.4813406592291298e-05,
+ "loss": 0.3318,
+ "step": 13210
+ },
+ {
+ "epoch": 35.997275204359674,
+ "grad_norm": 5.834189414978027,
+ "learning_rate": 1.4812633037339992e-05,
+ "loss": 0.2874,
+ "step": 13211
+ },
+ {
+ "epoch": 36.0,
+ "grad_norm": 4.096088409423828,
+ "learning_rate": 1.4811859444908053e-05,
+ "loss": 0.404,
+ "step": 13212
+ },
+ {
+ "epoch": 36.002724795640326,
+ "grad_norm": 4.21230936050415,
+ "learning_rate": 1.4811085815001507e-05,
+ "loss": 0.2251,
+ "step": 13213
+ },
+ {
+ "epoch": 36.00544959128065,
+ "grad_norm": 4.194154739379883,
+ "learning_rate": 1.481031214762638e-05,
+ "loss": 0.152,
+ "step": 13214
+ },
+ {
+ "epoch": 36.00817438692098,
+ "grad_norm": 4.156232833862305,
+ "learning_rate": 1.4809538442788698e-05,
+ "loss": 0.2318,
+ "step": 13215
+ },
+ {
+ "epoch": 36.01089918256131,
+ "grad_norm": 4.39467191696167,
+ "learning_rate": 1.4808764700494482e-05,
+ "loss": 0.2006,
+ "step": 13216
+ },
+ {
+ "epoch": 36.013623978201636,
+ "grad_norm": 3.4029335975646973,
+ "learning_rate": 1.4807990920749766e-05,
+ "loss": 0.2644,
+ "step": 13217
+ },
+ {
+ "epoch": 36.01634877384196,
+ "grad_norm": 4.232716083526611,
+ "learning_rate": 1.4807217103560567e-05,
+ "loss": 0.155,
+ "step": 13218
+ },
+ {
+ "epoch": 36.01907356948229,
+ "grad_norm": 3.0255064964294434,
+ "learning_rate": 1.4806443248932917e-05,
+ "loss": 0.3707,
+ "step": 13219
+ },
+ {
+ "epoch": 36.02179836512261,
+ "grad_norm": 3.7099480628967285,
+ "learning_rate": 1.4805669356872843e-05,
+ "loss": 0.1627,
+ "step": 13220
+ },
+ {
+ "epoch": 36.02452316076294,
+ "grad_norm": 4.019195079803467,
+ "learning_rate": 1.4804895427386371e-05,
+ "loss": 0.2578,
+ "step": 13221
+ },
+ {
+ "epoch": 36.02724795640327,
+ "grad_norm": 3.4400153160095215,
+ "learning_rate": 1.4804121460479528e-05,
+ "loss": 0.1352,
+ "step": 13222
+ },
+ {
+ "epoch": 36.0299727520436,
+ "grad_norm": 4.095520496368408,
+ "learning_rate": 1.4803347456158341e-05,
+ "loss": 0.2595,
+ "step": 13223
+ },
+ {
+ "epoch": 36.032697547683924,
+ "grad_norm": 3.9667015075683594,
+ "learning_rate": 1.4802573414428837e-05,
+ "loss": 0.1458,
+ "step": 13224
+ },
+ {
+ "epoch": 36.03542234332425,
+ "grad_norm": 3.8175363540649414,
+ "learning_rate": 1.4801799335297052e-05,
+ "loss": 0.1067,
+ "step": 13225
+ },
+ {
+ "epoch": 36.038147138964575,
+ "grad_norm": 3.382685899734497,
+ "learning_rate": 1.4801025218769001e-05,
+ "loss": 0.1021,
+ "step": 13226
+ },
+ {
+ "epoch": 36.0408719346049,
+ "grad_norm": 3.4560036659240723,
+ "learning_rate": 1.4800251064850723e-05,
+ "loss": 0.2549,
+ "step": 13227
+ },
+ {
+ "epoch": 36.043596730245234,
+ "grad_norm": 4.0026631355285645,
+ "learning_rate": 1.4799476873548244e-05,
+ "loss": 0.1104,
+ "step": 13228
+ },
+ {
+ "epoch": 36.04632152588556,
+ "grad_norm": 3.9980978965759277,
+ "learning_rate": 1.4798702644867596e-05,
+ "loss": 0.2515,
+ "step": 13229
+ },
+ {
+ "epoch": 36.049046321525886,
+ "grad_norm": 3.9909591674804688,
+ "learning_rate": 1.47979283788148e-05,
+ "loss": 0.1315,
+ "step": 13230
+ },
+ {
+ "epoch": 36.05177111716621,
+ "grad_norm": 4.102788925170898,
+ "learning_rate": 1.4797154075395899e-05,
+ "loss": 0.1382,
+ "step": 13231
+ },
+ {
+ "epoch": 36.05449591280654,
+ "grad_norm": 3.9231300354003906,
+ "learning_rate": 1.4796379734616912e-05,
+ "loss": 0.1415,
+ "step": 13232
+ },
+ {
+ "epoch": 36.05722070844686,
+ "grad_norm": 3.441337823867798,
+ "learning_rate": 1.4795605356483877e-05,
+ "loss": 0.2435,
+ "step": 13233
+ },
+ {
+ "epoch": 36.059945504087196,
+ "grad_norm": 3.0760581493377686,
+ "learning_rate": 1.479483094100282e-05,
+ "loss": 0.3199,
+ "step": 13234
+ },
+ {
+ "epoch": 36.06267029972752,
+ "grad_norm": 4.13790225982666,
+ "learning_rate": 1.4794056488179776e-05,
+ "loss": 0.1324,
+ "step": 13235
+ },
+ {
+ "epoch": 36.06539509536785,
+ "grad_norm": 3.2327780723571777,
+ "learning_rate": 1.479328199802077e-05,
+ "loss": 0.0965,
+ "step": 13236
+ },
+ {
+ "epoch": 36.06811989100817,
+ "grad_norm": 3.95591139793396,
+ "learning_rate": 1.4792507470531842e-05,
+ "loss": 0.1599,
+ "step": 13237
+ },
+ {
+ "epoch": 36.0708446866485,
+ "grad_norm": 5.664831161499023,
+ "learning_rate": 1.4791732905719019e-05,
+ "loss": 0.1754,
+ "step": 13238
+ },
+ {
+ "epoch": 36.073569482288825,
+ "grad_norm": 4.515031337738037,
+ "learning_rate": 1.4790958303588334e-05,
+ "loss": 0.1394,
+ "step": 13239
+ },
+ {
+ "epoch": 36.07629427792916,
+ "grad_norm": 4.38517951965332,
+ "learning_rate": 1.4790183664145821e-05,
+ "loss": 0.1711,
+ "step": 13240
+ },
+ {
+ "epoch": 36.079019073569484,
+ "grad_norm": 4.302331924438477,
+ "learning_rate": 1.4789408987397512e-05,
+ "loss": 0.2568,
+ "step": 13241
+ },
+ {
+ "epoch": 36.08174386920981,
+ "grad_norm": 3.523406505584717,
+ "learning_rate": 1.4788634273349439e-05,
+ "loss": 0.1885,
+ "step": 13242
+ },
+ {
+ "epoch": 36.084468664850135,
+ "grad_norm": 3.4162240028381348,
+ "learning_rate": 1.4787859522007638e-05,
+ "loss": 0.1563,
+ "step": 13243
+ },
+ {
+ "epoch": 36.08719346049046,
+ "grad_norm": 8.63239574432373,
+ "learning_rate": 1.4787084733378138e-05,
+ "loss": 0.1318,
+ "step": 13244
+ },
+ {
+ "epoch": 36.08991825613079,
+ "grad_norm": 3.589057445526123,
+ "learning_rate": 1.4786309907466975e-05,
+ "loss": 0.1782,
+ "step": 13245
+ },
+ {
+ "epoch": 36.09264305177112,
+ "grad_norm": 3.012946367263794,
+ "learning_rate": 1.4785535044280189e-05,
+ "loss": 0.1454,
+ "step": 13246
+ },
+ {
+ "epoch": 36.095367847411445,
+ "grad_norm": 2.9011435508728027,
+ "learning_rate": 1.478476014382381e-05,
+ "loss": 0.1337,
+ "step": 13247
+ },
+ {
+ "epoch": 36.09809264305177,
+ "grad_norm": 3.3638598918914795,
+ "learning_rate": 1.4783985206103868e-05,
+ "loss": 0.2326,
+ "step": 13248
+ },
+ {
+ "epoch": 36.1008174386921,
+ "grad_norm": 4.14896297454834,
+ "learning_rate": 1.4783210231126411e-05,
+ "loss": 0.1496,
+ "step": 13249
+ },
+ {
+ "epoch": 36.10354223433242,
+ "grad_norm": 3.273402452468872,
+ "learning_rate": 1.4782435218897459e-05,
+ "loss": 0.1448,
+ "step": 13250
+ },
+ {
+ "epoch": 36.10626702997275,
+ "grad_norm": 3.681029796600342,
+ "learning_rate": 1.478166016942306e-05,
+ "loss": 0.0961,
+ "step": 13251
+ },
+ {
+ "epoch": 36.10899182561308,
+ "grad_norm": 4.5384345054626465,
+ "learning_rate": 1.4780885082709242e-05,
+ "loss": 0.2035,
+ "step": 13252
+ },
+ {
+ "epoch": 36.11171662125341,
+ "grad_norm": 3.7394917011260986,
+ "learning_rate": 1.4780109958762049e-05,
+ "loss": 0.2035,
+ "step": 13253
+ },
+ {
+ "epoch": 36.11444141689373,
+ "grad_norm": 4.09307861328125,
+ "learning_rate": 1.4779334797587508e-05,
+ "loss": 0.1821,
+ "step": 13254
+ },
+ {
+ "epoch": 36.11716621253406,
+ "grad_norm": 3.8676226139068604,
+ "learning_rate": 1.4778559599191668e-05,
+ "loss": 0.125,
+ "step": 13255
+ },
+ {
+ "epoch": 36.119891008174385,
+ "grad_norm": 4.3630571365356445,
+ "learning_rate": 1.4777784363580553e-05,
+ "loss": 0.2372,
+ "step": 13256
+ },
+ {
+ "epoch": 36.12261580381471,
+ "grad_norm": 3.657651662826538,
+ "learning_rate": 1.4777009090760213e-05,
+ "loss": 0.1713,
+ "step": 13257
+ },
+ {
+ "epoch": 36.12534059945504,
+ "grad_norm": 3.490802526473999,
+ "learning_rate": 1.4776233780736677e-05,
+ "loss": 0.0937,
+ "step": 13258
+ },
+ {
+ "epoch": 36.12806539509537,
+ "grad_norm": 4.948929309844971,
+ "learning_rate": 1.4775458433515988e-05,
+ "loss": 0.1409,
+ "step": 13259
+ },
+ {
+ "epoch": 36.130790190735695,
+ "grad_norm": 4.975615501403809,
+ "learning_rate": 1.477468304910418e-05,
+ "loss": 0.1326,
+ "step": 13260
+ },
+ {
+ "epoch": 36.13351498637602,
+ "grad_norm": 3.330186605453491,
+ "learning_rate": 1.4773907627507296e-05,
+ "loss": 0.1422,
+ "step": 13261
+ },
+ {
+ "epoch": 36.13623978201635,
+ "grad_norm": 4.032511234283447,
+ "learning_rate": 1.4773132168731372e-05,
+ "loss": 0.118,
+ "step": 13262
+ },
+ {
+ "epoch": 36.13896457765667,
+ "grad_norm": 3.5689282417297363,
+ "learning_rate": 1.4772356672782448e-05,
+ "loss": 0.3096,
+ "step": 13263
+ },
+ {
+ "epoch": 36.141689373297005,
+ "grad_norm": 3.9850289821624756,
+ "learning_rate": 1.4771581139666564e-05,
+ "loss": 0.1754,
+ "step": 13264
+ },
+ {
+ "epoch": 36.14441416893733,
+ "grad_norm": 3.895552158355713,
+ "learning_rate": 1.4770805569389762e-05,
+ "loss": 0.1514,
+ "step": 13265
+ },
+ {
+ "epoch": 36.14713896457766,
+ "grad_norm": 4.133829116821289,
+ "learning_rate": 1.4770029961958075e-05,
+ "loss": 0.1958,
+ "step": 13266
+ },
+ {
+ "epoch": 36.14986376021798,
+ "grad_norm": 3.587510585784912,
+ "learning_rate": 1.4769254317377553e-05,
+ "loss": 0.1075,
+ "step": 13267
+ },
+ {
+ "epoch": 36.15258855585831,
+ "grad_norm": 4.030117511749268,
+ "learning_rate": 1.476847863565423e-05,
+ "loss": 0.2545,
+ "step": 13268
+ },
+ {
+ "epoch": 36.155313351498634,
+ "grad_norm": 4.739619731903076,
+ "learning_rate": 1.4767702916794149e-05,
+ "loss": 0.2273,
+ "step": 13269
+ },
+ {
+ "epoch": 36.15803814713897,
+ "grad_norm": 3.889606475830078,
+ "learning_rate": 1.476692716080335e-05,
+ "loss": 0.1911,
+ "step": 13270
+ },
+ {
+ "epoch": 36.16076294277929,
+ "grad_norm": 4.110728740692139,
+ "learning_rate": 1.4766151367687875e-05,
+ "loss": 0.1623,
+ "step": 13271
+ },
+ {
+ "epoch": 36.16348773841962,
+ "grad_norm": 3.324054718017578,
+ "learning_rate": 1.476537553745377e-05,
+ "loss": 0.4138,
+ "step": 13272
+ },
+ {
+ "epoch": 36.166212534059945,
+ "grad_norm": 4.502650737762451,
+ "learning_rate": 1.476459967010707e-05,
+ "loss": 0.1465,
+ "step": 13273
+ },
+ {
+ "epoch": 36.16893732970027,
+ "grad_norm": 3.8765835762023926,
+ "learning_rate": 1.4763823765653824e-05,
+ "loss": 0.1773,
+ "step": 13274
+ },
+ {
+ "epoch": 36.171662125340596,
+ "grad_norm": 3.7640018463134766,
+ "learning_rate": 1.476304782410007e-05,
+ "loss": 0.2011,
+ "step": 13275
+ },
+ {
+ "epoch": 36.17438692098093,
+ "grad_norm": 4.685653209686279,
+ "learning_rate": 1.4762271845451852e-05,
+ "loss": 0.1699,
+ "step": 13276
+ },
+ {
+ "epoch": 36.177111716621255,
+ "grad_norm": 3.5704195499420166,
+ "learning_rate": 1.4761495829715219e-05,
+ "loss": 0.2646,
+ "step": 13277
+ },
+ {
+ "epoch": 36.17983651226158,
+ "grad_norm": 3.8592772483825684,
+ "learning_rate": 1.4760719776896205e-05,
+ "loss": 0.2169,
+ "step": 13278
+ },
+ {
+ "epoch": 36.182561307901906,
+ "grad_norm": 3.268378496170044,
+ "learning_rate": 1.4759943687000858e-05,
+ "loss": 0.2969,
+ "step": 13279
+ },
+ {
+ "epoch": 36.18528610354223,
+ "grad_norm": 3.731829881668091,
+ "learning_rate": 1.4759167560035226e-05,
+ "loss": 0.1246,
+ "step": 13280
+ },
+ {
+ "epoch": 36.18801089918256,
+ "grad_norm": 4.124495983123779,
+ "learning_rate": 1.475839139600535e-05,
+ "loss": 0.2717,
+ "step": 13281
+ },
+ {
+ "epoch": 36.19073569482289,
+ "grad_norm": 3.3732287883758545,
+ "learning_rate": 1.4757615194917274e-05,
+ "loss": 0.184,
+ "step": 13282
+ },
+ {
+ "epoch": 36.19346049046322,
+ "grad_norm": 3.179875612258911,
+ "learning_rate": 1.4756838956777043e-05,
+ "loss": 0.1058,
+ "step": 13283
+ },
+ {
+ "epoch": 36.19618528610354,
+ "grad_norm": 5.262901306152344,
+ "learning_rate": 1.4756062681590706e-05,
+ "loss": 0.17,
+ "step": 13284
+ },
+ {
+ "epoch": 36.19891008174387,
+ "grad_norm": 3.8430941104888916,
+ "learning_rate": 1.4755286369364301e-05,
+ "loss": 0.2426,
+ "step": 13285
+ },
+ {
+ "epoch": 36.201634877384194,
+ "grad_norm": 4.139091968536377,
+ "learning_rate": 1.4754510020103883e-05,
+ "loss": 0.1778,
+ "step": 13286
+ },
+ {
+ "epoch": 36.20435967302452,
+ "grad_norm": 4.606226921081543,
+ "learning_rate": 1.4753733633815492e-05,
+ "loss": 0.1281,
+ "step": 13287
+ },
+ {
+ "epoch": 36.20708446866485,
+ "grad_norm": 3.6965019702911377,
+ "learning_rate": 1.4752957210505179e-05,
+ "loss": 0.0901,
+ "step": 13288
+ },
+ {
+ "epoch": 36.20980926430518,
+ "grad_norm": 3.6713852882385254,
+ "learning_rate": 1.4752180750178986e-05,
+ "loss": 0.1478,
+ "step": 13289
+ },
+ {
+ "epoch": 36.212534059945504,
+ "grad_norm": 3.5745744705200195,
+ "learning_rate": 1.4751404252842962e-05,
+ "loss": 0.1975,
+ "step": 13290
+ },
+ {
+ "epoch": 36.21525885558583,
+ "grad_norm": 3.807485580444336,
+ "learning_rate": 1.4750627718503156e-05,
+ "loss": 0.1805,
+ "step": 13291
+ },
+ {
+ "epoch": 36.217983651226156,
+ "grad_norm": 3.525071859359741,
+ "learning_rate": 1.4749851147165615e-05,
+ "loss": 0.2269,
+ "step": 13292
+ },
+ {
+ "epoch": 36.22070844686648,
+ "grad_norm": 4.04987096786499,
+ "learning_rate": 1.4749074538836384e-05,
+ "loss": 0.1631,
+ "step": 13293
+ },
+ {
+ "epoch": 36.223433242506815,
+ "grad_norm": 3.0343987941741943,
+ "learning_rate": 1.4748297893521515e-05,
+ "loss": 0.1569,
+ "step": 13294
+ },
+ {
+ "epoch": 36.22615803814714,
+ "grad_norm": 3.6291961669921875,
+ "learning_rate": 1.4747521211227054e-05,
+ "loss": 0.1768,
+ "step": 13295
+ },
+ {
+ "epoch": 36.228882833787466,
+ "grad_norm": 5.501640319824219,
+ "learning_rate": 1.4746744491959049e-05,
+ "loss": 0.2201,
+ "step": 13296
+ },
+ {
+ "epoch": 36.23160762942779,
+ "grad_norm": 3.980851173400879,
+ "learning_rate": 1.4745967735723552e-05,
+ "loss": 0.1795,
+ "step": 13297
+ },
+ {
+ "epoch": 36.23433242506812,
+ "grad_norm": 21.61891746520996,
+ "learning_rate": 1.4745190942526613e-05,
+ "loss": 0.1841,
+ "step": 13298
+ },
+ {
+ "epoch": 36.237057220708444,
+ "grad_norm": 3.1857070922851562,
+ "learning_rate": 1.4744414112374274e-05,
+ "loss": 0.1873,
+ "step": 13299
+ },
+ {
+ "epoch": 36.23978201634878,
+ "grad_norm": 6.417799472808838,
+ "learning_rate": 1.4743637245272598e-05,
+ "loss": 0.1888,
+ "step": 13300
+ },
+ {
+ "epoch": 36.2425068119891,
+ "grad_norm": 3.586683511734009,
+ "learning_rate": 1.4742860341227622e-05,
+ "loss": 0.1033,
+ "step": 13301
+ },
+ {
+ "epoch": 36.24523160762943,
+ "grad_norm": 3.476329803466797,
+ "learning_rate": 1.4742083400245406e-05,
+ "loss": 0.2597,
+ "step": 13302
+ },
+ {
+ "epoch": 36.247956403269754,
+ "grad_norm": 8.77595329284668,
+ "learning_rate": 1.4741306422331995e-05,
+ "loss": 0.1474,
+ "step": 13303
+ },
+ {
+ "epoch": 36.25068119891008,
+ "grad_norm": 3.690427303314209,
+ "learning_rate": 1.4740529407493444e-05,
+ "loss": 0.1428,
+ "step": 13304
+ },
+ {
+ "epoch": 36.253405994550405,
+ "grad_norm": 4.290805816650391,
+ "learning_rate": 1.4739752355735799e-05,
+ "loss": 0.2195,
+ "step": 13305
+ },
+ {
+ "epoch": 36.25613079019074,
+ "grad_norm": 3.914966344833374,
+ "learning_rate": 1.4738975267065118e-05,
+ "loss": 0.1913,
+ "step": 13306
+ },
+ {
+ "epoch": 36.258855585831064,
+ "grad_norm": 4.287760257720947,
+ "learning_rate": 1.473819814148745e-05,
+ "loss": 0.1471,
+ "step": 13307
+ },
+ {
+ "epoch": 36.26158038147139,
+ "grad_norm": 4.433450222015381,
+ "learning_rate": 1.473742097900885e-05,
+ "loss": 0.1621,
+ "step": 13308
+ },
+ {
+ "epoch": 36.264305177111716,
+ "grad_norm": 3.755520820617676,
+ "learning_rate": 1.4736643779635362e-05,
+ "loss": 0.1387,
+ "step": 13309
+ },
+ {
+ "epoch": 36.26702997275204,
+ "grad_norm": 5.102533340454102,
+ "learning_rate": 1.473586654337305e-05,
+ "loss": 0.2217,
+ "step": 13310
+ },
+ {
+ "epoch": 36.26975476839237,
+ "grad_norm": 4.227988243103027,
+ "learning_rate": 1.4735089270227958e-05,
+ "loss": 0.1976,
+ "step": 13311
+ },
+ {
+ "epoch": 36.2724795640327,
+ "grad_norm": 4.778122901916504,
+ "learning_rate": 1.4734311960206147e-05,
+ "loss": 0.2735,
+ "step": 13312
+ },
+ {
+ "epoch": 36.275204359673026,
+ "grad_norm": 3.8774847984313965,
+ "learning_rate": 1.473353461331366e-05,
+ "loss": 0.2245,
+ "step": 13313
+ },
+ {
+ "epoch": 36.27792915531335,
+ "grad_norm": 3.278872013092041,
+ "learning_rate": 1.4732757229556565e-05,
+ "loss": 0.1275,
+ "step": 13314
+ },
+ {
+ "epoch": 36.28065395095368,
+ "grad_norm": 3.930098295211792,
+ "learning_rate": 1.4731979808940906e-05,
+ "loss": 0.1877,
+ "step": 13315
+ },
+ {
+ "epoch": 36.283378746594,
+ "grad_norm": 4.465882301330566,
+ "learning_rate": 1.4731202351472741e-05,
+ "loss": 0.228,
+ "step": 13316
+ },
+ {
+ "epoch": 36.28610354223433,
+ "grad_norm": 4.767672538757324,
+ "learning_rate": 1.4730424857158123e-05,
+ "loss": 0.1614,
+ "step": 13317
+ },
+ {
+ "epoch": 36.28882833787466,
+ "grad_norm": 3.6608688831329346,
+ "learning_rate": 1.472964732600311e-05,
+ "loss": 0.1638,
+ "step": 13318
+ },
+ {
+ "epoch": 36.29155313351499,
+ "grad_norm": 4.27473783493042,
+ "learning_rate": 1.4728869758013758e-05,
+ "loss": 0.1923,
+ "step": 13319
+ },
+ {
+ "epoch": 36.294277929155314,
+ "grad_norm": 3.23496413230896,
+ "learning_rate": 1.4728092153196119e-05,
+ "loss": 0.1179,
+ "step": 13320
+ },
+ {
+ "epoch": 36.29700272479564,
+ "grad_norm": 3.953462600708008,
+ "learning_rate": 1.4727314511556248e-05,
+ "loss": 0.3975,
+ "step": 13321
+ },
+ {
+ "epoch": 36.299727520435965,
+ "grad_norm": 4.004514217376709,
+ "learning_rate": 1.4726536833100204e-05,
+ "loss": 0.2273,
+ "step": 13322
+ },
+ {
+ "epoch": 36.30245231607629,
+ "grad_norm": 4.7965850830078125,
+ "learning_rate": 1.4725759117834045e-05,
+ "loss": 0.1156,
+ "step": 13323
+ },
+ {
+ "epoch": 36.305177111716624,
+ "grad_norm": 3.986513376235962,
+ "learning_rate": 1.4724981365763826e-05,
+ "loss": 0.1906,
+ "step": 13324
+ },
+ {
+ "epoch": 36.30790190735695,
+ "grad_norm": 4.581685543060303,
+ "learning_rate": 1.4724203576895604e-05,
+ "loss": 0.1578,
+ "step": 13325
+ },
+ {
+ "epoch": 36.310626702997276,
+ "grad_norm": 4.146121025085449,
+ "learning_rate": 1.4723425751235436e-05,
+ "loss": 0.1663,
+ "step": 13326
+ },
+ {
+ "epoch": 36.3133514986376,
+ "grad_norm": 3.8283894062042236,
+ "learning_rate": 1.4722647888789382e-05,
+ "loss": 0.1357,
+ "step": 13327
+ },
+ {
+ "epoch": 36.31607629427793,
+ "grad_norm": 3.4876811504364014,
+ "learning_rate": 1.4721869989563495e-05,
+ "loss": 0.0965,
+ "step": 13328
+ },
+ {
+ "epoch": 36.31880108991825,
+ "grad_norm": 3.9417965412139893,
+ "learning_rate": 1.472109205356384e-05,
+ "loss": 0.1698,
+ "step": 13329
+ },
+ {
+ "epoch": 36.321525885558586,
+ "grad_norm": 3.705703020095825,
+ "learning_rate": 1.4720314080796469e-05,
+ "loss": 0.3075,
+ "step": 13330
+ },
+ {
+ "epoch": 36.32425068119891,
+ "grad_norm": 3.863615036010742,
+ "learning_rate": 1.4719536071267443e-05,
+ "loss": 0.1092,
+ "step": 13331
+ },
+ {
+ "epoch": 36.32697547683924,
+ "grad_norm": 4.152604579925537,
+ "learning_rate": 1.4718758024982827e-05,
+ "loss": 0.1353,
+ "step": 13332
+ },
+ {
+ "epoch": 36.32970027247956,
+ "grad_norm": 4.855541706085205,
+ "learning_rate": 1.4717979941948672e-05,
+ "loss": 0.1649,
+ "step": 13333
+ },
+ {
+ "epoch": 36.33242506811989,
+ "grad_norm": 4.244178771972656,
+ "learning_rate": 1.4717201822171043e-05,
+ "loss": 0.1951,
+ "step": 13334
+ },
+ {
+ "epoch": 36.335149863760215,
+ "grad_norm": 4.450632572174072,
+ "learning_rate": 1.4716423665655995e-05,
+ "loss": 0.1436,
+ "step": 13335
+ },
+ {
+ "epoch": 36.33787465940055,
+ "grad_norm": 3.718182325363159,
+ "learning_rate": 1.4715645472409594e-05,
+ "loss": 0.3408,
+ "step": 13336
+ },
+ {
+ "epoch": 36.34059945504087,
+ "grad_norm": 3.7080881595611572,
+ "learning_rate": 1.4714867242437896e-05,
+ "loss": 0.225,
+ "step": 13337
+ },
+ {
+ "epoch": 36.3433242506812,
+ "grad_norm": 4.332475662231445,
+ "learning_rate": 1.4714088975746967e-05,
+ "loss": 0.1246,
+ "step": 13338
+ },
+ {
+ "epoch": 36.346049046321525,
+ "grad_norm": 3.716726779937744,
+ "learning_rate": 1.471331067234286e-05,
+ "loss": 0.1576,
+ "step": 13339
+ },
+ {
+ "epoch": 36.34877384196185,
+ "grad_norm": 3.736865520477295,
+ "learning_rate": 1.4712532332231644e-05,
+ "loss": 0.1986,
+ "step": 13340
+ },
+ {
+ "epoch": 36.35149863760218,
+ "grad_norm": 3.717778205871582,
+ "learning_rate": 1.4711753955419376e-05,
+ "loss": 0.2131,
+ "step": 13341
+ },
+ {
+ "epoch": 36.35422343324251,
+ "grad_norm": 3.681694269180298,
+ "learning_rate": 1.4710975541912122e-05,
+ "loss": 0.1324,
+ "step": 13342
+ },
+ {
+ "epoch": 36.356948228882835,
+ "grad_norm": 3.6900510787963867,
+ "learning_rate": 1.471019709171594e-05,
+ "loss": 0.1608,
+ "step": 13343
+ },
+ {
+ "epoch": 36.35967302452316,
+ "grad_norm": 3.8685688972473145,
+ "learning_rate": 1.4709418604836898e-05,
+ "loss": 0.2571,
+ "step": 13344
+ },
+ {
+ "epoch": 36.36239782016349,
+ "grad_norm": 4.513291358947754,
+ "learning_rate": 1.4708640081281054e-05,
+ "loss": 0.1866,
+ "step": 13345
+ },
+ {
+ "epoch": 36.36512261580381,
+ "grad_norm": 3.9801948070526123,
+ "learning_rate": 1.4707861521054471e-05,
+ "loss": 0.0994,
+ "step": 13346
+ },
+ {
+ "epoch": 36.36784741144414,
+ "grad_norm": 4.186225414276123,
+ "learning_rate": 1.4707082924163216e-05,
+ "loss": 0.1832,
+ "step": 13347
+ },
+ {
+ "epoch": 36.37057220708447,
+ "grad_norm": 3.4475250244140625,
+ "learning_rate": 1.470630429061335e-05,
+ "loss": 0.138,
+ "step": 13348
+ },
+ {
+ "epoch": 36.3732970027248,
+ "grad_norm": 4.360630512237549,
+ "learning_rate": 1.4705525620410936e-05,
+ "loss": 0.157,
+ "step": 13349
+ },
+ {
+ "epoch": 36.37602179836512,
+ "grad_norm": 5.043584823608398,
+ "learning_rate": 1.4704746913562043e-05,
+ "loss": 0.208,
+ "step": 13350
+ },
+ {
+ "epoch": 36.37874659400545,
+ "grad_norm": 3.9643495082855225,
+ "learning_rate": 1.470396817007273e-05,
+ "loss": 0.1695,
+ "step": 13351
+ },
+ {
+ "epoch": 36.381471389645775,
+ "grad_norm": 3.089604377746582,
+ "learning_rate": 1.4703189389949066e-05,
+ "loss": 0.1065,
+ "step": 13352
+ },
+ {
+ "epoch": 36.3841961852861,
+ "grad_norm": 4.854663372039795,
+ "learning_rate": 1.4702410573197112e-05,
+ "loss": 0.198,
+ "step": 13353
+ },
+ {
+ "epoch": 36.38692098092643,
+ "grad_norm": 3.710354804992676,
+ "learning_rate": 1.4701631719822937e-05,
+ "loss": 0.2662,
+ "step": 13354
+ },
+ {
+ "epoch": 36.38964577656676,
+ "grad_norm": 4.002683639526367,
+ "learning_rate": 1.4700852829832605e-05,
+ "loss": 0.1231,
+ "step": 13355
+ },
+ {
+ "epoch": 36.392370572207085,
+ "grad_norm": 3.060931444168091,
+ "learning_rate": 1.4700073903232184e-05,
+ "loss": 0.1395,
+ "step": 13356
+ },
+ {
+ "epoch": 36.39509536784741,
+ "grad_norm": 3.4523203372955322,
+ "learning_rate": 1.469929494002774e-05,
+ "loss": 0.2213,
+ "step": 13357
+ },
+ {
+ "epoch": 36.39782016348774,
+ "grad_norm": 3.76961350440979,
+ "learning_rate": 1.4698515940225333e-05,
+ "loss": 0.1687,
+ "step": 13358
+ },
+ {
+ "epoch": 36.40054495912806,
+ "grad_norm": 4.353285312652588,
+ "learning_rate": 1.4697736903831041e-05,
+ "loss": 0.1317,
+ "step": 13359
+ },
+ {
+ "epoch": 36.403269754768395,
+ "grad_norm": 4.419847011566162,
+ "learning_rate": 1.4696957830850922e-05,
+ "loss": 0.2736,
+ "step": 13360
+ },
+ {
+ "epoch": 36.40599455040872,
+ "grad_norm": 4.326084136962891,
+ "learning_rate": 1.4696178721291048e-05,
+ "loss": 0.201,
+ "step": 13361
+ },
+ {
+ "epoch": 36.40871934604905,
+ "grad_norm": 3.967036485671997,
+ "learning_rate": 1.4695399575157483e-05,
+ "loss": 0.209,
+ "step": 13362
+ },
+ {
+ "epoch": 36.41144414168937,
+ "grad_norm": 3.385560989379883,
+ "learning_rate": 1.4694620392456299e-05,
+ "loss": 0.1077,
+ "step": 13363
+ },
+ {
+ "epoch": 36.4141689373297,
+ "grad_norm": 3.3493824005126953,
+ "learning_rate": 1.4693841173193562e-05,
+ "loss": 0.1292,
+ "step": 13364
+ },
+ {
+ "epoch": 36.416893732970024,
+ "grad_norm": 2.722726821899414,
+ "learning_rate": 1.469306191737534e-05,
+ "loss": 0.1452,
+ "step": 13365
+ },
+ {
+ "epoch": 36.41961852861036,
+ "grad_norm": 4.269332408905029,
+ "learning_rate": 1.4692282625007703e-05,
+ "loss": 0.1996,
+ "step": 13366
+ },
+ {
+ "epoch": 36.42234332425068,
+ "grad_norm": 4.754920959472656,
+ "learning_rate": 1.4691503296096721e-05,
+ "loss": 0.2089,
+ "step": 13367
+ },
+ {
+ "epoch": 36.42506811989101,
+ "grad_norm": 4.543644428253174,
+ "learning_rate": 1.4690723930648463e-05,
+ "loss": 0.1232,
+ "step": 13368
+ },
+ {
+ "epoch": 36.427792915531334,
+ "grad_norm": 4.042744159698486,
+ "learning_rate": 1.4689944528668997e-05,
+ "loss": 0.1249,
+ "step": 13369
+ },
+ {
+ "epoch": 36.43051771117166,
+ "grad_norm": 3.251384735107422,
+ "learning_rate": 1.4689165090164395e-05,
+ "loss": 0.1387,
+ "step": 13370
+ },
+ {
+ "epoch": 36.433242506811986,
+ "grad_norm": 4.840087890625,
+ "learning_rate": 1.4688385615140726e-05,
+ "loss": 0.1461,
+ "step": 13371
+ },
+ {
+ "epoch": 36.43596730245232,
+ "grad_norm": 3.4843881130218506,
+ "learning_rate": 1.4687606103604057e-05,
+ "loss": 0.1862,
+ "step": 13372
+ },
+ {
+ "epoch": 36.438692098092645,
+ "grad_norm": 3.512843608856201,
+ "learning_rate": 1.4686826555560468e-05,
+ "loss": 0.3033,
+ "step": 13373
+ },
+ {
+ "epoch": 36.44141689373297,
+ "grad_norm": 4.1708149909973145,
+ "learning_rate": 1.468604697101602e-05,
+ "loss": 0.1341,
+ "step": 13374
+ },
+ {
+ "epoch": 36.444141689373296,
+ "grad_norm": 3.8743348121643066,
+ "learning_rate": 1.468526734997679e-05,
+ "loss": 0.1181,
+ "step": 13375
+ },
+ {
+ "epoch": 36.44686648501362,
+ "grad_norm": 4.368322849273682,
+ "learning_rate": 1.468448769244885e-05,
+ "loss": 0.1237,
+ "step": 13376
+ },
+ {
+ "epoch": 36.44959128065395,
+ "grad_norm": 2.778435230255127,
+ "learning_rate": 1.4683707998438273e-05,
+ "loss": 0.2125,
+ "step": 13377
+ },
+ {
+ "epoch": 36.45231607629428,
+ "grad_norm": 3.5088212490081787,
+ "learning_rate": 1.4682928267951124e-05,
+ "loss": 0.1445,
+ "step": 13378
+ },
+ {
+ "epoch": 36.45504087193461,
+ "grad_norm": 3.834777355194092,
+ "learning_rate": 1.4682148500993486e-05,
+ "loss": 0.2702,
+ "step": 13379
+ },
+ {
+ "epoch": 36.45776566757493,
+ "grad_norm": 3.1854004859924316,
+ "learning_rate": 1.4681368697571421e-05,
+ "loss": 0.1339,
+ "step": 13380
+ },
+ {
+ "epoch": 36.46049046321526,
+ "grad_norm": 3.68837833404541,
+ "learning_rate": 1.4680588857691009e-05,
+ "loss": 0.2196,
+ "step": 13381
+ },
+ {
+ "epoch": 36.463215258855584,
+ "grad_norm": 3.3353567123413086,
+ "learning_rate": 1.4679808981358319e-05,
+ "loss": 0.0889,
+ "step": 13382
+ },
+ {
+ "epoch": 36.46594005449591,
+ "grad_norm": 3.6722841262817383,
+ "learning_rate": 1.4679029068579432e-05,
+ "loss": 0.2083,
+ "step": 13383
+ },
+ {
+ "epoch": 36.46866485013624,
+ "grad_norm": 4.5272064208984375,
+ "learning_rate": 1.4678249119360415e-05,
+ "loss": 0.2406,
+ "step": 13384
+ },
+ {
+ "epoch": 36.47138964577657,
+ "grad_norm": 3.8546009063720703,
+ "learning_rate": 1.4677469133707346e-05,
+ "loss": 0.2501,
+ "step": 13385
+ },
+ {
+ "epoch": 36.474114441416894,
+ "grad_norm": 3.446580648422241,
+ "learning_rate": 1.4676689111626294e-05,
+ "loss": 0.1844,
+ "step": 13386
+ },
+ {
+ "epoch": 36.47683923705722,
+ "grad_norm": 2.9624640941619873,
+ "learning_rate": 1.4675909053123342e-05,
+ "loss": 0.1363,
+ "step": 13387
+ },
+ {
+ "epoch": 36.479564032697546,
+ "grad_norm": 3.707512855529785,
+ "learning_rate": 1.4675128958204559e-05,
+ "loss": 0.1516,
+ "step": 13388
+ },
+ {
+ "epoch": 36.48228882833787,
+ "grad_norm": 4.680509090423584,
+ "learning_rate": 1.4674348826876024e-05,
+ "loss": 0.3815,
+ "step": 13389
+ },
+ {
+ "epoch": 36.485013623978205,
+ "grad_norm": 4.833968639373779,
+ "learning_rate": 1.4673568659143806e-05,
+ "loss": 0.197,
+ "step": 13390
+ },
+ {
+ "epoch": 36.48773841961853,
+ "grad_norm": 3.4053003787994385,
+ "learning_rate": 1.4672788455013988e-05,
+ "loss": 0.1563,
+ "step": 13391
+ },
+ {
+ "epoch": 36.490463215258856,
+ "grad_norm": 5.064488410949707,
+ "learning_rate": 1.4672008214492643e-05,
+ "loss": 0.2198,
+ "step": 13392
+ },
+ {
+ "epoch": 36.49318801089918,
+ "grad_norm": 3.743194818496704,
+ "learning_rate": 1.4671227937585853e-05,
+ "loss": 0.1803,
+ "step": 13393
+ },
+ {
+ "epoch": 36.49591280653951,
+ "grad_norm": 6.897176265716553,
+ "learning_rate": 1.4670447624299685e-05,
+ "loss": 0.1118,
+ "step": 13394
+ },
+ {
+ "epoch": 36.49863760217983,
+ "grad_norm": 2.5453877449035645,
+ "learning_rate": 1.4669667274640227e-05,
+ "loss": 0.1246,
+ "step": 13395
+ },
+ {
+ "epoch": 36.50136239782017,
+ "grad_norm": 5.025479793548584,
+ "learning_rate": 1.4668886888613545e-05,
+ "loss": 0.4499,
+ "step": 13396
+ },
+ {
+ "epoch": 36.50408719346049,
+ "grad_norm": 3.8040077686309814,
+ "learning_rate": 1.4668106466225724e-05,
+ "loss": 0.2349,
+ "step": 13397
+ },
+ {
+ "epoch": 36.50681198910082,
+ "grad_norm": 3.8106095790863037,
+ "learning_rate": 1.466732600748284e-05,
+ "loss": 0.1438,
+ "step": 13398
+ },
+ {
+ "epoch": 36.509536784741144,
+ "grad_norm": 3.945317029953003,
+ "learning_rate": 1.4666545512390971e-05,
+ "loss": 0.1835,
+ "step": 13399
+ },
+ {
+ "epoch": 36.51226158038147,
+ "grad_norm": 3.0905673503875732,
+ "learning_rate": 1.4665764980956196e-05,
+ "loss": 0.3525,
+ "step": 13400
+ },
+ {
+ "epoch": 36.514986376021795,
+ "grad_norm": 3.479957103729248,
+ "learning_rate": 1.4664984413184595e-05,
+ "loss": 0.1444,
+ "step": 13401
+ },
+ {
+ "epoch": 36.51771117166213,
+ "grad_norm": 4.514152526855469,
+ "learning_rate": 1.4664203809082242e-05,
+ "loss": 0.2497,
+ "step": 13402
+ },
+ {
+ "epoch": 36.520435967302454,
+ "grad_norm": 3.4301702976226807,
+ "learning_rate": 1.4663423168655224e-05,
+ "loss": 0.2262,
+ "step": 13403
+ },
+ {
+ "epoch": 36.52316076294278,
+ "grad_norm": 4.100943565368652,
+ "learning_rate": 1.4662642491909613e-05,
+ "loss": 0.143,
+ "step": 13404
+ },
+ {
+ "epoch": 36.525885558583106,
+ "grad_norm": 4.541743755340576,
+ "learning_rate": 1.4661861778851495e-05,
+ "loss": 0.2516,
+ "step": 13405
+ },
+ {
+ "epoch": 36.52861035422343,
+ "grad_norm": 3.199359655380249,
+ "learning_rate": 1.4661081029486947e-05,
+ "loss": 0.1646,
+ "step": 13406
+ },
+ {
+ "epoch": 36.53133514986376,
+ "grad_norm": 4.365713596343994,
+ "learning_rate": 1.466030024382205e-05,
+ "loss": 0.172,
+ "step": 13407
+ },
+ {
+ "epoch": 36.53405994550409,
+ "grad_norm": 4.049338340759277,
+ "learning_rate": 1.4659519421862882e-05,
+ "loss": 0.1629,
+ "step": 13408
+ },
+ {
+ "epoch": 36.536784741144416,
+ "grad_norm": 4.443026065826416,
+ "learning_rate": 1.4658738563615528e-05,
+ "loss": 0.2725,
+ "step": 13409
+ },
+ {
+ "epoch": 36.53950953678474,
+ "grad_norm": 4.234590530395508,
+ "learning_rate": 1.4657957669086068e-05,
+ "loss": 0.1356,
+ "step": 13410
+ },
+ {
+ "epoch": 36.54223433242507,
+ "grad_norm": 4.504232883453369,
+ "learning_rate": 1.4657176738280587e-05,
+ "loss": 0.347,
+ "step": 13411
+ },
+ {
+ "epoch": 36.54495912806539,
+ "grad_norm": 4.03237247467041,
+ "learning_rate": 1.4656395771205159e-05,
+ "loss": 0.1369,
+ "step": 13412
+ },
+ {
+ "epoch": 36.54768392370572,
+ "grad_norm": 4.532645225524902,
+ "learning_rate": 1.4655614767865874e-05,
+ "loss": 0.1295,
+ "step": 13413
+ },
+ {
+ "epoch": 36.55040871934605,
+ "grad_norm": 3.1397626399993896,
+ "learning_rate": 1.4654833728268809e-05,
+ "loss": 0.1451,
+ "step": 13414
+ },
+ {
+ "epoch": 36.55313351498638,
+ "grad_norm": 3.30374813079834,
+ "learning_rate": 1.465405265242005e-05,
+ "loss": 0.0919,
+ "step": 13415
+ },
+ {
+ "epoch": 36.555858310626704,
+ "grad_norm": 4.249808311462402,
+ "learning_rate": 1.4653271540325678e-05,
+ "loss": 0.1768,
+ "step": 13416
+ },
+ {
+ "epoch": 36.55858310626703,
+ "grad_norm": 3.4320273399353027,
+ "learning_rate": 1.4652490391991776e-05,
+ "loss": 0.346,
+ "step": 13417
+ },
+ {
+ "epoch": 36.561307901907355,
+ "grad_norm": 4.425020217895508,
+ "learning_rate": 1.4651709207424427e-05,
+ "loss": 0.1832,
+ "step": 13418
+ },
+ {
+ "epoch": 36.56403269754768,
+ "grad_norm": 5.566543102264404,
+ "learning_rate": 1.4650927986629721e-05,
+ "loss": 0.1694,
+ "step": 13419
+ },
+ {
+ "epoch": 36.566757493188014,
+ "grad_norm": 3.9243831634521484,
+ "learning_rate": 1.4650146729613735e-05,
+ "loss": 0.2172,
+ "step": 13420
+ },
+ {
+ "epoch": 36.56948228882834,
+ "grad_norm": 3.622451066970825,
+ "learning_rate": 1.4649365436382557e-05,
+ "loss": 0.0881,
+ "step": 13421
+ },
+ {
+ "epoch": 36.572207084468666,
+ "grad_norm": 4.510460376739502,
+ "learning_rate": 1.464858410694227e-05,
+ "loss": 0.2264,
+ "step": 13422
+ },
+ {
+ "epoch": 36.57493188010899,
+ "grad_norm": 3.514829397201538,
+ "learning_rate": 1.464780274129896e-05,
+ "loss": 0.4449,
+ "step": 13423
+ },
+ {
+ "epoch": 36.57765667574932,
+ "grad_norm": 3.9048430919647217,
+ "learning_rate": 1.464702133945871e-05,
+ "loss": 0.2004,
+ "step": 13424
+ },
+ {
+ "epoch": 36.58038147138964,
+ "grad_norm": 4.688483238220215,
+ "learning_rate": 1.4646239901427608e-05,
+ "loss": 0.1601,
+ "step": 13425
+ },
+ {
+ "epoch": 36.583106267029976,
+ "grad_norm": 3.5562589168548584,
+ "learning_rate": 1.464545842721174e-05,
+ "loss": 0.2791,
+ "step": 13426
+ },
+ {
+ "epoch": 36.5858310626703,
+ "grad_norm": 3.811058759689331,
+ "learning_rate": 1.4644676916817191e-05,
+ "loss": 0.2541,
+ "step": 13427
+ },
+ {
+ "epoch": 36.58855585831063,
+ "grad_norm": 4.028832912445068,
+ "learning_rate": 1.4643895370250048e-05,
+ "loss": 0.1322,
+ "step": 13428
+ },
+ {
+ "epoch": 36.59128065395095,
+ "grad_norm": 3.5302798748016357,
+ "learning_rate": 1.4643113787516398e-05,
+ "loss": 0.0919,
+ "step": 13429
+ },
+ {
+ "epoch": 36.59400544959128,
+ "grad_norm": 4.4097747802734375,
+ "learning_rate": 1.4642332168622326e-05,
+ "loss": 0.2089,
+ "step": 13430
+ },
+ {
+ "epoch": 36.596730245231605,
+ "grad_norm": 3.2241592407226562,
+ "learning_rate": 1.4641550513573922e-05,
+ "loss": 0.0869,
+ "step": 13431
+ },
+ {
+ "epoch": 36.59945504087194,
+ "grad_norm": 4.4640889167785645,
+ "learning_rate": 1.464076882237727e-05,
+ "loss": 0.1881,
+ "step": 13432
+ },
+ {
+ "epoch": 36.60217983651226,
+ "grad_norm": 4.14105749130249,
+ "learning_rate": 1.4639987095038462e-05,
+ "loss": 0.1975,
+ "step": 13433
+ },
+ {
+ "epoch": 36.60490463215259,
+ "grad_norm": 3.2423503398895264,
+ "learning_rate": 1.4639205331563582e-05,
+ "loss": 0.0995,
+ "step": 13434
+ },
+ {
+ "epoch": 36.607629427792915,
+ "grad_norm": 3.9527876377105713,
+ "learning_rate": 1.4638423531958719e-05,
+ "loss": 0.1973,
+ "step": 13435
+ },
+ {
+ "epoch": 36.61035422343324,
+ "grad_norm": 4.354691505432129,
+ "learning_rate": 1.4637641696229967e-05,
+ "loss": 0.1834,
+ "step": 13436
+ },
+ {
+ "epoch": 36.61307901907357,
+ "grad_norm": 3.7938477993011475,
+ "learning_rate": 1.4636859824383406e-05,
+ "loss": 0.1004,
+ "step": 13437
+ },
+ {
+ "epoch": 36.6158038147139,
+ "grad_norm": 4.060889720916748,
+ "learning_rate": 1.4636077916425133e-05,
+ "loss": 0.1408,
+ "step": 13438
+ },
+ {
+ "epoch": 36.618528610354225,
+ "grad_norm": 3.7075183391571045,
+ "learning_rate": 1.4635295972361235e-05,
+ "loss": 0.1895,
+ "step": 13439
+ },
+ {
+ "epoch": 36.62125340599455,
+ "grad_norm": 4.226568222045898,
+ "learning_rate": 1.4634513992197798e-05,
+ "loss": 0.1403,
+ "step": 13440
+ },
+ {
+ "epoch": 36.62397820163488,
+ "grad_norm": 3.314805269241333,
+ "learning_rate": 1.463373197594092e-05,
+ "loss": 0.1535,
+ "step": 13441
+ },
+ {
+ "epoch": 36.6267029972752,
+ "grad_norm": 4.650611400604248,
+ "learning_rate": 1.4632949923596681e-05,
+ "loss": 0.2147,
+ "step": 13442
+ },
+ {
+ "epoch": 36.62942779291553,
+ "grad_norm": 3.046078681945801,
+ "learning_rate": 1.4632167835171177e-05,
+ "loss": 0.1333,
+ "step": 13443
+ },
+ {
+ "epoch": 36.63215258855586,
+ "grad_norm": 3.856060743331909,
+ "learning_rate": 1.4631385710670505e-05,
+ "loss": 0.3176,
+ "step": 13444
+ },
+ {
+ "epoch": 36.63487738419619,
+ "grad_norm": 3.560877561569214,
+ "learning_rate": 1.4630603550100747e-05,
+ "loss": 0.2107,
+ "step": 13445
+ },
+ {
+ "epoch": 36.63760217983651,
+ "grad_norm": 4.023266315460205,
+ "learning_rate": 1.4629821353467997e-05,
+ "loss": 0.1332,
+ "step": 13446
+ },
+ {
+ "epoch": 36.64032697547684,
+ "grad_norm": 3.4434397220611572,
+ "learning_rate": 1.4629039120778345e-05,
+ "loss": 0.1237,
+ "step": 13447
+ },
+ {
+ "epoch": 36.643051771117165,
+ "grad_norm": 4.173470497131348,
+ "learning_rate": 1.462825685203789e-05,
+ "loss": 0.1134,
+ "step": 13448
+ },
+ {
+ "epoch": 36.64577656675749,
+ "grad_norm": 3.8889596462249756,
+ "learning_rate": 1.4627474547252718e-05,
+ "loss": 0.1942,
+ "step": 13449
+ },
+ {
+ "epoch": 36.64850136239782,
+ "grad_norm": 5.267218112945557,
+ "learning_rate": 1.462669220642892e-05,
+ "loss": 0.1904,
+ "step": 13450
+ },
+ {
+ "epoch": 36.65122615803815,
+ "grad_norm": 4.985625267028809,
+ "learning_rate": 1.4625909829572596e-05,
+ "loss": 0.1643,
+ "step": 13451
+ },
+ {
+ "epoch": 36.653950953678475,
+ "grad_norm": 3.837423324584961,
+ "learning_rate": 1.4625127416689834e-05,
+ "loss": 0.2161,
+ "step": 13452
+ },
+ {
+ "epoch": 36.6566757493188,
+ "grad_norm": 4.7848429679870605,
+ "learning_rate": 1.4624344967786729e-05,
+ "loss": 0.1115,
+ "step": 13453
+ },
+ {
+ "epoch": 36.65940054495913,
+ "grad_norm": 3.535413980484009,
+ "learning_rate": 1.4623562482869373e-05,
+ "loss": 0.1439,
+ "step": 13454
+ },
+ {
+ "epoch": 36.66212534059945,
+ "grad_norm": 3.6398262977600098,
+ "learning_rate": 1.462277996194386e-05,
+ "loss": 0.184,
+ "step": 13455
+ },
+ {
+ "epoch": 36.664850136239785,
+ "grad_norm": 3.655811309814453,
+ "learning_rate": 1.462199740501629e-05,
+ "loss": 0.131,
+ "step": 13456
+ },
+ {
+ "epoch": 36.66757493188011,
+ "grad_norm": 3.368325710296631,
+ "learning_rate": 1.462121481209275e-05,
+ "loss": 0.2542,
+ "step": 13457
+ },
+ {
+ "epoch": 36.67029972752044,
+ "grad_norm": 3.8434903621673584,
+ "learning_rate": 1.4620432183179339e-05,
+ "loss": 0.2081,
+ "step": 13458
+ },
+ {
+ "epoch": 36.67302452316076,
+ "grad_norm": 3.7903170585632324,
+ "learning_rate": 1.461964951828215e-05,
+ "loss": 0.2439,
+ "step": 13459
+ },
+ {
+ "epoch": 36.67574931880109,
+ "grad_norm": 4.329790115356445,
+ "learning_rate": 1.4618866817407276e-05,
+ "loss": 0.1321,
+ "step": 13460
+ },
+ {
+ "epoch": 36.678474114441414,
+ "grad_norm": 4.008922576904297,
+ "learning_rate": 1.4618084080560819e-05,
+ "loss": 0.1838,
+ "step": 13461
+ },
+ {
+ "epoch": 36.68119891008175,
+ "grad_norm": 3.628861665725708,
+ "learning_rate": 1.4617301307748872e-05,
+ "loss": 0.1577,
+ "step": 13462
+ },
+ {
+ "epoch": 36.68392370572207,
+ "grad_norm": 4.360490798950195,
+ "learning_rate": 1.461651849897753e-05,
+ "loss": 0.1249,
+ "step": 13463
+ },
+ {
+ "epoch": 36.6866485013624,
+ "grad_norm": 4.622376918792725,
+ "learning_rate": 1.4615735654252894e-05,
+ "loss": 0.1391,
+ "step": 13464
+ },
+ {
+ "epoch": 36.689373297002724,
+ "grad_norm": 3.725651741027832,
+ "learning_rate": 1.4614952773581052e-05,
+ "loss": 0.2265,
+ "step": 13465
+ },
+ {
+ "epoch": 36.69209809264305,
+ "grad_norm": 3.5234570503234863,
+ "learning_rate": 1.461416985696811e-05,
+ "loss": 0.2066,
+ "step": 13466
+ },
+ {
+ "epoch": 36.694822888283376,
+ "grad_norm": 4.215324878692627,
+ "learning_rate": 1.4613386904420161e-05,
+ "loss": 0.1765,
+ "step": 13467
+ },
+ {
+ "epoch": 36.69754768392371,
+ "grad_norm": 4.460766315460205,
+ "learning_rate": 1.4612603915943302e-05,
+ "loss": 0.2498,
+ "step": 13468
+ },
+ {
+ "epoch": 36.700272479564035,
+ "grad_norm": 3.3366076946258545,
+ "learning_rate": 1.4611820891543632e-05,
+ "loss": 0.1495,
+ "step": 13469
+ },
+ {
+ "epoch": 36.70299727520436,
+ "grad_norm": 3.540046215057373,
+ "learning_rate": 1.461103783122725e-05,
+ "loss": 0.1139,
+ "step": 13470
+ },
+ {
+ "epoch": 36.705722070844686,
+ "grad_norm": 4.124093055725098,
+ "learning_rate": 1.4610254735000254e-05,
+ "loss": 0.1486,
+ "step": 13471
+ },
+ {
+ "epoch": 36.70844686648501,
+ "grad_norm": 4.223927974700928,
+ "learning_rate": 1.4609471602868743e-05,
+ "loss": 0.3563,
+ "step": 13472
+ },
+ {
+ "epoch": 36.71117166212534,
+ "grad_norm": 4.150303363800049,
+ "learning_rate": 1.4608688434838816e-05,
+ "loss": 0.272,
+ "step": 13473
+ },
+ {
+ "epoch": 36.71389645776567,
+ "grad_norm": 5.3836259841918945,
+ "learning_rate": 1.460790523091657e-05,
+ "loss": 0.2477,
+ "step": 13474
+ },
+ {
+ "epoch": 36.716621253406,
+ "grad_norm": 4.451722145080566,
+ "learning_rate": 1.4607121991108109e-05,
+ "loss": 0.1231,
+ "step": 13475
+ },
+ {
+ "epoch": 36.71934604904632,
+ "grad_norm": 3.3016858100891113,
+ "learning_rate": 1.4606338715419528e-05,
+ "loss": 0.0688,
+ "step": 13476
+ },
+ {
+ "epoch": 36.72207084468665,
+ "grad_norm": 4.334693431854248,
+ "learning_rate": 1.4605555403856928e-05,
+ "loss": 0.2584,
+ "step": 13477
+ },
+ {
+ "epoch": 36.724795640326974,
+ "grad_norm": 3.6084015369415283,
+ "learning_rate": 1.4604772056426412e-05,
+ "loss": 0.1625,
+ "step": 13478
+ },
+ {
+ "epoch": 36.7275204359673,
+ "grad_norm": 3.132294178009033,
+ "learning_rate": 1.460398867313408e-05,
+ "loss": 0.1774,
+ "step": 13479
+ },
+ {
+ "epoch": 36.73024523160763,
+ "grad_norm": 3.671729803085327,
+ "learning_rate": 1.4603205253986034e-05,
+ "loss": 0.1941,
+ "step": 13480
+ },
+ {
+ "epoch": 36.73297002724796,
+ "grad_norm": 3.435577869415283,
+ "learning_rate": 1.4602421798988371e-05,
+ "loss": 0.2375,
+ "step": 13481
+ },
+ {
+ "epoch": 36.735694822888284,
+ "grad_norm": 3.2753610610961914,
+ "learning_rate": 1.4601638308147196e-05,
+ "loss": 0.1175,
+ "step": 13482
+ },
+ {
+ "epoch": 36.73841961852861,
+ "grad_norm": 4.2215776443481445,
+ "learning_rate": 1.4600854781468611e-05,
+ "loss": 0.1682,
+ "step": 13483
+ },
+ {
+ "epoch": 36.741144414168936,
+ "grad_norm": 3.8162243366241455,
+ "learning_rate": 1.4600071218958718e-05,
+ "loss": 0.1273,
+ "step": 13484
+ },
+ {
+ "epoch": 36.74386920980926,
+ "grad_norm": 3.372706174850464,
+ "learning_rate": 1.4599287620623612e-05,
+ "loss": 0.1038,
+ "step": 13485
+ },
+ {
+ "epoch": 36.746594005449595,
+ "grad_norm": 3.5140185356140137,
+ "learning_rate": 1.4598503986469408e-05,
+ "loss": 0.1516,
+ "step": 13486
+ },
+ {
+ "epoch": 36.74931880108992,
+ "grad_norm": 3.274367094039917,
+ "learning_rate": 1.4597720316502201e-05,
+ "loss": 0.2312,
+ "step": 13487
+ },
+ {
+ "epoch": 36.752043596730246,
+ "grad_norm": 3.167862892150879,
+ "learning_rate": 1.4596936610728098e-05,
+ "loss": 0.1186,
+ "step": 13488
+ },
+ {
+ "epoch": 36.75476839237057,
+ "grad_norm": 4.6436872482299805,
+ "learning_rate": 1.4596152869153199e-05,
+ "loss": 0.1323,
+ "step": 13489
+ },
+ {
+ "epoch": 36.7574931880109,
+ "grad_norm": 4.326160907745361,
+ "learning_rate": 1.459536909178361e-05,
+ "loss": 0.2074,
+ "step": 13490
+ },
+ {
+ "epoch": 36.76021798365122,
+ "grad_norm": 4.032314777374268,
+ "learning_rate": 1.4594585278625435e-05,
+ "loss": 0.1066,
+ "step": 13491
+ },
+ {
+ "epoch": 36.762942779291556,
+ "grad_norm": 5.255019187927246,
+ "learning_rate": 1.4593801429684778e-05,
+ "loss": 0.3739,
+ "step": 13492
+ },
+ {
+ "epoch": 36.76566757493188,
+ "grad_norm": 4.077271461486816,
+ "learning_rate": 1.4593017544967741e-05,
+ "loss": 0.1852,
+ "step": 13493
+ },
+ {
+ "epoch": 36.76839237057221,
+ "grad_norm": 3.6170263290405273,
+ "learning_rate": 1.4592233624480432e-05,
+ "loss": 0.1273,
+ "step": 13494
+ },
+ {
+ "epoch": 36.771117166212534,
+ "grad_norm": 3.940189838409424,
+ "learning_rate": 1.4591449668228954e-05,
+ "loss": 0.292,
+ "step": 13495
+ },
+ {
+ "epoch": 36.77384196185286,
+ "grad_norm": 3.8475749492645264,
+ "learning_rate": 1.4590665676219416e-05,
+ "loss": 0.1301,
+ "step": 13496
+ },
+ {
+ "epoch": 36.776566757493185,
+ "grad_norm": 3.862938404083252,
+ "learning_rate": 1.458988164845792e-05,
+ "loss": 0.1396,
+ "step": 13497
+ },
+ {
+ "epoch": 36.77929155313352,
+ "grad_norm": 4.414910793304443,
+ "learning_rate": 1.4589097584950576e-05,
+ "loss": 0.1434,
+ "step": 13498
+ },
+ {
+ "epoch": 36.782016348773844,
+ "grad_norm": 4.130340099334717,
+ "learning_rate": 1.4588313485703484e-05,
+ "loss": 0.0978,
+ "step": 13499
+ },
+ {
+ "epoch": 36.78474114441417,
+ "grad_norm": 3.879321575164795,
+ "learning_rate": 1.4587529350722758e-05,
+ "loss": 0.1652,
+ "step": 13500
+ },
+ {
+ "epoch": 36.787465940054496,
+ "grad_norm": 3.7773048877716064,
+ "learning_rate": 1.4586745180014498e-05,
+ "loss": 0.1335,
+ "step": 13501
+ },
+ {
+ "epoch": 36.79019073569482,
+ "grad_norm": 3.8057773113250732,
+ "learning_rate": 1.4585960973584819e-05,
+ "loss": 0.1269,
+ "step": 13502
+ },
+ {
+ "epoch": 36.79291553133515,
+ "grad_norm": 3.800973892211914,
+ "learning_rate": 1.4585176731439816e-05,
+ "loss": 0.4673,
+ "step": 13503
+ },
+ {
+ "epoch": 36.79564032697548,
+ "grad_norm": 3.4435291290283203,
+ "learning_rate": 1.458439245358561e-05,
+ "loss": 0.1462,
+ "step": 13504
+ },
+ {
+ "epoch": 36.798365122615806,
+ "grad_norm": 4.966694355010986,
+ "learning_rate": 1.4583608140028301e-05,
+ "loss": 0.3309,
+ "step": 13505
+ },
+ {
+ "epoch": 36.80108991825613,
+ "grad_norm": 3.473142623901367,
+ "learning_rate": 1.4582823790774001e-05,
+ "loss": 0.0981,
+ "step": 13506
+ },
+ {
+ "epoch": 36.80381471389646,
+ "grad_norm": 3.2586159706115723,
+ "learning_rate": 1.4582039405828813e-05,
+ "loss": 0.1138,
+ "step": 13507
+ },
+ {
+ "epoch": 36.80653950953678,
+ "grad_norm": 3.2870802879333496,
+ "learning_rate": 1.4581254985198854e-05,
+ "loss": 0.2569,
+ "step": 13508
+ },
+ {
+ "epoch": 36.80926430517711,
+ "grad_norm": 4.089127063751221,
+ "learning_rate": 1.4580470528890227e-05,
+ "loss": 0.1622,
+ "step": 13509
+ },
+ {
+ "epoch": 36.81198910081744,
+ "grad_norm": 4.598082542419434,
+ "learning_rate": 1.4579686036909043e-05,
+ "loss": 0.1425,
+ "step": 13510
+ },
+ {
+ "epoch": 36.81471389645777,
+ "grad_norm": 3.789857864379883,
+ "learning_rate": 1.457890150926141e-05,
+ "loss": 0.2039,
+ "step": 13511
+ },
+ {
+ "epoch": 36.817438692098094,
+ "grad_norm": 3.9066247940063477,
+ "learning_rate": 1.457811694595344e-05,
+ "loss": 0.1713,
+ "step": 13512
+ },
+ {
+ "epoch": 36.82016348773842,
+ "grad_norm": 3.189425230026245,
+ "learning_rate": 1.4577332346991244e-05,
+ "loss": 0.2195,
+ "step": 13513
+ },
+ {
+ "epoch": 36.822888283378745,
+ "grad_norm": 4.120590686798096,
+ "learning_rate": 1.4576547712380931e-05,
+ "loss": 0.1454,
+ "step": 13514
+ },
+ {
+ "epoch": 36.82561307901907,
+ "grad_norm": 3.924074411392212,
+ "learning_rate": 1.457576304212861e-05,
+ "loss": 0.2966,
+ "step": 13515
+ },
+ {
+ "epoch": 36.828337874659404,
+ "grad_norm": 3.997619390487671,
+ "learning_rate": 1.4574978336240396e-05,
+ "loss": 0.2138,
+ "step": 13516
+ },
+ {
+ "epoch": 36.83106267029973,
+ "grad_norm": 3.4480998516082764,
+ "learning_rate": 1.4574193594722394e-05,
+ "loss": 0.2517,
+ "step": 13517
+ },
+ {
+ "epoch": 36.833787465940055,
+ "grad_norm": 3.7926924228668213,
+ "learning_rate": 1.4573408817580723e-05,
+ "loss": 0.1767,
+ "step": 13518
+ },
+ {
+ "epoch": 36.83651226158038,
+ "grad_norm": 7.524046897888184,
+ "learning_rate": 1.457262400482149e-05,
+ "loss": 0.2719,
+ "step": 13519
+ },
+ {
+ "epoch": 36.83923705722071,
+ "grad_norm": 3.810415744781494,
+ "learning_rate": 1.4571839156450809e-05,
+ "loss": 0.1702,
+ "step": 13520
+ },
+ {
+ "epoch": 36.84196185286103,
+ "grad_norm": 4.968423843383789,
+ "learning_rate": 1.4571054272474791e-05,
+ "loss": 0.2055,
+ "step": 13521
+ },
+ {
+ "epoch": 36.844686648501366,
+ "grad_norm": 3.9061784744262695,
+ "learning_rate": 1.457026935289955e-05,
+ "loss": 0.1301,
+ "step": 13522
+ },
+ {
+ "epoch": 36.84741144414169,
+ "grad_norm": 3.393244504928589,
+ "learning_rate": 1.45694843977312e-05,
+ "loss": 0.1608,
+ "step": 13523
+ },
+ {
+ "epoch": 36.85013623978202,
+ "grad_norm": 3.4371418952941895,
+ "learning_rate": 1.456869940697585e-05,
+ "loss": 0.1093,
+ "step": 13524
+ },
+ {
+ "epoch": 36.85286103542234,
+ "grad_norm": 4.049927234649658,
+ "learning_rate": 1.4567914380639617e-05,
+ "loss": 0.2213,
+ "step": 13525
+ },
+ {
+ "epoch": 36.85558583106267,
+ "grad_norm": 4.070270538330078,
+ "learning_rate": 1.4567129318728614e-05,
+ "loss": 0.1292,
+ "step": 13526
+ },
+ {
+ "epoch": 36.858310626702995,
+ "grad_norm": 3.4641427993774414,
+ "learning_rate": 1.4566344221248957e-05,
+ "loss": 0.1772,
+ "step": 13527
+ },
+ {
+ "epoch": 36.86103542234333,
+ "grad_norm": 3.1217551231384277,
+ "learning_rate": 1.4565559088206753e-05,
+ "loss": 0.1086,
+ "step": 13528
+ },
+ {
+ "epoch": 36.86376021798365,
+ "grad_norm": 4.15180778503418,
+ "learning_rate": 1.4564773919608125e-05,
+ "loss": 0.1831,
+ "step": 13529
+ },
+ {
+ "epoch": 36.86648501362398,
+ "grad_norm": 3.3173909187316895,
+ "learning_rate": 1.4563988715459185e-05,
+ "loss": 0.1183,
+ "step": 13530
+ },
+ {
+ "epoch": 36.869209809264305,
+ "grad_norm": 4.210923671722412,
+ "learning_rate": 1.4563203475766048e-05,
+ "loss": 0.1552,
+ "step": 13531
+ },
+ {
+ "epoch": 36.87193460490463,
+ "grad_norm": 4.497961044311523,
+ "learning_rate": 1.4562418200534827e-05,
+ "loss": 0.2293,
+ "step": 13532
+ },
+ {
+ "epoch": 36.87465940054496,
+ "grad_norm": 4.365159034729004,
+ "learning_rate": 1.4561632889771642e-05,
+ "loss": 0.1356,
+ "step": 13533
+ },
+ {
+ "epoch": 36.87738419618529,
+ "grad_norm": 5.944848537445068,
+ "learning_rate": 1.4560847543482605e-05,
+ "loss": 0.2914,
+ "step": 13534
+ },
+ {
+ "epoch": 36.880108991825615,
+ "grad_norm": 3.3907597064971924,
+ "learning_rate": 1.4560062161673838e-05,
+ "loss": 0.1693,
+ "step": 13535
+ },
+ {
+ "epoch": 36.88283378746594,
+ "grad_norm": 3.378484010696411,
+ "learning_rate": 1.455927674435145e-05,
+ "loss": 0.1732,
+ "step": 13536
+ },
+ {
+ "epoch": 36.88555858310627,
+ "grad_norm": 3.708082914352417,
+ "learning_rate": 1.4558491291521561e-05,
+ "loss": 0.143,
+ "step": 13537
+ },
+ {
+ "epoch": 36.88828337874659,
+ "grad_norm": 4.335235595703125,
+ "learning_rate": 1.455770580319029e-05,
+ "loss": 0.3861,
+ "step": 13538
+ },
+ {
+ "epoch": 36.89100817438692,
+ "grad_norm": 3.5353078842163086,
+ "learning_rate": 1.4556920279363754e-05,
+ "loss": 0.2898,
+ "step": 13539
+ },
+ {
+ "epoch": 36.89373297002725,
+ "grad_norm": 3.623497724533081,
+ "learning_rate": 1.4556134720048066e-05,
+ "loss": 0.2376,
+ "step": 13540
+ },
+ {
+ "epoch": 36.89645776566758,
+ "grad_norm": 4.276484489440918,
+ "learning_rate": 1.4555349125249353e-05,
+ "loss": 0.1684,
+ "step": 13541
+ },
+ {
+ "epoch": 36.8991825613079,
+ "grad_norm": 3.87426495552063,
+ "learning_rate": 1.4554563494973721e-05,
+ "loss": 0.4075,
+ "step": 13542
+ },
+ {
+ "epoch": 36.90190735694823,
+ "grad_norm": 3.9037222862243652,
+ "learning_rate": 1.45537778292273e-05,
+ "loss": 0.1085,
+ "step": 13543
+ },
+ {
+ "epoch": 36.904632152588555,
+ "grad_norm": 3.4015347957611084,
+ "learning_rate": 1.4552992128016202e-05,
+ "loss": 0.1887,
+ "step": 13544
+ },
+ {
+ "epoch": 36.90735694822888,
+ "grad_norm": 3.685048818588257,
+ "learning_rate": 1.4552206391346548e-05,
+ "loss": 0.3499,
+ "step": 13545
+ },
+ {
+ "epoch": 36.91008174386921,
+ "grad_norm": 3.583357095718384,
+ "learning_rate": 1.4551420619224455e-05,
+ "loss": 0.1863,
+ "step": 13546
+ },
+ {
+ "epoch": 36.91280653950954,
+ "grad_norm": 4.528999328613281,
+ "learning_rate": 1.4550634811656048e-05,
+ "loss": 0.2398,
+ "step": 13547
+ },
+ {
+ "epoch": 36.915531335149865,
+ "grad_norm": 4.342367172241211,
+ "learning_rate": 1.4549848968647441e-05,
+ "loss": 0.1274,
+ "step": 13548
+ },
+ {
+ "epoch": 36.91825613079019,
+ "grad_norm": 3.757988214492798,
+ "learning_rate": 1.454906309020476e-05,
+ "loss": 0.1892,
+ "step": 13549
+ },
+ {
+ "epoch": 36.920980926430516,
+ "grad_norm": 3.583442449569702,
+ "learning_rate": 1.4548277176334121e-05,
+ "loss": 0.1993,
+ "step": 13550
+ },
+ {
+ "epoch": 36.92370572207084,
+ "grad_norm": 4.156078815460205,
+ "learning_rate": 1.4547491227041643e-05,
+ "loss": 0.1588,
+ "step": 13551
+ },
+ {
+ "epoch": 36.926430517711175,
+ "grad_norm": 3.6952269077301025,
+ "learning_rate": 1.4546705242333452e-05,
+ "loss": 0.1757,
+ "step": 13552
+ },
+ {
+ "epoch": 36.9291553133515,
+ "grad_norm": 3.931908130645752,
+ "learning_rate": 1.4545919222215668e-05,
+ "loss": 0.3447,
+ "step": 13553
+ },
+ {
+ "epoch": 36.93188010899183,
+ "grad_norm": 3.9014177322387695,
+ "learning_rate": 1.4545133166694408e-05,
+ "loss": 0.2841,
+ "step": 13554
+ },
+ {
+ "epoch": 36.93460490463215,
+ "grad_norm": 3.1277265548706055,
+ "learning_rate": 1.45443470757758e-05,
+ "loss": 0.1697,
+ "step": 13555
+ },
+ {
+ "epoch": 36.93732970027248,
+ "grad_norm": 3.729722738265991,
+ "learning_rate": 1.4543560949465963e-05,
+ "loss": 0.1486,
+ "step": 13556
+ },
+ {
+ "epoch": 36.940054495912804,
+ "grad_norm": 4.493760585784912,
+ "learning_rate": 1.4542774787771019e-05,
+ "loss": 0.1375,
+ "step": 13557
+ },
+ {
+ "epoch": 36.94277929155314,
+ "grad_norm": 4.078537464141846,
+ "learning_rate": 1.4541988590697089e-05,
+ "loss": 0.1694,
+ "step": 13558
+ },
+ {
+ "epoch": 36.94550408719346,
+ "grad_norm": 3.4051272869110107,
+ "learning_rate": 1.4541202358250301e-05,
+ "loss": 0.0969,
+ "step": 13559
+ },
+ {
+ "epoch": 36.94822888283379,
+ "grad_norm": 3.8305509090423584,
+ "learning_rate": 1.4540416090436775e-05,
+ "loss": 0.1869,
+ "step": 13560
+ },
+ {
+ "epoch": 36.950953678474114,
+ "grad_norm": 3.955962896347046,
+ "learning_rate": 1.4539629787262635e-05,
+ "loss": 0.1656,
+ "step": 13561
+ },
+ {
+ "epoch": 36.95367847411444,
+ "grad_norm": 3.583622932434082,
+ "learning_rate": 1.4538843448734001e-05,
+ "loss": 0.2388,
+ "step": 13562
+ },
+ {
+ "epoch": 36.956403269754766,
+ "grad_norm": 3.8269805908203125,
+ "learning_rate": 1.4538057074857004e-05,
+ "loss": 0.3071,
+ "step": 13563
+ },
+ {
+ "epoch": 36.95912806539509,
+ "grad_norm": 4.137397766113281,
+ "learning_rate": 1.4537270665637762e-05,
+ "loss": 0.1393,
+ "step": 13564
+ },
+ {
+ "epoch": 36.961852861035425,
+ "grad_norm": 4.37985897064209,
+ "learning_rate": 1.4536484221082403e-05,
+ "loss": 0.1255,
+ "step": 13565
+ },
+ {
+ "epoch": 36.96457765667575,
+ "grad_norm": 4.7956743240356445,
+ "learning_rate": 1.453569774119705e-05,
+ "loss": 0.2187,
+ "step": 13566
+ },
+ {
+ "epoch": 36.967302452316076,
+ "grad_norm": 3.3594775199890137,
+ "learning_rate": 1.4534911225987832e-05,
+ "loss": 0.2903,
+ "step": 13567
+ },
+ {
+ "epoch": 36.9700272479564,
+ "grad_norm": 4.3006815910339355,
+ "learning_rate": 1.4534124675460868e-05,
+ "loss": 0.2181,
+ "step": 13568
+ },
+ {
+ "epoch": 36.97275204359673,
+ "grad_norm": 4.630638122558594,
+ "learning_rate": 1.453333808962229e-05,
+ "loss": 0.1636,
+ "step": 13569
+ },
+ {
+ "epoch": 36.97547683923706,
+ "grad_norm": 3.693990468978882,
+ "learning_rate": 1.4532551468478217e-05,
+ "loss": 0.2182,
+ "step": 13570
+ },
+ {
+ "epoch": 36.97820163487739,
+ "grad_norm": 4.355136871337891,
+ "learning_rate": 1.4531764812034781e-05,
+ "loss": 0.2664,
+ "step": 13571
+ },
+ {
+ "epoch": 36.98092643051771,
+ "grad_norm": 4.216235637664795,
+ "learning_rate": 1.4530978120298108e-05,
+ "loss": 0.2115,
+ "step": 13572
+ },
+ {
+ "epoch": 36.98365122615804,
+ "grad_norm": 3.5452022552490234,
+ "learning_rate": 1.4530191393274322e-05,
+ "loss": 0.1575,
+ "step": 13573
+ },
+ {
+ "epoch": 36.986376021798364,
+ "grad_norm": 3.9963700771331787,
+ "learning_rate": 1.452940463096955e-05,
+ "loss": 0.0989,
+ "step": 13574
+ },
+ {
+ "epoch": 36.98910081743869,
+ "grad_norm": 4.0252485275268555,
+ "learning_rate": 1.4528617833389925e-05,
+ "loss": 0.1859,
+ "step": 13575
+ },
+ {
+ "epoch": 36.991825613079016,
+ "grad_norm": 3.4766037464141846,
+ "learning_rate": 1.4527831000541564e-05,
+ "loss": 0.1787,
+ "step": 13576
+ },
+ {
+ "epoch": 36.99455040871935,
+ "grad_norm": 3.454817771911621,
+ "learning_rate": 1.4527044132430605e-05,
+ "loss": 0.2873,
+ "step": 13577
+ },
+ {
+ "epoch": 36.997275204359674,
+ "grad_norm": 4.369072437286377,
+ "learning_rate": 1.452625722906317e-05,
+ "loss": 0.1637,
+ "step": 13578
+ },
+ {
+ "epoch": 37.0,
+ "grad_norm": 5.162822246551514,
+ "learning_rate": 1.4525470290445392e-05,
+ "loss": 0.126,
+ "step": 13579
+ },
+ {
+ "epoch": 37.002724795640326,
+ "grad_norm": 3.9214189052581787,
+ "learning_rate": 1.4524683316583392e-05,
+ "loss": 0.226,
+ "step": 13580
+ },
+ {
+ "epoch": 37.00544959128065,
+ "grad_norm": 3.6210668087005615,
+ "learning_rate": 1.4523896307483308e-05,
+ "loss": 0.1678,
+ "step": 13581
+ },
+ {
+ "epoch": 37.00817438692098,
+ "grad_norm": 3.429661273956299,
+ "learning_rate": 1.4523109263151262e-05,
+ "loss": 0.1477,
+ "step": 13582
+ },
+ {
+ "epoch": 37.01089918256131,
+ "grad_norm": 3.0631422996520996,
+ "learning_rate": 1.4522322183593392e-05,
+ "loss": 0.0947,
+ "step": 13583
+ },
+ {
+ "epoch": 37.013623978201636,
+ "grad_norm": 4.483565330505371,
+ "learning_rate": 1.4521535068815817e-05,
+ "loss": 0.183,
+ "step": 13584
+ },
+ {
+ "epoch": 37.01634877384196,
+ "grad_norm": 2.8948843479156494,
+ "learning_rate": 1.4520747918824676e-05,
+ "loss": 0.1443,
+ "step": 13585
+ },
+ {
+ "epoch": 37.01907356948229,
+ "grad_norm": 4.156149387359619,
+ "learning_rate": 1.4519960733626093e-05,
+ "loss": 0.122,
+ "step": 13586
+ },
+ {
+ "epoch": 37.02179836512261,
+ "grad_norm": 3.519052267074585,
+ "learning_rate": 1.4519173513226203e-05,
+ "loss": 0.185,
+ "step": 13587
+ },
+ {
+ "epoch": 37.02452316076294,
+ "grad_norm": 4.208253383636475,
+ "learning_rate": 1.4518386257631135e-05,
+ "loss": 0.1358,
+ "step": 13588
+ },
+ {
+ "epoch": 37.02724795640327,
+ "grad_norm": 3.543532609939575,
+ "learning_rate": 1.4517598966847019e-05,
+ "loss": 0.2307,
+ "step": 13589
+ },
+ {
+ "epoch": 37.0299727520436,
+ "grad_norm": 3.717811346054077,
+ "learning_rate": 1.4516811640879988e-05,
+ "loss": 0.1361,
+ "step": 13590
+ },
+ {
+ "epoch": 37.032697547683924,
+ "grad_norm": 3.3498079776763916,
+ "learning_rate": 1.4516024279736177e-05,
+ "loss": 0.2095,
+ "step": 13591
+ },
+ {
+ "epoch": 37.03542234332425,
+ "grad_norm": 3.3489181995391846,
+ "learning_rate": 1.4515236883421708e-05,
+ "loss": 0.098,
+ "step": 13592
+ },
+ {
+ "epoch": 37.038147138964575,
+ "grad_norm": 4.083990097045898,
+ "learning_rate": 1.4514449451942725e-05,
+ "loss": 0.2499,
+ "step": 13593
+ },
+ {
+ "epoch": 37.0408719346049,
+ "grad_norm": 3.6576101779937744,
+ "learning_rate": 1.4513661985305351e-05,
+ "loss": 0.2358,
+ "step": 13594
+ },
+ {
+ "epoch": 37.043596730245234,
+ "grad_norm": 3.7327957153320312,
+ "learning_rate": 1.4512874483515727e-05,
+ "loss": 0.1787,
+ "step": 13595
+ },
+ {
+ "epoch": 37.04632152588556,
+ "grad_norm": 3.6718173027038574,
+ "learning_rate": 1.4512086946579977e-05,
+ "loss": 0.3752,
+ "step": 13596
+ },
+ {
+ "epoch": 37.049046321525886,
+ "grad_norm": 3.0314362049102783,
+ "learning_rate": 1.4511299374504242e-05,
+ "loss": 0.2756,
+ "step": 13597
+ },
+ {
+ "epoch": 37.05177111716621,
+ "grad_norm": 3.7492446899414062,
+ "learning_rate": 1.4510511767294653e-05,
+ "loss": 0.119,
+ "step": 13598
+ },
+ {
+ "epoch": 37.05449591280654,
+ "grad_norm": 5.51267671585083,
+ "learning_rate": 1.4509724124957341e-05,
+ "loss": 0.1239,
+ "step": 13599
+ },
+ {
+ "epoch": 37.05722070844686,
+ "grad_norm": 3.4550974369049072,
+ "learning_rate": 1.4508936447498442e-05,
+ "loss": 0.1348,
+ "step": 13600
+ },
+ {
+ "epoch": 37.059945504087196,
+ "grad_norm": 3.4729130268096924,
+ "learning_rate": 1.4508148734924095e-05,
+ "loss": 0.1068,
+ "step": 13601
+ },
+ {
+ "epoch": 37.06267029972752,
+ "grad_norm": 3.6329762935638428,
+ "learning_rate": 1.4507360987240429e-05,
+ "loss": 0.1054,
+ "step": 13602
+ },
+ {
+ "epoch": 37.06539509536785,
+ "grad_norm": 4.169193744659424,
+ "learning_rate": 1.450657320445358e-05,
+ "loss": 0.2812,
+ "step": 13603
+ },
+ {
+ "epoch": 37.06811989100817,
+ "grad_norm": 3.4690418243408203,
+ "learning_rate": 1.4505785386569686e-05,
+ "loss": 0.1553,
+ "step": 13604
+ },
+ {
+ "epoch": 37.0708446866485,
+ "grad_norm": 136.98480224609375,
+ "learning_rate": 1.4504997533594881e-05,
+ "loss": 0.1713,
+ "step": 13605
+ },
+ {
+ "epoch": 37.073569482288825,
+ "grad_norm": 3.1124184131622314,
+ "learning_rate": 1.4504209645535294e-05,
+ "loss": 0.1095,
+ "step": 13606
+ },
+ {
+ "epoch": 37.07629427792916,
+ "grad_norm": 4.41782283782959,
+ "learning_rate": 1.450342172239707e-05,
+ "loss": 0.124,
+ "step": 13607
+ },
+ {
+ "epoch": 37.079019073569484,
+ "grad_norm": 5.9045329093933105,
+ "learning_rate": 1.4502633764186346e-05,
+ "loss": 0.2677,
+ "step": 13608
+ },
+ {
+ "epoch": 37.08174386920981,
+ "grad_norm": 4.932072162628174,
+ "learning_rate": 1.4501845770909253e-05,
+ "loss": 0.2198,
+ "step": 13609
+ },
+ {
+ "epoch": 37.084468664850135,
+ "grad_norm": 6.572333335876465,
+ "learning_rate": 1.4501057742571931e-05,
+ "loss": 0.274,
+ "step": 13610
+ },
+ {
+ "epoch": 37.08719346049046,
+ "grad_norm": 8.687525749206543,
+ "learning_rate": 1.4500269679180513e-05,
+ "loss": 0.4692,
+ "step": 13611
+ },
+ {
+ "epoch": 37.08991825613079,
+ "grad_norm": 9.101067543029785,
+ "learning_rate": 1.4499481580741144e-05,
+ "loss": 0.2629,
+ "step": 13612
+ },
+ {
+ "epoch": 37.09264305177112,
+ "grad_norm": 11.676504135131836,
+ "learning_rate": 1.4498693447259956e-05,
+ "loss": 0.4755,
+ "step": 13613
+ },
+ {
+ "epoch": 37.095367847411445,
+ "grad_norm": 11.828145980834961,
+ "learning_rate": 1.4497905278743086e-05,
+ "loss": 0.2631,
+ "step": 13614
+ },
+ {
+ "epoch": 37.09809264305177,
+ "grad_norm": 11.727564811706543,
+ "learning_rate": 1.4497117075196677e-05,
+ "loss": 0.293,
+ "step": 13615
+ },
+ {
+ "epoch": 37.1008174386921,
+ "grad_norm": 5.971573829650879,
+ "learning_rate": 1.4496328836626863e-05,
+ "loss": 0.3532,
+ "step": 13616
+ },
+ {
+ "epoch": 37.10354223433242,
+ "grad_norm": 8.028942108154297,
+ "learning_rate": 1.4495540563039785e-05,
+ "loss": 0.2767,
+ "step": 13617
+ },
+ {
+ "epoch": 37.10626702997275,
+ "grad_norm": 6.783729076385498,
+ "learning_rate": 1.4494752254441585e-05,
+ "loss": 0.2363,
+ "step": 13618
+ },
+ {
+ "epoch": 37.10899182561308,
+ "grad_norm": 7.514562129974365,
+ "learning_rate": 1.4493963910838393e-05,
+ "loss": 0.3821,
+ "step": 13619
+ },
+ {
+ "epoch": 37.11171662125341,
+ "grad_norm": 6.603673458099365,
+ "learning_rate": 1.4493175532236361e-05,
+ "loss": 0.2951,
+ "step": 13620
+ },
+ {
+ "epoch": 37.11444141689373,
+ "grad_norm": 6.976290225982666,
+ "learning_rate": 1.4492387118641618e-05,
+ "loss": 0.2869,
+ "step": 13621
+ },
+ {
+ "epoch": 37.11716621253406,
+ "grad_norm": 6.453617572784424,
+ "learning_rate": 1.4491598670060312e-05,
+ "loss": 0.3948,
+ "step": 13622
+ },
+ {
+ "epoch": 37.119891008174385,
+ "grad_norm": 4.888205051422119,
+ "learning_rate": 1.4490810186498577e-05,
+ "loss": 0.1988,
+ "step": 13623
+ },
+ {
+ "epoch": 37.12261580381471,
+ "grad_norm": 5.3571977615356445,
+ "learning_rate": 1.4490021667962556e-05,
+ "loss": 0.2,
+ "step": 13624
+ },
+ {
+ "epoch": 37.12534059945504,
+ "grad_norm": 4.008316993713379,
+ "learning_rate": 1.4489233114458394e-05,
+ "loss": 0.3258,
+ "step": 13625
+ },
+ {
+ "epoch": 37.12806539509537,
+ "grad_norm": 7.693735599517822,
+ "learning_rate": 1.448844452599223e-05,
+ "loss": 0.3745,
+ "step": 13626
+ },
+ {
+ "epoch": 37.130790190735695,
+ "grad_norm": 5.1889190673828125,
+ "learning_rate": 1.4487655902570201e-05,
+ "loss": 0.3711,
+ "step": 13627
+ },
+ {
+ "epoch": 37.13351498637602,
+ "grad_norm": 14.465084075927734,
+ "learning_rate": 1.4486867244198455e-05,
+ "loss": 0.3083,
+ "step": 13628
+ },
+ {
+ "epoch": 37.13623978201635,
+ "grad_norm": 5.561546325683594,
+ "learning_rate": 1.4486078550883129e-05,
+ "loss": 0.225,
+ "step": 13629
+ },
+ {
+ "epoch": 37.13896457765667,
+ "grad_norm": 5.784046173095703,
+ "learning_rate": 1.448528982263037e-05,
+ "loss": 0.2201,
+ "step": 13630
+ },
+ {
+ "epoch": 37.141689373297005,
+ "grad_norm": 5.5845112800598145,
+ "learning_rate": 1.4484501059446317e-05,
+ "loss": 0.3046,
+ "step": 13631
+ },
+ {
+ "epoch": 37.14441416893733,
+ "grad_norm": 5.912179470062256,
+ "learning_rate": 1.4483712261337113e-05,
+ "loss": 0.3351,
+ "step": 13632
+ },
+ {
+ "epoch": 37.14713896457766,
+ "grad_norm": 5.473941326141357,
+ "learning_rate": 1.4482923428308904e-05,
+ "loss": 0.3723,
+ "step": 13633
+ },
+ {
+ "epoch": 37.14986376021798,
+ "grad_norm": 6.63756799697876,
+ "learning_rate": 1.448213456036783e-05,
+ "loss": 0.2224,
+ "step": 13634
+ },
+ {
+ "epoch": 37.15258855585831,
+ "grad_norm": 4.4367194175720215,
+ "learning_rate": 1.4481345657520036e-05,
+ "loss": 0.4343,
+ "step": 13635
+ },
+ {
+ "epoch": 37.155313351498634,
+ "grad_norm": 6.380015850067139,
+ "learning_rate": 1.4480556719771671e-05,
+ "loss": 0.2412,
+ "step": 13636
+ },
+ {
+ "epoch": 37.15803814713897,
+ "grad_norm": 5.190423965454102,
+ "learning_rate": 1.447976774712887e-05,
+ "loss": 0.1585,
+ "step": 13637
+ },
+ {
+ "epoch": 37.16076294277929,
+ "grad_norm": 4.444990158081055,
+ "learning_rate": 1.4478978739597781e-05,
+ "loss": 0.2628,
+ "step": 13638
+ },
+ {
+ "epoch": 37.16348773841962,
+ "grad_norm": 5.372991561889648,
+ "learning_rate": 1.4478189697184553e-05,
+ "loss": 0.1707,
+ "step": 13639
+ },
+ {
+ "epoch": 37.166212534059945,
+ "grad_norm": 4.8876051902771,
+ "learning_rate": 1.4477400619895326e-05,
+ "loss": 0.1545,
+ "step": 13640
+ },
+ {
+ "epoch": 37.16893732970027,
+ "grad_norm": 5.498533248901367,
+ "learning_rate": 1.4476611507736244e-05,
+ "loss": 0.2173,
+ "step": 13641
+ },
+ {
+ "epoch": 37.171662125340596,
+ "grad_norm": 7.100074291229248,
+ "learning_rate": 1.4475822360713463e-05,
+ "loss": 0.3266,
+ "step": 13642
+ },
+ {
+ "epoch": 37.17438692098093,
+ "grad_norm": 5.136620998382568,
+ "learning_rate": 1.4475033178833116e-05,
+ "loss": 0.1561,
+ "step": 13643
+ },
+ {
+ "epoch": 37.177111716621255,
+ "grad_norm": 5.183481693267822,
+ "learning_rate": 1.4474243962101357e-05,
+ "loss": 0.1903,
+ "step": 13644
+ },
+ {
+ "epoch": 37.17983651226158,
+ "grad_norm": 4.034117698669434,
+ "learning_rate": 1.4473454710524328e-05,
+ "loss": 0.1739,
+ "step": 13645
+ },
+ {
+ "epoch": 37.182561307901906,
+ "grad_norm": 6.25624942779541,
+ "learning_rate": 1.447266542410818e-05,
+ "loss": 0.1482,
+ "step": 13646
+ },
+ {
+ "epoch": 37.18528610354223,
+ "grad_norm": 5.069846153259277,
+ "learning_rate": 1.4471876102859057e-05,
+ "loss": 0.1546,
+ "step": 13647
+ },
+ {
+ "epoch": 37.18801089918256,
+ "grad_norm": 6.476383209228516,
+ "learning_rate": 1.4471086746783108e-05,
+ "loss": 0.2393,
+ "step": 13648
+ },
+ {
+ "epoch": 37.19073569482289,
+ "grad_norm": 4.902276992797852,
+ "learning_rate": 1.4470297355886476e-05,
+ "loss": 0.2212,
+ "step": 13649
+ },
+ {
+ "epoch": 37.19346049046322,
+ "grad_norm": 4.5963215827941895,
+ "learning_rate": 1.4469507930175313e-05,
+ "loss": 0.1627,
+ "step": 13650
+ },
+ {
+ "epoch": 37.19618528610354,
+ "grad_norm": 4.550283432006836,
+ "learning_rate": 1.4468718469655766e-05,
+ "loss": 0.1452,
+ "step": 13651
+ },
+ {
+ "epoch": 37.19891008174387,
+ "grad_norm": 5.636865139007568,
+ "learning_rate": 1.4467928974333987e-05,
+ "loss": 0.2773,
+ "step": 13652
+ },
+ {
+ "epoch": 37.201634877384194,
+ "grad_norm": 3.9459500312805176,
+ "learning_rate": 1.4467139444216117e-05,
+ "loss": 0.1977,
+ "step": 13653
+ },
+ {
+ "epoch": 37.20435967302452,
+ "grad_norm": 6.835061550140381,
+ "learning_rate": 1.446634987930831e-05,
+ "loss": 0.1496,
+ "step": 13654
+ },
+ {
+ "epoch": 37.20708446866485,
+ "grad_norm": 3.4409658908843994,
+ "learning_rate": 1.4465560279616714e-05,
+ "loss": 0.0979,
+ "step": 13655
+ },
+ {
+ "epoch": 37.20980926430518,
+ "grad_norm": 5.043013572692871,
+ "learning_rate": 1.446477064514748e-05,
+ "loss": 0.1965,
+ "step": 13656
+ },
+ {
+ "epoch": 37.212534059945504,
+ "grad_norm": 3.9123849868774414,
+ "learning_rate": 1.4463980975906753e-05,
+ "loss": 0.1882,
+ "step": 13657
+ },
+ {
+ "epoch": 37.21525885558583,
+ "grad_norm": 4.5565948486328125,
+ "learning_rate": 1.4463191271900686e-05,
+ "loss": 0.1453,
+ "step": 13658
+ },
+ {
+ "epoch": 37.217983651226156,
+ "grad_norm": 6.310123443603516,
+ "learning_rate": 1.4462401533135429e-05,
+ "loss": 0.2004,
+ "step": 13659
+ },
+ {
+ "epoch": 37.22070844686648,
+ "grad_norm": 3.9266157150268555,
+ "learning_rate": 1.4461611759617134e-05,
+ "loss": 0.1372,
+ "step": 13660
+ },
+ {
+ "epoch": 37.223433242506815,
+ "grad_norm": 3.9669339656829834,
+ "learning_rate": 1.446082195135195e-05,
+ "loss": 0.2617,
+ "step": 13661
+ },
+ {
+ "epoch": 37.22615803814714,
+ "grad_norm": 3.8446948528289795,
+ "learning_rate": 1.4460032108346029e-05,
+ "loss": 0.1778,
+ "step": 13662
+ },
+ {
+ "epoch": 37.228882833787466,
+ "grad_norm": 3.846968650817871,
+ "learning_rate": 1.445924223060552e-05,
+ "loss": 0.3532,
+ "step": 13663
+ },
+ {
+ "epoch": 37.23160762942779,
+ "grad_norm": 4.566476821899414,
+ "learning_rate": 1.4458452318136575e-05,
+ "loss": 0.1484,
+ "step": 13664
+ },
+ {
+ "epoch": 37.23433242506812,
+ "grad_norm": 4.539822578430176,
+ "learning_rate": 1.4457662370945348e-05,
+ "loss": 0.1726,
+ "step": 13665
+ },
+ {
+ "epoch": 37.237057220708444,
+ "grad_norm": 4.865803241729736,
+ "learning_rate": 1.445687238903799e-05,
+ "loss": 0.2667,
+ "step": 13666
+ },
+ {
+ "epoch": 37.23978201634878,
+ "grad_norm": 4.7285966873168945,
+ "learning_rate": 1.4456082372420652e-05,
+ "loss": 0.2853,
+ "step": 13667
+ },
+ {
+ "epoch": 37.2425068119891,
+ "grad_norm": 4.605989456176758,
+ "learning_rate": 1.4455292321099492e-05,
+ "loss": 0.3934,
+ "step": 13668
+ },
+ {
+ "epoch": 37.24523160762943,
+ "grad_norm": 4.23125696182251,
+ "learning_rate": 1.4454502235080654e-05,
+ "loss": 0.1478,
+ "step": 13669
+ },
+ {
+ "epoch": 37.247956403269754,
+ "grad_norm": 4.252155303955078,
+ "learning_rate": 1.44537121143703e-05,
+ "loss": 0.2338,
+ "step": 13670
+ },
+ {
+ "epoch": 37.25068119891008,
+ "grad_norm": 4.304054260253906,
+ "learning_rate": 1.4452921958974578e-05,
+ "loss": 0.1544,
+ "step": 13671
+ },
+ {
+ "epoch": 37.253405994550405,
+ "grad_norm": 4.438971996307373,
+ "learning_rate": 1.4452131768899641e-05,
+ "loss": 0.1654,
+ "step": 13672
+ },
+ {
+ "epoch": 37.25613079019074,
+ "grad_norm": 8.163948059082031,
+ "learning_rate": 1.4451341544151647e-05,
+ "loss": 0.2383,
+ "step": 13673
+ },
+ {
+ "epoch": 37.258855585831064,
+ "grad_norm": 4.170433521270752,
+ "learning_rate": 1.4450551284736748e-05,
+ "loss": 0.2533,
+ "step": 13674
+ },
+ {
+ "epoch": 37.26158038147139,
+ "grad_norm": 5.544373512268066,
+ "learning_rate": 1.4449760990661098e-05,
+ "loss": 0.1784,
+ "step": 13675
+ },
+ {
+ "epoch": 37.264305177111716,
+ "grad_norm": 4.4691877365112305,
+ "learning_rate": 1.4448970661930854e-05,
+ "loss": 0.168,
+ "step": 13676
+ },
+ {
+ "epoch": 37.26702997275204,
+ "grad_norm": 3.2681074142456055,
+ "learning_rate": 1.4448180298552168e-05,
+ "loss": 0.1872,
+ "step": 13677
+ },
+ {
+ "epoch": 37.26975476839237,
+ "grad_norm": 3.8646252155303955,
+ "learning_rate": 1.4447389900531199e-05,
+ "loss": 0.1688,
+ "step": 13678
+ },
+ {
+ "epoch": 37.2724795640327,
+ "grad_norm": 6.324230194091797,
+ "learning_rate": 1.44465994678741e-05,
+ "loss": 0.1451,
+ "step": 13679
+ },
+ {
+ "epoch": 37.275204359673026,
+ "grad_norm": 6.928562164306641,
+ "learning_rate": 1.4445809000587027e-05,
+ "loss": 0.2151,
+ "step": 13680
+ },
+ {
+ "epoch": 37.27792915531335,
+ "grad_norm": 4.030706405639648,
+ "learning_rate": 1.4445018498676135e-05,
+ "loss": 0.3316,
+ "step": 13681
+ },
+ {
+ "epoch": 37.28065395095368,
+ "grad_norm": 4.580703258514404,
+ "learning_rate": 1.4444227962147586e-05,
+ "loss": 0.3085,
+ "step": 13682
+ },
+ {
+ "epoch": 37.283378746594,
+ "grad_norm": 3.975982427597046,
+ "learning_rate": 1.4443437391007531e-05,
+ "loss": 0.1449,
+ "step": 13683
+ },
+ {
+ "epoch": 37.28610354223433,
+ "grad_norm": 3.9343576431274414,
+ "learning_rate": 1.4442646785262126e-05,
+ "loss": 0.3039,
+ "step": 13684
+ },
+ {
+ "epoch": 37.28882833787466,
+ "grad_norm": 3.597008228302002,
+ "learning_rate": 1.4441856144917535e-05,
+ "loss": 0.3415,
+ "step": 13685
+ },
+ {
+ "epoch": 37.29155313351499,
+ "grad_norm": 4.057291030883789,
+ "learning_rate": 1.4441065469979908e-05,
+ "loss": 0.2201,
+ "step": 13686
+ },
+ {
+ "epoch": 37.294277929155314,
+ "grad_norm": 4.215564727783203,
+ "learning_rate": 1.4440274760455406e-05,
+ "loss": 0.216,
+ "step": 13687
+ },
+ {
+ "epoch": 37.29700272479564,
+ "grad_norm": 5.980860710144043,
+ "learning_rate": 1.443948401635019e-05,
+ "loss": 0.1419,
+ "step": 13688
+ },
+ {
+ "epoch": 37.299727520435965,
+ "grad_norm": 5.0181732177734375,
+ "learning_rate": 1.4438693237670412e-05,
+ "loss": 0.2537,
+ "step": 13689
+ },
+ {
+ "epoch": 37.30245231607629,
+ "grad_norm": 3.929222822189331,
+ "learning_rate": 1.4437902424422236e-05,
+ "loss": 0.3422,
+ "step": 13690
+ },
+ {
+ "epoch": 37.305177111716624,
+ "grad_norm": 3.7822935581207275,
+ "learning_rate": 1.443711157661182e-05,
+ "loss": 0.1243,
+ "step": 13691
+ },
+ {
+ "epoch": 37.30790190735695,
+ "grad_norm": 5.744969844818115,
+ "learning_rate": 1.4436320694245318e-05,
+ "loss": 0.122,
+ "step": 13692
+ },
+ {
+ "epoch": 37.310626702997276,
+ "grad_norm": 3.0580809116363525,
+ "learning_rate": 1.4435529777328895e-05,
+ "loss": 0.1142,
+ "step": 13693
+ },
+ {
+ "epoch": 37.3133514986376,
+ "grad_norm": 3.1427297592163086,
+ "learning_rate": 1.443473882586871e-05,
+ "loss": 0.1309,
+ "step": 13694
+ },
+ {
+ "epoch": 37.31607629427793,
+ "grad_norm": 4.490758895874023,
+ "learning_rate": 1.4433947839870924e-05,
+ "loss": 0.1434,
+ "step": 13695
+ },
+ {
+ "epoch": 37.31880108991825,
+ "grad_norm": 4.156644821166992,
+ "learning_rate": 1.443315681934169e-05,
+ "loss": 0.1656,
+ "step": 13696
+ },
+ {
+ "epoch": 37.321525885558586,
+ "grad_norm": 4.069386005401611,
+ "learning_rate": 1.4432365764287175e-05,
+ "loss": 0.2103,
+ "step": 13697
+ },
+ {
+ "epoch": 37.32425068119891,
+ "grad_norm": 4.8825459480285645,
+ "learning_rate": 1.4431574674713538e-05,
+ "loss": 0.2134,
+ "step": 13698
+ },
+ {
+ "epoch": 37.32697547683924,
+ "grad_norm": 3.769958734512329,
+ "learning_rate": 1.443078355062694e-05,
+ "loss": 0.1019,
+ "step": 13699
+ },
+ {
+ "epoch": 37.32970027247956,
+ "grad_norm": 3.573072671890259,
+ "learning_rate": 1.4429992392033545e-05,
+ "loss": 0.2784,
+ "step": 13700
+ },
+ {
+ "epoch": 37.33242506811989,
+ "grad_norm": 5.397806167602539,
+ "learning_rate": 1.4429201198939508e-05,
+ "loss": 0.174,
+ "step": 13701
+ },
+ {
+ "epoch": 37.335149863760215,
+ "grad_norm": 3.7042064666748047,
+ "learning_rate": 1.4428409971350994e-05,
+ "loss": 0.1967,
+ "step": 13702
+ },
+ {
+ "epoch": 37.33787465940055,
+ "grad_norm": 4.118574142456055,
+ "learning_rate": 1.442761870927417e-05,
+ "loss": 0.1065,
+ "step": 13703
+ },
+ {
+ "epoch": 37.34059945504087,
+ "grad_norm": 3.849665641784668,
+ "learning_rate": 1.4426827412715192e-05,
+ "loss": 0.24,
+ "step": 13704
+ },
+ {
+ "epoch": 37.3433242506812,
+ "grad_norm": 3.537505626678467,
+ "learning_rate": 1.4426036081680223e-05,
+ "loss": 0.1147,
+ "step": 13705
+ },
+ {
+ "epoch": 37.346049046321525,
+ "grad_norm": 3.852710008621216,
+ "learning_rate": 1.4425244716175431e-05,
+ "loss": 0.3307,
+ "step": 13706
+ },
+ {
+ "epoch": 37.34877384196185,
+ "grad_norm": 6.9562273025512695,
+ "learning_rate": 1.4424453316206974e-05,
+ "loss": 0.2578,
+ "step": 13707
+ },
+ {
+ "epoch": 37.35149863760218,
+ "grad_norm": 4.046548366546631,
+ "learning_rate": 1.4423661881781014e-05,
+ "loss": 0.1313,
+ "step": 13708
+ },
+ {
+ "epoch": 37.35422343324251,
+ "grad_norm": 3.6552982330322266,
+ "learning_rate": 1.442287041290372e-05,
+ "loss": 0.1704,
+ "step": 13709
+ },
+ {
+ "epoch": 37.356948228882835,
+ "grad_norm": 3.6742947101593018,
+ "learning_rate": 1.4422078909581253e-05,
+ "loss": 0.1478,
+ "step": 13710
+ },
+ {
+ "epoch": 37.35967302452316,
+ "grad_norm": 4.378448963165283,
+ "learning_rate": 1.4421287371819781e-05,
+ "loss": 0.1172,
+ "step": 13711
+ },
+ {
+ "epoch": 37.36239782016349,
+ "grad_norm": 3.7022619247436523,
+ "learning_rate": 1.442049579962546e-05,
+ "loss": 0.2186,
+ "step": 13712
+ },
+ {
+ "epoch": 37.36512261580381,
+ "grad_norm": 3.3775057792663574,
+ "learning_rate": 1.4419704193004464e-05,
+ "loss": 0.1882,
+ "step": 13713
+ },
+ {
+ "epoch": 37.36784741144414,
+ "grad_norm": 3.68023681640625,
+ "learning_rate": 1.4418912551962953e-05,
+ "loss": 0.1214,
+ "step": 13714
+ },
+ {
+ "epoch": 37.37057220708447,
+ "grad_norm": 4.048430442810059,
+ "learning_rate": 1.4418120876507093e-05,
+ "loss": 0.0951,
+ "step": 13715
+ },
+ {
+ "epoch": 37.3732970027248,
+ "grad_norm": 4.296184539794922,
+ "learning_rate": 1.441732916664305e-05,
+ "loss": 0.2137,
+ "step": 13716
+ },
+ {
+ "epoch": 37.37602179836512,
+ "grad_norm": 3.453005313873291,
+ "learning_rate": 1.441653742237699e-05,
+ "loss": 0.2222,
+ "step": 13717
+ },
+ {
+ "epoch": 37.37874659400545,
+ "grad_norm": 3.7848997116088867,
+ "learning_rate": 1.4415745643715075e-05,
+ "loss": 0.1671,
+ "step": 13718
+ },
+ {
+ "epoch": 37.381471389645775,
+ "grad_norm": 3.7209784984588623,
+ "learning_rate": 1.4414953830663478e-05,
+ "loss": 0.3453,
+ "step": 13719
+ },
+ {
+ "epoch": 37.3841961852861,
+ "grad_norm": 4.017925262451172,
+ "learning_rate": 1.4414161983228362e-05,
+ "loss": 0.1949,
+ "step": 13720
+ },
+ {
+ "epoch": 37.38692098092643,
+ "grad_norm": 2.9964468479156494,
+ "learning_rate": 1.4413370101415898e-05,
+ "loss": 0.1021,
+ "step": 13721
+ },
+ {
+ "epoch": 37.38964577656676,
+ "grad_norm": 3.8376822471618652,
+ "learning_rate": 1.4412578185232246e-05,
+ "loss": 0.2884,
+ "step": 13722
+ },
+ {
+ "epoch": 37.392370572207085,
+ "grad_norm": 3.443154811859131,
+ "learning_rate": 1.441178623468358e-05,
+ "loss": 0.2426,
+ "step": 13723
+ },
+ {
+ "epoch": 37.39509536784741,
+ "grad_norm": 4.092416286468506,
+ "learning_rate": 1.441099424977606e-05,
+ "loss": 0.1581,
+ "step": 13724
+ },
+ {
+ "epoch": 37.39782016348774,
+ "grad_norm": 3.992004632949829,
+ "learning_rate": 1.4410202230515865e-05,
+ "loss": 0.1026,
+ "step": 13725
+ },
+ {
+ "epoch": 37.40054495912806,
+ "grad_norm": 3.4627315998077393,
+ "learning_rate": 1.440941017690915e-05,
+ "loss": 0.1009,
+ "step": 13726
+ },
+ {
+ "epoch": 37.403269754768395,
+ "grad_norm": 4.61834192276001,
+ "learning_rate": 1.4408618088962096e-05,
+ "loss": 0.1855,
+ "step": 13727
+ },
+ {
+ "epoch": 37.40599455040872,
+ "grad_norm": 3.7064290046691895,
+ "learning_rate": 1.4407825966680862e-05,
+ "loss": 0.1368,
+ "step": 13728
+ },
+ {
+ "epoch": 37.40871934604905,
+ "grad_norm": 4.63818359375,
+ "learning_rate": 1.4407033810071624e-05,
+ "loss": 0.1777,
+ "step": 13729
+ },
+ {
+ "epoch": 37.41144414168937,
+ "grad_norm": 3.207643508911133,
+ "learning_rate": 1.4406241619140549e-05,
+ "loss": 0.097,
+ "step": 13730
+ },
+ {
+ "epoch": 37.4141689373297,
+ "grad_norm": 3.8176615238189697,
+ "learning_rate": 1.4405449393893807e-05,
+ "loss": 0.1744,
+ "step": 13731
+ },
+ {
+ "epoch": 37.416893732970024,
+ "grad_norm": 5.941243648529053,
+ "learning_rate": 1.4404657134337562e-05,
+ "loss": 0.1633,
+ "step": 13732
+ },
+ {
+ "epoch": 37.41961852861036,
+ "grad_norm": 3.400968074798584,
+ "learning_rate": 1.4403864840477994e-05,
+ "loss": 0.3347,
+ "step": 13733
+ },
+ {
+ "epoch": 37.42234332425068,
+ "grad_norm": 4.227228164672852,
+ "learning_rate": 1.4403072512321266e-05,
+ "loss": 0.1515,
+ "step": 13734
+ },
+ {
+ "epoch": 37.42506811989101,
+ "grad_norm": 3.9702014923095703,
+ "learning_rate": 1.440228014987355e-05,
+ "loss": 0.167,
+ "step": 13735
+ },
+ {
+ "epoch": 37.427792915531334,
+ "grad_norm": 4.838196277618408,
+ "learning_rate": 1.440148775314102e-05,
+ "loss": 0.2324,
+ "step": 13736
+ },
+ {
+ "epoch": 37.43051771117166,
+ "grad_norm": 4.762652397155762,
+ "learning_rate": 1.4400695322129846e-05,
+ "loss": 0.2298,
+ "step": 13737
+ },
+ {
+ "epoch": 37.433242506811986,
+ "grad_norm": 3.933837890625,
+ "learning_rate": 1.4399902856846195e-05,
+ "loss": 0.1418,
+ "step": 13738
+ },
+ {
+ "epoch": 37.43596730245232,
+ "grad_norm": 3.534647226333618,
+ "learning_rate": 1.4399110357296244e-05,
+ "loss": 0.1541,
+ "step": 13739
+ },
+ {
+ "epoch": 37.438692098092645,
+ "grad_norm": 3.502084732055664,
+ "learning_rate": 1.4398317823486164e-05,
+ "loss": 0.1309,
+ "step": 13740
+ },
+ {
+ "epoch": 37.44141689373297,
+ "grad_norm": 4.087728023529053,
+ "learning_rate": 1.4397525255422125e-05,
+ "loss": 0.2283,
+ "step": 13741
+ },
+ {
+ "epoch": 37.444141689373296,
+ "grad_norm": 4.141765594482422,
+ "learning_rate": 1.4396732653110302e-05,
+ "loss": 0.1299,
+ "step": 13742
+ },
+ {
+ "epoch": 37.44686648501362,
+ "grad_norm": 19.689311981201172,
+ "learning_rate": 1.4395940016556867e-05,
+ "loss": 0.2396,
+ "step": 13743
+ },
+ {
+ "epoch": 37.44959128065395,
+ "grad_norm": 4.546355247497559,
+ "learning_rate": 1.4395147345767988e-05,
+ "loss": 0.3043,
+ "step": 13744
+ },
+ {
+ "epoch": 37.45231607629428,
+ "grad_norm": 3.9201555252075195,
+ "learning_rate": 1.4394354640749848e-05,
+ "loss": 0.1183,
+ "step": 13745
+ },
+ {
+ "epoch": 37.45504087193461,
+ "grad_norm": 3.5579679012298584,
+ "learning_rate": 1.4393561901508613e-05,
+ "loss": 0.1549,
+ "step": 13746
+ },
+ {
+ "epoch": 37.45776566757493,
+ "grad_norm": 4.017496109008789,
+ "learning_rate": 1.439276912805046e-05,
+ "loss": 0.1315,
+ "step": 13747
+ },
+ {
+ "epoch": 37.46049046321526,
+ "grad_norm": 4.008517265319824,
+ "learning_rate": 1.4391976320381562e-05,
+ "loss": 0.113,
+ "step": 13748
+ },
+ {
+ "epoch": 37.463215258855584,
+ "grad_norm": 4.437317848205566,
+ "learning_rate": 1.4391183478508098e-05,
+ "loss": 0.2135,
+ "step": 13749
+ },
+ {
+ "epoch": 37.46594005449591,
+ "grad_norm": 4.764256477355957,
+ "learning_rate": 1.4390390602436232e-05,
+ "loss": 0.1275,
+ "step": 13750
+ },
+ {
+ "epoch": 37.46866485013624,
+ "grad_norm": 3.542659044265747,
+ "learning_rate": 1.438959769217215e-05,
+ "loss": 0.2563,
+ "step": 13751
+ },
+ {
+ "epoch": 37.47138964577657,
+ "grad_norm": 4.488884449005127,
+ "learning_rate": 1.4388804747722018e-05,
+ "loss": 0.1601,
+ "step": 13752
+ },
+ {
+ "epoch": 37.474114441416894,
+ "grad_norm": 4.110666751861572,
+ "learning_rate": 1.4388011769092018e-05,
+ "loss": 0.2171,
+ "step": 13753
+ },
+ {
+ "epoch": 37.47683923705722,
+ "grad_norm": 6.178955078125,
+ "learning_rate": 1.4387218756288323e-05,
+ "loss": 0.3044,
+ "step": 13754
+ },
+ {
+ "epoch": 37.479564032697546,
+ "grad_norm": 3.695258855819702,
+ "learning_rate": 1.4386425709317113e-05,
+ "loss": 0.1396,
+ "step": 13755
+ },
+ {
+ "epoch": 37.48228882833787,
+ "grad_norm": 5.342553615570068,
+ "learning_rate": 1.4385632628184558e-05,
+ "loss": 0.1424,
+ "step": 13756
+ },
+ {
+ "epoch": 37.485013623978205,
+ "grad_norm": 3.995234966278076,
+ "learning_rate": 1.4384839512896837e-05,
+ "loss": 0.1711,
+ "step": 13757
+ },
+ {
+ "epoch": 37.48773841961853,
+ "grad_norm": 3.840933322906494,
+ "learning_rate": 1.4384046363460129e-05,
+ "loss": 0.1574,
+ "step": 13758
+ },
+ {
+ "epoch": 37.490463215258856,
+ "grad_norm": 4.435140132904053,
+ "learning_rate": 1.4383253179880607e-05,
+ "loss": 0.397,
+ "step": 13759
+ },
+ {
+ "epoch": 37.49318801089918,
+ "grad_norm": 3.5478508472442627,
+ "learning_rate": 1.4382459962164452e-05,
+ "loss": 0.1445,
+ "step": 13760
+ },
+ {
+ "epoch": 37.49591280653951,
+ "grad_norm": 9.216634750366211,
+ "learning_rate": 1.4381666710317837e-05,
+ "loss": 0.1305,
+ "step": 13761
+ },
+ {
+ "epoch": 37.49863760217983,
+ "grad_norm": 3.4724879264831543,
+ "learning_rate": 1.4380873424346945e-05,
+ "loss": 0.1102,
+ "step": 13762
+ },
+ {
+ "epoch": 37.50136239782017,
+ "grad_norm": 4.251380920410156,
+ "learning_rate": 1.438008010425795e-05,
+ "loss": 0.2094,
+ "step": 13763
+ },
+ {
+ "epoch": 37.50408719346049,
+ "grad_norm": 3.5181286334991455,
+ "learning_rate": 1.4379286750057033e-05,
+ "loss": 0.1063,
+ "step": 13764
+ },
+ {
+ "epoch": 37.50681198910082,
+ "grad_norm": 3.8135831356048584,
+ "learning_rate": 1.4378493361750374e-05,
+ "loss": 0.2671,
+ "step": 13765
+ },
+ {
+ "epoch": 37.509536784741144,
+ "grad_norm": 4.358130931854248,
+ "learning_rate": 1.4377699939344148e-05,
+ "loss": 0.1694,
+ "step": 13766
+ },
+ {
+ "epoch": 37.51226158038147,
+ "grad_norm": 4.068714618682861,
+ "learning_rate": 1.4376906482844534e-05,
+ "loss": 0.1441,
+ "step": 13767
+ },
+ {
+ "epoch": 37.514986376021795,
+ "grad_norm": 3.616333484649658,
+ "learning_rate": 1.4376112992257714e-05,
+ "loss": 0.1491,
+ "step": 13768
+ },
+ {
+ "epoch": 37.51771117166213,
+ "grad_norm": 4.567919731140137,
+ "learning_rate": 1.4375319467589868e-05,
+ "loss": 0.2035,
+ "step": 13769
+ },
+ {
+ "epoch": 37.520435967302454,
+ "grad_norm": 3.6051013469696045,
+ "learning_rate": 1.4374525908847175e-05,
+ "loss": 0.1492,
+ "step": 13770
+ },
+ {
+ "epoch": 37.52316076294278,
+ "grad_norm": 6.12717866897583,
+ "learning_rate": 1.4373732316035813e-05,
+ "loss": 0.1595,
+ "step": 13771
+ },
+ {
+ "epoch": 37.525885558583106,
+ "grad_norm": 4.953482627868652,
+ "learning_rate": 1.4372938689161969e-05,
+ "loss": 0.2021,
+ "step": 13772
+ },
+ {
+ "epoch": 37.52861035422343,
+ "grad_norm": 4.247292995452881,
+ "learning_rate": 1.4372145028231814e-05,
+ "loss": 0.1378,
+ "step": 13773
+ },
+ {
+ "epoch": 37.53133514986376,
+ "grad_norm": 4.126422882080078,
+ "learning_rate": 1.4371351333251535e-05,
+ "loss": 0.1531,
+ "step": 13774
+ },
+ {
+ "epoch": 37.53405994550409,
+ "grad_norm": 4.7410502433776855,
+ "learning_rate": 1.4370557604227316e-05,
+ "loss": 0.2419,
+ "step": 13775
+ },
+ {
+ "epoch": 37.536784741144416,
+ "grad_norm": 3.9142682552337646,
+ "learning_rate": 1.4369763841165331e-05,
+ "loss": 0.1355,
+ "step": 13776
+ },
+ {
+ "epoch": 37.53950953678474,
+ "grad_norm": 4.724118709564209,
+ "learning_rate": 1.4368970044071768e-05,
+ "loss": 0.1264,
+ "step": 13777
+ },
+ {
+ "epoch": 37.54223433242507,
+ "grad_norm": 3.7173779010772705,
+ "learning_rate": 1.4368176212952807e-05,
+ "loss": 0.1532,
+ "step": 13778
+ },
+ {
+ "epoch": 37.54495912806539,
+ "grad_norm": 4.2904372215271,
+ "learning_rate": 1.4367382347814628e-05,
+ "loss": 0.131,
+ "step": 13779
+ },
+ {
+ "epoch": 37.54768392370572,
+ "grad_norm": 3.4420247077941895,
+ "learning_rate": 1.436658844866342e-05,
+ "loss": 0.1827,
+ "step": 13780
+ },
+ {
+ "epoch": 37.55040871934605,
+ "grad_norm": 4.2069268226623535,
+ "learning_rate": 1.4365794515505359e-05,
+ "loss": 0.1249,
+ "step": 13781
+ },
+ {
+ "epoch": 37.55313351498638,
+ "grad_norm": 3.7238781452178955,
+ "learning_rate": 1.4365000548346633e-05,
+ "loss": 0.1921,
+ "step": 13782
+ },
+ {
+ "epoch": 37.555858310626704,
+ "grad_norm": 4.958370685577393,
+ "learning_rate": 1.436420654719342e-05,
+ "loss": 0.172,
+ "step": 13783
+ },
+ {
+ "epoch": 37.55858310626703,
+ "grad_norm": 4.464888572692871,
+ "learning_rate": 1.4363412512051908e-05,
+ "loss": 0.1357,
+ "step": 13784
+ },
+ {
+ "epoch": 37.561307901907355,
+ "grad_norm": 4.024531841278076,
+ "learning_rate": 1.436261844292828e-05,
+ "loss": 0.2273,
+ "step": 13785
+ },
+ {
+ "epoch": 37.56403269754768,
+ "grad_norm": 3.9867231845855713,
+ "learning_rate": 1.4361824339828718e-05,
+ "loss": 0.1846,
+ "step": 13786
+ },
+ {
+ "epoch": 37.566757493188014,
+ "grad_norm": 3.882692337036133,
+ "learning_rate": 1.4361030202759412e-05,
+ "loss": 0.2549,
+ "step": 13787
+ },
+ {
+ "epoch": 37.56948228882834,
+ "grad_norm": 3.575737476348877,
+ "learning_rate": 1.4360236031726543e-05,
+ "loss": 0.1256,
+ "step": 13788
+ },
+ {
+ "epoch": 37.572207084468666,
+ "grad_norm": 4.325496673583984,
+ "learning_rate": 1.4359441826736294e-05,
+ "loss": 0.2311,
+ "step": 13789
+ },
+ {
+ "epoch": 37.57493188010899,
+ "grad_norm": 3.936809778213501,
+ "learning_rate": 1.4358647587794851e-05,
+ "loss": 0.1577,
+ "step": 13790
+ },
+ {
+ "epoch": 37.57765667574932,
+ "grad_norm": 2.886841058731079,
+ "learning_rate": 1.4357853314908403e-05,
+ "loss": 0.1526,
+ "step": 13791
+ },
+ {
+ "epoch": 37.58038147138964,
+ "grad_norm": 3.5534958839416504,
+ "learning_rate": 1.4357059008083133e-05,
+ "loss": 0.2715,
+ "step": 13792
+ },
+ {
+ "epoch": 37.583106267029976,
+ "grad_norm": 3.126446485519409,
+ "learning_rate": 1.4356264667325231e-05,
+ "loss": 0.1454,
+ "step": 13793
+ },
+ {
+ "epoch": 37.5858310626703,
+ "grad_norm": 4.307419776916504,
+ "learning_rate": 1.4355470292640879e-05,
+ "loss": 0.1996,
+ "step": 13794
+ },
+ {
+ "epoch": 37.58855585831063,
+ "grad_norm": 3.7611618041992188,
+ "learning_rate": 1.4354675884036257e-05,
+ "loss": 0.195,
+ "step": 13795
+ },
+ {
+ "epoch": 37.59128065395095,
+ "grad_norm": 5.082409381866455,
+ "learning_rate": 1.4353881441517568e-05,
+ "loss": 0.2603,
+ "step": 13796
+ },
+ {
+ "epoch": 37.59400544959128,
+ "grad_norm": 3.7108631134033203,
+ "learning_rate": 1.4353086965090985e-05,
+ "loss": 0.1178,
+ "step": 13797
+ },
+ {
+ "epoch": 37.596730245231605,
+ "grad_norm": 3.80521297454834,
+ "learning_rate": 1.4352292454762705e-05,
+ "loss": 0.1205,
+ "step": 13798
+ },
+ {
+ "epoch": 37.59945504087194,
+ "grad_norm": 4.373488903045654,
+ "learning_rate": 1.435149791053891e-05,
+ "loss": 0.2127,
+ "step": 13799
+ },
+ {
+ "epoch": 37.60217983651226,
+ "grad_norm": 4.658786296844482,
+ "learning_rate": 1.4350703332425791e-05,
+ "loss": 0.234,
+ "step": 13800
+ },
+ {
+ "epoch": 37.60490463215259,
+ "grad_norm": 4.0365891456604,
+ "learning_rate": 1.4349908720429531e-05,
+ "loss": 0.3097,
+ "step": 13801
+ },
+ {
+ "epoch": 37.607629427792915,
+ "grad_norm": 17.371225357055664,
+ "learning_rate": 1.4349114074556326e-05,
+ "loss": 0.2237,
+ "step": 13802
+ },
+ {
+ "epoch": 37.61035422343324,
+ "grad_norm": 3.6933209896087646,
+ "learning_rate": 1.4348319394812355e-05,
+ "loss": 0.1242,
+ "step": 13803
+ },
+ {
+ "epoch": 37.61307901907357,
+ "grad_norm": 4.228282451629639,
+ "learning_rate": 1.4347524681203817e-05,
+ "loss": 0.14,
+ "step": 13804
+ },
+ {
+ "epoch": 37.6158038147139,
+ "grad_norm": 3.784743547439575,
+ "learning_rate": 1.4346729933736894e-05,
+ "loss": 0.19,
+ "step": 13805
+ },
+ {
+ "epoch": 37.618528610354225,
+ "grad_norm": 3.968752384185791,
+ "learning_rate": 1.434593515241778e-05,
+ "loss": 0.2292,
+ "step": 13806
+ },
+ {
+ "epoch": 37.62125340599455,
+ "grad_norm": 4.894870758056641,
+ "learning_rate": 1.4345140337252663e-05,
+ "loss": 0.328,
+ "step": 13807
+ },
+ {
+ "epoch": 37.62397820163488,
+ "grad_norm": 4.501951217651367,
+ "learning_rate": 1.4344345488247733e-05,
+ "loss": 0.1818,
+ "step": 13808
+ },
+ {
+ "epoch": 37.6267029972752,
+ "grad_norm": 4.234055519104004,
+ "learning_rate": 1.4343550605409178e-05,
+ "loss": 0.263,
+ "step": 13809
+ },
+ {
+ "epoch": 37.62942779291553,
+ "grad_norm": 3.566692590713501,
+ "learning_rate": 1.4342755688743192e-05,
+ "loss": 0.1215,
+ "step": 13810
+ },
+ {
+ "epoch": 37.63215258855586,
+ "grad_norm": 3.8914411067962646,
+ "learning_rate": 1.4341960738255965e-05,
+ "loss": 0.2113,
+ "step": 13811
+ },
+ {
+ "epoch": 37.63487738419619,
+ "grad_norm": 3.8161048889160156,
+ "learning_rate": 1.4341165753953688e-05,
+ "loss": 0.114,
+ "step": 13812
+ },
+ {
+ "epoch": 37.63760217983651,
+ "grad_norm": 4.348979949951172,
+ "learning_rate": 1.4340370735842548e-05,
+ "loss": 0.2063,
+ "step": 13813
+ },
+ {
+ "epoch": 37.64032697547684,
+ "grad_norm": 11.030738830566406,
+ "learning_rate": 1.4339575683928746e-05,
+ "loss": 0.1227,
+ "step": 13814
+ },
+ {
+ "epoch": 37.643051771117165,
+ "grad_norm": 3.3263964653015137,
+ "learning_rate": 1.4338780598218465e-05,
+ "loss": 0.2663,
+ "step": 13815
+ },
+ {
+ "epoch": 37.64577656675749,
+ "grad_norm": 4.067708969116211,
+ "learning_rate": 1.4337985478717902e-05,
+ "loss": 0.1272,
+ "step": 13816
+ },
+ {
+ "epoch": 37.64850136239782,
+ "grad_norm": 4.712686538696289,
+ "learning_rate": 1.4337190325433246e-05,
+ "loss": 0.1427,
+ "step": 13817
+ },
+ {
+ "epoch": 37.65122615803815,
+ "grad_norm": 5.165581226348877,
+ "learning_rate": 1.4336395138370694e-05,
+ "loss": 0.1489,
+ "step": 13818
+ },
+ {
+ "epoch": 37.653950953678475,
+ "grad_norm": 3.2760162353515625,
+ "learning_rate": 1.4335599917536432e-05,
+ "loss": 0.1301,
+ "step": 13819
+ },
+ {
+ "epoch": 37.6566757493188,
+ "grad_norm": 4.683065414428711,
+ "learning_rate": 1.4334804662936661e-05,
+ "loss": 0.2448,
+ "step": 13820
+ },
+ {
+ "epoch": 37.65940054495913,
+ "grad_norm": 3.946953535079956,
+ "learning_rate": 1.4334009374577566e-05,
+ "loss": 0.1366,
+ "step": 13821
+ },
+ {
+ "epoch": 37.66212534059945,
+ "grad_norm": 7.503315448760986,
+ "learning_rate": 1.433321405246535e-05,
+ "loss": 0.1275,
+ "step": 13822
+ },
+ {
+ "epoch": 37.664850136239785,
+ "grad_norm": 3.595456123352051,
+ "learning_rate": 1.43324186966062e-05,
+ "loss": 0.1833,
+ "step": 13823
+ },
+ {
+ "epoch": 37.66757493188011,
+ "grad_norm": 4.496795654296875,
+ "learning_rate": 1.4331623307006316e-05,
+ "loss": 0.191,
+ "step": 13824
+ },
+ {
+ "epoch": 37.67029972752044,
+ "grad_norm": 3.5580058097839355,
+ "learning_rate": 1.4330827883671885e-05,
+ "loss": 0.1289,
+ "step": 13825
+ },
+ {
+ "epoch": 37.67302452316076,
+ "grad_norm": 3.644401788711548,
+ "learning_rate": 1.433003242660911e-05,
+ "loss": 0.2169,
+ "step": 13826
+ },
+ {
+ "epoch": 37.67574931880109,
+ "grad_norm": 4.263618469238281,
+ "learning_rate": 1.432923693582418e-05,
+ "loss": 0.2112,
+ "step": 13827
+ },
+ {
+ "epoch": 37.678474114441414,
+ "grad_norm": 4.403650283813477,
+ "learning_rate": 1.4328441411323292e-05,
+ "loss": 0.2536,
+ "step": 13828
+ },
+ {
+ "epoch": 37.68119891008175,
+ "grad_norm": 3.3255109786987305,
+ "learning_rate": 1.432764585311264e-05,
+ "loss": 0.1443,
+ "step": 13829
+ },
+ {
+ "epoch": 37.68392370572207,
+ "grad_norm": 4.5683112144470215,
+ "learning_rate": 1.4326850261198422e-05,
+ "loss": 0.3364,
+ "step": 13830
+ },
+ {
+ "epoch": 37.6866485013624,
+ "grad_norm": 4.293881893157959,
+ "learning_rate": 1.4326054635586833e-05,
+ "loss": 0.1157,
+ "step": 13831
+ },
+ {
+ "epoch": 37.689373297002724,
+ "grad_norm": 4.963428497314453,
+ "learning_rate": 1.4325258976284073e-05,
+ "loss": 0.2247,
+ "step": 13832
+ },
+ {
+ "epoch": 37.69209809264305,
+ "grad_norm": 4.392552375793457,
+ "learning_rate": 1.4324463283296332e-05,
+ "loss": 0.1328,
+ "step": 13833
+ },
+ {
+ "epoch": 37.694822888283376,
+ "grad_norm": 3.4991016387939453,
+ "learning_rate": 1.4323667556629812e-05,
+ "loss": 0.1227,
+ "step": 13834
+ },
+ {
+ "epoch": 37.69754768392371,
+ "grad_norm": 3.577772378921509,
+ "learning_rate": 1.4322871796290707e-05,
+ "loss": 0.2072,
+ "step": 13835
+ },
+ {
+ "epoch": 37.700272479564035,
+ "grad_norm": 3.821047067642212,
+ "learning_rate": 1.4322076002285216e-05,
+ "loss": 0.2069,
+ "step": 13836
+ },
+ {
+ "epoch": 37.70299727520436,
+ "grad_norm": 4.175758361816406,
+ "learning_rate": 1.4321280174619538e-05,
+ "loss": 0.226,
+ "step": 13837
+ },
+ {
+ "epoch": 37.705722070844686,
+ "grad_norm": 4.468236446380615,
+ "learning_rate": 1.4320484313299865e-05,
+ "loss": 0.1407,
+ "step": 13838
+ },
+ {
+ "epoch": 37.70844686648501,
+ "grad_norm": 3.8516058921813965,
+ "learning_rate": 1.43196884183324e-05,
+ "loss": 0.1234,
+ "step": 13839
+ },
+ {
+ "epoch": 37.71117166212534,
+ "grad_norm": 3.6252613067626953,
+ "learning_rate": 1.4318892489723345e-05,
+ "loss": 0.2804,
+ "step": 13840
+ },
+ {
+ "epoch": 37.71389645776567,
+ "grad_norm": 3.9538302421569824,
+ "learning_rate": 1.431809652747889e-05,
+ "loss": 0.1388,
+ "step": 13841
+ },
+ {
+ "epoch": 37.716621253406,
+ "grad_norm": 3.8053689002990723,
+ "learning_rate": 1.4317300531605241e-05,
+ "loss": 0.2345,
+ "step": 13842
+ },
+ {
+ "epoch": 37.71934604904632,
+ "grad_norm": 3.765848398208618,
+ "learning_rate": 1.4316504502108592e-05,
+ "loss": 0.184,
+ "step": 13843
+ },
+ {
+ "epoch": 37.72207084468665,
+ "grad_norm": 4.060519695281982,
+ "learning_rate": 1.4315708438995148e-05,
+ "loss": 0.1566,
+ "step": 13844
+ },
+ {
+ "epoch": 37.724795640326974,
+ "grad_norm": 3.5917482376098633,
+ "learning_rate": 1.4314912342271103e-05,
+ "loss": 0.2162,
+ "step": 13845
+ },
+ {
+ "epoch": 37.7275204359673,
+ "grad_norm": 3.81589412689209,
+ "learning_rate": 1.431411621194266e-05,
+ "loss": 0.1781,
+ "step": 13846
+ },
+ {
+ "epoch": 37.73024523160763,
+ "grad_norm": 4.5712151527404785,
+ "learning_rate": 1.431332004801602e-05,
+ "loss": 0.1571,
+ "step": 13847
+ },
+ {
+ "epoch": 37.73297002724796,
+ "grad_norm": 4.5237040519714355,
+ "learning_rate": 1.431252385049738e-05,
+ "loss": 0.1237,
+ "step": 13848
+ },
+ {
+ "epoch": 37.735694822888284,
+ "grad_norm": 3.8141791820526123,
+ "learning_rate": 1.4311727619392944e-05,
+ "loss": 0.2249,
+ "step": 13849
+ },
+ {
+ "epoch": 37.73841961852861,
+ "grad_norm": 3.922767162322998,
+ "learning_rate": 1.4310931354708913e-05,
+ "loss": 0.2817,
+ "step": 13850
+ },
+ {
+ "epoch": 37.741144414168936,
+ "grad_norm": 3.6525144577026367,
+ "learning_rate": 1.4310135056451486e-05,
+ "loss": 0.1879,
+ "step": 13851
+ },
+ {
+ "epoch": 37.74386920980926,
+ "grad_norm": 6.109289646148682,
+ "learning_rate": 1.4309338724626869e-05,
+ "loss": 0.4416,
+ "step": 13852
+ },
+ {
+ "epoch": 37.746594005449595,
+ "grad_norm": 4.089931488037109,
+ "learning_rate": 1.4308542359241257e-05,
+ "loss": 0.3258,
+ "step": 13853
+ },
+ {
+ "epoch": 37.74931880108992,
+ "grad_norm": 3.7562899589538574,
+ "learning_rate": 1.4307745960300858e-05,
+ "loss": 0.15,
+ "step": 13854
+ },
+ {
+ "epoch": 37.752043596730246,
+ "grad_norm": 3.251781463623047,
+ "learning_rate": 1.430694952781187e-05,
+ "loss": 0.0909,
+ "step": 13855
+ },
+ {
+ "epoch": 37.75476839237057,
+ "grad_norm": 3.651534080505371,
+ "learning_rate": 1.43061530617805e-05,
+ "loss": 0.0901,
+ "step": 13856
+ },
+ {
+ "epoch": 37.7574931880109,
+ "grad_norm": 3.279602527618408,
+ "learning_rate": 1.4305356562212947e-05,
+ "loss": 0.1261,
+ "step": 13857
+ },
+ {
+ "epoch": 37.76021798365122,
+ "grad_norm": 3.465087890625,
+ "learning_rate": 1.4304560029115415e-05,
+ "loss": 0.1126,
+ "step": 13858
+ },
+ {
+ "epoch": 37.762942779291556,
+ "grad_norm": 3.6901323795318604,
+ "learning_rate": 1.4303763462494112e-05,
+ "loss": 0.143,
+ "step": 13859
+ },
+ {
+ "epoch": 37.76566757493188,
+ "grad_norm": 4.600917816162109,
+ "learning_rate": 1.4302966862355235e-05,
+ "loss": 0.1294,
+ "step": 13860
+ },
+ {
+ "epoch": 37.76839237057221,
+ "grad_norm": 3.992794990539551,
+ "learning_rate": 1.4302170228704987e-05,
+ "loss": 0.1185,
+ "step": 13861
+ },
+ {
+ "epoch": 37.771117166212534,
+ "grad_norm": 3.8065483570098877,
+ "learning_rate": 1.430137356154958e-05,
+ "loss": 0.0977,
+ "step": 13862
+ },
+ {
+ "epoch": 37.77384196185286,
+ "grad_norm": 3.972712755203247,
+ "learning_rate": 1.4300576860895215e-05,
+ "loss": 0.2276,
+ "step": 13863
+ },
+ {
+ "epoch": 37.776566757493185,
+ "grad_norm": 4.062636852264404,
+ "learning_rate": 1.4299780126748091e-05,
+ "loss": 0.1578,
+ "step": 13864
+ },
+ {
+ "epoch": 37.77929155313352,
+ "grad_norm": 3.9515957832336426,
+ "learning_rate": 1.4298983359114422e-05,
+ "loss": 0.3831,
+ "step": 13865
+ },
+ {
+ "epoch": 37.782016348773844,
+ "grad_norm": 5.4728875160217285,
+ "learning_rate": 1.4298186558000406e-05,
+ "loss": 0.1519,
+ "step": 13866
+ },
+ {
+ "epoch": 37.78474114441417,
+ "grad_norm": 5.128042697906494,
+ "learning_rate": 1.4297389723412256e-05,
+ "loss": 0.2157,
+ "step": 13867
+ },
+ {
+ "epoch": 37.787465940054496,
+ "grad_norm": 3.5184903144836426,
+ "learning_rate": 1.429659285535617e-05,
+ "loss": 0.1444,
+ "step": 13868
+ },
+ {
+ "epoch": 37.79019073569482,
+ "grad_norm": 4.943660259246826,
+ "learning_rate": 1.4295795953838358e-05,
+ "loss": 0.194,
+ "step": 13869
+ },
+ {
+ "epoch": 37.79291553133515,
+ "grad_norm": 3.8278229236602783,
+ "learning_rate": 1.4294999018865023e-05,
+ "loss": 0.3073,
+ "step": 13870
+ },
+ {
+ "epoch": 37.79564032697548,
+ "grad_norm": 4.482454299926758,
+ "learning_rate": 1.4294202050442379e-05,
+ "loss": 0.3105,
+ "step": 13871
+ },
+ {
+ "epoch": 37.798365122615806,
+ "grad_norm": 4.888441562652588,
+ "learning_rate": 1.4293405048576623e-05,
+ "loss": 0.1676,
+ "step": 13872
+ },
+ {
+ "epoch": 37.80108991825613,
+ "grad_norm": 4.677815914154053,
+ "learning_rate": 1.4292608013273968e-05,
+ "loss": 0.1087,
+ "step": 13873
+ },
+ {
+ "epoch": 37.80381471389646,
+ "grad_norm": 3.675924777984619,
+ "learning_rate": 1.4291810944540619e-05,
+ "loss": 0.2351,
+ "step": 13874
+ },
+ {
+ "epoch": 37.80653950953678,
+ "grad_norm": 3.626073122024536,
+ "learning_rate": 1.4291013842382785e-05,
+ "loss": 0.4378,
+ "step": 13875
+ },
+ {
+ "epoch": 37.80926430517711,
+ "grad_norm": 4.120556831359863,
+ "learning_rate": 1.4290216706806675e-05,
+ "loss": 0.1228,
+ "step": 13876
+ },
+ {
+ "epoch": 37.81198910081744,
+ "grad_norm": 5.305412769317627,
+ "learning_rate": 1.4289419537818494e-05,
+ "loss": 0.1658,
+ "step": 13877
+ },
+ {
+ "epoch": 37.81471389645777,
+ "grad_norm": 3.477285861968994,
+ "learning_rate": 1.4288622335424452e-05,
+ "loss": 0.267,
+ "step": 13878
+ },
+ {
+ "epoch": 37.817438692098094,
+ "grad_norm": 4.175856113433838,
+ "learning_rate": 1.4287825099630759e-05,
+ "loss": 0.1285,
+ "step": 13879
+ },
+ {
+ "epoch": 37.82016348773842,
+ "grad_norm": 4.026623249053955,
+ "learning_rate": 1.4287027830443618e-05,
+ "loss": 0.1566,
+ "step": 13880
+ },
+ {
+ "epoch": 37.822888283378745,
+ "grad_norm": 3.693837881088257,
+ "learning_rate": 1.4286230527869245e-05,
+ "loss": 0.1036,
+ "step": 13881
+ },
+ {
+ "epoch": 37.82561307901907,
+ "grad_norm": 32.18373107910156,
+ "learning_rate": 1.4285433191913845e-05,
+ "loss": 0.1589,
+ "step": 13882
+ },
+ {
+ "epoch": 37.828337874659404,
+ "grad_norm": 3.992586374282837,
+ "learning_rate": 1.4284635822583632e-05,
+ "loss": 0.2132,
+ "step": 13883
+ },
+ {
+ "epoch": 37.83106267029973,
+ "grad_norm": 5.666081428527832,
+ "learning_rate": 1.428383841988481e-05,
+ "loss": 0.1624,
+ "step": 13884
+ },
+ {
+ "epoch": 37.833787465940055,
+ "grad_norm": 4.641520977020264,
+ "learning_rate": 1.4283040983823594e-05,
+ "loss": 0.1647,
+ "step": 13885
+ },
+ {
+ "epoch": 37.83651226158038,
+ "grad_norm": 4.850436210632324,
+ "learning_rate": 1.4282243514406192e-05,
+ "loss": 0.2141,
+ "step": 13886
+ },
+ {
+ "epoch": 37.83923705722071,
+ "grad_norm": 3.2379584312438965,
+ "learning_rate": 1.4281446011638818e-05,
+ "loss": 0.2595,
+ "step": 13887
+ },
+ {
+ "epoch": 37.84196185286103,
+ "grad_norm": 3.9466588497161865,
+ "learning_rate": 1.4280648475527676e-05,
+ "loss": 0.231,
+ "step": 13888
+ },
+ {
+ "epoch": 37.844686648501366,
+ "grad_norm": 4.38828706741333,
+ "learning_rate": 1.4279850906078984e-05,
+ "loss": 0.1602,
+ "step": 13889
+ },
+ {
+ "epoch": 37.84741144414169,
+ "grad_norm": 4.4213337898254395,
+ "learning_rate": 1.427905330329895e-05,
+ "loss": 0.1768,
+ "step": 13890
+ },
+ {
+ "epoch": 37.85013623978202,
+ "grad_norm": 3.5918869972229004,
+ "learning_rate": 1.4278255667193785e-05,
+ "loss": 0.1743,
+ "step": 13891
+ },
+ {
+ "epoch": 37.85286103542234,
+ "grad_norm": 5.897396087646484,
+ "learning_rate": 1.4277457997769707e-05,
+ "loss": 0.1109,
+ "step": 13892
+ },
+ {
+ "epoch": 37.85558583106267,
+ "grad_norm": 3.797861099243164,
+ "learning_rate": 1.427666029503292e-05,
+ "loss": 0.1531,
+ "step": 13893
+ },
+ {
+ "epoch": 37.858310626702995,
+ "grad_norm": 4.4702935218811035,
+ "learning_rate": 1.4275862558989642e-05,
+ "loss": 0.1806,
+ "step": 13894
+ },
+ {
+ "epoch": 37.86103542234333,
+ "grad_norm": 4.393936634063721,
+ "learning_rate": 1.4275064789646085e-05,
+ "loss": 0.1624,
+ "step": 13895
+ },
+ {
+ "epoch": 37.86376021798365,
+ "grad_norm": 4.080281734466553,
+ "learning_rate": 1.4274266987008459e-05,
+ "loss": 0.2493,
+ "step": 13896
+ },
+ {
+ "epoch": 37.86648501362398,
+ "grad_norm": 4.9120564460754395,
+ "learning_rate": 1.427346915108298e-05,
+ "loss": 0.3262,
+ "step": 13897
+ },
+ {
+ "epoch": 37.869209809264305,
+ "grad_norm": 3.816009283065796,
+ "learning_rate": 1.4272671281875857e-05,
+ "loss": 0.2228,
+ "step": 13898
+ },
+ {
+ "epoch": 37.87193460490463,
+ "grad_norm": 3.458791971206665,
+ "learning_rate": 1.4271873379393314e-05,
+ "loss": 0.1719,
+ "step": 13899
+ },
+ {
+ "epoch": 37.87465940054496,
+ "grad_norm": 4.165095806121826,
+ "learning_rate": 1.4271075443641552e-05,
+ "loss": 0.132,
+ "step": 13900
+ },
+ {
+ "epoch": 37.87738419618529,
+ "grad_norm": 5.342282295227051,
+ "learning_rate": 1.4270277474626799e-05,
+ "loss": 0.2589,
+ "step": 13901
+ },
+ {
+ "epoch": 37.880108991825615,
+ "grad_norm": 4.168907165527344,
+ "learning_rate": 1.4269479472355259e-05,
+ "loss": 0.1499,
+ "step": 13902
+ },
+ {
+ "epoch": 37.88283378746594,
+ "grad_norm": 3.2016820907592773,
+ "learning_rate": 1.4268681436833149e-05,
+ "loss": 0.123,
+ "step": 13903
+ },
+ {
+ "epoch": 37.88555858310627,
+ "grad_norm": 3.693169593811035,
+ "learning_rate": 1.4267883368066688e-05,
+ "loss": 0.191,
+ "step": 13904
+ },
+ {
+ "epoch": 37.88828337874659,
+ "grad_norm": 4.7284770011901855,
+ "learning_rate": 1.4267085266062088e-05,
+ "loss": 0.1691,
+ "step": 13905
+ },
+ {
+ "epoch": 37.89100817438692,
+ "grad_norm": 4.019737720489502,
+ "learning_rate": 1.4266287130825564e-05,
+ "loss": 0.196,
+ "step": 13906
+ },
+ {
+ "epoch": 37.89373297002725,
+ "grad_norm": 4.764437198638916,
+ "learning_rate": 1.4265488962363334e-05,
+ "loss": 0.1272,
+ "step": 13907
+ },
+ {
+ "epoch": 37.89645776566758,
+ "grad_norm": 3.994277000427246,
+ "learning_rate": 1.4264690760681613e-05,
+ "loss": 0.2906,
+ "step": 13908
+ },
+ {
+ "epoch": 37.8991825613079,
+ "grad_norm": 3.921323299407959,
+ "learning_rate": 1.4263892525786617e-05,
+ "loss": 0.2179,
+ "step": 13909
+ },
+ {
+ "epoch": 37.90190735694823,
+ "grad_norm": 3.702531337738037,
+ "learning_rate": 1.4263094257684564e-05,
+ "loss": 0.2753,
+ "step": 13910
+ },
+ {
+ "epoch": 37.904632152588555,
+ "grad_norm": 4.32727575302124,
+ "learning_rate": 1.426229595638167e-05,
+ "loss": 0.1404,
+ "step": 13911
+ },
+ {
+ "epoch": 37.90735694822888,
+ "grad_norm": 3.5643234252929688,
+ "learning_rate": 1.4261497621884152e-05,
+ "loss": 0.1199,
+ "step": 13912
+ },
+ {
+ "epoch": 37.91008174386921,
+ "grad_norm": 3.6292009353637695,
+ "learning_rate": 1.4260699254198231e-05,
+ "loss": 0.1803,
+ "step": 13913
+ },
+ {
+ "epoch": 37.91280653950954,
+ "grad_norm": 4.693570613861084,
+ "learning_rate": 1.4259900853330116e-05,
+ "loss": 0.1518,
+ "step": 13914
+ },
+ {
+ "epoch": 37.915531335149865,
+ "grad_norm": 4.540931224822998,
+ "learning_rate": 1.4259102419286035e-05,
+ "loss": 0.1297,
+ "step": 13915
+ },
+ {
+ "epoch": 37.91825613079019,
+ "grad_norm": 4.026116847991943,
+ "learning_rate": 1.4258303952072197e-05,
+ "loss": 0.2535,
+ "step": 13916
+ },
+ {
+ "epoch": 37.920980926430516,
+ "grad_norm": 3.6070289611816406,
+ "learning_rate": 1.4257505451694826e-05,
+ "loss": 0.111,
+ "step": 13917
+ },
+ {
+ "epoch": 37.92370572207084,
+ "grad_norm": 5.254515171051025,
+ "learning_rate": 1.4256706918160142e-05,
+ "loss": 0.1864,
+ "step": 13918
+ },
+ {
+ "epoch": 37.926430517711175,
+ "grad_norm": 4.385950088500977,
+ "learning_rate": 1.4255908351474358e-05,
+ "loss": 0.1098,
+ "step": 13919
+ },
+ {
+ "epoch": 37.9291553133515,
+ "grad_norm": 4.430014133453369,
+ "learning_rate": 1.4255109751643698e-05,
+ "loss": 0.146,
+ "step": 13920
+ },
+ {
+ "epoch": 37.93188010899183,
+ "grad_norm": 3.8786838054656982,
+ "learning_rate": 1.4254311118674385e-05,
+ "loss": 0.2114,
+ "step": 13921
+ },
+ {
+ "epoch": 37.93460490463215,
+ "grad_norm": 3.6663551330566406,
+ "learning_rate": 1.4253512452572629e-05,
+ "loss": 0.092,
+ "step": 13922
+ },
+ {
+ "epoch": 37.93732970027248,
+ "grad_norm": 3.970649003982544,
+ "learning_rate": 1.4252713753344656e-05,
+ "loss": 0.2159,
+ "step": 13923
+ },
+ {
+ "epoch": 37.940054495912804,
+ "grad_norm": 4.057528495788574,
+ "learning_rate": 1.4251915020996686e-05,
+ "loss": 0.2639,
+ "step": 13924
+ },
+ {
+ "epoch": 37.94277929155314,
+ "grad_norm": 3.7795891761779785,
+ "learning_rate": 1.4251116255534936e-05,
+ "loss": 0.2311,
+ "step": 13925
+ },
+ {
+ "epoch": 37.94550408719346,
+ "grad_norm": 4.222429275512695,
+ "learning_rate": 1.425031745696563e-05,
+ "loss": 0.1047,
+ "step": 13926
+ },
+ {
+ "epoch": 37.94822888283379,
+ "grad_norm": 7.097311496734619,
+ "learning_rate": 1.4249518625294991e-05,
+ "loss": 0.1843,
+ "step": 13927
+ },
+ {
+ "epoch": 37.950953678474114,
+ "grad_norm": 6.351932525634766,
+ "learning_rate": 1.4248719760529235e-05,
+ "loss": 0.1893,
+ "step": 13928
+ },
+ {
+ "epoch": 37.95367847411444,
+ "grad_norm": 4.553431034088135,
+ "learning_rate": 1.4247920862674589e-05,
+ "loss": 0.2645,
+ "step": 13929
+ },
+ {
+ "epoch": 37.956403269754766,
+ "grad_norm": 4.4812726974487305,
+ "learning_rate": 1.4247121931737269e-05,
+ "loss": 0.2726,
+ "step": 13930
+ },
+ {
+ "epoch": 37.95912806539509,
+ "grad_norm": 3.5234851837158203,
+ "learning_rate": 1.4246322967723502e-05,
+ "loss": 0.0825,
+ "step": 13931
+ },
+ {
+ "epoch": 37.961852861035425,
+ "grad_norm": 3.582716464996338,
+ "learning_rate": 1.424552397063951e-05,
+ "loss": 0.3342,
+ "step": 13932
+ },
+ {
+ "epoch": 37.96457765667575,
+ "grad_norm": 4.848450183868408,
+ "learning_rate": 1.424472494049151e-05,
+ "loss": 0.1987,
+ "step": 13933
+ },
+ {
+ "epoch": 37.967302452316076,
+ "grad_norm": 3.6350951194763184,
+ "learning_rate": 1.4243925877285731e-05,
+ "loss": 0.2132,
+ "step": 13934
+ },
+ {
+ "epoch": 37.9700272479564,
+ "grad_norm": 3.678093910217285,
+ "learning_rate": 1.4243126781028393e-05,
+ "loss": 0.2044,
+ "step": 13935
+ },
+ {
+ "epoch": 37.97275204359673,
+ "grad_norm": 4.635814189910889,
+ "learning_rate": 1.424232765172572e-05,
+ "loss": 0.1701,
+ "step": 13936
+ },
+ {
+ "epoch": 37.97547683923706,
+ "grad_norm": 4.402882099151611,
+ "learning_rate": 1.4241528489383938e-05,
+ "loss": 0.1616,
+ "step": 13937
+ },
+ {
+ "epoch": 37.97820163487739,
+ "grad_norm": 3.797759771347046,
+ "learning_rate": 1.4240729294009268e-05,
+ "loss": 0.1363,
+ "step": 13938
+ },
+ {
+ "epoch": 37.98092643051771,
+ "grad_norm": 3.8869829177856445,
+ "learning_rate": 1.4239930065607934e-05,
+ "loss": 0.2008,
+ "step": 13939
+ },
+ {
+ "epoch": 37.98365122615804,
+ "grad_norm": 4.241336345672607,
+ "learning_rate": 1.4239130804186161e-05,
+ "loss": 0.1398,
+ "step": 13940
+ },
+ {
+ "epoch": 37.986376021798364,
+ "grad_norm": 3.53261661529541,
+ "learning_rate": 1.4238331509750173e-05,
+ "loss": 0.322,
+ "step": 13941
+ },
+ {
+ "epoch": 37.98910081743869,
+ "grad_norm": 3.070261001586914,
+ "learning_rate": 1.4237532182306198e-05,
+ "loss": 0.1295,
+ "step": 13942
+ },
+ {
+ "epoch": 37.991825613079016,
+ "grad_norm": 3.4839630126953125,
+ "learning_rate": 1.4236732821860456e-05,
+ "loss": 0.0773,
+ "step": 13943
+ },
+ {
+ "epoch": 37.99455040871935,
+ "grad_norm": 3.621271848678589,
+ "learning_rate": 1.4235933428419177e-05,
+ "loss": 0.1267,
+ "step": 13944
+ },
+ {
+ "epoch": 37.997275204359674,
+ "grad_norm": 3.4866626262664795,
+ "learning_rate": 1.4235134001988584e-05,
+ "loss": 0.2113,
+ "step": 13945
+ },
+ {
+ "epoch": 38.0,
+ "grad_norm": 3.7373645305633545,
+ "learning_rate": 1.4234334542574906e-05,
+ "loss": 0.2012,
+ "step": 13946
+ },
+ {
+ "epoch": 38.002724795640326,
+ "grad_norm": 3.059135913848877,
+ "learning_rate": 1.4233535050184362e-05,
+ "loss": 0.0846,
+ "step": 13947
+ },
+ {
+ "epoch": 38.00544959128065,
+ "grad_norm": 3.3280210494995117,
+ "learning_rate": 1.4232735524823187e-05,
+ "loss": 0.1763,
+ "step": 13948
+ },
+ {
+ "epoch": 38.00817438692098,
+ "grad_norm": 3.515549659729004,
+ "learning_rate": 1.4231935966497603e-05,
+ "loss": 0.1041,
+ "step": 13949
+ },
+ {
+ "epoch": 38.01089918256131,
+ "grad_norm": 4.4094133377075195,
+ "learning_rate": 1.4231136375213837e-05,
+ "loss": 0.2286,
+ "step": 13950
+ },
+ {
+ "epoch": 38.013623978201636,
+ "grad_norm": 2.522056818008423,
+ "learning_rate": 1.4230336750978117e-05,
+ "loss": 0.0831,
+ "step": 13951
+ },
+ {
+ "epoch": 38.01634877384196,
+ "grad_norm": 3.7353692054748535,
+ "learning_rate": 1.4229537093796672e-05,
+ "loss": 0.2237,
+ "step": 13952
+ },
+ {
+ "epoch": 38.01907356948229,
+ "grad_norm": 3.7609612941741943,
+ "learning_rate": 1.4228737403675727e-05,
+ "loss": 0.1263,
+ "step": 13953
+ },
+ {
+ "epoch": 38.02179836512261,
+ "grad_norm": 3.293534994125366,
+ "learning_rate": 1.4227937680621513e-05,
+ "loss": 0.1628,
+ "step": 13954
+ },
+ {
+ "epoch": 38.02452316076294,
+ "grad_norm": 3.659073829650879,
+ "learning_rate": 1.4227137924640255e-05,
+ "loss": 0.1668,
+ "step": 13955
+ },
+ {
+ "epoch": 38.02724795640327,
+ "grad_norm": 3.2908935546875,
+ "learning_rate": 1.4226338135738185e-05,
+ "loss": 0.1407,
+ "step": 13956
+ },
+ {
+ "epoch": 38.0299727520436,
+ "grad_norm": 3.223428726196289,
+ "learning_rate": 1.4225538313921527e-05,
+ "loss": 0.1125,
+ "step": 13957
+ },
+ {
+ "epoch": 38.032697547683924,
+ "grad_norm": 3.28900146484375,
+ "learning_rate": 1.4224738459196514e-05,
+ "loss": 0.1963,
+ "step": 13958
+ },
+ {
+ "epoch": 38.03542234332425,
+ "grad_norm": 3.4377307891845703,
+ "learning_rate": 1.4223938571569371e-05,
+ "loss": 0.1454,
+ "step": 13959
+ },
+ {
+ "epoch": 38.038147138964575,
+ "grad_norm": 4.181890964508057,
+ "learning_rate": 1.4223138651046333e-05,
+ "loss": 0.1491,
+ "step": 13960
+ },
+ {
+ "epoch": 38.0408719346049,
+ "grad_norm": 3.581207036972046,
+ "learning_rate": 1.4222338697633627e-05,
+ "loss": 0.1425,
+ "step": 13961
+ },
+ {
+ "epoch": 38.043596730245234,
+ "grad_norm": 3.851961135864258,
+ "learning_rate": 1.4221538711337483e-05,
+ "loss": 0.0836,
+ "step": 13962
+ },
+ {
+ "epoch": 38.04632152588556,
+ "grad_norm": 3.688572883605957,
+ "learning_rate": 1.4220738692164132e-05,
+ "loss": 0.1597,
+ "step": 13963
+ },
+ {
+ "epoch": 38.049046321525886,
+ "grad_norm": 3.8746843338012695,
+ "learning_rate": 1.4219938640119803e-05,
+ "loss": 0.1721,
+ "step": 13964
+ },
+ {
+ "epoch": 38.05177111716621,
+ "grad_norm": 3.606640100479126,
+ "learning_rate": 1.4219138555210727e-05,
+ "loss": 0.2908,
+ "step": 13965
+ },
+ {
+ "epoch": 38.05449591280654,
+ "grad_norm": 3.3415415287017822,
+ "learning_rate": 1.421833843744314e-05,
+ "loss": 0.1646,
+ "step": 13966
+ },
+ {
+ "epoch": 38.05722070844686,
+ "grad_norm": 3.6494805812835693,
+ "learning_rate": 1.4217538286823262e-05,
+ "loss": 0.0761,
+ "step": 13967
+ },
+ {
+ "epoch": 38.059945504087196,
+ "grad_norm": 6.4264421463012695,
+ "learning_rate": 1.4216738103357335e-05,
+ "loss": 0.1544,
+ "step": 13968
+ },
+ {
+ "epoch": 38.06267029972752,
+ "grad_norm": 3.216275691986084,
+ "learning_rate": 1.4215937887051588e-05,
+ "loss": 0.1162,
+ "step": 13969
+ },
+ {
+ "epoch": 38.06539509536785,
+ "grad_norm": 3.799415349960327,
+ "learning_rate": 1.4215137637912252e-05,
+ "loss": 0.1818,
+ "step": 13970
+ },
+ {
+ "epoch": 38.06811989100817,
+ "grad_norm": 3.9421629905700684,
+ "learning_rate": 1.421433735594556e-05,
+ "loss": 0.1165,
+ "step": 13971
+ },
+ {
+ "epoch": 38.0708446866485,
+ "grad_norm": 3.0199246406555176,
+ "learning_rate": 1.4213537041157746e-05,
+ "loss": 0.0681,
+ "step": 13972
+ },
+ {
+ "epoch": 38.073569482288825,
+ "grad_norm": 4.179049968719482,
+ "learning_rate": 1.4212736693555038e-05,
+ "loss": 0.1386,
+ "step": 13973
+ },
+ {
+ "epoch": 38.07629427792916,
+ "grad_norm": 3.3011035919189453,
+ "learning_rate": 1.4211936313143673e-05,
+ "loss": 0.0982,
+ "step": 13974
+ },
+ {
+ "epoch": 38.079019073569484,
+ "grad_norm": 2.970900297164917,
+ "learning_rate": 1.4211135899929885e-05,
+ "loss": 0.1165,
+ "step": 13975
+ },
+ {
+ "epoch": 38.08174386920981,
+ "grad_norm": 3.672281265258789,
+ "learning_rate": 1.4210335453919904e-05,
+ "loss": 0.1081,
+ "step": 13976
+ },
+ {
+ "epoch": 38.084468664850135,
+ "grad_norm": 3.636178731918335,
+ "learning_rate": 1.4209534975119965e-05,
+ "loss": 0.1907,
+ "step": 13977
+ },
+ {
+ "epoch": 38.08719346049046,
+ "grad_norm": 2.916801929473877,
+ "learning_rate": 1.4208734463536307e-05,
+ "loss": 0.0916,
+ "step": 13978
+ },
+ {
+ "epoch": 38.08991825613079,
+ "grad_norm": 3.9426817893981934,
+ "learning_rate": 1.4207933919175157e-05,
+ "loss": 0.1705,
+ "step": 13979
+ },
+ {
+ "epoch": 38.09264305177112,
+ "grad_norm": 4.9384002685546875,
+ "learning_rate": 1.4207133342042756e-05,
+ "loss": 0.2246,
+ "step": 13980
+ },
+ {
+ "epoch": 38.095367847411445,
+ "grad_norm": 3.592484951019287,
+ "learning_rate": 1.4206332732145334e-05,
+ "loss": 0.2194,
+ "step": 13981
+ },
+ {
+ "epoch": 38.09809264305177,
+ "grad_norm": 3.037217617034912,
+ "learning_rate": 1.420553208948913e-05,
+ "loss": 0.1671,
+ "step": 13982
+ },
+ {
+ "epoch": 38.1008174386921,
+ "grad_norm": 3.022444486618042,
+ "learning_rate": 1.4204731414080375e-05,
+ "loss": 0.2889,
+ "step": 13983
+ },
+ {
+ "epoch": 38.10354223433242,
+ "grad_norm": 3.831031322479248,
+ "learning_rate": 1.4203930705925308e-05,
+ "loss": 0.2184,
+ "step": 13984
+ },
+ {
+ "epoch": 38.10626702997275,
+ "grad_norm": 6.260773181915283,
+ "learning_rate": 1.4203129965030162e-05,
+ "loss": 0.1042,
+ "step": 13985
+ },
+ {
+ "epoch": 38.10899182561308,
+ "grad_norm": 4.062816619873047,
+ "learning_rate": 1.4202329191401178e-05,
+ "loss": 0.1799,
+ "step": 13986
+ },
+ {
+ "epoch": 38.11171662125341,
+ "grad_norm": 3.943955898284912,
+ "learning_rate": 1.4201528385044589e-05,
+ "loss": 0.1751,
+ "step": 13987
+ },
+ {
+ "epoch": 38.11444141689373,
+ "grad_norm": 3.3669941425323486,
+ "learning_rate": 1.4200727545966633e-05,
+ "loss": 0.1539,
+ "step": 13988
+ },
+ {
+ "epoch": 38.11716621253406,
+ "grad_norm": 2.929950475692749,
+ "learning_rate": 1.4199926674173545e-05,
+ "loss": 0.1032,
+ "step": 13989
+ },
+ {
+ "epoch": 38.119891008174385,
+ "grad_norm": 3.755445957183838,
+ "learning_rate": 1.4199125769671564e-05,
+ "loss": 0.2383,
+ "step": 13990
+ },
+ {
+ "epoch": 38.12261580381471,
+ "grad_norm": 3.2988317012786865,
+ "learning_rate": 1.4198324832466925e-05,
+ "loss": 0.2148,
+ "step": 13991
+ },
+ {
+ "epoch": 38.12534059945504,
+ "grad_norm": 3.7238259315490723,
+ "learning_rate": 1.419752386256587e-05,
+ "loss": 0.3251,
+ "step": 13992
+ },
+ {
+ "epoch": 38.12806539509537,
+ "grad_norm": 3.183495283126831,
+ "learning_rate": 1.4196722859974629e-05,
+ "loss": 0.1247,
+ "step": 13993
+ },
+ {
+ "epoch": 38.130790190735695,
+ "grad_norm": 3.857301950454712,
+ "learning_rate": 1.419592182469945e-05,
+ "loss": 0.1187,
+ "step": 13994
+ },
+ {
+ "epoch": 38.13351498637602,
+ "grad_norm": 3.185110092163086,
+ "learning_rate": 1.4195120756746566e-05,
+ "loss": 0.1604,
+ "step": 13995
+ },
+ {
+ "epoch": 38.13623978201635,
+ "grad_norm": 3.2463865280151367,
+ "learning_rate": 1.4194319656122217e-05,
+ "loss": 0.1137,
+ "step": 13996
+ },
+ {
+ "epoch": 38.13896457765667,
+ "grad_norm": 3.548471212387085,
+ "learning_rate": 1.4193518522832642e-05,
+ "loss": 0.2027,
+ "step": 13997
+ },
+ {
+ "epoch": 38.141689373297005,
+ "grad_norm": 5.014037609100342,
+ "learning_rate": 1.4192717356884078e-05,
+ "loss": 0.1466,
+ "step": 13998
+ },
+ {
+ "epoch": 38.14441416893733,
+ "grad_norm": 3.353203296661377,
+ "learning_rate": 1.4191916158282766e-05,
+ "loss": 0.1958,
+ "step": 13999
+ },
+ {
+ "epoch": 38.14713896457766,
+ "grad_norm": 2.966999053955078,
+ "learning_rate": 1.4191114927034949e-05,
+ "loss": 0.0863,
+ "step": 14000
+ },
+ {
+ "epoch": 38.14986376021798,
+ "grad_norm": 4.661839008331299,
+ "learning_rate": 1.4190313663146859e-05,
+ "loss": 0.1388,
+ "step": 14001
+ },
+ {
+ "epoch": 38.15258855585831,
+ "grad_norm": 3.470564126968384,
+ "learning_rate": 1.4189512366624745e-05,
+ "loss": 0.221,
+ "step": 14002
+ },
+ {
+ "epoch": 38.155313351498634,
+ "grad_norm": 2.8508172035217285,
+ "learning_rate": 1.418871103747484e-05,
+ "loss": 0.1735,
+ "step": 14003
+ },
+ {
+ "epoch": 38.15803814713897,
+ "grad_norm": 3.3028764724731445,
+ "learning_rate": 1.4187909675703391e-05,
+ "loss": 0.1056,
+ "step": 14004
+ },
+ {
+ "epoch": 38.16076294277929,
+ "grad_norm": 2.9627201557159424,
+ "learning_rate": 1.4187108281316635e-05,
+ "loss": 0.2383,
+ "step": 14005
+ },
+ {
+ "epoch": 38.16348773841962,
+ "grad_norm": 3.0156006813049316,
+ "learning_rate": 1.4186306854320817e-05,
+ "loss": 0.2,
+ "step": 14006
+ },
+ {
+ "epoch": 38.166212534059945,
+ "grad_norm": 2.9936506748199463,
+ "learning_rate": 1.4185505394722172e-05,
+ "loss": 0.1288,
+ "step": 14007
+ },
+ {
+ "epoch": 38.16893732970027,
+ "grad_norm": 2.7741081714630127,
+ "learning_rate": 1.418470390252695e-05,
+ "loss": 0.1404,
+ "step": 14008
+ },
+ {
+ "epoch": 38.171662125340596,
+ "grad_norm": 3.207383155822754,
+ "learning_rate": 1.4183902377741385e-05,
+ "loss": 0.2158,
+ "step": 14009
+ },
+ {
+ "epoch": 38.17438692098093,
+ "grad_norm": 3.474409341812134,
+ "learning_rate": 1.4183100820371723e-05,
+ "loss": 0.138,
+ "step": 14010
+ },
+ {
+ "epoch": 38.177111716621255,
+ "grad_norm": 3.7058088779449463,
+ "learning_rate": 1.418229923042421e-05,
+ "loss": 0.2725,
+ "step": 14011
+ },
+ {
+ "epoch": 38.17983651226158,
+ "grad_norm": 3.1114790439605713,
+ "learning_rate": 1.4181497607905082e-05,
+ "loss": 0.1696,
+ "step": 14012
+ },
+ {
+ "epoch": 38.182561307901906,
+ "grad_norm": 3.3860650062561035,
+ "learning_rate": 1.4180695952820585e-05,
+ "loss": 0.1622,
+ "step": 14013
+ },
+ {
+ "epoch": 38.18528610354223,
+ "grad_norm": 3.6057450771331787,
+ "learning_rate": 1.4179894265176964e-05,
+ "loss": 0.2474,
+ "step": 14014
+ },
+ {
+ "epoch": 38.18801089918256,
+ "grad_norm": 3.9680912494659424,
+ "learning_rate": 1.4179092544980459e-05,
+ "loss": 0.1052,
+ "step": 14015
+ },
+ {
+ "epoch": 38.19073569482289,
+ "grad_norm": 3.6684770584106445,
+ "learning_rate": 1.4178290792237319e-05,
+ "loss": 0.1142,
+ "step": 14016
+ },
+ {
+ "epoch": 38.19346049046322,
+ "grad_norm": 4.198531627655029,
+ "learning_rate": 1.4177489006953784e-05,
+ "loss": 0.1268,
+ "step": 14017
+ },
+ {
+ "epoch": 38.19618528610354,
+ "grad_norm": 3.8875484466552734,
+ "learning_rate": 1.4176687189136098e-05,
+ "loss": 0.1107,
+ "step": 14018
+ },
+ {
+ "epoch": 38.19891008174387,
+ "grad_norm": 3.514880657196045,
+ "learning_rate": 1.4175885338790507e-05,
+ "loss": 0.3154,
+ "step": 14019
+ },
+ {
+ "epoch": 38.201634877384194,
+ "grad_norm": 3.1846511363983154,
+ "learning_rate": 1.4175083455923254e-05,
+ "loss": 0.1798,
+ "step": 14020
+ },
+ {
+ "epoch": 38.20435967302452,
+ "grad_norm": 3.657844305038452,
+ "learning_rate": 1.417428154054059e-05,
+ "loss": 0.2521,
+ "step": 14021
+ },
+ {
+ "epoch": 38.20708446866485,
+ "grad_norm": 3.5783708095550537,
+ "learning_rate": 1.417347959264875e-05,
+ "loss": 0.1199,
+ "step": 14022
+ },
+ {
+ "epoch": 38.20980926430518,
+ "grad_norm": 4.034553527832031,
+ "learning_rate": 1.4172677612253987e-05,
+ "loss": 0.1094,
+ "step": 14023
+ },
+ {
+ "epoch": 38.212534059945504,
+ "grad_norm": 3.6115174293518066,
+ "learning_rate": 1.4171875599362545e-05,
+ "loss": 0.2459,
+ "step": 14024
+ },
+ {
+ "epoch": 38.21525885558583,
+ "grad_norm": 3.292365074157715,
+ "learning_rate": 1.4171073553980673e-05,
+ "loss": 0.1998,
+ "step": 14025
+ },
+ {
+ "epoch": 38.217983651226156,
+ "grad_norm": 3.571512222290039,
+ "learning_rate": 1.4170271476114612e-05,
+ "loss": 0.0923,
+ "step": 14026
+ },
+ {
+ "epoch": 38.22070844686648,
+ "grad_norm": 3.6074600219726562,
+ "learning_rate": 1.416946936577061e-05,
+ "loss": 0.0848,
+ "step": 14027
+ },
+ {
+ "epoch": 38.223433242506815,
+ "grad_norm": 3.5145585536956787,
+ "learning_rate": 1.4168667222954915e-05,
+ "loss": 0.1504,
+ "step": 14028
+ },
+ {
+ "epoch": 38.22615803814714,
+ "grad_norm": 3.1022424697875977,
+ "learning_rate": 1.4167865047673775e-05,
+ "loss": 0.1477,
+ "step": 14029
+ },
+ {
+ "epoch": 38.228882833787466,
+ "grad_norm": 3.7328057289123535,
+ "learning_rate": 1.4167062839933437e-05,
+ "loss": 0.1648,
+ "step": 14030
+ },
+ {
+ "epoch": 38.23160762942779,
+ "grad_norm": 4.197993755340576,
+ "learning_rate": 1.4166260599740148e-05,
+ "loss": 0.1593,
+ "step": 14031
+ },
+ {
+ "epoch": 38.23433242506812,
+ "grad_norm": 2.6917848587036133,
+ "learning_rate": 1.4165458327100152e-05,
+ "loss": 0.1315,
+ "step": 14032
+ },
+ {
+ "epoch": 38.237057220708444,
+ "grad_norm": 3.2726354598999023,
+ "learning_rate": 1.4164656022019702e-05,
+ "loss": 0.2144,
+ "step": 14033
+ },
+ {
+ "epoch": 38.23978201634878,
+ "grad_norm": 2.9934921264648438,
+ "learning_rate": 1.4163853684505046e-05,
+ "loss": 0.2644,
+ "step": 14034
+ },
+ {
+ "epoch": 38.2425068119891,
+ "grad_norm": 3.7820656299591064,
+ "learning_rate": 1.416305131456243e-05,
+ "loss": 0.0657,
+ "step": 14035
+ },
+ {
+ "epoch": 38.24523160762943,
+ "grad_norm": 3.0555386543273926,
+ "learning_rate": 1.4162248912198107e-05,
+ "loss": 0.1645,
+ "step": 14036
+ },
+ {
+ "epoch": 38.247956403269754,
+ "grad_norm": 12.432686805725098,
+ "learning_rate": 1.4161446477418322e-05,
+ "loss": 0.1525,
+ "step": 14037
+ },
+ {
+ "epoch": 38.25068119891008,
+ "grad_norm": 3.0991299152374268,
+ "learning_rate": 1.4160644010229325e-05,
+ "loss": 0.1112,
+ "step": 14038
+ },
+ {
+ "epoch": 38.253405994550405,
+ "grad_norm": 3.934197425842285,
+ "learning_rate": 1.4159841510637368e-05,
+ "loss": 0.1361,
+ "step": 14039
+ },
+ {
+ "epoch": 38.25613079019074,
+ "grad_norm": 3.520111322402954,
+ "learning_rate": 1.4159038978648698e-05,
+ "loss": 0.2246,
+ "step": 14040
+ },
+ {
+ "epoch": 38.258855585831064,
+ "grad_norm": 3.0261425971984863,
+ "learning_rate": 1.4158236414269569e-05,
+ "loss": 0.1066,
+ "step": 14041
+ },
+ {
+ "epoch": 38.26158038147139,
+ "grad_norm": 3.9534575939178467,
+ "learning_rate": 1.4157433817506224e-05,
+ "loss": 0.1273,
+ "step": 14042
+ },
+ {
+ "epoch": 38.264305177111716,
+ "grad_norm": 3.1399805545806885,
+ "learning_rate": 1.4156631188364922e-05,
+ "loss": 0.1309,
+ "step": 14043
+ },
+ {
+ "epoch": 38.26702997275204,
+ "grad_norm": 3.619544506072998,
+ "learning_rate": 1.4155828526851907e-05,
+ "loss": 0.1056,
+ "step": 14044
+ },
+ {
+ "epoch": 38.26975476839237,
+ "grad_norm": 3.345036268234253,
+ "learning_rate": 1.4155025832973435e-05,
+ "loss": 0.1008,
+ "step": 14045
+ },
+ {
+ "epoch": 38.2724795640327,
+ "grad_norm": 3.0487160682678223,
+ "learning_rate": 1.4154223106735754e-05,
+ "loss": 0.1803,
+ "step": 14046
+ },
+ {
+ "epoch": 38.275204359673026,
+ "grad_norm": 3.3610665798187256,
+ "learning_rate": 1.415342034814512e-05,
+ "loss": 0.3279,
+ "step": 14047
+ },
+ {
+ "epoch": 38.27792915531335,
+ "grad_norm": 3.7378499507904053,
+ "learning_rate": 1.4152617557207779e-05,
+ "loss": 0.1471,
+ "step": 14048
+ },
+ {
+ "epoch": 38.28065395095368,
+ "grad_norm": 3.2335946559906006,
+ "learning_rate": 1.4151814733929989e-05,
+ "loss": 0.0901,
+ "step": 14049
+ },
+ {
+ "epoch": 38.283378746594,
+ "grad_norm": 3.7182748317718506,
+ "learning_rate": 1.4151011878317998e-05,
+ "loss": 0.1332,
+ "step": 14050
+ },
+ {
+ "epoch": 38.28610354223433,
+ "grad_norm": 3.2232236862182617,
+ "learning_rate": 1.415020899037806e-05,
+ "loss": 0.1323,
+ "step": 14051
+ },
+ {
+ "epoch": 38.28882833787466,
+ "grad_norm": 3.487349271774292,
+ "learning_rate": 1.4149406070116429e-05,
+ "loss": 0.212,
+ "step": 14052
+ },
+ {
+ "epoch": 38.29155313351499,
+ "grad_norm": 3.5872509479522705,
+ "learning_rate": 1.4148603117539354e-05,
+ "loss": 0.1007,
+ "step": 14053
+ },
+ {
+ "epoch": 38.294277929155314,
+ "grad_norm": 3.2778103351593018,
+ "learning_rate": 1.4147800132653093e-05,
+ "loss": 0.1906,
+ "step": 14054
+ },
+ {
+ "epoch": 38.29700272479564,
+ "grad_norm": 3.4826419353485107,
+ "learning_rate": 1.4146997115463901e-05,
+ "loss": 0.3082,
+ "step": 14055
+ },
+ {
+ "epoch": 38.299727520435965,
+ "grad_norm": 3.402775287628174,
+ "learning_rate": 1.4146194065978024e-05,
+ "loss": 0.2868,
+ "step": 14056
+ },
+ {
+ "epoch": 38.30245231607629,
+ "grad_norm": 2.4969677925109863,
+ "learning_rate": 1.4145390984201726e-05,
+ "loss": 0.0622,
+ "step": 14057
+ },
+ {
+ "epoch": 38.305177111716624,
+ "grad_norm": 3.5716655254364014,
+ "learning_rate": 1.4144587870141252e-05,
+ "loss": 0.2035,
+ "step": 14058
+ },
+ {
+ "epoch": 38.30790190735695,
+ "grad_norm": 4.016568660736084,
+ "learning_rate": 1.4143784723802865e-05,
+ "loss": 0.1863,
+ "step": 14059
+ },
+ {
+ "epoch": 38.310626702997276,
+ "grad_norm": 3.8425798416137695,
+ "learning_rate": 1.4142981545192814e-05,
+ "loss": 0.1159,
+ "step": 14060
+ },
+ {
+ "epoch": 38.3133514986376,
+ "grad_norm": 2.9862210750579834,
+ "learning_rate": 1.4142178334317356e-05,
+ "loss": 0.1891,
+ "step": 14061
+ },
+ {
+ "epoch": 38.31607629427793,
+ "grad_norm": 3.394881010055542,
+ "learning_rate": 1.4141375091182746e-05,
+ "loss": 0.1497,
+ "step": 14062
+ },
+ {
+ "epoch": 38.31880108991825,
+ "grad_norm": 5.067387104034424,
+ "learning_rate": 1.4140571815795242e-05,
+ "loss": 0.1472,
+ "step": 14063
+ },
+ {
+ "epoch": 38.321525885558586,
+ "grad_norm": 3.3960158824920654,
+ "learning_rate": 1.4139768508161097e-05,
+ "loss": 0.3012,
+ "step": 14064
+ },
+ {
+ "epoch": 38.32425068119891,
+ "grad_norm": 4.145744800567627,
+ "learning_rate": 1.4138965168286567e-05,
+ "loss": 0.1406,
+ "step": 14065
+ },
+ {
+ "epoch": 38.32697547683924,
+ "grad_norm": 5.199387550354004,
+ "learning_rate": 1.4138161796177909e-05,
+ "loss": 0.2895,
+ "step": 14066
+ },
+ {
+ "epoch": 38.32970027247956,
+ "grad_norm": 3.133888006210327,
+ "learning_rate": 1.4137358391841383e-05,
+ "loss": 0.1291,
+ "step": 14067
+ },
+ {
+ "epoch": 38.33242506811989,
+ "grad_norm": 4.087012767791748,
+ "learning_rate": 1.413655495528324e-05,
+ "loss": 0.134,
+ "step": 14068
+ },
+ {
+ "epoch": 38.335149863760215,
+ "grad_norm": 4.342003345489502,
+ "learning_rate": 1.4135751486509743e-05,
+ "loss": 0.274,
+ "step": 14069
+ },
+ {
+ "epoch": 38.33787465940055,
+ "grad_norm": 2.9868197441101074,
+ "learning_rate": 1.4134947985527143e-05,
+ "loss": 0.1433,
+ "step": 14070
+ },
+ {
+ "epoch": 38.34059945504087,
+ "grad_norm": 3.526642322540283,
+ "learning_rate": 1.4134144452341702e-05,
+ "loss": 0.2502,
+ "step": 14071
+ },
+ {
+ "epoch": 38.3433242506812,
+ "grad_norm": 3.386446952819824,
+ "learning_rate": 1.4133340886959678e-05,
+ "loss": 0.1227,
+ "step": 14072
+ },
+ {
+ "epoch": 38.346049046321525,
+ "grad_norm": 3.277831792831421,
+ "learning_rate": 1.4132537289387331e-05,
+ "loss": 0.1377,
+ "step": 14073
+ },
+ {
+ "epoch": 38.34877384196185,
+ "grad_norm": 3.315495491027832,
+ "learning_rate": 1.4131733659630913e-05,
+ "loss": 0.1386,
+ "step": 14074
+ },
+ {
+ "epoch": 38.35149863760218,
+ "grad_norm": 3.7335731983184814,
+ "learning_rate": 1.4130929997696688e-05,
+ "loss": 0.2096,
+ "step": 14075
+ },
+ {
+ "epoch": 38.35422343324251,
+ "grad_norm": 3.2648515701293945,
+ "learning_rate": 1.4130126303590911e-05,
+ "loss": 0.2482,
+ "step": 14076
+ },
+ {
+ "epoch": 38.356948228882835,
+ "grad_norm": 4.050138473510742,
+ "learning_rate": 1.4129322577319846e-05,
+ "loss": 0.1532,
+ "step": 14077
+ },
+ {
+ "epoch": 38.35967302452316,
+ "grad_norm": 3.956148147583008,
+ "learning_rate": 1.4128518818889747e-05,
+ "loss": 0.3308,
+ "step": 14078
+ },
+ {
+ "epoch": 38.36239782016349,
+ "grad_norm": 3.575655937194824,
+ "learning_rate": 1.4127715028306879e-05,
+ "loss": 0.0945,
+ "step": 14079
+ },
+ {
+ "epoch": 38.36512261580381,
+ "grad_norm": 2.982895851135254,
+ "learning_rate": 1.4126911205577495e-05,
+ "loss": 0.1234,
+ "step": 14080
+ },
+ {
+ "epoch": 38.36784741144414,
+ "grad_norm": 3.854867696762085,
+ "learning_rate": 1.4126107350707865e-05,
+ "loss": 0.1998,
+ "step": 14081
+ },
+ {
+ "epoch": 38.37057220708447,
+ "grad_norm": 3.97391414642334,
+ "learning_rate": 1.4125303463704241e-05,
+ "loss": 0.1184,
+ "step": 14082
+ },
+ {
+ "epoch": 38.3732970027248,
+ "grad_norm": 3.984618663787842,
+ "learning_rate": 1.4124499544572887e-05,
+ "loss": 0.2416,
+ "step": 14083
+ },
+ {
+ "epoch": 38.37602179836512,
+ "grad_norm": 3.8586905002593994,
+ "learning_rate": 1.4123695593320062e-05,
+ "loss": 0.1386,
+ "step": 14084
+ },
+ {
+ "epoch": 38.37874659400545,
+ "grad_norm": 3.3823049068450928,
+ "learning_rate": 1.4122891609952029e-05,
+ "loss": 0.1723,
+ "step": 14085
+ },
+ {
+ "epoch": 38.381471389645775,
+ "grad_norm": 3.272404909133911,
+ "learning_rate": 1.4122087594475049e-05,
+ "loss": 0.1025,
+ "step": 14086
+ },
+ {
+ "epoch": 38.3841961852861,
+ "grad_norm": 3.443622589111328,
+ "learning_rate": 1.4121283546895384e-05,
+ "loss": 0.1549,
+ "step": 14087
+ },
+ {
+ "epoch": 38.38692098092643,
+ "grad_norm": 4.269765377044678,
+ "learning_rate": 1.4120479467219292e-05,
+ "loss": 0.1114,
+ "step": 14088
+ },
+ {
+ "epoch": 38.38964577656676,
+ "grad_norm": 3.532210111618042,
+ "learning_rate": 1.4119675355453044e-05,
+ "loss": 0.1334,
+ "step": 14089
+ },
+ {
+ "epoch": 38.392370572207085,
+ "grad_norm": 3.226598024368286,
+ "learning_rate": 1.4118871211602893e-05,
+ "loss": 0.0941,
+ "step": 14090
+ },
+ {
+ "epoch": 38.39509536784741,
+ "grad_norm": 3.2404732704162598,
+ "learning_rate": 1.411806703567511e-05,
+ "loss": 0.1148,
+ "step": 14091
+ },
+ {
+ "epoch": 38.39782016348774,
+ "grad_norm": 2.952805995941162,
+ "learning_rate": 1.4117262827675946e-05,
+ "loss": 0.0711,
+ "step": 14092
+ },
+ {
+ "epoch": 38.40054495912806,
+ "grad_norm": 3.5664849281311035,
+ "learning_rate": 1.4116458587611678e-05,
+ "loss": 0.1479,
+ "step": 14093
+ },
+ {
+ "epoch": 38.403269754768395,
+ "grad_norm": 3.5273334980010986,
+ "learning_rate": 1.411565431548856e-05,
+ "loss": 0.1562,
+ "step": 14094
+ },
+ {
+ "epoch": 38.40599455040872,
+ "grad_norm": 3.8034098148345947,
+ "learning_rate": 1.4114850011312861e-05,
+ "loss": 0.2965,
+ "step": 14095
+ },
+ {
+ "epoch": 38.40871934604905,
+ "grad_norm": 3.9778847694396973,
+ "learning_rate": 1.411404567509084e-05,
+ "loss": 0.2363,
+ "step": 14096
+ },
+ {
+ "epoch": 38.41144414168937,
+ "grad_norm": 4.51084566116333,
+ "learning_rate": 1.4113241306828764e-05,
+ "loss": 0.1745,
+ "step": 14097
+ },
+ {
+ "epoch": 38.4141689373297,
+ "grad_norm": 3.8210151195526123,
+ "learning_rate": 1.4112436906532898e-05,
+ "loss": 0.1138,
+ "step": 14098
+ },
+ {
+ "epoch": 38.416893732970024,
+ "grad_norm": 2.6526262760162354,
+ "learning_rate": 1.4111632474209506e-05,
+ "loss": 0.1077,
+ "step": 14099
+ },
+ {
+ "epoch": 38.41961852861036,
+ "grad_norm": 2.829176664352417,
+ "learning_rate": 1.4110828009864853e-05,
+ "loss": 0.1169,
+ "step": 14100
+ },
+ {
+ "epoch": 38.42234332425068,
+ "grad_norm": 3.0856497287750244,
+ "learning_rate": 1.4110023513505201e-05,
+ "loss": 0.1114,
+ "step": 14101
+ },
+ {
+ "epoch": 38.42506811989101,
+ "grad_norm": 2.8988537788391113,
+ "learning_rate": 1.410921898513682e-05,
+ "loss": 0.0881,
+ "step": 14102
+ },
+ {
+ "epoch": 38.427792915531334,
+ "grad_norm": 4.134519100189209,
+ "learning_rate": 1.4108414424765978e-05,
+ "loss": 0.124,
+ "step": 14103
+ },
+ {
+ "epoch": 38.43051771117166,
+ "grad_norm": 3.695312738418579,
+ "learning_rate": 1.410760983239893e-05,
+ "loss": 0.2874,
+ "step": 14104
+ },
+ {
+ "epoch": 38.433242506811986,
+ "grad_norm": 3.5157079696655273,
+ "learning_rate": 1.4106805208041951e-05,
+ "loss": 0.1421,
+ "step": 14105
+ },
+ {
+ "epoch": 38.43596730245232,
+ "grad_norm": 3.6636745929718018,
+ "learning_rate": 1.4106000551701309e-05,
+ "loss": 0.0968,
+ "step": 14106
+ },
+ {
+ "epoch": 38.438692098092645,
+ "grad_norm": 2.5052852630615234,
+ "learning_rate": 1.4105195863383261e-05,
+ "loss": 0.0903,
+ "step": 14107
+ },
+ {
+ "epoch": 38.44141689373297,
+ "grad_norm": 12.370586395263672,
+ "learning_rate": 1.4104391143094086e-05,
+ "loss": 0.1335,
+ "step": 14108
+ },
+ {
+ "epoch": 38.444141689373296,
+ "grad_norm": 3.0852222442626953,
+ "learning_rate": 1.4103586390840038e-05,
+ "loss": 0.2783,
+ "step": 14109
+ },
+ {
+ "epoch": 38.44686648501362,
+ "grad_norm": 4.050045013427734,
+ "learning_rate": 1.4102781606627393e-05,
+ "loss": 0.1303,
+ "step": 14110
+ },
+ {
+ "epoch": 38.44959128065395,
+ "grad_norm": 3.53559947013855,
+ "learning_rate": 1.4101976790462421e-05,
+ "loss": 0.2197,
+ "step": 14111
+ },
+ {
+ "epoch": 38.45231607629428,
+ "grad_norm": 3.3790817260742188,
+ "learning_rate": 1.4101171942351381e-05,
+ "loss": 0.1668,
+ "step": 14112
+ },
+ {
+ "epoch": 38.45504087193461,
+ "grad_norm": 3.6089046001434326,
+ "learning_rate": 1.4100367062300548e-05,
+ "loss": 0.1647,
+ "step": 14113
+ },
+ {
+ "epoch": 38.45776566757493,
+ "grad_norm": 4.095571994781494,
+ "learning_rate": 1.409956215031619e-05,
+ "loss": 0.0996,
+ "step": 14114
+ },
+ {
+ "epoch": 38.46049046321526,
+ "grad_norm": 3.2654786109924316,
+ "learning_rate": 1.4098757206404572e-05,
+ "loss": 0.1357,
+ "step": 14115
+ },
+ {
+ "epoch": 38.463215258855584,
+ "grad_norm": 4.026064395904541,
+ "learning_rate": 1.4097952230571965e-05,
+ "loss": 0.31,
+ "step": 14116
+ },
+ {
+ "epoch": 38.46594005449591,
+ "grad_norm": 3.5454607009887695,
+ "learning_rate": 1.4097147222824639e-05,
+ "loss": 0.195,
+ "step": 14117
+ },
+ {
+ "epoch": 38.46866485013624,
+ "grad_norm": 3.3974058628082275,
+ "learning_rate": 1.4096342183168861e-05,
+ "loss": 0.1252,
+ "step": 14118
+ },
+ {
+ "epoch": 38.47138964577657,
+ "grad_norm": 3.9315719604492188,
+ "learning_rate": 1.4095537111610903e-05,
+ "loss": 0.1674,
+ "step": 14119
+ },
+ {
+ "epoch": 38.474114441416894,
+ "grad_norm": 3.9505958557128906,
+ "learning_rate": 1.4094732008157032e-05,
+ "loss": 0.1175,
+ "step": 14120
+ },
+ {
+ "epoch": 38.47683923705722,
+ "grad_norm": 3.608952283859253,
+ "learning_rate": 1.409392687281352e-05,
+ "loss": 0.1944,
+ "step": 14121
+ },
+ {
+ "epoch": 38.479564032697546,
+ "grad_norm": 3.767322540283203,
+ "learning_rate": 1.4093121705586637e-05,
+ "loss": 0.113,
+ "step": 14122
+ },
+ {
+ "epoch": 38.48228882833787,
+ "grad_norm": 3.8934152126312256,
+ "learning_rate": 1.4092316506482656e-05,
+ "loss": 0.2015,
+ "step": 14123
+ },
+ {
+ "epoch": 38.485013623978205,
+ "grad_norm": 3.617312431335449,
+ "learning_rate": 1.4091511275507845e-05,
+ "loss": 0.126,
+ "step": 14124
+ },
+ {
+ "epoch": 38.48773841961853,
+ "grad_norm": 2.7508749961853027,
+ "learning_rate": 1.4090706012668474e-05,
+ "loss": 0.0674,
+ "step": 14125
+ },
+ {
+ "epoch": 38.490463215258856,
+ "grad_norm": 4.594725608825684,
+ "learning_rate": 1.4089900717970819e-05,
+ "loss": 0.2272,
+ "step": 14126
+ },
+ {
+ "epoch": 38.49318801089918,
+ "grad_norm": 3.491703510284424,
+ "learning_rate": 1.4089095391421148e-05,
+ "loss": 0.2058,
+ "step": 14127
+ },
+ {
+ "epoch": 38.49591280653951,
+ "grad_norm": 3.9597549438476562,
+ "learning_rate": 1.4088290033025736e-05,
+ "loss": 0.1568,
+ "step": 14128
+ },
+ {
+ "epoch": 38.49863760217983,
+ "grad_norm": 2.3482649326324463,
+ "learning_rate": 1.4087484642790847e-05,
+ "loss": 0.0574,
+ "step": 14129
+ },
+ {
+ "epoch": 38.50136239782017,
+ "grad_norm": 4.405796051025391,
+ "learning_rate": 1.4086679220722765e-05,
+ "loss": 0.144,
+ "step": 14130
+ },
+ {
+ "epoch": 38.50408719346049,
+ "grad_norm": 4.322147369384766,
+ "learning_rate": 1.4085873766827751e-05,
+ "loss": 0.1557,
+ "step": 14131
+ },
+ {
+ "epoch": 38.50681198910082,
+ "grad_norm": 3.899132490158081,
+ "learning_rate": 1.408506828111209e-05,
+ "loss": 0.1666,
+ "step": 14132
+ },
+ {
+ "epoch": 38.509536784741144,
+ "grad_norm": 3.763018846511841,
+ "learning_rate": 1.4084262763582044e-05,
+ "loss": 0.3092,
+ "step": 14133
+ },
+ {
+ "epoch": 38.51226158038147,
+ "grad_norm": 4.216452121734619,
+ "learning_rate": 1.4083457214243891e-05,
+ "loss": 0.299,
+ "step": 14134
+ },
+ {
+ "epoch": 38.514986376021795,
+ "grad_norm": 5.236637592315674,
+ "learning_rate": 1.4082651633103908e-05,
+ "loss": 0.1125,
+ "step": 14135
+ },
+ {
+ "epoch": 38.51771117166213,
+ "grad_norm": 3.5705230236053467,
+ "learning_rate": 1.4081846020168363e-05,
+ "loss": 0.1856,
+ "step": 14136
+ },
+ {
+ "epoch": 38.520435967302454,
+ "grad_norm": 3.374195098876953,
+ "learning_rate": 1.4081040375443533e-05,
+ "loss": 0.1232,
+ "step": 14137
+ },
+ {
+ "epoch": 38.52316076294278,
+ "grad_norm": 2.988133430480957,
+ "learning_rate": 1.4080234698935693e-05,
+ "loss": 0.2142,
+ "step": 14138
+ },
+ {
+ "epoch": 38.525885558583106,
+ "grad_norm": 3.82804012298584,
+ "learning_rate": 1.4079428990651113e-05,
+ "loss": 0.1228,
+ "step": 14139
+ },
+ {
+ "epoch": 38.52861035422343,
+ "grad_norm": 3.421844482421875,
+ "learning_rate": 1.4078623250596075e-05,
+ "loss": 0.2056,
+ "step": 14140
+ },
+ {
+ "epoch": 38.53133514986376,
+ "grad_norm": 2.616758108139038,
+ "learning_rate": 1.407781747877685e-05,
+ "loss": 0.0646,
+ "step": 14141
+ },
+ {
+ "epoch": 38.53405994550409,
+ "grad_norm": 3.7466742992401123,
+ "learning_rate": 1.4077011675199713e-05,
+ "loss": 0.098,
+ "step": 14142
+ },
+ {
+ "epoch": 38.536784741144416,
+ "grad_norm": 3.2858541011810303,
+ "learning_rate": 1.407620583987094e-05,
+ "loss": 0.12,
+ "step": 14143
+ },
+ {
+ "epoch": 38.53950953678474,
+ "grad_norm": 3.852754831314087,
+ "learning_rate": 1.4075399972796808e-05,
+ "loss": 0.2134,
+ "step": 14144
+ },
+ {
+ "epoch": 38.54223433242507,
+ "grad_norm": 3.5313684940338135,
+ "learning_rate": 1.4074594073983592e-05,
+ "loss": 0.1515,
+ "step": 14145
+ },
+ {
+ "epoch": 38.54495912806539,
+ "grad_norm": 3.3226211071014404,
+ "learning_rate": 1.4073788143437568e-05,
+ "loss": 0.1385,
+ "step": 14146
+ },
+ {
+ "epoch": 38.54768392370572,
+ "grad_norm": 3.3102924823760986,
+ "learning_rate": 1.4072982181165014e-05,
+ "loss": 0.319,
+ "step": 14147
+ },
+ {
+ "epoch": 38.55040871934605,
+ "grad_norm": 4.194208145141602,
+ "learning_rate": 1.4072176187172204e-05,
+ "loss": 0.1247,
+ "step": 14148
+ },
+ {
+ "epoch": 38.55313351498638,
+ "grad_norm": 4.211638450622559,
+ "learning_rate": 1.4071370161465417e-05,
+ "loss": 0.1737,
+ "step": 14149
+ },
+ {
+ "epoch": 38.555858310626704,
+ "grad_norm": 3.292100667953491,
+ "learning_rate": 1.407056410405093e-05,
+ "loss": 0.2278,
+ "step": 14150
+ },
+ {
+ "epoch": 38.55858310626703,
+ "grad_norm": 4.150158405303955,
+ "learning_rate": 1.4069758014935023e-05,
+ "loss": 0.1026,
+ "step": 14151
+ },
+ {
+ "epoch": 38.561307901907355,
+ "grad_norm": 6.109641075134277,
+ "learning_rate": 1.406895189412397e-05,
+ "loss": 0.1116,
+ "step": 14152
+ },
+ {
+ "epoch": 38.56403269754768,
+ "grad_norm": 2.741910219192505,
+ "learning_rate": 1.406814574162405e-05,
+ "loss": 0.1162,
+ "step": 14153
+ },
+ {
+ "epoch": 38.566757493188014,
+ "grad_norm": 3.2188010215759277,
+ "learning_rate": 1.4067339557441546e-05,
+ "loss": 0.276,
+ "step": 14154
+ },
+ {
+ "epoch": 38.56948228882834,
+ "grad_norm": 4.283053398132324,
+ "learning_rate": 1.406653334158273e-05,
+ "loss": 0.1499,
+ "step": 14155
+ },
+ {
+ "epoch": 38.572207084468666,
+ "grad_norm": 3.4578025341033936,
+ "learning_rate": 1.406572709405388e-05,
+ "loss": 0.1433,
+ "step": 14156
+ },
+ {
+ "epoch": 38.57493188010899,
+ "grad_norm": 3.7715351581573486,
+ "learning_rate": 1.406492081486128e-05,
+ "loss": 0.1128,
+ "step": 14157
+ },
+ {
+ "epoch": 38.57765667574932,
+ "grad_norm": 3.338750123977661,
+ "learning_rate": 1.406411450401121e-05,
+ "loss": 0.1857,
+ "step": 14158
+ },
+ {
+ "epoch": 38.58038147138964,
+ "grad_norm": 4.0570759773254395,
+ "learning_rate": 1.4063308161509945e-05,
+ "loss": 0.1347,
+ "step": 14159
+ },
+ {
+ "epoch": 38.583106267029976,
+ "grad_norm": 4.748410701751709,
+ "learning_rate": 1.406250178736377e-05,
+ "loss": 0.1002,
+ "step": 14160
+ },
+ {
+ "epoch": 38.5858310626703,
+ "grad_norm": 3.248582124710083,
+ "learning_rate": 1.4061695381578957e-05,
+ "loss": 0.2823,
+ "step": 14161
+ },
+ {
+ "epoch": 38.58855585831063,
+ "grad_norm": 3.2258379459381104,
+ "learning_rate": 1.4060888944161795e-05,
+ "loss": 0.0795,
+ "step": 14162
+ },
+ {
+ "epoch": 38.59128065395095,
+ "grad_norm": 3.317054033279419,
+ "learning_rate": 1.4060082475118558e-05,
+ "loss": 0.1885,
+ "step": 14163
+ },
+ {
+ "epoch": 38.59400544959128,
+ "grad_norm": 2.9410512447357178,
+ "learning_rate": 1.4059275974455531e-05,
+ "loss": 0.208,
+ "step": 14164
+ },
+ {
+ "epoch": 38.596730245231605,
+ "grad_norm": 3.056736946105957,
+ "learning_rate": 1.4058469442178987e-05,
+ "loss": 0.2118,
+ "step": 14165
+ },
+ {
+ "epoch": 38.59945504087194,
+ "grad_norm": 3.7396657466888428,
+ "learning_rate": 1.4057662878295221e-05,
+ "loss": 0.1424,
+ "step": 14166
+ },
+ {
+ "epoch": 38.60217983651226,
+ "grad_norm": 3.341968536376953,
+ "learning_rate": 1.4056856282810505e-05,
+ "loss": 0.2172,
+ "step": 14167
+ },
+ {
+ "epoch": 38.60490463215259,
+ "grad_norm": 4.26236629486084,
+ "learning_rate": 1.4056049655731125e-05,
+ "loss": 0.2486,
+ "step": 14168
+ },
+ {
+ "epoch": 38.607629427792915,
+ "grad_norm": 3.9045603275299072,
+ "learning_rate": 1.4055242997063359e-05,
+ "loss": 0.1174,
+ "step": 14169
+ },
+ {
+ "epoch": 38.61035422343324,
+ "grad_norm": 3.2555673122406006,
+ "learning_rate": 1.4054436306813491e-05,
+ "loss": 0.1634,
+ "step": 14170
+ },
+ {
+ "epoch": 38.61307901907357,
+ "grad_norm": 3.620532751083374,
+ "learning_rate": 1.4053629584987802e-05,
+ "loss": 0.0899,
+ "step": 14171
+ },
+ {
+ "epoch": 38.6158038147139,
+ "grad_norm": 4.274932384490967,
+ "learning_rate": 1.405282283159258e-05,
+ "loss": 0.1334,
+ "step": 14172
+ },
+ {
+ "epoch": 38.618528610354225,
+ "grad_norm": 4.793648719787598,
+ "learning_rate": 1.4052016046634101e-05,
+ "loss": 0.1184,
+ "step": 14173
+ },
+ {
+ "epoch": 38.62125340599455,
+ "grad_norm": 3.1074957847595215,
+ "learning_rate": 1.4051209230118652e-05,
+ "loss": 0.2457,
+ "step": 14174
+ },
+ {
+ "epoch": 38.62397820163488,
+ "grad_norm": 3.921602964401245,
+ "learning_rate": 1.4050402382052514e-05,
+ "loss": 0.202,
+ "step": 14175
+ },
+ {
+ "epoch": 38.6267029972752,
+ "grad_norm": 3.6080305576324463,
+ "learning_rate": 1.4049595502441977e-05,
+ "loss": 0.1673,
+ "step": 14176
+ },
+ {
+ "epoch": 38.62942779291553,
+ "grad_norm": 4.508090972900391,
+ "learning_rate": 1.4048788591293318e-05,
+ "loss": 0.1927,
+ "step": 14177
+ },
+ {
+ "epoch": 38.63215258855586,
+ "grad_norm": 3.541712999343872,
+ "learning_rate": 1.4047981648612824e-05,
+ "loss": 0.1287,
+ "step": 14178
+ },
+ {
+ "epoch": 38.63487738419619,
+ "grad_norm": 3.69193172454834,
+ "learning_rate": 1.4047174674406781e-05,
+ "loss": 0.1346,
+ "step": 14179
+ },
+ {
+ "epoch": 38.63760217983651,
+ "grad_norm": 3.9724159240722656,
+ "learning_rate": 1.4046367668681471e-05,
+ "loss": 0.1795,
+ "step": 14180
+ },
+ {
+ "epoch": 38.64032697547684,
+ "grad_norm": 7.9792304039001465,
+ "learning_rate": 1.4045560631443178e-05,
+ "loss": 0.1594,
+ "step": 14181
+ },
+ {
+ "epoch": 38.643051771117165,
+ "grad_norm": 3.7866296768188477,
+ "learning_rate": 1.4044753562698188e-05,
+ "loss": 0.1183,
+ "step": 14182
+ },
+ {
+ "epoch": 38.64577656675749,
+ "grad_norm": 26.007862091064453,
+ "learning_rate": 1.4043946462452791e-05,
+ "loss": 0.2331,
+ "step": 14183
+ },
+ {
+ "epoch": 38.64850136239782,
+ "grad_norm": 2.849698781967163,
+ "learning_rate": 1.4043139330713267e-05,
+ "loss": 0.0728,
+ "step": 14184
+ },
+ {
+ "epoch": 38.65122615803815,
+ "grad_norm": 3.248002290725708,
+ "learning_rate": 1.4042332167485904e-05,
+ "loss": 0.0927,
+ "step": 14185
+ },
+ {
+ "epoch": 38.653950953678475,
+ "grad_norm": 3.942734956741333,
+ "learning_rate": 1.4041524972776989e-05,
+ "loss": 0.2007,
+ "step": 14186
+ },
+ {
+ "epoch": 38.6566757493188,
+ "grad_norm": 3.2506842613220215,
+ "learning_rate": 1.404071774659281e-05,
+ "loss": 0.107,
+ "step": 14187
+ },
+ {
+ "epoch": 38.65940054495913,
+ "grad_norm": 4.201396942138672,
+ "learning_rate": 1.4039910488939648e-05,
+ "loss": 0.3712,
+ "step": 14188
+ },
+ {
+ "epoch": 38.66212534059945,
+ "grad_norm": 3.947938919067383,
+ "learning_rate": 1.4039103199823794e-05,
+ "loss": 0.2209,
+ "step": 14189
+ },
+ {
+ "epoch": 38.664850136239785,
+ "grad_norm": 3.4256365299224854,
+ "learning_rate": 1.4038295879251535e-05,
+ "loss": 0.1008,
+ "step": 14190
+ },
+ {
+ "epoch": 38.66757493188011,
+ "grad_norm": 3.2697176933288574,
+ "learning_rate": 1.4037488527229156e-05,
+ "loss": 0.0955,
+ "step": 14191
+ },
+ {
+ "epoch": 38.67029972752044,
+ "grad_norm": 5.0567193031311035,
+ "learning_rate": 1.4036681143762945e-05,
+ "loss": 0.2175,
+ "step": 14192
+ },
+ {
+ "epoch": 38.67302452316076,
+ "grad_norm": 3.81673526763916,
+ "learning_rate": 1.4035873728859195e-05,
+ "loss": 0.3152,
+ "step": 14193
+ },
+ {
+ "epoch": 38.67574931880109,
+ "grad_norm": 2.870438575744629,
+ "learning_rate": 1.4035066282524188e-05,
+ "loss": 0.0627,
+ "step": 14194
+ },
+ {
+ "epoch": 38.678474114441414,
+ "grad_norm": 3.1877613067626953,
+ "learning_rate": 1.4034258804764217e-05,
+ "loss": 0.1245,
+ "step": 14195
+ },
+ {
+ "epoch": 38.68119891008175,
+ "grad_norm": 3.4831719398498535,
+ "learning_rate": 1.4033451295585565e-05,
+ "loss": 0.1785,
+ "step": 14196
+ },
+ {
+ "epoch": 38.68392370572207,
+ "grad_norm": 3.7880563735961914,
+ "learning_rate": 1.4032643754994526e-05,
+ "loss": 0.1263,
+ "step": 14197
+ },
+ {
+ "epoch": 38.6866485013624,
+ "grad_norm": 4.438798427581787,
+ "learning_rate": 1.4031836182997387e-05,
+ "loss": 0.1107,
+ "step": 14198
+ },
+ {
+ "epoch": 38.689373297002724,
+ "grad_norm": 3.8279316425323486,
+ "learning_rate": 1.4031028579600436e-05,
+ "loss": 0.1579,
+ "step": 14199
+ },
+ {
+ "epoch": 38.69209809264305,
+ "grad_norm": 3.151992082595825,
+ "learning_rate": 1.4030220944809965e-05,
+ "loss": 0.1023,
+ "step": 14200
+ },
+ {
+ "epoch": 38.694822888283376,
+ "grad_norm": 3.1303155422210693,
+ "learning_rate": 1.4029413278632266e-05,
+ "loss": 0.1888,
+ "step": 14201
+ },
+ {
+ "epoch": 38.69754768392371,
+ "grad_norm": 4.090818405151367,
+ "learning_rate": 1.4028605581073623e-05,
+ "loss": 0.2804,
+ "step": 14202
+ },
+ {
+ "epoch": 38.700272479564035,
+ "grad_norm": 4.053791046142578,
+ "learning_rate": 1.402779785214033e-05,
+ "loss": 0.2383,
+ "step": 14203
+ },
+ {
+ "epoch": 38.70299727520436,
+ "grad_norm": 4.969809532165527,
+ "learning_rate": 1.4026990091838677e-05,
+ "loss": 0.1803,
+ "step": 14204
+ },
+ {
+ "epoch": 38.705722070844686,
+ "grad_norm": 3.148733377456665,
+ "learning_rate": 1.4026182300174955e-05,
+ "loss": 0.1113,
+ "step": 14205
+ },
+ {
+ "epoch": 38.70844686648501,
+ "grad_norm": 3.3885080814361572,
+ "learning_rate": 1.4025374477155453e-05,
+ "loss": 0.131,
+ "step": 14206
+ },
+ {
+ "epoch": 38.71117166212534,
+ "grad_norm": 3.8979718685150146,
+ "learning_rate": 1.4024566622786468e-05,
+ "loss": 0.1262,
+ "step": 14207
+ },
+ {
+ "epoch": 38.71389645776567,
+ "grad_norm": 3.54727840423584,
+ "learning_rate": 1.4023758737074284e-05,
+ "loss": 0.1716,
+ "step": 14208
+ },
+ {
+ "epoch": 38.716621253406,
+ "grad_norm": 2.9371259212493896,
+ "learning_rate": 1.4022950820025198e-05,
+ "loss": 0.1163,
+ "step": 14209
+ },
+ {
+ "epoch": 38.71934604904632,
+ "grad_norm": 3.9172444343566895,
+ "learning_rate": 1.40221428716455e-05,
+ "loss": 0.1532,
+ "step": 14210
+ },
+ {
+ "epoch": 38.72207084468665,
+ "grad_norm": 3.5254898071289062,
+ "learning_rate": 1.4021334891941484e-05,
+ "loss": 0.1518,
+ "step": 14211
+ },
+ {
+ "epoch": 38.724795640326974,
+ "grad_norm": 3.2746753692626953,
+ "learning_rate": 1.4020526880919439e-05,
+ "loss": 0.1343,
+ "step": 14212
+ },
+ {
+ "epoch": 38.7275204359673,
+ "grad_norm": 11.781305313110352,
+ "learning_rate": 1.4019718838585664e-05,
+ "loss": 0.1027,
+ "step": 14213
+ },
+ {
+ "epoch": 38.73024523160763,
+ "grad_norm": 3.4439332485198975,
+ "learning_rate": 1.4018910764946445e-05,
+ "loss": 0.2248,
+ "step": 14214
+ },
+ {
+ "epoch": 38.73297002724796,
+ "grad_norm": 4.5536065101623535,
+ "learning_rate": 1.4018102660008079e-05,
+ "loss": 0.1225,
+ "step": 14215
+ },
+ {
+ "epoch": 38.735694822888284,
+ "grad_norm": 3.682762622833252,
+ "learning_rate": 1.4017294523776859e-05,
+ "loss": 0.1651,
+ "step": 14216
+ },
+ {
+ "epoch": 38.73841961852861,
+ "grad_norm": 3.384291648864746,
+ "learning_rate": 1.4016486356259076e-05,
+ "loss": 0.1036,
+ "step": 14217
+ },
+ {
+ "epoch": 38.741144414168936,
+ "grad_norm": 3.9116156101226807,
+ "learning_rate": 1.4015678157461028e-05,
+ "loss": 0.1844,
+ "step": 14218
+ },
+ {
+ "epoch": 38.74386920980926,
+ "grad_norm": 4.035623073577881,
+ "learning_rate": 1.401486992738901e-05,
+ "loss": 0.1654,
+ "step": 14219
+ },
+ {
+ "epoch": 38.746594005449595,
+ "grad_norm": 4.055289268493652,
+ "learning_rate": 1.4014061666049312e-05,
+ "loss": 0.1901,
+ "step": 14220
+ },
+ {
+ "epoch": 38.74931880108992,
+ "grad_norm": 3.1037003993988037,
+ "learning_rate": 1.4013253373448232e-05,
+ "loss": 0.0864,
+ "step": 14221
+ },
+ {
+ "epoch": 38.752043596730246,
+ "grad_norm": 3.087236166000366,
+ "learning_rate": 1.4012445049592063e-05,
+ "loss": 0.2223,
+ "step": 14222
+ },
+ {
+ "epoch": 38.75476839237057,
+ "grad_norm": 3.771319627761841,
+ "learning_rate": 1.4011636694487103e-05,
+ "loss": 0.1149,
+ "step": 14223
+ },
+ {
+ "epoch": 38.7574931880109,
+ "grad_norm": 3.1193697452545166,
+ "learning_rate": 1.4010828308139644e-05,
+ "loss": 0.0897,
+ "step": 14224
+ },
+ {
+ "epoch": 38.76021798365122,
+ "grad_norm": 3.007960319519043,
+ "learning_rate": 1.4010019890555982e-05,
+ "loss": 0.1468,
+ "step": 14225
+ },
+ {
+ "epoch": 38.762942779291556,
+ "grad_norm": 3.209429979324341,
+ "learning_rate": 1.4009211441742413e-05,
+ "loss": 0.2574,
+ "step": 14226
+ },
+ {
+ "epoch": 38.76566757493188,
+ "grad_norm": 3.637150287628174,
+ "learning_rate": 1.4008402961705238e-05,
+ "loss": 0.1484,
+ "step": 14227
+ },
+ {
+ "epoch": 38.76839237057221,
+ "grad_norm": 4.1046462059021,
+ "learning_rate": 1.400759445045075e-05,
+ "loss": 0.1505,
+ "step": 14228
+ },
+ {
+ "epoch": 38.771117166212534,
+ "grad_norm": 4.261979103088379,
+ "learning_rate": 1.4006785907985245e-05,
+ "loss": 0.1531,
+ "step": 14229
+ },
+ {
+ "epoch": 38.77384196185286,
+ "grad_norm": 3.6361358165740967,
+ "learning_rate": 1.4005977334315017e-05,
+ "loss": 0.2741,
+ "step": 14230
+ },
+ {
+ "epoch": 38.776566757493185,
+ "grad_norm": 3.318692445755005,
+ "learning_rate": 1.4005168729446372e-05,
+ "loss": 0.0873,
+ "step": 14231
+ },
+ {
+ "epoch": 38.77929155313352,
+ "grad_norm": 3.783543109893799,
+ "learning_rate": 1.4004360093385599e-05,
+ "loss": 0.217,
+ "step": 14232
+ },
+ {
+ "epoch": 38.782016348773844,
+ "grad_norm": 3.8267745971679688,
+ "learning_rate": 1.4003551426138998e-05,
+ "loss": 0.1379,
+ "step": 14233
+ },
+ {
+ "epoch": 38.78474114441417,
+ "grad_norm": 3.2316293716430664,
+ "learning_rate": 1.4002742727712863e-05,
+ "loss": 0.2041,
+ "step": 14234
+ },
+ {
+ "epoch": 38.787465940054496,
+ "grad_norm": 3.5665860176086426,
+ "learning_rate": 1.4001933998113504e-05,
+ "loss": 0.1132,
+ "step": 14235
+ },
+ {
+ "epoch": 38.79019073569482,
+ "grad_norm": 4.136230945587158,
+ "learning_rate": 1.4001125237347206e-05,
+ "loss": 0.1395,
+ "step": 14236
+ },
+ {
+ "epoch": 38.79291553133515,
+ "grad_norm": 3.4491498470306396,
+ "learning_rate": 1.4000316445420279e-05,
+ "loss": 0.1244,
+ "step": 14237
+ },
+ {
+ "epoch": 38.79564032697548,
+ "grad_norm": 3.396529197692871,
+ "learning_rate": 1.3999507622339013e-05,
+ "loss": 0.1009,
+ "step": 14238
+ },
+ {
+ "epoch": 38.798365122615806,
+ "grad_norm": 3.7777254581451416,
+ "learning_rate": 1.399869876810971e-05,
+ "loss": 0.2244,
+ "step": 14239
+ },
+ {
+ "epoch": 38.80108991825613,
+ "grad_norm": 3.855440139770508,
+ "learning_rate": 1.3997889882738671e-05,
+ "loss": 0.1579,
+ "step": 14240
+ },
+ {
+ "epoch": 38.80381471389646,
+ "grad_norm": 3.2241404056549072,
+ "learning_rate": 1.3997080966232195e-05,
+ "loss": 0.0724,
+ "step": 14241
+ },
+ {
+ "epoch": 38.80653950953678,
+ "grad_norm": 4.223465442657471,
+ "learning_rate": 1.399627201859658e-05,
+ "loss": 0.1087,
+ "step": 14242
+ },
+ {
+ "epoch": 38.80926430517711,
+ "grad_norm": 3.123835802078247,
+ "learning_rate": 1.3995463039838123e-05,
+ "loss": 0.1063,
+ "step": 14243
+ },
+ {
+ "epoch": 38.81198910081744,
+ "grad_norm": 3.682858467102051,
+ "learning_rate": 1.3994654029963132e-05,
+ "loss": 0.1964,
+ "step": 14244
+ },
+ {
+ "epoch": 38.81471389645777,
+ "grad_norm": 3.387737274169922,
+ "learning_rate": 1.3993844988977907e-05,
+ "loss": 0.1563,
+ "step": 14245
+ },
+ {
+ "epoch": 38.817438692098094,
+ "grad_norm": 3.8116676807403564,
+ "learning_rate": 1.3993035916888742e-05,
+ "loss": 0.1737,
+ "step": 14246
+ },
+ {
+ "epoch": 38.82016348773842,
+ "grad_norm": 4.3365607261657715,
+ "learning_rate": 1.3992226813701946e-05,
+ "loss": 0.1547,
+ "step": 14247
+ },
+ {
+ "epoch": 38.822888283378745,
+ "grad_norm": 3.5087904930114746,
+ "learning_rate": 1.399141767942381e-05,
+ "loss": 0.1202,
+ "step": 14248
+ },
+ {
+ "epoch": 38.82561307901907,
+ "grad_norm": 2.4336130619049072,
+ "learning_rate": 1.3990608514060646e-05,
+ "loss": 0.1804,
+ "step": 14249
+ },
+ {
+ "epoch": 38.828337874659404,
+ "grad_norm": 3.403745651245117,
+ "learning_rate": 1.3989799317618751e-05,
+ "loss": 0.146,
+ "step": 14250
+ },
+ {
+ "epoch": 38.83106267029973,
+ "grad_norm": 3.0786287784576416,
+ "learning_rate": 1.3988990090104424e-05,
+ "loss": 0.1428,
+ "step": 14251
+ },
+ {
+ "epoch": 38.833787465940055,
+ "grad_norm": 3.9339845180511475,
+ "learning_rate": 1.3988180831523972e-05,
+ "loss": 0.1017,
+ "step": 14252
+ },
+ {
+ "epoch": 38.83651226158038,
+ "grad_norm": 3.437091112136841,
+ "learning_rate": 1.3987371541883699e-05,
+ "loss": 0.1278,
+ "step": 14253
+ },
+ {
+ "epoch": 38.83923705722071,
+ "grad_norm": 3.5714242458343506,
+ "learning_rate": 1.3986562221189902e-05,
+ "loss": 0.1632,
+ "step": 14254
+ },
+ {
+ "epoch": 38.84196185286103,
+ "grad_norm": 3.8012349605560303,
+ "learning_rate": 1.3985752869448888e-05,
+ "loss": 0.2293,
+ "step": 14255
+ },
+ {
+ "epoch": 38.844686648501366,
+ "grad_norm": 4.050828456878662,
+ "learning_rate": 1.3984943486666957e-05,
+ "loss": 0.1268,
+ "step": 14256
+ },
+ {
+ "epoch": 38.84741144414169,
+ "grad_norm": 3.3075990676879883,
+ "learning_rate": 1.3984134072850417e-05,
+ "loss": 0.1633,
+ "step": 14257
+ },
+ {
+ "epoch": 38.85013623978202,
+ "grad_norm": 3.0918169021606445,
+ "learning_rate": 1.3983324628005566e-05,
+ "loss": 0.2224,
+ "step": 14258
+ },
+ {
+ "epoch": 38.85286103542234,
+ "grad_norm": 5.827859401702881,
+ "learning_rate": 1.3982515152138713e-05,
+ "loss": 0.1153,
+ "step": 14259
+ },
+ {
+ "epoch": 38.85558583106267,
+ "grad_norm": 3.2805252075195312,
+ "learning_rate": 1.3981705645256161e-05,
+ "loss": 0.1612,
+ "step": 14260
+ },
+ {
+ "epoch": 38.858310626702995,
+ "grad_norm": 3.268380880355835,
+ "learning_rate": 1.3980896107364212e-05,
+ "loss": 0.2473,
+ "step": 14261
+ },
+ {
+ "epoch": 38.86103542234333,
+ "grad_norm": 2.99371337890625,
+ "learning_rate": 1.398008653846917e-05,
+ "loss": 0.1911,
+ "step": 14262
+ },
+ {
+ "epoch": 38.86376021798365,
+ "grad_norm": 3.7174532413482666,
+ "learning_rate": 1.3979276938577347e-05,
+ "loss": 0.1442,
+ "step": 14263
+ },
+ {
+ "epoch": 38.86648501362398,
+ "grad_norm": 4.204202651977539,
+ "learning_rate": 1.397846730769504e-05,
+ "loss": 0.1616,
+ "step": 14264
+ },
+ {
+ "epoch": 38.869209809264305,
+ "grad_norm": 3.6148204803466797,
+ "learning_rate": 1.3977657645828559e-05,
+ "loss": 0.2167,
+ "step": 14265
+ },
+ {
+ "epoch": 38.87193460490463,
+ "grad_norm": 3.7188620567321777,
+ "learning_rate": 1.3976847952984209e-05,
+ "loss": 0.1033,
+ "step": 14266
+ },
+ {
+ "epoch": 38.87465940054496,
+ "grad_norm": 3.364196300506592,
+ "learning_rate": 1.3976038229168296e-05,
+ "loss": 0.1243,
+ "step": 14267
+ },
+ {
+ "epoch": 38.87738419618529,
+ "grad_norm": 3.57443904876709,
+ "learning_rate": 1.3975228474387122e-05,
+ "loss": 0.1162,
+ "step": 14268
+ },
+ {
+ "epoch": 38.880108991825615,
+ "grad_norm": 3.1089749336242676,
+ "learning_rate": 1.3974418688646997e-05,
+ "loss": 0.093,
+ "step": 14269
+ },
+ {
+ "epoch": 38.88283378746594,
+ "grad_norm": 3.724578619003296,
+ "learning_rate": 1.397360887195423e-05,
+ "loss": 0.103,
+ "step": 14270
+ },
+ {
+ "epoch": 38.88555858310627,
+ "grad_norm": 3.5159218311309814,
+ "learning_rate": 1.3972799024315123e-05,
+ "loss": 0.1961,
+ "step": 14271
+ },
+ {
+ "epoch": 38.88828337874659,
+ "grad_norm": 4.453646659851074,
+ "learning_rate": 1.3971989145735982e-05,
+ "loss": 0.2042,
+ "step": 14272
+ },
+ {
+ "epoch": 38.89100817438692,
+ "grad_norm": 4.219421863555908,
+ "learning_rate": 1.3971179236223122e-05,
+ "loss": 0.1488,
+ "step": 14273
+ },
+ {
+ "epoch": 38.89373297002725,
+ "grad_norm": 2.929407835006714,
+ "learning_rate": 1.3970369295782844e-05,
+ "loss": 0.1262,
+ "step": 14274
+ },
+ {
+ "epoch": 38.89645776566758,
+ "grad_norm": 3.4698379039764404,
+ "learning_rate": 1.396955932442146e-05,
+ "loss": 0.1277,
+ "step": 14275
+ },
+ {
+ "epoch": 38.8991825613079,
+ "grad_norm": 4.034002304077148,
+ "learning_rate": 1.3968749322145274e-05,
+ "loss": 0.2169,
+ "step": 14276
+ },
+ {
+ "epoch": 38.90190735694823,
+ "grad_norm": 2.952831745147705,
+ "learning_rate": 1.3967939288960595e-05,
+ "loss": 0.1738,
+ "step": 14277
+ },
+ {
+ "epoch": 38.904632152588555,
+ "grad_norm": 3.1070098876953125,
+ "learning_rate": 1.3967129224873734e-05,
+ "loss": 0.0786,
+ "step": 14278
+ },
+ {
+ "epoch": 38.90735694822888,
+ "grad_norm": 3.654160499572754,
+ "learning_rate": 1.3966319129890997e-05,
+ "loss": 0.0867,
+ "step": 14279
+ },
+ {
+ "epoch": 38.91008174386921,
+ "grad_norm": 3.6056809425354004,
+ "learning_rate": 1.3965509004018695e-05,
+ "loss": 0.1067,
+ "step": 14280
+ },
+ {
+ "epoch": 38.91280653950954,
+ "grad_norm": 3.532139778137207,
+ "learning_rate": 1.3964698847263138e-05,
+ "loss": 0.2212,
+ "step": 14281
+ },
+ {
+ "epoch": 38.915531335149865,
+ "grad_norm": 3.2069742679595947,
+ "learning_rate": 1.3963888659630634e-05,
+ "loss": 0.325,
+ "step": 14282
+ },
+ {
+ "epoch": 38.91825613079019,
+ "grad_norm": 4.047801494598389,
+ "learning_rate": 1.396307844112749e-05,
+ "loss": 0.1576,
+ "step": 14283
+ },
+ {
+ "epoch": 38.920980926430516,
+ "grad_norm": 4.179137706756592,
+ "learning_rate": 1.3962268191760019e-05,
+ "loss": 0.1776,
+ "step": 14284
+ },
+ {
+ "epoch": 38.92370572207084,
+ "grad_norm": 2.6677331924438477,
+ "learning_rate": 1.3961457911534532e-05,
+ "loss": 0.1459,
+ "step": 14285
+ },
+ {
+ "epoch": 38.926430517711175,
+ "grad_norm": 4.285943984985352,
+ "learning_rate": 1.3960647600457338e-05,
+ "loss": 0.1362,
+ "step": 14286
+ },
+ {
+ "epoch": 38.9291553133515,
+ "grad_norm": 9.723367691040039,
+ "learning_rate": 1.3959837258534748e-05,
+ "loss": 0.1489,
+ "step": 14287
+ },
+ {
+ "epoch": 38.93188010899183,
+ "grad_norm": 4.159480571746826,
+ "learning_rate": 1.3959026885773072e-05,
+ "loss": 0.1893,
+ "step": 14288
+ },
+ {
+ "epoch": 38.93460490463215,
+ "grad_norm": 3.212905168533325,
+ "learning_rate": 1.3958216482178623e-05,
+ "loss": 0.1156,
+ "step": 14289
+ },
+ {
+ "epoch": 38.93732970027248,
+ "grad_norm": 3.1981348991394043,
+ "learning_rate": 1.3957406047757714e-05,
+ "loss": 0.1766,
+ "step": 14290
+ },
+ {
+ "epoch": 38.940054495912804,
+ "grad_norm": 3.4070959091186523,
+ "learning_rate": 1.395659558251665e-05,
+ "loss": 0.1568,
+ "step": 14291
+ },
+ {
+ "epoch": 38.94277929155314,
+ "grad_norm": 2.981278657913208,
+ "learning_rate": 1.3955785086461749e-05,
+ "loss": 0.2,
+ "step": 14292
+ },
+ {
+ "epoch": 38.94550408719346,
+ "grad_norm": 3.3687124252319336,
+ "learning_rate": 1.395497455959932e-05,
+ "loss": 0.236,
+ "step": 14293
+ },
+ {
+ "epoch": 38.94822888283379,
+ "grad_norm": 6.6699113845825195,
+ "learning_rate": 1.3954164001935674e-05,
+ "loss": 0.3012,
+ "step": 14294
+ },
+ {
+ "epoch": 38.950953678474114,
+ "grad_norm": 3.454345464706421,
+ "learning_rate": 1.3953353413477132e-05,
+ "loss": 0.2467,
+ "step": 14295
+ },
+ {
+ "epoch": 38.95367847411444,
+ "grad_norm": 3.8459818363189697,
+ "learning_rate": 1.3952542794229998e-05,
+ "loss": 0.1634,
+ "step": 14296
+ },
+ {
+ "epoch": 38.956403269754766,
+ "grad_norm": 2.6928017139434814,
+ "learning_rate": 1.3951732144200584e-05,
+ "loss": 0.0871,
+ "step": 14297
+ },
+ {
+ "epoch": 38.95912806539509,
+ "grad_norm": 3.879988193511963,
+ "learning_rate": 1.3950921463395213e-05,
+ "loss": 0.0852,
+ "step": 14298
+ },
+ {
+ "epoch": 38.961852861035425,
+ "grad_norm": 3.9086458683013916,
+ "learning_rate": 1.3950110751820192e-05,
+ "loss": 0.1166,
+ "step": 14299
+ },
+ {
+ "epoch": 38.96457765667575,
+ "grad_norm": 3.403714418411255,
+ "learning_rate": 1.3949300009481835e-05,
+ "loss": 0.1229,
+ "step": 14300
+ },
+ {
+ "epoch": 38.967302452316076,
+ "grad_norm": 3.0470659732818604,
+ "learning_rate": 1.3948489236386454e-05,
+ "loss": 0.0755,
+ "step": 14301
+ },
+ {
+ "epoch": 38.9700272479564,
+ "grad_norm": 3.5446126461029053,
+ "learning_rate": 1.3947678432540368e-05,
+ "loss": 0.116,
+ "step": 14302
+ },
+ {
+ "epoch": 38.97275204359673,
+ "grad_norm": 3.444605588912964,
+ "learning_rate": 1.3946867597949887e-05,
+ "loss": 0.1523,
+ "step": 14303
+ },
+ {
+ "epoch": 38.97547683923706,
+ "grad_norm": 3.6973977088928223,
+ "learning_rate": 1.3946056732621332e-05,
+ "loss": 0.1572,
+ "step": 14304
+ },
+ {
+ "epoch": 38.97820163487739,
+ "grad_norm": 3.68373441696167,
+ "learning_rate": 1.3945245836561013e-05,
+ "loss": 0.4707,
+ "step": 14305
+ },
+ {
+ "epoch": 38.98092643051771,
+ "grad_norm": 3.770282030105591,
+ "learning_rate": 1.3944434909775247e-05,
+ "loss": 0.2455,
+ "step": 14306
+ },
+ {
+ "epoch": 38.98365122615804,
+ "grad_norm": 3.754664897918701,
+ "learning_rate": 1.3943623952270346e-05,
+ "loss": 0.1459,
+ "step": 14307
+ },
+ {
+ "epoch": 38.986376021798364,
+ "grad_norm": 3.504155158996582,
+ "learning_rate": 1.3942812964052632e-05,
+ "loss": 0.1488,
+ "step": 14308
+ },
+ {
+ "epoch": 38.98910081743869,
+ "grad_norm": 2.774792194366455,
+ "learning_rate": 1.3942001945128416e-05,
+ "loss": 0.0905,
+ "step": 14309
+ },
+ {
+ "epoch": 38.991825613079016,
+ "grad_norm": 2.839503765106201,
+ "learning_rate": 1.3941190895504015e-05,
+ "loss": 0.1672,
+ "step": 14310
+ },
+ {
+ "epoch": 38.99455040871935,
+ "grad_norm": 3.6763675212860107,
+ "learning_rate": 1.3940379815185746e-05,
+ "loss": 0.2144,
+ "step": 14311
+ },
+ {
+ "epoch": 38.997275204359674,
+ "grad_norm": 4.084926605224609,
+ "learning_rate": 1.3939568704179926e-05,
+ "loss": 0.2863,
+ "step": 14312
+ },
+ {
+ "epoch": 39.0,
+ "grad_norm": 3.3414433002471924,
+ "learning_rate": 1.3938757562492873e-05,
+ "loss": 0.0871,
+ "step": 14313
+ },
+ {
+ "epoch": 39.002724795640326,
+ "grad_norm": 3.280057430267334,
+ "learning_rate": 1.3937946390130904e-05,
+ "loss": 0.1244,
+ "step": 14314
+ },
+ {
+ "epoch": 39.00544959128065,
+ "grad_norm": 3.409501314163208,
+ "learning_rate": 1.3937135187100332e-05,
+ "loss": 0.0764,
+ "step": 14315
+ },
+ {
+ "epoch": 39.00817438692098,
+ "grad_norm": 3.1620845794677734,
+ "learning_rate": 1.393632395340748e-05,
+ "loss": 0.1796,
+ "step": 14316
+ },
+ {
+ "epoch": 39.01089918256131,
+ "grad_norm": 3.550021171569824,
+ "learning_rate": 1.3935512689058663e-05,
+ "loss": 0.1566,
+ "step": 14317
+ },
+ {
+ "epoch": 39.013623978201636,
+ "grad_norm": 3.7412750720977783,
+ "learning_rate": 1.3934701394060202e-05,
+ "loss": 0.1058,
+ "step": 14318
+ },
+ {
+ "epoch": 39.01634877384196,
+ "grad_norm": 3.2823586463928223,
+ "learning_rate": 1.393389006841841e-05,
+ "loss": 0.2687,
+ "step": 14319
+ },
+ {
+ "epoch": 39.01907356948229,
+ "grad_norm": 3.733161211013794,
+ "learning_rate": 1.3933078712139607e-05,
+ "loss": 0.0976,
+ "step": 14320
+ },
+ {
+ "epoch": 39.02179836512261,
+ "grad_norm": 3.361510753631592,
+ "learning_rate": 1.3932267325230118e-05,
+ "loss": 0.1352,
+ "step": 14321
+ },
+ {
+ "epoch": 39.02452316076294,
+ "grad_norm": 3.221450090408325,
+ "learning_rate": 1.3931455907696257e-05,
+ "loss": 0.0964,
+ "step": 14322
+ },
+ {
+ "epoch": 39.02724795640327,
+ "grad_norm": 3.6514344215393066,
+ "learning_rate": 1.3930644459544343e-05,
+ "loss": 0.218,
+ "step": 14323
+ },
+ {
+ "epoch": 39.0299727520436,
+ "grad_norm": 3.9863228797912598,
+ "learning_rate": 1.3929832980780696e-05,
+ "loss": 0.1826,
+ "step": 14324
+ },
+ {
+ "epoch": 39.032697547683924,
+ "grad_norm": 2.7949929237365723,
+ "learning_rate": 1.3929021471411637e-05,
+ "loss": 0.087,
+ "step": 14325
+ },
+ {
+ "epoch": 39.03542234332425,
+ "grad_norm": 3.853773832321167,
+ "learning_rate": 1.3928209931443487e-05,
+ "loss": 0.1124,
+ "step": 14326
+ },
+ {
+ "epoch": 39.038147138964575,
+ "grad_norm": 3.1689414978027344,
+ "learning_rate": 1.392739836088256e-05,
+ "loss": 0.181,
+ "step": 14327
+ },
+ {
+ "epoch": 39.0408719346049,
+ "grad_norm": 2.5113821029663086,
+ "learning_rate": 1.3926586759735185e-05,
+ "loss": 0.079,
+ "step": 14328
+ },
+ {
+ "epoch": 39.043596730245234,
+ "grad_norm": 2.9339067935943604,
+ "learning_rate": 1.3925775128007674e-05,
+ "loss": 0.1638,
+ "step": 14329
+ },
+ {
+ "epoch": 39.04632152588556,
+ "grad_norm": 2.6700806617736816,
+ "learning_rate": 1.3924963465706358e-05,
+ "loss": 0.2111,
+ "step": 14330
+ },
+ {
+ "epoch": 39.049046321525886,
+ "grad_norm": 2.6672699451446533,
+ "learning_rate": 1.3924151772837549e-05,
+ "loss": 0.1045,
+ "step": 14331
+ },
+ {
+ "epoch": 39.05177111716621,
+ "grad_norm": 3.227785587310791,
+ "learning_rate": 1.3923340049407576e-05,
+ "loss": 0.2175,
+ "step": 14332
+ },
+ {
+ "epoch": 39.05449591280654,
+ "grad_norm": 3.4597785472869873,
+ "learning_rate": 1.3922528295422755e-05,
+ "loss": 0.2453,
+ "step": 14333
+ },
+ {
+ "epoch": 39.05722070844686,
+ "grad_norm": 3.4480443000793457,
+ "learning_rate": 1.392171651088941e-05,
+ "loss": 0.1172,
+ "step": 14334
+ },
+ {
+ "epoch": 39.059945504087196,
+ "grad_norm": 3.541322708129883,
+ "learning_rate": 1.3920904695813861e-05,
+ "loss": 0.1002,
+ "step": 14335
+ },
+ {
+ "epoch": 39.06267029972752,
+ "grad_norm": 3.4819796085357666,
+ "learning_rate": 1.3920092850202435e-05,
+ "loss": 0.1358,
+ "step": 14336
+ },
+ {
+ "epoch": 39.06539509536785,
+ "grad_norm": 3.2888596057891846,
+ "learning_rate": 1.391928097406145e-05,
+ "loss": 0.1962,
+ "step": 14337
+ },
+ {
+ "epoch": 39.06811989100817,
+ "grad_norm": 3.396695137023926,
+ "learning_rate": 1.3918469067397233e-05,
+ "loss": 0.1076,
+ "step": 14338
+ },
+ {
+ "epoch": 39.0708446866485,
+ "grad_norm": 3.898594379425049,
+ "learning_rate": 1.3917657130216103e-05,
+ "loss": 0.127,
+ "step": 14339
+ },
+ {
+ "epoch": 39.073569482288825,
+ "grad_norm": 3.0390052795410156,
+ "learning_rate": 1.3916845162524388e-05,
+ "loss": 0.0973,
+ "step": 14340
+ },
+ {
+ "epoch": 39.07629427792916,
+ "grad_norm": 2.5301637649536133,
+ "learning_rate": 1.3916033164328406e-05,
+ "loss": 0.0648,
+ "step": 14341
+ },
+ {
+ "epoch": 39.079019073569484,
+ "grad_norm": 3.2571301460266113,
+ "learning_rate": 1.3915221135634487e-05,
+ "loss": 0.0995,
+ "step": 14342
+ },
+ {
+ "epoch": 39.08174386920981,
+ "grad_norm": 2.9791080951690674,
+ "learning_rate": 1.391440907644895e-05,
+ "loss": 0.2421,
+ "step": 14343
+ },
+ {
+ "epoch": 39.084468664850135,
+ "grad_norm": 3.5741279125213623,
+ "learning_rate": 1.3913596986778123e-05,
+ "loss": 0.1403,
+ "step": 14344
+ },
+ {
+ "epoch": 39.08719346049046,
+ "grad_norm": 3.027524709701538,
+ "learning_rate": 1.3912784866628326e-05,
+ "loss": 0.2081,
+ "step": 14345
+ },
+ {
+ "epoch": 39.08991825613079,
+ "grad_norm": 2.3582096099853516,
+ "learning_rate": 1.3911972716005887e-05,
+ "loss": 0.0731,
+ "step": 14346
+ },
+ {
+ "epoch": 39.09264305177112,
+ "grad_norm": 2.7368931770324707,
+ "learning_rate": 1.391116053491713e-05,
+ "loss": 0.1112,
+ "step": 14347
+ },
+ {
+ "epoch": 39.095367847411445,
+ "grad_norm": 3.7673096656799316,
+ "learning_rate": 1.3910348323368381e-05,
+ "loss": 0.0887,
+ "step": 14348
+ },
+ {
+ "epoch": 39.09809264305177,
+ "grad_norm": 2.8192830085754395,
+ "learning_rate": 1.3909536081365964e-05,
+ "loss": 0.1019,
+ "step": 14349
+ },
+ {
+ "epoch": 39.1008174386921,
+ "grad_norm": 3.0649523735046387,
+ "learning_rate": 1.390872380891621e-05,
+ "loss": 0.203,
+ "step": 14350
+ },
+ {
+ "epoch": 39.10354223433242,
+ "grad_norm": 2.7920703887939453,
+ "learning_rate": 1.3907911506025436e-05,
+ "loss": 0.1313,
+ "step": 14351
+ },
+ {
+ "epoch": 39.10626702997275,
+ "grad_norm": 3.869267225265503,
+ "learning_rate": 1.3907099172699976e-05,
+ "loss": 0.1314,
+ "step": 14352
+ },
+ {
+ "epoch": 39.10899182561308,
+ "grad_norm": 3.6431753635406494,
+ "learning_rate": 1.390628680894615e-05,
+ "loss": 0.1168,
+ "step": 14353
+ },
+ {
+ "epoch": 39.11171662125341,
+ "grad_norm": 3.4703104496002197,
+ "learning_rate": 1.390547441477029e-05,
+ "loss": 0.0835,
+ "step": 14354
+ },
+ {
+ "epoch": 39.11444141689373,
+ "grad_norm": 3.2260937690734863,
+ "learning_rate": 1.3904661990178723e-05,
+ "loss": 0.0924,
+ "step": 14355
+ },
+ {
+ "epoch": 39.11716621253406,
+ "grad_norm": 3.2402100563049316,
+ "learning_rate": 1.390384953517777e-05,
+ "loss": 0.0855,
+ "step": 14356
+ },
+ {
+ "epoch": 39.119891008174385,
+ "grad_norm": 3.4586997032165527,
+ "learning_rate": 1.3903037049773769e-05,
+ "loss": 0.2326,
+ "step": 14357
+ },
+ {
+ "epoch": 39.12261580381471,
+ "grad_norm": 3.06036639213562,
+ "learning_rate": 1.3902224533973034e-05,
+ "loss": 0.2609,
+ "step": 14358
+ },
+ {
+ "epoch": 39.12534059945504,
+ "grad_norm": 3.0915913581848145,
+ "learning_rate": 1.3901411987781904e-05,
+ "loss": 0.2691,
+ "step": 14359
+ },
+ {
+ "epoch": 39.12806539509537,
+ "grad_norm": 3.3583381175994873,
+ "learning_rate": 1.3900599411206704e-05,
+ "loss": 0.1284,
+ "step": 14360
+ },
+ {
+ "epoch": 39.130790190735695,
+ "grad_norm": 2.9110939502716064,
+ "learning_rate": 1.3899786804253757e-05,
+ "loss": 0.1381,
+ "step": 14361
+ },
+ {
+ "epoch": 39.13351498637602,
+ "grad_norm": 2.611429452896118,
+ "learning_rate": 1.3898974166929398e-05,
+ "loss": 0.246,
+ "step": 14362
+ },
+ {
+ "epoch": 39.13623978201635,
+ "grad_norm": 3.175748109817505,
+ "learning_rate": 1.3898161499239952e-05,
+ "loss": 0.1018,
+ "step": 14363
+ },
+ {
+ "epoch": 39.13896457765667,
+ "grad_norm": 3.0770649909973145,
+ "learning_rate": 1.3897348801191752e-05,
+ "loss": 0.2137,
+ "step": 14364
+ },
+ {
+ "epoch": 39.141689373297005,
+ "grad_norm": 3.6234662532806396,
+ "learning_rate": 1.3896536072791125e-05,
+ "loss": 0.1447,
+ "step": 14365
+ },
+ {
+ "epoch": 39.14441416893733,
+ "grad_norm": 3.2570273876190186,
+ "learning_rate": 1.38957233140444e-05,
+ "loss": 0.147,
+ "step": 14366
+ },
+ {
+ "epoch": 39.14713896457766,
+ "grad_norm": 3.6546313762664795,
+ "learning_rate": 1.3894910524957907e-05,
+ "loss": 0.1875,
+ "step": 14367
+ },
+ {
+ "epoch": 39.14986376021798,
+ "grad_norm": 2.7993295192718506,
+ "learning_rate": 1.3894097705537977e-05,
+ "loss": 0.0832,
+ "step": 14368
+ },
+ {
+ "epoch": 39.15258855585831,
+ "grad_norm": 2.88297700881958,
+ "learning_rate": 1.389328485579094e-05,
+ "loss": 0.2411,
+ "step": 14369
+ },
+ {
+ "epoch": 39.155313351498634,
+ "grad_norm": 3.4852476119995117,
+ "learning_rate": 1.3892471975723122e-05,
+ "loss": 0.1977,
+ "step": 14370
+ },
+ {
+ "epoch": 39.15803814713897,
+ "grad_norm": 3.5161662101745605,
+ "learning_rate": 1.3891659065340861e-05,
+ "loss": 0.1253,
+ "step": 14371
+ },
+ {
+ "epoch": 39.16076294277929,
+ "grad_norm": 4.418061256408691,
+ "learning_rate": 1.3890846124650482e-05,
+ "loss": 0.1671,
+ "step": 14372
+ },
+ {
+ "epoch": 39.16348773841962,
+ "grad_norm": 2.8839662075042725,
+ "learning_rate": 1.3890033153658322e-05,
+ "loss": 0.1383,
+ "step": 14373
+ },
+ {
+ "epoch": 39.166212534059945,
+ "grad_norm": 3.606159210205078,
+ "learning_rate": 1.3889220152370706e-05,
+ "loss": 0.0978,
+ "step": 14374
+ },
+ {
+ "epoch": 39.16893732970027,
+ "grad_norm": 3.531829833984375,
+ "learning_rate": 1.3888407120793973e-05,
+ "loss": 0.1138,
+ "step": 14375
+ },
+ {
+ "epoch": 39.171662125340596,
+ "grad_norm": 2.843752145767212,
+ "learning_rate": 1.3887594058934444e-05,
+ "loss": 0.1968,
+ "step": 14376
+ },
+ {
+ "epoch": 39.17438692098093,
+ "grad_norm": 2.9425771236419678,
+ "learning_rate": 1.3886780966798464e-05,
+ "loss": 0.1015,
+ "step": 14377
+ },
+ {
+ "epoch": 39.177111716621255,
+ "grad_norm": 3.0286924839019775,
+ "learning_rate": 1.3885967844392354e-05,
+ "loss": 0.1225,
+ "step": 14378
+ },
+ {
+ "epoch": 39.17983651226158,
+ "grad_norm": 3.4592978954315186,
+ "learning_rate": 1.3885154691722453e-05,
+ "loss": 0.1919,
+ "step": 14379
+ },
+ {
+ "epoch": 39.182561307901906,
+ "grad_norm": 3.838184356689453,
+ "learning_rate": 1.3884341508795092e-05,
+ "loss": 0.1462,
+ "step": 14380
+ },
+ {
+ "epoch": 39.18528610354223,
+ "grad_norm": 3.0588200092315674,
+ "learning_rate": 1.3883528295616606e-05,
+ "loss": 0.0916,
+ "step": 14381
+ },
+ {
+ "epoch": 39.18801089918256,
+ "grad_norm": 2.6581404209136963,
+ "learning_rate": 1.3882715052193323e-05,
+ "loss": 0.1851,
+ "step": 14382
+ },
+ {
+ "epoch": 39.19073569482289,
+ "grad_norm": 3.5518743991851807,
+ "learning_rate": 1.3881901778531583e-05,
+ "loss": 0.4609,
+ "step": 14383
+ },
+ {
+ "epoch": 39.19346049046322,
+ "grad_norm": 3.657963752746582,
+ "learning_rate": 1.3881088474637716e-05,
+ "loss": 0.1084,
+ "step": 14384
+ },
+ {
+ "epoch": 39.19618528610354,
+ "grad_norm": 2.8845670223236084,
+ "learning_rate": 1.388027514051806e-05,
+ "loss": 0.1058,
+ "step": 14385
+ },
+ {
+ "epoch": 39.19891008174387,
+ "grad_norm": 2.7217140197753906,
+ "learning_rate": 1.3879461776178943e-05,
+ "loss": 0.2088,
+ "step": 14386
+ },
+ {
+ "epoch": 39.201634877384194,
+ "grad_norm": 4.357101917266846,
+ "learning_rate": 1.3878648381626704e-05,
+ "loss": 0.1358,
+ "step": 14387
+ },
+ {
+ "epoch": 39.20435967302452,
+ "grad_norm": 2.8624587059020996,
+ "learning_rate": 1.3877834956867675e-05,
+ "loss": 0.1127,
+ "step": 14388
+ },
+ {
+ "epoch": 39.20708446866485,
+ "grad_norm": 6.234848499298096,
+ "learning_rate": 1.387702150190819e-05,
+ "loss": 0.098,
+ "step": 14389
+ },
+ {
+ "epoch": 39.20980926430518,
+ "grad_norm": 4.179355144500732,
+ "learning_rate": 1.3876208016754589e-05,
+ "loss": 0.0929,
+ "step": 14390
+ },
+ {
+ "epoch": 39.212534059945504,
+ "grad_norm": 3.782618999481201,
+ "learning_rate": 1.3875394501413206e-05,
+ "loss": 0.2134,
+ "step": 14391
+ },
+ {
+ "epoch": 39.21525885558583,
+ "grad_norm": 3.6591691970825195,
+ "learning_rate": 1.3874580955890374e-05,
+ "loss": 0.1039,
+ "step": 14392
+ },
+ {
+ "epoch": 39.217983651226156,
+ "grad_norm": 3.973867416381836,
+ "learning_rate": 1.3873767380192431e-05,
+ "loss": 0.0698,
+ "step": 14393
+ },
+ {
+ "epoch": 39.22070844686648,
+ "grad_norm": 3.3634088039398193,
+ "learning_rate": 1.387295377432571e-05,
+ "loss": 0.1183,
+ "step": 14394
+ },
+ {
+ "epoch": 39.223433242506815,
+ "grad_norm": 3.8122544288635254,
+ "learning_rate": 1.3872140138296553e-05,
+ "loss": 0.1215,
+ "step": 14395
+ },
+ {
+ "epoch": 39.22615803814714,
+ "grad_norm": 4.449495315551758,
+ "learning_rate": 1.387132647211129e-05,
+ "loss": 0.1402,
+ "step": 14396
+ },
+ {
+ "epoch": 39.228882833787466,
+ "grad_norm": 4.282369136810303,
+ "learning_rate": 1.3870512775776262e-05,
+ "loss": 0.1404,
+ "step": 14397
+ },
+ {
+ "epoch": 39.23160762942779,
+ "grad_norm": 3.1761720180511475,
+ "learning_rate": 1.3869699049297805e-05,
+ "loss": 0.0874,
+ "step": 14398
+ },
+ {
+ "epoch": 39.23433242506812,
+ "grad_norm": 3.323735237121582,
+ "learning_rate": 1.3868885292682258e-05,
+ "loss": 0.1464,
+ "step": 14399
+ },
+ {
+ "epoch": 39.237057220708444,
+ "grad_norm": 3.2029881477355957,
+ "learning_rate": 1.3868071505935953e-05,
+ "loss": 0.3029,
+ "step": 14400
+ },
+ {
+ "epoch": 39.23978201634878,
+ "grad_norm": 2.5351953506469727,
+ "learning_rate": 1.3867257689065237e-05,
+ "loss": 0.0873,
+ "step": 14401
+ },
+ {
+ "epoch": 39.2425068119891,
+ "grad_norm": 2.8364455699920654,
+ "learning_rate": 1.3866443842076438e-05,
+ "loss": 0.1137,
+ "step": 14402
+ },
+ {
+ "epoch": 39.24523160762943,
+ "grad_norm": 2.5234389305114746,
+ "learning_rate": 1.3865629964975901e-05,
+ "loss": 0.0727,
+ "step": 14403
+ },
+ {
+ "epoch": 39.247956403269754,
+ "grad_norm": 3.0652809143066406,
+ "learning_rate": 1.3864816057769961e-05,
+ "loss": 0.1981,
+ "step": 14404
+ },
+ {
+ "epoch": 39.25068119891008,
+ "grad_norm": 3.0669286251068115,
+ "learning_rate": 1.3864002120464962e-05,
+ "loss": 0.0688,
+ "step": 14405
+ },
+ {
+ "epoch": 39.253405994550405,
+ "grad_norm": 3.446439504623413,
+ "learning_rate": 1.3863188153067233e-05,
+ "loss": 0.1915,
+ "step": 14406
+ },
+ {
+ "epoch": 39.25613079019074,
+ "grad_norm": 3.1357011795043945,
+ "learning_rate": 1.3862374155583122e-05,
+ "loss": 0.1855,
+ "step": 14407
+ },
+ {
+ "epoch": 39.258855585831064,
+ "grad_norm": 3.2021942138671875,
+ "learning_rate": 1.3861560128018962e-05,
+ "loss": 0.1333,
+ "step": 14408
+ },
+ {
+ "epoch": 39.26158038147139,
+ "grad_norm": 2.312235116958618,
+ "learning_rate": 1.3860746070381101e-05,
+ "loss": 0.1817,
+ "step": 14409
+ },
+ {
+ "epoch": 39.264305177111716,
+ "grad_norm": 2.748317003250122,
+ "learning_rate": 1.3859931982675872e-05,
+ "loss": 0.0834,
+ "step": 14410
+ },
+ {
+ "epoch": 39.26702997275204,
+ "grad_norm": 2.8883371353149414,
+ "learning_rate": 1.3859117864909619e-05,
+ "loss": 0.0666,
+ "step": 14411
+ },
+ {
+ "epoch": 39.26975476839237,
+ "grad_norm": 2.5730323791503906,
+ "learning_rate": 1.3858303717088676e-05,
+ "loss": 0.0754,
+ "step": 14412
+ },
+ {
+ "epoch": 39.2724795640327,
+ "grad_norm": 2.9654018878936768,
+ "learning_rate": 1.3857489539219393e-05,
+ "loss": 0.1185,
+ "step": 14413
+ },
+ {
+ "epoch": 39.275204359673026,
+ "grad_norm": 2.8967690467834473,
+ "learning_rate": 1.38566753313081e-05,
+ "loss": 0.0977,
+ "step": 14414
+ },
+ {
+ "epoch": 39.27792915531335,
+ "grad_norm": 3.6722280979156494,
+ "learning_rate": 1.3855861093361144e-05,
+ "loss": 0.1406,
+ "step": 14415
+ },
+ {
+ "epoch": 39.28065395095368,
+ "grad_norm": 2.7885231971740723,
+ "learning_rate": 1.3855046825384868e-05,
+ "loss": 0.1562,
+ "step": 14416
+ },
+ {
+ "epoch": 39.283378746594,
+ "grad_norm": 3.1760785579681396,
+ "learning_rate": 1.3854232527385614e-05,
+ "loss": 0.1369,
+ "step": 14417
+ },
+ {
+ "epoch": 39.28610354223433,
+ "grad_norm": 2.5468480587005615,
+ "learning_rate": 1.3853418199369717e-05,
+ "loss": 0.0933,
+ "step": 14418
+ },
+ {
+ "epoch": 39.28882833787466,
+ "grad_norm": 2.790041446685791,
+ "learning_rate": 1.3852603841343525e-05,
+ "loss": 0.0962,
+ "step": 14419
+ },
+ {
+ "epoch": 39.29155313351499,
+ "grad_norm": 3.7377796173095703,
+ "learning_rate": 1.385178945331338e-05,
+ "loss": 0.1391,
+ "step": 14420
+ },
+ {
+ "epoch": 39.294277929155314,
+ "grad_norm": 3.693786382675171,
+ "learning_rate": 1.385097503528562e-05,
+ "loss": 0.1232,
+ "step": 14421
+ },
+ {
+ "epoch": 39.29700272479564,
+ "grad_norm": 2.295938730239868,
+ "learning_rate": 1.385016058726659e-05,
+ "loss": 0.16,
+ "step": 14422
+ },
+ {
+ "epoch": 39.299727520435965,
+ "grad_norm": 3.106149196624756,
+ "learning_rate": 1.3849346109262634e-05,
+ "loss": 0.1072,
+ "step": 14423
+ },
+ {
+ "epoch": 39.30245231607629,
+ "grad_norm": 3.1570847034454346,
+ "learning_rate": 1.3848531601280095e-05,
+ "loss": 0.0836,
+ "step": 14424
+ },
+ {
+ "epoch": 39.305177111716624,
+ "grad_norm": 2.899125814437866,
+ "learning_rate": 1.3847717063325316e-05,
+ "loss": 0.0872,
+ "step": 14425
+ },
+ {
+ "epoch": 39.30790190735695,
+ "grad_norm": 5.229333400726318,
+ "learning_rate": 1.384690249540464e-05,
+ "loss": 0.094,
+ "step": 14426
+ },
+ {
+ "epoch": 39.310626702997276,
+ "grad_norm": 3.3088202476501465,
+ "learning_rate": 1.3846087897524412e-05,
+ "loss": 0.0808,
+ "step": 14427
+ },
+ {
+ "epoch": 39.3133514986376,
+ "grad_norm": 3.214536666870117,
+ "learning_rate": 1.3845273269690973e-05,
+ "loss": 0.1465,
+ "step": 14428
+ },
+ {
+ "epoch": 39.31607629427793,
+ "grad_norm": 2.8457820415496826,
+ "learning_rate": 1.3844458611910671e-05,
+ "loss": 0.1038,
+ "step": 14429
+ },
+ {
+ "epoch": 39.31880108991825,
+ "grad_norm": 4.242368698120117,
+ "learning_rate": 1.384364392418985e-05,
+ "loss": 0.0965,
+ "step": 14430
+ },
+ {
+ "epoch": 39.321525885558586,
+ "grad_norm": 2.879315137863159,
+ "learning_rate": 1.3842829206534854e-05,
+ "loss": 0.2582,
+ "step": 14431
+ },
+ {
+ "epoch": 39.32425068119891,
+ "grad_norm": 3.940357208251953,
+ "learning_rate": 1.3842014458952025e-05,
+ "loss": 0.0945,
+ "step": 14432
+ },
+ {
+ "epoch": 39.32697547683924,
+ "grad_norm": 2.8503658771514893,
+ "learning_rate": 1.3841199681447713e-05,
+ "loss": 0.0775,
+ "step": 14433
+ },
+ {
+ "epoch": 39.32970027247956,
+ "grad_norm": 3.7933664321899414,
+ "learning_rate": 1.3840384874028262e-05,
+ "loss": 0.1907,
+ "step": 14434
+ },
+ {
+ "epoch": 39.33242506811989,
+ "grad_norm": 3.21034574508667,
+ "learning_rate": 1.3839570036700015e-05,
+ "loss": 0.1639,
+ "step": 14435
+ },
+ {
+ "epoch": 39.335149863760215,
+ "grad_norm": 3.231860637664795,
+ "learning_rate": 1.383875516946932e-05,
+ "loss": 0.151,
+ "step": 14436
+ },
+ {
+ "epoch": 39.33787465940055,
+ "grad_norm": 3.787703037261963,
+ "learning_rate": 1.3837940272342525e-05,
+ "loss": 0.2108,
+ "step": 14437
+ },
+ {
+ "epoch": 39.34059945504087,
+ "grad_norm": 2.9036107063293457,
+ "learning_rate": 1.3837125345325975e-05,
+ "loss": 0.1327,
+ "step": 14438
+ },
+ {
+ "epoch": 39.3433242506812,
+ "grad_norm": 3.507642984390259,
+ "learning_rate": 1.3836310388426015e-05,
+ "loss": 0.2513,
+ "step": 14439
+ },
+ {
+ "epoch": 39.346049046321525,
+ "grad_norm": 3.678231716156006,
+ "learning_rate": 1.3835495401648995e-05,
+ "loss": 0.161,
+ "step": 14440
+ },
+ {
+ "epoch": 39.34877384196185,
+ "grad_norm": 4.634098052978516,
+ "learning_rate": 1.3834680385001257e-05,
+ "loss": 0.1314,
+ "step": 14441
+ },
+ {
+ "epoch": 39.35149863760218,
+ "grad_norm": 3.846036434173584,
+ "learning_rate": 1.3833865338489156e-05,
+ "loss": 0.1859,
+ "step": 14442
+ },
+ {
+ "epoch": 39.35422343324251,
+ "grad_norm": 2.682065486907959,
+ "learning_rate": 1.383305026211903e-05,
+ "loss": 0.0647,
+ "step": 14443
+ },
+ {
+ "epoch": 39.356948228882835,
+ "grad_norm": 2.957157611846924,
+ "learning_rate": 1.3832235155897237e-05,
+ "loss": 0.2177,
+ "step": 14444
+ },
+ {
+ "epoch": 39.35967302452316,
+ "grad_norm": 3.2148690223693848,
+ "learning_rate": 1.3831420019830118e-05,
+ "loss": 0.1393,
+ "step": 14445
+ },
+ {
+ "epoch": 39.36239782016349,
+ "grad_norm": 4.354025363922119,
+ "learning_rate": 1.3830604853924021e-05,
+ "loss": 0.1621,
+ "step": 14446
+ },
+ {
+ "epoch": 39.36512261580381,
+ "grad_norm": 3.12267804145813,
+ "learning_rate": 1.3829789658185299e-05,
+ "loss": 0.158,
+ "step": 14447
+ },
+ {
+ "epoch": 39.36784741144414,
+ "grad_norm": 3.310739040374756,
+ "learning_rate": 1.3828974432620297e-05,
+ "loss": 0.1689,
+ "step": 14448
+ },
+ {
+ "epoch": 39.37057220708447,
+ "grad_norm": 3.876743793487549,
+ "learning_rate": 1.3828159177235368e-05,
+ "loss": 0.151,
+ "step": 14449
+ },
+ {
+ "epoch": 39.3732970027248,
+ "grad_norm": 3.2014734745025635,
+ "learning_rate": 1.3827343892036855e-05,
+ "loss": 0.088,
+ "step": 14450
+ },
+ {
+ "epoch": 39.37602179836512,
+ "grad_norm": 2.98020339012146,
+ "learning_rate": 1.3826528577031113e-05,
+ "loss": 0.1339,
+ "step": 14451
+ },
+ {
+ "epoch": 39.37874659400545,
+ "grad_norm": 2.874891996383667,
+ "learning_rate": 1.3825713232224492e-05,
+ "loss": 0.0745,
+ "step": 14452
+ },
+ {
+ "epoch": 39.381471389645775,
+ "grad_norm": 2.972795009613037,
+ "learning_rate": 1.3824897857623336e-05,
+ "loss": 0.1931,
+ "step": 14453
+ },
+ {
+ "epoch": 39.3841961852861,
+ "grad_norm": 2.808920383453369,
+ "learning_rate": 1.3824082453234e-05,
+ "loss": 0.1257,
+ "step": 14454
+ },
+ {
+ "epoch": 39.38692098092643,
+ "grad_norm": 2.4366037845611572,
+ "learning_rate": 1.3823267019062832e-05,
+ "loss": 0.0895,
+ "step": 14455
+ },
+ {
+ "epoch": 39.38964577656676,
+ "grad_norm": 4.152764797210693,
+ "learning_rate": 1.3822451555116186e-05,
+ "loss": 0.1356,
+ "step": 14456
+ },
+ {
+ "epoch": 39.392370572207085,
+ "grad_norm": 2.8351094722747803,
+ "learning_rate": 1.3821636061400407e-05,
+ "loss": 0.1949,
+ "step": 14457
+ },
+ {
+ "epoch": 39.39509536784741,
+ "grad_norm": 3.2473089694976807,
+ "learning_rate": 1.3820820537921849e-05,
+ "loss": 0.1758,
+ "step": 14458
+ },
+ {
+ "epoch": 39.39782016348774,
+ "grad_norm": 2.5387179851531982,
+ "learning_rate": 1.3820004984686866e-05,
+ "loss": 0.194,
+ "step": 14459
+ },
+ {
+ "epoch": 39.40054495912806,
+ "grad_norm": 3.81803035736084,
+ "learning_rate": 1.3819189401701807e-05,
+ "loss": 0.1731,
+ "step": 14460
+ },
+ {
+ "epoch": 39.403269754768395,
+ "grad_norm": 3.4588241577148438,
+ "learning_rate": 1.381837378897302e-05,
+ "loss": 0.2229,
+ "step": 14461
+ },
+ {
+ "epoch": 39.40599455040872,
+ "grad_norm": 3.006185293197632,
+ "learning_rate": 1.3817558146506867e-05,
+ "loss": 0.2576,
+ "step": 14462
+ },
+ {
+ "epoch": 39.40871934604905,
+ "grad_norm": 2.3683226108551025,
+ "learning_rate": 1.381674247430969e-05,
+ "loss": 0.0872,
+ "step": 14463
+ },
+ {
+ "epoch": 39.41144414168937,
+ "grad_norm": 2.422860622406006,
+ "learning_rate": 1.3815926772387847e-05,
+ "loss": 0.0643,
+ "step": 14464
+ },
+ {
+ "epoch": 39.4141689373297,
+ "grad_norm": 3.40771222114563,
+ "learning_rate": 1.3815111040747688e-05,
+ "loss": 0.0695,
+ "step": 14465
+ },
+ {
+ "epoch": 39.416893732970024,
+ "grad_norm": 3.150268077850342,
+ "learning_rate": 1.3814295279395566e-05,
+ "loss": 0.1261,
+ "step": 14466
+ },
+ {
+ "epoch": 39.41961852861036,
+ "grad_norm": 3.094414472579956,
+ "learning_rate": 1.3813479488337838e-05,
+ "loss": 0.1491,
+ "step": 14467
+ },
+ {
+ "epoch": 39.42234332425068,
+ "grad_norm": 4.12404203414917,
+ "learning_rate": 1.3812663667580855e-05,
+ "loss": 0.2272,
+ "step": 14468
+ },
+ {
+ "epoch": 39.42506811989101,
+ "grad_norm": 3.644502639770508,
+ "learning_rate": 1.3811847817130965e-05,
+ "loss": 0.2092,
+ "step": 14469
+ },
+ {
+ "epoch": 39.427792915531334,
+ "grad_norm": 3.8725149631500244,
+ "learning_rate": 1.3811031936994533e-05,
+ "loss": 0.1647,
+ "step": 14470
+ },
+ {
+ "epoch": 39.43051771117166,
+ "grad_norm": 3.100600481033325,
+ "learning_rate": 1.3810216027177903e-05,
+ "loss": 0.1058,
+ "step": 14471
+ },
+ {
+ "epoch": 39.433242506811986,
+ "grad_norm": 4.175043106079102,
+ "learning_rate": 1.3809400087687434e-05,
+ "loss": 0.3811,
+ "step": 14472
+ },
+ {
+ "epoch": 39.43596730245232,
+ "grad_norm": 5.819088459014893,
+ "learning_rate": 1.380858411852948e-05,
+ "loss": 0.205,
+ "step": 14473
+ },
+ {
+ "epoch": 39.438692098092645,
+ "grad_norm": 3.6199569702148438,
+ "learning_rate": 1.3807768119710397e-05,
+ "loss": 0.0813,
+ "step": 14474
+ },
+ {
+ "epoch": 39.44141689373297,
+ "grad_norm": 3.700812816619873,
+ "learning_rate": 1.3806952091236533e-05,
+ "loss": 0.2698,
+ "step": 14475
+ },
+ {
+ "epoch": 39.444141689373296,
+ "grad_norm": 4.550185203552246,
+ "learning_rate": 1.3806136033114255e-05,
+ "loss": 0.2258,
+ "step": 14476
+ },
+ {
+ "epoch": 39.44686648501362,
+ "grad_norm": 2.9854736328125,
+ "learning_rate": 1.3805319945349908e-05,
+ "loss": 0.1931,
+ "step": 14477
+ },
+ {
+ "epoch": 39.44959128065395,
+ "grad_norm": 3.4972336292266846,
+ "learning_rate": 1.3804503827949855e-05,
+ "loss": 0.1392,
+ "step": 14478
+ },
+ {
+ "epoch": 39.45231607629428,
+ "grad_norm": 3.41055965423584,
+ "learning_rate": 1.3803687680920446e-05,
+ "loss": 0.1325,
+ "step": 14479
+ },
+ {
+ "epoch": 39.45504087193461,
+ "grad_norm": 3.6330044269561768,
+ "learning_rate": 1.3802871504268043e-05,
+ "loss": 0.0987,
+ "step": 14480
+ },
+ {
+ "epoch": 39.45776566757493,
+ "grad_norm": 3.367213726043701,
+ "learning_rate": 1.3802055297998996e-05,
+ "loss": 0.1221,
+ "step": 14481
+ },
+ {
+ "epoch": 39.46049046321526,
+ "grad_norm": 3.675640106201172,
+ "learning_rate": 1.3801239062119666e-05,
+ "loss": 0.2293,
+ "step": 14482
+ },
+ {
+ "epoch": 39.463215258855584,
+ "grad_norm": 3.698559045791626,
+ "learning_rate": 1.3800422796636406e-05,
+ "loss": 0.1225,
+ "step": 14483
+ },
+ {
+ "epoch": 39.46594005449591,
+ "grad_norm": 3.0148515701293945,
+ "learning_rate": 1.3799606501555578e-05,
+ "loss": 0.0917,
+ "step": 14484
+ },
+ {
+ "epoch": 39.46866485013624,
+ "grad_norm": 4.080511569976807,
+ "learning_rate": 1.3798790176883536e-05,
+ "loss": 0.1033,
+ "step": 14485
+ },
+ {
+ "epoch": 39.47138964577657,
+ "grad_norm": 3.9746954441070557,
+ "learning_rate": 1.379797382262664e-05,
+ "loss": 0.2942,
+ "step": 14486
+ },
+ {
+ "epoch": 39.474114441416894,
+ "grad_norm": 3.6068904399871826,
+ "learning_rate": 1.3797157438791244e-05,
+ "loss": 0.1752,
+ "step": 14487
+ },
+ {
+ "epoch": 39.47683923705722,
+ "grad_norm": 3.3865113258361816,
+ "learning_rate": 1.379634102538371e-05,
+ "loss": 0.1252,
+ "step": 14488
+ },
+ {
+ "epoch": 39.479564032697546,
+ "grad_norm": 3.0101404190063477,
+ "learning_rate": 1.3795524582410394e-05,
+ "loss": 0.0967,
+ "step": 14489
+ },
+ {
+ "epoch": 39.48228882833787,
+ "grad_norm": 2.6312179565429688,
+ "learning_rate": 1.3794708109877655e-05,
+ "loss": 0.0852,
+ "step": 14490
+ },
+ {
+ "epoch": 39.485013623978205,
+ "grad_norm": 2.5860276222229004,
+ "learning_rate": 1.379389160779185e-05,
+ "loss": 0.0451,
+ "step": 14491
+ },
+ {
+ "epoch": 39.48773841961853,
+ "grad_norm": 4.185610771179199,
+ "learning_rate": 1.379307507615934e-05,
+ "loss": 0.1054,
+ "step": 14492
+ },
+ {
+ "epoch": 39.490463215258856,
+ "grad_norm": 4.486627101898193,
+ "learning_rate": 1.3792258514986481e-05,
+ "loss": 0.1563,
+ "step": 14493
+ },
+ {
+ "epoch": 39.49318801089918,
+ "grad_norm": 3.7792365550994873,
+ "learning_rate": 1.3791441924279638e-05,
+ "loss": 0.1206,
+ "step": 14494
+ },
+ {
+ "epoch": 39.49591280653951,
+ "grad_norm": 3.82649564743042,
+ "learning_rate": 1.3790625304045165e-05,
+ "loss": 0.1291,
+ "step": 14495
+ },
+ {
+ "epoch": 39.49863760217983,
+ "grad_norm": 3.493886709213257,
+ "learning_rate": 1.3789808654289428e-05,
+ "loss": 0.1249,
+ "step": 14496
+ },
+ {
+ "epoch": 39.50136239782017,
+ "grad_norm": 2.249948501586914,
+ "learning_rate": 1.3788991975018777e-05,
+ "loss": 0.1451,
+ "step": 14497
+ },
+ {
+ "epoch": 39.50408719346049,
+ "grad_norm": 3.8038878440856934,
+ "learning_rate": 1.3788175266239585e-05,
+ "loss": 0.1431,
+ "step": 14498
+ },
+ {
+ "epoch": 39.50681198910082,
+ "grad_norm": 3.8824501037597656,
+ "learning_rate": 1.37873585279582e-05,
+ "loss": 0.105,
+ "step": 14499
+ },
+ {
+ "epoch": 39.509536784741144,
+ "grad_norm": 2.8143088817596436,
+ "learning_rate": 1.3786541760180994e-05,
+ "loss": 0.082,
+ "step": 14500
+ },
+ {
+ "epoch": 39.51226158038147,
+ "grad_norm": 4.19122314453125,
+ "learning_rate": 1.3785724962914317e-05,
+ "loss": 0.1663,
+ "step": 14501
+ },
+ {
+ "epoch": 39.514986376021795,
+ "grad_norm": 2.926750898361206,
+ "learning_rate": 1.378490813616454e-05,
+ "loss": 0.1562,
+ "step": 14502
+ },
+ {
+ "epoch": 39.51771117166213,
+ "grad_norm": 3.034102439880371,
+ "learning_rate": 1.3784091279938019e-05,
+ "loss": 0.1195,
+ "step": 14503
+ },
+ {
+ "epoch": 39.520435967302454,
+ "grad_norm": 2.931443214416504,
+ "learning_rate": 1.3783274394241119e-05,
+ "loss": 0.2407,
+ "step": 14504
+ },
+ {
+ "epoch": 39.52316076294278,
+ "grad_norm": 3.912181854248047,
+ "learning_rate": 1.3782457479080198e-05,
+ "loss": 0.3895,
+ "step": 14505
+ },
+ {
+ "epoch": 39.525885558583106,
+ "grad_norm": 2.876732587814331,
+ "learning_rate": 1.378164053446162e-05,
+ "loss": 0.0999,
+ "step": 14506
+ },
+ {
+ "epoch": 39.52861035422343,
+ "grad_norm": 3.0108630657196045,
+ "learning_rate": 1.3780823560391748e-05,
+ "loss": 0.0565,
+ "step": 14507
+ },
+ {
+ "epoch": 39.53133514986376,
+ "grad_norm": 4.149706840515137,
+ "learning_rate": 1.378000655687694e-05,
+ "loss": 0.1976,
+ "step": 14508
+ },
+ {
+ "epoch": 39.53405994550409,
+ "grad_norm": 3.3090102672576904,
+ "learning_rate": 1.3779189523923568e-05,
+ "loss": 0.1097,
+ "step": 14509
+ },
+ {
+ "epoch": 39.536784741144416,
+ "grad_norm": 3.153085231781006,
+ "learning_rate": 1.3778372461537985e-05,
+ "loss": 0.1219,
+ "step": 14510
+ },
+ {
+ "epoch": 39.53950953678474,
+ "grad_norm": 3.9393463134765625,
+ "learning_rate": 1.3777555369726561e-05,
+ "loss": 0.1245,
+ "step": 14511
+ },
+ {
+ "epoch": 39.54223433242507,
+ "grad_norm": 3.6141135692596436,
+ "learning_rate": 1.3776738248495659e-05,
+ "loss": 0.1451,
+ "step": 14512
+ },
+ {
+ "epoch": 39.54495912806539,
+ "grad_norm": 2.8912007808685303,
+ "learning_rate": 1.3775921097851637e-05,
+ "loss": 0.1206,
+ "step": 14513
+ },
+ {
+ "epoch": 39.54768392370572,
+ "grad_norm": 3.293100595474243,
+ "learning_rate": 1.3775103917800867e-05,
+ "loss": 0.141,
+ "step": 14514
+ },
+ {
+ "epoch": 39.55040871934605,
+ "grad_norm": 2.785207509994507,
+ "learning_rate": 1.3774286708349706e-05,
+ "loss": 0.0813,
+ "step": 14515
+ },
+ {
+ "epoch": 39.55313351498638,
+ "grad_norm": 2.920821189880371,
+ "learning_rate": 1.3773469469504526e-05,
+ "loss": 0.1653,
+ "step": 14516
+ },
+ {
+ "epoch": 39.555858310626704,
+ "grad_norm": 8.18297290802002,
+ "learning_rate": 1.3772652201271684e-05,
+ "loss": 0.1449,
+ "step": 14517
+ },
+ {
+ "epoch": 39.55858310626703,
+ "grad_norm": 3.4139888286590576,
+ "learning_rate": 1.3771834903657548e-05,
+ "loss": 0.1035,
+ "step": 14518
+ },
+ {
+ "epoch": 39.561307901907355,
+ "grad_norm": 3.1998462677001953,
+ "learning_rate": 1.3771017576668485e-05,
+ "loss": 0.1727,
+ "step": 14519
+ },
+ {
+ "epoch": 39.56403269754768,
+ "grad_norm": 7.878898620605469,
+ "learning_rate": 1.3770200220310855e-05,
+ "loss": 0.094,
+ "step": 14520
+ },
+ {
+ "epoch": 39.566757493188014,
+ "grad_norm": 3.1782963275909424,
+ "learning_rate": 1.3769382834591029e-05,
+ "loss": 0.1323,
+ "step": 14521
+ },
+ {
+ "epoch": 39.56948228882834,
+ "grad_norm": 4.267875671386719,
+ "learning_rate": 1.3768565419515371e-05,
+ "loss": 0.204,
+ "step": 14522
+ },
+ {
+ "epoch": 39.572207084468666,
+ "grad_norm": 2.952256679534912,
+ "learning_rate": 1.3767747975090247e-05,
+ "loss": 0.3591,
+ "step": 14523
+ },
+ {
+ "epoch": 39.57493188010899,
+ "grad_norm": 3.4185352325439453,
+ "learning_rate": 1.3766930501322024e-05,
+ "loss": 0.107,
+ "step": 14524
+ },
+ {
+ "epoch": 39.57765667574932,
+ "grad_norm": 3.061511993408203,
+ "learning_rate": 1.3766112998217064e-05,
+ "loss": 0.0623,
+ "step": 14525
+ },
+ {
+ "epoch": 39.58038147138964,
+ "grad_norm": 2.6205015182495117,
+ "learning_rate": 1.376529546578174e-05,
+ "loss": 0.083,
+ "step": 14526
+ },
+ {
+ "epoch": 39.583106267029976,
+ "grad_norm": 3.3403451442718506,
+ "learning_rate": 1.3764477904022417e-05,
+ "loss": 0.1007,
+ "step": 14527
+ },
+ {
+ "epoch": 39.5858310626703,
+ "grad_norm": 3.2856366634368896,
+ "learning_rate": 1.3763660312945459e-05,
+ "loss": 0.1365,
+ "step": 14528
+ },
+ {
+ "epoch": 39.58855585831063,
+ "grad_norm": 8.040404319763184,
+ "learning_rate": 1.3762842692557236e-05,
+ "loss": 0.1118,
+ "step": 14529
+ },
+ {
+ "epoch": 39.59128065395095,
+ "grad_norm": 2.7413370609283447,
+ "learning_rate": 1.3762025042864116e-05,
+ "loss": 0.0806,
+ "step": 14530
+ },
+ {
+ "epoch": 39.59400544959128,
+ "grad_norm": 3.1602115631103516,
+ "learning_rate": 1.3761207363872466e-05,
+ "loss": 0.2107,
+ "step": 14531
+ },
+ {
+ "epoch": 39.596730245231605,
+ "grad_norm": 2.790731430053711,
+ "learning_rate": 1.3760389655588656e-05,
+ "loss": 0.0759,
+ "step": 14532
+ },
+ {
+ "epoch": 39.59945504087194,
+ "grad_norm": 3.5917227268218994,
+ "learning_rate": 1.3759571918019048e-05,
+ "loss": 0.145,
+ "step": 14533
+ },
+ {
+ "epoch": 39.60217983651226,
+ "grad_norm": 5.045289039611816,
+ "learning_rate": 1.375875415117002e-05,
+ "loss": 0.1023,
+ "step": 14534
+ },
+ {
+ "epoch": 39.60490463215259,
+ "grad_norm": 3.584707260131836,
+ "learning_rate": 1.375793635504793e-05,
+ "loss": 0.1157,
+ "step": 14535
+ },
+ {
+ "epoch": 39.607629427792915,
+ "grad_norm": 3.0276072025299072,
+ "learning_rate": 1.3757118529659155e-05,
+ "loss": 0.147,
+ "step": 14536
+ },
+ {
+ "epoch": 39.61035422343324,
+ "grad_norm": 2.8075852394104004,
+ "learning_rate": 1.3756300675010064e-05,
+ "loss": 0.1188,
+ "step": 14537
+ },
+ {
+ "epoch": 39.61307901907357,
+ "grad_norm": 3.068315029144287,
+ "learning_rate": 1.375548279110702e-05,
+ "loss": 0.1162,
+ "step": 14538
+ },
+ {
+ "epoch": 39.6158038147139,
+ "grad_norm": 3.394029378890991,
+ "learning_rate": 1.3754664877956401e-05,
+ "loss": 0.1195,
+ "step": 14539
+ },
+ {
+ "epoch": 39.618528610354225,
+ "grad_norm": 3.2169129848480225,
+ "learning_rate": 1.375384693556457e-05,
+ "loss": 0.1047,
+ "step": 14540
+ },
+ {
+ "epoch": 39.62125340599455,
+ "grad_norm": 3.189608573913574,
+ "learning_rate": 1.3753028963937902e-05,
+ "loss": 0.0905,
+ "step": 14541
+ },
+ {
+ "epoch": 39.62397820163488,
+ "grad_norm": 3.627978801727295,
+ "learning_rate": 1.3752210963082764e-05,
+ "loss": 0.1316,
+ "step": 14542
+ },
+ {
+ "epoch": 39.6267029972752,
+ "grad_norm": 3.1746878623962402,
+ "learning_rate": 1.375139293300553e-05,
+ "loss": 0.0889,
+ "step": 14543
+ },
+ {
+ "epoch": 39.62942779291553,
+ "grad_norm": 3.5888500213623047,
+ "learning_rate": 1.3750574873712563e-05,
+ "loss": 0.1407,
+ "step": 14544
+ },
+ {
+ "epoch": 39.63215258855586,
+ "grad_norm": 3.4999215602874756,
+ "learning_rate": 1.3749756785210245e-05,
+ "loss": 0.1391,
+ "step": 14545
+ },
+ {
+ "epoch": 39.63487738419619,
+ "grad_norm": 3.0695126056671143,
+ "learning_rate": 1.3748938667504938e-05,
+ "loss": 0.1111,
+ "step": 14546
+ },
+ {
+ "epoch": 39.63760217983651,
+ "grad_norm": 5.23260498046875,
+ "learning_rate": 1.3748120520603022e-05,
+ "loss": 0.3172,
+ "step": 14547
+ },
+ {
+ "epoch": 39.64032697547684,
+ "grad_norm": 3.124382972717285,
+ "learning_rate": 1.374730234451086e-05,
+ "loss": 0.1593,
+ "step": 14548
+ },
+ {
+ "epoch": 39.643051771117165,
+ "grad_norm": 3.0528290271759033,
+ "learning_rate": 1.374648413923483e-05,
+ "loss": 0.2113,
+ "step": 14549
+ },
+ {
+ "epoch": 39.64577656675749,
+ "grad_norm": 3.064786672592163,
+ "learning_rate": 1.37456659047813e-05,
+ "loss": 0.1069,
+ "step": 14550
+ },
+ {
+ "epoch": 39.64850136239782,
+ "grad_norm": 4.211847305297852,
+ "learning_rate": 1.3744847641156649e-05,
+ "loss": 0.0805,
+ "step": 14551
+ },
+ {
+ "epoch": 39.65122615803815,
+ "grad_norm": 3.6154606342315674,
+ "learning_rate": 1.3744029348367238e-05,
+ "loss": 0.0945,
+ "step": 14552
+ },
+ {
+ "epoch": 39.653950953678475,
+ "grad_norm": 3.882985830307007,
+ "learning_rate": 1.3743211026419452e-05,
+ "loss": 0.1277,
+ "step": 14553
+ },
+ {
+ "epoch": 39.6566757493188,
+ "grad_norm": 4.196113109588623,
+ "learning_rate": 1.3742392675319657e-05,
+ "loss": 0.2355,
+ "step": 14554
+ },
+ {
+ "epoch": 39.65940054495913,
+ "grad_norm": 3.5770938396453857,
+ "learning_rate": 1.3741574295074232e-05,
+ "loss": 0.2865,
+ "step": 14555
+ },
+ {
+ "epoch": 39.66212534059945,
+ "grad_norm": 3.404226779937744,
+ "learning_rate": 1.3740755885689541e-05,
+ "loss": 0.2353,
+ "step": 14556
+ },
+ {
+ "epoch": 39.664850136239785,
+ "grad_norm": 3.7059569358825684,
+ "learning_rate": 1.373993744717197e-05,
+ "loss": 0.3357,
+ "step": 14557
+ },
+ {
+ "epoch": 39.66757493188011,
+ "grad_norm": 3.202531576156616,
+ "learning_rate": 1.3739118979527882e-05,
+ "loss": 0.1623,
+ "step": 14558
+ },
+ {
+ "epoch": 39.67029972752044,
+ "grad_norm": 3.2610347270965576,
+ "learning_rate": 1.3738300482763655e-05,
+ "loss": 0.1217,
+ "step": 14559
+ },
+ {
+ "epoch": 39.67302452316076,
+ "grad_norm": 4.227707862854004,
+ "learning_rate": 1.3737481956885668e-05,
+ "loss": 0.1064,
+ "step": 14560
+ },
+ {
+ "epoch": 39.67574931880109,
+ "grad_norm": 3.825139284133911,
+ "learning_rate": 1.3736663401900288e-05,
+ "loss": 0.174,
+ "step": 14561
+ },
+ {
+ "epoch": 39.678474114441414,
+ "grad_norm": 3.2275919914245605,
+ "learning_rate": 1.3735844817813895e-05,
+ "loss": 0.1531,
+ "step": 14562
+ },
+ {
+ "epoch": 39.68119891008175,
+ "grad_norm": 2.78216814994812,
+ "learning_rate": 1.3735026204632864e-05,
+ "loss": 0.1038,
+ "step": 14563
+ },
+ {
+ "epoch": 39.68392370572207,
+ "grad_norm": 2.9793145656585693,
+ "learning_rate": 1.3734207562363568e-05,
+ "loss": 0.1402,
+ "step": 14564
+ },
+ {
+ "epoch": 39.6866485013624,
+ "grad_norm": 3.324450969696045,
+ "learning_rate": 1.3733388891012385e-05,
+ "loss": 0.1277,
+ "step": 14565
+ },
+ {
+ "epoch": 39.689373297002724,
+ "grad_norm": 3.247483015060425,
+ "learning_rate": 1.3732570190585689e-05,
+ "loss": 0.1044,
+ "step": 14566
+ },
+ {
+ "epoch": 39.69209809264305,
+ "grad_norm": 3.357638359069824,
+ "learning_rate": 1.3731751461089858e-05,
+ "loss": 0.0993,
+ "step": 14567
+ },
+ {
+ "epoch": 39.694822888283376,
+ "grad_norm": 4.475883483886719,
+ "learning_rate": 1.3730932702531261e-05,
+ "loss": 0.154,
+ "step": 14568
+ },
+ {
+ "epoch": 39.69754768392371,
+ "grad_norm": 3.037031888961792,
+ "learning_rate": 1.3730113914916285e-05,
+ "loss": 0.1692,
+ "step": 14569
+ },
+ {
+ "epoch": 39.700272479564035,
+ "grad_norm": 3.5867090225219727,
+ "learning_rate": 1.37292950982513e-05,
+ "loss": 0.0918,
+ "step": 14570
+ },
+ {
+ "epoch": 39.70299727520436,
+ "grad_norm": 3.238875150680542,
+ "learning_rate": 1.3728476252542687e-05,
+ "loss": 0.1501,
+ "step": 14571
+ },
+ {
+ "epoch": 39.705722070844686,
+ "grad_norm": 3.137681722640991,
+ "learning_rate": 1.3727657377796816e-05,
+ "loss": 0.1737,
+ "step": 14572
+ },
+ {
+ "epoch": 39.70844686648501,
+ "grad_norm": 3.3888509273529053,
+ "learning_rate": 1.3726838474020077e-05,
+ "loss": 0.1179,
+ "step": 14573
+ },
+ {
+ "epoch": 39.71117166212534,
+ "grad_norm": 4.459808349609375,
+ "learning_rate": 1.3726019541218833e-05,
+ "loss": 0.1965,
+ "step": 14574
+ },
+ {
+ "epoch": 39.71389645776567,
+ "grad_norm": 3.7163214683532715,
+ "learning_rate": 1.3725200579399474e-05,
+ "loss": 0.0802,
+ "step": 14575
+ },
+ {
+ "epoch": 39.716621253406,
+ "grad_norm": 3.183773994445801,
+ "learning_rate": 1.372438158856837e-05,
+ "loss": 0.1134,
+ "step": 14576
+ },
+ {
+ "epoch": 39.71934604904632,
+ "grad_norm": 3.0660958290100098,
+ "learning_rate": 1.3723562568731903e-05,
+ "loss": 0.1078,
+ "step": 14577
+ },
+ {
+ "epoch": 39.72207084468665,
+ "grad_norm": 4.291983604431152,
+ "learning_rate": 1.3722743519896447e-05,
+ "loss": 0.0954,
+ "step": 14578
+ },
+ {
+ "epoch": 39.724795640326974,
+ "grad_norm": 3.2720131874084473,
+ "learning_rate": 1.3721924442068388e-05,
+ "loss": 0.0766,
+ "step": 14579
+ },
+ {
+ "epoch": 39.7275204359673,
+ "grad_norm": 3.5988385677337646,
+ "learning_rate": 1.37211053352541e-05,
+ "loss": 0.1716,
+ "step": 14580
+ },
+ {
+ "epoch": 39.73024523160763,
+ "grad_norm": 3.2286157608032227,
+ "learning_rate": 1.3720286199459967e-05,
+ "loss": 0.1667,
+ "step": 14581
+ },
+ {
+ "epoch": 39.73297002724796,
+ "grad_norm": 4.682883262634277,
+ "learning_rate": 1.371946703469236e-05,
+ "loss": 0.1149,
+ "step": 14582
+ },
+ {
+ "epoch": 39.735694822888284,
+ "grad_norm": 5.655329704284668,
+ "learning_rate": 1.3718647840957666e-05,
+ "loss": 0.087,
+ "step": 14583
+ },
+ {
+ "epoch": 39.73841961852861,
+ "grad_norm": 2.745274305343628,
+ "learning_rate": 1.3717828618262261e-05,
+ "loss": 0.0744,
+ "step": 14584
+ },
+ {
+ "epoch": 39.741144414168936,
+ "grad_norm": 2.3141210079193115,
+ "learning_rate": 1.3717009366612528e-05,
+ "loss": 0.0589,
+ "step": 14585
+ },
+ {
+ "epoch": 39.74386920980926,
+ "grad_norm": 3.504027843475342,
+ "learning_rate": 1.3716190086014844e-05,
+ "loss": 0.101,
+ "step": 14586
+ },
+ {
+ "epoch": 39.746594005449595,
+ "grad_norm": 3.0541915893554688,
+ "learning_rate": 1.3715370776475591e-05,
+ "loss": 0.2061,
+ "step": 14587
+ },
+ {
+ "epoch": 39.74931880108992,
+ "grad_norm": 3.882248640060425,
+ "learning_rate": 1.3714551438001149e-05,
+ "loss": 0.1376,
+ "step": 14588
+ },
+ {
+ "epoch": 39.752043596730246,
+ "grad_norm": 3.286653995513916,
+ "learning_rate": 1.3713732070597902e-05,
+ "loss": 0.2456,
+ "step": 14589
+ },
+ {
+ "epoch": 39.75476839237057,
+ "grad_norm": 3.798718214035034,
+ "learning_rate": 1.371291267427223e-05,
+ "loss": 0.1869,
+ "step": 14590
+ },
+ {
+ "epoch": 39.7574931880109,
+ "grad_norm": 3.808666467666626,
+ "learning_rate": 1.3712093249030513e-05,
+ "loss": 0.0975,
+ "step": 14591
+ },
+ {
+ "epoch": 39.76021798365122,
+ "grad_norm": 3.458028554916382,
+ "learning_rate": 1.3711273794879131e-05,
+ "loss": 0.1514,
+ "step": 14592
+ },
+ {
+ "epoch": 39.762942779291556,
+ "grad_norm": 3.0355467796325684,
+ "learning_rate": 1.3710454311824472e-05,
+ "loss": 0.1087,
+ "step": 14593
+ },
+ {
+ "epoch": 39.76566757493188,
+ "grad_norm": 5.280908584594727,
+ "learning_rate": 1.370963479987291e-05,
+ "loss": 0.1353,
+ "step": 14594
+ },
+ {
+ "epoch": 39.76839237057221,
+ "grad_norm": 4.164149284362793,
+ "learning_rate": 1.3708815259030835e-05,
+ "loss": 0.1087,
+ "step": 14595
+ },
+ {
+ "epoch": 39.771117166212534,
+ "grad_norm": 3.940078020095825,
+ "learning_rate": 1.3707995689304626e-05,
+ "loss": 0.2051,
+ "step": 14596
+ },
+ {
+ "epoch": 39.77384196185286,
+ "grad_norm": 3.539515972137451,
+ "learning_rate": 1.3707176090700662e-05,
+ "loss": 0.2233,
+ "step": 14597
+ },
+ {
+ "epoch": 39.776566757493185,
+ "grad_norm": 3.402662754058838,
+ "learning_rate": 1.3706356463225332e-05,
+ "loss": 0.2817,
+ "step": 14598
+ },
+ {
+ "epoch": 39.77929155313352,
+ "grad_norm": 2.653688907623291,
+ "learning_rate": 1.3705536806885018e-05,
+ "loss": 0.1574,
+ "step": 14599
+ },
+ {
+ "epoch": 39.782016348773844,
+ "grad_norm": 3.0516958236694336,
+ "learning_rate": 1.3704717121686103e-05,
+ "loss": 0.0988,
+ "step": 14600
+ },
+ {
+ "epoch": 39.78474114441417,
+ "grad_norm": 3.0986874103546143,
+ "learning_rate": 1.3703897407634971e-05,
+ "loss": 0.3464,
+ "step": 14601
+ },
+ {
+ "epoch": 39.787465940054496,
+ "grad_norm": 7.319231986999512,
+ "learning_rate": 1.3703077664738002e-05,
+ "loss": 0.1302,
+ "step": 14602
+ },
+ {
+ "epoch": 39.79019073569482,
+ "grad_norm": 2.857408285140991,
+ "learning_rate": 1.3702257893001588e-05,
+ "loss": 0.0804,
+ "step": 14603
+ },
+ {
+ "epoch": 39.79291553133515,
+ "grad_norm": 3.1558830738067627,
+ "learning_rate": 1.3701438092432104e-05,
+ "loss": 0.0812,
+ "step": 14604
+ },
+ {
+ "epoch": 39.79564032697548,
+ "grad_norm": 7.200352191925049,
+ "learning_rate": 1.3700618263035941e-05,
+ "loss": 0.1259,
+ "step": 14605
+ },
+ {
+ "epoch": 39.798365122615806,
+ "grad_norm": 3.4074127674102783,
+ "learning_rate": 1.3699798404819485e-05,
+ "loss": 0.1935,
+ "step": 14606
+ },
+ {
+ "epoch": 39.80108991825613,
+ "grad_norm": 2.7210261821746826,
+ "learning_rate": 1.3698978517789112e-05,
+ "loss": 0.0766,
+ "step": 14607
+ },
+ {
+ "epoch": 39.80381471389646,
+ "grad_norm": 3.267975330352783,
+ "learning_rate": 1.3698158601951217e-05,
+ "loss": 0.2538,
+ "step": 14608
+ },
+ {
+ "epoch": 39.80653950953678,
+ "grad_norm": 3.254826307296753,
+ "learning_rate": 1.3697338657312183e-05,
+ "loss": 0.1575,
+ "step": 14609
+ },
+ {
+ "epoch": 39.80926430517711,
+ "grad_norm": 2.594285249710083,
+ "learning_rate": 1.3696518683878392e-05,
+ "loss": 0.0509,
+ "step": 14610
+ },
+ {
+ "epoch": 39.81198910081744,
+ "grad_norm": 3.387139081954956,
+ "learning_rate": 1.3695698681656233e-05,
+ "loss": 0.1109,
+ "step": 14611
+ },
+ {
+ "epoch": 39.81471389645777,
+ "grad_norm": 3.4637563228607178,
+ "learning_rate": 1.3694878650652092e-05,
+ "loss": 0.1481,
+ "step": 14612
+ },
+ {
+ "epoch": 39.817438692098094,
+ "grad_norm": 3.70220947265625,
+ "learning_rate": 1.3694058590872354e-05,
+ "loss": 0.107,
+ "step": 14613
+ },
+ {
+ "epoch": 39.82016348773842,
+ "grad_norm": 3.3779854774475098,
+ "learning_rate": 1.369323850232341e-05,
+ "loss": 0.3575,
+ "step": 14614
+ },
+ {
+ "epoch": 39.822888283378745,
+ "grad_norm": 3.1462080478668213,
+ "learning_rate": 1.3692418385011639e-05,
+ "loss": 0.0973,
+ "step": 14615
+ },
+ {
+ "epoch": 39.82561307901907,
+ "grad_norm": 3.272115468978882,
+ "learning_rate": 1.3691598238943438e-05,
+ "loss": 0.2151,
+ "step": 14616
+ },
+ {
+ "epoch": 39.828337874659404,
+ "grad_norm": 3.0490615367889404,
+ "learning_rate": 1.3690778064125182e-05,
+ "loss": 0.2051,
+ "step": 14617
+ },
+ {
+ "epoch": 39.83106267029973,
+ "grad_norm": 4.255982875823975,
+ "learning_rate": 1.368995786056327e-05,
+ "loss": 0.3818,
+ "step": 14618
+ },
+ {
+ "epoch": 39.833787465940055,
+ "grad_norm": 3.1575708389282227,
+ "learning_rate": 1.3689137628264085e-05,
+ "loss": 0.1589,
+ "step": 14619
+ },
+ {
+ "epoch": 39.83651226158038,
+ "grad_norm": 3.215009927749634,
+ "learning_rate": 1.368831736723401e-05,
+ "loss": 0.1132,
+ "step": 14620
+ },
+ {
+ "epoch": 39.83923705722071,
+ "grad_norm": 3.222836494445801,
+ "learning_rate": 1.3687497077479441e-05,
+ "loss": 0.082,
+ "step": 14621
+ },
+ {
+ "epoch": 39.84196185286103,
+ "grad_norm": 3.0055935382843018,
+ "learning_rate": 1.3686676759006764e-05,
+ "loss": 0.2145,
+ "step": 14622
+ },
+ {
+ "epoch": 39.844686648501366,
+ "grad_norm": 3.500380516052246,
+ "learning_rate": 1.3685856411822367e-05,
+ "loss": 0.1781,
+ "step": 14623
+ },
+ {
+ "epoch": 39.84741144414169,
+ "grad_norm": 3.176892042160034,
+ "learning_rate": 1.3685036035932639e-05,
+ "loss": 0.0711,
+ "step": 14624
+ },
+ {
+ "epoch": 39.85013623978202,
+ "grad_norm": 3.9945287704467773,
+ "learning_rate": 1.3684215631343969e-05,
+ "loss": 0.1981,
+ "step": 14625
+ },
+ {
+ "epoch": 39.85286103542234,
+ "grad_norm": 3.5638980865478516,
+ "learning_rate": 1.3683395198062747e-05,
+ "loss": 0.3114,
+ "step": 14626
+ },
+ {
+ "epoch": 39.85558583106267,
+ "grad_norm": 3.049150228500366,
+ "learning_rate": 1.368257473609536e-05,
+ "loss": 0.1445,
+ "step": 14627
+ },
+ {
+ "epoch": 39.858310626702995,
+ "grad_norm": 3.1515471935272217,
+ "learning_rate": 1.3681754245448202e-05,
+ "loss": 0.1657,
+ "step": 14628
+ },
+ {
+ "epoch": 39.86103542234333,
+ "grad_norm": 3.6733274459838867,
+ "learning_rate": 1.3680933726127656e-05,
+ "loss": 0.1595,
+ "step": 14629
+ },
+ {
+ "epoch": 39.86376021798365,
+ "grad_norm": 3.5498251914978027,
+ "learning_rate": 1.368011317814012e-05,
+ "loss": 0.094,
+ "step": 14630
+ },
+ {
+ "epoch": 39.86648501362398,
+ "grad_norm": 4.234216213226318,
+ "learning_rate": 1.367929260149198e-05,
+ "loss": 0.2036,
+ "step": 14631
+ },
+ {
+ "epoch": 39.869209809264305,
+ "grad_norm": 3.4043545722961426,
+ "learning_rate": 1.3678471996189629e-05,
+ "loss": 0.1217,
+ "step": 14632
+ },
+ {
+ "epoch": 39.87193460490463,
+ "grad_norm": 3.7236766815185547,
+ "learning_rate": 1.3677651362239454e-05,
+ "loss": 0.1121,
+ "step": 14633
+ },
+ {
+ "epoch": 39.87465940054496,
+ "grad_norm": 3.8254899978637695,
+ "learning_rate": 1.3676830699647852e-05,
+ "loss": 0.082,
+ "step": 14634
+ },
+ {
+ "epoch": 39.87738419618529,
+ "grad_norm": 3.487609624862671,
+ "learning_rate": 1.3676010008421208e-05,
+ "loss": 0.1269,
+ "step": 14635
+ },
+ {
+ "epoch": 39.880108991825615,
+ "grad_norm": 3.496410369873047,
+ "learning_rate": 1.3675189288565918e-05,
+ "loss": 0.151,
+ "step": 14636
+ },
+ {
+ "epoch": 39.88283378746594,
+ "grad_norm": 3.462085008621216,
+ "learning_rate": 1.367436854008837e-05,
+ "loss": 0.2545,
+ "step": 14637
+ },
+ {
+ "epoch": 39.88555858310627,
+ "grad_norm": 2.9873971939086914,
+ "learning_rate": 1.3673547762994959e-05,
+ "loss": 0.2011,
+ "step": 14638
+ },
+ {
+ "epoch": 39.88828337874659,
+ "grad_norm": 2.7445178031921387,
+ "learning_rate": 1.3672726957292075e-05,
+ "loss": 0.1872,
+ "step": 14639
+ },
+ {
+ "epoch": 39.89100817438692,
+ "grad_norm": 3.527808427810669,
+ "learning_rate": 1.3671906122986114e-05,
+ "loss": 0.4366,
+ "step": 14640
+ },
+ {
+ "epoch": 39.89373297002725,
+ "grad_norm": 2.8832008838653564,
+ "learning_rate": 1.3671085260083465e-05,
+ "loss": 0.0594,
+ "step": 14641
+ },
+ {
+ "epoch": 39.89645776566758,
+ "grad_norm": 3.19586181640625,
+ "learning_rate": 1.3670264368590524e-05,
+ "loss": 0.0877,
+ "step": 14642
+ },
+ {
+ "epoch": 39.8991825613079,
+ "grad_norm": 2.848024845123291,
+ "learning_rate": 1.3669443448513679e-05,
+ "loss": 0.1432,
+ "step": 14643
+ },
+ {
+ "epoch": 39.90190735694823,
+ "grad_norm": 2.5029561519622803,
+ "learning_rate": 1.366862249985933e-05,
+ "loss": 0.1446,
+ "step": 14644
+ },
+ {
+ "epoch": 39.904632152588555,
+ "grad_norm": 3.3008716106414795,
+ "learning_rate": 1.3667801522633866e-05,
+ "loss": 0.0931,
+ "step": 14645
+ },
+ {
+ "epoch": 39.90735694822888,
+ "grad_norm": 2.6956427097320557,
+ "learning_rate": 1.3666980516843682e-05,
+ "loss": 0.0845,
+ "step": 14646
+ },
+ {
+ "epoch": 39.91008174386921,
+ "grad_norm": 3.3345324993133545,
+ "learning_rate": 1.3666159482495167e-05,
+ "loss": 0.1567,
+ "step": 14647
+ },
+ {
+ "epoch": 39.91280653950954,
+ "grad_norm": 3.1125009059906006,
+ "learning_rate": 1.3665338419594725e-05,
+ "loss": 0.1081,
+ "step": 14648
+ },
+ {
+ "epoch": 39.915531335149865,
+ "grad_norm": 3.4968204498291016,
+ "learning_rate": 1.3664517328148744e-05,
+ "loss": 0.122,
+ "step": 14649
+ },
+ {
+ "epoch": 39.91825613079019,
+ "grad_norm": 2.9148643016815186,
+ "learning_rate": 1.3663696208163622e-05,
+ "loss": 0.1006,
+ "step": 14650
+ },
+ {
+ "epoch": 39.920980926430516,
+ "grad_norm": 4.0753912925720215,
+ "learning_rate": 1.366287505964575e-05,
+ "loss": 0.3066,
+ "step": 14651
+ },
+ {
+ "epoch": 39.92370572207084,
+ "grad_norm": 3.801762819290161,
+ "learning_rate": 1.3662053882601525e-05,
+ "loss": 0.1381,
+ "step": 14652
+ },
+ {
+ "epoch": 39.926430517711175,
+ "grad_norm": 3.871091604232788,
+ "learning_rate": 1.3661232677037341e-05,
+ "loss": 0.1985,
+ "step": 14653
+ },
+ {
+ "epoch": 39.9291553133515,
+ "grad_norm": 4.259714126586914,
+ "learning_rate": 1.3660411442959597e-05,
+ "loss": 0.1263,
+ "step": 14654
+ },
+ {
+ "epoch": 39.93188010899183,
+ "grad_norm": 2.8830509185791016,
+ "learning_rate": 1.3659590180374685e-05,
+ "loss": 0.0807,
+ "step": 14655
+ },
+ {
+ "epoch": 39.93460490463215,
+ "grad_norm": 3.6432809829711914,
+ "learning_rate": 1.3658768889289003e-05,
+ "loss": 0.16,
+ "step": 14656
+ },
+ {
+ "epoch": 39.93732970027248,
+ "grad_norm": 3.095949649810791,
+ "learning_rate": 1.3657947569708946e-05,
+ "loss": 0.1222,
+ "step": 14657
+ },
+ {
+ "epoch": 39.940054495912804,
+ "grad_norm": 4.3556013107299805,
+ "learning_rate": 1.3657126221640914e-05,
+ "loss": 0.1226,
+ "step": 14658
+ },
+ {
+ "epoch": 39.94277929155314,
+ "grad_norm": 3.957587480545044,
+ "learning_rate": 1.3656304845091298e-05,
+ "loss": 0.1913,
+ "step": 14659
+ },
+ {
+ "epoch": 39.94550408719346,
+ "grad_norm": 5.863476276397705,
+ "learning_rate": 1.3655483440066496e-05,
+ "loss": 0.1982,
+ "step": 14660
+ },
+ {
+ "epoch": 39.94822888283379,
+ "grad_norm": 3.439257860183716,
+ "learning_rate": 1.3654662006572909e-05,
+ "loss": 0.2286,
+ "step": 14661
+ },
+ {
+ "epoch": 39.950953678474114,
+ "grad_norm": 3.0223300457000732,
+ "learning_rate": 1.3653840544616932e-05,
+ "loss": 0.1506,
+ "step": 14662
+ },
+ {
+ "epoch": 39.95367847411444,
+ "grad_norm": 3.4834647178649902,
+ "learning_rate": 1.3653019054204962e-05,
+ "loss": 0.1258,
+ "step": 14663
+ },
+ {
+ "epoch": 39.956403269754766,
+ "grad_norm": 4.00076961517334,
+ "learning_rate": 1.3652197535343396e-05,
+ "loss": 0.154,
+ "step": 14664
+ },
+ {
+ "epoch": 39.95912806539509,
+ "grad_norm": 2.989978075027466,
+ "learning_rate": 1.3651375988038633e-05,
+ "loss": 0.1312,
+ "step": 14665
+ },
+ {
+ "epoch": 39.961852861035425,
+ "grad_norm": 2.8565776348114014,
+ "learning_rate": 1.3650554412297075e-05,
+ "loss": 0.0756,
+ "step": 14666
+ },
+ {
+ "epoch": 39.96457765667575,
+ "grad_norm": 4.164483547210693,
+ "learning_rate": 1.3649732808125114e-05,
+ "loss": 0.246,
+ "step": 14667
+ },
+ {
+ "epoch": 39.967302452316076,
+ "grad_norm": 3.2191851139068604,
+ "learning_rate": 1.3648911175529151e-05,
+ "loss": 0.1516,
+ "step": 14668
+ },
+ {
+ "epoch": 39.9700272479564,
+ "grad_norm": 3.4668197631835938,
+ "learning_rate": 1.3648089514515585e-05,
+ "loss": 0.1373,
+ "step": 14669
+ },
+ {
+ "epoch": 39.97275204359673,
+ "grad_norm": 2.512021780014038,
+ "learning_rate": 1.3647267825090818e-05,
+ "loss": 0.1911,
+ "step": 14670
+ },
+ {
+ "epoch": 39.97547683923706,
+ "grad_norm": 3.2696151733398438,
+ "learning_rate": 1.3646446107261244e-05,
+ "loss": 0.1143,
+ "step": 14671
+ },
+ {
+ "epoch": 39.97820163487739,
+ "grad_norm": 3.458655595779419,
+ "learning_rate": 1.3645624361033268e-05,
+ "loss": 0.1081,
+ "step": 14672
+ },
+ {
+ "epoch": 39.98092643051771,
+ "grad_norm": 2.8968582153320312,
+ "learning_rate": 1.364480258641328e-05,
+ "loss": 0.1477,
+ "step": 14673
+ },
+ {
+ "epoch": 39.98365122615804,
+ "grad_norm": 3.3163208961486816,
+ "learning_rate": 1.3643980783407693e-05,
+ "loss": 0.234,
+ "step": 14674
+ },
+ {
+ "epoch": 39.986376021798364,
+ "grad_norm": 3.228137731552124,
+ "learning_rate": 1.36431589520229e-05,
+ "loss": 0.1105,
+ "step": 14675
+ },
+ {
+ "epoch": 39.98910081743869,
+ "grad_norm": 3.248363494873047,
+ "learning_rate": 1.3642337092265301e-05,
+ "loss": 0.1384,
+ "step": 14676
+ },
+ {
+ "epoch": 39.991825613079016,
+ "grad_norm": 3.1538584232330322,
+ "learning_rate": 1.3641515204141297e-05,
+ "loss": 0.0816,
+ "step": 14677
+ },
+ {
+ "epoch": 39.99455040871935,
+ "grad_norm": 4.53725004196167,
+ "learning_rate": 1.3640693287657294e-05,
+ "loss": 0.1431,
+ "step": 14678
+ },
+ {
+ "epoch": 39.997275204359674,
+ "grad_norm": 2.6238272190093994,
+ "learning_rate": 1.3639871342819686e-05,
+ "loss": 0.3043,
+ "step": 14679
+ },
+ {
+ "epoch": 40.0,
+ "grad_norm": 3.3229753971099854,
+ "learning_rate": 1.3639049369634878e-05,
+ "loss": 0.1766,
+ "step": 14680
+ },
+ {
+ "epoch": 40.002724795640326,
+ "grad_norm": 2.7847683429718018,
+ "learning_rate": 1.3638227368109268e-05,
+ "loss": 0.1179,
+ "step": 14681
+ },
+ {
+ "epoch": 40.00544959128065,
+ "grad_norm": 2.722151756286621,
+ "learning_rate": 1.3637405338249261e-05,
+ "loss": 0.073,
+ "step": 14682
+ },
+ {
+ "epoch": 40.00817438692098,
+ "grad_norm": 3.2934072017669678,
+ "learning_rate": 1.3636583280061262e-05,
+ "loss": 0.21,
+ "step": 14683
+ },
+ {
+ "epoch": 40.01089918256131,
+ "grad_norm": 2.983292579650879,
+ "learning_rate": 1.3635761193551666e-05,
+ "loss": 0.0934,
+ "step": 14684
+ },
+ {
+ "epoch": 40.013623978201636,
+ "grad_norm": 5.050808906555176,
+ "learning_rate": 1.3634939078726878e-05,
+ "loss": 0.2312,
+ "step": 14685
+ },
+ {
+ "epoch": 40.01634877384196,
+ "grad_norm": 3.502277135848999,
+ "learning_rate": 1.3634116935593307e-05,
+ "loss": 0.2566,
+ "step": 14686
+ },
+ {
+ "epoch": 40.01907356948229,
+ "grad_norm": 4.3973588943481445,
+ "learning_rate": 1.3633294764157344e-05,
+ "loss": 0.2522,
+ "step": 14687
+ },
+ {
+ "epoch": 40.02179836512261,
+ "grad_norm": 3.033175230026245,
+ "learning_rate": 1.3632472564425402e-05,
+ "loss": 0.143,
+ "step": 14688
+ },
+ {
+ "epoch": 40.02452316076294,
+ "grad_norm": 3.3919620513916016,
+ "learning_rate": 1.3631650336403879e-05,
+ "loss": 0.1454,
+ "step": 14689
+ },
+ {
+ "epoch": 40.02724795640327,
+ "grad_norm": 2.8513388633728027,
+ "learning_rate": 1.3630828080099181e-05,
+ "loss": 0.0699,
+ "step": 14690
+ },
+ {
+ "epoch": 40.0299727520436,
+ "grad_norm": 3.830301523208618,
+ "learning_rate": 1.3630005795517712e-05,
+ "loss": 0.071,
+ "step": 14691
+ },
+ {
+ "epoch": 40.032697547683924,
+ "grad_norm": 3.580322027206421,
+ "learning_rate": 1.362918348266587e-05,
+ "loss": 0.1536,
+ "step": 14692
+ },
+ {
+ "epoch": 40.03542234332425,
+ "grad_norm": 3.9467005729675293,
+ "learning_rate": 1.3628361141550068e-05,
+ "loss": 0.1517,
+ "step": 14693
+ },
+ {
+ "epoch": 40.038147138964575,
+ "grad_norm": 2.668837785720825,
+ "learning_rate": 1.3627538772176705e-05,
+ "loss": 0.1843,
+ "step": 14694
+ },
+ {
+ "epoch": 40.0408719346049,
+ "grad_norm": 2.9095306396484375,
+ "learning_rate": 1.3626716374552187e-05,
+ "loss": 0.0654,
+ "step": 14695
+ },
+ {
+ "epoch": 40.043596730245234,
+ "grad_norm": 2.946396589279175,
+ "learning_rate": 1.3625893948682919e-05,
+ "loss": 0.1625,
+ "step": 14696
+ },
+ {
+ "epoch": 40.04632152588556,
+ "grad_norm": 3.2889926433563232,
+ "learning_rate": 1.3625071494575304e-05,
+ "loss": 0.0703,
+ "step": 14697
+ },
+ {
+ "epoch": 40.049046321525886,
+ "grad_norm": 2.4861154556274414,
+ "learning_rate": 1.3624249012235748e-05,
+ "loss": 0.0675,
+ "step": 14698
+ },
+ {
+ "epoch": 40.05177111716621,
+ "grad_norm": 3.8316423892974854,
+ "learning_rate": 1.3623426501670662e-05,
+ "loss": 0.1629,
+ "step": 14699
+ },
+ {
+ "epoch": 40.05449591280654,
+ "grad_norm": 2.575942277908325,
+ "learning_rate": 1.3622603962886443e-05,
+ "loss": 0.1536,
+ "step": 14700
+ },
+ {
+ "epoch": 40.05722070844686,
+ "grad_norm": 3.8390865325927734,
+ "learning_rate": 1.3621781395889503e-05,
+ "loss": 0.2184,
+ "step": 14701
+ },
+ {
+ "epoch": 40.059945504087196,
+ "grad_norm": 3.627209424972534,
+ "learning_rate": 1.3620958800686246e-05,
+ "loss": 0.2495,
+ "step": 14702
+ },
+ {
+ "epoch": 40.06267029972752,
+ "grad_norm": 3.416364908218384,
+ "learning_rate": 1.3620136177283078e-05,
+ "loss": 0.0794,
+ "step": 14703
+ },
+ {
+ "epoch": 40.06539509536785,
+ "grad_norm": 3.8214333057403564,
+ "learning_rate": 1.3619313525686407e-05,
+ "loss": 0.1197,
+ "step": 14704
+ },
+ {
+ "epoch": 40.06811989100817,
+ "grad_norm": 2.547147750854492,
+ "learning_rate": 1.3618490845902637e-05,
+ "loss": 0.0705,
+ "step": 14705
+ },
+ {
+ "epoch": 40.0708446866485,
+ "grad_norm": 2.836092233657837,
+ "learning_rate": 1.3617668137938176e-05,
+ "loss": 0.1506,
+ "step": 14706
+ },
+ {
+ "epoch": 40.073569482288825,
+ "grad_norm": 3.525500774383545,
+ "learning_rate": 1.3616845401799433e-05,
+ "loss": 0.1451,
+ "step": 14707
+ },
+ {
+ "epoch": 40.07629427792916,
+ "grad_norm": 1.9627777338027954,
+ "learning_rate": 1.3616022637492814e-05,
+ "loss": 0.1357,
+ "step": 14708
+ },
+ {
+ "epoch": 40.079019073569484,
+ "grad_norm": 3.992858648300171,
+ "learning_rate": 1.3615199845024729e-05,
+ "loss": 0.1752,
+ "step": 14709
+ },
+ {
+ "epoch": 40.08174386920981,
+ "grad_norm": 3.1615636348724365,
+ "learning_rate": 1.3614377024401581e-05,
+ "loss": 0.3003,
+ "step": 14710
+ },
+ {
+ "epoch": 40.084468664850135,
+ "grad_norm": 3.67834734916687,
+ "learning_rate": 1.3613554175629783e-05,
+ "loss": 0.1581,
+ "step": 14711
+ },
+ {
+ "epoch": 40.08719346049046,
+ "grad_norm": 3.391097068786621,
+ "learning_rate": 1.3612731298715743e-05,
+ "loss": 0.1074,
+ "step": 14712
+ },
+ {
+ "epoch": 40.08991825613079,
+ "grad_norm": 3.836423873901367,
+ "learning_rate": 1.3611908393665866e-05,
+ "loss": 0.2653,
+ "step": 14713
+ },
+ {
+ "epoch": 40.09264305177112,
+ "grad_norm": 2.818084478378296,
+ "learning_rate": 1.3611085460486561e-05,
+ "loss": 0.0557,
+ "step": 14714
+ },
+ {
+ "epoch": 40.095367847411445,
+ "grad_norm": 2.4768424034118652,
+ "learning_rate": 1.3610262499184241e-05,
+ "loss": 0.0829,
+ "step": 14715
+ },
+ {
+ "epoch": 40.09809264305177,
+ "grad_norm": 2.7345590591430664,
+ "learning_rate": 1.360943950976531e-05,
+ "loss": 0.0906,
+ "step": 14716
+ },
+ {
+ "epoch": 40.1008174386921,
+ "grad_norm": 3.138591766357422,
+ "learning_rate": 1.3608616492236185e-05,
+ "loss": 0.1866,
+ "step": 14717
+ },
+ {
+ "epoch": 40.10354223433242,
+ "grad_norm": 3.2754628658294678,
+ "learning_rate": 1.3607793446603266e-05,
+ "loss": 0.1835,
+ "step": 14718
+ },
+ {
+ "epoch": 40.10626702997275,
+ "grad_norm": 3.1507506370544434,
+ "learning_rate": 1.3606970372872972e-05,
+ "loss": 0.1562,
+ "step": 14719
+ },
+ {
+ "epoch": 40.10899182561308,
+ "grad_norm": 2.7832024097442627,
+ "learning_rate": 1.3606147271051706e-05,
+ "loss": 0.0645,
+ "step": 14720
+ },
+ {
+ "epoch": 40.11171662125341,
+ "grad_norm": 2.9990580081939697,
+ "learning_rate": 1.3605324141145883e-05,
+ "loss": 0.0928,
+ "step": 14721
+ },
+ {
+ "epoch": 40.11444141689373,
+ "grad_norm": 3.3058533668518066,
+ "learning_rate": 1.3604500983161911e-05,
+ "loss": 0.1916,
+ "step": 14722
+ },
+ {
+ "epoch": 40.11716621253406,
+ "grad_norm": 3.1005401611328125,
+ "learning_rate": 1.36036777971062e-05,
+ "loss": 0.1455,
+ "step": 14723
+ },
+ {
+ "epoch": 40.119891008174385,
+ "grad_norm": 3.2392845153808594,
+ "learning_rate": 1.3602854582985163e-05,
+ "loss": 0.2401,
+ "step": 14724
+ },
+ {
+ "epoch": 40.12261580381471,
+ "grad_norm": 4.0169291496276855,
+ "learning_rate": 1.360203134080521e-05,
+ "loss": 0.1102,
+ "step": 14725
+ },
+ {
+ "epoch": 40.12534059945504,
+ "grad_norm": 3.107980251312256,
+ "learning_rate": 1.3601208070572754e-05,
+ "loss": 0.1803,
+ "step": 14726
+ },
+ {
+ "epoch": 40.12806539509537,
+ "grad_norm": 2.8345987796783447,
+ "learning_rate": 1.3600384772294206e-05,
+ "loss": 0.23,
+ "step": 14727
+ },
+ {
+ "epoch": 40.130790190735695,
+ "grad_norm": 2.9933595657348633,
+ "learning_rate": 1.3599561445975975e-05,
+ "loss": 0.2686,
+ "step": 14728
+ },
+ {
+ "epoch": 40.13351498637602,
+ "grad_norm": 7.724529266357422,
+ "learning_rate": 1.3598738091624477e-05,
+ "loss": 0.127,
+ "step": 14729
+ },
+ {
+ "epoch": 40.13623978201635,
+ "grad_norm": 3.4170422554016113,
+ "learning_rate": 1.3597914709246122e-05,
+ "loss": 0.1804,
+ "step": 14730
+ },
+ {
+ "epoch": 40.13896457765667,
+ "grad_norm": 3.4260387420654297,
+ "learning_rate": 1.3597091298847325e-05,
+ "loss": 0.1225,
+ "step": 14731
+ },
+ {
+ "epoch": 40.141689373297005,
+ "grad_norm": 2.521115779876709,
+ "learning_rate": 1.3596267860434496e-05,
+ "loss": 0.1564,
+ "step": 14732
+ },
+ {
+ "epoch": 40.14441416893733,
+ "grad_norm": 3.5803918838500977,
+ "learning_rate": 1.3595444394014046e-05,
+ "loss": 0.1691,
+ "step": 14733
+ },
+ {
+ "epoch": 40.14713896457766,
+ "grad_norm": 2.8172495365142822,
+ "learning_rate": 1.359462089959239e-05,
+ "loss": 0.1012,
+ "step": 14734
+ },
+ {
+ "epoch": 40.14986376021798,
+ "grad_norm": 2.636270523071289,
+ "learning_rate": 1.3593797377175949e-05,
+ "loss": 0.2162,
+ "step": 14735
+ },
+ {
+ "epoch": 40.15258855585831,
+ "grad_norm": 2.540724277496338,
+ "learning_rate": 1.3592973826771123e-05,
+ "loss": 0.1087,
+ "step": 14736
+ },
+ {
+ "epoch": 40.155313351498634,
+ "grad_norm": 3.5870697498321533,
+ "learning_rate": 1.3592150248384334e-05,
+ "loss": 0.272,
+ "step": 14737
+ },
+ {
+ "epoch": 40.15803814713897,
+ "grad_norm": 2.714653968811035,
+ "learning_rate": 1.3591326642021996e-05,
+ "loss": 0.1003,
+ "step": 14738
+ },
+ {
+ "epoch": 40.16076294277929,
+ "grad_norm": 2.4694159030914307,
+ "learning_rate": 1.3590503007690523e-05,
+ "loss": 0.1618,
+ "step": 14739
+ },
+ {
+ "epoch": 40.16348773841962,
+ "grad_norm": 2.612447738647461,
+ "learning_rate": 1.3589679345396323e-05,
+ "loss": 0.0833,
+ "step": 14740
+ },
+ {
+ "epoch": 40.166212534059945,
+ "grad_norm": 98.7608413696289,
+ "learning_rate": 1.3588855655145819e-05,
+ "loss": 0.1724,
+ "step": 14741
+ },
+ {
+ "epoch": 40.16893732970027,
+ "grad_norm": 3.1076080799102783,
+ "learning_rate": 1.3588031936945423e-05,
+ "loss": 0.2296,
+ "step": 14742
+ },
+ {
+ "epoch": 40.171662125340596,
+ "grad_norm": 2.893986463546753,
+ "learning_rate": 1.3587208190801549e-05,
+ "loss": 0.1606,
+ "step": 14743
+ },
+ {
+ "epoch": 40.17438692098093,
+ "grad_norm": 3.023024320602417,
+ "learning_rate": 1.3586384416720612e-05,
+ "loss": 0.0788,
+ "step": 14744
+ },
+ {
+ "epoch": 40.177111716621255,
+ "grad_norm": 3.622058629989624,
+ "learning_rate": 1.3585560614709033e-05,
+ "loss": 0.0757,
+ "step": 14745
+ },
+ {
+ "epoch": 40.17983651226158,
+ "grad_norm": 5.085446357727051,
+ "learning_rate": 1.3584736784773218e-05,
+ "loss": 0.1675,
+ "step": 14746
+ },
+ {
+ "epoch": 40.182561307901906,
+ "grad_norm": 3.4219167232513428,
+ "learning_rate": 1.3583912926919594e-05,
+ "loss": 0.0779,
+ "step": 14747
+ },
+ {
+ "epoch": 40.18528610354223,
+ "grad_norm": 3.711840867996216,
+ "learning_rate": 1.3583089041154565e-05,
+ "loss": 0.2049,
+ "step": 14748
+ },
+ {
+ "epoch": 40.18801089918256,
+ "grad_norm": 5.555823802947998,
+ "learning_rate": 1.358226512748456e-05,
+ "loss": 0.1239,
+ "step": 14749
+ },
+ {
+ "epoch": 40.19073569482289,
+ "grad_norm": 4.0163726806640625,
+ "learning_rate": 1.3581441185915982e-05,
+ "loss": 0.1021,
+ "step": 14750
+ },
+ {
+ "epoch": 40.19346049046322,
+ "grad_norm": 3.461228132247925,
+ "learning_rate": 1.3580617216455261e-05,
+ "loss": 0.0956,
+ "step": 14751
+ },
+ {
+ "epoch": 40.19618528610354,
+ "grad_norm": 3.556161403656006,
+ "learning_rate": 1.3579793219108805e-05,
+ "loss": 0.1426,
+ "step": 14752
+ },
+ {
+ "epoch": 40.19891008174387,
+ "grad_norm": 3.572796106338501,
+ "learning_rate": 1.3578969193883038e-05,
+ "loss": 0.2435,
+ "step": 14753
+ },
+ {
+ "epoch": 40.201634877384194,
+ "grad_norm": 3.887681484222412,
+ "learning_rate": 1.3578145140784373e-05,
+ "loss": 0.1434,
+ "step": 14754
+ },
+ {
+ "epoch": 40.20435967302452,
+ "grad_norm": 5.360838413238525,
+ "learning_rate": 1.3577321059819229e-05,
+ "loss": 0.1991,
+ "step": 14755
+ },
+ {
+ "epoch": 40.20708446866485,
+ "grad_norm": 4.987587928771973,
+ "learning_rate": 1.3576496950994022e-05,
+ "loss": 0.125,
+ "step": 14756
+ },
+ {
+ "epoch": 40.20980926430518,
+ "grad_norm": 4.655491828918457,
+ "learning_rate": 1.3575672814315174e-05,
+ "loss": 0.1585,
+ "step": 14757
+ },
+ {
+ "epoch": 40.212534059945504,
+ "grad_norm": 3.5645833015441895,
+ "learning_rate": 1.3574848649789102e-05,
+ "loss": 0.0823,
+ "step": 14758
+ },
+ {
+ "epoch": 40.21525885558583,
+ "grad_norm": 4.838129997253418,
+ "learning_rate": 1.357402445742222e-05,
+ "loss": 0.2504,
+ "step": 14759
+ },
+ {
+ "epoch": 40.217983651226156,
+ "grad_norm": 3.655717134475708,
+ "learning_rate": 1.3573200237220951e-05,
+ "loss": 0.1488,
+ "step": 14760
+ },
+ {
+ "epoch": 40.22070844686648,
+ "grad_norm": 4.133580684661865,
+ "learning_rate": 1.3572375989191717e-05,
+ "loss": 0.2046,
+ "step": 14761
+ },
+ {
+ "epoch": 40.223433242506815,
+ "grad_norm": 3.9970407485961914,
+ "learning_rate": 1.3571551713340931e-05,
+ "loss": 0.2576,
+ "step": 14762
+ },
+ {
+ "epoch": 40.22615803814714,
+ "grad_norm": 3.249878168106079,
+ "learning_rate": 1.3570727409675018e-05,
+ "loss": 0.1037,
+ "step": 14763
+ },
+ {
+ "epoch": 40.228882833787466,
+ "grad_norm": 3.499098300933838,
+ "learning_rate": 1.3569903078200393e-05,
+ "loss": 0.0949,
+ "step": 14764
+ },
+ {
+ "epoch": 40.23160762942779,
+ "grad_norm": 3.8147714138031006,
+ "learning_rate": 1.3569078718923478e-05,
+ "loss": 0.1171,
+ "step": 14765
+ },
+ {
+ "epoch": 40.23433242506812,
+ "grad_norm": 3.384699583053589,
+ "learning_rate": 1.3568254331850693e-05,
+ "loss": 0.0835,
+ "step": 14766
+ },
+ {
+ "epoch": 40.237057220708444,
+ "grad_norm": 5.31102180480957,
+ "learning_rate": 1.3567429916988457e-05,
+ "loss": 0.1674,
+ "step": 14767
+ },
+ {
+ "epoch": 40.23978201634878,
+ "grad_norm": 4.646358489990234,
+ "learning_rate": 1.3566605474343192e-05,
+ "loss": 0.1515,
+ "step": 14768
+ },
+ {
+ "epoch": 40.2425068119891,
+ "grad_norm": 5.266685485839844,
+ "learning_rate": 1.3565781003921318e-05,
+ "loss": 0.1472,
+ "step": 14769
+ },
+ {
+ "epoch": 40.24523160762943,
+ "grad_norm": 3.413543939590454,
+ "learning_rate": 1.3564956505729259e-05,
+ "loss": 0.2801,
+ "step": 14770
+ },
+ {
+ "epoch": 40.247956403269754,
+ "grad_norm": 5.611678600311279,
+ "learning_rate": 1.356413197977343e-05,
+ "loss": 0.1162,
+ "step": 14771
+ },
+ {
+ "epoch": 40.25068119891008,
+ "grad_norm": 4.605590343475342,
+ "learning_rate": 1.3563307426060258e-05,
+ "loss": 0.1219,
+ "step": 14772
+ },
+ {
+ "epoch": 40.253405994550405,
+ "grad_norm": 3.130608320236206,
+ "learning_rate": 1.3562482844596164e-05,
+ "loss": 0.1317,
+ "step": 14773
+ },
+ {
+ "epoch": 40.25613079019074,
+ "grad_norm": 4.768927097320557,
+ "learning_rate": 1.3561658235387562e-05,
+ "loss": 0.2798,
+ "step": 14774
+ },
+ {
+ "epoch": 40.258855585831064,
+ "grad_norm": 3.5179522037506104,
+ "learning_rate": 1.3560833598440887e-05,
+ "loss": 0.082,
+ "step": 14775
+ },
+ {
+ "epoch": 40.26158038147139,
+ "grad_norm": 3.726066827774048,
+ "learning_rate": 1.3560008933762552e-05,
+ "loss": 0.1261,
+ "step": 14776
+ },
+ {
+ "epoch": 40.264305177111716,
+ "grad_norm": 4.146425247192383,
+ "learning_rate": 1.355918424135898e-05,
+ "loss": 0.1855,
+ "step": 14777
+ },
+ {
+ "epoch": 40.26702997275204,
+ "grad_norm": 4.4072465896606445,
+ "learning_rate": 1.35583595212366e-05,
+ "loss": 0.4134,
+ "step": 14778
+ },
+ {
+ "epoch": 40.26975476839237,
+ "grad_norm": 4.750302791595459,
+ "learning_rate": 1.3557534773401827e-05,
+ "loss": 0.1844,
+ "step": 14779
+ },
+ {
+ "epoch": 40.2724795640327,
+ "grad_norm": 2.9954981803894043,
+ "learning_rate": 1.355670999786109e-05,
+ "loss": 0.1277,
+ "step": 14780
+ },
+ {
+ "epoch": 40.275204359673026,
+ "grad_norm": 3.057671308517456,
+ "learning_rate": 1.3555885194620806e-05,
+ "loss": 0.1109,
+ "step": 14781
+ },
+ {
+ "epoch": 40.27792915531335,
+ "grad_norm": 2.7754459381103516,
+ "learning_rate": 1.3555060363687407e-05,
+ "loss": 0.0837,
+ "step": 14782
+ },
+ {
+ "epoch": 40.28065395095368,
+ "grad_norm": 4.504101276397705,
+ "learning_rate": 1.355423550506731e-05,
+ "loss": 0.1455,
+ "step": 14783
+ },
+ {
+ "epoch": 40.283378746594,
+ "grad_norm": 3.1066384315490723,
+ "learning_rate": 1.3553410618766942e-05,
+ "loss": 0.1213,
+ "step": 14784
+ },
+ {
+ "epoch": 40.28610354223433,
+ "grad_norm": 3.915963888168335,
+ "learning_rate": 1.3552585704792724e-05,
+ "loss": 0.1089,
+ "step": 14785
+ },
+ {
+ "epoch": 40.28882833787466,
+ "grad_norm": 4.7101216316223145,
+ "learning_rate": 1.3551760763151087e-05,
+ "loss": 0.1605,
+ "step": 14786
+ },
+ {
+ "epoch": 40.29155313351499,
+ "grad_norm": 3.7818028926849365,
+ "learning_rate": 1.355093579384845e-05,
+ "loss": 0.1632,
+ "step": 14787
+ },
+ {
+ "epoch": 40.294277929155314,
+ "grad_norm": 3.7118797302246094,
+ "learning_rate": 1.3550110796891241e-05,
+ "loss": 0.1245,
+ "step": 14788
+ },
+ {
+ "epoch": 40.29700272479564,
+ "grad_norm": 3.580415725708008,
+ "learning_rate": 1.3549285772285881e-05,
+ "loss": 0.1455,
+ "step": 14789
+ },
+ {
+ "epoch": 40.299727520435965,
+ "grad_norm": 3.8852710723876953,
+ "learning_rate": 1.3548460720038798e-05,
+ "loss": 0.1804,
+ "step": 14790
+ },
+ {
+ "epoch": 40.30245231607629,
+ "grad_norm": 3.8866052627563477,
+ "learning_rate": 1.3547635640156415e-05,
+ "loss": 0.0986,
+ "step": 14791
+ },
+ {
+ "epoch": 40.305177111716624,
+ "grad_norm": 3.512385845184326,
+ "learning_rate": 1.3546810532645166e-05,
+ "loss": 0.1013,
+ "step": 14792
+ },
+ {
+ "epoch": 40.30790190735695,
+ "grad_norm": 3.4301693439483643,
+ "learning_rate": 1.3545985397511464e-05,
+ "loss": 0.0864,
+ "step": 14793
+ },
+ {
+ "epoch": 40.310626702997276,
+ "grad_norm": 4.581264019012451,
+ "learning_rate": 1.3545160234761745e-05,
+ "loss": 0.0834,
+ "step": 14794
+ },
+ {
+ "epoch": 40.3133514986376,
+ "grad_norm": 4.129714488983154,
+ "learning_rate": 1.354433504440243e-05,
+ "loss": 0.0996,
+ "step": 14795
+ },
+ {
+ "epoch": 40.31607629427793,
+ "grad_norm": 5.488058567047119,
+ "learning_rate": 1.3543509826439951e-05,
+ "loss": 0.1283,
+ "step": 14796
+ },
+ {
+ "epoch": 40.31880108991825,
+ "grad_norm": 3.628899574279785,
+ "learning_rate": 1.3542684580880729e-05,
+ "loss": 0.1899,
+ "step": 14797
+ },
+ {
+ "epoch": 40.321525885558586,
+ "grad_norm": 3.6265342235565186,
+ "learning_rate": 1.3541859307731198e-05,
+ "loss": 0.2031,
+ "step": 14798
+ },
+ {
+ "epoch": 40.32425068119891,
+ "grad_norm": 3.799354314804077,
+ "learning_rate": 1.3541034006997776e-05,
+ "loss": 0.0982,
+ "step": 14799
+ },
+ {
+ "epoch": 40.32697547683924,
+ "grad_norm": 3.5766632556915283,
+ "learning_rate": 1.3540208678686897e-05,
+ "loss": 0.1034,
+ "step": 14800
+ },
+ {
+ "epoch": 40.32970027247956,
+ "grad_norm": 3.493321657180786,
+ "learning_rate": 1.3539383322804986e-05,
+ "loss": 0.1142,
+ "step": 14801
+ },
+ {
+ "epoch": 40.33242506811989,
+ "grad_norm": 4.11686897277832,
+ "learning_rate": 1.353855793935847e-05,
+ "loss": 0.186,
+ "step": 14802
+ },
+ {
+ "epoch": 40.335149863760215,
+ "grad_norm": 3.6271562576293945,
+ "learning_rate": 1.353773252835378e-05,
+ "loss": 0.1448,
+ "step": 14803
+ },
+ {
+ "epoch": 40.33787465940055,
+ "grad_norm": 3.5770363807678223,
+ "learning_rate": 1.3536907089797344e-05,
+ "loss": 0.0877,
+ "step": 14804
+ },
+ {
+ "epoch": 40.34059945504087,
+ "grad_norm": 4.120598316192627,
+ "learning_rate": 1.3536081623695588e-05,
+ "loss": 0.1803,
+ "step": 14805
+ },
+ {
+ "epoch": 40.3433242506812,
+ "grad_norm": 3.7732748985290527,
+ "learning_rate": 1.3535256130054943e-05,
+ "loss": 0.2133,
+ "step": 14806
+ },
+ {
+ "epoch": 40.346049046321525,
+ "grad_norm": 3.3589272499084473,
+ "learning_rate": 1.3534430608881837e-05,
+ "loss": 0.1134,
+ "step": 14807
+ },
+ {
+ "epoch": 40.34877384196185,
+ "grad_norm": 3.299001693725586,
+ "learning_rate": 1.35336050601827e-05,
+ "loss": 0.1543,
+ "step": 14808
+ },
+ {
+ "epoch": 40.35149863760218,
+ "grad_norm": 2.9567198753356934,
+ "learning_rate": 1.353277948396396e-05,
+ "loss": 0.0806,
+ "step": 14809
+ },
+ {
+ "epoch": 40.35422343324251,
+ "grad_norm": 3.5866153240203857,
+ "learning_rate": 1.3531953880232044e-05,
+ "loss": 0.2609,
+ "step": 14810
+ },
+ {
+ "epoch": 40.356948228882835,
+ "grad_norm": 4.902384281158447,
+ "learning_rate": 1.3531128248993388e-05,
+ "loss": 0.1142,
+ "step": 14811
+ },
+ {
+ "epoch": 40.35967302452316,
+ "grad_norm": 3.2207436561584473,
+ "learning_rate": 1.353030259025442e-05,
+ "loss": 0.1545,
+ "step": 14812
+ },
+ {
+ "epoch": 40.36239782016349,
+ "grad_norm": 4.589351177215576,
+ "learning_rate": 1.3529476904021565e-05,
+ "loss": 0.1389,
+ "step": 14813
+ },
+ {
+ "epoch": 40.36512261580381,
+ "grad_norm": 3.3374531269073486,
+ "learning_rate": 1.3528651190301263e-05,
+ "loss": 0.0805,
+ "step": 14814
+ },
+ {
+ "epoch": 40.36784741144414,
+ "grad_norm": 4.076810836791992,
+ "learning_rate": 1.3527825449099936e-05,
+ "loss": 0.2224,
+ "step": 14815
+ },
+ {
+ "epoch": 40.37057220708447,
+ "grad_norm": 3.64052152633667,
+ "learning_rate": 1.3526999680424018e-05,
+ "loss": 0.117,
+ "step": 14816
+ },
+ {
+ "epoch": 40.3732970027248,
+ "grad_norm": 3.4253365993499756,
+ "learning_rate": 1.352617388427994e-05,
+ "loss": 0.091,
+ "step": 14817
+ },
+ {
+ "epoch": 40.37602179836512,
+ "grad_norm": 3.670166254043579,
+ "learning_rate": 1.3525348060674133e-05,
+ "loss": 0.1189,
+ "step": 14818
+ },
+ {
+ "epoch": 40.37874659400545,
+ "grad_norm": 3.366314649581909,
+ "learning_rate": 1.3524522209613027e-05,
+ "loss": 0.2086,
+ "step": 14819
+ },
+ {
+ "epoch": 40.381471389645775,
+ "grad_norm": 3.1134965419769287,
+ "learning_rate": 1.3523696331103058e-05,
+ "loss": 0.1901,
+ "step": 14820
+ },
+ {
+ "epoch": 40.3841961852861,
+ "grad_norm": 4.003781795501709,
+ "learning_rate": 1.3522870425150654e-05,
+ "loss": 0.2161,
+ "step": 14821
+ },
+ {
+ "epoch": 40.38692098092643,
+ "grad_norm": 4.070357322692871,
+ "learning_rate": 1.3522044491762253e-05,
+ "loss": 0.1431,
+ "step": 14822
+ },
+ {
+ "epoch": 40.38964577656676,
+ "grad_norm": 3.5765798091888428,
+ "learning_rate": 1.3521218530944276e-05,
+ "loss": 0.2788,
+ "step": 14823
+ },
+ {
+ "epoch": 40.392370572207085,
+ "grad_norm": 3.388995885848999,
+ "learning_rate": 1.3520392542703167e-05,
+ "loss": 0.1777,
+ "step": 14824
+ },
+ {
+ "epoch": 40.39509536784741,
+ "grad_norm": 3.979374408721924,
+ "learning_rate": 1.3519566527045354e-05,
+ "loss": 0.1155,
+ "step": 14825
+ },
+ {
+ "epoch": 40.39782016348774,
+ "grad_norm": 3.3959057331085205,
+ "learning_rate": 1.3518740483977268e-05,
+ "loss": 0.1048,
+ "step": 14826
+ },
+ {
+ "epoch": 40.40054495912806,
+ "grad_norm": 3.5722081661224365,
+ "learning_rate": 1.3517914413505344e-05,
+ "loss": 0.2449,
+ "step": 14827
+ },
+ {
+ "epoch": 40.403269754768395,
+ "grad_norm": 3.764178991317749,
+ "learning_rate": 1.3517088315636018e-05,
+ "loss": 0.0984,
+ "step": 14828
+ },
+ {
+ "epoch": 40.40599455040872,
+ "grad_norm": 3.898322820663452,
+ "learning_rate": 1.351626219037572e-05,
+ "loss": 0.292,
+ "step": 14829
+ },
+ {
+ "epoch": 40.40871934604905,
+ "grad_norm": 3.1324949264526367,
+ "learning_rate": 1.3515436037730887e-05,
+ "loss": 0.1425,
+ "step": 14830
+ },
+ {
+ "epoch": 40.41144414168937,
+ "grad_norm": 2.866201639175415,
+ "learning_rate": 1.3514609857707948e-05,
+ "loss": 0.0773,
+ "step": 14831
+ },
+ {
+ "epoch": 40.4141689373297,
+ "grad_norm": 3.2203071117401123,
+ "learning_rate": 1.3513783650313346e-05,
+ "loss": 0.0789,
+ "step": 14832
+ },
+ {
+ "epoch": 40.416893732970024,
+ "grad_norm": 3.651625394821167,
+ "learning_rate": 1.3512957415553505e-05,
+ "loss": 0.164,
+ "step": 14833
+ },
+ {
+ "epoch": 40.41961852861036,
+ "grad_norm": 3.764089345932007,
+ "learning_rate": 1.3512131153434867e-05,
+ "loss": 0.1899,
+ "step": 14834
+ },
+ {
+ "epoch": 40.42234332425068,
+ "grad_norm": 3.4797189235687256,
+ "learning_rate": 1.3511304863963863e-05,
+ "loss": 0.1766,
+ "step": 14835
+ },
+ {
+ "epoch": 40.42506811989101,
+ "grad_norm": 3.270343542098999,
+ "learning_rate": 1.351047854714693e-05,
+ "loss": 0.1017,
+ "step": 14836
+ },
+ {
+ "epoch": 40.427792915531334,
+ "grad_norm": 5.760008335113525,
+ "learning_rate": 1.3509652202990501e-05,
+ "loss": 0.1021,
+ "step": 14837
+ },
+ {
+ "epoch": 40.43051771117166,
+ "grad_norm": 3.4929771423339844,
+ "learning_rate": 1.350882583150102e-05,
+ "loss": 0.1461,
+ "step": 14838
+ },
+ {
+ "epoch": 40.433242506811986,
+ "grad_norm": 3.3491101264953613,
+ "learning_rate": 1.3507999432684911e-05,
+ "loss": 0.1241,
+ "step": 14839
+ },
+ {
+ "epoch": 40.43596730245232,
+ "grad_norm": 3.0704777240753174,
+ "learning_rate": 1.350717300654862e-05,
+ "loss": 0.1531,
+ "step": 14840
+ },
+ {
+ "epoch": 40.438692098092645,
+ "grad_norm": 4.3303632736206055,
+ "learning_rate": 1.3506346553098573e-05,
+ "loss": 0.1497,
+ "step": 14841
+ },
+ {
+ "epoch": 40.44141689373297,
+ "grad_norm": 3.8526406288146973,
+ "learning_rate": 1.3505520072341216e-05,
+ "loss": 0.1055,
+ "step": 14842
+ },
+ {
+ "epoch": 40.444141689373296,
+ "grad_norm": 4.007032871246338,
+ "learning_rate": 1.3504693564282977e-05,
+ "loss": 0.2368,
+ "step": 14843
+ },
+ {
+ "epoch": 40.44686648501362,
+ "grad_norm": 3.326892137527466,
+ "learning_rate": 1.3503867028930305e-05,
+ "loss": 0.0792,
+ "step": 14844
+ },
+ {
+ "epoch": 40.44959128065395,
+ "grad_norm": 2.6804862022399902,
+ "learning_rate": 1.350304046628962e-05,
+ "loss": 0.0691,
+ "step": 14845
+ },
+ {
+ "epoch": 40.45231607629428,
+ "grad_norm": 2.7847352027893066,
+ "learning_rate": 1.3502213876367373e-05,
+ "loss": 0.1373,
+ "step": 14846
+ },
+ {
+ "epoch": 40.45504087193461,
+ "grad_norm": 3.308497428894043,
+ "learning_rate": 1.3501387259169998e-05,
+ "loss": 0.1129,
+ "step": 14847
+ },
+ {
+ "epoch": 40.45776566757493,
+ "grad_norm": 3.2861292362213135,
+ "learning_rate": 1.3500560614703932e-05,
+ "loss": 0.143,
+ "step": 14848
+ },
+ {
+ "epoch": 40.46049046321526,
+ "grad_norm": 2.9979801177978516,
+ "learning_rate": 1.3499733942975611e-05,
+ "loss": 0.1196,
+ "step": 14849
+ },
+ {
+ "epoch": 40.463215258855584,
+ "grad_norm": 3.944535970687866,
+ "learning_rate": 1.3498907243991477e-05,
+ "loss": 0.0958,
+ "step": 14850
+ },
+ {
+ "epoch": 40.46594005449591,
+ "grad_norm": 6.717879772186279,
+ "learning_rate": 1.3498080517757961e-05,
+ "loss": 0.2865,
+ "step": 14851
+ },
+ {
+ "epoch": 40.46866485013624,
+ "grad_norm": 3.174757957458496,
+ "learning_rate": 1.3497253764281513e-05,
+ "loss": 0.0714,
+ "step": 14852
+ },
+ {
+ "epoch": 40.47138964577657,
+ "grad_norm": 2.629763126373291,
+ "learning_rate": 1.3496426983568561e-05,
+ "loss": 0.0728,
+ "step": 14853
+ },
+ {
+ "epoch": 40.474114441416894,
+ "grad_norm": 3.4493515491485596,
+ "learning_rate": 1.3495600175625548e-05,
+ "loss": 0.1482,
+ "step": 14854
+ },
+ {
+ "epoch": 40.47683923705722,
+ "grad_norm": 3.489800214767456,
+ "learning_rate": 1.3494773340458916e-05,
+ "loss": 0.145,
+ "step": 14855
+ },
+ {
+ "epoch": 40.479564032697546,
+ "grad_norm": 3.0767569541931152,
+ "learning_rate": 1.34939464780751e-05,
+ "loss": 0.0946,
+ "step": 14856
+ },
+ {
+ "epoch": 40.48228882833787,
+ "grad_norm": 3.186567544937134,
+ "learning_rate": 1.3493119588480541e-05,
+ "loss": 0.3,
+ "step": 14857
+ },
+ {
+ "epoch": 40.485013623978205,
+ "grad_norm": 3.3569540977478027,
+ "learning_rate": 1.349229267168168e-05,
+ "loss": 0.0975,
+ "step": 14858
+ },
+ {
+ "epoch": 40.48773841961853,
+ "grad_norm": 3.3809542655944824,
+ "learning_rate": 1.3491465727684953e-05,
+ "loss": 0.2675,
+ "step": 14859
+ },
+ {
+ "epoch": 40.490463215258856,
+ "grad_norm": 3.2522244453430176,
+ "learning_rate": 1.3490638756496807e-05,
+ "loss": 0.1827,
+ "step": 14860
+ },
+ {
+ "epoch": 40.49318801089918,
+ "grad_norm": 3.7186036109924316,
+ "learning_rate": 1.348981175812368e-05,
+ "loss": 0.1224,
+ "step": 14861
+ },
+ {
+ "epoch": 40.49591280653951,
+ "grad_norm": 3.0743587017059326,
+ "learning_rate": 1.3488984732572006e-05,
+ "loss": 0.1002,
+ "step": 14862
+ },
+ {
+ "epoch": 40.49863760217983,
+ "grad_norm": 3.870176315307617,
+ "learning_rate": 1.3488157679848235e-05,
+ "loss": 0.1713,
+ "step": 14863
+ },
+ {
+ "epoch": 40.50136239782017,
+ "grad_norm": 2.9159040451049805,
+ "learning_rate": 1.3487330599958802e-05,
+ "loss": 0.156,
+ "step": 14864
+ },
+ {
+ "epoch": 40.50408719346049,
+ "grad_norm": 3.111279010772705,
+ "learning_rate": 1.3486503492910155e-05,
+ "loss": 0.0977,
+ "step": 14865
+ },
+ {
+ "epoch": 40.50681198910082,
+ "grad_norm": 3.272963285446167,
+ "learning_rate": 1.3485676358708725e-05,
+ "loss": 0.1519,
+ "step": 14866
+ },
+ {
+ "epoch": 40.509536784741144,
+ "grad_norm": 3.6467597484588623,
+ "learning_rate": 1.3484849197360964e-05,
+ "loss": 0.0921,
+ "step": 14867
+ },
+ {
+ "epoch": 40.51226158038147,
+ "grad_norm": 3.380493402481079,
+ "learning_rate": 1.348402200887331e-05,
+ "loss": 0.1818,
+ "step": 14868
+ },
+ {
+ "epoch": 40.514986376021795,
+ "grad_norm": 3.167201280593872,
+ "learning_rate": 1.3483194793252202e-05,
+ "loss": 0.2195,
+ "step": 14869
+ },
+ {
+ "epoch": 40.51771117166213,
+ "grad_norm": 4.466792106628418,
+ "learning_rate": 1.3482367550504085e-05,
+ "loss": 0.0643,
+ "step": 14870
+ },
+ {
+ "epoch": 40.520435967302454,
+ "grad_norm": 3.6125645637512207,
+ "learning_rate": 1.3481540280635403e-05,
+ "loss": 0.1672,
+ "step": 14871
+ },
+ {
+ "epoch": 40.52316076294278,
+ "grad_norm": 3.9646637439727783,
+ "learning_rate": 1.3480712983652598e-05,
+ "loss": 0.1557,
+ "step": 14872
+ },
+ {
+ "epoch": 40.525885558583106,
+ "grad_norm": 4.313509464263916,
+ "learning_rate": 1.3479885659562111e-05,
+ "loss": 0.1438,
+ "step": 14873
+ },
+ {
+ "epoch": 40.52861035422343,
+ "grad_norm": 3.7224154472351074,
+ "learning_rate": 1.3479058308370386e-05,
+ "loss": 0.2153,
+ "step": 14874
+ },
+ {
+ "epoch": 40.53133514986376,
+ "grad_norm": 3.079724073410034,
+ "learning_rate": 1.3478230930083868e-05,
+ "loss": 0.1251,
+ "step": 14875
+ },
+ {
+ "epoch": 40.53405994550409,
+ "grad_norm": 3.0325376987457275,
+ "learning_rate": 1.3477403524708998e-05,
+ "loss": 0.1155,
+ "step": 14876
+ },
+ {
+ "epoch": 40.536784741144416,
+ "grad_norm": 3.275496244430542,
+ "learning_rate": 1.3476576092252225e-05,
+ "loss": 0.0915,
+ "step": 14877
+ },
+ {
+ "epoch": 40.53950953678474,
+ "grad_norm": 3.3083181381225586,
+ "learning_rate": 1.3475748632719985e-05,
+ "loss": 0.1437,
+ "step": 14878
+ },
+ {
+ "epoch": 40.54223433242507,
+ "grad_norm": 3.210522174835205,
+ "learning_rate": 1.3474921146118728e-05,
+ "loss": 0.1498,
+ "step": 14879
+ },
+ {
+ "epoch": 40.54495912806539,
+ "grad_norm": 2.904879093170166,
+ "learning_rate": 1.3474093632454897e-05,
+ "loss": 0.1331,
+ "step": 14880
+ },
+ {
+ "epoch": 40.54768392370572,
+ "grad_norm": 5.160099506378174,
+ "learning_rate": 1.3473266091734938e-05,
+ "loss": 0.2234,
+ "step": 14881
+ },
+ {
+ "epoch": 40.55040871934605,
+ "grad_norm": 2.809971332550049,
+ "learning_rate": 1.3472438523965292e-05,
+ "loss": 0.0941,
+ "step": 14882
+ },
+ {
+ "epoch": 40.55313351498638,
+ "grad_norm": 3.65234637260437,
+ "learning_rate": 1.3471610929152408e-05,
+ "loss": 0.2621,
+ "step": 14883
+ },
+ {
+ "epoch": 40.555858310626704,
+ "grad_norm": 4.270843505859375,
+ "learning_rate": 1.3470783307302728e-05,
+ "loss": 0.1065,
+ "step": 14884
+ },
+ {
+ "epoch": 40.55858310626703,
+ "grad_norm": 2.774635076522827,
+ "learning_rate": 1.3469955658422701e-05,
+ "loss": 0.1952,
+ "step": 14885
+ },
+ {
+ "epoch": 40.561307901907355,
+ "grad_norm": 3.2634565830230713,
+ "learning_rate": 1.3469127982518773e-05,
+ "loss": 0.1007,
+ "step": 14886
+ },
+ {
+ "epoch": 40.56403269754768,
+ "grad_norm": 2.9703128337860107,
+ "learning_rate": 1.3468300279597386e-05,
+ "loss": 0.1971,
+ "step": 14887
+ },
+ {
+ "epoch": 40.566757493188014,
+ "grad_norm": 3.4170925617218018,
+ "learning_rate": 1.3467472549664983e-05,
+ "loss": 0.1307,
+ "step": 14888
+ },
+ {
+ "epoch": 40.56948228882834,
+ "grad_norm": 3.621896743774414,
+ "learning_rate": 1.3466644792728023e-05,
+ "loss": 0.1085,
+ "step": 14889
+ },
+ {
+ "epoch": 40.572207084468666,
+ "grad_norm": 3.8706700801849365,
+ "learning_rate": 1.3465817008792943e-05,
+ "loss": 0.0981,
+ "step": 14890
+ },
+ {
+ "epoch": 40.57493188010899,
+ "grad_norm": 3.278569221496582,
+ "learning_rate": 1.3464989197866192e-05,
+ "loss": 0.1713,
+ "step": 14891
+ },
+ {
+ "epoch": 40.57765667574932,
+ "grad_norm": 3.277909278869629,
+ "learning_rate": 1.3464161359954214e-05,
+ "loss": 0.1694,
+ "step": 14892
+ },
+ {
+ "epoch": 40.58038147138964,
+ "grad_norm": 3.016818046569824,
+ "learning_rate": 1.3463333495063462e-05,
+ "loss": 0.1289,
+ "step": 14893
+ },
+ {
+ "epoch": 40.583106267029976,
+ "grad_norm": 3.147191047668457,
+ "learning_rate": 1.3462505603200377e-05,
+ "loss": 0.1166,
+ "step": 14894
+ },
+ {
+ "epoch": 40.5858310626703,
+ "grad_norm": 3.473580837249756,
+ "learning_rate": 1.3461677684371412e-05,
+ "loss": 0.1706,
+ "step": 14895
+ },
+ {
+ "epoch": 40.58855585831063,
+ "grad_norm": 4.963214874267578,
+ "learning_rate": 1.346084973858301e-05,
+ "loss": 0.1367,
+ "step": 14896
+ },
+ {
+ "epoch": 40.59128065395095,
+ "grad_norm": 2.93817138671875,
+ "learning_rate": 1.3460021765841625e-05,
+ "loss": 0.0826,
+ "step": 14897
+ },
+ {
+ "epoch": 40.59400544959128,
+ "grad_norm": 3.4494705200195312,
+ "learning_rate": 1.34591937661537e-05,
+ "loss": 0.0899,
+ "step": 14898
+ },
+ {
+ "epoch": 40.596730245231605,
+ "grad_norm": 2.820005178451538,
+ "learning_rate": 1.3458365739525685e-05,
+ "loss": 0.0942,
+ "step": 14899
+ },
+ {
+ "epoch": 40.59945504087194,
+ "grad_norm": 3.430879592895508,
+ "learning_rate": 1.3457537685964029e-05,
+ "loss": 0.2417,
+ "step": 14900
+ },
+ {
+ "epoch": 40.60217983651226,
+ "grad_norm": 3.6186141967773438,
+ "learning_rate": 1.3456709605475182e-05,
+ "loss": 0.1512,
+ "step": 14901
+ },
+ {
+ "epoch": 40.60490463215259,
+ "grad_norm": 3.6986708641052246,
+ "learning_rate": 1.345588149806559e-05,
+ "loss": 0.1461,
+ "step": 14902
+ },
+ {
+ "epoch": 40.607629427792915,
+ "grad_norm": 3.421196460723877,
+ "learning_rate": 1.3455053363741708e-05,
+ "loss": 0.0885,
+ "step": 14903
+ },
+ {
+ "epoch": 40.61035422343324,
+ "grad_norm": 3.380784034729004,
+ "learning_rate": 1.3454225202509978e-05,
+ "loss": 0.1753,
+ "step": 14904
+ },
+ {
+ "epoch": 40.61307901907357,
+ "grad_norm": 3.068056106567383,
+ "learning_rate": 1.3453397014376854e-05,
+ "loss": 0.1406,
+ "step": 14905
+ },
+ {
+ "epoch": 40.6158038147139,
+ "grad_norm": 3.4337141513824463,
+ "learning_rate": 1.3452568799348783e-05,
+ "loss": 0.133,
+ "step": 14906
+ },
+ {
+ "epoch": 40.618528610354225,
+ "grad_norm": 3.3439910411834717,
+ "learning_rate": 1.3451740557432223e-05,
+ "loss": 0.1213,
+ "step": 14907
+ },
+ {
+ "epoch": 40.62125340599455,
+ "grad_norm": 3.414219856262207,
+ "learning_rate": 1.3450912288633614e-05,
+ "loss": 0.1845,
+ "step": 14908
+ },
+ {
+ "epoch": 40.62397820163488,
+ "grad_norm": 3.3287744522094727,
+ "learning_rate": 1.3450083992959413e-05,
+ "loss": 0.1096,
+ "step": 14909
+ },
+ {
+ "epoch": 40.6267029972752,
+ "grad_norm": 3.2261531352996826,
+ "learning_rate": 1.3449255670416068e-05,
+ "loss": 0.1962,
+ "step": 14910
+ },
+ {
+ "epoch": 40.62942779291553,
+ "grad_norm": 3.326547861099243,
+ "learning_rate": 1.3448427321010034e-05,
+ "loss": 0.0844,
+ "step": 14911
+ },
+ {
+ "epoch": 40.63215258855586,
+ "grad_norm": 2.750291109085083,
+ "learning_rate": 1.3447598944747755e-05,
+ "loss": 0.0665,
+ "step": 14912
+ },
+ {
+ "epoch": 40.63487738419619,
+ "grad_norm": 2.7609305381774902,
+ "learning_rate": 1.3446770541635687e-05,
+ "loss": 0.1262,
+ "step": 14913
+ },
+ {
+ "epoch": 40.63760217983651,
+ "grad_norm": 3.493389129638672,
+ "learning_rate": 1.3445942111680279e-05,
+ "loss": 0.119,
+ "step": 14914
+ },
+ {
+ "epoch": 40.64032697547684,
+ "grad_norm": 4.542502403259277,
+ "learning_rate": 1.3445113654887992e-05,
+ "loss": 0.1273,
+ "step": 14915
+ },
+ {
+ "epoch": 40.643051771117165,
+ "grad_norm": 4.942398548126221,
+ "learning_rate": 1.3444285171265263e-05,
+ "loss": 0.1041,
+ "step": 14916
+ },
+ {
+ "epoch": 40.64577656675749,
+ "grad_norm": 8.127179145812988,
+ "learning_rate": 1.3443456660818557e-05,
+ "loss": 0.1752,
+ "step": 14917
+ },
+ {
+ "epoch": 40.64850136239782,
+ "grad_norm": 3.075187921524048,
+ "learning_rate": 1.3442628123554318e-05,
+ "loss": 0.1504,
+ "step": 14918
+ },
+ {
+ "epoch": 40.65122615803815,
+ "grad_norm": 3.006960391998291,
+ "learning_rate": 1.3441799559479005e-05,
+ "loss": 0.1551,
+ "step": 14919
+ },
+ {
+ "epoch": 40.653950953678475,
+ "grad_norm": 3.111560106277466,
+ "learning_rate": 1.3440970968599065e-05,
+ "loss": 0.0942,
+ "step": 14920
+ },
+ {
+ "epoch": 40.6566757493188,
+ "grad_norm": 3.394669771194458,
+ "learning_rate": 1.3440142350920958e-05,
+ "loss": 0.1449,
+ "step": 14921
+ },
+ {
+ "epoch": 40.65940054495913,
+ "grad_norm": 3.70312237739563,
+ "learning_rate": 1.3439313706451125e-05,
+ "loss": 0.1263,
+ "step": 14922
+ },
+ {
+ "epoch": 40.66212534059945,
+ "grad_norm": 4.098132610321045,
+ "learning_rate": 1.3438485035196034e-05,
+ "loss": 0.1534,
+ "step": 14923
+ },
+ {
+ "epoch": 40.664850136239785,
+ "grad_norm": 3.7208523750305176,
+ "learning_rate": 1.343765633716213e-05,
+ "loss": 0.0932,
+ "step": 14924
+ },
+ {
+ "epoch": 40.66757493188011,
+ "grad_norm": 3.553236246109009,
+ "learning_rate": 1.3436827612355871e-05,
+ "loss": 0.1054,
+ "step": 14925
+ },
+ {
+ "epoch": 40.67029972752044,
+ "grad_norm": 2.97786283493042,
+ "learning_rate": 1.3435998860783708e-05,
+ "loss": 0.0785,
+ "step": 14926
+ },
+ {
+ "epoch": 40.67302452316076,
+ "grad_norm": 4.0134758949279785,
+ "learning_rate": 1.3435170082452097e-05,
+ "loss": 0.1407,
+ "step": 14927
+ },
+ {
+ "epoch": 40.67574931880109,
+ "grad_norm": 3.7470717430114746,
+ "learning_rate": 1.3434341277367491e-05,
+ "loss": 0.3112,
+ "step": 14928
+ },
+ {
+ "epoch": 40.678474114441414,
+ "grad_norm": 3.0631792545318604,
+ "learning_rate": 1.343351244553635e-05,
+ "loss": 0.1371,
+ "step": 14929
+ },
+ {
+ "epoch": 40.68119891008175,
+ "grad_norm": 2.7483084201812744,
+ "learning_rate": 1.343268358696512e-05,
+ "loss": 0.1298,
+ "step": 14930
+ },
+ {
+ "epoch": 40.68392370572207,
+ "grad_norm": 3.129152297973633,
+ "learning_rate": 1.3431854701660259e-05,
+ "loss": 0.1178,
+ "step": 14931
+ },
+ {
+ "epoch": 40.6866485013624,
+ "grad_norm": 4.2710723876953125,
+ "learning_rate": 1.3431025789628227e-05,
+ "loss": 0.1246,
+ "step": 14932
+ },
+ {
+ "epoch": 40.689373297002724,
+ "grad_norm": 4.343286037445068,
+ "learning_rate": 1.3430196850875476e-05,
+ "loss": 0.158,
+ "step": 14933
+ },
+ {
+ "epoch": 40.69209809264305,
+ "grad_norm": 3.0455493927001953,
+ "learning_rate": 1.3429367885408461e-05,
+ "loss": 0.0939,
+ "step": 14934
+ },
+ {
+ "epoch": 40.694822888283376,
+ "grad_norm": 3.7838780879974365,
+ "learning_rate": 1.3428538893233642e-05,
+ "loss": 0.2784,
+ "step": 14935
+ },
+ {
+ "epoch": 40.69754768392371,
+ "grad_norm": 3.4545738697052,
+ "learning_rate": 1.3427709874357471e-05,
+ "loss": 0.0995,
+ "step": 14936
+ },
+ {
+ "epoch": 40.700272479564035,
+ "grad_norm": 3.8315842151641846,
+ "learning_rate": 1.3426880828786408e-05,
+ "loss": 0.1968,
+ "step": 14937
+ },
+ {
+ "epoch": 40.70299727520436,
+ "grad_norm": 3.2948780059814453,
+ "learning_rate": 1.3426051756526905e-05,
+ "loss": 0.1943,
+ "step": 14938
+ },
+ {
+ "epoch": 40.705722070844686,
+ "grad_norm": 15.598267555236816,
+ "learning_rate": 1.342522265758542e-05,
+ "loss": 0.1862,
+ "step": 14939
+ },
+ {
+ "epoch": 40.70844686648501,
+ "grad_norm": 3.3405938148498535,
+ "learning_rate": 1.3424393531968415e-05,
+ "loss": 0.161,
+ "step": 14940
+ },
+ {
+ "epoch": 40.71117166212534,
+ "grad_norm": 3.7480008602142334,
+ "learning_rate": 1.3423564379682339e-05,
+ "loss": 0.3013,
+ "step": 14941
+ },
+ {
+ "epoch": 40.71389645776567,
+ "grad_norm": 5.75825309753418,
+ "learning_rate": 1.342273520073366e-05,
+ "loss": 0.2219,
+ "step": 14942
+ },
+ {
+ "epoch": 40.716621253406,
+ "grad_norm": 2.9516608715057373,
+ "learning_rate": 1.3421905995128821e-05,
+ "loss": 0.0591,
+ "step": 14943
+ },
+ {
+ "epoch": 40.71934604904632,
+ "grad_norm": 3.894503116607666,
+ "learning_rate": 1.3421076762874295e-05,
+ "loss": 0.1486,
+ "step": 14944
+ },
+ {
+ "epoch": 40.72207084468665,
+ "grad_norm": 3.548152446746826,
+ "learning_rate": 1.3420247503976529e-05,
+ "loss": 0.2047,
+ "step": 14945
+ },
+ {
+ "epoch": 40.724795640326974,
+ "grad_norm": 3.4489381313323975,
+ "learning_rate": 1.3419418218441989e-05,
+ "loss": 0.2329,
+ "step": 14946
+ },
+ {
+ "epoch": 40.7275204359673,
+ "grad_norm": 18.938074111938477,
+ "learning_rate": 1.3418588906277128e-05,
+ "loss": 0.0727,
+ "step": 14947
+ },
+ {
+ "epoch": 40.73024523160763,
+ "grad_norm": 3.2476961612701416,
+ "learning_rate": 1.3417759567488407e-05,
+ "loss": 0.1575,
+ "step": 14948
+ },
+ {
+ "epoch": 40.73297002724796,
+ "grad_norm": 3.270653486251831,
+ "learning_rate": 1.3416930202082284e-05,
+ "loss": 0.1314,
+ "step": 14949
+ },
+ {
+ "epoch": 40.735694822888284,
+ "grad_norm": 3.3480966091156006,
+ "learning_rate": 1.341610081006522e-05,
+ "loss": 0.1568,
+ "step": 14950
+ },
+ {
+ "epoch": 40.73841961852861,
+ "grad_norm": 3.1426241397857666,
+ "learning_rate": 1.341527139144367e-05,
+ "loss": 0.0815,
+ "step": 14951
+ },
+ {
+ "epoch": 40.741144414168936,
+ "grad_norm": 3.158989906311035,
+ "learning_rate": 1.3414441946224099e-05,
+ "loss": 0.1734,
+ "step": 14952
+ },
+ {
+ "epoch": 40.74386920980926,
+ "grad_norm": 3.3693764209747314,
+ "learning_rate": 1.3413612474412965e-05,
+ "loss": 0.1011,
+ "step": 14953
+ },
+ {
+ "epoch": 40.746594005449595,
+ "grad_norm": 2.83599591255188,
+ "learning_rate": 1.3412782976016724e-05,
+ "loss": 0.123,
+ "step": 14954
+ },
+ {
+ "epoch": 40.74931880108992,
+ "grad_norm": 3.918226718902588,
+ "learning_rate": 1.3411953451041839e-05,
+ "loss": 0.1616,
+ "step": 14955
+ },
+ {
+ "epoch": 40.752043596730246,
+ "grad_norm": 3.2864797115325928,
+ "learning_rate": 1.3411123899494772e-05,
+ "loss": 0.1143,
+ "step": 14956
+ },
+ {
+ "epoch": 40.75476839237057,
+ "grad_norm": 4.343465328216553,
+ "learning_rate": 1.341029432138198e-05,
+ "loss": 0.3072,
+ "step": 14957
+ },
+ {
+ "epoch": 40.7574931880109,
+ "grad_norm": 3.512840509414673,
+ "learning_rate": 1.3409464716709928e-05,
+ "loss": 0.1218,
+ "step": 14958
+ },
+ {
+ "epoch": 40.76021798365122,
+ "grad_norm": 3.1422860622406006,
+ "learning_rate": 1.3408635085485074e-05,
+ "loss": 0.0885,
+ "step": 14959
+ },
+ {
+ "epoch": 40.762942779291556,
+ "grad_norm": 2.9213359355926514,
+ "learning_rate": 1.340780542771388e-05,
+ "loss": 0.2931,
+ "step": 14960
+ },
+ {
+ "epoch": 40.76566757493188,
+ "grad_norm": 3.577897548675537,
+ "learning_rate": 1.3406975743402805e-05,
+ "loss": 0.1114,
+ "step": 14961
+ },
+ {
+ "epoch": 40.76839237057221,
+ "grad_norm": 3.807793140411377,
+ "learning_rate": 1.3406146032558316e-05,
+ "loss": 0.2336,
+ "step": 14962
+ },
+ {
+ "epoch": 40.771117166212534,
+ "grad_norm": 3.205376148223877,
+ "learning_rate": 1.3405316295186868e-05,
+ "loss": 0.1166,
+ "step": 14963
+ },
+ {
+ "epoch": 40.77384196185286,
+ "grad_norm": 3.5111122131347656,
+ "learning_rate": 1.3404486531294926e-05,
+ "loss": 0.0843,
+ "step": 14964
+ },
+ {
+ "epoch": 40.776566757493185,
+ "grad_norm": 2.6630635261535645,
+ "learning_rate": 1.3403656740888953e-05,
+ "loss": 0.118,
+ "step": 14965
+ },
+ {
+ "epoch": 40.77929155313352,
+ "grad_norm": 3.003377676010132,
+ "learning_rate": 1.3402826923975414e-05,
+ "loss": 0.0739,
+ "step": 14966
+ },
+ {
+ "epoch": 40.782016348773844,
+ "grad_norm": 3.19866681098938,
+ "learning_rate": 1.3401997080560765e-05,
+ "loss": 0.0795,
+ "step": 14967
+ },
+ {
+ "epoch": 40.78474114441417,
+ "grad_norm": 3.807825803756714,
+ "learning_rate": 1.3401167210651474e-05,
+ "loss": 0.1027,
+ "step": 14968
+ },
+ {
+ "epoch": 40.787465940054496,
+ "grad_norm": 3.7281863689422607,
+ "learning_rate": 1.3400337314254e-05,
+ "loss": 0.1122,
+ "step": 14969
+ },
+ {
+ "epoch": 40.79019073569482,
+ "grad_norm": 3.3738179206848145,
+ "learning_rate": 1.339950739137481e-05,
+ "loss": 0.1476,
+ "step": 14970
+ },
+ {
+ "epoch": 40.79291553133515,
+ "grad_norm": 20.895627975463867,
+ "learning_rate": 1.3398677442020367e-05,
+ "loss": 0.169,
+ "step": 14971
+ },
+ {
+ "epoch": 40.79564032697548,
+ "grad_norm": 4.790477752685547,
+ "learning_rate": 1.3397847466197133e-05,
+ "loss": 0.3124,
+ "step": 14972
+ },
+ {
+ "epoch": 40.798365122615806,
+ "grad_norm": 2.7995028495788574,
+ "learning_rate": 1.339701746391157e-05,
+ "loss": 0.1168,
+ "step": 14973
+ },
+ {
+ "epoch": 40.80108991825613,
+ "grad_norm": 3.622248649597168,
+ "learning_rate": 1.3396187435170144e-05,
+ "loss": 0.1575,
+ "step": 14974
+ },
+ {
+ "epoch": 40.80381471389646,
+ "grad_norm": 18.120290756225586,
+ "learning_rate": 1.3395357379979324e-05,
+ "loss": 0.1142,
+ "step": 14975
+ },
+ {
+ "epoch": 40.80653950953678,
+ "grad_norm": 5.425030708312988,
+ "learning_rate": 1.3394527298345567e-05,
+ "loss": 0.1193,
+ "step": 14976
+ },
+ {
+ "epoch": 40.80926430517711,
+ "grad_norm": 106.76289367675781,
+ "learning_rate": 1.339369719027534e-05,
+ "loss": 0.1089,
+ "step": 14977
+ },
+ {
+ "epoch": 40.81198910081744,
+ "grad_norm": 3.086984872817993,
+ "learning_rate": 1.3392867055775112e-05,
+ "loss": 0.0608,
+ "step": 14978
+ },
+ {
+ "epoch": 40.81471389645777,
+ "grad_norm": 3.4470722675323486,
+ "learning_rate": 1.3392036894851342e-05,
+ "loss": 0.1267,
+ "step": 14979
+ },
+ {
+ "epoch": 40.817438692098094,
+ "grad_norm": 3.843595266342163,
+ "learning_rate": 1.3391206707510498e-05,
+ "loss": 0.0807,
+ "step": 14980
+ },
+ {
+ "epoch": 40.82016348773842,
+ "grad_norm": 4.080315113067627,
+ "learning_rate": 1.3390376493759045e-05,
+ "loss": 0.156,
+ "step": 14981
+ },
+ {
+ "epoch": 40.822888283378745,
+ "grad_norm": 4.168814659118652,
+ "learning_rate": 1.3389546253603447e-05,
+ "loss": 0.1927,
+ "step": 14982
+ },
+ {
+ "epoch": 40.82561307901907,
+ "grad_norm": 5.623293876647949,
+ "learning_rate": 1.3388715987050174e-05,
+ "loss": 0.2161,
+ "step": 14983
+ },
+ {
+ "epoch": 40.828337874659404,
+ "grad_norm": 5.746096134185791,
+ "learning_rate": 1.338788569410569e-05,
+ "loss": 0.213,
+ "step": 14984
+ },
+ {
+ "epoch": 40.83106267029973,
+ "grad_norm": 6.889951705932617,
+ "learning_rate": 1.3387055374776463e-05,
+ "loss": 0.2731,
+ "step": 14985
+ },
+ {
+ "epoch": 40.833787465940055,
+ "grad_norm": 10.468439102172852,
+ "learning_rate": 1.3386225029068959e-05,
+ "loss": 0.1597,
+ "step": 14986
+ },
+ {
+ "epoch": 40.83651226158038,
+ "grad_norm": 5.377793788909912,
+ "learning_rate": 1.338539465698964e-05,
+ "loss": 0.1699,
+ "step": 14987
+ },
+ {
+ "epoch": 40.83923705722071,
+ "grad_norm": 7.506341457366943,
+ "learning_rate": 1.3384564258544976e-05,
+ "loss": 0.1909,
+ "step": 14988
+ },
+ {
+ "epoch": 40.84196185286103,
+ "grad_norm": 8.193270683288574,
+ "learning_rate": 1.3383733833741434e-05,
+ "loss": 0.189,
+ "step": 14989
+ },
+ {
+ "epoch": 40.844686648501366,
+ "grad_norm": 7.927051067352295,
+ "learning_rate": 1.3382903382585485e-05,
+ "loss": 0.2012,
+ "step": 14990
+ },
+ {
+ "epoch": 40.84741144414169,
+ "grad_norm": 5.357126712799072,
+ "learning_rate": 1.338207290508359e-05,
+ "loss": 0.1258,
+ "step": 14991
+ },
+ {
+ "epoch": 40.85013623978202,
+ "grad_norm": 8.291776657104492,
+ "learning_rate": 1.3381242401242222e-05,
+ "loss": 0.2003,
+ "step": 14992
+ },
+ {
+ "epoch": 40.85286103542234,
+ "grad_norm": 5.09745979309082,
+ "learning_rate": 1.3380411871067846e-05,
+ "loss": 0.3483,
+ "step": 14993
+ },
+ {
+ "epoch": 40.85558583106267,
+ "grad_norm": 8.878171920776367,
+ "learning_rate": 1.3379581314566931e-05,
+ "loss": 0.2278,
+ "step": 14994
+ },
+ {
+ "epoch": 40.858310626702995,
+ "grad_norm": 4.9039130210876465,
+ "learning_rate": 1.3378750731745949e-05,
+ "loss": 0.2437,
+ "step": 14995
+ },
+ {
+ "epoch": 40.86103542234333,
+ "grad_norm": 6.453513145446777,
+ "learning_rate": 1.337792012261136e-05,
+ "loss": 0.3845,
+ "step": 14996
+ },
+ {
+ "epoch": 40.86376021798365,
+ "grad_norm": 4.538636207580566,
+ "learning_rate": 1.337708948716964e-05,
+ "loss": 0.1674,
+ "step": 14997
+ },
+ {
+ "epoch": 40.86648501362398,
+ "grad_norm": 4.516594886779785,
+ "learning_rate": 1.3376258825427259e-05,
+ "loss": 0.2434,
+ "step": 14998
+ },
+ {
+ "epoch": 40.869209809264305,
+ "grad_norm": 5.99159049987793,
+ "learning_rate": 1.3375428137390678e-05,
+ "loss": 0.1838,
+ "step": 14999
+ },
+ {
+ "epoch": 40.87193460490463,
+ "grad_norm": 4.930017948150635,
+ "learning_rate": 1.3374597423066373e-05,
+ "loss": 0.218,
+ "step": 15000
+ },
+ {
+ "epoch": 40.87465940054496,
+ "grad_norm": 5.056175231933594,
+ "learning_rate": 1.3373766682460811e-05,
+ "loss": 0.1403,
+ "step": 15001
+ },
+ {
+ "epoch": 40.87738419618529,
+ "grad_norm": 5.388237953186035,
+ "learning_rate": 1.3372935915580465e-05,
+ "loss": 0.1776,
+ "step": 15002
+ },
+ {
+ "epoch": 40.880108991825615,
+ "grad_norm": 5.196503162384033,
+ "learning_rate": 1.3372105122431801e-05,
+ "loss": 0.2115,
+ "step": 15003
+ },
+ {
+ "epoch": 40.88283378746594,
+ "grad_norm": 4.4470720291137695,
+ "learning_rate": 1.3371274303021293e-05,
+ "loss": 0.1126,
+ "step": 15004
+ },
+ {
+ "epoch": 40.88555858310627,
+ "grad_norm": 5.426560401916504,
+ "learning_rate": 1.3370443457355407e-05,
+ "loss": 0.1507,
+ "step": 15005
+ },
+ {
+ "epoch": 40.88828337874659,
+ "grad_norm": 8.439411163330078,
+ "learning_rate": 1.3369612585440617e-05,
+ "loss": 0.1358,
+ "step": 15006
+ },
+ {
+ "epoch": 40.89100817438692,
+ "grad_norm": 4.914752006530762,
+ "learning_rate": 1.336878168728339e-05,
+ "loss": 0.1183,
+ "step": 15007
+ },
+ {
+ "epoch": 40.89373297002725,
+ "grad_norm": 4.43251371383667,
+ "learning_rate": 1.33679507628902e-05,
+ "loss": 0.1335,
+ "step": 15008
+ },
+ {
+ "epoch": 40.89645776566758,
+ "grad_norm": 4.9748215675354,
+ "learning_rate": 1.336711981226752e-05,
+ "loss": 0.1315,
+ "step": 15009
+ },
+ {
+ "epoch": 40.8991825613079,
+ "grad_norm": 6.777787685394287,
+ "learning_rate": 1.336628883542182e-05,
+ "loss": 0.1747,
+ "step": 15010
+ },
+ {
+ "epoch": 40.90190735694823,
+ "grad_norm": 3.705249309539795,
+ "learning_rate": 1.3365457832359567e-05,
+ "loss": 0.1096,
+ "step": 15011
+ },
+ {
+ "epoch": 40.904632152588555,
+ "grad_norm": 5.144952297210693,
+ "learning_rate": 1.3364626803087242e-05,
+ "loss": 0.231,
+ "step": 15012
+ },
+ {
+ "epoch": 40.90735694822888,
+ "grad_norm": 5.11735725402832,
+ "learning_rate": 1.3363795747611309e-05,
+ "loss": 0.3406,
+ "step": 15013
+ },
+ {
+ "epoch": 40.91008174386921,
+ "grad_norm": 4.974647045135498,
+ "learning_rate": 1.3362964665938246e-05,
+ "loss": 0.2015,
+ "step": 15014
+ },
+ {
+ "epoch": 40.91280653950954,
+ "grad_norm": 4.355781078338623,
+ "learning_rate": 1.3362133558074517e-05,
+ "loss": 0.1143,
+ "step": 15015
+ },
+ {
+ "epoch": 40.915531335149865,
+ "grad_norm": 4.873548984527588,
+ "learning_rate": 1.3361302424026603e-05,
+ "loss": 0.2118,
+ "step": 15016
+ },
+ {
+ "epoch": 40.91825613079019,
+ "grad_norm": 3.717289686203003,
+ "learning_rate": 1.3360471263800974e-05,
+ "loss": 0.1311,
+ "step": 15017
+ },
+ {
+ "epoch": 40.920980926430516,
+ "grad_norm": 8.181449890136719,
+ "learning_rate": 1.33596400774041e-05,
+ "loss": 0.2385,
+ "step": 15018
+ },
+ {
+ "epoch": 40.92370572207084,
+ "grad_norm": 3.547685146331787,
+ "learning_rate": 1.335880886484246e-05,
+ "loss": 0.1887,
+ "step": 15019
+ },
+ {
+ "epoch": 40.926430517711175,
+ "grad_norm": 4.2000412940979,
+ "learning_rate": 1.3357977626122525e-05,
+ "loss": 0.239,
+ "step": 15020
+ },
+ {
+ "epoch": 40.9291553133515,
+ "grad_norm": 4.941713809967041,
+ "learning_rate": 1.3357146361250765e-05,
+ "loss": 0.1957,
+ "step": 15021
+ },
+ {
+ "epoch": 40.93188010899183,
+ "grad_norm": 4.328118324279785,
+ "learning_rate": 1.3356315070233664e-05,
+ "loss": 0.1695,
+ "step": 15022
+ },
+ {
+ "epoch": 40.93460490463215,
+ "grad_norm": 5.107132434844971,
+ "learning_rate": 1.3355483753077682e-05,
+ "loss": 0.173,
+ "step": 15023
+ },
+ {
+ "epoch": 40.93732970027248,
+ "grad_norm": 5.059732913970947,
+ "learning_rate": 1.3354652409789303e-05,
+ "loss": 0.1257,
+ "step": 15024
+ },
+ {
+ "epoch": 40.940054495912804,
+ "grad_norm": 3.9745893478393555,
+ "learning_rate": 1.3353821040375001e-05,
+ "loss": 0.2233,
+ "step": 15025
+ },
+ {
+ "epoch": 40.94277929155314,
+ "grad_norm": 4.2427659034729,
+ "learning_rate": 1.3352989644841245e-05,
+ "loss": 0.202,
+ "step": 15026
+ },
+ {
+ "epoch": 40.94550408719346,
+ "grad_norm": 4.358948707580566,
+ "learning_rate": 1.3352158223194517e-05,
+ "loss": 0.3214,
+ "step": 15027
+ },
+ {
+ "epoch": 40.94822888283379,
+ "grad_norm": 5.485036849975586,
+ "learning_rate": 1.3351326775441287e-05,
+ "loss": 0.1991,
+ "step": 15028
+ },
+ {
+ "epoch": 40.950953678474114,
+ "grad_norm": 3.77752685546875,
+ "learning_rate": 1.3350495301588034e-05,
+ "loss": 0.1347,
+ "step": 15029
+ },
+ {
+ "epoch": 40.95367847411444,
+ "grad_norm": 5.442925930023193,
+ "learning_rate": 1.334966380164123e-05,
+ "loss": 0.1604,
+ "step": 15030
+ },
+ {
+ "epoch": 40.956403269754766,
+ "grad_norm": 4.632906913757324,
+ "learning_rate": 1.3348832275607351e-05,
+ "loss": 0.2991,
+ "step": 15031
+ },
+ {
+ "epoch": 40.95912806539509,
+ "grad_norm": 3.547370433807373,
+ "learning_rate": 1.3348000723492875e-05,
+ "loss": 0.1302,
+ "step": 15032
+ },
+ {
+ "epoch": 40.961852861035425,
+ "grad_norm": 4.727902889251709,
+ "learning_rate": 1.3347169145304277e-05,
+ "loss": 0.1079,
+ "step": 15033
+ },
+ {
+ "epoch": 40.96457765667575,
+ "grad_norm": 4.75783109664917,
+ "learning_rate": 1.3346337541048034e-05,
+ "loss": 0.1633,
+ "step": 15034
+ },
+ {
+ "epoch": 40.967302452316076,
+ "grad_norm": 3.4246315956115723,
+ "learning_rate": 1.3345505910730621e-05,
+ "loss": 0.237,
+ "step": 15035
+ },
+ {
+ "epoch": 40.9700272479564,
+ "grad_norm": 4.292584419250488,
+ "learning_rate": 1.3344674254358515e-05,
+ "loss": 0.1374,
+ "step": 15036
+ },
+ {
+ "epoch": 40.97275204359673,
+ "grad_norm": 4.395107269287109,
+ "learning_rate": 1.3343842571938197e-05,
+ "loss": 0.129,
+ "step": 15037
+ },
+ {
+ "epoch": 40.97547683923706,
+ "grad_norm": 3.1712393760681152,
+ "learning_rate": 1.3343010863476135e-05,
+ "loss": 0.2849,
+ "step": 15038
+ },
+ {
+ "epoch": 40.97820163487739,
+ "grad_norm": 4.7664008140563965,
+ "learning_rate": 1.3342179128978818e-05,
+ "loss": 0.1951,
+ "step": 15039
+ },
+ {
+ "epoch": 40.98092643051771,
+ "grad_norm": 4.312863349914551,
+ "learning_rate": 1.3341347368452712e-05,
+ "loss": 0.1529,
+ "step": 15040
+ },
+ {
+ "epoch": 40.98365122615804,
+ "grad_norm": 3.5997707843780518,
+ "learning_rate": 1.3340515581904306e-05,
+ "loss": 0.1287,
+ "step": 15041
+ },
+ {
+ "epoch": 40.986376021798364,
+ "grad_norm": 5.822768688201904,
+ "learning_rate": 1.3339683769340069e-05,
+ "loss": 0.0887,
+ "step": 15042
+ },
+ {
+ "epoch": 40.98910081743869,
+ "grad_norm": 4.255836009979248,
+ "learning_rate": 1.333885193076648e-05,
+ "loss": 0.1377,
+ "step": 15043
+ },
+ {
+ "epoch": 40.991825613079016,
+ "grad_norm": 3.9941649436950684,
+ "learning_rate": 1.3338020066190022e-05,
+ "loss": 0.2385,
+ "step": 15044
+ },
+ {
+ "epoch": 40.99455040871935,
+ "grad_norm": 4.23140811920166,
+ "learning_rate": 1.3337188175617173e-05,
+ "loss": 0.2453,
+ "step": 15045
+ },
+ {
+ "epoch": 40.997275204359674,
+ "grad_norm": 5.094249248504639,
+ "learning_rate": 1.3336356259054406e-05,
+ "loss": 0.3707,
+ "step": 15046
+ },
+ {
+ "epoch": 41.0,
+ "grad_norm": 4.141048431396484,
+ "learning_rate": 1.3335524316508208e-05,
+ "loss": 0.1496,
+ "step": 15047
+ },
+ {
+ "epoch": 41.002724795640326,
+ "grad_norm": 4.063958168029785,
+ "learning_rate": 1.3334692347985053e-05,
+ "loss": 0.1138,
+ "step": 15048
+ },
+ {
+ "epoch": 41.00544959128065,
+ "grad_norm": 3.8987278938293457,
+ "learning_rate": 1.333386035349142e-05,
+ "loss": 0.2199,
+ "step": 15049
+ },
+ {
+ "epoch": 41.00817438692098,
+ "grad_norm": 3.274005889892578,
+ "learning_rate": 1.3333028333033791e-05,
+ "loss": 0.1597,
+ "step": 15050
+ },
+ {
+ "epoch": 41.01089918256131,
+ "grad_norm": 4.153541088104248,
+ "learning_rate": 1.3332196286618641e-05,
+ "loss": 0.0967,
+ "step": 15051
+ },
+ {
+ "epoch": 41.013623978201636,
+ "grad_norm": 6.2700347900390625,
+ "learning_rate": 1.3331364214252456e-05,
+ "loss": 0.0936,
+ "step": 15052
+ },
+ {
+ "epoch": 41.01634877384196,
+ "grad_norm": 4.608465194702148,
+ "learning_rate": 1.3330532115941714e-05,
+ "loss": 0.089,
+ "step": 15053
+ },
+ {
+ "epoch": 41.01907356948229,
+ "grad_norm": 3.665262222290039,
+ "learning_rate": 1.3329699991692894e-05,
+ "loss": 0.2076,
+ "step": 15054
+ },
+ {
+ "epoch": 41.02179836512261,
+ "grad_norm": 3.7877440452575684,
+ "learning_rate": 1.3328867841512481e-05,
+ "loss": 0.1057,
+ "step": 15055
+ },
+ {
+ "epoch": 41.02452316076294,
+ "grad_norm": 3.8143110275268555,
+ "learning_rate": 1.3328035665406948e-05,
+ "loss": 0.0904,
+ "step": 15056
+ },
+ {
+ "epoch": 41.02724795640327,
+ "grad_norm": 4.035426616668701,
+ "learning_rate": 1.3327203463382786e-05,
+ "loss": 0.1882,
+ "step": 15057
+ },
+ {
+ "epoch": 41.0299727520436,
+ "grad_norm": 4.099752902984619,
+ "learning_rate": 1.3326371235446464e-05,
+ "loss": 0.1245,
+ "step": 15058
+ },
+ {
+ "epoch": 41.032697547683924,
+ "grad_norm": 5.499074935913086,
+ "learning_rate": 1.3325538981604475e-05,
+ "loss": 0.1233,
+ "step": 15059
+ },
+ {
+ "epoch": 41.03542234332425,
+ "grad_norm": 4.531176567077637,
+ "learning_rate": 1.332470670186329e-05,
+ "loss": 0.1634,
+ "step": 15060
+ },
+ {
+ "epoch": 41.038147138964575,
+ "grad_norm": 4.0297465324401855,
+ "learning_rate": 1.3323874396229402e-05,
+ "loss": 0.1097,
+ "step": 15061
+ },
+ {
+ "epoch": 41.0408719346049,
+ "grad_norm": 20.410839080810547,
+ "learning_rate": 1.3323042064709285e-05,
+ "loss": 0.2238,
+ "step": 15062
+ },
+ {
+ "epoch": 41.043596730245234,
+ "grad_norm": 3.154783010482788,
+ "learning_rate": 1.3322209707309424e-05,
+ "loss": 0.1589,
+ "step": 15063
+ },
+ {
+ "epoch": 41.04632152588556,
+ "grad_norm": 3.5390405654907227,
+ "learning_rate": 1.3321377324036297e-05,
+ "loss": 0.1513,
+ "step": 15064
+ },
+ {
+ "epoch": 41.049046321525886,
+ "grad_norm": 3.91922664642334,
+ "learning_rate": 1.3320544914896396e-05,
+ "loss": 0.0896,
+ "step": 15065
+ },
+ {
+ "epoch": 41.05177111716621,
+ "grad_norm": 3.2090299129486084,
+ "learning_rate": 1.3319712479896195e-05,
+ "loss": 0.0852,
+ "step": 15066
+ },
+ {
+ "epoch": 41.05449591280654,
+ "grad_norm": 3.7687909603118896,
+ "learning_rate": 1.3318880019042179e-05,
+ "loss": 0.1264,
+ "step": 15067
+ },
+ {
+ "epoch": 41.05722070844686,
+ "grad_norm": 3.816025495529175,
+ "learning_rate": 1.3318047532340833e-05,
+ "loss": 0.0968,
+ "step": 15068
+ },
+ {
+ "epoch": 41.059945504087196,
+ "grad_norm": 3.3944900035858154,
+ "learning_rate": 1.3317215019798639e-05,
+ "loss": 0.1228,
+ "step": 15069
+ },
+ {
+ "epoch": 41.06267029972752,
+ "grad_norm": 3.318511962890625,
+ "learning_rate": 1.3316382481422081e-05,
+ "loss": 0.2237,
+ "step": 15070
+ },
+ {
+ "epoch": 41.06539509536785,
+ "grad_norm": 3.6442513465881348,
+ "learning_rate": 1.3315549917217647e-05,
+ "loss": 0.0719,
+ "step": 15071
+ },
+ {
+ "epoch": 41.06811989100817,
+ "grad_norm": 3.293546199798584,
+ "learning_rate": 1.3314717327191814e-05,
+ "loss": 0.1801,
+ "step": 15072
+ },
+ {
+ "epoch": 41.0708446866485,
+ "grad_norm": 3.2978134155273438,
+ "learning_rate": 1.331388471135107e-05,
+ "loss": 0.2628,
+ "step": 15073
+ },
+ {
+ "epoch": 41.073569482288825,
+ "grad_norm": 4.108832836151123,
+ "learning_rate": 1.3313052069701896e-05,
+ "loss": 0.1271,
+ "step": 15074
+ },
+ {
+ "epoch": 41.07629427792916,
+ "grad_norm": 3.8666961193084717,
+ "learning_rate": 1.3312219402250781e-05,
+ "loss": 0.1132,
+ "step": 15075
+ },
+ {
+ "epoch": 41.079019073569484,
+ "grad_norm": 3.7872633934020996,
+ "learning_rate": 1.3311386709004208e-05,
+ "loss": 0.1643,
+ "step": 15076
+ },
+ {
+ "epoch": 41.08174386920981,
+ "grad_norm": 3.7094812393188477,
+ "learning_rate": 1.3310553989968662e-05,
+ "loss": 0.1424,
+ "step": 15077
+ },
+ {
+ "epoch": 41.084468664850135,
+ "grad_norm": 3.8079404830932617,
+ "learning_rate": 1.3309721245150627e-05,
+ "loss": 0.1306,
+ "step": 15078
+ },
+ {
+ "epoch": 41.08719346049046,
+ "grad_norm": 6.231881618499756,
+ "learning_rate": 1.3308888474556591e-05,
+ "loss": 0.1007,
+ "step": 15079
+ },
+ {
+ "epoch": 41.08991825613079,
+ "grad_norm": 6.169212818145752,
+ "learning_rate": 1.3308055678193037e-05,
+ "loss": 0.1682,
+ "step": 15080
+ },
+ {
+ "epoch": 41.09264305177112,
+ "grad_norm": 3.5099823474884033,
+ "learning_rate": 1.3307222856066452e-05,
+ "loss": 0.0922,
+ "step": 15081
+ },
+ {
+ "epoch": 41.095367847411445,
+ "grad_norm": 4.338865280151367,
+ "learning_rate": 1.3306390008183324e-05,
+ "loss": 0.2286,
+ "step": 15082
+ },
+ {
+ "epoch": 41.09809264305177,
+ "grad_norm": 3.6146249771118164,
+ "learning_rate": 1.3305557134550133e-05,
+ "loss": 0.1881,
+ "step": 15083
+ },
+ {
+ "epoch": 41.1008174386921,
+ "grad_norm": 3.4403491020202637,
+ "learning_rate": 1.3304724235173372e-05,
+ "loss": 0.1175,
+ "step": 15084
+ },
+ {
+ "epoch": 41.10354223433242,
+ "grad_norm": 3.5752806663513184,
+ "learning_rate": 1.3303891310059528e-05,
+ "loss": 0.1064,
+ "step": 15085
+ },
+ {
+ "epoch": 41.10626702997275,
+ "grad_norm": 3.3423097133636475,
+ "learning_rate": 1.3303058359215075e-05,
+ "loss": 0.1554,
+ "step": 15086
+ },
+ {
+ "epoch": 41.10899182561308,
+ "grad_norm": 3.5356979370117188,
+ "learning_rate": 1.3302225382646518e-05,
+ "loss": 0.0934,
+ "step": 15087
+ },
+ {
+ "epoch": 41.11171662125341,
+ "grad_norm": 3.843421220779419,
+ "learning_rate": 1.3301392380360334e-05,
+ "loss": 0.1344,
+ "step": 15088
+ },
+ {
+ "epoch": 41.11444141689373,
+ "grad_norm": 7.025874614715576,
+ "learning_rate": 1.3300559352363015e-05,
+ "loss": 0.1429,
+ "step": 15089
+ },
+ {
+ "epoch": 41.11716621253406,
+ "grad_norm": 4.2677202224731445,
+ "learning_rate": 1.3299726298661039e-05,
+ "loss": 0.1182,
+ "step": 15090
+ },
+ {
+ "epoch": 41.119891008174385,
+ "grad_norm": 4.152862071990967,
+ "learning_rate": 1.3298893219260909e-05,
+ "loss": 0.1366,
+ "step": 15091
+ },
+ {
+ "epoch": 41.12261580381471,
+ "grad_norm": 3.3635826110839844,
+ "learning_rate": 1.32980601141691e-05,
+ "loss": 0.1084,
+ "step": 15092
+ },
+ {
+ "epoch": 41.12534059945504,
+ "grad_norm": 4.933255672454834,
+ "learning_rate": 1.3297226983392106e-05,
+ "loss": 0.0969,
+ "step": 15093
+ },
+ {
+ "epoch": 41.12806539509537,
+ "grad_norm": 3.3002545833587646,
+ "learning_rate": 1.3296393826936416e-05,
+ "loss": 0.1544,
+ "step": 15094
+ },
+ {
+ "epoch": 41.130790190735695,
+ "grad_norm": 4.4337310791015625,
+ "learning_rate": 1.3295560644808511e-05,
+ "loss": 0.1332,
+ "step": 15095
+ },
+ {
+ "epoch": 41.13351498637602,
+ "grad_norm": 4.293772220611572,
+ "learning_rate": 1.3294727437014891e-05,
+ "loss": 0.1449,
+ "step": 15096
+ },
+ {
+ "epoch": 41.13623978201635,
+ "grad_norm": 4.079353332519531,
+ "learning_rate": 1.3293894203562041e-05,
+ "loss": 0.2572,
+ "step": 15097
+ },
+ {
+ "epoch": 41.13896457765667,
+ "grad_norm": 3.0863399505615234,
+ "learning_rate": 1.3293060944456446e-05,
+ "loss": 0.2431,
+ "step": 15098
+ },
+ {
+ "epoch": 41.141689373297005,
+ "grad_norm": 3.347914695739746,
+ "learning_rate": 1.32922276597046e-05,
+ "loss": 0.1026,
+ "step": 15099
+ },
+ {
+ "epoch": 41.14441416893733,
+ "grad_norm": 3.105147361755371,
+ "learning_rate": 1.3291394349312989e-05,
+ "loss": 0.0861,
+ "step": 15100
+ },
+ {
+ "epoch": 41.14713896457766,
+ "grad_norm": 3.1250219345092773,
+ "learning_rate": 1.3290561013288108e-05,
+ "loss": 0.1379,
+ "step": 15101
+ },
+ {
+ "epoch": 41.14986376021798,
+ "grad_norm": 3.2204556465148926,
+ "learning_rate": 1.3289727651636439e-05,
+ "loss": 0.0855,
+ "step": 15102
+ },
+ {
+ "epoch": 41.15258855585831,
+ "grad_norm": 3.0542612075805664,
+ "learning_rate": 1.328889426436448e-05,
+ "loss": 0.2303,
+ "step": 15103
+ },
+ {
+ "epoch": 41.155313351498634,
+ "grad_norm": 3.360044240951538,
+ "learning_rate": 1.328806085147872e-05,
+ "loss": 0.1084,
+ "step": 15104
+ },
+ {
+ "epoch": 41.15803814713897,
+ "grad_norm": 3.5166175365448,
+ "learning_rate": 1.3287227412985644e-05,
+ "loss": 0.1034,
+ "step": 15105
+ },
+ {
+ "epoch": 41.16076294277929,
+ "grad_norm": 3.623102903366089,
+ "learning_rate": 1.3286393948891749e-05,
+ "loss": 0.2188,
+ "step": 15106
+ },
+ {
+ "epoch": 41.16348773841962,
+ "grad_norm": 3.686460018157959,
+ "learning_rate": 1.3285560459203524e-05,
+ "loss": 0.0983,
+ "step": 15107
+ },
+ {
+ "epoch": 41.166212534059945,
+ "grad_norm": 3.478285551071167,
+ "learning_rate": 1.3284726943927458e-05,
+ "loss": 0.1175,
+ "step": 15108
+ },
+ {
+ "epoch": 41.16893732970027,
+ "grad_norm": 4.951805591583252,
+ "learning_rate": 1.3283893403070044e-05,
+ "loss": 0.1371,
+ "step": 15109
+ },
+ {
+ "epoch": 41.171662125340596,
+ "grad_norm": 3.349276304244995,
+ "learning_rate": 1.3283059836637775e-05,
+ "loss": 0.2218,
+ "step": 15110
+ },
+ {
+ "epoch": 41.17438692098093,
+ "grad_norm": 3.5513453483581543,
+ "learning_rate": 1.3282226244637141e-05,
+ "loss": 0.1332,
+ "step": 15111
+ },
+ {
+ "epoch": 41.177111716621255,
+ "grad_norm": 3.532456874847412,
+ "learning_rate": 1.3281392627074638e-05,
+ "loss": 0.0964,
+ "step": 15112
+ },
+ {
+ "epoch": 41.17983651226158,
+ "grad_norm": 4.759359836578369,
+ "learning_rate": 1.3280558983956753e-05,
+ "loss": 0.1562,
+ "step": 15113
+ },
+ {
+ "epoch": 41.182561307901906,
+ "grad_norm": 3.2467143535614014,
+ "learning_rate": 1.3279725315289979e-05,
+ "loss": 0.0774,
+ "step": 15114
+ },
+ {
+ "epoch": 41.18528610354223,
+ "grad_norm": 3.228471040725708,
+ "learning_rate": 1.327889162108081e-05,
+ "loss": 0.1894,
+ "step": 15115
+ },
+ {
+ "epoch": 41.18801089918256,
+ "grad_norm": 3.7846004962921143,
+ "learning_rate": 1.327805790133574e-05,
+ "loss": 0.1455,
+ "step": 15116
+ },
+ {
+ "epoch": 41.19073569482289,
+ "grad_norm": 4.632404804229736,
+ "learning_rate": 1.327722415606126e-05,
+ "loss": 0.0938,
+ "step": 15117
+ },
+ {
+ "epoch": 41.19346049046322,
+ "grad_norm": 2.724532127380371,
+ "learning_rate": 1.3276390385263862e-05,
+ "loss": 0.0828,
+ "step": 15118
+ },
+ {
+ "epoch": 41.19618528610354,
+ "grad_norm": 3.124376058578491,
+ "learning_rate": 1.3275556588950043e-05,
+ "loss": 0.2377,
+ "step": 15119
+ },
+ {
+ "epoch": 41.19891008174387,
+ "grad_norm": 2.9907352924346924,
+ "learning_rate": 1.3274722767126294e-05,
+ "loss": 0.0806,
+ "step": 15120
+ },
+ {
+ "epoch": 41.201634877384194,
+ "grad_norm": 4.026010513305664,
+ "learning_rate": 1.3273888919799109e-05,
+ "loss": 0.1667,
+ "step": 15121
+ },
+ {
+ "epoch": 41.20435967302452,
+ "grad_norm": 4.422990798950195,
+ "learning_rate": 1.3273055046974984e-05,
+ "loss": 0.1565,
+ "step": 15122
+ },
+ {
+ "epoch": 41.20708446866485,
+ "grad_norm": 4.278494358062744,
+ "learning_rate": 1.3272221148660409e-05,
+ "loss": 0.0964,
+ "step": 15123
+ },
+ {
+ "epoch": 41.20980926430518,
+ "grad_norm": 3.5143983364105225,
+ "learning_rate": 1.3271387224861885e-05,
+ "loss": 0.1996,
+ "step": 15124
+ },
+ {
+ "epoch": 41.212534059945504,
+ "grad_norm": 3.7627017498016357,
+ "learning_rate": 1.3270553275585897e-05,
+ "loss": 0.1396,
+ "step": 15125
+ },
+ {
+ "epoch": 41.21525885558583,
+ "grad_norm": 3.1896514892578125,
+ "learning_rate": 1.3269719300838952e-05,
+ "loss": 0.1107,
+ "step": 15126
+ },
+ {
+ "epoch": 41.217983651226156,
+ "grad_norm": 3.6305723190307617,
+ "learning_rate": 1.3268885300627534e-05,
+ "loss": 0.0972,
+ "step": 15127
+ },
+ {
+ "epoch": 41.22070844686648,
+ "grad_norm": 3.1414923667907715,
+ "learning_rate": 1.3268051274958145e-05,
+ "loss": 0.1614,
+ "step": 15128
+ },
+ {
+ "epoch": 41.223433242506815,
+ "grad_norm": 3.15079665184021,
+ "learning_rate": 1.3267217223837273e-05,
+ "loss": 0.1111,
+ "step": 15129
+ },
+ {
+ "epoch": 41.22615803814714,
+ "grad_norm": 5.272952556610107,
+ "learning_rate": 1.3266383147271422e-05,
+ "loss": 0.0911,
+ "step": 15130
+ },
+ {
+ "epoch": 41.228882833787466,
+ "grad_norm": 3.223296880722046,
+ "learning_rate": 1.3265549045267085e-05,
+ "loss": 0.102,
+ "step": 15131
+ },
+ {
+ "epoch": 41.23160762942779,
+ "grad_norm": 3.813524007797241,
+ "learning_rate": 1.3264714917830756e-05,
+ "loss": 0.1424,
+ "step": 15132
+ },
+ {
+ "epoch": 41.23433242506812,
+ "grad_norm": 3.443358898162842,
+ "learning_rate": 1.3263880764968933e-05,
+ "loss": 0.0998,
+ "step": 15133
+ },
+ {
+ "epoch": 41.237057220708444,
+ "grad_norm": 3.260328531265259,
+ "learning_rate": 1.326304658668811e-05,
+ "loss": 0.1205,
+ "step": 15134
+ },
+ {
+ "epoch": 41.23978201634878,
+ "grad_norm": 3.1876232624053955,
+ "learning_rate": 1.3262212382994786e-05,
+ "loss": 0.1094,
+ "step": 15135
+ },
+ {
+ "epoch": 41.2425068119891,
+ "grad_norm": 2.8319168090820312,
+ "learning_rate": 1.3261378153895459e-05,
+ "loss": 0.1227,
+ "step": 15136
+ },
+ {
+ "epoch": 41.24523160762943,
+ "grad_norm": 3.3156890869140625,
+ "learning_rate": 1.3260543899396618e-05,
+ "loss": 0.1046,
+ "step": 15137
+ },
+ {
+ "epoch": 41.247956403269754,
+ "grad_norm": 2.735217571258545,
+ "learning_rate": 1.3259709619504772e-05,
+ "loss": 0.1081,
+ "step": 15138
+ },
+ {
+ "epoch": 41.25068119891008,
+ "grad_norm": 3.476116180419922,
+ "learning_rate": 1.3258875314226409e-05,
+ "loss": 0.1708,
+ "step": 15139
+ },
+ {
+ "epoch": 41.253405994550405,
+ "grad_norm": 3.930833339691162,
+ "learning_rate": 1.3258040983568035e-05,
+ "loss": 0.2237,
+ "step": 15140
+ },
+ {
+ "epoch": 41.25613079019074,
+ "grad_norm": 4.663483142852783,
+ "learning_rate": 1.3257206627536137e-05,
+ "loss": 0.1679,
+ "step": 15141
+ },
+ {
+ "epoch": 41.258855585831064,
+ "grad_norm": 3.082674503326416,
+ "learning_rate": 1.3256372246137223e-05,
+ "loss": 0.0909,
+ "step": 15142
+ },
+ {
+ "epoch": 41.26158038147139,
+ "grad_norm": 2.3535845279693604,
+ "learning_rate": 1.3255537839377784e-05,
+ "loss": 0.0945,
+ "step": 15143
+ },
+ {
+ "epoch": 41.264305177111716,
+ "grad_norm": 8.917684555053711,
+ "learning_rate": 1.3254703407264322e-05,
+ "loss": 0.1936,
+ "step": 15144
+ },
+ {
+ "epoch": 41.26702997275204,
+ "grad_norm": 3.782207727432251,
+ "learning_rate": 1.3253868949803332e-05,
+ "loss": 0.0723,
+ "step": 15145
+ },
+ {
+ "epoch": 41.26975476839237,
+ "grad_norm": 2.6511588096618652,
+ "learning_rate": 1.3253034467001319e-05,
+ "loss": 0.1106,
+ "step": 15146
+ },
+ {
+ "epoch": 41.2724795640327,
+ "grad_norm": 4.227427005767822,
+ "learning_rate": 1.3252199958864775e-05,
+ "loss": 0.1658,
+ "step": 15147
+ },
+ {
+ "epoch": 41.275204359673026,
+ "grad_norm": 4.332622528076172,
+ "learning_rate": 1.3251365425400205e-05,
+ "loss": 0.1137,
+ "step": 15148
+ },
+ {
+ "epoch": 41.27792915531335,
+ "grad_norm": 3.501075029373169,
+ "learning_rate": 1.3250530866614104e-05,
+ "loss": 0.1777,
+ "step": 15149
+ },
+ {
+ "epoch": 41.28065395095368,
+ "grad_norm": 3.2031211853027344,
+ "learning_rate": 1.3249696282512976e-05,
+ "loss": 0.1065,
+ "step": 15150
+ },
+ {
+ "epoch": 41.283378746594,
+ "grad_norm": 3.43902325630188,
+ "learning_rate": 1.3248861673103315e-05,
+ "loss": 0.2068,
+ "step": 15151
+ },
+ {
+ "epoch": 41.28610354223433,
+ "grad_norm": 3.7883176803588867,
+ "learning_rate": 1.3248027038391626e-05,
+ "loss": 0.0872,
+ "step": 15152
+ },
+ {
+ "epoch": 41.28882833787466,
+ "grad_norm": 3.718719482421875,
+ "learning_rate": 1.3247192378384406e-05,
+ "loss": 0.1044,
+ "step": 15153
+ },
+ {
+ "epoch": 41.29155313351499,
+ "grad_norm": 3.294686794281006,
+ "learning_rate": 1.3246357693088155e-05,
+ "loss": 0.1613,
+ "step": 15154
+ },
+ {
+ "epoch": 41.294277929155314,
+ "grad_norm": 4.672397613525391,
+ "learning_rate": 1.3245522982509376e-05,
+ "loss": 0.1474,
+ "step": 15155
+ },
+ {
+ "epoch": 41.29700272479564,
+ "grad_norm": 2.7552268505096436,
+ "learning_rate": 1.3244688246654569e-05,
+ "loss": 0.1032,
+ "step": 15156
+ },
+ {
+ "epoch": 41.299727520435965,
+ "grad_norm": 2.6076462268829346,
+ "learning_rate": 1.3243853485530231e-05,
+ "loss": 0.3787,
+ "step": 15157
+ },
+ {
+ "epoch": 41.30245231607629,
+ "grad_norm": 3.8080954551696777,
+ "learning_rate": 1.3243018699142871e-05,
+ "loss": 0.118,
+ "step": 15158
+ },
+ {
+ "epoch": 41.305177111716624,
+ "grad_norm": 3.6266376972198486,
+ "learning_rate": 1.3242183887498983e-05,
+ "loss": 0.1104,
+ "step": 15159
+ },
+ {
+ "epoch": 41.30790190735695,
+ "grad_norm": 2.556832790374756,
+ "learning_rate": 1.3241349050605074e-05,
+ "loss": 0.0683,
+ "step": 15160
+ },
+ {
+ "epoch": 41.310626702997276,
+ "grad_norm": 3.460153818130493,
+ "learning_rate": 1.324051418846764e-05,
+ "loss": 0.1992,
+ "step": 15161
+ },
+ {
+ "epoch": 41.3133514986376,
+ "grad_norm": 3.463552236557007,
+ "learning_rate": 1.323967930109319e-05,
+ "loss": 0.0779,
+ "step": 15162
+ },
+ {
+ "epoch": 41.31607629427793,
+ "grad_norm": 3.1097302436828613,
+ "learning_rate": 1.3238844388488215e-05,
+ "loss": 0.1591,
+ "step": 15163
+ },
+ {
+ "epoch": 41.31880108991825,
+ "grad_norm": 2.946180820465088,
+ "learning_rate": 1.3238009450659228e-05,
+ "loss": 0.2249,
+ "step": 15164
+ },
+ {
+ "epoch": 41.321525885558586,
+ "grad_norm": 4.265206336975098,
+ "learning_rate": 1.3237174487612727e-05,
+ "loss": 0.102,
+ "step": 15165
+ },
+ {
+ "epoch": 41.32425068119891,
+ "grad_norm": 4.16032075881958,
+ "learning_rate": 1.3236339499355217e-05,
+ "loss": 0.0929,
+ "step": 15166
+ },
+ {
+ "epoch": 41.32697547683924,
+ "grad_norm": 4.009609222412109,
+ "learning_rate": 1.3235504485893198e-05,
+ "loss": 0.0815,
+ "step": 15167
+ },
+ {
+ "epoch": 41.32970027247956,
+ "grad_norm": 3.2409942150115967,
+ "learning_rate": 1.3234669447233175e-05,
+ "loss": 0.1305,
+ "step": 15168
+ },
+ {
+ "epoch": 41.33242506811989,
+ "grad_norm": 3.6888105869293213,
+ "learning_rate": 1.323383438338165e-05,
+ "loss": 0.1605,
+ "step": 15169
+ },
+ {
+ "epoch": 41.335149863760215,
+ "grad_norm": 3.6400034427642822,
+ "learning_rate": 1.3232999294345126e-05,
+ "loss": 0.182,
+ "step": 15170
+ },
+ {
+ "epoch": 41.33787465940055,
+ "grad_norm": 3.6942403316497803,
+ "learning_rate": 1.3232164180130108e-05,
+ "loss": 0.2848,
+ "step": 15171
+ },
+ {
+ "epoch": 41.34059945504087,
+ "grad_norm": 3.5814297199249268,
+ "learning_rate": 1.3231329040743099e-05,
+ "loss": 0.1921,
+ "step": 15172
+ },
+ {
+ "epoch": 41.3433242506812,
+ "grad_norm": 3.143367290496826,
+ "learning_rate": 1.3230493876190602e-05,
+ "loss": 0.0899,
+ "step": 15173
+ },
+ {
+ "epoch": 41.346049046321525,
+ "grad_norm": 3.104790449142456,
+ "learning_rate": 1.3229658686479128e-05,
+ "loss": 0.0997,
+ "step": 15174
+ },
+ {
+ "epoch": 41.34877384196185,
+ "grad_norm": 2.941096067428589,
+ "learning_rate": 1.322882347161517e-05,
+ "loss": 0.0815,
+ "step": 15175
+ },
+ {
+ "epoch": 41.35149863760218,
+ "grad_norm": 3.514127492904663,
+ "learning_rate": 1.3227988231605242e-05,
+ "loss": 0.1976,
+ "step": 15176
+ },
+ {
+ "epoch": 41.35422343324251,
+ "grad_norm": 4.303969860076904,
+ "learning_rate": 1.3227152966455844e-05,
+ "loss": 0.2214,
+ "step": 15177
+ },
+ {
+ "epoch": 41.356948228882835,
+ "grad_norm": 3.5096302032470703,
+ "learning_rate": 1.3226317676173485e-05,
+ "loss": 0.1827,
+ "step": 15178
+ },
+ {
+ "epoch": 41.35967302452316,
+ "grad_norm": 4.073300361633301,
+ "learning_rate": 1.3225482360764666e-05,
+ "loss": 0.1974,
+ "step": 15179
+ },
+ {
+ "epoch": 41.36239782016349,
+ "grad_norm": 3.2148098945617676,
+ "learning_rate": 1.3224647020235894e-05,
+ "loss": 0.1678,
+ "step": 15180
+ },
+ {
+ "epoch": 41.36512261580381,
+ "grad_norm": 2.721928358078003,
+ "learning_rate": 1.3223811654593677e-05,
+ "loss": 0.0997,
+ "step": 15181
+ },
+ {
+ "epoch": 41.36784741144414,
+ "grad_norm": 2.8931784629821777,
+ "learning_rate": 1.3222976263844517e-05,
+ "loss": 0.1306,
+ "step": 15182
+ },
+ {
+ "epoch": 41.37057220708447,
+ "grad_norm": 3.097466468811035,
+ "learning_rate": 1.3222140847994917e-05,
+ "loss": 0.1096,
+ "step": 15183
+ },
+ {
+ "epoch": 41.3732970027248,
+ "grad_norm": 3.585115671157837,
+ "learning_rate": 1.3221305407051395e-05,
+ "loss": 0.1947,
+ "step": 15184
+ },
+ {
+ "epoch": 41.37602179836512,
+ "grad_norm": 3.216701030731201,
+ "learning_rate": 1.3220469941020447e-05,
+ "loss": 0.205,
+ "step": 15185
+ },
+ {
+ "epoch": 41.37874659400545,
+ "grad_norm": 3.161019802093506,
+ "learning_rate": 1.3219634449908585e-05,
+ "loss": 0.2892,
+ "step": 15186
+ },
+ {
+ "epoch": 41.381471389645775,
+ "grad_norm": 3.258064031600952,
+ "learning_rate": 1.321879893372231e-05,
+ "loss": 0.1332,
+ "step": 15187
+ },
+ {
+ "epoch": 41.3841961852861,
+ "grad_norm": 3.684506416320801,
+ "learning_rate": 1.3217963392468135e-05,
+ "loss": 0.1343,
+ "step": 15188
+ },
+ {
+ "epoch": 41.38692098092643,
+ "grad_norm": 3.160670518875122,
+ "learning_rate": 1.3217127826152563e-05,
+ "loss": 0.1025,
+ "step": 15189
+ },
+ {
+ "epoch": 41.38964577656676,
+ "grad_norm": 2.7180099487304688,
+ "learning_rate": 1.3216292234782104e-05,
+ "loss": 0.1089,
+ "step": 15190
+ },
+ {
+ "epoch": 41.392370572207085,
+ "grad_norm": 2.970989465713501,
+ "learning_rate": 1.3215456618363264e-05,
+ "loss": 0.1656,
+ "step": 15191
+ },
+ {
+ "epoch": 41.39509536784741,
+ "grad_norm": 3.02868914604187,
+ "learning_rate": 1.3214620976902553e-05,
+ "loss": 0.078,
+ "step": 15192
+ },
+ {
+ "epoch": 41.39782016348774,
+ "grad_norm": 4.10238790512085,
+ "learning_rate": 1.3213785310406477e-05,
+ "loss": 0.1671,
+ "step": 15193
+ },
+ {
+ "epoch": 41.40054495912806,
+ "grad_norm": 2.6613996028900146,
+ "learning_rate": 1.321294961888154e-05,
+ "loss": 0.1139,
+ "step": 15194
+ },
+ {
+ "epoch": 41.403269754768395,
+ "grad_norm": 4.108104705810547,
+ "learning_rate": 1.321211390233426e-05,
+ "loss": 0.081,
+ "step": 15195
+ },
+ {
+ "epoch": 41.40599455040872,
+ "grad_norm": 3.1168487071990967,
+ "learning_rate": 1.321127816077114e-05,
+ "loss": 0.0993,
+ "step": 15196
+ },
+ {
+ "epoch": 41.40871934604905,
+ "grad_norm": 2.5514729022979736,
+ "learning_rate": 1.3210442394198686e-05,
+ "loss": 0.0629,
+ "step": 15197
+ },
+ {
+ "epoch": 41.41144414168937,
+ "grad_norm": 3.1951191425323486,
+ "learning_rate": 1.3209606602623411e-05,
+ "loss": 0.0953,
+ "step": 15198
+ },
+ {
+ "epoch": 41.4141689373297,
+ "grad_norm": 3.654484272003174,
+ "learning_rate": 1.3208770786051826e-05,
+ "loss": 0.131,
+ "step": 15199
+ },
+ {
+ "epoch": 41.416893732970024,
+ "grad_norm": 3.6561899185180664,
+ "learning_rate": 1.3207934944490433e-05,
+ "loss": 0.0868,
+ "step": 15200
+ },
+ {
+ "epoch": 41.41961852861036,
+ "grad_norm": 3.433523416519165,
+ "learning_rate": 1.3207099077945749e-05,
+ "loss": 0.1389,
+ "step": 15201
+ },
+ {
+ "epoch": 41.42234332425068,
+ "grad_norm": 2.5659024715423584,
+ "learning_rate": 1.3206263186424279e-05,
+ "loss": 0.1547,
+ "step": 15202
+ },
+ {
+ "epoch": 41.42506811989101,
+ "grad_norm": 3.881751775741577,
+ "learning_rate": 1.3205427269932535e-05,
+ "loss": 0.1996,
+ "step": 15203
+ },
+ {
+ "epoch": 41.427792915531334,
+ "grad_norm": 3.4369113445281982,
+ "learning_rate": 1.3204591328477028e-05,
+ "loss": 0.2233,
+ "step": 15204
+ },
+ {
+ "epoch": 41.43051771117166,
+ "grad_norm": 2.7599220275878906,
+ "learning_rate": 1.3203755362064263e-05,
+ "loss": 0.0899,
+ "step": 15205
+ },
+ {
+ "epoch": 41.433242506811986,
+ "grad_norm": 3.0202908515930176,
+ "learning_rate": 1.3202919370700758e-05,
+ "loss": 0.1172,
+ "step": 15206
+ },
+ {
+ "epoch": 41.43596730245232,
+ "grad_norm": 3.617220878601074,
+ "learning_rate": 1.3202083354393019e-05,
+ "loss": 0.0994,
+ "step": 15207
+ },
+ {
+ "epoch": 41.438692098092645,
+ "grad_norm": 3.743570327758789,
+ "learning_rate": 1.3201247313147559e-05,
+ "loss": 0.1374,
+ "step": 15208
+ },
+ {
+ "epoch": 41.44141689373297,
+ "grad_norm": 3.700831413269043,
+ "learning_rate": 1.3200411246970885e-05,
+ "loss": 0.1212,
+ "step": 15209
+ },
+ {
+ "epoch": 41.444141689373296,
+ "grad_norm": 2.9375882148742676,
+ "learning_rate": 1.3199575155869514e-05,
+ "loss": 0.0761,
+ "step": 15210
+ },
+ {
+ "epoch": 41.44686648501362,
+ "grad_norm": 3.6847100257873535,
+ "learning_rate": 1.3198739039849955e-05,
+ "loss": 0.1948,
+ "step": 15211
+ },
+ {
+ "epoch": 41.44959128065395,
+ "grad_norm": 3.699958562850952,
+ "learning_rate": 1.3197902898918718e-05,
+ "loss": 0.3178,
+ "step": 15212
+ },
+ {
+ "epoch": 41.45231607629428,
+ "grad_norm": 2.9728989601135254,
+ "learning_rate": 1.3197066733082316e-05,
+ "loss": 0.1012,
+ "step": 15213
+ },
+ {
+ "epoch": 41.45504087193461,
+ "grad_norm": 3.4643912315368652,
+ "learning_rate": 1.3196230542347259e-05,
+ "loss": 0.0859,
+ "step": 15214
+ },
+ {
+ "epoch": 41.45776566757493,
+ "grad_norm": 2.65470290184021,
+ "learning_rate": 1.3195394326720063e-05,
+ "loss": 0.1232,
+ "step": 15215
+ },
+ {
+ "epoch": 41.46049046321526,
+ "grad_norm": 2.5464346408843994,
+ "learning_rate": 1.3194558086207238e-05,
+ "loss": 0.2918,
+ "step": 15216
+ },
+ {
+ "epoch": 41.463215258855584,
+ "grad_norm": 3.0876612663269043,
+ "learning_rate": 1.3193721820815302e-05,
+ "loss": 0.1265,
+ "step": 15217
+ },
+ {
+ "epoch": 41.46594005449591,
+ "grad_norm": 3.3773205280303955,
+ "learning_rate": 1.3192885530550758e-05,
+ "loss": 0.3589,
+ "step": 15218
+ },
+ {
+ "epoch": 41.46866485013624,
+ "grad_norm": 3.734553098678589,
+ "learning_rate": 1.3192049215420129e-05,
+ "loss": 0.186,
+ "step": 15219
+ },
+ {
+ "epoch": 41.47138964577657,
+ "grad_norm": 3.2186238765716553,
+ "learning_rate": 1.319121287542992e-05,
+ "loss": 0.1106,
+ "step": 15220
+ },
+ {
+ "epoch": 41.474114441416894,
+ "grad_norm": 3.0683658123016357,
+ "learning_rate": 1.319037651058665e-05,
+ "loss": 0.2682,
+ "step": 15221
+ },
+ {
+ "epoch": 41.47683923705722,
+ "grad_norm": 2.638213634490967,
+ "learning_rate": 1.3189540120896829e-05,
+ "loss": 0.1402,
+ "step": 15222
+ },
+ {
+ "epoch": 41.479564032697546,
+ "grad_norm": 3.4371843338012695,
+ "learning_rate": 1.318870370636697e-05,
+ "loss": 0.0574,
+ "step": 15223
+ },
+ {
+ "epoch": 41.48228882833787,
+ "grad_norm": 3.3930160999298096,
+ "learning_rate": 1.3187867267003592e-05,
+ "loss": 0.2937,
+ "step": 15224
+ },
+ {
+ "epoch": 41.485013623978205,
+ "grad_norm": 3.583491563796997,
+ "learning_rate": 1.3187030802813206e-05,
+ "loss": 0.205,
+ "step": 15225
+ },
+ {
+ "epoch": 41.48773841961853,
+ "grad_norm": 3.30039119720459,
+ "learning_rate": 1.3186194313802325e-05,
+ "loss": 0.0761,
+ "step": 15226
+ },
+ {
+ "epoch": 41.490463215258856,
+ "grad_norm": 3.5668444633483887,
+ "learning_rate": 1.318535779997747e-05,
+ "loss": 0.1646,
+ "step": 15227
+ },
+ {
+ "epoch": 41.49318801089918,
+ "grad_norm": 2.937206983566284,
+ "learning_rate": 1.3184521261345146e-05,
+ "loss": 0.2081,
+ "step": 15228
+ },
+ {
+ "epoch": 41.49591280653951,
+ "grad_norm": 3.306910276412964,
+ "learning_rate": 1.3183684697911875e-05,
+ "loss": 0.1156,
+ "step": 15229
+ },
+ {
+ "epoch": 41.49863760217983,
+ "grad_norm": 3.666545867919922,
+ "learning_rate": 1.318284810968417e-05,
+ "loss": 0.228,
+ "step": 15230
+ },
+ {
+ "epoch": 41.50136239782017,
+ "grad_norm": 3.559070587158203,
+ "learning_rate": 1.3182011496668548e-05,
+ "loss": 0.1628,
+ "step": 15231
+ },
+ {
+ "epoch": 41.50408719346049,
+ "grad_norm": 3.0568957328796387,
+ "learning_rate": 1.3181174858871517e-05,
+ "loss": 0.1153,
+ "step": 15232
+ },
+ {
+ "epoch": 41.50681198910082,
+ "grad_norm": 3.383171796798706,
+ "learning_rate": 1.3180338196299603e-05,
+ "loss": 0.0848,
+ "step": 15233
+ },
+ {
+ "epoch": 41.509536784741144,
+ "grad_norm": 3.0174672603607178,
+ "learning_rate": 1.3179501508959315e-05,
+ "loss": 0.1091,
+ "step": 15234
+ },
+ {
+ "epoch": 41.51226158038147,
+ "grad_norm": 3.2700533866882324,
+ "learning_rate": 1.3178664796857176e-05,
+ "loss": 0.1271,
+ "step": 15235
+ },
+ {
+ "epoch": 41.514986376021795,
+ "grad_norm": 3.3253540992736816,
+ "learning_rate": 1.3177828059999695e-05,
+ "loss": 0.1511,
+ "step": 15236
+ },
+ {
+ "epoch": 41.51771117166213,
+ "grad_norm": 5.653471946716309,
+ "learning_rate": 1.3176991298393393e-05,
+ "loss": 0.2444,
+ "step": 15237
+ },
+ {
+ "epoch": 41.520435967302454,
+ "grad_norm": 2.9478652477264404,
+ "learning_rate": 1.3176154512044783e-05,
+ "loss": 0.1188,
+ "step": 15238
+ },
+ {
+ "epoch": 41.52316076294278,
+ "grad_norm": 6.0193986892700195,
+ "learning_rate": 1.3175317700960386e-05,
+ "loss": 0.2122,
+ "step": 15239
+ },
+ {
+ "epoch": 41.525885558583106,
+ "grad_norm": 3.7397427558898926,
+ "learning_rate": 1.3174480865146712e-05,
+ "loss": 0.2262,
+ "step": 15240
+ },
+ {
+ "epoch": 41.52861035422343,
+ "grad_norm": 3.3189055919647217,
+ "learning_rate": 1.317364400461029e-05,
+ "loss": 0.3147,
+ "step": 15241
+ },
+ {
+ "epoch": 41.53133514986376,
+ "grad_norm": 3.627485990524292,
+ "learning_rate": 1.3172807119357625e-05,
+ "loss": 0.1148,
+ "step": 15242
+ },
+ {
+ "epoch": 41.53405994550409,
+ "grad_norm": 3.287276268005371,
+ "learning_rate": 1.3171970209395243e-05,
+ "loss": 0.1838,
+ "step": 15243
+ },
+ {
+ "epoch": 41.536784741144416,
+ "grad_norm": 3.1880950927734375,
+ "learning_rate": 1.317113327472966e-05,
+ "loss": 0.276,
+ "step": 15244
+ },
+ {
+ "epoch": 41.53950953678474,
+ "grad_norm": 5.051136016845703,
+ "learning_rate": 1.3170296315367392e-05,
+ "loss": 0.0785,
+ "step": 15245
+ },
+ {
+ "epoch": 41.54223433242507,
+ "grad_norm": 3.7270307540893555,
+ "learning_rate": 1.3169459331314958e-05,
+ "loss": 0.1572,
+ "step": 15246
+ },
+ {
+ "epoch": 41.54495912806539,
+ "grad_norm": 4.251556873321533,
+ "learning_rate": 1.3168622322578879e-05,
+ "loss": 0.0827,
+ "step": 15247
+ },
+ {
+ "epoch": 41.54768392370572,
+ "grad_norm": 3.328911781311035,
+ "learning_rate": 1.316778528916567e-05,
+ "loss": 0.0997,
+ "step": 15248
+ },
+ {
+ "epoch": 41.55040871934605,
+ "grad_norm": 3.0779664516448975,
+ "learning_rate": 1.3166948231081849e-05,
+ "loss": 0.1192,
+ "step": 15249
+ },
+ {
+ "epoch": 41.55313351498638,
+ "grad_norm": 3.608757734298706,
+ "learning_rate": 1.3166111148333935e-05,
+ "loss": 0.0899,
+ "step": 15250
+ },
+ {
+ "epoch": 41.555858310626704,
+ "grad_norm": 3.3601341247558594,
+ "learning_rate": 1.3165274040928456e-05,
+ "loss": 0.1255,
+ "step": 15251
+ },
+ {
+ "epoch": 41.55858310626703,
+ "grad_norm": 3.7020857334136963,
+ "learning_rate": 1.316443690887192e-05,
+ "loss": 0.1105,
+ "step": 15252
+ },
+ {
+ "epoch": 41.561307901907355,
+ "grad_norm": 3.27146053314209,
+ "learning_rate": 1.3163599752170852e-05,
+ "loss": 0.0951,
+ "step": 15253
+ },
+ {
+ "epoch": 41.56403269754768,
+ "grad_norm": 3.083031177520752,
+ "learning_rate": 1.3162762570831773e-05,
+ "loss": 0.0805,
+ "step": 15254
+ },
+ {
+ "epoch": 41.566757493188014,
+ "grad_norm": 3.6091935634613037,
+ "learning_rate": 1.31619253648612e-05,
+ "loss": 0.094,
+ "step": 15255
+ },
+ {
+ "epoch": 41.56948228882834,
+ "grad_norm": 2.433016300201416,
+ "learning_rate": 1.3161088134265651e-05,
+ "loss": 0.1807,
+ "step": 15256
+ },
+ {
+ "epoch": 41.572207084468666,
+ "grad_norm": 3.698946475982666,
+ "learning_rate": 1.3160250879051655e-05,
+ "loss": 0.0897,
+ "step": 15257
+ },
+ {
+ "epoch": 41.57493188010899,
+ "grad_norm": 3.9808027744293213,
+ "learning_rate": 1.315941359922572e-05,
+ "loss": 0.1193,
+ "step": 15258
+ },
+ {
+ "epoch": 41.57765667574932,
+ "grad_norm": 2.9388999938964844,
+ "learning_rate": 1.3158576294794378e-05,
+ "loss": 0.1688,
+ "step": 15259
+ },
+ {
+ "epoch": 41.58038147138964,
+ "grad_norm": 3.2644996643066406,
+ "learning_rate": 1.3157738965764146e-05,
+ "loss": 0.3468,
+ "step": 15260
+ },
+ {
+ "epoch": 41.583106267029976,
+ "grad_norm": 4.62359619140625,
+ "learning_rate": 1.3156901612141543e-05,
+ "loss": 0.1745,
+ "step": 15261
+ },
+ {
+ "epoch": 41.5858310626703,
+ "grad_norm": 2.8692948818206787,
+ "learning_rate": 1.3156064233933093e-05,
+ "loss": 0.072,
+ "step": 15262
+ },
+ {
+ "epoch": 41.58855585831063,
+ "grad_norm": 3.7071259021759033,
+ "learning_rate": 1.3155226831145316e-05,
+ "loss": 0.1224,
+ "step": 15263
+ },
+ {
+ "epoch": 41.59128065395095,
+ "grad_norm": 3.0015344619750977,
+ "learning_rate": 1.3154389403784733e-05,
+ "loss": 0.0897,
+ "step": 15264
+ },
+ {
+ "epoch": 41.59400544959128,
+ "grad_norm": 3.226168394088745,
+ "learning_rate": 1.3153551951857869e-05,
+ "loss": 0.1973,
+ "step": 15265
+ },
+ {
+ "epoch": 41.596730245231605,
+ "grad_norm": 2.712505578994751,
+ "learning_rate": 1.315271447537124e-05,
+ "loss": 0.1346,
+ "step": 15266
+ },
+ {
+ "epoch": 41.59945504087194,
+ "grad_norm": 3.0109949111938477,
+ "learning_rate": 1.3151876974331375e-05,
+ "loss": 0.0698,
+ "step": 15267
+ },
+ {
+ "epoch": 41.60217983651226,
+ "grad_norm": 4.1873064041137695,
+ "learning_rate": 1.3151039448744794e-05,
+ "loss": 0.1242,
+ "step": 15268
+ },
+ {
+ "epoch": 41.60490463215259,
+ "grad_norm": 2.860076427459717,
+ "learning_rate": 1.3150201898618019e-05,
+ "loss": 0.2258,
+ "step": 15269
+ },
+ {
+ "epoch": 41.607629427792915,
+ "grad_norm": 3.9189558029174805,
+ "learning_rate": 1.314936432395757e-05,
+ "loss": 0.1345,
+ "step": 15270
+ },
+ {
+ "epoch": 41.61035422343324,
+ "grad_norm": 2.7372286319732666,
+ "learning_rate": 1.3148526724769976e-05,
+ "loss": 0.0771,
+ "step": 15271
+ },
+ {
+ "epoch": 41.61307901907357,
+ "grad_norm": 4.18353796005249,
+ "learning_rate": 1.3147689101061755e-05,
+ "loss": 0.3153,
+ "step": 15272
+ },
+ {
+ "epoch": 41.6158038147139,
+ "grad_norm": 2.8110647201538086,
+ "learning_rate": 1.3146851452839435e-05,
+ "loss": 0.0657,
+ "step": 15273
+ },
+ {
+ "epoch": 41.618528610354225,
+ "grad_norm": 6.645241737365723,
+ "learning_rate": 1.3146013780109536e-05,
+ "loss": 0.1158,
+ "step": 15274
+ },
+ {
+ "epoch": 41.62125340599455,
+ "grad_norm": 2.763091802597046,
+ "learning_rate": 1.3145176082878584e-05,
+ "loss": 0.0869,
+ "step": 15275
+ },
+ {
+ "epoch": 41.62397820163488,
+ "grad_norm": 2.6475908756256104,
+ "learning_rate": 1.3144338361153101e-05,
+ "loss": 0.1078,
+ "step": 15276
+ },
+ {
+ "epoch": 41.6267029972752,
+ "grad_norm": 3.071131706237793,
+ "learning_rate": 1.314350061493961e-05,
+ "loss": 0.2153,
+ "step": 15277
+ },
+ {
+ "epoch": 41.62942779291553,
+ "grad_norm": 4.23525333404541,
+ "learning_rate": 1.314266284424464e-05,
+ "loss": 0.1773,
+ "step": 15278
+ },
+ {
+ "epoch": 41.63215258855586,
+ "grad_norm": 3.873936176300049,
+ "learning_rate": 1.3141825049074712e-05,
+ "loss": 0.1618,
+ "step": 15279
+ },
+ {
+ "epoch": 41.63487738419619,
+ "grad_norm": 3.1964447498321533,
+ "learning_rate": 1.3140987229436353e-05,
+ "loss": 0.1267,
+ "step": 15280
+ },
+ {
+ "epoch": 41.63760217983651,
+ "grad_norm": 3.3715364933013916,
+ "learning_rate": 1.3140149385336085e-05,
+ "loss": 0.1383,
+ "step": 15281
+ },
+ {
+ "epoch": 41.64032697547684,
+ "grad_norm": 3.010953187942505,
+ "learning_rate": 1.3139311516780435e-05,
+ "loss": 0.2087,
+ "step": 15282
+ },
+ {
+ "epoch": 41.643051771117165,
+ "grad_norm": 3.2014291286468506,
+ "learning_rate": 1.3138473623775927e-05,
+ "loss": 0.1192,
+ "step": 15283
+ },
+ {
+ "epoch": 41.64577656675749,
+ "grad_norm": 4.259453773498535,
+ "learning_rate": 1.3137635706329091e-05,
+ "loss": 0.1326,
+ "step": 15284
+ },
+ {
+ "epoch": 41.64850136239782,
+ "grad_norm": 4.570309162139893,
+ "learning_rate": 1.3136797764446445e-05,
+ "loss": 0.1509,
+ "step": 15285
+ },
+ {
+ "epoch": 41.65122615803815,
+ "grad_norm": 2.4461452960968018,
+ "learning_rate": 1.3135959798134522e-05,
+ "loss": 0.1051,
+ "step": 15286
+ },
+ {
+ "epoch": 41.653950953678475,
+ "grad_norm": 4.032877445220947,
+ "learning_rate": 1.3135121807399842e-05,
+ "loss": 0.1139,
+ "step": 15287
+ },
+ {
+ "epoch": 41.6566757493188,
+ "grad_norm": 2.7379589080810547,
+ "learning_rate": 1.3134283792248939e-05,
+ "loss": 0.0577,
+ "step": 15288
+ },
+ {
+ "epoch": 41.65940054495913,
+ "grad_norm": 3.263279914855957,
+ "learning_rate": 1.3133445752688329e-05,
+ "loss": 0.1226,
+ "step": 15289
+ },
+ {
+ "epoch": 41.66212534059945,
+ "grad_norm": 3.6108052730560303,
+ "learning_rate": 1.3132607688724547e-05,
+ "loss": 0.0799,
+ "step": 15290
+ },
+ {
+ "epoch": 41.664850136239785,
+ "grad_norm": 4.610023498535156,
+ "learning_rate": 1.3131769600364117e-05,
+ "loss": 0.137,
+ "step": 15291
+ },
+ {
+ "epoch": 41.66757493188011,
+ "grad_norm": 3.1937034130096436,
+ "learning_rate": 1.3130931487613566e-05,
+ "loss": 0.1064,
+ "step": 15292
+ },
+ {
+ "epoch": 41.67029972752044,
+ "grad_norm": 3.5114119052886963,
+ "learning_rate": 1.313009335047942e-05,
+ "loss": 0.2106,
+ "step": 15293
+ },
+ {
+ "epoch": 41.67302452316076,
+ "grad_norm": 3.313265085220337,
+ "learning_rate": 1.312925518896821e-05,
+ "loss": 0.0815,
+ "step": 15294
+ },
+ {
+ "epoch": 41.67574931880109,
+ "grad_norm": 3.8177218437194824,
+ "learning_rate": 1.312841700308646e-05,
+ "loss": 0.1069,
+ "step": 15295
+ },
+ {
+ "epoch": 41.678474114441414,
+ "grad_norm": 4.413660049438477,
+ "learning_rate": 1.31275787928407e-05,
+ "loss": 0.1446,
+ "step": 15296
+ },
+ {
+ "epoch": 41.68119891008175,
+ "grad_norm": 4.913716793060303,
+ "learning_rate": 1.3126740558237459e-05,
+ "loss": 0.2385,
+ "step": 15297
+ },
+ {
+ "epoch": 41.68392370572207,
+ "grad_norm": 2.6267595291137695,
+ "learning_rate": 1.312590229928326e-05,
+ "loss": 0.1047,
+ "step": 15298
+ },
+ {
+ "epoch": 41.6866485013624,
+ "grad_norm": 2.582348346710205,
+ "learning_rate": 1.3125064015984634e-05,
+ "loss": 0.2048,
+ "step": 15299
+ },
+ {
+ "epoch": 41.689373297002724,
+ "grad_norm": 3.2220139503479004,
+ "learning_rate": 1.3124225708348113e-05,
+ "loss": 0.1292,
+ "step": 15300
+ },
+ {
+ "epoch": 41.69209809264305,
+ "grad_norm": 3.313261032104492,
+ "learning_rate": 1.3123387376380218e-05,
+ "loss": 0.2172,
+ "step": 15301
+ },
+ {
+ "epoch": 41.694822888283376,
+ "grad_norm": 2.457580804824829,
+ "learning_rate": 1.3122549020087487e-05,
+ "loss": 0.1292,
+ "step": 15302
+ },
+ {
+ "epoch": 41.69754768392371,
+ "grad_norm": 3.2573704719543457,
+ "learning_rate": 1.3121710639476444e-05,
+ "loss": 0.0794,
+ "step": 15303
+ },
+ {
+ "epoch": 41.700272479564035,
+ "grad_norm": 2.8478219509124756,
+ "learning_rate": 1.3120872234553618e-05,
+ "loss": 0.1885,
+ "step": 15304
+ },
+ {
+ "epoch": 41.70299727520436,
+ "grad_norm": 4.132429599761963,
+ "learning_rate": 1.3120033805325541e-05,
+ "loss": 0.1621,
+ "step": 15305
+ },
+ {
+ "epoch": 41.705722070844686,
+ "grad_norm": 3.154768466949463,
+ "learning_rate": 1.3119195351798742e-05,
+ "loss": 0.1515,
+ "step": 15306
+ },
+ {
+ "epoch": 41.70844686648501,
+ "grad_norm": 3.5002522468566895,
+ "learning_rate": 1.3118356873979745e-05,
+ "loss": 0.1044,
+ "step": 15307
+ },
+ {
+ "epoch": 41.71117166212534,
+ "grad_norm": 73.18011474609375,
+ "learning_rate": 1.3117518371875091e-05,
+ "loss": 0.1995,
+ "step": 15308
+ },
+ {
+ "epoch": 41.71389645776567,
+ "grad_norm": 4.34943151473999,
+ "learning_rate": 1.3116679845491298e-05,
+ "loss": 0.1202,
+ "step": 15309
+ },
+ {
+ "epoch": 41.716621253406,
+ "grad_norm": 3.3929522037506104,
+ "learning_rate": 1.3115841294834908e-05,
+ "loss": 0.1462,
+ "step": 15310
+ },
+ {
+ "epoch": 41.71934604904632,
+ "grad_norm": 3.599581003189087,
+ "learning_rate": 1.3115002719912443e-05,
+ "loss": 0.0873,
+ "step": 15311
+ },
+ {
+ "epoch": 41.72207084468665,
+ "grad_norm": 4.29304313659668,
+ "learning_rate": 1.3114164120730439e-05,
+ "loss": 0.2627,
+ "step": 15312
+ },
+ {
+ "epoch": 41.724795640326974,
+ "grad_norm": 4.730172157287598,
+ "learning_rate": 1.3113325497295424e-05,
+ "loss": 0.2526,
+ "step": 15313
+ },
+ {
+ "epoch": 41.7275204359673,
+ "grad_norm": 4.117900848388672,
+ "learning_rate": 1.3112486849613933e-05,
+ "loss": 0.0956,
+ "step": 15314
+ },
+ {
+ "epoch": 41.73024523160763,
+ "grad_norm": 4.212102890014648,
+ "learning_rate": 1.311164817769249e-05,
+ "loss": 0.1082,
+ "step": 15315
+ },
+ {
+ "epoch": 41.73297002724796,
+ "grad_norm": 5.299549102783203,
+ "learning_rate": 1.3110809481537637e-05,
+ "loss": 0.257,
+ "step": 15316
+ },
+ {
+ "epoch": 41.735694822888284,
+ "grad_norm": 4.193414688110352,
+ "learning_rate": 1.3109970761155895e-05,
+ "loss": 0.2593,
+ "step": 15317
+ },
+ {
+ "epoch": 41.73841961852861,
+ "grad_norm": 3.734405755996704,
+ "learning_rate": 1.31091320165538e-05,
+ "loss": 0.1818,
+ "step": 15318
+ },
+ {
+ "epoch": 41.741144414168936,
+ "grad_norm": 3.434394359588623,
+ "learning_rate": 1.3108293247737886e-05,
+ "loss": 0.168,
+ "step": 15319
+ },
+ {
+ "epoch": 41.74386920980926,
+ "grad_norm": 5.648651123046875,
+ "learning_rate": 1.3107454454714686e-05,
+ "loss": 0.1422,
+ "step": 15320
+ },
+ {
+ "epoch": 41.746594005449595,
+ "grad_norm": 5.161495208740234,
+ "learning_rate": 1.310661563749073e-05,
+ "loss": 0.1978,
+ "step": 15321
+ },
+ {
+ "epoch": 41.74931880108992,
+ "grad_norm": 4.784137725830078,
+ "learning_rate": 1.3105776796072554e-05,
+ "loss": 0.2248,
+ "step": 15322
+ },
+ {
+ "epoch": 41.752043596730246,
+ "grad_norm": 9.197917938232422,
+ "learning_rate": 1.3104937930466684e-05,
+ "loss": 0.1903,
+ "step": 15323
+ },
+ {
+ "epoch": 41.75476839237057,
+ "grad_norm": 5.131933212280273,
+ "learning_rate": 1.3104099040679658e-05,
+ "loss": 0.2535,
+ "step": 15324
+ },
+ {
+ "epoch": 41.7574931880109,
+ "grad_norm": 4.914931297302246,
+ "learning_rate": 1.3103260126718009e-05,
+ "loss": 0.228,
+ "step": 15325
+ },
+ {
+ "epoch": 41.76021798365122,
+ "grad_norm": 3.613996744155884,
+ "learning_rate": 1.310242118858827e-05,
+ "loss": 0.1168,
+ "step": 15326
+ },
+ {
+ "epoch": 41.762942779291556,
+ "grad_norm": 7.762661933898926,
+ "learning_rate": 1.3101582226296974e-05,
+ "loss": 0.2082,
+ "step": 15327
+ },
+ {
+ "epoch": 41.76566757493188,
+ "grad_norm": 6.866849899291992,
+ "learning_rate": 1.310074323985066e-05,
+ "loss": 0.1846,
+ "step": 15328
+ },
+ {
+ "epoch": 41.76839237057221,
+ "grad_norm": 4.368823528289795,
+ "learning_rate": 1.3099904229255853e-05,
+ "loss": 0.1032,
+ "step": 15329
+ },
+ {
+ "epoch": 41.771117166212534,
+ "grad_norm": 4.01397180557251,
+ "learning_rate": 1.3099065194519094e-05,
+ "loss": 0.1344,
+ "step": 15330
+ },
+ {
+ "epoch": 41.77384196185286,
+ "grad_norm": 5.0536394119262695,
+ "learning_rate": 1.3098226135646913e-05,
+ "loss": 0.2162,
+ "step": 15331
+ },
+ {
+ "epoch": 41.776566757493185,
+ "grad_norm": 4.647796630859375,
+ "learning_rate": 1.3097387052645848e-05,
+ "loss": 0.3636,
+ "step": 15332
+ },
+ {
+ "epoch": 41.77929155313352,
+ "grad_norm": 4.304924964904785,
+ "learning_rate": 1.3096547945522431e-05,
+ "loss": 0.1377,
+ "step": 15333
+ },
+ {
+ "epoch": 41.782016348773844,
+ "grad_norm": 5.244636535644531,
+ "learning_rate": 1.3095708814283199e-05,
+ "loss": 0.1216,
+ "step": 15334
+ },
+ {
+ "epoch": 41.78474114441417,
+ "grad_norm": 5.382554531097412,
+ "learning_rate": 1.3094869658934684e-05,
+ "loss": 0.1383,
+ "step": 15335
+ },
+ {
+ "epoch": 41.787465940054496,
+ "grad_norm": 5.917181491851807,
+ "learning_rate": 1.3094030479483428e-05,
+ "loss": 0.0965,
+ "step": 15336
+ },
+ {
+ "epoch": 41.79019073569482,
+ "grad_norm": 3.556555986404419,
+ "learning_rate": 1.3093191275935957e-05,
+ "loss": 0.0742,
+ "step": 15337
+ },
+ {
+ "epoch": 41.79291553133515,
+ "grad_norm": 4.494316101074219,
+ "learning_rate": 1.3092352048298817e-05,
+ "loss": 0.1598,
+ "step": 15338
+ },
+ {
+ "epoch": 41.79564032697548,
+ "grad_norm": 4.510070323944092,
+ "learning_rate": 1.3091512796578536e-05,
+ "loss": 0.1139,
+ "step": 15339
+ },
+ {
+ "epoch": 41.798365122615806,
+ "grad_norm": 4.499635696411133,
+ "learning_rate": 1.3090673520781656e-05,
+ "loss": 0.1198,
+ "step": 15340
+ },
+ {
+ "epoch": 41.80108991825613,
+ "grad_norm": 3.967008352279663,
+ "learning_rate": 1.3089834220914708e-05,
+ "loss": 0.1058,
+ "step": 15341
+ },
+ {
+ "epoch": 41.80381471389646,
+ "grad_norm": 3.254887342453003,
+ "learning_rate": 1.308899489698423e-05,
+ "loss": 0.1007,
+ "step": 15342
+ },
+ {
+ "epoch": 41.80653950953678,
+ "grad_norm": 3.3996055126190186,
+ "learning_rate": 1.308815554899676e-05,
+ "loss": 0.1166,
+ "step": 15343
+ },
+ {
+ "epoch": 41.80926430517711,
+ "grad_norm": 6.636096477508545,
+ "learning_rate": 1.3087316176958831e-05,
+ "loss": 0.282,
+ "step": 15344
+ },
+ {
+ "epoch": 41.81198910081744,
+ "grad_norm": 4.048243045806885,
+ "learning_rate": 1.3086476780876984e-05,
+ "loss": 0.1855,
+ "step": 15345
+ },
+ {
+ "epoch": 41.81471389645777,
+ "grad_norm": 4.113626956939697,
+ "learning_rate": 1.308563736075776e-05,
+ "loss": 0.1547,
+ "step": 15346
+ },
+ {
+ "epoch": 41.817438692098094,
+ "grad_norm": 4.251448154449463,
+ "learning_rate": 1.3084797916607685e-05,
+ "loss": 0.1403,
+ "step": 15347
+ },
+ {
+ "epoch": 41.82016348773842,
+ "grad_norm": 4.662040710449219,
+ "learning_rate": 1.3083958448433309e-05,
+ "loss": 0.2512,
+ "step": 15348
+ },
+ {
+ "epoch": 41.822888283378745,
+ "grad_norm": 4.403966426849365,
+ "learning_rate": 1.308311895624116e-05,
+ "loss": 0.1383,
+ "step": 15349
+ },
+ {
+ "epoch": 41.82561307901907,
+ "grad_norm": 3.2109880447387695,
+ "learning_rate": 1.3082279440037781e-05,
+ "loss": 0.2036,
+ "step": 15350
+ },
+ {
+ "epoch": 41.828337874659404,
+ "grad_norm": 4.118466854095459,
+ "learning_rate": 1.3081439899829708e-05,
+ "loss": 0.1248,
+ "step": 15351
+ },
+ {
+ "epoch": 41.83106267029973,
+ "grad_norm": 4.314377307891846,
+ "learning_rate": 1.3080600335623481e-05,
+ "loss": 0.2431,
+ "step": 15352
+ },
+ {
+ "epoch": 41.833787465940055,
+ "grad_norm": 3.2194201946258545,
+ "learning_rate": 1.3079760747425639e-05,
+ "loss": 0.163,
+ "step": 15353
+ },
+ {
+ "epoch": 41.83651226158038,
+ "grad_norm": 4.693909645080566,
+ "learning_rate": 1.3078921135242717e-05,
+ "loss": 0.2002,
+ "step": 15354
+ },
+ {
+ "epoch": 41.83923705722071,
+ "grad_norm": 3.7756757736206055,
+ "learning_rate": 1.307808149908126e-05,
+ "loss": 0.1093,
+ "step": 15355
+ },
+ {
+ "epoch": 41.84196185286103,
+ "grad_norm": 4.092785358428955,
+ "learning_rate": 1.3077241838947799e-05,
+ "loss": 0.0718,
+ "step": 15356
+ },
+ {
+ "epoch": 41.844686648501366,
+ "grad_norm": 4.106571197509766,
+ "learning_rate": 1.307640215484888e-05,
+ "loss": 0.1248,
+ "step": 15357
+ },
+ {
+ "epoch": 41.84741144414169,
+ "grad_norm": 3.903512477874756,
+ "learning_rate": 1.3075562446791042e-05,
+ "loss": 0.0885,
+ "step": 15358
+ },
+ {
+ "epoch": 41.85013623978202,
+ "grad_norm": 3.1092114448547363,
+ "learning_rate": 1.3074722714780817e-05,
+ "loss": 0.11,
+ "step": 15359
+ },
+ {
+ "epoch": 41.85286103542234,
+ "grad_norm": 4.112528324127197,
+ "learning_rate": 1.3073882958824755e-05,
+ "loss": 0.1069,
+ "step": 15360
+ },
+ {
+ "epoch": 41.85558583106267,
+ "grad_norm": 4.468888282775879,
+ "learning_rate": 1.307304317892939e-05,
+ "loss": 0.1745,
+ "step": 15361
+ },
+ {
+ "epoch": 41.858310626702995,
+ "grad_norm": 6.639834403991699,
+ "learning_rate": 1.3072203375101263e-05,
+ "loss": 0.1279,
+ "step": 15362
+ },
+ {
+ "epoch": 41.86103542234333,
+ "grad_norm": 3.4828073978424072,
+ "learning_rate": 1.3071363547346916e-05,
+ "loss": 0.255,
+ "step": 15363
+ },
+ {
+ "epoch": 41.86376021798365,
+ "grad_norm": 3.8889565467834473,
+ "learning_rate": 1.3070523695672887e-05,
+ "loss": 0.1398,
+ "step": 15364
+ },
+ {
+ "epoch": 41.86648501362398,
+ "grad_norm": 3.5858659744262695,
+ "learning_rate": 1.3069683820085722e-05,
+ "loss": 0.1909,
+ "step": 15365
+ },
+ {
+ "epoch": 41.869209809264305,
+ "grad_norm": 3.593209981918335,
+ "learning_rate": 1.3068843920591953e-05,
+ "loss": 0.1371,
+ "step": 15366
+ },
+ {
+ "epoch": 41.87193460490463,
+ "grad_norm": 3.2787539958953857,
+ "learning_rate": 1.306800399719813e-05,
+ "loss": 0.1374,
+ "step": 15367
+ },
+ {
+ "epoch": 41.87465940054496,
+ "grad_norm": 4.068887233734131,
+ "learning_rate": 1.306716404991079e-05,
+ "loss": 0.1427,
+ "step": 15368
+ },
+ {
+ "epoch": 41.87738419618529,
+ "grad_norm": 4.79254150390625,
+ "learning_rate": 1.3066324078736474e-05,
+ "loss": 0.1244,
+ "step": 15369
+ },
+ {
+ "epoch": 41.880108991825615,
+ "grad_norm": 3.1378061771392822,
+ "learning_rate": 1.3065484083681726e-05,
+ "loss": 0.1523,
+ "step": 15370
+ },
+ {
+ "epoch": 41.88283378746594,
+ "grad_norm": 4.785521030426025,
+ "learning_rate": 1.3064644064753088e-05,
+ "loss": 0.237,
+ "step": 15371
+ },
+ {
+ "epoch": 41.88555858310627,
+ "grad_norm": 3.739339590072632,
+ "learning_rate": 1.3063804021957096e-05,
+ "loss": 0.1064,
+ "step": 15372
+ },
+ {
+ "epoch": 41.88828337874659,
+ "grad_norm": 5.459422588348389,
+ "learning_rate": 1.3062963955300301e-05,
+ "loss": 0.1142,
+ "step": 15373
+ },
+ {
+ "epoch": 41.89100817438692,
+ "grad_norm": 3.9092392921447754,
+ "learning_rate": 1.306212386478924e-05,
+ "loss": 0.207,
+ "step": 15374
+ },
+ {
+ "epoch": 41.89373297002725,
+ "grad_norm": 3.615450859069824,
+ "learning_rate": 1.3061283750430457e-05,
+ "loss": 0.0767,
+ "step": 15375
+ },
+ {
+ "epoch": 41.89645776566758,
+ "grad_norm": 3.1115622520446777,
+ "learning_rate": 1.3060443612230493e-05,
+ "loss": 0.0876,
+ "step": 15376
+ },
+ {
+ "epoch": 41.8991825613079,
+ "grad_norm": 4.287469863891602,
+ "learning_rate": 1.3059603450195897e-05,
+ "loss": 0.1198,
+ "step": 15377
+ },
+ {
+ "epoch": 41.90190735694823,
+ "grad_norm": 3.6527934074401855,
+ "learning_rate": 1.30587632643332e-05,
+ "loss": 0.1334,
+ "step": 15378
+ },
+ {
+ "epoch": 41.904632152588555,
+ "grad_norm": 4.556772708892822,
+ "learning_rate": 1.305792305464896e-05,
+ "loss": 0.1425,
+ "step": 15379
+ },
+ {
+ "epoch": 41.90735694822888,
+ "grad_norm": 4.3607378005981445,
+ "learning_rate": 1.305708282114971e-05,
+ "loss": 0.2425,
+ "step": 15380
+ },
+ {
+ "epoch": 41.91008174386921,
+ "grad_norm": 14.646055221557617,
+ "learning_rate": 1.3056242563842003e-05,
+ "loss": 0.333,
+ "step": 15381
+ },
+ {
+ "epoch": 41.91280653950954,
+ "grad_norm": 4.261848449707031,
+ "learning_rate": 1.3055402282732372e-05,
+ "loss": 0.1053,
+ "step": 15382
+ },
+ {
+ "epoch": 41.915531335149865,
+ "grad_norm": 8.401264190673828,
+ "learning_rate": 1.3054561977827369e-05,
+ "loss": 0.1618,
+ "step": 15383
+ },
+ {
+ "epoch": 41.91825613079019,
+ "grad_norm": 3.269148349761963,
+ "learning_rate": 1.3053721649133536e-05,
+ "loss": 0.0883,
+ "step": 15384
+ },
+ {
+ "epoch": 41.920980926430516,
+ "grad_norm": 3.9153518676757812,
+ "learning_rate": 1.3052881296657414e-05,
+ "loss": 0.1052,
+ "step": 15385
+ },
+ {
+ "epoch": 41.92370572207084,
+ "grad_norm": 3.2361881732940674,
+ "learning_rate": 1.305204092040555e-05,
+ "loss": 0.0841,
+ "step": 15386
+ },
+ {
+ "epoch": 41.926430517711175,
+ "grad_norm": 4.215946674346924,
+ "learning_rate": 1.3051200520384493e-05,
+ "loss": 0.1167,
+ "step": 15387
+ },
+ {
+ "epoch": 41.9291553133515,
+ "grad_norm": 3.824403762817383,
+ "learning_rate": 1.3050360096600782e-05,
+ "loss": 0.1052,
+ "step": 15388
+ },
+ {
+ "epoch": 41.93188010899183,
+ "grad_norm": 4.1555304527282715,
+ "learning_rate": 1.304951964906097e-05,
+ "loss": 0.1624,
+ "step": 15389
+ },
+ {
+ "epoch": 41.93460490463215,
+ "grad_norm": 4.187368392944336,
+ "learning_rate": 1.304867917777159e-05,
+ "loss": 0.1913,
+ "step": 15390
+ },
+ {
+ "epoch": 41.93732970027248,
+ "grad_norm": 3.439427137374878,
+ "learning_rate": 1.30478386827392e-05,
+ "loss": 0.2313,
+ "step": 15391
+ },
+ {
+ "epoch": 41.940054495912804,
+ "grad_norm": 3.5098299980163574,
+ "learning_rate": 1.3046998163970337e-05,
+ "loss": 0.1153,
+ "step": 15392
+ },
+ {
+ "epoch": 41.94277929155314,
+ "grad_norm": 3.8741233348846436,
+ "learning_rate": 1.3046157621471554e-05,
+ "loss": 0.1314,
+ "step": 15393
+ },
+ {
+ "epoch": 41.94550408719346,
+ "grad_norm": 6.197341442108154,
+ "learning_rate": 1.304531705524939e-05,
+ "loss": 0.1521,
+ "step": 15394
+ },
+ {
+ "epoch": 41.94822888283379,
+ "grad_norm": 4.3119120597839355,
+ "learning_rate": 1.3044476465310393e-05,
+ "loss": 0.1832,
+ "step": 15395
+ },
+ {
+ "epoch": 41.950953678474114,
+ "grad_norm": 4.553133964538574,
+ "learning_rate": 1.3043635851661114e-05,
+ "loss": 0.1744,
+ "step": 15396
+ },
+ {
+ "epoch": 41.95367847411444,
+ "grad_norm": 3.353588819503784,
+ "learning_rate": 1.3042795214308099e-05,
+ "loss": 0.2248,
+ "step": 15397
+ },
+ {
+ "epoch": 41.956403269754766,
+ "grad_norm": 4.102766036987305,
+ "learning_rate": 1.304195455325789e-05,
+ "loss": 0.1846,
+ "step": 15398
+ },
+ {
+ "epoch": 41.95912806539509,
+ "grad_norm": 3.693760633468628,
+ "learning_rate": 1.3041113868517039e-05,
+ "loss": 0.2858,
+ "step": 15399
+ },
+ {
+ "epoch": 41.961852861035425,
+ "grad_norm": 3.2788641452789307,
+ "learning_rate": 1.3040273160092089e-05,
+ "loss": 0.1265,
+ "step": 15400
+ },
+ {
+ "epoch": 41.96457765667575,
+ "grad_norm": 2.983215570449829,
+ "learning_rate": 1.3039432427989592e-05,
+ "loss": 0.0757,
+ "step": 15401
+ },
+ {
+ "epoch": 41.967302452316076,
+ "grad_norm": 2.877533435821533,
+ "learning_rate": 1.303859167221609e-05,
+ "loss": 0.1976,
+ "step": 15402
+ },
+ {
+ "epoch": 41.9700272479564,
+ "grad_norm": 4.4829206466674805,
+ "learning_rate": 1.3037750892778136e-05,
+ "loss": 0.2635,
+ "step": 15403
+ },
+ {
+ "epoch": 41.97275204359673,
+ "grad_norm": 3.8609209060668945,
+ "learning_rate": 1.3036910089682272e-05,
+ "loss": 0.1287,
+ "step": 15404
+ },
+ {
+ "epoch": 41.97547683923706,
+ "grad_norm": 3.6533796787261963,
+ "learning_rate": 1.3036069262935057e-05,
+ "loss": 0.1324,
+ "step": 15405
+ },
+ {
+ "epoch": 41.97820163487739,
+ "grad_norm": 4.566494941711426,
+ "learning_rate": 1.3035228412543026e-05,
+ "loss": 0.2131,
+ "step": 15406
+ },
+ {
+ "epoch": 41.98092643051771,
+ "grad_norm": 4.530869960784912,
+ "learning_rate": 1.3034387538512739e-05,
+ "loss": 0.2924,
+ "step": 15407
+ },
+ {
+ "epoch": 41.98365122615804,
+ "grad_norm": 4.387035369873047,
+ "learning_rate": 1.3033546640850734e-05,
+ "loss": 0.1998,
+ "step": 15408
+ },
+ {
+ "epoch": 41.986376021798364,
+ "grad_norm": 3.805189847946167,
+ "learning_rate": 1.3032705719563569e-05,
+ "loss": 0.2346,
+ "step": 15409
+ },
+ {
+ "epoch": 41.98910081743869,
+ "grad_norm": 4.669410705566406,
+ "learning_rate": 1.3031864774657788e-05,
+ "loss": 0.0629,
+ "step": 15410
+ },
+ {
+ "epoch": 41.991825613079016,
+ "grad_norm": 4.315951347351074,
+ "learning_rate": 1.3031023806139944e-05,
+ "loss": 0.2019,
+ "step": 15411
+ },
+ {
+ "epoch": 41.99455040871935,
+ "grad_norm": 3.9831011295318604,
+ "learning_rate": 1.303018281401658e-05,
+ "loss": 0.0989,
+ "step": 15412
+ },
+ {
+ "epoch": 41.997275204359674,
+ "grad_norm": 3.465353012084961,
+ "learning_rate": 1.3029341798294253e-05,
+ "loss": 0.0855,
+ "step": 15413
+ },
+ {
+ "epoch": 42.0,
+ "grad_norm": 3.9033589363098145,
+ "learning_rate": 1.3028500758979507e-05,
+ "loss": 0.0988,
+ "step": 15414
+ },
+ {
+ "epoch": 42.002724795640326,
+ "grad_norm": 3.1742355823516846,
+ "learning_rate": 1.3027659696078898e-05,
+ "loss": 0.168,
+ "step": 15415
+ },
+ {
+ "epoch": 42.00544959128065,
+ "grad_norm": 4.614014148712158,
+ "learning_rate": 1.302681860959897e-05,
+ "loss": 0.1696,
+ "step": 15416
+ },
+ {
+ "epoch": 42.00817438692098,
+ "grad_norm": 3.552983283996582,
+ "learning_rate": 1.3025977499546279e-05,
+ "loss": 0.0651,
+ "step": 15417
+ },
+ {
+ "epoch": 42.01089918256131,
+ "grad_norm": 3.9310548305511475,
+ "learning_rate": 1.3025136365927371e-05,
+ "loss": 0.084,
+ "step": 15418
+ },
+ {
+ "epoch": 42.013623978201636,
+ "grad_norm": 9.503935813903809,
+ "learning_rate": 1.3024295208748798e-05,
+ "loss": 0.1907,
+ "step": 15419
+ },
+ {
+ "epoch": 42.01634877384196,
+ "grad_norm": 5.0625996589660645,
+ "learning_rate": 1.3023454028017111e-05,
+ "loss": 0.1255,
+ "step": 15420
+ },
+ {
+ "epoch": 42.01907356948229,
+ "grad_norm": 3.501913547515869,
+ "learning_rate": 1.3022612823738861e-05,
+ "loss": 0.183,
+ "step": 15421
+ },
+ {
+ "epoch": 42.02179836512261,
+ "grad_norm": 3.1586201190948486,
+ "learning_rate": 1.3021771595920601e-05,
+ "loss": 0.1648,
+ "step": 15422
+ },
+ {
+ "epoch": 42.02452316076294,
+ "grad_norm": 3.6308014392852783,
+ "learning_rate": 1.3020930344568881e-05,
+ "loss": 0.1545,
+ "step": 15423
+ },
+ {
+ "epoch": 42.02724795640327,
+ "grad_norm": 4.130557537078857,
+ "learning_rate": 1.3020089069690252e-05,
+ "loss": 0.1117,
+ "step": 15424
+ },
+ {
+ "epoch": 42.0299727520436,
+ "grad_norm": 4.02703332901001,
+ "learning_rate": 1.3019247771291265e-05,
+ "loss": 0.1372,
+ "step": 15425
+ },
+ {
+ "epoch": 42.032697547683924,
+ "grad_norm": 3.26668119430542,
+ "learning_rate": 1.3018406449378475e-05,
+ "loss": 0.147,
+ "step": 15426
+ },
+ {
+ "epoch": 42.03542234332425,
+ "grad_norm": 3.9211678504943848,
+ "learning_rate": 1.3017565103958434e-05,
+ "loss": 0.0785,
+ "step": 15427
+ },
+ {
+ "epoch": 42.038147138964575,
+ "grad_norm": 2.739295482635498,
+ "learning_rate": 1.3016723735037693e-05,
+ "loss": 0.0617,
+ "step": 15428
+ },
+ {
+ "epoch": 42.0408719346049,
+ "grad_norm": 5.188828945159912,
+ "learning_rate": 1.3015882342622804e-05,
+ "loss": 0.2598,
+ "step": 15429
+ },
+ {
+ "epoch": 42.043596730245234,
+ "grad_norm": 3.207833766937256,
+ "learning_rate": 1.3015040926720319e-05,
+ "loss": 0.0778,
+ "step": 15430
+ },
+ {
+ "epoch": 42.04632152588556,
+ "grad_norm": 4.082417964935303,
+ "learning_rate": 1.3014199487336792e-05,
+ "loss": 0.1177,
+ "step": 15431
+ },
+ {
+ "epoch": 42.049046321525886,
+ "grad_norm": 2.8504977226257324,
+ "learning_rate": 1.3013358024478778e-05,
+ "loss": 0.0846,
+ "step": 15432
+ },
+ {
+ "epoch": 42.05177111716621,
+ "grad_norm": 3.471031665802002,
+ "learning_rate": 1.3012516538152828e-05,
+ "loss": 0.0798,
+ "step": 15433
+ },
+ {
+ "epoch": 42.05449591280654,
+ "grad_norm": 4.720677852630615,
+ "learning_rate": 1.3011675028365496e-05,
+ "loss": 0.0915,
+ "step": 15434
+ },
+ {
+ "epoch": 42.05722070844686,
+ "grad_norm": 3.3215835094451904,
+ "learning_rate": 1.3010833495123339e-05,
+ "loss": 0.0843,
+ "step": 15435
+ },
+ {
+ "epoch": 42.059945504087196,
+ "grad_norm": 2.9701082706451416,
+ "learning_rate": 1.3009991938432904e-05,
+ "loss": 0.1043,
+ "step": 15436
+ },
+ {
+ "epoch": 42.06267029972752,
+ "grad_norm": 2.256822347640991,
+ "learning_rate": 1.3009150358300751e-05,
+ "loss": 0.0609,
+ "step": 15437
+ },
+ {
+ "epoch": 42.06539509536785,
+ "grad_norm": 7.208272933959961,
+ "learning_rate": 1.3008308754733428e-05,
+ "loss": 0.1055,
+ "step": 15438
+ },
+ {
+ "epoch": 42.06811989100817,
+ "grad_norm": 4.856968879699707,
+ "learning_rate": 1.3007467127737497e-05,
+ "loss": 0.2192,
+ "step": 15439
+ },
+ {
+ "epoch": 42.0708446866485,
+ "grad_norm": 3.506649971008301,
+ "learning_rate": 1.3006625477319509e-05,
+ "loss": 0.1047,
+ "step": 15440
+ },
+ {
+ "epoch": 42.073569482288825,
+ "grad_norm": 3.737814426422119,
+ "learning_rate": 1.3005783803486015e-05,
+ "loss": 0.0983,
+ "step": 15441
+ },
+ {
+ "epoch": 42.07629427792916,
+ "grad_norm": 3.901151657104492,
+ "learning_rate": 1.3004942106243576e-05,
+ "loss": 0.1362,
+ "step": 15442
+ },
+ {
+ "epoch": 42.079019073569484,
+ "grad_norm": 3.002098560333252,
+ "learning_rate": 1.3004100385598744e-05,
+ "loss": 0.1111,
+ "step": 15443
+ },
+ {
+ "epoch": 42.08174386920981,
+ "grad_norm": 3.8361096382141113,
+ "learning_rate": 1.3003258641558073e-05,
+ "loss": 0.229,
+ "step": 15444
+ },
+ {
+ "epoch": 42.084468664850135,
+ "grad_norm": 2.995012044906616,
+ "learning_rate": 1.3002416874128125e-05,
+ "loss": 0.1621,
+ "step": 15445
+ },
+ {
+ "epoch": 42.08719346049046,
+ "grad_norm": 4.957695960998535,
+ "learning_rate": 1.3001575083315447e-05,
+ "loss": 0.1868,
+ "step": 15446
+ },
+ {
+ "epoch": 42.08991825613079,
+ "grad_norm": 3.357353448867798,
+ "learning_rate": 1.30007332691266e-05,
+ "loss": 0.0927,
+ "step": 15447
+ },
+ {
+ "epoch": 42.09264305177112,
+ "grad_norm": 3.1168527603149414,
+ "learning_rate": 1.2999891431568141e-05,
+ "loss": 0.0697,
+ "step": 15448
+ },
+ {
+ "epoch": 42.095367847411445,
+ "grad_norm": 3.9983081817626953,
+ "learning_rate": 1.299904957064662e-05,
+ "loss": 0.0783,
+ "step": 15449
+ },
+ {
+ "epoch": 42.09809264305177,
+ "grad_norm": 2.999393939971924,
+ "learning_rate": 1.29982076863686e-05,
+ "loss": 0.101,
+ "step": 15450
+ },
+ {
+ "epoch": 42.1008174386921,
+ "grad_norm": 3.84466814994812,
+ "learning_rate": 1.2997365778740636e-05,
+ "loss": 0.1326,
+ "step": 15451
+ },
+ {
+ "epoch": 42.10354223433242,
+ "grad_norm": 2.826019525527954,
+ "learning_rate": 1.2996523847769283e-05,
+ "loss": 0.1084,
+ "step": 15452
+ },
+ {
+ "epoch": 42.10626702997275,
+ "grad_norm": 3.1953113079071045,
+ "learning_rate": 1.2995681893461098e-05,
+ "loss": 0.0951,
+ "step": 15453
+ },
+ {
+ "epoch": 42.10899182561308,
+ "grad_norm": 4.2031779289245605,
+ "learning_rate": 1.2994839915822639e-05,
+ "loss": 0.0863,
+ "step": 15454
+ },
+ {
+ "epoch": 42.11171662125341,
+ "grad_norm": 3.346635580062866,
+ "learning_rate": 1.2993997914860464e-05,
+ "loss": 0.0849,
+ "step": 15455
+ },
+ {
+ "epoch": 42.11444141689373,
+ "grad_norm": 4.016412734985352,
+ "learning_rate": 1.2993155890581127e-05,
+ "loss": 0.088,
+ "step": 15456
+ },
+ {
+ "epoch": 42.11716621253406,
+ "grad_norm": 3.733713150024414,
+ "learning_rate": 1.2992313842991189e-05,
+ "loss": 0.152,
+ "step": 15457
+ },
+ {
+ "epoch": 42.119891008174385,
+ "grad_norm": 2.9617199897766113,
+ "learning_rate": 1.2991471772097211e-05,
+ "loss": 0.0748,
+ "step": 15458
+ },
+ {
+ "epoch": 42.12261580381471,
+ "grad_norm": 3.73180890083313,
+ "learning_rate": 1.2990629677905741e-05,
+ "loss": 0.1034,
+ "step": 15459
+ },
+ {
+ "epoch": 42.12534059945504,
+ "grad_norm": 9.536408424377441,
+ "learning_rate": 1.298978756042335e-05,
+ "loss": 0.1307,
+ "step": 15460
+ },
+ {
+ "epoch": 42.12806539509537,
+ "grad_norm": 3.0497305393218994,
+ "learning_rate": 1.2988945419656585e-05,
+ "loss": 0.09,
+ "step": 15461
+ },
+ {
+ "epoch": 42.130790190735695,
+ "grad_norm": 4.133369445800781,
+ "learning_rate": 1.2988103255612012e-05,
+ "loss": 0.1131,
+ "step": 15462
+ },
+ {
+ "epoch": 42.13351498637602,
+ "grad_norm": 2.859269857406616,
+ "learning_rate": 1.2987261068296184e-05,
+ "loss": 0.0948,
+ "step": 15463
+ },
+ {
+ "epoch": 42.13623978201635,
+ "grad_norm": 28.01190185546875,
+ "learning_rate": 1.2986418857715663e-05,
+ "loss": 0.1197,
+ "step": 15464
+ },
+ {
+ "epoch": 42.13896457765667,
+ "grad_norm": 4.041991233825684,
+ "learning_rate": 1.298557662387701e-05,
+ "loss": 0.1109,
+ "step": 15465
+ },
+ {
+ "epoch": 42.141689373297005,
+ "grad_norm": 3.502408504486084,
+ "learning_rate": 1.2984734366786782e-05,
+ "loss": 0.1236,
+ "step": 15466
+ },
+ {
+ "epoch": 42.14441416893733,
+ "grad_norm": 3.561000108718872,
+ "learning_rate": 1.298389208645154e-05,
+ "loss": 0.1125,
+ "step": 15467
+ },
+ {
+ "epoch": 42.14713896457766,
+ "grad_norm": 8.574073791503906,
+ "learning_rate": 1.2983049782877842e-05,
+ "loss": 0.2344,
+ "step": 15468
+ },
+ {
+ "epoch": 42.14986376021798,
+ "grad_norm": 3.4600541591644287,
+ "learning_rate": 1.2982207456072246e-05,
+ "loss": 0.0853,
+ "step": 15469
+ },
+ {
+ "epoch": 42.15258855585831,
+ "grad_norm": 3.065230131149292,
+ "learning_rate": 1.2981365106041317e-05,
+ "loss": 0.1881,
+ "step": 15470
+ },
+ {
+ "epoch": 42.155313351498634,
+ "grad_norm": 4.190520763397217,
+ "learning_rate": 1.298052273279161e-05,
+ "loss": 0.1786,
+ "step": 15471
+ },
+ {
+ "epoch": 42.15803814713897,
+ "grad_norm": 2.7054169178009033,
+ "learning_rate": 1.297968033632969e-05,
+ "loss": 0.1392,
+ "step": 15472
+ },
+ {
+ "epoch": 42.16076294277929,
+ "grad_norm": 3.7768139839172363,
+ "learning_rate": 1.2978837916662113e-05,
+ "loss": 0.2504,
+ "step": 15473
+ },
+ {
+ "epoch": 42.16348773841962,
+ "grad_norm": 11.856160163879395,
+ "learning_rate": 1.2977995473795444e-05,
+ "loss": 0.1328,
+ "step": 15474
+ },
+ {
+ "epoch": 42.166212534059945,
+ "grad_norm": 4.816944599151611,
+ "learning_rate": 1.2977153007736241e-05,
+ "loss": 0.1376,
+ "step": 15475
+ },
+ {
+ "epoch": 42.16893732970027,
+ "grad_norm": 2.8558719158172607,
+ "learning_rate": 1.2976310518491068e-05,
+ "loss": 0.2704,
+ "step": 15476
+ },
+ {
+ "epoch": 42.171662125340596,
+ "grad_norm": 2.9081382751464844,
+ "learning_rate": 1.2975468006066483e-05,
+ "loss": 0.1976,
+ "step": 15477
+ },
+ {
+ "epoch": 42.17438692098093,
+ "grad_norm": 4.355327129364014,
+ "learning_rate": 1.2974625470469051e-05,
+ "loss": 0.2354,
+ "step": 15478
+ },
+ {
+ "epoch": 42.177111716621255,
+ "grad_norm": 3.23808217048645,
+ "learning_rate": 1.297378291170533e-05,
+ "loss": 0.1329,
+ "step": 15479
+ },
+ {
+ "epoch": 42.17983651226158,
+ "grad_norm": 3.679736375808716,
+ "learning_rate": 1.2972940329781885e-05,
+ "loss": 0.1644,
+ "step": 15480
+ },
+ {
+ "epoch": 42.182561307901906,
+ "grad_norm": 3.5791730880737305,
+ "learning_rate": 1.2972097724705273e-05,
+ "loss": 0.0772,
+ "step": 15481
+ },
+ {
+ "epoch": 42.18528610354223,
+ "grad_norm": 3.207584857940674,
+ "learning_rate": 1.2971255096482063e-05,
+ "loss": 0.1003,
+ "step": 15482
+ },
+ {
+ "epoch": 42.18801089918256,
+ "grad_norm": 4.792714595794678,
+ "learning_rate": 1.297041244511881e-05,
+ "loss": 0.0859,
+ "step": 15483
+ },
+ {
+ "epoch": 42.19073569482289,
+ "grad_norm": 3.2615530490875244,
+ "learning_rate": 1.2969569770622084e-05,
+ "loss": 0.2137,
+ "step": 15484
+ },
+ {
+ "epoch": 42.19346049046322,
+ "grad_norm": 3.8044896125793457,
+ "learning_rate": 1.2968727072998442e-05,
+ "loss": 0.172,
+ "step": 15485
+ },
+ {
+ "epoch": 42.19618528610354,
+ "grad_norm": 2.89691424369812,
+ "learning_rate": 1.2967884352254451e-05,
+ "loss": 0.1155,
+ "step": 15486
+ },
+ {
+ "epoch": 42.19891008174387,
+ "grad_norm": 4.018763542175293,
+ "learning_rate": 1.2967041608396671e-05,
+ "loss": 0.0971,
+ "step": 15487
+ },
+ {
+ "epoch": 42.201634877384194,
+ "grad_norm": 4.385662078857422,
+ "learning_rate": 1.296619884143167e-05,
+ "loss": 0.3136,
+ "step": 15488
+ },
+ {
+ "epoch": 42.20435967302452,
+ "grad_norm": 3.6733434200286865,
+ "learning_rate": 1.2965356051366003e-05,
+ "loss": 0.1871,
+ "step": 15489
+ },
+ {
+ "epoch": 42.20708446866485,
+ "grad_norm": 4.509671211242676,
+ "learning_rate": 1.296451323820624e-05,
+ "loss": 0.1213,
+ "step": 15490
+ },
+ {
+ "epoch": 42.20980926430518,
+ "grad_norm": 10.481316566467285,
+ "learning_rate": 1.2963670401958944e-05,
+ "loss": 0.0773,
+ "step": 15491
+ },
+ {
+ "epoch": 42.212534059945504,
+ "grad_norm": 3.5957376956939697,
+ "learning_rate": 1.2962827542630678e-05,
+ "loss": 0.1029,
+ "step": 15492
+ },
+ {
+ "epoch": 42.21525885558583,
+ "grad_norm": 2.6438710689544678,
+ "learning_rate": 1.2961984660228005e-05,
+ "loss": 0.0611,
+ "step": 15493
+ },
+ {
+ "epoch": 42.217983651226156,
+ "grad_norm": 3.6552844047546387,
+ "learning_rate": 1.2961141754757494e-05,
+ "loss": 0.2092,
+ "step": 15494
+ },
+ {
+ "epoch": 42.22070844686648,
+ "grad_norm": 5.091664791107178,
+ "learning_rate": 1.2960298826225701e-05,
+ "loss": 0.1993,
+ "step": 15495
+ },
+ {
+ "epoch": 42.223433242506815,
+ "grad_norm": 3.558835983276367,
+ "learning_rate": 1.2959455874639202e-05,
+ "loss": 0.2218,
+ "step": 15496
+ },
+ {
+ "epoch": 42.22615803814714,
+ "grad_norm": 3.2033820152282715,
+ "learning_rate": 1.2958612900004552e-05,
+ "loss": 0.1215,
+ "step": 15497
+ },
+ {
+ "epoch": 42.228882833787466,
+ "grad_norm": 5.174519062042236,
+ "learning_rate": 1.2957769902328324e-05,
+ "loss": 0.1006,
+ "step": 15498
+ },
+ {
+ "epoch": 42.23160762942779,
+ "grad_norm": 4.325557708740234,
+ "learning_rate": 1.2956926881617073e-05,
+ "loss": 0.0619,
+ "step": 15499
+ },
+ {
+ "epoch": 42.23433242506812,
+ "grad_norm": 3.6242713928222656,
+ "learning_rate": 1.2956083837877376e-05,
+ "loss": 0.078,
+ "step": 15500
+ },
+ {
+ "epoch": 42.237057220708444,
+ "grad_norm": 3.6170153617858887,
+ "learning_rate": 1.2955240771115788e-05,
+ "loss": 0.2429,
+ "step": 15501
+ },
+ {
+ "epoch": 42.23978201634878,
+ "grad_norm": 3.0377368927001953,
+ "learning_rate": 1.2954397681338886e-05,
+ "loss": 0.0609,
+ "step": 15502
+ },
+ {
+ "epoch": 42.2425068119891,
+ "grad_norm": 4.360470771789551,
+ "learning_rate": 1.2953554568553225e-05,
+ "loss": 0.1634,
+ "step": 15503
+ },
+ {
+ "epoch": 42.24523160762943,
+ "grad_norm": 3.6672377586364746,
+ "learning_rate": 1.2952711432765381e-05,
+ "loss": 0.1706,
+ "step": 15504
+ },
+ {
+ "epoch": 42.247956403269754,
+ "grad_norm": 3.257526159286499,
+ "learning_rate": 1.2951868273981912e-05,
+ "loss": 0.0799,
+ "step": 15505
+ },
+ {
+ "epoch": 42.25068119891008,
+ "grad_norm": 4.8078203201293945,
+ "learning_rate": 1.2951025092209391e-05,
+ "loss": 0.1951,
+ "step": 15506
+ },
+ {
+ "epoch": 42.253405994550405,
+ "grad_norm": 4.217690467834473,
+ "learning_rate": 1.2950181887454374e-05,
+ "loss": 0.0989,
+ "step": 15507
+ },
+ {
+ "epoch": 42.25613079019074,
+ "grad_norm": 3.297131061553955,
+ "learning_rate": 1.2949338659723443e-05,
+ "loss": 0.1601,
+ "step": 15508
+ },
+ {
+ "epoch": 42.258855585831064,
+ "grad_norm": 5.313800811767578,
+ "learning_rate": 1.2948495409023155e-05,
+ "loss": 0.2912,
+ "step": 15509
+ },
+ {
+ "epoch": 42.26158038147139,
+ "grad_norm": 3.5955193042755127,
+ "learning_rate": 1.294765213536008e-05,
+ "loss": 0.0813,
+ "step": 15510
+ },
+ {
+ "epoch": 42.264305177111716,
+ "grad_norm": 3.0958147048950195,
+ "learning_rate": 1.2946808838740786e-05,
+ "loss": 0.1046,
+ "step": 15511
+ },
+ {
+ "epoch": 42.26702997275204,
+ "grad_norm": 3.226656436920166,
+ "learning_rate": 1.2945965519171837e-05,
+ "loss": 0.1146,
+ "step": 15512
+ },
+ {
+ "epoch": 42.26975476839237,
+ "grad_norm": 5.8506669998168945,
+ "learning_rate": 1.2945122176659806e-05,
+ "loss": 0.0972,
+ "step": 15513
+ },
+ {
+ "epoch": 42.2724795640327,
+ "grad_norm": 3.015350580215454,
+ "learning_rate": 1.2944278811211257e-05,
+ "loss": 0.2195,
+ "step": 15514
+ },
+ {
+ "epoch": 42.275204359673026,
+ "grad_norm": 3.2294762134552,
+ "learning_rate": 1.2943435422832757e-05,
+ "loss": 0.1375,
+ "step": 15515
+ },
+ {
+ "epoch": 42.27792915531335,
+ "grad_norm": 4.236064434051514,
+ "learning_rate": 1.2942592011530878e-05,
+ "loss": 0.228,
+ "step": 15516
+ },
+ {
+ "epoch": 42.28065395095368,
+ "grad_norm": 2.9839282035827637,
+ "learning_rate": 1.294174857731219e-05,
+ "loss": 0.0701,
+ "step": 15517
+ },
+ {
+ "epoch": 42.283378746594,
+ "grad_norm": 3.223029136657715,
+ "learning_rate": 1.2940905120183256e-05,
+ "loss": 0.0738,
+ "step": 15518
+ },
+ {
+ "epoch": 42.28610354223433,
+ "grad_norm": 3.9331300258636475,
+ "learning_rate": 1.2940061640150647e-05,
+ "loss": 0.1109,
+ "step": 15519
+ },
+ {
+ "epoch": 42.28882833787466,
+ "grad_norm": 12.08202838897705,
+ "learning_rate": 1.2939218137220937e-05,
+ "loss": 0.1151,
+ "step": 15520
+ },
+ {
+ "epoch": 42.29155313351499,
+ "grad_norm": 3.228304624557495,
+ "learning_rate": 1.2938374611400686e-05,
+ "loss": 0.0711,
+ "step": 15521
+ },
+ {
+ "epoch": 42.294277929155314,
+ "grad_norm": 3.5593299865722656,
+ "learning_rate": 1.2937531062696472e-05,
+ "loss": 0.1475,
+ "step": 15522
+ },
+ {
+ "epoch": 42.29700272479564,
+ "grad_norm": 3.1819238662719727,
+ "learning_rate": 1.2936687491114857e-05,
+ "loss": 0.1053,
+ "step": 15523
+ },
+ {
+ "epoch": 42.299727520435965,
+ "grad_norm": 2.76425838470459,
+ "learning_rate": 1.2935843896662415e-05,
+ "loss": 0.1729,
+ "step": 15524
+ },
+ {
+ "epoch": 42.30245231607629,
+ "grad_norm": 2.878596067428589,
+ "learning_rate": 1.2935000279345717e-05,
+ "loss": 0.1885,
+ "step": 15525
+ },
+ {
+ "epoch": 42.305177111716624,
+ "grad_norm": 4.074594497680664,
+ "learning_rate": 1.293415663917133e-05,
+ "loss": 0.1342,
+ "step": 15526
+ },
+ {
+ "epoch": 42.30790190735695,
+ "grad_norm": 3.680797815322876,
+ "learning_rate": 1.2933312976145827e-05,
+ "loss": 0.1419,
+ "step": 15527
+ },
+ {
+ "epoch": 42.310626702997276,
+ "grad_norm": 5.727573394775391,
+ "learning_rate": 1.2932469290275776e-05,
+ "loss": 0.1229,
+ "step": 15528
+ },
+ {
+ "epoch": 42.3133514986376,
+ "grad_norm": 2.6009814739227295,
+ "learning_rate": 1.2931625581567748e-05,
+ "loss": 0.1378,
+ "step": 15529
+ },
+ {
+ "epoch": 42.31607629427793,
+ "grad_norm": 4.474771976470947,
+ "learning_rate": 1.2930781850028313e-05,
+ "loss": 0.1085,
+ "step": 15530
+ },
+ {
+ "epoch": 42.31880108991825,
+ "grad_norm": 7.4341325759887695,
+ "learning_rate": 1.2929938095664045e-05,
+ "loss": 0.1905,
+ "step": 15531
+ },
+ {
+ "epoch": 42.321525885558586,
+ "grad_norm": 3.9599878787994385,
+ "learning_rate": 1.2929094318481515e-05,
+ "loss": 0.1643,
+ "step": 15532
+ },
+ {
+ "epoch": 42.32425068119891,
+ "grad_norm": 3.4821643829345703,
+ "learning_rate": 1.2928250518487291e-05,
+ "loss": 0.154,
+ "step": 15533
+ },
+ {
+ "epoch": 42.32697547683924,
+ "grad_norm": 3.4598348140716553,
+ "learning_rate": 1.2927406695687945e-05,
+ "loss": 0.0898,
+ "step": 15534
+ },
+ {
+ "epoch": 42.32970027247956,
+ "grad_norm": 3.189786911010742,
+ "learning_rate": 1.2926562850090052e-05,
+ "loss": 0.0851,
+ "step": 15535
+ },
+ {
+ "epoch": 42.33242506811989,
+ "grad_norm": 3.2347185611724854,
+ "learning_rate": 1.2925718981700181e-05,
+ "loss": 0.0957,
+ "step": 15536
+ },
+ {
+ "epoch": 42.335149863760215,
+ "grad_norm": 3.279073715209961,
+ "learning_rate": 1.2924875090524904e-05,
+ "loss": 0.0749,
+ "step": 15537
+ },
+ {
+ "epoch": 42.33787465940055,
+ "grad_norm": 3.2201528549194336,
+ "learning_rate": 1.2924031176570794e-05,
+ "loss": 0.1865,
+ "step": 15538
+ },
+ {
+ "epoch": 42.34059945504087,
+ "grad_norm": 3.127079725265503,
+ "learning_rate": 1.2923187239844426e-05,
+ "loss": 0.1998,
+ "step": 15539
+ },
+ {
+ "epoch": 42.3433242506812,
+ "grad_norm": 3.3544318675994873,
+ "learning_rate": 1.2922343280352367e-05,
+ "loss": 0.1172,
+ "step": 15540
+ },
+ {
+ "epoch": 42.346049046321525,
+ "grad_norm": 5.234440803527832,
+ "learning_rate": 1.2921499298101193e-05,
+ "loss": 0.1144,
+ "step": 15541
+ },
+ {
+ "epoch": 42.34877384196185,
+ "grad_norm": 3.827716112136841,
+ "learning_rate": 1.2920655293097474e-05,
+ "loss": 0.1516,
+ "step": 15542
+ },
+ {
+ "epoch": 42.35149863760218,
+ "grad_norm": 2.7813785076141357,
+ "learning_rate": 1.291981126534779e-05,
+ "loss": 0.0923,
+ "step": 15543
+ },
+ {
+ "epoch": 42.35422343324251,
+ "grad_norm": 7.174185276031494,
+ "learning_rate": 1.2918967214858709e-05,
+ "loss": 0.0933,
+ "step": 15544
+ },
+ {
+ "epoch": 42.356948228882835,
+ "grad_norm": 3.288094997406006,
+ "learning_rate": 1.2918123141636804e-05,
+ "loss": 0.1749,
+ "step": 15545
+ },
+ {
+ "epoch": 42.35967302452316,
+ "grad_norm": 4.855967044830322,
+ "learning_rate": 1.291727904568865e-05,
+ "loss": 0.167,
+ "step": 15546
+ },
+ {
+ "epoch": 42.36239782016349,
+ "grad_norm": 4.854400157928467,
+ "learning_rate": 1.2916434927020825e-05,
+ "loss": 0.0606,
+ "step": 15547
+ },
+ {
+ "epoch": 42.36512261580381,
+ "grad_norm": 3.4712085723876953,
+ "learning_rate": 1.291559078563989e-05,
+ "loss": 0.127,
+ "step": 15548
+ },
+ {
+ "epoch": 42.36784741144414,
+ "grad_norm": 2.990720272064209,
+ "learning_rate": 1.2914746621552437e-05,
+ "loss": 0.0823,
+ "step": 15549
+ },
+ {
+ "epoch": 42.37057220708447,
+ "grad_norm": 4.013089656829834,
+ "learning_rate": 1.2913902434765023e-05,
+ "loss": 0.1557,
+ "step": 15550
+ },
+ {
+ "epoch": 42.3732970027248,
+ "grad_norm": 3.2428770065307617,
+ "learning_rate": 1.2913058225284237e-05,
+ "loss": 0.1403,
+ "step": 15551
+ },
+ {
+ "epoch": 42.37602179836512,
+ "grad_norm": 3.4551126956939697,
+ "learning_rate": 1.2912213993116642e-05,
+ "loss": 0.0673,
+ "step": 15552
+ },
+ {
+ "epoch": 42.37874659400545,
+ "grad_norm": 3.481874942779541,
+ "learning_rate": 1.2911369738268823e-05,
+ "loss": 0.2574,
+ "step": 15553
+ },
+ {
+ "epoch": 42.381471389645775,
+ "grad_norm": 3.792058229446411,
+ "learning_rate": 1.2910525460747346e-05,
+ "loss": 0.1721,
+ "step": 15554
+ },
+ {
+ "epoch": 42.3841961852861,
+ "grad_norm": 3.2958972454071045,
+ "learning_rate": 1.2909681160558794e-05,
+ "loss": 0.3522,
+ "step": 15555
+ },
+ {
+ "epoch": 42.38692098092643,
+ "grad_norm": 2.785099744796753,
+ "learning_rate": 1.2908836837709734e-05,
+ "loss": 0.0554,
+ "step": 15556
+ },
+ {
+ "epoch": 42.38964577656676,
+ "grad_norm": 2.6389453411102295,
+ "learning_rate": 1.2907992492206752e-05,
+ "loss": 0.1229,
+ "step": 15557
+ },
+ {
+ "epoch": 42.392370572207085,
+ "grad_norm": 3.0929954051971436,
+ "learning_rate": 1.2907148124056415e-05,
+ "loss": 0.0673,
+ "step": 15558
+ },
+ {
+ "epoch": 42.39509536784741,
+ "grad_norm": 3.421421766281128,
+ "learning_rate": 1.2906303733265301e-05,
+ "loss": 0.0921,
+ "step": 15559
+ },
+ {
+ "epoch": 42.39782016348774,
+ "grad_norm": 3.937851667404175,
+ "learning_rate": 1.2905459319839988e-05,
+ "loss": 0.1022,
+ "step": 15560
+ },
+ {
+ "epoch": 42.40054495912806,
+ "grad_norm": 3.1786091327667236,
+ "learning_rate": 1.2904614883787052e-05,
+ "loss": 0.1196,
+ "step": 15561
+ },
+ {
+ "epoch": 42.403269754768395,
+ "grad_norm": 3.5089151859283447,
+ "learning_rate": 1.290377042511307e-05,
+ "loss": 0.1169,
+ "step": 15562
+ },
+ {
+ "epoch": 42.40599455040872,
+ "grad_norm": 3.599698543548584,
+ "learning_rate": 1.2902925943824615e-05,
+ "loss": 0.1403,
+ "step": 15563
+ },
+ {
+ "epoch": 42.40871934604905,
+ "grad_norm": 3.2090048789978027,
+ "learning_rate": 1.2902081439928268e-05,
+ "loss": 0.1538,
+ "step": 15564
+ },
+ {
+ "epoch": 42.41144414168937,
+ "grad_norm": 3.0599913597106934,
+ "learning_rate": 1.2901236913430604e-05,
+ "loss": 0.1738,
+ "step": 15565
+ },
+ {
+ "epoch": 42.4141689373297,
+ "grad_norm": 2.8608505725860596,
+ "learning_rate": 1.2900392364338199e-05,
+ "loss": 0.0922,
+ "step": 15566
+ },
+ {
+ "epoch": 42.416893732970024,
+ "grad_norm": 3.039937734603882,
+ "learning_rate": 1.2899547792657633e-05,
+ "loss": 0.1015,
+ "step": 15567
+ },
+ {
+ "epoch": 42.41961852861036,
+ "grad_norm": 3.2900662422180176,
+ "learning_rate": 1.289870319839548e-05,
+ "loss": 0.067,
+ "step": 15568
+ },
+ {
+ "epoch": 42.42234332425068,
+ "grad_norm": 3.022922992706299,
+ "learning_rate": 1.289785858155832e-05,
+ "loss": 0.1194,
+ "step": 15569
+ },
+ {
+ "epoch": 42.42506811989101,
+ "grad_norm": 4.156905651092529,
+ "learning_rate": 1.2897013942152732e-05,
+ "loss": 0.1239,
+ "step": 15570
+ },
+ {
+ "epoch": 42.427792915531334,
+ "grad_norm": 5.4796648025512695,
+ "learning_rate": 1.2896169280185293e-05,
+ "loss": 0.1869,
+ "step": 15571
+ },
+ {
+ "epoch": 42.43051771117166,
+ "grad_norm": 3.8929624557495117,
+ "learning_rate": 1.289532459566258e-05,
+ "loss": 0.0745,
+ "step": 15572
+ },
+ {
+ "epoch": 42.433242506811986,
+ "grad_norm": 3.4701995849609375,
+ "learning_rate": 1.2894479888591173e-05,
+ "loss": 0.1217,
+ "step": 15573
+ },
+ {
+ "epoch": 42.43596730245232,
+ "grad_norm": 2.865375280380249,
+ "learning_rate": 1.289363515897765e-05,
+ "loss": 0.1564,
+ "step": 15574
+ },
+ {
+ "epoch": 42.438692098092645,
+ "grad_norm": 3.4071178436279297,
+ "learning_rate": 1.289279040682859e-05,
+ "loss": 0.1022,
+ "step": 15575
+ },
+ {
+ "epoch": 42.44141689373297,
+ "grad_norm": 3.377262592315674,
+ "learning_rate": 1.2891945632150566e-05,
+ "loss": 0.0747,
+ "step": 15576
+ },
+ {
+ "epoch": 42.444141689373296,
+ "grad_norm": 4.8660054206848145,
+ "learning_rate": 1.2891100834950167e-05,
+ "loss": 0.1123,
+ "step": 15577
+ },
+ {
+ "epoch": 42.44686648501362,
+ "grad_norm": 3.5932297706604004,
+ "learning_rate": 1.2890256015233968e-05,
+ "loss": 0.2087,
+ "step": 15578
+ },
+ {
+ "epoch": 42.44959128065395,
+ "grad_norm": 3.504842758178711,
+ "learning_rate": 1.2889411173008549e-05,
+ "loss": 0.1034,
+ "step": 15579
+ },
+ {
+ "epoch": 42.45231607629428,
+ "grad_norm": 2.9445858001708984,
+ "learning_rate": 1.2888566308280486e-05,
+ "loss": 0.1859,
+ "step": 15580
+ },
+ {
+ "epoch": 42.45504087193461,
+ "grad_norm": 4.133998870849609,
+ "learning_rate": 1.2887721421056366e-05,
+ "loss": 0.0618,
+ "step": 15581
+ },
+ {
+ "epoch": 42.45776566757493,
+ "grad_norm": 3.549724578857422,
+ "learning_rate": 1.2886876511342758e-05,
+ "loss": 0.2557,
+ "step": 15582
+ },
+ {
+ "epoch": 42.46049046321526,
+ "grad_norm": 2.4833855628967285,
+ "learning_rate": 1.2886031579146257e-05,
+ "loss": 0.1619,
+ "step": 15583
+ },
+ {
+ "epoch": 42.463215258855584,
+ "grad_norm": 3.947458028793335,
+ "learning_rate": 1.2885186624473428e-05,
+ "loss": 0.1669,
+ "step": 15584
+ },
+ {
+ "epoch": 42.46594005449591,
+ "grad_norm": 2.973510503768921,
+ "learning_rate": 1.288434164733086e-05,
+ "loss": 0.2985,
+ "step": 15585
+ },
+ {
+ "epoch": 42.46866485013624,
+ "grad_norm": 2.0910959243774414,
+ "learning_rate": 1.2883496647725131e-05,
+ "loss": 0.0608,
+ "step": 15586
+ },
+ {
+ "epoch": 42.47138964577657,
+ "grad_norm": 3.052903890609741,
+ "learning_rate": 1.2882651625662825e-05,
+ "loss": 0.0592,
+ "step": 15587
+ },
+ {
+ "epoch": 42.474114441416894,
+ "grad_norm": 2.7107369899749756,
+ "learning_rate": 1.288180658115052e-05,
+ "loss": 0.0575,
+ "step": 15588
+ },
+ {
+ "epoch": 42.47683923705722,
+ "grad_norm": 3.22662353515625,
+ "learning_rate": 1.28809615141948e-05,
+ "loss": 0.1528,
+ "step": 15589
+ },
+ {
+ "epoch": 42.479564032697546,
+ "grad_norm": 3.1326212882995605,
+ "learning_rate": 1.2880116424802242e-05,
+ "loss": 0.2233,
+ "step": 15590
+ },
+ {
+ "epoch": 42.48228882833787,
+ "grad_norm": 3.455268383026123,
+ "learning_rate": 1.287927131297943e-05,
+ "loss": 0.2371,
+ "step": 15591
+ },
+ {
+ "epoch": 42.485013623978205,
+ "grad_norm": 4.046286106109619,
+ "learning_rate": 1.2878426178732947e-05,
+ "loss": 0.1317,
+ "step": 15592
+ },
+ {
+ "epoch": 42.48773841961853,
+ "grad_norm": 2.677583694458008,
+ "learning_rate": 1.2877581022069374e-05,
+ "loss": 0.0609,
+ "step": 15593
+ },
+ {
+ "epoch": 42.490463215258856,
+ "grad_norm": 5.978700160980225,
+ "learning_rate": 1.2876735842995291e-05,
+ "loss": 0.0908,
+ "step": 15594
+ },
+ {
+ "epoch": 42.49318801089918,
+ "grad_norm": 3.001488447189331,
+ "learning_rate": 1.2875890641517283e-05,
+ "loss": 0.1179,
+ "step": 15595
+ },
+ {
+ "epoch": 42.49591280653951,
+ "grad_norm": 4.917922496795654,
+ "learning_rate": 1.2875045417641928e-05,
+ "loss": 0.0711,
+ "step": 15596
+ },
+ {
+ "epoch": 42.49863760217983,
+ "grad_norm": 3.2776474952697754,
+ "learning_rate": 1.2874200171375817e-05,
+ "loss": 0.1578,
+ "step": 15597
+ },
+ {
+ "epoch": 42.50136239782017,
+ "grad_norm": 3.504551887512207,
+ "learning_rate": 1.2873354902725525e-05,
+ "loss": 0.1888,
+ "step": 15598
+ },
+ {
+ "epoch": 42.50408719346049,
+ "grad_norm": 3.6083579063415527,
+ "learning_rate": 1.2872509611697639e-05,
+ "loss": 0.0817,
+ "step": 15599
+ },
+ {
+ "epoch": 42.50681198910082,
+ "grad_norm": 3.110363006591797,
+ "learning_rate": 1.287166429829874e-05,
+ "loss": 0.0995,
+ "step": 15600
+ },
+ {
+ "epoch": 42.509536784741144,
+ "grad_norm": 3.735283613204956,
+ "learning_rate": 1.2870818962535412e-05,
+ "loss": 0.0964,
+ "step": 15601
+ },
+ {
+ "epoch": 42.51226158038147,
+ "grad_norm": 4.08259391784668,
+ "learning_rate": 1.2869973604414237e-05,
+ "loss": 0.1292,
+ "step": 15602
+ },
+ {
+ "epoch": 42.514986376021795,
+ "grad_norm": 3.973139524459839,
+ "learning_rate": 1.28691282239418e-05,
+ "loss": 0.0902,
+ "step": 15603
+ },
+ {
+ "epoch": 42.51771117166213,
+ "grad_norm": 2.8776164054870605,
+ "learning_rate": 1.2868282821124685e-05,
+ "loss": 0.181,
+ "step": 15604
+ },
+ {
+ "epoch": 42.520435967302454,
+ "grad_norm": 2.1923294067382812,
+ "learning_rate": 1.2867437395969476e-05,
+ "loss": 0.0427,
+ "step": 15605
+ },
+ {
+ "epoch": 42.52316076294278,
+ "grad_norm": 3.1001341342926025,
+ "learning_rate": 1.2866591948482756e-05,
+ "loss": 0.1143,
+ "step": 15606
+ },
+ {
+ "epoch": 42.525885558583106,
+ "grad_norm": 3.025310516357422,
+ "learning_rate": 1.286574647867111e-05,
+ "loss": 0.0598,
+ "step": 15607
+ },
+ {
+ "epoch": 42.52861035422343,
+ "grad_norm": 3.074608564376831,
+ "learning_rate": 1.2864900986541125e-05,
+ "loss": 0.0872,
+ "step": 15608
+ },
+ {
+ "epoch": 42.53133514986376,
+ "grad_norm": 3.2438204288482666,
+ "learning_rate": 1.2864055472099381e-05,
+ "loss": 0.1342,
+ "step": 15609
+ },
+ {
+ "epoch": 42.53405994550409,
+ "grad_norm": 11.468170166015625,
+ "learning_rate": 1.2863209935352468e-05,
+ "loss": 0.1358,
+ "step": 15610
+ },
+ {
+ "epoch": 42.536784741144416,
+ "grad_norm": 2.8617279529571533,
+ "learning_rate": 1.2862364376306964e-05,
+ "loss": 0.057,
+ "step": 15611
+ },
+ {
+ "epoch": 42.53950953678474,
+ "grad_norm": 5.289137840270996,
+ "learning_rate": 1.2861518794969464e-05,
+ "loss": 0.3259,
+ "step": 15612
+ },
+ {
+ "epoch": 42.54223433242507,
+ "grad_norm": 3.3815901279449463,
+ "learning_rate": 1.2860673191346542e-05,
+ "loss": 0.2107,
+ "step": 15613
+ },
+ {
+ "epoch": 42.54495912806539,
+ "grad_norm": 3.522883653640747,
+ "learning_rate": 1.2859827565444793e-05,
+ "loss": 0.0758,
+ "step": 15614
+ },
+ {
+ "epoch": 42.54768392370572,
+ "grad_norm": 5.207448959350586,
+ "learning_rate": 1.2858981917270796e-05,
+ "loss": 0.1188,
+ "step": 15615
+ },
+ {
+ "epoch": 42.55040871934605,
+ "grad_norm": 3.581106662750244,
+ "learning_rate": 1.2858136246831144e-05,
+ "loss": 0.2223,
+ "step": 15616
+ },
+ {
+ "epoch": 42.55313351498638,
+ "grad_norm": 3.1284468173980713,
+ "learning_rate": 1.2857290554132416e-05,
+ "loss": 0.1693,
+ "step": 15617
+ },
+ {
+ "epoch": 42.555858310626704,
+ "grad_norm": 2.7940759658813477,
+ "learning_rate": 1.2856444839181199e-05,
+ "loss": 0.0865,
+ "step": 15618
+ },
+ {
+ "epoch": 42.55858310626703,
+ "grad_norm": 3.389892578125,
+ "learning_rate": 1.2855599101984085e-05,
+ "loss": 0.0904,
+ "step": 15619
+ },
+ {
+ "epoch": 42.561307901907355,
+ "grad_norm": 3.493091106414795,
+ "learning_rate": 1.2854753342547656e-05,
+ "loss": 0.1449,
+ "step": 15620
+ },
+ {
+ "epoch": 42.56403269754768,
+ "grad_norm": 3.64518666267395,
+ "learning_rate": 1.2853907560878499e-05,
+ "loss": 0.081,
+ "step": 15621
+ },
+ {
+ "epoch": 42.566757493188014,
+ "grad_norm": 4.063978672027588,
+ "learning_rate": 1.2853061756983203e-05,
+ "loss": 0.1721,
+ "step": 15622
+ },
+ {
+ "epoch": 42.56948228882834,
+ "grad_norm": 2.6713926792144775,
+ "learning_rate": 1.2852215930868353e-05,
+ "loss": 0.1067,
+ "step": 15623
+ },
+ {
+ "epoch": 42.572207084468666,
+ "grad_norm": 2.954101085662842,
+ "learning_rate": 1.285137008254054e-05,
+ "loss": 0.1165,
+ "step": 15624
+ },
+ {
+ "epoch": 42.57493188010899,
+ "grad_norm": 2.998579740524292,
+ "learning_rate": 1.2850524212006344e-05,
+ "loss": 0.2266,
+ "step": 15625
+ },
+ {
+ "epoch": 42.57765667574932,
+ "grad_norm": 3.7928199768066406,
+ "learning_rate": 1.284967831927236e-05,
+ "loss": 0.1764,
+ "step": 15626
+ },
+ {
+ "epoch": 42.58038147138964,
+ "grad_norm": 3.568537712097168,
+ "learning_rate": 1.2848832404345173e-05,
+ "loss": 0.0766,
+ "step": 15627
+ },
+ {
+ "epoch": 42.583106267029976,
+ "grad_norm": 42.84107208251953,
+ "learning_rate": 1.2847986467231368e-05,
+ "loss": 0.1905,
+ "step": 15628
+ },
+ {
+ "epoch": 42.5858310626703,
+ "grad_norm": 2.9747281074523926,
+ "learning_rate": 1.2847140507937539e-05,
+ "loss": 0.3382,
+ "step": 15629
+ },
+ {
+ "epoch": 42.58855585831063,
+ "grad_norm": 4.041215896606445,
+ "learning_rate": 1.2846294526470274e-05,
+ "loss": 0.152,
+ "step": 15630
+ },
+ {
+ "epoch": 42.59128065395095,
+ "grad_norm": 3.173274517059326,
+ "learning_rate": 1.2845448522836152e-05,
+ "loss": 0.0736,
+ "step": 15631
+ },
+ {
+ "epoch": 42.59400544959128,
+ "grad_norm": 3.667750358581543,
+ "learning_rate": 1.2844602497041774e-05,
+ "loss": 0.0787,
+ "step": 15632
+ },
+ {
+ "epoch": 42.596730245231605,
+ "grad_norm": 3.8761789798736572,
+ "learning_rate": 1.2843756449093721e-05,
+ "loss": 0.141,
+ "step": 15633
+ },
+ {
+ "epoch": 42.59945504087194,
+ "grad_norm": 3.9558064937591553,
+ "learning_rate": 1.2842910378998585e-05,
+ "loss": 0.2084,
+ "step": 15634
+ },
+ {
+ "epoch": 42.60217983651226,
+ "grad_norm": 3.862759828567505,
+ "learning_rate": 1.2842064286762954e-05,
+ "loss": 0.1498,
+ "step": 15635
+ },
+ {
+ "epoch": 42.60490463215259,
+ "grad_norm": 4.234774112701416,
+ "learning_rate": 1.2841218172393417e-05,
+ "loss": 0.0792,
+ "step": 15636
+ },
+ {
+ "epoch": 42.607629427792915,
+ "grad_norm": 3.768277168273926,
+ "learning_rate": 1.2840372035896565e-05,
+ "loss": 0.2659,
+ "step": 15637
+ },
+ {
+ "epoch": 42.61035422343324,
+ "grad_norm": 3.3350706100463867,
+ "learning_rate": 1.2839525877278986e-05,
+ "loss": 0.2289,
+ "step": 15638
+ },
+ {
+ "epoch": 42.61307901907357,
+ "grad_norm": 5.5832839012146,
+ "learning_rate": 1.2838679696547272e-05,
+ "loss": 0.1292,
+ "step": 15639
+ },
+ {
+ "epoch": 42.6158038147139,
+ "grad_norm": 3.373760461807251,
+ "learning_rate": 1.2837833493708013e-05,
+ "loss": 0.1444,
+ "step": 15640
+ },
+ {
+ "epoch": 42.618528610354225,
+ "grad_norm": 3.76164174079895,
+ "learning_rate": 1.2836987268767798e-05,
+ "loss": 0.1579,
+ "step": 15641
+ },
+ {
+ "epoch": 42.62125340599455,
+ "grad_norm": 5.27699613571167,
+ "learning_rate": 1.2836141021733217e-05,
+ "loss": 0.1357,
+ "step": 15642
+ },
+ {
+ "epoch": 42.62397820163488,
+ "grad_norm": 3.96787428855896,
+ "learning_rate": 1.283529475261086e-05,
+ "loss": 0.14,
+ "step": 15643
+ },
+ {
+ "epoch": 42.6267029972752,
+ "grad_norm": 3.462932586669922,
+ "learning_rate": 1.283444846140732e-05,
+ "loss": 0.1275,
+ "step": 15644
+ },
+ {
+ "epoch": 42.62942779291553,
+ "grad_norm": 3.557356595993042,
+ "learning_rate": 1.2833602148129184e-05,
+ "loss": 0.0925,
+ "step": 15645
+ },
+ {
+ "epoch": 42.63215258855586,
+ "grad_norm": 4.582718372344971,
+ "learning_rate": 1.2832755812783048e-05,
+ "loss": 0.2036,
+ "step": 15646
+ },
+ {
+ "epoch": 42.63487738419619,
+ "grad_norm": 4.599499702453613,
+ "learning_rate": 1.28319094553755e-05,
+ "loss": 0.0687,
+ "step": 15647
+ },
+ {
+ "epoch": 42.63760217983651,
+ "grad_norm": 3.564931631088257,
+ "learning_rate": 1.2831063075913134e-05,
+ "loss": 0.1635,
+ "step": 15648
+ },
+ {
+ "epoch": 42.64032697547684,
+ "grad_norm": 3.437415361404419,
+ "learning_rate": 1.2830216674402539e-05,
+ "loss": 0.1398,
+ "step": 15649
+ },
+ {
+ "epoch": 42.643051771117165,
+ "grad_norm": 4.013363361358643,
+ "learning_rate": 1.2829370250850306e-05,
+ "loss": 0.3138,
+ "step": 15650
+ },
+ {
+ "epoch": 42.64577656675749,
+ "grad_norm": 4.795948028564453,
+ "learning_rate": 1.282852380526303e-05,
+ "loss": 0.1433,
+ "step": 15651
+ },
+ {
+ "epoch": 42.64850136239782,
+ "grad_norm": 4.70443058013916,
+ "learning_rate": 1.2827677337647302e-05,
+ "loss": 0.0904,
+ "step": 15652
+ },
+ {
+ "epoch": 42.65122615803815,
+ "grad_norm": 3.9685912132263184,
+ "learning_rate": 1.2826830848009711e-05,
+ "loss": 0.0921,
+ "step": 15653
+ },
+ {
+ "epoch": 42.653950953678475,
+ "grad_norm": 3.4378881454467773,
+ "learning_rate": 1.2825984336356856e-05,
+ "loss": 0.1552,
+ "step": 15654
+ },
+ {
+ "epoch": 42.6566757493188,
+ "grad_norm": 3.466538667678833,
+ "learning_rate": 1.2825137802695324e-05,
+ "loss": 0.0814,
+ "step": 15655
+ },
+ {
+ "epoch": 42.65940054495913,
+ "grad_norm": 4.2976155281066895,
+ "learning_rate": 1.282429124703171e-05,
+ "loss": 0.0812,
+ "step": 15656
+ },
+ {
+ "epoch": 42.66212534059945,
+ "grad_norm": 5.214032173156738,
+ "learning_rate": 1.2823444669372606e-05,
+ "loss": 0.079,
+ "step": 15657
+ },
+ {
+ "epoch": 42.664850136239785,
+ "grad_norm": 3.481034994125366,
+ "learning_rate": 1.282259806972461e-05,
+ "loss": 0.1916,
+ "step": 15658
+ },
+ {
+ "epoch": 42.66757493188011,
+ "grad_norm": 4.315105438232422,
+ "learning_rate": 1.2821751448094306e-05,
+ "loss": 0.2197,
+ "step": 15659
+ },
+ {
+ "epoch": 42.67029972752044,
+ "grad_norm": 3.409212112426758,
+ "learning_rate": 1.2820904804488297e-05,
+ "loss": 0.1275,
+ "step": 15660
+ },
+ {
+ "epoch": 42.67302452316076,
+ "grad_norm": 3.0105443000793457,
+ "learning_rate": 1.2820058138913169e-05,
+ "loss": 0.1003,
+ "step": 15661
+ },
+ {
+ "epoch": 42.67574931880109,
+ "grad_norm": 3.894927501678467,
+ "learning_rate": 1.2819211451375517e-05,
+ "loss": 0.1499,
+ "step": 15662
+ },
+ {
+ "epoch": 42.678474114441414,
+ "grad_norm": 3.7467517852783203,
+ "learning_rate": 1.2818364741881939e-05,
+ "loss": 0.1193,
+ "step": 15663
+ },
+ {
+ "epoch": 42.68119891008175,
+ "grad_norm": 3.687185525894165,
+ "learning_rate": 1.281751801043903e-05,
+ "loss": 0.1815,
+ "step": 15664
+ },
+ {
+ "epoch": 42.68392370572207,
+ "grad_norm": 4.072238445281982,
+ "learning_rate": 1.2816671257053377e-05,
+ "loss": 0.1129,
+ "step": 15665
+ },
+ {
+ "epoch": 42.6866485013624,
+ "grad_norm": 3.568232297897339,
+ "learning_rate": 1.2815824481731582e-05,
+ "loss": 0.0849,
+ "step": 15666
+ },
+ {
+ "epoch": 42.689373297002724,
+ "grad_norm": 2.9845962524414062,
+ "learning_rate": 1.2814977684480234e-05,
+ "loss": 0.2036,
+ "step": 15667
+ },
+ {
+ "epoch": 42.69209809264305,
+ "grad_norm": 4.523996353149414,
+ "learning_rate": 1.2814130865305937e-05,
+ "loss": 0.1252,
+ "step": 15668
+ },
+ {
+ "epoch": 42.694822888283376,
+ "grad_norm": 4.280789375305176,
+ "learning_rate": 1.2813284024215272e-05,
+ "loss": 0.0889,
+ "step": 15669
+ },
+ {
+ "epoch": 42.69754768392371,
+ "grad_norm": 4.382587432861328,
+ "learning_rate": 1.2812437161214843e-05,
+ "loss": 0.2092,
+ "step": 15670
+ },
+ {
+ "epoch": 42.700272479564035,
+ "grad_norm": 3.025296211242676,
+ "learning_rate": 1.2811590276311243e-05,
+ "loss": 0.2036,
+ "step": 15671
+ },
+ {
+ "epoch": 42.70299727520436,
+ "grad_norm": 2.6724941730499268,
+ "learning_rate": 1.2810743369511071e-05,
+ "loss": 0.1045,
+ "step": 15672
+ },
+ {
+ "epoch": 42.705722070844686,
+ "grad_norm": 3.499209403991699,
+ "learning_rate": 1.280989644082092e-05,
+ "loss": 0.0607,
+ "step": 15673
+ },
+ {
+ "epoch": 42.70844686648501,
+ "grad_norm": 4.296411514282227,
+ "learning_rate": 1.2809049490247386e-05,
+ "loss": 0.1481,
+ "step": 15674
+ },
+ {
+ "epoch": 42.71117166212534,
+ "grad_norm": 3.8108842372894287,
+ "learning_rate": 1.2808202517797064e-05,
+ "loss": 0.1941,
+ "step": 15675
+ },
+ {
+ "epoch": 42.71389645776567,
+ "grad_norm": 3.182464122772217,
+ "learning_rate": 1.2807355523476552e-05,
+ "loss": 0.1829,
+ "step": 15676
+ },
+ {
+ "epoch": 42.716621253406,
+ "grad_norm": 4.532083034515381,
+ "learning_rate": 1.2806508507292444e-05,
+ "loss": 0.1033,
+ "step": 15677
+ },
+ {
+ "epoch": 42.71934604904632,
+ "grad_norm": 2.6337804794311523,
+ "learning_rate": 1.280566146925134e-05,
+ "loss": 0.0774,
+ "step": 15678
+ },
+ {
+ "epoch": 42.72207084468665,
+ "grad_norm": 3.3819549083709717,
+ "learning_rate": 1.2804814409359834e-05,
+ "loss": 0.0817,
+ "step": 15679
+ },
+ {
+ "epoch": 42.724795640326974,
+ "grad_norm": 5.344264030456543,
+ "learning_rate": 1.2803967327624522e-05,
+ "loss": 0.069,
+ "step": 15680
+ },
+ {
+ "epoch": 42.7275204359673,
+ "grad_norm": 3.6134543418884277,
+ "learning_rate": 1.2803120224052004e-05,
+ "loss": 0.0776,
+ "step": 15681
+ },
+ {
+ "epoch": 42.73024523160763,
+ "grad_norm": 3.5008225440979004,
+ "learning_rate": 1.2802273098648876e-05,
+ "loss": 0.0936,
+ "step": 15682
+ },
+ {
+ "epoch": 42.73297002724796,
+ "grad_norm": 3.000621795654297,
+ "learning_rate": 1.2801425951421736e-05,
+ "loss": 0.0881,
+ "step": 15683
+ },
+ {
+ "epoch": 42.735694822888284,
+ "grad_norm": 3.4330177307128906,
+ "learning_rate": 1.280057878237718e-05,
+ "loss": 0.131,
+ "step": 15684
+ },
+ {
+ "epoch": 42.73841961852861,
+ "grad_norm": 3.7648119926452637,
+ "learning_rate": 1.2799731591521805e-05,
+ "loss": 0.1683,
+ "step": 15685
+ },
+ {
+ "epoch": 42.741144414168936,
+ "grad_norm": 3.5383880138397217,
+ "learning_rate": 1.2798884378862213e-05,
+ "loss": 0.1245,
+ "step": 15686
+ },
+ {
+ "epoch": 42.74386920980926,
+ "grad_norm": 4.2039103507995605,
+ "learning_rate": 1.2798037144405e-05,
+ "loss": 0.0889,
+ "step": 15687
+ },
+ {
+ "epoch": 42.746594005449595,
+ "grad_norm": 3.0618245601654053,
+ "learning_rate": 1.2797189888156761e-05,
+ "loss": 0.078,
+ "step": 15688
+ },
+ {
+ "epoch": 42.74931880108992,
+ "grad_norm": 3.259033441543579,
+ "learning_rate": 1.27963426101241e-05,
+ "loss": 0.0602,
+ "step": 15689
+ },
+ {
+ "epoch": 42.752043596730246,
+ "grad_norm": 2.8273355960845947,
+ "learning_rate": 1.279549531031361e-05,
+ "loss": 0.0772,
+ "step": 15690
+ },
+ {
+ "epoch": 42.75476839237057,
+ "grad_norm": 3.491328001022339,
+ "learning_rate": 1.2794647988731896e-05,
+ "loss": 0.0989,
+ "step": 15691
+ },
+ {
+ "epoch": 42.7574931880109,
+ "grad_norm": 3.6088616847991943,
+ "learning_rate": 1.279380064538555e-05,
+ "loss": 0.1329,
+ "step": 15692
+ },
+ {
+ "epoch": 42.76021798365122,
+ "grad_norm": 2.9256339073181152,
+ "learning_rate": 1.2792953280281175e-05,
+ "loss": 0.0791,
+ "step": 15693
+ },
+ {
+ "epoch": 42.762942779291556,
+ "grad_norm": 3.362100839614868,
+ "learning_rate": 1.2792105893425373e-05,
+ "loss": 0.0696,
+ "step": 15694
+ },
+ {
+ "epoch": 42.76566757493188,
+ "grad_norm": 2.8778958320617676,
+ "learning_rate": 1.2791258484824736e-05,
+ "loss": 0.0671,
+ "step": 15695
+ },
+ {
+ "epoch": 42.76839237057221,
+ "grad_norm": 5.010435581207275,
+ "learning_rate": 1.279041105448587e-05,
+ "loss": 0.2583,
+ "step": 15696
+ },
+ {
+ "epoch": 42.771117166212534,
+ "grad_norm": 3.950134754180908,
+ "learning_rate": 1.278956360241537e-05,
+ "loss": 0.1117,
+ "step": 15697
+ },
+ {
+ "epoch": 42.77384196185286,
+ "grad_norm": 3.3407630920410156,
+ "learning_rate": 1.2788716128619843e-05,
+ "loss": 0.065,
+ "step": 15698
+ },
+ {
+ "epoch": 42.776566757493185,
+ "grad_norm": 3.0507709980010986,
+ "learning_rate": 1.2787868633105882e-05,
+ "loss": 0.2301,
+ "step": 15699
+ },
+ {
+ "epoch": 42.77929155313352,
+ "grad_norm": 3.1347591876983643,
+ "learning_rate": 1.278702111588009e-05,
+ "loss": 0.2372,
+ "step": 15700
+ },
+ {
+ "epoch": 42.782016348773844,
+ "grad_norm": 3.1317338943481445,
+ "learning_rate": 1.2786173576949068e-05,
+ "loss": 0.061,
+ "step": 15701
+ },
+ {
+ "epoch": 42.78474114441417,
+ "grad_norm": 4.6421427726745605,
+ "learning_rate": 1.2785326016319415e-05,
+ "loss": 0.0861,
+ "step": 15702
+ },
+ {
+ "epoch": 42.787465940054496,
+ "grad_norm": 3.0677013397216797,
+ "learning_rate": 1.2784478433997733e-05,
+ "loss": 0.0918,
+ "step": 15703
+ },
+ {
+ "epoch": 42.79019073569482,
+ "grad_norm": 3.449274778366089,
+ "learning_rate": 1.278363082999062e-05,
+ "loss": 0.2616,
+ "step": 15704
+ },
+ {
+ "epoch": 42.79291553133515,
+ "grad_norm": 3.220302104949951,
+ "learning_rate": 1.2782783204304683e-05,
+ "loss": 0.1598,
+ "step": 15705
+ },
+ {
+ "epoch": 42.79564032697548,
+ "grad_norm": 3.9886767864227295,
+ "learning_rate": 1.2781935556946517e-05,
+ "loss": 0.1391,
+ "step": 15706
+ },
+ {
+ "epoch": 42.798365122615806,
+ "grad_norm": 3.1909170150756836,
+ "learning_rate": 1.2781087887922729e-05,
+ "loss": 0.0817,
+ "step": 15707
+ },
+ {
+ "epoch": 42.80108991825613,
+ "grad_norm": 3.227504253387451,
+ "learning_rate": 1.2780240197239917e-05,
+ "loss": 0.0743,
+ "step": 15708
+ },
+ {
+ "epoch": 42.80381471389646,
+ "grad_norm": 4.358556747436523,
+ "learning_rate": 1.2779392484904685e-05,
+ "loss": 0.0875,
+ "step": 15709
+ },
+ {
+ "epoch": 42.80653950953678,
+ "grad_norm": 7.293618679046631,
+ "learning_rate": 1.2778544750923634e-05,
+ "loss": 0.0677,
+ "step": 15710
+ },
+ {
+ "epoch": 42.80926430517711,
+ "grad_norm": 3.8990023136138916,
+ "learning_rate": 1.2777696995303366e-05,
+ "loss": 0.1578,
+ "step": 15711
+ },
+ {
+ "epoch": 42.81198910081744,
+ "grad_norm": 3.2969536781311035,
+ "learning_rate": 1.277684921805048e-05,
+ "loss": 0.1289,
+ "step": 15712
+ },
+ {
+ "epoch": 42.81471389645777,
+ "grad_norm": 3.7783255577087402,
+ "learning_rate": 1.2776001419171585e-05,
+ "loss": 0.2169,
+ "step": 15713
+ },
+ {
+ "epoch": 42.817438692098094,
+ "grad_norm": 3.5159060955047607,
+ "learning_rate": 1.2775153598673276e-05,
+ "loss": 0.1911,
+ "step": 15714
+ },
+ {
+ "epoch": 42.82016348773842,
+ "grad_norm": 3.9097254276275635,
+ "learning_rate": 1.2774305756562164e-05,
+ "loss": 0.1572,
+ "step": 15715
+ },
+ {
+ "epoch": 42.822888283378745,
+ "grad_norm": 2.9993162155151367,
+ "learning_rate": 1.2773457892844844e-05,
+ "loss": 0.0759,
+ "step": 15716
+ },
+ {
+ "epoch": 42.82561307901907,
+ "grad_norm": 3.846860885620117,
+ "learning_rate": 1.277261000752793e-05,
+ "loss": 0.2268,
+ "step": 15717
+ },
+ {
+ "epoch": 42.828337874659404,
+ "grad_norm": 4.74245023727417,
+ "learning_rate": 1.2771762100618012e-05,
+ "loss": 0.2138,
+ "step": 15718
+ },
+ {
+ "epoch": 42.83106267029973,
+ "grad_norm": 3.1613078117370605,
+ "learning_rate": 1.2770914172121706e-05,
+ "loss": 0.1433,
+ "step": 15719
+ },
+ {
+ "epoch": 42.833787465940055,
+ "grad_norm": 4.3296284675598145,
+ "learning_rate": 1.2770066222045605e-05,
+ "loss": 0.1042,
+ "step": 15720
+ },
+ {
+ "epoch": 42.83651226158038,
+ "grad_norm": 3.0764243602752686,
+ "learning_rate": 1.276921825039632e-05,
+ "loss": 0.1377,
+ "step": 15721
+ },
+ {
+ "epoch": 42.83923705722071,
+ "grad_norm": 4.115959644317627,
+ "learning_rate": 1.2768370257180449e-05,
+ "loss": 0.2376,
+ "step": 15722
+ },
+ {
+ "epoch": 42.84196185286103,
+ "grad_norm": 3.711365222930908,
+ "learning_rate": 1.2767522242404602e-05,
+ "loss": 0.0947,
+ "step": 15723
+ },
+ {
+ "epoch": 42.844686648501366,
+ "grad_norm": 5.139861106872559,
+ "learning_rate": 1.276667420607538e-05,
+ "loss": 0.2299,
+ "step": 15724
+ },
+ {
+ "epoch": 42.84741144414169,
+ "grad_norm": 3.1659913063049316,
+ "learning_rate": 1.276582614819939e-05,
+ "loss": 0.2277,
+ "step": 15725
+ },
+ {
+ "epoch": 42.85013623978202,
+ "grad_norm": 3.9636480808258057,
+ "learning_rate": 1.2764978068783233e-05,
+ "loss": 0.2659,
+ "step": 15726
+ },
+ {
+ "epoch": 42.85286103542234,
+ "grad_norm": 3.5439350605010986,
+ "learning_rate": 1.2764129967833518e-05,
+ "loss": 0.1103,
+ "step": 15727
+ },
+ {
+ "epoch": 42.85558583106267,
+ "grad_norm": 2.941293954849243,
+ "learning_rate": 1.2763281845356845e-05,
+ "loss": 0.0827,
+ "step": 15728
+ },
+ {
+ "epoch": 42.858310626702995,
+ "grad_norm": 10.495644569396973,
+ "learning_rate": 1.2762433701359827e-05,
+ "loss": 0.1856,
+ "step": 15729
+ },
+ {
+ "epoch": 42.86103542234333,
+ "grad_norm": 3.0569467544555664,
+ "learning_rate": 1.2761585535849059e-05,
+ "loss": 0.1258,
+ "step": 15730
+ },
+ {
+ "epoch": 42.86376021798365,
+ "grad_norm": 4.529148578643799,
+ "learning_rate": 1.2760737348831152e-05,
+ "loss": 0.1576,
+ "step": 15731
+ },
+ {
+ "epoch": 42.86648501362398,
+ "grad_norm": 3.201810598373413,
+ "learning_rate": 1.2759889140312711e-05,
+ "loss": 0.2328,
+ "step": 15732
+ },
+ {
+ "epoch": 42.869209809264305,
+ "grad_norm": 2.7349746227264404,
+ "learning_rate": 1.2759040910300346e-05,
+ "loss": 0.0511,
+ "step": 15733
+ },
+ {
+ "epoch": 42.87193460490463,
+ "grad_norm": 4.032858848571777,
+ "learning_rate": 1.2758192658800657e-05,
+ "loss": 0.1798,
+ "step": 15734
+ },
+ {
+ "epoch": 42.87465940054496,
+ "grad_norm": 3.663466453552246,
+ "learning_rate": 1.2757344385820254e-05,
+ "loss": 0.1623,
+ "step": 15735
+ },
+ {
+ "epoch": 42.87738419618529,
+ "grad_norm": 3.5051097869873047,
+ "learning_rate": 1.2756496091365737e-05,
+ "loss": 0.1383,
+ "step": 15736
+ },
+ {
+ "epoch": 42.880108991825615,
+ "grad_norm": 3.9779505729675293,
+ "learning_rate": 1.2755647775443722e-05,
+ "loss": 0.1102,
+ "step": 15737
+ },
+ {
+ "epoch": 42.88283378746594,
+ "grad_norm": 5.55609655380249,
+ "learning_rate": 1.275479943806081e-05,
+ "loss": 0.1023,
+ "step": 15738
+ },
+ {
+ "epoch": 42.88555858310627,
+ "grad_norm": 3.2811455726623535,
+ "learning_rate": 1.2753951079223606e-05,
+ "loss": 0.1289,
+ "step": 15739
+ },
+ {
+ "epoch": 42.88828337874659,
+ "grad_norm": 2.606860637664795,
+ "learning_rate": 1.275310269893872e-05,
+ "loss": 0.086,
+ "step": 15740
+ },
+ {
+ "epoch": 42.89100817438692,
+ "grad_norm": 3.3573012351989746,
+ "learning_rate": 1.2752254297212762e-05,
+ "loss": 0.1068,
+ "step": 15741
+ },
+ {
+ "epoch": 42.89373297002725,
+ "grad_norm": 8.537073135375977,
+ "learning_rate": 1.2751405874052334e-05,
+ "loss": 0.1743,
+ "step": 15742
+ },
+ {
+ "epoch": 42.89645776566758,
+ "grad_norm": 3.5040316581726074,
+ "learning_rate": 1.2750557429464046e-05,
+ "loss": 0.1116,
+ "step": 15743
+ },
+ {
+ "epoch": 42.8991825613079,
+ "grad_norm": 4.379354953765869,
+ "learning_rate": 1.2749708963454507e-05,
+ "loss": 0.1341,
+ "step": 15744
+ },
+ {
+ "epoch": 42.90190735694823,
+ "grad_norm": 3.5129010677337646,
+ "learning_rate": 1.274886047603032e-05,
+ "loss": 0.1215,
+ "step": 15745
+ },
+ {
+ "epoch": 42.904632152588555,
+ "grad_norm": 5.493027687072754,
+ "learning_rate": 1.2748011967198097e-05,
+ "loss": 0.1568,
+ "step": 15746
+ },
+ {
+ "epoch": 42.90735694822888,
+ "grad_norm": 3.393096685409546,
+ "learning_rate": 1.2747163436964447e-05,
+ "loss": 0.104,
+ "step": 15747
+ },
+ {
+ "epoch": 42.91008174386921,
+ "grad_norm": 3.7406630516052246,
+ "learning_rate": 1.274631488533597e-05,
+ "loss": 0.0947,
+ "step": 15748
+ },
+ {
+ "epoch": 42.91280653950954,
+ "grad_norm": 3.891613245010376,
+ "learning_rate": 1.274546631231929e-05,
+ "loss": 0.1769,
+ "step": 15749
+ },
+ {
+ "epoch": 42.915531335149865,
+ "grad_norm": 5.533442497253418,
+ "learning_rate": 1.2744617717921002e-05,
+ "loss": 0.1626,
+ "step": 15750
+ },
+ {
+ "epoch": 42.91825613079019,
+ "grad_norm": 2.9155290126800537,
+ "learning_rate": 1.274376910214772e-05,
+ "loss": 0.0815,
+ "step": 15751
+ },
+ {
+ "epoch": 42.920980926430516,
+ "grad_norm": 3.2954301834106445,
+ "learning_rate": 1.2742920465006052e-05,
+ "loss": 0.1571,
+ "step": 15752
+ },
+ {
+ "epoch": 42.92370572207084,
+ "grad_norm": 2.8164684772491455,
+ "learning_rate": 1.2742071806502612e-05,
+ "loss": 0.0553,
+ "step": 15753
+ },
+ {
+ "epoch": 42.926430517711175,
+ "grad_norm": 4.154541492462158,
+ "learning_rate": 1.2741223126644e-05,
+ "loss": 0.0929,
+ "step": 15754
+ },
+ {
+ "epoch": 42.9291553133515,
+ "grad_norm": 3.486121892929077,
+ "learning_rate": 1.2740374425436834e-05,
+ "loss": 0.1297,
+ "step": 15755
+ },
+ {
+ "epoch": 42.93188010899183,
+ "grad_norm": 2.824097156524658,
+ "learning_rate": 1.2739525702887718e-05,
+ "loss": 0.1011,
+ "step": 15756
+ },
+ {
+ "epoch": 42.93460490463215,
+ "grad_norm": 4.209000110626221,
+ "learning_rate": 1.2738676959003262e-05,
+ "loss": 0.1812,
+ "step": 15757
+ },
+ {
+ "epoch": 42.93732970027248,
+ "grad_norm": 3.2499840259552,
+ "learning_rate": 1.2737828193790081e-05,
+ "loss": 0.2145,
+ "step": 15758
+ },
+ {
+ "epoch": 42.940054495912804,
+ "grad_norm": 4.717215538024902,
+ "learning_rate": 1.2736979407254782e-05,
+ "loss": 0.1412,
+ "step": 15759
+ },
+ {
+ "epoch": 42.94277929155314,
+ "grad_norm": 4.245742321014404,
+ "learning_rate": 1.2736130599403973e-05,
+ "loss": 0.1278,
+ "step": 15760
+ },
+ {
+ "epoch": 42.94550408719346,
+ "grad_norm": 3.6342246532440186,
+ "learning_rate": 1.2735281770244272e-05,
+ "loss": 0.0952,
+ "step": 15761
+ },
+ {
+ "epoch": 42.94822888283379,
+ "grad_norm": 3.2619898319244385,
+ "learning_rate": 1.273443291978228e-05,
+ "loss": 0.0899,
+ "step": 15762
+ },
+ {
+ "epoch": 42.950953678474114,
+ "grad_norm": 6.385715961456299,
+ "learning_rate": 1.2733584048024614e-05,
+ "loss": 0.0693,
+ "step": 15763
+ },
+ {
+ "epoch": 42.95367847411444,
+ "grad_norm": 5.318986415863037,
+ "learning_rate": 1.2732735154977885e-05,
+ "loss": 0.2213,
+ "step": 15764
+ },
+ {
+ "epoch": 42.956403269754766,
+ "grad_norm": 3.47227144241333,
+ "learning_rate": 1.2731886240648699e-05,
+ "loss": 0.2185,
+ "step": 15765
+ },
+ {
+ "epoch": 42.95912806539509,
+ "grad_norm": 3.9290716648101807,
+ "learning_rate": 1.2731037305043675e-05,
+ "loss": 0.11,
+ "step": 15766
+ },
+ {
+ "epoch": 42.961852861035425,
+ "grad_norm": 15.447677612304688,
+ "learning_rate": 1.2730188348169416e-05,
+ "loss": 0.0906,
+ "step": 15767
+ },
+ {
+ "epoch": 42.96457765667575,
+ "grad_norm": 2.6861648559570312,
+ "learning_rate": 1.272933937003254e-05,
+ "loss": 0.0723,
+ "step": 15768
+ },
+ {
+ "epoch": 42.967302452316076,
+ "grad_norm": 14.05892276763916,
+ "learning_rate": 1.2728490370639659e-05,
+ "loss": 0.1464,
+ "step": 15769
+ },
+ {
+ "epoch": 42.9700272479564,
+ "grad_norm": 5.244617938995361,
+ "learning_rate": 1.2727641349997379e-05,
+ "loss": 0.1579,
+ "step": 15770
+ },
+ {
+ "epoch": 42.97275204359673,
+ "grad_norm": 4.4247846603393555,
+ "learning_rate": 1.2726792308112319e-05,
+ "loss": 0.1498,
+ "step": 15771
+ },
+ {
+ "epoch": 42.97547683923706,
+ "grad_norm": 2.3100485801696777,
+ "learning_rate": 1.2725943244991089e-05,
+ "loss": 0.0588,
+ "step": 15772
+ },
+ {
+ "epoch": 42.97820163487739,
+ "grad_norm": 3.6390206813812256,
+ "learning_rate": 1.27250941606403e-05,
+ "loss": 0.114,
+ "step": 15773
+ },
+ {
+ "epoch": 42.98092643051771,
+ "grad_norm": 11.339212417602539,
+ "learning_rate": 1.2724245055066561e-05,
+ "loss": 0.0947,
+ "step": 15774
+ },
+ {
+ "epoch": 42.98365122615804,
+ "grad_norm": 4.238719940185547,
+ "learning_rate": 1.2723395928276492e-05,
+ "loss": 0.1464,
+ "step": 15775
+ },
+ {
+ "epoch": 42.986376021798364,
+ "grad_norm": 3.529782295227051,
+ "learning_rate": 1.2722546780276705e-05,
+ "loss": 0.07,
+ "step": 15776
+ },
+ {
+ "epoch": 42.98910081743869,
+ "grad_norm": 5.91493558883667,
+ "learning_rate": 1.272169761107381e-05,
+ "loss": 0.177,
+ "step": 15777
+ },
+ {
+ "epoch": 42.991825613079016,
+ "grad_norm": 4.44365119934082,
+ "learning_rate": 1.272084842067442e-05,
+ "loss": 0.1219,
+ "step": 15778
+ },
+ {
+ "epoch": 42.99455040871935,
+ "grad_norm": 3.6413962841033936,
+ "learning_rate": 1.2719999209085152e-05,
+ "loss": 0.3164,
+ "step": 15779
+ },
+ {
+ "epoch": 42.997275204359674,
+ "grad_norm": 3.3227553367614746,
+ "learning_rate": 1.2719149976312617e-05,
+ "loss": 0.3467,
+ "step": 15780
+ },
+ {
+ "epoch": 43.0,
+ "grad_norm": 2.627345323562622,
+ "learning_rate": 1.2718300722363431e-05,
+ "loss": 0.0515,
+ "step": 15781
+ },
+ {
+ "epoch": 43.002724795640326,
+ "grad_norm": 3.4053261280059814,
+ "learning_rate": 1.2717451447244205e-05,
+ "loss": 0.0828,
+ "step": 15782
+ },
+ {
+ "epoch": 43.00544959128065,
+ "grad_norm": 3.4314918518066406,
+ "learning_rate": 1.2716602150961553e-05,
+ "loss": 0.2114,
+ "step": 15783
+ },
+ {
+ "epoch": 43.00817438692098,
+ "grad_norm": 2.782090663909912,
+ "learning_rate": 1.2715752833522094e-05,
+ "loss": 0.0754,
+ "step": 15784
+ },
+ {
+ "epoch": 43.01089918256131,
+ "grad_norm": 3.8802902698516846,
+ "learning_rate": 1.2714903494932438e-05,
+ "loss": 0.0976,
+ "step": 15785
+ },
+ {
+ "epoch": 43.013623978201636,
+ "grad_norm": 3.1236722469329834,
+ "learning_rate": 1.2714054135199201e-05,
+ "loss": 0.0815,
+ "step": 15786
+ },
+ {
+ "epoch": 43.01634877384196,
+ "grad_norm": 2.659071922302246,
+ "learning_rate": 1.2713204754328996e-05,
+ "loss": 0.0536,
+ "step": 15787
+ },
+ {
+ "epoch": 43.01907356948229,
+ "grad_norm": 3.43519926071167,
+ "learning_rate": 1.2712355352328442e-05,
+ "loss": 0.2828,
+ "step": 15788
+ },
+ {
+ "epoch": 43.02179836512261,
+ "grad_norm": 3.1571013927459717,
+ "learning_rate": 1.2711505929204153e-05,
+ "loss": 0.0714,
+ "step": 15789
+ },
+ {
+ "epoch": 43.02452316076294,
+ "grad_norm": 3.8070714473724365,
+ "learning_rate": 1.2710656484962742e-05,
+ "loss": 0.1481,
+ "step": 15790
+ },
+ {
+ "epoch": 43.02724795640327,
+ "grad_norm": 3.5965735912323,
+ "learning_rate": 1.2709807019610823e-05,
+ "loss": 0.0805,
+ "step": 15791
+ },
+ {
+ "epoch": 43.0299727520436,
+ "grad_norm": 3.3077375888824463,
+ "learning_rate": 1.2708957533155014e-05,
+ "loss": 0.1686,
+ "step": 15792
+ },
+ {
+ "epoch": 43.032697547683924,
+ "grad_norm": 2.999852418899536,
+ "learning_rate": 1.2708108025601935e-05,
+ "loss": 0.0725,
+ "step": 15793
+ },
+ {
+ "epoch": 43.03542234332425,
+ "grad_norm": 3.2349658012390137,
+ "learning_rate": 1.2707258496958197e-05,
+ "loss": 0.1736,
+ "step": 15794
+ },
+ {
+ "epoch": 43.038147138964575,
+ "grad_norm": 3.3749401569366455,
+ "learning_rate": 1.2706408947230414e-05,
+ "loss": 0.0661,
+ "step": 15795
+ },
+ {
+ "epoch": 43.0408719346049,
+ "grad_norm": 3.043250322341919,
+ "learning_rate": 1.270555937642521e-05,
+ "loss": 0.1807,
+ "step": 15796
+ },
+ {
+ "epoch": 43.043596730245234,
+ "grad_norm": 3.328571081161499,
+ "learning_rate": 1.2704709784549192e-05,
+ "loss": 0.2472,
+ "step": 15797
+ },
+ {
+ "epoch": 43.04632152588556,
+ "grad_norm": 3.9302914142608643,
+ "learning_rate": 1.2703860171608985e-05,
+ "loss": 0.1562,
+ "step": 15798
+ },
+ {
+ "epoch": 43.049046321525886,
+ "grad_norm": 3.035996675491333,
+ "learning_rate": 1.2703010537611198e-05,
+ "loss": 0.0793,
+ "step": 15799
+ },
+ {
+ "epoch": 43.05177111716621,
+ "grad_norm": 3.7502429485321045,
+ "learning_rate": 1.2702160882562453e-05,
+ "loss": 0.1015,
+ "step": 15800
+ },
+ {
+ "epoch": 43.05449591280654,
+ "grad_norm": 2.774843454360962,
+ "learning_rate": 1.2701311206469366e-05,
+ "loss": 0.1915,
+ "step": 15801
+ },
+ {
+ "epoch": 43.05722070844686,
+ "grad_norm": 3.1792562007904053,
+ "learning_rate": 1.2700461509338555e-05,
+ "loss": 0.0746,
+ "step": 15802
+ },
+ {
+ "epoch": 43.059945504087196,
+ "grad_norm": 3.3948023319244385,
+ "learning_rate": 1.2699611791176638e-05,
+ "loss": 0.1371,
+ "step": 15803
+ },
+ {
+ "epoch": 43.06267029972752,
+ "grad_norm": 4.89182710647583,
+ "learning_rate": 1.269876205199023e-05,
+ "loss": 0.0545,
+ "step": 15804
+ },
+ {
+ "epoch": 43.06539509536785,
+ "grad_norm": 3.7218899726867676,
+ "learning_rate": 1.269791229178595e-05,
+ "loss": 0.1599,
+ "step": 15805
+ },
+ {
+ "epoch": 43.06811989100817,
+ "grad_norm": 4.639289379119873,
+ "learning_rate": 1.2697062510570417e-05,
+ "loss": 0.166,
+ "step": 15806
+ },
+ {
+ "epoch": 43.0708446866485,
+ "grad_norm": 3.0235321521759033,
+ "learning_rate": 1.2696212708350247e-05,
+ "loss": 0.1731,
+ "step": 15807
+ },
+ {
+ "epoch": 43.073569482288825,
+ "grad_norm": 4.024735927581787,
+ "learning_rate": 1.2695362885132057e-05,
+ "loss": 0.2707,
+ "step": 15808
+ },
+ {
+ "epoch": 43.07629427792916,
+ "grad_norm": 2.8919615745544434,
+ "learning_rate": 1.269451304092247e-05,
+ "loss": 0.0985,
+ "step": 15809
+ },
+ {
+ "epoch": 43.079019073569484,
+ "grad_norm": 4.22364616394043,
+ "learning_rate": 1.2693663175728102e-05,
+ "loss": 0.1906,
+ "step": 15810
+ },
+ {
+ "epoch": 43.08174386920981,
+ "grad_norm": 2.918194055557251,
+ "learning_rate": 1.269281328955557e-05,
+ "loss": 0.1818,
+ "step": 15811
+ },
+ {
+ "epoch": 43.084468664850135,
+ "grad_norm": 2.9532759189605713,
+ "learning_rate": 1.2691963382411498e-05,
+ "loss": 0.0805,
+ "step": 15812
+ },
+ {
+ "epoch": 43.08719346049046,
+ "grad_norm": 3.837831974029541,
+ "learning_rate": 1.26911134543025e-05,
+ "loss": 0.1911,
+ "step": 15813
+ },
+ {
+ "epoch": 43.08991825613079,
+ "grad_norm": 4.561805248260498,
+ "learning_rate": 1.2690263505235199e-05,
+ "loss": 0.1915,
+ "step": 15814
+ },
+ {
+ "epoch": 43.09264305177112,
+ "grad_norm": 4.4803786277771,
+ "learning_rate": 1.2689413535216209e-05,
+ "loss": 0.157,
+ "step": 15815
+ },
+ {
+ "epoch": 43.095367847411445,
+ "grad_norm": 3.322883367538452,
+ "learning_rate": 1.2688563544252156e-05,
+ "loss": 0.0981,
+ "step": 15816
+ },
+ {
+ "epoch": 43.09809264305177,
+ "grad_norm": 3.713918685913086,
+ "learning_rate": 1.2687713532349653e-05,
+ "loss": 0.2439,
+ "step": 15817
+ },
+ {
+ "epoch": 43.1008174386921,
+ "grad_norm": 2.951857566833496,
+ "learning_rate": 1.2686863499515327e-05,
+ "loss": 0.0848,
+ "step": 15818
+ },
+ {
+ "epoch": 43.10354223433242,
+ "grad_norm": 3.1318840980529785,
+ "learning_rate": 1.2686013445755792e-05,
+ "loss": 0.1192,
+ "step": 15819
+ },
+ {
+ "epoch": 43.10626702997275,
+ "grad_norm": 2.4734785556793213,
+ "learning_rate": 1.2685163371077676e-05,
+ "loss": 0.0759,
+ "step": 15820
+ },
+ {
+ "epoch": 43.10899182561308,
+ "grad_norm": 3.2993366718292236,
+ "learning_rate": 1.2684313275487587e-05,
+ "loss": 0.139,
+ "step": 15821
+ },
+ {
+ "epoch": 43.11171662125341,
+ "grad_norm": 3.350257635116577,
+ "learning_rate": 1.2683463158992157e-05,
+ "loss": 0.1295,
+ "step": 15822
+ },
+ {
+ "epoch": 43.11444141689373,
+ "grad_norm": 2.7190537452697754,
+ "learning_rate": 1.2682613021598e-05,
+ "loss": 0.1203,
+ "step": 15823
+ },
+ {
+ "epoch": 43.11716621253406,
+ "grad_norm": 3.587864637374878,
+ "learning_rate": 1.2681762863311744e-05,
+ "loss": 0.2044,
+ "step": 15824
+ },
+ {
+ "epoch": 43.119891008174385,
+ "grad_norm": 2.5878560543060303,
+ "learning_rate": 1.2680912684139996e-05,
+ "loss": 0.2092,
+ "step": 15825
+ },
+ {
+ "epoch": 43.12261580381471,
+ "grad_norm": 3.0288000106811523,
+ "learning_rate": 1.2680062484089396e-05,
+ "loss": 0.0965,
+ "step": 15826
+ },
+ {
+ "epoch": 43.12534059945504,
+ "grad_norm": 2.698408842086792,
+ "learning_rate": 1.2679212263166548e-05,
+ "loss": 0.0833,
+ "step": 15827
+ },
+ {
+ "epoch": 43.12806539509537,
+ "grad_norm": 2.469306707382202,
+ "learning_rate": 1.2678362021378085e-05,
+ "loss": 0.1202,
+ "step": 15828
+ },
+ {
+ "epoch": 43.130790190735695,
+ "grad_norm": 2.896528720855713,
+ "learning_rate": 1.2677511758730625e-05,
+ "loss": 0.2874,
+ "step": 15829
+ },
+ {
+ "epoch": 43.13351498637602,
+ "grad_norm": 3.548424243927002,
+ "learning_rate": 1.267666147523079e-05,
+ "loss": 0.1078,
+ "step": 15830
+ },
+ {
+ "epoch": 43.13623978201635,
+ "grad_norm": 3.3155882358551025,
+ "learning_rate": 1.2675811170885199e-05,
+ "loss": 0.1523,
+ "step": 15831
+ },
+ {
+ "epoch": 43.13896457765667,
+ "grad_norm": 2.703396797180176,
+ "learning_rate": 1.2674960845700478e-05,
+ "loss": 0.0706,
+ "step": 15832
+ },
+ {
+ "epoch": 43.141689373297005,
+ "grad_norm": 2.8026251792907715,
+ "learning_rate": 1.267411049968325e-05,
+ "loss": 0.0937,
+ "step": 15833
+ },
+ {
+ "epoch": 43.14441416893733,
+ "grad_norm": 3.0272340774536133,
+ "learning_rate": 1.2673260132840133e-05,
+ "loss": 0.1669,
+ "step": 15834
+ },
+ {
+ "epoch": 43.14713896457766,
+ "grad_norm": 2.2173011302948,
+ "learning_rate": 1.2672409745177752e-05,
+ "loss": 0.0608,
+ "step": 15835
+ },
+ {
+ "epoch": 43.14986376021798,
+ "grad_norm": 1.8766918182373047,
+ "learning_rate": 1.267155933670273e-05,
+ "loss": 0.0591,
+ "step": 15836
+ },
+ {
+ "epoch": 43.15258855585831,
+ "grad_norm": 3.0187346935272217,
+ "learning_rate": 1.267070890742169e-05,
+ "loss": 0.0663,
+ "step": 15837
+ },
+ {
+ "epoch": 43.155313351498634,
+ "grad_norm": 15.613056182861328,
+ "learning_rate": 1.2669858457341258e-05,
+ "loss": 0.1611,
+ "step": 15838
+ },
+ {
+ "epoch": 43.15803814713897,
+ "grad_norm": 3.486299991607666,
+ "learning_rate": 1.266900798646805e-05,
+ "loss": 0.0846,
+ "step": 15839
+ },
+ {
+ "epoch": 43.16076294277929,
+ "grad_norm": 3.063582181930542,
+ "learning_rate": 1.2668157494808697e-05,
+ "loss": 0.0609,
+ "step": 15840
+ },
+ {
+ "epoch": 43.16348773841962,
+ "grad_norm": 3.5328421592712402,
+ "learning_rate": 1.2667306982369816e-05,
+ "loss": 0.0853,
+ "step": 15841
+ },
+ {
+ "epoch": 43.166212534059945,
+ "grad_norm": 3.0072851181030273,
+ "learning_rate": 1.2666456449158037e-05,
+ "loss": 0.2076,
+ "step": 15842
+ },
+ {
+ "epoch": 43.16893732970027,
+ "grad_norm": 3.958845376968384,
+ "learning_rate": 1.2665605895179977e-05,
+ "loss": 0.217,
+ "step": 15843
+ },
+ {
+ "epoch": 43.171662125340596,
+ "grad_norm": 2.736316442489624,
+ "learning_rate": 1.266475532044227e-05,
+ "loss": 0.0729,
+ "step": 15844
+ },
+ {
+ "epoch": 43.17438692098093,
+ "grad_norm": 2.9079360961914062,
+ "learning_rate": 1.266390472495153e-05,
+ "loss": 0.1587,
+ "step": 15845
+ },
+ {
+ "epoch": 43.177111716621255,
+ "grad_norm": 2.661457061767578,
+ "learning_rate": 1.2663054108714388e-05,
+ "loss": 0.2046,
+ "step": 15846
+ },
+ {
+ "epoch": 43.17983651226158,
+ "grad_norm": 3.3228859901428223,
+ "learning_rate": 1.2662203471737465e-05,
+ "loss": 0.272,
+ "step": 15847
+ },
+ {
+ "epoch": 43.182561307901906,
+ "grad_norm": 2.7180087566375732,
+ "learning_rate": 1.2661352814027388e-05,
+ "loss": 0.1583,
+ "step": 15848
+ },
+ {
+ "epoch": 43.18528610354223,
+ "grad_norm": 3.396350145339966,
+ "learning_rate": 1.266050213559078e-05,
+ "loss": 0.1105,
+ "step": 15849
+ },
+ {
+ "epoch": 43.18801089918256,
+ "grad_norm": 3.4732699394226074,
+ "learning_rate": 1.2659651436434268e-05,
+ "loss": 0.1222,
+ "step": 15850
+ },
+ {
+ "epoch": 43.19073569482289,
+ "grad_norm": 3.158820152282715,
+ "learning_rate": 1.2658800716564474e-05,
+ "loss": 0.0827,
+ "step": 15851
+ },
+ {
+ "epoch": 43.19346049046322,
+ "grad_norm": 4.200428009033203,
+ "learning_rate": 1.2657949975988028e-05,
+ "loss": 0.1606,
+ "step": 15852
+ },
+ {
+ "epoch": 43.19618528610354,
+ "grad_norm": 2.985954761505127,
+ "learning_rate": 1.2657099214711553e-05,
+ "loss": 0.0739,
+ "step": 15853
+ },
+ {
+ "epoch": 43.19891008174387,
+ "grad_norm": 3.0069069862365723,
+ "learning_rate": 1.2656248432741674e-05,
+ "loss": 0.0703,
+ "step": 15854
+ },
+ {
+ "epoch": 43.201634877384194,
+ "grad_norm": 3.790745973587036,
+ "learning_rate": 1.2655397630085015e-05,
+ "loss": 0.0816,
+ "step": 15855
+ },
+ {
+ "epoch": 43.20435967302452,
+ "grad_norm": 2.6209158897399902,
+ "learning_rate": 1.2654546806748209e-05,
+ "loss": 0.0814,
+ "step": 15856
+ },
+ {
+ "epoch": 43.20708446866485,
+ "grad_norm": 3.196512460708618,
+ "learning_rate": 1.2653695962737875e-05,
+ "loss": 0.1133,
+ "step": 15857
+ },
+ {
+ "epoch": 43.20980926430518,
+ "grad_norm": 3.4734902381896973,
+ "learning_rate": 1.2652845098060647e-05,
+ "loss": 0.0911,
+ "step": 15858
+ },
+ {
+ "epoch": 43.212534059945504,
+ "grad_norm": 5.279588222503662,
+ "learning_rate": 1.2651994212723142e-05,
+ "loss": 0.1036,
+ "step": 15859
+ },
+ {
+ "epoch": 43.21525885558583,
+ "grad_norm": 3.614436626434326,
+ "learning_rate": 1.2651143306731992e-05,
+ "loss": 0.1079,
+ "step": 15860
+ },
+ {
+ "epoch": 43.217983651226156,
+ "grad_norm": 3.3685035705566406,
+ "learning_rate": 1.2650292380093826e-05,
+ "loss": 0.1162,
+ "step": 15861
+ },
+ {
+ "epoch": 43.22070844686648,
+ "grad_norm": 3.7627322673797607,
+ "learning_rate": 1.2649441432815266e-05,
+ "loss": 0.0865,
+ "step": 15862
+ },
+ {
+ "epoch": 43.223433242506815,
+ "grad_norm": 3.0178403854370117,
+ "learning_rate": 1.2648590464902941e-05,
+ "loss": 0.1262,
+ "step": 15863
+ },
+ {
+ "epoch": 43.22615803814714,
+ "grad_norm": 3.1500260829925537,
+ "learning_rate": 1.264773947636348e-05,
+ "loss": 0.1983,
+ "step": 15864
+ },
+ {
+ "epoch": 43.228882833787466,
+ "grad_norm": 3.4203460216522217,
+ "learning_rate": 1.2646888467203509e-05,
+ "loss": 0.0918,
+ "step": 15865
+ },
+ {
+ "epoch": 43.23160762942779,
+ "grad_norm": 7.338772773742676,
+ "learning_rate": 1.2646037437429654e-05,
+ "loss": 0.2069,
+ "step": 15866
+ },
+ {
+ "epoch": 43.23433242506812,
+ "grad_norm": 3.13272762298584,
+ "learning_rate": 1.2645186387048546e-05,
+ "loss": 0.0803,
+ "step": 15867
+ },
+ {
+ "epoch": 43.237057220708444,
+ "grad_norm": 3.2692627906799316,
+ "learning_rate": 1.2644335316066811e-05,
+ "loss": 0.1282,
+ "step": 15868
+ },
+ {
+ "epoch": 43.23978201634878,
+ "grad_norm": 3.218737840652466,
+ "learning_rate": 1.2643484224491078e-05,
+ "loss": 0.1428,
+ "step": 15869
+ },
+ {
+ "epoch": 43.2425068119891,
+ "grad_norm": 3.3984150886535645,
+ "learning_rate": 1.2642633112327975e-05,
+ "loss": 0.1676,
+ "step": 15870
+ },
+ {
+ "epoch": 43.24523160762943,
+ "grad_norm": 3.2092783451080322,
+ "learning_rate": 1.2641781979584133e-05,
+ "loss": 0.0568,
+ "step": 15871
+ },
+ {
+ "epoch": 43.247956403269754,
+ "grad_norm": 2.8498129844665527,
+ "learning_rate": 1.2640930826266176e-05,
+ "loss": 0.1057,
+ "step": 15872
+ },
+ {
+ "epoch": 43.25068119891008,
+ "grad_norm": 3.3885536193847656,
+ "learning_rate": 1.2640079652380734e-05,
+ "loss": 0.0996,
+ "step": 15873
+ },
+ {
+ "epoch": 43.253405994550405,
+ "grad_norm": 4.925052642822266,
+ "learning_rate": 1.2639228457934435e-05,
+ "loss": 0.0952,
+ "step": 15874
+ },
+ {
+ "epoch": 43.25613079019074,
+ "grad_norm": 3.2135350704193115,
+ "learning_rate": 1.2638377242933912e-05,
+ "loss": 0.2372,
+ "step": 15875
+ },
+ {
+ "epoch": 43.258855585831064,
+ "grad_norm": 3.064039945602417,
+ "learning_rate": 1.2637526007385792e-05,
+ "loss": 0.1174,
+ "step": 15876
+ },
+ {
+ "epoch": 43.26158038147139,
+ "grad_norm": 3.5446348190307617,
+ "learning_rate": 1.2636674751296702e-05,
+ "loss": 0.1423,
+ "step": 15877
+ },
+ {
+ "epoch": 43.264305177111716,
+ "grad_norm": 3.469653367996216,
+ "learning_rate": 1.2635823474673277e-05,
+ "loss": 0.0836,
+ "step": 15878
+ },
+ {
+ "epoch": 43.26702997275204,
+ "grad_norm": 3.0113861560821533,
+ "learning_rate": 1.2634972177522143e-05,
+ "loss": 0.1955,
+ "step": 15879
+ },
+ {
+ "epoch": 43.26975476839237,
+ "grad_norm": 3.3286073207855225,
+ "learning_rate": 1.263412085984993e-05,
+ "loss": 0.1775,
+ "step": 15880
+ },
+ {
+ "epoch": 43.2724795640327,
+ "grad_norm": 2.947270393371582,
+ "learning_rate": 1.2633269521663269e-05,
+ "loss": 0.0834,
+ "step": 15881
+ },
+ {
+ "epoch": 43.275204359673026,
+ "grad_norm": 3.6426942348480225,
+ "learning_rate": 1.2632418162968789e-05,
+ "loss": 0.1446,
+ "step": 15882
+ },
+ {
+ "epoch": 43.27792915531335,
+ "grad_norm": 3.199448347091675,
+ "learning_rate": 1.2631566783773121e-05,
+ "loss": 0.1044,
+ "step": 15883
+ },
+ {
+ "epoch": 43.28065395095368,
+ "grad_norm": 2.7437596321105957,
+ "learning_rate": 1.2630715384082897e-05,
+ "loss": 0.1814,
+ "step": 15884
+ },
+ {
+ "epoch": 43.283378746594,
+ "grad_norm": 3.0962471961975098,
+ "learning_rate": 1.2629863963904745e-05,
+ "loss": 0.0716,
+ "step": 15885
+ },
+ {
+ "epoch": 43.28610354223433,
+ "grad_norm": 2.8330252170562744,
+ "learning_rate": 1.2629012523245294e-05,
+ "loss": 0.346,
+ "step": 15886
+ },
+ {
+ "epoch": 43.28882833787466,
+ "grad_norm": 2.8586809635162354,
+ "learning_rate": 1.2628161062111182e-05,
+ "loss": 0.0881,
+ "step": 15887
+ },
+ {
+ "epoch": 43.29155313351499,
+ "grad_norm": 2.473142147064209,
+ "learning_rate": 1.2627309580509034e-05,
+ "loss": 0.0915,
+ "step": 15888
+ },
+ {
+ "epoch": 43.294277929155314,
+ "grad_norm": 2.52375864982605,
+ "learning_rate": 1.2626458078445487e-05,
+ "loss": 0.0613,
+ "step": 15889
+ },
+ {
+ "epoch": 43.29700272479564,
+ "grad_norm": 3.3829197883605957,
+ "learning_rate": 1.2625606555927167e-05,
+ "loss": 0.0656,
+ "step": 15890
+ },
+ {
+ "epoch": 43.299727520435965,
+ "grad_norm": 2.699552297592163,
+ "learning_rate": 1.2624755012960707e-05,
+ "loss": 0.1272,
+ "step": 15891
+ },
+ {
+ "epoch": 43.30245231607629,
+ "grad_norm": 2.336064577102661,
+ "learning_rate": 1.2623903449552741e-05,
+ "loss": 0.0575,
+ "step": 15892
+ },
+ {
+ "epoch": 43.305177111716624,
+ "grad_norm": 3.315103530883789,
+ "learning_rate": 1.2623051865709899e-05,
+ "loss": 0.2,
+ "step": 15893
+ },
+ {
+ "epoch": 43.30790190735695,
+ "grad_norm": 2.7759604454040527,
+ "learning_rate": 1.262220026143881e-05,
+ "loss": 0.0667,
+ "step": 15894
+ },
+ {
+ "epoch": 43.310626702997276,
+ "grad_norm": 2.701545000076294,
+ "learning_rate": 1.2621348636746115e-05,
+ "loss": 0.1046,
+ "step": 15895
+ },
+ {
+ "epoch": 43.3133514986376,
+ "grad_norm": 3.074169158935547,
+ "learning_rate": 1.2620496991638439e-05,
+ "loss": 0.0659,
+ "step": 15896
+ },
+ {
+ "epoch": 43.31607629427793,
+ "grad_norm": 3.0037648677825928,
+ "learning_rate": 1.2619645326122416e-05,
+ "loss": 0.1234,
+ "step": 15897
+ },
+ {
+ "epoch": 43.31880108991825,
+ "grad_norm": 1.977616548538208,
+ "learning_rate": 1.261879364020468e-05,
+ "loss": 0.0518,
+ "step": 15898
+ },
+ {
+ "epoch": 43.321525885558586,
+ "grad_norm": 4.094875335693359,
+ "learning_rate": 1.2617941933891867e-05,
+ "loss": 0.2237,
+ "step": 15899
+ },
+ {
+ "epoch": 43.32425068119891,
+ "grad_norm": 2.396916389465332,
+ "learning_rate": 1.26170902071906e-05,
+ "loss": 0.1021,
+ "step": 15900
+ },
+ {
+ "epoch": 43.32697547683924,
+ "grad_norm": 2.3406007289886475,
+ "learning_rate": 1.2616238460107525e-05,
+ "loss": 0.0789,
+ "step": 15901
+ },
+ {
+ "epoch": 43.32970027247956,
+ "grad_norm": 3.4132494926452637,
+ "learning_rate": 1.2615386692649263e-05,
+ "loss": 0.0722,
+ "step": 15902
+ },
+ {
+ "epoch": 43.33242506811989,
+ "grad_norm": 3.338123083114624,
+ "learning_rate": 1.2614534904822455e-05,
+ "loss": 0.1911,
+ "step": 15903
+ },
+ {
+ "epoch": 43.335149863760215,
+ "grad_norm": 3.6919007301330566,
+ "learning_rate": 1.2613683096633736e-05,
+ "loss": 0.2373,
+ "step": 15904
+ },
+ {
+ "epoch": 43.33787465940055,
+ "grad_norm": 3.2064146995544434,
+ "learning_rate": 1.2612831268089736e-05,
+ "loss": 0.0964,
+ "step": 15905
+ },
+ {
+ "epoch": 43.34059945504087,
+ "grad_norm": 4.3943047523498535,
+ "learning_rate": 1.261197941919709e-05,
+ "loss": 0.1341,
+ "step": 15906
+ },
+ {
+ "epoch": 43.3433242506812,
+ "grad_norm": 2.9569637775421143,
+ "learning_rate": 1.2611127549962432e-05,
+ "loss": 0.0628,
+ "step": 15907
+ },
+ {
+ "epoch": 43.346049046321525,
+ "grad_norm": 3.952716588973999,
+ "learning_rate": 1.2610275660392397e-05,
+ "loss": 0.16,
+ "step": 15908
+ },
+ {
+ "epoch": 43.34877384196185,
+ "grad_norm": 2.602510690689087,
+ "learning_rate": 1.2609423750493622e-05,
+ "loss": 0.1132,
+ "step": 15909
+ },
+ {
+ "epoch": 43.35149863760218,
+ "grad_norm": 2.767552375793457,
+ "learning_rate": 1.2608571820272734e-05,
+ "loss": 0.1173,
+ "step": 15910
+ },
+ {
+ "epoch": 43.35422343324251,
+ "grad_norm": 3.195824384689331,
+ "learning_rate": 1.2607719869736374e-05,
+ "loss": 0.1019,
+ "step": 15911
+ },
+ {
+ "epoch": 43.356948228882835,
+ "grad_norm": 3.8886680603027344,
+ "learning_rate": 1.2606867898891175e-05,
+ "loss": 0.0837,
+ "step": 15912
+ },
+ {
+ "epoch": 43.35967302452316,
+ "grad_norm": 4.240359306335449,
+ "learning_rate": 1.2606015907743777e-05,
+ "loss": 0.0738,
+ "step": 15913
+ },
+ {
+ "epoch": 43.36239782016349,
+ "grad_norm": 3.2453551292419434,
+ "learning_rate": 1.2605163896300808e-05,
+ "loss": 0.0875,
+ "step": 15914
+ },
+ {
+ "epoch": 43.36512261580381,
+ "grad_norm": 3.355379581451416,
+ "learning_rate": 1.2604311864568905e-05,
+ "loss": 0.1742,
+ "step": 15915
+ },
+ {
+ "epoch": 43.36784741144414,
+ "grad_norm": 3.202354669570923,
+ "learning_rate": 1.2603459812554708e-05,
+ "loss": 0.098,
+ "step": 15916
+ },
+ {
+ "epoch": 43.37057220708447,
+ "grad_norm": 3.1638927459716797,
+ "learning_rate": 1.260260774026485e-05,
+ "loss": 0.0717,
+ "step": 15917
+ },
+ {
+ "epoch": 43.3732970027248,
+ "grad_norm": 3.006913900375366,
+ "learning_rate": 1.2601755647705963e-05,
+ "loss": 0.0998,
+ "step": 15918
+ },
+ {
+ "epoch": 43.37602179836512,
+ "grad_norm": 3.8732352256774902,
+ "learning_rate": 1.260090353488469e-05,
+ "loss": 0.1502,
+ "step": 15919
+ },
+ {
+ "epoch": 43.37874659400545,
+ "grad_norm": 3.449373483657837,
+ "learning_rate": 1.2600051401807659e-05,
+ "loss": 0.227,
+ "step": 15920
+ },
+ {
+ "epoch": 43.381471389645775,
+ "grad_norm": 3.104627847671509,
+ "learning_rate": 1.2599199248481515e-05,
+ "loss": 0.1599,
+ "step": 15921
+ },
+ {
+ "epoch": 43.3841961852861,
+ "grad_norm": 3.114495277404785,
+ "learning_rate": 1.2598347074912891e-05,
+ "loss": 0.1224,
+ "step": 15922
+ },
+ {
+ "epoch": 43.38692098092643,
+ "grad_norm": 3.467994213104248,
+ "learning_rate": 1.2597494881108425e-05,
+ "loss": 0.131,
+ "step": 15923
+ },
+ {
+ "epoch": 43.38964577656676,
+ "grad_norm": 3.2575275897979736,
+ "learning_rate": 1.259664266707475e-05,
+ "loss": 0.1644,
+ "step": 15924
+ },
+ {
+ "epoch": 43.392370572207085,
+ "grad_norm": 4.1949310302734375,
+ "learning_rate": 1.2595790432818508e-05,
+ "loss": 0.2425,
+ "step": 15925
+ },
+ {
+ "epoch": 43.39509536784741,
+ "grad_norm": 6.824401378631592,
+ "learning_rate": 1.2594938178346332e-05,
+ "loss": 0.1271,
+ "step": 15926
+ },
+ {
+ "epoch": 43.39782016348774,
+ "grad_norm": 3.0114896297454834,
+ "learning_rate": 1.2594085903664861e-05,
+ "loss": 0.1368,
+ "step": 15927
+ },
+ {
+ "epoch": 43.40054495912806,
+ "grad_norm": 2.6770169734954834,
+ "learning_rate": 1.2593233608780733e-05,
+ "loss": 0.0841,
+ "step": 15928
+ },
+ {
+ "epoch": 43.403269754768395,
+ "grad_norm": 3.0328562259674072,
+ "learning_rate": 1.2592381293700583e-05,
+ "loss": 0.0533,
+ "step": 15929
+ },
+ {
+ "epoch": 43.40599455040872,
+ "grad_norm": 4.037158489227295,
+ "learning_rate": 1.2591528958431052e-05,
+ "loss": 0.0988,
+ "step": 15930
+ },
+ {
+ "epoch": 43.40871934604905,
+ "grad_norm": 2.9018964767456055,
+ "learning_rate": 1.259067660297878e-05,
+ "loss": 0.0774,
+ "step": 15931
+ },
+ {
+ "epoch": 43.41144414168937,
+ "grad_norm": 3.2149205207824707,
+ "learning_rate": 1.2589824227350397e-05,
+ "loss": 0.1619,
+ "step": 15932
+ },
+ {
+ "epoch": 43.4141689373297,
+ "grad_norm": 3.2354207038879395,
+ "learning_rate": 1.2588971831552553e-05,
+ "loss": 0.1493,
+ "step": 15933
+ },
+ {
+ "epoch": 43.416893732970024,
+ "grad_norm": 3.472376585006714,
+ "learning_rate": 1.2588119415591873e-05,
+ "loss": 0.1056,
+ "step": 15934
+ },
+ {
+ "epoch": 43.41961852861036,
+ "grad_norm": 3.359921932220459,
+ "learning_rate": 1.2587266979475006e-05,
+ "loss": 0.1343,
+ "step": 15935
+ },
+ {
+ "epoch": 43.42234332425068,
+ "grad_norm": 3.36545729637146,
+ "learning_rate": 1.2586414523208586e-05,
+ "loss": 0.1314,
+ "step": 15936
+ },
+ {
+ "epoch": 43.42506811989101,
+ "grad_norm": 3.343698024749756,
+ "learning_rate": 1.2585562046799251e-05,
+ "loss": 0.057,
+ "step": 15937
+ },
+ {
+ "epoch": 43.427792915531334,
+ "grad_norm": 3.2822048664093018,
+ "learning_rate": 1.2584709550253644e-05,
+ "loss": 0.1613,
+ "step": 15938
+ },
+ {
+ "epoch": 43.43051771117166,
+ "grad_norm": 4.329890251159668,
+ "learning_rate": 1.2583857033578401e-05,
+ "loss": 0.2325,
+ "step": 15939
+ },
+ {
+ "epoch": 43.433242506811986,
+ "grad_norm": 3.509294033050537,
+ "learning_rate": 1.2583004496780164e-05,
+ "loss": 0.1821,
+ "step": 15940
+ },
+ {
+ "epoch": 43.43596730245232,
+ "grad_norm": 3.1888256072998047,
+ "learning_rate": 1.258215193986557e-05,
+ "loss": 0.0509,
+ "step": 15941
+ },
+ {
+ "epoch": 43.438692098092645,
+ "grad_norm": 6.806349277496338,
+ "learning_rate": 1.2581299362841262e-05,
+ "loss": 0.0719,
+ "step": 15942
+ },
+ {
+ "epoch": 43.44141689373297,
+ "grad_norm": 4.555354118347168,
+ "learning_rate": 1.2580446765713876e-05,
+ "loss": 0.2435,
+ "step": 15943
+ },
+ {
+ "epoch": 43.444141689373296,
+ "grad_norm": 5.065535545349121,
+ "learning_rate": 1.2579594148490054e-05,
+ "loss": 0.0589,
+ "step": 15944
+ },
+ {
+ "epoch": 43.44686648501362,
+ "grad_norm": 4.211273670196533,
+ "learning_rate": 1.2578741511176436e-05,
+ "loss": 0.1228,
+ "step": 15945
+ },
+ {
+ "epoch": 43.44959128065395,
+ "grad_norm": 3.2076549530029297,
+ "learning_rate": 1.257788885377966e-05,
+ "loss": 0.0911,
+ "step": 15946
+ },
+ {
+ "epoch": 43.45231607629428,
+ "grad_norm": 3.4893760681152344,
+ "learning_rate": 1.257703617630637e-05,
+ "loss": 0.0984,
+ "step": 15947
+ },
+ {
+ "epoch": 43.45504087193461,
+ "grad_norm": 2.403320550918579,
+ "learning_rate": 1.2576183478763207e-05,
+ "loss": 0.0556,
+ "step": 15948
+ },
+ {
+ "epoch": 43.45776566757493,
+ "grad_norm": 3.3591248989105225,
+ "learning_rate": 1.2575330761156808e-05,
+ "loss": 0.333,
+ "step": 15949
+ },
+ {
+ "epoch": 43.46049046321526,
+ "grad_norm": 3.330505132675171,
+ "learning_rate": 1.2574478023493817e-05,
+ "loss": 0.1515,
+ "step": 15950
+ },
+ {
+ "epoch": 43.463215258855584,
+ "grad_norm": 6.1331000328063965,
+ "learning_rate": 1.2573625265780873e-05,
+ "loss": 0.1425,
+ "step": 15951
+ },
+ {
+ "epoch": 43.46594005449591,
+ "grad_norm": 3.485036611557007,
+ "learning_rate": 1.257277248802462e-05,
+ "loss": 0.2386,
+ "step": 15952
+ },
+ {
+ "epoch": 43.46866485013624,
+ "grad_norm": 3.397254705429077,
+ "learning_rate": 1.2571919690231695e-05,
+ "loss": 0.2839,
+ "step": 15953
+ },
+ {
+ "epoch": 43.47138964577657,
+ "grad_norm": 3.3949978351593018,
+ "learning_rate": 1.2571066872408744e-05,
+ "loss": 0.0623,
+ "step": 15954
+ },
+ {
+ "epoch": 43.474114441416894,
+ "grad_norm": 2.9776296615600586,
+ "learning_rate": 1.2570214034562404e-05,
+ "loss": 0.1234,
+ "step": 15955
+ },
+ {
+ "epoch": 43.47683923705722,
+ "grad_norm": 3.178307294845581,
+ "learning_rate": 1.2569361176699323e-05,
+ "loss": 0.1079,
+ "step": 15956
+ },
+ {
+ "epoch": 43.479564032697546,
+ "grad_norm": 3.005838394165039,
+ "learning_rate": 1.256850829882614e-05,
+ "loss": 0.1022,
+ "step": 15957
+ },
+ {
+ "epoch": 43.48228882833787,
+ "grad_norm": 4.107235431671143,
+ "learning_rate": 1.2567655400949497e-05,
+ "loss": 0.2057,
+ "step": 15958
+ },
+ {
+ "epoch": 43.485013623978205,
+ "grad_norm": 2.6124253273010254,
+ "learning_rate": 1.2566802483076033e-05,
+ "loss": 0.0782,
+ "step": 15959
+ },
+ {
+ "epoch": 43.48773841961853,
+ "grad_norm": 2.9140758514404297,
+ "learning_rate": 1.25659495452124e-05,
+ "loss": 0.1478,
+ "step": 15960
+ },
+ {
+ "epoch": 43.490463215258856,
+ "grad_norm": 3.2501001358032227,
+ "learning_rate": 1.256509658736523e-05,
+ "loss": 0.0603,
+ "step": 15961
+ },
+ {
+ "epoch": 43.49318801089918,
+ "grad_norm": 2.8687989711761475,
+ "learning_rate": 1.2564243609541172e-05,
+ "loss": 0.0849,
+ "step": 15962
+ },
+ {
+ "epoch": 43.49591280653951,
+ "grad_norm": 2.8265609741210938,
+ "learning_rate": 1.2563390611746864e-05,
+ "loss": 0.0404,
+ "step": 15963
+ },
+ {
+ "epoch": 43.49863760217983,
+ "grad_norm": 3.3696513175964355,
+ "learning_rate": 1.2562537593988954e-05,
+ "loss": 0.0972,
+ "step": 15964
+ },
+ {
+ "epoch": 43.50136239782017,
+ "grad_norm": 2.9844369888305664,
+ "learning_rate": 1.2561684556274085e-05,
+ "loss": 0.128,
+ "step": 15965
+ },
+ {
+ "epoch": 43.50408719346049,
+ "grad_norm": 2.8845913410186768,
+ "learning_rate": 1.25608314986089e-05,
+ "loss": 0.0445,
+ "step": 15966
+ },
+ {
+ "epoch": 43.50681198910082,
+ "grad_norm": 3.2457525730133057,
+ "learning_rate": 1.2559978421000039e-05,
+ "loss": 0.074,
+ "step": 15967
+ },
+ {
+ "epoch": 43.509536784741144,
+ "grad_norm": 3.3454747200012207,
+ "learning_rate": 1.2559125323454154e-05,
+ "loss": 0.1454,
+ "step": 15968
+ },
+ {
+ "epoch": 43.51226158038147,
+ "grad_norm": 2.8938045501708984,
+ "learning_rate": 1.2558272205977877e-05,
+ "loss": 0.1289,
+ "step": 15969
+ },
+ {
+ "epoch": 43.514986376021795,
+ "grad_norm": 2.5399129390716553,
+ "learning_rate": 1.255741906857786e-05,
+ "loss": 0.0447,
+ "step": 15970
+ },
+ {
+ "epoch": 43.51771117166213,
+ "grad_norm": 3.401662826538086,
+ "learning_rate": 1.2556565911260747e-05,
+ "loss": 0.1588,
+ "step": 15971
+ },
+ {
+ "epoch": 43.520435967302454,
+ "grad_norm": 3.6595802307128906,
+ "learning_rate": 1.2555712734033179e-05,
+ "loss": 0.0745,
+ "step": 15972
+ },
+ {
+ "epoch": 43.52316076294278,
+ "grad_norm": 3.78926944732666,
+ "learning_rate": 1.2554859536901805e-05,
+ "loss": 0.2056,
+ "step": 15973
+ },
+ {
+ "epoch": 43.525885558583106,
+ "grad_norm": 2.500774383544922,
+ "learning_rate": 1.2554006319873265e-05,
+ "loss": 0.0481,
+ "step": 15974
+ },
+ {
+ "epoch": 43.52861035422343,
+ "grad_norm": 2.8060548305511475,
+ "learning_rate": 1.2553153082954206e-05,
+ "loss": 0.188,
+ "step": 15975
+ },
+ {
+ "epoch": 43.53133514986376,
+ "grad_norm": 3.2901575565338135,
+ "learning_rate": 1.2552299826151275e-05,
+ "loss": 0.1311,
+ "step": 15976
+ },
+ {
+ "epoch": 43.53405994550409,
+ "grad_norm": 2.916433334350586,
+ "learning_rate": 1.2551446549471114e-05,
+ "loss": 0.1812,
+ "step": 15977
+ },
+ {
+ "epoch": 43.536784741144416,
+ "grad_norm": 3.2524704933166504,
+ "learning_rate": 1.2550593252920371e-05,
+ "loss": 0.1158,
+ "step": 15978
+ },
+ {
+ "epoch": 43.53950953678474,
+ "grad_norm": 3.257223129272461,
+ "learning_rate": 1.2549739936505687e-05,
+ "loss": 0.0802,
+ "step": 15979
+ },
+ {
+ "epoch": 43.54223433242507,
+ "grad_norm": 3.1468281745910645,
+ "learning_rate": 1.2548886600233709e-05,
+ "loss": 0.1063,
+ "step": 15980
+ },
+ {
+ "epoch": 43.54495912806539,
+ "grad_norm": 2.1983208656311035,
+ "learning_rate": 1.2548033244111087e-05,
+ "loss": 0.0449,
+ "step": 15981
+ },
+ {
+ "epoch": 43.54768392370572,
+ "grad_norm": 3.227219581604004,
+ "learning_rate": 1.2547179868144466e-05,
+ "loss": 0.075,
+ "step": 15982
+ },
+ {
+ "epoch": 43.55040871934605,
+ "grad_norm": 3.178009033203125,
+ "learning_rate": 1.2546326472340486e-05,
+ "loss": 0.078,
+ "step": 15983
+ },
+ {
+ "epoch": 43.55313351498638,
+ "grad_norm": 3.188004970550537,
+ "learning_rate": 1.2545473056705801e-05,
+ "loss": 0.0631,
+ "step": 15984
+ },
+ {
+ "epoch": 43.555858310626704,
+ "grad_norm": 3.1795873641967773,
+ "learning_rate": 1.2544619621247052e-05,
+ "loss": 0.0505,
+ "step": 15985
+ },
+ {
+ "epoch": 43.55858310626703,
+ "grad_norm": 3.682218551635742,
+ "learning_rate": 1.2543766165970887e-05,
+ "loss": 0.0651,
+ "step": 15986
+ },
+ {
+ "epoch": 43.561307901907355,
+ "grad_norm": 3.6823067665100098,
+ "learning_rate": 1.2542912690883954e-05,
+ "loss": 0.1586,
+ "step": 15987
+ },
+ {
+ "epoch": 43.56403269754768,
+ "grad_norm": 2.9260642528533936,
+ "learning_rate": 1.25420591959929e-05,
+ "loss": 0.1224,
+ "step": 15988
+ },
+ {
+ "epoch": 43.566757493188014,
+ "grad_norm": 2.662670612335205,
+ "learning_rate": 1.2541205681304364e-05,
+ "loss": 0.0715,
+ "step": 15989
+ },
+ {
+ "epoch": 43.56948228882834,
+ "grad_norm": 2.285619020462036,
+ "learning_rate": 1.2540352146825008e-05,
+ "loss": 0.0505,
+ "step": 15990
+ },
+ {
+ "epoch": 43.572207084468666,
+ "grad_norm": 3.6111443042755127,
+ "learning_rate": 1.2539498592561468e-05,
+ "loss": 0.108,
+ "step": 15991
+ },
+ {
+ "epoch": 43.57493188010899,
+ "grad_norm": 2.838327646255493,
+ "learning_rate": 1.2538645018520396e-05,
+ "loss": 0.0612,
+ "step": 15992
+ },
+ {
+ "epoch": 43.57765667574932,
+ "grad_norm": 4.754798889160156,
+ "learning_rate": 1.2537791424708436e-05,
+ "loss": 0.2196,
+ "step": 15993
+ },
+ {
+ "epoch": 43.58038147138964,
+ "grad_norm": 3.3089184761047363,
+ "learning_rate": 1.2536937811132242e-05,
+ "loss": 0.0837,
+ "step": 15994
+ },
+ {
+ "epoch": 43.583106267029976,
+ "grad_norm": 2.4091646671295166,
+ "learning_rate": 1.2536084177798453e-05,
+ "loss": 0.0559,
+ "step": 15995
+ },
+ {
+ "epoch": 43.5858310626703,
+ "grad_norm": 3.7022528648376465,
+ "learning_rate": 1.2535230524713726e-05,
+ "loss": 0.0764,
+ "step": 15996
+ },
+ {
+ "epoch": 43.58855585831063,
+ "grad_norm": 3.101780891418457,
+ "learning_rate": 1.2534376851884702e-05,
+ "loss": 0.1116,
+ "step": 15997
+ },
+ {
+ "epoch": 43.59128065395095,
+ "grad_norm": 3.391084909439087,
+ "learning_rate": 1.2533523159318035e-05,
+ "loss": 0.0748,
+ "step": 15998
+ },
+ {
+ "epoch": 43.59400544959128,
+ "grad_norm": 4.03339147567749,
+ "learning_rate": 1.2532669447020369e-05,
+ "loss": 0.0998,
+ "step": 15999
+ },
+ {
+ "epoch": 43.596730245231605,
+ "grad_norm": 5.724348545074463,
+ "learning_rate": 1.2531815714998356e-05,
+ "loss": 0.2266,
+ "step": 16000
+ },
+ {
+ "epoch": 43.59945504087194,
+ "grad_norm": 3.0274693965911865,
+ "learning_rate": 1.2530961963258643e-05,
+ "loss": 0.1011,
+ "step": 16001
+ },
+ {
+ "epoch": 43.60217983651226,
+ "grad_norm": 2.974632501602173,
+ "learning_rate": 1.2530108191807884e-05,
+ "loss": 0.1535,
+ "step": 16002
+ },
+ {
+ "epoch": 43.60490463215259,
+ "grad_norm": 3.3523025512695312,
+ "learning_rate": 1.252925440065272e-05,
+ "loss": 0.0956,
+ "step": 16003
+ },
+ {
+ "epoch": 43.607629427792915,
+ "grad_norm": 3.18112850189209,
+ "learning_rate": 1.2528400589799805e-05,
+ "loss": 0.1052,
+ "step": 16004
+ },
+ {
+ "epoch": 43.61035422343324,
+ "grad_norm": 2.7689199447631836,
+ "learning_rate": 1.2527546759255786e-05,
+ "loss": 0.078,
+ "step": 16005
+ },
+ {
+ "epoch": 43.61307901907357,
+ "grad_norm": 8.107528686523438,
+ "learning_rate": 1.2526692909027314e-05,
+ "loss": 0.1312,
+ "step": 16006
+ },
+ {
+ "epoch": 43.6158038147139,
+ "grad_norm": 2.650664806365967,
+ "learning_rate": 1.252583903912104e-05,
+ "loss": 0.2092,
+ "step": 16007
+ },
+ {
+ "epoch": 43.618528610354225,
+ "grad_norm": 2.6998093128204346,
+ "learning_rate": 1.2524985149543616e-05,
+ "loss": 0.233,
+ "step": 16008
+ },
+ {
+ "epoch": 43.62125340599455,
+ "grad_norm": 3.278795003890991,
+ "learning_rate": 1.2524131240301684e-05,
+ "loss": 0.1518,
+ "step": 16009
+ },
+ {
+ "epoch": 43.62397820163488,
+ "grad_norm": 3.297494888305664,
+ "learning_rate": 1.2523277311401903e-05,
+ "loss": 0.1576,
+ "step": 16010
+ },
+ {
+ "epoch": 43.6267029972752,
+ "grad_norm": 3.331165313720703,
+ "learning_rate": 1.2522423362850917e-05,
+ "loss": 0.1488,
+ "step": 16011
+ },
+ {
+ "epoch": 43.62942779291553,
+ "grad_norm": 3.698153018951416,
+ "learning_rate": 1.2521569394655382e-05,
+ "loss": 0.29,
+ "step": 16012
+ },
+ {
+ "epoch": 43.63215258855586,
+ "grad_norm": 3.9088289737701416,
+ "learning_rate": 1.2520715406821942e-05,
+ "loss": 0.0888,
+ "step": 16013
+ },
+ {
+ "epoch": 43.63487738419619,
+ "grad_norm": 3.5238325595855713,
+ "learning_rate": 1.2519861399357254e-05,
+ "loss": 0.0794,
+ "step": 16014
+ },
+ {
+ "epoch": 43.63760217983651,
+ "grad_norm": 3.2884693145751953,
+ "learning_rate": 1.2519007372267964e-05,
+ "loss": 0.144,
+ "step": 16015
+ },
+ {
+ "epoch": 43.64032697547684,
+ "grad_norm": 2.7937605381011963,
+ "learning_rate": 1.2518153325560726e-05,
+ "loss": 0.0645,
+ "step": 16016
+ },
+ {
+ "epoch": 43.643051771117165,
+ "grad_norm": 3.3499011993408203,
+ "learning_rate": 1.2517299259242193e-05,
+ "loss": 0.0965,
+ "step": 16017
+ },
+ {
+ "epoch": 43.64577656675749,
+ "grad_norm": 3.6325318813323975,
+ "learning_rate": 1.2516445173319013e-05,
+ "loss": 0.1268,
+ "step": 16018
+ },
+ {
+ "epoch": 43.64850136239782,
+ "grad_norm": 2.925403356552124,
+ "learning_rate": 1.2515591067797841e-05,
+ "loss": 0.1986,
+ "step": 16019
+ },
+ {
+ "epoch": 43.65122615803815,
+ "grad_norm": 3.6015114784240723,
+ "learning_rate": 1.2514736942685326e-05,
+ "loss": 0.0885,
+ "step": 16020
+ },
+ {
+ "epoch": 43.653950953678475,
+ "grad_norm": 3.3991777896881104,
+ "learning_rate": 1.251388279798812e-05,
+ "loss": 0.0862,
+ "step": 16021
+ },
+ {
+ "epoch": 43.6566757493188,
+ "grad_norm": 5.408688068389893,
+ "learning_rate": 1.2513028633712875e-05,
+ "loss": 0.1309,
+ "step": 16022
+ },
+ {
+ "epoch": 43.65940054495913,
+ "grad_norm": 3.7332587242126465,
+ "learning_rate": 1.2512174449866243e-05,
+ "loss": 0.3124,
+ "step": 16023
+ },
+ {
+ "epoch": 43.66212534059945,
+ "grad_norm": 4.7245306968688965,
+ "learning_rate": 1.251132024645488e-05,
+ "loss": 0.2228,
+ "step": 16024
+ },
+ {
+ "epoch": 43.664850136239785,
+ "grad_norm": 3.1114742755889893,
+ "learning_rate": 1.2510466023485435e-05,
+ "loss": 0.0949,
+ "step": 16025
+ },
+ {
+ "epoch": 43.66757493188011,
+ "grad_norm": 3.0631864070892334,
+ "learning_rate": 1.2509611780964562e-05,
+ "loss": 0.1265,
+ "step": 16026
+ },
+ {
+ "epoch": 43.67029972752044,
+ "grad_norm": 3.1786084175109863,
+ "learning_rate": 1.2508757518898911e-05,
+ "loss": 0.1101,
+ "step": 16027
+ },
+ {
+ "epoch": 43.67302452316076,
+ "grad_norm": 2.685044765472412,
+ "learning_rate": 1.2507903237295138e-05,
+ "loss": 0.074,
+ "step": 16028
+ },
+ {
+ "epoch": 43.67574931880109,
+ "grad_norm": 3.3083555698394775,
+ "learning_rate": 1.2507048936159896e-05,
+ "loss": 0.0931,
+ "step": 16029
+ },
+ {
+ "epoch": 43.678474114441414,
+ "grad_norm": 3.524705171585083,
+ "learning_rate": 1.2506194615499836e-05,
+ "loss": 0.1606,
+ "step": 16030
+ },
+ {
+ "epoch": 43.68119891008175,
+ "grad_norm": 5.751564025878906,
+ "learning_rate": 1.2505340275321613e-05,
+ "loss": 0.1896,
+ "step": 16031
+ },
+ {
+ "epoch": 43.68392370572207,
+ "grad_norm": 2.5893795490264893,
+ "learning_rate": 1.2504485915631883e-05,
+ "loss": 0.0691,
+ "step": 16032
+ },
+ {
+ "epoch": 43.6866485013624,
+ "grad_norm": 3.0518319606781006,
+ "learning_rate": 1.2503631536437295e-05,
+ "loss": 0.102,
+ "step": 16033
+ },
+ {
+ "epoch": 43.689373297002724,
+ "grad_norm": 3.149669647216797,
+ "learning_rate": 1.2502777137744507e-05,
+ "loss": 0.1136,
+ "step": 16034
+ },
+ {
+ "epoch": 43.69209809264305,
+ "grad_norm": 3.635786533355713,
+ "learning_rate": 1.250192271956017e-05,
+ "loss": 0.0869,
+ "step": 16035
+ },
+ {
+ "epoch": 43.694822888283376,
+ "grad_norm": 2.943920612335205,
+ "learning_rate": 1.250106828189094e-05,
+ "loss": 0.0764,
+ "step": 16036
+ },
+ {
+ "epoch": 43.69754768392371,
+ "grad_norm": 3.334703207015991,
+ "learning_rate": 1.2500213824743472e-05,
+ "loss": 0.1024,
+ "step": 16037
+ },
+ {
+ "epoch": 43.700272479564035,
+ "grad_norm": 2.8669371604919434,
+ "learning_rate": 1.2499359348124418e-05,
+ "loss": 0.0945,
+ "step": 16038
+ },
+ {
+ "epoch": 43.70299727520436,
+ "grad_norm": 3.2310965061187744,
+ "learning_rate": 1.2498504852040433e-05,
+ "loss": 0.1355,
+ "step": 16039
+ },
+ {
+ "epoch": 43.705722070844686,
+ "grad_norm": 3.356686592102051,
+ "learning_rate": 1.2497650336498172e-05,
+ "loss": 0.0769,
+ "step": 16040
+ },
+ {
+ "epoch": 43.70844686648501,
+ "grad_norm": 2.813737630844116,
+ "learning_rate": 1.2496795801504291e-05,
+ "loss": 0.082,
+ "step": 16041
+ },
+ {
+ "epoch": 43.71117166212534,
+ "grad_norm": 2.9044573307037354,
+ "learning_rate": 1.2495941247065445e-05,
+ "loss": 0.0661,
+ "step": 16042
+ },
+ {
+ "epoch": 43.71389645776567,
+ "grad_norm": 2.847262144088745,
+ "learning_rate": 1.2495086673188293e-05,
+ "loss": 0.156,
+ "step": 16043
+ },
+ {
+ "epoch": 43.716621253406,
+ "grad_norm": 3.5052716732025146,
+ "learning_rate": 1.249423207987948e-05,
+ "loss": 0.1164,
+ "step": 16044
+ },
+ {
+ "epoch": 43.71934604904632,
+ "grad_norm": 4.005692005157471,
+ "learning_rate": 1.249337746714567e-05,
+ "loss": 0.1255,
+ "step": 16045
+ },
+ {
+ "epoch": 43.72207084468665,
+ "grad_norm": 3.196399211883545,
+ "learning_rate": 1.2492522834993519e-05,
+ "loss": 0.1065,
+ "step": 16046
+ },
+ {
+ "epoch": 43.724795640326974,
+ "grad_norm": 3.101074457168579,
+ "learning_rate": 1.2491668183429679e-05,
+ "loss": 0.0894,
+ "step": 16047
+ },
+ {
+ "epoch": 43.7275204359673,
+ "grad_norm": 4.538997173309326,
+ "learning_rate": 1.2490813512460806e-05,
+ "loss": 0.1114,
+ "step": 16048
+ },
+ {
+ "epoch": 43.73024523160763,
+ "grad_norm": 2.718318223953247,
+ "learning_rate": 1.2489958822093557e-05,
+ "loss": 0.058,
+ "step": 16049
+ },
+ {
+ "epoch": 43.73297002724796,
+ "grad_norm": 6.886080741882324,
+ "learning_rate": 1.2489104112334588e-05,
+ "loss": 0.0759,
+ "step": 16050
+ },
+ {
+ "epoch": 43.735694822888284,
+ "grad_norm": 3.9172232151031494,
+ "learning_rate": 1.248824938319056e-05,
+ "loss": 0.115,
+ "step": 16051
+ },
+ {
+ "epoch": 43.73841961852861,
+ "grad_norm": 3.9740145206451416,
+ "learning_rate": 1.2487394634668123e-05,
+ "loss": 0.0918,
+ "step": 16052
+ },
+ {
+ "epoch": 43.741144414168936,
+ "grad_norm": 2.973994731903076,
+ "learning_rate": 1.2486539866773941e-05,
+ "loss": 0.1052,
+ "step": 16053
+ },
+ {
+ "epoch": 43.74386920980926,
+ "grad_norm": 5.534222602844238,
+ "learning_rate": 1.2485685079514663e-05,
+ "loss": 0.1109,
+ "step": 16054
+ },
+ {
+ "epoch": 43.746594005449595,
+ "grad_norm": 3.393031597137451,
+ "learning_rate": 1.2484830272896951e-05,
+ "loss": 0.1226,
+ "step": 16055
+ },
+ {
+ "epoch": 43.74931880108992,
+ "grad_norm": 2.8080520629882812,
+ "learning_rate": 1.2483975446927459e-05,
+ "loss": 0.1286,
+ "step": 16056
+ },
+ {
+ "epoch": 43.752043596730246,
+ "grad_norm": 3.269775629043579,
+ "learning_rate": 1.2483120601612847e-05,
+ "loss": 0.0882,
+ "step": 16057
+ },
+ {
+ "epoch": 43.75476839237057,
+ "grad_norm": 3.222670793533325,
+ "learning_rate": 1.2482265736959769e-05,
+ "loss": 0.0782,
+ "step": 16058
+ },
+ {
+ "epoch": 43.7574931880109,
+ "grad_norm": 3.475343704223633,
+ "learning_rate": 1.248141085297489e-05,
+ "loss": 0.2071,
+ "step": 16059
+ },
+ {
+ "epoch": 43.76021798365122,
+ "grad_norm": 3.9062869548797607,
+ "learning_rate": 1.2480555949664859e-05,
+ "loss": 0.0762,
+ "step": 16060
+ },
+ {
+ "epoch": 43.762942779291556,
+ "grad_norm": 3.350677490234375,
+ "learning_rate": 1.247970102703634e-05,
+ "loss": 0.1817,
+ "step": 16061
+ },
+ {
+ "epoch": 43.76566757493188,
+ "grad_norm": 3.766378164291382,
+ "learning_rate": 1.247884608509599e-05,
+ "loss": 0.0912,
+ "step": 16062
+ },
+ {
+ "epoch": 43.76839237057221,
+ "grad_norm": 3.163710832595825,
+ "learning_rate": 1.2477991123850468e-05,
+ "loss": 0.1254,
+ "step": 16063
+ },
+ {
+ "epoch": 43.771117166212534,
+ "grad_norm": 2.6450605392456055,
+ "learning_rate": 1.2477136143306428e-05,
+ "loss": 0.0599,
+ "step": 16064
+ },
+ {
+ "epoch": 43.77384196185286,
+ "grad_norm": 5.736313819885254,
+ "learning_rate": 1.2476281143470535e-05,
+ "loss": 0.0721,
+ "step": 16065
+ },
+ {
+ "epoch": 43.776566757493185,
+ "grad_norm": 4.018364429473877,
+ "learning_rate": 1.2475426124349436e-05,
+ "loss": 0.1097,
+ "step": 16066
+ },
+ {
+ "epoch": 43.77929155313352,
+ "grad_norm": 2.7099647521972656,
+ "learning_rate": 1.2474571085949807e-05,
+ "loss": 0.2073,
+ "step": 16067
+ },
+ {
+ "epoch": 43.782016348773844,
+ "grad_norm": 3.218003749847412,
+ "learning_rate": 1.2473716028278292e-05,
+ "loss": 0.4198,
+ "step": 16068
+ },
+ {
+ "epoch": 43.78474114441417,
+ "grad_norm": 3.19840145111084,
+ "learning_rate": 1.2472860951341561e-05,
+ "loss": 0.0782,
+ "step": 16069
+ },
+ {
+ "epoch": 43.787465940054496,
+ "grad_norm": 2.9666388034820557,
+ "learning_rate": 1.2472005855146265e-05,
+ "loss": 0.1053,
+ "step": 16070
+ },
+ {
+ "epoch": 43.79019073569482,
+ "grad_norm": 2.8169655799865723,
+ "learning_rate": 1.247115073969907e-05,
+ "loss": 0.0611,
+ "step": 16071
+ },
+ {
+ "epoch": 43.79291553133515,
+ "grad_norm": 3.6449246406555176,
+ "learning_rate": 1.2470295605006628e-05,
+ "loss": 0.097,
+ "step": 16072
+ },
+ {
+ "epoch": 43.79564032697548,
+ "grad_norm": 3.043303966522217,
+ "learning_rate": 1.246944045107561e-05,
+ "loss": 0.0728,
+ "step": 16073
+ },
+ {
+ "epoch": 43.798365122615806,
+ "grad_norm": 3.1731605529785156,
+ "learning_rate": 1.2468585277912664e-05,
+ "loss": 0.1109,
+ "step": 16074
+ },
+ {
+ "epoch": 43.80108991825613,
+ "grad_norm": 3.372666835784912,
+ "learning_rate": 1.2467730085524454e-05,
+ "loss": 0.0904,
+ "step": 16075
+ },
+ {
+ "epoch": 43.80381471389646,
+ "grad_norm": 3.7683265209198,
+ "learning_rate": 1.2466874873917646e-05,
+ "loss": 0.0883,
+ "step": 16076
+ },
+ {
+ "epoch": 43.80653950953678,
+ "grad_norm": 2.793830394744873,
+ "learning_rate": 1.2466019643098895e-05,
+ "loss": 0.0975,
+ "step": 16077
+ },
+ {
+ "epoch": 43.80926430517711,
+ "grad_norm": 3.9033257961273193,
+ "learning_rate": 1.2465164393074859e-05,
+ "loss": 0.1888,
+ "step": 16078
+ },
+ {
+ "epoch": 43.81198910081744,
+ "grad_norm": 2.6017003059387207,
+ "learning_rate": 1.2464309123852205e-05,
+ "loss": 0.0759,
+ "step": 16079
+ },
+ {
+ "epoch": 43.81471389645777,
+ "grad_norm": 3.231499433517456,
+ "learning_rate": 1.246345383543759e-05,
+ "loss": 0.1162,
+ "step": 16080
+ },
+ {
+ "epoch": 43.817438692098094,
+ "grad_norm": 3.0713694095611572,
+ "learning_rate": 1.2462598527837678e-05,
+ "loss": 0.1041,
+ "step": 16081
+ },
+ {
+ "epoch": 43.82016348773842,
+ "grad_norm": 4.138360977172852,
+ "learning_rate": 1.2461743201059124e-05,
+ "loss": 0.1691,
+ "step": 16082
+ },
+ {
+ "epoch": 43.822888283378745,
+ "grad_norm": 2.712207555770874,
+ "learning_rate": 1.2460887855108597e-05,
+ "loss": 0.1062,
+ "step": 16083
+ },
+ {
+ "epoch": 43.82561307901907,
+ "grad_norm": 3.4238572120666504,
+ "learning_rate": 1.2460032489992751e-05,
+ "loss": 0.0807,
+ "step": 16084
+ },
+ {
+ "epoch": 43.828337874659404,
+ "grad_norm": 3.186703681945801,
+ "learning_rate": 1.2459177105718251e-05,
+ "loss": 0.0693,
+ "step": 16085
+ },
+ {
+ "epoch": 43.83106267029973,
+ "grad_norm": 3.233128547668457,
+ "learning_rate": 1.2458321702291762e-05,
+ "loss": 0.0999,
+ "step": 16086
+ },
+ {
+ "epoch": 43.833787465940055,
+ "grad_norm": 3.0699033737182617,
+ "learning_rate": 1.2457466279719943e-05,
+ "loss": 0.0662,
+ "step": 16087
+ },
+ {
+ "epoch": 43.83651226158038,
+ "grad_norm": 4.1437835693359375,
+ "learning_rate": 1.2456610838009454e-05,
+ "loss": 0.1533,
+ "step": 16088
+ },
+ {
+ "epoch": 43.83923705722071,
+ "grad_norm": 2.482172727584839,
+ "learning_rate": 1.2455755377166958e-05,
+ "loss": 0.0592,
+ "step": 16089
+ },
+ {
+ "epoch": 43.84196185286103,
+ "grad_norm": 2.770381450653076,
+ "learning_rate": 1.245489989719912e-05,
+ "loss": 0.0499,
+ "step": 16090
+ },
+ {
+ "epoch": 43.844686648501366,
+ "grad_norm": 2.942105770111084,
+ "learning_rate": 1.2454044398112602e-05,
+ "loss": 0.1823,
+ "step": 16091
+ },
+ {
+ "epoch": 43.84741144414169,
+ "grad_norm": 3.1841623783111572,
+ "learning_rate": 1.245318887991406e-05,
+ "loss": 0.074,
+ "step": 16092
+ },
+ {
+ "epoch": 43.85013623978202,
+ "grad_norm": 3.91001558303833,
+ "learning_rate": 1.2452333342610166e-05,
+ "loss": 0.1168,
+ "step": 16093
+ },
+ {
+ "epoch": 43.85286103542234,
+ "grad_norm": 3.4322311878204346,
+ "learning_rate": 1.2451477786207575e-05,
+ "loss": 0.1082,
+ "step": 16094
+ },
+ {
+ "epoch": 43.85558583106267,
+ "grad_norm": 2.7400777339935303,
+ "learning_rate": 1.245062221071296e-05,
+ "loss": 0.1345,
+ "step": 16095
+ },
+ {
+ "epoch": 43.858310626702995,
+ "grad_norm": 3.277681827545166,
+ "learning_rate": 1.2449766616132973e-05,
+ "loss": 0.121,
+ "step": 16096
+ },
+ {
+ "epoch": 43.86103542234333,
+ "grad_norm": 3.3407142162323,
+ "learning_rate": 1.2448911002474287e-05,
+ "loss": 0.058,
+ "step": 16097
+ },
+ {
+ "epoch": 43.86376021798365,
+ "grad_norm": 2.5321927070617676,
+ "learning_rate": 1.2448055369743557e-05,
+ "loss": 0.0555,
+ "step": 16098
+ },
+ {
+ "epoch": 43.86648501362398,
+ "grad_norm": 3.200728178024292,
+ "learning_rate": 1.2447199717947451e-05,
+ "loss": 0.0812,
+ "step": 16099
+ },
+ {
+ "epoch": 43.869209809264305,
+ "grad_norm": 3.6914327144622803,
+ "learning_rate": 1.2446344047092633e-05,
+ "loss": 0.2435,
+ "step": 16100
+ },
+ {
+ "epoch": 43.87193460490463,
+ "grad_norm": 2.677368402481079,
+ "learning_rate": 1.2445488357185766e-05,
+ "loss": 0.0466,
+ "step": 16101
+ },
+ {
+ "epoch": 43.87465940054496,
+ "grad_norm": 3.35286545753479,
+ "learning_rate": 1.2444632648233517e-05,
+ "loss": 0.1626,
+ "step": 16102
+ },
+ {
+ "epoch": 43.87738419618529,
+ "grad_norm": 3.3959836959838867,
+ "learning_rate": 1.2443776920242542e-05,
+ "loss": 0.2226,
+ "step": 16103
+ },
+ {
+ "epoch": 43.880108991825615,
+ "grad_norm": 3.164189338684082,
+ "learning_rate": 1.2442921173219515e-05,
+ "loss": 0.0938,
+ "step": 16104
+ },
+ {
+ "epoch": 43.88283378746594,
+ "grad_norm": 3.514540433883667,
+ "learning_rate": 1.2442065407171098e-05,
+ "loss": 0.099,
+ "step": 16105
+ },
+ {
+ "epoch": 43.88555858310627,
+ "grad_norm": 2.9390058517456055,
+ "learning_rate": 1.2441209622103951e-05,
+ "loss": 0.1041,
+ "step": 16106
+ },
+ {
+ "epoch": 43.88828337874659,
+ "grad_norm": 3.9191110134124756,
+ "learning_rate": 1.2440353818024744e-05,
+ "loss": 0.1233,
+ "step": 16107
+ },
+ {
+ "epoch": 43.89100817438692,
+ "grad_norm": 3.7221174240112305,
+ "learning_rate": 1.243949799494014e-05,
+ "loss": 0.1464,
+ "step": 16108
+ },
+ {
+ "epoch": 43.89373297002725,
+ "grad_norm": 3.3282742500305176,
+ "learning_rate": 1.2438642152856802e-05,
+ "loss": 0.1556,
+ "step": 16109
+ },
+ {
+ "epoch": 43.89645776566758,
+ "grad_norm": 2.9121644496917725,
+ "learning_rate": 1.24377862917814e-05,
+ "loss": 0.0939,
+ "step": 16110
+ },
+ {
+ "epoch": 43.8991825613079,
+ "grad_norm": 3.4605414867401123,
+ "learning_rate": 1.2436930411720594e-05,
+ "loss": 0.0819,
+ "step": 16111
+ },
+ {
+ "epoch": 43.90190735694823,
+ "grad_norm": 2.7784550189971924,
+ "learning_rate": 1.2436074512681055e-05,
+ "loss": 0.0674,
+ "step": 16112
+ },
+ {
+ "epoch": 43.904632152588555,
+ "grad_norm": 4.3147454261779785,
+ "learning_rate": 1.2435218594669445e-05,
+ "loss": 0.088,
+ "step": 16113
+ },
+ {
+ "epoch": 43.90735694822888,
+ "grad_norm": 3.6245622634887695,
+ "learning_rate": 1.2434362657692432e-05,
+ "loss": 0.0917,
+ "step": 16114
+ },
+ {
+ "epoch": 43.91008174386921,
+ "grad_norm": 3.140012264251709,
+ "learning_rate": 1.2433506701756678e-05,
+ "loss": 0.154,
+ "step": 16115
+ },
+ {
+ "epoch": 43.91280653950954,
+ "grad_norm": 4.168983459472656,
+ "learning_rate": 1.2432650726868854e-05,
+ "loss": 0.2127,
+ "step": 16116
+ },
+ {
+ "epoch": 43.915531335149865,
+ "grad_norm": 3.2901320457458496,
+ "learning_rate": 1.2431794733035625e-05,
+ "loss": 0.0733,
+ "step": 16117
+ },
+ {
+ "epoch": 43.91825613079019,
+ "grad_norm": 3.4337501525878906,
+ "learning_rate": 1.2430938720263657e-05,
+ "loss": 0.1398,
+ "step": 16118
+ },
+ {
+ "epoch": 43.920980926430516,
+ "grad_norm": 3.2893893718719482,
+ "learning_rate": 1.2430082688559616e-05,
+ "loss": 0.1983,
+ "step": 16119
+ },
+ {
+ "epoch": 43.92370572207084,
+ "grad_norm": 2.928095817565918,
+ "learning_rate": 1.242922663793017e-05,
+ "loss": 0.1238,
+ "step": 16120
+ },
+ {
+ "epoch": 43.926430517711175,
+ "grad_norm": 4.040589809417725,
+ "learning_rate": 1.2428370568381982e-05,
+ "loss": 0.132,
+ "step": 16121
+ },
+ {
+ "epoch": 43.9291553133515,
+ "grad_norm": 3.972342014312744,
+ "learning_rate": 1.2427514479921725e-05,
+ "loss": 0.0863,
+ "step": 16122
+ },
+ {
+ "epoch": 43.93188010899183,
+ "grad_norm": 3.6250386238098145,
+ "learning_rate": 1.2426658372556062e-05,
+ "loss": 0.0785,
+ "step": 16123
+ },
+ {
+ "epoch": 43.93460490463215,
+ "grad_norm": 3.414656400680542,
+ "learning_rate": 1.2425802246291664e-05,
+ "loss": 0.1256,
+ "step": 16124
+ },
+ {
+ "epoch": 43.93732970027248,
+ "grad_norm": 3.021574020385742,
+ "learning_rate": 1.2424946101135193e-05,
+ "loss": 0.0991,
+ "step": 16125
+ },
+ {
+ "epoch": 43.940054495912804,
+ "grad_norm": 3.544116497039795,
+ "learning_rate": 1.2424089937093324e-05,
+ "loss": 0.1249,
+ "step": 16126
+ },
+ {
+ "epoch": 43.94277929155314,
+ "grad_norm": 2.649693727493286,
+ "learning_rate": 1.2423233754172714e-05,
+ "loss": 0.0751,
+ "step": 16127
+ },
+ {
+ "epoch": 43.94550408719346,
+ "grad_norm": 2.996037721633911,
+ "learning_rate": 1.2422377552380045e-05,
+ "loss": 0.1471,
+ "step": 16128
+ },
+ {
+ "epoch": 43.94822888283379,
+ "grad_norm": 3.955778121948242,
+ "learning_rate": 1.2421521331721974e-05,
+ "loss": 0.1304,
+ "step": 16129
+ },
+ {
+ "epoch": 43.950953678474114,
+ "grad_norm": 3.1556715965270996,
+ "learning_rate": 1.2420665092205173e-05,
+ "loss": 0.1984,
+ "step": 16130
+ },
+ {
+ "epoch": 43.95367847411444,
+ "grad_norm": 2.6070716381073,
+ "learning_rate": 1.2419808833836309e-05,
+ "loss": 0.0844,
+ "step": 16131
+ },
+ {
+ "epoch": 43.956403269754766,
+ "grad_norm": 3.165370225906372,
+ "learning_rate": 1.2418952556622055e-05,
+ "loss": 0.1581,
+ "step": 16132
+ },
+ {
+ "epoch": 43.95912806539509,
+ "grad_norm": 2.8659894466400146,
+ "learning_rate": 1.2418096260569072e-05,
+ "loss": 0.06,
+ "step": 16133
+ },
+ {
+ "epoch": 43.961852861035425,
+ "grad_norm": 2.8613643646240234,
+ "learning_rate": 1.2417239945684036e-05,
+ "loss": 0.1499,
+ "step": 16134
+ },
+ {
+ "epoch": 43.96457765667575,
+ "grad_norm": 3.186699628829956,
+ "learning_rate": 1.241638361197361e-05,
+ "loss": 0.0987,
+ "step": 16135
+ },
+ {
+ "epoch": 43.967302452316076,
+ "grad_norm": 2.6490190029144287,
+ "learning_rate": 1.2415527259444471e-05,
+ "loss": 0.1035,
+ "step": 16136
+ },
+ {
+ "epoch": 43.9700272479564,
+ "grad_norm": 3.823216199874878,
+ "learning_rate": 1.2414670888103279e-05,
+ "loss": 0.2041,
+ "step": 16137
+ },
+ {
+ "epoch": 43.97275204359673,
+ "grad_norm": 3.068735361099243,
+ "learning_rate": 1.241381449795671e-05,
+ "loss": 0.0863,
+ "step": 16138
+ },
+ {
+ "epoch": 43.97547683923706,
+ "grad_norm": 2.5379762649536133,
+ "learning_rate": 1.2412958089011433e-05,
+ "loss": 0.085,
+ "step": 16139
+ },
+ {
+ "epoch": 43.97820163487739,
+ "grad_norm": 3.7545859813690186,
+ "learning_rate": 1.2412101661274114e-05,
+ "loss": 0.207,
+ "step": 16140
+ },
+ {
+ "epoch": 43.98092643051771,
+ "grad_norm": 3.168057918548584,
+ "learning_rate": 1.2411245214751426e-05,
+ "loss": 0.1287,
+ "step": 16141
+ },
+ {
+ "epoch": 43.98365122615804,
+ "grad_norm": 3.1474170684814453,
+ "learning_rate": 1.2410388749450038e-05,
+ "loss": 0.1112,
+ "step": 16142
+ },
+ {
+ "epoch": 43.986376021798364,
+ "grad_norm": 5.204498767852783,
+ "learning_rate": 1.2409532265376618e-05,
+ "loss": 0.1145,
+ "step": 16143
+ },
+ {
+ "epoch": 43.98910081743869,
+ "grad_norm": 4.074854850769043,
+ "learning_rate": 1.2408675762537838e-05,
+ "loss": 0.1076,
+ "step": 16144
+ },
+ {
+ "epoch": 43.991825613079016,
+ "grad_norm": 2.973233222961426,
+ "learning_rate": 1.240781924094037e-05,
+ "loss": 0.1356,
+ "step": 16145
+ },
+ {
+ "epoch": 43.99455040871935,
+ "grad_norm": 3.1633095741271973,
+ "learning_rate": 1.2406962700590884e-05,
+ "loss": 0.1292,
+ "step": 16146
+ },
+ {
+ "epoch": 43.997275204359674,
+ "grad_norm": 2.1182827949523926,
+ "learning_rate": 1.2406106141496049e-05,
+ "loss": 0.046,
+ "step": 16147
+ },
+ {
+ "epoch": 44.0,
+ "grad_norm": 3.9092257022857666,
+ "learning_rate": 1.2405249563662539e-05,
+ "loss": 0.1039,
+ "step": 16148
+ },
+ {
+ "epoch": 44.002724795640326,
+ "grad_norm": 3.4239137172698975,
+ "learning_rate": 1.2404392967097019e-05,
+ "loss": 0.0673,
+ "step": 16149
+ },
+ {
+ "epoch": 44.00544959128065,
+ "grad_norm": 2.8760414123535156,
+ "learning_rate": 1.2403536351806167e-05,
+ "loss": 0.0658,
+ "step": 16150
+ },
+ {
+ "epoch": 44.00817438692098,
+ "grad_norm": 2.862652063369751,
+ "learning_rate": 1.2402679717796651e-05,
+ "loss": 0.2222,
+ "step": 16151
+ },
+ {
+ "epoch": 44.01089918256131,
+ "grad_norm": 4.01659631729126,
+ "learning_rate": 1.240182306507514e-05,
+ "loss": 0.156,
+ "step": 16152
+ },
+ {
+ "epoch": 44.013623978201636,
+ "grad_norm": 5.979721546173096,
+ "learning_rate": 1.2400966393648309e-05,
+ "loss": 0.1711,
+ "step": 16153
+ },
+ {
+ "epoch": 44.01634877384196,
+ "grad_norm": 3.1475019454956055,
+ "learning_rate": 1.2400109703522831e-05,
+ "loss": 0.0908,
+ "step": 16154
+ },
+ {
+ "epoch": 44.01907356948229,
+ "grad_norm": 2.8161003589630127,
+ "learning_rate": 1.2399252994705373e-05,
+ "loss": 0.0758,
+ "step": 16155
+ },
+ {
+ "epoch": 44.02179836512261,
+ "grad_norm": 3.531716823577881,
+ "learning_rate": 1.2398396267202615e-05,
+ "loss": 0.061,
+ "step": 16156
+ },
+ {
+ "epoch": 44.02452316076294,
+ "grad_norm": 2.881636619567871,
+ "learning_rate": 1.239753952102122e-05,
+ "loss": 0.0806,
+ "step": 16157
+ },
+ {
+ "epoch": 44.02724795640327,
+ "grad_norm": 3.003209114074707,
+ "learning_rate": 1.2396682756167866e-05,
+ "loss": 0.1528,
+ "step": 16158
+ },
+ {
+ "epoch": 44.0299727520436,
+ "grad_norm": 2.2352070808410645,
+ "learning_rate": 1.2395825972649222e-05,
+ "loss": 0.0548,
+ "step": 16159
+ },
+ {
+ "epoch": 44.032697547683924,
+ "grad_norm": 3.711057424545288,
+ "learning_rate": 1.2394969170471965e-05,
+ "loss": 0.1398,
+ "step": 16160
+ },
+ {
+ "epoch": 44.03542234332425,
+ "grad_norm": 2.73334002494812,
+ "learning_rate": 1.2394112349642761e-05,
+ "loss": 0.0452,
+ "step": 16161
+ },
+ {
+ "epoch": 44.038147138964575,
+ "grad_norm": 2.9782791137695312,
+ "learning_rate": 1.2393255510168292e-05,
+ "loss": 0.1102,
+ "step": 16162
+ },
+ {
+ "epoch": 44.0408719346049,
+ "grad_norm": 3.0706417560577393,
+ "learning_rate": 1.2392398652055222e-05,
+ "loss": 0.1085,
+ "step": 16163
+ },
+ {
+ "epoch": 44.043596730245234,
+ "grad_norm": 2.726313591003418,
+ "learning_rate": 1.2391541775310231e-05,
+ "loss": 0.1383,
+ "step": 16164
+ },
+ {
+ "epoch": 44.04632152588556,
+ "grad_norm": 3.247169256210327,
+ "learning_rate": 1.2390684879939989e-05,
+ "loss": 0.3237,
+ "step": 16165
+ },
+ {
+ "epoch": 44.049046321525886,
+ "grad_norm": 2.7690725326538086,
+ "learning_rate": 1.238982796595117e-05,
+ "loss": 0.1745,
+ "step": 16166
+ },
+ {
+ "epoch": 44.05177111716621,
+ "grad_norm": 2.569321632385254,
+ "learning_rate": 1.2388971033350447e-05,
+ "loss": 0.0587,
+ "step": 16167
+ },
+ {
+ "epoch": 44.05449591280654,
+ "grad_norm": 2.2494537830352783,
+ "learning_rate": 1.2388114082144495e-05,
+ "loss": 0.0686,
+ "step": 16168
+ },
+ {
+ "epoch": 44.05722070844686,
+ "grad_norm": 3.396137237548828,
+ "learning_rate": 1.2387257112339987e-05,
+ "loss": 0.1047,
+ "step": 16169
+ },
+ {
+ "epoch": 44.059945504087196,
+ "grad_norm": 3.1041862964630127,
+ "learning_rate": 1.2386400123943597e-05,
+ "loss": 0.2645,
+ "step": 16170
+ },
+ {
+ "epoch": 44.06267029972752,
+ "grad_norm": 3.082643508911133,
+ "learning_rate": 1.2385543116962002e-05,
+ "loss": 0.0715,
+ "step": 16171
+ },
+ {
+ "epoch": 44.06539509536785,
+ "grad_norm": 2.8364689350128174,
+ "learning_rate": 1.2384686091401874e-05,
+ "loss": 0.0731,
+ "step": 16172
+ },
+ {
+ "epoch": 44.06811989100817,
+ "grad_norm": 2.660191774368286,
+ "learning_rate": 1.2383829047269884e-05,
+ "loss": 0.1728,
+ "step": 16173
+ },
+ {
+ "epoch": 44.0708446866485,
+ "grad_norm": 5.508583068847656,
+ "learning_rate": 1.2382971984572714e-05,
+ "loss": 0.264,
+ "step": 16174
+ },
+ {
+ "epoch": 44.073569482288825,
+ "grad_norm": 3.1493091583251953,
+ "learning_rate": 1.2382114903317033e-05,
+ "loss": 0.0834,
+ "step": 16175
+ },
+ {
+ "epoch": 44.07629427792916,
+ "grad_norm": 3.552441358566284,
+ "learning_rate": 1.2381257803509518e-05,
+ "loss": 0.161,
+ "step": 16176
+ },
+ {
+ "epoch": 44.079019073569484,
+ "grad_norm": 2.7047719955444336,
+ "learning_rate": 1.2380400685156844e-05,
+ "loss": 0.0896,
+ "step": 16177
+ },
+ {
+ "epoch": 44.08174386920981,
+ "grad_norm": 2.5555336475372314,
+ "learning_rate": 1.2379543548265685e-05,
+ "loss": 0.1255,
+ "step": 16178
+ },
+ {
+ "epoch": 44.084468664850135,
+ "grad_norm": 2.345309019088745,
+ "learning_rate": 1.2378686392842718e-05,
+ "loss": 0.1431,
+ "step": 16179
+ },
+ {
+ "epoch": 44.08719346049046,
+ "grad_norm": 2.7408831119537354,
+ "learning_rate": 1.2377829218894618e-05,
+ "loss": 0.058,
+ "step": 16180
+ },
+ {
+ "epoch": 44.08991825613079,
+ "grad_norm": 3.1485280990600586,
+ "learning_rate": 1.237697202642806e-05,
+ "loss": 0.1114,
+ "step": 16181
+ },
+ {
+ "epoch": 44.09264305177112,
+ "grad_norm": 3.186803102493286,
+ "learning_rate": 1.2376114815449723e-05,
+ "loss": 0.1346,
+ "step": 16182
+ },
+ {
+ "epoch": 44.095367847411445,
+ "grad_norm": 2.377411365509033,
+ "learning_rate": 1.2375257585966275e-05,
+ "loss": 0.1479,
+ "step": 16183
+ },
+ {
+ "epoch": 44.09809264305177,
+ "grad_norm": 3.135748863220215,
+ "learning_rate": 1.2374400337984404e-05,
+ "loss": 0.2445,
+ "step": 16184
+ },
+ {
+ "epoch": 44.1008174386921,
+ "grad_norm": 2.589104652404785,
+ "learning_rate": 1.2373543071510774e-05,
+ "loss": 0.0439,
+ "step": 16185
+ },
+ {
+ "epoch": 44.10354223433242,
+ "grad_norm": 2.574885845184326,
+ "learning_rate": 1.2372685786552072e-05,
+ "loss": 0.0686,
+ "step": 16186
+ },
+ {
+ "epoch": 44.10626702997275,
+ "grad_norm": 2.5764594078063965,
+ "learning_rate": 1.2371828483114964e-05,
+ "loss": 0.053,
+ "step": 16187
+ },
+ {
+ "epoch": 44.10899182561308,
+ "grad_norm": 3.0751752853393555,
+ "learning_rate": 1.2370971161206136e-05,
+ "loss": 0.2773,
+ "step": 16188
+ },
+ {
+ "epoch": 44.11171662125341,
+ "grad_norm": 1.817478895187378,
+ "learning_rate": 1.237011382083226e-05,
+ "loss": 0.0553,
+ "step": 16189
+ },
+ {
+ "epoch": 44.11444141689373,
+ "grad_norm": 2.3913588523864746,
+ "learning_rate": 1.2369256462000015e-05,
+ "loss": 0.059,
+ "step": 16190
+ },
+ {
+ "epoch": 44.11716621253406,
+ "grad_norm": 3.092622756958008,
+ "learning_rate": 1.2368399084716072e-05,
+ "loss": 0.1449,
+ "step": 16191
+ },
+ {
+ "epoch": 44.119891008174385,
+ "grad_norm": 1.7544547319412231,
+ "learning_rate": 1.236754168898712e-05,
+ "loss": 0.0443,
+ "step": 16192
+ },
+ {
+ "epoch": 44.12261580381471,
+ "grad_norm": 2.8649418354034424,
+ "learning_rate": 1.2366684274819825e-05,
+ "loss": 0.0718,
+ "step": 16193
+ },
+ {
+ "epoch": 44.12534059945504,
+ "grad_norm": 2.908430814743042,
+ "learning_rate": 1.2365826842220872e-05,
+ "loss": 0.0611,
+ "step": 16194
+ },
+ {
+ "epoch": 44.12806539509537,
+ "grad_norm": 3.0513272285461426,
+ "learning_rate": 1.2364969391196933e-05,
+ "loss": 0.0821,
+ "step": 16195
+ },
+ {
+ "epoch": 44.130790190735695,
+ "grad_norm": 2.828406810760498,
+ "learning_rate": 1.2364111921754686e-05,
+ "loss": 0.0556,
+ "step": 16196
+ },
+ {
+ "epoch": 44.13351498637602,
+ "grad_norm": 2.731182813644409,
+ "learning_rate": 1.2363254433900818e-05,
+ "loss": 0.0695,
+ "step": 16197
+ },
+ {
+ "epoch": 44.13623978201635,
+ "grad_norm": 2.272573947906494,
+ "learning_rate": 1.2362396927641996e-05,
+ "loss": 0.0496,
+ "step": 16198
+ },
+ {
+ "epoch": 44.13896457765667,
+ "grad_norm": 2.187159776687622,
+ "learning_rate": 1.2361539402984905e-05,
+ "loss": 0.1788,
+ "step": 16199
+ },
+ {
+ "epoch": 44.141689373297005,
+ "grad_norm": 2.838644027709961,
+ "learning_rate": 1.2360681859936222e-05,
+ "loss": 0.078,
+ "step": 16200
+ },
+ {
+ "epoch": 44.14441416893733,
+ "grad_norm": 2.408221483230591,
+ "learning_rate": 1.2359824298502622e-05,
+ "loss": 0.0587,
+ "step": 16201
+ },
+ {
+ "epoch": 44.14713896457766,
+ "grad_norm": 3.236161231994629,
+ "learning_rate": 1.2358966718690786e-05,
+ "loss": 0.1371,
+ "step": 16202
+ },
+ {
+ "epoch": 44.14986376021798,
+ "grad_norm": 2.516582727432251,
+ "learning_rate": 1.2358109120507393e-05,
+ "loss": 0.0578,
+ "step": 16203
+ },
+ {
+ "epoch": 44.15258855585831,
+ "grad_norm": 2.473573923110962,
+ "learning_rate": 1.2357251503959125e-05,
+ "loss": 0.0524,
+ "step": 16204
+ },
+ {
+ "epoch": 44.155313351498634,
+ "grad_norm": 2.822120428085327,
+ "learning_rate": 1.2356393869052656e-05,
+ "loss": 0.1057,
+ "step": 16205
+ },
+ {
+ "epoch": 44.15803814713897,
+ "grad_norm": 2.9414663314819336,
+ "learning_rate": 1.2355536215794668e-05,
+ "loss": 0.1037,
+ "step": 16206
+ },
+ {
+ "epoch": 44.16076294277929,
+ "grad_norm": 2.1676483154296875,
+ "learning_rate": 1.235467854419184e-05,
+ "loss": 0.1128,
+ "step": 16207
+ },
+ {
+ "epoch": 44.16348773841962,
+ "grad_norm": 6.507018089294434,
+ "learning_rate": 1.235382085425085e-05,
+ "loss": 0.0733,
+ "step": 16208
+ },
+ {
+ "epoch": 44.166212534059945,
+ "grad_norm": 3.040388584136963,
+ "learning_rate": 1.2352963145978382e-05,
+ "loss": 0.1036,
+ "step": 16209
+ },
+ {
+ "epoch": 44.16893732970027,
+ "grad_norm": 3.156602382659912,
+ "learning_rate": 1.235210541938111e-05,
+ "loss": 0.0745,
+ "step": 16210
+ },
+ {
+ "epoch": 44.171662125340596,
+ "grad_norm": 2.686002731323242,
+ "learning_rate": 1.2351247674465719e-05,
+ "loss": 0.1691,
+ "step": 16211
+ },
+ {
+ "epoch": 44.17438692098093,
+ "grad_norm": 3.445991039276123,
+ "learning_rate": 1.2350389911238884e-05,
+ "loss": 0.1028,
+ "step": 16212
+ },
+ {
+ "epoch": 44.177111716621255,
+ "grad_norm": 2.142117738723755,
+ "learning_rate": 1.2349532129707289e-05,
+ "loss": 0.0927,
+ "step": 16213
+ },
+ {
+ "epoch": 44.17983651226158,
+ "grad_norm": 3.310786008834839,
+ "learning_rate": 1.2348674329877613e-05,
+ "loss": 0.0829,
+ "step": 16214
+ },
+ {
+ "epoch": 44.182561307901906,
+ "grad_norm": 2.7376914024353027,
+ "learning_rate": 1.2347816511756539e-05,
+ "loss": 0.0853,
+ "step": 16215
+ },
+ {
+ "epoch": 44.18528610354223,
+ "grad_norm": 2.919039487838745,
+ "learning_rate": 1.2346958675350742e-05,
+ "loss": 0.0637,
+ "step": 16216
+ },
+ {
+ "epoch": 44.18801089918256,
+ "grad_norm": 3.642732620239258,
+ "learning_rate": 1.234610082066691e-05,
+ "loss": 0.1429,
+ "step": 16217
+ },
+ {
+ "epoch": 44.19073569482289,
+ "grad_norm": 2.6089706420898438,
+ "learning_rate": 1.234524294771172e-05,
+ "loss": 0.0761,
+ "step": 16218
+ },
+ {
+ "epoch": 44.19346049046322,
+ "grad_norm": 2.981519937515259,
+ "learning_rate": 1.2344385056491852e-05,
+ "loss": 0.114,
+ "step": 16219
+ },
+ {
+ "epoch": 44.19618528610354,
+ "grad_norm": 3.4363672733306885,
+ "learning_rate": 1.2343527147013988e-05,
+ "loss": 0.0821,
+ "step": 16220
+ },
+ {
+ "epoch": 44.19891008174387,
+ "grad_norm": 2.7248482704162598,
+ "learning_rate": 1.234266921928481e-05,
+ "loss": 0.0566,
+ "step": 16221
+ },
+ {
+ "epoch": 44.201634877384194,
+ "grad_norm": 2.9162983894348145,
+ "learning_rate": 1.2341811273310999e-05,
+ "loss": 0.0959,
+ "step": 16222
+ },
+ {
+ "epoch": 44.20435967302452,
+ "grad_norm": 4.308824062347412,
+ "learning_rate": 1.234095330909924e-05,
+ "loss": 0.2894,
+ "step": 16223
+ },
+ {
+ "epoch": 44.20708446866485,
+ "grad_norm": 2.870410442352295,
+ "learning_rate": 1.234009532665621e-05,
+ "loss": 0.0603,
+ "step": 16224
+ },
+ {
+ "epoch": 44.20980926430518,
+ "grad_norm": 3.2972731590270996,
+ "learning_rate": 1.2339237325988595e-05,
+ "loss": 0.1182,
+ "step": 16225
+ },
+ {
+ "epoch": 44.212534059945504,
+ "grad_norm": 3.156996726989746,
+ "learning_rate": 1.2338379307103072e-05,
+ "loss": 0.0812,
+ "step": 16226
+ },
+ {
+ "epoch": 44.21525885558583,
+ "grad_norm": 2.558865547180176,
+ "learning_rate": 1.233752127000633e-05,
+ "loss": 0.0627,
+ "step": 16227
+ },
+ {
+ "epoch": 44.217983651226156,
+ "grad_norm": 4.1827616691589355,
+ "learning_rate": 1.2336663214705045e-05,
+ "loss": 0.0807,
+ "step": 16228
+ },
+ {
+ "epoch": 44.22070844686648,
+ "grad_norm": 3.232729434967041,
+ "learning_rate": 1.2335805141205901e-05,
+ "loss": 0.0685,
+ "step": 16229
+ },
+ {
+ "epoch": 44.223433242506815,
+ "grad_norm": 74.39045715332031,
+ "learning_rate": 1.2334947049515582e-05,
+ "loss": 0.3129,
+ "step": 16230
+ },
+ {
+ "epoch": 44.22615803814714,
+ "grad_norm": 2.7659192085266113,
+ "learning_rate": 1.2334088939640772e-05,
+ "loss": 0.0597,
+ "step": 16231
+ },
+ {
+ "epoch": 44.228882833787466,
+ "grad_norm": 2.9288790225982666,
+ "learning_rate": 1.233323081158815e-05,
+ "loss": 0.0849,
+ "step": 16232
+ },
+ {
+ "epoch": 44.23160762942779,
+ "grad_norm": 3.026132822036743,
+ "learning_rate": 1.2332372665364406e-05,
+ "loss": 0.1364,
+ "step": 16233
+ },
+ {
+ "epoch": 44.23433242506812,
+ "grad_norm": 3.5569889545440674,
+ "learning_rate": 1.2331514500976216e-05,
+ "loss": 0.1421,
+ "step": 16234
+ },
+ {
+ "epoch": 44.237057220708444,
+ "grad_norm": 3.4734232425689697,
+ "learning_rate": 1.2330656318430268e-05,
+ "loss": 0.0739,
+ "step": 16235
+ },
+ {
+ "epoch": 44.23978201634878,
+ "grad_norm": 8.937232971191406,
+ "learning_rate": 1.232979811773324e-05,
+ "loss": 0.1195,
+ "step": 16236
+ },
+ {
+ "epoch": 44.2425068119891,
+ "grad_norm": 3.6599390506744385,
+ "learning_rate": 1.2328939898891823e-05,
+ "loss": 0.1784,
+ "step": 16237
+ },
+ {
+ "epoch": 44.24523160762943,
+ "grad_norm": 4.477314472198486,
+ "learning_rate": 1.2328081661912692e-05,
+ "loss": 0.1926,
+ "step": 16238
+ },
+ {
+ "epoch": 44.247956403269754,
+ "grad_norm": 4.393899440765381,
+ "learning_rate": 1.2327223406802542e-05,
+ "loss": 0.2396,
+ "step": 16239
+ },
+ {
+ "epoch": 44.25068119891008,
+ "grad_norm": 4.326869010925293,
+ "learning_rate": 1.2326365133568046e-05,
+ "loss": 0.1343,
+ "step": 16240
+ },
+ {
+ "epoch": 44.253405994550405,
+ "grad_norm": 6.6287713050842285,
+ "learning_rate": 1.2325506842215897e-05,
+ "loss": 0.2331,
+ "step": 16241
+ },
+ {
+ "epoch": 44.25613079019074,
+ "grad_norm": 6.487010478973389,
+ "learning_rate": 1.2324648532752775e-05,
+ "loss": 0.1185,
+ "step": 16242
+ },
+ {
+ "epoch": 44.258855585831064,
+ "grad_norm": 8.087191581726074,
+ "learning_rate": 1.2323790205185364e-05,
+ "loss": 0.1979,
+ "step": 16243
+ },
+ {
+ "epoch": 44.26158038147139,
+ "grad_norm": 4.05342435836792,
+ "learning_rate": 1.2322931859520349e-05,
+ "loss": 0.0938,
+ "step": 16244
+ },
+ {
+ "epoch": 44.264305177111716,
+ "grad_norm": 4.101500988006592,
+ "learning_rate": 1.2322073495764418e-05,
+ "loss": 0.1011,
+ "step": 16245
+ },
+ {
+ "epoch": 44.26702997275204,
+ "grad_norm": 6.690335273742676,
+ "learning_rate": 1.232121511392425e-05,
+ "loss": 0.1586,
+ "step": 16246
+ },
+ {
+ "epoch": 44.26975476839237,
+ "grad_norm": 5.4969000816345215,
+ "learning_rate": 1.2320356714006534e-05,
+ "loss": 0.1536,
+ "step": 16247
+ },
+ {
+ "epoch": 44.2724795640327,
+ "grad_norm": 2.8979907035827637,
+ "learning_rate": 1.2319498296017955e-05,
+ "loss": 0.0606,
+ "step": 16248
+ },
+ {
+ "epoch": 44.275204359673026,
+ "grad_norm": 4.583102703094482,
+ "learning_rate": 1.2318639859965198e-05,
+ "loss": 0.1252,
+ "step": 16249
+ },
+ {
+ "epoch": 44.27792915531335,
+ "grad_norm": 6.921213150024414,
+ "learning_rate": 1.231778140585495e-05,
+ "loss": 0.08,
+ "step": 16250
+ },
+ {
+ "epoch": 44.28065395095368,
+ "grad_norm": 3.731557607650757,
+ "learning_rate": 1.2316922933693893e-05,
+ "loss": 0.1961,
+ "step": 16251
+ },
+ {
+ "epoch": 44.283378746594,
+ "grad_norm": 3.954416036605835,
+ "learning_rate": 1.2316064443488715e-05,
+ "loss": 0.1379,
+ "step": 16252
+ },
+ {
+ "epoch": 44.28610354223433,
+ "grad_norm": 2.957524538040161,
+ "learning_rate": 1.2315205935246103e-05,
+ "loss": 0.3444,
+ "step": 16253
+ },
+ {
+ "epoch": 44.28882833787466,
+ "grad_norm": 4.285735607147217,
+ "learning_rate": 1.2314347408972742e-05,
+ "loss": 0.0723,
+ "step": 16254
+ },
+ {
+ "epoch": 44.29155313351499,
+ "grad_norm": 4.990494251251221,
+ "learning_rate": 1.2313488864675318e-05,
+ "loss": 0.1552,
+ "step": 16255
+ },
+ {
+ "epoch": 44.294277929155314,
+ "grad_norm": 5.499634265899658,
+ "learning_rate": 1.2312630302360513e-05,
+ "loss": 0.1079,
+ "step": 16256
+ },
+ {
+ "epoch": 44.29700272479564,
+ "grad_norm": 4.02118444442749,
+ "learning_rate": 1.2311771722035022e-05,
+ "loss": 0.0954,
+ "step": 16257
+ },
+ {
+ "epoch": 44.299727520435965,
+ "grad_norm": 3.555302381515503,
+ "learning_rate": 1.2310913123705523e-05,
+ "loss": 0.1441,
+ "step": 16258
+ },
+ {
+ "epoch": 44.30245231607629,
+ "grad_norm": 5.787808895111084,
+ "learning_rate": 1.2310054507378712e-05,
+ "loss": 0.1377,
+ "step": 16259
+ },
+ {
+ "epoch": 44.305177111716624,
+ "grad_norm": 3.771728515625,
+ "learning_rate": 1.2309195873061269e-05,
+ "loss": 0.1403,
+ "step": 16260
+ },
+ {
+ "epoch": 44.30790190735695,
+ "grad_norm": 3.382127285003662,
+ "learning_rate": 1.2308337220759883e-05,
+ "loss": 0.1765,
+ "step": 16261
+ },
+ {
+ "epoch": 44.310626702997276,
+ "grad_norm": 4.084686756134033,
+ "learning_rate": 1.230747855048124e-05,
+ "loss": 0.1399,
+ "step": 16262
+ },
+ {
+ "epoch": 44.3133514986376,
+ "grad_norm": 3.626293659210205,
+ "learning_rate": 1.230661986223203e-05,
+ "loss": 0.2669,
+ "step": 16263
+ },
+ {
+ "epoch": 44.31607629427793,
+ "grad_norm": 4.873316764831543,
+ "learning_rate": 1.2305761156018936e-05,
+ "loss": 0.2149,
+ "step": 16264
+ },
+ {
+ "epoch": 44.31880108991825,
+ "grad_norm": 4.250701427459717,
+ "learning_rate": 1.230490243184865e-05,
+ "loss": 0.1163,
+ "step": 16265
+ },
+ {
+ "epoch": 44.321525885558586,
+ "grad_norm": 2.7836849689483643,
+ "learning_rate": 1.2304043689727859e-05,
+ "loss": 0.1176,
+ "step": 16266
+ },
+ {
+ "epoch": 44.32425068119891,
+ "grad_norm": 3.2909255027770996,
+ "learning_rate": 1.2303184929663248e-05,
+ "loss": 0.0517,
+ "step": 16267
+ },
+ {
+ "epoch": 44.32697547683924,
+ "grad_norm": 4.581793785095215,
+ "learning_rate": 1.2302326151661509e-05,
+ "loss": 0.1937,
+ "step": 16268
+ },
+ {
+ "epoch": 44.32970027247956,
+ "grad_norm": 3.6550939083099365,
+ "learning_rate": 1.2301467355729327e-05,
+ "loss": 0.2229,
+ "step": 16269
+ },
+ {
+ "epoch": 44.33242506811989,
+ "grad_norm": 3.634798765182495,
+ "learning_rate": 1.2300608541873391e-05,
+ "loss": 0.1235,
+ "step": 16270
+ },
+ {
+ "epoch": 44.335149863760215,
+ "grad_norm": 6.25985050201416,
+ "learning_rate": 1.229974971010039e-05,
+ "loss": 0.1482,
+ "step": 16271
+ },
+ {
+ "epoch": 44.33787465940055,
+ "grad_norm": 3.145923376083374,
+ "learning_rate": 1.2298890860417013e-05,
+ "loss": 0.0783,
+ "step": 16272
+ },
+ {
+ "epoch": 44.34059945504087,
+ "grad_norm": 4.797497272491455,
+ "learning_rate": 1.2298031992829945e-05,
+ "loss": 0.1718,
+ "step": 16273
+ },
+ {
+ "epoch": 44.3433242506812,
+ "grad_norm": 4.1703081130981445,
+ "learning_rate": 1.229717310734588e-05,
+ "loss": 0.1405,
+ "step": 16274
+ },
+ {
+ "epoch": 44.346049046321525,
+ "grad_norm": 2.8082034587860107,
+ "learning_rate": 1.2296314203971505e-05,
+ "loss": 0.0589,
+ "step": 16275
+ },
+ {
+ "epoch": 44.34877384196185,
+ "grad_norm": 3.2366180419921875,
+ "learning_rate": 1.2295455282713509e-05,
+ "loss": 0.1667,
+ "step": 16276
+ },
+ {
+ "epoch": 44.35149863760218,
+ "grad_norm": 6.065915584564209,
+ "learning_rate": 1.229459634357858e-05,
+ "loss": 0.2694,
+ "step": 16277
+ },
+ {
+ "epoch": 44.35422343324251,
+ "grad_norm": 2.9645462036132812,
+ "learning_rate": 1.229373738657341e-05,
+ "loss": 0.0873,
+ "step": 16278
+ },
+ {
+ "epoch": 44.356948228882835,
+ "grad_norm": 2.6623375415802,
+ "learning_rate": 1.2292878411704688e-05,
+ "loss": 0.0706,
+ "step": 16279
+ },
+ {
+ "epoch": 44.35967302452316,
+ "grad_norm": 4.046746730804443,
+ "learning_rate": 1.2292019418979098e-05,
+ "loss": 0.0733,
+ "step": 16280
+ },
+ {
+ "epoch": 44.36239782016349,
+ "grad_norm": 2.9969656467437744,
+ "learning_rate": 1.2291160408403338e-05,
+ "loss": 0.2134,
+ "step": 16281
+ },
+ {
+ "epoch": 44.36512261580381,
+ "grad_norm": 2.7104485034942627,
+ "learning_rate": 1.2290301379984095e-05,
+ "loss": 0.0546,
+ "step": 16282
+ },
+ {
+ "epoch": 44.36784741144414,
+ "grad_norm": 3.5889508724212646,
+ "learning_rate": 1.2289442333728057e-05,
+ "loss": 0.1244,
+ "step": 16283
+ },
+ {
+ "epoch": 44.37057220708447,
+ "grad_norm": 3.326298713684082,
+ "learning_rate": 1.2288583269641917e-05,
+ "loss": 0.0764,
+ "step": 16284
+ },
+ {
+ "epoch": 44.3732970027248,
+ "grad_norm": 3.4014904499053955,
+ "learning_rate": 1.2287724187732364e-05,
+ "loss": 0.1113,
+ "step": 16285
+ },
+ {
+ "epoch": 44.37602179836512,
+ "grad_norm": 3.7485647201538086,
+ "learning_rate": 1.2286865088006088e-05,
+ "loss": 0.08,
+ "step": 16286
+ },
+ {
+ "epoch": 44.37874659400545,
+ "grad_norm": 3.059293031692505,
+ "learning_rate": 1.2286005970469779e-05,
+ "loss": 0.0862,
+ "step": 16287
+ },
+ {
+ "epoch": 44.381471389645775,
+ "grad_norm": 2.929184675216675,
+ "learning_rate": 1.2285146835130133e-05,
+ "loss": 0.0741,
+ "step": 16288
+ },
+ {
+ "epoch": 44.3841961852861,
+ "grad_norm": 2.8662657737731934,
+ "learning_rate": 1.2284287681993834e-05,
+ "loss": 0.1198,
+ "step": 16289
+ },
+ {
+ "epoch": 44.38692098092643,
+ "grad_norm": 2.923544406890869,
+ "learning_rate": 1.2283428511067574e-05,
+ "loss": 0.1938,
+ "step": 16290
+ },
+ {
+ "epoch": 44.38964577656676,
+ "grad_norm": 2.8161308765411377,
+ "learning_rate": 1.228256932235805e-05,
+ "loss": 0.1188,
+ "step": 16291
+ },
+ {
+ "epoch": 44.392370572207085,
+ "grad_norm": 3.649897575378418,
+ "learning_rate": 1.228171011587195e-05,
+ "loss": 0.0702,
+ "step": 16292
+ },
+ {
+ "epoch": 44.39509536784741,
+ "grad_norm": 4.633885383605957,
+ "learning_rate": 1.228085089161596e-05,
+ "loss": 0.1758,
+ "step": 16293
+ },
+ {
+ "epoch": 44.39782016348774,
+ "grad_norm": 3.6293060779571533,
+ "learning_rate": 1.2279991649596784e-05,
+ "loss": 0.1111,
+ "step": 16294
+ },
+ {
+ "epoch": 44.40054495912806,
+ "grad_norm": 2.8622469902038574,
+ "learning_rate": 1.22791323898211e-05,
+ "loss": 0.1976,
+ "step": 16295
+ },
+ {
+ "epoch": 44.403269754768395,
+ "grad_norm": 3.0185694694519043,
+ "learning_rate": 1.2278273112295612e-05,
+ "loss": 0.0939,
+ "step": 16296
+ },
+ {
+ "epoch": 44.40599455040872,
+ "grad_norm": 3.7690939903259277,
+ "learning_rate": 1.2277413817027003e-05,
+ "loss": 0.1715,
+ "step": 16297
+ },
+ {
+ "epoch": 44.40871934604905,
+ "grad_norm": 3.9431557655334473,
+ "learning_rate": 1.2276554504021969e-05,
+ "loss": 0.1376,
+ "step": 16298
+ },
+ {
+ "epoch": 44.41144414168937,
+ "grad_norm": 2.7524266242980957,
+ "learning_rate": 1.22756951732872e-05,
+ "loss": 0.1728,
+ "step": 16299
+ },
+ {
+ "epoch": 44.4141689373297,
+ "grad_norm": 3.307047128677368,
+ "learning_rate": 1.2274835824829392e-05,
+ "loss": 0.0929,
+ "step": 16300
+ },
+ {
+ "epoch": 44.416893732970024,
+ "grad_norm": 3.039268970489502,
+ "learning_rate": 1.2273976458655233e-05,
+ "loss": 0.0703,
+ "step": 16301
+ },
+ {
+ "epoch": 44.41961852861036,
+ "grad_norm": 3.721269130706787,
+ "learning_rate": 1.2273117074771423e-05,
+ "loss": 0.084,
+ "step": 16302
+ },
+ {
+ "epoch": 44.42234332425068,
+ "grad_norm": 2.9756603240966797,
+ "learning_rate": 1.2272257673184646e-05,
+ "loss": 0.0503,
+ "step": 16303
+ },
+ {
+ "epoch": 44.42506811989101,
+ "grad_norm": 2.8437812328338623,
+ "learning_rate": 1.2271398253901606e-05,
+ "loss": 0.1027,
+ "step": 16304
+ },
+ {
+ "epoch": 44.427792915531334,
+ "grad_norm": 2.4376485347747803,
+ "learning_rate": 1.2270538816928982e-05,
+ "loss": 0.1602,
+ "step": 16305
+ },
+ {
+ "epoch": 44.43051771117166,
+ "grad_norm": 5.031126976013184,
+ "learning_rate": 1.226967936227348e-05,
+ "loss": 0.1663,
+ "step": 16306
+ },
+ {
+ "epoch": 44.433242506811986,
+ "grad_norm": 3.584353446960449,
+ "learning_rate": 1.2268819889941784e-05,
+ "loss": 0.1723,
+ "step": 16307
+ },
+ {
+ "epoch": 44.43596730245232,
+ "grad_norm": 4.325202465057373,
+ "learning_rate": 1.2267960399940595e-05,
+ "loss": 0.0788,
+ "step": 16308
+ },
+ {
+ "epoch": 44.438692098092645,
+ "grad_norm": 4.188876628875732,
+ "learning_rate": 1.2267100892276602e-05,
+ "loss": 0.0929,
+ "step": 16309
+ },
+ {
+ "epoch": 44.44141689373297,
+ "grad_norm": 3.460188865661621,
+ "learning_rate": 1.2266241366956503e-05,
+ "loss": 0.0986,
+ "step": 16310
+ },
+ {
+ "epoch": 44.444141689373296,
+ "grad_norm": 3.3681814670562744,
+ "learning_rate": 1.2265381823986984e-05,
+ "loss": 0.0912,
+ "step": 16311
+ },
+ {
+ "epoch": 44.44686648501362,
+ "grad_norm": 3.1192963123321533,
+ "learning_rate": 1.2264522263374747e-05,
+ "loss": 0.0623,
+ "step": 16312
+ },
+ {
+ "epoch": 44.44959128065395,
+ "grad_norm": 3.446944236755371,
+ "learning_rate": 1.2263662685126486e-05,
+ "loss": 0.1106,
+ "step": 16313
+ },
+ {
+ "epoch": 44.45231607629428,
+ "grad_norm": 3.2644829750061035,
+ "learning_rate": 1.2262803089248889e-05,
+ "loss": 0.1006,
+ "step": 16314
+ },
+ {
+ "epoch": 44.45504087193461,
+ "grad_norm": 3.397054672241211,
+ "learning_rate": 1.2261943475748657e-05,
+ "loss": 0.1879,
+ "step": 16315
+ },
+ {
+ "epoch": 44.45776566757493,
+ "grad_norm": 3.360900640487671,
+ "learning_rate": 1.226108384463248e-05,
+ "loss": 0.0758,
+ "step": 16316
+ },
+ {
+ "epoch": 44.46049046321526,
+ "grad_norm": 3.3524527549743652,
+ "learning_rate": 1.2260224195907053e-05,
+ "loss": 0.1182,
+ "step": 16317
+ },
+ {
+ "epoch": 44.463215258855584,
+ "grad_norm": 7.943295001983643,
+ "learning_rate": 1.2259364529579076e-05,
+ "loss": 0.157,
+ "step": 16318
+ },
+ {
+ "epoch": 44.46594005449591,
+ "grad_norm": 4.067416667938232,
+ "learning_rate": 1.2258504845655242e-05,
+ "loss": 0.089,
+ "step": 16319
+ },
+ {
+ "epoch": 44.46866485013624,
+ "grad_norm": 3.368337869644165,
+ "learning_rate": 1.2257645144142244e-05,
+ "loss": 0.1712,
+ "step": 16320
+ },
+ {
+ "epoch": 44.47138964577657,
+ "grad_norm": 4.385191440582275,
+ "learning_rate": 1.2256785425046776e-05,
+ "loss": 0.1281,
+ "step": 16321
+ },
+ {
+ "epoch": 44.474114441416894,
+ "grad_norm": 3.971771001815796,
+ "learning_rate": 1.2255925688375539e-05,
+ "loss": 0.1445,
+ "step": 16322
+ },
+ {
+ "epoch": 44.47683923705722,
+ "grad_norm": 3.803821325302124,
+ "learning_rate": 1.2255065934135222e-05,
+ "loss": 0.1413,
+ "step": 16323
+ },
+ {
+ "epoch": 44.479564032697546,
+ "grad_norm": 3.005871057510376,
+ "learning_rate": 1.2254206162332523e-05,
+ "loss": 0.0891,
+ "step": 16324
+ },
+ {
+ "epoch": 44.48228882833787,
+ "grad_norm": 2.6608824729919434,
+ "learning_rate": 1.2253346372974142e-05,
+ "loss": 0.0574,
+ "step": 16325
+ },
+ {
+ "epoch": 44.485013623978205,
+ "grad_norm": 3.134885787963867,
+ "learning_rate": 1.2252486566066772e-05,
+ "loss": 0.1069,
+ "step": 16326
+ },
+ {
+ "epoch": 44.48773841961853,
+ "grad_norm": 2.8412396907806396,
+ "learning_rate": 1.2251626741617106e-05,
+ "loss": 0.0684,
+ "step": 16327
+ },
+ {
+ "epoch": 44.490463215258856,
+ "grad_norm": 4.515639305114746,
+ "learning_rate": 1.225076689963185e-05,
+ "loss": 0.1631,
+ "step": 16328
+ },
+ {
+ "epoch": 44.49318801089918,
+ "grad_norm": 4.42960786819458,
+ "learning_rate": 1.2249907040117688e-05,
+ "loss": 0.1087,
+ "step": 16329
+ },
+ {
+ "epoch": 44.49591280653951,
+ "grad_norm": 2.503058671951294,
+ "learning_rate": 1.2249047163081325e-05,
+ "loss": 0.0611,
+ "step": 16330
+ },
+ {
+ "epoch": 44.49863760217983,
+ "grad_norm": 2.531858444213867,
+ "learning_rate": 1.2248187268529453e-05,
+ "loss": 0.1247,
+ "step": 16331
+ },
+ {
+ "epoch": 44.50136239782017,
+ "grad_norm": 5.403724670410156,
+ "learning_rate": 1.2247327356468772e-05,
+ "loss": 0.1288,
+ "step": 16332
+ },
+ {
+ "epoch": 44.50408719346049,
+ "grad_norm": 2.5948827266693115,
+ "learning_rate": 1.2246467426905978e-05,
+ "loss": 0.2672,
+ "step": 16333
+ },
+ {
+ "epoch": 44.50681198910082,
+ "grad_norm": 3.851266622543335,
+ "learning_rate": 1.2245607479847767e-05,
+ "loss": 0.1514,
+ "step": 16334
+ },
+ {
+ "epoch": 44.509536784741144,
+ "grad_norm": 2.758700132369995,
+ "learning_rate": 1.2244747515300839e-05,
+ "loss": 0.0682,
+ "step": 16335
+ },
+ {
+ "epoch": 44.51226158038147,
+ "grad_norm": 3.1032259464263916,
+ "learning_rate": 1.224388753327189e-05,
+ "loss": 0.0795,
+ "step": 16336
+ },
+ {
+ "epoch": 44.514986376021795,
+ "grad_norm": 5.264296054840088,
+ "learning_rate": 1.2243027533767614e-05,
+ "loss": 0.1359,
+ "step": 16337
+ },
+ {
+ "epoch": 44.51771117166213,
+ "grad_norm": 3.3823482990264893,
+ "learning_rate": 1.2242167516794714e-05,
+ "loss": 0.0794,
+ "step": 16338
+ },
+ {
+ "epoch": 44.520435967302454,
+ "grad_norm": 4.0481648445129395,
+ "learning_rate": 1.2241307482359885e-05,
+ "loss": 0.1419,
+ "step": 16339
+ },
+ {
+ "epoch": 44.52316076294278,
+ "grad_norm": 3.1494944095611572,
+ "learning_rate": 1.2240447430469829e-05,
+ "loss": 0.0833,
+ "step": 16340
+ },
+ {
+ "epoch": 44.525885558583106,
+ "grad_norm": 3.443079710006714,
+ "learning_rate": 1.2239587361131235e-05,
+ "loss": 0.1059,
+ "step": 16341
+ },
+ {
+ "epoch": 44.52861035422343,
+ "grad_norm": 6.133141040802002,
+ "learning_rate": 1.2238727274350808e-05,
+ "loss": 0.1172,
+ "step": 16342
+ },
+ {
+ "epoch": 44.53133514986376,
+ "grad_norm": 4.204970836639404,
+ "learning_rate": 1.2237867170135244e-05,
+ "loss": 0.1076,
+ "step": 16343
+ },
+ {
+ "epoch": 44.53405994550409,
+ "grad_norm": 4.500120162963867,
+ "learning_rate": 1.2237007048491248e-05,
+ "loss": 0.1295,
+ "step": 16344
+ },
+ {
+ "epoch": 44.536784741144416,
+ "grad_norm": 3.538550853729248,
+ "learning_rate": 1.2236146909425508e-05,
+ "loss": 0.1823,
+ "step": 16345
+ },
+ {
+ "epoch": 44.53950953678474,
+ "grad_norm": 3.6367928981781006,
+ "learning_rate": 1.223528675294473e-05,
+ "loss": 0.1539,
+ "step": 16346
+ },
+ {
+ "epoch": 44.54223433242507,
+ "grad_norm": 3.927931547164917,
+ "learning_rate": 1.223442657905561e-05,
+ "loss": 0.1238,
+ "step": 16347
+ },
+ {
+ "epoch": 44.54495912806539,
+ "grad_norm": 3.0266735553741455,
+ "learning_rate": 1.2233566387764847e-05,
+ "loss": 0.0609,
+ "step": 16348
+ },
+ {
+ "epoch": 44.54768392370572,
+ "grad_norm": 3.116879463195801,
+ "learning_rate": 1.2232706179079142e-05,
+ "loss": 0.1062,
+ "step": 16349
+ },
+ {
+ "epoch": 44.55040871934605,
+ "grad_norm": 3.130445718765259,
+ "learning_rate": 1.2231845953005192e-05,
+ "loss": 0.081,
+ "step": 16350
+ },
+ {
+ "epoch": 44.55313351498638,
+ "grad_norm": 3.2738285064697266,
+ "learning_rate": 1.2230985709549699e-05,
+ "loss": 0.1966,
+ "step": 16351
+ },
+ {
+ "epoch": 44.555858310626704,
+ "grad_norm": 3.0134377479553223,
+ "learning_rate": 1.2230125448719358e-05,
+ "loss": 0.2136,
+ "step": 16352
+ },
+ {
+ "epoch": 44.55858310626703,
+ "grad_norm": 3.4388277530670166,
+ "learning_rate": 1.2229265170520876e-05,
+ "loss": 0.1902,
+ "step": 16353
+ },
+ {
+ "epoch": 44.561307901907355,
+ "grad_norm": 3.236258029937744,
+ "learning_rate": 1.2228404874960949e-05,
+ "loss": 0.1028,
+ "step": 16354
+ },
+ {
+ "epoch": 44.56403269754768,
+ "grad_norm": 3.427520990371704,
+ "learning_rate": 1.2227544562046272e-05,
+ "loss": 0.1066,
+ "step": 16355
+ },
+ {
+ "epoch": 44.566757493188014,
+ "grad_norm": 3.0525763034820557,
+ "learning_rate": 1.2226684231783551e-05,
+ "loss": 0.1812,
+ "step": 16356
+ },
+ {
+ "epoch": 44.56948228882834,
+ "grad_norm": 3.4213473796844482,
+ "learning_rate": 1.2225823884179486e-05,
+ "loss": 0.1039,
+ "step": 16357
+ },
+ {
+ "epoch": 44.572207084468666,
+ "grad_norm": 3.7217206954956055,
+ "learning_rate": 1.222496351924078e-05,
+ "loss": 0.1184,
+ "step": 16358
+ },
+ {
+ "epoch": 44.57493188010899,
+ "grad_norm": 2.964245319366455,
+ "learning_rate": 1.2224103136974123e-05,
+ "loss": 0.0774,
+ "step": 16359
+ },
+ {
+ "epoch": 44.57765667574932,
+ "grad_norm": 2.6594088077545166,
+ "learning_rate": 1.2223242737386225e-05,
+ "loss": 0.0406,
+ "step": 16360
+ },
+ {
+ "epoch": 44.58038147138964,
+ "grad_norm": 3.534209728240967,
+ "learning_rate": 1.2222382320483785e-05,
+ "loss": 0.1871,
+ "step": 16361
+ },
+ {
+ "epoch": 44.583106267029976,
+ "grad_norm": 4.971428871154785,
+ "learning_rate": 1.2221521886273502e-05,
+ "loss": 0.1083,
+ "step": 16362
+ },
+ {
+ "epoch": 44.5858310626703,
+ "grad_norm": 4.393173694610596,
+ "learning_rate": 1.222066143476208e-05,
+ "loss": 0.1171,
+ "step": 16363
+ },
+ {
+ "epoch": 44.58855585831063,
+ "grad_norm": 3.0477633476257324,
+ "learning_rate": 1.2219800965956217e-05,
+ "loss": 0.1181,
+ "step": 16364
+ },
+ {
+ "epoch": 44.59128065395095,
+ "grad_norm": 2.8552069664001465,
+ "learning_rate": 1.2218940479862613e-05,
+ "loss": 0.1665,
+ "step": 16365
+ },
+ {
+ "epoch": 44.59400544959128,
+ "grad_norm": 2.9086170196533203,
+ "learning_rate": 1.2218079976487976e-05,
+ "loss": 0.0982,
+ "step": 16366
+ },
+ {
+ "epoch": 44.596730245231605,
+ "grad_norm": 3.963552474975586,
+ "learning_rate": 1.2217219455839e-05,
+ "loss": 0.1138,
+ "step": 16367
+ },
+ {
+ "epoch": 44.59945504087194,
+ "grad_norm": 2.9100141525268555,
+ "learning_rate": 1.221635891792239e-05,
+ "loss": 0.088,
+ "step": 16368
+ },
+ {
+ "epoch": 44.60217983651226,
+ "grad_norm": 3.5979323387145996,
+ "learning_rate": 1.2215498362744853e-05,
+ "loss": 0.1571,
+ "step": 16369
+ },
+ {
+ "epoch": 44.60490463215259,
+ "grad_norm": 2.9783453941345215,
+ "learning_rate": 1.2214637790313081e-05,
+ "loss": 0.4059,
+ "step": 16370
+ },
+ {
+ "epoch": 44.607629427792915,
+ "grad_norm": 2.9741382598876953,
+ "learning_rate": 1.2213777200633786e-05,
+ "loss": 0.0718,
+ "step": 16371
+ },
+ {
+ "epoch": 44.61035422343324,
+ "grad_norm": 3.2324371337890625,
+ "learning_rate": 1.221291659371366e-05,
+ "loss": 0.0946,
+ "step": 16372
+ },
+ {
+ "epoch": 44.61307901907357,
+ "grad_norm": 3.3301503658294678,
+ "learning_rate": 1.2212055969559415e-05,
+ "loss": 0.0893,
+ "step": 16373
+ },
+ {
+ "epoch": 44.6158038147139,
+ "grad_norm": 3.9918150901794434,
+ "learning_rate": 1.2211195328177747e-05,
+ "loss": 0.1259,
+ "step": 16374
+ },
+ {
+ "epoch": 44.618528610354225,
+ "grad_norm": 2.5081939697265625,
+ "learning_rate": 1.2210334669575362e-05,
+ "loss": 0.0768,
+ "step": 16375
+ },
+ {
+ "epoch": 44.62125340599455,
+ "grad_norm": 2.8576817512512207,
+ "learning_rate": 1.2209473993758957e-05,
+ "loss": 0.0829,
+ "step": 16376
+ },
+ {
+ "epoch": 44.62397820163488,
+ "grad_norm": 3.8166561126708984,
+ "learning_rate": 1.2208613300735247e-05,
+ "loss": 0.1245,
+ "step": 16377
+ },
+ {
+ "epoch": 44.6267029972752,
+ "grad_norm": 4.645926475524902,
+ "learning_rate": 1.2207752590510924e-05,
+ "loss": 0.2166,
+ "step": 16378
+ },
+ {
+ "epoch": 44.62942779291553,
+ "grad_norm": 3.1788721084594727,
+ "learning_rate": 1.2206891863092697e-05,
+ "loss": 0.1022,
+ "step": 16379
+ },
+ {
+ "epoch": 44.63215258855586,
+ "grad_norm": 3.123617649078369,
+ "learning_rate": 1.2206031118487264e-05,
+ "loss": 0.1215,
+ "step": 16380
+ },
+ {
+ "epoch": 44.63487738419619,
+ "grad_norm": 5.775550365447998,
+ "learning_rate": 1.2205170356701337e-05,
+ "loss": 0.0865,
+ "step": 16381
+ },
+ {
+ "epoch": 44.63760217983651,
+ "grad_norm": 3.215189218521118,
+ "learning_rate": 1.220430957774161e-05,
+ "loss": 0.1293,
+ "step": 16382
+ },
+ {
+ "epoch": 44.64032697547684,
+ "grad_norm": 4.484073162078857,
+ "learning_rate": 1.220344878161479e-05,
+ "loss": 0.1384,
+ "step": 16383
+ },
+ {
+ "epoch": 44.643051771117165,
+ "grad_norm": 2.052863359451294,
+ "learning_rate": 1.2202587968327584e-05,
+ "loss": 0.1097,
+ "step": 16384
+ },
+ {
+ "epoch": 44.64577656675749,
+ "grad_norm": 2.763705253601074,
+ "learning_rate": 1.2201727137886692e-05,
+ "loss": 0.1437,
+ "step": 16385
+ },
+ {
+ "epoch": 44.64850136239782,
+ "grad_norm": 2.994436264038086,
+ "learning_rate": 1.220086629029882e-05,
+ "loss": 0.2113,
+ "step": 16386
+ },
+ {
+ "epoch": 44.65122615803815,
+ "grad_norm": 3.4886302947998047,
+ "learning_rate": 1.2200005425570676e-05,
+ "loss": 0.1356,
+ "step": 16387
+ },
+ {
+ "epoch": 44.653950953678475,
+ "grad_norm": 3.013478994369507,
+ "learning_rate": 1.2199144543708957e-05,
+ "loss": 0.0821,
+ "step": 16388
+ },
+ {
+ "epoch": 44.6566757493188,
+ "grad_norm": 3.0457606315612793,
+ "learning_rate": 1.2198283644720373e-05,
+ "loss": 0.1914,
+ "step": 16389
+ },
+ {
+ "epoch": 44.65940054495913,
+ "grad_norm": 2.7016658782958984,
+ "learning_rate": 1.2197422728611624e-05,
+ "loss": 0.2585,
+ "step": 16390
+ },
+ {
+ "epoch": 44.66212534059945,
+ "grad_norm": 8.722550392150879,
+ "learning_rate": 1.219656179538942e-05,
+ "loss": 0.1168,
+ "step": 16391
+ },
+ {
+ "epoch": 44.664850136239785,
+ "grad_norm": 3.813955783843994,
+ "learning_rate": 1.2195700845060462e-05,
+ "loss": 0.2675,
+ "step": 16392
+ },
+ {
+ "epoch": 44.66757493188011,
+ "grad_norm": 3.8874683380126953,
+ "learning_rate": 1.2194839877631457e-05,
+ "loss": 0.0534,
+ "step": 16393
+ },
+ {
+ "epoch": 44.67029972752044,
+ "grad_norm": 2.8622231483459473,
+ "learning_rate": 1.2193978893109109e-05,
+ "loss": 0.1748,
+ "step": 16394
+ },
+ {
+ "epoch": 44.67302452316076,
+ "grad_norm": 3.116316318511963,
+ "learning_rate": 1.2193117891500126e-05,
+ "loss": 0.1944,
+ "step": 16395
+ },
+ {
+ "epoch": 44.67574931880109,
+ "grad_norm": 3.30139422416687,
+ "learning_rate": 1.2192256872811207e-05,
+ "loss": 0.0585,
+ "step": 16396
+ },
+ {
+ "epoch": 44.678474114441414,
+ "grad_norm": 4.459347724914551,
+ "learning_rate": 1.219139583704907e-05,
+ "loss": 0.179,
+ "step": 16397
+ },
+ {
+ "epoch": 44.68119891008175,
+ "grad_norm": 3.1846373081207275,
+ "learning_rate": 1.2190534784220405e-05,
+ "loss": 0.1294,
+ "step": 16398
+ },
+ {
+ "epoch": 44.68392370572207,
+ "grad_norm": 2.5054991245269775,
+ "learning_rate": 1.2189673714331932e-05,
+ "loss": 0.0545,
+ "step": 16399
+ },
+ {
+ "epoch": 44.6866485013624,
+ "grad_norm": 3.9883792400360107,
+ "learning_rate": 1.2188812627390345e-05,
+ "loss": 0.1328,
+ "step": 16400
+ },
+ {
+ "epoch": 44.689373297002724,
+ "grad_norm": 4.850978851318359,
+ "learning_rate": 1.218795152340236e-05,
+ "loss": 0.109,
+ "step": 16401
+ },
+ {
+ "epoch": 44.69209809264305,
+ "grad_norm": 3.771267890930176,
+ "learning_rate": 1.2187090402374673e-05,
+ "loss": 0.2061,
+ "step": 16402
+ },
+ {
+ "epoch": 44.694822888283376,
+ "grad_norm": 3.2852439880371094,
+ "learning_rate": 1.2186229264314e-05,
+ "loss": 0.0946,
+ "step": 16403
+ },
+ {
+ "epoch": 44.69754768392371,
+ "grad_norm": 2.7866621017456055,
+ "learning_rate": 1.2185368109227044e-05,
+ "loss": 0.1705,
+ "step": 16404
+ },
+ {
+ "epoch": 44.700272479564035,
+ "grad_norm": 2.7078959941864014,
+ "learning_rate": 1.2184506937120513e-05,
+ "loss": 0.0828,
+ "step": 16405
+ },
+ {
+ "epoch": 44.70299727520436,
+ "grad_norm": 3.046513080596924,
+ "learning_rate": 1.2183645748001109e-05,
+ "loss": 0.0833,
+ "step": 16406
+ },
+ {
+ "epoch": 44.705722070844686,
+ "grad_norm": 3.939838409423828,
+ "learning_rate": 1.2182784541875546e-05,
+ "loss": 0.0909,
+ "step": 16407
+ },
+ {
+ "epoch": 44.70844686648501,
+ "grad_norm": 2.9457855224609375,
+ "learning_rate": 1.2181923318750523e-05,
+ "loss": 0.1809,
+ "step": 16408
+ },
+ {
+ "epoch": 44.71117166212534,
+ "grad_norm": 2.868462085723877,
+ "learning_rate": 1.2181062078632757e-05,
+ "loss": 0.1111,
+ "step": 16409
+ },
+ {
+ "epoch": 44.71389645776567,
+ "grad_norm": 2.721851348876953,
+ "learning_rate": 1.2180200821528942e-05,
+ "loss": 0.1439,
+ "step": 16410
+ },
+ {
+ "epoch": 44.716621253406,
+ "grad_norm": 3.8915956020355225,
+ "learning_rate": 1.2179339547445799e-05,
+ "loss": 0.0919,
+ "step": 16411
+ },
+ {
+ "epoch": 44.71934604904632,
+ "grad_norm": 3.785297393798828,
+ "learning_rate": 1.2178478256390028e-05,
+ "loss": 0.0575,
+ "step": 16412
+ },
+ {
+ "epoch": 44.72207084468665,
+ "grad_norm": 3.427711248397827,
+ "learning_rate": 1.2177616948368341e-05,
+ "loss": 0.09,
+ "step": 16413
+ },
+ {
+ "epoch": 44.724795640326974,
+ "grad_norm": 3.5590598583221436,
+ "learning_rate": 1.217675562338744e-05,
+ "loss": 0.1647,
+ "step": 16414
+ },
+ {
+ "epoch": 44.7275204359673,
+ "grad_norm": 2.9706268310546875,
+ "learning_rate": 1.2175894281454037e-05,
+ "loss": 0.1332,
+ "step": 16415
+ },
+ {
+ "epoch": 44.73024523160763,
+ "grad_norm": 7.152397155761719,
+ "learning_rate": 1.2175032922574839e-05,
+ "loss": 0.1572,
+ "step": 16416
+ },
+ {
+ "epoch": 44.73297002724796,
+ "grad_norm": 3.8014848232269287,
+ "learning_rate": 1.2174171546756557e-05,
+ "loss": 0.2021,
+ "step": 16417
+ },
+ {
+ "epoch": 44.735694822888284,
+ "grad_norm": 3.514738082885742,
+ "learning_rate": 1.2173310154005896e-05,
+ "loss": 0.0783,
+ "step": 16418
+ },
+ {
+ "epoch": 44.73841961852861,
+ "grad_norm": 5.537026882171631,
+ "learning_rate": 1.217244874432956e-05,
+ "loss": 0.0793,
+ "step": 16419
+ },
+ {
+ "epoch": 44.741144414168936,
+ "grad_norm": 3.0548064708709717,
+ "learning_rate": 1.2171587317734268e-05,
+ "loss": 0.0783,
+ "step": 16420
+ },
+ {
+ "epoch": 44.74386920980926,
+ "grad_norm": 2.99078106880188,
+ "learning_rate": 1.2170725874226725e-05,
+ "loss": 0.1064,
+ "step": 16421
+ },
+ {
+ "epoch": 44.746594005449595,
+ "grad_norm": 3.426892042160034,
+ "learning_rate": 1.2169864413813637e-05,
+ "loss": 0.084,
+ "step": 16422
+ },
+ {
+ "epoch": 44.74931880108992,
+ "grad_norm": 5.200826644897461,
+ "learning_rate": 1.2169002936501717e-05,
+ "loss": 0.1334,
+ "step": 16423
+ },
+ {
+ "epoch": 44.752043596730246,
+ "grad_norm": 5.238041877746582,
+ "learning_rate": 1.216814144229767e-05,
+ "loss": 0.1223,
+ "step": 16424
+ },
+ {
+ "epoch": 44.75476839237057,
+ "grad_norm": 3.290930986404419,
+ "learning_rate": 1.2167279931208209e-05,
+ "loss": 0.0609,
+ "step": 16425
+ },
+ {
+ "epoch": 44.7574931880109,
+ "grad_norm": 4.584853649139404,
+ "learning_rate": 1.2166418403240038e-05,
+ "loss": 0.0822,
+ "step": 16426
+ },
+ {
+ "epoch": 44.76021798365122,
+ "grad_norm": 2.920445680618286,
+ "learning_rate": 1.2165556858399874e-05,
+ "loss": 0.0893,
+ "step": 16427
+ },
+ {
+ "epoch": 44.762942779291556,
+ "grad_norm": 2.723066806793213,
+ "learning_rate": 1.216469529669442e-05,
+ "loss": 0.1612,
+ "step": 16428
+ },
+ {
+ "epoch": 44.76566757493188,
+ "grad_norm": 2.4763967990875244,
+ "learning_rate": 1.216383371813039e-05,
+ "loss": 0.057,
+ "step": 16429
+ },
+ {
+ "epoch": 44.76839237057221,
+ "grad_norm": 3.126718044281006,
+ "learning_rate": 1.2162972122714491e-05,
+ "loss": 0.1034,
+ "step": 16430
+ },
+ {
+ "epoch": 44.771117166212534,
+ "grad_norm": 3.1335973739624023,
+ "learning_rate": 1.2162110510453437e-05,
+ "loss": 0.1631,
+ "step": 16431
+ },
+ {
+ "epoch": 44.77384196185286,
+ "grad_norm": 2.9010531902313232,
+ "learning_rate": 1.2161248881353936e-05,
+ "loss": 0.1011,
+ "step": 16432
+ },
+ {
+ "epoch": 44.776566757493185,
+ "grad_norm": 2.95917010307312,
+ "learning_rate": 1.2160387235422697e-05,
+ "loss": 0.2758,
+ "step": 16433
+ },
+ {
+ "epoch": 44.77929155313352,
+ "grad_norm": 2.741062641143799,
+ "learning_rate": 1.215952557266643e-05,
+ "loss": 0.2575,
+ "step": 16434
+ },
+ {
+ "epoch": 44.782016348773844,
+ "grad_norm": 3.6716413497924805,
+ "learning_rate": 1.2158663893091851e-05,
+ "loss": 0.1049,
+ "step": 16435
+ },
+ {
+ "epoch": 44.78474114441417,
+ "grad_norm": 3.2005553245544434,
+ "learning_rate": 1.2157802196705666e-05,
+ "loss": 0.0625,
+ "step": 16436
+ },
+ {
+ "epoch": 44.787465940054496,
+ "grad_norm": 2.9694690704345703,
+ "learning_rate": 1.2156940483514585e-05,
+ "loss": 0.1044,
+ "step": 16437
+ },
+ {
+ "epoch": 44.79019073569482,
+ "grad_norm": 3.152423620223999,
+ "learning_rate": 1.2156078753525325e-05,
+ "loss": 0.0806,
+ "step": 16438
+ },
+ {
+ "epoch": 44.79291553133515,
+ "grad_norm": 3.487405776977539,
+ "learning_rate": 1.2155217006744588e-05,
+ "loss": 0.0856,
+ "step": 16439
+ },
+ {
+ "epoch": 44.79564032697548,
+ "grad_norm": 3.596632242202759,
+ "learning_rate": 1.2154355243179092e-05,
+ "loss": 0.1369,
+ "step": 16440
+ },
+ {
+ "epoch": 44.798365122615806,
+ "grad_norm": 3.2471396923065186,
+ "learning_rate": 1.215349346283555e-05,
+ "loss": 0.0695,
+ "step": 16441
+ },
+ {
+ "epoch": 44.80108991825613,
+ "grad_norm": 2.973168134689331,
+ "learning_rate": 1.2152631665720667e-05,
+ "loss": 0.1351,
+ "step": 16442
+ },
+ {
+ "epoch": 44.80381471389646,
+ "grad_norm": 3.281651020050049,
+ "learning_rate": 1.215176985184116e-05,
+ "loss": 0.0686,
+ "step": 16443
+ },
+ {
+ "epoch": 44.80653950953678,
+ "grad_norm": 3.376415491104126,
+ "learning_rate": 1.2150908021203736e-05,
+ "loss": 0.2235,
+ "step": 16444
+ },
+ {
+ "epoch": 44.80926430517711,
+ "grad_norm": 3.072640895843506,
+ "learning_rate": 1.215004617381511e-05,
+ "loss": 0.0963,
+ "step": 16445
+ },
+ {
+ "epoch": 44.81198910081744,
+ "grad_norm": 3.6920182704925537,
+ "learning_rate": 1.2149184309681995e-05,
+ "loss": 0.1028,
+ "step": 16446
+ },
+ {
+ "epoch": 44.81471389645777,
+ "grad_norm": 3.303715467453003,
+ "learning_rate": 1.21483224288111e-05,
+ "loss": 0.0739,
+ "step": 16447
+ },
+ {
+ "epoch": 44.817438692098094,
+ "grad_norm": 2.8739778995513916,
+ "learning_rate": 1.2147460531209141e-05,
+ "loss": 0.0648,
+ "step": 16448
+ },
+ {
+ "epoch": 44.82016348773842,
+ "grad_norm": 2.8762435913085938,
+ "learning_rate": 1.2146598616882828e-05,
+ "loss": 0.101,
+ "step": 16449
+ },
+ {
+ "epoch": 44.822888283378745,
+ "grad_norm": 3.5661585330963135,
+ "learning_rate": 1.2145736685838876e-05,
+ "loss": 0.0912,
+ "step": 16450
+ },
+ {
+ "epoch": 44.82561307901907,
+ "grad_norm": 5.520371913909912,
+ "learning_rate": 1.2144874738083992e-05,
+ "loss": 0.1779,
+ "step": 16451
+ },
+ {
+ "epoch": 44.828337874659404,
+ "grad_norm": 3.7217514514923096,
+ "learning_rate": 1.2144012773624893e-05,
+ "loss": 0.1023,
+ "step": 16452
+ },
+ {
+ "epoch": 44.83106267029973,
+ "grad_norm": 5.955410480499268,
+ "learning_rate": 1.2143150792468292e-05,
+ "loss": 0.0913,
+ "step": 16453
+ },
+ {
+ "epoch": 44.833787465940055,
+ "grad_norm": 2.5974643230438232,
+ "learning_rate": 1.2142288794620903e-05,
+ "loss": 0.123,
+ "step": 16454
+ },
+ {
+ "epoch": 44.83651226158038,
+ "grad_norm": 2.973301410675049,
+ "learning_rate": 1.2141426780089436e-05,
+ "loss": 0.1855,
+ "step": 16455
+ },
+ {
+ "epoch": 44.83923705722071,
+ "grad_norm": 3.447610378265381,
+ "learning_rate": 1.2140564748880608e-05,
+ "loss": 0.132,
+ "step": 16456
+ },
+ {
+ "epoch": 44.84196185286103,
+ "grad_norm": 3.221714735031128,
+ "learning_rate": 1.213970270100113e-05,
+ "loss": 0.2166,
+ "step": 16457
+ },
+ {
+ "epoch": 44.844686648501366,
+ "grad_norm": 2.825873613357544,
+ "learning_rate": 1.2138840636457716e-05,
+ "loss": 0.1195,
+ "step": 16458
+ },
+ {
+ "epoch": 44.84741144414169,
+ "grad_norm": 2.6053221225738525,
+ "learning_rate": 1.213797855525708e-05,
+ "loss": 0.1334,
+ "step": 16459
+ },
+ {
+ "epoch": 44.85013623978202,
+ "grad_norm": 5.568309783935547,
+ "learning_rate": 1.2137116457405938e-05,
+ "loss": 0.0886,
+ "step": 16460
+ },
+ {
+ "epoch": 44.85286103542234,
+ "grad_norm": 3.725358724594116,
+ "learning_rate": 1.2136254342910997e-05,
+ "loss": 0.1265,
+ "step": 16461
+ },
+ {
+ "epoch": 44.85558583106267,
+ "grad_norm": 2.934250593185425,
+ "learning_rate": 1.2135392211778978e-05,
+ "loss": 0.2113,
+ "step": 16462
+ },
+ {
+ "epoch": 44.858310626702995,
+ "grad_norm": 6.387988567352295,
+ "learning_rate": 1.2134530064016594e-05,
+ "loss": 0.0786,
+ "step": 16463
+ },
+ {
+ "epoch": 44.86103542234333,
+ "grad_norm": 2.9093120098114014,
+ "learning_rate": 1.213366789963056e-05,
+ "loss": 0.0564,
+ "step": 16464
+ },
+ {
+ "epoch": 44.86376021798365,
+ "grad_norm": 2.9012527465820312,
+ "learning_rate": 1.2132805718627588e-05,
+ "loss": 0.0712,
+ "step": 16465
+ },
+ {
+ "epoch": 44.86648501362398,
+ "grad_norm": 4.4518866539001465,
+ "learning_rate": 1.2131943521014393e-05,
+ "loss": 0.0469,
+ "step": 16466
+ },
+ {
+ "epoch": 44.869209809264305,
+ "grad_norm": 2.363471269607544,
+ "learning_rate": 1.213108130679769e-05,
+ "loss": 0.063,
+ "step": 16467
+ },
+ {
+ "epoch": 44.87193460490463,
+ "grad_norm": 5.626222610473633,
+ "learning_rate": 1.2130219075984195e-05,
+ "loss": 0.1534,
+ "step": 16468
+ },
+ {
+ "epoch": 44.87465940054496,
+ "grad_norm": 2.7398524284362793,
+ "learning_rate": 1.2129356828580623e-05,
+ "loss": 0.0667,
+ "step": 16469
+ },
+ {
+ "epoch": 44.87738419618529,
+ "grad_norm": 4.474040508270264,
+ "learning_rate": 1.212849456459369e-05,
+ "loss": 0.2994,
+ "step": 16470
+ },
+ {
+ "epoch": 44.880108991825615,
+ "grad_norm": 3.315720796585083,
+ "learning_rate": 1.2127632284030101e-05,
+ "loss": 0.1841,
+ "step": 16471
+ },
+ {
+ "epoch": 44.88283378746594,
+ "grad_norm": 2.355161190032959,
+ "learning_rate": 1.212676998689659e-05,
+ "loss": 0.066,
+ "step": 16472
+ },
+ {
+ "epoch": 44.88555858310627,
+ "grad_norm": 2.9384677410125732,
+ "learning_rate": 1.212590767319986e-05,
+ "loss": 0.1736,
+ "step": 16473
+ },
+ {
+ "epoch": 44.88828337874659,
+ "grad_norm": 2.8491668701171875,
+ "learning_rate": 1.2125045342946628e-05,
+ "loss": 0.0648,
+ "step": 16474
+ },
+ {
+ "epoch": 44.89100817438692,
+ "grad_norm": 2.8912761211395264,
+ "learning_rate": 1.2124182996143615e-05,
+ "loss": 0.0574,
+ "step": 16475
+ },
+ {
+ "epoch": 44.89373297002725,
+ "grad_norm": 3.6161468029022217,
+ "learning_rate": 1.2123320632797531e-05,
+ "loss": 0.2342,
+ "step": 16476
+ },
+ {
+ "epoch": 44.89645776566758,
+ "grad_norm": 2.2781779766082764,
+ "learning_rate": 1.2122458252915093e-05,
+ "loss": 0.0768,
+ "step": 16477
+ },
+ {
+ "epoch": 44.8991825613079,
+ "grad_norm": 3.4443013668060303,
+ "learning_rate": 1.2121595856503018e-05,
+ "loss": 0.0998,
+ "step": 16478
+ },
+ {
+ "epoch": 44.90190735694823,
+ "grad_norm": 2.9003970623016357,
+ "learning_rate": 1.2120733443568023e-05,
+ "loss": 0.0832,
+ "step": 16479
+ },
+ {
+ "epoch": 44.904632152588555,
+ "grad_norm": 4.621079444885254,
+ "learning_rate": 1.2119871014116827e-05,
+ "loss": 0.0486,
+ "step": 16480
+ },
+ {
+ "epoch": 44.90735694822888,
+ "grad_norm": 3.4478907585144043,
+ "learning_rate": 1.211900856815614e-05,
+ "loss": 0.0837,
+ "step": 16481
+ },
+ {
+ "epoch": 44.91008174386921,
+ "grad_norm": 2.3126678466796875,
+ "learning_rate": 1.2118146105692687e-05,
+ "loss": 0.0618,
+ "step": 16482
+ },
+ {
+ "epoch": 44.91280653950954,
+ "grad_norm": 3.2970409393310547,
+ "learning_rate": 1.2117283626733176e-05,
+ "loss": 0.1073,
+ "step": 16483
+ },
+ {
+ "epoch": 44.915531335149865,
+ "grad_norm": 2.1315486431121826,
+ "learning_rate": 1.2116421131284332e-05,
+ "loss": 0.0683,
+ "step": 16484
+ },
+ {
+ "epoch": 44.91825613079019,
+ "grad_norm": 3.264350414276123,
+ "learning_rate": 1.2115558619352866e-05,
+ "loss": 0.0723,
+ "step": 16485
+ },
+ {
+ "epoch": 44.920980926430516,
+ "grad_norm": 3.173365592956543,
+ "learning_rate": 1.2114696090945499e-05,
+ "loss": 0.116,
+ "step": 16486
+ },
+ {
+ "epoch": 44.92370572207084,
+ "grad_norm": 5.332544326782227,
+ "learning_rate": 1.2113833546068945e-05,
+ "loss": 0.0916,
+ "step": 16487
+ },
+ {
+ "epoch": 44.926430517711175,
+ "grad_norm": 3.072880268096924,
+ "learning_rate": 1.2112970984729924e-05,
+ "loss": 0.0779,
+ "step": 16488
+ },
+ {
+ "epoch": 44.9291553133515,
+ "grad_norm": 4.1760640144348145,
+ "learning_rate": 1.2112108406935152e-05,
+ "loss": 0.0791,
+ "step": 16489
+ },
+ {
+ "epoch": 44.93188010899183,
+ "grad_norm": 3.456834077835083,
+ "learning_rate": 1.211124581269135e-05,
+ "loss": 0.3423,
+ "step": 16490
+ },
+ {
+ "epoch": 44.93460490463215,
+ "grad_norm": 4.496623992919922,
+ "learning_rate": 1.2110383202005234e-05,
+ "loss": 0.1275,
+ "step": 16491
+ },
+ {
+ "epoch": 44.93732970027248,
+ "grad_norm": 2.91316294670105,
+ "learning_rate": 1.2109520574883519e-05,
+ "loss": 0.4105,
+ "step": 16492
+ },
+ {
+ "epoch": 44.940054495912804,
+ "grad_norm": 3.448993444442749,
+ "learning_rate": 1.2108657931332928e-05,
+ "loss": 0.111,
+ "step": 16493
+ },
+ {
+ "epoch": 44.94277929155314,
+ "grad_norm": 2.7782485485076904,
+ "learning_rate": 1.2107795271360177e-05,
+ "loss": 0.1554,
+ "step": 16494
+ },
+ {
+ "epoch": 44.94550408719346,
+ "grad_norm": 3.2890784740448,
+ "learning_rate": 1.210693259497198e-05,
+ "loss": 0.1921,
+ "step": 16495
+ },
+ {
+ "epoch": 44.94822888283379,
+ "grad_norm": 5.3513875007629395,
+ "learning_rate": 1.2106069902175061e-05,
+ "loss": 0.1054,
+ "step": 16496
+ },
+ {
+ "epoch": 44.950953678474114,
+ "grad_norm": 3.4420268535614014,
+ "learning_rate": 1.210520719297614e-05,
+ "loss": 0.0826,
+ "step": 16497
+ },
+ {
+ "epoch": 44.95367847411444,
+ "grad_norm": 2.746809959411621,
+ "learning_rate": 1.2104344467381934e-05,
+ "loss": 0.0695,
+ "step": 16498
+ },
+ {
+ "epoch": 44.956403269754766,
+ "grad_norm": 3.5084633827209473,
+ "learning_rate": 1.2103481725399158e-05,
+ "loss": 0.1064,
+ "step": 16499
+ },
+ {
+ "epoch": 44.95912806539509,
+ "grad_norm": 3.5531294345855713,
+ "learning_rate": 1.2102618967034537e-05,
+ "loss": 0.1168,
+ "step": 16500
+ },
+ {
+ "epoch": 44.961852861035425,
+ "grad_norm": 2.7614905834198,
+ "learning_rate": 1.2101756192294785e-05,
+ "loss": 0.1648,
+ "step": 16501
+ },
+ {
+ "epoch": 44.96457765667575,
+ "grad_norm": 3.0529773235321045,
+ "learning_rate": 1.2100893401186624e-05,
+ "loss": 0.1154,
+ "step": 16502
+ },
+ {
+ "epoch": 44.967302452316076,
+ "grad_norm": 3.0180764198303223,
+ "learning_rate": 1.2100030593716772e-05,
+ "loss": 0.0908,
+ "step": 16503
+ },
+ {
+ "epoch": 44.9700272479564,
+ "grad_norm": 3.3153834342956543,
+ "learning_rate": 1.2099167769891952e-05,
+ "loss": 0.1512,
+ "step": 16504
+ },
+ {
+ "epoch": 44.97275204359673,
+ "grad_norm": 2.8533530235290527,
+ "learning_rate": 1.2098304929718875e-05,
+ "loss": 0.1092,
+ "step": 16505
+ },
+ {
+ "epoch": 44.97547683923706,
+ "grad_norm": 3.0964009761810303,
+ "learning_rate": 1.2097442073204272e-05,
+ "loss": 0.1008,
+ "step": 16506
+ },
+ {
+ "epoch": 44.97820163487739,
+ "grad_norm": 3.0468597412109375,
+ "learning_rate": 1.2096579200354857e-05,
+ "loss": 0.1632,
+ "step": 16507
+ },
+ {
+ "epoch": 44.98092643051771,
+ "grad_norm": 2.4162752628326416,
+ "learning_rate": 1.209571631117735e-05,
+ "loss": 0.1672,
+ "step": 16508
+ },
+ {
+ "epoch": 44.98365122615804,
+ "grad_norm": 2.5352137088775635,
+ "learning_rate": 1.2094853405678471e-05,
+ "loss": 0.0786,
+ "step": 16509
+ },
+ {
+ "epoch": 44.986376021798364,
+ "grad_norm": 3.2644355297088623,
+ "learning_rate": 1.2093990483864944e-05,
+ "loss": 0.0895,
+ "step": 16510
+ },
+ {
+ "epoch": 44.98910081743869,
+ "grad_norm": 3.0335607528686523,
+ "learning_rate": 1.2093127545743482e-05,
+ "loss": 0.1052,
+ "step": 16511
+ },
+ {
+ "epoch": 44.991825613079016,
+ "grad_norm": 3.6361353397369385,
+ "learning_rate": 1.2092264591320815e-05,
+ "loss": 0.136,
+ "step": 16512
+ },
+ {
+ "epoch": 44.99455040871935,
+ "grad_norm": 2.4920060634613037,
+ "learning_rate": 1.2091401620603656e-05,
+ "loss": 0.1224,
+ "step": 16513
+ },
+ {
+ "epoch": 44.997275204359674,
+ "grad_norm": 3.8500492572784424,
+ "learning_rate": 1.2090538633598728e-05,
+ "loss": 0.2066,
+ "step": 16514
+ },
+ {
+ "epoch": 45.0,
+ "grad_norm": 3.407466411590576,
+ "learning_rate": 1.2089675630312755e-05,
+ "loss": 0.0765,
+ "step": 16515
+ },
+ {
+ "epoch": 45.002724795640326,
+ "grad_norm": 2.890751600265503,
+ "learning_rate": 1.2088812610752449e-05,
+ "loss": 0.0766,
+ "step": 16516
+ },
+ {
+ "epoch": 45.00544959128065,
+ "grad_norm": 2.44423508644104,
+ "learning_rate": 1.2087949574924542e-05,
+ "loss": 0.0592,
+ "step": 16517
+ },
+ {
+ "epoch": 45.00817438692098,
+ "grad_norm": 2.4049808979034424,
+ "learning_rate": 1.2087086522835754e-05,
+ "loss": 0.0951,
+ "step": 16518
+ },
+ {
+ "epoch": 45.01089918256131,
+ "grad_norm": 2.5491411685943604,
+ "learning_rate": 1.2086223454492798e-05,
+ "loss": 0.0902,
+ "step": 16519
+ },
+ {
+ "epoch": 45.013623978201636,
+ "grad_norm": 3.6166939735412598,
+ "learning_rate": 1.2085360369902405e-05,
+ "loss": 0.1672,
+ "step": 16520
+ },
+ {
+ "epoch": 45.01634877384196,
+ "grad_norm": 3.2467856407165527,
+ "learning_rate": 1.208449726907129e-05,
+ "loss": 0.1558,
+ "step": 16521
+ },
+ {
+ "epoch": 45.01907356948229,
+ "grad_norm": 2.810112953186035,
+ "learning_rate": 1.2083634152006178e-05,
+ "loss": 0.1921,
+ "step": 16522
+ },
+ {
+ "epoch": 45.02179836512261,
+ "grad_norm": 3.313887119293213,
+ "learning_rate": 1.208277101871379e-05,
+ "loss": 0.1473,
+ "step": 16523
+ },
+ {
+ "epoch": 45.02452316076294,
+ "grad_norm": 2.7547566890716553,
+ "learning_rate": 1.208190786920085e-05,
+ "loss": 0.2308,
+ "step": 16524
+ },
+ {
+ "epoch": 45.02724795640327,
+ "grad_norm": 3.4855825901031494,
+ "learning_rate": 1.2081044703474079e-05,
+ "loss": 0.1062,
+ "step": 16525
+ },
+ {
+ "epoch": 45.0299727520436,
+ "grad_norm": 4.43713903427124,
+ "learning_rate": 1.2080181521540195e-05,
+ "loss": 0.2093,
+ "step": 16526
+ },
+ {
+ "epoch": 45.032697547683924,
+ "grad_norm": 3.1449525356292725,
+ "learning_rate": 1.2079318323405926e-05,
+ "loss": 0.0667,
+ "step": 16527
+ },
+ {
+ "epoch": 45.03542234332425,
+ "grad_norm": 2.5413870811462402,
+ "learning_rate": 1.2078455109077994e-05,
+ "loss": 0.168,
+ "step": 16528
+ },
+ {
+ "epoch": 45.038147138964575,
+ "grad_norm": 2.94400691986084,
+ "learning_rate": 1.207759187856312e-05,
+ "loss": 0.0482,
+ "step": 16529
+ },
+ {
+ "epoch": 45.0408719346049,
+ "grad_norm": 5.762238025665283,
+ "learning_rate": 1.2076728631868026e-05,
+ "loss": 0.098,
+ "step": 16530
+ },
+ {
+ "epoch": 45.043596730245234,
+ "grad_norm": 2.7185819149017334,
+ "learning_rate": 1.2075865368999437e-05,
+ "loss": 0.0772,
+ "step": 16531
+ },
+ {
+ "epoch": 45.04632152588556,
+ "grad_norm": 2.6767666339874268,
+ "learning_rate": 1.2075002089964077e-05,
+ "loss": 0.2155,
+ "step": 16532
+ },
+ {
+ "epoch": 45.049046321525886,
+ "grad_norm": 2.8329901695251465,
+ "learning_rate": 1.2074138794768666e-05,
+ "loss": 0.0729,
+ "step": 16533
+ },
+ {
+ "epoch": 45.05177111716621,
+ "grad_norm": 3.2482194900512695,
+ "learning_rate": 1.2073275483419928e-05,
+ "loss": 0.1562,
+ "step": 16534
+ },
+ {
+ "epoch": 45.05449591280654,
+ "grad_norm": 3.013538122177124,
+ "learning_rate": 1.2072412155924589e-05,
+ "loss": 0.2036,
+ "step": 16535
+ },
+ {
+ "epoch": 45.05722070844686,
+ "grad_norm": 2.558934450149536,
+ "learning_rate": 1.207154881228937e-05,
+ "loss": 0.0431,
+ "step": 16536
+ },
+ {
+ "epoch": 45.059945504087196,
+ "grad_norm": 3.7464547157287598,
+ "learning_rate": 1.2070685452520996e-05,
+ "loss": 0.1334,
+ "step": 16537
+ },
+ {
+ "epoch": 45.06267029972752,
+ "grad_norm": 2.9682421684265137,
+ "learning_rate": 1.206982207662619e-05,
+ "loss": 0.0541,
+ "step": 16538
+ },
+ {
+ "epoch": 45.06539509536785,
+ "grad_norm": 3.2697067260742188,
+ "learning_rate": 1.2068958684611675e-05,
+ "loss": 0.1463,
+ "step": 16539
+ },
+ {
+ "epoch": 45.06811989100817,
+ "grad_norm": 2.3783199787139893,
+ "learning_rate": 1.2068095276484177e-05,
+ "loss": 0.0497,
+ "step": 16540
+ },
+ {
+ "epoch": 45.0708446866485,
+ "grad_norm": 2.6204755306243896,
+ "learning_rate": 1.2067231852250421e-05,
+ "loss": 0.1686,
+ "step": 16541
+ },
+ {
+ "epoch": 45.073569482288825,
+ "grad_norm": 2.9343340396881104,
+ "learning_rate": 1.206636841191713e-05,
+ "loss": 0.0482,
+ "step": 16542
+ },
+ {
+ "epoch": 45.07629427792916,
+ "grad_norm": 2.717257022857666,
+ "learning_rate": 1.2065504955491028e-05,
+ "loss": 0.0459,
+ "step": 16543
+ },
+ {
+ "epoch": 45.079019073569484,
+ "grad_norm": 2.3560304641723633,
+ "learning_rate": 1.2064641482978835e-05,
+ "loss": 0.1323,
+ "step": 16544
+ },
+ {
+ "epoch": 45.08174386920981,
+ "grad_norm": 2.883267402648926,
+ "learning_rate": 1.206377799438729e-05,
+ "loss": 0.1004,
+ "step": 16545
+ },
+ {
+ "epoch": 45.084468664850135,
+ "grad_norm": 3.030815839767456,
+ "learning_rate": 1.20629144897231e-05,
+ "loss": 0.1645,
+ "step": 16546
+ },
+ {
+ "epoch": 45.08719346049046,
+ "grad_norm": 2.525696039199829,
+ "learning_rate": 1.2062050968993002e-05,
+ "loss": 0.0884,
+ "step": 16547
+ },
+ {
+ "epoch": 45.08991825613079,
+ "grad_norm": 3.041459083557129,
+ "learning_rate": 1.2061187432203716e-05,
+ "loss": 0.1396,
+ "step": 16548
+ },
+ {
+ "epoch": 45.09264305177112,
+ "grad_norm": 2.868438482284546,
+ "learning_rate": 1.206032387936197e-05,
+ "loss": 0.0784,
+ "step": 16549
+ },
+ {
+ "epoch": 45.095367847411445,
+ "grad_norm": 2.77864933013916,
+ "learning_rate": 1.2059460310474486e-05,
+ "loss": 0.063,
+ "step": 16550
+ },
+ {
+ "epoch": 45.09809264305177,
+ "grad_norm": 3.100841760635376,
+ "learning_rate": 1.2058596725547993e-05,
+ "loss": 0.1685,
+ "step": 16551
+ },
+ {
+ "epoch": 45.1008174386921,
+ "grad_norm": 3.000182867050171,
+ "learning_rate": 1.2057733124589216e-05,
+ "loss": 0.0838,
+ "step": 16552
+ },
+ {
+ "epoch": 45.10354223433242,
+ "grad_norm": 2.57462739944458,
+ "learning_rate": 1.2056869507604878e-05,
+ "loss": 0.126,
+ "step": 16553
+ },
+ {
+ "epoch": 45.10626702997275,
+ "grad_norm": 2.6186399459838867,
+ "learning_rate": 1.2056005874601706e-05,
+ "loss": 0.0824,
+ "step": 16554
+ },
+ {
+ "epoch": 45.10899182561308,
+ "grad_norm": 2.4519996643066406,
+ "learning_rate": 1.205514222558643e-05,
+ "loss": 0.0618,
+ "step": 16555
+ },
+ {
+ "epoch": 45.11171662125341,
+ "grad_norm": 2.9088079929351807,
+ "learning_rate": 1.2054278560565768e-05,
+ "loss": 0.0516,
+ "step": 16556
+ },
+ {
+ "epoch": 45.11444141689373,
+ "grad_norm": 3.545365810394287,
+ "learning_rate": 1.2053414879546453e-05,
+ "loss": 0.0987,
+ "step": 16557
+ },
+ {
+ "epoch": 45.11716621253406,
+ "grad_norm": 3.1729679107666016,
+ "learning_rate": 1.2052551182535205e-05,
+ "loss": 0.1541,
+ "step": 16558
+ },
+ {
+ "epoch": 45.119891008174385,
+ "grad_norm": 2.9971234798431396,
+ "learning_rate": 1.2051687469538762e-05,
+ "loss": 0.1011,
+ "step": 16559
+ },
+ {
+ "epoch": 45.12261580381471,
+ "grad_norm": 2.3939225673675537,
+ "learning_rate": 1.2050823740563838e-05,
+ "loss": 0.1474,
+ "step": 16560
+ },
+ {
+ "epoch": 45.12534059945504,
+ "grad_norm": 3.1177189350128174,
+ "learning_rate": 1.2049959995617166e-05,
+ "loss": 0.0647,
+ "step": 16561
+ },
+ {
+ "epoch": 45.12806539509537,
+ "grad_norm": 3.52839994430542,
+ "learning_rate": 1.2049096234705471e-05,
+ "loss": 0.1718,
+ "step": 16562
+ },
+ {
+ "epoch": 45.130790190735695,
+ "grad_norm": 2.8340492248535156,
+ "learning_rate": 1.2048232457835482e-05,
+ "loss": 0.1344,
+ "step": 16563
+ },
+ {
+ "epoch": 45.13351498637602,
+ "grad_norm": 2.37522292137146,
+ "learning_rate": 1.204736866501392e-05,
+ "loss": 0.1339,
+ "step": 16564
+ },
+ {
+ "epoch": 45.13623978201635,
+ "grad_norm": 3.3282501697540283,
+ "learning_rate": 1.204650485624752e-05,
+ "loss": 0.1359,
+ "step": 16565
+ },
+ {
+ "epoch": 45.13896457765667,
+ "grad_norm": 2.2546908855438232,
+ "learning_rate": 1.2045641031543008e-05,
+ "loss": 0.0325,
+ "step": 16566
+ },
+ {
+ "epoch": 45.141689373297005,
+ "grad_norm": 3.882960796356201,
+ "learning_rate": 1.2044777190907107e-05,
+ "loss": 0.068,
+ "step": 16567
+ },
+ {
+ "epoch": 45.14441416893733,
+ "grad_norm": 3.34743070602417,
+ "learning_rate": 1.2043913334346546e-05,
+ "loss": 0.092,
+ "step": 16568
+ },
+ {
+ "epoch": 45.14713896457766,
+ "grad_norm": 3.1676342487335205,
+ "learning_rate": 1.2043049461868057e-05,
+ "loss": 0.2109,
+ "step": 16569
+ },
+ {
+ "epoch": 45.14986376021798,
+ "grad_norm": 3.4123475551605225,
+ "learning_rate": 1.2042185573478362e-05,
+ "loss": 0.0617,
+ "step": 16570
+ },
+ {
+ "epoch": 45.15258855585831,
+ "grad_norm": 2.850924253463745,
+ "learning_rate": 1.2041321669184194e-05,
+ "loss": 0.326,
+ "step": 16571
+ },
+ {
+ "epoch": 45.155313351498634,
+ "grad_norm": 3.365802049636841,
+ "learning_rate": 1.2040457748992275e-05,
+ "loss": 0.0544,
+ "step": 16572
+ },
+ {
+ "epoch": 45.15803814713897,
+ "grad_norm": 2.6519787311553955,
+ "learning_rate": 1.203959381290934e-05,
+ "loss": 0.1267,
+ "step": 16573
+ },
+ {
+ "epoch": 45.16076294277929,
+ "grad_norm": 2.632171154022217,
+ "learning_rate": 1.203872986094211e-05,
+ "loss": 0.0682,
+ "step": 16574
+ },
+ {
+ "epoch": 45.16348773841962,
+ "grad_norm": 3.068737506866455,
+ "learning_rate": 1.2037865893097323e-05,
+ "loss": 0.1464,
+ "step": 16575
+ },
+ {
+ "epoch": 45.166212534059945,
+ "grad_norm": 2.814425230026245,
+ "learning_rate": 1.2037001909381699e-05,
+ "loss": 0.0822,
+ "step": 16576
+ },
+ {
+ "epoch": 45.16893732970027,
+ "grad_norm": 3.3870325088500977,
+ "learning_rate": 1.203613790980197e-05,
+ "loss": 0.0784,
+ "step": 16577
+ },
+ {
+ "epoch": 45.171662125340596,
+ "grad_norm": 3.7428476810455322,
+ "learning_rate": 1.2035273894364863e-05,
+ "loss": 0.0673,
+ "step": 16578
+ },
+ {
+ "epoch": 45.17438692098093,
+ "grad_norm": 2.2333590984344482,
+ "learning_rate": 1.203440986307711e-05,
+ "loss": 0.048,
+ "step": 16579
+ },
+ {
+ "epoch": 45.177111716621255,
+ "grad_norm": 3.056241750717163,
+ "learning_rate": 1.2033545815945438e-05,
+ "loss": 0.1227,
+ "step": 16580
+ },
+ {
+ "epoch": 45.17983651226158,
+ "grad_norm": 2.1115987300872803,
+ "learning_rate": 1.2032681752976578e-05,
+ "loss": 0.0377,
+ "step": 16581
+ },
+ {
+ "epoch": 45.182561307901906,
+ "grad_norm": 2.817174196243286,
+ "learning_rate": 1.2031817674177255e-05,
+ "loss": 0.1235,
+ "step": 16582
+ },
+ {
+ "epoch": 45.18528610354223,
+ "grad_norm": 3.1548447608947754,
+ "learning_rate": 1.2030953579554203e-05,
+ "loss": 0.0671,
+ "step": 16583
+ },
+ {
+ "epoch": 45.18801089918256,
+ "grad_norm": 3.1117610931396484,
+ "learning_rate": 1.203008946911415e-05,
+ "loss": 0.1268,
+ "step": 16584
+ },
+ {
+ "epoch": 45.19073569482289,
+ "grad_norm": 1.8627084493637085,
+ "learning_rate": 1.2029225342863826e-05,
+ "loss": 0.0707,
+ "step": 16585
+ },
+ {
+ "epoch": 45.19346049046322,
+ "grad_norm": 2.992319107055664,
+ "learning_rate": 1.202836120080996e-05,
+ "loss": 0.0748,
+ "step": 16586
+ },
+ {
+ "epoch": 45.19618528610354,
+ "grad_norm": 3.1559107303619385,
+ "learning_rate": 1.2027497042959281e-05,
+ "loss": 0.0728,
+ "step": 16587
+ },
+ {
+ "epoch": 45.19891008174387,
+ "grad_norm": 2.906529188156128,
+ "learning_rate": 1.2026632869318522e-05,
+ "loss": 0.0619,
+ "step": 16588
+ },
+ {
+ "epoch": 45.201634877384194,
+ "grad_norm": 2.39351749420166,
+ "learning_rate": 1.2025768679894412e-05,
+ "loss": 0.197,
+ "step": 16589
+ },
+ {
+ "epoch": 45.20435967302452,
+ "grad_norm": 2.537943124771118,
+ "learning_rate": 1.2024904474693679e-05,
+ "loss": 0.1173,
+ "step": 16590
+ },
+ {
+ "epoch": 45.20708446866485,
+ "grad_norm": 2.5408031940460205,
+ "learning_rate": 1.2024040253723054e-05,
+ "loss": 0.1222,
+ "step": 16591
+ },
+ {
+ "epoch": 45.20980926430518,
+ "grad_norm": 3.055729866027832,
+ "learning_rate": 1.2023176016989272e-05,
+ "loss": 0.0753,
+ "step": 16592
+ },
+ {
+ "epoch": 45.212534059945504,
+ "grad_norm": 2.6908767223358154,
+ "learning_rate": 1.2022311764499057e-05,
+ "loss": 0.0713,
+ "step": 16593
+ },
+ {
+ "epoch": 45.21525885558583,
+ "grad_norm": 2.709963083267212,
+ "learning_rate": 1.2021447496259145e-05,
+ "loss": 0.0804,
+ "step": 16594
+ },
+ {
+ "epoch": 45.217983651226156,
+ "grad_norm": 2.4345407485961914,
+ "learning_rate": 1.2020583212276267e-05,
+ "loss": 0.0752,
+ "step": 16595
+ },
+ {
+ "epoch": 45.22070844686648,
+ "grad_norm": 2.575911521911621,
+ "learning_rate": 1.2019718912557148e-05,
+ "loss": 0.055,
+ "step": 16596
+ },
+ {
+ "epoch": 45.223433242506815,
+ "grad_norm": 3.520292282104492,
+ "learning_rate": 1.2018854597108524e-05,
+ "loss": 0.1234,
+ "step": 16597
+ },
+ {
+ "epoch": 45.22615803814714,
+ "grad_norm": 3.1117990016937256,
+ "learning_rate": 1.2017990265937127e-05,
+ "loss": 0.1761,
+ "step": 16598
+ },
+ {
+ "epoch": 45.228882833787466,
+ "grad_norm": 1.9323114156723022,
+ "learning_rate": 1.2017125919049688e-05,
+ "loss": 0.0459,
+ "step": 16599
+ },
+ {
+ "epoch": 45.23160762942779,
+ "grad_norm": 4.140061855316162,
+ "learning_rate": 1.2016261556452933e-05,
+ "loss": 0.1713,
+ "step": 16600
+ },
+ {
+ "epoch": 45.23433242506812,
+ "grad_norm": 3.2628655433654785,
+ "learning_rate": 1.2015397178153601e-05,
+ "loss": 0.2337,
+ "step": 16601
+ },
+ {
+ "epoch": 45.237057220708444,
+ "grad_norm": 2.5598061084747314,
+ "learning_rate": 1.2014532784158419e-05,
+ "loss": 0.0871,
+ "step": 16602
+ },
+ {
+ "epoch": 45.23978201634878,
+ "grad_norm": 2.3329720497131348,
+ "learning_rate": 1.201366837447412e-05,
+ "loss": 0.1027,
+ "step": 16603
+ },
+ {
+ "epoch": 45.2425068119891,
+ "grad_norm": 2.8852131366729736,
+ "learning_rate": 1.2012803949107437e-05,
+ "loss": 0.0619,
+ "step": 16604
+ },
+ {
+ "epoch": 45.24523160762943,
+ "grad_norm": 2.0412230491638184,
+ "learning_rate": 1.2011939508065105e-05,
+ "loss": 0.0444,
+ "step": 16605
+ },
+ {
+ "epoch": 45.247956403269754,
+ "grad_norm": 3.333332061767578,
+ "learning_rate": 1.2011075051353848e-05,
+ "loss": 0.0665,
+ "step": 16606
+ },
+ {
+ "epoch": 45.25068119891008,
+ "grad_norm": 2.951378107070923,
+ "learning_rate": 1.2010210578980407e-05,
+ "loss": 0.1006,
+ "step": 16607
+ },
+ {
+ "epoch": 45.253405994550405,
+ "grad_norm": 3.572838306427002,
+ "learning_rate": 1.2009346090951507e-05,
+ "loss": 0.138,
+ "step": 16608
+ },
+ {
+ "epoch": 45.25613079019074,
+ "grad_norm": 3.4753365516662598,
+ "learning_rate": 1.2008481587273889e-05,
+ "loss": 0.0996,
+ "step": 16609
+ },
+ {
+ "epoch": 45.258855585831064,
+ "grad_norm": 3.5344655513763428,
+ "learning_rate": 1.2007617067954279e-05,
+ "loss": 0.0772,
+ "step": 16610
+ },
+ {
+ "epoch": 45.26158038147139,
+ "grad_norm": 3.930788993835449,
+ "learning_rate": 1.2006752532999408e-05,
+ "loss": 0.0667,
+ "step": 16611
+ },
+ {
+ "epoch": 45.264305177111716,
+ "grad_norm": 2.9915852546691895,
+ "learning_rate": 1.2005887982416018e-05,
+ "loss": 0.1048,
+ "step": 16612
+ },
+ {
+ "epoch": 45.26702997275204,
+ "grad_norm": 2.4784672260284424,
+ "learning_rate": 1.2005023416210835e-05,
+ "loss": 0.1263,
+ "step": 16613
+ },
+ {
+ "epoch": 45.26975476839237,
+ "grad_norm": 3.149545192718506,
+ "learning_rate": 1.2004158834390594e-05,
+ "loss": 0.0566,
+ "step": 16614
+ },
+ {
+ "epoch": 45.2724795640327,
+ "grad_norm": 2.451742649078369,
+ "learning_rate": 1.2003294236962028e-05,
+ "loss": 0.2936,
+ "step": 16615
+ },
+ {
+ "epoch": 45.275204359673026,
+ "grad_norm": 2.852996587753296,
+ "learning_rate": 1.2002429623931874e-05,
+ "loss": 0.176,
+ "step": 16616
+ },
+ {
+ "epoch": 45.27792915531335,
+ "grad_norm": 2.4323928356170654,
+ "learning_rate": 1.2001564995306858e-05,
+ "loss": 0.0399,
+ "step": 16617
+ },
+ {
+ "epoch": 45.28065395095368,
+ "grad_norm": 3.6809706687927246,
+ "learning_rate": 1.2000700351093722e-05,
+ "loss": 0.2047,
+ "step": 16618
+ },
+ {
+ "epoch": 45.283378746594,
+ "grad_norm": 2.895249843597412,
+ "learning_rate": 1.1999835691299193e-05,
+ "loss": 0.0879,
+ "step": 16619
+ },
+ {
+ "epoch": 45.28610354223433,
+ "grad_norm": 2.264709234237671,
+ "learning_rate": 1.1998971015930012e-05,
+ "loss": 0.0487,
+ "step": 16620
+ },
+ {
+ "epoch": 45.28882833787466,
+ "grad_norm": 2.653745412826538,
+ "learning_rate": 1.1998106324992906e-05,
+ "loss": 0.166,
+ "step": 16621
+ },
+ {
+ "epoch": 45.29155313351499,
+ "grad_norm": 2.230900764465332,
+ "learning_rate": 1.1997241618494613e-05,
+ "loss": 0.0488,
+ "step": 16622
+ },
+ {
+ "epoch": 45.294277929155314,
+ "grad_norm": 3.099991798400879,
+ "learning_rate": 1.1996376896441867e-05,
+ "loss": 0.048,
+ "step": 16623
+ },
+ {
+ "epoch": 45.29700272479564,
+ "grad_norm": 2.4526588916778564,
+ "learning_rate": 1.1995512158841402e-05,
+ "loss": 0.1411,
+ "step": 16624
+ },
+ {
+ "epoch": 45.299727520435965,
+ "grad_norm": 2.3200132846832275,
+ "learning_rate": 1.1994647405699951e-05,
+ "loss": 0.0909,
+ "step": 16625
+ },
+ {
+ "epoch": 45.30245231607629,
+ "grad_norm": 3.055072784423828,
+ "learning_rate": 1.1993782637024251e-05,
+ "loss": 0.1195,
+ "step": 16626
+ },
+ {
+ "epoch": 45.305177111716624,
+ "grad_norm": 2.3414156436920166,
+ "learning_rate": 1.1992917852821035e-05,
+ "loss": 0.0693,
+ "step": 16627
+ },
+ {
+ "epoch": 45.30790190735695,
+ "grad_norm": 2.4925355911254883,
+ "learning_rate": 1.1992053053097041e-05,
+ "loss": 0.0444,
+ "step": 16628
+ },
+ {
+ "epoch": 45.310626702997276,
+ "grad_norm": 3.3536500930786133,
+ "learning_rate": 1.1991188237859e-05,
+ "loss": 0.1908,
+ "step": 16629
+ },
+ {
+ "epoch": 45.3133514986376,
+ "grad_norm": 3.9857001304626465,
+ "learning_rate": 1.199032340711365e-05,
+ "loss": 0.1255,
+ "step": 16630
+ },
+ {
+ "epoch": 45.31607629427793,
+ "grad_norm": 2.8826746940612793,
+ "learning_rate": 1.1989458560867726e-05,
+ "loss": 0.0982,
+ "step": 16631
+ },
+ {
+ "epoch": 45.31880108991825,
+ "grad_norm": 2.2033021450042725,
+ "learning_rate": 1.1988593699127964e-05,
+ "loss": 0.0478,
+ "step": 16632
+ },
+ {
+ "epoch": 45.321525885558586,
+ "grad_norm": 2.4730420112609863,
+ "learning_rate": 1.1987728821901094e-05,
+ "loss": 0.0489,
+ "step": 16633
+ },
+ {
+ "epoch": 45.32425068119891,
+ "grad_norm": 2.4726576805114746,
+ "learning_rate": 1.198686392919386e-05,
+ "loss": 0.119,
+ "step": 16634
+ },
+ {
+ "epoch": 45.32697547683924,
+ "grad_norm": 3.2300825119018555,
+ "learning_rate": 1.198599902101299e-05,
+ "loss": 0.1995,
+ "step": 16635
+ },
+ {
+ "epoch": 45.32970027247956,
+ "grad_norm": 3.4165151119232178,
+ "learning_rate": 1.1985134097365227e-05,
+ "loss": 0.1711,
+ "step": 16636
+ },
+ {
+ "epoch": 45.33242506811989,
+ "grad_norm": 3.0325205326080322,
+ "learning_rate": 1.1984269158257303e-05,
+ "loss": 0.0672,
+ "step": 16637
+ },
+ {
+ "epoch": 45.335149863760215,
+ "grad_norm": 3.553009271621704,
+ "learning_rate": 1.1983404203695954e-05,
+ "loss": 0.127,
+ "step": 16638
+ },
+ {
+ "epoch": 45.33787465940055,
+ "grad_norm": 2.7984237670898438,
+ "learning_rate": 1.1982539233687916e-05,
+ "loss": 0.0709,
+ "step": 16639
+ },
+ {
+ "epoch": 45.34059945504087,
+ "grad_norm": 2.4789226055145264,
+ "learning_rate": 1.1981674248239927e-05,
+ "loss": 0.1534,
+ "step": 16640
+ },
+ {
+ "epoch": 45.3433242506812,
+ "grad_norm": 2.706575870513916,
+ "learning_rate": 1.1980809247358722e-05,
+ "loss": 0.0788,
+ "step": 16641
+ },
+ {
+ "epoch": 45.346049046321525,
+ "grad_norm": 3.9739253520965576,
+ "learning_rate": 1.197994423105104e-05,
+ "loss": 0.226,
+ "step": 16642
+ },
+ {
+ "epoch": 45.34877384196185,
+ "grad_norm": 2.2117927074432373,
+ "learning_rate": 1.1979079199323613e-05,
+ "loss": 0.0471,
+ "step": 16643
+ },
+ {
+ "epoch": 45.35149863760218,
+ "grad_norm": 2.2672410011291504,
+ "learning_rate": 1.1978214152183186e-05,
+ "loss": 0.0583,
+ "step": 16644
+ },
+ {
+ "epoch": 45.35422343324251,
+ "grad_norm": 2.834831953048706,
+ "learning_rate": 1.1977349089636484e-05,
+ "loss": 0.0889,
+ "step": 16645
+ },
+ {
+ "epoch": 45.356948228882835,
+ "grad_norm": 2.9563047885894775,
+ "learning_rate": 1.1976484011690257e-05,
+ "loss": 0.096,
+ "step": 16646
+ },
+ {
+ "epoch": 45.35967302452316,
+ "grad_norm": 2.8599865436553955,
+ "learning_rate": 1.1975618918351234e-05,
+ "loss": 0.066,
+ "step": 16647
+ },
+ {
+ "epoch": 45.36239782016349,
+ "grad_norm": 3.3421008586883545,
+ "learning_rate": 1.1974753809626157e-05,
+ "loss": 0.1306,
+ "step": 16648
+ },
+ {
+ "epoch": 45.36512261580381,
+ "grad_norm": 2.8014004230499268,
+ "learning_rate": 1.1973888685521755e-05,
+ "loss": 0.1288,
+ "step": 16649
+ },
+ {
+ "epoch": 45.36784741144414,
+ "grad_norm": 2.769298553466797,
+ "learning_rate": 1.1973023546044777e-05,
+ "loss": 0.0869,
+ "step": 16650
+ },
+ {
+ "epoch": 45.37057220708447,
+ "grad_norm": 2.1482114791870117,
+ "learning_rate": 1.1972158391201949e-05,
+ "loss": 0.054,
+ "step": 16651
+ },
+ {
+ "epoch": 45.3732970027248,
+ "grad_norm": 2.974673271179199,
+ "learning_rate": 1.197129322100002e-05,
+ "loss": 0.1658,
+ "step": 16652
+ },
+ {
+ "epoch": 45.37602179836512,
+ "grad_norm": 3.280747652053833,
+ "learning_rate": 1.197042803544572e-05,
+ "loss": 0.0881,
+ "step": 16653
+ },
+ {
+ "epoch": 45.37874659400545,
+ "grad_norm": 2.945436477661133,
+ "learning_rate": 1.1969562834545794e-05,
+ "loss": 0.2042,
+ "step": 16654
+ },
+ {
+ "epoch": 45.381471389645775,
+ "grad_norm": 2.786709785461426,
+ "learning_rate": 1.1968697618306973e-05,
+ "loss": 0.108,
+ "step": 16655
+ },
+ {
+ "epoch": 45.3841961852861,
+ "grad_norm": 2.7908966541290283,
+ "learning_rate": 1.1967832386736001e-05,
+ "loss": 0.0908,
+ "step": 16656
+ },
+ {
+ "epoch": 45.38692098092643,
+ "grad_norm": 3.076099395751953,
+ "learning_rate": 1.1966967139839609e-05,
+ "loss": 0.3016,
+ "step": 16657
+ },
+ {
+ "epoch": 45.38964577656676,
+ "grad_norm": 2.4349358081817627,
+ "learning_rate": 1.1966101877624545e-05,
+ "loss": 0.0506,
+ "step": 16658
+ },
+ {
+ "epoch": 45.392370572207085,
+ "grad_norm": 3.444193124771118,
+ "learning_rate": 1.1965236600097541e-05,
+ "loss": 0.2811,
+ "step": 16659
+ },
+ {
+ "epoch": 45.39509536784741,
+ "grad_norm": 3.0821874141693115,
+ "learning_rate": 1.1964371307265334e-05,
+ "loss": 0.1417,
+ "step": 16660
+ },
+ {
+ "epoch": 45.39782016348774,
+ "grad_norm": 3.603827476501465,
+ "learning_rate": 1.196350599913467e-05,
+ "loss": 0.204,
+ "step": 16661
+ },
+ {
+ "epoch": 45.40054495912806,
+ "grad_norm": 2.915435314178467,
+ "learning_rate": 1.1962640675712284e-05,
+ "loss": 0.0662,
+ "step": 16662
+ },
+ {
+ "epoch": 45.403269754768395,
+ "grad_norm": 2.69297456741333,
+ "learning_rate": 1.1961775337004916e-05,
+ "loss": 0.1153,
+ "step": 16663
+ },
+ {
+ "epoch": 45.40599455040872,
+ "grad_norm": 3.105175495147705,
+ "learning_rate": 1.1960909983019307e-05,
+ "loss": 0.0642,
+ "step": 16664
+ },
+ {
+ "epoch": 45.40871934604905,
+ "grad_norm": 1.8391669988632202,
+ "learning_rate": 1.1960044613762189e-05,
+ "loss": 0.0555,
+ "step": 16665
+ },
+ {
+ "epoch": 45.41144414168937,
+ "grad_norm": 2.5607290267944336,
+ "learning_rate": 1.1959179229240312e-05,
+ "loss": 0.0566,
+ "step": 16666
+ },
+ {
+ "epoch": 45.4141689373297,
+ "grad_norm": 2.7348692417144775,
+ "learning_rate": 1.1958313829460406e-05,
+ "loss": 0.0594,
+ "step": 16667
+ },
+ {
+ "epoch": 45.416893732970024,
+ "grad_norm": 2.765315055847168,
+ "learning_rate": 1.1957448414429218e-05,
+ "loss": 0.0499,
+ "step": 16668
+ },
+ {
+ "epoch": 45.41961852861036,
+ "grad_norm": 3.190347671508789,
+ "learning_rate": 1.1956582984153479e-05,
+ "loss": 0.0708,
+ "step": 16669
+ },
+ {
+ "epoch": 45.42234332425068,
+ "grad_norm": 3.0375659465789795,
+ "learning_rate": 1.1955717538639941e-05,
+ "loss": 0.1235,
+ "step": 16670
+ },
+ {
+ "epoch": 45.42506811989101,
+ "grad_norm": 2.8862404823303223,
+ "learning_rate": 1.1954852077895335e-05,
+ "loss": 0.0982,
+ "step": 16671
+ },
+ {
+ "epoch": 45.427792915531334,
+ "grad_norm": 2.5782454013824463,
+ "learning_rate": 1.1953986601926406e-05,
+ "loss": 0.0645,
+ "step": 16672
+ },
+ {
+ "epoch": 45.43051771117166,
+ "grad_norm": 2.413180351257324,
+ "learning_rate": 1.195312111073989e-05,
+ "loss": 0.0936,
+ "step": 16673
+ },
+ {
+ "epoch": 45.433242506811986,
+ "grad_norm": 3.3797154426574707,
+ "learning_rate": 1.195225560434253e-05,
+ "loss": 0.0851,
+ "step": 16674
+ },
+ {
+ "epoch": 45.43596730245232,
+ "grad_norm": 3.8783788681030273,
+ "learning_rate": 1.1951390082741068e-05,
+ "loss": 0.1508,
+ "step": 16675
+ },
+ {
+ "epoch": 45.438692098092645,
+ "grad_norm": 3.2792305946350098,
+ "learning_rate": 1.1950524545942241e-05,
+ "loss": 0.1316,
+ "step": 16676
+ },
+ {
+ "epoch": 45.44141689373297,
+ "grad_norm": 3.0357844829559326,
+ "learning_rate": 1.1949658993952793e-05,
+ "loss": 0.0661,
+ "step": 16677
+ },
+ {
+ "epoch": 45.444141689373296,
+ "grad_norm": 2.6011650562286377,
+ "learning_rate": 1.1948793426779462e-05,
+ "loss": 0.1408,
+ "step": 16678
+ },
+ {
+ "epoch": 45.44686648501362,
+ "grad_norm": 2.859628200531006,
+ "learning_rate": 1.194792784442899e-05,
+ "loss": 0.0632,
+ "step": 16679
+ },
+ {
+ "epoch": 45.44959128065395,
+ "grad_norm": 2.8607404232025146,
+ "learning_rate": 1.1947062246908121e-05,
+ "loss": 0.0806,
+ "step": 16680
+ },
+ {
+ "epoch": 45.45231607629428,
+ "grad_norm": 2.5366296768188477,
+ "learning_rate": 1.1946196634223592e-05,
+ "loss": 0.0701,
+ "step": 16681
+ },
+ {
+ "epoch": 45.45504087193461,
+ "grad_norm": 2.7465343475341797,
+ "learning_rate": 1.1945331006382148e-05,
+ "loss": 0.0776,
+ "step": 16682
+ },
+ {
+ "epoch": 45.45776566757493,
+ "grad_norm": 3.4747183322906494,
+ "learning_rate": 1.1944465363390527e-05,
+ "loss": 0.0836,
+ "step": 16683
+ },
+ {
+ "epoch": 45.46049046321526,
+ "grad_norm": 2.4223506450653076,
+ "learning_rate": 1.1943599705255476e-05,
+ "loss": 0.0789,
+ "step": 16684
+ },
+ {
+ "epoch": 45.463215258855584,
+ "grad_norm": 2.3778491020202637,
+ "learning_rate": 1.1942734031983727e-05,
+ "loss": 0.159,
+ "step": 16685
+ },
+ {
+ "epoch": 45.46594005449591,
+ "grad_norm": 3.279966354370117,
+ "learning_rate": 1.1941868343582031e-05,
+ "loss": 0.1093,
+ "step": 16686
+ },
+ {
+ "epoch": 45.46866485013624,
+ "grad_norm": 3.0196704864501953,
+ "learning_rate": 1.1941002640057126e-05,
+ "loss": 0.0783,
+ "step": 16687
+ },
+ {
+ "epoch": 45.47138964577657,
+ "grad_norm": 2.6855292320251465,
+ "learning_rate": 1.1940136921415757e-05,
+ "loss": 0.1789,
+ "step": 16688
+ },
+ {
+ "epoch": 45.474114441416894,
+ "grad_norm": 2.550710439682007,
+ "learning_rate": 1.1939271187664664e-05,
+ "loss": 0.0774,
+ "step": 16689
+ },
+ {
+ "epoch": 45.47683923705722,
+ "grad_norm": 2.2856638431549072,
+ "learning_rate": 1.1938405438810588e-05,
+ "loss": 0.0515,
+ "step": 16690
+ },
+ {
+ "epoch": 45.479564032697546,
+ "grad_norm": 3.00111985206604,
+ "learning_rate": 1.193753967486027e-05,
+ "loss": 0.1319,
+ "step": 16691
+ },
+ {
+ "epoch": 45.48228882833787,
+ "grad_norm": 3.0009803771972656,
+ "learning_rate": 1.1936673895820461e-05,
+ "loss": 0.1149,
+ "step": 16692
+ },
+ {
+ "epoch": 45.485013623978205,
+ "grad_norm": 2.432760000228882,
+ "learning_rate": 1.1935808101697895e-05,
+ "loss": 0.0752,
+ "step": 16693
+ },
+ {
+ "epoch": 45.48773841961853,
+ "grad_norm": 2.1018800735473633,
+ "learning_rate": 1.1934942292499316e-05,
+ "loss": 0.0537,
+ "step": 16694
+ },
+ {
+ "epoch": 45.490463215258856,
+ "grad_norm": 2.8626649379730225,
+ "learning_rate": 1.1934076468231472e-05,
+ "loss": 0.096,
+ "step": 16695
+ },
+ {
+ "epoch": 45.49318801089918,
+ "grad_norm": 17.45563507080078,
+ "learning_rate": 1.1933210628901101e-05,
+ "loss": 0.1153,
+ "step": 16696
+ },
+ {
+ "epoch": 45.49591280653951,
+ "grad_norm": 2.7457473278045654,
+ "learning_rate": 1.193234477451495e-05,
+ "loss": 0.0545,
+ "step": 16697
+ },
+ {
+ "epoch": 45.49863760217983,
+ "grad_norm": 2.6637063026428223,
+ "learning_rate": 1.1931478905079758e-05,
+ "loss": 0.1335,
+ "step": 16698
+ },
+ {
+ "epoch": 45.50136239782017,
+ "grad_norm": 2.9527390003204346,
+ "learning_rate": 1.1930613020602273e-05,
+ "loss": 0.0837,
+ "step": 16699
+ },
+ {
+ "epoch": 45.50408719346049,
+ "grad_norm": 2.694767951965332,
+ "learning_rate": 1.1929747121089232e-05,
+ "loss": 0.0624,
+ "step": 16700
+ },
+ {
+ "epoch": 45.50681198910082,
+ "grad_norm": 2.0819694995880127,
+ "learning_rate": 1.1928881206547385e-05,
+ "loss": 0.0545,
+ "step": 16701
+ },
+ {
+ "epoch": 45.509536784741144,
+ "grad_norm": 3.3149962425231934,
+ "learning_rate": 1.1928015276983473e-05,
+ "loss": 0.1038,
+ "step": 16702
+ },
+ {
+ "epoch": 45.51226158038147,
+ "grad_norm": 2.862412929534912,
+ "learning_rate": 1.192714933240424e-05,
+ "loss": 0.1126,
+ "step": 16703
+ },
+ {
+ "epoch": 45.514986376021795,
+ "grad_norm": 3.04461669921875,
+ "learning_rate": 1.192628337281643e-05,
+ "loss": 0.0663,
+ "step": 16704
+ },
+ {
+ "epoch": 45.51771117166213,
+ "grad_norm": 2.9301598072052,
+ "learning_rate": 1.1925417398226786e-05,
+ "loss": 0.2074,
+ "step": 16705
+ },
+ {
+ "epoch": 45.520435967302454,
+ "grad_norm": 2.3617725372314453,
+ "learning_rate": 1.1924551408642054e-05,
+ "loss": 0.0752,
+ "step": 16706
+ },
+ {
+ "epoch": 45.52316076294278,
+ "grad_norm": 2.8438985347747803,
+ "learning_rate": 1.1923685404068978e-05,
+ "loss": 0.0462,
+ "step": 16707
+ },
+ {
+ "epoch": 45.525885558583106,
+ "grad_norm": 3.721480131149292,
+ "learning_rate": 1.1922819384514302e-05,
+ "loss": 0.1045,
+ "step": 16708
+ },
+ {
+ "epoch": 45.52861035422343,
+ "grad_norm": 5.619823932647705,
+ "learning_rate": 1.1921953349984771e-05,
+ "loss": 0.1956,
+ "step": 16709
+ },
+ {
+ "epoch": 45.53133514986376,
+ "grad_norm": 2.987964391708374,
+ "learning_rate": 1.1921087300487128e-05,
+ "loss": 0.0654,
+ "step": 16710
+ },
+ {
+ "epoch": 45.53405994550409,
+ "grad_norm": 3.1839730739593506,
+ "learning_rate": 1.1920221236028122e-05,
+ "loss": 0.1027,
+ "step": 16711
+ },
+ {
+ "epoch": 45.536784741144416,
+ "grad_norm": 3.163536310195923,
+ "learning_rate": 1.1919355156614488e-05,
+ "loss": 0.206,
+ "step": 16712
+ },
+ {
+ "epoch": 45.53950953678474,
+ "grad_norm": 3.4473304748535156,
+ "learning_rate": 1.1918489062252982e-05,
+ "loss": 0.2064,
+ "step": 16713
+ },
+ {
+ "epoch": 45.54223433242507,
+ "grad_norm": 2.228466272354126,
+ "learning_rate": 1.1917622952950346e-05,
+ "loss": 0.039,
+ "step": 16714
+ },
+ {
+ "epoch": 45.54495912806539,
+ "grad_norm": 3.231025218963623,
+ "learning_rate": 1.1916756828713326e-05,
+ "loss": 0.0784,
+ "step": 16715
+ },
+ {
+ "epoch": 45.54768392370572,
+ "grad_norm": 2.3898978233337402,
+ "learning_rate": 1.191589068954866e-05,
+ "loss": 0.1637,
+ "step": 16716
+ },
+ {
+ "epoch": 45.55040871934605,
+ "grad_norm": 2.9177451133728027,
+ "learning_rate": 1.1915024535463102e-05,
+ "loss": 0.1198,
+ "step": 16717
+ },
+ {
+ "epoch": 45.55313351498638,
+ "grad_norm": 7.111724376678467,
+ "learning_rate": 1.1914158366463392e-05,
+ "loss": 0.0729,
+ "step": 16718
+ },
+ {
+ "epoch": 45.555858310626704,
+ "grad_norm": 2.9408998489379883,
+ "learning_rate": 1.1913292182556284e-05,
+ "loss": 0.0609,
+ "step": 16719
+ },
+ {
+ "epoch": 45.55858310626703,
+ "grad_norm": 3.4024877548217773,
+ "learning_rate": 1.191242598374851e-05,
+ "loss": 0.1254,
+ "step": 16720
+ },
+ {
+ "epoch": 45.561307901907355,
+ "grad_norm": 3.2503578662872314,
+ "learning_rate": 1.191155977004683e-05,
+ "loss": 0.1428,
+ "step": 16721
+ },
+ {
+ "epoch": 45.56403269754768,
+ "grad_norm": 3.08888840675354,
+ "learning_rate": 1.1910693541457983e-05,
+ "loss": 0.0608,
+ "step": 16722
+ },
+ {
+ "epoch": 45.566757493188014,
+ "grad_norm": 2.765517234802246,
+ "learning_rate": 1.1909827297988716e-05,
+ "loss": 0.0661,
+ "step": 16723
+ },
+ {
+ "epoch": 45.56948228882834,
+ "grad_norm": 3.1784656047821045,
+ "learning_rate": 1.1908961039645774e-05,
+ "loss": 0.1945,
+ "step": 16724
+ },
+ {
+ "epoch": 45.572207084468666,
+ "grad_norm": 3.3831019401550293,
+ "learning_rate": 1.1908094766435906e-05,
+ "loss": 0.1262,
+ "step": 16725
+ },
+ {
+ "epoch": 45.57493188010899,
+ "grad_norm": 2.7335007190704346,
+ "learning_rate": 1.190722847836586e-05,
+ "loss": 0.1291,
+ "step": 16726
+ },
+ {
+ "epoch": 45.57765667574932,
+ "grad_norm": 2.594881772994995,
+ "learning_rate": 1.1906362175442376e-05,
+ "loss": 0.1015,
+ "step": 16727
+ },
+ {
+ "epoch": 45.58038147138964,
+ "grad_norm": 2.034952163696289,
+ "learning_rate": 1.1905495857672207e-05,
+ "loss": 0.0508,
+ "step": 16728
+ },
+ {
+ "epoch": 45.583106267029976,
+ "grad_norm": 3.1674740314483643,
+ "learning_rate": 1.1904629525062094e-05,
+ "loss": 0.0689,
+ "step": 16729
+ },
+ {
+ "epoch": 45.5858310626703,
+ "grad_norm": 3.0371036529541016,
+ "learning_rate": 1.190376317761879e-05,
+ "loss": 0.1477,
+ "step": 16730
+ },
+ {
+ "epoch": 45.58855585831063,
+ "grad_norm": 3.039764642715454,
+ "learning_rate": 1.1902896815349041e-05,
+ "loss": 0.2291,
+ "step": 16731
+ },
+ {
+ "epoch": 45.59128065395095,
+ "grad_norm": 3.625026226043701,
+ "learning_rate": 1.1902030438259593e-05,
+ "loss": 0.1758,
+ "step": 16732
+ },
+ {
+ "epoch": 45.59400544959128,
+ "grad_norm": 6.789809703826904,
+ "learning_rate": 1.1901164046357192e-05,
+ "loss": 0.1911,
+ "step": 16733
+ },
+ {
+ "epoch": 45.596730245231605,
+ "grad_norm": 2.300203323364258,
+ "learning_rate": 1.1900297639648586e-05,
+ "loss": 0.1386,
+ "step": 16734
+ },
+ {
+ "epoch": 45.59945504087194,
+ "grad_norm": 2.52565336227417,
+ "learning_rate": 1.1899431218140525e-05,
+ "loss": 0.0571,
+ "step": 16735
+ },
+ {
+ "epoch": 45.60217983651226,
+ "grad_norm": 2.686793565750122,
+ "learning_rate": 1.1898564781839752e-05,
+ "loss": 0.1075,
+ "step": 16736
+ },
+ {
+ "epoch": 45.60490463215259,
+ "grad_norm": 4.99215841293335,
+ "learning_rate": 1.1897698330753017e-05,
+ "loss": 0.154,
+ "step": 16737
+ },
+ {
+ "epoch": 45.607629427792915,
+ "grad_norm": 3.3753929138183594,
+ "learning_rate": 1.1896831864887071e-05,
+ "loss": 0.0584,
+ "step": 16738
+ },
+ {
+ "epoch": 45.61035422343324,
+ "grad_norm": 10.69469928741455,
+ "learning_rate": 1.189596538424866e-05,
+ "loss": 0.0526,
+ "step": 16739
+ },
+ {
+ "epoch": 45.61307901907357,
+ "grad_norm": 2.753396511077881,
+ "learning_rate": 1.189509888884453e-05,
+ "loss": 0.1721,
+ "step": 16740
+ },
+ {
+ "epoch": 45.6158038147139,
+ "grad_norm": 2.9938485622406006,
+ "learning_rate": 1.1894232378681432e-05,
+ "loss": 0.0729,
+ "step": 16741
+ },
+ {
+ "epoch": 45.618528610354225,
+ "grad_norm": 2.5327301025390625,
+ "learning_rate": 1.1893365853766112e-05,
+ "loss": 0.0633,
+ "step": 16742
+ },
+ {
+ "epoch": 45.62125340599455,
+ "grad_norm": 3.068667411804199,
+ "learning_rate": 1.1892499314105322e-05,
+ "loss": 0.0785,
+ "step": 16743
+ },
+ {
+ "epoch": 45.62397820163488,
+ "grad_norm": 2.6132335662841797,
+ "learning_rate": 1.1891632759705806e-05,
+ "loss": 0.1621,
+ "step": 16744
+ },
+ {
+ "epoch": 45.6267029972752,
+ "grad_norm": 2.8408961296081543,
+ "learning_rate": 1.1890766190574317e-05,
+ "loss": 0.1928,
+ "step": 16745
+ },
+ {
+ "epoch": 45.62942779291553,
+ "grad_norm": 3.347186803817749,
+ "learning_rate": 1.1889899606717596e-05,
+ "loss": 0.0955,
+ "step": 16746
+ },
+ {
+ "epoch": 45.63215258855586,
+ "grad_norm": 2.910905361175537,
+ "learning_rate": 1.1889033008142405e-05,
+ "loss": 0.0799,
+ "step": 16747
+ },
+ {
+ "epoch": 45.63487738419619,
+ "grad_norm": 3.0595028400421143,
+ "learning_rate": 1.188816639485548e-05,
+ "loss": 0.0808,
+ "step": 16748
+ },
+ {
+ "epoch": 45.63760217983651,
+ "grad_norm": 4.1445231437683105,
+ "learning_rate": 1.188729976686358e-05,
+ "loss": 0.2482,
+ "step": 16749
+ },
+ {
+ "epoch": 45.64032697547684,
+ "grad_norm": 3.5138559341430664,
+ "learning_rate": 1.1886433124173448e-05,
+ "loss": 0.0991,
+ "step": 16750
+ },
+ {
+ "epoch": 45.643051771117165,
+ "grad_norm": 4.571988105773926,
+ "learning_rate": 1.1885566466791838e-05,
+ "loss": 0.0726,
+ "step": 16751
+ },
+ {
+ "epoch": 45.64577656675749,
+ "grad_norm": 4.608826637268066,
+ "learning_rate": 1.1884699794725494e-05,
+ "loss": 0.1025,
+ "step": 16752
+ },
+ {
+ "epoch": 45.64850136239782,
+ "grad_norm": 2.3876688480377197,
+ "learning_rate": 1.1883833107981173e-05,
+ "loss": 0.0577,
+ "step": 16753
+ },
+ {
+ "epoch": 45.65122615803815,
+ "grad_norm": 3.069201946258545,
+ "learning_rate": 1.1882966406565615e-05,
+ "loss": 0.2042,
+ "step": 16754
+ },
+ {
+ "epoch": 45.653950953678475,
+ "grad_norm": 4.569705009460449,
+ "learning_rate": 1.1882099690485578e-05,
+ "loss": 0.1129,
+ "step": 16755
+ },
+ {
+ "epoch": 45.6566757493188,
+ "grad_norm": 3.269584894180298,
+ "learning_rate": 1.1881232959747808e-05,
+ "loss": 0.0862,
+ "step": 16756
+ },
+ {
+ "epoch": 45.65940054495913,
+ "grad_norm": 2.355178117752075,
+ "learning_rate": 1.1880366214359058e-05,
+ "loss": 0.0582,
+ "step": 16757
+ },
+ {
+ "epoch": 45.66212534059945,
+ "grad_norm": 2.9501373767852783,
+ "learning_rate": 1.1879499454326075e-05,
+ "loss": 0.0468,
+ "step": 16758
+ },
+ {
+ "epoch": 45.664850136239785,
+ "grad_norm": 2.6920409202575684,
+ "learning_rate": 1.1878632679655613e-05,
+ "loss": 0.0596,
+ "step": 16759
+ },
+ {
+ "epoch": 45.66757493188011,
+ "grad_norm": 3.0794029235839844,
+ "learning_rate": 1.1877765890354416e-05,
+ "loss": 0.0417,
+ "step": 16760
+ },
+ {
+ "epoch": 45.67029972752044,
+ "grad_norm": 3.846254587173462,
+ "learning_rate": 1.1876899086429243e-05,
+ "loss": 0.1109,
+ "step": 16761
+ },
+ {
+ "epoch": 45.67302452316076,
+ "grad_norm": 4.155247688293457,
+ "learning_rate": 1.1876032267886836e-05,
+ "loss": 0.1066,
+ "step": 16762
+ },
+ {
+ "epoch": 45.67574931880109,
+ "grad_norm": 5.453339099884033,
+ "learning_rate": 1.1875165434733953e-05,
+ "loss": 0.2101,
+ "step": 16763
+ },
+ {
+ "epoch": 45.678474114441414,
+ "grad_norm": 2.805898666381836,
+ "learning_rate": 1.1874298586977343e-05,
+ "loss": 0.0988,
+ "step": 16764
+ },
+ {
+ "epoch": 45.68119891008175,
+ "grad_norm": 3.337223768234253,
+ "learning_rate": 1.1873431724623754e-05,
+ "loss": 0.0867,
+ "step": 16765
+ },
+ {
+ "epoch": 45.68392370572207,
+ "grad_norm": 2.5902702808380127,
+ "learning_rate": 1.1872564847679937e-05,
+ "loss": 0.0584,
+ "step": 16766
+ },
+ {
+ "epoch": 45.6866485013624,
+ "grad_norm": 3.858363389968872,
+ "learning_rate": 1.187169795615265e-05,
+ "loss": 0.2519,
+ "step": 16767
+ },
+ {
+ "epoch": 45.689373297002724,
+ "grad_norm": 4.544195652008057,
+ "learning_rate": 1.1870831050048637e-05,
+ "loss": 0.099,
+ "step": 16768
+ },
+ {
+ "epoch": 45.69209809264305,
+ "grad_norm": 2.9737188816070557,
+ "learning_rate": 1.1869964129374653e-05,
+ "loss": 0.0751,
+ "step": 16769
+ },
+ {
+ "epoch": 45.694822888283376,
+ "grad_norm": 4.128628253936768,
+ "learning_rate": 1.1869097194137447e-05,
+ "loss": 0.0873,
+ "step": 16770
+ },
+ {
+ "epoch": 45.69754768392371,
+ "grad_norm": 3.2894160747528076,
+ "learning_rate": 1.1868230244343772e-05,
+ "loss": 0.1043,
+ "step": 16771
+ },
+ {
+ "epoch": 45.700272479564035,
+ "grad_norm": 2.5615689754486084,
+ "learning_rate": 1.1867363280000381e-05,
+ "loss": 0.057,
+ "step": 16772
+ },
+ {
+ "epoch": 45.70299727520436,
+ "grad_norm": 3.408125877380371,
+ "learning_rate": 1.1866496301114024e-05,
+ "loss": 0.0959,
+ "step": 16773
+ },
+ {
+ "epoch": 45.705722070844686,
+ "grad_norm": 3.30910587310791,
+ "learning_rate": 1.1865629307691457e-05,
+ "loss": 0.0798,
+ "step": 16774
+ },
+ {
+ "epoch": 45.70844686648501,
+ "grad_norm": 4.538061141967773,
+ "learning_rate": 1.1864762299739425e-05,
+ "loss": 0.1184,
+ "step": 16775
+ },
+ {
+ "epoch": 45.71117166212534,
+ "grad_norm": 4.762657165527344,
+ "learning_rate": 1.1863895277264685e-05,
+ "loss": 0.1597,
+ "step": 16776
+ },
+ {
+ "epoch": 45.71389645776567,
+ "grad_norm": 4.24841833114624,
+ "learning_rate": 1.186302824027399e-05,
+ "loss": 0.2056,
+ "step": 16777
+ },
+ {
+ "epoch": 45.716621253406,
+ "grad_norm": 3.6817119121551514,
+ "learning_rate": 1.186216118877409e-05,
+ "loss": 0.1447,
+ "step": 16778
+ },
+ {
+ "epoch": 45.71934604904632,
+ "grad_norm": 3.944566249847412,
+ "learning_rate": 1.186129412277174e-05,
+ "loss": 0.0815,
+ "step": 16779
+ },
+ {
+ "epoch": 45.72207084468665,
+ "grad_norm": 2.764177083969116,
+ "learning_rate": 1.186042704227369e-05,
+ "loss": 0.0667,
+ "step": 16780
+ },
+ {
+ "epoch": 45.724795640326974,
+ "grad_norm": 3.289802312850952,
+ "learning_rate": 1.1859559947286695e-05,
+ "loss": 0.1187,
+ "step": 16781
+ },
+ {
+ "epoch": 45.7275204359673,
+ "grad_norm": 2.475689649581909,
+ "learning_rate": 1.1858692837817507e-05,
+ "loss": 0.1734,
+ "step": 16782
+ },
+ {
+ "epoch": 45.73024523160763,
+ "grad_norm": 3.835944890975952,
+ "learning_rate": 1.1857825713872877e-05,
+ "loss": 0.108,
+ "step": 16783
+ },
+ {
+ "epoch": 45.73297002724796,
+ "grad_norm": 3.364121675491333,
+ "learning_rate": 1.1856958575459563e-05,
+ "loss": 0.1276,
+ "step": 16784
+ },
+ {
+ "epoch": 45.735694822888284,
+ "grad_norm": 2.972400665283203,
+ "learning_rate": 1.1856091422584312e-05,
+ "loss": 0.122,
+ "step": 16785
+ },
+ {
+ "epoch": 45.73841961852861,
+ "grad_norm": 2.9779438972473145,
+ "learning_rate": 1.1855224255253885e-05,
+ "loss": 0.0884,
+ "step": 16786
+ },
+ {
+ "epoch": 45.741144414168936,
+ "grad_norm": 5.791435718536377,
+ "learning_rate": 1.185435707347503e-05,
+ "loss": 0.0785,
+ "step": 16787
+ },
+ {
+ "epoch": 45.74386920980926,
+ "grad_norm": 3.2473738193511963,
+ "learning_rate": 1.1853489877254498e-05,
+ "loss": 0.0917,
+ "step": 16788
+ },
+ {
+ "epoch": 45.746594005449595,
+ "grad_norm": 2.670070171356201,
+ "learning_rate": 1.1852622666599047e-05,
+ "loss": 0.1465,
+ "step": 16789
+ },
+ {
+ "epoch": 45.74931880108992,
+ "grad_norm": 3.4070675373077393,
+ "learning_rate": 1.1851755441515435e-05,
+ "loss": 0.1116,
+ "step": 16790
+ },
+ {
+ "epoch": 45.752043596730246,
+ "grad_norm": 3.476510763168335,
+ "learning_rate": 1.1850888202010408e-05,
+ "loss": 0.0768,
+ "step": 16791
+ },
+ {
+ "epoch": 45.75476839237057,
+ "grad_norm": 2.2928335666656494,
+ "learning_rate": 1.1850020948090725e-05,
+ "loss": 0.0967,
+ "step": 16792
+ },
+ {
+ "epoch": 45.7574931880109,
+ "grad_norm": 2.7685813903808594,
+ "learning_rate": 1.1849153679763137e-05,
+ "loss": 0.0689,
+ "step": 16793
+ },
+ {
+ "epoch": 45.76021798365122,
+ "grad_norm": 3.9588420391082764,
+ "learning_rate": 1.1848286397034402e-05,
+ "loss": 0.0862,
+ "step": 16794
+ },
+ {
+ "epoch": 45.762942779291556,
+ "grad_norm": 3.0907046794891357,
+ "learning_rate": 1.1847419099911268e-05,
+ "loss": 0.2033,
+ "step": 16795
+ },
+ {
+ "epoch": 45.76566757493188,
+ "grad_norm": 3.2923078536987305,
+ "learning_rate": 1.1846551788400497e-05,
+ "loss": 0.1668,
+ "step": 16796
+ },
+ {
+ "epoch": 45.76839237057221,
+ "grad_norm": 3.0714807510375977,
+ "learning_rate": 1.1845684462508838e-05,
+ "loss": 0.1055,
+ "step": 16797
+ },
+ {
+ "epoch": 45.771117166212534,
+ "grad_norm": 3.1456856727600098,
+ "learning_rate": 1.1844817122243047e-05,
+ "loss": 0.0521,
+ "step": 16798
+ },
+ {
+ "epoch": 45.77384196185286,
+ "grad_norm": 2.6661524772644043,
+ "learning_rate": 1.1843949767609879e-05,
+ "loss": 0.1167,
+ "step": 16799
+ },
+ {
+ "epoch": 45.776566757493185,
+ "grad_norm": 3.6226041316986084,
+ "learning_rate": 1.1843082398616095e-05,
+ "loss": 0.2829,
+ "step": 16800
+ },
+ {
+ "epoch": 45.77929155313352,
+ "grad_norm": 2.3243446350097656,
+ "learning_rate": 1.1842215015268439e-05,
+ "loss": 0.0573,
+ "step": 16801
+ },
+ {
+ "epoch": 45.782016348773844,
+ "grad_norm": 2.8449487686157227,
+ "learning_rate": 1.1841347617573675e-05,
+ "loss": 0.0698,
+ "step": 16802
+ },
+ {
+ "epoch": 45.78474114441417,
+ "grad_norm": 3.182638168334961,
+ "learning_rate": 1.1840480205538553e-05,
+ "loss": 0.1209,
+ "step": 16803
+ },
+ {
+ "epoch": 45.787465940054496,
+ "grad_norm": 2.531667947769165,
+ "learning_rate": 1.1839612779169833e-05,
+ "loss": 0.0833,
+ "step": 16804
+ },
+ {
+ "epoch": 45.79019073569482,
+ "grad_norm": 2.771512746810913,
+ "learning_rate": 1.1838745338474266e-05,
+ "loss": 0.3421,
+ "step": 16805
+ },
+ {
+ "epoch": 45.79291553133515,
+ "grad_norm": 3.5168192386627197,
+ "learning_rate": 1.183787788345861e-05,
+ "loss": 0.0794,
+ "step": 16806
+ },
+ {
+ "epoch": 45.79564032697548,
+ "grad_norm": 2.9356696605682373,
+ "learning_rate": 1.1837010414129617e-05,
+ "loss": 0.1183,
+ "step": 16807
+ },
+ {
+ "epoch": 45.798365122615806,
+ "grad_norm": 2.028149127960205,
+ "learning_rate": 1.1836142930494053e-05,
+ "loss": 0.0446,
+ "step": 16808
+ },
+ {
+ "epoch": 45.80108991825613,
+ "grad_norm": 3.0049898624420166,
+ "learning_rate": 1.1835275432558659e-05,
+ "loss": 0.1151,
+ "step": 16809
+ },
+ {
+ "epoch": 45.80381471389646,
+ "grad_norm": 2.3828110694885254,
+ "learning_rate": 1.1834407920330207e-05,
+ "loss": 0.1493,
+ "step": 16810
+ },
+ {
+ "epoch": 45.80653950953678,
+ "grad_norm": 4.133638858795166,
+ "learning_rate": 1.1833540393815438e-05,
+ "loss": 0.2902,
+ "step": 16811
+ },
+ {
+ "epoch": 45.80926430517711,
+ "grad_norm": 2.2309751510620117,
+ "learning_rate": 1.183267285302112e-05,
+ "loss": 0.2086,
+ "step": 16812
+ },
+ {
+ "epoch": 45.81198910081744,
+ "grad_norm": 2.6023318767547607,
+ "learning_rate": 1.1831805297954001e-05,
+ "loss": 0.0662,
+ "step": 16813
+ },
+ {
+ "epoch": 45.81471389645777,
+ "grad_norm": 2.6330177783966064,
+ "learning_rate": 1.1830937728620847e-05,
+ "loss": 0.104,
+ "step": 16814
+ },
+ {
+ "epoch": 45.817438692098094,
+ "grad_norm": 2.5253684520721436,
+ "learning_rate": 1.1830070145028404e-05,
+ "loss": 0.058,
+ "step": 16815
+ },
+ {
+ "epoch": 45.82016348773842,
+ "grad_norm": 2.7085747718811035,
+ "learning_rate": 1.1829202547183434e-05,
+ "loss": 0.0757,
+ "step": 16816
+ },
+ {
+ "epoch": 45.822888283378745,
+ "grad_norm": 2.2845003604888916,
+ "learning_rate": 1.1828334935092694e-05,
+ "loss": 0.0531,
+ "step": 16817
+ },
+ {
+ "epoch": 45.82561307901907,
+ "grad_norm": 2.895815849304199,
+ "learning_rate": 1.1827467308762943e-05,
+ "loss": 0.0873,
+ "step": 16818
+ },
+ {
+ "epoch": 45.828337874659404,
+ "grad_norm": 2.8777458667755127,
+ "learning_rate": 1.1826599668200931e-05,
+ "loss": 0.1429,
+ "step": 16819
+ },
+ {
+ "epoch": 45.83106267029973,
+ "grad_norm": 2.6922647953033447,
+ "learning_rate": 1.1825732013413424e-05,
+ "loss": 0.1248,
+ "step": 16820
+ },
+ {
+ "epoch": 45.833787465940055,
+ "grad_norm": 2.392413854598999,
+ "learning_rate": 1.1824864344407172e-05,
+ "loss": 0.0415,
+ "step": 16821
+ },
+ {
+ "epoch": 45.83651226158038,
+ "grad_norm": 2.961989164352417,
+ "learning_rate": 1.1823996661188936e-05,
+ "loss": 0.1384,
+ "step": 16822
+ },
+ {
+ "epoch": 45.83923705722071,
+ "grad_norm": 3.208449363708496,
+ "learning_rate": 1.1823128963765472e-05,
+ "loss": 0.0921,
+ "step": 16823
+ },
+ {
+ "epoch": 45.84196185286103,
+ "grad_norm": 2.413182258605957,
+ "learning_rate": 1.1822261252143538e-05,
+ "loss": 0.0681,
+ "step": 16824
+ },
+ {
+ "epoch": 45.844686648501366,
+ "grad_norm": 3.2372665405273438,
+ "learning_rate": 1.1821393526329894e-05,
+ "loss": 0.0593,
+ "step": 16825
+ },
+ {
+ "epoch": 45.84741144414169,
+ "grad_norm": 2.7608673572540283,
+ "learning_rate": 1.1820525786331296e-05,
+ "loss": 0.054,
+ "step": 16826
+ },
+ {
+ "epoch": 45.85013623978202,
+ "grad_norm": 2.6361424922943115,
+ "learning_rate": 1.18196580321545e-05,
+ "loss": 0.0587,
+ "step": 16827
+ },
+ {
+ "epoch": 45.85286103542234,
+ "grad_norm": 2.136610507965088,
+ "learning_rate": 1.1818790263806265e-05,
+ "loss": 0.0403,
+ "step": 16828
+ },
+ {
+ "epoch": 45.85558583106267,
+ "grad_norm": 3.7684109210968018,
+ "learning_rate": 1.1817922481293351e-05,
+ "loss": 0.1524,
+ "step": 16829
+ },
+ {
+ "epoch": 45.858310626702995,
+ "grad_norm": 3.5983951091766357,
+ "learning_rate": 1.1817054684622518e-05,
+ "loss": 0.0912,
+ "step": 16830
+ },
+ {
+ "epoch": 45.86103542234333,
+ "grad_norm": 3.3669097423553467,
+ "learning_rate": 1.1816186873800519e-05,
+ "loss": 0.1253,
+ "step": 16831
+ },
+ {
+ "epoch": 45.86376021798365,
+ "grad_norm": 2.8445003032684326,
+ "learning_rate": 1.1815319048834113e-05,
+ "loss": 0.2226,
+ "step": 16832
+ },
+ {
+ "epoch": 45.86648501362398,
+ "grad_norm": 2.7046456336975098,
+ "learning_rate": 1.1814451209730062e-05,
+ "loss": 0.0648,
+ "step": 16833
+ },
+ {
+ "epoch": 45.869209809264305,
+ "grad_norm": 2.4508683681488037,
+ "learning_rate": 1.1813583356495126e-05,
+ "loss": 0.0616,
+ "step": 16834
+ },
+ {
+ "epoch": 45.87193460490463,
+ "grad_norm": 2.943197011947632,
+ "learning_rate": 1.1812715489136059e-05,
+ "loss": 0.0769,
+ "step": 16835
+ },
+ {
+ "epoch": 45.87465940054496,
+ "grad_norm": 3.3920745849609375,
+ "learning_rate": 1.1811847607659623e-05,
+ "loss": 0.2077,
+ "step": 16836
+ },
+ {
+ "epoch": 45.87738419618529,
+ "grad_norm": 3.0366625785827637,
+ "learning_rate": 1.1810979712072577e-05,
+ "loss": 0.2325,
+ "step": 16837
+ },
+ {
+ "epoch": 45.880108991825615,
+ "grad_norm": 3.57977557182312,
+ "learning_rate": 1.1810111802381679e-05,
+ "loss": 0.0835,
+ "step": 16838
+ },
+ {
+ "epoch": 45.88283378746594,
+ "grad_norm": 3.264902353286743,
+ "learning_rate": 1.1809243878593686e-05,
+ "loss": 0.1458,
+ "step": 16839
+ },
+ {
+ "epoch": 45.88555858310627,
+ "grad_norm": 2.357858419418335,
+ "learning_rate": 1.1808375940715362e-05,
+ "loss": 0.0727,
+ "step": 16840
+ },
+ {
+ "epoch": 45.88828337874659,
+ "grad_norm": 2.724946975708008,
+ "learning_rate": 1.1807507988753461e-05,
+ "loss": 0.0512,
+ "step": 16841
+ },
+ {
+ "epoch": 45.89100817438692,
+ "grad_norm": 3.42793607711792,
+ "learning_rate": 1.180664002271475e-05,
+ "loss": 0.098,
+ "step": 16842
+ },
+ {
+ "epoch": 45.89373297002725,
+ "grad_norm": 2.8697497844696045,
+ "learning_rate": 1.1805772042605982e-05,
+ "loss": 0.0673,
+ "step": 16843
+ },
+ {
+ "epoch": 45.89645776566758,
+ "grad_norm": 2.569232225418091,
+ "learning_rate": 1.1804904048433925e-05,
+ "loss": 0.0547,
+ "step": 16844
+ },
+ {
+ "epoch": 45.8991825613079,
+ "grad_norm": 2.841993808746338,
+ "learning_rate": 1.1804036040205326e-05,
+ "loss": 0.1263,
+ "step": 16845
+ },
+ {
+ "epoch": 45.90190735694823,
+ "grad_norm": 3.391345739364624,
+ "learning_rate": 1.1803168017926958e-05,
+ "loss": 0.0721,
+ "step": 16846
+ },
+ {
+ "epoch": 45.904632152588555,
+ "grad_norm": 2.9366352558135986,
+ "learning_rate": 1.1802299981605572e-05,
+ "loss": 0.0832,
+ "step": 16847
+ },
+ {
+ "epoch": 45.90735694822888,
+ "grad_norm": 2.232724666595459,
+ "learning_rate": 1.1801431931247936e-05,
+ "loss": 0.0449,
+ "step": 16848
+ },
+ {
+ "epoch": 45.91008174386921,
+ "grad_norm": 2.706179618835449,
+ "learning_rate": 1.1800563866860802e-05,
+ "loss": 0.0816,
+ "step": 16849
+ },
+ {
+ "epoch": 45.91280653950954,
+ "grad_norm": 2.1850194931030273,
+ "learning_rate": 1.1799695788450933e-05,
+ "loss": 0.0695,
+ "step": 16850
+ },
+ {
+ "epoch": 45.915531335149865,
+ "grad_norm": 2.804032802581787,
+ "learning_rate": 1.1798827696025094e-05,
+ "loss": 0.094,
+ "step": 16851
+ },
+ {
+ "epoch": 45.91825613079019,
+ "grad_norm": 2.3116488456726074,
+ "learning_rate": 1.1797959589590042e-05,
+ "loss": 0.2289,
+ "step": 16852
+ },
+ {
+ "epoch": 45.920980926430516,
+ "grad_norm": 5.112425804138184,
+ "learning_rate": 1.179709146915254e-05,
+ "loss": 0.0723,
+ "step": 16853
+ },
+ {
+ "epoch": 45.92370572207084,
+ "grad_norm": 3.1547420024871826,
+ "learning_rate": 1.1796223334719347e-05,
+ "loss": 0.2713,
+ "step": 16854
+ },
+ {
+ "epoch": 45.926430517711175,
+ "grad_norm": 2.576367139816284,
+ "learning_rate": 1.1795355186297223e-05,
+ "loss": 0.0875,
+ "step": 16855
+ },
+ {
+ "epoch": 45.9291553133515,
+ "grad_norm": 4.246774673461914,
+ "learning_rate": 1.1794487023892934e-05,
+ "loss": 0.063,
+ "step": 16856
+ },
+ {
+ "epoch": 45.93188010899183,
+ "grad_norm": 3.1988933086395264,
+ "learning_rate": 1.1793618847513235e-05,
+ "loss": 0.0621,
+ "step": 16857
+ },
+ {
+ "epoch": 45.93460490463215,
+ "grad_norm": 3.58183217048645,
+ "learning_rate": 1.179275065716489e-05,
+ "loss": 0.181,
+ "step": 16858
+ },
+ {
+ "epoch": 45.93732970027248,
+ "grad_norm": 2.534698486328125,
+ "learning_rate": 1.1791882452854661e-05,
+ "loss": 0.1288,
+ "step": 16859
+ },
+ {
+ "epoch": 45.940054495912804,
+ "grad_norm": 3.112786293029785,
+ "learning_rate": 1.1791014234589308e-05,
+ "loss": 0.1354,
+ "step": 16860
+ },
+ {
+ "epoch": 45.94277929155314,
+ "grad_norm": 3.178510904312134,
+ "learning_rate": 1.1790146002375597e-05,
+ "loss": 0.2012,
+ "step": 16861
+ },
+ {
+ "epoch": 45.94550408719346,
+ "grad_norm": 2.372830629348755,
+ "learning_rate": 1.1789277756220283e-05,
+ "loss": 0.0547,
+ "step": 16862
+ },
+ {
+ "epoch": 45.94822888283379,
+ "grad_norm": 3.0872740745544434,
+ "learning_rate": 1.1788409496130133e-05,
+ "loss": 0.0632,
+ "step": 16863
+ },
+ {
+ "epoch": 45.950953678474114,
+ "grad_norm": 2.5911574363708496,
+ "learning_rate": 1.178754122211191e-05,
+ "loss": 0.1956,
+ "step": 16864
+ },
+ {
+ "epoch": 45.95367847411444,
+ "grad_norm": 9.722040176391602,
+ "learning_rate": 1.178667293417237e-05,
+ "loss": 0.1197,
+ "step": 16865
+ },
+ {
+ "epoch": 45.956403269754766,
+ "grad_norm": 2.5803961753845215,
+ "learning_rate": 1.1785804632318278e-05,
+ "loss": 0.0441,
+ "step": 16866
+ },
+ {
+ "epoch": 45.95912806539509,
+ "grad_norm": 2.2330307960510254,
+ "learning_rate": 1.1784936316556399e-05,
+ "loss": 0.0767,
+ "step": 16867
+ },
+ {
+ "epoch": 45.961852861035425,
+ "grad_norm": 2.4360930919647217,
+ "learning_rate": 1.1784067986893492e-05,
+ "loss": 0.0726,
+ "step": 16868
+ },
+ {
+ "epoch": 45.96457765667575,
+ "grad_norm": 4.033812046051025,
+ "learning_rate": 1.1783199643336324e-05,
+ "loss": 0.0632,
+ "step": 16869
+ },
+ {
+ "epoch": 45.967302452316076,
+ "grad_norm": 2.9241061210632324,
+ "learning_rate": 1.1782331285891654e-05,
+ "loss": 0.2501,
+ "step": 16870
+ },
+ {
+ "epoch": 45.9700272479564,
+ "grad_norm": 3.126584053039551,
+ "learning_rate": 1.1781462914566244e-05,
+ "loss": 0.1078,
+ "step": 16871
+ },
+ {
+ "epoch": 45.97275204359673,
+ "grad_norm": 2.5806212425231934,
+ "learning_rate": 1.1780594529366856e-05,
+ "loss": 0.079,
+ "step": 16872
+ },
+ {
+ "epoch": 45.97547683923706,
+ "grad_norm": 4.516685485839844,
+ "learning_rate": 1.177972613030026e-05,
+ "loss": 0.136,
+ "step": 16873
+ },
+ {
+ "epoch": 45.97820163487739,
+ "grad_norm": 3.515037775039673,
+ "learning_rate": 1.1778857717373208e-05,
+ "loss": 0.151,
+ "step": 16874
+ },
+ {
+ "epoch": 45.98092643051771,
+ "grad_norm": 4.469524383544922,
+ "learning_rate": 1.1777989290592474e-05,
+ "loss": 0.0352,
+ "step": 16875
+ },
+ {
+ "epoch": 45.98365122615804,
+ "grad_norm": 2.8900094032287598,
+ "learning_rate": 1.1777120849964814e-05,
+ "loss": 0.0644,
+ "step": 16876
+ },
+ {
+ "epoch": 45.986376021798364,
+ "grad_norm": 3.401128053665161,
+ "learning_rate": 1.1776252395496996e-05,
+ "loss": 0.1323,
+ "step": 16877
+ },
+ {
+ "epoch": 45.98910081743869,
+ "grad_norm": 3.809197425842285,
+ "learning_rate": 1.177538392719578e-05,
+ "loss": 0.2213,
+ "step": 16878
+ },
+ {
+ "epoch": 45.991825613079016,
+ "grad_norm": 2.9543964862823486,
+ "learning_rate": 1.1774515445067933e-05,
+ "loss": 0.0559,
+ "step": 16879
+ },
+ {
+ "epoch": 45.99455040871935,
+ "grad_norm": 3.25065016746521,
+ "learning_rate": 1.1773646949120217e-05,
+ "loss": 0.1317,
+ "step": 16880
+ },
+ {
+ "epoch": 45.997275204359674,
+ "grad_norm": 4.542169094085693,
+ "learning_rate": 1.1772778439359393e-05,
+ "loss": 0.0803,
+ "step": 16881
+ },
+ {
+ "epoch": 46.0,
+ "grad_norm": 3.069019079208374,
+ "learning_rate": 1.177190991579223e-05,
+ "loss": 0.0686,
+ "step": 16882
+ },
+ {
+ "epoch": 46.002724795640326,
+ "grad_norm": 2.6269891262054443,
+ "learning_rate": 1.177104137842549e-05,
+ "loss": 0.0489,
+ "step": 16883
+ },
+ {
+ "epoch": 46.00544959128065,
+ "grad_norm": 2.626964569091797,
+ "learning_rate": 1.177017282726593e-05,
+ "loss": 0.1099,
+ "step": 16884
+ },
+ {
+ "epoch": 46.00817438692098,
+ "grad_norm": 3.7479567527770996,
+ "learning_rate": 1.1769304262320329e-05,
+ "loss": 0.1562,
+ "step": 16885
+ },
+ {
+ "epoch": 46.01089918256131,
+ "grad_norm": 2.2504074573516846,
+ "learning_rate": 1.1768435683595441e-05,
+ "loss": 0.0782,
+ "step": 16886
+ },
+ {
+ "epoch": 46.013623978201636,
+ "grad_norm": 3.319395065307617,
+ "learning_rate": 1.1767567091098033e-05,
+ "loss": 0.0657,
+ "step": 16887
+ },
+ {
+ "epoch": 46.01634877384196,
+ "grad_norm": 2.58231258392334,
+ "learning_rate": 1.1766698484834869e-05,
+ "loss": 0.0613,
+ "step": 16888
+ },
+ {
+ "epoch": 46.01907356948229,
+ "grad_norm": 2.660454511642456,
+ "learning_rate": 1.1765829864812714e-05,
+ "loss": 0.1956,
+ "step": 16889
+ },
+ {
+ "epoch": 46.02179836512261,
+ "grad_norm": 2.91058349609375,
+ "learning_rate": 1.1764961231038331e-05,
+ "loss": 0.1056,
+ "step": 16890
+ },
+ {
+ "epoch": 46.02452316076294,
+ "grad_norm": 3.6378252506256104,
+ "learning_rate": 1.1764092583518491e-05,
+ "loss": 0.0935,
+ "step": 16891
+ },
+ {
+ "epoch": 46.02724795640327,
+ "grad_norm": 3.153869390487671,
+ "learning_rate": 1.176322392225995e-05,
+ "loss": 0.0726,
+ "step": 16892
+ },
+ {
+ "epoch": 46.0299727520436,
+ "grad_norm": 3.476090431213379,
+ "learning_rate": 1.1762355247269481e-05,
+ "loss": 0.2231,
+ "step": 16893
+ },
+ {
+ "epoch": 46.032697547683924,
+ "grad_norm": 3.1227545738220215,
+ "learning_rate": 1.1761486558553843e-05,
+ "loss": 0.1153,
+ "step": 16894
+ },
+ {
+ "epoch": 46.03542234332425,
+ "grad_norm": 3.1061058044433594,
+ "learning_rate": 1.1760617856119809e-05,
+ "loss": 0.1569,
+ "step": 16895
+ },
+ {
+ "epoch": 46.038147138964575,
+ "grad_norm": 3.382739782333374,
+ "learning_rate": 1.1759749139974137e-05,
+ "loss": 0.2145,
+ "step": 16896
+ },
+ {
+ "epoch": 46.0408719346049,
+ "grad_norm": 2.0179216861724854,
+ "learning_rate": 1.1758880410123596e-05,
+ "loss": 0.0543,
+ "step": 16897
+ },
+ {
+ "epoch": 46.043596730245234,
+ "grad_norm": 2.8091375827789307,
+ "learning_rate": 1.175801166657495e-05,
+ "loss": 0.1176,
+ "step": 16898
+ },
+ {
+ "epoch": 46.04632152588556,
+ "grad_norm": 3.005152463912964,
+ "learning_rate": 1.175714290933497e-05,
+ "loss": 0.1287,
+ "step": 16899
+ },
+ {
+ "epoch": 46.049046321525886,
+ "grad_norm": 3.349978446960449,
+ "learning_rate": 1.175627413841041e-05,
+ "loss": 0.1077,
+ "step": 16900
+ },
+ {
+ "epoch": 46.05177111716621,
+ "grad_norm": 3.415827512741089,
+ "learning_rate": 1.1755405353808047e-05,
+ "loss": 0.0783,
+ "step": 16901
+ },
+ {
+ "epoch": 46.05449591280654,
+ "grad_norm": 2.737689256668091,
+ "learning_rate": 1.1754536555534643e-05,
+ "loss": 0.0655,
+ "step": 16902
+ },
+ {
+ "epoch": 46.05722070844686,
+ "grad_norm": 2.526170253753662,
+ "learning_rate": 1.1753667743596969e-05,
+ "loss": 0.0894,
+ "step": 16903
+ },
+ {
+ "epoch": 46.059945504087196,
+ "grad_norm": 3.3743221759796143,
+ "learning_rate": 1.1752798918001782e-05,
+ "loss": 0.209,
+ "step": 16904
+ },
+ {
+ "epoch": 46.06267029972752,
+ "grad_norm": 4.210108757019043,
+ "learning_rate": 1.1751930078755858e-05,
+ "loss": 0.0704,
+ "step": 16905
+ },
+ {
+ "epoch": 46.06539509536785,
+ "grad_norm": 3.5254218578338623,
+ "learning_rate": 1.1751061225865953e-05,
+ "loss": 0.0849,
+ "step": 16906
+ },
+ {
+ "epoch": 46.06811989100817,
+ "grad_norm": 3.425496816635132,
+ "learning_rate": 1.1750192359338845e-05,
+ "loss": 0.1634,
+ "step": 16907
+ },
+ {
+ "epoch": 46.0708446866485,
+ "grad_norm": 2.591372013092041,
+ "learning_rate": 1.1749323479181291e-05,
+ "loss": 0.1312,
+ "step": 16908
+ },
+ {
+ "epoch": 46.073569482288825,
+ "grad_norm": 3.095695734024048,
+ "learning_rate": 1.1748454585400064e-05,
+ "loss": 0.0529,
+ "step": 16909
+ },
+ {
+ "epoch": 46.07629427792916,
+ "grad_norm": 5.057247161865234,
+ "learning_rate": 1.1747585678001929e-05,
+ "loss": 0.1521,
+ "step": 16910
+ },
+ {
+ "epoch": 46.079019073569484,
+ "grad_norm": 2.982008457183838,
+ "learning_rate": 1.1746716756993654e-05,
+ "loss": 0.1338,
+ "step": 16911
+ },
+ {
+ "epoch": 46.08174386920981,
+ "grad_norm": 2.421123504638672,
+ "learning_rate": 1.1745847822382004e-05,
+ "loss": 0.0432,
+ "step": 16912
+ },
+ {
+ "epoch": 46.084468664850135,
+ "grad_norm": 7.2049970626831055,
+ "learning_rate": 1.174497887417375e-05,
+ "loss": 0.3055,
+ "step": 16913
+ },
+ {
+ "epoch": 46.08719346049046,
+ "grad_norm": 3.003171682357788,
+ "learning_rate": 1.1744109912375652e-05,
+ "loss": 0.1671,
+ "step": 16914
+ },
+ {
+ "epoch": 46.08991825613079,
+ "grad_norm": 2.6577820777893066,
+ "learning_rate": 1.1743240936994484e-05,
+ "loss": 0.0492,
+ "step": 16915
+ },
+ {
+ "epoch": 46.09264305177112,
+ "grad_norm": 2.546238899230957,
+ "learning_rate": 1.1742371948037012e-05,
+ "loss": 0.0688,
+ "step": 16916
+ },
+ {
+ "epoch": 46.095367847411445,
+ "grad_norm": 2.8761138916015625,
+ "learning_rate": 1.1741502945510005e-05,
+ "loss": 0.0464,
+ "step": 16917
+ },
+ {
+ "epoch": 46.09809264305177,
+ "grad_norm": 8.43749713897705,
+ "learning_rate": 1.1740633929420225e-05,
+ "loss": 0.0564,
+ "step": 16918
+ },
+ {
+ "epoch": 46.1008174386921,
+ "grad_norm": 3.1997058391571045,
+ "learning_rate": 1.1739764899774447e-05,
+ "loss": 0.0561,
+ "step": 16919
+ },
+ {
+ "epoch": 46.10354223433242,
+ "grad_norm": 3.0413131713867188,
+ "learning_rate": 1.1738895856579436e-05,
+ "loss": 0.1589,
+ "step": 16920
+ },
+ {
+ "epoch": 46.10626702997275,
+ "grad_norm": 3.0787861347198486,
+ "learning_rate": 1.173802679984196e-05,
+ "loss": 0.0492,
+ "step": 16921
+ },
+ {
+ "epoch": 46.10899182561308,
+ "grad_norm": 2.7101380825042725,
+ "learning_rate": 1.1737157729568787e-05,
+ "loss": 0.1137,
+ "step": 16922
+ },
+ {
+ "epoch": 46.11171662125341,
+ "grad_norm": 3.56730580329895,
+ "learning_rate": 1.1736288645766684e-05,
+ "loss": 0.0727,
+ "step": 16923
+ },
+ {
+ "epoch": 46.11444141689373,
+ "grad_norm": 2.3976614475250244,
+ "learning_rate": 1.1735419548442422e-05,
+ "loss": 0.1208,
+ "step": 16924
+ },
+ {
+ "epoch": 46.11716621253406,
+ "grad_norm": 3.3400278091430664,
+ "learning_rate": 1.173455043760277e-05,
+ "loss": 0.0718,
+ "step": 16925
+ },
+ {
+ "epoch": 46.119891008174385,
+ "grad_norm": 3.651416540145874,
+ "learning_rate": 1.1733681313254492e-05,
+ "loss": 0.1462,
+ "step": 16926
+ },
+ {
+ "epoch": 46.12261580381471,
+ "grad_norm": 3.370382308959961,
+ "learning_rate": 1.1732812175404362e-05,
+ "loss": 0.1016,
+ "step": 16927
+ },
+ {
+ "epoch": 46.12534059945504,
+ "grad_norm": 2.773139715194702,
+ "learning_rate": 1.1731943024059145e-05,
+ "loss": 0.0892,
+ "step": 16928
+ },
+ {
+ "epoch": 46.12806539509537,
+ "grad_norm": 2.6830224990844727,
+ "learning_rate": 1.1731073859225612e-05,
+ "loss": 0.0972,
+ "step": 16929
+ },
+ {
+ "epoch": 46.130790190735695,
+ "grad_norm": 2.9452526569366455,
+ "learning_rate": 1.173020468091053e-05,
+ "loss": 0.1364,
+ "step": 16930
+ },
+ {
+ "epoch": 46.13351498637602,
+ "grad_norm": 3.0297131538391113,
+ "learning_rate": 1.1729335489120675e-05,
+ "loss": 0.06,
+ "step": 16931
+ },
+ {
+ "epoch": 46.13623978201635,
+ "grad_norm": 2.932814359664917,
+ "learning_rate": 1.1728466283862806e-05,
+ "loss": 0.0544,
+ "step": 16932
+ },
+ {
+ "epoch": 46.13896457765667,
+ "grad_norm": 2.595360040664673,
+ "learning_rate": 1.1727597065143699e-05,
+ "loss": 0.149,
+ "step": 16933
+ },
+ {
+ "epoch": 46.141689373297005,
+ "grad_norm": 3.1727001667022705,
+ "learning_rate": 1.1726727832970121e-05,
+ "loss": 0.0874,
+ "step": 16934
+ },
+ {
+ "epoch": 46.14441416893733,
+ "grad_norm": 3.7145655155181885,
+ "learning_rate": 1.1725858587348842e-05,
+ "loss": 0.1009,
+ "step": 16935
+ },
+ {
+ "epoch": 46.14713896457766,
+ "grad_norm": 3.260016679763794,
+ "learning_rate": 1.1724989328286633e-05,
+ "loss": 0.0786,
+ "step": 16936
+ },
+ {
+ "epoch": 46.14986376021798,
+ "grad_norm": 2.1845481395721436,
+ "learning_rate": 1.172412005579026e-05,
+ "loss": 0.0471,
+ "step": 16937
+ },
+ {
+ "epoch": 46.15258855585831,
+ "grad_norm": 3.519724130630493,
+ "learning_rate": 1.1723250769866499e-05,
+ "loss": 0.0928,
+ "step": 16938
+ },
+ {
+ "epoch": 46.155313351498634,
+ "grad_norm": 2.7967240810394287,
+ "learning_rate": 1.1722381470522116e-05,
+ "loss": 0.053,
+ "step": 16939
+ },
+ {
+ "epoch": 46.15803814713897,
+ "grad_norm": 2.3894925117492676,
+ "learning_rate": 1.1721512157763878e-05,
+ "loss": 0.0554,
+ "step": 16940
+ },
+ {
+ "epoch": 46.16076294277929,
+ "grad_norm": 2.652651786804199,
+ "learning_rate": 1.1720642831598563e-05,
+ "loss": 0.1351,
+ "step": 16941
+ },
+ {
+ "epoch": 46.16348773841962,
+ "grad_norm": 3.419609785079956,
+ "learning_rate": 1.1719773492032933e-05,
+ "loss": 0.1272,
+ "step": 16942
+ },
+ {
+ "epoch": 46.166212534059945,
+ "grad_norm": 2.785417079925537,
+ "learning_rate": 1.1718904139073765e-05,
+ "loss": 0.0502,
+ "step": 16943
+ },
+ {
+ "epoch": 46.16893732970027,
+ "grad_norm": 5.423637390136719,
+ "learning_rate": 1.1718034772727827e-05,
+ "loss": 0.0866,
+ "step": 16944
+ },
+ {
+ "epoch": 46.171662125340596,
+ "grad_norm": 3.0619957447052,
+ "learning_rate": 1.1717165393001888e-05,
+ "loss": 0.1059,
+ "step": 16945
+ },
+ {
+ "epoch": 46.17438692098093,
+ "grad_norm": 2.631425380706787,
+ "learning_rate": 1.1716295999902722e-05,
+ "loss": 0.0592,
+ "step": 16946
+ },
+ {
+ "epoch": 46.177111716621255,
+ "grad_norm": 5.059213161468506,
+ "learning_rate": 1.1715426593437096e-05,
+ "loss": 0.2063,
+ "step": 16947
+ },
+ {
+ "epoch": 46.17983651226158,
+ "grad_norm": 2.2388787269592285,
+ "learning_rate": 1.1714557173611785e-05,
+ "loss": 0.1182,
+ "step": 16948
+ },
+ {
+ "epoch": 46.182561307901906,
+ "grad_norm": 2.690915107727051,
+ "learning_rate": 1.1713687740433557e-05,
+ "loss": 0.0436,
+ "step": 16949
+ },
+ {
+ "epoch": 46.18528610354223,
+ "grad_norm": 3.908388376235962,
+ "learning_rate": 1.1712818293909182e-05,
+ "loss": 0.1569,
+ "step": 16950
+ },
+ {
+ "epoch": 46.18801089918256,
+ "grad_norm": 2.549912214279175,
+ "learning_rate": 1.1711948834045437e-05,
+ "loss": 0.0877,
+ "step": 16951
+ },
+ {
+ "epoch": 46.19073569482289,
+ "grad_norm": 2.833170175552368,
+ "learning_rate": 1.1711079360849086e-05,
+ "loss": 0.0484,
+ "step": 16952
+ },
+ {
+ "epoch": 46.19346049046322,
+ "grad_norm": 2.4011056423187256,
+ "learning_rate": 1.1710209874326905e-05,
+ "loss": 0.1225,
+ "step": 16953
+ },
+ {
+ "epoch": 46.19618528610354,
+ "grad_norm": 2.646827459335327,
+ "learning_rate": 1.1709340374485667e-05,
+ "loss": 0.0685,
+ "step": 16954
+ },
+ {
+ "epoch": 46.19891008174387,
+ "grad_norm": 3.986689567565918,
+ "learning_rate": 1.170847086133214e-05,
+ "loss": 0.1127,
+ "step": 16955
+ },
+ {
+ "epoch": 46.201634877384194,
+ "grad_norm": 2.614210367202759,
+ "learning_rate": 1.1707601334873097e-05,
+ "loss": 0.0682,
+ "step": 16956
+ },
+ {
+ "epoch": 46.20435967302452,
+ "grad_norm": 2.5095770359039307,
+ "learning_rate": 1.1706731795115309e-05,
+ "loss": 0.1918,
+ "step": 16957
+ },
+ {
+ "epoch": 46.20708446866485,
+ "grad_norm": 3.5267069339752197,
+ "learning_rate": 1.1705862242065549e-05,
+ "loss": 0.202,
+ "step": 16958
+ },
+ {
+ "epoch": 46.20980926430518,
+ "grad_norm": 2.4048514366149902,
+ "learning_rate": 1.1704992675730588e-05,
+ "loss": 0.1427,
+ "step": 16959
+ },
+ {
+ "epoch": 46.212534059945504,
+ "grad_norm": 2.36020827293396,
+ "learning_rate": 1.17041230961172e-05,
+ "loss": 0.0846,
+ "step": 16960
+ },
+ {
+ "epoch": 46.21525885558583,
+ "grad_norm": 2.805004358291626,
+ "learning_rate": 1.1703253503232152e-05,
+ "loss": 0.1145,
+ "step": 16961
+ },
+ {
+ "epoch": 46.217983651226156,
+ "grad_norm": 7.060424327850342,
+ "learning_rate": 1.1702383897082229e-05,
+ "loss": 0.0598,
+ "step": 16962
+ },
+ {
+ "epoch": 46.22070844686648,
+ "grad_norm": 9.050069808959961,
+ "learning_rate": 1.1701514277674186e-05,
+ "loss": 0.3747,
+ "step": 16963
+ },
+ {
+ "epoch": 46.223433242506815,
+ "grad_norm": 2.4657697677612305,
+ "learning_rate": 1.1700644645014812e-05,
+ "loss": 0.0576,
+ "step": 16964
+ },
+ {
+ "epoch": 46.22615803814714,
+ "grad_norm": 2.0219619274139404,
+ "learning_rate": 1.1699774999110867e-05,
+ "loss": 0.1178,
+ "step": 16965
+ },
+ {
+ "epoch": 46.228882833787466,
+ "grad_norm": 4.540931224822998,
+ "learning_rate": 1.1698905339969133e-05,
+ "loss": 0.2518,
+ "step": 16966
+ },
+ {
+ "epoch": 46.23160762942779,
+ "grad_norm": 2.713365077972412,
+ "learning_rate": 1.1698035667596378e-05,
+ "loss": 0.0587,
+ "step": 16967
+ },
+ {
+ "epoch": 46.23433242506812,
+ "grad_norm": 3.2522082328796387,
+ "learning_rate": 1.1697165981999375e-05,
+ "loss": 0.1942,
+ "step": 16968
+ },
+ {
+ "epoch": 46.237057220708444,
+ "grad_norm": 2.700408458709717,
+ "learning_rate": 1.1696296283184896e-05,
+ "loss": 0.1505,
+ "step": 16969
+ },
+ {
+ "epoch": 46.23978201634878,
+ "grad_norm": 2.8211727142333984,
+ "learning_rate": 1.1695426571159718e-05,
+ "loss": 0.1148,
+ "step": 16970
+ },
+ {
+ "epoch": 46.2425068119891,
+ "grad_norm": 3.7321205139160156,
+ "learning_rate": 1.1694556845930611e-05,
+ "loss": 0.1835,
+ "step": 16971
+ },
+ {
+ "epoch": 46.24523160762943,
+ "grad_norm": 2.708782196044922,
+ "learning_rate": 1.1693687107504352e-05,
+ "loss": 0.0735,
+ "step": 16972
+ },
+ {
+ "epoch": 46.247956403269754,
+ "grad_norm": 2.222942352294922,
+ "learning_rate": 1.169281735588771e-05,
+ "loss": 0.0679,
+ "step": 16973
+ },
+ {
+ "epoch": 46.25068119891008,
+ "grad_norm": 2.2312543392181396,
+ "learning_rate": 1.1691947591087464e-05,
+ "loss": 0.1298,
+ "step": 16974
+ },
+ {
+ "epoch": 46.253405994550405,
+ "grad_norm": 2.7317166328430176,
+ "learning_rate": 1.1691077813110383e-05,
+ "loss": 0.0837,
+ "step": 16975
+ },
+ {
+ "epoch": 46.25613079019074,
+ "grad_norm": 3.0678586959838867,
+ "learning_rate": 1.1690208021963243e-05,
+ "loss": 0.304,
+ "step": 16976
+ },
+ {
+ "epoch": 46.258855585831064,
+ "grad_norm": 3.0467426776885986,
+ "learning_rate": 1.1689338217652818e-05,
+ "loss": 0.0618,
+ "step": 16977
+ },
+ {
+ "epoch": 46.26158038147139,
+ "grad_norm": 4.324429988861084,
+ "learning_rate": 1.1688468400185877e-05,
+ "loss": 0.272,
+ "step": 16978
+ },
+ {
+ "epoch": 46.264305177111716,
+ "grad_norm": 2.813305139541626,
+ "learning_rate": 1.16875985695692e-05,
+ "loss": 0.0441,
+ "step": 16979
+ },
+ {
+ "epoch": 46.26702997275204,
+ "grad_norm": 5.086302757263184,
+ "learning_rate": 1.1686728725809564e-05,
+ "loss": 0.0694,
+ "step": 16980
+ },
+ {
+ "epoch": 46.26975476839237,
+ "grad_norm": 2.595808267593384,
+ "learning_rate": 1.1685858868913736e-05,
+ "loss": 0.1424,
+ "step": 16981
+ },
+ {
+ "epoch": 46.2724795640327,
+ "grad_norm": 3.154871940612793,
+ "learning_rate": 1.1684988998888494e-05,
+ "loss": 0.1726,
+ "step": 16982
+ },
+ {
+ "epoch": 46.275204359673026,
+ "grad_norm": 3.2945501804351807,
+ "learning_rate": 1.168411911574061e-05,
+ "loss": 0.1655,
+ "step": 16983
+ },
+ {
+ "epoch": 46.27792915531335,
+ "grad_norm": 2.6695265769958496,
+ "learning_rate": 1.1683249219476864e-05,
+ "loss": 0.0829,
+ "step": 16984
+ },
+ {
+ "epoch": 46.28065395095368,
+ "grad_norm": 2.858109712600708,
+ "learning_rate": 1.1682379310104023e-05,
+ "loss": 0.0617,
+ "step": 16985
+ },
+ {
+ "epoch": 46.283378746594,
+ "grad_norm": 3.068746566772461,
+ "learning_rate": 1.1681509387628871e-05,
+ "loss": 0.0986,
+ "step": 16986
+ },
+ {
+ "epoch": 46.28610354223433,
+ "grad_norm": 2.3143153190612793,
+ "learning_rate": 1.1680639452058172e-05,
+ "loss": 0.1611,
+ "step": 16987
+ },
+ {
+ "epoch": 46.28882833787466,
+ "grad_norm": 3.3153817653656006,
+ "learning_rate": 1.1679769503398712e-05,
+ "loss": 0.1546,
+ "step": 16988
+ },
+ {
+ "epoch": 46.29155313351499,
+ "grad_norm": 2.6133828163146973,
+ "learning_rate": 1.1678899541657258e-05,
+ "loss": 0.1531,
+ "step": 16989
+ },
+ {
+ "epoch": 46.294277929155314,
+ "grad_norm": 1.9607720375061035,
+ "learning_rate": 1.1678029566840593e-05,
+ "loss": 0.1425,
+ "step": 16990
+ },
+ {
+ "epoch": 46.29700272479564,
+ "grad_norm": 3.381265163421631,
+ "learning_rate": 1.1677159578955481e-05,
+ "loss": 0.0865,
+ "step": 16991
+ },
+ {
+ "epoch": 46.299727520435965,
+ "grad_norm": 2.9767282009124756,
+ "learning_rate": 1.167628957800871e-05,
+ "loss": 0.2779,
+ "step": 16992
+ },
+ {
+ "epoch": 46.30245231607629,
+ "grad_norm": 2.8521618843078613,
+ "learning_rate": 1.1675419564007046e-05,
+ "loss": 0.0586,
+ "step": 16993
+ },
+ {
+ "epoch": 46.305177111716624,
+ "grad_norm": 3.492716073989868,
+ "learning_rate": 1.167454953695727e-05,
+ "loss": 0.0945,
+ "step": 16994
+ },
+ {
+ "epoch": 46.30790190735695,
+ "grad_norm": 3.1201634407043457,
+ "learning_rate": 1.1673679496866155e-05,
+ "loss": 0.0534,
+ "step": 16995
+ },
+ {
+ "epoch": 46.310626702997276,
+ "grad_norm": 3.1314995288848877,
+ "learning_rate": 1.167280944374048e-05,
+ "loss": 0.0739,
+ "step": 16996
+ },
+ {
+ "epoch": 46.3133514986376,
+ "grad_norm": 3.40720272064209,
+ "learning_rate": 1.1671939377587018e-05,
+ "loss": 0.0803,
+ "step": 16997
+ },
+ {
+ "epoch": 46.31607629427793,
+ "grad_norm": 3.089838743209839,
+ "learning_rate": 1.1671069298412548e-05,
+ "loss": 0.081,
+ "step": 16998
+ },
+ {
+ "epoch": 46.31880108991825,
+ "grad_norm": 1.9788936376571655,
+ "learning_rate": 1.167019920622384e-05,
+ "loss": 0.0921,
+ "step": 16999
+ },
+ {
+ "epoch": 46.321525885558586,
+ "grad_norm": 3.3233752250671387,
+ "learning_rate": 1.1669329101027677e-05,
+ "loss": 0.0756,
+ "step": 17000
+ },
+ {
+ "epoch": 46.32425068119891,
+ "grad_norm": 3.3043479919433594,
+ "learning_rate": 1.166845898283083e-05,
+ "loss": 0.1242,
+ "step": 17001
+ },
+ {
+ "epoch": 46.32697547683924,
+ "grad_norm": 2.6012275218963623,
+ "learning_rate": 1.1667588851640084e-05,
+ "loss": 0.054,
+ "step": 17002
+ },
+ {
+ "epoch": 46.32970027247956,
+ "grad_norm": 2.6964774131774902,
+ "learning_rate": 1.1666718707462203e-05,
+ "loss": 0.0956,
+ "step": 17003
+ },
+ {
+ "epoch": 46.33242506811989,
+ "grad_norm": 2.525609254837036,
+ "learning_rate": 1.1665848550303972e-05,
+ "loss": 0.1447,
+ "step": 17004
+ },
+ {
+ "epoch": 46.335149863760215,
+ "grad_norm": 2.534562587738037,
+ "learning_rate": 1.1664978380172167e-05,
+ "loss": 0.0427,
+ "step": 17005
+ },
+ {
+ "epoch": 46.33787465940055,
+ "grad_norm": 3.1933963298797607,
+ "learning_rate": 1.1664108197073566e-05,
+ "loss": 0.0739,
+ "step": 17006
+ },
+ {
+ "epoch": 46.34059945504087,
+ "grad_norm": 2.3929355144500732,
+ "learning_rate": 1.166323800101494e-05,
+ "loss": 0.0465,
+ "step": 17007
+ },
+ {
+ "epoch": 46.3433242506812,
+ "grad_norm": 2.625593900680542,
+ "learning_rate": 1.1662367792003075e-05,
+ "loss": 0.072,
+ "step": 17008
+ },
+ {
+ "epoch": 46.346049046321525,
+ "grad_norm": 3.211843252182007,
+ "learning_rate": 1.1661497570044737e-05,
+ "loss": 0.1004,
+ "step": 17009
+ },
+ {
+ "epoch": 46.34877384196185,
+ "grad_norm": 2.9420738220214844,
+ "learning_rate": 1.1660627335146714e-05,
+ "loss": 0.197,
+ "step": 17010
+ },
+ {
+ "epoch": 46.35149863760218,
+ "grad_norm": 2.914764165878296,
+ "learning_rate": 1.1659757087315777e-05,
+ "loss": 0.1094,
+ "step": 17011
+ },
+ {
+ "epoch": 46.35422343324251,
+ "grad_norm": 3.191218376159668,
+ "learning_rate": 1.1658886826558709e-05,
+ "loss": 0.1593,
+ "step": 17012
+ },
+ {
+ "epoch": 46.356948228882835,
+ "grad_norm": 2.3588168621063232,
+ "learning_rate": 1.1658016552882279e-05,
+ "loss": 0.0863,
+ "step": 17013
+ },
+ {
+ "epoch": 46.35967302452316,
+ "grad_norm": 1.8584392070770264,
+ "learning_rate": 1.1657146266293269e-05,
+ "loss": 0.0423,
+ "step": 17014
+ },
+ {
+ "epoch": 46.36239782016349,
+ "grad_norm": 2.7332539558410645,
+ "learning_rate": 1.1656275966798458e-05,
+ "loss": 0.0704,
+ "step": 17015
+ },
+ {
+ "epoch": 46.36512261580381,
+ "grad_norm": 2.92928147315979,
+ "learning_rate": 1.1655405654404625e-05,
+ "loss": 0.2756,
+ "step": 17016
+ },
+ {
+ "epoch": 46.36784741144414,
+ "grad_norm": 3.399219274520874,
+ "learning_rate": 1.1654535329118545e-05,
+ "loss": 0.0609,
+ "step": 17017
+ },
+ {
+ "epoch": 46.37057220708447,
+ "grad_norm": 2.9586572647094727,
+ "learning_rate": 1.1653664990947e-05,
+ "loss": 0.0808,
+ "step": 17018
+ },
+ {
+ "epoch": 46.3732970027248,
+ "grad_norm": 2.3984220027923584,
+ "learning_rate": 1.165279463989676e-05,
+ "loss": 0.1171,
+ "step": 17019
+ },
+ {
+ "epoch": 46.37602179836512,
+ "grad_norm": 2.811058282852173,
+ "learning_rate": 1.1651924275974613e-05,
+ "loss": 0.0759,
+ "step": 17020
+ },
+ {
+ "epoch": 46.37874659400545,
+ "grad_norm": 3.08483624458313,
+ "learning_rate": 1.165105389918733e-05,
+ "loss": 0.0653,
+ "step": 17021
+ },
+ {
+ "epoch": 46.381471389645775,
+ "grad_norm": 3.0451033115386963,
+ "learning_rate": 1.1650183509541693e-05,
+ "loss": 0.0646,
+ "step": 17022
+ },
+ {
+ "epoch": 46.3841961852861,
+ "grad_norm": 3.171231746673584,
+ "learning_rate": 1.1649313107044483e-05,
+ "loss": 0.0538,
+ "step": 17023
+ },
+ {
+ "epoch": 46.38692098092643,
+ "grad_norm": 2.5242016315460205,
+ "learning_rate": 1.164844269170247e-05,
+ "loss": 0.0706,
+ "step": 17024
+ },
+ {
+ "epoch": 46.38964577656676,
+ "grad_norm": 3.052356243133545,
+ "learning_rate": 1.1647572263522445e-05,
+ "loss": 0.076,
+ "step": 17025
+ },
+ {
+ "epoch": 46.392370572207085,
+ "grad_norm": 3.015949010848999,
+ "learning_rate": 1.1646701822511173e-05,
+ "loss": 0.0793,
+ "step": 17026
+ },
+ {
+ "epoch": 46.39509536784741,
+ "grad_norm": 2.9493660926818848,
+ "learning_rate": 1.1645831368675445e-05,
+ "loss": 0.0688,
+ "step": 17027
+ },
+ {
+ "epoch": 46.39782016348774,
+ "grad_norm": 3.517650842666626,
+ "learning_rate": 1.1644960902022033e-05,
+ "loss": 0.0727,
+ "step": 17028
+ },
+ {
+ "epoch": 46.40054495912806,
+ "grad_norm": 2.0208559036254883,
+ "learning_rate": 1.1644090422557719e-05,
+ "loss": 0.1848,
+ "step": 17029
+ },
+ {
+ "epoch": 46.403269754768395,
+ "grad_norm": 3.8644092082977295,
+ "learning_rate": 1.1643219930289282e-05,
+ "loss": 0.0702,
+ "step": 17030
+ },
+ {
+ "epoch": 46.40599455040872,
+ "grad_norm": 2.9504754543304443,
+ "learning_rate": 1.1642349425223502e-05,
+ "loss": 0.0907,
+ "step": 17031
+ },
+ {
+ "epoch": 46.40871934604905,
+ "grad_norm": 2.4011847972869873,
+ "learning_rate": 1.1641478907367157e-05,
+ "loss": 0.0629,
+ "step": 17032
+ },
+ {
+ "epoch": 46.41144414168937,
+ "grad_norm": 2.567525625228882,
+ "learning_rate": 1.1640608376727028e-05,
+ "loss": 0.0472,
+ "step": 17033
+ },
+ {
+ "epoch": 46.4141689373297,
+ "grad_norm": 2.723294258117676,
+ "learning_rate": 1.163973783330989e-05,
+ "loss": 0.0557,
+ "step": 17034
+ },
+ {
+ "epoch": 46.416893732970024,
+ "grad_norm": 2.099271535873413,
+ "learning_rate": 1.1638867277122532e-05,
+ "loss": 0.0707,
+ "step": 17035
+ },
+ {
+ "epoch": 46.41961852861036,
+ "grad_norm": 3.1325042247772217,
+ "learning_rate": 1.1637996708171725e-05,
+ "loss": 0.067,
+ "step": 17036
+ },
+ {
+ "epoch": 46.42234332425068,
+ "grad_norm": 3.4447317123413086,
+ "learning_rate": 1.163712612646425e-05,
+ "loss": 0.1765,
+ "step": 17037
+ },
+ {
+ "epoch": 46.42506811989101,
+ "grad_norm": 3.8644297122955322,
+ "learning_rate": 1.1636255532006892e-05,
+ "loss": 0.0587,
+ "step": 17038
+ },
+ {
+ "epoch": 46.427792915531334,
+ "grad_norm": 3.410985231399536,
+ "learning_rate": 1.163538492480643e-05,
+ "loss": 0.0645,
+ "step": 17039
+ },
+ {
+ "epoch": 46.43051771117166,
+ "grad_norm": 2.6761794090270996,
+ "learning_rate": 1.1634514304869641e-05,
+ "loss": 0.1301,
+ "step": 17040
+ },
+ {
+ "epoch": 46.433242506811986,
+ "grad_norm": 2.740987777709961,
+ "learning_rate": 1.163364367220331e-05,
+ "loss": 0.0775,
+ "step": 17041
+ },
+ {
+ "epoch": 46.43596730245232,
+ "grad_norm": 5.943276405334473,
+ "learning_rate": 1.1632773026814213e-05,
+ "loss": 0.1119,
+ "step": 17042
+ },
+ {
+ "epoch": 46.438692098092645,
+ "grad_norm": 2.899068832397461,
+ "learning_rate": 1.1631902368709132e-05,
+ "loss": 0.0594,
+ "step": 17043
+ },
+ {
+ "epoch": 46.44141689373297,
+ "grad_norm": 3.699373960494995,
+ "learning_rate": 1.1631031697894846e-05,
+ "loss": 0.1008,
+ "step": 17044
+ },
+ {
+ "epoch": 46.444141689373296,
+ "grad_norm": 2.897064685821533,
+ "learning_rate": 1.1630161014378141e-05,
+ "loss": 0.0638,
+ "step": 17045
+ },
+ {
+ "epoch": 46.44686648501362,
+ "grad_norm": 2.151148796081543,
+ "learning_rate": 1.1629290318165792e-05,
+ "loss": 0.0582,
+ "step": 17046
+ },
+ {
+ "epoch": 46.44959128065395,
+ "grad_norm": 4.242058753967285,
+ "learning_rate": 1.1628419609264582e-05,
+ "loss": 0.1418,
+ "step": 17047
+ },
+ {
+ "epoch": 46.45231607629428,
+ "grad_norm": 2.3225595951080322,
+ "learning_rate": 1.1627548887681292e-05,
+ "loss": 0.043,
+ "step": 17048
+ },
+ {
+ "epoch": 46.45504087193461,
+ "grad_norm": 3.361703872680664,
+ "learning_rate": 1.1626678153422708e-05,
+ "loss": 0.1057,
+ "step": 17049
+ },
+ {
+ "epoch": 46.45776566757493,
+ "grad_norm": 2.269904375076294,
+ "learning_rate": 1.1625807406495605e-05,
+ "loss": 0.0499,
+ "step": 17050
+ },
+ {
+ "epoch": 46.46049046321526,
+ "grad_norm": 2.244356393814087,
+ "learning_rate": 1.1624936646906766e-05,
+ "loss": 0.0364,
+ "step": 17051
+ },
+ {
+ "epoch": 46.463215258855584,
+ "grad_norm": 1.7891228199005127,
+ "learning_rate": 1.1624065874662971e-05,
+ "loss": 0.0265,
+ "step": 17052
+ },
+ {
+ "epoch": 46.46594005449591,
+ "grad_norm": 3.4184882640838623,
+ "learning_rate": 1.1623195089771005e-05,
+ "loss": 0.1702,
+ "step": 17053
+ },
+ {
+ "epoch": 46.46866485013624,
+ "grad_norm": 2.914834976196289,
+ "learning_rate": 1.1622324292237645e-05,
+ "loss": 0.0788,
+ "step": 17054
+ },
+ {
+ "epoch": 46.47138964577657,
+ "grad_norm": 3.0574889183044434,
+ "learning_rate": 1.162145348206968e-05,
+ "loss": 0.0832,
+ "step": 17055
+ },
+ {
+ "epoch": 46.474114441416894,
+ "grad_norm": 2.7681894302368164,
+ "learning_rate": 1.1620582659273882e-05,
+ "loss": 0.0633,
+ "step": 17056
+ },
+ {
+ "epoch": 46.47683923705722,
+ "grad_norm": 2.602687120437622,
+ "learning_rate": 1.1619711823857045e-05,
+ "loss": 0.0697,
+ "step": 17057
+ },
+ {
+ "epoch": 46.479564032697546,
+ "grad_norm": 3.0172529220581055,
+ "learning_rate": 1.161884097582594e-05,
+ "loss": 0.0815,
+ "step": 17058
+ },
+ {
+ "epoch": 46.48228882833787,
+ "grad_norm": 3.328904628753662,
+ "learning_rate": 1.1617970115187356e-05,
+ "loss": 0.0565,
+ "step": 17059
+ },
+ {
+ "epoch": 46.485013623978205,
+ "grad_norm": 2.3259708881378174,
+ "learning_rate": 1.1617099241948072e-05,
+ "loss": 0.0622,
+ "step": 17060
+ },
+ {
+ "epoch": 46.48773841961853,
+ "grad_norm": 2.352989673614502,
+ "learning_rate": 1.1616228356114871e-05,
+ "loss": 0.0549,
+ "step": 17061
+ },
+ {
+ "epoch": 46.490463215258856,
+ "grad_norm": 2.9136781692504883,
+ "learning_rate": 1.1615357457694534e-05,
+ "loss": 0.0979,
+ "step": 17062
+ },
+ {
+ "epoch": 46.49318801089918,
+ "grad_norm": 2.8358445167541504,
+ "learning_rate": 1.1614486546693848e-05,
+ "loss": 0.0877,
+ "step": 17063
+ },
+ {
+ "epoch": 46.49591280653951,
+ "grad_norm": 3.5308380126953125,
+ "learning_rate": 1.1613615623119587e-05,
+ "loss": 0.0824,
+ "step": 17064
+ },
+ {
+ "epoch": 46.49863760217983,
+ "grad_norm": 3.1050777435302734,
+ "learning_rate": 1.1612744686978545e-05,
+ "loss": 0.1562,
+ "step": 17065
+ },
+ {
+ "epoch": 46.50136239782017,
+ "grad_norm": 2.8586416244506836,
+ "learning_rate": 1.1611873738277499e-05,
+ "loss": 0.0697,
+ "step": 17066
+ },
+ {
+ "epoch": 46.50408719346049,
+ "grad_norm": 3.945406198501587,
+ "learning_rate": 1.1611002777023233e-05,
+ "loss": 0.1775,
+ "step": 17067
+ },
+ {
+ "epoch": 46.50681198910082,
+ "grad_norm": 4.915589332580566,
+ "learning_rate": 1.1610131803222527e-05,
+ "loss": 0.1073,
+ "step": 17068
+ },
+ {
+ "epoch": 46.509536784741144,
+ "grad_norm": 2.562346935272217,
+ "learning_rate": 1.1609260816882167e-05,
+ "loss": 0.0638,
+ "step": 17069
+ },
+ {
+ "epoch": 46.51226158038147,
+ "grad_norm": 2.897296190261841,
+ "learning_rate": 1.1608389818008933e-05,
+ "loss": 0.1267,
+ "step": 17070
+ },
+ {
+ "epoch": 46.514986376021795,
+ "grad_norm": 3.8431193828582764,
+ "learning_rate": 1.1607518806609614e-05,
+ "loss": 0.0775,
+ "step": 17071
+ },
+ {
+ "epoch": 46.51771117166213,
+ "grad_norm": 2.518490791320801,
+ "learning_rate": 1.160664778269099e-05,
+ "loss": 0.1781,
+ "step": 17072
+ },
+ {
+ "epoch": 46.520435967302454,
+ "grad_norm": 3.233832359313965,
+ "learning_rate": 1.1605776746259844e-05,
+ "loss": 0.1659,
+ "step": 17073
+ },
+ {
+ "epoch": 46.52316076294278,
+ "grad_norm": 3.3392019271850586,
+ "learning_rate": 1.1604905697322959e-05,
+ "loss": 0.1102,
+ "step": 17074
+ },
+ {
+ "epoch": 46.525885558583106,
+ "grad_norm": 2.321369171142578,
+ "learning_rate": 1.1604034635887123e-05,
+ "loss": 0.0739,
+ "step": 17075
+ },
+ {
+ "epoch": 46.52861035422343,
+ "grad_norm": 2.8480491638183594,
+ "learning_rate": 1.1603163561959114e-05,
+ "loss": 0.1827,
+ "step": 17076
+ },
+ {
+ "epoch": 46.53133514986376,
+ "grad_norm": 3.4583568572998047,
+ "learning_rate": 1.1602292475545722e-05,
+ "loss": 0.0722,
+ "step": 17077
+ },
+ {
+ "epoch": 46.53405994550409,
+ "grad_norm": 2.518859386444092,
+ "learning_rate": 1.1601421376653725e-05,
+ "loss": 0.1384,
+ "step": 17078
+ },
+ {
+ "epoch": 46.536784741144416,
+ "grad_norm": 2.8729310035705566,
+ "learning_rate": 1.1600550265289913e-05,
+ "loss": 0.1156,
+ "step": 17079
+ },
+ {
+ "epoch": 46.53950953678474,
+ "grad_norm": 1.8541958332061768,
+ "learning_rate": 1.1599679141461064e-05,
+ "loss": 0.0397,
+ "step": 17080
+ },
+ {
+ "epoch": 46.54223433242507,
+ "grad_norm": 2.3610405921936035,
+ "learning_rate": 1.1598808005173967e-05,
+ "loss": 0.087,
+ "step": 17081
+ },
+ {
+ "epoch": 46.54495912806539,
+ "grad_norm": 3.2910873889923096,
+ "learning_rate": 1.15979368564354e-05,
+ "loss": 0.0893,
+ "step": 17082
+ },
+ {
+ "epoch": 46.54768392370572,
+ "grad_norm": 2.0027120113372803,
+ "learning_rate": 1.1597065695252158e-05,
+ "loss": 0.0385,
+ "step": 17083
+ },
+ {
+ "epoch": 46.55040871934605,
+ "grad_norm": 2.9734742641448975,
+ "learning_rate": 1.1596194521631018e-05,
+ "loss": 0.1025,
+ "step": 17084
+ },
+ {
+ "epoch": 46.55313351498638,
+ "grad_norm": 3.4642248153686523,
+ "learning_rate": 1.159532333557877e-05,
+ "loss": 0.0573,
+ "step": 17085
+ },
+ {
+ "epoch": 46.555858310626704,
+ "grad_norm": 3.357607126235962,
+ "learning_rate": 1.159445213710219e-05,
+ "loss": 0.0776,
+ "step": 17086
+ },
+ {
+ "epoch": 46.55858310626703,
+ "grad_norm": 2.669051170349121,
+ "learning_rate": 1.159358092620807e-05,
+ "loss": 0.041,
+ "step": 17087
+ },
+ {
+ "epoch": 46.561307901907355,
+ "grad_norm": 2.4038541316986084,
+ "learning_rate": 1.1592709702903193e-05,
+ "loss": 0.0478,
+ "step": 17088
+ },
+ {
+ "epoch": 46.56403269754768,
+ "grad_norm": 3.177166700363159,
+ "learning_rate": 1.1591838467194345e-05,
+ "loss": 0.147,
+ "step": 17089
+ },
+ {
+ "epoch": 46.566757493188014,
+ "grad_norm": 2.4460501670837402,
+ "learning_rate": 1.1590967219088305e-05,
+ "loss": 0.1984,
+ "step": 17090
+ },
+ {
+ "epoch": 46.56948228882834,
+ "grad_norm": 3.5197012424468994,
+ "learning_rate": 1.159009595859187e-05,
+ "loss": 0.0944,
+ "step": 17091
+ },
+ {
+ "epoch": 46.572207084468666,
+ "grad_norm": 2.6984775066375732,
+ "learning_rate": 1.1589224685711816e-05,
+ "loss": 0.0711,
+ "step": 17092
+ },
+ {
+ "epoch": 46.57493188010899,
+ "grad_norm": 3.079129695892334,
+ "learning_rate": 1.1588353400454933e-05,
+ "loss": 0.0706,
+ "step": 17093
+ },
+ {
+ "epoch": 46.57765667574932,
+ "grad_norm": 2.5527961254119873,
+ "learning_rate": 1.1587482102828002e-05,
+ "loss": 0.151,
+ "step": 17094
+ },
+ {
+ "epoch": 46.58038147138964,
+ "grad_norm": 3.098464012145996,
+ "learning_rate": 1.1586610792837813e-05,
+ "loss": 0.2205,
+ "step": 17095
+ },
+ {
+ "epoch": 46.583106267029976,
+ "grad_norm": 2.933976888656616,
+ "learning_rate": 1.158573947049115e-05,
+ "loss": 0.0758,
+ "step": 17096
+ },
+ {
+ "epoch": 46.5858310626703,
+ "grad_norm": 3.858299970626831,
+ "learning_rate": 1.1584868135794801e-05,
+ "loss": 0.131,
+ "step": 17097
+ },
+ {
+ "epoch": 46.58855585831063,
+ "grad_norm": 3.1826562881469727,
+ "learning_rate": 1.1583996788755548e-05,
+ "loss": 0.0649,
+ "step": 17098
+ },
+ {
+ "epoch": 46.59128065395095,
+ "grad_norm": 2.777003049850464,
+ "learning_rate": 1.1583125429380177e-05,
+ "loss": 0.06,
+ "step": 17099
+ },
+ {
+ "epoch": 46.59400544959128,
+ "grad_norm": 2.9832522869110107,
+ "learning_rate": 1.1582254057675481e-05,
+ "loss": 0.0608,
+ "step": 17100
+ },
+ {
+ "epoch": 46.596730245231605,
+ "grad_norm": 2.3055171966552734,
+ "learning_rate": 1.1581382673648238e-05,
+ "loss": 0.1087,
+ "step": 17101
+ },
+ {
+ "epoch": 46.59945504087194,
+ "grad_norm": 2.2541098594665527,
+ "learning_rate": 1.1580511277305238e-05,
+ "loss": 0.0547,
+ "step": 17102
+ },
+ {
+ "epoch": 46.60217983651226,
+ "grad_norm": 2.811206579208374,
+ "learning_rate": 1.1579639868653267e-05,
+ "loss": 0.0735,
+ "step": 17103
+ },
+ {
+ "epoch": 46.60490463215259,
+ "grad_norm": 3.273664951324463,
+ "learning_rate": 1.1578768447699111e-05,
+ "loss": 0.1071,
+ "step": 17104
+ },
+ {
+ "epoch": 46.607629427792915,
+ "grad_norm": 2.4480061531066895,
+ "learning_rate": 1.1577897014449562e-05,
+ "loss": 0.0846,
+ "step": 17105
+ },
+ {
+ "epoch": 46.61035422343324,
+ "grad_norm": 2.4391257762908936,
+ "learning_rate": 1.1577025568911395e-05,
+ "loss": 0.0807,
+ "step": 17106
+ },
+ {
+ "epoch": 46.61307901907357,
+ "grad_norm": 2.697392225265503,
+ "learning_rate": 1.1576154111091407e-05,
+ "loss": 0.1197,
+ "step": 17107
+ },
+ {
+ "epoch": 46.6158038147139,
+ "grad_norm": 4.192628383636475,
+ "learning_rate": 1.157528264099638e-05,
+ "loss": 0.1097,
+ "step": 17108
+ },
+ {
+ "epoch": 46.618528610354225,
+ "grad_norm": 3.8190343379974365,
+ "learning_rate": 1.1574411158633105e-05,
+ "loss": 0.1706,
+ "step": 17109
+ },
+ {
+ "epoch": 46.62125340599455,
+ "grad_norm": 2.769253969192505,
+ "learning_rate": 1.1573539664008367e-05,
+ "loss": 0.0858,
+ "step": 17110
+ },
+ {
+ "epoch": 46.62397820163488,
+ "grad_norm": 2.9855949878692627,
+ "learning_rate": 1.157266815712895e-05,
+ "loss": 0.036,
+ "step": 17111
+ },
+ {
+ "epoch": 46.6267029972752,
+ "grad_norm": 3.584730386734009,
+ "learning_rate": 1.1571796638001644e-05,
+ "loss": 0.0596,
+ "step": 17112
+ },
+ {
+ "epoch": 46.62942779291553,
+ "grad_norm": 3.353942394256592,
+ "learning_rate": 1.1570925106633237e-05,
+ "loss": 0.1018,
+ "step": 17113
+ },
+ {
+ "epoch": 46.63215258855586,
+ "grad_norm": 2.9191997051239014,
+ "learning_rate": 1.1570053563030516e-05,
+ "loss": 0.0573,
+ "step": 17114
+ },
+ {
+ "epoch": 46.63487738419619,
+ "grad_norm": 2.569658041000366,
+ "learning_rate": 1.156918200720027e-05,
+ "loss": 0.0659,
+ "step": 17115
+ },
+ {
+ "epoch": 46.63760217983651,
+ "grad_norm": 2.942351818084717,
+ "learning_rate": 1.1568310439149281e-05,
+ "loss": 0.0744,
+ "step": 17116
+ },
+ {
+ "epoch": 46.64032697547684,
+ "grad_norm": 2.77483868598938,
+ "learning_rate": 1.1567438858884344e-05,
+ "loss": 0.1168,
+ "step": 17117
+ },
+ {
+ "epoch": 46.643051771117165,
+ "grad_norm": 3.512773036956787,
+ "learning_rate": 1.1566567266412243e-05,
+ "loss": 0.133,
+ "step": 17118
+ },
+ {
+ "epoch": 46.64577656675749,
+ "grad_norm": 2.5987348556518555,
+ "learning_rate": 1.1565695661739767e-05,
+ "loss": 0.1007,
+ "step": 17119
+ },
+ {
+ "epoch": 46.64850136239782,
+ "grad_norm": 2.3028147220611572,
+ "learning_rate": 1.1564824044873704e-05,
+ "loss": 0.0406,
+ "step": 17120
+ },
+ {
+ "epoch": 46.65122615803815,
+ "grad_norm": 2.657203435897827,
+ "learning_rate": 1.156395241582084e-05,
+ "loss": 0.08,
+ "step": 17121
+ },
+ {
+ "epoch": 46.653950953678475,
+ "grad_norm": 4.464833736419678,
+ "learning_rate": 1.1563080774587967e-05,
+ "loss": 0.0597,
+ "step": 17122
+ },
+ {
+ "epoch": 46.6566757493188,
+ "grad_norm": 3.6179921627044678,
+ "learning_rate": 1.1562209121181869e-05,
+ "loss": 0.0672,
+ "step": 17123
+ },
+ {
+ "epoch": 46.65940054495913,
+ "grad_norm": 2.3319027423858643,
+ "learning_rate": 1.1561337455609336e-05,
+ "loss": 0.066,
+ "step": 17124
+ },
+ {
+ "epoch": 46.66212534059945,
+ "grad_norm": 4.665820598602295,
+ "learning_rate": 1.1560465777877159e-05,
+ "loss": 0.1665,
+ "step": 17125
+ },
+ {
+ "epoch": 46.664850136239785,
+ "grad_norm": 2.3707425594329834,
+ "learning_rate": 1.1559594087992126e-05,
+ "loss": 0.078,
+ "step": 17126
+ },
+ {
+ "epoch": 46.66757493188011,
+ "grad_norm": 2.5511789321899414,
+ "learning_rate": 1.1558722385961022e-05,
+ "loss": 0.0468,
+ "step": 17127
+ },
+ {
+ "epoch": 46.67029972752044,
+ "grad_norm": 3.443943977355957,
+ "learning_rate": 1.1557850671790638e-05,
+ "loss": 0.2236,
+ "step": 17128
+ },
+ {
+ "epoch": 46.67302452316076,
+ "grad_norm": 2.2736120223999023,
+ "learning_rate": 1.1556978945487764e-05,
+ "loss": 0.0963,
+ "step": 17129
+ },
+ {
+ "epoch": 46.67574931880109,
+ "grad_norm": 2.168689727783203,
+ "learning_rate": 1.1556107207059191e-05,
+ "loss": 0.0666,
+ "step": 17130
+ },
+ {
+ "epoch": 46.678474114441414,
+ "grad_norm": 2.8231046199798584,
+ "learning_rate": 1.15552354565117e-05,
+ "loss": 0.1664,
+ "step": 17131
+ },
+ {
+ "epoch": 46.68119891008175,
+ "grad_norm": 2.745741128921509,
+ "learning_rate": 1.1554363693852088e-05,
+ "loss": 0.1818,
+ "step": 17132
+ },
+ {
+ "epoch": 46.68392370572207,
+ "grad_norm": 4.5182905197143555,
+ "learning_rate": 1.155349191908714e-05,
+ "loss": 0.0571,
+ "step": 17133
+ },
+ {
+ "epoch": 46.6866485013624,
+ "grad_norm": 2.6534016132354736,
+ "learning_rate": 1.1552620132223649e-05,
+ "loss": 0.0714,
+ "step": 17134
+ },
+ {
+ "epoch": 46.689373297002724,
+ "grad_norm": 2.6831116676330566,
+ "learning_rate": 1.1551748333268402e-05,
+ "loss": 0.0763,
+ "step": 17135
+ },
+ {
+ "epoch": 46.69209809264305,
+ "grad_norm": 4.321683883666992,
+ "learning_rate": 1.155087652222819e-05,
+ "loss": 0.2293,
+ "step": 17136
+ },
+ {
+ "epoch": 46.694822888283376,
+ "grad_norm": 3.1551806926727295,
+ "learning_rate": 1.1550004699109797e-05,
+ "loss": 0.0962,
+ "step": 17137
+ },
+ {
+ "epoch": 46.69754768392371,
+ "grad_norm": 3.762584686279297,
+ "learning_rate": 1.1549132863920023e-05,
+ "loss": 0.1063,
+ "step": 17138
+ },
+ {
+ "epoch": 46.700272479564035,
+ "grad_norm": 2.4231765270233154,
+ "learning_rate": 1.154826101666565e-05,
+ "loss": 0.0422,
+ "step": 17139
+ },
+ {
+ "epoch": 46.70299727520436,
+ "grad_norm": 2.252134084701538,
+ "learning_rate": 1.154738915735347e-05,
+ "loss": 0.1226,
+ "step": 17140
+ },
+ {
+ "epoch": 46.705722070844686,
+ "grad_norm": 3.1259894371032715,
+ "learning_rate": 1.1546517285990272e-05,
+ "loss": 0.0755,
+ "step": 17141
+ },
+ {
+ "epoch": 46.70844686648501,
+ "grad_norm": 4.502803325653076,
+ "learning_rate": 1.1545645402582845e-05,
+ "loss": 0.1483,
+ "step": 17142
+ },
+ {
+ "epoch": 46.71117166212534,
+ "grad_norm": 3.398407220840454,
+ "learning_rate": 1.1544773507137985e-05,
+ "loss": 0.0971,
+ "step": 17143
+ },
+ {
+ "epoch": 46.71389645776567,
+ "grad_norm": 7.126631259918213,
+ "learning_rate": 1.1543901599662476e-05,
+ "loss": 0.1048,
+ "step": 17144
+ },
+ {
+ "epoch": 46.716621253406,
+ "grad_norm": 2.3712968826293945,
+ "learning_rate": 1.1543029680163114e-05,
+ "loss": 0.0412,
+ "step": 17145
+ },
+ {
+ "epoch": 46.71934604904632,
+ "grad_norm": 3.165818214416504,
+ "learning_rate": 1.1542157748646687e-05,
+ "loss": 0.0749,
+ "step": 17146
+ },
+ {
+ "epoch": 46.72207084468665,
+ "grad_norm": 4.168436050415039,
+ "learning_rate": 1.1541285805119981e-05,
+ "loss": 0.1132,
+ "step": 17147
+ },
+ {
+ "epoch": 46.724795640326974,
+ "grad_norm": 4.351560592651367,
+ "learning_rate": 1.1540413849589795e-05,
+ "loss": 0.1209,
+ "step": 17148
+ },
+ {
+ "epoch": 46.7275204359673,
+ "grad_norm": 2.520702838897705,
+ "learning_rate": 1.1539541882062912e-05,
+ "loss": 0.059,
+ "step": 17149
+ },
+ {
+ "epoch": 46.73024523160763,
+ "grad_norm": 3.3219428062438965,
+ "learning_rate": 1.1538669902546128e-05,
+ "loss": 0.1408,
+ "step": 17150
+ },
+ {
+ "epoch": 46.73297002724796,
+ "grad_norm": 2.8687148094177246,
+ "learning_rate": 1.153779791104623e-05,
+ "loss": 0.1565,
+ "step": 17151
+ },
+ {
+ "epoch": 46.735694822888284,
+ "grad_norm": 3.277597427368164,
+ "learning_rate": 1.1536925907570017e-05,
+ "loss": 0.113,
+ "step": 17152
+ },
+ {
+ "epoch": 46.73841961852861,
+ "grad_norm": 2.333498954772949,
+ "learning_rate": 1.1536053892124271e-05,
+ "loss": 0.0503,
+ "step": 17153
+ },
+ {
+ "epoch": 46.741144414168936,
+ "grad_norm": 2.3934733867645264,
+ "learning_rate": 1.1535181864715787e-05,
+ "loss": 0.1037,
+ "step": 17154
+ },
+ {
+ "epoch": 46.74386920980926,
+ "grad_norm": 3.3765909671783447,
+ "learning_rate": 1.1534309825351355e-05,
+ "loss": 0.0647,
+ "step": 17155
+ },
+ {
+ "epoch": 46.746594005449595,
+ "grad_norm": 2.78175687789917,
+ "learning_rate": 1.153343777403777e-05,
+ "loss": 0.1227,
+ "step": 17156
+ },
+ {
+ "epoch": 46.74931880108992,
+ "grad_norm": 2.841015577316284,
+ "learning_rate": 1.153256571078182e-05,
+ "loss": 0.3004,
+ "step": 17157
+ },
+ {
+ "epoch": 46.752043596730246,
+ "grad_norm": 3.4508755207061768,
+ "learning_rate": 1.1531693635590297e-05,
+ "loss": 0.1242,
+ "step": 17158
+ },
+ {
+ "epoch": 46.75476839237057,
+ "grad_norm": 4.560981750488281,
+ "learning_rate": 1.153082154846999e-05,
+ "loss": 0.0524,
+ "step": 17159
+ },
+ {
+ "epoch": 46.7574931880109,
+ "grad_norm": 2.2892067432403564,
+ "learning_rate": 1.15299494494277e-05,
+ "loss": 0.0841,
+ "step": 17160
+ },
+ {
+ "epoch": 46.76021798365122,
+ "grad_norm": 2.751690626144409,
+ "learning_rate": 1.152907733847021e-05,
+ "loss": 0.0604,
+ "step": 17161
+ },
+ {
+ "epoch": 46.762942779291556,
+ "grad_norm": 2.8938543796539307,
+ "learning_rate": 1.1528205215604315e-05,
+ "loss": 0.0673,
+ "step": 17162
+ },
+ {
+ "epoch": 46.76566757493188,
+ "grad_norm": 2.8185486793518066,
+ "learning_rate": 1.1527333080836807e-05,
+ "loss": 0.1707,
+ "step": 17163
+ },
+ {
+ "epoch": 46.76839237057221,
+ "grad_norm": 2.6655664443969727,
+ "learning_rate": 1.152646093417448e-05,
+ "loss": 0.0508,
+ "step": 17164
+ },
+ {
+ "epoch": 46.771117166212534,
+ "grad_norm": 2.246617555618286,
+ "learning_rate": 1.1525588775624122e-05,
+ "loss": 0.0518,
+ "step": 17165
+ },
+ {
+ "epoch": 46.77384196185286,
+ "grad_norm": 2.3349545001983643,
+ "learning_rate": 1.1524716605192527e-05,
+ "loss": 0.043,
+ "step": 17166
+ },
+ {
+ "epoch": 46.776566757493185,
+ "grad_norm": 2.468385934829712,
+ "learning_rate": 1.152384442288649e-05,
+ "loss": 0.0502,
+ "step": 17167
+ },
+ {
+ "epoch": 46.77929155313352,
+ "grad_norm": 2.3526806831359863,
+ "learning_rate": 1.1522972228712799e-05,
+ "loss": 0.062,
+ "step": 17168
+ },
+ {
+ "epoch": 46.782016348773844,
+ "grad_norm": 2.794058084487915,
+ "learning_rate": 1.1522100022678251e-05,
+ "loss": 0.1578,
+ "step": 17169
+ },
+ {
+ "epoch": 46.78474114441417,
+ "grad_norm": 3.025059700012207,
+ "learning_rate": 1.1521227804789638e-05,
+ "loss": 0.2657,
+ "step": 17170
+ },
+ {
+ "epoch": 46.787465940054496,
+ "grad_norm": 3.5528392791748047,
+ "learning_rate": 1.1520355575053752e-05,
+ "loss": 0.1268,
+ "step": 17171
+ },
+ {
+ "epoch": 46.79019073569482,
+ "grad_norm": 2.250263214111328,
+ "learning_rate": 1.1519483333477385e-05,
+ "loss": 0.0576,
+ "step": 17172
+ },
+ {
+ "epoch": 46.79291553133515,
+ "grad_norm": 2.642306327819824,
+ "learning_rate": 1.151861108006733e-05,
+ "loss": 0.1444,
+ "step": 17173
+ },
+ {
+ "epoch": 46.79564032697548,
+ "grad_norm": 2.9106287956237793,
+ "learning_rate": 1.1517738814830383e-05,
+ "loss": 0.1077,
+ "step": 17174
+ },
+ {
+ "epoch": 46.798365122615806,
+ "grad_norm": 3.057082414627075,
+ "learning_rate": 1.1516866537773332e-05,
+ "loss": 0.1535,
+ "step": 17175
+ },
+ {
+ "epoch": 46.80108991825613,
+ "grad_norm": 2.884396553039551,
+ "learning_rate": 1.1515994248902976e-05,
+ "loss": 0.1832,
+ "step": 17176
+ },
+ {
+ "epoch": 46.80381471389646,
+ "grad_norm": 3.0862350463867188,
+ "learning_rate": 1.1515121948226102e-05,
+ "loss": 0.0673,
+ "step": 17177
+ },
+ {
+ "epoch": 46.80653950953678,
+ "grad_norm": 3.053504228591919,
+ "learning_rate": 1.1514249635749511e-05,
+ "loss": 0.0885,
+ "step": 17178
+ },
+ {
+ "epoch": 46.80926430517711,
+ "grad_norm": 2.583094835281372,
+ "learning_rate": 1.1513377311479989e-05,
+ "loss": 0.0654,
+ "step": 17179
+ },
+ {
+ "epoch": 46.81198910081744,
+ "grad_norm": 2.2151455879211426,
+ "learning_rate": 1.1512504975424338e-05,
+ "loss": 0.0815,
+ "step": 17180
+ },
+ {
+ "epoch": 46.81471389645777,
+ "grad_norm": 3.2332496643066406,
+ "learning_rate": 1.1511632627589343e-05,
+ "loss": 0.071,
+ "step": 17181
+ },
+ {
+ "epoch": 46.817438692098094,
+ "grad_norm": 3.664963960647583,
+ "learning_rate": 1.1510760267981803e-05,
+ "loss": 0.3222,
+ "step": 17182
+ },
+ {
+ "epoch": 46.82016348773842,
+ "grad_norm": 2.6490883827209473,
+ "learning_rate": 1.1509887896608509e-05,
+ "loss": 0.0744,
+ "step": 17183
+ },
+ {
+ "epoch": 46.822888283378745,
+ "grad_norm": 3.394263982772827,
+ "learning_rate": 1.150901551347626e-05,
+ "loss": 0.1009,
+ "step": 17184
+ },
+ {
+ "epoch": 46.82561307901907,
+ "grad_norm": 3.498683452606201,
+ "learning_rate": 1.1508143118591843e-05,
+ "loss": 0.1528,
+ "step": 17185
+ },
+ {
+ "epoch": 46.828337874659404,
+ "grad_norm": 2.652958393096924,
+ "learning_rate": 1.1507270711962057e-05,
+ "loss": 0.1051,
+ "step": 17186
+ },
+ {
+ "epoch": 46.83106267029973,
+ "grad_norm": 2.511693000793457,
+ "learning_rate": 1.1506398293593697e-05,
+ "loss": 0.0698,
+ "step": 17187
+ },
+ {
+ "epoch": 46.833787465940055,
+ "grad_norm": 2.0262303352355957,
+ "learning_rate": 1.1505525863493554e-05,
+ "loss": 0.0668,
+ "step": 17188
+ },
+ {
+ "epoch": 46.83651226158038,
+ "grad_norm": 2.3623979091644287,
+ "learning_rate": 1.1504653421668424e-05,
+ "loss": 0.1474,
+ "step": 17189
+ },
+ {
+ "epoch": 46.83923705722071,
+ "grad_norm": 2.9458396434783936,
+ "learning_rate": 1.1503780968125105e-05,
+ "loss": 0.1093,
+ "step": 17190
+ },
+ {
+ "epoch": 46.84196185286103,
+ "grad_norm": 4.092609405517578,
+ "learning_rate": 1.1502908502870382e-05,
+ "loss": 0.2126,
+ "step": 17191
+ },
+ {
+ "epoch": 46.844686648501366,
+ "grad_norm": 2.560701370239258,
+ "learning_rate": 1.1502036025911061e-05,
+ "loss": 0.0636,
+ "step": 17192
+ },
+ {
+ "epoch": 46.84741144414169,
+ "grad_norm": 2.839046001434326,
+ "learning_rate": 1.150116353725393e-05,
+ "loss": 0.0746,
+ "step": 17193
+ },
+ {
+ "epoch": 46.85013623978202,
+ "grad_norm": 2.584174633026123,
+ "learning_rate": 1.1500291036905784e-05,
+ "loss": 0.0502,
+ "step": 17194
+ },
+ {
+ "epoch": 46.85286103542234,
+ "grad_norm": 3.190070867538452,
+ "learning_rate": 1.1499418524873422e-05,
+ "loss": 0.219,
+ "step": 17195
+ },
+ {
+ "epoch": 46.85558583106267,
+ "grad_norm": 2.8543541431427,
+ "learning_rate": 1.1498546001163638e-05,
+ "loss": 0.0665,
+ "step": 17196
+ },
+ {
+ "epoch": 46.858310626702995,
+ "grad_norm": 3.369868755340576,
+ "learning_rate": 1.1497673465783224e-05,
+ "loss": 0.1485,
+ "step": 17197
+ },
+ {
+ "epoch": 46.86103542234333,
+ "grad_norm": 3.296168565750122,
+ "learning_rate": 1.1496800918738977e-05,
+ "loss": 0.0523,
+ "step": 17198
+ },
+ {
+ "epoch": 46.86376021798365,
+ "grad_norm": 4.221715927124023,
+ "learning_rate": 1.1495928360037691e-05,
+ "loss": 0.2831,
+ "step": 17199
+ },
+ {
+ "epoch": 46.86648501362398,
+ "grad_norm": 2.786876678466797,
+ "learning_rate": 1.1495055789686167e-05,
+ "loss": 0.1129,
+ "step": 17200
+ },
+ {
+ "epoch": 46.869209809264305,
+ "grad_norm": 2.957418441772461,
+ "learning_rate": 1.1494183207691194e-05,
+ "loss": 0.1348,
+ "step": 17201
+ },
+ {
+ "epoch": 46.87193460490463,
+ "grad_norm": 3.9561595916748047,
+ "learning_rate": 1.149331061405957e-05,
+ "loss": 0.1821,
+ "step": 17202
+ },
+ {
+ "epoch": 46.87465940054496,
+ "grad_norm": 2.3793606758117676,
+ "learning_rate": 1.1492438008798093e-05,
+ "loss": 0.1795,
+ "step": 17203
+ },
+ {
+ "epoch": 46.87738419618529,
+ "grad_norm": 2.577460765838623,
+ "learning_rate": 1.1491565391913555e-05,
+ "loss": 0.1732,
+ "step": 17204
+ },
+ {
+ "epoch": 46.880108991825615,
+ "grad_norm": 2.461723566055298,
+ "learning_rate": 1.1490692763412757e-05,
+ "loss": 0.2816,
+ "step": 17205
+ },
+ {
+ "epoch": 46.88283378746594,
+ "grad_norm": 2.5665111541748047,
+ "learning_rate": 1.1489820123302487e-05,
+ "loss": 0.0714,
+ "step": 17206
+ },
+ {
+ "epoch": 46.88555858310627,
+ "grad_norm": 3.9260666370391846,
+ "learning_rate": 1.148894747158955e-05,
+ "loss": 0.0884,
+ "step": 17207
+ },
+ {
+ "epoch": 46.88828337874659,
+ "grad_norm": 3.041675090789795,
+ "learning_rate": 1.1488074808280736e-05,
+ "loss": 0.1291,
+ "step": 17208
+ },
+ {
+ "epoch": 46.89100817438692,
+ "grad_norm": 2.968337297439575,
+ "learning_rate": 1.1487202133382844e-05,
+ "loss": 0.0813,
+ "step": 17209
+ },
+ {
+ "epoch": 46.89373297002725,
+ "grad_norm": 3.273066759109497,
+ "learning_rate": 1.1486329446902669e-05,
+ "loss": 0.0738,
+ "step": 17210
+ },
+ {
+ "epoch": 46.89645776566758,
+ "grad_norm": 2.594087600708008,
+ "learning_rate": 1.1485456748847006e-05,
+ "loss": 0.1,
+ "step": 17211
+ },
+ {
+ "epoch": 46.8991825613079,
+ "grad_norm": 3.5929081439971924,
+ "learning_rate": 1.1484584039222655e-05,
+ "loss": 0.207,
+ "step": 17212
+ },
+ {
+ "epoch": 46.90190735694823,
+ "grad_norm": 3.045988082885742,
+ "learning_rate": 1.1483711318036414e-05,
+ "loss": 0.0538,
+ "step": 17213
+ },
+ {
+ "epoch": 46.904632152588555,
+ "grad_norm": 2.7803854942321777,
+ "learning_rate": 1.1482838585295076e-05,
+ "loss": 0.0691,
+ "step": 17214
+ },
+ {
+ "epoch": 46.90735694822888,
+ "grad_norm": 3.548527717590332,
+ "learning_rate": 1.1481965841005438e-05,
+ "loss": 0.1143,
+ "step": 17215
+ },
+ {
+ "epoch": 46.91008174386921,
+ "grad_norm": 2.3967480659484863,
+ "learning_rate": 1.1481093085174298e-05,
+ "loss": 0.0864,
+ "step": 17216
+ },
+ {
+ "epoch": 46.91280653950954,
+ "grad_norm": 3.1456551551818848,
+ "learning_rate": 1.1480220317808453e-05,
+ "loss": 0.1618,
+ "step": 17217
+ },
+ {
+ "epoch": 46.915531335149865,
+ "grad_norm": 3.6474170684814453,
+ "learning_rate": 1.1479347538914698e-05,
+ "loss": 0.1034,
+ "step": 17218
+ },
+ {
+ "epoch": 46.91825613079019,
+ "grad_norm": 2.5154266357421875,
+ "learning_rate": 1.1478474748499833e-05,
+ "loss": 0.0639,
+ "step": 17219
+ },
+ {
+ "epoch": 46.920980926430516,
+ "grad_norm": 3.6528542041778564,
+ "learning_rate": 1.1477601946570651e-05,
+ "loss": 0.1617,
+ "step": 17220
+ },
+ {
+ "epoch": 46.92370572207084,
+ "grad_norm": 2.8661913871765137,
+ "learning_rate": 1.147672913313396e-05,
+ "loss": 0.071,
+ "step": 17221
+ },
+ {
+ "epoch": 46.926430517711175,
+ "grad_norm": 2.6706740856170654,
+ "learning_rate": 1.1475856308196544e-05,
+ "loss": 0.0921,
+ "step": 17222
+ },
+ {
+ "epoch": 46.9291553133515,
+ "grad_norm": 2.5091376304626465,
+ "learning_rate": 1.1474983471765211e-05,
+ "loss": 0.1059,
+ "step": 17223
+ },
+ {
+ "epoch": 46.93188010899183,
+ "grad_norm": 2.5499391555786133,
+ "learning_rate": 1.147411062384675e-05,
+ "loss": 0.1567,
+ "step": 17224
+ },
+ {
+ "epoch": 46.93460490463215,
+ "grad_norm": 2.4571268558502197,
+ "learning_rate": 1.1473237764447967e-05,
+ "loss": 0.086,
+ "step": 17225
+ },
+ {
+ "epoch": 46.93732970027248,
+ "grad_norm": 3.9788694381713867,
+ "learning_rate": 1.1472364893575651e-05,
+ "loss": 0.055,
+ "step": 17226
+ },
+ {
+ "epoch": 46.940054495912804,
+ "grad_norm": 3.033494472503662,
+ "learning_rate": 1.1471492011236609e-05,
+ "loss": 0.1172,
+ "step": 17227
+ },
+ {
+ "epoch": 46.94277929155314,
+ "grad_norm": 2.9792864322662354,
+ "learning_rate": 1.1470619117437629e-05,
+ "loss": 0.0974,
+ "step": 17228
+ },
+ {
+ "epoch": 46.94550408719346,
+ "grad_norm": 2.6296958923339844,
+ "learning_rate": 1.146974621218552e-05,
+ "loss": 0.1155,
+ "step": 17229
+ },
+ {
+ "epoch": 46.94822888283379,
+ "grad_norm": 2.7685441970825195,
+ "learning_rate": 1.1468873295487073e-05,
+ "loss": 0.195,
+ "step": 17230
+ },
+ {
+ "epoch": 46.950953678474114,
+ "grad_norm": 2.74686861038208,
+ "learning_rate": 1.1468000367349088e-05,
+ "loss": 0.1022,
+ "step": 17231
+ },
+ {
+ "epoch": 46.95367847411444,
+ "grad_norm": 2.509007453918457,
+ "learning_rate": 1.1467127427778364e-05,
+ "loss": 0.0459,
+ "step": 17232
+ },
+ {
+ "epoch": 46.956403269754766,
+ "grad_norm": 2.999617338180542,
+ "learning_rate": 1.14662544767817e-05,
+ "loss": 0.1063,
+ "step": 17233
+ },
+ {
+ "epoch": 46.95912806539509,
+ "grad_norm": 4.282141208648682,
+ "learning_rate": 1.146538151436589e-05,
+ "loss": 0.0768,
+ "step": 17234
+ },
+ {
+ "epoch": 46.961852861035425,
+ "grad_norm": 2.5955333709716797,
+ "learning_rate": 1.1464508540537738e-05,
+ "loss": 0.1342,
+ "step": 17235
+ },
+ {
+ "epoch": 46.96457765667575,
+ "grad_norm": 2.9923465251922607,
+ "learning_rate": 1.1463635555304038e-05,
+ "loss": 0.0611,
+ "step": 17236
+ },
+ {
+ "epoch": 46.967302452316076,
+ "grad_norm": 3.462531328201294,
+ "learning_rate": 1.1462762558671594e-05,
+ "loss": 0.069,
+ "step": 17237
+ },
+ {
+ "epoch": 46.9700272479564,
+ "grad_norm": 2.7188425064086914,
+ "learning_rate": 1.14618895506472e-05,
+ "loss": 0.1019,
+ "step": 17238
+ },
+ {
+ "epoch": 46.97275204359673,
+ "grad_norm": 2.224890947341919,
+ "learning_rate": 1.1461016531237661e-05,
+ "loss": 0.0484,
+ "step": 17239
+ },
+ {
+ "epoch": 46.97547683923706,
+ "grad_norm": 2.9244225025177,
+ "learning_rate": 1.1460143500449769e-05,
+ "loss": 0.2415,
+ "step": 17240
+ },
+ {
+ "epoch": 46.97820163487739,
+ "grad_norm": 3.94286847114563,
+ "learning_rate": 1.145927045829033e-05,
+ "loss": 0.1619,
+ "step": 17241
+ },
+ {
+ "epoch": 46.98092643051771,
+ "grad_norm": 2.875850200653076,
+ "learning_rate": 1.1458397404766138e-05,
+ "loss": 0.0601,
+ "step": 17242
+ },
+ {
+ "epoch": 46.98365122615804,
+ "grad_norm": 2.0453877449035645,
+ "learning_rate": 1.1457524339883994e-05,
+ "loss": 0.1353,
+ "step": 17243
+ },
+ {
+ "epoch": 46.986376021798364,
+ "grad_norm": 5.691519737243652,
+ "learning_rate": 1.1456651263650695e-05,
+ "loss": 0.0673,
+ "step": 17244
+ },
+ {
+ "epoch": 46.98910081743869,
+ "grad_norm": 2.610470771789551,
+ "learning_rate": 1.1455778176073045e-05,
+ "loss": 0.0779,
+ "step": 17245
+ },
+ {
+ "epoch": 46.991825613079016,
+ "grad_norm": 2.6388072967529297,
+ "learning_rate": 1.1454905077157839e-05,
+ "loss": 0.0538,
+ "step": 17246
+ },
+ {
+ "epoch": 46.99455040871935,
+ "grad_norm": 4.500870704650879,
+ "learning_rate": 1.1454031966911883e-05,
+ "loss": 0.085,
+ "step": 17247
+ },
+ {
+ "epoch": 46.997275204359674,
+ "grad_norm": 2.7130391597747803,
+ "learning_rate": 1.1453158845341969e-05,
+ "loss": 0.0541,
+ "step": 17248
+ },
+ {
+ "epoch": 47.0,
+ "grad_norm": 2.2755744457244873,
+ "learning_rate": 1.1452285712454905e-05,
+ "loss": 0.0442,
+ "step": 17249
+ },
+ {
+ "epoch": 47.002724795640326,
+ "grad_norm": 4.036783218383789,
+ "learning_rate": 1.1451412568257484e-05,
+ "loss": 0.0843,
+ "step": 17250
+ },
+ {
+ "epoch": 47.00544959128065,
+ "grad_norm": 1.5764706134796143,
+ "learning_rate": 1.1450539412756509e-05,
+ "loss": 0.0291,
+ "step": 17251
+ },
+ {
+ "epoch": 47.00817438692098,
+ "grad_norm": 2.4137492179870605,
+ "learning_rate": 1.144966624595878e-05,
+ "loss": 0.0639,
+ "step": 17252
+ },
+ {
+ "epoch": 47.01089918256131,
+ "grad_norm": 2.95823073387146,
+ "learning_rate": 1.1448793067871095e-05,
+ "loss": 0.0669,
+ "step": 17253
+ },
+ {
+ "epoch": 47.013623978201636,
+ "grad_norm": 2.4471585750579834,
+ "learning_rate": 1.1447919878500255e-05,
+ "loss": 0.0669,
+ "step": 17254
+ },
+ {
+ "epoch": 47.01634877384196,
+ "grad_norm": 2.5213522911071777,
+ "learning_rate": 1.1447046677853066e-05,
+ "loss": 0.2697,
+ "step": 17255
+ },
+ {
+ "epoch": 47.01907356948229,
+ "grad_norm": 2.7867918014526367,
+ "learning_rate": 1.1446173465936322e-05,
+ "loss": 0.1052,
+ "step": 17256
+ },
+ {
+ "epoch": 47.02179836512261,
+ "grad_norm": 1.9856781959533691,
+ "learning_rate": 1.1445300242756827e-05,
+ "loss": 0.0446,
+ "step": 17257
+ },
+ {
+ "epoch": 47.02452316076294,
+ "grad_norm": 2.603065013885498,
+ "learning_rate": 1.144442700832138e-05,
+ "loss": 0.12,
+ "step": 17258
+ },
+ {
+ "epoch": 47.02724795640327,
+ "grad_norm": 1.7283841371536255,
+ "learning_rate": 1.1443553762636781e-05,
+ "loss": 0.039,
+ "step": 17259
+ },
+ {
+ "epoch": 47.0299727520436,
+ "grad_norm": 2.968872308731079,
+ "learning_rate": 1.1442680505709832e-05,
+ "loss": 0.0598,
+ "step": 17260
+ },
+ {
+ "epoch": 47.032697547683924,
+ "grad_norm": 2.6774446964263916,
+ "learning_rate": 1.1441807237547336e-05,
+ "loss": 0.0684,
+ "step": 17261
+ },
+ {
+ "epoch": 47.03542234332425,
+ "grad_norm": 2.5980489253997803,
+ "learning_rate": 1.1440933958156087e-05,
+ "loss": 0.13,
+ "step": 17262
+ },
+ {
+ "epoch": 47.038147138964575,
+ "grad_norm": 3.0348329544067383,
+ "learning_rate": 1.1440060667542891e-05,
+ "loss": 0.0518,
+ "step": 17263
+ },
+ {
+ "epoch": 47.0408719346049,
+ "grad_norm": 2.503164768218994,
+ "learning_rate": 1.143918736571455e-05,
+ "loss": 0.05,
+ "step": 17264
+ },
+ {
+ "epoch": 47.043596730245234,
+ "grad_norm": 2.3268754482269287,
+ "learning_rate": 1.1438314052677865e-05,
+ "loss": 0.0712,
+ "step": 17265
+ },
+ {
+ "epoch": 47.04632152588556,
+ "grad_norm": 2.487248182296753,
+ "learning_rate": 1.1437440728439637e-05,
+ "loss": 0.0459,
+ "step": 17266
+ },
+ {
+ "epoch": 47.049046321525886,
+ "grad_norm": 3.7758982181549072,
+ "learning_rate": 1.1436567393006668e-05,
+ "loss": 0.1356,
+ "step": 17267
+ },
+ {
+ "epoch": 47.05177111716621,
+ "grad_norm": 3.240025520324707,
+ "learning_rate": 1.1435694046385755e-05,
+ "loss": 0.0773,
+ "step": 17268
+ },
+ {
+ "epoch": 47.05449591280654,
+ "grad_norm": 3.2546215057373047,
+ "learning_rate": 1.1434820688583706e-05,
+ "loss": 0.0542,
+ "step": 17269
+ },
+ {
+ "epoch": 47.05722070844686,
+ "grad_norm": 2.4813854694366455,
+ "learning_rate": 1.1433947319607318e-05,
+ "loss": 0.0707,
+ "step": 17270
+ },
+ {
+ "epoch": 47.059945504087196,
+ "grad_norm": 2.329991340637207,
+ "learning_rate": 1.1433073939463393e-05,
+ "loss": 0.0515,
+ "step": 17271
+ },
+ {
+ "epoch": 47.06267029972752,
+ "grad_norm": 2.8097517490386963,
+ "learning_rate": 1.1432200548158738e-05,
+ "loss": 0.1063,
+ "step": 17272
+ },
+ {
+ "epoch": 47.06539509536785,
+ "grad_norm": 2.6612868309020996,
+ "learning_rate": 1.1431327145700149e-05,
+ "loss": 0.1947,
+ "step": 17273
+ },
+ {
+ "epoch": 47.06811989100817,
+ "grad_norm": 1.9818838834762573,
+ "learning_rate": 1.143045373209443e-05,
+ "loss": 0.0461,
+ "step": 17274
+ },
+ {
+ "epoch": 47.0708446866485,
+ "grad_norm": 2.408891439437866,
+ "learning_rate": 1.1429580307348385e-05,
+ "loss": 0.128,
+ "step": 17275
+ },
+ {
+ "epoch": 47.073569482288825,
+ "grad_norm": 3.7022008895874023,
+ "learning_rate": 1.142870687146881e-05,
+ "loss": 0.0601,
+ "step": 17276
+ },
+ {
+ "epoch": 47.07629427792916,
+ "grad_norm": 3.4577178955078125,
+ "learning_rate": 1.1427833424462517e-05,
+ "loss": 0.1065,
+ "step": 17277
+ },
+ {
+ "epoch": 47.079019073569484,
+ "grad_norm": 4.946778297424316,
+ "learning_rate": 1.1426959966336302e-05,
+ "loss": 0.1022,
+ "step": 17278
+ },
+ {
+ "epoch": 47.08174386920981,
+ "grad_norm": 2.4843122959136963,
+ "learning_rate": 1.1426086497096966e-05,
+ "loss": 0.0903,
+ "step": 17279
+ },
+ {
+ "epoch": 47.084468664850135,
+ "grad_norm": 1.9901041984558105,
+ "learning_rate": 1.1425213016751317e-05,
+ "loss": 0.05,
+ "step": 17280
+ },
+ {
+ "epoch": 47.08719346049046,
+ "grad_norm": 2.039046049118042,
+ "learning_rate": 1.1424339525306154e-05,
+ "loss": 0.0522,
+ "step": 17281
+ },
+ {
+ "epoch": 47.08991825613079,
+ "grad_norm": 2.392059803009033,
+ "learning_rate": 1.1423466022768283e-05,
+ "loss": 0.0524,
+ "step": 17282
+ },
+ {
+ "epoch": 47.09264305177112,
+ "grad_norm": 2.8223798274993896,
+ "learning_rate": 1.14225925091445e-05,
+ "loss": 0.1265,
+ "step": 17283
+ },
+ {
+ "epoch": 47.095367847411445,
+ "grad_norm": 2.765836715698242,
+ "learning_rate": 1.1421718984441617e-05,
+ "loss": 0.2082,
+ "step": 17284
+ },
+ {
+ "epoch": 47.09809264305177,
+ "grad_norm": 2.5023694038391113,
+ "learning_rate": 1.1420845448666429e-05,
+ "loss": 0.2417,
+ "step": 17285
+ },
+ {
+ "epoch": 47.1008174386921,
+ "grad_norm": 1.9745312929153442,
+ "learning_rate": 1.1419971901825743e-05,
+ "loss": 0.1091,
+ "step": 17286
+ },
+ {
+ "epoch": 47.10354223433242,
+ "grad_norm": 3.104161500930786,
+ "learning_rate": 1.1419098343926363e-05,
+ "loss": 0.1402,
+ "step": 17287
+ },
+ {
+ "epoch": 47.10626702997275,
+ "grad_norm": 2.339876413345337,
+ "learning_rate": 1.1418224774975088e-05,
+ "loss": 0.0574,
+ "step": 17288
+ },
+ {
+ "epoch": 47.10899182561308,
+ "grad_norm": 3.1297030448913574,
+ "learning_rate": 1.1417351194978725e-05,
+ "loss": 0.181,
+ "step": 17289
+ },
+ {
+ "epoch": 47.11171662125341,
+ "grad_norm": 2.5031256675720215,
+ "learning_rate": 1.1416477603944078e-05,
+ "loss": 0.0566,
+ "step": 17290
+ },
+ {
+ "epoch": 47.11444141689373,
+ "grad_norm": 2.108046054840088,
+ "learning_rate": 1.1415604001877949e-05,
+ "loss": 0.1276,
+ "step": 17291
+ },
+ {
+ "epoch": 47.11716621253406,
+ "grad_norm": 3.3281211853027344,
+ "learning_rate": 1.1414730388787143e-05,
+ "loss": 0.075,
+ "step": 17292
+ },
+ {
+ "epoch": 47.119891008174385,
+ "grad_norm": 2.7752957344055176,
+ "learning_rate": 1.1413856764678459e-05,
+ "loss": 0.0775,
+ "step": 17293
+ },
+ {
+ "epoch": 47.12261580381471,
+ "grad_norm": 3.1550331115722656,
+ "learning_rate": 1.1412983129558706e-05,
+ "loss": 0.159,
+ "step": 17294
+ },
+ {
+ "epoch": 47.12534059945504,
+ "grad_norm": 2.574998617172241,
+ "learning_rate": 1.1412109483434683e-05,
+ "loss": 0.0792,
+ "step": 17295
+ },
+ {
+ "epoch": 47.12806539509537,
+ "grad_norm": 17.580171585083008,
+ "learning_rate": 1.1411235826313201e-05,
+ "loss": 0.1118,
+ "step": 17296
+ },
+ {
+ "epoch": 47.130790190735695,
+ "grad_norm": 3.251979112625122,
+ "learning_rate": 1.1410362158201054e-05,
+ "loss": 0.2301,
+ "step": 17297
+ },
+ {
+ "epoch": 47.13351498637602,
+ "grad_norm": 2.2546610832214355,
+ "learning_rate": 1.1409488479105058e-05,
+ "loss": 0.051,
+ "step": 17298
+ },
+ {
+ "epoch": 47.13623978201635,
+ "grad_norm": 1.9009510278701782,
+ "learning_rate": 1.1408614789032007e-05,
+ "loss": 0.0744,
+ "step": 17299
+ },
+ {
+ "epoch": 47.13896457765667,
+ "grad_norm": 3.2430012226104736,
+ "learning_rate": 1.1407741087988713e-05,
+ "loss": 0.0467,
+ "step": 17300
+ },
+ {
+ "epoch": 47.141689373297005,
+ "grad_norm": 2.560361385345459,
+ "learning_rate": 1.1406867375981975e-05,
+ "loss": 0.1713,
+ "step": 17301
+ },
+ {
+ "epoch": 47.14441416893733,
+ "grad_norm": 2.5025529861450195,
+ "learning_rate": 1.1405993653018599e-05,
+ "loss": 0.0578,
+ "step": 17302
+ },
+ {
+ "epoch": 47.14713896457766,
+ "grad_norm": 3.2581725120544434,
+ "learning_rate": 1.1405119919105387e-05,
+ "loss": 0.1751,
+ "step": 17303
+ },
+ {
+ "epoch": 47.14986376021798,
+ "grad_norm": 3.042536497116089,
+ "learning_rate": 1.1404246174249152e-05,
+ "loss": 0.0548,
+ "step": 17304
+ },
+ {
+ "epoch": 47.15258855585831,
+ "grad_norm": 4.7469706535339355,
+ "learning_rate": 1.1403372418456687e-05,
+ "loss": 0.1467,
+ "step": 17305
+ },
+ {
+ "epoch": 47.155313351498634,
+ "grad_norm": 2.3862175941467285,
+ "learning_rate": 1.1402498651734806e-05,
+ "loss": 0.0545,
+ "step": 17306
+ },
+ {
+ "epoch": 47.15803814713897,
+ "grad_norm": 3.3881897926330566,
+ "learning_rate": 1.1401624874090309e-05,
+ "loss": 0.0919,
+ "step": 17307
+ },
+ {
+ "epoch": 47.16076294277929,
+ "grad_norm": 3.966330051422119,
+ "learning_rate": 1.1400751085530006e-05,
+ "loss": 0.057,
+ "step": 17308
+ },
+ {
+ "epoch": 47.16348773841962,
+ "grad_norm": 2.6897926330566406,
+ "learning_rate": 1.1399877286060695e-05,
+ "loss": 0.056,
+ "step": 17309
+ },
+ {
+ "epoch": 47.166212534059945,
+ "grad_norm": 1.7546682357788086,
+ "learning_rate": 1.1399003475689187e-05,
+ "loss": 0.0339,
+ "step": 17310
+ },
+ {
+ "epoch": 47.16893732970027,
+ "grad_norm": 3.1685333251953125,
+ "learning_rate": 1.1398129654422284e-05,
+ "loss": 0.132,
+ "step": 17311
+ },
+ {
+ "epoch": 47.171662125340596,
+ "grad_norm": 3.047367572784424,
+ "learning_rate": 1.1397255822266794e-05,
+ "loss": 0.0498,
+ "step": 17312
+ },
+ {
+ "epoch": 47.17438692098093,
+ "grad_norm": 3.0987958908081055,
+ "learning_rate": 1.1396381979229518e-05,
+ "loss": 0.1118,
+ "step": 17313
+ },
+ {
+ "epoch": 47.177111716621255,
+ "grad_norm": 2.40488338470459,
+ "learning_rate": 1.1395508125317266e-05,
+ "loss": 0.0499,
+ "step": 17314
+ },
+ {
+ "epoch": 47.17983651226158,
+ "grad_norm": 3.309950113296509,
+ "learning_rate": 1.1394634260536839e-05,
+ "loss": 0.0583,
+ "step": 17315
+ },
+ {
+ "epoch": 47.182561307901906,
+ "grad_norm": 2.69107985496521,
+ "learning_rate": 1.1393760384895047e-05,
+ "loss": 0.0515,
+ "step": 17316
+ },
+ {
+ "epoch": 47.18528610354223,
+ "grad_norm": 2.981142997741699,
+ "learning_rate": 1.1392886498398695e-05,
+ "loss": 0.0837,
+ "step": 17317
+ },
+ {
+ "epoch": 47.18801089918256,
+ "grad_norm": 2.3740482330322266,
+ "learning_rate": 1.1392012601054588e-05,
+ "loss": 0.0607,
+ "step": 17318
+ },
+ {
+ "epoch": 47.19073569482289,
+ "grad_norm": 3.448303461074829,
+ "learning_rate": 1.1391138692869528e-05,
+ "loss": 0.0778,
+ "step": 17319
+ },
+ {
+ "epoch": 47.19346049046322,
+ "grad_norm": 3.148115396499634,
+ "learning_rate": 1.1390264773850329e-05,
+ "loss": 0.085,
+ "step": 17320
+ },
+ {
+ "epoch": 47.19618528610354,
+ "grad_norm": 3.1850531101226807,
+ "learning_rate": 1.1389390844003789e-05,
+ "loss": 0.0745,
+ "step": 17321
+ },
+ {
+ "epoch": 47.19891008174387,
+ "grad_norm": 2.775909900665283,
+ "learning_rate": 1.138851690333672e-05,
+ "loss": 0.132,
+ "step": 17322
+ },
+ {
+ "epoch": 47.201634877384194,
+ "grad_norm": 4.510265827178955,
+ "learning_rate": 1.1387642951855925e-05,
+ "loss": 0.1094,
+ "step": 17323
+ },
+ {
+ "epoch": 47.20435967302452,
+ "grad_norm": 3.281630754470825,
+ "learning_rate": 1.1386768989568213e-05,
+ "loss": 0.057,
+ "step": 17324
+ },
+ {
+ "epoch": 47.20708446866485,
+ "grad_norm": 2.4624252319335938,
+ "learning_rate": 1.1385895016480387e-05,
+ "loss": 0.12,
+ "step": 17325
+ },
+ {
+ "epoch": 47.20980926430518,
+ "grad_norm": 2.8787269592285156,
+ "learning_rate": 1.1385021032599257e-05,
+ "loss": 0.0756,
+ "step": 17326
+ },
+ {
+ "epoch": 47.212534059945504,
+ "grad_norm": 3.644456386566162,
+ "learning_rate": 1.1384147037931625e-05,
+ "loss": 0.0508,
+ "step": 17327
+ },
+ {
+ "epoch": 47.21525885558583,
+ "grad_norm": 3.1293067932128906,
+ "learning_rate": 1.1383273032484302e-05,
+ "loss": 0.1497,
+ "step": 17328
+ },
+ {
+ "epoch": 47.217983651226156,
+ "grad_norm": 2.9675722122192383,
+ "learning_rate": 1.1382399016264093e-05,
+ "loss": 0.0607,
+ "step": 17329
+ },
+ {
+ "epoch": 47.22070844686648,
+ "grad_norm": 2.9031078815460205,
+ "learning_rate": 1.1381524989277807e-05,
+ "loss": 0.1798,
+ "step": 17330
+ },
+ {
+ "epoch": 47.223433242506815,
+ "grad_norm": 2.7518718242645264,
+ "learning_rate": 1.1380650951532243e-05,
+ "loss": 0.053,
+ "step": 17331
+ },
+ {
+ "epoch": 47.22615803814714,
+ "grad_norm": 2.865051746368408,
+ "learning_rate": 1.137977690303422e-05,
+ "loss": 0.2731,
+ "step": 17332
+ },
+ {
+ "epoch": 47.228882833787466,
+ "grad_norm": 2.6309916973114014,
+ "learning_rate": 1.1378902843790538e-05,
+ "loss": 0.186,
+ "step": 17333
+ },
+ {
+ "epoch": 47.23160762942779,
+ "grad_norm": 2.056133508682251,
+ "learning_rate": 1.1378028773808003e-05,
+ "loss": 0.0554,
+ "step": 17334
+ },
+ {
+ "epoch": 47.23433242506812,
+ "grad_norm": 2.451977252960205,
+ "learning_rate": 1.1377154693093426e-05,
+ "loss": 0.0339,
+ "step": 17335
+ },
+ {
+ "epoch": 47.237057220708444,
+ "grad_norm": 2.0942957401275635,
+ "learning_rate": 1.1376280601653613e-05,
+ "loss": 0.0512,
+ "step": 17336
+ },
+ {
+ "epoch": 47.23978201634878,
+ "grad_norm": 3.3514645099639893,
+ "learning_rate": 1.1375406499495367e-05,
+ "loss": 0.2283,
+ "step": 17337
+ },
+ {
+ "epoch": 47.2425068119891,
+ "grad_norm": 2.695680618286133,
+ "learning_rate": 1.1374532386625506e-05,
+ "loss": 0.0754,
+ "step": 17338
+ },
+ {
+ "epoch": 47.24523160762943,
+ "grad_norm": 2.9595701694488525,
+ "learning_rate": 1.1373658263050824e-05,
+ "loss": 0.1402,
+ "step": 17339
+ },
+ {
+ "epoch": 47.247956403269754,
+ "grad_norm": 3.130528211593628,
+ "learning_rate": 1.1372784128778138e-05,
+ "loss": 0.1329,
+ "step": 17340
+ },
+ {
+ "epoch": 47.25068119891008,
+ "grad_norm": 2.666588544845581,
+ "learning_rate": 1.1371909983814253e-05,
+ "loss": 0.1051,
+ "step": 17341
+ },
+ {
+ "epoch": 47.253405994550405,
+ "grad_norm": 2.185654640197754,
+ "learning_rate": 1.1371035828165979e-05,
+ "loss": 0.1003,
+ "step": 17342
+ },
+ {
+ "epoch": 47.25613079019074,
+ "grad_norm": 2.528934955596924,
+ "learning_rate": 1.1370161661840122e-05,
+ "loss": 0.1825,
+ "step": 17343
+ },
+ {
+ "epoch": 47.258855585831064,
+ "grad_norm": 2.415579080581665,
+ "learning_rate": 1.136928748484349e-05,
+ "loss": 0.1056,
+ "step": 17344
+ },
+ {
+ "epoch": 47.26158038147139,
+ "grad_norm": 2.384828567504883,
+ "learning_rate": 1.1368413297182892e-05,
+ "loss": 0.1012,
+ "step": 17345
+ },
+ {
+ "epoch": 47.264305177111716,
+ "grad_norm": 2.6950340270996094,
+ "learning_rate": 1.1367539098865134e-05,
+ "loss": 0.0626,
+ "step": 17346
+ },
+ {
+ "epoch": 47.26702997275204,
+ "grad_norm": 2.9147887229919434,
+ "learning_rate": 1.1366664889897025e-05,
+ "loss": 0.0991,
+ "step": 17347
+ },
+ {
+ "epoch": 47.26975476839237,
+ "grad_norm": 2.905837297439575,
+ "learning_rate": 1.1365790670285372e-05,
+ "loss": 0.1299,
+ "step": 17348
+ },
+ {
+ "epoch": 47.2724795640327,
+ "grad_norm": 2.538835048675537,
+ "learning_rate": 1.136491644003699e-05,
+ "loss": 0.0443,
+ "step": 17349
+ },
+ {
+ "epoch": 47.275204359673026,
+ "grad_norm": 2.9323060512542725,
+ "learning_rate": 1.1364042199158681e-05,
+ "loss": 0.0508,
+ "step": 17350
+ },
+ {
+ "epoch": 47.27792915531335,
+ "grad_norm": 2.7492620944976807,
+ "learning_rate": 1.1363167947657252e-05,
+ "loss": 0.063,
+ "step": 17351
+ },
+ {
+ "epoch": 47.28065395095368,
+ "grad_norm": 2.6618330478668213,
+ "learning_rate": 1.1362293685539518e-05,
+ "loss": 0.0609,
+ "step": 17352
+ },
+ {
+ "epoch": 47.283378746594,
+ "grad_norm": 2.3398683071136475,
+ "learning_rate": 1.1361419412812284e-05,
+ "loss": 0.0569,
+ "step": 17353
+ },
+ {
+ "epoch": 47.28610354223433,
+ "grad_norm": 4.235284805297852,
+ "learning_rate": 1.1360545129482361e-05,
+ "loss": 0.105,
+ "step": 17354
+ },
+ {
+ "epoch": 47.28882833787466,
+ "grad_norm": 2.3547167778015137,
+ "learning_rate": 1.1359670835556554e-05,
+ "loss": 0.0974,
+ "step": 17355
+ },
+ {
+ "epoch": 47.29155313351499,
+ "grad_norm": 2.1292388439178467,
+ "learning_rate": 1.1358796531041676e-05,
+ "loss": 0.045,
+ "step": 17356
+ },
+ {
+ "epoch": 47.294277929155314,
+ "grad_norm": 3.1206748485565186,
+ "learning_rate": 1.1357922215944533e-05,
+ "loss": 0.0654,
+ "step": 17357
+ },
+ {
+ "epoch": 47.29700272479564,
+ "grad_norm": 2.6258625984191895,
+ "learning_rate": 1.1357047890271935e-05,
+ "loss": 0.2364,
+ "step": 17358
+ },
+ {
+ "epoch": 47.299727520435965,
+ "grad_norm": 2.2485029697418213,
+ "learning_rate": 1.1356173554030692e-05,
+ "loss": 0.0433,
+ "step": 17359
+ },
+ {
+ "epoch": 47.30245231607629,
+ "grad_norm": 2.8789405822753906,
+ "learning_rate": 1.1355299207227612e-05,
+ "loss": 0.1057,
+ "step": 17360
+ },
+ {
+ "epoch": 47.305177111716624,
+ "grad_norm": 2.4508163928985596,
+ "learning_rate": 1.135442484986951e-05,
+ "loss": 0.0645,
+ "step": 17361
+ },
+ {
+ "epoch": 47.30790190735695,
+ "grad_norm": 2.0694892406463623,
+ "learning_rate": 1.1353550481963188e-05,
+ "loss": 0.1191,
+ "step": 17362
+ },
+ {
+ "epoch": 47.310626702997276,
+ "grad_norm": 2.7476706504821777,
+ "learning_rate": 1.1352676103515456e-05,
+ "loss": 0.0671,
+ "step": 17363
+ },
+ {
+ "epoch": 47.3133514986376,
+ "grad_norm": 2.4952943325042725,
+ "learning_rate": 1.135180171453313e-05,
+ "loss": 0.1332,
+ "step": 17364
+ },
+ {
+ "epoch": 47.31607629427793,
+ "grad_norm": 2.574416160583496,
+ "learning_rate": 1.1350927315023012e-05,
+ "loss": 0.1294,
+ "step": 17365
+ },
+ {
+ "epoch": 47.31880108991825,
+ "grad_norm": 2.268399238586426,
+ "learning_rate": 1.1350052904991917e-05,
+ "loss": 0.1483,
+ "step": 17366
+ },
+ {
+ "epoch": 47.321525885558586,
+ "grad_norm": 2.9332435131073,
+ "learning_rate": 1.1349178484446655e-05,
+ "loss": 0.1704,
+ "step": 17367
+ },
+ {
+ "epoch": 47.32425068119891,
+ "grad_norm": 2.779252529144287,
+ "learning_rate": 1.1348304053394032e-05,
+ "loss": 0.0633,
+ "step": 17368
+ },
+ {
+ "epoch": 47.32697547683924,
+ "grad_norm": 2.623027801513672,
+ "learning_rate": 1.1347429611840863e-05,
+ "loss": 0.0733,
+ "step": 17369
+ },
+ {
+ "epoch": 47.32970027247956,
+ "grad_norm": 2.8476290702819824,
+ "learning_rate": 1.1346555159793956e-05,
+ "loss": 0.3431,
+ "step": 17370
+ },
+ {
+ "epoch": 47.33242506811989,
+ "grad_norm": 2.640658140182495,
+ "learning_rate": 1.134568069726012e-05,
+ "loss": 0.0965,
+ "step": 17371
+ },
+ {
+ "epoch": 47.335149863760215,
+ "grad_norm": 2.0303773880004883,
+ "learning_rate": 1.1344806224246164e-05,
+ "loss": 0.0335,
+ "step": 17372
+ },
+ {
+ "epoch": 47.33787465940055,
+ "grad_norm": 2.7480435371398926,
+ "learning_rate": 1.13439317407589e-05,
+ "loss": 0.137,
+ "step": 17373
+ },
+ {
+ "epoch": 47.34059945504087,
+ "grad_norm": 2.8585455417633057,
+ "learning_rate": 1.134305724680514e-05,
+ "loss": 0.0534,
+ "step": 17374
+ },
+ {
+ "epoch": 47.3433242506812,
+ "grad_norm": 2.8530051708221436,
+ "learning_rate": 1.1342182742391693e-05,
+ "loss": 0.2069,
+ "step": 17375
+ },
+ {
+ "epoch": 47.346049046321525,
+ "grad_norm": 2.349867105484009,
+ "learning_rate": 1.1341308227525372e-05,
+ "loss": 0.0722,
+ "step": 17376
+ },
+ {
+ "epoch": 47.34877384196185,
+ "grad_norm": 2.161571741104126,
+ "learning_rate": 1.1340433702212986e-05,
+ "loss": 0.0695,
+ "step": 17377
+ },
+ {
+ "epoch": 47.35149863760218,
+ "grad_norm": 3.320021629333496,
+ "learning_rate": 1.1339559166461343e-05,
+ "loss": 0.1452,
+ "step": 17378
+ },
+ {
+ "epoch": 47.35422343324251,
+ "grad_norm": 2.23953914642334,
+ "learning_rate": 1.1338684620277259e-05,
+ "loss": 0.0497,
+ "step": 17379
+ },
+ {
+ "epoch": 47.356948228882835,
+ "grad_norm": 3.74613356590271,
+ "learning_rate": 1.1337810063667539e-05,
+ "loss": 0.0815,
+ "step": 17380
+ },
+ {
+ "epoch": 47.35967302452316,
+ "grad_norm": 3.0376832485198975,
+ "learning_rate": 1.1336935496639e-05,
+ "loss": 0.1241,
+ "step": 17381
+ },
+ {
+ "epoch": 47.36239782016349,
+ "grad_norm": 2.1717846393585205,
+ "learning_rate": 1.1336060919198449e-05,
+ "loss": 0.0466,
+ "step": 17382
+ },
+ {
+ "epoch": 47.36512261580381,
+ "grad_norm": 2.2589638233184814,
+ "learning_rate": 1.1335186331352697e-05,
+ "loss": 0.0428,
+ "step": 17383
+ },
+ {
+ "epoch": 47.36784741144414,
+ "grad_norm": 2.9791994094848633,
+ "learning_rate": 1.1334311733108558e-05,
+ "loss": 0.0586,
+ "step": 17384
+ },
+ {
+ "epoch": 47.37057220708447,
+ "grad_norm": 2.779087543487549,
+ "learning_rate": 1.1333437124472844e-05,
+ "loss": 0.0556,
+ "step": 17385
+ },
+ {
+ "epoch": 47.3732970027248,
+ "grad_norm": 2.5089547634124756,
+ "learning_rate": 1.1332562505452363e-05,
+ "loss": 0.0772,
+ "step": 17386
+ },
+ {
+ "epoch": 47.37602179836512,
+ "grad_norm": 2.1826696395874023,
+ "learning_rate": 1.1331687876053929e-05,
+ "loss": 0.0703,
+ "step": 17387
+ },
+ {
+ "epoch": 47.37874659400545,
+ "grad_norm": 2.56069016456604,
+ "learning_rate": 1.1330813236284353e-05,
+ "loss": 0.0745,
+ "step": 17388
+ },
+ {
+ "epoch": 47.381471389645775,
+ "grad_norm": 2.577807664871216,
+ "learning_rate": 1.1329938586150445e-05,
+ "loss": 0.1045,
+ "step": 17389
+ },
+ {
+ "epoch": 47.3841961852861,
+ "grad_norm": 3.7020347118377686,
+ "learning_rate": 1.1329063925659018e-05,
+ "loss": 0.1053,
+ "step": 17390
+ },
+ {
+ "epoch": 47.38692098092643,
+ "grad_norm": 2.601081371307373,
+ "learning_rate": 1.1328189254816883e-05,
+ "loss": 0.0744,
+ "step": 17391
+ },
+ {
+ "epoch": 47.38964577656676,
+ "grad_norm": 3.3807342052459717,
+ "learning_rate": 1.1327314573630855e-05,
+ "loss": 0.077,
+ "step": 17392
+ },
+ {
+ "epoch": 47.392370572207085,
+ "grad_norm": 1.7726649045944214,
+ "learning_rate": 1.1326439882107745e-05,
+ "loss": 0.0853,
+ "step": 17393
+ },
+ {
+ "epoch": 47.39509536784741,
+ "grad_norm": 2.4840471744537354,
+ "learning_rate": 1.132556518025436e-05,
+ "loss": 0.1994,
+ "step": 17394
+ },
+ {
+ "epoch": 47.39782016348774,
+ "grad_norm": 3.3093326091766357,
+ "learning_rate": 1.132469046807752e-05,
+ "loss": 0.0681,
+ "step": 17395
+ },
+ {
+ "epoch": 47.40054495912806,
+ "grad_norm": 2.64337420463562,
+ "learning_rate": 1.132381574558403e-05,
+ "loss": 0.0848,
+ "step": 17396
+ },
+ {
+ "epoch": 47.403269754768395,
+ "grad_norm": 2.978961706161499,
+ "learning_rate": 1.1322941012780707e-05,
+ "loss": 0.1624,
+ "step": 17397
+ },
+ {
+ "epoch": 47.40599455040872,
+ "grad_norm": 1.7001557350158691,
+ "learning_rate": 1.1322066269674364e-05,
+ "loss": 0.1149,
+ "step": 17398
+ },
+ {
+ "epoch": 47.40871934604905,
+ "grad_norm": 2.032486915588379,
+ "learning_rate": 1.132119151627181e-05,
+ "loss": 0.046,
+ "step": 17399
+ },
+ {
+ "epoch": 47.41144414168937,
+ "grad_norm": 3.0266432762145996,
+ "learning_rate": 1.1320316752579854e-05,
+ "loss": 0.0874,
+ "step": 17400
+ },
+ {
+ "epoch": 47.4141689373297,
+ "grad_norm": 2.690242052078247,
+ "learning_rate": 1.1319441978605319e-05,
+ "loss": 0.2054,
+ "step": 17401
+ },
+ {
+ "epoch": 47.416893732970024,
+ "grad_norm": 2.615290641784668,
+ "learning_rate": 1.131856719435501e-05,
+ "loss": 0.1004,
+ "step": 17402
+ },
+ {
+ "epoch": 47.41961852861036,
+ "grad_norm": 4.606714248657227,
+ "learning_rate": 1.1317692399835746e-05,
+ "loss": 0.2902,
+ "step": 17403
+ },
+ {
+ "epoch": 47.42234332425068,
+ "grad_norm": 2.725951671600342,
+ "learning_rate": 1.1316817595054333e-05,
+ "loss": 0.0584,
+ "step": 17404
+ },
+ {
+ "epoch": 47.42506811989101,
+ "grad_norm": 2.580773115158081,
+ "learning_rate": 1.1315942780017589e-05,
+ "loss": 0.1655,
+ "step": 17405
+ },
+ {
+ "epoch": 47.427792915531334,
+ "grad_norm": 2.4359841346740723,
+ "learning_rate": 1.1315067954732322e-05,
+ "loss": 0.2597,
+ "step": 17406
+ },
+ {
+ "epoch": 47.43051771117166,
+ "grad_norm": 2.767242908477783,
+ "learning_rate": 1.131419311920535e-05,
+ "loss": 0.1184,
+ "step": 17407
+ },
+ {
+ "epoch": 47.433242506811986,
+ "grad_norm": 2.5002427101135254,
+ "learning_rate": 1.131331827344348e-05,
+ "loss": 0.2253,
+ "step": 17408
+ },
+ {
+ "epoch": 47.43596730245232,
+ "grad_norm": 2.3485124111175537,
+ "learning_rate": 1.1312443417453535e-05,
+ "loss": 0.0423,
+ "step": 17409
+ },
+ {
+ "epoch": 47.438692098092645,
+ "grad_norm": 2.9458892345428467,
+ "learning_rate": 1.131156855124232e-05,
+ "loss": 0.1087,
+ "step": 17410
+ },
+ {
+ "epoch": 47.44141689373297,
+ "grad_norm": 3.1424405574798584,
+ "learning_rate": 1.1310693674816655e-05,
+ "loss": 0.1374,
+ "step": 17411
+ },
+ {
+ "epoch": 47.444141689373296,
+ "grad_norm": 3.4906973838806152,
+ "learning_rate": 1.1309818788183347e-05,
+ "loss": 0.0827,
+ "step": 17412
+ },
+ {
+ "epoch": 47.44686648501362,
+ "grad_norm": 2.7378768920898438,
+ "learning_rate": 1.1308943891349213e-05,
+ "loss": 0.0907,
+ "step": 17413
+ },
+ {
+ "epoch": 47.44959128065395,
+ "grad_norm": 2.7743594646453857,
+ "learning_rate": 1.1308068984321066e-05,
+ "loss": 0.0795,
+ "step": 17414
+ },
+ {
+ "epoch": 47.45231607629428,
+ "grad_norm": 2.5084009170532227,
+ "learning_rate": 1.130719406710572e-05,
+ "loss": 0.0472,
+ "step": 17415
+ },
+ {
+ "epoch": 47.45504087193461,
+ "grad_norm": 2.8244402408599854,
+ "learning_rate": 1.130631913970999e-05,
+ "loss": 0.2118,
+ "step": 17416
+ },
+ {
+ "epoch": 47.45776566757493,
+ "grad_norm": 2.6146347522735596,
+ "learning_rate": 1.1305444202140685e-05,
+ "loss": 0.1122,
+ "step": 17417
+ },
+ {
+ "epoch": 47.46049046321526,
+ "grad_norm": 2.5417840480804443,
+ "learning_rate": 1.1304569254404623e-05,
+ "loss": 0.0461,
+ "step": 17418
+ },
+ {
+ "epoch": 47.463215258855584,
+ "grad_norm": 2.4873814582824707,
+ "learning_rate": 1.130369429650862e-05,
+ "loss": 0.0511,
+ "step": 17419
+ },
+ {
+ "epoch": 47.46594005449591,
+ "grad_norm": 2.042783737182617,
+ "learning_rate": 1.1302819328459487e-05,
+ "loss": 0.0577,
+ "step": 17420
+ },
+ {
+ "epoch": 47.46866485013624,
+ "grad_norm": 2.9776833057403564,
+ "learning_rate": 1.130194435026404e-05,
+ "loss": 0.1041,
+ "step": 17421
+ },
+ {
+ "epoch": 47.47138964577657,
+ "grad_norm": 1.9004592895507812,
+ "learning_rate": 1.1301069361929092e-05,
+ "loss": 0.038,
+ "step": 17422
+ },
+ {
+ "epoch": 47.474114441416894,
+ "grad_norm": 3.011411666870117,
+ "learning_rate": 1.1300194363461456e-05,
+ "loss": 0.0618,
+ "step": 17423
+ },
+ {
+ "epoch": 47.47683923705722,
+ "grad_norm": 2.6970667839050293,
+ "learning_rate": 1.129931935486795e-05,
+ "loss": 0.0957,
+ "step": 17424
+ },
+ {
+ "epoch": 47.479564032697546,
+ "grad_norm": 2.2647407054901123,
+ "learning_rate": 1.1298444336155388e-05,
+ "loss": 0.0403,
+ "step": 17425
+ },
+ {
+ "epoch": 47.48228882833787,
+ "grad_norm": 3.222248077392578,
+ "learning_rate": 1.1297569307330576e-05,
+ "loss": 0.099,
+ "step": 17426
+ },
+ {
+ "epoch": 47.485013623978205,
+ "grad_norm": 2.9249391555786133,
+ "learning_rate": 1.1296694268400343e-05,
+ "loss": 0.0511,
+ "step": 17427
+ },
+ {
+ "epoch": 47.48773841961853,
+ "grad_norm": 2.9981000423431396,
+ "learning_rate": 1.1295819219371494e-05,
+ "loss": 0.1953,
+ "step": 17428
+ },
+ {
+ "epoch": 47.490463215258856,
+ "grad_norm": 4.262014389038086,
+ "learning_rate": 1.1294944160250849e-05,
+ "loss": 0.272,
+ "step": 17429
+ },
+ {
+ "epoch": 47.49318801089918,
+ "grad_norm": 4.005654335021973,
+ "learning_rate": 1.129406909104522e-05,
+ "loss": 0.0623,
+ "step": 17430
+ },
+ {
+ "epoch": 47.49591280653951,
+ "grad_norm": 2.652388334274292,
+ "learning_rate": 1.1293194011761421e-05,
+ "loss": 0.0963,
+ "step": 17431
+ },
+ {
+ "epoch": 47.49863760217983,
+ "grad_norm": 2.784369468688965,
+ "learning_rate": 1.1292318922406268e-05,
+ "loss": 0.0843,
+ "step": 17432
+ },
+ {
+ "epoch": 47.50136239782017,
+ "grad_norm": 2.6811742782592773,
+ "learning_rate": 1.129144382298658e-05,
+ "loss": 0.1014,
+ "step": 17433
+ },
+ {
+ "epoch": 47.50408719346049,
+ "grad_norm": 2.484971761703491,
+ "learning_rate": 1.1290568713509166e-05,
+ "loss": 0.1256,
+ "step": 17434
+ },
+ {
+ "epoch": 47.50681198910082,
+ "grad_norm": 2.9556312561035156,
+ "learning_rate": 1.1289693593980843e-05,
+ "loss": 0.0629,
+ "step": 17435
+ },
+ {
+ "epoch": 47.509536784741144,
+ "grad_norm": 3.107506513595581,
+ "learning_rate": 1.1288818464408433e-05,
+ "loss": 0.0707,
+ "step": 17436
+ },
+ {
+ "epoch": 47.51226158038147,
+ "grad_norm": 2.7318670749664307,
+ "learning_rate": 1.1287943324798744e-05,
+ "loss": 0.0595,
+ "step": 17437
+ },
+ {
+ "epoch": 47.514986376021795,
+ "grad_norm": 2.461946487426758,
+ "learning_rate": 1.1287068175158594e-05,
+ "loss": 0.0606,
+ "step": 17438
+ },
+ {
+ "epoch": 47.51771117166213,
+ "grad_norm": 3.766435384750366,
+ "learning_rate": 1.1286193015494797e-05,
+ "loss": 0.2851,
+ "step": 17439
+ },
+ {
+ "epoch": 47.520435967302454,
+ "grad_norm": 2.7208516597747803,
+ "learning_rate": 1.128531784581417e-05,
+ "loss": 0.064,
+ "step": 17440
+ },
+ {
+ "epoch": 47.52316076294278,
+ "grad_norm": 2.0805909633636475,
+ "learning_rate": 1.1284442666123532e-05,
+ "loss": 0.0423,
+ "step": 17441
+ },
+ {
+ "epoch": 47.525885558583106,
+ "grad_norm": 2.3465497493743896,
+ "learning_rate": 1.1283567476429693e-05,
+ "loss": 0.0543,
+ "step": 17442
+ },
+ {
+ "epoch": 47.52861035422343,
+ "grad_norm": 2.3623464107513428,
+ "learning_rate": 1.1282692276739473e-05,
+ "loss": 0.0586,
+ "step": 17443
+ },
+ {
+ "epoch": 47.53133514986376,
+ "grad_norm": 2.0605475902557373,
+ "learning_rate": 1.1281817067059689e-05,
+ "loss": 0.0383,
+ "step": 17444
+ },
+ {
+ "epoch": 47.53405994550409,
+ "grad_norm": 2.275054454803467,
+ "learning_rate": 1.128094184739715e-05,
+ "loss": 0.0728,
+ "step": 17445
+ },
+ {
+ "epoch": 47.536784741144416,
+ "grad_norm": 2.888916254043579,
+ "learning_rate": 1.128006661775868e-05,
+ "loss": 0.0777,
+ "step": 17446
+ },
+ {
+ "epoch": 47.53950953678474,
+ "grad_norm": 2.8713924884796143,
+ "learning_rate": 1.1279191378151094e-05,
+ "loss": 0.0525,
+ "step": 17447
+ },
+ {
+ "epoch": 47.54223433242507,
+ "grad_norm": 4.373905181884766,
+ "learning_rate": 1.1278316128581205e-05,
+ "loss": 0.2958,
+ "step": 17448
+ },
+ {
+ "epoch": 47.54495912806539,
+ "grad_norm": 2.2184183597564697,
+ "learning_rate": 1.1277440869055829e-05,
+ "loss": 0.0484,
+ "step": 17449
+ },
+ {
+ "epoch": 47.54768392370572,
+ "grad_norm": 2.825129985809326,
+ "learning_rate": 1.1276565599581786e-05,
+ "loss": 0.101,
+ "step": 17450
+ },
+ {
+ "epoch": 47.55040871934605,
+ "grad_norm": 2.276465654373169,
+ "learning_rate": 1.1275690320165891e-05,
+ "loss": 0.0866,
+ "step": 17451
+ },
+ {
+ "epoch": 47.55313351498638,
+ "grad_norm": 3.3487136363983154,
+ "learning_rate": 1.1274815030814962e-05,
+ "loss": 0.0815,
+ "step": 17452
+ },
+ {
+ "epoch": 47.555858310626704,
+ "grad_norm": 2.99627423286438,
+ "learning_rate": 1.1273939731535814e-05,
+ "loss": 0.0992,
+ "step": 17453
+ },
+ {
+ "epoch": 47.55858310626703,
+ "grad_norm": 2.9918155670166016,
+ "learning_rate": 1.1273064422335266e-05,
+ "loss": 0.1007,
+ "step": 17454
+ },
+ {
+ "epoch": 47.561307901907355,
+ "grad_norm": 2.744769334793091,
+ "learning_rate": 1.127218910322013e-05,
+ "loss": 0.0683,
+ "step": 17455
+ },
+ {
+ "epoch": 47.56403269754768,
+ "grad_norm": 2.5170676708221436,
+ "learning_rate": 1.127131377419723e-05,
+ "loss": 0.0645,
+ "step": 17456
+ },
+ {
+ "epoch": 47.566757493188014,
+ "grad_norm": 2.389218330383301,
+ "learning_rate": 1.1270438435273376e-05,
+ "loss": 0.2167,
+ "step": 17457
+ },
+ {
+ "epoch": 47.56948228882834,
+ "grad_norm": 4.271505355834961,
+ "learning_rate": 1.1269563086455393e-05,
+ "loss": 0.0774,
+ "step": 17458
+ },
+ {
+ "epoch": 47.572207084468666,
+ "grad_norm": 5.25460958480835,
+ "learning_rate": 1.126868772775009e-05,
+ "loss": 0.109,
+ "step": 17459
+ },
+ {
+ "epoch": 47.57493188010899,
+ "grad_norm": 3.006955146789551,
+ "learning_rate": 1.1267812359164286e-05,
+ "loss": 0.058,
+ "step": 17460
+ },
+ {
+ "epoch": 47.57765667574932,
+ "grad_norm": 1.687778353691101,
+ "learning_rate": 1.1266936980704801e-05,
+ "loss": 0.0388,
+ "step": 17461
+ },
+ {
+ "epoch": 47.58038147138964,
+ "grad_norm": 2.8829736709594727,
+ "learning_rate": 1.1266061592378455e-05,
+ "loss": 0.1023,
+ "step": 17462
+ },
+ {
+ "epoch": 47.583106267029976,
+ "grad_norm": 2.391129970550537,
+ "learning_rate": 1.1265186194192058e-05,
+ "loss": 0.0419,
+ "step": 17463
+ },
+ {
+ "epoch": 47.5858310626703,
+ "grad_norm": 2.5367374420166016,
+ "learning_rate": 1.1264310786152434e-05,
+ "loss": 0.0753,
+ "step": 17464
+ },
+ {
+ "epoch": 47.58855585831063,
+ "grad_norm": 2.762000322341919,
+ "learning_rate": 1.1263435368266397e-05,
+ "loss": 0.0929,
+ "step": 17465
+ },
+ {
+ "epoch": 47.59128065395095,
+ "grad_norm": 2.849350929260254,
+ "learning_rate": 1.1262559940540769e-05,
+ "loss": 0.1979,
+ "step": 17466
+ },
+ {
+ "epoch": 47.59400544959128,
+ "grad_norm": 4.2369794845581055,
+ "learning_rate": 1.1261684502982363e-05,
+ "loss": 0.094,
+ "step": 17467
+ },
+ {
+ "epoch": 47.596730245231605,
+ "grad_norm": 3.0777063369750977,
+ "learning_rate": 1.1260809055597996e-05,
+ "loss": 0.218,
+ "step": 17468
+ },
+ {
+ "epoch": 47.59945504087194,
+ "grad_norm": 3.006535530090332,
+ "learning_rate": 1.125993359839449e-05,
+ "loss": 0.0665,
+ "step": 17469
+ },
+ {
+ "epoch": 47.60217983651226,
+ "grad_norm": 4.403693199157715,
+ "learning_rate": 1.1259058131378665e-05,
+ "loss": 0.1,
+ "step": 17470
+ },
+ {
+ "epoch": 47.60490463215259,
+ "grad_norm": 2.545358657836914,
+ "learning_rate": 1.125818265455733e-05,
+ "loss": 0.1289,
+ "step": 17471
+ },
+ {
+ "epoch": 47.607629427792915,
+ "grad_norm": 3.178333282470703,
+ "learning_rate": 1.1257307167937315e-05,
+ "loss": 0.0901,
+ "step": 17472
+ },
+ {
+ "epoch": 47.61035422343324,
+ "grad_norm": 2.213599920272827,
+ "learning_rate": 1.1256431671525428e-05,
+ "loss": 0.0752,
+ "step": 17473
+ },
+ {
+ "epoch": 47.61307901907357,
+ "grad_norm": 3.2589449882507324,
+ "learning_rate": 1.1255556165328494e-05,
+ "loss": 0.0916,
+ "step": 17474
+ },
+ {
+ "epoch": 47.6158038147139,
+ "grad_norm": 1.9802409410476685,
+ "learning_rate": 1.125468064935333e-05,
+ "loss": 0.0378,
+ "step": 17475
+ },
+ {
+ "epoch": 47.618528610354225,
+ "grad_norm": 3.6562583446502686,
+ "learning_rate": 1.1253805123606752e-05,
+ "loss": 0.0925,
+ "step": 17476
+ },
+ {
+ "epoch": 47.62125340599455,
+ "grad_norm": 2.295644998550415,
+ "learning_rate": 1.1252929588095576e-05,
+ "loss": 0.0478,
+ "step": 17477
+ },
+ {
+ "epoch": 47.62397820163488,
+ "grad_norm": 3.1513900756835938,
+ "learning_rate": 1.1252054042826627e-05,
+ "loss": 0.182,
+ "step": 17478
+ },
+ {
+ "epoch": 47.6267029972752,
+ "grad_norm": 2.6366379261016846,
+ "learning_rate": 1.1251178487806725e-05,
+ "loss": 0.2133,
+ "step": 17479
+ },
+ {
+ "epoch": 47.62942779291553,
+ "grad_norm": 2.661428928375244,
+ "learning_rate": 1.1250302923042684e-05,
+ "loss": 0.0498,
+ "step": 17480
+ },
+ {
+ "epoch": 47.63215258855586,
+ "grad_norm": 2.834806442260742,
+ "learning_rate": 1.1249427348541322e-05,
+ "loss": 0.042,
+ "step": 17481
+ },
+ {
+ "epoch": 47.63487738419619,
+ "grad_norm": 2.473759651184082,
+ "learning_rate": 1.124855176430946e-05,
+ "loss": 0.04,
+ "step": 17482
+ },
+ {
+ "epoch": 47.63760217983651,
+ "grad_norm": 3.1461997032165527,
+ "learning_rate": 1.1247676170353918e-05,
+ "loss": 0.1049,
+ "step": 17483
+ },
+ {
+ "epoch": 47.64032697547684,
+ "grad_norm": 2.7429757118225098,
+ "learning_rate": 1.1246800566681517e-05,
+ "loss": 0.0789,
+ "step": 17484
+ },
+ {
+ "epoch": 47.643051771117165,
+ "grad_norm": 4.321488857269287,
+ "learning_rate": 1.124592495329907e-05,
+ "loss": 0.1065,
+ "step": 17485
+ },
+ {
+ "epoch": 47.64577656675749,
+ "grad_norm": 2.774259090423584,
+ "learning_rate": 1.1245049330213398e-05,
+ "loss": 0.0669,
+ "step": 17486
+ },
+ {
+ "epoch": 47.64850136239782,
+ "grad_norm": 3.3800675868988037,
+ "learning_rate": 1.1244173697431324e-05,
+ "loss": 0.1262,
+ "step": 17487
+ },
+ {
+ "epoch": 47.65122615803815,
+ "grad_norm": 3.013646364212036,
+ "learning_rate": 1.1243298054959666e-05,
+ "loss": 0.1557,
+ "step": 17488
+ },
+ {
+ "epoch": 47.653950953678475,
+ "grad_norm": 2.9568490982055664,
+ "learning_rate": 1.1242422402805242e-05,
+ "loss": 0.1045,
+ "step": 17489
+ },
+ {
+ "epoch": 47.6566757493188,
+ "grad_norm": 2.1368658542633057,
+ "learning_rate": 1.1241546740974874e-05,
+ "loss": 0.0541,
+ "step": 17490
+ },
+ {
+ "epoch": 47.65940054495913,
+ "grad_norm": 4.297214984893799,
+ "learning_rate": 1.1240671069475377e-05,
+ "loss": 0.1496,
+ "step": 17491
+ },
+ {
+ "epoch": 47.66212534059945,
+ "grad_norm": 2.4155001640319824,
+ "learning_rate": 1.1239795388313578e-05,
+ "loss": 0.0727,
+ "step": 17492
+ },
+ {
+ "epoch": 47.664850136239785,
+ "grad_norm": 3.162046432495117,
+ "learning_rate": 1.1238919697496287e-05,
+ "loss": 0.0989,
+ "step": 17493
+ },
+ {
+ "epoch": 47.66757493188011,
+ "grad_norm": 3.3390300273895264,
+ "learning_rate": 1.123804399703033e-05,
+ "loss": 0.0842,
+ "step": 17494
+ },
+ {
+ "epoch": 47.67029972752044,
+ "grad_norm": 1.5416027307510376,
+ "learning_rate": 1.1237168286922528e-05,
+ "loss": 0.036,
+ "step": 17495
+ },
+ {
+ "epoch": 47.67302452316076,
+ "grad_norm": 2.5452752113342285,
+ "learning_rate": 1.1236292567179702e-05,
+ "loss": 0.1225,
+ "step": 17496
+ },
+ {
+ "epoch": 47.67574931880109,
+ "grad_norm": 2.647254228591919,
+ "learning_rate": 1.1235416837808663e-05,
+ "loss": 0.0568,
+ "step": 17497
+ },
+ {
+ "epoch": 47.678474114441414,
+ "grad_norm": 2.9616472721099854,
+ "learning_rate": 1.1234541098816244e-05,
+ "loss": 0.0527,
+ "step": 17498
+ },
+ {
+ "epoch": 47.68119891008175,
+ "grad_norm": 3.8103203773498535,
+ "learning_rate": 1.1233665350209253e-05,
+ "loss": 0.0858,
+ "step": 17499
+ },
+ {
+ "epoch": 47.68392370572207,
+ "grad_norm": 2.931959867477417,
+ "learning_rate": 1.123278959199452e-05,
+ "loss": 0.0458,
+ "step": 17500
+ },
+ {
+ "epoch": 47.6866485013624,
+ "grad_norm": 2.277695894241333,
+ "learning_rate": 1.1231913824178859e-05,
+ "loss": 0.0482,
+ "step": 17501
+ },
+ {
+ "epoch": 47.689373297002724,
+ "grad_norm": 2.943812131881714,
+ "learning_rate": 1.1231038046769094e-05,
+ "loss": 0.108,
+ "step": 17502
+ },
+ {
+ "epoch": 47.69209809264305,
+ "grad_norm": 2.432861566543579,
+ "learning_rate": 1.1230162259772041e-05,
+ "loss": 0.0901,
+ "step": 17503
+ },
+ {
+ "epoch": 47.694822888283376,
+ "grad_norm": 2.923929452896118,
+ "learning_rate": 1.1229286463194529e-05,
+ "loss": 0.0774,
+ "step": 17504
+ },
+ {
+ "epoch": 47.69754768392371,
+ "grad_norm": 2.279658794403076,
+ "learning_rate": 1.1228410657043369e-05,
+ "loss": 0.0537,
+ "step": 17505
+ },
+ {
+ "epoch": 47.700272479564035,
+ "grad_norm": 2.510606050491333,
+ "learning_rate": 1.1227534841325392e-05,
+ "loss": 0.0884,
+ "step": 17506
+ },
+ {
+ "epoch": 47.70299727520436,
+ "grad_norm": 2.7028608322143555,
+ "learning_rate": 1.1226659016047408e-05,
+ "loss": 0.0949,
+ "step": 17507
+ },
+ {
+ "epoch": 47.705722070844686,
+ "grad_norm": 2.968425750732422,
+ "learning_rate": 1.1225783181216245e-05,
+ "loss": 0.0901,
+ "step": 17508
+ },
+ {
+ "epoch": 47.70844686648501,
+ "grad_norm": 2.7960586547851562,
+ "learning_rate": 1.1224907336838722e-05,
+ "loss": 0.1996,
+ "step": 17509
+ },
+ {
+ "epoch": 47.71117166212534,
+ "grad_norm": 3.2108068466186523,
+ "learning_rate": 1.1224031482921662e-05,
+ "loss": 0.0479,
+ "step": 17510
+ },
+ {
+ "epoch": 47.71389645776567,
+ "grad_norm": 2.5051069259643555,
+ "learning_rate": 1.122315561947188e-05,
+ "loss": 0.0752,
+ "step": 17511
+ },
+ {
+ "epoch": 47.716621253406,
+ "grad_norm": 4.027047634124756,
+ "learning_rate": 1.1222279746496203e-05,
+ "loss": 0.0452,
+ "step": 17512
+ },
+ {
+ "epoch": 47.71934604904632,
+ "grad_norm": 3.1387217044830322,
+ "learning_rate": 1.1221403864001452e-05,
+ "loss": 0.0814,
+ "step": 17513
+ },
+ {
+ "epoch": 47.72207084468665,
+ "grad_norm": 3.1164121627807617,
+ "learning_rate": 1.1220527971994447e-05,
+ "loss": 0.2027,
+ "step": 17514
+ },
+ {
+ "epoch": 47.724795640326974,
+ "grad_norm": 2.7821836471557617,
+ "learning_rate": 1.1219652070482007e-05,
+ "loss": 0.0874,
+ "step": 17515
+ },
+ {
+ "epoch": 47.7275204359673,
+ "grad_norm": 2.6584253311157227,
+ "learning_rate": 1.1218776159470962e-05,
+ "loss": 0.0944,
+ "step": 17516
+ },
+ {
+ "epoch": 47.73024523160763,
+ "grad_norm": 3.583146333694458,
+ "learning_rate": 1.121790023896812e-05,
+ "loss": 0.1921,
+ "step": 17517
+ },
+ {
+ "epoch": 47.73297002724796,
+ "grad_norm": 2.0450379848480225,
+ "learning_rate": 1.1217024308980316e-05,
+ "loss": 0.0405,
+ "step": 17518
+ },
+ {
+ "epoch": 47.735694822888284,
+ "grad_norm": 2.88075852394104,
+ "learning_rate": 1.1216148369514361e-05,
+ "loss": 0.1122,
+ "step": 17519
+ },
+ {
+ "epoch": 47.73841961852861,
+ "grad_norm": 2.3719606399536133,
+ "learning_rate": 1.1215272420577084e-05,
+ "loss": 0.0525,
+ "step": 17520
+ },
+ {
+ "epoch": 47.741144414168936,
+ "grad_norm": 2.26948618888855,
+ "learning_rate": 1.1214396462175306e-05,
+ "loss": 0.143,
+ "step": 17521
+ },
+ {
+ "epoch": 47.74386920980926,
+ "grad_norm": 3.0690248012542725,
+ "learning_rate": 1.1213520494315844e-05,
+ "loss": 0.0568,
+ "step": 17522
+ },
+ {
+ "epoch": 47.746594005449595,
+ "grad_norm": 3.0862627029418945,
+ "learning_rate": 1.1212644517005527e-05,
+ "loss": 0.0554,
+ "step": 17523
+ },
+ {
+ "epoch": 47.74931880108992,
+ "grad_norm": 2.4806926250457764,
+ "learning_rate": 1.121176853025117e-05,
+ "loss": 0.0468,
+ "step": 17524
+ },
+ {
+ "epoch": 47.752043596730246,
+ "grad_norm": 2.8695473670959473,
+ "learning_rate": 1.1210892534059599e-05,
+ "loss": 0.1505,
+ "step": 17525
+ },
+ {
+ "epoch": 47.75476839237057,
+ "grad_norm": 3.4686691761016846,
+ "learning_rate": 1.121001652843764e-05,
+ "loss": 0.1894,
+ "step": 17526
+ },
+ {
+ "epoch": 47.7574931880109,
+ "grad_norm": 2.2075233459472656,
+ "learning_rate": 1.1209140513392108e-05,
+ "loss": 0.1495,
+ "step": 17527
+ },
+ {
+ "epoch": 47.76021798365122,
+ "grad_norm": 3.3527071475982666,
+ "learning_rate": 1.120826448892983e-05,
+ "loss": 0.2109,
+ "step": 17528
+ },
+ {
+ "epoch": 47.762942779291556,
+ "grad_norm": 2.6841094493865967,
+ "learning_rate": 1.1207388455057623e-05,
+ "loss": 0.0661,
+ "step": 17529
+ },
+ {
+ "epoch": 47.76566757493188,
+ "grad_norm": 2.7696077823638916,
+ "learning_rate": 1.1206512411782318e-05,
+ "loss": 0.1084,
+ "step": 17530
+ },
+ {
+ "epoch": 47.76839237057221,
+ "grad_norm": 2.9208016395568848,
+ "learning_rate": 1.1205636359110733e-05,
+ "loss": 0.0929,
+ "step": 17531
+ },
+ {
+ "epoch": 47.771117166212534,
+ "grad_norm": 2.309746742248535,
+ "learning_rate": 1.1204760297049688e-05,
+ "loss": 0.1116,
+ "step": 17532
+ },
+ {
+ "epoch": 47.77384196185286,
+ "grad_norm": 2.5888595581054688,
+ "learning_rate": 1.1203884225606012e-05,
+ "loss": 0.0714,
+ "step": 17533
+ },
+ {
+ "epoch": 47.776566757493185,
+ "grad_norm": 3.504862070083618,
+ "learning_rate": 1.1203008144786523e-05,
+ "loss": 0.0683,
+ "step": 17534
+ },
+ {
+ "epoch": 47.77929155313352,
+ "grad_norm": 1.9906659126281738,
+ "learning_rate": 1.1202132054598045e-05,
+ "loss": 0.0386,
+ "step": 17535
+ },
+ {
+ "epoch": 47.782016348773844,
+ "grad_norm": 2.2432546615600586,
+ "learning_rate": 1.1201255955047401e-05,
+ "loss": 0.0509,
+ "step": 17536
+ },
+ {
+ "epoch": 47.78474114441417,
+ "grad_norm": 2.1952035427093506,
+ "learning_rate": 1.1200379846141414e-05,
+ "loss": 0.097,
+ "step": 17537
+ },
+ {
+ "epoch": 47.787465940054496,
+ "grad_norm": 3.3117876052856445,
+ "learning_rate": 1.1199503727886906e-05,
+ "loss": 0.0912,
+ "step": 17538
+ },
+ {
+ "epoch": 47.79019073569482,
+ "grad_norm": 3.573143482208252,
+ "learning_rate": 1.1198627600290705e-05,
+ "loss": 0.0987,
+ "step": 17539
+ },
+ {
+ "epoch": 47.79291553133515,
+ "grad_norm": 2.2487454414367676,
+ "learning_rate": 1.119775146335963e-05,
+ "loss": 0.0527,
+ "step": 17540
+ },
+ {
+ "epoch": 47.79564032697548,
+ "grad_norm": 2.7377703189849854,
+ "learning_rate": 1.1196875317100507e-05,
+ "loss": 0.2207,
+ "step": 17541
+ },
+ {
+ "epoch": 47.798365122615806,
+ "grad_norm": 3.0041089057922363,
+ "learning_rate": 1.1195999161520152e-05,
+ "loss": 0.093,
+ "step": 17542
+ },
+ {
+ "epoch": 47.80108991825613,
+ "grad_norm": 2.0277180671691895,
+ "learning_rate": 1.1195122996625399e-05,
+ "loss": 0.038,
+ "step": 17543
+ },
+ {
+ "epoch": 47.80381471389646,
+ "grad_norm": 2.6679916381835938,
+ "learning_rate": 1.1194246822423063e-05,
+ "loss": 0.0478,
+ "step": 17544
+ },
+ {
+ "epoch": 47.80653950953678,
+ "grad_norm": 2.201544761657715,
+ "learning_rate": 1.1193370638919976e-05,
+ "loss": 0.08,
+ "step": 17545
+ },
+ {
+ "epoch": 47.80926430517711,
+ "grad_norm": 3.2336440086364746,
+ "learning_rate": 1.119249444612295e-05,
+ "loss": 0.1063,
+ "step": 17546
+ },
+ {
+ "epoch": 47.81198910081744,
+ "grad_norm": 2.636674165725708,
+ "learning_rate": 1.119161824403882e-05,
+ "loss": 0.0737,
+ "step": 17547
+ },
+ {
+ "epoch": 47.81471389645777,
+ "grad_norm": 2.1710410118103027,
+ "learning_rate": 1.1190742032674404e-05,
+ "loss": 0.0793,
+ "step": 17548
+ },
+ {
+ "epoch": 47.817438692098094,
+ "grad_norm": 2.7095329761505127,
+ "learning_rate": 1.118986581203653e-05,
+ "loss": 0.1282,
+ "step": 17549
+ },
+ {
+ "epoch": 47.82016348773842,
+ "grad_norm": 2.8484389781951904,
+ "learning_rate": 1.1188989582132016e-05,
+ "loss": 0.0617,
+ "step": 17550
+ },
+ {
+ "epoch": 47.822888283378745,
+ "grad_norm": 2.3287220001220703,
+ "learning_rate": 1.1188113342967692e-05,
+ "loss": 0.1591,
+ "step": 17551
+ },
+ {
+ "epoch": 47.82561307901907,
+ "grad_norm": 2.6710386276245117,
+ "learning_rate": 1.1187237094550378e-05,
+ "loss": 0.0731,
+ "step": 17552
+ },
+ {
+ "epoch": 47.828337874659404,
+ "grad_norm": 2.3600080013275146,
+ "learning_rate": 1.1186360836886903e-05,
+ "loss": 0.0739,
+ "step": 17553
+ },
+ {
+ "epoch": 47.83106267029973,
+ "grad_norm": 1.9637635946273804,
+ "learning_rate": 1.1185484569984082e-05,
+ "loss": 0.0294,
+ "step": 17554
+ },
+ {
+ "epoch": 47.833787465940055,
+ "grad_norm": 4.005996227264404,
+ "learning_rate": 1.1184608293848748e-05,
+ "loss": 0.1082,
+ "step": 17555
+ },
+ {
+ "epoch": 47.83651226158038,
+ "grad_norm": 2.2963595390319824,
+ "learning_rate": 1.1183732008487723e-05,
+ "loss": 0.056,
+ "step": 17556
+ },
+ {
+ "epoch": 47.83923705722071,
+ "grad_norm": 2.4077768325805664,
+ "learning_rate": 1.1182855713907831e-05,
+ "loss": 0.0638,
+ "step": 17557
+ },
+ {
+ "epoch": 47.84196185286103,
+ "grad_norm": 3.0728704929351807,
+ "learning_rate": 1.1181979410115897e-05,
+ "loss": 0.1252,
+ "step": 17558
+ },
+ {
+ "epoch": 47.844686648501366,
+ "grad_norm": 2.3760573863983154,
+ "learning_rate": 1.1181103097118748e-05,
+ "loss": 0.1075,
+ "step": 17559
+ },
+ {
+ "epoch": 47.84741144414169,
+ "grad_norm": 3.1397500038146973,
+ "learning_rate": 1.1180226774923204e-05,
+ "loss": 0.1517,
+ "step": 17560
+ },
+ {
+ "epoch": 47.85013623978202,
+ "grad_norm": 2.8059825897216797,
+ "learning_rate": 1.1179350443536092e-05,
+ "loss": 0.1372,
+ "step": 17561
+ },
+ {
+ "epoch": 47.85286103542234,
+ "grad_norm": 2.592531681060791,
+ "learning_rate": 1.1178474102964236e-05,
+ "loss": 0.0747,
+ "step": 17562
+ },
+ {
+ "epoch": 47.85558583106267,
+ "grad_norm": 3.044501304626465,
+ "learning_rate": 1.117759775321446e-05,
+ "loss": 0.0705,
+ "step": 17563
+ },
+ {
+ "epoch": 47.858310626702995,
+ "grad_norm": 2.167360305786133,
+ "learning_rate": 1.1176721394293593e-05,
+ "loss": 0.0426,
+ "step": 17564
+ },
+ {
+ "epoch": 47.86103542234333,
+ "grad_norm": 3.274238109588623,
+ "learning_rate": 1.1175845026208458e-05,
+ "loss": 0.1494,
+ "step": 17565
+ },
+ {
+ "epoch": 47.86376021798365,
+ "grad_norm": 2.6146585941314697,
+ "learning_rate": 1.1174968648965878e-05,
+ "loss": 0.0753,
+ "step": 17566
+ },
+ {
+ "epoch": 47.86648501362398,
+ "grad_norm": 3.11510968208313,
+ "learning_rate": 1.1174092262572682e-05,
+ "loss": 0.1171,
+ "step": 17567
+ },
+ {
+ "epoch": 47.869209809264305,
+ "grad_norm": 2.624591827392578,
+ "learning_rate": 1.1173215867035692e-05,
+ "loss": 0.0374,
+ "step": 17568
+ },
+ {
+ "epoch": 47.87193460490463,
+ "grad_norm": 3.0111894607543945,
+ "learning_rate": 1.1172339462361735e-05,
+ "loss": 0.0614,
+ "step": 17569
+ },
+ {
+ "epoch": 47.87465940054496,
+ "grad_norm": 3.3964011669158936,
+ "learning_rate": 1.1171463048557636e-05,
+ "loss": 0.0752,
+ "step": 17570
+ },
+ {
+ "epoch": 47.87738419618529,
+ "grad_norm": 3.0458803176879883,
+ "learning_rate": 1.117058662563022e-05,
+ "loss": 0.0719,
+ "step": 17571
+ },
+ {
+ "epoch": 47.880108991825615,
+ "grad_norm": 3.257497549057007,
+ "learning_rate": 1.1169710193586312e-05,
+ "loss": 0.0808,
+ "step": 17572
+ },
+ {
+ "epoch": 47.88283378746594,
+ "grad_norm": 2.6234142780303955,
+ "learning_rate": 1.1168833752432739e-05,
+ "loss": 0.0442,
+ "step": 17573
+ },
+ {
+ "epoch": 47.88555858310627,
+ "grad_norm": 2.6439101696014404,
+ "learning_rate": 1.1167957302176327e-05,
+ "loss": 0.0739,
+ "step": 17574
+ },
+ {
+ "epoch": 47.88828337874659,
+ "grad_norm": 2.9201455116271973,
+ "learning_rate": 1.1167080842823904e-05,
+ "loss": 0.1135,
+ "step": 17575
+ },
+ {
+ "epoch": 47.89100817438692,
+ "grad_norm": 3.0905673503875732,
+ "learning_rate": 1.1166204374382287e-05,
+ "loss": 0.1206,
+ "step": 17576
+ },
+ {
+ "epoch": 47.89373297002725,
+ "grad_norm": 2.903468370437622,
+ "learning_rate": 1.1165327896858314e-05,
+ "loss": 0.1566,
+ "step": 17577
+ },
+ {
+ "epoch": 47.89645776566758,
+ "grad_norm": 2.847524404525757,
+ "learning_rate": 1.1164451410258801e-05,
+ "loss": 0.1864,
+ "step": 17578
+ },
+ {
+ "epoch": 47.8991825613079,
+ "grad_norm": 2.541430711746216,
+ "learning_rate": 1.116357491459058e-05,
+ "loss": 0.0568,
+ "step": 17579
+ },
+ {
+ "epoch": 47.90190735694823,
+ "grad_norm": 2.596785068511963,
+ "learning_rate": 1.1162698409860471e-05,
+ "loss": 0.1223,
+ "step": 17580
+ },
+ {
+ "epoch": 47.904632152588555,
+ "grad_norm": 2.228508472442627,
+ "learning_rate": 1.1161821896075309e-05,
+ "loss": 0.0562,
+ "step": 17581
+ },
+ {
+ "epoch": 47.90735694822888,
+ "grad_norm": 2.1208534240722656,
+ "learning_rate": 1.1160945373241914e-05,
+ "loss": 0.0647,
+ "step": 17582
+ },
+ {
+ "epoch": 47.91008174386921,
+ "grad_norm": 2.277470350265503,
+ "learning_rate": 1.1160068841367113e-05,
+ "loss": 0.057,
+ "step": 17583
+ },
+ {
+ "epoch": 47.91280653950954,
+ "grad_norm": 3.4231932163238525,
+ "learning_rate": 1.1159192300457733e-05,
+ "loss": 0.0976,
+ "step": 17584
+ },
+ {
+ "epoch": 47.915531335149865,
+ "grad_norm": 2.5316081047058105,
+ "learning_rate": 1.1158315750520603e-05,
+ "loss": 0.0892,
+ "step": 17585
+ },
+ {
+ "epoch": 47.91825613079019,
+ "grad_norm": 2.3603148460388184,
+ "learning_rate": 1.1157439191562545e-05,
+ "loss": 0.0359,
+ "step": 17586
+ },
+ {
+ "epoch": 47.920980926430516,
+ "grad_norm": 2.3762567043304443,
+ "learning_rate": 1.1156562623590391e-05,
+ "loss": 0.1704,
+ "step": 17587
+ },
+ {
+ "epoch": 47.92370572207084,
+ "grad_norm": 2.6970038414001465,
+ "learning_rate": 1.1155686046610958e-05,
+ "loss": 0.0466,
+ "step": 17588
+ },
+ {
+ "epoch": 47.926430517711175,
+ "grad_norm": 4.565173625946045,
+ "learning_rate": 1.1154809460631084e-05,
+ "loss": 0.0518,
+ "step": 17589
+ },
+ {
+ "epoch": 47.9291553133515,
+ "grad_norm": 2.0279719829559326,
+ "learning_rate": 1.1153932865657588e-05,
+ "loss": 0.0401,
+ "step": 17590
+ },
+ {
+ "epoch": 47.93188010899183,
+ "grad_norm": 2.665750026702881,
+ "learning_rate": 1.1153056261697303e-05,
+ "loss": 0.2282,
+ "step": 17591
+ },
+ {
+ "epoch": 47.93460490463215,
+ "grad_norm": 2.643846035003662,
+ "learning_rate": 1.1152179648757053e-05,
+ "loss": 0.1155,
+ "step": 17592
+ },
+ {
+ "epoch": 47.93732970027248,
+ "grad_norm": 2.392054319381714,
+ "learning_rate": 1.1151303026843664e-05,
+ "loss": 0.1791,
+ "step": 17593
+ },
+ {
+ "epoch": 47.940054495912804,
+ "grad_norm": 2.1424243450164795,
+ "learning_rate": 1.1150426395963964e-05,
+ "loss": 0.0457,
+ "step": 17594
+ },
+ {
+ "epoch": 47.94277929155314,
+ "grad_norm": 2.9089035987854004,
+ "learning_rate": 1.1149549756124782e-05,
+ "loss": 0.0727,
+ "step": 17595
+ },
+ {
+ "epoch": 47.94550408719346,
+ "grad_norm": 2.9449031352996826,
+ "learning_rate": 1.1148673107332942e-05,
+ "loss": 0.0925,
+ "step": 17596
+ },
+ {
+ "epoch": 47.94822888283379,
+ "grad_norm": 2.8393332958221436,
+ "learning_rate": 1.1147796449595274e-05,
+ "loss": 0.1344,
+ "step": 17597
+ },
+ {
+ "epoch": 47.950953678474114,
+ "grad_norm": 3.3989670276641846,
+ "learning_rate": 1.1146919782918601e-05,
+ "loss": 0.0866,
+ "step": 17598
+ },
+ {
+ "epoch": 47.95367847411444,
+ "grad_norm": 3.287322998046875,
+ "learning_rate": 1.1146043107309754e-05,
+ "loss": 0.081,
+ "step": 17599
+ },
+ {
+ "epoch": 47.956403269754766,
+ "grad_norm": 3.1032190322875977,
+ "learning_rate": 1.1145166422775562e-05,
+ "loss": 0.1813,
+ "step": 17600
+ },
+ {
+ "epoch": 47.95912806539509,
+ "grad_norm": 1.8288978338241577,
+ "learning_rate": 1.1144289729322853e-05,
+ "loss": 0.1559,
+ "step": 17601
+ },
+ {
+ "epoch": 47.961852861035425,
+ "grad_norm": 2.5142335891723633,
+ "learning_rate": 1.1143413026958448e-05,
+ "loss": 0.0425,
+ "step": 17602
+ },
+ {
+ "epoch": 47.96457765667575,
+ "grad_norm": 2.694533348083496,
+ "learning_rate": 1.1142536315689183e-05,
+ "loss": 0.0899,
+ "step": 17603
+ },
+ {
+ "epoch": 47.967302452316076,
+ "grad_norm": 2.9677228927612305,
+ "learning_rate": 1.1141659595521879e-05,
+ "loss": 0.0538,
+ "step": 17604
+ },
+ {
+ "epoch": 47.9700272479564,
+ "grad_norm": 2.9357500076293945,
+ "learning_rate": 1.1140782866463369e-05,
+ "loss": 0.057,
+ "step": 17605
+ },
+ {
+ "epoch": 47.97275204359673,
+ "grad_norm": 2.619701623916626,
+ "learning_rate": 1.1139906128520476e-05,
+ "loss": 0.0781,
+ "step": 17606
+ },
+ {
+ "epoch": 47.97547683923706,
+ "grad_norm": 2.9285197257995605,
+ "learning_rate": 1.1139029381700033e-05,
+ "loss": 0.0716,
+ "step": 17607
+ },
+ {
+ "epoch": 47.97820163487739,
+ "grad_norm": 2.4408817291259766,
+ "learning_rate": 1.1138152626008864e-05,
+ "loss": 0.1272,
+ "step": 17608
+ },
+ {
+ "epoch": 47.98092643051771,
+ "grad_norm": 2.393932819366455,
+ "learning_rate": 1.1137275861453801e-05,
+ "loss": 0.0882,
+ "step": 17609
+ },
+ {
+ "epoch": 47.98365122615804,
+ "grad_norm": 3.1361756324768066,
+ "learning_rate": 1.113639908804167e-05,
+ "loss": 0.0598,
+ "step": 17610
+ },
+ {
+ "epoch": 47.986376021798364,
+ "grad_norm": 3.0295157432556152,
+ "learning_rate": 1.1135522305779298e-05,
+ "loss": 0.1286,
+ "step": 17611
+ },
+ {
+ "epoch": 47.98910081743869,
+ "grad_norm": 3.1915509700775146,
+ "learning_rate": 1.1134645514673514e-05,
+ "loss": 0.3047,
+ "step": 17612
+ },
+ {
+ "epoch": 47.991825613079016,
+ "grad_norm": 1.9568201303482056,
+ "learning_rate": 1.1133768714731152e-05,
+ "loss": 0.0392,
+ "step": 17613
+ },
+ {
+ "epoch": 47.99455040871935,
+ "grad_norm": 3.591489553451538,
+ "learning_rate": 1.113289190595903e-05,
+ "loss": 0.0708,
+ "step": 17614
+ },
+ {
+ "epoch": 47.997275204359674,
+ "grad_norm": 2.7456860542297363,
+ "learning_rate": 1.1132015088363985e-05,
+ "loss": 0.1623,
+ "step": 17615
+ },
+ {
+ "epoch": 48.0,
+ "grad_norm": 2.9107398986816406,
+ "learning_rate": 1.1131138261952845e-05,
+ "loss": 0.0448,
+ "step": 17616
+ },
+ {
+ "epoch": 48.002724795640326,
+ "grad_norm": 2.903590202331543,
+ "learning_rate": 1.1130261426732433e-05,
+ "loss": 0.1717,
+ "step": 17617
+ },
+ {
+ "epoch": 48.00544959128065,
+ "grad_norm": 2.7714970111846924,
+ "learning_rate": 1.1129384582709584e-05,
+ "loss": 0.1366,
+ "step": 17618
+ },
+ {
+ "epoch": 48.00817438692098,
+ "grad_norm": 2.7199482917785645,
+ "learning_rate": 1.1128507729891123e-05,
+ "loss": 0.0434,
+ "step": 17619
+ },
+ {
+ "epoch": 48.01089918256131,
+ "grad_norm": 3.194709062576294,
+ "learning_rate": 1.1127630868283881e-05,
+ "loss": 0.1311,
+ "step": 17620
+ },
+ {
+ "epoch": 48.013623978201636,
+ "grad_norm": 2.931918144226074,
+ "learning_rate": 1.1126753997894683e-05,
+ "loss": 0.2326,
+ "step": 17621
+ },
+ {
+ "epoch": 48.01634877384196,
+ "grad_norm": 2.3366427421569824,
+ "learning_rate": 1.1125877118730363e-05,
+ "loss": 0.0751,
+ "step": 17622
+ },
+ {
+ "epoch": 48.01907356948229,
+ "grad_norm": 2.2673075199127197,
+ "learning_rate": 1.112500023079775e-05,
+ "loss": 0.1368,
+ "step": 17623
+ },
+ {
+ "epoch": 48.02179836512261,
+ "grad_norm": 3.2362582683563232,
+ "learning_rate": 1.1124123334103671e-05,
+ "loss": 0.0662,
+ "step": 17624
+ },
+ {
+ "epoch": 48.02452316076294,
+ "grad_norm": 2.1153931617736816,
+ "learning_rate": 1.1123246428654953e-05,
+ "loss": 0.0582,
+ "step": 17625
+ },
+ {
+ "epoch": 48.02724795640327,
+ "grad_norm": 2.253751754760742,
+ "learning_rate": 1.1122369514458431e-05,
+ "loss": 0.1488,
+ "step": 17626
+ },
+ {
+ "epoch": 48.0299727520436,
+ "grad_norm": 2.4317288398742676,
+ "learning_rate": 1.112149259152093e-05,
+ "loss": 0.0437,
+ "step": 17627
+ },
+ {
+ "epoch": 48.032697547683924,
+ "grad_norm": 2.394043207168579,
+ "learning_rate": 1.112061565984928e-05,
+ "loss": 0.056,
+ "step": 17628
+ },
+ {
+ "epoch": 48.03542234332425,
+ "grad_norm": 2.7741312980651855,
+ "learning_rate": 1.1119738719450312e-05,
+ "loss": 0.1503,
+ "step": 17629
+ },
+ {
+ "epoch": 48.038147138964575,
+ "grad_norm": 2.9651966094970703,
+ "learning_rate": 1.1118861770330857e-05,
+ "loss": 0.076,
+ "step": 17630
+ },
+ {
+ "epoch": 48.0408719346049,
+ "grad_norm": 2.7154738903045654,
+ "learning_rate": 1.1117984812497739e-05,
+ "loss": 0.0772,
+ "step": 17631
+ },
+ {
+ "epoch": 48.043596730245234,
+ "grad_norm": 2.314377784729004,
+ "learning_rate": 1.1117107845957793e-05,
+ "loss": 0.0614,
+ "step": 17632
+ },
+ {
+ "epoch": 48.04632152588556,
+ "grad_norm": 2.4928646087646484,
+ "learning_rate": 1.1116230870717846e-05,
+ "loss": 0.1722,
+ "step": 17633
+ },
+ {
+ "epoch": 48.049046321525886,
+ "grad_norm": 2.9796788692474365,
+ "learning_rate": 1.1115353886784732e-05,
+ "loss": 0.1059,
+ "step": 17634
+ },
+ {
+ "epoch": 48.05177111716621,
+ "grad_norm": 2.420727491378784,
+ "learning_rate": 1.1114476894165273e-05,
+ "loss": 0.1043,
+ "step": 17635
+ },
+ {
+ "epoch": 48.05449591280654,
+ "grad_norm": 5.780261039733887,
+ "learning_rate": 1.1113599892866309e-05,
+ "loss": 0.0456,
+ "step": 17636
+ },
+ {
+ "epoch": 48.05722070844686,
+ "grad_norm": 2.543757677078247,
+ "learning_rate": 1.1112722882894662e-05,
+ "loss": 0.0771,
+ "step": 17637
+ },
+ {
+ "epoch": 48.059945504087196,
+ "grad_norm": 2.0290160179138184,
+ "learning_rate": 1.1111845864257166e-05,
+ "loss": 0.0456,
+ "step": 17638
+ },
+ {
+ "epoch": 48.06267029972752,
+ "grad_norm": 2.1223647594451904,
+ "learning_rate": 1.111096883696065e-05,
+ "loss": 0.1813,
+ "step": 17639
+ },
+ {
+ "epoch": 48.06539509536785,
+ "grad_norm": 2.831057071685791,
+ "learning_rate": 1.1110091801011944e-05,
+ "loss": 0.0771,
+ "step": 17640
+ },
+ {
+ "epoch": 48.06811989100817,
+ "grad_norm": 2.410842180252075,
+ "learning_rate": 1.1109214756417876e-05,
+ "loss": 0.055,
+ "step": 17641
+ },
+ {
+ "epoch": 48.0708446866485,
+ "grad_norm": 2.486776351928711,
+ "learning_rate": 1.1108337703185283e-05,
+ "loss": 0.1497,
+ "step": 17642
+ },
+ {
+ "epoch": 48.073569482288825,
+ "grad_norm": 2.0264265537261963,
+ "learning_rate": 1.1107460641320991e-05,
+ "loss": 0.0445,
+ "step": 17643
+ },
+ {
+ "epoch": 48.07629427792916,
+ "grad_norm": 2.6673502922058105,
+ "learning_rate": 1.110658357083183e-05,
+ "loss": 0.0776,
+ "step": 17644
+ },
+ {
+ "epoch": 48.079019073569484,
+ "grad_norm": 3.021986722946167,
+ "learning_rate": 1.1105706491724633e-05,
+ "loss": 0.2134,
+ "step": 17645
+ },
+ {
+ "epoch": 48.08174386920981,
+ "grad_norm": 2.08530330657959,
+ "learning_rate": 1.110482940400623e-05,
+ "loss": 0.0426,
+ "step": 17646
+ },
+ {
+ "epoch": 48.084468664850135,
+ "grad_norm": 2.507469892501831,
+ "learning_rate": 1.110395230768345e-05,
+ "loss": 0.0388,
+ "step": 17647
+ },
+ {
+ "epoch": 48.08719346049046,
+ "grad_norm": 3.0488438606262207,
+ "learning_rate": 1.1103075202763125e-05,
+ "loss": 0.1004,
+ "step": 17648
+ },
+ {
+ "epoch": 48.08991825613079,
+ "grad_norm": 2.734882116317749,
+ "learning_rate": 1.1102198089252083e-05,
+ "loss": 0.0549,
+ "step": 17649
+ },
+ {
+ "epoch": 48.09264305177112,
+ "grad_norm": 3.0436623096466064,
+ "learning_rate": 1.1101320967157164e-05,
+ "loss": 0.2053,
+ "step": 17650
+ },
+ {
+ "epoch": 48.095367847411445,
+ "grad_norm": 2.8632917404174805,
+ "learning_rate": 1.1100443836485187e-05,
+ "loss": 0.0512,
+ "step": 17651
+ },
+ {
+ "epoch": 48.09809264305177,
+ "grad_norm": 2.4841208457946777,
+ "learning_rate": 1.1099566697242992e-05,
+ "loss": 0.0587,
+ "step": 17652
+ },
+ {
+ "epoch": 48.1008174386921,
+ "grad_norm": 2.3314192295074463,
+ "learning_rate": 1.1098689549437406e-05,
+ "loss": 0.0672,
+ "step": 17653
+ },
+ {
+ "epoch": 48.10354223433242,
+ "grad_norm": 3.0436718463897705,
+ "learning_rate": 1.1097812393075262e-05,
+ "loss": 0.1296,
+ "step": 17654
+ },
+ {
+ "epoch": 48.10626702997275,
+ "grad_norm": 2.497135877609253,
+ "learning_rate": 1.1096935228163387e-05,
+ "loss": 0.0471,
+ "step": 17655
+ },
+ {
+ "epoch": 48.10899182561308,
+ "grad_norm": 3.466273307800293,
+ "learning_rate": 1.1096058054708624e-05,
+ "loss": 0.1103,
+ "step": 17656
+ },
+ {
+ "epoch": 48.11171662125341,
+ "grad_norm": 2.539485454559326,
+ "learning_rate": 1.1095180872717789e-05,
+ "loss": 0.1372,
+ "step": 17657
+ },
+ {
+ "epoch": 48.11444141689373,
+ "grad_norm": 2.061994791030884,
+ "learning_rate": 1.109430368219772e-05,
+ "loss": 0.0633,
+ "step": 17658
+ },
+ {
+ "epoch": 48.11716621253406,
+ "grad_norm": 2.027218818664551,
+ "learning_rate": 1.109342648315525e-05,
+ "loss": 0.0416,
+ "step": 17659
+ },
+ {
+ "epoch": 48.119891008174385,
+ "grad_norm": 2.8767969608306885,
+ "learning_rate": 1.1092549275597213e-05,
+ "loss": 0.1841,
+ "step": 17660
+ },
+ {
+ "epoch": 48.12261580381471,
+ "grad_norm": 2.693504810333252,
+ "learning_rate": 1.1091672059530434e-05,
+ "loss": 0.0653,
+ "step": 17661
+ },
+ {
+ "epoch": 48.12534059945504,
+ "grad_norm": 2.8586342334747314,
+ "learning_rate": 1.1090794834961751e-05,
+ "loss": 0.1032,
+ "step": 17662
+ },
+ {
+ "epoch": 48.12806539509537,
+ "grad_norm": 3.367985963821411,
+ "learning_rate": 1.108991760189799e-05,
+ "loss": 0.0568,
+ "step": 17663
+ },
+ {
+ "epoch": 48.130790190735695,
+ "grad_norm": 3.3162691593170166,
+ "learning_rate": 1.1089040360345991e-05,
+ "loss": 0.1003,
+ "step": 17664
+ },
+ {
+ "epoch": 48.13351498637602,
+ "grad_norm": 2.513723850250244,
+ "learning_rate": 1.1088163110312576e-05,
+ "loss": 0.081,
+ "step": 17665
+ },
+ {
+ "epoch": 48.13623978201635,
+ "grad_norm": 2.372418165206909,
+ "learning_rate": 1.108728585180458e-05,
+ "loss": 0.1984,
+ "step": 17666
+ },
+ {
+ "epoch": 48.13896457765667,
+ "grad_norm": 3.2496721744537354,
+ "learning_rate": 1.1086408584828841e-05,
+ "loss": 0.0903,
+ "step": 17667
+ },
+ {
+ "epoch": 48.141689373297005,
+ "grad_norm": 3.3568248748779297,
+ "learning_rate": 1.1085531309392186e-05,
+ "loss": 0.0403,
+ "step": 17668
+ },
+ {
+ "epoch": 48.14441416893733,
+ "grad_norm": 2.4645586013793945,
+ "learning_rate": 1.1084654025501446e-05,
+ "loss": 0.048,
+ "step": 17669
+ },
+ {
+ "epoch": 48.14713896457766,
+ "grad_norm": 2.369861364364624,
+ "learning_rate": 1.1083776733163459e-05,
+ "loss": 0.0575,
+ "step": 17670
+ },
+ {
+ "epoch": 48.14986376021798,
+ "grad_norm": 2.466414213180542,
+ "learning_rate": 1.108289943238505e-05,
+ "loss": 0.0597,
+ "step": 17671
+ },
+ {
+ "epoch": 48.15258855585831,
+ "grad_norm": 2.7669613361358643,
+ "learning_rate": 1.108202212317306e-05,
+ "loss": 0.0551,
+ "step": 17672
+ },
+ {
+ "epoch": 48.155313351498634,
+ "grad_norm": 3.18776273727417,
+ "learning_rate": 1.1081144805534311e-05,
+ "loss": 0.1807,
+ "step": 17673
+ },
+ {
+ "epoch": 48.15803814713897,
+ "grad_norm": 3.1143743991851807,
+ "learning_rate": 1.1080267479475644e-05,
+ "loss": 0.1147,
+ "step": 17674
+ },
+ {
+ "epoch": 48.16076294277929,
+ "grad_norm": 2.552971124649048,
+ "learning_rate": 1.1079390145003886e-05,
+ "loss": 0.0471,
+ "step": 17675
+ },
+ {
+ "epoch": 48.16348773841962,
+ "grad_norm": 2.364525318145752,
+ "learning_rate": 1.1078512802125876e-05,
+ "loss": 0.1964,
+ "step": 17676
+ },
+ {
+ "epoch": 48.166212534059945,
+ "grad_norm": 2.631500244140625,
+ "learning_rate": 1.1077635450848439e-05,
+ "loss": 0.1111,
+ "step": 17677
+ },
+ {
+ "epoch": 48.16893732970027,
+ "grad_norm": 2.4427199363708496,
+ "learning_rate": 1.1076758091178417e-05,
+ "loss": 0.1022,
+ "step": 17678
+ },
+ {
+ "epoch": 48.171662125340596,
+ "grad_norm": 2.415092945098877,
+ "learning_rate": 1.1075880723122632e-05,
+ "loss": 0.0402,
+ "step": 17679
+ },
+ {
+ "epoch": 48.17438692098093,
+ "grad_norm": 2.5717267990112305,
+ "learning_rate": 1.1075003346687927e-05,
+ "loss": 0.0723,
+ "step": 17680
+ },
+ {
+ "epoch": 48.177111716621255,
+ "grad_norm": 3.1179327964782715,
+ "learning_rate": 1.1074125961881128e-05,
+ "loss": 0.2309,
+ "step": 17681
+ },
+ {
+ "epoch": 48.17983651226158,
+ "grad_norm": 3.280992269515991,
+ "learning_rate": 1.1073248568709073e-05,
+ "loss": 0.0759,
+ "step": 17682
+ },
+ {
+ "epoch": 48.182561307901906,
+ "grad_norm": 2.6756906509399414,
+ "learning_rate": 1.1072371167178591e-05,
+ "loss": 0.1411,
+ "step": 17683
+ },
+ {
+ "epoch": 48.18528610354223,
+ "grad_norm": 3.8031489849090576,
+ "learning_rate": 1.1071493757296515e-05,
+ "loss": 0.1039,
+ "step": 17684
+ },
+ {
+ "epoch": 48.18801089918256,
+ "grad_norm": 2.356520414352417,
+ "learning_rate": 1.1070616339069683e-05,
+ "loss": 0.2361,
+ "step": 17685
+ },
+ {
+ "epoch": 48.19073569482289,
+ "grad_norm": 2.0115084648132324,
+ "learning_rate": 1.1069738912504924e-05,
+ "loss": 0.0296,
+ "step": 17686
+ },
+ {
+ "epoch": 48.19346049046322,
+ "grad_norm": 2.303874969482422,
+ "learning_rate": 1.1068861477609072e-05,
+ "loss": 0.1156,
+ "step": 17687
+ },
+ {
+ "epoch": 48.19618528610354,
+ "grad_norm": 2.641449213027954,
+ "learning_rate": 1.1067984034388963e-05,
+ "loss": 0.0541,
+ "step": 17688
+ },
+ {
+ "epoch": 48.19891008174387,
+ "grad_norm": 2.4543309211730957,
+ "learning_rate": 1.1067106582851427e-05,
+ "loss": 0.0607,
+ "step": 17689
+ },
+ {
+ "epoch": 48.201634877384194,
+ "grad_norm": 2.88311767578125,
+ "learning_rate": 1.1066229123003302e-05,
+ "loss": 0.119,
+ "step": 17690
+ },
+ {
+ "epoch": 48.20435967302452,
+ "grad_norm": 2.8944692611694336,
+ "learning_rate": 1.1065351654851417e-05,
+ "loss": 0.0561,
+ "step": 17691
+ },
+ {
+ "epoch": 48.20708446866485,
+ "grad_norm": 3.29729962348938,
+ "learning_rate": 1.1064474178402607e-05,
+ "loss": 0.1472,
+ "step": 17692
+ },
+ {
+ "epoch": 48.20980926430518,
+ "grad_norm": 2.2976253032684326,
+ "learning_rate": 1.106359669366371e-05,
+ "loss": 0.0524,
+ "step": 17693
+ },
+ {
+ "epoch": 48.212534059945504,
+ "grad_norm": 2.9988787174224854,
+ "learning_rate": 1.1062719200641551e-05,
+ "loss": 0.1423,
+ "step": 17694
+ },
+ {
+ "epoch": 48.21525885558583,
+ "grad_norm": 2.7540953159332275,
+ "learning_rate": 1.1061841699342975e-05,
+ "loss": 0.0552,
+ "step": 17695
+ },
+ {
+ "epoch": 48.217983651226156,
+ "grad_norm": 2.389970541000366,
+ "learning_rate": 1.1060964189774804e-05,
+ "loss": 0.0405,
+ "step": 17696
+ },
+ {
+ "epoch": 48.22070844686648,
+ "grad_norm": 2.737276315689087,
+ "learning_rate": 1.106008667194388e-05,
+ "loss": 0.0944,
+ "step": 17697
+ },
+ {
+ "epoch": 48.223433242506815,
+ "grad_norm": 2.804260730743408,
+ "learning_rate": 1.1059209145857038e-05,
+ "loss": 0.0689,
+ "step": 17698
+ },
+ {
+ "epoch": 48.22615803814714,
+ "grad_norm": 2.8365588188171387,
+ "learning_rate": 1.1058331611521105e-05,
+ "loss": 0.0845,
+ "step": 17699
+ },
+ {
+ "epoch": 48.228882833787466,
+ "grad_norm": 1.5584291219711304,
+ "learning_rate": 1.1057454068942923e-05,
+ "loss": 0.0335,
+ "step": 17700
+ },
+ {
+ "epoch": 48.23160762942779,
+ "grad_norm": 3.317922592163086,
+ "learning_rate": 1.105657651812932e-05,
+ "loss": 0.1869,
+ "step": 17701
+ },
+ {
+ "epoch": 48.23433242506812,
+ "grad_norm": 1.9530268907546997,
+ "learning_rate": 1.1055698959087133e-05,
+ "loss": 0.0338,
+ "step": 17702
+ },
+ {
+ "epoch": 48.237057220708444,
+ "grad_norm": 2.4548656940460205,
+ "learning_rate": 1.10548213918232e-05,
+ "loss": 0.0278,
+ "step": 17703
+ },
+ {
+ "epoch": 48.23978201634878,
+ "grad_norm": 2.8422160148620605,
+ "learning_rate": 1.1053943816344348e-05,
+ "loss": 0.2553,
+ "step": 17704
+ },
+ {
+ "epoch": 48.2425068119891,
+ "grad_norm": 2.7732906341552734,
+ "learning_rate": 1.1053066232657419e-05,
+ "loss": 0.1124,
+ "step": 17705
+ },
+ {
+ "epoch": 48.24523160762943,
+ "grad_norm": 3.1899681091308594,
+ "learning_rate": 1.1052188640769242e-05,
+ "loss": 0.0517,
+ "step": 17706
+ },
+ {
+ "epoch": 48.247956403269754,
+ "grad_norm": 2.3398516178131104,
+ "learning_rate": 1.1051311040686654e-05,
+ "loss": 0.1035,
+ "step": 17707
+ },
+ {
+ "epoch": 48.25068119891008,
+ "grad_norm": 3.0454494953155518,
+ "learning_rate": 1.105043343241649e-05,
+ "loss": 0.0728,
+ "step": 17708
+ },
+ {
+ "epoch": 48.253405994550405,
+ "grad_norm": 3.171614646911621,
+ "learning_rate": 1.1049555815965583e-05,
+ "loss": 0.1456,
+ "step": 17709
+ },
+ {
+ "epoch": 48.25613079019074,
+ "grad_norm": 2.6111159324645996,
+ "learning_rate": 1.1048678191340766e-05,
+ "loss": 0.0539,
+ "step": 17710
+ },
+ {
+ "epoch": 48.258855585831064,
+ "grad_norm": 3.311800003051758,
+ "learning_rate": 1.1047800558548885e-05,
+ "loss": 0.1703,
+ "step": 17711
+ },
+ {
+ "epoch": 48.26158038147139,
+ "grad_norm": 4.08121395111084,
+ "learning_rate": 1.1046922917596761e-05,
+ "loss": 0.1161,
+ "step": 17712
+ },
+ {
+ "epoch": 48.264305177111716,
+ "grad_norm": 2.4243297576904297,
+ "learning_rate": 1.1046045268491237e-05,
+ "loss": 0.0546,
+ "step": 17713
+ },
+ {
+ "epoch": 48.26702997275204,
+ "grad_norm": 2.5127270221710205,
+ "learning_rate": 1.1045167611239145e-05,
+ "loss": 0.2161,
+ "step": 17714
+ },
+ {
+ "epoch": 48.26975476839237,
+ "grad_norm": 3.577413320541382,
+ "learning_rate": 1.1044289945847325e-05,
+ "loss": 0.0715,
+ "step": 17715
+ },
+ {
+ "epoch": 48.2724795640327,
+ "grad_norm": 3.1249701976776123,
+ "learning_rate": 1.1043412272322605e-05,
+ "loss": 0.1861,
+ "step": 17716
+ },
+ {
+ "epoch": 48.275204359673026,
+ "grad_norm": 9.04617691040039,
+ "learning_rate": 1.1042534590671825e-05,
+ "loss": 0.1328,
+ "step": 17717
+ },
+ {
+ "epoch": 48.27792915531335,
+ "grad_norm": 2.6262733936309814,
+ "learning_rate": 1.1041656900901816e-05,
+ "loss": 0.2216,
+ "step": 17718
+ },
+ {
+ "epoch": 48.28065395095368,
+ "grad_norm": 2.6830668449401855,
+ "learning_rate": 1.104077920301942e-05,
+ "loss": 0.0845,
+ "step": 17719
+ },
+ {
+ "epoch": 48.283378746594,
+ "grad_norm": 2.40034556388855,
+ "learning_rate": 1.1039901497031466e-05,
+ "loss": 0.077,
+ "step": 17720
+ },
+ {
+ "epoch": 48.28610354223433,
+ "grad_norm": 2.0013227462768555,
+ "learning_rate": 1.1039023782944798e-05,
+ "loss": 0.0343,
+ "step": 17721
+ },
+ {
+ "epoch": 48.28882833787466,
+ "grad_norm": 3.3431665897369385,
+ "learning_rate": 1.1038146060766241e-05,
+ "loss": 0.0823,
+ "step": 17722
+ },
+ {
+ "epoch": 48.29155313351499,
+ "grad_norm": 2.5641798973083496,
+ "learning_rate": 1.103726833050264e-05,
+ "loss": 0.1519,
+ "step": 17723
+ },
+ {
+ "epoch": 48.294277929155314,
+ "grad_norm": 6.528947353363037,
+ "learning_rate": 1.1036390592160824e-05,
+ "loss": 0.1608,
+ "step": 17724
+ },
+ {
+ "epoch": 48.29700272479564,
+ "grad_norm": 2.7799038887023926,
+ "learning_rate": 1.1035512845747633e-05,
+ "loss": 0.2025,
+ "step": 17725
+ },
+ {
+ "epoch": 48.299727520435965,
+ "grad_norm": 3.6612563133239746,
+ "learning_rate": 1.1034635091269899e-05,
+ "loss": 0.1105,
+ "step": 17726
+ },
+ {
+ "epoch": 48.30245231607629,
+ "grad_norm": 2.4793922901153564,
+ "learning_rate": 1.1033757328734461e-05,
+ "loss": 0.1446,
+ "step": 17727
+ },
+ {
+ "epoch": 48.305177111716624,
+ "grad_norm": 2.482189893722534,
+ "learning_rate": 1.1032879558148152e-05,
+ "loss": 0.0902,
+ "step": 17728
+ },
+ {
+ "epoch": 48.30790190735695,
+ "grad_norm": 2.688755512237549,
+ "learning_rate": 1.1032001779517816e-05,
+ "loss": 0.1476,
+ "step": 17729
+ },
+ {
+ "epoch": 48.310626702997276,
+ "grad_norm": 3.12800669670105,
+ "learning_rate": 1.1031123992850279e-05,
+ "loss": 0.057,
+ "step": 17730
+ },
+ {
+ "epoch": 48.3133514986376,
+ "grad_norm": 1.9182682037353516,
+ "learning_rate": 1.1030246198152386e-05,
+ "loss": 0.0266,
+ "step": 17731
+ },
+ {
+ "epoch": 48.31607629427793,
+ "grad_norm": 3.003298759460449,
+ "learning_rate": 1.1029368395430964e-05,
+ "loss": 0.0317,
+ "step": 17732
+ },
+ {
+ "epoch": 48.31880108991825,
+ "grad_norm": 2.708616018295288,
+ "learning_rate": 1.1028490584692856e-05,
+ "loss": 0.0575,
+ "step": 17733
+ },
+ {
+ "epoch": 48.321525885558586,
+ "grad_norm": 2.6594913005828857,
+ "learning_rate": 1.1027612765944895e-05,
+ "loss": 0.0666,
+ "step": 17734
+ },
+ {
+ "epoch": 48.32425068119891,
+ "grad_norm": 2.6902201175689697,
+ "learning_rate": 1.102673493919392e-05,
+ "loss": 0.0634,
+ "step": 17735
+ },
+ {
+ "epoch": 48.32697547683924,
+ "grad_norm": 3.0443451404571533,
+ "learning_rate": 1.1025857104446765e-05,
+ "loss": 0.0873,
+ "step": 17736
+ },
+ {
+ "epoch": 48.32970027247956,
+ "grad_norm": 2.821826696395874,
+ "learning_rate": 1.1024979261710271e-05,
+ "loss": 0.0569,
+ "step": 17737
+ },
+ {
+ "epoch": 48.33242506811989,
+ "grad_norm": 3.054486036300659,
+ "learning_rate": 1.1024101410991268e-05,
+ "loss": 0.0532,
+ "step": 17738
+ },
+ {
+ "epoch": 48.335149863760215,
+ "grad_norm": 3.5306148529052734,
+ "learning_rate": 1.1023223552296601e-05,
+ "loss": 0.2654,
+ "step": 17739
+ },
+ {
+ "epoch": 48.33787465940055,
+ "grad_norm": 2.4695746898651123,
+ "learning_rate": 1.1022345685633095e-05,
+ "loss": 0.0934,
+ "step": 17740
+ },
+ {
+ "epoch": 48.34059945504087,
+ "grad_norm": 2.610743761062622,
+ "learning_rate": 1.10214678110076e-05,
+ "loss": 0.0519,
+ "step": 17741
+ },
+ {
+ "epoch": 48.3433242506812,
+ "grad_norm": 2.4257991313934326,
+ "learning_rate": 1.1020589928426944e-05,
+ "loss": 0.0753,
+ "step": 17742
+ },
+ {
+ "epoch": 48.346049046321525,
+ "grad_norm": 2.312136173248291,
+ "learning_rate": 1.1019712037897967e-05,
+ "loss": 0.1212,
+ "step": 17743
+ },
+ {
+ "epoch": 48.34877384196185,
+ "grad_norm": 3.9203591346740723,
+ "learning_rate": 1.1018834139427503e-05,
+ "loss": 0.0698,
+ "step": 17744
+ },
+ {
+ "epoch": 48.35149863760218,
+ "grad_norm": 2.3973469734191895,
+ "learning_rate": 1.1017956233022394e-05,
+ "loss": 0.0401,
+ "step": 17745
+ },
+ {
+ "epoch": 48.35422343324251,
+ "grad_norm": 2.3771846294403076,
+ "learning_rate": 1.1017078318689473e-05,
+ "loss": 0.0491,
+ "step": 17746
+ },
+ {
+ "epoch": 48.356948228882835,
+ "grad_norm": 3.736830472946167,
+ "learning_rate": 1.1016200396435581e-05,
+ "loss": 0.0823,
+ "step": 17747
+ },
+ {
+ "epoch": 48.35967302452316,
+ "grad_norm": 1.629069209098816,
+ "learning_rate": 1.1015322466267552e-05,
+ "loss": 0.0588,
+ "step": 17748
+ },
+ {
+ "epoch": 48.36239782016349,
+ "grad_norm": 3.4204349517822266,
+ "learning_rate": 1.1014444528192225e-05,
+ "loss": 0.1871,
+ "step": 17749
+ },
+ {
+ "epoch": 48.36512261580381,
+ "grad_norm": 1.9188722372055054,
+ "learning_rate": 1.1013566582216436e-05,
+ "loss": 0.0328,
+ "step": 17750
+ },
+ {
+ "epoch": 48.36784741144414,
+ "grad_norm": 2.4082891941070557,
+ "learning_rate": 1.1012688628347024e-05,
+ "loss": 0.0368,
+ "step": 17751
+ },
+ {
+ "epoch": 48.37057220708447,
+ "grad_norm": 2.2963814735412598,
+ "learning_rate": 1.1011810666590824e-05,
+ "loss": 0.045,
+ "step": 17752
+ },
+ {
+ "epoch": 48.3732970027248,
+ "grad_norm": 2.687026262283325,
+ "learning_rate": 1.1010932696954675e-05,
+ "loss": 0.0691,
+ "step": 17753
+ },
+ {
+ "epoch": 48.37602179836512,
+ "grad_norm": 2.288076162338257,
+ "learning_rate": 1.1010054719445414e-05,
+ "loss": 0.1816,
+ "step": 17754
+ },
+ {
+ "epoch": 48.37874659400545,
+ "grad_norm": 4.379790782928467,
+ "learning_rate": 1.1009176734069881e-05,
+ "loss": 0.1052,
+ "step": 17755
+ },
+ {
+ "epoch": 48.381471389645775,
+ "grad_norm": 4.132734298706055,
+ "learning_rate": 1.1008298740834911e-05,
+ "loss": 0.0835,
+ "step": 17756
+ },
+ {
+ "epoch": 48.3841961852861,
+ "grad_norm": 2.7016830444335938,
+ "learning_rate": 1.1007420739747345e-05,
+ "loss": 0.165,
+ "step": 17757
+ },
+ {
+ "epoch": 48.38692098092643,
+ "grad_norm": 3.323949098587036,
+ "learning_rate": 1.1006542730814017e-05,
+ "loss": 0.0714,
+ "step": 17758
+ },
+ {
+ "epoch": 48.38964577656676,
+ "grad_norm": 3.124953269958496,
+ "learning_rate": 1.1005664714041768e-05,
+ "loss": 0.1607,
+ "step": 17759
+ },
+ {
+ "epoch": 48.392370572207085,
+ "grad_norm": 3.034266471862793,
+ "learning_rate": 1.1004786689437433e-05,
+ "loss": 0.1215,
+ "step": 17760
+ },
+ {
+ "epoch": 48.39509536784741,
+ "grad_norm": 2.705019235610962,
+ "learning_rate": 1.1003908657007851e-05,
+ "loss": 0.0511,
+ "step": 17761
+ },
+ {
+ "epoch": 48.39782016348774,
+ "grad_norm": 2.658473014831543,
+ "learning_rate": 1.100303061675986e-05,
+ "loss": 0.0442,
+ "step": 17762
+ },
+ {
+ "epoch": 48.40054495912806,
+ "grad_norm": 2.5651304721832275,
+ "learning_rate": 1.1002152568700302e-05,
+ "loss": 0.0707,
+ "step": 17763
+ },
+ {
+ "epoch": 48.403269754768395,
+ "grad_norm": 1.6745004653930664,
+ "learning_rate": 1.1001274512836011e-05,
+ "loss": 0.0426,
+ "step": 17764
+ },
+ {
+ "epoch": 48.40599455040872,
+ "grad_norm": 3.1021180152893066,
+ "learning_rate": 1.1000396449173826e-05,
+ "loss": 0.1604,
+ "step": 17765
+ },
+ {
+ "epoch": 48.40871934604905,
+ "grad_norm": 2.701617956161499,
+ "learning_rate": 1.0999518377720587e-05,
+ "loss": 0.0583,
+ "step": 17766
+ },
+ {
+ "epoch": 48.41144414168937,
+ "grad_norm": 2.993894338607788,
+ "learning_rate": 1.099864029848313e-05,
+ "loss": 0.1019,
+ "step": 17767
+ },
+ {
+ "epoch": 48.4141689373297,
+ "grad_norm": 3.5074760913848877,
+ "learning_rate": 1.0997762211468292e-05,
+ "loss": 0.2251,
+ "step": 17768
+ },
+ {
+ "epoch": 48.416893732970024,
+ "grad_norm": 3.1653077602386475,
+ "learning_rate": 1.0996884116682918e-05,
+ "loss": 0.1184,
+ "step": 17769
+ },
+ {
+ "epoch": 48.41961852861036,
+ "grad_norm": 3.141177177429199,
+ "learning_rate": 1.099600601413384e-05,
+ "loss": 0.0783,
+ "step": 17770
+ },
+ {
+ "epoch": 48.42234332425068,
+ "grad_norm": 3.017300605773926,
+ "learning_rate": 1.09951279038279e-05,
+ "loss": 0.109,
+ "step": 17771
+ },
+ {
+ "epoch": 48.42506811989101,
+ "grad_norm": 2.4923689365386963,
+ "learning_rate": 1.0994249785771938e-05,
+ "loss": 0.1439,
+ "step": 17772
+ },
+ {
+ "epoch": 48.427792915531334,
+ "grad_norm": 3.3310654163360596,
+ "learning_rate": 1.0993371659972786e-05,
+ "loss": 0.0975,
+ "step": 17773
+ },
+ {
+ "epoch": 48.43051771117166,
+ "grad_norm": 2.687316417694092,
+ "learning_rate": 1.0992493526437288e-05,
+ "loss": 0.0753,
+ "step": 17774
+ },
+ {
+ "epoch": 48.433242506811986,
+ "grad_norm": 2.984131097793579,
+ "learning_rate": 1.0991615385172288e-05,
+ "loss": 0.042,
+ "step": 17775
+ },
+ {
+ "epoch": 48.43596730245232,
+ "grad_norm": 2.7118654251098633,
+ "learning_rate": 1.0990737236184615e-05,
+ "loss": 0.0614,
+ "step": 17776
+ },
+ {
+ "epoch": 48.438692098092645,
+ "grad_norm": 2.6971795558929443,
+ "learning_rate": 1.0989859079481114e-05,
+ "loss": 0.0813,
+ "step": 17777
+ },
+ {
+ "epoch": 48.44141689373297,
+ "grad_norm": 2.5233898162841797,
+ "learning_rate": 1.098898091506862e-05,
+ "loss": 0.0649,
+ "step": 17778
+ },
+ {
+ "epoch": 48.444141689373296,
+ "grad_norm": 5.518234729766846,
+ "learning_rate": 1.0988102742953975e-05,
+ "loss": 0.1494,
+ "step": 17779
+ },
+ {
+ "epoch": 48.44686648501362,
+ "grad_norm": 2.4809885025024414,
+ "learning_rate": 1.0987224563144018e-05,
+ "loss": 0.0362,
+ "step": 17780
+ },
+ {
+ "epoch": 48.44959128065395,
+ "grad_norm": 2.948784351348877,
+ "learning_rate": 1.098634637564559e-05,
+ "loss": 0.0671,
+ "step": 17781
+ },
+ {
+ "epoch": 48.45231607629428,
+ "grad_norm": 2.6223673820495605,
+ "learning_rate": 1.0985468180465525e-05,
+ "loss": 0.0701,
+ "step": 17782
+ },
+ {
+ "epoch": 48.45504087193461,
+ "grad_norm": 3.1557023525238037,
+ "learning_rate": 1.0984589977610669e-05,
+ "loss": 0.1168,
+ "step": 17783
+ },
+ {
+ "epoch": 48.45776566757493,
+ "grad_norm": 3.0509653091430664,
+ "learning_rate": 1.0983711767087856e-05,
+ "loss": 0.0695,
+ "step": 17784
+ },
+ {
+ "epoch": 48.46049046321526,
+ "grad_norm": 3.1287906169891357,
+ "learning_rate": 1.0982833548903926e-05,
+ "loss": 0.0476,
+ "step": 17785
+ },
+ {
+ "epoch": 48.463215258855584,
+ "grad_norm": 2.042253017425537,
+ "learning_rate": 1.0981955323065723e-05,
+ "loss": 0.0763,
+ "step": 17786
+ },
+ {
+ "epoch": 48.46594005449591,
+ "grad_norm": 2.9440574645996094,
+ "learning_rate": 1.098107708958008e-05,
+ "loss": 0.1122,
+ "step": 17787
+ },
+ {
+ "epoch": 48.46866485013624,
+ "grad_norm": 2.840392589569092,
+ "learning_rate": 1.0980198848453843e-05,
+ "loss": 0.0628,
+ "step": 17788
+ },
+ {
+ "epoch": 48.47138964577657,
+ "grad_norm": 2.7567789554595947,
+ "learning_rate": 1.0979320599693848e-05,
+ "loss": 0.074,
+ "step": 17789
+ },
+ {
+ "epoch": 48.474114441416894,
+ "grad_norm": 2.927389144897461,
+ "learning_rate": 1.0978442343306936e-05,
+ "loss": 0.1602,
+ "step": 17790
+ },
+ {
+ "epoch": 48.47683923705722,
+ "grad_norm": 2.2287235260009766,
+ "learning_rate": 1.0977564079299946e-05,
+ "loss": 0.0648,
+ "step": 17791
+ },
+ {
+ "epoch": 48.479564032697546,
+ "grad_norm": 2.375643491744995,
+ "learning_rate": 1.0976685807679719e-05,
+ "loss": 0.1754,
+ "step": 17792
+ },
+ {
+ "epoch": 48.48228882833787,
+ "grad_norm": 2.436002016067505,
+ "learning_rate": 1.0975807528453093e-05,
+ "loss": 0.047,
+ "step": 17793
+ },
+ {
+ "epoch": 48.485013623978205,
+ "grad_norm": 2.446694850921631,
+ "learning_rate": 1.0974929241626912e-05,
+ "loss": 0.1412,
+ "step": 17794
+ },
+ {
+ "epoch": 48.48773841961853,
+ "grad_norm": 2.0197956562042236,
+ "learning_rate": 1.0974050947208012e-05,
+ "loss": 0.0416,
+ "step": 17795
+ },
+ {
+ "epoch": 48.490463215258856,
+ "grad_norm": 2.695866823196411,
+ "learning_rate": 1.0973172645203235e-05,
+ "loss": 0.0614,
+ "step": 17796
+ },
+ {
+ "epoch": 48.49318801089918,
+ "grad_norm": 3.04227614402771,
+ "learning_rate": 1.0972294335619418e-05,
+ "loss": 0.0447,
+ "step": 17797
+ },
+ {
+ "epoch": 48.49591280653951,
+ "grad_norm": 2.8660571575164795,
+ "learning_rate": 1.0971416018463406e-05,
+ "loss": 0.0562,
+ "step": 17798
+ },
+ {
+ "epoch": 48.49863760217983,
+ "grad_norm": 2.094625234603882,
+ "learning_rate": 1.0970537693742038e-05,
+ "loss": 0.209,
+ "step": 17799
+ },
+ {
+ "epoch": 48.50136239782017,
+ "grad_norm": 2.8557190895080566,
+ "learning_rate": 1.0969659361462153e-05,
+ "loss": 0.0594,
+ "step": 17800
+ },
+ {
+ "epoch": 48.50408719346049,
+ "grad_norm": 2.1815083026885986,
+ "learning_rate": 1.0968781021630593e-05,
+ "loss": 0.1292,
+ "step": 17801
+ },
+ {
+ "epoch": 48.50681198910082,
+ "grad_norm": 2.604773998260498,
+ "learning_rate": 1.0967902674254199e-05,
+ "loss": 0.0632,
+ "step": 17802
+ },
+ {
+ "epoch": 48.509536784741144,
+ "grad_norm": 3.236093282699585,
+ "learning_rate": 1.0967024319339806e-05,
+ "loss": 0.0788,
+ "step": 17803
+ },
+ {
+ "epoch": 48.51226158038147,
+ "grad_norm": 2.849534511566162,
+ "learning_rate": 1.0966145956894259e-05,
+ "loss": 0.0625,
+ "step": 17804
+ },
+ {
+ "epoch": 48.514986376021795,
+ "grad_norm": 4.085686206817627,
+ "learning_rate": 1.0965267586924399e-05,
+ "loss": 0.1606,
+ "step": 17805
+ },
+ {
+ "epoch": 48.51771117166213,
+ "grad_norm": 2.8375589847564697,
+ "learning_rate": 1.096438920943707e-05,
+ "loss": 0.1644,
+ "step": 17806
+ },
+ {
+ "epoch": 48.520435967302454,
+ "grad_norm": 2.97253680229187,
+ "learning_rate": 1.0963510824439102e-05,
+ "loss": 0.0487,
+ "step": 17807
+ },
+ {
+ "epoch": 48.52316076294278,
+ "grad_norm": 2.359736442565918,
+ "learning_rate": 1.096263243193735e-05,
+ "loss": 0.0553,
+ "step": 17808
+ },
+ {
+ "epoch": 48.525885558583106,
+ "grad_norm": 2.850938320159912,
+ "learning_rate": 1.0961754031938643e-05,
+ "loss": 0.173,
+ "step": 17809
+ },
+ {
+ "epoch": 48.52861035422343,
+ "grad_norm": 2.7836015224456787,
+ "learning_rate": 1.096087562444983e-05,
+ "loss": 0.0885,
+ "step": 17810
+ },
+ {
+ "epoch": 48.53133514986376,
+ "grad_norm": 3.713740110397339,
+ "learning_rate": 1.0959997209477747e-05,
+ "loss": 0.0647,
+ "step": 17811
+ },
+ {
+ "epoch": 48.53405994550409,
+ "grad_norm": 2.7944793701171875,
+ "learning_rate": 1.0959118787029236e-05,
+ "loss": 0.0693,
+ "step": 17812
+ },
+ {
+ "epoch": 48.536784741144416,
+ "grad_norm": 2.8329474925994873,
+ "learning_rate": 1.0958240357111136e-05,
+ "loss": 0.1097,
+ "step": 17813
+ },
+ {
+ "epoch": 48.53950953678474,
+ "grad_norm": 2.4837000370025635,
+ "learning_rate": 1.0957361919730293e-05,
+ "loss": 0.0676,
+ "step": 17814
+ },
+ {
+ "epoch": 48.54223433242507,
+ "grad_norm": 2.423370599746704,
+ "learning_rate": 1.095648347489355e-05,
+ "loss": 0.0881,
+ "step": 17815
+ },
+ {
+ "epoch": 48.54495912806539,
+ "grad_norm": 2.1597983837127686,
+ "learning_rate": 1.095560502260774e-05,
+ "loss": 0.0404,
+ "step": 17816
+ },
+ {
+ "epoch": 48.54768392370572,
+ "grad_norm": 2.1059272289276123,
+ "learning_rate": 1.0954726562879709e-05,
+ "loss": 0.0337,
+ "step": 17817
+ },
+ {
+ "epoch": 48.55040871934605,
+ "grad_norm": 2.5681369304656982,
+ "learning_rate": 1.0953848095716302e-05,
+ "loss": 0.0438,
+ "step": 17818
+ },
+ {
+ "epoch": 48.55313351498638,
+ "grad_norm": 2.710106611251831,
+ "learning_rate": 1.0952969621124354e-05,
+ "loss": 0.0681,
+ "step": 17819
+ },
+ {
+ "epoch": 48.555858310626704,
+ "grad_norm": 3.556849241256714,
+ "learning_rate": 1.095209113911071e-05,
+ "loss": 0.1952,
+ "step": 17820
+ },
+ {
+ "epoch": 48.55858310626703,
+ "grad_norm": 2.113124370574951,
+ "learning_rate": 1.0951212649682208e-05,
+ "loss": 0.0251,
+ "step": 17821
+ },
+ {
+ "epoch": 48.561307901907355,
+ "grad_norm": 2.689687490463257,
+ "learning_rate": 1.0950334152845697e-05,
+ "loss": 0.1058,
+ "step": 17822
+ },
+ {
+ "epoch": 48.56403269754768,
+ "grad_norm": 3.3139922618865967,
+ "learning_rate": 1.094945564860801e-05,
+ "loss": 0.0849,
+ "step": 17823
+ },
+ {
+ "epoch": 48.566757493188014,
+ "grad_norm": 2.7741191387176514,
+ "learning_rate": 1.0948577136975998e-05,
+ "loss": 0.1542,
+ "step": 17824
+ },
+ {
+ "epoch": 48.56948228882834,
+ "grad_norm": 2.6584763526916504,
+ "learning_rate": 1.0947698617956493e-05,
+ "loss": 0.0919,
+ "step": 17825
+ },
+ {
+ "epoch": 48.572207084468666,
+ "grad_norm": 3.2591023445129395,
+ "learning_rate": 1.0946820091556345e-05,
+ "loss": 0.1624,
+ "step": 17826
+ },
+ {
+ "epoch": 48.57493188010899,
+ "grad_norm": 2.7699971199035645,
+ "learning_rate": 1.0945941557782389e-05,
+ "loss": 0.0666,
+ "step": 17827
+ },
+ {
+ "epoch": 48.57765667574932,
+ "grad_norm": 4.140091419219971,
+ "learning_rate": 1.0945063016641475e-05,
+ "loss": 0.0627,
+ "step": 17828
+ },
+ {
+ "epoch": 48.58038147138964,
+ "grad_norm": 4.009925842285156,
+ "learning_rate": 1.0944184468140438e-05,
+ "loss": 0.1392,
+ "step": 17829
+ },
+ {
+ "epoch": 48.583106267029976,
+ "grad_norm": 2.853372812271118,
+ "learning_rate": 1.0943305912286122e-05,
+ "loss": 0.1071,
+ "step": 17830
+ },
+ {
+ "epoch": 48.5858310626703,
+ "grad_norm": 3.162996768951416,
+ "learning_rate": 1.0942427349085369e-05,
+ "loss": 0.1274,
+ "step": 17831
+ },
+ {
+ "epoch": 48.58855585831063,
+ "grad_norm": 2.7406845092773438,
+ "learning_rate": 1.0941548778545026e-05,
+ "loss": 0.1524,
+ "step": 17832
+ },
+ {
+ "epoch": 48.59128065395095,
+ "grad_norm": 2.9043288230895996,
+ "learning_rate": 1.0940670200671927e-05,
+ "loss": 0.0628,
+ "step": 17833
+ },
+ {
+ "epoch": 48.59400544959128,
+ "grad_norm": 2.62968111038208,
+ "learning_rate": 1.0939791615472925e-05,
+ "loss": 0.0741,
+ "step": 17834
+ },
+ {
+ "epoch": 48.596730245231605,
+ "grad_norm": 2.5662577152252197,
+ "learning_rate": 1.093891302295485e-05,
+ "loss": 0.0579,
+ "step": 17835
+ },
+ {
+ "epoch": 48.59945504087194,
+ "grad_norm": 2.309643507003784,
+ "learning_rate": 1.0938034423124556e-05,
+ "loss": 0.1543,
+ "step": 17836
+ },
+ {
+ "epoch": 48.60217983651226,
+ "grad_norm": 2.7618892192840576,
+ "learning_rate": 1.0937155815988876e-05,
+ "loss": 0.0588,
+ "step": 17837
+ },
+ {
+ "epoch": 48.60490463215259,
+ "grad_norm": 2.946566343307495,
+ "learning_rate": 1.0936277201554659e-05,
+ "loss": 0.0772,
+ "step": 17838
+ },
+ {
+ "epoch": 48.607629427792915,
+ "grad_norm": 2.144101858139038,
+ "learning_rate": 1.093539857982874e-05,
+ "loss": 0.0415,
+ "step": 17839
+ },
+ {
+ "epoch": 48.61035422343324,
+ "grad_norm": 2.370884656906128,
+ "learning_rate": 1.0934519950817975e-05,
+ "loss": 0.0632,
+ "step": 17840
+ },
+ {
+ "epoch": 48.61307901907357,
+ "grad_norm": 2.778916597366333,
+ "learning_rate": 1.0933641314529194e-05,
+ "loss": 0.0582,
+ "step": 17841
+ },
+ {
+ "epoch": 48.6158038147139,
+ "grad_norm": 3.6477177143096924,
+ "learning_rate": 1.0932762670969245e-05,
+ "loss": 0.2491,
+ "step": 17842
+ },
+ {
+ "epoch": 48.618528610354225,
+ "grad_norm": 2.3540098667144775,
+ "learning_rate": 1.093188402014497e-05,
+ "loss": 0.0476,
+ "step": 17843
+ },
+ {
+ "epoch": 48.62125340599455,
+ "grad_norm": 2.7567856311798096,
+ "learning_rate": 1.0931005362063215e-05,
+ "loss": 0.0296,
+ "step": 17844
+ },
+ {
+ "epoch": 48.62397820163488,
+ "grad_norm": 2.3983259201049805,
+ "learning_rate": 1.0930126696730817e-05,
+ "loss": 0.1187,
+ "step": 17845
+ },
+ {
+ "epoch": 48.6267029972752,
+ "grad_norm": 3.1358048915863037,
+ "learning_rate": 1.0929248024154623e-05,
+ "loss": 0.0563,
+ "step": 17846
+ },
+ {
+ "epoch": 48.62942779291553,
+ "grad_norm": 4.009900093078613,
+ "learning_rate": 1.0928369344341475e-05,
+ "loss": 0.0642,
+ "step": 17847
+ },
+ {
+ "epoch": 48.63215258855586,
+ "grad_norm": 3.2491374015808105,
+ "learning_rate": 1.0927490657298217e-05,
+ "loss": 0.2078,
+ "step": 17848
+ },
+ {
+ "epoch": 48.63487738419619,
+ "grad_norm": 3.351067543029785,
+ "learning_rate": 1.092661196303169e-05,
+ "loss": 0.0854,
+ "step": 17849
+ },
+ {
+ "epoch": 48.63760217983651,
+ "grad_norm": 2.5168473720550537,
+ "learning_rate": 1.092573326154874e-05,
+ "loss": 0.09,
+ "step": 17850
+ },
+ {
+ "epoch": 48.64032697547684,
+ "grad_norm": 2.4439194202423096,
+ "learning_rate": 1.092485455285621e-05,
+ "loss": 0.1335,
+ "step": 17851
+ },
+ {
+ "epoch": 48.643051771117165,
+ "grad_norm": 2.272148370742798,
+ "learning_rate": 1.0923975836960945e-05,
+ "loss": 0.0735,
+ "step": 17852
+ },
+ {
+ "epoch": 48.64577656675749,
+ "grad_norm": 2.5930135250091553,
+ "learning_rate": 1.092309711386978e-05,
+ "loss": 0.081,
+ "step": 17853
+ },
+ {
+ "epoch": 48.64850136239782,
+ "grad_norm": 11.102985382080078,
+ "learning_rate": 1.0922218383589571e-05,
+ "loss": 0.1038,
+ "step": 17854
+ },
+ {
+ "epoch": 48.65122615803815,
+ "grad_norm": 2.807741641998291,
+ "learning_rate": 1.0921339646127149e-05,
+ "loss": 0.1169,
+ "step": 17855
+ },
+ {
+ "epoch": 48.653950953678475,
+ "grad_norm": 3.1355643272399902,
+ "learning_rate": 1.0920460901489367e-05,
+ "loss": 0.0907,
+ "step": 17856
+ },
+ {
+ "epoch": 48.6566757493188,
+ "grad_norm": 2.5485007762908936,
+ "learning_rate": 1.0919582149683063e-05,
+ "loss": 0.055,
+ "step": 17857
+ },
+ {
+ "epoch": 48.65940054495913,
+ "grad_norm": 2.8033759593963623,
+ "learning_rate": 1.0918703390715085e-05,
+ "loss": 0.0622,
+ "step": 17858
+ },
+ {
+ "epoch": 48.66212534059945,
+ "grad_norm": 2.1833548545837402,
+ "learning_rate": 1.0917824624592275e-05,
+ "loss": 0.0804,
+ "step": 17859
+ },
+ {
+ "epoch": 48.664850136239785,
+ "grad_norm": 3.533519744873047,
+ "learning_rate": 1.0916945851321472e-05,
+ "loss": 0.0763,
+ "step": 17860
+ },
+ {
+ "epoch": 48.66757493188011,
+ "grad_norm": 4.07701301574707,
+ "learning_rate": 1.0916067070909526e-05,
+ "loss": 0.2056,
+ "step": 17861
+ },
+ {
+ "epoch": 48.67029972752044,
+ "grad_norm": 2.676598310470581,
+ "learning_rate": 1.0915188283363282e-05,
+ "loss": 0.0966,
+ "step": 17862
+ },
+ {
+ "epoch": 48.67302452316076,
+ "grad_norm": 3.3538646697998047,
+ "learning_rate": 1.0914309488689578e-05,
+ "loss": 0.1155,
+ "step": 17863
+ },
+ {
+ "epoch": 48.67574931880109,
+ "grad_norm": 2.1572792530059814,
+ "learning_rate": 1.0913430686895261e-05,
+ "loss": 0.0289,
+ "step": 17864
+ },
+ {
+ "epoch": 48.678474114441414,
+ "grad_norm": 3.184718370437622,
+ "learning_rate": 1.0912551877987179e-05,
+ "loss": 0.0684,
+ "step": 17865
+ },
+ {
+ "epoch": 48.68119891008175,
+ "grad_norm": 3.8212289810180664,
+ "learning_rate": 1.0911673061972168e-05,
+ "loss": 0.16,
+ "step": 17866
+ },
+ {
+ "epoch": 48.68392370572207,
+ "grad_norm": 3.1432225704193115,
+ "learning_rate": 1.0910794238857082e-05,
+ "loss": 0.1835,
+ "step": 17867
+ },
+ {
+ "epoch": 48.6866485013624,
+ "grad_norm": 3.0590755939483643,
+ "learning_rate": 1.0909915408648755e-05,
+ "loss": 0.1087,
+ "step": 17868
+ },
+ {
+ "epoch": 48.689373297002724,
+ "grad_norm": 2.3475704193115234,
+ "learning_rate": 1.0909036571354037e-05,
+ "loss": 0.0372,
+ "step": 17869
+ },
+ {
+ "epoch": 48.69209809264305,
+ "grad_norm": 4.45289421081543,
+ "learning_rate": 1.0908157726979772e-05,
+ "loss": 0.202,
+ "step": 17870
+ },
+ {
+ "epoch": 48.694822888283376,
+ "grad_norm": 4.035921573638916,
+ "learning_rate": 1.0907278875532805e-05,
+ "loss": 0.1849,
+ "step": 17871
+ },
+ {
+ "epoch": 48.69754768392371,
+ "grad_norm": 2.8902225494384766,
+ "learning_rate": 1.0906400017019976e-05,
+ "loss": 0.1237,
+ "step": 17872
+ },
+ {
+ "epoch": 48.700272479564035,
+ "grad_norm": 5.581071376800537,
+ "learning_rate": 1.0905521151448132e-05,
+ "loss": 0.1235,
+ "step": 17873
+ },
+ {
+ "epoch": 48.70299727520436,
+ "grad_norm": 3.2748868465423584,
+ "learning_rate": 1.0904642278824122e-05,
+ "loss": 0.0849,
+ "step": 17874
+ },
+ {
+ "epoch": 48.705722070844686,
+ "grad_norm": 3.1764209270477295,
+ "learning_rate": 1.0903763399154787e-05,
+ "loss": 0.0762,
+ "step": 17875
+ },
+ {
+ "epoch": 48.70844686648501,
+ "grad_norm": 2.2713372707366943,
+ "learning_rate": 1.0902884512446972e-05,
+ "loss": 0.0468,
+ "step": 17876
+ },
+ {
+ "epoch": 48.71117166212534,
+ "grad_norm": 3.8586575984954834,
+ "learning_rate": 1.090200561870752e-05,
+ "loss": 0.0626,
+ "step": 17877
+ },
+ {
+ "epoch": 48.71389645776567,
+ "grad_norm": 2.6073431968688965,
+ "learning_rate": 1.0901126717943277e-05,
+ "loss": 0.0596,
+ "step": 17878
+ },
+ {
+ "epoch": 48.716621253406,
+ "grad_norm": 3.3192265033721924,
+ "learning_rate": 1.0900247810161091e-05,
+ "loss": 0.0733,
+ "step": 17879
+ },
+ {
+ "epoch": 48.71934604904632,
+ "grad_norm": 2.336872100830078,
+ "learning_rate": 1.0899368895367802e-05,
+ "loss": 0.0665,
+ "step": 17880
+ },
+ {
+ "epoch": 48.72207084468665,
+ "grad_norm": 6.574321746826172,
+ "learning_rate": 1.0898489973570256e-05,
+ "loss": 0.0709,
+ "step": 17881
+ },
+ {
+ "epoch": 48.724795640326974,
+ "grad_norm": 2.4883387088775635,
+ "learning_rate": 1.0897611044775299e-05,
+ "loss": 0.0623,
+ "step": 17882
+ },
+ {
+ "epoch": 48.7275204359673,
+ "grad_norm": 3.0205628871917725,
+ "learning_rate": 1.0896732108989775e-05,
+ "loss": 0.0495,
+ "step": 17883
+ },
+ {
+ "epoch": 48.73024523160763,
+ "grad_norm": 3.3506083488464355,
+ "learning_rate": 1.0895853166220534e-05,
+ "loss": 0.058,
+ "step": 17884
+ },
+ {
+ "epoch": 48.73297002724796,
+ "grad_norm": 2.8703784942626953,
+ "learning_rate": 1.0894974216474415e-05,
+ "loss": 0.1023,
+ "step": 17885
+ },
+ {
+ "epoch": 48.735694822888284,
+ "grad_norm": 2.990722894668579,
+ "learning_rate": 1.0894095259758267e-05,
+ "loss": 0.0669,
+ "step": 17886
+ },
+ {
+ "epoch": 48.73841961852861,
+ "grad_norm": 3.781271457672119,
+ "learning_rate": 1.0893216296078932e-05,
+ "loss": 0.1235,
+ "step": 17887
+ },
+ {
+ "epoch": 48.741144414168936,
+ "grad_norm": 2.8118247985839844,
+ "learning_rate": 1.0892337325443259e-05,
+ "loss": 0.1514,
+ "step": 17888
+ },
+ {
+ "epoch": 48.74386920980926,
+ "grad_norm": 2.9285037517547607,
+ "learning_rate": 1.089145834785809e-05,
+ "loss": 0.1355,
+ "step": 17889
+ },
+ {
+ "epoch": 48.746594005449595,
+ "grad_norm": 2.3810455799102783,
+ "learning_rate": 1.0890579363330271e-05,
+ "loss": 0.0676,
+ "step": 17890
+ },
+ {
+ "epoch": 48.74931880108992,
+ "grad_norm": 2.4691665172576904,
+ "learning_rate": 1.0889700371866651e-05,
+ "loss": 0.3039,
+ "step": 17891
+ },
+ {
+ "epoch": 48.752043596730246,
+ "grad_norm": 2.4086694717407227,
+ "learning_rate": 1.0888821373474073e-05,
+ "loss": 0.1022,
+ "step": 17892
+ },
+ {
+ "epoch": 48.75476839237057,
+ "grad_norm": 2.6359357833862305,
+ "learning_rate": 1.0887942368159384e-05,
+ "loss": 0.0598,
+ "step": 17893
+ },
+ {
+ "epoch": 48.7574931880109,
+ "grad_norm": 2.9705278873443604,
+ "learning_rate": 1.0887063355929426e-05,
+ "loss": 0.0459,
+ "step": 17894
+ },
+ {
+ "epoch": 48.76021798365122,
+ "grad_norm": 2.694471836090088,
+ "learning_rate": 1.0886184336791047e-05,
+ "loss": 0.1089,
+ "step": 17895
+ },
+ {
+ "epoch": 48.762942779291556,
+ "grad_norm": 2.0046329498291016,
+ "learning_rate": 1.0885305310751095e-05,
+ "loss": 0.0311,
+ "step": 17896
+ },
+ {
+ "epoch": 48.76566757493188,
+ "grad_norm": 3.4832046031951904,
+ "learning_rate": 1.0884426277816413e-05,
+ "loss": 0.0584,
+ "step": 17897
+ },
+ {
+ "epoch": 48.76839237057221,
+ "grad_norm": 3.3515944480895996,
+ "learning_rate": 1.0883547237993846e-05,
+ "loss": 0.1435,
+ "step": 17898
+ },
+ {
+ "epoch": 48.771117166212534,
+ "grad_norm": 2.9549195766448975,
+ "learning_rate": 1.0882668191290242e-05,
+ "loss": 0.1361,
+ "step": 17899
+ },
+ {
+ "epoch": 48.77384196185286,
+ "grad_norm": 2.9547908306121826,
+ "learning_rate": 1.0881789137712446e-05,
+ "loss": 0.0592,
+ "step": 17900
+ },
+ {
+ "epoch": 48.776566757493185,
+ "grad_norm": 3.0550906658172607,
+ "learning_rate": 1.0880910077267308e-05,
+ "loss": 0.0556,
+ "step": 17901
+ },
+ {
+ "epoch": 48.77929155313352,
+ "grad_norm": 3.159517526626587,
+ "learning_rate": 1.0880031009961667e-05,
+ "loss": 0.1117,
+ "step": 17902
+ },
+ {
+ "epoch": 48.782016348773844,
+ "grad_norm": 2.3347957134246826,
+ "learning_rate": 1.0879151935802376e-05,
+ "loss": 0.1053,
+ "step": 17903
+ },
+ {
+ "epoch": 48.78474114441417,
+ "grad_norm": 2.6541810035705566,
+ "learning_rate": 1.0878272854796274e-05,
+ "loss": 0.091,
+ "step": 17904
+ },
+ {
+ "epoch": 48.787465940054496,
+ "grad_norm": 2.464064598083496,
+ "learning_rate": 1.0877393766950217e-05,
+ "loss": 0.0342,
+ "step": 17905
+ },
+ {
+ "epoch": 48.79019073569482,
+ "grad_norm": 2.5267932415008545,
+ "learning_rate": 1.0876514672271041e-05,
+ "loss": 0.1226,
+ "step": 17906
+ },
+ {
+ "epoch": 48.79291553133515,
+ "grad_norm": 3.080045700073242,
+ "learning_rate": 1.0875635570765596e-05,
+ "loss": 0.2346,
+ "step": 17907
+ },
+ {
+ "epoch": 48.79564032697548,
+ "grad_norm": 2.4043846130371094,
+ "learning_rate": 1.0874756462440732e-05,
+ "loss": 0.087,
+ "step": 17908
+ },
+ {
+ "epoch": 48.798365122615806,
+ "grad_norm": 1.9824086427688599,
+ "learning_rate": 1.0873877347303295e-05,
+ "loss": 0.0917,
+ "step": 17909
+ },
+ {
+ "epoch": 48.80108991825613,
+ "grad_norm": 1.9359489679336548,
+ "learning_rate": 1.0872998225360126e-05,
+ "loss": 0.0371,
+ "step": 17910
+ },
+ {
+ "epoch": 48.80381471389646,
+ "grad_norm": 2.5726704597473145,
+ "learning_rate": 1.0872119096618074e-05,
+ "loss": 0.0641,
+ "step": 17911
+ },
+ {
+ "epoch": 48.80653950953678,
+ "grad_norm": 2.7403299808502197,
+ "learning_rate": 1.0871239961083987e-05,
+ "loss": 0.0547,
+ "step": 17912
+ },
+ {
+ "epoch": 48.80926430517711,
+ "grad_norm": 3.1691062450408936,
+ "learning_rate": 1.0870360818764716e-05,
+ "loss": 0.0675,
+ "step": 17913
+ },
+ {
+ "epoch": 48.81198910081744,
+ "grad_norm": 2.805917739868164,
+ "learning_rate": 1.0869481669667097e-05,
+ "loss": 0.0549,
+ "step": 17914
+ },
+ {
+ "epoch": 48.81471389645777,
+ "grad_norm": 2.3877837657928467,
+ "learning_rate": 1.0868602513797989e-05,
+ "loss": 0.1119,
+ "step": 17915
+ },
+ {
+ "epoch": 48.817438692098094,
+ "grad_norm": 3.035651445388794,
+ "learning_rate": 1.0867723351164223e-05,
+ "loss": 0.1852,
+ "step": 17916
+ },
+ {
+ "epoch": 48.82016348773842,
+ "grad_norm": 2.4301979541778564,
+ "learning_rate": 1.0866844181772662e-05,
+ "loss": 0.145,
+ "step": 17917
+ },
+ {
+ "epoch": 48.822888283378745,
+ "grad_norm": 2.723475217819214,
+ "learning_rate": 1.0865965005630147e-05,
+ "loss": 0.055,
+ "step": 17918
+ },
+ {
+ "epoch": 48.82561307901907,
+ "grad_norm": 2.5858099460601807,
+ "learning_rate": 1.0865085822743521e-05,
+ "loss": 0.0598,
+ "step": 17919
+ },
+ {
+ "epoch": 48.828337874659404,
+ "grad_norm": 2.83579683303833,
+ "learning_rate": 1.0864206633119637e-05,
+ "loss": 0.201,
+ "step": 17920
+ },
+ {
+ "epoch": 48.83106267029973,
+ "grad_norm": 3.2286229133605957,
+ "learning_rate": 1.0863327436765342e-05,
+ "loss": 0.1112,
+ "step": 17921
+ },
+ {
+ "epoch": 48.833787465940055,
+ "grad_norm": 2.3076088428497314,
+ "learning_rate": 1.0862448233687475e-05,
+ "loss": 0.0562,
+ "step": 17922
+ },
+ {
+ "epoch": 48.83651226158038,
+ "grad_norm": 2.5334246158599854,
+ "learning_rate": 1.0861569023892893e-05,
+ "loss": 0.0629,
+ "step": 17923
+ },
+ {
+ "epoch": 48.83923705722071,
+ "grad_norm": 2.6522464752197266,
+ "learning_rate": 1.0860689807388437e-05,
+ "loss": 0.1417,
+ "step": 17924
+ },
+ {
+ "epoch": 48.84196185286103,
+ "grad_norm": 9.687566757202148,
+ "learning_rate": 1.0859810584180956e-05,
+ "loss": 0.1034,
+ "step": 17925
+ },
+ {
+ "epoch": 48.844686648501366,
+ "grad_norm": 3.2899329662323,
+ "learning_rate": 1.0858931354277299e-05,
+ "loss": 0.054,
+ "step": 17926
+ },
+ {
+ "epoch": 48.84741144414169,
+ "grad_norm": 2.7329773902893066,
+ "learning_rate": 1.0858052117684312e-05,
+ "loss": 0.0632,
+ "step": 17927
+ },
+ {
+ "epoch": 48.85013623978202,
+ "grad_norm": 3.535761594772339,
+ "learning_rate": 1.0857172874408844e-05,
+ "loss": 0.1577,
+ "step": 17928
+ },
+ {
+ "epoch": 48.85286103542234,
+ "grad_norm": 3.3936078548431396,
+ "learning_rate": 1.0856293624457741e-05,
+ "loss": 0.3215,
+ "step": 17929
+ },
+ {
+ "epoch": 48.85558583106267,
+ "grad_norm": 2.147068977355957,
+ "learning_rate": 1.0855414367837848e-05,
+ "loss": 0.0364,
+ "step": 17930
+ },
+ {
+ "epoch": 48.858310626702995,
+ "grad_norm": 2.9419517517089844,
+ "learning_rate": 1.0854535104556019e-05,
+ "loss": 0.0443,
+ "step": 17931
+ },
+ {
+ "epoch": 48.86103542234333,
+ "grad_norm": 2.0291099548339844,
+ "learning_rate": 1.0853655834619096e-05,
+ "loss": 0.0639,
+ "step": 17932
+ },
+ {
+ "epoch": 48.86376021798365,
+ "grad_norm": 2.6877336502075195,
+ "learning_rate": 1.0852776558033928e-05,
+ "loss": 0.1417,
+ "step": 17933
+ },
+ {
+ "epoch": 48.86648501362398,
+ "grad_norm": 2.369535207748413,
+ "learning_rate": 1.0851897274807368e-05,
+ "loss": 0.1559,
+ "step": 17934
+ },
+ {
+ "epoch": 48.869209809264305,
+ "grad_norm": 2.372760772705078,
+ "learning_rate": 1.0851017984946255e-05,
+ "loss": 0.0483,
+ "step": 17935
+ },
+ {
+ "epoch": 48.87193460490463,
+ "grad_norm": 6.041565418243408,
+ "learning_rate": 1.0850138688457441e-05,
+ "loss": 0.0814,
+ "step": 17936
+ },
+ {
+ "epoch": 48.87465940054496,
+ "grad_norm": 3.055098533630371,
+ "learning_rate": 1.0849259385347779e-05,
+ "loss": 0.1773,
+ "step": 17937
+ },
+ {
+ "epoch": 48.87738419618529,
+ "grad_norm": 2.576897621154785,
+ "learning_rate": 1.0848380075624107e-05,
+ "loss": 0.1794,
+ "step": 17938
+ },
+ {
+ "epoch": 48.880108991825615,
+ "grad_norm": 2.4560325145721436,
+ "learning_rate": 1.0847500759293284e-05,
+ "loss": 0.0411,
+ "step": 17939
+ },
+ {
+ "epoch": 48.88283378746594,
+ "grad_norm": 2.105100154876709,
+ "learning_rate": 1.0846621436362146e-05,
+ "loss": 0.0435,
+ "step": 17940
+ },
+ {
+ "epoch": 48.88555858310627,
+ "grad_norm": 3.1624839305877686,
+ "learning_rate": 1.0845742106837553e-05,
+ "loss": 0.0696,
+ "step": 17941
+ },
+ {
+ "epoch": 48.88828337874659,
+ "grad_norm": 2.800328254699707,
+ "learning_rate": 1.0844862770726344e-05,
+ "loss": 0.0664,
+ "step": 17942
+ },
+ {
+ "epoch": 48.89100817438692,
+ "grad_norm": 2.778909683227539,
+ "learning_rate": 1.0843983428035372e-05,
+ "loss": 0.0584,
+ "step": 17943
+ },
+ {
+ "epoch": 48.89373297002725,
+ "grad_norm": 2.8630378246307373,
+ "learning_rate": 1.0843104078771485e-05,
+ "loss": 0.0537,
+ "step": 17944
+ },
+ {
+ "epoch": 48.89645776566758,
+ "grad_norm": 2.3671908378601074,
+ "learning_rate": 1.0842224722941528e-05,
+ "loss": 0.039,
+ "step": 17945
+ },
+ {
+ "epoch": 48.8991825613079,
+ "grad_norm": 2.8138668537139893,
+ "learning_rate": 1.0841345360552356e-05,
+ "loss": 0.0997,
+ "step": 17946
+ },
+ {
+ "epoch": 48.90190735694823,
+ "grad_norm": 1.9510724544525146,
+ "learning_rate": 1.084046599161081e-05,
+ "loss": 0.0369,
+ "step": 17947
+ },
+ {
+ "epoch": 48.904632152588555,
+ "grad_norm": 2.6869006156921387,
+ "learning_rate": 1.0839586616123743e-05,
+ "loss": 0.1036,
+ "step": 17948
+ },
+ {
+ "epoch": 48.90735694822888,
+ "grad_norm": 3.641577959060669,
+ "learning_rate": 1.0838707234098003e-05,
+ "loss": 0.0711,
+ "step": 17949
+ },
+ {
+ "epoch": 48.91008174386921,
+ "grad_norm": 4.093544006347656,
+ "learning_rate": 1.0837827845540438e-05,
+ "loss": 0.1162,
+ "step": 17950
+ },
+ {
+ "epoch": 48.91280653950954,
+ "grad_norm": 2.4413418769836426,
+ "learning_rate": 1.0836948450457894e-05,
+ "loss": 0.0738,
+ "step": 17951
+ },
+ {
+ "epoch": 48.915531335149865,
+ "grad_norm": 3.017456531524658,
+ "learning_rate": 1.0836069048857228e-05,
+ "loss": 0.1082,
+ "step": 17952
+ },
+ {
+ "epoch": 48.91825613079019,
+ "grad_norm": 3.821998357772827,
+ "learning_rate": 1.0835189640745279e-05,
+ "loss": 0.0749,
+ "step": 17953
+ },
+ {
+ "epoch": 48.920980926430516,
+ "grad_norm": 3.2439849376678467,
+ "learning_rate": 1.08343102261289e-05,
+ "loss": 0.1469,
+ "step": 17954
+ },
+ {
+ "epoch": 48.92370572207084,
+ "grad_norm": 2.5044164657592773,
+ "learning_rate": 1.0833430805014942e-05,
+ "loss": 0.0346,
+ "step": 17955
+ },
+ {
+ "epoch": 48.926430517711175,
+ "grad_norm": 3.585897445678711,
+ "learning_rate": 1.0832551377410252e-05,
+ "loss": 0.1867,
+ "step": 17956
+ },
+ {
+ "epoch": 48.9291553133515,
+ "grad_norm": 2.6106739044189453,
+ "learning_rate": 1.0831671943321675e-05,
+ "loss": 0.0501,
+ "step": 17957
+ },
+ {
+ "epoch": 48.93188010899183,
+ "grad_norm": 3.037031650543213,
+ "learning_rate": 1.0830792502756067e-05,
+ "loss": 0.0607,
+ "step": 17958
+ },
+ {
+ "epoch": 48.93460490463215,
+ "grad_norm": 2.613008737564087,
+ "learning_rate": 1.0829913055720268e-05,
+ "loss": 0.0481,
+ "step": 17959
+ },
+ {
+ "epoch": 48.93732970027248,
+ "grad_norm": 2.793196201324463,
+ "learning_rate": 1.082903360222114e-05,
+ "loss": 0.1378,
+ "step": 17960
+ },
+ {
+ "epoch": 48.940054495912804,
+ "grad_norm": 2.9219963550567627,
+ "learning_rate": 1.0828154142265521e-05,
+ "loss": 0.1228,
+ "step": 17961
+ },
+ {
+ "epoch": 48.94277929155314,
+ "grad_norm": 2.5682876110076904,
+ "learning_rate": 1.0827274675860266e-05,
+ "loss": 0.06,
+ "step": 17962
+ },
+ {
+ "epoch": 48.94550408719346,
+ "grad_norm": 2.5881314277648926,
+ "learning_rate": 1.0826395203012223e-05,
+ "loss": 0.0424,
+ "step": 17963
+ },
+ {
+ "epoch": 48.94822888283379,
+ "grad_norm": 3.5474681854248047,
+ "learning_rate": 1.0825515723728241e-05,
+ "loss": 0.1186,
+ "step": 17964
+ },
+ {
+ "epoch": 48.950953678474114,
+ "grad_norm": 3.666215658187866,
+ "learning_rate": 1.0824636238015167e-05,
+ "loss": 0.1099,
+ "step": 17965
+ },
+ {
+ "epoch": 48.95367847411444,
+ "grad_norm": 2.969519853591919,
+ "learning_rate": 1.0823756745879853e-05,
+ "loss": 0.0568,
+ "step": 17966
+ },
+ {
+ "epoch": 48.956403269754766,
+ "grad_norm": 2.1339099407196045,
+ "learning_rate": 1.0822877247329148e-05,
+ "loss": 0.1288,
+ "step": 17967
+ },
+ {
+ "epoch": 48.95912806539509,
+ "grad_norm": 2.9214351177215576,
+ "learning_rate": 1.0821997742369901e-05,
+ "loss": 0.0596,
+ "step": 17968
+ },
+ {
+ "epoch": 48.961852861035425,
+ "grad_norm": 2.629012107849121,
+ "learning_rate": 1.0821118231008963e-05,
+ "loss": 0.1311,
+ "step": 17969
+ },
+ {
+ "epoch": 48.96457765667575,
+ "grad_norm": 4.255996227264404,
+ "learning_rate": 1.0820238713253184e-05,
+ "loss": 0.035,
+ "step": 17970
+ },
+ {
+ "epoch": 48.967302452316076,
+ "grad_norm": 2.6998727321624756,
+ "learning_rate": 1.0819359189109409e-05,
+ "loss": 0.0892,
+ "step": 17971
+ },
+ {
+ "epoch": 48.9700272479564,
+ "grad_norm": 2.5916850566864014,
+ "learning_rate": 1.0818479658584495e-05,
+ "loss": 0.0528,
+ "step": 17972
+ },
+ {
+ "epoch": 48.97275204359673,
+ "grad_norm": 3.710885524749756,
+ "learning_rate": 1.0817600121685283e-05,
+ "loss": 0.0762,
+ "step": 17973
+ },
+ {
+ "epoch": 48.97547683923706,
+ "grad_norm": 2.649653911590576,
+ "learning_rate": 1.0816720578418634e-05,
+ "loss": 0.0693,
+ "step": 17974
+ },
+ {
+ "epoch": 48.97820163487739,
+ "grad_norm": 3.9290616512298584,
+ "learning_rate": 1.0815841028791385e-05,
+ "loss": 0.0612,
+ "step": 17975
+ },
+ {
+ "epoch": 48.98092643051771,
+ "grad_norm": 4.300586223602295,
+ "learning_rate": 1.0814961472810395e-05,
+ "loss": 0.0693,
+ "step": 17976
+ },
+ {
+ "epoch": 48.98365122615804,
+ "grad_norm": 3.4678168296813965,
+ "learning_rate": 1.081408191048251e-05,
+ "loss": 0.0954,
+ "step": 17977
+ },
+ {
+ "epoch": 48.986376021798364,
+ "grad_norm": 2.094954013824463,
+ "learning_rate": 1.0813202341814583e-05,
+ "loss": 0.0778,
+ "step": 17978
+ },
+ {
+ "epoch": 48.98910081743869,
+ "grad_norm": 4.051567554473877,
+ "learning_rate": 1.081232276681346e-05,
+ "loss": 0.1531,
+ "step": 17979
+ },
+ {
+ "epoch": 48.991825613079016,
+ "grad_norm": 2.7455790042877197,
+ "learning_rate": 1.0811443185485998e-05,
+ "loss": 0.0562,
+ "step": 17980
+ },
+ {
+ "epoch": 48.99455040871935,
+ "grad_norm": 3.4105377197265625,
+ "learning_rate": 1.0810563597839039e-05,
+ "loss": 0.1814,
+ "step": 17981
+ },
+ {
+ "epoch": 48.997275204359674,
+ "grad_norm": 3.8774077892303467,
+ "learning_rate": 1.0809684003879438e-05,
+ "loss": 0.0984,
+ "step": 17982
+ },
+ {
+ "epoch": 49.0,
+ "grad_norm": 2.2154862880706787,
+ "learning_rate": 1.0808804403614044e-05,
+ "loss": 0.1079,
+ "step": 17983
+ },
+ {
+ "epoch": 49.002724795640326,
+ "grad_norm": 4.0823493003845215,
+ "learning_rate": 1.0807924797049707e-05,
+ "loss": 0.2469,
+ "step": 17984
+ },
+ {
+ "epoch": 49.00544959128065,
+ "grad_norm": 2.1167755126953125,
+ "learning_rate": 1.0807045184193274e-05,
+ "loss": 0.1738,
+ "step": 17985
+ },
+ {
+ "epoch": 49.00817438692098,
+ "grad_norm": 2.260573387145996,
+ "learning_rate": 1.0806165565051606e-05,
+ "loss": 0.0427,
+ "step": 17986
+ },
+ {
+ "epoch": 49.01089918256131,
+ "grad_norm": 2.5716118812561035,
+ "learning_rate": 1.0805285939631543e-05,
+ "loss": 0.159,
+ "step": 17987
+ },
+ {
+ "epoch": 49.013623978201636,
+ "grad_norm": 2.685030937194824,
+ "learning_rate": 1.0804406307939938e-05,
+ "loss": 0.1053,
+ "step": 17988
+ },
+ {
+ "epoch": 49.01634877384196,
+ "grad_norm": 3.2969532012939453,
+ "learning_rate": 1.0803526669983644e-05,
+ "loss": 0.0851,
+ "step": 17989
+ },
+ {
+ "epoch": 49.01907356948229,
+ "grad_norm": 2.3806509971618652,
+ "learning_rate": 1.0802647025769511e-05,
+ "loss": 0.1075,
+ "step": 17990
+ },
+ {
+ "epoch": 49.02179836512261,
+ "grad_norm": 3.550131320953369,
+ "learning_rate": 1.0801767375304388e-05,
+ "loss": 0.2859,
+ "step": 17991
+ },
+ {
+ "epoch": 49.02452316076294,
+ "grad_norm": 3.319887399673462,
+ "learning_rate": 1.0800887718595127e-05,
+ "loss": 0.1037,
+ "step": 17992
+ },
+ {
+ "epoch": 49.02724795640327,
+ "grad_norm": 2.359722375869751,
+ "learning_rate": 1.0800008055648574e-05,
+ "loss": 0.0721,
+ "step": 17993
+ },
+ {
+ "epoch": 49.0299727520436,
+ "grad_norm": 2.858422040939331,
+ "learning_rate": 1.0799128386471589e-05,
+ "loss": 0.0946,
+ "step": 17994
+ },
+ {
+ "epoch": 49.032697547683924,
+ "grad_norm": 2.566873788833618,
+ "learning_rate": 1.0798248711071016e-05,
+ "loss": 0.1742,
+ "step": 17995
+ },
+ {
+ "epoch": 49.03542234332425,
+ "grad_norm": 2.390108823776245,
+ "learning_rate": 1.0797369029453708e-05,
+ "loss": 0.0631,
+ "step": 17996
+ },
+ {
+ "epoch": 49.038147138964575,
+ "grad_norm": 2.828551769256592,
+ "learning_rate": 1.0796489341626515e-05,
+ "loss": 0.0723,
+ "step": 17997
+ },
+ {
+ "epoch": 49.0408719346049,
+ "grad_norm": 4.140299320220947,
+ "learning_rate": 1.0795609647596292e-05,
+ "loss": 0.097,
+ "step": 17998
+ },
+ {
+ "epoch": 49.043596730245234,
+ "grad_norm": 2.3399221897125244,
+ "learning_rate": 1.0794729947369881e-05,
+ "loss": 0.0564,
+ "step": 17999
+ },
+ {
+ "epoch": 49.04632152588556,
+ "grad_norm": 2.424687623977661,
+ "learning_rate": 1.0793850240954145e-05,
+ "loss": 0.0402,
+ "step": 18000
+ },
+ {
+ "epoch": 49.049046321525886,
+ "grad_norm": 3.1560752391815186,
+ "learning_rate": 1.0792970528355925e-05,
+ "loss": 0.1748,
+ "step": 18001
+ },
+ {
+ "epoch": 49.05177111716621,
+ "grad_norm": 2.613957405090332,
+ "learning_rate": 1.0792090809582077e-05,
+ "loss": 0.066,
+ "step": 18002
+ },
+ {
+ "epoch": 49.05449591280654,
+ "grad_norm": 2.159745454788208,
+ "learning_rate": 1.0791211084639448e-05,
+ "loss": 0.0327,
+ "step": 18003
+ },
+ {
+ "epoch": 49.05722070844686,
+ "grad_norm": 3.88759183883667,
+ "learning_rate": 1.0790331353534898e-05,
+ "loss": 0.0786,
+ "step": 18004
+ },
+ {
+ "epoch": 49.059945504087196,
+ "grad_norm": 3.2257261276245117,
+ "learning_rate": 1.0789451616275268e-05,
+ "loss": 0.0624,
+ "step": 18005
+ },
+ {
+ "epoch": 49.06267029972752,
+ "grad_norm": 2.3023555278778076,
+ "learning_rate": 1.078857187286742e-05,
+ "loss": 0.0488,
+ "step": 18006
+ },
+ {
+ "epoch": 49.06539509536785,
+ "grad_norm": 2.3962624073028564,
+ "learning_rate": 1.0787692123318193e-05,
+ "loss": 0.1205,
+ "step": 18007
+ },
+ {
+ "epoch": 49.06811989100817,
+ "grad_norm": 9.462297439575195,
+ "learning_rate": 1.0786812367634452e-05,
+ "loss": 0.0472,
+ "step": 18008
+ },
+ {
+ "epoch": 49.0708446866485,
+ "grad_norm": 5.581429958343506,
+ "learning_rate": 1.0785932605823035e-05,
+ "loss": 0.0659,
+ "step": 18009
+ },
+ {
+ "epoch": 49.073569482288825,
+ "grad_norm": 1.627170443534851,
+ "learning_rate": 1.0785052837890805e-05,
+ "loss": 0.0461,
+ "step": 18010
+ },
+ {
+ "epoch": 49.07629427792916,
+ "grad_norm": 2.72904372215271,
+ "learning_rate": 1.0784173063844606e-05,
+ "loss": 0.1522,
+ "step": 18011
+ },
+ {
+ "epoch": 49.079019073569484,
+ "grad_norm": 2.3903868198394775,
+ "learning_rate": 1.0783293283691292e-05,
+ "loss": 0.0635,
+ "step": 18012
+ },
+ {
+ "epoch": 49.08174386920981,
+ "grad_norm": 2.04001784324646,
+ "learning_rate": 1.0782413497437716e-05,
+ "loss": 0.0633,
+ "step": 18013
+ },
+ {
+ "epoch": 49.084468664850135,
+ "grad_norm": 2.892338275909424,
+ "learning_rate": 1.078153370509073e-05,
+ "loss": 0.0505,
+ "step": 18014
+ },
+ {
+ "epoch": 49.08719346049046,
+ "grad_norm": 1.8432247638702393,
+ "learning_rate": 1.0780653906657183e-05,
+ "loss": 0.0243,
+ "step": 18015
+ },
+ {
+ "epoch": 49.08991825613079,
+ "grad_norm": 2.7124269008636475,
+ "learning_rate": 1.0779774102143932e-05,
+ "loss": 0.0579,
+ "step": 18016
+ },
+ {
+ "epoch": 49.09264305177112,
+ "grad_norm": 2.9684815406799316,
+ "learning_rate": 1.0778894291557822e-05,
+ "loss": 0.0946,
+ "step": 18017
+ },
+ {
+ "epoch": 49.095367847411445,
+ "grad_norm": 3.0524988174438477,
+ "learning_rate": 1.077801447490571e-05,
+ "loss": 0.0399,
+ "step": 18018
+ },
+ {
+ "epoch": 49.09809264305177,
+ "grad_norm": 2.952394485473633,
+ "learning_rate": 1.0777134652194444e-05,
+ "loss": 0.1112,
+ "step": 18019
+ },
+ {
+ "epoch": 49.1008174386921,
+ "grad_norm": 2.8768815994262695,
+ "learning_rate": 1.077625482343088e-05,
+ "loss": 0.1203,
+ "step": 18020
+ },
+ {
+ "epoch": 49.10354223433242,
+ "grad_norm": 3.3877227306365967,
+ "learning_rate": 1.077537498862187e-05,
+ "loss": 0.1263,
+ "step": 18021
+ },
+ {
+ "epoch": 49.10626702997275,
+ "grad_norm": 2.9158217906951904,
+ "learning_rate": 1.077449514777426e-05,
+ "loss": 0.0724,
+ "step": 18022
+ },
+ {
+ "epoch": 49.10899182561308,
+ "grad_norm": 2.7689499855041504,
+ "learning_rate": 1.0773615300894908e-05,
+ "loss": 0.1082,
+ "step": 18023
+ },
+ {
+ "epoch": 49.11171662125341,
+ "grad_norm": 3.1098785400390625,
+ "learning_rate": 1.0772735447990669e-05,
+ "loss": 0.0501,
+ "step": 18024
+ },
+ {
+ "epoch": 49.11444141689373,
+ "grad_norm": 2.5867416858673096,
+ "learning_rate": 1.0771855589068387e-05,
+ "loss": 0.0706,
+ "step": 18025
+ },
+ {
+ "epoch": 49.11716621253406,
+ "grad_norm": 2.5096120834350586,
+ "learning_rate": 1.0770975724134922e-05,
+ "loss": 0.0756,
+ "step": 18026
+ },
+ {
+ "epoch": 49.119891008174385,
+ "grad_norm": 2.935767889022827,
+ "learning_rate": 1.077009585319712e-05,
+ "loss": 0.0541,
+ "step": 18027
+ },
+ {
+ "epoch": 49.12261580381471,
+ "grad_norm": 2.2678866386413574,
+ "learning_rate": 1.0769215976261838e-05,
+ "loss": 0.1015,
+ "step": 18028
+ },
+ {
+ "epoch": 49.12534059945504,
+ "grad_norm": 2.1786959171295166,
+ "learning_rate": 1.0768336093335928e-05,
+ "loss": 0.0364,
+ "step": 18029
+ },
+ {
+ "epoch": 49.12806539509537,
+ "grad_norm": 3.8094286918640137,
+ "learning_rate": 1.0767456204426236e-05,
+ "loss": 0.13,
+ "step": 18030
+ },
+ {
+ "epoch": 49.130790190735695,
+ "grad_norm": 2.1899261474609375,
+ "learning_rate": 1.0766576309539624e-05,
+ "loss": 0.0344,
+ "step": 18031
+ },
+ {
+ "epoch": 49.13351498637602,
+ "grad_norm": 2.6382479667663574,
+ "learning_rate": 1.076569640868294e-05,
+ "loss": 0.0944,
+ "step": 18032
+ },
+ {
+ "epoch": 49.13623978201635,
+ "grad_norm": 2.410459280014038,
+ "learning_rate": 1.0764816501863038e-05,
+ "loss": 0.0659,
+ "step": 18033
+ },
+ {
+ "epoch": 49.13896457765667,
+ "grad_norm": 1.9222785234451294,
+ "learning_rate": 1.0763936589086765e-05,
+ "loss": 0.1209,
+ "step": 18034
+ },
+ {
+ "epoch": 49.141689373297005,
+ "grad_norm": 3.2226333618164062,
+ "learning_rate": 1.0763056670360983e-05,
+ "loss": 0.0635,
+ "step": 18035
+ },
+ {
+ "epoch": 49.14441416893733,
+ "grad_norm": 3.0227575302124023,
+ "learning_rate": 1.0762176745692536e-05,
+ "loss": 0.1021,
+ "step": 18036
+ },
+ {
+ "epoch": 49.14713896457766,
+ "grad_norm": 3.629605531692505,
+ "learning_rate": 1.0761296815088286e-05,
+ "loss": 0.1574,
+ "step": 18037
+ },
+ {
+ "epoch": 49.14986376021798,
+ "grad_norm": 1.9410324096679688,
+ "learning_rate": 1.0760416878555078e-05,
+ "loss": 0.1816,
+ "step": 18038
+ },
+ {
+ "epoch": 49.15258855585831,
+ "grad_norm": 2.0810742378234863,
+ "learning_rate": 1.075953693609977e-05,
+ "loss": 0.0406,
+ "step": 18039
+ },
+ {
+ "epoch": 49.155313351498634,
+ "grad_norm": 2.6079559326171875,
+ "learning_rate": 1.075865698772921e-05,
+ "loss": 0.1025,
+ "step": 18040
+ },
+ {
+ "epoch": 49.15803814713897,
+ "grad_norm": 3.135256767272949,
+ "learning_rate": 1.0757777033450257e-05,
+ "loss": 0.0629,
+ "step": 18041
+ },
+ {
+ "epoch": 49.16076294277929,
+ "grad_norm": 2.617326259613037,
+ "learning_rate": 1.0756897073269758e-05,
+ "loss": 0.0507,
+ "step": 18042
+ },
+ {
+ "epoch": 49.16348773841962,
+ "grad_norm": 1.9044108390808105,
+ "learning_rate": 1.0756017107194569e-05,
+ "loss": 0.0402,
+ "step": 18043
+ },
+ {
+ "epoch": 49.166212534059945,
+ "grad_norm": 2.8923892974853516,
+ "learning_rate": 1.0755137135231544e-05,
+ "loss": 0.134,
+ "step": 18044
+ },
+ {
+ "epoch": 49.16893732970027,
+ "grad_norm": 3.792111396789551,
+ "learning_rate": 1.0754257157387536e-05,
+ "loss": 0.1966,
+ "step": 18045
+ },
+ {
+ "epoch": 49.171662125340596,
+ "grad_norm": 1.9150642156600952,
+ "learning_rate": 1.0753377173669394e-05,
+ "loss": 0.0566,
+ "step": 18046
+ },
+ {
+ "epoch": 49.17438692098093,
+ "grad_norm": 6.026059627532959,
+ "learning_rate": 1.075249718408398e-05,
+ "loss": 0.0978,
+ "step": 18047
+ },
+ {
+ "epoch": 49.177111716621255,
+ "grad_norm": 2.7620849609375,
+ "learning_rate": 1.075161718863814e-05,
+ "loss": 0.0813,
+ "step": 18048
+ },
+ {
+ "epoch": 49.17983651226158,
+ "grad_norm": 2.233771800994873,
+ "learning_rate": 1.0750737187338729e-05,
+ "loss": 0.0892,
+ "step": 18049
+ },
+ {
+ "epoch": 49.182561307901906,
+ "grad_norm": 4.253141403198242,
+ "learning_rate": 1.0749857180192603e-05,
+ "loss": 0.0963,
+ "step": 18050
+ },
+ {
+ "epoch": 49.18528610354223,
+ "grad_norm": 2.8558902740478516,
+ "learning_rate": 1.0748977167206611e-05,
+ "loss": 0.0701,
+ "step": 18051
+ },
+ {
+ "epoch": 49.18801089918256,
+ "grad_norm": 8.496492385864258,
+ "learning_rate": 1.0748097148387609e-05,
+ "loss": 0.0757,
+ "step": 18052
+ },
+ {
+ "epoch": 49.19073569482289,
+ "grad_norm": 2.721975088119507,
+ "learning_rate": 1.0747217123742453e-05,
+ "loss": 0.09,
+ "step": 18053
+ },
+ {
+ "epoch": 49.19346049046322,
+ "grad_norm": 3.3704452514648438,
+ "learning_rate": 1.0746337093277987e-05,
+ "loss": 0.2617,
+ "step": 18054
+ },
+ {
+ "epoch": 49.19618528610354,
+ "grad_norm": 2.671886682510376,
+ "learning_rate": 1.0745457057001079e-05,
+ "loss": 0.0316,
+ "step": 18055
+ },
+ {
+ "epoch": 49.19891008174387,
+ "grad_norm": 2.4180612564086914,
+ "learning_rate": 1.0744577014918574e-05,
+ "loss": 0.0923,
+ "step": 18056
+ },
+ {
+ "epoch": 49.201634877384194,
+ "grad_norm": 2.3776164054870605,
+ "learning_rate": 1.0743696967037327e-05,
+ "loss": 0.1491,
+ "step": 18057
+ },
+ {
+ "epoch": 49.20435967302452,
+ "grad_norm": 2.2513551712036133,
+ "learning_rate": 1.074281691336419e-05,
+ "loss": 0.107,
+ "step": 18058
+ },
+ {
+ "epoch": 49.20708446866485,
+ "grad_norm": 2.501389265060425,
+ "learning_rate": 1.0741936853906023e-05,
+ "loss": 0.0492,
+ "step": 18059
+ },
+ {
+ "epoch": 49.20980926430518,
+ "grad_norm": 2.0730526447296143,
+ "learning_rate": 1.0741056788669671e-05,
+ "loss": 0.1828,
+ "step": 18060
+ },
+ {
+ "epoch": 49.212534059945504,
+ "grad_norm": 2.1425158977508545,
+ "learning_rate": 1.0740176717661997e-05,
+ "loss": 0.0474,
+ "step": 18061
+ },
+ {
+ "epoch": 49.21525885558583,
+ "grad_norm": 3.0483102798461914,
+ "learning_rate": 1.0739296640889844e-05,
+ "loss": 0.1276,
+ "step": 18062
+ },
+ {
+ "epoch": 49.217983651226156,
+ "grad_norm": 3.257850170135498,
+ "learning_rate": 1.073841655836008e-05,
+ "loss": 0.2463,
+ "step": 18063
+ },
+ {
+ "epoch": 49.22070844686648,
+ "grad_norm": 2.9637699127197266,
+ "learning_rate": 1.0737536470079545e-05,
+ "loss": 0.1176,
+ "step": 18064
+ },
+ {
+ "epoch": 49.223433242506815,
+ "grad_norm": 2.808152437210083,
+ "learning_rate": 1.0736656376055106e-05,
+ "loss": 0.0392,
+ "step": 18065
+ },
+ {
+ "epoch": 49.22615803814714,
+ "grad_norm": 2.766263484954834,
+ "learning_rate": 1.0735776276293605e-05,
+ "loss": 0.178,
+ "step": 18066
+ },
+ {
+ "epoch": 49.228882833787466,
+ "grad_norm": 2.4036030769348145,
+ "learning_rate": 1.0734896170801907e-05,
+ "loss": 0.0935,
+ "step": 18067
+ },
+ {
+ "epoch": 49.23160762942779,
+ "grad_norm": 11.932389259338379,
+ "learning_rate": 1.0734016059586858e-05,
+ "loss": 0.0402,
+ "step": 18068
+ },
+ {
+ "epoch": 49.23433242506812,
+ "grad_norm": 2.098254442214966,
+ "learning_rate": 1.0733135942655318e-05,
+ "loss": 0.1543,
+ "step": 18069
+ },
+ {
+ "epoch": 49.237057220708444,
+ "grad_norm": 2.2416603565216064,
+ "learning_rate": 1.0732255820014137e-05,
+ "loss": 0.0706,
+ "step": 18070
+ },
+ {
+ "epoch": 49.23978201634878,
+ "grad_norm": 2.219024658203125,
+ "learning_rate": 1.073137569167017e-05,
+ "loss": 0.2219,
+ "step": 18071
+ },
+ {
+ "epoch": 49.2425068119891,
+ "grad_norm": 1.4801074266433716,
+ "learning_rate": 1.0730495557630275e-05,
+ "loss": 0.024,
+ "step": 18072
+ },
+ {
+ "epoch": 49.24523160762943,
+ "grad_norm": 3.2994651794433594,
+ "learning_rate": 1.0729615417901305e-05,
+ "loss": 0.237,
+ "step": 18073
+ },
+ {
+ "epoch": 49.247956403269754,
+ "grad_norm": 3.032609462738037,
+ "learning_rate": 1.072873527249011e-05,
+ "loss": 0.1854,
+ "step": 18074
+ },
+ {
+ "epoch": 49.25068119891008,
+ "grad_norm": 2.448770046234131,
+ "learning_rate": 1.0727855121403551e-05,
+ "loss": 0.0707,
+ "step": 18075
+ },
+ {
+ "epoch": 49.253405994550405,
+ "grad_norm": 1.9964734315872192,
+ "learning_rate": 1.0726974964648478e-05,
+ "loss": 0.0408,
+ "step": 18076
+ },
+ {
+ "epoch": 49.25613079019074,
+ "grad_norm": 3.1488847732543945,
+ "learning_rate": 1.072609480223175e-05,
+ "loss": 0.0624,
+ "step": 18077
+ },
+ {
+ "epoch": 49.258855585831064,
+ "grad_norm": 1.6416841745376587,
+ "learning_rate": 1.0725214634160217e-05,
+ "loss": 0.031,
+ "step": 18078
+ },
+ {
+ "epoch": 49.26158038147139,
+ "grad_norm": 2.8256547451019287,
+ "learning_rate": 1.0724334460440735e-05,
+ "loss": 0.0895,
+ "step": 18079
+ },
+ {
+ "epoch": 49.264305177111716,
+ "grad_norm": 1.8936188220977783,
+ "learning_rate": 1.0723454281080161e-05,
+ "loss": 0.0364,
+ "step": 18080
+ },
+ {
+ "epoch": 49.26702997275204,
+ "grad_norm": 1.9424725770950317,
+ "learning_rate": 1.072257409608535e-05,
+ "loss": 0.074,
+ "step": 18081
+ },
+ {
+ "epoch": 49.26975476839237,
+ "grad_norm": 1.4117745161056519,
+ "learning_rate": 1.072169390546315e-05,
+ "loss": 0.0221,
+ "step": 18082
+ },
+ {
+ "epoch": 49.2724795640327,
+ "grad_norm": 2.9095077514648438,
+ "learning_rate": 1.0720813709220427e-05,
+ "loss": 0.0609,
+ "step": 18083
+ },
+ {
+ "epoch": 49.275204359673026,
+ "grad_norm": 6.850194931030273,
+ "learning_rate": 1.0719933507364027e-05,
+ "loss": 0.0719,
+ "step": 18084
+ },
+ {
+ "epoch": 49.27792915531335,
+ "grad_norm": 2.249072313308716,
+ "learning_rate": 1.071905329990081e-05,
+ "loss": 0.0478,
+ "step": 18085
+ },
+ {
+ "epoch": 49.28065395095368,
+ "grad_norm": 2.324329376220703,
+ "learning_rate": 1.0718173086837626e-05,
+ "loss": 0.1172,
+ "step": 18086
+ },
+ {
+ "epoch": 49.283378746594,
+ "grad_norm": 3.3625905513763428,
+ "learning_rate": 1.0717292868181335e-05,
+ "loss": 0.0849,
+ "step": 18087
+ },
+ {
+ "epoch": 49.28610354223433,
+ "grad_norm": 2.395294427871704,
+ "learning_rate": 1.0716412643938788e-05,
+ "loss": 0.3134,
+ "step": 18088
+ },
+ {
+ "epoch": 49.28882833787466,
+ "grad_norm": 2.809342384338379,
+ "learning_rate": 1.0715532414116844e-05,
+ "loss": 0.0693,
+ "step": 18089
+ },
+ {
+ "epoch": 49.29155313351499,
+ "grad_norm": 3.719207286834717,
+ "learning_rate": 1.0714652178722357e-05,
+ "loss": 0.0514,
+ "step": 18090
+ },
+ {
+ "epoch": 49.294277929155314,
+ "grad_norm": 2.224536657333374,
+ "learning_rate": 1.0713771937762183e-05,
+ "loss": 0.0377,
+ "step": 18091
+ },
+ {
+ "epoch": 49.29700272479564,
+ "grad_norm": 2.48049259185791,
+ "learning_rate": 1.0712891691243174e-05,
+ "loss": 0.0678,
+ "step": 18092
+ },
+ {
+ "epoch": 49.299727520435965,
+ "grad_norm": 2.665151834487915,
+ "learning_rate": 1.0712011439172189e-05,
+ "loss": 0.1075,
+ "step": 18093
+ },
+ {
+ "epoch": 49.30245231607629,
+ "grad_norm": 2.6266636848449707,
+ "learning_rate": 1.071113118155608e-05,
+ "loss": 0.1796,
+ "step": 18094
+ },
+ {
+ "epoch": 49.305177111716624,
+ "grad_norm": 3.689288377761841,
+ "learning_rate": 1.0710250918401709e-05,
+ "loss": 0.0598,
+ "step": 18095
+ },
+ {
+ "epoch": 49.30790190735695,
+ "grad_norm": 2.5031282901763916,
+ "learning_rate": 1.070937064971592e-05,
+ "loss": 0.0648,
+ "step": 18096
+ },
+ {
+ "epoch": 49.310626702997276,
+ "grad_norm": 2.3678388595581055,
+ "learning_rate": 1.0708490375505577e-05,
+ "loss": 0.0468,
+ "step": 18097
+ },
+ {
+ "epoch": 49.3133514986376,
+ "grad_norm": 1.6969726085662842,
+ "learning_rate": 1.0707610095777534e-05,
+ "loss": 0.0368,
+ "step": 18098
+ },
+ {
+ "epoch": 49.31607629427793,
+ "grad_norm": 27.373132705688477,
+ "learning_rate": 1.070672981053865e-05,
+ "loss": 0.0543,
+ "step": 18099
+ },
+ {
+ "epoch": 49.31880108991825,
+ "grad_norm": 3.0106987953186035,
+ "learning_rate": 1.0705849519795772e-05,
+ "loss": 0.0727,
+ "step": 18100
+ },
+ {
+ "epoch": 49.321525885558586,
+ "grad_norm": 2.8800997734069824,
+ "learning_rate": 1.0704969223555766e-05,
+ "loss": 0.0378,
+ "step": 18101
+ },
+ {
+ "epoch": 49.32425068119891,
+ "grad_norm": 2.3144326210021973,
+ "learning_rate": 1.0704088921825477e-05,
+ "loss": 0.0539,
+ "step": 18102
+ },
+ {
+ "epoch": 49.32697547683924,
+ "grad_norm": 2.4963059425354004,
+ "learning_rate": 1.0703208614611772e-05,
+ "loss": 0.0563,
+ "step": 18103
+ },
+ {
+ "epoch": 49.32970027247956,
+ "grad_norm": 2.9827358722686768,
+ "learning_rate": 1.0702328301921495e-05,
+ "loss": 0.1586,
+ "step": 18104
+ },
+ {
+ "epoch": 49.33242506811989,
+ "grad_norm": 4.122383117675781,
+ "learning_rate": 1.0701447983761508e-05,
+ "loss": 0.1399,
+ "step": 18105
+ },
+ {
+ "epoch": 49.335149863760215,
+ "grad_norm": 1.8213895559310913,
+ "learning_rate": 1.0700567660138672e-05,
+ "loss": 0.0429,
+ "step": 18106
+ },
+ {
+ "epoch": 49.33787465940055,
+ "grad_norm": 2.6293859481811523,
+ "learning_rate": 1.0699687331059834e-05,
+ "loss": 0.1354,
+ "step": 18107
+ },
+ {
+ "epoch": 49.34059945504087,
+ "grad_norm": 9.40906810760498,
+ "learning_rate": 1.0698806996531855e-05,
+ "loss": 0.0471,
+ "step": 18108
+ },
+ {
+ "epoch": 49.3433242506812,
+ "grad_norm": 3.1191518306732178,
+ "learning_rate": 1.0697926656561588e-05,
+ "loss": 0.1349,
+ "step": 18109
+ },
+ {
+ "epoch": 49.346049046321525,
+ "grad_norm": 2.6165785789489746,
+ "learning_rate": 1.069704631115589e-05,
+ "loss": 0.1027,
+ "step": 18110
+ },
+ {
+ "epoch": 49.34877384196185,
+ "grad_norm": 2.6145029067993164,
+ "learning_rate": 1.069616596032162e-05,
+ "loss": 0.1713,
+ "step": 18111
+ },
+ {
+ "epoch": 49.35149863760218,
+ "grad_norm": 5.416059494018555,
+ "learning_rate": 1.069528560406563e-05,
+ "loss": 0.2319,
+ "step": 18112
+ },
+ {
+ "epoch": 49.35422343324251,
+ "grad_norm": 2.9023492336273193,
+ "learning_rate": 1.069440524239478e-05,
+ "loss": 0.0866,
+ "step": 18113
+ },
+ {
+ "epoch": 49.356948228882835,
+ "grad_norm": 2.9564085006713867,
+ "learning_rate": 1.0693524875315924e-05,
+ "loss": 0.0864,
+ "step": 18114
+ },
+ {
+ "epoch": 49.35967302452316,
+ "grad_norm": 2.5526630878448486,
+ "learning_rate": 1.0692644502835915e-05,
+ "loss": 0.0345,
+ "step": 18115
+ },
+ {
+ "epoch": 49.36239782016349,
+ "grad_norm": 2.4949824810028076,
+ "learning_rate": 1.0691764124961618e-05,
+ "loss": 0.1504,
+ "step": 18116
+ },
+ {
+ "epoch": 49.36512261580381,
+ "grad_norm": 3.9554972648620605,
+ "learning_rate": 1.0690883741699882e-05,
+ "loss": 0.1738,
+ "step": 18117
+ },
+ {
+ "epoch": 49.36784741144414,
+ "grad_norm": 3.2877514362335205,
+ "learning_rate": 1.0690003353057567e-05,
+ "loss": 0.0506,
+ "step": 18118
+ },
+ {
+ "epoch": 49.37057220708447,
+ "grad_norm": 2.9768662452697754,
+ "learning_rate": 1.0689122959041524e-05,
+ "loss": 0.2857,
+ "step": 18119
+ },
+ {
+ "epoch": 49.3732970027248,
+ "grad_norm": 2.892071485519409,
+ "learning_rate": 1.068824255965862e-05,
+ "loss": 0.1697,
+ "step": 18120
+ },
+ {
+ "epoch": 49.37602179836512,
+ "grad_norm": 8.649165153503418,
+ "learning_rate": 1.0687362154915699e-05,
+ "loss": 0.11,
+ "step": 18121
+ },
+ {
+ "epoch": 49.37874659400545,
+ "grad_norm": 2.5021755695343018,
+ "learning_rate": 1.0686481744819626e-05,
+ "loss": 0.063,
+ "step": 18122
+ },
+ {
+ "epoch": 49.381471389645775,
+ "grad_norm": 2.984523057937622,
+ "learning_rate": 1.0685601329377255e-05,
+ "loss": 0.1252,
+ "step": 18123
+ },
+ {
+ "epoch": 49.3841961852861,
+ "grad_norm": 4.922848224639893,
+ "learning_rate": 1.0684720908595445e-05,
+ "loss": 0.1627,
+ "step": 18124
+ },
+ {
+ "epoch": 49.38692098092643,
+ "grad_norm": 4.668880462646484,
+ "learning_rate": 1.0683840482481046e-05,
+ "loss": 0.0936,
+ "step": 18125
+ },
+ {
+ "epoch": 49.38964577656676,
+ "grad_norm": 2.26086163520813,
+ "learning_rate": 1.0682960051040924e-05,
+ "loss": 0.0596,
+ "step": 18126
+ },
+ {
+ "epoch": 49.392370572207085,
+ "grad_norm": 2.7219901084899902,
+ "learning_rate": 1.068207961428193e-05,
+ "loss": 0.102,
+ "step": 18127
+ },
+ {
+ "epoch": 49.39509536784741,
+ "grad_norm": 3.0982301235198975,
+ "learning_rate": 1.0681199172210923e-05,
+ "loss": 0.08,
+ "step": 18128
+ },
+ {
+ "epoch": 49.39782016348774,
+ "grad_norm": 11.979656219482422,
+ "learning_rate": 1.0680318724834758e-05,
+ "loss": 0.1497,
+ "step": 18129
+ },
+ {
+ "epoch": 49.40054495912806,
+ "grad_norm": 3.9809603691101074,
+ "learning_rate": 1.0679438272160292e-05,
+ "loss": 0.0561,
+ "step": 18130
+ },
+ {
+ "epoch": 49.403269754768395,
+ "grad_norm": 4.239670753479004,
+ "learning_rate": 1.067855781419438e-05,
+ "loss": 0.1513,
+ "step": 18131
+ },
+ {
+ "epoch": 49.40599455040872,
+ "grad_norm": 3.0404295921325684,
+ "learning_rate": 1.0677677350943885e-05,
+ "loss": 0.0689,
+ "step": 18132
+ },
+ {
+ "epoch": 49.40871934604905,
+ "grad_norm": 4.467679500579834,
+ "learning_rate": 1.0676796882415658e-05,
+ "loss": 0.0692,
+ "step": 18133
+ },
+ {
+ "epoch": 49.41144414168937,
+ "grad_norm": 2.3008649349212646,
+ "learning_rate": 1.0675916408616562e-05,
+ "loss": 0.068,
+ "step": 18134
+ },
+ {
+ "epoch": 49.4141689373297,
+ "grad_norm": 4.015974998474121,
+ "learning_rate": 1.0675035929553448e-05,
+ "loss": 0.0655,
+ "step": 18135
+ },
+ {
+ "epoch": 49.416893732970024,
+ "grad_norm": 3.3710062503814697,
+ "learning_rate": 1.0674155445233179e-05,
+ "loss": 0.0867,
+ "step": 18136
+ },
+ {
+ "epoch": 49.41961852861036,
+ "grad_norm": 14.047161102294922,
+ "learning_rate": 1.0673274955662604e-05,
+ "loss": 0.0766,
+ "step": 18137
+ },
+ {
+ "epoch": 49.42234332425068,
+ "grad_norm": 3.375401258468628,
+ "learning_rate": 1.067239446084859e-05,
+ "loss": 0.1224,
+ "step": 18138
+ },
+ {
+ "epoch": 49.42506811989101,
+ "grad_norm": 3.1120481491088867,
+ "learning_rate": 1.0671513960797988e-05,
+ "loss": 0.0884,
+ "step": 18139
+ },
+ {
+ "epoch": 49.427792915531334,
+ "grad_norm": 3.219355344772339,
+ "learning_rate": 1.0670633455517656e-05,
+ "loss": 0.0827,
+ "step": 18140
+ },
+ {
+ "epoch": 49.43051771117166,
+ "grad_norm": 2.8637678623199463,
+ "learning_rate": 1.0669752945014452e-05,
+ "loss": 0.0467,
+ "step": 18141
+ },
+ {
+ "epoch": 49.433242506811986,
+ "grad_norm": 4.707312107086182,
+ "learning_rate": 1.0668872429295236e-05,
+ "loss": 0.1172,
+ "step": 18142
+ },
+ {
+ "epoch": 49.43596730245232,
+ "grad_norm": 3.4437966346740723,
+ "learning_rate": 1.0667991908366861e-05,
+ "loss": 0.055,
+ "step": 18143
+ },
+ {
+ "epoch": 49.438692098092645,
+ "grad_norm": 3.818251132965088,
+ "learning_rate": 1.0667111382236187e-05,
+ "loss": 0.0816,
+ "step": 18144
+ },
+ {
+ "epoch": 49.44141689373297,
+ "grad_norm": 2.4278759956359863,
+ "learning_rate": 1.0666230850910071e-05,
+ "loss": 0.0804,
+ "step": 18145
+ },
+ {
+ "epoch": 49.444141689373296,
+ "grad_norm": 3.1228320598602295,
+ "learning_rate": 1.0665350314395373e-05,
+ "loss": 0.085,
+ "step": 18146
+ },
+ {
+ "epoch": 49.44686648501362,
+ "grad_norm": 5.196975231170654,
+ "learning_rate": 1.0664469772698945e-05,
+ "loss": 0.0655,
+ "step": 18147
+ },
+ {
+ "epoch": 49.44959128065395,
+ "grad_norm": 2.759883165359497,
+ "learning_rate": 1.0663589225827648e-05,
+ "loss": 0.0847,
+ "step": 18148
+ },
+ {
+ "epoch": 49.45231607629428,
+ "grad_norm": 3.022484540939331,
+ "learning_rate": 1.066270867378834e-05,
+ "loss": 0.0875,
+ "step": 18149
+ },
+ {
+ "epoch": 49.45504087193461,
+ "grad_norm": 2.3835649490356445,
+ "learning_rate": 1.0661828116587878e-05,
+ "loss": 0.0827,
+ "step": 18150
+ },
+ {
+ "epoch": 49.45776566757493,
+ "grad_norm": 3.2728209495544434,
+ "learning_rate": 1.066094755423312e-05,
+ "loss": 0.0604,
+ "step": 18151
+ },
+ {
+ "epoch": 49.46049046321526,
+ "grad_norm": 3.524674892425537,
+ "learning_rate": 1.0660066986730925e-05,
+ "loss": 0.1225,
+ "step": 18152
+ },
+ {
+ "epoch": 49.463215258855584,
+ "grad_norm": 2.3928911685943604,
+ "learning_rate": 1.0659186414088148e-05,
+ "loss": 0.0725,
+ "step": 18153
+ },
+ {
+ "epoch": 49.46594005449591,
+ "grad_norm": 2.408026695251465,
+ "learning_rate": 1.0658305836311651e-05,
+ "loss": 0.1208,
+ "step": 18154
+ },
+ {
+ "epoch": 49.46866485013624,
+ "grad_norm": 6.224653720855713,
+ "learning_rate": 1.0657425253408287e-05,
+ "loss": 0.1466,
+ "step": 18155
+ },
+ {
+ "epoch": 49.47138964577657,
+ "grad_norm": 2.678178310394287,
+ "learning_rate": 1.0656544665384918e-05,
+ "loss": 0.0492,
+ "step": 18156
+ },
+ {
+ "epoch": 49.474114441416894,
+ "grad_norm": 2.8446311950683594,
+ "learning_rate": 1.0655664072248395e-05,
+ "loss": 0.049,
+ "step": 18157
+ },
+ {
+ "epoch": 49.47683923705722,
+ "grad_norm": 2.5954911708831787,
+ "learning_rate": 1.0654783474005587e-05,
+ "loss": 0.0911,
+ "step": 18158
+ },
+ {
+ "epoch": 49.479564032697546,
+ "grad_norm": 2.9927210807800293,
+ "learning_rate": 1.0653902870663344e-05,
+ "loss": 0.0592,
+ "step": 18159
+ },
+ {
+ "epoch": 49.48228882833787,
+ "grad_norm": 3.0707528591156006,
+ "learning_rate": 1.0653022262228527e-05,
+ "loss": 0.1814,
+ "step": 18160
+ },
+ {
+ "epoch": 49.485013623978205,
+ "grad_norm": 2.985183000564575,
+ "learning_rate": 1.0652141648707994e-05,
+ "loss": 0.1076,
+ "step": 18161
+ },
+ {
+ "epoch": 49.48773841961853,
+ "grad_norm": 2.7508254051208496,
+ "learning_rate": 1.0651261030108603e-05,
+ "loss": 0.0543,
+ "step": 18162
+ },
+ {
+ "epoch": 49.490463215258856,
+ "grad_norm": 3.5518839359283447,
+ "learning_rate": 1.065038040643721e-05,
+ "loss": 0.0656,
+ "step": 18163
+ },
+ {
+ "epoch": 49.49318801089918,
+ "grad_norm": 2.823397397994995,
+ "learning_rate": 1.064949977770068e-05,
+ "loss": 0.1714,
+ "step": 18164
+ },
+ {
+ "epoch": 49.49591280653951,
+ "grad_norm": 2.761187791824341,
+ "learning_rate": 1.064861914390586e-05,
+ "loss": 0.0343,
+ "step": 18165
+ },
+ {
+ "epoch": 49.49863760217983,
+ "grad_norm": 3.510981321334839,
+ "learning_rate": 1.064773850505962e-05,
+ "loss": 0.1732,
+ "step": 18166
+ },
+ {
+ "epoch": 49.50136239782017,
+ "grad_norm": 3.217832088470459,
+ "learning_rate": 1.0646857861168811e-05,
+ "loss": 0.1439,
+ "step": 18167
+ },
+ {
+ "epoch": 49.50408719346049,
+ "grad_norm": 2.917675256729126,
+ "learning_rate": 1.0645977212240297e-05,
+ "loss": 0.1126,
+ "step": 18168
+ },
+ {
+ "epoch": 49.50681198910082,
+ "grad_norm": 2.300577402114868,
+ "learning_rate": 1.064509655828093e-05,
+ "loss": 0.1585,
+ "step": 18169
+ },
+ {
+ "epoch": 49.509536784741144,
+ "grad_norm": 2.931518793106079,
+ "learning_rate": 1.0644215899297574e-05,
+ "loss": 0.2066,
+ "step": 18170
+ },
+ {
+ "epoch": 49.51226158038147,
+ "grad_norm": 2.0847439765930176,
+ "learning_rate": 1.0643335235297084e-05,
+ "loss": 0.0484,
+ "step": 18171
+ },
+ {
+ "epoch": 49.514986376021795,
+ "grad_norm": 3.0077829360961914,
+ "learning_rate": 1.0642454566286319e-05,
+ "loss": 0.1994,
+ "step": 18172
+ },
+ {
+ "epoch": 49.51771117166213,
+ "grad_norm": 3.645716667175293,
+ "learning_rate": 1.064157389227214e-05,
+ "loss": 0.1204,
+ "step": 18173
+ },
+ {
+ "epoch": 49.520435967302454,
+ "grad_norm": 3.1746840476989746,
+ "learning_rate": 1.0640693213261404e-05,
+ "loss": 0.0663,
+ "step": 18174
+ },
+ {
+ "epoch": 49.52316076294278,
+ "grad_norm": 1.9686912298202515,
+ "learning_rate": 1.0639812529260967e-05,
+ "loss": 0.0313,
+ "step": 18175
+ },
+ {
+ "epoch": 49.525885558583106,
+ "grad_norm": 2.844243049621582,
+ "learning_rate": 1.0638931840277696e-05,
+ "loss": 0.1209,
+ "step": 18176
+ },
+ {
+ "epoch": 49.52861035422343,
+ "grad_norm": 2.0375542640686035,
+ "learning_rate": 1.063805114631844e-05,
+ "loss": 0.1229,
+ "step": 18177
+ },
+ {
+ "epoch": 49.53133514986376,
+ "grad_norm": 1.8174880743026733,
+ "learning_rate": 1.0637170447390063e-05,
+ "loss": 0.0255,
+ "step": 18178
+ },
+ {
+ "epoch": 49.53405994550409,
+ "grad_norm": 2.335846424102783,
+ "learning_rate": 1.0636289743499424e-05,
+ "loss": 0.1099,
+ "step": 18179
+ },
+ {
+ "epoch": 49.536784741144416,
+ "grad_norm": 2.112422227859497,
+ "learning_rate": 1.0635409034653383e-05,
+ "loss": 0.0429,
+ "step": 18180
+ },
+ {
+ "epoch": 49.53950953678474,
+ "grad_norm": 3.0569186210632324,
+ "learning_rate": 1.0634528320858792e-05,
+ "loss": 0.0669,
+ "step": 18181
+ },
+ {
+ "epoch": 49.54223433242507,
+ "grad_norm": 2.718153238296509,
+ "learning_rate": 1.0633647602122521e-05,
+ "loss": 0.0383,
+ "step": 18182
+ },
+ {
+ "epoch": 49.54495912806539,
+ "grad_norm": 2.56766939163208,
+ "learning_rate": 1.0632766878451416e-05,
+ "loss": 0.1541,
+ "step": 18183
+ },
+ {
+ "epoch": 49.54768392370572,
+ "grad_norm": 3.433192253112793,
+ "learning_rate": 1.0631886149852347e-05,
+ "loss": 0.1097,
+ "step": 18184
+ },
+ {
+ "epoch": 49.55040871934605,
+ "grad_norm": 2.9851059913635254,
+ "learning_rate": 1.0631005416332168e-05,
+ "loss": 0.056,
+ "step": 18185
+ },
+ {
+ "epoch": 49.55313351498638,
+ "grad_norm": 3.850802421569824,
+ "learning_rate": 1.063012467789774e-05,
+ "loss": 0.1127,
+ "step": 18186
+ },
+ {
+ "epoch": 49.555858310626704,
+ "grad_norm": 2.7492735385894775,
+ "learning_rate": 1.0629243934555919e-05,
+ "loss": 0.1277,
+ "step": 18187
+ },
+ {
+ "epoch": 49.55858310626703,
+ "grad_norm": 2.8928418159484863,
+ "learning_rate": 1.062836318631357e-05,
+ "loss": 0.1575,
+ "step": 18188
+ },
+ {
+ "epoch": 49.561307901907355,
+ "grad_norm": 2.9919307231903076,
+ "learning_rate": 1.0627482433177547e-05,
+ "loss": 0.2781,
+ "step": 18189
+ },
+ {
+ "epoch": 49.56403269754768,
+ "grad_norm": 3.150834321975708,
+ "learning_rate": 1.0626601675154713e-05,
+ "loss": 0.0925,
+ "step": 18190
+ },
+ {
+ "epoch": 49.566757493188014,
+ "grad_norm": 2.0984275341033936,
+ "learning_rate": 1.0625720912251923e-05,
+ "loss": 0.0662,
+ "step": 18191
+ },
+ {
+ "epoch": 49.56948228882834,
+ "grad_norm": 2.429816246032715,
+ "learning_rate": 1.0624840144476036e-05,
+ "loss": 0.0581,
+ "step": 18192
+ },
+ {
+ "epoch": 49.572207084468666,
+ "grad_norm": 5.370753765106201,
+ "learning_rate": 1.0623959371833918e-05,
+ "loss": 0.0696,
+ "step": 18193
+ },
+ {
+ "epoch": 49.57493188010899,
+ "grad_norm": 2.522432565689087,
+ "learning_rate": 1.0623078594332424e-05,
+ "loss": 0.0848,
+ "step": 18194
+ },
+ {
+ "epoch": 49.57765667574932,
+ "grad_norm": 2.896318197250366,
+ "learning_rate": 1.0622197811978414e-05,
+ "loss": 0.045,
+ "step": 18195
+ },
+ {
+ "epoch": 49.58038147138964,
+ "grad_norm": 2.4604084491729736,
+ "learning_rate": 1.0621317024778747e-05,
+ "loss": 0.157,
+ "step": 18196
+ },
+ {
+ "epoch": 49.583106267029976,
+ "grad_norm": 2.632185220718384,
+ "learning_rate": 1.062043623274028e-05,
+ "loss": 0.0769,
+ "step": 18197
+ },
+ {
+ "epoch": 49.5858310626703,
+ "grad_norm": 2.654392719268799,
+ "learning_rate": 1.0619555435869881e-05,
+ "loss": 0.0414,
+ "step": 18198
+ },
+ {
+ "epoch": 49.58855585831063,
+ "grad_norm": 2.8149254322052,
+ "learning_rate": 1.0618674634174402e-05,
+ "loss": 0.0751,
+ "step": 18199
+ },
+ {
+ "epoch": 49.59128065395095,
+ "grad_norm": 2.6992998123168945,
+ "learning_rate": 1.0617793827660703e-05,
+ "loss": 0.0951,
+ "step": 18200
+ },
+ {
+ "epoch": 49.59400544959128,
+ "grad_norm": 2.7289037704467773,
+ "learning_rate": 1.0616913016335648e-05,
+ "loss": 0.04,
+ "step": 18201
+ },
+ {
+ "epoch": 49.596730245231605,
+ "grad_norm": 2.868316888809204,
+ "learning_rate": 1.0616032200206094e-05,
+ "loss": 0.1297,
+ "step": 18202
+ },
+ {
+ "epoch": 49.59945504087194,
+ "grad_norm": 3.9736900329589844,
+ "learning_rate": 1.0615151379278901e-05,
+ "loss": 0.1122,
+ "step": 18203
+ },
+ {
+ "epoch": 49.60217983651226,
+ "grad_norm": 2.5158445835113525,
+ "learning_rate": 1.0614270553560928e-05,
+ "loss": 0.0475,
+ "step": 18204
+ },
+ {
+ "epoch": 49.60490463215259,
+ "grad_norm": 2.6366872787475586,
+ "learning_rate": 1.0613389723059037e-05,
+ "loss": 0.0751,
+ "step": 18205
+ },
+ {
+ "epoch": 49.607629427792915,
+ "grad_norm": 2.116210460662842,
+ "learning_rate": 1.0612508887780085e-05,
+ "loss": 0.0915,
+ "step": 18206
+ },
+ {
+ "epoch": 49.61035422343324,
+ "grad_norm": 5.195136547088623,
+ "learning_rate": 1.0611628047730935e-05,
+ "loss": 0.0931,
+ "step": 18207
+ },
+ {
+ "epoch": 49.61307901907357,
+ "grad_norm": 1.9826781749725342,
+ "learning_rate": 1.0610747202918442e-05,
+ "loss": 0.0528,
+ "step": 18208
+ },
+ {
+ "epoch": 49.6158038147139,
+ "grad_norm": 4.925271987915039,
+ "learning_rate": 1.0609866353349474e-05,
+ "loss": 0.0743,
+ "step": 18209
+ },
+ {
+ "epoch": 49.618528610354225,
+ "grad_norm": 1.7959606647491455,
+ "learning_rate": 1.0608985499030884e-05,
+ "loss": 0.0487,
+ "step": 18210
+ },
+ {
+ "epoch": 49.62125340599455,
+ "grad_norm": 3.011258602142334,
+ "learning_rate": 1.0608104639969535e-05,
+ "loss": 0.0686,
+ "step": 18211
+ },
+ {
+ "epoch": 49.62397820163488,
+ "grad_norm": 2.5110697746276855,
+ "learning_rate": 1.0607223776172287e-05,
+ "loss": 0.1906,
+ "step": 18212
+ },
+ {
+ "epoch": 49.6267029972752,
+ "grad_norm": 3.3020517826080322,
+ "learning_rate": 1.0606342907645998e-05,
+ "loss": 0.0757,
+ "step": 18213
+ },
+ {
+ "epoch": 49.62942779291553,
+ "grad_norm": 2.9320969581604004,
+ "learning_rate": 1.0605462034397532e-05,
+ "loss": 0.0678,
+ "step": 18214
+ },
+ {
+ "epoch": 49.63215258855586,
+ "grad_norm": 2.5724847316741943,
+ "learning_rate": 1.060458115643375e-05,
+ "loss": 0.1102,
+ "step": 18215
+ },
+ {
+ "epoch": 49.63487738419619,
+ "grad_norm": 2.44553279876709,
+ "learning_rate": 1.0603700273761504e-05,
+ "loss": 0.0507,
+ "step": 18216
+ },
+ {
+ "epoch": 49.63760217983651,
+ "grad_norm": 2.7090349197387695,
+ "learning_rate": 1.060281938638766e-05,
+ "loss": 0.0447,
+ "step": 18217
+ },
+ {
+ "epoch": 49.64032697547684,
+ "grad_norm": 3.240654706954956,
+ "learning_rate": 1.060193849431908e-05,
+ "loss": 0.0835,
+ "step": 18218
+ },
+ {
+ "epoch": 49.643051771117165,
+ "grad_norm": 2.8823654651641846,
+ "learning_rate": 1.0601057597562623e-05,
+ "loss": 0.1293,
+ "step": 18219
+ },
+ {
+ "epoch": 49.64577656675749,
+ "grad_norm": 2.7110345363616943,
+ "learning_rate": 1.0600176696125146e-05,
+ "loss": 0.0418,
+ "step": 18220
+ },
+ {
+ "epoch": 49.64850136239782,
+ "grad_norm": 3.6128015518188477,
+ "learning_rate": 1.0599295790013515e-05,
+ "loss": 0.079,
+ "step": 18221
+ },
+ {
+ "epoch": 49.65122615803815,
+ "grad_norm": 2.660897731781006,
+ "learning_rate": 1.0598414879234587e-05,
+ "loss": 0.1703,
+ "step": 18222
+ },
+ {
+ "epoch": 49.653950953678475,
+ "grad_norm": 2.3718464374542236,
+ "learning_rate": 1.0597533963795223e-05,
+ "loss": 0.0402,
+ "step": 18223
+ },
+ {
+ "epoch": 49.6566757493188,
+ "grad_norm": 2.32999324798584,
+ "learning_rate": 1.0596653043702283e-05,
+ "loss": 0.0633,
+ "step": 18224
+ },
+ {
+ "epoch": 49.65940054495913,
+ "grad_norm": 3.236966371536255,
+ "learning_rate": 1.0595772118962631e-05,
+ "loss": 0.1858,
+ "step": 18225
+ },
+ {
+ "epoch": 49.66212534059945,
+ "grad_norm": 2.72224760055542,
+ "learning_rate": 1.0594891189583119e-05,
+ "loss": 0.0434,
+ "step": 18226
+ },
+ {
+ "epoch": 49.664850136239785,
+ "grad_norm": 3.0341007709503174,
+ "learning_rate": 1.059401025557062e-05,
+ "loss": 0.0948,
+ "step": 18227
+ },
+ {
+ "epoch": 49.66757493188011,
+ "grad_norm": 2.1307897567749023,
+ "learning_rate": 1.0593129316931985e-05,
+ "loss": 0.0723,
+ "step": 18228
+ },
+ {
+ "epoch": 49.67029972752044,
+ "grad_norm": 3.4893736839294434,
+ "learning_rate": 1.0592248373674078e-05,
+ "loss": 0.1208,
+ "step": 18229
+ },
+ {
+ "epoch": 49.67302452316076,
+ "grad_norm": 2.3940815925598145,
+ "learning_rate": 1.059136742580376e-05,
+ "loss": 0.0311,
+ "step": 18230
+ },
+ {
+ "epoch": 49.67574931880109,
+ "grad_norm": 2.583322525024414,
+ "learning_rate": 1.0590486473327894e-05,
+ "loss": 0.0822,
+ "step": 18231
+ },
+ {
+ "epoch": 49.678474114441414,
+ "grad_norm": 3.374169111251831,
+ "learning_rate": 1.0589605516253333e-05,
+ "loss": 0.1504,
+ "step": 18232
+ },
+ {
+ "epoch": 49.68119891008175,
+ "grad_norm": 2.542870283126831,
+ "learning_rate": 1.058872455458695e-05,
+ "loss": 0.0608,
+ "step": 18233
+ },
+ {
+ "epoch": 49.68392370572207,
+ "grad_norm": 2.6383116245269775,
+ "learning_rate": 1.0587843588335592e-05,
+ "loss": 0.23,
+ "step": 18234
+ },
+ {
+ "epoch": 49.6866485013624,
+ "grad_norm": 2.8919520378112793,
+ "learning_rate": 1.058696261750613e-05,
+ "loss": 0.1063,
+ "step": 18235
+ },
+ {
+ "epoch": 49.689373297002724,
+ "grad_norm": 3.709644079208374,
+ "learning_rate": 1.0586081642105423e-05,
+ "loss": 0.1499,
+ "step": 18236
+ },
+ {
+ "epoch": 49.69209809264305,
+ "grad_norm": 2.73941969871521,
+ "learning_rate": 1.0585200662140334e-05,
+ "loss": 0.2553,
+ "step": 18237
+ },
+ {
+ "epoch": 49.694822888283376,
+ "grad_norm": 2.6044528484344482,
+ "learning_rate": 1.0584319677617716e-05,
+ "loss": 0.0951,
+ "step": 18238
+ },
+ {
+ "epoch": 49.69754768392371,
+ "grad_norm": 2.6637303829193115,
+ "learning_rate": 1.0583438688544438e-05,
+ "loss": 0.086,
+ "step": 18239
+ },
+ {
+ "epoch": 49.700272479564035,
+ "grad_norm": 5.2626752853393555,
+ "learning_rate": 1.0582557694927356e-05,
+ "loss": 0.069,
+ "step": 18240
+ },
+ {
+ "epoch": 49.70299727520436,
+ "grad_norm": 2.814640998840332,
+ "learning_rate": 1.0581676696773336e-05,
+ "loss": 0.2491,
+ "step": 18241
+ },
+ {
+ "epoch": 49.705722070844686,
+ "grad_norm": 2.3116977214813232,
+ "learning_rate": 1.0580795694089234e-05,
+ "loss": 0.0629,
+ "step": 18242
+ },
+ {
+ "epoch": 49.70844686648501,
+ "grad_norm": 2.7548296451568604,
+ "learning_rate": 1.0579914686881914e-05,
+ "loss": 0.0348,
+ "step": 18243
+ },
+ {
+ "epoch": 49.71117166212534,
+ "grad_norm": 2.191967248916626,
+ "learning_rate": 1.0579033675158236e-05,
+ "loss": 0.0553,
+ "step": 18244
+ },
+ {
+ "epoch": 49.71389645776567,
+ "grad_norm": 3.665259599685669,
+ "learning_rate": 1.0578152658925066e-05,
+ "loss": 0.1714,
+ "step": 18245
+ },
+ {
+ "epoch": 49.716621253406,
+ "grad_norm": 2.5727107524871826,
+ "learning_rate": 1.0577271638189261e-05,
+ "loss": 0.1359,
+ "step": 18246
+ },
+ {
+ "epoch": 49.71934604904632,
+ "grad_norm": 5.302357196807861,
+ "learning_rate": 1.0576390612957683e-05,
+ "loss": 0.0681,
+ "step": 18247
+ },
+ {
+ "epoch": 49.72207084468665,
+ "grad_norm": 2.1885123252868652,
+ "learning_rate": 1.057550958323719e-05,
+ "loss": 0.0361,
+ "step": 18248
+ },
+ {
+ "epoch": 49.724795640326974,
+ "grad_norm": 2.7223637104034424,
+ "learning_rate": 1.0574628549034653e-05,
+ "loss": 0.0914,
+ "step": 18249
+ },
+ {
+ "epoch": 49.7275204359673,
+ "grad_norm": 2.37020206451416,
+ "learning_rate": 1.0573747510356923e-05,
+ "loss": 0.0732,
+ "step": 18250
+ },
+ {
+ "epoch": 49.73024523160763,
+ "grad_norm": 2.7177391052246094,
+ "learning_rate": 1.0572866467210865e-05,
+ "loss": 0.0689,
+ "step": 18251
+ },
+ {
+ "epoch": 49.73297002724796,
+ "grad_norm": 2.4021239280700684,
+ "learning_rate": 1.0571985419603343e-05,
+ "loss": 0.1432,
+ "step": 18252
+ },
+ {
+ "epoch": 49.735694822888284,
+ "grad_norm": 3.0656371116638184,
+ "learning_rate": 1.0571104367541218e-05,
+ "loss": 0.1586,
+ "step": 18253
+ },
+ {
+ "epoch": 49.73841961852861,
+ "grad_norm": 3.2765085697174072,
+ "learning_rate": 1.0570223311031349e-05,
+ "loss": 0.0689,
+ "step": 18254
+ },
+ {
+ "epoch": 49.741144414168936,
+ "grad_norm": 2.5492563247680664,
+ "learning_rate": 1.0569342250080601e-05,
+ "loss": 0.0425,
+ "step": 18255
+ },
+ {
+ "epoch": 49.74386920980926,
+ "grad_norm": 3.220200538635254,
+ "learning_rate": 1.0568461184695833e-05,
+ "loss": 0.1811,
+ "step": 18256
+ },
+ {
+ "epoch": 49.746594005449595,
+ "grad_norm": 2.745443820953369,
+ "learning_rate": 1.056758011488391e-05,
+ "loss": 0.1427,
+ "step": 18257
+ },
+ {
+ "epoch": 49.74931880108992,
+ "grad_norm": 2.2784128189086914,
+ "learning_rate": 1.0566699040651687e-05,
+ "loss": 0.0412,
+ "step": 18258
+ },
+ {
+ "epoch": 49.752043596730246,
+ "grad_norm": 2.3517327308654785,
+ "learning_rate": 1.0565817962006033e-05,
+ "loss": 0.1631,
+ "step": 18259
+ },
+ {
+ "epoch": 49.75476839237057,
+ "grad_norm": 3.011260747909546,
+ "learning_rate": 1.0564936878953804e-05,
+ "loss": 0.1062,
+ "step": 18260
+ },
+ {
+ "epoch": 49.7574931880109,
+ "grad_norm": 3.184122085571289,
+ "learning_rate": 1.0564055791501867e-05,
+ "loss": 0.0845,
+ "step": 18261
+ },
+ {
+ "epoch": 49.76021798365122,
+ "grad_norm": 2.529841661453247,
+ "learning_rate": 1.056317469965708e-05,
+ "loss": 0.0987,
+ "step": 18262
+ },
+ {
+ "epoch": 49.762942779291556,
+ "grad_norm": 15.118707656860352,
+ "learning_rate": 1.0562293603426309e-05,
+ "loss": 0.0733,
+ "step": 18263
+ },
+ {
+ "epoch": 49.76566757493188,
+ "grad_norm": 3.2699484825134277,
+ "learning_rate": 1.0561412502816411e-05,
+ "loss": 0.1345,
+ "step": 18264
+ },
+ {
+ "epoch": 49.76839237057221,
+ "grad_norm": 2.6171483993530273,
+ "learning_rate": 1.0560531397834253e-05,
+ "loss": 0.141,
+ "step": 18265
+ },
+ {
+ "epoch": 49.771117166212534,
+ "grad_norm": 2.346344470977783,
+ "learning_rate": 1.055965028848669e-05,
+ "loss": 0.0558,
+ "step": 18266
+ },
+ {
+ "epoch": 49.77384196185286,
+ "grad_norm": 2.346492290496826,
+ "learning_rate": 1.0558769174780595e-05,
+ "loss": 0.0386,
+ "step": 18267
+ },
+ {
+ "epoch": 49.776566757493185,
+ "grad_norm": 3.0589983463287354,
+ "learning_rate": 1.055788805672282e-05,
+ "loss": 0.1115,
+ "step": 18268
+ },
+ {
+ "epoch": 49.77929155313352,
+ "grad_norm": 2.7905168533325195,
+ "learning_rate": 1.0557006934320232e-05,
+ "loss": 0.0478,
+ "step": 18269
+ },
+ {
+ "epoch": 49.782016348773844,
+ "grad_norm": 2.9464163780212402,
+ "learning_rate": 1.0556125807579691e-05,
+ "loss": 0.0945,
+ "step": 18270
+ },
+ {
+ "epoch": 49.78474114441417,
+ "grad_norm": 2.067387342453003,
+ "learning_rate": 1.0555244676508058e-05,
+ "loss": 0.0319,
+ "step": 18271
+ },
+ {
+ "epoch": 49.787465940054496,
+ "grad_norm": 2.699509859085083,
+ "learning_rate": 1.05543635411122e-05,
+ "loss": 0.0443,
+ "step": 18272
+ },
+ {
+ "epoch": 49.79019073569482,
+ "grad_norm": 2.6682114601135254,
+ "learning_rate": 1.0553482401398979e-05,
+ "loss": 0.0915,
+ "step": 18273
+ },
+ {
+ "epoch": 49.79291553133515,
+ "grad_norm": 2.613781452178955,
+ "learning_rate": 1.055260125737525e-05,
+ "loss": 0.0401,
+ "step": 18274
+ },
+ {
+ "epoch": 49.79564032697548,
+ "grad_norm": 3.124507427215576,
+ "learning_rate": 1.0551720109047883e-05,
+ "loss": 0.0724,
+ "step": 18275
+ },
+ {
+ "epoch": 49.798365122615806,
+ "grad_norm": 2.8962631225585938,
+ "learning_rate": 1.0550838956423737e-05,
+ "loss": 0.0518,
+ "step": 18276
+ },
+ {
+ "epoch": 49.80108991825613,
+ "grad_norm": 2.248539447784424,
+ "learning_rate": 1.0549957799509674e-05,
+ "loss": 0.1857,
+ "step": 18277
+ },
+ {
+ "epoch": 49.80381471389646,
+ "grad_norm": 2.6016740798950195,
+ "learning_rate": 1.0549076638312558e-05,
+ "loss": 0.2204,
+ "step": 18278
+ },
+ {
+ "epoch": 49.80653950953678,
+ "grad_norm": 3.329979419708252,
+ "learning_rate": 1.0548195472839251e-05,
+ "loss": 0.1425,
+ "step": 18279
+ },
+ {
+ "epoch": 49.80926430517711,
+ "grad_norm": 4.969999313354492,
+ "learning_rate": 1.0547314303096617e-05,
+ "loss": 0.0495,
+ "step": 18280
+ },
+ {
+ "epoch": 49.81198910081744,
+ "grad_norm": 3.020908832550049,
+ "learning_rate": 1.0546433129091514e-05,
+ "loss": 0.0711,
+ "step": 18281
+ },
+ {
+ "epoch": 49.81471389645777,
+ "grad_norm": 2.687854528427124,
+ "learning_rate": 1.0545551950830808e-05,
+ "loss": 0.1018,
+ "step": 18282
+ },
+ {
+ "epoch": 49.817438692098094,
+ "grad_norm": 2.6868820190429688,
+ "learning_rate": 1.054467076832136e-05,
+ "loss": 0.0516,
+ "step": 18283
+ },
+ {
+ "epoch": 49.82016348773842,
+ "grad_norm": 2.490438222885132,
+ "learning_rate": 1.0543789581570037e-05,
+ "loss": 0.0442,
+ "step": 18284
+ },
+ {
+ "epoch": 49.822888283378745,
+ "grad_norm": 2.1212518215179443,
+ "learning_rate": 1.0542908390583694e-05,
+ "loss": 0.0503,
+ "step": 18285
+ },
+ {
+ "epoch": 49.82561307901907,
+ "grad_norm": 3.3616809844970703,
+ "learning_rate": 1.0542027195369202e-05,
+ "loss": 0.1631,
+ "step": 18286
+ },
+ {
+ "epoch": 49.828337874659404,
+ "grad_norm": 2.0669944286346436,
+ "learning_rate": 1.0541145995933416e-05,
+ "loss": 0.0681,
+ "step": 18287
+ },
+ {
+ "epoch": 49.83106267029973,
+ "grad_norm": 2.7934625148773193,
+ "learning_rate": 1.0540264792283205e-05,
+ "loss": 0.0313,
+ "step": 18288
+ },
+ {
+ "epoch": 49.833787465940055,
+ "grad_norm": 2.343611240386963,
+ "learning_rate": 1.0539383584425428e-05,
+ "loss": 0.049,
+ "step": 18289
+ },
+ {
+ "epoch": 49.83651226158038,
+ "grad_norm": 2.296492338180542,
+ "learning_rate": 1.053850237236695e-05,
+ "loss": 0.0558,
+ "step": 18290
+ },
+ {
+ "epoch": 49.83923705722071,
+ "grad_norm": 3.257873296737671,
+ "learning_rate": 1.0537621156114633e-05,
+ "loss": 0.1065,
+ "step": 18291
+ },
+ {
+ "epoch": 49.84196185286103,
+ "grad_norm": 3.2095859050750732,
+ "learning_rate": 1.053673993567534e-05,
+ "loss": 0.0759,
+ "step": 18292
+ },
+ {
+ "epoch": 49.844686648501366,
+ "grad_norm": 3.7985434532165527,
+ "learning_rate": 1.0535858711055931e-05,
+ "loss": 0.1088,
+ "step": 18293
+ },
+ {
+ "epoch": 49.84741144414169,
+ "grad_norm": 3.6755411624908447,
+ "learning_rate": 1.0534977482263274e-05,
+ "loss": 0.0563,
+ "step": 18294
+ },
+ {
+ "epoch": 49.85013623978202,
+ "grad_norm": 2.4541587829589844,
+ "learning_rate": 1.0534096249304226e-05,
+ "loss": 0.083,
+ "step": 18295
+ },
+ {
+ "epoch": 49.85286103542234,
+ "grad_norm": 2.7235770225524902,
+ "learning_rate": 1.0533215012185659e-05,
+ "loss": 0.1057,
+ "step": 18296
+ },
+ {
+ "epoch": 49.85558583106267,
+ "grad_norm": 2.2592461109161377,
+ "learning_rate": 1.0532333770914427e-05,
+ "loss": 0.0367,
+ "step": 18297
+ },
+ {
+ "epoch": 49.858310626702995,
+ "grad_norm": 2.339188575744629,
+ "learning_rate": 1.0531452525497398e-05,
+ "loss": 0.0584,
+ "step": 18298
+ },
+ {
+ "epoch": 49.86103542234333,
+ "grad_norm": 2.4814627170562744,
+ "learning_rate": 1.0530571275941433e-05,
+ "loss": 0.107,
+ "step": 18299
+ },
+ {
+ "epoch": 49.86376021798365,
+ "grad_norm": 2.487588882446289,
+ "learning_rate": 1.0529690022253398e-05,
+ "loss": 0.1452,
+ "step": 18300
+ },
+ {
+ "epoch": 49.86648501362398,
+ "grad_norm": 2.308091402053833,
+ "learning_rate": 1.0528808764440152e-05,
+ "loss": 0.0813,
+ "step": 18301
+ },
+ {
+ "epoch": 49.869209809264305,
+ "grad_norm": 2.127250909805298,
+ "learning_rate": 1.052792750250856e-05,
+ "loss": 0.1715,
+ "step": 18302
+ },
+ {
+ "epoch": 49.87193460490463,
+ "grad_norm": 2.538487672805786,
+ "learning_rate": 1.0527046236465485e-05,
+ "loss": 0.149,
+ "step": 18303
+ },
+ {
+ "epoch": 49.87465940054496,
+ "grad_norm": 2.6063482761383057,
+ "learning_rate": 1.0526164966317796e-05,
+ "loss": 0.1398,
+ "step": 18304
+ },
+ {
+ "epoch": 49.87738419618529,
+ "grad_norm": 3.165769100189209,
+ "learning_rate": 1.0525283692072346e-05,
+ "loss": 0.0904,
+ "step": 18305
+ },
+ {
+ "epoch": 49.880108991825615,
+ "grad_norm": 10.427550315856934,
+ "learning_rate": 1.0524402413736005e-05,
+ "loss": 0.1664,
+ "step": 18306
+ },
+ {
+ "epoch": 49.88283378746594,
+ "grad_norm": 2.7045373916625977,
+ "learning_rate": 1.0523521131315636e-05,
+ "loss": 0.0919,
+ "step": 18307
+ },
+ {
+ "epoch": 49.88555858310627,
+ "grad_norm": 2.0682504177093506,
+ "learning_rate": 1.0522639844818102e-05,
+ "loss": 0.0758,
+ "step": 18308
+ },
+ {
+ "epoch": 49.88828337874659,
+ "grad_norm": 3.210679054260254,
+ "learning_rate": 1.0521758554250263e-05,
+ "loss": 0.0625,
+ "step": 18309
+ },
+ {
+ "epoch": 49.89100817438692,
+ "grad_norm": 1.9264754056930542,
+ "learning_rate": 1.0520877259618986e-05,
+ "loss": 0.0346,
+ "step": 18310
+ },
+ {
+ "epoch": 49.89373297002725,
+ "grad_norm": 2.4800987243652344,
+ "learning_rate": 1.0519995960931136e-05,
+ "loss": 0.0659,
+ "step": 18311
+ },
+ {
+ "epoch": 49.89645776566758,
+ "grad_norm": 2.3443424701690674,
+ "learning_rate": 1.051911465819357e-05,
+ "loss": 0.059,
+ "step": 18312
+ },
+ {
+ "epoch": 49.8991825613079,
+ "grad_norm": 2.627199172973633,
+ "learning_rate": 1.0518233351413155e-05,
+ "loss": 0.0388,
+ "step": 18313
+ },
+ {
+ "epoch": 49.90190735694823,
+ "grad_norm": 2.9303841590881348,
+ "learning_rate": 1.0517352040596762e-05,
+ "loss": 0.2302,
+ "step": 18314
+ },
+ {
+ "epoch": 49.904632152588555,
+ "grad_norm": 2.8889286518096924,
+ "learning_rate": 1.0516470725751242e-05,
+ "loss": 0.0617,
+ "step": 18315
+ },
+ {
+ "epoch": 49.90735694822888,
+ "grad_norm": 2.781851053237915,
+ "learning_rate": 1.0515589406883469e-05,
+ "loss": 0.0656,
+ "step": 18316
+ },
+ {
+ "epoch": 49.91008174386921,
+ "grad_norm": 2.7878544330596924,
+ "learning_rate": 1.05147080840003e-05,
+ "loss": 0.2179,
+ "step": 18317
+ },
+ {
+ "epoch": 49.91280653950954,
+ "grad_norm": 2.365365743637085,
+ "learning_rate": 1.0513826757108603e-05,
+ "loss": 0.0381,
+ "step": 18318
+ },
+ {
+ "epoch": 49.915531335149865,
+ "grad_norm": 1.9531092643737793,
+ "learning_rate": 1.0512945426215234e-05,
+ "loss": 0.037,
+ "step": 18319
+ },
+ {
+ "epoch": 49.91825613079019,
+ "grad_norm": 5.329381465911865,
+ "learning_rate": 1.0512064091327067e-05,
+ "loss": 0.0508,
+ "step": 18320
+ },
+ {
+ "epoch": 49.920980926430516,
+ "grad_norm": 2.40765380859375,
+ "learning_rate": 1.0511182752450959e-05,
+ "loss": 0.095,
+ "step": 18321
+ },
+ {
+ "epoch": 49.92370572207084,
+ "grad_norm": 4.351596832275391,
+ "learning_rate": 1.0510301409593779e-05,
+ "loss": 0.0622,
+ "step": 18322
+ },
+ {
+ "epoch": 49.926430517711175,
+ "grad_norm": 2.5622384548187256,
+ "learning_rate": 1.0509420062762388e-05,
+ "loss": 0.0822,
+ "step": 18323
+ },
+ {
+ "epoch": 49.9291553133515,
+ "grad_norm": 4.085280418395996,
+ "learning_rate": 1.0508538711963649e-05,
+ "loss": 0.1775,
+ "step": 18324
+ },
+ {
+ "epoch": 49.93188010899183,
+ "grad_norm": 2.4044787883758545,
+ "learning_rate": 1.0507657357204426e-05,
+ "loss": 0.0565,
+ "step": 18325
+ },
+ {
+ "epoch": 49.93460490463215,
+ "grad_norm": 2.4247682094573975,
+ "learning_rate": 1.0506775998491587e-05,
+ "loss": 0.1162,
+ "step": 18326
+ },
+ {
+ "epoch": 49.93732970027248,
+ "grad_norm": 3.9967451095581055,
+ "learning_rate": 1.050589463583199e-05,
+ "loss": 0.0723,
+ "step": 18327
+ },
+ {
+ "epoch": 49.940054495912804,
+ "grad_norm": 4.197859287261963,
+ "learning_rate": 1.0505013269232503e-05,
+ "loss": 0.2064,
+ "step": 18328
+ },
+ {
+ "epoch": 49.94277929155314,
+ "grad_norm": 2.3218400478363037,
+ "learning_rate": 1.0504131898699985e-05,
+ "loss": 0.0557,
+ "step": 18329
+ },
+ {
+ "epoch": 49.94550408719346,
+ "grad_norm": 3.565688133239746,
+ "learning_rate": 1.0503250524241308e-05,
+ "loss": 0.1325,
+ "step": 18330
+ },
+ {
+ "epoch": 49.94822888283379,
+ "grad_norm": 3.017334461212158,
+ "learning_rate": 1.0502369145863332e-05,
+ "loss": 0.0679,
+ "step": 18331
+ },
+ {
+ "epoch": 49.950953678474114,
+ "grad_norm": 2.400122880935669,
+ "learning_rate": 1.0501487763572921e-05,
+ "loss": 0.0668,
+ "step": 18332
+ },
+ {
+ "epoch": 49.95367847411444,
+ "grad_norm": 3.4552183151245117,
+ "learning_rate": 1.0500606377376938e-05,
+ "loss": 0.2092,
+ "step": 18333
+ },
+ {
+ "epoch": 49.956403269754766,
+ "grad_norm": 3.427994728088379,
+ "learning_rate": 1.0499724987282251e-05,
+ "loss": 0.0564,
+ "step": 18334
+ },
+ {
+ "epoch": 49.95912806539509,
+ "grad_norm": 2.838963508605957,
+ "learning_rate": 1.049884359329572e-05,
+ "loss": 0.1134,
+ "step": 18335
+ },
+ {
+ "epoch": 49.961852861035425,
+ "grad_norm": 3.203843593597412,
+ "learning_rate": 1.0497962195424214e-05,
+ "loss": 0.0655,
+ "step": 18336
+ },
+ {
+ "epoch": 49.96457765667575,
+ "grad_norm": 2.092912435531616,
+ "learning_rate": 1.049708079367459e-05,
+ "loss": 0.0385,
+ "step": 18337
+ },
+ {
+ "epoch": 49.967302452316076,
+ "grad_norm": 2.4752376079559326,
+ "learning_rate": 1.0496199388053718e-05,
+ "loss": 0.0716,
+ "step": 18338
+ },
+ {
+ "epoch": 49.9700272479564,
+ "grad_norm": 3.8172695636749268,
+ "learning_rate": 1.0495317978568462e-05,
+ "loss": 0.0921,
+ "step": 18339
+ },
+ {
+ "epoch": 49.97275204359673,
+ "grad_norm": 2.1981022357940674,
+ "learning_rate": 1.0494436565225688e-05,
+ "loss": 0.1062,
+ "step": 18340
+ },
+ {
+ "epoch": 49.97547683923706,
+ "grad_norm": 3.3180713653564453,
+ "learning_rate": 1.0493555148032253e-05,
+ "loss": 0.0734,
+ "step": 18341
+ },
+ {
+ "epoch": 49.97820163487739,
+ "grad_norm": 4.741184711456299,
+ "learning_rate": 1.0492673726995031e-05,
+ "loss": 0.1419,
+ "step": 18342
+ },
+ {
+ "epoch": 49.98092643051771,
+ "grad_norm": 2.4075767993927,
+ "learning_rate": 1.0491792302120879e-05,
+ "loss": 0.0559,
+ "step": 18343
+ },
+ {
+ "epoch": 49.98365122615804,
+ "grad_norm": 3.578977346420288,
+ "learning_rate": 1.0490910873416667e-05,
+ "loss": 0.1064,
+ "step": 18344
+ },
+ {
+ "epoch": 49.986376021798364,
+ "grad_norm": 3.2731740474700928,
+ "learning_rate": 1.0490029440889254e-05,
+ "loss": 0.0562,
+ "step": 18345
+ },
+ {
+ "epoch": 49.98910081743869,
+ "grad_norm": 2.735992908477783,
+ "learning_rate": 1.0489148004545506e-05,
+ "loss": 0.1216,
+ "step": 18346
+ },
+ {
+ "epoch": 49.991825613079016,
+ "grad_norm": 3.218329429626465,
+ "learning_rate": 1.0488266564392293e-05,
+ "loss": 0.0721,
+ "step": 18347
+ },
+ {
+ "epoch": 49.99455040871935,
+ "grad_norm": 2.4931321144104004,
+ "learning_rate": 1.0487385120436475e-05,
+ "loss": 0.1202,
+ "step": 18348
+ },
+ {
+ "epoch": 49.997275204359674,
+ "grad_norm": 2.6227409839630127,
+ "learning_rate": 1.0486503672684915e-05,
+ "loss": 0.0523,
+ "step": 18349
+ },
+ {
+ "epoch": 50.0,
+ "grad_norm": 2.497035264968872,
+ "learning_rate": 1.0485622221144485e-05,
+ "loss": 0.0717,
+ "step": 18350
+ },
+ {
+ "epoch": 50.002724795640326,
+ "grad_norm": 2.7382426261901855,
+ "learning_rate": 1.048474076582204e-05,
+ "loss": 0.0533,
+ "step": 18351
+ },
+ {
+ "epoch": 50.00544959128065,
+ "grad_norm": 2.055645704269409,
+ "learning_rate": 1.0483859306724451e-05,
+ "loss": 0.1337,
+ "step": 18352
+ },
+ {
+ "epoch": 50.00817438692098,
+ "grad_norm": 2.851384401321411,
+ "learning_rate": 1.0482977843858581e-05,
+ "loss": 0.2001,
+ "step": 18353
+ },
+ {
+ "epoch": 50.01089918256131,
+ "grad_norm": 2.7616798877716064,
+ "learning_rate": 1.0482096377231297e-05,
+ "loss": 0.0493,
+ "step": 18354
+ },
+ {
+ "epoch": 50.013623978201636,
+ "grad_norm": 11.681770324707031,
+ "learning_rate": 1.0481214906849459e-05,
+ "loss": 0.0581,
+ "step": 18355
+ },
+ {
+ "epoch": 50.01634877384196,
+ "grad_norm": 2.102198600769043,
+ "learning_rate": 1.0480333432719936e-05,
+ "loss": 0.0728,
+ "step": 18356
+ },
+ {
+ "epoch": 50.01907356948229,
+ "grad_norm": 2.3789117336273193,
+ "learning_rate": 1.0479451954849594e-05,
+ "loss": 0.1553,
+ "step": 18357
+ },
+ {
+ "epoch": 50.02179836512261,
+ "grad_norm": 1.7027740478515625,
+ "learning_rate": 1.0478570473245293e-05,
+ "loss": 0.0375,
+ "step": 18358
+ },
+ {
+ "epoch": 50.02452316076294,
+ "grad_norm": 1.8035557270050049,
+ "learning_rate": 1.0477688987913898e-05,
+ "loss": 0.0333,
+ "step": 18359
+ },
+ {
+ "epoch": 50.02724795640327,
+ "grad_norm": 2.8411169052124023,
+ "learning_rate": 1.0476807498862282e-05,
+ "loss": 0.0443,
+ "step": 18360
+ },
+ {
+ "epoch": 50.0299727520436,
+ "grad_norm": 2.550978660583496,
+ "learning_rate": 1.04759260060973e-05,
+ "loss": 0.1251,
+ "step": 18361
+ },
+ {
+ "epoch": 50.032697547683924,
+ "grad_norm": 2.3933842182159424,
+ "learning_rate": 1.0475044509625827e-05,
+ "loss": 0.0491,
+ "step": 18362
+ },
+ {
+ "epoch": 50.03542234332425,
+ "grad_norm": 2.4329655170440674,
+ "learning_rate": 1.0474163009454718e-05,
+ "loss": 0.0418,
+ "step": 18363
+ },
+ {
+ "epoch": 50.038147138964575,
+ "grad_norm": 2.248729705810547,
+ "learning_rate": 1.0473281505590844e-05,
+ "loss": 0.0496,
+ "step": 18364
+ },
+ {
+ "epoch": 50.0408719346049,
+ "grad_norm": 3.35878324508667,
+ "learning_rate": 1.047239999804107e-05,
+ "loss": 0.0783,
+ "step": 18365
+ },
+ {
+ "epoch": 50.043596730245234,
+ "grad_norm": 5.551647663116455,
+ "learning_rate": 1.0471518486812258e-05,
+ "loss": 0.1904,
+ "step": 18366
+ },
+ {
+ "epoch": 50.04632152588556,
+ "grad_norm": 2.7577645778656006,
+ "learning_rate": 1.0470636971911277e-05,
+ "loss": 0.139,
+ "step": 18367
+ },
+ {
+ "epoch": 50.049046321525886,
+ "grad_norm": 3.198946714401245,
+ "learning_rate": 1.0469755453344989e-05,
+ "loss": 0.1103,
+ "step": 18368
+ },
+ {
+ "epoch": 50.05177111716621,
+ "grad_norm": 2.0614304542541504,
+ "learning_rate": 1.0468873931120263e-05,
+ "loss": 0.1084,
+ "step": 18369
+ },
+ {
+ "epoch": 50.05449591280654,
+ "grad_norm": 3.423579692840576,
+ "learning_rate": 1.0467992405243959e-05,
+ "loss": 0.1539,
+ "step": 18370
+ },
+ {
+ "epoch": 50.05722070844686,
+ "grad_norm": 2.1848981380462646,
+ "learning_rate": 1.0467110875722945e-05,
+ "loss": 0.0546,
+ "step": 18371
+ },
+ {
+ "epoch": 50.059945504087196,
+ "grad_norm": 2.0654144287109375,
+ "learning_rate": 1.0466229342564087e-05,
+ "loss": 0.0464,
+ "step": 18372
+ },
+ {
+ "epoch": 50.06267029972752,
+ "grad_norm": 3.6344377994537354,
+ "learning_rate": 1.0465347805774253e-05,
+ "loss": 0.0949,
+ "step": 18373
+ },
+ {
+ "epoch": 50.06539509536785,
+ "grad_norm": 2.7336039543151855,
+ "learning_rate": 1.0464466265360303e-05,
+ "loss": 0.0996,
+ "step": 18374
+ },
+ {
+ "epoch": 50.06811989100817,
+ "grad_norm": 1.922207236289978,
+ "learning_rate": 1.0463584721329105e-05,
+ "loss": 0.1721,
+ "step": 18375
+ },
+ {
+ "epoch": 50.0708446866485,
+ "grad_norm": 2.345367670059204,
+ "learning_rate": 1.0462703173687523e-05,
+ "loss": 0.1472,
+ "step": 18376
+ },
+ {
+ "epoch": 50.073569482288825,
+ "grad_norm": 2.992401361465454,
+ "learning_rate": 1.0461821622442425e-05,
+ "loss": 0.1516,
+ "step": 18377
+ },
+ {
+ "epoch": 50.07629427792916,
+ "grad_norm": 2.587277412414551,
+ "learning_rate": 1.0460940067600676e-05,
+ "loss": 0.0443,
+ "step": 18378
+ },
+ {
+ "epoch": 50.079019073569484,
+ "grad_norm": 3.0097262859344482,
+ "learning_rate": 1.046005850916914e-05,
+ "loss": 0.0463,
+ "step": 18379
+ },
+ {
+ "epoch": 50.08174386920981,
+ "grad_norm": 1.9484574794769287,
+ "learning_rate": 1.0459176947154682e-05,
+ "loss": 0.0433,
+ "step": 18380
+ },
+ {
+ "epoch": 50.084468664850135,
+ "grad_norm": 2.5740487575531006,
+ "learning_rate": 1.0458295381564168e-05,
+ "loss": 0.1052,
+ "step": 18381
+ },
+ {
+ "epoch": 50.08719346049046,
+ "grad_norm": 2.2405457496643066,
+ "learning_rate": 1.0457413812404461e-05,
+ "loss": 0.1687,
+ "step": 18382
+ },
+ {
+ "epoch": 50.08991825613079,
+ "grad_norm": 2.4644837379455566,
+ "learning_rate": 1.0456532239682438e-05,
+ "loss": 0.0648,
+ "step": 18383
+ },
+ {
+ "epoch": 50.09264305177112,
+ "grad_norm": 2.123737096786499,
+ "learning_rate": 1.0455650663404953e-05,
+ "loss": 0.0402,
+ "step": 18384
+ },
+ {
+ "epoch": 50.095367847411445,
+ "grad_norm": 2.560478448867798,
+ "learning_rate": 1.0454769083578876e-05,
+ "loss": 0.1169,
+ "step": 18385
+ },
+ {
+ "epoch": 50.09809264305177,
+ "grad_norm": 2.5579330921173096,
+ "learning_rate": 1.0453887500211068e-05,
+ "loss": 0.0521,
+ "step": 18386
+ },
+ {
+ "epoch": 50.1008174386921,
+ "grad_norm": 2.451930046081543,
+ "learning_rate": 1.0453005913308405e-05,
+ "loss": 0.0416,
+ "step": 18387
+ },
+ {
+ "epoch": 50.10354223433242,
+ "grad_norm": 1.9753013849258423,
+ "learning_rate": 1.0452124322877741e-05,
+ "loss": 0.0381,
+ "step": 18388
+ },
+ {
+ "epoch": 50.10626702997275,
+ "grad_norm": 3.0260844230651855,
+ "learning_rate": 1.045124272892595e-05,
+ "loss": 0.0914,
+ "step": 18389
+ },
+ {
+ "epoch": 50.10899182561308,
+ "grad_norm": 2.535451650619507,
+ "learning_rate": 1.0450361131459894e-05,
+ "loss": 0.1717,
+ "step": 18390
+ },
+ {
+ "epoch": 50.11171662125341,
+ "grad_norm": 2.7841782569885254,
+ "learning_rate": 1.0449479530486441e-05,
+ "loss": 0.0523,
+ "step": 18391
+ },
+ {
+ "epoch": 50.11444141689373,
+ "grad_norm": 2.7811994552612305,
+ "learning_rate": 1.0448597926012456e-05,
+ "loss": 0.1449,
+ "step": 18392
+ },
+ {
+ "epoch": 50.11716621253406,
+ "grad_norm": 2.9627504348754883,
+ "learning_rate": 1.0447716318044807e-05,
+ "loss": 0.0685,
+ "step": 18393
+ },
+ {
+ "epoch": 50.119891008174385,
+ "grad_norm": 2.274914026260376,
+ "learning_rate": 1.0446834706590354e-05,
+ "loss": 0.0283,
+ "step": 18394
+ },
+ {
+ "epoch": 50.12261580381471,
+ "grad_norm": 2.739741325378418,
+ "learning_rate": 1.044595309165597e-05,
+ "loss": 0.0619,
+ "step": 18395
+ },
+ {
+ "epoch": 50.12534059945504,
+ "grad_norm": 2.592038154602051,
+ "learning_rate": 1.0445071473248514e-05,
+ "loss": 0.0632,
+ "step": 18396
+ },
+ {
+ "epoch": 50.12806539509537,
+ "grad_norm": 2.699920177459717,
+ "learning_rate": 1.0444189851374857e-05,
+ "loss": 0.1231,
+ "step": 18397
+ },
+ {
+ "epoch": 50.130790190735695,
+ "grad_norm": 2.5383782386779785,
+ "learning_rate": 1.0443308226041862e-05,
+ "loss": 0.0732,
+ "step": 18398
+ },
+ {
+ "epoch": 50.13351498637602,
+ "grad_norm": 2.4484448432922363,
+ "learning_rate": 1.04424265972564e-05,
+ "loss": 0.0477,
+ "step": 18399
+ },
+ {
+ "epoch": 50.13623978201635,
+ "grad_norm": 2.1556901931762695,
+ "learning_rate": 1.0441544965025332e-05,
+ "loss": 0.0513,
+ "step": 18400
+ },
+ {
+ "epoch": 50.13896457765667,
+ "grad_norm": 2.2474966049194336,
+ "learning_rate": 1.044066332935553e-05,
+ "loss": 0.0367,
+ "step": 18401
+ },
+ {
+ "epoch": 50.141689373297005,
+ "grad_norm": 1.9213807582855225,
+ "learning_rate": 1.0439781690253851e-05,
+ "loss": 0.0447,
+ "step": 18402
+ },
+ {
+ "epoch": 50.14441416893733,
+ "grad_norm": 2.100672483444214,
+ "learning_rate": 1.043890004772717e-05,
+ "loss": 0.035,
+ "step": 18403
+ },
+ {
+ "epoch": 50.14713896457766,
+ "grad_norm": 2.6824138164520264,
+ "learning_rate": 1.0438018401782347e-05,
+ "loss": 0.1383,
+ "step": 18404
+ },
+ {
+ "epoch": 50.14986376021798,
+ "grad_norm": 2.8125243186950684,
+ "learning_rate": 1.0437136752426254e-05,
+ "loss": 0.1723,
+ "step": 18405
+ },
+ {
+ "epoch": 50.15258855585831,
+ "grad_norm": 2.896599531173706,
+ "learning_rate": 1.0436255099665746e-05,
+ "loss": 0.098,
+ "step": 18406
+ },
+ {
+ "epoch": 50.155313351498634,
+ "grad_norm": 2.9471707344055176,
+ "learning_rate": 1.0435373443507706e-05,
+ "loss": 0.0882,
+ "step": 18407
+ },
+ {
+ "epoch": 50.15803814713897,
+ "grad_norm": 2.7819342613220215,
+ "learning_rate": 1.0434491783958986e-05,
+ "loss": 0.1315,
+ "step": 18408
+ },
+ {
+ "epoch": 50.16076294277929,
+ "grad_norm": 3.500412940979004,
+ "learning_rate": 1.0433610121026463e-05,
+ "loss": 0.0479,
+ "step": 18409
+ },
+ {
+ "epoch": 50.16348773841962,
+ "grad_norm": 3.049957036972046,
+ "learning_rate": 1.0432728454716996e-05,
+ "loss": 0.0647,
+ "step": 18410
+ },
+ {
+ "epoch": 50.166212534059945,
+ "grad_norm": 2.6368160247802734,
+ "learning_rate": 1.0431846785037454e-05,
+ "loss": 0.1139,
+ "step": 18411
+ },
+ {
+ "epoch": 50.16893732970027,
+ "grad_norm": 2.345170497894287,
+ "learning_rate": 1.0430965111994701e-05,
+ "loss": 0.0437,
+ "step": 18412
+ },
+ {
+ "epoch": 50.171662125340596,
+ "grad_norm": 2.745213508605957,
+ "learning_rate": 1.0430083435595611e-05,
+ "loss": 0.0953,
+ "step": 18413
+ },
+ {
+ "epoch": 50.17438692098093,
+ "grad_norm": 2.4639525413513184,
+ "learning_rate": 1.0429201755847041e-05,
+ "loss": 0.05,
+ "step": 18414
+ },
+ {
+ "epoch": 50.177111716621255,
+ "grad_norm": 2.3686094284057617,
+ "learning_rate": 1.042832007275586e-05,
+ "loss": 0.1473,
+ "step": 18415
+ },
+ {
+ "epoch": 50.17983651226158,
+ "grad_norm": 2.0209176540374756,
+ "learning_rate": 1.0427438386328936e-05,
+ "loss": 0.0725,
+ "step": 18416
+ },
+ {
+ "epoch": 50.182561307901906,
+ "grad_norm": 2.3460288047790527,
+ "learning_rate": 1.042655669657314e-05,
+ "loss": 0.0649,
+ "step": 18417
+ },
+ {
+ "epoch": 50.18528610354223,
+ "grad_norm": 2.1154983043670654,
+ "learning_rate": 1.0425675003495333e-05,
+ "loss": 0.0326,
+ "step": 18418
+ },
+ {
+ "epoch": 50.18801089918256,
+ "grad_norm": 2.990619421005249,
+ "learning_rate": 1.0424793307102381e-05,
+ "loss": 0.056,
+ "step": 18419
+ },
+ {
+ "epoch": 50.19073569482289,
+ "grad_norm": 2.965803384780884,
+ "learning_rate": 1.0423911607401152e-05,
+ "loss": 0.1185,
+ "step": 18420
+ },
+ {
+ "epoch": 50.19346049046322,
+ "grad_norm": 2.026495933532715,
+ "learning_rate": 1.0423029904398515e-05,
+ "loss": 0.0653,
+ "step": 18421
+ },
+ {
+ "epoch": 50.19618528610354,
+ "grad_norm": 3.0666627883911133,
+ "learning_rate": 1.0422148198101334e-05,
+ "loss": 0.0884,
+ "step": 18422
+ },
+ {
+ "epoch": 50.19891008174387,
+ "grad_norm": 2.437859535217285,
+ "learning_rate": 1.0421266488516478e-05,
+ "loss": 0.189,
+ "step": 18423
+ },
+ {
+ "epoch": 50.201634877384194,
+ "grad_norm": 3.6139631271362305,
+ "learning_rate": 1.0420384775650806e-05,
+ "loss": 0.0746,
+ "step": 18424
+ },
+ {
+ "epoch": 50.20435967302452,
+ "grad_norm": 1.747441053390503,
+ "learning_rate": 1.0419503059511197e-05,
+ "loss": 0.034,
+ "step": 18425
+ },
+ {
+ "epoch": 50.20708446866485,
+ "grad_norm": 3.101374387741089,
+ "learning_rate": 1.0418621340104509e-05,
+ "loss": 0.0332,
+ "step": 18426
+ },
+ {
+ "epoch": 50.20980926430518,
+ "grad_norm": 2.166839838027954,
+ "learning_rate": 1.0417739617437615e-05,
+ "loss": 0.0425,
+ "step": 18427
+ },
+ {
+ "epoch": 50.212534059945504,
+ "grad_norm": 2.478234052658081,
+ "learning_rate": 1.0416857891517374e-05,
+ "loss": 0.0436,
+ "step": 18428
+ },
+ {
+ "epoch": 50.21525885558583,
+ "grad_norm": 2.878437042236328,
+ "learning_rate": 1.041597616235066e-05,
+ "loss": 0.0466,
+ "step": 18429
+ },
+ {
+ "epoch": 50.217983651226156,
+ "grad_norm": 2.7601749897003174,
+ "learning_rate": 1.0415094429944334e-05,
+ "loss": 0.1946,
+ "step": 18430
+ },
+ {
+ "epoch": 50.22070844686648,
+ "grad_norm": 2.289506673812866,
+ "learning_rate": 1.041421269430527e-05,
+ "loss": 0.0534,
+ "step": 18431
+ },
+ {
+ "epoch": 50.223433242506815,
+ "grad_norm": 2.0180861949920654,
+ "learning_rate": 1.0413330955440328e-05,
+ "loss": 0.1135,
+ "step": 18432
+ },
+ {
+ "epoch": 50.22615803814714,
+ "grad_norm": 2.2103798389434814,
+ "learning_rate": 1.0412449213356377e-05,
+ "loss": 0.0416,
+ "step": 18433
+ },
+ {
+ "epoch": 50.228882833787466,
+ "grad_norm": 2.308856725692749,
+ "learning_rate": 1.0411567468060288e-05,
+ "loss": 0.1565,
+ "step": 18434
+ },
+ {
+ "epoch": 50.23160762942779,
+ "grad_norm": 2.42596435546875,
+ "learning_rate": 1.0410685719558921e-05,
+ "loss": 0.0357,
+ "step": 18435
+ },
+ {
+ "epoch": 50.23433242506812,
+ "grad_norm": 2.700526714324951,
+ "learning_rate": 1.040980396785915e-05,
+ "loss": 0.052,
+ "step": 18436
+ },
+ {
+ "epoch": 50.237057220708444,
+ "grad_norm": 2.233126163482666,
+ "learning_rate": 1.0408922212967839e-05,
+ "loss": 0.0561,
+ "step": 18437
+ },
+ {
+ "epoch": 50.23978201634878,
+ "grad_norm": 2.1621365547180176,
+ "learning_rate": 1.0408040454891852e-05,
+ "loss": 0.0328,
+ "step": 18438
+ },
+ {
+ "epoch": 50.2425068119891,
+ "grad_norm": 2.4609968662261963,
+ "learning_rate": 1.0407158693638065e-05,
+ "loss": 0.1569,
+ "step": 18439
+ },
+ {
+ "epoch": 50.24523160762943,
+ "grad_norm": 3.4706506729125977,
+ "learning_rate": 1.0406276929213332e-05,
+ "loss": 0.103,
+ "step": 18440
+ },
+ {
+ "epoch": 50.247956403269754,
+ "grad_norm": 2.143491268157959,
+ "learning_rate": 1.040539516162453e-05,
+ "loss": 0.0917,
+ "step": 18441
+ },
+ {
+ "epoch": 50.25068119891008,
+ "grad_norm": 3.1182641983032227,
+ "learning_rate": 1.0404513390878526e-05,
+ "loss": 0.0726,
+ "step": 18442
+ },
+ {
+ "epoch": 50.253405994550405,
+ "grad_norm": 2.6423211097717285,
+ "learning_rate": 1.0403631616982181e-05,
+ "loss": 0.0866,
+ "step": 18443
+ },
+ {
+ "epoch": 50.25613079019074,
+ "grad_norm": 2.3535332679748535,
+ "learning_rate": 1.040274983994237e-05,
+ "loss": 0.0787,
+ "step": 18444
+ },
+ {
+ "epoch": 50.258855585831064,
+ "grad_norm": 2.0707881450653076,
+ "learning_rate": 1.0401868059765952e-05,
+ "loss": 0.0382,
+ "step": 18445
+ },
+ {
+ "epoch": 50.26158038147139,
+ "grad_norm": 2.7861201763153076,
+ "learning_rate": 1.04009862764598e-05,
+ "loss": 0.1025,
+ "step": 18446
+ },
+ {
+ "epoch": 50.264305177111716,
+ "grad_norm": 2.5795931816101074,
+ "learning_rate": 1.0400104490030781e-05,
+ "loss": 0.1687,
+ "step": 18447
+ },
+ {
+ "epoch": 50.26702997275204,
+ "grad_norm": 2.807596206665039,
+ "learning_rate": 1.039922270048576e-05,
+ "loss": 0.066,
+ "step": 18448
+ },
+ {
+ "epoch": 50.26975476839237,
+ "grad_norm": 2.071350574493408,
+ "learning_rate": 1.0398340907831603e-05,
+ "loss": 0.0545,
+ "step": 18449
+ },
+ {
+ "epoch": 50.2724795640327,
+ "grad_norm": 2.0975537300109863,
+ "learning_rate": 1.0397459112075186e-05,
+ "loss": 0.0684,
+ "step": 18450
+ },
+ {
+ "epoch": 50.275204359673026,
+ "grad_norm": 2.8144443035125732,
+ "learning_rate": 1.0396577313223367e-05,
+ "loss": 0.0701,
+ "step": 18451
+ },
+ {
+ "epoch": 50.27792915531335,
+ "grad_norm": 2.822052478790283,
+ "learning_rate": 1.0395695511283015e-05,
+ "loss": 0.0776,
+ "step": 18452
+ },
+ {
+ "epoch": 50.28065395095368,
+ "grad_norm": 2.965564250946045,
+ "learning_rate": 1.0394813706261e-05,
+ "loss": 0.0388,
+ "step": 18453
+ },
+ {
+ "epoch": 50.283378746594,
+ "grad_norm": 7.128743648529053,
+ "learning_rate": 1.0393931898164191e-05,
+ "loss": 0.0932,
+ "step": 18454
+ },
+ {
+ "epoch": 50.28610354223433,
+ "grad_norm": 2.1813242435455322,
+ "learning_rate": 1.039305008699945e-05,
+ "loss": 0.0338,
+ "step": 18455
+ },
+ {
+ "epoch": 50.28882833787466,
+ "grad_norm": 2.243227243423462,
+ "learning_rate": 1.0392168272773651e-05,
+ "loss": 0.097,
+ "step": 18456
+ },
+ {
+ "epoch": 50.29155313351499,
+ "grad_norm": 2.31799578666687,
+ "learning_rate": 1.0391286455493655e-05,
+ "loss": 0.184,
+ "step": 18457
+ },
+ {
+ "epoch": 50.294277929155314,
+ "grad_norm": 2.1909704208374023,
+ "learning_rate": 1.0390404635166331e-05,
+ "loss": 0.0477,
+ "step": 18458
+ },
+ {
+ "epoch": 50.29700272479564,
+ "grad_norm": 2.9072105884552,
+ "learning_rate": 1.0389522811798551e-05,
+ "loss": 0.0734,
+ "step": 18459
+ },
+ {
+ "epoch": 50.299727520435965,
+ "grad_norm": 1.8163877725601196,
+ "learning_rate": 1.038864098539718e-05,
+ "loss": 0.0455,
+ "step": 18460
+ },
+ {
+ "epoch": 50.30245231607629,
+ "grad_norm": 3.1926028728485107,
+ "learning_rate": 1.0387759155969085e-05,
+ "loss": 0.1768,
+ "step": 18461
+ },
+ {
+ "epoch": 50.305177111716624,
+ "grad_norm": 3.3836171627044678,
+ "learning_rate": 1.0386877323521137e-05,
+ "loss": 0.0884,
+ "step": 18462
+ },
+ {
+ "epoch": 50.30790190735695,
+ "grad_norm": 2.5015742778778076,
+ "learning_rate": 1.0385995488060196e-05,
+ "loss": 0.0512,
+ "step": 18463
+ },
+ {
+ "epoch": 50.310626702997276,
+ "grad_norm": 2.741772174835205,
+ "learning_rate": 1.0385113649593137e-05,
+ "loss": 0.0901,
+ "step": 18464
+ },
+ {
+ "epoch": 50.3133514986376,
+ "grad_norm": 4.481037139892578,
+ "learning_rate": 1.0384231808126826e-05,
+ "loss": 0.2084,
+ "step": 18465
+ },
+ {
+ "epoch": 50.31607629427793,
+ "grad_norm": 2.4062962532043457,
+ "learning_rate": 1.038334996366813e-05,
+ "loss": 0.055,
+ "step": 18466
+ },
+ {
+ "epoch": 50.31880108991825,
+ "grad_norm": 2.2869701385498047,
+ "learning_rate": 1.0382468116223912e-05,
+ "loss": 0.0975,
+ "step": 18467
+ },
+ {
+ "epoch": 50.321525885558586,
+ "grad_norm": 2.195385456085205,
+ "learning_rate": 1.0381586265801049e-05,
+ "loss": 0.1831,
+ "step": 18468
+ },
+ {
+ "epoch": 50.32425068119891,
+ "grad_norm": 2.6136679649353027,
+ "learning_rate": 1.0380704412406405e-05,
+ "loss": 0.0542,
+ "step": 18469
+ },
+ {
+ "epoch": 50.32697547683924,
+ "grad_norm": 3.4063122272491455,
+ "learning_rate": 1.0379822556046845e-05,
+ "loss": 0.1736,
+ "step": 18470
+ },
+ {
+ "epoch": 50.32970027247956,
+ "grad_norm": 3.3712000846862793,
+ "learning_rate": 1.037894069672924e-05,
+ "loss": 0.1529,
+ "step": 18471
+ },
+ {
+ "epoch": 50.33242506811989,
+ "grad_norm": 2.3530004024505615,
+ "learning_rate": 1.037805883446046e-05,
+ "loss": 0.0406,
+ "step": 18472
+ },
+ {
+ "epoch": 50.335149863760215,
+ "grad_norm": 2.2870171070098877,
+ "learning_rate": 1.0377176969247366e-05,
+ "loss": 0.0375,
+ "step": 18473
+ },
+ {
+ "epoch": 50.33787465940055,
+ "grad_norm": 1.9355655908584595,
+ "learning_rate": 1.0376295101096834e-05,
+ "loss": 0.0662,
+ "step": 18474
+ },
+ {
+ "epoch": 50.34059945504087,
+ "grad_norm": 1.8795355558395386,
+ "learning_rate": 1.037541323001572e-05,
+ "loss": 0.0396,
+ "step": 18475
+ },
+ {
+ "epoch": 50.3433242506812,
+ "grad_norm": 2.5143375396728516,
+ "learning_rate": 1.0374531356010909e-05,
+ "loss": 0.0946,
+ "step": 18476
+ },
+ {
+ "epoch": 50.346049046321525,
+ "grad_norm": 2.3057658672332764,
+ "learning_rate": 1.0373649479089254e-05,
+ "loss": 0.052,
+ "step": 18477
+ },
+ {
+ "epoch": 50.34877384196185,
+ "grad_norm": 2.909342050552368,
+ "learning_rate": 1.0372767599257631e-05,
+ "loss": 0.067,
+ "step": 18478
+ },
+ {
+ "epoch": 50.35149863760218,
+ "grad_norm": 2.1016788482666016,
+ "learning_rate": 1.0371885716522904e-05,
+ "loss": 0.0645,
+ "step": 18479
+ },
+ {
+ "epoch": 50.35422343324251,
+ "grad_norm": 2.659306526184082,
+ "learning_rate": 1.0371003830891948e-05,
+ "loss": 0.0676,
+ "step": 18480
+ },
+ {
+ "epoch": 50.356948228882835,
+ "grad_norm": 2.766587734222412,
+ "learning_rate": 1.0370121942371623e-05,
+ "loss": 0.0792,
+ "step": 18481
+ },
+ {
+ "epoch": 50.35967302452316,
+ "grad_norm": 2.730769634246826,
+ "learning_rate": 1.0369240050968802e-05,
+ "loss": 0.0603,
+ "step": 18482
+ },
+ {
+ "epoch": 50.36239782016349,
+ "grad_norm": 2.0089962482452393,
+ "learning_rate": 1.0368358156690345e-05,
+ "loss": 0.0742,
+ "step": 18483
+ },
+ {
+ "epoch": 50.36512261580381,
+ "grad_norm": 2.097205877304077,
+ "learning_rate": 1.0367476259543133e-05,
+ "loss": 0.0393,
+ "step": 18484
+ },
+ {
+ "epoch": 50.36784741144414,
+ "grad_norm": 2.2193689346313477,
+ "learning_rate": 1.0366594359534024e-05,
+ "loss": 0.0528,
+ "step": 18485
+ },
+ {
+ "epoch": 50.37057220708447,
+ "grad_norm": 2.5391082763671875,
+ "learning_rate": 1.0365712456669891e-05,
+ "loss": 0.0568,
+ "step": 18486
+ },
+ {
+ "epoch": 50.3732970027248,
+ "grad_norm": 3.7621257305145264,
+ "learning_rate": 1.0364830550957603e-05,
+ "loss": 0.0638,
+ "step": 18487
+ },
+ {
+ "epoch": 50.37602179836512,
+ "grad_norm": 3.178337335586548,
+ "learning_rate": 1.0363948642404025e-05,
+ "loss": 0.0688,
+ "step": 18488
+ },
+ {
+ "epoch": 50.37874659400545,
+ "grad_norm": 4.452877521514893,
+ "learning_rate": 1.0363066731016026e-05,
+ "loss": 0.2419,
+ "step": 18489
+ },
+ {
+ "epoch": 50.381471389645775,
+ "grad_norm": 2.473254680633545,
+ "learning_rate": 1.0362184816800478e-05,
+ "loss": 0.0642,
+ "step": 18490
+ },
+ {
+ "epoch": 50.3841961852861,
+ "grad_norm": 2.6699624061584473,
+ "learning_rate": 1.0361302899764242e-05,
+ "loss": 0.1467,
+ "step": 18491
+ },
+ {
+ "epoch": 50.38692098092643,
+ "grad_norm": 2.076582670211792,
+ "learning_rate": 1.0360420979914192e-05,
+ "loss": 0.0958,
+ "step": 18492
+ },
+ {
+ "epoch": 50.38964577656676,
+ "grad_norm": 2.5097506046295166,
+ "learning_rate": 1.0359539057257196e-05,
+ "loss": 0.2019,
+ "step": 18493
+ },
+ {
+ "epoch": 50.392370572207085,
+ "grad_norm": 2.775413990020752,
+ "learning_rate": 1.0358657131800123e-05,
+ "loss": 0.0746,
+ "step": 18494
+ },
+ {
+ "epoch": 50.39509536784741,
+ "grad_norm": 2.563009262084961,
+ "learning_rate": 1.0357775203549836e-05,
+ "loss": 0.1736,
+ "step": 18495
+ },
+ {
+ "epoch": 50.39782016348774,
+ "grad_norm": 2.426431179046631,
+ "learning_rate": 1.0356893272513211e-05,
+ "loss": 0.0525,
+ "step": 18496
+ },
+ {
+ "epoch": 50.40054495912806,
+ "grad_norm": 2.647047758102417,
+ "learning_rate": 1.0356011338697109e-05,
+ "loss": 0.0508,
+ "step": 18497
+ },
+ {
+ "epoch": 50.403269754768395,
+ "grad_norm": 2.795377254486084,
+ "learning_rate": 1.0355129402108407e-05,
+ "loss": 0.0773,
+ "step": 18498
+ },
+ {
+ "epoch": 50.40599455040872,
+ "grad_norm": 2.821824550628662,
+ "learning_rate": 1.0354247462753964e-05,
+ "loss": 0.0482,
+ "step": 18499
+ },
+ {
+ "epoch": 50.40871934604905,
+ "grad_norm": 2.197463035583496,
+ "learning_rate": 1.0353365520640657e-05,
+ "loss": 0.0722,
+ "step": 18500
+ },
+ {
+ "epoch": 50.41144414168937,
+ "grad_norm": 2.189786434173584,
+ "learning_rate": 1.0352483575775345e-05,
+ "loss": 0.0377,
+ "step": 18501
+ },
+ {
+ "epoch": 50.4141689373297,
+ "grad_norm": 3.054358959197998,
+ "learning_rate": 1.0351601628164906e-05,
+ "loss": 0.0533,
+ "step": 18502
+ },
+ {
+ "epoch": 50.416893732970024,
+ "grad_norm": 2.0162103176116943,
+ "learning_rate": 1.0350719677816206e-05,
+ "loss": 0.0318,
+ "step": 18503
+ },
+ {
+ "epoch": 50.41961852861036,
+ "grad_norm": 3.032081365585327,
+ "learning_rate": 1.0349837724736111e-05,
+ "loss": 0.1016,
+ "step": 18504
+ },
+ {
+ "epoch": 50.42234332425068,
+ "grad_norm": 2.7034435272216797,
+ "learning_rate": 1.034895576893149e-05,
+ "loss": 0.0786,
+ "step": 18505
+ },
+ {
+ "epoch": 50.42506811989101,
+ "grad_norm": 3.370467185974121,
+ "learning_rate": 1.0348073810409216e-05,
+ "loss": 0.1231,
+ "step": 18506
+ },
+ {
+ "epoch": 50.427792915531334,
+ "grad_norm": 2.8061726093292236,
+ "learning_rate": 1.0347191849176151e-05,
+ "loss": 0.0533,
+ "step": 18507
+ },
+ {
+ "epoch": 50.43051771117166,
+ "grad_norm": 2.3041927814483643,
+ "learning_rate": 1.034630988523917e-05,
+ "loss": 0.0442,
+ "step": 18508
+ },
+ {
+ "epoch": 50.433242506811986,
+ "grad_norm": 2.281059503555298,
+ "learning_rate": 1.0345427918605136e-05,
+ "loss": 0.0314,
+ "step": 18509
+ },
+ {
+ "epoch": 50.43596730245232,
+ "grad_norm": 2.4599618911743164,
+ "learning_rate": 1.034454594928092e-05,
+ "loss": 0.035,
+ "step": 18510
+ },
+ {
+ "epoch": 50.438692098092645,
+ "grad_norm": 2.14058256149292,
+ "learning_rate": 1.0343663977273393e-05,
+ "loss": 0.0333,
+ "step": 18511
+ },
+ {
+ "epoch": 50.44141689373297,
+ "grad_norm": 2.660867214202881,
+ "learning_rate": 1.0342782002589421e-05,
+ "loss": 0.1409,
+ "step": 18512
+ },
+ {
+ "epoch": 50.444141689373296,
+ "grad_norm": 2.476963996887207,
+ "learning_rate": 1.0341900025235876e-05,
+ "loss": 0.09,
+ "step": 18513
+ },
+ {
+ "epoch": 50.44686648501362,
+ "grad_norm": 3.7153618335723877,
+ "learning_rate": 1.0341018045219624e-05,
+ "loss": 0.0594,
+ "step": 18514
+ },
+ {
+ "epoch": 50.44959128065395,
+ "grad_norm": 4.526004314422607,
+ "learning_rate": 1.0340136062547531e-05,
+ "loss": 0.1724,
+ "step": 18515
+ },
+ {
+ "epoch": 50.45231607629428,
+ "grad_norm": 3.5125892162323,
+ "learning_rate": 1.0339254077226474e-05,
+ "loss": 0.1791,
+ "step": 18516
+ },
+ {
+ "epoch": 50.45504087193461,
+ "grad_norm": 2.2232775688171387,
+ "learning_rate": 1.0338372089263316e-05,
+ "loss": 0.1032,
+ "step": 18517
+ },
+ {
+ "epoch": 50.45776566757493,
+ "grad_norm": 3.4223265647888184,
+ "learning_rate": 1.0337490098664923e-05,
+ "loss": 0.1222,
+ "step": 18518
+ },
+ {
+ "epoch": 50.46049046321526,
+ "grad_norm": 2.0142087936401367,
+ "learning_rate": 1.0336608105438172e-05,
+ "loss": 0.2012,
+ "step": 18519
+ },
+ {
+ "epoch": 50.463215258855584,
+ "grad_norm": 3.98681378364563,
+ "learning_rate": 1.0335726109589927e-05,
+ "loss": 0.0829,
+ "step": 18520
+ },
+ {
+ "epoch": 50.46594005449591,
+ "grad_norm": 2.8620338439941406,
+ "learning_rate": 1.0334844111127058e-05,
+ "loss": 0.0692,
+ "step": 18521
+ },
+ {
+ "epoch": 50.46866485013624,
+ "grad_norm": 2.9453318119049072,
+ "learning_rate": 1.0333962110056435e-05,
+ "loss": 0.0696,
+ "step": 18522
+ },
+ {
+ "epoch": 50.47138964577657,
+ "grad_norm": 2.8745362758636475,
+ "learning_rate": 1.0333080106384922e-05,
+ "loss": 0.068,
+ "step": 18523
+ },
+ {
+ "epoch": 50.474114441416894,
+ "grad_norm": 2.9902760982513428,
+ "learning_rate": 1.0332198100119398e-05,
+ "loss": 0.0607,
+ "step": 18524
+ },
+ {
+ "epoch": 50.47683923705722,
+ "grad_norm": 3.131500005722046,
+ "learning_rate": 1.033131609126672e-05,
+ "loss": 0.064,
+ "step": 18525
+ },
+ {
+ "epoch": 50.479564032697546,
+ "grad_norm": 2.7947354316711426,
+ "learning_rate": 1.0330434079833767e-05,
+ "loss": 0.1815,
+ "step": 18526
+ },
+ {
+ "epoch": 50.48228882833787,
+ "grad_norm": 2.322235107421875,
+ "learning_rate": 1.0329552065827402e-05,
+ "loss": 0.0411,
+ "step": 18527
+ },
+ {
+ "epoch": 50.485013623978205,
+ "grad_norm": 2.714871406555176,
+ "learning_rate": 1.0328670049254494e-05,
+ "loss": 0.0957,
+ "step": 18528
+ },
+ {
+ "epoch": 50.48773841961853,
+ "grad_norm": 3.3679752349853516,
+ "learning_rate": 1.032778803012192e-05,
+ "loss": 0.0542,
+ "step": 18529
+ },
+ {
+ "epoch": 50.490463215258856,
+ "grad_norm": 2.3518271446228027,
+ "learning_rate": 1.0326906008436537e-05,
+ "loss": 0.053,
+ "step": 18530
+ },
+ {
+ "epoch": 50.49318801089918,
+ "grad_norm": 1.939316987991333,
+ "learning_rate": 1.0326023984205224e-05,
+ "loss": 0.1913,
+ "step": 18531
+ },
+ {
+ "epoch": 50.49591280653951,
+ "grad_norm": 2.5885722637176514,
+ "learning_rate": 1.0325141957434848e-05,
+ "loss": 0.0432,
+ "step": 18532
+ },
+ {
+ "epoch": 50.49863760217983,
+ "grad_norm": 2.777711868286133,
+ "learning_rate": 1.0324259928132273e-05,
+ "loss": 0.1896,
+ "step": 18533
+ },
+ {
+ "epoch": 50.50136239782017,
+ "grad_norm": 3.0301260948181152,
+ "learning_rate": 1.0323377896304377e-05,
+ "loss": 0.0838,
+ "step": 18534
+ },
+ {
+ "epoch": 50.50408719346049,
+ "grad_norm": 3.14876651763916,
+ "learning_rate": 1.032249586195802e-05,
+ "loss": 0.1029,
+ "step": 18535
+ },
+ {
+ "epoch": 50.50681198910082,
+ "grad_norm": 2.291651487350464,
+ "learning_rate": 1.0321613825100076e-05,
+ "loss": 0.0671,
+ "step": 18536
+ },
+ {
+ "epoch": 50.509536784741144,
+ "grad_norm": 2.5837008953094482,
+ "learning_rate": 1.0320731785737417e-05,
+ "loss": 0.207,
+ "step": 18537
+ },
+ {
+ "epoch": 50.51226158038147,
+ "grad_norm": 2.252709150314331,
+ "learning_rate": 1.0319849743876904e-05,
+ "loss": 0.0643,
+ "step": 18538
+ },
+ {
+ "epoch": 50.514986376021795,
+ "grad_norm": 2.474256992340088,
+ "learning_rate": 1.0318967699525418e-05,
+ "loss": 0.0347,
+ "step": 18539
+ },
+ {
+ "epoch": 50.51771117166213,
+ "grad_norm": 2.087270736694336,
+ "learning_rate": 1.0318085652689817e-05,
+ "loss": 0.1557,
+ "step": 18540
+ },
+ {
+ "epoch": 50.520435967302454,
+ "grad_norm": 3.2081549167633057,
+ "learning_rate": 1.0317203603376978e-05,
+ "loss": 0.0469,
+ "step": 18541
+ },
+ {
+ "epoch": 50.52316076294278,
+ "grad_norm": 3.378793239593506,
+ "learning_rate": 1.0316321551593765e-05,
+ "loss": 0.0873,
+ "step": 18542
+ },
+ {
+ "epoch": 50.525885558583106,
+ "grad_norm": 2.0692546367645264,
+ "learning_rate": 1.0315439497347053e-05,
+ "loss": 0.0398,
+ "step": 18543
+ },
+ {
+ "epoch": 50.52861035422343,
+ "grad_norm": 4.412168979644775,
+ "learning_rate": 1.0314557440643704e-05,
+ "loss": 0.0978,
+ "step": 18544
+ },
+ {
+ "epoch": 50.53133514986376,
+ "grad_norm": 2.892982244491577,
+ "learning_rate": 1.0313675381490596e-05,
+ "loss": 0.1654,
+ "step": 18545
+ },
+ {
+ "epoch": 50.53405994550409,
+ "grad_norm": 3.252068042755127,
+ "learning_rate": 1.031279331989459e-05,
+ "loss": 0.0512,
+ "step": 18546
+ },
+ {
+ "epoch": 50.536784741144416,
+ "grad_norm": 2.117605209350586,
+ "learning_rate": 1.0311911255862563e-05,
+ "loss": 0.0404,
+ "step": 18547
+ },
+ {
+ "epoch": 50.53950953678474,
+ "grad_norm": 3.4846999645233154,
+ "learning_rate": 1.0311029189401381e-05,
+ "loss": 0.1281,
+ "step": 18548
+ },
+ {
+ "epoch": 50.54223433242507,
+ "grad_norm": 2.8161637783050537,
+ "learning_rate": 1.0310147120517914e-05,
+ "loss": 0.271,
+ "step": 18549
+ },
+ {
+ "epoch": 50.54495912806539,
+ "grad_norm": 2.3538458347320557,
+ "learning_rate": 1.030926504921903e-05,
+ "loss": 0.1002,
+ "step": 18550
+ },
+ {
+ "epoch": 50.54768392370572,
+ "grad_norm": 2.717883825302124,
+ "learning_rate": 1.0308382975511601e-05,
+ "loss": 0.0756,
+ "step": 18551
+ },
+ {
+ "epoch": 50.55040871934605,
+ "grad_norm": 3.1137967109680176,
+ "learning_rate": 1.0307500899402493e-05,
+ "loss": 0.1991,
+ "step": 18552
+ },
+ {
+ "epoch": 50.55313351498638,
+ "grad_norm": 2.3722102642059326,
+ "learning_rate": 1.0306618820898579e-05,
+ "loss": 0.0484,
+ "step": 18553
+ },
+ {
+ "epoch": 50.555858310626704,
+ "grad_norm": 2.7273824214935303,
+ "learning_rate": 1.0305736740006726e-05,
+ "loss": 0.0747,
+ "step": 18554
+ },
+ {
+ "epoch": 50.55858310626703,
+ "grad_norm": 3.3231875896453857,
+ "learning_rate": 1.0304854656733808e-05,
+ "loss": 0.0636,
+ "step": 18555
+ },
+ {
+ "epoch": 50.561307901907355,
+ "grad_norm": 3.0324549674987793,
+ "learning_rate": 1.0303972571086691e-05,
+ "loss": 0.056,
+ "step": 18556
+ },
+ {
+ "epoch": 50.56403269754768,
+ "grad_norm": 3.2079200744628906,
+ "learning_rate": 1.0303090483072247e-05,
+ "loss": 0.0779,
+ "step": 18557
+ },
+ {
+ "epoch": 50.566757493188014,
+ "grad_norm": 2.3272249698638916,
+ "learning_rate": 1.0302208392697339e-05,
+ "loss": 0.0514,
+ "step": 18558
+ },
+ {
+ "epoch": 50.56948228882834,
+ "grad_norm": 2.1616129875183105,
+ "learning_rate": 1.0301326299968848e-05,
+ "loss": 0.0552,
+ "step": 18559
+ },
+ {
+ "epoch": 50.572207084468666,
+ "grad_norm": 3.248126983642578,
+ "learning_rate": 1.0300444204893634e-05,
+ "loss": 0.1552,
+ "step": 18560
+ },
+ {
+ "epoch": 50.57493188010899,
+ "grad_norm": 2.7528839111328125,
+ "learning_rate": 1.0299562107478569e-05,
+ "loss": 0.0405,
+ "step": 18561
+ },
+ {
+ "epoch": 50.57765667574932,
+ "grad_norm": 2.810513973236084,
+ "learning_rate": 1.0298680007730525e-05,
+ "loss": 0.1699,
+ "step": 18562
+ },
+ {
+ "epoch": 50.58038147138964,
+ "grad_norm": 2.795664072036743,
+ "learning_rate": 1.029779790565637e-05,
+ "loss": 0.1624,
+ "step": 18563
+ },
+ {
+ "epoch": 50.583106267029976,
+ "grad_norm": 1.9692680835723877,
+ "learning_rate": 1.0296915801262977e-05,
+ "loss": 0.0244,
+ "step": 18564
+ },
+ {
+ "epoch": 50.5858310626703,
+ "grad_norm": 2.2593255043029785,
+ "learning_rate": 1.0296033694557213e-05,
+ "loss": 0.1039,
+ "step": 18565
+ },
+ {
+ "epoch": 50.58855585831063,
+ "grad_norm": 2.9348626136779785,
+ "learning_rate": 1.0295151585545947e-05,
+ "loss": 0.0533,
+ "step": 18566
+ },
+ {
+ "epoch": 50.59128065395095,
+ "grad_norm": 2.3301448822021484,
+ "learning_rate": 1.029426947423605e-05,
+ "loss": 0.05,
+ "step": 18567
+ },
+ {
+ "epoch": 50.59400544959128,
+ "grad_norm": 2.060809373855591,
+ "learning_rate": 1.029338736063439e-05,
+ "loss": 0.0315,
+ "step": 18568
+ },
+ {
+ "epoch": 50.596730245231605,
+ "grad_norm": 3.8944640159606934,
+ "learning_rate": 1.0292505244747843e-05,
+ "loss": 0.0607,
+ "step": 18569
+ },
+ {
+ "epoch": 50.59945504087194,
+ "grad_norm": 2.4633071422576904,
+ "learning_rate": 1.0291623126583268e-05,
+ "loss": 0.0686,
+ "step": 18570
+ },
+ {
+ "epoch": 50.60217983651226,
+ "grad_norm": 1.8553107976913452,
+ "learning_rate": 1.0290741006147548e-05,
+ "loss": 0.0357,
+ "step": 18571
+ },
+ {
+ "epoch": 50.60490463215259,
+ "grad_norm": 3.145019054412842,
+ "learning_rate": 1.0289858883447544e-05,
+ "loss": 0.104,
+ "step": 18572
+ },
+ {
+ "epoch": 50.607629427792915,
+ "grad_norm": 2.2033746242523193,
+ "learning_rate": 1.0288976758490129e-05,
+ "loss": 0.0742,
+ "step": 18573
+ },
+ {
+ "epoch": 50.61035422343324,
+ "grad_norm": 2.453275203704834,
+ "learning_rate": 1.028809463128217e-05,
+ "loss": 0.0685,
+ "step": 18574
+ },
+ {
+ "epoch": 50.61307901907357,
+ "grad_norm": 2.553650379180908,
+ "learning_rate": 1.0287212501830543e-05,
+ "loss": 0.1545,
+ "step": 18575
+ },
+ {
+ "epoch": 50.6158038147139,
+ "grad_norm": 2.7481062412261963,
+ "learning_rate": 1.0286330370142111e-05,
+ "loss": 0.0531,
+ "step": 18576
+ },
+ {
+ "epoch": 50.618528610354225,
+ "grad_norm": 2.9447884559631348,
+ "learning_rate": 1.028544823622375e-05,
+ "loss": 0.0994,
+ "step": 18577
+ },
+ {
+ "epoch": 50.62125340599455,
+ "grad_norm": 2.1282808780670166,
+ "learning_rate": 1.0284566100082323e-05,
+ "loss": 0.1908,
+ "step": 18578
+ },
+ {
+ "epoch": 50.62397820163488,
+ "grad_norm": 3.6866748332977295,
+ "learning_rate": 1.0283683961724708e-05,
+ "loss": 0.0622,
+ "step": 18579
+ },
+ {
+ "epoch": 50.6267029972752,
+ "grad_norm": 2.606356143951416,
+ "learning_rate": 1.0282801821157768e-05,
+ "loss": 0.1002,
+ "step": 18580
+ },
+ {
+ "epoch": 50.62942779291553,
+ "grad_norm": 2.4442028999328613,
+ "learning_rate": 1.0281919678388382e-05,
+ "loss": 0.1303,
+ "step": 18581
+ },
+ {
+ "epoch": 50.63215258855586,
+ "grad_norm": 2.184096336364746,
+ "learning_rate": 1.028103753342341e-05,
+ "loss": 0.0379,
+ "step": 18582
+ },
+ {
+ "epoch": 50.63487738419619,
+ "grad_norm": 2.4806909561157227,
+ "learning_rate": 1.028015538626973e-05,
+ "loss": 0.1148,
+ "step": 18583
+ },
+ {
+ "epoch": 50.63760217983651,
+ "grad_norm": 4.143237590789795,
+ "learning_rate": 1.0279273236934207e-05,
+ "loss": 0.0941,
+ "step": 18584
+ },
+ {
+ "epoch": 50.64032697547684,
+ "grad_norm": 2.4464685916900635,
+ "learning_rate": 1.0278391085423715e-05,
+ "loss": 0.0535,
+ "step": 18585
+ },
+ {
+ "epoch": 50.643051771117165,
+ "grad_norm": 2.3714137077331543,
+ "learning_rate": 1.0277508931745119e-05,
+ "loss": 0.1585,
+ "step": 18586
+ },
+ {
+ "epoch": 50.64577656675749,
+ "grad_norm": 3.1024022102355957,
+ "learning_rate": 1.0276626775905294e-05,
+ "loss": 0.1127,
+ "step": 18587
+ },
+ {
+ "epoch": 50.64850136239782,
+ "grad_norm": 3.2057077884674072,
+ "learning_rate": 1.0275744617911107e-05,
+ "loss": 0.1187,
+ "step": 18588
+ },
+ {
+ "epoch": 50.65122615803815,
+ "grad_norm": 1.9499858617782593,
+ "learning_rate": 1.0274862457769432e-05,
+ "loss": 0.0654,
+ "step": 18589
+ },
+ {
+ "epoch": 50.653950953678475,
+ "grad_norm": 2.8086812496185303,
+ "learning_rate": 1.0273980295487136e-05,
+ "loss": 0.0627,
+ "step": 18590
+ },
+ {
+ "epoch": 50.6566757493188,
+ "grad_norm": 3.22538685798645,
+ "learning_rate": 1.027309813107109e-05,
+ "loss": 0.0854,
+ "step": 18591
+ },
+ {
+ "epoch": 50.65940054495913,
+ "grad_norm": 3.8813469409942627,
+ "learning_rate": 1.0272215964528168e-05,
+ "loss": 0.0464,
+ "step": 18592
+ },
+ {
+ "epoch": 50.66212534059945,
+ "grad_norm": 2.948739767074585,
+ "learning_rate": 1.0271333795865233e-05,
+ "loss": 0.0724,
+ "step": 18593
+ },
+ {
+ "epoch": 50.664850136239785,
+ "grad_norm": 2.8052353858947754,
+ "learning_rate": 1.027045162508916e-05,
+ "loss": 0.1141,
+ "step": 18594
+ },
+ {
+ "epoch": 50.66757493188011,
+ "grad_norm": 1.9035429954528809,
+ "learning_rate": 1.026956945220682e-05,
+ "loss": 0.0385,
+ "step": 18595
+ },
+ {
+ "epoch": 50.67029972752044,
+ "grad_norm": 2.290461301803589,
+ "learning_rate": 1.026868727722508e-05,
+ "loss": 0.0516,
+ "step": 18596
+ },
+ {
+ "epoch": 50.67302452316076,
+ "grad_norm": 2.4504435062408447,
+ "learning_rate": 1.0267805100150811e-05,
+ "loss": 0.1031,
+ "step": 18597
+ },
+ {
+ "epoch": 50.67574931880109,
+ "grad_norm": 2.1904091835021973,
+ "learning_rate": 1.0266922920990885e-05,
+ "loss": 0.0454,
+ "step": 18598
+ },
+ {
+ "epoch": 50.678474114441414,
+ "grad_norm": 2.863295316696167,
+ "learning_rate": 1.0266040739752175e-05,
+ "loss": 0.1758,
+ "step": 18599
+ },
+ {
+ "epoch": 50.68119891008175,
+ "grad_norm": 2.688398838043213,
+ "learning_rate": 1.0265158556441543e-05,
+ "loss": 0.0514,
+ "step": 18600
+ },
+ {
+ "epoch": 50.68392370572207,
+ "grad_norm": 3.1517744064331055,
+ "learning_rate": 1.0264276371065871e-05,
+ "loss": 0.0709,
+ "step": 18601
+ },
+ {
+ "epoch": 50.6866485013624,
+ "grad_norm": 2.768472671508789,
+ "learning_rate": 1.026339418363202e-05,
+ "loss": 0.0751,
+ "step": 18602
+ },
+ {
+ "epoch": 50.689373297002724,
+ "grad_norm": 3.54105281829834,
+ "learning_rate": 1.0262511994146862e-05,
+ "loss": 0.1115,
+ "step": 18603
+ },
+ {
+ "epoch": 50.69209809264305,
+ "grad_norm": 2.9117233753204346,
+ "learning_rate": 1.026162980261727e-05,
+ "loss": 0.1122,
+ "step": 18604
+ },
+ {
+ "epoch": 50.694822888283376,
+ "grad_norm": 2.1712419986724854,
+ "learning_rate": 1.0260747609050115e-05,
+ "loss": 0.0355,
+ "step": 18605
+ },
+ {
+ "epoch": 50.69754768392371,
+ "grad_norm": 2.4386489391326904,
+ "learning_rate": 1.0259865413452266e-05,
+ "loss": 0.0489,
+ "step": 18606
+ },
+ {
+ "epoch": 50.700272479564035,
+ "grad_norm": 2.4155285358428955,
+ "learning_rate": 1.0258983215830591e-05,
+ "loss": 0.0903,
+ "step": 18607
+ },
+ {
+ "epoch": 50.70299727520436,
+ "grad_norm": 3.977341890335083,
+ "learning_rate": 1.0258101016191963e-05,
+ "loss": 0.2003,
+ "step": 18608
+ },
+ {
+ "epoch": 50.705722070844686,
+ "grad_norm": 2.8866870403289795,
+ "learning_rate": 1.0257218814543257e-05,
+ "loss": 0.1432,
+ "step": 18609
+ },
+ {
+ "epoch": 50.70844686648501,
+ "grad_norm": 1.8772621154785156,
+ "learning_rate": 1.0256336610891336e-05,
+ "loss": 0.0357,
+ "step": 18610
+ },
+ {
+ "epoch": 50.71117166212534,
+ "grad_norm": 2.637296199798584,
+ "learning_rate": 1.0255454405243076e-05,
+ "loss": 0.0487,
+ "step": 18611
+ },
+ {
+ "epoch": 50.71389645776567,
+ "grad_norm": 2.401993751525879,
+ "learning_rate": 1.0254572197605341e-05,
+ "loss": 0.0553,
+ "step": 18612
+ },
+ {
+ "epoch": 50.716621253406,
+ "grad_norm": 2.663942337036133,
+ "learning_rate": 1.0253689987985008e-05,
+ "loss": 0.0285,
+ "step": 18613
+ },
+ {
+ "epoch": 50.71934604904632,
+ "grad_norm": 2.629486560821533,
+ "learning_rate": 1.0252807776388946e-05,
+ "loss": 0.0567,
+ "step": 18614
+ },
+ {
+ "epoch": 50.72207084468665,
+ "grad_norm": 3.0123536586761475,
+ "learning_rate": 1.0251925562824026e-05,
+ "loss": 0.0717,
+ "step": 18615
+ },
+ {
+ "epoch": 50.724795640326974,
+ "grad_norm": 2.1439590454101562,
+ "learning_rate": 1.0251043347297118e-05,
+ "loss": 0.034,
+ "step": 18616
+ },
+ {
+ "epoch": 50.7275204359673,
+ "grad_norm": 2.8541646003723145,
+ "learning_rate": 1.0250161129815091e-05,
+ "loss": 0.051,
+ "step": 18617
+ },
+ {
+ "epoch": 50.73024523160763,
+ "grad_norm": 2.314472198486328,
+ "learning_rate": 1.0249278910384817e-05,
+ "loss": 0.1459,
+ "step": 18618
+ },
+ {
+ "epoch": 50.73297002724796,
+ "grad_norm": 23.229934692382812,
+ "learning_rate": 1.0248396689013166e-05,
+ "loss": 0.049,
+ "step": 18619
+ },
+ {
+ "epoch": 50.735694822888284,
+ "grad_norm": 2.5702264308929443,
+ "learning_rate": 1.0247514465707012e-05,
+ "loss": 0.0355,
+ "step": 18620
+ },
+ {
+ "epoch": 50.73841961852861,
+ "grad_norm": 3.3073246479034424,
+ "learning_rate": 1.024663224047322e-05,
+ "loss": 0.0699,
+ "step": 18621
+ },
+ {
+ "epoch": 50.741144414168936,
+ "grad_norm": 3.421494245529175,
+ "learning_rate": 1.0245750013318669e-05,
+ "loss": 0.0754,
+ "step": 18622
+ },
+ {
+ "epoch": 50.74386920980926,
+ "grad_norm": 2.487290382385254,
+ "learning_rate": 1.024486778425022e-05,
+ "loss": 0.1118,
+ "step": 18623
+ },
+ {
+ "epoch": 50.746594005449595,
+ "grad_norm": 2.6211984157562256,
+ "learning_rate": 1.0243985553274753e-05,
+ "loss": 0.0638,
+ "step": 18624
+ },
+ {
+ "epoch": 50.74931880108992,
+ "grad_norm": 2.966541290283203,
+ "learning_rate": 1.024310332039913e-05,
+ "loss": 0.0649,
+ "step": 18625
+ },
+ {
+ "epoch": 50.752043596730246,
+ "grad_norm": 2.7331910133361816,
+ "learning_rate": 1.024222108563023e-05,
+ "loss": 0.0879,
+ "step": 18626
+ },
+ {
+ "epoch": 50.75476839237057,
+ "grad_norm": 2.787611246109009,
+ "learning_rate": 1.0241338848974918e-05,
+ "loss": 0.1236,
+ "step": 18627
+ },
+ {
+ "epoch": 50.7574931880109,
+ "grad_norm": 2.4648876190185547,
+ "learning_rate": 1.0240456610440068e-05,
+ "loss": 0.2556,
+ "step": 18628
+ },
+ {
+ "epoch": 50.76021798365122,
+ "grad_norm": 2.3832249641418457,
+ "learning_rate": 1.0239574370032548e-05,
+ "loss": 0.1083,
+ "step": 18629
+ },
+ {
+ "epoch": 50.762942779291556,
+ "grad_norm": 4.190240383148193,
+ "learning_rate": 1.0238692127759232e-05,
+ "loss": 0.1152,
+ "step": 18630
+ },
+ {
+ "epoch": 50.76566757493188,
+ "grad_norm": 2.576481819152832,
+ "learning_rate": 1.0237809883626986e-05,
+ "loss": 0.0658,
+ "step": 18631
+ },
+ {
+ "epoch": 50.76839237057221,
+ "grad_norm": 2.9102063179016113,
+ "learning_rate": 1.0236927637642688e-05,
+ "loss": 0.147,
+ "step": 18632
+ },
+ {
+ "epoch": 50.771117166212534,
+ "grad_norm": 3.111394166946411,
+ "learning_rate": 1.0236045389813203e-05,
+ "loss": 0.134,
+ "step": 18633
+ },
+ {
+ "epoch": 50.77384196185286,
+ "grad_norm": 2.8597259521484375,
+ "learning_rate": 1.0235163140145408e-05,
+ "loss": 0.2513,
+ "step": 18634
+ },
+ {
+ "epoch": 50.776566757493185,
+ "grad_norm": 4.155642509460449,
+ "learning_rate": 1.0234280888646166e-05,
+ "loss": 0.047,
+ "step": 18635
+ },
+ {
+ "epoch": 50.77929155313352,
+ "grad_norm": 2.6606147289276123,
+ "learning_rate": 1.0233398635322353e-05,
+ "loss": 0.0691,
+ "step": 18636
+ },
+ {
+ "epoch": 50.782016348773844,
+ "grad_norm": 3.1796674728393555,
+ "learning_rate": 1.023251638018084e-05,
+ "loss": 0.0587,
+ "step": 18637
+ },
+ {
+ "epoch": 50.78474114441417,
+ "grad_norm": 1.7821273803710938,
+ "learning_rate": 1.0231634123228498e-05,
+ "loss": 0.0338,
+ "step": 18638
+ },
+ {
+ "epoch": 50.787465940054496,
+ "grad_norm": 3.346360683441162,
+ "learning_rate": 1.023075186447219e-05,
+ "loss": 0.0504,
+ "step": 18639
+ },
+ {
+ "epoch": 50.79019073569482,
+ "grad_norm": 2.9172120094299316,
+ "learning_rate": 1.0229869603918801e-05,
+ "loss": 0.1426,
+ "step": 18640
+ },
+ {
+ "epoch": 50.79291553133515,
+ "grad_norm": 2.524420976638794,
+ "learning_rate": 1.0228987341575192e-05,
+ "loss": 0.0662,
+ "step": 18641
+ },
+ {
+ "epoch": 50.79564032697548,
+ "grad_norm": 2.583484172821045,
+ "learning_rate": 1.0228105077448239e-05,
+ "loss": 0.06,
+ "step": 18642
+ },
+ {
+ "epoch": 50.798365122615806,
+ "grad_norm": 2.5588879585266113,
+ "learning_rate": 1.0227222811544807e-05,
+ "loss": 0.0554,
+ "step": 18643
+ },
+ {
+ "epoch": 50.80108991825613,
+ "grad_norm": 2.4002718925476074,
+ "learning_rate": 1.0226340543871778e-05,
+ "loss": 0.1344,
+ "step": 18644
+ },
+ {
+ "epoch": 50.80381471389646,
+ "grad_norm": 2.36399245262146,
+ "learning_rate": 1.0225458274436011e-05,
+ "loss": 0.0753,
+ "step": 18645
+ },
+ {
+ "epoch": 50.80653950953678,
+ "grad_norm": 3.08683443069458,
+ "learning_rate": 1.0224576003244385e-05,
+ "loss": 0.1273,
+ "step": 18646
+ },
+ {
+ "epoch": 50.80926430517711,
+ "grad_norm": 2.7668070793151855,
+ "learning_rate": 1.0223693730303761e-05,
+ "loss": 0.1882,
+ "step": 18647
+ },
+ {
+ "epoch": 50.81198910081744,
+ "grad_norm": 2.8716928958892822,
+ "learning_rate": 1.0222811455621025e-05,
+ "loss": 0.3281,
+ "step": 18648
+ },
+ {
+ "epoch": 50.81471389645777,
+ "grad_norm": 2.880939483642578,
+ "learning_rate": 1.0221929179203038e-05,
+ "loss": 0.0598,
+ "step": 18649
+ },
+ {
+ "epoch": 50.817438692098094,
+ "grad_norm": 2.5432136058807373,
+ "learning_rate": 1.0221046901056675e-05,
+ "loss": 0.0622,
+ "step": 18650
+ },
+ {
+ "epoch": 50.82016348773842,
+ "grad_norm": 9.822779655456543,
+ "learning_rate": 1.0220164621188804e-05,
+ "loss": 0.0969,
+ "step": 18651
+ },
+ {
+ "epoch": 50.822888283378745,
+ "grad_norm": 2.362976551055908,
+ "learning_rate": 1.02192823396063e-05,
+ "loss": 0.1758,
+ "step": 18652
+ },
+ {
+ "epoch": 50.82561307901907,
+ "grad_norm": 3.1506354808807373,
+ "learning_rate": 1.0218400056316031e-05,
+ "loss": 0.0803,
+ "step": 18653
+ },
+ {
+ "epoch": 50.828337874659404,
+ "grad_norm": 3.226322650909424,
+ "learning_rate": 1.0217517771324871e-05,
+ "loss": 0.3215,
+ "step": 18654
+ },
+ {
+ "epoch": 50.83106267029973,
+ "grad_norm": 2.9490394592285156,
+ "learning_rate": 1.0216635484639688e-05,
+ "loss": 0.0832,
+ "step": 18655
+ },
+ {
+ "epoch": 50.833787465940055,
+ "grad_norm": 3.1897695064544678,
+ "learning_rate": 1.0215753196267352e-05,
+ "loss": 0.11,
+ "step": 18656
+ },
+ {
+ "epoch": 50.83651226158038,
+ "grad_norm": 3.257929563522339,
+ "learning_rate": 1.021487090621474e-05,
+ "loss": 0.0331,
+ "step": 18657
+ },
+ {
+ "epoch": 50.83923705722071,
+ "grad_norm": 2.536243438720703,
+ "learning_rate": 1.0213988614488721e-05,
+ "loss": 0.0882,
+ "step": 18658
+ },
+ {
+ "epoch": 50.84196185286103,
+ "grad_norm": 2.5097768306732178,
+ "learning_rate": 1.0213106321096164e-05,
+ "loss": 0.1198,
+ "step": 18659
+ },
+ {
+ "epoch": 50.844686648501366,
+ "grad_norm": 2.333688259124756,
+ "learning_rate": 1.0212224026043945e-05,
+ "loss": 0.0495,
+ "step": 18660
+ },
+ {
+ "epoch": 50.84741144414169,
+ "grad_norm": 2.5300302505493164,
+ "learning_rate": 1.0211341729338928e-05,
+ "loss": 0.0726,
+ "step": 18661
+ },
+ {
+ "epoch": 50.85013623978202,
+ "grad_norm": 2.3558905124664307,
+ "learning_rate": 1.0210459430987993e-05,
+ "loss": 0.046,
+ "step": 18662
+ },
+ {
+ "epoch": 50.85286103542234,
+ "grad_norm": 2.9479422569274902,
+ "learning_rate": 1.0209577130998002e-05,
+ "loss": 0.1341,
+ "step": 18663
+ },
+ {
+ "epoch": 50.85558583106267,
+ "grad_norm": 3.1183485984802246,
+ "learning_rate": 1.0208694829375834e-05,
+ "loss": 0.1232,
+ "step": 18664
+ },
+ {
+ "epoch": 50.858310626702995,
+ "grad_norm": 1.7531911134719849,
+ "learning_rate": 1.0207812526128356e-05,
+ "loss": 0.0456,
+ "step": 18665
+ },
+ {
+ "epoch": 50.86103542234333,
+ "grad_norm": 2.973123788833618,
+ "learning_rate": 1.020693022126244e-05,
+ "loss": 0.0617,
+ "step": 18666
+ },
+ {
+ "epoch": 50.86376021798365,
+ "grad_norm": 2.424668550491333,
+ "learning_rate": 1.020604791478496e-05,
+ "loss": 0.1455,
+ "step": 18667
+ },
+ {
+ "epoch": 50.86648501362398,
+ "grad_norm": 4.784740447998047,
+ "learning_rate": 1.0205165606702784e-05,
+ "loss": 0.148,
+ "step": 18668
+ },
+ {
+ "epoch": 50.869209809264305,
+ "grad_norm": 2.3737030029296875,
+ "learning_rate": 1.0204283297022786e-05,
+ "loss": 0.1046,
+ "step": 18669
+ },
+ {
+ "epoch": 50.87193460490463,
+ "grad_norm": 3.33147931098938,
+ "learning_rate": 1.0203400985751838e-05,
+ "loss": 0.155,
+ "step": 18670
+ },
+ {
+ "epoch": 50.87465940054496,
+ "grad_norm": 7.86328649520874,
+ "learning_rate": 1.0202518672896806e-05,
+ "loss": 0.0466,
+ "step": 18671
+ },
+ {
+ "epoch": 50.87738419618529,
+ "grad_norm": 5.010165214538574,
+ "learning_rate": 1.0201636358464568e-05,
+ "loss": 0.1775,
+ "step": 18672
+ },
+ {
+ "epoch": 50.880108991825615,
+ "grad_norm": 3.451033353805542,
+ "learning_rate": 1.0200754042461988e-05,
+ "loss": 0.0908,
+ "step": 18673
+ },
+ {
+ "epoch": 50.88283378746594,
+ "grad_norm": 2.4235682487487793,
+ "learning_rate": 1.0199871724895947e-05,
+ "loss": 0.0658,
+ "step": 18674
+ },
+ {
+ "epoch": 50.88555858310627,
+ "grad_norm": 2.6110141277313232,
+ "learning_rate": 1.0198989405773307e-05,
+ "loss": 0.1092,
+ "step": 18675
+ },
+ {
+ "epoch": 50.88828337874659,
+ "grad_norm": 2.681318759918213,
+ "learning_rate": 1.019810708510095e-05,
+ "loss": 0.0884,
+ "step": 18676
+ },
+ {
+ "epoch": 50.89100817438692,
+ "grad_norm": 3.879671812057495,
+ "learning_rate": 1.0197224762885735e-05,
+ "loss": 0.1735,
+ "step": 18677
+ },
+ {
+ "epoch": 50.89373297002725,
+ "grad_norm": 4.141540050506592,
+ "learning_rate": 1.0196342439134546e-05,
+ "loss": 0.2147,
+ "step": 18678
+ },
+ {
+ "epoch": 50.89645776566758,
+ "grad_norm": 4.442753314971924,
+ "learning_rate": 1.0195460113854243e-05,
+ "loss": 0.0672,
+ "step": 18679
+ },
+ {
+ "epoch": 50.8991825613079,
+ "grad_norm": 4.286531925201416,
+ "learning_rate": 1.0194577787051708e-05,
+ "loss": 0.0673,
+ "step": 18680
+ },
+ {
+ "epoch": 50.90190735694823,
+ "grad_norm": 2.579247236251831,
+ "learning_rate": 1.0193695458733803e-05,
+ "loss": 0.062,
+ "step": 18681
+ },
+ {
+ "epoch": 50.904632152588555,
+ "grad_norm": 2.333807945251465,
+ "learning_rate": 1.0192813128907404e-05,
+ "loss": 0.21,
+ "step": 18682
+ },
+ {
+ "epoch": 50.90735694822888,
+ "grad_norm": 2.9556875228881836,
+ "learning_rate": 1.0191930797579386e-05,
+ "loss": 0.0711,
+ "step": 18683
+ },
+ {
+ "epoch": 50.91008174386921,
+ "grad_norm": 3.0188279151916504,
+ "learning_rate": 1.0191048464756615e-05,
+ "loss": 0.052,
+ "step": 18684
+ },
+ {
+ "epoch": 50.91280653950954,
+ "grad_norm": 2.3107049465179443,
+ "learning_rate": 1.0190166130445964e-05,
+ "loss": 0.0384,
+ "step": 18685
+ },
+ {
+ "epoch": 50.915531335149865,
+ "grad_norm": 2.539520025253296,
+ "learning_rate": 1.0189283794654308e-05,
+ "loss": 0.0928,
+ "step": 18686
+ },
+ {
+ "epoch": 50.91825613079019,
+ "grad_norm": 2.8103156089782715,
+ "learning_rate": 1.0188401457388513e-05,
+ "loss": 0.0668,
+ "step": 18687
+ },
+ {
+ "epoch": 50.920980926430516,
+ "grad_norm": 1.784070372581482,
+ "learning_rate": 1.0187519118655456e-05,
+ "loss": 0.1218,
+ "step": 18688
+ },
+ {
+ "epoch": 50.92370572207084,
+ "grad_norm": 2.322049617767334,
+ "learning_rate": 1.0186636778462002e-05,
+ "loss": 0.0435,
+ "step": 18689
+ },
+ {
+ "epoch": 50.926430517711175,
+ "grad_norm": 2.8020389080047607,
+ "learning_rate": 1.0185754436815029e-05,
+ "loss": 0.0586,
+ "step": 18690
+ },
+ {
+ "epoch": 50.9291553133515,
+ "grad_norm": 3.118302822113037,
+ "learning_rate": 1.018487209372141e-05,
+ "loss": 0.1204,
+ "step": 18691
+ },
+ {
+ "epoch": 50.93188010899183,
+ "grad_norm": 3.0160865783691406,
+ "learning_rate": 1.0183989749188009e-05,
+ "loss": 0.0975,
+ "step": 18692
+ },
+ {
+ "epoch": 50.93460490463215,
+ "grad_norm": 2.4401586055755615,
+ "learning_rate": 1.0183107403221703e-05,
+ "loss": 0.071,
+ "step": 18693
+ },
+ {
+ "epoch": 50.93732970027248,
+ "grad_norm": 10.61763858795166,
+ "learning_rate": 1.0182225055829362e-05,
+ "loss": 0.1202,
+ "step": 18694
+ },
+ {
+ "epoch": 50.940054495912804,
+ "grad_norm": 2.191387414932251,
+ "learning_rate": 1.0181342707017859e-05,
+ "loss": 0.04,
+ "step": 18695
+ },
+ {
+ "epoch": 50.94277929155314,
+ "grad_norm": 3.287543773651123,
+ "learning_rate": 1.0180460356794065e-05,
+ "loss": 0.1414,
+ "step": 18696
+ },
+ {
+ "epoch": 50.94550408719346,
+ "grad_norm": 2.9383716583251953,
+ "learning_rate": 1.017957800516485e-05,
+ "loss": 0.1467,
+ "step": 18697
+ },
+ {
+ "epoch": 50.94822888283379,
+ "grad_norm": 3.2955079078674316,
+ "learning_rate": 1.0178695652137088e-05,
+ "loss": 0.1239,
+ "step": 18698
+ },
+ {
+ "epoch": 50.950953678474114,
+ "grad_norm": 2.4495925903320312,
+ "learning_rate": 1.0177813297717649e-05,
+ "loss": 0.0607,
+ "step": 18699
+ },
+ {
+ "epoch": 50.95367847411444,
+ "grad_norm": 3.510603427886963,
+ "learning_rate": 1.0176930941913405e-05,
+ "loss": 0.1061,
+ "step": 18700
+ },
+ {
+ "epoch": 50.956403269754766,
+ "grad_norm": 2.7640275955200195,
+ "learning_rate": 1.0176048584731233e-05,
+ "loss": 0.0552,
+ "step": 18701
+ },
+ {
+ "epoch": 50.95912806539509,
+ "grad_norm": 2.268214464187622,
+ "learning_rate": 1.0175166226177998e-05,
+ "loss": 0.0561,
+ "step": 18702
+ },
+ {
+ "epoch": 50.961852861035425,
+ "grad_norm": 2.3939943313598633,
+ "learning_rate": 1.0174283866260575e-05,
+ "loss": 0.0707,
+ "step": 18703
+ },
+ {
+ "epoch": 50.96457765667575,
+ "grad_norm": 2.748398542404175,
+ "learning_rate": 1.0173401504985831e-05,
+ "loss": 0.1885,
+ "step": 18704
+ },
+ {
+ "epoch": 50.967302452316076,
+ "grad_norm": 2.0087780952453613,
+ "learning_rate": 1.0172519142360646e-05,
+ "loss": 0.0462,
+ "step": 18705
+ },
+ {
+ "epoch": 50.9700272479564,
+ "grad_norm": 2.812889814376831,
+ "learning_rate": 1.0171636778391885e-05,
+ "loss": 0.0688,
+ "step": 18706
+ },
+ {
+ "epoch": 50.97275204359673,
+ "grad_norm": 3.395920991897583,
+ "learning_rate": 1.0170754413086422e-05,
+ "loss": 0.0668,
+ "step": 18707
+ },
+ {
+ "epoch": 50.97547683923706,
+ "grad_norm": 2.818784713745117,
+ "learning_rate": 1.016987204645113e-05,
+ "loss": 0.2136,
+ "step": 18708
+ },
+ {
+ "epoch": 50.97820163487739,
+ "grad_norm": 2.8888304233551025,
+ "learning_rate": 1.016898967849288e-05,
+ "loss": 0.1182,
+ "step": 18709
+ },
+ {
+ "epoch": 50.98092643051771,
+ "grad_norm": 3.5326015949249268,
+ "learning_rate": 1.0168107309218545e-05,
+ "loss": 0.0559,
+ "step": 18710
+ },
+ {
+ "epoch": 50.98365122615804,
+ "grad_norm": 2.6787590980529785,
+ "learning_rate": 1.0167224938634994e-05,
+ "loss": 0.148,
+ "step": 18711
+ },
+ {
+ "epoch": 50.986376021798364,
+ "grad_norm": 3.578200101852417,
+ "learning_rate": 1.01663425667491e-05,
+ "loss": 0.1116,
+ "step": 18712
+ },
+ {
+ "epoch": 50.98910081743869,
+ "grad_norm": 2.2192118167877197,
+ "learning_rate": 1.016546019356774e-05,
+ "loss": 0.0394,
+ "step": 18713
+ },
+ {
+ "epoch": 50.991825613079016,
+ "grad_norm": 3.779367685317993,
+ "learning_rate": 1.0164577819097777e-05,
+ "loss": 0.1056,
+ "step": 18714
+ },
+ {
+ "epoch": 50.99455040871935,
+ "grad_norm": 2.4295153617858887,
+ "learning_rate": 1.016369544334609e-05,
+ "loss": 0.0605,
+ "step": 18715
+ },
+ {
+ "epoch": 50.997275204359674,
+ "grad_norm": 2.3781232833862305,
+ "learning_rate": 1.0162813066319542e-05,
+ "loss": 0.0516,
+ "step": 18716
+ },
+ {
+ "epoch": 51.0,
+ "grad_norm": 2.862762212753296,
+ "learning_rate": 1.0161930688025018e-05,
+ "loss": 0.0763,
+ "step": 18717
+ },
+ {
+ "epoch": 51.002724795640326,
+ "grad_norm": 2.2284607887268066,
+ "learning_rate": 1.0161048308469378e-05,
+ "loss": 0.1124,
+ "step": 18718
+ },
+ {
+ "epoch": 51.00544959128065,
+ "grad_norm": 2.6494698524475098,
+ "learning_rate": 1.0160165927659502e-05,
+ "loss": 0.0661,
+ "step": 18719
+ },
+ {
+ "epoch": 51.00817438692098,
+ "grad_norm": 2.8065054416656494,
+ "learning_rate": 1.015928354560226e-05,
+ "loss": 0.106,
+ "step": 18720
+ },
+ {
+ "epoch": 51.01089918256131,
+ "grad_norm": 2.6253273487091064,
+ "learning_rate": 1.0158401162304522e-05,
+ "loss": 0.2454,
+ "step": 18721
+ },
+ {
+ "epoch": 51.013623978201636,
+ "grad_norm": 2.650952100753784,
+ "learning_rate": 1.0157518777773159e-05,
+ "loss": 0.0366,
+ "step": 18722
+ },
+ {
+ "epoch": 51.01634877384196,
+ "grad_norm": 2.948214292526245,
+ "learning_rate": 1.0156636392015046e-05,
+ "loss": 0.1697,
+ "step": 18723
+ },
+ {
+ "epoch": 51.01907356948229,
+ "grad_norm": 3.20794677734375,
+ "learning_rate": 1.0155754005037054e-05,
+ "loss": 0.0661,
+ "step": 18724
+ },
+ {
+ "epoch": 51.02179836512261,
+ "grad_norm": 3.0082881450653076,
+ "learning_rate": 1.015487161684605e-05,
+ "loss": 0.2154,
+ "step": 18725
+ },
+ {
+ "epoch": 51.02452316076294,
+ "grad_norm": 3.5602872371673584,
+ "learning_rate": 1.0153989227448917e-05,
+ "loss": 0.1069,
+ "step": 18726
+ },
+ {
+ "epoch": 51.02724795640327,
+ "grad_norm": 2.742253065109253,
+ "learning_rate": 1.015310683685252e-05,
+ "loss": 0.045,
+ "step": 18727
+ },
+ {
+ "epoch": 51.0299727520436,
+ "grad_norm": 3.2902886867523193,
+ "learning_rate": 1.015222444506373e-05,
+ "loss": 0.1133,
+ "step": 18728
+ },
+ {
+ "epoch": 51.032697547683924,
+ "grad_norm": 2.9077887535095215,
+ "learning_rate": 1.0151342052089423e-05,
+ "loss": 0.144,
+ "step": 18729
+ },
+ {
+ "epoch": 51.03542234332425,
+ "grad_norm": 3.729621648788452,
+ "learning_rate": 1.0150459657936469e-05,
+ "loss": 0.0654,
+ "step": 18730
+ },
+ {
+ "epoch": 51.038147138964575,
+ "grad_norm": 3.848940849304199,
+ "learning_rate": 1.0149577262611739e-05,
+ "loss": 0.0646,
+ "step": 18731
+ },
+ {
+ "epoch": 51.0408719346049,
+ "grad_norm": 2.967320203781128,
+ "learning_rate": 1.0148694866122107e-05,
+ "loss": 0.1076,
+ "step": 18732
+ },
+ {
+ "epoch": 51.043596730245234,
+ "grad_norm": 4.041110515594482,
+ "learning_rate": 1.014781246847444e-05,
+ "loss": 0.1251,
+ "step": 18733
+ },
+ {
+ "epoch": 51.04632152588556,
+ "grad_norm": 2.4757273197174072,
+ "learning_rate": 1.0146930069675617e-05,
+ "loss": 0.037,
+ "step": 18734
+ },
+ {
+ "epoch": 51.049046321525886,
+ "grad_norm": 2.0562996864318848,
+ "learning_rate": 1.014604766973251e-05,
+ "loss": 0.0406,
+ "step": 18735
+ },
+ {
+ "epoch": 51.05177111716621,
+ "grad_norm": 3.958940267562866,
+ "learning_rate": 1.0145165268651985e-05,
+ "loss": 0.0769,
+ "step": 18736
+ },
+ {
+ "epoch": 51.05449591280654,
+ "grad_norm": 2.5543556213378906,
+ "learning_rate": 1.014428286644092e-05,
+ "loss": 0.0859,
+ "step": 18737
+ },
+ {
+ "epoch": 51.05722070844686,
+ "grad_norm": 2.970541477203369,
+ "learning_rate": 1.0143400463106184e-05,
+ "loss": 0.051,
+ "step": 18738
+ },
+ {
+ "epoch": 51.059945504087196,
+ "grad_norm": 1.7725374698638916,
+ "learning_rate": 1.0142518058654649e-05,
+ "loss": 0.0312,
+ "step": 18739
+ },
+ {
+ "epoch": 51.06267029972752,
+ "grad_norm": 2.5775790214538574,
+ "learning_rate": 1.0141635653093187e-05,
+ "loss": 0.058,
+ "step": 18740
+ },
+ {
+ "epoch": 51.06539509536785,
+ "grad_norm": 2.7342469692230225,
+ "learning_rate": 1.0140753246428674e-05,
+ "loss": 0.0565,
+ "step": 18741
+ },
+ {
+ "epoch": 51.06811989100817,
+ "grad_norm": 3.0570068359375,
+ "learning_rate": 1.0139870838667976e-05,
+ "loss": 0.1484,
+ "step": 18742
+ },
+ {
+ "epoch": 51.0708446866485,
+ "grad_norm": 2.0890145301818848,
+ "learning_rate": 1.0138988429817971e-05,
+ "loss": 0.0841,
+ "step": 18743
+ },
+ {
+ "epoch": 51.073569482288825,
+ "grad_norm": 2.430305242538452,
+ "learning_rate": 1.0138106019885525e-05,
+ "loss": 0.0833,
+ "step": 18744
+ },
+ {
+ "epoch": 51.07629427792916,
+ "grad_norm": 2.227113723754883,
+ "learning_rate": 1.0137223608877518e-05,
+ "loss": 0.1315,
+ "step": 18745
+ },
+ {
+ "epoch": 51.079019073569484,
+ "grad_norm": 2.940873384475708,
+ "learning_rate": 1.0136341196800816e-05,
+ "loss": 0.1173,
+ "step": 18746
+ },
+ {
+ "epoch": 51.08174386920981,
+ "grad_norm": 2.7617294788360596,
+ "learning_rate": 1.0135458783662296e-05,
+ "loss": 0.0655,
+ "step": 18747
+ },
+ {
+ "epoch": 51.084468664850135,
+ "grad_norm": 3.351942777633667,
+ "learning_rate": 1.0134576369468823e-05,
+ "loss": 0.0448,
+ "step": 18748
+ },
+ {
+ "epoch": 51.08719346049046,
+ "grad_norm": 2.6121158599853516,
+ "learning_rate": 1.0133693954227278e-05,
+ "loss": 0.1022,
+ "step": 18749
+ },
+ {
+ "epoch": 51.08991825613079,
+ "grad_norm": 2.3541975021362305,
+ "learning_rate": 1.0132811537944522e-05,
+ "loss": 0.0622,
+ "step": 18750
+ },
+ {
+ "epoch": 51.09264305177112,
+ "grad_norm": 1.8576161861419678,
+ "learning_rate": 1.0131929120627439e-05,
+ "loss": 0.0721,
+ "step": 18751
+ },
+ {
+ "epoch": 51.095367847411445,
+ "grad_norm": 2.4380533695220947,
+ "learning_rate": 1.0131046702282893e-05,
+ "loss": 0.0434,
+ "step": 18752
+ },
+ {
+ "epoch": 51.09809264305177,
+ "grad_norm": 2.974144458770752,
+ "learning_rate": 1.0130164282917764e-05,
+ "loss": 0.1626,
+ "step": 18753
+ },
+ {
+ "epoch": 51.1008174386921,
+ "grad_norm": 2.7503466606140137,
+ "learning_rate": 1.0129281862538918e-05,
+ "loss": 0.1421,
+ "step": 18754
+ },
+ {
+ "epoch": 51.10354223433242,
+ "grad_norm": 2.0543336868286133,
+ "learning_rate": 1.012839944115323e-05,
+ "loss": 0.058,
+ "step": 18755
+ },
+ {
+ "epoch": 51.10626702997275,
+ "grad_norm": 2.127936601638794,
+ "learning_rate": 1.0127517018767569e-05,
+ "loss": 0.0339,
+ "step": 18756
+ },
+ {
+ "epoch": 51.10899182561308,
+ "grad_norm": 2.1997835636138916,
+ "learning_rate": 1.0126634595388812e-05,
+ "loss": 0.0627,
+ "step": 18757
+ },
+ {
+ "epoch": 51.11171662125341,
+ "grad_norm": 2.7996881008148193,
+ "learning_rate": 1.0125752171023825e-05,
+ "loss": 0.114,
+ "step": 18758
+ },
+ {
+ "epoch": 51.11444141689373,
+ "grad_norm": 3.0545969009399414,
+ "learning_rate": 1.0124869745679486e-05,
+ "loss": 0.0721,
+ "step": 18759
+ },
+ {
+ "epoch": 51.11716621253406,
+ "grad_norm": 2.794571876525879,
+ "learning_rate": 1.0123987319362668e-05,
+ "loss": 0.088,
+ "step": 18760
+ },
+ {
+ "epoch": 51.119891008174385,
+ "grad_norm": 2.540412187576294,
+ "learning_rate": 1.0123104892080237e-05,
+ "loss": 0.0671,
+ "step": 18761
+ },
+ {
+ "epoch": 51.12261580381471,
+ "grad_norm": 2.765181303024292,
+ "learning_rate": 1.012222246383907e-05,
+ "loss": 0.0846,
+ "step": 18762
+ },
+ {
+ "epoch": 51.12534059945504,
+ "grad_norm": 2.1313793659210205,
+ "learning_rate": 1.0121340034646041e-05,
+ "loss": 0.075,
+ "step": 18763
+ },
+ {
+ "epoch": 51.12806539509537,
+ "grad_norm": 2.7620999813079834,
+ "learning_rate": 1.0120457604508019e-05,
+ "loss": 0.0831,
+ "step": 18764
+ },
+ {
+ "epoch": 51.130790190735695,
+ "grad_norm": 2.342684268951416,
+ "learning_rate": 1.0119575173431877e-05,
+ "loss": 0.0851,
+ "step": 18765
+ },
+ {
+ "epoch": 51.13351498637602,
+ "grad_norm": 3.2810885906219482,
+ "learning_rate": 1.0118692741424485e-05,
+ "loss": 0.0659,
+ "step": 18766
+ },
+ {
+ "epoch": 51.13623978201635,
+ "grad_norm": 3.2975850105285645,
+ "learning_rate": 1.0117810308492719e-05,
+ "loss": 0.2255,
+ "step": 18767
+ },
+ {
+ "epoch": 51.13896457765667,
+ "grad_norm": 1.9119350910186768,
+ "learning_rate": 1.011692787464345e-05,
+ "loss": 0.0316,
+ "step": 18768
+ },
+ {
+ "epoch": 51.141689373297005,
+ "grad_norm": 1.8824745416641235,
+ "learning_rate": 1.011604543988355e-05,
+ "loss": 0.0372,
+ "step": 18769
+ },
+ {
+ "epoch": 51.14441416893733,
+ "grad_norm": 2.3289952278137207,
+ "learning_rate": 1.0115163004219894e-05,
+ "loss": 0.0902,
+ "step": 18770
+ },
+ {
+ "epoch": 51.14713896457766,
+ "grad_norm": 1.99750816822052,
+ "learning_rate": 1.011428056765935e-05,
+ "loss": 0.0435,
+ "step": 18771
+ },
+ {
+ "epoch": 51.14986376021798,
+ "grad_norm": 2.4101455211639404,
+ "learning_rate": 1.0113398130208791e-05,
+ "loss": 0.079,
+ "step": 18772
+ },
+ {
+ "epoch": 51.15258855585831,
+ "grad_norm": 2.911222457885742,
+ "learning_rate": 1.0112515691875095e-05,
+ "loss": 0.0648,
+ "step": 18773
+ },
+ {
+ "epoch": 51.155313351498634,
+ "grad_norm": 2.1906516551971436,
+ "learning_rate": 1.0111633252665129e-05,
+ "loss": 0.2075,
+ "step": 18774
+ },
+ {
+ "epoch": 51.15803814713897,
+ "grad_norm": 2.851034164428711,
+ "learning_rate": 1.0110750812585767e-05,
+ "loss": 0.2213,
+ "step": 18775
+ },
+ {
+ "epoch": 51.16076294277929,
+ "grad_norm": 2.242652416229248,
+ "learning_rate": 1.010986837164388e-05,
+ "loss": 0.1269,
+ "step": 18776
+ },
+ {
+ "epoch": 51.16348773841962,
+ "grad_norm": 3.1508498191833496,
+ "learning_rate": 1.0108985929846342e-05,
+ "loss": 0.0992,
+ "step": 18777
+ },
+ {
+ "epoch": 51.166212534059945,
+ "grad_norm": 3.595994472503662,
+ "learning_rate": 1.0108103487200028e-05,
+ "loss": 0.0873,
+ "step": 18778
+ },
+ {
+ "epoch": 51.16893732970027,
+ "grad_norm": 2.914717674255371,
+ "learning_rate": 1.0107221043711804e-05,
+ "loss": 0.1123,
+ "step": 18779
+ },
+ {
+ "epoch": 51.171662125340596,
+ "grad_norm": 2.747021436691284,
+ "learning_rate": 1.0106338599388548e-05,
+ "loss": 0.21,
+ "step": 18780
+ },
+ {
+ "epoch": 51.17438692098093,
+ "grad_norm": 2.355086088180542,
+ "learning_rate": 1.010545615423713e-05,
+ "loss": 0.1887,
+ "step": 18781
+ },
+ {
+ "epoch": 51.177111716621255,
+ "grad_norm": 2.3854260444641113,
+ "learning_rate": 1.010457370826442e-05,
+ "loss": 0.0586,
+ "step": 18782
+ },
+ {
+ "epoch": 51.17983651226158,
+ "grad_norm": 1.873380184173584,
+ "learning_rate": 1.0103691261477297e-05,
+ "loss": 0.0498,
+ "step": 18783
+ },
+ {
+ "epoch": 51.182561307901906,
+ "grad_norm": 1.411076307296753,
+ "learning_rate": 1.0102808813882627e-05,
+ "loss": 0.032,
+ "step": 18784
+ },
+ {
+ "epoch": 51.18528610354223,
+ "grad_norm": 2.419597625732422,
+ "learning_rate": 1.0101926365487287e-05,
+ "loss": 0.0755,
+ "step": 18785
+ },
+ {
+ "epoch": 51.18801089918256,
+ "grad_norm": 4.179108142852783,
+ "learning_rate": 1.0101043916298148e-05,
+ "loss": 0.0562,
+ "step": 18786
+ },
+ {
+ "epoch": 51.19073569482289,
+ "grad_norm": 2.731205940246582,
+ "learning_rate": 1.0100161466322082e-05,
+ "loss": 0.0878,
+ "step": 18787
+ },
+ {
+ "epoch": 51.19346049046322,
+ "grad_norm": 2.3465683460235596,
+ "learning_rate": 1.0099279015565961e-05,
+ "loss": 0.0708,
+ "step": 18788
+ },
+ {
+ "epoch": 51.19618528610354,
+ "grad_norm": 3.0337975025177,
+ "learning_rate": 1.0098396564036658e-05,
+ "loss": 0.0617,
+ "step": 18789
+ },
+ {
+ "epoch": 51.19891008174387,
+ "grad_norm": 22.76420021057129,
+ "learning_rate": 1.0097514111741048e-05,
+ "loss": 0.071,
+ "step": 18790
+ },
+ {
+ "epoch": 51.201634877384194,
+ "grad_norm": 2.7614967823028564,
+ "learning_rate": 1.0096631658686e-05,
+ "loss": 0.1254,
+ "step": 18791
+ },
+ {
+ "epoch": 51.20435967302452,
+ "grad_norm": 3.2513623237609863,
+ "learning_rate": 1.009574920487839e-05,
+ "loss": 0.0885,
+ "step": 18792
+ },
+ {
+ "epoch": 51.20708446866485,
+ "grad_norm": 2.0811920166015625,
+ "learning_rate": 1.009486675032508e-05,
+ "loss": 0.0381,
+ "step": 18793
+ },
+ {
+ "epoch": 51.20980926430518,
+ "grad_norm": 2.1645023822784424,
+ "learning_rate": 1.009398429503296e-05,
+ "loss": 0.0472,
+ "step": 18794
+ },
+ {
+ "epoch": 51.212534059945504,
+ "grad_norm": 2.0359294414520264,
+ "learning_rate": 1.0093101839008888e-05,
+ "loss": 0.1243,
+ "step": 18795
+ },
+ {
+ "epoch": 51.21525885558583,
+ "grad_norm": 3.207397699356079,
+ "learning_rate": 1.0092219382259745e-05,
+ "loss": 0.1661,
+ "step": 18796
+ },
+ {
+ "epoch": 51.217983651226156,
+ "grad_norm": 2.002716302871704,
+ "learning_rate": 1.00913369247924e-05,
+ "loss": 0.0653,
+ "step": 18797
+ },
+ {
+ "epoch": 51.22070844686648,
+ "grad_norm": 2.5234241485595703,
+ "learning_rate": 1.0090454466613724e-05,
+ "loss": 0.0451,
+ "step": 18798
+ },
+ {
+ "epoch": 51.223433242506815,
+ "grad_norm": 5.223841190338135,
+ "learning_rate": 1.0089572007730593e-05,
+ "loss": 0.19,
+ "step": 18799
+ },
+ {
+ "epoch": 51.22615803814714,
+ "grad_norm": 3.160416603088379,
+ "learning_rate": 1.0088689548149878e-05,
+ "loss": 0.1277,
+ "step": 18800
+ },
+ {
+ "epoch": 51.228882833787466,
+ "grad_norm": 2.6127712726593018,
+ "learning_rate": 1.008780708787845e-05,
+ "loss": 0.0831,
+ "step": 18801
+ },
+ {
+ "epoch": 51.23160762942779,
+ "grad_norm": 2.7284741401672363,
+ "learning_rate": 1.0086924626923185e-05,
+ "loss": 0.1126,
+ "step": 18802
+ },
+ {
+ "epoch": 51.23433242506812,
+ "grad_norm": 2.770735740661621,
+ "learning_rate": 1.0086042165290952e-05,
+ "loss": 0.0939,
+ "step": 18803
+ },
+ {
+ "epoch": 51.237057220708444,
+ "grad_norm": 2.4538941383361816,
+ "learning_rate": 1.0085159702988627e-05,
+ "loss": 0.0973,
+ "step": 18804
+ },
+ {
+ "epoch": 51.23978201634878,
+ "grad_norm": 3.02398681640625,
+ "learning_rate": 1.0084277240023082e-05,
+ "loss": 0.0681,
+ "step": 18805
+ },
+ {
+ "epoch": 51.2425068119891,
+ "grad_norm": 2.3772804737091064,
+ "learning_rate": 1.0083394776401188e-05,
+ "loss": 0.0841,
+ "step": 18806
+ },
+ {
+ "epoch": 51.24523160762943,
+ "grad_norm": 2.4825572967529297,
+ "learning_rate": 1.0082512312129815e-05,
+ "loss": 0.1218,
+ "step": 18807
+ },
+ {
+ "epoch": 51.247956403269754,
+ "grad_norm": 3.6024744510650635,
+ "learning_rate": 1.0081629847215843e-05,
+ "loss": 0.0526,
+ "step": 18808
+ },
+ {
+ "epoch": 51.25068119891008,
+ "grad_norm": 3.1049587726593018,
+ "learning_rate": 1.0080747381666138e-05,
+ "loss": 0.0765,
+ "step": 18809
+ },
+ {
+ "epoch": 51.253405994550405,
+ "grad_norm": 2.4857711791992188,
+ "learning_rate": 1.0079864915487576e-05,
+ "loss": 0.1184,
+ "step": 18810
+ },
+ {
+ "epoch": 51.25613079019074,
+ "grad_norm": 1.8683897256851196,
+ "learning_rate": 1.0078982448687027e-05,
+ "loss": 0.0829,
+ "step": 18811
+ },
+ {
+ "epoch": 51.258855585831064,
+ "grad_norm": 3.1150007247924805,
+ "learning_rate": 1.0078099981271367e-05,
+ "loss": 0.2108,
+ "step": 18812
+ },
+ {
+ "epoch": 51.26158038147139,
+ "grad_norm": 2.427694320678711,
+ "learning_rate": 1.0077217513247465e-05,
+ "loss": 0.0398,
+ "step": 18813
+ },
+ {
+ "epoch": 51.264305177111716,
+ "grad_norm": 2.5022544860839844,
+ "learning_rate": 1.0076335044622198e-05,
+ "loss": 0.0479,
+ "step": 18814
+ },
+ {
+ "epoch": 51.26702997275204,
+ "grad_norm": 2.49086332321167,
+ "learning_rate": 1.0075452575402433e-05,
+ "loss": 0.0496,
+ "step": 18815
+ },
+ {
+ "epoch": 51.26975476839237,
+ "grad_norm": 2.732241153717041,
+ "learning_rate": 1.007457010559505e-05,
+ "loss": 0.0687,
+ "step": 18816
+ },
+ {
+ "epoch": 51.2724795640327,
+ "grad_norm": 2.383281707763672,
+ "learning_rate": 1.0073687635206915e-05,
+ "loss": 0.0926,
+ "step": 18817
+ },
+ {
+ "epoch": 51.275204359673026,
+ "grad_norm": 2.3660061359405518,
+ "learning_rate": 1.0072805164244903e-05,
+ "loss": 0.0548,
+ "step": 18818
+ },
+ {
+ "epoch": 51.27792915531335,
+ "grad_norm": 2.877800464630127,
+ "learning_rate": 1.0071922692715885e-05,
+ "loss": 0.0605,
+ "step": 18819
+ },
+ {
+ "epoch": 51.28065395095368,
+ "grad_norm": 2.3278017044067383,
+ "learning_rate": 1.0071040220626738e-05,
+ "loss": 0.0482,
+ "step": 18820
+ },
+ {
+ "epoch": 51.283378746594,
+ "grad_norm": 2.9219565391540527,
+ "learning_rate": 1.0070157747984331e-05,
+ "loss": 0.1165,
+ "step": 18821
+ },
+ {
+ "epoch": 51.28610354223433,
+ "grad_norm": 3.382221221923828,
+ "learning_rate": 1.006927527479554e-05,
+ "loss": 0.1374,
+ "step": 18822
+ },
+ {
+ "epoch": 51.28882833787466,
+ "grad_norm": 3.0333003997802734,
+ "learning_rate": 1.0068392801067232e-05,
+ "loss": 0.0651,
+ "step": 18823
+ },
+ {
+ "epoch": 51.29155313351499,
+ "grad_norm": 2.6935784816741943,
+ "learning_rate": 1.0067510326806284e-05,
+ "loss": 0.1435,
+ "step": 18824
+ },
+ {
+ "epoch": 51.294277929155314,
+ "grad_norm": 2.649193048477173,
+ "learning_rate": 1.0066627852019567e-05,
+ "loss": 0.1543,
+ "step": 18825
+ },
+ {
+ "epoch": 51.29700272479564,
+ "grad_norm": 2.760855197906494,
+ "learning_rate": 1.0065745376713958e-05,
+ "loss": 0.0976,
+ "step": 18826
+ },
+ {
+ "epoch": 51.299727520435965,
+ "grad_norm": 2.318749189376831,
+ "learning_rate": 1.0064862900896323e-05,
+ "loss": 0.0457,
+ "step": 18827
+ },
+ {
+ "epoch": 51.30245231607629,
+ "grad_norm": 2.1295392513275146,
+ "learning_rate": 1.0063980424573538e-05,
+ "loss": 0.0474,
+ "step": 18828
+ },
+ {
+ "epoch": 51.305177111716624,
+ "grad_norm": 2.6850154399871826,
+ "learning_rate": 1.0063097947752477e-05,
+ "loss": 0.0803,
+ "step": 18829
+ },
+ {
+ "epoch": 51.30790190735695,
+ "grad_norm": 2.477027654647827,
+ "learning_rate": 1.0062215470440011e-05,
+ "loss": 0.0498,
+ "step": 18830
+ },
+ {
+ "epoch": 51.310626702997276,
+ "grad_norm": 2.630479335784912,
+ "learning_rate": 1.006133299264301e-05,
+ "loss": 0.0669,
+ "step": 18831
+ },
+ {
+ "epoch": 51.3133514986376,
+ "grad_norm": 3.329608678817749,
+ "learning_rate": 1.0060450514368355e-05,
+ "loss": 0.0494,
+ "step": 18832
+ },
+ {
+ "epoch": 51.31607629427793,
+ "grad_norm": 1.7856462001800537,
+ "learning_rate": 1.005956803562291e-05,
+ "loss": 0.0932,
+ "step": 18833
+ },
+ {
+ "epoch": 51.31880108991825,
+ "grad_norm": 2.568901777267456,
+ "learning_rate": 1.005868555641355e-05,
+ "loss": 0.0622,
+ "step": 18834
+ },
+ {
+ "epoch": 51.321525885558586,
+ "grad_norm": 2.571798086166382,
+ "learning_rate": 1.005780307674715e-05,
+ "loss": 0.2771,
+ "step": 18835
+ },
+ {
+ "epoch": 51.32425068119891,
+ "grad_norm": 2.7121739387512207,
+ "learning_rate": 1.0056920596630582e-05,
+ "loss": 0.0619,
+ "step": 18836
+ },
+ {
+ "epoch": 51.32697547683924,
+ "grad_norm": 3.0109708309173584,
+ "learning_rate": 1.0056038116070716e-05,
+ "loss": 0.067,
+ "step": 18837
+ },
+ {
+ "epoch": 51.32970027247956,
+ "grad_norm": 2.9687893390655518,
+ "learning_rate": 1.005515563507443e-05,
+ "loss": 0.0507,
+ "step": 18838
+ },
+ {
+ "epoch": 51.33242506811989,
+ "grad_norm": 3.4513463973999023,
+ "learning_rate": 1.0054273153648591e-05,
+ "loss": 0.074,
+ "step": 18839
+ },
+ {
+ "epoch": 51.335149863760215,
+ "grad_norm": 2.5092666149139404,
+ "learning_rate": 1.005339067180008e-05,
+ "loss": 0.1017,
+ "step": 18840
+ },
+ {
+ "epoch": 51.33787465940055,
+ "grad_norm": 2.777754545211792,
+ "learning_rate": 1.0052508189535758e-05,
+ "loss": 0.0466,
+ "step": 18841
+ },
+ {
+ "epoch": 51.34059945504087,
+ "grad_norm": 2.2272322177886963,
+ "learning_rate": 1.0051625706862507e-05,
+ "loss": 0.0312,
+ "step": 18842
+ },
+ {
+ "epoch": 51.3433242506812,
+ "grad_norm": 2.1849958896636963,
+ "learning_rate": 1.0050743223787193e-05,
+ "loss": 0.1128,
+ "step": 18843
+ },
+ {
+ "epoch": 51.346049046321525,
+ "grad_norm": 2.3576831817626953,
+ "learning_rate": 1.0049860740316697e-05,
+ "loss": 0.0516,
+ "step": 18844
+ },
+ {
+ "epoch": 51.34877384196185,
+ "grad_norm": 2.0491600036621094,
+ "learning_rate": 1.0048978256457885e-05,
+ "loss": 0.0448,
+ "step": 18845
+ },
+ {
+ "epoch": 51.35149863760218,
+ "grad_norm": 2.479088306427002,
+ "learning_rate": 1.004809577221763e-05,
+ "loss": 0.121,
+ "step": 18846
+ },
+ {
+ "epoch": 51.35422343324251,
+ "grad_norm": 2.7174155712127686,
+ "learning_rate": 1.0047213287602809e-05,
+ "loss": 0.0439,
+ "step": 18847
+ },
+ {
+ "epoch": 51.356948228882835,
+ "grad_norm": 1.886997103691101,
+ "learning_rate": 1.0046330802620292e-05,
+ "loss": 0.0283,
+ "step": 18848
+ },
+ {
+ "epoch": 51.35967302452316,
+ "grad_norm": 2.4204885959625244,
+ "learning_rate": 1.0045448317276951e-05,
+ "loss": 0.0519,
+ "step": 18849
+ },
+ {
+ "epoch": 51.36239782016349,
+ "grad_norm": 2.1639647483825684,
+ "learning_rate": 1.0044565831579664e-05,
+ "loss": 0.0454,
+ "step": 18850
+ },
+ {
+ "epoch": 51.36512261580381,
+ "grad_norm": 1.7184144258499146,
+ "learning_rate": 1.0043683345535295e-05,
+ "loss": 0.0281,
+ "step": 18851
+ },
+ {
+ "epoch": 51.36784741144414,
+ "grad_norm": 2.2807133197784424,
+ "learning_rate": 1.0042800859150726e-05,
+ "loss": 0.0635,
+ "step": 18852
+ },
+ {
+ "epoch": 51.37057220708447,
+ "grad_norm": 3.025064706802368,
+ "learning_rate": 1.0041918372432822e-05,
+ "loss": 0.12,
+ "step": 18853
+ },
+ {
+ "epoch": 51.3732970027248,
+ "grad_norm": 4.936983585357666,
+ "learning_rate": 1.004103588538846e-05,
+ "loss": 0.1003,
+ "step": 18854
+ },
+ {
+ "epoch": 51.37602179836512,
+ "grad_norm": 2.5984113216400146,
+ "learning_rate": 1.0040153398024513e-05,
+ "loss": 0.0779,
+ "step": 18855
+ },
+ {
+ "epoch": 51.37874659400545,
+ "grad_norm": 2.0882132053375244,
+ "learning_rate": 1.0039270910347849e-05,
+ "loss": 0.1368,
+ "step": 18856
+ },
+ {
+ "epoch": 51.381471389645775,
+ "grad_norm": 4.072432994842529,
+ "learning_rate": 1.0038388422365348e-05,
+ "loss": 0.0393,
+ "step": 18857
+ },
+ {
+ "epoch": 51.3841961852861,
+ "grad_norm": 2.92830491065979,
+ "learning_rate": 1.0037505934083877e-05,
+ "loss": 0.0475,
+ "step": 18858
+ },
+ {
+ "epoch": 51.38692098092643,
+ "grad_norm": 2.048457384109497,
+ "learning_rate": 1.0036623445510312e-05,
+ "loss": 0.0752,
+ "step": 18859
+ },
+ {
+ "epoch": 51.38964577656676,
+ "grad_norm": 2.551622152328491,
+ "learning_rate": 1.0035740956651527e-05,
+ "loss": 0.1237,
+ "step": 18860
+ },
+ {
+ "epoch": 51.392370572207085,
+ "grad_norm": 2.4319188594818115,
+ "learning_rate": 1.0034858467514388e-05,
+ "loss": 0.0531,
+ "step": 18861
+ },
+ {
+ "epoch": 51.39509536784741,
+ "grad_norm": 3.7192184925079346,
+ "learning_rate": 1.0033975978105774e-05,
+ "loss": 0.0727,
+ "step": 18862
+ },
+ {
+ "epoch": 51.39782016348774,
+ "grad_norm": 2.1317036151885986,
+ "learning_rate": 1.003309348843256e-05,
+ "loss": 0.0485,
+ "step": 18863
+ },
+ {
+ "epoch": 51.40054495912806,
+ "grad_norm": 2.4394521713256836,
+ "learning_rate": 1.0032210998501611e-05,
+ "loss": 0.0616,
+ "step": 18864
+ },
+ {
+ "epoch": 51.403269754768395,
+ "grad_norm": 2.991551160812378,
+ "learning_rate": 1.0031328508319805e-05,
+ "loss": 0.076,
+ "step": 18865
+ },
+ {
+ "epoch": 51.40599455040872,
+ "grad_norm": 1.6177767515182495,
+ "learning_rate": 1.0030446017894014e-05,
+ "loss": 0.0288,
+ "step": 18866
+ },
+ {
+ "epoch": 51.40871934604905,
+ "grad_norm": 2.543459892272949,
+ "learning_rate": 1.002956352723111e-05,
+ "loss": 0.1446,
+ "step": 18867
+ },
+ {
+ "epoch": 51.41144414168937,
+ "grad_norm": 2.192700147628784,
+ "learning_rate": 1.0028681036337963e-05,
+ "loss": 0.0543,
+ "step": 18868
+ },
+ {
+ "epoch": 51.4141689373297,
+ "grad_norm": 2.591515302658081,
+ "learning_rate": 1.0027798545221453e-05,
+ "loss": 0.1123,
+ "step": 18869
+ },
+ {
+ "epoch": 51.416893732970024,
+ "grad_norm": 1.9682934284210205,
+ "learning_rate": 1.0026916053888449e-05,
+ "loss": 0.0433,
+ "step": 18870
+ },
+ {
+ "epoch": 51.41961852861036,
+ "grad_norm": 2.5184996128082275,
+ "learning_rate": 1.0026033562345821e-05,
+ "loss": 0.0735,
+ "step": 18871
+ },
+ {
+ "epoch": 51.42234332425068,
+ "grad_norm": 2.4503860473632812,
+ "learning_rate": 1.0025151070600445e-05,
+ "loss": 0.0731,
+ "step": 18872
+ },
+ {
+ "epoch": 51.42506811989101,
+ "grad_norm": 2.3597331047058105,
+ "learning_rate": 1.0024268578659195e-05,
+ "loss": 0.043,
+ "step": 18873
+ },
+ {
+ "epoch": 51.427792915531334,
+ "grad_norm": 2.7530791759490967,
+ "learning_rate": 1.0023386086528942e-05,
+ "loss": 0.0858,
+ "step": 18874
+ },
+ {
+ "epoch": 51.43051771117166,
+ "grad_norm": 2.579472303390503,
+ "learning_rate": 1.0022503594216558e-05,
+ "loss": 0.0356,
+ "step": 18875
+ },
+ {
+ "epoch": 51.433242506811986,
+ "grad_norm": 2.56326961517334,
+ "learning_rate": 1.0021621101728916e-05,
+ "loss": 0.0944,
+ "step": 18876
+ },
+ {
+ "epoch": 51.43596730245232,
+ "grad_norm": 2.2264485359191895,
+ "learning_rate": 1.002073860907289e-05,
+ "loss": 0.0727,
+ "step": 18877
+ },
+ {
+ "epoch": 51.438692098092645,
+ "grad_norm": 2.274899959564209,
+ "learning_rate": 1.0019856116255353e-05,
+ "loss": 0.087,
+ "step": 18878
+ },
+ {
+ "epoch": 51.44141689373297,
+ "grad_norm": 2.1309814453125,
+ "learning_rate": 1.0018973623283177e-05,
+ "loss": 0.0652,
+ "step": 18879
+ },
+ {
+ "epoch": 51.444141689373296,
+ "grad_norm": 3.5111050605773926,
+ "learning_rate": 1.0018091130163235e-05,
+ "loss": 0.0986,
+ "step": 18880
+ },
+ {
+ "epoch": 51.44686648501362,
+ "grad_norm": 1.98800790309906,
+ "learning_rate": 1.0017208636902401e-05,
+ "loss": 0.0285,
+ "step": 18881
+ },
+ {
+ "epoch": 51.44959128065395,
+ "grad_norm": 3.176222562789917,
+ "learning_rate": 1.0016326143507544e-05,
+ "loss": 0.1758,
+ "step": 18882
+ },
+ {
+ "epoch": 51.45231607629428,
+ "grad_norm": 2.4646100997924805,
+ "learning_rate": 1.0015443649985542e-05,
+ "loss": 0.0536,
+ "step": 18883
+ },
+ {
+ "epoch": 51.45504087193461,
+ "grad_norm": 2.7669591903686523,
+ "learning_rate": 1.0014561156343263e-05,
+ "loss": 0.0363,
+ "step": 18884
+ },
+ {
+ "epoch": 51.45776566757493,
+ "grad_norm": 2.811537981033325,
+ "learning_rate": 1.0013678662587584e-05,
+ "loss": 0.0724,
+ "step": 18885
+ },
+ {
+ "epoch": 51.46049046321526,
+ "grad_norm": 2.3504323959350586,
+ "learning_rate": 1.0012796168725376e-05,
+ "loss": 0.1484,
+ "step": 18886
+ },
+ {
+ "epoch": 51.463215258855584,
+ "grad_norm": 2.399199962615967,
+ "learning_rate": 1.0011913674763514e-05,
+ "loss": 0.0704,
+ "step": 18887
+ },
+ {
+ "epoch": 51.46594005449591,
+ "grad_norm": 2.4450666904449463,
+ "learning_rate": 1.0011031180708861e-05,
+ "loss": 0.0394,
+ "step": 18888
+ },
+ {
+ "epoch": 51.46866485013624,
+ "grad_norm": 1.9324781894683838,
+ "learning_rate": 1.0010148686568304e-05,
+ "loss": 0.039,
+ "step": 18889
+ },
+ {
+ "epoch": 51.47138964577657,
+ "grad_norm": 3.258615255355835,
+ "learning_rate": 1.0009266192348708e-05,
+ "loss": 0.1143,
+ "step": 18890
+ },
+ {
+ "epoch": 51.474114441416894,
+ "grad_norm": 4.297921657562256,
+ "learning_rate": 1.0008383698056949e-05,
+ "loss": 0.112,
+ "step": 18891
+ },
+ {
+ "epoch": 51.47683923705722,
+ "grad_norm": 2.3299076557159424,
+ "learning_rate": 1.0007501203699894e-05,
+ "loss": 0.0496,
+ "step": 18892
+ },
+ {
+ "epoch": 51.479564032697546,
+ "grad_norm": 2.5536344051361084,
+ "learning_rate": 1.0006618709284424e-05,
+ "loss": 0.1136,
+ "step": 18893
+ },
+ {
+ "epoch": 51.48228882833787,
+ "grad_norm": 2.792233467102051,
+ "learning_rate": 1.0005736214817405e-05,
+ "loss": 0.0521,
+ "step": 18894
+ },
+ {
+ "epoch": 51.485013623978205,
+ "grad_norm": 3.284877061843872,
+ "learning_rate": 1.0004853720305717e-05,
+ "loss": 0.099,
+ "step": 18895
+ },
+ {
+ "epoch": 51.48773841961853,
+ "grad_norm": 2.7318975925445557,
+ "learning_rate": 1.0003971225756222e-05,
+ "loss": 0.0599,
+ "step": 18896
+ },
+ {
+ "epoch": 51.490463215258856,
+ "grad_norm": 2.7449262142181396,
+ "learning_rate": 1.00030887311758e-05,
+ "loss": 0.0444,
+ "step": 18897
+ },
+ {
+ "epoch": 51.49318801089918,
+ "grad_norm": 2.1043663024902344,
+ "learning_rate": 1.0002206236571325e-05,
+ "loss": 0.0609,
+ "step": 18898
+ },
+ {
+ "epoch": 51.49591280653951,
+ "grad_norm": 2.8072028160095215,
+ "learning_rate": 1.0001323741949669e-05,
+ "loss": 0.0525,
+ "step": 18899
+ },
+ {
+ "epoch": 51.49863760217983,
+ "grad_norm": 1.908186912536621,
+ "learning_rate": 1.0000441247317702e-05,
+ "loss": 0.0457,
+ "step": 18900
+ },
+ {
+ "epoch": 51.50136239782017,
+ "grad_norm": 2.6478326320648193,
+ "learning_rate": 9.9995587526823e-06,
+ "loss": 0.0397,
+ "step": 18901
+ },
+ {
+ "epoch": 51.50408719346049,
+ "grad_norm": 1.6158487796783447,
+ "learning_rate": 9.998676258050331e-06,
+ "loss": 0.0265,
+ "step": 18902
+ },
+ {
+ "epoch": 51.50681198910082,
+ "grad_norm": 2.1636714935302734,
+ "learning_rate": 9.997793763428676e-06,
+ "loss": 0.1989,
+ "step": 18903
+ },
+ {
+ "epoch": 51.509536784741144,
+ "grad_norm": 3.217568874359131,
+ "learning_rate": 9.9969112688242e-06,
+ "loss": 0.102,
+ "step": 18904
+ },
+ {
+ "epoch": 51.51226158038147,
+ "grad_norm": 2.183145523071289,
+ "learning_rate": 9.996028774243781e-06,
+ "loss": 0.0583,
+ "step": 18905
+ },
+ {
+ "epoch": 51.514986376021795,
+ "grad_norm": 3.8883323669433594,
+ "learning_rate": 9.995146279694288e-06,
+ "loss": 0.1637,
+ "step": 18906
+ },
+ {
+ "epoch": 51.51771117166213,
+ "grad_norm": 2.4734079837799072,
+ "learning_rate": 9.994263785182597e-06,
+ "loss": 0.0486,
+ "step": 18907
+ },
+ {
+ "epoch": 51.520435967302454,
+ "grad_norm": 2.0062716007232666,
+ "learning_rate": 9.993381290715578e-06,
+ "loss": 0.0517,
+ "step": 18908
+ },
+ {
+ "epoch": 51.52316076294278,
+ "grad_norm": 9.434966087341309,
+ "learning_rate": 9.992498796300108e-06,
+ "loss": 0.039,
+ "step": 18909
+ },
+ {
+ "epoch": 51.525885558583106,
+ "grad_norm": 2.0676214694976807,
+ "learning_rate": 9.991616301943054e-06,
+ "loss": 0.1534,
+ "step": 18910
+ },
+ {
+ "epoch": 51.52861035422343,
+ "grad_norm": 2.3137779235839844,
+ "learning_rate": 9.990733807651295e-06,
+ "loss": 0.0432,
+ "step": 18911
+ },
+ {
+ "epoch": 51.53133514986376,
+ "grad_norm": 2.51523494720459,
+ "learning_rate": 9.989851313431698e-06,
+ "loss": 0.0871,
+ "step": 18912
+ },
+ {
+ "epoch": 51.53405994550409,
+ "grad_norm": 2.5724451541900635,
+ "learning_rate": 9.98896881929114e-06,
+ "loss": 0.1209,
+ "step": 18913
+ },
+ {
+ "epoch": 51.536784741144416,
+ "grad_norm": 3.2845444679260254,
+ "learning_rate": 9.988086325236493e-06,
+ "loss": 0.0946,
+ "step": 18914
+ },
+ {
+ "epoch": 51.53950953678474,
+ "grad_norm": 2.1750731468200684,
+ "learning_rate": 9.987203831274629e-06,
+ "loss": 0.1611,
+ "step": 18915
+ },
+ {
+ "epoch": 51.54223433242507,
+ "grad_norm": 2.355794668197632,
+ "learning_rate": 9.98632133741242e-06,
+ "loss": 0.1582,
+ "step": 18916
+ },
+ {
+ "epoch": 51.54495912806539,
+ "grad_norm": 7.715234279632568,
+ "learning_rate": 9.985438843656742e-06,
+ "loss": 0.0505,
+ "step": 18917
+ },
+ {
+ "epoch": 51.54768392370572,
+ "grad_norm": 2.8258655071258545,
+ "learning_rate": 9.98455635001446e-06,
+ "loss": 0.1479,
+ "step": 18918
+ },
+ {
+ "epoch": 51.55040871934605,
+ "grad_norm": 2.687775135040283,
+ "learning_rate": 9.983673856492457e-06,
+ "loss": 0.1583,
+ "step": 18919
+ },
+ {
+ "epoch": 51.55313351498638,
+ "grad_norm": 2.228022575378418,
+ "learning_rate": 9.982791363097602e-06,
+ "loss": 0.1242,
+ "step": 18920
+ },
+ {
+ "epoch": 51.555858310626704,
+ "grad_norm": 2.194807291030884,
+ "learning_rate": 9.981908869836767e-06,
+ "loss": 0.0978,
+ "step": 18921
+ },
+ {
+ "epoch": 51.55858310626703,
+ "grad_norm": 3.3436546325683594,
+ "learning_rate": 9.981026376716823e-06,
+ "loss": 0.1729,
+ "step": 18922
+ },
+ {
+ "epoch": 51.561307901907355,
+ "grad_norm": 3.097306728363037,
+ "learning_rate": 9.98014388374465e-06,
+ "loss": 0.136,
+ "step": 18923
+ },
+ {
+ "epoch": 51.56403269754768,
+ "grad_norm": 3.945737361907959,
+ "learning_rate": 9.97926139092711e-06,
+ "loss": 0.1133,
+ "step": 18924
+ },
+ {
+ "epoch": 51.566757493188014,
+ "grad_norm": 3.4219563007354736,
+ "learning_rate": 9.978378898271087e-06,
+ "loss": 0.0596,
+ "step": 18925
+ },
+ {
+ "epoch": 51.56948228882834,
+ "grad_norm": 1.9949588775634766,
+ "learning_rate": 9.977496405783444e-06,
+ "loss": 0.0352,
+ "step": 18926
+ },
+ {
+ "epoch": 51.572207084468666,
+ "grad_norm": 2.254958391189575,
+ "learning_rate": 9.976613913471063e-06,
+ "loss": 0.1469,
+ "step": 18927
+ },
+ {
+ "epoch": 51.57493188010899,
+ "grad_norm": 3.60614275932312,
+ "learning_rate": 9.975731421340807e-06,
+ "loss": 0.0565,
+ "step": 18928
+ },
+ {
+ "epoch": 51.57765667574932,
+ "grad_norm": 4.374655246734619,
+ "learning_rate": 9.974848929399557e-06,
+ "loss": 0.1992,
+ "step": 18929
+ },
+ {
+ "epoch": 51.58038147138964,
+ "grad_norm": 2.1518146991729736,
+ "learning_rate": 9.973966437654182e-06,
+ "loss": 0.041,
+ "step": 18930
+ },
+ {
+ "epoch": 51.583106267029976,
+ "grad_norm": 3.044917345046997,
+ "learning_rate": 9.973083946111556e-06,
+ "loss": 0.1476,
+ "step": 18931
+ },
+ {
+ "epoch": 51.5858310626703,
+ "grad_norm": 2.0564780235290527,
+ "learning_rate": 9.97220145477855e-06,
+ "loss": 0.0434,
+ "step": 18932
+ },
+ {
+ "epoch": 51.58855585831063,
+ "grad_norm": 2.3076012134552,
+ "learning_rate": 9.971318963662042e-06,
+ "loss": 0.0521,
+ "step": 18933
+ },
+ {
+ "epoch": 51.59128065395095,
+ "grad_norm": 2.171574592590332,
+ "learning_rate": 9.970436472768895e-06,
+ "loss": 0.0436,
+ "step": 18934
+ },
+ {
+ "epoch": 51.59400544959128,
+ "grad_norm": 2.780240058898926,
+ "learning_rate": 9.969553982105988e-06,
+ "loss": 0.1393,
+ "step": 18935
+ },
+ {
+ "epoch": 51.596730245231605,
+ "grad_norm": 1.6483296155929565,
+ "learning_rate": 9.968671491680197e-06,
+ "loss": 0.0285,
+ "step": 18936
+ },
+ {
+ "epoch": 51.59945504087194,
+ "grad_norm": 3.0889546871185303,
+ "learning_rate": 9.96778900149839e-06,
+ "loss": 0.0568,
+ "step": 18937
+ },
+ {
+ "epoch": 51.60217983651226,
+ "grad_norm": 2.401411771774292,
+ "learning_rate": 9.966906511567443e-06,
+ "loss": 0.1178,
+ "step": 18938
+ },
+ {
+ "epoch": 51.60490463215259,
+ "grad_norm": 2.2151811122894287,
+ "learning_rate": 9.966024021894226e-06,
+ "loss": 0.0474,
+ "step": 18939
+ },
+ {
+ "epoch": 51.607629427792915,
+ "grad_norm": 2.9434993267059326,
+ "learning_rate": 9.965141532485613e-06,
+ "loss": 0.1064,
+ "step": 18940
+ },
+ {
+ "epoch": 51.61035422343324,
+ "grad_norm": 2.321208953857422,
+ "learning_rate": 9.964259043348476e-06,
+ "loss": 0.0569,
+ "step": 18941
+ },
+ {
+ "epoch": 51.61307901907357,
+ "grad_norm": 1.7575876712799072,
+ "learning_rate": 9.963376554489691e-06,
+ "loss": 0.0301,
+ "step": 18942
+ },
+ {
+ "epoch": 51.6158038147139,
+ "grad_norm": 2.364811897277832,
+ "learning_rate": 9.962494065916124e-06,
+ "loss": 0.0537,
+ "step": 18943
+ },
+ {
+ "epoch": 51.618528610354225,
+ "grad_norm": 2.6547110080718994,
+ "learning_rate": 9.961611577634655e-06,
+ "loss": 0.0395,
+ "step": 18944
+ },
+ {
+ "epoch": 51.62125340599455,
+ "grad_norm": 2.1212782859802246,
+ "learning_rate": 9.960729089652153e-06,
+ "loss": 0.0604,
+ "step": 18945
+ },
+ {
+ "epoch": 51.62397820163488,
+ "grad_norm": 2.830687999725342,
+ "learning_rate": 9.95984660197549e-06,
+ "loss": 0.0493,
+ "step": 18946
+ },
+ {
+ "epoch": 51.6267029972752,
+ "grad_norm": 2.374377489089966,
+ "learning_rate": 9.958964114611544e-06,
+ "loss": 0.239,
+ "step": 18947
+ },
+ {
+ "epoch": 51.62942779291553,
+ "grad_norm": 2.3894693851470947,
+ "learning_rate": 9.958081627567182e-06,
+ "loss": 0.027,
+ "step": 18948
+ },
+ {
+ "epoch": 51.63215258855586,
+ "grad_norm": 1.849291205406189,
+ "learning_rate": 9.95719914084928e-06,
+ "loss": 0.0293,
+ "step": 18949
+ },
+ {
+ "epoch": 51.63487738419619,
+ "grad_norm": 1.7331037521362305,
+ "learning_rate": 9.956316654464706e-06,
+ "loss": 0.028,
+ "step": 18950
+ },
+ {
+ "epoch": 51.63760217983651,
+ "grad_norm": 2.431533098220825,
+ "learning_rate": 9.95543416842034e-06,
+ "loss": 0.1472,
+ "step": 18951
+ },
+ {
+ "epoch": 51.64032697547684,
+ "grad_norm": 3.522338390350342,
+ "learning_rate": 9.95455168272305e-06,
+ "loss": 0.1858,
+ "step": 18952
+ },
+ {
+ "epoch": 51.643051771117165,
+ "grad_norm": 2.0495755672454834,
+ "learning_rate": 9.953669197379708e-06,
+ "loss": 0.0617,
+ "step": 18953
+ },
+ {
+ "epoch": 51.64577656675749,
+ "grad_norm": 3.9087412357330322,
+ "learning_rate": 9.952786712397191e-06,
+ "loss": 0.076,
+ "step": 18954
+ },
+ {
+ "epoch": 51.64850136239782,
+ "grad_norm": 2.6392276287078857,
+ "learning_rate": 9.95190422778237e-06,
+ "loss": 0.0401,
+ "step": 18955
+ },
+ {
+ "epoch": 51.65122615803815,
+ "grad_norm": 2.298158884048462,
+ "learning_rate": 9.951021743542119e-06,
+ "loss": 0.0728,
+ "step": 18956
+ },
+ {
+ "epoch": 51.653950953678475,
+ "grad_norm": 2.7786262035369873,
+ "learning_rate": 9.950139259683305e-06,
+ "loss": 0.1616,
+ "step": 18957
+ },
+ {
+ "epoch": 51.6566757493188,
+ "grad_norm": 2.4286837577819824,
+ "learning_rate": 9.949256776212808e-06,
+ "loss": 0.0793,
+ "step": 18958
+ },
+ {
+ "epoch": 51.65940054495913,
+ "grad_norm": 2.3694143295288086,
+ "learning_rate": 9.948374293137496e-06,
+ "loss": 0.0717,
+ "step": 18959
+ },
+ {
+ "epoch": 51.66212534059945,
+ "grad_norm": 2.4213650226593018,
+ "learning_rate": 9.947491810464246e-06,
+ "loss": 0.1194,
+ "step": 18960
+ },
+ {
+ "epoch": 51.664850136239785,
+ "grad_norm": 2.2838876247406006,
+ "learning_rate": 9.946609328199926e-06,
+ "loss": 0.0365,
+ "step": 18961
+ },
+ {
+ "epoch": 51.66757493188011,
+ "grad_norm": 2.3854079246520996,
+ "learning_rate": 9.94572684635141e-06,
+ "loss": 0.0662,
+ "step": 18962
+ },
+ {
+ "epoch": 51.67029972752044,
+ "grad_norm": 2.170199394226074,
+ "learning_rate": 9.944844364925574e-06,
+ "loss": 0.2034,
+ "step": 18963
+ },
+ {
+ "epoch": 51.67302452316076,
+ "grad_norm": 3.1565380096435547,
+ "learning_rate": 9.943961883929287e-06,
+ "loss": 0.1738,
+ "step": 18964
+ },
+ {
+ "epoch": 51.67574931880109,
+ "grad_norm": 4.035930633544922,
+ "learning_rate": 9.943079403369423e-06,
+ "loss": 0.0916,
+ "step": 18965
+ },
+ {
+ "epoch": 51.678474114441414,
+ "grad_norm": 2.0342631340026855,
+ "learning_rate": 9.942196923252855e-06,
+ "loss": 0.0486,
+ "step": 18966
+ },
+ {
+ "epoch": 51.68119891008175,
+ "grad_norm": 3.917903423309326,
+ "learning_rate": 9.941314443586454e-06,
+ "loss": 0.0468,
+ "step": 18967
+ },
+ {
+ "epoch": 51.68392370572207,
+ "grad_norm": 3.9632959365844727,
+ "learning_rate": 9.940431964377094e-06,
+ "loss": 0.0909,
+ "step": 18968
+ },
+ {
+ "epoch": 51.6866485013624,
+ "grad_norm": 1.7977666854858398,
+ "learning_rate": 9.93954948563165e-06,
+ "loss": 0.0866,
+ "step": 18969
+ },
+ {
+ "epoch": 51.689373297002724,
+ "grad_norm": 3.308637857437134,
+ "learning_rate": 9.938667007356991e-06,
+ "loss": 0.1488,
+ "step": 18970
+ },
+ {
+ "epoch": 51.69209809264305,
+ "grad_norm": 2.6114485263824463,
+ "learning_rate": 9.93778452955999e-06,
+ "loss": 0.0532,
+ "step": 18971
+ },
+ {
+ "epoch": 51.694822888283376,
+ "grad_norm": 2.066448211669922,
+ "learning_rate": 9.936902052247525e-06,
+ "loss": 0.0386,
+ "step": 18972
+ },
+ {
+ "epoch": 51.69754768392371,
+ "grad_norm": 2.3119046688079834,
+ "learning_rate": 9.936019575426462e-06,
+ "loss": 0.0365,
+ "step": 18973
+ },
+ {
+ "epoch": 51.700272479564035,
+ "grad_norm": 2.725661039352417,
+ "learning_rate": 9.935137099103678e-06,
+ "loss": 0.0861,
+ "step": 18974
+ },
+ {
+ "epoch": 51.70299727520436,
+ "grad_norm": 2.453627586364746,
+ "learning_rate": 9.934254623286043e-06,
+ "loss": 0.1163,
+ "step": 18975
+ },
+ {
+ "epoch": 51.705722070844686,
+ "grad_norm": 2.9297287464141846,
+ "learning_rate": 9.933372147980434e-06,
+ "loss": 0.189,
+ "step": 18976
+ },
+ {
+ "epoch": 51.70844686648501,
+ "grad_norm": 2.5474743843078613,
+ "learning_rate": 9.932489673193718e-06,
+ "loss": 0.067,
+ "step": 18977
+ },
+ {
+ "epoch": 51.71117166212534,
+ "grad_norm": 5.532418727874756,
+ "learning_rate": 9.931607198932771e-06,
+ "loss": 0.0766,
+ "step": 18978
+ },
+ {
+ "epoch": 51.71389645776567,
+ "grad_norm": 3.3489646911621094,
+ "learning_rate": 9.930724725204463e-06,
+ "loss": 0.066,
+ "step": 18979
+ },
+ {
+ "epoch": 51.716621253406,
+ "grad_norm": 1.6992483139038086,
+ "learning_rate": 9.929842252015672e-06,
+ "loss": 0.0386,
+ "step": 18980
+ },
+ {
+ "epoch": 51.71934604904632,
+ "grad_norm": 2.8625848293304443,
+ "learning_rate": 9.928959779373265e-06,
+ "loss": 0.0431,
+ "step": 18981
+ },
+ {
+ "epoch": 51.72207084468665,
+ "grad_norm": 1.798404574394226,
+ "learning_rate": 9.928077307284118e-06,
+ "loss": 0.1787,
+ "step": 18982
+ },
+ {
+ "epoch": 51.724795640326974,
+ "grad_norm": 3.031081438064575,
+ "learning_rate": 9.9271948357551e-06,
+ "loss": 0.0575,
+ "step": 18983
+ },
+ {
+ "epoch": 51.7275204359673,
+ "grad_norm": 2.6872572898864746,
+ "learning_rate": 9.92631236479309e-06,
+ "loss": 0.2038,
+ "step": 18984
+ },
+ {
+ "epoch": 51.73024523160763,
+ "grad_norm": 3.036290168762207,
+ "learning_rate": 9.925429894404955e-06,
+ "loss": 0.1212,
+ "step": 18985
+ },
+ {
+ "epoch": 51.73297002724796,
+ "grad_norm": 2.0422191619873047,
+ "learning_rate": 9.92454742459757e-06,
+ "loss": 0.0692,
+ "step": 18986
+ },
+ {
+ "epoch": 51.735694822888284,
+ "grad_norm": 2.0378830432891846,
+ "learning_rate": 9.923664955377802e-06,
+ "loss": 0.0478,
+ "step": 18987
+ },
+ {
+ "epoch": 51.73841961852861,
+ "grad_norm": 2.9656219482421875,
+ "learning_rate": 9.922782486752537e-06,
+ "loss": 0.0775,
+ "step": 18988
+ },
+ {
+ "epoch": 51.741144414168936,
+ "grad_norm": 3.1947617530822754,
+ "learning_rate": 9.921900018728634e-06,
+ "loss": 0.1384,
+ "step": 18989
+ },
+ {
+ "epoch": 51.74386920980926,
+ "grad_norm": 2.7121341228485107,
+ "learning_rate": 9.921017551312975e-06,
+ "loss": 0.1224,
+ "step": 18990
+ },
+ {
+ "epoch": 51.746594005449595,
+ "grad_norm": 2.4603726863861084,
+ "learning_rate": 9.920135084512425e-06,
+ "loss": 0.0716,
+ "step": 18991
+ },
+ {
+ "epoch": 51.74931880108992,
+ "grad_norm": 2.387716770172119,
+ "learning_rate": 9.919252618333864e-06,
+ "loss": 0.065,
+ "step": 18992
+ },
+ {
+ "epoch": 51.752043596730246,
+ "grad_norm": 3.5422720909118652,
+ "learning_rate": 9.918370152784158e-06,
+ "loss": 0.0525,
+ "step": 18993
+ },
+ {
+ "epoch": 51.75476839237057,
+ "grad_norm": 2.5449180603027344,
+ "learning_rate": 9.917487687870187e-06,
+ "loss": 0.1394,
+ "step": 18994
+ },
+ {
+ "epoch": 51.7574931880109,
+ "grad_norm": 2.7209982872009277,
+ "learning_rate": 9.916605223598815e-06,
+ "loss": 0.0661,
+ "step": 18995
+ },
+ {
+ "epoch": 51.76021798365122,
+ "grad_norm": 2.1206276416778564,
+ "learning_rate": 9.915722759976923e-06,
+ "loss": 0.0446,
+ "step": 18996
+ },
+ {
+ "epoch": 51.762942779291556,
+ "grad_norm": 2.275244951248169,
+ "learning_rate": 9.914840297011376e-06,
+ "loss": 0.0515,
+ "step": 18997
+ },
+ {
+ "epoch": 51.76566757493188,
+ "grad_norm": 2.1160268783569336,
+ "learning_rate": 9.913957834709051e-06,
+ "loss": 0.1462,
+ "step": 18998
+ },
+ {
+ "epoch": 51.76839237057221,
+ "grad_norm": 3.08219313621521,
+ "learning_rate": 9.913075373076819e-06,
+ "loss": 0.0601,
+ "step": 18999
+ },
+ {
+ "epoch": 51.771117166212534,
+ "grad_norm": 2.892716646194458,
+ "learning_rate": 9.912192912121554e-06,
+ "loss": 0.0918,
+ "step": 19000
+ },
+ {
+ "epoch": 51.77384196185286,
+ "grad_norm": 2.291033983230591,
+ "learning_rate": 9.911310451850127e-06,
+ "loss": 0.0424,
+ "step": 19001
+ },
+ {
+ "epoch": 51.776566757493185,
+ "grad_norm": 2.8563029766082764,
+ "learning_rate": 9.910427992269414e-06,
+ "loss": 0.2022,
+ "step": 19002
+ },
+ {
+ "epoch": 51.77929155313352,
+ "grad_norm": 3.4059736728668213,
+ "learning_rate": 9.909545533386281e-06,
+ "loss": 0.2169,
+ "step": 19003
+ },
+ {
+ "epoch": 51.782016348773844,
+ "grad_norm": 2.4023067951202393,
+ "learning_rate": 9.908663075207602e-06,
+ "loss": 0.0873,
+ "step": 19004
+ },
+ {
+ "epoch": 51.78474114441417,
+ "grad_norm": 2.6413583755493164,
+ "learning_rate": 9.907780617740258e-06,
+ "loss": 0.097,
+ "step": 19005
+ },
+ {
+ "epoch": 51.787465940054496,
+ "grad_norm": 2.203923225402832,
+ "learning_rate": 9.906898160991113e-06,
+ "loss": 0.0517,
+ "step": 19006
+ },
+ {
+ "epoch": 51.79019073569482,
+ "grad_norm": 2.666367530822754,
+ "learning_rate": 9.906015704967044e-06,
+ "loss": 0.049,
+ "step": 19007
+ },
+ {
+ "epoch": 51.79291553133515,
+ "grad_norm": 2.8069639205932617,
+ "learning_rate": 9.90513324967492e-06,
+ "loss": 0.1072,
+ "step": 19008
+ },
+ {
+ "epoch": 51.79564032697548,
+ "grad_norm": 2.6156833171844482,
+ "learning_rate": 9.904250795121614e-06,
+ "loss": 0.2135,
+ "step": 19009
+ },
+ {
+ "epoch": 51.798365122615806,
+ "grad_norm": 2.060487985610962,
+ "learning_rate": 9.903368341314003e-06,
+ "loss": 0.0449,
+ "step": 19010
+ },
+ {
+ "epoch": 51.80108991825613,
+ "grad_norm": 2.995082378387451,
+ "learning_rate": 9.902485888258953e-06,
+ "loss": 0.0739,
+ "step": 19011
+ },
+ {
+ "epoch": 51.80381471389646,
+ "grad_norm": 3.427727222442627,
+ "learning_rate": 9.901603435963343e-06,
+ "loss": 0.0486,
+ "step": 19012
+ },
+ {
+ "epoch": 51.80653950953678,
+ "grad_norm": 2.609792470932007,
+ "learning_rate": 9.90072098443404e-06,
+ "loss": 0.0528,
+ "step": 19013
+ },
+ {
+ "epoch": 51.80926430517711,
+ "grad_norm": 1.6156412363052368,
+ "learning_rate": 9.899838533677922e-06,
+ "loss": 0.0265,
+ "step": 19014
+ },
+ {
+ "epoch": 51.81198910081744,
+ "grad_norm": 1.7490965127944946,
+ "learning_rate": 9.898956083701854e-06,
+ "loss": 0.038,
+ "step": 19015
+ },
+ {
+ "epoch": 51.81471389645777,
+ "grad_norm": 2.1492931842803955,
+ "learning_rate": 9.898073634512717e-06,
+ "loss": 0.0317,
+ "step": 19016
+ },
+ {
+ "epoch": 51.817438692098094,
+ "grad_norm": 2.8334267139434814,
+ "learning_rate": 9.897191186117375e-06,
+ "loss": 0.1577,
+ "step": 19017
+ },
+ {
+ "epoch": 51.82016348773842,
+ "grad_norm": 2.53041672706604,
+ "learning_rate": 9.896308738522708e-06,
+ "loss": 0.0495,
+ "step": 19018
+ },
+ {
+ "epoch": 51.822888283378745,
+ "grad_norm": 2.528789758682251,
+ "learning_rate": 9.895426291735582e-06,
+ "loss": 0.2353,
+ "step": 19019
+ },
+ {
+ "epoch": 51.82561307901907,
+ "grad_norm": 2.4779326915740967,
+ "learning_rate": 9.894543845762876e-06,
+ "loss": 0.0695,
+ "step": 19020
+ },
+ {
+ "epoch": 51.828337874659404,
+ "grad_norm": 2.9567792415618896,
+ "learning_rate": 9.893661400611456e-06,
+ "loss": 0.0551,
+ "step": 19021
+ },
+ {
+ "epoch": 51.83106267029973,
+ "grad_norm": 5.830719470977783,
+ "learning_rate": 9.892778956288198e-06,
+ "loss": 0.0846,
+ "step": 19022
+ },
+ {
+ "epoch": 51.833787465940055,
+ "grad_norm": 2.820223093032837,
+ "learning_rate": 9.891896512799975e-06,
+ "loss": 0.1415,
+ "step": 19023
+ },
+ {
+ "epoch": 51.83651226158038,
+ "grad_norm": 2.1692309379577637,
+ "learning_rate": 9.891014070153658e-06,
+ "loss": 0.0412,
+ "step": 19024
+ },
+ {
+ "epoch": 51.83923705722071,
+ "grad_norm": 2.1583456993103027,
+ "learning_rate": 9.890131628356122e-06,
+ "loss": 0.0562,
+ "step": 19025
+ },
+ {
+ "epoch": 51.84196185286103,
+ "grad_norm": 2.5917696952819824,
+ "learning_rate": 9.889249187414235e-06,
+ "loss": 0.0468,
+ "step": 19026
+ },
+ {
+ "epoch": 51.844686648501366,
+ "grad_norm": 2.47902512550354,
+ "learning_rate": 9.888366747334873e-06,
+ "loss": 0.0487,
+ "step": 19027
+ },
+ {
+ "epoch": 51.84741144414169,
+ "grad_norm": 2.542861223220825,
+ "learning_rate": 9.887484308124907e-06,
+ "loss": 0.0575,
+ "step": 19028
+ },
+ {
+ "epoch": 51.85013623978202,
+ "grad_norm": 2.381164073944092,
+ "learning_rate": 9.88660186979121e-06,
+ "loss": 0.1651,
+ "step": 19029
+ },
+ {
+ "epoch": 51.85286103542234,
+ "grad_norm": 2.0955488681793213,
+ "learning_rate": 9.885719432340652e-06,
+ "loss": 0.1325,
+ "step": 19030
+ },
+ {
+ "epoch": 51.85558583106267,
+ "grad_norm": 2.6642277240753174,
+ "learning_rate": 9.884836995780109e-06,
+ "loss": 0.1978,
+ "step": 19031
+ },
+ {
+ "epoch": 51.858310626702995,
+ "grad_norm": 2.937927007675171,
+ "learning_rate": 9.883954560116452e-06,
+ "loss": 0.0537,
+ "step": 19032
+ },
+ {
+ "epoch": 51.86103542234333,
+ "grad_norm": 5.493134498596191,
+ "learning_rate": 9.883072125356552e-06,
+ "loss": 0.0376,
+ "step": 19033
+ },
+ {
+ "epoch": 51.86376021798365,
+ "grad_norm": 2.2883830070495605,
+ "learning_rate": 9.882189691507285e-06,
+ "loss": 0.0675,
+ "step": 19034
+ },
+ {
+ "epoch": 51.86648501362398,
+ "grad_norm": 1.9431875944137573,
+ "learning_rate": 9.881307258575519e-06,
+ "loss": 0.0401,
+ "step": 19035
+ },
+ {
+ "epoch": 51.869209809264305,
+ "grad_norm": 2.073688268661499,
+ "learning_rate": 9.880424826568128e-06,
+ "loss": 0.0461,
+ "step": 19036
+ },
+ {
+ "epoch": 51.87193460490463,
+ "grad_norm": 3.138981342315674,
+ "learning_rate": 9.879542395491984e-06,
+ "loss": 0.0506,
+ "step": 19037
+ },
+ {
+ "epoch": 51.87465940054496,
+ "grad_norm": 3.064486026763916,
+ "learning_rate": 9.878659965353964e-06,
+ "loss": 0.0577,
+ "step": 19038
+ },
+ {
+ "epoch": 51.87738419618529,
+ "grad_norm": 3.0599002838134766,
+ "learning_rate": 9.87777753616093e-06,
+ "loss": 0.0911,
+ "step": 19039
+ },
+ {
+ "epoch": 51.880108991825615,
+ "grad_norm": 2.433084726333618,
+ "learning_rate": 9.876895107919761e-06,
+ "loss": 0.05,
+ "step": 19040
+ },
+ {
+ "epoch": 51.88283378746594,
+ "grad_norm": 2.6262974739074707,
+ "learning_rate": 9.876012680637334e-06,
+ "loss": 0.062,
+ "step": 19041
+ },
+ {
+ "epoch": 51.88555858310627,
+ "grad_norm": 3.0307178497314453,
+ "learning_rate": 9.875130254320513e-06,
+ "loss": 0.0481,
+ "step": 19042
+ },
+ {
+ "epoch": 51.88828337874659,
+ "grad_norm": 2.297208309173584,
+ "learning_rate": 9.874247828976176e-06,
+ "loss": 0.2096,
+ "step": 19043
+ },
+ {
+ "epoch": 51.89100817438692,
+ "grad_norm": 2.818955659866333,
+ "learning_rate": 9.87336540461119e-06,
+ "loss": 0.1936,
+ "step": 19044
+ },
+ {
+ "epoch": 51.89373297002725,
+ "grad_norm": 2.5621001720428467,
+ "learning_rate": 9.872482981232434e-06,
+ "loss": 0.057,
+ "step": 19045
+ },
+ {
+ "epoch": 51.89645776566758,
+ "grad_norm": 2.5616114139556885,
+ "learning_rate": 9.871600558846772e-06,
+ "loss": 0.1344,
+ "step": 19046
+ },
+ {
+ "epoch": 51.8991825613079,
+ "grad_norm": 2.1055800914764404,
+ "learning_rate": 9.870718137461085e-06,
+ "loss": 0.0605,
+ "step": 19047
+ },
+ {
+ "epoch": 51.90190735694823,
+ "grad_norm": 2.5870749950408936,
+ "learning_rate": 9.869835717082237e-06,
+ "loss": 0.0613,
+ "step": 19048
+ },
+ {
+ "epoch": 51.904632152588555,
+ "grad_norm": 2.043525457382202,
+ "learning_rate": 9.868953297717108e-06,
+ "loss": 0.0358,
+ "step": 19049
+ },
+ {
+ "epoch": 51.90735694822888,
+ "grad_norm": 2.003561496734619,
+ "learning_rate": 9.868070879372563e-06,
+ "loss": 0.0399,
+ "step": 19050
+ },
+ {
+ "epoch": 51.91008174386921,
+ "grad_norm": 2.9481897354125977,
+ "learning_rate": 9.867188462055482e-06,
+ "loss": 0.0922,
+ "step": 19051
+ },
+ {
+ "epoch": 51.91280653950954,
+ "grad_norm": 2.2971293926239014,
+ "learning_rate": 9.86630604577273e-06,
+ "loss": 0.073,
+ "step": 19052
+ },
+ {
+ "epoch": 51.915531335149865,
+ "grad_norm": 2.5365874767303467,
+ "learning_rate": 9.865423630531182e-06,
+ "loss": 0.0375,
+ "step": 19053
+ },
+ {
+ "epoch": 51.91825613079019,
+ "grad_norm": 2.0723912715911865,
+ "learning_rate": 9.86454121633771e-06,
+ "loss": 0.0649,
+ "step": 19054
+ },
+ {
+ "epoch": 51.920980926430516,
+ "grad_norm": 2.5480291843414307,
+ "learning_rate": 9.863658803199187e-06,
+ "loss": 0.0461,
+ "step": 19055
+ },
+ {
+ "epoch": 51.92370572207084,
+ "grad_norm": 1.9463211297988892,
+ "learning_rate": 9.862776391122482e-06,
+ "loss": 0.0472,
+ "step": 19056
+ },
+ {
+ "epoch": 51.926430517711175,
+ "grad_norm": 2.219831705093384,
+ "learning_rate": 9.861893980114475e-06,
+ "loss": 0.0418,
+ "step": 19057
+ },
+ {
+ "epoch": 51.9291553133515,
+ "grad_norm": 2.2796175479888916,
+ "learning_rate": 9.86101157018203e-06,
+ "loss": 0.0416,
+ "step": 19058
+ },
+ {
+ "epoch": 51.93188010899183,
+ "grad_norm": 3.2271766662597656,
+ "learning_rate": 9.860129161332026e-06,
+ "loss": 0.2217,
+ "step": 19059
+ },
+ {
+ "epoch": 51.93460490463215,
+ "grad_norm": 2.315675973892212,
+ "learning_rate": 9.859246753571328e-06,
+ "loss": 0.0476,
+ "step": 19060
+ },
+ {
+ "epoch": 51.93732970027248,
+ "grad_norm": 2.56036376953125,
+ "learning_rate": 9.858364346906814e-06,
+ "loss": 0.0541,
+ "step": 19061
+ },
+ {
+ "epoch": 51.940054495912804,
+ "grad_norm": 2.7747342586517334,
+ "learning_rate": 9.857481941345354e-06,
+ "loss": 0.0902,
+ "step": 19062
+ },
+ {
+ "epoch": 51.94277929155314,
+ "grad_norm": 2.120832920074463,
+ "learning_rate": 9.856599536893821e-06,
+ "loss": 0.0587,
+ "step": 19063
+ },
+ {
+ "epoch": 51.94550408719346,
+ "grad_norm": 3.270843982696533,
+ "learning_rate": 9.855717133559082e-06,
+ "loss": 0.2901,
+ "step": 19064
+ },
+ {
+ "epoch": 51.94822888283379,
+ "grad_norm": 1.8834240436553955,
+ "learning_rate": 9.854834731348017e-06,
+ "loss": 0.0357,
+ "step": 19065
+ },
+ {
+ "epoch": 51.950953678474114,
+ "grad_norm": 2.5294008255004883,
+ "learning_rate": 9.853952330267494e-06,
+ "loss": 0.0519,
+ "step": 19066
+ },
+ {
+ "epoch": 51.95367847411444,
+ "grad_norm": 2.5097670555114746,
+ "learning_rate": 9.853069930324386e-06,
+ "loss": 0.0762,
+ "step": 19067
+ },
+ {
+ "epoch": 51.956403269754766,
+ "grad_norm": 2.551447868347168,
+ "learning_rate": 9.852187531525563e-06,
+ "loss": 0.0755,
+ "step": 19068
+ },
+ {
+ "epoch": 51.95912806539509,
+ "grad_norm": 2.2220497131347656,
+ "learning_rate": 9.8513051338779e-06,
+ "loss": 0.0443,
+ "step": 19069
+ },
+ {
+ "epoch": 51.961852861035425,
+ "grad_norm": 3.055201292037964,
+ "learning_rate": 9.850422737388266e-06,
+ "loss": 0.0645,
+ "step": 19070
+ },
+ {
+ "epoch": 51.96457765667575,
+ "grad_norm": 2.8203787803649902,
+ "learning_rate": 9.849540342063538e-06,
+ "loss": 0.0475,
+ "step": 19071
+ },
+ {
+ "epoch": 51.967302452316076,
+ "grad_norm": 2.519509792327881,
+ "learning_rate": 9.848657947910582e-06,
+ "loss": 0.1006,
+ "step": 19072
+ },
+ {
+ "epoch": 51.9700272479564,
+ "grad_norm": 2.029930830001831,
+ "learning_rate": 9.84777555493627e-06,
+ "loss": 0.2435,
+ "step": 19073
+ },
+ {
+ "epoch": 51.97275204359673,
+ "grad_norm": 2.94417405128479,
+ "learning_rate": 9.84689316314748e-06,
+ "loss": 0.0413,
+ "step": 19074
+ },
+ {
+ "epoch": 51.97547683923706,
+ "grad_norm": 2.0514931678771973,
+ "learning_rate": 9.846010772551083e-06,
+ "loss": 0.0456,
+ "step": 19075
+ },
+ {
+ "epoch": 51.97820163487739,
+ "grad_norm": 2.6659953594207764,
+ "learning_rate": 9.845128383153948e-06,
+ "loss": 0.0567,
+ "step": 19076
+ },
+ {
+ "epoch": 51.98092643051771,
+ "grad_norm": 2.395395517349243,
+ "learning_rate": 9.84424599496295e-06,
+ "loss": 0.1335,
+ "step": 19077
+ },
+ {
+ "epoch": 51.98365122615804,
+ "grad_norm": 2.4223453998565674,
+ "learning_rate": 9.843363607984957e-06,
+ "loss": 0.0723,
+ "step": 19078
+ },
+ {
+ "epoch": 51.986376021798364,
+ "grad_norm": 2.181818962097168,
+ "learning_rate": 9.842481222226845e-06,
+ "loss": 0.0599,
+ "step": 19079
+ },
+ {
+ "epoch": 51.98910081743869,
+ "grad_norm": 5.835550308227539,
+ "learning_rate": 9.841598837695482e-06,
+ "loss": 0.068,
+ "step": 19080
+ },
+ {
+ "epoch": 51.991825613079016,
+ "grad_norm": 3.034038543701172,
+ "learning_rate": 9.840716454397746e-06,
+ "loss": 0.0795,
+ "step": 19081
+ },
+ {
+ "epoch": 51.99455040871935,
+ "grad_norm": 2.700390100479126,
+ "learning_rate": 9.8398340723405e-06,
+ "loss": 0.0631,
+ "step": 19082
+ },
+ {
+ "epoch": 51.997275204359674,
+ "grad_norm": 2.3030827045440674,
+ "learning_rate": 9.838951691530625e-06,
+ "loss": 0.0469,
+ "step": 19083
+ },
+ {
+ "epoch": 52.0,
+ "grad_norm": 2.7661163806915283,
+ "learning_rate": 9.838069311974986e-06,
+ "loss": 0.0382,
+ "step": 19084
+ },
+ {
+ "epoch": 52.002724795640326,
+ "grad_norm": 2.7143571376800537,
+ "learning_rate": 9.837186933680462e-06,
+ "loss": 0.062,
+ "step": 19085
+ },
+ {
+ "epoch": 52.00544959128065,
+ "grad_norm": 1.8123778104782104,
+ "learning_rate": 9.836304556653916e-06,
+ "loss": 0.1242,
+ "step": 19086
+ },
+ {
+ "epoch": 52.00817438692098,
+ "grad_norm": 3.082768440246582,
+ "learning_rate": 9.835422180902228e-06,
+ "loss": 0.0988,
+ "step": 19087
+ },
+ {
+ "epoch": 52.01089918256131,
+ "grad_norm": 2.208719491958618,
+ "learning_rate": 9.834539806432267e-06,
+ "loss": 0.0418,
+ "step": 19088
+ },
+ {
+ "epoch": 52.013623978201636,
+ "grad_norm": 2.0091753005981445,
+ "learning_rate": 9.833657433250904e-06,
+ "loss": 0.1136,
+ "step": 19089
+ },
+ {
+ "epoch": 52.01634877384196,
+ "grad_norm": 2.2976772785186768,
+ "learning_rate": 9.832775061365008e-06,
+ "loss": 0.0401,
+ "step": 19090
+ },
+ {
+ "epoch": 52.01907356948229,
+ "grad_norm": 1.8580418825149536,
+ "learning_rate": 9.831892690781457e-06,
+ "loss": 0.0653,
+ "step": 19091
+ },
+ {
+ "epoch": 52.02179836512261,
+ "grad_norm": 28.690593719482422,
+ "learning_rate": 9.831010321507122e-06,
+ "loss": 0.135,
+ "step": 19092
+ },
+ {
+ "epoch": 52.02452316076294,
+ "grad_norm": 2.191596031188965,
+ "learning_rate": 9.830127953548872e-06,
+ "loss": 0.0407,
+ "step": 19093
+ },
+ {
+ "epoch": 52.02724795640327,
+ "grad_norm": 2.2804856300354004,
+ "learning_rate": 9.82924558691358e-06,
+ "loss": 0.0517,
+ "step": 19094
+ },
+ {
+ "epoch": 52.0299727520436,
+ "grad_norm": 2.69991397857666,
+ "learning_rate": 9.828363221608118e-06,
+ "loss": 0.0472,
+ "step": 19095
+ },
+ {
+ "epoch": 52.032697547683924,
+ "grad_norm": 2.2915029525756836,
+ "learning_rate": 9.827480857639357e-06,
+ "loss": 0.0788,
+ "step": 19096
+ },
+ {
+ "epoch": 52.03542234332425,
+ "grad_norm": 2.866312026977539,
+ "learning_rate": 9.826598495014172e-06,
+ "loss": 0.0639,
+ "step": 19097
+ },
+ {
+ "epoch": 52.038147138964575,
+ "grad_norm": 2.5231566429138184,
+ "learning_rate": 9.82571613373943e-06,
+ "loss": 0.0556,
+ "step": 19098
+ },
+ {
+ "epoch": 52.0408719346049,
+ "grad_norm": 2.982900857925415,
+ "learning_rate": 9.824833773822006e-06,
+ "loss": 0.1903,
+ "step": 19099
+ },
+ {
+ "epoch": 52.043596730245234,
+ "grad_norm": 3.0458099842071533,
+ "learning_rate": 9.82395141526877e-06,
+ "loss": 0.0632,
+ "step": 19100
+ },
+ {
+ "epoch": 52.04632152588556,
+ "grad_norm": 2.1493985652923584,
+ "learning_rate": 9.823069058086597e-06,
+ "loss": 0.1154,
+ "step": 19101
+ },
+ {
+ "epoch": 52.049046321525886,
+ "grad_norm": 2.1910831928253174,
+ "learning_rate": 9.822186702282354e-06,
+ "loss": 0.2053,
+ "step": 19102
+ },
+ {
+ "epoch": 52.05177111716621,
+ "grad_norm": 3.5950028896331787,
+ "learning_rate": 9.821304347862917e-06,
+ "loss": 0.0839,
+ "step": 19103
+ },
+ {
+ "epoch": 52.05449591280654,
+ "grad_norm": 3.014310836791992,
+ "learning_rate": 9.820421994835154e-06,
+ "loss": 0.1713,
+ "step": 19104
+ },
+ {
+ "epoch": 52.05722070844686,
+ "grad_norm": 2.4017786979675293,
+ "learning_rate": 9.81953964320594e-06,
+ "loss": 0.0416,
+ "step": 19105
+ },
+ {
+ "epoch": 52.059945504087196,
+ "grad_norm": 2.7370097637176514,
+ "learning_rate": 9.818657292982146e-06,
+ "loss": 0.044,
+ "step": 19106
+ },
+ {
+ "epoch": 52.06267029972752,
+ "grad_norm": 3.65400767326355,
+ "learning_rate": 9.817774944170638e-06,
+ "loss": 0.0808,
+ "step": 19107
+ },
+ {
+ "epoch": 52.06539509536785,
+ "grad_norm": 2.294445037841797,
+ "learning_rate": 9.816892596778299e-06,
+ "loss": 0.0805,
+ "step": 19108
+ },
+ {
+ "epoch": 52.06811989100817,
+ "grad_norm": 2.4163525104522705,
+ "learning_rate": 9.816010250811993e-06,
+ "loss": 0.1247,
+ "step": 19109
+ },
+ {
+ "epoch": 52.0708446866485,
+ "grad_norm": 2.902838706970215,
+ "learning_rate": 9.815127906278595e-06,
+ "loss": 0.108,
+ "step": 19110
+ },
+ {
+ "epoch": 52.073569482288825,
+ "grad_norm": 2.8715434074401855,
+ "learning_rate": 9.814245563184971e-06,
+ "loss": 0.1132,
+ "step": 19111
+ },
+ {
+ "epoch": 52.07629427792916,
+ "grad_norm": 2.2053656578063965,
+ "learning_rate": 9.813363221538e-06,
+ "loss": 0.1058,
+ "step": 19112
+ },
+ {
+ "epoch": 52.079019073569484,
+ "grad_norm": 1.8813954591751099,
+ "learning_rate": 9.812480881344548e-06,
+ "loss": 0.0601,
+ "step": 19113
+ },
+ {
+ "epoch": 52.08174386920981,
+ "grad_norm": 2.205298662185669,
+ "learning_rate": 9.811598542611492e-06,
+ "loss": 0.1385,
+ "step": 19114
+ },
+ {
+ "epoch": 52.084468664850135,
+ "grad_norm": 2.7368013858795166,
+ "learning_rate": 9.810716205345695e-06,
+ "loss": 0.1248,
+ "step": 19115
+ },
+ {
+ "epoch": 52.08719346049046,
+ "grad_norm": 2.141465663909912,
+ "learning_rate": 9.80983386955404e-06,
+ "loss": 0.0752,
+ "step": 19116
+ },
+ {
+ "epoch": 52.08991825613079,
+ "grad_norm": 3.20174503326416,
+ "learning_rate": 9.808951535243389e-06,
+ "loss": 0.1182,
+ "step": 19117
+ },
+ {
+ "epoch": 52.09264305177112,
+ "grad_norm": 3.0807106494903564,
+ "learning_rate": 9.808069202420617e-06,
+ "loss": 0.0695,
+ "step": 19118
+ },
+ {
+ "epoch": 52.095367847411445,
+ "grad_norm": 3.170603036880493,
+ "learning_rate": 9.8071868710926e-06,
+ "loss": 0.1336,
+ "step": 19119
+ },
+ {
+ "epoch": 52.09809264305177,
+ "grad_norm": 2.8674962520599365,
+ "learning_rate": 9.8063045412662e-06,
+ "loss": 0.0825,
+ "step": 19120
+ },
+ {
+ "epoch": 52.1008174386921,
+ "grad_norm": 1.800222396850586,
+ "learning_rate": 9.805422212948297e-06,
+ "loss": 0.0298,
+ "step": 19121
+ },
+ {
+ "epoch": 52.10354223433242,
+ "grad_norm": 2.0392096042633057,
+ "learning_rate": 9.80453988614576e-06,
+ "loss": 0.0393,
+ "step": 19122
+ },
+ {
+ "epoch": 52.10626702997275,
+ "grad_norm": 2.6290528774261475,
+ "learning_rate": 9.80365756086546e-06,
+ "loss": 0.1514,
+ "step": 19123
+ },
+ {
+ "epoch": 52.10899182561308,
+ "grad_norm": 2.545975685119629,
+ "learning_rate": 9.802775237114267e-06,
+ "loss": 0.0401,
+ "step": 19124
+ },
+ {
+ "epoch": 52.11171662125341,
+ "grad_norm": 2.7095930576324463,
+ "learning_rate": 9.801892914899051e-06,
+ "loss": 0.1248,
+ "step": 19125
+ },
+ {
+ "epoch": 52.11444141689373,
+ "grad_norm": 2.1136295795440674,
+ "learning_rate": 9.801010594226691e-06,
+ "loss": 0.0511,
+ "step": 19126
+ },
+ {
+ "epoch": 52.11716621253406,
+ "grad_norm": 1.417776346206665,
+ "learning_rate": 9.800128275104054e-06,
+ "loss": 0.0249,
+ "step": 19127
+ },
+ {
+ "epoch": 52.119891008174385,
+ "grad_norm": 1.749893307685852,
+ "learning_rate": 9.799245957538013e-06,
+ "loss": 0.0357,
+ "step": 19128
+ },
+ {
+ "epoch": 52.12261580381471,
+ "grad_norm": 3.0674848556518555,
+ "learning_rate": 9.798363641535434e-06,
+ "loss": 0.1052,
+ "step": 19129
+ },
+ {
+ "epoch": 52.12534059945504,
+ "grad_norm": 2.3643083572387695,
+ "learning_rate": 9.797481327103196e-06,
+ "loss": 0.0939,
+ "step": 19130
+ },
+ {
+ "epoch": 52.12806539509537,
+ "grad_norm": 2.2763144969940186,
+ "learning_rate": 9.796599014248164e-06,
+ "loss": 0.0391,
+ "step": 19131
+ },
+ {
+ "epoch": 52.130790190735695,
+ "grad_norm": 2.3077893257141113,
+ "learning_rate": 9.795716702977216e-06,
+ "loss": 0.0309,
+ "step": 19132
+ },
+ {
+ "epoch": 52.13351498637602,
+ "grad_norm": 2.789299488067627,
+ "learning_rate": 9.794834393297217e-06,
+ "loss": 0.0923,
+ "step": 19133
+ },
+ {
+ "epoch": 52.13623978201635,
+ "grad_norm": 2.48203182220459,
+ "learning_rate": 9.793952085215043e-06,
+ "loss": 0.0669,
+ "step": 19134
+ },
+ {
+ "epoch": 52.13896457765667,
+ "grad_norm": 3.1909079551696777,
+ "learning_rate": 9.793069778737561e-06,
+ "loss": 0.1323,
+ "step": 19135
+ },
+ {
+ "epoch": 52.141689373297005,
+ "grad_norm": 3.0375287532806396,
+ "learning_rate": 9.792187473871648e-06,
+ "loss": 0.2599,
+ "step": 19136
+ },
+ {
+ "epoch": 52.14441416893733,
+ "grad_norm": 2.3030037879943848,
+ "learning_rate": 9.79130517062417e-06,
+ "loss": 0.1159,
+ "step": 19137
+ },
+ {
+ "epoch": 52.14713896457766,
+ "grad_norm": 2.3653199672698975,
+ "learning_rate": 9.790422869002003e-06,
+ "loss": 0.0692,
+ "step": 19138
+ },
+ {
+ "epoch": 52.14986376021798,
+ "grad_norm": 2.5181987285614014,
+ "learning_rate": 9.789540569012012e-06,
+ "loss": 0.0412,
+ "step": 19139
+ },
+ {
+ "epoch": 52.15258855585831,
+ "grad_norm": 2.166254758834839,
+ "learning_rate": 9.788658270661076e-06,
+ "loss": 0.0443,
+ "step": 19140
+ },
+ {
+ "epoch": 52.155313351498634,
+ "grad_norm": 3.67846941947937,
+ "learning_rate": 9.78777597395606e-06,
+ "loss": 0.1536,
+ "step": 19141
+ },
+ {
+ "epoch": 52.15803814713897,
+ "grad_norm": 2.4900782108306885,
+ "learning_rate": 9.786893678903835e-06,
+ "loss": 0.052,
+ "step": 19142
+ },
+ {
+ "epoch": 52.16076294277929,
+ "grad_norm": 2.134281873703003,
+ "learning_rate": 9.786011385511279e-06,
+ "loss": 0.0356,
+ "step": 19143
+ },
+ {
+ "epoch": 52.16348773841962,
+ "grad_norm": 4.112092971801758,
+ "learning_rate": 9.78512909378526e-06,
+ "loss": 0.0542,
+ "step": 19144
+ },
+ {
+ "epoch": 52.166212534059945,
+ "grad_norm": 2.671893835067749,
+ "learning_rate": 9.784246803732648e-06,
+ "loss": 0.0921,
+ "step": 19145
+ },
+ {
+ "epoch": 52.16893732970027,
+ "grad_norm": 3.0067803859710693,
+ "learning_rate": 9.783364515360315e-06,
+ "loss": 0.1119,
+ "step": 19146
+ },
+ {
+ "epoch": 52.171662125340596,
+ "grad_norm": 2.2790000438690186,
+ "learning_rate": 9.782482228675132e-06,
+ "loss": 0.1231,
+ "step": 19147
+ },
+ {
+ "epoch": 52.17438692098093,
+ "grad_norm": 2.4479219913482666,
+ "learning_rate": 9.781599943683972e-06,
+ "loss": 0.0513,
+ "step": 19148
+ },
+ {
+ "epoch": 52.177111716621255,
+ "grad_norm": 2.6366541385650635,
+ "learning_rate": 9.780717660393702e-06,
+ "loss": 0.1498,
+ "step": 19149
+ },
+ {
+ "epoch": 52.17983651226158,
+ "grad_norm": 2.29569411277771,
+ "learning_rate": 9.7798353788112e-06,
+ "loss": 0.0521,
+ "step": 19150
+ },
+ {
+ "epoch": 52.182561307901906,
+ "grad_norm": 2.4435312747955322,
+ "learning_rate": 9.778953098943329e-06,
+ "loss": 0.1223,
+ "step": 19151
+ },
+ {
+ "epoch": 52.18528610354223,
+ "grad_norm": 2.0505971908569336,
+ "learning_rate": 9.778070820796966e-06,
+ "loss": 0.0425,
+ "step": 19152
+ },
+ {
+ "epoch": 52.18801089918256,
+ "grad_norm": 2.359196186065674,
+ "learning_rate": 9.777188544378979e-06,
+ "loss": 0.1786,
+ "step": 19153
+ },
+ {
+ "epoch": 52.19073569482289,
+ "grad_norm": 3.0749714374542236,
+ "learning_rate": 9.77630626969624e-06,
+ "loss": 0.0641,
+ "step": 19154
+ },
+ {
+ "epoch": 52.19346049046322,
+ "grad_norm": 2.1745007038116455,
+ "learning_rate": 9.775423996755622e-06,
+ "loss": 0.0669,
+ "step": 19155
+ },
+ {
+ "epoch": 52.19618528610354,
+ "grad_norm": 2.469118595123291,
+ "learning_rate": 9.774541725563996e-06,
+ "loss": 0.0394,
+ "step": 19156
+ },
+ {
+ "epoch": 52.19891008174387,
+ "grad_norm": 3.1347122192382812,
+ "learning_rate": 9.773659456128229e-06,
+ "loss": 0.1425,
+ "step": 19157
+ },
+ {
+ "epoch": 52.201634877384194,
+ "grad_norm": 3.08113431930542,
+ "learning_rate": 9.772777188455196e-06,
+ "loss": 0.05,
+ "step": 19158
+ },
+ {
+ "epoch": 52.20435967302452,
+ "grad_norm": 2.235389471054077,
+ "learning_rate": 9.771894922551763e-06,
+ "loss": 0.2147,
+ "step": 19159
+ },
+ {
+ "epoch": 52.20708446866485,
+ "grad_norm": 2.5611720085144043,
+ "learning_rate": 9.771012658424808e-06,
+ "loss": 0.0442,
+ "step": 19160
+ },
+ {
+ "epoch": 52.20980926430518,
+ "grad_norm": 2.5288827419281006,
+ "learning_rate": 9.7701303960812e-06,
+ "loss": 0.1316,
+ "step": 19161
+ },
+ {
+ "epoch": 52.212534059945504,
+ "grad_norm": 3.0282843112945557,
+ "learning_rate": 9.76924813552781e-06,
+ "loss": 0.094,
+ "step": 19162
+ },
+ {
+ "epoch": 52.21525885558583,
+ "grad_norm": 3.1223175525665283,
+ "learning_rate": 9.768365876771505e-06,
+ "loss": 0.0546,
+ "step": 19163
+ },
+ {
+ "epoch": 52.217983651226156,
+ "grad_norm": 1.9844022989273071,
+ "learning_rate": 9.767483619819164e-06,
+ "loss": 0.0815,
+ "step": 19164
+ },
+ {
+ "epoch": 52.22070844686648,
+ "grad_norm": 2.5727250576019287,
+ "learning_rate": 9.766601364677648e-06,
+ "loss": 0.0751,
+ "step": 19165
+ },
+ {
+ "epoch": 52.223433242506815,
+ "grad_norm": 2.461308717727661,
+ "learning_rate": 9.765719111353837e-06,
+ "loss": 0.1835,
+ "step": 19166
+ },
+ {
+ "epoch": 52.22615803814714,
+ "grad_norm": 2.599323034286499,
+ "learning_rate": 9.764836859854595e-06,
+ "loss": 0.0983,
+ "step": 19167
+ },
+ {
+ "epoch": 52.228882833787466,
+ "grad_norm": 2.469376802444458,
+ "learning_rate": 9.763954610186798e-06,
+ "loss": 0.0478,
+ "step": 19168
+ },
+ {
+ "epoch": 52.23160762942779,
+ "grad_norm": 3.1060991287231445,
+ "learning_rate": 9.763072362357315e-06,
+ "loss": 0.0708,
+ "step": 19169
+ },
+ {
+ "epoch": 52.23433242506812,
+ "grad_norm": 3.683716297149658,
+ "learning_rate": 9.762190116373017e-06,
+ "loss": 0.2163,
+ "step": 19170
+ },
+ {
+ "epoch": 52.237057220708444,
+ "grad_norm": 1.5595203638076782,
+ "learning_rate": 9.761307872240773e-06,
+ "loss": 0.0876,
+ "step": 19171
+ },
+ {
+ "epoch": 52.23978201634878,
+ "grad_norm": 2.2832465171813965,
+ "learning_rate": 9.760425629967457e-06,
+ "loss": 0.0601,
+ "step": 19172
+ },
+ {
+ "epoch": 52.2425068119891,
+ "grad_norm": 2.2269411087036133,
+ "learning_rate": 9.759543389559937e-06,
+ "loss": 0.041,
+ "step": 19173
+ },
+ {
+ "epoch": 52.24523160762943,
+ "grad_norm": 2.932615280151367,
+ "learning_rate": 9.758661151025087e-06,
+ "loss": 0.0403,
+ "step": 19174
+ },
+ {
+ "epoch": 52.247956403269754,
+ "grad_norm": 2.6302309036254883,
+ "learning_rate": 9.757778914369774e-06,
+ "loss": 0.038,
+ "step": 19175
+ },
+ {
+ "epoch": 52.25068119891008,
+ "grad_norm": 2.2434003353118896,
+ "learning_rate": 9.75689667960087e-06,
+ "loss": 0.0889,
+ "step": 19176
+ },
+ {
+ "epoch": 52.253405994550405,
+ "grad_norm": 2.437620162963867,
+ "learning_rate": 9.75601444672525e-06,
+ "loss": 0.0642,
+ "step": 19177
+ },
+ {
+ "epoch": 52.25613079019074,
+ "grad_norm": 2.194697380065918,
+ "learning_rate": 9.75513221574978e-06,
+ "loss": 0.053,
+ "step": 19178
+ },
+ {
+ "epoch": 52.258855585831064,
+ "grad_norm": 1.6035573482513428,
+ "learning_rate": 9.754249986681334e-06,
+ "loss": 0.026,
+ "step": 19179
+ },
+ {
+ "epoch": 52.26158038147139,
+ "grad_norm": 2.5342857837677,
+ "learning_rate": 9.75336775952678e-06,
+ "loss": 0.0742,
+ "step": 19180
+ },
+ {
+ "epoch": 52.264305177111716,
+ "grad_norm": 1.9136642217636108,
+ "learning_rate": 9.752485534292992e-06,
+ "loss": 0.0312,
+ "step": 19181
+ },
+ {
+ "epoch": 52.26702997275204,
+ "grad_norm": 2.7598845958709717,
+ "learning_rate": 9.751603310986835e-06,
+ "loss": 0.0873,
+ "step": 19182
+ },
+ {
+ "epoch": 52.26975476839237,
+ "grad_norm": 3.0727250576019287,
+ "learning_rate": 9.750721089615185e-06,
+ "loss": 0.0368,
+ "step": 19183
+ },
+ {
+ "epoch": 52.2724795640327,
+ "grad_norm": 2.758974313735962,
+ "learning_rate": 9.749838870184914e-06,
+ "loss": 0.0537,
+ "step": 19184
+ },
+ {
+ "epoch": 52.275204359673026,
+ "grad_norm": 2.2884390354156494,
+ "learning_rate": 9.748956652702885e-06,
+ "loss": 0.082,
+ "step": 19185
+ },
+ {
+ "epoch": 52.27792915531335,
+ "grad_norm": 4.4253153800964355,
+ "learning_rate": 9.748074437175979e-06,
+ "loss": 0.2698,
+ "step": 19186
+ },
+ {
+ "epoch": 52.28065395095368,
+ "grad_norm": 3.2324154376983643,
+ "learning_rate": 9.747192223611055e-06,
+ "loss": 0.1438,
+ "step": 19187
+ },
+ {
+ "epoch": 52.283378746594,
+ "grad_norm": 2.4014697074890137,
+ "learning_rate": 9.746310012014996e-06,
+ "loss": 0.0847,
+ "step": 19188
+ },
+ {
+ "epoch": 52.28610354223433,
+ "grad_norm": 2.666675090789795,
+ "learning_rate": 9.745427802394662e-06,
+ "loss": 0.1204,
+ "step": 19189
+ },
+ {
+ "epoch": 52.28882833787466,
+ "grad_norm": 2.1587202548980713,
+ "learning_rate": 9.74454559475693e-06,
+ "loss": 0.0405,
+ "step": 19190
+ },
+ {
+ "epoch": 52.29155313351499,
+ "grad_norm": 2.9209234714508057,
+ "learning_rate": 9.743663389108667e-06,
+ "loss": 0.0514,
+ "step": 19191
+ },
+ {
+ "epoch": 52.294277929155314,
+ "grad_norm": 3.301055669784546,
+ "learning_rate": 9.742781185456748e-06,
+ "loss": 0.1226,
+ "step": 19192
+ },
+ {
+ "epoch": 52.29700272479564,
+ "grad_norm": 2.055111885070801,
+ "learning_rate": 9.74189898380804e-06,
+ "loss": 0.0357,
+ "step": 19193
+ },
+ {
+ "epoch": 52.299727520435965,
+ "grad_norm": 1.7981451749801636,
+ "learning_rate": 9.741016784169409e-06,
+ "loss": 0.0483,
+ "step": 19194
+ },
+ {
+ "epoch": 52.30245231607629,
+ "grad_norm": 3.292200803756714,
+ "learning_rate": 9.740134586547737e-06,
+ "loss": 0.075,
+ "step": 19195
+ },
+ {
+ "epoch": 52.305177111716624,
+ "grad_norm": 2.858973979949951,
+ "learning_rate": 9.739252390949887e-06,
+ "loss": 0.1892,
+ "step": 19196
+ },
+ {
+ "epoch": 52.30790190735695,
+ "grad_norm": 2.143068313598633,
+ "learning_rate": 9.738370197382732e-06,
+ "loss": 0.0333,
+ "step": 19197
+ },
+ {
+ "epoch": 52.310626702997276,
+ "grad_norm": 1.889665126800537,
+ "learning_rate": 9.73748800585314e-06,
+ "loss": 0.0367,
+ "step": 19198
+ },
+ {
+ "epoch": 52.3133514986376,
+ "grad_norm": 2.9359078407287598,
+ "learning_rate": 9.736605816367984e-06,
+ "loss": 0.0996,
+ "step": 19199
+ },
+ {
+ "epoch": 52.31607629427793,
+ "grad_norm": 2.1729328632354736,
+ "learning_rate": 9.735723628934132e-06,
+ "loss": 0.0342,
+ "step": 19200
+ },
+ {
+ "epoch": 52.31880108991825,
+ "grad_norm": 3.04559326171875,
+ "learning_rate": 9.73484144355846e-06,
+ "loss": 0.2477,
+ "step": 19201
+ },
+ {
+ "epoch": 52.321525885558586,
+ "grad_norm": 2.723592758178711,
+ "learning_rate": 9.733959260247829e-06,
+ "loss": 0.07,
+ "step": 19202
+ },
+ {
+ "epoch": 52.32425068119891,
+ "grad_norm": 2.4956979751586914,
+ "learning_rate": 9.733077079009118e-06,
+ "loss": 0.1432,
+ "step": 19203
+ },
+ {
+ "epoch": 52.32697547683924,
+ "grad_norm": 2.080479860305786,
+ "learning_rate": 9.732194899849192e-06,
+ "loss": 0.1393,
+ "step": 19204
+ },
+ {
+ "epoch": 52.32970027247956,
+ "grad_norm": 2.108461856842041,
+ "learning_rate": 9.731312722774924e-06,
+ "loss": 0.0506,
+ "step": 19205
+ },
+ {
+ "epoch": 52.33242506811989,
+ "grad_norm": 2.3983516693115234,
+ "learning_rate": 9.730430547793186e-06,
+ "loss": 0.0813,
+ "step": 19206
+ },
+ {
+ "epoch": 52.335149863760215,
+ "grad_norm": 2.3082780838012695,
+ "learning_rate": 9.729548374910843e-06,
+ "loss": 0.1814,
+ "step": 19207
+ },
+ {
+ "epoch": 52.33787465940055,
+ "grad_norm": 2.3215584754943848,
+ "learning_rate": 9.728666204134772e-06,
+ "loss": 0.081,
+ "step": 19208
+ },
+ {
+ "epoch": 52.34059945504087,
+ "grad_norm": 3.1427719593048096,
+ "learning_rate": 9.727784035471837e-06,
+ "loss": 0.128,
+ "step": 19209
+ },
+ {
+ "epoch": 52.3433242506812,
+ "grad_norm": 1.5243710279464722,
+ "learning_rate": 9.726901868928913e-06,
+ "loss": 0.0711,
+ "step": 19210
+ },
+ {
+ "epoch": 52.346049046321525,
+ "grad_norm": 2.728058338165283,
+ "learning_rate": 9.726019704512864e-06,
+ "loss": 0.0285,
+ "step": 19211
+ },
+ {
+ "epoch": 52.34877384196185,
+ "grad_norm": 2.6983611583709717,
+ "learning_rate": 9.725137542230568e-06,
+ "loss": 0.1675,
+ "step": 19212
+ },
+ {
+ "epoch": 52.35149863760218,
+ "grad_norm": 2.252480983734131,
+ "learning_rate": 9.724255382088895e-06,
+ "loss": 0.199,
+ "step": 19213
+ },
+ {
+ "epoch": 52.35422343324251,
+ "grad_norm": 2.894031286239624,
+ "learning_rate": 9.723373224094708e-06,
+ "loss": 0.1191,
+ "step": 19214
+ },
+ {
+ "epoch": 52.356948228882835,
+ "grad_norm": 1.85772705078125,
+ "learning_rate": 9.722491068254883e-06,
+ "loss": 0.0581,
+ "step": 19215
+ },
+ {
+ "epoch": 52.35967302452316,
+ "grad_norm": 2.3829479217529297,
+ "learning_rate": 9.721608914576288e-06,
+ "loss": 0.0435,
+ "step": 19216
+ },
+ {
+ "epoch": 52.36239782016349,
+ "grad_norm": 2.349857807159424,
+ "learning_rate": 9.720726763065795e-06,
+ "loss": 0.1173,
+ "step": 19217
+ },
+ {
+ "epoch": 52.36512261580381,
+ "grad_norm": 2.669268846511841,
+ "learning_rate": 9.719844613730273e-06,
+ "loss": 0.1283,
+ "step": 19218
+ },
+ {
+ "epoch": 52.36784741144414,
+ "grad_norm": 2.5215466022491455,
+ "learning_rate": 9.718962466576592e-06,
+ "loss": 0.1062,
+ "step": 19219
+ },
+ {
+ "epoch": 52.37057220708447,
+ "grad_norm": 2.6064453125,
+ "learning_rate": 9.71808032161162e-06,
+ "loss": 0.0993,
+ "step": 19220
+ },
+ {
+ "epoch": 52.3732970027248,
+ "grad_norm": 3.0583574771881104,
+ "learning_rate": 9.717198178842234e-06,
+ "loss": 0.1404,
+ "step": 19221
+ },
+ {
+ "epoch": 52.37602179836512,
+ "grad_norm": 1.4960284233093262,
+ "learning_rate": 9.716316038275295e-06,
+ "loss": 0.0291,
+ "step": 19222
+ },
+ {
+ "epoch": 52.37874659400545,
+ "grad_norm": 2.021129846572876,
+ "learning_rate": 9.715433899917682e-06,
+ "loss": 0.2218,
+ "step": 19223
+ },
+ {
+ "epoch": 52.381471389645775,
+ "grad_norm": 3.175762414932251,
+ "learning_rate": 9.714551763776255e-06,
+ "loss": 0.0951,
+ "step": 19224
+ },
+ {
+ "epoch": 52.3841961852861,
+ "grad_norm": 2.3241751194000244,
+ "learning_rate": 9.713669629857894e-06,
+ "loss": 0.0427,
+ "step": 19225
+ },
+ {
+ "epoch": 52.38692098092643,
+ "grad_norm": 10.485630989074707,
+ "learning_rate": 9.712787498169462e-06,
+ "loss": 0.0294,
+ "step": 19226
+ },
+ {
+ "epoch": 52.38964577656676,
+ "grad_norm": 2.814267873764038,
+ "learning_rate": 9.711905368717834e-06,
+ "loss": 0.0781,
+ "step": 19227
+ },
+ {
+ "epoch": 52.392370572207085,
+ "grad_norm": 2.2245802879333496,
+ "learning_rate": 9.711023241509873e-06,
+ "loss": 0.0407,
+ "step": 19228
+ },
+ {
+ "epoch": 52.39509536784741,
+ "grad_norm": 2.747870922088623,
+ "learning_rate": 9.710141116552458e-06,
+ "loss": 0.0989,
+ "step": 19229
+ },
+ {
+ "epoch": 52.39782016348774,
+ "grad_norm": 2.725015640258789,
+ "learning_rate": 9.709258993852453e-06,
+ "loss": 0.1025,
+ "step": 19230
+ },
+ {
+ "epoch": 52.40054495912806,
+ "grad_norm": 2.014695882797241,
+ "learning_rate": 9.708376873416732e-06,
+ "loss": 0.0555,
+ "step": 19231
+ },
+ {
+ "epoch": 52.403269754768395,
+ "grad_norm": 2.4770731925964355,
+ "learning_rate": 9.70749475525216e-06,
+ "loss": 0.0726,
+ "step": 19232
+ },
+ {
+ "epoch": 52.40599455040872,
+ "grad_norm": 2.0316579341888428,
+ "learning_rate": 9.706612639365613e-06,
+ "loss": 0.1185,
+ "step": 19233
+ },
+ {
+ "epoch": 52.40871934604905,
+ "grad_norm": 2.7401022911071777,
+ "learning_rate": 9.705730525763951e-06,
+ "loss": 0.0814,
+ "step": 19234
+ },
+ {
+ "epoch": 52.41144414168937,
+ "grad_norm": 2.2901313304901123,
+ "learning_rate": 9.704848414454057e-06,
+ "loss": 0.0973,
+ "step": 19235
+ },
+ {
+ "epoch": 52.4141689373297,
+ "grad_norm": 2.1307926177978516,
+ "learning_rate": 9.70396630544279e-06,
+ "loss": 0.0398,
+ "step": 19236
+ },
+ {
+ "epoch": 52.416893732970024,
+ "grad_norm": 3.283026933670044,
+ "learning_rate": 9.703084198737028e-06,
+ "loss": 0.0743,
+ "step": 19237
+ },
+ {
+ "epoch": 52.41961852861036,
+ "grad_norm": 2.0139970779418945,
+ "learning_rate": 9.702202094343631e-06,
+ "loss": 0.1025,
+ "step": 19238
+ },
+ {
+ "epoch": 52.42234332425068,
+ "grad_norm": 2.2103466987609863,
+ "learning_rate": 9.701319992269479e-06,
+ "loss": 0.0439,
+ "step": 19239
+ },
+ {
+ "epoch": 52.42506811989101,
+ "grad_norm": 2.4948318004608154,
+ "learning_rate": 9.700437892521434e-06,
+ "loss": 0.1138,
+ "step": 19240
+ },
+ {
+ "epoch": 52.427792915531334,
+ "grad_norm": 1.8803588151931763,
+ "learning_rate": 9.699555795106373e-06,
+ "loss": 0.0356,
+ "step": 19241
+ },
+ {
+ "epoch": 52.43051771117166,
+ "grad_norm": 2.2399439811706543,
+ "learning_rate": 9.698673700031159e-06,
+ "loss": 0.0327,
+ "step": 19242
+ },
+ {
+ "epoch": 52.433242506811986,
+ "grad_norm": 4.303890228271484,
+ "learning_rate": 9.697791607302665e-06,
+ "loss": 0.0564,
+ "step": 19243
+ },
+ {
+ "epoch": 52.43596730245232,
+ "grad_norm": 2.601051092147827,
+ "learning_rate": 9.69690951692776e-06,
+ "loss": 0.0912,
+ "step": 19244
+ },
+ {
+ "epoch": 52.438692098092645,
+ "grad_norm": 2.794412612915039,
+ "learning_rate": 9.69602742891331e-06,
+ "loss": 0.1048,
+ "step": 19245
+ },
+ {
+ "epoch": 52.44141689373297,
+ "grad_norm": 2.9999096393585205,
+ "learning_rate": 9.695145343266195e-06,
+ "loss": 0.0811,
+ "step": 19246
+ },
+ {
+ "epoch": 52.444141689373296,
+ "grad_norm": 1.6846057176589966,
+ "learning_rate": 9.694263259993274e-06,
+ "loss": 0.0296,
+ "step": 19247
+ },
+ {
+ "epoch": 52.44686648501362,
+ "grad_norm": 2.1255738735198975,
+ "learning_rate": 9.693381179101423e-06,
+ "loss": 0.0344,
+ "step": 19248
+ },
+ {
+ "epoch": 52.44959128065395,
+ "grad_norm": 2.832900285720825,
+ "learning_rate": 9.69249910059751e-06,
+ "loss": 0.0773,
+ "step": 19249
+ },
+ {
+ "epoch": 52.45231607629428,
+ "grad_norm": 2.2677133083343506,
+ "learning_rate": 9.691617024488402e-06,
+ "loss": 0.0377,
+ "step": 19250
+ },
+ {
+ "epoch": 52.45504087193461,
+ "grad_norm": 1.9817249774932861,
+ "learning_rate": 9.690734950780973e-06,
+ "loss": 0.0457,
+ "step": 19251
+ },
+ {
+ "epoch": 52.45776566757493,
+ "grad_norm": 5.31567907333374,
+ "learning_rate": 9.689852879482089e-06,
+ "loss": 0.0401,
+ "step": 19252
+ },
+ {
+ "epoch": 52.46049046321526,
+ "grad_norm": 2.4476613998413086,
+ "learning_rate": 9.688970810598622e-06,
+ "loss": 0.0791,
+ "step": 19253
+ },
+ {
+ "epoch": 52.463215258855584,
+ "grad_norm": 1.8942393064498901,
+ "learning_rate": 9.688088744137438e-06,
+ "loss": 0.039,
+ "step": 19254
+ },
+ {
+ "epoch": 52.46594005449591,
+ "grad_norm": 2.6088109016418457,
+ "learning_rate": 9.687206680105412e-06,
+ "loss": 0.034,
+ "step": 19255
+ },
+ {
+ "epoch": 52.46866485013624,
+ "grad_norm": 2.0114853382110596,
+ "learning_rate": 9.686324618509409e-06,
+ "loss": 0.0312,
+ "step": 19256
+ },
+ {
+ "epoch": 52.47138964577657,
+ "grad_norm": 2.95035457611084,
+ "learning_rate": 9.6854425593563e-06,
+ "loss": 0.0417,
+ "step": 19257
+ },
+ {
+ "epoch": 52.474114441416894,
+ "grad_norm": 2.3254363536834717,
+ "learning_rate": 9.684560502652952e-06,
+ "loss": 0.0439,
+ "step": 19258
+ },
+ {
+ "epoch": 52.47683923705722,
+ "grad_norm": 3.6394894123077393,
+ "learning_rate": 9.68367844840624e-06,
+ "loss": 0.0802,
+ "step": 19259
+ },
+ {
+ "epoch": 52.479564032697546,
+ "grad_norm": 2.43530011177063,
+ "learning_rate": 9.682796396623027e-06,
+ "loss": 0.1221,
+ "step": 19260
+ },
+ {
+ "epoch": 52.48228882833787,
+ "grad_norm": 2.033824920654297,
+ "learning_rate": 9.681914347310188e-06,
+ "loss": 0.0307,
+ "step": 19261
+ },
+ {
+ "epoch": 52.485013623978205,
+ "grad_norm": 1.7505159378051758,
+ "learning_rate": 9.681032300474586e-06,
+ "loss": 0.0402,
+ "step": 19262
+ },
+ {
+ "epoch": 52.48773841961853,
+ "grad_norm": 2.9315433502197266,
+ "learning_rate": 9.680150256123096e-06,
+ "loss": 0.0915,
+ "step": 19263
+ },
+ {
+ "epoch": 52.490463215258856,
+ "grad_norm": 3.2886955738067627,
+ "learning_rate": 9.679268214262588e-06,
+ "loss": 0.0789,
+ "step": 19264
+ },
+ {
+ "epoch": 52.49318801089918,
+ "grad_norm": 2.3684067726135254,
+ "learning_rate": 9.678386174899925e-06,
+ "loss": 0.091,
+ "step": 19265
+ },
+ {
+ "epoch": 52.49591280653951,
+ "grad_norm": 2.1321990489959717,
+ "learning_rate": 9.677504138041984e-06,
+ "loss": 0.0397,
+ "step": 19266
+ },
+ {
+ "epoch": 52.49863760217983,
+ "grad_norm": 2.8655195236206055,
+ "learning_rate": 9.676622103695628e-06,
+ "loss": 0.0567,
+ "step": 19267
+ },
+ {
+ "epoch": 52.50136239782017,
+ "grad_norm": 2.2821128368377686,
+ "learning_rate": 9.675740071867729e-06,
+ "loss": 0.1593,
+ "step": 19268
+ },
+ {
+ "epoch": 52.50408719346049,
+ "grad_norm": 2.20466685295105,
+ "learning_rate": 9.674858042565157e-06,
+ "loss": 0.1036,
+ "step": 19269
+ },
+ {
+ "epoch": 52.50681198910082,
+ "grad_norm": 3.9170889854431152,
+ "learning_rate": 9.673976015794778e-06,
+ "loss": 0.0715,
+ "step": 19270
+ },
+ {
+ "epoch": 52.509536784741144,
+ "grad_norm": 3.9013681411743164,
+ "learning_rate": 9.673093991563466e-06,
+ "loss": 0.0786,
+ "step": 19271
+ },
+ {
+ "epoch": 52.51226158038147,
+ "grad_norm": 2.604205369949341,
+ "learning_rate": 9.672211969878085e-06,
+ "loss": 0.2097,
+ "step": 19272
+ },
+ {
+ "epoch": 52.514986376021795,
+ "grad_norm": 3.2574222087860107,
+ "learning_rate": 9.671329950745509e-06,
+ "loss": 0.2703,
+ "step": 19273
+ },
+ {
+ "epoch": 52.51771117166213,
+ "grad_norm": 3.047499895095825,
+ "learning_rate": 9.670447934172602e-06,
+ "loss": 0.1189,
+ "step": 19274
+ },
+ {
+ "epoch": 52.520435967302454,
+ "grad_norm": 2.220416307449341,
+ "learning_rate": 9.669565920166238e-06,
+ "loss": 0.0756,
+ "step": 19275
+ },
+ {
+ "epoch": 52.52316076294278,
+ "grad_norm": 2.433084487915039,
+ "learning_rate": 9.668683908733283e-06,
+ "loss": 0.1039,
+ "step": 19276
+ },
+ {
+ "epoch": 52.525885558583106,
+ "grad_norm": 2.1801440715789795,
+ "learning_rate": 9.667801899880609e-06,
+ "loss": 0.0678,
+ "step": 19277
+ },
+ {
+ "epoch": 52.52861035422343,
+ "grad_norm": 2.2098560333251953,
+ "learning_rate": 9.66691989361508e-06,
+ "loss": 0.0651,
+ "step": 19278
+ },
+ {
+ "epoch": 52.53133514986376,
+ "grad_norm": 2.5681605339050293,
+ "learning_rate": 9.66603788994357e-06,
+ "loss": 0.0584,
+ "step": 19279
+ },
+ {
+ "epoch": 52.53405994550409,
+ "grad_norm": 2.045865774154663,
+ "learning_rate": 9.665155888872943e-06,
+ "loss": 0.048,
+ "step": 19280
+ },
+ {
+ "epoch": 52.536784741144416,
+ "grad_norm": 2.9007716178894043,
+ "learning_rate": 9.664273890410073e-06,
+ "loss": 0.0494,
+ "step": 19281
+ },
+ {
+ "epoch": 52.53950953678474,
+ "grad_norm": 6.613916397094727,
+ "learning_rate": 9.66339189456183e-06,
+ "loss": 0.0595,
+ "step": 19282
+ },
+ {
+ "epoch": 52.54223433242507,
+ "grad_norm": 3.491119861602783,
+ "learning_rate": 9.662509901335077e-06,
+ "loss": 0.0656,
+ "step": 19283
+ },
+ {
+ "epoch": 52.54495912806539,
+ "grad_norm": 2.6432385444641113,
+ "learning_rate": 9.661627910736689e-06,
+ "loss": 0.0775,
+ "step": 19284
+ },
+ {
+ "epoch": 52.54768392370572,
+ "grad_norm": 1.803709626197815,
+ "learning_rate": 9.660745922773529e-06,
+ "loss": 0.0323,
+ "step": 19285
+ },
+ {
+ "epoch": 52.55040871934605,
+ "grad_norm": 2.5865659713745117,
+ "learning_rate": 9.65986393745247e-06,
+ "loss": 0.1088,
+ "step": 19286
+ },
+ {
+ "epoch": 52.55313351498638,
+ "grad_norm": 1.8550461530685425,
+ "learning_rate": 9.65898195478038e-06,
+ "loss": 0.0464,
+ "step": 19287
+ },
+ {
+ "epoch": 52.555858310626704,
+ "grad_norm": 2.7175886631011963,
+ "learning_rate": 9.658099974764129e-06,
+ "loss": 0.0461,
+ "step": 19288
+ },
+ {
+ "epoch": 52.55858310626703,
+ "grad_norm": 3.0431036949157715,
+ "learning_rate": 9.65721799741058e-06,
+ "loss": 0.1297,
+ "step": 19289
+ },
+ {
+ "epoch": 52.561307901907355,
+ "grad_norm": 2.2129993438720703,
+ "learning_rate": 9.656336022726612e-06,
+ "loss": 0.0296,
+ "step": 19290
+ },
+ {
+ "epoch": 52.56403269754768,
+ "grad_norm": 2.1675772666931152,
+ "learning_rate": 9.655454050719081e-06,
+ "loss": 0.0882,
+ "step": 19291
+ },
+ {
+ "epoch": 52.566757493188014,
+ "grad_norm": 1.903780460357666,
+ "learning_rate": 9.654572081394868e-06,
+ "loss": 0.0301,
+ "step": 19292
+ },
+ {
+ "epoch": 52.56948228882834,
+ "grad_norm": 2.7481610774993896,
+ "learning_rate": 9.653690114760835e-06,
+ "loss": 0.0327,
+ "step": 19293
+ },
+ {
+ "epoch": 52.572207084468666,
+ "grad_norm": 2.816952705383301,
+ "learning_rate": 9.65280815082385e-06,
+ "loss": 0.1886,
+ "step": 19294
+ },
+ {
+ "epoch": 52.57493188010899,
+ "grad_norm": 3.3644001483917236,
+ "learning_rate": 9.65192618959079e-06,
+ "loss": 0.1795,
+ "step": 19295
+ },
+ {
+ "epoch": 52.57765667574932,
+ "grad_norm": 2.6442222595214844,
+ "learning_rate": 9.651044231068513e-06,
+ "loss": 0.0494,
+ "step": 19296
+ },
+ {
+ "epoch": 52.58038147138964,
+ "grad_norm": 2.597496271133423,
+ "learning_rate": 9.650162275263889e-06,
+ "loss": 0.1577,
+ "step": 19297
+ },
+ {
+ "epoch": 52.583106267029976,
+ "grad_norm": 1.9127278327941895,
+ "learning_rate": 9.649280322183796e-06,
+ "loss": 0.0352,
+ "step": 19298
+ },
+ {
+ "epoch": 52.5858310626703,
+ "grad_norm": 2.5099709033966064,
+ "learning_rate": 9.648398371835094e-06,
+ "loss": 0.122,
+ "step": 19299
+ },
+ {
+ "epoch": 52.58855585831063,
+ "grad_norm": 2.6633732318878174,
+ "learning_rate": 9.647516424224655e-06,
+ "loss": 0.2206,
+ "step": 19300
+ },
+ {
+ "epoch": 52.59128065395095,
+ "grad_norm": 3.9511332511901855,
+ "learning_rate": 9.646634479359346e-06,
+ "loss": 0.0424,
+ "step": 19301
+ },
+ {
+ "epoch": 52.59400544959128,
+ "grad_norm": 2.2530124187469482,
+ "learning_rate": 9.645752537246038e-06,
+ "loss": 0.044,
+ "step": 19302
+ },
+ {
+ "epoch": 52.596730245231605,
+ "grad_norm": 10.582465171813965,
+ "learning_rate": 9.644870597891597e-06,
+ "loss": 0.0741,
+ "step": 19303
+ },
+ {
+ "epoch": 52.59945504087194,
+ "grad_norm": 2.3031883239746094,
+ "learning_rate": 9.643988661302893e-06,
+ "loss": 0.2154,
+ "step": 19304
+ },
+ {
+ "epoch": 52.60217983651226,
+ "grad_norm": 1.9878026247024536,
+ "learning_rate": 9.643106727486792e-06,
+ "loss": 0.0443,
+ "step": 19305
+ },
+ {
+ "epoch": 52.60490463215259,
+ "grad_norm": 2.056410312652588,
+ "learning_rate": 9.642224796450165e-06,
+ "loss": 0.1996,
+ "step": 19306
+ },
+ {
+ "epoch": 52.607629427792915,
+ "grad_norm": 1.7927830219268799,
+ "learning_rate": 9.64134286819988e-06,
+ "loss": 0.0495,
+ "step": 19307
+ },
+ {
+ "epoch": 52.61035422343324,
+ "grad_norm": 1.971624732017517,
+ "learning_rate": 9.640460942742806e-06,
+ "loss": 0.0288,
+ "step": 19308
+ },
+ {
+ "epoch": 52.61307901907357,
+ "grad_norm": 3.1943275928497314,
+ "learning_rate": 9.63957902008581e-06,
+ "loss": 0.0953,
+ "step": 19309
+ },
+ {
+ "epoch": 52.6158038147139,
+ "grad_norm": 2.75439715385437,
+ "learning_rate": 9.638697100235763e-06,
+ "loss": 0.0659,
+ "step": 19310
+ },
+ {
+ "epoch": 52.618528610354225,
+ "grad_norm": 2.174086570739746,
+ "learning_rate": 9.637815183199527e-06,
+ "loss": 0.0368,
+ "step": 19311
+ },
+ {
+ "epoch": 52.62125340599455,
+ "grad_norm": 1.832973837852478,
+ "learning_rate": 9.636933268983979e-06,
+ "loss": 0.0495,
+ "step": 19312
+ },
+ {
+ "epoch": 52.62397820163488,
+ "grad_norm": 1.5588326454162598,
+ "learning_rate": 9.636051357595979e-06,
+ "loss": 0.0306,
+ "step": 19313
+ },
+ {
+ "epoch": 52.6267029972752,
+ "grad_norm": 9.785394668579102,
+ "learning_rate": 9.635169449042398e-06,
+ "loss": 0.1291,
+ "step": 19314
+ },
+ {
+ "epoch": 52.62942779291553,
+ "grad_norm": 3.3805031776428223,
+ "learning_rate": 9.634287543330109e-06,
+ "loss": 0.0576,
+ "step": 19315
+ },
+ {
+ "epoch": 52.63215258855586,
+ "grad_norm": 6.736478328704834,
+ "learning_rate": 9.633405640465977e-06,
+ "loss": 0.0923,
+ "step": 19316
+ },
+ {
+ "epoch": 52.63487738419619,
+ "grad_norm": 2.068934917449951,
+ "learning_rate": 9.632523740456869e-06,
+ "loss": 0.1337,
+ "step": 19317
+ },
+ {
+ "epoch": 52.63760217983651,
+ "grad_norm": 2.5227227210998535,
+ "learning_rate": 9.631641843309656e-06,
+ "loss": 0.0473,
+ "step": 19318
+ },
+ {
+ "epoch": 52.64032697547684,
+ "grad_norm": 3.760159730911255,
+ "learning_rate": 9.630759949031203e-06,
+ "loss": 0.0677,
+ "step": 19319
+ },
+ {
+ "epoch": 52.643051771117165,
+ "grad_norm": 1.9070627689361572,
+ "learning_rate": 9.629878057628382e-06,
+ "loss": 0.0689,
+ "step": 19320
+ },
+ {
+ "epoch": 52.64577656675749,
+ "grad_norm": 2.063629150390625,
+ "learning_rate": 9.628996169108056e-06,
+ "loss": 0.0542,
+ "step": 19321
+ },
+ {
+ "epoch": 52.64850136239782,
+ "grad_norm": 2.154963731765747,
+ "learning_rate": 9.628114283477098e-06,
+ "loss": 0.0351,
+ "step": 19322
+ },
+ {
+ "epoch": 52.65122615803815,
+ "grad_norm": 4.7306318283081055,
+ "learning_rate": 9.62723240074237e-06,
+ "loss": 0.0659,
+ "step": 19323
+ },
+ {
+ "epoch": 52.653950953678475,
+ "grad_norm": 1.5335769653320312,
+ "learning_rate": 9.62635052091075e-06,
+ "loss": 0.0234,
+ "step": 19324
+ },
+ {
+ "epoch": 52.6566757493188,
+ "grad_norm": 1.9983878135681152,
+ "learning_rate": 9.625468643989096e-06,
+ "loss": 0.0561,
+ "step": 19325
+ },
+ {
+ "epoch": 52.65940054495913,
+ "grad_norm": 1.8781403303146362,
+ "learning_rate": 9.624586769984282e-06,
+ "loss": 0.0389,
+ "step": 19326
+ },
+ {
+ "epoch": 52.66212534059945,
+ "grad_norm": 1.9963499307632446,
+ "learning_rate": 9.623704898903173e-06,
+ "loss": 0.04,
+ "step": 19327
+ },
+ {
+ "epoch": 52.664850136239785,
+ "grad_norm": 3.0735270977020264,
+ "learning_rate": 9.622823030752639e-06,
+ "loss": 0.0627,
+ "step": 19328
+ },
+ {
+ "epoch": 52.66757493188011,
+ "grad_norm": 2.6777608394622803,
+ "learning_rate": 9.621941165539546e-06,
+ "loss": 0.0575,
+ "step": 19329
+ },
+ {
+ "epoch": 52.67029972752044,
+ "grad_norm": 1.7215937376022339,
+ "learning_rate": 9.621059303270765e-06,
+ "loss": 0.046,
+ "step": 19330
+ },
+ {
+ "epoch": 52.67302452316076,
+ "grad_norm": 1.9468967914581299,
+ "learning_rate": 9.620177443953156e-06,
+ "loss": 0.0309,
+ "step": 19331
+ },
+ {
+ "epoch": 52.67574931880109,
+ "grad_norm": 2.1237611770629883,
+ "learning_rate": 9.619295587593597e-06,
+ "loss": 0.0522,
+ "step": 19332
+ },
+ {
+ "epoch": 52.678474114441414,
+ "grad_norm": 3.193106174468994,
+ "learning_rate": 9.61841373419895e-06,
+ "loss": 0.057,
+ "step": 19333
+ },
+ {
+ "epoch": 52.68119891008175,
+ "grad_norm": 2.4807417392730713,
+ "learning_rate": 9.617531883776088e-06,
+ "loss": 0.1045,
+ "step": 19334
+ },
+ {
+ "epoch": 52.68392370572207,
+ "grad_norm": 2.6089866161346436,
+ "learning_rate": 9.616650036331874e-06,
+ "loss": 0.1753,
+ "step": 19335
+ },
+ {
+ "epoch": 52.6866485013624,
+ "grad_norm": 2.3515093326568604,
+ "learning_rate": 9.615768191873179e-06,
+ "loss": 0.1419,
+ "step": 19336
+ },
+ {
+ "epoch": 52.689373297002724,
+ "grad_norm": 2.1598591804504395,
+ "learning_rate": 9.614886350406865e-06,
+ "loss": 0.0374,
+ "step": 19337
+ },
+ {
+ "epoch": 52.69209809264305,
+ "grad_norm": 2.2143988609313965,
+ "learning_rate": 9.614004511939807e-06,
+ "loss": 0.029,
+ "step": 19338
+ },
+ {
+ "epoch": 52.694822888283376,
+ "grad_norm": 2.8116793632507324,
+ "learning_rate": 9.613122676478868e-06,
+ "loss": 0.0551,
+ "step": 19339
+ },
+ {
+ "epoch": 52.69754768392371,
+ "grad_norm": 1.9915086030960083,
+ "learning_rate": 9.612240844030917e-06,
+ "loss": 0.0589,
+ "step": 19340
+ },
+ {
+ "epoch": 52.700272479564035,
+ "grad_norm": 2.1612186431884766,
+ "learning_rate": 9.611359014602823e-06,
+ "loss": 0.054,
+ "step": 19341
+ },
+ {
+ "epoch": 52.70299727520436,
+ "grad_norm": 2.231459617614746,
+ "learning_rate": 9.610477188201452e-06,
+ "loss": 0.0355,
+ "step": 19342
+ },
+ {
+ "epoch": 52.705722070844686,
+ "grad_norm": 2.2654449939727783,
+ "learning_rate": 9.609595364833672e-06,
+ "loss": 0.1337,
+ "step": 19343
+ },
+ {
+ "epoch": 52.70844686648501,
+ "grad_norm": 2.425809621810913,
+ "learning_rate": 9.60871354450635e-06,
+ "loss": 0.0358,
+ "step": 19344
+ },
+ {
+ "epoch": 52.71117166212534,
+ "grad_norm": 1.8000664710998535,
+ "learning_rate": 9.607831727226355e-06,
+ "loss": 0.0501,
+ "step": 19345
+ },
+ {
+ "epoch": 52.71389645776567,
+ "grad_norm": 2.7712178230285645,
+ "learning_rate": 9.606949913000554e-06,
+ "loss": 0.2143,
+ "step": 19346
+ },
+ {
+ "epoch": 52.716621253406,
+ "grad_norm": 2.5158023834228516,
+ "learning_rate": 9.606068101835815e-06,
+ "loss": 0.0619,
+ "step": 19347
+ },
+ {
+ "epoch": 52.71934604904632,
+ "grad_norm": 4.023554801940918,
+ "learning_rate": 9.605186293739e-06,
+ "loss": 0.0694,
+ "step": 19348
+ },
+ {
+ "epoch": 52.72207084468665,
+ "grad_norm": 1.9870562553405762,
+ "learning_rate": 9.604304488716987e-06,
+ "loss": 0.0583,
+ "step": 19349
+ },
+ {
+ "epoch": 52.724795640326974,
+ "grad_norm": 2.538252592086792,
+ "learning_rate": 9.603422686776635e-06,
+ "loss": 0.087,
+ "step": 19350
+ },
+ {
+ "epoch": 52.7275204359673,
+ "grad_norm": 3.5860378742218018,
+ "learning_rate": 9.602540887924819e-06,
+ "loss": 0.0474,
+ "step": 19351
+ },
+ {
+ "epoch": 52.73024523160763,
+ "grad_norm": 2.0790224075317383,
+ "learning_rate": 9.601659092168395e-06,
+ "loss": 0.0247,
+ "step": 19352
+ },
+ {
+ "epoch": 52.73297002724796,
+ "grad_norm": 2.3471126556396484,
+ "learning_rate": 9.600777299514243e-06,
+ "loss": 0.1681,
+ "step": 19353
+ },
+ {
+ "epoch": 52.735694822888284,
+ "grad_norm": 1.808556079864502,
+ "learning_rate": 9.59989550996922e-06,
+ "loss": 0.0265,
+ "step": 19354
+ },
+ {
+ "epoch": 52.73841961852861,
+ "grad_norm": 2.6146914958953857,
+ "learning_rate": 9.599013723540204e-06,
+ "loss": 0.0591,
+ "step": 19355
+ },
+ {
+ "epoch": 52.741144414168936,
+ "grad_norm": 3.145578622817993,
+ "learning_rate": 9.598131940234051e-06,
+ "loss": 0.1195,
+ "step": 19356
+ },
+ {
+ "epoch": 52.74386920980926,
+ "grad_norm": 4.707976341247559,
+ "learning_rate": 9.597250160057635e-06,
+ "loss": 0.0534,
+ "step": 19357
+ },
+ {
+ "epoch": 52.746594005449595,
+ "grad_norm": 2.6301088333129883,
+ "learning_rate": 9.596368383017822e-06,
+ "loss": 0.0376,
+ "step": 19358
+ },
+ {
+ "epoch": 52.74931880108992,
+ "grad_norm": 2.243314266204834,
+ "learning_rate": 9.595486609121477e-06,
+ "loss": 0.237,
+ "step": 19359
+ },
+ {
+ "epoch": 52.752043596730246,
+ "grad_norm": 2.4939401149749756,
+ "learning_rate": 9.594604838375473e-06,
+ "loss": 0.0891,
+ "step": 19360
+ },
+ {
+ "epoch": 52.75476839237057,
+ "grad_norm": 2.2945616245269775,
+ "learning_rate": 9.593723070786671e-06,
+ "loss": 0.0537,
+ "step": 19361
+ },
+ {
+ "epoch": 52.7574931880109,
+ "grad_norm": 3.2083170413970947,
+ "learning_rate": 9.592841306361942e-06,
+ "loss": 0.0898,
+ "step": 19362
+ },
+ {
+ "epoch": 52.76021798365122,
+ "grad_norm": 2.1221439838409424,
+ "learning_rate": 9.59195954510815e-06,
+ "loss": 0.0389,
+ "step": 19363
+ },
+ {
+ "epoch": 52.762942779291556,
+ "grad_norm": 2.3233678340911865,
+ "learning_rate": 9.591077787032166e-06,
+ "loss": 0.1345,
+ "step": 19364
+ },
+ {
+ "epoch": 52.76566757493188,
+ "grad_norm": 5.913759708404541,
+ "learning_rate": 9.590196032140852e-06,
+ "loss": 0.0518,
+ "step": 19365
+ },
+ {
+ "epoch": 52.76839237057221,
+ "grad_norm": 2.5399913787841797,
+ "learning_rate": 9.589314280441077e-06,
+ "loss": 0.0595,
+ "step": 19366
+ },
+ {
+ "epoch": 52.771117166212534,
+ "grad_norm": 1.9807261228561401,
+ "learning_rate": 9.588432531939714e-06,
+ "loss": 0.0357,
+ "step": 19367
+ },
+ {
+ "epoch": 52.77384196185286,
+ "grad_norm": 2.9118566513061523,
+ "learning_rate": 9.587550786643623e-06,
+ "loss": 0.0745,
+ "step": 19368
+ },
+ {
+ "epoch": 52.776566757493185,
+ "grad_norm": 2.643418073654175,
+ "learning_rate": 9.586669044559674e-06,
+ "loss": 0.1461,
+ "step": 19369
+ },
+ {
+ "epoch": 52.77929155313352,
+ "grad_norm": 2.8369996547698975,
+ "learning_rate": 9.585787305694733e-06,
+ "loss": 0.2528,
+ "step": 19370
+ },
+ {
+ "epoch": 52.782016348773844,
+ "grad_norm": 3.773036479949951,
+ "learning_rate": 9.584905570055668e-06,
+ "loss": 0.0887,
+ "step": 19371
+ },
+ {
+ "epoch": 52.78474114441417,
+ "grad_norm": 2.692723512649536,
+ "learning_rate": 9.584023837649343e-06,
+ "loss": 0.037,
+ "step": 19372
+ },
+ {
+ "epoch": 52.787465940054496,
+ "grad_norm": 6.642039775848389,
+ "learning_rate": 9.58314210848263e-06,
+ "loss": 0.0501,
+ "step": 19373
+ },
+ {
+ "epoch": 52.79019073569482,
+ "grad_norm": 3.3806254863739014,
+ "learning_rate": 9.582260382562389e-06,
+ "loss": 0.0501,
+ "step": 19374
+ },
+ {
+ "epoch": 52.79291553133515,
+ "grad_norm": 2.658536434173584,
+ "learning_rate": 9.581378659895494e-06,
+ "loss": 0.0441,
+ "step": 19375
+ },
+ {
+ "epoch": 52.79564032697548,
+ "grad_norm": 2.6577529907226562,
+ "learning_rate": 9.580496940488806e-06,
+ "loss": 0.0681,
+ "step": 19376
+ },
+ {
+ "epoch": 52.798365122615806,
+ "grad_norm": 3.1774730682373047,
+ "learning_rate": 9.579615224349195e-06,
+ "loss": 0.0854,
+ "step": 19377
+ },
+ {
+ "epoch": 52.80108991825613,
+ "grad_norm": 3.15779972076416,
+ "learning_rate": 9.578733511483527e-06,
+ "loss": 0.0383,
+ "step": 19378
+ },
+ {
+ "epoch": 52.80381471389646,
+ "grad_norm": 2.147559642791748,
+ "learning_rate": 9.577851801898669e-06,
+ "loss": 0.0453,
+ "step": 19379
+ },
+ {
+ "epoch": 52.80653950953678,
+ "grad_norm": 3.2170302867889404,
+ "learning_rate": 9.57697009560149e-06,
+ "loss": 0.1302,
+ "step": 19380
+ },
+ {
+ "epoch": 52.80926430517711,
+ "grad_norm": 2.4110541343688965,
+ "learning_rate": 9.57608839259885e-06,
+ "loss": 0.171,
+ "step": 19381
+ },
+ {
+ "epoch": 52.81198910081744,
+ "grad_norm": 2.61696457862854,
+ "learning_rate": 9.575206692897624e-06,
+ "loss": 0.1559,
+ "step": 19382
+ },
+ {
+ "epoch": 52.81471389645777,
+ "grad_norm": 2.558314800262451,
+ "learning_rate": 9.574324996504669e-06,
+ "loss": 0.1021,
+ "step": 19383
+ },
+ {
+ "epoch": 52.817438692098094,
+ "grad_norm": 2.7919373512268066,
+ "learning_rate": 9.573443303426861e-06,
+ "loss": 0.0773,
+ "step": 19384
+ },
+ {
+ "epoch": 52.82016348773842,
+ "grad_norm": 2.63722825050354,
+ "learning_rate": 9.572561613671063e-06,
+ "loss": 0.0522,
+ "step": 19385
+ },
+ {
+ "epoch": 52.822888283378745,
+ "grad_norm": 3.390321731567383,
+ "learning_rate": 9.571679927244141e-06,
+ "loss": 0.2115,
+ "step": 19386
+ },
+ {
+ "epoch": 52.82561307901907,
+ "grad_norm": 3.038709878921509,
+ "learning_rate": 9.570798244152962e-06,
+ "loss": 0.1119,
+ "step": 19387
+ },
+ {
+ "epoch": 52.828337874659404,
+ "grad_norm": 3.642442464828491,
+ "learning_rate": 9.569916564404392e-06,
+ "loss": 0.0485,
+ "step": 19388
+ },
+ {
+ "epoch": 52.83106267029973,
+ "grad_norm": 3.0865373611450195,
+ "learning_rate": 9.5690348880053e-06,
+ "loss": 0.0757,
+ "step": 19389
+ },
+ {
+ "epoch": 52.833787465940055,
+ "grad_norm": 3.050865411758423,
+ "learning_rate": 9.568153214962548e-06,
+ "loss": 0.0471,
+ "step": 19390
+ },
+ {
+ "epoch": 52.83651226158038,
+ "grad_norm": 3.1210060119628906,
+ "learning_rate": 9.567271545283008e-06,
+ "loss": 0.0993,
+ "step": 19391
+ },
+ {
+ "epoch": 52.83923705722071,
+ "grad_norm": 1.968131184577942,
+ "learning_rate": 9.566389878973539e-06,
+ "loss": 0.033,
+ "step": 19392
+ },
+ {
+ "epoch": 52.84196185286103,
+ "grad_norm": 3.950549364089966,
+ "learning_rate": 9.565508216041015e-06,
+ "loss": 0.0591,
+ "step": 19393
+ },
+ {
+ "epoch": 52.844686648501366,
+ "grad_norm": 3.82480788230896,
+ "learning_rate": 9.564626556492297e-06,
+ "loss": 0.0419,
+ "step": 19394
+ },
+ {
+ "epoch": 52.84741144414169,
+ "grad_norm": 1.791062355041504,
+ "learning_rate": 9.563744900334256e-06,
+ "loss": 0.0722,
+ "step": 19395
+ },
+ {
+ "epoch": 52.85013623978202,
+ "grad_norm": 2.334421157836914,
+ "learning_rate": 9.562863247573753e-06,
+ "loss": 0.041,
+ "step": 19396
+ },
+ {
+ "epoch": 52.85286103542234,
+ "grad_norm": 2.2769012451171875,
+ "learning_rate": 9.561981598217658e-06,
+ "loss": 0.1431,
+ "step": 19397
+ },
+ {
+ "epoch": 52.85558583106267,
+ "grad_norm": 3.0323326587677,
+ "learning_rate": 9.561099952272835e-06,
+ "loss": 0.0515,
+ "step": 19398
+ },
+ {
+ "epoch": 52.858310626702995,
+ "grad_norm": 2.855241060256958,
+ "learning_rate": 9.560218309746154e-06,
+ "loss": 0.1106,
+ "step": 19399
+ },
+ {
+ "epoch": 52.86103542234333,
+ "grad_norm": 2.0660223960876465,
+ "learning_rate": 9.559336670644472e-06,
+ "loss": 0.1342,
+ "step": 19400
+ },
+ {
+ "epoch": 52.86376021798365,
+ "grad_norm": 3.176835298538208,
+ "learning_rate": 9.558455034974667e-06,
+ "loss": 0.0467,
+ "step": 19401
+ },
+ {
+ "epoch": 52.86648501362398,
+ "grad_norm": 2.4615352153778076,
+ "learning_rate": 9.5575734027436e-06,
+ "loss": 0.134,
+ "step": 19402
+ },
+ {
+ "epoch": 52.869209809264305,
+ "grad_norm": 3.205413818359375,
+ "learning_rate": 9.556691773958137e-06,
+ "loss": 0.0649,
+ "step": 19403
+ },
+ {
+ "epoch": 52.87193460490463,
+ "grad_norm": 2.200235366821289,
+ "learning_rate": 9.555810148625144e-06,
+ "loss": 0.0928,
+ "step": 19404
+ },
+ {
+ "epoch": 52.87465940054496,
+ "grad_norm": 2.5695762634277344,
+ "learning_rate": 9.55492852675149e-06,
+ "loss": 0.0965,
+ "step": 19405
+ },
+ {
+ "epoch": 52.87738419618529,
+ "grad_norm": 2.558344841003418,
+ "learning_rate": 9.554046908344034e-06,
+ "loss": 0.039,
+ "step": 19406
+ },
+ {
+ "epoch": 52.880108991825615,
+ "grad_norm": 4.443081378936768,
+ "learning_rate": 9.55316529340965e-06,
+ "loss": 0.1303,
+ "step": 19407
+ },
+ {
+ "epoch": 52.88283378746594,
+ "grad_norm": 2.771044969558716,
+ "learning_rate": 9.552283681955196e-06,
+ "loss": 0.1022,
+ "step": 19408
+ },
+ {
+ "epoch": 52.88555858310627,
+ "grad_norm": 1.850396752357483,
+ "learning_rate": 9.551402073987547e-06,
+ "loss": 0.0653,
+ "step": 19409
+ },
+ {
+ "epoch": 52.88828337874659,
+ "grad_norm": 2.418403387069702,
+ "learning_rate": 9.55052046951356e-06,
+ "loss": 0.0955,
+ "step": 19410
+ },
+ {
+ "epoch": 52.89100817438692,
+ "grad_norm": 2.2210841178894043,
+ "learning_rate": 9.549638868540108e-06,
+ "loss": 0.0701,
+ "step": 19411
+ },
+ {
+ "epoch": 52.89373297002725,
+ "grad_norm": 4.8933587074279785,
+ "learning_rate": 9.548757271074053e-06,
+ "loss": 0.0835,
+ "step": 19412
+ },
+ {
+ "epoch": 52.89645776566758,
+ "grad_norm": 3.6658525466918945,
+ "learning_rate": 9.547875677122262e-06,
+ "loss": 0.1006,
+ "step": 19413
+ },
+ {
+ "epoch": 52.8991825613079,
+ "grad_norm": 2.0841684341430664,
+ "learning_rate": 9.5469940866916e-06,
+ "loss": 0.0616,
+ "step": 19414
+ },
+ {
+ "epoch": 52.90190735694823,
+ "grad_norm": 1.823482632637024,
+ "learning_rate": 9.546112499788935e-06,
+ "loss": 0.0412,
+ "step": 19415
+ },
+ {
+ "epoch": 52.904632152588555,
+ "grad_norm": 2.523869037628174,
+ "learning_rate": 9.545230916421129e-06,
+ "loss": 0.1037,
+ "step": 19416
+ },
+ {
+ "epoch": 52.90735694822888,
+ "grad_norm": 2.898244619369507,
+ "learning_rate": 9.544349336595049e-06,
+ "loss": 0.0717,
+ "step": 19417
+ },
+ {
+ "epoch": 52.91008174386921,
+ "grad_norm": 3.119903326034546,
+ "learning_rate": 9.543467760317565e-06,
+ "loss": 0.0832,
+ "step": 19418
+ },
+ {
+ "epoch": 52.91280653950954,
+ "grad_norm": 1.9957491159439087,
+ "learning_rate": 9.542586187595537e-06,
+ "loss": 0.0644,
+ "step": 19419
+ },
+ {
+ "epoch": 52.915531335149865,
+ "grad_norm": 2.9185380935668945,
+ "learning_rate": 9.541704618435834e-06,
+ "loss": 0.0926,
+ "step": 19420
+ },
+ {
+ "epoch": 52.91825613079019,
+ "grad_norm": 2.425431251525879,
+ "learning_rate": 9.540823052845323e-06,
+ "loss": 0.0538,
+ "step": 19421
+ },
+ {
+ "epoch": 52.920980926430516,
+ "grad_norm": 2.729757785797119,
+ "learning_rate": 9.539941490830863e-06,
+ "loss": 0.1292,
+ "step": 19422
+ },
+ {
+ "epoch": 52.92370572207084,
+ "grad_norm": 2.278898000717163,
+ "learning_rate": 9.539059932399328e-06,
+ "loss": 0.0491,
+ "step": 19423
+ },
+ {
+ "epoch": 52.926430517711175,
+ "grad_norm": 2.702746629714966,
+ "learning_rate": 9.538178377557577e-06,
+ "loss": 0.1066,
+ "step": 19424
+ },
+ {
+ "epoch": 52.9291553133515,
+ "grad_norm": 2.5572240352630615,
+ "learning_rate": 9.53729682631248e-06,
+ "loss": 0.0478,
+ "step": 19425
+ },
+ {
+ "epoch": 52.93188010899183,
+ "grad_norm": 3.3699727058410645,
+ "learning_rate": 9.536415278670897e-06,
+ "loss": 0.0458,
+ "step": 19426
+ },
+ {
+ "epoch": 52.93460490463215,
+ "grad_norm": 2.702944278717041,
+ "learning_rate": 9.5355337346397e-06,
+ "loss": 0.0991,
+ "step": 19427
+ },
+ {
+ "epoch": 52.93732970027248,
+ "grad_norm": 2.2959651947021484,
+ "learning_rate": 9.53465219422575e-06,
+ "loss": 0.1322,
+ "step": 19428
+ },
+ {
+ "epoch": 52.940054495912804,
+ "grad_norm": 1.9163693189620972,
+ "learning_rate": 9.533770657435914e-06,
+ "loss": 0.0387,
+ "step": 19429
+ },
+ {
+ "epoch": 52.94277929155314,
+ "grad_norm": 2.8302838802337646,
+ "learning_rate": 9.532889124277057e-06,
+ "loss": 0.2037,
+ "step": 19430
+ },
+ {
+ "epoch": 52.94550408719346,
+ "grad_norm": 2.6316425800323486,
+ "learning_rate": 9.532007594756046e-06,
+ "loss": 0.0631,
+ "step": 19431
+ },
+ {
+ "epoch": 52.94822888283379,
+ "grad_norm": 3.767286777496338,
+ "learning_rate": 9.531126068879742e-06,
+ "loss": 0.0579,
+ "step": 19432
+ },
+ {
+ "epoch": 52.950953678474114,
+ "grad_norm": 2.430703639984131,
+ "learning_rate": 9.530244546655016e-06,
+ "loss": 0.1008,
+ "step": 19433
+ },
+ {
+ "epoch": 52.95367847411444,
+ "grad_norm": 2.731851816177368,
+ "learning_rate": 9.529363028088725e-06,
+ "loss": 0.034,
+ "step": 19434
+ },
+ {
+ "epoch": 52.956403269754766,
+ "grad_norm": 2.8577821254730225,
+ "learning_rate": 9.528481513187744e-06,
+ "loss": 0.082,
+ "step": 19435
+ },
+ {
+ "epoch": 52.95912806539509,
+ "grad_norm": 2.8610217571258545,
+ "learning_rate": 9.527600001958933e-06,
+ "loss": 0.097,
+ "step": 19436
+ },
+ {
+ "epoch": 52.961852861035425,
+ "grad_norm": 6.426231384277344,
+ "learning_rate": 9.526718494409158e-06,
+ "loss": 0.0802,
+ "step": 19437
+ },
+ {
+ "epoch": 52.96457765667575,
+ "grad_norm": 1.8382799625396729,
+ "learning_rate": 9.525836990545283e-06,
+ "loss": 0.0384,
+ "step": 19438
+ },
+ {
+ "epoch": 52.967302452316076,
+ "grad_norm": 4.294992923736572,
+ "learning_rate": 9.524955490374176e-06,
+ "loss": 0.1134,
+ "step": 19439
+ },
+ {
+ "epoch": 52.9700272479564,
+ "grad_norm": 1.6766995191574097,
+ "learning_rate": 9.5240739939027e-06,
+ "loss": 0.056,
+ "step": 19440
+ },
+ {
+ "epoch": 52.97275204359673,
+ "grad_norm": 2.8755221366882324,
+ "learning_rate": 9.52319250113772e-06,
+ "loss": 0.0611,
+ "step": 19441
+ },
+ {
+ "epoch": 52.97547683923706,
+ "grad_norm": 1.90351402759552,
+ "learning_rate": 9.522311012086103e-06,
+ "loss": 0.0435,
+ "step": 19442
+ },
+ {
+ "epoch": 52.97820163487739,
+ "grad_norm": 3.242047071456909,
+ "learning_rate": 9.52142952675471e-06,
+ "loss": 0.0943,
+ "step": 19443
+ },
+ {
+ "epoch": 52.98092643051771,
+ "grad_norm": 3.10324764251709,
+ "learning_rate": 9.52054804515041e-06,
+ "loss": 0.1195,
+ "step": 19444
+ },
+ {
+ "epoch": 52.98365122615804,
+ "grad_norm": 3.225119113922119,
+ "learning_rate": 9.519666567280067e-06,
+ "loss": 0.0838,
+ "step": 19445
+ },
+ {
+ "epoch": 52.986376021798364,
+ "grad_norm": 2.6599884033203125,
+ "learning_rate": 9.518785093150543e-06,
+ "loss": 0.0553,
+ "step": 19446
+ },
+ {
+ "epoch": 52.98910081743869,
+ "grad_norm": 1.7592592239379883,
+ "learning_rate": 9.517903622768707e-06,
+ "loss": 0.0597,
+ "step": 19447
+ },
+ {
+ "epoch": 52.991825613079016,
+ "grad_norm": 2.88277268409729,
+ "learning_rate": 9.51702215614142e-06,
+ "loss": 0.0543,
+ "step": 19448
+ },
+ {
+ "epoch": 52.99455040871935,
+ "grad_norm": 2.2909984588623047,
+ "learning_rate": 9.516140693275554e-06,
+ "loss": 0.0636,
+ "step": 19449
+ },
+ {
+ "epoch": 52.997275204359674,
+ "grad_norm": 2.463326930999756,
+ "learning_rate": 9.515259234177964e-06,
+ "loss": 0.0691,
+ "step": 19450
+ },
+ {
+ "epoch": 53.0,
+ "grad_norm": 2.5258514881134033,
+ "learning_rate": 9.514377778855521e-06,
+ "loss": 0.1491,
+ "step": 19451
+ },
+ {
+ "epoch": 53.002724795640326,
+ "grad_norm": 6.15228796005249,
+ "learning_rate": 9.513496327315085e-06,
+ "loss": 0.0429,
+ "step": 19452
+ },
+ {
+ "epoch": 53.00544959128065,
+ "grad_norm": 1.8393267393112183,
+ "learning_rate": 9.512614879563525e-06,
+ "loss": 0.029,
+ "step": 19453
+ },
+ {
+ "epoch": 53.00817438692098,
+ "grad_norm": 2.703932046890259,
+ "learning_rate": 9.511733435607709e-06,
+ "loss": 0.1341,
+ "step": 19454
+ },
+ {
+ "epoch": 53.01089918256131,
+ "grad_norm": 4.04781436920166,
+ "learning_rate": 9.510851995454494e-06,
+ "loss": 0.1524,
+ "step": 19455
+ },
+ {
+ "epoch": 53.013623978201636,
+ "grad_norm": 2.2447667121887207,
+ "learning_rate": 9.50997055911075e-06,
+ "loss": 0.0328,
+ "step": 19456
+ },
+ {
+ "epoch": 53.01634877384196,
+ "grad_norm": 2.4266114234924316,
+ "learning_rate": 9.509089126583336e-06,
+ "loss": 0.0559,
+ "step": 19457
+ },
+ {
+ "epoch": 53.01907356948229,
+ "grad_norm": 2.132646322250366,
+ "learning_rate": 9.508207697879123e-06,
+ "loss": 0.0398,
+ "step": 19458
+ },
+ {
+ "epoch": 53.02179836512261,
+ "grad_norm": 2.411353826522827,
+ "learning_rate": 9.507326273004972e-06,
+ "loss": 0.039,
+ "step": 19459
+ },
+ {
+ "epoch": 53.02452316076294,
+ "grad_norm": 3.0257623195648193,
+ "learning_rate": 9.50644485196775e-06,
+ "loss": 0.0871,
+ "step": 19460
+ },
+ {
+ "epoch": 53.02724795640327,
+ "grad_norm": 2.136542797088623,
+ "learning_rate": 9.505563434774317e-06,
+ "loss": 0.0687,
+ "step": 19461
+ },
+ {
+ "epoch": 53.0299727520436,
+ "grad_norm": 2.796483039855957,
+ "learning_rate": 9.504682021431541e-06,
+ "loss": 0.0633,
+ "step": 19462
+ },
+ {
+ "epoch": 53.032697547683924,
+ "grad_norm": 2.163130283355713,
+ "learning_rate": 9.503800611946285e-06,
+ "loss": 0.0376,
+ "step": 19463
+ },
+ {
+ "epoch": 53.03542234332425,
+ "grad_norm": 2.2654638290405273,
+ "learning_rate": 9.502919206325415e-06,
+ "loss": 0.1904,
+ "step": 19464
+ },
+ {
+ "epoch": 53.038147138964575,
+ "grad_norm": 4.510808944702148,
+ "learning_rate": 9.502037804575793e-06,
+ "loss": 0.1469,
+ "step": 19465
+ },
+ {
+ "epoch": 53.0408719346049,
+ "grad_norm": 3.2492096424102783,
+ "learning_rate": 9.501156406704284e-06,
+ "loss": 0.1671,
+ "step": 19466
+ },
+ {
+ "epoch": 53.043596730245234,
+ "grad_norm": 2.58743953704834,
+ "learning_rate": 9.500275012717754e-06,
+ "loss": 0.0747,
+ "step": 19467
+ },
+ {
+ "epoch": 53.04632152588556,
+ "grad_norm": 2.801464080810547,
+ "learning_rate": 9.499393622623064e-06,
+ "loss": 0.1782,
+ "step": 19468
+ },
+ {
+ "epoch": 53.049046321525886,
+ "grad_norm": 2.2837560176849365,
+ "learning_rate": 9.49851223642708e-06,
+ "loss": 0.152,
+ "step": 19469
+ },
+ {
+ "epoch": 53.05177111716621,
+ "grad_norm": 2.5508511066436768,
+ "learning_rate": 9.49763085413667e-06,
+ "loss": 0.11,
+ "step": 19470
+ },
+ {
+ "epoch": 53.05449591280654,
+ "grad_norm": 1.992436408996582,
+ "learning_rate": 9.496749475758692e-06,
+ "loss": 0.1398,
+ "step": 19471
+ },
+ {
+ "epoch": 53.05722070844686,
+ "grad_norm": 1.9695086479187012,
+ "learning_rate": 9.495868101300015e-06,
+ "loss": 0.0594,
+ "step": 19472
+ },
+ {
+ "epoch": 53.059945504087196,
+ "grad_norm": 1.651104211807251,
+ "learning_rate": 9.4949867307675e-06,
+ "loss": 0.1034,
+ "step": 19473
+ },
+ {
+ "epoch": 53.06267029972752,
+ "grad_norm": 4.110095500946045,
+ "learning_rate": 9.494105364168014e-06,
+ "loss": 0.2036,
+ "step": 19474
+ },
+ {
+ "epoch": 53.06539509536785,
+ "grad_norm": 2.595247745513916,
+ "learning_rate": 9.493224001508416e-06,
+ "loss": 0.0889,
+ "step": 19475
+ },
+ {
+ "epoch": 53.06811989100817,
+ "grad_norm": 2.352219820022583,
+ "learning_rate": 9.492342642795576e-06,
+ "loss": 0.0355,
+ "step": 19476
+ },
+ {
+ "epoch": 53.0708446866485,
+ "grad_norm": 3.922657012939453,
+ "learning_rate": 9.491461288036353e-06,
+ "loss": 0.0803,
+ "step": 19477
+ },
+ {
+ "epoch": 53.073569482288825,
+ "grad_norm": 2.179656744003296,
+ "learning_rate": 9.490579937237615e-06,
+ "loss": 0.1503,
+ "step": 19478
+ },
+ {
+ "epoch": 53.07629427792916,
+ "grad_norm": 2.2736284732818604,
+ "learning_rate": 9.489698590406223e-06,
+ "loss": 0.1598,
+ "step": 19479
+ },
+ {
+ "epoch": 53.079019073569484,
+ "grad_norm": 3.040252447128296,
+ "learning_rate": 9.488817247549043e-06,
+ "loss": 0.0383,
+ "step": 19480
+ },
+ {
+ "epoch": 53.08174386920981,
+ "grad_norm": 2.3655965328216553,
+ "learning_rate": 9.487935908672936e-06,
+ "loss": 0.0534,
+ "step": 19481
+ },
+ {
+ "epoch": 53.084468664850135,
+ "grad_norm": 2.184185266494751,
+ "learning_rate": 9.487054573784769e-06,
+ "loss": 0.1689,
+ "step": 19482
+ },
+ {
+ "epoch": 53.08719346049046,
+ "grad_norm": 1.6504507064819336,
+ "learning_rate": 9.486173242891404e-06,
+ "loss": 0.0482,
+ "step": 19483
+ },
+ {
+ "epoch": 53.08991825613079,
+ "grad_norm": 2.5911097526550293,
+ "learning_rate": 9.485291915999706e-06,
+ "loss": 0.0452,
+ "step": 19484
+ },
+ {
+ "epoch": 53.09264305177112,
+ "grad_norm": 2.6252996921539307,
+ "learning_rate": 9.484410593116536e-06,
+ "loss": 0.1137,
+ "step": 19485
+ },
+ {
+ "epoch": 53.095367847411445,
+ "grad_norm": 2.607675790786743,
+ "learning_rate": 9.483529274248758e-06,
+ "loss": 0.0793,
+ "step": 19486
+ },
+ {
+ "epoch": 53.09809264305177,
+ "grad_norm": 2.5496973991394043,
+ "learning_rate": 9.48264795940324e-06,
+ "loss": 0.0487,
+ "step": 19487
+ },
+ {
+ "epoch": 53.1008174386921,
+ "grad_norm": 2.3485803604125977,
+ "learning_rate": 9.481766648586844e-06,
+ "loss": 0.1324,
+ "step": 19488
+ },
+ {
+ "epoch": 53.10354223433242,
+ "grad_norm": 2.2632219791412354,
+ "learning_rate": 9.48088534180643e-06,
+ "loss": 0.0476,
+ "step": 19489
+ },
+ {
+ "epoch": 53.10626702997275,
+ "grad_norm": 2.426309823989868,
+ "learning_rate": 9.480004039068868e-06,
+ "loss": 0.0835,
+ "step": 19490
+ },
+ {
+ "epoch": 53.10899182561308,
+ "grad_norm": 1.9662997722625732,
+ "learning_rate": 9.479122740381015e-06,
+ "loss": 0.056,
+ "step": 19491
+ },
+ {
+ "epoch": 53.11171662125341,
+ "grad_norm": 1.6628806591033936,
+ "learning_rate": 9.478241445749739e-06,
+ "loss": 0.0387,
+ "step": 19492
+ },
+ {
+ "epoch": 53.11444141689373,
+ "grad_norm": 4.7273335456848145,
+ "learning_rate": 9.477360155181902e-06,
+ "loss": 0.1362,
+ "step": 19493
+ },
+ {
+ "epoch": 53.11716621253406,
+ "grad_norm": 2.368149518966675,
+ "learning_rate": 9.476478868684366e-06,
+ "loss": 0.0693,
+ "step": 19494
+ },
+ {
+ "epoch": 53.119891008174385,
+ "grad_norm": 2.8609259128570557,
+ "learning_rate": 9.475597586263996e-06,
+ "loss": 0.0762,
+ "step": 19495
+ },
+ {
+ "epoch": 53.12261580381471,
+ "grad_norm": 1.6861741542816162,
+ "learning_rate": 9.474716307927657e-06,
+ "loss": 0.0322,
+ "step": 19496
+ },
+ {
+ "epoch": 53.12534059945504,
+ "grad_norm": 2.4989328384399414,
+ "learning_rate": 9.473835033682209e-06,
+ "loss": 0.0557,
+ "step": 19497
+ },
+ {
+ "epoch": 53.12806539509537,
+ "grad_norm": 2.4875001907348633,
+ "learning_rate": 9.472953763534517e-06,
+ "loss": 0.0328,
+ "step": 19498
+ },
+ {
+ "epoch": 53.130790190735695,
+ "grad_norm": 2.287020206451416,
+ "learning_rate": 9.472072497491443e-06,
+ "loss": 0.032,
+ "step": 19499
+ },
+ {
+ "epoch": 53.13351498637602,
+ "grad_norm": 2.457498550415039,
+ "learning_rate": 9.471191235559853e-06,
+ "loss": 0.0453,
+ "step": 19500
+ },
+ {
+ "epoch": 53.13623978201635,
+ "grad_norm": 2.3089396953582764,
+ "learning_rate": 9.470309977746607e-06,
+ "loss": 0.0903,
+ "step": 19501
+ },
+ {
+ "epoch": 53.13896457765667,
+ "grad_norm": 2.8926310539245605,
+ "learning_rate": 9.46942872405857e-06,
+ "loss": 0.1172,
+ "step": 19502
+ },
+ {
+ "epoch": 53.141689373297005,
+ "grad_norm": 1.780134677886963,
+ "learning_rate": 9.468547474502603e-06,
+ "loss": 0.083,
+ "step": 19503
+ },
+ {
+ "epoch": 53.14441416893733,
+ "grad_norm": 2.4300730228424072,
+ "learning_rate": 9.467666229085574e-06,
+ "loss": 0.0567,
+ "step": 19504
+ },
+ {
+ "epoch": 53.14713896457766,
+ "grad_norm": 1.7408441305160522,
+ "learning_rate": 9.466784987814345e-06,
+ "loss": 0.0695,
+ "step": 19505
+ },
+ {
+ "epoch": 53.14986376021798,
+ "grad_norm": 2.026883840560913,
+ "learning_rate": 9.465903750695774e-06,
+ "loss": 0.0336,
+ "step": 19506
+ },
+ {
+ "epoch": 53.15258855585831,
+ "grad_norm": 2.888192892074585,
+ "learning_rate": 9.465022517736728e-06,
+ "loss": 0.0411,
+ "step": 19507
+ },
+ {
+ "epoch": 53.155313351498634,
+ "grad_norm": 3.0095126628875732,
+ "learning_rate": 9.464141288944072e-06,
+ "loss": 0.0873,
+ "step": 19508
+ },
+ {
+ "epoch": 53.15803814713897,
+ "grad_norm": 4.436251640319824,
+ "learning_rate": 9.463260064324663e-06,
+ "loss": 0.0525,
+ "step": 19509
+ },
+ {
+ "epoch": 53.16076294277929,
+ "grad_norm": 1.611110806465149,
+ "learning_rate": 9.46237884388537e-06,
+ "loss": 0.0258,
+ "step": 19510
+ },
+ {
+ "epoch": 53.16348773841962,
+ "grad_norm": 2.8804755210876465,
+ "learning_rate": 9.461497627633051e-06,
+ "loss": 0.1074,
+ "step": 19511
+ },
+ {
+ "epoch": 53.166212534059945,
+ "grad_norm": 2.9073824882507324,
+ "learning_rate": 9.460616415574575e-06,
+ "loss": 0.1224,
+ "step": 19512
+ },
+ {
+ "epoch": 53.16893732970027,
+ "grad_norm": 1.96123206615448,
+ "learning_rate": 9.459735207716796e-06,
+ "loss": 0.0364,
+ "step": 19513
+ },
+ {
+ "epoch": 53.171662125340596,
+ "grad_norm": 1.8982599973678589,
+ "learning_rate": 9.458854004066586e-06,
+ "loss": 0.0234,
+ "step": 19514
+ },
+ {
+ "epoch": 53.17438692098093,
+ "grad_norm": 1.8885579109191895,
+ "learning_rate": 9.457972804630801e-06,
+ "loss": 0.0434,
+ "step": 19515
+ },
+ {
+ "epoch": 53.177111716621255,
+ "grad_norm": 2.0053417682647705,
+ "learning_rate": 9.457091609416309e-06,
+ "loss": 0.0326,
+ "step": 19516
+ },
+ {
+ "epoch": 53.17983651226158,
+ "grad_norm": 2.3478732109069824,
+ "learning_rate": 9.456210418429968e-06,
+ "loss": 0.1199,
+ "step": 19517
+ },
+ {
+ "epoch": 53.182561307901906,
+ "grad_norm": 2.9201512336730957,
+ "learning_rate": 9.455329231678644e-06,
+ "loss": 0.1521,
+ "step": 19518
+ },
+ {
+ "epoch": 53.18528610354223,
+ "grad_norm": 2.207432985305786,
+ "learning_rate": 9.454448049169195e-06,
+ "loss": 0.1552,
+ "step": 19519
+ },
+ {
+ "epoch": 53.18801089918256,
+ "grad_norm": 1.9007363319396973,
+ "learning_rate": 9.453566870908488e-06,
+ "loss": 0.1594,
+ "step": 19520
+ },
+ {
+ "epoch": 53.19073569482289,
+ "grad_norm": 2.4501452445983887,
+ "learning_rate": 9.452685696903387e-06,
+ "loss": 0.1055,
+ "step": 19521
+ },
+ {
+ "epoch": 53.19346049046322,
+ "grad_norm": 1.5193750858306885,
+ "learning_rate": 9.45180452716075e-06,
+ "loss": 0.0954,
+ "step": 19522
+ },
+ {
+ "epoch": 53.19618528610354,
+ "grad_norm": 2.426198720932007,
+ "learning_rate": 9.450923361687443e-06,
+ "loss": 0.0499,
+ "step": 19523
+ },
+ {
+ "epoch": 53.19891008174387,
+ "grad_norm": 2.297342538833618,
+ "learning_rate": 9.450042200490328e-06,
+ "loss": 0.1206,
+ "step": 19524
+ },
+ {
+ "epoch": 53.201634877384194,
+ "grad_norm": 1.7860106229782104,
+ "learning_rate": 9.449161043576266e-06,
+ "loss": 0.091,
+ "step": 19525
+ },
+ {
+ "epoch": 53.20435967302452,
+ "grad_norm": 2.456651210784912,
+ "learning_rate": 9.44827989095212e-06,
+ "loss": 0.1043,
+ "step": 19526
+ },
+ {
+ "epoch": 53.20708446866485,
+ "grad_norm": 1.9694178104400635,
+ "learning_rate": 9.447398742624754e-06,
+ "loss": 0.0499,
+ "step": 19527
+ },
+ {
+ "epoch": 53.20980926430518,
+ "grad_norm": 2.226588010787964,
+ "learning_rate": 9.446517598601026e-06,
+ "loss": 0.1135,
+ "step": 19528
+ },
+ {
+ "epoch": 53.212534059945504,
+ "grad_norm": 2.2941181659698486,
+ "learning_rate": 9.445636458887804e-06,
+ "loss": 0.118,
+ "step": 19529
+ },
+ {
+ "epoch": 53.21525885558583,
+ "grad_norm": 1.9590528011322021,
+ "learning_rate": 9.444755323491944e-06,
+ "loss": 0.0783,
+ "step": 19530
+ },
+ {
+ "epoch": 53.217983651226156,
+ "grad_norm": 2.7270402908325195,
+ "learning_rate": 9.443874192420312e-06,
+ "loss": 0.1329,
+ "step": 19531
+ },
+ {
+ "epoch": 53.22070844686648,
+ "grad_norm": 1.3562333583831787,
+ "learning_rate": 9.442993065679773e-06,
+ "loss": 0.0202,
+ "step": 19532
+ },
+ {
+ "epoch": 53.223433242506815,
+ "grad_norm": 2.1392340660095215,
+ "learning_rate": 9.442111943277183e-06,
+ "loss": 0.1319,
+ "step": 19533
+ },
+ {
+ "epoch": 53.22615803814714,
+ "grad_norm": 2.4832215309143066,
+ "learning_rate": 9.44123082521941e-06,
+ "loss": 0.0395,
+ "step": 19534
+ },
+ {
+ "epoch": 53.228882833787466,
+ "grad_norm": 1.3895155191421509,
+ "learning_rate": 9.440349711513311e-06,
+ "loss": 0.0192,
+ "step": 19535
+ },
+ {
+ "epoch": 53.23160762942779,
+ "grad_norm": 2.8254878520965576,
+ "learning_rate": 9.439468602165752e-06,
+ "loss": 0.0882,
+ "step": 19536
+ },
+ {
+ "epoch": 53.23433242506812,
+ "grad_norm": 1.8659132719039917,
+ "learning_rate": 9.438587497183592e-06,
+ "loss": 0.0558,
+ "step": 19537
+ },
+ {
+ "epoch": 53.237057220708444,
+ "grad_norm": 2.0751688480377197,
+ "learning_rate": 9.437706396573693e-06,
+ "loss": 0.0475,
+ "step": 19538
+ },
+ {
+ "epoch": 53.23978201634878,
+ "grad_norm": 2.2299113273620605,
+ "learning_rate": 9.43682530034292e-06,
+ "loss": 0.1478,
+ "step": 19539
+ },
+ {
+ "epoch": 53.2425068119891,
+ "grad_norm": 3.5168585777282715,
+ "learning_rate": 9.435944208498135e-06,
+ "loss": 0.1875,
+ "step": 19540
+ },
+ {
+ "epoch": 53.24523160762943,
+ "grad_norm": 2.3008124828338623,
+ "learning_rate": 9.435063121046198e-06,
+ "loss": 0.0962,
+ "step": 19541
+ },
+ {
+ "epoch": 53.247956403269754,
+ "grad_norm": 2.2701330184936523,
+ "learning_rate": 9.43418203799397e-06,
+ "loss": 0.029,
+ "step": 19542
+ },
+ {
+ "epoch": 53.25068119891008,
+ "grad_norm": 1.764454960823059,
+ "learning_rate": 9.433300959348317e-06,
+ "loss": 0.0874,
+ "step": 19543
+ },
+ {
+ "epoch": 53.253405994550405,
+ "grad_norm": 2.305368185043335,
+ "learning_rate": 9.432419885116094e-06,
+ "loss": 0.0583,
+ "step": 19544
+ },
+ {
+ "epoch": 53.25613079019074,
+ "grad_norm": 2.0332911014556885,
+ "learning_rate": 9.43153881530417e-06,
+ "loss": 0.0284,
+ "step": 19545
+ },
+ {
+ "epoch": 53.258855585831064,
+ "grad_norm": 2.030040979385376,
+ "learning_rate": 9.4306577499194e-06,
+ "loss": 0.0566,
+ "step": 19546
+ },
+ {
+ "epoch": 53.26158038147139,
+ "grad_norm": 1.6853327751159668,
+ "learning_rate": 9.429776688968654e-06,
+ "loss": 0.0354,
+ "step": 19547
+ },
+ {
+ "epoch": 53.264305177111716,
+ "grad_norm": 2.1725730895996094,
+ "learning_rate": 9.428895632458785e-06,
+ "loss": 0.1183,
+ "step": 19548
+ },
+ {
+ "epoch": 53.26702997275204,
+ "grad_norm": 2.700989007949829,
+ "learning_rate": 9.428014580396659e-06,
+ "loss": 0.0884,
+ "step": 19549
+ },
+ {
+ "epoch": 53.26975476839237,
+ "grad_norm": 1.9754804372787476,
+ "learning_rate": 9.427133532789137e-06,
+ "loss": 0.0847,
+ "step": 19550
+ },
+ {
+ "epoch": 53.2724795640327,
+ "grad_norm": 2.2669334411621094,
+ "learning_rate": 9.426252489643082e-06,
+ "loss": 0.0573,
+ "step": 19551
+ },
+ {
+ "epoch": 53.275204359673026,
+ "grad_norm": 2.9294519424438477,
+ "learning_rate": 9.425371450965352e-06,
+ "loss": 0.1093,
+ "step": 19552
+ },
+ {
+ "epoch": 53.27792915531335,
+ "grad_norm": 2.229907751083374,
+ "learning_rate": 9.424490416762811e-06,
+ "loss": 0.044,
+ "step": 19553
+ },
+ {
+ "epoch": 53.28065395095368,
+ "grad_norm": 2.4395806789398193,
+ "learning_rate": 9.423609387042322e-06,
+ "loss": 0.0676,
+ "step": 19554
+ },
+ {
+ "epoch": 53.283378746594,
+ "grad_norm": 2.492172956466675,
+ "learning_rate": 9.42272836181074e-06,
+ "loss": 0.1518,
+ "step": 19555
+ },
+ {
+ "epoch": 53.28610354223433,
+ "grad_norm": 1.843397855758667,
+ "learning_rate": 9.421847341074933e-06,
+ "loss": 0.1055,
+ "step": 19556
+ },
+ {
+ "epoch": 53.28882833787466,
+ "grad_norm": 2.164602041244507,
+ "learning_rate": 9.420966324841762e-06,
+ "loss": 0.0639,
+ "step": 19557
+ },
+ {
+ "epoch": 53.29155313351499,
+ "grad_norm": 1.1913955211639404,
+ "learning_rate": 9.420085313118086e-06,
+ "loss": 0.0178,
+ "step": 19558
+ },
+ {
+ "epoch": 53.294277929155314,
+ "grad_norm": 1.9365332126617432,
+ "learning_rate": 9.419204305910769e-06,
+ "loss": 0.1485,
+ "step": 19559
+ },
+ {
+ "epoch": 53.29700272479564,
+ "grad_norm": 2.0457797050476074,
+ "learning_rate": 9.418323303226667e-06,
+ "loss": 0.0371,
+ "step": 19560
+ },
+ {
+ "epoch": 53.299727520435965,
+ "grad_norm": 2.238787889480591,
+ "learning_rate": 9.417442305072647e-06,
+ "loss": 0.0344,
+ "step": 19561
+ },
+ {
+ "epoch": 53.30245231607629,
+ "grad_norm": 2.515655994415283,
+ "learning_rate": 9.416561311455565e-06,
+ "loss": 0.0293,
+ "step": 19562
+ },
+ {
+ "epoch": 53.305177111716624,
+ "grad_norm": 2.239564895629883,
+ "learning_rate": 9.415680322382289e-06,
+ "loss": 0.0606,
+ "step": 19563
+ },
+ {
+ "epoch": 53.30790190735695,
+ "grad_norm": 2.167081356048584,
+ "learning_rate": 9.41479933785967e-06,
+ "loss": 0.0405,
+ "step": 19564
+ },
+ {
+ "epoch": 53.310626702997276,
+ "grad_norm": 1.7086117267608643,
+ "learning_rate": 9.413918357894579e-06,
+ "loss": 0.0219,
+ "step": 19565
+ },
+ {
+ "epoch": 53.3133514986376,
+ "grad_norm": 2.7973248958587646,
+ "learning_rate": 9.41303738249387e-06,
+ "loss": 0.2917,
+ "step": 19566
+ },
+ {
+ "epoch": 53.31607629427793,
+ "grad_norm": 2.056171417236328,
+ "learning_rate": 9.41215641166441e-06,
+ "loss": 0.1068,
+ "step": 19567
+ },
+ {
+ "epoch": 53.31880108991825,
+ "grad_norm": 2.34356951713562,
+ "learning_rate": 9.411275445413055e-06,
+ "loss": 0.1035,
+ "step": 19568
+ },
+ {
+ "epoch": 53.321525885558586,
+ "grad_norm": 2.5406782627105713,
+ "learning_rate": 9.41039448374667e-06,
+ "loss": 0.0332,
+ "step": 19569
+ },
+ {
+ "epoch": 53.32425068119891,
+ "grad_norm": 1.8950986862182617,
+ "learning_rate": 9.409513526672111e-06,
+ "loss": 0.04,
+ "step": 19570
+ },
+ {
+ "epoch": 53.32697547683924,
+ "grad_norm": 2.4875943660736084,
+ "learning_rate": 9.408632574196245e-06,
+ "loss": 0.1384,
+ "step": 19571
+ },
+ {
+ "epoch": 53.32970027247956,
+ "grad_norm": 2.5494449138641357,
+ "learning_rate": 9.407751626325922e-06,
+ "loss": 0.1182,
+ "step": 19572
+ },
+ {
+ "epoch": 53.33242506811989,
+ "grad_norm": 1.9768933057785034,
+ "learning_rate": 9.406870683068017e-06,
+ "loss": 0.0546,
+ "step": 19573
+ },
+ {
+ "epoch": 53.335149863760215,
+ "grad_norm": 2.315608263015747,
+ "learning_rate": 9.40598974442938e-06,
+ "loss": 0.0392,
+ "step": 19574
+ },
+ {
+ "epoch": 53.33787465940055,
+ "grad_norm": 2.79134464263916,
+ "learning_rate": 9.405108810416881e-06,
+ "loss": 0.0406,
+ "step": 19575
+ },
+ {
+ "epoch": 53.34059945504087,
+ "grad_norm": 2.86594295501709,
+ "learning_rate": 9.404227881037372e-06,
+ "loss": 0.0688,
+ "step": 19576
+ },
+ {
+ "epoch": 53.3433242506812,
+ "grad_norm": 2.2367608547210693,
+ "learning_rate": 9.40334695629772e-06,
+ "loss": 0.1016,
+ "step": 19577
+ },
+ {
+ "epoch": 53.346049046321525,
+ "grad_norm": 1.7314616441726685,
+ "learning_rate": 9.402466036204779e-06,
+ "loss": 0.0293,
+ "step": 19578
+ },
+ {
+ "epoch": 53.34877384196185,
+ "grad_norm": 2.1894755363464355,
+ "learning_rate": 9.401585120765416e-06,
+ "loss": 0.0311,
+ "step": 19579
+ },
+ {
+ "epoch": 53.35149863760218,
+ "grad_norm": 2.4152393341064453,
+ "learning_rate": 9.400704209986488e-06,
+ "loss": 0.0803,
+ "step": 19580
+ },
+ {
+ "epoch": 53.35422343324251,
+ "grad_norm": 2.0267276763916016,
+ "learning_rate": 9.399823303874856e-06,
+ "loss": 0.0366,
+ "step": 19581
+ },
+ {
+ "epoch": 53.356948228882835,
+ "grad_norm": 2.6885902881622314,
+ "learning_rate": 9.398942402437382e-06,
+ "loss": 0.1478,
+ "step": 19582
+ },
+ {
+ "epoch": 53.35967302452316,
+ "grad_norm": 2.2547807693481445,
+ "learning_rate": 9.398061505680925e-06,
+ "loss": 0.0357,
+ "step": 19583
+ },
+ {
+ "epoch": 53.36239782016349,
+ "grad_norm": 2.372467041015625,
+ "learning_rate": 9.397180613612343e-06,
+ "loss": 0.0383,
+ "step": 19584
+ },
+ {
+ "epoch": 53.36512261580381,
+ "grad_norm": 2.7570977210998535,
+ "learning_rate": 9.396299726238501e-06,
+ "loss": 0.0563,
+ "step": 19585
+ },
+ {
+ "epoch": 53.36784741144414,
+ "grad_norm": 2.137174129486084,
+ "learning_rate": 9.395418843566257e-06,
+ "loss": 0.0587,
+ "step": 19586
+ },
+ {
+ "epoch": 53.37057220708447,
+ "grad_norm": 1.8282086849212646,
+ "learning_rate": 9.394537965602473e-06,
+ "loss": 0.0281,
+ "step": 19587
+ },
+ {
+ "epoch": 53.3732970027248,
+ "grad_norm": 3.0089292526245117,
+ "learning_rate": 9.393657092354005e-06,
+ "loss": 0.0595,
+ "step": 19588
+ },
+ {
+ "epoch": 53.37602179836512,
+ "grad_norm": 2.253026008605957,
+ "learning_rate": 9.392776223827715e-06,
+ "loss": 0.0741,
+ "step": 19589
+ },
+ {
+ "epoch": 53.37874659400545,
+ "grad_norm": 2.7236275672912598,
+ "learning_rate": 9.391895360030467e-06,
+ "loss": 0.0509,
+ "step": 19590
+ },
+ {
+ "epoch": 53.381471389645775,
+ "grad_norm": 2.4988787174224854,
+ "learning_rate": 9.391014500969118e-06,
+ "loss": 0.063,
+ "step": 19591
+ },
+ {
+ "epoch": 53.3841961852861,
+ "grad_norm": 2.461725950241089,
+ "learning_rate": 9.39013364665053e-06,
+ "loss": 0.1296,
+ "step": 19592
+ },
+ {
+ "epoch": 53.38692098092643,
+ "grad_norm": 2.569134473800659,
+ "learning_rate": 9.389252797081558e-06,
+ "loss": 0.0395,
+ "step": 19593
+ },
+ {
+ "epoch": 53.38964577656676,
+ "grad_norm": 2.59150767326355,
+ "learning_rate": 9.388371952269068e-06,
+ "loss": 0.1288,
+ "step": 19594
+ },
+ {
+ "epoch": 53.392370572207085,
+ "grad_norm": 1.9067697525024414,
+ "learning_rate": 9.387491112219917e-06,
+ "loss": 0.0988,
+ "step": 19595
+ },
+ {
+ "epoch": 53.39509536784741,
+ "grad_norm": 2.0372848510742188,
+ "learning_rate": 9.386610276940966e-06,
+ "loss": 0.0813,
+ "step": 19596
+ },
+ {
+ "epoch": 53.39782016348774,
+ "grad_norm": 1.9124276638031006,
+ "learning_rate": 9.385729446439074e-06,
+ "loss": 0.0688,
+ "step": 19597
+ },
+ {
+ "epoch": 53.40054495912806,
+ "grad_norm": 1.8603287935256958,
+ "learning_rate": 9.384848620721102e-06,
+ "loss": 0.0358,
+ "step": 19598
+ },
+ {
+ "epoch": 53.403269754768395,
+ "grad_norm": 2.715649366378784,
+ "learning_rate": 9.38396779979391e-06,
+ "loss": 0.1538,
+ "step": 19599
+ },
+ {
+ "epoch": 53.40599455040872,
+ "grad_norm": 2.085322618484497,
+ "learning_rate": 9.383086983664354e-06,
+ "loss": 0.1439,
+ "step": 19600
+ },
+ {
+ "epoch": 53.40871934604905,
+ "grad_norm": 3.0355520248413086,
+ "learning_rate": 9.3822061723393e-06,
+ "loss": 0.0537,
+ "step": 19601
+ },
+ {
+ "epoch": 53.41144414168937,
+ "grad_norm": 2.716320276260376,
+ "learning_rate": 9.381325365825601e-06,
+ "loss": 0.0308,
+ "step": 19602
+ },
+ {
+ "epoch": 53.4141689373297,
+ "grad_norm": 1.9399898052215576,
+ "learning_rate": 9.380444564130124e-06,
+ "loss": 0.1248,
+ "step": 19603
+ },
+ {
+ "epoch": 53.416893732970024,
+ "grad_norm": 2.485546112060547,
+ "learning_rate": 9.379563767259721e-06,
+ "loss": 0.0706,
+ "step": 19604
+ },
+ {
+ "epoch": 53.41961852861036,
+ "grad_norm": 1.6398659944534302,
+ "learning_rate": 9.378682975221258e-06,
+ "loss": 0.0298,
+ "step": 19605
+ },
+ {
+ "epoch": 53.42234332425068,
+ "grad_norm": 2.3411335945129395,
+ "learning_rate": 9.37780218802159e-06,
+ "loss": 0.0966,
+ "step": 19606
+ },
+ {
+ "epoch": 53.42506811989101,
+ "grad_norm": 2.879350423812866,
+ "learning_rate": 9.376921405667576e-06,
+ "loss": 0.0681,
+ "step": 19607
+ },
+ {
+ "epoch": 53.427792915531334,
+ "grad_norm": 1.5600498914718628,
+ "learning_rate": 9.376040628166084e-06,
+ "loss": 0.0215,
+ "step": 19608
+ },
+ {
+ "epoch": 53.43051771117166,
+ "grad_norm": 1.7704826593399048,
+ "learning_rate": 9.375159855523964e-06,
+ "loss": 0.0308,
+ "step": 19609
+ },
+ {
+ "epoch": 53.433242506811986,
+ "grad_norm": 2.5143682956695557,
+ "learning_rate": 9.374279087748082e-06,
+ "loss": 0.0744,
+ "step": 19610
+ },
+ {
+ "epoch": 53.43596730245232,
+ "grad_norm": 3.8349459171295166,
+ "learning_rate": 9.37339832484529e-06,
+ "loss": 0.0666,
+ "step": 19611
+ },
+ {
+ "epoch": 53.438692098092645,
+ "grad_norm": 2.163703680038452,
+ "learning_rate": 9.372517566822456e-06,
+ "loss": 0.0976,
+ "step": 19612
+ },
+ {
+ "epoch": 53.44141689373297,
+ "grad_norm": 2.770625591278076,
+ "learning_rate": 9.371636813686433e-06,
+ "loss": 0.095,
+ "step": 19613
+ },
+ {
+ "epoch": 53.444141689373296,
+ "grad_norm": 2.7361185550689697,
+ "learning_rate": 9.370756065444083e-06,
+ "loss": 0.0965,
+ "step": 19614
+ },
+ {
+ "epoch": 53.44686648501362,
+ "grad_norm": 2.866441488265991,
+ "learning_rate": 9.369875322102262e-06,
+ "loss": 0.0771,
+ "step": 19615
+ },
+ {
+ "epoch": 53.44959128065395,
+ "grad_norm": 2.077071189880371,
+ "learning_rate": 9.368994583667835e-06,
+ "loss": 0.0787,
+ "step": 19616
+ },
+ {
+ "epoch": 53.45231607629428,
+ "grad_norm": 2.0627479553222656,
+ "learning_rate": 9.368113850147655e-06,
+ "loss": 0.0382,
+ "step": 19617
+ },
+ {
+ "epoch": 53.45504087193461,
+ "grad_norm": 2.026026725769043,
+ "learning_rate": 9.367233121548585e-06,
+ "loss": 0.0595,
+ "step": 19618
+ },
+ {
+ "epoch": 53.45776566757493,
+ "grad_norm": 2.172414541244507,
+ "learning_rate": 9.366352397877485e-06,
+ "loss": 0.0668,
+ "step": 19619
+ },
+ {
+ "epoch": 53.46049046321526,
+ "grad_norm": 2.296494960784912,
+ "learning_rate": 9.36547167914121e-06,
+ "loss": 0.0997,
+ "step": 19620
+ },
+ {
+ "epoch": 53.463215258855584,
+ "grad_norm": 2.4783873558044434,
+ "learning_rate": 9.364590965346622e-06,
+ "loss": 0.0505,
+ "step": 19621
+ },
+ {
+ "epoch": 53.46594005449591,
+ "grad_norm": 2.0330007076263428,
+ "learning_rate": 9.363710256500578e-06,
+ "loss": 0.045,
+ "step": 19622
+ },
+ {
+ "epoch": 53.46866485013624,
+ "grad_norm": 1.8184881210327148,
+ "learning_rate": 9.36282955260994e-06,
+ "loss": 0.0388,
+ "step": 19623
+ },
+ {
+ "epoch": 53.47138964577657,
+ "grad_norm": 2.100782871246338,
+ "learning_rate": 9.361948853681562e-06,
+ "loss": 0.0788,
+ "step": 19624
+ },
+ {
+ "epoch": 53.474114441416894,
+ "grad_norm": 2.4429171085357666,
+ "learning_rate": 9.361068159722306e-06,
+ "loss": 0.0767,
+ "step": 19625
+ },
+ {
+ "epoch": 53.47683923705722,
+ "grad_norm": 1.8915714025497437,
+ "learning_rate": 9.360187470739033e-06,
+ "loss": 0.0313,
+ "step": 19626
+ },
+ {
+ "epoch": 53.479564032697546,
+ "grad_norm": 2.613842725753784,
+ "learning_rate": 9.359306786738598e-06,
+ "loss": 0.0753,
+ "step": 19627
+ },
+ {
+ "epoch": 53.48228882833787,
+ "grad_norm": 1.8366929292678833,
+ "learning_rate": 9.358426107727862e-06,
+ "loss": 0.0581,
+ "step": 19628
+ },
+ {
+ "epoch": 53.485013623978205,
+ "grad_norm": 1.9857792854309082,
+ "learning_rate": 9.357545433713683e-06,
+ "loss": 0.0519,
+ "step": 19629
+ },
+ {
+ "epoch": 53.48773841961853,
+ "grad_norm": 2.501182794570923,
+ "learning_rate": 9.35666476470292e-06,
+ "loss": 0.1037,
+ "step": 19630
+ },
+ {
+ "epoch": 53.490463215258856,
+ "grad_norm": 2.2017760276794434,
+ "learning_rate": 9.35578410070243e-06,
+ "loss": 0.0988,
+ "step": 19631
+ },
+ {
+ "epoch": 53.49318801089918,
+ "grad_norm": 2.9633054733276367,
+ "learning_rate": 9.354903441719073e-06,
+ "loss": 0.0921,
+ "step": 19632
+ },
+ {
+ "epoch": 53.49591280653951,
+ "grad_norm": 2.6600046157836914,
+ "learning_rate": 9.354022787759707e-06,
+ "loss": 0.0711,
+ "step": 19633
+ },
+ {
+ "epoch": 53.49863760217983,
+ "grad_norm": 2.380194664001465,
+ "learning_rate": 9.353142138831192e-06,
+ "loss": 0.094,
+ "step": 19634
+ },
+ {
+ "epoch": 53.50136239782017,
+ "grad_norm": 1.4678584337234497,
+ "learning_rate": 9.352261494940382e-06,
+ "loss": 0.0197,
+ "step": 19635
+ },
+ {
+ "epoch": 53.50408719346049,
+ "grad_norm": 3.1108171939849854,
+ "learning_rate": 9.351380856094142e-06,
+ "loss": 0.1237,
+ "step": 19636
+ },
+ {
+ "epoch": 53.50681198910082,
+ "grad_norm": 6.049625396728516,
+ "learning_rate": 9.350500222299326e-06,
+ "loss": 0.0722,
+ "step": 19637
+ },
+ {
+ "epoch": 53.509536784741144,
+ "grad_norm": 1.7614415884017944,
+ "learning_rate": 9.349619593562793e-06,
+ "loss": 0.0584,
+ "step": 19638
+ },
+ {
+ "epoch": 53.51226158038147,
+ "grad_norm": 2.0790133476257324,
+ "learning_rate": 9.348738969891402e-06,
+ "loss": 0.0513,
+ "step": 19639
+ },
+ {
+ "epoch": 53.514986376021795,
+ "grad_norm": 2.442708730697632,
+ "learning_rate": 9.34785835129201e-06,
+ "loss": 0.0349,
+ "step": 19640
+ },
+ {
+ "epoch": 53.51771117166213,
+ "grad_norm": 2.4963772296905518,
+ "learning_rate": 9.346977737771471e-06,
+ "loss": 0.1438,
+ "step": 19641
+ },
+ {
+ "epoch": 53.520435967302454,
+ "grad_norm": 2.2713704109191895,
+ "learning_rate": 9.346097129336658e-06,
+ "loss": 0.1326,
+ "step": 19642
+ },
+ {
+ "epoch": 53.52316076294278,
+ "grad_norm": 2.399904489517212,
+ "learning_rate": 9.345216525994415e-06,
+ "loss": 0.0477,
+ "step": 19643
+ },
+ {
+ "epoch": 53.525885558583106,
+ "grad_norm": 2.52372670173645,
+ "learning_rate": 9.344335927751606e-06,
+ "loss": 0.1672,
+ "step": 19644
+ },
+ {
+ "epoch": 53.52861035422343,
+ "grad_norm": 3.8516716957092285,
+ "learning_rate": 9.343455334615085e-06,
+ "loss": 0.1356,
+ "step": 19645
+ },
+ {
+ "epoch": 53.53133514986376,
+ "grad_norm": 2.2704710960388184,
+ "learning_rate": 9.342574746591718e-06,
+ "loss": 0.0414,
+ "step": 19646
+ },
+ {
+ "epoch": 53.53405994550409,
+ "grad_norm": 1.6247299909591675,
+ "learning_rate": 9.341694163688352e-06,
+ "loss": 0.1383,
+ "step": 19647
+ },
+ {
+ "epoch": 53.536784741144416,
+ "grad_norm": 2.5968704223632812,
+ "learning_rate": 9.340813585911854e-06,
+ "loss": 0.1323,
+ "step": 19648
+ },
+ {
+ "epoch": 53.53950953678474,
+ "grad_norm": 2.498697519302368,
+ "learning_rate": 9.339933013269076e-06,
+ "loss": 0.0641,
+ "step": 19649
+ },
+ {
+ "epoch": 53.54223433242507,
+ "grad_norm": 2.3188328742980957,
+ "learning_rate": 9.339052445766883e-06,
+ "loss": 0.0516,
+ "step": 19650
+ },
+ {
+ "epoch": 53.54495912806539,
+ "grad_norm": 1.9664748907089233,
+ "learning_rate": 9.338171883412123e-06,
+ "loss": 0.0337,
+ "step": 19651
+ },
+ {
+ "epoch": 53.54768392370572,
+ "grad_norm": 2.4697864055633545,
+ "learning_rate": 9.337291326211663e-06,
+ "loss": 0.0462,
+ "step": 19652
+ },
+ {
+ "epoch": 53.55040871934605,
+ "grad_norm": 2.4101650714874268,
+ "learning_rate": 9.336410774172356e-06,
+ "loss": 0.046,
+ "step": 19653
+ },
+ {
+ "epoch": 53.55313351498638,
+ "grad_norm": 9.224514961242676,
+ "learning_rate": 9.33553022730106e-06,
+ "loss": 0.0245,
+ "step": 19654
+ },
+ {
+ "epoch": 53.555858310626704,
+ "grad_norm": 3.354036569595337,
+ "learning_rate": 9.334649685604632e-06,
+ "loss": 0.2396,
+ "step": 19655
+ },
+ {
+ "epoch": 53.55858310626703,
+ "grad_norm": 2.4946460723876953,
+ "learning_rate": 9.333769149089934e-06,
+ "loss": 0.0317,
+ "step": 19656
+ },
+ {
+ "epoch": 53.561307901907355,
+ "grad_norm": 1.9953597784042358,
+ "learning_rate": 9.332888617763816e-06,
+ "loss": 0.0521,
+ "step": 19657
+ },
+ {
+ "epoch": 53.56403269754768,
+ "grad_norm": 2.672818422317505,
+ "learning_rate": 9.33200809163314e-06,
+ "loss": 0.0856,
+ "step": 19658
+ },
+ {
+ "epoch": 53.566757493188014,
+ "grad_norm": 2.3943088054656982,
+ "learning_rate": 9.331127570704765e-06,
+ "loss": 0.0578,
+ "step": 19659
+ },
+ {
+ "epoch": 53.56948228882834,
+ "grad_norm": 10.690425872802734,
+ "learning_rate": 9.330247054985548e-06,
+ "loss": 0.0333,
+ "step": 19660
+ },
+ {
+ "epoch": 53.572207084468666,
+ "grad_norm": 2.3924756050109863,
+ "learning_rate": 9.329366544482346e-06,
+ "loss": 0.1006,
+ "step": 19661
+ },
+ {
+ "epoch": 53.57493188010899,
+ "grad_norm": 4.041621208190918,
+ "learning_rate": 9.328486039202015e-06,
+ "loss": 0.0785,
+ "step": 19662
+ },
+ {
+ "epoch": 53.57765667574932,
+ "grad_norm": 2.219719171524048,
+ "learning_rate": 9.327605539151412e-06,
+ "loss": 0.0474,
+ "step": 19663
+ },
+ {
+ "epoch": 53.58038147138964,
+ "grad_norm": 1.9250259399414062,
+ "learning_rate": 9.326725044337398e-06,
+ "loss": 0.1291,
+ "step": 19664
+ },
+ {
+ "epoch": 53.583106267029976,
+ "grad_norm": 2.4022228717803955,
+ "learning_rate": 9.325844554766826e-06,
+ "loss": 0.0538,
+ "step": 19665
+ },
+ {
+ "epoch": 53.5858310626703,
+ "grad_norm": 1.9270833730697632,
+ "learning_rate": 9.324964070446556e-06,
+ "loss": 0.044,
+ "step": 19666
+ },
+ {
+ "epoch": 53.58855585831063,
+ "grad_norm": 2.642576217651367,
+ "learning_rate": 9.324083591383441e-06,
+ "loss": 0.0637,
+ "step": 19667
+ },
+ {
+ "epoch": 53.59128065395095,
+ "grad_norm": 1.9235405921936035,
+ "learning_rate": 9.323203117584345e-06,
+ "loss": 0.0515,
+ "step": 19668
+ },
+ {
+ "epoch": 53.59400544959128,
+ "grad_norm": 2.571953773498535,
+ "learning_rate": 9.322322649056118e-06,
+ "loss": 0.0439,
+ "step": 19669
+ },
+ {
+ "epoch": 53.596730245231605,
+ "grad_norm": 1.9366233348846436,
+ "learning_rate": 9.321442185805625e-06,
+ "loss": 0.0505,
+ "step": 19670
+ },
+ {
+ "epoch": 53.59945504087194,
+ "grad_norm": 2.3768057823181152,
+ "learning_rate": 9.320561727839715e-06,
+ "loss": 0.043,
+ "step": 19671
+ },
+ {
+ "epoch": 53.60217983651226,
+ "grad_norm": 3.0929694175720215,
+ "learning_rate": 9.31968127516525e-06,
+ "loss": 0.1052,
+ "step": 19672
+ },
+ {
+ "epoch": 53.60490463215259,
+ "grad_norm": 2.623072385787964,
+ "learning_rate": 9.318800827789082e-06,
+ "loss": 0.0534,
+ "step": 19673
+ },
+ {
+ "epoch": 53.607629427792915,
+ "grad_norm": 2.3726954460144043,
+ "learning_rate": 9.317920385718075e-06,
+ "loss": 0.0427,
+ "step": 19674
+ },
+ {
+ "epoch": 53.61035422343324,
+ "grad_norm": 2.5545077323913574,
+ "learning_rate": 9.317039948959077e-06,
+ "loss": 0.1598,
+ "step": 19675
+ },
+ {
+ "epoch": 53.61307901907357,
+ "grad_norm": 1.606134057044983,
+ "learning_rate": 9.316159517518952e-06,
+ "loss": 0.0377,
+ "step": 19676
+ },
+ {
+ "epoch": 53.6158038147139,
+ "grad_norm": 2.1293370723724365,
+ "learning_rate": 9.315279091404558e-06,
+ "loss": 0.0464,
+ "step": 19677
+ },
+ {
+ "epoch": 53.618528610354225,
+ "grad_norm": 1.9684182405471802,
+ "learning_rate": 9.314398670622745e-06,
+ "loss": 0.0475,
+ "step": 19678
+ },
+ {
+ "epoch": 53.62125340599455,
+ "grad_norm": 1.836662769317627,
+ "learning_rate": 9.313518255180375e-06,
+ "loss": 0.0371,
+ "step": 19679
+ },
+ {
+ "epoch": 53.62397820163488,
+ "grad_norm": 2.577711582183838,
+ "learning_rate": 9.312637845084303e-06,
+ "loss": 0.2999,
+ "step": 19680
+ },
+ {
+ "epoch": 53.6267029972752,
+ "grad_norm": 2.235924243927002,
+ "learning_rate": 9.311757440341384e-06,
+ "loss": 0.0337,
+ "step": 19681
+ },
+ {
+ "epoch": 53.62942779291553,
+ "grad_norm": 2.4647085666656494,
+ "learning_rate": 9.310877040958478e-06,
+ "loss": 0.2182,
+ "step": 19682
+ },
+ {
+ "epoch": 53.63215258855586,
+ "grad_norm": 1.8259167671203613,
+ "learning_rate": 9.309996646942437e-06,
+ "loss": 0.1093,
+ "step": 19683
+ },
+ {
+ "epoch": 53.63487738419619,
+ "grad_norm": 2.287522077560425,
+ "learning_rate": 9.309116258300121e-06,
+ "loss": 0.0565,
+ "step": 19684
+ },
+ {
+ "epoch": 53.63760217983651,
+ "grad_norm": 2.4461019039154053,
+ "learning_rate": 9.308235875038385e-06,
+ "loss": 0.0379,
+ "step": 19685
+ },
+ {
+ "epoch": 53.64032697547684,
+ "grad_norm": 2.074802875518799,
+ "learning_rate": 9.307355497164086e-06,
+ "loss": 0.0622,
+ "step": 19686
+ },
+ {
+ "epoch": 53.643051771117165,
+ "grad_norm": 1.8279881477355957,
+ "learning_rate": 9.30647512468408e-06,
+ "loss": 0.0395,
+ "step": 19687
+ },
+ {
+ "epoch": 53.64577656675749,
+ "grad_norm": 2.433382272720337,
+ "learning_rate": 9.305594757605226e-06,
+ "loss": 0.1098,
+ "step": 19688
+ },
+ {
+ "epoch": 53.64850136239782,
+ "grad_norm": 2.781404495239258,
+ "learning_rate": 9.304714395934373e-06,
+ "loss": 0.0526,
+ "step": 19689
+ },
+ {
+ "epoch": 53.65122615803815,
+ "grad_norm": 2.2660348415374756,
+ "learning_rate": 9.303834039678386e-06,
+ "loss": 0.0618,
+ "step": 19690
+ },
+ {
+ "epoch": 53.653950953678475,
+ "grad_norm": 1.7611048221588135,
+ "learning_rate": 9.302953688844113e-06,
+ "loss": 0.0725,
+ "step": 19691
+ },
+ {
+ "epoch": 53.6566757493188,
+ "grad_norm": 2.3267927169799805,
+ "learning_rate": 9.302073343438414e-06,
+ "loss": 0.041,
+ "step": 19692
+ },
+ {
+ "epoch": 53.65940054495913,
+ "grad_norm": 1.963337779045105,
+ "learning_rate": 9.301193003468148e-06,
+ "loss": 0.1611,
+ "step": 19693
+ },
+ {
+ "epoch": 53.66212534059945,
+ "grad_norm": 2.323141574859619,
+ "learning_rate": 9.300312668940168e-06,
+ "loss": 0.1171,
+ "step": 19694
+ },
+ {
+ "epoch": 53.664850136239785,
+ "grad_norm": 1.9084380865097046,
+ "learning_rate": 9.299432339861332e-06,
+ "loss": 0.0296,
+ "step": 19695
+ },
+ {
+ "epoch": 53.66757493188011,
+ "grad_norm": 2.051236391067505,
+ "learning_rate": 9.298552016238492e-06,
+ "loss": 0.0805,
+ "step": 19696
+ },
+ {
+ "epoch": 53.67029972752044,
+ "grad_norm": 2.6408162117004395,
+ "learning_rate": 9.297671698078508e-06,
+ "loss": 0.0834,
+ "step": 19697
+ },
+ {
+ "epoch": 53.67302452316076,
+ "grad_norm": 1.5521423816680908,
+ "learning_rate": 9.296791385388233e-06,
+ "loss": 0.033,
+ "step": 19698
+ },
+ {
+ "epoch": 53.67574931880109,
+ "grad_norm": 1.5592353343963623,
+ "learning_rate": 9.295911078174526e-06,
+ "loss": 0.0249,
+ "step": 19699
+ },
+ {
+ "epoch": 53.678474114441414,
+ "grad_norm": 2.6670637130737305,
+ "learning_rate": 9.295030776444239e-06,
+ "loss": 0.0372,
+ "step": 19700
+ },
+ {
+ "epoch": 53.68119891008175,
+ "grad_norm": 1.7121764421463013,
+ "learning_rate": 9.29415048020423e-06,
+ "loss": 0.0318,
+ "step": 19701
+ },
+ {
+ "epoch": 53.68392370572207,
+ "grad_norm": 1.7293109893798828,
+ "learning_rate": 9.293270189461354e-06,
+ "loss": 0.0581,
+ "step": 19702
+ },
+ {
+ "epoch": 53.6866485013624,
+ "grad_norm": 3.6132540702819824,
+ "learning_rate": 9.292389904222468e-06,
+ "loss": 0.0664,
+ "step": 19703
+ },
+ {
+ "epoch": 53.689373297002724,
+ "grad_norm": 2.8182387351989746,
+ "learning_rate": 9.291509624494426e-06,
+ "loss": 0.1154,
+ "step": 19704
+ },
+ {
+ "epoch": 53.69209809264305,
+ "grad_norm": 3.0405113697052,
+ "learning_rate": 9.290629350284083e-06,
+ "loss": 0.2122,
+ "step": 19705
+ },
+ {
+ "epoch": 53.694822888283376,
+ "grad_norm": 2.7197930812835693,
+ "learning_rate": 9.289749081598298e-06,
+ "loss": 0.0545,
+ "step": 19706
+ },
+ {
+ "epoch": 53.69754768392371,
+ "grad_norm": 1.7691274881362915,
+ "learning_rate": 9.288868818443923e-06,
+ "loss": 0.0398,
+ "step": 19707
+ },
+ {
+ "epoch": 53.700272479564035,
+ "grad_norm": 1.5248658657073975,
+ "learning_rate": 9.287988560827814e-06,
+ "loss": 0.0272,
+ "step": 19708
+ },
+ {
+ "epoch": 53.70299727520436,
+ "grad_norm": 2.1526129245758057,
+ "learning_rate": 9.28710830875683e-06,
+ "loss": 0.039,
+ "step": 19709
+ },
+ {
+ "epoch": 53.705722070844686,
+ "grad_norm": 2.1251957416534424,
+ "learning_rate": 9.286228062237817e-06,
+ "loss": 0.1856,
+ "step": 19710
+ },
+ {
+ "epoch": 53.70844686648501,
+ "grad_norm": 2.0073840618133545,
+ "learning_rate": 9.285347821277644e-06,
+ "loss": 0.1285,
+ "step": 19711
+ },
+ {
+ "epoch": 53.71117166212534,
+ "grad_norm": 1.7754219770431519,
+ "learning_rate": 9.284467585883156e-06,
+ "loss": 0.0616,
+ "step": 19712
+ },
+ {
+ "epoch": 53.71389645776567,
+ "grad_norm": 2.546035051345825,
+ "learning_rate": 9.283587356061214e-06,
+ "loss": 0.0441,
+ "step": 19713
+ },
+ {
+ "epoch": 53.716621253406,
+ "grad_norm": 2.815772771835327,
+ "learning_rate": 9.282707131818666e-06,
+ "loss": 0.0572,
+ "step": 19714
+ },
+ {
+ "epoch": 53.71934604904632,
+ "grad_norm": 2.1469149589538574,
+ "learning_rate": 9.281826913162377e-06,
+ "loss": 0.045,
+ "step": 19715
+ },
+ {
+ "epoch": 53.72207084468665,
+ "grad_norm": 3.622260570526123,
+ "learning_rate": 9.280946700099194e-06,
+ "loss": 0.0688,
+ "step": 19716
+ },
+ {
+ "epoch": 53.724795640326974,
+ "grad_norm": 2.120177745819092,
+ "learning_rate": 9.280066492635976e-06,
+ "loss": 0.037,
+ "step": 19717
+ },
+ {
+ "epoch": 53.7275204359673,
+ "grad_norm": 2.758284330368042,
+ "learning_rate": 9.279186290779576e-06,
+ "loss": 0.0476,
+ "step": 19718
+ },
+ {
+ "epoch": 53.73024523160763,
+ "grad_norm": 1.8275502920150757,
+ "learning_rate": 9.278306094536851e-06,
+ "loss": 0.0381,
+ "step": 19719
+ },
+ {
+ "epoch": 53.73297002724796,
+ "grad_norm": 2.3045341968536377,
+ "learning_rate": 9.277425903914654e-06,
+ "loss": 0.0395,
+ "step": 19720
+ },
+ {
+ "epoch": 53.735694822888284,
+ "grad_norm": 2.322319507598877,
+ "learning_rate": 9.276545718919842e-06,
+ "loss": 0.0465,
+ "step": 19721
+ },
+ {
+ "epoch": 53.73841961852861,
+ "grad_norm": 2.581332206726074,
+ "learning_rate": 9.275665539559268e-06,
+ "loss": 0.064,
+ "step": 19722
+ },
+ {
+ "epoch": 53.741144414168936,
+ "grad_norm": 1.7737714052200317,
+ "learning_rate": 9.274785365839788e-06,
+ "loss": 0.0351,
+ "step": 19723
+ },
+ {
+ "epoch": 53.74386920980926,
+ "grad_norm": 2.491161346435547,
+ "learning_rate": 9.273905197768254e-06,
+ "loss": 0.0685,
+ "step": 19724
+ },
+ {
+ "epoch": 53.746594005449595,
+ "grad_norm": 2.8109817504882812,
+ "learning_rate": 9.273025035351526e-06,
+ "loss": 0.0967,
+ "step": 19725
+ },
+ {
+ "epoch": 53.74931880108992,
+ "grad_norm": 3.3395791053771973,
+ "learning_rate": 9.272144878596454e-06,
+ "loss": 0.0526,
+ "step": 19726
+ },
+ {
+ "epoch": 53.752043596730246,
+ "grad_norm": 2.483640670776367,
+ "learning_rate": 9.27126472750989e-06,
+ "loss": 0.0563,
+ "step": 19727
+ },
+ {
+ "epoch": 53.75476839237057,
+ "grad_norm": 3.037386417388916,
+ "learning_rate": 9.270384582098697e-06,
+ "loss": 0.0337,
+ "step": 19728
+ },
+ {
+ "epoch": 53.7574931880109,
+ "grad_norm": 1.9387871026992798,
+ "learning_rate": 9.269504442369727e-06,
+ "loss": 0.0437,
+ "step": 19729
+ },
+ {
+ "epoch": 53.76021798365122,
+ "grad_norm": 2.568800449371338,
+ "learning_rate": 9.26862430832983e-06,
+ "loss": 0.0558,
+ "step": 19730
+ },
+ {
+ "epoch": 53.762942779291556,
+ "grad_norm": 2.871338129043579,
+ "learning_rate": 9.267744179985866e-06,
+ "loss": 0.0848,
+ "step": 19731
+ },
+ {
+ "epoch": 53.76566757493188,
+ "grad_norm": 2.661219358444214,
+ "learning_rate": 9.266864057344684e-06,
+ "loss": 0.0292,
+ "step": 19732
+ },
+ {
+ "epoch": 53.76839237057221,
+ "grad_norm": 3.1716957092285156,
+ "learning_rate": 9.265983940413145e-06,
+ "loss": 0.1444,
+ "step": 19733
+ },
+ {
+ "epoch": 53.771117166212534,
+ "grad_norm": 1.6554569005966187,
+ "learning_rate": 9.265103829198096e-06,
+ "loss": 0.0397,
+ "step": 19734
+ },
+ {
+ "epoch": 53.77384196185286,
+ "grad_norm": 2.086195707321167,
+ "learning_rate": 9.264223723706397e-06,
+ "loss": 0.0359,
+ "step": 19735
+ },
+ {
+ "epoch": 53.776566757493185,
+ "grad_norm": 3.3630521297454834,
+ "learning_rate": 9.263343623944899e-06,
+ "loss": 0.0333,
+ "step": 19736
+ },
+ {
+ "epoch": 53.77929155313352,
+ "grad_norm": 2.1334540843963623,
+ "learning_rate": 9.262463529920458e-06,
+ "loss": 0.0953,
+ "step": 19737
+ },
+ {
+ "epoch": 53.782016348773844,
+ "grad_norm": 2.9471089839935303,
+ "learning_rate": 9.261583441639926e-06,
+ "loss": 0.1159,
+ "step": 19738
+ },
+ {
+ "epoch": 53.78474114441417,
+ "grad_norm": 2.5800364017486572,
+ "learning_rate": 9.26070335911016e-06,
+ "loss": 0.1018,
+ "step": 19739
+ },
+ {
+ "epoch": 53.787465940054496,
+ "grad_norm": 2.9630801677703857,
+ "learning_rate": 9.259823282338008e-06,
+ "loss": 0.0706,
+ "step": 19740
+ },
+ {
+ "epoch": 53.79019073569482,
+ "grad_norm": 1.8297141790390015,
+ "learning_rate": 9.258943211330334e-06,
+ "loss": 0.0269,
+ "step": 19741
+ },
+ {
+ "epoch": 53.79291553133515,
+ "grad_norm": 2.391658306121826,
+ "learning_rate": 9.258063146093984e-06,
+ "loss": 0.162,
+ "step": 19742
+ },
+ {
+ "epoch": 53.79564032697548,
+ "grad_norm": 2.3330769538879395,
+ "learning_rate": 9.257183086635814e-06,
+ "loss": 0.119,
+ "step": 19743
+ },
+ {
+ "epoch": 53.798365122615806,
+ "grad_norm": 2.4855058193206787,
+ "learning_rate": 9.256303032962676e-06,
+ "loss": 0.1054,
+ "step": 19744
+ },
+ {
+ "epoch": 53.80108991825613,
+ "grad_norm": 2.553321361541748,
+ "learning_rate": 9.255422985081426e-06,
+ "loss": 0.0491,
+ "step": 19745
+ },
+ {
+ "epoch": 53.80381471389646,
+ "grad_norm": 2.560450315475464,
+ "learning_rate": 9.254542942998921e-06,
+ "loss": 0.0868,
+ "step": 19746
+ },
+ {
+ "epoch": 53.80653950953678,
+ "grad_norm": 4.429399490356445,
+ "learning_rate": 9.253662906722013e-06,
+ "loss": 0.0801,
+ "step": 19747
+ },
+ {
+ "epoch": 53.80926430517711,
+ "grad_norm": 2.056278944015503,
+ "learning_rate": 9.25278287625755e-06,
+ "loss": 0.0435,
+ "step": 19748
+ },
+ {
+ "epoch": 53.81198910081744,
+ "grad_norm": 2.113776206970215,
+ "learning_rate": 9.251902851612394e-06,
+ "loss": 0.0873,
+ "step": 19749
+ },
+ {
+ "epoch": 53.81471389645777,
+ "grad_norm": 1.872866153717041,
+ "learning_rate": 9.251022832793392e-06,
+ "loss": 0.1251,
+ "step": 19750
+ },
+ {
+ "epoch": 53.817438692098094,
+ "grad_norm": 2.9217300415039062,
+ "learning_rate": 9.250142819807402e-06,
+ "loss": 0.064,
+ "step": 19751
+ },
+ {
+ "epoch": 53.82016348773842,
+ "grad_norm": 2.4359779357910156,
+ "learning_rate": 9.249262812661273e-06,
+ "loss": 0.189,
+ "step": 19752
+ },
+ {
+ "epoch": 53.822888283378745,
+ "grad_norm": 2.3449504375457764,
+ "learning_rate": 9.248382811361864e-06,
+ "loss": 0.068,
+ "step": 19753
+ },
+ {
+ "epoch": 53.82561307901907,
+ "grad_norm": 2.4298040866851807,
+ "learning_rate": 9.247502815916023e-06,
+ "loss": 0.0778,
+ "step": 19754
+ },
+ {
+ "epoch": 53.828337874659404,
+ "grad_norm": 1.9975199699401855,
+ "learning_rate": 9.246622826330607e-06,
+ "loss": 0.0334,
+ "step": 19755
+ },
+ {
+ "epoch": 53.83106267029973,
+ "grad_norm": 3.361619234085083,
+ "learning_rate": 9.245742842612467e-06,
+ "loss": 0.0432,
+ "step": 19756
+ },
+ {
+ "epoch": 53.833787465940055,
+ "grad_norm": 2.6567227840423584,
+ "learning_rate": 9.24486286476846e-06,
+ "loss": 0.1015,
+ "step": 19757
+ },
+ {
+ "epoch": 53.83651226158038,
+ "grad_norm": 2.248793601989746,
+ "learning_rate": 9.243982892805434e-06,
+ "loss": 0.0953,
+ "step": 19758
+ },
+ {
+ "epoch": 53.83923705722071,
+ "grad_norm": 1.9941734075546265,
+ "learning_rate": 9.243102926730247e-06,
+ "loss": 0.0344,
+ "step": 19759
+ },
+ {
+ "epoch": 53.84196185286103,
+ "grad_norm": 2.583352565765381,
+ "learning_rate": 9.242222966549748e-06,
+ "loss": 0.0601,
+ "step": 19760
+ },
+ {
+ "epoch": 53.844686648501366,
+ "grad_norm": 9.211976051330566,
+ "learning_rate": 9.24134301227079e-06,
+ "loss": 0.0582,
+ "step": 19761
+ },
+ {
+ "epoch": 53.84741144414169,
+ "grad_norm": 2.4986205101013184,
+ "learning_rate": 9.240463063900233e-06,
+ "loss": 0.0427,
+ "step": 19762
+ },
+ {
+ "epoch": 53.85013623978202,
+ "grad_norm": 1.998684048652649,
+ "learning_rate": 9.239583121444923e-06,
+ "loss": 0.03,
+ "step": 19763
+ },
+ {
+ "epoch": 53.85286103542234,
+ "grad_norm": 5.523169994354248,
+ "learning_rate": 9.238703184911717e-06,
+ "loss": 0.1331,
+ "step": 19764
+ },
+ {
+ "epoch": 53.85558583106267,
+ "grad_norm": 2.254234790802002,
+ "learning_rate": 9.237823254307464e-06,
+ "loss": 0.0475,
+ "step": 19765
+ },
+ {
+ "epoch": 53.858310626702995,
+ "grad_norm": 2.288024663925171,
+ "learning_rate": 9.236943329639022e-06,
+ "loss": 0.2076,
+ "step": 19766
+ },
+ {
+ "epoch": 53.86103542234333,
+ "grad_norm": 2.0510616302490234,
+ "learning_rate": 9.236063410913237e-06,
+ "loss": 0.0765,
+ "step": 19767
+ },
+ {
+ "epoch": 53.86376021798365,
+ "grad_norm": 2.912844181060791,
+ "learning_rate": 9.235183498136965e-06,
+ "loss": 0.0814,
+ "step": 19768
+ },
+ {
+ "epoch": 53.86648501362398,
+ "grad_norm": 2.51546311378479,
+ "learning_rate": 9.234303591317064e-06,
+ "loss": 0.0322,
+ "step": 19769
+ },
+ {
+ "epoch": 53.869209809264305,
+ "grad_norm": 2.5354840755462646,
+ "learning_rate": 9.233423690460377e-06,
+ "loss": 0.0619,
+ "step": 19770
+ },
+ {
+ "epoch": 53.87193460490463,
+ "grad_norm": 1.880324363708496,
+ "learning_rate": 9.232543795573766e-06,
+ "loss": 0.0323,
+ "step": 19771
+ },
+ {
+ "epoch": 53.87465940054496,
+ "grad_norm": 2.1648244857788086,
+ "learning_rate": 9.231663906664077e-06,
+ "loss": 0.0502,
+ "step": 19772
+ },
+ {
+ "epoch": 53.87738419618529,
+ "grad_norm": 2.152881383895874,
+ "learning_rate": 9.230784023738165e-06,
+ "loss": 0.0955,
+ "step": 19773
+ },
+ {
+ "epoch": 53.880108991825615,
+ "grad_norm": 4.354531764984131,
+ "learning_rate": 9.229904146802882e-06,
+ "loss": 0.0585,
+ "step": 19774
+ },
+ {
+ "epoch": 53.88283378746594,
+ "grad_norm": 2.3862464427948,
+ "learning_rate": 9.229024275865083e-06,
+ "loss": 0.0578,
+ "step": 19775
+ },
+ {
+ "epoch": 53.88555858310627,
+ "grad_norm": 2.6107943058013916,
+ "learning_rate": 9.228144410931615e-06,
+ "loss": 0.0866,
+ "step": 19776
+ },
+ {
+ "epoch": 53.88828337874659,
+ "grad_norm": 2.048727035522461,
+ "learning_rate": 9.227264552009336e-06,
+ "loss": 0.1906,
+ "step": 19777
+ },
+ {
+ "epoch": 53.89100817438692,
+ "grad_norm": 2.649508476257324,
+ "learning_rate": 9.226384699105094e-06,
+ "loss": 0.0286,
+ "step": 19778
+ },
+ {
+ "epoch": 53.89373297002725,
+ "grad_norm": 2.7260758876800537,
+ "learning_rate": 9.22550485222574e-06,
+ "loss": 0.0502,
+ "step": 19779
+ },
+ {
+ "epoch": 53.89645776566758,
+ "grad_norm": 2.936368465423584,
+ "learning_rate": 9.224625011378134e-06,
+ "loss": 0.0609,
+ "step": 19780
+ },
+ {
+ "epoch": 53.8991825613079,
+ "grad_norm": 2.060107946395874,
+ "learning_rate": 9.223745176569121e-06,
+ "loss": 0.0344,
+ "step": 19781
+ },
+ {
+ "epoch": 53.90190735694823,
+ "grad_norm": 2.2588095664978027,
+ "learning_rate": 9.222865347805559e-06,
+ "loss": 0.0753,
+ "step": 19782
+ },
+ {
+ "epoch": 53.904632152588555,
+ "grad_norm": 1.9193929433822632,
+ "learning_rate": 9.221985525094292e-06,
+ "loss": 0.0372,
+ "step": 19783
+ },
+ {
+ "epoch": 53.90735694822888,
+ "grad_norm": 2.1710832118988037,
+ "learning_rate": 9.221105708442181e-06,
+ "loss": 0.0359,
+ "step": 19784
+ },
+ {
+ "epoch": 53.91008174386921,
+ "grad_norm": 2.4441604614257812,
+ "learning_rate": 9.220225897856071e-06,
+ "loss": 0.0434,
+ "step": 19785
+ },
+ {
+ "epoch": 53.91280653950954,
+ "grad_norm": 2.757624864578247,
+ "learning_rate": 9.21934609334282e-06,
+ "loss": 0.106,
+ "step": 19786
+ },
+ {
+ "epoch": 53.915531335149865,
+ "grad_norm": 2.6174235343933105,
+ "learning_rate": 9.218466294909271e-06,
+ "loss": 0.0735,
+ "step": 19787
+ },
+ {
+ "epoch": 53.91825613079019,
+ "grad_norm": 2.3917012214660645,
+ "learning_rate": 9.217586502562288e-06,
+ "loss": 0.0605,
+ "step": 19788
+ },
+ {
+ "epoch": 53.920980926430516,
+ "grad_norm": 2.114473819732666,
+ "learning_rate": 9.21670671630871e-06,
+ "loss": 0.0361,
+ "step": 19789
+ },
+ {
+ "epoch": 53.92370572207084,
+ "grad_norm": 2.330634593963623,
+ "learning_rate": 9.215826936155396e-06,
+ "loss": 0.0498,
+ "step": 19790
+ },
+ {
+ "epoch": 53.926430517711175,
+ "grad_norm": 2.5290987491607666,
+ "learning_rate": 9.2149471621092e-06,
+ "loss": 0.0667,
+ "step": 19791
+ },
+ {
+ "epoch": 53.9291553133515,
+ "grad_norm": 2.986093282699585,
+ "learning_rate": 9.214067394176967e-06,
+ "loss": 0.1974,
+ "step": 19792
+ },
+ {
+ "epoch": 53.93188010899183,
+ "grad_norm": 2.296387195587158,
+ "learning_rate": 9.213187632365555e-06,
+ "loss": 0.0678,
+ "step": 19793
+ },
+ {
+ "epoch": 53.93460490463215,
+ "grad_norm": 2.079270839691162,
+ "learning_rate": 9.212307876681808e-06,
+ "loss": 0.0732,
+ "step": 19794
+ },
+ {
+ "epoch": 53.93732970027248,
+ "grad_norm": 2.161888599395752,
+ "learning_rate": 9.211428127132587e-06,
+ "loss": 0.0379,
+ "step": 19795
+ },
+ {
+ "epoch": 53.940054495912804,
+ "grad_norm": 2.2117464542388916,
+ "learning_rate": 9.210548383724732e-06,
+ "loss": 0.0338,
+ "step": 19796
+ },
+ {
+ "epoch": 53.94277929155314,
+ "grad_norm": 2.3179192543029785,
+ "learning_rate": 9.209668646465103e-06,
+ "loss": 0.0736,
+ "step": 19797
+ },
+ {
+ "epoch": 53.94550408719346,
+ "grad_norm": 2.724027395248413,
+ "learning_rate": 9.208788915360552e-06,
+ "loss": 0.0431,
+ "step": 19798
+ },
+ {
+ "epoch": 53.94822888283379,
+ "grad_norm": 2.556659698486328,
+ "learning_rate": 9.207909190417924e-06,
+ "loss": 0.2025,
+ "step": 19799
+ },
+ {
+ "epoch": 53.950953678474114,
+ "grad_norm": 2.9058291912078857,
+ "learning_rate": 9.207029471644077e-06,
+ "loss": 0.1966,
+ "step": 19800
+ },
+ {
+ "epoch": 53.95367847411444,
+ "grad_norm": 1.680421233177185,
+ "learning_rate": 9.206149759045859e-06,
+ "loss": 0.0327,
+ "step": 19801
+ },
+ {
+ "epoch": 53.956403269754766,
+ "grad_norm": 2.008385419845581,
+ "learning_rate": 9.20527005263012e-06,
+ "loss": 0.0841,
+ "step": 19802
+ },
+ {
+ "epoch": 53.95912806539509,
+ "grad_norm": 2.234656572341919,
+ "learning_rate": 9.204390352403711e-06,
+ "loss": 0.0906,
+ "step": 19803
+ },
+ {
+ "epoch": 53.961852861035425,
+ "grad_norm": 2.0597879886627197,
+ "learning_rate": 9.203510658373487e-06,
+ "loss": 0.0557,
+ "step": 19804
+ },
+ {
+ "epoch": 53.96457765667575,
+ "grad_norm": 2.0760793685913086,
+ "learning_rate": 9.202630970546293e-06,
+ "loss": 0.0379,
+ "step": 19805
+ },
+ {
+ "epoch": 53.967302452316076,
+ "grad_norm": 1.974612832069397,
+ "learning_rate": 9.201751288928988e-06,
+ "loss": 0.0343,
+ "step": 19806
+ },
+ {
+ "epoch": 53.9700272479564,
+ "grad_norm": 2.6962080001831055,
+ "learning_rate": 9.200871613528414e-06,
+ "loss": 0.0713,
+ "step": 19807
+ },
+ {
+ "epoch": 53.97275204359673,
+ "grad_norm": 2.0434510707855225,
+ "learning_rate": 9.19999194435143e-06,
+ "loss": 0.0393,
+ "step": 19808
+ },
+ {
+ "epoch": 53.97547683923706,
+ "grad_norm": 2.028308153152466,
+ "learning_rate": 9.199112281404878e-06,
+ "loss": 0.0872,
+ "step": 19809
+ },
+ {
+ "epoch": 53.97820163487739,
+ "grad_norm": 2.7676804065704346,
+ "learning_rate": 9.198232624695618e-06,
+ "loss": 0.1003,
+ "step": 19810
+ },
+ {
+ "epoch": 53.98092643051771,
+ "grad_norm": 2.916457414627075,
+ "learning_rate": 9.197352974230494e-06,
+ "loss": 0.1129,
+ "step": 19811
+ },
+ {
+ "epoch": 53.98365122615804,
+ "grad_norm": 2.1364123821258545,
+ "learning_rate": 9.196473330016361e-06,
+ "loss": 0.0946,
+ "step": 19812
+ },
+ {
+ "epoch": 53.986376021798364,
+ "grad_norm": 2.19230055809021,
+ "learning_rate": 9.195593692060062e-06,
+ "loss": 0.113,
+ "step": 19813
+ },
+ {
+ "epoch": 53.98910081743869,
+ "grad_norm": 2.234753131866455,
+ "learning_rate": 9.194714060368458e-06,
+ "loss": 0.0618,
+ "step": 19814
+ },
+ {
+ "epoch": 53.991825613079016,
+ "grad_norm": 3.9343395233154297,
+ "learning_rate": 9.193834434948396e-06,
+ "loss": 0.061,
+ "step": 19815
+ },
+ {
+ "epoch": 53.99455040871935,
+ "grad_norm": 1.7825289964675903,
+ "learning_rate": 9.192954815806724e-06,
+ "loss": 0.0296,
+ "step": 19816
+ },
+ {
+ "epoch": 53.997275204359674,
+ "grad_norm": 5.156072616577148,
+ "learning_rate": 9.192075202950297e-06,
+ "loss": 0.0318,
+ "step": 19817
+ },
+ {
+ "epoch": 54.0,
+ "grad_norm": 2.1622228622436523,
+ "learning_rate": 9.19119559638596e-06,
+ "loss": 0.0519,
+ "step": 19818
+ },
+ {
+ "epoch": 54.002724795640326,
+ "grad_norm": 2.408742904663086,
+ "learning_rate": 9.190315996120565e-06,
+ "loss": 0.1728,
+ "step": 19819
+ },
+ {
+ "epoch": 54.00544959128065,
+ "grad_norm": 2.2695364952087402,
+ "learning_rate": 9.189436402160964e-06,
+ "loss": 0.0641,
+ "step": 19820
+ },
+ {
+ "epoch": 54.00817438692098,
+ "grad_norm": 2.1158225536346436,
+ "learning_rate": 9.188556814514004e-06,
+ "loss": 0.0314,
+ "step": 19821
+ },
+ {
+ "epoch": 54.01089918256131,
+ "grad_norm": 1.5955783128738403,
+ "learning_rate": 9.187677233186541e-06,
+ "loss": 0.028,
+ "step": 19822
+ },
+ {
+ "epoch": 54.013623978201636,
+ "grad_norm": 2.1103403568267822,
+ "learning_rate": 9.18679765818542e-06,
+ "loss": 0.0497,
+ "step": 19823
+ },
+ {
+ "epoch": 54.01634877384196,
+ "grad_norm": 3.103428363800049,
+ "learning_rate": 9.185918089517493e-06,
+ "loss": 0.0645,
+ "step": 19824
+ },
+ {
+ "epoch": 54.01907356948229,
+ "grad_norm": 1.4687799215316772,
+ "learning_rate": 9.18503852718961e-06,
+ "loss": 0.03,
+ "step": 19825
+ },
+ {
+ "epoch": 54.02179836512261,
+ "grad_norm": 2.132404327392578,
+ "learning_rate": 9.18415897120862e-06,
+ "loss": 0.0306,
+ "step": 19826
+ },
+ {
+ "epoch": 54.02452316076294,
+ "grad_norm": 1.913790225982666,
+ "learning_rate": 9.183279421581373e-06,
+ "loss": 0.0513,
+ "step": 19827
+ },
+ {
+ "epoch": 54.02724795640327,
+ "grad_norm": 1.7601040601730347,
+ "learning_rate": 9.18239987831472e-06,
+ "loss": 0.0308,
+ "step": 19828
+ },
+ {
+ "epoch": 54.0299727520436,
+ "grad_norm": 2.6210460662841797,
+ "learning_rate": 9.181520341415511e-06,
+ "loss": 0.1184,
+ "step": 19829
+ },
+ {
+ "epoch": 54.032697547683924,
+ "grad_norm": 2.4804654121398926,
+ "learning_rate": 9.180640810890591e-06,
+ "loss": 0.0819,
+ "step": 19830
+ },
+ {
+ "epoch": 54.03542234332425,
+ "grad_norm": 2.082981586456299,
+ "learning_rate": 9.17976128674682e-06,
+ "loss": 0.1912,
+ "step": 19831
+ },
+ {
+ "epoch": 54.038147138964575,
+ "grad_norm": 2.565772294998169,
+ "learning_rate": 9.178881768991037e-06,
+ "loss": 0.0359,
+ "step": 19832
+ },
+ {
+ "epoch": 54.0408719346049,
+ "grad_norm": 2.132354497909546,
+ "learning_rate": 9.178002257630099e-06,
+ "loss": 0.0676,
+ "step": 19833
+ },
+ {
+ "epoch": 54.043596730245234,
+ "grad_norm": 1.8332815170288086,
+ "learning_rate": 9.177122752670854e-06,
+ "loss": 0.02,
+ "step": 19834
+ },
+ {
+ "epoch": 54.04632152588556,
+ "grad_norm": 3.019965648651123,
+ "learning_rate": 9.176243254120148e-06,
+ "loss": 0.1868,
+ "step": 19835
+ },
+ {
+ "epoch": 54.049046321525886,
+ "grad_norm": 2.0549817085266113,
+ "learning_rate": 9.175363761984836e-06,
+ "loss": 0.0291,
+ "step": 19836
+ },
+ {
+ "epoch": 54.05177111716621,
+ "grad_norm": 1.7415319681167603,
+ "learning_rate": 9.174484276271762e-06,
+ "loss": 0.0336,
+ "step": 19837
+ },
+ {
+ "epoch": 54.05449591280654,
+ "grad_norm": 2.217653751373291,
+ "learning_rate": 9.17360479698778e-06,
+ "loss": 0.0436,
+ "step": 19838
+ },
+ {
+ "epoch": 54.05722070844686,
+ "grad_norm": 2.5792691707611084,
+ "learning_rate": 9.172725324139735e-06,
+ "loss": 0.1343,
+ "step": 19839
+ },
+ {
+ "epoch": 54.059945504087196,
+ "grad_norm": 3.3122971057891846,
+ "learning_rate": 9.17184585773448e-06,
+ "loss": 0.1489,
+ "step": 19840
+ },
+ {
+ "epoch": 54.06267029972752,
+ "grad_norm": 1.7964528799057007,
+ "learning_rate": 9.170966397778862e-06,
+ "loss": 0.0368,
+ "step": 19841
+ },
+ {
+ "epoch": 54.06539509536785,
+ "grad_norm": 2.159963846206665,
+ "learning_rate": 9.170086944279733e-06,
+ "loss": 0.0441,
+ "step": 19842
+ },
+ {
+ "epoch": 54.06811989100817,
+ "grad_norm": 1.9328163862228394,
+ "learning_rate": 9.169207497243939e-06,
+ "loss": 0.0472,
+ "step": 19843
+ },
+ {
+ "epoch": 54.0708446866485,
+ "grad_norm": 2.5520291328430176,
+ "learning_rate": 9.16832805667833e-06,
+ "loss": 0.1099,
+ "step": 19844
+ },
+ {
+ "epoch": 54.073569482288825,
+ "grad_norm": 2.1097450256347656,
+ "learning_rate": 9.167448622589754e-06,
+ "loss": 0.0391,
+ "step": 19845
+ },
+ {
+ "epoch": 54.07629427792916,
+ "grad_norm": 1.7126795053482056,
+ "learning_rate": 9.166569194985065e-06,
+ "loss": 0.029,
+ "step": 19846
+ },
+ {
+ "epoch": 54.079019073569484,
+ "grad_norm": 2.055748701095581,
+ "learning_rate": 9.1656897738711e-06,
+ "loss": 0.1448,
+ "step": 19847
+ },
+ {
+ "epoch": 54.08174386920981,
+ "grad_norm": 1.834686279296875,
+ "learning_rate": 9.164810359254721e-06,
+ "loss": 0.0357,
+ "step": 19848
+ },
+ {
+ "epoch": 54.084468664850135,
+ "grad_norm": 1.8671526908874512,
+ "learning_rate": 9.163930951142777e-06,
+ "loss": 0.0343,
+ "step": 19849
+ },
+ {
+ "epoch": 54.08719346049046,
+ "grad_norm": 2.825345277786255,
+ "learning_rate": 9.163051549542106e-06,
+ "loss": 0.0498,
+ "step": 19850
+ },
+ {
+ "epoch": 54.08991825613079,
+ "grad_norm": 2.136807680130005,
+ "learning_rate": 9.162172154459565e-06,
+ "loss": 0.025,
+ "step": 19851
+ },
+ {
+ "epoch": 54.09264305177112,
+ "grad_norm": 3.2644567489624023,
+ "learning_rate": 9.161292765901998e-06,
+ "loss": 0.0619,
+ "step": 19852
+ },
+ {
+ "epoch": 54.095367847411445,
+ "grad_norm": 2.60627818107605,
+ "learning_rate": 9.16041338387626e-06,
+ "loss": 0.1629,
+ "step": 19853
+ },
+ {
+ "epoch": 54.09809264305177,
+ "grad_norm": 2.0850353240966797,
+ "learning_rate": 9.159534008389193e-06,
+ "loss": 0.0392,
+ "step": 19854
+ },
+ {
+ "epoch": 54.1008174386921,
+ "grad_norm": 2.2050819396972656,
+ "learning_rate": 9.158654639447647e-06,
+ "loss": 0.0384,
+ "step": 19855
+ },
+ {
+ "epoch": 54.10354223433242,
+ "grad_norm": 2.551730155944824,
+ "learning_rate": 9.157775277058474e-06,
+ "loss": 0.0403,
+ "step": 19856
+ },
+ {
+ "epoch": 54.10626702997275,
+ "grad_norm": 2.3822009563446045,
+ "learning_rate": 9.156895921228519e-06,
+ "loss": 0.1806,
+ "step": 19857
+ },
+ {
+ "epoch": 54.10899182561308,
+ "grad_norm": 1.7177506685256958,
+ "learning_rate": 9.156016571964633e-06,
+ "loss": 0.035,
+ "step": 19858
+ },
+ {
+ "epoch": 54.11171662125341,
+ "grad_norm": 2.2272908687591553,
+ "learning_rate": 9.155137229273658e-06,
+ "loss": 0.1151,
+ "step": 19859
+ },
+ {
+ "epoch": 54.11444141689373,
+ "grad_norm": 1.8871304988861084,
+ "learning_rate": 9.154257893162452e-06,
+ "loss": 0.0674,
+ "step": 19860
+ },
+ {
+ "epoch": 54.11716621253406,
+ "grad_norm": 2.5069212913513184,
+ "learning_rate": 9.153378563637855e-06,
+ "loss": 0.058,
+ "step": 19861
+ },
+ {
+ "epoch": 54.119891008174385,
+ "grad_norm": 2.2810046672821045,
+ "learning_rate": 9.152499240706723e-06,
+ "loss": 0.0514,
+ "step": 19862
+ },
+ {
+ "epoch": 54.12261580381471,
+ "grad_norm": 2.57492733001709,
+ "learning_rate": 9.151619924375896e-06,
+ "loss": 0.0322,
+ "step": 19863
+ },
+ {
+ "epoch": 54.12534059945504,
+ "grad_norm": 1.2003593444824219,
+ "learning_rate": 9.150740614652228e-06,
+ "loss": 0.0217,
+ "step": 19864
+ },
+ {
+ "epoch": 54.12806539509537,
+ "grad_norm": 2.07271409034729,
+ "learning_rate": 9.149861311542559e-06,
+ "loss": 0.082,
+ "step": 19865
+ },
+ {
+ "epoch": 54.130790190735695,
+ "grad_norm": 2.7496302127838135,
+ "learning_rate": 9.148982015053746e-06,
+ "loss": 0.1265,
+ "step": 19866
+ },
+ {
+ "epoch": 54.13351498637602,
+ "grad_norm": 2.9683539867401123,
+ "learning_rate": 9.148102725192635e-06,
+ "loss": 0.0861,
+ "step": 19867
+ },
+ {
+ "epoch": 54.13623978201635,
+ "grad_norm": 2.615151882171631,
+ "learning_rate": 9.147223441966072e-06,
+ "loss": 0.0444,
+ "step": 19868
+ },
+ {
+ "epoch": 54.13896457765667,
+ "grad_norm": 2.2258477210998535,
+ "learning_rate": 9.146344165380907e-06,
+ "loss": 0.0365,
+ "step": 19869
+ },
+ {
+ "epoch": 54.141689373297005,
+ "grad_norm": 2.475271463394165,
+ "learning_rate": 9.145464895443983e-06,
+ "loss": 0.0394,
+ "step": 19870
+ },
+ {
+ "epoch": 54.14441416893733,
+ "grad_norm": 1.9952642917633057,
+ "learning_rate": 9.144585632162153e-06,
+ "loss": 0.0267,
+ "step": 19871
+ },
+ {
+ "epoch": 54.14713896457766,
+ "grad_norm": 2.4723424911499023,
+ "learning_rate": 9.143706375542262e-06,
+ "loss": 0.1792,
+ "step": 19872
+ },
+ {
+ "epoch": 54.14986376021798,
+ "grad_norm": 2.1603987216949463,
+ "learning_rate": 9.14282712559116e-06,
+ "loss": 0.0878,
+ "step": 19873
+ },
+ {
+ "epoch": 54.15258855585831,
+ "grad_norm": 2.700655460357666,
+ "learning_rate": 9.141947882315691e-06,
+ "loss": 0.06,
+ "step": 19874
+ },
+ {
+ "epoch": 54.155313351498634,
+ "grad_norm": 2.731328248977661,
+ "learning_rate": 9.141068645722705e-06,
+ "loss": 0.0657,
+ "step": 19875
+ },
+ {
+ "epoch": 54.15803814713897,
+ "grad_norm": 2.919395685195923,
+ "learning_rate": 9.140189415819046e-06,
+ "loss": 0.0857,
+ "step": 19876
+ },
+ {
+ "epoch": 54.16076294277929,
+ "grad_norm": 2.170475482940674,
+ "learning_rate": 9.139310192611565e-06,
+ "loss": 0.0903,
+ "step": 19877
+ },
+ {
+ "epoch": 54.16348773841962,
+ "grad_norm": 2.374829053878784,
+ "learning_rate": 9.138430976107112e-06,
+ "loss": 0.0408,
+ "step": 19878
+ },
+ {
+ "epoch": 54.166212534059945,
+ "grad_norm": 2.7189455032348633,
+ "learning_rate": 9.137551766312528e-06,
+ "loss": 0.0563,
+ "step": 19879
+ },
+ {
+ "epoch": 54.16893732970027,
+ "grad_norm": 2.7829298973083496,
+ "learning_rate": 9.136672563234665e-06,
+ "loss": 0.0518,
+ "step": 19880
+ },
+ {
+ "epoch": 54.171662125340596,
+ "grad_norm": 1.9589145183563232,
+ "learning_rate": 9.135793366880365e-06,
+ "loss": 0.1246,
+ "step": 19881
+ },
+ {
+ "epoch": 54.17438692098093,
+ "grad_norm": 1.2130239009857178,
+ "learning_rate": 9.134914177256477e-06,
+ "loss": 0.0315,
+ "step": 19882
+ },
+ {
+ "epoch": 54.177111716621255,
+ "grad_norm": 2.462796926498413,
+ "learning_rate": 9.134034994369855e-06,
+ "loss": 0.0706,
+ "step": 19883
+ },
+ {
+ "epoch": 54.17983651226158,
+ "grad_norm": 1.9165021181106567,
+ "learning_rate": 9.133155818227338e-06,
+ "loss": 0.0347,
+ "step": 19884
+ },
+ {
+ "epoch": 54.182561307901906,
+ "grad_norm": 1.6894607543945312,
+ "learning_rate": 9.132276648835777e-06,
+ "loss": 0.0872,
+ "step": 19885
+ },
+ {
+ "epoch": 54.18528610354223,
+ "grad_norm": 2.577744245529175,
+ "learning_rate": 9.131397486202016e-06,
+ "loss": 0.1092,
+ "step": 19886
+ },
+ {
+ "epoch": 54.18801089918256,
+ "grad_norm": 2.157041072845459,
+ "learning_rate": 9.130518330332906e-06,
+ "loss": 0.0448,
+ "step": 19887
+ },
+ {
+ "epoch": 54.19073569482289,
+ "grad_norm": 1.9532028436660767,
+ "learning_rate": 9.129639181235287e-06,
+ "loss": 0.0859,
+ "step": 19888
+ },
+ {
+ "epoch": 54.19346049046322,
+ "grad_norm": 2.447089433670044,
+ "learning_rate": 9.128760038916014e-06,
+ "loss": 0.0898,
+ "step": 19889
+ },
+ {
+ "epoch": 54.19618528610354,
+ "grad_norm": 2.157031297683716,
+ "learning_rate": 9.127880903381929e-06,
+ "loss": 0.0504,
+ "step": 19890
+ },
+ {
+ "epoch": 54.19891008174387,
+ "grad_norm": 2.3384406566619873,
+ "learning_rate": 9.127001774639879e-06,
+ "loss": 0.0796,
+ "step": 19891
+ },
+ {
+ "epoch": 54.201634877384194,
+ "grad_norm": 1.791693925857544,
+ "learning_rate": 9.12612265269671e-06,
+ "loss": 0.0502,
+ "step": 19892
+ },
+ {
+ "epoch": 54.20435967302452,
+ "grad_norm": 2.6145577430725098,
+ "learning_rate": 9.125243537559271e-06,
+ "loss": 0.0261,
+ "step": 19893
+ },
+ {
+ "epoch": 54.20708446866485,
+ "grad_norm": 2.6787915229797363,
+ "learning_rate": 9.124364429234405e-06,
+ "loss": 0.2012,
+ "step": 19894
+ },
+ {
+ "epoch": 54.20980926430518,
+ "grad_norm": 2.548933506011963,
+ "learning_rate": 9.123485327728964e-06,
+ "loss": 0.0992,
+ "step": 19895
+ },
+ {
+ "epoch": 54.212534059945504,
+ "grad_norm": 1.2145193815231323,
+ "learning_rate": 9.122606233049788e-06,
+ "loss": 0.0204,
+ "step": 19896
+ },
+ {
+ "epoch": 54.21525885558583,
+ "grad_norm": 2.8234658241271973,
+ "learning_rate": 9.12172714520373e-06,
+ "loss": 0.1364,
+ "step": 19897
+ },
+ {
+ "epoch": 54.217983651226156,
+ "grad_norm": 1.6631338596343994,
+ "learning_rate": 9.120848064197629e-06,
+ "loss": 0.0367,
+ "step": 19898
+ },
+ {
+ "epoch": 54.22070844686648,
+ "grad_norm": 1.7198406457901,
+ "learning_rate": 9.119968990038333e-06,
+ "loss": 0.0336,
+ "step": 19899
+ },
+ {
+ "epoch": 54.223433242506815,
+ "grad_norm": 2.6944754123687744,
+ "learning_rate": 9.119089922732692e-06,
+ "loss": 0.0455,
+ "step": 19900
+ },
+ {
+ "epoch": 54.22615803814714,
+ "grad_norm": 2.233306884765625,
+ "learning_rate": 9.118210862287554e-06,
+ "loss": 0.1239,
+ "step": 19901
+ },
+ {
+ "epoch": 54.228882833787466,
+ "grad_norm": 2.772667646408081,
+ "learning_rate": 9.117331808709758e-06,
+ "loss": 0.107,
+ "step": 19902
+ },
+ {
+ "epoch": 54.23160762942779,
+ "grad_norm": 2.229613780975342,
+ "learning_rate": 9.116452762006156e-06,
+ "loss": 0.0833,
+ "step": 19903
+ },
+ {
+ "epoch": 54.23433242506812,
+ "grad_norm": 2.3356173038482666,
+ "learning_rate": 9.11557372218359e-06,
+ "loss": 0.0621,
+ "step": 19904
+ },
+ {
+ "epoch": 54.237057220708444,
+ "grad_norm": 1.9368072748184204,
+ "learning_rate": 9.114694689248908e-06,
+ "loss": 0.129,
+ "step": 19905
+ },
+ {
+ "epoch": 54.23978201634878,
+ "grad_norm": 2.4166088104248047,
+ "learning_rate": 9.113815663208954e-06,
+ "loss": 0.0682,
+ "step": 19906
+ },
+ {
+ "epoch": 54.2425068119891,
+ "grad_norm": 3.0321199893951416,
+ "learning_rate": 9.112936644070577e-06,
+ "loss": 0.1655,
+ "step": 19907
+ },
+ {
+ "epoch": 54.24523160762943,
+ "grad_norm": 1.534135341644287,
+ "learning_rate": 9.11205763184062e-06,
+ "loss": 0.0274,
+ "step": 19908
+ },
+ {
+ "epoch": 54.247956403269754,
+ "grad_norm": 3.307431697845459,
+ "learning_rate": 9.11117862652593e-06,
+ "loss": 0.0365,
+ "step": 19909
+ },
+ {
+ "epoch": 54.25068119891008,
+ "grad_norm": 2.48236083984375,
+ "learning_rate": 9.11029962813335e-06,
+ "loss": 0.105,
+ "step": 19910
+ },
+ {
+ "epoch": 54.253405994550405,
+ "grad_norm": 2.4416215419769287,
+ "learning_rate": 9.10942063666973e-06,
+ "loss": 0.0537,
+ "step": 19911
+ },
+ {
+ "epoch": 54.25613079019074,
+ "grad_norm": 1.9118107557296753,
+ "learning_rate": 9.108541652141913e-06,
+ "loss": 0.1403,
+ "step": 19912
+ },
+ {
+ "epoch": 54.258855585831064,
+ "grad_norm": 2.461374282836914,
+ "learning_rate": 9.107662674556746e-06,
+ "loss": 0.1246,
+ "step": 19913
+ },
+ {
+ "epoch": 54.26158038147139,
+ "grad_norm": 2.321561813354492,
+ "learning_rate": 9.106783703921073e-06,
+ "loss": 0.0387,
+ "step": 19914
+ },
+ {
+ "epoch": 54.264305177111716,
+ "grad_norm": 2.1776232719421387,
+ "learning_rate": 9.10590474024174e-06,
+ "loss": 0.0519,
+ "step": 19915
+ },
+ {
+ "epoch": 54.26702997275204,
+ "grad_norm": 1.6877409219741821,
+ "learning_rate": 9.105025783525587e-06,
+ "loss": 0.0541,
+ "step": 19916
+ },
+ {
+ "epoch": 54.26975476839237,
+ "grad_norm": 2.0375754833221436,
+ "learning_rate": 9.104146833779467e-06,
+ "loss": 0.0581,
+ "step": 19917
+ },
+ {
+ "epoch": 54.2724795640327,
+ "grad_norm": 2.4059340953826904,
+ "learning_rate": 9.103267891010226e-06,
+ "loss": 0.055,
+ "step": 19918
+ },
+ {
+ "epoch": 54.275204359673026,
+ "grad_norm": 2.4328970909118652,
+ "learning_rate": 9.102388955224703e-06,
+ "loss": 0.0571,
+ "step": 19919
+ },
+ {
+ "epoch": 54.27792915531335,
+ "grad_norm": 2.5222108364105225,
+ "learning_rate": 9.101510026429747e-06,
+ "loss": 0.043,
+ "step": 19920
+ },
+ {
+ "epoch": 54.28065395095368,
+ "grad_norm": 3.599274158477783,
+ "learning_rate": 9.100631104632203e-06,
+ "loss": 0.1382,
+ "step": 19921
+ },
+ {
+ "epoch": 54.283378746594,
+ "grad_norm": 2.4388608932495117,
+ "learning_rate": 9.099752189838912e-06,
+ "loss": 0.1444,
+ "step": 19922
+ },
+ {
+ "epoch": 54.28610354223433,
+ "grad_norm": 2.8404905796051025,
+ "learning_rate": 9.098873282056726e-06,
+ "loss": 0.1308,
+ "step": 19923
+ },
+ {
+ "epoch": 54.28882833787466,
+ "grad_norm": 2.609819173812866,
+ "learning_rate": 9.097994381292483e-06,
+ "loss": 0.1085,
+ "step": 19924
+ },
+ {
+ "epoch": 54.29155313351499,
+ "grad_norm": 2.2193074226379395,
+ "learning_rate": 9.097115487553033e-06,
+ "loss": 0.0516,
+ "step": 19925
+ },
+ {
+ "epoch": 54.294277929155314,
+ "grad_norm": 1.8604563474655151,
+ "learning_rate": 9.096236600845214e-06,
+ "loss": 0.0258,
+ "step": 19926
+ },
+ {
+ "epoch": 54.29700272479564,
+ "grad_norm": 2.225578546524048,
+ "learning_rate": 9.095357721175881e-06,
+ "loss": 0.0937,
+ "step": 19927
+ },
+ {
+ "epoch": 54.299727520435965,
+ "grad_norm": 2.8089938163757324,
+ "learning_rate": 9.09447884855187e-06,
+ "loss": 0.0469,
+ "step": 19928
+ },
+ {
+ "epoch": 54.30245231607629,
+ "grad_norm": 1.2501941919326782,
+ "learning_rate": 9.093599982980029e-06,
+ "loss": 0.0214,
+ "step": 19929
+ },
+ {
+ "epoch": 54.305177111716624,
+ "grad_norm": 2.8465023040771484,
+ "learning_rate": 9.092721124467202e-06,
+ "loss": 0.0802,
+ "step": 19930
+ },
+ {
+ "epoch": 54.30790190735695,
+ "grad_norm": 2.261523962020874,
+ "learning_rate": 9.091842273020233e-06,
+ "loss": 0.0428,
+ "step": 19931
+ },
+ {
+ "epoch": 54.310626702997276,
+ "grad_norm": 2.684704303741455,
+ "learning_rate": 9.090963428645968e-06,
+ "loss": 0.1193,
+ "step": 19932
+ },
+ {
+ "epoch": 54.3133514986376,
+ "grad_norm": 1.4751821756362915,
+ "learning_rate": 9.090084591351247e-06,
+ "loss": 0.0251,
+ "step": 19933
+ },
+ {
+ "epoch": 54.31607629427793,
+ "grad_norm": 2.235544204711914,
+ "learning_rate": 9.089205761142923e-06,
+ "loss": 0.0443,
+ "step": 19934
+ },
+ {
+ "epoch": 54.31880108991825,
+ "grad_norm": 2.068721294403076,
+ "learning_rate": 9.088326938027832e-06,
+ "loss": 0.0536,
+ "step": 19935
+ },
+ {
+ "epoch": 54.321525885558586,
+ "grad_norm": 2.378217935562134,
+ "learning_rate": 9.087448122012824e-06,
+ "loss": 0.352,
+ "step": 19936
+ },
+ {
+ "epoch": 54.32425068119891,
+ "grad_norm": 2.077406644821167,
+ "learning_rate": 9.086569313104739e-06,
+ "loss": 0.042,
+ "step": 19937
+ },
+ {
+ "epoch": 54.32697547683924,
+ "grad_norm": 2.1678919792175293,
+ "learning_rate": 9.085690511310425e-06,
+ "loss": 0.0639,
+ "step": 19938
+ },
+ {
+ "epoch": 54.32970027247956,
+ "grad_norm": 2.04587459564209,
+ "learning_rate": 9.084811716636721e-06,
+ "loss": 0.0271,
+ "step": 19939
+ },
+ {
+ "epoch": 54.33242506811989,
+ "grad_norm": 1.8391417264938354,
+ "learning_rate": 9.083932929090476e-06,
+ "loss": 0.0489,
+ "step": 19940
+ },
+ {
+ "epoch": 54.335149863760215,
+ "grad_norm": 1.8391821384429932,
+ "learning_rate": 9.083054148678531e-06,
+ "loss": 0.0655,
+ "step": 19941
+ },
+ {
+ "epoch": 54.33787465940055,
+ "grad_norm": 2.1930460929870605,
+ "learning_rate": 9.082175375407729e-06,
+ "loss": 0.1165,
+ "step": 19942
+ },
+ {
+ "epoch": 54.34059945504087,
+ "grad_norm": 2.9708385467529297,
+ "learning_rate": 9.081296609284919e-06,
+ "loss": 0.0542,
+ "step": 19943
+ },
+ {
+ "epoch": 54.3433242506812,
+ "grad_norm": 1.818195104598999,
+ "learning_rate": 9.080417850316938e-06,
+ "loss": 0.1009,
+ "step": 19944
+ },
+ {
+ "epoch": 54.346049046321525,
+ "grad_norm": 4.159575939178467,
+ "learning_rate": 9.079539098510637e-06,
+ "loss": 0.0838,
+ "step": 19945
+ },
+ {
+ "epoch": 54.34877384196185,
+ "grad_norm": 2.2902133464813232,
+ "learning_rate": 9.078660353872853e-06,
+ "loss": 0.0812,
+ "step": 19946
+ },
+ {
+ "epoch": 54.35149863760218,
+ "grad_norm": 2.8508026599884033,
+ "learning_rate": 9.077781616410436e-06,
+ "loss": 0.0478,
+ "step": 19947
+ },
+ {
+ "epoch": 54.35422343324251,
+ "grad_norm": 2.074798822402954,
+ "learning_rate": 9.076902886130221e-06,
+ "loss": 0.0546,
+ "step": 19948
+ },
+ {
+ "epoch": 54.356948228882835,
+ "grad_norm": 3.6851930618286133,
+ "learning_rate": 9.076024163039061e-06,
+ "loss": 0.0725,
+ "step": 19949
+ },
+ {
+ "epoch": 54.35967302452316,
+ "grad_norm": 1.6347739696502686,
+ "learning_rate": 9.075145447143792e-06,
+ "loss": 0.0487,
+ "step": 19950
+ },
+ {
+ "epoch": 54.36239782016349,
+ "grad_norm": 1.8811054229736328,
+ "learning_rate": 9.074266738451258e-06,
+ "loss": 0.0318,
+ "step": 19951
+ },
+ {
+ "epoch": 54.36512261580381,
+ "grad_norm": 5.093660831451416,
+ "learning_rate": 9.07338803696831e-06,
+ "loss": 0.0511,
+ "step": 19952
+ },
+ {
+ "epoch": 54.36784741144414,
+ "grad_norm": 2.1455612182617188,
+ "learning_rate": 9.072509342701785e-06,
+ "loss": 0.0443,
+ "step": 19953
+ },
+ {
+ "epoch": 54.37057220708447,
+ "grad_norm": 2.0642316341400146,
+ "learning_rate": 9.071630655658528e-06,
+ "loss": 0.0381,
+ "step": 19954
+ },
+ {
+ "epoch": 54.3732970027248,
+ "grad_norm": 1.9815986156463623,
+ "learning_rate": 9.070751975845378e-06,
+ "loss": 0.032,
+ "step": 19955
+ },
+ {
+ "epoch": 54.37602179836512,
+ "grad_norm": 1.7638202905654907,
+ "learning_rate": 9.069873303269187e-06,
+ "loss": 0.1871,
+ "step": 19956
+ },
+ {
+ "epoch": 54.37874659400545,
+ "grad_norm": 2.46620774269104,
+ "learning_rate": 9.068994637936789e-06,
+ "loss": 0.1451,
+ "step": 19957
+ },
+ {
+ "epoch": 54.381471389645775,
+ "grad_norm": 3.0521368980407715,
+ "learning_rate": 9.068115979855033e-06,
+ "loss": 0.0976,
+ "step": 19958
+ },
+ {
+ "epoch": 54.3841961852861,
+ "grad_norm": 2.0271358489990234,
+ "learning_rate": 9.067237329030758e-06,
+ "loss": 0.0432,
+ "step": 19959
+ },
+ {
+ "epoch": 54.38692098092643,
+ "grad_norm": 1.7110095024108887,
+ "learning_rate": 9.066358685470811e-06,
+ "loss": 0.0355,
+ "step": 19960
+ },
+ {
+ "epoch": 54.38964577656676,
+ "grad_norm": 2.4316437244415283,
+ "learning_rate": 9.06548004918203e-06,
+ "loss": 0.0498,
+ "step": 19961
+ },
+ {
+ "epoch": 54.392370572207085,
+ "grad_norm": 2.338111639022827,
+ "learning_rate": 9.064601420171262e-06,
+ "loss": 0.1535,
+ "step": 19962
+ },
+ {
+ "epoch": 54.39509536784741,
+ "grad_norm": 1.1803510189056396,
+ "learning_rate": 9.063722798445346e-06,
+ "loss": 0.0154,
+ "step": 19963
+ },
+ {
+ "epoch": 54.39782016348774,
+ "grad_norm": 7.260927677154541,
+ "learning_rate": 9.062844184011128e-06,
+ "loss": 0.0544,
+ "step": 19964
+ },
+ {
+ "epoch": 54.40054495912806,
+ "grad_norm": 2.2557883262634277,
+ "learning_rate": 9.06196557687545e-06,
+ "loss": 0.0498,
+ "step": 19965
+ },
+ {
+ "epoch": 54.403269754768395,
+ "grad_norm": 1.740099310874939,
+ "learning_rate": 9.061086977045151e-06,
+ "loss": 0.0422,
+ "step": 19966
+ },
+ {
+ "epoch": 54.40599455040872,
+ "grad_norm": 2.270204782485962,
+ "learning_rate": 9.06020838452708e-06,
+ "loss": 0.0558,
+ "step": 19967
+ },
+ {
+ "epoch": 54.40871934604905,
+ "grad_norm": 3.2670228481292725,
+ "learning_rate": 9.059329799328071e-06,
+ "loss": 0.0433,
+ "step": 19968
+ },
+ {
+ "epoch": 54.41144414168937,
+ "grad_norm": 2.696568250656128,
+ "learning_rate": 9.058451221454974e-06,
+ "loss": 0.0472,
+ "step": 19969
+ },
+ {
+ "epoch": 54.4141689373297,
+ "grad_norm": 2.957484722137451,
+ "learning_rate": 9.057572650914631e-06,
+ "loss": 0.0982,
+ "step": 19970
+ },
+ {
+ "epoch": 54.416893732970024,
+ "grad_norm": 2.5025737285614014,
+ "learning_rate": 9.05669408771388e-06,
+ "loss": 0.0444,
+ "step": 19971
+ },
+ {
+ "epoch": 54.41961852861036,
+ "grad_norm": 1.9417071342468262,
+ "learning_rate": 9.055815531859565e-06,
+ "loss": 0.0411,
+ "step": 19972
+ },
+ {
+ "epoch": 54.42234332425068,
+ "grad_norm": 1.9951194524765015,
+ "learning_rate": 9.054936983358528e-06,
+ "loss": 0.0831,
+ "step": 19973
+ },
+ {
+ "epoch": 54.42506811989101,
+ "grad_norm": 2.3879992961883545,
+ "learning_rate": 9.054058442217613e-06,
+ "loss": 0.0298,
+ "step": 19974
+ },
+ {
+ "epoch": 54.427792915531334,
+ "grad_norm": 3.6897966861724854,
+ "learning_rate": 9.053179908443658e-06,
+ "loss": 0.075,
+ "step": 19975
+ },
+ {
+ "epoch": 54.43051771117166,
+ "grad_norm": 1.9053328037261963,
+ "learning_rate": 9.05230138204351e-06,
+ "loss": 0.2614,
+ "step": 19976
+ },
+ {
+ "epoch": 54.433242506811986,
+ "grad_norm": 2.6902902126312256,
+ "learning_rate": 9.051422863024006e-06,
+ "loss": 0.047,
+ "step": 19977
+ },
+ {
+ "epoch": 54.43596730245232,
+ "grad_norm": 1.8518224954605103,
+ "learning_rate": 9.050544351391992e-06,
+ "loss": 0.0391,
+ "step": 19978
+ },
+ {
+ "epoch": 54.438692098092645,
+ "grad_norm": 1.972528100013733,
+ "learning_rate": 9.049665847154307e-06,
+ "loss": 0.1343,
+ "step": 19979
+ },
+ {
+ "epoch": 54.44141689373297,
+ "grad_norm": 2.715541362762451,
+ "learning_rate": 9.048787350317795e-06,
+ "loss": 0.1824,
+ "step": 19980
+ },
+ {
+ "epoch": 54.444141689373296,
+ "grad_norm": 2.577176570892334,
+ "learning_rate": 9.047908860889295e-06,
+ "loss": 0.1331,
+ "step": 19981
+ },
+ {
+ "epoch": 54.44686648501362,
+ "grad_norm": 2.9279825687408447,
+ "learning_rate": 9.047030378875652e-06,
+ "loss": 0.109,
+ "step": 19982
+ },
+ {
+ "epoch": 54.44959128065395,
+ "grad_norm": 4.871722221374512,
+ "learning_rate": 9.046151904283703e-06,
+ "loss": 0.0476,
+ "step": 19983
+ },
+ {
+ "epoch": 54.45231607629428,
+ "grad_norm": 2.2102649211883545,
+ "learning_rate": 9.045273437120294e-06,
+ "loss": 0.0326,
+ "step": 19984
+ },
+ {
+ "epoch": 54.45504087193461,
+ "grad_norm": 2.1785786151885986,
+ "learning_rate": 9.04439497739226e-06,
+ "loss": 0.0609,
+ "step": 19985
+ },
+ {
+ "epoch": 54.45776566757493,
+ "grad_norm": 2.4301559925079346,
+ "learning_rate": 9.043516525106452e-06,
+ "loss": 0.0534,
+ "step": 19986
+ },
+ {
+ "epoch": 54.46049046321526,
+ "grad_norm": 2.0273239612579346,
+ "learning_rate": 9.042638080269705e-06,
+ "loss": 0.0441,
+ "step": 19987
+ },
+ {
+ "epoch": 54.463215258855584,
+ "grad_norm": 1.7404361963272095,
+ "learning_rate": 9.041759642888866e-06,
+ "loss": 0.0295,
+ "step": 19988
+ },
+ {
+ "epoch": 54.46594005449591,
+ "grad_norm": 2.233821153640747,
+ "learning_rate": 9.040881212970766e-06,
+ "loss": 0.0498,
+ "step": 19989
+ },
+ {
+ "epoch": 54.46866485013624,
+ "grad_norm": 2.9067938327789307,
+ "learning_rate": 9.040002790522257e-06,
+ "loss": 0.0643,
+ "step": 19990
+ },
+ {
+ "epoch": 54.47138964577657,
+ "grad_norm": 1.6771705150604248,
+ "learning_rate": 9.039124375550172e-06,
+ "loss": 0.0938,
+ "step": 19991
+ },
+ {
+ "epoch": 54.474114441416894,
+ "grad_norm": 2.2939810752868652,
+ "learning_rate": 9.038245968061359e-06,
+ "loss": 0.0378,
+ "step": 19992
+ },
+ {
+ "epoch": 54.47683923705722,
+ "grad_norm": 2.221298933029175,
+ "learning_rate": 9.037367568062653e-06,
+ "loss": 0.042,
+ "step": 19993
+ },
+ {
+ "epoch": 54.479564032697546,
+ "grad_norm": 2.3498635292053223,
+ "learning_rate": 9.0364891755609e-06,
+ "loss": 0.0599,
+ "step": 19994
+ },
+ {
+ "epoch": 54.48228882833787,
+ "grad_norm": 2.4432170391082764,
+ "learning_rate": 9.035610790562934e-06,
+ "loss": 0.1372,
+ "step": 19995
+ },
+ {
+ "epoch": 54.485013623978205,
+ "grad_norm": 2.8394508361816406,
+ "learning_rate": 9.034732413075603e-06,
+ "loss": 0.1555,
+ "step": 19996
+ },
+ {
+ "epoch": 54.48773841961853,
+ "grad_norm": 2.3370137214660645,
+ "learning_rate": 9.033854043105743e-06,
+ "loss": 0.0469,
+ "step": 19997
+ },
+ {
+ "epoch": 54.490463215258856,
+ "grad_norm": 1.7658350467681885,
+ "learning_rate": 9.032975680660199e-06,
+ "loss": 0.032,
+ "step": 19998
+ },
+ {
+ "epoch": 54.49318801089918,
+ "grad_norm": 2.249786376953125,
+ "learning_rate": 9.032097325745808e-06,
+ "loss": 0.0334,
+ "step": 19999
+ },
+ {
+ "epoch": 54.49591280653951,
+ "grad_norm": 2.3821280002593994,
+ "learning_rate": 9.031218978369412e-06,
+ "loss": 0.0888,
+ "step": 20000
+ },
+ {
+ "epoch": 54.49863760217983,
+ "grad_norm": 2.176561117172241,
+ "learning_rate": 9.03034063853785e-06,
+ "loss": 0.0374,
+ "step": 20001
+ },
+ {
+ "epoch": 54.50136239782017,
+ "grad_norm": 2.091151475906372,
+ "learning_rate": 9.029462306257964e-06,
+ "loss": 0.0419,
+ "step": 20002
+ },
+ {
+ "epoch": 54.50408719346049,
+ "grad_norm": 8.2155179977417,
+ "learning_rate": 9.028583981536596e-06,
+ "loss": 0.0518,
+ "step": 20003
+ },
+ {
+ "epoch": 54.50681198910082,
+ "grad_norm": 4.130445957183838,
+ "learning_rate": 9.027705664380584e-06,
+ "loss": 0.0565,
+ "step": 20004
+ },
+ {
+ "epoch": 54.509536784741144,
+ "grad_norm": 2.234790086746216,
+ "learning_rate": 9.02682735479677e-06,
+ "loss": 0.111,
+ "step": 20005
+ },
+ {
+ "epoch": 54.51226158038147,
+ "grad_norm": 1.1561241149902344,
+ "learning_rate": 9.025949052791991e-06,
+ "loss": 0.024,
+ "step": 20006
+ },
+ {
+ "epoch": 54.514986376021795,
+ "grad_norm": 2.7396867275238037,
+ "learning_rate": 9.02507075837309e-06,
+ "loss": 0.0571,
+ "step": 20007
+ },
+ {
+ "epoch": 54.51771117166213,
+ "grad_norm": 2.2979090213775635,
+ "learning_rate": 9.024192471546908e-06,
+ "loss": 0.0528,
+ "step": 20008
+ },
+ {
+ "epoch": 54.520435967302454,
+ "grad_norm": 2.3368918895721436,
+ "learning_rate": 9.023314192320283e-06,
+ "loss": 0.143,
+ "step": 20009
+ },
+ {
+ "epoch": 54.52316076294278,
+ "grad_norm": 1.9170209169387817,
+ "learning_rate": 9.022435920700056e-06,
+ "loss": 0.0377,
+ "step": 20010
+ },
+ {
+ "epoch": 54.525885558583106,
+ "grad_norm": 2.592256546020508,
+ "learning_rate": 9.021557656693066e-06,
+ "loss": 0.077,
+ "step": 20011
+ },
+ {
+ "epoch": 54.52861035422343,
+ "grad_norm": 2.395226240158081,
+ "learning_rate": 9.020679400306156e-06,
+ "loss": 0.1745,
+ "step": 20012
+ },
+ {
+ "epoch": 54.53133514986376,
+ "grad_norm": 3.3926138877868652,
+ "learning_rate": 9.019801151546158e-06,
+ "loss": 0.0566,
+ "step": 20013
+ },
+ {
+ "epoch": 54.53405994550409,
+ "grad_norm": 1.8144128322601318,
+ "learning_rate": 9.018922910419922e-06,
+ "loss": 0.0313,
+ "step": 20014
+ },
+ {
+ "epoch": 54.536784741144416,
+ "grad_norm": 2.355710029602051,
+ "learning_rate": 9.01804467693428e-06,
+ "loss": 0.1049,
+ "step": 20015
+ },
+ {
+ "epoch": 54.53950953678474,
+ "grad_norm": 2.1403446197509766,
+ "learning_rate": 9.017166451096077e-06,
+ "loss": 0.0341,
+ "step": 20016
+ },
+ {
+ "epoch": 54.54223433242507,
+ "grad_norm": 2.3192813396453857,
+ "learning_rate": 9.016288232912149e-06,
+ "loss": 0.21,
+ "step": 20017
+ },
+ {
+ "epoch": 54.54495912806539,
+ "grad_norm": 3.363741397857666,
+ "learning_rate": 9.015410022389336e-06,
+ "loss": 0.0642,
+ "step": 20018
+ },
+ {
+ "epoch": 54.54768392370572,
+ "grad_norm": 1.9762569665908813,
+ "learning_rate": 9.014531819534476e-06,
+ "loss": 0.0272,
+ "step": 20019
+ },
+ {
+ "epoch": 54.55040871934605,
+ "grad_norm": 2.2226264476776123,
+ "learning_rate": 9.013653624354412e-06,
+ "loss": 0.0378,
+ "step": 20020
+ },
+ {
+ "epoch": 54.55313351498638,
+ "grad_norm": 1.7604990005493164,
+ "learning_rate": 9.012775436855983e-06,
+ "loss": 0.053,
+ "step": 20021
+ },
+ {
+ "epoch": 54.555858310626704,
+ "grad_norm": 3.0390257835388184,
+ "learning_rate": 9.011897257046025e-06,
+ "loss": 0.2529,
+ "step": 20022
+ },
+ {
+ "epoch": 54.55858310626703,
+ "grad_norm": 3.143223285675049,
+ "learning_rate": 9.011019084931383e-06,
+ "loss": 0.0762,
+ "step": 20023
+ },
+ {
+ "epoch": 54.561307901907355,
+ "grad_norm": 1.3835240602493286,
+ "learning_rate": 9.01014092051889e-06,
+ "loss": 0.0271,
+ "step": 20024
+ },
+ {
+ "epoch": 54.56403269754768,
+ "grad_norm": 2.565128803253174,
+ "learning_rate": 9.009262763815389e-06,
+ "loss": 0.1424,
+ "step": 20025
+ },
+ {
+ "epoch": 54.566757493188014,
+ "grad_norm": 1.8565847873687744,
+ "learning_rate": 9.008384614827716e-06,
+ "loss": 0.0209,
+ "step": 20026
+ },
+ {
+ "epoch": 54.56948228882834,
+ "grad_norm": 1.8213295936584473,
+ "learning_rate": 9.007506473562713e-06,
+ "loss": 0.024,
+ "step": 20027
+ },
+ {
+ "epoch": 54.572207084468666,
+ "grad_norm": 2.738518476486206,
+ "learning_rate": 9.006628340027215e-06,
+ "loss": 0.1337,
+ "step": 20028
+ },
+ {
+ "epoch": 54.57493188010899,
+ "grad_norm": 1.8350962400436401,
+ "learning_rate": 9.005750214228066e-06,
+ "loss": 0.0516,
+ "step": 20029
+ },
+ {
+ "epoch": 54.57765667574932,
+ "grad_norm": 2.6710212230682373,
+ "learning_rate": 9.004872096172103e-06,
+ "loss": 0.1046,
+ "step": 20030
+ },
+ {
+ "epoch": 54.58038147138964,
+ "grad_norm": 1.9752790927886963,
+ "learning_rate": 9.003993985866162e-06,
+ "loss": 0.0348,
+ "step": 20031
+ },
+ {
+ "epoch": 54.583106267029976,
+ "grad_norm": 2.411229372024536,
+ "learning_rate": 9.003115883317085e-06,
+ "loss": 0.1345,
+ "step": 20032
+ },
+ {
+ "epoch": 54.5858310626703,
+ "grad_norm": 1.9754809141159058,
+ "learning_rate": 9.00223778853171e-06,
+ "loss": 0.0602,
+ "step": 20033
+ },
+ {
+ "epoch": 54.58855585831063,
+ "grad_norm": 3.38496470451355,
+ "learning_rate": 9.001359701516876e-06,
+ "loss": 0.1046,
+ "step": 20034
+ },
+ {
+ "epoch": 54.59128065395095,
+ "grad_norm": 2.3185346126556396,
+ "learning_rate": 9.000481622279417e-06,
+ "loss": 0.0419,
+ "step": 20035
+ },
+ {
+ "epoch": 54.59400544959128,
+ "grad_norm": 2.5788278579711914,
+ "learning_rate": 8.999603550826179e-06,
+ "loss": 0.0891,
+ "step": 20036
+ },
+ {
+ "epoch": 54.596730245231605,
+ "grad_norm": 2.192044973373413,
+ "learning_rate": 8.998725487163989e-06,
+ "loss": 0.0411,
+ "step": 20037
+ },
+ {
+ "epoch": 54.59945504087194,
+ "grad_norm": 1.9578498601913452,
+ "learning_rate": 8.997847431299697e-06,
+ "loss": 0.0634,
+ "step": 20038
+ },
+ {
+ "epoch": 54.60217983651226,
+ "grad_norm": 2.37115740776062,
+ "learning_rate": 8.996969383240139e-06,
+ "loss": 0.1483,
+ "step": 20039
+ },
+ {
+ "epoch": 54.60490463215259,
+ "grad_norm": 2.6630725860595703,
+ "learning_rate": 8.996091342992149e-06,
+ "loss": 0.1025,
+ "step": 20040
+ },
+ {
+ "epoch": 54.607629427792915,
+ "grad_norm": 1.7902921438217163,
+ "learning_rate": 8.99521331056257e-06,
+ "loss": 0.035,
+ "step": 20041
+ },
+ {
+ "epoch": 54.61035422343324,
+ "grad_norm": 1.9126635789871216,
+ "learning_rate": 8.994335285958235e-06,
+ "loss": 0.0377,
+ "step": 20042
+ },
+ {
+ "epoch": 54.61307901907357,
+ "grad_norm": 2.8064992427825928,
+ "learning_rate": 8.993457269185985e-06,
+ "loss": 0.1485,
+ "step": 20043
+ },
+ {
+ "epoch": 54.6158038147139,
+ "grad_norm": 4.688305377960205,
+ "learning_rate": 8.992579260252657e-06,
+ "loss": 0.1198,
+ "step": 20044
+ },
+ {
+ "epoch": 54.618528610354225,
+ "grad_norm": 2.033010482788086,
+ "learning_rate": 8.99170125916509e-06,
+ "loss": 0.0965,
+ "step": 20045
+ },
+ {
+ "epoch": 54.62125340599455,
+ "grad_norm": 2.6993203163146973,
+ "learning_rate": 8.99082326593012e-06,
+ "loss": 0.0843,
+ "step": 20046
+ },
+ {
+ "epoch": 54.62397820163488,
+ "grad_norm": 3.3861851692199707,
+ "learning_rate": 8.98994528055459e-06,
+ "loss": 0.1766,
+ "step": 20047
+ },
+ {
+ "epoch": 54.6267029972752,
+ "grad_norm": 2.5927841663360596,
+ "learning_rate": 8.989067303045329e-06,
+ "loss": 0.0981,
+ "step": 20048
+ },
+ {
+ "epoch": 54.62942779291553,
+ "grad_norm": 2.6226038932800293,
+ "learning_rate": 8.988189333409181e-06,
+ "loss": 0.0418,
+ "step": 20049
+ },
+ {
+ "epoch": 54.63215258855586,
+ "grad_norm": 1.865850567817688,
+ "learning_rate": 8.987311371652981e-06,
+ "loss": 0.0617,
+ "step": 20050
+ },
+ {
+ "epoch": 54.63487738419619,
+ "grad_norm": 2.417933702468872,
+ "learning_rate": 8.986433417783568e-06,
+ "loss": 0.05,
+ "step": 20051
+ },
+ {
+ "epoch": 54.63760217983651,
+ "grad_norm": 2.06666898727417,
+ "learning_rate": 8.98555547180778e-06,
+ "loss": 0.0332,
+ "step": 20052
+ },
+ {
+ "epoch": 54.64032697547684,
+ "grad_norm": 3.037860870361328,
+ "learning_rate": 8.98467753373245e-06,
+ "loss": 0.0393,
+ "step": 20053
+ },
+ {
+ "epoch": 54.643051771117165,
+ "grad_norm": 2.7527503967285156,
+ "learning_rate": 8.983799603564419e-06,
+ "loss": 0.0314,
+ "step": 20054
+ },
+ {
+ "epoch": 54.64577656675749,
+ "grad_norm": 2.50146746635437,
+ "learning_rate": 8.982921681310527e-06,
+ "loss": 0.0662,
+ "step": 20055
+ },
+ {
+ "epoch": 54.64850136239782,
+ "grad_norm": 2.462761878967285,
+ "learning_rate": 8.982043766977606e-06,
+ "loss": 0.1031,
+ "step": 20056
+ },
+ {
+ "epoch": 54.65122615803815,
+ "grad_norm": 2.3663573265075684,
+ "learning_rate": 8.981165860572498e-06,
+ "loss": 0.0676,
+ "step": 20057
+ },
+ {
+ "epoch": 54.653950953678475,
+ "grad_norm": 1.921129822731018,
+ "learning_rate": 8.980287962102035e-06,
+ "loss": 0.1055,
+ "step": 20058
+ },
+ {
+ "epoch": 54.6566757493188,
+ "grad_norm": 2.179280996322632,
+ "learning_rate": 8.979410071573059e-06,
+ "loss": 0.1209,
+ "step": 20059
+ },
+ {
+ "epoch": 54.65940054495913,
+ "grad_norm": 2.308009624481201,
+ "learning_rate": 8.978532188992403e-06,
+ "loss": 0.0483,
+ "step": 20060
+ },
+ {
+ "epoch": 54.66212534059945,
+ "grad_norm": 4.115198135375977,
+ "learning_rate": 8.977654314366906e-06,
+ "loss": 0.0557,
+ "step": 20061
+ },
+ {
+ "epoch": 54.664850136239785,
+ "grad_norm": 2.334864377975464,
+ "learning_rate": 8.976776447703402e-06,
+ "loss": 0.1106,
+ "step": 20062
+ },
+ {
+ "epoch": 54.66757493188011,
+ "grad_norm": 10.093795776367188,
+ "learning_rate": 8.975898589008734e-06,
+ "loss": 0.1176,
+ "step": 20063
+ },
+ {
+ "epoch": 54.67029972752044,
+ "grad_norm": 2.1100351810455322,
+ "learning_rate": 8.975020738289732e-06,
+ "loss": 0.1057,
+ "step": 20064
+ },
+ {
+ "epoch": 54.67302452316076,
+ "grad_norm": 2.293112277984619,
+ "learning_rate": 8.974142895553237e-06,
+ "loss": 0.0371,
+ "step": 20065
+ },
+ {
+ "epoch": 54.67574931880109,
+ "grad_norm": 2.1385247707366943,
+ "learning_rate": 8.973265060806083e-06,
+ "loss": 0.0335,
+ "step": 20066
+ },
+ {
+ "epoch": 54.678474114441414,
+ "grad_norm": 2.425642251968384,
+ "learning_rate": 8.972387234055109e-06,
+ "loss": 0.0435,
+ "step": 20067
+ },
+ {
+ "epoch": 54.68119891008175,
+ "grad_norm": 3.223165273666382,
+ "learning_rate": 8.971509415307147e-06,
+ "loss": 0.0583,
+ "step": 20068
+ },
+ {
+ "epoch": 54.68392370572207,
+ "grad_norm": 3.914524555206299,
+ "learning_rate": 8.970631604569041e-06,
+ "loss": 0.1345,
+ "step": 20069
+ },
+ {
+ "epoch": 54.6866485013624,
+ "grad_norm": 1.3357810974121094,
+ "learning_rate": 8.96975380184762e-06,
+ "loss": 0.0226,
+ "step": 20070
+ },
+ {
+ "epoch": 54.689373297002724,
+ "grad_norm": 2.1522393226623535,
+ "learning_rate": 8.968876007149721e-06,
+ "loss": 0.1392,
+ "step": 20071
+ },
+ {
+ "epoch": 54.69209809264305,
+ "grad_norm": 2.5478098392486572,
+ "learning_rate": 8.967998220482184e-06,
+ "loss": 0.0886,
+ "step": 20072
+ },
+ {
+ "epoch": 54.694822888283376,
+ "grad_norm": 2.101048231124878,
+ "learning_rate": 8.967120441851846e-06,
+ "loss": 0.1527,
+ "step": 20073
+ },
+ {
+ "epoch": 54.69754768392371,
+ "grad_norm": 2.329963207244873,
+ "learning_rate": 8.966242671265539e-06,
+ "loss": 0.0685,
+ "step": 20074
+ },
+ {
+ "epoch": 54.700272479564035,
+ "grad_norm": 2.342663526535034,
+ "learning_rate": 8.965364908730103e-06,
+ "loss": 0.0815,
+ "step": 20075
+ },
+ {
+ "epoch": 54.70299727520436,
+ "grad_norm": 3.468040943145752,
+ "learning_rate": 8.96448715425237e-06,
+ "loss": 0.1121,
+ "step": 20076
+ },
+ {
+ "epoch": 54.705722070844686,
+ "grad_norm": 2.4572534561157227,
+ "learning_rate": 8.96360940783918e-06,
+ "loss": 0.0689,
+ "step": 20077
+ },
+ {
+ "epoch": 54.70844686648501,
+ "grad_norm": 2.290416955947876,
+ "learning_rate": 8.962731669497364e-06,
+ "loss": 0.0453,
+ "step": 20078
+ },
+ {
+ "epoch": 54.71117166212534,
+ "grad_norm": 1.4338525533676147,
+ "learning_rate": 8.96185393923376e-06,
+ "loss": 0.019,
+ "step": 20079
+ },
+ {
+ "epoch": 54.71389645776567,
+ "grad_norm": 2.6068410873413086,
+ "learning_rate": 8.960976217055205e-06,
+ "loss": 0.0568,
+ "step": 20080
+ },
+ {
+ "epoch": 54.716621253406,
+ "grad_norm": 2.0071170330047607,
+ "learning_rate": 8.960098502968535e-06,
+ "loss": 0.0518,
+ "step": 20081
+ },
+ {
+ "epoch": 54.71934604904632,
+ "grad_norm": 2.2597815990448,
+ "learning_rate": 8.959220796980583e-06,
+ "loss": 0.0778,
+ "step": 20082
+ },
+ {
+ "epoch": 54.72207084468665,
+ "grad_norm": 2.540914297103882,
+ "learning_rate": 8.958343099098187e-06,
+ "loss": 0.0725,
+ "step": 20083
+ },
+ {
+ "epoch": 54.724795640326974,
+ "grad_norm": 1.9122810363769531,
+ "learning_rate": 8.95746540932818e-06,
+ "loss": 0.0349,
+ "step": 20084
+ },
+ {
+ "epoch": 54.7275204359673,
+ "grad_norm": 2.397792339324951,
+ "learning_rate": 8.956587727677402e-06,
+ "loss": 0.0812,
+ "step": 20085
+ },
+ {
+ "epoch": 54.73024523160763,
+ "grad_norm": 2.2188079357147217,
+ "learning_rate": 8.95571005415268e-06,
+ "loss": 0.1288,
+ "step": 20086
+ },
+ {
+ "epoch": 54.73297002724796,
+ "grad_norm": 2.7192606925964355,
+ "learning_rate": 8.954832388760859e-06,
+ "loss": 0.0861,
+ "step": 20087
+ },
+ {
+ "epoch": 54.735694822888284,
+ "grad_norm": 1.688978910446167,
+ "learning_rate": 8.953954731508765e-06,
+ "loss": 0.0224,
+ "step": 20088
+ },
+ {
+ "epoch": 54.73841961852861,
+ "grad_norm": 2.3782248497009277,
+ "learning_rate": 8.95307708240324e-06,
+ "loss": 0.1034,
+ "step": 20089
+ },
+ {
+ "epoch": 54.741144414168936,
+ "grad_norm": 2.3861238956451416,
+ "learning_rate": 8.95219944145112e-06,
+ "loss": 0.0656,
+ "step": 20090
+ },
+ {
+ "epoch": 54.74386920980926,
+ "grad_norm": 2.2641217708587646,
+ "learning_rate": 8.951321808659232e-06,
+ "loss": 0.151,
+ "step": 20091
+ },
+ {
+ "epoch": 54.746594005449595,
+ "grad_norm": 2.0336108207702637,
+ "learning_rate": 8.950444184034422e-06,
+ "loss": 0.19,
+ "step": 20092
+ },
+ {
+ "epoch": 54.74931880108992,
+ "grad_norm": 1.9101070165634155,
+ "learning_rate": 8.949566567583514e-06,
+ "loss": 0.0457,
+ "step": 20093
+ },
+ {
+ "epoch": 54.752043596730246,
+ "grad_norm": 2.5095882415771484,
+ "learning_rate": 8.948688959313348e-06,
+ "loss": 0.0679,
+ "step": 20094
+ },
+ {
+ "epoch": 54.75476839237057,
+ "grad_norm": 2.77236270904541,
+ "learning_rate": 8.947811359230762e-06,
+ "loss": 0.1781,
+ "step": 20095
+ },
+ {
+ "epoch": 54.7574931880109,
+ "grad_norm": 1.941487193107605,
+ "learning_rate": 8.946933767342584e-06,
+ "loss": 0.024,
+ "step": 20096
+ },
+ {
+ "epoch": 54.76021798365122,
+ "grad_norm": 1.8824816942214966,
+ "learning_rate": 8.946056183655654e-06,
+ "loss": 0.0195,
+ "step": 20097
+ },
+ {
+ "epoch": 54.762942779291556,
+ "grad_norm": 2.526986837387085,
+ "learning_rate": 8.945178608176804e-06,
+ "loss": 0.0426,
+ "step": 20098
+ },
+ {
+ "epoch": 54.76566757493188,
+ "grad_norm": 4.977381229400635,
+ "learning_rate": 8.94430104091287e-06,
+ "loss": 0.0527,
+ "step": 20099
+ },
+ {
+ "epoch": 54.76839237057221,
+ "grad_norm": 1.488520860671997,
+ "learning_rate": 8.943423481870684e-06,
+ "loss": 0.037,
+ "step": 20100
+ },
+ {
+ "epoch": 54.771117166212534,
+ "grad_norm": 2.0489397048950195,
+ "learning_rate": 8.942545931057083e-06,
+ "loss": 0.0619,
+ "step": 20101
+ },
+ {
+ "epoch": 54.77384196185286,
+ "grad_norm": 2.494022846221924,
+ "learning_rate": 8.941668388478898e-06,
+ "loss": 0.1162,
+ "step": 20102
+ },
+ {
+ "epoch": 54.776566757493185,
+ "grad_norm": 1.8415285348892212,
+ "learning_rate": 8.940790854142967e-06,
+ "loss": 0.0413,
+ "step": 20103
+ },
+ {
+ "epoch": 54.77929155313352,
+ "grad_norm": 2.333876848220825,
+ "learning_rate": 8.939913328056123e-06,
+ "loss": 0.0914,
+ "step": 20104
+ },
+ {
+ "epoch": 54.782016348773844,
+ "grad_norm": 2.7908847332000732,
+ "learning_rate": 8.939035810225196e-06,
+ "loss": 0.0451,
+ "step": 20105
+ },
+ {
+ "epoch": 54.78474114441417,
+ "grad_norm": 1.9634920358657837,
+ "learning_rate": 8.93815830065703e-06,
+ "loss": 0.0608,
+ "step": 20106
+ },
+ {
+ "epoch": 54.787465940054496,
+ "grad_norm": 1.7775524854660034,
+ "learning_rate": 8.937280799358449e-06,
+ "loss": 0.1876,
+ "step": 20107
+ },
+ {
+ "epoch": 54.79019073569482,
+ "grad_norm": 2.0622334480285645,
+ "learning_rate": 8.936403306336294e-06,
+ "loss": 0.0351,
+ "step": 20108
+ },
+ {
+ "epoch": 54.79291553133515,
+ "grad_norm": 3.773918628692627,
+ "learning_rate": 8.935525821597393e-06,
+ "loss": 0.1757,
+ "step": 20109
+ },
+ {
+ "epoch": 54.79564032697548,
+ "grad_norm": 3.084001302719116,
+ "learning_rate": 8.934648345148586e-06,
+ "loss": 0.0913,
+ "step": 20110
+ },
+ {
+ "epoch": 54.798365122615806,
+ "grad_norm": 2.1120593547821045,
+ "learning_rate": 8.933770876996701e-06,
+ "loss": 0.0575,
+ "step": 20111
+ },
+ {
+ "epoch": 54.80108991825613,
+ "grad_norm": 2.0362884998321533,
+ "learning_rate": 8.932893417148574e-06,
+ "loss": 0.0931,
+ "step": 20112
+ },
+ {
+ "epoch": 54.80381471389646,
+ "grad_norm": 2.3703789710998535,
+ "learning_rate": 8.932015965611039e-06,
+ "loss": 0.0628,
+ "step": 20113
+ },
+ {
+ "epoch": 54.80653950953678,
+ "grad_norm": 2.344231605529785,
+ "learning_rate": 8.931138522390932e-06,
+ "loss": 0.0569,
+ "step": 20114
+ },
+ {
+ "epoch": 54.80926430517711,
+ "grad_norm": 16.18416404724121,
+ "learning_rate": 8.93026108749508e-06,
+ "loss": 0.0389,
+ "step": 20115
+ },
+ {
+ "epoch": 54.81198910081744,
+ "grad_norm": 2.9628071784973145,
+ "learning_rate": 8.92938366093032e-06,
+ "loss": 0.0486,
+ "step": 20116
+ },
+ {
+ "epoch": 54.81471389645777,
+ "grad_norm": 1.9855397939682007,
+ "learning_rate": 8.928506242703487e-06,
+ "loss": 0.0562,
+ "step": 20117
+ },
+ {
+ "epoch": 54.817438692098094,
+ "grad_norm": 2.4758031368255615,
+ "learning_rate": 8.927628832821414e-06,
+ "loss": 0.0685,
+ "step": 20118
+ },
+ {
+ "epoch": 54.82016348773842,
+ "grad_norm": 2.8759400844573975,
+ "learning_rate": 8.926751431290932e-06,
+ "loss": 0.1303,
+ "step": 20119
+ },
+ {
+ "epoch": 54.822888283378745,
+ "grad_norm": 3.2605504989624023,
+ "learning_rate": 8.925874038118874e-06,
+ "loss": 0.0978,
+ "step": 20120
+ },
+ {
+ "epoch": 54.82561307901907,
+ "grad_norm": 3.6295688152313232,
+ "learning_rate": 8.924996653312078e-06,
+ "loss": 0.0848,
+ "step": 20121
+ },
+ {
+ "epoch": 54.828337874659404,
+ "grad_norm": 1.7623504400253296,
+ "learning_rate": 8.92411927687737e-06,
+ "loss": 0.1299,
+ "step": 20122
+ },
+ {
+ "epoch": 54.83106267029973,
+ "grad_norm": 2.645852565765381,
+ "learning_rate": 8.923241908821585e-06,
+ "loss": 0.048,
+ "step": 20123
+ },
+ {
+ "epoch": 54.833787465940055,
+ "grad_norm": 2.207475423812866,
+ "learning_rate": 8.92236454915156e-06,
+ "loss": 0.1171,
+ "step": 20124
+ },
+ {
+ "epoch": 54.83651226158038,
+ "grad_norm": 2.803717851638794,
+ "learning_rate": 8.921487197874125e-06,
+ "loss": 0.0912,
+ "step": 20125
+ },
+ {
+ "epoch": 54.83923705722071,
+ "grad_norm": 2.638176679611206,
+ "learning_rate": 8.920609854996114e-06,
+ "loss": 0.0896,
+ "step": 20126
+ },
+ {
+ "epoch": 54.84196185286103,
+ "grad_norm": 2.9214019775390625,
+ "learning_rate": 8.919732520524358e-06,
+ "loss": 0.0343,
+ "step": 20127
+ },
+ {
+ "epoch": 54.844686648501366,
+ "grad_norm": 2.1068570613861084,
+ "learning_rate": 8.918855194465692e-06,
+ "loss": 0.0621,
+ "step": 20128
+ },
+ {
+ "epoch": 54.84741144414169,
+ "grad_norm": 2.449967622756958,
+ "learning_rate": 8.917977876826944e-06,
+ "loss": 0.034,
+ "step": 20129
+ },
+ {
+ "epoch": 54.85013623978202,
+ "grad_norm": 2.631528854370117,
+ "learning_rate": 8.917100567614953e-06,
+ "loss": 0.0887,
+ "step": 20130
+ },
+ {
+ "epoch": 54.85286103542234,
+ "grad_norm": 1.545359492301941,
+ "learning_rate": 8.916223266836545e-06,
+ "loss": 0.0249,
+ "step": 20131
+ },
+ {
+ "epoch": 54.85558583106267,
+ "grad_norm": 1.876952886581421,
+ "learning_rate": 8.915345974498555e-06,
+ "loss": 0.0559,
+ "step": 20132
+ },
+ {
+ "epoch": 54.858310626702995,
+ "grad_norm": 2.832508087158203,
+ "learning_rate": 8.914468690607817e-06,
+ "loss": 0.0488,
+ "step": 20133
+ },
+ {
+ "epoch": 54.86103542234333,
+ "grad_norm": 1.8362020254135132,
+ "learning_rate": 8.913591415171164e-06,
+ "loss": 0.1249,
+ "step": 20134
+ },
+ {
+ "epoch": 54.86376021798365,
+ "grad_norm": 3.079726219177246,
+ "learning_rate": 8.912714148195421e-06,
+ "loss": 0.0459,
+ "step": 20135
+ },
+ {
+ "epoch": 54.86648501362398,
+ "grad_norm": 2.887085437774658,
+ "learning_rate": 8.91183688968743e-06,
+ "loss": 0.0892,
+ "step": 20136
+ },
+ {
+ "epoch": 54.869209809264305,
+ "grad_norm": 2.0791168212890625,
+ "learning_rate": 8.910959639654016e-06,
+ "loss": 0.0526,
+ "step": 20137
+ },
+ {
+ "epoch": 54.87193460490463,
+ "grad_norm": 2.110004186630249,
+ "learning_rate": 8.910082398102011e-06,
+ "loss": 0.0726,
+ "step": 20138
+ },
+ {
+ "epoch": 54.87465940054496,
+ "grad_norm": 2.1887946128845215,
+ "learning_rate": 8.909205165038254e-06,
+ "loss": 0.0449,
+ "step": 20139
+ },
+ {
+ "epoch": 54.87738419618529,
+ "grad_norm": 2.3762032985687256,
+ "learning_rate": 8.908327940469566e-06,
+ "loss": 0.0591,
+ "step": 20140
+ },
+ {
+ "epoch": 54.880108991825615,
+ "grad_norm": 2.196963310241699,
+ "learning_rate": 8.907450724402787e-06,
+ "loss": 0.0942,
+ "step": 20141
+ },
+ {
+ "epoch": 54.88283378746594,
+ "grad_norm": 1.8782764673233032,
+ "learning_rate": 8.906573516844749e-06,
+ "loss": 0.0262,
+ "step": 20142
+ },
+ {
+ "epoch": 54.88555858310627,
+ "grad_norm": 2.6664276123046875,
+ "learning_rate": 8.90569631780228e-06,
+ "loss": 0.2343,
+ "step": 20143
+ },
+ {
+ "epoch": 54.88828337874659,
+ "grad_norm": 2.1531665325164795,
+ "learning_rate": 8.904819127282215e-06,
+ "loss": 0.0557,
+ "step": 20144
+ },
+ {
+ "epoch": 54.89100817438692,
+ "grad_norm": 2.3223114013671875,
+ "learning_rate": 8.903941945291381e-06,
+ "loss": 0.0405,
+ "step": 20145
+ },
+ {
+ "epoch": 54.89373297002725,
+ "grad_norm": 2.065804958343506,
+ "learning_rate": 8.903064771836614e-06,
+ "loss": 0.0352,
+ "step": 20146
+ },
+ {
+ "epoch": 54.89645776566758,
+ "grad_norm": 2.5212559700012207,
+ "learning_rate": 8.90218760692474e-06,
+ "loss": 0.0531,
+ "step": 20147
+ },
+ {
+ "epoch": 54.8991825613079,
+ "grad_norm": 3.537992477416992,
+ "learning_rate": 8.901310450562596e-06,
+ "loss": 0.0617,
+ "step": 20148
+ },
+ {
+ "epoch": 54.90190735694823,
+ "grad_norm": 2.1773507595062256,
+ "learning_rate": 8.900433302757011e-06,
+ "loss": 0.141,
+ "step": 20149
+ },
+ {
+ "epoch": 54.904632152588555,
+ "grad_norm": 1.928658127784729,
+ "learning_rate": 8.899556163514816e-06,
+ "loss": 0.0392,
+ "step": 20150
+ },
+ {
+ "epoch": 54.90735694822888,
+ "grad_norm": 3.035580635070801,
+ "learning_rate": 8.898679032842841e-06,
+ "loss": 0.0283,
+ "step": 20151
+ },
+ {
+ "epoch": 54.91008174386921,
+ "grad_norm": 3.1824076175689697,
+ "learning_rate": 8.89780191074792e-06,
+ "loss": 0.3276,
+ "step": 20152
+ },
+ {
+ "epoch": 54.91280653950954,
+ "grad_norm": 2.878652811050415,
+ "learning_rate": 8.89692479723688e-06,
+ "loss": 0.04,
+ "step": 20153
+ },
+ {
+ "epoch": 54.915531335149865,
+ "grad_norm": 2.7689967155456543,
+ "learning_rate": 8.896047692316556e-06,
+ "loss": 0.0385,
+ "step": 20154
+ },
+ {
+ "epoch": 54.91825613079019,
+ "grad_norm": 3.4229397773742676,
+ "learning_rate": 8.895170595993775e-06,
+ "loss": 0.0452,
+ "step": 20155
+ },
+ {
+ "epoch": 54.920980926430516,
+ "grad_norm": 1.7456871271133423,
+ "learning_rate": 8.894293508275372e-06,
+ "loss": 0.028,
+ "step": 20156
+ },
+ {
+ "epoch": 54.92370572207084,
+ "grad_norm": 2.503218412399292,
+ "learning_rate": 8.89341642916817e-06,
+ "loss": 0.0415,
+ "step": 20157
+ },
+ {
+ "epoch": 54.926430517711175,
+ "grad_norm": 2.477001667022705,
+ "learning_rate": 8.89253935867901e-06,
+ "loss": 0.0494,
+ "step": 20158
+ },
+ {
+ "epoch": 54.9291553133515,
+ "grad_norm": 2.5573489665985107,
+ "learning_rate": 8.891662296814718e-06,
+ "loss": 0.0334,
+ "step": 20159
+ },
+ {
+ "epoch": 54.93188010899183,
+ "grad_norm": 102.74515533447266,
+ "learning_rate": 8.890785243582126e-06,
+ "loss": 0.0709,
+ "step": 20160
+ },
+ {
+ "epoch": 54.93460490463215,
+ "grad_norm": 2.331223487854004,
+ "learning_rate": 8.88990819898806e-06,
+ "loss": 0.0575,
+ "step": 20161
+ },
+ {
+ "epoch": 54.93732970027248,
+ "grad_norm": 7.738840103149414,
+ "learning_rate": 8.889031163039354e-06,
+ "loss": 0.1015,
+ "step": 20162
+ },
+ {
+ "epoch": 54.940054495912804,
+ "grad_norm": 2.3014369010925293,
+ "learning_rate": 8.888154135742837e-06,
+ "loss": 0.1495,
+ "step": 20163
+ },
+ {
+ "epoch": 54.94277929155314,
+ "grad_norm": 2.409651279449463,
+ "learning_rate": 8.887277117105342e-06,
+ "loss": 0.0528,
+ "step": 20164
+ },
+ {
+ "epoch": 54.94550408719346,
+ "grad_norm": 3.208707332611084,
+ "learning_rate": 8.886400107133693e-06,
+ "loss": 0.0613,
+ "step": 20165
+ },
+ {
+ "epoch": 54.94822888283379,
+ "grad_norm": 3.95151424407959,
+ "learning_rate": 8.885523105834728e-06,
+ "loss": 0.055,
+ "step": 20166
+ },
+ {
+ "epoch": 54.950953678474114,
+ "grad_norm": 4.362948894500732,
+ "learning_rate": 8.884646113215271e-06,
+ "loss": 0.0861,
+ "step": 20167
+ },
+ {
+ "epoch": 54.95367847411444,
+ "grad_norm": 4.160459518432617,
+ "learning_rate": 8.883769129282157e-06,
+ "loss": 0.1041,
+ "step": 20168
+ },
+ {
+ "epoch": 54.956403269754766,
+ "grad_norm": 4.4721550941467285,
+ "learning_rate": 8.88289215404221e-06,
+ "loss": 0.0661,
+ "step": 20169
+ },
+ {
+ "epoch": 54.95912806539509,
+ "grad_norm": 4.152753829956055,
+ "learning_rate": 8.882015187502266e-06,
+ "loss": 0.0929,
+ "step": 20170
+ },
+ {
+ "epoch": 54.961852861035425,
+ "grad_norm": 5.519103527069092,
+ "learning_rate": 8.881138229669148e-06,
+ "loss": 0.0801,
+ "step": 20171
+ },
+ {
+ "epoch": 54.96457765667575,
+ "grad_norm": 3.4758434295654297,
+ "learning_rate": 8.880261280549693e-06,
+ "loss": 0.0709,
+ "step": 20172
+ },
+ {
+ "epoch": 54.967302452316076,
+ "grad_norm": 3.6811280250549316,
+ "learning_rate": 8.879384340150723e-06,
+ "loss": 0.0445,
+ "step": 20173
+ },
+ {
+ "epoch": 54.9700272479564,
+ "grad_norm": 3.159067392349243,
+ "learning_rate": 8.878507408479071e-06,
+ "loss": 0.0878,
+ "step": 20174
+ },
+ {
+ "epoch": 54.97275204359673,
+ "grad_norm": 4.748511791229248,
+ "learning_rate": 8.877630485541572e-06,
+ "loss": 0.0857,
+ "step": 20175
+ },
+ {
+ "epoch": 54.97547683923706,
+ "grad_norm": 5.368686676025391,
+ "learning_rate": 8.876753571345047e-06,
+ "loss": 0.1061,
+ "step": 20176
+ },
+ {
+ "epoch": 54.97820163487739,
+ "grad_norm": 4.173523426055908,
+ "learning_rate": 8.875876665896332e-06,
+ "loss": 0.1091,
+ "step": 20177
+ },
+ {
+ "epoch": 54.98092643051771,
+ "grad_norm": 4.668532371520996,
+ "learning_rate": 8.874999769202252e-06,
+ "loss": 0.0856,
+ "step": 20178
+ },
+ {
+ "epoch": 54.98365122615804,
+ "grad_norm": 7.759694576263428,
+ "learning_rate": 8.874122881269638e-06,
+ "loss": 0.0606,
+ "step": 20179
+ },
+ {
+ "epoch": 54.986376021798364,
+ "grad_norm": 3.876049757003784,
+ "learning_rate": 8.873246002105318e-06,
+ "loss": 0.0846,
+ "step": 20180
+ },
+ {
+ "epoch": 54.98910081743869,
+ "grad_norm": 6.200564861297607,
+ "learning_rate": 8.872369131716122e-06,
+ "loss": 0.1017,
+ "step": 20181
+ },
+ {
+ "epoch": 54.991825613079016,
+ "grad_norm": 6.355111598968506,
+ "learning_rate": 8.87149227010888e-06,
+ "loss": 0.0667,
+ "step": 20182
+ },
+ {
+ "epoch": 54.99455040871935,
+ "grad_norm": 3.9665040969848633,
+ "learning_rate": 8.870615417290418e-06,
+ "loss": 0.0789,
+ "step": 20183
+ },
+ {
+ "epoch": 54.997275204359674,
+ "grad_norm": 4.39323091506958,
+ "learning_rate": 8.86973857326757e-06,
+ "loss": 0.3198,
+ "step": 20184
+ },
+ {
+ "epoch": 55.0,
+ "grad_norm": 3.1647040843963623,
+ "learning_rate": 8.868861738047158e-06,
+ "loss": 0.2906,
+ "step": 20185
+ },
+ {
+ "epoch": 55.002724795640326,
+ "grad_norm": 3.943697929382324,
+ "learning_rate": 8.867984911636018e-06,
+ "loss": 0.0494,
+ "step": 20186
+ },
+ {
+ "epoch": 55.00544959128065,
+ "grad_norm": 3.1861867904663086,
+ "learning_rate": 8.867108094040971e-06,
+ "loss": 0.133,
+ "step": 20187
+ },
+ {
+ "epoch": 55.00817438692098,
+ "grad_norm": 3.9842445850372314,
+ "learning_rate": 8.866231285268853e-06,
+ "loss": 0.0518,
+ "step": 20188
+ },
+ {
+ "epoch": 55.01089918256131,
+ "grad_norm": 3.7787723541259766,
+ "learning_rate": 8.865354485326488e-06,
+ "loss": 0.0936,
+ "step": 20189
+ },
+ {
+ "epoch": 55.013623978201636,
+ "grad_norm": 6.026356220245361,
+ "learning_rate": 8.864477694220707e-06,
+ "loss": 0.1734,
+ "step": 20190
+ },
+ {
+ "epoch": 55.01634877384196,
+ "grad_norm": 3.7436671257019043,
+ "learning_rate": 8.863600911958332e-06,
+ "loss": 0.1854,
+ "step": 20191
+ },
+ {
+ "epoch": 55.01907356948229,
+ "grad_norm": 4.973622798919678,
+ "learning_rate": 8.862724138546199e-06,
+ "loss": 0.215,
+ "step": 20192
+ },
+ {
+ "epoch": 55.02179836512261,
+ "grad_norm": 4.169313907623291,
+ "learning_rate": 8.861847373991137e-06,
+ "loss": 0.1511,
+ "step": 20193
+ },
+ {
+ "epoch": 55.02452316076294,
+ "grad_norm": 3.9769768714904785,
+ "learning_rate": 8.860970618299969e-06,
+ "loss": 0.0627,
+ "step": 20194
+ },
+ {
+ "epoch": 55.02724795640327,
+ "grad_norm": 3.1294867992401123,
+ "learning_rate": 8.860093871479527e-06,
+ "loss": 0.0792,
+ "step": 20195
+ },
+ {
+ "epoch": 55.0299727520436,
+ "grad_norm": 2.855649948120117,
+ "learning_rate": 8.859217133536634e-06,
+ "loss": 0.0488,
+ "step": 20196
+ },
+ {
+ "epoch": 55.032697547683924,
+ "grad_norm": 2.741525173187256,
+ "learning_rate": 8.858340404478125e-06,
+ "loss": 0.1147,
+ "step": 20197
+ },
+ {
+ "epoch": 55.03542234332425,
+ "grad_norm": 18.70888328552246,
+ "learning_rate": 8.857463684310822e-06,
+ "loss": 0.0975,
+ "step": 20198
+ },
+ {
+ "epoch": 55.038147138964575,
+ "grad_norm": 3.9645609855651855,
+ "learning_rate": 8.856586973041554e-06,
+ "loss": 0.0748,
+ "step": 20199
+ },
+ {
+ "epoch": 55.0408719346049,
+ "grad_norm": 2.7615418434143066,
+ "learning_rate": 8.855710270677152e-06,
+ "loss": 0.04,
+ "step": 20200
+ },
+ {
+ "epoch": 55.043596730245234,
+ "grad_norm": 3.231832981109619,
+ "learning_rate": 8.854833577224441e-06,
+ "loss": 0.0809,
+ "step": 20201
+ },
+ {
+ "epoch": 55.04632152588556,
+ "grad_norm": 3.880399703979492,
+ "learning_rate": 8.853956892690247e-06,
+ "loss": 0.0739,
+ "step": 20202
+ },
+ {
+ "epoch": 55.049046321525886,
+ "grad_norm": 2.858058214187622,
+ "learning_rate": 8.8530802170814e-06,
+ "loss": 0.0535,
+ "step": 20203
+ },
+ {
+ "epoch": 55.05177111716621,
+ "grad_norm": 3.8072454929351807,
+ "learning_rate": 8.852203550404733e-06,
+ "loss": 0.0517,
+ "step": 20204
+ },
+ {
+ "epoch": 55.05449591280654,
+ "grad_norm": 4.619277477264404,
+ "learning_rate": 8.851326892667061e-06,
+ "loss": 0.0878,
+ "step": 20205
+ },
+ {
+ "epoch": 55.05722070844686,
+ "grad_norm": 2.930051565170288,
+ "learning_rate": 8.850450243875223e-06,
+ "loss": 0.1564,
+ "step": 20206
+ },
+ {
+ "epoch": 55.059945504087196,
+ "grad_norm": 3.1861343383789062,
+ "learning_rate": 8.849573604036037e-06,
+ "loss": 0.1396,
+ "step": 20207
+ },
+ {
+ "epoch": 55.06267029972752,
+ "grad_norm": 4.159079074859619,
+ "learning_rate": 8.84869697315634e-06,
+ "loss": 0.0832,
+ "step": 20208
+ },
+ {
+ "epoch": 55.06539509536785,
+ "grad_norm": 3.0841922760009766,
+ "learning_rate": 8.847820351242949e-06,
+ "loss": 0.028,
+ "step": 20209
+ },
+ {
+ "epoch": 55.06811989100817,
+ "grad_norm": 2.564162492752075,
+ "learning_rate": 8.846943738302697e-06,
+ "loss": 0.0485,
+ "step": 20210
+ },
+ {
+ "epoch": 55.0708446866485,
+ "grad_norm": 2.450270414352417,
+ "learning_rate": 8.846067134342412e-06,
+ "loss": 0.053,
+ "step": 20211
+ },
+ {
+ "epoch": 55.073569482288825,
+ "grad_norm": 2.9612932205200195,
+ "learning_rate": 8.845190539368918e-06,
+ "loss": 0.0551,
+ "step": 20212
+ },
+ {
+ "epoch": 55.07629427792916,
+ "grad_norm": 3.672757625579834,
+ "learning_rate": 8.844313953389043e-06,
+ "loss": 0.0472,
+ "step": 20213
+ },
+ {
+ "epoch": 55.079019073569484,
+ "grad_norm": 2.7413313388824463,
+ "learning_rate": 8.843437376409614e-06,
+ "loss": 0.0652,
+ "step": 20214
+ },
+ {
+ "epoch": 55.08174386920981,
+ "grad_norm": 2.9071173667907715,
+ "learning_rate": 8.842560808437459e-06,
+ "loss": 0.0515,
+ "step": 20215
+ },
+ {
+ "epoch": 55.084468664850135,
+ "grad_norm": 2.984543800354004,
+ "learning_rate": 8.841684249479399e-06,
+ "loss": 0.0584,
+ "step": 20216
+ },
+ {
+ "epoch": 55.08719346049046,
+ "grad_norm": 4.243727684020996,
+ "learning_rate": 8.840807699542269e-06,
+ "loss": 0.0601,
+ "step": 20217
+ },
+ {
+ "epoch": 55.08991825613079,
+ "grad_norm": 3.1064791679382324,
+ "learning_rate": 8.839931158632889e-06,
+ "loss": 0.0592,
+ "step": 20218
+ },
+ {
+ "epoch": 55.09264305177112,
+ "grad_norm": 2.4570260047912598,
+ "learning_rate": 8.83905462675809e-06,
+ "loss": 0.1986,
+ "step": 20219
+ },
+ {
+ "epoch": 55.095367847411445,
+ "grad_norm": 2.3606085777282715,
+ "learning_rate": 8.838178103924694e-06,
+ "loss": 0.036,
+ "step": 20220
+ },
+ {
+ "epoch": 55.09809264305177,
+ "grad_norm": 3.4526076316833496,
+ "learning_rate": 8.83730159013953e-06,
+ "loss": 0.1915,
+ "step": 20221
+ },
+ {
+ "epoch": 55.1008174386921,
+ "grad_norm": 9.722543716430664,
+ "learning_rate": 8.836425085409424e-06,
+ "loss": 0.0395,
+ "step": 20222
+ },
+ {
+ "epoch": 55.10354223433242,
+ "grad_norm": 3.0091657638549805,
+ "learning_rate": 8.835548589741204e-06,
+ "loss": 0.0479,
+ "step": 20223
+ },
+ {
+ "epoch": 55.10626702997275,
+ "grad_norm": 4.093650817871094,
+ "learning_rate": 8.834672103141691e-06,
+ "loss": 0.191,
+ "step": 20224
+ },
+ {
+ "epoch": 55.10899182561308,
+ "grad_norm": 3.1652543544769287,
+ "learning_rate": 8.833795625617715e-06,
+ "loss": 0.0975,
+ "step": 20225
+ },
+ {
+ "epoch": 55.11171662125341,
+ "grad_norm": 2.293940305709839,
+ "learning_rate": 8.832919157176098e-06,
+ "loss": 0.1952,
+ "step": 20226
+ },
+ {
+ "epoch": 55.11444141689373,
+ "grad_norm": 2.587665557861328,
+ "learning_rate": 8.832042697823673e-06,
+ "loss": 0.0554,
+ "step": 20227
+ },
+ {
+ "epoch": 55.11716621253406,
+ "grad_norm": 3.2048180103302,
+ "learning_rate": 8.83116624756726e-06,
+ "loss": 0.0605,
+ "step": 20228
+ },
+ {
+ "epoch": 55.119891008174385,
+ "grad_norm": 4.039882183074951,
+ "learning_rate": 8.83028980641369e-06,
+ "loss": 0.1765,
+ "step": 20229
+ },
+ {
+ "epoch": 55.12261580381471,
+ "grad_norm": 2.3961057662963867,
+ "learning_rate": 8.829413374369782e-06,
+ "loss": 0.0379,
+ "step": 20230
+ },
+ {
+ "epoch": 55.12534059945504,
+ "grad_norm": 2.2965104579925537,
+ "learning_rate": 8.828536951442367e-06,
+ "loss": 0.0323,
+ "step": 20231
+ },
+ {
+ "epoch": 55.12806539509537,
+ "grad_norm": 3.056424140930176,
+ "learning_rate": 8.827660537638267e-06,
+ "loss": 0.0792,
+ "step": 20232
+ },
+ {
+ "epoch": 55.130790190735695,
+ "grad_norm": 2.6936888694763184,
+ "learning_rate": 8.826784132964313e-06,
+ "loss": 0.0332,
+ "step": 20233
+ },
+ {
+ "epoch": 55.13351498637602,
+ "grad_norm": 4.096400260925293,
+ "learning_rate": 8.825907737427321e-06,
+ "loss": 0.1388,
+ "step": 20234
+ },
+ {
+ "epoch": 55.13623978201635,
+ "grad_norm": 4.747122764587402,
+ "learning_rate": 8.825031351034125e-06,
+ "loss": 0.115,
+ "step": 20235
+ },
+ {
+ "epoch": 55.13896457765667,
+ "grad_norm": 3.096954345703125,
+ "learning_rate": 8.824154973791545e-06,
+ "loss": 0.0449,
+ "step": 20236
+ },
+ {
+ "epoch": 55.141689373297005,
+ "grad_norm": 2.6540844440460205,
+ "learning_rate": 8.823278605706412e-06,
+ "loss": 0.0657,
+ "step": 20237
+ },
+ {
+ "epoch": 55.14441416893733,
+ "grad_norm": 3.6592347621917725,
+ "learning_rate": 8.822402246785542e-06,
+ "loss": 0.0978,
+ "step": 20238
+ },
+ {
+ "epoch": 55.14713896457766,
+ "grad_norm": 2.3283658027648926,
+ "learning_rate": 8.82152589703577e-06,
+ "loss": 0.0428,
+ "step": 20239
+ },
+ {
+ "epoch": 55.14986376021798,
+ "grad_norm": 3.224055528640747,
+ "learning_rate": 8.820649556463913e-06,
+ "loss": 0.0319,
+ "step": 20240
+ },
+ {
+ "epoch": 55.15258855585831,
+ "grad_norm": 3.95828914642334,
+ "learning_rate": 8.819773225076803e-06,
+ "loss": 0.0442,
+ "step": 20241
+ },
+ {
+ "epoch": 55.155313351498634,
+ "grad_norm": 4.406844139099121,
+ "learning_rate": 8.818896902881257e-06,
+ "loss": 0.0393,
+ "step": 20242
+ },
+ {
+ "epoch": 55.15803814713897,
+ "grad_norm": 3.3675594329833984,
+ "learning_rate": 8.818020589884103e-06,
+ "loss": 0.1556,
+ "step": 20243
+ },
+ {
+ "epoch": 55.16076294277929,
+ "grad_norm": 2.4992847442626953,
+ "learning_rate": 8.817144286092167e-06,
+ "loss": 0.086,
+ "step": 20244
+ },
+ {
+ "epoch": 55.16348773841962,
+ "grad_norm": 3.559192419052124,
+ "learning_rate": 8.816267991512277e-06,
+ "loss": 0.0794,
+ "step": 20245
+ },
+ {
+ "epoch": 55.166212534059945,
+ "grad_norm": 2.852226495742798,
+ "learning_rate": 8.815391706151252e-06,
+ "loss": 0.1097,
+ "step": 20246
+ },
+ {
+ "epoch": 55.16893732970027,
+ "grad_norm": 3.463001251220703,
+ "learning_rate": 8.81451543001592e-06,
+ "loss": 0.0566,
+ "step": 20247
+ },
+ {
+ "epoch": 55.171662125340596,
+ "grad_norm": 2.010502815246582,
+ "learning_rate": 8.8136391631131e-06,
+ "loss": 0.0449,
+ "step": 20248
+ },
+ {
+ "epoch": 55.17438692098093,
+ "grad_norm": 2.0331592559814453,
+ "learning_rate": 8.812762905449624e-06,
+ "loss": 0.1249,
+ "step": 20249
+ },
+ {
+ "epoch": 55.177111716621255,
+ "grad_norm": 2.7877349853515625,
+ "learning_rate": 8.81188665703231e-06,
+ "loss": 0.0898,
+ "step": 20250
+ },
+ {
+ "epoch": 55.17983651226158,
+ "grad_norm": 2.2630615234375,
+ "learning_rate": 8.811010417867987e-06,
+ "loss": 0.0269,
+ "step": 20251
+ },
+ {
+ "epoch": 55.182561307901906,
+ "grad_norm": 3.0941901206970215,
+ "learning_rate": 8.810134187963472e-06,
+ "loss": 0.1211,
+ "step": 20252
+ },
+ {
+ "epoch": 55.18528610354223,
+ "grad_norm": 2.5243773460388184,
+ "learning_rate": 8.809257967325598e-06,
+ "loss": 0.0519,
+ "step": 20253
+ },
+ {
+ "epoch": 55.18801089918256,
+ "grad_norm": 3.460495948791504,
+ "learning_rate": 8.808381755961183e-06,
+ "loss": 0.0676,
+ "step": 20254
+ },
+ {
+ "epoch": 55.19073569482289,
+ "grad_norm": 2.8354990482330322,
+ "learning_rate": 8.807505553877054e-06,
+ "loss": 0.1326,
+ "step": 20255
+ },
+ {
+ "epoch": 55.19346049046322,
+ "grad_norm": 2.5535387992858887,
+ "learning_rate": 8.80662936108003e-06,
+ "loss": 0.0958,
+ "step": 20256
+ },
+ {
+ "epoch": 55.19618528610354,
+ "grad_norm": 1.867370843887329,
+ "learning_rate": 8.805753177576942e-06,
+ "loss": 0.0217,
+ "step": 20257
+ },
+ {
+ "epoch": 55.19891008174387,
+ "grad_norm": 2.0437614917755127,
+ "learning_rate": 8.804877003374606e-06,
+ "loss": 0.0374,
+ "step": 20258
+ },
+ {
+ "epoch": 55.201634877384194,
+ "grad_norm": 2.1501636505126953,
+ "learning_rate": 8.804000838479853e-06,
+ "loss": 0.106,
+ "step": 20259
+ },
+ {
+ "epoch": 55.20435967302452,
+ "grad_norm": 2.1756861209869385,
+ "learning_rate": 8.803124682899498e-06,
+ "loss": 0.0569,
+ "step": 20260
+ },
+ {
+ "epoch": 55.20708446866485,
+ "grad_norm": 2.497002601623535,
+ "learning_rate": 8.802248536640371e-06,
+ "loss": 0.1026,
+ "step": 20261
+ },
+ {
+ "epoch": 55.20980926430518,
+ "grad_norm": 3.059906482696533,
+ "learning_rate": 8.801372399709297e-06,
+ "loss": 0.1682,
+ "step": 20262
+ },
+ {
+ "epoch": 55.212534059945504,
+ "grad_norm": 2.467034339904785,
+ "learning_rate": 8.800496272113094e-06,
+ "loss": 0.0793,
+ "step": 20263
+ },
+ {
+ "epoch": 55.21525885558583,
+ "grad_norm": 1.9022740125656128,
+ "learning_rate": 8.79962015385859e-06,
+ "loss": 0.0962,
+ "step": 20264
+ },
+ {
+ "epoch": 55.217983651226156,
+ "grad_norm": 2.5468101501464844,
+ "learning_rate": 8.798744044952602e-06,
+ "loss": 0.0572,
+ "step": 20265
+ },
+ {
+ "epoch": 55.22070844686648,
+ "grad_norm": 2.7528560161590576,
+ "learning_rate": 8.797867945401957e-06,
+ "loss": 0.0959,
+ "step": 20266
+ },
+ {
+ "epoch": 55.223433242506815,
+ "grad_norm": 2.896012783050537,
+ "learning_rate": 8.79699185521348e-06,
+ "loss": 0.0457,
+ "step": 20267
+ },
+ {
+ "epoch": 55.22615803814714,
+ "grad_norm": 4.581964492797852,
+ "learning_rate": 8.79611577439399e-06,
+ "loss": 0.0651,
+ "step": 20268
+ },
+ {
+ "epoch": 55.228882833787466,
+ "grad_norm": 2.363874673843384,
+ "learning_rate": 8.795239702950314e-06,
+ "loss": 0.0667,
+ "step": 20269
+ },
+ {
+ "epoch": 55.23160762942779,
+ "grad_norm": 2.3791990280151367,
+ "learning_rate": 8.794363640889269e-06,
+ "loss": 0.1421,
+ "step": 20270
+ },
+ {
+ "epoch": 55.23433242506812,
+ "grad_norm": 2.5254099369049072,
+ "learning_rate": 8.793487588217686e-06,
+ "loss": 0.0588,
+ "step": 20271
+ },
+ {
+ "epoch": 55.237057220708444,
+ "grad_norm": 2.7740373611450195,
+ "learning_rate": 8.792611544942378e-06,
+ "loss": 0.2375,
+ "step": 20272
+ },
+ {
+ "epoch": 55.23978201634878,
+ "grad_norm": 2.1755709648132324,
+ "learning_rate": 8.791735511070175e-06,
+ "loss": 0.0338,
+ "step": 20273
+ },
+ {
+ "epoch": 55.2425068119891,
+ "grad_norm": 3.326023578643799,
+ "learning_rate": 8.790859486607896e-06,
+ "loss": 0.0627,
+ "step": 20274
+ },
+ {
+ "epoch": 55.24523160762943,
+ "grad_norm": 2.2725532054901123,
+ "learning_rate": 8.789983471562365e-06,
+ "loss": 0.1114,
+ "step": 20275
+ },
+ {
+ "epoch": 55.247956403269754,
+ "grad_norm": 2.7059264183044434,
+ "learning_rate": 8.789107465940403e-06,
+ "loss": 0.0357,
+ "step": 20276
+ },
+ {
+ "epoch": 55.25068119891008,
+ "grad_norm": 2.5586791038513184,
+ "learning_rate": 8.788231469748831e-06,
+ "loss": 0.2874,
+ "step": 20277
+ },
+ {
+ "epoch": 55.253405994550405,
+ "grad_norm": 3.0360124111175537,
+ "learning_rate": 8.787355482994477e-06,
+ "loss": 0.0491,
+ "step": 20278
+ },
+ {
+ "epoch": 55.25613079019074,
+ "grad_norm": 3.6437947750091553,
+ "learning_rate": 8.786479505684157e-06,
+ "loss": 0.0975,
+ "step": 20279
+ },
+ {
+ "epoch": 55.258855585831064,
+ "grad_norm": 3.7733571529388428,
+ "learning_rate": 8.785603537824699e-06,
+ "loss": 0.0701,
+ "step": 20280
+ },
+ {
+ "epoch": 55.26158038147139,
+ "grad_norm": 3.652235269546509,
+ "learning_rate": 8.784727579422918e-06,
+ "loss": 0.0447,
+ "step": 20281
+ },
+ {
+ "epoch": 55.264305177111716,
+ "grad_norm": 2.444307565689087,
+ "learning_rate": 8.783851630485642e-06,
+ "loss": 0.0303,
+ "step": 20282
+ },
+ {
+ "epoch": 55.26702997275204,
+ "grad_norm": 2.368260622024536,
+ "learning_rate": 8.782975691019689e-06,
+ "loss": 0.0706,
+ "step": 20283
+ },
+ {
+ "epoch": 55.26975476839237,
+ "grad_norm": 2.786334276199341,
+ "learning_rate": 8.782099761031882e-06,
+ "loss": 0.053,
+ "step": 20284
+ },
+ {
+ "epoch": 55.2724795640327,
+ "grad_norm": 3.100574493408203,
+ "learning_rate": 8.781223840529043e-06,
+ "loss": 0.1246,
+ "step": 20285
+ },
+ {
+ "epoch": 55.275204359673026,
+ "grad_norm": 2.567122220993042,
+ "learning_rate": 8.780347929517995e-06,
+ "loss": 0.0473,
+ "step": 20286
+ },
+ {
+ "epoch": 55.27792915531335,
+ "grad_norm": 2.4149653911590576,
+ "learning_rate": 8.779472028005555e-06,
+ "loss": 0.0423,
+ "step": 20287
+ },
+ {
+ "epoch": 55.28065395095368,
+ "grad_norm": 2.228759527206421,
+ "learning_rate": 8.778596135998551e-06,
+ "loss": 0.0334,
+ "step": 20288
+ },
+ {
+ "epoch": 55.283378746594,
+ "grad_norm": 2.2847495079040527,
+ "learning_rate": 8.777720253503799e-06,
+ "loss": 0.0869,
+ "step": 20289
+ },
+ {
+ "epoch": 55.28610354223433,
+ "grad_norm": 3.0842862129211426,
+ "learning_rate": 8.776844380528123e-06,
+ "loss": 0.0916,
+ "step": 20290
+ },
+ {
+ "epoch": 55.28882833787466,
+ "grad_norm": 4.052604675292969,
+ "learning_rate": 8.775968517078343e-06,
+ "loss": 0.2828,
+ "step": 20291
+ },
+ {
+ "epoch": 55.29155313351499,
+ "grad_norm": 2.699522018432617,
+ "learning_rate": 8.77509266316128e-06,
+ "loss": 0.0501,
+ "step": 20292
+ },
+ {
+ "epoch": 55.294277929155314,
+ "grad_norm": 2.8618884086608887,
+ "learning_rate": 8.774216818783758e-06,
+ "loss": 0.0578,
+ "step": 20293
+ },
+ {
+ "epoch": 55.29700272479564,
+ "grad_norm": 1.9673746824264526,
+ "learning_rate": 8.773340983952595e-06,
+ "loss": 0.0313,
+ "step": 20294
+ },
+ {
+ "epoch": 55.299727520435965,
+ "grad_norm": 2.931713581085205,
+ "learning_rate": 8.77246515867461e-06,
+ "loss": 0.0821,
+ "step": 20295
+ },
+ {
+ "epoch": 55.30245231607629,
+ "grad_norm": 2.6576414108276367,
+ "learning_rate": 8.77158934295663e-06,
+ "loss": 0.0796,
+ "step": 20296
+ },
+ {
+ "epoch": 55.305177111716624,
+ "grad_norm": 2.0844202041625977,
+ "learning_rate": 8.770713536805471e-06,
+ "loss": 0.0353,
+ "step": 20297
+ },
+ {
+ "epoch": 55.30790190735695,
+ "grad_norm": 3.0335426330566406,
+ "learning_rate": 8.769837740227959e-06,
+ "loss": 0.0387,
+ "step": 20298
+ },
+ {
+ "epoch": 55.310626702997276,
+ "grad_norm": 2.1454715728759766,
+ "learning_rate": 8.768961953230908e-06,
+ "loss": 0.0442,
+ "step": 20299
+ },
+ {
+ "epoch": 55.3133514986376,
+ "grad_norm": 1.9906611442565918,
+ "learning_rate": 8.768086175821145e-06,
+ "loss": 0.0229,
+ "step": 20300
+ },
+ {
+ "epoch": 55.31607629427793,
+ "grad_norm": 2.3566031455993652,
+ "learning_rate": 8.767210408005483e-06,
+ "loss": 0.0505,
+ "step": 20301
+ },
+ {
+ "epoch": 55.31880108991825,
+ "grad_norm": 2.492905616760254,
+ "learning_rate": 8.76633464979075e-06,
+ "loss": 0.061,
+ "step": 20302
+ },
+ {
+ "epoch": 55.321525885558586,
+ "grad_norm": 2.200145721435547,
+ "learning_rate": 8.76545890118376e-06,
+ "loss": 0.0504,
+ "step": 20303
+ },
+ {
+ "epoch": 55.32425068119891,
+ "grad_norm": 1.9991446733474731,
+ "learning_rate": 8.764583162191338e-06,
+ "loss": 0.1175,
+ "step": 20304
+ },
+ {
+ "epoch": 55.32697547683924,
+ "grad_norm": 2.408846616744995,
+ "learning_rate": 8.763707432820303e-06,
+ "loss": 0.0487,
+ "step": 20305
+ },
+ {
+ "epoch": 55.32970027247956,
+ "grad_norm": 2.162026882171631,
+ "learning_rate": 8.762831713077475e-06,
+ "loss": 0.0492,
+ "step": 20306
+ },
+ {
+ "epoch": 55.33242506811989,
+ "grad_norm": 2.7582178115844727,
+ "learning_rate": 8.761956002969672e-06,
+ "loss": 0.0561,
+ "step": 20307
+ },
+ {
+ "epoch": 55.335149863760215,
+ "grad_norm": 1.7732020616531372,
+ "learning_rate": 8.761080302503718e-06,
+ "loss": 0.0463,
+ "step": 20308
+ },
+ {
+ "epoch": 55.33787465940055,
+ "grad_norm": 24.210363388061523,
+ "learning_rate": 8.760204611686429e-06,
+ "loss": 0.0678,
+ "step": 20309
+ },
+ {
+ "epoch": 55.34059945504087,
+ "grad_norm": 1.5150028467178345,
+ "learning_rate": 8.759328930524628e-06,
+ "loss": 0.0214,
+ "step": 20310
+ },
+ {
+ "epoch": 55.3433242506812,
+ "grad_norm": 2.5485100746154785,
+ "learning_rate": 8.75845325902513e-06,
+ "loss": 0.0434,
+ "step": 20311
+ },
+ {
+ "epoch": 55.346049046321525,
+ "grad_norm": 3.4678285121917725,
+ "learning_rate": 8.757577597194758e-06,
+ "loss": 0.1025,
+ "step": 20312
+ },
+ {
+ "epoch": 55.34877384196185,
+ "grad_norm": 2.435654401779175,
+ "learning_rate": 8.756701945040334e-06,
+ "loss": 0.0609,
+ "step": 20313
+ },
+ {
+ "epoch": 55.35149863760218,
+ "grad_norm": 3.0077953338623047,
+ "learning_rate": 8.755826302568676e-06,
+ "loss": 0.0952,
+ "step": 20314
+ },
+ {
+ "epoch": 55.35422343324251,
+ "grad_norm": 2.2221553325653076,
+ "learning_rate": 8.754950669786602e-06,
+ "loss": 0.0439,
+ "step": 20315
+ },
+ {
+ "epoch": 55.356948228882835,
+ "grad_norm": 2.8174076080322266,
+ "learning_rate": 8.754075046700934e-06,
+ "loss": 0.0476,
+ "step": 20316
+ },
+ {
+ "epoch": 55.35967302452316,
+ "grad_norm": 2.416801691055298,
+ "learning_rate": 8.753199433318486e-06,
+ "loss": 0.0467,
+ "step": 20317
+ },
+ {
+ "epoch": 55.36239782016349,
+ "grad_norm": 2.3192033767700195,
+ "learning_rate": 8.752323829646084e-06,
+ "loss": 0.0655,
+ "step": 20318
+ },
+ {
+ "epoch": 55.36512261580381,
+ "grad_norm": 3.4442925453186035,
+ "learning_rate": 8.751448235690541e-06,
+ "loss": 0.1636,
+ "step": 20319
+ },
+ {
+ "epoch": 55.36784741144414,
+ "grad_norm": 2.8491780757904053,
+ "learning_rate": 8.750572651458682e-06,
+ "loss": 0.1072,
+ "step": 20320
+ },
+ {
+ "epoch": 55.37057220708447,
+ "grad_norm": 2.2357633113861084,
+ "learning_rate": 8.74969707695732e-06,
+ "loss": 0.0323,
+ "step": 20321
+ },
+ {
+ "epoch": 55.3732970027248,
+ "grad_norm": 2.753896713256836,
+ "learning_rate": 8.74882151219328e-06,
+ "loss": 0.0463,
+ "step": 20322
+ },
+ {
+ "epoch": 55.37602179836512,
+ "grad_norm": 3.6318562030792236,
+ "learning_rate": 8.747945957173374e-06,
+ "loss": 0.0975,
+ "step": 20323
+ },
+ {
+ "epoch": 55.37874659400545,
+ "grad_norm": 2.8872921466827393,
+ "learning_rate": 8.747070411904429e-06,
+ "loss": 0.0431,
+ "step": 20324
+ },
+ {
+ "epoch": 55.381471389645775,
+ "grad_norm": 1.8094468116760254,
+ "learning_rate": 8.746194876393255e-06,
+ "loss": 0.0332,
+ "step": 20325
+ },
+ {
+ "epoch": 55.3841961852861,
+ "grad_norm": 3.3519539833068848,
+ "learning_rate": 8.745319350646678e-06,
+ "loss": 0.0717,
+ "step": 20326
+ },
+ {
+ "epoch": 55.38692098092643,
+ "grad_norm": 2.3661301136016846,
+ "learning_rate": 8.744443834671509e-06,
+ "loss": 0.0768,
+ "step": 20327
+ },
+ {
+ "epoch": 55.38964577656676,
+ "grad_norm": 2.988692045211792,
+ "learning_rate": 8.743568328474577e-06,
+ "loss": 0.0541,
+ "step": 20328
+ },
+ {
+ "epoch": 55.392370572207085,
+ "grad_norm": 2.213244915008545,
+ "learning_rate": 8.742692832062688e-06,
+ "loss": 0.0212,
+ "step": 20329
+ },
+ {
+ "epoch": 55.39509536784741,
+ "grad_norm": 4.779205322265625,
+ "learning_rate": 8.74181734544267e-06,
+ "loss": 0.0442,
+ "step": 20330
+ },
+ {
+ "epoch": 55.39782016348774,
+ "grad_norm": 2.9346601963043213,
+ "learning_rate": 8.740941868621337e-06,
+ "loss": 0.0611,
+ "step": 20331
+ },
+ {
+ "epoch": 55.40054495912806,
+ "grad_norm": 3.410214900970459,
+ "learning_rate": 8.740066401605512e-06,
+ "loss": 0.083,
+ "step": 20332
+ },
+ {
+ "epoch": 55.403269754768395,
+ "grad_norm": 14.429762840270996,
+ "learning_rate": 8.739190944402006e-06,
+ "loss": 0.1444,
+ "step": 20333
+ },
+ {
+ "epoch": 55.40599455040872,
+ "grad_norm": 2.5689902305603027,
+ "learning_rate": 8.738315497017642e-06,
+ "loss": 0.0623,
+ "step": 20334
+ },
+ {
+ "epoch": 55.40871934604905,
+ "grad_norm": 3.557610034942627,
+ "learning_rate": 8.737440059459234e-06,
+ "loss": 0.0513,
+ "step": 20335
+ },
+ {
+ "epoch": 55.41144414168937,
+ "grad_norm": 3.1952364444732666,
+ "learning_rate": 8.736564631733606e-06,
+ "loss": 0.1127,
+ "step": 20336
+ },
+ {
+ "epoch": 55.4141689373297,
+ "grad_norm": 7.291930198669434,
+ "learning_rate": 8.735689213847569e-06,
+ "loss": 0.1434,
+ "step": 20337
+ },
+ {
+ "epoch": 55.416893732970024,
+ "grad_norm": 2.82944917678833,
+ "learning_rate": 8.734813805807944e-06,
+ "loss": 0.1651,
+ "step": 20338
+ },
+ {
+ "epoch": 55.41961852861036,
+ "grad_norm": 2.962190628051758,
+ "learning_rate": 8.73393840762155e-06,
+ "loss": 0.151,
+ "step": 20339
+ },
+ {
+ "epoch": 55.42234332425068,
+ "grad_norm": 4.979565143585205,
+ "learning_rate": 8.733063019295202e-06,
+ "loss": 0.073,
+ "step": 20340
+ },
+ {
+ "epoch": 55.42506811989101,
+ "grad_norm": 2.2783262729644775,
+ "learning_rate": 8.732187640835718e-06,
+ "loss": 0.0325,
+ "step": 20341
+ },
+ {
+ "epoch": 55.427792915531334,
+ "grad_norm": 4.4915690422058105,
+ "learning_rate": 8.731312272249916e-06,
+ "loss": 0.0698,
+ "step": 20342
+ },
+ {
+ "epoch": 55.43051771117166,
+ "grad_norm": 3.5131795406341553,
+ "learning_rate": 8.730436913544612e-06,
+ "loss": 0.1687,
+ "step": 20343
+ },
+ {
+ "epoch": 55.433242506811986,
+ "grad_norm": 7.863950252532959,
+ "learning_rate": 8.729561564726627e-06,
+ "loss": 0.1377,
+ "step": 20344
+ },
+ {
+ "epoch": 55.43596730245232,
+ "grad_norm": 6.379323482513428,
+ "learning_rate": 8.728686225802773e-06,
+ "loss": 0.0732,
+ "step": 20345
+ },
+ {
+ "epoch": 55.438692098092645,
+ "grad_norm": 4.014640808105469,
+ "learning_rate": 8.727810896779869e-06,
+ "loss": 0.0523,
+ "step": 20346
+ },
+ {
+ "epoch": 55.44141689373297,
+ "grad_norm": 2.731750011444092,
+ "learning_rate": 8.726935577664737e-06,
+ "loss": 0.0634,
+ "step": 20347
+ },
+ {
+ "epoch": 55.444141689373296,
+ "grad_norm": 3.197571039199829,
+ "learning_rate": 8.726060268464186e-06,
+ "loss": 0.0669,
+ "step": 20348
+ },
+ {
+ "epoch": 55.44686648501362,
+ "grad_norm": 2.935802459716797,
+ "learning_rate": 8.72518496918504e-06,
+ "loss": 0.0933,
+ "step": 20349
+ },
+ {
+ "epoch": 55.44959128065395,
+ "grad_norm": 2.689546585083008,
+ "learning_rate": 8.724309679834109e-06,
+ "loss": 0.0504,
+ "step": 20350
+ },
+ {
+ "epoch": 55.45231607629428,
+ "grad_norm": 2.178647041320801,
+ "learning_rate": 8.723434400418215e-06,
+ "loss": 0.0439,
+ "step": 20351
+ },
+ {
+ "epoch": 55.45504087193461,
+ "grad_norm": 3.182891845703125,
+ "learning_rate": 8.722559130944173e-06,
+ "loss": 0.0995,
+ "step": 20352
+ },
+ {
+ "epoch": 55.45776566757493,
+ "grad_norm": 4.989989280700684,
+ "learning_rate": 8.721683871418797e-06,
+ "loss": 0.089,
+ "step": 20353
+ },
+ {
+ "epoch": 55.46049046321526,
+ "grad_norm": 2.107557773590088,
+ "learning_rate": 8.720808621848911e-06,
+ "loss": 0.1065,
+ "step": 20354
+ },
+ {
+ "epoch": 55.463215258855584,
+ "grad_norm": 2.968999147415161,
+ "learning_rate": 8.719933382241322e-06,
+ "loss": 0.0476,
+ "step": 20355
+ },
+ {
+ "epoch": 55.46594005449591,
+ "grad_norm": 24.377925872802734,
+ "learning_rate": 8.719058152602853e-06,
+ "loss": 0.0655,
+ "step": 20356
+ },
+ {
+ "epoch": 55.46866485013624,
+ "grad_norm": 3.000243663787842,
+ "learning_rate": 8.718182932940316e-06,
+ "loss": 0.1308,
+ "step": 20357
+ },
+ {
+ "epoch": 55.47138964577657,
+ "grad_norm": 4.0126190185546875,
+ "learning_rate": 8.71730772326053e-06,
+ "loss": 0.1054,
+ "step": 20358
+ },
+ {
+ "epoch": 55.474114441416894,
+ "grad_norm": 7.615103721618652,
+ "learning_rate": 8.71643252357031e-06,
+ "loss": 0.1618,
+ "step": 20359
+ },
+ {
+ "epoch": 55.47683923705722,
+ "grad_norm": 3.2556891441345215,
+ "learning_rate": 8.715557333876472e-06,
+ "loss": 0.0764,
+ "step": 20360
+ },
+ {
+ "epoch": 55.479564032697546,
+ "grad_norm": 2.1927247047424316,
+ "learning_rate": 8.714682154185831e-06,
+ "loss": 0.0334,
+ "step": 20361
+ },
+ {
+ "epoch": 55.48228882833787,
+ "grad_norm": 2.015519142150879,
+ "learning_rate": 8.713806984505208e-06,
+ "loss": 0.0458,
+ "step": 20362
+ },
+ {
+ "epoch": 55.485013623978205,
+ "grad_norm": 2.3072845935821533,
+ "learning_rate": 8.712931824841411e-06,
+ "loss": 0.0422,
+ "step": 20363
+ },
+ {
+ "epoch": 55.48773841961853,
+ "grad_norm": 2.692699432373047,
+ "learning_rate": 8.712056675201257e-06,
+ "loss": 0.0583,
+ "step": 20364
+ },
+ {
+ "epoch": 55.490463215258856,
+ "grad_norm": 3.701953887939453,
+ "learning_rate": 8.711181535591569e-06,
+ "loss": 0.0572,
+ "step": 20365
+ },
+ {
+ "epoch": 55.49318801089918,
+ "grad_norm": 9.810215950012207,
+ "learning_rate": 8.710306406019155e-06,
+ "loss": 0.0726,
+ "step": 20366
+ },
+ {
+ "epoch": 55.49591280653951,
+ "grad_norm": 3.4938573837280273,
+ "learning_rate": 8.709431286490837e-06,
+ "loss": 0.1148,
+ "step": 20367
+ },
+ {
+ "epoch": 55.49863760217983,
+ "grad_norm": 5.765433311462402,
+ "learning_rate": 8.708556177013423e-06,
+ "loss": 0.0772,
+ "step": 20368
+ },
+ {
+ "epoch": 55.50136239782017,
+ "grad_norm": 3.4583232402801514,
+ "learning_rate": 8.707681077593735e-06,
+ "loss": 0.2792,
+ "step": 20369
+ },
+ {
+ "epoch": 55.50408719346049,
+ "grad_norm": 2.3087668418884277,
+ "learning_rate": 8.70680598823858e-06,
+ "loss": 0.0433,
+ "step": 20370
+ },
+ {
+ "epoch": 55.50681198910082,
+ "grad_norm": 4.864282608032227,
+ "learning_rate": 8.705930908954786e-06,
+ "loss": 0.1104,
+ "step": 20371
+ },
+ {
+ "epoch": 55.509536784741144,
+ "grad_norm": 2.4622766971588135,
+ "learning_rate": 8.705055839749154e-06,
+ "loss": 0.0315,
+ "step": 20372
+ },
+ {
+ "epoch": 55.51226158038147,
+ "grad_norm": 3.364635705947876,
+ "learning_rate": 8.704180780628508e-06,
+ "loss": 0.0541,
+ "step": 20373
+ },
+ {
+ "epoch": 55.514986376021795,
+ "grad_norm": 2.3641884326934814,
+ "learning_rate": 8.70330573159966e-06,
+ "loss": 0.0636,
+ "step": 20374
+ },
+ {
+ "epoch": 55.51771117166213,
+ "grad_norm": 2.4297351837158203,
+ "learning_rate": 8.702430692669426e-06,
+ "loss": 0.0351,
+ "step": 20375
+ },
+ {
+ "epoch": 55.520435967302454,
+ "grad_norm": 2.244913339614868,
+ "learning_rate": 8.701555663844619e-06,
+ "loss": 0.224,
+ "step": 20376
+ },
+ {
+ "epoch": 55.52316076294278,
+ "grad_norm": 3.4123854637145996,
+ "learning_rate": 8.700680645132053e-06,
+ "loss": 0.2273,
+ "step": 20377
+ },
+ {
+ "epoch": 55.525885558583106,
+ "grad_norm": 3.631024122238159,
+ "learning_rate": 8.699805636538548e-06,
+ "loss": 0.0478,
+ "step": 20378
+ },
+ {
+ "epoch": 55.52861035422343,
+ "grad_norm": 29.903772354125977,
+ "learning_rate": 8.698930638070913e-06,
+ "loss": 0.0379,
+ "step": 20379
+ },
+ {
+ "epoch": 55.53133514986376,
+ "grad_norm": 2.4092772006988525,
+ "learning_rate": 8.698055649735964e-06,
+ "loss": 0.0697,
+ "step": 20380
+ },
+ {
+ "epoch": 55.53405994550409,
+ "grad_norm": 3.0009284019470215,
+ "learning_rate": 8.697180671540513e-06,
+ "loss": 0.0442,
+ "step": 20381
+ },
+ {
+ "epoch": 55.536784741144416,
+ "grad_norm": 2.5581629276275635,
+ "learning_rate": 8.69630570349138e-06,
+ "loss": 0.0722,
+ "step": 20382
+ },
+ {
+ "epoch": 55.53950953678474,
+ "grad_norm": 5.474605083465576,
+ "learning_rate": 8.695430745595377e-06,
+ "loss": 0.1267,
+ "step": 20383
+ },
+ {
+ "epoch": 55.54223433242507,
+ "grad_norm": 2.297297716140747,
+ "learning_rate": 8.694555797859317e-06,
+ "loss": 0.0395,
+ "step": 20384
+ },
+ {
+ "epoch": 55.54495912806539,
+ "grad_norm": 2.714020252227783,
+ "learning_rate": 8.693680860290016e-06,
+ "loss": 0.0575,
+ "step": 20385
+ },
+ {
+ "epoch": 55.54768392370572,
+ "grad_norm": 2.690788507461548,
+ "learning_rate": 8.692805932894281e-06,
+ "loss": 0.2056,
+ "step": 20386
+ },
+ {
+ "epoch": 55.55040871934605,
+ "grad_norm": 4.726029872894287,
+ "learning_rate": 8.691931015678938e-06,
+ "loss": 0.0673,
+ "step": 20387
+ },
+ {
+ "epoch": 55.55313351498638,
+ "grad_norm": 3.712827682495117,
+ "learning_rate": 8.69105610865079e-06,
+ "loss": 0.0576,
+ "step": 20388
+ },
+ {
+ "epoch": 55.555858310626704,
+ "grad_norm": 2.9035720825195312,
+ "learning_rate": 8.690181211816657e-06,
+ "loss": 0.0516,
+ "step": 20389
+ },
+ {
+ "epoch": 55.55858310626703,
+ "grad_norm": 2.467743158340454,
+ "learning_rate": 8.689306325183348e-06,
+ "loss": 0.088,
+ "step": 20390
+ },
+ {
+ "epoch": 55.561307901907355,
+ "grad_norm": 3.452564239501953,
+ "learning_rate": 8.688431448757682e-06,
+ "loss": 0.2426,
+ "step": 20391
+ },
+ {
+ "epoch": 55.56403269754768,
+ "grad_norm": 3.024784564971924,
+ "learning_rate": 8.687556582546467e-06,
+ "loss": 0.1184,
+ "step": 20392
+ },
+ {
+ "epoch": 55.566757493188014,
+ "grad_norm": 2.6938464641571045,
+ "learning_rate": 8.686681726556521e-06,
+ "loss": 0.1678,
+ "step": 20393
+ },
+ {
+ "epoch": 55.56948228882834,
+ "grad_norm": 3.133256196975708,
+ "learning_rate": 8.685806880794654e-06,
+ "loss": 0.1413,
+ "step": 20394
+ },
+ {
+ "epoch": 55.572207084468666,
+ "grad_norm": 5.474874019622803,
+ "learning_rate": 8.684932045267683e-06,
+ "loss": 0.058,
+ "step": 20395
+ },
+ {
+ "epoch": 55.57493188010899,
+ "grad_norm": 3.6744208335876465,
+ "learning_rate": 8.684057219982416e-06,
+ "loss": 0.0514,
+ "step": 20396
+ },
+ {
+ "epoch": 55.57765667574932,
+ "grad_norm": 3.9740896224975586,
+ "learning_rate": 8.683182404945672e-06,
+ "loss": 0.1406,
+ "step": 20397
+ },
+ {
+ "epoch": 55.58038147138964,
+ "grad_norm": 2.322848081588745,
+ "learning_rate": 8.682307600164255e-06,
+ "loss": 0.049,
+ "step": 20398
+ },
+ {
+ "epoch": 55.583106267029976,
+ "grad_norm": 2.8928961753845215,
+ "learning_rate": 8.68143280564499e-06,
+ "loss": 0.0932,
+ "step": 20399
+ },
+ {
+ "epoch": 55.5858310626703,
+ "grad_norm": 2.8043878078460693,
+ "learning_rate": 8.68055802139468e-06,
+ "loss": 0.0374,
+ "step": 20400
+ },
+ {
+ "epoch": 55.58855585831063,
+ "grad_norm": 3.9522464275360107,
+ "learning_rate": 8.679683247420146e-06,
+ "loss": 0.0366,
+ "step": 20401
+ },
+ {
+ "epoch": 55.59128065395095,
+ "grad_norm": 2.6054821014404297,
+ "learning_rate": 8.678808483728194e-06,
+ "loss": 0.0406,
+ "step": 20402
+ },
+ {
+ "epoch": 55.59400544959128,
+ "grad_norm": 3.3808200359344482,
+ "learning_rate": 8.67793373032564e-06,
+ "loss": 0.1836,
+ "step": 20403
+ },
+ {
+ "epoch": 55.596730245231605,
+ "grad_norm": 3.640059471130371,
+ "learning_rate": 8.677058987219294e-06,
+ "loss": 0.0711,
+ "step": 20404
+ },
+ {
+ "epoch": 55.59945504087194,
+ "grad_norm": 3.5811100006103516,
+ "learning_rate": 8.676184254415973e-06,
+ "loss": 0.0681,
+ "step": 20405
+ },
+ {
+ "epoch": 55.60217983651226,
+ "grad_norm": 3.394984245300293,
+ "learning_rate": 8.675309531922484e-06,
+ "loss": 0.152,
+ "step": 20406
+ },
+ {
+ "epoch": 55.60490463215259,
+ "grad_norm": 3.336758613586426,
+ "learning_rate": 8.674434819745643e-06,
+ "loss": 0.0719,
+ "step": 20407
+ },
+ {
+ "epoch": 55.607629427792915,
+ "grad_norm": 2.9202232360839844,
+ "learning_rate": 8.67356011789226e-06,
+ "loss": 0.2254,
+ "step": 20408
+ },
+ {
+ "epoch": 55.61035422343324,
+ "grad_norm": 2.6424624919891357,
+ "learning_rate": 8.672685426369148e-06,
+ "loss": 0.0485,
+ "step": 20409
+ },
+ {
+ "epoch": 55.61307901907357,
+ "grad_norm": 6.119407653808594,
+ "learning_rate": 8.671810745183118e-06,
+ "loss": 0.0473,
+ "step": 20410
+ },
+ {
+ "epoch": 55.6158038147139,
+ "grad_norm": 3.4249775409698486,
+ "learning_rate": 8.670936074340987e-06,
+ "loss": 0.0823,
+ "step": 20411
+ },
+ {
+ "epoch": 55.618528610354225,
+ "grad_norm": 2.6213161945343018,
+ "learning_rate": 8.67006141384956e-06,
+ "loss": 0.0861,
+ "step": 20412
+ },
+ {
+ "epoch": 55.62125340599455,
+ "grad_norm": 2.8514702320098877,
+ "learning_rate": 8.669186763715654e-06,
+ "loss": 0.0778,
+ "step": 20413
+ },
+ {
+ "epoch": 55.62397820163488,
+ "grad_norm": 3.7410075664520264,
+ "learning_rate": 8.668312123946076e-06,
+ "loss": 0.0967,
+ "step": 20414
+ },
+ {
+ "epoch": 55.6267029972752,
+ "grad_norm": 2.6527106761932373,
+ "learning_rate": 8.667437494547637e-06,
+ "loss": 0.0496,
+ "step": 20415
+ },
+ {
+ "epoch": 55.62942779291553,
+ "grad_norm": 2.352189302444458,
+ "learning_rate": 8.666562875527158e-06,
+ "loss": 0.0704,
+ "step": 20416
+ },
+ {
+ "epoch": 55.63215258855586,
+ "grad_norm": 3.263643741607666,
+ "learning_rate": 8.665688266891442e-06,
+ "loss": 0.0859,
+ "step": 20417
+ },
+ {
+ "epoch": 55.63487738419619,
+ "grad_norm": 2.7968125343322754,
+ "learning_rate": 8.664813668647303e-06,
+ "loss": 0.037,
+ "step": 20418
+ },
+ {
+ "epoch": 55.63760217983651,
+ "grad_norm": 3.034111261367798,
+ "learning_rate": 8.663939080801555e-06,
+ "loss": 0.0324,
+ "step": 20419
+ },
+ {
+ "epoch": 55.64032697547684,
+ "grad_norm": 2.7979559898376465,
+ "learning_rate": 8.663064503361003e-06,
+ "loss": 0.0856,
+ "step": 20420
+ },
+ {
+ "epoch": 55.643051771117165,
+ "grad_norm": 2.7677876949310303,
+ "learning_rate": 8.662189936332463e-06,
+ "loss": 0.0828,
+ "step": 20421
+ },
+ {
+ "epoch": 55.64577656675749,
+ "grad_norm": 2.3885369300842285,
+ "learning_rate": 8.661315379722745e-06,
+ "loss": 0.0656,
+ "step": 20422
+ },
+ {
+ "epoch": 55.64850136239782,
+ "grad_norm": 104.0846939086914,
+ "learning_rate": 8.66044083353866e-06,
+ "loss": 0.0687,
+ "step": 20423
+ },
+ {
+ "epoch": 55.65122615803815,
+ "grad_norm": 2.756653308868408,
+ "learning_rate": 8.659566297787016e-06,
+ "loss": 0.0363,
+ "step": 20424
+ },
+ {
+ "epoch": 55.653950953678475,
+ "grad_norm": 2.246764659881592,
+ "learning_rate": 8.658691772474632e-06,
+ "loss": 0.1017,
+ "step": 20425
+ },
+ {
+ "epoch": 55.6566757493188,
+ "grad_norm": 3.3863799571990967,
+ "learning_rate": 8.657817257608308e-06,
+ "loss": 0.0662,
+ "step": 20426
+ },
+ {
+ "epoch": 55.65940054495913,
+ "grad_norm": 4.000697612762451,
+ "learning_rate": 8.656942753194863e-06,
+ "loss": 0.0969,
+ "step": 20427
+ },
+ {
+ "epoch": 55.66212534059945,
+ "grad_norm": 3.3467180728912354,
+ "learning_rate": 8.656068259241102e-06,
+ "loss": 0.0456,
+ "step": 20428
+ },
+ {
+ "epoch": 55.664850136239785,
+ "grad_norm": 3.579864978790283,
+ "learning_rate": 8.655193775753841e-06,
+ "loss": 0.1222,
+ "step": 20429
+ },
+ {
+ "epoch": 55.66757493188011,
+ "grad_norm": 5.391087055206299,
+ "learning_rate": 8.654319302739887e-06,
+ "loss": 0.1021,
+ "step": 20430
+ },
+ {
+ "epoch": 55.67029972752044,
+ "grad_norm": 7.880346775054932,
+ "learning_rate": 8.65344484020605e-06,
+ "loss": 0.1219,
+ "step": 20431
+ },
+ {
+ "epoch": 55.67302452316076,
+ "grad_norm": 3.7400450706481934,
+ "learning_rate": 8.652570388159139e-06,
+ "loss": 0.1323,
+ "step": 20432
+ },
+ {
+ "epoch": 55.67574931880109,
+ "grad_norm": 3.690230369567871,
+ "learning_rate": 8.651695946605967e-06,
+ "loss": 0.1084,
+ "step": 20433
+ },
+ {
+ "epoch": 55.678474114441414,
+ "grad_norm": 3.6763105392456055,
+ "learning_rate": 8.650821515553347e-06,
+ "loss": 0.1763,
+ "step": 20434
+ },
+ {
+ "epoch": 55.68119891008175,
+ "grad_norm": 8.295631408691406,
+ "learning_rate": 8.649947095008083e-06,
+ "loss": 0.1726,
+ "step": 20435
+ },
+ {
+ "epoch": 55.68392370572207,
+ "grad_norm": 5.9327216148376465,
+ "learning_rate": 8.64907268497699e-06,
+ "loss": 0.1041,
+ "step": 20436
+ },
+ {
+ "epoch": 55.6866485013624,
+ "grad_norm": 28.028514862060547,
+ "learning_rate": 8.648198285466874e-06,
+ "loss": 0.1252,
+ "step": 20437
+ },
+ {
+ "epoch": 55.689373297002724,
+ "grad_norm": 23.07686424255371,
+ "learning_rate": 8.647323896484547e-06,
+ "loss": 0.2129,
+ "step": 20438
+ },
+ {
+ "epoch": 55.69209809264305,
+ "grad_norm": 5.927446365356445,
+ "learning_rate": 8.646449518036817e-06,
+ "loss": 0.0915,
+ "step": 20439
+ },
+ {
+ "epoch": 55.694822888283376,
+ "grad_norm": 4.42392635345459,
+ "learning_rate": 8.645575150130494e-06,
+ "loss": 0.1093,
+ "step": 20440
+ },
+ {
+ "epoch": 55.69754768392371,
+ "grad_norm": 4.118308067321777,
+ "learning_rate": 8.64470079277239e-06,
+ "loss": 0.2173,
+ "step": 20441
+ },
+ {
+ "epoch": 55.700272479564035,
+ "grad_norm": 3.302528142929077,
+ "learning_rate": 8.64382644596931e-06,
+ "loss": 0.0686,
+ "step": 20442
+ },
+ {
+ "epoch": 55.70299727520436,
+ "grad_norm": 4.279078960418701,
+ "learning_rate": 8.642952109728069e-06,
+ "loss": 0.0838,
+ "step": 20443
+ },
+ {
+ "epoch": 55.705722070844686,
+ "grad_norm": 4.326389312744141,
+ "learning_rate": 8.642077784055472e-06,
+ "loss": 0.1072,
+ "step": 20444
+ },
+ {
+ "epoch": 55.70844686648501,
+ "grad_norm": 7.119607448577881,
+ "learning_rate": 8.64120346895833e-06,
+ "loss": 0.2893,
+ "step": 20445
+ },
+ {
+ "epoch": 55.71117166212534,
+ "grad_norm": 4.272476673126221,
+ "learning_rate": 8.64032916444345e-06,
+ "loss": 0.118,
+ "step": 20446
+ },
+ {
+ "epoch": 55.71389645776567,
+ "grad_norm": 5.625648498535156,
+ "learning_rate": 8.639454870517646e-06,
+ "loss": 0.1147,
+ "step": 20447
+ },
+ {
+ "epoch": 55.716621253406,
+ "grad_norm": 4.154967308044434,
+ "learning_rate": 8.63858058718772e-06,
+ "loss": 0.103,
+ "step": 20448
+ },
+ {
+ "epoch": 55.71934604904632,
+ "grad_norm": 3.5507259368896484,
+ "learning_rate": 8.637706314460487e-06,
+ "loss": 0.1575,
+ "step": 20449
+ },
+ {
+ "epoch": 55.72207084468665,
+ "grad_norm": 4.0768818855285645,
+ "learning_rate": 8.63683205234275e-06,
+ "loss": 0.0923,
+ "step": 20450
+ },
+ {
+ "epoch": 55.724795640326974,
+ "grad_norm": 2.798717737197876,
+ "learning_rate": 8.63595780084132e-06,
+ "loss": 0.294,
+ "step": 20451
+ },
+ {
+ "epoch": 55.7275204359673,
+ "grad_norm": 4.829318523406982,
+ "learning_rate": 8.635083559963014e-06,
+ "loss": 0.0768,
+ "step": 20452
+ },
+ {
+ "epoch": 55.73024523160763,
+ "grad_norm": 3.929849624633789,
+ "learning_rate": 8.634209329714628e-06,
+ "loss": 0.0604,
+ "step": 20453
+ },
+ {
+ "epoch": 55.73297002724796,
+ "grad_norm": 4.418501853942871,
+ "learning_rate": 8.633335110102979e-06,
+ "loss": 0.1486,
+ "step": 20454
+ },
+ {
+ "epoch": 55.735694822888284,
+ "grad_norm": 3.514481544494629,
+ "learning_rate": 8.632460901134869e-06,
+ "loss": 0.107,
+ "step": 20455
+ },
+ {
+ "epoch": 55.73841961852861,
+ "grad_norm": 3.90201997756958,
+ "learning_rate": 8.631586702817111e-06,
+ "loss": 0.0694,
+ "step": 20456
+ },
+ {
+ "epoch": 55.741144414168936,
+ "grad_norm": 3.74560546875,
+ "learning_rate": 8.630712515156512e-06,
+ "loss": 0.1447,
+ "step": 20457
+ },
+ {
+ "epoch": 55.74386920980926,
+ "grad_norm": 4.011105537414551,
+ "learning_rate": 8.62983833815988e-06,
+ "loss": 0.1991,
+ "step": 20458
+ },
+ {
+ "epoch": 55.746594005449595,
+ "grad_norm": 6.806775093078613,
+ "learning_rate": 8.628964171834023e-06,
+ "loss": 0.1076,
+ "step": 20459
+ },
+ {
+ "epoch": 55.74931880108992,
+ "grad_norm": 4.257572650909424,
+ "learning_rate": 8.628090016185748e-06,
+ "loss": 0.1022,
+ "step": 20460
+ },
+ {
+ "epoch": 55.752043596730246,
+ "grad_norm": 41.50615310668945,
+ "learning_rate": 8.627215871221864e-06,
+ "loss": 0.0879,
+ "step": 20461
+ },
+ {
+ "epoch": 55.75476839237057,
+ "grad_norm": 3.3194193840026855,
+ "learning_rate": 8.62634173694918e-06,
+ "loss": 0.0718,
+ "step": 20462
+ },
+ {
+ "epoch": 55.7574931880109,
+ "grad_norm": 3.9582347869873047,
+ "learning_rate": 8.625467613374501e-06,
+ "loss": 0.1215,
+ "step": 20463
+ },
+ {
+ "epoch": 55.76021798365122,
+ "grad_norm": 2.578914165496826,
+ "learning_rate": 8.624593500504634e-06,
+ "loss": 0.1377,
+ "step": 20464
+ },
+ {
+ "epoch": 55.762942779291556,
+ "grad_norm": 3.5530483722686768,
+ "learning_rate": 8.623719398346394e-06,
+ "loss": 0.164,
+ "step": 20465
+ },
+ {
+ "epoch": 55.76566757493188,
+ "grad_norm": 9.512057304382324,
+ "learning_rate": 8.622845306906577e-06,
+ "loss": 0.0618,
+ "step": 20466
+ },
+ {
+ "epoch": 55.76839237057221,
+ "grad_norm": 2.61985445022583,
+ "learning_rate": 8.621971226191995e-06,
+ "loss": 0.1167,
+ "step": 20467
+ },
+ {
+ "epoch": 55.771117166212534,
+ "grad_norm": 3.001112699508667,
+ "learning_rate": 8.621097156209463e-06,
+ "loss": 0.0472,
+ "step": 20468
+ },
+ {
+ "epoch": 55.77384196185286,
+ "grad_norm": 2.8825292587280273,
+ "learning_rate": 8.62022309696578e-06,
+ "loss": 0.0809,
+ "step": 20469
+ },
+ {
+ "epoch": 55.776566757493185,
+ "grad_norm": 4.011526584625244,
+ "learning_rate": 8.619349048467755e-06,
+ "loss": 0.1662,
+ "step": 20470
+ },
+ {
+ "epoch": 55.77929155313352,
+ "grad_norm": 2.8058223724365234,
+ "learning_rate": 8.618475010722194e-06,
+ "loss": 0.0623,
+ "step": 20471
+ },
+ {
+ "epoch": 55.782016348773844,
+ "grad_norm": 3.4031450748443604,
+ "learning_rate": 8.617600983735908e-06,
+ "loss": 0.0523,
+ "step": 20472
+ },
+ {
+ "epoch": 55.78474114441417,
+ "grad_norm": 5.883828639984131,
+ "learning_rate": 8.6167269675157e-06,
+ "loss": 0.0543,
+ "step": 20473
+ },
+ {
+ "epoch": 55.787465940054496,
+ "grad_norm": 3.5101943016052246,
+ "learning_rate": 8.615852962068377e-06,
+ "loss": 0.0527,
+ "step": 20474
+ },
+ {
+ "epoch": 55.79019073569482,
+ "grad_norm": 3.6151554584503174,
+ "learning_rate": 8.614978967400747e-06,
+ "loss": 0.154,
+ "step": 20475
+ },
+ {
+ "epoch": 55.79291553133515,
+ "grad_norm": 3.2551286220550537,
+ "learning_rate": 8.614104983519617e-06,
+ "loss": 0.1751,
+ "step": 20476
+ },
+ {
+ "epoch": 55.79564032697548,
+ "grad_norm": 4.0339789390563965,
+ "learning_rate": 8.61323101043179e-06,
+ "loss": 0.0514,
+ "step": 20477
+ },
+ {
+ "epoch": 55.798365122615806,
+ "grad_norm": 3.705734968185425,
+ "learning_rate": 8.612357048144079e-06,
+ "loss": 0.0716,
+ "step": 20478
+ },
+ {
+ "epoch": 55.80108991825613,
+ "grad_norm": 3.341442108154297,
+ "learning_rate": 8.611483096663284e-06,
+ "loss": 0.0905,
+ "step": 20479
+ },
+ {
+ "epoch": 55.80381471389646,
+ "grad_norm": 4.540669918060303,
+ "learning_rate": 8.610609155996216e-06,
+ "loss": 0.0905,
+ "step": 20480
+ },
+ {
+ "epoch": 55.80653950953678,
+ "grad_norm": 3.9841837882995605,
+ "learning_rate": 8.609735226149676e-06,
+ "loss": 0.0825,
+ "step": 20481
+ },
+ {
+ "epoch": 55.80926430517711,
+ "grad_norm": 3.9929444789886475,
+ "learning_rate": 8.608861307130477e-06,
+ "loss": 0.0767,
+ "step": 20482
+ },
+ {
+ "epoch": 55.81198910081744,
+ "grad_norm": 3.55513596534729,
+ "learning_rate": 8.607987398945419e-06,
+ "loss": 0.153,
+ "step": 20483
+ },
+ {
+ "epoch": 55.81471389645777,
+ "grad_norm": 6.872756004333496,
+ "learning_rate": 8.607113501601307e-06,
+ "loss": 0.1236,
+ "step": 20484
+ },
+ {
+ "epoch": 55.817438692098094,
+ "grad_norm": 3.0314323902130127,
+ "learning_rate": 8.606239615104953e-06,
+ "loss": 0.057,
+ "step": 20485
+ },
+ {
+ "epoch": 55.82016348773842,
+ "grad_norm": 4.890347957611084,
+ "learning_rate": 8.605365739463163e-06,
+ "loss": 0.0656,
+ "step": 20486
+ },
+ {
+ "epoch": 55.822888283378745,
+ "grad_norm": 4.264582633972168,
+ "learning_rate": 8.604491874682736e-06,
+ "loss": 0.1314,
+ "step": 20487
+ },
+ {
+ "epoch": 55.82561307901907,
+ "grad_norm": 4.397229194641113,
+ "learning_rate": 8.603618020770486e-06,
+ "loss": 0.2053,
+ "step": 20488
+ },
+ {
+ "epoch": 55.828337874659404,
+ "grad_norm": 3.0207324028015137,
+ "learning_rate": 8.602744177733208e-06,
+ "loss": 0.0651,
+ "step": 20489
+ },
+ {
+ "epoch": 55.83106267029973,
+ "grad_norm": 3.2151224613189697,
+ "learning_rate": 8.601870345577719e-06,
+ "loss": 0.0502,
+ "step": 20490
+ },
+ {
+ "epoch": 55.833787465940055,
+ "grad_norm": 4.062087535858154,
+ "learning_rate": 8.600996524310814e-06,
+ "loss": 0.0797,
+ "step": 20491
+ },
+ {
+ "epoch": 55.83651226158038,
+ "grad_norm": 3.6404521465301514,
+ "learning_rate": 8.600122713939307e-06,
+ "loss": 0.0615,
+ "step": 20492
+ },
+ {
+ "epoch": 55.83923705722071,
+ "grad_norm": 4.404694080352783,
+ "learning_rate": 8.599248914469997e-06,
+ "loss": 0.0615,
+ "step": 20493
+ },
+ {
+ "epoch": 55.84196185286103,
+ "grad_norm": 3.428596258163452,
+ "learning_rate": 8.598375125909693e-06,
+ "loss": 0.077,
+ "step": 20494
+ },
+ {
+ "epoch": 55.844686648501366,
+ "grad_norm": 4.678455829620361,
+ "learning_rate": 8.597501348265197e-06,
+ "loss": 0.0555,
+ "step": 20495
+ },
+ {
+ "epoch": 55.84741144414169,
+ "grad_norm": 4.156827926635742,
+ "learning_rate": 8.596627581543317e-06,
+ "loss": 0.0983,
+ "step": 20496
+ },
+ {
+ "epoch": 55.85013623978202,
+ "grad_norm": 4.002730369567871,
+ "learning_rate": 8.595753825750853e-06,
+ "loss": 0.0743,
+ "step": 20497
+ },
+ {
+ "epoch": 55.85286103542234,
+ "grad_norm": 3.235837697982788,
+ "learning_rate": 8.594880080894616e-06,
+ "loss": 0.0652,
+ "step": 20498
+ },
+ {
+ "epoch": 55.85558583106267,
+ "grad_norm": 2.5663905143737793,
+ "learning_rate": 8.594006346981406e-06,
+ "loss": 0.0721,
+ "step": 20499
+ },
+ {
+ "epoch": 55.858310626702995,
+ "grad_norm": 4.276483535766602,
+ "learning_rate": 8.59313262401803e-06,
+ "loss": 0.1069,
+ "step": 20500
+ },
+ {
+ "epoch": 55.86103542234333,
+ "grad_norm": 3.3870081901550293,
+ "learning_rate": 8.59225891201129e-06,
+ "loss": 0.0552,
+ "step": 20501
+ },
+ {
+ "epoch": 55.86376021798365,
+ "grad_norm": 3.3666799068450928,
+ "learning_rate": 8.591385210967993e-06,
+ "loss": 0.1186,
+ "step": 20502
+ },
+ {
+ "epoch": 55.86648501362398,
+ "grad_norm": 8.544636726379395,
+ "learning_rate": 8.590511520894946e-06,
+ "loss": 0.0933,
+ "step": 20503
+ },
+ {
+ "epoch": 55.869209809264305,
+ "grad_norm": 3.681668758392334,
+ "learning_rate": 8.589637841798945e-06,
+ "loss": 0.0852,
+ "step": 20504
+ },
+ {
+ "epoch": 55.87193460490463,
+ "grad_norm": 2.910865306854248,
+ "learning_rate": 8.5887641736868e-06,
+ "loss": 0.056,
+ "step": 20505
+ },
+ {
+ "epoch": 55.87465940054496,
+ "grad_norm": 3.214862823486328,
+ "learning_rate": 8.587890516565319e-06,
+ "loss": 0.0768,
+ "step": 20506
+ },
+ {
+ "epoch": 55.87738419618529,
+ "grad_norm": 2.4955966472625732,
+ "learning_rate": 8.587016870441297e-06,
+ "loss": 0.0513,
+ "step": 20507
+ },
+ {
+ "epoch": 55.880108991825615,
+ "grad_norm": 2.752106189727783,
+ "learning_rate": 8.586143235321545e-06,
+ "loss": 0.0458,
+ "step": 20508
+ },
+ {
+ "epoch": 55.88283378746594,
+ "grad_norm": 3.9310410022735596,
+ "learning_rate": 8.585269611212862e-06,
+ "loss": 0.0692,
+ "step": 20509
+ },
+ {
+ "epoch": 55.88555858310627,
+ "grad_norm": 2.7172627449035645,
+ "learning_rate": 8.584395998122054e-06,
+ "loss": 0.1011,
+ "step": 20510
+ },
+ {
+ "epoch": 55.88828337874659,
+ "grad_norm": 3.1608898639678955,
+ "learning_rate": 8.583522396055923e-06,
+ "loss": 0.093,
+ "step": 20511
+ },
+ {
+ "epoch": 55.89100817438692,
+ "grad_norm": 2.9542112350463867,
+ "learning_rate": 8.582648805021276e-06,
+ "loss": 0.0938,
+ "step": 20512
+ },
+ {
+ "epoch": 55.89373297002725,
+ "grad_norm": 2.9722490310668945,
+ "learning_rate": 8.581775225024914e-06,
+ "loss": 0.0912,
+ "step": 20513
+ },
+ {
+ "epoch": 55.89645776566758,
+ "grad_norm": 3.349339485168457,
+ "learning_rate": 8.58090165607364e-06,
+ "loss": 0.0535,
+ "step": 20514
+ },
+ {
+ "epoch": 55.8991825613079,
+ "grad_norm": 3.676875352859497,
+ "learning_rate": 8.58002809817426e-06,
+ "loss": 0.1212,
+ "step": 20515
+ },
+ {
+ "epoch": 55.90190735694823,
+ "grad_norm": 2.664891481399536,
+ "learning_rate": 8.579154551333574e-06,
+ "loss": 0.0668,
+ "step": 20516
+ },
+ {
+ "epoch": 55.904632152588555,
+ "grad_norm": 3.0295116901397705,
+ "learning_rate": 8.578281015558388e-06,
+ "loss": 0.0554,
+ "step": 20517
+ },
+ {
+ "epoch": 55.90735694822888,
+ "grad_norm": 3.1258485317230225,
+ "learning_rate": 8.577407490855499e-06,
+ "loss": 0.1315,
+ "step": 20518
+ },
+ {
+ "epoch": 55.91008174386921,
+ "grad_norm": 3.0943851470947266,
+ "learning_rate": 8.57653397723172e-06,
+ "loss": 0.0697,
+ "step": 20519
+ },
+ {
+ "epoch": 55.91280653950954,
+ "grad_norm": 4.538146495819092,
+ "learning_rate": 8.575660474693846e-06,
+ "loss": 0.0491,
+ "step": 20520
+ },
+ {
+ "epoch": 55.915531335149865,
+ "grad_norm": 2.7519185543060303,
+ "learning_rate": 8.574786983248684e-06,
+ "loss": 0.0641,
+ "step": 20521
+ },
+ {
+ "epoch": 55.91825613079019,
+ "grad_norm": 3.406585931777954,
+ "learning_rate": 8.573913502903034e-06,
+ "loss": 0.1468,
+ "step": 20522
+ },
+ {
+ "epoch": 55.920980926430516,
+ "grad_norm": 2.450366973876953,
+ "learning_rate": 8.573040033663702e-06,
+ "loss": 0.043,
+ "step": 20523
+ },
+ {
+ "epoch": 55.92370572207084,
+ "grad_norm": 2.882218360900879,
+ "learning_rate": 8.572166575537484e-06,
+ "loss": 0.0441,
+ "step": 20524
+ },
+ {
+ "epoch": 55.926430517711175,
+ "grad_norm": 4.7175140380859375,
+ "learning_rate": 8.571293128531191e-06,
+ "loss": 0.0959,
+ "step": 20525
+ },
+ {
+ "epoch": 55.9291553133515,
+ "grad_norm": 3.0406744480133057,
+ "learning_rate": 8.57041969265162e-06,
+ "loss": 0.0621,
+ "step": 20526
+ },
+ {
+ "epoch": 55.93188010899183,
+ "grad_norm": 4.914891242980957,
+ "learning_rate": 8.569546267905571e-06,
+ "loss": 0.0786,
+ "step": 20527
+ },
+ {
+ "epoch": 55.93460490463215,
+ "grad_norm": 3.1082944869995117,
+ "learning_rate": 8.568672854299856e-06,
+ "loss": 0.1723,
+ "step": 20528
+ },
+ {
+ "epoch": 55.93732970027248,
+ "grad_norm": 3.2055375576019287,
+ "learning_rate": 8.567799451841266e-06,
+ "loss": 0.0456,
+ "step": 20529
+ },
+ {
+ "epoch": 55.940054495912804,
+ "grad_norm": 2.747576951980591,
+ "learning_rate": 8.566926060536608e-06,
+ "loss": 0.0472,
+ "step": 20530
+ },
+ {
+ "epoch": 55.94277929155314,
+ "grad_norm": 3.4815165996551514,
+ "learning_rate": 8.566052680392684e-06,
+ "loss": 0.1722,
+ "step": 20531
+ },
+ {
+ "epoch": 55.94550408719346,
+ "grad_norm": 1.9644190073013306,
+ "learning_rate": 8.565179311416298e-06,
+ "loss": 0.0597,
+ "step": 20532
+ },
+ {
+ "epoch": 55.94822888283379,
+ "grad_norm": 2.649864673614502,
+ "learning_rate": 8.564305953614246e-06,
+ "loss": 0.0363,
+ "step": 20533
+ },
+ {
+ "epoch": 55.950953678474114,
+ "grad_norm": 3.513690710067749,
+ "learning_rate": 8.563432606993337e-06,
+ "loss": 0.1765,
+ "step": 20534
+ },
+ {
+ "epoch": 55.95367847411444,
+ "grad_norm": 2.800788402557373,
+ "learning_rate": 8.562559271560365e-06,
+ "loss": 0.0567,
+ "step": 20535
+ },
+ {
+ "epoch": 55.956403269754766,
+ "grad_norm": 3.6321089267730713,
+ "learning_rate": 8.561685947322133e-06,
+ "loss": 0.0829,
+ "step": 20536
+ },
+ {
+ "epoch": 55.95912806539509,
+ "grad_norm": 4.189486503601074,
+ "learning_rate": 8.560812634285451e-06,
+ "loss": 0.0482,
+ "step": 20537
+ },
+ {
+ "epoch": 55.961852861035425,
+ "grad_norm": 2.4395861625671387,
+ "learning_rate": 8.559939332457109e-06,
+ "loss": 0.0436,
+ "step": 20538
+ },
+ {
+ "epoch": 55.96457765667575,
+ "grad_norm": 3.1815173625946045,
+ "learning_rate": 8.559066041843917e-06,
+ "loss": 0.1148,
+ "step": 20539
+ },
+ {
+ "epoch": 55.967302452316076,
+ "grad_norm": 2.6192049980163574,
+ "learning_rate": 8.558192762452669e-06,
+ "loss": 0.0423,
+ "step": 20540
+ },
+ {
+ "epoch": 55.9700272479564,
+ "grad_norm": 5.64263391494751,
+ "learning_rate": 8.557319494290171e-06,
+ "loss": 0.08,
+ "step": 20541
+ },
+ {
+ "epoch": 55.97275204359673,
+ "grad_norm": 2.7219245433807373,
+ "learning_rate": 8.55644623736322e-06,
+ "loss": 0.06,
+ "step": 20542
+ },
+ {
+ "epoch": 55.97547683923706,
+ "grad_norm": 3.6037518978118896,
+ "learning_rate": 8.555572991678624e-06,
+ "loss": 0.0648,
+ "step": 20543
+ },
+ {
+ "epoch": 55.97820163487739,
+ "grad_norm": 3.719003438949585,
+ "learning_rate": 8.554699757243174e-06,
+ "loss": 0.0921,
+ "step": 20544
+ },
+ {
+ "epoch": 55.98092643051771,
+ "grad_norm": 2.8958046436309814,
+ "learning_rate": 8.553826534063681e-06,
+ "loss": 0.0914,
+ "step": 20545
+ },
+ {
+ "epoch": 55.98365122615804,
+ "grad_norm": 2.881096601486206,
+ "learning_rate": 8.552953322146936e-06,
+ "loss": 0.0719,
+ "step": 20546
+ },
+ {
+ "epoch": 55.986376021798364,
+ "grad_norm": 2.6342363357543945,
+ "learning_rate": 8.552080121499747e-06,
+ "loss": 0.0548,
+ "step": 20547
+ },
+ {
+ "epoch": 55.98910081743869,
+ "grad_norm": 3.4788002967834473,
+ "learning_rate": 8.551206932128908e-06,
+ "loss": 0.0808,
+ "step": 20548
+ },
+ {
+ "epoch": 55.991825613079016,
+ "grad_norm": 3.0843818187713623,
+ "learning_rate": 8.550333754041227e-06,
+ "loss": 0.0968,
+ "step": 20549
+ },
+ {
+ "epoch": 55.99455040871935,
+ "grad_norm": 2.777289390563965,
+ "learning_rate": 8.549460587243496e-06,
+ "loss": 0.0455,
+ "step": 20550
+ },
+ {
+ "epoch": 55.997275204359674,
+ "grad_norm": 4.32223653793335,
+ "learning_rate": 8.54858743174252e-06,
+ "loss": 0.0695,
+ "step": 20551
+ },
+ {
+ "epoch": 56.0,
+ "grad_norm": 2.705508232116699,
+ "learning_rate": 8.5477142875451e-06,
+ "loss": 0.0248,
+ "step": 20552
+ },
+ {
+ "epoch": 56.002724795640326,
+ "grad_norm": 2.444056749343872,
+ "learning_rate": 8.546841154658031e-06,
+ "loss": 0.1231,
+ "step": 20553
+ },
+ {
+ "epoch": 56.00544959128065,
+ "grad_norm": 2.582623243331909,
+ "learning_rate": 8.545968033088117e-06,
+ "loss": 0.0344,
+ "step": 20554
+ },
+ {
+ "epoch": 56.00817438692098,
+ "grad_norm": 3.6686673164367676,
+ "learning_rate": 8.545094922842161e-06,
+ "loss": 0.1088,
+ "step": 20555
+ },
+ {
+ "epoch": 56.01089918256131,
+ "grad_norm": 1.8390010595321655,
+ "learning_rate": 8.544221823926957e-06,
+ "loss": 0.0317,
+ "step": 20556
+ },
+ {
+ "epoch": 56.013623978201636,
+ "grad_norm": 3.1465914249420166,
+ "learning_rate": 8.543348736349307e-06,
+ "loss": 0.0368,
+ "step": 20557
+ },
+ {
+ "epoch": 56.01634877384196,
+ "grad_norm": 2.5141797065734863,
+ "learning_rate": 8.54247566011601e-06,
+ "loss": 0.0526,
+ "step": 20558
+ },
+ {
+ "epoch": 56.01907356948229,
+ "grad_norm": 3.39721941947937,
+ "learning_rate": 8.541602595233866e-06,
+ "loss": 0.0517,
+ "step": 20559
+ },
+ {
+ "epoch": 56.02179836512261,
+ "grad_norm": 3.6756551265716553,
+ "learning_rate": 8.540729541709673e-06,
+ "loss": 0.0521,
+ "step": 20560
+ },
+ {
+ "epoch": 56.02452316076294,
+ "grad_norm": 3.0021440982818604,
+ "learning_rate": 8.539856499550233e-06,
+ "loss": 0.0397,
+ "step": 20561
+ },
+ {
+ "epoch": 56.02724795640327,
+ "grad_norm": 2.3593082427978516,
+ "learning_rate": 8.53898346876234e-06,
+ "loss": 0.1254,
+ "step": 20562
+ },
+ {
+ "epoch": 56.0299727520436,
+ "grad_norm": 2.4953644275665283,
+ "learning_rate": 8.5381104493528e-06,
+ "loss": 0.1172,
+ "step": 20563
+ },
+ {
+ "epoch": 56.032697547683924,
+ "grad_norm": 2.611631393432617,
+ "learning_rate": 8.537237441328409e-06,
+ "loss": 0.0365,
+ "step": 20564
+ },
+ {
+ "epoch": 56.03542234332425,
+ "grad_norm": 2.3411648273468018,
+ "learning_rate": 8.536364444695967e-06,
+ "loss": 0.0587,
+ "step": 20565
+ },
+ {
+ "epoch": 56.038147138964575,
+ "grad_norm": 3.1423861980438232,
+ "learning_rate": 8.535491459462267e-06,
+ "loss": 0.11,
+ "step": 20566
+ },
+ {
+ "epoch": 56.0408719346049,
+ "grad_norm": 2.944500684738159,
+ "learning_rate": 8.534618485634115e-06,
+ "loss": 0.0557,
+ "step": 20567
+ },
+ {
+ "epoch": 56.043596730245234,
+ "grad_norm": 1.9541003704071045,
+ "learning_rate": 8.533745523218307e-06,
+ "loss": 0.0371,
+ "step": 20568
+ },
+ {
+ "epoch": 56.04632152588556,
+ "grad_norm": 2.167980909347534,
+ "learning_rate": 8.532872572221642e-06,
+ "loss": 0.0701,
+ "step": 20569
+ },
+ {
+ "epoch": 56.049046321525886,
+ "grad_norm": 2.74839186668396,
+ "learning_rate": 8.531999632650912e-06,
+ "loss": 0.2549,
+ "step": 20570
+ },
+ {
+ "epoch": 56.05177111716621,
+ "grad_norm": 2.0840060710906982,
+ "learning_rate": 8.531126704512928e-06,
+ "loss": 0.1265,
+ "step": 20571
+ },
+ {
+ "epoch": 56.05449591280654,
+ "grad_norm": 2.955068349838257,
+ "learning_rate": 8.53025378781448e-06,
+ "loss": 0.0977,
+ "step": 20572
+ },
+ {
+ "epoch": 56.05722070844686,
+ "grad_norm": 3.3737564086914062,
+ "learning_rate": 8.529380882562371e-06,
+ "loss": 0.0771,
+ "step": 20573
+ },
+ {
+ "epoch": 56.059945504087196,
+ "grad_norm": 2.5731282234191895,
+ "learning_rate": 8.528507988763393e-06,
+ "loss": 0.0523,
+ "step": 20574
+ },
+ {
+ "epoch": 56.06267029972752,
+ "grad_norm": 2.1656298637390137,
+ "learning_rate": 8.527635106424352e-06,
+ "loss": 0.0613,
+ "step": 20575
+ },
+ {
+ "epoch": 56.06539509536785,
+ "grad_norm": 3.1821811199188232,
+ "learning_rate": 8.526762235552038e-06,
+ "loss": 0.0871,
+ "step": 20576
+ },
+ {
+ "epoch": 56.06811989100817,
+ "grad_norm": 5.005757808685303,
+ "learning_rate": 8.525889376153253e-06,
+ "loss": 0.0497,
+ "step": 20577
+ },
+ {
+ "epoch": 56.0708446866485,
+ "grad_norm": 2.5638303756713867,
+ "learning_rate": 8.525016528234794e-06,
+ "loss": 0.0511,
+ "step": 20578
+ },
+ {
+ "epoch": 56.073569482288825,
+ "grad_norm": 2.2201926708221436,
+ "learning_rate": 8.52414369180346e-06,
+ "loss": 0.0413,
+ "step": 20579
+ },
+ {
+ "epoch": 56.07629427792916,
+ "grad_norm": 1.9597461223602295,
+ "learning_rate": 8.523270866866044e-06,
+ "loss": 0.0345,
+ "step": 20580
+ },
+ {
+ "epoch": 56.079019073569484,
+ "grad_norm": 2.8702890872955322,
+ "learning_rate": 8.52239805342935e-06,
+ "loss": 0.155,
+ "step": 20581
+ },
+ {
+ "epoch": 56.08174386920981,
+ "grad_norm": 3.183558940887451,
+ "learning_rate": 8.521525251500172e-06,
+ "loss": 0.0457,
+ "step": 20582
+ },
+ {
+ "epoch": 56.084468664850135,
+ "grad_norm": 2.012411594390869,
+ "learning_rate": 8.520652461085309e-06,
+ "loss": 0.0409,
+ "step": 20583
+ },
+ {
+ "epoch": 56.08719346049046,
+ "grad_norm": 2.439301013946533,
+ "learning_rate": 8.519779682191554e-06,
+ "loss": 0.0529,
+ "step": 20584
+ },
+ {
+ "epoch": 56.08991825613079,
+ "grad_norm": 2.6224725246429443,
+ "learning_rate": 8.518906914825709e-06,
+ "loss": 0.0536,
+ "step": 20585
+ },
+ {
+ "epoch": 56.09264305177112,
+ "grad_norm": 2.129708766937256,
+ "learning_rate": 8.518034158994567e-06,
+ "loss": 0.0325,
+ "step": 20586
+ },
+ {
+ "epoch": 56.095367847411445,
+ "grad_norm": 2.9877612590789795,
+ "learning_rate": 8.517161414704926e-06,
+ "loss": 0.0348,
+ "step": 20587
+ },
+ {
+ "epoch": 56.09809264305177,
+ "grad_norm": 2.912266731262207,
+ "learning_rate": 8.516288681963589e-06,
+ "loss": 0.131,
+ "step": 20588
+ },
+ {
+ "epoch": 56.1008174386921,
+ "grad_norm": 3.3581833839416504,
+ "learning_rate": 8.515415960777345e-06,
+ "loss": 0.0509,
+ "step": 20589
+ },
+ {
+ "epoch": 56.10354223433242,
+ "grad_norm": 2.0328946113586426,
+ "learning_rate": 8.514543251152996e-06,
+ "loss": 0.1917,
+ "step": 20590
+ },
+ {
+ "epoch": 56.10626702997275,
+ "grad_norm": 1.7305907011032104,
+ "learning_rate": 8.513670553097335e-06,
+ "loss": 0.1304,
+ "step": 20591
+ },
+ {
+ "epoch": 56.10899182561308,
+ "grad_norm": 2.8216845989227295,
+ "learning_rate": 8.512797866617158e-06,
+ "loss": 0.0682,
+ "step": 20592
+ },
+ {
+ "epoch": 56.11171662125341,
+ "grad_norm": 1.9632965326309204,
+ "learning_rate": 8.511925191719268e-06,
+ "loss": 0.1193,
+ "step": 20593
+ },
+ {
+ "epoch": 56.11444141689373,
+ "grad_norm": 2.593263626098633,
+ "learning_rate": 8.511052528410453e-06,
+ "loss": 0.0881,
+ "step": 20594
+ },
+ {
+ "epoch": 56.11716621253406,
+ "grad_norm": 2.770054817199707,
+ "learning_rate": 8.510179876697514e-06,
+ "loss": 0.1841,
+ "step": 20595
+ },
+ {
+ "epoch": 56.119891008174385,
+ "grad_norm": 3.0248944759368896,
+ "learning_rate": 8.509307236587247e-06,
+ "loss": 0.0957,
+ "step": 20596
+ },
+ {
+ "epoch": 56.12261580381471,
+ "grad_norm": 3.145876407623291,
+ "learning_rate": 8.508434608086448e-06,
+ "loss": 0.0777,
+ "step": 20597
+ },
+ {
+ "epoch": 56.12534059945504,
+ "grad_norm": 1.9750829935073853,
+ "learning_rate": 8.507561991201908e-06,
+ "loss": 0.0334,
+ "step": 20598
+ },
+ {
+ "epoch": 56.12806539509537,
+ "grad_norm": 1.664153814315796,
+ "learning_rate": 8.506689385940433e-06,
+ "loss": 0.0283,
+ "step": 20599
+ },
+ {
+ "epoch": 56.130790190735695,
+ "grad_norm": 1.8593064546585083,
+ "learning_rate": 8.50581679230881e-06,
+ "loss": 0.032,
+ "step": 20600
+ },
+ {
+ "epoch": 56.13351498637602,
+ "grad_norm": 2.934802770614624,
+ "learning_rate": 8.504944210313838e-06,
+ "loss": 0.0561,
+ "step": 20601
+ },
+ {
+ "epoch": 56.13623978201635,
+ "grad_norm": 1.8253824710845947,
+ "learning_rate": 8.50407163996231e-06,
+ "loss": 0.0307,
+ "step": 20602
+ },
+ {
+ "epoch": 56.13896457765667,
+ "grad_norm": 3.0419840812683105,
+ "learning_rate": 8.503199081261028e-06,
+ "loss": 0.0473,
+ "step": 20603
+ },
+ {
+ "epoch": 56.141689373297005,
+ "grad_norm": 2.0137298107147217,
+ "learning_rate": 8.50232653421678e-06,
+ "loss": 0.0311,
+ "step": 20604
+ },
+ {
+ "epoch": 56.14441416893733,
+ "grad_norm": 2.3562755584716797,
+ "learning_rate": 8.501453998836364e-06,
+ "loss": 0.041,
+ "step": 20605
+ },
+ {
+ "epoch": 56.14713896457766,
+ "grad_norm": 2.3670244216918945,
+ "learning_rate": 8.50058147512658e-06,
+ "loss": 0.0562,
+ "step": 20606
+ },
+ {
+ "epoch": 56.14986376021798,
+ "grad_norm": 3.0103213787078857,
+ "learning_rate": 8.499708963094216e-06,
+ "loss": 0.2595,
+ "step": 20607
+ },
+ {
+ "epoch": 56.15258855585831,
+ "grad_norm": 2.6325552463531494,
+ "learning_rate": 8.498836462746074e-06,
+ "loss": 0.0345,
+ "step": 20608
+ },
+ {
+ "epoch": 56.155313351498634,
+ "grad_norm": 2.847313404083252,
+ "learning_rate": 8.497963974088942e-06,
+ "loss": 0.09,
+ "step": 20609
+ },
+ {
+ "epoch": 56.15803814713897,
+ "grad_norm": 2.8092024326324463,
+ "learning_rate": 8.497091497129621e-06,
+ "loss": 0.0565,
+ "step": 20610
+ },
+ {
+ "epoch": 56.16076294277929,
+ "grad_norm": 2.2805490493774414,
+ "learning_rate": 8.496219031874899e-06,
+ "loss": 0.0721,
+ "step": 20611
+ },
+ {
+ "epoch": 56.16348773841962,
+ "grad_norm": 2.5317776203155518,
+ "learning_rate": 8.49534657833158e-06,
+ "loss": 0.0665,
+ "step": 20612
+ },
+ {
+ "epoch": 56.166212534059945,
+ "grad_norm": 2.707932949066162,
+ "learning_rate": 8.49447413650645e-06,
+ "loss": 0.0573,
+ "step": 20613
+ },
+ {
+ "epoch": 56.16893732970027,
+ "grad_norm": 3.0782687664031982,
+ "learning_rate": 8.493601706406305e-06,
+ "loss": 0.0918,
+ "step": 20614
+ },
+ {
+ "epoch": 56.171662125340596,
+ "grad_norm": 2.7679502964019775,
+ "learning_rate": 8.492729288037945e-06,
+ "loss": 0.0353,
+ "step": 20615
+ },
+ {
+ "epoch": 56.17438692098093,
+ "grad_norm": 1.8996922969818115,
+ "learning_rate": 8.491856881408159e-06,
+ "loss": 0.0355,
+ "step": 20616
+ },
+ {
+ "epoch": 56.177111716621255,
+ "grad_norm": 4.048816204071045,
+ "learning_rate": 8.490984486523746e-06,
+ "loss": 0.0518,
+ "step": 20617
+ },
+ {
+ "epoch": 56.17983651226158,
+ "grad_norm": 2.383310079574585,
+ "learning_rate": 8.490112103391494e-06,
+ "loss": 0.0356,
+ "step": 20618
+ },
+ {
+ "epoch": 56.182561307901906,
+ "grad_norm": 2.0713369846343994,
+ "learning_rate": 8.489239732018202e-06,
+ "loss": 0.0421,
+ "step": 20619
+ },
+ {
+ "epoch": 56.18528610354223,
+ "grad_norm": 3.171442985534668,
+ "learning_rate": 8.48836737241066e-06,
+ "loss": 0.0738,
+ "step": 20620
+ },
+ {
+ "epoch": 56.18801089918256,
+ "grad_norm": 1.9766919612884521,
+ "learning_rate": 8.487495024575668e-06,
+ "loss": 0.029,
+ "step": 20621
+ },
+ {
+ "epoch": 56.19073569482289,
+ "grad_norm": 2.457634925842285,
+ "learning_rate": 8.48662268852001e-06,
+ "loss": 0.1003,
+ "step": 20622
+ },
+ {
+ "epoch": 56.19346049046322,
+ "grad_norm": 2.8911070823669434,
+ "learning_rate": 8.48575036425049e-06,
+ "loss": 0.1638,
+ "step": 20623
+ },
+ {
+ "epoch": 56.19618528610354,
+ "grad_norm": 2.8728299140930176,
+ "learning_rate": 8.4848780517739e-06,
+ "loss": 0.0478,
+ "step": 20624
+ },
+ {
+ "epoch": 56.19891008174387,
+ "grad_norm": 1.995609164237976,
+ "learning_rate": 8.484005751097026e-06,
+ "loss": 0.1003,
+ "step": 20625
+ },
+ {
+ "epoch": 56.201634877384194,
+ "grad_norm": 2.500788927078247,
+ "learning_rate": 8.483133462226671e-06,
+ "loss": 0.0897,
+ "step": 20626
+ },
+ {
+ "epoch": 56.20435967302452,
+ "grad_norm": 3.7274577617645264,
+ "learning_rate": 8.48226118516962e-06,
+ "loss": 0.0597,
+ "step": 20627
+ },
+ {
+ "epoch": 56.20708446866485,
+ "grad_norm": 2.0847866535186768,
+ "learning_rate": 8.481388919932672e-06,
+ "loss": 0.0336,
+ "step": 20628
+ },
+ {
+ "epoch": 56.20980926430518,
+ "grad_norm": 2.4570963382720947,
+ "learning_rate": 8.480516666522616e-06,
+ "loss": 0.0988,
+ "step": 20629
+ },
+ {
+ "epoch": 56.212534059945504,
+ "grad_norm": 4.067639350891113,
+ "learning_rate": 8.479644424946251e-06,
+ "loss": 0.0325,
+ "step": 20630
+ },
+ {
+ "epoch": 56.21525885558583,
+ "grad_norm": 2.854677677154541,
+ "learning_rate": 8.478772195210365e-06,
+ "loss": 0.0556,
+ "step": 20631
+ },
+ {
+ "epoch": 56.217983651226156,
+ "grad_norm": 1.8370753526687622,
+ "learning_rate": 8.477899977321752e-06,
+ "loss": 0.0425,
+ "step": 20632
+ },
+ {
+ "epoch": 56.22070844686648,
+ "grad_norm": 2.923435926437378,
+ "learning_rate": 8.477027771287203e-06,
+ "loss": 0.1015,
+ "step": 20633
+ },
+ {
+ "epoch": 56.223433242506815,
+ "grad_norm": 2.7340807914733887,
+ "learning_rate": 8.476155577113515e-06,
+ "loss": 0.077,
+ "step": 20634
+ },
+ {
+ "epoch": 56.22615803814714,
+ "grad_norm": 2.5731863975524902,
+ "learning_rate": 8.475283394807478e-06,
+ "loss": 0.0591,
+ "step": 20635
+ },
+ {
+ "epoch": 56.228882833787466,
+ "grad_norm": 2.880963087081909,
+ "learning_rate": 8.474411224375885e-06,
+ "loss": 0.1502,
+ "step": 20636
+ },
+ {
+ "epoch": 56.23160762942779,
+ "grad_norm": 2.0543406009674072,
+ "learning_rate": 8.473539065825527e-06,
+ "loss": 0.0711,
+ "step": 20637
+ },
+ {
+ "epoch": 56.23433242506812,
+ "grad_norm": 1.7565157413482666,
+ "learning_rate": 8.472666919163198e-06,
+ "loss": 0.0378,
+ "step": 20638
+ },
+ {
+ "epoch": 56.237057220708444,
+ "grad_norm": 2.9041974544525146,
+ "learning_rate": 8.471794784395687e-06,
+ "loss": 0.0382,
+ "step": 20639
+ },
+ {
+ "epoch": 56.23978201634878,
+ "grad_norm": 2.8214023113250732,
+ "learning_rate": 8.470922661529792e-06,
+ "loss": 0.0229,
+ "step": 20640
+ },
+ {
+ "epoch": 56.2425068119891,
+ "grad_norm": 4.390793800354004,
+ "learning_rate": 8.470050550572302e-06,
+ "loss": 0.059,
+ "step": 20641
+ },
+ {
+ "epoch": 56.24523160762943,
+ "grad_norm": 2.334028959274292,
+ "learning_rate": 8.46917845153001e-06,
+ "loss": 0.0536,
+ "step": 20642
+ },
+ {
+ "epoch": 56.247956403269754,
+ "grad_norm": 2.511768341064453,
+ "learning_rate": 8.468306364409704e-06,
+ "loss": 0.0348,
+ "step": 20643
+ },
+ {
+ "epoch": 56.25068119891008,
+ "grad_norm": 2.7089121341705322,
+ "learning_rate": 8.467434289218184e-06,
+ "loss": 0.0375,
+ "step": 20644
+ },
+ {
+ "epoch": 56.253405994550405,
+ "grad_norm": 1.9299030303955078,
+ "learning_rate": 8.466562225962233e-06,
+ "loss": 0.0769,
+ "step": 20645
+ },
+ {
+ "epoch": 56.25613079019074,
+ "grad_norm": 2.6587960720062256,
+ "learning_rate": 8.465690174648648e-06,
+ "loss": 0.0654,
+ "step": 20646
+ },
+ {
+ "epoch": 56.258855585831064,
+ "grad_norm": 8.406907081604004,
+ "learning_rate": 8.464818135284214e-06,
+ "loss": 0.0647,
+ "step": 20647
+ },
+ {
+ "epoch": 56.26158038147139,
+ "grad_norm": 3.0533714294433594,
+ "learning_rate": 8.463946107875732e-06,
+ "loss": 0.2211,
+ "step": 20648
+ },
+ {
+ "epoch": 56.264305177111716,
+ "grad_norm": 1.6724638938903809,
+ "learning_rate": 8.463074092429986e-06,
+ "loss": 0.0283,
+ "step": 20649
+ },
+ {
+ "epoch": 56.26702997275204,
+ "grad_norm": 2.9044241905212402,
+ "learning_rate": 8.462202088953771e-06,
+ "loss": 0.0455,
+ "step": 20650
+ },
+ {
+ "epoch": 56.26975476839237,
+ "grad_norm": 2.7840428352355957,
+ "learning_rate": 8.461330097453875e-06,
+ "loss": 0.0865,
+ "step": 20651
+ },
+ {
+ "epoch": 56.2724795640327,
+ "grad_norm": 3.0028581619262695,
+ "learning_rate": 8.460458117937091e-06,
+ "loss": 0.0519,
+ "step": 20652
+ },
+ {
+ "epoch": 56.275204359673026,
+ "grad_norm": 2.021851062774658,
+ "learning_rate": 8.45958615041021e-06,
+ "loss": 0.0414,
+ "step": 20653
+ },
+ {
+ "epoch": 56.27792915531335,
+ "grad_norm": 2.4991307258605957,
+ "learning_rate": 8.458714194880022e-06,
+ "loss": 0.0251,
+ "step": 20654
+ },
+ {
+ "epoch": 56.28065395095368,
+ "grad_norm": 1.98735773563385,
+ "learning_rate": 8.45784225135332e-06,
+ "loss": 0.0272,
+ "step": 20655
+ },
+ {
+ "epoch": 56.283378746594,
+ "grad_norm": 1.6085554361343384,
+ "learning_rate": 8.456970319836886e-06,
+ "loss": 0.0274,
+ "step": 20656
+ },
+ {
+ "epoch": 56.28610354223433,
+ "grad_norm": 2.1118195056915283,
+ "learning_rate": 8.456098400337522e-06,
+ "loss": 0.1424,
+ "step": 20657
+ },
+ {
+ "epoch": 56.28882833787466,
+ "grad_norm": 2.2711551189422607,
+ "learning_rate": 8.455226492862017e-06,
+ "loss": 0.0279,
+ "step": 20658
+ },
+ {
+ "epoch": 56.29155313351499,
+ "grad_norm": 2.240755796432495,
+ "learning_rate": 8.454354597417155e-06,
+ "loss": 0.1145,
+ "step": 20659
+ },
+ {
+ "epoch": 56.294277929155314,
+ "grad_norm": 3.496342182159424,
+ "learning_rate": 8.453482714009732e-06,
+ "loss": 0.0815,
+ "step": 20660
+ },
+ {
+ "epoch": 56.29700272479564,
+ "grad_norm": 3.7883739471435547,
+ "learning_rate": 8.452610842646533e-06,
+ "loss": 0.0582,
+ "step": 20661
+ },
+ {
+ "epoch": 56.299727520435965,
+ "grad_norm": 1.3206783533096313,
+ "learning_rate": 8.451738983334355e-06,
+ "loss": 0.0216,
+ "step": 20662
+ },
+ {
+ "epoch": 56.30245231607629,
+ "grad_norm": 2.747035503387451,
+ "learning_rate": 8.45086713607998e-06,
+ "loss": 0.0703,
+ "step": 20663
+ },
+ {
+ "epoch": 56.305177111716624,
+ "grad_norm": 2.587695598602295,
+ "learning_rate": 8.449995300890204e-06,
+ "loss": 0.0423,
+ "step": 20664
+ },
+ {
+ "epoch": 56.30790190735695,
+ "grad_norm": 2.573094129562378,
+ "learning_rate": 8.449123477771815e-06,
+ "loss": 0.1329,
+ "step": 20665
+ },
+ {
+ "epoch": 56.310626702997276,
+ "grad_norm": 2.045747756958008,
+ "learning_rate": 8.448251666731601e-06,
+ "loss": 0.033,
+ "step": 20666
+ },
+ {
+ "epoch": 56.3133514986376,
+ "grad_norm": 2.5283401012420654,
+ "learning_rate": 8.447379867776353e-06,
+ "loss": 0.0318,
+ "step": 20667
+ },
+ {
+ "epoch": 56.31607629427793,
+ "grad_norm": 2.691645860671997,
+ "learning_rate": 8.446508080912863e-06,
+ "loss": 0.0935,
+ "step": 20668
+ },
+ {
+ "epoch": 56.31880108991825,
+ "grad_norm": 1.28969144821167,
+ "learning_rate": 8.445636306147915e-06,
+ "loss": 0.0199,
+ "step": 20669
+ },
+ {
+ "epoch": 56.321525885558586,
+ "grad_norm": 1.9561066627502441,
+ "learning_rate": 8.444764543488303e-06,
+ "loss": 0.0252,
+ "step": 20670
+ },
+ {
+ "epoch": 56.32425068119891,
+ "grad_norm": 2.6769919395446777,
+ "learning_rate": 8.443892792940815e-06,
+ "loss": 0.0441,
+ "step": 20671
+ },
+ {
+ "epoch": 56.32697547683924,
+ "grad_norm": 2.5082263946533203,
+ "learning_rate": 8.44302105451224e-06,
+ "loss": 0.068,
+ "step": 20672
+ },
+ {
+ "epoch": 56.32970027247956,
+ "grad_norm": 3.352229595184326,
+ "learning_rate": 8.442149328209364e-06,
+ "loss": 0.0883,
+ "step": 20673
+ },
+ {
+ "epoch": 56.33242506811989,
+ "grad_norm": 2.2220711708068848,
+ "learning_rate": 8.44127761403898e-06,
+ "loss": 0.0359,
+ "step": 20674
+ },
+ {
+ "epoch": 56.335149863760215,
+ "grad_norm": 2.260627269744873,
+ "learning_rate": 8.440405912007879e-06,
+ "loss": 0.0253,
+ "step": 20675
+ },
+ {
+ "epoch": 56.33787465940055,
+ "grad_norm": 3.4348692893981934,
+ "learning_rate": 8.439534222122843e-06,
+ "loss": 0.2498,
+ "step": 20676
+ },
+ {
+ "epoch": 56.34059945504087,
+ "grad_norm": 2.510993003845215,
+ "learning_rate": 8.438662544390666e-06,
+ "loss": 0.027,
+ "step": 20677
+ },
+ {
+ "epoch": 56.3433242506812,
+ "grad_norm": 2.1311519145965576,
+ "learning_rate": 8.437790878818135e-06,
+ "loss": 0.0855,
+ "step": 20678
+ },
+ {
+ "epoch": 56.346049046321525,
+ "grad_norm": 2.283508062362671,
+ "learning_rate": 8.436919225412038e-06,
+ "loss": 0.0719,
+ "step": 20679
+ },
+ {
+ "epoch": 56.34877384196185,
+ "grad_norm": 2.0969996452331543,
+ "learning_rate": 8.436047584179163e-06,
+ "loss": 0.0366,
+ "step": 20680
+ },
+ {
+ "epoch": 56.35149863760218,
+ "grad_norm": 1.7256392240524292,
+ "learning_rate": 8.4351759551263e-06,
+ "loss": 0.0571,
+ "step": 20681
+ },
+ {
+ "epoch": 56.35422343324251,
+ "grad_norm": 3.415335178375244,
+ "learning_rate": 8.434304338260237e-06,
+ "loss": 0.1321,
+ "step": 20682
+ },
+ {
+ "epoch": 56.356948228882835,
+ "grad_norm": 1.5256974697113037,
+ "learning_rate": 8.433432733587758e-06,
+ "loss": 0.0274,
+ "step": 20683
+ },
+ {
+ "epoch": 56.35967302452316,
+ "grad_norm": 2.084994316101074,
+ "learning_rate": 8.43256114111566e-06,
+ "loss": 0.1191,
+ "step": 20684
+ },
+ {
+ "epoch": 56.36239782016349,
+ "grad_norm": 2.285475492477417,
+ "learning_rate": 8.43168956085072e-06,
+ "loss": 0.0539,
+ "step": 20685
+ },
+ {
+ "epoch": 56.36512261580381,
+ "grad_norm": 2.0213992595672607,
+ "learning_rate": 8.430817992799735e-06,
+ "loss": 0.0467,
+ "step": 20686
+ },
+ {
+ "epoch": 56.36784741144414,
+ "grad_norm": 3.1540327072143555,
+ "learning_rate": 8.429946436969488e-06,
+ "loss": 0.0599,
+ "step": 20687
+ },
+ {
+ "epoch": 56.37057220708447,
+ "grad_norm": 1.7139548063278198,
+ "learning_rate": 8.429074893366768e-06,
+ "loss": 0.0265,
+ "step": 20688
+ },
+ {
+ "epoch": 56.3732970027248,
+ "grad_norm": 2.546781301498413,
+ "learning_rate": 8.42820336199836e-06,
+ "loss": 0.0316,
+ "step": 20689
+ },
+ {
+ "epoch": 56.37602179836512,
+ "grad_norm": 2.4416143894195557,
+ "learning_rate": 8.427331842871052e-06,
+ "loss": 0.0451,
+ "step": 20690
+ },
+ {
+ "epoch": 56.37874659400545,
+ "grad_norm": 3.303438425064087,
+ "learning_rate": 8.426460335991638e-06,
+ "loss": 0.0354,
+ "step": 20691
+ },
+ {
+ "epoch": 56.381471389645775,
+ "grad_norm": 2.1460156440734863,
+ "learning_rate": 8.425588841366896e-06,
+ "loss": 0.0308,
+ "step": 20692
+ },
+ {
+ "epoch": 56.3841961852861,
+ "grad_norm": 2.587888479232788,
+ "learning_rate": 8.424717359003621e-06,
+ "loss": 0.1154,
+ "step": 20693
+ },
+ {
+ "epoch": 56.38692098092643,
+ "grad_norm": 2.5340893268585205,
+ "learning_rate": 8.423845888908595e-06,
+ "loss": 0.0692,
+ "step": 20694
+ },
+ {
+ "epoch": 56.38964577656676,
+ "grad_norm": 3.2634527683258057,
+ "learning_rate": 8.422974431088607e-06,
+ "loss": 0.1393,
+ "step": 20695
+ },
+ {
+ "epoch": 56.392370572207085,
+ "grad_norm": 2.417414426803589,
+ "learning_rate": 8.422102985550441e-06,
+ "loss": 0.0737,
+ "step": 20696
+ },
+ {
+ "epoch": 56.39509536784741,
+ "grad_norm": 1.854806661605835,
+ "learning_rate": 8.42123155230089e-06,
+ "loss": 0.0283,
+ "step": 20697
+ },
+ {
+ "epoch": 56.39782016348774,
+ "grad_norm": 5.431143283843994,
+ "learning_rate": 8.420360131346735e-06,
+ "loss": 0.156,
+ "step": 20698
+ },
+ {
+ "epoch": 56.40054495912806,
+ "grad_norm": 2.864485263824463,
+ "learning_rate": 8.419488722694765e-06,
+ "loss": 0.0479,
+ "step": 20699
+ },
+ {
+ "epoch": 56.403269754768395,
+ "grad_norm": 2.7509684562683105,
+ "learning_rate": 8.418617326351765e-06,
+ "loss": 0.0946,
+ "step": 20700
+ },
+ {
+ "epoch": 56.40599455040872,
+ "grad_norm": 2.9687201976776123,
+ "learning_rate": 8.417745942324522e-06,
+ "loss": 0.1313,
+ "step": 20701
+ },
+ {
+ "epoch": 56.40871934604905,
+ "grad_norm": 3.037707805633545,
+ "learning_rate": 8.416874570619825e-06,
+ "loss": 0.0539,
+ "step": 20702
+ },
+ {
+ "epoch": 56.41144414168937,
+ "grad_norm": 4.167900562286377,
+ "learning_rate": 8.416003211244454e-06,
+ "loss": 0.0461,
+ "step": 20703
+ },
+ {
+ "epoch": 56.4141689373297,
+ "grad_norm": 5.195884704589844,
+ "learning_rate": 8.415131864205204e-06,
+ "loss": 0.0332,
+ "step": 20704
+ },
+ {
+ "epoch": 56.416893732970024,
+ "grad_norm": 2.512338161468506,
+ "learning_rate": 8.414260529508851e-06,
+ "loss": 0.167,
+ "step": 20705
+ },
+ {
+ "epoch": 56.41961852861036,
+ "grad_norm": 5.752065181732178,
+ "learning_rate": 8.41338920716219e-06,
+ "loss": 0.1377,
+ "step": 20706
+ },
+ {
+ "epoch": 56.42234332425068,
+ "grad_norm": 2.0384113788604736,
+ "learning_rate": 8.412517897172001e-06,
+ "loss": 0.0292,
+ "step": 20707
+ },
+ {
+ "epoch": 56.42506811989101,
+ "grad_norm": 2.420250177383423,
+ "learning_rate": 8.411646599545068e-06,
+ "loss": 0.0213,
+ "step": 20708
+ },
+ {
+ "epoch": 56.427792915531334,
+ "grad_norm": 2.1440045833587646,
+ "learning_rate": 8.410775314288186e-06,
+ "loss": 0.088,
+ "step": 20709
+ },
+ {
+ "epoch": 56.43051771117166,
+ "grad_norm": 2.6589221954345703,
+ "learning_rate": 8.40990404140813e-06,
+ "loss": 0.0774,
+ "step": 20710
+ },
+ {
+ "epoch": 56.433242506811986,
+ "grad_norm": 2.568405866622925,
+ "learning_rate": 8.409032780911695e-06,
+ "loss": 0.0357,
+ "step": 20711
+ },
+ {
+ "epoch": 56.43596730245232,
+ "grad_norm": 3.223337411880493,
+ "learning_rate": 8.408161532805659e-06,
+ "loss": 0.0334,
+ "step": 20712
+ },
+ {
+ "epoch": 56.438692098092645,
+ "grad_norm": 2.51682186126709,
+ "learning_rate": 8.40729029709681e-06,
+ "loss": 0.0405,
+ "step": 20713
+ },
+ {
+ "epoch": 56.44141689373297,
+ "grad_norm": 2.534538984298706,
+ "learning_rate": 8.406419073791933e-06,
+ "loss": 0.0409,
+ "step": 20714
+ },
+ {
+ "epoch": 56.444141689373296,
+ "grad_norm": 2.168865442276001,
+ "learning_rate": 8.405547862897813e-06,
+ "loss": 0.0312,
+ "step": 20715
+ },
+ {
+ "epoch": 56.44686648501362,
+ "grad_norm": 2.3076016902923584,
+ "learning_rate": 8.404676664421235e-06,
+ "loss": 0.0306,
+ "step": 20716
+ },
+ {
+ "epoch": 56.44959128065395,
+ "grad_norm": 3.3099610805511475,
+ "learning_rate": 8.403805478368983e-06,
+ "loss": 0.0886,
+ "step": 20717
+ },
+ {
+ "epoch": 56.45231607629428,
+ "grad_norm": 2.0427534580230713,
+ "learning_rate": 8.402934304747843e-06,
+ "loss": 0.0264,
+ "step": 20718
+ },
+ {
+ "epoch": 56.45504087193461,
+ "grad_norm": 2.8177707195281982,
+ "learning_rate": 8.402063143564601e-06,
+ "loss": 0.0572,
+ "step": 20719
+ },
+ {
+ "epoch": 56.45776566757493,
+ "grad_norm": 3.1211230754852295,
+ "learning_rate": 8.401191994826036e-06,
+ "loss": 0.1158,
+ "step": 20720
+ },
+ {
+ "epoch": 56.46049046321526,
+ "grad_norm": 2.853895902633667,
+ "learning_rate": 8.40032085853894e-06,
+ "loss": 0.0777,
+ "step": 20721
+ },
+ {
+ "epoch": 56.463215258855584,
+ "grad_norm": 2.133054256439209,
+ "learning_rate": 8.399449734710092e-06,
+ "loss": 0.0336,
+ "step": 20722
+ },
+ {
+ "epoch": 56.46594005449591,
+ "grad_norm": 2.890558958053589,
+ "learning_rate": 8.398578623346277e-06,
+ "loss": 0.0756,
+ "step": 20723
+ },
+ {
+ "epoch": 56.46866485013624,
+ "grad_norm": 2.1591196060180664,
+ "learning_rate": 8.397707524454283e-06,
+ "loss": 0.125,
+ "step": 20724
+ },
+ {
+ "epoch": 56.47138964577657,
+ "grad_norm": 1.9825929403305054,
+ "learning_rate": 8.396836438040885e-06,
+ "loss": 0.0408,
+ "step": 20725
+ },
+ {
+ "epoch": 56.474114441416894,
+ "grad_norm": 5.621065139770508,
+ "learning_rate": 8.395965364112877e-06,
+ "loss": 0.159,
+ "step": 20726
+ },
+ {
+ "epoch": 56.47683923705722,
+ "grad_norm": 2.358057737350464,
+ "learning_rate": 8.395094302677041e-06,
+ "loss": 0.052,
+ "step": 20727
+ },
+ {
+ "epoch": 56.479564032697546,
+ "grad_norm": 2.3166141510009766,
+ "learning_rate": 8.394223253740157e-06,
+ "loss": 0.03,
+ "step": 20728
+ },
+ {
+ "epoch": 56.48228882833787,
+ "grad_norm": 2.3523435592651367,
+ "learning_rate": 8.393352217309012e-06,
+ "loss": 0.1132,
+ "step": 20729
+ },
+ {
+ "epoch": 56.485013623978205,
+ "grad_norm": 3.071815013885498,
+ "learning_rate": 8.392481193390388e-06,
+ "loss": 0.21,
+ "step": 20730
+ },
+ {
+ "epoch": 56.48773841961853,
+ "grad_norm": 2.6350560188293457,
+ "learning_rate": 8.391610181991068e-06,
+ "loss": 0.0811,
+ "step": 20731
+ },
+ {
+ "epoch": 56.490463215258856,
+ "grad_norm": 2.4097328186035156,
+ "learning_rate": 8.390739183117836e-06,
+ "loss": 0.0344,
+ "step": 20732
+ },
+ {
+ "epoch": 56.49318801089918,
+ "grad_norm": 2.6954739093780518,
+ "learning_rate": 8.389868196777478e-06,
+ "loss": 0.0667,
+ "step": 20733
+ },
+ {
+ "epoch": 56.49591280653951,
+ "grad_norm": 2.6090426445007324,
+ "learning_rate": 8.388997222976772e-06,
+ "loss": 0.0333,
+ "step": 20734
+ },
+ {
+ "epoch": 56.49863760217983,
+ "grad_norm": 4.972955226898193,
+ "learning_rate": 8.388126261722504e-06,
+ "loss": 0.2109,
+ "step": 20735
+ },
+ {
+ "epoch": 56.50136239782017,
+ "grad_norm": 2.6081371307373047,
+ "learning_rate": 8.387255313021456e-06,
+ "loss": 0.0399,
+ "step": 20736
+ },
+ {
+ "epoch": 56.50408719346049,
+ "grad_norm": 2.5897276401519775,
+ "learning_rate": 8.386384376880414e-06,
+ "loss": 0.0492,
+ "step": 20737
+ },
+ {
+ "epoch": 56.50681198910082,
+ "grad_norm": 2.5252935886383057,
+ "learning_rate": 8.385513453306157e-06,
+ "loss": 0.0287,
+ "step": 20738
+ },
+ {
+ "epoch": 56.509536784741144,
+ "grad_norm": 3.5735011100769043,
+ "learning_rate": 8.38464254230547e-06,
+ "loss": 0.0765,
+ "step": 20739
+ },
+ {
+ "epoch": 56.51226158038147,
+ "grad_norm": 2.167325019836426,
+ "learning_rate": 8.383771643885134e-06,
+ "loss": 0.0501,
+ "step": 20740
+ },
+ {
+ "epoch": 56.514986376021795,
+ "grad_norm": 2.2223894596099854,
+ "learning_rate": 8.382900758051934e-06,
+ "loss": 0.2183,
+ "step": 20741
+ },
+ {
+ "epoch": 56.51771117166213,
+ "grad_norm": 2.201897621154785,
+ "learning_rate": 8.382029884812644e-06,
+ "loss": 0.2116,
+ "step": 20742
+ },
+ {
+ "epoch": 56.520435967302454,
+ "grad_norm": 4.780940055847168,
+ "learning_rate": 8.381159024174061e-06,
+ "loss": 0.1011,
+ "step": 20743
+ },
+ {
+ "epoch": 56.52316076294278,
+ "grad_norm": 2.0703508853912354,
+ "learning_rate": 8.380288176142957e-06,
+ "loss": 0.0285,
+ "step": 20744
+ },
+ {
+ "epoch": 56.525885558583106,
+ "grad_norm": 2.778540849685669,
+ "learning_rate": 8.379417340726118e-06,
+ "loss": 0.0926,
+ "step": 20745
+ },
+ {
+ "epoch": 56.52861035422343,
+ "grad_norm": 3.066202163696289,
+ "learning_rate": 8.378546517930322e-06,
+ "loss": 0.0636,
+ "step": 20746
+ },
+ {
+ "epoch": 56.53133514986376,
+ "grad_norm": 2.7569167613983154,
+ "learning_rate": 8.377675707762356e-06,
+ "loss": 0.0472,
+ "step": 20747
+ },
+ {
+ "epoch": 56.53405994550409,
+ "grad_norm": 2.920428514480591,
+ "learning_rate": 8.376804910228996e-06,
+ "loss": 0.0687,
+ "step": 20748
+ },
+ {
+ "epoch": 56.536784741144416,
+ "grad_norm": 2.2418415546417236,
+ "learning_rate": 8.375934125337032e-06,
+ "loss": 0.0332,
+ "step": 20749
+ },
+ {
+ "epoch": 56.53950953678474,
+ "grad_norm": 2.39054012298584,
+ "learning_rate": 8.375063353093237e-06,
+ "loss": 0.0778,
+ "step": 20750
+ },
+ {
+ "epoch": 56.54223433242507,
+ "grad_norm": 2.1297268867492676,
+ "learning_rate": 8.3741925935044e-06,
+ "loss": 0.0818,
+ "step": 20751
+ },
+ {
+ "epoch": 56.54495912806539,
+ "grad_norm": 4.210344314575195,
+ "learning_rate": 8.373321846577295e-06,
+ "loss": 0.1679,
+ "step": 20752
+ },
+ {
+ "epoch": 56.54768392370572,
+ "grad_norm": 2.438084363937378,
+ "learning_rate": 8.37245111231871e-06,
+ "loss": 0.0413,
+ "step": 20753
+ },
+ {
+ "epoch": 56.55040871934605,
+ "grad_norm": 3.1605098247528076,
+ "learning_rate": 8.371580390735421e-06,
+ "loss": 0.0614,
+ "step": 20754
+ },
+ {
+ "epoch": 56.55313351498638,
+ "grad_norm": 1.9434828758239746,
+ "learning_rate": 8.370709681834213e-06,
+ "loss": 0.0407,
+ "step": 20755
+ },
+ {
+ "epoch": 56.555858310626704,
+ "grad_norm": 2.3503246307373047,
+ "learning_rate": 8.369838985621864e-06,
+ "loss": 0.041,
+ "step": 20756
+ },
+ {
+ "epoch": 56.55858310626703,
+ "grad_norm": 1.9912911653518677,
+ "learning_rate": 8.368968302105158e-06,
+ "loss": 0.0315,
+ "step": 20757
+ },
+ {
+ "epoch": 56.561307901907355,
+ "grad_norm": 2.3158204555511475,
+ "learning_rate": 8.368097631290873e-06,
+ "loss": 0.052,
+ "step": 20758
+ },
+ {
+ "epoch": 56.56403269754768,
+ "grad_norm": 4.852734565734863,
+ "learning_rate": 8.367226973185789e-06,
+ "loss": 0.0453,
+ "step": 20759
+ },
+ {
+ "epoch": 56.566757493188014,
+ "grad_norm": 2.388352155685425,
+ "learning_rate": 8.366356327796693e-06,
+ "loss": 0.1193,
+ "step": 20760
+ },
+ {
+ "epoch": 56.56948228882834,
+ "grad_norm": 1.8992738723754883,
+ "learning_rate": 8.36548569513036e-06,
+ "loss": 0.0388,
+ "step": 20761
+ },
+ {
+ "epoch": 56.572207084468666,
+ "grad_norm": 2.091442823410034,
+ "learning_rate": 8.364615075193572e-06,
+ "loss": 0.0279,
+ "step": 20762
+ },
+ {
+ "epoch": 56.57493188010899,
+ "grad_norm": 3.6499760150909424,
+ "learning_rate": 8.363744467993108e-06,
+ "loss": 0.0497,
+ "step": 20763
+ },
+ {
+ "epoch": 56.57765667574932,
+ "grad_norm": 2.720569133758545,
+ "learning_rate": 8.362873873535751e-06,
+ "loss": 0.0538,
+ "step": 20764
+ },
+ {
+ "epoch": 56.58038147138964,
+ "grad_norm": 2.8812406063079834,
+ "learning_rate": 8.362003291828279e-06,
+ "loss": 0.0779,
+ "step": 20765
+ },
+ {
+ "epoch": 56.583106267029976,
+ "grad_norm": 2.473677635192871,
+ "learning_rate": 8.361132722877471e-06,
+ "loss": 0.1427,
+ "step": 20766
+ },
+ {
+ "epoch": 56.5858310626703,
+ "grad_norm": 3.450197219848633,
+ "learning_rate": 8.360262166690112e-06,
+ "loss": 0.0545,
+ "step": 20767
+ },
+ {
+ "epoch": 56.58855585831063,
+ "grad_norm": 1.5093131065368652,
+ "learning_rate": 8.359391623272976e-06,
+ "loss": 0.0303,
+ "step": 20768
+ },
+ {
+ "epoch": 56.59128065395095,
+ "grad_norm": 3.3580117225646973,
+ "learning_rate": 8.358521092632848e-06,
+ "loss": 0.0976,
+ "step": 20769
+ },
+ {
+ "epoch": 56.59400544959128,
+ "grad_norm": 2.2803025245666504,
+ "learning_rate": 8.357650574776501e-06,
+ "loss": 0.0366,
+ "step": 20770
+ },
+ {
+ "epoch": 56.596730245231605,
+ "grad_norm": 2.3760855197906494,
+ "learning_rate": 8.356780069710721e-06,
+ "loss": 0.0725,
+ "step": 20771
+ },
+ {
+ "epoch": 56.59945504087194,
+ "grad_norm": 2.828434944152832,
+ "learning_rate": 8.355909577442284e-06,
+ "loss": 0.1305,
+ "step": 20772
+ },
+ {
+ "epoch": 56.60217983651226,
+ "grad_norm": 3.4503304958343506,
+ "learning_rate": 8.355039097977972e-06,
+ "loss": 0.157,
+ "step": 20773
+ },
+ {
+ "epoch": 56.60490463215259,
+ "grad_norm": 2.234631299972534,
+ "learning_rate": 8.35416863132456e-06,
+ "loss": 0.151,
+ "step": 20774
+ },
+ {
+ "epoch": 56.607629427792915,
+ "grad_norm": 2.127695083618164,
+ "learning_rate": 8.353298177488832e-06,
+ "loss": 0.0415,
+ "step": 20775
+ },
+ {
+ "epoch": 56.61035422343324,
+ "grad_norm": 2.5823354721069336,
+ "learning_rate": 8.35242773647756e-06,
+ "loss": 0.032,
+ "step": 20776
+ },
+ {
+ "epoch": 56.61307901907357,
+ "grad_norm": 2.3928072452545166,
+ "learning_rate": 8.35155730829753e-06,
+ "loss": 0.0402,
+ "step": 20777
+ },
+ {
+ "epoch": 56.6158038147139,
+ "grad_norm": 2.4320483207702637,
+ "learning_rate": 8.350686892955522e-06,
+ "loss": 0.0833,
+ "step": 20778
+ },
+ {
+ "epoch": 56.618528610354225,
+ "grad_norm": 2.9904024600982666,
+ "learning_rate": 8.349816490458307e-06,
+ "loss": 0.1462,
+ "step": 20779
+ },
+ {
+ "epoch": 56.62125340599455,
+ "grad_norm": 4.9901862144470215,
+ "learning_rate": 8.348946100812672e-06,
+ "loss": 0.0557,
+ "step": 20780
+ },
+ {
+ "epoch": 56.62397820163488,
+ "grad_norm": 2.481612205505371,
+ "learning_rate": 8.34807572402539e-06,
+ "loss": 0.069,
+ "step": 20781
+ },
+ {
+ "epoch": 56.6267029972752,
+ "grad_norm": 1.8919349908828735,
+ "learning_rate": 8.347205360103243e-06,
+ "loss": 0.0443,
+ "step": 20782
+ },
+ {
+ "epoch": 56.62942779291553,
+ "grad_norm": 2.32675838470459,
+ "learning_rate": 8.346335009053004e-06,
+ "loss": 0.0751,
+ "step": 20783
+ },
+ {
+ "epoch": 56.63215258855586,
+ "grad_norm": 3.732267379760742,
+ "learning_rate": 8.345464670881458e-06,
+ "loss": 0.0774,
+ "step": 20784
+ },
+ {
+ "epoch": 56.63487738419619,
+ "grad_norm": 2.2233803272247314,
+ "learning_rate": 8.344594345595377e-06,
+ "loss": 0.1248,
+ "step": 20785
+ },
+ {
+ "epoch": 56.63760217983651,
+ "grad_norm": 2.493915319442749,
+ "learning_rate": 8.343724033201544e-06,
+ "loss": 0.0673,
+ "step": 20786
+ },
+ {
+ "epoch": 56.64032697547684,
+ "grad_norm": 2.4105465412139893,
+ "learning_rate": 8.342853733706733e-06,
+ "loss": 0.0739,
+ "step": 20787
+ },
+ {
+ "epoch": 56.643051771117165,
+ "grad_norm": 19.311742782592773,
+ "learning_rate": 8.341983447117725e-06,
+ "loss": 0.0494,
+ "step": 20788
+ },
+ {
+ "epoch": 56.64577656675749,
+ "grad_norm": 1.7920578718185425,
+ "learning_rate": 8.341113173441298e-06,
+ "loss": 0.0389,
+ "step": 20789
+ },
+ {
+ "epoch": 56.64850136239782,
+ "grad_norm": 2.2788193225860596,
+ "learning_rate": 8.340242912684225e-06,
+ "loss": 0.0307,
+ "step": 20790
+ },
+ {
+ "epoch": 56.65122615803815,
+ "grad_norm": 3.23583984375,
+ "learning_rate": 8.33937266485329e-06,
+ "loss": 0.0695,
+ "step": 20791
+ },
+ {
+ "epoch": 56.653950953678475,
+ "grad_norm": 2.4873406887054443,
+ "learning_rate": 8.338502429955264e-06,
+ "loss": 0.0379,
+ "step": 20792
+ },
+ {
+ "epoch": 56.6566757493188,
+ "grad_norm": 2.785691499710083,
+ "learning_rate": 8.337632207996932e-06,
+ "loss": 0.0631,
+ "step": 20793
+ },
+ {
+ "epoch": 56.65940054495913,
+ "grad_norm": 2.0126373767852783,
+ "learning_rate": 8.336761998985059e-06,
+ "loss": 0.0386,
+ "step": 20794
+ },
+ {
+ "epoch": 56.66212534059945,
+ "grad_norm": 2.5082225799560547,
+ "learning_rate": 8.335891802926435e-06,
+ "loss": 0.04,
+ "step": 20795
+ },
+ {
+ "epoch": 56.664850136239785,
+ "grad_norm": 3.539358615875244,
+ "learning_rate": 8.335021619827833e-06,
+ "loss": 0.1545,
+ "step": 20796
+ },
+ {
+ "epoch": 56.66757493188011,
+ "grad_norm": 2.532963514328003,
+ "learning_rate": 8.334151449696027e-06,
+ "loss": 0.0891,
+ "step": 20797
+ },
+ {
+ "epoch": 56.67029972752044,
+ "grad_norm": 2.255316734313965,
+ "learning_rate": 8.333281292537799e-06,
+ "loss": 0.0551,
+ "step": 20798
+ },
+ {
+ "epoch": 56.67302452316076,
+ "grad_norm": 2.9347267150878906,
+ "learning_rate": 8.33241114835992e-06,
+ "loss": 0.0899,
+ "step": 20799
+ },
+ {
+ "epoch": 56.67574931880109,
+ "grad_norm": 4.750433921813965,
+ "learning_rate": 8.331541017169172e-06,
+ "loss": 0.0504,
+ "step": 20800
+ },
+ {
+ "epoch": 56.678474114441414,
+ "grad_norm": 2.0265555381774902,
+ "learning_rate": 8.330670898972326e-06,
+ "loss": 0.028,
+ "step": 20801
+ },
+ {
+ "epoch": 56.68119891008175,
+ "grad_norm": 2.701486110687256,
+ "learning_rate": 8.329800793776163e-06,
+ "loss": 0.038,
+ "step": 20802
+ },
+ {
+ "epoch": 56.68392370572207,
+ "grad_norm": 3.468280553817749,
+ "learning_rate": 8.328930701587456e-06,
+ "loss": 0.0482,
+ "step": 20803
+ },
+ {
+ "epoch": 56.6866485013624,
+ "grad_norm": 2.5730199813842773,
+ "learning_rate": 8.328060622412986e-06,
+ "loss": 0.0261,
+ "step": 20804
+ },
+ {
+ "epoch": 56.689373297002724,
+ "grad_norm": 7.425819396972656,
+ "learning_rate": 8.327190556259523e-06,
+ "loss": 0.0467,
+ "step": 20805
+ },
+ {
+ "epoch": 56.69209809264305,
+ "grad_norm": 7.018514633178711,
+ "learning_rate": 8.326320503133847e-06,
+ "loss": 0.0602,
+ "step": 20806
+ },
+ {
+ "epoch": 56.694822888283376,
+ "grad_norm": 2.188319683074951,
+ "learning_rate": 8.325450463042733e-06,
+ "loss": 0.0482,
+ "step": 20807
+ },
+ {
+ "epoch": 56.69754768392371,
+ "grad_norm": 2.079368829727173,
+ "learning_rate": 8.324580435992959e-06,
+ "loss": 0.0245,
+ "step": 20808
+ },
+ {
+ "epoch": 56.700272479564035,
+ "grad_norm": 3.7903432846069336,
+ "learning_rate": 8.323710421991295e-06,
+ "loss": 0.3294,
+ "step": 20809
+ },
+ {
+ "epoch": 56.70299727520436,
+ "grad_norm": 2.687669277191162,
+ "learning_rate": 8.32284042104452e-06,
+ "loss": 0.1986,
+ "step": 20810
+ },
+ {
+ "epoch": 56.705722070844686,
+ "grad_norm": 1.7331440448760986,
+ "learning_rate": 8.321970433159409e-06,
+ "loss": 0.0353,
+ "step": 20811
+ },
+ {
+ "epoch": 56.70844686648501,
+ "grad_norm": 3.055025339126587,
+ "learning_rate": 8.321100458342742e-06,
+ "loss": 0.0474,
+ "step": 20812
+ },
+ {
+ "epoch": 56.71117166212534,
+ "grad_norm": 2.7950997352600098,
+ "learning_rate": 8.32023049660129e-06,
+ "loss": 0.0932,
+ "step": 20813
+ },
+ {
+ "epoch": 56.71389645776567,
+ "grad_norm": 5.86964225769043,
+ "learning_rate": 8.319360547941828e-06,
+ "loss": 0.052,
+ "step": 20814
+ },
+ {
+ "epoch": 56.716621253406,
+ "grad_norm": 2.816690444946289,
+ "learning_rate": 8.318490612371132e-06,
+ "loss": 0.0497,
+ "step": 20815
+ },
+ {
+ "epoch": 56.71934604904632,
+ "grad_norm": 3.4259767532348633,
+ "learning_rate": 8.317620689895979e-06,
+ "loss": 0.0657,
+ "step": 20816
+ },
+ {
+ "epoch": 56.72207084468665,
+ "grad_norm": 4.289257049560547,
+ "learning_rate": 8.316750780523139e-06,
+ "loss": 0.0314,
+ "step": 20817
+ },
+ {
+ "epoch": 56.724795640326974,
+ "grad_norm": 2.8622705936431885,
+ "learning_rate": 8.315880884259393e-06,
+ "loss": 0.1138,
+ "step": 20818
+ },
+ {
+ "epoch": 56.7275204359673,
+ "grad_norm": 2.092207670211792,
+ "learning_rate": 8.31501100111151e-06,
+ "loss": 0.0792,
+ "step": 20819
+ },
+ {
+ "epoch": 56.73024523160763,
+ "grad_norm": 2.9541103839874268,
+ "learning_rate": 8.314141131086268e-06,
+ "loss": 0.105,
+ "step": 20820
+ },
+ {
+ "epoch": 56.73297002724796,
+ "grad_norm": 2.1705892086029053,
+ "learning_rate": 8.31327127419044e-06,
+ "loss": 0.1505,
+ "step": 20821
+ },
+ {
+ "epoch": 56.735694822888284,
+ "grad_norm": 2.2566874027252197,
+ "learning_rate": 8.312401430430802e-06,
+ "loss": 0.0698,
+ "step": 20822
+ },
+ {
+ "epoch": 56.73841961852861,
+ "grad_norm": 2.7813735008239746,
+ "learning_rate": 8.311531599814125e-06,
+ "loss": 0.0458,
+ "step": 20823
+ },
+ {
+ "epoch": 56.741144414168936,
+ "grad_norm": 2.7275731563568115,
+ "learning_rate": 8.310661782347189e-06,
+ "loss": 0.0512,
+ "step": 20824
+ },
+ {
+ "epoch": 56.74386920980926,
+ "grad_norm": 2.1151087284088135,
+ "learning_rate": 8.309791978036762e-06,
+ "loss": 0.0301,
+ "step": 20825
+ },
+ {
+ "epoch": 56.746594005449595,
+ "grad_norm": 2.2018349170684814,
+ "learning_rate": 8.308922186889623e-06,
+ "loss": 0.2252,
+ "step": 20826
+ },
+ {
+ "epoch": 56.74931880108992,
+ "grad_norm": 2.043675422668457,
+ "learning_rate": 8.308052408912541e-06,
+ "loss": 0.0394,
+ "step": 20827
+ },
+ {
+ "epoch": 56.752043596730246,
+ "grad_norm": 2.9567458629608154,
+ "learning_rate": 8.307182644112291e-06,
+ "loss": 0.051,
+ "step": 20828
+ },
+ {
+ "epoch": 56.75476839237057,
+ "grad_norm": 2.055121421813965,
+ "learning_rate": 8.306312892495648e-06,
+ "loss": 0.0292,
+ "step": 20829
+ },
+ {
+ "epoch": 56.7574931880109,
+ "grad_norm": 2.4291434288024902,
+ "learning_rate": 8.30544315406939e-06,
+ "loss": 0.0294,
+ "step": 20830
+ },
+ {
+ "epoch": 56.76021798365122,
+ "grad_norm": 1.9938892126083374,
+ "learning_rate": 8.304573428840284e-06,
+ "loss": 0.0393,
+ "step": 20831
+ },
+ {
+ "epoch": 56.762942779291556,
+ "grad_norm": 2.501277208328247,
+ "learning_rate": 8.303703716815108e-06,
+ "loss": 0.0431,
+ "step": 20832
+ },
+ {
+ "epoch": 56.76566757493188,
+ "grad_norm": 2.556356906890869,
+ "learning_rate": 8.302834018000628e-06,
+ "loss": 0.1263,
+ "step": 20833
+ },
+ {
+ "epoch": 56.76839237057221,
+ "grad_norm": 2.0183589458465576,
+ "learning_rate": 8.301964332403627e-06,
+ "loss": 0.1559,
+ "step": 20834
+ },
+ {
+ "epoch": 56.771117166212534,
+ "grad_norm": 3.1807539463043213,
+ "learning_rate": 8.301094660030869e-06,
+ "loss": 0.0329,
+ "step": 20835
+ },
+ {
+ "epoch": 56.77384196185286,
+ "grad_norm": 2.7493278980255127,
+ "learning_rate": 8.300225000889135e-06,
+ "loss": 0.0404,
+ "step": 20836
+ },
+ {
+ "epoch": 56.776566757493185,
+ "grad_norm": 3.3814709186553955,
+ "learning_rate": 8.299355354985192e-06,
+ "loss": 0.0419,
+ "step": 20837
+ },
+ {
+ "epoch": 56.77929155313352,
+ "grad_norm": 2.3809821605682373,
+ "learning_rate": 8.298485722325815e-06,
+ "loss": 0.066,
+ "step": 20838
+ },
+ {
+ "epoch": 56.782016348773844,
+ "grad_norm": 2.8231775760650635,
+ "learning_rate": 8.297616102917776e-06,
+ "loss": 0.1279,
+ "step": 20839
+ },
+ {
+ "epoch": 56.78474114441417,
+ "grad_norm": 3.062037944793701,
+ "learning_rate": 8.29674649676785e-06,
+ "loss": 0.0508,
+ "step": 20840
+ },
+ {
+ "epoch": 56.787465940054496,
+ "grad_norm": 2.0650081634521484,
+ "learning_rate": 8.295876903882804e-06,
+ "loss": 0.0218,
+ "step": 20841
+ },
+ {
+ "epoch": 56.79019073569482,
+ "grad_norm": 3.0778565406799316,
+ "learning_rate": 8.295007324269417e-06,
+ "loss": 0.0584,
+ "step": 20842
+ },
+ {
+ "epoch": 56.79291553133515,
+ "grad_norm": 3.4615602493286133,
+ "learning_rate": 8.294137757934456e-06,
+ "loss": 0.0732,
+ "step": 20843
+ },
+ {
+ "epoch": 56.79564032697548,
+ "grad_norm": 3.7161262035369873,
+ "learning_rate": 8.293268204884696e-06,
+ "loss": 0.034,
+ "step": 20844
+ },
+ {
+ "epoch": 56.798365122615806,
+ "grad_norm": 2.8164594173431396,
+ "learning_rate": 8.292398665126906e-06,
+ "loss": 0.0538,
+ "step": 20845
+ },
+ {
+ "epoch": 56.80108991825613,
+ "grad_norm": 2.560220956802368,
+ "learning_rate": 8.291529138667862e-06,
+ "loss": 0.1252,
+ "step": 20846
+ },
+ {
+ "epoch": 56.80381471389646,
+ "grad_norm": 2.0368690490722656,
+ "learning_rate": 8.290659625514336e-06,
+ "loss": 0.0546,
+ "step": 20847
+ },
+ {
+ "epoch": 56.80653950953678,
+ "grad_norm": 1.9523507356643677,
+ "learning_rate": 8.289790125673095e-06,
+ "loss": 0.0494,
+ "step": 20848
+ },
+ {
+ "epoch": 56.80926430517711,
+ "grad_norm": 1.6560359001159668,
+ "learning_rate": 8.288920639150915e-06,
+ "loss": 0.022,
+ "step": 20849
+ },
+ {
+ "epoch": 56.81198910081744,
+ "grad_norm": 2.370422601699829,
+ "learning_rate": 8.288051165954566e-06,
+ "loss": 0.0594,
+ "step": 20850
+ },
+ {
+ "epoch": 56.81471389645777,
+ "grad_norm": 2.056720018386841,
+ "learning_rate": 8.28718170609082e-06,
+ "loss": 0.0962,
+ "step": 20851
+ },
+ {
+ "epoch": 56.817438692098094,
+ "grad_norm": 2.366223096847534,
+ "learning_rate": 8.286312259566445e-06,
+ "loss": 0.027,
+ "step": 20852
+ },
+ {
+ "epoch": 56.82016348773842,
+ "grad_norm": 1.507956624031067,
+ "learning_rate": 8.285442826388217e-06,
+ "loss": 0.0275,
+ "step": 20853
+ },
+ {
+ "epoch": 56.822888283378745,
+ "grad_norm": 6.268251419067383,
+ "learning_rate": 8.284573406562906e-06,
+ "loss": 0.113,
+ "step": 20854
+ },
+ {
+ "epoch": 56.82561307901907,
+ "grad_norm": 1.768964409828186,
+ "learning_rate": 8.283704000097281e-06,
+ "loss": 0.0335,
+ "step": 20855
+ },
+ {
+ "epoch": 56.828337874659404,
+ "grad_norm": 2.228614091873169,
+ "learning_rate": 8.282834606998114e-06,
+ "loss": 0.135,
+ "step": 20856
+ },
+ {
+ "epoch": 56.83106267029973,
+ "grad_norm": 2.733891248703003,
+ "learning_rate": 8.281965227272176e-06,
+ "loss": 0.0311,
+ "step": 20857
+ },
+ {
+ "epoch": 56.833787465940055,
+ "grad_norm": 3.0152063369750977,
+ "learning_rate": 8.281095860926238e-06,
+ "loss": 0.0491,
+ "step": 20858
+ },
+ {
+ "epoch": 56.83651226158038,
+ "grad_norm": 2.6993327140808105,
+ "learning_rate": 8.280226507967068e-06,
+ "loss": 0.0307,
+ "step": 20859
+ },
+ {
+ "epoch": 56.83923705722071,
+ "grad_norm": 2.6646416187286377,
+ "learning_rate": 8.279357168401442e-06,
+ "loss": 0.0327,
+ "step": 20860
+ },
+ {
+ "epoch": 56.84196185286103,
+ "grad_norm": 2.333613395690918,
+ "learning_rate": 8.278487842236125e-06,
+ "loss": 0.0857,
+ "step": 20861
+ },
+ {
+ "epoch": 56.844686648501366,
+ "grad_norm": 2.282902479171753,
+ "learning_rate": 8.277618529477886e-06,
+ "loss": 0.0769,
+ "step": 20862
+ },
+ {
+ "epoch": 56.84741144414169,
+ "grad_norm": 2.9541547298431396,
+ "learning_rate": 8.276749230133503e-06,
+ "loss": 0.0547,
+ "step": 20863
+ },
+ {
+ "epoch": 56.85013623978202,
+ "grad_norm": 1.8247817754745483,
+ "learning_rate": 8.27587994420974e-06,
+ "loss": 0.1708,
+ "step": 20864
+ },
+ {
+ "epoch": 56.85286103542234,
+ "grad_norm": 2.1009860038757324,
+ "learning_rate": 8.27501067171337e-06,
+ "loss": 0.0427,
+ "step": 20865
+ },
+ {
+ "epoch": 56.85558583106267,
+ "grad_norm": 3.201190233230591,
+ "learning_rate": 8.27414141265116e-06,
+ "loss": 0.0645,
+ "step": 20866
+ },
+ {
+ "epoch": 56.858310626702995,
+ "grad_norm": 2.668006420135498,
+ "learning_rate": 8.273272167029882e-06,
+ "loss": 0.0875,
+ "step": 20867
+ },
+ {
+ "epoch": 56.86103542234333,
+ "grad_norm": 2.7157092094421387,
+ "learning_rate": 8.272402934856302e-06,
+ "loss": 0.0729,
+ "step": 20868
+ },
+ {
+ "epoch": 56.86376021798365,
+ "grad_norm": 1.8871477842330933,
+ "learning_rate": 8.271533716137196e-06,
+ "loss": 0.1012,
+ "step": 20869
+ },
+ {
+ "epoch": 56.86648501362398,
+ "grad_norm": 2.2199900150299072,
+ "learning_rate": 8.270664510879328e-06,
+ "loss": 0.0431,
+ "step": 20870
+ },
+ {
+ "epoch": 56.869209809264305,
+ "grad_norm": 2.595329523086548,
+ "learning_rate": 8.269795319089471e-06,
+ "loss": 0.0349,
+ "step": 20871
+ },
+ {
+ "epoch": 56.87193460490463,
+ "grad_norm": 3.597379684448242,
+ "learning_rate": 8.26892614077439e-06,
+ "loss": 0.1493,
+ "step": 20872
+ },
+ {
+ "epoch": 56.87465940054496,
+ "grad_norm": 11.5690336227417,
+ "learning_rate": 8.26805697594086e-06,
+ "loss": 0.0339,
+ "step": 20873
+ },
+ {
+ "epoch": 56.87738419618529,
+ "grad_norm": 3.29909610748291,
+ "learning_rate": 8.267187824595641e-06,
+ "loss": 0.1456,
+ "step": 20874
+ },
+ {
+ "epoch": 56.880108991825615,
+ "grad_norm": 2.2447328567504883,
+ "learning_rate": 8.26631868674551e-06,
+ "loss": 0.0229,
+ "step": 20875
+ },
+ {
+ "epoch": 56.88283378746594,
+ "grad_norm": 3.354140281677246,
+ "learning_rate": 8.265449562397235e-06,
+ "loss": 0.0433,
+ "step": 20876
+ },
+ {
+ "epoch": 56.88555858310627,
+ "grad_norm": 2.8985211849212646,
+ "learning_rate": 8.26458045155758e-06,
+ "loss": 0.0313,
+ "step": 20877
+ },
+ {
+ "epoch": 56.88828337874659,
+ "grad_norm": 2.0361244678497314,
+ "learning_rate": 8.26371135423332e-06,
+ "loss": 0.1009,
+ "step": 20878
+ },
+ {
+ "epoch": 56.89100817438692,
+ "grad_norm": 2.502659320831299,
+ "learning_rate": 8.262842270431218e-06,
+ "loss": 0.0444,
+ "step": 20879
+ },
+ {
+ "epoch": 56.89373297002725,
+ "grad_norm": 2.110105037689209,
+ "learning_rate": 8.26197320015804e-06,
+ "loss": 0.0355,
+ "step": 20880
+ },
+ {
+ "epoch": 56.89645776566758,
+ "grad_norm": 3.2308454513549805,
+ "learning_rate": 8.261104143420566e-06,
+ "loss": 0.1302,
+ "step": 20881
+ },
+ {
+ "epoch": 56.8991825613079,
+ "grad_norm": 2.495879888534546,
+ "learning_rate": 8.260235100225553e-06,
+ "loss": 0.0528,
+ "step": 20882
+ },
+ {
+ "epoch": 56.90190735694823,
+ "grad_norm": 2.0051941871643066,
+ "learning_rate": 8.259366070579775e-06,
+ "loss": 0.1941,
+ "step": 20883
+ },
+ {
+ "epoch": 56.904632152588555,
+ "grad_norm": 1.781348466873169,
+ "learning_rate": 8.258497054489998e-06,
+ "loss": 0.0249,
+ "step": 20884
+ },
+ {
+ "epoch": 56.90735694822888,
+ "grad_norm": 2.745511054992676,
+ "learning_rate": 8.25762805196299e-06,
+ "loss": 0.0413,
+ "step": 20885
+ },
+ {
+ "epoch": 56.91008174386921,
+ "grad_norm": 2.58117413520813,
+ "learning_rate": 8.256759063005517e-06,
+ "loss": 0.0522,
+ "step": 20886
+ },
+ {
+ "epoch": 56.91280653950954,
+ "grad_norm": 1.5642834901809692,
+ "learning_rate": 8.255890087624352e-06,
+ "loss": 0.029,
+ "step": 20887
+ },
+ {
+ "epoch": 56.915531335149865,
+ "grad_norm": 2.9588351249694824,
+ "learning_rate": 8.255021125826255e-06,
+ "loss": 0.0482,
+ "step": 20888
+ },
+ {
+ "epoch": 56.91825613079019,
+ "grad_norm": 3.5047223567962646,
+ "learning_rate": 8.254152177618e-06,
+ "loss": 0.0503,
+ "step": 20889
+ },
+ {
+ "epoch": 56.920980926430516,
+ "grad_norm": 2.1992976665496826,
+ "learning_rate": 8.253283243006349e-06,
+ "loss": 0.1203,
+ "step": 20890
+ },
+ {
+ "epoch": 56.92370572207084,
+ "grad_norm": 2.678426504135132,
+ "learning_rate": 8.252414321998075e-06,
+ "loss": 0.0436,
+ "step": 20891
+ },
+ {
+ "epoch": 56.926430517711175,
+ "grad_norm": 2.232238292694092,
+ "learning_rate": 8.25154541459994e-06,
+ "loss": 0.0267,
+ "step": 20892
+ },
+ {
+ "epoch": 56.9291553133515,
+ "grad_norm": 2.0743160247802734,
+ "learning_rate": 8.250676520818712e-06,
+ "loss": 0.1233,
+ "step": 20893
+ },
+ {
+ "epoch": 56.93188010899183,
+ "grad_norm": 4.238556861877441,
+ "learning_rate": 8.24980764066116e-06,
+ "loss": 0.1991,
+ "step": 20894
+ },
+ {
+ "epoch": 56.93460490463215,
+ "grad_norm": 2.9712307453155518,
+ "learning_rate": 8.248938774134052e-06,
+ "loss": 0.0881,
+ "step": 20895
+ },
+ {
+ "epoch": 56.93732970027248,
+ "grad_norm": 4.094025135040283,
+ "learning_rate": 8.248069921244148e-06,
+ "loss": 0.083,
+ "step": 20896
+ },
+ {
+ "epoch": 56.940054495912804,
+ "grad_norm": 1.8340392112731934,
+ "learning_rate": 8.247201081998218e-06,
+ "loss": 0.0583,
+ "step": 20897
+ },
+ {
+ "epoch": 56.94277929155314,
+ "grad_norm": 2.69498348236084,
+ "learning_rate": 8.246332256403033e-06,
+ "loss": 0.0736,
+ "step": 20898
+ },
+ {
+ "epoch": 56.94550408719346,
+ "grad_norm": 2.15995454788208,
+ "learning_rate": 8.245463444465357e-06,
+ "loss": 0.0674,
+ "step": 20899
+ },
+ {
+ "epoch": 56.94822888283379,
+ "grad_norm": 1.7203888893127441,
+ "learning_rate": 8.244594646191953e-06,
+ "loss": 0.0675,
+ "step": 20900
+ },
+ {
+ "epoch": 56.950953678474114,
+ "grad_norm": 2.4894559383392334,
+ "learning_rate": 8.243725861589592e-06,
+ "loss": 0.0753,
+ "step": 20901
+ },
+ {
+ "epoch": 56.95367847411444,
+ "grad_norm": 2.2006258964538574,
+ "learning_rate": 8.242857090665036e-06,
+ "loss": 0.0318,
+ "step": 20902
+ },
+ {
+ "epoch": 56.956403269754766,
+ "grad_norm": 1.7957969903945923,
+ "learning_rate": 8.241988333425052e-06,
+ "loss": 0.0604,
+ "step": 20903
+ },
+ {
+ "epoch": 56.95912806539509,
+ "grad_norm": 2.330315589904785,
+ "learning_rate": 8.241119589876406e-06,
+ "loss": 0.1361,
+ "step": 20904
+ },
+ {
+ "epoch": 56.961852861035425,
+ "grad_norm": 1.5417507886886597,
+ "learning_rate": 8.240250860025866e-06,
+ "loss": 0.0303,
+ "step": 20905
+ },
+ {
+ "epoch": 56.96457765667575,
+ "grad_norm": 2.0778861045837402,
+ "learning_rate": 8.239382143880195e-06,
+ "loss": 0.0344,
+ "step": 20906
+ },
+ {
+ "epoch": 56.967302452316076,
+ "grad_norm": 4.799381256103516,
+ "learning_rate": 8.238513441446158e-06,
+ "loss": 0.1165,
+ "step": 20907
+ },
+ {
+ "epoch": 56.9700272479564,
+ "grad_norm": 1.3894482851028442,
+ "learning_rate": 8.237644752730522e-06,
+ "loss": 0.0284,
+ "step": 20908
+ },
+ {
+ "epoch": 56.97275204359673,
+ "grad_norm": 2.3884670734405518,
+ "learning_rate": 8.236776077740054e-06,
+ "loss": 0.0374,
+ "step": 20909
+ },
+ {
+ "epoch": 56.97547683923706,
+ "grad_norm": 1.9196860790252686,
+ "learning_rate": 8.235907416481514e-06,
+ "loss": 0.0418,
+ "step": 20910
+ },
+ {
+ "epoch": 56.97820163487739,
+ "grad_norm": 2.005380630493164,
+ "learning_rate": 8.235038768961672e-06,
+ "loss": 0.1762,
+ "step": 20911
+ },
+ {
+ "epoch": 56.98092643051771,
+ "grad_norm": 2.495137929916382,
+ "learning_rate": 8.23417013518729e-06,
+ "loss": 0.0294,
+ "step": 20912
+ },
+ {
+ "epoch": 56.98365122615804,
+ "grad_norm": 3.014270782470703,
+ "learning_rate": 8.233301515165138e-06,
+ "loss": 0.06,
+ "step": 20913
+ },
+ {
+ "epoch": 56.986376021798364,
+ "grad_norm": 2.0996055603027344,
+ "learning_rate": 8.23243290890197e-06,
+ "loss": 0.0321,
+ "step": 20914
+ },
+ {
+ "epoch": 56.98910081743869,
+ "grad_norm": 2.653420925140381,
+ "learning_rate": 8.23156431640456e-06,
+ "loss": 0.0906,
+ "step": 20915
+ },
+ {
+ "epoch": 56.991825613079016,
+ "grad_norm": 1.6294243335723877,
+ "learning_rate": 8.230695737679671e-06,
+ "loss": 0.0203,
+ "step": 20916
+ },
+ {
+ "epoch": 56.99455040871935,
+ "grad_norm": 3.245906114578247,
+ "learning_rate": 8.229827172734068e-06,
+ "loss": 0.1944,
+ "step": 20917
+ },
+ {
+ "epoch": 56.997275204359674,
+ "grad_norm": 2.1729485988616943,
+ "learning_rate": 8.228958621574514e-06,
+ "loss": 0.1808,
+ "step": 20918
+ },
+ {
+ "epoch": 57.0,
+ "grad_norm": 2.2420663833618164,
+ "learning_rate": 8.228090084207773e-06,
+ "loss": 0.0367,
+ "step": 20919
+ },
+ {
+ "epoch": 57.002724795640326,
+ "grad_norm": 4.189292907714844,
+ "learning_rate": 8.227221560640609e-06,
+ "loss": 0.0632,
+ "step": 20920
+ },
+ {
+ "epoch": 57.00544959128065,
+ "grad_norm": 3.07879376411438,
+ "learning_rate": 8.226353050879788e-06,
+ "loss": 0.0259,
+ "step": 20921
+ },
+ {
+ "epoch": 57.00817438692098,
+ "grad_norm": 2.5718867778778076,
+ "learning_rate": 8.225484554932069e-06,
+ "loss": 0.0469,
+ "step": 20922
+ },
+ {
+ "epoch": 57.01089918256131,
+ "grad_norm": 1.8573572635650635,
+ "learning_rate": 8.224616072804223e-06,
+ "loss": 0.0524,
+ "step": 20923
+ },
+ {
+ "epoch": 57.013623978201636,
+ "grad_norm": 1.7739444971084595,
+ "learning_rate": 8.223747604503005e-06,
+ "loss": 0.0332,
+ "step": 20924
+ },
+ {
+ "epoch": 57.01634877384196,
+ "grad_norm": 2.551164150238037,
+ "learning_rate": 8.22287915003519e-06,
+ "loss": 0.1202,
+ "step": 20925
+ },
+ {
+ "epoch": 57.01907356948229,
+ "grad_norm": 2.6704888343811035,
+ "learning_rate": 8.22201070940753e-06,
+ "loss": 0.1277,
+ "step": 20926
+ },
+ {
+ "epoch": 57.02179836512261,
+ "grad_norm": 3.1271657943725586,
+ "learning_rate": 8.221142282626795e-06,
+ "loss": 0.0545,
+ "step": 20927
+ },
+ {
+ "epoch": 57.02452316076294,
+ "grad_norm": 1.2829865217208862,
+ "learning_rate": 8.220273869699746e-06,
+ "loss": 0.0239,
+ "step": 20928
+ },
+ {
+ "epoch": 57.02724795640327,
+ "grad_norm": 2.1092469692230225,
+ "learning_rate": 8.219405470633149e-06,
+ "loss": 0.1383,
+ "step": 20929
+ },
+ {
+ "epoch": 57.0299727520436,
+ "grad_norm": 2.7003543376922607,
+ "learning_rate": 8.218537085433761e-06,
+ "loss": 0.0537,
+ "step": 20930
+ },
+ {
+ "epoch": 57.032697547683924,
+ "grad_norm": 2.3894262313842773,
+ "learning_rate": 8.217668714108348e-06,
+ "loss": 0.0251,
+ "step": 20931
+ },
+ {
+ "epoch": 57.03542234332425,
+ "grad_norm": 1.8427870273590088,
+ "learning_rate": 8.216800356663679e-06,
+ "loss": 0.077,
+ "step": 20932
+ },
+ {
+ "epoch": 57.038147138964575,
+ "grad_norm": 2.0688560009002686,
+ "learning_rate": 8.215932013106508e-06,
+ "loss": 0.0833,
+ "step": 20933
+ },
+ {
+ "epoch": 57.0408719346049,
+ "grad_norm": 2.014604330062866,
+ "learning_rate": 8.215063683443603e-06,
+ "loss": 0.0525,
+ "step": 20934
+ },
+ {
+ "epoch": 57.043596730245234,
+ "grad_norm": 180.33067321777344,
+ "learning_rate": 8.214195367681723e-06,
+ "loss": 0.0433,
+ "step": 20935
+ },
+ {
+ "epoch": 57.04632152588556,
+ "grad_norm": 2.3076910972595215,
+ "learning_rate": 8.213327065827635e-06,
+ "loss": 0.0353,
+ "step": 20936
+ },
+ {
+ "epoch": 57.049046321525886,
+ "grad_norm": 2.263047933578491,
+ "learning_rate": 8.212458777888094e-06,
+ "loss": 0.0338,
+ "step": 20937
+ },
+ {
+ "epoch": 57.05177111716621,
+ "grad_norm": 2.00123929977417,
+ "learning_rate": 8.21159050386987e-06,
+ "loss": 0.0343,
+ "step": 20938
+ },
+ {
+ "epoch": 57.05449591280654,
+ "grad_norm": 3.1213138103485107,
+ "learning_rate": 8.210722243779719e-06,
+ "loss": 0.083,
+ "step": 20939
+ },
+ {
+ "epoch": 57.05722070844686,
+ "grad_norm": 3.5313754081726074,
+ "learning_rate": 8.209853997624406e-06,
+ "loss": 0.0621,
+ "step": 20940
+ },
+ {
+ "epoch": 57.059945504087196,
+ "grad_norm": 3.6424551010131836,
+ "learning_rate": 8.208985765410694e-06,
+ "loss": 0.0844,
+ "step": 20941
+ },
+ {
+ "epoch": 57.06267029972752,
+ "grad_norm": 5.068265438079834,
+ "learning_rate": 8.20811754714534e-06,
+ "loss": 0.0851,
+ "step": 20942
+ },
+ {
+ "epoch": 57.06539509536785,
+ "grad_norm": 3.910783052444458,
+ "learning_rate": 8.207249342835114e-06,
+ "loss": 0.0795,
+ "step": 20943
+ },
+ {
+ "epoch": 57.06811989100817,
+ "grad_norm": 6.68663215637207,
+ "learning_rate": 8.206381152486769e-06,
+ "loss": 0.1267,
+ "step": 20944
+ },
+ {
+ "epoch": 57.0708446866485,
+ "grad_norm": 4.127377033233643,
+ "learning_rate": 8.205512976107071e-06,
+ "loss": 0.0651,
+ "step": 20945
+ },
+ {
+ "epoch": 57.073569482288825,
+ "grad_norm": 4.0515851974487305,
+ "learning_rate": 8.204644813702778e-06,
+ "loss": 0.0747,
+ "step": 20946
+ },
+ {
+ "epoch": 57.07629427792916,
+ "grad_norm": 4.917068958282471,
+ "learning_rate": 8.203776665280656e-06,
+ "loss": 0.1138,
+ "step": 20947
+ },
+ {
+ "epoch": 57.079019073569484,
+ "grad_norm": 5.808668613433838,
+ "learning_rate": 8.202908530847463e-06,
+ "loss": 0.1247,
+ "step": 20948
+ },
+ {
+ "epoch": 57.08174386920981,
+ "grad_norm": 3.633122682571411,
+ "learning_rate": 8.202040410409957e-06,
+ "loss": 0.1886,
+ "step": 20949
+ },
+ {
+ "epoch": 57.084468664850135,
+ "grad_norm": 8.102224349975586,
+ "learning_rate": 8.201172303974907e-06,
+ "loss": 0.2038,
+ "step": 20950
+ },
+ {
+ "epoch": 57.08719346049046,
+ "grad_norm": 3.5224497318267822,
+ "learning_rate": 8.200304211549066e-06,
+ "loss": 0.0988,
+ "step": 20951
+ },
+ {
+ "epoch": 57.08991825613079,
+ "grad_norm": 3.5284407138824463,
+ "learning_rate": 8.199436133139202e-06,
+ "loss": 0.1167,
+ "step": 20952
+ },
+ {
+ "epoch": 57.09264305177112,
+ "grad_norm": 3.777663469314575,
+ "learning_rate": 8.198568068752069e-06,
+ "loss": 0.1869,
+ "step": 20953
+ },
+ {
+ "epoch": 57.095367847411445,
+ "grad_norm": 4.036108016967773,
+ "learning_rate": 8.19770001839443e-06,
+ "loss": 0.1278,
+ "step": 20954
+ },
+ {
+ "epoch": 57.09809264305177,
+ "grad_norm": 4.82944917678833,
+ "learning_rate": 8.196831982073044e-06,
+ "loss": 0.1084,
+ "step": 20955
+ },
+ {
+ "epoch": 57.1008174386921,
+ "grad_norm": 4.445448398590088,
+ "learning_rate": 8.195963959794675e-06,
+ "loss": 0.0753,
+ "step": 20956
+ },
+ {
+ "epoch": 57.10354223433242,
+ "grad_norm": 3.768475294113159,
+ "learning_rate": 8.19509595156608e-06,
+ "loss": 0.0671,
+ "step": 20957
+ },
+ {
+ "epoch": 57.10626702997275,
+ "grad_norm": 4.0944085121154785,
+ "learning_rate": 8.19422795739402e-06,
+ "loss": 0.1149,
+ "step": 20958
+ },
+ {
+ "epoch": 57.10899182561308,
+ "grad_norm": 3.297511100769043,
+ "learning_rate": 8.193359977285252e-06,
+ "loss": 0.0627,
+ "step": 20959
+ },
+ {
+ "epoch": 57.11171662125341,
+ "grad_norm": 4.805819511413574,
+ "learning_rate": 8.19249201124654e-06,
+ "loss": 0.137,
+ "step": 20960
+ },
+ {
+ "epoch": 57.11444141689373,
+ "grad_norm": 3.7722373008728027,
+ "learning_rate": 8.191624059284643e-06,
+ "loss": 0.1912,
+ "step": 20961
+ },
+ {
+ "epoch": 57.11716621253406,
+ "grad_norm": 2.411750078201294,
+ "learning_rate": 8.190756121406317e-06,
+ "loss": 0.0407,
+ "step": 20962
+ },
+ {
+ "epoch": 57.119891008174385,
+ "grad_norm": 3.3380675315856934,
+ "learning_rate": 8.189888197618326e-06,
+ "loss": 0.0731,
+ "step": 20963
+ },
+ {
+ "epoch": 57.12261580381471,
+ "grad_norm": 2.6617536544799805,
+ "learning_rate": 8.189020287927428e-06,
+ "loss": 0.0578,
+ "step": 20964
+ },
+ {
+ "epoch": 57.12534059945504,
+ "grad_norm": 4.367624282836914,
+ "learning_rate": 8.18815239234038e-06,
+ "loss": 0.064,
+ "step": 20965
+ },
+ {
+ "epoch": 57.12806539509537,
+ "grad_norm": 3.9145865440368652,
+ "learning_rate": 8.187284510863943e-06,
+ "loss": 0.1527,
+ "step": 20966
+ },
+ {
+ "epoch": 57.130790190735695,
+ "grad_norm": 3.0844852924346924,
+ "learning_rate": 8.186416643504874e-06,
+ "loss": 0.1276,
+ "step": 20967
+ },
+ {
+ "epoch": 57.13351498637602,
+ "grad_norm": 2.8756299018859863,
+ "learning_rate": 8.185548790269937e-06,
+ "loss": 0.069,
+ "step": 20968
+ },
+ {
+ "epoch": 57.13623978201635,
+ "grad_norm": 3.0673000812530518,
+ "learning_rate": 8.184680951165887e-06,
+ "loss": 0.0698,
+ "step": 20969
+ },
+ {
+ "epoch": 57.13896457765667,
+ "grad_norm": 4.796577453613281,
+ "learning_rate": 8.183813126199485e-06,
+ "loss": 0.077,
+ "step": 20970
+ },
+ {
+ "epoch": 57.141689373297005,
+ "grad_norm": 9.892295837402344,
+ "learning_rate": 8.182945315377484e-06,
+ "loss": 0.1176,
+ "step": 20971
+ },
+ {
+ "epoch": 57.14441416893733,
+ "grad_norm": 2.8652665615081787,
+ "learning_rate": 8.18207751870665e-06,
+ "loss": 0.097,
+ "step": 20972
+ },
+ {
+ "epoch": 57.14713896457766,
+ "grad_norm": 4.523515701293945,
+ "learning_rate": 8.181209736193736e-06,
+ "loss": 0.0952,
+ "step": 20973
+ },
+ {
+ "epoch": 57.14986376021798,
+ "grad_norm": 3.974118232727051,
+ "learning_rate": 8.180341967845505e-06,
+ "loss": 0.0844,
+ "step": 20974
+ },
+ {
+ "epoch": 57.15258855585831,
+ "grad_norm": 2.272758960723877,
+ "learning_rate": 8.179474213668709e-06,
+ "loss": 0.0669,
+ "step": 20975
+ },
+ {
+ "epoch": 57.155313351498634,
+ "grad_norm": 3.232013702392578,
+ "learning_rate": 8.17860647367011e-06,
+ "loss": 0.1139,
+ "step": 20976
+ },
+ {
+ "epoch": 57.15803814713897,
+ "grad_norm": 3.8660507202148438,
+ "learning_rate": 8.177738747856464e-06,
+ "loss": 0.0621,
+ "step": 20977
+ },
+ {
+ "epoch": 57.16076294277929,
+ "grad_norm": 3.3339223861694336,
+ "learning_rate": 8.176871036234533e-06,
+ "loss": 0.0678,
+ "step": 20978
+ },
+ {
+ "epoch": 57.16348773841962,
+ "grad_norm": 4.419252872467041,
+ "learning_rate": 8.176003338811069e-06,
+ "loss": 0.1196,
+ "step": 20979
+ },
+ {
+ "epoch": 57.166212534059945,
+ "grad_norm": 3.663262367248535,
+ "learning_rate": 8.175135655592833e-06,
+ "loss": 0.0966,
+ "step": 20980
+ },
+ {
+ "epoch": 57.16893732970027,
+ "grad_norm": 2.4827892780303955,
+ "learning_rate": 8.17426798658658e-06,
+ "loss": 0.0385,
+ "step": 20981
+ },
+ {
+ "epoch": 57.171662125340596,
+ "grad_norm": 3.0613882541656494,
+ "learning_rate": 8.173400331799072e-06,
+ "loss": 0.048,
+ "step": 20982
+ },
+ {
+ "epoch": 57.17438692098093,
+ "grad_norm": 4.107855319976807,
+ "learning_rate": 8.172532691237058e-06,
+ "loss": 0.0791,
+ "step": 20983
+ },
+ {
+ "epoch": 57.177111716621255,
+ "grad_norm": 3.435267925262451,
+ "learning_rate": 8.171665064907306e-06,
+ "loss": 0.0826,
+ "step": 20984
+ },
+ {
+ "epoch": 57.17983651226158,
+ "grad_norm": 5.114736557006836,
+ "learning_rate": 8.170797452816566e-06,
+ "loss": 0.0831,
+ "step": 20985
+ },
+ {
+ "epoch": 57.182561307901906,
+ "grad_norm": 3.398550271987915,
+ "learning_rate": 8.169929854971598e-06,
+ "loss": 0.1434,
+ "step": 20986
+ },
+ {
+ "epoch": 57.18528610354223,
+ "grad_norm": 3.0135645866394043,
+ "learning_rate": 8.169062271379155e-06,
+ "loss": 0.1713,
+ "step": 20987
+ },
+ {
+ "epoch": 57.18801089918256,
+ "grad_norm": 2.5155866146087646,
+ "learning_rate": 8.168194702046e-06,
+ "loss": 0.0484,
+ "step": 20988
+ },
+ {
+ "epoch": 57.19073569482289,
+ "grad_norm": 2.8594396114349365,
+ "learning_rate": 8.167327146978883e-06,
+ "loss": 0.0612,
+ "step": 20989
+ },
+ {
+ "epoch": 57.19346049046322,
+ "grad_norm": 2.7956182956695557,
+ "learning_rate": 8.166459606184565e-06,
+ "loss": 0.0356,
+ "step": 20990
+ },
+ {
+ "epoch": 57.19618528610354,
+ "grad_norm": 2.2375898361206055,
+ "learning_rate": 8.165592079669798e-06,
+ "loss": 0.0517,
+ "step": 20991
+ },
+ {
+ "epoch": 57.19891008174387,
+ "grad_norm": 4.1796345710754395,
+ "learning_rate": 8.164724567441343e-06,
+ "loss": 0.2113,
+ "step": 20992
+ },
+ {
+ "epoch": 57.201634877384194,
+ "grad_norm": 3.491035223007202,
+ "learning_rate": 8.163857069505952e-06,
+ "loss": 0.1024,
+ "step": 20993
+ },
+ {
+ "epoch": 57.20435967302452,
+ "grad_norm": 3.8526387214660645,
+ "learning_rate": 8.162989585870385e-06,
+ "loss": 0.106,
+ "step": 20994
+ },
+ {
+ "epoch": 57.20708446866485,
+ "grad_norm": 2.6737542152404785,
+ "learning_rate": 8.162122116541394e-06,
+ "loss": 0.0773,
+ "step": 20995
+ },
+ {
+ "epoch": 57.20980926430518,
+ "grad_norm": 3.965137481689453,
+ "learning_rate": 8.161254661525739e-06,
+ "loss": 0.069,
+ "step": 20996
+ },
+ {
+ "epoch": 57.212534059945504,
+ "grad_norm": 3.304584503173828,
+ "learning_rate": 8.160387220830172e-06,
+ "loss": 0.0637,
+ "step": 20997
+ },
+ {
+ "epoch": 57.21525885558583,
+ "grad_norm": 3.790163040161133,
+ "learning_rate": 8.159519794461452e-06,
+ "loss": 0.0454,
+ "step": 20998
+ },
+ {
+ "epoch": 57.217983651226156,
+ "grad_norm": 1.858001470565796,
+ "learning_rate": 8.158652382426328e-06,
+ "loss": 0.1115,
+ "step": 20999
+ },
+ {
+ "epoch": 57.22070844686648,
+ "grad_norm": 3.8037219047546387,
+ "learning_rate": 8.157784984731561e-06,
+ "loss": 0.1108,
+ "step": 21000
+ },
+ {
+ "epoch": 57.223433242506815,
+ "grad_norm": 3.0992517471313477,
+ "learning_rate": 8.156917601383909e-06,
+ "loss": 0.1667,
+ "step": 21001
+ },
+ {
+ "epoch": 57.22615803814714,
+ "grad_norm": 2.488919496536255,
+ "learning_rate": 8.156050232390121e-06,
+ "loss": 0.0742,
+ "step": 21002
+ },
+ {
+ "epoch": 57.228882833787466,
+ "grad_norm": 2.594763994216919,
+ "learning_rate": 8.155182877756953e-06,
+ "loss": 0.1312,
+ "step": 21003
+ },
+ {
+ "epoch": 57.23160762942779,
+ "grad_norm": 4.037342071533203,
+ "learning_rate": 8.154315537491166e-06,
+ "loss": 0.0912,
+ "step": 21004
+ },
+ {
+ "epoch": 57.23433242506812,
+ "grad_norm": 6.0710577964782715,
+ "learning_rate": 8.153448211599506e-06,
+ "loss": 0.0911,
+ "step": 21005
+ },
+ {
+ "epoch": 57.237057220708444,
+ "grad_norm": 4.658319473266602,
+ "learning_rate": 8.152580900088735e-06,
+ "loss": 0.0601,
+ "step": 21006
+ },
+ {
+ "epoch": 57.23978201634878,
+ "grad_norm": 2.7739837169647217,
+ "learning_rate": 8.151713602965602e-06,
+ "loss": 0.0699,
+ "step": 21007
+ },
+ {
+ "epoch": 57.2425068119891,
+ "grad_norm": 3.6633126735687256,
+ "learning_rate": 8.150846320236866e-06,
+ "loss": 0.1395,
+ "step": 21008
+ },
+ {
+ "epoch": 57.24523160762943,
+ "grad_norm": 3.9689273834228516,
+ "learning_rate": 8.149979051909278e-06,
+ "loss": 0.0465,
+ "step": 21009
+ },
+ {
+ "epoch": 57.247956403269754,
+ "grad_norm": 2.4107301235198975,
+ "learning_rate": 8.149111797989595e-06,
+ "loss": 0.0736,
+ "step": 21010
+ },
+ {
+ "epoch": 57.25068119891008,
+ "grad_norm": 3.059943675994873,
+ "learning_rate": 8.148244558484568e-06,
+ "loss": 0.1071,
+ "step": 21011
+ },
+ {
+ "epoch": 57.253405994550405,
+ "grad_norm": 2.2856011390686035,
+ "learning_rate": 8.147377333400955e-06,
+ "loss": 0.0454,
+ "step": 21012
+ },
+ {
+ "epoch": 57.25613079019074,
+ "grad_norm": 1.885221242904663,
+ "learning_rate": 8.146510122745504e-06,
+ "loss": 0.1229,
+ "step": 21013
+ },
+ {
+ "epoch": 57.258855585831064,
+ "grad_norm": 3.339482545852661,
+ "learning_rate": 8.145642926524977e-06,
+ "loss": 0.054,
+ "step": 21014
+ },
+ {
+ "epoch": 57.26158038147139,
+ "grad_norm": 2.562401294708252,
+ "learning_rate": 8.14477574474612e-06,
+ "loss": 0.0347,
+ "step": 21015
+ },
+ {
+ "epoch": 57.264305177111716,
+ "grad_norm": 1.935423493385315,
+ "learning_rate": 8.143908577415691e-06,
+ "loss": 0.0391,
+ "step": 21016
+ },
+ {
+ "epoch": 57.26702997275204,
+ "grad_norm": 2.822117328643799,
+ "learning_rate": 8.14304142454044e-06,
+ "loss": 0.0842,
+ "step": 21017
+ },
+ {
+ "epoch": 57.26975476839237,
+ "grad_norm": 2.9311492443084717,
+ "learning_rate": 8.142174286127123e-06,
+ "loss": 0.0496,
+ "step": 21018
+ },
+ {
+ "epoch": 57.2724795640327,
+ "grad_norm": 2.203108072280884,
+ "learning_rate": 8.141307162182496e-06,
+ "loss": 0.0449,
+ "step": 21019
+ },
+ {
+ "epoch": 57.275204359673026,
+ "grad_norm": 3.8627078533172607,
+ "learning_rate": 8.140440052713307e-06,
+ "loss": 0.1755,
+ "step": 21020
+ },
+ {
+ "epoch": 57.27792915531335,
+ "grad_norm": 2.968916654586792,
+ "learning_rate": 8.139572957726312e-06,
+ "loss": 0.0591,
+ "step": 21021
+ },
+ {
+ "epoch": 57.28065395095368,
+ "grad_norm": 2.618196725845337,
+ "learning_rate": 8.138705877228262e-06,
+ "loss": 0.164,
+ "step": 21022
+ },
+ {
+ "epoch": 57.283378746594,
+ "grad_norm": 3.0670242309570312,
+ "learning_rate": 8.137838811225913e-06,
+ "loss": 0.0469,
+ "step": 21023
+ },
+ {
+ "epoch": 57.28610354223433,
+ "grad_norm": 2.843203544616699,
+ "learning_rate": 8.136971759726011e-06,
+ "loss": 0.0808,
+ "step": 21024
+ },
+ {
+ "epoch": 57.28882833787466,
+ "grad_norm": 2.5789895057678223,
+ "learning_rate": 8.136104722735317e-06,
+ "loss": 0.0367,
+ "step": 21025
+ },
+ {
+ "epoch": 57.29155313351499,
+ "grad_norm": 2.258441209793091,
+ "learning_rate": 8.135237700260578e-06,
+ "loss": 0.041,
+ "step": 21026
+ },
+ {
+ "epoch": 57.294277929155314,
+ "grad_norm": 2.9373462200164795,
+ "learning_rate": 8.134370692308547e-06,
+ "loss": 0.0757,
+ "step": 21027
+ },
+ {
+ "epoch": 57.29700272479564,
+ "grad_norm": 2.634385824203491,
+ "learning_rate": 8.133503698885978e-06,
+ "loss": 0.1688,
+ "step": 21028
+ },
+ {
+ "epoch": 57.299727520435965,
+ "grad_norm": 3.710176467895508,
+ "learning_rate": 8.132636719999622e-06,
+ "loss": 0.0483,
+ "step": 21029
+ },
+ {
+ "epoch": 57.30245231607629,
+ "grad_norm": 29.706933975219727,
+ "learning_rate": 8.131769755656231e-06,
+ "loss": 0.047,
+ "step": 21030
+ },
+ {
+ "epoch": 57.305177111716624,
+ "grad_norm": 4.706148147583008,
+ "learning_rate": 8.130902805862556e-06,
+ "loss": 0.0942,
+ "step": 21031
+ },
+ {
+ "epoch": 57.30790190735695,
+ "grad_norm": 3.7373077869415283,
+ "learning_rate": 8.130035870625352e-06,
+ "loss": 0.0543,
+ "step": 21032
+ },
+ {
+ "epoch": 57.310626702997276,
+ "grad_norm": 2.8798205852508545,
+ "learning_rate": 8.129168949951368e-06,
+ "loss": 0.1053,
+ "step": 21033
+ },
+ {
+ "epoch": 57.3133514986376,
+ "grad_norm": 1.302497386932373,
+ "learning_rate": 8.128302043847355e-06,
+ "loss": 0.0201,
+ "step": 21034
+ },
+ {
+ "epoch": 57.31607629427793,
+ "grad_norm": 2.107565402984619,
+ "learning_rate": 8.127435152320061e-06,
+ "loss": 0.0289,
+ "step": 21035
+ },
+ {
+ "epoch": 57.31880108991825,
+ "grad_norm": 2.414942979812622,
+ "learning_rate": 8.126568275376247e-06,
+ "loss": 0.0536,
+ "step": 21036
+ },
+ {
+ "epoch": 57.321525885558586,
+ "grad_norm": 8.451323509216309,
+ "learning_rate": 8.125701413022659e-06,
+ "loss": 0.2156,
+ "step": 21037
+ },
+ {
+ "epoch": 57.32425068119891,
+ "grad_norm": 4.350429534912109,
+ "learning_rate": 8.124834565266048e-06,
+ "loss": 0.0474,
+ "step": 21038
+ },
+ {
+ "epoch": 57.32697547683924,
+ "grad_norm": 3.378709316253662,
+ "learning_rate": 8.123967732113165e-06,
+ "loss": 0.1526,
+ "step": 21039
+ },
+ {
+ "epoch": 57.32970027247956,
+ "grad_norm": 1.866207480430603,
+ "learning_rate": 8.12310091357076e-06,
+ "loss": 0.0256,
+ "step": 21040
+ },
+ {
+ "epoch": 57.33242506811989,
+ "grad_norm": 3.3515186309814453,
+ "learning_rate": 8.122234109645586e-06,
+ "loss": 0.0698,
+ "step": 21041
+ },
+ {
+ "epoch": 57.335149863760215,
+ "grad_norm": 7.097870349884033,
+ "learning_rate": 8.12136732034439e-06,
+ "loss": 0.0816,
+ "step": 21042
+ },
+ {
+ "epoch": 57.33787465940055,
+ "grad_norm": 4.234493255615234,
+ "learning_rate": 8.120500545673928e-06,
+ "loss": 0.0502,
+ "step": 21043
+ },
+ {
+ "epoch": 57.34059945504087,
+ "grad_norm": 3.808809518814087,
+ "learning_rate": 8.119633785640944e-06,
+ "loss": 0.1269,
+ "step": 21044
+ },
+ {
+ "epoch": 57.3433242506812,
+ "grad_norm": 2.693981170654297,
+ "learning_rate": 8.118767040252196e-06,
+ "loss": 0.0441,
+ "step": 21045
+ },
+ {
+ "epoch": 57.346049046321525,
+ "grad_norm": 2.78717303276062,
+ "learning_rate": 8.117900309514425e-06,
+ "loss": 0.0306,
+ "step": 21046
+ },
+ {
+ "epoch": 57.34877384196185,
+ "grad_norm": 2.1586101055145264,
+ "learning_rate": 8.117033593434389e-06,
+ "loss": 0.0576,
+ "step": 21047
+ },
+ {
+ "epoch": 57.35149863760218,
+ "grad_norm": 4.34012508392334,
+ "learning_rate": 8.116166892018834e-06,
+ "loss": 0.0661,
+ "step": 21048
+ },
+ {
+ "epoch": 57.35422343324251,
+ "grad_norm": 2.811373710632324,
+ "learning_rate": 8.115300205274508e-06,
+ "loss": 0.0738,
+ "step": 21049
+ },
+ {
+ "epoch": 57.356948228882835,
+ "grad_norm": 3.1375019550323486,
+ "learning_rate": 8.114433533208165e-06,
+ "loss": 0.0653,
+ "step": 21050
+ },
+ {
+ "epoch": 57.35967302452316,
+ "grad_norm": 2.984532594680786,
+ "learning_rate": 8.113566875826553e-06,
+ "loss": 0.1102,
+ "step": 21051
+ },
+ {
+ "epoch": 57.36239782016349,
+ "grad_norm": 3.07688307762146,
+ "learning_rate": 8.11270023313642e-06,
+ "loss": 0.0427,
+ "step": 21052
+ },
+ {
+ "epoch": 57.36512261580381,
+ "grad_norm": 2.0242273807525635,
+ "learning_rate": 8.11183360514452e-06,
+ "loss": 0.0481,
+ "step": 21053
+ },
+ {
+ "epoch": 57.36784741144414,
+ "grad_norm": 3.7911312580108643,
+ "learning_rate": 8.110966991857596e-06,
+ "loss": 0.0797,
+ "step": 21054
+ },
+ {
+ "epoch": 57.37057220708447,
+ "grad_norm": 2.7200639247894287,
+ "learning_rate": 8.110100393282402e-06,
+ "loss": 0.046,
+ "step": 21055
+ },
+ {
+ "epoch": 57.3732970027248,
+ "grad_norm": 3.740577220916748,
+ "learning_rate": 8.109233809425686e-06,
+ "loss": 0.0774,
+ "step": 21056
+ },
+ {
+ "epoch": 57.37602179836512,
+ "grad_norm": 4.832794189453125,
+ "learning_rate": 8.108367240294197e-06,
+ "loss": 0.1561,
+ "step": 21057
+ },
+ {
+ "epoch": 57.37874659400545,
+ "grad_norm": 2.687976121902466,
+ "learning_rate": 8.10750068589468e-06,
+ "loss": 0.0554,
+ "step": 21058
+ },
+ {
+ "epoch": 57.381471389645775,
+ "grad_norm": 2.870914936065674,
+ "learning_rate": 8.10663414623389e-06,
+ "loss": 0.052,
+ "step": 21059
+ },
+ {
+ "epoch": 57.3841961852861,
+ "grad_norm": 4.617345809936523,
+ "learning_rate": 8.10576762131857e-06,
+ "loss": 0.1671,
+ "step": 21060
+ },
+ {
+ "epoch": 57.38692098092643,
+ "grad_norm": 19.893692016601562,
+ "learning_rate": 8.104901111155472e-06,
+ "loss": 0.0427,
+ "step": 21061
+ },
+ {
+ "epoch": 57.38964577656676,
+ "grad_norm": 2.2858848571777344,
+ "learning_rate": 8.104034615751343e-06,
+ "loss": 0.0487,
+ "step": 21062
+ },
+ {
+ "epoch": 57.392370572207085,
+ "grad_norm": 2.523271083831787,
+ "learning_rate": 8.103168135112932e-06,
+ "loss": 0.0786,
+ "step": 21063
+ },
+ {
+ "epoch": 57.39509536784741,
+ "grad_norm": 2.7784969806671143,
+ "learning_rate": 8.102301669246984e-06,
+ "loss": 0.2017,
+ "step": 21064
+ },
+ {
+ "epoch": 57.39782016348774,
+ "grad_norm": 4.028162479400635,
+ "learning_rate": 8.101435218160253e-06,
+ "loss": 0.0475,
+ "step": 21065
+ },
+ {
+ "epoch": 57.40054495912806,
+ "grad_norm": 3.0779473781585693,
+ "learning_rate": 8.10056878185948e-06,
+ "loss": 0.0789,
+ "step": 21066
+ },
+ {
+ "epoch": 57.403269754768395,
+ "grad_norm": 2.278118371963501,
+ "learning_rate": 8.09970236035142e-06,
+ "loss": 0.0333,
+ "step": 21067
+ },
+ {
+ "epoch": 57.40599455040872,
+ "grad_norm": 2.255213737487793,
+ "learning_rate": 8.098835953642813e-06,
+ "loss": 0.0864,
+ "step": 21068
+ },
+ {
+ "epoch": 57.40871934604905,
+ "grad_norm": 3.3764965534210205,
+ "learning_rate": 8.097969561740409e-06,
+ "loss": 0.0536,
+ "step": 21069
+ },
+ {
+ "epoch": 57.41144414168937,
+ "grad_norm": 1.7783499956130981,
+ "learning_rate": 8.097103184650959e-06,
+ "loss": 0.0485,
+ "step": 21070
+ },
+ {
+ "epoch": 57.4141689373297,
+ "grad_norm": 2.192190408706665,
+ "learning_rate": 8.09623682238121e-06,
+ "loss": 0.0465,
+ "step": 21071
+ },
+ {
+ "epoch": 57.416893732970024,
+ "grad_norm": 2.7393951416015625,
+ "learning_rate": 8.095370474937904e-06,
+ "loss": 0.0414,
+ "step": 21072
+ },
+ {
+ "epoch": 57.41961852861036,
+ "grad_norm": 3.780320167541504,
+ "learning_rate": 8.094504142327796e-06,
+ "loss": 0.0907,
+ "step": 21073
+ },
+ {
+ "epoch": 57.42234332425068,
+ "grad_norm": 2.466608762741089,
+ "learning_rate": 8.093637824557625e-06,
+ "loss": 0.0514,
+ "step": 21074
+ },
+ {
+ "epoch": 57.42506811989101,
+ "grad_norm": 2.41591215133667,
+ "learning_rate": 8.092771521634143e-06,
+ "loss": 0.046,
+ "step": 21075
+ },
+ {
+ "epoch": 57.427792915531334,
+ "grad_norm": 3.161233425140381,
+ "learning_rate": 8.091905233564095e-06,
+ "loss": 0.0382,
+ "step": 21076
+ },
+ {
+ "epoch": 57.43051771117166,
+ "grad_norm": 2.009263277053833,
+ "learning_rate": 8.091038960354227e-06,
+ "loss": 0.0335,
+ "step": 21077
+ },
+ {
+ "epoch": 57.433242506811986,
+ "grad_norm": 2.109229564666748,
+ "learning_rate": 8.090172702011287e-06,
+ "loss": 0.1361,
+ "step": 21078
+ },
+ {
+ "epoch": 57.43596730245232,
+ "grad_norm": 2.6371829509735107,
+ "learning_rate": 8.08930645854202e-06,
+ "loss": 0.0717,
+ "step": 21079
+ },
+ {
+ "epoch": 57.438692098092645,
+ "grad_norm": 6.136563301086426,
+ "learning_rate": 8.088440229953172e-06,
+ "loss": 0.0566,
+ "step": 21080
+ },
+ {
+ "epoch": 57.44141689373297,
+ "grad_norm": 3.1998116970062256,
+ "learning_rate": 8.087574016251492e-06,
+ "loss": 0.0525,
+ "step": 21081
+ },
+ {
+ "epoch": 57.444141689373296,
+ "grad_norm": 3.095475673675537,
+ "learning_rate": 8.086707817443722e-06,
+ "loss": 0.0433,
+ "step": 21082
+ },
+ {
+ "epoch": 57.44686648501362,
+ "grad_norm": 2.9920122623443604,
+ "learning_rate": 8.085841633536611e-06,
+ "loss": 0.0371,
+ "step": 21083
+ },
+ {
+ "epoch": 57.44959128065395,
+ "grad_norm": 3.262359380722046,
+ "learning_rate": 8.084975464536902e-06,
+ "loss": 0.0633,
+ "step": 21084
+ },
+ {
+ "epoch": 57.45231607629428,
+ "grad_norm": 2.9780280590057373,
+ "learning_rate": 8.084109310451345e-06,
+ "loss": 0.0605,
+ "step": 21085
+ },
+ {
+ "epoch": 57.45504087193461,
+ "grad_norm": 2.809014081954956,
+ "learning_rate": 8.083243171286679e-06,
+ "loss": 0.3305,
+ "step": 21086
+ },
+ {
+ "epoch": 57.45776566757493,
+ "grad_norm": 1.9704408645629883,
+ "learning_rate": 8.082377047049654e-06,
+ "loss": 0.0654,
+ "step": 21087
+ },
+ {
+ "epoch": 57.46049046321526,
+ "grad_norm": 3.0478789806365967,
+ "learning_rate": 8.08151093774702e-06,
+ "loss": 0.0758,
+ "step": 21088
+ },
+ {
+ "epoch": 57.463215258855584,
+ "grad_norm": 2.7477214336395264,
+ "learning_rate": 8.080644843385512e-06,
+ "loss": 0.1091,
+ "step": 21089
+ },
+ {
+ "epoch": 57.46594005449591,
+ "grad_norm": 2.7675974369049072,
+ "learning_rate": 8.079778763971882e-06,
+ "loss": 0.0643,
+ "step": 21090
+ },
+ {
+ "epoch": 57.46866485013624,
+ "grad_norm": 3.2738137245178223,
+ "learning_rate": 8.078912699512874e-06,
+ "loss": 0.0941,
+ "step": 21091
+ },
+ {
+ "epoch": 57.47138964577657,
+ "grad_norm": 2.3208718299865723,
+ "learning_rate": 8.07804665001523e-06,
+ "loss": 0.0519,
+ "step": 21092
+ },
+ {
+ "epoch": 57.474114441416894,
+ "grad_norm": 6.2093658447265625,
+ "learning_rate": 8.0771806154857e-06,
+ "loss": 0.1431,
+ "step": 21093
+ },
+ {
+ "epoch": 57.47683923705722,
+ "grad_norm": 1.796402931213379,
+ "learning_rate": 8.076314595931024e-06,
+ "loss": 0.0204,
+ "step": 21094
+ },
+ {
+ "epoch": 57.479564032697546,
+ "grad_norm": 2.3748831748962402,
+ "learning_rate": 8.075448591357948e-06,
+ "loss": 0.0286,
+ "step": 21095
+ },
+ {
+ "epoch": 57.48228882833787,
+ "grad_norm": 2.757874011993408,
+ "learning_rate": 8.074582601773215e-06,
+ "loss": 0.0479,
+ "step": 21096
+ },
+ {
+ "epoch": 57.485013623978205,
+ "grad_norm": 3.800487518310547,
+ "learning_rate": 8.073716627183575e-06,
+ "loss": 0.0379,
+ "step": 21097
+ },
+ {
+ "epoch": 57.48773841961853,
+ "grad_norm": 3.4152936935424805,
+ "learning_rate": 8.072850667595763e-06,
+ "loss": 0.0591,
+ "step": 21098
+ },
+ {
+ "epoch": 57.490463215258856,
+ "grad_norm": 4.744851589202881,
+ "learning_rate": 8.071984723016532e-06,
+ "loss": 0.0281,
+ "step": 21099
+ },
+ {
+ "epoch": 57.49318801089918,
+ "grad_norm": 1.9423454999923706,
+ "learning_rate": 8.071118793452618e-06,
+ "loss": 0.0437,
+ "step": 21100
+ },
+ {
+ "epoch": 57.49591280653951,
+ "grad_norm": 2.6029934883117676,
+ "learning_rate": 8.070252878910771e-06,
+ "loss": 0.0387,
+ "step": 21101
+ },
+ {
+ "epoch": 57.49863760217983,
+ "grad_norm": 2.0885252952575684,
+ "learning_rate": 8.069386979397732e-06,
+ "loss": 0.0524,
+ "step": 21102
+ },
+ {
+ "epoch": 57.50136239782017,
+ "grad_norm": 2.687270402908325,
+ "learning_rate": 8.068521094920242e-06,
+ "loss": 0.0461,
+ "step": 21103
+ },
+ {
+ "epoch": 57.50408719346049,
+ "grad_norm": 2.8992786407470703,
+ "learning_rate": 8.067655225485052e-06,
+ "loss": 0.0721,
+ "step": 21104
+ },
+ {
+ "epoch": 57.50681198910082,
+ "grad_norm": 3.569467067718506,
+ "learning_rate": 8.066789371098899e-06,
+ "loss": 0.0918,
+ "step": 21105
+ },
+ {
+ "epoch": 57.509536784741144,
+ "grad_norm": 2.298762321472168,
+ "learning_rate": 8.065923531768529e-06,
+ "loss": 0.0485,
+ "step": 21106
+ },
+ {
+ "epoch": 57.51226158038147,
+ "grad_norm": 2.4000496864318848,
+ "learning_rate": 8.065057707500684e-06,
+ "loss": 0.06,
+ "step": 21107
+ },
+ {
+ "epoch": 57.514986376021795,
+ "grad_norm": 2.7315635681152344,
+ "learning_rate": 8.064191898302109e-06,
+ "loss": 0.1452,
+ "step": 21108
+ },
+ {
+ "epoch": 57.51771117166213,
+ "grad_norm": 2.8182148933410645,
+ "learning_rate": 8.063326104179542e-06,
+ "loss": 0.041,
+ "step": 21109
+ },
+ {
+ "epoch": 57.520435967302454,
+ "grad_norm": 2.1455438137054443,
+ "learning_rate": 8.062460325139731e-06,
+ "loss": 0.0351,
+ "step": 21110
+ },
+ {
+ "epoch": 57.52316076294278,
+ "grad_norm": 11.29668140411377,
+ "learning_rate": 8.061594561189417e-06,
+ "loss": 0.0724,
+ "step": 21111
+ },
+ {
+ "epoch": 57.525885558583106,
+ "grad_norm": 2.7010738849639893,
+ "learning_rate": 8.06072881233534e-06,
+ "loss": 0.0992,
+ "step": 21112
+ },
+ {
+ "epoch": 57.52861035422343,
+ "grad_norm": 2.8165152072906494,
+ "learning_rate": 8.059863078584245e-06,
+ "loss": 0.0761,
+ "step": 21113
+ },
+ {
+ "epoch": 57.53133514986376,
+ "grad_norm": 3.114109992980957,
+ "learning_rate": 8.058997359942875e-06,
+ "loss": 0.063,
+ "step": 21114
+ },
+ {
+ "epoch": 57.53405994550409,
+ "grad_norm": 2.0979883670806885,
+ "learning_rate": 8.058131656417972e-06,
+ "loss": 0.0348,
+ "step": 21115
+ },
+ {
+ "epoch": 57.536784741144416,
+ "grad_norm": 4.14935827255249,
+ "learning_rate": 8.057265968016275e-06,
+ "loss": 0.0646,
+ "step": 21116
+ },
+ {
+ "epoch": 57.53950953678474,
+ "grad_norm": 4.610604286193848,
+ "learning_rate": 8.056400294744531e-06,
+ "loss": 0.0424,
+ "step": 21117
+ },
+ {
+ "epoch": 57.54223433242507,
+ "grad_norm": 3.0349299907684326,
+ "learning_rate": 8.055534636609476e-06,
+ "loss": 0.0344,
+ "step": 21118
+ },
+ {
+ "epoch": 57.54495912806539,
+ "grad_norm": 2.8489203453063965,
+ "learning_rate": 8.054668993617857e-06,
+ "loss": 0.0491,
+ "step": 21119
+ },
+ {
+ "epoch": 57.54768392370572,
+ "grad_norm": 2.6979176998138428,
+ "learning_rate": 8.053803365776412e-06,
+ "loss": 0.0537,
+ "step": 21120
+ },
+ {
+ "epoch": 57.55040871934605,
+ "grad_norm": 2.5363662242889404,
+ "learning_rate": 8.052937753091879e-06,
+ "loss": 0.184,
+ "step": 21121
+ },
+ {
+ "epoch": 57.55313351498638,
+ "grad_norm": 2.7442116737365723,
+ "learning_rate": 8.05207215557101e-06,
+ "loss": 0.0572,
+ "step": 21122
+ },
+ {
+ "epoch": 57.555858310626704,
+ "grad_norm": 2.551673412322998,
+ "learning_rate": 8.05120657322054e-06,
+ "loss": 0.1478,
+ "step": 21123
+ },
+ {
+ "epoch": 57.55858310626703,
+ "grad_norm": 3.6128060817718506,
+ "learning_rate": 8.050341006047209e-06,
+ "loss": 0.1087,
+ "step": 21124
+ },
+ {
+ "epoch": 57.561307901907355,
+ "grad_norm": 2.256990432739258,
+ "learning_rate": 8.04947545405776e-06,
+ "loss": 0.0435,
+ "step": 21125
+ },
+ {
+ "epoch": 57.56403269754768,
+ "grad_norm": 2.383962392807007,
+ "learning_rate": 8.048609917258935e-06,
+ "loss": 0.0639,
+ "step": 21126
+ },
+ {
+ "epoch": 57.566757493188014,
+ "grad_norm": 2.161757469177246,
+ "learning_rate": 8.047744395657471e-06,
+ "loss": 0.0413,
+ "step": 21127
+ },
+ {
+ "epoch": 57.56948228882834,
+ "grad_norm": 1.698662281036377,
+ "learning_rate": 8.046878889260112e-06,
+ "loss": 0.043,
+ "step": 21128
+ },
+ {
+ "epoch": 57.572207084468666,
+ "grad_norm": 2.2235660552978516,
+ "learning_rate": 8.046013398073597e-06,
+ "loss": 0.1673,
+ "step": 21129
+ },
+ {
+ "epoch": 57.57493188010899,
+ "grad_norm": 2.5285484790802,
+ "learning_rate": 8.045147922104668e-06,
+ "loss": 0.0466,
+ "step": 21130
+ },
+ {
+ "epoch": 57.57765667574932,
+ "grad_norm": 2.3021700382232666,
+ "learning_rate": 8.044282461360062e-06,
+ "loss": 0.4543,
+ "step": 21131
+ },
+ {
+ "epoch": 57.58038147138964,
+ "grad_norm": 3.094179153442383,
+ "learning_rate": 8.043417015846523e-06,
+ "loss": 0.0743,
+ "step": 21132
+ },
+ {
+ "epoch": 57.583106267029976,
+ "grad_norm": 2.6132521629333496,
+ "learning_rate": 8.042551585570787e-06,
+ "loss": 0.0504,
+ "step": 21133
+ },
+ {
+ "epoch": 57.5858310626703,
+ "grad_norm": 2.403550863265991,
+ "learning_rate": 8.041686170539599e-06,
+ "loss": 0.0495,
+ "step": 21134
+ },
+ {
+ "epoch": 57.58855585831063,
+ "grad_norm": 2.828885078430176,
+ "learning_rate": 8.040820770759694e-06,
+ "loss": 0.0408,
+ "step": 21135
+ },
+ {
+ "epoch": 57.59128065395095,
+ "grad_norm": 4.264162540435791,
+ "learning_rate": 8.039955386237813e-06,
+ "loss": 0.0299,
+ "step": 21136
+ },
+ {
+ "epoch": 57.59400544959128,
+ "grad_norm": 2.240438461303711,
+ "learning_rate": 8.0390900169807e-06,
+ "loss": 0.0442,
+ "step": 21137
+ },
+ {
+ "epoch": 57.596730245231605,
+ "grad_norm": 2.6937811374664307,
+ "learning_rate": 8.038224662995084e-06,
+ "loss": 0.0675,
+ "step": 21138
+ },
+ {
+ "epoch": 57.59945504087194,
+ "grad_norm": 2.558171510696411,
+ "learning_rate": 8.037359324287716e-06,
+ "loss": 0.0649,
+ "step": 21139
+ },
+ {
+ "epoch": 57.60217983651226,
+ "grad_norm": 3.2288477420806885,
+ "learning_rate": 8.036494000865331e-06,
+ "loss": 0.0949,
+ "step": 21140
+ },
+ {
+ "epoch": 57.60490463215259,
+ "grad_norm": 2.0205624103546143,
+ "learning_rate": 8.035628692734666e-06,
+ "loss": 0.0347,
+ "step": 21141
+ },
+ {
+ "epoch": 57.607629427792915,
+ "grad_norm": 2.4247467517852783,
+ "learning_rate": 8.034763399902462e-06,
+ "loss": 0.1811,
+ "step": 21142
+ },
+ {
+ "epoch": 57.61035422343324,
+ "grad_norm": 2.249375104904175,
+ "learning_rate": 8.033898122375458e-06,
+ "loss": 0.0305,
+ "step": 21143
+ },
+ {
+ "epoch": 57.61307901907357,
+ "grad_norm": 7.736526012420654,
+ "learning_rate": 8.033032860160394e-06,
+ "loss": 0.0341,
+ "step": 21144
+ },
+ {
+ "epoch": 57.6158038147139,
+ "grad_norm": 2.5673105716705322,
+ "learning_rate": 8.032167613264002e-06,
+ "loss": 0.0405,
+ "step": 21145
+ },
+ {
+ "epoch": 57.618528610354225,
+ "grad_norm": 2.404249906539917,
+ "learning_rate": 8.03130238169303e-06,
+ "loss": 0.1054,
+ "step": 21146
+ },
+ {
+ "epoch": 57.62125340599455,
+ "grad_norm": 3.001824378967285,
+ "learning_rate": 8.030437165454208e-06,
+ "loss": 0.0767,
+ "step": 21147
+ },
+ {
+ "epoch": 57.62397820163488,
+ "grad_norm": 2.7434942722320557,
+ "learning_rate": 8.029571964554282e-06,
+ "loss": 0.0814,
+ "step": 21148
+ },
+ {
+ "epoch": 57.6267029972752,
+ "grad_norm": 3.762064218521118,
+ "learning_rate": 8.028706778999982e-06,
+ "loss": 0.0416,
+ "step": 21149
+ },
+ {
+ "epoch": 57.62942779291553,
+ "grad_norm": 2.0549395084381104,
+ "learning_rate": 8.027841608798053e-06,
+ "loss": 0.053,
+ "step": 21150
+ },
+ {
+ "epoch": 57.63215258855586,
+ "grad_norm": 2.5216264724731445,
+ "learning_rate": 8.026976453955228e-06,
+ "loss": 0.0315,
+ "step": 21151
+ },
+ {
+ "epoch": 57.63487738419619,
+ "grad_norm": 2.152392864227295,
+ "learning_rate": 8.026111314478249e-06,
+ "loss": 0.0266,
+ "step": 21152
+ },
+ {
+ "epoch": 57.63760217983651,
+ "grad_norm": 2.4079606533050537,
+ "learning_rate": 8.02524619037385e-06,
+ "loss": 0.0412,
+ "step": 21153
+ },
+ {
+ "epoch": 57.64032697547684,
+ "grad_norm": 1.9555752277374268,
+ "learning_rate": 8.02438108164877e-06,
+ "loss": 0.0325,
+ "step": 21154
+ },
+ {
+ "epoch": 57.643051771117165,
+ "grad_norm": 1.7784439325332642,
+ "learning_rate": 8.023515988309742e-06,
+ "loss": 0.0474,
+ "step": 21155
+ },
+ {
+ "epoch": 57.64577656675749,
+ "grad_norm": 2.8752400875091553,
+ "learning_rate": 8.022650910363516e-06,
+ "loss": 0.0758,
+ "step": 21156
+ },
+ {
+ "epoch": 57.64850136239782,
+ "grad_norm": 2.49324631690979,
+ "learning_rate": 8.021785847816816e-06,
+ "loss": 0.1229,
+ "step": 21157
+ },
+ {
+ "epoch": 57.65122615803815,
+ "grad_norm": 3.5844919681549072,
+ "learning_rate": 8.020920800676387e-06,
+ "loss": 0.06,
+ "step": 21158
+ },
+ {
+ "epoch": 57.653950953678475,
+ "grad_norm": 2.315361261367798,
+ "learning_rate": 8.020055768948963e-06,
+ "loss": 0.204,
+ "step": 21159
+ },
+ {
+ "epoch": 57.6566757493188,
+ "grad_norm": 2.917799234390259,
+ "learning_rate": 8.01919075264128e-06,
+ "loss": 0.0472,
+ "step": 21160
+ },
+ {
+ "epoch": 57.65940054495913,
+ "grad_norm": 3.0093960762023926,
+ "learning_rate": 8.018325751760075e-06,
+ "loss": 0.142,
+ "step": 21161
+ },
+ {
+ "epoch": 57.66212534059945,
+ "grad_norm": 2.5561156272888184,
+ "learning_rate": 8.017460766312088e-06,
+ "loss": 0.1957,
+ "step": 21162
+ },
+ {
+ "epoch": 57.664850136239785,
+ "grad_norm": 1.813188910484314,
+ "learning_rate": 8.01659579630405e-06,
+ "loss": 0.0233,
+ "step": 21163
+ },
+ {
+ "epoch": 57.66757493188011,
+ "grad_norm": 2.4973225593566895,
+ "learning_rate": 8.015730841742702e-06,
+ "loss": 0.0438,
+ "step": 21164
+ },
+ {
+ "epoch": 57.67029972752044,
+ "grad_norm": 2.826347589492798,
+ "learning_rate": 8.014865902634774e-06,
+ "loss": 0.061,
+ "step": 21165
+ },
+ {
+ "epoch": 57.67302452316076,
+ "grad_norm": 3.343771457672119,
+ "learning_rate": 8.014000978987013e-06,
+ "loss": 0.1699,
+ "step": 21166
+ },
+ {
+ "epoch": 57.67574931880109,
+ "grad_norm": 3.5801970958709717,
+ "learning_rate": 8.013136070806143e-06,
+ "loss": 0.072,
+ "step": 21167
+ },
+ {
+ "epoch": 57.678474114441414,
+ "grad_norm": 4.694897651672363,
+ "learning_rate": 8.01227117809891e-06,
+ "loss": 0.0441,
+ "step": 21168
+ },
+ {
+ "epoch": 57.68119891008175,
+ "grad_norm": 2.623826503753662,
+ "learning_rate": 8.011406300872041e-06,
+ "loss": 0.1275,
+ "step": 21169
+ },
+ {
+ "epoch": 57.68392370572207,
+ "grad_norm": 1.6386638879776,
+ "learning_rate": 8.01054143913228e-06,
+ "loss": 0.0379,
+ "step": 21170
+ },
+ {
+ "epoch": 57.6866485013624,
+ "grad_norm": 1.6905165910720825,
+ "learning_rate": 8.009676592886353e-06,
+ "loss": 0.0581,
+ "step": 21171
+ },
+ {
+ "epoch": 57.689373297002724,
+ "grad_norm": 1.8993396759033203,
+ "learning_rate": 8.008811762141e-06,
+ "loss": 0.0356,
+ "step": 21172
+ },
+ {
+ "epoch": 57.69209809264305,
+ "grad_norm": 3.698244094848633,
+ "learning_rate": 8.00794694690296e-06,
+ "loss": 0.041,
+ "step": 21173
+ },
+ {
+ "epoch": 57.694822888283376,
+ "grad_norm": 2.0565102100372314,
+ "learning_rate": 8.007082147178967e-06,
+ "loss": 0.0365,
+ "step": 21174
+ },
+ {
+ "epoch": 57.69754768392371,
+ "grad_norm": 2.971383571624756,
+ "learning_rate": 8.006217362975752e-06,
+ "loss": 0.0442,
+ "step": 21175
+ },
+ {
+ "epoch": 57.700272479564035,
+ "grad_norm": 4.191947937011719,
+ "learning_rate": 8.005352594300052e-06,
+ "loss": 0.0769,
+ "step": 21176
+ },
+ {
+ "epoch": 57.70299727520436,
+ "grad_norm": 2.7116525173187256,
+ "learning_rate": 8.004487841158601e-06,
+ "loss": 0.2684,
+ "step": 21177
+ },
+ {
+ "epoch": 57.705722070844686,
+ "grad_norm": 3.469273567199707,
+ "learning_rate": 8.003623103558137e-06,
+ "loss": 0.0648,
+ "step": 21178
+ },
+ {
+ "epoch": 57.70844686648501,
+ "grad_norm": 2.712843179702759,
+ "learning_rate": 8.002758381505389e-06,
+ "loss": 0.107,
+ "step": 21179
+ },
+ {
+ "epoch": 57.71117166212534,
+ "grad_norm": 2.7341036796569824,
+ "learning_rate": 8.001893675007098e-06,
+ "loss": 0.0431,
+ "step": 21180
+ },
+ {
+ "epoch": 57.71389645776567,
+ "grad_norm": 2.010369300842285,
+ "learning_rate": 8.001028984069991e-06,
+ "loss": 0.0334,
+ "step": 21181
+ },
+ {
+ "epoch": 57.716621253406,
+ "grad_norm": 3.8369619846343994,
+ "learning_rate": 8.00016430870081e-06,
+ "loss": 0.0408,
+ "step": 21182
+ },
+ {
+ "epoch": 57.71934604904632,
+ "grad_norm": 2.3499181270599365,
+ "learning_rate": 7.999299648906281e-06,
+ "loss": 0.0543,
+ "step": 21183
+ },
+ {
+ "epoch": 57.72207084468665,
+ "grad_norm": 2.2508885860443115,
+ "learning_rate": 7.998435004693144e-06,
+ "loss": 0.034,
+ "step": 21184
+ },
+ {
+ "epoch": 57.724795640326974,
+ "grad_norm": 2.181194305419922,
+ "learning_rate": 7.99757037606813e-06,
+ "loss": 0.1414,
+ "step": 21185
+ },
+ {
+ "epoch": 57.7275204359673,
+ "grad_norm": 2.0579848289489746,
+ "learning_rate": 7.996705763037975e-06,
+ "loss": 0.0274,
+ "step": 21186
+ },
+ {
+ "epoch": 57.73024523160763,
+ "grad_norm": 2.2776975631713867,
+ "learning_rate": 7.99584116560941e-06,
+ "loss": 0.0537,
+ "step": 21187
+ },
+ {
+ "epoch": 57.73297002724796,
+ "grad_norm": 2.273754119873047,
+ "learning_rate": 7.99497658378917e-06,
+ "loss": 0.0521,
+ "step": 21188
+ },
+ {
+ "epoch": 57.735694822888284,
+ "grad_norm": 3.601112127304077,
+ "learning_rate": 7.994112017583984e-06,
+ "loss": 0.0384,
+ "step": 21189
+ },
+ {
+ "epoch": 57.73841961852861,
+ "grad_norm": 2.3234620094299316,
+ "learning_rate": 7.99324746700059e-06,
+ "loss": 0.1602,
+ "step": 21190
+ },
+ {
+ "epoch": 57.741144414168936,
+ "grad_norm": 3.5036566257476807,
+ "learning_rate": 7.992382932045726e-06,
+ "loss": 0.108,
+ "step": 21191
+ },
+ {
+ "epoch": 57.74386920980926,
+ "grad_norm": 1.9434713125228882,
+ "learning_rate": 7.991518412726113e-06,
+ "loss": 0.0379,
+ "step": 21192
+ },
+ {
+ "epoch": 57.746594005449595,
+ "grad_norm": 1.7634479999542236,
+ "learning_rate": 7.990653909048494e-06,
+ "loss": 0.0286,
+ "step": 21193
+ },
+ {
+ "epoch": 57.74931880108992,
+ "grad_norm": 2.150554656982422,
+ "learning_rate": 7.989789421019596e-06,
+ "loss": 0.1107,
+ "step": 21194
+ },
+ {
+ "epoch": 57.752043596730246,
+ "grad_norm": 1.7669358253479004,
+ "learning_rate": 7.988924948646153e-06,
+ "loss": 0.0347,
+ "step": 21195
+ },
+ {
+ "epoch": 57.75476839237057,
+ "grad_norm": 2.6823604106903076,
+ "learning_rate": 7.988060491934899e-06,
+ "loss": 0.0955,
+ "step": 21196
+ },
+ {
+ "epoch": 57.7574931880109,
+ "grad_norm": 2.549980640411377,
+ "learning_rate": 7.987196050892565e-06,
+ "loss": 0.047,
+ "step": 21197
+ },
+ {
+ "epoch": 57.76021798365122,
+ "grad_norm": 1.9706356525421143,
+ "learning_rate": 7.986331625525883e-06,
+ "loss": 0.0231,
+ "step": 21198
+ },
+ {
+ "epoch": 57.762942779291556,
+ "grad_norm": 3.000706672668457,
+ "learning_rate": 7.985467215841584e-06,
+ "loss": 0.0966,
+ "step": 21199
+ },
+ {
+ "epoch": 57.76566757493188,
+ "grad_norm": 2.2449820041656494,
+ "learning_rate": 7.984602821846404e-06,
+ "loss": 0.0557,
+ "step": 21200
+ },
+ {
+ "epoch": 57.76839237057221,
+ "grad_norm": 1.6533360481262207,
+ "learning_rate": 7.983738443547068e-06,
+ "loss": 0.025,
+ "step": 21201
+ },
+ {
+ "epoch": 57.771117166212534,
+ "grad_norm": 1.4314603805541992,
+ "learning_rate": 7.982874080950319e-06,
+ "loss": 0.026,
+ "step": 21202
+ },
+ {
+ "epoch": 57.77384196185286,
+ "grad_norm": 2.3609981536865234,
+ "learning_rate": 7.982009734062876e-06,
+ "loss": 0.158,
+ "step": 21203
+ },
+ {
+ "epoch": 57.776566757493185,
+ "grad_norm": 2.1252825260162354,
+ "learning_rate": 7.98114540289148e-06,
+ "loss": 0.0471,
+ "step": 21204
+ },
+ {
+ "epoch": 57.77929155313352,
+ "grad_norm": 4.789887428283691,
+ "learning_rate": 7.980281087442855e-06,
+ "loss": 0.051,
+ "step": 21205
+ },
+ {
+ "epoch": 57.782016348773844,
+ "grad_norm": 2.8740503787994385,
+ "learning_rate": 7.97941678772374e-06,
+ "loss": 0.0393,
+ "step": 21206
+ },
+ {
+ "epoch": 57.78474114441417,
+ "grad_norm": 2.5027835369110107,
+ "learning_rate": 7.978552503740855e-06,
+ "loss": 0.0961,
+ "step": 21207
+ },
+ {
+ "epoch": 57.787465940054496,
+ "grad_norm": 2.251155376434326,
+ "learning_rate": 7.977688235500942e-06,
+ "loss": 0.055,
+ "step": 21208
+ },
+ {
+ "epoch": 57.79019073569482,
+ "grad_norm": 2.8523597717285156,
+ "learning_rate": 7.97682398301073e-06,
+ "loss": 0.0782,
+ "step": 21209
+ },
+ {
+ "epoch": 57.79291553133515,
+ "grad_norm": 2.721550941467285,
+ "learning_rate": 7.975959746276946e-06,
+ "loss": 0.1463,
+ "step": 21210
+ },
+ {
+ "epoch": 57.79564032697548,
+ "grad_norm": 2.7162561416625977,
+ "learning_rate": 7.975095525306324e-06,
+ "loss": 0.0493,
+ "step": 21211
+ },
+ {
+ "epoch": 57.798365122615806,
+ "grad_norm": 2.0879080295562744,
+ "learning_rate": 7.97423132010559e-06,
+ "loss": 0.0279,
+ "step": 21212
+ },
+ {
+ "epoch": 57.80108991825613,
+ "grad_norm": 2.5339062213897705,
+ "learning_rate": 7.97336713068148e-06,
+ "loss": 0.0511,
+ "step": 21213
+ },
+ {
+ "epoch": 57.80381471389646,
+ "grad_norm": 2.227099895477295,
+ "learning_rate": 7.97250295704072e-06,
+ "loss": 0.2568,
+ "step": 21214
+ },
+ {
+ "epoch": 57.80653950953678,
+ "grad_norm": 4.492751598358154,
+ "learning_rate": 7.971638799190044e-06,
+ "loss": 0.1238,
+ "step": 21215
+ },
+ {
+ "epoch": 57.80926430517711,
+ "grad_norm": 1.7814489603042603,
+ "learning_rate": 7.970774657136177e-06,
+ "loss": 0.1168,
+ "step": 21216
+ },
+ {
+ "epoch": 57.81198910081744,
+ "grad_norm": 1.8419840335845947,
+ "learning_rate": 7.969910530885854e-06,
+ "loss": 0.0666,
+ "step": 21217
+ },
+ {
+ "epoch": 57.81471389645777,
+ "grad_norm": 2.499321937561035,
+ "learning_rate": 7.9690464204458e-06,
+ "loss": 0.0995,
+ "step": 21218
+ },
+ {
+ "epoch": 57.817438692098094,
+ "grad_norm": 2.4542698860168457,
+ "learning_rate": 7.968182325822748e-06,
+ "loss": 0.0791,
+ "step": 21219
+ },
+ {
+ "epoch": 57.82016348773842,
+ "grad_norm": 2.259418249130249,
+ "learning_rate": 7.967318247023427e-06,
+ "loss": 0.0347,
+ "step": 21220
+ },
+ {
+ "epoch": 57.822888283378745,
+ "grad_norm": 1.9838428497314453,
+ "learning_rate": 7.966454184054566e-06,
+ "loss": 0.0297,
+ "step": 21221
+ },
+ {
+ "epoch": 57.82561307901907,
+ "grad_norm": 1.730881929397583,
+ "learning_rate": 7.965590136922894e-06,
+ "loss": 0.0249,
+ "step": 21222
+ },
+ {
+ "epoch": 57.828337874659404,
+ "grad_norm": 4.596040725708008,
+ "learning_rate": 7.96472610563514e-06,
+ "loss": 0.0248,
+ "step": 21223
+ },
+ {
+ "epoch": 57.83106267029973,
+ "grad_norm": 1.9409500360488892,
+ "learning_rate": 7.96386209019803e-06,
+ "loss": 0.0616,
+ "step": 21224
+ },
+ {
+ "epoch": 57.833787465940055,
+ "grad_norm": 2.97624135017395,
+ "learning_rate": 7.962998090618303e-06,
+ "loss": 0.0349,
+ "step": 21225
+ },
+ {
+ "epoch": 57.83651226158038,
+ "grad_norm": 3.21820068359375,
+ "learning_rate": 7.962134106902679e-06,
+ "loss": 0.0909,
+ "step": 21226
+ },
+ {
+ "epoch": 57.83923705722071,
+ "grad_norm": 2.447091579437256,
+ "learning_rate": 7.961270139057891e-06,
+ "loss": 0.0277,
+ "step": 21227
+ },
+ {
+ "epoch": 57.84196185286103,
+ "grad_norm": 2.0796873569488525,
+ "learning_rate": 7.960406187090661e-06,
+ "loss": 0.0818,
+ "step": 21228
+ },
+ {
+ "epoch": 57.844686648501366,
+ "grad_norm": 2.25553822517395,
+ "learning_rate": 7.959542251007726e-06,
+ "loss": 0.0368,
+ "step": 21229
+ },
+ {
+ "epoch": 57.84741144414169,
+ "grad_norm": 2.1942853927612305,
+ "learning_rate": 7.95867833081581e-06,
+ "loss": 0.0564,
+ "step": 21230
+ },
+ {
+ "epoch": 57.85013623978202,
+ "grad_norm": 1.8815428018569946,
+ "learning_rate": 7.95781442652164e-06,
+ "loss": 0.0326,
+ "step": 21231
+ },
+ {
+ "epoch": 57.85286103542234,
+ "grad_norm": 2.289034605026245,
+ "learning_rate": 7.956950538131946e-06,
+ "loss": 0.0299,
+ "step": 21232
+ },
+ {
+ "epoch": 57.85558583106267,
+ "grad_norm": 2.239926815032959,
+ "learning_rate": 7.956086665653456e-06,
+ "loss": 0.0511,
+ "step": 21233
+ },
+ {
+ "epoch": 57.858310626702995,
+ "grad_norm": 2.8251500129699707,
+ "learning_rate": 7.955222809092895e-06,
+ "loss": 0.066,
+ "step": 21234
+ },
+ {
+ "epoch": 57.86103542234333,
+ "grad_norm": 2.212529420852661,
+ "learning_rate": 7.954358968456997e-06,
+ "loss": 0.151,
+ "step": 21235
+ },
+ {
+ "epoch": 57.86376021798365,
+ "grad_norm": 2.2048518657684326,
+ "learning_rate": 7.953495143752481e-06,
+ "loss": 0.0647,
+ "step": 21236
+ },
+ {
+ "epoch": 57.86648501362398,
+ "grad_norm": 2.168747663497925,
+ "learning_rate": 7.952631334986083e-06,
+ "loss": 0.0313,
+ "step": 21237
+ },
+ {
+ "epoch": 57.869209809264305,
+ "grad_norm": 2.179305076599121,
+ "learning_rate": 7.951767542164524e-06,
+ "loss": 0.0709,
+ "step": 21238
+ },
+ {
+ "epoch": 57.87193460490463,
+ "grad_norm": 2.0092501640319824,
+ "learning_rate": 7.950903765294534e-06,
+ "loss": 0.0955,
+ "step": 21239
+ },
+ {
+ "epoch": 57.87465940054496,
+ "grad_norm": 1.8909237384796143,
+ "learning_rate": 7.950040004382838e-06,
+ "loss": 0.027,
+ "step": 21240
+ },
+ {
+ "epoch": 57.87738419618529,
+ "grad_norm": 2.3084309101104736,
+ "learning_rate": 7.949176259436164e-06,
+ "loss": 0.0464,
+ "step": 21241
+ },
+ {
+ "epoch": 57.880108991825615,
+ "grad_norm": 2.8828775882720947,
+ "learning_rate": 7.94831253046124e-06,
+ "loss": 0.0556,
+ "step": 21242
+ },
+ {
+ "epoch": 57.88283378746594,
+ "grad_norm": 1.9351611137390137,
+ "learning_rate": 7.947448817464793e-06,
+ "loss": 0.1596,
+ "step": 21243
+ },
+ {
+ "epoch": 57.88555858310627,
+ "grad_norm": 2.526383638381958,
+ "learning_rate": 7.946585120453549e-06,
+ "loss": 0.0356,
+ "step": 21244
+ },
+ {
+ "epoch": 57.88828337874659,
+ "grad_norm": 2.500676155090332,
+ "learning_rate": 7.945721439434234e-06,
+ "loss": 0.0361,
+ "step": 21245
+ },
+ {
+ "epoch": 57.89100817438692,
+ "grad_norm": 2.1384363174438477,
+ "learning_rate": 7.944857774413573e-06,
+ "loss": 0.0445,
+ "step": 21246
+ },
+ {
+ "epoch": 57.89373297002725,
+ "grad_norm": 2.067753314971924,
+ "learning_rate": 7.943994125398295e-06,
+ "loss": 0.0305,
+ "step": 21247
+ },
+ {
+ "epoch": 57.89645776566758,
+ "grad_norm": 1.9107786417007446,
+ "learning_rate": 7.943130492395123e-06,
+ "loss": 0.0371,
+ "step": 21248
+ },
+ {
+ "epoch": 57.8991825613079,
+ "grad_norm": 2.0884668827056885,
+ "learning_rate": 7.942266875410788e-06,
+ "loss": 0.0672,
+ "step": 21249
+ },
+ {
+ "epoch": 57.90190735694823,
+ "grad_norm": 2.9047083854675293,
+ "learning_rate": 7.941403274452009e-06,
+ "loss": 0.0386,
+ "step": 21250
+ },
+ {
+ "epoch": 57.904632152588555,
+ "grad_norm": 3.002500057220459,
+ "learning_rate": 7.940539689525517e-06,
+ "loss": 0.0399,
+ "step": 21251
+ },
+ {
+ "epoch": 57.90735694822888,
+ "grad_norm": 2.65754771232605,
+ "learning_rate": 7.939676120638033e-06,
+ "loss": 0.0362,
+ "step": 21252
+ },
+ {
+ "epoch": 57.91008174386921,
+ "grad_norm": 4.844908714294434,
+ "learning_rate": 7.938812567796289e-06,
+ "loss": 0.1033,
+ "step": 21253
+ },
+ {
+ "epoch": 57.91280653950954,
+ "grad_norm": 2.8697991371154785,
+ "learning_rate": 7.937949031007001e-06,
+ "loss": 0.1497,
+ "step": 21254
+ },
+ {
+ "epoch": 57.915531335149865,
+ "grad_norm": 2.118797779083252,
+ "learning_rate": 7.937085510276904e-06,
+ "loss": 0.0565,
+ "step": 21255
+ },
+ {
+ "epoch": 57.91825613079019,
+ "grad_norm": 1.7942376136779785,
+ "learning_rate": 7.936222005612717e-06,
+ "loss": 0.0234,
+ "step": 21256
+ },
+ {
+ "epoch": 57.920980926430516,
+ "grad_norm": 2.0363500118255615,
+ "learning_rate": 7.935358517021168e-06,
+ "loss": 0.0555,
+ "step": 21257
+ },
+ {
+ "epoch": 57.92370572207084,
+ "grad_norm": 2.00536847114563,
+ "learning_rate": 7.934495044508976e-06,
+ "loss": 0.046,
+ "step": 21258
+ },
+ {
+ "epoch": 57.926430517711175,
+ "grad_norm": 2.109297275543213,
+ "learning_rate": 7.933631588082872e-06,
+ "loss": 0.0555,
+ "step": 21259
+ },
+ {
+ "epoch": 57.9291553133515,
+ "grad_norm": 2.6770381927490234,
+ "learning_rate": 7.932768147749582e-06,
+ "loss": 0.0346,
+ "step": 21260
+ },
+ {
+ "epoch": 57.93188010899183,
+ "grad_norm": 2.376979351043701,
+ "learning_rate": 7.931904723515825e-06,
+ "loss": 0.1474,
+ "step": 21261
+ },
+ {
+ "epoch": 57.93460490463215,
+ "grad_norm": 2.818664312362671,
+ "learning_rate": 7.931041315388328e-06,
+ "loss": 0.0619,
+ "step": 21262
+ },
+ {
+ "epoch": 57.93732970027248,
+ "grad_norm": 3.4146978855133057,
+ "learning_rate": 7.930177923373814e-06,
+ "loss": 0.0375,
+ "step": 21263
+ },
+ {
+ "epoch": 57.940054495912804,
+ "grad_norm": 2.591921806335449,
+ "learning_rate": 7.929314547479007e-06,
+ "loss": 0.1011,
+ "step": 21264
+ },
+ {
+ "epoch": 57.94277929155314,
+ "grad_norm": 2.28962779045105,
+ "learning_rate": 7.928451187710632e-06,
+ "loss": 0.0306,
+ "step": 21265
+ },
+ {
+ "epoch": 57.94550408719346,
+ "grad_norm": 2.782501697540283,
+ "learning_rate": 7.927587844075413e-06,
+ "loss": 0.1091,
+ "step": 21266
+ },
+ {
+ "epoch": 57.94822888283379,
+ "grad_norm": 3.3190038204193115,
+ "learning_rate": 7.926724516580076e-06,
+ "loss": 0.1015,
+ "step": 21267
+ },
+ {
+ "epoch": 57.950953678474114,
+ "grad_norm": 1.7303284406661987,
+ "learning_rate": 7.925861205231336e-06,
+ "loss": 0.0425,
+ "step": 21268
+ },
+ {
+ "epoch": 57.95367847411444,
+ "grad_norm": 2.241946220397949,
+ "learning_rate": 7.924997910035928e-06,
+ "loss": 0.0356,
+ "step": 21269
+ },
+ {
+ "epoch": 57.956403269754766,
+ "grad_norm": 2.6345841884613037,
+ "learning_rate": 7.924134631000566e-06,
+ "loss": 0.0434,
+ "step": 21270
+ },
+ {
+ "epoch": 57.95912806539509,
+ "grad_norm": 3.0955214500427246,
+ "learning_rate": 7.923271368131977e-06,
+ "loss": 0.0616,
+ "step": 21271
+ },
+ {
+ "epoch": 57.961852861035425,
+ "grad_norm": 2.8268356323242188,
+ "learning_rate": 7.922408121436883e-06,
+ "loss": 0.0396,
+ "step": 21272
+ },
+ {
+ "epoch": 57.96457765667575,
+ "grad_norm": 3.150498867034912,
+ "learning_rate": 7.921544890922011e-06,
+ "loss": 0.0319,
+ "step": 21273
+ },
+ {
+ "epoch": 57.967302452316076,
+ "grad_norm": 2.9290952682495117,
+ "learning_rate": 7.920681676594077e-06,
+ "loss": 0.0719,
+ "step": 21274
+ },
+ {
+ "epoch": 57.9700272479564,
+ "grad_norm": 2.350827693939209,
+ "learning_rate": 7.919818478459805e-06,
+ "loss": 0.166,
+ "step": 21275
+ },
+ {
+ "epoch": 57.97275204359673,
+ "grad_norm": 1.9489392042160034,
+ "learning_rate": 7.918955296525925e-06,
+ "loss": 0.0278,
+ "step": 21276
+ },
+ {
+ "epoch": 57.97547683923706,
+ "grad_norm": 1.6865017414093018,
+ "learning_rate": 7.91809213079915e-06,
+ "loss": 0.0376,
+ "step": 21277
+ },
+ {
+ "epoch": 57.97820163487739,
+ "grad_norm": 2.262932300567627,
+ "learning_rate": 7.917228981286212e-06,
+ "loss": 0.0317,
+ "step": 21278
+ },
+ {
+ "epoch": 57.98092643051771,
+ "grad_norm": 2.439934015274048,
+ "learning_rate": 7.916365847993823e-06,
+ "loss": 0.0653,
+ "step": 21279
+ },
+ {
+ "epoch": 57.98365122615804,
+ "grad_norm": 3.1156649589538574,
+ "learning_rate": 7.915502730928713e-06,
+ "loss": 0.0497,
+ "step": 21280
+ },
+ {
+ "epoch": 57.986376021798364,
+ "grad_norm": 2.4230360984802246,
+ "learning_rate": 7.914639630097598e-06,
+ "loss": 0.0476,
+ "step": 21281
+ },
+ {
+ "epoch": 57.98910081743869,
+ "grad_norm": 3.554121971130371,
+ "learning_rate": 7.913776545507204e-06,
+ "loss": 0.1344,
+ "step": 21282
+ },
+ {
+ "epoch": 57.991825613079016,
+ "grad_norm": 6.642258167266846,
+ "learning_rate": 7.91291347716425e-06,
+ "loss": 0.1086,
+ "step": 21283
+ },
+ {
+ "epoch": 57.99455040871935,
+ "grad_norm": 2.6686851978302,
+ "learning_rate": 7.91205042507546e-06,
+ "loss": 0.028,
+ "step": 21284
+ },
+ {
+ "epoch": 57.997275204359674,
+ "grad_norm": 1.9570413827896118,
+ "learning_rate": 7.911187389247553e-06,
+ "loss": 0.034,
+ "step": 21285
+ },
+ {
+ "epoch": 58.0,
+ "grad_norm": 2.350947856903076,
+ "learning_rate": 7.91032436968725e-06,
+ "loss": 0.0989,
+ "step": 21286
+ },
+ {
+ "epoch": 58.002724795640326,
+ "grad_norm": 1.7172086238861084,
+ "learning_rate": 7.909461366401277e-06,
+ "loss": 0.0297,
+ "step": 21287
+ },
+ {
+ "epoch": 58.00544959128065,
+ "grad_norm": 2.9354546070098877,
+ "learning_rate": 7.908598379396349e-06,
+ "loss": 0.0205,
+ "step": 21288
+ },
+ {
+ "epoch": 58.00817438692098,
+ "grad_norm": 2.7402563095092773,
+ "learning_rate": 7.90773540867919e-06,
+ "loss": 0.0465,
+ "step": 21289
+ },
+ {
+ "epoch": 58.01089918256131,
+ "grad_norm": 1.8259797096252441,
+ "learning_rate": 7.90687245425652e-06,
+ "loss": 0.0333,
+ "step": 21290
+ },
+ {
+ "epoch": 58.013623978201636,
+ "grad_norm": 1.94265878200531,
+ "learning_rate": 7.90600951613506e-06,
+ "loss": 0.0669,
+ "step": 21291
+ },
+ {
+ "epoch": 58.01634877384196,
+ "grad_norm": 2.674187183380127,
+ "learning_rate": 7.905146594321532e-06,
+ "loss": 0.0293,
+ "step": 21292
+ },
+ {
+ "epoch": 58.01907356948229,
+ "grad_norm": 2.4613773822784424,
+ "learning_rate": 7.904283688822651e-06,
+ "loss": 0.0339,
+ "step": 21293
+ },
+ {
+ "epoch": 58.02179836512261,
+ "grad_norm": 1.7437459230422974,
+ "learning_rate": 7.903420799645145e-06,
+ "loss": 0.0337,
+ "step": 21294
+ },
+ {
+ "epoch": 58.02452316076294,
+ "grad_norm": 1.6902283430099487,
+ "learning_rate": 7.90255792679573e-06,
+ "loss": 0.0448,
+ "step": 21295
+ },
+ {
+ "epoch": 58.02724795640327,
+ "grad_norm": 2.4785408973693848,
+ "learning_rate": 7.901695070281126e-06,
+ "loss": 0.0437,
+ "step": 21296
+ },
+ {
+ "epoch": 58.0299727520436,
+ "grad_norm": 1.9778372049331665,
+ "learning_rate": 7.90083223010805e-06,
+ "loss": 0.0426,
+ "step": 21297
+ },
+ {
+ "epoch": 58.032697547683924,
+ "grad_norm": 2.2993924617767334,
+ "learning_rate": 7.899969406283232e-06,
+ "loss": 0.1689,
+ "step": 21298
+ },
+ {
+ "epoch": 58.03542234332425,
+ "grad_norm": 2.012148857116699,
+ "learning_rate": 7.89910659881338e-06,
+ "loss": 0.1302,
+ "step": 21299
+ },
+ {
+ "epoch": 58.038147138964575,
+ "grad_norm": 1.6363649368286133,
+ "learning_rate": 7.898243807705219e-06,
+ "loss": 0.0358,
+ "step": 21300
+ },
+ {
+ "epoch": 58.0408719346049,
+ "grad_norm": 1.4601547718048096,
+ "learning_rate": 7.897381032965467e-06,
+ "loss": 0.0264,
+ "step": 21301
+ },
+ {
+ "epoch": 58.043596730245234,
+ "grad_norm": 1.8033039569854736,
+ "learning_rate": 7.896518274600845e-06,
+ "loss": 0.0496,
+ "step": 21302
+ },
+ {
+ "epoch": 58.04632152588556,
+ "grad_norm": 2.268251895904541,
+ "learning_rate": 7.89565553261807e-06,
+ "loss": 0.1243,
+ "step": 21303
+ },
+ {
+ "epoch": 58.049046321525886,
+ "grad_norm": 2.700159788131714,
+ "learning_rate": 7.894792807023864e-06,
+ "loss": 0.0344,
+ "step": 21304
+ },
+ {
+ "epoch": 58.05177111716621,
+ "grad_norm": 1.9671779870986938,
+ "learning_rate": 7.89393009782494e-06,
+ "loss": 0.0274,
+ "step": 21305
+ },
+ {
+ "epoch": 58.05449591280654,
+ "grad_norm": 2.077565908432007,
+ "learning_rate": 7.893067405028025e-06,
+ "loss": 0.0561,
+ "step": 21306
+ },
+ {
+ "epoch": 58.05722070844686,
+ "grad_norm": 1.8887159824371338,
+ "learning_rate": 7.89220472863983e-06,
+ "loss": 0.0193,
+ "step": 21307
+ },
+ {
+ "epoch": 58.059945504087196,
+ "grad_norm": 2.032059907913208,
+ "learning_rate": 7.891342068667079e-06,
+ "loss": 0.0318,
+ "step": 21308
+ },
+ {
+ "epoch": 58.06267029972752,
+ "grad_norm": 1.9449037313461304,
+ "learning_rate": 7.890479425116484e-06,
+ "loss": 0.0847,
+ "step": 21309
+ },
+ {
+ "epoch": 58.06539509536785,
+ "grad_norm": 1.8887825012207031,
+ "learning_rate": 7.889616797994768e-06,
+ "loss": 0.0308,
+ "step": 21310
+ },
+ {
+ "epoch": 58.06811989100817,
+ "grad_norm": 2.6162326335906982,
+ "learning_rate": 7.88875418730865e-06,
+ "loss": 0.2102,
+ "step": 21311
+ },
+ {
+ "epoch": 58.0708446866485,
+ "grad_norm": 1.9880342483520508,
+ "learning_rate": 7.887891593064848e-06,
+ "loss": 0.0219,
+ "step": 21312
+ },
+ {
+ "epoch": 58.073569482288825,
+ "grad_norm": 1.9218113422393799,
+ "learning_rate": 7.887029015270076e-06,
+ "loss": 0.0278,
+ "step": 21313
+ },
+ {
+ "epoch": 58.07629427792916,
+ "grad_norm": 2.324369430541992,
+ "learning_rate": 7.886166453931056e-06,
+ "loss": 0.0886,
+ "step": 21314
+ },
+ {
+ "epoch": 58.079019073569484,
+ "grad_norm": 1.8903151750564575,
+ "learning_rate": 7.885303909054504e-06,
+ "loss": 0.026,
+ "step": 21315
+ },
+ {
+ "epoch": 58.08174386920981,
+ "grad_norm": 1.4897536039352417,
+ "learning_rate": 7.884441380647136e-06,
+ "loss": 0.0224,
+ "step": 21316
+ },
+ {
+ "epoch": 58.084468664850135,
+ "grad_norm": 1.9680343866348267,
+ "learning_rate": 7.88357886871567e-06,
+ "loss": 0.0878,
+ "step": 21317
+ },
+ {
+ "epoch": 58.08719346049046,
+ "grad_norm": 2.4236695766448975,
+ "learning_rate": 7.882716373266825e-06,
+ "loss": 0.0375,
+ "step": 21318
+ },
+ {
+ "epoch": 58.08991825613079,
+ "grad_norm": 2.1958487033843994,
+ "learning_rate": 7.881853894307316e-06,
+ "loss": 0.1619,
+ "step": 21319
+ },
+ {
+ "epoch": 58.09264305177112,
+ "grad_norm": 2.5010645389556885,
+ "learning_rate": 7.880991431843861e-06,
+ "loss": 0.0333,
+ "step": 21320
+ },
+ {
+ "epoch": 58.095367847411445,
+ "grad_norm": 1.625545859336853,
+ "learning_rate": 7.880128985883175e-06,
+ "loss": 0.0303,
+ "step": 21321
+ },
+ {
+ "epoch": 58.09809264305177,
+ "grad_norm": 1.4988152980804443,
+ "learning_rate": 7.879266556431979e-06,
+ "loss": 0.0168,
+ "step": 21322
+ },
+ {
+ "epoch": 58.1008174386921,
+ "grad_norm": 2.232743978500366,
+ "learning_rate": 7.878404143496985e-06,
+ "loss": 0.0347,
+ "step": 21323
+ },
+ {
+ "epoch": 58.10354223433242,
+ "grad_norm": 1.380751609802246,
+ "learning_rate": 7.877541747084912e-06,
+ "loss": 0.0224,
+ "step": 21324
+ },
+ {
+ "epoch": 58.10626702997275,
+ "grad_norm": 2.7193048000335693,
+ "learning_rate": 7.876679367202474e-06,
+ "loss": 0.1077,
+ "step": 21325
+ },
+ {
+ "epoch": 58.10899182561308,
+ "grad_norm": 1.302163004875183,
+ "learning_rate": 7.875817003856392e-06,
+ "loss": 0.0147,
+ "step": 21326
+ },
+ {
+ "epoch": 58.11171662125341,
+ "grad_norm": 2.450270175933838,
+ "learning_rate": 7.874954657053373e-06,
+ "loss": 0.138,
+ "step": 21327
+ },
+ {
+ "epoch": 58.11444141689373,
+ "grad_norm": 1.9793018102645874,
+ "learning_rate": 7.874092326800142e-06,
+ "loss": 0.0607,
+ "step": 21328
+ },
+ {
+ "epoch": 58.11716621253406,
+ "grad_norm": 2.5006790161132812,
+ "learning_rate": 7.87323001310341e-06,
+ "loss": 0.1855,
+ "step": 21329
+ },
+ {
+ "epoch": 58.119891008174385,
+ "grad_norm": 1.5526539087295532,
+ "learning_rate": 7.872367715969897e-06,
+ "loss": 0.0242,
+ "step": 21330
+ },
+ {
+ "epoch": 58.12261580381471,
+ "grad_norm": 2.4691832065582275,
+ "learning_rate": 7.871505435406315e-06,
+ "loss": 0.123,
+ "step": 21331
+ },
+ {
+ "epoch": 58.12534059945504,
+ "grad_norm": 1.7410720586776733,
+ "learning_rate": 7.870643171419382e-06,
+ "loss": 0.0218,
+ "step": 21332
+ },
+ {
+ "epoch": 58.12806539509537,
+ "grad_norm": 1.889326572418213,
+ "learning_rate": 7.869780924015807e-06,
+ "loss": 0.0266,
+ "step": 21333
+ },
+ {
+ "epoch": 58.130790190735695,
+ "grad_norm": 3.4017140865325928,
+ "learning_rate": 7.868918693202314e-06,
+ "loss": 0.0511,
+ "step": 21334
+ },
+ {
+ "epoch": 58.13351498637602,
+ "grad_norm": 1.8918582201004028,
+ "learning_rate": 7.86805647898561e-06,
+ "loss": 0.061,
+ "step": 21335
+ },
+ {
+ "epoch": 58.13623978201635,
+ "grad_norm": 2.706906795501709,
+ "learning_rate": 7.867194281372417e-06,
+ "loss": 0.0611,
+ "step": 21336
+ },
+ {
+ "epoch": 58.13896457765667,
+ "grad_norm": 3.6651406288146973,
+ "learning_rate": 7.866332100369444e-06,
+ "loss": 0.1086,
+ "step": 21337
+ },
+ {
+ "epoch": 58.141689373297005,
+ "grad_norm": 1.3749679327011108,
+ "learning_rate": 7.865469935983409e-06,
+ "loss": 0.046,
+ "step": 21338
+ },
+ {
+ "epoch": 58.14441416893733,
+ "grad_norm": 2.4738919734954834,
+ "learning_rate": 7.864607788221024e-06,
+ "loss": 0.0792,
+ "step": 21339
+ },
+ {
+ "epoch": 58.14713896457766,
+ "grad_norm": 2.09600830078125,
+ "learning_rate": 7.863745657089008e-06,
+ "loss": 0.0839,
+ "step": 21340
+ },
+ {
+ "epoch": 58.14986376021798,
+ "grad_norm": 1.6354560852050781,
+ "learning_rate": 7.862883542594067e-06,
+ "loss": 0.0252,
+ "step": 21341
+ },
+ {
+ "epoch": 58.15258855585831,
+ "grad_norm": 1.435293436050415,
+ "learning_rate": 7.862021444742924e-06,
+ "loss": 0.0367,
+ "step": 21342
+ },
+ {
+ "epoch": 58.155313351498634,
+ "grad_norm": 1.5711159706115723,
+ "learning_rate": 7.861159363542289e-06,
+ "loss": 0.0693,
+ "step": 21343
+ },
+ {
+ "epoch": 58.15803814713897,
+ "grad_norm": 1.5336220264434814,
+ "learning_rate": 7.860297298998871e-06,
+ "loss": 0.0222,
+ "step": 21344
+ },
+ {
+ "epoch": 58.16076294277929,
+ "grad_norm": 2.5056862831115723,
+ "learning_rate": 7.859435251119394e-06,
+ "loss": 0.1213,
+ "step": 21345
+ },
+ {
+ "epoch": 58.16348773841962,
+ "grad_norm": 2.409665822982788,
+ "learning_rate": 7.858573219910565e-06,
+ "loss": 0.1294,
+ "step": 21346
+ },
+ {
+ "epoch": 58.166212534059945,
+ "grad_norm": 2.3158464431762695,
+ "learning_rate": 7.857711205379099e-06,
+ "loss": 0.0977,
+ "step": 21347
+ },
+ {
+ "epoch": 58.16893732970027,
+ "grad_norm": 1.5136346817016602,
+ "learning_rate": 7.85684920753171e-06,
+ "loss": 0.0957,
+ "step": 21348
+ },
+ {
+ "epoch": 58.171662125340596,
+ "grad_norm": 1.9174063205718994,
+ "learning_rate": 7.85598722637511e-06,
+ "loss": 0.0692,
+ "step": 21349
+ },
+ {
+ "epoch": 58.17438692098093,
+ "grad_norm": 3.0842652320861816,
+ "learning_rate": 7.855125261916012e-06,
+ "loss": 0.0473,
+ "step": 21350
+ },
+ {
+ "epoch": 58.177111716621255,
+ "grad_norm": 1.8409278392791748,
+ "learning_rate": 7.854263314161128e-06,
+ "loss": 0.0207,
+ "step": 21351
+ },
+ {
+ "epoch": 58.17983651226158,
+ "grad_norm": 1.8013767004013062,
+ "learning_rate": 7.853401383117176e-06,
+ "loss": 0.1145,
+ "step": 21352
+ },
+ {
+ "epoch": 58.182561307901906,
+ "grad_norm": 2.1727113723754883,
+ "learning_rate": 7.85253946879086e-06,
+ "loss": 0.0495,
+ "step": 21353
+ },
+ {
+ "epoch": 58.18528610354223,
+ "grad_norm": 2.234278678894043,
+ "learning_rate": 7.851677571188903e-06,
+ "loss": 0.1059,
+ "step": 21354
+ },
+ {
+ "epoch": 58.18801089918256,
+ "grad_norm": 2.022228956222534,
+ "learning_rate": 7.850815690318007e-06,
+ "loss": 0.0262,
+ "step": 21355
+ },
+ {
+ "epoch": 58.19073569482289,
+ "grad_norm": 2.455186605453491,
+ "learning_rate": 7.849953826184892e-06,
+ "loss": 0.2203,
+ "step": 21356
+ },
+ {
+ "epoch": 58.19346049046322,
+ "grad_norm": 2.45652437210083,
+ "learning_rate": 7.849091978796267e-06,
+ "loss": 0.034,
+ "step": 21357
+ },
+ {
+ "epoch": 58.19618528610354,
+ "grad_norm": 1.9167811870574951,
+ "learning_rate": 7.848230148158847e-06,
+ "loss": 0.1,
+ "step": 21358
+ },
+ {
+ "epoch": 58.19891008174387,
+ "grad_norm": 2.218451499938965,
+ "learning_rate": 7.847368334279336e-06,
+ "loss": 0.0438,
+ "step": 21359
+ },
+ {
+ "epoch": 58.201634877384194,
+ "grad_norm": 2.3814642429351807,
+ "learning_rate": 7.846506537164456e-06,
+ "loss": 0.0432,
+ "step": 21360
+ },
+ {
+ "epoch": 58.20435967302452,
+ "grad_norm": 2.650606155395508,
+ "learning_rate": 7.845644756820908e-06,
+ "loss": 0.0859,
+ "step": 21361
+ },
+ {
+ "epoch": 58.20708446866485,
+ "grad_norm": 1.2411437034606934,
+ "learning_rate": 7.844782993255412e-06,
+ "loss": 0.0182,
+ "step": 21362
+ },
+ {
+ "epoch": 58.20980926430518,
+ "grad_norm": 2.2458765506744385,
+ "learning_rate": 7.843921246474679e-06,
+ "loss": 0.0386,
+ "step": 21363
+ },
+ {
+ "epoch": 58.212534059945504,
+ "grad_norm": 1.549630880355835,
+ "learning_rate": 7.843059516485415e-06,
+ "loss": 0.017,
+ "step": 21364
+ },
+ {
+ "epoch": 58.21525885558583,
+ "grad_norm": 3.197328567504883,
+ "learning_rate": 7.842197803294337e-06,
+ "loss": 0.0706,
+ "step": 21365
+ },
+ {
+ "epoch": 58.217983651226156,
+ "grad_norm": 3.264719247817993,
+ "learning_rate": 7.84133610690815e-06,
+ "loss": 0.0287,
+ "step": 21366
+ },
+ {
+ "epoch": 58.22070844686648,
+ "grad_norm": 2.022225856781006,
+ "learning_rate": 7.840474427333573e-06,
+ "loss": 0.039,
+ "step": 21367
+ },
+ {
+ "epoch": 58.223433242506815,
+ "grad_norm": 1.4830855131149292,
+ "learning_rate": 7.839612764577305e-06,
+ "loss": 0.0229,
+ "step": 21368
+ },
+ {
+ "epoch": 58.22615803814714,
+ "grad_norm": 1.3550142049789429,
+ "learning_rate": 7.838751118646068e-06,
+ "loss": 0.024,
+ "step": 21369
+ },
+ {
+ "epoch": 58.228882833787466,
+ "grad_norm": 1.5133583545684814,
+ "learning_rate": 7.837889489546566e-06,
+ "loss": 0.022,
+ "step": 21370
+ },
+ {
+ "epoch": 58.23160762942779,
+ "grad_norm": 2.700192928314209,
+ "learning_rate": 7.837027877285512e-06,
+ "loss": 0.101,
+ "step": 21371
+ },
+ {
+ "epoch": 58.23433242506812,
+ "grad_norm": 1.9770888090133667,
+ "learning_rate": 7.836166281869614e-06,
+ "loss": 0.0451,
+ "step": 21372
+ },
+ {
+ "epoch": 58.237057220708444,
+ "grad_norm": 5.306911945343018,
+ "learning_rate": 7.835304703305583e-06,
+ "loss": 0.1377,
+ "step": 21373
+ },
+ {
+ "epoch": 58.23978201634878,
+ "grad_norm": 1.960842490196228,
+ "learning_rate": 7.834443141600131e-06,
+ "loss": 0.0205,
+ "step": 21374
+ },
+ {
+ "epoch": 58.2425068119891,
+ "grad_norm": 1.7437938451766968,
+ "learning_rate": 7.833581596759965e-06,
+ "loss": 0.0483,
+ "step": 21375
+ },
+ {
+ "epoch": 58.24523160762943,
+ "grad_norm": 1.560953140258789,
+ "learning_rate": 7.832720068791798e-06,
+ "loss": 0.021,
+ "step": 21376
+ },
+ {
+ "epoch": 58.247956403269754,
+ "grad_norm": 2.3287198543548584,
+ "learning_rate": 7.831858557702332e-06,
+ "loss": 0.0382,
+ "step": 21377
+ },
+ {
+ "epoch": 58.25068119891008,
+ "grad_norm": 2.3126614093780518,
+ "learning_rate": 7.830997063498288e-06,
+ "loss": 0.0402,
+ "step": 21378
+ },
+ {
+ "epoch": 58.253405994550405,
+ "grad_norm": 2.246802568435669,
+ "learning_rate": 7.830135586186363e-06,
+ "loss": 0.0366,
+ "step": 21379
+ },
+ {
+ "epoch": 58.25613079019074,
+ "grad_norm": 2.0984959602355957,
+ "learning_rate": 7.829274125773275e-06,
+ "loss": 0.0808,
+ "step": 21380
+ },
+ {
+ "epoch": 58.258855585831064,
+ "grad_norm": 2.26249098777771,
+ "learning_rate": 7.82841268226573e-06,
+ "loss": 0.0877,
+ "step": 21381
+ },
+ {
+ "epoch": 58.26158038147139,
+ "grad_norm": 2.1466188430786133,
+ "learning_rate": 7.827551255670438e-06,
+ "loss": 0.053,
+ "step": 21382
+ },
+ {
+ "epoch": 58.264305177111716,
+ "grad_norm": 2.851383686065674,
+ "learning_rate": 7.82668984599411e-06,
+ "loss": 0.054,
+ "step": 21383
+ },
+ {
+ "epoch": 58.26702997275204,
+ "grad_norm": 1.8919548988342285,
+ "learning_rate": 7.825828453243446e-06,
+ "loss": 0.0286,
+ "step": 21384
+ },
+ {
+ "epoch": 58.26975476839237,
+ "grad_norm": 2.0569636821746826,
+ "learning_rate": 7.824967077425163e-06,
+ "loss": 0.0285,
+ "step": 21385
+ },
+ {
+ "epoch": 58.2724795640327,
+ "grad_norm": 191.47093200683594,
+ "learning_rate": 7.824105718545964e-06,
+ "loss": 0.044,
+ "step": 21386
+ },
+ {
+ "epoch": 58.275204359673026,
+ "grad_norm": 2.455397129058838,
+ "learning_rate": 7.823244376612565e-06,
+ "loss": 0.0278,
+ "step": 21387
+ },
+ {
+ "epoch": 58.27792915531335,
+ "grad_norm": 1.7611243724822998,
+ "learning_rate": 7.822383051631664e-06,
+ "loss": 0.02,
+ "step": 21388
+ },
+ {
+ "epoch": 58.28065395095368,
+ "grad_norm": 2.3751742839813232,
+ "learning_rate": 7.821521743609974e-06,
+ "loss": 0.0762,
+ "step": 21389
+ },
+ {
+ "epoch": 58.283378746594,
+ "grad_norm": 1.8133372068405151,
+ "learning_rate": 7.820660452554203e-06,
+ "loss": 0.0749,
+ "step": 21390
+ },
+ {
+ "epoch": 58.28610354223433,
+ "grad_norm": 3.0598838329315186,
+ "learning_rate": 7.81979917847106e-06,
+ "loss": 0.0529,
+ "step": 21391
+ },
+ {
+ "epoch": 58.28882833787466,
+ "grad_norm": 2.0502231121063232,
+ "learning_rate": 7.81893792136725e-06,
+ "loss": 0.0701,
+ "step": 21392
+ },
+ {
+ "epoch": 58.29155313351499,
+ "grad_norm": 6.004947662353516,
+ "learning_rate": 7.81807668124948e-06,
+ "loss": 0.0484,
+ "step": 21393
+ },
+ {
+ "epoch": 58.294277929155314,
+ "grad_norm": 3.1579110622406006,
+ "learning_rate": 7.817215458124459e-06,
+ "loss": 0.0356,
+ "step": 21394
+ },
+ {
+ "epoch": 58.29700272479564,
+ "grad_norm": 2.576709032058716,
+ "learning_rate": 7.816354251998896e-06,
+ "loss": 0.0639,
+ "step": 21395
+ },
+ {
+ "epoch": 58.299727520435965,
+ "grad_norm": 4.055290222167969,
+ "learning_rate": 7.815493062879488e-06,
+ "loss": 0.0898,
+ "step": 21396
+ },
+ {
+ "epoch": 58.30245231607629,
+ "grad_norm": 2.749274253845215,
+ "learning_rate": 7.814631890772957e-06,
+ "loss": 0.1345,
+ "step": 21397
+ },
+ {
+ "epoch": 58.305177111716624,
+ "grad_norm": 4.214979648590088,
+ "learning_rate": 7.813770735686e-06,
+ "loss": 0.1904,
+ "step": 21398
+ },
+ {
+ "epoch": 58.30790190735695,
+ "grad_norm": 2.873530626296997,
+ "learning_rate": 7.812909597625328e-06,
+ "loss": 0.083,
+ "step": 21399
+ },
+ {
+ "epoch": 58.310626702997276,
+ "grad_norm": 4.1929426193237305,
+ "learning_rate": 7.812048476597645e-06,
+ "loss": 0.0412,
+ "step": 21400
+ },
+ {
+ "epoch": 58.3133514986376,
+ "grad_norm": 2.9833741188049316,
+ "learning_rate": 7.811187372609658e-06,
+ "loss": 0.0466,
+ "step": 21401
+ },
+ {
+ "epoch": 58.31607629427793,
+ "grad_norm": 4.092068195343018,
+ "learning_rate": 7.810326285668073e-06,
+ "loss": 0.0721,
+ "step": 21402
+ },
+ {
+ "epoch": 58.31880108991825,
+ "grad_norm": 3.8128671646118164,
+ "learning_rate": 7.809465215779596e-06,
+ "loss": 0.0477,
+ "step": 21403
+ },
+ {
+ "epoch": 58.321525885558586,
+ "grad_norm": 3.667264223098755,
+ "learning_rate": 7.808604162950934e-06,
+ "loss": 0.0705,
+ "step": 21404
+ },
+ {
+ "epoch": 58.32425068119891,
+ "grad_norm": 3.6622257232666016,
+ "learning_rate": 7.807743127188794e-06,
+ "loss": 0.049,
+ "step": 21405
+ },
+ {
+ "epoch": 58.32697547683924,
+ "grad_norm": 3.991621255874634,
+ "learning_rate": 7.806882108499878e-06,
+ "loss": 0.1148,
+ "step": 21406
+ },
+ {
+ "epoch": 58.32970027247956,
+ "grad_norm": 2.671076536178589,
+ "learning_rate": 7.806021106890894e-06,
+ "loss": 0.0358,
+ "step": 21407
+ },
+ {
+ "epoch": 58.33242506811989,
+ "grad_norm": 4.769007682800293,
+ "learning_rate": 7.805160122368548e-06,
+ "loss": 0.0648,
+ "step": 21408
+ },
+ {
+ "epoch": 58.335149863760215,
+ "grad_norm": 3.6334657669067383,
+ "learning_rate": 7.804299154939543e-06,
+ "loss": 0.1082,
+ "step": 21409
+ },
+ {
+ "epoch": 58.33787465940055,
+ "grad_norm": 3.35072922706604,
+ "learning_rate": 7.803438204610584e-06,
+ "loss": 0.0541,
+ "step": 21410
+ },
+ {
+ "epoch": 58.34059945504087,
+ "grad_norm": 3.4032328128814697,
+ "learning_rate": 7.802577271388381e-06,
+ "loss": 0.093,
+ "step": 21411
+ },
+ {
+ "epoch": 58.3433242506812,
+ "grad_norm": 3.467787981033325,
+ "learning_rate": 7.801716355279632e-06,
+ "loss": 0.0841,
+ "step": 21412
+ },
+ {
+ "epoch": 58.346049046321525,
+ "grad_norm": 3.166562795639038,
+ "learning_rate": 7.800855456291045e-06,
+ "loss": 0.1407,
+ "step": 21413
+ },
+ {
+ "epoch": 58.34877384196185,
+ "grad_norm": 2.9849791526794434,
+ "learning_rate": 7.799994574429329e-06,
+ "loss": 0.0703,
+ "step": 21414
+ },
+ {
+ "epoch": 58.35149863760218,
+ "grad_norm": 7.570132255554199,
+ "learning_rate": 7.79913370970118e-06,
+ "loss": 0.0589,
+ "step": 21415
+ },
+ {
+ "epoch": 58.35422343324251,
+ "grad_norm": 2.3788492679595947,
+ "learning_rate": 7.798272862113308e-06,
+ "loss": 0.038,
+ "step": 21416
+ },
+ {
+ "epoch": 58.356948228882835,
+ "grad_norm": 2.744842767715454,
+ "learning_rate": 7.797412031672419e-06,
+ "loss": 0.0325,
+ "step": 21417
+ },
+ {
+ "epoch": 58.35967302452316,
+ "grad_norm": 2.1262171268463135,
+ "learning_rate": 7.796551218385211e-06,
+ "loss": 0.0309,
+ "step": 21418
+ },
+ {
+ "epoch": 58.36239782016349,
+ "grad_norm": 2.693434000015259,
+ "learning_rate": 7.795690422258394e-06,
+ "loss": 0.0344,
+ "step": 21419
+ },
+ {
+ "epoch": 58.36512261580381,
+ "grad_norm": 42.840267181396484,
+ "learning_rate": 7.794829643298667e-06,
+ "loss": 0.094,
+ "step": 21420
+ },
+ {
+ "epoch": 58.36784741144414,
+ "grad_norm": 11.667839050292969,
+ "learning_rate": 7.793968881512737e-06,
+ "loss": 0.0545,
+ "step": 21421
+ },
+ {
+ "epoch": 58.37057220708447,
+ "grad_norm": 2.4000203609466553,
+ "learning_rate": 7.793108136907306e-06,
+ "loss": 0.042,
+ "step": 21422
+ },
+ {
+ "epoch": 58.3732970027248,
+ "grad_norm": 3.6074085235595703,
+ "learning_rate": 7.79224740948908e-06,
+ "loss": 0.0636,
+ "step": 21423
+ },
+ {
+ "epoch": 58.37602179836512,
+ "grad_norm": 3.019568920135498,
+ "learning_rate": 7.791386699264756e-06,
+ "loss": 0.1297,
+ "step": 21424
+ },
+ {
+ "epoch": 58.37874659400545,
+ "grad_norm": 2.388225793838501,
+ "learning_rate": 7.790526006241044e-06,
+ "loss": 0.0761,
+ "step": 21425
+ },
+ {
+ "epoch": 58.381471389645775,
+ "grad_norm": 3.913135290145874,
+ "learning_rate": 7.789665330424643e-06,
+ "loss": 0.1132,
+ "step": 21426
+ },
+ {
+ "epoch": 58.3841961852861,
+ "grad_norm": 3.1081464290618896,
+ "learning_rate": 7.78880467182226e-06,
+ "loss": 0.057,
+ "step": 21427
+ },
+ {
+ "epoch": 58.38692098092643,
+ "grad_norm": 3.059847831726074,
+ "learning_rate": 7.78794403044059e-06,
+ "loss": 0.1945,
+ "step": 21428
+ },
+ {
+ "epoch": 58.38964577656676,
+ "grad_norm": 2.065755844116211,
+ "learning_rate": 7.787083406286345e-06,
+ "loss": 0.1186,
+ "step": 21429
+ },
+ {
+ "epoch": 58.392370572207085,
+ "grad_norm": 2.284446954727173,
+ "learning_rate": 7.78622279936622e-06,
+ "loss": 0.177,
+ "step": 21430
+ },
+ {
+ "epoch": 58.39509536784741,
+ "grad_norm": 3.1781015396118164,
+ "learning_rate": 7.78536220968692e-06,
+ "loss": 0.0964,
+ "step": 21431
+ },
+ {
+ "epoch": 58.39782016348774,
+ "grad_norm": 3.189263343811035,
+ "learning_rate": 7.78450163725515e-06,
+ "loss": 0.0509,
+ "step": 21432
+ },
+ {
+ "epoch": 58.40054495912806,
+ "grad_norm": 3.4305577278137207,
+ "learning_rate": 7.78364108207761e-06,
+ "loss": 0.0461,
+ "step": 21433
+ },
+ {
+ "epoch": 58.403269754768395,
+ "grad_norm": 3.508859395980835,
+ "learning_rate": 7.782780544161002e-06,
+ "loss": 0.1113,
+ "step": 21434
+ },
+ {
+ "epoch": 58.40599455040872,
+ "grad_norm": 3.390941619873047,
+ "learning_rate": 7.781920023512027e-06,
+ "loss": 0.1704,
+ "step": 21435
+ },
+ {
+ "epoch": 58.40871934604905,
+ "grad_norm": 2.4030063152313232,
+ "learning_rate": 7.78105952013739e-06,
+ "loss": 0.0548,
+ "step": 21436
+ },
+ {
+ "epoch": 58.41144414168937,
+ "grad_norm": 2.068002462387085,
+ "learning_rate": 7.780199034043787e-06,
+ "loss": 0.1533,
+ "step": 21437
+ },
+ {
+ "epoch": 58.4141689373297,
+ "grad_norm": 3.1031606197357178,
+ "learning_rate": 7.779338565237922e-06,
+ "loss": 0.1861,
+ "step": 21438
+ },
+ {
+ "epoch": 58.416893732970024,
+ "grad_norm": 4.734791278839111,
+ "learning_rate": 7.778478113726501e-06,
+ "loss": 0.1521,
+ "step": 21439
+ },
+ {
+ "epoch": 58.41961852861036,
+ "grad_norm": 9.122775077819824,
+ "learning_rate": 7.777617679516217e-06,
+ "loss": 0.1644,
+ "step": 21440
+ },
+ {
+ "epoch": 58.42234332425068,
+ "grad_norm": 2.936645030975342,
+ "learning_rate": 7.776757262613777e-06,
+ "loss": 0.0429,
+ "step": 21441
+ },
+ {
+ "epoch": 58.42506811989101,
+ "grad_norm": 3.1332006454467773,
+ "learning_rate": 7.77589686302588e-06,
+ "loss": 0.0479,
+ "step": 21442
+ },
+ {
+ "epoch": 58.427792915531334,
+ "grad_norm": 4.105042934417725,
+ "learning_rate": 7.775036480759226e-06,
+ "loss": 0.0685,
+ "step": 21443
+ },
+ {
+ "epoch": 58.43051771117166,
+ "grad_norm": 2.3841710090637207,
+ "learning_rate": 7.774176115820516e-06,
+ "loss": 0.0313,
+ "step": 21444
+ },
+ {
+ "epoch": 58.433242506811986,
+ "grad_norm": 2.212765693664551,
+ "learning_rate": 7.773315768216452e-06,
+ "loss": 0.0209,
+ "step": 21445
+ },
+ {
+ "epoch": 58.43596730245232,
+ "grad_norm": 2.28586745262146,
+ "learning_rate": 7.772455437953731e-06,
+ "loss": 0.0282,
+ "step": 21446
+ },
+ {
+ "epoch": 58.438692098092645,
+ "grad_norm": 3.1287598609924316,
+ "learning_rate": 7.771595125039053e-06,
+ "loss": 0.1895,
+ "step": 21447
+ },
+ {
+ "epoch": 58.44141689373297,
+ "grad_norm": 1.8975571393966675,
+ "learning_rate": 7.770734829479126e-06,
+ "loss": 0.0355,
+ "step": 21448
+ },
+ {
+ "epoch": 58.444141689373296,
+ "grad_norm": 3.059744119644165,
+ "learning_rate": 7.76987455128064e-06,
+ "loss": 0.1762,
+ "step": 21449
+ },
+ {
+ "epoch": 58.44686648501362,
+ "grad_norm": 2.8331754207611084,
+ "learning_rate": 7.769014290450303e-06,
+ "loss": 0.0593,
+ "step": 21450
+ },
+ {
+ "epoch": 58.44959128065395,
+ "grad_norm": 3.2393651008605957,
+ "learning_rate": 7.76815404699481e-06,
+ "loss": 0.0651,
+ "step": 21451
+ },
+ {
+ "epoch": 58.45231607629428,
+ "grad_norm": 3.655819892883301,
+ "learning_rate": 7.76729382092086e-06,
+ "loss": 0.0402,
+ "step": 21452
+ },
+ {
+ "epoch": 58.45504087193461,
+ "grad_norm": 2.294581651687622,
+ "learning_rate": 7.766433612235154e-06,
+ "loss": 0.042,
+ "step": 21453
+ },
+ {
+ "epoch": 58.45776566757493,
+ "grad_norm": 2.072118043899536,
+ "learning_rate": 7.765573420944394e-06,
+ "loss": 0.0273,
+ "step": 21454
+ },
+ {
+ "epoch": 58.46049046321526,
+ "grad_norm": 4.6825270652771,
+ "learning_rate": 7.764713247055273e-06,
+ "loss": 0.0593,
+ "step": 21455
+ },
+ {
+ "epoch": 58.463215258855584,
+ "grad_norm": 2.9635086059570312,
+ "learning_rate": 7.763853090574495e-06,
+ "loss": 0.0443,
+ "step": 21456
+ },
+ {
+ "epoch": 58.46594005449591,
+ "grad_norm": 2.011329174041748,
+ "learning_rate": 7.762992951508757e-06,
+ "loss": 0.0307,
+ "step": 21457
+ },
+ {
+ "epoch": 58.46866485013624,
+ "grad_norm": 2.536632776260376,
+ "learning_rate": 7.762132829864757e-06,
+ "loss": 0.041,
+ "step": 21458
+ },
+ {
+ "epoch": 58.47138964577657,
+ "grad_norm": 2.6182758808135986,
+ "learning_rate": 7.761272725649195e-06,
+ "loss": 0.0376,
+ "step": 21459
+ },
+ {
+ "epoch": 58.474114441416894,
+ "grad_norm": 3.116442918777466,
+ "learning_rate": 7.760412638868768e-06,
+ "loss": 0.0541,
+ "step": 21460
+ },
+ {
+ "epoch": 58.47683923705722,
+ "grad_norm": 1.9262471199035645,
+ "learning_rate": 7.759552569530178e-06,
+ "loss": 0.0325,
+ "step": 21461
+ },
+ {
+ "epoch": 58.479564032697546,
+ "grad_norm": 120.25289154052734,
+ "learning_rate": 7.758692517640117e-06,
+ "loss": 0.0509,
+ "step": 21462
+ },
+ {
+ "epoch": 58.48228882833787,
+ "grad_norm": 2.284986972808838,
+ "learning_rate": 7.75783248320529e-06,
+ "loss": 0.0373,
+ "step": 21463
+ },
+ {
+ "epoch": 58.485013623978205,
+ "grad_norm": 2.4601173400878906,
+ "learning_rate": 7.756972466232388e-06,
+ "loss": 0.0291,
+ "step": 21464
+ },
+ {
+ "epoch": 58.48773841961853,
+ "grad_norm": 3.0323102474212646,
+ "learning_rate": 7.756112466728111e-06,
+ "loss": 0.0841,
+ "step": 21465
+ },
+ {
+ "epoch": 58.490463215258856,
+ "grad_norm": 5.296110153198242,
+ "learning_rate": 7.755252484699163e-06,
+ "loss": 0.0878,
+ "step": 21466
+ },
+ {
+ "epoch": 58.49318801089918,
+ "grad_norm": 4.330109119415283,
+ "learning_rate": 7.754392520152233e-06,
+ "loss": 0.0705,
+ "step": 21467
+ },
+ {
+ "epoch": 58.49591280653951,
+ "grad_norm": 1.9812688827514648,
+ "learning_rate": 7.753532573094024e-06,
+ "loss": 0.0332,
+ "step": 21468
+ },
+ {
+ "epoch": 58.49863760217983,
+ "grad_norm": 3.0659735202789307,
+ "learning_rate": 7.75267264353123e-06,
+ "loss": 0.0625,
+ "step": 21469
+ },
+ {
+ "epoch": 58.50136239782017,
+ "grad_norm": 21.85101890563965,
+ "learning_rate": 7.751812731470549e-06,
+ "loss": 0.0712,
+ "step": 21470
+ },
+ {
+ "epoch": 58.50408719346049,
+ "grad_norm": 3.152966260910034,
+ "learning_rate": 7.750952836918679e-06,
+ "loss": 0.0644,
+ "step": 21471
+ },
+ {
+ "epoch": 58.50681198910082,
+ "grad_norm": 4.0522780418396,
+ "learning_rate": 7.750092959882315e-06,
+ "loss": 0.0893,
+ "step": 21472
+ },
+ {
+ "epoch": 58.509536784741144,
+ "grad_norm": 3.241903066635132,
+ "learning_rate": 7.749233100368155e-06,
+ "loss": 0.0525,
+ "step": 21473
+ },
+ {
+ "epoch": 58.51226158038147,
+ "grad_norm": 3.728628158569336,
+ "learning_rate": 7.748373258382896e-06,
+ "loss": 0.1999,
+ "step": 21474
+ },
+ {
+ "epoch": 58.514986376021795,
+ "grad_norm": 5.200041770935059,
+ "learning_rate": 7.747513433933231e-06,
+ "loss": 0.0789,
+ "step": 21475
+ },
+ {
+ "epoch": 58.51771117166213,
+ "grad_norm": 3.337999105453491,
+ "learning_rate": 7.746653627025861e-06,
+ "loss": 0.1411,
+ "step": 21476
+ },
+ {
+ "epoch": 58.520435967302454,
+ "grad_norm": 10.729714393615723,
+ "learning_rate": 7.745793837667479e-06,
+ "loss": 0.0634,
+ "step": 21477
+ },
+ {
+ "epoch": 58.52316076294278,
+ "grad_norm": 7.118719577789307,
+ "learning_rate": 7.744934065864785e-06,
+ "loss": 0.0745,
+ "step": 21478
+ },
+ {
+ "epoch": 58.525885558583106,
+ "grad_norm": 2.527693748474121,
+ "learning_rate": 7.744074311624468e-06,
+ "loss": 0.0365,
+ "step": 21479
+ },
+ {
+ "epoch": 58.52861035422343,
+ "grad_norm": 3.71494722366333,
+ "learning_rate": 7.743214574953231e-06,
+ "loss": 0.0376,
+ "step": 21480
+ },
+ {
+ "epoch": 58.53133514986376,
+ "grad_norm": 2.8432867527008057,
+ "learning_rate": 7.742354855857761e-06,
+ "loss": 0.0415,
+ "step": 21481
+ },
+ {
+ "epoch": 58.53405994550409,
+ "grad_norm": 3.0487635135650635,
+ "learning_rate": 7.74149515434476e-06,
+ "loss": 0.0733,
+ "step": 21482
+ },
+ {
+ "epoch": 58.536784741144416,
+ "grad_norm": 4.491422653198242,
+ "learning_rate": 7.740635470420922e-06,
+ "loss": 0.1434,
+ "step": 21483
+ },
+ {
+ "epoch": 58.53950953678474,
+ "grad_norm": 7.242132186889648,
+ "learning_rate": 7.739775804092947e-06,
+ "loss": 0.0594,
+ "step": 21484
+ },
+ {
+ "epoch": 58.54223433242507,
+ "grad_norm": 3.160189390182495,
+ "learning_rate": 7.738916155367522e-06,
+ "loss": 0.0754,
+ "step": 21485
+ },
+ {
+ "epoch": 58.54495912806539,
+ "grad_norm": 4.193337917327881,
+ "learning_rate": 7.738056524251347e-06,
+ "loss": 0.2521,
+ "step": 21486
+ },
+ {
+ "epoch": 58.54768392370572,
+ "grad_norm": 5.652463912963867,
+ "learning_rate": 7.737196910751113e-06,
+ "loss": 0.0648,
+ "step": 21487
+ },
+ {
+ "epoch": 58.55040871934605,
+ "grad_norm": 2.9470674991607666,
+ "learning_rate": 7.73633731487352e-06,
+ "loss": 0.1313,
+ "step": 21488
+ },
+ {
+ "epoch": 58.55313351498638,
+ "grad_norm": 3.0041074752807617,
+ "learning_rate": 7.735477736625255e-06,
+ "loss": 0.0867,
+ "step": 21489
+ },
+ {
+ "epoch": 58.555858310626704,
+ "grad_norm": 2.7692418098449707,
+ "learning_rate": 7.73461817601302e-06,
+ "loss": 0.138,
+ "step": 21490
+ },
+ {
+ "epoch": 58.55858310626703,
+ "grad_norm": 3.2348742485046387,
+ "learning_rate": 7.733758633043502e-06,
+ "loss": 0.0997,
+ "step": 21491
+ },
+ {
+ "epoch": 58.561307901907355,
+ "grad_norm": 5.199392318725586,
+ "learning_rate": 7.732899107723401e-06,
+ "loss": 0.0737,
+ "step": 21492
+ },
+ {
+ "epoch": 58.56403269754768,
+ "grad_norm": 2.5761775970458984,
+ "learning_rate": 7.732039600059407e-06,
+ "loss": 0.0675,
+ "step": 21493
+ },
+ {
+ "epoch": 58.566757493188014,
+ "grad_norm": 4.526670455932617,
+ "learning_rate": 7.73118011005822e-06,
+ "loss": 0.0336,
+ "step": 21494
+ },
+ {
+ "epoch": 58.56948228882834,
+ "grad_norm": 3.1940295696258545,
+ "learning_rate": 7.730320637726524e-06,
+ "loss": 0.0671,
+ "step": 21495
+ },
+ {
+ "epoch": 58.572207084468666,
+ "grad_norm": 4.685419082641602,
+ "learning_rate": 7.729461183071023e-06,
+ "loss": 0.0912,
+ "step": 21496
+ },
+ {
+ "epoch": 58.57493188010899,
+ "grad_norm": 3.5167338848114014,
+ "learning_rate": 7.728601746098401e-06,
+ "loss": 0.1352,
+ "step": 21497
+ },
+ {
+ "epoch": 58.57765667574932,
+ "grad_norm": 3.3475358486175537,
+ "learning_rate": 7.727742326815357e-06,
+ "loss": 0.0448,
+ "step": 21498
+ },
+ {
+ "epoch": 58.58038147138964,
+ "grad_norm": 2.2082934379577637,
+ "learning_rate": 7.72688292522858e-06,
+ "loss": 0.0433,
+ "step": 21499
+ },
+ {
+ "epoch": 58.583106267029976,
+ "grad_norm": 4.580643177032471,
+ "learning_rate": 7.726023541344767e-06,
+ "loss": 0.0932,
+ "step": 21500
+ },
+ {
+ "epoch": 58.5858310626703,
+ "grad_norm": 2.962451934814453,
+ "learning_rate": 7.725164175170611e-06,
+ "loss": 0.129,
+ "step": 21501
+ },
+ {
+ "epoch": 58.58855585831063,
+ "grad_norm": 2.3987207412719727,
+ "learning_rate": 7.724304826712802e-06,
+ "loss": 0.0777,
+ "step": 21502
+ },
+ {
+ "epoch": 58.59128065395095,
+ "grad_norm": 3.2313315868377686,
+ "learning_rate": 7.723445495978034e-06,
+ "loss": 0.0535,
+ "step": 21503
+ },
+ {
+ "epoch": 58.59400544959128,
+ "grad_norm": 3.0143625736236572,
+ "learning_rate": 7.722586182973002e-06,
+ "loss": 0.0627,
+ "step": 21504
+ },
+ {
+ "epoch": 58.596730245231605,
+ "grad_norm": 3.0362558364868164,
+ "learning_rate": 7.721726887704392e-06,
+ "loss": 0.0595,
+ "step": 21505
+ },
+ {
+ "epoch": 58.59945504087194,
+ "grad_norm": 2.936662197113037,
+ "learning_rate": 7.720867610178902e-06,
+ "loss": 0.0673,
+ "step": 21506
+ },
+ {
+ "epoch": 58.60217983651226,
+ "grad_norm": 3.660651922225952,
+ "learning_rate": 7.72000835040322e-06,
+ "loss": 0.1343,
+ "step": 21507
+ },
+ {
+ "epoch": 58.60490463215259,
+ "grad_norm": 2.6684248447418213,
+ "learning_rate": 7.719149108384042e-06,
+ "loss": 0.105,
+ "step": 21508
+ },
+ {
+ "epoch": 58.607629427792915,
+ "grad_norm": 2.9238457679748535,
+ "learning_rate": 7.718289884128054e-06,
+ "loss": 0.0398,
+ "step": 21509
+ },
+ {
+ "epoch": 58.61035422343324,
+ "grad_norm": 3.07651948928833,
+ "learning_rate": 7.717430677641954e-06,
+ "loss": 0.0431,
+ "step": 21510
+ },
+ {
+ "epoch": 58.61307901907357,
+ "grad_norm": 3.4656612873077393,
+ "learning_rate": 7.716571488932427e-06,
+ "loss": 0.0514,
+ "step": 21511
+ },
+ {
+ "epoch": 58.6158038147139,
+ "grad_norm": 2.6011886596679688,
+ "learning_rate": 7.715712318006173e-06,
+ "loss": 0.0722,
+ "step": 21512
+ },
+ {
+ "epoch": 58.618528610354225,
+ "grad_norm": 2.497008800506592,
+ "learning_rate": 7.714853164869872e-06,
+ "loss": 0.0246,
+ "step": 21513
+ },
+ {
+ "epoch": 58.62125340599455,
+ "grad_norm": 2.7664084434509277,
+ "learning_rate": 7.713994029530224e-06,
+ "loss": 0.1138,
+ "step": 21514
+ },
+ {
+ "epoch": 58.62397820163488,
+ "grad_norm": 2.740528106689453,
+ "learning_rate": 7.713134911993917e-06,
+ "loss": 0.112,
+ "step": 21515
+ },
+ {
+ "epoch": 58.6267029972752,
+ "grad_norm": 2.6035585403442383,
+ "learning_rate": 7.712275812267638e-06,
+ "loss": 0.1072,
+ "step": 21516
+ },
+ {
+ "epoch": 58.62942779291553,
+ "grad_norm": 2.6341023445129395,
+ "learning_rate": 7.711416730358085e-06,
+ "loss": 0.0422,
+ "step": 21517
+ },
+ {
+ "epoch": 58.63215258855586,
+ "grad_norm": 3.1803650856018066,
+ "learning_rate": 7.710557666271943e-06,
+ "loss": 0.0863,
+ "step": 21518
+ },
+ {
+ "epoch": 58.63487738419619,
+ "grad_norm": 3.1433959007263184,
+ "learning_rate": 7.709698620015908e-06,
+ "loss": 0.0614,
+ "step": 21519
+ },
+ {
+ "epoch": 58.63760217983651,
+ "grad_norm": 3.1882758140563965,
+ "learning_rate": 7.708839591596663e-06,
+ "loss": 0.0683,
+ "step": 21520
+ },
+ {
+ "epoch": 58.64032697547684,
+ "grad_norm": 2.4298319816589355,
+ "learning_rate": 7.707980581020903e-06,
+ "loss": 0.0504,
+ "step": 21521
+ },
+ {
+ "epoch": 58.643051771117165,
+ "grad_norm": 2.0079338550567627,
+ "learning_rate": 7.707121588295316e-06,
+ "loss": 0.0298,
+ "step": 21522
+ },
+ {
+ "epoch": 58.64577656675749,
+ "grad_norm": 3.325286865234375,
+ "learning_rate": 7.706262613426594e-06,
+ "loss": 0.0331,
+ "step": 21523
+ },
+ {
+ "epoch": 58.64850136239782,
+ "grad_norm": 3.510437488555908,
+ "learning_rate": 7.705403656421422e-06,
+ "loss": 0.0884,
+ "step": 21524
+ },
+ {
+ "epoch": 58.65122615803815,
+ "grad_norm": 2.0673434734344482,
+ "learning_rate": 7.704544717286493e-06,
+ "loss": 0.0265,
+ "step": 21525
+ },
+ {
+ "epoch": 58.653950953678475,
+ "grad_norm": 2.3083136081695557,
+ "learning_rate": 7.703685796028499e-06,
+ "loss": 0.036,
+ "step": 21526
+ },
+ {
+ "epoch": 58.6566757493188,
+ "grad_norm": 2.352524757385254,
+ "learning_rate": 7.702826892654121e-06,
+ "loss": 0.0257,
+ "step": 21527
+ },
+ {
+ "epoch": 58.65940054495913,
+ "grad_norm": 3.578261613845825,
+ "learning_rate": 7.701968007170057e-06,
+ "loss": 0.0469,
+ "step": 21528
+ },
+ {
+ "epoch": 58.66212534059945,
+ "grad_norm": 3.0623321533203125,
+ "learning_rate": 7.701109139582993e-06,
+ "loss": 0.15,
+ "step": 21529
+ },
+ {
+ "epoch": 58.664850136239785,
+ "grad_norm": 2.3339152336120605,
+ "learning_rate": 7.700250289899614e-06,
+ "loss": 0.0682,
+ "step": 21530
+ },
+ {
+ "epoch": 58.66757493188011,
+ "grad_norm": 3.077481508255005,
+ "learning_rate": 7.699391458126612e-06,
+ "loss": 0.1256,
+ "step": 21531
+ },
+ {
+ "epoch": 58.67029972752044,
+ "grad_norm": 11.423670768737793,
+ "learning_rate": 7.698532644270678e-06,
+ "loss": 0.2215,
+ "step": 21532
+ },
+ {
+ "epoch": 58.67302452316076,
+ "grad_norm": 3.3275110721588135,
+ "learning_rate": 7.697673848338495e-06,
+ "loss": 0.0509,
+ "step": 21533
+ },
+ {
+ "epoch": 58.67574931880109,
+ "grad_norm": 2.4686343669891357,
+ "learning_rate": 7.696815070336752e-06,
+ "loss": 0.036,
+ "step": 21534
+ },
+ {
+ "epoch": 58.678474114441414,
+ "grad_norm": 3.0218520164489746,
+ "learning_rate": 7.695956310272143e-06,
+ "loss": 0.0689,
+ "step": 21535
+ },
+ {
+ "epoch": 58.68119891008175,
+ "grad_norm": 3.18528151512146,
+ "learning_rate": 7.695097568151351e-06,
+ "loss": 0.0982,
+ "step": 21536
+ },
+ {
+ "epoch": 58.68392370572207,
+ "grad_norm": 2.8117973804473877,
+ "learning_rate": 7.694238843981066e-06,
+ "loss": 0.0514,
+ "step": 21537
+ },
+ {
+ "epoch": 58.6866485013624,
+ "grad_norm": 4.177014350891113,
+ "learning_rate": 7.693380137767971e-06,
+ "loss": 0.051,
+ "step": 21538
+ },
+ {
+ "epoch": 58.689373297002724,
+ "grad_norm": 2.6624693870544434,
+ "learning_rate": 7.692521449518763e-06,
+ "loss": 0.0345,
+ "step": 21539
+ },
+ {
+ "epoch": 58.69209809264305,
+ "grad_norm": 2.6261472702026367,
+ "learning_rate": 7.691662779240119e-06,
+ "loss": 0.0787,
+ "step": 21540
+ },
+ {
+ "epoch": 58.694822888283376,
+ "grad_norm": 3.211308002471924,
+ "learning_rate": 7.690804126938733e-06,
+ "loss": 0.0444,
+ "step": 21541
+ },
+ {
+ "epoch": 58.69754768392371,
+ "grad_norm": 4.3777923583984375,
+ "learning_rate": 7.68994549262129e-06,
+ "loss": 0.0585,
+ "step": 21542
+ },
+ {
+ "epoch": 58.700272479564035,
+ "grad_norm": 2.3906261920928955,
+ "learning_rate": 7.689086876294479e-06,
+ "loss": 0.2415,
+ "step": 21543
+ },
+ {
+ "epoch": 58.70299727520436,
+ "grad_norm": 2.769740581512451,
+ "learning_rate": 7.68822827796498e-06,
+ "loss": 0.0324,
+ "step": 21544
+ },
+ {
+ "epoch": 58.705722070844686,
+ "grad_norm": 4.067683219909668,
+ "learning_rate": 7.68736969763949e-06,
+ "loss": 0.0683,
+ "step": 21545
+ },
+ {
+ "epoch": 58.70844686648501,
+ "grad_norm": 2.5457561016082764,
+ "learning_rate": 7.686511135324687e-06,
+ "loss": 0.035,
+ "step": 21546
+ },
+ {
+ "epoch": 58.71117166212534,
+ "grad_norm": 3.7460553646087646,
+ "learning_rate": 7.685652591027261e-06,
+ "loss": 0.0742,
+ "step": 21547
+ },
+ {
+ "epoch": 58.71389645776567,
+ "grad_norm": 2.3619496822357178,
+ "learning_rate": 7.6847940647539e-06,
+ "loss": 0.0992,
+ "step": 21548
+ },
+ {
+ "epoch": 58.716621253406,
+ "grad_norm": 1.919956088066101,
+ "learning_rate": 7.683935556511287e-06,
+ "loss": 0.0458,
+ "step": 21549
+ },
+ {
+ "epoch": 58.71934604904632,
+ "grad_norm": 3.259500503540039,
+ "learning_rate": 7.68307706630611e-06,
+ "loss": 0.1706,
+ "step": 21550
+ },
+ {
+ "epoch": 58.72207084468665,
+ "grad_norm": 2.850440263748169,
+ "learning_rate": 7.682218594145052e-06,
+ "loss": 0.1628,
+ "step": 21551
+ },
+ {
+ "epoch": 58.724795640326974,
+ "grad_norm": 1.7772876024246216,
+ "learning_rate": 7.681360140034802e-06,
+ "loss": 0.1118,
+ "step": 21552
+ },
+ {
+ "epoch": 58.7275204359673,
+ "grad_norm": 2.295637845993042,
+ "learning_rate": 7.680501703982047e-06,
+ "loss": 0.1262,
+ "step": 21553
+ },
+ {
+ "epoch": 58.73024523160763,
+ "grad_norm": 2.3949661254882812,
+ "learning_rate": 7.679643285993466e-06,
+ "loss": 0.0561,
+ "step": 21554
+ },
+ {
+ "epoch": 58.73297002724796,
+ "grad_norm": 2.7665817737579346,
+ "learning_rate": 7.678784886075753e-06,
+ "loss": 0.0693,
+ "step": 21555
+ },
+ {
+ "epoch": 58.735694822888284,
+ "grad_norm": 2.8356969356536865,
+ "learning_rate": 7.677926504235585e-06,
+ "loss": 0.035,
+ "step": 21556
+ },
+ {
+ "epoch": 58.73841961852861,
+ "grad_norm": 2.777632474899292,
+ "learning_rate": 7.677068140479654e-06,
+ "loss": 0.1695,
+ "step": 21557
+ },
+ {
+ "epoch": 58.741144414168936,
+ "grad_norm": 2.359386920928955,
+ "learning_rate": 7.676209794814639e-06,
+ "loss": 0.0285,
+ "step": 21558
+ },
+ {
+ "epoch": 58.74386920980926,
+ "grad_norm": 3.929917573928833,
+ "learning_rate": 7.67535146724723e-06,
+ "loss": 0.045,
+ "step": 21559
+ },
+ {
+ "epoch": 58.746594005449595,
+ "grad_norm": 2.3769686222076416,
+ "learning_rate": 7.674493157784106e-06,
+ "loss": 0.0622,
+ "step": 21560
+ },
+ {
+ "epoch": 58.74931880108992,
+ "grad_norm": 4.042645454406738,
+ "learning_rate": 7.673634866431955e-06,
+ "loss": 0.0463,
+ "step": 21561
+ },
+ {
+ "epoch": 58.752043596730246,
+ "grad_norm": 3.065056324005127,
+ "learning_rate": 7.672776593197463e-06,
+ "loss": 0.0866,
+ "step": 21562
+ },
+ {
+ "epoch": 58.75476839237057,
+ "grad_norm": 4.539422035217285,
+ "learning_rate": 7.67191833808731e-06,
+ "loss": 0.0529,
+ "step": 21563
+ },
+ {
+ "epoch": 58.7574931880109,
+ "grad_norm": 2.1391661167144775,
+ "learning_rate": 7.671060101108182e-06,
+ "loss": 0.0358,
+ "step": 21564
+ },
+ {
+ "epoch": 58.76021798365122,
+ "grad_norm": 12.793122291564941,
+ "learning_rate": 7.670201882266764e-06,
+ "loss": 0.0459,
+ "step": 21565
+ },
+ {
+ "epoch": 58.762942779291556,
+ "grad_norm": 4.673292636871338,
+ "learning_rate": 7.669343681569737e-06,
+ "loss": 0.0883,
+ "step": 21566
+ },
+ {
+ "epoch": 58.76566757493188,
+ "grad_norm": 2.472615957260132,
+ "learning_rate": 7.66848549902379e-06,
+ "loss": 0.1051,
+ "step": 21567
+ },
+ {
+ "epoch": 58.76839237057221,
+ "grad_norm": 1.8613640069961548,
+ "learning_rate": 7.667627334635595e-06,
+ "loss": 0.0237,
+ "step": 21568
+ },
+ {
+ "epoch": 58.771117166212534,
+ "grad_norm": 2.8041224479675293,
+ "learning_rate": 7.666769188411848e-06,
+ "loss": 0.0485,
+ "step": 21569
+ },
+ {
+ "epoch": 58.77384196185286,
+ "grad_norm": 3.1028263568878174,
+ "learning_rate": 7.665911060359228e-06,
+ "loss": 0.1236,
+ "step": 21570
+ },
+ {
+ "epoch": 58.776566757493185,
+ "grad_norm": 3.3375024795532227,
+ "learning_rate": 7.66505295048442e-06,
+ "loss": 0.0664,
+ "step": 21571
+ },
+ {
+ "epoch": 58.77929155313352,
+ "grad_norm": 2.8683290481567383,
+ "learning_rate": 7.6641948587941e-06,
+ "loss": 0.0808,
+ "step": 21572
+ },
+ {
+ "epoch": 58.782016348773844,
+ "grad_norm": 2.9646830558776855,
+ "learning_rate": 7.663336785294959e-06,
+ "loss": 0.0558,
+ "step": 21573
+ },
+ {
+ "epoch": 58.78474114441417,
+ "grad_norm": 3.293139934539795,
+ "learning_rate": 7.662478729993673e-06,
+ "loss": 0.1405,
+ "step": 21574
+ },
+ {
+ "epoch": 58.787465940054496,
+ "grad_norm": 2.481926918029785,
+ "learning_rate": 7.661620692896931e-06,
+ "loss": 0.0509,
+ "step": 21575
+ },
+ {
+ "epoch": 58.79019073569482,
+ "grad_norm": 2.7436981201171875,
+ "learning_rate": 7.660762674011408e-06,
+ "loss": 0.0821,
+ "step": 21576
+ },
+ {
+ "epoch": 58.79291553133515,
+ "grad_norm": 3.259918451309204,
+ "learning_rate": 7.659904673343793e-06,
+ "loss": 0.0553,
+ "step": 21577
+ },
+ {
+ "epoch": 58.79564032697548,
+ "grad_norm": 2.4369513988494873,
+ "learning_rate": 7.659046690900762e-06,
+ "loss": 0.1204,
+ "step": 21578
+ },
+ {
+ "epoch": 58.798365122615806,
+ "grad_norm": 2.808702230453491,
+ "learning_rate": 7.658188726689003e-06,
+ "loss": 0.0538,
+ "step": 21579
+ },
+ {
+ "epoch": 58.80108991825613,
+ "grad_norm": 2.7337429523468018,
+ "learning_rate": 7.657330780715193e-06,
+ "loss": 0.101,
+ "step": 21580
+ },
+ {
+ "epoch": 58.80381471389646,
+ "grad_norm": 4.326451301574707,
+ "learning_rate": 7.656472852986017e-06,
+ "loss": 0.0607,
+ "step": 21581
+ },
+ {
+ "epoch": 58.80653950953678,
+ "grad_norm": 3.5991668701171875,
+ "learning_rate": 7.655614943508153e-06,
+ "loss": 0.0576,
+ "step": 21582
+ },
+ {
+ "epoch": 58.80926430517711,
+ "grad_norm": 2.6510167121887207,
+ "learning_rate": 7.654757052288286e-06,
+ "loss": 0.0307,
+ "step": 21583
+ },
+ {
+ "epoch": 58.81198910081744,
+ "grad_norm": 2.3271803855895996,
+ "learning_rate": 7.653899179333094e-06,
+ "loss": 0.0832,
+ "step": 21584
+ },
+ {
+ "epoch": 58.81471389645777,
+ "grad_norm": 2.1004087924957275,
+ "learning_rate": 7.653041324649258e-06,
+ "loss": 0.0291,
+ "step": 21585
+ },
+ {
+ "epoch": 58.817438692098094,
+ "grad_norm": 3.038374900817871,
+ "learning_rate": 7.652183488243465e-06,
+ "loss": 0.033,
+ "step": 21586
+ },
+ {
+ "epoch": 58.82016348773842,
+ "grad_norm": 2.152613401412964,
+ "learning_rate": 7.651325670122389e-06,
+ "loss": 0.0371,
+ "step": 21587
+ },
+ {
+ "epoch": 58.822888283378745,
+ "grad_norm": 2.5146989822387695,
+ "learning_rate": 7.650467870292711e-06,
+ "loss": 0.0419,
+ "step": 21588
+ },
+ {
+ "epoch": 58.82561307901907,
+ "grad_norm": 2.8988776206970215,
+ "learning_rate": 7.649610088761117e-06,
+ "loss": 0.26,
+ "step": 21589
+ },
+ {
+ "epoch": 58.828337874659404,
+ "grad_norm": 2.4771888256073,
+ "learning_rate": 7.648752325534283e-06,
+ "loss": 0.04,
+ "step": 21590
+ },
+ {
+ "epoch": 58.83106267029973,
+ "grad_norm": 2.808974504470825,
+ "learning_rate": 7.647894580618892e-06,
+ "loss": 0.0398,
+ "step": 21591
+ },
+ {
+ "epoch": 58.833787465940055,
+ "grad_norm": 2.7568278312683105,
+ "learning_rate": 7.64703685402162e-06,
+ "loss": 0.115,
+ "step": 21592
+ },
+ {
+ "epoch": 58.83651226158038,
+ "grad_norm": 3.1542253494262695,
+ "learning_rate": 7.646179145749151e-06,
+ "loss": 0.0382,
+ "step": 21593
+ },
+ {
+ "epoch": 58.83923705722071,
+ "grad_norm": 2.9639861583709717,
+ "learning_rate": 7.645321455808162e-06,
+ "loss": 0.0461,
+ "step": 21594
+ },
+ {
+ "epoch": 58.84196185286103,
+ "grad_norm": 2.5628085136413574,
+ "learning_rate": 7.644463784205335e-06,
+ "loss": 0.0346,
+ "step": 21595
+ },
+ {
+ "epoch": 58.844686648501366,
+ "grad_norm": 2.060296058654785,
+ "learning_rate": 7.643606130947345e-06,
+ "loss": 0.0217,
+ "step": 21596
+ },
+ {
+ "epoch": 58.84741144414169,
+ "grad_norm": 3.021660566329956,
+ "learning_rate": 7.64274849604088e-06,
+ "loss": 0.1755,
+ "step": 21597
+ },
+ {
+ "epoch": 58.85013623978202,
+ "grad_norm": 2.3943886756896973,
+ "learning_rate": 7.641890879492608e-06,
+ "loss": 0.0309,
+ "step": 21598
+ },
+ {
+ "epoch": 58.85286103542234,
+ "grad_norm": 2.8328983783721924,
+ "learning_rate": 7.641033281309219e-06,
+ "loss": 0.0553,
+ "step": 21599
+ },
+ {
+ "epoch": 58.85558583106267,
+ "grad_norm": 1.7499319314956665,
+ "learning_rate": 7.640175701497383e-06,
+ "loss": 0.0213,
+ "step": 21600
+ },
+ {
+ "epoch": 58.858310626702995,
+ "grad_norm": 2.763207197189331,
+ "learning_rate": 7.639318140063785e-06,
+ "loss": 0.0271,
+ "step": 21601
+ },
+ {
+ "epoch": 58.86103542234333,
+ "grad_norm": 3.118647336959839,
+ "learning_rate": 7.638460597015097e-06,
+ "loss": 0.0483,
+ "step": 21602
+ },
+ {
+ "epoch": 58.86376021798365,
+ "grad_norm": 2.8396658897399902,
+ "learning_rate": 7.637603072358004e-06,
+ "loss": 0.0689,
+ "step": 21603
+ },
+ {
+ "epoch": 58.86648501362398,
+ "grad_norm": 2.8603508472442627,
+ "learning_rate": 7.636745566099185e-06,
+ "loss": 0.0429,
+ "step": 21604
+ },
+ {
+ "epoch": 58.869209809264305,
+ "grad_norm": 2.4617860317230225,
+ "learning_rate": 7.635888078245312e-06,
+ "loss": 0.0403,
+ "step": 21605
+ },
+ {
+ "epoch": 58.87193460490463,
+ "grad_norm": 1.96005117893219,
+ "learning_rate": 7.635030608803072e-06,
+ "loss": 0.0284,
+ "step": 21606
+ },
+ {
+ "epoch": 58.87465940054496,
+ "grad_norm": 2.943983316421509,
+ "learning_rate": 7.634173157779133e-06,
+ "loss": 0.1975,
+ "step": 21607
+ },
+ {
+ "epoch": 58.87738419618529,
+ "grad_norm": 2.079272508621216,
+ "learning_rate": 7.633315725180178e-06,
+ "loss": 0.0341,
+ "step": 21608
+ },
+ {
+ "epoch": 58.880108991825615,
+ "grad_norm": 3.928229808807373,
+ "learning_rate": 7.632458311012884e-06,
+ "loss": 0.1574,
+ "step": 21609
+ },
+ {
+ "epoch": 58.88283378746594,
+ "grad_norm": 2.3215270042419434,
+ "learning_rate": 7.63160091528393e-06,
+ "loss": 0.0313,
+ "step": 21610
+ },
+ {
+ "epoch": 58.88555858310627,
+ "grad_norm": 2.701824188232422,
+ "learning_rate": 7.630743537999989e-06,
+ "loss": 0.0331,
+ "step": 21611
+ },
+ {
+ "epoch": 58.88828337874659,
+ "grad_norm": 2.559706687927246,
+ "learning_rate": 7.629886179167743e-06,
+ "loss": 0.1054,
+ "step": 21612
+ },
+ {
+ "epoch": 58.89100817438692,
+ "grad_norm": 1.9146041870117188,
+ "learning_rate": 7.629028838793867e-06,
+ "loss": 0.0554,
+ "step": 21613
+ },
+ {
+ "epoch": 58.89373297002725,
+ "grad_norm": 1.928780436515808,
+ "learning_rate": 7.6281715168850364e-06,
+ "loss": 0.0341,
+ "step": 21614
+ },
+ {
+ "epoch": 58.89645776566758,
+ "grad_norm": 3.554821252822876,
+ "learning_rate": 7.6273142134479335e-06,
+ "loss": 0.1257,
+ "step": 21615
+ },
+ {
+ "epoch": 58.8991825613079,
+ "grad_norm": 2.499284029006958,
+ "learning_rate": 7.626456928489228e-06,
+ "loss": 0.0353,
+ "step": 21616
+ },
+ {
+ "epoch": 58.90190735694823,
+ "grad_norm": 2.2935142517089844,
+ "learning_rate": 7.625599662015602e-06,
+ "loss": 0.0404,
+ "step": 21617
+ },
+ {
+ "epoch": 58.904632152588555,
+ "grad_norm": 14.466222763061523,
+ "learning_rate": 7.624742414033725e-06,
+ "loss": 0.0605,
+ "step": 21618
+ },
+ {
+ "epoch": 58.90735694822888,
+ "grad_norm": 2.2436203956604004,
+ "learning_rate": 7.623885184550283e-06,
+ "loss": 0.0455,
+ "step": 21619
+ },
+ {
+ "epoch": 58.91008174386921,
+ "grad_norm": 2.184229850769043,
+ "learning_rate": 7.6230279735719415e-06,
+ "loss": 0.063,
+ "step": 21620
+ },
+ {
+ "epoch": 58.91280653950954,
+ "grad_norm": 2.619995594024658,
+ "learning_rate": 7.622170781105382e-06,
+ "loss": 0.0884,
+ "step": 21621
+ },
+ {
+ "epoch": 58.915531335149865,
+ "grad_norm": 3.064615488052368,
+ "learning_rate": 7.621313607157285e-06,
+ "loss": 0.0753,
+ "step": 21622
+ },
+ {
+ "epoch": 58.91825613079019,
+ "grad_norm": 2.4268176555633545,
+ "learning_rate": 7.620456451734317e-06,
+ "loss": 0.0442,
+ "step": 21623
+ },
+ {
+ "epoch": 58.920980926430516,
+ "grad_norm": 3.729375123977661,
+ "learning_rate": 7.61959931484316e-06,
+ "loss": 0.0287,
+ "step": 21624
+ },
+ {
+ "epoch": 58.92370572207084,
+ "grad_norm": 2.550609827041626,
+ "learning_rate": 7.618742196490485e-06,
+ "loss": 0.1087,
+ "step": 21625
+ },
+ {
+ "epoch": 58.926430517711175,
+ "grad_norm": 3.8394391536712646,
+ "learning_rate": 7.617885096682971e-06,
+ "loss": 0.1688,
+ "step": 21626
+ },
+ {
+ "epoch": 58.9291553133515,
+ "grad_norm": 2.6041245460510254,
+ "learning_rate": 7.617028015427289e-06,
+ "loss": 0.0426,
+ "step": 21627
+ },
+ {
+ "epoch": 58.93188010899183,
+ "grad_norm": 3.4257326126098633,
+ "learning_rate": 7.616170952730118e-06,
+ "loss": 0.0992,
+ "step": 21628
+ },
+ {
+ "epoch": 58.93460490463215,
+ "grad_norm": 2.844250202178955,
+ "learning_rate": 7.6153139085981296e-06,
+ "loss": 0.1398,
+ "step": 21629
+ },
+ {
+ "epoch": 58.93732970027248,
+ "grad_norm": 2.070683002471924,
+ "learning_rate": 7.614456883038001e-06,
+ "loss": 0.0293,
+ "step": 21630
+ },
+ {
+ "epoch": 58.940054495912804,
+ "grad_norm": 2.313575029373169,
+ "learning_rate": 7.613599876056404e-06,
+ "loss": 0.0569,
+ "step": 21631
+ },
+ {
+ "epoch": 58.94277929155314,
+ "grad_norm": 2.460676908493042,
+ "learning_rate": 7.612742887660016e-06,
+ "loss": 0.05,
+ "step": 21632
+ },
+ {
+ "epoch": 58.94550408719346,
+ "grad_norm": 1.6540063619613647,
+ "learning_rate": 7.611885917855508e-06,
+ "loss": 0.0229,
+ "step": 21633
+ },
+ {
+ "epoch": 58.94822888283379,
+ "grad_norm": 2.966891288757324,
+ "learning_rate": 7.611028966649555e-06,
+ "loss": 0.0383,
+ "step": 21634
+ },
+ {
+ "epoch": 58.950953678474114,
+ "grad_norm": 2.8487436771392822,
+ "learning_rate": 7.610172034048835e-06,
+ "loss": 0.0581,
+ "step": 21635
+ },
+ {
+ "epoch": 58.95367847411444,
+ "grad_norm": 1.6127744913101196,
+ "learning_rate": 7.609315120060015e-06,
+ "loss": 0.0415,
+ "step": 21636
+ },
+ {
+ "epoch": 58.956403269754766,
+ "grad_norm": 1.7472819089889526,
+ "learning_rate": 7.60845822468977e-06,
+ "loss": 0.0253,
+ "step": 21637
+ },
+ {
+ "epoch": 58.95912806539509,
+ "grad_norm": 1.8267289400100708,
+ "learning_rate": 7.6076013479447786e-06,
+ "loss": 0.0452,
+ "step": 21638
+ },
+ {
+ "epoch": 58.961852861035425,
+ "grad_norm": 2.92063045501709,
+ "learning_rate": 7.60674448983171e-06,
+ "loss": 0.0572,
+ "step": 21639
+ },
+ {
+ "epoch": 58.96457765667575,
+ "grad_norm": 2.8240244388580322,
+ "learning_rate": 7.60588765035724e-06,
+ "loss": 0.0268,
+ "step": 21640
+ },
+ {
+ "epoch": 58.967302452316076,
+ "grad_norm": 3.347069025039673,
+ "learning_rate": 7.605030829528037e-06,
+ "loss": 0.093,
+ "step": 21641
+ },
+ {
+ "epoch": 58.9700272479564,
+ "grad_norm": 3.107154130935669,
+ "learning_rate": 7.60417402735078e-06,
+ "loss": 0.0603,
+ "step": 21642
+ },
+ {
+ "epoch": 58.97275204359673,
+ "grad_norm": 2.555111885070801,
+ "learning_rate": 7.603317243832137e-06,
+ "loss": 0.0596,
+ "step": 21643
+ },
+ {
+ "epoch": 58.97547683923706,
+ "grad_norm": 2.3941566944122314,
+ "learning_rate": 7.6024604789787835e-06,
+ "loss": 0.0284,
+ "step": 21644
+ },
+ {
+ "epoch": 58.97820163487739,
+ "grad_norm": 2.303985834121704,
+ "learning_rate": 7.601603732797389e-06,
+ "loss": 0.1043,
+ "step": 21645
+ },
+ {
+ "epoch": 58.98092643051771,
+ "grad_norm": 2.016369342803955,
+ "learning_rate": 7.600747005294629e-06,
+ "loss": 0.0311,
+ "step": 21646
+ },
+ {
+ "epoch": 58.98365122615804,
+ "grad_norm": 2.1495368480682373,
+ "learning_rate": 7.599890296477171e-06,
+ "loss": 0.0733,
+ "step": 21647
+ },
+ {
+ "epoch": 58.986376021798364,
+ "grad_norm": 1.8376060724258423,
+ "learning_rate": 7.599033606351694e-06,
+ "loss": 0.0207,
+ "step": 21648
+ },
+ {
+ "epoch": 58.98910081743869,
+ "grad_norm": 2.7066190242767334,
+ "learning_rate": 7.598176934924863e-06,
+ "loss": 0.1304,
+ "step": 21649
+ },
+ {
+ "epoch": 58.991825613079016,
+ "grad_norm": 2.705211639404297,
+ "learning_rate": 7.5973202822033554e-06,
+ "loss": 0.1077,
+ "step": 21650
+ },
+ {
+ "epoch": 58.99455040871935,
+ "grad_norm": 1.9837160110473633,
+ "learning_rate": 7.596463648193837e-06,
+ "loss": 0.1119,
+ "step": 21651
+ },
+ {
+ "epoch": 58.997275204359674,
+ "grad_norm": 2.4356777667999268,
+ "learning_rate": 7.595607032902985e-06,
+ "loss": 0.0651,
+ "step": 21652
+ },
+ {
+ "epoch": 59.0,
+ "grad_norm": 2.0317940711975098,
+ "learning_rate": 7.594750436337467e-06,
+ "loss": 0.0212,
+ "step": 21653
+ },
+ {
+ "epoch": 59.002724795640326,
+ "grad_norm": 2.3906238079071045,
+ "learning_rate": 7.593893858503951e-06,
+ "loss": 0.0309,
+ "step": 21654
+ },
+ {
+ "epoch": 59.00544959128065,
+ "grad_norm": 2.6265718936920166,
+ "learning_rate": 7.593037299409116e-06,
+ "loss": 0.0456,
+ "step": 21655
+ },
+ {
+ "epoch": 59.00817438692098,
+ "grad_norm": 2.272932291030884,
+ "learning_rate": 7.592180759059631e-06,
+ "loss": 0.0271,
+ "step": 21656
+ },
+ {
+ "epoch": 59.01089918256131,
+ "grad_norm": 2.245856761932373,
+ "learning_rate": 7.591324237462162e-06,
+ "loss": 0.0547,
+ "step": 21657
+ },
+ {
+ "epoch": 59.013623978201636,
+ "grad_norm": 3.697932720184326,
+ "learning_rate": 7.5904677346233845e-06,
+ "loss": 0.0783,
+ "step": 21658
+ },
+ {
+ "epoch": 59.01634877384196,
+ "grad_norm": 2.1769561767578125,
+ "learning_rate": 7.589611250549965e-06,
+ "loss": 0.0266,
+ "step": 21659
+ },
+ {
+ "epoch": 59.01907356948229,
+ "grad_norm": 2.1970677375793457,
+ "learning_rate": 7.588754785248578e-06,
+ "loss": 0.102,
+ "step": 21660
+ },
+ {
+ "epoch": 59.02179836512261,
+ "grad_norm": 1.7048373222351074,
+ "learning_rate": 7.587898338725887e-06,
+ "loss": 0.0221,
+ "step": 21661
+ },
+ {
+ "epoch": 59.02452316076294,
+ "grad_norm": 2.0966572761535645,
+ "learning_rate": 7.587041910988571e-06,
+ "loss": 0.0253,
+ "step": 21662
+ },
+ {
+ "epoch": 59.02724795640327,
+ "grad_norm": 1.4147800207138062,
+ "learning_rate": 7.586185502043291e-06,
+ "loss": 0.0316,
+ "step": 21663
+ },
+ {
+ "epoch": 59.0299727520436,
+ "grad_norm": 3.467154026031494,
+ "learning_rate": 7.5853291118967236e-06,
+ "loss": 0.0404,
+ "step": 21664
+ },
+ {
+ "epoch": 59.032697547683924,
+ "grad_norm": 1.4367470741271973,
+ "learning_rate": 7.584472740555533e-06,
+ "loss": 0.033,
+ "step": 21665
+ },
+ {
+ "epoch": 59.03542234332425,
+ "grad_norm": 1.7864084243774414,
+ "learning_rate": 7.583616388026392e-06,
+ "loss": 0.0248,
+ "step": 21666
+ },
+ {
+ "epoch": 59.038147138964575,
+ "grad_norm": 2.470409870147705,
+ "learning_rate": 7.582760054315968e-06,
+ "loss": 0.059,
+ "step": 21667
+ },
+ {
+ "epoch": 59.0408719346049,
+ "grad_norm": 3.305509328842163,
+ "learning_rate": 7.5819037394309326e-06,
+ "loss": 0.0592,
+ "step": 21668
+ },
+ {
+ "epoch": 59.043596730245234,
+ "grad_norm": 1.8170075416564941,
+ "learning_rate": 7.581047443377951e-06,
+ "loss": 0.0175,
+ "step": 21669
+ },
+ {
+ "epoch": 59.04632152588556,
+ "grad_norm": 3.4434096813201904,
+ "learning_rate": 7.580191166163696e-06,
+ "loss": 0.0534,
+ "step": 21670
+ },
+ {
+ "epoch": 59.049046321525886,
+ "grad_norm": 4.031411170959473,
+ "learning_rate": 7.5793349077948296e-06,
+ "loss": 0.0417,
+ "step": 21671
+ },
+ {
+ "epoch": 59.05177111716621,
+ "grad_norm": 2.7785699367523193,
+ "learning_rate": 7.578478668278029e-06,
+ "loss": 0.0323,
+ "step": 21672
+ },
+ {
+ "epoch": 59.05449591280654,
+ "grad_norm": 2.361947536468506,
+ "learning_rate": 7.5776224476199585e-06,
+ "loss": 0.0846,
+ "step": 21673
+ },
+ {
+ "epoch": 59.05722070844686,
+ "grad_norm": 2.4240171909332275,
+ "learning_rate": 7.576766245827285e-06,
+ "loss": 0.0247,
+ "step": 21674
+ },
+ {
+ "epoch": 59.059945504087196,
+ "grad_norm": 1.8656425476074219,
+ "learning_rate": 7.5759100629066785e-06,
+ "loss": 0.0826,
+ "step": 21675
+ },
+ {
+ "epoch": 59.06267029972752,
+ "grad_norm": 1.8975636959075928,
+ "learning_rate": 7.575053898864808e-06,
+ "loss": 0.0323,
+ "step": 21676
+ },
+ {
+ "epoch": 59.06539509536785,
+ "grad_norm": 3.3828704357147217,
+ "learning_rate": 7.574197753708339e-06,
+ "loss": 0.1482,
+ "step": 21677
+ },
+ {
+ "epoch": 59.06811989100817,
+ "grad_norm": 2.258082389831543,
+ "learning_rate": 7.57334162744394e-06,
+ "loss": 0.0463,
+ "step": 21678
+ },
+ {
+ "epoch": 59.0708446866485,
+ "grad_norm": 2.904008626937866,
+ "learning_rate": 7.572485520078277e-06,
+ "loss": 0.0305,
+ "step": 21679
+ },
+ {
+ "epoch": 59.073569482288825,
+ "grad_norm": 2.0173609256744385,
+ "learning_rate": 7.571629431618021e-06,
+ "loss": 0.1828,
+ "step": 21680
+ },
+ {
+ "epoch": 59.07629427792916,
+ "grad_norm": 1.7837131023406982,
+ "learning_rate": 7.570773362069835e-06,
+ "loss": 0.0215,
+ "step": 21681
+ },
+ {
+ "epoch": 59.079019073569484,
+ "grad_norm": 1.9205653667449951,
+ "learning_rate": 7.569917311440389e-06,
+ "loss": 0.0331,
+ "step": 21682
+ },
+ {
+ "epoch": 59.08174386920981,
+ "grad_norm": 2.947437047958374,
+ "learning_rate": 7.569061279736347e-06,
+ "loss": 0.0459,
+ "step": 21683
+ },
+ {
+ "epoch": 59.084468664850135,
+ "grad_norm": 2.205134391784668,
+ "learning_rate": 7.568205266964379e-06,
+ "loss": 0.0502,
+ "step": 21684
+ },
+ {
+ "epoch": 59.08719346049046,
+ "grad_norm": 3.9281039237976074,
+ "learning_rate": 7.567349273131149e-06,
+ "loss": 0.1591,
+ "step": 21685
+ },
+ {
+ "epoch": 59.08991825613079,
+ "grad_norm": 3.876446008682251,
+ "learning_rate": 7.566493298243327e-06,
+ "loss": 0.049,
+ "step": 21686
+ },
+ {
+ "epoch": 59.09264305177112,
+ "grad_norm": 1.882807970046997,
+ "learning_rate": 7.565637342307574e-06,
+ "loss": 0.023,
+ "step": 21687
+ },
+ {
+ "epoch": 59.095367847411445,
+ "grad_norm": 2.20778489112854,
+ "learning_rate": 7.5647814053305565e-06,
+ "loss": 0.0277,
+ "step": 21688
+ },
+ {
+ "epoch": 59.09809264305177,
+ "grad_norm": 2.3415703773498535,
+ "learning_rate": 7.563925487318948e-06,
+ "loss": 0.0268,
+ "step": 21689
+ },
+ {
+ "epoch": 59.1008174386921,
+ "grad_norm": 2.1389691829681396,
+ "learning_rate": 7.563069588279407e-06,
+ "loss": 0.0296,
+ "step": 21690
+ },
+ {
+ "epoch": 59.10354223433242,
+ "grad_norm": 2.403531551361084,
+ "learning_rate": 7.562213708218604e-06,
+ "loss": 0.0358,
+ "step": 21691
+ },
+ {
+ "epoch": 59.10626702997275,
+ "grad_norm": 1.8819376230239868,
+ "learning_rate": 7.5613578471432e-06,
+ "loss": 0.0176,
+ "step": 21692
+ },
+ {
+ "epoch": 59.10899182561308,
+ "grad_norm": 2.699406623840332,
+ "learning_rate": 7.560502005059864e-06,
+ "loss": 0.0249,
+ "step": 21693
+ },
+ {
+ "epoch": 59.11171662125341,
+ "grad_norm": 3.3741750717163086,
+ "learning_rate": 7.559646181975259e-06,
+ "loss": 0.1091,
+ "step": 21694
+ },
+ {
+ "epoch": 59.11444141689373,
+ "grad_norm": 2.4280331134796143,
+ "learning_rate": 7.558790377896052e-06,
+ "loss": 0.0534,
+ "step": 21695
+ },
+ {
+ "epoch": 59.11716621253406,
+ "grad_norm": 3.056077003479004,
+ "learning_rate": 7.557934592828906e-06,
+ "loss": 0.0437,
+ "step": 21696
+ },
+ {
+ "epoch": 59.119891008174385,
+ "grad_norm": 1.6542754173278809,
+ "learning_rate": 7.557078826780488e-06,
+ "loss": 0.1254,
+ "step": 21697
+ },
+ {
+ "epoch": 59.12261580381471,
+ "grad_norm": 2.552506923675537,
+ "learning_rate": 7.5562230797574585e-06,
+ "loss": 0.0361,
+ "step": 21698
+ },
+ {
+ "epoch": 59.12534059945504,
+ "grad_norm": 1.8527368307113647,
+ "learning_rate": 7.555367351766487e-06,
+ "loss": 0.0266,
+ "step": 21699
+ },
+ {
+ "epoch": 59.12806539509537,
+ "grad_norm": 2.059629440307617,
+ "learning_rate": 7.5545116428142375e-06,
+ "loss": 0.0279,
+ "step": 21700
+ },
+ {
+ "epoch": 59.130790190735695,
+ "grad_norm": 1.8156847953796387,
+ "learning_rate": 7.553655952907369e-06,
+ "loss": 0.0226,
+ "step": 21701
+ },
+ {
+ "epoch": 59.13351498637602,
+ "grad_norm": 1.524317741394043,
+ "learning_rate": 7.552800282052552e-06,
+ "loss": 0.0233,
+ "step": 21702
+ },
+ {
+ "epoch": 59.13623978201635,
+ "grad_norm": 2.0314292907714844,
+ "learning_rate": 7.551944630256447e-06,
+ "loss": 0.0714,
+ "step": 21703
+ },
+ {
+ "epoch": 59.13896457765667,
+ "grad_norm": 1.7619796991348267,
+ "learning_rate": 7.55108899752572e-06,
+ "loss": 0.1937,
+ "step": 21704
+ },
+ {
+ "epoch": 59.141689373297005,
+ "grad_norm": 2.8857500553131104,
+ "learning_rate": 7.5502333838670295e-06,
+ "loss": 0.202,
+ "step": 21705
+ },
+ {
+ "epoch": 59.14441416893733,
+ "grad_norm": 2.187293529510498,
+ "learning_rate": 7.5493777892870416e-06,
+ "loss": 0.0479,
+ "step": 21706
+ },
+ {
+ "epoch": 59.14713896457766,
+ "grad_norm": 1.5934306383132935,
+ "learning_rate": 7.548522213792424e-06,
+ "loss": 0.0261,
+ "step": 21707
+ },
+ {
+ "epoch": 59.14986376021798,
+ "grad_norm": 2.3120357990264893,
+ "learning_rate": 7.547666657389836e-06,
+ "loss": 0.0726,
+ "step": 21708
+ },
+ {
+ "epoch": 59.15258855585831,
+ "grad_norm": 2.1427853107452393,
+ "learning_rate": 7.546811120085942e-06,
+ "loss": 0.0181,
+ "step": 21709
+ },
+ {
+ "epoch": 59.155313351498634,
+ "grad_norm": 1.6967788934707642,
+ "learning_rate": 7.545955601887402e-06,
+ "loss": 0.0274,
+ "step": 21710
+ },
+ {
+ "epoch": 59.15803814713897,
+ "grad_norm": 2.5138769149780273,
+ "learning_rate": 7.545100102800883e-06,
+ "loss": 0.0586,
+ "step": 21711
+ },
+ {
+ "epoch": 59.16076294277929,
+ "grad_norm": 5.631899356842041,
+ "learning_rate": 7.544244622833043e-06,
+ "loss": 0.0226,
+ "step": 21712
+ },
+ {
+ "epoch": 59.16348773841962,
+ "grad_norm": 1.8880447149276733,
+ "learning_rate": 7.5433891619905496e-06,
+ "loss": 0.0324,
+ "step": 21713
+ },
+ {
+ "epoch": 59.166212534059945,
+ "grad_norm": 2.2018380165100098,
+ "learning_rate": 7.542533720280061e-06,
+ "loss": 0.0321,
+ "step": 21714
+ },
+ {
+ "epoch": 59.16893732970027,
+ "grad_norm": 4.066023826599121,
+ "learning_rate": 7.541678297708241e-06,
+ "loss": 0.0554,
+ "step": 21715
+ },
+ {
+ "epoch": 59.171662125340596,
+ "grad_norm": 1.3196372985839844,
+ "learning_rate": 7.5408228942817495e-06,
+ "loss": 0.0193,
+ "step": 21716
+ },
+ {
+ "epoch": 59.17438692098093,
+ "grad_norm": 2.501569986343384,
+ "learning_rate": 7.539967510007253e-06,
+ "loss": 0.0366,
+ "step": 21717
+ },
+ {
+ "epoch": 59.177111716621255,
+ "grad_norm": 2.121246337890625,
+ "learning_rate": 7.5391121448914085e-06,
+ "loss": 0.1714,
+ "step": 21718
+ },
+ {
+ "epoch": 59.17983651226158,
+ "grad_norm": 1.857397198677063,
+ "learning_rate": 7.5382567989408805e-06,
+ "loss": 0.074,
+ "step": 21719
+ },
+ {
+ "epoch": 59.182561307901906,
+ "grad_norm": 2.777033805847168,
+ "learning_rate": 7.537401472162328e-06,
+ "loss": 0.0177,
+ "step": 21720
+ },
+ {
+ "epoch": 59.18528610354223,
+ "grad_norm": 3.018859624862671,
+ "learning_rate": 7.536546164562413e-06,
+ "loss": 0.0365,
+ "step": 21721
+ },
+ {
+ "epoch": 59.18801089918256,
+ "grad_norm": 3.611368179321289,
+ "learning_rate": 7.5356908761478e-06,
+ "loss": 0.076,
+ "step": 21722
+ },
+ {
+ "epoch": 59.19073569482289,
+ "grad_norm": 2.981947422027588,
+ "learning_rate": 7.534835606925141e-06,
+ "loss": 0.0831,
+ "step": 21723
+ },
+ {
+ "epoch": 59.19346049046322,
+ "grad_norm": 1.4143248796463013,
+ "learning_rate": 7.533980356901107e-06,
+ "loss": 0.0221,
+ "step": 21724
+ },
+ {
+ "epoch": 59.19618528610354,
+ "grad_norm": 2.3894147872924805,
+ "learning_rate": 7.533125126082357e-06,
+ "loss": 0.0398,
+ "step": 21725
+ },
+ {
+ "epoch": 59.19891008174387,
+ "grad_norm": 2.0235440731048584,
+ "learning_rate": 7.532269914475546e-06,
+ "loss": 0.2097,
+ "step": 21726
+ },
+ {
+ "epoch": 59.201634877384194,
+ "grad_norm": 2.3154497146606445,
+ "learning_rate": 7.53141472208734e-06,
+ "loss": 0.0301,
+ "step": 21727
+ },
+ {
+ "epoch": 59.20435967302452,
+ "grad_norm": 2.515028715133667,
+ "learning_rate": 7.530559548924395e-06,
+ "loss": 0.0354,
+ "step": 21728
+ },
+ {
+ "epoch": 59.20708446866485,
+ "grad_norm": 1.8636095523834229,
+ "learning_rate": 7.529704394993374e-06,
+ "loss": 0.0338,
+ "step": 21729
+ },
+ {
+ "epoch": 59.20980926430518,
+ "grad_norm": 1.7882027626037598,
+ "learning_rate": 7.528849260300933e-06,
+ "loss": 0.0257,
+ "step": 21730
+ },
+ {
+ "epoch": 59.212534059945504,
+ "grad_norm": 2.289390802383423,
+ "learning_rate": 7.527994144853738e-06,
+ "loss": 0.1761,
+ "step": 21731
+ },
+ {
+ "epoch": 59.21525885558583,
+ "grad_norm": 1.88228178024292,
+ "learning_rate": 7.5271390486584416e-06,
+ "loss": 0.0305,
+ "step": 21732
+ },
+ {
+ "epoch": 59.217983651226156,
+ "grad_norm": 2.8658719062805176,
+ "learning_rate": 7.526283971721711e-06,
+ "loss": 0.1824,
+ "step": 21733
+ },
+ {
+ "epoch": 59.22070844686648,
+ "grad_norm": 2.3882949352264404,
+ "learning_rate": 7.5254289140501965e-06,
+ "loss": 0.1088,
+ "step": 21734
+ },
+ {
+ "epoch": 59.223433242506815,
+ "grad_norm": 2.0557217597961426,
+ "learning_rate": 7.524573875650565e-06,
+ "loss": 0.0245,
+ "step": 21735
+ },
+ {
+ "epoch": 59.22615803814714,
+ "grad_norm": 34.44456481933594,
+ "learning_rate": 7.523718856529471e-06,
+ "loss": 0.0334,
+ "step": 21736
+ },
+ {
+ "epoch": 59.228882833787466,
+ "grad_norm": 2.2586421966552734,
+ "learning_rate": 7.522863856693577e-06,
+ "loss": 0.2148,
+ "step": 21737
+ },
+ {
+ "epoch": 59.23160762942779,
+ "grad_norm": 1.897787094116211,
+ "learning_rate": 7.522008876149537e-06,
+ "loss": 0.0381,
+ "step": 21738
+ },
+ {
+ "epoch": 59.23433242506812,
+ "grad_norm": 2.4473376274108887,
+ "learning_rate": 7.521153914904013e-06,
+ "loss": 0.0431,
+ "step": 21739
+ },
+ {
+ "epoch": 59.237057220708444,
+ "grad_norm": 6.197749614715576,
+ "learning_rate": 7.520298972963657e-06,
+ "loss": 0.0815,
+ "step": 21740
+ },
+ {
+ "epoch": 59.23978201634878,
+ "grad_norm": 1.5630625486373901,
+ "learning_rate": 7.51944405033514e-06,
+ "loss": 0.027,
+ "step": 21741
+ },
+ {
+ "epoch": 59.2425068119891,
+ "grad_norm": 1.9694435596466064,
+ "learning_rate": 7.51858914702511e-06,
+ "loss": 0.0853,
+ "step": 21742
+ },
+ {
+ "epoch": 59.24523160762943,
+ "grad_norm": 2.6040866374969482,
+ "learning_rate": 7.517734263040232e-06,
+ "loss": 0.0362,
+ "step": 21743
+ },
+ {
+ "epoch": 59.247956403269754,
+ "grad_norm": 2.521496295928955,
+ "learning_rate": 7.516879398387156e-06,
+ "loss": 0.0346,
+ "step": 21744
+ },
+ {
+ "epoch": 59.25068119891008,
+ "grad_norm": 1.9004883766174316,
+ "learning_rate": 7.516024553072545e-06,
+ "loss": 0.0169,
+ "step": 21745
+ },
+ {
+ "epoch": 59.253405994550405,
+ "grad_norm": 2.4515109062194824,
+ "learning_rate": 7.515169727103052e-06,
+ "loss": 0.0608,
+ "step": 21746
+ },
+ {
+ "epoch": 59.25613079019074,
+ "grad_norm": 3.089179277420044,
+ "learning_rate": 7.514314920485341e-06,
+ "loss": 0.1098,
+ "step": 21747
+ },
+ {
+ "epoch": 59.258855585831064,
+ "grad_norm": 1.7746987342834473,
+ "learning_rate": 7.513460133226062e-06,
+ "loss": 0.0303,
+ "step": 21748
+ },
+ {
+ "epoch": 59.26158038147139,
+ "grad_norm": 2.47904896736145,
+ "learning_rate": 7.512605365331878e-06,
+ "loss": 0.0515,
+ "step": 21749
+ },
+ {
+ "epoch": 59.264305177111716,
+ "grad_norm": 1.8044254779815674,
+ "learning_rate": 7.511750616809443e-06,
+ "loss": 0.0299,
+ "step": 21750
+ },
+ {
+ "epoch": 59.26702997275204,
+ "grad_norm": 3.2225935459136963,
+ "learning_rate": 7.510895887665413e-06,
+ "loss": 0.0855,
+ "step": 21751
+ },
+ {
+ "epoch": 59.26975476839237,
+ "grad_norm": 1.439475178718567,
+ "learning_rate": 7.510041177906446e-06,
+ "loss": 0.0274,
+ "step": 21752
+ },
+ {
+ "epoch": 59.2724795640327,
+ "grad_norm": 2.2381792068481445,
+ "learning_rate": 7.5091864875392e-06,
+ "loss": 0.0993,
+ "step": 21753
+ },
+ {
+ "epoch": 59.275204359673026,
+ "grad_norm": 2.9947123527526855,
+ "learning_rate": 7.508331816570327e-06,
+ "loss": 0.04,
+ "step": 21754
+ },
+ {
+ "epoch": 59.27792915531335,
+ "grad_norm": 1.822904348373413,
+ "learning_rate": 7.507477165006487e-06,
+ "loss": 0.0264,
+ "step": 21755
+ },
+ {
+ "epoch": 59.28065395095368,
+ "grad_norm": 2.8858768939971924,
+ "learning_rate": 7.506622532854333e-06,
+ "loss": 0.1585,
+ "step": 21756
+ },
+ {
+ "epoch": 59.283378746594,
+ "grad_norm": 2.2321293354034424,
+ "learning_rate": 7.50576792012052e-06,
+ "loss": 0.0226,
+ "step": 21757
+ },
+ {
+ "epoch": 59.28610354223433,
+ "grad_norm": 2.019930124282837,
+ "learning_rate": 7.504913326811711e-06,
+ "loss": 0.0398,
+ "step": 21758
+ },
+ {
+ "epoch": 59.28882833787466,
+ "grad_norm": 2.636875867843628,
+ "learning_rate": 7.504058752934555e-06,
+ "loss": 0.0269,
+ "step": 21759
+ },
+ {
+ "epoch": 59.29155313351499,
+ "grad_norm": 2.342794179916382,
+ "learning_rate": 7.5032041984957106e-06,
+ "loss": 0.0275,
+ "step": 21760
+ },
+ {
+ "epoch": 59.294277929155314,
+ "grad_norm": 2.0334041118621826,
+ "learning_rate": 7.502349663501829e-06,
+ "loss": 0.0434,
+ "step": 21761
+ },
+ {
+ "epoch": 59.29700272479564,
+ "grad_norm": 2.2521913051605225,
+ "learning_rate": 7.5014951479595684e-06,
+ "loss": 0.0218,
+ "step": 21762
+ },
+ {
+ "epoch": 59.299727520435965,
+ "grad_norm": 2.2697505950927734,
+ "learning_rate": 7.5006406518755856e-06,
+ "loss": 0.0273,
+ "step": 21763
+ },
+ {
+ "epoch": 59.30245231607629,
+ "grad_norm": 2.5380048751831055,
+ "learning_rate": 7.499786175256531e-06,
+ "loss": 0.0526,
+ "step": 21764
+ },
+ {
+ "epoch": 59.305177111716624,
+ "grad_norm": 2.1708860397338867,
+ "learning_rate": 7.498931718109063e-06,
+ "loss": 0.0752,
+ "step": 21765
+ },
+ {
+ "epoch": 59.30790190735695,
+ "grad_norm": 2.9677538871765137,
+ "learning_rate": 7.49807728043983e-06,
+ "loss": 0.0352,
+ "step": 21766
+ },
+ {
+ "epoch": 59.310626702997276,
+ "grad_norm": 2.056480884552002,
+ "learning_rate": 7.497222862255496e-06,
+ "loss": 0.0311,
+ "step": 21767
+ },
+ {
+ "epoch": 59.3133514986376,
+ "grad_norm": 3.494718313217163,
+ "learning_rate": 7.496368463562706e-06,
+ "loss": 0.0524,
+ "step": 21768
+ },
+ {
+ "epoch": 59.31607629427793,
+ "grad_norm": 3.077859401702881,
+ "learning_rate": 7.495514084368121e-06,
+ "loss": 0.0269,
+ "step": 21769
+ },
+ {
+ "epoch": 59.31880108991825,
+ "grad_norm": 1.8814605474472046,
+ "learning_rate": 7.494659724678388e-06,
+ "loss": 0.0227,
+ "step": 21770
+ },
+ {
+ "epoch": 59.321525885558586,
+ "grad_norm": 4.260807991027832,
+ "learning_rate": 7.493805384500168e-06,
+ "loss": 0.0502,
+ "step": 21771
+ },
+ {
+ "epoch": 59.32425068119891,
+ "grad_norm": 3.393765926361084,
+ "learning_rate": 7.492951063840109e-06,
+ "loss": 0.0491,
+ "step": 21772
+ },
+ {
+ "epoch": 59.32697547683924,
+ "grad_norm": 1.514844298362732,
+ "learning_rate": 7.492096762704868e-06,
+ "loss": 0.1565,
+ "step": 21773
+ },
+ {
+ "epoch": 59.32970027247956,
+ "grad_norm": 1.7522764205932617,
+ "learning_rate": 7.491242481101091e-06,
+ "loss": 0.0275,
+ "step": 21774
+ },
+ {
+ "epoch": 59.33242506811989,
+ "grad_norm": 2.0079171657562256,
+ "learning_rate": 7.490388219035441e-06,
+ "loss": 0.0957,
+ "step": 21775
+ },
+ {
+ "epoch": 59.335149863760215,
+ "grad_norm": 2.0875420570373535,
+ "learning_rate": 7.489533976514568e-06,
+ "loss": 0.0214,
+ "step": 21776
+ },
+ {
+ "epoch": 59.33787465940055,
+ "grad_norm": 1.6688698530197144,
+ "learning_rate": 7.488679753545121e-06,
+ "loss": 0.0294,
+ "step": 21777
+ },
+ {
+ "epoch": 59.34059945504087,
+ "grad_norm": 1.9747960567474365,
+ "learning_rate": 7.487825550133759e-06,
+ "loss": 0.0192,
+ "step": 21778
+ },
+ {
+ "epoch": 59.3433242506812,
+ "grad_norm": 2.8721957206726074,
+ "learning_rate": 7.486971366287128e-06,
+ "loss": 0.1219,
+ "step": 21779
+ },
+ {
+ "epoch": 59.346049046321525,
+ "grad_norm": 1.8291996717453003,
+ "learning_rate": 7.4861172020118844e-06,
+ "loss": 0.0364,
+ "step": 21780
+ },
+ {
+ "epoch": 59.34877384196185,
+ "grad_norm": 2.1164770126342773,
+ "learning_rate": 7.485263057314678e-06,
+ "loss": 0.0409,
+ "step": 21781
+ },
+ {
+ "epoch": 59.35149863760218,
+ "grad_norm": 2.4804365634918213,
+ "learning_rate": 7.484408932202163e-06,
+ "loss": 0.0828,
+ "step": 21782
+ },
+ {
+ "epoch": 59.35422343324251,
+ "grad_norm": 1.9303582906723022,
+ "learning_rate": 7.483554826680989e-06,
+ "loss": 0.0722,
+ "step": 21783
+ },
+ {
+ "epoch": 59.356948228882835,
+ "grad_norm": 3.5324482917785645,
+ "learning_rate": 7.482700740757811e-06,
+ "loss": 0.08,
+ "step": 21784
+ },
+ {
+ "epoch": 59.35967302452316,
+ "grad_norm": 2.94490909576416,
+ "learning_rate": 7.481846674439276e-06,
+ "loss": 0.0396,
+ "step": 21785
+ },
+ {
+ "epoch": 59.36239782016349,
+ "grad_norm": 2.2757370471954346,
+ "learning_rate": 7.480992627732038e-06,
+ "loss": 0.0719,
+ "step": 21786
+ },
+ {
+ "epoch": 59.36512261580381,
+ "grad_norm": 2.4897515773773193,
+ "learning_rate": 7.4801386006427515e-06,
+ "loss": 0.0274,
+ "step": 21787
+ },
+ {
+ "epoch": 59.36784741144414,
+ "grad_norm": 3.403836488723755,
+ "learning_rate": 7.479284593178061e-06,
+ "loss": 0.0595,
+ "step": 21788
+ },
+ {
+ "epoch": 59.37057220708447,
+ "grad_norm": 1.974462866783142,
+ "learning_rate": 7.478430605344624e-06,
+ "loss": 0.0676,
+ "step": 21789
+ },
+ {
+ "epoch": 59.3732970027248,
+ "grad_norm": 4.413393974304199,
+ "learning_rate": 7.477576637149086e-06,
+ "loss": 0.0595,
+ "step": 21790
+ },
+ {
+ "epoch": 59.37602179836512,
+ "grad_norm": 2.0811140537261963,
+ "learning_rate": 7.476722688598102e-06,
+ "loss": 0.0281,
+ "step": 21791
+ },
+ {
+ "epoch": 59.37874659400545,
+ "grad_norm": 2.2042741775512695,
+ "learning_rate": 7.475868759698317e-06,
+ "loss": 0.0945,
+ "step": 21792
+ },
+ {
+ "epoch": 59.381471389645775,
+ "grad_norm": 1.9813995361328125,
+ "learning_rate": 7.4750148504563855e-06,
+ "loss": 0.0347,
+ "step": 21793
+ },
+ {
+ "epoch": 59.3841961852861,
+ "grad_norm": 2.2402453422546387,
+ "learning_rate": 7.47416096087896e-06,
+ "loss": 0.1522,
+ "step": 21794
+ },
+ {
+ "epoch": 59.38692098092643,
+ "grad_norm": 2.1699390411376953,
+ "learning_rate": 7.473307090972685e-06,
+ "loss": 0.0247,
+ "step": 21795
+ },
+ {
+ "epoch": 59.38964577656676,
+ "grad_norm": 2.7793807983398438,
+ "learning_rate": 7.4724532407442165e-06,
+ "loss": 0.0521,
+ "step": 21796
+ },
+ {
+ "epoch": 59.392370572207085,
+ "grad_norm": 2.6261372566223145,
+ "learning_rate": 7.471599410200198e-06,
+ "loss": 0.1399,
+ "step": 21797
+ },
+ {
+ "epoch": 59.39509536784741,
+ "grad_norm": 1.8876782655715942,
+ "learning_rate": 7.470745599347285e-06,
+ "loss": 0.0243,
+ "step": 21798
+ },
+ {
+ "epoch": 59.39782016348774,
+ "grad_norm": 2.5352022647857666,
+ "learning_rate": 7.46989180819212e-06,
+ "loss": 0.0419,
+ "step": 21799
+ },
+ {
+ "epoch": 59.40054495912806,
+ "grad_norm": 2.76574444770813,
+ "learning_rate": 7.469038036741358e-06,
+ "loss": 0.0514,
+ "step": 21800
+ },
+ {
+ "epoch": 59.403269754768395,
+ "grad_norm": 2.4494149684906006,
+ "learning_rate": 7.468184285001645e-06,
+ "loss": 0.135,
+ "step": 21801
+ },
+ {
+ "epoch": 59.40599455040872,
+ "grad_norm": 2.1252002716064453,
+ "learning_rate": 7.467330552979635e-06,
+ "loss": 0.0278,
+ "step": 21802
+ },
+ {
+ "epoch": 59.40871934604905,
+ "grad_norm": 2.2362124919891357,
+ "learning_rate": 7.4664768406819685e-06,
+ "loss": 0.0637,
+ "step": 21803
+ },
+ {
+ "epoch": 59.41144414168937,
+ "grad_norm": 1.998886227607727,
+ "learning_rate": 7.465623148115302e-06,
+ "loss": 0.0324,
+ "step": 21804
+ },
+ {
+ "epoch": 59.4141689373297,
+ "grad_norm": 7.556716442108154,
+ "learning_rate": 7.46476947528628e-06,
+ "loss": 0.0399,
+ "step": 21805
+ },
+ {
+ "epoch": 59.416893732970024,
+ "grad_norm": 1.8062561750411987,
+ "learning_rate": 7.463915822201552e-06,
+ "loss": 0.0226,
+ "step": 21806
+ },
+ {
+ "epoch": 59.41961852861036,
+ "grad_norm": 2.5942234992980957,
+ "learning_rate": 7.463062188867764e-06,
+ "loss": 0.0384,
+ "step": 21807
+ },
+ {
+ "epoch": 59.42234332425068,
+ "grad_norm": 2.2429847717285156,
+ "learning_rate": 7.462208575291567e-06,
+ "loss": 0.0243,
+ "step": 21808
+ },
+ {
+ "epoch": 59.42506811989101,
+ "grad_norm": 2.326117753982544,
+ "learning_rate": 7.461354981479605e-06,
+ "loss": 0.0641,
+ "step": 21809
+ },
+ {
+ "epoch": 59.427792915531334,
+ "grad_norm": 2.814317226409912,
+ "learning_rate": 7.460501407438533e-06,
+ "loss": 0.0347,
+ "step": 21810
+ },
+ {
+ "epoch": 59.43051771117166,
+ "grad_norm": 1.7528518438339233,
+ "learning_rate": 7.459647853174993e-06,
+ "loss": 0.0325,
+ "step": 21811
+ },
+ {
+ "epoch": 59.433242506811986,
+ "grad_norm": 1.6542097330093384,
+ "learning_rate": 7.458794318695635e-06,
+ "loss": 0.0205,
+ "step": 21812
+ },
+ {
+ "epoch": 59.43596730245232,
+ "grad_norm": 2.470125675201416,
+ "learning_rate": 7.457940804007104e-06,
+ "loss": 0.0457,
+ "step": 21813
+ },
+ {
+ "epoch": 59.438692098092645,
+ "grad_norm": 2.795057773590088,
+ "learning_rate": 7.457087309116049e-06,
+ "loss": 0.0355,
+ "step": 21814
+ },
+ {
+ "epoch": 59.44141689373297,
+ "grad_norm": 3.1185057163238525,
+ "learning_rate": 7.456233834029114e-06,
+ "loss": 0.0346,
+ "step": 21815
+ },
+ {
+ "epoch": 59.444141689373296,
+ "grad_norm": 1.9835617542266846,
+ "learning_rate": 7.455380378752951e-06,
+ "loss": 0.0233,
+ "step": 21816
+ },
+ {
+ "epoch": 59.44686648501362,
+ "grad_norm": 1.7593142986297607,
+ "learning_rate": 7.454526943294201e-06,
+ "loss": 0.0198,
+ "step": 21817
+ },
+ {
+ "epoch": 59.44959128065395,
+ "grad_norm": 2.2032978534698486,
+ "learning_rate": 7.4536735276595165e-06,
+ "loss": 0.0358,
+ "step": 21818
+ },
+ {
+ "epoch": 59.45231607629428,
+ "grad_norm": 2.326097249984741,
+ "learning_rate": 7.452820131855538e-06,
+ "loss": 0.0968,
+ "step": 21819
+ },
+ {
+ "epoch": 59.45504087193461,
+ "grad_norm": 1.4510927200317383,
+ "learning_rate": 7.451966755888915e-06,
+ "loss": 0.0186,
+ "step": 21820
+ },
+ {
+ "epoch": 59.45776566757493,
+ "grad_norm": 2.821772575378418,
+ "learning_rate": 7.451113399766292e-06,
+ "loss": 0.1235,
+ "step": 21821
+ },
+ {
+ "epoch": 59.46049046321526,
+ "grad_norm": 2.1778461933135986,
+ "learning_rate": 7.450260063494319e-06,
+ "loss": 0.0316,
+ "step": 21822
+ },
+ {
+ "epoch": 59.463215258855584,
+ "grad_norm": 2.9232749938964844,
+ "learning_rate": 7.449406747079635e-06,
+ "loss": 0.0549,
+ "step": 21823
+ },
+ {
+ "epoch": 59.46594005449591,
+ "grad_norm": 1.9235742092132568,
+ "learning_rate": 7.448553450528891e-06,
+ "loss": 0.0233,
+ "step": 21824
+ },
+ {
+ "epoch": 59.46866485013624,
+ "grad_norm": 2.499454975128174,
+ "learning_rate": 7.447700173848729e-06,
+ "loss": 0.0363,
+ "step": 21825
+ },
+ {
+ "epoch": 59.47138964577657,
+ "grad_norm": 3.3167004585266113,
+ "learning_rate": 7.446846917045793e-06,
+ "loss": 0.1067,
+ "step": 21826
+ },
+ {
+ "epoch": 59.474114441416894,
+ "grad_norm": 2.012392282485962,
+ "learning_rate": 7.445993680126735e-06,
+ "loss": 0.0369,
+ "step": 21827
+ },
+ {
+ "epoch": 59.47683923705722,
+ "grad_norm": 2.94258189201355,
+ "learning_rate": 7.445140463098197e-06,
+ "loss": 0.0398,
+ "step": 21828
+ },
+ {
+ "epoch": 59.479564032697546,
+ "grad_norm": 2.5311386585235596,
+ "learning_rate": 7.444287265966821e-06,
+ "loss": 0.0404,
+ "step": 21829
+ },
+ {
+ "epoch": 59.48228882833787,
+ "grad_norm": 7.945481777191162,
+ "learning_rate": 7.443434088739256e-06,
+ "loss": 0.0729,
+ "step": 21830
+ },
+ {
+ "epoch": 59.485013623978205,
+ "grad_norm": 2.895001173019409,
+ "learning_rate": 7.442580931422141e-06,
+ "loss": 0.0427,
+ "step": 21831
+ },
+ {
+ "epoch": 59.48773841961853,
+ "grad_norm": 3.152796745300293,
+ "learning_rate": 7.441727794022126e-06,
+ "loss": 0.0423,
+ "step": 21832
+ },
+ {
+ "epoch": 59.490463215258856,
+ "grad_norm": 3.9044687747955322,
+ "learning_rate": 7.44087467654585e-06,
+ "loss": 0.0368,
+ "step": 21833
+ },
+ {
+ "epoch": 59.49318801089918,
+ "grad_norm": 2.628950834274292,
+ "learning_rate": 7.440021578999963e-06,
+ "loss": 0.0818,
+ "step": 21834
+ },
+ {
+ "epoch": 59.49591280653951,
+ "grad_norm": 5.242530822753906,
+ "learning_rate": 7.439168501391103e-06,
+ "loss": 0.1042,
+ "step": 21835
+ },
+ {
+ "epoch": 59.49863760217983,
+ "grad_norm": 1.6679661273956299,
+ "learning_rate": 7.438315443725919e-06,
+ "loss": 0.0186,
+ "step": 21836
+ },
+ {
+ "epoch": 59.50136239782017,
+ "grad_norm": 8.902632713317871,
+ "learning_rate": 7.437462406011048e-06,
+ "loss": 0.0286,
+ "step": 21837
+ },
+ {
+ "epoch": 59.50408719346049,
+ "grad_norm": 1.9723854064941406,
+ "learning_rate": 7.43660938825314e-06,
+ "loss": 0.0288,
+ "step": 21838
+ },
+ {
+ "epoch": 59.50681198910082,
+ "grad_norm": 1.5166420936584473,
+ "learning_rate": 7.435756390458834e-06,
+ "loss": 0.0189,
+ "step": 21839
+ },
+ {
+ "epoch": 59.509536784741144,
+ "grad_norm": 2.0773985385894775,
+ "learning_rate": 7.4349034126347765e-06,
+ "loss": 0.0441,
+ "step": 21840
+ },
+ {
+ "epoch": 59.51226158038147,
+ "grad_norm": 2.3489513397216797,
+ "learning_rate": 7.4340504547876066e-06,
+ "loss": 0.1353,
+ "step": 21841
+ },
+ {
+ "epoch": 59.514986376021795,
+ "grad_norm": 2.4623053073883057,
+ "learning_rate": 7.433197516923971e-06,
+ "loss": 0.0503,
+ "step": 21842
+ },
+ {
+ "epoch": 59.51771117166213,
+ "grad_norm": 1.7211887836456299,
+ "learning_rate": 7.432344599050507e-06,
+ "loss": 0.0385,
+ "step": 21843
+ },
+ {
+ "epoch": 59.520435967302454,
+ "grad_norm": 2.958826780319214,
+ "learning_rate": 7.431491701173862e-06,
+ "loss": 0.0831,
+ "step": 21844
+ },
+ {
+ "epoch": 59.52316076294278,
+ "grad_norm": 2.0321483612060547,
+ "learning_rate": 7.43063882330068e-06,
+ "loss": 0.0376,
+ "step": 21845
+ },
+ {
+ "epoch": 59.525885558583106,
+ "grad_norm": 5.575500011444092,
+ "learning_rate": 7.429785965437597e-06,
+ "loss": 0.0896,
+ "step": 21846
+ },
+ {
+ "epoch": 59.52861035422343,
+ "grad_norm": 2.4931833744049072,
+ "learning_rate": 7.42893312759126e-06,
+ "loss": 0.0897,
+ "step": 21847
+ },
+ {
+ "epoch": 59.53133514986376,
+ "grad_norm": 2.261945962905884,
+ "learning_rate": 7.4280803097683086e-06,
+ "loss": 0.0222,
+ "step": 21848
+ },
+ {
+ "epoch": 59.53405994550409,
+ "grad_norm": 2.390223979949951,
+ "learning_rate": 7.427227511975383e-06,
+ "loss": 0.0479,
+ "step": 21849
+ },
+ {
+ "epoch": 59.536784741144416,
+ "grad_norm": 3.477689027786255,
+ "learning_rate": 7.426374734219131e-06,
+ "loss": 0.0376,
+ "step": 21850
+ },
+ {
+ "epoch": 59.53950953678474,
+ "grad_norm": 3.2214105129241943,
+ "learning_rate": 7.425521976506186e-06,
+ "loss": 0.1196,
+ "step": 21851
+ },
+ {
+ "epoch": 59.54223433242507,
+ "grad_norm": 2.1146652698516846,
+ "learning_rate": 7.424669238843195e-06,
+ "loss": 0.0813,
+ "step": 21852
+ },
+ {
+ "epoch": 59.54495912806539,
+ "grad_norm": 4.040227890014648,
+ "learning_rate": 7.423816521236795e-06,
+ "loss": 0.0552,
+ "step": 21853
+ },
+ {
+ "epoch": 59.54768392370572,
+ "grad_norm": 2.087672472000122,
+ "learning_rate": 7.4229638236936315e-06,
+ "loss": 0.0662,
+ "step": 21854
+ },
+ {
+ "epoch": 59.55040871934605,
+ "grad_norm": 2.5247223377227783,
+ "learning_rate": 7.422111146220342e-06,
+ "loss": 0.0344,
+ "step": 21855
+ },
+ {
+ "epoch": 59.55313351498638,
+ "grad_norm": 3.5763347148895264,
+ "learning_rate": 7.421258488823568e-06,
+ "loss": 0.0437,
+ "step": 21856
+ },
+ {
+ "epoch": 59.555858310626704,
+ "grad_norm": 2.3095576763153076,
+ "learning_rate": 7.42040585150995e-06,
+ "loss": 0.0597,
+ "step": 21857
+ },
+ {
+ "epoch": 59.55858310626703,
+ "grad_norm": 2.2192909717559814,
+ "learning_rate": 7.419553234286129e-06,
+ "loss": 0.0356,
+ "step": 21858
+ },
+ {
+ "epoch": 59.561307901907355,
+ "grad_norm": 2.618168830871582,
+ "learning_rate": 7.418700637158742e-06,
+ "loss": 0.1024,
+ "step": 21859
+ },
+ {
+ "epoch": 59.56403269754768,
+ "grad_norm": 2.2957382202148438,
+ "learning_rate": 7.417848060134429e-06,
+ "loss": 0.1921,
+ "step": 21860
+ },
+ {
+ "epoch": 59.566757493188014,
+ "grad_norm": 2.0045981407165527,
+ "learning_rate": 7.416995503219837e-06,
+ "loss": 0.0258,
+ "step": 21861
+ },
+ {
+ "epoch": 59.56948228882834,
+ "grad_norm": 1.7484928369522095,
+ "learning_rate": 7.416142966421599e-06,
+ "loss": 0.0201,
+ "step": 21862
+ },
+ {
+ "epoch": 59.572207084468666,
+ "grad_norm": 2.30379056930542,
+ "learning_rate": 7.4152904497463575e-06,
+ "loss": 0.1349,
+ "step": 21863
+ },
+ {
+ "epoch": 59.57493188010899,
+ "grad_norm": 2.211411714553833,
+ "learning_rate": 7.414437953200751e-06,
+ "loss": 0.0293,
+ "step": 21864
+ },
+ {
+ "epoch": 59.57765667574932,
+ "grad_norm": 1.8082847595214844,
+ "learning_rate": 7.413585476791418e-06,
+ "loss": 0.1106,
+ "step": 21865
+ },
+ {
+ "epoch": 59.58038147138964,
+ "grad_norm": 1.1785682439804077,
+ "learning_rate": 7.412733020524997e-06,
+ "loss": 0.0199,
+ "step": 21866
+ },
+ {
+ "epoch": 59.583106267029976,
+ "grad_norm": 2.384856700897217,
+ "learning_rate": 7.411880584408131e-06,
+ "loss": 0.1634,
+ "step": 21867
+ },
+ {
+ "epoch": 59.5858310626703,
+ "grad_norm": 4.192415237426758,
+ "learning_rate": 7.4110281684474515e-06,
+ "loss": 0.0722,
+ "step": 21868
+ },
+ {
+ "epoch": 59.58855585831063,
+ "grad_norm": 2.159928560256958,
+ "learning_rate": 7.410175772649605e-06,
+ "loss": 0.0427,
+ "step": 21869
+ },
+ {
+ "epoch": 59.59128065395095,
+ "grad_norm": 2.356689691543579,
+ "learning_rate": 7.4093233970212245e-06,
+ "loss": 0.027,
+ "step": 21870
+ },
+ {
+ "epoch": 59.59400544959128,
+ "grad_norm": 7.3180694580078125,
+ "learning_rate": 7.40847104156895e-06,
+ "loss": 0.0303,
+ "step": 21871
+ },
+ {
+ "epoch": 59.596730245231605,
+ "grad_norm": 1.653246283531189,
+ "learning_rate": 7.40761870629942e-06,
+ "loss": 0.0241,
+ "step": 21872
+ },
+ {
+ "epoch": 59.59945504087194,
+ "grad_norm": 1.9848426580429077,
+ "learning_rate": 7.40676639121927e-06,
+ "loss": 0.1191,
+ "step": 21873
+ },
+ {
+ "epoch": 59.60217983651226,
+ "grad_norm": 2.5631978511810303,
+ "learning_rate": 7.405914096335143e-06,
+ "loss": 0.0596,
+ "step": 21874
+ },
+ {
+ "epoch": 59.60490463215259,
+ "grad_norm": 2.2215967178344727,
+ "learning_rate": 7.405061821653672e-06,
+ "loss": 0.0882,
+ "step": 21875
+ },
+ {
+ "epoch": 59.607629427792915,
+ "grad_norm": 2.388868570327759,
+ "learning_rate": 7.404209567181497e-06,
+ "loss": 0.0912,
+ "step": 21876
+ },
+ {
+ "epoch": 59.61035422343324,
+ "grad_norm": 1.223140835762024,
+ "learning_rate": 7.403357332925251e-06,
+ "loss": 0.0161,
+ "step": 21877
+ },
+ {
+ "epoch": 59.61307901907357,
+ "grad_norm": 2.127819061279297,
+ "learning_rate": 7.402505118891575e-06,
+ "loss": 0.06,
+ "step": 21878
+ },
+ {
+ "epoch": 59.6158038147139,
+ "grad_norm": 8.9898042678833,
+ "learning_rate": 7.40165292508711e-06,
+ "loss": 0.0236,
+ "step": 21879
+ },
+ {
+ "epoch": 59.618528610354225,
+ "grad_norm": 1.7036832571029663,
+ "learning_rate": 7.400800751518484e-06,
+ "loss": 0.0232,
+ "step": 21880
+ },
+ {
+ "epoch": 59.62125340599455,
+ "grad_norm": 2.304771900177002,
+ "learning_rate": 7.399948598192341e-06,
+ "loss": 0.0405,
+ "step": 21881
+ },
+ {
+ "epoch": 59.62397820163488,
+ "grad_norm": 5.674117088317871,
+ "learning_rate": 7.3990964651153144e-06,
+ "loss": 0.0675,
+ "step": 21882
+ },
+ {
+ "epoch": 59.6267029972752,
+ "grad_norm": 2.565638542175293,
+ "learning_rate": 7.39824435229404e-06,
+ "loss": 0.1113,
+ "step": 21883
+ },
+ {
+ "epoch": 59.62942779291553,
+ "grad_norm": 2.3418307304382324,
+ "learning_rate": 7.397392259735155e-06,
+ "loss": 0.0465,
+ "step": 21884
+ },
+ {
+ "epoch": 59.63215258855586,
+ "grad_norm": 1.7469556331634521,
+ "learning_rate": 7.396540187445296e-06,
+ "loss": 0.1315,
+ "step": 21885
+ },
+ {
+ "epoch": 59.63487738419619,
+ "grad_norm": 3.4097797870635986,
+ "learning_rate": 7.395688135431096e-06,
+ "loss": 0.0666,
+ "step": 21886
+ },
+ {
+ "epoch": 59.63760217983651,
+ "grad_norm": 2.1318023204803467,
+ "learning_rate": 7.394836103699197e-06,
+ "loss": 0.0975,
+ "step": 21887
+ },
+ {
+ "epoch": 59.64032697547684,
+ "grad_norm": 2.7355048656463623,
+ "learning_rate": 7.393984092256227e-06,
+ "loss": 0.0317,
+ "step": 21888
+ },
+ {
+ "epoch": 59.643051771117165,
+ "grad_norm": 3.4460079669952393,
+ "learning_rate": 7.393132101108827e-06,
+ "loss": 0.0828,
+ "step": 21889
+ },
+ {
+ "epoch": 59.64577656675749,
+ "grad_norm": 4.025564193725586,
+ "learning_rate": 7.392280130263628e-06,
+ "loss": 0.0258,
+ "step": 21890
+ },
+ {
+ "epoch": 59.64850136239782,
+ "grad_norm": 2.499191999435425,
+ "learning_rate": 7.39142817972727e-06,
+ "loss": 0.0286,
+ "step": 21891
+ },
+ {
+ "epoch": 59.65122615803815,
+ "grad_norm": 1.9381250143051147,
+ "learning_rate": 7.390576249506384e-06,
+ "loss": 0.0162,
+ "step": 21892
+ },
+ {
+ "epoch": 59.653950953678475,
+ "grad_norm": 2.5298750400543213,
+ "learning_rate": 7.389724339607606e-06,
+ "loss": 0.1053,
+ "step": 21893
+ },
+ {
+ "epoch": 59.6566757493188,
+ "grad_norm": 1.6337168216705322,
+ "learning_rate": 7.388872450037572e-06,
+ "loss": 0.1479,
+ "step": 21894
+ },
+ {
+ "epoch": 59.65940054495913,
+ "grad_norm": 4.326839447021484,
+ "learning_rate": 7.38802058080291e-06,
+ "loss": 0.0686,
+ "step": 21895
+ },
+ {
+ "epoch": 59.66212534059945,
+ "grad_norm": 2.1241910457611084,
+ "learning_rate": 7.387168731910264e-06,
+ "loss": 0.1851,
+ "step": 21896
+ },
+ {
+ "epoch": 59.664850136239785,
+ "grad_norm": 1.358424425125122,
+ "learning_rate": 7.386316903366266e-06,
+ "loss": 0.0179,
+ "step": 21897
+ },
+ {
+ "epoch": 59.66757493188011,
+ "grad_norm": 2.4569780826568604,
+ "learning_rate": 7.385465095177544e-06,
+ "loss": 0.035,
+ "step": 21898
+ },
+ {
+ "epoch": 59.67029972752044,
+ "grad_norm": 2.7317845821380615,
+ "learning_rate": 7.384613307350739e-06,
+ "loss": 0.025,
+ "step": 21899
+ },
+ {
+ "epoch": 59.67302452316076,
+ "grad_norm": 3.0000534057617188,
+ "learning_rate": 7.38376153989248e-06,
+ "loss": 0.0543,
+ "step": 21900
+ },
+ {
+ "epoch": 59.67574931880109,
+ "grad_norm": 2.1020166873931885,
+ "learning_rate": 7.382909792809403e-06,
+ "loss": 0.07,
+ "step": 21901
+ },
+ {
+ "epoch": 59.678474114441414,
+ "grad_norm": 2.3242077827453613,
+ "learning_rate": 7.3820580661081375e-06,
+ "loss": 0.0248,
+ "step": 21902
+ },
+ {
+ "epoch": 59.68119891008175,
+ "grad_norm": 3.2344114780426025,
+ "learning_rate": 7.381206359795323e-06,
+ "loss": 0.0396,
+ "step": 21903
+ },
+ {
+ "epoch": 59.68392370572207,
+ "grad_norm": 2.373487710952759,
+ "learning_rate": 7.380354673877585e-06,
+ "loss": 0.1226,
+ "step": 21904
+ },
+ {
+ "epoch": 59.6866485013624,
+ "grad_norm": 4.3078389167785645,
+ "learning_rate": 7.379503008361565e-06,
+ "loss": 0.0582,
+ "step": 21905
+ },
+ {
+ "epoch": 59.689373297002724,
+ "grad_norm": 1.75370192527771,
+ "learning_rate": 7.378651363253888e-06,
+ "loss": 0.0295,
+ "step": 21906
+ },
+ {
+ "epoch": 59.69209809264305,
+ "grad_norm": 2.181732177734375,
+ "learning_rate": 7.377799738561192e-06,
+ "loss": 0.0358,
+ "step": 21907
+ },
+ {
+ "epoch": 59.694822888283376,
+ "grad_norm": 3.2245795726776123,
+ "learning_rate": 7.376948134290105e-06,
+ "loss": 0.0519,
+ "step": 21908
+ },
+ {
+ "epoch": 59.69754768392371,
+ "grad_norm": 2.9284563064575195,
+ "learning_rate": 7.376096550447265e-06,
+ "loss": 0.0451,
+ "step": 21909
+ },
+ {
+ "epoch": 59.700272479564035,
+ "grad_norm": 1.2457709312438965,
+ "learning_rate": 7.375244987039296e-06,
+ "loss": 0.0147,
+ "step": 21910
+ },
+ {
+ "epoch": 59.70299727520436,
+ "grad_norm": 2.8799829483032227,
+ "learning_rate": 7.3743934440728385e-06,
+ "loss": 0.1277,
+ "step": 21911
+ },
+ {
+ "epoch": 59.705722070844686,
+ "grad_norm": 1.7110213041305542,
+ "learning_rate": 7.373541921554515e-06,
+ "loss": 0.0274,
+ "step": 21912
+ },
+ {
+ "epoch": 59.70844686648501,
+ "grad_norm": 9.07761287689209,
+ "learning_rate": 7.3726904194909655e-06,
+ "loss": 0.0821,
+ "step": 21913
+ },
+ {
+ "epoch": 59.71117166212534,
+ "grad_norm": 1.9290190935134888,
+ "learning_rate": 7.3718389378888175e-06,
+ "loss": 0.0477,
+ "step": 21914
+ },
+ {
+ "epoch": 59.71389645776567,
+ "grad_norm": 5.402926445007324,
+ "learning_rate": 7.370987476754707e-06,
+ "loss": 0.0389,
+ "step": 21915
+ },
+ {
+ "epoch": 59.716621253406,
+ "grad_norm": 2.7021677494049072,
+ "learning_rate": 7.370136036095258e-06,
+ "loss": 0.121,
+ "step": 21916
+ },
+ {
+ "epoch": 59.71934604904632,
+ "grad_norm": 2.800405740737915,
+ "learning_rate": 7.369284615917107e-06,
+ "loss": 0.2881,
+ "step": 21917
+ },
+ {
+ "epoch": 59.72207084468665,
+ "grad_norm": 2.1230690479278564,
+ "learning_rate": 7.368433216226881e-06,
+ "loss": 0.0442,
+ "step": 21918
+ },
+ {
+ "epoch": 59.724795640326974,
+ "grad_norm": 1.9035769701004028,
+ "learning_rate": 7.3675818370312145e-06,
+ "loss": 0.0433,
+ "step": 21919
+ },
+ {
+ "epoch": 59.7275204359673,
+ "grad_norm": 5.028702735900879,
+ "learning_rate": 7.3667304783367345e-06,
+ "loss": 0.1832,
+ "step": 21920
+ },
+ {
+ "epoch": 59.73024523160763,
+ "grad_norm": 2.2712175846099854,
+ "learning_rate": 7.365879140150074e-06,
+ "loss": 0.0254,
+ "step": 21921
+ },
+ {
+ "epoch": 59.73297002724796,
+ "grad_norm": 1.5996971130371094,
+ "learning_rate": 7.36502782247786e-06,
+ "loss": 0.0417,
+ "step": 21922
+ },
+ {
+ "epoch": 59.735694822888284,
+ "grad_norm": 4.158304691314697,
+ "learning_rate": 7.364176525326726e-06,
+ "loss": 0.0889,
+ "step": 21923
+ },
+ {
+ "epoch": 59.73841961852861,
+ "grad_norm": 2.893571376800537,
+ "learning_rate": 7.3633252487032995e-06,
+ "loss": 0.0683,
+ "step": 21924
+ },
+ {
+ "epoch": 59.741144414168936,
+ "grad_norm": 2.0534825325012207,
+ "learning_rate": 7.362473992614212e-06,
+ "loss": 0.045,
+ "step": 21925
+ },
+ {
+ "epoch": 59.74386920980926,
+ "grad_norm": 1.9701112508773804,
+ "learning_rate": 7.361622757066092e-06,
+ "loss": 0.0249,
+ "step": 21926
+ },
+ {
+ "epoch": 59.746594005449595,
+ "grad_norm": 2.4702887535095215,
+ "learning_rate": 7.360771542065569e-06,
+ "loss": 0.0422,
+ "step": 21927
+ },
+ {
+ "epoch": 59.74931880108992,
+ "grad_norm": 2.2465932369232178,
+ "learning_rate": 7.3599203476192716e-06,
+ "loss": 0.0444,
+ "step": 21928
+ },
+ {
+ "epoch": 59.752043596730246,
+ "grad_norm": 2.7946159839630127,
+ "learning_rate": 7.3590691737338264e-06,
+ "loss": 0.0838,
+ "step": 21929
+ },
+ {
+ "epoch": 59.75476839237057,
+ "grad_norm": 2.346273899078369,
+ "learning_rate": 7.35821802041587e-06,
+ "loss": 0.0249,
+ "step": 21930
+ },
+ {
+ "epoch": 59.7574931880109,
+ "grad_norm": 2.148380756378174,
+ "learning_rate": 7.357366887672025e-06,
+ "loss": 0.0413,
+ "step": 21931
+ },
+ {
+ "epoch": 59.76021798365122,
+ "grad_norm": 3.067819356918335,
+ "learning_rate": 7.356515775508924e-06,
+ "loss": 0.1455,
+ "step": 21932
+ },
+ {
+ "epoch": 59.762942779291556,
+ "grad_norm": 1.6427249908447266,
+ "learning_rate": 7.35566468393319e-06,
+ "loss": 0.0191,
+ "step": 21933
+ },
+ {
+ "epoch": 59.76566757493188,
+ "grad_norm": 4.7301788330078125,
+ "learning_rate": 7.354813612951457e-06,
+ "loss": 0.0636,
+ "step": 21934
+ },
+ {
+ "epoch": 59.76839237057221,
+ "grad_norm": 2.8582684993743896,
+ "learning_rate": 7.353962562570347e-06,
+ "loss": 0.0255,
+ "step": 21935
+ },
+ {
+ "epoch": 59.771117166212534,
+ "grad_norm": 1.9099048376083374,
+ "learning_rate": 7.353111532796493e-06,
+ "loss": 0.0592,
+ "step": 21936
+ },
+ {
+ "epoch": 59.77384196185286,
+ "grad_norm": 2.729358196258545,
+ "learning_rate": 7.352260523636523e-06,
+ "loss": 0.0657,
+ "step": 21937
+ },
+ {
+ "epoch": 59.776566757493185,
+ "grad_norm": 1.8485515117645264,
+ "learning_rate": 7.351409535097061e-06,
+ "loss": 0.0484,
+ "step": 21938
+ },
+ {
+ "epoch": 59.77929155313352,
+ "grad_norm": 3.2902562618255615,
+ "learning_rate": 7.350558567184738e-06,
+ "loss": 0.0554,
+ "step": 21939
+ },
+ {
+ "epoch": 59.782016348773844,
+ "grad_norm": 3.7358105182647705,
+ "learning_rate": 7.349707619906177e-06,
+ "loss": 0.0496,
+ "step": 21940
+ },
+ {
+ "epoch": 59.78474114441417,
+ "grad_norm": 2.4445977210998535,
+ "learning_rate": 7.348856693268011e-06,
+ "loss": 0.0234,
+ "step": 21941
+ },
+ {
+ "epoch": 59.787465940054496,
+ "grad_norm": 25.688512802124023,
+ "learning_rate": 7.348005787276861e-06,
+ "loss": 0.0241,
+ "step": 21942
+ },
+ {
+ "epoch": 59.79019073569482,
+ "grad_norm": 1.561798095703125,
+ "learning_rate": 7.347154901939358e-06,
+ "loss": 0.0217,
+ "step": 21943
+ },
+ {
+ "epoch": 59.79291553133515,
+ "grad_norm": 1.9557701349258423,
+ "learning_rate": 7.346304037262127e-06,
+ "loss": 0.0166,
+ "step": 21944
+ },
+ {
+ "epoch": 59.79564032697548,
+ "grad_norm": 2.760680913925171,
+ "learning_rate": 7.345453193251795e-06,
+ "loss": 0.0241,
+ "step": 21945
+ },
+ {
+ "epoch": 59.798365122615806,
+ "grad_norm": 1.9964431524276733,
+ "learning_rate": 7.344602369914984e-06,
+ "loss": 0.0443,
+ "step": 21946
+ },
+ {
+ "epoch": 59.80108991825613,
+ "grad_norm": 2.546928644180298,
+ "learning_rate": 7.343751567258328e-06,
+ "loss": 0.1249,
+ "step": 21947
+ },
+ {
+ "epoch": 59.80381471389646,
+ "grad_norm": 2.9672937393188477,
+ "learning_rate": 7.34290078528845e-06,
+ "loss": 0.0678,
+ "step": 21948
+ },
+ {
+ "epoch": 59.80653950953678,
+ "grad_norm": 3.602682113647461,
+ "learning_rate": 7.342050024011973e-06,
+ "loss": 0.067,
+ "step": 21949
+ },
+ {
+ "epoch": 59.80926430517711,
+ "grad_norm": 3.215146780014038,
+ "learning_rate": 7.3411992834355276e-06,
+ "loss": 0.0271,
+ "step": 21950
+ },
+ {
+ "epoch": 59.81198910081744,
+ "grad_norm": 4.0443806648254395,
+ "learning_rate": 7.340348563565735e-06,
+ "loss": 0.0627,
+ "step": 21951
+ },
+ {
+ "epoch": 59.81471389645777,
+ "grad_norm": 1.459176778793335,
+ "learning_rate": 7.339497864409223e-06,
+ "loss": 0.0167,
+ "step": 21952
+ },
+ {
+ "epoch": 59.817438692098094,
+ "grad_norm": 2.047147750854492,
+ "learning_rate": 7.338647185972614e-06,
+ "loss": 0.0215,
+ "step": 21953
+ },
+ {
+ "epoch": 59.82016348773842,
+ "grad_norm": 3.3127541542053223,
+ "learning_rate": 7.337796528262537e-06,
+ "loss": 0.1418,
+ "step": 21954
+ },
+ {
+ "epoch": 59.822888283378745,
+ "grad_norm": 1.926203966140747,
+ "learning_rate": 7.336945891285614e-06,
+ "loss": 0.0214,
+ "step": 21955
+ },
+ {
+ "epoch": 59.82561307901907,
+ "grad_norm": 2.548579216003418,
+ "learning_rate": 7.336095275048474e-06,
+ "loss": 0.101,
+ "step": 21956
+ },
+ {
+ "epoch": 59.828337874659404,
+ "grad_norm": 1.6139768362045288,
+ "learning_rate": 7.3352446795577336e-06,
+ "loss": 0.0977,
+ "step": 21957
+ },
+ {
+ "epoch": 59.83106267029973,
+ "grad_norm": 4.257946491241455,
+ "learning_rate": 7.334394104820024e-06,
+ "loss": 0.0306,
+ "step": 21958
+ },
+ {
+ "epoch": 59.833787465940055,
+ "grad_norm": 2.1342601776123047,
+ "learning_rate": 7.333543550841966e-06,
+ "loss": 0.0452,
+ "step": 21959
+ },
+ {
+ "epoch": 59.83651226158038,
+ "grad_norm": 1.8712303638458252,
+ "learning_rate": 7.332693017630185e-06,
+ "loss": 0.0851,
+ "step": 21960
+ },
+ {
+ "epoch": 59.83923705722071,
+ "grad_norm": 1.5025303363800049,
+ "learning_rate": 7.331842505191308e-06,
+ "loss": 0.0298,
+ "step": 21961
+ },
+ {
+ "epoch": 59.84196185286103,
+ "grad_norm": 2.386803150177002,
+ "learning_rate": 7.330992013531952e-06,
+ "loss": 0.0765,
+ "step": 21962
+ },
+ {
+ "epoch": 59.844686648501366,
+ "grad_norm": 2.826340675354004,
+ "learning_rate": 7.330141542658747e-06,
+ "loss": 0.0596,
+ "step": 21963
+ },
+ {
+ "epoch": 59.84741144414169,
+ "grad_norm": 1.935585379600525,
+ "learning_rate": 7.32929109257831e-06,
+ "loss": 0.0289,
+ "step": 21964
+ },
+ {
+ "epoch": 59.85013623978202,
+ "grad_norm": 2.4856796264648438,
+ "learning_rate": 7.328440663297269e-06,
+ "loss": 0.1134,
+ "step": 21965
+ },
+ {
+ "epoch": 59.85286103542234,
+ "grad_norm": 2.873974561691284,
+ "learning_rate": 7.32759025482225e-06,
+ "loss": 0.0322,
+ "step": 21966
+ },
+ {
+ "epoch": 59.85558583106267,
+ "grad_norm": 2.349881172180176,
+ "learning_rate": 7.3267398671598685e-06,
+ "loss": 0.0392,
+ "step": 21967
+ },
+ {
+ "epoch": 59.858310626702995,
+ "grad_norm": 2.010082960128784,
+ "learning_rate": 7.325889500316754e-06,
+ "loss": 0.0627,
+ "step": 21968
+ },
+ {
+ "epoch": 59.86103542234333,
+ "grad_norm": 2.215298652648926,
+ "learning_rate": 7.325039154299522e-06,
+ "loss": 0.2131,
+ "step": 21969
+ },
+ {
+ "epoch": 59.86376021798365,
+ "grad_norm": 3.1088664531707764,
+ "learning_rate": 7.324188829114804e-06,
+ "loss": 0.1585,
+ "step": 21970
+ },
+ {
+ "epoch": 59.86648501362398,
+ "grad_norm": 3.3505115509033203,
+ "learning_rate": 7.323338524769212e-06,
+ "loss": 0.0231,
+ "step": 21971
+ },
+ {
+ "epoch": 59.869209809264305,
+ "grad_norm": 61.80009841918945,
+ "learning_rate": 7.322488241269379e-06,
+ "loss": 0.028,
+ "step": 21972
+ },
+ {
+ "epoch": 59.87193460490463,
+ "grad_norm": 57.920989990234375,
+ "learning_rate": 7.321637978621916e-06,
+ "loss": 0.0752,
+ "step": 21973
+ },
+ {
+ "epoch": 59.87465940054496,
+ "grad_norm": 2.286543607711792,
+ "learning_rate": 7.320787736833455e-06,
+ "loss": 0.0381,
+ "step": 21974
+ },
+ {
+ "epoch": 59.87738419618529,
+ "grad_norm": 2.1691548824310303,
+ "learning_rate": 7.319937515910609e-06,
+ "loss": 0.0386,
+ "step": 21975
+ },
+ {
+ "epoch": 59.880108991825615,
+ "grad_norm": 2.119908094406128,
+ "learning_rate": 7.319087315860005e-06,
+ "loss": 0.0291,
+ "step": 21976
+ },
+ {
+ "epoch": 59.88283378746594,
+ "grad_norm": 2.7863504886627197,
+ "learning_rate": 7.318237136688262e-06,
+ "loss": 0.0363,
+ "step": 21977
+ },
+ {
+ "epoch": 59.88555858310627,
+ "grad_norm": 2.5753395557403564,
+ "learning_rate": 7.317386978402003e-06,
+ "loss": 0.0359,
+ "step": 21978
+ },
+ {
+ "epoch": 59.88828337874659,
+ "grad_norm": 3.2744603157043457,
+ "learning_rate": 7.316536841007847e-06,
+ "loss": 0.0761,
+ "step": 21979
+ },
+ {
+ "epoch": 59.89100817438692,
+ "grad_norm": 2.225037097930908,
+ "learning_rate": 7.3156867245124175e-06,
+ "loss": 0.1061,
+ "step": 21980
+ },
+ {
+ "epoch": 59.89373297002725,
+ "grad_norm": 3.1315863132476807,
+ "learning_rate": 7.314836628922326e-06,
+ "loss": 0.0608,
+ "step": 21981
+ },
+ {
+ "epoch": 59.89645776566758,
+ "grad_norm": 2.6410772800445557,
+ "learning_rate": 7.313986554244208e-06,
+ "loss": 0.0641,
+ "step": 21982
+ },
+ {
+ "epoch": 59.8991825613079,
+ "grad_norm": 2.869687795639038,
+ "learning_rate": 7.313136500484673e-06,
+ "loss": 0.0401,
+ "step": 21983
+ },
+ {
+ "epoch": 59.90190735694823,
+ "grad_norm": 2.859328031539917,
+ "learning_rate": 7.312286467650347e-06,
+ "loss": 0.0439,
+ "step": 21984
+ },
+ {
+ "epoch": 59.904632152588555,
+ "grad_norm": 2.2563390731811523,
+ "learning_rate": 7.311436455747847e-06,
+ "loss": 0.0367,
+ "step": 21985
+ },
+ {
+ "epoch": 59.90735694822888,
+ "grad_norm": 2.969977378845215,
+ "learning_rate": 7.3105864647837934e-06,
+ "loss": 0.0335,
+ "step": 21986
+ },
+ {
+ "epoch": 59.91008174386921,
+ "grad_norm": 3.115579605102539,
+ "learning_rate": 7.309736494764804e-06,
+ "loss": 0.0451,
+ "step": 21987
+ },
+ {
+ "epoch": 59.91280653950954,
+ "grad_norm": 4.807703495025635,
+ "learning_rate": 7.308886545697502e-06,
+ "loss": 0.1353,
+ "step": 21988
+ },
+ {
+ "epoch": 59.915531335149865,
+ "grad_norm": 2.112199068069458,
+ "learning_rate": 7.308036617588504e-06,
+ "loss": 0.0344,
+ "step": 21989
+ },
+ {
+ "epoch": 59.91825613079019,
+ "grad_norm": 3.1096861362457275,
+ "learning_rate": 7.307186710444432e-06,
+ "loss": 0.1103,
+ "step": 21990
+ },
+ {
+ "epoch": 59.920980926430516,
+ "grad_norm": 3.6003177165985107,
+ "learning_rate": 7.3063368242719005e-06,
+ "loss": 0.0449,
+ "step": 21991
+ },
+ {
+ "epoch": 59.92370572207084,
+ "grad_norm": 3.2146692276000977,
+ "learning_rate": 7.3054869590775344e-06,
+ "loss": 0.0422,
+ "step": 21992
+ },
+ {
+ "epoch": 59.926430517711175,
+ "grad_norm": 2.72499418258667,
+ "learning_rate": 7.304637114867946e-06,
+ "loss": 0.0395,
+ "step": 21993
+ },
+ {
+ "epoch": 59.9291553133515,
+ "grad_norm": 2.1352198123931885,
+ "learning_rate": 7.30378729164976e-06,
+ "loss": 0.0209,
+ "step": 21994
+ },
+ {
+ "epoch": 59.93188010899183,
+ "grad_norm": 4.4527459144592285,
+ "learning_rate": 7.302937489429589e-06,
+ "loss": 0.0462,
+ "step": 21995
+ },
+ {
+ "epoch": 59.93460490463215,
+ "grad_norm": 3.115173101425171,
+ "learning_rate": 7.302087708214056e-06,
+ "loss": 0.1071,
+ "step": 21996
+ },
+ {
+ "epoch": 59.93732970027248,
+ "grad_norm": 2.689356565475464,
+ "learning_rate": 7.301237948009775e-06,
+ "loss": 0.0358,
+ "step": 21997
+ },
+ {
+ "epoch": 59.940054495912804,
+ "grad_norm": 3.24954891204834,
+ "learning_rate": 7.3003882088233635e-06,
+ "loss": 0.1604,
+ "step": 21998
+ },
+ {
+ "epoch": 59.94277929155314,
+ "grad_norm": 4.533755302429199,
+ "learning_rate": 7.299538490661446e-06,
+ "loss": 0.0619,
+ "step": 21999
+ },
+ {
+ "epoch": 59.94550408719346,
+ "grad_norm": 2.144399881362915,
+ "learning_rate": 7.298688793530635e-06,
+ "loss": 0.0777,
+ "step": 22000
+ },
+ {
+ "epoch": 59.94822888283379,
+ "grad_norm": 2.544394016265869,
+ "learning_rate": 7.297839117437549e-06,
+ "loss": 0.0805,
+ "step": 22001
+ },
+ {
+ "epoch": 59.950953678474114,
+ "grad_norm": 2.3073601722717285,
+ "learning_rate": 7.296989462388806e-06,
+ "loss": 0.1612,
+ "step": 22002
+ },
+ {
+ "epoch": 59.95367847411444,
+ "grad_norm": 2.9807851314544678,
+ "learning_rate": 7.2961398283910204e-06,
+ "loss": 0.1453,
+ "step": 22003
+ },
+ {
+ "epoch": 59.956403269754766,
+ "grad_norm": 2.8637328147888184,
+ "learning_rate": 7.295290215450812e-06,
+ "loss": 0.0402,
+ "step": 22004
+ },
+ {
+ "epoch": 59.95912806539509,
+ "grad_norm": 2.4882168769836426,
+ "learning_rate": 7.2944406235747945e-06,
+ "loss": 0.0522,
+ "step": 22005
+ },
+ {
+ "epoch": 59.961852861035425,
+ "grad_norm": 2.5956735610961914,
+ "learning_rate": 7.293591052769589e-06,
+ "loss": 0.0651,
+ "step": 22006
+ },
+ {
+ "epoch": 59.96457765667575,
+ "grad_norm": 2.526373863220215,
+ "learning_rate": 7.292741503041807e-06,
+ "loss": 0.0525,
+ "step": 22007
+ },
+ {
+ "epoch": 59.967302452316076,
+ "grad_norm": 2.720773220062256,
+ "learning_rate": 7.2918919743980685e-06,
+ "loss": 0.1369,
+ "step": 22008
+ },
+ {
+ "epoch": 59.9700272479564,
+ "grad_norm": 4.978150367736816,
+ "learning_rate": 7.291042466844986e-06,
+ "loss": 0.0623,
+ "step": 22009
+ },
+ {
+ "epoch": 59.97275204359673,
+ "grad_norm": 2.08091402053833,
+ "learning_rate": 7.290192980389182e-06,
+ "loss": 0.096,
+ "step": 22010
+ },
+ {
+ "epoch": 59.97547683923706,
+ "grad_norm": 2.060152769088745,
+ "learning_rate": 7.289343515037263e-06,
+ "loss": 0.1045,
+ "step": 22011
+ },
+ {
+ "epoch": 59.97820163487739,
+ "grad_norm": 2.868919849395752,
+ "learning_rate": 7.288494070795853e-06,
+ "loss": 0.0455,
+ "step": 22012
+ },
+ {
+ "epoch": 59.98092643051771,
+ "grad_norm": 1.9682955741882324,
+ "learning_rate": 7.287644647671561e-06,
+ "loss": 0.1192,
+ "step": 22013
+ },
+ {
+ "epoch": 59.98365122615804,
+ "grad_norm": 3.429980516433716,
+ "learning_rate": 7.2867952456710075e-06,
+ "loss": 0.1148,
+ "step": 22014
+ },
+ {
+ "epoch": 59.986376021798364,
+ "grad_norm": 2.7559564113616943,
+ "learning_rate": 7.2859458648008005e-06,
+ "loss": 0.0623,
+ "step": 22015
+ },
+ {
+ "epoch": 59.98910081743869,
+ "grad_norm": 2.8109524250030518,
+ "learning_rate": 7.285096505067564e-06,
+ "loss": 0.1969,
+ "step": 22016
+ },
+ {
+ "epoch": 59.991825613079016,
+ "grad_norm": 3.3134636878967285,
+ "learning_rate": 7.284247166477909e-06,
+ "loss": 0.0588,
+ "step": 22017
+ },
+ {
+ "epoch": 59.99455040871935,
+ "grad_norm": 3.0938918590545654,
+ "learning_rate": 7.283397849038447e-06,
+ "loss": 0.1383,
+ "step": 22018
+ },
+ {
+ "epoch": 59.997275204359674,
+ "grad_norm": 1.5627973079681396,
+ "learning_rate": 7.282548552755799e-06,
+ "loss": 0.0206,
+ "step": 22019
+ },
+ {
+ "epoch": 60.0,
+ "grad_norm": 1.6823599338531494,
+ "learning_rate": 7.2816992776365714e-06,
+ "loss": 0.0229,
+ "step": 22020
+ },
+ {
+ "epoch": 60.002724795640326,
+ "grad_norm": 3.2398977279663086,
+ "learning_rate": 7.280850023687387e-06,
+ "loss": 0.1166,
+ "step": 22021
+ },
+ {
+ "epoch": 60.00544959128065,
+ "grad_norm": 4.0668182373046875,
+ "learning_rate": 7.280000790914851e-06,
+ "loss": 0.0513,
+ "step": 22022
+ },
+ {
+ "epoch": 60.00817438692098,
+ "grad_norm": 2.300875663757324,
+ "learning_rate": 7.27915157932558e-06,
+ "loss": 0.1671,
+ "step": 22023
+ },
+ {
+ "epoch": 60.01089918256131,
+ "grad_norm": 2.258364677429199,
+ "learning_rate": 7.278302388926194e-06,
+ "loss": 0.0273,
+ "step": 22024
+ },
+ {
+ "epoch": 60.013623978201636,
+ "grad_norm": 2.5527184009552,
+ "learning_rate": 7.277453219723298e-06,
+ "loss": 0.0902,
+ "step": 22025
+ },
+ {
+ "epoch": 60.01634877384196,
+ "grad_norm": 2.553588628768921,
+ "learning_rate": 7.276604071723511e-06,
+ "loss": 0.0406,
+ "step": 22026
+ },
+ {
+ "epoch": 60.01907356948229,
+ "grad_norm": 1.5406901836395264,
+ "learning_rate": 7.275754944933442e-06,
+ "loss": 0.036,
+ "step": 22027
+ },
+ {
+ "epoch": 60.02179836512261,
+ "grad_norm": 2.426330804824829,
+ "learning_rate": 7.274905839359707e-06,
+ "loss": 0.041,
+ "step": 22028
+ },
+ {
+ "epoch": 60.02452316076294,
+ "grad_norm": 1.9398462772369385,
+ "learning_rate": 7.274056755008916e-06,
+ "loss": 0.2407,
+ "step": 22029
+ },
+ {
+ "epoch": 60.02724795640327,
+ "grad_norm": 3.1811892986297607,
+ "learning_rate": 7.273207691887685e-06,
+ "loss": 0.0485,
+ "step": 22030
+ },
+ {
+ "epoch": 60.0299727520436,
+ "grad_norm": 2.42561674118042,
+ "learning_rate": 7.272358650002623e-06,
+ "loss": 0.08,
+ "step": 22031
+ },
+ {
+ "epoch": 60.032697547683924,
+ "grad_norm": 1.8519953489303589,
+ "learning_rate": 7.271509629360343e-06,
+ "loss": 0.0183,
+ "step": 22032
+ },
+ {
+ "epoch": 60.03542234332425,
+ "grad_norm": 2.844485282897949,
+ "learning_rate": 7.270660629967461e-06,
+ "loss": 0.0295,
+ "step": 22033
+ },
+ {
+ "epoch": 60.038147138964575,
+ "grad_norm": 1.3753477334976196,
+ "learning_rate": 7.269811651830584e-06,
+ "loss": 0.0223,
+ "step": 22034
+ },
+ {
+ "epoch": 60.0408719346049,
+ "grad_norm": 2.953296422958374,
+ "learning_rate": 7.268962694956329e-06,
+ "loss": 0.179,
+ "step": 22035
+ },
+ {
+ "epoch": 60.043596730245234,
+ "grad_norm": 2.489459991455078,
+ "learning_rate": 7.2681137593513005e-06,
+ "loss": 0.0966,
+ "step": 22036
+ },
+ {
+ "epoch": 60.04632152588556,
+ "grad_norm": 1.6408772468566895,
+ "learning_rate": 7.2672648450221195e-06,
+ "loss": 0.0194,
+ "step": 22037
+ },
+ {
+ "epoch": 60.049046321525886,
+ "grad_norm": 2.2408220767974854,
+ "learning_rate": 7.266415951975388e-06,
+ "loss": 0.062,
+ "step": 22038
+ },
+ {
+ "epoch": 60.05177111716621,
+ "grad_norm": 3.1908164024353027,
+ "learning_rate": 7.265567080217723e-06,
+ "loss": 0.0188,
+ "step": 22039
+ },
+ {
+ "epoch": 60.05449591280654,
+ "grad_norm": 2.297990083694458,
+ "learning_rate": 7.264718229755731e-06,
+ "loss": 0.0391,
+ "step": 22040
+ },
+ {
+ "epoch": 60.05722070844686,
+ "grad_norm": 3.440293788909912,
+ "learning_rate": 7.263869400596028e-06,
+ "loss": 0.1144,
+ "step": 22041
+ },
+ {
+ "epoch": 60.059945504087196,
+ "grad_norm": 1.998288869857788,
+ "learning_rate": 7.26302059274522e-06,
+ "loss": 0.0218,
+ "step": 22042
+ },
+ {
+ "epoch": 60.06267029972752,
+ "grad_norm": 1.9744524955749512,
+ "learning_rate": 7.262171806209924e-06,
+ "loss": 0.0359,
+ "step": 22043
+ },
+ {
+ "epoch": 60.06539509536785,
+ "grad_norm": 2.3341996669769287,
+ "learning_rate": 7.261323040996739e-06,
+ "loss": 0.0706,
+ "step": 22044
+ },
+ {
+ "epoch": 60.06811989100817,
+ "grad_norm": 3.268862009048462,
+ "learning_rate": 7.260474297112286e-06,
+ "loss": 0.0715,
+ "step": 22045
+ },
+ {
+ "epoch": 60.0708446866485,
+ "grad_norm": 2.099651336669922,
+ "learning_rate": 7.259625574563172e-06,
+ "loss": 0.021,
+ "step": 22046
+ },
+ {
+ "epoch": 60.073569482288825,
+ "grad_norm": 1.4483468532562256,
+ "learning_rate": 7.258776873356003e-06,
+ "loss": 0.0169,
+ "step": 22047
+ },
+ {
+ "epoch": 60.07629427792916,
+ "grad_norm": 2.3181116580963135,
+ "learning_rate": 7.257928193497394e-06,
+ "loss": 0.1066,
+ "step": 22048
+ },
+ {
+ "epoch": 60.079019073569484,
+ "grad_norm": 2.014277219772339,
+ "learning_rate": 7.257079534993949e-06,
+ "loss": 0.0991,
+ "step": 22049
+ },
+ {
+ "epoch": 60.08174386920981,
+ "grad_norm": 1.9099525213241577,
+ "learning_rate": 7.2562308978522784e-06,
+ "loss": 0.0868,
+ "step": 22050
+ },
+ {
+ "epoch": 60.084468664850135,
+ "grad_norm": 2.1944446563720703,
+ "learning_rate": 7.255382282079e-06,
+ "loss": 0.0468,
+ "step": 22051
+ },
+ {
+ "epoch": 60.08719346049046,
+ "grad_norm": 1.6159353256225586,
+ "learning_rate": 7.254533687680712e-06,
+ "loss": 0.0115,
+ "step": 22052
+ },
+ {
+ "epoch": 60.08991825613079,
+ "grad_norm": 2.713303565979004,
+ "learning_rate": 7.253685114664029e-06,
+ "loss": 0.0852,
+ "step": 22053
+ },
+ {
+ "epoch": 60.09264305177112,
+ "grad_norm": 2.804295301437378,
+ "learning_rate": 7.252836563035557e-06,
+ "loss": 0.1009,
+ "step": 22054
+ },
+ {
+ "epoch": 60.095367847411445,
+ "grad_norm": 2.252511739730835,
+ "learning_rate": 7.251988032801906e-06,
+ "loss": 0.0381,
+ "step": 22055
+ },
+ {
+ "epoch": 60.09809264305177,
+ "grad_norm": 3.0858805179595947,
+ "learning_rate": 7.251139523969683e-06,
+ "loss": 0.0299,
+ "step": 22056
+ },
+ {
+ "epoch": 60.1008174386921,
+ "grad_norm": 2.334627628326416,
+ "learning_rate": 7.2502910365454985e-06,
+ "loss": 0.1934,
+ "step": 22057
+ },
+ {
+ "epoch": 60.10354223433242,
+ "grad_norm": 13.700309753417969,
+ "learning_rate": 7.249442570535956e-06,
+ "loss": 0.1911,
+ "step": 22058
+ },
+ {
+ "epoch": 60.10626702997275,
+ "grad_norm": 2.5898914337158203,
+ "learning_rate": 7.24859412594767e-06,
+ "loss": 0.0866,
+ "step": 22059
+ },
+ {
+ "epoch": 60.10899182561308,
+ "grad_norm": 1.6799453496932983,
+ "learning_rate": 7.247745702787241e-06,
+ "loss": 0.0351,
+ "step": 22060
+ },
+ {
+ "epoch": 60.11171662125341,
+ "grad_norm": 2.467136859893799,
+ "learning_rate": 7.246897301061282e-06,
+ "loss": 0.168,
+ "step": 22061
+ },
+ {
+ "epoch": 60.11444141689373,
+ "grad_norm": 2.5998342037200928,
+ "learning_rate": 7.2460489207763965e-06,
+ "loss": 0.0543,
+ "step": 22062
+ },
+ {
+ "epoch": 60.11716621253406,
+ "grad_norm": 2.316824436187744,
+ "learning_rate": 7.245200561939196e-06,
+ "loss": 0.0489,
+ "step": 22063
+ },
+ {
+ "epoch": 60.119891008174385,
+ "grad_norm": 2.0856094360351562,
+ "learning_rate": 7.244352224556281e-06,
+ "loss": 0.087,
+ "step": 22064
+ },
+ {
+ "epoch": 60.12261580381471,
+ "grad_norm": 2.8968212604522705,
+ "learning_rate": 7.243503908634266e-06,
+ "loss": 0.0354,
+ "step": 22065
+ },
+ {
+ "epoch": 60.12534059945504,
+ "grad_norm": 1.6289125680923462,
+ "learning_rate": 7.2426556141797525e-06,
+ "loss": 0.0319,
+ "step": 22066
+ },
+ {
+ "epoch": 60.12806539509537,
+ "grad_norm": 1.700653314590454,
+ "learning_rate": 7.241807341199345e-06,
+ "loss": 0.1921,
+ "step": 22067
+ },
+ {
+ "epoch": 60.130790190735695,
+ "grad_norm": 2.8341064453125,
+ "learning_rate": 7.240959089699655e-06,
+ "loss": 0.053,
+ "step": 22068
+ },
+ {
+ "epoch": 60.13351498637602,
+ "grad_norm": 2.563744306564331,
+ "learning_rate": 7.2401108596872885e-06,
+ "loss": 0.0361,
+ "step": 22069
+ },
+ {
+ "epoch": 60.13623978201635,
+ "grad_norm": 1.6389639377593994,
+ "learning_rate": 7.239262651168848e-06,
+ "loss": 0.0409,
+ "step": 22070
+ },
+ {
+ "epoch": 60.13896457765667,
+ "grad_norm": 3.7086379528045654,
+ "learning_rate": 7.238414464150945e-06,
+ "loss": 0.0447,
+ "step": 22071
+ },
+ {
+ "epoch": 60.141689373297005,
+ "grad_norm": 2.211578130722046,
+ "learning_rate": 7.237566298640178e-06,
+ "loss": 0.0373,
+ "step": 22072
+ },
+ {
+ "epoch": 60.14441416893733,
+ "grad_norm": 3.035041570663452,
+ "learning_rate": 7.2367181546431565e-06,
+ "loss": 0.0353,
+ "step": 22073
+ },
+ {
+ "epoch": 60.14713896457766,
+ "grad_norm": 2.4309768676757812,
+ "learning_rate": 7.2358700321664845e-06,
+ "loss": 0.0509,
+ "step": 22074
+ },
+ {
+ "epoch": 60.14986376021798,
+ "grad_norm": 1.7298613786697388,
+ "learning_rate": 7.2350219312167695e-06,
+ "loss": 0.0241,
+ "step": 22075
+ },
+ {
+ "epoch": 60.15258855585831,
+ "grad_norm": 3.1753647327423096,
+ "learning_rate": 7.234173851800612e-06,
+ "loss": 0.0605,
+ "step": 22076
+ },
+ {
+ "epoch": 60.155313351498634,
+ "grad_norm": 1.7607156038284302,
+ "learning_rate": 7.2333257939246226e-06,
+ "loss": 0.0232,
+ "step": 22077
+ },
+ {
+ "epoch": 60.15803814713897,
+ "grad_norm": 2.2599029541015625,
+ "learning_rate": 7.2324777575954e-06,
+ "loss": 0.0218,
+ "step": 22078
+ },
+ {
+ "epoch": 60.16076294277929,
+ "grad_norm": 3.1070573329925537,
+ "learning_rate": 7.231629742819554e-06,
+ "loss": 0.0548,
+ "step": 22079
+ },
+ {
+ "epoch": 60.16348773841962,
+ "grad_norm": 10.807502746582031,
+ "learning_rate": 7.230781749603685e-06,
+ "loss": 0.0404,
+ "step": 22080
+ },
+ {
+ "epoch": 60.166212534059945,
+ "grad_norm": 3.5893166065216064,
+ "learning_rate": 7.2299337779543995e-06,
+ "loss": 0.2294,
+ "step": 22081
+ },
+ {
+ "epoch": 60.16893732970027,
+ "grad_norm": 5.226335048675537,
+ "learning_rate": 7.2290858278783e-06,
+ "loss": 0.1076,
+ "step": 22082
+ },
+ {
+ "epoch": 60.171662125340596,
+ "grad_norm": 1.9509162902832031,
+ "learning_rate": 7.2282378993819916e-06,
+ "loss": 0.0199,
+ "step": 22083
+ },
+ {
+ "epoch": 60.17438692098093,
+ "grad_norm": 1.9179598093032837,
+ "learning_rate": 7.227389992472073e-06,
+ "loss": 0.0536,
+ "step": 22084
+ },
+ {
+ "epoch": 60.177111716621255,
+ "grad_norm": 2.207465171813965,
+ "learning_rate": 7.226542107155154e-06,
+ "loss": 0.0495,
+ "step": 22085
+ },
+ {
+ "epoch": 60.17983651226158,
+ "grad_norm": 2.003126382827759,
+ "learning_rate": 7.2256942434378395e-06,
+ "loss": 0.0349,
+ "step": 22086
+ },
+ {
+ "epoch": 60.182561307901906,
+ "grad_norm": 2.5865070819854736,
+ "learning_rate": 7.224846401326725e-06,
+ "loss": 0.0226,
+ "step": 22087
+ },
+ {
+ "epoch": 60.18528610354223,
+ "grad_norm": 1.8921812772750854,
+ "learning_rate": 7.223998580828418e-06,
+ "loss": 0.0492,
+ "step": 22088
+ },
+ {
+ "epoch": 60.18801089918256,
+ "grad_norm": 2.0194311141967773,
+ "learning_rate": 7.223150781949524e-06,
+ "loss": 0.0909,
+ "step": 22089
+ },
+ {
+ "epoch": 60.19073569482289,
+ "grad_norm": 8.190629005432129,
+ "learning_rate": 7.222303004696638e-06,
+ "loss": 0.044,
+ "step": 22090
+ },
+ {
+ "epoch": 60.19346049046322,
+ "grad_norm": 1.6974294185638428,
+ "learning_rate": 7.22145524907637e-06,
+ "loss": 0.034,
+ "step": 22091
+ },
+ {
+ "epoch": 60.19618528610354,
+ "grad_norm": 2.5855226516723633,
+ "learning_rate": 7.220607515095317e-06,
+ "loss": 0.0902,
+ "step": 22092
+ },
+ {
+ "epoch": 60.19891008174387,
+ "grad_norm": 2.1800148487091064,
+ "learning_rate": 7.219759802760085e-06,
+ "loss": 0.021,
+ "step": 22093
+ },
+ {
+ "epoch": 60.201634877384194,
+ "grad_norm": 2.002150058746338,
+ "learning_rate": 7.2189121120772735e-06,
+ "loss": 0.0219,
+ "step": 22094
+ },
+ {
+ "epoch": 60.20435967302452,
+ "grad_norm": 1.7807129621505737,
+ "learning_rate": 7.218064443053485e-06,
+ "loss": 0.0199,
+ "step": 22095
+ },
+ {
+ "epoch": 60.20708446866485,
+ "grad_norm": 2.404022216796875,
+ "learning_rate": 7.217216795695321e-06,
+ "loss": 0.0382,
+ "step": 22096
+ },
+ {
+ "epoch": 60.20980926430518,
+ "grad_norm": 2.7518961429595947,
+ "learning_rate": 7.216369170009384e-06,
+ "loss": 0.0336,
+ "step": 22097
+ },
+ {
+ "epoch": 60.212534059945504,
+ "grad_norm": 2.140996217727661,
+ "learning_rate": 7.215521566002273e-06,
+ "loss": 0.0823,
+ "step": 22098
+ },
+ {
+ "epoch": 60.21525885558583,
+ "grad_norm": 1.658980131149292,
+ "learning_rate": 7.214673983680591e-06,
+ "loss": 0.0231,
+ "step": 22099
+ },
+ {
+ "epoch": 60.217983651226156,
+ "grad_norm": 2.674773931503296,
+ "learning_rate": 7.213826423050937e-06,
+ "loss": 0.0558,
+ "step": 22100
+ },
+ {
+ "epoch": 60.22070844686648,
+ "grad_norm": 1.559118390083313,
+ "learning_rate": 7.212978884119911e-06,
+ "loss": 0.1277,
+ "step": 22101
+ },
+ {
+ "epoch": 60.223433242506815,
+ "grad_norm": 2.343963146209717,
+ "learning_rate": 7.212131366894121e-06,
+ "loss": 0.037,
+ "step": 22102
+ },
+ {
+ "epoch": 60.22615803814714,
+ "grad_norm": 2.1192994117736816,
+ "learning_rate": 7.211283871380159e-06,
+ "loss": 0.228,
+ "step": 22103
+ },
+ {
+ "epoch": 60.228882833787466,
+ "grad_norm": 1.7171218395233154,
+ "learning_rate": 7.21043639758463e-06,
+ "loss": 0.0193,
+ "step": 22104
+ },
+ {
+ "epoch": 60.23160762942779,
+ "grad_norm": 2.325385093688965,
+ "learning_rate": 7.2095889455141325e-06,
+ "loss": 0.0564,
+ "step": 22105
+ },
+ {
+ "epoch": 60.23433242506812,
+ "grad_norm": 3.0220072269439697,
+ "learning_rate": 7.208741515175266e-06,
+ "loss": 0.0507,
+ "step": 22106
+ },
+ {
+ "epoch": 60.237057220708444,
+ "grad_norm": 2.1577396392822266,
+ "learning_rate": 7.207894106574631e-06,
+ "loss": 0.0302,
+ "step": 22107
+ },
+ {
+ "epoch": 60.23978201634878,
+ "grad_norm": 2.940560817718506,
+ "learning_rate": 7.207046719718827e-06,
+ "loss": 0.0906,
+ "step": 22108
+ },
+ {
+ "epoch": 60.2425068119891,
+ "grad_norm": 2.159417152404785,
+ "learning_rate": 7.206199354614453e-06,
+ "loss": 0.0403,
+ "step": 22109
+ },
+ {
+ "epoch": 60.24523160762943,
+ "grad_norm": 1.930489420890808,
+ "learning_rate": 7.205352011268107e-06,
+ "loss": 0.0255,
+ "step": 22110
+ },
+ {
+ "epoch": 60.247956403269754,
+ "grad_norm": 2.0277271270751953,
+ "learning_rate": 7.204504689686392e-06,
+ "loss": 0.0168,
+ "step": 22111
+ },
+ {
+ "epoch": 60.25068119891008,
+ "grad_norm": 2.730210542678833,
+ "learning_rate": 7.203657389875902e-06,
+ "loss": 0.0335,
+ "step": 22112
+ },
+ {
+ "epoch": 60.253405994550405,
+ "grad_norm": 2.2015161514282227,
+ "learning_rate": 7.2028101118432414e-06,
+ "loss": 0.0359,
+ "step": 22113
+ },
+ {
+ "epoch": 60.25613079019074,
+ "grad_norm": 2.3820102214813232,
+ "learning_rate": 7.201962855595004e-06,
+ "loss": 0.031,
+ "step": 22114
+ },
+ {
+ "epoch": 60.258855585831064,
+ "grad_norm": 4.314913749694824,
+ "learning_rate": 7.2011156211377904e-06,
+ "loss": 0.0286,
+ "step": 22115
+ },
+ {
+ "epoch": 60.26158038147139,
+ "grad_norm": 2.8772289752960205,
+ "learning_rate": 7.200268408478197e-06,
+ "loss": 0.0511,
+ "step": 22116
+ },
+ {
+ "epoch": 60.264305177111716,
+ "grad_norm": 1.672746181488037,
+ "learning_rate": 7.199421217622825e-06,
+ "loss": 0.0206,
+ "step": 22117
+ },
+ {
+ "epoch": 60.26702997275204,
+ "grad_norm": 2.982311487197876,
+ "learning_rate": 7.198574048578268e-06,
+ "loss": 0.0352,
+ "step": 22118
+ },
+ {
+ "epoch": 60.26975476839237,
+ "grad_norm": 2.440772294998169,
+ "learning_rate": 7.1977269013511254e-06,
+ "loss": 0.1404,
+ "step": 22119
+ },
+ {
+ "epoch": 60.2724795640327,
+ "grad_norm": 2.8254506587982178,
+ "learning_rate": 7.196879775947998e-06,
+ "loss": 0.0916,
+ "step": 22120
+ },
+ {
+ "epoch": 60.275204359673026,
+ "grad_norm": 2.6894478797912598,
+ "learning_rate": 7.196032672375479e-06,
+ "loss": 0.0643,
+ "step": 22121
+ },
+ {
+ "epoch": 60.27792915531335,
+ "grad_norm": 2.8892157077789307,
+ "learning_rate": 7.19518559064017e-06,
+ "loss": 0.0512,
+ "step": 22122
+ },
+ {
+ "epoch": 60.28065395095368,
+ "grad_norm": 3.4282281398773193,
+ "learning_rate": 7.194338530748663e-06,
+ "loss": 0.0225,
+ "step": 22123
+ },
+ {
+ "epoch": 60.283378746594,
+ "grad_norm": 1.9847115278244019,
+ "learning_rate": 7.1934914927075595e-06,
+ "loss": 0.0416,
+ "step": 22124
+ },
+ {
+ "epoch": 60.28610354223433,
+ "grad_norm": 2.2255382537841797,
+ "learning_rate": 7.1926444765234515e-06,
+ "loss": 0.0441,
+ "step": 22125
+ },
+ {
+ "epoch": 60.28882833787466,
+ "grad_norm": 3.1593189239501953,
+ "learning_rate": 7.19179748220294e-06,
+ "loss": 0.0524,
+ "step": 22126
+ },
+ {
+ "epoch": 60.29155313351499,
+ "grad_norm": 2.229228973388672,
+ "learning_rate": 7.190950509752617e-06,
+ "loss": 0.1723,
+ "step": 22127
+ },
+ {
+ "epoch": 60.294277929155314,
+ "grad_norm": 1.6642812490463257,
+ "learning_rate": 7.190103559179085e-06,
+ "loss": 0.0988,
+ "step": 22128
+ },
+ {
+ "epoch": 60.29700272479564,
+ "grad_norm": 2.12465763092041,
+ "learning_rate": 7.1892566304889315e-06,
+ "loss": 0.1244,
+ "step": 22129
+ },
+ {
+ "epoch": 60.299727520435965,
+ "grad_norm": 2.333627700805664,
+ "learning_rate": 7.188409723688759e-06,
+ "loss": 0.0408,
+ "step": 22130
+ },
+ {
+ "epoch": 60.30245231607629,
+ "grad_norm": 2.5083484649658203,
+ "learning_rate": 7.18756283878516e-06,
+ "loss": 0.0244,
+ "step": 22131
+ },
+ {
+ "epoch": 60.305177111716624,
+ "grad_norm": 3.49600887298584,
+ "learning_rate": 7.186715975784731e-06,
+ "loss": 0.0857,
+ "step": 22132
+ },
+ {
+ "epoch": 60.30790190735695,
+ "grad_norm": 2.888213872909546,
+ "learning_rate": 7.18586913469407e-06,
+ "loss": 0.0241,
+ "step": 22133
+ },
+ {
+ "epoch": 60.310626702997276,
+ "grad_norm": 4.418033599853516,
+ "learning_rate": 7.185022315519768e-06,
+ "loss": 0.0758,
+ "step": 22134
+ },
+ {
+ "epoch": 60.3133514986376,
+ "grad_norm": 1.9535391330718994,
+ "learning_rate": 7.1841755182684215e-06,
+ "loss": 0.0334,
+ "step": 22135
+ },
+ {
+ "epoch": 60.31607629427793,
+ "grad_norm": 1.9297291040420532,
+ "learning_rate": 7.183328742946623e-06,
+ "loss": 0.0728,
+ "step": 22136
+ },
+ {
+ "epoch": 60.31880108991825,
+ "grad_norm": 2.833613395690918,
+ "learning_rate": 7.182481989560971e-06,
+ "loss": 0.0257,
+ "step": 22137
+ },
+ {
+ "epoch": 60.321525885558586,
+ "grad_norm": 1.6294703483581543,
+ "learning_rate": 7.18163525811806e-06,
+ "loss": 0.0373,
+ "step": 22138
+ },
+ {
+ "epoch": 60.32425068119891,
+ "grad_norm": 10.354475021362305,
+ "learning_rate": 7.180788548624482e-06,
+ "loss": 0.0823,
+ "step": 22139
+ },
+ {
+ "epoch": 60.32697547683924,
+ "grad_norm": 3.2036287784576416,
+ "learning_rate": 7.179941861086834e-06,
+ "loss": 0.1137,
+ "step": 22140
+ },
+ {
+ "epoch": 60.32970027247956,
+ "grad_norm": 1.613546371459961,
+ "learning_rate": 7.179095195511707e-06,
+ "loss": 0.0451,
+ "step": 22141
+ },
+ {
+ "epoch": 60.33242506811989,
+ "grad_norm": 1.9399296045303345,
+ "learning_rate": 7.178248551905696e-06,
+ "loss": 0.0314,
+ "step": 22142
+ },
+ {
+ "epoch": 60.335149863760215,
+ "grad_norm": 1.5403608083724976,
+ "learning_rate": 7.177401930275394e-06,
+ "loss": 0.017,
+ "step": 22143
+ },
+ {
+ "epoch": 60.33787465940055,
+ "grad_norm": 3.1528384685516357,
+ "learning_rate": 7.176555330627395e-06,
+ "loss": 0.0613,
+ "step": 22144
+ },
+ {
+ "epoch": 60.34059945504087,
+ "grad_norm": 1.6675516366958618,
+ "learning_rate": 7.175708752968291e-06,
+ "loss": 0.1774,
+ "step": 22145
+ },
+ {
+ "epoch": 60.3433242506812,
+ "grad_norm": 4.045810699462891,
+ "learning_rate": 7.174862197304679e-06,
+ "loss": 0.0607,
+ "step": 22146
+ },
+ {
+ "epoch": 60.346049046321525,
+ "grad_norm": 1.9677788019180298,
+ "learning_rate": 7.174015663643147e-06,
+ "loss": 0.0319,
+ "step": 22147
+ },
+ {
+ "epoch": 60.34877384196185,
+ "grad_norm": 1.5508817434310913,
+ "learning_rate": 7.173169151990291e-06,
+ "loss": 0.018,
+ "step": 22148
+ },
+ {
+ "epoch": 60.35149863760218,
+ "grad_norm": 2.494908571243286,
+ "learning_rate": 7.172322662352702e-06,
+ "loss": 0.0384,
+ "step": 22149
+ },
+ {
+ "epoch": 60.35422343324251,
+ "grad_norm": 3.9148340225219727,
+ "learning_rate": 7.171476194736975e-06,
+ "loss": 0.0744,
+ "step": 22150
+ },
+ {
+ "epoch": 60.356948228882835,
+ "grad_norm": 1.8101519346237183,
+ "learning_rate": 7.170629749149698e-06,
+ "loss": 0.0226,
+ "step": 22151
+ },
+ {
+ "epoch": 60.35967302452316,
+ "grad_norm": 2.4348337650299072,
+ "learning_rate": 7.169783325597468e-06,
+ "loss": 0.03,
+ "step": 22152
+ },
+ {
+ "epoch": 60.36239782016349,
+ "grad_norm": 2.07472562789917,
+ "learning_rate": 7.168936924086866e-06,
+ "loss": 0.0429,
+ "step": 22153
+ },
+ {
+ "epoch": 60.36512261580381,
+ "grad_norm": 1.764729380607605,
+ "learning_rate": 7.168090544624501e-06,
+ "loss": 0.065,
+ "step": 22154
+ },
+ {
+ "epoch": 60.36784741144414,
+ "grad_norm": 3.583106279373169,
+ "learning_rate": 7.167244187216953e-06,
+ "loss": 0.1474,
+ "step": 22155
+ },
+ {
+ "epoch": 60.37057220708447,
+ "grad_norm": 1.8611454963684082,
+ "learning_rate": 7.166397851870816e-06,
+ "loss": 0.035,
+ "step": 22156
+ },
+ {
+ "epoch": 60.3732970027248,
+ "grad_norm": 2.8216235637664795,
+ "learning_rate": 7.165551538592682e-06,
+ "loss": 0.1157,
+ "step": 22157
+ },
+ {
+ "epoch": 60.37602179836512,
+ "grad_norm": 2.128490686416626,
+ "learning_rate": 7.164705247389143e-06,
+ "loss": 0.1145,
+ "step": 22158
+ },
+ {
+ "epoch": 60.37874659400545,
+ "grad_norm": 2.3129169940948486,
+ "learning_rate": 7.163858978266786e-06,
+ "loss": 0.0306,
+ "step": 22159
+ },
+ {
+ "epoch": 60.381471389645775,
+ "grad_norm": 1.8828394412994385,
+ "learning_rate": 7.163012731232206e-06,
+ "loss": 0.0228,
+ "step": 22160
+ },
+ {
+ "epoch": 60.3841961852861,
+ "grad_norm": 1.9650882482528687,
+ "learning_rate": 7.162166506291988e-06,
+ "loss": 0.1045,
+ "step": 22161
+ },
+ {
+ "epoch": 60.38692098092643,
+ "grad_norm": 2.0500175952911377,
+ "learning_rate": 7.16132030345273e-06,
+ "loss": 0.0748,
+ "step": 22162
+ },
+ {
+ "epoch": 60.38964577656676,
+ "grad_norm": 2.309939384460449,
+ "learning_rate": 7.160474122721015e-06,
+ "loss": 0.1539,
+ "step": 22163
+ },
+ {
+ "epoch": 60.392370572207085,
+ "grad_norm": 2.306363821029663,
+ "learning_rate": 7.159627964103439e-06,
+ "loss": 0.0206,
+ "step": 22164
+ },
+ {
+ "epoch": 60.39509536784741,
+ "grad_norm": 2.070624589920044,
+ "learning_rate": 7.158781827606587e-06,
+ "loss": 0.0439,
+ "step": 22165
+ },
+ {
+ "epoch": 60.39782016348774,
+ "grad_norm": 1.9535043239593506,
+ "learning_rate": 7.157935713237052e-06,
+ "loss": 0.0847,
+ "step": 22166
+ },
+ {
+ "epoch": 60.40054495912806,
+ "grad_norm": 2.1646225452423096,
+ "learning_rate": 7.15708962100142e-06,
+ "loss": 0.0405,
+ "step": 22167
+ },
+ {
+ "epoch": 60.403269754768395,
+ "grad_norm": 2.400301694869995,
+ "learning_rate": 7.156243550906285e-06,
+ "loss": 0.0567,
+ "step": 22168
+ },
+ {
+ "epoch": 60.40599455040872,
+ "grad_norm": 1.890462040901184,
+ "learning_rate": 7.155397502958231e-06,
+ "loss": 0.0202,
+ "step": 22169
+ },
+ {
+ "epoch": 60.40871934604905,
+ "grad_norm": 1.5231188535690308,
+ "learning_rate": 7.154551477163848e-06,
+ "loss": 0.0246,
+ "step": 22170
+ },
+ {
+ "epoch": 60.41144414168937,
+ "grad_norm": 1.4940664768218994,
+ "learning_rate": 7.153705473529731e-06,
+ "loss": 0.0287,
+ "step": 22171
+ },
+ {
+ "epoch": 60.4141689373297,
+ "grad_norm": 1.3897590637207031,
+ "learning_rate": 7.152859492062462e-06,
+ "loss": 0.0163,
+ "step": 22172
+ },
+ {
+ "epoch": 60.416893732970024,
+ "grad_norm": 2.3564653396606445,
+ "learning_rate": 7.152013532768633e-06,
+ "loss": 0.1038,
+ "step": 22173
+ },
+ {
+ "epoch": 60.41961852861036,
+ "grad_norm": 1.8617568016052246,
+ "learning_rate": 7.15116759565483e-06,
+ "loss": 0.0195,
+ "step": 22174
+ },
+ {
+ "epoch": 60.42234332425068,
+ "grad_norm": 2.37644362449646,
+ "learning_rate": 7.150321680727641e-06,
+ "loss": 0.1037,
+ "step": 22175
+ },
+ {
+ "epoch": 60.42506811989101,
+ "grad_norm": 3.2036221027374268,
+ "learning_rate": 7.149475787993658e-06,
+ "loss": 0.0952,
+ "step": 22176
+ },
+ {
+ "epoch": 60.427792915531334,
+ "grad_norm": 2.337826728820801,
+ "learning_rate": 7.148629917459464e-06,
+ "loss": 0.051,
+ "step": 22177
+ },
+ {
+ "epoch": 60.43051771117166,
+ "grad_norm": 1.9196900129318237,
+ "learning_rate": 7.147784069131649e-06,
+ "loss": 0.1518,
+ "step": 22178
+ },
+ {
+ "epoch": 60.433242506811986,
+ "grad_norm": 2.506983518600464,
+ "learning_rate": 7.146938243016799e-06,
+ "loss": 0.105,
+ "step": 22179
+ },
+ {
+ "epoch": 60.43596730245232,
+ "grad_norm": 3.7314817905426025,
+ "learning_rate": 7.146092439121504e-06,
+ "loss": 0.0432,
+ "step": 22180
+ },
+ {
+ "epoch": 60.438692098092645,
+ "grad_norm": 3.04512357711792,
+ "learning_rate": 7.145246657452347e-06,
+ "loss": 0.0274,
+ "step": 22181
+ },
+ {
+ "epoch": 60.44141689373297,
+ "grad_norm": 3.982968330383301,
+ "learning_rate": 7.144400898015919e-06,
+ "loss": 0.0342,
+ "step": 22182
+ },
+ {
+ "epoch": 60.444141689373296,
+ "grad_norm": 2.1075565814971924,
+ "learning_rate": 7.1435551608188035e-06,
+ "loss": 0.0456,
+ "step": 22183
+ },
+ {
+ "epoch": 60.44686648501362,
+ "grad_norm": 1.8748283386230469,
+ "learning_rate": 7.14270944586759e-06,
+ "loss": 0.1121,
+ "step": 22184
+ },
+ {
+ "epoch": 60.44959128065395,
+ "grad_norm": 2.2180426120758057,
+ "learning_rate": 7.141863753168861e-06,
+ "loss": 0.0458,
+ "step": 22185
+ },
+ {
+ "epoch": 60.45231607629428,
+ "grad_norm": 1.587331771850586,
+ "learning_rate": 7.141018082729208e-06,
+ "loss": 0.0248,
+ "step": 22186
+ },
+ {
+ "epoch": 60.45504087193461,
+ "grad_norm": 2.829760789871216,
+ "learning_rate": 7.14017243455521e-06,
+ "loss": 0.0384,
+ "step": 22187
+ },
+ {
+ "epoch": 60.45776566757493,
+ "grad_norm": 2.4166014194488525,
+ "learning_rate": 7.13932680865346e-06,
+ "loss": 0.0332,
+ "step": 22188
+ },
+ {
+ "epoch": 60.46049046321526,
+ "grad_norm": 2.2505035400390625,
+ "learning_rate": 7.1384812050305404e-06,
+ "loss": 0.0604,
+ "step": 22189
+ },
+ {
+ "epoch": 60.463215258855584,
+ "grad_norm": 2.1702616214752197,
+ "learning_rate": 7.137635623693036e-06,
+ "loss": 0.1203,
+ "step": 22190
+ },
+ {
+ "epoch": 60.46594005449591,
+ "grad_norm": 1.9510258436203003,
+ "learning_rate": 7.136790064647536e-06,
+ "loss": 0.0449,
+ "step": 22191
+ },
+ {
+ "epoch": 60.46866485013624,
+ "grad_norm": 1.65886390209198,
+ "learning_rate": 7.1359445279006205e-06,
+ "loss": 0.0225,
+ "step": 22192
+ },
+ {
+ "epoch": 60.47138964577657,
+ "grad_norm": 2.3048312664031982,
+ "learning_rate": 7.135099013458879e-06,
+ "loss": 0.0601,
+ "step": 22193
+ },
+ {
+ "epoch": 60.474114441416894,
+ "grad_norm": 1.559122085571289,
+ "learning_rate": 7.134253521328891e-06,
+ "loss": 0.032,
+ "step": 22194
+ },
+ {
+ "epoch": 60.47683923705722,
+ "grad_norm": 2.7654953002929688,
+ "learning_rate": 7.133408051517247e-06,
+ "loss": 0.0329,
+ "step": 22195
+ },
+ {
+ "epoch": 60.479564032697546,
+ "grad_norm": 2.550490140914917,
+ "learning_rate": 7.1325626040305275e-06,
+ "loss": 0.0746,
+ "step": 22196
+ },
+ {
+ "epoch": 60.48228882833787,
+ "grad_norm": 2.645333766937256,
+ "learning_rate": 7.131717178875317e-06,
+ "loss": 0.0325,
+ "step": 22197
+ },
+ {
+ "epoch": 60.485013623978205,
+ "grad_norm": 2.5199623107910156,
+ "learning_rate": 7.130871776058204e-06,
+ "loss": 0.022,
+ "step": 22198
+ },
+ {
+ "epoch": 60.48773841961853,
+ "grad_norm": 3.751945734024048,
+ "learning_rate": 7.130026395585766e-06,
+ "loss": 0.0331,
+ "step": 22199
+ },
+ {
+ "epoch": 60.490463215258856,
+ "grad_norm": 2.1614956855773926,
+ "learning_rate": 7.129181037464593e-06,
+ "loss": 0.0254,
+ "step": 22200
+ },
+ {
+ "epoch": 60.49318801089918,
+ "grad_norm": 1.9715930223464966,
+ "learning_rate": 7.1283357017012635e-06,
+ "loss": 0.0467,
+ "step": 22201
+ },
+ {
+ "epoch": 60.49591280653951,
+ "grad_norm": 1.836290717124939,
+ "learning_rate": 7.127490388302365e-06,
+ "loss": 0.0659,
+ "step": 22202
+ },
+ {
+ "epoch": 60.49863760217983,
+ "grad_norm": 2.091784954071045,
+ "learning_rate": 7.126645097274477e-06,
+ "loss": 0.036,
+ "step": 22203
+ },
+ {
+ "epoch": 60.50136239782017,
+ "grad_norm": 2.0912435054779053,
+ "learning_rate": 7.125799828624187e-06,
+ "loss": 0.0266,
+ "step": 22204
+ },
+ {
+ "epoch": 60.50408719346049,
+ "grad_norm": 2.0901739597320557,
+ "learning_rate": 7.12495458235807e-06,
+ "loss": 0.0413,
+ "step": 22205
+ },
+ {
+ "epoch": 60.50681198910082,
+ "grad_norm": 2.268709897994995,
+ "learning_rate": 7.1241093584827185e-06,
+ "loss": 0.0244,
+ "step": 22206
+ },
+ {
+ "epoch": 60.509536784741144,
+ "grad_norm": 2.412081241607666,
+ "learning_rate": 7.12326415700471e-06,
+ "loss": 0.027,
+ "step": 22207
+ },
+ {
+ "epoch": 60.51226158038147,
+ "grad_norm": 1.4695374965667725,
+ "learning_rate": 7.122418977930628e-06,
+ "loss": 0.0579,
+ "step": 22208
+ },
+ {
+ "epoch": 60.514986376021795,
+ "grad_norm": 1.563288927078247,
+ "learning_rate": 7.121573821267055e-06,
+ "loss": 0.097,
+ "step": 22209
+ },
+ {
+ "epoch": 60.51771117166213,
+ "grad_norm": 2.2236452102661133,
+ "learning_rate": 7.120728687020571e-06,
+ "loss": 0.0454,
+ "step": 22210
+ },
+ {
+ "epoch": 60.520435967302454,
+ "grad_norm": 2.2350666522979736,
+ "learning_rate": 7.1198835751977615e-06,
+ "loss": 0.0703,
+ "step": 22211
+ },
+ {
+ "epoch": 60.52316076294278,
+ "grad_norm": 1.6844977140426636,
+ "learning_rate": 7.119038485805203e-06,
+ "loss": 0.0702,
+ "step": 22212
+ },
+ {
+ "epoch": 60.525885558583106,
+ "grad_norm": 2.3411245346069336,
+ "learning_rate": 7.118193418849483e-06,
+ "loss": 0.0346,
+ "step": 22213
+ },
+ {
+ "epoch": 60.52861035422343,
+ "grad_norm": 10.103995323181152,
+ "learning_rate": 7.1173483743371775e-06,
+ "loss": 0.0277,
+ "step": 22214
+ },
+ {
+ "epoch": 60.53133514986376,
+ "grad_norm": 2.0480849742889404,
+ "learning_rate": 7.116503352274873e-06,
+ "loss": 0.0272,
+ "step": 22215
+ },
+ {
+ "epoch": 60.53405994550409,
+ "grad_norm": 2.512927532196045,
+ "learning_rate": 7.115658352669143e-06,
+ "loss": 0.027,
+ "step": 22216
+ },
+ {
+ "epoch": 60.536784741144416,
+ "grad_norm": 7.095827102661133,
+ "learning_rate": 7.114813375526577e-06,
+ "loss": 0.0352,
+ "step": 22217
+ },
+ {
+ "epoch": 60.53950953678474,
+ "grad_norm": 2.0549135208129883,
+ "learning_rate": 7.113968420853749e-06,
+ "loss": 0.0236,
+ "step": 22218
+ },
+ {
+ "epoch": 60.54223433242507,
+ "grad_norm": 2.486626148223877,
+ "learning_rate": 7.113123488657243e-06,
+ "loss": 0.124,
+ "step": 22219
+ },
+ {
+ "epoch": 60.54495912806539,
+ "grad_norm": 1.6475368738174438,
+ "learning_rate": 7.11227857894364e-06,
+ "loss": 0.0269,
+ "step": 22220
+ },
+ {
+ "epoch": 60.54768392370572,
+ "grad_norm": 2.2917754650115967,
+ "learning_rate": 7.111433691719516e-06,
+ "loss": 0.0199,
+ "step": 22221
+ },
+ {
+ "epoch": 60.55040871934605,
+ "grad_norm": 1.5521379709243774,
+ "learning_rate": 7.110588826991451e-06,
+ "loss": 0.0197,
+ "step": 22222
+ },
+ {
+ "epoch": 60.55313351498638,
+ "grad_norm": 2.5787289142608643,
+ "learning_rate": 7.109743984766032e-06,
+ "loss": 0.0536,
+ "step": 22223
+ },
+ {
+ "epoch": 60.555858310626704,
+ "grad_norm": 2.678534507751465,
+ "learning_rate": 7.108899165049832e-06,
+ "loss": 0.0476,
+ "step": 22224
+ },
+ {
+ "epoch": 60.55858310626703,
+ "grad_norm": 2.090538740158081,
+ "learning_rate": 7.108054367849434e-06,
+ "loss": 0.091,
+ "step": 22225
+ },
+ {
+ "epoch": 60.561307901907355,
+ "grad_norm": 2.1621406078338623,
+ "learning_rate": 7.107209593171414e-06,
+ "loss": 0.0316,
+ "step": 22226
+ },
+ {
+ "epoch": 60.56403269754768,
+ "grad_norm": 1.7134954929351807,
+ "learning_rate": 7.106364841022353e-06,
+ "loss": 0.0868,
+ "step": 22227
+ },
+ {
+ "epoch": 60.566757493188014,
+ "grad_norm": 2.216686487197876,
+ "learning_rate": 7.105520111408828e-06,
+ "loss": 0.0302,
+ "step": 22228
+ },
+ {
+ "epoch": 60.56948228882834,
+ "grad_norm": 1.7230137586593628,
+ "learning_rate": 7.104675404337422e-06,
+ "loss": 0.0277,
+ "step": 22229
+ },
+ {
+ "epoch": 60.572207084468666,
+ "grad_norm": 2.068504571914673,
+ "learning_rate": 7.103830719814708e-06,
+ "loss": 0.0386,
+ "step": 22230
+ },
+ {
+ "epoch": 60.57493188010899,
+ "grad_norm": 3.2674307823181152,
+ "learning_rate": 7.102986057847269e-06,
+ "loss": 0.0296,
+ "step": 22231
+ },
+ {
+ "epoch": 60.57765667574932,
+ "grad_norm": 2.118699789047241,
+ "learning_rate": 7.102141418441681e-06,
+ "loss": 0.1156,
+ "step": 22232
+ },
+ {
+ "epoch": 60.58038147138964,
+ "grad_norm": 3.2428271770477295,
+ "learning_rate": 7.101296801604525e-06,
+ "loss": 0.084,
+ "step": 22233
+ },
+ {
+ "epoch": 60.583106267029976,
+ "grad_norm": 1.9198036193847656,
+ "learning_rate": 7.100452207342372e-06,
+ "loss": 0.027,
+ "step": 22234
+ },
+ {
+ "epoch": 60.5858310626703,
+ "grad_norm": 1.9519309997558594,
+ "learning_rate": 7.099607635661806e-06,
+ "loss": 0.0257,
+ "step": 22235
+ },
+ {
+ "epoch": 60.58855585831063,
+ "grad_norm": 2.4559638500213623,
+ "learning_rate": 7.0987630865694e-06,
+ "loss": 0.0643,
+ "step": 22236
+ },
+ {
+ "epoch": 60.59128065395095,
+ "grad_norm": 3.5479865074157715,
+ "learning_rate": 7.097918560071737e-06,
+ "loss": 0.1305,
+ "step": 22237
+ },
+ {
+ "epoch": 60.59400544959128,
+ "grad_norm": 2.116528272628784,
+ "learning_rate": 7.0970740561753884e-06,
+ "loss": 0.0846,
+ "step": 22238
+ },
+ {
+ "epoch": 60.596730245231605,
+ "grad_norm": 1.9751482009887695,
+ "learning_rate": 7.0962295748869314e-06,
+ "loss": 0.1671,
+ "step": 22239
+ },
+ {
+ "epoch": 60.59945504087194,
+ "grad_norm": 3.3659658432006836,
+ "learning_rate": 7.095385116212947e-06,
+ "loss": 0.1114,
+ "step": 22240
+ },
+ {
+ "epoch": 60.60217983651226,
+ "grad_norm": 2.8783650398254395,
+ "learning_rate": 7.094540680160013e-06,
+ "loss": 0.1019,
+ "step": 22241
+ },
+ {
+ "epoch": 60.60490463215259,
+ "grad_norm": 1.7179865837097168,
+ "learning_rate": 7.0936962667347e-06,
+ "loss": 0.0237,
+ "step": 22242
+ },
+ {
+ "epoch": 60.607629427792915,
+ "grad_norm": 2.4205119609832764,
+ "learning_rate": 7.092851875943587e-06,
+ "loss": 0.0855,
+ "step": 22243
+ },
+ {
+ "epoch": 60.61035422343324,
+ "grad_norm": 1.9008276462554932,
+ "learning_rate": 7.09200750779325e-06,
+ "loss": 0.0182,
+ "step": 22244
+ },
+ {
+ "epoch": 60.61307901907357,
+ "grad_norm": 1.7630565166473389,
+ "learning_rate": 7.091163162290266e-06,
+ "loss": 0.0234,
+ "step": 22245
+ },
+ {
+ "epoch": 60.6158038147139,
+ "grad_norm": 2.4510951042175293,
+ "learning_rate": 7.090318839441209e-06,
+ "loss": 0.0688,
+ "step": 22246
+ },
+ {
+ "epoch": 60.618528610354225,
+ "grad_norm": 1.9586855173110962,
+ "learning_rate": 7.089474539252656e-06,
+ "loss": 0.0855,
+ "step": 22247
+ },
+ {
+ "epoch": 60.62125340599455,
+ "grad_norm": 6.366184711456299,
+ "learning_rate": 7.088630261731181e-06,
+ "loss": 0.0958,
+ "step": 22248
+ },
+ {
+ "epoch": 60.62397820163488,
+ "grad_norm": 1.6478339433670044,
+ "learning_rate": 7.087786006883362e-06,
+ "loss": 0.0518,
+ "step": 22249
+ },
+ {
+ "epoch": 60.6267029972752,
+ "grad_norm": 3.137711763381958,
+ "learning_rate": 7.086941774715767e-06,
+ "loss": 0.0489,
+ "step": 22250
+ },
+ {
+ "epoch": 60.62942779291553,
+ "grad_norm": 2.1345691680908203,
+ "learning_rate": 7.08609756523498e-06,
+ "loss": 0.026,
+ "step": 22251
+ },
+ {
+ "epoch": 60.63215258855586,
+ "grad_norm": 3.0977253913879395,
+ "learning_rate": 7.0852533784475695e-06,
+ "loss": 0.0965,
+ "step": 22252
+ },
+ {
+ "epoch": 60.63487738419619,
+ "grad_norm": 1.679879903793335,
+ "learning_rate": 7.084409214360113e-06,
+ "loss": 0.0265,
+ "step": 22253
+ },
+ {
+ "epoch": 60.63760217983651,
+ "grad_norm": 2.303305149078369,
+ "learning_rate": 7.083565072979182e-06,
+ "loss": 0.0376,
+ "step": 22254
+ },
+ {
+ "epoch": 60.64032697547684,
+ "grad_norm": 2.2975454330444336,
+ "learning_rate": 7.082720954311354e-06,
+ "loss": 0.0249,
+ "step": 22255
+ },
+ {
+ "epoch": 60.643051771117165,
+ "grad_norm": 2.331575393676758,
+ "learning_rate": 7.081876858363197e-06,
+ "loss": 0.0409,
+ "step": 22256
+ },
+ {
+ "epoch": 60.64577656675749,
+ "grad_norm": 2.2383158206939697,
+ "learning_rate": 7.081032785141292e-06,
+ "loss": 0.1633,
+ "step": 22257
+ },
+ {
+ "epoch": 60.64850136239782,
+ "grad_norm": 2.748145341873169,
+ "learning_rate": 7.0801887346522116e-06,
+ "loss": 0.0432,
+ "step": 22258
+ },
+ {
+ "epoch": 60.65122615803815,
+ "grad_norm": 1.5489648580551147,
+ "learning_rate": 7.079344706902525e-06,
+ "loss": 0.0176,
+ "step": 22259
+ },
+ {
+ "epoch": 60.653950953678475,
+ "grad_norm": 2.4607839584350586,
+ "learning_rate": 7.0785007018988095e-06,
+ "loss": 0.0221,
+ "step": 22260
+ },
+ {
+ "epoch": 60.6566757493188,
+ "grad_norm": 2.3340868949890137,
+ "learning_rate": 7.077656719647636e-06,
+ "loss": 0.0795,
+ "step": 22261
+ },
+ {
+ "epoch": 60.65940054495913,
+ "grad_norm": 1.499947428703308,
+ "learning_rate": 7.076812760155576e-06,
+ "loss": 0.0128,
+ "step": 22262
+ },
+ {
+ "epoch": 60.66212534059945,
+ "grad_norm": 2.951647996902466,
+ "learning_rate": 7.075968823429207e-06,
+ "loss": 0.0525,
+ "step": 22263
+ },
+ {
+ "epoch": 60.664850136239785,
+ "grad_norm": 1.8586317300796509,
+ "learning_rate": 7.075124909475097e-06,
+ "loss": 0.0243,
+ "step": 22264
+ },
+ {
+ "epoch": 60.66757493188011,
+ "grad_norm": 1.4509999752044678,
+ "learning_rate": 7.074281018299822e-06,
+ "loss": 0.0363,
+ "step": 22265
+ },
+ {
+ "epoch": 60.67029972752044,
+ "grad_norm": 2.150059938430786,
+ "learning_rate": 7.07343714990995e-06,
+ "loss": 0.0232,
+ "step": 22266
+ },
+ {
+ "epoch": 60.67302452316076,
+ "grad_norm": 4.438474655151367,
+ "learning_rate": 7.072593304312057e-06,
+ "loss": 0.0316,
+ "step": 22267
+ },
+ {
+ "epoch": 60.67574931880109,
+ "grad_norm": 2.190931558609009,
+ "learning_rate": 7.071749481512712e-06,
+ "loss": 0.0398,
+ "step": 22268
+ },
+ {
+ "epoch": 60.678474114441414,
+ "grad_norm": 2.6557021141052246,
+ "learning_rate": 7.070905681518489e-06,
+ "loss": 0.1554,
+ "step": 22269
+ },
+ {
+ "epoch": 60.68119891008175,
+ "grad_norm": 2.165910482406616,
+ "learning_rate": 7.070061904335956e-06,
+ "loss": 0.0373,
+ "step": 22270
+ },
+ {
+ "epoch": 60.68392370572207,
+ "grad_norm": 2.312858819961548,
+ "learning_rate": 7.0692181499716905e-06,
+ "loss": 0.0326,
+ "step": 22271
+ },
+ {
+ "epoch": 60.6866485013624,
+ "grad_norm": 2.0056562423706055,
+ "learning_rate": 7.0683744184322554e-06,
+ "loss": 0.0289,
+ "step": 22272
+ },
+ {
+ "epoch": 60.689373297002724,
+ "grad_norm": 2.6399388313293457,
+ "learning_rate": 7.067530709724226e-06,
+ "loss": 0.1301,
+ "step": 22273
+ },
+ {
+ "epoch": 60.69209809264305,
+ "grad_norm": 3.8146791458129883,
+ "learning_rate": 7.066687023854175e-06,
+ "loss": 0.0372,
+ "step": 22274
+ },
+ {
+ "epoch": 60.694822888283376,
+ "grad_norm": 2.54620099067688,
+ "learning_rate": 7.06584336082867e-06,
+ "loss": 0.0288,
+ "step": 22275
+ },
+ {
+ "epoch": 60.69754768392371,
+ "grad_norm": 3.5424695014953613,
+ "learning_rate": 7.0649997206542855e-06,
+ "loss": 0.0514,
+ "step": 22276
+ },
+ {
+ "epoch": 60.700272479564035,
+ "grad_norm": 2.2829227447509766,
+ "learning_rate": 7.064156103337585e-06,
+ "loss": 0.0686,
+ "step": 22277
+ },
+ {
+ "epoch": 60.70299727520436,
+ "grad_norm": 2.9107346534729004,
+ "learning_rate": 7.063312508885145e-06,
+ "loss": 0.0862,
+ "step": 22278
+ },
+ {
+ "epoch": 60.705722070844686,
+ "grad_norm": 1.8104077577590942,
+ "learning_rate": 7.06246893730353e-06,
+ "loss": 0.0332,
+ "step": 22279
+ },
+ {
+ "epoch": 60.70844686648501,
+ "grad_norm": 2.4279370307922363,
+ "learning_rate": 7.061625388599316e-06,
+ "loss": 0.043,
+ "step": 22280
+ },
+ {
+ "epoch": 60.71117166212534,
+ "grad_norm": 1.9249563217163086,
+ "learning_rate": 7.060781862779067e-06,
+ "loss": 0.0367,
+ "step": 22281
+ },
+ {
+ "epoch": 60.71389645776567,
+ "grad_norm": 2.2628350257873535,
+ "learning_rate": 7.0599383598493545e-06,
+ "loss": 0.0274,
+ "step": 22282
+ },
+ {
+ "epoch": 60.716621253406,
+ "grad_norm": 2.2239556312561035,
+ "learning_rate": 7.059094879816746e-06,
+ "loss": 0.0411,
+ "step": 22283
+ },
+ {
+ "epoch": 60.71934604904632,
+ "grad_norm": 2.8584325313568115,
+ "learning_rate": 7.058251422687812e-06,
+ "loss": 0.0591,
+ "step": 22284
+ },
+ {
+ "epoch": 60.72207084468665,
+ "grad_norm": 2.03960919380188,
+ "learning_rate": 7.057407988469123e-06,
+ "loss": 0.1004,
+ "step": 22285
+ },
+ {
+ "epoch": 60.724795640326974,
+ "grad_norm": 5.233191967010498,
+ "learning_rate": 7.056564577167244e-06,
+ "loss": 0.0386,
+ "step": 22286
+ },
+ {
+ "epoch": 60.7275204359673,
+ "grad_norm": 2.008301258087158,
+ "learning_rate": 7.055721188788749e-06,
+ "loss": 0.0319,
+ "step": 22287
+ },
+ {
+ "epoch": 60.73024523160763,
+ "grad_norm": 2.832472085952759,
+ "learning_rate": 7.054877823340199e-06,
+ "loss": 0.1176,
+ "step": 22288
+ },
+ {
+ "epoch": 60.73297002724796,
+ "grad_norm": 2.9371817111968994,
+ "learning_rate": 7.054034480828167e-06,
+ "loss": 0.0495,
+ "step": 22289
+ },
+ {
+ "epoch": 60.735694822888284,
+ "grad_norm": 2.4106454849243164,
+ "learning_rate": 7.0531911612592184e-06,
+ "loss": 0.0705,
+ "step": 22290
+ },
+ {
+ "epoch": 60.73841961852861,
+ "grad_norm": 2.2850213050842285,
+ "learning_rate": 7.052347864639921e-06,
+ "loss": 0.0213,
+ "step": 22291
+ },
+ {
+ "epoch": 60.741144414168936,
+ "grad_norm": 2.565596103668213,
+ "learning_rate": 7.051504590976847e-06,
+ "loss": 0.0323,
+ "step": 22292
+ },
+ {
+ "epoch": 60.74386920980926,
+ "grad_norm": 2.4593546390533447,
+ "learning_rate": 7.050661340276558e-06,
+ "loss": 0.0228,
+ "step": 22293
+ },
+ {
+ "epoch": 60.746594005449595,
+ "grad_norm": 3.565633773803711,
+ "learning_rate": 7.049818112545626e-06,
+ "loss": 0.019,
+ "step": 22294
+ },
+ {
+ "epoch": 60.74931880108992,
+ "grad_norm": 2.257486581802368,
+ "learning_rate": 7.048974907790613e-06,
+ "loss": 0.06,
+ "step": 22295
+ },
+ {
+ "epoch": 60.752043596730246,
+ "grad_norm": 1.1748231649398804,
+ "learning_rate": 7.0481317260180905e-06,
+ "loss": 0.0162,
+ "step": 22296
+ },
+ {
+ "epoch": 60.75476839237057,
+ "grad_norm": 2.6165881156921387,
+ "learning_rate": 7.047288567234621e-06,
+ "loss": 0.0457,
+ "step": 22297
+ },
+ {
+ "epoch": 60.7574931880109,
+ "grad_norm": 1.5514558553695679,
+ "learning_rate": 7.046445431446775e-06,
+ "loss": 0.0235,
+ "step": 22298
+ },
+ {
+ "epoch": 60.76021798365122,
+ "grad_norm": 3.0405237674713135,
+ "learning_rate": 7.045602318661117e-06,
+ "loss": 0.0489,
+ "step": 22299
+ },
+ {
+ "epoch": 60.762942779291556,
+ "grad_norm": 3.606215238571167,
+ "learning_rate": 7.044759228884212e-06,
+ "loss": 0.053,
+ "step": 22300
+ },
+ {
+ "epoch": 60.76566757493188,
+ "grad_norm": 3.857496976852417,
+ "learning_rate": 7.043916162122628e-06,
+ "loss": 0.0252,
+ "step": 22301
+ },
+ {
+ "epoch": 60.76839237057221,
+ "grad_norm": 1.5116370916366577,
+ "learning_rate": 7.0430731183829305e-06,
+ "loss": 0.1278,
+ "step": 22302
+ },
+ {
+ "epoch": 60.771117166212534,
+ "grad_norm": 1.5944349765777588,
+ "learning_rate": 7.042230097671681e-06,
+ "loss": 0.0305,
+ "step": 22303
+ },
+ {
+ "epoch": 60.77384196185286,
+ "grad_norm": 1.8664342164993286,
+ "learning_rate": 7.0413870999954525e-06,
+ "loss": 0.023,
+ "step": 22304
+ },
+ {
+ "epoch": 60.776566757493185,
+ "grad_norm": 1.9926131963729858,
+ "learning_rate": 7.040544125360802e-06,
+ "loss": 0.041,
+ "step": 22305
+ },
+ {
+ "epoch": 60.77929155313352,
+ "grad_norm": 2.3918869495391846,
+ "learning_rate": 7.0397011737743e-06,
+ "loss": 0.0264,
+ "step": 22306
+ },
+ {
+ "epoch": 60.782016348773844,
+ "grad_norm": 3.3346316814422607,
+ "learning_rate": 7.038858245242513e-06,
+ "loss": 0.0744,
+ "step": 22307
+ },
+ {
+ "epoch": 60.78474114441417,
+ "grad_norm": 2.254549980163574,
+ "learning_rate": 7.038015339771996e-06,
+ "loss": 0.0657,
+ "step": 22308
+ },
+ {
+ "epoch": 60.787465940054496,
+ "grad_norm": 1.2482938766479492,
+ "learning_rate": 7.037172457369324e-06,
+ "loss": 0.0159,
+ "step": 22309
+ },
+ {
+ "epoch": 60.79019073569482,
+ "grad_norm": 1.7642590999603271,
+ "learning_rate": 7.036329598041058e-06,
+ "loss": 0.0302,
+ "step": 22310
+ },
+ {
+ "epoch": 60.79291553133515,
+ "grad_norm": 2.210522413253784,
+ "learning_rate": 7.035486761793762e-06,
+ "loss": 0.0781,
+ "step": 22311
+ },
+ {
+ "epoch": 60.79564032697548,
+ "grad_norm": 1.7688322067260742,
+ "learning_rate": 7.034643948634e-06,
+ "loss": 0.0296,
+ "step": 22312
+ },
+ {
+ "epoch": 60.798365122615806,
+ "grad_norm": 1.9997386932373047,
+ "learning_rate": 7.033801158568334e-06,
+ "loss": 0.0288,
+ "step": 22313
+ },
+ {
+ "epoch": 60.80108991825613,
+ "grad_norm": 1.619143009185791,
+ "learning_rate": 7.03295839160333e-06,
+ "loss": 0.1086,
+ "step": 22314
+ },
+ {
+ "epoch": 60.80381471389646,
+ "grad_norm": 1.7917600870132446,
+ "learning_rate": 7.03211564774555e-06,
+ "loss": 0.065,
+ "step": 22315
+ },
+ {
+ "epoch": 60.80653950953678,
+ "grad_norm": 1.8518394231796265,
+ "learning_rate": 7.03127292700156e-06,
+ "loss": 0.0447,
+ "step": 22316
+ },
+ {
+ "epoch": 60.80926430517711,
+ "grad_norm": 4.026852607727051,
+ "learning_rate": 7.030430229377918e-06,
+ "loss": 0.0307,
+ "step": 22317
+ },
+ {
+ "epoch": 60.81198910081744,
+ "grad_norm": 1.493260145187378,
+ "learning_rate": 7.029587554881192e-06,
+ "loss": 0.0236,
+ "step": 22318
+ },
+ {
+ "epoch": 60.81471389645777,
+ "grad_norm": 1.8024816513061523,
+ "learning_rate": 7.028744903517941e-06,
+ "loss": 0.0401,
+ "step": 22319
+ },
+ {
+ "epoch": 60.817438692098094,
+ "grad_norm": 2.528129816055298,
+ "learning_rate": 7.027902275294731e-06,
+ "loss": 0.0722,
+ "step": 22320
+ },
+ {
+ "epoch": 60.82016348773842,
+ "grad_norm": 1.839281678199768,
+ "learning_rate": 7.027059670218119e-06,
+ "loss": 0.1508,
+ "step": 22321
+ },
+ {
+ "epoch": 60.822888283378745,
+ "grad_norm": 2.4776031970977783,
+ "learning_rate": 7.026217088294674e-06,
+ "loss": 0.0378,
+ "step": 22322
+ },
+ {
+ "epoch": 60.82561307901907,
+ "grad_norm": 2.0078771114349365,
+ "learning_rate": 7.025374529530952e-06,
+ "loss": 0.128,
+ "step": 22323
+ },
+ {
+ "epoch": 60.828337874659404,
+ "grad_norm": 1.797383189201355,
+ "learning_rate": 7.024531993933521e-06,
+ "loss": 0.0403,
+ "step": 22324
+ },
+ {
+ "epoch": 60.83106267029973,
+ "grad_norm": 3.189854860305786,
+ "learning_rate": 7.023689481508931e-06,
+ "loss": 0.054,
+ "step": 22325
+ },
+ {
+ "epoch": 60.833787465940055,
+ "grad_norm": 2.0537712574005127,
+ "learning_rate": 7.0228469922637586e-06,
+ "loss": 0.0349,
+ "step": 22326
+ },
+ {
+ "epoch": 60.83651226158038,
+ "grad_norm": 1.9602528810501099,
+ "learning_rate": 7.0220045262045555e-06,
+ "loss": 0.0334,
+ "step": 22327
+ },
+ {
+ "epoch": 60.83923705722071,
+ "grad_norm": 2.797947883605957,
+ "learning_rate": 7.021162083337888e-06,
+ "loss": 0.1596,
+ "step": 22328
+ },
+ {
+ "epoch": 60.84196185286103,
+ "grad_norm": 2.406471014022827,
+ "learning_rate": 7.020319663670312e-06,
+ "loss": 0.0435,
+ "step": 22329
+ },
+ {
+ "epoch": 60.844686648501366,
+ "grad_norm": 2.0497593879699707,
+ "learning_rate": 7.019477267208392e-06,
+ "loss": 0.0415,
+ "step": 22330
+ },
+ {
+ "epoch": 60.84741144414169,
+ "grad_norm": 4.106441020965576,
+ "learning_rate": 7.018634893958686e-06,
+ "loss": 0.1734,
+ "step": 22331
+ },
+ {
+ "epoch": 60.85013623978202,
+ "grad_norm": 1.350679874420166,
+ "learning_rate": 7.0177925439277574e-06,
+ "loss": 0.0198,
+ "step": 22332
+ },
+ {
+ "epoch": 60.85286103542234,
+ "grad_norm": 2.3159544467926025,
+ "learning_rate": 7.016950217122162e-06,
+ "loss": 0.0536,
+ "step": 22333
+ },
+ {
+ "epoch": 60.85558583106267,
+ "grad_norm": 1.7440626621246338,
+ "learning_rate": 7.016107913548465e-06,
+ "loss": 0.0478,
+ "step": 22334
+ },
+ {
+ "epoch": 60.858310626702995,
+ "grad_norm": 1.7568151950836182,
+ "learning_rate": 7.01526563321322e-06,
+ "loss": 0.1011,
+ "step": 22335
+ },
+ {
+ "epoch": 60.86103542234333,
+ "grad_norm": 1.8500936031341553,
+ "learning_rate": 7.014423376122993e-06,
+ "loss": 0.073,
+ "step": 22336
+ },
+ {
+ "epoch": 60.86376021798365,
+ "grad_norm": 2.206705093383789,
+ "learning_rate": 7.0135811422843384e-06,
+ "loss": 0.0201,
+ "step": 22337
+ },
+ {
+ "epoch": 60.86648501362398,
+ "grad_norm": 2.035299062728882,
+ "learning_rate": 7.0127389317038205e-06,
+ "loss": 0.0442,
+ "step": 22338
+ },
+ {
+ "epoch": 60.869209809264305,
+ "grad_norm": 1.8242703676223755,
+ "learning_rate": 7.011896744387993e-06,
+ "loss": 0.0465,
+ "step": 22339
+ },
+ {
+ "epoch": 60.87193460490463,
+ "grad_norm": 2.9132516384124756,
+ "learning_rate": 7.011054580343421e-06,
+ "loss": 0.0338,
+ "step": 22340
+ },
+ {
+ "epoch": 60.87465940054496,
+ "grad_norm": 4.505248546600342,
+ "learning_rate": 7.010212439576656e-06,
+ "loss": 0.0389,
+ "step": 22341
+ },
+ {
+ "epoch": 60.87738419618529,
+ "grad_norm": 2.416707992553711,
+ "learning_rate": 7.009370322094258e-06,
+ "loss": 0.0395,
+ "step": 22342
+ },
+ {
+ "epoch": 60.880108991825615,
+ "grad_norm": 1.7231948375701904,
+ "learning_rate": 7.008528227902793e-06,
+ "loss": 0.0768,
+ "step": 22343
+ },
+ {
+ "epoch": 60.88283378746594,
+ "grad_norm": 1.5866787433624268,
+ "learning_rate": 7.00768615700881e-06,
+ "loss": 0.0181,
+ "step": 22344
+ },
+ {
+ "epoch": 60.88555858310627,
+ "grad_norm": 1.5507469177246094,
+ "learning_rate": 7.006844109418876e-06,
+ "loss": 0.0375,
+ "step": 22345
+ },
+ {
+ "epoch": 60.88828337874659,
+ "grad_norm": 2.1541810035705566,
+ "learning_rate": 7.0060020851395395e-06,
+ "loss": 0.182,
+ "step": 22346
+ },
+ {
+ "epoch": 60.89100817438692,
+ "grad_norm": 1.899961233139038,
+ "learning_rate": 7.005160084177365e-06,
+ "loss": 0.0797,
+ "step": 22347
+ },
+ {
+ "epoch": 60.89373297002725,
+ "grad_norm": 1.87232506275177,
+ "learning_rate": 7.0043181065389055e-06,
+ "loss": 0.0373,
+ "step": 22348
+ },
+ {
+ "epoch": 60.89645776566758,
+ "grad_norm": 1.7813332080841064,
+ "learning_rate": 7.003476152230719e-06,
+ "loss": 0.27,
+ "step": 22349
+ },
+ {
+ "epoch": 60.8991825613079,
+ "grad_norm": 2.217878580093384,
+ "learning_rate": 7.0026342212593674e-06,
+ "loss": 0.0297,
+ "step": 22350
+ },
+ {
+ "epoch": 60.90190735694823,
+ "grad_norm": 2.223421573638916,
+ "learning_rate": 7.001792313631401e-06,
+ "loss": 0.0372,
+ "step": 22351
+ },
+ {
+ "epoch": 60.904632152588555,
+ "grad_norm": 2.607304096221924,
+ "learning_rate": 7.000950429353382e-06,
+ "loss": 0.0232,
+ "step": 22352
+ },
+ {
+ "epoch": 60.90735694822888,
+ "grad_norm": 1.8581109046936035,
+ "learning_rate": 7.000108568431863e-06,
+ "loss": 0.0248,
+ "step": 22353
+ },
+ {
+ "epoch": 60.91008174386921,
+ "grad_norm": 1.2242933511734009,
+ "learning_rate": 6.999266730873403e-06,
+ "loss": 0.0165,
+ "step": 22354
+ },
+ {
+ "epoch": 60.91280653950954,
+ "grad_norm": 1.8398268222808838,
+ "learning_rate": 6.998424916684557e-06,
+ "loss": 0.0831,
+ "step": 22355
+ },
+ {
+ "epoch": 60.915531335149865,
+ "grad_norm": 2.9094533920288086,
+ "learning_rate": 6.997583125871881e-06,
+ "loss": 0.0218,
+ "step": 22356
+ },
+ {
+ "epoch": 60.91825613079019,
+ "grad_norm": 2.6210992336273193,
+ "learning_rate": 6.996741358441928e-06,
+ "loss": 0.0274,
+ "step": 22357
+ },
+ {
+ "epoch": 60.920980926430516,
+ "grad_norm": 4.3155903816223145,
+ "learning_rate": 6.995899614401261e-06,
+ "loss": 0.0474,
+ "step": 22358
+ },
+ {
+ "epoch": 60.92370572207084,
+ "grad_norm": 2.318830728530884,
+ "learning_rate": 6.9950578937564265e-06,
+ "loss": 0.0848,
+ "step": 22359
+ },
+ {
+ "epoch": 60.926430517711175,
+ "grad_norm": 1.3673741817474365,
+ "learning_rate": 6.994216196513985e-06,
+ "loss": 0.0294,
+ "step": 22360
+ },
+ {
+ "epoch": 60.9291553133515,
+ "grad_norm": 2.0711772441864014,
+ "learning_rate": 6.9933745226804965e-06,
+ "loss": 0.0452,
+ "step": 22361
+ },
+ {
+ "epoch": 60.93188010899183,
+ "grad_norm": 3.1061627864837646,
+ "learning_rate": 6.992532872262505e-06,
+ "loss": 0.0755,
+ "step": 22362
+ },
+ {
+ "epoch": 60.93460490463215,
+ "grad_norm": 1.7958829402923584,
+ "learning_rate": 6.991691245266574e-06,
+ "loss": 0.0195,
+ "step": 22363
+ },
+ {
+ "epoch": 60.93732970027248,
+ "grad_norm": 1.4402395486831665,
+ "learning_rate": 6.990849641699254e-06,
+ "loss": 0.0244,
+ "step": 22364
+ },
+ {
+ "epoch": 60.940054495912804,
+ "grad_norm": 1.9573312997817993,
+ "learning_rate": 6.9900080615671e-06,
+ "loss": 0.049,
+ "step": 22365
+ },
+ {
+ "epoch": 60.94277929155314,
+ "grad_norm": 2.0000557899475098,
+ "learning_rate": 6.989166504876665e-06,
+ "loss": 0.0388,
+ "step": 22366
+ },
+ {
+ "epoch": 60.94550408719346,
+ "grad_norm": 1.465518593788147,
+ "learning_rate": 6.988324971634506e-06,
+ "loss": 0.022,
+ "step": 22367
+ },
+ {
+ "epoch": 60.94822888283379,
+ "grad_norm": 2.6160800457000732,
+ "learning_rate": 6.987483461847174e-06,
+ "loss": 0.0662,
+ "step": 22368
+ },
+ {
+ "epoch": 60.950953678474114,
+ "grad_norm": 3.072357416152954,
+ "learning_rate": 6.9866419755212264e-06,
+ "loss": 0.035,
+ "step": 22369
+ },
+ {
+ "epoch": 60.95367847411444,
+ "grad_norm": 1.7659028768539429,
+ "learning_rate": 6.98580051266321e-06,
+ "loss": 0.0389,
+ "step": 22370
+ },
+ {
+ "epoch": 60.956403269754766,
+ "grad_norm": 1.598663091659546,
+ "learning_rate": 6.9849590732796845e-06,
+ "loss": 0.0228,
+ "step": 22371
+ },
+ {
+ "epoch": 60.95912806539509,
+ "grad_norm": 2.5422909259796143,
+ "learning_rate": 6.9841176573772005e-06,
+ "loss": 0.138,
+ "step": 22372
+ },
+ {
+ "epoch": 60.961852861035425,
+ "grad_norm": 2.991769790649414,
+ "learning_rate": 6.983276264962311e-06,
+ "loss": 0.0521,
+ "step": 22373
+ },
+ {
+ "epoch": 60.96457765667575,
+ "grad_norm": 1.9579100608825684,
+ "learning_rate": 6.98243489604157e-06,
+ "loss": 0.0269,
+ "step": 22374
+ },
+ {
+ "epoch": 60.967302452316076,
+ "grad_norm": 1.9248785972595215,
+ "learning_rate": 6.981593550621528e-06,
+ "loss": 0.0348,
+ "step": 22375
+ },
+ {
+ "epoch": 60.9700272479564,
+ "grad_norm": 1.4020494222640991,
+ "learning_rate": 6.980752228708738e-06,
+ "loss": 0.0126,
+ "step": 22376
+ },
+ {
+ "epoch": 60.97275204359673,
+ "grad_norm": 1.6798808574676514,
+ "learning_rate": 6.97991093030975e-06,
+ "loss": 0.02,
+ "step": 22377
+ },
+ {
+ "epoch": 60.97547683923706,
+ "grad_norm": 2.283238410949707,
+ "learning_rate": 6.979069655431121e-06,
+ "loss": 0.0308,
+ "step": 22378
+ },
+ {
+ "epoch": 60.97820163487739,
+ "grad_norm": 1.5575485229492188,
+ "learning_rate": 6.978228404079401e-06,
+ "loss": 0.0182,
+ "step": 22379
+ },
+ {
+ "epoch": 60.98092643051771,
+ "grad_norm": 2.1952030658721924,
+ "learning_rate": 6.97738717626114e-06,
+ "loss": 0.0224,
+ "step": 22380
+ },
+ {
+ "epoch": 60.98365122615804,
+ "grad_norm": 2.241351842880249,
+ "learning_rate": 6.976545971982892e-06,
+ "loss": 0.0514,
+ "step": 22381
+ },
+ {
+ "epoch": 60.986376021798364,
+ "grad_norm": 2.459512948989868,
+ "learning_rate": 6.975704791251205e-06,
+ "loss": 0.0375,
+ "step": 22382
+ },
+ {
+ "epoch": 60.98910081743869,
+ "grad_norm": 1.9750030040740967,
+ "learning_rate": 6.974863634072633e-06,
+ "loss": 0.0415,
+ "step": 22383
+ },
+ {
+ "epoch": 60.991825613079016,
+ "grad_norm": 2.0007641315460205,
+ "learning_rate": 6.974022500453723e-06,
+ "loss": 0.0207,
+ "step": 22384
+ },
+ {
+ "epoch": 60.99455040871935,
+ "grad_norm": 1.8376237154006958,
+ "learning_rate": 6.973181390401031e-06,
+ "loss": 0.0335,
+ "step": 22385
+ },
+ {
+ "epoch": 60.997275204359674,
+ "grad_norm": 2.3588898181915283,
+ "learning_rate": 6.972340303921105e-06,
+ "loss": 0.0641,
+ "step": 22386
+ },
+ {
+ "epoch": 61.0,
+ "grad_norm": 1.5751723051071167,
+ "learning_rate": 6.971499241020495e-06,
+ "loss": 0.0172,
+ "step": 22387
+ },
+ {
+ "epoch": 61.002724795640326,
+ "grad_norm": 1.9629234075546265,
+ "learning_rate": 6.97065820170575e-06,
+ "loss": 0.0194,
+ "step": 22388
+ },
+ {
+ "epoch": 61.00544959128065,
+ "grad_norm": 1.9875829219818115,
+ "learning_rate": 6.969817185983423e-06,
+ "loss": 0.0802,
+ "step": 22389
+ },
+ {
+ "epoch": 61.00817438692098,
+ "grad_norm": 2.0828309059143066,
+ "learning_rate": 6.968976193860061e-06,
+ "loss": 0.0362,
+ "step": 22390
+ },
+ {
+ "epoch": 61.01089918256131,
+ "grad_norm": 2.966578245162964,
+ "learning_rate": 6.968135225342216e-06,
+ "loss": 0.0325,
+ "step": 22391
+ },
+ {
+ "epoch": 61.013623978201636,
+ "grad_norm": 1.2247406244277954,
+ "learning_rate": 6.967294280436435e-06,
+ "loss": 0.0193,
+ "step": 22392
+ },
+ {
+ "epoch": 61.01634877384196,
+ "grad_norm": 1.8722141981124878,
+ "learning_rate": 6.9664533591492676e-06,
+ "loss": 0.0472,
+ "step": 22393
+ },
+ {
+ "epoch": 61.01907356948229,
+ "grad_norm": 2.1688408851623535,
+ "learning_rate": 6.965612461487263e-06,
+ "loss": 0.0383,
+ "step": 22394
+ },
+ {
+ "epoch": 61.02179836512261,
+ "grad_norm": 1.3115323781967163,
+ "learning_rate": 6.9647715874569744e-06,
+ "loss": 0.0161,
+ "step": 22395
+ },
+ {
+ "epoch": 61.02452316076294,
+ "grad_norm": 1.7461615800857544,
+ "learning_rate": 6.963930737064946e-06,
+ "loss": 0.1201,
+ "step": 22396
+ },
+ {
+ "epoch": 61.02724795640327,
+ "grad_norm": 2.1791136264801025,
+ "learning_rate": 6.9630899103177275e-06,
+ "loss": 0.0159,
+ "step": 22397
+ },
+ {
+ "epoch": 61.0299727520436,
+ "grad_norm": 2.0558433532714844,
+ "learning_rate": 6.962249107221865e-06,
+ "loss": 0.0337,
+ "step": 22398
+ },
+ {
+ "epoch": 61.032697547683924,
+ "grad_norm": 1.7500873804092407,
+ "learning_rate": 6.961408327783911e-06,
+ "loss": 0.0414,
+ "step": 22399
+ },
+ {
+ "epoch": 61.03542234332425,
+ "grad_norm": 1.8597514629364014,
+ "learning_rate": 6.9605675720104105e-06,
+ "loss": 0.1048,
+ "step": 22400
+ },
+ {
+ "epoch": 61.038147138964575,
+ "grad_norm": 1.825549602508545,
+ "learning_rate": 6.959726839907914e-06,
+ "loss": 0.0313,
+ "step": 22401
+ },
+ {
+ "epoch": 61.0408719346049,
+ "grad_norm": 3.201993465423584,
+ "learning_rate": 6.958886131482964e-06,
+ "loss": 0.0261,
+ "step": 22402
+ },
+ {
+ "epoch": 61.043596730245234,
+ "grad_norm": 1.9795671701431274,
+ "learning_rate": 6.958045446742112e-06,
+ "loss": 0.0301,
+ "step": 22403
+ },
+ {
+ "epoch": 61.04632152588556,
+ "grad_norm": 1.9091261625289917,
+ "learning_rate": 6.957204785691905e-06,
+ "loss": 0.0655,
+ "step": 22404
+ },
+ {
+ "epoch": 61.049046321525886,
+ "grad_norm": 2.2369942665100098,
+ "learning_rate": 6.956364148338888e-06,
+ "loss": 0.0394,
+ "step": 22405
+ },
+ {
+ "epoch": 61.05177111716621,
+ "grad_norm": 4.187967300415039,
+ "learning_rate": 6.955523534689609e-06,
+ "loss": 0.0315,
+ "step": 22406
+ },
+ {
+ "epoch": 61.05449591280654,
+ "grad_norm": 1.796702265739441,
+ "learning_rate": 6.954682944750616e-06,
+ "loss": 0.0237,
+ "step": 22407
+ },
+ {
+ "epoch": 61.05722070844686,
+ "grad_norm": 2.4351890087127686,
+ "learning_rate": 6.9538423785284524e-06,
+ "loss": 0.0749,
+ "step": 22408
+ },
+ {
+ "epoch": 61.059945504087196,
+ "grad_norm": 1.936820149421692,
+ "learning_rate": 6.953001836029668e-06,
+ "loss": 0.0473,
+ "step": 22409
+ },
+ {
+ "epoch": 61.06267029972752,
+ "grad_norm": 6.01869535446167,
+ "learning_rate": 6.952161317260806e-06,
+ "loss": 0.0386,
+ "step": 22410
+ },
+ {
+ "epoch": 61.06539509536785,
+ "grad_norm": 3.667424440383911,
+ "learning_rate": 6.951320822228411e-06,
+ "loss": 0.0368,
+ "step": 22411
+ },
+ {
+ "epoch": 61.06811989100817,
+ "grad_norm": 1.606785774230957,
+ "learning_rate": 6.9504803509390324e-06,
+ "loss": 0.0226,
+ "step": 22412
+ },
+ {
+ "epoch": 61.0708446866485,
+ "grad_norm": 1.4506185054779053,
+ "learning_rate": 6.949639903399219e-06,
+ "loss": 0.0326,
+ "step": 22413
+ },
+ {
+ "epoch": 61.073569482288825,
+ "grad_norm": 1.9527279138565063,
+ "learning_rate": 6.948799479615508e-06,
+ "loss": 0.0988,
+ "step": 22414
+ },
+ {
+ "epoch": 61.07629427792916,
+ "grad_norm": 1.5655763149261475,
+ "learning_rate": 6.947959079594451e-06,
+ "loss": 0.0167,
+ "step": 22415
+ },
+ {
+ "epoch": 61.079019073569484,
+ "grad_norm": 1.175885796546936,
+ "learning_rate": 6.947118703342588e-06,
+ "loss": 0.0311,
+ "step": 22416
+ },
+ {
+ "epoch": 61.08174386920981,
+ "grad_norm": 3.262662649154663,
+ "learning_rate": 6.9462783508664685e-06,
+ "loss": 0.0145,
+ "step": 22417
+ },
+ {
+ "epoch": 61.084468664850135,
+ "grad_norm": 3.497673749923706,
+ "learning_rate": 6.9454380221726335e-06,
+ "loss": 0.1514,
+ "step": 22418
+ },
+ {
+ "epoch": 61.08719346049046,
+ "grad_norm": 2.300168752670288,
+ "learning_rate": 6.9445977172676315e-06,
+ "loss": 0.1061,
+ "step": 22419
+ },
+ {
+ "epoch": 61.08991825613079,
+ "grad_norm": 1.418641448020935,
+ "learning_rate": 6.943757436158001e-06,
+ "loss": 0.197,
+ "step": 22420
+ },
+ {
+ "epoch": 61.09264305177112,
+ "grad_norm": 2.3047773838043213,
+ "learning_rate": 6.94291717885029e-06,
+ "loss": 0.0161,
+ "step": 22421
+ },
+ {
+ "epoch": 61.095367847411445,
+ "grad_norm": 1.392716884613037,
+ "learning_rate": 6.9420769453510415e-06,
+ "loss": 0.0169,
+ "step": 22422
+ },
+ {
+ "epoch": 61.09809264305177,
+ "grad_norm": 1.5533162355422974,
+ "learning_rate": 6.941236735666801e-06,
+ "loss": 0.0364,
+ "step": 22423
+ },
+ {
+ "epoch": 61.1008174386921,
+ "grad_norm": 2.1090519428253174,
+ "learning_rate": 6.9403965498041084e-06,
+ "loss": 0.0893,
+ "step": 22424
+ },
+ {
+ "epoch": 61.10354223433242,
+ "grad_norm": 2.433978796005249,
+ "learning_rate": 6.939556387769511e-06,
+ "loss": 0.2526,
+ "step": 22425
+ },
+ {
+ "epoch": 61.10626702997275,
+ "grad_norm": 2.3940269947052,
+ "learning_rate": 6.9387162495695485e-06,
+ "loss": 0.1639,
+ "step": 22426
+ },
+ {
+ "epoch": 61.10899182561308,
+ "grad_norm": 2.0278496742248535,
+ "learning_rate": 6.937876135210766e-06,
+ "loss": 0.0778,
+ "step": 22427
+ },
+ {
+ "epoch": 61.11171662125341,
+ "grad_norm": 2.20428466796875,
+ "learning_rate": 6.937036044699701e-06,
+ "loss": 0.0263,
+ "step": 22428
+ },
+ {
+ "epoch": 61.11444141689373,
+ "grad_norm": 2.4413647651672363,
+ "learning_rate": 6.936195978042904e-06,
+ "loss": 0.0957,
+ "step": 22429
+ },
+ {
+ "epoch": 61.11716621253406,
+ "grad_norm": 2.2020087242126465,
+ "learning_rate": 6.9353559352469165e-06,
+ "loss": 0.1039,
+ "step": 22430
+ },
+ {
+ "epoch": 61.119891008174385,
+ "grad_norm": 1.8824230432510376,
+ "learning_rate": 6.934515916318276e-06,
+ "loss": 0.023,
+ "step": 22431
+ },
+ {
+ "epoch": 61.12261580381471,
+ "grad_norm": 2.299020528793335,
+ "learning_rate": 6.9336759212635285e-06,
+ "loss": 0.0191,
+ "step": 22432
+ },
+ {
+ "epoch": 61.12534059945504,
+ "grad_norm": 2.261070966720581,
+ "learning_rate": 6.932835950089211e-06,
+ "loss": 0.0489,
+ "step": 22433
+ },
+ {
+ "epoch": 61.12806539509537,
+ "grad_norm": 1.8754559755325317,
+ "learning_rate": 6.93199600280187e-06,
+ "loss": 0.0263,
+ "step": 22434
+ },
+ {
+ "epoch": 61.130790190735695,
+ "grad_norm": 2.0133204460144043,
+ "learning_rate": 6.931156079408048e-06,
+ "loss": 0.022,
+ "step": 22435
+ },
+ {
+ "epoch": 61.13351498637602,
+ "grad_norm": 2.0767574310302734,
+ "learning_rate": 6.930316179914281e-06,
+ "loss": 0.1192,
+ "step": 22436
+ },
+ {
+ "epoch": 61.13623978201635,
+ "grad_norm": 2.2190468311309814,
+ "learning_rate": 6.9294763043271156e-06,
+ "loss": 0.0494,
+ "step": 22437
+ },
+ {
+ "epoch": 61.13896457765667,
+ "grad_norm": 2.2056140899658203,
+ "learning_rate": 6.928636452653086e-06,
+ "loss": 0.154,
+ "step": 22438
+ },
+ {
+ "epoch": 61.141689373297005,
+ "grad_norm": 4.062867164611816,
+ "learning_rate": 6.9277966248987405e-06,
+ "loss": 0.1003,
+ "step": 22439
+ },
+ {
+ "epoch": 61.14441416893733,
+ "grad_norm": 2.2924392223358154,
+ "learning_rate": 6.926956821070614e-06,
+ "loss": 0.0371,
+ "step": 22440
+ },
+ {
+ "epoch": 61.14713896457766,
+ "grad_norm": 2.1697542667388916,
+ "learning_rate": 6.92611704117525e-06,
+ "loss": 0.0257,
+ "step": 22441
+ },
+ {
+ "epoch": 61.14986376021798,
+ "grad_norm": 3.0479092597961426,
+ "learning_rate": 6.925277285219185e-06,
+ "loss": 0.0318,
+ "step": 22442
+ },
+ {
+ "epoch": 61.15258855585831,
+ "grad_norm": 2.3117456436157227,
+ "learning_rate": 6.924437553208965e-06,
+ "loss": 0.0569,
+ "step": 22443
+ },
+ {
+ "epoch": 61.155313351498634,
+ "grad_norm": 1.7632365226745605,
+ "learning_rate": 6.923597845151124e-06,
+ "loss": 0.0327,
+ "step": 22444
+ },
+ {
+ "epoch": 61.15803814713897,
+ "grad_norm": 1.5353211164474487,
+ "learning_rate": 6.922758161052201e-06,
+ "loss": 0.0309,
+ "step": 22445
+ },
+ {
+ "epoch": 61.16076294277929,
+ "grad_norm": 1.9547009468078613,
+ "learning_rate": 6.921918500918743e-06,
+ "loss": 0.0498,
+ "step": 22446
+ },
+ {
+ "epoch": 61.16348773841962,
+ "grad_norm": 1.686248779296875,
+ "learning_rate": 6.921078864757282e-06,
+ "loss": 0.1006,
+ "step": 22447
+ },
+ {
+ "epoch": 61.166212534059945,
+ "grad_norm": 1.7088121175765991,
+ "learning_rate": 6.920239252574364e-06,
+ "loss": 0.0181,
+ "step": 22448
+ },
+ {
+ "epoch": 61.16893732970027,
+ "grad_norm": 1.4767382144927979,
+ "learning_rate": 6.9193996643765195e-06,
+ "loss": 0.0623,
+ "step": 22449
+ },
+ {
+ "epoch": 61.171662125340596,
+ "grad_norm": 1.599284291267395,
+ "learning_rate": 6.918560100170293e-06,
+ "loss": 0.0336,
+ "step": 22450
+ },
+ {
+ "epoch": 61.17438692098093,
+ "grad_norm": 2.079446315765381,
+ "learning_rate": 6.91772055996222e-06,
+ "loss": 0.0155,
+ "step": 22451
+ },
+ {
+ "epoch": 61.177111716621255,
+ "grad_norm": 2.63967227935791,
+ "learning_rate": 6.916881043758843e-06,
+ "loss": 0.1212,
+ "step": 22452
+ },
+ {
+ "epoch": 61.17983651226158,
+ "grad_norm": 2.1463043689727783,
+ "learning_rate": 6.916041551566695e-06,
+ "loss": 0.0277,
+ "step": 22453
+ },
+ {
+ "epoch": 61.182561307901906,
+ "grad_norm": 2.5048153400421143,
+ "learning_rate": 6.915202083392316e-06,
+ "loss": 0.0283,
+ "step": 22454
+ },
+ {
+ "epoch": 61.18528610354223,
+ "grad_norm": 1.8808609247207642,
+ "learning_rate": 6.9143626392422446e-06,
+ "loss": 0.0474,
+ "step": 22455
+ },
+ {
+ "epoch": 61.18801089918256,
+ "grad_norm": 3.160059928894043,
+ "learning_rate": 6.9135232191230176e-06,
+ "loss": 0.0255,
+ "step": 22456
+ },
+ {
+ "epoch": 61.19073569482289,
+ "grad_norm": 3.140342950820923,
+ "learning_rate": 6.912683823041171e-06,
+ "loss": 0.0732,
+ "step": 22457
+ },
+ {
+ "epoch": 61.19346049046322,
+ "grad_norm": 2.7100565433502197,
+ "learning_rate": 6.911844451003244e-06,
+ "loss": 0.0598,
+ "step": 22458
+ },
+ {
+ "epoch": 61.19618528610354,
+ "grad_norm": 1.9119696617126465,
+ "learning_rate": 6.911005103015775e-06,
+ "loss": 0.1513,
+ "step": 22459
+ },
+ {
+ "epoch": 61.19891008174387,
+ "grad_norm": 3.389190673828125,
+ "learning_rate": 6.910165779085296e-06,
+ "loss": 0.0805,
+ "step": 22460
+ },
+ {
+ "epoch": 61.201634877384194,
+ "grad_norm": 1.793669581413269,
+ "learning_rate": 6.90932647921835e-06,
+ "loss": 0.0357,
+ "step": 22461
+ },
+ {
+ "epoch": 61.20435967302452,
+ "grad_norm": 1.3936567306518555,
+ "learning_rate": 6.908487203421465e-06,
+ "loss": 0.0369,
+ "step": 22462
+ },
+ {
+ "epoch": 61.20708446866485,
+ "grad_norm": 1.5322694778442383,
+ "learning_rate": 6.907647951701183e-06,
+ "loss": 0.0137,
+ "step": 22463
+ },
+ {
+ "epoch": 61.20980926430518,
+ "grad_norm": 1.7670660018920898,
+ "learning_rate": 6.906808724064043e-06,
+ "loss": 0.0225,
+ "step": 22464
+ },
+ {
+ "epoch": 61.212534059945504,
+ "grad_norm": 1.5903620719909668,
+ "learning_rate": 6.905969520516573e-06,
+ "loss": 0.0326,
+ "step": 22465
+ },
+ {
+ "epoch": 61.21525885558583,
+ "grad_norm": 1.6857876777648926,
+ "learning_rate": 6.905130341065317e-06,
+ "loss": 0.0276,
+ "step": 22466
+ },
+ {
+ "epoch": 61.217983651226156,
+ "grad_norm": 2.1751887798309326,
+ "learning_rate": 6.904291185716802e-06,
+ "loss": 0.0323,
+ "step": 22467
+ },
+ {
+ "epoch": 61.22070844686648,
+ "grad_norm": 1.6384081840515137,
+ "learning_rate": 6.903452054477573e-06,
+ "loss": 0.0259,
+ "step": 22468
+ },
+ {
+ "epoch": 61.223433242506815,
+ "grad_norm": 1.8028063774108887,
+ "learning_rate": 6.902612947354155e-06,
+ "loss": 0.0462,
+ "step": 22469
+ },
+ {
+ "epoch": 61.22615803814714,
+ "grad_norm": 1.5785092115402222,
+ "learning_rate": 6.9017738643530905e-06,
+ "loss": 0.0218,
+ "step": 22470
+ },
+ {
+ "epoch": 61.228882833787466,
+ "grad_norm": 2.2915735244750977,
+ "learning_rate": 6.90093480548091e-06,
+ "loss": 0.0368,
+ "step": 22471
+ },
+ {
+ "epoch": 61.23160762942779,
+ "grad_norm": 2.210981607437134,
+ "learning_rate": 6.9000957707441505e-06,
+ "loss": 0.0676,
+ "step": 22472
+ },
+ {
+ "epoch": 61.23433242506812,
+ "grad_norm": 1.872844934463501,
+ "learning_rate": 6.899256760149345e-06,
+ "loss": 0.0262,
+ "step": 22473
+ },
+ {
+ "epoch": 61.237057220708444,
+ "grad_norm": 1.6500301361083984,
+ "learning_rate": 6.898417773703028e-06,
+ "loss": 0.0237,
+ "step": 22474
+ },
+ {
+ "epoch": 61.23978201634878,
+ "grad_norm": 1.419887661933899,
+ "learning_rate": 6.897578811411732e-06,
+ "loss": 0.0199,
+ "step": 22475
+ },
+ {
+ "epoch": 61.2425068119891,
+ "grad_norm": 2.33601975440979,
+ "learning_rate": 6.896739873281995e-06,
+ "loss": 0.0409,
+ "step": 22476
+ },
+ {
+ "epoch": 61.24523160762943,
+ "grad_norm": 2.5602681636810303,
+ "learning_rate": 6.8959009593203454e-06,
+ "loss": 0.171,
+ "step": 22477
+ },
+ {
+ "epoch": 61.247956403269754,
+ "grad_norm": 3.286353826522827,
+ "learning_rate": 6.895062069533322e-06,
+ "loss": 0.0657,
+ "step": 22478
+ },
+ {
+ "epoch": 61.25068119891008,
+ "grad_norm": 2.3696413040161133,
+ "learning_rate": 6.894223203927452e-06,
+ "loss": 0.0894,
+ "step": 22479
+ },
+ {
+ "epoch": 61.253405994550405,
+ "grad_norm": 1.621720314025879,
+ "learning_rate": 6.8933843625092715e-06,
+ "loss": 0.0267,
+ "step": 22480
+ },
+ {
+ "epoch": 61.25613079019074,
+ "grad_norm": 2.0779099464416504,
+ "learning_rate": 6.892545545285314e-06,
+ "loss": 0.0307,
+ "step": 22481
+ },
+ {
+ "epoch": 61.258855585831064,
+ "grad_norm": 3.2780275344848633,
+ "learning_rate": 6.891706752262115e-06,
+ "loss": 0.0812,
+ "step": 22482
+ },
+ {
+ "epoch": 61.26158038147139,
+ "grad_norm": 1.9065289497375488,
+ "learning_rate": 6.890867983446201e-06,
+ "loss": 0.0261,
+ "step": 22483
+ },
+ {
+ "epoch": 61.264305177111716,
+ "grad_norm": 2.561950445175171,
+ "learning_rate": 6.8900292388441095e-06,
+ "loss": 0.0416,
+ "step": 22484
+ },
+ {
+ "epoch": 61.26702997275204,
+ "grad_norm": 1.6244688034057617,
+ "learning_rate": 6.889190518462368e-06,
+ "loss": 0.022,
+ "step": 22485
+ },
+ {
+ "epoch": 61.26975476839237,
+ "grad_norm": 2.013483762741089,
+ "learning_rate": 6.888351822307513e-06,
+ "loss": 0.0298,
+ "step": 22486
+ },
+ {
+ "epoch": 61.2724795640327,
+ "grad_norm": 2.0162510871887207,
+ "learning_rate": 6.887513150386072e-06,
+ "loss": 0.0242,
+ "step": 22487
+ },
+ {
+ "epoch": 61.275204359673026,
+ "grad_norm": 2.37302827835083,
+ "learning_rate": 6.886674502704579e-06,
+ "loss": 0.0792,
+ "step": 22488
+ },
+ {
+ "epoch": 61.27792915531335,
+ "grad_norm": 1.6928058862686157,
+ "learning_rate": 6.885835879269564e-06,
+ "loss": 0.0256,
+ "step": 22489
+ },
+ {
+ "epoch": 61.28065395095368,
+ "grad_norm": 2.2793843746185303,
+ "learning_rate": 6.88499728008756e-06,
+ "loss": 0.0619,
+ "step": 22490
+ },
+ {
+ "epoch": 61.283378746594,
+ "grad_norm": 1.5228102207183838,
+ "learning_rate": 6.884158705165096e-06,
+ "loss": 0.0222,
+ "step": 22491
+ },
+ {
+ "epoch": 61.28610354223433,
+ "grad_norm": 2.028301477432251,
+ "learning_rate": 6.883320154508705e-06,
+ "loss": 0.0272,
+ "step": 22492
+ },
+ {
+ "epoch": 61.28882833787466,
+ "grad_norm": 1.4442472457885742,
+ "learning_rate": 6.882481628124915e-06,
+ "loss": 0.0263,
+ "step": 22493
+ },
+ {
+ "epoch": 61.29155313351499,
+ "grad_norm": 3.3833305835723877,
+ "learning_rate": 6.881643126020259e-06,
+ "loss": 0.1218,
+ "step": 22494
+ },
+ {
+ "epoch": 61.294277929155314,
+ "grad_norm": 1.2789602279663086,
+ "learning_rate": 6.8808046482012645e-06,
+ "loss": 0.0279,
+ "step": 22495
+ },
+ {
+ "epoch": 61.29700272479564,
+ "grad_norm": 2.2182860374450684,
+ "learning_rate": 6.8799661946744646e-06,
+ "loss": 0.0457,
+ "step": 22496
+ },
+ {
+ "epoch": 61.299727520435965,
+ "grad_norm": 1.5548244714736938,
+ "learning_rate": 6.879127765446383e-06,
+ "loss": 0.167,
+ "step": 22497
+ },
+ {
+ "epoch": 61.30245231607629,
+ "grad_norm": 1.0479422807693481,
+ "learning_rate": 6.878289360523556e-06,
+ "loss": 0.0117,
+ "step": 22498
+ },
+ {
+ "epoch": 61.305177111716624,
+ "grad_norm": 1.843062400817871,
+ "learning_rate": 6.877450979912512e-06,
+ "loss": 0.0213,
+ "step": 22499
+ },
+ {
+ "epoch": 61.30790190735695,
+ "grad_norm": 2.0217554569244385,
+ "learning_rate": 6.876612623619782e-06,
+ "loss": 0.0204,
+ "step": 22500
+ },
+ {
+ "epoch": 61.310626702997276,
+ "grad_norm": 2.3189985752105713,
+ "learning_rate": 6.875774291651889e-06,
+ "loss": 0.1171,
+ "step": 22501
+ },
+ {
+ "epoch": 61.3133514986376,
+ "grad_norm": 1.7337456941604614,
+ "learning_rate": 6.874935984015367e-06,
+ "loss": 0.039,
+ "step": 22502
+ },
+ {
+ "epoch": 61.31607629427793,
+ "grad_norm": 2.2422070503234863,
+ "learning_rate": 6.874097700716742e-06,
+ "loss": 0.046,
+ "step": 22503
+ },
+ {
+ "epoch": 61.31880108991825,
+ "grad_norm": 1.4085386991500854,
+ "learning_rate": 6.873259441762545e-06,
+ "loss": 0.0181,
+ "step": 22504
+ },
+ {
+ "epoch": 61.321525885558586,
+ "grad_norm": 1.8418525457382202,
+ "learning_rate": 6.872421207159301e-06,
+ "loss": 0.026,
+ "step": 22505
+ },
+ {
+ "epoch": 61.32425068119891,
+ "grad_norm": 2.653505563735962,
+ "learning_rate": 6.871582996913542e-06,
+ "loss": 0.0322,
+ "step": 22506
+ },
+ {
+ "epoch": 61.32697547683924,
+ "grad_norm": 1.1896374225616455,
+ "learning_rate": 6.870744811031791e-06,
+ "loss": 0.0155,
+ "step": 22507
+ },
+ {
+ "epoch": 61.32970027247956,
+ "grad_norm": 2.867475986480713,
+ "learning_rate": 6.869906649520581e-06,
+ "loss": 0.0453,
+ "step": 22508
+ },
+ {
+ "epoch": 61.33242506811989,
+ "grad_norm": 2.619243621826172,
+ "learning_rate": 6.869068512386436e-06,
+ "loss": 0.0338,
+ "step": 22509
+ },
+ {
+ "epoch": 61.335149863760215,
+ "grad_norm": 3.1160647869110107,
+ "learning_rate": 6.868230399635887e-06,
+ "loss": 0.169,
+ "step": 22510
+ },
+ {
+ "epoch": 61.33787465940055,
+ "grad_norm": 2.242542028427124,
+ "learning_rate": 6.867392311275456e-06,
+ "loss": 0.0956,
+ "step": 22511
+ },
+ {
+ "epoch": 61.34059945504087,
+ "grad_norm": 2.2135210037231445,
+ "learning_rate": 6.866554247311676e-06,
+ "loss": 0.0516,
+ "step": 22512
+ },
+ {
+ "epoch": 61.3433242506812,
+ "grad_norm": 2.3541057109832764,
+ "learning_rate": 6.865716207751068e-06,
+ "loss": 0.0277,
+ "step": 22513
+ },
+ {
+ "epoch": 61.346049046321525,
+ "grad_norm": 2.3229002952575684,
+ "learning_rate": 6.864878192600158e-06,
+ "loss": 0.1074,
+ "step": 22514
+ },
+ {
+ "epoch": 61.34877384196185,
+ "grad_norm": 2.2161314487457275,
+ "learning_rate": 6.864040201865481e-06,
+ "loss": 0.0259,
+ "step": 22515
+ },
+ {
+ "epoch": 61.35149863760218,
+ "grad_norm": 2.5575063228607178,
+ "learning_rate": 6.863202235553557e-06,
+ "loss": 0.0323,
+ "step": 22516
+ },
+ {
+ "epoch": 61.35422343324251,
+ "grad_norm": 25.836278915405273,
+ "learning_rate": 6.862364293670913e-06,
+ "loss": 0.0537,
+ "step": 22517
+ },
+ {
+ "epoch": 61.356948228882835,
+ "grad_norm": 2.1679136753082275,
+ "learning_rate": 6.8615263762240735e-06,
+ "loss": 0.05,
+ "step": 22518
+ },
+ {
+ "epoch": 61.35967302452316,
+ "grad_norm": 1.7739524841308594,
+ "learning_rate": 6.860688483219569e-06,
+ "loss": 0.0437,
+ "step": 22519
+ },
+ {
+ "epoch": 61.36239782016349,
+ "grad_norm": 1.989777684211731,
+ "learning_rate": 6.859850614663916e-06,
+ "loss": 0.0395,
+ "step": 22520
+ },
+ {
+ "epoch": 61.36512261580381,
+ "grad_norm": 1.6384317874908447,
+ "learning_rate": 6.859012770563649e-06,
+ "loss": 0.0188,
+ "step": 22521
+ },
+ {
+ "epoch": 61.36784741144414,
+ "grad_norm": 1.723526120185852,
+ "learning_rate": 6.858174950925289e-06,
+ "loss": 0.0279,
+ "step": 22522
+ },
+ {
+ "epoch": 61.37057220708447,
+ "grad_norm": 1.9844071865081787,
+ "learning_rate": 6.857337155755362e-06,
+ "loss": 0.0211,
+ "step": 22523
+ },
+ {
+ "epoch": 61.3732970027248,
+ "grad_norm": 1.641653060913086,
+ "learning_rate": 6.856499385060392e-06,
+ "loss": 0.1066,
+ "step": 22524
+ },
+ {
+ "epoch": 61.37602179836512,
+ "grad_norm": 1.6293281316757202,
+ "learning_rate": 6.855661638846902e-06,
+ "loss": 0.0258,
+ "step": 22525
+ },
+ {
+ "epoch": 61.37874659400545,
+ "grad_norm": 1.932193636894226,
+ "learning_rate": 6.854823917121421e-06,
+ "loss": 0.0271,
+ "step": 22526
+ },
+ {
+ "epoch": 61.381471389645775,
+ "grad_norm": 2.0864713191986084,
+ "learning_rate": 6.853986219890467e-06,
+ "loss": 0.03,
+ "step": 22527
+ },
+ {
+ "epoch": 61.3841961852861,
+ "grad_norm": 2.424278736114502,
+ "learning_rate": 6.853148547160568e-06,
+ "loss": 0.026,
+ "step": 22528
+ },
+ {
+ "epoch": 61.38692098092643,
+ "grad_norm": 2.6643435955047607,
+ "learning_rate": 6.852310898938248e-06,
+ "loss": 0.0535,
+ "step": 22529
+ },
+ {
+ "epoch": 61.38964577656676,
+ "grad_norm": 1.857762098312378,
+ "learning_rate": 6.85147327523003e-06,
+ "loss": 0.0648,
+ "step": 22530
+ },
+ {
+ "epoch": 61.392370572207085,
+ "grad_norm": 1.927832841873169,
+ "learning_rate": 6.850635676042432e-06,
+ "loss": 0.0463,
+ "step": 22531
+ },
+ {
+ "epoch": 61.39509536784741,
+ "grad_norm": 2.33500337600708,
+ "learning_rate": 6.849798101381984e-06,
+ "loss": 0.0178,
+ "step": 22532
+ },
+ {
+ "epoch": 61.39782016348774,
+ "grad_norm": 2.8801393508911133,
+ "learning_rate": 6.848960551255208e-06,
+ "loss": 0.0248,
+ "step": 22533
+ },
+ {
+ "epoch": 61.40054495912806,
+ "grad_norm": 2.0224344730377197,
+ "learning_rate": 6.848123025668626e-06,
+ "loss": 0.1697,
+ "step": 22534
+ },
+ {
+ "epoch": 61.403269754768395,
+ "grad_norm": 1.7856637239456177,
+ "learning_rate": 6.847285524628762e-06,
+ "loss": 0.0321,
+ "step": 22535
+ },
+ {
+ "epoch": 61.40599455040872,
+ "grad_norm": 1.5992523431777954,
+ "learning_rate": 6.846448048142134e-06,
+ "loss": 0.0305,
+ "step": 22536
+ },
+ {
+ "epoch": 61.40871934604905,
+ "grad_norm": 3.273334264755249,
+ "learning_rate": 6.84561059621527e-06,
+ "loss": 0.1354,
+ "step": 22537
+ },
+ {
+ "epoch": 61.41144414168937,
+ "grad_norm": 2.289907932281494,
+ "learning_rate": 6.844773168854686e-06,
+ "loss": 0.0633,
+ "step": 22538
+ },
+ {
+ "epoch": 61.4141689373297,
+ "grad_norm": 2.2640600204467773,
+ "learning_rate": 6.843935766066911e-06,
+ "loss": 0.0318,
+ "step": 22539
+ },
+ {
+ "epoch": 61.416893732970024,
+ "grad_norm": 1.2916557788848877,
+ "learning_rate": 6.8430983878584585e-06,
+ "loss": 0.0246,
+ "step": 22540
+ },
+ {
+ "epoch": 61.41961852861036,
+ "grad_norm": 2.1303021907806396,
+ "learning_rate": 6.842261034235858e-06,
+ "loss": 0.0521,
+ "step": 22541
+ },
+ {
+ "epoch": 61.42234332425068,
+ "grad_norm": 2.197815179824829,
+ "learning_rate": 6.841423705205623e-06,
+ "loss": 0.0569,
+ "step": 22542
+ },
+ {
+ "epoch": 61.42506811989101,
+ "grad_norm": 2.5137650966644287,
+ "learning_rate": 6.840586400774282e-06,
+ "loss": 0.1322,
+ "step": 22543
+ },
+ {
+ "epoch": 61.427792915531334,
+ "grad_norm": 2.0341622829437256,
+ "learning_rate": 6.839749120948349e-06,
+ "loss": 0.0487,
+ "step": 22544
+ },
+ {
+ "epoch": 61.43051771117166,
+ "grad_norm": 1.5367568731307983,
+ "learning_rate": 6.8389118657343505e-06,
+ "loss": 0.0421,
+ "step": 22545
+ },
+ {
+ "epoch": 61.433242506811986,
+ "grad_norm": 2.1281986236572266,
+ "learning_rate": 6.838074635138805e-06,
+ "loss": 0.0359,
+ "step": 22546
+ },
+ {
+ "epoch": 61.43596730245232,
+ "grad_norm": 61.03743362426758,
+ "learning_rate": 6.8372374291682306e-06,
+ "loss": 0.1523,
+ "step": 22547
+ },
+ {
+ "epoch": 61.438692098092645,
+ "grad_norm": 1.73598051071167,
+ "learning_rate": 6.836400247829151e-06,
+ "loss": 0.0325,
+ "step": 22548
+ },
+ {
+ "epoch": 61.44141689373297,
+ "grad_norm": 1.9652063846588135,
+ "learning_rate": 6.835563091128081e-06,
+ "loss": 0.0485,
+ "step": 22549
+ },
+ {
+ "epoch": 61.444141689373296,
+ "grad_norm": 2.1327598094940186,
+ "learning_rate": 6.834725959071546e-06,
+ "loss": 0.0879,
+ "step": 22550
+ },
+ {
+ "epoch": 61.44686648501362,
+ "grad_norm": 3.0861313343048096,
+ "learning_rate": 6.833888851666064e-06,
+ "loss": 0.0362,
+ "step": 22551
+ },
+ {
+ "epoch": 61.44959128065395,
+ "grad_norm": 2.3059756755828857,
+ "learning_rate": 6.8330517689181534e-06,
+ "loss": 0.1065,
+ "step": 22552
+ },
+ {
+ "epoch": 61.45231607629428,
+ "grad_norm": 1.9438672065734863,
+ "learning_rate": 6.832214710834335e-06,
+ "loss": 0.0344,
+ "step": 22553
+ },
+ {
+ "epoch": 61.45504087193461,
+ "grad_norm": 1.5301624536514282,
+ "learning_rate": 6.831377677421125e-06,
+ "loss": 0.0815,
+ "step": 22554
+ },
+ {
+ "epoch": 61.45776566757493,
+ "grad_norm": 1.9831254482269287,
+ "learning_rate": 6.830540668685045e-06,
+ "loss": 0.056,
+ "step": 22555
+ },
+ {
+ "epoch": 61.46049046321526,
+ "grad_norm": 2.0276546478271484,
+ "learning_rate": 6.82970368463261e-06,
+ "loss": 0.075,
+ "step": 22556
+ },
+ {
+ "epoch": 61.463215258855584,
+ "grad_norm": 1.999906301498413,
+ "learning_rate": 6.8288667252703435e-06,
+ "loss": 0.0231,
+ "step": 22557
+ },
+ {
+ "epoch": 61.46594005449591,
+ "grad_norm": 2.99190616607666,
+ "learning_rate": 6.828029790604757e-06,
+ "loss": 0.1179,
+ "step": 22558
+ },
+ {
+ "epoch": 61.46866485013624,
+ "grad_norm": 1.0159262418746948,
+ "learning_rate": 6.827192880642377e-06,
+ "loss": 0.0132,
+ "step": 22559
+ },
+ {
+ "epoch": 61.47138964577657,
+ "grad_norm": 2.061126708984375,
+ "learning_rate": 6.826355995389714e-06,
+ "loss": 0.0835,
+ "step": 22560
+ },
+ {
+ "epoch": 61.474114441416894,
+ "grad_norm": 1.670363187789917,
+ "learning_rate": 6.82551913485329e-06,
+ "loss": 0.0493,
+ "step": 22561
+ },
+ {
+ "epoch": 61.47683923705722,
+ "grad_norm": 1.4942911863327026,
+ "learning_rate": 6.824682299039619e-06,
+ "loss": 0.0208,
+ "step": 22562
+ },
+ {
+ "epoch": 61.479564032697546,
+ "grad_norm": 1.872321605682373,
+ "learning_rate": 6.823845487955222e-06,
+ "loss": 0.0859,
+ "step": 22563
+ },
+ {
+ "epoch": 61.48228882833787,
+ "grad_norm": 2.8062148094177246,
+ "learning_rate": 6.823008701606612e-06,
+ "loss": 0.046,
+ "step": 22564
+ },
+ {
+ "epoch": 61.485013623978205,
+ "grad_norm": 2.6070775985717773,
+ "learning_rate": 6.822171940000311e-06,
+ "loss": 0.022,
+ "step": 22565
+ },
+ {
+ "epoch": 61.48773841961853,
+ "grad_norm": 2.1246089935302734,
+ "learning_rate": 6.821335203142825e-06,
+ "loss": 0.0342,
+ "step": 22566
+ },
+ {
+ "epoch": 61.490463215258856,
+ "grad_norm": 5.9173502922058105,
+ "learning_rate": 6.820498491040684e-06,
+ "loss": 0.0489,
+ "step": 22567
+ },
+ {
+ "epoch": 61.49318801089918,
+ "grad_norm": 19.502450942993164,
+ "learning_rate": 6.819661803700397e-06,
+ "loss": 0.0188,
+ "step": 22568
+ },
+ {
+ "epoch": 61.49591280653951,
+ "grad_norm": 2.5197012424468994,
+ "learning_rate": 6.818825141128484e-06,
+ "loss": 0.0502,
+ "step": 22569
+ },
+ {
+ "epoch": 61.49863760217983,
+ "grad_norm": 3.028984308242798,
+ "learning_rate": 6.817988503331456e-06,
+ "loss": 0.1984,
+ "step": 22570
+ },
+ {
+ "epoch": 61.50136239782017,
+ "grad_norm": 2.9086947441101074,
+ "learning_rate": 6.817151890315834e-06,
+ "loss": 0.1775,
+ "step": 22571
+ },
+ {
+ "epoch": 61.50408719346049,
+ "grad_norm": 1.8135905265808105,
+ "learning_rate": 6.816315302088128e-06,
+ "loss": 0.026,
+ "step": 22572
+ },
+ {
+ "epoch": 61.50681198910082,
+ "grad_norm": 1.981508493423462,
+ "learning_rate": 6.815478738654857e-06,
+ "loss": 0.0517,
+ "step": 22573
+ },
+ {
+ "epoch": 61.509536784741144,
+ "grad_norm": 2.190747022628784,
+ "learning_rate": 6.814642200022534e-06,
+ "loss": 0.1343,
+ "step": 22574
+ },
+ {
+ "epoch": 61.51226158038147,
+ "grad_norm": 1.9964202642440796,
+ "learning_rate": 6.813805686197677e-06,
+ "loss": 0.0554,
+ "step": 22575
+ },
+ {
+ "epoch": 61.514986376021795,
+ "grad_norm": 2.415062427520752,
+ "learning_rate": 6.812969197186796e-06,
+ "loss": 0.0241,
+ "step": 22576
+ },
+ {
+ "epoch": 61.51771117166213,
+ "grad_norm": 2.362863063812256,
+ "learning_rate": 6.8121327329964116e-06,
+ "loss": 0.04,
+ "step": 22577
+ },
+ {
+ "epoch": 61.520435967302454,
+ "grad_norm": 2.1957623958587646,
+ "learning_rate": 6.811296293633032e-06,
+ "loss": 0.056,
+ "step": 22578
+ },
+ {
+ "epoch": 61.52316076294278,
+ "grad_norm": 1.5270532369613647,
+ "learning_rate": 6.810459879103177e-06,
+ "loss": 0.0179,
+ "step": 22579
+ },
+ {
+ "epoch": 61.525885558583106,
+ "grad_norm": 1.3593844175338745,
+ "learning_rate": 6.8096234894133554e-06,
+ "loss": 0.0147,
+ "step": 22580
+ },
+ {
+ "epoch": 61.52861035422343,
+ "grad_norm": 1.803031086921692,
+ "learning_rate": 6.808787124570086e-06,
+ "loss": 0.0945,
+ "step": 22581
+ },
+ {
+ "epoch": 61.53133514986376,
+ "grad_norm": 1.7195485830307007,
+ "learning_rate": 6.8079507845798754e-06,
+ "loss": 0.0602,
+ "step": 22582
+ },
+ {
+ "epoch": 61.53405994550409,
+ "grad_norm": 1.9834874868392944,
+ "learning_rate": 6.807114469449242e-06,
+ "loss": 0.0584,
+ "step": 22583
+ },
+ {
+ "epoch": 61.536784741144416,
+ "grad_norm": 1.649855375289917,
+ "learning_rate": 6.806278179184701e-06,
+ "loss": 0.0238,
+ "step": 22584
+ },
+ {
+ "epoch": 61.53950953678474,
+ "grad_norm": 3.3921430110931396,
+ "learning_rate": 6.805441913792761e-06,
+ "loss": 0.0415,
+ "step": 22585
+ },
+ {
+ "epoch": 61.54223433242507,
+ "grad_norm": 1.9280860424041748,
+ "learning_rate": 6.804605673279937e-06,
+ "loss": 0.0396,
+ "step": 22586
+ },
+ {
+ "epoch": 61.54495912806539,
+ "grad_norm": 2.024338483810425,
+ "learning_rate": 6.803769457652743e-06,
+ "loss": 0.0197,
+ "step": 22587
+ },
+ {
+ "epoch": 61.54768392370572,
+ "grad_norm": 2.637011766433716,
+ "learning_rate": 6.802933266917687e-06,
+ "loss": 0.0734,
+ "step": 22588
+ },
+ {
+ "epoch": 61.55040871934605,
+ "grad_norm": 2.242365598678589,
+ "learning_rate": 6.802097101081286e-06,
+ "loss": 0.0384,
+ "step": 22589
+ },
+ {
+ "epoch": 61.55313351498638,
+ "grad_norm": 2.0500407218933105,
+ "learning_rate": 6.801260960150049e-06,
+ "loss": 0.0266,
+ "step": 22590
+ },
+ {
+ "epoch": 61.555858310626704,
+ "grad_norm": 1.5199658870697021,
+ "learning_rate": 6.80042484413049e-06,
+ "loss": 0.0773,
+ "step": 22591
+ },
+ {
+ "epoch": 61.55858310626703,
+ "grad_norm": 1.6200264692306519,
+ "learning_rate": 6.799588753029116e-06,
+ "loss": 0.0199,
+ "step": 22592
+ },
+ {
+ "epoch": 61.561307901907355,
+ "grad_norm": 2.5865156650543213,
+ "learning_rate": 6.798752686852445e-06,
+ "loss": 0.0745,
+ "step": 22593
+ },
+ {
+ "epoch": 61.56403269754768,
+ "grad_norm": 2.460791826248169,
+ "learning_rate": 6.797916645606984e-06,
+ "loss": 0.0637,
+ "step": 22594
+ },
+ {
+ "epoch": 61.566757493188014,
+ "grad_norm": 3.261644124984741,
+ "learning_rate": 6.797080629299246e-06,
+ "loss": 0.0277,
+ "step": 22595
+ },
+ {
+ "epoch": 61.56948228882834,
+ "grad_norm": 2.667501449584961,
+ "learning_rate": 6.796244637935738e-06,
+ "loss": 0.1144,
+ "step": 22596
+ },
+ {
+ "epoch": 61.572207084468666,
+ "grad_norm": 2.4013819694519043,
+ "learning_rate": 6.7954086715229785e-06,
+ "loss": 0.0491,
+ "step": 22597
+ },
+ {
+ "epoch": 61.57493188010899,
+ "grad_norm": 1.7992186546325684,
+ "learning_rate": 6.794572730067468e-06,
+ "loss": 0.03,
+ "step": 22598
+ },
+ {
+ "epoch": 61.57765667574932,
+ "grad_norm": 2.4108965396881104,
+ "learning_rate": 6.793736813575726e-06,
+ "loss": 0.0426,
+ "step": 22599
+ },
+ {
+ "epoch": 61.58038147138964,
+ "grad_norm": 1.7794959545135498,
+ "learning_rate": 6.792900922054253e-06,
+ "loss": 0.0223,
+ "step": 22600
+ },
+ {
+ "epoch": 61.583106267029976,
+ "grad_norm": 2.0885260105133057,
+ "learning_rate": 6.792065055509567e-06,
+ "loss": 0.0395,
+ "step": 22601
+ },
+ {
+ "epoch": 61.5858310626703,
+ "grad_norm": 2.322711229324341,
+ "learning_rate": 6.791229213948178e-06,
+ "loss": 0.0293,
+ "step": 22602
+ },
+ {
+ "epoch": 61.58855585831063,
+ "grad_norm": 2.036811590194702,
+ "learning_rate": 6.79039339737659e-06,
+ "loss": 0.0293,
+ "step": 22603
+ },
+ {
+ "epoch": 61.59128065395095,
+ "grad_norm": 2.5450267791748047,
+ "learning_rate": 6.7895576058013165e-06,
+ "loss": 0.0554,
+ "step": 22604
+ },
+ {
+ "epoch": 61.59400544959128,
+ "grad_norm": 2.6264185905456543,
+ "learning_rate": 6.788721839228863e-06,
+ "loss": 0.0347,
+ "step": 22605
+ },
+ {
+ "epoch": 61.596730245231605,
+ "grad_norm": 1.9108670949935913,
+ "learning_rate": 6.787886097665744e-06,
+ "loss": 0.0207,
+ "step": 22606
+ },
+ {
+ "epoch": 61.59945504087194,
+ "grad_norm": 2.9795830249786377,
+ "learning_rate": 6.787050381118461e-06,
+ "loss": 0.0356,
+ "step": 22607
+ },
+ {
+ "epoch": 61.60217983651226,
+ "grad_norm": 1.9623130559921265,
+ "learning_rate": 6.786214689593527e-06,
+ "loss": 0.1026,
+ "step": 22608
+ },
+ {
+ "epoch": 61.60490463215259,
+ "grad_norm": 1.9707322120666504,
+ "learning_rate": 6.785379023097451e-06,
+ "loss": 0.0959,
+ "step": 22609
+ },
+ {
+ "epoch": 61.607629427792915,
+ "grad_norm": 1.9082854986190796,
+ "learning_rate": 6.784543381636738e-06,
+ "loss": 0.0251,
+ "step": 22610
+ },
+ {
+ "epoch": 61.61035422343324,
+ "grad_norm": 1.7421194314956665,
+ "learning_rate": 6.783707765217899e-06,
+ "loss": 0.0237,
+ "step": 22611
+ },
+ {
+ "epoch": 61.61307901907357,
+ "grad_norm": 2.0385637283325195,
+ "learning_rate": 6.782872173847439e-06,
+ "loss": 0.0468,
+ "step": 22612
+ },
+ {
+ "epoch": 61.6158038147139,
+ "grad_norm": 1.9061470031738281,
+ "learning_rate": 6.78203660753187e-06,
+ "loss": 0.023,
+ "step": 22613
+ },
+ {
+ "epoch": 61.618528610354225,
+ "grad_norm": 2.430309295654297,
+ "learning_rate": 6.781201066277693e-06,
+ "loss": 0.0494,
+ "step": 22614
+ },
+ {
+ "epoch": 61.62125340599455,
+ "grad_norm": 2.330453872680664,
+ "learning_rate": 6.780365550091421e-06,
+ "loss": 0.1372,
+ "step": 22615
+ },
+ {
+ "epoch": 61.62397820163488,
+ "grad_norm": 1.964902400970459,
+ "learning_rate": 6.779530058979555e-06,
+ "loss": 0.0402,
+ "step": 22616
+ },
+ {
+ "epoch": 61.6267029972752,
+ "grad_norm": 1.5970609188079834,
+ "learning_rate": 6.778694592948605e-06,
+ "loss": 0.0177,
+ "step": 22617
+ },
+ {
+ "epoch": 61.62942779291553,
+ "grad_norm": 1.8127045631408691,
+ "learning_rate": 6.777859152005082e-06,
+ "loss": 0.0401,
+ "step": 22618
+ },
+ {
+ "epoch": 61.63215258855586,
+ "grad_norm": 1.9279998540878296,
+ "learning_rate": 6.777023736155486e-06,
+ "loss": 0.1222,
+ "step": 22619
+ },
+ {
+ "epoch": 61.63487738419619,
+ "grad_norm": 1.200833797454834,
+ "learning_rate": 6.776188345406327e-06,
+ "loss": 0.0211,
+ "step": 22620
+ },
+ {
+ "epoch": 61.63760217983651,
+ "grad_norm": 1.9975347518920898,
+ "learning_rate": 6.775352979764107e-06,
+ "loss": 0.029,
+ "step": 22621
+ },
+ {
+ "epoch": 61.64032697547684,
+ "grad_norm": 1.582732081413269,
+ "learning_rate": 6.7745176392353365e-06,
+ "loss": 0.0221,
+ "step": 22622
+ },
+ {
+ "epoch": 61.643051771117165,
+ "grad_norm": 3.138223648071289,
+ "learning_rate": 6.773682323826517e-06,
+ "loss": 0.0843,
+ "step": 22623
+ },
+ {
+ "epoch": 61.64577656675749,
+ "grad_norm": 2.043445348739624,
+ "learning_rate": 6.772847033544157e-06,
+ "loss": 0.0228,
+ "step": 22624
+ },
+ {
+ "epoch": 61.64850136239782,
+ "grad_norm": 1.901571273803711,
+ "learning_rate": 6.77201176839476e-06,
+ "loss": 0.0427,
+ "step": 22625
+ },
+ {
+ "epoch": 61.65122615803815,
+ "grad_norm": 32.550907135009766,
+ "learning_rate": 6.771176528384833e-06,
+ "loss": 0.0258,
+ "step": 22626
+ },
+ {
+ "epoch": 61.653950953678475,
+ "grad_norm": 2.9802072048187256,
+ "learning_rate": 6.770341313520877e-06,
+ "loss": 0.0715,
+ "step": 22627
+ },
+ {
+ "epoch": 61.6566757493188,
+ "grad_norm": 1.5331844091415405,
+ "learning_rate": 6.7695061238094e-06,
+ "loss": 0.0137,
+ "step": 22628
+ },
+ {
+ "epoch": 61.65940054495913,
+ "grad_norm": 1.5838056802749634,
+ "learning_rate": 6.7686709592569046e-06,
+ "loss": 0.0154,
+ "step": 22629
+ },
+ {
+ "epoch": 61.66212534059945,
+ "grad_norm": 1.4820133447647095,
+ "learning_rate": 6.767835819869897e-06,
+ "loss": 0.0187,
+ "step": 22630
+ },
+ {
+ "epoch": 61.664850136239785,
+ "grad_norm": 2.3250060081481934,
+ "learning_rate": 6.767000705654878e-06,
+ "loss": 0.1394,
+ "step": 22631
+ },
+ {
+ "epoch": 61.66757493188011,
+ "grad_norm": 1.9270460605621338,
+ "learning_rate": 6.766165616618354e-06,
+ "loss": 0.0453,
+ "step": 22632
+ },
+ {
+ "epoch": 61.67029972752044,
+ "grad_norm": 3.388057231903076,
+ "learning_rate": 6.76533055276683e-06,
+ "loss": 0.1038,
+ "step": 22633
+ },
+ {
+ "epoch": 61.67302452316076,
+ "grad_norm": 1.9097222089767456,
+ "learning_rate": 6.764495514106806e-06,
+ "loss": 0.0156,
+ "step": 22634
+ },
+ {
+ "epoch": 61.67574931880109,
+ "grad_norm": 1.3384814262390137,
+ "learning_rate": 6.763660500644783e-06,
+ "loss": 0.0156,
+ "step": 22635
+ },
+ {
+ "epoch": 61.678474114441414,
+ "grad_norm": 1.292805552482605,
+ "learning_rate": 6.762825512387273e-06,
+ "loss": 0.0932,
+ "step": 22636
+ },
+ {
+ "epoch": 61.68119891008175,
+ "grad_norm": 1.9934523105621338,
+ "learning_rate": 6.761990549340771e-06,
+ "loss": 0.0738,
+ "step": 22637
+ },
+ {
+ "epoch": 61.68392370572207,
+ "grad_norm": 1.7588810920715332,
+ "learning_rate": 6.761155611511786e-06,
+ "loss": 0.0204,
+ "step": 22638
+ },
+ {
+ "epoch": 61.6866485013624,
+ "grad_norm": 2.656836748123169,
+ "learning_rate": 6.760320698906815e-06,
+ "loss": 0.0496,
+ "step": 22639
+ },
+ {
+ "epoch": 61.689373297002724,
+ "grad_norm": 230.8816375732422,
+ "learning_rate": 6.7594858115323625e-06,
+ "loss": 0.0277,
+ "step": 22640
+ },
+ {
+ "epoch": 61.69209809264305,
+ "grad_norm": 2.420361280441284,
+ "learning_rate": 6.758650949394929e-06,
+ "loss": 0.062,
+ "step": 22641
+ },
+ {
+ "epoch": 61.694822888283376,
+ "grad_norm": 1.9783400297164917,
+ "learning_rate": 6.757816112501019e-06,
+ "loss": 0.1394,
+ "step": 22642
+ },
+ {
+ "epoch": 61.69754768392371,
+ "grad_norm": 3.542457103729248,
+ "learning_rate": 6.756981300857131e-06,
+ "loss": 0.0391,
+ "step": 22643
+ },
+ {
+ "epoch": 61.700272479564035,
+ "grad_norm": 2.4269464015960693,
+ "learning_rate": 6.75614651446977e-06,
+ "loss": 0.1037,
+ "step": 22644
+ },
+ {
+ "epoch": 61.70299727520436,
+ "grad_norm": 2.354921340942383,
+ "learning_rate": 6.7553117533454345e-06,
+ "loss": 0.0285,
+ "step": 22645
+ },
+ {
+ "epoch": 61.705722070844686,
+ "grad_norm": 2.9319584369659424,
+ "learning_rate": 6.754477017490628e-06,
+ "loss": 0.0812,
+ "step": 22646
+ },
+ {
+ "epoch": 61.70844686648501,
+ "grad_norm": 2.63912296295166,
+ "learning_rate": 6.753642306911849e-06,
+ "loss": 0.0377,
+ "step": 22647
+ },
+ {
+ "epoch": 61.71117166212534,
+ "grad_norm": 2.9943294525146484,
+ "learning_rate": 6.7528076216156e-06,
+ "loss": 0.0398,
+ "step": 22648
+ },
+ {
+ "epoch": 61.71389645776567,
+ "grad_norm": 2.8944168090820312,
+ "learning_rate": 6.751972961608379e-06,
+ "loss": 0.0289,
+ "step": 22649
+ },
+ {
+ "epoch": 61.716621253406,
+ "grad_norm": 3.521620512008667,
+ "learning_rate": 6.75113832689669e-06,
+ "loss": 0.1594,
+ "step": 22650
+ },
+ {
+ "epoch": 61.71934604904632,
+ "grad_norm": 4.129425525665283,
+ "learning_rate": 6.75030371748703e-06,
+ "loss": 0.1269,
+ "step": 22651
+ },
+ {
+ "epoch": 61.72207084468665,
+ "grad_norm": 5.5993194580078125,
+ "learning_rate": 6.7494691333858964e-06,
+ "loss": 0.0781,
+ "step": 22652
+ },
+ {
+ "epoch": 61.724795640326974,
+ "grad_norm": 3.443220615386963,
+ "learning_rate": 6.748634574599795e-06,
+ "loss": 0.0699,
+ "step": 22653
+ },
+ {
+ "epoch": 61.7275204359673,
+ "grad_norm": 3.3221867084503174,
+ "learning_rate": 6.747800041135226e-06,
+ "loss": 0.0407,
+ "step": 22654
+ },
+ {
+ "epoch": 61.73024523160763,
+ "grad_norm": 3.4916138648986816,
+ "learning_rate": 6.746965532998684e-06,
+ "loss": 0.0513,
+ "step": 22655
+ },
+ {
+ "epoch": 61.73297002724796,
+ "grad_norm": 2.6795058250427246,
+ "learning_rate": 6.7461310501966695e-06,
+ "loss": 0.0541,
+ "step": 22656
+ },
+ {
+ "epoch": 61.735694822888284,
+ "grad_norm": 3.2988834381103516,
+ "learning_rate": 6.745296592735682e-06,
+ "loss": 0.1973,
+ "step": 22657
+ },
+ {
+ "epoch": 61.73841961852861,
+ "grad_norm": 5.541263580322266,
+ "learning_rate": 6.744462160622221e-06,
+ "loss": 0.1127,
+ "step": 22658
+ },
+ {
+ "epoch": 61.741144414168936,
+ "grad_norm": 4.120975017547607,
+ "learning_rate": 6.7436277538627804e-06,
+ "loss": 0.128,
+ "step": 22659
+ },
+ {
+ "epoch": 61.74386920980926,
+ "grad_norm": 5.196717262268066,
+ "learning_rate": 6.742793372463865e-06,
+ "loss": 0.0942,
+ "step": 22660
+ },
+ {
+ "epoch": 61.746594005449595,
+ "grad_norm": 3.7869162559509277,
+ "learning_rate": 6.741959016431969e-06,
+ "loss": 0.0491,
+ "step": 22661
+ },
+ {
+ "epoch": 61.74931880108992,
+ "grad_norm": 3.1700668334960938,
+ "learning_rate": 6.741124685773592e-06,
+ "loss": 0.1058,
+ "step": 22662
+ },
+ {
+ "epoch": 61.752043596730246,
+ "grad_norm": 4.281494140625,
+ "learning_rate": 6.7402903804952304e-06,
+ "loss": 0.1105,
+ "step": 22663
+ },
+ {
+ "epoch": 61.75476839237057,
+ "grad_norm": 4.370480060577393,
+ "learning_rate": 6.739456100603383e-06,
+ "loss": 0.0529,
+ "step": 22664
+ },
+ {
+ "epoch": 61.7574931880109,
+ "grad_norm": 5.230910301208496,
+ "learning_rate": 6.738621846104546e-06,
+ "loss": 0.0388,
+ "step": 22665
+ },
+ {
+ "epoch": 61.76021798365122,
+ "grad_norm": 3.782536029815674,
+ "learning_rate": 6.737787617005218e-06,
+ "loss": 0.0501,
+ "step": 22666
+ },
+ {
+ "epoch": 61.762942779291556,
+ "grad_norm": 3.1445348262786865,
+ "learning_rate": 6.736953413311893e-06,
+ "loss": 0.1431,
+ "step": 22667
+ },
+ {
+ "epoch": 61.76566757493188,
+ "grad_norm": 4.784895896911621,
+ "learning_rate": 6.736119235031074e-06,
+ "loss": 0.0481,
+ "step": 22668
+ },
+ {
+ "epoch": 61.76839237057221,
+ "grad_norm": 3.6668882369995117,
+ "learning_rate": 6.735285082169246e-06,
+ "loss": 0.059,
+ "step": 22669
+ },
+ {
+ "epoch": 61.771117166212534,
+ "grad_norm": 4.514746189117432,
+ "learning_rate": 6.734450954732916e-06,
+ "loss": 0.0832,
+ "step": 22670
+ },
+ {
+ "epoch": 61.77384196185286,
+ "grad_norm": 5.342491149902344,
+ "learning_rate": 6.7336168527285796e-06,
+ "loss": 0.1273,
+ "step": 22671
+ },
+ {
+ "epoch": 61.776566757493185,
+ "grad_norm": 3.5849053859710693,
+ "learning_rate": 6.732782776162727e-06,
+ "loss": 0.0789,
+ "step": 22672
+ },
+ {
+ "epoch": 61.77929155313352,
+ "grad_norm": 4.34608793258667,
+ "learning_rate": 6.731948725041858e-06,
+ "loss": 0.0499,
+ "step": 22673
+ },
+ {
+ "epoch": 61.782016348773844,
+ "grad_norm": 2.883944272994995,
+ "learning_rate": 6.73111469937247e-06,
+ "loss": 0.0518,
+ "step": 22674
+ },
+ {
+ "epoch": 61.78474114441417,
+ "grad_norm": 6.110548496246338,
+ "learning_rate": 6.730280699161052e-06,
+ "loss": 0.0619,
+ "step": 22675
+ },
+ {
+ "epoch": 61.787465940054496,
+ "grad_norm": 4.144836902618408,
+ "learning_rate": 6.729446724414103e-06,
+ "loss": 0.0539,
+ "step": 22676
+ },
+ {
+ "epoch": 61.79019073569482,
+ "grad_norm": 2.8058536052703857,
+ "learning_rate": 6.7286127751381194e-06,
+ "loss": 0.1343,
+ "step": 22677
+ },
+ {
+ "epoch": 61.79291553133515,
+ "grad_norm": 3.5287489891052246,
+ "learning_rate": 6.727778851339592e-06,
+ "loss": 0.0571,
+ "step": 22678
+ },
+ {
+ "epoch": 61.79564032697548,
+ "grad_norm": 4.850423812866211,
+ "learning_rate": 6.726944953025019e-06,
+ "loss": 0.1054,
+ "step": 22679
+ },
+ {
+ "epoch": 61.798365122615806,
+ "grad_norm": 2.4637844562530518,
+ "learning_rate": 6.7261110802008944e-06,
+ "loss": 0.0326,
+ "step": 22680
+ },
+ {
+ "epoch": 61.80108991825613,
+ "grad_norm": 3.072594404220581,
+ "learning_rate": 6.72527723287371e-06,
+ "loss": 0.1636,
+ "step": 22681
+ },
+ {
+ "epoch": 61.80381471389646,
+ "grad_norm": 2.724639892578125,
+ "learning_rate": 6.724443411049961e-06,
+ "loss": 0.1178,
+ "step": 22682
+ },
+ {
+ "epoch": 61.80653950953678,
+ "grad_norm": 2.3940508365631104,
+ "learning_rate": 6.723609614736142e-06,
+ "loss": 0.0224,
+ "step": 22683
+ },
+ {
+ "epoch": 61.80926430517711,
+ "grad_norm": 2.152684211730957,
+ "learning_rate": 6.722775843938745e-06,
+ "loss": 0.0181,
+ "step": 22684
+ },
+ {
+ "epoch": 61.81198910081744,
+ "grad_norm": 3.3848471641540527,
+ "learning_rate": 6.721942098664264e-06,
+ "loss": 0.0745,
+ "step": 22685
+ },
+ {
+ "epoch": 61.81471389645777,
+ "grad_norm": 2.8656322956085205,
+ "learning_rate": 6.72110837891919e-06,
+ "loss": 0.051,
+ "step": 22686
+ },
+ {
+ "epoch": 61.817438692098094,
+ "grad_norm": 2.9621005058288574,
+ "learning_rate": 6.720274684710023e-06,
+ "loss": 0.1464,
+ "step": 22687
+ },
+ {
+ "epoch": 61.82016348773842,
+ "grad_norm": 5.583528518676758,
+ "learning_rate": 6.719441016043249e-06,
+ "loss": 0.0805,
+ "step": 22688
+ },
+ {
+ "epoch": 61.822888283378745,
+ "grad_norm": 3.469735860824585,
+ "learning_rate": 6.718607372925365e-06,
+ "loss": 0.191,
+ "step": 22689
+ },
+ {
+ "epoch": 61.82561307901907,
+ "grad_norm": 2.295095205307007,
+ "learning_rate": 6.717773755362857e-06,
+ "loss": 0.0603,
+ "step": 22690
+ },
+ {
+ "epoch": 61.828337874659404,
+ "grad_norm": 4.201513290405273,
+ "learning_rate": 6.716940163362227e-06,
+ "loss": 0.0284,
+ "step": 22691
+ },
+ {
+ "epoch": 61.83106267029973,
+ "grad_norm": 2.135514497756958,
+ "learning_rate": 6.716106596929956e-06,
+ "loss": 0.0328,
+ "step": 22692
+ },
+ {
+ "epoch": 61.833787465940055,
+ "grad_norm": 2.383734703063965,
+ "learning_rate": 6.715273056072545e-06,
+ "loss": 0.042,
+ "step": 22693
+ },
+ {
+ "epoch": 61.83651226158038,
+ "grad_norm": 2.8600516319274902,
+ "learning_rate": 6.71443954079648e-06,
+ "loss": 0.026,
+ "step": 22694
+ },
+ {
+ "epoch": 61.83923705722071,
+ "grad_norm": 2.3742969036102295,
+ "learning_rate": 6.713606051108252e-06,
+ "loss": 0.1377,
+ "step": 22695
+ },
+ {
+ "epoch": 61.84196185286103,
+ "grad_norm": 2.3730812072753906,
+ "learning_rate": 6.712772587014359e-06,
+ "loss": 0.0796,
+ "step": 22696
+ },
+ {
+ "epoch": 61.844686648501366,
+ "grad_norm": 2.3614630699157715,
+ "learning_rate": 6.711939148521284e-06,
+ "loss": 0.0519,
+ "step": 22697
+ },
+ {
+ "epoch": 61.84741144414169,
+ "grad_norm": 5.57280158996582,
+ "learning_rate": 6.711105735635524e-06,
+ "loss": 0.083,
+ "step": 22698
+ },
+ {
+ "epoch": 61.85013623978202,
+ "grad_norm": 2.7564094066619873,
+ "learning_rate": 6.710272348363563e-06,
+ "loss": 0.0667,
+ "step": 22699
+ },
+ {
+ "epoch": 61.85286103542234,
+ "grad_norm": 2.055699348449707,
+ "learning_rate": 6.709438986711898e-06,
+ "loss": 0.028,
+ "step": 22700
+ },
+ {
+ "epoch": 61.85558583106267,
+ "grad_norm": 3.221670150756836,
+ "learning_rate": 6.708605650687014e-06,
+ "loss": 0.0397,
+ "step": 22701
+ },
+ {
+ "epoch": 61.858310626702995,
+ "grad_norm": 3.474926471710205,
+ "learning_rate": 6.707772340295404e-06,
+ "loss": 0.0282,
+ "step": 22702
+ },
+ {
+ "epoch": 61.86103542234333,
+ "grad_norm": 2.787043571472168,
+ "learning_rate": 6.706939055543558e-06,
+ "loss": 0.062,
+ "step": 22703
+ },
+ {
+ "epoch": 61.86376021798365,
+ "grad_norm": 2.9837465286254883,
+ "learning_rate": 6.70610579643796e-06,
+ "loss": 0.0338,
+ "step": 22704
+ },
+ {
+ "epoch": 61.86648501362398,
+ "grad_norm": 3.281346082687378,
+ "learning_rate": 6.70527256298511e-06,
+ "loss": 0.0636,
+ "step": 22705
+ },
+ {
+ "epoch": 61.869209809264305,
+ "grad_norm": 3.207984447479248,
+ "learning_rate": 6.704439355191487e-06,
+ "loss": 0.1049,
+ "step": 22706
+ },
+ {
+ "epoch": 61.87193460490463,
+ "grad_norm": 16.86895751953125,
+ "learning_rate": 6.703606173063588e-06,
+ "loss": 0.1166,
+ "step": 22707
+ },
+ {
+ "epoch": 61.87465940054496,
+ "grad_norm": 2.731025218963623,
+ "learning_rate": 6.702773016607895e-06,
+ "loss": 0.0432,
+ "step": 22708
+ },
+ {
+ "epoch": 61.87738419618529,
+ "grad_norm": 2.921461820602417,
+ "learning_rate": 6.701939885830904e-06,
+ "loss": 0.0619,
+ "step": 22709
+ },
+ {
+ "epoch": 61.880108991825615,
+ "grad_norm": 2.7210400104522705,
+ "learning_rate": 6.701106780739095e-06,
+ "loss": 0.0393,
+ "step": 22710
+ },
+ {
+ "epoch": 61.88283378746594,
+ "grad_norm": 2.595778465270996,
+ "learning_rate": 6.700273701338962e-06,
+ "loss": 0.0343,
+ "step": 22711
+ },
+ {
+ "epoch": 61.88555858310627,
+ "grad_norm": 4.686801910400391,
+ "learning_rate": 6.69944064763699e-06,
+ "loss": 0.028,
+ "step": 22712
+ },
+ {
+ "epoch": 61.88828337874659,
+ "grad_norm": 2.4257051944732666,
+ "learning_rate": 6.69860761963967e-06,
+ "loss": 0.0714,
+ "step": 22713
+ },
+ {
+ "epoch": 61.89100817438692,
+ "grad_norm": 2.5374913215637207,
+ "learning_rate": 6.697774617353485e-06,
+ "loss": 0.0476,
+ "step": 22714
+ },
+ {
+ "epoch": 61.89373297002725,
+ "grad_norm": 1.5696215629577637,
+ "learning_rate": 6.6969416407849264e-06,
+ "loss": 0.1233,
+ "step": 22715
+ },
+ {
+ "epoch": 61.89645776566758,
+ "grad_norm": 4.048803329467773,
+ "learning_rate": 6.696108689940479e-06,
+ "loss": 0.0488,
+ "step": 22716
+ },
+ {
+ "epoch": 61.8991825613079,
+ "grad_norm": 2.285104751586914,
+ "learning_rate": 6.695275764826633e-06,
+ "loss": 0.0392,
+ "step": 22717
+ },
+ {
+ "epoch": 61.90190735694823,
+ "grad_norm": 2.4629876613616943,
+ "learning_rate": 6.694442865449871e-06,
+ "loss": 0.1204,
+ "step": 22718
+ },
+ {
+ "epoch": 61.904632152588555,
+ "grad_norm": 3.34283185005188,
+ "learning_rate": 6.6936099918166805e-06,
+ "loss": 0.1483,
+ "step": 22719
+ },
+ {
+ "epoch": 61.90735694822888,
+ "grad_norm": 1.5223777294158936,
+ "learning_rate": 6.692777143933552e-06,
+ "loss": 0.0193,
+ "step": 22720
+ },
+ {
+ "epoch": 61.91008174386921,
+ "grad_norm": 3.2039778232574463,
+ "learning_rate": 6.6919443218069645e-06,
+ "loss": 0.076,
+ "step": 22721
+ },
+ {
+ "epoch": 61.91280653950954,
+ "grad_norm": 2.509147882461548,
+ "learning_rate": 6.69111152544341e-06,
+ "loss": 0.0636,
+ "step": 22722
+ },
+ {
+ "epoch": 61.915531335149865,
+ "grad_norm": 2.346625328063965,
+ "learning_rate": 6.690278754849374e-06,
+ "loss": 0.0326,
+ "step": 22723
+ },
+ {
+ "epoch": 61.91825613079019,
+ "grad_norm": 2.233981132507324,
+ "learning_rate": 6.68944601003134e-06,
+ "loss": 0.0696,
+ "step": 22724
+ },
+ {
+ "epoch": 61.920980926430516,
+ "grad_norm": 1.734760046005249,
+ "learning_rate": 6.688613290995794e-06,
+ "loss": 0.024,
+ "step": 22725
+ },
+ {
+ "epoch": 61.92370572207084,
+ "grad_norm": 2.4145164489746094,
+ "learning_rate": 6.6877805977492205e-06,
+ "loss": 0.0418,
+ "step": 22726
+ },
+ {
+ "epoch": 61.926430517711175,
+ "grad_norm": 2.455000162124634,
+ "learning_rate": 6.686947930298107e-06,
+ "loss": 0.0284,
+ "step": 22727
+ },
+ {
+ "epoch": 61.9291553133515,
+ "grad_norm": 4.450302600860596,
+ "learning_rate": 6.686115288648933e-06,
+ "loss": 0.051,
+ "step": 22728
+ },
+ {
+ "epoch": 61.93188010899183,
+ "grad_norm": 2.4035959243774414,
+ "learning_rate": 6.68528267280819e-06,
+ "loss": 0.0376,
+ "step": 22729
+ },
+ {
+ "epoch": 61.93460490463215,
+ "grad_norm": 3.5143983364105225,
+ "learning_rate": 6.684450082782357e-06,
+ "loss": 0.1262,
+ "step": 22730
+ },
+ {
+ "epoch": 61.93732970027248,
+ "grad_norm": 2.4047353267669678,
+ "learning_rate": 6.6836175185779205e-06,
+ "loss": 0.1089,
+ "step": 22731
+ },
+ {
+ "epoch": 61.940054495912804,
+ "grad_norm": 1.8325798511505127,
+ "learning_rate": 6.682784980201363e-06,
+ "loss": 0.0219,
+ "step": 22732
+ },
+ {
+ "epoch": 61.94277929155314,
+ "grad_norm": 2.846041679382324,
+ "learning_rate": 6.681952467659172e-06,
+ "loss": 0.0551,
+ "step": 22733
+ },
+ {
+ "epoch": 61.94550408719346,
+ "grad_norm": 2.323126792907715,
+ "learning_rate": 6.681119980957825e-06,
+ "loss": 0.1529,
+ "step": 22734
+ },
+ {
+ "epoch": 61.94822888283379,
+ "grad_norm": 4.5325608253479,
+ "learning_rate": 6.680287520103812e-06,
+ "loss": 0.0511,
+ "step": 22735
+ },
+ {
+ "epoch": 61.950953678474114,
+ "grad_norm": 1.832663893699646,
+ "learning_rate": 6.67945508510361e-06,
+ "loss": 0.0242,
+ "step": 22736
+ },
+ {
+ "epoch": 61.95367847411444,
+ "grad_norm": 4.11060905456543,
+ "learning_rate": 6.678622675963708e-06,
+ "loss": 0.0309,
+ "step": 22737
+ },
+ {
+ "epoch": 61.956403269754766,
+ "grad_norm": 2.815330982208252,
+ "learning_rate": 6.677790292690578e-06,
+ "loss": 0.0583,
+ "step": 22738
+ },
+ {
+ "epoch": 61.95912806539509,
+ "grad_norm": 1.7284191846847534,
+ "learning_rate": 6.676957935290718e-06,
+ "loss": 0.0541,
+ "step": 22739
+ },
+ {
+ "epoch": 61.961852861035425,
+ "grad_norm": 2.374342918395996,
+ "learning_rate": 6.676125603770599e-06,
+ "loss": 0.1259,
+ "step": 22740
+ },
+ {
+ "epoch": 61.96457765667575,
+ "grad_norm": 2.6156601905822754,
+ "learning_rate": 6.675293298136709e-06,
+ "loss": 0.0368,
+ "step": 22741
+ },
+ {
+ "epoch": 61.967302452316076,
+ "grad_norm": 2.2382540702819824,
+ "learning_rate": 6.674461018395527e-06,
+ "loss": 0.0665,
+ "step": 22742
+ },
+ {
+ "epoch": 61.9700272479564,
+ "grad_norm": 3.6103553771972656,
+ "learning_rate": 6.6736287645535365e-06,
+ "loss": 0.0777,
+ "step": 22743
+ },
+ {
+ "epoch": 61.97275204359673,
+ "grad_norm": 21.796878814697266,
+ "learning_rate": 6.672796536617218e-06,
+ "loss": 0.101,
+ "step": 22744
+ },
+ {
+ "epoch": 61.97547683923706,
+ "grad_norm": 2.279762029647827,
+ "learning_rate": 6.671964334593053e-06,
+ "loss": 0.0301,
+ "step": 22745
+ },
+ {
+ "epoch": 61.97820163487739,
+ "grad_norm": 2.9274179935455322,
+ "learning_rate": 6.671132158487521e-06,
+ "loss": 0.0406,
+ "step": 22746
+ },
+ {
+ "epoch": 61.98092643051771,
+ "grad_norm": 3.064366102218628,
+ "learning_rate": 6.6703000083071065e-06,
+ "loss": 0.048,
+ "step": 22747
+ },
+ {
+ "epoch": 61.98365122615804,
+ "grad_norm": 3.197705030441284,
+ "learning_rate": 6.669467884058287e-06,
+ "loss": 0.0931,
+ "step": 22748
+ },
+ {
+ "epoch": 61.986376021798364,
+ "grad_norm": 2.4383928775787354,
+ "learning_rate": 6.668635785747547e-06,
+ "loss": 0.0289,
+ "step": 22749
+ },
+ {
+ "epoch": 61.98910081743869,
+ "grad_norm": 3.928455352783203,
+ "learning_rate": 6.667803713381362e-06,
+ "loss": 0.1031,
+ "step": 22750
+ },
+ {
+ "epoch": 61.991825613079016,
+ "grad_norm": 2.2609143257141113,
+ "learning_rate": 6.666971666966216e-06,
+ "loss": 0.0356,
+ "step": 22751
+ },
+ {
+ "epoch": 61.99455040871935,
+ "grad_norm": 4.106645107269287,
+ "learning_rate": 6.6661396465085845e-06,
+ "loss": 0.0398,
+ "step": 22752
+ },
+ {
+ "epoch": 61.997275204359674,
+ "grad_norm": 5.119528293609619,
+ "learning_rate": 6.665307652014954e-06,
+ "loss": 0.0735,
+ "step": 22753
+ },
+ {
+ "epoch": 62.0,
+ "grad_norm": 2.0170986652374268,
+ "learning_rate": 6.664475683491797e-06,
+ "loss": 0.0283,
+ "step": 22754
+ },
+ {
+ "epoch": 62.002724795640326,
+ "grad_norm": 1.8329190015792847,
+ "learning_rate": 6.663643740945594e-06,
+ "loss": 0.0312,
+ "step": 22755
+ },
+ {
+ "epoch": 62.00544959128065,
+ "grad_norm": 2.066864252090454,
+ "learning_rate": 6.66281182438283e-06,
+ "loss": 0.0653,
+ "step": 22756
+ },
+ {
+ "epoch": 62.00817438692098,
+ "grad_norm": 2.3493590354919434,
+ "learning_rate": 6.6619799338099785e-06,
+ "loss": 0.0611,
+ "step": 22757
+ },
+ {
+ "epoch": 62.01089918256131,
+ "grad_norm": 2.370511770248413,
+ "learning_rate": 6.661148069233522e-06,
+ "loss": 0.0293,
+ "step": 22758
+ },
+ {
+ "epoch": 62.013623978201636,
+ "grad_norm": 2.293320894241333,
+ "learning_rate": 6.660316230659935e-06,
+ "loss": 0.111,
+ "step": 22759
+ },
+ {
+ "epoch": 62.01634877384196,
+ "grad_norm": 2.7459592819213867,
+ "learning_rate": 6.6594844180956965e-06,
+ "loss": 0.0329,
+ "step": 22760
+ },
+ {
+ "epoch": 62.01907356948229,
+ "grad_norm": 2.654275894165039,
+ "learning_rate": 6.658652631547288e-06,
+ "loss": 0.1046,
+ "step": 22761
+ },
+ {
+ "epoch": 62.02179836512261,
+ "grad_norm": 1.894126534461975,
+ "learning_rate": 6.657820871021185e-06,
+ "loss": 0.0493,
+ "step": 22762
+ },
+ {
+ "epoch": 62.02452316076294,
+ "grad_norm": 3.0662171840667725,
+ "learning_rate": 6.656989136523866e-06,
+ "loss": 0.049,
+ "step": 22763
+ },
+ {
+ "epoch": 62.02724795640327,
+ "grad_norm": 1.7917289733886719,
+ "learning_rate": 6.656157428061807e-06,
+ "loss": 0.0628,
+ "step": 22764
+ },
+ {
+ "epoch": 62.0299727520436,
+ "grad_norm": 1.9233194589614868,
+ "learning_rate": 6.655325745641488e-06,
+ "loss": 0.0461,
+ "step": 22765
+ },
+ {
+ "epoch": 62.032697547683924,
+ "grad_norm": 2.6187100410461426,
+ "learning_rate": 6.654494089269381e-06,
+ "loss": 0.1599,
+ "step": 22766
+ },
+ {
+ "epoch": 62.03542234332425,
+ "grad_norm": 3.212883710861206,
+ "learning_rate": 6.6536624589519705e-06,
+ "loss": 0.0254,
+ "step": 22767
+ },
+ {
+ "epoch": 62.038147138964575,
+ "grad_norm": 1.8728365898132324,
+ "learning_rate": 6.652830854695726e-06,
+ "loss": 0.0297,
+ "step": 22768
+ },
+ {
+ "epoch": 62.0408719346049,
+ "grad_norm": 2.872628688812256,
+ "learning_rate": 6.651999276507129e-06,
+ "loss": 0.3105,
+ "step": 22769
+ },
+ {
+ "epoch": 62.043596730245234,
+ "grad_norm": 4.030917167663574,
+ "learning_rate": 6.651167724392653e-06,
+ "loss": 0.0671,
+ "step": 22770
+ },
+ {
+ "epoch": 62.04632152588556,
+ "grad_norm": 4.068236827850342,
+ "learning_rate": 6.650336198358776e-06,
+ "loss": 0.0401,
+ "step": 22771
+ },
+ {
+ "epoch": 62.049046321525886,
+ "grad_norm": 2.463329792022705,
+ "learning_rate": 6.649504698411968e-06,
+ "loss": 0.0251,
+ "step": 22772
+ },
+ {
+ "epoch": 62.05177111716621,
+ "grad_norm": 1.6900700330734253,
+ "learning_rate": 6.648673224558713e-06,
+ "loss": 0.0393,
+ "step": 22773
+ },
+ {
+ "epoch": 62.05449591280654,
+ "grad_norm": 3.5069398880004883,
+ "learning_rate": 6.647841776805485e-06,
+ "loss": 0.052,
+ "step": 22774
+ },
+ {
+ "epoch": 62.05722070844686,
+ "grad_norm": 2.950352668762207,
+ "learning_rate": 6.647010355158754e-06,
+ "loss": 0.1258,
+ "step": 22775
+ },
+ {
+ "epoch": 62.059945504087196,
+ "grad_norm": 1.5299322605133057,
+ "learning_rate": 6.646178959625002e-06,
+ "loss": 0.067,
+ "step": 22776
+ },
+ {
+ "epoch": 62.06267029972752,
+ "grad_norm": 1.766417145729065,
+ "learning_rate": 6.6453475902106975e-06,
+ "loss": 0.0529,
+ "step": 22777
+ },
+ {
+ "epoch": 62.06539509536785,
+ "grad_norm": 1.8339983224868774,
+ "learning_rate": 6.644516246922321e-06,
+ "loss": 0.0198,
+ "step": 22778
+ },
+ {
+ "epoch": 62.06811989100817,
+ "grad_norm": 1.4076124429702759,
+ "learning_rate": 6.643684929766341e-06,
+ "loss": 0.0176,
+ "step": 22779
+ },
+ {
+ "epoch": 62.0708446866485,
+ "grad_norm": 2.091317892074585,
+ "learning_rate": 6.642853638749236e-06,
+ "loss": 0.032,
+ "step": 22780
+ },
+ {
+ "epoch": 62.073569482288825,
+ "grad_norm": 36.7926025390625,
+ "learning_rate": 6.642022373877479e-06,
+ "loss": 0.0306,
+ "step": 22781
+ },
+ {
+ "epoch": 62.07629427792916,
+ "grad_norm": 15.386452674865723,
+ "learning_rate": 6.641191135157542e-06,
+ "loss": 0.0735,
+ "step": 22782
+ },
+ {
+ "epoch": 62.079019073569484,
+ "grad_norm": 1.9884388446807861,
+ "learning_rate": 6.640359922595902e-06,
+ "loss": 0.0355,
+ "step": 22783
+ },
+ {
+ "epoch": 62.08174386920981,
+ "grad_norm": 1.3148939609527588,
+ "learning_rate": 6.6395287361990305e-06,
+ "loss": 0.0181,
+ "step": 22784
+ },
+ {
+ "epoch": 62.084468664850135,
+ "grad_norm": 2.245433807373047,
+ "learning_rate": 6.638697575973402e-06,
+ "loss": 0.0542,
+ "step": 22785
+ },
+ {
+ "epoch": 62.08719346049046,
+ "grad_norm": 17.036048889160156,
+ "learning_rate": 6.637866441925487e-06,
+ "loss": 0.0279,
+ "step": 22786
+ },
+ {
+ "epoch": 62.08991825613079,
+ "grad_norm": 2.670304298400879,
+ "learning_rate": 6.637035334061761e-06,
+ "loss": 0.0317,
+ "step": 22787
+ },
+ {
+ "epoch": 62.09264305177112,
+ "grad_norm": 1.936025619506836,
+ "learning_rate": 6.636204252388694e-06,
+ "loss": 0.0279,
+ "step": 22788
+ },
+ {
+ "epoch": 62.095367847411445,
+ "grad_norm": 2.595755100250244,
+ "learning_rate": 6.6353731969127625e-06,
+ "loss": 0.046,
+ "step": 22789
+ },
+ {
+ "epoch": 62.09809264305177,
+ "grad_norm": 5.007326126098633,
+ "learning_rate": 6.634542167640432e-06,
+ "loss": 0.0288,
+ "step": 22790
+ },
+ {
+ "epoch": 62.1008174386921,
+ "grad_norm": 1.7644370794296265,
+ "learning_rate": 6.63371116457818e-06,
+ "loss": 0.1313,
+ "step": 22791
+ },
+ {
+ "epoch": 62.10354223433242,
+ "grad_norm": 1.6949944496154785,
+ "learning_rate": 6.63288018773248e-06,
+ "loss": 0.0268,
+ "step": 22792
+ },
+ {
+ "epoch": 62.10626702997275,
+ "grad_norm": 1.6945101022720337,
+ "learning_rate": 6.632049237109799e-06,
+ "loss": 0.0189,
+ "step": 22793
+ },
+ {
+ "epoch": 62.10899182561308,
+ "grad_norm": 2.637058734893799,
+ "learning_rate": 6.6312183127166115e-06,
+ "loss": 0.1678,
+ "step": 22794
+ },
+ {
+ "epoch": 62.11171662125341,
+ "grad_norm": 1.3698723316192627,
+ "learning_rate": 6.630387414559386e-06,
+ "loss": 0.0231,
+ "step": 22795
+ },
+ {
+ "epoch": 62.11444141689373,
+ "grad_norm": 2.7357101440429688,
+ "learning_rate": 6.629556542644597e-06,
+ "loss": 0.0693,
+ "step": 22796
+ },
+ {
+ "epoch": 62.11716621253406,
+ "grad_norm": 2.532717227935791,
+ "learning_rate": 6.628725696978711e-06,
+ "loss": 0.0279,
+ "step": 22797
+ },
+ {
+ "epoch": 62.119891008174385,
+ "grad_norm": 2.4534006118774414,
+ "learning_rate": 6.627894877568202e-06,
+ "loss": 0.0521,
+ "step": 22798
+ },
+ {
+ "epoch": 62.12261580381471,
+ "grad_norm": 2.8291966915130615,
+ "learning_rate": 6.6270640844195365e-06,
+ "loss": 0.0993,
+ "step": 22799
+ },
+ {
+ "epoch": 62.12534059945504,
+ "grad_norm": 2.97200870513916,
+ "learning_rate": 6.626233317539191e-06,
+ "loss": 0.1137,
+ "step": 22800
+ },
+ {
+ "epoch": 62.12806539509537,
+ "grad_norm": 1.9714057445526123,
+ "learning_rate": 6.6254025769336295e-06,
+ "loss": 0.0228,
+ "step": 22801
+ },
+ {
+ "epoch": 62.130790190735695,
+ "grad_norm": 2.945868968963623,
+ "learning_rate": 6.624571862609326e-06,
+ "loss": 0.1403,
+ "step": 22802
+ },
+ {
+ "epoch": 62.13351498637602,
+ "grad_norm": 2.347957134246826,
+ "learning_rate": 6.623741174572746e-06,
+ "loss": 0.0399,
+ "step": 22803
+ },
+ {
+ "epoch": 62.13623978201635,
+ "grad_norm": 3.1777241230010986,
+ "learning_rate": 6.622910512830363e-06,
+ "loss": 0.031,
+ "step": 22804
+ },
+ {
+ "epoch": 62.13896457765667,
+ "grad_norm": 1.5352649688720703,
+ "learning_rate": 6.622079877388643e-06,
+ "loss": 0.0306,
+ "step": 22805
+ },
+ {
+ "epoch": 62.141689373297005,
+ "grad_norm": 1.2149871587753296,
+ "learning_rate": 6.6212492682540555e-06,
+ "loss": 0.1005,
+ "step": 22806
+ },
+ {
+ "epoch": 62.14441416893733,
+ "grad_norm": 2.192308187484741,
+ "learning_rate": 6.620418685433068e-06,
+ "loss": 0.1058,
+ "step": 22807
+ },
+ {
+ "epoch": 62.14713896457766,
+ "grad_norm": 3.6944706439971924,
+ "learning_rate": 6.619588128932155e-06,
+ "loss": 0.18,
+ "step": 22808
+ },
+ {
+ "epoch": 62.14986376021798,
+ "grad_norm": 3.1399662494659424,
+ "learning_rate": 6.618757598757779e-06,
+ "loss": 0.0408,
+ "step": 22809
+ },
+ {
+ "epoch": 62.15258855585831,
+ "grad_norm": 2.7444114685058594,
+ "learning_rate": 6.617927094916412e-06,
+ "loss": 0.0558,
+ "step": 22810
+ },
+ {
+ "epoch": 62.155313351498634,
+ "grad_norm": 1.8867963552474976,
+ "learning_rate": 6.617096617414518e-06,
+ "loss": 0.0502,
+ "step": 22811
+ },
+ {
+ "epoch": 62.15803814713897,
+ "grad_norm": 2.3751680850982666,
+ "learning_rate": 6.616266166258568e-06,
+ "loss": 0.1696,
+ "step": 22812
+ },
+ {
+ "epoch": 62.16076294277929,
+ "grad_norm": 2.6512250900268555,
+ "learning_rate": 6.615435741455026e-06,
+ "loss": 0.0288,
+ "step": 22813
+ },
+ {
+ "epoch": 62.16348773841962,
+ "grad_norm": 1.6117563247680664,
+ "learning_rate": 6.614605343010365e-06,
+ "loss": 0.0627,
+ "step": 22814
+ },
+ {
+ "epoch": 62.166212534059945,
+ "grad_norm": 2.5908472537994385,
+ "learning_rate": 6.6137749709310465e-06,
+ "loss": 0.048,
+ "step": 22815
+ },
+ {
+ "epoch": 62.16893732970027,
+ "grad_norm": 1.9890637397766113,
+ "learning_rate": 6.612944625223541e-06,
+ "loss": 0.0564,
+ "step": 22816
+ },
+ {
+ "epoch": 62.171662125340596,
+ "grad_norm": 4.130444526672363,
+ "learning_rate": 6.61211430589431e-06,
+ "loss": 0.0453,
+ "step": 22817
+ },
+ {
+ "epoch": 62.17438692098093,
+ "grad_norm": 2.5755152702331543,
+ "learning_rate": 6.611284012949828e-06,
+ "loss": 0.0437,
+ "step": 22818
+ },
+ {
+ "epoch": 62.177111716621255,
+ "grad_norm": 1.961026668548584,
+ "learning_rate": 6.610453746396555e-06,
+ "loss": 0.043,
+ "step": 22819
+ },
+ {
+ "epoch": 62.17983651226158,
+ "grad_norm": 1.5838978290557861,
+ "learning_rate": 6.609623506240961e-06,
+ "loss": 0.0806,
+ "step": 22820
+ },
+ {
+ "epoch": 62.182561307901906,
+ "grad_norm": 2.535766363143921,
+ "learning_rate": 6.608793292489506e-06,
+ "loss": 0.0538,
+ "step": 22821
+ },
+ {
+ "epoch": 62.18528610354223,
+ "grad_norm": 2.0299072265625,
+ "learning_rate": 6.607963105148664e-06,
+ "loss": 0.0293,
+ "step": 22822
+ },
+ {
+ "epoch": 62.18801089918256,
+ "grad_norm": 2.110037088394165,
+ "learning_rate": 6.607132944224893e-06,
+ "loss": 0.0657,
+ "step": 22823
+ },
+ {
+ "epoch": 62.19073569482289,
+ "grad_norm": 3.056575298309326,
+ "learning_rate": 6.6063028097246605e-06,
+ "loss": 0.0331,
+ "step": 22824
+ },
+ {
+ "epoch": 62.19346049046322,
+ "grad_norm": 1.884275197982788,
+ "learning_rate": 6.605472701654434e-06,
+ "loss": 0.0211,
+ "step": 22825
+ },
+ {
+ "epoch": 62.19618528610354,
+ "grad_norm": 2.7828357219696045,
+ "learning_rate": 6.604642620020678e-06,
+ "loss": 0.0527,
+ "step": 22826
+ },
+ {
+ "epoch": 62.19891008174387,
+ "grad_norm": 2.1038355827331543,
+ "learning_rate": 6.603812564829855e-06,
+ "loss": 0.0261,
+ "step": 22827
+ },
+ {
+ "epoch": 62.201634877384194,
+ "grad_norm": 2.3491368293762207,
+ "learning_rate": 6.602982536088431e-06,
+ "loss": 0.0455,
+ "step": 22828
+ },
+ {
+ "epoch": 62.20435967302452,
+ "grad_norm": 2.8241045475006104,
+ "learning_rate": 6.60215253380287e-06,
+ "loss": 0.0728,
+ "step": 22829
+ },
+ {
+ "epoch": 62.20708446866485,
+ "grad_norm": 3.0763161182403564,
+ "learning_rate": 6.601322557979636e-06,
+ "loss": 0.1369,
+ "step": 22830
+ },
+ {
+ "epoch": 62.20980926430518,
+ "grad_norm": 1.6010996103286743,
+ "learning_rate": 6.600492608625191e-06,
+ "loss": 0.025,
+ "step": 22831
+ },
+ {
+ "epoch": 62.212534059945504,
+ "grad_norm": 2.61745548248291,
+ "learning_rate": 6.599662685746002e-06,
+ "loss": 0.0434,
+ "step": 22832
+ },
+ {
+ "epoch": 62.21525885558583,
+ "grad_norm": 2.9148380756378174,
+ "learning_rate": 6.598832789348529e-06,
+ "loss": 0.0534,
+ "step": 22833
+ },
+ {
+ "epoch": 62.217983651226156,
+ "grad_norm": 2.9415106773376465,
+ "learning_rate": 6.598002919439239e-06,
+ "loss": 0.0753,
+ "step": 22834
+ },
+ {
+ "epoch": 62.22070844686648,
+ "grad_norm": 2.0774502754211426,
+ "learning_rate": 6.59717307602459e-06,
+ "loss": 0.0412,
+ "step": 22835
+ },
+ {
+ "epoch": 62.223433242506815,
+ "grad_norm": 2.4154887199401855,
+ "learning_rate": 6.59634325911105e-06,
+ "loss": 0.1005,
+ "step": 22836
+ },
+ {
+ "epoch": 62.22615803814714,
+ "grad_norm": 2.33443021774292,
+ "learning_rate": 6.595513468705077e-06,
+ "loss": 0.0223,
+ "step": 22837
+ },
+ {
+ "epoch": 62.228882833787466,
+ "grad_norm": 4.080283164978027,
+ "learning_rate": 6.594683704813137e-06,
+ "loss": 0.1505,
+ "step": 22838
+ },
+ {
+ "epoch": 62.23160762942779,
+ "grad_norm": 2.8543741703033447,
+ "learning_rate": 6.5938539674416905e-06,
+ "loss": 0.0544,
+ "step": 22839
+ },
+ {
+ "epoch": 62.23433242506812,
+ "grad_norm": 2.624668836593628,
+ "learning_rate": 6.5930242565972e-06,
+ "loss": 0.0305,
+ "step": 22840
+ },
+ {
+ "epoch": 62.237057220708444,
+ "grad_norm": 2.10197114944458,
+ "learning_rate": 6.592194572286124e-06,
+ "loss": 0.032,
+ "step": 22841
+ },
+ {
+ "epoch": 62.23978201634878,
+ "grad_norm": 1.8208422660827637,
+ "learning_rate": 6.591364914514927e-06,
+ "loss": 0.0328,
+ "step": 22842
+ },
+ {
+ "epoch": 62.2425068119891,
+ "grad_norm": 1.8467549085617065,
+ "learning_rate": 6.590535283290074e-06,
+ "loss": 0.0921,
+ "step": 22843
+ },
+ {
+ "epoch": 62.24523160762943,
+ "grad_norm": 2.702537775039673,
+ "learning_rate": 6.58970567861802e-06,
+ "loss": 0.034,
+ "step": 22844
+ },
+ {
+ "epoch": 62.247956403269754,
+ "grad_norm": 2.321485996246338,
+ "learning_rate": 6.5888761005052304e-06,
+ "loss": 0.068,
+ "step": 22845
+ },
+ {
+ "epoch": 62.25068119891008,
+ "grad_norm": 3.025134325027466,
+ "learning_rate": 6.588046548958162e-06,
+ "loss": 0.0773,
+ "step": 22846
+ },
+ {
+ "epoch": 62.253405994550405,
+ "grad_norm": 2.7616875171661377,
+ "learning_rate": 6.587217023983278e-06,
+ "loss": 0.0461,
+ "step": 22847
+ },
+ {
+ "epoch": 62.25613079019074,
+ "grad_norm": 2.003253936767578,
+ "learning_rate": 6.586387525587039e-06,
+ "loss": 0.0242,
+ "step": 22848
+ },
+ {
+ "epoch": 62.258855585831064,
+ "grad_norm": 2.2463009357452393,
+ "learning_rate": 6.585558053775902e-06,
+ "loss": 0.0571,
+ "step": 22849
+ },
+ {
+ "epoch": 62.26158038147139,
+ "grad_norm": 2.247230291366577,
+ "learning_rate": 6.584728608556332e-06,
+ "loss": 0.2196,
+ "step": 22850
+ },
+ {
+ "epoch": 62.264305177111716,
+ "grad_norm": 3.2215380668640137,
+ "learning_rate": 6.583899189934784e-06,
+ "loss": 0.1349,
+ "step": 22851
+ },
+ {
+ "epoch": 62.26702997275204,
+ "grad_norm": 1.994508981704712,
+ "learning_rate": 6.58306979791772e-06,
+ "loss": 0.0444,
+ "step": 22852
+ },
+ {
+ "epoch": 62.26975476839237,
+ "grad_norm": 2.9084417819976807,
+ "learning_rate": 6.5822404325115964e-06,
+ "loss": 0.0687,
+ "step": 22853
+ },
+ {
+ "epoch": 62.2724795640327,
+ "grad_norm": 2.5679659843444824,
+ "learning_rate": 6.581411093722876e-06,
+ "loss": 0.0512,
+ "step": 22854
+ },
+ {
+ "epoch": 62.275204359673026,
+ "grad_norm": 2.59405517578125,
+ "learning_rate": 6.580581781558015e-06,
+ "loss": 0.0382,
+ "step": 22855
+ },
+ {
+ "epoch": 62.27792915531335,
+ "grad_norm": 1.575154185295105,
+ "learning_rate": 6.579752496023475e-06,
+ "loss": 0.0308,
+ "step": 22856
+ },
+ {
+ "epoch": 62.28065395095368,
+ "grad_norm": 2.8456666469573975,
+ "learning_rate": 6.578923237125709e-06,
+ "loss": 0.0289,
+ "step": 22857
+ },
+ {
+ "epoch": 62.283378746594,
+ "grad_norm": 2.01763653755188,
+ "learning_rate": 6.578094004871178e-06,
+ "loss": 0.0417,
+ "step": 22858
+ },
+ {
+ "epoch": 62.28610354223433,
+ "grad_norm": 1.7768104076385498,
+ "learning_rate": 6.577264799266345e-06,
+ "loss": 0.0236,
+ "step": 22859
+ },
+ {
+ "epoch": 62.28882833787466,
+ "grad_norm": 2.180147171020508,
+ "learning_rate": 6.5764356203176605e-06,
+ "loss": 0.0673,
+ "step": 22860
+ },
+ {
+ "epoch": 62.29155313351499,
+ "grad_norm": 1.9418102502822876,
+ "learning_rate": 6.575606468031589e-06,
+ "loss": 0.0302,
+ "step": 22861
+ },
+ {
+ "epoch": 62.294277929155314,
+ "grad_norm": 2.328031301498413,
+ "learning_rate": 6.57477734241458e-06,
+ "loss": 0.0269,
+ "step": 22862
+ },
+ {
+ "epoch": 62.29700272479564,
+ "grad_norm": 1.742129921913147,
+ "learning_rate": 6.573948243473098e-06,
+ "loss": 0.0377,
+ "step": 22863
+ },
+ {
+ "epoch": 62.299727520435965,
+ "grad_norm": 2.3883402347564697,
+ "learning_rate": 6.5731191712135954e-06,
+ "loss": 0.0259,
+ "step": 22864
+ },
+ {
+ "epoch": 62.30245231607629,
+ "grad_norm": 2.4664523601531982,
+ "learning_rate": 6.572290125642531e-06,
+ "loss": 0.1045,
+ "step": 22865
+ },
+ {
+ "epoch": 62.305177111716624,
+ "grad_norm": 2.499922752380371,
+ "learning_rate": 6.571461106766359e-06,
+ "loss": 0.0187,
+ "step": 22866
+ },
+ {
+ "epoch": 62.30790190735695,
+ "grad_norm": 1.6065669059753418,
+ "learning_rate": 6.570632114591541e-06,
+ "loss": 0.034,
+ "step": 22867
+ },
+ {
+ "epoch": 62.310626702997276,
+ "grad_norm": 2.0651285648345947,
+ "learning_rate": 6.569803149124526e-06,
+ "loss": 0.0472,
+ "step": 22868
+ },
+ {
+ "epoch": 62.3133514986376,
+ "grad_norm": 2.2247490882873535,
+ "learning_rate": 6.568974210371775e-06,
+ "loss": 0.0265,
+ "step": 22869
+ },
+ {
+ "epoch": 62.31607629427793,
+ "grad_norm": 2.019850492477417,
+ "learning_rate": 6.568145298339743e-06,
+ "loss": 0.1664,
+ "step": 22870
+ },
+ {
+ "epoch": 62.31880108991825,
+ "grad_norm": 1.9901598691940308,
+ "learning_rate": 6.567316413034884e-06,
+ "loss": 0.0991,
+ "step": 22871
+ },
+ {
+ "epoch": 62.321525885558586,
+ "grad_norm": 1.5778312683105469,
+ "learning_rate": 6.566487554463657e-06,
+ "loss": 0.0294,
+ "step": 22872
+ },
+ {
+ "epoch": 62.32425068119891,
+ "grad_norm": 2.799419641494751,
+ "learning_rate": 6.5656587226325105e-06,
+ "loss": 0.0386,
+ "step": 22873
+ },
+ {
+ "epoch": 62.32697547683924,
+ "grad_norm": 1.647688388824463,
+ "learning_rate": 6.564829917547907e-06,
+ "loss": 0.018,
+ "step": 22874
+ },
+ {
+ "epoch": 62.32970027247956,
+ "grad_norm": 1.8750032186508179,
+ "learning_rate": 6.564001139216294e-06,
+ "loss": 0.0281,
+ "step": 22875
+ },
+ {
+ "epoch": 62.33242506811989,
+ "grad_norm": 2.11152982711792,
+ "learning_rate": 6.563172387644129e-06,
+ "loss": 0.0225,
+ "step": 22876
+ },
+ {
+ "epoch": 62.335149863760215,
+ "grad_norm": 2.346611499786377,
+ "learning_rate": 6.56234366283787e-06,
+ "loss": 0.062,
+ "step": 22877
+ },
+ {
+ "epoch": 62.33787465940055,
+ "grad_norm": 2.0655922889709473,
+ "learning_rate": 6.561514964803966e-06,
+ "loss": 0.0244,
+ "step": 22878
+ },
+ {
+ "epoch": 62.34059945504087,
+ "grad_norm": 3.3258533477783203,
+ "learning_rate": 6.560686293548875e-06,
+ "loss": 0.0271,
+ "step": 22879
+ },
+ {
+ "epoch": 62.3433242506812,
+ "grad_norm": 2.017139196395874,
+ "learning_rate": 6.5598576490790465e-06,
+ "loss": 0.0583,
+ "step": 22880
+ },
+ {
+ "epoch": 62.346049046321525,
+ "grad_norm": 2.503473997116089,
+ "learning_rate": 6.559029031400937e-06,
+ "loss": 0.068,
+ "step": 22881
+ },
+ {
+ "epoch": 62.34877384196185,
+ "grad_norm": 2.677175998687744,
+ "learning_rate": 6.558200440520999e-06,
+ "loss": 0.0394,
+ "step": 22882
+ },
+ {
+ "epoch": 62.35149863760218,
+ "grad_norm": 2.1662659645080566,
+ "learning_rate": 6.557371876445684e-06,
+ "loss": 0.0297,
+ "step": 22883
+ },
+ {
+ "epoch": 62.35422343324251,
+ "grad_norm": 2.748776912689209,
+ "learning_rate": 6.556543339181447e-06,
+ "loss": 0.0519,
+ "step": 22884
+ },
+ {
+ "epoch": 62.356948228882835,
+ "grad_norm": 2.014500379562378,
+ "learning_rate": 6.555714828734739e-06,
+ "loss": 0.0895,
+ "step": 22885
+ },
+ {
+ "epoch": 62.35967302452316,
+ "grad_norm": 4.2649078369140625,
+ "learning_rate": 6.554886345112013e-06,
+ "loss": 0.0647,
+ "step": 22886
+ },
+ {
+ "epoch": 62.36239782016349,
+ "grad_norm": 2.9875376224517822,
+ "learning_rate": 6.554057888319723e-06,
+ "loss": 0.0627,
+ "step": 22887
+ },
+ {
+ "epoch": 62.36512261580381,
+ "grad_norm": 2.161076545715332,
+ "learning_rate": 6.553229458364317e-06,
+ "loss": 0.0259,
+ "step": 22888
+ },
+ {
+ "epoch": 62.36784741144414,
+ "grad_norm": 3.267967939376831,
+ "learning_rate": 6.552401055252251e-06,
+ "loss": 0.059,
+ "step": 22889
+ },
+ {
+ "epoch": 62.37057220708447,
+ "grad_norm": 1.237341046333313,
+ "learning_rate": 6.551572678989973e-06,
+ "loss": 0.0154,
+ "step": 22890
+ },
+ {
+ "epoch": 62.3732970027248,
+ "grad_norm": 1.6759690046310425,
+ "learning_rate": 6.550744329583934e-06,
+ "loss": 0.0488,
+ "step": 22891
+ },
+ {
+ "epoch": 62.37602179836512,
+ "grad_norm": 2.767777919769287,
+ "learning_rate": 6.5499160070405906e-06,
+ "loss": 0.1133,
+ "step": 22892
+ },
+ {
+ "epoch": 62.37874659400545,
+ "grad_norm": 3.2072505950927734,
+ "learning_rate": 6.549087711366386e-06,
+ "loss": 0.15,
+ "step": 22893
+ },
+ {
+ "epoch": 62.381471389645775,
+ "grad_norm": 1.8526103496551514,
+ "learning_rate": 6.548259442567778e-06,
+ "loss": 0.1125,
+ "step": 22894
+ },
+ {
+ "epoch": 62.3841961852861,
+ "grad_norm": 1.3772876262664795,
+ "learning_rate": 6.547431200651217e-06,
+ "loss": 0.0202,
+ "step": 22895
+ },
+ {
+ "epoch": 62.38692098092643,
+ "grad_norm": 2.834193468093872,
+ "learning_rate": 6.546602985623147e-06,
+ "loss": 0.0302,
+ "step": 22896
+ },
+ {
+ "epoch": 62.38964577656676,
+ "grad_norm": 2.077956438064575,
+ "learning_rate": 6.545774797490025e-06,
+ "loss": 0.0304,
+ "step": 22897
+ },
+ {
+ "epoch": 62.392370572207085,
+ "grad_norm": 3.6662421226501465,
+ "learning_rate": 6.544946636258295e-06,
+ "loss": 0.039,
+ "step": 22898
+ },
+ {
+ "epoch": 62.39509536784741,
+ "grad_norm": 1.637887954711914,
+ "learning_rate": 6.544118501934411e-06,
+ "loss": 0.0208,
+ "step": 22899
+ },
+ {
+ "epoch": 62.39782016348774,
+ "grad_norm": 3.826730489730835,
+ "learning_rate": 6.543290394524819e-06,
+ "loss": 0.0687,
+ "step": 22900
+ },
+ {
+ "epoch": 62.40054495912806,
+ "grad_norm": 1.8637757301330566,
+ "learning_rate": 6.542462314035973e-06,
+ "loss": 0.0288,
+ "step": 22901
+ },
+ {
+ "epoch": 62.403269754768395,
+ "grad_norm": 3.098798990249634,
+ "learning_rate": 6.541634260474317e-06,
+ "loss": 0.1634,
+ "step": 22902
+ },
+ {
+ "epoch": 62.40599455040872,
+ "grad_norm": 3.949308395385742,
+ "learning_rate": 6.540806233846304e-06,
+ "loss": 0.0528,
+ "step": 22903
+ },
+ {
+ "epoch": 62.40871934604905,
+ "grad_norm": 3.016284942626953,
+ "learning_rate": 6.539978234158379e-06,
+ "loss": 0.1207,
+ "step": 22904
+ },
+ {
+ "epoch": 62.41144414168937,
+ "grad_norm": 2.2393250465393066,
+ "learning_rate": 6.539150261416993e-06,
+ "loss": 0.1017,
+ "step": 22905
+ },
+ {
+ "epoch": 62.4141689373297,
+ "grad_norm": 1.9822908639907837,
+ "learning_rate": 6.538322315628593e-06,
+ "loss": 0.056,
+ "step": 22906
+ },
+ {
+ "epoch": 62.416893732970024,
+ "grad_norm": 2.2774553298950195,
+ "learning_rate": 6.537494396799629e-06,
+ "loss": 0.0634,
+ "step": 22907
+ },
+ {
+ "epoch": 62.41961852861036,
+ "grad_norm": 3.511004686355591,
+ "learning_rate": 6.536666504936544e-06,
+ "loss": 0.0387,
+ "step": 22908
+ },
+ {
+ "epoch": 62.42234332425068,
+ "grad_norm": 2.2565667629241943,
+ "learning_rate": 6.53583864004579e-06,
+ "loss": 0.0847,
+ "step": 22909
+ },
+ {
+ "epoch": 62.42506811989101,
+ "grad_norm": 2.2247090339660645,
+ "learning_rate": 6.53501080213381e-06,
+ "loss": 0.0693,
+ "step": 22910
+ },
+ {
+ "epoch": 62.427792915531334,
+ "grad_norm": 1.4608837366104126,
+ "learning_rate": 6.534182991207059e-06,
+ "loss": 0.0311,
+ "step": 22911
+ },
+ {
+ "epoch": 62.43051771117166,
+ "grad_norm": 1.3440959453582764,
+ "learning_rate": 6.533355207271976e-06,
+ "loss": 0.0208,
+ "step": 22912
+ },
+ {
+ "epoch": 62.433242506811986,
+ "grad_norm": 1.3674530982971191,
+ "learning_rate": 6.532527450335016e-06,
+ "loss": 0.0493,
+ "step": 22913
+ },
+ {
+ "epoch": 62.43596730245232,
+ "grad_norm": 1.319697618484497,
+ "learning_rate": 6.531699720402616e-06,
+ "loss": 0.0165,
+ "step": 22914
+ },
+ {
+ "epoch": 62.438692098092645,
+ "grad_norm": 1.9022551774978638,
+ "learning_rate": 6.530872017481232e-06,
+ "loss": 0.0488,
+ "step": 22915
+ },
+ {
+ "epoch": 62.44141689373297,
+ "grad_norm": 1.7737370729446411,
+ "learning_rate": 6.5300443415772995e-06,
+ "loss": 0.0276,
+ "step": 22916
+ },
+ {
+ "epoch": 62.444141689373296,
+ "grad_norm": 1.3452422618865967,
+ "learning_rate": 6.529216692697273e-06,
+ "loss": 0.023,
+ "step": 22917
+ },
+ {
+ "epoch": 62.44686648501362,
+ "grad_norm": 2.76774525642395,
+ "learning_rate": 6.528389070847594e-06,
+ "loss": 0.1032,
+ "step": 22918
+ },
+ {
+ "epoch": 62.44959128065395,
+ "grad_norm": 1.467433214187622,
+ "learning_rate": 6.527561476034712e-06,
+ "loss": 0.0173,
+ "step": 22919
+ },
+ {
+ "epoch": 62.45231607629428,
+ "grad_norm": 1.8661813735961914,
+ "learning_rate": 6.526733908265066e-06,
+ "loss": 0.0826,
+ "step": 22920
+ },
+ {
+ "epoch": 62.45504087193461,
+ "grad_norm": 1.9902722835540771,
+ "learning_rate": 6.525906367545108e-06,
+ "loss": 0.0763,
+ "step": 22921
+ },
+ {
+ "epoch": 62.45776566757493,
+ "grad_norm": 2.0381791591644287,
+ "learning_rate": 6.525078853881275e-06,
+ "loss": 0.0224,
+ "step": 22922
+ },
+ {
+ "epoch": 62.46049046321526,
+ "grad_norm": 1.7295013666152954,
+ "learning_rate": 6.524251367280019e-06,
+ "loss": 0.0268,
+ "step": 22923
+ },
+ {
+ "epoch": 62.463215258855584,
+ "grad_norm": 1.930057168006897,
+ "learning_rate": 6.5234239077477815e-06,
+ "loss": 0.0549,
+ "step": 22924
+ },
+ {
+ "epoch": 62.46594005449591,
+ "grad_norm": 2.0777108669281006,
+ "learning_rate": 6.522596475291005e-06,
+ "loss": 0.1769,
+ "step": 22925
+ },
+ {
+ "epoch": 62.46866485013624,
+ "grad_norm": 1.2857882976531982,
+ "learning_rate": 6.521769069916136e-06,
+ "loss": 0.0144,
+ "step": 22926
+ },
+ {
+ "epoch": 62.47138964577657,
+ "grad_norm": 1.7848670482635498,
+ "learning_rate": 6.520941691629615e-06,
+ "loss": 0.0216,
+ "step": 22927
+ },
+ {
+ "epoch": 62.474114441416894,
+ "grad_norm": 2.1347875595092773,
+ "learning_rate": 6.520114340437892e-06,
+ "loss": 0.0482,
+ "step": 22928
+ },
+ {
+ "epoch": 62.47683923705722,
+ "grad_norm": 1.725404977798462,
+ "learning_rate": 6.5192870163474045e-06,
+ "loss": 0.0489,
+ "step": 22929
+ },
+ {
+ "epoch": 62.479564032697546,
+ "grad_norm": 2.0995919704437256,
+ "learning_rate": 6.518459719364601e-06,
+ "loss": 0.1753,
+ "step": 22930
+ },
+ {
+ "epoch": 62.48228882833787,
+ "grad_norm": 2.2850804328918457,
+ "learning_rate": 6.517632449495916e-06,
+ "loss": 0.0375,
+ "step": 22931
+ },
+ {
+ "epoch": 62.485013623978205,
+ "grad_norm": 1.322705626487732,
+ "learning_rate": 6.516805206747803e-06,
+ "loss": 0.0173,
+ "step": 22932
+ },
+ {
+ "epoch": 62.48773841961853,
+ "grad_norm": 3.0661425590515137,
+ "learning_rate": 6.515977991126694e-06,
+ "loss": 0.0559,
+ "step": 22933
+ },
+ {
+ "epoch": 62.490463215258856,
+ "grad_norm": 2.4941446781158447,
+ "learning_rate": 6.515150802639037e-06,
+ "loss": 0.0462,
+ "step": 22934
+ },
+ {
+ "epoch": 62.49318801089918,
+ "grad_norm": 2.5121378898620605,
+ "learning_rate": 6.514323641291276e-06,
+ "loss": 0.0207,
+ "step": 22935
+ },
+ {
+ "epoch": 62.49591280653951,
+ "grad_norm": 2.9467570781707764,
+ "learning_rate": 6.5134965070898496e-06,
+ "loss": 0.12,
+ "step": 22936
+ },
+ {
+ "epoch": 62.49863760217983,
+ "grad_norm": 2.352813959121704,
+ "learning_rate": 6.512669400041201e-06,
+ "loss": 0.0696,
+ "step": 22937
+ },
+ {
+ "epoch": 62.50136239782017,
+ "grad_norm": 2.3857507705688477,
+ "learning_rate": 6.5118423201517665e-06,
+ "loss": 0.1141,
+ "step": 22938
+ },
+ {
+ "epoch": 62.50408719346049,
+ "grad_norm": 1.9703357219696045,
+ "learning_rate": 6.511015267427997e-06,
+ "loss": 0.0287,
+ "step": 22939
+ },
+ {
+ "epoch": 62.50681198910082,
+ "grad_norm": 0.9750725626945496,
+ "learning_rate": 6.510188241876325e-06,
+ "loss": 0.0139,
+ "step": 22940
+ },
+ {
+ "epoch": 62.509536784741144,
+ "grad_norm": 2.015580415725708,
+ "learning_rate": 6.509361243503196e-06,
+ "loss": 0.0764,
+ "step": 22941
+ },
+ {
+ "epoch": 62.51226158038147,
+ "grad_norm": 1.9719252586364746,
+ "learning_rate": 6.508534272315048e-06,
+ "loss": 0.0493,
+ "step": 22942
+ },
+ {
+ "epoch": 62.514986376021795,
+ "grad_norm": 2.097059488296509,
+ "learning_rate": 6.507707328318326e-06,
+ "loss": 0.0377,
+ "step": 22943
+ },
+ {
+ "epoch": 62.51771117166213,
+ "grad_norm": 1.712147831916809,
+ "learning_rate": 6.50688041151946e-06,
+ "loss": 0.0217,
+ "step": 22944
+ },
+ {
+ "epoch": 62.520435967302454,
+ "grad_norm": 3.012880325317383,
+ "learning_rate": 6.506053521924902e-06,
+ "loss": 0.1041,
+ "step": 22945
+ },
+ {
+ "epoch": 62.52316076294278,
+ "grad_norm": 3.751210927963257,
+ "learning_rate": 6.5052266595410874e-06,
+ "loss": 0.0214,
+ "step": 22946
+ },
+ {
+ "epoch": 62.525885558583106,
+ "grad_norm": 2.3271024227142334,
+ "learning_rate": 6.504399824374453e-06,
+ "loss": 0.0506,
+ "step": 22947
+ },
+ {
+ "epoch": 62.52861035422343,
+ "grad_norm": 1.7616724967956543,
+ "learning_rate": 6.503573016431442e-06,
+ "loss": 0.0765,
+ "step": 22948
+ },
+ {
+ "epoch": 62.53133514986376,
+ "grad_norm": 1.835694432258606,
+ "learning_rate": 6.502746235718491e-06,
+ "loss": 0.027,
+ "step": 22949
+ },
+ {
+ "epoch": 62.53405994550409,
+ "grad_norm": 2.880202531814575,
+ "learning_rate": 6.50191948224204e-06,
+ "loss": 0.1697,
+ "step": 22950
+ },
+ {
+ "epoch": 62.536784741144416,
+ "grad_norm": 2.0381364822387695,
+ "learning_rate": 6.501092756008527e-06,
+ "loss": 0.015,
+ "step": 22951
+ },
+ {
+ "epoch": 62.53950953678474,
+ "grad_norm": 2.3571910858154297,
+ "learning_rate": 6.500266057024392e-06,
+ "loss": 0.0429,
+ "step": 22952
+ },
+ {
+ "epoch": 62.54223433242507,
+ "grad_norm": 2.165522575378418,
+ "learning_rate": 6.499439385296071e-06,
+ "loss": 0.0301,
+ "step": 22953
+ },
+ {
+ "epoch": 62.54495912806539,
+ "grad_norm": 2.800168991088867,
+ "learning_rate": 6.498612740830006e-06,
+ "loss": 0.1153,
+ "step": 22954
+ },
+ {
+ "epoch": 62.54768392370572,
+ "grad_norm": 2.0589847564697266,
+ "learning_rate": 6.497786123632628e-06,
+ "loss": 0.1761,
+ "step": 22955
+ },
+ {
+ "epoch": 62.55040871934605,
+ "grad_norm": 8.731778144836426,
+ "learning_rate": 6.4969595337103806e-06,
+ "loss": 0.0447,
+ "step": 22956
+ },
+ {
+ "epoch": 62.55313351498638,
+ "grad_norm": 4.106476783752441,
+ "learning_rate": 6.496132971069702e-06,
+ "loss": 0.0536,
+ "step": 22957
+ },
+ {
+ "epoch": 62.555858310626704,
+ "grad_norm": 1.7994362115859985,
+ "learning_rate": 6.495306435717024e-06,
+ "loss": 0.0676,
+ "step": 22958
+ },
+ {
+ "epoch": 62.55858310626703,
+ "grad_norm": 1.8261828422546387,
+ "learning_rate": 6.49447992765879e-06,
+ "loss": 0.0269,
+ "step": 22959
+ },
+ {
+ "epoch": 62.561307901907355,
+ "grad_norm": 1.9813681840896606,
+ "learning_rate": 6.49365344690143e-06,
+ "loss": 0.0875,
+ "step": 22960
+ },
+ {
+ "epoch": 62.56403269754768,
+ "grad_norm": 1.8104290962219238,
+ "learning_rate": 6.492826993451386e-06,
+ "loss": 0.0267,
+ "step": 22961
+ },
+ {
+ "epoch": 62.566757493188014,
+ "grad_norm": 2.1993188858032227,
+ "learning_rate": 6.49200056731509e-06,
+ "loss": 0.0314,
+ "step": 22962
+ },
+ {
+ "epoch": 62.56948228882834,
+ "grad_norm": 1.9601480960845947,
+ "learning_rate": 6.491174168498981e-06,
+ "loss": 0.0298,
+ "step": 22963
+ },
+ {
+ "epoch": 62.572207084468666,
+ "grad_norm": 2.2463111877441406,
+ "learning_rate": 6.490347797009498e-06,
+ "loss": 0.0988,
+ "step": 22964
+ },
+ {
+ "epoch": 62.57493188010899,
+ "grad_norm": 2.0400123596191406,
+ "learning_rate": 6.489521452853071e-06,
+ "loss": 0.0268,
+ "step": 22965
+ },
+ {
+ "epoch": 62.57765667574932,
+ "grad_norm": 3.618638753890991,
+ "learning_rate": 6.488695136036141e-06,
+ "loss": 0.0986,
+ "step": 22966
+ },
+ {
+ "epoch": 62.58038147138964,
+ "grad_norm": 1.5309263467788696,
+ "learning_rate": 6.487868846565136e-06,
+ "loss": 0.0281,
+ "step": 22967
+ },
+ {
+ "epoch": 62.583106267029976,
+ "grad_norm": 1.895965337753296,
+ "learning_rate": 6.487042584446499e-06,
+ "loss": 0.0411,
+ "step": 22968
+ },
+ {
+ "epoch": 62.5858310626703,
+ "grad_norm": 1.158380150794983,
+ "learning_rate": 6.486216349686658e-06,
+ "loss": 0.0146,
+ "step": 22969
+ },
+ {
+ "epoch": 62.58855585831063,
+ "grad_norm": 1.8465139865875244,
+ "learning_rate": 6.485390142292054e-06,
+ "loss": 0.1065,
+ "step": 22970
+ },
+ {
+ "epoch": 62.59128065395095,
+ "grad_norm": 2.282928466796875,
+ "learning_rate": 6.484563962269117e-06,
+ "loss": 0.1448,
+ "step": 22971
+ },
+ {
+ "epoch": 62.59400544959128,
+ "grad_norm": 2.0712685585021973,
+ "learning_rate": 6.483737809624284e-06,
+ "loss": 0.0302,
+ "step": 22972
+ },
+ {
+ "epoch": 62.596730245231605,
+ "grad_norm": 1.4463768005371094,
+ "learning_rate": 6.482911684363985e-06,
+ "loss": 0.0192,
+ "step": 22973
+ },
+ {
+ "epoch": 62.59945504087194,
+ "grad_norm": 1.5804821252822876,
+ "learning_rate": 6.48208558649466e-06,
+ "loss": 0.0408,
+ "step": 22974
+ },
+ {
+ "epoch": 62.60217983651226,
+ "grad_norm": 1.061888337135315,
+ "learning_rate": 6.481259516022735e-06,
+ "loss": 0.0133,
+ "step": 22975
+ },
+ {
+ "epoch": 62.60490463215259,
+ "grad_norm": 1.8623110055923462,
+ "learning_rate": 6.480433472954653e-06,
+ "loss": 0.0218,
+ "step": 22976
+ },
+ {
+ "epoch": 62.607629427792915,
+ "grad_norm": 2.2384769916534424,
+ "learning_rate": 6.479607457296837e-06,
+ "loss": 0.1015,
+ "step": 22977
+ },
+ {
+ "epoch": 62.61035422343324,
+ "grad_norm": 2.026137113571167,
+ "learning_rate": 6.4787814690557264e-06,
+ "loss": 0.0257,
+ "step": 22978
+ },
+ {
+ "epoch": 62.61307901907357,
+ "grad_norm": 1.9807649850845337,
+ "learning_rate": 6.47795550823775e-06,
+ "loss": 0.0457,
+ "step": 22979
+ },
+ {
+ "epoch": 62.6158038147139,
+ "grad_norm": 2.209599733352661,
+ "learning_rate": 6.4771295748493455e-06,
+ "loss": 0.0679,
+ "step": 22980
+ },
+ {
+ "epoch": 62.618528610354225,
+ "grad_norm": 2.139369249343872,
+ "learning_rate": 6.476303668896941e-06,
+ "loss": 0.0902,
+ "step": 22981
+ },
+ {
+ "epoch": 62.62125340599455,
+ "grad_norm": 2.1566309928894043,
+ "learning_rate": 6.475477790386974e-06,
+ "loss": 0.0337,
+ "step": 22982
+ },
+ {
+ "epoch": 62.62397820163488,
+ "grad_norm": 1.7404741048812866,
+ "learning_rate": 6.474651939325869e-06,
+ "loss": 0.0168,
+ "step": 22983
+ },
+ {
+ "epoch": 62.6267029972752,
+ "grad_norm": 1.928384780883789,
+ "learning_rate": 6.473826115720063e-06,
+ "loss": 0.0272,
+ "step": 22984
+ },
+ {
+ "epoch": 62.62942779291553,
+ "grad_norm": 1.4428085088729858,
+ "learning_rate": 6.473000319575985e-06,
+ "loss": 0.0192,
+ "step": 22985
+ },
+ {
+ "epoch": 62.63215258855586,
+ "grad_norm": 2.029496669769287,
+ "learning_rate": 6.472174550900068e-06,
+ "loss": 0.058,
+ "step": 22986
+ },
+ {
+ "epoch": 62.63487738419619,
+ "grad_norm": 3.414435386657715,
+ "learning_rate": 6.471348809698741e-06,
+ "loss": 0.0434,
+ "step": 22987
+ },
+ {
+ "epoch": 62.63760217983651,
+ "grad_norm": 3.7716896533966064,
+ "learning_rate": 6.470523095978436e-06,
+ "loss": 0.0418,
+ "step": 22988
+ },
+ {
+ "epoch": 62.64032697547684,
+ "grad_norm": 1.7681019306182861,
+ "learning_rate": 6.469697409745583e-06,
+ "loss": 0.0231,
+ "step": 22989
+ },
+ {
+ "epoch": 62.643051771117165,
+ "grad_norm": 1.6684380769729614,
+ "learning_rate": 6.468871751006614e-06,
+ "loss": 0.1162,
+ "step": 22990
+ },
+ {
+ "epoch": 62.64577656675749,
+ "grad_norm": 2.0776121616363525,
+ "learning_rate": 6.468046119767956e-06,
+ "loss": 0.0287,
+ "step": 22991
+ },
+ {
+ "epoch": 62.64850136239782,
+ "grad_norm": 2.4404096603393555,
+ "learning_rate": 6.4672205160360455e-06,
+ "loss": 0.0392,
+ "step": 22992
+ },
+ {
+ "epoch": 62.65122615803815,
+ "grad_norm": 7.1236371994018555,
+ "learning_rate": 6.466394939817305e-06,
+ "loss": 0.038,
+ "step": 22993
+ },
+ {
+ "epoch": 62.653950953678475,
+ "grad_norm": 1.4581724405288696,
+ "learning_rate": 6.465569391118167e-06,
+ "loss": 0.0251,
+ "step": 22994
+ },
+ {
+ "epoch": 62.6566757493188,
+ "grad_norm": 1.919740915298462,
+ "learning_rate": 6.464743869945061e-06,
+ "loss": 0.1517,
+ "step": 22995
+ },
+ {
+ "epoch": 62.65940054495913,
+ "grad_norm": 1.351522445678711,
+ "learning_rate": 6.4639183763044124e-06,
+ "loss": 0.0279,
+ "step": 22996
+ },
+ {
+ "epoch": 62.66212534059945,
+ "grad_norm": 2.569756031036377,
+ "learning_rate": 6.463092910202656e-06,
+ "loss": 0.0314,
+ "step": 22997
+ },
+ {
+ "epoch": 62.664850136239785,
+ "grad_norm": 1.8367096185684204,
+ "learning_rate": 6.462267471646221e-06,
+ "loss": 0.0269,
+ "step": 22998
+ },
+ {
+ "epoch": 62.66757493188011,
+ "grad_norm": 1.4018319845199585,
+ "learning_rate": 6.461442060641531e-06,
+ "loss": 0.0278,
+ "step": 22999
+ },
+ {
+ "epoch": 62.67029972752044,
+ "grad_norm": 2.0142617225646973,
+ "learning_rate": 6.460616677195017e-06,
+ "loss": 0.1618,
+ "step": 23000
+ },
+ {
+ "epoch": 62.67302452316076,
+ "grad_norm": 2.5130691528320312,
+ "learning_rate": 6.4597913213131054e-06,
+ "loss": 0.0691,
+ "step": 23001
+ },
+ {
+ "epoch": 62.67574931880109,
+ "grad_norm": 1.866016149520874,
+ "learning_rate": 6.458965993002228e-06,
+ "loss": 0.0476,
+ "step": 23002
+ },
+ {
+ "epoch": 62.678474114441414,
+ "grad_norm": 1.5038368701934814,
+ "learning_rate": 6.458140692268806e-06,
+ "loss": 0.0389,
+ "step": 23003
+ },
+ {
+ "epoch": 62.68119891008175,
+ "grad_norm": 2.024502992630005,
+ "learning_rate": 6.457315419119273e-06,
+ "loss": 0.1232,
+ "step": 23004
+ },
+ {
+ "epoch": 62.68392370572207,
+ "grad_norm": 1.6649489402770996,
+ "learning_rate": 6.45649017356005e-06,
+ "loss": 0.0148,
+ "step": 23005
+ },
+ {
+ "epoch": 62.6866485013624,
+ "grad_norm": 2.0356884002685547,
+ "learning_rate": 6.4556649555975715e-06,
+ "loss": 0.0345,
+ "step": 23006
+ },
+ {
+ "epoch": 62.689373297002724,
+ "grad_norm": 2.3853237628936768,
+ "learning_rate": 6.454839765238259e-06,
+ "loss": 0.0369,
+ "step": 23007
+ },
+ {
+ "epoch": 62.69209809264305,
+ "grad_norm": 10.888466835021973,
+ "learning_rate": 6.45401460248854e-06,
+ "loss": 0.0268,
+ "step": 23008
+ },
+ {
+ "epoch": 62.694822888283376,
+ "grad_norm": 1.7015737295150757,
+ "learning_rate": 6.45318946735484e-06,
+ "loss": 0.0231,
+ "step": 23009
+ },
+ {
+ "epoch": 62.69754768392371,
+ "grad_norm": 2.018620491027832,
+ "learning_rate": 6.452364359843588e-06,
+ "loss": 0.0744,
+ "step": 23010
+ },
+ {
+ "epoch": 62.700272479564035,
+ "grad_norm": 2.7502167224884033,
+ "learning_rate": 6.451539279961206e-06,
+ "loss": 0.0462,
+ "step": 23011
+ },
+ {
+ "epoch": 62.70299727520436,
+ "grad_norm": 2.330012321472168,
+ "learning_rate": 6.450714227714126e-06,
+ "loss": 0.0195,
+ "step": 23012
+ },
+ {
+ "epoch": 62.705722070844686,
+ "grad_norm": 2.507704496383667,
+ "learning_rate": 6.449889203108762e-06,
+ "loss": 0.0349,
+ "step": 23013
+ },
+ {
+ "epoch": 62.70844686648501,
+ "grad_norm": 1.941423773765564,
+ "learning_rate": 6.449064206151551e-06,
+ "loss": 0.035,
+ "step": 23014
+ },
+ {
+ "epoch": 62.71117166212534,
+ "grad_norm": 1.2020955085754395,
+ "learning_rate": 6.448239236848914e-06,
+ "loss": 0.012,
+ "step": 23015
+ },
+ {
+ "epoch": 62.71389645776567,
+ "grad_norm": 3.059718132019043,
+ "learning_rate": 6.447414295207275e-06,
+ "loss": 0.0382,
+ "step": 23016
+ },
+ {
+ "epoch": 62.716621253406,
+ "grad_norm": 1.6623862981796265,
+ "learning_rate": 6.446589381233062e-06,
+ "loss": 0.0423,
+ "step": 23017
+ },
+ {
+ "epoch": 62.71934604904632,
+ "grad_norm": 1.9917211532592773,
+ "learning_rate": 6.445764494932691e-06,
+ "loss": 0.0262,
+ "step": 23018
+ },
+ {
+ "epoch": 62.72207084468665,
+ "grad_norm": 1.4317024946212769,
+ "learning_rate": 6.444939636312597e-06,
+ "loss": 0.0231,
+ "step": 23019
+ },
+ {
+ "epoch": 62.724795640326974,
+ "grad_norm": 2.208904266357422,
+ "learning_rate": 6.4441148053791956e-06,
+ "loss": 0.0999,
+ "step": 23020
+ },
+ {
+ "epoch": 62.7275204359673,
+ "grad_norm": 1.9855479001998901,
+ "learning_rate": 6.443290002138913e-06,
+ "loss": 0.1023,
+ "step": 23021
+ },
+ {
+ "epoch": 62.73024523160763,
+ "grad_norm": 1.572440505027771,
+ "learning_rate": 6.442465226598177e-06,
+ "loss": 0.0158,
+ "step": 23022
+ },
+ {
+ "epoch": 62.73297002724796,
+ "grad_norm": 1.8311876058578491,
+ "learning_rate": 6.4416404787634045e-06,
+ "loss": 0.0254,
+ "step": 23023
+ },
+ {
+ "epoch": 62.735694822888284,
+ "grad_norm": 1.9474363327026367,
+ "learning_rate": 6.440815758641023e-06,
+ "loss": 0.0305,
+ "step": 23024
+ },
+ {
+ "epoch": 62.73841961852861,
+ "grad_norm": 1.6442605257034302,
+ "learning_rate": 6.439991066237452e-06,
+ "loss": 0.0228,
+ "step": 23025
+ },
+ {
+ "epoch": 62.741144414168936,
+ "grad_norm": 1.973140835762024,
+ "learning_rate": 6.4391664015591185e-06,
+ "loss": 0.0487,
+ "step": 23026
+ },
+ {
+ "epoch": 62.74386920980926,
+ "grad_norm": 1.9395607709884644,
+ "learning_rate": 6.438341764612439e-06,
+ "loss": 0.031,
+ "step": 23027
+ },
+ {
+ "epoch": 62.746594005449595,
+ "grad_norm": 1.2907744646072388,
+ "learning_rate": 6.437517155403843e-06,
+ "loss": 0.0138,
+ "step": 23028
+ },
+ {
+ "epoch": 62.74931880108992,
+ "grad_norm": 1.4190521240234375,
+ "learning_rate": 6.4366925739397456e-06,
+ "loss": 0.0201,
+ "step": 23029
+ },
+ {
+ "epoch": 62.752043596730246,
+ "grad_norm": 1.7052805423736572,
+ "learning_rate": 6.4358680202265696e-06,
+ "loss": 0.0207,
+ "step": 23030
+ },
+ {
+ "epoch": 62.75476839237057,
+ "grad_norm": 2.0561435222625732,
+ "learning_rate": 6.435043494270743e-06,
+ "loss": 0.0275,
+ "step": 23031
+ },
+ {
+ "epoch": 62.7574931880109,
+ "grad_norm": 1.5908010005950928,
+ "learning_rate": 6.434218996078683e-06,
+ "loss": 0.1154,
+ "step": 23032
+ },
+ {
+ "epoch": 62.76021798365122,
+ "grad_norm": 1.1319327354431152,
+ "learning_rate": 6.43339452565681e-06,
+ "loss": 0.0171,
+ "step": 23033
+ },
+ {
+ "epoch": 62.762942779291556,
+ "grad_norm": 1.3068573474884033,
+ "learning_rate": 6.432570083011544e-06,
+ "loss": 0.0093,
+ "step": 23034
+ },
+ {
+ "epoch": 62.76566757493188,
+ "grad_norm": 1.4871233701705933,
+ "learning_rate": 6.43174566814931e-06,
+ "loss": 0.0423,
+ "step": 23035
+ },
+ {
+ "epoch": 62.76839237057221,
+ "grad_norm": 2.022487163543701,
+ "learning_rate": 6.4309212810765245e-06,
+ "loss": 0.0204,
+ "step": 23036
+ },
+ {
+ "epoch": 62.771117166212534,
+ "grad_norm": 2.4005632400512695,
+ "learning_rate": 6.43009692179961e-06,
+ "loss": 0.0214,
+ "step": 23037
+ },
+ {
+ "epoch": 62.77384196185286,
+ "grad_norm": 2.763737916946411,
+ "learning_rate": 6.429272590324984e-06,
+ "loss": 0.0247,
+ "step": 23038
+ },
+ {
+ "epoch": 62.776566757493185,
+ "grad_norm": 3.107404947280884,
+ "learning_rate": 6.428448286659071e-06,
+ "loss": 0.0472,
+ "step": 23039
+ },
+ {
+ "epoch": 62.77929155313352,
+ "grad_norm": 2.6678738594055176,
+ "learning_rate": 6.427624010808284e-06,
+ "loss": 0.0442,
+ "step": 23040
+ },
+ {
+ "epoch": 62.782016348773844,
+ "grad_norm": 2.5203213691711426,
+ "learning_rate": 6.42679976277905e-06,
+ "loss": 0.0428,
+ "step": 23041
+ },
+ {
+ "epoch": 62.78474114441417,
+ "grad_norm": 1.3749741315841675,
+ "learning_rate": 6.425975542577783e-06,
+ "loss": 0.0173,
+ "step": 23042
+ },
+ {
+ "epoch": 62.787465940054496,
+ "grad_norm": 2.1115705966949463,
+ "learning_rate": 6.425151350210901e-06,
+ "loss": 0.0934,
+ "step": 23043
+ },
+ {
+ "epoch": 62.79019073569482,
+ "grad_norm": 2.4776971340179443,
+ "learning_rate": 6.424327185684829e-06,
+ "loss": 0.0473,
+ "step": 23044
+ },
+ {
+ "epoch": 62.79291553133515,
+ "grad_norm": 1.8242825269699097,
+ "learning_rate": 6.423503049005979e-06,
+ "loss": 0.0314,
+ "step": 23045
+ },
+ {
+ "epoch": 62.79564032697548,
+ "grad_norm": 1.3651087284088135,
+ "learning_rate": 6.422678940180775e-06,
+ "loss": 0.0156,
+ "step": 23046
+ },
+ {
+ "epoch": 62.798365122615806,
+ "grad_norm": 2.1449661254882812,
+ "learning_rate": 6.42185485921563e-06,
+ "loss": 0.1642,
+ "step": 23047
+ },
+ {
+ "epoch": 62.80108991825613,
+ "grad_norm": 1.8731728792190552,
+ "learning_rate": 6.421030806116961e-06,
+ "loss": 0.0308,
+ "step": 23048
+ },
+ {
+ "epoch": 62.80381471389646,
+ "grad_norm": 1.7583208084106445,
+ "learning_rate": 6.420206780891194e-06,
+ "loss": 0.0878,
+ "step": 23049
+ },
+ {
+ "epoch": 62.80653950953678,
+ "grad_norm": 1.5400804281234741,
+ "learning_rate": 6.419382783544739e-06,
+ "loss": 0.0273,
+ "step": 23050
+ },
+ {
+ "epoch": 62.80926430517711,
+ "grad_norm": 1.0502700805664062,
+ "learning_rate": 6.4185588140840195e-06,
+ "loss": 0.0155,
+ "step": 23051
+ },
+ {
+ "epoch": 62.81198910081744,
+ "grad_norm": 1.6970314979553223,
+ "learning_rate": 6.417734872515444e-06,
+ "loss": 0.1064,
+ "step": 23052
+ },
+ {
+ "epoch": 62.81471389645777,
+ "grad_norm": 1.4410903453826904,
+ "learning_rate": 6.416910958845437e-06,
+ "loss": 0.0234,
+ "step": 23053
+ },
+ {
+ "epoch": 62.817438692098094,
+ "grad_norm": 1.714158058166504,
+ "learning_rate": 6.41608707308041e-06,
+ "loss": 0.0216,
+ "step": 23054
+ },
+ {
+ "epoch": 62.82016348773842,
+ "grad_norm": 9.255804061889648,
+ "learning_rate": 6.415263215226785e-06,
+ "loss": 0.0316,
+ "step": 23055
+ },
+ {
+ "epoch": 62.822888283378745,
+ "grad_norm": 1.5248570442199707,
+ "learning_rate": 6.414439385290971e-06,
+ "loss": 0.0385,
+ "step": 23056
+ },
+ {
+ "epoch": 62.82561307901907,
+ "grad_norm": 1.5541476011276245,
+ "learning_rate": 6.41361558327939e-06,
+ "loss": 0.0524,
+ "step": 23057
+ },
+ {
+ "epoch": 62.828337874659404,
+ "grad_norm": 1.7881866693496704,
+ "learning_rate": 6.412791809198453e-06,
+ "loss": 0.155,
+ "step": 23058
+ },
+ {
+ "epoch": 62.83106267029973,
+ "grad_norm": 1.7681828737258911,
+ "learning_rate": 6.411968063054581e-06,
+ "loss": 0.164,
+ "step": 23059
+ },
+ {
+ "epoch": 62.833787465940055,
+ "grad_norm": 1.6102309226989746,
+ "learning_rate": 6.411144344854183e-06,
+ "loss": 0.0269,
+ "step": 23060
+ },
+ {
+ "epoch": 62.83651226158038,
+ "grad_norm": 1.9925553798675537,
+ "learning_rate": 6.410320654603681e-06,
+ "loss": 0.0563,
+ "step": 23061
+ },
+ {
+ "epoch": 62.83923705722071,
+ "grad_norm": 1.9043070077896118,
+ "learning_rate": 6.409496992309483e-06,
+ "loss": 0.0285,
+ "step": 23062
+ },
+ {
+ "epoch": 62.84196185286103,
+ "grad_norm": 41.13208770751953,
+ "learning_rate": 6.408673357978009e-06,
+ "loss": 0.0437,
+ "step": 23063
+ },
+ {
+ "epoch": 62.844686648501366,
+ "grad_norm": 1.1788111925125122,
+ "learning_rate": 6.407849751615669e-06,
+ "loss": 0.0152,
+ "step": 23064
+ },
+ {
+ "epoch": 62.84741144414169,
+ "grad_norm": 2.8239290714263916,
+ "learning_rate": 6.407026173228878e-06,
+ "loss": 0.1269,
+ "step": 23065
+ },
+ {
+ "epoch": 62.85013623978202,
+ "grad_norm": 2.2240958213806152,
+ "learning_rate": 6.406202622824054e-06,
+ "loss": 0.0762,
+ "step": 23066
+ },
+ {
+ "epoch": 62.85286103542234,
+ "grad_norm": 1.7635234594345093,
+ "learning_rate": 6.40537910040761e-06,
+ "loss": 0.0194,
+ "step": 23067
+ },
+ {
+ "epoch": 62.85558583106267,
+ "grad_norm": 12.859148025512695,
+ "learning_rate": 6.404555605985956e-06,
+ "loss": 0.0593,
+ "step": 23068
+ },
+ {
+ "epoch": 62.858310626702995,
+ "grad_norm": 1.7925300598144531,
+ "learning_rate": 6.403732139565508e-06,
+ "loss": 0.0767,
+ "step": 23069
+ },
+ {
+ "epoch": 62.86103542234333,
+ "grad_norm": 4.014009952545166,
+ "learning_rate": 6.402908701152677e-06,
+ "loss": 0.0446,
+ "step": 23070
+ },
+ {
+ "epoch": 62.86376021798365,
+ "grad_norm": 1.2239246368408203,
+ "learning_rate": 6.40208529075388e-06,
+ "loss": 0.017,
+ "step": 23071
+ },
+ {
+ "epoch": 62.86648501362398,
+ "grad_norm": 1.4607328176498413,
+ "learning_rate": 6.401261908375525e-06,
+ "loss": 0.0432,
+ "step": 23072
+ },
+ {
+ "epoch": 62.869209809264305,
+ "grad_norm": 3.0785200595855713,
+ "learning_rate": 6.4004385540240266e-06,
+ "loss": 0.1136,
+ "step": 23073
+ },
+ {
+ "epoch": 62.87193460490463,
+ "grad_norm": 2.323744058609009,
+ "learning_rate": 6.399615227705797e-06,
+ "loss": 0.0642,
+ "step": 23074
+ },
+ {
+ "epoch": 62.87465940054496,
+ "grad_norm": 1.8489865064620972,
+ "learning_rate": 6.398791929427248e-06,
+ "loss": 0.0706,
+ "step": 23075
+ },
+ {
+ "epoch": 62.87738419618529,
+ "grad_norm": 1.5804264545440674,
+ "learning_rate": 6.397968659194792e-06,
+ "loss": 0.0175,
+ "step": 23076
+ },
+ {
+ "epoch": 62.880108991825615,
+ "grad_norm": 2.6543498039245605,
+ "learning_rate": 6.397145417014841e-06,
+ "loss": 0.1719,
+ "step": 23077
+ },
+ {
+ "epoch": 62.88283378746594,
+ "grad_norm": 2.2001142501831055,
+ "learning_rate": 6.3963222028938035e-06,
+ "loss": 0.0607,
+ "step": 23078
+ },
+ {
+ "epoch": 62.88555858310627,
+ "grad_norm": 1.9768741130828857,
+ "learning_rate": 6.395499016838095e-06,
+ "loss": 0.1223,
+ "step": 23079
+ },
+ {
+ "epoch": 62.88828337874659,
+ "grad_norm": 2.5268473625183105,
+ "learning_rate": 6.394675858854121e-06,
+ "loss": 0.0252,
+ "step": 23080
+ },
+ {
+ "epoch": 62.89100817438692,
+ "grad_norm": 2.208691120147705,
+ "learning_rate": 6.393852728948299e-06,
+ "loss": 0.0248,
+ "step": 23081
+ },
+ {
+ "epoch": 62.89373297002725,
+ "grad_norm": 1.8417878150939941,
+ "learning_rate": 6.393029627127031e-06,
+ "loss": 0.0784,
+ "step": 23082
+ },
+ {
+ "epoch": 62.89645776566758,
+ "grad_norm": 2.403280735015869,
+ "learning_rate": 6.392206553396734e-06,
+ "loss": 0.053,
+ "step": 23083
+ },
+ {
+ "epoch": 62.8991825613079,
+ "grad_norm": 2.5330371856689453,
+ "learning_rate": 6.391383507763816e-06,
+ "loss": 0.0336,
+ "step": 23084
+ },
+ {
+ "epoch": 62.90190735694823,
+ "grad_norm": 2.249795913696289,
+ "learning_rate": 6.39056049023469e-06,
+ "loss": 0.0206,
+ "step": 23085
+ },
+ {
+ "epoch": 62.904632152588555,
+ "grad_norm": 2.328019142150879,
+ "learning_rate": 6.3897375008157606e-06,
+ "loss": 0.0248,
+ "step": 23086
+ },
+ {
+ "epoch": 62.90735694822888,
+ "grad_norm": 1.9938608407974243,
+ "learning_rate": 6.388914539513441e-06,
+ "loss": 0.0367,
+ "step": 23087
+ },
+ {
+ "epoch": 62.91008174386921,
+ "grad_norm": 2.566476583480835,
+ "learning_rate": 6.388091606334137e-06,
+ "loss": 0.1918,
+ "step": 23088
+ },
+ {
+ "epoch": 62.91280653950954,
+ "grad_norm": 2.9487125873565674,
+ "learning_rate": 6.387268701284261e-06,
+ "loss": 0.0844,
+ "step": 23089
+ },
+ {
+ "epoch": 62.915531335149865,
+ "grad_norm": 2.788599967956543,
+ "learning_rate": 6.386445824370218e-06,
+ "loss": 0.0442,
+ "step": 23090
+ },
+ {
+ "epoch": 62.91825613079019,
+ "grad_norm": 1.9741172790527344,
+ "learning_rate": 6.385622975598422e-06,
+ "loss": 0.0272,
+ "step": 23091
+ },
+ {
+ "epoch": 62.920980926430516,
+ "grad_norm": 2.019543409347534,
+ "learning_rate": 6.384800154975274e-06,
+ "loss": 0.0758,
+ "step": 23092
+ },
+ {
+ "epoch": 62.92370572207084,
+ "grad_norm": 2.9114151000976562,
+ "learning_rate": 6.38397736250719e-06,
+ "loss": 0.0532,
+ "step": 23093
+ },
+ {
+ "epoch": 62.926430517711175,
+ "grad_norm": 1.7815196514129639,
+ "learning_rate": 6.38315459820057e-06,
+ "loss": 0.0206,
+ "step": 23094
+ },
+ {
+ "epoch": 62.9291553133515,
+ "grad_norm": 2.607203960418701,
+ "learning_rate": 6.382331862061828e-06,
+ "loss": 0.1402,
+ "step": 23095
+ },
+ {
+ "epoch": 62.93188010899183,
+ "grad_norm": 2.4145383834838867,
+ "learning_rate": 6.381509154097367e-06,
+ "loss": 0.1863,
+ "step": 23096
+ },
+ {
+ "epoch": 62.93460490463215,
+ "grad_norm": 2.1817140579223633,
+ "learning_rate": 6.3806864743136e-06,
+ "loss": 0.0574,
+ "step": 23097
+ },
+ {
+ "epoch": 62.93732970027248,
+ "grad_norm": 2.3111159801483154,
+ "learning_rate": 6.379863822716926e-06,
+ "loss": 0.0399,
+ "step": 23098
+ },
+ {
+ "epoch": 62.940054495912804,
+ "grad_norm": 1.9696258306503296,
+ "learning_rate": 6.379041199313756e-06,
+ "loss": 0.0207,
+ "step": 23099
+ },
+ {
+ "epoch": 62.94277929155314,
+ "grad_norm": 2.170783519744873,
+ "learning_rate": 6.378218604110499e-06,
+ "loss": 0.0941,
+ "step": 23100
+ },
+ {
+ "epoch": 62.94550408719346,
+ "grad_norm": 2.2742905616760254,
+ "learning_rate": 6.377396037113557e-06,
+ "loss": 0.091,
+ "step": 23101
+ },
+ {
+ "epoch": 62.94822888283379,
+ "grad_norm": 2.158107042312622,
+ "learning_rate": 6.376573498329341e-06,
+ "loss": 0.0204,
+ "step": 23102
+ },
+ {
+ "epoch": 62.950953678474114,
+ "grad_norm": 1.905707597732544,
+ "learning_rate": 6.375750987764251e-06,
+ "loss": 0.0219,
+ "step": 23103
+ },
+ {
+ "epoch": 62.95367847411444,
+ "grad_norm": 1.9057852029800415,
+ "learning_rate": 6.374928505424698e-06,
+ "loss": 0.022,
+ "step": 23104
+ },
+ {
+ "epoch": 62.956403269754766,
+ "grad_norm": 1.4343081712722778,
+ "learning_rate": 6.374106051317084e-06,
+ "loss": 0.0379,
+ "step": 23105
+ },
+ {
+ "epoch": 62.95912806539509,
+ "grad_norm": 1.9395034313201904,
+ "learning_rate": 6.373283625447817e-06,
+ "loss": 0.0292,
+ "step": 23106
+ },
+ {
+ "epoch": 62.961852861035425,
+ "grad_norm": 1.601516842842102,
+ "learning_rate": 6.372461227823297e-06,
+ "loss": 0.0207,
+ "step": 23107
+ },
+ {
+ "epoch": 62.96457765667575,
+ "grad_norm": 2.5253498554229736,
+ "learning_rate": 6.371638858449933e-06,
+ "loss": 0.029,
+ "step": 23108
+ },
+ {
+ "epoch": 62.967302452316076,
+ "grad_norm": 2.0850813388824463,
+ "learning_rate": 6.370816517334132e-06,
+ "loss": 0.0425,
+ "step": 23109
+ },
+ {
+ "epoch": 62.9700272479564,
+ "grad_norm": 2.189824342727661,
+ "learning_rate": 6.369994204482293e-06,
+ "loss": 0.0198,
+ "step": 23110
+ },
+ {
+ "epoch": 62.97275204359673,
+ "grad_norm": 2.3927481174468994,
+ "learning_rate": 6.369171919900823e-06,
+ "loss": 0.0376,
+ "step": 23111
+ },
+ {
+ "epoch": 62.97547683923706,
+ "grad_norm": 1.6316529512405396,
+ "learning_rate": 6.368349663596123e-06,
+ "loss": 0.0262,
+ "step": 23112
+ },
+ {
+ "epoch": 62.97820163487739,
+ "grad_norm": 2.2630863189697266,
+ "learning_rate": 6.367527435574602e-06,
+ "loss": 0.0804,
+ "step": 23113
+ },
+ {
+ "epoch": 62.98092643051771,
+ "grad_norm": 2.308218002319336,
+ "learning_rate": 6.366705235842658e-06,
+ "loss": 0.0288,
+ "step": 23114
+ },
+ {
+ "epoch": 62.98365122615804,
+ "grad_norm": 3.2873218059539795,
+ "learning_rate": 6.365883064406699e-06,
+ "loss": 0.0396,
+ "step": 23115
+ },
+ {
+ "epoch": 62.986376021798364,
+ "grad_norm": 3.5017824172973633,
+ "learning_rate": 6.365060921273121e-06,
+ "loss": 0.0283,
+ "step": 23116
+ },
+ {
+ "epoch": 62.98910081743869,
+ "grad_norm": 2.1721370220184326,
+ "learning_rate": 6.364238806448335e-06,
+ "loss": 0.0311,
+ "step": 23117
+ },
+ {
+ "epoch": 62.991825613079016,
+ "grad_norm": 2.472202777862549,
+ "learning_rate": 6.3634167199387405e-06,
+ "loss": 0.0742,
+ "step": 23118
+ },
+ {
+ "epoch": 62.99455040871935,
+ "grad_norm": 5.650823593139648,
+ "learning_rate": 6.362594661750739e-06,
+ "loss": 0.093,
+ "step": 23119
+ },
+ {
+ "epoch": 62.997275204359674,
+ "grad_norm": 1.5006712675094604,
+ "learning_rate": 6.361772631890735e-06,
+ "loss": 0.0222,
+ "step": 23120
+ },
+ {
+ "epoch": 63.0,
+ "grad_norm": 1.5615804195404053,
+ "learning_rate": 6.360950630365126e-06,
+ "loss": 0.0252,
+ "step": 23121
+ },
+ {
+ "epoch": 63.002724795640326,
+ "grad_norm": 1.5731377601623535,
+ "learning_rate": 6.360128657180318e-06,
+ "loss": 0.1361,
+ "step": 23122
+ },
+ {
+ "epoch": 63.00544959128065,
+ "grad_norm": 3.1710281372070312,
+ "learning_rate": 6.3593067123427096e-06,
+ "loss": 0.0877,
+ "step": 23123
+ },
+ {
+ "epoch": 63.00817438692098,
+ "grad_norm": 1.1909493207931519,
+ "learning_rate": 6.358484795858704e-06,
+ "loss": 0.0152,
+ "step": 23124
+ },
+ {
+ "epoch": 63.01089918256131,
+ "grad_norm": 1.2301257848739624,
+ "learning_rate": 6.357662907734701e-06,
+ "loss": 0.0267,
+ "step": 23125
+ },
+ {
+ "epoch": 63.013623978201636,
+ "grad_norm": 2.2955784797668457,
+ "learning_rate": 6.356841047977104e-06,
+ "loss": 0.0538,
+ "step": 23126
+ },
+ {
+ "epoch": 63.01634877384196,
+ "grad_norm": 1.8583242893218994,
+ "learning_rate": 6.35601921659231e-06,
+ "loss": 0.0314,
+ "step": 23127
+ },
+ {
+ "epoch": 63.01907356948229,
+ "grad_norm": 1.827971339225769,
+ "learning_rate": 6.355197413586722e-06,
+ "loss": 0.016,
+ "step": 23128
+ },
+ {
+ "epoch": 63.02179836512261,
+ "grad_norm": 1.3488885164260864,
+ "learning_rate": 6.354375638966739e-06,
+ "loss": 0.011,
+ "step": 23129
+ },
+ {
+ "epoch": 63.02452316076294,
+ "grad_norm": 2.28116774559021,
+ "learning_rate": 6.35355389273876e-06,
+ "loss": 0.0374,
+ "step": 23130
+ },
+ {
+ "epoch": 63.02724795640327,
+ "grad_norm": 2.10119366645813,
+ "learning_rate": 6.352732174909187e-06,
+ "loss": 0.0623,
+ "step": 23131
+ },
+ {
+ "epoch": 63.0299727520436,
+ "grad_norm": 2.406144618988037,
+ "learning_rate": 6.351910485484417e-06,
+ "loss": 0.0397,
+ "step": 23132
+ },
+ {
+ "epoch": 63.032697547683924,
+ "grad_norm": 1.3977197408676147,
+ "learning_rate": 6.351088824470853e-06,
+ "loss": 0.0174,
+ "step": 23133
+ },
+ {
+ "epoch": 63.03542234332425,
+ "grad_norm": 1.5811264514923096,
+ "learning_rate": 6.3502671918748886e-06,
+ "loss": 0.118,
+ "step": 23134
+ },
+ {
+ "epoch": 63.038147138964575,
+ "grad_norm": 1.6673686504364014,
+ "learning_rate": 6.349445587702927e-06,
+ "loss": 0.014,
+ "step": 23135
+ },
+ {
+ "epoch": 63.0408719346049,
+ "grad_norm": 1.809087872505188,
+ "learning_rate": 6.348624011961368e-06,
+ "loss": 0.0217,
+ "step": 23136
+ },
+ {
+ "epoch": 63.043596730245234,
+ "grad_norm": 2.196381092071533,
+ "learning_rate": 6.347802464656605e-06,
+ "loss": 0.1596,
+ "step": 23137
+ },
+ {
+ "epoch": 63.04632152588556,
+ "grad_norm": 1.0886234045028687,
+ "learning_rate": 6.3469809457950406e-06,
+ "loss": 0.0147,
+ "step": 23138
+ },
+ {
+ "epoch": 63.049046321525886,
+ "grad_norm": 1.7888543605804443,
+ "learning_rate": 6.346159455383069e-06,
+ "loss": 0.0263,
+ "step": 23139
+ },
+ {
+ "epoch": 63.05177111716621,
+ "grad_norm": 1.331861138343811,
+ "learning_rate": 6.3453379934270945e-06,
+ "loss": 0.0231,
+ "step": 23140
+ },
+ {
+ "epoch": 63.05449591280654,
+ "grad_norm": 1.4528077840805054,
+ "learning_rate": 6.3445165599335044e-06,
+ "loss": 0.0195,
+ "step": 23141
+ },
+ {
+ "epoch": 63.05722070844686,
+ "grad_norm": 1.3079653978347778,
+ "learning_rate": 6.343695154908707e-06,
+ "loss": 0.0198,
+ "step": 23142
+ },
+ {
+ "epoch": 63.059945504087196,
+ "grad_norm": 4.988930702209473,
+ "learning_rate": 6.34287377835909e-06,
+ "loss": 0.0229,
+ "step": 23143
+ },
+ {
+ "epoch": 63.06267029972752,
+ "grad_norm": 2.993654489517212,
+ "learning_rate": 6.342052430291057e-06,
+ "loss": 0.0205,
+ "step": 23144
+ },
+ {
+ "epoch": 63.06539509536785,
+ "grad_norm": 1.2258070707321167,
+ "learning_rate": 6.341231110710999e-06,
+ "loss": 0.0134,
+ "step": 23145
+ },
+ {
+ "epoch": 63.06811989100817,
+ "grad_norm": 1.666306495666504,
+ "learning_rate": 6.340409819625319e-06,
+ "loss": 0.0293,
+ "step": 23146
+ },
+ {
+ "epoch": 63.0708446866485,
+ "grad_norm": 1.9203699827194214,
+ "learning_rate": 6.339588557040406e-06,
+ "loss": 0.0288,
+ "step": 23147
+ },
+ {
+ "epoch": 63.073569482288825,
+ "grad_norm": 1.6634159088134766,
+ "learning_rate": 6.338767322962662e-06,
+ "loss": 0.0277,
+ "step": 23148
+ },
+ {
+ "epoch": 63.07629427792916,
+ "grad_norm": 1.393005609512329,
+ "learning_rate": 6.337946117398479e-06,
+ "loss": 0.0652,
+ "step": 23149
+ },
+ {
+ "epoch": 63.079019073569484,
+ "grad_norm": 3.1553566455841064,
+ "learning_rate": 6.3371249403542556e-06,
+ "loss": 0.0268,
+ "step": 23150
+ },
+ {
+ "epoch": 63.08174386920981,
+ "grad_norm": 1.7141743898391724,
+ "learning_rate": 6.336303791836379e-06,
+ "loss": 0.0181,
+ "step": 23151
+ },
+ {
+ "epoch": 63.084468664850135,
+ "grad_norm": 2.8339309692382812,
+ "learning_rate": 6.335482671851256e-06,
+ "loss": 0.0407,
+ "step": 23152
+ },
+ {
+ "epoch": 63.08719346049046,
+ "grad_norm": 1.4237128496170044,
+ "learning_rate": 6.334661580405274e-06,
+ "loss": 0.0937,
+ "step": 23153
+ },
+ {
+ "epoch": 63.08991825613079,
+ "grad_norm": 1.9651434421539307,
+ "learning_rate": 6.333840517504832e-06,
+ "loss": 0.0286,
+ "step": 23154
+ },
+ {
+ "epoch": 63.09264305177112,
+ "grad_norm": 2.1936163902282715,
+ "learning_rate": 6.333019483156322e-06,
+ "loss": 0.0281,
+ "step": 23155
+ },
+ {
+ "epoch": 63.095367847411445,
+ "grad_norm": 1.8476998805999756,
+ "learning_rate": 6.332198477366138e-06,
+ "loss": 0.0422,
+ "step": 23156
+ },
+ {
+ "epoch": 63.09809264305177,
+ "grad_norm": 1.819933533668518,
+ "learning_rate": 6.331377500140672e-06,
+ "loss": 0.0421,
+ "step": 23157
+ },
+ {
+ "epoch": 63.1008174386921,
+ "grad_norm": 1.4583276510238647,
+ "learning_rate": 6.330556551486324e-06,
+ "loss": 0.0184,
+ "step": 23158
+ },
+ {
+ "epoch": 63.10354223433242,
+ "grad_norm": 2.3580539226531982,
+ "learning_rate": 6.329735631409479e-06,
+ "loss": 0.0336,
+ "step": 23159
+ },
+ {
+ "epoch": 63.10626702997275,
+ "grad_norm": 2.1278433799743652,
+ "learning_rate": 6.328914739916539e-06,
+ "loss": 0.0591,
+ "step": 23160
+ },
+ {
+ "epoch": 63.10899182561308,
+ "grad_norm": 4.101580619812012,
+ "learning_rate": 6.328093877013888e-06,
+ "loss": 0.0191,
+ "step": 23161
+ },
+ {
+ "epoch": 63.11171662125341,
+ "grad_norm": 1.7342156171798706,
+ "learning_rate": 6.327273042707929e-06,
+ "loss": 0.0367,
+ "step": 23162
+ },
+ {
+ "epoch": 63.11444141689373,
+ "grad_norm": 2.509524345397949,
+ "learning_rate": 6.3264522370050455e-06,
+ "loss": 0.051,
+ "step": 23163
+ },
+ {
+ "epoch": 63.11716621253406,
+ "grad_norm": 2.3433525562286377,
+ "learning_rate": 6.325631459911635e-06,
+ "loss": 0.1567,
+ "step": 23164
+ },
+ {
+ "epoch": 63.119891008174385,
+ "grad_norm": 2.314093828201294,
+ "learning_rate": 6.324810711434089e-06,
+ "loss": 0.0282,
+ "step": 23165
+ },
+ {
+ "epoch": 63.12261580381471,
+ "grad_norm": 1.5323421955108643,
+ "learning_rate": 6.3239899915787985e-06,
+ "loss": 0.0681,
+ "step": 23166
+ },
+ {
+ "epoch": 63.12534059945504,
+ "grad_norm": 2.4682233333587646,
+ "learning_rate": 6.323169300352155e-06,
+ "loss": 0.0356,
+ "step": 23167
+ },
+ {
+ "epoch": 63.12806539509537,
+ "grad_norm": 2.6241774559020996,
+ "learning_rate": 6.322348637760547e-06,
+ "loss": 0.03,
+ "step": 23168
+ },
+ {
+ "epoch": 63.130790190735695,
+ "grad_norm": 2.2604780197143555,
+ "learning_rate": 6.321528003810375e-06,
+ "loss": 0.0256,
+ "step": 23169
+ },
+ {
+ "epoch": 63.13351498637602,
+ "grad_norm": 1.4767556190490723,
+ "learning_rate": 6.320707398508021e-06,
+ "loss": 0.1189,
+ "step": 23170
+ },
+ {
+ "epoch": 63.13623978201635,
+ "grad_norm": 1.2959991693496704,
+ "learning_rate": 6.3198868218598805e-06,
+ "loss": 0.0147,
+ "step": 23171
+ },
+ {
+ "epoch": 63.13896457765667,
+ "grad_norm": 1.6116245985031128,
+ "learning_rate": 6.319066273872345e-06,
+ "loss": 0.0204,
+ "step": 23172
+ },
+ {
+ "epoch": 63.141689373297005,
+ "grad_norm": 1.5816155672073364,
+ "learning_rate": 6.318245754551802e-06,
+ "loss": 0.0954,
+ "step": 23173
+ },
+ {
+ "epoch": 63.14441416893733,
+ "grad_norm": 2.4078595638275146,
+ "learning_rate": 6.317425263904643e-06,
+ "loss": 0.1771,
+ "step": 23174
+ },
+ {
+ "epoch": 63.14713896457766,
+ "grad_norm": 1.791966438293457,
+ "learning_rate": 6.316604801937256e-06,
+ "loss": 0.0437,
+ "step": 23175
+ },
+ {
+ "epoch": 63.14986376021798,
+ "grad_norm": 2.1774191856384277,
+ "learning_rate": 6.315784368656034e-06,
+ "loss": 0.0446,
+ "step": 23176
+ },
+ {
+ "epoch": 63.15258855585831,
+ "grad_norm": 3.7666475772857666,
+ "learning_rate": 6.314963964067362e-06,
+ "loss": 0.0449,
+ "step": 23177
+ },
+ {
+ "epoch": 63.155313351498634,
+ "grad_norm": 1.5087199211120605,
+ "learning_rate": 6.314143588177636e-06,
+ "loss": 0.0174,
+ "step": 23178
+ },
+ {
+ "epoch": 63.15803814713897,
+ "grad_norm": 2.2203760147094727,
+ "learning_rate": 6.313323240993237e-06,
+ "loss": 0.0774,
+ "step": 23179
+ },
+ {
+ "epoch": 63.16076294277929,
+ "grad_norm": 1.5406326055526733,
+ "learning_rate": 6.312502922520562e-06,
+ "loss": 0.016,
+ "step": 23180
+ },
+ {
+ "epoch": 63.16348773841962,
+ "grad_norm": 1.4561660289764404,
+ "learning_rate": 6.311682632765992e-06,
+ "loss": 0.0275,
+ "step": 23181
+ },
+ {
+ "epoch": 63.166212534059945,
+ "grad_norm": 1.495814323425293,
+ "learning_rate": 6.310862371735922e-06,
+ "loss": 0.0712,
+ "step": 23182
+ },
+ {
+ "epoch": 63.16893732970027,
+ "grad_norm": 2.0516555309295654,
+ "learning_rate": 6.310042139436734e-06,
+ "loss": 0.0365,
+ "step": 23183
+ },
+ {
+ "epoch": 63.171662125340596,
+ "grad_norm": 1.1874924898147583,
+ "learning_rate": 6.309221935874822e-06,
+ "loss": 0.0951,
+ "step": 23184
+ },
+ {
+ "epoch": 63.17438692098093,
+ "grad_norm": 2.4307737350463867,
+ "learning_rate": 6.308401761056566e-06,
+ "loss": 0.0753,
+ "step": 23185
+ },
+ {
+ "epoch": 63.177111716621255,
+ "grad_norm": 1.3652799129486084,
+ "learning_rate": 6.30758161498836e-06,
+ "loss": 0.016,
+ "step": 23186
+ },
+ {
+ "epoch": 63.17983651226158,
+ "grad_norm": 1.0595239400863647,
+ "learning_rate": 6.306761497676594e-06,
+ "loss": 0.013,
+ "step": 23187
+ },
+ {
+ "epoch": 63.182561307901906,
+ "grad_norm": 2.291311264038086,
+ "learning_rate": 6.305941409127646e-06,
+ "loss": 0.0829,
+ "step": 23188
+ },
+ {
+ "epoch": 63.18528610354223,
+ "grad_norm": 1.5480984449386597,
+ "learning_rate": 6.305121349347911e-06,
+ "loss": 0.0426,
+ "step": 23189
+ },
+ {
+ "epoch": 63.18801089918256,
+ "grad_norm": 1.147931456565857,
+ "learning_rate": 6.304301318343768e-06,
+ "loss": 0.0255,
+ "step": 23190
+ },
+ {
+ "epoch": 63.19073569482289,
+ "grad_norm": 1.1886659860610962,
+ "learning_rate": 6.303481316121611e-06,
+ "loss": 0.0121,
+ "step": 23191
+ },
+ {
+ "epoch": 63.19346049046322,
+ "grad_norm": 1.5568816661834717,
+ "learning_rate": 6.302661342687822e-06,
+ "loss": 0.0944,
+ "step": 23192
+ },
+ {
+ "epoch": 63.19618528610354,
+ "grad_norm": 1.6158080101013184,
+ "learning_rate": 6.3018413980487865e-06,
+ "loss": 0.0158,
+ "step": 23193
+ },
+ {
+ "epoch": 63.19891008174387,
+ "grad_norm": 1.592377781867981,
+ "learning_rate": 6.301021482210889e-06,
+ "loss": 0.0945,
+ "step": 23194
+ },
+ {
+ "epoch": 63.201634877384194,
+ "grad_norm": 1.700764775276184,
+ "learning_rate": 6.30020159518052e-06,
+ "loss": 0.0383,
+ "step": 23195
+ },
+ {
+ "epoch": 63.20435967302452,
+ "grad_norm": 1.936616063117981,
+ "learning_rate": 6.2993817369640616e-06,
+ "loss": 0.028,
+ "step": 23196
+ },
+ {
+ "epoch": 63.20708446866485,
+ "grad_norm": 2.456009864807129,
+ "learning_rate": 6.2985619075678985e-06,
+ "loss": 0.0773,
+ "step": 23197
+ },
+ {
+ "epoch": 63.20980926430518,
+ "grad_norm": 1.7474576234817505,
+ "learning_rate": 6.297742106998417e-06,
+ "loss": 0.023,
+ "step": 23198
+ },
+ {
+ "epoch": 63.212534059945504,
+ "grad_norm": 1.4822169542312622,
+ "learning_rate": 6.296922335262e-06,
+ "loss": 0.021,
+ "step": 23199
+ },
+ {
+ "epoch": 63.21525885558583,
+ "grad_norm": 5.753752708435059,
+ "learning_rate": 6.296102592365035e-06,
+ "loss": 0.0357,
+ "step": 23200
+ },
+ {
+ "epoch": 63.217983651226156,
+ "grad_norm": 1.8350698947906494,
+ "learning_rate": 6.2952828783139e-06,
+ "loss": 0.031,
+ "step": 23201
+ },
+ {
+ "epoch": 63.22070844686648,
+ "grad_norm": 1.6709849834442139,
+ "learning_rate": 6.294463193114982e-06,
+ "loss": 0.0465,
+ "step": 23202
+ },
+ {
+ "epoch": 63.223433242506815,
+ "grad_norm": 2.0836150646209717,
+ "learning_rate": 6.293643536774668e-06,
+ "loss": 0.0443,
+ "step": 23203
+ },
+ {
+ "epoch": 63.22615803814714,
+ "grad_norm": 2.1634559631347656,
+ "learning_rate": 6.292823909299338e-06,
+ "loss": 0.0407,
+ "step": 23204
+ },
+ {
+ "epoch": 63.228882833787466,
+ "grad_norm": 2.2786707878112793,
+ "learning_rate": 6.2920043106953774e-06,
+ "loss": 0.0219,
+ "step": 23205
+ },
+ {
+ "epoch": 63.23160762942779,
+ "grad_norm": 1.585465908050537,
+ "learning_rate": 6.291184740969168e-06,
+ "loss": 0.0133,
+ "step": 23206
+ },
+ {
+ "epoch": 63.23433242506812,
+ "grad_norm": 3.7051796913146973,
+ "learning_rate": 6.290365200127091e-06,
+ "loss": 0.0231,
+ "step": 23207
+ },
+ {
+ "epoch": 63.237057220708444,
+ "grad_norm": 1.354328989982605,
+ "learning_rate": 6.289545688175531e-06,
+ "loss": 0.1386,
+ "step": 23208
+ },
+ {
+ "epoch": 63.23978201634878,
+ "grad_norm": 2.1179893016815186,
+ "learning_rate": 6.2887262051208704e-06,
+ "loss": 0.0644,
+ "step": 23209
+ },
+ {
+ "epoch": 63.2425068119891,
+ "grad_norm": 2.321617603302002,
+ "learning_rate": 6.28790675096949e-06,
+ "loss": 0.0254,
+ "step": 23210
+ },
+ {
+ "epoch": 63.24523160762943,
+ "grad_norm": 1.912070870399475,
+ "learning_rate": 6.287087325727774e-06,
+ "loss": 0.0347,
+ "step": 23211
+ },
+ {
+ "epoch": 63.247956403269754,
+ "grad_norm": 1.8036057949066162,
+ "learning_rate": 6.286267929402099e-06,
+ "loss": 0.0202,
+ "step": 23212
+ },
+ {
+ "epoch": 63.25068119891008,
+ "grad_norm": 2.1411845684051514,
+ "learning_rate": 6.285448561998853e-06,
+ "loss": 0.0444,
+ "step": 23213
+ },
+ {
+ "epoch": 63.253405994550405,
+ "grad_norm": 2.4142658710479736,
+ "learning_rate": 6.284629223524412e-06,
+ "loss": 0.0284,
+ "step": 23214
+ },
+ {
+ "epoch": 63.25613079019074,
+ "grad_norm": 2.8085391521453857,
+ "learning_rate": 6.283809913985162e-06,
+ "loss": 0.1105,
+ "step": 23215
+ },
+ {
+ "epoch": 63.258855585831064,
+ "grad_norm": 2.179999828338623,
+ "learning_rate": 6.282990633387477e-06,
+ "loss": 0.0325,
+ "step": 23216
+ },
+ {
+ "epoch": 63.26158038147139,
+ "grad_norm": 1.7540485858917236,
+ "learning_rate": 6.282171381737742e-06,
+ "loss": 0.0532,
+ "step": 23217
+ },
+ {
+ "epoch": 63.264305177111716,
+ "grad_norm": 2.4900286197662354,
+ "learning_rate": 6.281352159042338e-06,
+ "loss": 0.0182,
+ "step": 23218
+ },
+ {
+ "epoch": 63.26702997275204,
+ "grad_norm": 1.6385259628295898,
+ "learning_rate": 6.280532965307643e-06,
+ "loss": 0.0648,
+ "step": 23219
+ },
+ {
+ "epoch": 63.26975476839237,
+ "grad_norm": 1.8548864126205444,
+ "learning_rate": 6.279713800540035e-06,
+ "loss": 0.0439,
+ "step": 23220
+ },
+ {
+ "epoch": 63.2724795640327,
+ "grad_norm": 2.013521432876587,
+ "learning_rate": 6.2788946647459e-06,
+ "loss": 0.0452,
+ "step": 23221
+ },
+ {
+ "epoch": 63.275204359673026,
+ "grad_norm": 1.9578286409378052,
+ "learning_rate": 6.278075557931611e-06,
+ "loss": 0.0207,
+ "step": 23222
+ },
+ {
+ "epoch": 63.27792915531335,
+ "grad_norm": 54.84714126586914,
+ "learning_rate": 6.277256480103553e-06,
+ "loss": 0.0509,
+ "step": 23223
+ },
+ {
+ "epoch": 63.28065395095368,
+ "grad_norm": 1.8086299896240234,
+ "learning_rate": 6.276437431268098e-06,
+ "loss": 0.0214,
+ "step": 23224
+ },
+ {
+ "epoch": 63.283378746594,
+ "grad_norm": 1.9585689306259155,
+ "learning_rate": 6.275618411431633e-06,
+ "loss": 0.073,
+ "step": 23225
+ },
+ {
+ "epoch": 63.28610354223433,
+ "grad_norm": 1.8249289989471436,
+ "learning_rate": 6.274799420600528e-06,
+ "loss": 0.019,
+ "step": 23226
+ },
+ {
+ "epoch": 63.28882833787466,
+ "grad_norm": 3.164020538330078,
+ "learning_rate": 6.273980458781168e-06,
+ "loss": 0.0247,
+ "step": 23227
+ },
+ {
+ "epoch": 63.29155313351499,
+ "grad_norm": 1.7721182107925415,
+ "learning_rate": 6.273161525979926e-06,
+ "loss": 0.0175,
+ "step": 23228
+ },
+ {
+ "epoch": 63.294277929155314,
+ "grad_norm": 1.4118907451629639,
+ "learning_rate": 6.2723426222031845e-06,
+ "loss": 0.0324,
+ "step": 23229
+ },
+ {
+ "epoch": 63.29700272479564,
+ "grad_norm": 1.4178656339645386,
+ "learning_rate": 6.271523747457316e-06,
+ "loss": 0.0182,
+ "step": 23230
+ },
+ {
+ "epoch": 63.299727520435965,
+ "grad_norm": 1.3449374437332153,
+ "learning_rate": 6.270704901748703e-06,
+ "loss": 0.1616,
+ "step": 23231
+ },
+ {
+ "epoch": 63.30245231607629,
+ "grad_norm": 2.370455026626587,
+ "learning_rate": 6.269886085083718e-06,
+ "loss": 0.0301,
+ "step": 23232
+ },
+ {
+ "epoch": 63.305177111716624,
+ "grad_norm": 1.351995587348938,
+ "learning_rate": 6.269067297468743e-06,
+ "loss": 0.0197,
+ "step": 23233
+ },
+ {
+ "epoch": 63.30790190735695,
+ "grad_norm": 1.7846704721450806,
+ "learning_rate": 6.268248538910149e-06,
+ "loss": 0.0932,
+ "step": 23234
+ },
+ {
+ "epoch": 63.310626702997276,
+ "grad_norm": 1.7647783756256104,
+ "learning_rate": 6.267429809414316e-06,
+ "loss": 0.0221,
+ "step": 23235
+ },
+ {
+ "epoch": 63.3133514986376,
+ "grad_norm": 1.78054940700531,
+ "learning_rate": 6.266611108987619e-06,
+ "loss": 0.0246,
+ "step": 23236
+ },
+ {
+ "epoch": 63.31607629427793,
+ "grad_norm": 2.2716434001922607,
+ "learning_rate": 6.265792437636433e-06,
+ "loss": 0.086,
+ "step": 23237
+ },
+ {
+ "epoch": 63.31880108991825,
+ "grad_norm": 3.252768039703369,
+ "learning_rate": 6.264973795367136e-06,
+ "loss": 0.1894,
+ "step": 23238
+ },
+ {
+ "epoch": 63.321525885558586,
+ "grad_norm": 1.5825839042663574,
+ "learning_rate": 6.264155182186105e-06,
+ "loss": 0.0198,
+ "step": 23239
+ },
+ {
+ "epoch": 63.32425068119891,
+ "grad_norm": 1.9841119050979614,
+ "learning_rate": 6.263336598099712e-06,
+ "loss": 0.0156,
+ "step": 23240
+ },
+ {
+ "epoch": 63.32697547683924,
+ "grad_norm": 3.5777857303619385,
+ "learning_rate": 6.262518043114336e-06,
+ "loss": 0.0536,
+ "step": 23241
+ },
+ {
+ "epoch": 63.32970027247956,
+ "grad_norm": 3.379183292388916,
+ "learning_rate": 6.261699517236346e-06,
+ "loss": 0.0282,
+ "step": 23242
+ },
+ {
+ "epoch": 63.33242506811989,
+ "grad_norm": 4.431500434875488,
+ "learning_rate": 6.260881020472122e-06,
+ "loss": 0.123,
+ "step": 23243
+ },
+ {
+ "epoch": 63.335149863760215,
+ "grad_norm": 1.878844141960144,
+ "learning_rate": 6.260062552828035e-06,
+ "loss": 0.0282,
+ "step": 23244
+ },
+ {
+ "epoch": 63.33787465940055,
+ "grad_norm": 1.442433476448059,
+ "learning_rate": 6.259244114310461e-06,
+ "loss": 0.0141,
+ "step": 23245
+ },
+ {
+ "epoch": 63.34059945504087,
+ "grad_norm": 1.0960161685943604,
+ "learning_rate": 6.2584257049257724e-06,
+ "loss": 0.0135,
+ "step": 23246
+ },
+ {
+ "epoch": 63.3433242506812,
+ "grad_norm": 3.4297757148742676,
+ "learning_rate": 6.257607324680346e-06,
+ "loss": 0.0418,
+ "step": 23247
+ },
+ {
+ "epoch": 63.346049046321525,
+ "grad_norm": 1.4526147842407227,
+ "learning_rate": 6.25678897358055e-06,
+ "loss": 0.0177,
+ "step": 23248
+ },
+ {
+ "epoch": 63.34877384196185,
+ "grad_norm": 2.4080393314361572,
+ "learning_rate": 6.255970651632764e-06,
+ "loss": 0.1079,
+ "step": 23249
+ },
+ {
+ "epoch": 63.35149863760218,
+ "grad_norm": 1.8713998794555664,
+ "learning_rate": 6.255152358843357e-06,
+ "loss": 0.033,
+ "step": 23250
+ },
+ {
+ "epoch": 63.35422343324251,
+ "grad_norm": 2.746436834335327,
+ "learning_rate": 6.254334095218704e-06,
+ "loss": 0.1126,
+ "step": 23251
+ },
+ {
+ "epoch": 63.356948228882835,
+ "grad_norm": 2.1590638160705566,
+ "learning_rate": 6.253515860765175e-06,
+ "loss": 0.1059,
+ "step": 23252
+ },
+ {
+ "epoch": 63.35967302452316,
+ "grad_norm": 1.985795497894287,
+ "learning_rate": 6.252697655489145e-06,
+ "loss": 0.0939,
+ "step": 23253
+ },
+ {
+ "epoch": 63.36239782016349,
+ "grad_norm": 1.9154716730117798,
+ "learning_rate": 6.2518794793969815e-06,
+ "loss": 0.0224,
+ "step": 23254
+ },
+ {
+ "epoch": 63.36512261580381,
+ "grad_norm": 1.7584341764450073,
+ "learning_rate": 6.251061332495062e-06,
+ "loss": 0.0563,
+ "step": 23255
+ },
+ {
+ "epoch": 63.36784741144414,
+ "grad_norm": 1.4316596984863281,
+ "learning_rate": 6.2502432147897584e-06,
+ "loss": 0.0325,
+ "step": 23256
+ },
+ {
+ "epoch": 63.37057220708447,
+ "grad_norm": 3.1759910583496094,
+ "learning_rate": 6.249425126287439e-06,
+ "loss": 0.0234,
+ "step": 23257
+ },
+ {
+ "epoch": 63.3732970027248,
+ "grad_norm": 2.0711607933044434,
+ "learning_rate": 6.248607066994473e-06,
+ "loss": 0.0417,
+ "step": 23258
+ },
+ {
+ "epoch": 63.37602179836512,
+ "grad_norm": 1.723002314567566,
+ "learning_rate": 6.247789036917238e-06,
+ "loss": 0.0295,
+ "step": 23259
+ },
+ {
+ "epoch": 63.37874659400545,
+ "grad_norm": 2.504901647567749,
+ "learning_rate": 6.246971036062099e-06,
+ "loss": 0.0357,
+ "step": 23260
+ },
+ {
+ "epoch": 63.381471389645775,
+ "grad_norm": 2.0430424213409424,
+ "learning_rate": 6.246153064435432e-06,
+ "loss": 0.0226,
+ "step": 23261
+ },
+ {
+ "epoch": 63.3841961852861,
+ "grad_norm": 1.6244263648986816,
+ "learning_rate": 6.245335122043601e-06,
+ "loss": 0.0157,
+ "step": 23262
+ },
+ {
+ "epoch": 63.38692098092643,
+ "grad_norm": 1.9477545022964478,
+ "learning_rate": 6.244517208892982e-06,
+ "loss": 0.0186,
+ "step": 23263
+ },
+ {
+ "epoch": 63.38964577656676,
+ "grad_norm": 2.104729175567627,
+ "learning_rate": 6.243699324989941e-06,
+ "loss": 0.1243,
+ "step": 23264
+ },
+ {
+ "epoch": 63.392370572207085,
+ "grad_norm": 1.8368297815322876,
+ "learning_rate": 6.242881470340848e-06,
+ "loss": 0.1609,
+ "step": 23265
+ },
+ {
+ "epoch": 63.39509536784741,
+ "grad_norm": 2.8845903873443604,
+ "learning_rate": 6.242063644952072e-06,
+ "loss": 0.1447,
+ "step": 23266
+ },
+ {
+ "epoch": 63.39782016348774,
+ "grad_norm": 1.9899946451187134,
+ "learning_rate": 6.241245848829986e-06,
+ "loss": 0.0938,
+ "step": 23267
+ },
+ {
+ "epoch": 63.40054495912806,
+ "grad_norm": 2.2183785438537598,
+ "learning_rate": 6.240428081980955e-06,
+ "loss": 0.0442,
+ "step": 23268
+ },
+ {
+ "epoch": 63.403269754768395,
+ "grad_norm": 2.8757941722869873,
+ "learning_rate": 6.239610344411351e-06,
+ "loss": 0.0862,
+ "step": 23269
+ },
+ {
+ "epoch": 63.40599455040872,
+ "grad_norm": 1.7816137075424194,
+ "learning_rate": 6.238792636127537e-06,
+ "loss": 0.0185,
+ "step": 23270
+ },
+ {
+ "epoch": 63.40871934604905,
+ "grad_norm": 2.0671703815460205,
+ "learning_rate": 6.2379749571358835e-06,
+ "loss": 0.0354,
+ "step": 23271
+ },
+ {
+ "epoch": 63.41144414168937,
+ "grad_norm": 1.6595348119735718,
+ "learning_rate": 6.237157307442765e-06,
+ "loss": 0.0194,
+ "step": 23272
+ },
+ {
+ "epoch": 63.4141689373297,
+ "grad_norm": 3.359687328338623,
+ "learning_rate": 6.2363396870545425e-06,
+ "loss": 0.0395,
+ "step": 23273
+ },
+ {
+ "epoch": 63.416893732970024,
+ "grad_norm": 2.2326042652130127,
+ "learning_rate": 6.235522095977586e-06,
+ "loss": 0.0195,
+ "step": 23274
+ },
+ {
+ "epoch": 63.41961852861036,
+ "grad_norm": 2.0988943576812744,
+ "learning_rate": 6.234704534218261e-06,
+ "loss": 0.0246,
+ "step": 23275
+ },
+ {
+ "epoch": 63.42234332425068,
+ "grad_norm": 2.7817211151123047,
+ "learning_rate": 6.233887001782937e-06,
+ "loss": 0.0411,
+ "step": 23276
+ },
+ {
+ "epoch": 63.42506811989101,
+ "grad_norm": 1.9098435640335083,
+ "learning_rate": 6.233069498677979e-06,
+ "loss": 0.0525,
+ "step": 23277
+ },
+ {
+ "epoch": 63.427792915531334,
+ "grad_norm": 2.417476177215576,
+ "learning_rate": 6.232252024909755e-06,
+ "loss": 0.0758,
+ "step": 23278
+ },
+ {
+ "epoch": 63.43051771117166,
+ "grad_norm": 1.4829936027526855,
+ "learning_rate": 6.231434580484631e-06,
+ "loss": 0.0385,
+ "step": 23279
+ },
+ {
+ "epoch": 63.433242506811986,
+ "grad_norm": 2.724877119064331,
+ "learning_rate": 6.230617165408974e-06,
+ "loss": 0.068,
+ "step": 23280
+ },
+ {
+ "epoch": 63.43596730245232,
+ "grad_norm": 2.026930809020996,
+ "learning_rate": 6.229799779689147e-06,
+ "loss": 0.1738,
+ "step": 23281
+ },
+ {
+ "epoch": 63.438692098092645,
+ "grad_norm": 1.6638484001159668,
+ "learning_rate": 6.228982423331519e-06,
+ "loss": 0.0188,
+ "step": 23282
+ },
+ {
+ "epoch": 63.44141689373297,
+ "grad_norm": 5.155297756195068,
+ "learning_rate": 6.228165096342456e-06,
+ "loss": 0.1328,
+ "step": 23283
+ },
+ {
+ "epoch": 63.444141689373296,
+ "grad_norm": 1.5937691926956177,
+ "learning_rate": 6.227347798728319e-06,
+ "loss": 0.0297,
+ "step": 23284
+ },
+ {
+ "epoch": 63.44686648501362,
+ "grad_norm": 1.8147104978561401,
+ "learning_rate": 6.2265305304954796e-06,
+ "loss": 0.0279,
+ "step": 23285
+ },
+ {
+ "epoch": 63.44959128065395,
+ "grad_norm": 3.751678943634033,
+ "learning_rate": 6.225713291650295e-06,
+ "loss": 0.0351,
+ "step": 23286
+ },
+ {
+ "epoch": 63.45231607629428,
+ "grad_norm": 1.9648996591567993,
+ "learning_rate": 6.224896082199137e-06,
+ "loss": 0.0261,
+ "step": 23287
+ },
+ {
+ "epoch": 63.45504087193461,
+ "grad_norm": 2.231144428253174,
+ "learning_rate": 6.224078902148365e-06,
+ "loss": 0.0779,
+ "step": 23288
+ },
+ {
+ "epoch": 63.45776566757493,
+ "grad_norm": 2.353212594985962,
+ "learning_rate": 6.223261751504343e-06,
+ "loss": 0.036,
+ "step": 23289
+ },
+ {
+ "epoch": 63.46049046321526,
+ "grad_norm": 2.69644832611084,
+ "learning_rate": 6.222444630273439e-06,
+ "loss": 0.0428,
+ "step": 23290
+ },
+ {
+ "epoch": 63.463215258855584,
+ "grad_norm": 2.2398929595947266,
+ "learning_rate": 6.221627538462015e-06,
+ "loss": 0.0895,
+ "step": 23291
+ },
+ {
+ "epoch": 63.46594005449591,
+ "grad_norm": 1.994147539138794,
+ "learning_rate": 6.2208104760764355e-06,
+ "loss": 0.0214,
+ "step": 23292
+ },
+ {
+ "epoch": 63.46866485013624,
+ "grad_norm": 2.708786725997925,
+ "learning_rate": 6.21999344312306e-06,
+ "loss": 0.1762,
+ "step": 23293
+ },
+ {
+ "epoch": 63.47138964577657,
+ "grad_norm": 2.8405601978302,
+ "learning_rate": 6.219176439608256e-06,
+ "loss": 0.0391,
+ "step": 23294
+ },
+ {
+ "epoch": 63.474114441416894,
+ "grad_norm": 2.413914918899536,
+ "learning_rate": 6.218359465538383e-06,
+ "loss": 0.1495,
+ "step": 23295
+ },
+ {
+ "epoch": 63.47683923705722,
+ "grad_norm": 2.054935932159424,
+ "learning_rate": 6.217542520919806e-06,
+ "loss": 0.0788,
+ "step": 23296
+ },
+ {
+ "epoch": 63.479564032697546,
+ "grad_norm": 2.3776495456695557,
+ "learning_rate": 6.2167256057588845e-06,
+ "loss": 0.1343,
+ "step": 23297
+ },
+ {
+ "epoch": 63.48228882833787,
+ "grad_norm": 2.695943593978882,
+ "learning_rate": 6.215908720061983e-06,
+ "loss": 0.0668,
+ "step": 23298
+ },
+ {
+ "epoch": 63.485013623978205,
+ "grad_norm": 1.8043245077133179,
+ "learning_rate": 6.2150918638354614e-06,
+ "loss": 0.0265,
+ "step": 23299
+ },
+ {
+ "epoch": 63.48773841961853,
+ "grad_norm": 1.730655312538147,
+ "learning_rate": 6.214275037085685e-06,
+ "loss": 0.0227,
+ "step": 23300
+ },
+ {
+ "epoch": 63.490463215258856,
+ "grad_norm": 1.8202217817306519,
+ "learning_rate": 6.21345823981901e-06,
+ "loss": 0.1056,
+ "step": 23301
+ },
+ {
+ "epoch": 63.49318801089918,
+ "grad_norm": 3.9278371334075928,
+ "learning_rate": 6.212641472041803e-06,
+ "loss": 0.043,
+ "step": 23302
+ },
+ {
+ "epoch": 63.49591280653951,
+ "grad_norm": 2.880432367324829,
+ "learning_rate": 6.211824733760421e-06,
+ "loss": 0.043,
+ "step": 23303
+ },
+ {
+ "epoch": 63.49863760217983,
+ "grad_norm": 1.5632514953613281,
+ "learning_rate": 6.211008024981224e-06,
+ "loss": 0.0154,
+ "step": 23304
+ },
+ {
+ "epoch": 63.50136239782017,
+ "grad_norm": 1.7499524354934692,
+ "learning_rate": 6.210191345710579e-06,
+ "loss": 0.0241,
+ "step": 23305
+ },
+ {
+ "epoch": 63.50408719346049,
+ "grad_norm": 1.8799846172332764,
+ "learning_rate": 6.209374695954835e-06,
+ "loss": 0.0218,
+ "step": 23306
+ },
+ {
+ "epoch": 63.50681198910082,
+ "grad_norm": 1.7941888570785522,
+ "learning_rate": 6.208558075720363e-06,
+ "loss": 0.0458,
+ "step": 23307
+ },
+ {
+ "epoch": 63.509536784741144,
+ "grad_norm": 1.9949264526367188,
+ "learning_rate": 6.20774148501352e-06,
+ "loss": 0.0234,
+ "step": 23308
+ },
+ {
+ "epoch": 63.51226158038147,
+ "grad_norm": 2.2424511909484863,
+ "learning_rate": 6.206924923840663e-06,
+ "loss": 0.0285,
+ "step": 23309
+ },
+ {
+ "epoch": 63.514986376021795,
+ "grad_norm": 2.67582368850708,
+ "learning_rate": 6.206108392208153e-06,
+ "loss": 0.0528,
+ "step": 23310
+ },
+ {
+ "epoch": 63.51771117166213,
+ "grad_norm": 2.4068644046783447,
+ "learning_rate": 6.205291890122348e-06,
+ "loss": 0.0597,
+ "step": 23311
+ },
+ {
+ "epoch": 63.520435967302454,
+ "grad_norm": 2.148583173751831,
+ "learning_rate": 6.204475417589609e-06,
+ "loss": 0.022,
+ "step": 23312
+ },
+ {
+ "epoch": 63.52316076294278,
+ "grad_norm": 1.9230669736862183,
+ "learning_rate": 6.203658974616291e-06,
+ "loss": 0.0851,
+ "step": 23313
+ },
+ {
+ "epoch": 63.525885558583106,
+ "grad_norm": 2.180738925933838,
+ "learning_rate": 6.202842561208759e-06,
+ "loss": 0.0282,
+ "step": 23314
+ },
+ {
+ "epoch": 63.52861035422343,
+ "grad_norm": 1.8112906217575073,
+ "learning_rate": 6.2020261773733624e-06,
+ "loss": 0.1398,
+ "step": 23315
+ },
+ {
+ "epoch": 63.53133514986376,
+ "grad_norm": 2.682020664215088,
+ "learning_rate": 6.201209823116466e-06,
+ "loss": 0.0476,
+ "step": 23316
+ },
+ {
+ "epoch": 63.53405994550409,
+ "grad_norm": 3.2618467807769775,
+ "learning_rate": 6.200393498444424e-06,
+ "loss": 0.0365,
+ "step": 23317
+ },
+ {
+ "epoch": 63.536784741144416,
+ "grad_norm": 1.7816630601882935,
+ "learning_rate": 6.199577203363597e-06,
+ "loss": 0.046,
+ "step": 23318
+ },
+ {
+ "epoch": 63.53950953678474,
+ "grad_norm": 2.7830941677093506,
+ "learning_rate": 6.198760937880339e-06,
+ "loss": 0.0401,
+ "step": 23319
+ },
+ {
+ "epoch": 63.54223433242507,
+ "grad_norm": 1.8289686441421509,
+ "learning_rate": 6.197944702001009e-06,
+ "loss": 0.0546,
+ "step": 23320
+ },
+ {
+ "epoch": 63.54495912806539,
+ "grad_norm": 1.6870653629302979,
+ "learning_rate": 6.197128495731964e-06,
+ "loss": 0.0341,
+ "step": 23321
+ },
+ {
+ "epoch": 63.54768392370572,
+ "grad_norm": 1.7838412523269653,
+ "learning_rate": 6.196312319079558e-06,
+ "loss": 0.0305,
+ "step": 23322
+ },
+ {
+ "epoch": 63.55040871934605,
+ "grad_norm": 2.6121814250946045,
+ "learning_rate": 6.195496172050146e-06,
+ "loss": 0.0931,
+ "step": 23323
+ },
+ {
+ "epoch": 63.55313351498638,
+ "grad_norm": 1.9211134910583496,
+ "learning_rate": 6.194680054650093e-06,
+ "loss": 0.0295,
+ "step": 23324
+ },
+ {
+ "epoch": 63.555858310626704,
+ "grad_norm": 2.526691198348999,
+ "learning_rate": 6.193863966885746e-06,
+ "loss": 0.0301,
+ "step": 23325
+ },
+ {
+ "epoch": 63.55858310626703,
+ "grad_norm": 2.3478305339813232,
+ "learning_rate": 6.193047908763468e-06,
+ "loss": 0.0457,
+ "step": 23326
+ },
+ {
+ "epoch": 63.561307901907355,
+ "grad_norm": 2.467607259750366,
+ "learning_rate": 6.192231880289605e-06,
+ "loss": 0.1272,
+ "step": 23327
+ },
+ {
+ "epoch": 63.56403269754768,
+ "grad_norm": 1.8157119750976562,
+ "learning_rate": 6.191415881470523e-06,
+ "loss": 0.0396,
+ "step": 23328
+ },
+ {
+ "epoch": 63.566757493188014,
+ "grad_norm": 2.069833517074585,
+ "learning_rate": 6.190599912312567e-06,
+ "loss": 0.0275,
+ "step": 23329
+ },
+ {
+ "epoch": 63.56948228882834,
+ "grad_norm": 2.027933359146118,
+ "learning_rate": 6.189783972822101e-06,
+ "loss": 0.1269,
+ "step": 23330
+ },
+ {
+ "epoch": 63.572207084468666,
+ "grad_norm": 2.0909576416015625,
+ "learning_rate": 6.18896806300547e-06,
+ "loss": 0.0488,
+ "step": 23331
+ },
+ {
+ "epoch": 63.57493188010899,
+ "grad_norm": 2.2980000972747803,
+ "learning_rate": 6.188152182869037e-06,
+ "loss": 0.0284,
+ "step": 23332
+ },
+ {
+ "epoch": 63.57765667574932,
+ "grad_norm": 1.5634405612945557,
+ "learning_rate": 6.187336332419149e-06,
+ "loss": 0.0228,
+ "step": 23333
+ },
+ {
+ "epoch": 63.58038147138964,
+ "grad_norm": 1.9333994388580322,
+ "learning_rate": 6.186520511662167e-06,
+ "loss": 0.0237,
+ "step": 23334
+ },
+ {
+ "epoch": 63.583106267029976,
+ "grad_norm": 1.9272345304489136,
+ "learning_rate": 6.185704720604435e-06,
+ "loss": 0.017,
+ "step": 23335
+ },
+ {
+ "epoch": 63.5858310626703,
+ "grad_norm": 2.6535215377807617,
+ "learning_rate": 6.184888959252317e-06,
+ "loss": 0.0316,
+ "step": 23336
+ },
+ {
+ "epoch": 63.58855585831063,
+ "grad_norm": 1.6827208995819092,
+ "learning_rate": 6.184073227612158e-06,
+ "loss": 0.0142,
+ "step": 23337
+ },
+ {
+ "epoch": 63.59128065395095,
+ "grad_norm": 2.074445962905884,
+ "learning_rate": 6.183257525690316e-06,
+ "loss": 0.0665,
+ "step": 23338
+ },
+ {
+ "epoch": 63.59400544959128,
+ "grad_norm": 2.059246778488159,
+ "learning_rate": 6.182441853493139e-06,
+ "loss": 0.0426,
+ "step": 23339
+ },
+ {
+ "epoch": 63.596730245231605,
+ "grad_norm": 2.190584182739258,
+ "learning_rate": 6.181626211026979e-06,
+ "loss": 0.0858,
+ "step": 23340
+ },
+ {
+ "epoch": 63.59945504087194,
+ "grad_norm": 1.4991832971572876,
+ "learning_rate": 6.180810598298197e-06,
+ "loss": 0.0529,
+ "step": 23341
+ },
+ {
+ "epoch": 63.60217983651226,
+ "grad_norm": 2.0709643363952637,
+ "learning_rate": 6.179995015313136e-06,
+ "loss": 0.0247,
+ "step": 23342
+ },
+ {
+ "epoch": 63.60490463215259,
+ "grad_norm": 2.757026195526123,
+ "learning_rate": 6.1791794620781535e-06,
+ "loss": 0.0296,
+ "step": 23343
+ },
+ {
+ "epoch": 63.607629427792915,
+ "grad_norm": 2.0602152347564697,
+ "learning_rate": 6.178363938599596e-06,
+ "loss": 0.1347,
+ "step": 23344
+ },
+ {
+ "epoch": 63.61035422343324,
+ "grad_norm": 1.5427017211914062,
+ "learning_rate": 6.1775484448838184e-06,
+ "loss": 0.0227,
+ "step": 23345
+ },
+ {
+ "epoch": 63.61307901907357,
+ "grad_norm": 2.3747570514678955,
+ "learning_rate": 6.176732980937171e-06,
+ "loss": 0.017,
+ "step": 23346
+ },
+ {
+ "epoch": 63.6158038147139,
+ "grad_norm": 2.0203917026519775,
+ "learning_rate": 6.175917546766003e-06,
+ "loss": 0.027,
+ "step": 23347
+ },
+ {
+ "epoch": 63.618528610354225,
+ "grad_norm": 1.706546425819397,
+ "learning_rate": 6.175102142376668e-06,
+ "loss": 0.0407,
+ "step": 23348
+ },
+ {
+ "epoch": 63.62125340599455,
+ "grad_norm": 1.275816559791565,
+ "learning_rate": 6.1742867677755115e-06,
+ "loss": 0.0265,
+ "step": 23349
+ },
+ {
+ "epoch": 63.62397820163488,
+ "grad_norm": 2.5487072467803955,
+ "learning_rate": 6.173471422968889e-06,
+ "loss": 0.0557,
+ "step": 23350
+ },
+ {
+ "epoch": 63.6267029972752,
+ "grad_norm": 1.5275999307632446,
+ "learning_rate": 6.172656107963146e-06,
+ "loss": 0.0238,
+ "step": 23351
+ },
+ {
+ "epoch": 63.62942779291553,
+ "grad_norm": 1.9649118185043335,
+ "learning_rate": 6.171840822764636e-06,
+ "loss": 0.0316,
+ "step": 23352
+ },
+ {
+ "epoch": 63.63215258855586,
+ "grad_norm": 1.726703405380249,
+ "learning_rate": 6.1710255673797045e-06,
+ "loss": 0.051,
+ "step": 23353
+ },
+ {
+ "epoch": 63.63487738419619,
+ "grad_norm": 1.6148295402526855,
+ "learning_rate": 6.170210341814705e-06,
+ "loss": 0.0156,
+ "step": 23354
+ },
+ {
+ "epoch": 63.63760217983651,
+ "grad_norm": 1.7329256534576416,
+ "learning_rate": 6.169395146075982e-06,
+ "loss": 0.1246,
+ "step": 23355
+ },
+ {
+ "epoch": 63.64032697547684,
+ "grad_norm": 1.7117750644683838,
+ "learning_rate": 6.168579980169889e-06,
+ "loss": 0.0406,
+ "step": 23356
+ },
+ {
+ "epoch": 63.643051771117165,
+ "grad_norm": 2.297767400741577,
+ "learning_rate": 6.167764844102766e-06,
+ "loss": 0.0332,
+ "step": 23357
+ },
+ {
+ "epoch": 63.64577656675749,
+ "grad_norm": 2.001070022583008,
+ "learning_rate": 6.16694973788097e-06,
+ "loss": 0.0801,
+ "step": 23358
+ },
+ {
+ "epoch": 63.64850136239782,
+ "grad_norm": 2.310532331466675,
+ "learning_rate": 6.166134661510848e-06,
+ "loss": 0.1581,
+ "step": 23359
+ },
+ {
+ "epoch": 63.65122615803815,
+ "grad_norm": 2.275498390197754,
+ "learning_rate": 6.165319614998744e-06,
+ "loss": 0.0279,
+ "step": 23360
+ },
+ {
+ "epoch": 63.653950953678475,
+ "grad_norm": 2.0945677757263184,
+ "learning_rate": 6.164504598351009e-06,
+ "loss": 0.035,
+ "step": 23361
+ },
+ {
+ "epoch": 63.6566757493188,
+ "grad_norm": 2.153949022293091,
+ "learning_rate": 6.1636896115739864e-06,
+ "loss": 0.0478,
+ "step": 23362
+ },
+ {
+ "epoch": 63.65940054495913,
+ "grad_norm": 1.5129265785217285,
+ "learning_rate": 6.1628746546740295e-06,
+ "loss": 0.0355,
+ "step": 23363
+ },
+ {
+ "epoch": 63.66212534059945,
+ "grad_norm": 2.0653462409973145,
+ "learning_rate": 6.162059727657477e-06,
+ "loss": 0.0221,
+ "step": 23364
+ },
+ {
+ "epoch": 63.664850136239785,
+ "grad_norm": 1.1519036293029785,
+ "learning_rate": 6.161244830530684e-06,
+ "loss": 0.0174,
+ "step": 23365
+ },
+ {
+ "epoch": 63.66757493188011,
+ "grad_norm": 1.608416199684143,
+ "learning_rate": 6.1604299632999885e-06,
+ "loss": 0.0321,
+ "step": 23366
+ },
+ {
+ "epoch": 63.67029972752044,
+ "grad_norm": 2.131471872329712,
+ "learning_rate": 6.159615125971743e-06,
+ "loss": 0.0198,
+ "step": 23367
+ },
+ {
+ "epoch": 63.67302452316076,
+ "grad_norm": 2.1423773765563965,
+ "learning_rate": 6.158800318552292e-06,
+ "loss": 0.0261,
+ "step": 23368
+ },
+ {
+ "epoch": 63.67574931880109,
+ "grad_norm": 1.3174821138381958,
+ "learning_rate": 6.157985541047979e-06,
+ "loss": 0.0327,
+ "step": 23369
+ },
+ {
+ "epoch": 63.678474114441414,
+ "grad_norm": 2.1056745052337646,
+ "learning_rate": 6.157170793465153e-06,
+ "loss": 0.0231,
+ "step": 23370
+ },
+ {
+ "epoch": 63.68119891008175,
+ "grad_norm": 1.2408262491226196,
+ "learning_rate": 6.156356075810154e-06,
+ "loss": 0.0181,
+ "step": 23371
+ },
+ {
+ "epoch": 63.68392370572207,
+ "grad_norm": 1.4696738719940186,
+ "learning_rate": 6.155541388089332e-06,
+ "loss": 0.0216,
+ "step": 23372
+ },
+ {
+ "epoch": 63.6866485013624,
+ "grad_norm": 1.9276447296142578,
+ "learning_rate": 6.15472673030903e-06,
+ "loss": 0.0927,
+ "step": 23373
+ },
+ {
+ "epoch": 63.689373297002724,
+ "grad_norm": 2.1825766563415527,
+ "learning_rate": 6.1539121024755935e-06,
+ "loss": 0.0314,
+ "step": 23374
+ },
+ {
+ "epoch": 63.69209809264305,
+ "grad_norm": 2.586421489715576,
+ "learning_rate": 6.153097504595362e-06,
+ "loss": 0.085,
+ "step": 23375
+ },
+ {
+ "epoch": 63.694822888283376,
+ "grad_norm": 1.9905104637145996,
+ "learning_rate": 6.152282936674685e-06,
+ "loss": 0.0432,
+ "step": 23376
+ },
+ {
+ "epoch": 63.69754768392371,
+ "grad_norm": 2.144240617752075,
+ "learning_rate": 6.151468398719906e-06,
+ "loss": 0.0422,
+ "step": 23377
+ },
+ {
+ "epoch": 63.700272479564035,
+ "grad_norm": 2.567589044570923,
+ "learning_rate": 6.1506538907373655e-06,
+ "loss": 0.0496,
+ "step": 23378
+ },
+ {
+ "epoch": 63.70299727520436,
+ "grad_norm": 2.664142608642578,
+ "learning_rate": 6.149839412733411e-06,
+ "loss": 0.0747,
+ "step": 23379
+ },
+ {
+ "epoch": 63.705722070844686,
+ "grad_norm": 1.9527229070663452,
+ "learning_rate": 6.149024964714382e-06,
+ "loss": 0.0275,
+ "step": 23380
+ },
+ {
+ "epoch": 63.70844686648501,
+ "grad_norm": 2.0968925952911377,
+ "learning_rate": 6.148210546686624e-06,
+ "loss": 0.0648,
+ "step": 23381
+ },
+ {
+ "epoch": 63.71117166212534,
+ "grad_norm": 1.6855792999267578,
+ "learning_rate": 6.1473961586564756e-06,
+ "loss": 0.0359,
+ "step": 23382
+ },
+ {
+ "epoch": 63.71389645776567,
+ "grad_norm": 2.131962537765503,
+ "learning_rate": 6.146581800630285e-06,
+ "loss": 0.0201,
+ "step": 23383
+ },
+ {
+ "epoch": 63.716621253406,
+ "grad_norm": 2.1409990787506104,
+ "learning_rate": 6.145767472614389e-06,
+ "loss": 0.0217,
+ "step": 23384
+ },
+ {
+ "epoch": 63.71934604904632,
+ "grad_norm": 1.8234124183654785,
+ "learning_rate": 6.144953174615133e-06,
+ "loss": 0.034,
+ "step": 23385
+ },
+ {
+ "epoch": 63.72207084468665,
+ "grad_norm": 2.887147903442383,
+ "learning_rate": 6.1441389066388565e-06,
+ "loss": 0.0176,
+ "step": 23386
+ },
+ {
+ "epoch": 63.724795640326974,
+ "grad_norm": 2.44392728805542,
+ "learning_rate": 6.143324668691904e-06,
+ "loss": 0.0455,
+ "step": 23387
+ },
+ {
+ "epoch": 63.7275204359673,
+ "grad_norm": 2.050638437271118,
+ "learning_rate": 6.142510460780613e-06,
+ "loss": 0.0513,
+ "step": 23388
+ },
+ {
+ "epoch": 63.73024523160763,
+ "grad_norm": 2.4240951538085938,
+ "learning_rate": 6.141696282911328e-06,
+ "loss": 0.0755,
+ "step": 23389
+ },
+ {
+ "epoch": 63.73297002724796,
+ "grad_norm": 1.5272631645202637,
+ "learning_rate": 6.140882135090387e-06,
+ "loss": 0.0343,
+ "step": 23390
+ },
+ {
+ "epoch": 63.735694822888284,
+ "grad_norm": 2.1212990283966064,
+ "learning_rate": 6.14006801732413e-06,
+ "loss": 0.0598,
+ "step": 23391
+ },
+ {
+ "epoch": 63.73841961852861,
+ "grad_norm": 1.973994493484497,
+ "learning_rate": 6.139253929618898e-06,
+ "loss": 0.0343,
+ "step": 23392
+ },
+ {
+ "epoch": 63.741144414168936,
+ "grad_norm": 1.732708215713501,
+ "learning_rate": 6.138439871981037e-06,
+ "loss": 0.0325,
+ "step": 23393
+ },
+ {
+ "epoch": 63.74386920980926,
+ "grad_norm": 1.316652774810791,
+ "learning_rate": 6.137625844416879e-06,
+ "loss": 0.0274,
+ "step": 23394
+ },
+ {
+ "epoch": 63.746594005449595,
+ "grad_norm": 2.531451940536499,
+ "learning_rate": 6.136811846932769e-06,
+ "loss": 0.1203,
+ "step": 23395
+ },
+ {
+ "epoch": 63.74931880108992,
+ "grad_norm": 2.3501124382019043,
+ "learning_rate": 6.135997879535042e-06,
+ "loss": 0.0566,
+ "step": 23396
+ },
+ {
+ "epoch": 63.752043596730246,
+ "grad_norm": 1.5224674940109253,
+ "learning_rate": 6.135183942230041e-06,
+ "loss": 0.0364,
+ "step": 23397
+ },
+ {
+ "epoch": 63.75476839237057,
+ "grad_norm": 1.9684693813323975,
+ "learning_rate": 6.1343700350241005e-06,
+ "loss": 0.0694,
+ "step": 23398
+ },
+ {
+ "epoch": 63.7574931880109,
+ "grad_norm": 2.5461878776550293,
+ "learning_rate": 6.133556157923565e-06,
+ "loss": 0.1385,
+ "step": 23399
+ },
+ {
+ "epoch": 63.76021798365122,
+ "grad_norm": 1.9635040760040283,
+ "learning_rate": 6.132742310934767e-06,
+ "loss": 0.0599,
+ "step": 23400
+ },
+ {
+ "epoch": 63.762942779291556,
+ "grad_norm": 2.0328726768493652,
+ "learning_rate": 6.131928494064049e-06,
+ "loss": 0.0277,
+ "step": 23401
+ },
+ {
+ "epoch": 63.76566757493188,
+ "grad_norm": 1.39298415184021,
+ "learning_rate": 6.1311147073177455e-06,
+ "loss": 0.0202,
+ "step": 23402
+ },
+ {
+ "epoch": 63.76839237057221,
+ "grad_norm": 1.9625595808029175,
+ "learning_rate": 6.1303009507021995e-06,
+ "loss": 0.0264,
+ "step": 23403
+ },
+ {
+ "epoch": 63.771117166212534,
+ "grad_norm": 1.409645438194275,
+ "learning_rate": 6.129487224223741e-06,
+ "loss": 0.031,
+ "step": 23404
+ },
+ {
+ "epoch": 63.77384196185286,
+ "grad_norm": 1.4819620847702026,
+ "learning_rate": 6.128673527888715e-06,
+ "loss": 0.0986,
+ "step": 23405
+ },
+ {
+ "epoch": 63.776566757493185,
+ "grad_norm": 2.3551290035247803,
+ "learning_rate": 6.127859861703453e-06,
+ "loss": 0.0218,
+ "step": 23406
+ },
+ {
+ "epoch": 63.77929155313352,
+ "grad_norm": 3.496570110321045,
+ "learning_rate": 6.127046225674294e-06,
+ "loss": 0.0335,
+ "step": 23407
+ },
+ {
+ "epoch": 63.782016348773844,
+ "grad_norm": 2.244701623916626,
+ "learning_rate": 6.126232619807575e-06,
+ "loss": 0.0271,
+ "step": 23408
+ },
+ {
+ "epoch": 63.78474114441417,
+ "grad_norm": 2.17472505569458,
+ "learning_rate": 6.125419044109629e-06,
+ "loss": 0.0827,
+ "step": 23409
+ },
+ {
+ "epoch": 63.787465940054496,
+ "grad_norm": 1.5917409658432007,
+ "learning_rate": 6.124605498586796e-06,
+ "loss": 0.0244,
+ "step": 23410
+ },
+ {
+ "epoch": 63.79019073569482,
+ "grad_norm": 1.9930446147918701,
+ "learning_rate": 6.123791983245411e-06,
+ "loss": 0.1484,
+ "step": 23411
+ },
+ {
+ "epoch": 63.79291553133515,
+ "grad_norm": 1.683990716934204,
+ "learning_rate": 6.122978498091809e-06,
+ "loss": 0.0371,
+ "step": 23412
+ },
+ {
+ "epoch": 63.79564032697548,
+ "grad_norm": 1.6561030149459839,
+ "learning_rate": 6.1221650431323285e-06,
+ "loss": 0.0165,
+ "step": 23413
+ },
+ {
+ "epoch": 63.798365122615806,
+ "grad_norm": 1.604533314704895,
+ "learning_rate": 6.121351618373299e-06,
+ "loss": 0.0798,
+ "step": 23414
+ },
+ {
+ "epoch": 63.80108991825613,
+ "grad_norm": 1.8890620470046997,
+ "learning_rate": 6.12053822382106e-06,
+ "loss": 0.073,
+ "step": 23415
+ },
+ {
+ "epoch": 63.80381471389646,
+ "grad_norm": 1.4159318208694458,
+ "learning_rate": 6.119724859481943e-06,
+ "loss": 0.1708,
+ "step": 23416
+ },
+ {
+ "epoch": 63.80653950953678,
+ "grad_norm": 1.655693769454956,
+ "learning_rate": 6.1189115253622855e-06,
+ "loss": 0.0256,
+ "step": 23417
+ },
+ {
+ "epoch": 63.80926430517711,
+ "grad_norm": 1.6807457208633423,
+ "learning_rate": 6.118098221468418e-06,
+ "loss": 0.0278,
+ "step": 23418
+ },
+ {
+ "epoch": 63.81198910081744,
+ "grad_norm": 2.16815185546875,
+ "learning_rate": 6.117284947806679e-06,
+ "loss": 0.1071,
+ "step": 23419
+ },
+ {
+ "epoch": 63.81471389645777,
+ "grad_norm": 1.2063204050064087,
+ "learning_rate": 6.116471704383397e-06,
+ "loss": 0.0184,
+ "step": 23420
+ },
+ {
+ "epoch": 63.817438692098094,
+ "grad_norm": 2.0648977756500244,
+ "learning_rate": 6.1156584912049126e-06,
+ "loss": 0.04,
+ "step": 23421
+ },
+ {
+ "epoch": 63.82016348773842,
+ "grad_norm": 1.440682053565979,
+ "learning_rate": 6.114845308277551e-06,
+ "loss": 0.0602,
+ "step": 23422
+ },
+ {
+ "epoch": 63.822888283378745,
+ "grad_norm": 2.1333086490631104,
+ "learning_rate": 6.114032155607651e-06,
+ "loss": 0.0278,
+ "step": 23423
+ },
+ {
+ "epoch": 63.82561307901907,
+ "grad_norm": 1.8964455127716064,
+ "learning_rate": 6.113219033201543e-06,
+ "loss": 0.0185,
+ "step": 23424
+ },
+ {
+ "epoch": 63.828337874659404,
+ "grad_norm": 2.2221946716308594,
+ "learning_rate": 6.11240594106556e-06,
+ "loss": 0.0433,
+ "step": 23425
+ },
+ {
+ "epoch": 63.83106267029973,
+ "grad_norm": 1.3396990299224854,
+ "learning_rate": 6.111592879206032e-06,
+ "loss": 0.0191,
+ "step": 23426
+ },
+ {
+ "epoch": 63.833787465940055,
+ "grad_norm": 2.0160281658172607,
+ "learning_rate": 6.110779847629295e-06,
+ "loss": 0.0538,
+ "step": 23427
+ },
+ {
+ "epoch": 63.83651226158038,
+ "grad_norm": 2.07218599319458,
+ "learning_rate": 6.109966846341681e-06,
+ "loss": 0.0694,
+ "step": 23428
+ },
+ {
+ "epoch": 63.83923705722071,
+ "grad_norm": 2.148660898208618,
+ "learning_rate": 6.109153875349518e-06,
+ "loss": 0.0575,
+ "step": 23429
+ },
+ {
+ "epoch": 63.84196185286103,
+ "grad_norm": 1.7673535346984863,
+ "learning_rate": 6.1083409346591425e-06,
+ "loss": 0.1047,
+ "step": 23430
+ },
+ {
+ "epoch": 63.844686648501366,
+ "grad_norm": 1.2809398174285889,
+ "learning_rate": 6.1075280242768785e-06,
+ "loss": 0.0207,
+ "step": 23431
+ },
+ {
+ "epoch": 63.84741144414169,
+ "grad_norm": 1.9502772092819214,
+ "learning_rate": 6.106715144209063e-06,
+ "loss": 0.0229,
+ "step": 23432
+ },
+ {
+ "epoch": 63.85013623978202,
+ "grad_norm": 1.373421549797058,
+ "learning_rate": 6.105902294462026e-06,
+ "loss": 0.0207,
+ "step": 23433
+ },
+ {
+ "epoch": 63.85286103542234,
+ "grad_norm": 2.0442869663238525,
+ "learning_rate": 6.105089475042094e-06,
+ "loss": 0.0261,
+ "step": 23434
+ },
+ {
+ "epoch": 63.85558583106267,
+ "grad_norm": 1.5261701345443726,
+ "learning_rate": 6.104276685955603e-06,
+ "loss": 0.0781,
+ "step": 23435
+ },
+ {
+ "epoch": 63.858310626702995,
+ "grad_norm": 1.3601734638214111,
+ "learning_rate": 6.103463927208877e-06,
+ "loss": 0.044,
+ "step": 23436
+ },
+ {
+ "epoch": 63.86103542234333,
+ "grad_norm": 0.9020306468009949,
+ "learning_rate": 6.10265119880825e-06,
+ "loss": 0.0162,
+ "step": 23437
+ },
+ {
+ "epoch": 63.86376021798365,
+ "grad_norm": 1.9606302976608276,
+ "learning_rate": 6.101838500760049e-06,
+ "loss": 0.0552,
+ "step": 23438
+ },
+ {
+ "epoch": 63.86648501362398,
+ "grad_norm": 1.684071660041809,
+ "learning_rate": 6.101025833070607e-06,
+ "loss": 0.0271,
+ "step": 23439
+ },
+ {
+ "epoch": 63.869209809264305,
+ "grad_norm": 1.897269606590271,
+ "learning_rate": 6.100213195746246e-06,
+ "loss": 0.0275,
+ "step": 23440
+ },
+ {
+ "epoch": 63.87193460490463,
+ "grad_norm": 1.8842260837554932,
+ "learning_rate": 6.099400588793303e-06,
+ "loss": 0.1605,
+ "step": 23441
+ },
+ {
+ "epoch": 63.87465940054496,
+ "grad_norm": 1.6963528394699097,
+ "learning_rate": 6.0985880122181e-06,
+ "loss": 0.181,
+ "step": 23442
+ },
+ {
+ "epoch": 63.87738419618529,
+ "grad_norm": 5.119819164276123,
+ "learning_rate": 6.097775466026966e-06,
+ "loss": 0.0743,
+ "step": 23443
+ },
+ {
+ "epoch": 63.880108991825615,
+ "grad_norm": 1.2195173501968384,
+ "learning_rate": 6.096962950226235e-06,
+ "loss": 0.0202,
+ "step": 23444
+ },
+ {
+ "epoch": 63.88283378746594,
+ "grad_norm": 1.6584399938583374,
+ "learning_rate": 6.096150464822229e-06,
+ "loss": 0.0475,
+ "step": 23445
+ },
+ {
+ "epoch": 63.88555858310627,
+ "grad_norm": 1.6290390491485596,
+ "learning_rate": 6.095338009821279e-06,
+ "loss": 0.0308,
+ "step": 23446
+ },
+ {
+ "epoch": 63.88828337874659,
+ "grad_norm": 1.434589147567749,
+ "learning_rate": 6.09452558522971e-06,
+ "loss": 0.0307,
+ "step": 23447
+ },
+ {
+ "epoch": 63.89100817438692,
+ "grad_norm": 3.1169729232788086,
+ "learning_rate": 6.093713191053852e-06,
+ "loss": 0.1081,
+ "step": 23448
+ },
+ {
+ "epoch": 63.89373297002725,
+ "grad_norm": 1.0424237251281738,
+ "learning_rate": 6.092900827300028e-06,
+ "loss": 0.0127,
+ "step": 23449
+ },
+ {
+ "epoch": 63.89645776566758,
+ "grad_norm": 1.9725178480148315,
+ "learning_rate": 6.0920884939745675e-06,
+ "loss": 0.0204,
+ "step": 23450
+ },
+ {
+ "epoch": 63.8991825613079,
+ "grad_norm": 1.965431571006775,
+ "learning_rate": 6.091276191083795e-06,
+ "loss": 0.0912,
+ "step": 23451
+ },
+ {
+ "epoch": 63.90190735694823,
+ "grad_norm": 1.5329208374023438,
+ "learning_rate": 6.090463918634039e-06,
+ "loss": 0.021,
+ "step": 23452
+ },
+ {
+ "epoch": 63.904632152588555,
+ "grad_norm": 1.4032388925552368,
+ "learning_rate": 6.089651676631623e-06,
+ "loss": 0.0128,
+ "step": 23453
+ },
+ {
+ "epoch": 63.90735694822888,
+ "grad_norm": 1.7063673734664917,
+ "learning_rate": 6.088839465082873e-06,
+ "loss": 0.228,
+ "step": 23454
+ },
+ {
+ "epoch": 63.91008174386921,
+ "grad_norm": 1.3782106637954712,
+ "learning_rate": 6.088027283994118e-06,
+ "loss": 0.0195,
+ "step": 23455
+ },
+ {
+ "epoch": 63.91280653950954,
+ "grad_norm": 1.3990674018859863,
+ "learning_rate": 6.087215133371678e-06,
+ "loss": 0.0189,
+ "step": 23456
+ },
+ {
+ "epoch": 63.915531335149865,
+ "grad_norm": 1.6541087627410889,
+ "learning_rate": 6.086403013221883e-06,
+ "loss": 0.0233,
+ "step": 23457
+ },
+ {
+ "epoch": 63.91825613079019,
+ "grad_norm": 2.3763363361358643,
+ "learning_rate": 6.085590923551053e-06,
+ "loss": 0.0337,
+ "step": 23458
+ },
+ {
+ "epoch": 63.920980926430516,
+ "grad_norm": 2.4630658626556396,
+ "learning_rate": 6.084778864365517e-06,
+ "loss": 0.0451,
+ "step": 23459
+ },
+ {
+ "epoch": 63.92370572207084,
+ "grad_norm": 2.355365514755249,
+ "learning_rate": 6.083966835671595e-06,
+ "loss": 0.1219,
+ "step": 23460
+ },
+ {
+ "epoch": 63.926430517711175,
+ "grad_norm": 1.1893585920333862,
+ "learning_rate": 6.083154837475613e-06,
+ "loss": 0.0138,
+ "step": 23461
+ },
+ {
+ "epoch": 63.9291553133515,
+ "grad_norm": 1.9722833633422852,
+ "learning_rate": 6.082342869783897e-06,
+ "loss": 0.0641,
+ "step": 23462
+ },
+ {
+ "epoch": 63.93188010899183,
+ "grad_norm": 9.661918640136719,
+ "learning_rate": 6.081530932602768e-06,
+ "loss": 0.0147,
+ "step": 23463
+ },
+ {
+ "epoch": 63.93460490463215,
+ "grad_norm": 2.6636722087860107,
+ "learning_rate": 6.0807190259385506e-06,
+ "loss": 0.0256,
+ "step": 23464
+ },
+ {
+ "epoch": 63.93732970027248,
+ "grad_norm": 1.311886191368103,
+ "learning_rate": 6.079907149797567e-06,
+ "loss": 0.0172,
+ "step": 23465
+ },
+ {
+ "epoch": 63.940054495912804,
+ "grad_norm": 4.192531108856201,
+ "learning_rate": 6.079095304186141e-06,
+ "loss": 0.0722,
+ "step": 23466
+ },
+ {
+ "epoch": 63.94277929155314,
+ "grad_norm": 1.8788368701934814,
+ "learning_rate": 6.078283489110593e-06,
+ "loss": 0.0265,
+ "step": 23467
+ },
+ {
+ "epoch": 63.94550408719346,
+ "grad_norm": 1.5297014713287354,
+ "learning_rate": 6.077471704577249e-06,
+ "loss": 0.0192,
+ "step": 23468
+ },
+ {
+ "epoch": 63.94822888283379,
+ "grad_norm": 1.742236852645874,
+ "learning_rate": 6.076659950592427e-06,
+ "loss": 0.0191,
+ "step": 23469
+ },
+ {
+ "epoch": 63.950953678474114,
+ "grad_norm": 1.427634596824646,
+ "learning_rate": 6.075848227162453e-06,
+ "loss": 0.03,
+ "step": 23470
+ },
+ {
+ "epoch": 63.95367847411444,
+ "grad_norm": 1.6404322385787964,
+ "learning_rate": 6.075036534293646e-06,
+ "loss": 0.0472,
+ "step": 23471
+ },
+ {
+ "epoch": 63.956403269754766,
+ "grad_norm": 2.739321231842041,
+ "learning_rate": 6.074224871992329e-06,
+ "loss": 0.0325,
+ "step": 23472
+ },
+ {
+ "epoch": 63.95912806539509,
+ "grad_norm": 2.3380348682403564,
+ "learning_rate": 6.0734132402648205e-06,
+ "loss": 0.1716,
+ "step": 23473
+ },
+ {
+ "epoch": 63.961852861035425,
+ "grad_norm": 2.1064937114715576,
+ "learning_rate": 6.0726016391174435e-06,
+ "loss": 0.0699,
+ "step": 23474
+ },
+ {
+ "epoch": 63.96457765667575,
+ "grad_norm": 1.898909568786621,
+ "learning_rate": 6.071790068556519e-06,
+ "loss": 0.0246,
+ "step": 23475
+ },
+ {
+ "epoch": 63.967302452316076,
+ "grad_norm": 2.352954149246216,
+ "learning_rate": 6.070978528588369e-06,
+ "loss": 0.1805,
+ "step": 23476
+ },
+ {
+ "epoch": 63.9700272479564,
+ "grad_norm": 1.4739787578582764,
+ "learning_rate": 6.0701670192193085e-06,
+ "loss": 0.0117,
+ "step": 23477
+ },
+ {
+ "epoch": 63.97275204359673,
+ "grad_norm": 2.4022138118743896,
+ "learning_rate": 6.06935554045566e-06,
+ "loss": 0.0656,
+ "step": 23478
+ },
+ {
+ "epoch": 63.97547683923706,
+ "grad_norm": 1.7810442447662354,
+ "learning_rate": 6.068544092303744e-06,
+ "loss": 0.03,
+ "step": 23479
+ },
+ {
+ "epoch": 63.97820163487739,
+ "grad_norm": 1.3616849184036255,
+ "learning_rate": 6.067732674769884e-06,
+ "loss": 0.0156,
+ "step": 23480
+ },
+ {
+ "epoch": 63.98092643051771,
+ "grad_norm": 1.2600090503692627,
+ "learning_rate": 6.0669212878603925e-06,
+ "loss": 0.0159,
+ "step": 23481
+ },
+ {
+ "epoch": 63.98365122615804,
+ "grad_norm": 1.6757488250732422,
+ "learning_rate": 6.066109931581594e-06,
+ "loss": 0.0421,
+ "step": 23482
+ },
+ {
+ "epoch": 63.986376021798364,
+ "grad_norm": 1.3161979913711548,
+ "learning_rate": 6.065298605939803e-06,
+ "loss": 0.0183,
+ "step": 23483
+ },
+ {
+ "epoch": 63.98910081743869,
+ "grad_norm": 1.8751509189605713,
+ "learning_rate": 6.064487310941341e-06,
+ "loss": 0.0948,
+ "step": 23484
+ },
+ {
+ "epoch": 63.991825613079016,
+ "grad_norm": 1.5392162799835205,
+ "learning_rate": 6.063676046592523e-06,
+ "loss": 0.0293,
+ "step": 23485
+ },
+ {
+ "epoch": 63.99455040871935,
+ "grad_norm": 1.7878614664077759,
+ "learning_rate": 6.062864812899672e-06,
+ "loss": 0.0243,
+ "step": 23486
+ },
+ {
+ "epoch": 63.997275204359674,
+ "grad_norm": 2.110687494277954,
+ "learning_rate": 6.0620536098690995e-06,
+ "loss": 0.1294,
+ "step": 23487
+ },
+ {
+ "epoch": 64.0,
+ "grad_norm": 1.5619590282440186,
+ "learning_rate": 6.061242437507131e-06,
+ "loss": 0.0632,
+ "step": 23488
+ },
+ {
+ "epoch": 64.00272479564033,
+ "grad_norm": 3.3525636196136475,
+ "learning_rate": 6.0604312958200754e-06,
+ "loss": 0.0437,
+ "step": 23489
+ },
+ {
+ "epoch": 64.00544959128065,
+ "grad_norm": 3.3104097843170166,
+ "learning_rate": 6.059620184814258e-06,
+ "loss": 0.0295,
+ "step": 23490
+ },
+ {
+ "epoch": 64.00817438692098,
+ "grad_norm": 1.701200008392334,
+ "learning_rate": 6.058809104495988e-06,
+ "loss": 0.0797,
+ "step": 23491
+ },
+ {
+ "epoch": 64.0108991825613,
+ "grad_norm": 2.333894968032837,
+ "learning_rate": 6.05799805487159e-06,
+ "loss": 0.0512,
+ "step": 23492
+ },
+ {
+ "epoch": 64.01362397820164,
+ "grad_norm": 2.341613292694092,
+ "learning_rate": 6.057187035947373e-06,
+ "loss": 0.0297,
+ "step": 23493
+ },
+ {
+ "epoch": 64.01634877384195,
+ "grad_norm": 2.4151954650878906,
+ "learning_rate": 6.056376047729658e-06,
+ "loss": 0.0173,
+ "step": 23494
+ },
+ {
+ "epoch": 64.01907356948229,
+ "grad_norm": 1.1983225345611572,
+ "learning_rate": 6.055565090224756e-06,
+ "loss": 0.016,
+ "step": 23495
+ },
+ {
+ "epoch": 64.02179836512262,
+ "grad_norm": 3.3390347957611084,
+ "learning_rate": 6.054754163438989e-06,
+ "loss": 0.1021,
+ "step": 23496
+ },
+ {
+ "epoch": 64.02452316076294,
+ "grad_norm": 225.50929260253906,
+ "learning_rate": 6.053943267378668e-06,
+ "loss": 0.0139,
+ "step": 23497
+ },
+ {
+ "epoch": 64.02724795640327,
+ "grad_norm": 2.1733148097991943,
+ "learning_rate": 6.053132402050112e-06,
+ "loss": 0.0685,
+ "step": 23498
+ },
+ {
+ "epoch": 64.02997275204359,
+ "grad_norm": 1.4872686862945557,
+ "learning_rate": 6.052321567459634e-06,
+ "loss": 0.0141,
+ "step": 23499
+ },
+ {
+ "epoch": 64.03269754768392,
+ "grad_norm": 2.468593120574951,
+ "learning_rate": 6.051510763613548e-06,
+ "loss": 0.0878,
+ "step": 23500
+ },
+ {
+ "epoch": 64.03542234332426,
+ "grad_norm": 1.7920390367507935,
+ "learning_rate": 6.0506999905181696e-06,
+ "loss": 0.0239,
+ "step": 23501
+ },
+ {
+ "epoch": 64.03814713896458,
+ "grad_norm": 2.054257869720459,
+ "learning_rate": 6.049889248179812e-06,
+ "loss": 0.039,
+ "step": 23502
+ },
+ {
+ "epoch": 64.04087193460491,
+ "grad_norm": 1.8132596015930176,
+ "learning_rate": 6.049078536604789e-06,
+ "loss": 0.0213,
+ "step": 23503
+ },
+ {
+ "epoch": 64.04359673024523,
+ "grad_norm": 2.076725959777832,
+ "learning_rate": 6.048267855799417e-06,
+ "loss": 0.0369,
+ "step": 23504
+ },
+ {
+ "epoch": 64.04632152588556,
+ "grad_norm": 2.2015223503112793,
+ "learning_rate": 6.047457205770006e-06,
+ "loss": 0.0458,
+ "step": 23505
+ },
+ {
+ "epoch": 64.04904632152588,
+ "grad_norm": 1.5919333696365356,
+ "learning_rate": 6.046646586522873e-06,
+ "loss": 0.03,
+ "step": 23506
+ },
+ {
+ "epoch": 64.05177111716621,
+ "grad_norm": 3.115950107574463,
+ "learning_rate": 6.045835998064326e-06,
+ "loss": 0.041,
+ "step": 23507
+ },
+ {
+ "epoch": 64.05449591280654,
+ "grad_norm": 1.9749987125396729,
+ "learning_rate": 6.045025440400684e-06,
+ "loss": 0.0302,
+ "step": 23508
+ },
+ {
+ "epoch": 64.05722070844686,
+ "grad_norm": 2.0687825679779053,
+ "learning_rate": 6.044214913538255e-06,
+ "loss": 0.03,
+ "step": 23509
+ },
+ {
+ "epoch": 64.0599455040872,
+ "grad_norm": 1.7457873821258545,
+ "learning_rate": 6.043404417483356e-06,
+ "loss": 0.0243,
+ "step": 23510
+ },
+ {
+ "epoch": 64.06267029972751,
+ "grad_norm": 4.077377796173096,
+ "learning_rate": 6.0425939522422926e-06,
+ "loss": 0.0429,
+ "step": 23511
+ },
+ {
+ "epoch": 64.06539509536785,
+ "grad_norm": 2.4884374141693115,
+ "learning_rate": 6.041783517821378e-06,
+ "loss": 0.057,
+ "step": 23512
+ },
+ {
+ "epoch": 64.06811989100818,
+ "grad_norm": 1.5012216567993164,
+ "learning_rate": 6.040973114226929e-06,
+ "loss": 0.0155,
+ "step": 23513
+ },
+ {
+ "epoch": 64.0708446866485,
+ "grad_norm": 1.9774256944656372,
+ "learning_rate": 6.0401627414652545e-06,
+ "loss": 0.1828,
+ "step": 23514
+ },
+ {
+ "epoch": 64.07356948228883,
+ "grad_norm": 1.5298981666564941,
+ "learning_rate": 6.039352399542665e-06,
+ "loss": 0.0206,
+ "step": 23515
+ },
+ {
+ "epoch": 64.07629427792915,
+ "grad_norm": 2.2143571376800537,
+ "learning_rate": 6.03854208846547e-06,
+ "loss": 0.0273,
+ "step": 23516
+ },
+ {
+ "epoch": 64.07901907356948,
+ "grad_norm": 1.4700251817703247,
+ "learning_rate": 6.037731808239984e-06,
+ "loss": 0.0205,
+ "step": 23517
+ },
+ {
+ "epoch": 64.0817438692098,
+ "grad_norm": 5.778901100158691,
+ "learning_rate": 6.036921558872513e-06,
+ "loss": 0.032,
+ "step": 23518
+ },
+ {
+ "epoch": 64.08446866485014,
+ "grad_norm": 2.943141222000122,
+ "learning_rate": 6.03611134036937e-06,
+ "loss": 0.131,
+ "step": 23519
+ },
+ {
+ "epoch": 64.08719346049047,
+ "grad_norm": 2.5471081733703613,
+ "learning_rate": 6.035301152736866e-06,
+ "loss": 0.0287,
+ "step": 23520
+ },
+ {
+ "epoch": 64.08991825613079,
+ "grad_norm": 1.6932073831558228,
+ "learning_rate": 6.0344909959813055e-06,
+ "loss": 0.0675,
+ "step": 23521
+ },
+ {
+ "epoch": 64.09264305177112,
+ "grad_norm": 3.040344476699829,
+ "learning_rate": 6.033680870109005e-06,
+ "loss": 0.0314,
+ "step": 23522
+ },
+ {
+ "epoch": 64.09536784741144,
+ "grad_norm": 3.18621826171875,
+ "learning_rate": 6.032870775126268e-06,
+ "loss": 0.0318,
+ "step": 23523
+ },
+ {
+ "epoch": 64.09809264305177,
+ "grad_norm": 2.254574775695801,
+ "learning_rate": 6.032060711039408e-06,
+ "loss": 0.0219,
+ "step": 23524
+ },
+ {
+ "epoch": 64.1008174386921,
+ "grad_norm": 2.950350761413574,
+ "learning_rate": 6.03125067785473e-06,
+ "loss": 0.1134,
+ "step": 23525
+ },
+ {
+ "epoch": 64.10354223433242,
+ "grad_norm": 2.4653210639953613,
+ "learning_rate": 6.030440675578544e-06,
+ "loss": 0.126,
+ "step": 23526
+ },
+ {
+ "epoch": 64.10626702997276,
+ "grad_norm": 2.2051022052764893,
+ "learning_rate": 6.0296307042171575e-06,
+ "loss": 0.1507,
+ "step": 23527
+ },
+ {
+ "epoch": 64.10899182561307,
+ "grad_norm": 3.547858238220215,
+ "learning_rate": 6.0288207637768815e-06,
+ "loss": 0.0394,
+ "step": 23528
+ },
+ {
+ "epoch": 64.11171662125341,
+ "grad_norm": 2.438713550567627,
+ "learning_rate": 6.028010854264017e-06,
+ "loss": 0.0496,
+ "step": 23529
+ },
+ {
+ "epoch": 64.11444141689373,
+ "grad_norm": 2.454364538192749,
+ "learning_rate": 6.027200975684879e-06,
+ "loss": 0.0252,
+ "step": 23530
+ },
+ {
+ "epoch": 64.11716621253406,
+ "grad_norm": 1.6995021104812622,
+ "learning_rate": 6.026391128045773e-06,
+ "loss": 0.0531,
+ "step": 23531
+ },
+ {
+ "epoch": 64.11989100817439,
+ "grad_norm": 2.3260889053344727,
+ "learning_rate": 6.0255813113530036e-06,
+ "loss": 0.1227,
+ "step": 23532
+ },
+ {
+ "epoch": 64.12261580381471,
+ "grad_norm": 2.2693963050842285,
+ "learning_rate": 6.02477152561288e-06,
+ "loss": 0.0352,
+ "step": 23533
+ },
+ {
+ "epoch": 64.12534059945504,
+ "grad_norm": 2.243685007095337,
+ "learning_rate": 6.023961770831708e-06,
+ "loss": 0.1769,
+ "step": 23534
+ },
+ {
+ "epoch": 64.12806539509536,
+ "grad_norm": 2.3290212154388428,
+ "learning_rate": 6.023152047015794e-06,
+ "loss": 0.0259,
+ "step": 23535
+ },
+ {
+ "epoch": 64.1307901907357,
+ "grad_norm": 1.3964065313339233,
+ "learning_rate": 6.022342354171442e-06,
+ "loss": 0.0803,
+ "step": 23536
+ },
+ {
+ "epoch": 64.13351498637603,
+ "grad_norm": 2.19454288482666,
+ "learning_rate": 6.021532692304962e-06,
+ "loss": 0.0399,
+ "step": 23537
+ },
+ {
+ "epoch": 64.13623978201635,
+ "grad_norm": 1.8686907291412354,
+ "learning_rate": 6.020723061422655e-06,
+ "loss": 0.0204,
+ "step": 23538
+ },
+ {
+ "epoch": 64.13896457765668,
+ "grad_norm": 1.7992308139801025,
+ "learning_rate": 6.019913461530831e-06,
+ "loss": 0.0256,
+ "step": 23539
+ },
+ {
+ "epoch": 64.141689373297,
+ "grad_norm": 2.176311731338501,
+ "learning_rate": 6.019103892635791e-06,
+ "loss": 0.169,
+ "step": 23540
+ },
+ {
+ "epoch": 64.14441416893733,
+ "grad_norm": 2.4206736087799072,
+ "learning_rate": 6.018294354743842e-06,
+ "loss": 0.0641,
+ "step": 23541
+ },
+ {
+ "epoch": 64.14713896457765,
+ "grad_norm": 1.2569878101348877,
+ "learning_rate": 6.01748484786129e-06,
+ "loss": 0.0208,
+ "step": 23542
+ },
+ {
+ "epoch": 64.14986376021798,
+ "grad_norm": 2.406934976577759,
+ "learning_rate": 6.016675371994436e-06,
+ "loss": 0.0248,
+ "step": 23543
+ },
+ {
+ "epoch": 64.15258855585832,
+ "grad_norm": 1.6311570405960083,
+ "learning_rate": 6.015865927149588e-06,
+ "loss": 0.021,
+ "step": 23544
+ },
+ {
+ "epoch": 64.15531335149863,
+ "grad_norm": 1.503029227256775,
+ "learning_rate": 6.015056513333046e-06,
+ "loss": 0.0222,
+ "step": 23545
+ },
+ {
+ "epoch": 64.15803814713897,
+ "grad_norm": 2.3151423931121826,
+ "learning_rate": 6.014247130551117e-06,
+ "loss": 0.0629,
+ "step": 23546
+ },
+ {
+ "epoch": 64.16076294277929,
+ "grad_norm": 2.4885897636413574,
+ "learning_rate": 6.013437778810101e-06,
+ "loss": 0.0473,
+ "step": 23547
+ },
+ {
+ "epoch": 64.16348773841962,
+ "grad_norm": 1.9828487634658813,
+ "learning_rate": 6.012628458116303e-06,
+ "loss": 0.0378,
+ "step": 23548
+ },
+ {
+ "epoch": 64.16621253405995,
+ "grad_norm": 2.209428548812866,
+ "learning_rate": 6.011819168476028e-06,
+ "loss": 0.1816,
+ "step": 23549
+ },
+ {
+ "epoch": 64.16893732970027,
+ "grad_norm": 1.586296796798706,
+ "learning_rate": 6.011009909895576e-06,
+ "loss": 0.041,
+ "step": 23550
+ },
+ {
+ "epoch": 64.1716621253406,
+ "grad_norm": 2.4690206050872803,
+ "learning_rate": 6.010200682381253e-06,
+ "loss": 0.039,
+ "step": 23551
+ },
+ {
+ "epoch": 64.17438692098092,
+ "grad_norm": 2.214116096496582,
+ "learning_rate": 6.009391485939355e-06,
+ "loss": 0.0311,
+ "step": 23552
+ },
+ {
+ "epoch": 64.17711171662125,
+ "grad_norm": 2.1516871452331543,
+ "learning_rate": 6.008582320576191e-06,
+ "loss": 0.0327,
+ "step": 23553
+ },
+ {
+ "epoch": 64.17983651226157,
+ "grad_norm": 2.110696792602539,
+ "learning_rate": 6.0077731862980584e-06,
+ "loss": 0.0347,
+ "step": 23554
+ },
+ {
+ "epoch": 64.1825613079019,
+ "grad_norm": 1.9014217853546143,
+ "learning_rate": 6.00696408311126e-06,
+ "loss": 0.0183,
+ "step": 23555
+ },
+ {
+ "epoch": 64.18528610354224,
+ "grad_norm": 2.1551408767700195,
+ "learning_rate": 6.0061550110220954e-06,
+ "loss": 0.1134,
+ "step": 23556
+ },
+ {
+ "epoch": 64.18801089918256,
+ "grad_norm": 1.489501714706421,
+ "learning_rate": 6.005345970036868e-06,
+ "loss": 0.0187,
+ "step": 23557
+ },
+ {
+ "epoch": 64.19073569482289,
+ "grad_norm": 2.2775638103485107,
+ "learning_rate": 6.004536960161877e-06,
+ "loss": 0.0925,
+ "step": 23558
+ },
+ {
+ "epoch": 64.19346049046321,
+ "grad_norm": 1.6332736015319824,
+ "learning_rate": 6.003727981403427e-06,
+ "loss": 0.0221,
+ "step": 23559
+ },
+ {
+ "epoch": 64.19618528610354,
+ "grad_norm": 2.46966814994812,
+ "learning_rate": 6.0029190337678104e-06,
+ "loss": 0.0338,
+ "step": 23560
+ },
+ {
+ "epoch": 64.19891008174388,
+ "grad_norm": 2.8220975399017334,
+ "learning_rate": 6.002110117261334e-06,
+ "loss": 0.2817,
+ "step": 23561
+ },
+ {
+ "epoch": 64.2016348773842,
+ "grad_norm": 1.2849136590957642,
+ "learning_rate": 6.001301231890293e-06,
+ "loss": 0.0161,
+ "step": 23562
+ },
+ {
+ "epoch": 64.20435967302453,
+ "grad_norm": 1.9750442504882812,
+ "learning_rate": 6.0004923776609925e-06,
+ "loss": 0.0654,
+ "step": 23563
+ },
+ {
+ "epoch": 64.20708446866485,
+ "grad_norm": 2.008054494857788,
+ "learning_rate": 5.999683554579723e-06,
+ "loss": 0.0333,
+ "step": 23564
+ },
+ {
+ "epoch": 64.20980926430518,
+ "grad_norm": 2.950214147567749,
+ "learning_rate": 5.998874762652793e-06,
+ "loss": 0.0668,
+ "step": 23565
+ },
+ {
+ "epoch": 64.2125340599455,
+ "grad_norm": 1.5923957824707031,
+ "learning_rate": 5.998066001886498e-06,
+ "loss": 0.0406,
+ "step": 23566
+ },
+ {
+ "epoch": 64.21525885558583,
+ "grad_norm": 1.965442419052124,
+ "learning_rate": 5.997257272287136e-06,
+ "loss": 0.0272,
+ "step": 23567
+ },
+ {
+ "epoch": 64.21798365122616,
+ "grad_norm": 2.6716713905334473,
+ "learning_rate": 5.9964485738610045e-06,
+ "loss": 0.0574,
+ "step": 23568
+ },
+ {
+ "epoch": 64.22070844686648,
+ "grad_norm": 1.8160067796707153,
+ "learning_rate": 5.995639906614406e-06,
+ "loss": 0.018,
+ "step": 23569
+ },
+ {
+ "epoch": 64.22343324250681,
+ "grad_norm": 2.790663719177246,
+ "learning_rate": 5.994831270553632e-06,
+ "loss": 0.0352,
+ "step": 23570
+ },
+ {
+ "epoch": 64.22615803814713,
+ "grad_norm": 2.1898200511932373,
+ "learning_rate": 5.994022665684985e-06,
+ "loss": 0.0295,
+ "step": 23571
+ },
+ {
+ "epoch": 64.22888283378747,
+ "grad_norm": 5.050644397735596,
+ "learning_rate": 5.993214092014758e-06,
+ "loss": 0.027,
+ "step": 23572
+ },
+ {
+ "epoch": 64.2316076294278,
+ "grad_norm": 2.256417751312256,
+ "learning_rate": 5.992405549549253e-06,
+ "loss": 0.0247,
+ "step": 23573
+ },
+ {
+ "epoch": 64.23433242506812,
+ "grad_norm": 2.363283634185791,
+ "learning_rate": 5.991597038294763e-06,
+ "loss": 0.0308,
+ "step": 23574
+ },
+ {
+ "epoch": 64.23705722070845,
+ "grad_norm": 2.6039416790008545,
+ "learning_rate": 5.9907885582575875e-06,
+ "loss": 0.072,
+ "step": 23575
+ },
+ {
+ "epoch": 64.23978201634877,
+ "grad_norm": 2.8263919353485107,
+ "learning_rate": 5.989980109444021e-06,
+ "loss": 0.0634,
+ "step": 23576
+ },
+ {
+ "epoch": 64.2425068119891,
+ "grad_norm": 1.755186915397644,
+ "learning_rate": 5.98917169186036e-06,
+ "loss": 0.1086,
+ "step": 23577
+ },
+ {
+ "epoch": 64.24523160762942,
+ "grad_norm": 2.582397937774658,
+ "learning_rate": 5.988363305512901e-06,
+ "loss": 0.0428,
+ "step": 23578
+ },
+ {
+ "epoch": 64.24795640326975,
+ "grad_norm": 1.8449159860610962,
+ "learning_rate": 5.987554950407941e-06,
+ "loss": 0.0278,
+ "step": 23579
+ },
+ {
+ "epoch": 64.25068119891009,
+ "grad_norm": 1.8789074420928955,
+ "learning_rate": 5.986746626551772e-06,
+ "loss": 0.0233,
+ "step": 23580
+ },
+ {
+ "epoch": 64.2534059945504,
+ "grad_norm": 2.5938756465911865,
+ "learning_rate": 5.985938333950689e-06,
+ "loss": 0.0297,
+ "step": 23581
+ },
+ {
+ "epoch": 64.25613079019074,
+ "grad_norm": 2.3403921127319336,
+ "learning_rate": 5.985130072610993e-06,
+ "loss": 0.0647,
+ "step": 23582
+ },
+ {
+ "epoch": 64.25885558583106,
+ "grad_norm": 1.7774924039840698,
+ "learning_rate": 5.984321842538973e-06,
+ "loss": 0.0226,
+ "step": 23583
+ },
+ {
+ "epoch": 64.26158038147139,
+ "grad_norm": 2.4413232803344727,
+ "learning_rate": 5.9835136437409235e-06,
+ "loss": 0.0857,
+ "step": 23584
+ },
+ {
+ "epoch": 64.26430517711172,
+ "grad_norm": 1.7309285402297974,
+ "learning_rate": 5.982705476223145e-06,
+ "loss": 0.0273,
+ "step": 23585
+ },
+ {
+ "epoch": 64.26702997275204,
+ "grad_norm": 3.088890790939331,
+ "learning_rate": 5.981897339991923e-06,
+ "loss": 0.0286,
+ "step": 23586
+ },
+ {
+ "epoch": 64.26975476839237,
+ "grad_norm": 1.2417443990707397,
+ "learning_rate": 5.981089235053559e-06,
+ "loss": 0.0203,
+ "step": 23587
+ },
+ {
+ "epoch": 64.2724795640327,
+ "grad_norm": 1.5977617502212524,
+ "learning_rate": 5.980281161414339e-06,
+ "loss": 0.0245,
+ "step": 23588
+ },
+ {
+ "epoch": 64.27520435967303,
+ "grad_norm": 1.5502660274505615,
+ "learning_rate": 5.9794731190805635e-06,
+ "loss": 0.0962,
+ "step": 23589
+ },
+ {
+ "epoch": 64.27792915531334,
+ "grad_norm": 2.1440000534057617,
+ "learning_rate": 5.978665108058519e-06,
+ "loss": 0.069,
+ "step": 23590
+ },
+ {
+ "epoch": 64.28065395095368,
+ "grad_norm": 1.2899996042251587,
+ "learning_rate": 5.977857128354503e-06,
+ "loss": 0.0181,
+ "step": 23591
+ },
+ {
+ "epoch": 64.28337874659401,
+ "grad_norm": 1.9926526546478271,
+ "learning_rate": 5.977049179974804e-06,
+ "loss": 0.0474,
+ "step": 23592
+ },
+ {
+ "epoch": 64.28610354223433,
+ "grad_norm": 2.051173448562622,
+ "learning_rate": 5.976241262925719e-06,
+ "loss": 0.0217,
+ "step": 23593
+ },
+ {
+ "epoch": 64.28882833787466,
+ "grad_norm": 1.7425869703292847,
+ "learning_rate": 5.975433377213537e-06,
+ "loss": 0.0342,
+ "step": 23594
+ },
+ {
+ "epoch": 64.29155313351498,
+ "grad_norm": 2.0548555850982666,
+ "learning_rate": 5.974625522844551e-06,
+ "loss": 0.097,
+ "step": 23595
+ },
+ {
+ "epoch": 64.29427792915531,
+ "grad_norm": 3.4214019775390625,
+ "learning_rate": 5.9738176998250484e-06,
+ "loss": 0.1114,
+ "step": 23596
+ },
+ {
+ "epoch": 64.29700272479565,
+ "grad_norm": 1.4964486360549927,
+ "learning_rate": 5.973009908161328e-06,
+ "loss": 0.0404,
+ "step": 23597
+ },
+ {
+ "epoch": 64.29972752043597,
+ "grad_norm": 5.625494480133057,
+ "learning_rate": 5.972202147859671e-06,
+ "loss": 0.0314,
+ "step": 23598
+ },
+ {
+ "epoch": 64.3024523160763,
+ "grad_norm": 2.5002219676971436,
+ "learning_rate": 5.971394418926379e-06,
+ "loss": 0.0475,
+ "step": 23599
+ },
+ {
+ "epoch": 64.30517711171662,
+ "grad_norm": 3.6179587841033936,
+ "learning_rate": 5.970586721367737e-06,
+ "loss": 0.1505,
+ "step": 23600
+ },
+ {
+ "epoch": 64.30790190735695,
+ "grad_norm": 1.7506752014160156,
+ "learning_rate": 5.969779055190035e-06,
+ "loss": 0.0191,
+ "step": 23601
+ },
+ {
+ "epoch": 64.31062670299727,
+ "grad_norm": 1.6938704252243042,
+ "learning_rate": 5.968971420399565e-06,
+ "loss": 0.0213,
+ "step": 23602
+ },
+ {
+ "epoch": 64.3133514986376,
+ "grad_norm": 2.0760509967803955,
+ "learning_rate": 5.968163817002616e-06,
+ "loss": 0.0362,
+ "step": 23603
+ },
+ {
+ "epoch": 64.31607629427793,
+ "grad_norm": 1.3969857692718506,
+ "learning_rate": 5.967356245005478e-06,
+ "loss": 0.0195,
+ "step": 23604
+ },
+ {
+ "epoch": 64.31880108991825,
+ "grad_norm": 2.606356620788574,
+ "learning_rate": 5.966548704414436e-06,
+ "loss": 0.03,
+ "step": 23605
+ },
+ {
+ "epoch": 64.32152588555859,
+ "grad_norm": 1.910689115524292,
+ "learning_rate": 5.965741195235786e-06,
+ "loss": 0.0258,
+ "step": 23606
+ },
+ {
+ "epoch": 64.3242506811989,
+ "grad_norm": 1.855642318725586,
+ "learning_rate": 5.964933717475814e-06,
+ "loss": 0.1027,
+ "step": 23607
+ },
+ {
+ "epoch": 64.32697547683924,
+ "grad_norm": 1.6124192476272583,
+ "learning_rate": 5.964126271140807e-06,
+ "loss": 0.1649,
+ "step": 23608
+ },
+ {
+ "epoch": 64.32970027247957,
+ "grad_norm": 3.467219352722168,
+ "learning_rate": 5.963318856237057e-06,
+ "loss": 0.0858,
+ "step": 23609
+ },
+ {
+ "epoch": 64.33242506811989,
+ "grad_norm": 1.7685306072235107,
+ "learning_rate": 5.962511472770847e-06,
+ "loss": 0.028,
+ "step": 23610
+ },
+ {
+ "epoch": 64.33514986376022,
+ "grad_norm": 1.891403317451477,
+ "learning_rate": 5.96170412074847e-06,
+ "loss": 0.0718,
+ "step": 23611
+ },
+ {
+ "epoch": 64.33787465940054,
+ "grad_norm": 1.8968538045883179,
+ "learning_rate": 5.96089680017621e-06,
+ "loss": 0.0464,
+ "step": 23612
+ },
+ {
+ "epoch": 64.34059945504087,
+ "grad_norm": 2.7695956230163574,
+ "learning_rate": 5.960089511060357e-06,
+ "loss": 0.3111,
+ "step": 23613
+ },
+ {
+ "epoch": 64.34332425068119,
+ "grad_norm": 3.3405063152313232,
+ "learning_rate": 5.959282253407195e-06,
+ "loss": 0.0726,
+ "step": 23614
+ },
+ {
+ "epoch": 64.34604904632153,
+ "grad_norm": 1.883156418800354,
+ "learning_rate": 5.958475027223009e-06,
+ "loss": 0.0283,
+ "step": 23615
+ },
+ {
+ "epoch": 64.34877384196186,
+ "grad_norm": 1.9489076137542725,
+ "learning_rate": 5.9576678325140956e-06,
+ "loss": 0.0253,
+ "step": 23616
+ },
+ {
+ "epoch": 64.35149863760218,
+ "grad_norm": 1.8857412338256836,
+ "learning_rate": 5.9568606692867325e-06,
+ "loss": 0.0351,
+ "step": 23617
+ },
+ {
+ "epoch": 64.35422343324251,
+ "grad_norm": 2.401679039001465,
+ "learning_rate": 5.95605353754721e-06,
+ "loss": 0.0881,
+ "step": 23618
+ },
+ {
+ "epoch": 64.35694822888283,
+ "grad_norm": 1.732349157333374,
+ "learning_rate": 5.955246437301811e-06,
+ "loss": 0.2519,
+ "step": 23619
+ },
+ {
+ "epoch": 64.35967302452316,
+ "grad_norm": 2.1596903800964355,
+ "learning_rate": 5.954439368556826e-06,
+ "loss": 0.0289,
+ "step": 23620
+ },
+ {
+ "epoch": 64.3623978201635,
+ "grad_norm": 2.1449859142303467,
+ "learning_rate": 5.953632331318533e-06,
+ "loss": 0.0337,
+ "step": 23621
+ },
+ {
+ "epoch": 64.36512261580381,
+ "grad_norm": 2.083400011062622,
+ "learning_rate": 5.952825325593223e-06,
+ "loss": 0.0288,
+ "step": 23622
+ },
+ {
+ "epoch": 64.36784741144415,
+ "grad_norm": 1.58384108543396,
+ "learning_rate": 5.952018351387178e-06,
+ "loss": 0.0492,
+ "step": 23623
+ },
+ {
+ "epoch": 64.37057220708446,
+ "grad_norm": 1.6173374652862549,
+ "learning_rate": 5.951211408706685e-06,
+ "loss": 0.0191,
+ "step": 23624
+ },
+ {
+ "epoch": 64.3732970027248,
+ "grad_norm": 4.541933059692383,
+ "learning_rate": 5.9504044975580265e-06,
+ "loss": 0.0507,
+ "step": 23625
+ },
+ {
+ "epoch": 64.37602179836512,
+ "grad_norm": 1.524340271949768,
+ "learning_rate": 5.9495976179474876e-06,
+ "loss": 0.0178,
+ "step": 23626
+ },
+ {
+ "epoch": 64.37874659400545,
+ "grad_norm": 1.9175097942352295,
+ "learning_rate": 5.948790769881352e-06,
+ "loss": 0.1108,
+ "step": 23627
+ },
+ {
+ "epoch": 64.38147138964578,
+ "grad_norm": 1.564657211303711,
+ "learning_rate": 5.947983953365902e-06,
+ "loss": 0.0391,
+ "step": 23628
+ },
+ {
+ "epoch": 64.3841961852861,
+ "grad_norm": 2.063427686691284,
+ "learning_rate": 5.947177168407425e-06,
+ "loss": 0.0181,
+ "step": 23629
+ },
+ {
+ "epoch": 64.38692098092643,
+ "grad_norm": 2.4404654502868652,
+ "learning_rate": 5.9463704150122e-06,
+ "loss": 0.0371,
+ "step": 23630
+ },
+ {
+ "epoch": 64.38964577656675,
+ "grad_norm": 1.9750703573226929,
+ "learning_rate": 5.945563693186514e-06,
+ "loss": 0.0586,
+ "step": 23631
+ },
+ {
+ "epoch": 64.39237057220708,
+ "grad_norm": 2.156714677810669,
+ "learning_rate": 5.9447570029366456e-06,
+ "loss": 0.0863,
+ "step": 23632
+ },
+ {
+ "epoch": 64.39509536784742,
+ "grad_norm": 1.523685336112976,
+ "learning_rate": 5.943950344268876e-06,
+ "loss": 0.1779,
+ "step": 23633
+ },
+ {
+ "epoch": 64.39782016348774,
+ "grad_norm": 1.6193625926971436,
+ "learning_rate": 5.943143717189495e-06,
+ "loss": 0.0162,
+ "step": 23634
+ },
+ {
+ "epoch": 64.40054495912807,
+ "grad_norm": 2.4640514850616455,
+ "learning_rate": 5.942337121704779e-06,
+ "loss": 0.1347,
+ "step": 23635
+ },
+ {
+ "epoch": 64.40326975476839,
+ "grad_norm": 2.53055739402771,
+ "learning_rate": 5.941530557821012e-06,
+ "loss": 0.1437,
+ "step": 23636
+ },
+ {
+ "epoch": 64.40599455040872,
+ "grad_norm": 1.4927127361297607,
+ "learning_rate": 5.940724025544473e-06,
+ "loss": 0.0188,
+ "step": 23637
+ },
+ {
+ "epoch": 64.40871934604904,
+ "grad_norm": 1.5061341524124146,
+ "learning_rate": 5.939917524881446e-06,
+ "loss": 0.0196,
+ "step": 23638
+ },
+ {
+ "epoch": 64.41144414168937,
+ "grad_norm": 1.5975004434585571,
+ "learning_rate": 5.939111055838209e-06,
+ "loss": 0.1048,
+ "step": 23639
+ },
+ {
+ "epoch": 64.4141689373297,
+ "grad_norm": 1.6414061784744263,
+ "learning_rate": 5.9383046184210465e-06,
+ "loss": 0.06,
+ "step": 23640
+ },
+ {
+ "epoch": 64.41689373297002,
+ "grad_norm": 1.640093207359314,
+ "learning_rate": 5.937498212636235e-06,
+ "loss": 0.0739,
+ "step": 23641
+ },
+ {
+ "epoch": 64.41961852861036,
+ "grad_norm": 2.5547094345092773,
+ "learning_rate": 5.936691838490058e-06,
+ "loss": 0.1049,
+ "step": 23642
+ },
+ {
+ "epoch": 64.42234332425068,
+ "grad_norm": 2.501770257949829,
+ "learning_rate": 5.935885495988794e-06,
+ "loss": 0.0466,
+ "step": 23643
+ },
+ {
+ "epoch": 64.42506811989101,
+ "grad_norm": 1.4338629245758057,
+ "learning_rate": 5.935079185138722e-06,
+ "loss": 0.0137,
+ "step": 23644
+ },
+ {
+ "epoch": 64.42779291553134,
+ "grad_norm": 1.502241611480713,
+ "learning_rate": 5.934272905946123e-06,
+ "loss": 0.0306,
+ "step": 23645
+ },
+ {
+ "epoch": 64.43051771117166,
+ "grad_norm": 3.0494773387908936,
+ "learning_rate": 5.933466658417277e-06,
+ "loss": 0.0502,
+ "step": 23646
+ },
+ {
+ "epoch": 64.433242506812,
+ "grad_norm": 1.7371852397918701,
+ "learning_rate": 5.932660442558459e-06,
+ "loss": 0.0298,
+ "step": 23647
+ },
+ {
+ "epoch": 64.43596730245231,
+ "grad_norm": 2.1895487308502197,
+ "learning_rate": 5.931854258375953e-06,
+ "loss": 0.0595,
+ "step": 23648
+ },
+ {
+ "epoch": 64.43869209809264,
+ "grad_norm": 1.5880109071731567,
+ "learning_rate": 5.931048105876034e-06,
+ "loss": 0.0397,
+ "step": 23649
+ },
+ {
+ "epoch": 64.44141689373296,
+ "grad_norm": 1.9861952066421509,
+ "learning_rate": 5.930241985064978e-06,
+ "loss": 0.0717,
+ "step": 23650
+ },
+ {
+ "epoch": 64.4441416893733,
+ "grad_norm": 1.622717022895813,
+ "learning_rate": 5.92943589594907e-06,
+ "loss": 0.0318,
+ "step": 23651
+ },
+ {
+ "epoch": 64.44686648501363,
+ "grad_norm": 2.3286445140838623,
+ "learning_rate": 5.928629838534586e-06,
+ "loss": 0.0394,
+ "step": 23652
+ },
+ {
+ "epoch": 64.44959128065395,
+ "grad_norm": 2.0463783740997314,
+ "learning_rate": 5.9278238128277975e-06,
+ "loss": 0.0328,
+ "step": 23653
+ },
+ {
+ "epoch": 64.45231607629428,
+ "grad_norm": 2.7745883464813232,
+ "learning_rate": 5.92701781883499e-06,
+ "loss": 0.0347,
+ "step": 23654
+ },
+ {
+ "epoch": 64.4550408719346,
+ "grad_norm": 1.9597136974334717,
+ "learning_rate": 5.926211856562434e-06,
+ "loss": 0.0225,
+ "step": 23655
+ },
+ {
+ "epoch": 64.45776566757493,
+ "grad_norm": 2.537780284881592,
+ "learning_rate": 5.925405926016412e-06,
+ "loss": 0.0276,
+ "step": 23656
+ },
+ {
+ "epoch": 64.46049046321527,
+ "grad_norm": 1.801784634590149,
+ "learning_rate": 5.924600027203194e-06,
+ "loss": 0.0261,
+ "step": 23657
+ },
+ {
+ "epoch": 64.46321525885558,
+ "grad_norm": 2.1607673168182373,
+ "learning_rate": 5.923794160129064e-06,
+ "loss": 0.0279,
+ "step": 23658
+ },
+ {
+ "epoch": 64.46594005449592,
+ "grad_norm": 1.140751600265503,
+ "learning_rate": 5.92298832480029e-06,
+ "loss": 0.0142,
+ "step": 23659
+ },
+ {
+ "epoch": 64.46866485013624,
+ "grad_norm": 2.008544921875,
+ "learning_rate": 5.9221825212231545e-06,
+ "loss": 0.0646,
+ "step": 23660
+ },
+ {
+ "epoch": 64.47138964577657,
+ "grad_norm": 2.7924435138702393,
+ "learning_rate": 5.921376749403928e-06,
+ "loss": 0.0838,
+ "step": 23661
+ },
+ {
+ "epoch": 64.47411444141689,
+ "grad_norm": 2.5832531452178955,
+ "learning_rate": 5.92057100934889e-06,
+ "loss": 0.0282,
+ "step": 23662
+ },
+ {
+ "epoch": 64.47683923705722,
+ "grad_norm": 1.9846882820129395,
+ "learning_rate": 5.919765301064311e-06,
+ "loss": 0.0236,
+ "step": 23663
+ },
+ {
+ "epoch": 64.47956403269755,
+ "grad_norm": 2.619896411895752,
+ "learning_rate": 5.9189596245564715e-06,
+ "loss": 0.0454,
+ "step": 23664
+ },
+ {
+ "epoch": 64.48228882833787,
+ "grad_norm": 2.225499153137207,
+ "learning_rate": 5.918153979831641e-06,
+ "loss": 0.0215,
+ "step": 23665
+ },
+ {
+ "epoch": 64.4850136239782,
+ "grad_norm": 3.059386730194092,
+ "learning_rate": 5.917348366896099e-06,
+ "loss": 0.033,
+ "step": 23666
+ },
+ {
+ "epoch": 64.48773841961852,
+ "grad_norm": 0.971503734588623,
+ "learning_rate": 5.9165427857561096e-06,
+ "loss": 0.0108,
+ "step": 23667
+ },
+ {
+ "epoch": 64.49046321525886,
+ "grad_norm": 1.2835911512374878,
+ "learning_rate": 5.915737236417957e-06,
+ "loss": 0.0173,
+ "step": 23668
+ },
+ {
+ "epoch": 64.49318801089919,
+ "grad_norm": 1.8303158283233643,
+ "learning_rate": 5.914931718887915e-06,
+ "loss": 0.0429,
+ "step": 23669
+ },
+ {
+ "epoch": 64.49591280653951,
+ "grad_norm": 1.9778597354888916,
+ "learning_rate": 5.9141262331722505e-06,
+ "loss": 0.0242,
+ "step": 23670
+ },
+ {
+ "epoch": 64.49863760217984,
+ "grad_norm": 2.249831438064575,
+ "learning_rate": 5.913320779277239e-06,
+ "loss": 0.0269,
+ "step": 23671
+ },
+ {
+ "epoch": 64.50136239782016,
+ "grad_norm": 1.961737036705017,
+ "learning_rate": 5.912515357209154e-06,
+ "loss": 0.0366,
+ "step": 23672
+ },
+ {
+ "epoch": 64.50408719346049,
+ "grad_norm": 2.126152992248535,
+ "learning_rate": 5.911709966974269e-06,
+ "loss": 0.0577,
+ "step": 23673
+ },
+ {
+ "epoch": 64.50681198910081,
+ "grad_norm": 1.812860131263733,
+ "learning_rate": 5.910904608578855e-06,
+ "loss": 0.0302,
+ "step": 23674
+ },
+ {
+ "epoch": 64.50953678474114,
+ "grad_norm": 1.6512278318405151,
+ "learning_rate": 5.910099282029182e-06,
+ "loss": 0.1026,
+ "step": 23675
+ },
+ {
+ "epoch": 64.51226158038148,
+ "grad_norm": 1.7046626806259155,
+ "learning_rate": 5.909293987331528e-06,
+ "loss": 0.0505,
+ "step": 23676
+ },
+ {
+ "epoch": 64.5149863760218,
+ "grad_norm": 2.1445493698120117,
+ "learning_rate": 5.908488724492157e-06,
+ "loss": 0.0288,
+ "step": 23677
+ },
+ {
+ "epoch": 64.51771117166213,
+ "grad_norm": 2.391000509262085,
+ "learning_rate": 5.907683493517348e-06,
+ "loss": 0.0469,
+ "step": 23678
+ },
+ {
+ "epoch": 64.52043596730245,
+ "grad_norm": 2.4789068698883057,
+ "learning_rate": 5.906878294413365e-06,
+ "loss": 0.0585,
+ "step": 23679
+ },
+ {
+ "epoch": 64.52316076294278,
+ "grad_norm": 1.4757999181747437,
+ "learning_rate": 5.906073127186485e-06,
+ "loss": 0.0187,
+ "step": 23680
+ },
+ {
+ "epoch": 64.52588555858311,
+ "grad_norm": 1.849755883216858,
+ "learning_rate": 5.905267991842972e-06,
+ "loss": 0.0276,
+ "step": 23681
+ },
+ {
+ "epoch": 64.52861035422343,
+ "grad_norm": 2.1404366493225098,
+ "learning_rate": 5.904462888389104e-06,
+ "loss": 0.0687,
+ "step": 23682
+ },
+ {
+ "epoch": 64.53133514986376,
+ "grad_norm": 1.9287784099578857,
+ "learning_rate": 5.903657816831143e-06,
+ "loss": 0.0193,
+ "step": 23683
+ },
+ {
+ "epoch": 64.53405994550408,
+ "grad_norm": 1.8980143070220947,
+ "learning_rate": 5.902852777175363e-06,
+ "loss": 0.0333,
+ "step": 23684
+ },
+ {
+ "epoch": 64.53678474114442,
+ "grad_norm": 2.0413107872009277,
+ "learning_rate": 5.9020477694280375e-06,
+ "loss": 0.02,
+ "step": 23685
+ },
+ {
+ "epoch": 64.53950953678473,
+ "grad_norm": 2.490272045135498,
+ "learning_rate": 5.9012427935954295e-06,
+ "loss": 0.0903,
+ "step": 23686
+ },
+ {
+ "epoch": 64.54223433242507,
+ "grad_norm": 1.670483946800232,
+ "learning_rate": 5.900437849683813e-06,
+ "loss": 0.0322,
+ "step": 23687
+ },
+ {
+ "epoch": 64.5449591280654,
+ "grad_norm": 1.525977611541748,
+ "learning_rate": 5.899632937699453e-06,
+ "loss": 0.0357,
+ "step": 23688
+ },
+ {
+ "epoch": 64.54768392370572,
+ "grad_norm": 1.7357932329177856,
+ "learning_rate": 5.89882805764862e-06,
+ "loss": 0.0279,
+ "step": 23689
+ },
+ {
+ "epoch": 64.55040871934605,
+ "grad_norm": 1.655776858329773,
+ "learning_rate": 5.898023209537582e-06,
+ "loss": 0.0377,
+ "step": 23690
+ },
+ {
+ "epoch": 64.55313351498637,
+ "grad_norm": 1.9854955673217773,
+ "learning_rate": 5.897218393372609e-06,
+ "loss": 0.0352,
+ "step": 23691
+ },
+ {
+ "epoch": 64.5558583106267,
+ "grad_norm": 1.7698122262954712,
+ "learning_rate": 5.896413609159964e-06,
+ "loss": 0.0254,
+ "step": 23692
+ },
+ {
+ "epoch": 64.55858310626704,
+ "grad_norm": 1.682753086090088,
+ "learning_rate": 5.8956088569059185e-06,
+ "loss": 0.0414,
+ "step": 23693
+ },
+ {
+ "epoch": 64.56130790190736,
+ "grad_norm": 1.3539310693740845,
+ "learning_rate": 5.89480413661674e-06,
+ "loss": 0.0134,
+ "step": 23694
+ },
+ {
+ "epoch": 64.56403269754769,
+ "grad_norm": 3.1410937309265137,
+ "learning_rate": 5.893999448298695e-06,
+ "loss": 0.0558,
+ "step": 23695
+ },
+ {
+ "epoch": 64.566757493188,
+ "grad_norm": 1.5123733282089233,
+ "learning_rate": 5.893194791958051e-06,
+ "loss": 0.0206,
+ "step": 23696
+ },
+ {
+ "epoch": 64.56948228882834,
+ "grad_norm": 1.672925353050232,
+ "learning_rate": 5.892390167601072e-06,
+ "loss": 0.0338,
+ "step": 23697
+ },
+ {
+ "epoch": 64.57220708446866,
+ "grad_norm": 2.1117119789123535,
+ "learning_rate": 5.891585575234028e-06,
+ "loss": 0.0284,
+ "step": 23698
+ },
+ {
+ "epoch": 64.57493188010899,
+ "grad_norm": 1.9187448024749756,
+ "learning_rate": 5.89078101486318e-06,
+ "loss": 0.03,
+ "step": 23699
+ },
+ {
+ "epoch": 64.57765667574932,
+ "grad_norm": 2.203547716140747,
+ "learning_rate": 5.889976486494802e-06,
+ "loss": 0.0288,
+ "step": 23700
+ },
+ {
+ "epoch": 64.58038147138964,
+ "grad_norm": 1.5611618757247925,
+ "learning_rate": 5.889171990135149e-06,
+ "loss": 0.0183,
+ "step": 23701
+ },
+ {
+ "epoch": 64.58310626702998,
+ "grad_norm": 1.7493621110916138,
+ "learning_rate": 5.8883675257904936e-06,
+ "loss": 0.0182,
+ "step": 23702
+ },
+ {
+ "epoch": 64.5858310626703,
+ "grad_norm": 3.565206289291382,
+ "learning_rate": 5.887563093467103e-06,
+ "loss": 0.11,
+ "step": 23703
+ },
+ {
+ "epoch": 64.58855585831063,
+ "grad_norm": 1.433695912361145,
+ "learning_rate": 5.886758693171236e-06,
+ "loss": 0.0174,
+ "step": 23704
+ },
+ {
+ "epoch": 64.59128065395096,
+ "grad_norm": 2.3134992122650146,
+ "learning_rate": 5.885954324909162e-06,
+ "loss": 0.0198,
+ "step": 23705
+ },
+ {
+ "epoch": 64.59400544959128,
+ "grad_norm": 1.8307361602783203,
+ "learning_rate": 5.885149988687141e-06,
+ "loss": 0.0708,
+ "step": 23706
+ },
+ {
+ "epoch": 64.59673024523161,
+ "grad_norm": 1.6707000732421875,
+ "learning_rate": 5.8843456845114426e-06,
+ "loss": 0.0243,
+ "step": 23707
+ },
+ {
+ "epoch": 64.59945504087193,
+ "grad_norm": 1.5438371896743774,
+ "learning_rate": 5.883541412388325e-06,
+ "loss": 0.0159,
+ "step": 23708
+ },
+ {
+ "epoch": 64.60217983651226,
+ "grad_norm": 2.1374988555908203,
+ "learning_rate": 5.8827371723240555e-06,
+ "loss": 0.0297,
+ "step": 23709
+ },
+ {
+ "epoch": 64.60490463215258,
+ "grad_norm": 1.6313316822052002,
+ "learning_rate": 5.881932964324895e-06,
+ "loss": 0.0249,
+ "step": 23710
+ },
+ {
+ "epoch": 64.60762942779292,
+ "grad_norm": 1.380707025527954,
+ "learning_rate": 5.88112878839711e-06,
+ "loss": 0.0181,
+ "step": 23711
+ },
+ {
+ "epoch": 64.61035422343325,
+ "grad_norm": 1.7183849811553955,
+ "learning_rate": 5.88032464454696e-06,
+ "loss": 0.0269,
+ "step": 23712
+ },
+ {
+ "epoch": 64.61307901907357,
+ "grad_norm": 2.0367932319641113,
+ "learning_rate": 5.879520532780709e-06,
+ "loss": 0.087,
+ "step": 23713
+ },
+ {
+ "epoch": 64.6158038147139,
+ "grad_norm": 1.7261096239089966,
+ "learning_rate": 5.87871645310462e-06,
+ "loss": 0.0318,
+ "step": 23714
+ },
+ {
+ "epoch": 64.61852861035422,
+ "grad_norm": 1.4703251123428345,
+ "learning_rate": 5.877912405524954e-06,
+ "loss": 0.0135,
+ "step": 23715
+ },
+ {
+ "epoch": 64.62125340599455,
+ "grad_norm": 1.9311970472335815,
+ "learning_rate": 5.8771083900479744e-06,
+ "loss": 0.0792,
+ "step": 23716
+ },
+ {
+ "epoch": 64.62397820163488,
+ "grad_norm": 2.3567795753479004,
+ "learning_rate": 5.876304406679941e-06,
+ "loss": 0.0252,
+ "step": 23717
+ },
+ {
+ "epoch": 64.6267029972752,
+ "grad_norm": 2.0078811645507812,
+ "learning_rate": 5.875500455427119e-06,
+ "loss": 0.0463,
+ "step": 23718
+ },
+ {
+ "epoch": 64.62942779291554,
+ "grad_norm": 2.7085094451904297,
+ "learning_rate": 5.874696536295761e-06,
+ "loss": 0.0367,
+ "step": 23719
+ },
+ {
+ "epoch": 64.63215258855585,
+ "grad_norm": 1.7971662282943726,
+ "learning_rate": 5.873892649292136e-06,
+ "loss": 0.023,
+ "step": 23720
+ },
+ {
+ "epoch": 64.63487738419619,
+ "grad_norm": 1.6812915802001953,
+ "learning_rate": 5.873088794422505e-06,
+ "loss": 0.0181,
+ "step": 23721
+ },
+ {
+ "epoch": 64.6376021798365,
+ "grad_norm": 2.0677359104156494,
+ "learning_rate": 5.872284971693123e-06,
+ "loss": 0.0292,
+ "step": 23722
+ },
+ {
+ "epoch": 64.64032697547684,
+ "grad_norm": 2.3314449787139893,
+ "learning_rate": 5.871481181110255e-06,
+ "loss": 0.0486,
+ "step": 23723
+ },
+ {
+ "epoch": 64.64305177111717,
+ "grad_norm": 1.7804057598114014,
+ "learning_rate": 5.870677422680156e-06,
+ "loss": 0.0193,
+ "step": 23724
+ },
+ {
+ "epoch": 64.64577656675749,
+ "grad_norm": 1.7453396320343018,
+ "learning_rate": 5.8698736964090904e-06,
+ "loss": 0.0344,
+ "step": 23725
+ },
+ {
+ "epoch": 64.64850136239782,
+ "grad_norm": 2.462930679321289,
+ "learning_rate": 5.8690700023033145e-06,
+ "loss": 0.1395,
+ "step": 23726
+ },
+ {
+ "epoch": 64.65122615803814,
+ "grad_norm": 1.716685175895691,
+ "learning_rate": 5.86826634036909e-06,
+ "loss": 0.0182,
+ "step": 23727
+ },
+ {
+ "epoch": 64.65395095367847,
+ "grad_norm": 2.973278760910034,
+ "learning_rate": 5.867462710612672e-06,
+ "loss": 0.0334,
+ "step": 23728
+ },
+ {
+ "epoch": 64.65667574931881,
+ "grad_norm": 2.418403148651123,
+ "learning_rate": 5.866659113040324e-06,
+ "loss": 0.1755,
+ "step": 23729
+ },
+ {
+ "epoch": 64.65940054495913,
+ "grad_norm": 1.842740774154663,
+ "learning_rate": 5.8658555476582994e-06,
+ "loss": 0.1206,
+ "step": 23730
+ },
+ {
+ "epoch": 64.66212534059946,
+ "grad_norm": 1.8564093112945557,
+ "learning_rate": 5.865052014472861e-06,
+ "loss": 0.0221,
+ "step": 23731
+ },
+ {
+ "epoch": 64.66485013623978,
+ "grad_norm": 2.4129104614257812,
+ "learning_rate": 5.8642485134902625e-06,
+ "loss": 0.0277,
+ "step": 23732
+ },
+ {
+ "epoch": 64.66757493188011,
+ "grad_norm": 1.2615585327148438,
+ "learning_rate": 5.863445044716764e-06,
+ "loss": 0.0151,
+ "step": 23733
+ },
+ {
+ "epoch": 64.67029972752043,
+ "grad_norm": 1.2207324504852295,
+ "learning_rate": 5.862641608158623e-06,
+ "loss": 0.0128,
+ "step": 23734
+ },
+ {
+ "epoch": 64.67302452316076,
+ "grad_norm": 1.735542893409729,
+ "learning_rate": 5.861838203822096e-06,
+ "loss": 0.0226,
+ "step": 23735
+ },
+ {
+ "epoch": 64.6757493188011,
+ "grad_norm": 2.4135777950286865,
+ "learning_rate": 5.861034831713435e-06,
+ "loss": 0.0403,
+ "step": 23736
+ },
+ {
+ "epoch": 64.67847411444141,
+ "grad_norm": 1.2518620491027832,
+ "learning_rate": 5.860231491838906e-06,
+ "loss": 0.0292,
+ "step": 23737
+ },
+ {
+ "epoch": 64.68119891008175,
+ "grad_norm": 2.250140428543091,
+ "learning_rate": 5.859428184204761e-06,
+ "loss": 0.1256,
+ "step": 23738
+ },
+ {
+ "epoch": 64.68392370572207,
+ "grad_norm": 2.0681920051574707,
+ "learning_rate": 5.858624908817257e-06,
+ "loss": 0.0373,
+ "step": 23739
+ },
+ {
+ "epoch": 64.6866485013624,
+ "grad_norm": 2.6802587509155273,
+ "learning_rate": 5.857821665682646e-06,
+ "loss": 0.0717,
+ "step": 23740
+ },
+ {
+ "epoch": 64.68937329700273,
+ "grad_norm": 1.4602577686309814,
+ "learning_rate": 5.857018454807189e-06,
+ "loss": 0.0143,
+ "step": 23741
+ },
+ {
+ "epoch": 64.69209809264305,
+ "grad_norm": 1.5165417194366455,
+ "learning_rate": 5.856215276197137e-06,
+ "loss": 0.0327,
+ "step": 23742
+ },
+ {
+ "epoch": 64.69482288828338,
+ "grad_norm": 3.2763569355010986,
+ "learning_rate": 5.855412129858749e-06,
+ "loss": 0.0735,
+ "step": 23743
+ },
+ {
+ "epoch": 64.6975476839237,
+ "grad_norm": 1.565470814704895,
+ "learning_rate": 5.854609015798277e-06,
+ "loss": 0.0198,
+ "step": 23744
+ },
+ {
+ "epoch": 64.70027247956403,
+ "grad_norm": 2.4342293739318848,
+ "learning_rate": 5.8538059340219774e-06,
+ "loss": 0.0365,
+ "step": 23745
+ },
+ {
+ "epoch": 64.70299727520435,
+ "grad_norm": 2.7241082191467285,
+ "learning_rate": 5.853002884536103e-06,
+ "loss": 0.0462,
+ "step": 23746
+ },
+ {
+ "epoch": 64.70572207084469,
+ "grad_norm": 1.84480881690979,
+ "learning_rate": 5.852199867346909e-06,
+ "loss": 0.0601,
+ "step": 23747
+ },
+ {
+ "epoch": 64.70844686648502,
+ "grad_norm": 2.498885154724121,
+ "learning_rate": 5.851396882460648e-06,
+ "loss": 0.1402,
+ "step": 23748
+ },
+ {
+ "epoch": 64.71117166212534,
+ "grad_norm": 2.2449798583984375,
+ "learning_rate": 5.850593929883577e-06,
+ "loss": 0.0366,
+ "step": 23749
+ },
+ {
+ "epoch": 64.71389645776567,
+ "grad_norm": 2.5145106315612793,
+ "learning_rate": 5.8497910096219435e-06,
+ "loss": 0.0821,
+ "step": 23750
+ },
+ {
+ "epoch": 64.71662125340599,
+ "grad_norm": 1.5223854780197144,
+ "learning_rate": 5.848988121682006e-06,
+ "loss": 0.0229,
+ "step": 23751
+ },
+ {
+ "epoch": 64.71934604904632,
+ "grad_norm": 2.069664478302002,
+ "learning_rate": 5.848185266070016e-06,
+ "loss": 0.0501,
+ "step": 23752
+ },
+ {
+ "epoch": 64.72207084468666,
+ "grad_norm": 1.802673101425171,
+ "learning_rate": 5.847382442792222e-06,
+ "loss": 0.0225,
+ "step": 23753
+ },
+ {
+ "epoch": 64.72479564032697,
+ "grad_norm": 1.8798763751983643,
+ "learning_rate": 5.8465796518548824e-06,
+ "loss": 0.024,
+ "step": 23754
+ },
+ {
+ "epoch": 64.7275204359673,
+ "grad_norm": 1.7579302787780762,
+ "learning_rate": 5.845776893264247e-06,
+ "loss": 0.0787,
+ "step": 23755
+ },
+ {
+ "epoch": 64.73024523160763,
+ "grad_norm": 2.193053722381592,
+ "learning_rate": 5.844974167026567e-06,
+ "loss": 0.0766,
+ "step": 23756
+ },
+ {
+ "epoch": 64.73297002724796,
+ "grad_norm": 2.689924716949463,
+ "learning_rate": 5.8441714731480944e-06,
+ "loss": 0.1262,
+ "step": 23757
+ },
+ {
+ "epoch": 64.73569482288828,
+ "grad_norm": 2.3120534420013428,
+ "learning_rate": 5.843368811635081e-06,
+ "loss": 0.0198,
+ "step": 23758
+ },
+ {
+ "epoch": 64.73841961852861,
+ "grad_norm": 1.7465131282806396,
+ "learning_rate": 5.842566182493779e-06,
+ "loss": 0.1048,
+ "step": 23759
+ },
+ {
+ "epoch": 64.74114441416894,
+ "grad_norm": 2.5709683895111084,
+ "learning_rate": 5.8417635857304355e-06,
+ "loss": 0.1684,
+ "step": 23760
+ },
+ {
+ "epoch": 64.74386920980926,
+ "grad_norm": 2.8062164783477783,
+ "learning_rate": 5.840961021351305e-06,
+ "loss": 0.0993,
+ "step": 23761
+ },
+ {
+ "epoch": 64.7465940054496,
+ "grad_norm": 1.5783640146255493,
+ "learning_rate": 5.840158489362635e-06,
+ "loss": 0.0263,
+ "step": 23762
+ },
+ {
+ "epoch": 64.74931880108991,
+ "grad_norm": 2.3834826946258545,
+ "learning_rate": 5.839355989770678e-06,
+ "loss": 0.0468,
+ "step": 23763
+ },
+ {
+ "epoch": 64.75204359673025,
+ "grad_norm": 1.9882630109786987,
+ "learning_rate": 5.8385535225816805e-06,
+ "loss": 0.038,
+ "step": 23764
+ },
+ {
+ "epoch": 64.75476839237058,
+ "grad_norm": 2.016876459121704,
+ "learning_rate": 5.8377510878018975e-06,
+ "loss": 0.0258,
+ "step": 23765
+ },
+ {
+ "epoch": 64.7574931880109,
+ "grad_norm": 1.9203394651412964,
+ "learning_rate": 5.836948685437573e-06,
+ "loss": 0.0193,
+ "step": 23766
+ },
+ {
+ "epoch": 64.76021798365123,
+ "grad_norm": 1.6346347332000732,
+ "learning_rate": 5.836146315494956e-06,
+ "loss": 0.0262,
+ "step": 23767
+ },
+ {
+ "epoch": 64.76294277929155,
+ "grad_norm": 3.0076794624328613,
+ "learning_rate": 5.835343977980301e-06,
+ "loss": 0.1136,
+ "step": 23768
+ },
+ {
+ "epoch": 64.76566757493188,
+ "grad_norm": 2.827022075653076,
+ "learning_rate": 5.834541672899852e-06,
+ "loss": 0.0401,
+ "step": 23769
+ },
+ {
+ "epoch": 64.7683923705722,
+ "grad_norm": 1.2868648767471313,
+ "learning_rate": 5.8337394002598565e-06,
+ "loss": 0.0133,
+ "step": 23770
+ },
+ {
+ "epoch": 64.77111716621253,
+ "grad_norm": 2.1217095851898193,
+ "learning_rate": 5.832937160066567e-06,
+ "loss": 0.1075,
+ "step": 23771
+ },
+ {
+ "epoch": 64.77384196185287,
+ "grad_norm": 1.5550261735916138,
+ "learning_rate": 5.832134952326225e-06,
+ "loss": 0.0295,
+ "step": 23772
+ },
+ {
+ "epoch": 64.77656675749319,
+ "grad_norm": 8.64826774597168,
+ "learning_rate": 5.831332777045087e-06,
+ "loss": 0.0204,
+ "step": 23773
+ },
+ {
+ "epoch": 64.77929155313352,
+ "grad_norm": 1.760615348815918,
+ "learning_rate": 5.830530634229393e-06,
+ "loss": 0.039,
+ "step": 23774
+ },
+ {
+ "epoch": 64.78201634877384,
+ "grad_norm": 3.378669500350952,
+ "learning_rate": 5.829728523885394e-06,
+ "loss": 0.1068,
+ "step": 23775
+ },
+ {
+ "epoch": 64.78474114441417,
+ "grad_norm": 1.834916114807129,
+ "learning_rate": 5.82892644601933e-06,
+ "loss": 0.0183,
+ "step": 23776
+ },
+ {
+ "epoch": 64.7874659400545,
+ "grad_norm": 2.084787130355835,
+ "learning_rate": 5.828124400637456e-06,
+ "loss": 0.02,
+ "step": 23777
+ },
+ {
+ "epoch": 64.79019073569482,
+ "grad_norm": 1.854650616645813,
+ "learning_rate": 5.827322387746018e-06,
+ "loss": 0.0783,
+ "step": 23778
+ },
+ {
+ "epoch": 64.79291553133515,
+ "grad_norm": 2.7811479568481445,
+ "learning_rate": 5.826520407351252e-06,
+ "loss": 0.0223,
+ "step": 23779
+ },
+ {
+ "epoch": 64.79564032697547,
+ "grad_norm": 2.1911661624908447,
+ "learning_rate": 5.8257184594594155e-06,
+ "loss": 0.037,
+ "step": 23780
+ },
+ {
+ "epoch": 64.7983651226158,
+ "grad_norm": 2.110008478164673,
+ "learning_rate": 5.824916544076749e-06,
+ "loss": 0.0257,
+ "step": 23781
+ },
+ {
+ "epoch": 64.80108991825612,
+ "grad_norm": 2.1072657108306885,
+ "learning_rate": 5.824114661209498e-06,
+ "loss": 0.1411,
+ "step": 23782
+ },
+ {
+ "epoch": 64.80381471389646,
+ "grad_norm": 2.870584726333618,
+ "learning_rate": 5.823312810863904e-06,
+ "loss": 0.0282,
+ "step": 23783
+ },
+ {
+ "epoch": 64.80653950953679,
+ "grad_norm": 7.657654285430908,
+ "learning_rate": 5.82251099304622e-06,
+ "loss": 0.0373,
+ "step": 23784
+ },
+ {
+ "epoch": 64.80926430517711,
+ "grad_norm": 1.5143541097640991,
+ "learning_rate": 5.821709207762684e-06,
+ "loss": 0.0264,
+ "step": 23785
+ },
+ {
+ "epoch": 64.81198910081744,
+ "grad_norm": 1.5931376218795776,
+ "learning_rate": 5.8209074550195445e-06,
+ "loss": 0.0239,
+ "step": 23786
+ },
+ {
+ "epoch": 64.81471389645776,
+ "grad_norm": 1.2761014699935913,
+ "learning_rate": 5.8201057348230384e-06,
+ "loss": 0.0856,
+ "step": 23787
+ },
+ {
+ "epoch": 64.8174386920981,
+ "grad_norm": 2.941314697265625,
+ "learning_rate": 5.819304047179415e-06,
+ "loss": 0.0939,
+ "step": 23788
+ },
+ {
+ "epoch": 64.82016348773843,
+ "grad_norm": 1.675869345664978,
+ "learning_rate": 5.81850239209492e-06,
+ "loss": 0.029,
+ "step": 23789
+ },
+ {
+ "epoch": 64.82288828337875,
+ "grad_norm": 1.803663730621338,
+ "learning_rate": 5.8177007695757935e-06,
+ "loss": 0.0174,
+ "step": 23790
+ },
+ {
+ "epoch": 64.82561307901908,
+ "grad_norm": 1.4759974479675293,
+ "learning_rate": 5.816899179628279e-06,
+ "loss": 0.0224,
+ "step": 23791
+ },
+ {
+ "epoch": 64.8283378746594,
+ "grad_norm": 1.7367501258850098,
+ "learning_rate": 5.816097622258616e-06,
+ "loss": 0.0381,
+ "step": 23792
+ },
+ {
+ "epoch": 64.83106267029973,
+ "grad_norm": 1.6400959491729736,
+ "learning_rate": 5.815296097473053e-06,
+ "loss": 0.0413,
+ "step": 23793
+ },
+ {
+ "epoch": 64.83378746594005,
+ "grad_norm": 1.5837897062301636,
+ "learning_rate": 5.814494605277831e-06,
+ "loss": 0.0493,
+ "step": 23794
+ },
+ {
+ "epoch": 64.83651226158038,
+ "grad_norm": 4.521369457244873,
+ "learning_rate": 5.813693145679189e-06,
+ "loss": 0.036,
+ "step": 23795
+ },
+ {
+ "epoch": 64.83923705722071,
+ "grad_norm": 2.034454584121704,
+ "learning_rate": 5.8128917186833655e-06,
+ "loss": 0.0548,
+ "step": 23796
+ },
+ {
+ "epoch": 64.84196185286103,
+ "grad_norm": 2.0204052925109863,
+ "learning_rate": 5.812090324296611e-06,
+ "loss": 0.0339,
+ "step": 23797
+ },
+ {
+ "epoch": 64.84468664850137,
+ "grad_norm": 2.4671874046325684,
+ "learning_rate": 5.811288962525163e-06,
+ "loss": 0.0163,
+ "step": 23798
+ },
+ {
+ "epoch": 64.84741144414168,
+ "grad_norm": 2.13287091255188,
+ "learning_rate": 5.810487633375261e-06,
+ "loss": 0.0423,
+ "step": 23799
+ },
+ {
+ "epoch": 64.85013623978202,
+ "grad_norm": 2.203026294708252,
+ "learning_rate": 5.8096863368531435e-06,
+ "loss": 0.0494,
+ "step": 23800
+ },
+ {
+ "epoch": 64.85286103542235,
+ "grad_norm": 1.3614833354949951,
+ "learning_rate": 5.808885072965057e-06,
+ "loss": 0.0207,
+ "step": 23801
+ },
+ {
+ "epoch": 64.85558583106267,
+ "grad_norm": 2.3738698959350586,
+ "learning_rate": 5.808083841717238e-06,
+ "loss": 0.0364,
+ "step": 23802
+ },
+ {
+ "epoch": 64.858310626703,
+ "grad_norm": 1.9528080224990845,
+ "learning_rate": 5.807282643115924e-06,
+ "loss": 0.0286,
+ "step": 23803
+ },
+ {
+ "epoch": 64.86103542234332,
+ "grad_norm": 1.826361894607544,
+ "learning_rate": 5.8064814771673625e-06,
+ "loss": 0.0293,
+ "step": 23804
+ },
+ {
+ "epoch": 64.86376021798365,
+ "grad_norm": 1.5348117351531982,
+ "learning_rate": 5.805680343877783e-06,
+ "loss": 0.1172,
+ "step": 23805
+ },
+ {
+ "epoch": 64.86648501362397,
+ "grad_norm": 2.0364701747894287,
+ "learning_rate": 5.804879243253435e-06,
+ "loss": 0.0313,
+ "step": 23806
+ },
+ {
+ "epoch": 64.8692098092643,
+ "grad_norm": 1.5615718364715576,
+ "learning_rate": 5.804078175300551e-06,
+ "loss": 0.0301,
+ "step": 23807
+ },
+ {
+ "epoch": 64.87193460490464,
+ "grad_norm": 2.2209653854370117,
+ "learning_rate": 5.80327714002537e-06,
+ "loss": 0.1309,
+ "step": 23808
+ },
+ {
+ "epoch": 64.87465940054496,
+ "grad_norm": 2.006291627883911,
+ "learning_rate": 5.802476137434132e-06,
+ "loss": 0.0328,
+ "step": 23809
+ },
+ {
+ "epoch": 64.87738419618529,
+ "grad_norm": 2.2455086708068848,
+ "learning_rate": 5.801675167533078e-06,
+ "loss": 0.0215,
+ "step": 23810
+ },
+ {
+ "epoch": 64.88010899182561,
+ "grad_norm": 2.096162796020508,
+ "learning_rate": 5.80087423032844e-06,
+ "loss": 0.0171,
+ "step": 23811
+ },
+ {
+ "epoch": 64.88283378746594,
+ "grad_norm": 1.6564465761184692,
+ "learning_rate": 5.800073325826456e-06,
+ "loss": 0.1634,
+ "step": 23812
+ },
+ {
+ "epoch": 64.88555858310627,
+ "grad_norm": 1.2337162494659424,
+ "learning_rate": 5.799272454033369e-06,
+ "loss": 0.0196,
+ "step": 23813
+ },
+ {
+ "epoch": 64.88828337874659,
+ "grad_norm": 3.56858229637146,
+ "learning_rate": 5.7984716149554136e-06,
+ "loss": 0.0612,
+ "step": 23814
+ },
+ {
+ "epoch": 64.89100817438693,
+ "grad_norm": 1.145932912826538,
+ "learning_rate": 5.797670808598826e-06,
+ "loss": 0.0122,
+ "step": 23815
+ },
+ {
+ "epoch": 64.89373297002724,
+ "grad_norm": 1.637919306755066,
+ "learning_rate": 5.796870034969837e-06,
+ "loss": 0.0229,
+ "step": 23816
+ },
+ {
+ "epoch": 64.89645776566758,
+ "grad_norm": 2.3348348140716553,
+ "learning_rate": 5.796069294074696e-06,
+ "loss": 0.1144,
+ "step": 23817
+ },
+ {
+ "epoch": 64.8991825613079,
+ "grad_norm": 1.0543314218521118,
+ "learning_rate": 5.795268585919629e-06,
+ "loss": 0.0143,
+ "step": 23818
+ },
+ {
+ "epoch": 64.90190735694823,
+ "grad_norm": 1.4800496101379395,
+ "learning_rate": 5.794467910510877e-06,
+ "loss": 0.016,
+ "step": 23819
+ },
+ {
+ "epoch": 64.90463215258856,
+ "grad_norm": 1.5355348587036133,
+ "learning_rate": 5.793667267854668e-06,
+ "loss": 0.0754,
+ "step": 23820
+ },
+ {
+ "epoch": 64.90735694822888,
+ "grad_norm": 3.0119311809539795,
+ "learning_rate": 5.792866657957248e-06,
+ "loss": 0.0348,
+ "step": 23821
+ },
+ {
+ "epoch": 64.91008174386921,
+ "grad_norm": 1.4990712404251099,
+ "learning_rate": 5.792066080824843e-06,
+ "loss": 0.0416,
+ "step": 23822
+ },
+ {
+ "epoch": 64.91280653950953,
+ "grad_norm": 1.6069693565368652,
+ "learning_rate": 5.791265536463696e-06,
+ "loss": 0.0515,
+ "step": 23823
+ },
+ {
+ "epoch": 64.91553133514986,
+ "grad_norm": 1.8886858224868774,
+ "learning_rate": 5.790465024880033e-06,
+ "loss": 0.1756,
+ "step": 23824
+ },
+ {
+ "epoch": 64.9182561307902,
+ "grad_norm": 2.531160593032837,
+ "learning_rate": 5.7896645460800986e-06,
+ "loss": 0.0315,
+ "step": 23825
+ },
+ {
+ "epoch": 64.92098092643052,
+ "grad_norm": 1.7963871955871582,
+ "learning_rate": 5.7888641000701195e-06,
+ "loss": 0.0276,
+ "step": 23826
+ },
+ {
+ "epoch": 64.92370572207085,
+ "grad_norm": 1.4964224100112915,
+ "learning_rate": 5.788063686856331e-06,
+ "loss": 0.0411,
+ "step": 23827
+ },
+ {
+ "epoch": 64.92643051771117,
+ "grad_norm": 1.8793976306915283,
+ "learning_rate": 5.7872633064449635e-06,
+ "loss": 0.0561,
+ "step": 23828
+ },
+ {
+ "epoch": 64.9291553133515,
+ "grad_norm": 1.7080104351043701,
+ "learning_rate": 5.786462958842258e-06,
+ "loss": 0.0711,
+ "step": 23829
+ },
+ {
+ "epoch": 64.93188010899182,
+ "grad_norm": 1.6412321329116821,
+ "learning_rate": 5.785662644054444e-06,
+ "loss": 0.0234,
+ "step": 23830
+ },
+ {
+ "epoch": 64.93460490463215,
+ "grad_norm": 1.6887611150741577,
+ "learning_rate": 5.784862362087752e-06,
+ "loss": 0.0194,
+ "step": 23831
+ },
+ {
+ "epoch": 64.93732970027249,
+ "grad_norm": 1.8776289224624634,
+ "learning_rate": 5.7840621129484135e-06,
+ "loss": 0.047,
+ "step": 23832
+ },
+ {
+ "epoch": 64.9400544959128,
+ "grad_norm": 1.6163326501846313,
+ "learning_rate": 5.783261896642667e-06,
+ "loss": 0.0197,
+ "step": 23833
+ },
+ {
+ "epoch": 64.94277929155314,
+ "grad_norm": 1.8777517080307007,
+ "learning_rate": 5.7824617131767415e-06,
+ "loss": 0.055,
+ "step": 23834
+ },
+ {
+ "epoch": 64.94550408719346,
+ "grad_norm": 2.603266954421997,
+ "learning_rate": 5.7816615625568684e-06,
+ "loss": 0.0709,
+ "step": 23835
+ },
+ {
+ "epoch": 64.94822888283379,
+ "grad_norm": 1.871061086654663,
+ "learning_rate": 5.780861444789275e-06,
+ "loss": 0.0892,
+ "step": 23836
+ },
+ {
+ "epoch": 64.95095367847412,
+ "grad_norm": 1.7204787731170654,
+ "learning_rate": 5.780061359880201e-06,
+ "loss": 0.0195,
+ "step": 23837
+ },
+ {
+ "epoch": 64.95367847411444,
+ "grad_norm": 1.5655583143234253,
+ "learning_rate": 5.779261307835873e-06,
+ "loss": 0.0584,
+ "step": 23838
+ },
+ {
+ "epoch": 64.95640326975477,
+ "grad_norm": 1.52986741065979,
+ "learning_rate": 5.778461288662519e-06,
+ "loss": 0.046,
+ "step": 23839
+ },
+ {
+ "epoch": 64.95912806539509,
+ "grad_norm": 1.824897050857544,
+ "learning_rate": 5.777661302366375e-06,
+ "loss": 0.016,
+ "step": 23840
+ },
+ {
+ "epoch": 64.96185286103542,
+ "grad_norm": 2.2865898609161377,
+ "learning_rate": 5.7768613489536664e-06,
+ "loss": 0.1743,
+ "step": 23841
+ },
+ {
+ "epoch": 64.96457765667574,
+ "grad_norm": 1.4752951860427856,
+ "learning_rate": 5.776061428430631e-06,
+ "loss": 0.0176,
+ "step": 23842
+ },
+ {
+ "epoch": 64.96730245231608,
+ "grad_norm": 1.598523497581482,
+ "learning_rate": 5.77526154080349e-06,
+ "loss": 0.0147,
+ "step": 23843
+ },
+ {
+ "epoch": 64.97002724795641,
+ "grad_norm": 1.9758360385894775,
+ "learning_rate": 5.774461686078474e-06,
+ "loss": 0.0373,
+ "step": 23844
+ },
+ {
+ "epoch": 64.97275204359673,
+ "grad_norm": 1.2867381572723389,
+ "learning_rate": 5.773661864261818e-06,
+ "loss": 0.0133,
+ "step": 23845
+ },
+ {
+ "epoch": 64.97547683923706,
+ "grad_norm": 1.7542613744735718,
+ "learning_rate": 5.772862075359748e-06,
+ "loss": 0.0374,
+ "step": 23846
+ },
+ {
+ "epoch": 64.97820163487738,
+ "grad_norm": 1.6937898397445679,
+ "learning_rate": 5.772062319378491e-06,
+ "loss": 0.0377,
+ "step": 23847
+ },
+ {
+ "epoch": 64.98092643051771,
+ "grad_norm": 1.735448956489563,
+ "learning_rate": 5.771262596324273e-06,
+ "loss": 0.1734,
+ "step": 23848
+ },
+ {
+ "epoch": 64.98365122615803,
+ "grad_norm": 1.3433789014816284,
+ "learning_rate": 5.77046290620333e-06,
+ "loss": 0.0147,
+ "step": 23849
+ },
+ {
+ "epoch": 64.98637602179836,
+ "grad_norm": 4.174967288970947,
+ "learning_rate": 5.7696632490218865e-06,
+ "loss": 0.0418,
+ "step": 23850
+ },
+ {
+ "epoch": 64.9891008174387,
+ "grad_norm": 2.1468684673309326,
+ "learning_rate": 5.768863624786167e-06,
+ "loss": 0.046,
+ "step": 23851
+ },
+ {
+ "epoch": 64.99182561307902,
+ "grad_norm": 1.7777776718139648,
+ "learning_rate": 5.768064033502401e-06,
+ "loss": 0.2167,
+ "step": 23852
+ },
+ {
+ "epoch": 64.99455040871935,
+ "grad_norm": 1.2873115539550781,
+ "learning_rate": 5.767264475176817e-06,
+ "loss": 0.013,
+ "step": 23853
+ },
+ {
+ "epoch": 64.99727520435967,
+ "grad_norm": 0.9146227836608887,
+ "learning_rate": 5.766464949815642e-06,
+ "loss": 0.0122,
+ "step": 23854
+ },
+ {
+ "epoch": 65.0,
+ "grad_norm": 1.8148996829986572,
+ "learning_rate": 5.765665457425102e-06,
+ "loss": 0.057,
+ "step": 23855
+ },
+ {
+ "epoch": 65.00272479564033,
+ "grad_norm": 1.8733807802200317,
+ "learning_rate": 5.76486599801142e-06,
+ "loss": 0.0595,
+ "step": 23856
+ },
+ {
+ "epoch": 65.00544959128065,
+ "grad_norm": 2.143559694290161,
+ "learning_rate": 5.7640665715808244e-06,
+ "loss": 0.0205,
+ "step": 23857
+ },
+ {
+ "epoch": 65.00817438692098,
+ "grad_norm": 1.2678637504577637,
+ "learning_rate": 5.763267178139544e-06,
+ "loss": 0.013,
+ "step": 23858
+ },
+ {
+ "epoch": 65.0108991825613,
+ "grad_norm": 1.308029055595398,
+ "learning_rate": 5.7624678176938055e-06,
+ "loss": 0.0215,
+ "step": 23859
+ },
+ {
+ "epoch": 65.01362397820164,
+ "grad_norm": 1.645445466041565,
+ "learning_rate": 5.76166849024983e-06,
+ "loss": 0.0176,
+ "step": 23860
+ },
+ {
+ "epoch": 65.01634877384195,
+ "grad_norm": 1.2985948324203491,
+ "learning_rate": 5.760869195813841e-06,
+ "loss": 0.0268,
+ "step": 23861
+ },
+ {
+ "epoch": 65.01907356948229,
+ "grad_norm": 0.7143044471740723,
+ "learning_rate": 5.760069934392068e-06,
+ "loss": 0.0097,
+ "step": 23862
+ },
+ {
+ "epoch": 65.02179836512262,
+ "grad_norm": 2.019139528274536,
+ "learning_rate": 5.759270705990736e-06,
+ "loss": 0.1143,
+ "step": 23863
+ },
+ {
+ "epoch": 65.02452316076294,
+ "grad_norm": 2.4412355422973633,
+ "learning_rate": 5.758471510616066e-06,
+ "loss": 0.0393,
+ "step": 23864
+ },
+ {
+ "epoch": 65.02724795640327,
+ "grad_norm": 4.84037971496582,
+ "learning_rate": 5.75767234827428e-06,
+ "loss": 0.0792,
+ "step": 23865
+ },
+ {
+ "epoch": 65.02997275204359,
+ "grad_norm": 1.7308943271636963,
+ "learning_rate": 5.756873218971609e-06,
+ "loss": 0.0397,
+ "step": 23866
+ },
+ {
+ "epoch": 65.03269754768392,
+ "grad_norm": 2.0557522773742676,
+ "learning_rate": 5.756074122714272e-06,
+ "loss": 0.0346,
+ "step": 23867
+ },
+ {
+ "epoch": 65.03542234332426,
+ "grad_norm": 1.7879467010498047,
+ "learning_rate": 5.755275059508492e-06,
+ "loss": 0.0355,
+ "step": 23868
+ },
+ {
+ "epoch": 65.03814713896458,
+ "grad_norm": 1.4314184188842773,
+ "learning_rate": 5.754476029360494e-06,
+ "loss": 0.0167,
+ "step": 23869
+ },
+ {
+ "epoch": 65.04087193460491,
+ "grad_norm": 1.514151692390442,
+ "learning_rate": 5.753677032276501e-06,
+ "loss": 0.0089,
+ "step": 23870
+ },
+ {
+ "epoch": 65.04359673024523,
+ "grad_norm": 1.9462950229644775,
+ "learning_rate": 5.752878068262735e-06,
+ "loss": 0.02,
+ "step": 23871
+ },
+ {
+ "epoch": 65.04632152588556,
+ "grad_norm": 2.0544252395629883,
+ "learning_rate": 5.7520791373254135e-06,
+ "loss": 0.1082,
+ "step": 23872
+ },
+ {
+ "epoch": 65.04904632152588,
+ "grad_norm": 1.7667289972305298,
+ "learning_rate": 5.751280239470768e-06,
+ "loss": 0.0682,
+ "step": 23873
+ },
+ {
+ "epoch": 65.05177111716621,
+ "grad_norm": 1.4543218612670898,
+ "learning_rate": 5.750481374705011e-06,
+ "loss": 0.0282,
+ "step": 23874
+ },
+ {
+ "epoch": 65.05449591280654,
+ "grad_norm": 1.2878077030181885,
+ "learning_rate": 5.749682543034371e-06,
+ "loss": 0.0291,
+ "step": 23875
+ },
+ {
+ "epoch": 65.05722070844686,
+ "grad_norm": 1.597861647605896,
+ "learning_rate": 5.748883744465067e-06,
+ "loss": 0.0366,
+ "step": 23876
+ },
+ {
+ "epoch": 65.0599455040872,
+ "grad_norm": 1.4381829500198364,
+ "learning_rate": 5.748084979003317e-06,
+ "loss": 0.0299,
+ "step": 23877
+ },
+ {
+ "epoch": 65.06267029972751,
+ "grad_norm": 1.4160534143447876,
+ "learning_rate": 5.7472862466553455e-06,
+ "loss": 0.0188,
+ "step": 23878
+ },
+ {
+ "epoch": 65.06539509536785,
+ "grad_norm": 1.6496524810791016,
+ "learning_rate": 5.746487547427375e-06,
+ "loss": 0.0192,
+ "step": 23879
+ },
+ {
+ "epoch": 65.06811989100818,
+ "grad_norm": 1.8824506998062134,
+ "learning_rate": 5.74568888132562e-06,
+ "loss": 0.0292,
+ "step": 23880
+ },
+ {
+ "epoch": 65.0708446866485,
+ "grad_norm": 1.5965312719345093,
+ "learning_rate": 5.744890248356301e-06,
+ "loss": 0.112,
+ "step": 23881
+ },
+ {
+ "epoch": 65.07356948228883,
+ "grad_norm": 1.2588415145874023,
+ "learning_rate": 5.744091648525642e-06,
+ "loss": 0.0971,
+ "step": 23882
+ },
+ {
+ "epoch": 65.07629427792915,
+ "grad_norm": 1.9246535301208496,
+ "learning_rate": 5.743293081839862e-06,
+ "loss": 0.1013,
+ "step": 23883
+ },
+ {
+ "epoch": 65.07901907356948,
+ "grad_norm": 1.5566660165786743,
+ "learning_rate": 5.742494548305177e-06,
+ "loss": 0.0285,
+ "step": 23884
+ },
+ {
+ "epoch": 65.0817438692098,
+ "grad_norm": 1.010831356048584,
+ "learning_rate": 5.741696047927805e-06,
+ "loss": 0.0663,
+ "step": 23885
+ },
+ {
+ "epoch": 65.08446866485014,
+ "grad_norm": 2.0028746128082275,
+ "learning_rate": 5.74089758071397e-06,
+ "loss": 0.041,
+ "step": 23886
+ },
+ {
+ "epoch": 65.08719346049047,
+ "grad_norm": 2.213080883026123,
+ "learning_rate": 5.740099146669888e-06,
+ "loss": 0.0478,
+ "step": 23887
+ },
+ {
+ "epoch": 65.08991825613079,
+ "grad_norm": 1.4685055017471313,
+ "learning_rate": 5.739300745801777e-06,
+ "loss": 0.0201,
+ "step": 23888
+ },
+ {
+ "epoch": 65.09264305177112,
+ "grad_norm": 2.0127594470977783,
+ "learning_rate": 5.73850237811585e-06,
+ "loss": 0.0539,
+ "step": 23889
+ },
+ {
+ "epoch": 65.09536784741144,
+ "grad_norm": 1.622306227684021,
+ "learning_rate": 5.737704043618333e-06,
+ "loss": 0.0711,
+ "step": 23890
+ },
+ {
+ "epoch": 65.09809264305177,
+ "grad_norm": 1.8819780349731445,
+ "learning_rate": 5.7369057423154376e-06,
+ "loss": 0.0733,
+ "step": 23891
+ },
+ {
+ "epoch": 65.1008174386921,
+ "grad_norm": 2.124048948287964,
+ "learning_rate": 5.736107474213385e-06,
+ "loss": 0.0447,
+ "step": 23892
+ },
+ {
+ "epoch": 65.10354223433242,
+ "grad_norm": 1.5940946340560913,
+ "learning_rate": 5.735309239318387e-06,
+ "loss": 0.0174,
+ "step": 23893
+ },
+ {
+ "epoch": 65.10626702997276,
+ "grad_norm": 1.5557355880737305,
+ "learning_rate": 5.734511037636667e-06,
+ "loss": 0.0369,
+ "step": 23894
+ },
+ {
+ "epoch": 65.10899182561307,
+ "grad_norm": 1.794878602027893,
+ "learning_rate": 5.7337128691744395e-06,
+ "loss": 0.0536,
+ "step": 23895
+ },
+ {
+ "epoch": 65.11171662125341,
+ "grad_norm": 1.3462039232254028,
+ "learning_rate": 5.732914733937917e-06,
+ "loss": 0.0771,
+ "step": 23896
+ },
+ {
+ "epoch": 65.11444141689373,
+ "grad_norm": 1.1686784029006958,
+ "learning_rate": 5.732116631933313e-06,
+ "loss": 0.0167,
+ "step": 23897
+ },
+ {
+ "epoch": 65.11716621253406,
+ "grad_norm": 2.7702462673187256,
+ "learning_rate": 5.731318563166852e-06,
+ "loss": 0.1112,
+ "step": 23898
+ },
+ {
+ "epoch": 65.11989100817439,
+ "grad_norm": 1.482635498046875,
+ "learning_rate": 5.7305205276447454e-06,
+ "loss": 0.0641,
+ "step": 23899
+ },
+ {
+ "epoch": 65.12261580381471,
+ "grad_norm": 2.0533549785614014,
+ "learning_rate": 5.729722525373206e-06,
+ "loss": 0.0553,
+ "step": 23900
+ },
+ {
+ "epoch": 65.12534059945504,
+ "grad_norm": 2.2930450439453125,
+ "learning_rate": 5.728924556358446e-06,
+ "loss": 0.0387,
+ "step": 23901
+ },
+ {
+ "epoch": 65.12806539509536,
+ "grad_norm": 2.384227752685547,
+ "learning_rate": 5.72812662060669e-06,
+ "loss": 0.0635,
+ "step": 23902
+ },
+ {
+ "epoch": 65.1307901907357,
+ "grad_norm": 1.302087664604187,
+ "learning_rate": 5.727328718124146e-06,
+ "loss": 0.0223,
+ "step": 23903
+ },
+ {
+ "epoch": 65.13351498637603,
+ "grad_norm": 1.977304220199585,
+ "learning_rate": 5.726530848917027e-06,
+ "loss": 0.0286,
+ "step": 23904
+ },
+ {
+ "epoch": 65.13623978201635,
+ "grad_norm": 2.0363152027130127,
+ "learning_rate": 5.725733012991545e-06,
+ "loss": 0.0556,
+ "step": 23905
+ },
+ {
+ "epoch": 65.13896457765668,
+ "grad_norm": 2.5141892433166504,
+ "learning_rate": 5.724935210353921e-06,
+ "loss": 0.019,
+ "step": 23906
+ },
+ {
+ "epoch": 65.141689373297,
+ "grad_norm": 1.4035639762878418,
+ "learning_rate": 5.724137441010363e-06,
+ "loss": 0.0202,
+ "step": 23907
+ },
+ {
+ "epoch": 65.14441416893733,
+ "grad_norm": 2.3554258346557617,
+ "learning_rate": 5.723339704967081e-06,
+ "loss": 0.1341,
+ "step": 23908
+ },
+ {
+ "epoch": 65.14713896457765,
+ "grad_norm": 1.82404363155365,
+ "learning_rate": 5.722542002230293e-06,
+ "loss": 0.0167,
+ "step": 23909
+ },
+ {
+ "epoch": 65.14986376021798,
+ "grad_norm": 2.0270347595214844,
+ "learning_rate": 5.721744332806214e-06,
+ "loss": 0.0672,
+ "step": 23910
+ },
+ {
+ "epoch": 65.15258855585832,
+ "grad_norm": 1.4108020067214966,
+ "learning_rate": 5.720946696701053e-06,
+ "loss": 0.0285,
+ "step": 23911
+ },
+ {
+ "epoch": 65.15531335149863,
+ "grad_norm": 0.8344237804412842,
+ "learning_rate": 5.720149093921019e-06,
+ "loss": 0.0296,
+ "step": 23912
+ },
+ {
+ "epoch": 65.15803814713897,
+ "grad_norm": 1.686086654663086,
+ "learning_rate": 5.719351524472326e-06,
+ "loss": 0.0335,
+ "step": 23913
+ },
+ {
+ "epoch": 65.16076294277929,
+ "grad_norm": 1.3584409952163696,
+ "learning_rate": 5.718553988361186e-06,
+ "loss": 0.0127,
+ "step": 23914
+ },
+ {
+ "epoch": 65.16348773841962,
+ "grad_norm": 2.409972667694092,
+ "learning_rate": 5.717756485593811e-06,
+ "loss": 0.1014,
+ "step": 23915
+ },
+ {
+ "epoch": 65.16621253405995,
+ "grad_norm": 2.612624406814575,
+ "learning_rate": 5.71695901617641e-06,
+ "loss": 0.1066,
+ "step": 23916
+ },
+ {
+ "epoch": 65.16893732970027,
+ "grad_norm": 1.911440372467041,
+ "learning_rate": 5.716161580115192e-06,
+ "loss": 0.0305,
+ "step": 23917
+ },
+ {
+ "epoch": 65.1716621253406,
+ "grad_norm": 1.4712733030319214,
+ "learning_rate": 5.715364177416371e-06,
+ "loss": 0.0176,
+ "step": 23918
+ },
+ {
+ "epoch": 65.17438692098092,
+ "grad_norm": 1.8914521932601929,
+ "learning_rate": 5.714566808086159e-06,
+ "loss": 0.0219,
+ "step": 23919
+ },
+ {
+ "epoch": 65.17711171662125,
+ "grad_norm": 0.8856232762336731,
+ "learning_rate": 5.713769472130761e-06,
+ "loss": 0.0111,
+ "step": 23920
+ },
+ {
+ "epoch": 65.17983651226157,
+ "grad_norm": 1.5340352058410645,
+ "learning_rate": 5.712972169556384e-06,
+ "loss": 0.0933,
+ "step": 23921
+ },
+ {
+ "epoch": 65.1825613079019,
+ "grad_norm": 1.982503056526184,
+ "learning_rate": 5.712174900369247e-06,
+ "loss": 0.0343,
+ "step": 23922
+ },
+ {
+ "epoch": 65.18528610354224,
+ "grad_norm": 2.1724348068237305,
+ "learning_rate": 5.7113776645755525e-06,
+ "loss": 0.0407,
+ "step": 23923
+ },
+ {
+ "epoch": 65.18801089918256,
+ "grad_norm": 1.3289490938186646,
+ "learning_rate": 5.710580462181512e-06,
+ "loss": 0.0188,
+ "step": 23924
+ },
+ {
+ "epoch": 65.19073569482289,
+ "grad_norm": 1.202426552772522,
+ "learning_rate": 5.709783293193329e-06,
+ "loss": 0.0106,
+ "step": 23925
+ },
+ {
+ "epoch": 65.19346049046321,
+ "grad_norm": 2.257153272628784,
+ "learning_rate": 5.708986157617215e-06,
+ "loss": 0.0624,
+ "step": 23926
+ },
+ {
+ "epoch": 65.19618528610354,
+ "grad_norm": 1.6293476819992065,
+ "learning_rate": 5.708189055459382e-06,
+ "loss": 0.0569,
+ "step": 23927
+ },
+ {
+ "epoch": 65.19891008174388,
+ "grad_norm": 1.6536710262298584,
+ "learning_rate": 5.707391986726035e-06,
+ "loss": 0.06,
+ "step": 23928
+ },
+ {
+ "epoch": 65.2016348773842,
+ "grad_norm": 1.6041510105133057,
+ "learning_rate": 5.706594951423381e-06,
+ "loss": 0.107,
+ "step": 23929
+ },
+ {
+ "epoch": 65.20435967302453,
+ "grad_norm": 1.158810019493103,
+ "learning_rate": 5.705797949557624e-06,
+ "loss": 0.0156,
+ "step": 23930
+ },
+ {
+ "epoch": 65.20708446866485,
+ "grad_norm": 1.5937738418579102,
+ "learning_rate": 5.705000981134979e-06,
+ "loss": 0.0499,
+ "step": 23931
+ },
+ {
+ "epoch": 65.20980926430518,
+ "grad_norm": 2.2451517581939697,
+ "learning_rate": 5.704204046161647e-06,
+ "loss": 0.0275,
+ "step": 23932
+ },
+ {
+ "epoch": 65.2125340599455,
+ "grad_norm": 1.9782122373580933,
+ "learning_rate": 5.703407144643831e-06,
+ "loss": 0.0207,
+ "step": 23933
+ },
+ {
+ "epoch": 65.21525885558583,
+ "grad_norm": 1.3211140632629395,
+ "learning_rate": 5.702610276587747e-06,
+ "loss": 0.0218,
+ "step": 23934
+ },
+ {
+ "epoch": 65.21798365122616,
+ "grad_norm": 1.9111523628234863,
+ "learning_rate": 5.701813441999596e-06,
+ "loss": 0.0414,
+ "step": 23935
+ },
+ {
+ "epoch": 65.22070844686648,
+ "grad_norm": 2.0720913410186768,
+ "learning_rate": 5.701016640885582e-06,
+ "loss": 0.0836,
+ "step": 23936
+ },
+ {
+ "epoch": 65.22343324250681,
+ "grad_norm": 1.5412578582763672,
+ "learning_rate": 5.700219873251909e-06,
+ "loss": 0.0214,
+ "step": 23937
+ },
+ {
+ "epoch": 65.22615803814713,
+ "grad_norm": 1.401303768157959,
+ "learning_rate": 5.699423139104789e-06,
+ "loss": 0.0178,
+ "step": 23938
+ },
+ {
+ "epoch": 65.22888283378747,
+ "grad_norm": 1.5397884845733643,
+ "learning_rate": 5.698626438450423e-06,
+ "loss": 0.0233,
+ "step": 23939
+ },
+ {
+ "epoch": 65.2316076294278,
+ "grad_norm": 1.7637156248092651,
+ "learning_rate": 5.697829771295017e-06,
+ "loss": 0.0222,
+ "step": 23940
+ },
+ {
+ "epoch": 65.23433242506812,
+ "grad_norm": 1.882187008857727,
+ "learning_rate": 5.69703313764477e-06,
+ "loss": 0.052,
+ "step": 23941
+ },
+ {
+ "epoch": 65.23705722070845,
+ "grad_norm": 2.960939645767212,
+ "learning_rate": 5.69623653750589e-06,
+ "loss": 0.0452,
+ "step": 23942
+ },
+ {
+ "epoch": 65.23978201634877,
+ "grad_norm": 1.4603931903839111,
+ "learning_rate": 5.695439970884585e-06,
+ "loss": 0.0196,
+ "step": 23943
+ },
+ {
+ "epoch": 65.2425068119891,
+ "grad_norm": 1.8698638677597046,
+ "learning_rate": 5.694643437787055e-06,
+ "loss": 0.0308,
+ "step": 23944
+ },
+ {
+ "epoch": 65.24523160762942,
+ "grad_norm": 3.247584581375122,
+ "learning_rate": 5.693846938219505e-06,
+ "loss": 0.068,
+ "step": 23945
+ },
+ {
+ "epoch": 65.24795640326975,
+ "grad_norm": 1.678651213645935,
+ "learning_rate": 5.693050472188131e-06,
+ "loss": 0.1733,
+ "step": 23946
+ },
+ {
+ "epoch": 65.25068119891009,
+ "grad_norm": 2.2935140132904053,
+ "learning_rate": 5.692254039699146e-06,
+ "loss": 0.0355,
+ "step": 23947
+ },
+ {
+ "epoch": 65.2534059945504,
+ "grad_norm": 1.9504164457321167,
+ "learning_rate": 5.691457640758746e-06,
+ "loss": 0.0998,
+ "step": 23948
+ },
+ {
+ "epoch": 65.25613079019074,
+ "grad_norm": 1.7271662950515747,
+ "learning_rate": 5.690661275373137e-06,
+ "loss": 0.0298,
+ "step": 23949
+ },
+ {
+ "epoch": 65.25885558583106,
+ "grad_norm": 1.799682855606079,
+ "learning_rate": 5.689864943548514e-06,
+ "loss": 0.0585,
+ "step": 23950
+ },
+ {
+ "epoch": 65.26158038147139,
+ "grad_norm": 1.0246896743774414,
+ "learning_rate": 5.68906864529109e-06,
+ "loss": 0.0131,
+ "step": 23951
+ },
+ {
+ "epoch": 65.26430517711172,
+ "grad_norm": 1.6514414548873901,
+ "learning_rate": 5.688272380607058e-06,
+ "loss": 0.0207,
+ "step": 23952
+ },
+ {
+ "epoch": 65.26702997275204,
+ "grad_norm": 1.6408708095550537,
+ "learning_rate": 5.68747614950262e-06,
+ "loss": 0.0382,
+ "step": 23953
+ },
+ {
+ "epoch": 65.26975476839237,
+ "grad_norm": 2.469822883605957,
+ "learning_rate": 5.686679951983982e-06,
+ "loss": 0.1085,
+ "step": 23954
+ },
+ {
+ "epoch": 65.2724795640327,
+ "grad_norm": 2.17673659324646,
+ "learning_rate": 5.6858837880573426e-06,
+ "loss": 0.0534,
+ "step": 23955
+ },
+ {
+ "epoch": 65.27520435967303,
+ "grad_norm": 1.2158116102218628,
+ "learning_rate": 5.685087657728902e-06,
+ "loss": 0.022,
+ "step": 23956
+ },
+ {
+ "epoch": 65.27792915531334,
+ "grad_norm": 3.4442837238311768,
+ "learning_rate": 5.6842915610048545e-06,
+ "loss": 0.0297,
+ "step": 23957
+ },
+ {
+ "epoch": 65.28065395095368,
+ "grad_norm": 1.4420353174209595,
+ "learning_rate": 5.683495497891409e-06,
+ "loss": 0.0246,
+ "step": 23958
+ },
+ {
+ "epoch": 65.28337874659401,
+ "grad_norm": 1.0967471599578857,
+ "learning_rate": 5.682699468394763e-06,
+ "loss": 0.016,
+ "step": 23959
+ },
+ {
+ "epoch": 65.28610354223433,
+ "grad_norm": 1.667142391204834,
+ "learning_rate": 5.6819034725211096e-06,
+ "loss": 0.0245,
+ "step": 23960
+ },
+ {
+ "epoch": 65.28882833787466,
+ "grad_norm": 1.2729954719543457,
+ "learning_rate": 5.681107510276658e-06,
+ "loss": 0.0156,
+ "step": 23961
+ },
+ {
+ "epoch": 65.29155313351498,
+ "grad_norm": 1.8771530389785767,
+ "learning_rate": 5.680311581667597e-06,
+ "loss": 0.0421,
+ "step": 23962
+ },
+ {
+ "epoch": 65.29427792915531,
+ "grad_norm": 1.4516624212265015,
+ "learning_rate": 5.6795156867001345e-06,
+ "loss": 0.0238,
+ "step": 23963
+ },
+ {
+ "epoch": 65.29700272479565,
+ "grad_norm": 15.654653549194336,
+ "learning_rate": 5.678719825380466e-06,
+ "loss": 0.0278,
+ "step": 23964
+ },
+ {
+ "epoch": 65.29972752043597,
+ "grad_norm": 1.7806390523910522,
+ "learning_rate": 5.677923997714787e-06,
+ "loss": 0.0492,
+ "step": 23965
+ },
+ {
+ "epoch": 65.3024523160763,
+ "grad_norm": 1.246377944946289,
+ "learning_rate": 5.677128203709294e-06,
+ "loss": 0.0188,
+ "step": 23966
+ },
+ {
+ "epoch": 65.30517711171662,
+ "grad_norm": 2.231471538543701,
+ "learning_rate": 5.676332443370189e-06,
+ "loss": 0.1334,
+ "step": 23967
+ },
+ {
+ "epoch": 65.30790190735695,
+ "grad_norm": 4.643612384796143,
+ "learning_rate": 5.675536716703671e-06,
+ "loss": 0.0192,
+ "step": 23968
+ },
+ {
+ "epoch": 65.31062670299727,
+ "grad_norm": 1.9536374807357788,
+ "learning_rate": 5.674741023715933e-06,
+ "loss": 0.0747,
+ "step": 23969
+ },
+ {
+ "epoch": 65.3133514986376,
+ "grad_norm": 1.9786834716796875,
+ "learning_rate": 5.673945364413167e-06,
+ "loss": 0.0252,
+ "step": 23970
+ },
+ {
+ "epoch": 65.31607629427793,
+ "grad_norm": 1.588884711265564,
+ "learning_rate": 5.67314973880158e-06,
+ "loss": 0.0221,
+ "step": 23971
+ },
+ {
+ "epoch": 65.31880108991825,
+ "grad_norm": 1.7369295358657837,
+ "learning_rate": 5.672354146887364e-06,
+ "loss": 0.1073,
+ "step": 23972
+ },
+ {
+ "epoch": 65.32152588555859,
+ "grad_norm": 1.6561384201049805,
+ "learning_rate": 5.671558588676715e-06,
+ "loss": 0.0268,
+ "step": 23973
+ },
+ {
+ "epoch": 65.3242506811989,
+ "grad_norm": 1.5523159503936768,
+ "learning_rate": 5.6707630641758235e-06,
+ "loss": 0.0181,
+ "step": 23974
+ },
+ {
+ "epoch": 65.32697547683924,
+ "grad_norm": 1.8549326658248901,
+ "learning_rate": 5.6699675733908955e-06,
+ "loss": 0.1599,
+ "step": 23975
+ },
+ {
+ "epoch": 65.32970027247957,
+ "grad_norm": 1.4225406646728516,
+ "learning_rate": 5.6691721163281185e-06,
+ "loss": 0.0503,
+ "step": 23976
+ },
+ {
+ "epoch": 65.33242506811989,
+ "grad_norm": 1.7244014739990234,
+ "learning_rate": 5.668376692993687e-06,
+ "loss": 0.0186,
+ "step": 23977
+ },
+ {
+ "epoch": 65.33514986376022,
+ "grad_norm": 1.6283658742904663,
+ "learning_rate": 5.6675813033937985e-06,
+ "loss": 0.0353,
+ "step": 23978
+ },
+ {
+ "epoch": 65.33787465940054,
+ "grad_norm": 1.2865842580795288,
+ "learning_rate": 5.666785947534651e-06,
+ "loss": 0.0322,
+ "step": 23979
+ },
+ {
+ "epoch": 65.34059945504087,
+ "grad_norm": 1.6242769956588745,
+ "learning_rate": 5.6659906254224355e-06,
+ "loss": 0.0137,
+ "step": 23980
+ },
+ {
+ "epoch": 65.34332425068119,
+ "grad_norm": 2.4839162826538086,
+ "learning_rate": 5.665195337063345e-06,
+ "loss": 0.0171,
+ "step": 23981
+ },
+ {
+ "epoch": 65.34604904632153,
+ "grad_norm": 1.810097336769104,
+ "learning_rate": 5.664400082463569e-06,
+ "loss": 0.222,
+ "step": 23982
+ },
+ {
+ "epoch": 65.34877384196186,
+ "grad_norm": 1.4222970008850098,
+ "learning_rate": 5.663604861629311e-06,
+ "loss": 0.0168,
+ "step": 23983
+ },
+ {
+ "epoch": 65.35149863760218,
+ "grad_norm": 1.5606610774993896,
+ "learning_rate": 5.662809674566759e-06,
+ "loss": 0.0167,
+ "step": 23984
+ },
+ {
+ "epoch": 65.35422343324251,
+ "grad_norm": 2.8472986221313477,
+ "learning_rate": 5.662014521282104e-06,
+ "loss": 0.072,
+ "step": 23985
+ },
+ {
+ "epoch": 65.35694822888283,
+ "grad_norm": 1.7151211500167847,
+ "learning_rate": 5.661219401781537e-06,
+ "loss": 0.1032,
+ "step": 23986
+ },
+ {
+ "epoch": 65.35967302452316,
+ "grad_norm": 7.22019624710083,
+ "learning_rate": 5.660424316071258e-06,
+ "loss": 0.0272,
+ "step": 23987
+ },
+ {
+ "epoch": 65.3623978201635,
+ "grad_norm": 1.9050699472427368,
+ "learning_rate": 5.6596292641574535e-06,
+ "loss": 0.0264,
+ "step": 23988
+ },
+ {
+ "epoch": 65.36512261580381,
+ "grad_norm": 1.5122694969177246,
+ "learning_rate": 5.658834246046319e-06,
+ "loss": 0.0281,
+ "step": 23989
+ },
+ {
+ "epoch": 65.36784741144415,
+ "grad_norm": 1.3322696685791016,
+ "learning_rate": 5.658039261744037e-06,
+ "loss": 0.0135,
+ "step": 23990
+ },
+ {
+ "epoch": 65.37057220708446,
+ "grad_norm": 2.2564103603363037,
+ "learning_rate": 5.657244311256811e-06,
+ "loss": 0.0273,
+ "step": 23991
+ },
+ {
+ "epoch": 65.3732970027248,
+ "grad_norm": 1.5185288190841675,
+ "learning_rate": 5.656449394590826e-06,
+ "loss": 0.0171,
+ "step": 23992
+ },
+ {
+ "epoch": 65.37602179836512,
+ "grad_norm": 1.850968837738037,
+ "learning_rate": 5.655654511752274e-06,
+ "loss": 0.0166,
+ "step": 23993
+ },
+ {
+ "epoch": 65.37874659400545,
+ "grad_norm": 2.8955862522125244,
+ "learning_rate": 5.654859662747337e-06,
+ "loss": 0.032,
+ "step": 23994
+ },
+ {
+ "epoch": 65.38147138964578,
+ "grad_norm": 3.326239585876465,
+ "learning_rate": 5.6540648475822215e-06,
+ "loss": 0.0333,
+ "step": 23995
+ },
+ {
+ "epoch": 65.3841961852861,
+ "grad_norm": 2.289945363998413,
+ "learning_rate": 5.653270066263108e-06,
+ "loss": 0.0866,
+ "step": 23996
+ },
+ {
+ "epoch": 65.38692098092643,
+ "grad_norm": 5.9366912841796875,
+ "learning_rate": 5.6524753187961865e-06,
+ "loss": 0.0192,
+ "step": 23997
+ },
+ {
+ "epoch": 65.38964577656675,
+ "grad_norm": 1.75386381149292,
+ "learning_rate": 5.6516806051876435e-06,
+ "loss": 0.1678,
+ "step": 23998
+ },
+ {
+ "epoch": 65.39237057220708,
+ "grad_norm": 1.680172324180603,
+ "learning_rate": 5.650885925443678e-06,
+ "loss": 0.0787,
+ "step": 23999
+ },
+ {
+ "epoch": 65.39509536784742,
+ "grad_norm": 1.7128089666366577,
+ "learning_rate": 5.650091279570471e-06,
+ "loss": 0.0184,
+ "step": 24000
+ },
+ {
+ "epoch": 65.39782016348774,
+ "grad_norm": 2.4267420768737793,
+ "learning_rate": 5.649296667574215e-06,
+ "loss": 0.0435,
+ "step": 24001
+ },
+ {
+ "epoch": 65.40054495912807,
+ "grad_norm": 1.1746773719787598,
+ "learning_rate": 5.6485020894610905e-06,
+ "loss": 0.0213,
+ "step": 24002
+ },
+ {
+ "epoch": 65.40326975476839,
+ "grad_norm": 1.199209451675415,
+ "learning_rate": 5.647707545237298e-06,
+ "loss": 0.0084,
+ "step": 24003
+ },
+ {
+ "epoch": 65.40599455040872,
+ "grad_norm": 1.1901177167892456,
+ "learning_rate": 5.646913034909016e-06,
+ "loss": 0.0131,
+ "step": 24004
+ },
+ {
+ "epoch": 65.40871934604904,
+ "grad_norm": 1.658762812614441,
+ "learning_rate": 5.646118558482438e-06,
+ "loss": 0.0178,
+ "step": 24005
+ },
+ {
+ "epoch": 65.41144414168937,
+ "grad_norm": 1.494570255279541,
+ "learning_rate": 5.645324115963742e-06,
+ "loss": 0.0156,
+ "step": 24006
+ },
+ {
+ "epoch": 65.4141689373297,
+ "grad_norm": 2.0456929206848145,
+ "learning_rate": 5.644529707359128e-06,
+ "loss": 0.0749,
+ "step": 24007
+ },
+ {
+ "epoch": 65.41689373297002,
+ "grad_norm": 1.9392716884613037,
+ "learning_rate": 5.643735332674776e-06,
+ "loss": 0.0516,
+ "step": 24008
+ },
+ {
+ "epoch": 65.41961852861036,
+ "grad_norm": 2.4649271965026855,
+ "learning_rate": 5.6429409919168724e-06,
+ "loss": 0.0787,
+ "step": 24009
+ },
+ {
+ "epoch": 65.42234332425068,
+ "grad_norm": 1.3504976034164429,
+ "learning_rate": 5.6421466850915986e-06,
+ "loss": 0.0723,
+ "step": 24010
+ },
+ {
+ "epoch": 65.42506811989101,
+ "grad_norm": 1.6556237936019897,
+ "learning_rate": 5.641352412205148e-06,
+ "loss": 0.0779,
+ "step": 24011
+ },
+ {
+ "epoch": 65.42779291553134,
+ "grad_norm": 2.1391611099243164,
+ "learning_rate": 5.640558173263708e-06,
+ "loss": 0.024,
+ "step": 24012
+ },
+ {
+ "epoch": 65.43051771117166,
+ "grad_norm": 2.656092643737793,
+ "learning_rate": 5.6397639682734616e-06,
+ "loss": 0.0661,
+ "step": 24013
+ },
+ {
+ "epoch": 65.433242506812,
+ "grad_norm": 2.0034642219543457,
+ "learning_rate": 5.638969797240592e-06,
+ "loss": 0.1082,
+ "step": 24014
+ },
+ {
+ "epoch": 65.43596730245231,
+ "grad_norm": 1.3261052370071411,
+ "learning_rate": 5.638175660171281e-06,
+ "loss": 0.0225,
+ "step": 24015
+ },
+ {
+ "epoch": 65.43869209809264,
+ "grad_norm": 2.1208512783050537,
+ "learning_rate": 5.637381557071724e-06,
+ "loss": 0.0819,
+ "step": 24016
+ },
+ {
+ "epoch": 65.44141689373296,
+ "grad_norm": 2.316370964050293,
+ "learning_rate": 5.636587487948096e-06,
+ "loss": 0.0676,
+ "step": 24017
+ },
+ {
+ "epoch": 65.4441416893733,
+ "grad_norm": 1.2309339046478271,
+ "learning_rate": 5.635793452806582e-06,
+ "loss": 0.0608,
+ "step": 24018
+ },
+ {
+ "epoch": 65.44686648501363,
+ "grad_norm": 1.6152894496917725,
+ "learning_rate": 5.634999451653372e-06,
+ "loss": 0.0235,
+ "step": 24019
+ },
+ {
+ "epoch": 65.44959128065395,
+ "grad_norm": 1.416690707206726,
+ "learning_rate": 5.6342054844946446e-06,
+ "loss": 0.0326,
+ "step": 24020
+ },
+ {
+ "epoch": 65.45231607629428,
+ "grad_norm": 1.1335103511810303,
+ "learning_rate": 5.633411551336585e-06,
+ "loss": 0.0122,
+ "step": 24021
+ },
+ {
+ "epoch": 65.4550408719346,
+ "grad_norm": 2.074540376663208,
+ "learning_rate": 5.6326176521853725e-06,
+ "loss": 0.0462,
+ "step": 24022
+ },
+ {
+ "epoch": 65.45776566757493,
+ "grad_norm": 1.1698473691940308,
+ "learning_rate": 5.631823787047196e-06,
+ "loss": 0.0161,
+ "step": 24023
+ },
+ {
+ "epoch": 65.46049046321527,
+ "grad_norm": 1.8177380561828613,
+ "learning_rate": 5.631029955928236e-06,
+ "loss": 0.0338,
+ "step": 24024
+ },
+ {
+ "epoch": 65.46321525885558,
+ "grad_norm": 1.9146959781646729,
+ "learning_rate": 5.630236158834675e-06,
+ "loss": 0.0237,
+ "step": 24025
+ },
+ {
+ "epoch": 65.46594005449592,
+ "grad_norm": 1.4265724420547485,
+ "learning_rate": 5.629442395772688e-06,
+ "loss": 0.02,
+ "step": 24026
+ },
+ {
+ "epoch": 65.46866485013624,
+ "grad_norm": 1.3791720867156982,
+ "learning_rate": 5.628648666748467e-06,
+ "loss": 0.0162,
+ "step": 24027
+ },
+ {
+ "epoch": 65.47138964577657,
+ "grad_norm": 1.5359541177749634,
+ "learning_rate": 5.627854971768187e-06,
+ "loss": 0.021,
+ "step": 24028
+ },
+ {
+ "epoch": 65.47411444141689,
+ "grad_norm": 1.2796446084976196,
+ "learning_rate": 5.627061310838035e-06,
+ "loss": 0.0237,
+ "step": 24029
+ },
+ {
+ "epoch": 65.47683923705722,
+ "grad_norm": 1.6833970546722412,
+ "learning_rate": 5.626267683964188e-06,
+ "loss": 0.1297,
+ "step": 24030
+ },
+ {
+ "epoch": 65.47956403269755,
+ "grad_norm": 1.86417818069458,
+ "learning_rate": 5.625474091152825e-06,
+ "loss": 0.0239,
+ "step": 24031
+ },
+ {
+ "epoch": 65.48228882833787,
+ "grad_norm": 1.5451184511184692,
+ "learning_rate": 5.624680532410134e-06,
+ "loss": 0.0188,
+ "step": 24032
+ },
+ {
+ "epoch": 65.4850136239782,
+ "grad_norm": 1.8960864543914795,
+ "learning_rate": 5.623887007742288e-06,
+ "loss": 0.0406,
+ "step": 24033
+ },
+ {
+ "epoch": 65.48773841961852,
+ "grad_norm": 2.200434923171997,
+ "learning_rate": 5.623093517155469e-06,
+ "loss": 0.034,
+ "step": 24034
+ },
+ {
+ "epoch": 65.49046321525886,
+ "grad_norm": 1.851019024848938,
+ "learning_rate": 5.622300060655854e-06,
+ "loss": 0.0212,
+ "step": 24035
+ },
+ {
+ "epoch": 65.49318801089919,
+ "grad_norm": 1.996330976486206,
+ "learning_rate": 5.62150663824963e-06,
+ "loss": 0.0578,
+ "step": 24036
+ },
+ {
+ "epoch": 65.49591280653951,
+ "grad_norm": 1.7704271078109741,
+ "learning_rate": 5.620713249942968e-06,
+ "loss": 0.0154,
+ "step": 24037
+ },
+ {
+ "epoch": 65.49863760217984,
+ "grad_norm": 1.392599105834961,
+ "learning_rate": 5.619919895742054e-06,
+ "loss": 0.0363,
+ "step": 24038
+ },
+ {
+ "epoch": 65.50136239782016,
+ "grad_norm": 1.738369345664978,
+ "learning_rate": 5.619126575653058e-06,
+ "loss": 0.0174,
+ "step": 24039
+ },
+ {
+ "epoch": 65.50408719346049,
+ "grad_norm": 1.6718114614486694,
+ "learning_rate": 5.618333289682165e-06,
+ "loss": 0.1018,
+ "step": 24040
+ },
+ {
+ "epoch": 65.50681198910081,
+ "grad_norm": 1.6190698146820068,
+ "learning_rate": 5.617540037835553e-06,
+ "loss": 0.0376,
+ "step": 24041
+ },
+ {
+ "epoch": 65.50953678474114,
+ "grad_norm": 1.6873481273651123,
+ "learning_rate": 5.616746820119394e-06,
+ "loss": 0.0146,
+ "step": 24042
+ },
+ {
+ "epoch": 65.51226158038148,
+ "grad_norm": 2.346768379211426,
+ "learning_rate": 5.615953636539875e-06,
+ "loss": 0.0222,
+ "step": 24043
+ },
+ {
+ "epoch": 65.5149863760218,
+ "grad_norm": 1.7113797664642334,
+ "learning_rate": 5.6151604871031665e-06,
+ "loss": 0.0266,
+ "step": 24044
+ },
+ {
+ "epoch": 65.51771117166213,
+ "grad_norm": 1.6326115131378174,
+ "learning_rate": 5.6143673718154466e-06,
+ "loss": 0.0841,
+ "step": 24045
+ },
+ {
+ "epoch": 65.52043596730245,
+ "grad_norm": 2.461071491241455,
+ "learning_rate": 5.6135742906828894e-06,
+ "loss": 0.02,
+ "step": 24046
+ },
+ {
+ "epoch": 65.52316076294278,
+ "grad_norm": 2.4469594955444336,
+ "learning_rate": 5.612781243711673e-06,
+ "loss": 0.0325,
+ "step": 24047
+ },
+ {
+ "epoch": 65.52588555858311,
+ "grad_norm": 2.0061562061309814,
+ "learning_rate": 5.611988230907981e-06,
+ "loss": 0.0347,
+ "step": 24048
+ },
+ {
+ "epoch": 65.52861035422343,
+ "grad_norm": 1.6560651063919067,
+ "learning_rate": 5.611195252277983e-06,
+ "loss": 0.0261,
+ "step": 24049
+ },
+ {
+ "epoch": 65.53133514986376,
+ "grad_norm": 1.6774922609329224,
+ "learning_rate": 5.610402307827855e-06,
+ "loss": 0.0406,
+ "step": 24050
+ },
+ {
+ "epoch": 65.53405994550408,
+ "grad_norm": 2.6542763710021973,
+ "learning_rate": 5.609609397563768e-06,
+ "loss": 0.0637,
+ "step": 24051
+ },
+ {
+ "epoch": 65.53678474114442,
+ "grad_norm": 2.7218196392059326,
+ "learning_rate": 5.6088165214919064e-06,
+ "loss": 0.0892,
+ "step": 24052
+ },
+ {
+ "epoch": 65.53950953678473,
+ "grad_norm": 3.6189146041870117,
+ "learning_rate": 5.6080236796184395e-06,
+ "loss": 0.0101,
+ "step": 24053
+ },
+ {
+ "epoch": 65.54223433242507,
+ "grad_norm": 2.3353323936462402,
+ "learning_rate": 5.607230871949543e-06,
+ "loss": 0.0314,
+ "step": 24054
+ },
+ {
+ "epoch": 65.5449591280654,
+ "grad_norm": 1.953412652015686,
+ "learning_rate": 5.606438098491388e-06,
+ "loss": 0.1303,
+ "step": 24055
+ },
+ {
+ "epoch": 65.54768392370572,
+ "grad_norm": 4.799600601196289,
+ "learning_rate": 5.605645359250155e-06,
+ "loss": 0.2522,
+ "step": 24056
+ },
+ {
+ "epoch": 65.55040871934605,
+ "grad_norm": 1.9325088262557983,
+ "learning_rate": 5.604852654232015e-06,
+ "loss": 0.022,
+ "step": 24057
+ },
+ {
+ "epoch": 65.55313351498637,
+ "grad_norm": 1.6524401903152466,
+ "learning_rate": 5.604059983443139e-06,
+ "loss": 0.0222,
+ "step": 24058
+ },
+ {
+ "epoch": 65.5558583106267,
+ "grad_norm": 1.279539942741394,
+ "learning_rate": 5.6032673468897015e-06,
+ "loss": 0.0945,
+ "step": 24059
+ },
+ {
+ "epoch": 65.55858310626704,
+ "grad_norm": 3.7460319995880127,
+ "learning_rate": 5.602474744577879e-06,
+ "loss": 0.0725,
+ "step": 24060
+ },
+ {
+ "epoch": 65.56130790190736,
+ "grad_norm": 1.866079568862915,
+ "learning_rate": 5.6016821765138405e-06,
+ "loss": 0.0553,
+ "step": 24061
+ },
+ {
+ "epoch": 65.56403269754769,
+ "grad_norm": 2.3795478343963623,
+ "learning_rate": 5.600889642703758e-06,
+ "loss": 0.0198,
+ "step": 24062
+ },
+ {
+ "epoch": 65.566757493188,
+ "grad_norm": 1.340385913848877,
+ "learning_rate": 5.6000971431538045e-06,
+ "loss": 0.02,
+ "step": 24063
+ },
+ {
+ "epoch": 65.56948228882834,
+ "grad_norm": 0.9903956055641174,
+ "learning_rate": 5.599304677870157e-06,
+ "loss": 0.0098,
+ "step": 24064
+ },
+ {
+ "epoch": 65.57220708446866,
+ "grad_norm": 1.6676584482192993,
+ "learning_rate": 5.598512246858982e-06,
+ "loss": 0.0305,
+ "step": 24065
+ },
+ {
+ "epoch": 65.57493188010899,
+ "grad_norm": 1.855392575263977,
+ "learning_rate": 5.5977198501264514e-06,
+ "loss": 0.1194,
+ "step": 24066
+ },
+ {
+ "epoch": 65.57765667574932,
+ "grad_norm": 1.7243852615356445,
+ "learning_rate": 5.5969274876787334e-06,
+ "loss": 0.0161,
+ "step": 24067
+ },
+ {
+ "epoch": 65.58038147138964,
+ "grad_norm": 1.419854760169983,
+ "learning_rate": 5.5961351595220084e-06,
+ "loss": 0.0301,
+ "step": 24068
+ },
+ {
+ "epoch": 65.58310626702998,
+ "grad_norm": 1.594313383102417,
+ "learning_rate": 5.595342865662439e-06,
+ "loss": 0.0154,
+ "step": 24069
+ },
+ {
+ "epoch": 65.5858310626703,
+ "grad_norm": 0.8518718481063843,
+ "learning_rate": 5.594550606106198e-06,
+ "loss": 0.0099,
+ "step": 24070
+ },
+ {
+ "epoch": 65.58855585831063,
+ "grad_norm": 2.053449869155884,
+ "learning_rate": 5.593758380859452e-06,
+ "loss": 0.0278,
+ "step": 24071
+ },
+ {
+ "epoch": 65.59128065395096,
+ "grad_norm": 1.924013614654541,
+ "learning_rate": 5.5929661899283775e-06,
+ "loss": 0.0351,
+ "step": 24072
+ },
+ {
+ "epoch": 65.59400544959128,
+ "grad_norm": 2.9540603160858154,
+ "learning_rate": 5.59217403331914e-06,
+ "loss": 0.0491,
+ "step": 24073
+ },
+ {
+ "epoch": 65.59673024523161,
+ "grad_norm": 1.9705939292907715,
+ "learning_rate": 5.5913819110379095e-06,
+ "loss": 0.2181,
+ "step": 24074
+ },
+ {
+ "epoch": 65.59945504087193,
+ "grad_norm": 1.2401853799819946,
+ "learning_rate": 5.5905898230908506e-06,
+ "loss": 0.0272,
+ "step": 24075
+ },
+ {
+ "epoch": 65.60217983651226,
+ "grad_norm": 1.2159323692321777,
+ "learning_rate": 5.5897977694841396e-06,
+ "loss": 0.0151,
+ "step": 24076
+ },
+ {
+ "epoch": 65.60490463215258,
+ "grad_norm": 1.5752997398376465,
+ "learning_rate": 5.589005750223944e-06,
+ "loss": 0.0144,
+ "step": 24077
+ },
+ {
+ "epoch": 65.60762942779292,
+ "grad_norm": 1.375836968421936,
+ "learning_rate": 5.588213765316427e-06,
+ "loss": 0.0142,
+ "step": 24078
+ },
+ {
+ "epoch": 65.61035422343325,
+ "grad_norm": 2.4057538509368896,
+ "learning_rate": 5.587421814767757e-06,
+ "loss": 0.0487,
+ "step": 24079
+ },
+ {
+ "epoch": 65.61307901907357,
+ "grad_norm": 1.2515490055084229,
+ "learning_rate": 5.586629898584103e-06,
+ "loss": 0.0322,
+ "step": 24080
+ },
+ {
+ "epoch": 65.6158038147139,
+ "grad_norm": 2.285785436630249,
+ "learning_rate": 5.5858380167716375e-06,
+ "loss": 0.1196,
+ "step": 24081
+ },
+ {
+ "epoch": 65.61852861035422,
+ "grad_norm": 1.7656583786010742,
+ "learning_rate": 5.585046169336524e-06,
+ "loss": 0.0444,
+ "step": 24082
+ },
+ {
+ "epoch": 65.62125340599455,
+ "grad_norm": 2.592852830886841,
+ "learning_rate": 5.584254356284924e-06,
+ "loss": 0.0622,
+ "step": 24083
+ },
+ {
+ "epoch": 65.62397820163488,
+ "grad_norm": 2.751730442047119,
+ "learning_rate": 5.583462577623013e-06,
+ "loss": 0.0164,
+ "step": 24084
+ },
+ {
+ "epoch": 65.6267029972752,
+ "grad_norm": 1.597642421722412,
+ "learning_rate": 5.5826708333569536e-06,
+ "loss": 0.0282,
+ "step": 24085
+ },
+ {
+ "epoch": 65.62942779291554,
+ "grad_norm": 1.4419804811477661,
+ "learning_rate": 5.58187912349291e-06,
+ "loss": 0.0231,
+ "step": 24086
+ },
+ {
+ "epoch": 65.63215258855585,
+ "grad_norm": 1.4019933938980103,
+ "learning_rate": 5.581087448037048e-06,
+ "loss": 0.0681,
+ "step": 24087
+ },
+ {
+ "epoch": 65.63487738419619,
+ "grad_norm": 1.9230319261550903,
+ "learning_rate": 5.580295806995538e-06,
+ "loss": 0.0442,
+ "step": 24088
+ },
+ {
+ "epoch": 65.6376021798365,
+ "grad_norm": 1.6214927434921265,
+ "learning_rate": 5.5795042003745416e-06,
+ "loss": 0.0161,
+ "step": 24089
+ },
+ {
+ "epoch": 65.64032697547684,
+ "grad_norm": 4.054281711578369,
+ "learning_rate": 5.578712628180225e-06,
+ "loss": 0.0155,
+ "step": 24090
+ },
+ {
+ "epoch": 65.64305177111717,
+ "grad_norm": 1.9175299406051636,
+ "learning_rate": 5.577921090418747e-06,
+ "loss": 0.0192,
+ "step": 24091
+ },
+ {
+ "epoch": 65.64577656675749,
+ "grad_norm": 1.6876134872436523,
+ "learning_rate": 5.577129587096282e-06,
+ "loss": 0.0889,
+ "step": 24092
+ },
+ {
+ "epoch": 65.64850136239782,
+ "grad_norm": 1.7289905548095703,
+ "learning_rate": 5.576338118218989e-06,
+ "loss": 0.0759,
+ "step": 24093
+ },
+ {
+ "epoch": 65.65122615803814,
+ "grad_norm": 2.275433301925659,
+ "learning_rate": 5.575546683793034e-06,
+ "loss": 0.063,
+ "step": 24094
+ },
+ {
+ "epoch": 65.65395095367847,
+ "grad_norm": 2.8143503665924072,
+ "learning_rate": 5.574755283824573e-06,
+ "loss": 0.2271,
+ "step": 24095
+ },
+ {
+ "epoch": 65.65667574931881,
+ "grad_norm": 1.5990889072418213,
+ "learning_rate": 5.5739639183197805e-06,
+ "loss": 0.0174,
+ "step": 24096
+ },
+ {
+ "epoch": 65.65940054495913,
+ "grad_norm": 2.3965373039245605,
+ "learning_rate": 5.57317258728481e-06,
+ "loss": 0.0426,
+ "step": 24097
+ },
+ {
+ "epoch": 65.66212534059946,
+ "grad_norm": 2.1258764266967773,
+ "learning_rate": 5.572381290725833e-06,
+ "loss": 0.0361,
+ "step": 24098
+ },
+ {
+ "epoch": 65.66485013623978,
+ "grad_norm": 2.2383673191070557,
+ "learning_rate": 5.571590028649008e-06,
+ "loss": 0.0249,
+ "step": 24099
+ },
+ {
+ "epoch": 65.66757493188011,
+ "grad_norm": 1.0497920513153076,
+ "learning_rate": 5.570798801060493e-06,
+ "loss": 0.0369,
+ "step": 24100
+ },
+ {
+ "epoch": 65.67029972752043,
+ "grad_norm": 1.8001549243927002,
+ "learning_rate": 5.570007607966459e-06,
+ "loss": 0.1453,
+ "step": 24101
+ },
+ {
+ "epoch": 65.67302452316076,
+ "grad_norm": 1.8766497373580933,
+ "learning_rate": 5.569216449373063e-06,
+ "loss": 0.0197,
+ "step": 24102
+ },
+ {
+ "epoch": 65.6757493188011,
+ "grad_norm": 1.282988429069519,
+ "learning_rate": 5.568425325286466e-06,
+ "loss": 0.013,
+ "step": 24103
+ },
+ {
+ "epoch": 65.67847411444141,
+ "grad_norm": 5.7863688468933105,
+ "learning_rate": 5.567634235712827e-06,
+ "loss": 0.038,
+ "step": 24104
+ },
+ {
+ "epoch": 65.68119891008175,
+ "grad_norm": 1.2000964879989624,
+ "learning_rate": 5.566843180658313e-06,
+ "loss": 0.0102,
+ "step": 24105
+ },
+ {
+ "epoch": 65.68392370572207,
+ "grad_norm": 1.7562464475631714,
+ "learning_rate": 5.566052160129083e-06,
+ "loss": 0.0206,
+ "step": 24106
+ },
+ {
+ "epoch": 65.6866485013624,
+ "grad_norm": 1.4619660377502441,
+ "learning_rate": 5.565261174131291e-06,
+ "loss": 0.0239,
+ "step": 24107
+ },
+ {
+ "epoch": 65.68937329700273,
+ "grad_norm": 1.1956688165664673,
+ "learning_rate": 5.5644702226711055e-06,
+ "loss": 0.0297,
+ "step": 24108
+ },
+ {
+ "epoch": 65.69209809264305,
+ "grad_norm": 2.6355223655700684,
+ "learning_rate": 5.563679305754685e-06,
+ "loss": 0.0497,
+ "step": 24109
+ },
+ {
+ "epoch": 65.69482288828338,
+ "grad_norm": 2.101957321166992,
+ "learning_rate": 5.5628884233881865e-06,
+ "loss": 0.0705,
+ "step": 24110
+ },
+ {
+ "epoch": 65.6975476839237,
+ "grad_norm": 1.1809228658676147,
+ "learning_rate": 5.562097575577765e-06,
+ "loss": 0.0159,
+ "step": 24111
+ },
+ {
+ "epoch": 65.70027247956403,
+ "grad_norm": 1.5892935991287231,
+ "learning_rate": 5.56130676232959e-06,
+ "loss": 0.0602,
+ "step": 24112
+ },
+ {
+ "epoch": 65.70299727520435,
+ "grad_norm": 1.2930151224136353,
+ "learning_rate": 5.5605159836498145e-06,
+ "loss": 0.0109,
+ "step": 24113
+ },
+ {
+ "epoch": 65.70572207084469,
+ "grad_norm": 1.5546777248382568,
+ "learning_rate": 5.559725239544595e-06,
+ "loss": 0.0144,
+ "step": 24114
+ },
+ {
+ "epoch": 65.70844686648502,
+ "grad_norm": 3.6567494869232178,
+ "learning_rate": 5.558934530020095e-06,
+ "loss": 0.0562,
+ "step": 24115
+ },
+ {
+ "epoch": 65.71117166212534,
+ "grad_norm": 1.8206892013549805,
+ "learning_rate": 5.558143855082466e-06,
+ "loss": 0.0615,
+ "step": 24116
+ },
+ {
+ "epoch": 65.71389645776567,
+ "grad_norm": 1.503445029258728,
+ "learning_rate": 5.557353214737875e-06,
+ "loss": 0.0274,
+ "step": 24117
+ },
+ {
+ "epoch": 65.71662125340599,
+ "grad_norm": 1.9906681776046753,
+ "learning_rate": 5.556562608992473e-06,
+ "loss": 0.1008,
+ "step": 24118
+ },
+ {
+ "epoch": 65.71934604904632,
+ "grad_norm": 1.8682634830474854,
+ "learning_rate": 5.555772037852418e-06,
+ "loss": 0.0234,
+ "step": 24119
+ },
+ {
+ "epoch": 65.72207084468666,
+ "grad_norm": 1.7854633331298828,
+ "learning_rate": 5.554981501323864e-06,
+ "loss": 0.0269,
+ "step": 24120
+ },
+ {
+ "epoch": 65.72479564032697,
+ "grad_norm": 1.599360704421997,
+ "learning_rate": 5.554190999412975e-06,
+ "loss": 0.0395,
+ "step": 24121
+ },
+ {
+ "epoch": 65.7275204359673,
+ "grad_norm": 2.4342992305755615,
+ "learning_rate": 5.553400532125903e-06,
+ "loss": 0.103,
+ "step": 24122
+ },
+ {
+ "epoch": 65.73024523160763,
+ "grad_norm": 2.350985288619995,
+ "learning_rate": 5.552610099468805e-06,
+ "loss": 0.0417,
+ "step": 24123
+ },
+ {
+ "epoch": 65.73297002724796,
+ "grad_norm": 1.8169951438903809,
+ "learning_rate": 5.551819701447832e-06,
+ "loss": 0.1035,
+ "step": 24124
+ },
+ {
+ "epoch": 65.73569482288828,
+ "grad_norm": 2.139601230621338,
+ "learning_rate": 5.5510293380691494e-06,
+ "loss": 0.0555,
+ "step": 24125
+ },
+ {
+ "epoch": 65.73841961852861,
+ "grad_norm": 2.335941791534424,
+ "learning_rate": 5.550239009338905e-06,
+ "loss": 0.0186,
+ "step": 24126
+ },
+ {
+ "epoch": 65.74114441416894,
+ "grad_norm": 1.9059703350067139,
+ "learning_rate": 5.549448715263252e-06,
+ "loss": 0.087,
+ "step": 24127
+ },
+ {
+ "epoch": 65.74386920980926,
+ "grad_norm": 1.7616281509399414,
+ "learning_rate": 5.548658455848355e-06,
+ "loss": 0.0484,
+ "step": 24128
+ },
+ {
+ "epoch": 65.7465940054496,
+ "grad_norm": 1.748361587524414,
+ "learning_rate": 5.547868231100362e-06,
+ "loss": 0.104,
+ "step": 24129
+ },
+ {
+ "epoch": 65.74931880108991,
+ "grad_norm": 1.5811481475830078,
+ "learning_rate": 5.547078041025429e-06,
+ "loss": 0.0107,
+ "step": 24130
+ },
+ {
+ "epoch": 65.75204359673025,
+ "grad_norm": 1.5997687578201294,
+ "learning_rate": 5.546287885629702e-06,
+ "loss": 0.0311,
+ "step": 24131
+ },
+ {
+ "epoch": 65.75476839237058,
+ "grad_norm": 1.961602807044983,
+ "learning_rate": 5.545497764919345e-06,
+ "loss": 0.0333,
+ "step": 24132
+ },
+ {
+ "epoch": 65.7574931880109,
+ "grad_norm": 2.5520029067993164,
+ "learning_rate": 5.544707678900509e-06,
+ "loss": 0.0321,
+ "step": 24133
+ },
+ {
+ "epoch": 65.76021798365123,
+ "grad_norm": 1.952789306640625,
+ "learning_rate": 5.543917627579347e-06,
+ "loss": 0.1315,
+ "step": 24134
+ },
+ {
+ "epoch": 65.76294277929155,
+ "grad_norm": 1.2139146327972412,
+ "learning_rate": 5.543127610962012e-06,
+ "loss": 0.0172,
+ "step": 24135
+ },
+ {
+ "epoch": 65.76566757493188,
+ "grad_norm": 1.648496150970459,
+ "learning_rate": 5.5423376290546526e-06,
+ "loss": 0.0313,
+ "step": 24136
+ },
+ {
+ "epoch": 65.7683923705722,
+ "grad_norm": 1.028080701828003,
+ "learning_rate": 5.541547681863426e-06,
+ "loss": 0.0104,
+ "step": 24137
+ },
+ {
+ "epoch": 65.77111716621253,
+ "grad_norm": 1.8115652799606323,
+ "learning_rate": 5.5407577693944835e-06,
+ "loss": 0.0427,
+ "step": 24138
+ },
+ {
+ "epoch": 65.77384196185287,
+ "grad_norm": 1.606353998184204,
+ "learning_rate": 5.539967891653976e-06,
+ "loss": 0.0257,
+ "step": 24139
+ },
+ {
+ "epoch": 65.77656675749319,
+ "grad_norm": 1.4590612649917603,
+ "learning_rate": 5.539178048648051e-06,
+ "loss": 0.0289,
+ "step": 24140
+ },
+ {
+ "epoch": 65.77929155313352,
+ "grad_norm": 2.1576473712921143,
+ "learning_rate": 5.538388240382868e-06,
+ "loss": 0.0315,
+ "step": 24141
+ },
+ {
+ "epoch": 65.78201634877384,
+ "grad_norm": 2.277019739151001,
+ "learning_rate": 5.537598466864574e-06,
+ "loss": 0.0802,
+ "step": 24142
+ },
+ {
+ "epoch": 65.78474114441417,
+ "grad_norm": 0.9964087605476379,
+ "learning_rate": 5.536808728099318e-06,
+ "loss": 0.0149,
+ "step": 24143
+ },
+ {
+ "epoch": 65.7874659400545,
+ "grad_norm": 1.382768154144287,
+ "learning_rate": 5.536019024093251e-06,
+ "loss": 0.0289,
+ "step": 24144
+ },
+ {
+ "epoch": 65.79019073569482,
+ "grad_norm": 0.9875184893608093,
+ "learning_rate": 5.535229354852525e-06,
+ "loss": 0.0179,
+ "step": 24145
+ },
+ {
+ "epoch": 65.79291553133515,
+ "grad_norm": 1.9355429410934448,
+ "learning_rate": 5.534439720383291e-06,
+ "loss": 0.0607,
+ "step": 24146
+ },
+ {
+ "epoch": 65.79564032697547,
+ "grad_norm": 1.6414179801940918,
+ "learning_rate": 5.533650120691696e-06,
+ "loss": 0.0287,
+ "step": 24147
+ },
+ {
+ "epoch": 65.7983651226158,
+ "grad_norm": 1.3989732265472412,
+ "learning_rate": 5.532860555783885e-06,
+ "loss": 0.0202,
+ "step": 24148
+ },
+ {
+ "epoch": 65.80108991825612,
+ "grad_norm": 1.1475874185562134,
+ "learning_rate": 5.5320710256660146e-06,
+ "loss": 0.0148,
+ "step": 24149
+ },
+ {
+ "epoch": 65.80381471389646,
+ "grad_norm": 2.330589532852173,
+ "learning_rate": 5.531281530344235e-06,
+ "loss": 0.0202,
+ "step": 24150
+ },
+ {
+ "epoch": 65.80653950953679,
+ "grad_norm": 1.7185832262039185,
+ "learning_rate": 5.530492069824689e-06,
+ "loss": 0.0382,
+ "step": 24151
+ },
+ {
+ "epoch": 65.80926430517711,
+ "grad_norm": 1.885694146156311,
+ "learning_rate": 5.529702644113524e-06,
+ "loss": 0.0766,
+ "step": 24152
+ },
+ {
+ "epoch": 65.81198910081744,
+ "grad_norm": 1.2836247682571411,
+ "learning_rate": 5.528913253216896e-06,
+ "loss": 0.0124,
+ "step": 24153
+ },
+ {
+ "epoch": 65.81471389645776,
+ "grad_norm": 2.345731496810913,
+ "learning_rate": 5.528123897140946e-06,
+ "loss": 0.0984,
+ "step": 24154
+ },
+ {
+ "epoch": 65.8174386920981,
+ "grad_norm": 1.6822437047958374,
+ "learning_rate": 5.5273345758918225e-06,
+ "loss": 0.1583,
+ "step": 24155
+ },
+ {
+ "epoch": 65.82016348773843,
+ "grad_norm": 1.6389297246932983,
+ "learning_rate": 5.526545289475671e-06,
+ "loss": 0.0489,
+ "step": 24156
+ },
+ {
+ "epoch": 65.82288828337875,
+ "grad_norm": 1.731408715248108,
+ "learning_rate": 5.525756037898645e-06,
+ "loss": 0.1099,
+ "step": 24157
+ },
+ {
+ "epoch": 65.82561307901908,
+ "grad_norm": 1.3149816989898682,
+ "learning_rate": 5.524966821166886e-06,
+ "loss": 0.0815,
+ "step": 24158
+ },
+ {
+ "epoch": 65.8283378746594,
+ "grad_norm": 2.249696969985962,
+ "learning_rate": 5.524177639286543e-06,
+ "loss": 0.0613,
+ "step": 24159
+ },
+ {
+ "epoch": 65.83106267029973,
+ "grad_norm": 1.4644140005111694,
+ "learning_rate": 5.523388492263755e-06,
+ "loss": 0.0202,
+ "step": 24160
+ },
+ {
+ "epoch": 65.83378746594005,
+ "grad_norm": 1.8959547281265259,
+ "learning_rate": 5.522599380104677e-06,
+ "loss": 0.0194,
+ "step": 24161
+ },
+ {
+ "epoch": 65.83651226158038,
+ "grad_norm": 1.7307898998260498,
+ "learning_rate": 5.521810302815453e-06,
+ "loss": 0.0213,
+ "step": 24162
+ },
+ {
+ "epoch": 65.83923705722071,
+ "grad_norm": 1.2276042699813843,
+ "learning_rate": 5.521021260402224e-06,
+ "loss": 0.0109,
+ "step": 24163
+ },
+ {
+ "epoch": 65.84196185286103,
+ "grad_norm": 1.612969160079956,
+ "learning_rate": 5.520232252871134e-06,
+ "loss": 0.0292,
+ "step": 24164
+ },
+ {
+ "epoch": 65.84468664850137,
+ "grad_norm": 1.607504963874817,
+ "learning_rate": 5.519443280228335e-06,
+ "loss": 0.049,
+ "step": 24165
+ },
+ {
+ "epoch": 65.84741144414168,
+ "grad_norm": 1.7301859855651855,
+ "learning_rate": 5.5186543424799636e-06,
+ "loss": 0.0816,
+ "step": 24166
+ },
+ {
+ "epoch": 65.85013623978202,
+ "grad_norm": 2.5549585819244385,
+ "learning_rate": 5.517865439632173e-06,
+ "loss": 0.0444,
+ "step": 24167
+ },
+ {
+ "epoch": 65.85286103542235,
+ "grad_norm": 1.481102705001831,
+ "learning_rate": 5.517076571691096e-06,
+ "loss": 0.0467,
+ "step": 24168
+ },
+ {
+ "epoch": 65.85558583106267,
+ "grad_norm": 1.3948112726211548,
+ "learning_rate": 5.516287738662887e-06,
+ "loss": 0.0141,
+ "step": 24169
+ },
+ {
+ "epoch": 65.858310626703,
+ "grad_norm": 1.2381840944290161,
+ "learning_rate": 5.515498940553685e-06,
+ "loss": 0.0349,
+ "step": 24170
+ },
+ {
+ "epoch": 65.86103542234332,
+ "grad_norm": 1.8075847625732422,
+ "learning_rate": 5.514710177369634e-06,
+ "loss": 0.0406,
+ "step": 24171
+ },
+ {
+ "epoch": 65.86376021798365,
+ "grad_norm": 1.9491994380950928,
+ "learning_rate": 5.513921449116871e-06,
+ "loss": 0.019,
+ "step": 24172
+ },
+ {
+ "epoch": 65.86648501362397,
+ "grad_norm": 1.3183695077896118,
+ "learning_rate": 5.5131327558015466e-06,
+ "loss": 0.0288,
+ "step": 24173
+ },
+ {
+ "epoch": 65.8692098092643,
+ "grad_norm": 1.9738186597824097,
+ "learning_rate": 5.512344097429801e-06,
+ "loss": 0.0962,
+ "step": 24174
+ },
+ {
+ "epoch": 65.87193460490464,
+ "grad_norm": 1.4954285621643066,
+ "learning_rate": 5.511555474007775e-06,
+ "loss": 0.0302,
+ "step": 24175
+ },
+ {
+ "epoch": 65.87465940054496,
+ "grad_norm": 1.3692611455917358,
+ "learning_rate": 5.510766885541606e-06,
+ "loss": 0.0184,
+ "step": 24176
+ },
+ {
+ "epoch": 65.87738419618529,
+ "grad_norm": 1.8646847009658813,
+ "learning_rate": 5.509978332037444e-06,
+ "loss": 0.1123,
+ "step": 24177
+ },
+ {
+ "epoch": 65.88010899182561,
+ "grad_norm": 2.046912908554077,
+ "learning_rate": 5.509189813501427e-06,
+ "loss": 0.0338,
+ "step": 24178
+ },
+ {
+ "epoch": 65.88283378746594,
+ "grad_norm": 1.4460746049880981,
+ "learning_rate": 5.508401329939695e-06,
+ "loss": 0.0167,
+ "step": 24179
+ },
+ {
+ "epoch": 65.88555858310627,
+ "grad_norm": 2.0004453659057617,
+ "learning_rate": 5.507612881358384e-06,
+ "loss": 0.2003,
+ "step": 24180
+ },
+ {
+ "epoch": 65.88828337874659,
+ "grad_norm": 1.1063865423202515,
+ "learning_rate": 5.506824467763645e-06,
+ "loss": 0.0193,
+ "step": 24181
+ },
+ {
+ "epoch": 65.89100817438693,
+ "grad_norm": 1.3491113185882568,
+ "learning_rate": 5.506036089161611e-06,
+ "loss": 0.022,
+ "step": 24182
+ },
+ {
+ "epoch": 65.89373297002724,
+ "grad_norm": 1.6664026975631714,
+ "learning_rate": 5.505247745558419e-06,
+ "loss": 0.0239,
+ "step": 24183
+ },
+ {
+ "epoch": 65.89645776566758,
+ "grad_norm": 1.5702956914901733,
+ "learning_rate": 5.504459436960218e-06,
+ "loss": 0.0238,
+ "step": 24184
+ },
+ {
+ "epoch": 65.8991825613079,
+ "grad_norm": 1.5617505311965942,
+ "learning_rate": 5.5036711633731375e-06,
+ "loss": 0.0225,
+ "step": 24185
+ },
+ {
+ "epoch": 65.90190735694823,
+ "grad_norm": 2.2005536556243896,
+ "learning_rate": 5.502882924803326e-06,
+ "loss": 0.0398,
+ "step": 24186
+ },
+ {
+ "epoch": 65.90463215258856,
+ "grad_norm": 2.196323871612549,
+ "learning_rate": 5.502094721256916e-06,
+ "loss": 0.03,
+ "step": 24187
+ },
+ {
+ "epoch": 65.90735694822888,
+ "grad_norm": 0.9705361723899841,
+ "learning_rate": 5.501306552740049e-06,
+ "loss": 0.0121,
+ "step": 24188
+ },
+ {
+ "epoch": 65.91008174386921,
+ "grad_norm": 1.0684670209884644,
+ "learning_rate": 5.500518419258858e-06,
+ "loss": 0.1003,
+ "step": 24189
+ },
+ {
+ "epoch": 65.91280653950953,
+ "grad_norm": 1.2890361547470093,
+ "learning_rate": 5.499730320819487e-06,
+ "loss": 0.017,
+ "step": 24190
+ },
+ {
+ "epoch": 65.91553133514986,
+ "grad_norm": 1.6082159280776978,
+ "learning_rate": 5.498942257428073e-06,
+ "loss": 0.0431,
+ "step": 24191
+ },
+ {
+ "epoch": 65.9182561307902,
+ "grad_norm": 1.6980682611465454,
+ "learning_rate": 5.498154229090748e-06,
+ "loss": 0.0318,
+ "step": 24192
+ },
+ {
+ "epoch": 65.92098092643052,
+ "grad_norm": 1.709813117980957,
+ "learning_rate": 5.497366235813656e-06,
+ "loss": 0.0881,
+ "step": 24193
+ },
+ {
+ "epoch": 65.92370572207085,
+ "grad_norm": 1.5327459573745728,
+ "learning_rate": 5.496578277602931e-06,
+ "loss": 0.0138,
+ "step": 24194
+ },
+ {
+ "epoch": 65.92643051771117,
+ "grad_norm": 1.570581078529358,
+ "learning_rate": 5.49579035446471e-06,
+ "loss": 0.0364,
+ "step": 24195
+ },
+ {
+ "epoch": 65.9291553133515,
+ "grad_norm": 1.2189630270004272,
+ "learning_rate": 5.495002466405124e-06,
+ "loss": 0.0132,
+ "step": 24196
+ },
+ {
+ "epoch": 65.93188010899182,
+ "grad_norm": 2.0572900772094727,
+ "learning_rate": 5.4942146134303175e-06,
+ "loss": 0.0549,
+ "step": 24197
+ },
+ {
+ "epoch": 65.93460490463215,
+ "grad_norm": 1.9994819164276123,
+ "learning_rate": 5.493426795546424e-06,
+ "loss": 0.0645,
+ "step": 24198
+ },
+ {
+ "epoch": 65.93732970027249,
+ "grad_norm": 1.2253408432006836,
+ "learning_rate": 5.492639012759576e-06,
+ "loss": 0.0926,
+ "step": 24199
+ },
+ {
+ "epoch": 65.9400544959128,
+ "grad_norm": 1.5941829681396484,
+ "learning_rate": 5.491851265075907e-06,
+ "loss": 0.0256,
+ "step": 24200
+ },
+ {
+ "epoch": 65.94277929155314,
+ "grad_norm": 2.1128671169281006,
+ "learning_rate": 5.491063552501557e-06,
+ "loss": 0.0456,
+ "step": 24201
+ },
+ {
+ "epoch": 65.94550408719346,
+ "grad_norm": 1.3803824186325073,
+ "learning_rate": 5.4902758750426596e-06,
+ "loss": 0.1241,
+ "step": 24202
+ },
+ {
+ "epoch": 65.94822888283379,
+ "grad_norm": 1.3127923011779785,
+ "learning_rate": 5.489488232705351e-06,
+ "loss": 0.0142,
+ "step": 24203
+ },
+ {
+ "epoch": 65.95095367847412,
+ "grad_norm": 1.4292573928833008,
+ "learning_rate": 5.488700625495762e-06,
+ "loss": 0.0307,
+ "step": 24204
+ },
+ {
+ "epoch": 65.95367847411444,
+ "grad_norm": 1.5286792516708374,
+ "learning_rate": 5.487913053420023e-06,
+ "loss": 0.0234,
+ "step": 24205
+ },
+ {
+ "epoch": 65.95640326975477,
+ "grad_norm": 2.8672406673431396,
+ "learning_rate": 5.487125516484276e-06,
+ "loss": 0.0651,
+ "step": 24206
+ },
+ {
+ "epoch": 65.95912806539509,
+ "grad_norm": 1.2740075588226318,
+ "learning_rate": 5.486338014694651e-06,
+ "loss": 0.0121,
+ "step": 24207
+ },
+ {
+ "epoch": 65.96185286103542,
+ "grad_norm": 1.6192704439163208,
+ "learning_rate": 5.485550548057279e-06,
+ "loss": 0.0191,
+ "step": 24208
+ },
+ {
+ "epoch": 65.96457765667574,
+ "grad_norm": 1.150772213935852,
+ "learning_rate": 5.4847631165782914e-06,
+ "loss": 0.0171,
+ "step": 24209
+ },
+ {
+ "epoch": 65.96730245231608,
+ "grad_norm": 1.2158236503601074,
+ "learning_rate": 5.483975720263827e-06,
+ "loss": 0.0395,
+ "step": 24210
+ },
+ {
+ "epoch": 65.97002724795641,
+ "grad_norm": 1.8530575037002563,
+ "learning_rate": 5.4831883591200145e-06,
+ "loss": 0.0777,
+ "step": 24211
+ },
+ {
+ "epoch": 65.97275204359673,
+ "grad_norm": 1.289551019668579,
+ "learning_rate": 5.482401033152984e-06,
+ "loss": 0.155,
+ "step": 24212
+ },
+ {
+ "epoch": 65.97547683923706,
+ "grad_norm": 1.2991619110107422,
+ "learning_rate": 5.481613742368868e-06,
+ "loss": 0.1594,
+ "step": 24213
+ },
+ {
+ "epoch": 65.97820163487738,
+ "grad_norm": 1.6271849870681763,
+ "learning_rate": 5.4808264867738e-06,
+ "loss": 0.0223,
+ "step": 24214
+ },
+ {
+ "epoch": 65.98092643051771,
+ "grad_norm": 0.6770450472831726,
+ "learning_rate": 5.480039266373911e-06,
+ "loss": 0.0068,
+ "step": 24215
+ },
+ {
+ "epoch": 65.98365122615803,
+ "grad_norm": 1.7843492031097412,
+ "learning_rate": 5.479252081175326e-06,
+ "loss": 0.1077,
+ "step": 24216
+ },
+ {
+ "epoch": 65.98637602179836,
+ "grad_norm": 1.2631930112838745,
+ "learning_rate": 5.478464931184186e-06,
+ "loss": 0.0276,
+ "step": 24217
+ },
+ {
+ "epoch": 65.9891008174387,
+ "grad_norm": 1.8448091745376587,
+ "learning_rate": 5.477677816406611e-06,
+ "loss": 0.0481,
+ "step": 24218
+ },
+ {
+ "epoch": 65.99182561307902,
+ "grad_norm": 1.6419901847839355,
+ "learning_rate": 5.4768907368487375e-06,
+ "loss": 0.0229,
+ "step": 24219
+ },
+ {
+ "epoch": 65.99455040871935,
+ "grad_norm": 4.150572299957275,
+ "learning_rate": 5.4761036925166946e-06,
+ "loss": 0.0266,
+ "step": 24220
+ },
+ {
+ "epoch": 65.99727520435967,
+ "grad_norm": 1.4925336837768555,
+ "learning_rate": 5.475316683416607e-06,
+ "loss": 0.0177,
+ "step": 24221
+ },
+ {
+ "epoch": 66.0,
+ "grad_norm": 2.2911179065704346,
+ "learning_rate": 5.4745297095546125e-06,
+ "loss": 0.1,
+ "step": 24222
+ },
+ {
+ "epoch": 66.00272479564033,
+ "grad_norm": 1.8888612985610962,
+ "learning_rate": 5.4737427709368326e-06,
+ "loss": 0.0194,
+ "step": 24223
+ },
+ {
+ "epoch": 66.00544959128065,
+ "grad_norm": 1.4271801710128784,
+ "learning_rate": 5.472955867569399e-06,
+ "loss": 0.0323,
+ "step": 24224
+ },
+ {
+ "epoch": 66.00817438692098,
+ "grad_norm": 1.741093397140503,
+ "learning_rate": 5.472168999458436e-06,
+ "loss": 0.0449,
+ "step": 24225
+ },
+ {
+ "epoch": 66.0108991825613,
+ "grad_norm": 1.6312942504882812,
+ "learning_rate": 5.4713821666100794e-06,
+ "loss": 0.0181,
+ "step": 24226
+ },
+ {
+ "epoch": 66.01362397820164,
+ "grad_norm": 1.8817003965377808,
+ "learning_rate": 5.470595369030453e-06,
+ "loss": 0.0155,
+ "step": 24227
+ },
+ {
+ "epoch": 66.01634877384195,
+ "grad_norm": 1.2590198516845703,
+ "learning_rate": 5.4698086067256826e-06,
+ "loss": 0.0159,
+ "step": 24228
+ },
+ {
+ "epoch": 66.01907356948229,
+ "grad_norm": 1.4518071413040161,
+ "learning_rate": 5.469021879701895e-06,
+ "loss": 0.0314,
+ "step": 24229
+ },
+ {
+ "epoch": 66.02179836512262,
+ "grad_norm": 1.492693305015564,
+ "learning_rate": 5.468235187965221e-06,
+ "loss": 0.0322,
+ "step": 24230
+ },
+ {
+ "epoch": 66.02452316076294,
+ "grad_norm": 1.4209156036376953,
+ "learning_rate": 5.467448531521786e-06,
+ "loss": 0.0352,
+ "step": 24231
+ },
+ {
+ "epoch": 66.02724795640327,
+ "grad_norm": 1.6465928554534912,
+ "learning_rate": 5.466661910377717e-06,
+ "loss": 0.0295,
+ "step": 24232
+ },
+ {
+ "epoch": 66.02997275204359,
+ "grad_norm": 1.751296043395996,
+ "learning_rate": 5.465875324539134e-06,
+ "loss": 0.1294,
+ "step": 24233
+ },
+ {
+ "epoch": 66.03269754768392,
+ "grad_norm": 2.099451780319214,
+ "learning_rate": 5.465088774012172e-06,
+ "loss": 0.0197,
+ "step": 24234
+ },
+ {
+ "epoch": 66.03542234332426,
+ "grad_norm": 0.7576842904090881,
+ "learning_rate": 5.464302258802949e-06,
+ "loss": 0.0088,
+ "step": 24235
+ },
+ {
+ "epoch": 66.03814713896458,
+ "grad_norm": 1.5006067752838135,
+ "learning_rate": 5.463515778917598e-06,
+ "loss": 0.0267,
+ "step": 24236
+ },
+ {
+ "epoch": 66.04087193460491,
+ "grad_norm": 1.6755595207214355,
+ "learning_rate": 5.462729334362238e-06,
+ "loss": 0.0594,
+ "step": 24237
+ },
+ {
+ "epoch": 66.04359673024523,
+ "grad_norm": 1.7333229780197144,
+ "learning_rate": 5.461942925142998e-06,
+ "loss": 0.047,
+ "step": 24238
+ },
+ {
+ "epoch": 66.04632152588556,
+ "grad_norm": 1.1427693367004395,
+ "learning_rate": 5.4611565512660005e-06,
+ "loss": 0.0187,
+ "step": 24239
+ },
+ {
+ "epoch": 66.04904632152588,
+ "grad_norm": 1.8260281085968018,
+ "learning_rate": 5.46037021273737e-06,
+ "loss": 0.0357,
+ "step": 24240
+ },
+ {
+ "epoch": 66.05177111716621,
+ "grad_norm": 2.030900239944458,
+ "learning_rate": 5.459583909563226e-06,
+ "loss": 0.019,
+ "step": 24241
+ },
+ {
+ "epoch": 66.05449591280654,
+ "grad_norm": 1.9425435066223145,
+ "learning_rate": 5.4587976417497e-06,
+ "loss": 0.0849,
+ "step": 24242
+ },
+ {
+ "epoch": 66.05722070844686,
+ "grad_norm": 1.9045474529266357,
+ "learning_rate": 5.458011409302913e-06,
+ "loss": 0.0189,
+ "step": 24243
+ },
+ {
+ "epoch": 66.0599455040872,
+ "grad_norm": 1.4386680126190186,
+ "learning_rate": 5.457225212228987e-06,
+ "loss": 0.0186,
+ "step": 24244
+ },
+ {
+ "epoch": 66.06267029972751,
+ "grad_norm": 1.3220690488815308,
+ "learning_rate": 5.456439050534039e-06,
+ "loss": 0.0291,
+ "step": 24245
+ },
+ {
+ "epoch": 66.06539509536785,
+ "grad_norm": 11.7991361618042,
+ "learning_rate": 5.455652924224203e-06,
+ "loss": 0.0508,
+ "step": 24246
+ },
+ {
+ "epoch": 66.06811989100818,
+ "grad_norm": 1.5166254043579102,
+ "learning_rate": 5.4548668333055954e-06,
+ "loss": 0.0168,
+ "step": 24247
+ },
+ {
+ "epoch": 66.0708446866485,
+ "grad_norm": 1.9580872058868408,
+ "learning_rate": 5.454080777784338e-06,
+ "loss": 0.0523,
+ "step": 24248
+ },
+ {
+ "epoch": 66.07356948228883,
+ "grad_norm": 1.4869930744171143,
+ "learning_rate": 5.453294757666549e-06,
+ "loss": 0.018,
+ "step": 24249
+ },
+ {
+ "epoch": 66.07629427792915,
+ "grad_norm": 0.7562743425369263,
+ "learning_rate": 5.45250877295836e-06,
+ "loss": 0.0109,
+ "step": 24250
+ },
+ {
+ "epoch": 66.07901907356948,
+ "grad_norm": 1.4313998222351074,
+ "learning_rate": 5.451722823665885e-06,
+ "loss": 0.1167,
+ "step": 24251
+ },
+ {
+ "epoch": 66.0817438692098,
+ "grad_norm": 1.5456514358520508,
+ "learning_rate": 5.4509369097952425e-06,
+ "loss": 0.0391,
+ "step": 24252
+ },
+ {
+ "epoch": 66.08446866485014,
+ "grad_norm": 1.4788260459899902,
+ "learning_rate": 5.45015103135256e-06,
+ "loss": 0.0279,
+ "step": 24253
+ },
+ {
+ "epoch": 66.08719346049047,
+ "grad_norm": 1.8597185611724854,
+ "learning_rate": 5.449365188343952e-06,
+ "loss": 0.0855,
+ "step": 24254
+ },
+ {
+ "epoch": 66.08991825613079,
+ "grad_norm": 1.448919653892517,
+ "learning_rate": 5.448579380775545e-06,
+ "loss": 0.0163,
+ "step": 24255
+ },
+ {
+ "epoch": 66.09264305177112,
+ "grad_norm": 1.285521388053894,
+ "learning_rate": 5.447793608653456e-06,
+ "loss": 0.0133,
+ "step": 24256
+ },
+ {
+ "epoch": 66.09536784741144,
+ "grad_norm": 1.2943403720855713,
+ "learning_rate": 5.447007871983799e-06,
+ "loss": 0.0206,
+ "step": 24257
+ },
+ {
+ "epoch": 66.09809264305177,
+ "grad_norm": 1.3667006492614746,
+ "learning_rate": 5.446222170772703e-06,
+ "loss": 0.0144,
+ "step": 24258
+ },
+ {
+ "epoch": 66.1008174386921,
+ "grad_norm": 0.8837658762931824,
+ "learning_rate": 5.44543650502628e-06,
+ "loss": 0.009,
+ "step": 24259
+ },
+ {
+ "epoch": 66.10354223433242,
+ "grad_norm": 1.2521817684173584,
+ "learning_rate": 5.444650874750654e-06,
+ "loss": 0.0621,
+ "step": 24260
+ },
+ {
+ "epoch": 66.10626702997276,
+ "grad_norm": 1.8257954120635986,
+ "learning_rate": 5.443865279951934e-06,
+ "loss": 0.0343,
+ "step": 24261
+ },
+ {
+ "epoch": 66.10899182561307,
+ "grad_norm": 1.657636284828186,
+ "learning_rate": 5.443079720636251e-06,
+ "loss": 0.0813,
+ "step": 24262
+ },
+ {
+ "epoch": 66.11171662125341,
+ "grad_norm": 1.387143850326538,
+ "learning_rate": 5.4422941968097145e-06,
+ "loss": 0.0349,
+ "step": 24263
+ },
+ {
+ "epoch": 66.11444141689373,
+ "grad_norm": 2.561318874359131,
+ "learning_rate": 5.441508708478444e-06,
+ "loss": 0.1397,
+ "step": 24264
+ },
+ {
+ "epoch": 66.11716621253406,
+ "grad_norm": 1.350508451461792,
+ "learning_rate": 5.440723255648553e-06,
+ "loss": 0.0212,
+ "step": 24265
+ },
+ {
+ "epoch": 66.11989100817439,
+ "grad_norm": 1.6894378662109375,
+ "learning_rate": 5.439937838326168e-06,
+ "loss": 0.02,
+ "step": 24266
+ },
+ {
+ "epoch": 66.12261580381471,
+ "grad_norm": 1.577636480331421,
+ "learning_rate": 5.439152456517398e-06,
+ "loss": 0.0316,
+ "step": 24267
+ },
+ {
+ "epoch": 66.12534059945504,
+ "grad_norm": 1.366362452507019,
+ "learning_rate": 5.438367110228364e-06,
+ "loss": 0.0241,
+ "step": 24268
+ },
+ {
+ "epoch": 66.12806539509536,
+ "grad_norm": 1.398724913597107,
+ "learning_rate": 5.437581799465174e-06,
+ "loss": 0.0919,
+ "step": 24269
+ },
+ {
+ "epoch": 66.1307901907357,
+ "grad_norm": 1.9866470098495483,
+ "learning_rate": 5.436796524233953e-06,
+ "loss": 0.07,
+ "step": 24270
+ },
+ {
+ "epoch": 66.13351498637603,
+ "grad_norm": 2.5341765880584717,
+ "learning_rate": 5.436011284540816e-06,
+ "loss": 0.0164,
+ "step": 24271
+ },
+ {
+ "epoch": 66.13623978201635,
+ "grad_norm": 2.2858119010925293,
+ "learning_rate": 5.435226080391876e-06,
+ "loss": 0.0445,
+ "step": 24272
+ },
+ {
+ "epoch": 66.13896457765668,
+ "grad_norm": 1.7223758697509766,
+ "learning_rate": 5.43444091179325e-06,
+ "loss": 0.0207,
+ "step": 24273
+ },
+ {
+ "epoch": 66.141689373297,
+ "grad_norm": 1.5186704397201538,
+ "learning_rate": 5.433655778751046e-06,
+ "loss": 0.1092,
+ "step": 24274
+ },
+ {
+ "epoch": 66.14441416893733,
+ "grad_norm": 2.089057445526123,
+ "learning_rate": 5.432870681271387e-06,
+ "loss": 0.1497,
+ "step": 24275
+ },
+ {
+ "epoch": 66.14713896457765,
+ "grad_norm": 1.7316601276397705,
+ "learning_rate": 5.432085619360385e-06,
+ "loss": 0.0444,
+ "step": 24276
+ },
+ {
+ "epoch": 66.14986376021798,
+ "grad_norm": 1.2200390100479126,
+ "learning_rate": 5.431300593024154e-06,
+ "loss": 0.035,
+ "step": 24277
+ },
+ {
+ "epoch": 66.15258855585832,
+ "grad_norm": 1.5898423194885254,
+ "learning_rate": 5.430515602268802e-06,
+ "loss": 0.0455,
+ "step": 24278
+ },
+ {
+ "epoch": 66.15531335149863,
+ "grad_norm": 14.549177169799805,
+ "learning_rate": 5.429730647100452e-06,
+ "loss": 0.1448,
+ "step": 24279
+ },
+ {
+ "epoch": 66.15803814713897,
+ "grad_norm": 1.0222342014312744,
+ "learning_rate": 5.428945727525212e-06,
+ "loss": 0.0072,
+ "step": 24280
+ },
+ {
+ "epoch": 66.16076294277929,
+ "grad_norm": 1.6872119903564453,
+ "learning_rate": 5.428160843549192e-06,
+ "loss": 0.0813,
+ "step": 24281
+ },
+ {
+ "epoch": 66.16348773841962,
+ "grad_norm": 1.853369951248169,
+ "learning_rate": 5.427375995178512e-06,
+ "loss": 0.0204,
+ "step": 24282
+ },
+ {
+ "epoch": 66.16621253405995,
+ "grad_norm": 1.4285489320755005,
+ "learning_rate": 5.426591182419281e-06,
+ "loss": 0.1625,
+ "step": 24283
+ },
+ {
+ "epoch": 66.16893732970027,
+ "grad_norm": 2.4409642219543457,
+ "learning_rate": 5.425806405277609e-06,
+ "loss": 0.1367,
+ "step": 24284
+ },
+ {
+ "epoch": 66.1716621253406,
+ "grad_norm": 1.3194302320480347,
+ "learning_rate": 5.425021663759607e-06,
+ "loss": 0.0163,
+ "step": 24285
+ },
+ {
+ "epoch": 66.17438692098092,
+ "grad_norm": 1.4248631000518799,
+ "learning_rate": 5.424236957871389e-06,
+ "loss": 0.0544,
+ "step": 24286
+ },
+ {
+ "epoch": 66.17711171662125,
+ "grad_norm": 1.9970794916152954,
+ "learning_rate": 5.423452287619071e-06,
+ "loss": 0.0213,
+ "step": 24287
+ },
+ {
+ "epoch": 66.17983651226157,
+ "grad_norm": 1.8272532224655151,
+ "learning_rate": 5.4226676530087575e-06,
+ "loss": 0.0127,
+ "step": 24288
+ },
+ {
+ "epoch": 66.1825613079019,
+ "grad_norm": 1.3733787536621094,
+ "learning_rate": 5.421883054046562e-06,
+ "loss": 0.0208,
+ "step": 24289
+ },
+ {
+ "epoch": 66.18528610354224,
+ "grad_norm": 2.5201239585876465,
+ "learning_rate": 5.42109849073859e-06,
+ "loss": 0.0196,
+ "step": 24290
+ },
+ {
+ "epoch": 66.18801089918256,
+ "grad_norm": 1.7730869054794312,
+ "learning_rate": 5.42031396309096e-06,
+ "loss": 0.0674,
+ "step": 24291
+ },
+ {
+ "epoch": 66.19073569482289,
+ "grad_norm": 2.142868757247925,
+ "learning_rate": 5.4195294711097766e-06,
+ "loss": 0.0192,
+ "step": 24292
+ },
+ {
+ "epoch": 66.19346049046321,
+ "grad_norm": 1.607856273651123,
+ "learning_rate": 5.41874501480115e-06,
+ "loss": 0.0235,
+ "step": 24293
+ },
+ {
+ "epoch": 66.19618528610354,
+ "grad_norm": 2.204760789871216,
+ "learning_rate": 5.417960594171187e-06,
+ "loss": 0.0335,
+ "step": 24294
+ },
+ {
+ "epoch": 66.19891008174388,
+ "grad_norm": 1.6764276027679443,
+ "learning_rate": 5.417176209226003e-06,
+ "loss": 0.0148,
+ "step": 24295
+ },
+ {
+ "epoch": 66.2016348773842,
+ "grad_norm": 2.299558639526367,
+ "learning_rate": 5.4163918599717015e-06,
+ "loss": 0.0147,
+ "step": 24296
+ },
+ {
+ "epoch": 66.20435967302453,
+ "grad_norm": 1.7724584341049194,
+ "learning_rate": 5.415607546414394e-06,
+ "loss": 0.0443,
+ "step": 24297
+ },
+ {
+ "epoch": 66.20708446866485,
+ "grad_norm": 1.6446233987808228,
+ "learning_rate": 5.414823268560184e-06,
+ "loss": 0.1149,
+ "step": 24298
+ },
+ {
+ "epoch": 66.20980926430518,
+ "grad_norm": 1.1751112937927246,
+ "learning_rate": 5.414039026415187e-06,
+ "loss": 0.0118,
+ "step": 24299
+ },
+ {
+ "epoch": 66.2125340599455,
+ "grad_norm": 1.3980494737625122,
+ "learning_rate": 5.413254819985505e-06,
+ "loss": 0.0263,
+ "step": 24300
+ },
+ {
+ "epoch": 66.21525885558583,
+ "grad_norm": 1.4233546257019043,
+ "learning_rate": 5.412470649277244e-06,
+ "loss": 0.0216,
+ "step": 24301
+ },
+ {
+ "epoch": 66.21798365122616,
+ "grad_norm": 1.2770497798919678,
+ "learning_rate": 5.411686514296517e-06,
+ "loss": 0.0216,
+ "step": 24302
+ },
+ {
+ "epoch": 66.22070844686648,
+ "grad_norm": 1.717175006866455,
+ "learning_rate": 5.410902415049428e-06,
+ "loss": 0.0196,
+ "step": 24303
+ },
+ {
+ "epoch": 66.22343324250681,
+ "grad_norm": 1.613610029220581,
+ "learning_rate": 5.410118351542079e-06,
+ "loss": 0.0267,
+ "step": 24304
+ },
+ {
+ "epoch": 66.22615803814713,
+ "grad_norm": 1.2811676263809204,
+ "learning_rate": 5.409334323780586e-06,
+ "loss": 0.0213,
+ "step": 24305
+ },
+ {
+ "epoch": 66.22888283378747,
+ "grad_norm": 2.6505908966064453,
+ "learning_rate": 5.408550331771045e-06,
+ "loss": 0.141,
+ "step": 24306
+ },
+ {
+ "epoch": 66.2316076294278,
+ "grad_norm": 1.9367648363113403,
+ "learning_rate": 5.40776637551957e-06,
+ "loss": 0.0172,
+ "step": 24307
+ },
+ {
+ "epoch": 66.23433242506812,
+ "grad_norm": 1.355636477470398,
+ "learning_rate": 5.4069824550322614e-06,
+ "loss": 0.0382,
+ "step": 24308
+ },
+ {
+ "epoch": 66.23705722070845,
+ "grad_norm": 4.628684043884277,
+ "learning_rate": 5.406198570315227e-06,
+ "loss": 0.0292,
+ "step": 24309
+ },
+ {
+ "epoch": 66.23978201634877,
+ "grad_norm": 0.9034616947174072,
+ "learning_rate": 5.405414721374567e-06,
+ "loss": 0.0072,
+ "step": 24310
+ },
+ {
+ "epoch": 66.2425068119891,
+ "grad_norm": 1.4663053750991821,
+ "learning_rate": 5.404630908216391e-06,
+ "loss": 0.0632,
+ "step": 24311
+ },
+ {
+ "epoch": 66.24523160762942,
+ "grad_norm": 1.5819799900054932,
+ "learning_rate": 5.4038471308468046e-06,
+ "loss": 0.0178,
+ "step": 24312
+ },
+ {
+ "epoch": 66.24795640326975,
+ "grad_norm": 1.6509456634521484,
+ "learning_rate": 5.403063389271906e-06,
+ "loss": 0.0789,
+ "step": 24313
+ },
+ {
+ "epoch": 66.25068119891009,
+ "grad_norm": 2.6131584644317627,
+ "learning_rate": 5.4022796834978e-06,
+ "loss": 0.1221,
+ "step": 24314
+ },
+ {
+ "epoch": 66.2534059945504,
+ "grad_norm": 2.3241307735443115,
+ "learning_rate": 5.401496013530594e-06,
+ "loss": 0.0734,
+ "step": 24315
+ },
+ {
+ "epoch": 66.25613079019074,
+ "grad_norm": 1.6116302013397217,
+ "learning_rate": 5.400712379376389e-06,
+ "loss": 0.0426,
+ "step": 24316
+ },
+ {
+ "epoch": 66.25885558583106,
+ "grad_norm": 2.4266397953033447,
+ "learning_rate": 5.399928781041289e-06,
+ "loss": 0.0709,
+ "step": 24317
+ },
+ {
+ "epoch": 66.26158038147139,
+ "grad_norm": 0.9970799684524536,
+ "learning_rate": 5.399145218531392e-06,
+ "loss": 0.0101,
+ "step": 24318
+ },
+ {
+ "epoch": 66.26430517711172,
+ "grad_norm": 1.060912847518921,
+ "learning_rate": 5.398361691852807e-06,
+ "loss": 0.016,
+ "step": 24319
+ },
+ {
+ "epoch": 66.26702997275204,
+ "grad_norm": 2.460125684738159,
+ "learning_rate": 5.397578201011633e-06,
+ "loss": 0.1679,
+ "step": 24320
+ },
+ {
+ "epoch": 66.26975476839237,
+ "grad_norm": 2.3198840618133545,
+ "learning_rate": 5.396794746013969e-06,
+ "loss": 0.0601,
+ "step": 24321
+ },
+ {
+ "epoch": 66.2724795640327,
+ "grad_norm": 1.62663996219635,
+ "learning_rate": 5.396011326865918e-06,
+ "loss": 0.0454,
+ "step": 24322
+ },
+ {
+ "epoch": 66.27520435967303,
+ "grad_norm": 0.7050876021385193,
+ "learning_rate": 5.395227943573588e-06,
+ "loss": 0.0101,
+ "step": 24323
+ },
+ {
+ "epoch": 66.27792915531334,
+ "grad_norm": 1.7202494144439697,
+ "learning_rate": 5.394444596143074e-06,
+ "loss": 0.0353,
+ "step": 24324
+ },
+ {
+ "epoch": 66.28065395095368,
+ "grad_norm": 2.198603868484497,
+ "learning_rate": 5.393661284580476e-06,
+ "loss": 0.1068,
+ "step": 24325
+ },
+ {
+ "epoch": 66.28337874659401,
+ "grad_norm": 1.519546627998352,
+ "learning_rate": 5.392878008891893e-06,
+ "loss": 0.0264,
+ "step": 24326
+ },
+ {
+ "epoch": 66.28610354223433,
+ "grad_norm": 2.0428292751312256,
+ "learning_rate": 5.3920947690834315e-06,
+ "loss": 0.1193,
+ "step": 24327
+ },
+ {
+ "epoch": 66.28882833787466,
+ "grad_norm": 1.922265887260437,
+ "learning_rate": 5.391311565161188e-06,
+ "loss": 0.0298,
+ "step": 24328
+ },
+ {
+ "epoch": 66.29155313351498,
+ "grad_norm": 1.5999006032943726,
+ "learning_rate": 5.39052839713126e-06,
+ "loss": 0.207,
+ "step": 24329
+ },
+ {
+ "epoch": 66.29427792915531,
+ "grad_norm": 1.5286766290664673,
+ "learning_rate": 5.389745264999746e-06,
+ "loss": 0.0163,
+ "step": 24330
+ },
+ {
+ "epoch": 66.29700272479565,
+ "grad_norm": 1.138032078742981,
+ "learning_rate": 5.388962168772751e-06,
+ "loss": 0.015,
+ "step": 24331
+ },
+ {
+ "epoch": 66.29972752043597,
+ "grad_norm": 1.8013111352920532,
+ "learning_rate": 5.388179108456371e-06,
+ "loss": 0.0445,
+ "step": 24332
+ },
+ {
+ "epoch": 66.3024523160763,
+ "grad_norm": 1.5043202638626099,
+ "learning_rate": 5.387396084056704e-06,
+ "loss": 0.0747,
+ "step": 24333
+ },
+ {
+ "epoch": 66.30517711171662,
+ "grad_norm": 1.8697929382324219,
+ "learning_rate": 5.386613095579842e-06,
+ "loss": 0.0203,
+ "step": 24334
+ },
+ {
+ "epoch": 66.30790190735695,
+ "grad_norm": 1.5131778717041016,
+ "learning_rate": 5.385830143031894e-06,
+ "loss": 0.023,
+ "step": 24335
+ },
+ {
+ "epoch": 66.31062670299727,
+ "grad_norm": 1.4611785411834717,
+ "learning_rate": 5.385047226418953e-06,
+ "loss": 0.0164,
+ "step": 24336
+ },
+ {
+ "epoch": 66.3133514986376,
+ "grad_norm": 1.2447975873947144,
+ "learning_rate": 5.384264345747114e-06,
+ "loss": 0.0167,
+ "step": 24337
+ },
+ {
+ "epoch": 66.31607629427793,
+ "grad_norm": 1.629892349243164,
+ "learning_rate": 5.383481501022472e-06,
+ "loss": 0.017,
+ "step": 24338
+ },
+ {
+ "epoch": 66.31880108991825,
+ "grad_norm": 1.8316700458526611,
+ "learning_rate": 5.382698692251128e-06,
+ "loss": 0.0267,
+ "step": 24339
+ },
+ {
+ "epoch": 66.32152588555859,
+ "grad_norm": 1.6787619590759277,
+ "learning_rate": 5.381915919439182e-06,
+ "loss": 0.1405,
+ "step": 24340
+ },
+ {
+ "epoch": 66.3242506811989,
+ "grad_norm": 1.2015866041183472,
+ "learning_rate": 5.381133182592726e-06,
+ "loss": 0.0149,
+ "step": 24341
+ },
+ {
+ "epoch": 66.32697547683924,
+ "grad_norm": 4.761872291564941,
+ "learning_rate": 5.380350481717853e-06,
+ "loss": 0.0931,
+ "step": 24342
+ },
+ {
+ "epoch": 66.32970027247957,
+ "grad_norm": 1.6351215839385986,
+ "learning_rate": 5.379567816820664e-06,
+ "loss": 0.0267,
+ "step": 24343
+ },
+ {
+ "epoch": 66.33242506811989,
+ "grad_norm": 1.1913237571716309,
+ "learning_rate": 5.378785187907254e-06,
+ "loss": 0.1058,
+ "step": 24344
+ },
+ {
+ "epoch": 66.33514986376022,
+ "grad_norm": 0.9242094159126282,
+ "learning_rate": 5.378002594983715e-06,
+ "loss": 0.0103,
+ "step": 24345
+ },
+ {
+ "epoch": 66.33787465940054,
+ "grad_norm": 1.8560539484024048,
+ "learning_rate": 5.377220038056139e-06,
+ "loss": 0.0238,
+ "step": 24346
+ },
+ {
+ "epoch": 66.34059945504087,
+ "grad_norm": 1.5676121711730957,
+ "learning_rate": 5.376437517130629e-06,
+ "loss": 0.023,
+ "step": 24347
+ },
+ {
+ "epoch": 66.34332425068119,
+ "grad_norm": 1.9304401874542236,
+ "learning_rate": 5.375655032213275e-06,
+ "loss": 0.0739,
+ "step": 24348
+ },
+ {
+ "epoch": 66.34604904632153,
+ "grad_norm": 2.253999948501587,
+ "learning_rate": 5.374872583310171e-06,
+ "loss": 0.0754,
+ "step": 24349
+ },
+ {
+ "epoch": 66.34877384196186,
+ "grad_norm": 1.302359938621521,
+ "learning_rate": 5.3740901704274065e-06,
+ "loss": 0.12,
+ "step": 24350
+ },
+ {
+ "epoch": 66.35149863760218,
+ "grad_norm": 2.67948842048645,
+ "learning_rate": 5.373307793571081e-06,
+ "loss": 0.0483,
+ "step": 24351
+ },
+ {
+ "epoch": 66.35422343324251,
+ "grad_norm": 1.5562870502471924,
+ "learning_rate": 5.372525452747288e-06,
+ "loss": 0.0307,
+ "step": 24352
+ },
+ {
+ "epoch": 66.35694822888283,
+ "grad_norm": 2.0496973991394043,
+ "learning_rate": 5.3717431479621164e-06,
+ "loss": 0.1052,
+ "step": 24353
+ },
+ {
+ "epoch": 66.35967302452316,
+ "grad_norm": 0.8028293251991272,
+ "learning_rate": 5.370960879221656e-06,
+ "loss": 0.0072,
+ "step": 24354
+ },
+ {
+ "epoch": 66.3623978201635,
+ "grad_norm": 1.081930160522461,
+ "learning_rate": 5.370178646532004e-06,
+ "loss": 0.0179,
+ "step": 24355
+ },
+ {
+ "epoch": 66.36512261580381,
+ "grad_norm": 1.4857187271118164,
+ "learning_rate": 5.369396449899256e-06,
+ "loss": 0.0562,
+ "step": 24356
+ },
+ {
+ "epoch": 66.36784741144415,
+ "grad_norm": 1.8935909271240234,
+ "learning_rate": 5.368614289329499e-06,
+ "loss": 0.0273,
+ "step": 24357
+ },
+ {
+ "epoch": 66.37057220708446,
+ "grad_norm": 1.8185847997665405,
+ "learning_rate": 5.367832164828823e-06,
+ "loss": 0.0495,
+ "step": 24358
+ },
+ {
+ "epoch": 66.3732970027248,
+ "grad_norm": 2.185758590698242,
+ "learning_rate": 5.367050076403319e-06,
+ "loss": 0.0269,
+ "step": 24359
+ },
+ {
+ "epoch": 66.37602179836512,
+ "grad_norm": 1.729148507118225,
+ "learning_rate": 5.366268024059085e-06,
+ "loss": 0.0194,
+ "step": 24360
+ },
+ {
+ "epoch": 66.37874659400545,
+ "grad_norm": 3.0319032669067383,
+ "learning_rate": 5.365486007802206e-06,
+ "loss": 0.0469,
+ "step": 24361
+ },
+ {
+ "epoch": 66.38147138964578,
+ "grad_norm": 1.8443410396575928,
+ "learning_rate": 5.3647040276387695e-06,
+ "loss": 0.0762,
+ "step": 24362
+ },
+ {
+ "epoch": 66.3841961852861,
+ "grad_norm": 1.3080227375030518,
+ "learning_rate": 5.363922083574869e-06,
+ "loss": 0.0174,
+ "step": 24363
+ },
+ {
+ "epoch": 66.38692098092643,
+ "grad_norm": 1.758029818534851,
+ "learning_rate": 5.363140175616596e-06,
+ "loss": 0.0629,
+ "step": 24364
+ },
+ {
+ "epoch": 66.38964577656675,
+ "grad_norm": 2.8025949001312256,
+ "learning_rate": 5.362358303770039e-06,
+ "loss": 0.0833,
+ "step": 24365
+ },
+ {
+ "epoch": 66.39237057220708,
+ "grad_norm": 1.340047836303711,
+ "learning_rate": 5.36157646804128e-06,
+ "loss": 0.0159,
+ "step": 24366
+ },
+ {
+ "epoch": 66.39509536784742,
+ "grad_norm": 2.2411324977874756,
+ "learning_rate": 5.360794668436421e-06,
+ "loss": 0.0964,
+ "step": 24367
+ },
+ {
+ "epoch": 66.39782016348774,
+ "grad_norm": 1.7656344175338745,
+ "learning_rate": 5.360012904961542e-06,
+ "loss": 0.0194,
+ "step": 24368
+ },
+ {
+ "epoch": 66.40054495912807,
+ "grad_norm": 1.7136507034301758,
+ "learning_rate": 5.359231177622735e-06,
+ "loss": 0.118,
+ "step": 24369
+ },
+ {
+ "epoch": 66.40326975476839,
+ "grad_norm": 1.477491855621338,
+ "learning_rate": 5.358449486426081e-06,
+ "loss": 0.0299,
+ "step": 24370
+ },
+ {
+ "epoch": 66.40599455040872,
+ "grad_norm": 1.6492068767547607,
+ "learning_rate": 5.357667831377676e-06,
+ "loss": 0.0141,
+ "step": 24371
+ },
+ {
+ "epoch": 66.40871934604904,
+ "grad_norm": 2.9035375118255615,
+ "learning_rate": 5.356886212483603e-06,
+ "loss": 0.2746,
+ "step": 24372
+ },
+ {
+ "epoch": 66.41144414168937,
+ "grad_norm": 2.660414695739746,
+ "learning_rate": 5.356104629749952e-06,
+ "loss": 0.1021,
+ "step": 24373
+ },
+ {
+ "epoch": 66.4141689373297,
+ "grad_norm": 1.6572052240371704,
+ "learning_rate": 5.355323083182811e-06,
+ "loss": 0.0563,
+ "step": 24374
+ },
+ {
+ "epoch": 66.41689373297002,
+ "grad_norm": 0.895801842212677,
+ "learning_rate": 5.354541572788259e-06,
+ "loss": 0.0091,
+ "step": 24375
+ },
+ {
+ "epoch": 66.41961852861036,
+ "grad_norm": 2.0437839031219482,
+ "learning_rate": 5.353760098572393e-06,
+ "loss": 0.0216,
+ "step": 24376
+ },
+ {
+ "epoch": 66.42234332425068,
+ "grad_norm": 6.749046802520752,
+ "learning_rate": 5.3529786605412925e-06,
+ "loss": 0.0457,
+ "step": 24377
+ },
+ {
+ "epoch": 66.42506811989101,
+ "grad_norm": 1.5361768007278442,
+ "learning_rate": 5.352197258701044e-06,
+ "loss": 0.0164,
+ "step": 24378
+ },
+ {
+ "epoch": 66.42779291553134,
+ "grad_norm": 2.745257616043091,
+ "learning_rate": 5.351415893057732e-06,
+ "loss": 0.0645,
+ "step": 24379
+ },
+ {
+ "epoch": 66.43051771117166,
+ "grad_norm": 1.2246814966201782,
+ "learning_rate": 5.350634563617445e-06,
+ "loss": 0.0131,
+ "step": 24380
+ },
+ {
+ "epoch": 66.433242506812,
+ "grad_norm": 1.3032490015029907,
+ "learning_rate": 5.3498532703862685e-06,
+ "loss": 0.0269,
+ "step": 24381
+ },
+ {
+ "epoch": 66.43596730245231,
+ "grad_norm": 2.9628682136535645,
+ "learning_rate": 5.349072013370285e-06,
+ "loss": 0.0539,
+ "step": 24382
+ },
+ {
+ "epoch": 66.43869209809264,
+ "grad_norm": 2.4955294132232666,
+ "learning_rate": 5.348290792575572e-06,
+ "loss": 0.1513,
+ "step": 24383
+ },
+ {
+ "epoch": 66.44141689373296,
+ "grad_norm": 1.343919038772583,
+ "learning_rate": 5.347509608008228e-06,
+ "loss": 0.0166,
+ "step": 24384
+ },
+ {
+ "epoch": 66.4441416893733,
+ "grad_norm": 1.3166221380233765,
+ "learning_rate": 5.346728459674327e-06,
+ "loss": 0.0869,
+ "step": 24385
+ },
+ {
+ "epoch": 66.44686648501363,
+ "grad_norm": 1.2762991189956665,
+ "learning_rate": 5.3459473475799565e-06,
+ "loss": 0.0205,
+ "step": 24386
+ },
+ {
+ "epoch": 66.44959128065395,
+ "grad_norm": 1.950786828994751,
+ "learning_rate": 5.345166271731195e-06,
+ "loss": 0.0275,
+ "step": 24387
+ },
+ {
+ "epoch": 66.45231607629428,
+ "grad_norm": 1.8260420560836792,
+ "learning_rate": 5.344385232134131e-06,
+ "loss": 0.0356,
+ "step": 24388
+ },
+ {
+ "epoch": 66.4550408719346,
+ "grad_norm": 1.197803020477295,
+ "learning_rate": 5.343604228794845e-06,
+ "loss": 0.0188,
+ "step": 24389
+ },
+ {
+ "epoch": 66.45776566757493,
+ "grad_norm": 1.1394779682159424,
+ "learning_rate": 5.342823261719417e-06,
+ "loss": 0.0108,
+ "step": 24390
+ },
+ {
+ "epoch": 66.46049046321527,
+ "grad_norm": 4.7748847007751465,
+ "learning_rate": 5.34204233091393e-06,
+ "loss": 0.0205,
+ "step": 24391
+ },
+ {
+ "epoch": 66.46321525885558,
+ "grad_norm": 0.891670286655426,
+ "learning_rate": 5.341261436384472e-06,
+ "loss": 0.0108,
+ "step": 24392
+ },
+ {
+ "epoch": 66.46594005449592,
+ "grad_norm": 1.8393478393554688,
+ "learning_rate": 5.34048057813712e-06,
+ "loss": 0.1087,
+ "step": 24393
+ },
+ {
+ "epoch": 66.46866485013624,
+ "grad_norm": 1.4151724576950073,
+ "learning_rate": 5.339699756177954e-06,
+ "loss": 0.0174,
+ "step": 24394
+ },
+ {
+ "epoch": 66.47138964577657,
+ "grad_norm": 1.1459622383117676,
+ "learning_rate": 5.338918970513054e-06,
+ "loss": 0.0975,
+ "step": 24395
+ },
+ {
+ "epoch": 66.47411444141689,
+ "grad_norm": 2.0934488773345947,
+ "learning_rate": 5.338138221148506e-06,
+ "loss": 0.0374,
+ "step": 24396
+ },
+ {
+ "epoch": 66.47683923705722,
+ "grad_norm": 1.7033249139785767,
+ "learning_rate": 5.337357508090388e-06,
+ "loss": 0.027,
+ "step": 24397
+ },
+ {
+ "epoch": 66.47956403269755,
+ "grad_norm": 2.188991069793701,
+ "learning_rate": 5.336576831344781e-06,
+ "loss": 0.0391,
+ "step": 24398
+ },
+ {
+ "epoch": 66.48228882833787,
+ "grad_norm": 1.3671222925186157,
+ "learning_rate": 5.335796190917758e-06,
+ "loss": 0.0188,
+ "step": 24399
+ },
+ {
+ "epoch": 66.4850136239782,
+ "grad_norm": 1.5722448825836182,
+ "learning_rate": 5.335015586815408e-06,
+ "loss": 0.0203,
+ "step": 24400
+ },
+ {
+ "epoch": 66.48773841961852,
+ "grad_norm": 1.231913685798645,
+ "learning_rate": 5.3342350190438075e-06,
+ "loss": 0.0366,
+ "step": 24401
+ },
+ {
+ "epoch": 66.49046321525886,
+ "grad_norm": 3.2483160495758057,
+ "learning_rate": 5.333454487609033e-06,
+ "loss": 0.0242,
+ "step": 24402
+ },
+ {
+ "epoch": 66.49318801089919,
+ "grad_norm": 1.973442554473877,
+ "learning_rate": 5.332673992517162e-06,
+ "loss": 0.1315,
+ "step": 24403
+ },
+ {
+ "epoch": 66.49591280653951,
+ "grad_norm": 1.305861234664917,
+ "learning_rate": 5.3318935337742795e-06,
+ "loss": 0.0124,
+ "step": 24404
+ },
+ {
+ "epoch": 66.49863760217984,
+ "grad_norm": 1.656656265258789,
+ "learning_rate": 5.331113111386459e-06,
+ "loss": 0.016,
+ "step": 24405
+ },
+ {
+ "epoch": 66.50136239782016,
+ "grad_norm": 1.4248768091201782,
+ "learning_rate": 5.330332725359781e-06,
+ "loss": 0.0199,
+ "step": 24406
+ },
+ {
+ "epoch": 66.50408719346049,
+ "grad_norm": 2.454483985900879,
+ "learning_rate": 5.329552375700313e-06,
+ "loss": 0.0295,
+ "step": 24407
+ },
+ {
+ "epoch": 66.50681198910081,
+ "grad_norm": 1.6751371622085571,
+ "learning_rate": 5.328772062414147e-06,
+ "loss": 0.0502,
+ "step": 24408
+ },
+ {
+ "epoch": 66.50953678474114,
+ "grad_norm": 1.8173454999923706,
+ "learning_rate": 5.327991785507356e-06,
+ "loss": 0.0229,
+ "step": 24409
+ },
+ {
+ "epoch": 66.51226158038148,
+ "grad_norm": 1.5927238464355469,
+ "learning_rate": 5.327211544986013e-06,
+ "loss": 0.0369,
+ "step": 24410
+ },
+ {
+ "epoch": 66.5149863760218,
+ "grad_norm": 2.3097541332244873,
+ "learning_rate": 5.326431340856193e-06,
+ "loss": 0.0216,
+ "step": 24411
+ },
+ {
+ "epoch": 66.51771117166213,
+ "grad_norm": 2.7667531967163086,
+ "learning_rate": 5.3256511731239805e-06,
+ "loss": 0.0336,
+ "step": 24412
+ },
+ {
+ "epoch": 66.52043596730245,
+ "grad_norm": 2.0466299057006836,
+ "learning_rate": 5.324871041795444e-06,
+ "loss": 0.1855,
+ "step": 24413
+ },
+ {
+ "epoch": 66.52316076294278,
+ "grad_norm": 1.8321374654769897,
+ "learning_rate": 5.3240909468766625e-06,
+ "loss": 0.1387,
+ "step": 24414
+ },
+ {
+ "epoch": 66.52588555858311,
+ "grad_norm": 1.5223941802978516,
+ "learning_rate": 5.323310888373705e-06,
+ "loss": 0.0313,
+ "step": 24415
+ },
+ {
+ "epoch": 66.52861035422343,
+ "grad_norm": 1.7833616733551025,
+ "learning_rate": 5.322530866292658e-06,
+ "loss": 0.0128,
+ "step": 24416
+ },
+ {
+ "epoch": 66.53133514986376,
+ "grad_norm": 1.4574024677276611,
+ "learning_rate": 5.321750880639589e-06,
+ "loss": 0.0209,
+ "step": 24417
+ },
+ {
+ "epoch": 66.53405994550408,
+ "grad_norm": 2.106764078140259,
+ "learning_rate": 5.320970931420572e-06,
+ "loss": 0.0242,
+ "step": 24418
+ },
+ {
+ "epoch": 66.53678474114442,
+ "grad_norm": 1.2817738056182861,
+ "learning_rate": 5.3201910186416805e-06,
+ "loss": 0.1004,
+ "step": 24419
+ },
+ {
+ "epoch": 66.53950953678473,
+ "grad_norm": 1.4411121606826782,
+ "learning_rate": 5.3194111423089945e-06,
+ "loss": 0.0238,
+ "step": 24420
+ },
+ {
+ "epoch": 66.54223433242507,
+ "grad_norm": 2.040238380432129,
+ "learning_rate": 5.318631302428584e-06,
+ "loss": 0.041,
+ "step": 24421
+ },
+ {
+ "epoch": 66.5449591280654,
+ "grad_norm": 2.2095344066619873,
+ "learning_rate": 5.317851499006522e-06,
+ "loss": 0.1073,
+ "step": 24422
+ },
+ {
+ "epoch": 66.54768392370572,
+ "grad_norm": 1.4393826723098755,
+ "learning_rate": 5.317071732048877e-06,
+ "loss": 0.0532,
+ "step": 24423
+ },
+ {
+ "epoch": 66.55040871934605,
+ "grad_norm": 1.1244145631790161,
+ "learning_rate": 5.316292001561729e-06,
+ "loss": 0.0117,
+ "step": 24424
+ },
+ {
+ "epoch": 66.55313351498637,
+ "grad_norm": 0.7510209083557129,
+ "learning_rate": 5.315512307551149e-06,
+ "loss": 0.0084,
+ "step": 24425
+ },
+ {
+ "epoch": 66.5558583106267,
+ "grad_norm": 1.870347023010254,
+ "learning_rate": 5.314732650023211e-06,
+ "loss": 0.0726,
+ "step": 24426
+ },
+ {
+ "epoch": 66.55858310626704,
+ "grad_norm": 2.0053834915161133,
+ "learning_rate": 5.3139530289839835e-06,
+ "loss": 0.037,
+ "step": 24427
+ },
+ {
+ "epoch": 66.56130790190736,
+ "grad_norm": 2.1974332332611084,
+ "learning_rate": 5.313173444439535e-06,
+ "loss": 0.0921,
+ "step": 24428
+ },
+ {
+ "epoch": 66.56403269754769,
+ "grad_norm": 1.5224531888961792,
+ "learning_rate": 5.312393896395944e-06,
+ "loss": 0.0614,
+ "step": 24429
+ },
+ {
+ "epoch": 66.566757493188,
+ "grad_norm": 1.7106281518936157,
+ "learning_rate": 5.311614384859281e-06,
+ "loss": 0.0168,
+ "step": 24430
+ },
+ {
+ "epoch": 66.56948228882834,
+ "grad_norm": 1.0532894134521484,
+ "learning_rate": 5.310834909835607e-06,
+ "loss": 0.0095,
+ "step": 24431
+ },
+ {
+ "epoch": 66.57220708446866,
+ "grad_norm": 1.5841269493103027,
+ "learning_rate": 5.310055471331006e-06,
+ "loss": 0.0409,
+ "step": 24432
+ },
+ {
+ "epoch": 66.57493188010899,
+ "grad_norm": 1.5735106468200684,
+ "learning_rate": 5.309276069351541e-06,
+ "loss": 0.0341,
+ "step": 24433
+ },
+ {
+ "epoch": 66.57765667574932,
+ "grad_norm": 1.6542357206344604,
+ "learning_rate": 5.308496703903283e-06,
+ "loss": 0.12,
+ "step": 24434
+ },
+ {
+ "epoch": 66.58038147138964,
+ "grad_norm": 1.385417103767395,
+ "learning_rate": 5.307717374992298e-06,
+ "loss": 0.0326,
+ "step": 24435
+ },
+ {
+ "epoch": 66.58310626702998,
+ "grad_norm": 1.6149276494979858,
+ "learning_rate": 5.306938082624663e-06,
+ "loss": 0.0287,
+ "step": 24436
+ },
+ {
+ "epoch": 66.5858310626703,
+ "grad_norm": 1.5328630208969116,
+ "learning_rate": 5.306158826806443e-06,
+ "loss": 0.0529,
+ "step": 24437
+ },
+ {
+ "epoch": 66.58855585831063,
+ "grad_norm": 2.0813839435577393,
+ "learning_rate": 5.305379607543708e-06,
+ "loss": 0.1164,
+ "step": 24438
+ },
+ {
+ "epoch": 66.59128065395096,
+ "grad_norm": 1.081493616104126,
+ "learning_rate": 5.30460042484252e-06,
+ "loss": 0.017,
+ "step": 24439
+ },
+ {
+ "epoch": 66.59400544959128,
+ "grad_norm": 1.9223116636276245,
+ "learning_rate": 5.303821278708958e-06,
+ "loss": 0.0187,
+ "step": 24440
+ },
+ {
+ "epoch": 66.59673024523161,
+ "grad_norm": 1.6843793392181396,
+ "learning_rate": 5.303042169149079e-06,
+ "loss": 0.0713,
+ "step": 24441
+ },
+ {
+ "epoch": 66.59945504087193,
+ "grad_norm": 1.3852782249450684,
+ "learning_rate": 5.302263096168963e-06,
+ "loss": 0.0128,
+ "step": 24442
+ },
+ {
+ "epoch": 66.60217983651226,
+ "grad_norm": 2.0105583667755127,
+ "learning_rate": 5.301484059774668e-06,
+ "loss": 0.0228,
+ "step": 24443
+ },
+ {
+ "epoch": 66.60490463215258,
+ "grad_norm": 1.6422861814498901,
+ "learning_rate": 5.300705059972262e-06,
+ "loss": 0.0907,
+ "step": 24444
+ },
+ {
+ "epoch": 66.60762942779292,
+ "grad_norm": 3.439143419265747,
+ "learning_rate": 5.299926096767818e-06,
+ "loss": 0.0222,
+ "step": 24445
+ },
+ {
+ "epoch": 66.61035422343325,
+ "grad_norm": 1.4593801498413086,
+ "learning_rate": 5.299147170167397e-06,
+ "loss": 0.0153,
+ "step": 24446
+ },
+ {
+ "epoch": 66.61307901907357,
+ "grad_norm": 1.5968657732009888,
+ "learning_rate": 5.298368280177066e-06,
+ "loss": 0.0201,
+ "step": 24447
+ },
+ {
+ "epoch": 66.6158038147139,
+ "grad_norm": 1.591119647026062,
+ "learning_rate": 5.297589426802889e-06,
+ "loss": 0.0364,
+ "step": 24448
+ },
+ {
+ "epoch": 66.61852861035422,
+ "grad_norm": 2.236046552658081,
+ "learning_rate": 5.296810610050937e-06,
+ "loss": 0.0739,
+ "step": 24449
+ },
+ {
+ "epoch": 66.62125340599455,
+ "grad_norm": 1.051313042640686,
+ "learning_rate": 5.296031829927274e-06,
+ "loss": 0.0129,
+ "step": 24450
+ },
+ {
+ "epoch": 66.62397820163488,
+ "grad_norm": 1.3285554647445679,
+ "learning_rate": 5.2952530864379595e-06,
+ "loss": 0.0233,
+ "step": 24451
+ },
+ {
+ "epoch": 66.6267029972752,
+ "grad_norm": 2.099316358566284,
+ "learning_rate": 5.294474379589065e-06,
+ "loss": 0.0842,
+ "step": 24452
+ },
+ {
+ "epoch": 66.62942779291554,
+ "grad_norm": 1.287039041519165,
+ "learning_rate": 5.293695709386655e-06,
+ "loss": 0.1219,
+ "step": 24453
+ },
+ {
+ "epoch": 66.63215258855585,
+ "grad_norm": 1.1826972961425781,
+ "learning_rate": 5.292917075836789e-06,
+ "loss": 0.048,
+ "step": 24454
+ },
+ {
+ "epoch": 66.63487738419619,
+ "grad_norm": 1.5821980237960815,
+ "learning_rate": 5.29213847894553e-06,
+ "loss": 0.0291,
+ "step": 24455
+ },
+ {
+ "epoch": 66.6376021798365,
+ "grad_norm": 2.766885995864868,
+ "learning_rate": 5.29135991871895e-06,
+ "loss": 0.0752,
+ "step": 24456
+ },
+ {
+ "epoch": 66.64032697547684,
+ "grad_norm": 1.1418769359588623,
+ "learning_rate": 5.290581395163106e-06,
+ "loss": 0.0151,
+ "step": 24457
+ },
+ {
+ "epoch": 66.64305177111717,
+ "grad_norm": 1.1591253280639648,
+ "learning_rate": 5.289802908284063e-06,
+ "loss": 0.0118,
+ "step": 24458
+ },
+ {
+ "epoch": 66.64577656675749,
+ "grad_norm": 1.817427158355713,
+ "learning_rate": 5.289024458087879e-06,
+ "loss": 0.0907,
+ "step": 24459
+ },
+ {
+ "epoch": 66.64850136239782,
+ "grad_norm": 1.5493953227996826,
+ "learning_rate": 5.288246044580623e-06,
+ "loss": 0.073,
+ "step": 24460
+ },
+ {
+ "epoch": 66.65122615803814,
+ "grad_norm": 3.006450653076172,
+ "learning_rate": 5.287467667768357e-06,
+ "loss": 0.0185,
+ "step": 24461
+ },
+ {
+ "epoch": 66.65395095367847,
+ "grad_norm": 1.2168818712234497,
+ "learning_rate": 5.286689327657142e-06,
+ "loss": 0.0119,
+ "step": 24462
+ },
+ {
+ "epoch": 66.65667574931881,
+ "grad_norm": 1.5196647644042969,
+ "learning_rate": 5.285911024253039e-06,
+ "loss": 0.0186,
+ "step": 24463
+ },
+ {
+ "epoch": 66.65940054495913,
+ "grad_norm": 1.8257871866226196,
+ "learning_rate": 5.2851327575621035e-06,
+ "loss": 0.0322,
+ "step": 24464
+ },
+ {
+ "epoch": 66.66212534059946,
+ "grad_norm": 0.9414905309677124,
+ "learning_rate": 5.284354527590407e-06,
+ "loss": 0.0091,
+ "step": 24465
+ },
+ {
+ "epoch": 66.66485013623978,
+ "grad_norm": 1.347357153892517,
+ "learning_rate": 5.283576334344007e-06,
+ "loss": 0.0248,
+ "step": 24466
+ },
+ {
+ "epoch": 66.66757493188011,
+ "grad_norm": 1.689489722251892,
+ "learning_rate": 5.2827981778289625e-06,
+ "loss": 0.008,
+ "step": 24467
+ },
+ {
+ "epoch": 66.67029972752043,
+ "grad_norm": 1.0352435111999512,
+ "learning_rate": 5.2820200580513295e-06,
+ "loss": 0.019,
+ "step": 24468
+ },
+ {
+ "epoch": 66.67302452316076,
+ "grad_norm": 1.6792556047439575,
+ "learning_rate": 5.281241975017176e-06,
+ "loss": 0.0304,
+ "step": 24469
+ },
+ {
+ "epoch": 66.6757493188011,
+ "grad_norm": 2.179490566253662,
+ "learning_rate": 5.280463928732557e-06,
+ "loss": 0.0435,
+ "step": 24470
+ },
+ {
+ "epoch": 66.67847411444141,
+ "grad_norm": 1.9749995470046997,
+ "learning_rate": 5.279685919203537e-06,
+ "loss": 0.0559,
+ "step": 24471
+ },
+ {
+ "epoch": 66.68119891008175,
+ "grad_norm": 1.4275997877120972,
+ "learning_rate": 5.278907946436164e-06,
+ "loss": 0.0124,
+ "step": 24472
+ },
+ {
+ "epoch": 66.68392370572207,
+ "grad_norm": 1.6895062923431396,
+ "learning_rate": 5.278130010436509e-06,
+ "loss": 0.0156,
+ "step": 24473
+ },
+ {
+ "epoch": 66.6866485013624,
+ "grad_norm": 2.196322202682495,
+ "learning_rate": 5.277352111210625e-06,
+ "loss": 0.0877,
+ "step": 24474
+ },
+ {
+ "epoch": 66.68937329700273,
+ "grad_norm": 1.2285116910934448,
+ "learning_rate": 5.276574248764567e-06,
+ "loss": 0.0136,
+ "step": 24475
+ },
+ {
+ "epoch": 66.69209809264305,
+ "grad_norm": 1.2192291021347046,
+ "learning_rate": 5.275796423104396e-06,
+ "loss": 0.0179,
+ "step": 24476
+ },
+ {
+ "epoch": 66.69482288828338,
+ "grad_norm": 2.371520519256592,
+ "learning_rate": 5.275018634236176e-06,
+ "loss": 0.0454,
+ "step": 24477
+ },
+ {
+ "epoch": 66.6975476839237,
+ "grad_norm": 1.6800787448883057,
+ "learning_rate": 5.274240882165958e-06,
+ "loss": 0.0387,
+ "step": 24478
+ },
+ {
+ "epoch": 66.70027247956403,
+ "grad_norm": 1.8080143928527832,
+ "learning_rate": 5.2734631668998e-06,
+ "loss": 0.0587,
+ "step": 24479
+ },
+ {
+ "epoch": 66.70299727520435,
+ "grad_norm": 2.1049368381500244,
+ "learning_rate": 5.272685488443754e-06,
+ "loss": 0.0394,
+ "step": 24480
+ },
+ {
+ "epoch": 66.70572207084469,
+ "grad_norm": 1.0981671810150146,
+ "learning_rate": 5.271907846803885e-06,
+ "loss": 0.0501,
+ "step": 24481
+ },
+ {
+ "epoch": 66.70844686648502,
+ "grad_norm": 1.491863489151001,
+ "learning_rate": 5.271130241986246e-06,
+ "loss": 0.042,
+ "step": 24482
+ },
+ {
+ "epoch": 66.71117166212534,
+ "grad_norm": 1.7397221326828003,
+ "learning_rate": 5.270352673996893e-06,
+ "loss": 0.0723,
+ "step": 24483
+ },
+ {
+ "epoch": 66.71389645776567,
+ "grad_norm": 2.6666338443756104,
+ "learning_rate": 5.269575142841876e-06,
+ "loss": 0.0167,
+ "step": 24484
+ },
+ {
+ "epoch": 66.71662125340599,
+ "grad_norm": 2.2015697956085205,
+ "learning_rate": 5.268797648527262e-06,
+ "loss": 0.0578,
+ "step": 24485
+ },
+ {
+ "epoch": 66.71934604904632,
+ "grad_norm": 2.1878039836883545,
+ "learning_rate": 5.268020191059098e-06,
+ "loss": 0.0283,
+ "step": 24486
+ },
+ {
+ "epoch": 66.72207084468666,
+ "grad_norm": 1.6755436658859253,
+ "learning_rate": 5.26724277044344e-06,
+ "loss": 0.0468,
+ "step": 24487
+ },
+ {
+ "epoch": 66.72479564032697,
+ "grad_norm": 1.6322999000549316,
+ "learning_rate": 5.266465386686339e-06,
+ "loss": 0.0805,
+ "step": 24488
+ },
+ {
+ "epoch": 66.7275204359673,
+ "grad_norm": 2.3152222633361816,
+ "learning_rate": 5.2656880397938585e-06,
+ "loss": 0.0142,
+ "step": 24489
+ },
+ {
+ "epoch": 66.73024523160763,
+ "grad_norm": 1.3531476259231567,
+ "learning_rate": 5.264910729772046e-06,
+ "loss": 0.0536,
+ "step": 24490
+ },
+ {
+ "epoch": 66.73297002724796,
+ "grad_norm": 1.5718878507614136,
+ "learning_rate": 5.264133456626956e-06,
+ "loss": 0.0297,
+ "step": 24491
+ },
+ {
+ "epoch": 66.73569482288828,
+ "grad_norm": 1.4098750352859497,
+ "learning_rate": 5.2633562203646395e-06,
+ "loss": 0.0272,
+ "step": 24492
+ },
+ {
+ "epoch": 66.73841961852861,
+ "grad_norm": 1.7951350212097168,
+ "learning_rate": 5.262579020991152e-06,
+ "loss": 0.0361,
+ "step": 24493
+ },
+ {
+ "epoch": 66.74114441416894,
+ "grad_norm": 11.156868934631348,
+ "learning_rate": 5.26180185851255e-06,
+ "loss": 0.0366,
+ "step": 24494
+ },
+ {
+ "epoch": 66.74386920980926,
+ "grad_norm": 0.9865410923957825,
+ "learning_rate": 5.261024732934883e-06,
+ "loss": 0.0095,
+ "step": 24495
+ },
+ {
+ "epoch": 66.7465940054496,
+ "grad_norm": 1.977963924407959,
+ "learning_rate": 5.260247644264199e-06,
+ "loss": 0.0399,
+ "step": 24496
+ },
+ {
+ "epoch": 66.74931880108991,
+ "grad_norm": 1.959768295288086,
+ "learning_rate": 5.259470592506558e-06,
+ "loss": 0.0336,
+ "step": 24497
+ },
+ {
+ "epoch": 66.75204359673025,
+ "grad_norm": 0.8314064741134644,
+ "learning_rate": 5.258693577668007e-06,
+ "loss": 0.0087,
+ "step": 24498
+ },
+ {
+ "epoch": 66.75476839237058,
+ "grad_norm": 1.5071289539337158,
+ "learning_rate": 5.257916599754598e-06,
+ "loss": 0.0218,
+ "step": 24499
+ },
+ {
+ "epoch": 66.7574931880109,
+ "grad_norm": 1.2006996870040894,
+ "learning_rate": 5.257139658772378e-06,
+ "loss": 0.0253,
+ "step": 24500
+ },
+ {
+ "epoch": 66.76021798365123,
+ "grad_norm": 1.5376811027526855,
+ "learning_rate": 5.256362754727405e-06,
+ "loss": 0.0333,
+ "step": 24501
+ },
+ {
+ "epoch": 66.76294277929155,
+ "grad_norm": 1.941856026649475,
+ "learning_rate": 5.255585887625727e-06,
+ "loss": 0.0597,
+ "step": 24502
+ },
+ {
+ "epoch": 66.76566757493188,
+ "grad_norm": 2.0968377590179443,
+ "learning_rate": 5.254809057473393e-06,
+ "loss": 0.0391,
+ "step": 24503
+ },
+ {
+ "epoch": 66.7683923705722,
+ "grad_norm": 2.260561466217041,
+ "learning_rate": 5.25403226427645e-06,
+ "loss": 0.0231,
+ "step": 24504
+ },
+ {
+ "epoch": 66.77111716621253,
+ "grad_norm": 1.631389856338501,
+ "learning_rate": 5.253255508040954e-06,
+ "loss": 0.0615,
+ "step": 24505
+ },
+ {
+ "epoch": 66.77384196185287,
+ "grad_norm": 1.588674783706665,
+ "learning_rate": 5.252478788772952e-06,
+ "loss": 0.0433,
+ "step": 24506
+ },
+ {
+ "epoch": 66.77656675749319,
+ "grad_norm": 1.4386781454086304,
+ "learning_rate": 5.251702106478491e-06,
+ "loss": 0.0153,
+ "step": 24507
+ },
+ {
+ "epoch": 66.77929155313352,
+ "grad_norm": 1.6068652868270874,
+ "learning_rate": 5.2509254611636196e-06,
+ "loss": 0.0194,
+ "step": 24508
+ },
+ {
+ "epoch": 66.78201634877384,
+ "grad_norm": 1.3141535520553589,
+ "learning_rate": 5.25014885283439e-06,
+ "loss": 0.0221,
+ "step": 24509
+ },
+ {
+ "epoch": 66.78474114441417,
+ "grad_norm": 3.135648727416992,
+ "learning_rate": 5.249372281496845e-06,
+ "loss": 0.0215,
+ "step": 24510
+ },
+ {
+ "epoch": 66.7874659400545,
+ "grad_norm": 1.5247037410736084,
+ "learning_rate": 5.248595747157039e-06,
+ "loss": 0.1096,
+ "step": 24511
+ },
+ {
+ "epoch": 66.79019073569482,
+ "grad_norm": 1.5268630981445312,
+ "learning_rate": 5.247819249821018e-06,
+ "loss": 0.0941,
+ "step": 24512
+ },
+ {
+ "epoch": 66.79291553133515,
+ "grad_norm": 0.8024718761444092,
+ "learning_rate": 5.247042789494823e-06,
+ "loss": 0.0098,
+ "step": 24513
+ },
+ {
+ "epoch": 66.79564032697547,
+ "grad_norm": 1.7857979536056519,
+ "learning_rate": 5.2462663661845094e-06,
+ "loss": 0.0978,
+ "step": 24514
+ },
+ {
+ "epoch": 66.7983651226158,
+ "grad_norm": 2.1465697288513184,
+ "learning_rate": 5.24548997989612e-06,
+ "loss": 0.042,
+ "step": 24515
+ },
+ {
+ "epoch": 66.80108991825612,
+ "grad_norm": 1.2310502529144287,
+ "learning_rate": 5.244713630635698e-06,
+ "loss": 0.0212,
+ "step": 24516
+ },
+ {
+ "epoch": 66.80381471389646,
+ "grad_norm": 1.1824995279312134,
+ "learning_rate": 5.243937318409298e-06,
+ "loss": 0.018,
+ "step": 24517
+ },
+ {
+ "epoch": 66.80653950953679,
+ "grad_norm": 2.0055227279663086,
+ "learning_rate": 5.24316104322296e-06,
+ "loss": 0.0464,
+ "step": 24518
+ },
+ {
+ "epoch": 66.80926430517711,
+ "grad_norm": 1.3840879201889038,
+ "learning_rate": 5.242384805082731e-06,
+ "loss": 0.016,
+ "step": 24519
+ },
+ {
+ "epoch": 66.81198910081744,
+ "grad_norm": 1.522397756576538,
+ "learning_rate": 5.241608603994652e-06,
+ "loss": 0.0575,
+ "step": 24520
+ },
+ {
+ "epoch": 66.81471389645776,
+ "grad_norm": 1.447427749633789,
+ "learning_rate": 5.240832439964776e-06,
+ "loss": 0.0136,
+ "step": 24521
+ },
+ {
+ "epoch": 66.8174386920981,
+ "grad_norm": 1.8007737398147583,
+ "learning_rate": 5.240056312999145e-06,
+ "loss": 0.0245,
+ "step": 24522
+ },
+ {
+ "epoch": 66.82016348773843,
+ "grad_norm": 1.4952369928359985,
+ "learning_rate": 5.2392802231037996e-06,
+ "loss": 0.0246,
+ "step": 24523
+ },
+ {
+ "epoch": 66.82288828337875,
+ "grad_norm": 1.5158107280731201,
+ "learning_rate": 5.238504170284785e-06,
+ "loss": 0.0257,
+ "step": 24524
+ },
+ {
+ "epoch": 66.82561307901908,
+ "grad_norm": 1.1797521114349365,
+ "learning_rate": 5.23772815454815e-06,
+ "loss": 0.0197,
+ "step": 24525
+ },
+ {
+ "epoch": 66.8283378746594,
+ "grad_norm": 1.6768994331359863,
+ "learning_rate": 5.236952175899934e-06,
+ "loss": 0.0599,
+ "step": 24526
+ },
+ {
+ "epoch": 66.83106267029973,
+ "grad_norm": 1.8135778903961182,
+ "learning_rate": 5.236176234346179e-06,
+ "loss": 0.0243,
+ "step": 24527
+ },
+ {
+ "epoch": 66.83378746594005,
+ "grad_norm": 0.7393589019775391,
+ "learning_rate": 5.2354003298929325e-06,
+ "loss": 0.0109,
+ "step": 24528
+ },
+ {
+ "epoch": 66.83651226158038,
+ "grad_norm": 1.291560411453247,
+ "learning_rate": 5.234624462546232e-06,
+ "loss": 0.0118,
+ "step": 24529
+ },
+ {
+ "epoch": 66.83923705722071,
+ "grad_norm": 1.3496183156967163,
+ "learning_rate": 5.2338486323121266e-06,
+ "loss": 0.019,
+ "step": 24530
+ },
+ {
+ "epoch": 66.84196185286103,
+ "grad_norm": 1.4517790079116821,
+ "learning_rate": 5.233072839196653e-06,
+ "loss": 0.0409,
+ "step": 24531
+ },
+ {
+ "epoch": 66.84468664850137,
+ "grad_norm": 2.0715830326080322,
+ "learning_rate": 5.2322970832058565e-06,
+ "loss": 0.0533,
+ "step": 24532
+ },
+ {
+ "epoch": 66.84741144414168,
+ "grad_norm": 1.1927087306976318,
+ "learning_rate": 5.231521364345772e-06,
+ "loss": 0.014,
+ "step": 24533
+ },
+ {
+ "epoch": 66.85013623978202,
+ "grad_norm": 1.7037330865859985,
+ "learning_rate": 5.23074568262245e-06,
+ "loss": 0.0236,
+ "step": 24534
+ },
+ {
+ "epoch": 66.85286103542235,
+ "grad_norm": 1.5906367301940918,
+ "learning_rate": 5.229970038041927e-06,
+ "loss": 0.0175,
+ "step": 24535
+ },
+ {
+ "epoch": 66.85558583106267,
+ "grad_norm": 0.980337381362915,
+ "learning_rate": 5.229194430610243e-06,
+ "loss": 0.0131,
+ "step": 24536
+ },
+ {
+ "epoch": 66.858310626703,
+ "grad_norm": 1.7459028959274292,
+ "learning_rate": 5.228418860333437e-06,
+ "loss": 0.0487,
+ "step": 24537
+ },
+ {
+ "epoch": 66.86103542234332,
+ "grad_norm": 1.5929096937179565,
+ "learning_rate": 5.227643327217555e-06,
+ "loss": 0.0136,
+ "step": 24538
+ },
+ {
+ "epoch": 66.86376021798365,
+ "grad_norm": 1.2488529682159424,
+ "learning_rate": 5.226867831268632e-06,
+ "loss": 0.0146,
+ "step": 24539
+ },
+ {
+ "epoch": 66.86648501362397,
+ "grad_norm": 1.4487850666046143,
+ "learning_rate": 5.226092372492706e-06,
+ "loss": 0.1392,
+ "step": 24540
+ },
+ {
+ "epoch": 66.8692098092643,
+ "grad_norm": 1.164577841758728,
+ "learning_rate": 5.2253169508958225e-06,
+ "loss": 0.0135,
+ "step": 24541
+ },
+ {
+ "epoch": 66.87193460490464,
+ "grad_norm": 1.9939806461334229,
+ "learning_rate": 5.224541566484016e-06,
+ "loss": 0.0858,
+ "step": 24542
+ },
+ {
+ "epoch": 66.87465940054496,
+ "grad_norm": 2.614824056625366,
+ "learning_rate": 5.223766219263327e-06,
+ "loss": 0.0381,
+ "step": 24543
+ },
+ {
+ "epoch": 66.87738419618529,
+ "grad_norm": 1.322619080543518,
+ "learning_rate": 5.222990909239789e-06,
+ "loss": 0.0163,
+ "step": 24544
+ },
+ {
+ "epoch": 66.88010899182561,
+ "grad_norm": 2.0707859992980957,
+ "learning_rate": 5.222215636419443e-06,
+ "loss": 0.103,
+ "step": 24545
+ },
+ {
+ "epoch": 66.88283378746594,
+ "grad_norm": 1.7243835926055908,
+ "learning_rate": 5.221440400808333e-06,
+ "loss": 0.0524,
+ "step": 24546
+ },
+ {
+ "epoch": 66.88555858310627,
+ "grad_norm": 2.01124906539917,
+ "learning_rate": 5.220665202412492e-06,
+ "loss": 0.0165,
+ "step": 24547
+ },
+ {
+ "epoch": 66.88828337874659,
+ "grad_norm": 1.545057773590088,
+ "learning_rate": 5.219890041237955e-06,
+ "loss": 0.1114,
+ "step": 24548
+ },
+ {
+ "epoch": 66.89100817438693,
+ "grad_norm": 1.2977083921432495,
+ "learning_rate": 5.219114917290757e-06,
+ "loss": 0.0113,
+ "step": 24549
+ },
+ {
+ "epoch": 66.89373297002724,
+ "grad_norm": 1.7025457620620728,
+ "learning_rate": 5.2183398305769424e-06,
+ "loss": 0.0161,
+ "step": 24550
+ },
+ {
+ "epoch": 66.89645776566758,
+ "grad_norm": 1.6280888319015503,
+ "learning_rate": 5.217564781102543e-06,
+ "loss": 0.0906,
+ "step": 24551
+ },
+ {
+ "epoch": 66.8991825613079,
+ "grad_norm": 1.478726863861084,
+ "learning_rate": 5.216789768873596e-06,
+ "loss": 0.1375,
+ "step": 24552
+ },
+ {
+ "epoch": 66.90190735694823,
+ "grad_norm": 3.687340497970581,
+ "learning_rate": 5.2160147938961315e-06,
+ "loss": 0.1701,
+ "step": 24553
+ },
+ {
+ "epoch": 66.90463215258856,
+ "grad_norm": 1.5641919374465942,
+ "learning_rate": 5.215239856176194e-06,
+ "loss": 0.0421,
+ "step": 24554
+ },
+ {
+ "epoch": 66.90735694822888,
+ "grad_norm": 2.3343796730041504,
+ "learning_rate": 5.214464955719814e-06,
+ "loss": 0.0274,
+ "step": 24555
+ },
+ {
+ "epoch": 66.91008174386921,
+ "grad_norm": 1.569677710533142,
+ "learning_rate": 5.213690092533028e-06,
+ "loss": 0.0368,
+ "step": 24556
+ },
+ {
+ "epoch": 66.91280653950953,
+ "grad_norm": 1.08354651927948,
+ "learning_rate": 5.212915266621865e-06,
+ "loss": 0.012,
+ "step": 24557
+ },
+ {
+ "epoch": 66.91553133514986,
+ "grad_norm": 1.265305519104004,
+ "learning_rate": 5.212140477992367e-06,
+ "loss": 0.0151,
+ "step": 24558
+ },
+ {
+ "epoch": 66.9182561307902,
+ "grad_norm": 1.8441252708435059,
+ "learning_rate": 5.2113657266505655e-06,
+ "loss": 0.02,
+ "step": 24559
+ },
+ {
+ "epoch": 66.92098092643052,
+ "grad_norm": 1.7029216289520264,
+ "learning_rate": 5.210591012602495e-06,
+ "loss": 0.0416,
+ "step": 24560
+ },
+ {
+ "epoch": 66.92370572207085,
+ "grad_norm": 1.5439115762710571,
+ "learning_rate": 5.209816335854181e-06,
+ "loss": 0.0209,
+ "step": 24561
+ },
+ {
+ "epoch": 66.92643051771117,
+ "grad_norm": 0.8839330077171326,
+ "learning_rate": 5.209041696411665e-06,
+ "loss": 0.0112,
+ "step": 24562
+ },
+ {
+ "epoch": 66.9291553133515,
+ "grad_norm": 1.0492041110992432,
+ "learning_rate": 5.208267094280981e-06,
+ "loss": 0.0146,
+ "step": 24563
+ },
+ {
+ "epoch": 66.93188010899182,
+ "grad_norm": 1.1371618509292603,
+ "learning_rate": 5.20749252946816e-06,
+ "loss": 0.0189,
+ "step": 24564
+ },
+ {
+ "epoch": 66.93460490463215,
+ "grad_norm": 2.340172052383423,
+ "learning_rate": 5.206718001979228e-06,
+ "loss": 0.0319,
+ "step": 24565
+ },
+ {
+ "epoch": 66.93732970027249,
+ "grad_norm": 1.3870553970336914,
+ "learning_rate": 5.2059435118202264e-06,
+ "loss": 0.0242,
+ "step": 24566
+ },
+ {
+ "epoch": 66.9400544959128,
+ "grad_norm": 1.146804690361023,
+ "learning_rate": 5.205169058997181e-06,
+ "loss": 0.0132,
+ "step": 24567
+ },
+ {
+ "epoch": 66.94277929155314,
+ "grad_norm": 2.406467914581299,
+ "learning_rate": 5.204394643516127e-06,
+ "loss": 0.1529,
+ "step": 24568
+ },
+ {
+ "epoch": 66.94550408719346,
+ "grad_norm": 1.1805188655853271,
+ "learning_rate": 5.203620265383087e-06,
+ "loss": 0.0119,
+ "step": 24569
+ },
+ {
+ "epoch": 66.94822888283379,
+ "grad_norm": 1.4211182594299316,
+ "learning_rate": 5.202845924604103e-06,
+ "loss": 0.086,
+ "step": 24570
+ },
+ {
+ "epoch": 66.95095367847412,
+ "grad_norm": 1.665793776512146,
+ "learning_rate": 5.2020716211852e-06,
+ "loss": 0.0227,
+ "step": 24571
+ },
+ {
+ "epoch": 66.95367847411444,
+ "grad_norm": 2.358765125274658,
+ "learning_rate": 5.201297355132411e-06,
+ "loss": 0.0477,
+ "step": 24572
+ },
+ {
+ "epoch": 66.95640326975477,
+ "grad_norm": 1.4585614204406738,
+ "learning_rate": 5.200523126451757e-06,
+ "loss": 0.0635,
+ "step": 24573
+ },
+ {
+ "epoch": 66.95912806539509,
+ "grad_norm": 1.2692887783050537,
+ "learning_rate": 5.19974893514928e-06,
+ "loss": 0.0142,
+ "step": 24574
+ },
+ {
+ "epoch": 66.96185286103542,
+ "grad_norm": 1.3357048034667969,
+ "learning_rate": 5.198974781231003e-06,
+ "loss": 0.0288,
+ "step": 24575
+ },
+ {
+ "epoch": 66.96457765667574,
+ "grad_norm": 2.3957507610321045,
+ "learning_rate": 5.1982006647029555e-06,
+ "loss": 0.1198,
+ "step": 24576
+ },
+ {
+ "epoch": 66.96730245231608,
+ "grad_norm": 1.4890782833099365,
+ "learning_rate": 5.197426585571163e-06,
+ "loss": 0.019,
+ "step": 24577
+ },
+ {
+ "epoch": 66.97002724795641,
+ "grad_norm": 1.7925047874450684,
+ "learning_rate": 5.196652543841663e-06,
+ "loss": 0.0694,
+ "step": 24578
+ },
+ {
+ "epoch": 66.97275204359673,
+ "grad_norm": 1.608115553855896,
+ "learning_rate": 5.195878539520474e-06,
+ "loss": 0.0295,
+ "step": 24579
+ },
+ {
+ "epoch": 66.97547683923706,
+ "grad_norm": 1.3691719770431519,
+ "learning_rate": 5.195104572613631e-06,
+ "loss": 0.0329,
+ "step": 24580
+ },
+ {
+ "epoch": 66.97820163487738,
+ "grad_norm": 1.7589664459228516,
+ "learning_rate": 5.194330643127155e-06,
+ "loss": 0.0159,
+ "step": 24581
+ },
+ {
+ "epoch": 66.98092643051771,
+ "grad_norm": 1.670162320137024,
+ "learning_rate": 5.193556751067082e-06,
+ "loss": 0.0227,
+ "step": 24582
+ },
+ {
+ "epoch": 66.98365122615803,
+ "grad_norm": 1.3774687051773071,
+ "learning_rate": 5.192782896439434e-06,
+ "loss": 0.0168,
+ "step": 24583
+ },
+ {
+ "epoch": 66.98637602179836,
+ "grad_norm": 1.076026439666748,
+ "learning_rate": 5.192009079250239e-06,
+ "loss": 0.0085,
+ "step": 24584
+ },
+ {
+ "epoch": 66.9891008174387,
+ "grad_norm": 1.3821905851364136,
+ "learning_rate": 5.191235299505517e-06,
+ "loss": 0.0163,
+ "step": 24585
+ },
+ {
+ "epoch": 66.99182561307902,
+ "grad_norm": 1.8514517545700073,
+ "learning_rate": 5.190461557211306e-06,
+ "loss": 0.0208,
+ "step": 24586
+ },
+ {
+ "epoch": 66.99455040871935,
+ "grad_norm": 1.2445783615112305,
+ "learning_rate": 5.189687852373624e-06,
+ "loss": 0.0156,
+ "step": 24587
+ },
+ {
+ "epoch": 66.99727520435967,
+ "grad_norm": 1.4942835569381714,
+ "learning_rate": 5.188914184998497e-06,
+ "loss": 0.0234,
+ "step": 24588
+ },
+ {
+ "epoch": 67.0,
+ "grad_norm": 1.1019811630249023,
+ "learning_rate": 5.18814055509195e-06,
+ "loss": 0.0118,
+ "step": 24589
+ },
+ {
+ "epoch": 67.00272479564033,
+ "grad_norm": 1.571677327156067,
+ "learning_rate": 5.187366962660012e-06,
+ "loss": 0.0757,
+ "step": 24590
+ },
+ {
+ "epoch": 67.00544959128065,
+ "grad_norm": 2.1163384914398193,
+ "learning_rate": 5.186593407708706e-06,
+ "loss": 0.0136,
+ "step": 24591
+ },
+ {
+ "epoch": 67.00817438692098,
+ "grad_norm": 1.3206701278686523,
+ "learning_rate": 5.185819890244055e-06,
+ "loss": 0.0404,
+ "step": 24592
+ },
+ {
+ "epoch": 67.0108991825613,
+ "grad_norm": 1.6830108165740967,
+ "learning_rate": 5.185046410272079e-06,
+ "loss": 0.0452,
+ "step": 24593
+ },
+ {
+ "epoch": 67.01362397820164,
+ "grad_norm": 1.5064642429351807,
+ "learning_rate": 5.184272967798811e-06,
+ "loss": 0.018,
+ "step": 24594
+ },
+ {
+ "epoch": 67.01634877384195,
+ "grad_norm": 1.7431896924972534,
+ "learning_rate": 5.183499562830272e-06,
+ "loss": 0.0167,
+ "step": 24595
+ },
+ {
+ "epoch": 67.01907356948229,
+ "grad_norm": 1.2953094244003296,
+ "learning_rate": 5.182726195372478e-06,
+ "loss": 0.1379,
+ "step": 24596
+ },
+ {
+ "epoch": 67.02179836512262,
+ "grad_norm": 2.029541015625,
+ "learning_rate": 5.181952865431461e-06,
+ "loss": 0.0974,
+ "step": 24597
+ },
+ {
+ "epoch": 67.02452316076294,
+ "grad_norm": 1.1315516233444214,
+ "learning_rate": 5.1811795730132365e-06,
+ "loss": 0.0099,
+ "step": 24598
+ },
+ {
+ "epoch": 67.02724795640327,
+ "grad_norm": 1.7497642040252686,
+ "learning_rate": 5.180406318123834e-06,
+ "loss": 0.0602,
+ "step": 24599
+ },
+ {
+ "epoch": 67.02997275204359,
+ "grad_norm": 1.135597586631775,
+ "learning_rate": 5.1796331007692725e-06,
+ "loss": 0.0418,
+ "step": 24600
+ },
+ {
+ "epoch": 67.03269754768392,
+ "grad_norm": 1.3690004348754883,
+ "learning_rate": 5.1788599209555726e-06,
+ "loss": 0.0209,
+ "step": 24601
+ },
+ {
+ "epoch": 67.03542234332426,
+ "grad_norm": 1.3251372575759888,
+ "learning_rate": 5.1780867786887535e-06,
+ "loss": 0.0136,
+ "step": 24602
+ },
+ {
+ "epoch": 67.03814713896458,
+ "grad_norm": 1.7084330320358276,
+ "learning_rate": 5.177313673974842e-06,
+ "loss": 0.1947,
+ "step": 24603
+ },
+ {
+ "epoch": 67.04087193460491,
+ "grad_norm": 1.275894284248352,
+ "learning_rate": 5.176540606819857e-06,
+ "loss": 0.0227,
+ "step": 24604
+ },
+ {
+ "epoch": 67.04359673024523,
+ "grad_norm": 2.3214261531829834,
+ "learning_rate": 5.175767577229815e-06,
+ "loss": 0.0991,
+ "step": 24605
+ },
+ {
+ "epoch": 67.04632152588556,
+ "grad_norm": 1.1277343034744263,
+ "learning_rate": 5.174994585210744e-06,
+ "loss": 0.0103,
+ "step": 24606
+ },
+ {
+ "epoch": 67.04904632152588,
+ "grad_norm": 4.700556755065918,
+ "learning_rate": 5.174221630768659e-06,
+ "loss": 0.0929,
+ "step": 24607
+ },
+ {
+ "epoch": 67.05177111716621,
+ "grad_norm": 1.7022757530212402,
+ "learning_rate": 5.1734487139095815e-06,
+ "loss": 0.0188,
+ "step": 24608
+ },
+ {
+ "epoch": 67.05449591280654,
+ "grad_norm": 3.9603166580200195,
+ "learning_rate": 5.172675834639524e-06,
+ "loss": 0.1263,
+ "step": 24609
+ },
+ {
+ "epoch": 67.05722070844686,
+ "grad_norm": 0.9461039304733276,
+ "learning_rate": 5.171902992964518e-06,
+ "loss": 0.011,
+ "step": 24610
+ },
+ {
+ "epoch": 67.0599455040872,
+ "grad_norm": 3.0418291091918945,
+ "learning_rate": 5.1711301888905765e-06,
+ "loss": 0.0682,
+ "step": 24611
+ },
+ {
+ "epoch": 67.06267029972751,
+ "grad_norm": 2.2545149326324463,
+ "learning_rate": 5.170357422423715e-06,
+ "loss": 0.1093,
+ "step": 24612
+ },
+ {
+ "epoch": 67.06539509536785,
+ "grad_norm": 2.661691665649414,
+ "learning_rate": 5.169584693569953e-06,
+ "loss": 0.1758,
+ "step": 24613
+ },
+ {
+ "epoch": 67.06811989100818,
+ "grad_norm": 2.0120484828948975,
+ "learning_rate": 5.168812002335309e-06,
+ "loss": 0.0321,
+ "step": 24614
+ },
+ {
+ "epoch": 67.0708446866485,
+ "grad_norm": 1.392167568206787,
+ "learning_rate": 5.168039348725806e-06,
+ "loss": 0.0177,
+ "step": 24615
+ },
+ {
+ "epoch": 67.07356948228883,
+ "grad_norm": 0.9779276251792908,
+ "learning_rate": 5.167266732747455e-06,
+ "loss": 0.0132,
+ "step": 24616
+ },
+ {
+ "epoch": 67.07629427792915,
+ "grad_norm": 1.5923601388931274,
+ "learning_rate": 5.166494154406279e-06,
+ "loss": 0.0304,
+ "step": 24617
+ },
+ {
+ "epoch": 67.07901907356948,
+ "grad_norm": 1.8311585187911987,
+ "learning_rate": 5.165721613708283e-06,
+ "loss": 0.019,
+ "step": 24618
+ },
+ {
+ "epoch": 67.0817438692098,
+ "grad_norm": 1.6985770463943481,
+ "learning_rate": 5.164949110659498e-06,
+ "loss": 0.0517,
+ "step": 24619
+ },
+ {
+ "epoch": 67.08446866485014,
+ "grad_norm": 1.6108850240707397,
+ "learning_rate": 5.164176645265932e-06,
+ "loss": 0.0668,
+ "step": 24620
+ },
+ {
+ "epoch": 67.08719346049047,
+ "grad_norm": 1.5901429653167725,
+ "learning_rate": 5.163404217533604e-06,
+ "loss": 0.0225,
+ "step": 24621
+ },
+ {
+ "epoch": 67.08991825613079,
+ "grad_norm": 2.0848350524902344,
+ "learning_rate": 5.162631827468525e-06,
+ "loss": 0.0377,
+ "step": 24622
+ },
+ {
+ "epoch": 67.09264305177112,
+ "grad_norm": 1.7764184474945068,
+ "learning_rate": 5.161859475076716e-06,
+ "loss": 0.052,
+ "step": 24623
+ },
+ {
+ "epoch": 67.09536784741144,
+ "grad_norm": 1.2846482992172241,
+ "learning_rate": 5.16108716036419e-06,
+ "loss": 0.0323,
+ "step": 24624
+ },
+ {
+ "epoch": 67.09809264305177,
+ "grad_norm": 1.4541034698486328,
+ "learning_rate": 5.160314883336958e-06,
+ "loss": 0.0777,
+ "step": 24625
+ },
+ {
+ "epoch": 67.1008174386921,
+ "grad_norm": 1.938956379890442,
+ "learning_rate": 5.159542644001041e-06,
+ "loss": 0.1029,
+ "step": 24626
+ },
+ {
+ "epoch": 67.10354223433242,
+ "grad_norm": 1.5415668487548828,
+ "learning_rate": 5.158770442362451e-06,
+ "loss": 0.0175,
+ "step": 24627
+ },
+ {
+ "epoch": 67.10626702997276,
+ "grad_norm": 1.1556744575500488,
+ "learning_rate": 5.1579982784272005e-06,
+ "loss": 0.0154,
+ "step": 24628
+ },
+ {
+ "epoch": 67.10899182561307,
+ "grad_norm": 2.184248208999634,
+ "learning_rate": 5.1572261522013e-06,
+ "loss": 0.1039,
+ "step": 24629
+ },
+ {
+ "epoch": 67.11171662125341,
+ "grad_norm": 1.5619059801101685,
+ "learning_rate": 5.15645406369077e-06,
+ "loss": 0.113,
+ "step": 24630
+ },
+ {
+ "epoch": 67.11444141689373,
+ "grad_norm": 1.4811911582946777,
+ "learning_rate": 5.155682012901616e-06,
+ "loss": 0.0229,
+ "step": 24631
+ },
+ {
+ "epoch": 67.11716621253406,
+ "grad_norm": 1.9367084503173828,
+ "learning_rate": 5.1549099998398585e-06,
+ "loss": 0.032,
+ "step": 24632
+ },
+ {
+ "epoch": 67.11989100817439,
+ "grad_norm": 2.770411968231201,
+ "learning_rate": 5.154138024511506e-06,
+ "loss": 0.0379,
+ "step": 24633
+ },
+ {
+ "epoch": 67.12261580381471,
+ "grad_norm": 1.201864242553711,
+ "learning_rate": 5.153366086922567e-06,
+ "loss": 0.0128,
+ "step": 24634
+ },
+ {
+ "epoch": 67.12534059945504,
+ "grad_norm": 1.595775842666626,
+ "learning_rate": 5.152594187079061e-06,
+ "loss": 0.015,
+ "step": 24635
+ },
+ {
+ "epoch": 67.12806539509536,
+ "grad_norm": 1.2723294496536255,
+ "learning_rate": 5.151822324986995e-06,
+ "loss": 0.0125,
+ "step": 24636
+ },
+ {
+ "epoch": 67.1307901907357,
+ "grad_norm": 1.5292924642562866,
+ "learning_rate": 5.151050500652379e-06,
+ "loss": 0.0145,
+ "step": 24637
+ },
+ {
+ "epoch": 67.13351498637603,
+ "grad_norm": 1.7455888986587524,
+ "learning_rate": 5.150278714081224e-06,
+ "loss": 0.0208,
+ "step": 24638
+ },
+ {
+ "epoch": 67.13623978201635,
+ "grad_norm": 1.3146380186080933,
+ "learning_rate": 5.149506965279545e-06,
+ "loss": 0.0266,
+ "step": 24639
+ },
+ {
+ "epoch": 67.13896457765668,
+ "grad_norm": 1.8340904712677002,
+ "learning_rate": 5.148735254253351e-06,
+ "loss": 0.067,
+ "step": 24640
+ },
+ {
+ "epoch": 67.141689373297,
+ "grad_norm": 0.9489344358444214,
+ "learning_rate": 5.14796358100865e-06,
+ "loss": 0.0108,
+ "step": 24641
+ },
+ {
+ "epoch": 67.14441416893733,
+ "grad_norm": 2.0448949337005615,
+ "learning_rate": 5.1471919455514484e-06,
+ "loss": 0.0199,
+ "step": 24642
+ },
+ {
+ "epoch": 67.14713896457765,
+ "grad_norm": 1.6527825593948364,
+ "learning_rate": 5.146420347887764e-06,
+ "loss": 0.0257,
+ "step": 24643
+ },
+ {
+ "epoch": 67.14986376021798,
+ "grad_norm": 0.9665658473968506,
+ "learning_rate": 5.145648788023601e-06,
+ "loss": 0.1117,
+ "step": 24644
+ },
+ {
+ "epoch": 67.15258855585832,
+ "grad_norm": 1.058640480041504,
+ "learning_rate": 5.144877265964969e-06,
+ "loss": 0.0154,
+ "step": 24645
+ },
+ {
+ "epoch": 67.15531335149863,
+ "grad_norm": 1.6250134706497192,
+ "learning_rate": 5.144105781717874e-06,
+ "loss": 0.0299,
+ "step": 24646
+ },
+ {
+ "epoch": 67.15803814713897,
+ "grad_norm": 2.1328978538513184,
+ "learning_rate": 5.14333433528833e-06,
+ "loss": 0.0888,
+ "step": 24647
+ },
+ {
+ "epoch": 67.16076294277929,
+ "grad_norm": 2.021373748779297,
+ "learning_rate": 5.142562926682338e-06,
+ "loss": 0.0521,
+ "step": 24648
+ },
+ {
+ "epoch": 67.16348773841962,
+ "grad_norm": 2.182166814804077,
+ "learning_rate": 5.141791555905913e-06,
+ "loss": 0.0322,
+ "step": 24649
+ },
+ {
+ "epoch": 67.16621253405995,
+ "grad_norm": 1.5687121152877808,
+ "learning_rate": 5.141020222965056e-06,
+ "loss": 0.0802,
+ "step": 24650
+ },
+ {
+ "epoch": 67.16893732970027,
+ "grad_norm": 1.3966925144195557,
+ "learning_rate": 5.140248927865781e-06,
+ "loss": 0.0214,
+ "step": 24651
+ },
+ {
+ "epoch": 67.1716621253406,
+ "grad_norm": 1.0811034440994263,
+ "learning_rate": 5.139477670614091e-06,
+ "loss": 0.0196,
+ "step": 24652
+ },
+ {
+ "epoch": 67.17438692098092,
+ "grad_norm": 2.0260732173919678,
+ "learning_rate": 5.138706451215991e-06,
+ "loss": 0.0909,
+ "step": 24653
+ },
+ {
+ "epoch": 67.17711171662125,
+ "grad_norm": 1.4440816640853882,
+ "learning_rate": 5.137935269677485e-06,
+ "loss": 0.0261,
+ "step": 24654
+ },
+ {
+ "epoch": 67.17983651226157,
+ "grad_norm": 2.0465123653411865,
+ "learning_rate": 5.137164126004588e-06,
+ "loss": 0.0198,
+ "step": 24655
+ },
+ {
+ "epoch": 67.1825613079019,
+ "grad_norm": 1.7071162462234497,
+ "learning_rate": 5.136393020203298e-06,
+ "loss": 0.0752,
+ "step": 24656
+ },
+ {
+ "epoch": 67.18528610354224,
+ "grad_norm": 2.343754768371582,
+ "learning_rate": 5.1356219522796255e-06,
+ "loss": 0.016,
+ "step": 24657
+ },
+ {
+ "epoch": 67.18801089918256,
+ "grad_norm": 2.309356212615967,
+ "learning_rate": 5.134850922239567e-06,
+ "loss": 0.035,
+ "step": 24658
+ },
+ {
+ "epoch": 67.19073569482289,
+ "grad_norm": 1.0933619737625122,
+ "learning_rate": 5.134079930089138e-06,
+ "loss": 0.0155,
+ "step": 24659
+ },
+ {
+ "epoch": 67.19346049046321,
+ "grad_norm": 1.2807679176330566,
+ "learning_rate": 5.1333089758343364e-06,
+ "loss": 0.0179,
+ "step": 24660
+ },
+ {
+ "epoch": 67.19618528610354,
+ "grad_norm": 1.9336751699447632,
+ "learning_rate": 5.1325380594811694e-06,
+ "loss": 0.1286,
+ "step": 24661
+ },
+ {
+ "epoch": 67.19891008174388,
+ "grad_norm": 1.4811937808990479,
+ "learning_rate": 5.131767181035634e-06,
+ "loss": 0.0929,
+ "step": 24662
+ },
+ {
+ "epoch": 67.2016348773842,
+ "grad_norm": 1.37766432762146,
+ "learning_rate": 5.130996340503744e-06,
+ "loss": 0.0125,
+ "step": 24663
+ },
+ {
+ "epoch": 67.20435967302453,
+ "grad_norm": 1.036981463432312,
+ "learning_rate": 5.130225537891498e-06,
+ "loss": 0.0142,
+ "step": 24664
+ },
+ {
+ "epoch": 67.20708446866485,
+ "grad_norm": 1.5239735841751099,
+ "learning_rate": 5.129454773204894e-06,
+ "loss": 0.0702,
+ "step": 24665
+ },
+ {
+ "epoch": 67.20980926430518,
+ "grad_norm": 2.3296666145324707,
+ "learning_rate": 5.1286840464499435e-06,
+ "loss": 0.056,
+ "step": 24666
+ },
+ {
+ "epoch": 67.2125340599455,
+ "grad_norm": 1.0499602556228638,
+ "learning_rate": 5.127913357632641e-06,
+ "loss": 0.0093,
+ "step": 24667
+ },
+ {
+ "epoch": 67.21525885558583,
+ "grad_norm": 2.245041608810425,
+ "learning_rate": 5.127142706758996e-06,
+ "loss": 0.0287,
+ "step": 24668
+ },
+ {
+ "epoch": 67.21798365122616,
+ "grad_norm": 0.9884676933288574,
+ "learning_rate": 5.1263720938350085e-06,
+ "loss": 0.0121,
+ "step": 24669
+ },
+ {
+ "epoch": 67.22070844686648,
+ "grad_norm": 1.4455187320709229,
+ "learning_rate": 5.125601518866672e-06,
+ "loss": 0.0208,
+ "step": 24670
+ },
+ {
+ "epoch": 67.22343324250681,
+ "grad_norm": 1.4272998571395874,
+ "learning_rate": 5.12483098186e-06,
+ "loss": 0.0691,
+ "step": 24671
+ },
+ {
+ "epoch": 67.22615803814713,
+ "grad_norm": 1.8576542139053345,
+ "learning_rate": 5.124060482820986e-06,
+ "loss": 0.1071,
+ "step": 24672
+ },
+ {
+ "epoch": 67.22888283378747,
+ "grad_norm": 1.558747410774231,
+ "learning_rate": 5.123290021755634e-06,
+ "loss": 0.066,
+ "step": 24673
+ },
+ {
+ "epoch": 67.2316076294278,
+ "grad_norm": 1.53018057346344,
+ "learning_rate": 5.122519598669935e-06,
+ "loss": 0.035,
+ "step": 24674
+ },
+ {
+ "epoch": 67.23433242506812,
+ "grad_norm": 1.699807047843933,
+ "learning_rate": 5.121749213569903e-06,
+ "loss": 0.1314,
+ "step": 24675
+ },
+ {
+ "epoch": 67.23705722070845,
+ "grad_norm": 0.8232681155204773,
+ "learning_rate": 5.1209788664615304e-06,
+ "loss": 0.0097,
+ "step": 24676
+ },
+ {
+ "epoch": 67.23978201634877,
+ "grad_norm": 1.3377656936645508,
+ "learning_rate": 5.1202085573508175e-06,
+ "loss": 0.0138,
+ "step": 24677
+ },
+ {
+ "epoch": 67.2425068119891,
+ "grad_norm": 1.0411794185638428,
+ "learning_rate": 5.119438286243759e-06,
+ "loss": 0.0221,
+ "step": 24678
+ },
+ {
+ "epoch": 67.24523160762942,
+ "grad_norm": 1.667531967163086,
+ "learning_rate": 5.118668053146363e-06,
+ "loss": 0.0298,
+ "step": 24679
+ },
+ {
+ "epoch": 67.24795640326975,
+ "grad_norm": 1.6104098558425903,
+ "learning_rate": 5.117897858064622e-06,
+ "loss": 0.0282,
+ "step": 24680
+ },
+ {
+ "epoch": 67.25068119891009,
+ "grad_norm": 2.4279890060424805,
+ "learning_rate": 5.117127701004535e-06,
+ "loss": 0.119,
+ "step": 24681
+ },
+ {
+ "epoch": 67.2534059945504,
+ "grad_norm": 1.8025521039962769,
+ "learning_rate": 5.116357581972098e-06,
+ "loss": 0.0495,
+ "step": 24682
+ },
+ {
+ "epoch": 67.25613079019074,
+ "grad_norm": 1.5980381965637207,
+ "learning_rate": 5.115587500973309e-06,
+ "loss": 0.0489,
+ "step": 24683
+ },
+ {
+ "epoch": 67.25885558583106,
+ "grad_norm": 2.214289903640747,
+ "learning_rate": 5.114817458014172e-06,
+ "loss": 0.0223,
+ "step": 24684
+ },
+ {
+ "epoch": 67.26158038147139,
+ "grad_norm": 1.8148527145385742,
+ "learning_rate": 5.11404745310068e-06,
+ "loss": 0.0162,
+ "step": 24685
+ },
+ {
+ "epoch": 67.26430517711172,
+ "grad_norm": 2.330483913421631,
+ "learning_rate": 5.113277486238829e-06,
+ "loss": 0.0574,
+ "step": 24686
+ },
+ {
+ "epoch": 67.26702997275204,
+ "grad_norm": 1.7498233318328857,
+ "learning_rate": 5.112507557434612e-06,
+ "loss": 0.0846,
+ "step": 24687
+ },
+ {
+ "epoch": 67.26975476839237,
+ "grad_norm": 2.2397594451904297,
+ "learning_rate": 5.111737666694032e-06,
+ "loss": 0.048,
+ "step": 24688
+ },
+ {
+ "epoch": 67.2724795640327,
+ "grad_norm": 2.0872209072113037,
+ "learning_rate": 5.110967814023083e-06,
+ "loss": 0.0263,
+ "step": 24689
+ },
+ {
+ "epoch": 67.27520435967303,
+ "grad_norm": 1.3161611557006836,
+ "learning_rate": 5.110197999427753e-06,
+ "loss": 0.0178,
+ "step": 24690
+ },
+ {
+ "epoch": 67.27792915531334,
+ "grad_norm": 1.5333105325698853,
+ "learning_rate": 5.10942822291405e-06,
+ "loss": 0.0206,
+ "step": 24691
+ },
+ {
+ "epoch": 67.28065395095368,
+ "grad_norm": 2.4255125522613525,
+ "learning_rate": 5.108658484487961e-06,
+ "loss": 0.0712,
+ "step": 24692
+ },
+ {
+ "epoch": 67.28337874659401,
+ "grad_norm": 1.8330594301223755,
+ "learning_rate": 5.107888784155484e-06,
+ "loss": 0.0308,
+ "step": 24693
+ },
+ {
+ "epoch": 67.28610354223433,
+ "grad_norm": 1.6994534730911255,
+ "learning_rate": 5.107119121922606e-06,
+ "loss": 0.0898,
+ "step": 24694
+ },
+ {
+ "epoch": 67.28882833787466,
+ "grad_norm": 1.3509007692337036,
+ "learning_rate": 5.106349497795331e-06,
+ "loss": 0.0167,
+ "step": 24695
+ },
+ {
+ "epoch": 67.29155313351498,
+ "grad_norm": 1.6772229671478271,
+ "learning_rate": 5.105579911779649e-06,
+ "loss": 0.0243,
+ "step": 24696
+ },
+ {
+ "epoch": 67.29427792915531,
+ "grad_norm": 2.356813669204712,
+ "learning_rate": 5.104810363881554e-06,
+ "loss": 0.0869,
+ "step": 24697
+ },
+ {
+ "epoch": 67.29700272479565,
+ "grad_norm": 1.1081843376159668,
+ "learning_rate": 5.104040854107033e-06,
+ "loss": 0.0187,
+ "step": 24698
+ },
+ {
+ "epoch": 67.29972752043597,
+ "grad_norm": 1.9273127317428589,
+ "learning_rate": 5.103271382462085e-06,
+ "loss": 0.0559,
+ "step": 24699
+ },
+ {
+ "epoch": 67.3024523160763,
+ "grad_norm": 1.3556338548660278,
+ "learning_rate": 5.102501948952705e-06,
+ "loss": 0.0197,
+ "step": 24700
+ },
+ {
+ "epoch": 67.30517711171662,
+ "grad_norm": 1.121038556098938,
+ "learning_rate": 5.101732553584883e-06,
+ "loss": 0.0107,
+ "step": 24701
+ },
+ {
+ "epoch": 67.30790190735695,
+ "grad_norm": 3.927399158477783,
+ "learning_rate": 5.10096319636461e-06,
+ "loss": 0.0496,
+ "step": 24702
+ },
+ {
+ "epoch": 67.31062670299727,
+ "grad_norm": 1.4364409446716309,
+ "learning_rate": 5.100193877297874e-06,
+ "loss": 0.0278,
+ "step": 24703
+ },
+ {
+ "epoch": 67.3133514986376,
+ "grad_norm": 1.2939507961273193,
+ "learning_rate": 5.099424596390674e-06,
+ "loss": 0.0214,
+ "step": 24704
+ },
+ {
+ "epoch": 67.31607629427793,
+ "grad_norm": 2.0226104259490967,
+ "learning_rate": 5.098655353648998e-06,
+ "loss": 0.0402,
+ "step": 24705
+ },
+ {
+ "epoch": 67.31880108991825,
+ "grad_norm": 2.036558151245117,
+ "learning_rate": 5.097886149078836e-06,
+ "loss": 0.1262,
+ "step": 24706
+ },
+ {
+ "epoch": 67.32152588555859,
+ "grad_norm": 1.5485798120498657,
+ "learning_rate": 5.097116982686175e-06,
+ "loss": 0.0288,
+ "step": 24707
+ },
+ {
+ "epoch": 67.3242506811989,
+ "grad_norm": 1.4443540573120117,
+ "learning_rate": 5.0963478544770125e-06,
+ "loss": 0.034,
+ "step": 24708
+ },
+ {
+ "epoch": 67.32697547683924,
+ "grad_norm": 1.4456266164779663,
+ "learning_rate": 5.095578764457336e-06,
+ "loss": 0.0257,
+ "step": 24709
+ },
+ {
+ "epoch": 67.32970027247957,
+ "grad_norm": 1.435346245765686,
+ "learning_rate": 5.0948097126331355e-06,
+ "loss": 0.0276,
+ "step": 24710
+ },
+ {
+ "epoch": 67.33242506811989,
+ "grad_norm": 1.2628679275512695,
+ "learning_rate": 5.094040699010393e-06,
+ "loss": 0.0157,
+ "step": 24711
+ },
+ {
+ "epoch": 67.33514986376022,
+ "grad_norm": 1.6401995420455933,
+ "learning_rate": 5.093271723595108e-06,
+ "loss": 0.0582,
+ "step": 24712
+ },
+ {
+ "epoch": 67.33787465940054,
+ "grad_norm": 1.4409626722335815,
+ "learning_rate": 5.092502786393265e-06,
+ "loss": 0.0678,
+ "step": 24713
+ },
+ {
+ "epoch": 67.34059945504087,
+ "grad_norm": 3.2281577587127686,
+ "learning_rate": 5.091733887410849e-06,
+ "loss": 0.0188,
+ "step": 24714
+ },
+ {
+ "epoch": 67.34332425068119,
+ "grad_norm": 1.1711153984069824,
+ "learning_rate": 5.090965026653856e-06,
+ "loss": 0.024,
+ "step": 24715
+ },
+ {
+ "epoch": 67.34604904632153,
+ "grad_norm": 2.2223899364471436,
+ "learning_rate": 5.090196204128268e-06,
+ "loss": 0.0475,
+ "step": 24716
+ },
+ {
+ "epoch": 67.34877384196186,
+ "grad_norm": 2.419390916824341,
+ "learning_rate": 5.08942741984007e-06,
+ "loss": 0.0544,
+ "step": 24717
+ },
+ {
+ "epoch": 67.35149863760218,
+ "grad_norm": 1.0406692028045654,
+ "learning_rate": 5.088658673795257e-06,
+ "loss": 0.0153,
+ "step": 24718
+ },
+ {
+ "epoch": 67.35422343324251,
+ "grad_norm": 1.2692662477493286,
+ "learning_rate": 5.08788996599981e-06,
+ "loss": 0.0116,
+ "step": 24719
+ },
+ {
+ "epoch": 67.35694822888283,
+ "grad_norm": 1.6887539625167847,
+ "learning_rate": 5.0871212964597205e-06,
+ "loss": 0.0376,
+ "step": 24720
+ },
+ {
+ "epoch": 67.35967302452316,
+ "grad_norm": 1.3792093992233276,
+ "learning_rate": 5.086352665180971e-06,
+ "loss": 0.0294,
+ "step": 24721
+ },
+ {
+ "epoch": 67.3623978201635,
+ "grad_norm": 1.7966461181640625,
+ "learning_rate": 5.085584072169551e-06,
+ "loss": 0.0268,
+ "step": 24722
+ },
+ {
+ "epoch": 67.36512261580381,
+ "grad_norm": 1.0668398141860962,
+ "learning_rate": 5.08481551743144e-06,
+ "loss": 0.0167,
+ "step": 24723
+ },
+ {
+ "epoch": 67.36784741144415,
+ "grad_norm": 1.0407301187515259,
+ "learning_rate": 5.0840470009726295e-06,
+ "loss": 0.0093,
+ "step": 24724
+ },
+ {
+ "epoch": 67.37057220708446,
+ "grad_norm": 1.5017039775848389,
+ "learning_rate": 5.083278522799104e-06,
+ "loss": 0.0205,
+ "step": 24725
+ },
+ {
+ "epoch": 67.3732970027248,
+ "grad_norm": 1.3962677717208862,
+ "learning_rate": 5.082510082916847e-06,
+ "loss": 0.0626,
+ "step": 24726
+ },
+ {
+ "epoch": 67.37602179836512,
+ "grad_norm": 1.5052844285964966,
+ "learning_rate": 5.08174168133184e-06,
+ "loss": 0.0271,
+ "step": 24727
+ },
+ {
+ "epoch": 67.37874659400545,
+ "grad_norm": 1.769282579421997,
+ "learning_rate": 5.080973318050073e-06,
+ "loss": 0.0195,
+ "step": 24728
+ },
+ {
+ "epoch": 67.38147138964578,
+ "grad_norm": 1.113714575767517,
+ "learning_rate": 5.080204993077529e-06,
+ "loss": 0.0125,
+ "step": 24729
+ },
+ {
+ "epoch": 67.3841961852861,
+ "grad_norm": 2.3569774627685547,
+ "learning_rate": 5.0794367064201885e-06,
+ "loss": 0.0684,
+ "step": 24730
+ },
+ {
+ "epoch": 67.38692098092643,
+ "grad_norm": 1.736480474472046,
+ "learning_rate": 5.078668458084035e-06,
+ "loss": 0.1078,
+ "step": 24731
+ },
+ {
+ "epoch": 67.38964577656675,
+ "grad_norm": 1.3785474300384521,
+ "learning_rate": 5.077900248075055e-06,
+ "loss": 0.0691,
+ "step": 24732
+ },
+ {
+ "epoch": 67.39237057220708,
+ "grad_norm": 2.8475565910339355,
+ "learning_rate": 5.07713207639923e-06,
+ "loss": 0.0653,
+ "step": 24733
+ },
+ {
+ "epoch": 67.39509536784742,
+ "grad_norm": 3.8975203037261963,
+ "learning_rate": 5.0763639430625385e-06,
+ "loss": 0.0271,
+ "step": 24734
+ },
+ {
+ "epoch": 67.39782016348774,
+ "grad_norm": 1.6085609197616577,
+ "learning_rate": 5.0755958480709665e-06,
+ "loss": 0.0238,
+ "step": 24735
+ },
+ {
+ "epoch": 67.40054495912807,
+ "grad_norm": 1.1805577278137207,
+ "learning_rate": 5.074827791430499e-06,
+ "loss": 0.0159,
+ "step": 24736
+ },
+ {
+ "epoch": 67.40326975476839,
+ "grad_norm": 2.0049660205841064,
+ "learning_rate": 5.0740597731471134e-06,
+ "loss": 0.0411,
+ "step": 24737
+ },
+ {
+ "epoch": 67.40599455040872,
+ "grad_norm": 1.478649377822876,
+ "learning_rate": 5.073291793226791e-06,
+ "loss": 0.0149,
+ "step": 24738
+ },
+ {
+ "epoch": 67.40871934604904,
+ "grad_norm": 1.4042991399765015,
+ "learning_rate": 5.072523851675512e-06,
+ "loss": 0.0156,
+ "step": 24739
+ },
+ {
+ "epoch": 67.41144414168937,
+ "grad_norm": 1.425377607345581,
+ "learning_rate": 5.071755948499262e-06,
+ "loss": 0.0155,
+ "step": 24740
+ },
+ {
+ "epoch": 67.4141689373297,
+ "grad_norm": 2.670213222503662,
+ "learning_rate": 5.070988083704016e-06,
+ "loss": 0.0635,
+ "step": 24741
+ },
+ {
+ "epoch": 67.41689373297002,
+ "grad_norm": 3.5536344051361084,
+ "learning_rate": 5.070220257295758e-06,
+ "loss": 0.0259,
+ "step": 24742
+ },
+ {
+ "epoch": 67.41961852861036,
+ "grad_norm": 2.63844895362854,
+ "learning_rate": 5.0694524692804625e-06,
+ "loss": 0.0297,
+ "step": 24743
+ },
+ {
+ "epoch": 67.42234332425068,
+ "grad_norm": 2.1904335021972656,
+ "learning_rate": 5.068684719664116e-06,
+ "loss": 0.0884,
+ "step": 24744
+ },
+ {
+ "epoch": 67.42506811989101,
+ "grad_norm": 1.4846519231796265,
+ "learning_rate": 5.067917008452693e-06,
+ "loss": 0.023,
+ "step": 24745
+ },
+ {
+ "epoch": 67.42779291553134,
+ "grad_norm": 1.659479022026062,
+ "learning_rate": 5.067149335652175e-06,
+ "loss": 0.0421,
+ "step": 24746
+ },
+ {
+ "epoch": 67.43051771117166,
+ "grad_norm": 1.8537373542785645,
+ "learning_rate": 5.066381701268535e-06,
+ "loss": 0.0606,
+ "step": 24747
+ },
+ {
+ "epoch": 67.433242506812,
+ "grad_norm": 0.7418370246887207,
+ "learning_rate": 5.065614105307759e-06,
+ "loss": 0.0104,
+ "step": 24748
+ },
+ {
+ "epoch": 67.43596730245231,
+ "grad_norm": 1.343578577041626,
+ "learning_rate": 5.064846547775821e-06,
+ "loss": 0.0403,
+ "step": 24749
+ },
+ {
+ "epoch": 67.43869209809264,
+ "grad_norm": 1.723156452178955,
+ "learning_rate": 5.0640790286787e-06,
+ "loss": 0.1141,
+ "step": 24750
+ },
+ {
+ "epoch": 67.44141689373296,
+ "grad_norm": 2.0908761024475098,
+ "learning_rate": 5.0633115480223695e-06,
+ "loss": 0.0224,
+ "step": 24751
+ },
+ {
+ "epoch": 67.4441416893733,
+ "grad_norm": 1.4315781593322754,
+ "learning_rate": 5.062544105812807e-06,
+ "loss": 0.0314,
+ "step": 24752
+ },
+ {
+ "epoch": 67.44686648501363,
+ "grad_norm": 1.1112451553344727,
+ "learning_rate": 5.061776702055998e-06,
+ "loss": 0.0134,
+ "step": 24753
+ },
+ {
+ "epoch": 67.44959128065395,
+ "grad_norm": 1.9277801513671875,
+ "learning_rate": 5.0610093367579135e-06,
+ "loss": 0.0566,
+ "step": 24754
+ },
+ {
+ "epoch": 67.45231607629428,
+ "grad_norm": 1.4550938606262207,
+ "learning_rate": 5.060242009924525e-06,
+ "loss": 0.0198,
+ "step": 24755
+ },
+ {
+ "epoch": 67.4550408719346,
+ "grad_norm": 1.1457222700119019,
+ "learning_rate": 5.059474721561817e-06,
+ "loss": 0.0141,
+ "step": 24756
+ },
+ {
+ "epoch": 67.45776566757493,
+ "grad_norm": 0.8416242599487305,
+ "learning_rate": 5.058707471675761e-06,
+ "loss": 0.0097,
+ "step": 24757
+ },
+ {
+ "epoch": 67.46049046321527,
+ "grad_norm": 1.3116097450256348,
+ "learning_rate": 5.0579402602723315e-06,
+ "loss": 0.0224,
+ "step": 24758
+ },
+ {
+ "epoch": 67.46321525885558,
+ "grad_norm": 1.3832882642745972,
+ "learning_rate": 5.057173087357501e-06,
+ "loss": 0.0255,
+ "step": 24759
+ },
+ {
+ "epoch": 67.46594005449592,
+ "grad_norm": 2.0850391387939453,
+ "learning_rate": 5.05640595293725e-06,
+ "loss": 0.0519,
+ "step": 24760
+ },
+ {
+ "epoch": 67.46866485013624,
+ "grad_norm": 9.48072338104248,
+ "learning_rate": 5.055638857017551e-06,
+ "loss": 0.0281,
+ "step": 24761
+ },
+ {
+ "epoch": 67.47138964577657,
+ "grad_norm": 1.4678922891616821,
+ "learning_rate": 5.054871799604377e-06,
+ "loss": 0.0972,
+ "step": 24762
+ },
+ {
+ "epoch": 67.47411444141689,
+ "grad_norm": 1.2614924907684326,
+ "learning_rate": 5.054104780703699e-06,
+ "loss": 0.0249,
+ "step": 24763
+ },
+ {
+ "epoch": 67.47683923705722,
+ "grad_norm": 1.528222680091858,
+ "learning_rate": 5.053337800321498e-06,
+ "loss": 0.0397,
+ "step": 24764
+ },
+ {
+ "epoch": 67.47956403269755,
+ "grad_norm": 0.8961524963378906,
+ "learning_rate": 5.052570858463743e-06,
+ "loss": 0.0086,
+ "step": 24765
+ },
+ {
+ "epoch": 67.48228882833787,
+ "grad_norm": 1.3153631687164307,
+ "learning_rate": 5.051803955136406e-06,
+ "loss": 0.0217,
+ "step": 24766
+ },
+ {
+ "epoch": 67.4850136239782,
+ "grad_norm": 0.9343996644020081,
+ "learning_rate": 5.051037090345457e-06,
+ "loss": 0.0116,
+ "step": 24767
+ },
+ {
+ "epoch": 67.48773841961852,
+ "grad_norm": 1.2249442338943481,
+ "learning_rate": 5.050270264096873e-06,
+ "loss": 0.0166,
+ "step": 24768
+ },
+ {
+ "epoch": 67.49046321525886,
+ "grad_norm": 1.0382336378097534,
+ "learning_rate": 5.049503476396627e-06,
+ "loss": 0.0103,
+ "step": 24769
+ },
+ {
+ "epoch": 67.49318801089919,
+ "grad_norm": 1.8016328811645508,
+ "learning_rate": 5.048736727250689e-06,
+ "loss": 0.0511,
+ "step": 24770
+ },
+ {
+ "epoch": 67.49591280653951,
+ "grad_norm": 1.18902587890625,
+ "learning_rate": 5.047970016665031e-06,
+ "loss": 0.0173,
+ "step": 24771
+ },
+ {
+ "epoch": 67.49863760217984,
+ "grad_norm": 1.8010835647583008,
+ "learning_rate": 5.0472033446456195e-06,
+ "loss": 0.0591,
+ "step": 24772
+ },
+ {
+ "epoch": 67.50136239782016,
+ "grad_norm": 1.2618408203125,
+ "learning_rate": 5.046436711198431e-06,
+ "loss": 0.0152,
+ "step": 24773
+ },
+ {
+ "epoch": 67.50408719346049,
+ "grad_norm": 1.2070578336715698,
+ "learning_rate": 5.045670116329435e-06,
+ "loss": 0.0285,
+ "step": 24774
+ },
+ {
+ "epoch": 67.50681198910081,
+ "grad_norm": 1.7487696409225464,
+ "learning_rate": 5.0449035600446005e-06,
+ "loss": 0.0231,
+ "step": 24775
+ },
+ {
+ "epoch": 67.50953678474114,
+ "grad_norm": 7.346922874450684,
+ "learning_rate": 5.044137042349895e-06,
+ "loss": 0.0397,
+ "step": 24776
+ },
+ {
+ "epoch": 67.51226158038148,
+ "grad_norm": 2.9370830059051514,
+ "learning_rate": 5.043370563251293e-06,
+ "loss": 0.0927,
+ "step": 24777
+ },
+ {
+ "epoch": 67.5149863760218,
+ "grad_norm": 1.9930368661880493,
+ "learning_rate": 5.0426041227547616e-06,
+ "loss": 0.1402,
+ "step": 24778
+ },
+ {
+ "epoch": 67.51771117166213,
+ "grad_norm": 1.6820391416549683,
+ "learning_rate": 5.041837720866266e-06,
+ "loss": 0.0251,
+ "step": 24779
+ },
+ {
+ "epoch": 67.52043596730245,
+ "grad_norm": 1.8124146461486816,
+ "learning_rate": 5.041071357591782e-06,
+ "loss": 0.0197,
+ "step": 24780
+ },
+ {
+ "epoch": 67.52316076294278,
+ "grad_norm": 1.9864850044250488,
+ "learning_rate": 5.040305032937276e-06,
+ "loss": 0.1147,
+ "step": 24781
+ },
+ {
+ "epoch": 67.52588555858311,
+ "grad_norm": 1.567430853843689,
+ "learning_rate": 5.039538746908714e-06,
+ "loss": 0.0166,
+ "step": 24782
+ },
+ {
+ "epoch": 67.52861035422343,
+ "grad_norm": 1.6824814081192017,
+ "learning_rate": 5.038772499512061e-06,
+ "loss": 0.1955,
+ "step": 24783
+ },
+ {
+ "epoch": 67.53133514986376,
+ "grad_norm": 1.8028833866119385,
+ "learning_rate": 5.038006290753292e-06,
+ "loss": 0.0551,
+ "step": 24784
+ },
+ {
+ "epoch": 67.53405994550408,
+ "grad_norm": 1.863398551940918,
+ "learning_rate": 5.037240120638367e-06,
+ "loss": 0.0658,
+ "step": 24785
+ },
+ {
+ "epoch": 67.53678474114442,
+ "grad_norm": 1.4567835330963135,
+ "learning_rate": 5.0364739891732584e-06,
+ "loss": 0.0372,
+ "step": 24786
+ },
+ {
+ "epoch": 67.53950953678473,
+ "grad_norm": 1.417376160621643,
+ "learning_rate": 5.0357078963639325e-06,
+ "loss": 0.036,
+ "step": 24787
+ },
+ {
+ "epoch": 67.54223433242507,
+ "grad_norm": 1.9009464979171753,
+ "learning_rate": 5.034941842216349e-06,
+ "loss": 0.0458,
+ "step": 24788
+ },
+ {
+ "epoch": 67.5449591280654,
+ "grad_norm": 1.883802890777588,
+ "learning_rate": 5.034175826736483e-06,
+ "loss": 0.0473,
+ "step": 24789
+ },
+ {
+ "epoch": 67.54768392370572,
+ "grad_norm": 1.803724765777588,
+ "learning_rate": 5.033409849930295e-06,
+ "loss": 0.0267,
+ "step": 24790
+ },
+ {
+ "epoch": 67.55040871934605,
+ "grad_norm": 2.5073790550231934,
+ "learning_rate": 5.032643911803752e-06,
+ "loss": 0.0708,
+ "step": 24791
+ },
+ {
+ "epoch": 67.55313351498637,
+ "grad_norm": 1.1876837015151978,
+ "learning_rate": 5.031878012362813e-06,
+ "loss": 0.0312,
+ "step": 24792
+ },
+ {
+ "epoch": 67.5558583106267,
+ "grad_norm": 1.4394910335540771,
+ "learning_rate": 5.031112151613453e-06,
+ "loss": 0.02,
+ "step": 24793
+ },
+ {
+ "epoch": 67.55858310626704,
+ "grad_norm": 1.630826473236084,
+ "learning_rate": 5.030346329561632e-06,
+ "loss": 0.0297,
+ "step": 24794
+ },
+ {
+ "epoch": 67.56130790190736,
+ "grad_norm": 1.7558305263519287,
+ "learning_rate": 5.029580546213313e-06,
+ "loss": 0.0369,
+ "step": 24795
+ },
+ {
+ "epoch": 67.56403269754769,
+ "grad_norm": 1.788783073425293,
+ "learning_rate": 5.0288148015744576e-06,
+ "loss": 0.0288,
+ "step": 24796
+ },
+ {
+ "epoch": 67.566757493188,
+ "grad_norm": 1.4043900966644287,
+ "learning_rate": 5.028049095651036e-06,
+ "loss": 0.0315,
+ "step": 24797
+ },
+ {
+ "epoch": 67.56948228882834,
+ "grad_norm": 1.406780481338501,
+ "learning_rate": 5.027283428449008e-06,
+ "loss": 0.0158,
+ "step": 24798
+ },
+ {
+ "epoch": 67.57220708446866,
+ "grad_norm": 1.4435386657714844,
+ "learning_rate": 5.0265177999743335e-06,
+ "loss": 0.0063,
+ "step": 24799
+ },
+ {
+ "epoch": 67.57493188010899,
+ "grad_norm": 1.3609340190887451,
+ "learning_rate": 5.025752210232982e-06,
+ "loss": 0.1474,
+ "step": 24800
+ },
+ {
+ "epoch": 67.57765667574932,
+ "grad_norm": 0.812436580657959,
+ "learning_rate": 5.0249866592309115e-06,
+ "loss": 0.0092,
+ "step": 24801
+ },
+ {
+ "epoch": 67.58038147138964,
+ "grad_norm": 1.0790495872497559,
+ "learning_rate": 5.024221146974084e-06,
+ "loss": 0.0076,
+ "step": 24802
+ },
+ {
+ "epoch": 67.58310626702998,
+ "grad_norm": 1.5146105289459229,
+ "learning_rate": 5.02345567346846e-06,
+ "loss": 0.0307,
+ "step": 24803
+ },
+ {
+ "epoch": 67.5858310626703,
+ "grad_norm": 1.0473041534423828,
+ "learning_rate": 5.022690238720002e-06,
+ "loss": 0.0113,
+ "step": 24804
+ },
+ {
+ "epoch": 67.58855585831063,
+ "grad_norm": 1.1663401126861572,
+ "learning_rate": 5.021924842734678e-06,
+ "loss": 0.0195,
+ "step": 24805
+ },
+ {
+ "epoch": 67.59128065395096,
+ "grad_norm": 1.3731380701065063,
+ "learning_rate": 5.021159485518441e-06,
+ "loss": 0.013,
+ "step": 24806
+ },
+ {
+ "epoch": 67.59400544959128,
+ "grad_norm": 1.749855875968933,
+ "learning_rate": 5.020394167077254e-06,
+ "loss": 0.0857,
+ "step": 24807
+ },
+ {
+ "epoch": 67.59673024523161,
+ "grad_norm": 1.9804373979568481,
+ "learning_rate": 5.019628887417073e-06,
+ "loss": 0.0159,
+ "step": 24808
+ },
+ {
+ "epoch": 67.59945504087193,
+ "grad_norm": 1.9518810510635376,
+ "learning_rate": 5.0188636465438654e-06,
+ "loss": 0.0149,
+ "step": 24809
+ },
+ {
+ "epoch": 67.60217983651226,
+ "grad_norm": 0.9590237736701965,
+ "learning_rate": 5.018098444463588e-06,
+ "loss": 0.0151,
+ "step": 24810
+ },
+ {
+ "epoch": 67.60490463215258,
+ "grad_norm": 1.2624377012252808,
+ "learning_rate": 5.0173332811822e-06,
+ "loss": 0.0127,
+ "step": 24811
+ },
+ {
+ "epoch": 67.60762942779292,
+ "grad_norm": 1.5488392114639282,
+ "learning_rate": 5.016568156705657e-06,
+ "loss": 0.143,
+ "step": 24812
+ },
+ {
+ "epoch": 67.61035422343325,
+ "grad_norm": 1.9735403060913086,
+ "learning_rate": 5.015803071039923e-06,
+ "loss": 0.0301,
+ "step": 24813
+ },
+ {
+ "epoch": 67.61307901907357,
+ "grad_norm": 1.3927407264709473,
+ "learning_rate": 5.015038024190954e-06,
+ "loss": 0.0186,
+ "step": 24814
+ },
+ {
+ "epoch": 67.6158038147139,
+ "grad_norm": 1.1426396369934082,
+ "learning_rate": 5.014273016164711e-06,
+ "loss": 0.0416,
+ "step": 24815
+ },
+ {
+ "epoch": 67.61852861035422,
+ "grad_norm": 1.866880178451538,
+ "learning_rate": 5.013508046967142e-06,
+ "loss": 0.0893,
+ "step": 24816
+ },
+ {
+ "epoch": 67.62125340599455,
+ "grad_norm": 1.790256142616272,
+ "learning_rate": 5.012743116604217e-06,
+ "loss": 0.0172,
+ "step": 24817
+ },
+ {
+ "epoch": 67.62397820163488,
+ "grad_norm": 1.2179765701293945,
+ "learning_rate": 5.011978225081889e-06,
+ "loss": 0.019,
+ "step": 24818
+ },
+ {
+ "epoch": 67.6267029972752,
+ "grad_norm": 1.1609183549880981,
+ "learning_rate": 5.011213372406114e-06,
+ "loss": 0.0176,
+ "step": 24819
+ },
+ {
+ "epoch": 67.62942779291554,
+ "grad_norm": 1.5623537302017212,
+ "learning_rate": 5.010448558582841e-06,
+ "loss": 0.0154,
+ "step": 24820
+ },
+ {
+ "epoch": 67.63215258855585,
+ "grad_norm": 1.2702966928482056,
+ "learning_rate": 5.00968378361804e-06,
+ "loss": 0.0328,
+ "step": 24821
+ },
+ {
+ "epoch": 67.63487738419619,
+ "grad_norm": 1.86949622631073,
+ "learning_rate": 5.008919047517663e-06,
+ "loss": 0.0159,
+ "step": 24822
+ },
+ {
+ "epoch": 67.6376021798365,
+ "grad_norm": 1.5526357889175415,
+ "learning_rate": 5.008154350287663e-06,
+ "loss": 0.0196,
+ "step": 24823
+ },
+ {
+ "epoch": 67.64032697547684,
+ "grad_norm": 1.1194738149642944,
+ "learning_rate": 5.007389691933992e-06,
+ "loss": 0.0108,
+ "step": 24824
+ },
+ {
+ "epoch": 67.64305177111717,
+ "grad_norm": 1.0673004388809204,
+ "learning_rate": 5.006625072462615e-06,
+ "loss": 0.0123,
+ "step": 24825
+ },
+ {
+ "epoch": 67.64577656675749,
+ "grad_norm": 1.612930417060852,
+ "learning_rate": 5.005860491879479e-06,
+ "loss": 0.0194,
+ "step": 24826
+ },
+ {
+ "epoch": 67.64850136239782,
+ "grad_norm": 1.4844390153884888,
+ "learning_rate": 5.005095950190541e-06,
+ "loss": 0.0336,
+ "step": 24827
+ },
+ {
+ "epoch": 67.65122615803814,
+ "grad_norm": 3.249699592590332,
+ "learning_rate": 5.004331447401754e-06,
+ "loss": 0.026,
+ "step": 24828
+ },
+ {
+ "epoch": 67.65395095367847,
+ "grad_norm": 1.439937949180603,
+ "learning_rate": 5.003566983519073e-06,
+ "loss": 0.0528,
+ "step": 24829
+ },
+ {
+ "epoch": 67.65667574931881,
+ "grad_norm": 1.2326706647872925,
+ "learning_rate": 5.002802558548454e-06,
+ "loss": 0.0212,
+ "step": 24830
+ },
+ {
+ "epoch": 67.65940054495913,
+ "grad_norm": 1.9400315284729004,
+ "learning_rate": 5.002038172495847e-06,
+ "loss": 0.0173,
+ "step": 24831
+ },
+ {
+ "epoch": 67.66212534059946,
+ "grad_norm": 0.9235400557518005,
+ "learning_rate": 5.001273825367201e-06,
+ "loss": 0.0768,
+ "step": 24832
+ },
+ {
+ "epoch": 67.66485013623978,
+ "grad_norm": 1.7233002185821533,
+ "learning_rate": 5.0005095171684794e-06,
+ "loss": 0.0209,
+ "step": 24833
+ },
+ {
+ "epoch": 67.66757493188011,
+ "grad_norm": 2.5312986373901367,
+ "learning_rate": 4.999745247905627e-06,
+ "loss": 0.1236,
+ "step": 24834
+ },
+ {
+ "epoch": 67.67029972752043,
+ "grad_norm": 1.040313720703125,
+ "learning_rate": 4.998981017584598e-06,
+ "loss": 0.0174,
+ "step": 24835
+ },
+ {
+ "epoch": 67.67302452316076,
+ "grad_norm": 1.67762291431427,
+ "learning_rate": 4.99821682621134e-06,
+ "loss": 0.0951,
+ "step": 24836
+ },
+ {
+ "epoch": 67.6757493188011,
+ "grad_norm": 3.3999781608581543,
+ "learning_rate": 4.9974526737918095e-06,
+ "loss": 0.0388,
+ "step": 24837
+ },
+ {
+ "epoch": 67.67847411444141,
+ "grad_norm": 1.8680027723312378,
+ "learning_rate": 4.996688560331961e-06,
+ "loss": 0.0449,
+ "step": 24838
+ },
+ {
+ "epoch": 67.68119891008175,
+ "grad_norm": 1.3216164112091064,
+ "learning_rate": 4.995924485837739e-06,
+ "loss": 0.0941,
+ "step": 24839
+ },
+ {
+ "epoch": 67.68392370572207,
+ "grad_norm": 1.1712284088134766,
+ "learning_rate": 4.995160450315093e-06,
+ "loss": 0.0235,
+ "step": 24840
+ },
+ {
+ "epoch": 67.6866485013624,
+ "grad_norm": 0.7749657034873962,
+ "learning_rate": 4.99439645376998e-06,
+ "loss": 0.0117,
+ "step": 24841
+ },
+ {
+ "epoch": 67.68937329700273,
+ "grad_norm": 2.0384364128112793,
+ "learning_rate": 4.993632496208346e-06,
+ "loss": 0.1216,
+ "step": 24842
+ },
+ {
+ "epoch": 67.69209809264305,
+ "grad_norm": 1.1318020820617676,
+ "learning_rate": 4.992868577636142e-06,
+ "loss": 0.0102,
+ "step": 24843
+ },
+ {
+ "epoch": 67.69482288828338,
+ "grad_norm": 0.9840686321258545,
+ "learning_rate": 4.992104698059312e-06,
+ "loss": 0.0106,
+ "step": 24844
+ },
+ {
+ "epoch": 67.6975476839237,
+ "grad_norm": 1.2400990724563599,
+ "learning_rate": 4.991340857483813e-06,
+ "loss": 0.0184,
+ "step": 24845
+ },
+ {
+ "epoch": 67.70027247956403,
+ "grad_norm": 1.1373149156570435,
+ "learning_rate": 4.9905770559155905e-06,
+ "loss": 0.0151,
+ "step": 24846
+ },
+ {
+ "epoch": 67.70299727520435,
+ "grad_norm": 1.418765664100647,
+ "learning_rate": 4.989813293360593e-06,
+ "loss": 0.0174,
+ "step": 24847
+ },
+ {
+ "epoch": 67.70572207084469,
+ "grad_norm": 2.6830148696899414,
+ "learning_rate": 4.989049569824763e-06,
+ "loss": 0.0163,
+ "step": 24848
+ },
+ {
+ "epoch": 67.70844686648502,
+ "grad_norm": 0.9604282379150391,
+ "learning_rate": 4.988285885314059e-06,
+ "loss": 0.0088,
+ "step": 24849
+ },
+ {
+ "epoch": 67.71117166212534,
+ "grad_norm": 2.8456645011901855,
+ "learning_rate": 4.987522239834422e-06,
+ "loss": 0.1508,
+ "step": 24850
+ },
+ {
+ "epoch": 67.71389645776567,
+ "grad_norm": 1.1561788320541382,
+ "learning_rate": 4.9867586333918015e-06,
+ "loss": 0.0122,
+ "step": 24851
+ },
+ {
+ "epoch": 67.71662125340599,
+ "grad_norm": 1.836125373840332,
+ "learning_rate": 4.985995065992139e-06,
+ "loss": 0.03,
+ "step": 24852
+ },
+ {
+ "epoch": 67.71934604904632,
+ "grad_norm": 2.1526272296905518,
+ "learning_rate": 4.985231537641391e-06,
+ "loss": 0.0316,
+ "step": 24853
+ },
+ {
+ "epoch": 67.72207084468666,
+ "grad_norm": 1.3313676118850708,
+ "learning_rate": 4.984468048345493e-06,
+ "loss": 0.0117,
+ "step": 24854
+ },
+ {
+ "epoch": 67.72479564032697,
+ "grad_norm": 1.5326484441757202,
+ "learning_rate": 4.983704598110401e-06,
+ "loss": 0.0297,
+ "step": 24855
+ },
+ {
+ "epoch": 67.7275204359673,
+ "grad_norm": 2.5222055912017822,
+ "learning_rate": 4.982941186942055e-06,
+ "loss": 0.0573,
+ "step": 24856
+ },
+ {
+ "epoch": 67.73024523160763,
+ "grad_norm": 1.1926920413970947,
+ "learning_rate": 4.982177814846398e-06,
+ "loss": 0.0234,
+ "step": 24857
+ },
+ {
+ "epoch": 67.73297002724796,
+ "grad_norm": 0.8933512568473816,
+ "learning_rate": 4.981414481829383e-06,
+ "loss": 0.0107,
+ "step": 24858
+ },
+ {
+ "epoch": 67.73569482288828,
+ "grad_norm": 1.890838623046875,
+ "learning_rate": 4.980651187896951e-06,
+ "loss": 0.1494,
+ "step": 24859
+ },
+ {
+ "epoch": 67.73841961852861,
+ "grad_norm": 1.2350085973739624,
+ "learning_rate": 4.979887933055045e-06,
+ "loss": 0.0259,
+ "step": 24860
+ },
+ {
+ "epoch": 67.74114441416894,
+ "grad_norm": 1.4289393424987793,
+ "learning_rate": 4.979124717309607e-06,
+ "loss": 0.0224,
+ "step": 24861
+ },
+ {
+ "epoch": 67.74386920980926,
+ "grad_norm": 2.030320167541504,
+ "learning_rate": 4.978361540666587e-06,
+ "loss": 0.0167,
+ "step": 24862
+ },
+ {
+ "epoch": 67.7465940054496,
+ "grad_norm": 1.3926193714141846,
+ "learning_rate": 4.977598403131927e-06,
+ "loss": 0.1151,
+ "step": 24863
+ },
+ {
+ "epoch": 67.74931880108991,
+ "grad_norm": 1.8121997117996216,
+ "learning_rate": 4.976835304711565e-06,
+ "loss": 0.1091,
+ "step": 24864
+ },
+ {
+ "epoch": 67.75204359673025,
+ "grad_norm": 2.1384520530700684,
+ "learning_rate": 4.976072245411451e-06,
+ "loss": 0.0143,
+ "step": 24865
+ },
+ {
+ "epoch": 67.75476839237058,
+ "grad_norm": 2.3780624866485596,
+ "learning_rate": 4.9753092252375245e-06,
+ "loss": 0.0277,
+ "step": 24866
+ },
+ {
+ "epoch": 67.7574931880109,
+ "grad_norm": 1.2299690246582031,
+ "learning_rate": 4.974546244195729e-06,
+ "loss": 0.0136,
+ "step": 24867
+ },
+ {
+ "epoch": 67.76021798365123,
+ "grad_norm": 1.9167064428329468,
+ "learning_rate": 4.973783302292001e-06,
+ "loss": 0.1384,
+ "step": 24868
+ },
+ {
+ "epoch": 67.76294277929155,
+ "grad_norm": 1.0118112564086914,
+ "learning_rate": 4.973020399532291e-06,
+ "loss": 0.0106,
+ "step": 24869
+ },
+ {
+ "epoch": 67.76566757493188,
+ "grad_norm": 2.023843765258789,
+ "learning_rate": 4.972257535922537e-06,
+ "loss": 0.061,
+ "step": 24870
+ },
+ {
+ "epoch": 67.7683923705722,
+ "grad_norm": 1.8789981603622437,
+ "learning_rate": 4.971494711468675e-06,
+ "loss": 0.0423,
+ "step": 24871
+ },
+ {
+ "epoch": 67.77111716621253,
+ "grad_norm": 1.627807378768921,
+ "learning_rate": 4.970731926176654e-06,
+ "loss": 0.1026,
+ "step": 24872
+ },
+ {
+ "epoch": 67.77384196185287,
+ "grad_norm": 1.3647431135177612,
+ "learning_rate": 4.969969180052407e-06,
+ "loss": 0.0157,
+ "step": 24873
+ },
+ {
+ "epoch": 67.77656675749319,
+ "grad_norm": 1.1940453052520752,
+ "learning_rate": 4.969206473101883e-06,
+ "loss": 0.0096,
+ "step": 24874
+ },
+ {
+ "epoch": 67.77929155313352,
+ "grad_norm": 1.5791149139404297,
+ "learning_rate": 4.9684438053310155e-06,
+ "loss": 0.0144,
+ "step": 24875
+ },
+ {
+ "epoch": 67.78201634877384,
+ "grad_norm": 2.122701406478882,
+ "learning_rate": 4.967681176745747e-06,
+ "loss": 0.0313,
+ "step": 24876
+ },
+ {
+ "epoch": 67.78474114441417,
+ "grad_norm": 1.4620633125305176,
+ "learning_rate": 4.966918587352011e-06,
+ "loss": 0.0344,
+ "step": 24877
+ },
+ {
+ "epoch": 67.7874659400545,
+ "grad_norm": 1.4565562009811401,
+ "learning_rate": 4.966156037155755e-06,
+ "loss": 0.0443,
+ "step": 24878
+ },
+ {
+ "epoch": 67.79019073569482,
+ "grad_norm": 1.6986234188079834,
+ "learning_rate": 4.965393526162914e-06,
+ "loss": 0.0195,
+ "step": 24879
+ },
+ {
+ "epoch": 67.79291553133515,
+ "grad_norm": 1.7204115390777588,
+ "learning_rate": 4.964631054379426e-06,
+ "loss": 0.0904,
+ "step": 24880
+ },
+ {
+ "epoch": 67.79564032697547,
+ "grad_norm": 1.4039475917816162,
+ "learning_rate": 4.963868621811227e-06,
+ "loss": 0.1253,
+ "step": 24881
+ },
+ {
+ "epoch": 67.7983651226158,
+ "grad_norm": 1.4942307472229004,
+ "learning_rate": 4.963106228464258e-06,
+ "loss": 0.0155,
+ "step": 24882
+ },
+ {
+ "epoch": 67.80108991825612,
+ "grad_norm": 2.02583909034729,
+ "learning_rate": 4.962343874344457e-06,
+ "loss": 0.024,
+ "step": 24883
+ },
+ {
+ "epoch": 67.80381471389646,
+ "grad_norm": 1.5845253467559814,
+ "learning_rate": 4.96158155945776e-06,
+ "loss": 0.027,
+ "step": 24884
+ },
+ {
+ "epoch": 67.80653950953679,
+ "grad_norm": 2.012871742248535,
+ "learning_rate": 4.9608192838100995e-06,
+ "loss": 0.0237,
+ "step": 24885
+ },
+ {
+ "epoch": 67.80926430517711,
+ "grad_norm": 1.8689802885055542,
+ "learning_rate": 4.96005704740742e-06,
+ "loss": 0.0278,
+ "step": 24886
+ },
+ {
+ "epoch": 67.81198910081744,
+ "grad_norm": 2.2512946128845215,
+ "learning_rate": 4.959294850255655e-06,
+ "loss": 0.0307,
+ "step": 24887
+ },
+ {
+ "epoch": 67.81471389645776,
+ "grad_norm": 1.9260731935501099,
+ "learning_rate": 4.958532692360735e-06,
+ "loss": 0.0405,
+ "step": 24888
+ },
+ {
+ "epoch": 67.8174386920981,
+ "grad_norm": 0.9744316339492798,
+ "learning_rate": 4.9577705737286e-06,
+ "loss": 0.0152,
+ "step": 24889
+ },
+ {
+ "epoch": 67.82016348773843,
+ "grad_norm": 1.4431250095367432,
+ "learning_rate": 4.957008494365189e-06,
+ "loss": 0.0467,
+ "step": 24890
+ },
+ {
+ "epoch": 67.82288828337875,
+ "grad_norm": 1.012550711631775,
+ "learning_rate": 4.956246454276433e-06,
+ "loss": 0.0153,
+ "step": 24891
+ },
+ {
+ "epoch": 67.82561307901908,
+ "grad_norm": 1.6779162883758545,
+ "learning_rate": 4.955484453468269e-06,
+ "loss": 0.0208,
+ "step": 24892
+ },
+ {
+ "epoch": 67.8283378746594,
+ "grad_norm": 1.1777461767196655,
+ "learning_rate": 4.954722491946624e-06,
+ "loss": 0.0129,
+ "step": 24893
+ },
+ {
+ "epoch": 67.83106267029973,
+ "grad_norm": 1.2685554027557373,
+ "learning_rate": 4.953960569717443e-06,
+ "loss": 0.0163,
+ "step": 24894
+ },
+ {
+ "epoch": 67.83378746594005,
+ "grad_norm": 1.0344417095184326,
+ "learning_rate": 4.953198686786655e-06,
+ "loss": 0.0127,
+ "step": 24895
+ },
+ {
+ "epoch": 67.83651226158038,
+ "grad_norm": 1.3319501876831055,
+ "learning_rate": 4.952436843160192e-06,
+ "loss": 0.0215,
+ "step": 24896
+ },
+ {
+ "epoch": 67.83923705722071,
+ "grad_norm": 1.5595462322235107,
+ "learning_rate": 4.951675038843985e-06,
+ "loss": 0.042,
+ "step": 24897
+ },
+ {
+ "epoch": 67.84196185286103,
+ "grad_norm": 2.387413501739502,
+ "learning_rate": 4.950913273843975e-06,
+ "loss": 0.0631,
+ "step": 24898
+ },
+ {
+ "epoch": 67.84468664850137,
+ "grad_norm": 1.7588199377059937,
+ "learning_rate": 4.950151548166089e-06,
+ "loss": 0.0158,
+ "step": 24899
+ },
+ {
+ "epoch": 67.84741144414168,
+ "grad_norm": 2.0484619140625,
+ "learning_rate": 4.949389861816259e-06,
+ "loss": 0.0257,
+ "step": 24900
+ },
+ {
+ "epoch": 67.85013623978202,
+ "grad_norm": 1.0782607793807983,
+ "learning_rate": 4.948628214800417e-06,
+ "loss": 0.0096,
+ "step": 24901
+ },
+ {
+ "epoch": 67.85286103542235,
+ "grad_norm": 0.8774116039276123,
+ "learning_rate": 4.947866607124499e-06,
+ "loss": 0.0112,
+ "step": 24902
+ },
+ {
+ "epoch": 67.85558583106267,
+ "grad_norm": 0.9063315987586975,
+ "learning_rate": 4.947105038794432e-06,
+ "loss": 0.0147,
+ "step": 24903
+ },
+ {
+ "epoch": 67.858310626703,
+ "grad_norm": 2.4160635471343994,
+ "learning_rate": 4.946343509816149e-06,
+ "loss": 0.0152,
+ "step": 24904
+ },
+ {
+ "epoch": 67.86103542234332,
+ "grad_norm": 1.6394069194793701,
+ "learning_rate": 4.945582020195576e-06,
+ "loss": 0.0269,
+ "step": 24905
+ },
+ {
+ "epoch": 67.86376021798365,
+ "grad_norm": 1.180335283279419,
+ "learning_rate": 4.944820569938649e-06,
+ "loss": 0.0126,
+ "step": 24906
+ },
+ {
+ "epoch": 67.86648501362397,
+ "grad_norm": 1.693973183631897,
+ "learning_rate": 4.9440591590512984e-06,
+ "loss": 0.0364,
+ "step": 24907
+ },
+ {
+ "epoch": 67.8692098092643,
+ "grad_norm": 1.554350733757019,
+ "learning_rate": 4.9432977875394536e-06,
+ "loss": 0.077,
+ "step": 24908
+ },
+ {
+ "epoch": 67.87193460490464,
+ "grad_norm": 2.3741912841796875,
+ "learning_rate": 4.942536455409039e-06,
+ "loss": 0.0433,
+ "step": 24909
+ },
+ {
+ "epoch": 67.87465940054496,
+ "grad_norm": 1.639587163925171,
+ "learning_rate": 4.941775162665992e-06,
+ "loss": 0.0213,
+ "step": 24910
+ },
+ {
+ "epoch": 67.87738419618529,
+ "grad_norm": 2.195801258087158,
+ "learning_rate": 4.941013909316236e-06,
+ "loss": 0.1706,
+ "step": 24911
+ },
+ {
+ "epoch": 67.88010899182561,
+ "grad_norm": 1.7595418691635132,
+ "learning_rate": 4.940252695365701e-06,
+ "loss": 0.0218,
+ "step": 24912
+ },
+ {
+ "epoch": 67.88283378746594,
+ "grad_norm": 1.947717308998108,
+ "learning_rate": 4.939491520820312e-06,
+ "loss": 0.0232,
+ "step": 24913
+ },
+ {
+ "epoch": 67.88555858310627,
+ "grad_norm": 1.9713345766067505,
+ "learning_rate": 4.938730385686004e-06,
+ "loss": 0.0278,
+ "step": 24914
+ },
+ {
+ "epoch": 67.88828337874659,
+ "grad_norm": 1.5344189405441284,
+ "learning_rate": 4.937969289968701e-06,
+ "loss": 0.0489,
+ "step": 24915
+ },
+ {
+ "epoch": 67.89100817438693,
+ "grad_norm": 1.5603550672531128,
+ "learning_rate": 4.93720823367433e-06,
+ "loss": 0.0671,
+ "step": 24916
+ },
+ {
+ "epoch": 67.89373297002724,
+ "grad_norm": 1.4542757272720337,
+ "learning_rate": 4.936447216808814e-06,
+ "loss": 0.0217,
+ "step": 24917
+ },
+ {
+ "epoch": 67.89645776566758,
+ "grad_norm": 3.2427070140838623,
+ "learning_rate": 4.935686239378088e-06,
+ "loss": 0.046,
+ "step": 24918
+ },
+ {
+ "epoch": 67.8991825613079,
+ "grad_norm": 1.1150707006454468,
+ "learning_rate": 4.934925301388075e-06,
+ "loss": 0.0099,
+ "step": 24919
+ },
+ {
+ "epoch": 67.90190735694823,
+ "grad_norm": 1.028519630432129,
+ "learning_rate": 4.9341644028447014e-06,
+ "loss": 0.0115,
+ "step": 24920
+ },
+ {
+ "epoch": 67.90463215258856,
+ "grad_norm": 0.7460854649543762,
+ "learning_rate": 4.933403543753887e-06,
+ "loss": 0.0086,
+ "step": 24921
+ },
+ {
+ "epoch": 67.90735694822888,
+ "grad_norm": 2.3885955810546875,
+ "learning_rate": 4.9326427241215666e-06,
+ "loss": 0.0363,
+ "step": 24922
+ },
+ {
+ "epoch": 67.91008174386921,
+ "grad_norm": 2.310915946960449,
+ "learning_rate": 4.9318819439536596e-06,
+ "loss": 0.0339,
+ "step": 24923
+ },
+ {
+ "epoch": 67.91280653950953,
+ "grad_norm": 1.8721590042114258,
+ "learning_rate": 4.931121203256095e-06,
+ "loss": 0.0218,
+ "step": 24924
+ },
+ {
+ "epoch": 67.91553133514986,
+ "grad_norm": 2.1360185146331787,
+ "learning_rate": 4.9303605020347965e-06,
+ "loss": 0.0446,
+ "step": 24925
+ },
+ {
+ "epoch": 67.9182561307902,
+ "grad_norm": 1.539886236190796,
+ "learning_rate": 4.929599840295683e-06,
+ "loss": 0.0394,
+ "step": 24926
+ },
+ {
+ "epoch": 67.92098092643052,
+ "grad_norm": 1.6597963571548462,
+ "learning_rate": 4.928839218044687e-06,
+ "loss": 0.038,
+ "step": 24927
+ },
+ {
+ "epoch": 67.92370572207085,
+ "grad_norm": 1.3875486850738525,
+ "learning_rate": 4.928078635287726e-06,
+ "loss": 0.0233,
+ "step": 24928
+ },
+ {
+ "epoch": 67.92643051771117,
+ "grad_norm": 1.9473533630371094,
+ "learning_rate": 4.927318092030723e-06,
+ "loss": 0.0239,
+ "step": 24929
+ },
+ {
+ "epoch": 67.9291553133515,
+ "grad_norm": 1.75277578830719,
+ "learning_rate": 4.9265575882796076e-06,
+ "loss": 0.0298,
+ "step": 24930
+ },
+ {
+ "epoch": 67.93188010899182,
+ "grad_norm": 3.2413125038146973,
+ "learning_rate": 4.925797124040298e-06,
+ "loss": 0.1562,
+ "step": 24931
+ },
+ {
+ "epoch": 67.93460490463215,
+ "grad_norm": 1.9779630899429321,
+ "learning_rate": 4.925036699318717e-06,
+ "loss": 0.0243,
+ "step": 24932
+ },
+ {
+ "epoch": 67.93732970027249,
+ "grad_norm": 1.8907397985458374,
+ "learning_rate": 4.9242763141207825e-06,
+ "loss": 0.0622,
+ "step": 24933
+ },
+ {
+ "epoch": 67.9400544959128,
+ "grad_norm": 1.5389516353607178,
+ "learning_rate": 4.9235159684524256e-06,
+ "loss": 0.1392,
+ "step": 24934
+ },
+ {
+ "epoch": 67.94277929155314,
+ "grad_norm": 1.627701759338379,
+ "learning_rate": 4.922755662319563e-06,
+ "loss": 0.0098,
+ "step": 24935
+ },
+ {
+ "epoch": 67.94550408719346,
+ "grad_norm": 1.6016812324523926,
+ "learning_rate": 4.9219953957281154e-06,
+ "loss": 0.0281,
+ "step": 24936
+ },
+ {
+ "epoch": 67.94822888283379,
+ "grad_norm": 1.3950711488723755,
+ "learning_rate": 4.9212351686839995e-06,
+ "loss": 0.0167,
+ "step": 24937
+ },
+ {
+ "epoch": 67.95095367847412,
+ "grad_norm": 1.7387731075286865,
+ "learning_rate": 4.920474981193146e-06,
+ "loss": 0.0478,
+ "step": 24938
+ },
+ {
+ "epoch": 67.95367847411444,
+ "grad_norm": 2.361875057220459,
+ "learning_rate": 4.9197148332614685e-06,
+ "loss": 0.0316,
+ "step": 24939
+ },
+ {
+ "epoch": 67.95640326975477,
+ "grad_norm": 1.0658220052719116,
+ "learning_rate": 4.9189547248948856e-06,
+ "loss": 0.0095,
+ "step": 24940
+ },
+ {
+ "epoch": 67.95912806539509,
+ "grad_norm": 1.1691794395446777,
+ "learning_rate": 4.918194656099323e-06,
+ "loss": 0.0359,
+ "step": 24941
+ },
+ {
+ "epoch": 67.96185286103542,
+ "grad_norm": 1.7580950260162354,
+ "learning_rate": 4.917434626880694e-06,
+ "loss": 0.0605,
+ "step": 24942
+ },
+ {
+ "epoch": 67.96457765667574,
+ "grad_norm": 1.5403895378112793,
+ "learning_rate": 4.916674637244924e-06,
+ "loss": 0.0368,
+ "step": 24943
+ },
+ {
+ "epoch": 67.96730245231608,
+ "grad_norm": 1.292778491973877,
+ "learning_rate": 4.915914687197928e-06,
+ "loss": 0.0276,
+ "step": 24944
+ },
+ {
+ "epoch": 67.97002724795641,
+ "grad_norm": 1.4527238607406616,
+ "learning_rate": 4.915154776745624e-06,
+ "loss": 0.0228,
+ "step": 24945
+ },
+ {
+ "epoch": 67.97275204359673,
+ "grad_norm": 1.6443601846694946,
+ "learning_rate": 4.914394905893928e-06,
+ "loss": 0.162,
+ "step": 24946
+ },
+ {
+ "epoch": 67.97547683923706,
+ "grad_norm": 1.223017692565918,
+ "learning_rate": 4.913635074648765e-06,
+ "loss": 0.0258,
+ "step": 24947
+ },
+ {
+ "epoch": 67.97820163487738,
+ "grad_norm": 0.969547688961029,
+ "learning_rate": 4.912875283016047e-06,
+ "loss": 0.0075,
+ "step": 24948
+ },
+ {
+ "epoch": 67.98092643051771,
+ "grad_norm": 2.062025308609009,
+ "learning_rate": 4.9121155310016946e-06,
+ "loss": 0.0724,
+ "step": 24949
+ },
+ {
+ "epoch": 67.98365122615803,
+ "grad_norm": 2.51444673538208,
+ "learning_rate": 4.911355818611618e-06,
+ "loss": 0.0939,
+ "step": 24950
+ },
+ {
+ "epoch": 67.98637602179836,
+ "grad_norm": 1.594890832901001,
+ "learning_rate": 4.9105961458517415e-06,
+ "loss": 0.1009,
+ "step": 24951
+ },
+ {
+ "epoch": 67.9891008174387,
+ "grad_norm": 1.2918676137924194,
+ "learning_rate": 4.909836512727978e-06,
+ "loss": 0.0325,
+ "step": 24952
+ },
+ {
+ "epoch": 67.99182561307902,
+ "grad_norm": 1.581079363822937,
+ "learning_rate": 4.909076919246242e-06,
+ "loss": 0.068,
+ "step": 24953
+ },
+ {
+ "epoch": 67.99455040871935,
+ "grad_norm": 2.1510848999023438,
+ "learning_rate": 4.908317365412454e-06,
+ "loss": 0.0843,
+ "step": 24954
+ },
+ {
+ "epoch": 67.99727520435967,
+ "grad_norm": 1.3395655155181885,
+ "learning_rate": 4.9075578512325266e-06,
+ "loss": 0.0393,
+ "step": 24955
+ },
+ {
+ "epoch": 68.0,
+ "grad_norm": 1.9803168773651123,
+ "learning_rate": 4.9067983767123736e-06,
+ "loss": 0.1122,
+ "step": 24956
+ },
+ {
+ "epoch": 68.00272479564033,
+ "grad_norm": 1.8310704231262207,
+ "learning_rate": 4.906038941857908e-06,
+ "loss": 0.0276,
+ "step": 24957
+ },
+ {
+ "epoch": 68.00544959128065,
+ "grad_norm": 1.3307384252548218,
+ "learning_rate": 4.905279546675047e-06,
+ "loss": 0.0238,
+ "step": 24958
+ },
+ {
+ "epoch": 68.00817438692098,
+ "grad_norm": 1.1205881834030151,
+ "learning_rate": 4.904520191169709e-06,
+ "loss": 0.0442,
+ "step": 24959
+ },
+ {
+ "epoch": 68.0108991825613,
+ "grad_norm": 1.4545480012893677,
+ "learning_rate": 4.903760875347804e-06,
+ "loss": 0.0173,
+ "step": 24960
+ },
+ {
+ "epoch": 68.01362397820164,
+ "grad_norm": 1.2769328355789185,
+ "learning_rate": 4.9030015992152445e-06,
+ "loss": 0.0146,
+ "step": 24961
+ },
+ {
+ "epoch": 68.01634877384195,
+ "grad_norm": 1.6081352233886719,
+ "learning_rate": 4.902242362777942e-06,
+ "loss": 0.1146,
+ "step": 24962
+ },
+ {
+ "epoch": 68.01907356948229,
+ "grad_norm": 1.1331170797348022,
+ "learning_rate": 4.901483166041815e-06,
+ "loss": 0.0612,
+ "step": 24963
+ },
+ {
+ "epoch": 68.02179836512262,
+ "grad_norm": 1.2509974241256714,
+ "learning_rate": 4.900724009012773e-06,
+ "loss": 0.0122,
+ "step": 24964
+ },
+ {
+ "epoch": 68.02452316076294,
+ "grad_norm": 1.7770732641220093,
+ "learning_rate": 4.899964891696728e-06,
+ "loss": 0.0323,
+ "step": 24965
+ },
+ {
+ "epoch": 68.02724795640327,
+ "grad_norm": 1.4764822721481323,
+ "learning_rate": 4.899205814099589e-06,
+ "loss": 0.0225,
+ "step": 24966
+ },
+ {
+ "epoch": 68.02997275204359,
+ "grad_norm": 2.826836109161377,
+ "learning_rate": 4.898446776227274e-06,
+ "loss": 0.059,
+ "step": 24967
+ },
+ {
+ "epoch": 68.03269754768392,
+ "grad_norm": 1.6119133234024048,
+ "learning_rate": 4.897687778085693e-06,
+ "loss": 0.0961,
+ "step": 24968
+ },
+ {
+ "epoch": 68.03542234332426,
+ "grad_norm": 1.1347800493240356,
+ "learning_rate": 4.896928819680753e-06,
+ "loss": 0.0115,
+ "step": 24969
+ },
+ {
+ "epoch": 68.03814713896458,
+ "grad_norm": 1.7655669450759888,
+ "learning_rate": 4.896169901018365e-06,
+ "loss": 0.0329,
+ "step": 24970
+ },
+ {
+ "epoch": 68.04087193460491,
+ "grad_norm": 1.5746241807937622,
+ "learning_rate": 4.8954110221044435e-06,
+ "loss": 0.0175,
+ "step": 24971
+ },
+ {
+ "epoch": 68.04359673024523,
+ "grad_norm": 2.063277244567871,
+ "learning_rate": 4.894652182944899e-06,
+ "loss": 0.0691,
+ "step": 24972
+ },
+ {
+ "epoch": 68.04632152588556,
+ "grad_norm": 2.2294719219207764,
+ "learning_rate": 4.893893383545633e-06,
+ "loss": 0.1196,
+ "step": 24973
+ },
+ {
+ "epoch": 68.04904632152588,
+ "grad_norm": 1.3342620134353638,
+ "learning_rate": 4.893134623912567e-06,
+ "loss": 0.075,
+ "step": 24974
+ },
+ {
+ "epoch": 68.05177111716621,
+ "grad_norm": 1.739820957183838,
+ "learning_rate": 4.892375904051599e-06,
+ "loss": 0.1162,
+ "step": 24975
+ },
+ {
+ "epoch": 68.05449591280654,
+ "grad_norm": 1.0800721645355225,
+ "learning_rate": 4.891617223968648e-06,
+ "loss": 0.0122,
+ "step": 24976
+ },
+ {
+ "epoch": 68.05722070844686,
+ "grad_norm": 1.7068392038345337,
+ "learning_rate": 4.890858583669616e-06,
+ "loss": 0.0549,
+ "step": 24977
+ },
+ {
+ "epoch": 68.0599455040872,
+ "grad_norm": 2.543778419494629,
+ "learning_rate": 4.89009998316041e-06,
+ "loss": 0.055,
+ "step": 24978
+ },
+ {
+ "epoch": 68.06267029972751,
+ "grad_norm": 1.1498717069625854,
+ "learning_rate": 4.889341422446945e-06,
+ "loss": 0.0833,
+ "step": 24979
+ },
+ {
+ "epoch": 68.06539509536785,
+ "grad_norm": 1.8368891477584839,
+ "learning_rate": 4.888582901535124e-06,
+ "loss": 0.1306,
+ "step": 24980
+ },
+ {
+ "epoch": 68.06811989100818,
+ "grad_norm": 1.345865249633789,
+ "learning_rate": 4.887824420430854e-06,
+ "loss": 0.0353,
+ "step": 24981
+ },
+ {
+ "epoch": 68.0708446866485,
+ "grad_norm": 0.8878278732299805,
+ "learning_rate": 4.887065979140041e-06,
+ "loss": 0.0094,
+ "step": 24982
+ },
+ {
+ "epoch": 68.07356948228883,
+ "grad_norm": 2.1715688705444336,
+ "learning_rate": 4.886307577668596e-06,
+ "loss": 0.0889,
+ "step": 24983
+ },
+ {
+ "epoch": 68.07629427792915,
+ "grad_norm": 2.6297850608825684,
+ "learning_rate": 4.885549216022424e-06,
+ "loss": 0.0182,
+ "step": 24984
+ },
+ {
+ "epoch": 68.07901907356948,
+ "grad_norm": 1.266608476638794,
+ "learning_rate": 4.8847908942074295e-06,
+ "loss": 0.0183,
+ "step": 24985
+ },
+ {
+ "epoch": 68.0817438692098,
+ "grad_norm": 1.2247298955917358,
+ "learning_rate": 4.884032612229515e-06,
+ "loss": 0.0128,
+ "step": 24986
+ },
+ {
+ "epoch": 68.08446866485014,
+ "grad_norm": 1.3954142332077026,
+ "learning_rate": 4.8832743700945936e-06,
+ "loss": 0.0193,
+ "step": 24987
+ },
+ {
+ "epoch": 68.08719346049047,
+ "grad_norm": 1.4383271932601929,
+ "learning_rate": 4.882516167808568e-06,
+ "loss": 0.0311,
+ "step": 24988
+ },
+ {
+ "epoch": 68.08991825613079,
+ "grad_norm": 1.496853232383728,
+ "learning_rate": 4.881758005377339e-06,
+ "loss": 0.0245,
+ "step": 24989
+ },
+ {
+ "epoch": 68.09264305177112,
+ "grad_norm": 1.8278729915618896,
+ "learning_rate": 4.880999882806813e-06,
+ "loss": 0.0211,
+ "step": 24990
+ },
+ {
+ "epoch": 68.09536784741144,
+ "grad_norm": 1.617810606956482,
+ "learning_rate": 4.880241800102897e-06,
+ "loss": 0.0151,
+ "step": 24991
+ },
+ {
+ "epoch": 68.09809264305177,
+ "grad_norm": 1.6843702793121338,
+ "learning_rate": 4.87948375727149e-06,
+ "loss": 0.2775,
+ "step": 24992
+ },
+ {
+ "epoch": 68.1008174386921,
+ "grad_norm": 1.81972336769104,
+ "learning_rate": 4.878725754318503e-06,
+ "loss": 0.0316,
+ "step": 24993
+ },
+ {
+ "epoch": 68.10354223433242,
+ "grad_norm": 1.8845739364624023,
+ "learning_rate": 4.877967791249831e-06,
+ "loss": 0.1348,
+ "step": 24994
+ },
+ {
+ "epoch": 68.10626702997276,
+ "grad_norm": 1.1668636798858643,
+ "learning_rate": 4.877209868071384e-06,
+ "loss": 0.0157,
+ "step": 24995
+ },
+ {
+ "epoch": 68.10899182561307,
+ "grad_norm": 1.9008736610412598,
+ "learning_rate": 4.876451984789062e-06,
+ "loss": 0.1088,
+ "step": 24996
+ },
+ {
+ "epoch": 68.11171662125341,
+ "grad_norm": 1.9492629766464233,
+ "learning_rate": 4.8756941414087675e-06,
+ "loss": 0.0823,
+ "step": 24997
+ },
+ {
+ "epoch": 68.11444141689373,
+ "grad_norm": 1.143001914024353,
+ "learning_rate": 4.874936337936399e-06,
+ "loss": 0.0159,
+ "step": 24998
+ },
+ {
+ "epoch": 68.11716621253406,
+ "grad_norm": 0.7910826206207275,
+ "learning_rate": 4.874178574377863e-06,
+ "loss": 0.0091,
+ "step": 24999
+ },
+ {
+ "epoch": 68.11989100817439,
+ "grad_norm": 1.151232123374939,
+ "learning_rate": 4.8734208507390615e-06,
+ "loss": 0.0156,
+ "step": 25000
+ },
+ {
+ "epoch": 68.12261580381471,
+ "grad_norm": 1.2978821992874146,
+ "learning_rate": 4.872663167025893e-06,
+ "loss": 0.0419,
+ "step": 25001
+ },
+ {
+ "epoch": 68.12534059945504,
+ "grad_norm": 1.074131965637207,
+ "learning_rate": 4.8719055232442535e-06,
+ "loss": 0.0147,
+ "step": 25002
+ },
+ {
+ "epoch": 68.12806539509536,
+ "grad_norm": 1.656753659248352,
+ "learning_rate": 4.871147919400053e-06,
+ "loss": 0.0192,
+ "step": 25003
+ },
+ {
+ "epoch": 68.1307901907357,
+ "grad_norm": 1.8670350313186646,
+ "learning_rate": 4.8703903554991885e-06,
+ "loss": 0.0135,
+ "step": 25004
+ },
+ {
+ "epoch": 68.13351498637603,
+ "grad_norm": 1.8216155767440796,
+ "learning_rate": 4.869632831547559e-06,
+ "loss": 0.0103,
+ "step": 25005
+ },
+ {
+ "epoch": 68.13623978201635,
+ "grad_norm": 1.189877986907959,
+ "learning_rate": 4.868875347551059e-06,
+ "loss": 0.0122,
+ "step": 25006
+ },
+ {
+ "epoch": 68.13896457765668,
+ "grad_norm": 0.939598798751831,
+ "learning_rate": 4.868117903515596e-06,
+ "loss": 0.0122,
+ "step": 25007
+ },
+ {
+ "epoch": 68.141689373297,
+ "grad_norm": 1.2746562957763672,
+ "learning_rate": 4.867360499447066e-06,
+ "loss": 0.016,
+ "step": 25008
+ },
+ {
+ "epoch": 68.14441416893733,
+ "grad_norm": 1.7886549234390259,
+ "learning_rate": 4.866603135351365e-06,
+ "loss": 0.0728,
+ "step": 25009
+ },
+ {
+ "epoch": 68.14713896457765,
+ "grad_norm": 1.2277213335037231,
+ "learning_rate": 4.8658458112343955e-06,
+ "loss": 0.0221,
+ "step": 25010
+ },
+ {
+ "epoch": 68.14986376021798,
+ "grad_norm": 1.40022611618042,
+ "learning_rate": 4.8650885271020495e-06,
+ "loss": 0.0329,
+ "step": 25011
+ },
+ {
+ "epoch": 68.15258855585832,
+ "grad_norm": 0.7169376015663147,
+ "learning_rate": 4.864331282960234e-06,
+ "loss": 0.0148,
+ "step": 25012
+ },
+ {
+ "epoch": 68.15531335149863,
+ "grad_norm": 1.3107035160064697,
+ "learning_rate": 4.86357407881484e-06,
+ "loss": 0.0276,
+ "step": 25013
+ },
+ {
+ "epoch": 68.15803814713897,
+ "grad_norm": 1.8274657726287842,
+ "learning_rate": 4.862816914671763e-06,
+ "loss": 0.092,
+ "step": 25014
+ },
+ {
+ "epoch": 68.16076294277929,
+ "grad_norm": 1.1153755187988281,
+ "learning_rate": 4.862059790536906e-06,
+ "loss": 0.0247,
+ "step": 25015
+ },
+ {
+ "epoch": 68.16348773841962,
+ "grad_norm": 1.9827494621276855,
+ "learning_rate": 4.861302706416162e-06,
+ "loss": 0.0269,
+ "step": 25016
+ },
+ {
+ "epoch": 68.16621253405995,
+ "grad_norm": 1.6195356845855713,
+ "learning_rate": 4.860545662315427e-06,
+ "loss": 0.0265,
+ "step": 25017
+ },
+ {
+ "epoch": 68.16893732970027,
+ "grad_norm": 1.936737298965454,
+ "learning_rate": 4.859788658240593e-06,
+ "loss": 0.0743,
+ "step": 25018
+ },
+ {
+ "epoch": 68.1716621253406,
+ "grad_norm": 1.0631808042526245,
+ "learning_rate": 4.859031694197563e-06,
+ "loss": 0.0421,
+ "step": 25019
+ },
+ {
+ "epoch": 68.17438692098092,
+ "grad_norm": 1.4467954635620117,
+ "learning_rate": 4.858274770192229e-06,
+ "loss": 0.0469,
+ "step": 25020
+ },
+ {
+ "epoch": 68.17711171662125,
+ "grad_norm": 0.9795704483985901,
+ "learning_rate": 4.8575178862304865e-06,
+ "loss": 0.0135,
+ "step": 25021
+ },
+ {
+ "epoch": 68.17983651226157,
+ "grad_norm": 2.1791369915008545,
+ "learning_rate": 4.856761042318224e-06,
+ "loss": 0.0998,
+ "step": 25022
+ },
+ {
+ "epoch": 68.1825613079019,
+ "grad_norm": 1.510029911994934,
+ "learning_rate": 4.856004238461346e-06,
+ "loss": 0.0196,
+ "step": 25023
+ },
+ {
+ "epoch": 68.18528610354224,
+ "grad_norm": 1.4550195932388306,
+ "learning_rate": 4.8552474746657405e-06,
+ "loss": 0.0578,
+ "step": 25024
+ },
+ {
+ "epoch": 68.18801089918256,
+ "grad_norm": 1.8224983215332031,
+ "learning_rate": 4.854490750937303e-06,
+ "loss": 0.0877,
+ "step": 25025
+ },
+ {
+ "epoch": 68.19073569482289,
+ "grad_norm": 1.0781198740005493,
+ "learning_rate": 4.853734067281922e-06,
+ "loss": 0.0158,
+ "step": 25026
+ },
+ {
+ "epoch": 68.19346049046321,
+ "grad_norm": 0.6547746062278748,
+ "learning_rate": 4.852977423705494e-06,
+ "loss": 0.0091,
+ "step": 25027
+ },
+ {
+ "epoch": 68.19618528610354,
+ "grad_norm": 1.4796974658966064,
+ "learning_rate": 4.852220820213915e-06,
+ "loss": 0.024,
+ "step": 25028
+ },
+ {
+ "epoch": 68.19891008174388,
+ "grad_norm": 1.5794391632080078,
+ "learning_rate": 4.851464256813076e-06,
+ "loss": 0.037,
+ "step": 25029
+ },
+ {
+ "epoch": 68.2016348773842,
+ "grad_norm": 1.2255322933197021,
+ "learning_rate": 4.850707733508868e-06,
+ "loss": 0.0172,
+ "step": 25030
+ },
+ {
+ "epoch": 68.20435967302453,
+ "grad_norm": 1.3931597471237183,
+ "learning_rate": 4.8499512503071775e-06,
+ "loss": 0.0222,
+ "step": 25031
+ },
+ {
+ "epoch": 68.20708446866485,
+ "grad_norm": 1.3833080530166626,
+ "learning_rate": 4.849194807213905e-06,
+ "loss": 0.0369,
+ "step": 25032
+ },
+ {
+ "epoch": 68.20980926430518,
+ "grad_norm": 1.6733335256576538,
+ "learning_rate": 4.848438404234937e-06,
+ "loss": 0.0774,
+ "step": 25033
+ },
+ {
+ "epoch": 68.2125340599455,
+ "grad_norm": 1.3734543323516846,
+ "learning_rate": 4.847682041376166e-06,
+ "loss": 0.0155,
+ "step": 25034
+ },
+ {
+ "epoch": 68.21525885558583,
+ "grad_norm": 1.015913486480713,
+ "learning_rate": 4.846925718643478e-06,
+ "loss": 0.0218,
+ "step": 25035
+ },
+ {
+ "epoch": 68.21798365122616,
+ "grad_norm": 1.4292066097259521,
+ "learning_rate": 4.846169436042769e-06,
+ "loss": 0.0152,
+ "step": 25036
+ },
+ {
+ "epoch": 68.22070844686648,
+ "grad_norm": 1.670823335647583,
+ "learning_rate": 4.845413193579927e-06,
+ "loss": 0.0219,
+ "step": 25037
+ },
+ {
+ "epoch": 68.22343324250681,
+ "grad_norm": 1.3049288988113403,
+ "learning_rate": 4.844656991260838e-06,
+ "loss": 0.0145,
+ "step": 25038
+ },
+ {
+ "epoch": 68.22615803814713,
+ "grad_norm": 2.073105573654175,
+ "learning_rate": 4.843900829091398e-06,
+ "loss": 0.036,
+ "step": 25039
+ },
+ {
+ "epoch": 68.22888283378747,
+ "grad_norm": 1.9801476001739502,
+ "learning_rate": 4.8431447070774916e-06,
+ "loss": 0.0425,
+ "step": 25040
+ },
+ {
+ "epoch": 68.2316076294278,
+ "grad_norm": 1.556613564491272,
+ "learning_rate": 4.842388625225009e-06,
+ "loss": 0.0547,
+ "step": 25041
+ },
+ {
+ "epoch": 68.23433242506812,
+ "grad_norm": 1.1517592668533325,
+ "learning_rate": 4.841632583539835e-06,
+ "loss": 0.0194,
+ "step": 25042
+ },
+ {
+ "epoch": 68.23705722070845,
+ "grad_norm": 1.2661693096160889,
+ "learning_rate": 4.840876582027863e-06,
+ "loss": 0.0135,
+ "step": 25043
+ },
+ {
+ "epoch": 68.23978201634877,
+ "grad_norm": 1.73773992061615,
+ "learning_rate": 4.840120620694974e-06,
+ "loss": 0.0415,
+ "step": 25044
+ },
+ {
+ "epoch": 68.2425068119891,
+ "grad_norm": 1.4424635171890259,
+ "learning_rate": 4.839364699547064e-06,
+ "loss": 0.0906,
+ "step": 25045
+ },
+ {
+ "epoch": 68.24523160762942,
+ "grad_norm": 2.7330992221832275,
+ "learning_rate": 4.8386088185900145e-06,
+ "loss": 0.1126,
+ "step": 25046
+ },
+ {
+ "epoch": 68.24795640326975,
+ "grad_norm": 1.4517390727996826,
+ "learning_rate": 4.837852977829711e-06,
+ "loss": 0.0411,
+ "step": 25047
+ },
+ {
+ "epoch": 68.25068119891009,
+ "grad_norm": 1.4747194051742554,
+ "learning_rate": 4.837097177272045e-06,
+ "loss": 0.0205,
+ "step": 25048
+ },
+ {
+ "epoch": 68.2534059945504,
+ "grad_norm": 1.2641745805740356,
+ "learning_rate": 4.836341416922901e-06,
+ "loss": 0.0341,
+ "step": 25049
+ },
+ {
+ "epoch": 68.25613079019074,
+ "grad_norm": 1.5389817953109741,
+ "learning_rate": 4.835585696788163e-06,
+ "loss": 0.0599,
+ "step": 25050
+ },
+ {
+ "epoch": 68.25885558583106,
+ "grad_norm": 1.4847640991210938,
+ "learning_rate": 4.834830016873715e-06,
+ "loss": 0.0222,
+ "step": 25051
+ },
+ {
+ "epoch": 68.26158038147139,
+ "grad_norm": 1.2190124988555908,
+ "learning_rate": 4.834074377185447e-06,
+ "loss": 0.016,
+ "step": 25052
+ },
+ {
+ "epoch": 68.26430517711172,
+ "grad_norm": 1.641715407371521,
+ "learning_rate": 4.833318777729243e-06,
+ "loss": 0.0704,
+ "step": 25053
+ },
+ {
+ "epoch": 68.26702997275204,
+ "grad_norm": 2.311363935470581,
+ "learning_rate": 4.832563218510984e-06,
+ "loss": 0.0352,
+ "step": 25054
+ },
+ {
+ "epoch": 68.26975476839237,
+ "grad_norm": 4.0733819007873535,
+ "learning_rate": 4.831807699536554e-06,
+ "loss": 0.0318,
+ "step": 25055
+ },
+ {
+ "epoch": 68.2724795640327,
+ "grad_norm": 1.779191493988037,
+ "learning_rate": 4.831052220811843e-06,
+ "loss": 0.0214,
+ "step": 25056
+ },
+ {
+ "epoch": 68.27520435967303,
+ "grad_norm": 2.0070621967315674,
+ "learning_rate": 4.8302967823427296e-06,
+ "loss": 0.0413,
+ "step": 25057
+ },
+ {
+ "epoch": 68.27792915531334,
+ "grad_norm": 2.037531614303589,
+ "learning_rate": 4.8295413841350995e-06,
+ "loss": 0.0391,
+ "step": 25058
+ },
+ {
+ "epoch": 68.28065395095368,
+ "grad_norm": 1.2664707899093628,
+ "learning_rate": 4.8287860261948305e-06,
+ "loss": 0.0324,
+ "step": 25059
+ },
+ {
+ "epoch": 68.28337874659401,
+ "grad_norm": 0.7456591129302979,
+ "learning_rate": 4.828030708527814e-06,
+ "loss": 0.0077,
+ "step": 25060
+ },
+ {
+ "epoch": 68.28610354223433,
+ "grad_norm": 1.2819310426712036,
+ "learning_rate": 4.827275431139924e-06,
+ "loss": 0.0137,
+ "step": 25061
+ },
+ {
+ "epoch": 68.28882833787466,
+ "grad_norm": 1.3775421380996704,
+ "learning_rate": 4.82652019403705e-06,
+ "loss": 0.0534,
+ "step": 25062
+ },
+ {
+ "epoch": 68.29155313351498,
+ "grad_norm": 2.3944671154022217,
+ "learning_rate": 4.825764997225066e-06,
+ "loss": 0.05,
+ "step": 25063
+ },
+ {
+ "epoch": 68.29427792915531,
+ "grad_norm": 0.9368292689323425,
+ "learning_rate": 4.825009840709862e-06,
+ "loss": 0.0102,
+ "step": 25064
+ },
+ {
+ "epoch": 68.29700272479565,
+ "grad_norm": 1.295026183128357,
+ "learning_rate": 4.8242547244973144e-06,
+ "loss": 0.0143,
+ "step": 25065
+ },
+ {
+ "epoch": 68.29972752043597,
+ "grad_norm": 0.8941887617111206,
+ "learning_rate": 4.823499648593305e-06,
+ "loss": 0.0117,
+ "step": 25066
+ },
+ {
+ "epoch": 68.3024523160763,
+ "grad_norm": 1.692528486251831,
+ "learning_rate": 4.82274461300371e-06,
+ "loss": 0.048,
+ "step": 25067
+ },
+ {
+ "epoch": 68.30517711171662,
+ "grad_norm": 1.5085351467132568,
+ "learning_rate": 4.821989617734416e-06,
+ "loss": 0.0186,
+ "step": 25068
+ },
+ {
+ "epoch": 68.30790190735695,
+ "grad_norm": 1.8446414470672607,
+ "learning_rate": 4.821234662791302e-06,
+ "loss": 0.0222,
+ "step": 25069
+ },
+ {
+ "epoch": 68.31062670299727,
+ "grad_norm": 1.3027634620666504,
+ "learning_rate": 4.820479748180245e-06,
+ "loss": 0.0198,
+ "step": 25070
+ },
+ {
+ "epoch": 68.3133514986376,
+ "grad_norm": 1.4364900588989258,
+ "learning_rate": 4.81972487390712e-06,
+ "loss": 0.0135,
+ "step": 25071
+ },
+ {
+ "epoch": 68.31607629427793,
+ "grad_norm": 1.652768611907959,
+ "learning_rate": 4.818970039977816e-06,
+ "loss": 0.0413,
+ "step": 25072
+ },
+ {
+ "epoch": 68.31880108991825,
+ "grad_norm": 1.0440086126327515,
+ "learning_rate": 4.818215246398206e-06,
+ "loss": 0.0122,
+ "step": 25073
+ },
+ {
+ "epoch": 68.32152588555859,
+ "grad_norm": 1.811306357383728,
+ "learning_rate": 4.81746049317417e-06,
+ "loss": 0.0276,
+ "step": 25074
+ },
+ {
+ "epoch": 68.3242506811989,
+ "grad_norm": 1.0450485944747925,
+ "learning_rate": 4.81670578031158e-06,
+ "loss": 0.0143,
+ "step": 25075
+ },
+ {
+ "epoch": 68.32697547683924,
+ "grad_norm": 1.0803793668746948,
+ "learning_rate": 4.815951107816322e-06,
+ "loss": 0.0131,
+ "step": 25076
+ },
+ {
+ "epoch": 68.32970027247957,
+ "grad_norm": 1.5994822978973389,
+ "learning_rate": 4.815196475694271e-06,
+ "loss": 0.0235,
+ "step": 25077
+ },
+ {
+ "epoch": 68.33242506811989,
+ "grad_norm": 1.64213228225708,
+ "learning_rate": 4.814441883951299e-06,
+ "loss": 0.0273,
+ "step": 25078
+ },
+ {
+ "epoch": 68.33514986376022,
+ "grad_norm": 1.6227035522460938,
+ "learning_rate": 4.813687332593287e-06,
+ "loss": 0.0266,
+ "step": 25079
+ },
+ {
+ "epoch": 68.33787465940054,
+ "grad_norm": 3.162851572036743,
+ "learning_rate": 4.812932821626113e-06,
+ "loss": 0.105,
+ "step": 25080
+ },
+ {
+ "epoch": 68.34059945504087,
+ "grad_norm": 1.6079845428466797,
+ "learning_rate": 4.812178351055653e-06,
+ "loss": 0.0324,
+ "step": 25081
+ },
+ {
+ "epoch": 68.34332425068119,
+ "grad_norm": 2.762925148010254,
+ "learning_rate": 4.811423920887781e-06,
+ "loss": 0.0211,
+ "step": 25082
+ },
+ {
+ "epoch": 68.34604904632153,
+ "grad_norm": 2.090258836746216,
+ "learning_rate": 4.810669531128368e-06,
+ "loss": 0.0175,
+ "step": 25083
+ },
+ {
+ "epoch": 68.34877384196186,
+ "grad_norm": 1.287258505821228,
+ "learning_rate": 4.809915181783298e-06,
+ "loss": 0.017,
+ "step": 25084
+ },
+ {
+ "epoch": 68.35149863760218,
+ "grad_norm": 1.7159905433654785,
+ "learning_rate": 4.809160872858442e-06,
+ "loss": 0.0296,
+ "step": 25085
+ },
+ {
+ "epoch": 68.35422343324251,
+ "grad_norm": 1.7354284524917603,
+ "learning_rate": 4.8084066043596724e-06,
+ "loss": 0.0229,
+ "step": 25086
+ },
+ {
+ "epoch": 68.35694822888283,
+ "grad_norm": 1.958431363105774,
+ "learning_rate": 4.807652376292863e-06,
+ "loss": 0.03,
+ "step": 25087
+ },
+ {
+ "epoch": 68.35967302452316,
+ "grad_norm": 2.3015058040618896,
+ "learning_rate": 4.806898188663892e-06,
+ "loss": 0.1433,
+ "step": 25088
+ },
+ {
+ "epoch": 68.3623978201635,
+ "grad_norm": 1.9163309335708618,
+ "learning_rate": 4.806144041478632e-06,
+ "loss": 0.031,
+ "step": 25089
+ },
+ {
+ "epoch": 68.36512261580381,
+ "grad_norm": 2.799983263015747,
+ "learning_rate": 4.8053899347429545e-06,
+ "loss": 0.1457,
+ "step": 25090
+ },
+ {
+ "epoch": 68.36784741144415,
+ "grad_norm": 2.2345664501190186,
+ "learning_rate": 4.804635868462729e-06,
+ "loss": 0.0291,
+ "step": 25091
+ },
+ {
+ "epoch": 68.37057220708446,
+ "grad_norm": 1.8876771926879883,
+ "learning_rate": 4.803881842643837e-06,
+ "loss": 0.0446,
+ "step": 25092
+ },
+ {
+ "epoch": 68.3732970027248,
+ "grad_norm": 1.1975080966949463,
+ "learning_rate": 4.803127857292145e-06,
+ "loss": 0.0244,
+ "step": 25093
+ },
+ {
+ "epoch": 68.37602179836512,
+ "grad_norm": 1.5534034967422485,
+ "learning_rate": 4.802373912413526e-06,
+ "loss": 0.0155,
+ "step": 25094
+ },
+ {
+ "epoch": 68.37874659400545,
+ "grad_norm": 1.3472527265548706,
+ "learning_rate": 4.801620008013849e-06,
+ "loss": 0.0155,
+ "step": 25095
+ },
+ {
+ "epoch": 68.38147138964578,
+ "grad_norm": 1.8677878379821777,
+ "learning_rate": 4.800866144098987e-06,
+ "loss": 0.1302,
+ "step": 25096
+ },
+ {
+ "epoch": 68.3841961852861,
+ "grad_norm": 2.0489988327026367,
+ "learning_rate": 4.800112320674816e-06,
+ "loss": 0.0293,
+ "step": 25097
+ },
+ {
+ "epoch": 68.38692098092643,
+ "grad_norm": 1.8856241703033447,
+ "learning_rate": 4.799358537747203e-06,
+ "loss": 0.0169,
+ "step": 25098
+ },
+ {
+ "epoch": 68.38964577656675,
+ "grad_norm": 2.219674587249756,
+ "learning_rate": 4.798604795322018e-06,
+ "loss": 0.144,
+ "step": 25099
+ },
+ {
+ "epoch": 68.39237057220708,
+ "grad_norm": 1.3585273027420044,
+ "learning_rate": 4.797851093405128e-06,
+ "loss": 0.0244,
+ "step": 25100
+ },
+ {
+ "epoch": 68.39509536784742,
+ "grad_norm": 0.7631403803825378,
+ "learning_rate": 4.79709743200241e-06,
+ "loss": 0.0086,
+ "step": 25101
+ },
+ {
+ "epoch": 68.39782016348774,
+ "grad_norm": 1.2802306413650513,
+ "learning_rate": 4.7963438111197306e-06,
+ "loss": 0.02,
+ "step": 25102
+ },
+ {
+ "epoch": 68.40054495912807,
+ "grad_norm": 1.1982191801071167,
+ "learning_rate": 4.795590230762953e-06,
+ "loss": 0.0156,
+ "step": 25103
+ },
+ {
+ "epoch": 68.40326975476839,
+ "grad_norm": 1.6075944900512695,
+ "learning_rate": 4.794836690937955e-06,
+ "loss": 0.0149,
+ "step": 25104
+ },
+ {
+ "epoch": 68.40599455040872,
+ "grad_norm": 1.240289568901062,
+ "learning_rate": 4.794083191650603e-06,
+ "loss": 0.1461,
+ "step": 25105
+ },
+ {
+ "epoch": 68.40871934604904,
+ "grad_norm": 2.0032384395599365,
+ "learning_rate": 4.793329732906762e-06,
+ "loss": 0.0237,
+ "step": 25106
+ },
+ {
+ "epoch": 68.41144414168937,
+ "grad_norm": 1.9882220029830933,
+ "learning_rate": 4.792576314712297e-06,
+ "loss": 0.0302,
+ "step": 25107
+ },
+ {
+ "epoch": 68.4141689373297,
+ "grad_norm": 1.5716345310211182,
+ "learning_rate": 4.7918229370730855e-06,
+ "loss": 0.0253,
+ "step": 25108
+ },
+ {
+ "epoch": 68.41689373297002,
+ "grad_norm": 1.5521477460861206,
+ "learning_rate": 4.7910695999949885e-06,
+ "loss": 0.0218,
+ "step": 25109
+ },
+ {
+ "epoch": 68.41961852861036,
+ "grad_norm": 1.1686300039291382,
+ "learning_rate": 4.790316303483874e-06,
+ "loss": 0.0275,
+ "step": 25110
+ },
+ {
+ "epoch": 68.42234332425068,
+ "grad_norm": 1.0401527881622314,
+ "learning_rate": 4.7895630475456054e-06,
+ "loss": 0.0134,
+ "step": 25111
+ },
+ {
+ "epoch": 68.42506811989101,
+ "grad_norm": 1.0782524347305298,
+ "learning_rate": 4.788809832186051e-06,
+ "loss": 0.0131,
+ "step": 25112
+ },
+ {
+ "epoch": 68.42779291553134,
+ "grad_norm": 1.8605310916900635,
+ "learning_rate": 4.788056657411082e-06,
+ "loss": 0.0311,
+ "step": 25113
+ },
+ {
+ "epoch": 68.43051771117166,
+ "grad_norm": 1.3251031637191772,
+ "learning_rate": 4.787303523226561e-06,
+ "loss": 0.1471,
+ "step": 25114
+ },
+ {
+ "epoch": 68.433242506812,
+ "grad_norm": 2.1667141914367676,
+ "learning_rate": 4.786550429638352e-06,
+ "loss": 0.0563,
+ "step": 25115
+ },
+ {
+ "epoch": 68.43596730245231,
+ "grad_norm": 1.7329235076904297,
+ "learning_rate": 4.785797376652315e-06,
+ "loss": 0.0325,
+ "step": 25116
+ },
+ {
+ "epoch": 68.43869209809264,
+ "grad_norm": 2.020015001296997,
+ "learning_rate": 4.785044364274325e-06,
+ "loss": 0.0828,
+ "step": 25117
+ },
+ {
+ "epoch": 68.44141689373296,
+ "grad_norm": 1.1611921787261963,
+ "learning_rate": 4.784291392510242e-06,
+ "loss": 0.0171,
+ "step": 25118
+ },
+ {
+ "epoch": 68.4441416893733,
+ "grad_norm": 0.8906365036964417,
+ "learning_rate": 4.78353846136593e-06,
+ "loss": 0.0081,
+ "step": 25119
+ },
+ {
+ "epoch": 68.44686648501363,
+ "grad_norm": 1.8274785280227661,
+ "learning_rate": 4.782785570847248e-06,
+ "loss": 0.0285,
+ "step": 25120
+ },
+ {
+ "epoch": 68.44959128065395,
+ "grad_norm": 0.8450594544410706,
+ "learning_rate": 4.782032720960068e-06,
+ "loss": 0.0089,
+ "step": 25121
+ },
+ {
+ "epoch": 68.45231607629428,
+ "grad_norm": 1.5169668197631836,
+ "learning_rate": 4.781279911710249e-06,
+ "loss": 0.0323,
+ "step": 25122
+ },
+ {
+ "epoch": 68.4550408719346,
+ "grad_norm": 1.3727020025253296,
+ "learning_rate": 4.780527143103655e-06,
+ "loss": 0.1033,
+ "step": 25123
+ },
+ {
+ "epoch": 68.45776566757493,
+ "grad_norm": 3.203716278076172,
+ "learning_rate": 4.779774415146143e-06,
+ "loss": 0.0433,
+ "step": 25124
+ },
+ {
+ "epoch": 68.46049046321527,
+ "grad_norm": 1.894265055656433,
+ "learning_rate": 4.779021727843585e-06,
+ "loss": 0.0223,
+ "step": 25125
+ },
+ {
+ "epoch": 68.46321525885558,
+ "grad_norm": 1.0596266984939575,
+ "learning_rate": 4.7782690812018365e-06,
+ "loss": 0.0227,
+ "step": 25126
+ },
+ {
+ "epoch": 68.46594005449592,
+ "grad_norm": 1.514128565788269,
+ "learning_rate": 4.777516475226757e-06,
+ "loss": 0.0424,
+ "step": 25127
+ },
+ {
+ "epoch": 68.46866485013624,
+ "grad_norm": 1.023906946182251,
+ "learning_rate": 4.776763909924215e-06,
+ "loss": 0.0085,
+ "step": 25128
+ },
+ {
+ "epoch": 68.47138964577657,
+ "grad_norm": 1.4168349504470825,
+ "learning_rate": 4.776011385300066e-06,
+ "loss": 0.0715,
+ "step": 25129
+ },
+ {
+ "epoch": 68.47411444141689,
+ "grad_norm": 1.426096796989441,
+ "learning_rate": 4.77525890136017e-06,
+ "loss": 0.0182,
+ "step": 25130
+ },
+ {
+ "epoch": 68.47683923705722,
+ "grad_norm": 1.623916506767273,
+ "learning_rate": 4.774506458110393e-06,
+ "loss": 0.0383,
+ "step": 25131
+ },
+ {
+ "epoch": 68.47956403269755,
+ "grad_norm": 1.3178248405456543,
+ "learning_rate": 4.773754055556588e-06,
+ "loss": 0.0183,
+ "step": 25132
+ },
+ {
+ "epoch": 68.48228882833787,
+ "grad_norm": 1.0251545906066895,
+ "learning_rate": 4.773001693704622e-06,
+ "loss": 0.0148,
+ "step": 25133
+ },
+ {
+ "epoch": 68.4850136239782,
+ "grad_norm": 1.4363889694213867,
+ "learning_rate": 4.77224937256035e-06,
+ "loss": 0.0999,
+ "step": 25134
+ },
+ {
+ "epoch": 68.48773841961852,
+ "grad_norm": 1.8714494705200195,
+ "learning_rate": 4.7714970921296325e-06,
+ "loss": 0.0406,
+ "step": 25135
+ },
+ {
+ "epoch": 68.49046321525886,
+ "grad_norm": 1.2236504554748535,
+ "learning_rate": 4.770744852418322e-06,
+ "loss": 0.0379,
+ "step": 25136
+ },
+ {
+ "epoch": 68.49318801089919,
+ "grad_norm": 2.4985718727111816,
+ "learning_rate": 4.769992653432287e-06,
+ "loss": 0.0597,
+ "step": 25137
+ },
+ {
+ "epoch": 68.49591280653951,
+ "grad_norm": 1.5915600061416626,
+ "learning_rate": 4.7692404951773815e-06,
+ "loss": 0.0327,
+ "step": 25138
+ },
+ {
+ "epoch": 68.49863760217984,
+ "grad_norm": 1.2911595106124878,
+ "learning_rate": 4.768488377659462e-06,
+ "loss": 0.0175,
+ "step": 25139
+ },
+ {
+ "epoch": 68.50136239782016,
+ "grad_norm": 1.8329871892929077,
+ "learning_rate": 4.767736300884384e-06,
+ "loss": 0.0356,
+ "step": 25140
+ },
+ {
+ "epoch": 68.50408719346049,
+ "grad_norm": 0.9223028421401978,
+ "learning_rate": 4.766984264858009e-06,
+ "loss": 0.0104,
+ "step": 25141
+ },
+ {
+ "epoch": 68.50681198910081,
+ "grad_norm": 1.7959213256835938,
+ "learning_rate": 4.766232269586193e-06,
+ "loss": 0.1132,
+ "step": 25142
+ },
+ {
+ "epoch": 68.50953678474114,
+ "grad_norm": 1.4428962469100952,
+ "learning_rate": 4.7654803150747915e-06,
+ "loss": 0.0329,
+ "step": 25143
+ },
+ {
+ "epoch": 68.51226158038148,
+ "grad_norm": 1.4827162027359009,
+ "learning_rate": 4.764728401329658e-06,
+ "loss": 0.0467,
+ "step": 25144
+ },
+ {
+ "epoch": 68.5149863760218,
+ "grad_norm": 2.0324957370758057,
+ "learning_rate": 4.763976528356654e-06,
+ "loss": 0.0196,
+ "step": 25145
+ },
+ {
+ "epoch": 68.51771117166213,
+ "grad_norm": 1.0332467555999756,
+ "learning_rate": 4.763224696161633e-06,
+ "loss": 0.009,
+ "step": 25146
+ },
+ {
+ "epoch": 68.52043596730245,
+ "grad_norm": 1.9920814037322998,
+ "learning_rate": 4.7624729047504455e-06,
+ "loss": 0.0299,
+ "step": 25147
+ },
+ {
+ "epoch": 68.52316076294278,
+ "grad_norm": 2.277684211730957,
+ "learning_rate": 4.76172115412895e-06,
+ "loss": 0.0647,
+ "step": 25148
+ },
+ {
+ "epoch": 68.52588555858311,
+ "grad_norm": 1.5683376789093018,
+ "learning_rate": 4.760969444303006e-06,
+ "loss": 0.0547,
+ "step": 25149
+ },
+ {
+ "epoch": 68.52861035422343,
+ "grad_norm": 1.4760408401489258,
+ "learning_rate": 4.7602177752784625e-06,
+ "loss": 0.0954,
+ "step": 25150
+ },
+ {
+ "epoch": 68.53133514986376,
+ "grad_norm": 1.3335310220718384,
+ "learning_rate": 4.7594661470611745e-06,
+ "loss": 0.0155,
+ "step": 25151
+ },
+ {
+ "epoch": 68.53405994550408,
+ "grad_norm": 1.2381653785705566,
+ "learning_rate": 4.7587145596569925e-06,
+ "loss": 0.0112,
+ "step": 25152
+ },
+ {
+ "epoch": 68.53678474114442,
+ "grad_norm": 1.0957460403442383,
+ "learning_rate": 4.757963013071777e-06,
+ "loss": 0.011,
+ "step": 25153
+ },
+ {
+ "epoch": 68.53950953678473,
+ "grad_norm": 2.2065670490264893,
+ "learning_rate": 4.7572115073113755e-06,
+ "loss": 0.1102,
+ "step": 25154
+ },
+ {
+ "epoch": 68.54223433242507,
+ "grad_norm": 1.4955567121505737,
+ "learning_rate": 4.756460042381642e-06,
+ "loss": 0.0213,
+ "step": 25155
+ },
+ {
+ "epoch": 68.5449591280654,
+ "grad_norm": 1.8610793352127075,
+ "learning_rate": 4.755708618288426e-06,
+ "loss": 0.0277,
+ "step": 25156
+ },
+ {
+ "epoch": 68.54768392370572,
+ "grad_norm": 1.3852640390396118,
+ "learning_rate": 4.7549572350375864e-06,
+ "loss": 0.0136,
+ "step": 25157
+ },
+ {
+ "epoch": 68.55040871934605,
+ "grad_norm": 1.715808629989624,
+ "learning_rate": 4.75420589263497e-06,
+ "loss": 0.0285,
+ "step": 25158
+ },
+ {
+ "epoch": 68.55313351498637,
+ "grad_norm": 1.6961519718170166,
+ "learning_rate": 4.753454591086431e-06,
+ "loss": 0.0412,
+ "step": 25159
+ },
+ {
+ "epoch": 68.5558583106267,
+ "grad_norm": 1.29707670211792,
+ "learning_rate": 4.7527033303978135e-06,
+ "loss": 0.015,
+ "step": 25160
+ },
+ {
+ "epoch": 68.55858310626704,
+ "grad_norm": 1.1808468103408813,
+ "learning_rate": 4.751952110574977e-06,
+ "loss": 0.0215,
+ "step": 25161
+ },
+ {
+ "epoch": 68.56130790190736,
+ "grad_norm": 1.3469040393829346,
+ "learning_rate": 4.75120093162377e-06,
+ "loss": 0.0147,
+ "step": 25162
+ },
+ {
+ "epoch": 68.56403269754769,
+ "grad_norm": 1.3176567554473877,
+ "learning_rate": 4.75044979355004e-06,
+ "loss": 0.0259,
+ "step": 25163
+ },
+ {
+ "epoch": 68.566757493188,
+ "grad_norm": 1.720221996307373,
+ "learning_rate": 4.749698696359635e-06,
+ "loss": 0.0314,
+ "step": 25164
+ },
+ {
+ "epoch": 68.56948228882834,
+ "grad_norm": 1.2290288209915161,
+ "learning_rate": 4.748947640058407e-06,
+ "loss": 0.0231,
+ "step": 25165
+ },
+ {
+ "epoch": 68.57220708446866,
+ "grad_norm": 1.2227084636688232,
+ "learning_rate": 4.74819662465221e-06,
+ "loss": 0.1276,
+ "step": 25166
+ },
+ {
+ "epoch": 68.57493188010899,
+ "grad_norm": 1.0720374584197998,
+ "learning_rate": 4.747445650146888e-06,
+ "loss": 0.0094,
+ "step": 25167
+ },
+ {
+ "epoch": 68.57765667574932,
+ "grad_norm": 1.9713537693023682,
+ "learning_rate": 4.746694716548288e-06,
+ "loss": 0.1389,
+ "step": 25168
+ },
+ {
+ "epoch": 68.58038147138964,
+ "grad_norm": 2.1770920753479004,
+ "learning_rate": 4.745943823862263e-06,
+ "loss": 0.1059,
+ "step": 25169
+ },
+ {
+ "epoch": 68.58310626702998,
+ "grad_norm": 1.5402675867080688,
+ "learning_rate": 4.745192972094659e-06,
+ "loss": 0.0123,
+ "step": 25170
+ },
+ {
+ "epoch": 68.5858310626703,
+ "grad_norm": 1.2354974746704102,
+ "learning_rate": 4.744442161251323e-06,
+ "loss": 0.1433,
+ "step": 25171
+ },
+ {
+ "epoch": 68.58855585831063,
+ "grad_norm": 1.6574640274047852,
+ "learning_rate": 4.743691391338096e-06,
+ "loss": 0.1599,
+ "step": 25172
+ },
+ {
+ "epoch": 68.59128065395096,
+ "grad_norm": 1.786641001701355,
+ "learning_rate": 4.742940662360839e-06,
+ "loss": 0.0146,
+ "step": 25173
+ },
+ {
+ "epoch": 68.59400544959128,
+ "grad_norm": 1.9577512741088867,
+ "learning_rate": 4.742189974325389e-06,
+ "loss": 0.0629,
+ "step": 25174
+ },
+ {
+ "epoch": 68.59673024523161,
+ "grad_norm": 1.2382479906082153,
+ "learning_rate": 4.741439327237594e-06,
+ "loss": 0.0189,
+ "step": 25175
+ },
+ {
+ "epoch": 68.59945504087193,
+ "grad_norm": 3.1973910331726074,
+ "learning_rate": 4.740688721103297e-06,
+ "loss": 0.0561,
+ "step": 25176
+ },
+ {
+ "epoch": 68.60217983651226,
+ "grad_norm": 2.1210856437683105,
+ "learning_rate": 4.739938155928351e-06,
+ "loss": 0.1591,
+ "step": 25177
+ },
+ {
+ "epoch": 68.60490463215258,
+ "grad_norm": 1.505834937095642,
+ "learning_rate": 4.739187631718597e-06,
+ "loss": 0.0118,
+ "step": 25178
+ },
+ {
+ "epoch": 68.60762942779292,
+ "grad_norm": 2.0322072505950928,
+ "learning_rate": 4.73843714847988e-06,
+ "loss": 0.0587,
+ "step": 25179
+ },
+ {
+ "epoch": 68.61035422343325,
+ "grad_norm": 0.924760103225708,
+ "learning_rate": 4.737686706218042e-06,
+ "loss": 0.012,
+ "step": 25180
+ },
+ {
+ "epoch": 68.61307901907357,
+ "grad_norm": 1.4835504293441772,
+ "learning_rate": 4.736936304938929e-06,
+ "loss": 0.0556,
+ "step": 25181
+ },
+ {
+ "epoch": 68.6158038147139,
+ "grad_norm": 1.7709708213806152,
+ "learning_rate": 4.736185944648393e-06,
+ "loss": 0.0243,
+ "step": 25182
+ },
+ {
+ "epoch": 68.61852861035422,
+ "grad_norm": 1.4109495878219604,
+ "learning_rate": 4.73543562535227e-06,
+ "loss": 0.0201,
+ "step": 25183
+ },
+ {
+ "epoch": 68.62125340599455,
+ "grad_norm": 1.860097885131836,
+ "learning_rate": 4.734685347056405e-06,
+ "loss": 0.0332,
+ "step": 25184
+ },
+ {
+ "epoch": 68.62397820163488,
+ "grad_norm": 1.9509159326553345,
+ "learning_rate": 4.7339351097666375e-06,
+ "loss": 0.0168,
+ "step": 25185
+ },
+ {
+ "epoch": 68.6267029972752,
+ "grad_norm": 1.5127604007720947,
+ "learning_rate": 4.733184913488817e-06,
+ "loss": 0.0159,
+ "step": 25186
+ },
+ {
+ "epoch": 68.62942779291554,
+ "grad_norm": 1.5233898162841797,
+ "learning_rate": 4.732434758228784e-06,
+ "loss": 0.1051,
+ "step": 25187
+ },
+ {
+ "epoch": 68.63215258855585,
+ "grad_norm": 3.6313533782958984,
+ "learning_rate": 4.731684643992375e-06,
+ "loss": 0.0181,
+ "step": 25188
+ },
+ {
+ "epoch": 68.63487738419619,
+ "grad_norm": 0.5003183484077454,
+ "learning_rate": 4.730934570785441e-06,
+ "loss": 0.0061,
+ "step": 25189
+ },
+ {
+ "epoch": 68.6376021798365,
+ "grad_norm": 2.5273399353027344,
+ "learning_rate": 4.730184538613818e-06,
+ "loss": 0.0574,
+ "step": 25190
+ },
+ {
+ "epoch": 68.64032697547684,
+ "grad_norm": 1.5264039039611816,
+ "learning_rate": 4.729434547483349e-06,
+ "loss": 0.0443,
+ "step": 25191
+ },
+ {
+ "epoch": 68.64305177111717,
+ "grad_norm": 1.5601310729980469,
+ "learning_rate": 4.72868459739987e-06,
+ "loss": 0.1169,
+ "step": 25192
+ },
+ {
+ "epoch": 68.64577656675749,
+ "grad_norm": 1.2541892528533936,
+ "learning_rate": 4.727934688369229e-06,
+ "loss": 0.0163,
+ "step": 25193
+ },
+ {
+ "epoch": 68.64850136239782,
+ "grad_norm": 1.2081656455993652,
+ "learning_rate": 4.727184820397263e-06,
+ "loss": 0.0154,
+ "step": 25194
+ },
+ {
+ "epoch": 68.65122615803814,
+ "grad_norm": 1.9569278955459595,
+ "learning_rate": 4.7264349934898124e-06,
+ "loss": 0.0584,
+ "step": 25195
+ },
+ {
+ "epoch": 68.65395095367847,
+ "grad_norm": 2.1728060245513916,
+ "learning_rate": 4.725685207652712e-06,
+ "loss": 0.0304,
+ "step": 25196
+ },
+ {
+ "epoch": 68.65667574931881,
+ "grad_norm": 1.4109293222427368,
+ "learning_rate": 4.724935462891811e-06,
+ "loss": 0.0285,
+ "step": 25197
+ },
+ {
+ "epoch": 68.65940054495913,
+ "grad_norm": 1.6359004974365234,
+ "learning_rate": 4.724185759212937e-06,
+ "loss": 0.0314,
+ "step": 25198
+ },
+ {
+ "epoch": 68.66212534059946,
+ "grad_norm": 1.335678219795227,
+ "learning_rate": 4.723436096621939e-06,
+ "loss": 0.0152,
+ "step": 25199
+ },
+ {
+ "epoch": 68.66485013623978,
+ "grad_norm": 1.3688236474990845,
+ "learning_rate": 4.722686475124652e-06,
+ "loss": 0.0215,
+ "step": 25200
+ },
+ {
+ "epoch": 68.66757493188011,
+ "grad_norm": 1.3768210411071777,
+ "learning_rate": 4.7219368947269074e-06,
+ "loss": 0.0335,
+ "step": 25201
+ },
+ {
+ "epoch": 68.67029972752043,
+ "grad_norm": 1.1064934730529785,
+ "learning_rate": 4.721187355434553e-06,
+ "loss": 0.0094,
+ "step": 25202
+ },
+ {
+ "epoch": 68.67302452316076,
+ "grad_norm": 1.7825454473495483,
+ "learning_rate": 4.720437857253421e-06,
+ "loss": 0.1207,
+ "step": 25203
+ },
+ {
+ "epoch": 68.6757493188011,
+ "grad_norm": 2.3377959728240967,
+ "learning_rate": 4.719688400189349e-06,
+ "loss": 0.033,
+ "step": 25204
+ },
+ {
+ "epoch": 68.67847411444141,
+ "grad_norm": 2.418731212615967,
+ "learning_rate": 4.718938984248171e-06,
+ "loss": 0.0178,
+ "step": 25205
+ },
+ {
+ "epoch": 68.68119891008175,
+ "grad_norm": 1.515897512435913,
+ "learning_rate": 4.71818960943573e-06,
+ "loss": 0.0925,
+ "step": 25206
+ },
+ {
+ "epoch": 68.68392370572207,
+ "grad_norm": 1.9335837364196777,
+ "learning_rate": 4.717440275757858e-06,
+ "loss": 0.0311,
+ "step": 25207
+ },
+ {
+ "epoch": 68.6866485013624,
+ "grad_norm": 2.0679678916931152,
+ "learning_rate": 4.71669098322039e-06,
+ "loss": 0.0291,
+ "step": 25208
+ },
+ {
+ "epoch": 68.68937329700273,
+ "grad_norm": 1.3117830753326416,
+ "learning_rate": 4.7159417318291604e-06,
+ "loss": 0.0294,
+ "step": 25209
+ },
+ {
+ "epoch": 68.69209809264305,
+ "grad_norm": 1.1076480150222778,
+ "learning_rate": 4.715192521590009e-06,
+ "loss": 0.0306,
+ "step": 25210
+ },
+ {
+ "epoch": 68.69482288828338,
+ "grad_norm": 1.5228755474090576,
+ "learning_rate": 4.7144433525087685e-06,
+ "loss": 0.1013,
+ "step": 25211
+ },
+ {
+ "epoch": 68.6975476839237,
+ "grad_norm": 1.568915605545044,
+ "learning_rate": 4.71369422459127e-06,
+ "loss": 0.0497,
+ "step": 25212
+ },
+ {
+ "epoch": 68.70027247956403,
+ "grad_norm": 1.66825270652771,
+ "learning_rate": 4.712945137843353e-06,
+ "loss": 0.0767,
+ "step": 25213
+ },
+ {
+ "epoch": 68.70299727520435,
+ "grad_norm": 1.301878809928894,
+ "learning_rate": 4.712196092270851e-06,
+ "loss": 0.0252,
+ "step": 25214
+ },
+ {
+ "epoch": 68.70572207084469,
+ "grad_norm": 1.5150753259658813,
+ "learning_rate": 4.711447087879595e-06,
+ "loss": 0.0249,
+ "step": 25215
+ },
+ {
+ "epoch": 68.70844686648502,
+ "grad_norm": 1.3542333841323853,
+ "learning_rate": 4.710698124675415e-06,
+ "loss": 0.0708,
+ "step": 25216
+ },
+ {
+ "epoch": 68.71117166212534,
+ "grad_norm": 1.6323105096817017,
+ "learning_rate": 4.709949202664147e-06,
+ "loss": 0.0185,
+ "step": 25217
+ },
+ {
+ "epoch": 68.71389645776567,
+ "grad_norm": 1.442865252494812,
+ "learning_rate": 4.709200321851629e-06,
+ "loss": 0.0084,
+ "step": 25218
+ },
+ {
+ "epoch": 68.71662125340599,
+ "grad_norm": 1.3300292491912842,
+ "learning_rate": 4.708451482243688e-06,
+ "loss": 0.0138,
+ "step": 25219
+ },
+ {
+ "epoch": 68.71934604904632,
+ "grad_norm": 2.695969820022583,
+ "learning_rate": 4.707702683846157e-06,
+ "loss": 0.1345,
+ "step": 25220
+ },
+ {
+ "epoch": 68.72207084468666,
+ "grad_norm": 1.31377375125885,
+ "learning_rate": 4.706953926664863e-06,
+ "loss": 0.0153,
+ "step": 25221
+ },
+ {
+ "epoch": 68.72479564032697,
+ "grad_norm": 1.6504260301589966,
+ "learning_rate": 4.706205210705645e-06,
+ "loss": 0.0232,
+ "step": 25222
+ },
+ {
+ "epoch": 68.7275204359673,
+ "grad_norm": 2.994518995285034,
+ "learning_rate": 4.705456535974332e-06,
+ "loss": 0.0253,
+ "step": 25223
+ },
+ {
+ "epoch": 68.73024523160763,
+ "grad_norm": 1.095881700515747,
+ "learning_rate": 4.704707902476752e-06,
+ "loss": 0.0186,
+ "step": 25224
+ },
+ {
+ "epoch": 68.73297002724796,
+ "grad_norm": 1.5574238300323486,
+ "learning_rate": 4.703959310218733e-06,
+ "loss": 0.0126,
+ "step": 25225
+ },
+ {
+ "epoch": 68.73569482288828,
+ "grad_norm": 1.553891658782959,
+ "learning_rate": 4.703210759206112e-06,
+ "loss": 0.0336,
+ "step": 25226
+ },
+ {
+ "epoch": 68.73841961852861,
+ "grad_norm": 1.3633685111999512,
+ "learning_rate": 4.702462249444714e-06,
+ "loss": 0.0712,
+ "step": 25227
+ },
+ {
+ "epoch": 68.74114441416894,
+ "grad_norm": 2.0345122814178467,
+ "learning_rate": 4.7017137809403716e-06,
+ "loss": 0.0932,
+ "step": 25228
+ },
+ {
+ "epoch": 68.74386920980926,
+ "grad_norm": 2.825486183166504,
+ "learning_rate": 4.700965353698907e-06,
+ "loss": 0.0443,
+ "step": 25229
+ },
+ {
+ "epoch": 68.7465940054496,
+ "grad_norm": 2.416398286819458,
+ "learning_rate": 4.700216967726157e-06,
+ "loss": 0.0431,
+ "step": 25230
+ },
+ {
+ "epoch": 68.74931880108991,
+ "grad_norm": 1.9248931407928467,
+ "learning_rate": 4.699468623027948e-06,
+ "loss": 0.0498,
+ "step": 25231
+ },
+ {
+ "epoch": 68.75204359673025,
+ "grad_norm": 1.5071842670440674,
+ "learning_rate": 4.698720319610105e-06,
+ "loss": 0.0656,
+ "step": 25232
+ },
+ {
+ "epoch": 68.75476839237058,
+ "grad_norm": 1.3151891231536865,
+ "learning_rate": 4.69797205747845e-06,
+ "loss": 0.0106,
+ "step": 25233
+ },
+ {
+ "epoch": 68.7574931880109,
+ "grad_norm": 1.5884486436843872,
+ "learning_rate": 4.6972238366388265e-06,
+ "loss": 0.0147,
+ "step": 25234
+ },
+ {
+ "epoch": 68.76021798365123,
+ "grad_norm": 1.4056637287139893,
+ "learning_rate": 4.696475657097054e-06,
+ "loss": 0.0245,
+ "step": 25235
+ },
+ {
+ "epoch": 68.76294277929155,
+ "grad_norm": 1.5320779085159302,
+ "learning_rate": 4.6957275188589565e-06,
+ "loss": 0.011,
+ "step": 25236
+ },
+ {
+ "epoch": 68.76566757493188,
+ "grad_norm": 1.2139892578125,
+ "learning_rate": 4.69497942193036e-06,
+ "loss": 0.0195,
+ "step": 25237
+ },
+ {
+ "epoch": 68.7683923705722,
+ "grad_norm": 1.8683608770370483,
+ "learning_rate": 4.694231366317096e-06,
+ "loss": 0.0288,
+ "step": 25238
+ },
+ {
+ "epoch": 68.77111716621253,
+ "grad_norm": 0.8696718215942383,
+ "learning_rate": 4.693483352024988e-06,
+ "loss": 0.0091,
+ "step": 25239
+ },
+ {
+ "epoch": 68.77384196185287,
+ "grad_norm": 1.5676591396331787,
+ "learning_rate": 4.6927353790598595e-06,
+ "loss": 0.0229,
+ "step": 25240
+ },
+ {
+ "epoch": 68.77656675749319,
+ "grad_norm": 1.6601784229278564,
+ "learning_rate": 4.691987447427535e-06,
+ "loss": 0.0552,
+ "step": 25241
+ },
+ {
+ "epoch": 68.77929155313352,
+ "grad_norm": 0.9277983903884888,
+ "learning_rate": 4.691239557133844e-06,
+ "loss": 0.0072,
+ "step": 25242
+ },
+ {
+ "epoch": 68.78201634877384,
+ "grad_norm": 2.402552843093872,
+ "learning_rate": 4.690491708184609e-06,
+ "loss": 0.0634,
+ "step": 25243
+ },
+ {
+ "epoch": 68.78474114441417,
+ "grad_norm": 1.5910775661468506,
+ "learning_rate": 4.689743900585654e-06,
+ "loss": 0.0368,
+ "step": 25244
+ },
+ {
+ "epoch": 68.7874659400545,
+ "grad_norm": 1.7355797290802002,
+ "learning_rate": 4.688996134342798e-06,
+ "loss": 0.026,
+ "step": 25245
+ },
+ {
+ "epoch": 68.79019073569482,
+ "grad_norm": 1.3243285417556763,
+ "learning_rate": 4.688248409461873e-06,
+ "loss": 0.0543,
+ "step": 25246
+ },
+ {
+ "epoch": 68.79291553133515,
+ "grad_norm": 1.0394656658172607,
+ "learning_rate": 4.687500725948699e-06,
+ "loss": 0.0108,
+ "step": 25247
+ },
+ {
+ "epoch": 68.79564032697547,
+ "grad_norm": 2.094489097595215,
+ "learning_rate": 4.686753083809098e-06,
+ "loss": 0.1268,
+ "step": 25248
+ },
+ {
+ "epoch": 68.7983651226158,
+ "grad_norm": 1.8448446989059448,
+ "learning_rate": 4.68600548304889e-06,
+ "loss": 0.029,
+ "step": 25249
+ },
+ {
+ "epoch": 68.80108991825612,
+ "grad_norm": 3.052872657775879,
+ "learning_rate": 4.685257923673899e-06,
+ "loss": 0.1512,
+ "step": 25250
+ },
+ {
+ "epoch": 68.80381471389646,
+ "grad_norm": 2.8949472904205322,
+ "learning_rate": 4.6845104056899525e-06,
+ "loss": 0.0177,
+ "step": 25251
+ },
+ {
+ "epoch": 68.80653950953679,
+ "grad_norm": 1.1533114910125732,
+ "learning_rate": 4.683762929102868e-06,
+ "loss": 0.0113,
+ "step": 25252
+ },
+ {
+ "epoch": 68.80926430517711,
+ "grad_norm": 1.7116124629974365,
+ "learning_rate": 4.683015493918463e-06,
+ "loss": 0.0207,
+ "step": 25253
+ },
+ {
+ "epoch": 68.81198910081744,
+ "grad_norm": 1.6796244382858276,
+ "learning_rate": 4.682268100142567e-06,
+ "loss": 0.1103,
+ "step": 25254
+ },
+ {
+ "epoch": 68.81471389645776,
+ "grad_norm": 0.8731359243392944,
+ "learning_rate": 4.681520747780994e-06,
+ "loss": 0.0168,
+ "step": 25255
+ },
+ {
+ "epoch": 68.8174386920981,
+ "grad_norm": 1.1784716844558716,
+ "learning_rate": 4.680773436839566e-06,
+ "loss": 0.0171,
+ "step": 25256
+ },
+ {
+ "epoch": 68.82016348773843,
+ "grad_norm": 2.295872688293457,
+ "learning_rate": 4.680026167324101e-06,
+ "loss": 0.0874,
+ "step": 25257
+ },
+ {
+ "epoch": 68.82288828337875,
+ "grad_norm": 1.5142799615859985,
+ "learning_rate": 4.6792789392404246e-06,
+ "loss": 0.0248,
+ "step": 25258
+ },
+ {
+ "epoch": 68.82561307901908,
+ "grad_norm": 1.1958948373794556,
+ "learning_rate": 4.678531752594352e-06,
+ "loss": 0.0156,
+ "step": 25259
+ },
+ {
+ "epoch": 68.8283378746594,
+ "grad_norm": 1.2133386135101318,
+ "learning_rate": 4.677784607391703e-06,
+ "loss": 0.0247,
+ "step": 25260
+ },
+ {
+ "epoch": 68.83106267029973,
+ "grad_norm": 1.277726411819458,
+ "learning_rate": 4.677037503638293e-06,
+ "loss": 0.1352,
+ "step": 25261
+ },
+ {
+ "epoch": 68.83378746594005,
+ "grad_norm": 3.5649147033691406,
+ "learning_rate": 4.6762904413399465e-06,
+ "loss": 0.0599,
+ "step": 25262
+ },
+ {
+ "epoch": 68.83651226158038,
+ "grad_norm": 1.367871880531311,
+ "learning_rate": 4.675543420502479e-06,
+ "loss": 0.0267,
+ "step": 25263
+ },
+ {
+ "epoch": 68.83923705722071,
+ "grad_norm": 1.050128698348999,
+ "learning_rate": 4.674796441131708e-06,
+ "loss": 0.0114,
+ "step": 25264
+ },
+ {
+ "epoch": 68.84196185286103,
+ "grad_norm": 1.8090046644210815,
+ "learning_rate": 4.6740495032334475e-06,
+ "loss": 0.0199,
+ "step": 25265
+ },
+ {
+ "epoch": 68.84468664850137,
+ "grad_norm": 1.4006166458129883,
+ "learning_rate": 4.673302606813521e-06,
+ "loss": 0.0249,
+ "step": 25266
+ },
+ {
+ "epoch": 68.84741144414168,
+ "grad_norm": 1.4349087476730347,
+ "learning_rate": 4.67255575187774e-06,
+ "loss": 0.0606,
+ "step": 25267
+ },
+ {
+ "epoch": 68.85013623978202,
+ "grad_norm": 1.8103010654449463,
+ "learning_rate": 4.671808938431927e-06,
+ "loss": 0.12,
+ "step": 25268
+ },
+ {
+ "epoch": 68.85286103542235,
+ "grad_norm": 1.2470194101333618,
+ "learning_rate": 4.671062166481893e-06,
+ "loss": 0.0848,
+ "step": 25269
+ },
+ {
+ "epoch": 68.85558583106267,
+ "grad_norm": 1.6713128089904785,
+ "learning_rate": 4.670315436033452e-06,
+ "loss": 0.0214,
+ "step": 25270
+ },
+ {
+ "epoch": 68.858310626703,
+ "grad_norm": 0.8903434872627258,
+ "learning_rate": 4.669568747092427e-06,
+ "loss": 0.0149,
+ "step": 25271
+ },
+ {
+ "epoch": 68.86103542234332,
+ "grad_norm": 1.2199655771255493,
+ "learning_rate": 4.668822099664629e-06,
+ "loss": 0.0266,
+ "step": 25272
+ },
+ {
+ "epoch": 68.86376021798365,
+ "grad_norm": 1.0514771938323975,
+ "learning_rate": 4.668075493755873e-06,
+ "loss": 0.0107,
+ "step": 25273
+ },
+ {
+ "epoch": 68.86648501362397,
+ "grad_norm": 2.257753849029541,
+ "learning_rate": 4.667328929371969e-06,
+ "loss": 0.0519,
+ "step": 25274
+ },
+ {
+ "epoch": 68.8692098092643,
+ "grad_norm": 1.4011600017547607,
+ "learning_rate": 4.66658240651874e-06,
+ "loss": 0.0149,
+ "step": 25275
+ },
+ {
+ "epoch": 68.87193460490464,
+ "grad_norm": 0.8811075687408447,
+ "learning_rate": 4.665835925201995e-06,
+ "loss": 0.0073,
+ "step": 25276
+ },
+ {
+ "epoch": 68.87465940054496,
+ "grad_norm": 1.5530093908309937,
+ "learning_rate": 4.665089485427546e-06,
+ "loss": 0.0168,
+ "step": 25277
+ },
+ {
+ "epoch": 68.87738419618529,
+ "grad_norm": 1.2580119371414185,
+ "learning_rate": 4.664343087201211e-06,
+ "loss": 0.0173,
+ "step": 25278
+ },
+ {
+ "epoch": 68.88010899182561,
+ "grad_norm": 1.8839491605758667,
+ "learning_rate": 4.6635967305288e-06,
+ "loss": 0.0115,
+ "step": 25279
+ },
+ {
+ "epoch": 68.88283378746594,
+ "grad_norm": 2.0319061279296875,
+ "learning_rate": 4.662850415416127e-06,
+ "loss": 0.0842,
+ "step": 25280
+ },
+ {
+ "epoch": 68.88555858310627,
+ "grad_norm": 1.0122029781341553,
+ "learning_rate": 4.6621041418689985e-06,
+ "loss": 0.0105,
+ "step": 25281
+ },
+ {
+ "epoch": 68.88828337874659,
+ "grad_norm": 1.752841591835022,
+ "learning_rate": 4.661357909893235e-06,
+ "loss": 0.0369,
+ "step": 25282
+ },
+ {
+ "epoch": 68.89100817438693,
+ "grad_norm": 1.5442034006118774,
+ "learning_rate": 4.660611719494644e-06,
+ "loss": 0.0338,
+ "step": 25283
+ },
+ {
+ "epoch": 68.89373297002724,
+ "grad_norm": 1.21194326877594,
+ "learning_rate": 4.6598655706790344e-06,
+ "loss": 0.0277,
+ "step": 25284
+ },
+ {
+ "epoch": 68.89645776566758,
+ "grad_norm": 1.8029407262802124,
+ "learning_rate": 4.659119463452223e-06,
+ "loss": 0.0266,
+ "step": 25285
+ },
+ {
+ "epoch": 68.8991825613079,
+ "grad_norm": 1.4728342294692993,
+ "learning_rate": 4.658373397820013e-06,
+ "loss": 0.0206,
+ "step": 25286
+ },
+ {
+ "epoch": 68.90190735694823,
+ "grad_norm": 1.6141769886016846,
+ "learning_rate": 4.657627373788224e-06,
+ "loss": 0.0158,
+ "step": 25287
+ },
+ {
+ "epoch": 68.90463215258856,
+ "grad_norm": 0.6122933030128479,
+ "learning_rate": 4.65688139136266e-06,
+ "loss": 0.0062,
+ "step": 25288
+ },
+ {
+ "epoch": 68.90735694822888,
+ "grad_norm": 1.8106380701065063,
+ "learning_rate": 4.656135450549132e-06,
+ "loss": 0.0792,
+ "step": 25289
+ },
+ {
+ "epoch": 68.91008174386921,
+ "grad_norm": 1.3393288850784302,
+ "learning_rate": 4.655389551353445e-06,
+ "loss": 0.0207,
+ "step": 25290
+ },
+ {
+ "epoch": 68.91280653950953,
+ "grad_norm": 1.9946061372756958,
+ "learning_rate": 4.654643693781417e-06,
+ "loss": 0.1228,
+ "step": 25291
+ },
+ {
+ "epoch": 68.91553133514986,
+ "grad_norm": 1.9061391353607178,
+ "learning_rate": 4.653897877838851e-06,
+ "loss": 0.0171,
+ "step": 25292
+ },
+ {
+ "epoch": 68.9182561307902,
+ "grad_norm": 1.3707623481750488,
+ "learning_rate": 4.6531521035315566e-06,
+ "loss": 0.0213,
+ "step": 25293
+ },
+ {
+ "epoch": 68.92098092643052,
+ "grad_norm": 203.0663604736328,
+ "learning_rate": 4.652406370865338e-06,
+ "loss": 0.0215,
+ "step": 25294
+ },
+ {
+ "epoch": 68.92370572207085,
+ "grad_norm": 1.3791428804397583,
+ "learning_rate": 4.65166067984601e-06,
+ "loss": 0.0209,
+ "step": 25295
+ },
+ {
+ "epoch": 68.92643051771117,
+ "grad_norm": 2.123527765274048,
+ "learning_rate": 4.650915030479376e-06,
+ "loss": 0.0954,
+ "step": 25296
+ },
+ {
+ "epoch": 68.9291553133515,
+ "grad_norm": 1.8303015232086182,
+ "learning_rate": 4.65016942277124e-06,
+ "loss": 0.2118,
+ "step": 25297
+ },
+ {
+ "epoch": 68.93188010899182,
+ "grad_norm": 1.2010921239852905,
+ "learning_rate": 4.6494238567274175e-06,
+ "loss": 0.0105,
+ "step": 25298
+ },
+ {
+ "epoch": 68.93460490463215,
+ "grad_norm": 2.194054365158081,
+ "learning_rate": 4.648678332353709e-06,
+ "loss": 0.0283,
+ "step": 25299
+ },
+ {
+ "epoch": 68.93732970027249,
+ "grad_norm": 1.9671858549118042,
+ "learning_rate": 4.64793284965592e-06,
+ "loss": 0.0167,
+ "step": 25300
+ },
+ {
+ "epoch": 68.9400544959128,
+ "grad_norm": 1.8127894401550293,
+ "learning_rate": 4.647187408639855e-06,
+ "loss": 0.0191,
+ "step": 25301
+ },
+ {
+ "epoch": 68.94277929155314,
+ "grad_norm": 1.1538844108581543,
+ "learning_rate": 4.646442009311323e-06,
+ "loss": 0.0215,
+ "step": 25302
+ },
+ {
+ "epoch": 68.94550408719346,
+ "grad_norm": 3.1881561279296875,
+ "learning_rate": 4.645696651676131e-06,
+ "loss": 0.1048,
+ "step": 25303
+ },
+ {
+ "epoch": 68.94822888283379,
+ "grad_norm": 2.629476547241211,
+ "learning_rate": 4.644951335740081e-06,
+ "loss": 0.0599,
+ "step": 25304
+ },
+ {
+ "epoch": 68.95095367847412,
+ "grad_norm": 2.0501413345336914,
+ "learning_rate": 4.644206061508979e-06,
+ "loss": 0.1092,
+ "step": 25305
+ },
+ {
+ "epoch": 68.95367847411444,
+ "grad_norm": 1.107588529586792,
+ "learning_rate": 4.643460828988623e-06,
+ "loss": 0.0121,
+ "step": 25306
+ },
+ {
+ "epoch": 68.95640326975477,
+ "grad_norm": 1.991797924041748,
+ "learning_rate": 4.642715638184825e-06,
+ "loss": 0.0208,
+ "step": 25307
+ },
+ {
+ "epoch": 68.95912806539509,
+ "grad_norm": 1.463992714881897,
+ "learning_rate": 4.641970489103386e-06,
+ "loss": 0.0199,
+ "step": 25308
+ },
+ {
+ "epoch": 68.96185286103542,
+ "grad_norm": 2.7777340412139893,
+ "learning_rate": 4.641225381750109e-06,
+ "loss": 0.0214,
+ "step": 25309
+ },
+ {
+ "epoch": 68.96457765667574,
+ "grad_norm": 2.251861572265625,
+ "learning_rate": 4.640480316130791e-06,
+ "loss": 0.0574,
+ "step": 25310
+ },
+ {
+ "epoch": 68.96730245231608,
+ "grad_norm": 1.835971474647522,
+ "learning_rate": 4.639735292251244e-06,
+ "loss": 0.0199,
+ "step": 25311
+ },
+ {
+ "epoch": 68.97002724795641,
+ "grad_norm": 1.9061579704284668,
+ "learning_rate": 4.638990310117266e-06,
+ "loss": 0.0278,
+ "step": 25312
+ },
+ {
+ "epoch": 68.97275204359673,
+ "grad_norm": 2.073045253753662,
+ "learning_rate": 4.638245369734659e-06,
+ "loss": 0.0194,
+ "step": 25313
+ },
+ {
+ "epoch": 68.97547683923706,
+ "grad_norm": 2.4687821865081787,
+ "learning_rate": 4.637500471109222e-06,
+ "loss": 0.0221,
+ "step": 25314
+ },
+ {
+ "epoch": 68.97820163487738,
+ "grad_norm": 1.8098225593566895,
+ "learning_rate": 4.6367556142467605e-06,
+ "loss": 0.0322,
+ "step": 25315
+ },
+ {
+ "epoch": 68.98092643051771,
+ "grad_norm": 1.8050918579101562,
+ "learning_rate": 4.636010799153074e-06,
+ "loss": 0.0198,
+ "step": 25316
+ },
+ {
+ "epoch": 68.98365122615803,
+ "grad_norm": 2.9421634674072266,
+ "learning_rate": 4.635266025833964e-06,
+ "loss": 0.0252,
+ "step": 25317
+ },
+ {
+ "epoch": 68.98637602179836,
+ "grad_norm": 2.0530121326446533,
+ "learning_rate": 4.6345212942952255e-06,
+ "loss": 0.0271,
+ "step": 25318
+ },
+ {
+ "epoch": 68.9891008174387,
+ "grad_norm": 2.0613796710968018,
+ "learning_rate": 4.6337766045426625e-06,
+ "loss": 0.0406,
+ "step": 25319
+ },
+ {
+ "epoch": 68.99182561307902,
+ "grad_norm": 1.8388749361038208,
+ "learning_rate": 4.633031956582077e-06,
+ "loss": 0.0224,
+ "step": 25320
+ },
+ {
+ "epoch": 68.99455040871935,
+ "grad_norm": 1.949136734008789,
+ "learning_rate": 4.6322873504192675e-06,
+ "loss": 0.0569,
+ "step": 25321
+ },
+ {
+ "epoch": 68.99727520435967,
+ "grad_norm": 2.7628066539764404,
+ "learning_rate": 4.631542786060027e-06,
+ "loss": 0.0364,
+ "step": 25322
+ },
+ {
+ "epoch": 69.0,
+ "grad_norm": 2.0940101146698,
+ "learning_rate": 4.630798263510162e-06,
+ "loss": 0.0845,
+ "step": 25323
+ },
+ {
+ "epoch": 69.00272479564033,
+ "grad_norm": 2.4331583976745605,
+ "learning_rate": 4.630053782775467e-06,
+ "loss": 0.0281,
+ "step": 25324
+ },
+ {
+ "epoch": 69.00544959128065,
+ "grad_norm": 2.55979585647583,
+ "learning_rate": 4.629309343861741e-06,
+ "loss": 0.0266,
+ "step": 25325
+ },
+ {
+ "epoch": 69.00817438692098,
+ "grad_norm": 1.5005422830581665,
+ "learning_rate": 4.628564946774777e-06,
+ "loss": 0.0207,
+ "step": 25326
+ },
+ {
+ "epoch": 69.0108991825613,
+ "grad_norm": 2.1506543159484863,
+ "learning_rate": 4.62782059152038e-06,
+ "loss": 0.0457,
+ "step": 25327
+ },
+ {
+ "epoch": 69.01362397820164,
+ "grad_norm": 1.4061781167984009,
+ "learning_rate": 4.627076278104343e-06,
+ "loss": 0.0171,
+ "step": 25328
+ },
+ {
+ "epoch": 69.01634877384195,
+ "grad_norm": 2.159322500228882,
+ "learning_rate": 4.626332006532464e-06,
+ "loss": 0.0387,
+ "step": 25329
+ },
+ {
+ "epoch": 69.01907356948229,
+ "grad_norm": 2.500032901763916,
+ "learning_rate": 4.625587776810534e-06,
+ "loss": 0.0411,
+ "step": 25330
+ },
+ {
+ "epoch": 69.02179836512262,
+ "grad_norm": 1.9971626996994019,
+ "learning_rate": 4.624843588944357e-06,
+ "loss": 0.0239,
+ "step": 25331
+ },
+ {
+ "epoch": 69.02452316076294,
+ "grad_norm": 1.2674638032913208,
+ "learning_rate": 4.624099442939726e-06,
+ "loss": 0.022,
+ "step": 25332
+ },
+ {
+ "epoch": 69.02724795640327,
+ "grad_norm": 1.9919391870498657,
+ "learning_rate": 4.623355338802434e-06,
+ "loss": 0.0323,
+ "step": 25333
+ },
+ {
+ "epoch": 69.02997275204359,
+ "grad_norm": 1.3129711151123047,
+ "learning_rate": 4.622611276538276e-06,
+ "loss": 0.0233,
+ "step": 25334
+ },
+ {
+ "epoch": 69.03269754768392,
+ "grad_norm": 2.417346954345703,
+ "learning_rate": 4.62186725615305e-06,
+ "loss": 0.0217,
+ "step": 25335
+ },
+ {
+ "epoch": 69.03542234332426,
+ "grad_norm": 2.936107635498047,
+ "learning_rate": 4.621123277652546e-06,
+ "loss": 0.0216,
+ "step": 25336
+ },
+ {
+ "epoch": 69.03814713896458,
+ "grad_norm": 1.9264425039291382,
+ "learning_rate": 4.620379341042564e-06,
+ "loss": 0.0417,
+ "step": 25337
+ },
+ {
+ "epoch": 69.04087193460491,
+ "grad_norm": 2.6183865070343018,
+ "learning_rate": 4.619635446328895e-06,
+ "loss": 0.0222,
+ "step": 25338
+ },
+ {
+ "epoch": 69.04359673024523,
+ "grad_norm": 2.069636344909668,
+ "learning_rate": 4.618891593517328e-06,
+ "loss": 0.0454,
+ "step": 25339
+ },
+ {
+ "epoch": 69.04632152588556,
+ "grad_norm": 1.7607002258300781,
+ "learning_rate": 4.618147782613663e-06,
+ "loss": 0.0225,
+ "step": 25340
+ },
+ {
+ "epoch": 69.04904632152588,
+ "grad_norm": 1.3247473239898682,
+ "learning_rate": 4.617404013623692e-06,
+ "loss": 0.0131,
+ "step": 25341
+ },
+ {
+ "epoch": 69.05177111716621,
+ "grad_norm": 1.3513870239257812,
+ "learning_rate": 4.616660286553199e-06,
+ "loss": 0.0194,
+ "step": 25342
+ },
+ {
+ "epoch": 69.05449591280654,
+ "grad_norm": 2.2951035499572754,
+ "learning_rate": 4.615916601407988e-06,
+ "loss": 0.0525,
+ "step": 25343
+ },
+ {
+ "epoch": 69.05722070844686,
+ "grad_norm": 1.8518122434616089,
+ "learning_rate": 4.615172958193845e-06,
+ "loss": 0.0333,
+ "step": 25344
+ },
+ {
+ "epoch": 69.0599455040872,
+ "grad_norm": 2.1977031230926514,
+ "learning_rate": 4.614429356916561e-06,
+ "loss": 0.0551,
+ "step": 25345
+ },
+ {
+ "epoch": 69.06267029972751,
+ "grad_norm": 2.450535297393799,
+ "learning_rate": 4.613685797581925e-06,
+ "loss": 0.0331,
+ "step": 25346
+ },
+ {
+ "epoch": 69.06539509536785,
+ "grad_norm": 2.029362678527832,
+ "learning_rate": 4.612942280195734e-06,
+ "loss": 0.0505,
+ "step": 25347
+ },
+ {
+ "epoch": 69.06811989100818,
+ "grad_norm": 3.3982901573181152,
+ "learning_rate": 4.612198804763775e-06,
+ "loss": 0.0555,
+ "step": 25348
+ },
+ {
+ "epoch": 69.0708446866485,
+ "grad_norm": 2.373955249786377,
+ "learning_rate": 4.611455371291837e-06,
+ "loss": 0.1442,
+ "step": 25349
+ },
+ {
+ "epoch": 69.07356948228883,
+ "grad_norm": 2.162957191467285,
+ "learning_rate": 4.6107119797857096e-06,
+ "loss": 0.0229,
+ "step": 25350
+ },
+ {
+ "epoch": 69.07629427792915,
+ "grad_norm": 1.7145531177520752,
+ "learning_rate": 4.609968630251187e-06,
+ "loss": 0.0253,
+ "step": 25351
+ },
+ {
+ "epoch": 69.07901907356948,
+ "grad_norm": 1.3388299942016602,
+ "learning_rate": 4.6092253226940546e-06,
+ "loss": 0.0147,
+ "step": 25352
+ },
+ {
+ "epoch": 69.0817438692098,
+ "grad_norm": 2.00292706489563,
+ "learning_rate": 4.608482057120099e-06,
+ "loss": 0.0579,
+ "step": 25353
+ },
+ {
+ "epoch": 69.08446866485014,
+ "grad_norm": 1.6571016311645508,
+ "learning_rate": 4.6077388335351144e-06,
+ "loss": 0.0185,
+ "step": 25354
+ },
+ {
+ "epoch": 69.08719346049047,
+ "grad_norm": 1.5642179250717163,
+ "learning_rate": 4.6069956519448836e-06,
+ "loss": 0.0259,
+ "step": 25355
+ },
+ {
+ "epoch": 69.08991825613079,
+ "grad_norm": 1.9417765140533447,
+ "learning_rate": 4.606252512355199e-06,
+ "loss": 0.0224,
+ "step": 25356
+ },
+ {
+ "epoch": 69.09264305177112,
+ "grad_norm": 1.5484719276428223,
+ "learning_rate": 4.605509414771847e-06,
+ "loss": 0.1556,
+ "step": 25357
+ },
+ {
+ "epoch": 69.09536784741144,
+ "grad_norm": 1.0176548957824707,
+ "learning_rate": 4.6047663592006155e-06,
+ "loss": 0.014,
+ "step": 25358
+ },
+ {
+ "epoch": 69.09809264305177,
+ "grad_norm": 1.7311290502548218,
+ "learning_rate": 4.6040233456472845e-06,
+ "loss": 0.0161,
+ "step": 25359
+ },
+ {
+ "epoch": 69.1008174386921,
+ "grad_norm": 1.6195895671844482,
+ "learning_rate": 4.60328037411765e-06,
+ "loss": 0.0165,
+ "step": 25360
+ },
+ {
+ "epoch": 69.10354223433242,
+ "grad_norm": 2.018592357635498,
+ "learning_rate": 4.602537444617495e-06,
+ "loss": 0.044,
+ "step": 25361
+ },
+ {
+ "epoch": 69.10626702997276,
+ "grad_norm": 1.548484206199646,
+ "learning_rate": 4.601794557152602e-06,
+ "loss": 0.0634,
+ "step": 25362
+ },
+ {
+ "epoch": 69.10899182561307,
+ "grad_norm": 3.5425281524658203,
+ "learning_rate": 4.60105171172876e-06,
+ "loss": 0.0549,
+ "step": 25363
+ },
+ {
+ "epoch": 69.11171662125341,
+ "grad_norm": 1.0264264345169067,
+ "learning_rate": 4.6003089083517545e-06,
+ "loss": 0.0121,
+ "step": 25364
+ },
+ {
+ "epoch": 69.11444141689373,
+ "grad_norm": 2.3353281021118164,
+ "learning_rate": 4.599566147027371e-06,
+ "loss": 0.1058,
+ "step": 25365
+ },
+ {
+ "epoch": 69.11716621253406,
+ "grad_norm": 1.4783263206481934,
+ "learning_rate": 4.598823427761387e-06,
+ "loss": 0.0204,
+ "step": 25366
+ },
+ {
+ "epoch": 69.11989100817439,
+ "grad_norm": 1.316915512084961,
+ "learning_rate": 4.598080750559597e-06,
+ "loss": 0.0157,
+ "step": 25367
+ },
+ {
+ "epoch": 69.12261580381471,
+ "grad_norm": 0.917526125907898,
+ "learning_rate": 4.597338115427781e-06,
+ "loss": 0.0097,
+ "step": 25368
+ },
+ {
+ "epoch": 69.12534059945504,
+ "grad_norm": 1.6872724294662476,
+ "learning_rate": 4.596595522371721e-06,
+ "loss": 0.023,
+ "step": 25369
+ },
+ {
+ "epoch": 69.12806539509536,
+ "grad_norm": 1.3500981330871582,
+ "learning_rate": 4.595852971397197e-06,
+ "loss": 0.0248,
+ "step": 25370
+ },
+ {
+ "epoch": 69.1307901907357,
+ "grad_norm": 1.6973998546600342,
+ "learning_rate": 4.595110462509996e-06,
+ "loss": 0.0175,
+ "step": 25371
+ },
+ {
+ "epoch": 69.13351498637603,
+ "grad_norm": 1.3534506559371948,
+ "learning_rate": 4.594367995715905e-06,
+ "loss": 0.0662,
+ "step": 25372
+ },
+ {
+ "epoch": 69.13623978201635,
+ "grad_norm": 3.4794585704803467,
+ "learning_rate": 4.593625571020702e-06,
+ "loss": 0.0401,
+ "step": 25373
+ },
+ {
+ "epoch": 69.13896457765668,
+ "grad_norm": 1.4349665641784668,
+ "learning_rate": 4.592883188430169e-06,
+ "loss": 0.0134,
+ "step": 25374
+ },
+ {
+ "epoch": 69.141689373297,
+ "grad_norm": 1.4066561460494995,
+ "learning_rate": 4.592140847950085e-06,
+ "loss": 0.0179,
+ "step": 25375
+ },
+ {
+ "epoch": 69.14441416893733,
+ "grad_norm": 1.7502706050872803,
+ "learning_rate": 4.591398549586237e-06,
+ "loss": 0.1598,
+ "step": 25376
+ },
+ {
+ "epoch": 69.14713896457765,
+ "grad_norm": 2.6054329872131348,
+ "learning_rate": 4.5906562933444024e-06,
+ "loss": 0.0597,
+ "step": 25377
+ },
+ {
+ "epoch": 69.14986376021798,
+ "grad_norm": 1.936765193939209,
+ "learning_rate": 4.589914079230363e-06,
+ "loss": 0.035,
+ "step": 25378
+ },
+ {
+ "epoch": 69.15258855585832,
+ "grad_norm": 2.0636680126190186,
+ "learning_rate": 4.589171907249895e-06,
+ "loss": 0.0224,
+ "step": 25379
+ },
+ {
+ "epoch": 69.15531335149863,
+ "grad_norm": 1.7290648221969604,
+ "learning_rate": 4.588429777408785e-06,
+ "loss": 0.0999,
+ "step": 25380
+ },
+ {
+ "epoch": 69.15803814713897,
+ "grad_norm": 1.4974722862243652,
+ "learning_rate": 4.587687689712811e-06,
+ "loss": 0.0149,
+ "step": 25381
+ },
+ {
+ "epoch": 69.16076294277929,
+ "grad_norm": 0.8033252358436584,
+ "learning_rate": 4.586945644167751e-06,
+ "loss": 0.0106,
+ "step": 25382
+ },
+ {
+ "epoch": 69.16348773841962,
+ "grad_norm": 1.501081943511963,
+ "learning_rate": 4.586203640779379e-06,
+ "loss": 0.1348,
+ "step": 25383
+ },
+ {
+ "epoch": 69.16621253405995,
+ "grad_norm": 1.5387723445892334,
+ "learning_rate": 4.5854616795534835e-06,
+ "loss": 0.0123,
+ "step": 25384
+ },
+ {
+ "epoch": 69.16893732970027,
+ "grad_norm": 1.6354947090148926,
+ "learning_rate": 4.584719760495838e-06,
+ "loss": 0.0255,
+ "step": 25385
+ },
+ {
+ "epoch": 69.1716621253406,
+ "grad_norm": 1.915472149848938,
+ "learning_rate": 4.583977883612218e-06,
+ "loss": 0.0378,
+ "step": 25386
+ },
+ {
+ "epoch": 69.17438692098092,
+ "grad_norm": 1.6130841970443726,
+ "learning_rate": 4.583236048908406e-06,
+ "loss": 0.1016,
+ "step": 25387
+ },
+ {
+ "epoch": 69.17711171662125,
+ "grad_norm": 1.165579080581665,
+ "learning_rate": 4.582494256390174e-06,
+ "loss": 0.014,
+ "step": 25388
+ },
+ {
+ "epoch": 69.17983651226157,
+ "grad_norm": 1.4991449117660522,
+ "learning_rate": 4.581752506063306e-06,
+ "loss": 0.0222,
+ "step": 25389
+ },
+ {
+ "epoch": 69.1825613079019,
+ "grad_norm": 1.8167710304260254,
+ "learning_rate": 4.581010797933576e-06,
+ "loss": 0.0357,
+ "step": 25390
+ },
+ {
+ "epoch": 69.18528610354224,
+ "grad_norm": 1.7010102272033691,
+ "learning_rate": 4.580269132006755e-06,
+ "loss": 0.0121,
+ "step": 25391
+ },
+ {
+ "epoch": 69.18801089918256,
+ "grad_norm": 2.2328810691833496,
+ "learning_rate": 4.579527508288627e-06,
+ "loss": 0.0173,
+ "step": 25392
+ },
+ {
+ "epoch": 69.19073569482289,
+ "grad_norm": 1.147802472114563,
+ "learning_rate": 4.578785926784966e-06,
+ "loss": 0.0158,
+ "step": 25393
+ },
+ {
+ "epoch": 69.19346049046321,
+ "grad_norm": 3.030881404876709,
+ "learning_rate": 4.578044387501544e-06,
+ "loss": 0.0219,
+ "step": 25394
+ },
+ {
+ "epoch": 69.19618528610354,
+ "grad_norm": 1.7236889600753784,
+ "learning_rate": 4.577302890444134e-06,
+ "loss": 0.0144,
+ "step": 25395
+ },
+ {
+ "epoch": 69.19891008174388,
+ "grad_norm": 1.6047297716140747,
+ "learning_rate": 4.57656143561852e-06,
+ "loss": 0.0156,
+ "step": 25396
+ },
+ {
+ "epoch": 69.2016348773842,
+ "grad_norm": 2.688304901123047,
+ "learning_rate": 4.575820023030469e-06,
+ "loss": 0.1077,
+ "step": 25397
+ },
+ {
+ "epoch": 69.20435967302453,
+ "grad_norm": 1.2770936489105225,
+ "learning_rate": 4.575078652685758e-06,
+ "loss": 0.0166,
+ "step": 25398
+ },
+ {
+ "epoch": 69.20708446866485,
+ "grad_norm": 1.9294191598892212,
+ "learning_rate": 4.574337324590156e-06,
+ "loss": 0.0549,
+ "step": 25399
+ },
+ {
+ "epoch": 69.20980926430518,
+ "grad_norm": 1.662245750427246,
+ "learning_rate": 4.573596038749444e-06,
+ "loss": 0.1427,
+ "step": 25400
+ },
+ {
+ "epoch": 69.2125340599455,
+ "grad_norm": 2.442545175552368,
+ "learning_rate": 4.572854795169392e-06,
+ "loss": 0.0342,
+ "step": 25401
+ },
+ {
+ "epoch": 69.21525885558583,
+ "grad_norm": 2.049081325531006,
+ "learning_rate": 4.572113593855771e-06,
+ "loss": 0.0289,
+ "step": 25402
+ },
+ {
+ "epoch": 69.21798365122616,
+ "grad_norm": 1.1642065048217773,
+ "learning_rate": 4.571372434814352e-06,
+ "loss": 0.0161,
+ "step": 25403
+ },
+ {
+ "epoch": 69.22070844686648,
+ "grad_norm": 3.5030300617218018,
+ "learning_rate": 4.570631318050913e-06,
+ "loss": 0.0793,
+ "step": 25404
+ },
+ {
+ "epoch": 69.22343324250681,
+ "grad_norm": 1.8481252193450928,
+ "learning_rate": 4.569890243571219e-06,
+ "loss": 0.0226,
+ "step": 25405
+ },
+ {
+ "epoch": 69.22615803814713,
+ "grad_norm": 1.3345392942428589,
+ "learning_rate": 4.569149211381049e-06,
+ "loss": 0.0304,
+ "step": 25406
+ },
+ {
+ "epoch": 69.22888283378747,
+ "grad_norm": 2.078444242477417,
+ "learning_rate": 4.568408221486166e-06,
+ "loss": 0.078,
+ "step": 25407
+ },
+ {
+ "epoch": 69.2316076294278,
+ "grad_norm": 1.247416615486145,
+ "learning_rate": 4.567667273892349e-06,
+ "loss": 0.0122,
+ "step": 25408
+ },
+ {
+ "epoch": 69.23433242506812,
+ "grad_norm": 1.4486244916915894,
+ "learning_rate": 4.566926368605364e-06,
+ "loss": 0.0246,
+ "step": 25409
+ },
+ {
+ "epoch": 69.23705722070845,
+ "grad_norm": 1.719653844833374,
+ "learning_rate": 4.566185505630981e-06,
+ "loss": 0.1194,
+ "step": 25410
+ },
+ {
+ "epoch": 69.23978201634877,
+ "grad_norm": 2.5796706676483154,
+ "learning_rate": 4.565444684974968e-06,
+ "loss": 0.1011,
+ "step": 25411
+ },
+ {
+ "epoch": 69.2425068119891,
+ "grad_norm": 2.393554449081421,
+ "learning_rate": 4.5647039066431e-06,
+ "loss": 0.0411,
+ "step": 25412
+ },
+ {
+ "epoch": 69.24523160762942,
+ "grad_norm": 2.403130531311035,
+ "learning_rate": 4.563963170641143e-06,
+ "loss": 0.1124,
+ "step": 25413
+ },
+ {
+ "epoch": 69.24795640326975,
+ "grad_norm": 1.5087889432907104,
+ "learning_rate": 4.563222476974866e-06,
+ "loss": 0.1334,
+ "step": 25414
+ },
+ {
+ "epoch": 69.25068119891009,
+ "grad_norm": 1.921161413192749,
+ "learning_rate": 4.562481825650034e-06,
+ "loss": 0.0195,
+ "step": 25415
+ },
+ {
+ "epoch": 69.2534059945504,
+ "grad_norm": 1.4475829601287842,
+ "learning_rate": 4.561741216672422e-06,
+ "loss": 0.0195,
+ "step": 25416
+ },
+ {
+ "epoch": 69.25613079019074,
+ "grad_norm": 1.8142728805541992,
+ "learning_rate": 4.5610006500477935e-06,
+ "loss": 0.0212,
+ "step": 25417
+ },
+ {
+ "epoch": 69.25885558583106,
+ "grad_norm": 1.6590710878372192,
+ "learning_rate": 4.560260125781918e-06,
+ "loss": 0.017,
+ "step": 25418
+ },
+ {
+ "epoch": 69.26158038147139,
+ "grad_norm": 1.2882970571517944,
+ "learning_rate": 4.559519643880556e-06,
+ "loss": 0.0131,
+ "step": 25419
+ },
+ {
+ "epoch": 69.26430517711172,
+ "grad_norm": 1.7427349090576172,
+ "learning_rate": 4.5587792043494865e-06,
+ "loss": 0.0289,
+ "step": 25420
+ },
+ {
+ "epoch": 69.26702997275204,
+ "grad_norm": 1.8105838298797607,
+ "learning_rate": 4.558038807194468e-06,
+ "loss": 0.0774,
+ "step": 25421
+ },
+ {
+ "epoch": 69.26975476839237,
+ "grad_norm": 1.9343677759170532,
+ "learning_rate": 4.557298452421264e-06,
+ "loss": 0.0186,
+ "step": 25422
+ },
+ {
+ "epoch": 69.2724795640327,
+ "grad_norm": 1.381667137145996,
+ "learning_rate": 4.556558140035648e-06,
+ "loss": 0.0321,
+ "step": 25423
+ },
+ {
+ "epoch": 69.27520435967303,
+ "grad_norm": 1.6456488370895386,
+ "learning_rate": 4.555817870043379e-06,
+ "loss": 0.0391,
+ "step": 25424
+ },
+ {
+ "epoch": 69.27792915531334,
+ "grad_norm": 1.8796371221542358,
+ "learning_rate": 4.55507764245023e-06,
+ "loss": 0.0169,
+ "step": 25425
+ },
+ {
+ "epoch": 69.28065395095368,
+ "grad_norm": 1.9745827913284302,
+ "learning_rate": 4.554337457261959e-06,
+ "loss": 0.027,
+ "step": 25426
+ },
+ {
+ "epoch": 69.28337874659401,
+ "grad_norm": 1.6748461723327637,
+ "learning_rate": 4.55359731448433e-06,
+ "loss": 0.0269,
+ "step": 25427
+ },
+ {
+ "epoch": 69.28610354223433,
+ "grad_norm": 1.6346914768218994,
+ "learning_rate": 4.552857214123115e-06,
+ "loss": 0.0189,
+ "step": 25428
+ },
+ {
+ "epoch": 69.28882833787466,
+ "grad_norm": 1.7418920993804932,
+ "learning_rate": 4.5521171561840696e-06,
+ "loss": 0.0363,
+ "step": 25429
+ },
+ {
+ "epoch": 69.29155313351498,
+ "grad_norm": 2.564302921295166,
+ "learning_rate": 4.551377140672962e-06,
+ "loss": 0.0612,
+ "step": 25430
+ },
+ {
+ "epoch": 69.29427792915531,
+ "grad_norm": 1.6410073041915894,
+ "learning_rate": 4.55063716759555e-06,
+ "loss": 0.0252,
+ "step": 25431
+ },
+ {
+ "epoch": 69.29700272479565,
+ "grad_norm": 1.1627624034881592,
+ "learning_rate": 4.549897236957605e-06,
+ "loss": 0.0149,
+ "step": 25432
+ },
+ {
+ "epoch": 69.29972752043597,
+ "grad_norm": 1.1689609289169312,
+ "learning_rate": 4.549157348764883e-06,
+ "loss": 0.0152,
+ "step": 25433
+ },
+ {
+ "epoch": 69.3024523160763,
+ "grad_norm": 1.487563133239746,
+ "learning_rate": 4.548417503023149e-06,
+ "loss": 0.0324,
+ "step": 25434
+ },
+ {
+ "epoch": 69.30517711171662,
+ "grad_norm": 2.003352403640747,
+ "learning_rate": 4.5476776997381615e-06,
+ "loss": 0.024,
+ "step": 25435
+ },
+ {
+ "epoch": 69.30790190735695,
+ "grad_norm": 1.4120010137557983,
+ "learning_rate": 4.5469379389156865e-06,
+ "loss": 0.0172,
+ "step": 25436
+ },
+ {
+ "epoch": 69.31062670299727,
+ "grad_norm": 2.17726469039917,
+ "learning_rate": 4.546198220561484e-06,
+ "loss": 0.0305,
+ "step": 25437
+ },
+ {
+ "epoch": 69.3133514986376,
+ "grad_norm": 1.2227535247802734,
+ "learning_rate": 4.545458544681315e-06,
+ "loss": 0.0163,
+ "step": 25438
+ },
+ {
+ "epoch": 69.31607629427793,
+ "grad_norm": 2.0018136501312256,
+ "learning_rate": 4.544718911280934e-06,
+ "loss": 0.0502,
+ "step": 25439
+ },
+ {
+ "epoch": 69.31880108991825,
+ "grad_norm": 1.9760526418685913,
+ "learning_rate": 4.5439793203661075e-06,
+ "loss": 0.0226,
+ "step": 25440
+ },
+ {
+ "epoch": 69.32152588555859,
+ "grad_norm": 1.6027960777282715,
+ "learning_rate": 4.543239771942599e-06,
+ "loss": 0.0208,
+ "step": 25441
+ },
+ {
+ "epoch": 69.3242506811989,
+ "grad_norm": 2.161454439163208,
+ "learning_rate": 4.542500266016162e-06,
+ "loss": 0.1143,
+ "step": 25442
+ },
+ {
+ "epoch": 69.32697547683924,
+ "grad_norm": 2.120209217071533,
+ "learning_rate": 4.541760802592558e-06,
+ "loss": 0.0181,
+ "step": 25443
+ },
+ {
+ "epoch": 69.32970027247957,
+ "grad_norm": 1.721305251121521,
+ "learning_rate": 4.541021381677541e-06,
+ "loss": 0.1171,
+ "step": 25444
+ },
+ {
+ "epoch": 69.33242506811989,
+ "grad_norm": 1.9090218544006348,
+ "learning_rate": 4.540282003276879e-06,
+ "loss": 0.1226,
+ "step": 25445
+ },
+ {
+ "epoch": 69.33514986376022,
+ "grad_norm": 1.764641284942627,
+ "learning_rate": 4.5395426673963235e-06,
+ "loss": 0.0883,
+ "step": 25446
+ },
+ {
+ "epoch": 69.33787465940054,
+ "grad_norm": 1.6216400861740112,
+ "learning_rate": 4.5388033740416344e-06,
+ "loss": 0.0205,
+ "step": 25447
+ },
+ {
+ "epoch": 69.34059945504087,
+ "grad_norm": 1.199304223060608,
+ "learning_rate": 4.538064123218565e-06,
+ "loss": 0.016,
+ "step": 25448
+ },
+ {
+ "epoch": 69.34332425068119,
+ "grad_norm": 1.81783926486969,
+ "learning_rate": 4.53732491493288e-06,
+ "loss": 0.0531,
+ "step": 25449
+ },
+ {
+ "epoch": 69.34604904632153,
+ "grad_norm": 1.4405258893966675,
+ "learning_rate": 4.536585749190334e-06,
+ "loss": 0.0255,
+ "step": 25450
+ },
+ {
+ "epoch": 69.34877384196186,
+ "grad_norm": 1.659808874130249,
+ "learning_rate": 4.535846625996677e-06,
+ "loss": 0.0353,
+ "step": 25451
+ },
+ {
+ "epoch": 69.35149863760218,
+ "grad_norm": 1.3050206899642944,
+ "learning_rate": 4.5351075453576755e-06,
+ "loss": 0.0665,
+ "step": 25452
+ },
+ {
+ "epoch": 69.35422343324251,
+ "grad_norm": 2.022916078567505,
+ "learning_rate": 4.53436850727908e-06,
+ "loss": 0.029,
+ "step": 25453
+ },
+ {
+ "epoch": 69.35694822888283,
+ "grad_norm": 2.032628297805786,
+ "learning_rate": 4.533629511766646e-06,
+ "loss": 0.0789,
+ "step": 25454
+ },
+ {
+ "epoch": 69.35967302452316,
+ "grad_norm": 1.6775652170181274,
+ "learning_rate": 4.5328905588261274e-06,
+ "loss": 0.0339,
+ "step": 25455
+ },
+ {
+ "epoch": 69.3623978201635,
+ "grad_norm": 1.6709784269332886,
+ "learning_rate": 4.53215164846328e-06,
+ "loss": 0.0265,
+ "step": 25456
+ },
+ {
+ "epoch": 69.36512261580381,
+ "grad_norm": 1.2264022827148438,
+ "learning_rate": 4.531412780683864e-06,
+ "loss": 0.0342,
+ "step": 25457
+ },
+ {
+ "epoch": 69.36784741144415,
+ "grad_norm": 1.7759289741516113,
+ "learning_rate": 4.530673955493629e-06,
+ "loss": 0.036,
+ "step": 25458
+ },
+ {
+ "epoch": 69.37057220708446,
+ "grad_norm": 1.105054259300232,
+ "learning_rate": 4.529935172898327e-06,
+ "loss": 0.0127,
+ "step": 25459
+ },
+ {
+ "epoch": 69.3732970027248,
+ "grad_norm": 1.7075462341308594,
+ "learning_rate": 4.529196432903713e-06,
+ "loss": 0.0226,
+ "step": 25460
+ },
+ {
+ "epoch": 69.37602179836512,
+ "grad_norm": 1.7212969064712524,
+ "learning_rate": 4.528457735515544e-06,
+ "loss": 0.0194,
+ "step": 25461
+ },
+ {
+ "epoch": 69.37874659400545,
+ "grad_norm": 0.8639616966247559,
+ "learning_rate": 4.5277190807395695e-06,
+ "loss": 0.0136,
+ "step": 25462
+ },
+ {
+ "epoch": 69.38147138964578,
+ "grad_norm": 1.4877732992172241,
+ "learning_rate": 4.526980468581543e-06,
+ "loss": 0.033,
+ "step": 25463
+ },
+ {
+ "epoch": 69.3841961852861,
+ "grad_norm": 1.4413628578186035,
+ "learning_rate": 4.526241899047213e-06,
+ "loss": 0.0163,
+ "step": 25464
+ },
+ {
+ "epoch": 69.38692098092643,
+ "grad_norm": 1.9547761678695679,
+ "learning_rate": 4.525503372142339e-06,
+ "loss": 0.0259,
+ "step": 25465
+ },
+ {
+ "epoch": 69.38964577656675,
+ "grad_norm": 2.3272809982299805,
+ "learning_rate": 4.524764887872668e-06,
+ "loss": 0.0357,
+ "step": 25466
+ },
+ {
+ "epoch": 69.39237057220708,
+ "grad_norm": 2.6187405586242676,
+ "learning_rate": 4.524026446243952e-06,
+ "loss": 0.0326,
+ "step": 25467
+ },
+ {
+ "epoch": 69.39509536784742,
+ "grad_norm": 2.005866765975952,
+ "learning_rate": 4.523288047261938e-06,
+ "loss": 0.0568,
+ "step": 25468
+ },
+ {
+ "epoch": 69.39782016348774,
+ "grad_norm": 1.8212671279907227,
+ "learning_rate": 4.522549690932384e-06,
+ "loss": 0.0534,
+ "step": 25469
+ },
+ {
+ "epoch": 69.40054495912807,
+ "grad_norm": 1.8032686710357666,
+ "learning_rate": 4.521811377261036e-06,
+ "loss": 0.0144,
+ "step": 25470
+ },
+ {
+ "epoch": 69.40326975476839,
+ "grad_norm": 1.3082023859024048,
+ "learning_rate": 4.521073106253643e-06,
+ "loss": 0.0158,
+ "step": 25471
+ },
+ {
+ "epoch": 69.40599455040872,
+ "grad_norm": 1.9137229919433594,
+ "learning_rate": 4.5203348779159585e-06,
+ "loss": 0.0292,
+ "step": 25472
+ },
+ {
+ "epoch": 69.40871934604904,
+ "grad_norm": 1.3507086038589478,
+ "learning_rate": 4.51959669225373e-06,
+ "loss": 0.0177,
+ "step": 25473
+ },
+ {
+ "epoch": 69.41144414168937,
+ "grad_norm": 0.5645122528076172,
+ "learning_rate": 4.518858549272702e-06,
+ "loss": 0.0062,
+ "step": 25474
+ },
+ {
+ "epoch": 69.4141689373297,
+ "grad_norm": 1.846602439880371,
+ "learning_rate": 4.518120448978631e-06,
+ "loss": 0.0194,
+ "step": 25475
+ },
+ {
+ "epoch": 69.41689373297002,
+ "grad_norm": 1.5392895936965942,
+ "learning_rate": 4.517382391377258e-06,
+ "loss": 0.0216,
+ "step": 25476
+ },
+ {
+ "epoch": 69.41961852861036,
+ "grad_norm": 2.4167399406433105,
+ "learning_rate": 4.516644376474339e-06,
+ "loss": 0.0474,
+ "step": 25477
+ },
+ {
+ "epoch": 69.42234332425068,
+ "grad_norm": 1.1509349346160889,
+ "learning_rate": 4.515906404275615e-06,
+ "loss": 0.0126,
+ "step": 25478
+ },
+ {
+ "epoch": 69.42506811989101,
+ "grad_norm": 1.5143756866455078,
+ "learning_rate": 4.515168474786837e-06,
+ "loss": 0.0145,
+ "step": 25479
+ },
+ {
+ "epoch": 69.42779291553134,
+ "grad_norm": 2.026817798614502,
+ "learning_rate": 4.514430588013746e-06,
+ "loss": 0.1135,
+ "step": 25480
+ },
+ {
+ "epoch": 69.43051771117166,
+ "grad_norm": 1.999630331993103,
+ "learning_rate": 4.513692743962097e-06,
+ "loss": 0.0418,
+ "step": 25481
+ },
+ {
+ "epoch": 69.433242506812,
+ "grad_norm": 0.8042027950286865,
+ "learning_rate": 4.512954942637633e-06,
+ "loss": 0.0085,
+ "step": 25482
+ },
+ {
+ "epoch": 69.43596730245231,
+ "grad_norm": 2.1297829151153564,
+ "learning_rate": 4.512217184046098e-06,
+ "loss": 0.0269,
+ "step": 25483
+ },
+ {
+ "epoch": 69.43869209809264,
+ "grad_norm": 1.695241093635559,
+ "learning_rate": 4.511479468193236e-06,
+ "loss": 0.1068,
+ "step": 25484
+ },
+ {
+ "epoch": 69.44141689373296,
+ "grad_norm": 1.536124348640442,
+ "learning_rate": 4.5107417950848e-06,
+ "loss": 0.0311,
+ "step": 25485
+ },
+ {
+ "epoch": 69.4441416893733,
+ "grad_norm": 2.6238863468170166,
+ "learning_rate": 4.510004164726529e-06,
+ "loss": 0.2321,
+ "step": 25486
+ },
+ {
+ "epoch": 69.44686648501363,
+ "grad_norm": 1.0205305814743042,
+ "learning_rate": 4.509266577124169e-06,
+ "loss": 0.0108,
+ "step": 25487
+ },
+ {
+ "epoch": 69.44959128065395,
+ "grad_norm": 1.7267554998397827,
+ "learning_rate": 4.508529032283461e-06,
+ "loss": 0.0205,
+ "step": 25488
+ },
+ {
+ "epoch": 69.45231607629428,
+ "grad_norm": 1.156014084815979,
+ "learning_rate": 4.507791530210155e-06,
+ "loss": 0.0208,
+ "step": 25489
+ },
+ {
+ "epoch": 69.4550408719346,
+ "grad_norm": 0.9904470443725586,
+ "learning_rate": 4.507054070909992e-06,
+ "loss": 0.011,
+ "step": 25490
+ },
+ {
+ "epoch": 69.45776566757493,
+ "grad_norm": 2.25018048286438,
+ "learning_rate": 4.506316654388712e-06,
+ "loss": 0.0211,
+ "step": 25491
+ },
+ {
+ "epoch": 69.46049046321527,
+ "grad_norm": 1.6690529584884644,
+ "learning_rate": 4.50557928065206e-06,
+ "loss": 0.0222,
+ "step": 25492
+ },
+ {
+ "epoch": 69.46321525885558,
+ "grad_norm": 1.5211117267608643,
+ "learning_rate": 4.5048419497057835e-06,
+ "loss": 0.026,
+ "step": 25493
+ },
+ {
+ "epoch": 69.46594005449592,
+ "grad_norm": 1.337707281112671,
+ "learning_rate": 4.504104661555623e-06,
+ "loss": 0.0132,
+ "step": 25494
+ },
+ {
+ "epoch": 69.46866485013624,
+ "grad_norm": 1.941486120223999,
+ "learning_rate": 4.503367416207316e-06,
+ "loss": 0.012,
+ "step": 25495
+ },
+ {
+ "epoch": 69.47138964577657,
+ "grad_norm": 2.1606578826904297,
+ "learning_rate": 4.502630213666605e-06,
+ "loss": 0.0421,
+ "step": 25496
+ },
+ {
+ "epoch": 69.47411444141689,
+ "grad_norm": 1.3434104919433594,
+ "learning_rate": 4.501893053939236e-06,
+ "loss": 0.0206,
+ "step": 25497
+ },
+ {
+ "epoch": 69.47683923705722,
+ "grad_norm": 2.7076194286346436,
+ "learning_rate": 4.501155937030946e-06,
+ "loss": 0.0312,
+ "step": 25498
+ },
+ {
+ "epoch": 69.47956403269755,
+ "grad_norm": 1.7914690971374512,
+ "learning_rate": 4.500418862947478e-06,
+ "loss": 0.0216,
+ "step": 25499
+ },
+ {
+ "epoch": 69.48228882833787,
+ "grad_norm": 1.2162501811981201,
+ "learning_rate": 4.4996818316945655e-06,
+ "loss": 0.0115,
+ "step": 25500
+ },
+ {
+ "epoch": 69.4850136239782,
+ "grad_norm": 1.9595059156417847,
+ "learning_rate": 4.498944843277959e-06,
+ "loss": 0.0384,
+ "step": 25501
+ },
+ {
+ "epoch": 69.48773841961852,
+ "grad_norm": 1.760869026184082,
+ "learning_rate": 4.498207897703393e-06,
+ "loss": 0.0193,
+ "step": 25502
+ },
+ {
+ "epoch": 69.49046321525886,
+ "grad_norm": 1.8932640552520752,
+ "learning_rate": 4.497470994976606e-06,
+ "loss": 0.0736,
+ "step": 25503
+ },
+ {
+ "epoch": 69.49318801089919,
+ "grad_norm": 1.8346282243728638,
+ "learning_rate": 4.496734135103334e-06,
+ "loss": 0.0769,
+ "step": 25504
+ },
+ {
+ "epoch": 69.49591280653951,
+ "grad_norm": 1.1325645446777344,
+ "learning_rate": 4.4959973180893235e-06,
+ "loss": 0.018,
+ "step": 25505
+ },
+ {
+ "epoch": 69.49863760217984,
+ "grad_norm": 1.4483758211135864,
+ "learning_rate": 4.49526054394031e-06,
+ "loss": 0.021,
+ "step": 25506
+ },
+ {
+ "epoch": 69.50136239782016,
+ "grad_norm": 1.6259287595748901,
+ "learning_rate": 4.4945238126620286e-06,
+ "loss": 0.0175,
+ "step": 25507
+ },
+ {
+ "epoch": 69.50408719346049,
+ "grad_norm": 2.053666591644287,
+ "learning_rate": 4.493787124260215e-06,
+ "loss": 0.0208,
+ "step": 25508
+ },
+ {
+ "epoch": 69.50681198910081,
+ "grad_norm": 1.8819794654846191,
+ "learning_rate": 4.49305047874061e-06,
+ "loss": 0.1256,
+ "step": 25509
+ },
+ {
+ "epoch": 69.50953678474114,
+ "grad_norm": 1.5871649980545044,
+ "learning_rate": 4.492313876108955e-06,
+ "loss": 0.0116,
+ "step": 25510
+ },
+ {
+ "epoch": 69.51226158038148,
+ "grad_norm": 1.7161200046539307,
+ "learning_rate": 4.491577316370982e-06,
+ "loss": 0.0163,
+ "step": 25511
+ },
+ {
+ "epoch": 69.5149863760218,
+ "grad_norm": 1.7838164567947388,
+ "learning_rate": 4.490840799532428e-06,
+ "loss": 0.0705,
+ "step": 25512
+ },
+ {
+ "epoch": 69.51771117166213,
+ "grad_norm": 1.5646319389343262,
+ "learning_rate": 4.490104325599024e-06,
+ "loss": 0.0225,
+ "step": 25513
+ },
+ {
+ "epoch": 69.52043596730245,
+ "grad_norm": 3.2805755138397217,
+ "learning_rate": 4.489367894576514e-06,
+ "loss": 0.026,
+ "step": 25514
+ },
+ {
+ "epoch": 69.52316076294278,
+ "grad_norm": 1.5074712038040161,
+ "learning_rate": 4.4886315064706295e-06,
+ "loss": 0.0472,
+ "step": 25515
+ },
+ {
+ "epoch": 69.52588555858311,
+ "grad_norm": 1.1777675151824951,
+ "learning_rate": 4.487895161287102e-06,
+ "loss": 0.0194,
+ "step": 25516
+ },
+ {
+ "epoch": 69.52861035422343,
+ "grad_norm": 1.6304457187652588,
+ "learning_rate": 4.487158859031674e-06,
+ "loss": 0.0165,
+ "step": 25517
+ },
+ {
+ "epoch": 69.53133514986376,
+ "grad_norm": 1.6153769493103027,
+ "learning_rate": 4.486422599710074e-06,
+ "loss": 0.041,
+ "step": 25518
+ },
+ {
+ "epoch": 69.53405994550408,
+ "grad_norm": 2.6361193656921387,
+ "learning_rate": 4.4856863833280384e-06,
+ "loss": 0.0298,
+ "step": 25519
+ },
+ {
+ "epoch": 69.53678474114442,
+ "grad_norm": 1.0885549783706665,
+ "learning_rate": 4.484950209891296e-06,
+ "loss": 0.0123,
+ "step": 25520
+ },
+ {
+ "epoch": 69.53950953678473,
+ "grad_norm": 2.224118709564209,
+ "learning_rate": 4.484214079405587e-06,
+ "loss": 0.0649,
+ "step": 25521
+ },
+ {
+ "epoch": 69.54223433242507,
+ "grad_norm": 2.3837528228759766,
+ "learning_rate": 4.483477991876642e-06,
+ "loss": 0.1243,
+ "step": 25522
+ },
+ {
+ "epoch": 69.5449591280654,
+ "grad_norm": 1.9569886922836304,
+ "learning_rate": 4.482741947310192e-06,
+ "loss": 0.0337,
+ "step": 25523
+ },
+ {
+ "epoch": 69.54768392370572,
+ "grad_norm": 1.8647689819335938,
+ "learning_rate": 4.482005945711965e-06,
+ "loss": 0.1251,
+ "step": 25524
+ },
+ {
+ "epoch": 69.55040871934605,
+ "grad_norm": 1.122204065322876,
+ "learning_rate": 4.4812699870877e-06,
+ "loss": 0.0141,
+ "step": 25525
+ },
+ {
+ "epoch": 69.55313351498637,
+ "grad_norm": 1.5631746053695679,
+ "learning_rate": 4.480534071443131e-06,
+ "loss": 0.0408,
+ "step": 25526
+ },
+ {
+ "epoch": 69.5558583106267,
+ "grad_norm": 1.1626636981964111,
+ "learning_rate": 4.479798198783984e-06,
+ "loss": 0.0118,
+ "step": 25527
+ },
+ {
+ "epoch": 69.55858310626704,
+ "grad_norm": 2.0658442974090576,
+ "learning_rate": 4.479062369115992e-06,
+ "loss": 0.0371,
+ "step": 25528
+ },
+ {
+ "epoch": 69.56130790190736,
+ "grad_norm": 1.7634135484695435,
+ "learning_rate": 4.4783265824448795e-06,
+ "loss": 0.0434,
+ "step": 25529
+ },
+ {
+ "epoch": 69.56403269754769,
+ "grad_norm": 1.5297731161117554,
+ "learning_rate": 4.477590838776386e-06,
+ "loss": 0.0161,
+ "step": 25530
+ },
+ {
+ "epoch": 69.566757493188,
+ "grad_norm": 2.2987747192382812,
+ "learning_rate": 4.4768551381162375e-06,
+ "loss": 0.0852,
+ "step": 25531
+ },
+ {
+ "epoch": 69.56948228882834,
+ "grad_norm": 1.8437305688858032,
+ "learning_rate": 4.476119480470163e-06,
+ "loss": 0.1271,
+ "step": 25532
+ },
+ {
+ "epoch": 69.57220708446866,
+ "grad_norm": 1.0996450185775757,
+ "learning_rate": 4.475383865843889e-06,
+ "loss": 0.0124,
+ "step": 25533
+ },
+ {
+ "epoch": 69.57493188010899,
+ "grad_norm": 1.6074496507644653,
+ "learning_rate": 4.474648294243151e-06,
+ "loss": 0.1877,
+ "step": 25534
+ },
+ {
+ "epoch": 69.57765667574932,
+ "grad_norm": 1.6837542057037354,
+ "learning_rate": 4.473912765673676e-06,
+ "loss": 0.0624,
+ "step": 25535
+ },
+ {
+ "epoch": 69.58038147138964,
+ "grad_norm": 1.5208696126937866,
+ "learning_rate": 4.473177280141185e-06,
+ "loss": 0.0167,
+ "step": 25536
+ },
+ {
+ "epoch": 69.58310626702998,
+ "grad_norm": 2.1597681045532227,
+ "learning_rate": 4.4724418376514155e-06,
+ "loss": 0.1035,
+ "step": 25537
+ },
+ {
+ "epoch": 69.5858310626703,
+ "grad_norm": 1.5613071918487549,
+ "learning_rate": 4.47170643821009e-06,
+ "loss": 0.0439,
+ "step": 25538
+ },
+ {
+ "epoch": 69.58855585831063,
+ "grad_norm": 1.6280372142791748,
+ "learning_rate": 4.470971081822938e-06,
+ "loss": 0.0154,
+ "step": 25539
+ },
+ {
+ "epoch": 69.59128065395096,
+ "grad_norm": 1.1326510906219482,
+ "learning_rate": 4.470235768495682e-06,
+ "loss": 0.0963,
+ "step": 25540
+ },
+ {
+ "epoch": 69.59400544959128,
+ "grad_norm": 1.5956705808639526,
+ "learning_rate": 4.469500498234055e-06,
+ "loss": 0.0169,
+ "step": 25541
+ },
+ {
+ "epoch": 69.59673024523161,
+ "grad_norm": 2.3011913299560547,
+ "learning_rate": 4.468765271043777e-06,
+ "loss": 0.0174,
+ "step": 25542
+ },
+ {
+ "epoch": 69.59945504087193,
+ "grad_norm": 1.5669221878051758,
+ "learning_rate": 4.4680300869305795e-06,
+ "loss": 0.0357,
+ "step": 25543
+ },
+ {
+ "epoch": 69.60217983651226,
+ "grad_norm": 2.2706525325775146,
+ "learning_rate": 4.467294945900186e-06,
+ "loss": 0.0215,
+ "step": 25544
+ },
+ {
+ "epoch": 69.60490463215258,
+ "grad_norm": 1.4188426733016968,
+ "learning_rate": 4.466559847958318e-06,
+ "loss": 0.095,
+ "step": 25545
+ },
+ {
+ "epoch": 69.60762942779292,
+ "grad_norm": 2.7064807415008545,
+ "learning_rate": 4.465824793110708e-06,
+ "loss": 0.175,
+ "step": 25546
+ },
+ {
+ "epoch": 69.61035422343325,
+ "grad_norm": 1.6200109720230103,
+ "learning_rate": 4.4650897813630755e-06,
+ "loss": 0.0236,
+ "step": 25547
+ },
+ {
+ "epoch": 69.61307901907357,
+ "grad_norm": 0.9511815309524536,
+ "learning_rate": 4.464354812721147e-06,
+ "loss": 0.0167,
+ "step": 25548
+ },
+ {
+ "epoch": 69.6158038147139,
+ "grad_norm": 5.7763142585754395,
+ "learning_rate": 4.46361988719064e-06,
+ "loss": 0.0295,
+ "step": 25549
+ },
+ {
+ "epoch": 69.61852861035422,
+ "grad_norm": 1.1994560956954956,
+ "learning_rate": 4.462885004777287e-06,
+ "loss": 0.1436,
+ "step": 25550
+ },
+ {
+ "epoch": 69.62125340599455,
+ "grad_norm": 1.412441611289978,
+ "learning_rate": 4.462150165486807e-06,
+ "loss": 0.0128,
+ "step": 25551
+ },
+ {
+ "epoch": 69.62397820163488,
+ "grad_norm": 2.7298076152801514,
+ "learning_rate": 4.461415369324924e-06,
+ "loss": 0.0798,
+ "step": 25552
+ },
+ {
+ "epoch": 69.6267029972752,
+ "grad_norm": 1.9438767433166504,
+ "learning_rate": 4.460680616297356e-06,
+ "loss": 0.069,
+ "step": 25553
+ },
+ {
+ "epoch": 69.62942779291554,
+ "grad_norm": 1.5074583292007446,
+ "learning_rate": 4.459945906409832e-06,
+ "loss": 0.0886,
+ "step": 25554
+ },
+ {
+ "epoch": 69.63215258855585,
+ "grad_norm": 1.8867658376693726,
+ "learning_rate": 4.459211239668073e-06,
+ "loss": 0.028,
+ "step": 25555
+ },
+ {
+ "epoch": 69.63487738419619,
+ "grad_norm": 1.1622612476348877,
+ "learning_rate": 4.458476616077797e-06,
+ "loss": 0.0788,
+ "step": 25556
+ },
+ {
+ "epoch": 69.6376021798365,
+ "grad_norm": 1.8398467302322388,
+ "learning_rate": 4.457742035644723e-06,
+ "loss": 0.0555,
+ "step": 25557
+ },
+ {
+ "epoch": 69.64032697547684,
+ "grad_norm": 3.102893114089966,
+ "learning_rate": 4.4570074983745806e-06,
+ "loss": 0.0922,
+ "step": 25558
+ },
+ {
+ "epoch": 69.64305177111717,
+ "grad_norm": 1.472262978553772,
+ "learning_rate": 4.456273004273084e-06,
+ "loss": 0.0231,
+ "step": 25559
+ },
+ {
+ "epoch": 69.64577656675749,
+ "grad_norm": 1.4196076393127441,
+ "learning_rate": 4.4555385533459515e-06,
+ "loss": 0.0821,
+ "step": 25560
+ },
+ {
+ "epoch": 69.64850136239782,
+ "grad_norm": 2.123286247253418,
+ "learning_rate": 4.454804145598907e-06,
+ "loss": 0.0727,
+ "step": 25561
+ },
+ {
+ "epoch": 69.65122615803814,
+ "grad_norm": 1.4126378297805786,
+ "learning_rate": 4.4540697810376724e-06,
+ "loss": 0.0502,
+ "step": 25562
+ },
+ {
+ "epoch": 69.65395095367847,
+ "grad_norm": 1.302329421043396,
+ "learning_rate": 4.453335459667964e-06,
+ "loss": 0.0137,
+ "step": 25563
+ },
+ {
+ "epoch": 69.65667574931881,
+ "grad_norm": 1.786242961883545,
+ "learning_rate": 4.4526011814955e-06,
+ "loss": 0.0249,
+ "step": 25564
+ },
+ {
+ "epoch": 69.65940054495913,
+ "grad_norm": 2.8718972206115723,
+ "learning_rate": 4.451866946525997e-06,
+ "loss": 0.022,
+ "step": 25565
+ },
+ {
+ "epoch": 69.66212534059946,
+ "grad_norm": 2.2466769218444824,
+ "learning_rate": 4.451132754765179e-06,
+ "loss": 0.0759,
+ "step": 25566
+ },
+ {
+ "epoch": 69.66485013623978,
+ "grad_norm": 2.007413148880005,
+ "learning_rate": 4.450398606218759e-06,
+ "loss": 0.074,
+ "step": 25567
+ },
+ {
+ "epoch": 69.66757493188011,
+ "grad_norm": 1.6706684827804565,
+ "learning_rate": 4.449664500892456e-06,
+ "loss": 0.0785,
+ "step": 25568
+ },
+ {
+ "epoch": 69.67029972752043,
+ "grad_norm": 2.107938766479492,
+ "learning_rate": 4.448930438791986e-06,
+ "loss": 0.0896,
+ "step": 25569
+ },
+ {
+ "epoch": 69.67302452316076,
+ "grad_norm": 0.8555177450180054,
+ "learning_rate": 4.448196419923068e-06,
+ "loss": 0.0139,
+ "step": 25570
+ },
+ {
+ "epoch": 69.6757493188011,
+ "grad_norm": 1.446227788925171,
+ "learning_rate": 4.447462444291421e-06,
+ "loss": 0.0205,
+ "step": 25571
+ },
+ {
+ "epoch": 69.67847411444141,
+ "grad_norm": 1.6693400144577026,
+ "learning_rate": 4.446728511902755e-06,
+ "loss": 0.0099,
+ "step": 25572
+ },
+ {
+ "epoch": 69.68119891008175,
+ "grad_norm": 3.1161277294158936,
+ "learning_rate": 4.445994622762786e-06,
+ "loss": 0.0265,
+ "step": 25573
+ },
+ {
+ "epoch": 69.68392370572207,
+ "grad_norm": 1.5052263736724854,
+ "learning_rate": 4.445260776877235e-06,
+ "loss": 0.0572,
+ "step": 25574
+ },
+ {
+ "epoch": 69.6866485013624,
+ "grad_norm": 1.3783434629440308,
+ "learning_rate": 4.444526974251815e-06,
+ "loss": 0.0517,
+ "step": 25575
+ },
+ {
+ "epoch": 69.68937329700273,
+ "grad_norm": 6.5454607009887695,
+ "learning_rate": 4.443793214892241e-06,
+ "loss": 0.1071,
+ "step": 25576
+ },
+ {
+ "epoch": 69.69209809264305,
+ "grad_norm": 1.302241325378418,
+ "learning_rate": 4.443059498804217e-06,
+ "loss": 0.0141,
+ "step": 25577
+ },
+ {
+ "epoch": 69.69482288828338,
+ "grad_norm": 1.24799644947052,
+ "learning_rate": 4.442325825993475e-06,
+ "loss": 0.0545,
+ "step": 25578
+ },
+ {
+ "epoch": 69.6975476839237,
+ "grad_norm": 2.942439317703247,
+ "learning_rate": 4.441592196465722e-06,
+ "loss": 0.0811,
+ "step": 25579
+ },
+ {
+ "epoch": 69.70027247956403,
+ "grad_norm": 1.1232168674468994,
+ "learning_rate": 4.440858610226668e-06,
+ "loss": 0.0138,
+ "step": 25580
+ },
+ {
+ "epoch": 69.70299727520435,
+ "grad_norm": 1.7882858514785767,
+ "learning_rate": 4.440125067282026e-06,
+ "loss": 0.0939,
+ "step": 25581
+ },
+ {
+ "epoch": 69.70572207084469,
+ "grad_norm": 3.276773691177368,
+ "learning_rate": 4.439391567637514e-06,
+ "loss": 0.0381,
+ "step": 25582
+ },
+ {
+ "epoch": 69.70844686648502,
+ "grad_norm": 1.2637181282043457,
+ "learning_rate": 4.438658111298842e-06,
+ "loss": 0.0164,
+ "step": 25583
+ },
+ {
+ "epoch": 69.71117166212534,
+ "grad_norm": 1.5981879234313965,
+ "learning_rate": 4.437924698271721e-06,
+ "loss": 0.0432,
+ "step": 25584
+ },
+ {
+ "epoch": 69.71389645776567,
+ "grad_norm": 1.671640157699585,
+ "learning_rate": 4.437191328561861e-06,
+ "loss": 0.0366,
+ "step": 25585
+ },
+ {
+ "epoch": 69.71662125340599,
+ "grad_norm": 2.3656840324401855,
+ "learning_rate": 4.436458002174978e-06,
+ "loss": 0.0432,
+ "step": 25586
+ },
+ {
+ "epoch": 69.71934604904632,
+ "grad_norm": 1.5600495338439941,
+ "learning_rate": 4.435724719116781e-06,
+ "loss": 0.0792,
+ "step": 25587
+ },
+ {
+ "epoch": 69.72207084468666,
+ "grad_norm": 2.9029476642608643,
+ "learning_rate": 4.434991479392983e-06,
+ "loss": 0.0237,
+ "step": 25588
+ },
+ {
+ "epoch": 69.72479564032697,
+ "grad_norm": 1.4314910173416138,
+ "learning_rate": 4.434258283009287e-06,
+ "loss": 0.0196,
+ "step": 25589
+ },
+ {
+ "epoch": 69.7275204359673,
+ "grad_norm": 1.4588127136230469,
+ "learning_rate": 4.433525129971412e-06,
+ "loss": 0.0105,
+ "step": 25590
+ },
+ {
+ "epoch": 69.73024523160763,
+ "grad_norm": 2.0740251541137695,
+ "learning_rate": 4.432792020285065e-06,
+ "loss": 0.036,
+ "step": 25591
+ },
+ {
+ "epoch": 69.73297002724796,
+ "grad_norm": 1.8558874130249023,
+ "learning_rate": 4.432058953955954e-06,
+ "loss": 0.0249,
+ "step": 25592
+ },
+ {
+ "epoch": 69.73569482288828,
+ "grad_norm": 2.1660261154174805,
+ "learning_rate": 4.4313259309897835e-06,
+ "loss": 0.071,
+ "step": 25593
+ },
+ {
+ "epoch": 69.73841961852861,
+ "grad_norm": 1.7608106136322021,
+ "learning_rate": 4.430592951392269e-06,
+ "loss": 0.0702,
+ "step": 25594
+ },
+ {
+ "epoch": 69.74114441416894,
+ "grad_norm": 1.1573394536972046,
+ "learning_rate": 4.42986001516912e-06,
+ "loss": 0.1084,
+ "step": 25595
+ },
+ {
+ "epoch": 69.74386920980926,
+ "grad_norm": 0.930066704750061,
+ "learning_rate": 4.429127122326044e-06,
+ "loss": 0.0114,
+ "step": 25596
+ },
+ {
+ "epoch": 69.7465940054496,
+ "grad_norm": 1.6020973920822144,
+ "learning_rate": 4.428394272868746e-06,
+ "loss": 0.0243,
+ "step": 25597
+ },
+ {
+ "epoch": 69.74931880108991,
+ "grad_norm": 1.168182373046875,
+ "learning_rate": 4.427661466802929e-06,
+ "loss": 0.0159,
+ "step": 25598
+ },
+ {
+ "epoch": 69.75204359673025,
+ "grad_norm": 1.6268821954727173,
+ "learning_rate": 4.4269287041343105e-06,
+ "loss": 0.0365,
+ "step": 25599
+ },
+ {
+ "epoch": 69.75476839237058,
+ "grad_norm": 1.556496500968933,
+ "learning_rate": 4.42619598486859e-06,
+ "loss": 0.0154,
+ "step": 25600
+ },
+ {
+ "epoch": 69.7574931880109,
+ "grad_norm": 1.2610888481140137,
+ "learning_rate": 4.425463309011473e-06,
+ "loss": 0.0698,
+ "step": 25601
+ },
+ {
+ "epoch": 69.76021798365123,
+ "grad_norm": 1.6439151763916016,
+ "learning_rate": 4.4247306765686726e-06,
+ "loss": 0.0684,
+ "step": 25602
+ },
+ {
+ "epoch": 69.76294277929155,
+ "grad_norm": 1.2785029411315918,
+ "learning_rate": 4.42399808754589e-06,
+ "loss": 0.0199,
+ "step": 25603
+ },
+ {
+ "epoch": 69.76566757493188,
+ "grad_norm": 1.9432932138442993,
+ "learning_rate": 4.42326554194883e-06,
+ "loss": 0.0811,
+ "step": 25604
+ },
+ {
+ "epoch": 69.7683923705722,
+ "grad_norm": 0.9073695540428162,
+ "learning_rate": 4.422533039783196e-06,
+ "loss": 0.0097,
+ "step": 25605
+ },
+ {
+ "epoch": 69.77111716621253,
+ "grad_norm": 2.0512077808380127,
+ "learning_rate": 4.421800581054698e-06,
+ "loss": 0.0503,
+ "step": 25606
+ },
+ {
+ "epoch": 69.77384196185287,
+ "grad_norm": 1.8149431943893433,
+ "learning_rate": 4.421068165769038e-06,
+ "loss": 0.0279,
+ "step": 25607
+ },
+ {
+ "epoch": 69.77656675749319,
+ "grad_norm": 0.893211841583252,
+ "learning_rate": 4.420335793931919e-06,
+ "loss": 0.0091,
+ "step": 25608
+ },
+ {
+ "epoch": 69.77929155313352,
+ "grad_norm": 1.377799391746521,
+ "learning_rate": 4.419603465549042e-06,
+ "loss": 0.0381,
+ "step": 25609
+ },
+ {
+ "epoch": 69.78201634877384,
+ "grad_norm": 2.0203685760498047,
+ "learning_rate": 4.418871180626118e-06,
+ "loss": 0.0338,
+ "step": 25610
+ },
+ {
+ "epoch": 69.78474114441417,
+ "grad_norm": 1.538214087486267,
+ "learning_rate": 4.41813893916884e-06,
+ "loss": 0.0184,
+ "step": 25611
+ },
+ {
+ "epoch": 69.7874659400545,
+ "grad_norm": 0.8715608716011047,
+ "learning_rate": 4.417406741182921e-06,
+ "loss": 0.0086,
+ "step": 25612
+ },
+ {
+ "epoch": 69.79019073569482,
+ "grad_norm": 1.6437482833862305,
+ "learning_rate": 4.416674586674059e-06,
+ "loss": 0.0336,
+ "step": 25613
+ },
+ {
+ "epoch": 69.79291553133515,
+ "grad_norm": 1.2273050546646118,
+ "learning_rate": 4.415942475647952e-06,
+ "loss": 0.0156,
+ "step": 25614
+ },
+ {
+ "epoch": 69.79564032697547,
+ "grad_norm": 1.2996455430984497,
+ "learning_rate": 4.415210408110308e-06,
+ "loss": 0.0174,
+ "step": 25615
+ },
+ {
+ "epoch": 69.7983651226158,
+ "grad_norm": 1.8862237930297852,
+ "learning_rate": 4.414478384066827e-06,
+ "loss": 0.0834,
+ "step": 25616
+ },
+ {
+ "epoch": 69.80108991825612,
+ "grad_norm": 1.9379146099090576,
+ "learning_rate": 4.413746403523208e-06,
+ "loss": 0.0193,
+ "step": 25617
+ },
+ {
+ "epoch": 69.80381471389646,
+ "grad_norm": 1.6510542631149292,
+ "learning_rate": 4.413014466485148e-06,
+ "loss": 0.1651,
+ "step": 25618
+ },
+ {
+ "epoch": 69.80653950953679,
+ "grad_norm": 1.5204145908355713,
+ "learning_rate": 4.412282572958356e-06,
+ "loss": 0.0148,
+ "step": 25619
+ },
+ {
+ "epoch": 69.80926430517711,
+ "grad_norm": 1.4033856391906738,
+ "learning_rate": 4.411550722948527e-06,
+ "loss": 0.1149,
+ "step": 25620
+ },
+ {
+ "epoch": 69.81198910081744,
+ "grad_norm": 1.6877580881118774,
+ "learning_rate": 4.410818916461362e-06,
+ "loss": 0.0503,
+ "step": 25621
+ },
+ {
+ "epoch": 69.81471389645776,
+ "grad_norm": 2.0462286472320557,
+ "learning_rate": 4.410087153502554e-06,
+ "loss": 0.0351,
+ "step": 25622
+ },
+ {
+ "epoch": 69.8174386920981,
+ "grad_norm": 2.0952036380767822,
+ "learning_rate": 4.409355434077812e-06,
+ "loss": 0.0711,
+ "step": 25623
+ },
+ {
+ "epoch": 69.82016348773843,
+ "grad_norm": 1.5113054513931274,
+ "learning_rate": 4.408623758192829e-06,
+ "loss": 0.0233,
+ "step": 25624
+ },
+ {
+ "epoch": 69.82288828337875,
+ "grad_norm": 1.6642965078353882,
+ "learning_rate": 4.4078921258533026e-06,
+ "loss": 0.0438,
+ "step": 25625
+ },
+ {
+ "epoch": 69.82561307901908,
+ "grad_norm": 1.5512555837631226,
+ "learning_rate": 4.407160537064934e-06,
+ "loss": 0.0188,
+ "step": 25626
+ },
+ {
+ "epoch": 69.8283378746594,
+ "grad_norm": 1.5072258710861206,
+ "learning_rate": 4.40642899183342e-06,
+ "loss": 0.1158,
+ "step": 25627
+ },
+ {
+ "epoch": 69.83106267029973,
+ "grad_norm": 1.634395718574524,
+ "learning_rate": 4.4056974901644565e-06,
+ "loss": 0.0262,
+ "step": 25628
+ },
+ {
+ "epoch": 69.83378746594005,
+ "grad_norm": 1.5792872905731201,
+ "learning_rate": 4.404966032063738e-06,
+ "loss": 0.0322,
+ "step": 25629
+ },
+ {
+ "epoch": 69.83651226158038,
+ "grad_norm": 1.637690782546997,
+ "learning_rate": 4.404234617536964e-06,
+ "loss": 0.047,
+ "step": 25630
+ },
+ {
+ "epoch": 69.83923705722071,
+ "grad_norm": 1.502867341041565,
+ "learning_rate": 4.403503246589834e-06,
+ "loss": 0.0403,
+ "step": 25631
+ },
+ {
+ "epoch": 69.84196185286103,
+ "grad_norm": 1.094830870628357,
+ "learning_rate": 4.402771919228041e-06,
+ "loss": 0.0177,
+ "step": 25632
+ },
+ {
+ "epoch": 69.84468664850137,
+ "grad_norm": 1.7265979051589966,
+ "learning_rate": 4.402040635457279e-06,
+ "loss": 0.0299,
+ "step": 25633
+ },
+ {
+ "epoch": 69.84741144414168,
+ "grad_norm": 1.7376505136489868,
+ "learning_rate": 4.401309395283241e-06,
+ "loss": 0.0488,
+ "step": 25634
+ },
+ {
+ "epoch": 69.85013623978202,
+ "grad_norm": 1.91915762424469,
+ "learning_rate": 4.400578198711629e-06,
+ "loss": 0.0608,
+ "step": 25635
+ },
+ {
+ "epoch": 69.85286103542235,
+ "grad_norm": 1.1518924236297607,
+ "learning_rate": 4.399847045748134e-06,
+ "loss": 0.0105,
+ "step": 25636
+ },
+ {
+ "epoch": 69.85558583106267,
+ "grad_norm": 1.936537742614746,
+ "learning_rate": 4.39911593639845e-06,
+ "loss": 0.021,
+ "step": 25637
+ },
+ {
+ "epoch": 69.858310626703,
+ "grad_norm": 2.1188018321990967,
+ "learning_rate": 4.398384870668267e-06,
+ "loss": 0.1665,
+ "step": 25638
+ },
+ {
+ "epoch": 69.86103542234332,
+ "grad_norm": 2.0203404426574707,
+ "learning_rate": 4.397653848563287e-06,
+ "loss": 0.035,
+ "step": 25639
+ },
+ {
+ "epoch": 69.86376021798365,
+ "grad_norm": 2.900775909423828,
+ "learning_rate": 4.396922870089197e-06,
+ "loss": 0.0254,
+ "step": 25640
+ },
+ {
+ "epoch": 69.86648501362397,
+ "grad_norm": 2.179499387741089,
+ "learning_rate": 4.396191935251694e-06,
+ "loss": 0.0986,
+ "step": 25641
+ },
+ {
+ "epoch": 69.8692098092643,
+ "grad_norm": 1.2998508214950562,
+ "learning_rate": 4.395461044056462e-06,
+ "loss": 0.024,
+ "step": 25642
+ },
+ {
+ "epoch": 69.87193460490464,
+ "grad_norm": 1.6339603662490845,
+ "learning_rate": 4.394730196509203e-06,
+ "loss": 0.0255,
+ "step": 25643
+ },
+ {
+ "epoch": 69.87465940054496,
+ "grad_norm": 2.456413984298706,
+ "learning_rate": 4.393999392615604e-06,
+ "loss": 0.0361,
+ "step": 25644
+ },
+ {
+ "epoch": 69.87738419618529,
+ "grad_norm": 1.9755444526672363,
+ "learning_rate": 4.393268632381354e-06,
+ "loss": 0.0129,
+ "step": 25645
+ },
+ {
+ "epoch": 69.88010899182561,
+ "grad_norm": 2.07155442237854,
+ "learning_rate": 4.392537915812148e-06,
+ "loss": 0.1128,
+ "step": 25646
+ },
+ {
+ "epoch": 69.88283378746594,
+ "grad_norm": 1.0632455348968506,
+ "learning_rate": 4.391807242913679e-06,
+ "loss": 0.0127,
+ "step": 25647
+ },
+ {
+ "epoch": 69.88555858310627,
+ "grad_norm": 1.8287676572799683,
+ "learning_rate": 4.391076613691636e-06,
+ "loss": 0.0203,
+ "step": 25648
+ },
+ {
+ "epoch": 69.88828337874659,
+ "grad_norm": 1.7842833995819092,
+ "learning_rate": 4.390346028151708e-06,
+ "loss": 0.0162,
+ "step": 25649
+ },
+ {
+ "epoch": 69.89100817438693,
+ "grad_norm": 1.196655035018921,
+ "learning_rate": 4.38961548629958e-06,
+ "loss": 0.0104,
+ "step": 25650
+ },
+ {
+ "epoch": 69.89373297002724,
+ "grad_norm": 2.932917833328247,
+ "learning_rate": 4.38888498814095e-06,
+ "loss": 0.082,
+ "step": 25651
+ },
+ {
+ "epoch": 69.89645776566758,
+ "grad_norm": 1.8006856441497803,
+ "learning_rate": 4.388154533681502e-06,
+ "loss": 0.0451,
+ "step": 25652
+ },
+ {
+ "epoch": 69.8991825613079,
+ "grad_norm": 1.5961247682571411,
+ "learning_rate": 4.387424122926927e-06,
+ "loss": 0.0447,
+ "step": 25653
+ },
+ {
+ "epoch": 69.90190735694823,
+ "grad_norm": 1.412878155708313,
+ "learning_rate": 4.386693755882909e-06,
+ "loss": 0.0153,
+ "step": 25654
+ },
+ {
+ "epoch": 69.90463215258856,
+ "grad_norm": 1.6714342832565308,
+ "learning_rate": 4.385963432555142e-06,
+ "loss": 0.0221,
+ "step": 25655
+ },
+ {
+ "epoch": 69.90735694822888,
+ "grad_norm": 1.5926790237426758,
+ "learning_rate": 4.385233152949312e-06,
+ "loss": 0.0374,
+ "step": 25656
+ },
+ {
+ "epoch": 69.91008174386921,
+ "grad_norm": 1.627118468284607,
+ "learning_rate": 4.3845029170711055e-06,
+ "loss": 0.0197,
+ "step": 25657
+ },
+ {
+ "epoch": 69.91280653950953,
+ "grad_norm": 1.1246057748794556,
+ "learning_rate": 4.383772724926205e-06,
+ "loss": 0.1495,
+ "step": 25658
+ },
+ {
+ "epoch": 69.91553133514986,
+ "grad_norm": 1.6724798679351807,
+ "learning_rate": 4.383042576520306e-06,
+ "loss": 0.0654,
+ "step": 25659
+ },
+ {
+ "epoch": 69.9182561307902,
+ "grad_norm": 1.2692406177520752,
+ "learning_rate": 4.38231247185909e-06,
+ "loss": 0.0639,
+ "step": 25660
+ },
+ {
+ "epoch": 69.92098092643052,
+ "grad_norm": 1.4762475490570068,
+ "learning_rate": 4.381582410948244e-06,
+ "loss": 0.0648,
+ "step": 25661
+ },
+ {
+ "epoch": 69.92370572207085,
+ "grad_norm": 0.9411290884017944,
+ "learning_rate": 4.380852393793451e-06,
+ "loss": 0.0108,
+ "step": 25662
+ },
+ {
+ "epoch": 69.92643051771117,
+ "grad_norm": 1.3685959577560425,
+ "learning_rate": 4.380122420400397e-06,
+ "loss": 0.0213,
+ "step": 25663
+ },
+ {
+ "epoch": 69.9291553133515,
+ "grad_norm": 2.211432695388794,
+ "learning_rate": 4.379392490774773e-06,
+ "loss": 0.1032,
+ "step": 25664
+ },
+ {
+ "epoch": 69.93188010899182,
+ "grad_norm": 1.642522931098938,
+ "learning_rate": 4.378662604922259e-06,
+ "loss": 0.0727,
+ "step": 25665
+ },
+ {
+ "epoch": 69.93460490463215,
+ "grad_norm": 1.787738561630249,
+ "learning_rate": 4.3779327628485365e-06,
+ "loss": 0.0371,
+ "step": 25666
+ },
+ {
+ "epoch": 69.93732970027249,
+ "grad_norm": 1.5134296417236328,
+ "learning_rate": 4.377202964559296e-06,
+ "loss": 0.0562,
+ "step": 25667
+ },
+ {
+ "epoch": 69.9400544959128,
+ "grad_norm": 0.9241958260536194,
+ "learning_rate": 4.376473210060219e-06,
+ "loss": 0.0113,
+ "step": 25668
+ },
+ {
+ "epoch": 69.94277929155314,
+ "grad_norm": 0.9650863409042358,
+ "learning_rate": 4.375743499356987e-06,
+ "loss": 0.0262,
+ "step": 25669
+ },
+ {
+ "epoch": 69.94550408719346,
+ "grad_norm": 1.3248642683029175,
+ "learning_rate": 4.375013832455279e-06,
+ "loss": 0.0578,
+ "step": 25670
+ },
+ {
+ "epoch": 69.94822888283379,
+ "grad_norm": 2.1187307834625244,
+ "learning_rate": 4.374284209360787e-06,
+ "loss": 0.023,
+ "step": 25671
+ },
+ {
+ "epoch": 69.95095367847412,
+ "grad_norm": 1.8474966287612915,
+ "learning_rate": 4.373554630079187e-06,
+ "loss": 0.0307,
+ "step": 25672
+ },
+ {
+ "epoch": 69.95367847411444,
+ "grad_norm": 1.6454722881317139,
+ "learning_rate": 4.3728250946161644e-06,
+ "loss": 0.0451,
+ "step": 25673
+ },
+ {
+ "epoch": 69.95640326975477,
+ "grad_norm": 1.1283448934555054,
+ "learning_rate": 4.372095602977394e-06,
+ "loss": 0.0166,
+ "step": 25674
+ },
+ {
+ "epoch": 69.95912806539509,
+ "grad_norm": 1.4734750986099243,
+ "learning_rate": 4.371366155168566e-06,
+ "loss": 0.0213,
+ "step": 25675
+ },
+ {
+ "epoch": 69.96185286103542,
+ "grad_norm": 1.1851685047149658,
+ "learning_rate": 4.370636751195357e-06,
+ "loss": 0.015,
+ "step": 25676
+ },
+ {
+ "epoch": 69.96457765667574,
+ "grad_norm": 1.8067235946655273,
+ "learning_rate": 4.369907391063448e-06,
+ "loss": 0.0547,
+ "step": 25677
+ },
+ {
+ "epoch": 69.96730245231608,
+ "grad_norm": 1.5915831327438354,
+ "learning_rate": 4.369178074778516e-06,
+ "loss": 0.0661,
+ "step": 25678
+ },
+ {
+ "epoch": 69.97002724795641,
+ "grad_norm": 1.7049072980880737,
+ "learning_rate": 4.368448802346247e-06,
+ "loss": 0.063,
+ "step": 25679
+ },
+ {
+ "epoch": 69.97275204359673,
+ "grad_norm": 1.5072258710861206,
+ "learning_rate": 4.367719573772313e-06,
+ "loss": 0.0361,
+ "step": 25680
+ },
+ {
+ "epoch": 69.97547683923706,
+ "grad_norm": 1.1121985912322998,
+ "learning_rate": 4.366990389062402e-06,
+ "loss": 0.0146,
+ "step": 25681
+ },
+ {
+ "epoch": 69.97820163487738,
+ "grad_norm": 1.819605827331543,
+ "learning_rate": 4.366261248222189e-06,
+ "loss": 0.0182,
+ "step": 25682
+ },
+ {
+ "epoch": 69.98092643051771,
+ "grad_norm": 1.473260760307312,
+ "learning_rate": 4.3655321512573475e-06,
+ "loss": 0.0356,
+ "step": 25683
+ },
+ {
+ "epoch": 69.98365122615803,
+ "grad_norm": 1.8162950277328491,
+ "learning_rate": 4.364803098173566e-06,
+ "loss": 0.0358,
+ "step": 25684
+ },
+ {
+ "epoch": 69.98637602179836,
+ "grad_norm": 1.0297048091888428,
+ "learning_rate": 4.364074088976514e-06,
+ "loss": 0.0097,
+ "step": 25685
+ },
+ {
+ "epoch": 69.9891008174387,
+ "grad_norm": 1.9639580249786377,
+ "learning_rate": 4.363345123671873e-06,
+ "loss": 0.0299,
+ "step": 25686
+ },
+ {
+ "epoch": 69.99182561307902,
+ "grad_norm": 1.2388126850128174,
+ "learning_rate": 4.362616202265315e-06,
+ "loss": 0.0174,
+ "step": 25687
+ },
+ {
+ "epoch": 69.99455040871935,
+ "grad_norm": 1.7585886716842651,
+ "learning_rate": 4.3618873247625235e-06,
+ "loss": 0.0288,
+ "step": 25688
+ },
+ {
+ "epoch": 69.99727520435967,
+ "grad_norm": 1.4259759187698364,
+ "learning_rate": 4.361158491169173e-06,
+ "loss": 0.0488,
+ "step": 25689
+ },
+ {
+ "epoch": 70.0,
+ "grad_norm": 1.3755066394805908,
+ "learning_rate": 4.360429701490935e-06,
+ "loss": 0.023,
+ "step": 25690
+ },
+ {
+ "epoch": 70.00272479564033,
+ "grad_norm": 1.3952059745788574,
+ "learning_rate": 4.359700955733492e-06,
+ "loss": 0.0174,
+ "step": 25691
+ },
+ {
+ "epoch": 70.00544959128065,
+ "grad_norm": 1.898809790611267,
+ "learning_rate": 4.358972253902517e-06,
+ "loss": 0.0503,
+ "step": 25692
+ },
+ {
+ "epoch": 70.00817438692098,
+ "grad_norm": 1.2340081930160522,
+ "learning_rate": 4.358243596003684e-06,
+ "loss": 0.0168,
+ "step": 25693
+ },
+ {
+ "epoch": 70.0108991825613,
+ "grad_norm": 1.6564013957977295,
+ "learning_rate": 4.357514982042664e-06,
+ "loss": 0.0357,
+ "step": 25694
+ },
+ {
+ "epoch": 70.01362397820164,
+ "grad_norm": 1.7956446409225464,
+ "learning_rate": 4.3567864120251406e-06,
+ "loss": 0.0585,
+ "step": 25695
+ },
+ {
+ "epoch": 70.01634877384195,
+ "grad_norm": 2.89603590965271,
+ "learning_rate": 4.356057885956782e-06,
+ "loss": 0.1192,
+ "step": 25696
+ },
+ {
+ "epoch": 70.01907356948229,
+ "grad_norm": 1.5098893642425537,
+ "learning_rate": 4.35532940384326e-06,
+ "loss": 0.0222,
+ "step": 25697
+ },
+ {
+ "epoch": 70.02179836512262,
+ "grad_norm": 1.449484944343567,
+ "learning_rate": 4.354600965690255e-06,
+ "loss": 0.0935,
+ "step": 25698
+ },
+ {
+ "epoch": 70.02452316076294,
+ "grad_norm": 1.4571237564086914,
+ "learning_rate": 4.353872571503431e-06,
+ "loss": 0.1254,
+ "step": 25699
+ },
+ {
+ "epoch": 70.02724795640327,
+ "grad_norm": 1.6277809143066406,
+ "learning_rate": 4.353144221288471e-06,
+ "loss": 0.0219,
+ "step": 25700
+ },
+ {
+ "epoch": 70.02997275204359,
+ "grad_norm": 2.5492472648620605,
+ "learning_rate": 4.352415915051042e-06,
+ "loss": 0.0687,
+ "step": 25701
+ },
+ {
+ "epoch": 70.03269754768392,
+ "grad_norm": 1.9705259799957275,
+ "learning_rate": 4.351687652796815e-06,
+ "loss": 0.0331,
+ "step": 25702
+ },
+ {
+ "epoch": 70.03542234332426,
+ "grad_norm": 1.1719311475753784,
+ "learning_rate": 4.3509594345314585e-06,
+ "loss": 0.0126,
+ "step": 25703
+ },
+ {
+ "epoch": 70.03814713896458,
+ "grad_norm": 1.0920994281768799,
+ "learning_rate": 4.350231260260653e-06,
+ "loss": 0.0116,
+ "step": 25704
+ },
+ {
+ "epoch": 70.04087193460491,
+ "grad_norm": 1.189394235610962,
+ "learning_rate": 4.349503129990065e-06,
+ "loss": 0.0198,
+ "step": 25705
+ },
+ {
+ "epoch": 70.04359673024523,
+ "grad_norm": 2.163149833679199,
+ "learning_rate": 4.348775043725364e-06,
+ "loss": 0.0213,
+ "step": 25706
+ },
+ {
+ "epoch": 70.04632152588556,
+ "grad_norm": 0.7555617690086365,
+ "learning_rate": 4.348047001472217e-06,
+ "loss": 0.0093,
+ "step": 25707
+ },
+ {
+ "epoch": 70.04904632152588,
+ "grad_norm": 2.423433542251587,
+ "learning_rate": 4.347319003236302e-06,
+ "loss": 0.0491,
+ "step": 25708
+ },
+ {
+ "epoch": 70.05177111716621,
+ "grad_norm": 1.4306249618530273,
+ "learning_rate": 4.346591049023286e-06,
+ "loss": 0.0453,
+ "step": 25709
+ },
+ {
+ "epoch": 70.05449591280654,
+ "grad_norm": 1.5476775169372559,
+ "learning_rate": 4.34586313883883e-06,
+ "loss": 0.0241,
+ "step": 25710
+ },
+ {
+ "epoch": 70.05722070844686,
+ "grad_norm": 0.8604248762130737,
+ "learning_rate": 4.345135272688617e-06,
+ "loss": 0.0091,
+ "step": 25711
+ },
+ {
+ "epoch": 70.0599455040872,
+ "grad_norm": 1.3568611145019531,
+ "learning_rate": 4.3444074505783075e-06,
+ "loss": 0.0199,
+ "step": 25712
+ },
+ {
+ "epoch": 70.06267029972751,
+ "grad_norm": 1.468930959701538,
+ "learning_rate": 4.34367967251357e-06,
+ "loss": 0.0272,
+ "step": 25713
+ },
+ {
+ "epoch": 70.06539509536785,
+ "grad_norm": 1.5441904067993164,
+ "learning_rate": 4.342951938500069e-06,
+ "loss": 0.0184,
+ "step": 25714
+ },
+ {
+ "epoch": 70.06811989100818,
+ "grad_norm": 1.6030669212341309,
+ "learning_rate": 4.342224248543479e-06,
+ "loss": 0.0233,
+ "step": 25715
+ },
+ {
+ "epoch": 70.0708446866485,
+ "grad_norm": 1.9875730276107788,
+ "learning_rate": 4.341496602649465e-06,
+ "loss": 0.0215,
+ "step": 25716
+ },
+ {
+ "epoch": 70.07356948228883,
+ "grad_norm": 1.0586199760437012,
+ "learning_rate": 4.340769000823695e-06,
+ "loss": 0.0147,
+ "step": 25717
+ },
+ {
+ "epoch": 70.07629427792915,
+ "grad_norm": 0.7717596292495728,
+ "learning_rate": 4.340041443071833e-06,
+ "loss": 0.0078,
+ "step": 25718
+ },
+ {
+ "epoch": 70.07901907356948,
+ "grad_norm": 1.9315496683120728,
+ "learning_rate": 4.339313929399543e-06,
+ "loss": 0.0509,
+ "step": 25719
+ },
+ {
+ "epoch": 70.0817438692098,
+ "grad_norm": 1.369428038597107,
+ "learning_rate": 4.338586459812498e-06,
+ "loss": 0.0214,
+ "step": 25720
+ },
+ {
+ "epoch": 70.08446866485014,
+ "grad_norm": 1.0970889329910278,
+ "learning_rate": 4.337859034316358e-06,
+ "loss": 0.0408,
+ "step": 25721
+ },
+ {
+ "epoch": 70.08719346049047,
+ "grad_norm": 2.3362975120544434,
+ "learning_rate": 4.337131652916792e-06,
+ "loss": 0.0499,
+ "step": 25722
+ },
+ {
+ "epoch": 70.08991825613079,
+ "grad_norm": 1.5447286367416382,
+ "learning_rate": 4.3364043156194566e-06,
+ "loss": 0.014,
+ "step": 25723
+ },
+ {
+ "epoch": 70.09264305177112,
+ "grad_norm": 1.9875041246414185,
+ "learning_rate": 4.335677022430027e-06,
+ "loss": 0.0326,
+ "step": 25724
+ },
+ {
+ "epoch": 70.09536784741144,
+ "grad_norm": 1.457663655281067,
+ "learning_rate": 4.334949773354161e-06,
+ "loss": 0.032,
+ "step": 25725
+ },
+ {
+ "epoch": 70.09809264305177,
+ "grad_norm": 1.3702753782272339,
+ "learning_rate": 4.334222568397526e-06,
+ "loss": 0.0143,
+ "step": 25726
+ },
+ {
+ "epoch": 70.1008174386921,
+ "grad_norm": 1.626405954360962,
+ "learning_rate": 4.333495407565777e-06,
+ "loss": 0.209,
+ "step": 25727
+ },
+ {
+ "epoch": 70.10354223433242,
+ "grad_norm": 1.4559513330459595,
+ "learning_rate": 4.33276829086459e-06,
+ "loss": 0.0364,
+ "step": 25728
+ },
+ {
+ "epoch": 70.10626702997276,
+ "grad_norm": 1.543286681175232,
+ "learning_rate": 4.332041218299619e-06,
+ "loss": 0.1077,
+ "step": 25729
+ },
+ {
+ "epoch": 70.10899182561307,
+ "grad_norm": 1.9175587892532349,
+ "learning_rate": 4.331314189876528e-06,
+ "loss": 0.0163,
+ "step": 25730
+ },
+ {
+ "epoch": 70.11171662125341,
+ "grad_norm": 1.2482143640518188,
+ "learning_rate": 4.330587205600978e-06,
+ "loss": 0.0183,
+ "step": 25731
+ },
+ {
+ "epoch": 70.11444141689373,
+ "grad_norm": 1.9215672016143799,
+ "learning_rate": 4.329860265478631e-06,
+ "loss": 0.0923,
+ "step": 25732
+ },
+ {
+ "epoch": 70.11716621253406,
+ "grad_norm": 1.348219871520996,
+ "learning_rate": 4.329133369515155e-06,
+ "loss": 0.0235,
+ "step": 25733
+ },
+ {
+ "epoch": 70.11989100817439,
+ "grad_norm": 1.9702473878860474,
+ "learning_rate": 4.328406517716205e-06,
+ "loss": 0.0875,
+ "step": 25734
+ },
+ {
+ "epoch": 70.12261580381471,
+ "grad_norm": 1.9873847961425781,
+ "learning_rate": 4.327679710087438e-06,
+ "loss": 0.0594,
+ "step": 25735
+ },
+ {
+ "epoch": 70.12534059945504,
+ "grad_norm": 1.996045708656311,
+ "learning_rate": 4.326952946634523e-06,
+ "loss": 0.0762,
+ "step": 25736
+ },
+ {
+ "epoch": 70.12806539509536,
+ "grad_norm": 1.189359426498413,
+ "learning_rate": 4.326226227363116e-06,
+ "loss": 0.0461,
+ "step": 25737
+ },
+ {
+ "epoch": 70.1307901907357,
+ "grad_norm": 1.0724072456359863,
+ "learning_rate": 4.325499552278877e-06,
+ "loss": 0.0142,
+ "step": 25738
+ },
+ {
+ "epoch": 70.13351498637603,
+ "grad_norm": 1.4749656915664673,
+ "learning_rate": 4.32477292138746e-06,
+ "loss": 0.0914,
+ "step": 25739
+ },
+ {
+ "epoch": 70.13623978201635,
+ "grad_norm": 1.4523439407348633,
+ "learning_rate": 4.324046334694534e-06,
+ "loss": 0.018,
+ "step": 25740
+ },
+ {
+ "epoch": 70.13896457765668,
+ "grad_norm": 2.2948861122131348,
+ "learning_rate": 4.3233197922057515e-06,
+ "loss": 0.0426,
+ "step": 25741
+ },
+ {
+ "epoch": 70.141689373297,
+ "grad_norm": 1.4242417812347412,
+ "learning_rate": 4.322593293926771e-06,
+ "loss": 0.0985,
+ "step": 25742
+ },
+ {
+ "epoch": 70.14441416893733,
+ "grad_norm": 0.953602135181427,
+ "learning_rate": 4.321866839863249e-06,
+ "loss": 0.0131,
+ "step": 25743
+ },
+ {
+ "epoch": 70.14713896457765,
+ "grad_norm": 1.5984370708465576,
+ "learning_rate": 4.321140430020848e-06,
+ "loss": 0.0227,
+ "step": 25744
+ },
+ {
+ "epoch": 70.14986376021798,
+ "grad_norm": 1.2445807456970215,
+ "learning_rate": 4.320414064405223e-06,
+ "loss": 0.0115,
+ "step": 25745
+ },
+ {
+ "epoch": 70.15258855585832,
+ "grad_norm": 1.8903340101242065,
+ "learning_rate": 4.319687743022031e-06,
+ "loss": 0.0151,
+ "step": 25746
+ },
+ {
+ "epoch": 70.15531335149863,
+ "grad_norm": 1.167959451675415,
+ "learning_rate": 4.318961465876924e-06,
+ "loss": 0.1378,
+ "step": 25747
+ },
+ {
+ "epoch": 70.15803814713897,
+ "grad_norm": 3.456315517425537,
+ "learning_rate": 4.318235232975565e-06,
+ "loss": 0.0169,
+ "step": 25748
+ },
+ {
+ "epoch": 70.16076294277929,
+ "grad_norm": 0.9842459559440613,
+ "learning_rate": 4.317509044323605e-06,
+ "loss": 0.0108,
+ "step": 25749
+ },
+ {
+ "epoch": 70.16348773841962,
+ "grad_norm": 1.8077656030654907,
+ "learning_rate": 4.316782899926704e-06,
+ "loss": 0.0222,
+ "step": 25750
+ },
+ {
+ "epoch": 70.16621253405995,
+ "grad_norm": 1.022696614265442,
+ "learning_rate": 4.316056799790512e-06,
+ "loss": 0.0121,
+ "step": 25751
+ },
+ {
+ "epoch": 70.16893732970027,
+ "grad_norm": 0.8282576203346252,
+ "learning_rate": 4.31533074392069e-06,
+ "loss": 0.0128,
+ "step": 25752
+ },
+ {
+ "epoch": 70.1716621253406,
+ "grad_norm": 2.077439308166504,
+ "learning_rate": 4.314604732322889e-06,
+ "loss": 0.0222,
+ "step": 25753
+ },
+ {
+ "epoch": 70.17438692098092,
+ "grad_norm": 2.0179648399353027,
+ "learning_rate": 4.313878765002763e-06,
+ "loss": 0.0282,
+ "step": 25754
+ },
+ {
+ "epoch": 70.17711171662125,
+ "grad_norm": 1.147133231163025,
+ "learning_rate": 4.313152841965964e-06,
+ "loss": 0.0132,
+ "step": 25755
+ },
+ {
+ "epoch": 70.17983651226157,
+ "grad_norm": 2.7429442405700684,
+ "learning_rate": 4.31242696321815e-06,
+ "loss": 0.0287,
+ "step": 25756
+ },
+ {
+ "epoch": 70.1825613079019,
+ "grad_norm": 1.6488178968429565,
+ "learning_rate": 4.311701128764974e-06,
+ "loss": 0.065,
+ "step": 25757
+ },
+ {
+ "epoch": 70.18528610354224,
+ "grad_norm": 1.3394337892532349,
+ "learning_rate": 4.310975338612084e-06,
+ "loss": 0.0479,
+ "step": 25758
+ },
+ {
+ "epoch": 70.18801089918256,
+ "grad_norm": 1.2525962591171265,
+ "learning_rate": 4.310249592765133e-06,
+ "loss": 0.0201,
+ "step": 25759
+ },
+ {
+ "epoch": 70.19073569482289,
+ "grad_norm": 1.267626166343689,
+ "learning_rate": 4.309523891229777e-06,
+ "loss": 0.0168,
+ "step": 25760
+ },
+ {
+ "epoch": 70.19346049046321,
+ "grad_norm": 0.8055083751678467,
+ "learning_rate": 4.3087982340116665e-06,
+ "loss": 0.0137,
+ "step": 25761
+ },
+ {
+ "epoch": 70.19618528610354,
+ "grad_norm": 1.7633090019226074,
+ "learning_rate": 4.308072621116453e-06,
+ "loss": 0.0431,
+ "step": 25762
+ },
+ {
+ "epoch": 70.19891008174388,
+ "grad_norm": 2.76570200920105,
+ "learning_rate": 4.307347052549781e-06,
+ "loss": 0.0494,
+ "step": 25763
+ },
+ {
+ "epoch": 70.2016348773842,
+ "grad_norm": 1.388796091079712,
+ "learning_rate": 4.306621528317312e-06,
+ "loss": 0.0121,
+ "step": 25764
+ },
+ {
+ "epoch": 70.20435967302453,
+ "grad_norm": 0.8132423162460327,
+ "learning_rate": 4.305896048424691e-06,
+ "loss": 0.01,
+ "step": 25765
+ },
+ {
+ "epoch": 70.20708446866485,
+ "grad_norm": 1.222830057144165,
+ "learning_rate": 4.305170612877565e-06,
+ "loss": 0.0126,
+ "step": 25766
+ },
+ {
+ "epoch": 70.20980926430518,
+ "grad_norm": 1.0668857097625732,
+ "learning_rate": 4.304445221681592e-06,
+ "loss": 0.0096,
+ "step": 25767
+ },
+ {
+ "epoch": 70.2125340599455,
+ "grad_norm": 1.3880501985549927,
+ "learning_rate": 4.3037198748424105e-06,
+ "loss": 0.0143,
+ "step": 25768
+ },
+ {
+ "epoch": 70.21525885558583,
+ "grad_norm": 1.3867751359939575,
+ "learning_rate": 4.30299457236568e-06,
+ "loss": 0.0321,
+ "step": 25769
+ },
+ {
+ "epoch": 70.21798365122616,
+ "grad_norm": 1.5501062870025635,
+ "learning_rate": 4.302269314257044e-06,
+ "loss": 0.0256,
+ "step": 25770
+ },
+ {
+ "epoch": 70.22070844686648,
+ "grad_norm": 1.0515704154968262,
+ "learning_rate": 4.301544100522151e-06,
+ "loss": 0.0174,
+ "step": 25771
+ },
+ {
+ "epoch": 70.22343324250681,
+ "grad_norm": 1.4691765308380127,
+ "learning_rate": 4.300818931166645e-06,
+ "loss": 0.0229,
+ "step": 25772
+ },
+ {
+ "epoch": 70.22615803814713,
+ "grad_norm": 1.2919331789016724,
+ "learning_rate": 4.3000938061961825e-06,
+ "loss": 0.0161,
+ "step": 25773
+ },
+ {
+ "epoch": 70.22888283378747,
+ "grad_norm": 1.7618796825408936,
+ "learning_rate": 4.299368725616405e-06,
+ "loss": 0.0308,
+ "step": 25774
+ },
+ {
+ "epoch": 70.2316076294278,
+ "grad_norm": 1.6042433977127075,
+ "learning_rate": 4.298643689432958e-06,
+ "loss": 0.0239,
+ "step": 25775
+ },
+ {
+ "epoch": 70.23433242506812,
+ "grad_norm": 2.1468868255615234,
+ "learning_rate": 4.297918697651493e-06,
+ "loss": 0.1246,
+ "step": 25776
+ },
+ {
+ "epoch": 70.23705722070845,
+ "grad_norm": 1.6174490451812744,
+ "learning_rate": 4.297193750277654e-06,
+ "loss": 0.0578,
+ "step": 25777
+ },
+ {
+ "epoch": 70.23978201634877,
+ "grad_norm": 1.9597121477127075,
+ "learning_rate": 4.296468847317086e-06,
+ "loss": 0.0359,
+ "step": 25778
+ },
+ {
+ "epoch": 70.2425068119891,
+ "grad_norm": 0.9702286124229431,
+ "learning_rate": 4.2957439887754325e-06,
+ "loss": 0.0167,
+ "step": 25779
+ },
+ {
+ "epoch": 70.24523160762942,
+ "grad_norm": 1.47027587890625,
+ "learning_rate": 4.295019174658344e-06,
+ "loss": 0.1216,
+ "step": 25780
+ },
+ {
+ "epoch": 70.24795640326975,
+ "grad_norm": 1.735202431678772,
+ "learning_rate": 4.294294404971464e-06,
+ "loss": 0.1084,
+ "step": 25781
+ },
+ {
+ "epoch": 70.25068119891009,
+ "grad_norm": 2.094320774078369,
+ "learning_rate": 4.293569679720434e-06,
+ "loss": 0.1177,
+ "step": 25782
+ },
+ {
+ "epoch": 70.2534059945504,
+ "grad_norm": 1.1733574867248535,
+ "learning_rate": 4.2928449989108964e-06,
+ "loss": 0.0292,
+ "step": 25783
+ },
+ {
+ "epoch": 70.25613079019074,
+ "grad_norm": 1.1175378561019897,
+ "learning_rate": 4.292120362548499e-06,
+ "loss": 0.0095,
+ "step": 25784
+ },
+ {
+ "epoch": 70.25885558583106,
+ "grad_norm": 1.2638603448867798,
+ "learning_rate": 4.291395770638888e-06,
+ "loss": 0.0131,
+ "step": 25785
+ },
+ {
+ "epoch": 70.26158038147139,
+ "grad_norm": 1.521732211112976,
+ "learning_rate": 4.290671223187701e-06,
+ "loss": 0.072,
+ "step": 25786
+ },
+ {
+ "epoch": 70.26430517711172,
+ "grad_norm": 1.43979012966156,
+ "learning_rate": 4.289946720200585e-06,
+ "loss": 0.0719,
+ "step": 25787
+ },
+ {
+ "epoch": 70.26702997275204,
+ "grad_norm": 1.6108235120773315,
+ "learning_rate": 4.289222261683176e-06,
+ "loss": 0.0213,
+ "step": 25788
+ },
+ {
+ "epoch": 70.26975476839237,
+ "grad_norm": 1.4305036067962646,
+ "learning_rate": 4.2884978476411235e-06,
+ "loss": 0.0272,
+ "step": 25789
+ },
+ {
+ "epoch": 70.2724795640327,
+ "grad_norm": 1.3842494487762451,
+ "learning_rate": 4.287773478080066e-06,
+ "loss": 0.0267,
+ "step": 25790
+ },
+ {
+ "epoch": 70.27520435967303,
+ "grad_norm": 1.2348272800445557,
+ "learning_rate": 4.287049153005644e-06,
+ "loss": 0.0921,
+ "step": 25791
+ },
+ {
+ "epoch": 70.27792915531334,
+ "grad_norm": 1.940852403640747,
+ "learning_rate": 4.286324872423497e-06,
+ "loss": 0.0704,
+ "step": 25792
+ },
+ {
+ "epoch": 70.28065395095368,
+ "grad_norm": 1.8134040832519531,
+ "learning_rate": 4.2856006363392726e-06,
+ "loss": 0.0902,
+ "step": 25793
+ },
+ {
+ "epoch": 70.28337874659401,
+ "grad_norm": 2.828098773956299,
+ "learning_rate": 4.284876444758604e-06,
+ "loss": 0.0314,
+ "step": 25794
+ },
+ {
+ "epoch": 70.28610354223433,
+ "grad_norm": 1.6803630590438843,
+ "learning_rate": 4.284152297687136e-06,
+ "loss": 0.0245,
+ "step": 25795
+ },
+ {
+ "epoch": 70.28882833787466,
+ "grad_norm": 1.1910679340362549,
+ "learning_rate": 4.2834281951305e-06,
+ "loss": 0.0249,
+ "step": 25796
+ },
+ {
+ "epoch": 70.29155313351498,
+ "grad_norm": 1.2956552505493164,
+ "learning_rate": 4.2827041370943465e-06,
+ "loss": 0.0132,
+ "step": 25797
+ },
+ {
+ "epoch": 70.29427792915531,
+ "grad_norm": 1.2774674892425537,
+ "learning_rate": 4.2819801235843086e-06,
+ "loss": 0.0209,
+ "step": 25798
+ },
+ {
+ "epoch": 70.29700272479565,
+ "grad_norm": 1.6430796384811401,
+ "learning_rate": 4.2812561546060215e-06,
+ "loss": 0.0539,
+ "step": 25799
+ },
+ {
+ "epoch": 70.29972752043597,
+ "grad_norm": 1.4388459920883179,
+ "learning_rate": 4.280532230165132e-06,
+ "loss": 0.0595,
+ "step": 25800
+ },
+ {
+ "epoch": 70.3024523160763,
+ "grad_norm": 1.9970476627349854,
+ "learning_rate": 4.2798083502672695e-06,
+ "loss": 0.0223,
+ "step": 25801
+ },
+ {
+ "epoch": 70.30517711171662,
+ "grad_norm": 1.2017982006072998,
+ "learning_rate": 4.2790845149180784e-06,
+ "loss": 0.0197,
+ "step": 25802
+ },
+ {
+ "epoch": 70.30790190735695,
+ "grad_norm": 1.732566475868225,
+ "learning_rate": 4.278360724123194e-06,
+ "loss": 0.0205,
+ "step": 25803
+ },
+ {
+ "epoch": 70.31062670299727,
+ "grad_norm": 0.9791429042816162,
+ "learning_rate": 4.277636977888248e-06,
+ "loss": 0.0088,
+ "step": 25804
+ },
+ {
+ "epoch": 70.3133514986376,
+ "grad_norm": 10.661685943603516,
+ "learning_rate": 4.276913276218885e-06,
+ "loss": 0.0433,
+ "step": 25805
+ },
+ {
+ "epoch": 70.31607629427793,
+ "grad_norm": 1.1543577909469604,
+ "learning_rate": 4.2761896191207365e-06,
+ "loss": 0.0136,
+ "step": 25806
+ },
+ {
+ "epoch": 70.31880108991825,
+ "grad_norm": 1.023016095161438,
+ "learning_rate": 4.275466006599439e-06,
+ "loss": 0.0109,
+ "step": 25807
+ },
+ {
+ "epoch": 70.32152588555859,
+ "grad_norm": 1.5195499658584595,
+ "learning_rate": 4.2747424386606255e-06,
+ "loss": 0.0162,
+ "step": 25808
+ },
+ {
+ "epoch": 70.3242506811989,
+ "grad_norm": 1.870869517326355,
+ "learning_rate": 4.2740189153099355e-06,
+ "loss": 0.0965,
+ "step": 25809
+ },
+ {
+ "epoch": 70.32697547683924,
+ "grad_norm": 1.5064311027526855,
+ "learning_rate": 4.2732954365530034e-06,
+ "loss": 0.0736,
+ "step": 25810
+ },
+ {
+ "epoch": 70.32970027247957,
+ "grad_norm": 0.9201691746711731,
+ "learning_rate": 4.2725720023954605e-06,
+ "loss": 0.0106,
+ "step": 25811
+ },
+ {
+ "epoch": 70.33242506811989,
+ "grad_norm": 1.5782822370529175,
+ "learning_rate": 4.27184861284294e-06,
+ "loss": 0.0498,
+ "step": 25812
+ },
+ {
+ "epoch": 70.33514986376022,
+ "grad_norm": 1.3706427812576294,
+ "learning_rate": 4.271125267901083e-06,
+ "loss": 0.0229,
+ "step": 25813
+ },
+ {
+ "epoch": 70.33787465940054,
+ "grad_norm": 1.2491129636764526,
+ "learning_rate": 4.2704019675755155e-06,
+ "loss": 0.0155,
+ "step": 25814
+ },
+ {
+ "epoch": 70.34059945504087,
+ "grad_norm": 1.6879082918167114,
+ "learning_rate": 4.269678711871875e-06,
+ "loss": 0.0311,
+ "step": 25815
+ },
+ {
+ "epoch": 70.34332425068119,
+ "grad_norm": 1.233951210975647,
+ "learning_rate": 4.268955500795788e-06,
+ "loss": 0.0182,
+ "step": 25816
+ },
+ {
+ "epoch": 70.34604904632153,
+ "grad_norm": 1.5089586973190308,
+ "learning_rate": 4.268232334352895e-06,
+ "loss": 0.0197,
+ "step": 25817
+ },
+ {
+ "epoch": 70.34877384196186,
+ "grad_norm": 1.17656672000885,
+ "learning_rate": 4.267509212548821e-06,
+ "loss": 0.0103,
+ "step": 25818
+ },
+ {
+ "epoch": 70.35149863760218,
+ "grad_norm": 1.1131908893585205,
+ "learning_rate": 4.266786135389203e-06,
+ "loss": 0.1263,
+ "step": 25819
+ },
+ {
+ "epoch": 70.35422343324251,
+ "grad_norm": 1.4315202236175537,
+ "learning_rate": 4.2660631028796665e-06,
+ "loss": 0.1117,
+ "step": 25820
+ },
+ {
+ "epoch": 70.35694822888283,
+ "grad_norm": 1.1368002891540527,
+ "learning_rate": 4.26534011502585e-06,
+ "loss": 0.018,
+ "step": 25821
+ },
+ {
+ "epoch": 70.35967302452316,
+ "grad_norm": 2.1280782222747803,
+ "learning_rate": 4.264617171833381e-06,
+ "loss": 0.1845,
+ "step": 25822
+ },
+ {
+ "epoch": 70.3623978201635,
+ "grad_norm": 1.693379282951355,
+ "learning_rate": 4.263894273307888e-06,
+ "loss": 0.0959,
+ "step": 25823
+ },
+ {
+ "epoch": 70.36512261580381,
+ "grad_norm": 2.0901458263397217,
+ "learning_rate": 4.263171419454998e-06,
+ "loss": 0.0185,
+ "step": 25824
+ },
+ {
+ "epoch": 70.36784741144415,
+ "grad_norm": 1.6026443243026733,
+ "learning_rate": 4.262448610280348e-06,
+ "loss": 0.0343,
+ "step": 25825
+ },
+ {
+ "epoch": 70.37057220708446,
+ "grad_norm": 1.8749750852584839,
+ "learning_rate": 4.261725845789564e-06,
+ "loss": 0.0143,
+ "step": 25826
+ },
+ {
+ "epoch": 70.3732970027248,
+ "grad_norm": 1.3103787899017334,
+ "learning_rate": 4.261003125988275e-06,
+ "loss": 0.0225,
+ "step": 25827
+ },
+ {
+ "epoch": 70.37602179836512,
+ "grad_norm": 1.8744372129440308,
+ "learning_rate": 4.260280450882104e-06,
+ "loss": 0.0282,
+ "step": 25828
+ },
+ {
+ "epoch": 70.37874659400545,
+ "grad_norm": 1.8052961826324463,
+ "learning_rate": 4.259557820476689e-06,
+ "loss": 0.022,
+ "step": 25829
+ },
+ {
+ "epoch": 70.38147138964578,
+ "grad_norm": 1.0968092679977417,
+ "learning_rate": 4.258835234777652e-06,
+ "loss": 0.0151,
+ "step": 25830
+ },
+ {
+ "epoch": 70.3841961852861,
+ "grad_norm": 1.2998303174972534,
+ "learning_rate": 4.2581126937906215e-06,
+ "loss": 0.0149,
+ "step": 25831
+ },
+ {
+ "epoch": 70.38692098092643,
+ "grad_norm": 2.079179048538208,
+ "learning_rate": 4.257390197521221e-06,
+ "loss": 0.0518,
+ "step": 25832
+ },
+ {
+ "epoch": 70.38964577656675,
+ "grad_norm": 1.8788728713989258,
+ "learning_rate": 4.256667745975084e-06,
+ "loss": 0.0727,
+ "step": 25833
+ },
+ {
+ "epoch": 70.39237057220708,
+ "grad_norm": 2.4956600666046143,
+ "learning_rate": 4.255945339157834e-06,
+ "loss": 0.0232,
+ "step": 25834
+ },
+ {
+ "epoch": 70.39509536784742,
+ "grad_norm": 2.342189073562622,
+ "learning_rate": 4.255222977075094e-06,
+ "loss": 0.0448,
+ "step": 25835
+ },
+ {
+ "epoch": 70.39782016348774,
+ "grad_norm": 1.05767822265625,
+ "learning_rate": 4.254500659732496e-06,
+ "loss": 0.0143,
+ "step": 25836
+ },
+ {
+ "epoch": 70.40054495912807,
+ "grad_norm": 1.9603577852249146,
+ "learning_rate": 4.253778387135658e-06,
+ "loss": 0.053,
+ "step": 25837
+ },
+ {
+ "epoch": 70.40326975476839,
+ "grad_norm": 1.1806777715682983,
+ "learning_rate": 4.253056159290212e-06,
+ "loss": 0.0311,
+ "step": 25838
+ },
+ {
+ "epoch": 70.40599455040872,
+ "grad_norm": 1.3933203220367432,
+ "learning_rate": 4.2523339762017805e-06,
+ "loss": 0.0177,
+ "step": 25839
+ },
+ {
+ "epoch": 70.40871934604904,
+ "grad_norm": 1.6884299516677856,
+ "learning_rate": 4.2516118378759826e-06,
+ "loss": 0.0591,
+ "step": 25840
+ },
+ {
+ "epoch": 70.41144414168937,
+ "grad_norm": 2.3381001949310303,
+ "learning_rate": 4.250889744318452e-06,
+ "loss": 0.0206,
+ "step": 25841
+ },
+ {
+ "epoch": 70.4141689373297,
+ "grad_norm": 2.14009690284729,
+ "learning_rate": 4.2501676955348045e-06,
+ "loss": 0.0307,
+ "step": 25842
+ },
+ {
+ "epoch": 70.41689373297002,
+ "grad_norm": 2.1921017169952393,
+ "learning_rate": 4.249445691530668e-06,
+ "loss": 0.0302,
+ "step": 25843
+ },
+ {
+ "epoch": 70.41961852861036,
+ "grad_norm": 1.4484177827835083,
+ "learning_rate": 4.24872373231166e-06,
+ "loss": 0.0154,
+ "step": 25844
+ },
+ {
+ "epoch": 70.42234332425068,
+ "grad_norm": 1.4473401308059692,
+ "learning_rate": 4.248001817883408e-06,
+ "loss": 0.0139,
+ "step": 25845
+ },
+ {
+ "epoch": 70.42506811989101,
+ "grad_norm": 1.8202407360076904,
+ "learning_rate": 4.247279948251534e-06,
+ "loss": 0.1247,
+ "step": 25846
+ },
+ {
+ "epoch": 70.42779291553134,
+ "grad_norm": 1.3334722518920898,
+ "learning_rate": 4.246558123421658e-06,
+ "loss": 0.0894,
+ "step": 25847
+ },
+ {
+ "epoch": 70.43051771117166,
+ "grad_norm": 2.8276753425598145,
+ "learning_rate": 4.2458363433994e-06,
+ "loss": 0.016,
+ "step": 25848
+ },
+ {
+ "epoch": 70.433242506812,
+ "grad_norm": 2.148005962371826,
+ "learning_rate": 4.245114608190385e-06,
+ "loss": 0.0452,
+ "step": 25849
+ },
+ {
+ "epoch": 70.43596730245231,
+ "grad_norm": 1.9681165218353271,
+ "learning_rate": 4.244392917800234e-06,
+ "loss": 0.0398,
+ "step": 25850
+ },
+ {
+ "epoch": 70.43869209809264,
+ "grad_norm": 0.6908787488937378,
+ "learning_rate": 4.243671272234564e-06,
+ "loss": 0.008,
+ "step": 25851
+ },
+ {
+ "epoch": 70.44141689373296,
+ "grad_norm": 1.4032905101776123,
+ "learning_rate": 4.242949671498995e-06,
+ "loss": 0.0208,
+ "step": 25852
+ },
+ {
+ "epoch": 70.4441416893733,
+ "grad_norm": 0.9831303954124451,
+ "learning_rate": 4.242228115599147e-06,
+ "loss": 0.0108,
+ "step": 25853
+ },
+ {
+ "epoch": 70.44686648501363,
+ "grad_norm": 1.2421963214874268,
+ "learning_rate": 4.2415066045406465e-06,
+ "loss": 0.0285,
+ "step": 25854
+ },
+ {
+ "epoch": 70.44959128065395,
+ "grad_norm": 1.9273358583450317,
+ "learning_rate": 4.2407851383291055e-06,
+ "loss": 0.0475,
+ "step": 25855
+ },
+ {
+ "epoch": 70.45231607629428,
+ "grad_norm": 2.1507279872894287,
+ "learning_rate": 4.240063716970145e-06,
+ "loss": 0.114,
+ "step": 25856
+ },
+ {
+ "epoch": 70.4550408719346,
+ "grad_norm": 1.5297099351882935,
+ "learning_rate": 4.23934234046938e-06,
+ "loss": 0.1023,
+ "step": 25857
+ },
+ {
+ "epoch": 70.45776566757493,
+ "grad_norm": 0.9264373779296875,
+ "learning_rate": 4.2386210088324355e-06,
+ "loss": 0.0122,
+ "step": 25858
+ },
+ {
+ "epoch": 70.46049046321527,
+ "grad_norm": 1.5053647756576538,
+ "learning_rate": 4.2378997220649235e-06,
+ "loss": 0.0163,
+ "step": 25859
+ },
+ {
+ "epoch": 70.46321525885558,
+ "grad_norm": 1.7384350299835205,
+ "learning_rate": 4.23717848017246e-06,
+ "loss": 0.0191,
+ "step": 25860
+ },
+ {
+ "epoch": 70.46594005449592,
+ "grad_norm": 1.5721604824066162,
+ "learning_rate": 4.236457283160669e-06,
+ "loss": 0.0175,
+ "step": 25861
+ },
+ {
+ "epoch": 70.46866485013624,
+ "grad_norm": 1.6441996097564697,
+ "learning_rate": 4.235736131035163e-06,
+ "loss": 0.0471,
+ "step": 25862
+ },
+ {
+ "epoch": 70.47138964577657,
+ "grad_norm": 1.176135778427124,
+ "learning_rate": 4.235015023801558e-06,
+ "loss": 0.011,
+ "step": 25863
+ },
+ {
+ "epoch": 70.47411444141689,
+ "grad_norm": 1.688717246055603,
+ "learning_rate": 4.234293961465467e-06,
+ "loss": 0.0252,
+ "step": 25864
+ },
+ {
+ "epoch": 70.47683923705722,
+ "grad_norm": 1.4468010663986206,
+ "learning_rate": 4.233572944032514e-06,
+ "loss": 0.0846,
+ "step": 25865
+ },
+ {
+ "epoch": 70.47956403269755,
+ "grad_norm": 1.7168304920196533,
+ "learning_rate": 4.232851971508308e-06,
+ "loss": 0.0258,
+ "step": 25866
+ },
+ {
+ "epoch": 70.48228882833787,
+ "grad_norm": 1.8824219703674316,
+ "learning_rate": 4.232131043898465e-06,
+ "loss": 0.0487,
+ "step": 25867
+ },
+ {
+ "epoch": 70.4850136239782,
+ "grad_norm": 1.2139534950256348,
+ "learning_rate": 4.231410161208598e-06,
+ "loss": 0.0128,
+ "step": 25868
+ },
+ {
+ "epoch": 70.48773841961852,
+ "grad_norm": 1.5943429470062256,
+ "learning_rate": 4.230689323444322e-06,
+ "loss": 0.1381,
+ "step": 25869
+ },
+ {
+ "epoch": 70.49046321525886,
+ "grad_norm": 1.6373732089996338,
+ "learning_rate": 4.229968530611255e-06,
+ "loss": 0.0756,
+ "step": 25870
+ },
+ {
+ "epoch": 70.49318801089919,
+ "grad_norm": 1.2066798210144043,
+ "learning_rate": 4.229247782715007e-06,
+ "loss": 0.0179,
+ "step": 25871
+ },
+ {
+ "epoch": 70.49591280653951,
+ "grad_norm": 0.6469790935516357,
+ "learning_rate": 4.228527079761192e-06,
+ "loss": 0.0075,
+ "step": 25872
+ },
+ {
+ "epoch": 70.49863760217984,
+ "grad_norm": 1.5673739910125732,
+ "learning_rate": 4.2278064217554194e-06,
+ "loss": 0.0385,
+ "step": 25873
+ },
+ {
+ "epoch": 70.50136239782016,
+ "grad_norm": 1.1692315340042114,
+ "learning_rate": 4.2270858087033075e-06,
+ "loss": 0.0216,
+ "step": 25874
+ },
+ {
+ "epoch": 70.50408719346049,
+ "grad_norm": 0.8341361880302429,
+ "learning_rate": 4.226365240610465e-06,
+ "loss": 0.0108,
+ "step": 25875
+ },
+ {
+ "epoch": 70.50681198910081,
+ "grad_norm": 10.462350845336914,
+ "learning_rate": 4.225644717482506e-06,
+ "loss": 0.1038,
+ "step": 25876
+ },
+ {
+ "epoch": 70.50953678474114,
+ "grad_norm": 1.2337071895599365,
+ "learning_rate": 4.224924239325035e-06,
+ "loss": 0.0161,
+ "step": 25877
+ },
+ {
+ "epoch": 70.51226158038148,
+ "grad_norm": 1.285123348236084,
+ "learning_rate": 4.224203806143672e-06,
+ "loss": 0.0181,
+ "step": 25878
+ },
+ {
+ "epoch": 70.5149863760218,
+ "grad_norm": 1.9561787843704224,
+ "learning_rate": 4.223483417944024e-06,
+ "loss": 0.044,
+ "step": 25879
+ },
+ {
+ "epoch": 70.51771117166213,
+ "grad_norm": 1.3440288305282593,
+ "learning_rate": 4.2227630747317e-06,
+ "loss": 0.025,
+ "step": 25880
+ },
+ {
+ "epoch": 70.52043596730245,
+ "grad_norm": 1.3206069469451904,
+ "learning_rate": 4.22204277651231e-06,
+ "loss": 0.0155,
+ "step": 25881
+ },
+ {
+ "epoch": 70.52316076294278,
+ "grad_norm": 1.4206441640853882,
+ "learning_rate": 4.221322523291467e-06,
+ "loss": 0.0263,
+ "step": 25882
+ },
+ {
+ "epoch": 70.52588555858311,
+ "grad_norm": 1.3508304357528687,
+ "learning_rate": 4.2206023150747785e-06,
+ "loss": 0.0951,
+ "step": 25883
+ },
+ {
+ "epoch": 70.52861035422343,
+ "grad_norm": 2.0592992305755615,
+ "learning_rate": 4.21988215186785e-06,
+ "loss": 0.031,
+ "step": 25884
+ },
+ {
+ "epoch": 70.53133514986376,
+ "grad_norm": 2.2499306201934814,
+ "learning_rate": 4.2191620336762965e-06,
+ "loss": 0.0646,
+ "step": 25885
+ },
+ {
+ "epoch": 70.53405994550408,
+ "grad_norm": 2.0871105194091797,
+ "learning_rate": 4.218441960505723e-06,
+ "loss": 0.0626,
+ "step": 25886
+ },
+ {
+ "epoch": 70.53678474114442,
+ "grad_norm": 1.2007070779800415,
+ "learning_rate": 4.217721932361734e-06,
+ "loss": 0.0131,
+ "step": 25887
+ },
+ {
+ "epoch": 70.53950953678473,
+ "grad_norm": 0.7885409593582153,
+ "learning_rate": 4.217001949249944e-06,
+ "loss": 0.0148,
+ "step": 25888
+ },
+ {
+ "epoch": 70.54223433242507,
+ "grad_norm": 1.1322120428085327,
+ "learning_rate": 4.216282011175953e-06,
+ "loss": 0.067,
+ "step": 25889
+ },
+ {
+ "epoch": 70.5449591280654,
+ "grad_norm": 1.330941915512085,
+ "learning_rate": 4.2155621181453745e-06,
+ "loss": 0.0388,
+ "step": 25890
+ },
+ {
+ "epoch": 70.54768392370572,
+ "grad_norm": 1.8707786798477173,
+ "learning_rate": 4.214842270163813e-06,
+ "loss": 0.105,
+ "step": 25891
+ },
+ {
+ "epoch": 70.55040871934605,
+ "grad_norm": 1.7131696939468384,
+ "learning_rate": 4.214122467236873e-06,
+ "loss": 0.0859,
+ "step": 25892
+ },
+ {
+ "epoch": 70.55313351498637,
+ "grad_norm": 1.2503266334533691,
+ "learning_rate": 4.213402709370158e-06,
+ "loss": 0.1286,
+ "step": 25893
+ },
+ {
+ "epoch": 70.5558583106267,
+ "grad_norm": 1.4554108381271362,
+ "learning_rate": 4.21268299656928e-06,
+ "loss": 0.076,
+ "step": 25894
+ },
+ {
+ "epoch": 70.55858310626704,
+ "grad_norm": 2.0748445987701416,
+ "learning_rate": 4.211963328839841e-06,
+ "loss": 0.0462,
+ "step": 25895
+ },
+ {
+ "epoch": 70.56130790190736,
+ "grad_norm": 1.5024100542068481,
+ "learning_rate": 4.2112437061874445e-06,
+ "loss": 0.0251,
+ "step": 25896
+ },
+ {
+ "epoch": 70.56403269754769,
+ "grad_norm": 1.6779636144638062,
+ "learning_rate": 4.210524128617692e-06,
+ "loss": 0.0218,
+ "step": 25897
+ },
+ {
+ "epoch": 70.566757493188,
+ "grad_norm": 2.492083787918091,
+ "learning_rate": 4.209804596136195e-06,
+ "loss": 0.0386,
+ "step": 25898
+ },
+ {
+ "epoch": 70.56948228882834,
+ "grad_norm": 2.0773744583129883,
+ "learning_rate": 4.209085108748554e-06,
+ "loss": 0.0358,
+ "step": 25899
+ },
+ {
+ "epoch": 70.57220708446866,
+ "grad_norm": 1.8618146181106567,
+ "learning_rate": 4.208365666460371e-06,
+ "loss": 0.0291,
+ "step": 25900
+ },
+ {
+ "epoch": 70.57493188010899,
+ "grad_norm": 1.305694818496704,
+ "learning_rate": 4.207646269277247e-06,
+ "loss": 0.0172,
+ "step": 25901
+ },
+ {
+ "epoch": 70.57765667574932,
+ "grad_norm": 1.5719980001449585,
+ "learning_rate": 4.206926917204791e-06,
+ "loss": 0.0151,
+ "step": 25902
+ },
+ {
+ "epoch": 70.58038147138964,
+ "grad_norm": 1.4947649240493774,
+ "learning_rate": 4.2062076102486016e-06,
+ "loss": 0.0714,
+ "step": 25903
+ },
+ {
+ "epoch": 70.58310626702998,
+ "grad_norm": 1.4840861558914185,
+ "learning_rate": 4.2054883484142765e-06,
+ "loss": 0.015,
+ "step": 25904
+ },
+ {
+ "epoch": 70.5858310626703,
+ "grad_norm": 1.144762396812439,
+ "learning_rate": 4.204769131707422e-06,
+ "loss": 0.0174,
+ "step": 25905
+ },
+ {
+ "epoch": 70.58855585831063,
+ "grad_norm": 1.481791615486145,
+ "learning_rate": 4.204049960133643e-06,
+ "loss": 0.0423,
+ "step": 25906
+ },
+ {
+ "epoch": 70.59128065395096,
+ "grad_norm": 1.9784799814224243,
+ "learning_rate": 4.203330833698536e-06,
+ "loss": 0.0433,
+ "step": 25907
+ },
+ {
+ "epoch": 70.59400544959128,
+ "grad_norm": 1.4544273614883423,
+ "learning_rate": 4.202611752407703e-06,
+ "loss": 0.0266,
+ "step": 25908
+ },
+ {
+ "epoch": 70.59673024523161,
+ "grad_norm": 1.3618336915969849,
+ "learning_rate": 4.201892716266738e-06,
+ "loss": 0.0225,
+ "step": 25909
+ },
+ {
+ "epoch": 70.59945504087193,
+ "grad_norm": 1.9177131652832031,
+ "learning_rate": 4.2011737252812505e-06,
+ "loss": 0.0817,
+ "step": 25910
+ },
+ {
+ "epoch": 70.60217983651226,
+ "grad_norm": 1.731751561164856,
+ "learning_rate": 4.200454779456834e-06,
+ "loss": 0.022,
+ "step": 25911
+ },
+ {
+ "epoch": 70.60490463215258,
+ "grad_norm": 1.7821431159973145,
+ "learning_rate": 4.199735878799091e-06,
+ "loss": 0.1344,
+ "step": 25912
+ },
+ {
+ "epoch": 70.60762942779292,
+ "grad_norm": 1.2081396579742432,
+ "learning_rate": 4.199017023313613e-06,
+ "loss": 0.0185,
+ "step": 25913
+ },
+ {
+ "epoch": 70.61035422343325,
+ "grad_norm": 1.6539403200149536,
+ "learning_rate": 4.198298213006008e-06,
+ "loss": 0.0212,
+ "step": 25914
+ },
+ {
+ "epoch": 70.61307901907357,
+ "grad_norm": 1.5517117977142334,
+ "learning_rate": 4.19757944788187e-06,
+ "loss": 0.0267,
+ "step": 25915
+ },
+ {
+ "epoch": 70.6158038147139,
+ "grad_norm": 1.1456053256988525,
+ "learning_rate": 4.196860727946796e-06,
+ "loss": 0.0104,
+ "step": 25916
+ },
+ {
+ "epoch": 70.61852861035422,
+ "grad_norm": 1.9889694452285767,
+ "learning_rate": 4.19614205320638e-06,
+ "loss": 0.018,
+ "step": 25917
+ },
+ {
+ "epoch": 70.62125340599455,
+ "grad_norm": 1.2970917224884033,
+ "learning_rate": 4.1954234236662275e-06,
+ "loss": 0.0135,
+ "step": 25918
+ },
+ {
+ "epoch": 70.62397820163488,
+ "grad_norm": 2.148793935775757,
+ "learning_rate": 4.19470483933193e-06,
+ "loss": 0.0592,
+ "step": 25919
+ },
+ {
+ "epoch": 70.6267029972752,
+ "grad_norm": 1.4716796875,
+ "learning_rate": 4.193986300209084e-06,
+ "loss": 0.0126,
+ "step": 25920
+ },
+ {
+ "epoch": 70.62942779291554,
+ "grad_norm": 4.774728775024414,
+ "learning_rate": 4.193267806303283e-06,
+ "loss": 0.0241,
+ "step": 25921
+ },
+ {
+ "epoch": 70.63215258855585,
+ "grad_norm": 1.2630934715270996,
+ "learning_rate": 4.192549357620125e-06,
+ "loss": 0.0198,
+ "step": 25922
+ },
+ {
+ "epoch": 70.63487738419619,
+ "grad_norm": 2.0335075855255127,
+ "learning_rate": 4.19183095416521e-06,
+ "loss": 0.0403,
+ "step": 25923
+ },
+ {
+ "epoch": 70.6376021798365,
+ "grad_norm": 1.4687467813491821,
+ "learning_rate": 4.191112595944128e-06,
+ "loss": 0.018,
+ "step": 25924
+ },
+ {
+ "epoch": 70.64032697547684,
+ "grad_norm": 1.8739699125289917,
+ "learning_rate": 4.19039428296247e-06,
+ "loss": 0.0213,
+ "step": 25925
+ },
+ {
+ "epoch": 70.64305177111717,
+ "grad_norm": 1.105077862739563,
+ "learning_rate": 4.189676015225838e-06,
+ "loss": 0.0464,
+ "step": 25926
+ },
+ {
+ "epoch": 70.64577656675749,
+ "grad_norm": 2.1864969730377197,
+ "learning_rate": 4.188957792739824e-06,
+ "loss": 0.0316,
+ "step": 25927
+ },
+ {
+ "epoch": 70.64850136239782,
+ "grad_norm": 1.119065761566162,
+ "learning_rate": 4.188239615510018e-06,
+ "loss": 0.0103,
+ "step": 25928
+ },
+ {
+ "epoch": 70.65122615803814,
+ "grad_norm": 0.8644948601722717,
+ "learning_rate": 4.1875214835420116e-06,
+ "loss": 0.007,
+ "step": 25929
+ },
+ {
+ "epoch": 70.65395095367847,
+ "grad_norm": 3.6352827548980713,
+ "learning_rate": 4.186803396841405e-06,
+ "loss": 0.0735,
+ "step": 25930
+ },
+ {
+ "epoch": 70.65667574931881,
+ "grad_norm": 1.1644340753555298,
+ "learning_rate": 4.186085355413785e-06,
+ "loss": 0.0219,
+ "step": 25931
+ },
+ {
+ "epoch": 70.65940054495913,
+ "grad_norm": 1.182641863822937,
+ "learning_rate": 4.185367359264746e-06,
+ "loss": 0.011,
+ "step": 25932
+ },
+ {
+ "epoch": 70.66212534059946,
+ "grad_norm": 0.9268163442611694,
+ "learning_rate": 4.184649408399876e-06,
+ "loss": 0.0207,
+ "step": 25933
+ },
+ {
+ "epoch": 70.66485013623978,
+ "grad_norm": 1.7079434394836426,
+ "learning_rate": 4.183931502824772e-06,
+ "loss": 0.0201,
+ "step": 25934
+ },
+ {
+ "epoch": 70.66757493188011,
+ "grad_norm": 2.0662341117858887,
+ "learning_rate": 4.183213642545022e-06,
+ "loss": 0.0309,
+ "step": 25935
+ },
+ {
+ "epoch": 70.67029972752043,
+ "grad_norm": 1.1983224153518677,
+ "learning_rate": 4.182495827566217e-06,
+ "loss": 0.0171,
+ "step": 25936
+ },
+ {
+ "epoch": 70.67302452316076,
+ "grad_norm": 1.7114752531051636,
+ "learning_rate": 4.181778057893944e-06,
+ "loss": 0.017,
+ "step": 25937
+ },
+ {
+ "epoch": 70.6757493188011,
+ "grad_norm": 2.011420726776123,
+ "learning_rate": 4.1810603335337964e-06,
+ "loss": 0.0556,
+ "step": 25938
+ },
+ {
+ "epoch": 70.67847411444141,
+ "grad_norm": 1.657208800315857,
+ "learning_rate": 4.180342654491368e-06,
+ "loss": 0.0398,
+ "step": 25939
+ },
+ {
+ "epoch": 70.68119891008175,
+ "grad_norm": 0.9429657459259033,
+ "learning_rate": 4.179625020772242e-06,
+ "loss": 0.0233,
+ "step": 25940
+ },
+ {
+ "epoch": 70.68392370572207,
+ "grad_norm": 1.2573583126068115,
+ "learning_rate": 4.17890743238201e-06,
+ "loss": 0.0545,
+ "step": 25941
+ },
+ {
+ "epoch": 70.6866485013624,
+ "grad_norm": 2.3626132011413574,
+ "learning_rate": 4.178189889326255e-06,
+ "loss": 0.0933,
+ "step": 25942
+ },
+ {
+ "epoch": 70.68937329700273,
+ "grad_norm": 1.7388980388641357,
+ "learning_rate": 4.177472391610574e-06,
+ "loss": 0.0675,
+ "step": 25943
+ },
+ {
+ "epoch": 70.69209809264305,
+ "grad_norm": 1.1440335512161255,
+ "learning_rate": 4.176754939240552e-06,
+ "loss": 0.0096,
+ "step": 25944
+ },
+ {
+ "epoch": 70.69482288828338,
+ "grad_norm": 1.4307785034179688,
+ "learning_rate": 4.176037532221772e-06,
+ "loss": 0.0288,
+ "step": 25945
+ },
+ {
+ "epoch": 70.6975476839237,
+ "grad_norm": 0.8578829765319824,
+ "learning_rate": 4.175320170559823e-06,
+ "loss": 0.0094,
+ "step": 25946
+ },
+ {
+ "epoch": 70.70027247956403,
+ "grad_norm": 1.0858768224716187,
+ "learning_rate": 4.174602854260297e-06,
+ "loss": 0.0229,
+ "step": 25947
+ },
+ {
+ "epoch": 70.70299727520435,
+ "grad_norm": 1.7392833232879639,
+ "learning_rate": 4.173885583328775e-06,
+ "loss": 0.0239,
+ "step": 25948
+ },
+ {
+ "epoch": 70.70572207084469,
+ "grad_norm": 1.5247116088867188,
+ "learning_rate": 4.173168357770842e-06,
+ "loss": 0.0452,
+ "step": 25949
+ },
+ {
+ "epoch": 70.70844686648502,
+ "grad_norm": 1.2080806493759155,
+ "learning_rate": 4.172451177592089e-06,
+ "loss": 0.029,
+ "step": 25950
+ },
+ {
+ "epoch": 70.71117166212534,
+ "grad_norm": 1.1856576204299927,
+ "learning_rate": 4.171734042798099e-06,
+ "loss": 0.0202,
+ "step": 25951
+ },
+ {
+ "epoch": 70.71389645776567,
+ "grad_norm": 0.9044938087463379,
+ "learning_rate": 4.171016953394457e-06,
+ "loss": 0.0076,
+ "step": 25952
+ },
+ {
+ "epoch": 70.71662125340599,
+ "grad_norm": 1.4394643306732178,
+ "learning_rate": 4.1702999093867435e-06,
+ "loss": 0.0169,
+ "step": 25953
+ },
+ {
+ "epoch": 70.71934604904632,
+ "grad_norm": 2.080658197402954,
+ "learning_rate": 4.169582910780551e-06,
+ "loss": 0.0312,
+ "step": 25954
+ },
+ {
+ "epoch": 70.72207084468666,
+ "grad_norm": 1.7403883934020996,
+ "learning_rate": 4.1688659575814535e-06,
+ "loss": 0.0179,
+ "step": 25955
+ },
+ {
+ "epoch": 70.72479564032697,
+ "grad_norm": 1.2604217529296875,
+ "learning_rate": 4.168149049795046e-06,
+ "loss": 0.0182,
+ "step": 25956
+ },
+ {
+ "epoch": 70.7275204359673,
+ "grad_norm": 1.5000890493392944,
+ "learning_rate": 4.1674321874269055e-06,
+ "loss": 0.0283,
+ "step": 25957
+ },
+ {
+ "epoch": 70.73024523160763,
+ "grad_norm": 1.1672396659851074,
+ "learning_rate": 4.166715370482611e-06,
+ "loss": 0.0183,
+ "step": 25958
+ },
+ {
+ "epoch": 70.73297002724796,
+ "grad_norm": 1.634969711303711,
+ "learning_rate": 4.165998598967754e-06,
+ "loss": 0.0188,
+ "step": 25959
+ },
+ {
+ "epoch": 70.73569482288828,
+ "grad_norm": 1.269171953201294,
+ "learning_rate": 4.165281872887912e-06,
+ "loss": 0.0256,
+ "step": 25960
+ },
+ {
+ "epoch": 70.73841961852861,
+ "grad_norm": 1.1936577558517456,
+ "learning_rate": 4.164565192248667e-06,
+ "loss": 0.014,
+ "step": 25961
+ },
+ {
+ "epoch": 70.74114441416894,
+ "grad_norm": 3.3899686336517334,
+ "learning_rate": 4.163848557055596e-06,
+ "loss": 0.0752,
+ "step": 25962
+ },
+ {
+ "epoch": 70.74386920980926,
+ "grad_norm": 0.7528261542320251,
+ "learning_rate": 4.163131967314289e-06,
+ "loss": 0.0097,
+ "step": 25963
+ },
+ {
+ "epoch": 70.7465940054496,
+ "grad_norm": 1.38047456741333,
+ "learning_rate": 4.162415423030321e-06,
+ "loss": 0.0152,
+ "step": 25964
+ },
+ {
+ "epoch": 70.74931880108991,
+ "grad_norm": 1.5354872941970825,
+ "learning_rate": 4.161698924209275e-06,
+ "loss": 0.0341,
+ "step": 25965
+ },
+ {
+ "epoch": 70.75204359673025,
+ "grad_norm": 1.7851566076278687,
+ "learning_rate": 4.160982470856727e-06,
+ "loss": 0.1128,
+ "step": 25966
+ },
+ {
+ "epoch": 70.75476839237058,
+ "grad_norm": 1.5791345834732056,
+ "learning_rate": 4.160266062978262e-06,
+ "loss": 0.1227,
+ "step": 25967
+ },
+ {
+ "epoch": 70.7574931880109,
+ "grad_norm": 0.8729092478752136,
+ "learning_rate": 4.159549700579459e-06,
+ "loss": 0.0074,
+ "step": 25968
+ },
+ {
+ "epoch": 70.76021798365123,
+ "grad_norm": 1.0549224615097046,
+ "learning_rate": 4.158833383665892e-06,
+ "loss": 0.0532,
+ "step": 25969
+ },
+ {
+ "epoch": 70.76294277929155,
+ "grad_norm": 2.2009689807891846,
+ "learning_rate": 4.158117112243141e-06,
+ "loss": 0.0379,
+ "step": 25970
+ },
+ {
+ "epoch": 70.76566757493188,
+ "grad_norm": 2.139723777770996,
+ "learning_rate": 4.157400886316788e-06,
+ "loss": 0.0807,
+ "step": 25971
+ },
+ {
+ "epoch": 70.7683923705722,
+ "grad_norm": 1.3650288581848145,
+ "learning_rate": 4.15668470589241e-06,
+ "loss": 0.0137,
+ "step": 25972
+ },
+ {
+ "epoch": 70.77111716621253,
+ "grad_norm": 2.1362812519073486,
+ "learning_rate": 4.15596857097558e-06,
+ "loss": 0.0236,
+ "step": 25973
+ },
+ {
+ "epoch": 70.77384196185287,
+ "grad_norm": 1.6628350019454956,
+ "learning_rate": 4.155252481571878e-06,
+ "loss": 0.0136,
+ "step": 25974
+ },
+ {
+ "epoch": 70.77656675749319,
+ "grad_norm": 1.5180473327636719,
+ "learning_rate": 4.154536437686886e-06,
+ "loss": 0.0146,
+ "step": 25975
+ },
+ {
+ "epoch": 70.77929155313352,
+ "grad_norm": 1.3217389583587646,
+ "learning_rate": 4.153820439326175e-06,
+ "loss": 0.0157,
+ "step": 25976
+ },
+ {
+ "epoch": 70.78201634877384,
+ "grad_norm": 1.7222386598587036,
+ "learning_rate": 4.153104486495323e-06,
+ "loss": 0.0201,
+ "step": 25977
+ },
+ {
+ "epoch": 70.78474114441417,
+ "grad_norm": 1.2863141298294067,
+ "learning_rate": 4.152388579199901e-06,
+ "loss": 0.0136,
+ "step": 25978
+ },
+ {
+ "epoch": 70.7874659400545,
+ "grad_norm": 2.059894323348999,
+ "learning_rate": 4.151672717445493e-06,
+ "loss": 0.0828,
+ "step": 25979
+ },
+ {
+ "epoch": 70.79019073569482,
+ "grad_norm": 1.7785395383834839,
+ "learning_rate": 4.150956901237668e-06,
+ "loss": 0.079,
+ "step": 25980
+ },
+ {
+ "epoch": 70.79291553133515,
+ "grad_norm": 1.4112170934677124,
+ "learning_rate": 4.1502411305820035e-06,
+ "loss": 0.0223,
+ "step": 25981
+ },
+ {
+ "epoch": 70.79564032697547,
+ "grad_norm": 1.9280885457992554,
+ "learning_rate": 4.1495254054840696e-06,
+ "loss": 0.1555,
+ "step": 25982
+ },
+ {
+ "epoch": 70.7983651226158,
+ "grad_norm": 2.2762997150421143,
+ "learning_rate": 4.1488097259494464e-06,
+ "loss": 0.02,
+ "step": 25983
+ },
+ {
+ "epoch": 70.80108991825612,
+ "grad_norm": 1.5982915163040161,
+ "learning_rate": 4.148094091983705e-06,
+ "loss": 0.1231,
+ "step": 25984
+ },
+ {
+ "epoch": 70.80381471389646,
+ "grad_norm": 5.652202129364014,
+ "learning_rate": 4.147378503592418e-06,
+ "loss": 0.1137,
+ "step": 25985
+ },
+ {
+ "epoch": 70.80653950953679,
+ "grad_norm": 1.6023211479187012,
+ "learning_rate": 4.146662960781156e-06,
+ "loss": 0.0299,
+ "step": 25986
+ },
+ {
+ "epoch": 70.80926430517711,
+ "grad_norm": 1.194678783416748,
+ "learning_rate": 4.145947463555499e-06,
+ "loss": 0.0131,
+ "step": 25987
+ },
+ {
+ "epoch": 70.81198910081744,
+ "grad_norm": 0.6412039399147034,
+ "learning_rate": 4.1452320119210136e-06,
+ "loss": 0.0057,
+ "step": 25988
+ },
+ {
+ "epoch": 70.81471389645776,
+ "grad_norm": 1.6849157810211182,
+ "learning_rate": 4.144516605883272e-06,
+ "loss": 0.0298,
+ "step": 25989
+ },
+ {
+ "epoch": 70.8174386920981,
+ "grad_norm": 1.3417772054672241,
+ "learning_rate": 4.14380124544784e-06,
+ "loss": 0.0715,
+ "step": 25990
+ },
+ {
+ "epoch": 70.82016348773843,
+ "grad_norm": 1.3372151851654053,
+ "learning_rate": 4.143085930620303e-06,
+ "loss": 0.0414,
+ "step": 25991
+ },
+ {
+ "epoch": 70.82288828337875,
+ "grad_norm": 1.6000103950500488,
+ "learning_rate": 4.142370661406224e-06,
+ "loss": 0.0195,
+ "step": 25992
+ },
+ {
+ "epoch": 70.82561307901908,
+ "grad_norm": 1.5152924060821533,
+ "learning_rate": 4.141655437811174e-06,
+ "loss": 0.0601,
+ "step": 25993
+ },
+ {
+ "epoch": 70.8283378746594,
+ "grad_norm": 1.7508121728897095,
+ "learning_rate": 4.140940259840719e-06,
+ "loss": 0.0703,
+ "step": 25994
+ },
+ {
+ "epoch": 70.83106267029973,
+ "grad_norm": 1.569810390472412,
+ "learning_rate": 4.140225127500437e-06,
+ "loss": 0.0202,
+ "step": 25995
+ },
+ {
+ "epoch": 70.83378746594005,
+ "grad_norm": 1.1117582321166992,
+ "learning_rate": 4.139510040795892e-06,
+ "loss": 0.0217,
+ "step": 25996
+ },
+ {
+ "epoch": 70.83651226158038,
+ "grad_norm": 1.0949742794036865,
+ "learning_rate": 4.138794999732655e-06,
+ "loss": 0.0112,
+ "step": 25997
+ },
+ {
+ "epoch": 70.83923705722071,
+ "grad_norm": 0.9541957378387451,
+ "learning_rate": 4.138080004316289e-06,
+ "loss": 0.0269,
+ "step": 25998
+ },
+ {
+ "epoch": 70.84196185286103,
+ "grad_norm": 1.6662964820861816,
+ "learning_rate": 4.137365054552372e-06,
+ "loss": 0.0534,
+ "step": 25999
+ },
+ {
+ "epoch": 70.84468664850137,
+ "grad_norm": 1.0651229619979858,
+ "learning_rate": 4.136650150446468e-06,
+ "loss": 0.0106,
+ "step": 26000
+ },
+ {
+ "epoch": 70.84741144414168,
+ "grad_norm": 1.2264267206192017,
+ "learning_rate": 4.135935292004143e-06,
+ "loss": 0.1406,
+ "step": 26001
+ },
+ {
+ "epoch": 70.85013623978202,
+ "grad_norm": 1.8295128345489502,
+ "learning_rate": 4.135220479230961e-06,
+ "loss": 0.0347,
+ "step": 26002
+ },
+ {
+ "epoch": 70.85286103542235,
+ "grad_norm": 0.9435645341873169,
+ "learning_rate": 4.134505712132498e-06,
+ "loss": 0.011,
+ "step": 26003
+ },
+ {
+ "epoch": 70.85558583106267,
+ "grad_norm": 1.0113232135772705,
+ "learning_rate": 4.133790990714316e-06,
+ "loss": 0.0144,
+ "step": 26004
+ },
+ {
+ "epoch": 70.858310626703,
+ "grad_norm": 3.2293732166290283,
+ "learning_rate": 4.133076314981979e-06,
+ "loss": 0.0492,
+ "step": 26005
+ },
+ {
+ "epoch": 70.86103542234332,
+ "grad_norm": 1.1887787580490112,
+ "learning_rate": 4.132361684941053e-06,
+ "loss": 0.0148,
+ "step": 26006
+ },
+ {
+ "epoch": 70.86376021798365,
+ "grad_norm": 1.167280912399292,
+ "learning_rate": 4.131647100597105e-06,
+ "loss": 0.013,
+ "step": 26007
+ },
+ {
+ "epoch": 70.86648501362397,
+ "grad_norm": 2.5727953910827637,
+ "learning_rate": 4.130932561955704e-06,
+ "loss": 0.109,
+ "step": 26008
+ },
+ {
+ "epoch": 70.8692098092643,
+ "grad_norm": 1.1368250846862793,
+ "learning_rate": 4.13021806902241e-06,
+ "loss": 0.0144,
+ "step": 26009
+ },
+ {
+ "epoch": 70.87193460490464,
+ "grad_norm": 1.4113008975982666,
+ "learning_rate": 4.12950362180279e-06,
+ "loss": 0.0314,
+ "step": 26010
+ },
+ {
+ "epoch": 70.87465940054496,
+ "grad_norm": 1.3948971033096313,
+ "learning_rate": 4.128789220302404e-06,
+ "loss": 0.0332,
+ "step": 26011
+ },
+ {
+ "epoch": 70.87738419618529,
+ "grad_norm": 16.161685943603516,
+ "learning_rate": 4.128074864526821e-06,
+ "loss": 0.0214,
+ "step": 26012
+ },
+ {
+ "epoch": 70.88010899182561,
+ "grad_norm": 1.6439940929412842,
+ "learning_rate": 4.127360554481602e-06,
+ "loss": 0.0539,
+ "step": 26013
+ },
+ {
+ "epoch": 70.88283378746594,
+ "grad_norm": 1.3660832643508911,
+ "learning_rate": 4.126646290172307e-06,
+ "loss": 0.0329,
+ "step": 26014
+ },
+ {
+ "epoch": 70.88555858310627,
+ "grad_norm": 1.2860567569732666,
+ "learning_rate": 4.125932071604505e-06,
+ "loss": 0.0286,
+ "step": 26015
+ },
+ {
+ "epoch": 70.88828337874659,
+ "grad_norm": 1.2568936347961426,
+ "learning_rate": 4.125217898783756e-06,
+ "loss": 0.0236,
+ "step": 26016
+ },
+ {
+ "epoch": 70.89100817438693,
+ "grad_norm": 1.2523064613342285,
+ "learning_rate": 4.124503771715619e-06,
+ "loss": 0.0103,
+ "step": 26017
+ },
+ {
+ "epoch": 70.89373297002724,
+ "grad_norm": 1.7754963636398315,
+ "learning_rate": 4.123789690405655e-06,
+ "loss": 0.0335,
+ "step": 26018
+ },
+ {
+ "epoch": 70.89645776566758,
+ "grad_norm": 1.7982138395309448,
+ "learning_rate": 4.1230756548594305e-06,
+ "loss": 0.0398,
+ "step": 26019
+ },
+ {
+ "epoch": 70.8991825613079,
+ "grad_norm": 1.6007778644561768,
+ "learning_rate": 4.122361665082505e-06,
+ "loss": 0.0137,
+ "step": 26020
+ },
+ {
+ "epoch": 70.90190735694823,
+ "grad_norm": 1.4220296144485474,
+ "learning_rate": 4.121647721080436e-06,
+ "loss": 0.01,
+ "step": 26021
+ },
+ {
+ "epoch": 70.90463215258856,
+ "grad_norm": 2.709214925765991,
+ "learning_rate": 4.120933822858784e-06,
+ "loss": 0.0582,
+ "step": 26022
+ },
+ {
+ "epoch": 70.90735694822888,
+ "grad_norm": 1.927711844444275,
+ "learning_rate": 4.120219970423112e-06,
+ "loss": 0.1371,
+ "step": 26023
+ },
+ {
+ "epoch": 70.91008174386921,
+ "grad_norm": 1.435118556022644,
+ "learning_rate": 4.119506163778974e-06,
+ "loss": 0.0241,
+ "step": 26024
+ },
+ {
+ "epoch": 70.91280653950953,
+ "grad_norm": 0.9965248107910156,
+ "learning_rate": 4.1187924029319374e-06,
+ "loss": 0.0108,
+ "step": 26025
+ },
+ {
+ "epoch": 70.91553133514986,
+ "grad_norm": 0.914718508720398,
+ "learning_rate": 4.118078687887556e-06,
+ "loss": 0.0137,
+ "step": 26026
+ },
+ {
+ "epoch": 70.9182561307902,
+ "grad_norm": 1.4320087432861328,
+ "learning_rate": 4.117365018651385e-06,
+ "loss": 0.0267,
+ "step": 26027
+ },
+ {
+ "epoch": 70.92098092643052,
+ "grad_norm": 1.0770695209503174,
+ "learning_rate": 4.1166513952289885e-06,
+ "loss": 0.0102,
+ "step": 26028
+ },
+ {
+ "epoch": 70.92370572207085,
+ "grad_norm": 1.8807873725891113,
+ "learning_rate": 4.115937817625923e-06,
+ "loss": 0.0337,
+ "step": 26029
+ },
+ {
+ "epoch": 70.92643051771117,
+ "grad_norm": 2.403297185897827,
+ "learning_rate": 4.1152242858477435e-06,
+ "loss": 0.1465,
+ "step": 26030
+ },
+ {
+ "epoch": 70.9291553133515,
+ "grad_norm": 1.2184500694274902,
+ "learning_rate": 4.114510799900003e-06,
+ "loss": 0.013,
+ "step": 26031
+ },
+ {
+ "epoch": 70.93188010899182,
+ "grad_norm": 1.6460332870483398,
+ "learning_rate": 4.113797359788269e-06,
+ "loss": 0.0244,
+ "step": 26032
+ },
+ {
+ "epoch": 70.93460490463215,
+ "grad_norm": 2.254734992980957,
+ "learning_rate": 4.11308396551809e-06,
+ "loss": 0.0103,
+ "step": 26033
+ },
+ {
+ "epoch": 70.93732970027249,
+ "grad_norm": 0.9869784116744995,
+ "learning_rate": 4.11237061709502e-06,
+ "loss": 0.0117,
+ "step": 26034
+ },
+ {
+ "epoch": 70.9400544959128,
+ "grad_norm": 1.8459827899932861,
+ "learning_rate": 4.111657314524623e-06,
+ "loss": 0.023,
+ "step": 26035
+ },
+ {
+ "epoch": 70.94277929155314,
+ "grad_norm": 1.1577929258346558,
+ "learning_rate": 4.110944057812449e-06,
+ "loss": 0.017,
+ "step": 26036
+ },
+ {
+ "epoch": 70.94550408719346,
+ "grad_norm": 1.8252640962600708,
+ "learning_rate": 4.110230846964053e-06,
+ "loss": 0.0373,
+ "step": 26037
+ },
+ {
+ "epoch": 70.94822888283379,
+ "grad_norm": 1.2719117403030396,
+ "learning_rate": 4.109517681984986e-06,
+ "loss": 0.019,
+ "step": 26038
+ },
+ {
+ "epoch": 70.95095367847412,
+ "grad_norm": 1.553989291191101,
+ "learning_rate": 4.108804562880809e-06,
+ "loss": 0.0136,
+ "step": 26039
+ },
+ {
+ "epoch": 70.95367847411444,
+ "grad_norm": 1.9324101209640503,
+ "learning_rate": 4.108091489657072e-06,
+ "loss": 0.0296,
+ "step": 26040
+ },
+ {
+ "epoch": 70.95640326975477,
+ "grad_norm": 2.002316474914551,
+ "learning_rate": 4.107378462319327e-06,
+ "loss": 0.0738,
+ "step": 26041
+ },
+ {
+ "epoch": 70.95912806539509,
+ "grad_norm": 1.35040283203125,
+ "learning_rate": 4.106665480873131e-06,
+ "loss": 0.0227,
+ "step": 26042
+ },
+ {
+ "epoch": 70.96185286103542,
+ "grad_norm": 2.0319857597351074,
+ "learning_rate": 4.105952545324031e-06,
+ "loss": 0.0366,
+ "step": 26043
+ },
+ {
+ "epoch": 70.96457765667574,
+ "grad_norm": 2.355994939804077,
+ "learning_rate": 4.105239655677586e-06,
+ "loss": 0.0404,
+ "step": 26044
+ },
+ {
+ "epoch": 70.96730245231608,
+ "grad_norm": 1.1498031616210938,
+ "learning_rate": 4.104526811939347e-06,
+ "loss": 0.0223,
+ "step": 26045
+ },
+ {
+ "epoch": 70.97002724795641,
+ "grad_norm": 1.4016085863113403,
+ "learning_rate": 4.103814014114862e-06,
+ "loss": 0.0165,
+ "step": 26046
+ },
+ {
+ "epoch": 70.97275204359673,
+ "grad_norm": 2.1511032581329346,
+ "learning_rate": 4.10310126220968e-06,
+ "loss": 0.1375,
+ "step": 26047
+ },
+ {
+ "epoch": 70.97547683923706,
+ "grad_norm": 1.6969364881515503,
+ "learning_rate": 4.102388556229358e-06,
+ "loss": 0.0489,
+ "step": 26048
+ },
+ {
+ "epoch": 70.97820163487738,
+ "grad_norm": 1.265753149986267,
+ "learning_rate": 4.101675896179446e-06,
+ "loss": 0.0117,
+ "step": 26049
+ },
+ {
+ "epoch": 70.98092643051771,
+ "grad_norm": 1.6410236358642578,
+ "learning_rate": 4.100963282065492e-06,
+ "loss": 0.1778,
+ "step": 26050
+ },
+ {
+ "epoch": 70.98365122615803,
+ "grad_norm": 1.1499834060668945,
+ "learning_rate": 4.100250713893043e-06,
+ "loss": 0.0086,
+ "step": 26051
+ },
+ {
+ "epoch": 70.98637602179836,
+ "grad_norm": 1.6855428218841553,
+ "learning_rate": 4.099538191667654e-06,
+ "loss": 0.0242,
+ "step": 26052
+ },
+ {
+ "epoch": 70.9891008174387,
+ "grad_norm": 1.1762759685516357,
+ "learning_rate": 4.098825715394872e-06,
+ "loss": 0.0106,
+ "step": 26053
+ },
+ {
+ "epoch": 70.99182561307902,
+ "grad_norm": 1.4614777565002441,
+ "learning_rate": 4.098113285080245e-06,
+ "loss": 0.0206,
+ "step": 26054
+ },
+ {
+ "epoch": 70.99455040871935,
+ "grad_norm": 1.483368992805481,
+ "learning_rate": 4.097400900729319e-06,
+ "loss": 0.17,
+ "step": 26055
+ },
+ {
+ "epoch": 70.99727520435967,
+ "grad_norm": 1.469200611114502,
+ "learning_rate": 4.0966885623476485e-06,
+ "loss": 0.0401,
+ "step": 26056
+ },
+ {
+ "epoch": 71.0,
+ "grad_norm": 1.602545976638794,
+ "learning_rate": 4.095976269940777e-06,
+ "loss": 0.0987,
+ "step": 26057
+ },
+ {
+ "epoch": 71.00272479564033,
+ "grad_norm": 2.0742201805114746,
+ "learning_rate": 4.095264023514248e-06,
+ "loss": 0.0388,
+ "step": 26058
+ },
+ {
+ "epoch": 71.00544959128065,
+ "grad_norm": 1.481127381324768,
+ "learning_rate": 4.094551823073614e-06,
+ "loss": 0.0591,
+ "step": 26059
+ },
+ {
+ "epoch": 71.00817438692098,
+ "grad_norm": 1.4008381366729736,
+ "learning_rate": 4.093839668624423e-06,
+ "loss": 0.0139,
+ "step": 26060
+ },
+ {
+ "epoch": 71.0108991825613,
+ "grad_norm": 1.0178264379501343,
+ "learning_rate": 4.0931275601722194e-06,
+ "loss": 0.0121,
+ "step": 26061
+ },
+ {
+ "epoch": 71.01362397820164,
+ "grad_norm": 1.3631744384765625,
+ "learning_rate": 4.092415497722547e-06,
+ "loss": 0.0153,
+ "step": 26062
+ },
+ {
+ "epoch": 71.01634877384195,
+ "grad_norm": 1.2212117910385132,
+ "learning_rate": 4.091703481280951e-06,
+ "loss": 0.0232,
+ "step": 26063
+ },
+ {
+ "epoch": 71.01907356948229,
+ "grad_norm": 1.7683310508728027,
+ "learning_rate": 4.090991510852979e-06,
+ "loss": 0.0472,
+ "step": 26064
+ },
+ {
+ "epoch": 71.02179836512262,
+ "grad_norm": 2.022319793701172,
+ "learning_rate": 4.090279586444177e-06,
+ "loss": 0.027,
+ "step": 26065
+ },
+ {
+ "epoch": 71.02452316076294,
+ "grad_norm": 1.4224048852920532,
+ "learning_rate": 4.089567708060087e-06,
+ "loss": 0.0242,
+ "step": 26066
+ },
+ {
+ "epoch": 71.02724795640327,
+ "grad_norm": 2.0244522094726562,
+ "learning_rate": 4.08885587570625e-06,
+ "loss": 0.0202,
+ "step": 26067
+ },
+ {
+ "epoch": 71.02997275204359,
+ "grad_norm": 2.953336477279663,
+ "learning_rate": 4.088144089388216e-06,
+ "loss": 0.0408,
+ "step": 26068
+ },
+ {
+ "epoch": 71.03269754768392,
+ "grad_norm": 1.1558104753494263,
+ "learning_rate": 4.087432349111527e-06,
+ "loss": 0.0114,
+ "step": 26069
+ },
+ {
+ "epoch": 71.03542234332426,
+ "grad_norm": 1.3287136554718018,
+ "learning_rate": 4.086720654881723e-06,
+ "loss": 0.0136,
+ "step": 26070
+ },
+ {
+ "epoch": 71.03814713896458,
+ "grad_norm": 1.9921374320983887,
+ "learning_rate": 4.086009006704345e-06,
+ "loss": 0.0142,
+ "step": 26071
+ },
+ {
+ "epoch": 71.04087193460491,
+ "grad_norm": 0.9931190013885498,
+ "learning_rate": 4.085297404584942e-06,
+ "loss": 0.0092,
+ "step": 26072
+ },
+ {
+ "epoch": 71.04359673024523,
+ "grad_norm": 1.0979264974594116,
+ "learning_rate": 4.0845858485290535e-06,
+ "loss": 0.0399,
+ "step": 26073
+ },
+ {
+ "epoch": 71.04632152588556,
+ "grad_norm": 1.4461572170257568,
+ "learning_rate": 4.0838743385422195e-06,
+ "loss": 0.0157,
+ "step": 26074
+ },
+ {
+ "epoch": 71.04904632152588,
+ "grad_norm": 1.0386254787445068,
+ "learning_rate": 4.083162874629978e-06,
+ "loss": 0.0111,
+ "step": 26075
+ },
+ {
+ "epoch": 71.05177111716621,
+ "grad_norm": 1.6738852262496948,
+ "learning_rate": 4.082451456797873e-06,
+ "loss": 0.0142,
+ "step": 26076
+ },
+ {
+ "epoch": 71.05449591280654,
+ "grad_norm": 2.95519757270813,
+ "learning_rate": 4.08174008505145e-06,
+ "loss": 0.1163,
+ "step": 26077
+ },
+ {
+ "epoch": 71.05722070844686,
+ "grad_norm": 1.148704171180725,
+ "learning_rate": 4.081028759396245e-06,
+ "loss": 0.0181,
+ "step": 26078
+ },
+ {
+ "epoch": 71.0599455040872,
+ "grad_norm": 0.8258702158927917,
+ "learning_rate": 4.080317479837793e-06,
+ "loss": 0.0144,
+ "step": 26079
+ },
+ {
+ "epoch": 71.06267029972751,
+ "grad_norm": 1.9168049097061157,
+ "learning_rate": 4.079606246381642e-06,
+ "loss": 0.027,
+ "step": 26080
+ },
+ {
+ "epoch": 71.06539509536785,
+ "grad_norm": 1.6525815725326538,
+ "learning_rate": 4.0788950590333265e-06,
+ "loss": 0.0144,
+ "step": 26081
+ },
+ {
+ "epoch": 71.06811989100818,
+ "grad_norm": 2.1681315898895264,
+ "learning_rate": 4.078183917798386e-06,
+ "loss": 0.0236,
+ "step": 26082
+ },
+ {
+ "epoch": 71.0708446866485,
+ "grad_norm": 1.2791929244995117,
+ "learning_rate": 4.0774728226823556e-06,
+ "loss": 0.0132,
+ "step": 26083
+ },
+ {
+ "epoch": 71.07356948228883,
+ "grad_norm": 1.43775475025177,
+ "learning_rate": 4.07676177369078e-06,
+ "loss": 0.0584,
+ "step": 26084
+ },
+ {
+ "epoch": 71.07629427792915,
+ "grad_norm": 1.6189467906951904,
+ "learning_rate": 4.076050770829192e-06,
+ "loss": 0.0207,
+ "step": 26085
+ },
+ {
+ "epoch": 71.07901907356948,
+ "grad_norm": 1.1358128786087036,
+ "learning_rate": 4.075339814103131e-06,
+ "loss": 0.023,
+ "step": 26086
+ },
+ {
+ "epoch": 71.0817438692098,
+ "grad_norm": 1.5362638235092163,
+ "learning_rate": 4.074628903518129e-06,
+ "loss": 0.0148,
+ "step": 26087
+ },
+ {
+ "epoch": 71.08446866485014,
+ "grad_norm": 1.2832839488983154,
+ "learning_rate": 4.073918039079729e-06,
+ "loss": 0.0539,
+ "step": 26088
+ },
+ {
+ "epoch": 71.08719346049047,
+ "grad_norm": 1.1979966163635254,
+ "learning_rate": 4.073207220793465e-06,
+ "loss": 0.0201,
+ "step": 26089
+ },
+ {
+ "epoch": 71.08991825613079,
+ "grad_norm": 1.6395787000656128,
+ "learning_rate": 4.072496448664872e-06,
+ "loss": 0.0695,
+ "step": 26090
+ },
+ {
+ "epoch": 71.09264305177112,
+ "grad_norm": 1.2449768781661987,
+ "learning_rate": 4.071785722699484e-06,
+ "loss": 0.0187,
+ "step": 26091
+ },
+ {
+ "epoch": 71.09536784741144,
+ "grad_norm": 1.608792781829834,
+ "learning_rate": 4.07107504290284e-06,
+ "loss": 0.0215,
+ "step": 26092
+ },
+ {
+ "epoch": 71.09809264305177,
+ "grad_norm": 1.797042727470398,
+ "learning_rate": 4.0703644092804695e-06,
+ "loss": 0.1444,
+ "step": 26093
+ },
+ {
+ "epoch": 71.1008174386921,
+ "grad_norm": 1.9141243696212769,
+ "learning_rate": 4.069653821837913e-06,
+ "loss": 0.0259,
+ "step": 26094
+ },
+ {
+ "epoch": 71.10354223433242,
+ "grad_norm": 2.2559986114501953,
+ "learning_rate": 4.068943280580704e-06,
+ "loss": 0.06,
+ "step": 26095
+ },
+ {
+ "epoch": 71.10626702997276,
+ "grad_norm": 1.6298030614852905,
+ "learning_rate": 4.068232785514368e-06,
+ "loss": 0.0253,
+ "step": 26096
+ },
+ {
+ "epoch": 71.10899182561307,
+ "grad_norm": 1.59772527217865,
+ "learning_rate": 4.067522336644448e-06,
+ "loss": 0.0152,
+ "step": 26097
+ },
+ {
+ "epoch": 71.11171662125341,
+ "grad_norm": 1.8733516931533813,
+ "learning_rate": 4.0668119339764746e-06,
+ "loss": 0.0317,
+ "step": 26098
+ },
+ {
+ "epoch": 71.11444141689373,
+ "grad_norm": 0.9162262678146362,
+ "learning_rate": 4.066101577515974e-06,
+ "loss": 0.0075,
+ "step": 26099
+ },
+ {
+ "epoch": 71.11716621253406,
+ "grad_norm": 1.4007841348648071,
+ "learning_rate": 4.065391267268487e-06,
+ "loss": 0.0219,
+ "step": 26100
+ },
+ {
+ "epoch": 71.11989100817439,
+ "grad_norm": 1.6843153238296509,
+ "learning_rate": 4.064681003239543e-06,
+ "loss": 0.0639,
+ "step": 26101
+ },
+ {
+ "epoch": 71.12261580381471,
+ "grad_norm": 1.748854160308838,
+ "learning_rate": 4.063970785434671e-06,
+ "loss": 0.096,
+ "step": 26102
+ },
+ {
+ "epoch": 71.12534059945504,
+ "grad_norm": 2.145225763320923,
+ "learning_rate": 4.0632606138594e-06,
+ "loss": 0.0219,
+ "step": 26103
+ },
+ {
+ "epoch": 71.12806539509536,
+ "grad_norm": 0.8550347685813904,
+ "learning_rate": 4.062550488519268e-06,
+ "loss": 0.0101,
+ "step": 26104
+ },
+ {
+ "epoch": 71.1307901907357,
+ "grad_norm": 1.5420914888381958,
+ "learning_rate": 4.0618404094198016e-06,
+ "loss": 0.024,
+ "step": 26105
+ },
+ {
+ "epoch": 71.13351498637603,
+ "grad_norm": 0.8454663157463074,
+ "learning_rate": 4.061130376566531e-06,
+ "loss": 0.0068,
+ "step": 26106
+ },
+ {
+ "epoch": 71.13623978201635,
+ "grad_norm": 1.7283613681793213,
+ "learning_rate": 4.060420389964983e-06,
+ "loss": 0.0205,
+ "step": 26107
+ },
+ {
+ "epoch": 71.13896457765668,
+ "grad_norm": 0.9620832204818726,
+ "learning_rate": 4.059710449620693e-06,
+ "loss": 0.1542,
+ "step": 26108
+ },
+ {
+ "epoch": 71.141689373297,
+ "grad_norm": 1.7242462635040283,
+ "learning_rate": 4.0590005555391865e-06,
+ "loss": 0.0363,
+ "step": 26109
+ },
+ {
+ "epoch": 71.14441416893733,
+ "grad_norm": 1.8117144107818604,
+ "learning_rate": 4.0582907077259894e-06,
+ "loss": 0.0982,
+ "step": 26110
+ },
+ {
+ "epoch": 71.14713896457765,
+ "grad_norm": 2.0661096572875977,
+ "learning_rate": 4.057580906186637e-06,
+ "loss": 0.0278,
+ "step": 26111
+ },
+ {
+ "epoch": 71.14986376021798,
+ "grad_norm": 2.2118947505950928,
+ "learning_rate": 4.056871150926649e-06,
+ "loss": 0.0373,
+ "step": 26112
+ },
+ {
+ "epoch": 71.15258855585832,
+ "grad_norm": 2.461894989013672,
+ "learning_rate": 4.056161441951561e-06,
+ "loss": 0.0446,
+ "step": 26113
+ },
+ {
+ "epoch": 71.15531335149863,
+ "grad_norm": 2.899510383605957,
+ "learning_rate": 4.055451779266896e-06,
+ "loss": 0.0637,
+ "step": 26114
+ },
+ {
+ "epoch": 71.15803814713897,
+ "grad_norm": 0.9075875878334045,
+ "learning_rate": 4.0547421628781815e-06,
+ "loss": 0.0107,
+ "step": 26115
+ },
+ {
+ "epoch": 71.16076294277929,
+ "grad_norm": 1.3415950536727905,
+ "learning_rate": 4.05403259279094e-06,
+ "loss": 0.0539,
+ "step": 26116
+ },
+ {
+ "epoch": 71.16348773841962,
+ "grad_norm": 1.6435667276382446,
+ "learning_rate": 4.053323069010704e-06,
+ "loss": 0.0194,
+ "step": 26117
+ },
+ {
+ "epoch": 71.16621253405995,
+ "grad_norm": 0.7970269918441772,
+ "learning_rate": 4.052613591542998e-06,
+ "loss": 0.0082,
+ "step": 26118
+ },
+ {
+ "epoch": 71.16893732970027,
+ "grad_norm": 2.241084337234497,
+ "learning_rate": 4.051904160393345e-06,
+ "loss": 0.0573,
+ "step": 26119
+ },
+ {
+ "epoch": 71.1716621253406,
+ "grad_norm": 1.1427783966064453,
+ "learning_rate": 4.051194775567268e-06,
+ "loss": 0.0098,
+ "step": 26120
+ },
+ {
+ "epoch": 71.17438692098092,
+ "grad_norm": 1.5944044589996338,
+ "learning_rate": 4.050485437070297e-06,
+ "loss": 0.0374,
+ "step": 26121
+ },
+ {
+ "epoch": 71.17711171662125,
+ "grad_norm": 2.116370916366577,
+ "learning_rate": 4.049776144907955e-06,
+ "loss": 0.0218,
+ "step": 26122
+ },
+ {
+ "epoch": 71.17983651226157,
+ "grad_norm": 1.980747938156128,
+ "learning_rate": 4.0490668990857615e-06,
+ "loss": 0.0502,
+ "step": 26123
+ },
+ {
+ "epoch": 71.1825613079019,
+ "grad_norm": 1.6822067499160767,
+ "learning_rate": 4.048357699609246e-06,
+ "loss": 0.0578,
+ "step": 26124
+ },
+ {
+ "epoch": 71.18528610354224,
+ "grad_norm": 0.9914113283157349,
+ "learning_rate": 4.047648546483931e-06,
+ "loss": 0.0138,
+ "step": 26125
+ },
+ {
+ "epoch": 71.18801089918256,
+ "grad_norm": 1.4046696424484253,
+ "learning_rate": 4.046939439715336e-06,
+ "loss": 0.0138,
+ "step": 26126
+ },
+ {
+ "epoch": 71.19073569482289,
+ "grad_norm": 1.4945238828659058,
+ "learning_rate": 4.046230379308982e-06,
+ "loss": 0.0864,
+ "step": 26127
+ },
+ {
+ "epoch": 71.19346049046321,
+ "grad_norm": 1.517984390258789,
+ "learning_rate": 4.045521365270394e-06,
+ "loss": 0.0985,
+ "step": 26128
+ },
+ {
+ "epoch": 71.19618528610354,
+ "grad_norm": 1.2046754360198975,
+ "learning_rate": 4.044812397605098e-06,
+ "loss": 0.0104,
+ "step": 26129
+ },
+ {
+ "epoch": 71.19891008174388,
+ "grad_norm": 1.5820962190628052,
+ "learning_rate": 4.044103476318612e-06,
+ "loss": 0.0194,
+ "step": 26130
+ },
+ {
+ "epoch": 71.2016348773842,
+ "grad_norm": 1.0689986944198608,
+ "learning_rate": 4.043394601416455e-06,
+ "loss": 0.0108,
+ "step": 26131
+ },
+ {
+ "epoch": 71.20435967302453,
+ "grad_norm": 0.8419798016548157,
+ "learning_rate": 4.042685772904146e-06,
+ "loss": 0.0116,
+ "step": 26132
+ },
+ {
+ "epoch": 71.20708446866485,
+ "grad_norm": 1.4269641637802124,
+ "learning_rate": 4.041976990787212e-06,
+ "loss": 0.0154,
+ "step": 26133
+ },
+ {
+ "epoch": 71.20980926430518,
+ "grad_norm": 1.94810950756073,
+ "learning_rate": 4.041268255071169e-06,
+ "loss": 0.0795,
+ "step": 26134
+ },
+ {
+ "epoch": 71.2125340599455,
+ "grad_norm": 1.2101612091064453,
+ "learning_rate": 4.040559565761537e-06,
+ "loss": 0.0248,
+ "step": 26135
+ },
+ {
+ "epoch": 71.21525885558583,
+ "grad_norm": 2.0708260536193848,
+ "learning_rate": 4.039850922863833e-06,
+ "loss": 0.0703,
+ "step": 26136
+ },
+ {
+ "epoch": 71.21798365122616,
+ "grad_norm": 1.2251814603805542,
+ "learning_rate": 4.03914232638358e-06,
+ "loss": 0.0186,
+ "step": 26137
+ },
+ {
+ "epoch": 71.22070844686648,
+ "grad_norm": 0.7113569378852844,
+ "learning_rate": 4.038433776326297e-06,
+ "loss": 0.0085,
+ "step": 26138
+ },
+ {
+ "epoch": 71.22343324250681,
+ "grad_norm": 1.1255258321762085,
+ "learning_rate": 4.037725272697498e-06,
+ "loss": 0.0881,
+ "step": 26139
+ },
+ {
+ "epoch": 71.22615803814713,
+ "grad_norm": 1.3126517534255981,
+ "learning_rate": 4.0370168155027e-06,
+ "loss": 0.0143,
+ "step": 26140
+ },
+ {
+ "epoch": 71.22888283378747,
+ "grad_norm": 1.5968025922775269,
+ "learning_rate": 4.036308404747426e-06,
+ "loss": 0.0492,
+ "step": 26141
+ },
+ {
+ "epoch": 71.2316076294278,
+ "grad_norm": 1.4621212482452393,
+ "learning_rate": 4.035600040437191e-06,
+ "loss": 0.033,
+ "step": 26142
+ },
+ {
+ "epoch": 71.23433242506812,
+ "grad_norm": 1.923327922821045,
+ "learning_rate": 4.034891722577511e-06,
+ "loss": 0.0246,
+ "step": 26143
+ },
+ {
+ "epoch": 71.23705722070845,
+ "grad_norm": 1.2151654958724976,
+ "learning_rate": 4.034183451173897e-06,
+ "loss": 0.0142,
+ "step": 26144
+ },
+ {
+ "epoch": 71.23978201634877,
+ "grad_norm": 1.257883906364441,
+ "learning_rate": 4.033475226231872e-06,
+ "loss": 0.0203,
+ "step": 26145
+ },
+ {
+ "epoch": 71.2425068119891,
+ "grad_norm": 0.989088773727417,
+ "learning_rate": 4.032767047756954e-06,
+ "loss": 0.0186,
+ "step": 26146
+ },
+ {
+ "epoch": 71.24523160762942,
+ "grad_norm": 1.738734245300293,
+ "learning_rate": 4.032058915754653e-06,
+ "loss": 0.066,
+ "step": 26147
+ },
+ {
+ "epoch": 71.24795640326975,
+ "grad_norm": 1.2281105518341064,
+ "learning_rate": 4.031350830230482e-06,
+ "loss": 0.0133,
+ "step": 26148
+ },
+ {
+ "epoch": 71.25068119891009,
+ "grad_norm": 1.6524379253387451,
+ "learning_rate": 4.0306427911899614e-06,
+ "loss": 0.0363,
+ "step": 26149
+ },
+ {
+ "epoch": 71.2534059945504,
+ "grad_norm": 1.425186038017273,
+ "learning_rate": 4.029934798638603e-06,
+ "loss": 0.1261,
+ "step": 26150
+ },
+ {
+ "epoch": 71.25613079019074,
+ "grad_norm": 0.9839799404144287,
+ "learning_rate": 4.02922685258192e-06,
+ "loss": 0.0075,
+ "step": 26151
+ },
+ {
+ "epoch": 71.25885558583106,
+ "grad_norm": 3.9212653636932373,
+ "learning_rate": 4.028518953025423e-06,
+ "loss": 0.0188,
+ "step": 26152
+ },
+ {
+ "epoch": 71.26158038147139,
+ "grad_norm": 1.3625445365905762,
+ "learning_rate": 4.027811099974632e-06,
+ "loss": 0.0253,
+ "step": 26153
+ },
+ {
+ "epoch": 71.26430517711172,
+ "grad_norm": 1.8668583631515503,
+ "learning_rate": 4.0271032934350555e-06,
+ "loss": 0.0368,
+ "step": 26154
+ },
+ {
+ "epoch": 71.26702997275204,
+ "grad_norm": 1.3115588426589966,
+ "learning_rate": 4.026395533412206e-06,
+ "loss": 0.0135,
+ "step": 26155
+ },
+ {
+ "epoch": 71.26975476839237,
+ "grad_norm": 1.352772831916809,
+ "learning_rate": 4.025687819911593e-06,
+ "loss": 0.012,
+ "step": 26156
+ },
+ {
+ "epoch": 71.2724795640327,
+ "grad_norm": 1.4850956201553345,
+ "learning_rate": 4.024980152938735e-06,
+ "loss": 0.0103,
+ "step": 26157
+ },
+ {
+ "epoch": 71.27520435967303,
+ "grad_norm": 0.9487407207489014,
+ "learning_rate": 4.0242725324991375e-06,
+ "loss": 0.0102,
+ "step": 26158
+ },
+ {
+ "epoch": 71.27792915531334,
+ "grad_norm": 1.0661958456039429,
+ "learning_rate": 4.023564958598314e-06,
+ "loss": 0.0136,
+ "step": 26159
+ },
+ {
+ "epoch": 71.28065395095368,
+ "grad_norm": 0.8288411498069763,
+ "learning_rate": 4.0228574312417715e-06,
+ "loss": 0.0081,
+ "step": 26160
+ },
+ {
+ "epoch": 71.28337874659401,
+ "grad_norm": 0.922279953956604,
+ "learning_rate": 4.022149950435026e-06,
+ "loss": 0.0085,
+ "step": 26161
+ },
+ {
+ "epoch": 71.28610354223433,
+ "grad_norm": 1.3048334121704102,
+ "learning_rate": 4.02144251618358e-06,
+ "loss": 0.0518,
+ "step": 26162
+ },
+ {
+ "epoch": 71.28882833787466,
+ "grad_norm": 2.335592746734619,
+ "learning_rate": 4.020735128492952e-06,
+ "loss": 0.0514,
+ "step": 26163
+ },
+ {
+ "epoch": 71.29155313351498,
+ "grad_norm": 1.4333584308624268,
+ "learning_rate": 4.020027787368642e-06,
+ "loss": 0.0328,
+ "step": 26164
+ },
+ {
+ "epoch": 71.29427792915531,
+ "grad_norm": 1.838970422744751,
+ "learning_rate": 4.019320492816167e-06,
+ "loss": 0.0126,
+ "step": 26165
+ },
+ {
+ "epoch": 71.29700272479565,
+ "grad_norm": 1.8687559366226196,
+ "learning_rate": 4.018613244841031e-06,
+ "loss": 0.0617,
+ "step": 26166
+ },
+ {
+ "epoch": 71.29972752043597,
+ "grad_norm": 2.6422948837280273,
+ "learning_rate": 4.017906043448744e-06,
+ "loss": 0.0226,
+ "step": 26167
+ },
+ {
+ "epoch": 71.3024523160763,
+ "grad_norm": 2.2029354572296143,
+ "learning_rate": 4.017198888644807e-06,
+ "loss": 0.014,
+ "step": 26168
+ },
+ {
+ "epoch": 71.30517711171662,
+ "grad_norm": 2.7957370281219482,
+ "learning_rate": 4.016491780434737e-06,
+ "loss": 0.102,
+ "step": 26169
+ },
+ {
+ "epoch": 71.30790190735695,
+ "grad_norm": 1.3171052932739258,
+ "learning_rate": 4.015784718824036e-06,
+ "loss": 0.0234,
+ "step": 26170
+ },
+ {
+ "epoch": 71.31062670299727,
+ "grad_norm": 1.2335399389266968,
+ "learning_rate": 4.015077703818211e-06,
+ "loss": 0.0113,
+ "step": 26171
+ },
+ {
+ "epoch": 71.3133514986376,
+ "grad_norm": 1.4294570684432983,
+ "learning_rate": 4.014370735422765e-06,
+ "loss": 0.1905,
+ "step": 26172
+ },
+ {
+ "epoch": 71.31607629427793,
+ "grad_norm": 1.482164740562439,
+ "learning_rate": 4.013663813643211e-06,
+ "loss": 0.0191,
+ "step": 26173
+ },
+ {
+ "epoch": 71.31880108991825,
+ "grad_norm": 1.648977518081665,
+ "learning_rate": 4.012956938485049e-06,
+ "loss": 0.0278,
+ "step": 26174
+ },
+ {
+ "epoch": 71.32152588555859,
+ "grad_norm": 1.5765844583511353,
+ "learning_rate": 4.012250109953787e-06,
+ "loss": 0.0148,
+ "step": 26175
+ },
+ {
+ "epoch": 71.3242506811989,
+ "grad_norm": 1.1029653549194336,
+ "learning_rate": 4.011543328054924e-06,
+ "loss": 0.1161,
+ "step": 26176
+ },
+ {
+ "epoch": 71.32697547683924,
+ "grad_norm": 1.7210744619369507,
+ "learning_rate": 4.010836592793973e-06,
+ "loss": 0.1144,
+ "step": 26177
+ },
+ {
+ "epoch": 71.32970027247957,
+ "grad_norm": 1.4789369106292725,
+ "learning_rate": 4.010129904176434e-06,
+ "loss": 0.0151,
+ "step": 26178
+ },
+ {
+ "epoch": 71.33242506811989,
+ "grad_norm": 0.8296172618865967,
+ "learning_rate": 4.009423262207807e-06,
+ "loss": 0.0109,
+ "step": 26179
+ },
+ {
+ "epoch": 71.33514986376022,
+ "grad_norm": 1.5505448579788208,
+ "learning_rate": 4.0087166668936015e-06,
+ "loss": 0.0752,
+ "step": 26180
+ },
+ {
+ "epoch": 71.33787465940054,
+ "grad_norm": 1.0747650861740112,
+ "learning_rate": 4.008010118239315e-06,
+ "loss": 0.0078,
+ "step": 26181
+ },
+ {
+ "epoch": 71.34059945504087,
+ "grad_norm": 1.1608500480651855,
+ "learning_rate": 4.007303616250457e-06,
+ "loss": 0.0749,
+ "step": 26182
+ },
+ {
+ "epoch": 71.34332425068119,
+ "grad_norm": 1.7502143383026123,
+ "learning_rate": 4.006597160932524e-06,
+ "loss": 0.2086,
+ "step": 26183
+ },
+ {
+ "epoch": 71.34604904632153,
+ "grad_norm": 1.637658953666687,
+ "learning_rate": 4.005890752291021e-06,
+ "loss": 0.0703,
+ "step": 26184
+ },
+ {
+ "epoch": 71.34877384196186,
+ "grad_norm": 1.8448556661605835,
+ "learning_rate": 4.005184390331443e-06,
+ "loss": 0.0199,
+ "step": 26185
+ },
+ {
+ "epoch": 71.35149863760218,
+ "grad_norm": 1.7737858295440674,
+ "learning_rate": 4.0044780750593005e-06,
+ "loss": 0.0273,
+ "step": 26186
+ },
+ {
+ "epoch": 71.35422343324251,
+ "grad_norm": 1.2493573427200317,
+ "learning_rate": 4.003771806480089e-06,
+ "loss": 0.0148,
+ "step": 26187
+ },
+ {
+ "epoch": 71.35694822888283,
+ "grad_norm": 1.3300665616989136,
+ "learning_rate": 4.003065584599306e-06,
+ "loss": 0.0318,
+ "step": 26188
+ },
+ {
+ "epoch": 71.35967302452316,
+ "grad_norm": 1.2546465396881104,
+ "learning_rate": 4.002359409422459e-06,
+ "loss": 0.0249,
+ "step": 26189
+ },
+ {
+ "epoch": 71.3623978201635,
+ "grad_norm": 1.7923640012741089,
+ "learning_rate": 4.001653280955045e-06,
+ "loss": 0.0119,
+ "step": 26190
+ },
+ {
+ "epoch": 71.36512261580381,
+ "grad_norm": 1.045331597328186,
+ "learning_rate": 4.000947199202562e-06,
+ "loss": 0.011,
+ "step": 26191
+ },
+ {
+ "epoch": 71.36784741144415,
+ "grad_norm": 1.2104140520095825,
+ "learning_rate": 4.000241164170504e-06,
+ "loss": 0.0195,
+ "step": 26192
+ },
+ {
+ "epoch": 71.37057220708446,
+ "grad_norm": 1.8952561616897583,
+ "learning_rate": 3.99953517586438e-06,
+ "loss": 0.068,
+ "step": 26193
+ },
+ {
+ "epoch": 71.3732970027248,
+ "grad_norm": 1.410151720046997,
+ "learning_rate": 3.9988292342896816e-06,
+ "loss": 0.0142,
+ "step": 26194
+ },
+ {
+ "epoch": 71.37602179836512,
+ "grad_norm": 2.592226505279541,
+ "learning_rate": 3.998123339451908e-06,
+ "loss": 0.1449,
+ "step": 26195
+ },
+ {
+ "epoch": 71.37874659400545,
+ "grad_norm": 1.5944788455963135,
+ "learning_rate": 3.997417491356554e-06,
+ "loss": 0.0509,
+ "step": 26196
+ },
+ {
+ "epoch": 71.38147138964578,
+ "grad_norm": 1.0004335641860962,
+ "learning_rate": 3.996711690009119e-06,
+ "loss": 0.0122,
+ "step": 26197
+ },
+ {
+ "epoch": 71.3841961852861,
+ "grad_norm": 0.8946349620819092,
+ "learning_rate": 3.996005935415104e-06,
+ "loss": 0.0184,
+ "step": 26198
+ },
+ {
+ "epoch": 71.38692098092643,
+ "grad_norm": 1.483757734298706,
+ "learning_rate": 3.995300227580001e-06,
+ "loss": 0.0417,
+ "step": 26199
+ },
+ {
+ "epoch": 71.38964577656675,
+ "grad_norm": 1.8611695766448975,
+ "learning_rate": 3.994594566509306e-06,
+ "loss": 0.1478,
+ "step": 26200
+ },
+ {
+ "epoch": 71.39237057220708,
+ "grad_norm": 0.9337136745452881,
+ "learning_rate": 3.993888952208513e-06,
+ "loss": 0.01,
+ "step": 26201
+ },
+ {
+ "epoch": 71.39509536784742,
+ "grad_norm": 1.279174566268921,
+ "learning_rate": 3.993183384683124e-06,
+ "loss": 0.019,
+ "step": 26202
+ },
+ {
+ "epoch": 71.39782016348774,
+ "grad_norm": 5.647134780883789,
+ "learning_rate": 3.992477863938628e-06,
+ "loss": 0.0503,
+ "step": 26203
+ },
+ {
+ "epoch": 71.40054495912807,
+ "grad_norm": 2.0077638626098633,
+ "learning_rate": 3.99177238998052e-06,
+ "loss": 0.0661,
+ "step": 26204
+ },
+ {
+ "epoch": 71.40326975476839,
+ "grad_norm": 2.1069788932800293,
+ "learning_rate": 3.991066962814293e-06,
+ "loss": 0.2233,
+ "step": 26205
+ },
+ {
+ "epoch": 71.40599455040872,
+ "grad_norm": 1.5020653009414673,
+ "learning_rate": 3.990361582445447e-06,
+ "loss": 0.0155,
+ "step": 26206
+ },
+ {
+ "epoch": 71.40871934604904,
+ "grad_norm": 1.3419537544250488,
+ "learning_rate": 3.989656248879471e-06,
+ "loss": 0.0152,
+ "step": 26207
+ },
+ {
+ "epoch": 71.41144414168937,
+ "grad_norm": 1.1337846517562866,
+ "learning_rate": 3.988950962121855e-06,
+ "loss": 0.019,
+ "step": 26208
+ },
+ {
+ "epoch": 71.4141689373297,
+ "grad_norm": 1.5153937339782715,
+ "learning_rate": 3.988245722178099e-06,
+ "loss": 0.0229,
+ "step": 26209
+ },
+ {
+ "epoch": 71.41689373297002,
+ "grad_norm": 1.4541800022125244,
+ "learning_rate": 3.987540529053692e-06,
+ "loss": 0.0748,
+ "step": 26210
+ },
+ {
+ "epoch": 71.41961852861036,
+ "grad_norm": 1.3142108917236328,
+ "learning_rate": 3.9868353827541265e-06,
+ "loss": 0.0215,
+ "step": 26211
+ },
+ {
+ "epoch": 71.42234332425068,
+ "grad_norm": 1.4109646081924438,
+ "learning_rate": 3.986130283284889e-06,
+ "loss": 0.0189,
+ "step": 26212
+ },
+ {
+ "epoch": 71.42506811989101,
+ "grad_norm": 1.3492043018341064,
+ "learning_rate": 3.985425230651479e-06,
+ "loss": 0.0231,
+ "step": 26213
+ },
+ {
+ "epoch": 71.42779291553134,
+ "grad_norm": 1.416506052017212,
+ "learning_rate": 3.98472022485938e-06,
+ "loss": 0.0133,
+ "step": 26214
+ },
+ {
+ "epoch": 71.43051771117166,
+ "grad_norm": 2.096459150314331,
+ "learning_rate": 3.984015265914091e-06,
+ "loss": 0.0242,
+ "step": 26215
+ },
+ {
+ "epoch": 71.433242506812,
+ "grad_norm": 1.2788257598876953,
+ "learning_rate": 3.983310353821096e-06,
+ "loss": 0.014,
+ "step": 26216
+ },
+ {
+ "epoch": 71.43596730245231,
+ "grad_norm": 1.3523991107940674,
+ "learning_rate": 3.982605488585884e-06,
+ "loss": 0.0254,
+ "step": 26217
+ },
+ {
+ "epoch": 71.43869209809264,
+ "grad_norm": 1.5162110328674316,
+ "learning_rate": 3.9819006702139496e-06,
+ "loss": 0.0937,
+ "step": 26218
+ },
+ {
+ "epoch": 71.44141689373296,
+ "grad_norm": 1.5342859029769897,
+ "learning_rate": 3.981195898710779e-06,
+ "loss": 0.0426,
+ "step": 26219
+ },
+ {
+ "epoch": 71.4441416893733,
+ "grad_norm": 1.0337839126586914,
+ "learning_rate": 3.9804911740818604e-06,
+ "loss": 0.0932,
+ "step": 26220
+ },
+ {
+ "epoch": 71.44686648501363,
+ "grad_norm": 1.282143473625183,
+ "learning_rate": 3.97978649633268e-06,
+ "loss": 0.0231,
+ "step": 26221
+ },
+ {
+ "epoch": 71.44959128065395,
+ "grad_norm": 1.6369537115097046,
+ "learning_rate": 3.979081865468733e-06,
+ "loss": 0.0638,
+ "step": 26222
+ },
+ {
+ "epoch": 71.45231607629428,
+ "grad_norm": 2.0469844341278076,
+ "learning_rate": 3.978377281495502e-06,
+ "loss": 0.026,
+ "step": 26223
+ },
+ {
+ "epoch": 71.4550408719346,
+ "grad_norm": 2.0292582511901855,
+ "learning_rate": 3.977672744418475e-06,
+ "loss": 0.0399,
+ "step": 26224
+ },
+ {
+ "epoch": 71.45776566757493,
+ "grad_norm": 1.2165614366531372,
+ "learning_rate": 3.976968254243135e-06,
+ "loss": 0.0166,
+ "step": 26225
+ },
+ {
+ "epoch": 71.46049046321527,
+ "grad_norm": 2.044311761856079,
+ "learning_rate": 3.976263810974977e-06,
+ "loss": 0.0156,
+ "step": 26226
+ },
+ {
+ "epoch": 71.46321525885558,
+ "grad_norm": 1.4900121688842773,
+ "learning_rate": 3.975559414619481e-06,
+ "loss": 0.0367,
+ "step": 26227
+ },
+ {
+ "epoch": 71.46594005449592,
+ "grad_norm": 1.1144543886184692,
+ "learning_rate": 3.974855065182135e-06,
+ "loss": 0.0135,
+ "step": 26228
+ },
+ {
+ "epoch": 71.46866485013624,
+ "grad_norm": 1.0528050661087036,
+ "learning_rate": 3.974150762668421e-06,
+ "loss": 0.0102,
+ "step": 26229
+ },
+ {
+ "epoch": 71.47138964577657,
+ "grad_norm": 0.7115939259529114,
+ "learning_rate": 3.973446507083829e-06,
+ "loss": 0.0098,
+ "step": 26230
+ },
+ {
+ "epoch": 71.47411444141689,
+ "grad_norm": 1.447872519493103,
+ "learning_rate": 3.9727422984338384e-06,
+ "loss": 0.0341,
+ "step": 26231
+ },
+ {
+ "epoch": 71.47683923705722,
+ "grad_norm": 1.0760903358459473,
+ "learning_rate": 3.972038136723941e-06,
+ "loss": 0.0119,
+ "step": 26232
+ },
+ {
+ "epoch": 71.47956403269755,
+ "grad_norm": 1.4421820640563965,
+ "learning_rate": 3.971334021959613e-06,
+ "loss": 0.0227,
+ "step": 26233
+ },
+ {
+ "epoch": 71.48228882833787,
+ "grad_norm": 1.881272315979004,
+ "learning_rate": 3.970629954146346e-06,
+ "loss": 0.0139,
+ "step": 26234
+ },
+ {
+ "epoch": 71.4850136239782,
+ "grad_norm": 1.3068293333053589,
+ "learning_rate": 3.9699259332896165e-06,
+ "loss": 0.026,
+ "step": 26235
+ },
+ {
+ "epoch": 71.48773841961852,
+ "grad_norm": 2.022392988204956,
+ "learning_rate": 3.969221959394911e-06,
+ "loss": 0.1447,
+ "step": 26236
+ },
+ {
+ "epoch": 71.49046321525886,
+ "grad_norm": 1.538809895515442,
+ "learning_rate": 3.9685180324677074e-06,
+ "loss": 0.0977,
+ "step": 26237
+ },
+ {
+ "epoch": 71.49318801089919,
+ "grad_norm": 1.0287480354309082,
+ "learning_rate": 3.967814152513495e-06,
+ "loss": 0.0105,
+ "step": 26238
+ },
+ {
+ "epoch": 71.49591280653951,
+ "grad_norm": 0.9661115407943726,
+ "learning_rate": 3.967110319537751e-06,
+ "loss": 0.0133,
+ "step": 26239
+ },
+ {
+ "epoch": 71.49863760217984,
+ "grad_norm": 1.827789306640625,
+ "learning_rate": 3.966406533545959e-06,
+ "loss": 0.0356,
+ "step": 26240
+ },
+ {
+ "epoch": 71.50136239782016,
+ "grad_norm": 1.092270016670227,
+ "learning_rate": 3.965702794543594e-06,
+ "loss": 0.0386,
+ "step": 26241
+ },
+ {
+ "epoch": 71.50408719346049,
+ "grad_norm": 1.1108534336090088,
+ "learning_rate": 3.964999102536145e-06,
+ "loss": 0.0186,
+ "step": 26242
+ },
+ {
+ "epoch": 71.50681198910081,
+ "grad_norm": 2.2043964862823486,
+ "learning_rate": 3.96429545752909e-06,
+ "loss": 0.025,
+ "step": 26243
+ },
+ {
+ "epoch": 71.50953678474114,
+ "grad_norm": 1.1183091402053833,
+ "learning_rate": 3.963591859527906e-06,
+ "loss": 0.0305,
+ "step": 26244
+ },
+ {
+ "epoch": 71.51226158038148,
+ "grad_norm": 1.6148035526275635,
+ "learning_rate": 3.962888308538072e-06,
+ "loss": 0.0849,
+ "step": 26245
+ },
+ {
+ "epoch": 71.5149863760218,
+ "grad_norm": 1.0135269165039062,
+ "learning_rate": 3.962184804565074e-06,
+ "loss": 0.0093,
+ "step": 26246
+ },
+ {
+ "epoch": 71.51771117166213,
+ "grad_norm": 1.1068103313446045,
+ "learning_rate": 3.961481347614385e-06,
+ "loss": 0.0196,
+ "step": 26247
+ },
+ {
+ "epoch": 71.52043596730245,
+ "grad_norm": 2.220489263534546,
+ "learning_rate": 3.9607779376914814e-06,
+ "loss": 0.0407,
+ "step": 26248
+ },
+ {
+ "epoch": 71.52316076294278,
+ "grad_norm": 0.8655511140823364,
+ "learning_rate": 3.9600745748018445e-06,
+ "loss": 0.0114,
+ "step": 26249
+ },
+ {
+ "epoch": 71.52588555858311,
+ "grad_norm": 2.113290548324585,
+ "learning_rate": 3.959371258950956e-06,
+ "loss": 0.1059,
+ "step": 26250
+ },
+ {
+ "epoch": 71.52861035422343,
+ "grad_norm": 1.0015840530395508,
+ "learning_rate": 3.95866799014429e-06,
+ "loss": 0.0094,
+ "step": 26251
+ },
+ {
+ "epoch": 71.53133514986376,
+ "grad_norm": 1.2897162437438965,
+ "learning_rate": 3.957964768387324e-06,
+ "loss": 0.02,
+ "step": 26252
+ },
+ {
+ "epoch": 71.53405994550408,
+ "grad_norm": 1.0283410549163818,
+ "learning_rate": 3.957261593685529e-06,
+ "loss": 0.0212,
+ "step": 26253
+ },
+ {
+ "epoch": 71.53678474114442,
+ "grad_norm": 1.9388185739517212,
+ "learning_rate": 3.9565584660443916e-06,
+ "loss": 0.0166,
+ "step": 26254
+ },
+ {
+ "epoch": 71.53950953678473,
+ "grad_norm": 2.2432351112365723,
+ "learning_rate": 3.955855385469381e-06,
+ "loss": 0.0157,
+ "step": 26255
+ },
+ {
+ "epoch": 71.54223433242507,
+ "grad_norm": 1.6917701959609985,
+ "learning_rate": 3.955152351965974e-06,
+ "loss": 0.0655,
+ "step": 26256
+ },
+ {
+ "epoch": 71.5449591280654,
+ "grad_norm": 1.2279574871063232,
+ "learning_rate": 3.954449365539643e-06,
+ "loss": 0.0271,
+ "step": 26257
+ },
+ {
+ "epoch": 71.54768392370572,
+ "grad_norm": 4.825351715087891,
+ "learning_rate": 3.953746426195869e-06,
+ "loss": 0.0699,
+ "step": 26258
+ },
+ {
+ "epoch": 71.55040871934605,
+ "grad_norm": 1.4365836381912231,
+ "learning_rate": 3.953043533940124e-06,
+ "loss": 0.0285,
+ "step": 26259
+ },
+ {
+ "epoch": 71.55313351498637,
+ "grad_norm": 2.117095470428467,
+ "learning_rate": 3.95234068877788e-06,
+ "loss": 0.0295,
+ "step": 26260
+ },
+ {
+ "epoch": 71.5558583106267,
+ "grad_norm": 2.536057472229004,
+ "learning_rate": 3.951637890714609e-06,
+ "loss": 0.0719,
+ "step": 26261
+ },
+ {
+ "epoch": 71.55858310626704,
+ "grad_norm": 1.2509047985076904,
+ "learning_rate": 3.950935139755792e-06,
+ "loss": 0.0334,
+ "step": 26262
+ },
+ {
+ "epoch": 71.56130790190736,
+ "grad_norm": 1.1519496440887451,
+ "learning_rate": 3.950232435906897e-06,
+ "loss": 0.0094,
+ "step": 26263
+ },
+ {
+ "epoch": 71.56403269754769,
+ "grad_norm": 0.8407151699066162,
+ "learning_rate": 3.949529779173397e-06,
+ "loss": 0.0114,
+ "step": 26264
+ },
+ {
+ "epoch": 71.566757493188,
+ "grad_norm": 1.6823724508285522,
+ "learning_rate": 3.9488271695607595e-06,
+ "loss": 0.0649,
+ "step": 26265
+ },
+ {
+ "epoch": 71.56948228882834,
+ "grad_norm": 1.9866987466812134,
+ "learning_rate": 3.948124607074463e-06,
+ "loss": 0.0281,
+ "step": 26266
+ },
+ {
+ "epoch": 71.57220708446866,
+ "grad_norm": 1.5112754106521606,
+ "learning_rate": 3.94742209171998e-06,
+ "loss": 0.0184,
+ "step": 26267
+ },
+ {
+ "epoch": 71.57493188010899,
+ "grad_norm": 0.9462478160858154,
+ "learning_rate": 3.946719623502778e-06,
+ "loss": 0.011,
+ "step": 26268
+ },
+ {
+ "epoch": 71.57765667574932,
+ "grad_norm": 1.5100865364074707,
+ "learning_rate": 3.9460172024283305e-06,
+ "loss": 0.0097,
+ "step": 26269
+ },
+ {
+ "epoch": 71.58038147138964,
+ "grad_norm": 1.039736270904541,
+ "learning_rate": 3.945314828502102e-06,
+ "loss": 0.0092,
+ "step": 26270
+ },
+ {
+ "epoch": 71.58310626702998,
+ "grad_norm": 1.254364252090454,
+ "learning_rate": 3.94461250172957e-06,
+ "loss": 0.0112,
+ "step": 26271
+ },
+ {
+ "epoch": 71.5858310626703,
+ "grad_norm": 2.0190629959106445,
+ "learning_rate": 3.9439102221162016e-06,
+ "loss": 0.0409,
+ "step": 26272
+ },
+ {
+ "epoch": 71.58855585831063,
+ "grad_norm": 1.6365042924880981,
+ "learning_rate": 3.943207989667461e-06,
+ "loss": 0.028,
+ "step": 26273
+ },
+ {
+ "epoch": 71.59128065395096,
+ "grad_norm": 1.356303095817566,
+ "learning_rate": 3.9425058043888265e-06,
+ "loss": 0.0825,
+ "step": 26274
+ },
+ {
+ "epoch": 71.59400544959128,
+ "grad_norm": 0.9060015082359314,
+ "learning_rate": 3.941803666285761e-06,
+ "loss": 0.0102,
+ "step": 26275
+ },
+ {
+ "epoch": 71.59673024523161,
+ "grad_norm": 2.069345712661743,
+ "learning_rate": 3.941101575363733e-06,
+ "loss": 0.0872,
+ "step": 26276
+ },
+ {
+ "epoch": 71.59945504087193,
+ "grad_norm": 1.5578397512435913,
+ "learning_rate": 3.9403995316282084e-06,
+ "loss": 0.0151,
+ "step": 26277
+ },
+ {
+ "epoch": 71.60217983651226,
+ "grad_norm": 1.1284468173980713,
+ "learning_rate": 3.93969753508466e-06,
+ "loss": 0.0166,
+ "step": 26278
+ },
+ {
+ "epoch": 71.60490463215258,
+ "grad_norm": 1.226115345954895,
+ "learning_rate": 3.938995585738552e-06,
+ "loss": 0.0523,
+ "step": 26279
+ },
+ {
+ "epoch": 71.60762942779292,
+ "grad_norm": 0.901167094707489,
+ "learning_rate": 3.9382936835953525e-06,
+ "loss": 0.0121,
+ "step": 26280
+ },
+ {
+ "epoch": 71.61035422343325,
+ "grad_norm": 1.5379002094268799,
+ "learning_rate": 3.937591828660522e-06,
+ "loss": 0.022,
+ "step": 26281
+ },
+ {
+ "epoch": 71.61307901907357,
+ "grad_norm": 2.118368148803711,
+ "learning_rate": 3.9368900209395325e-06,
+ "loss": 0.0369,
+ "step": 26282
+ },
+ {
+ "epoch": 71.6158038147139,
+ "grad_norm": 1.1054248809814453,
+ "learning_rate": 3.936188260437851e-06,
+ "loss": 0.0122,
+ "step": 26283
+ },
+ {
+ "epoch": 71.61852861035422,
+ "grad_norm": 1.6093049049377441,
+ "learning_rate": 3.935486547160939e-06,
+ "loss": 0.0761,
+ "step": 26284
+ },
+ {
+ "epoch": 71.62125340599455,
+ "grad_norm": 1.632544994354248,
+ "learning_rate": 3.9347848811142644e-06,
+ "loss": 0.0604,
+ "step": 26285
+ },
+ {
+ "epoch": 71.62397820163488,
+ "grad_norm": 1.8605846166610718,
+ "learning_rate": 3.934083262303287e-06,
+ "loss": 0.0424,
+ "step": 26286
+ },
+ {
+ "epoch": 71.6267029972752,
+ "grad_norm": 1.3432339429855347,
+ "learning_rate": 3.933381690733475e-06,
+ "loss": 0.0178,
+ "step": 26287
+ },
+ {
+ "epoch": 71.62942779291554,
+ "grad_norm": 3.772735357284546,
+ "learning_rate": 3.932680166410293e-06,
+ "loss": 0.092,
+ "step": 26288
+ },
+ {
+ "epoch": 71.63215258855585,
+ "grad_norm": 2.1265132427215576,
+ "learning_rate": 3.931978689339202e-06,
+ "loss": 0.0386,
+ "step": 26289
+ },
+ {
+ "epoch": 71.63487738419619,
+ "grad_norm": 1.2988773584365845,
+ "learning_rate": 3.9312772595256624e-06,
+ "loss": 0.0145,
+ "step": 26290
+ },
+ {
+ "epoch": 71.6376021798365,
+ "grad_norm": 1.6851640939712524,
+ "learning_rate": 3.9305758769751436e-06,
+ "loss": 0.0144,
+ "step": 26291
+ },
+ {
+ "epoch": 71.64032697547684,
+ "grad_norm": 1.7908259630203247,
+ "learning_rate": 3.9298745416931035e-06,
+ "loss": 0.0402,
+ "step": 26292
+ },
+ {
+ "epoch": 71.64305177111717,
+ "grad_norm": 1.635419249534607,
+ "learning_rate": 3.9291732536850056e-06,
+ "loss": 0.0279,
+ "step": 26293
+ },
+ {
+ "epoch": 71.64577656675749,
+ "grad_norm": 1.743915319442749,
+ "learning_rate": 3.928472012956308e-06,
+ "loss": 0.0266,
+ "step": 26294
+ },
+ {
+ "epoch": 71.64850136239782,
+ "grad_norm": 1.5342369079589844,
+ "learning_rate": 3.927770819512479e-06,
+ "loss": 0.0427,
+ "step": 26295
+ },
+ {
+ "epoch": 71.65122615803814,
+ "grad_norm": 2.007218360900879,
+ "learning_rate": 3.927069673358976e-06,
+ "loss": 0.0357,
+ "step": 26296
+ },
+ {
+ "epoch": 71.65395095367847,
+ "grad_norm": 1.9007644653320312,
+ "learning_rate": 3.926368574501253e-06,
+ "loss": 0.0352,
+ "step": 26297
+ },
+ {
+ "epoch": 71.65667574931881,
+ "grad_norm": 0.8928267359733582,
+ "learning_rate": 3.925667522944781e-06,
+ "loss": 0.0104,
+ "step": 26298
+ },
+ {
+ "epoch": 71.65940054495913,
+ "grad_norm": 1.234500527381897,
+ "learning_rate": 3.924966518695015e-06,
+ "loss": 0.0217,
+ "step": 26299
+ },
+ {
+ "epoch": 71.66212534059946,
+ "grad_norm": 1.5415781736373901,
+ "learning_rate": 3.92426556175741e-06,
+ "loss": 0.0151,
+ "step": 26300
+ },
+ {
+ "epoch": 71.66485013623978,
+ "grad_norm": 1.450955867767334,
+ "learning_rate": 3.923564652137433e-06,
+ "loss": 0.0221,
+ "step": 26301
+ },
+ {
+ "epoch": 71.66757493188011,
+ "grad_norm": 2.278723955154419,
+ "learning_rate": 3.922863789840535e-06,
+ "loss": 0.0611,
+ "step": 26302
+ },
+ {
+ "epoch": 71.67029972752043,
+ "grad_norm": 1.6736479997634888,
+ "learning_rate": 3.922162974872182e-06,
+ "loss": 0.0514,
+ "step": 26303
+ },
+ {
+ "epoch": 71.67302452316076,
+ "grad_norm": 0.862838089466095,
+ "learning_rate": 3.9214622072378285e-06,
+ "loss": 0.012,
+ "step": 26304
+ },
+ {
+ "epoch": 71.6757493188011,
+ "grad_norm": 1.5385475158691406,
+ "learning_rate": 3.920761486942932e-06,
+ "loss": 0.074,
+ "step": 26305
+ },
+ {
+ "epoch": 71.67847411444141,
+ "grad_norm": 1.253670573234558,
+ "learning_rate": 3.920060813992945e-06,
+ "loss": 0.0283,
+ "step": 26306
+ },
+ {
+ "epoch": 71.68119891008175,
+ "grad_norm": 1.5977778434753418,
+ "learning_rate": 3.9193601883933326e-06,
+ "loss": 0.0232,
+ "step": 26307
+ },
+ {
+ "epoch": 71.68392370572207,
+ "grad_norm": 1.4180091619491577,
+ "learning_rate": 3.918659610149548e-06,
+ "loss": 0.0227,
+ "step": 26308
+ },
+ {
+ "epoch": 71.6866485013624,
+ "grad_norm": 1.8163914680480957,
+ "learning_rate": 3.917959079267045e-06,
+ "loss": 0.0144,
+ "step": 26309
+ },
+ {
+ "epoch": 71.68937329700273,
+ "grad_norm": 1.313603162765503,
+ "learning_rate": 3.917258595751279e-06,
+ "loss": 0.0313,
+ "step": 26310
+ },
+ {
+ "epoch": 71.69209809264305,
+ "grad_norm": 1.617401361465454,
+ "learning_rate": 3.916558159607711e-06,
+ "loss": 0.1,
+ "step": 26311
+ },
+ {
+ "epoch": 71.69482288828338,
+ "grad_norm": 1.8350651264190674,
+ "learning_rate": 3.9158577708417915e-06,
+ "loss": 0.0859,
+ "step": 26312
+ },
+ {
+ "epoch": 71.6975476839237,
+ "grad_norm": 2.088927984237671,
+ "learning_rate": 3.915157429458976e-06,
+ "loss": 0.077,
+ "step": 26313
+ },
+ {
+ "epoch": 71.70027247956403,
+ "grad_norm": 1.0165544748306274,
+ "learning_rate": 3.914457135464714e-06,
+ "loss": 0.0184,
+ "step": 26314
+ },
+ {
+ "epoch": 71.70299727520435,
+ "grad_norm": 1.6307564973831177,
+ "learning_rate": 3.913756888864468e-06,
+ "loss": 0.1048,
+ "step": 26315
+ },
+ {
+ "epoch": 71.70572207084469,
+ "grad_norm": 1.6565589904785156,
+ "learning_rate": 3.913056689663689e-06,
+ "loss": 0.0321,
+ "step": 26316
+ },
+ {
+ "epoch": 71.70844686648502,
+ "grad_norm": 1.31356942653656,
+ "learning_rate": 3.912356537867823e-06,
+ "loss": 0.0273,
+ "step": 26317
+ },
+ {
+ "epoch": 71.71117166212534,
+ "grad_norm": 1.2461836338043213,
+ "learning_rate": 3.91165643348233e-06,
+ "loss": 0.013,
+ "step": 26318
+ },
+ {
+ "epoch": 71.71389645776567,
+ "grad_norm": 1.2573047876358032,
+ "learning_rate": 3.910956376512664e-06,
+ "loss": 0.0167,
+ "step": 26319
+ },
+ {
+ "epoch": 71.71662125340599,
+ "grad_norm": 1.7412090301513672,
+ "learning_rate": 3.9102563669642724e-06,
+ "loss": 0.0142,
+ "step": 26320
+ },
+ {
+ "epoch": 71.71934604904632,
+ "grad_norm": 1.642682433128357,
+ "learning_rate": 3.909556404842609e-06,
+ "loss": 0.0273,
+ "step": 26321
+ },
+ {
+ "epoch": 71.72207084468666,
+ "grad_norm": 1.9757850170135498,
+ "learning_rate": 3.908856490153121e-06,
+ "loss": 0.1344,
+ "step": 26322
+ },
+ {
+ "epoch": 71.72479564032697,
+ "grad_norm": 1.6263879537582397,
+ "learning_rate": 3.9081566229012665e-06,
+ "loss": 0.0419,
+ "step": 26323
+ },
+ {
+ "epoch": 71.7275204359673,
+ "grad_norm": 1.1319217681884766,
+ "learning_rate": 3.907456803092492e-06,
+ "loss": 0.1355,
+ "step": 26324
+ },
+ {
+ "epoch": 71.73024523160763,
+ "grad_norm": 1.1433744430541992,
+ "learning_rate": 3.906757030732248e-06,
+ "loss": 0.0108,
+ "step": 26325
+ },
+ {
+ "epoch": 71.73297002724796,
+ "grad_norm": 1.016618013381958,
+ "learning_rate": 3.90605730582598e-06,
+ "loss": 0.0119,
+ "step": 26326
+ },
+ {
+ "epoch": 71.73569482288828,
+ "grad_norm": 1.0866345167160034,
+ "learning_rate": 3.905357628379145e-06,
+ "loss": 0.0088,
+ "step": 26327
+ },
+ {
+ "epoch": 71.73841961852861,
+ "grad_norm": 1.305197834968567,
+ "learning_rate": 3.9046579983971885e-06,
+ "loss": 0.0159,
+ "step": 26328
+ },
+ {
+ "epoch": 71.74114441416894,
+ "grad_norm": 1.537607192993164,
+ "learning_rate": 3.90395841588556e-06,
+ "loss": 0.0201,
+ "step": 26329
+ },
+ {
+ "epoch": 71.74386920980926,
+ "grad_norm": 2.4615180492401123,
+ "learning_rate": 3.903258880849703e-06,
+ "loss": 0.0898,
+ "step": 26330
+ },
+ {
+ "epoch": 71.7465940054496,
+ "grad_norm": 1.4269615411758423,
+ "learning_rate": 3.902559393295073e-06,
+ "loss": 0.0417,
+ "step": 26331
+ },
+ {
+ "epoch": 71.74931880108991,
+ "grad_norm": 2.5746142864227295,
+ "learning_rate": 3.901859953227115e-06,
+ "loss": 0.0749,
+ "step": 26332
+ },
+ {
+ "epoch": 71.75204359673025,
+ "grad_norm": 1.1960408687591553,
+ "learning_rate": 3.901160560651276e-06,
+ "loss": 0.014,
+ "step": 26333
+ },
+ {
+ "epoch": 71.75476839237058,
+ "grad_norm": 2.010399103164673,
+ "learning_rate": 3.900461215572997e-06,
+ "loss": 0.1704,
+ "step": 26334
+ },
+ {
+ "epoch": 71.7574931880109,
+ "grad_norm": 1.2992916107177734,
+ "learning_rate": 3.899761917997731e-06,
+ "loss": 0.0185,
+ "step": 26335
+ },
+ {
+ "epoch": 71.76021798365123,
+ "grad_norm": 1.1827375888824463,
+ "learning_rate": 3.899062667930926e-06,
+ "loss": 0.0602,
+ "step": 26336
+ },
+ {
+ "epoch": 71.76294277929155,
+ "grad_norm": 1.6133100986480713,
+ "learning_rate": 3.898363465378025e-06,
+ "loss": 0.1065,
+ "step": 26337
+ },
+ {
+ "epoch": 71.76566757493188,
+ "grad_norm": 2.2801573276519775,
+ "learning_rate": 3.897664310344469e-06,
+ "loss": 0.0301,
+ "step": 26338
+ },
+ {
+ "epoch": 71.7683923705722,
+ "grad_norm": 3.9963529109954834,
+ "learning_rate": 3.89696520283571e-06,
+ "loss": 0.0498,
+ "step": 26339
+ },
+ {
+ "epoch": 71.77111716621253,
+ "grad_norm": 5.5988545417785645,
+ "learning_rate": 3.89626614285719e-06,
+ "loss": 0.0181,
+ "step": 26340
+ },
+ {
+ "epoch": 71.77384196185287,
+ "grad_norm": 1.083262324333191,
+ "learning_rate": 3.895567130414353e-06,
+ "loss": 0.0187,
+ "step": 26341
+ },
+ {
+ "epoch": 71.77656675749319,
+ "grad_norm": 1.4803800582885742,
+ "learning_rate": 3.894868165512639e-06,
+ "loss": 0.0212,
+ "step": 26342
+ },
+ {
+ "epoch": 71.77929155313352,
+ "grad_norm": 2.443227767944336,
+ "learning_rate": 3.894169248157499e-06,
+ "loss": 0.0568,
+ "step": 26343
+ },
+ {
+ "epoch": 71.78201634877384,
+ "grad_norm": 1.1668777465820312,
+ "learning_rate": 3.8934703783543715e-06,
+ "loss": 0.0121,
+ "step": 26344
+ },
+ {
+ "epoch": 71.78474114441417,
+ "grad_norm": 1.5105572938919067,
+ "learning_rate": 3.8927715561087006e-06,
+ "loss": 0.0262,
+ "step": 26345
+ },
+ {
+ "epoch": 71.7874659400545,
+ "grad_norm": 1.1023918390274048,
+ "learning_rate": 3.892072781425926e-06,
+ "loss": 0.0141,
+ "step": 26346
+ },
+ {
+ "epoch": 71.79019073569482,
+ "grad_norm": 1.073654294013977,
+ "learning_rate": 3.8913740543114945e-06,
+ "loss": 0.0093,
+ "step": 26347
+ },
+ {
+ "epoch": 71.79291553133515,
+ "grad_norm": 2.592024803161621,
+ "learning_rate": 3.8906753747708446e-06,
+ "loss": 0.0485,
+ "step": 26348
+ },
+ {
+ "epoch": 71.79564032697547,
+ "grad_norm": 2.0060946941375732,
+ "learning_rate": 3.889976742809419e-06,
+ "loss": 0.1407,
+ "step": 26349
+ },
+ {
+ "epoch": 71.7983651226158,
+ "grad_norm": 0.9550496935844421,
+ "learning_rate": 3.889278158432654e-06,
+ "loss": 0.0083,
+ "step": 26350
+ },
+ {
+ "epoch": 71.80108991825612,
+ "grad_norm": 0.9777756333351135,
+ "learning_rate": 3.888579621645995e-06,
+ "loss": 0.0092,
+ "step": 26351
+ },
+ {
+ "epoch": 71.80381471389646,
+ "grad_norm": 1.7608275413513184,
+ "learning_rate": 3.887881132454884e-06,
+ "loss": 0.119,
+ "step": 26352
+ },
+ {
+ "epoch": 71.80653950953679,
+ "grad_norm": 1.1783101558685303,
+ "learning_rate": 3.887182690864758e-06,
+ "loss": 0.0784,
+ "step": 26353
+ },
+ {
+ "epoch": 71.80926430517711,
+ "grad_norm": 1.3825039863586426,
+ "learning_rate": 3.886484296881057e-06,
+ "loss": 0.0311,
+ "step": 26354
+ },
+ {
+ "epoch": 71.81198910081744,
+ "grad_norm": 1.2359912395477295,
+ "learning_rate": 3.885785950509215e-06,
+ "loss": 0.018,
+ "step": 26355
+ },
+ {
+ "epoch": 71.81471389645776,
+ "grad_norm": 1.8653959035873413,
+ "learning_rate": 3.88508765175468e-06,
+ "loss": 0.1036,
+ "step": 26356
+ },
+ {
+ "epoch": 71.8174386920981,
+ "grad_norm": 0.9016221761703491,
+ "learning_rate": 3.8843894006228844e-06,
+ "loss": 0.0109,
+ "step": 26357
+ },
+ {
+ "epoch": 71.82016348773843,
+ "grad_norm": 1.2703698873519897,
+ "learning_rate": 3.883691197119267e-06,
+ "loss": 0.0231,
+ "step": 26358
+ },
+ {
+ "epoch": 71.82288828337875,
+ "grad_norm": 2.1642441749572754,
+ "learning_rate": 3.8829930412492636e-06,
+ "loss": 0.035,
+ "step": 26359
+ },
+ {
+ "epoch": 71.82561307901908,
+ "grad_norm": 1.8612263202667236,
+ "learning_rate": 3.882294933018316e-06,
+ "loss": 0.0403,
+ "step": 26360
+ },
+ {
+ "epoch": 71.8283378746594,
+ "grad_norm": 0.9989765882492065,
+ "learning_rate": 3.881596872431859e-06,
+ "loss": 0.0111,
+ "step": 26361
+ },
+ {
+ "epoch": 71.83106267029973,
+ "grad_norm": 1.189489722251892,
+ "learning_rate": 3.880898859495326e-06,
+ "loss": 0.0385,
+ "step": 26362
+ },
+ {
+ "epoch": 71.83378746594005,
+ "grad_norm": 1.462558627128601,
+ "learning_rate": 3.880200894214158e-06,
+ "loss": 0.0453,
+ "step": 26363
+ },
+ {
+ "epoch": 71.83651226158038,
+ "grad_norm": 2.839057207107544,
+ "learning_rate": 3.879502976593788e-06,
+ "loss": 0.0517,
+ "step": 26364
+ },
+ {
+ "epoch": 71.83923705722071,
+ "grad_norm": 1.5579946041107178,
+ "learning_rate": 3.878805106639653e-06,
+ "loss": 0.0266,
+ "step": 26365
+ },
+ {
+ "epoch": 71.84196185286103,
+ "grad_norm": 2.2266242504119873,
+ "learning_rate": 3.8781072843571836e-06,
+ "loss": 0.0482,
+ "step": 26366
+ },
+ {
+ "epoch": 71.84468664850137,
+ "grad_norm": 2.1506807804107666,
+ "learning_rate": 3.87740950975182e-06,
+ "loss": 0.0806,
+ "step": 26367
+ },
+ {
+ "epoch": 71.84741144414168,
+ "grad_norm": 1.0857911109924316,
+ "learning_rate": 3.876711782828991e-06,
+ "loss": 0.0104,
+ "step": 26368
+ },
+ {
+ "epoch": 71.85013623978202,
+ "grad_norm": 1.5073323249816895,
+ "learning_rate": 3.876014103594137e-06,
+ "loss": 0.0162,
+ "step": 26369
+ },
+ {
+ "epoch": 71.85286103542235,
+ "grad_norm": 1.1000714302062988,
+ "learning_rate": 3.8753164720526884e-06,
+ "loss": 0.0107,
+ "step": 26370
+ },
+ {
+ "epoch": 71.85558583106267,
+ "grad_norm": 1.6282438039779663,
+ "learning_rate": 3.874618888210074e-06,
+ "loss": 0.048,
+ "step": 26371
+ },
+ {
+ "epoch": 71.858310626703,
+ "grad_norm": 1.8581920862197876,
+ "learning_rate": 3.873921352071734e-06,
+ "loss": 0.1094,
+ "step": 26372
+ },
+ {
+ "epoch": 71.86103542234332,
+ "grad_norm": 2.395339012145996,
+ "learning_rate": 3.873223863643098e-06,
+ "loss": 0.1604,
+ "step": 26373
+ },
+ {
+ "epoch": 71.86376021798365,
+ "grad_norm": 0.8712912797927856,
+ "learning_rate": 3.872526422929597e-06,
+ "loss": 0.0112,
+ "step": 26374
+ },
+ {
+ "epoch": 71.86648501362397,
+ "grad_norm": 1.4951149225234985,
+ "learning_rate": 3.87182902993666e-06,
+ "loss": 0.0478,
+ "step": 26375
+ },
+ {
+ "epoch": 71.8692098092643,
+ "grad_norm": 1.673077940940857,
+ "learning_rate": 3.871131684669725e-06,
+ "loss": 0.0995,
+ "step": 26376
+ },
+ {
+ "epoch": 71.87193460490464,
+ "grad_norm": 0.8115631341934204,
+ "learning_rate": 3.870434387134217e-06,
+ "loss": 0.0108,
+ "step": 26377
+ },
+ {
+ "epoch": 71.87465940054496,
+ "grad_norm": 1.0734792947769165,
+ "learning_rate": 3.869737137335571e-06,
+ "loss": 0.0114,
+ "step": 26378
+ },
+ {
+ "epoch": 71.87738419618529,
+ "grad_norm": 0.9704106450080872,
+ "learning_rate": 3.869039935279211e-06,
+ "loss": 0.016,
+ "step": 26379
+ },
+ {
+ "epoch": 71.88010899182561,
+ "grad_norm": 2.544384717941284,
+ "learning_rate": 3.868342780970573e-06,
+ "loss": 0.0502,
+ "step": 26380
+ },
+ {
+ "epoch": 71.88283378746594,
+ "grad_norm": 1.4441921710968018,
+ "learning_rate": 3.867645674415085e-06,
+ "loss": 0.013,
+ "step": 26381
+ },
+ {
+ "epoch": 71.88555858310627,
+ "grad_norm": 0.7929157614707947,
+ "learning_rate": 3.8669486156181715e-06,
+ "loss": 0.0115,
+ "step": 26382
+ },
+ {
+ "epoch": 71.88828337874659,
+ "grad_norm": 2.091632604598999,
+ "learning_rate": 3.866251604585268e-06,
+ "loss": 0.0496,
+ "step": 26383
+ },
+ {
+ "epoch": 71.89100817438693,
+ "grad_norm": 1.2835640907287598,
+ "learning_rate": 3.8655546413218005e-06,
+ "loss": 0.0157,
+ "step": 26384
+ },
+ {
+ "epoch": 71.89373297002724,
+ "grad_norm": 1.2114100456237793,
+ "learning_rate": 3.864857725833194e-06,
+ "loss": 0.0125,
+ "step": 26385
+ },
+ {
+ "epoch": 71.89645776566758,
+ "grad_norm": 1.8582260608673096,
+ "learning_rate": 3.864160858124877e-06,
+ "loss": 0.0222,
+ "step": 26386
+ },
+ {
+ "epoch": 71.8991825613079,
+ "grad_norm": 1.3677500486373901,
+ "learning_rate": 3.863464038202276e-06,
+ "loss": 0.089,
+ "step": 26387
+ },
+ {
+ "epoch": 71.90190735694823,
+ "grad_norm": 4.47391939163208,
+ "learning_rate": 3.862767266070824e-06,
+ "loss": 0.1347,
+ "step": 26388
+ },
+ {
+ "epoch": 71.90463215258856,
+ "grad_norm": 1.896376609802246,
+ "learning_rate": 3.8620705417359425e-06,
+ "loss": 0.031,
+ "step": 26389
+ },
+ {
+ "epoch": 71.90735694822888,
+ "grad_norm": 2.2315080165863037,
+ "learning_rate": 3.861373865203059e-06,
+ "loss": 0.1682,
+ "step": 26390
+ },
+ {
+ "epoch": 71.91008174386921,
+ "grad_norm": 1.0350768566131592,
+ "learning_rate": 3.860677236477594e-06,
+ "loss": 0.0088,
+ "step": 26391
+ },
+ {
+ "epoch": 71.91280653950953,
+ "grad_norm": 1.7950676679611206,
+ "learning_rate": 3.859980655564981e-06,
+ "loss": 0.0298,
+ "step": 26392
+ },
+ {
+ "epoch": 71.91553133514986,
+ "grad_norm": 1.0138187408447266,
+ "learning_rate": 3.85928412247064e-06,
+ "loss": 0.0128,
+ "step": 26393
+ },
+ {
+ "epoch": 71.9182561307902,
+ "grad_norm": 1.7906683683395386,
+ "learning_rate": 3.858587637199998e-06,
+ "loss": 0.0242,
+ "step": 26394
+ },
+ {
+ "epoch": 71.92098092643052,
+ "grad_norm": 1.9756791591644287,
+ "learning_rate": 3.857891199758473e-06,
+ "loss": 0.0211,
+ "step": 26395
+ },
+ {
+ "epoch": 71.92370572207085,
+ "grad_norm": 1.5662087202072144,
+ "learning_rate": 3.8571948101514975e-06,
+ "loss": 0.0335,
+ "step": 26396
+ },
+ {
+ "epoch": 71.92643051771117,
+ "grad_norm": 1.5943063497543335,
+ "learning_rate": 3.856498468384492e-06,
+ "loss": 0.0196,
+ "step": 26397
+ },
+ {
+ "epoch": 71.9291553133515,
+ "grad_norm": 0.8371425867080688,
+ "learning_rate": 3.855802174462878e-06,
+ "loss": 0.0124,
+ "step": 26398
+ },
+ {
+ "epoch": 71.93188010899182,
+ "grad_norm": 1.307281255722046,
+ "learning_rate": 3.8551059283920755e-06,
+ "loss": 0.0336,
+ "step": 26399
+ },
+ {
+ "epoch": 71.93460490463215,
+ "grad_norm": 1.670754075050354,
+ "learning_rate": 3.854409730177514e-06,
+ "loss": 0.0332,
+ "step": 26400
+ },
+ {
+ "epoch": 71.93732970027249,
+ "grad_norm": 1.148181676864624,
+ "learning_rate": 3.8537135798246114e-06,
+ "loss": 0.0126,
+ "step": 26401
+ },
+ {
+ "epoch": 71.9400544959128,
+ "grad_norm": 1.1004871129989624,
+ "learning_rate": 3.8530174773387894e-06,
+ "loss": 0.0186,
+ "step": 26402
+ },
+ {
+ "epoch": 71.94277929155314,
+ "grad_norm": 1.1367872953414917,
+ "learning_rate": 3.852321422725464e-06,
+ "loss": 0.0113,
+ "step": 26403
+ },
+ {
+ "epoch": 71.94550408719346,
+ "grad_norm": 0.9564248919487,
+ "learning_rate": 3.851625415990066e-06,
+ "loss": 0.0119,
+ "step": 26404
+ },
+ {
+ "epoch": 71.94822888283379,
+ "grad_norm": 1.8052042722702026,
+ "learning_rate": 3.850929457138013e-06,
+ "loss": 0.0634,
+ "step": 26405
+ },
+ {
+ "epoch": 71.95095367847412,
+ "grad_norm": 1.4355988502502441,
+ "learning_rate": 3.850233546174723e-06,
+ "loss": 0.0082,
+ "step": 26406
+ },
+ {
+ "epoch": 71.95367847411444,
+ "grad_norm": 1.9912664890289307,
+ "learning_rate": 3.849537683105613e-06,
+ "loss": 0.0092,
+ "step": 26407
+ },
+ {
+ "epoch": 71.95640326975477,
+ "grad_norm": 1.5045493841171265,
+ "learning_rate": 3.848841867936109e-06,
+ "loss": 0.086,
+ "step": 26408
+ },
+ {
+ "epoch": 71.95912806539509,
+ "grad_norm": 1.847098708152771,
+ "learning_rate": 3.848146100671627e-06,
+ "loss": 0.02,
+ "step": 26409
+ },
+ {
+ "epoch": 71.96185286103542,
+ "grad_norm": 0.8918415307998657,
+ "learning_rate": 3.847450381317584e-06,
+ "loss": 0.011,
+ "step": 26410
+ },
+ {
+ "epoch": 71.96457765667574,
+ "grad_norm": 1.4698964357376099,
+ "learning_rate": 3.8467547098793965e-06,
+ "loss": 0.0206,
+ "step": 26411
+ },
+ {
+ "epoch": 71.96730245231608,
+ "grad_norm": 1.4101264476776123,
+ "learning_rate": 3.846059086362489e-06,
+ "loss": 0.0531,
+ "step": 26412
+ },
+ {
+ "epoch": 71.97002724795641,
+ "grad_norm": 1.9930609464645386,
+ "learning_rate": 3.845363510772275e-06,
+ "loss": 0.1103,
+ "step": 26413
+ },
+ {
+ "epoch": 71.97275204359673,
+ "grad_norm": 1.4591147899627686,
+ "learning_rate": 3.844667983114172e-06,
+ "loss": 0.0245,
+ "step": 26414
+ },
+ {
+ "epoch": 71.97547683923706,
+ "grad_norm": 1.4823271036148071,
+ "learning_rate": 3.843972503393594e-06,
+ "loss": 0.0259,
+ "step": 26415
+ },
+ {
+ "epoch": 71.97820163487738,
+ "grad_norm": 1.6148217916488647,
+ "learning_rate": 3.843277071615962e-06,
+ "loss": 0.0148,
+ "step": 26416
+ },
+ {
+ "epoch": 71.98092643051771,
+ "grad_norm": 1.3901013135910034,
+ "learning_rate": 3.842581687786691e-06,
+ "loss": 0.0166,
+ "step": 26417
+ },
+ {
+ "epoch": 71.98365122615803,
+ "grad_norm": 1.9366790056228638,
+ "learning_rate": 3.841886351911195e-06,
+ "loss": 0.0279,
+ "step": 26418
+ },
+ {
+ "epoch": 71.98637602179836,
+ "grad_norm": 1.7721729278564453,
+ "learning_rate": 3.841191063994887e-06,
+ "loss": 0.0192,
+ "step": 26419
+ },
+ {
+ "epoch": 71.9891008174387,
+ "grad_norm": 1.3709228038787842,
+ "learning_rate": 3.840495824043184e-06,
+ "loss": 0.0235,
+ "step": 26420
+ },
+ {
+ "epoch": 71.99182561307902,
+ "grad_norm": 1.1202332973480225,
+ "learning_rate": 3.839800632061505e-06,
+ "loss": 0.0068,
+ "step": 26421
+ },
+ {
+ "epoch": 71.99455040871935,
+ "grad_norm": 1.4877219200134277,
+ "learning_rate": 3.839105488055261e-06,
+ "loss": 0.0262,
+ "step": 26422
+ },
+ {
+ "epoch": 71.99727520435967,
+ "grad_norm": 1.0598182678222656,
+ "learning_rate": 3.838410392029861e-06,
+ "loss": 0.0123,
+ "step": 26423
+ },
+ {
+ "epoch": 72.0,
+ "grad_norm": 1.73102605342865,
+ "learning_rate": 3.837715343990727e-06,
+ "loss": 0.0931,
+ "step": 26424
+ },
+ {
+ "epoch": 72.00272479564033,
+ "grad_norm": 1.7781505584716797,
+ "learning_rate": 3.837020343943267e-06,
+ "loss": 0.0327,
+ "step": 26425
+ },
+ {
+ "epoch": 72.00544959128065,
+ "grad_norm": 1.5768980979919434,
+ "learning_rate": 3.836325391892894e-06,
+ "loss": 0.0986,
+ "step": 26426
+ },
+ {
+ "epoch": 72.00817438692098,
+ "grad_norm": 1.1979589462280273,
+ "learning_rate": 3.8356304878450165e-06,
+ "loss": 0.0172,
+ "step": 26427
+ },
+ {
+ "epoch": 72.0108991825613,
+ "grad_norm": 1.3924486637115479,
+ "learning_rate": 3.834935631805055e-06,
+ "loss": 0.0433,
+ "step": 26428
+ },
+ {
+ "epoch": 72.01362397820164,
+ "grad_norm": 0.9566327333450317,
+ "learning_rate": 3.834240823778416e-06,
+ "loss": 0.011,
+ "step": 26429
+ },
+ {
+ "epoch": 72.01634877384195,
+ "grad_norm": 1.3996365070343018,
+ "learning_rate": 3.833546063770512e-06,
+ "loss": 0.0139,
+ "step": 26430
+ },
+ {
+ "epoch": 72.01907356948229,
+ "grad_norm": 1.1611493825912476,
+ "learning_rate": 3.8328513517867475e-06,
+ "loss": 0.0129,
+ "step": 26431
+ },
+ {
+ "epoch": 72.02179836512262,
+ "grad_norm": 1.390105962753296,
+ "learning_rate": 3.832156687832542e-06,
+ "loss": 0.0436,
+ "step": 26432
+ },
+ {
+ "epoch": 72.02452316076294,
+ "grad_norm": 1.3504163026809692,
+ "learning_rate": 3.831462071913302e-06,
+ "loss": 0.034,
+ "step": 26433
+ },
+ {
+ "epoch": 72.02724795640327,
+ "grad_norm": 1.8577570915222168,
+ "learning_rate": 3.8307675040344375e-06,
+ "loss": 0.0216,
+ "step": 26434
+ },
+ {
+ "epoch": 72.02997275204359,
+ "grad_norm": 1.1441876888275146,
+ "learning_rate": 3.830072984201353e-06,
+ "loss": 0.0071,
+ "step": 26435
+ },
+ {
+ "epoch": 72.03269754768392,
+ "grad_norm": 1.7976516485214233,
+ "learning_rate": 3.829378512419465e-06,
+ "loss": 0.0556,
+ "step": 26436
+ },
+ {
+ "epoch": 72.03542234332426,
+ "grad_norm": 1.9029868841171265,
+ "learning_rate": 3.828684088694174e-06,
+ "loss": 0.0527,
+ "step": 26437
+ },
+ {
+ "epoch": 72.03814713896458,
+ "grad_norm": 1.3553543090820312,
+ "learning_rate": 3.827989713030899e-06,
+ "loss": 0.038,
+ "step": 26438
+ },
+ {
+ "epoch": 72.04087193460491,
+ "grad_norm": 0.7604783177375793,
+ "learning_rate": 3.827295385435039e-06,
+ "loss": 0.0661,
+ "step": 26439
+ },
+ {
+ "epoch": 72.04359673024523,
+ "grad_norm": 1.704574704170227,
+ "learning_rate": 3.826601105912001e-06,
+ "loss": 0.0265,
+ "step": 26440
+ },
+ {
+ "epoch": 72.04632152588556,
+ "grad_norm": 1.8700414896011353,
+ "learning_rate": 3.825906874467199e-06,
+ "loss": 0.0231,
+ "step": 26441
+ },
+ {
+ "epoch": 72.04904632152588,
+ "grad_norm": 1.3272247314453125,
+ "learning_rate": 3.825212691106035e-06,
+ "loss": 0.0179,
+ "step": 26442
+ },
+ {
+ "epoch": 72.05177111716621,
+ "grad_norm": 1.5235869884490967,
+ "learning_rate": 3.824518555833916e-06,
+ "loss": 0.0464,
+ "step": 26443
+ },
+ {
+ "epoch": 72.05449591280654,
+ "grad_norm": 0.9689397215843201,
+ "learning_rate": 3.823824468656245e-06,
+ "loss": 0.0096,
+ "step": 26444
+ },
+ {
+ "epoch": 72.05722070844686,
+ "grad_norm": 1.1093796491622925,
+ "learning_rate": 3.823130429578432e-06,
+ "loss": 0.0141,
+ "step": 26445
+ },
+ {
+ "epoch": 72.0599455040872,
+ "grad_norm": 1.4514397382736206,
+ "learning_rate": 3.822436438605881e-06,
+ "loss": 0.014,
+ "step": 26446
+ },
+ {
+ "epoch": 72.06267029972751,
+ "grad_norm": 1.4896420240402222,
+ "learning_rate": 3.821742495743993e-06,
+ "loss": 0.0212,
+ "step": 26447
+ },
+ {
+ "epoch": 72.06539509536785,
+ "grad_norm": 2.197162389755249,
+ "learning_rate": 3.82104860099818e-06,
+ "loss": 0.0237,
+ "step": 26448
+ },
+ {
+ "epoch": 72.06811989100818,
+ "grad_norm": 1.2018966674804688,
+ "learning_rate": 3.8203547543738395e-06,
+ "loss": 0.018,
+ "step": 26449
+ },
+ {
+ "epoch": 72.0708446866485,
+ "grad_norm": 1.6985814571380615,
+ "learning_rate": 3.819660955876378e-06,
+ "loss": 0.0098,
+ "step": 26450
+ },
+ {
+ "epoch": 72.07356948228883,
+ "grad_norm": 1.5370759963989258,
+ "learning_rate": 3.818967205511196e-06,
+ "loss": 0.0346,
+ "step": 26451
+ },
+ {
+ "epoch": 72.07629427792915,
+ "grad_norm": 1.85963773727417,
+ "learning_rate": 3.8182735032837015e-06,
+ "loss": 0.0544,
+ "step": 26452
+ },
+ {
+ "epoch": 72.07901907356948,
+ "grad_norm": 1.8027689456939697,
+ "learning_rate": 3.817579849199293e-06,
+ "loss": 0.0166,
+ "step": 26453
+ },
+ {
+ "epoch": 72.0817438692098,
+ "grad_norm": 1.7440201044082642,
+ "learning_rate": 3.816886243263371e-06,
+ "loss": 0.0202,
+ "step": 26454
+ },
+ {
+ "epoch": 72.08446866485014,
+ "grad_norm": 1.8433420658111572,
+ "learning_rate": 3.816192685481344e-06,
+ "loss": 0.0185,
+ "step": 26455
+ },
+ {
+ "epoch": 72.08719346049047,
+ "grad_norm": 1.443068504333496,
+ "learning_rate": 3.8154991758586045e-06,
+ "loss": 0.0228,
+ "step": 26456
+ },
+ {
+ "epoch": 72.08991825613079,
+ "grad_norm": 1.1961045265197754,
+ "learning_rate": 3.814805714400562e-06,
+ "loss": 0.0103,
+ "step": 26457
+ },
+ {
+ "epoch": 72.09264305177112,
+ "grad_norm": 1.2628685235977173,
+ "learning_rate": 3.8141123011126135e-06,
+ "loss": 0.1363,
+ "step": 26458
+ },
+ {
+ "epoch": 72.09536784741144,
+ "grad_norm": 1.5852124691009521,
+ "learning_rate": 3.8134189360001604e-06,
+ "loss": 0.0545,
+ "step": 26459
+ },
+ {
+ "epoch": 72.09809264305177,
+ "grad_norm": 1.4393991231918335,
+ "learning_rate": 3.8127256190685968e-06,
+ "loss": 0.1781,
+ "step": 26460
+ },
+ {
+ "epoch": 72.1008174386921,
+ "grad_norm": 1.2271795272827148,
+ "learning_rate": 3.8120323503233315e-06,
+ "loss": 0.0119,
+ "step": 26461
+ },
+ {
+ "epoch": 72.10354223433242,
+ "grad_norm": 1.590436339378357,
+ "learning_rate": 3.811339129769758e-06,
+ "loss": 0.0411,
+ "step": 26462
+ },
+ {
+ "epoch": 72.10626702997276,
+ "grad_norm": 0.8444018363952637,
+ "learning_rate": 3.810645957413277e-06,
+ "loss": 0.0068,
+ "step": 26463
+ },
+ {
+ "epoch": 72.10899182561307,
+ "grad_norm": 1.6873846054077148,
+ "learning_rate": 3.8099528332592817e-06,
+ "loss": 0.0273,
+ "step": 26464
+ },
+ {
+ "epoch": 72.11171662125341,
+ "grad_norm": 1.4712756872177124,
+ "learning_rate": 3.8092597573131796e-06,
+ "loss": 0.0126,
+ "step": 26465
+ },
+ {
+ "epoch": 72.11444141689373,
+ "grad_norm": 1.7524689435958862,
+ "learning_rate": 3.8085667295803617e-06,
+ "loss": 0.0846,
+ "step": 26466
+ },
+ {
+ "epoch": 72.11716621253406,
+ "grad_norm": 1.6642565727233887,
+ "learning_rate": 3.8078737500662277e-06,
+ "loss": 0.0372,
+ "step": 26467
+ },
+ {
+ "epoch": 72.11989100817439,
+ "grad_norm": 1.5081899166107178,
+ "learning_rate": 3.8071808187761696e-06,
+ "loss": 0.0135,
+ "step": 26468
+ },
+ {
+ "epoch": 72.12261580381471,
+ "grad_norm": 1.361541748046875,
+ "learning_rate": 3.8064879357155915e-06,
+ "loss": 0.1126,
+ "step": 26469
+ },
+ {
+ "epoch": 72.12534059945504,
+ "grad_norm": 1.3909680843353271,
+ "learning_rate": 3.8057951008898865e-06,
+ "loss": 0.0598,
+ "step": 26470
+ },
+ {
+ "epoch": 72.12806539509536,
+ "grad_norm": 0.9216810464859009,
+ "learning_rate": 3.8051023143044464e-06,
+ "loss": 0.0054,
+ "step": 26471
+ },
+ {
+ "epoch": 72.1307901907357,
+ "grad_norm": 1.064292073249817,
+ "learning_rate": 3.8044095759646694e-06,
+ "loss": 0.0167,
+ "step": 26472
+ },
+ {
+ "epoch": 72.13351498637603,
+ "grad_norm": 1.999131202697754,
+ "learning_rate": 3.8037168858759556e-06,
+ "loss": 0.1884,
+ "step": 26473
+ },
+ {
+ "epoch": 72.13623978201635,
+ "grad_norm": 1.1804378032684326,
+ "learning_rate": 3.8030242440436947e-06,
+ "loss": 0.0333,
+ "step": 26474
+ },
+ {
+ "epoch": 72.13896457765668,
+ "grad_norm": 1.6175843477249146,
+ "learning_rate": 3.802331650473281e-06,
+ "loss": 0.0265,
+ "step": 26475
+ },
+ {
+ "epoch": 72.141689373297,
+ "grad_norm": 1.771360993385315,
+ "learning_rate": 3.8016391051701063e-06,
+ "loss": 0.0291,
+ "step": 26476
+ },
+ {
+ "epoch": 72.14441416893733,
+ "grad_norm": 1.0556108951568604,
+ "learning_rate": 3.8009466081395707e-06,
+ "loss": 0.0239,
+ "step": 26477
+ },
+ {
+ "epoch": 72.14713896457765,
+ "grad_norm": 1.7890031337738037,
+ "learning_rate": 3.8002541593870624e-06,
+ "loss": 0.0145,
+ "step": 26478
+ },
+ {
+ "epoch": 72.14986376021798,
+ "grad_norm": 2.17488169670105,
+ "learning_rate": 3.7995617589179758e-06,
+ "loss": 0.0436,
+ "step": 26479
+ },
+ {
+ "epoch": 72.15258855585832,
+ "grad_norm": 2.0764353275299072,
+ "learning_rate": 3.7988694067376987e-06,
+ "loss": 0.019,
+ "step": 26480
+ },
+ {
+ "epoch": 72.15531335149863,
+ "grad_norm": 1.6888645887374878,
+ "learning_rate": 3.7981771028516303e-06,
+ "loss": 0.0142,
+ "step": 26481
+ },
+ {
+ "epoch": 72.15803814713897,
+ "grad_norm": 2.5684759616851807,
+ "learning_rate": 3.7974848472651593e-06,
+ "loss": 0.0538,
+ "step": 26482
+ },
+ {
+ "epoch": 72.16076294277929,
+ "grad_norm": 1.640965461730957,
+ "learning_rate": 3.7967926399836775e-06,
+ "loss": 0.0273,
+ "step": 26483
+ },
+ {
+ "epoch": 72.16348773841962,
+ "grad_norm": 0.9806878566741943,
+ "learning_rate": 3.796100481012571e-06,
+ "loss": 0.0113,
+ "step": 26484
+ },
+ {
+ "epoch": 72.16621253405995,
+ "grad_norm": 1.3790879249572754,
+ "learning_rate": 3.7954083703572376e-06,
+ "loss": 0.0206,
+ "step": 26485
+ },
+ {
+ "epoch": 72.16893732970027,
+ "grad_norm": 1.8670237064361572,
+ "learning_rate": 3.7947163080230644e-06,
+ "loss": 0.0565,
+ "step": 26486
+ },
+ {
+ "epoch": 72.1716621253406,
+ "grad_norm": 1.27764892578125,
+ "learning_rate": 3.7940242940154404e-06,
+ "loss": 0.0204,
+ "step": 26487
+ },
+ {
+ "epoch": 72.17438692098092,
+ "grad_norm": 1.2938437461853027,
+ "learning_rate": 3.793332328339752e-06,
+ "loss": 0.0114,
+ "step": 26488
+ },
+ {
+ "epoch": 72.17711171662125,
+ "grad_norm": 3.005842924118042,
+ "learning_rate": 3.7926404110013913e-06,
+ "loss": 0.0533,
+ "step": 26489
+ },
+ {
+ "epoch": 72.17983651226157,
+ "grad_norm": 1.3412846326828003,
+ "learning_rate": 3.791948542005751e-06,
+ "loss": 0.0243,
+ "step": 26490
+ },
+ {
+ "epoch": 72.1825613079019,
+ "grad_norm": 1.7148683071136475,
+ "learning_rate": 3.791256721358216e-06,
+ "loss": 0.0433,
+ "step": 26491
+ },
+ {
+ "epoch": 72.18528610354224,
+ "grad_norm": 1.2569129467010498,
+ "learning_rate": 3.7905649490641704e-06,
+ "loss": 0.0927,
+ "step": 26492
+ },
+ {
+ "epoch": 72.18801089918256,
+ "grad_norm": 0.9402910470962524,
+ "learning_rate": 3.789873225129009e-06,
+ "loss": 0.0093,
+ "step": 26493
+ },
+ {
+ "epoch": 72.19073569482289,
+ "grad_norm": 1.4241782426834106,
+ "learning_rate": 3.789181549558114e-06,
+ "loss": 0.0669,
+ "step": 26494
+ },
+ {
+ "epoch": 72.19346049046321,
+ "grad_norm": 1.1895034313201904,
+ "learning_rate": 3.788489922356874e-06,
+ "loss": 0.0135,
+ "step": 26495
+ },
+ {
+ "epoch": 72.19618528610354,
+ "grad_norm": 1.4970037937164307,
+ "learning_rate": 3.787798343530671e-06,
+ "loss": 0.012,
+ "step": 26496
+ },
+ {
+ "epoch": 72.19891008174388,
+ "grad_norm": 2.461583375930786,
+ "learning_rate": 3.7871068130848974e-06,
+ "loss": 0.0245,
+ "step": 26497
+ },
+ {
+ "epoch": 72.2016348773842,
+ "grad_norm": 1.714522361755371,
+ "learning_rate": 3.7864153310249376e-06,
+ "loss": 0.0542,
+ "step": 26498
+ },
+ {
+ "epoch": 72.20435967302453,
+ "grad_norm": 1.1652871370315552,
+ "learning_rate": 3.7857238973561737e-06,
+ "loss": 0.0125,
+ "step": 26499
+ },
+ {
+ "epoch": 72.20708446866485,
+ "grad_norm": 1.205812692642212,
+ "learning_rate": 3.785032512083989e-06,
+ "loss": 0.0202,
+ "step": 26500
+ },
+ {
+ "epoch": 72.20980926430518,
+ "grad_norm": 0.9344871044158936,
+ "learning_rate": 3.784341175213775e-06,
+ "loss": 0.013,
+ "step": 26501
+ },
+ {
+ "epoch": 72.2125340599455,
+ "grad_norm": 1.1782797574996948,
+ "learning_rate": 3.7836498867509108e-06,
+ "loss": 0.0629,
+ "step": 26502
+ },
+ {
+ "epoch": 72.21525885558583,
+ "grad_norm": 2.434677839279175,
+ "learning_rate": 3.782958646700782e-06,
+ "loss": 0.0563,
+ "step": 26503
+ },
+ {
+ "epoch": 72.21798365122616,
+ "grad_norm": 2.9076972007751465,
+ "learning_rate": 3.7822674550687676e-06,
+ "loss": 0.0288,
+ "step": 26504
+ },
+ {
+ "epoch": 72.22070844686648,
+ "grad_norm": 0.865658164024353,
+ "learning_rate": 3.7815763118602577e-06,
+ "loss": 0.0077,
+ "step": 26505
+ },
+ {
+ "epoch": 72.22343324250681,
+ "grad_norm": 1.0163655281066895,
+ "learning_rate": 3.7808852170806275e-06,
+ "loss": 0.0126,
+ "step": 26506
+ },
+ {
+ "epoch": 72.22615803814713,
+ "grad_norm": 1.0585347414016724,
+ "learning_rate": 3.780194170735266e-06,
+ "loss": 0.0108,
+ "step": 26507
+ },
+ {
+ "epoch": 72.22888283378747,
+ "grad_norm": 1.4861466884613037,
+ "learning_rate": 3.779503172829554e-06,
+ "loss": 0.0403,
+ "step": 26508
+ },
+ {
+ "epoch": 72.2316076294278,
+ "grad_norm": 2.0863685607910156,
+ "learning_rate": 3.778812223368866e-06,
+ "loss": 0.0573,
+ "step": 26509
+ },
+ {
+ "epoch": 72.23433242506812,
+ "grad_norm": 1.4625351428985596,
+ "learning_rate": 3.778121322358592e-06,
+ "loss": 0.0119,
+ "step": 26510
+ },
+ {
+ "epoch": 72.23705722070845,
+ "grad_norm": 0.6692936420440674,
+ "learning_rate": 3.77743046980411e-06,
+ "loss": 0.0073,
+ "step": 26511
+ },
+ {
+ "epoch": 72.23978201634877,
+ "grad_norm": 1.275282859802246,
+ "learning_rate": 3.7767396657107936e-06,
+ "loss": 0.0222,
+ "step": 26512
+ },
+ {
+ "epoch": 72.2425068119891,
+ "grad_norm": 1.4180877208709717,
+ "learning_rate": 3.776048910084034e-06,
+ "loss": 0.0209,
+ "step": 26513
+ },
+ {
+ "epoch": 72.24523160762942,
+ "grad_norm": 1.2667667865753174,
+ "learning_rate": 3.7753582029292037e-06,
+ "loss": 0.0864,
+ "step": 26514
+ },
+ {
+ "epoch": 72.24795640326975,
+ "grad_norm": 1.037386178970337,
+ "learning_rate": 3.774667544251683e-06,
+ "loss": 0.015,
+ "step": 26515
+ },
+ {
+ "epoch": 72.25068119891009,
+ "grad_norm": 0.9625787734985352,
+ "learning_rate": 3.773976934056849e-06,
+ "loss": 0.0147,
+ "step": 26516
+ },
+ {
+ "epoch": 72.2534059945504,
+ "grad_norm": 1.286232352256775,
+ "learning_rate": 3.7732863723500855e-06,
+ "loss": 0.0151,
+ "step": 26517
+ },
+ {
+ "epoch": 72.25613079019074,
+ "grad_norm": 1.3339887857437134,
+ "learning_rate": 3.7725958591367675e-06,
+ "loss": 0.0115,
+ "step": 26518
+ },
+ {
+ "epoch": 72.25885558583106,
+ "grad_norm": 1.3996771574020386,
+ "learning_rate": 3.7719053944222715e-06,
+ "loss": 0.0425,
+ "step": 26519
+ },
+ {
+ "epoch": 72.26158038147139,
+ "grad_norm": 1.096875548362732,
+ "learning_rate": 3.7712149782119733e-06,
+ "loss": 0.0091,
+ "step": 26520
+ },
+ {
+ "epoch": 72.26430517711172,
+ "grad_norm": 1.5591872930526733,
+ "learning_rate": 3.770524610511256e-06,
+ "loss": 0.0132,
+ "step": 26521
+ },
+ {
+ "epoch": 72.26702997275204,
+ "grad_norm": 1.4918768405914307,
+ "learning_rate": 3.7698342913254913e-06,
+ "loss": 0.0714,
+ "step": 26522
+ },
+ {
+ "epoch": 72.26975476839237,
+ "grad_norm": 1.1540303230285645,
+ "learning_rate": 3.769144020660054e-06,
+ "loss": 0.0344,
+ "step": 26523
+ },
+ {
+ "epoch": 72.2724795640327,
+ "grad_norm": 1.318822979927063,
+ "learning_rate": 3.7684537985203263e-06,
+ "loss": 0.0445,
+ "step": 26524
+ },
+ {
+ "epoch": 72.27520435967303,
+ "grad_norm": 1.1548603773117065,
+ "learning_rate": 3.7677636249116755e-06,
+ "loss": 0.0125,
+ "step": 26525
+ },
+ {
+ "epoch": 72.27792915531334,
+ "grad_norm": 1.7131145000457764,
+ "learning_rate": 3.7670734998394855e-06,
+ "loss": 0.0717,
+ "step": 26526
+ },
+ {
+ "epoch": 72.28065395095368,
+ "grad_norm": 1.8953056335449219,
+ "learning_rate": 3.7663834233091257e-06,
+ "loss": 0.0181,
+ "step": 26527
+ },
+ {
+ "epoch": 72.28337874659401,
+ "grad_norm": 0.9788191318511963,
+ "learning_rate": 3.765693395325971e-06,
+ "loss": 0.0105,
+ "step": 26528
+ },
+ {
+ "epoch": 72.28610354223433,
+ "grad_norm": 1.1896437406539917,
+ "learning_rate": 3.765003415895393e-06,
+ "loss": 0.0602,
+ "step": 26529
+ },
+ {
+ "epoch": 72.28882833787466,
+ "grad_norm": 1.0786694288253784,
+ "learning_rate": 3.76431348502277e-06,
+ "loss": 0.0129,
+ "step": 26530
+ },
+ {
+ "epoch": 72.29155313351498,
+ "grad_norm": 2.4346015453338623,
+ "learning_rate": 3.7636236027134733e-06,
+ "loss": 0.029,
+ "step": 26531
+ },
+ {
+ "epoch": 72.29427792915531,
+ "grad_norm": 1.0185109376907349,
+ "learning_rate": 3.7629337689728763e-06,
+ "loss": 0.0127,
+ "step": 26532
+ },
+ {
+ "epoch": 72.29700272479565,
+ "grad_norm": 0.7436034679412842,
+ "learning_rate": 3.762243983806345e-06,
+ "loss": 0.0094,
+ "step": 26533
+ },
+ {
+ "epoch": 72.29972752043597,
+ "grad_norm": 8.408117294311523,
+ "learning_rate": 3.7615542472192624e-06,
+ "loss": 0.0202,
+ "step": 26534
+ },
+ {
+ "epoch": 72.3024523160763,
+ "grad_norm": 1.4992198944091797,
+ "learning_rate": 3.7608645592169935e-06,
+ "loss": 0.0195,
+ "step": 26535
+ },
+ {
+ "epoch": 72.30517711171662,
+ "grad_norm": 1.2188254594802856,
+ "learning_rate": 3.7601749198049064e-06,
+ "loss": 0.0179,
+ "step": 26536
+ },
+ {
+ "epoch": 72.30790190735695,
+ "grad_norm": 1.9437785148620605,
+ "learning_rate": 3.75948532898838e-06,
+ "loss": 0.0636,
+ "step": 26537
+ },
+ {
+ "epoch": 72.31062670299727,
+ "grad_norm": 1.0437198877334595,
+ "learning_rate": 3.758795786772782e-06,
+ "loss": 0.0093,
+ "step": 26538
+ },
+ {
+ "epoch": 72.3133514986376,
+ "grad_norm": 1.2610844373703003,
+ "learning_rate": 3.7581062931634802e-06,
+ "loss": 0.0492,
+ "step": 26539
+ },
+ {
+ "epoch": 72.31607629427793,
+ "grad_norm": 1.4077996015548706,
+ "learning_rate": 3.7574168481658435e-06,
+ "loss": 0.022,
+ "step": 26540
+ },
+ {
+ "epoch": 72.31880108991825,
+ "grad_norm": 1.5455275774002075,
+ "learning_rate": 3.7567274517852424e-06,
+ "loss": 0.0229,
+ "step": 26541
+ },
+ {
+ "epoch": 72.32152588555859,
+ "grad_norm": 1.925561547279358,
+ "learning_rate": 3.756038104027051e-06,
+ "loss": 0.1139,
+ "step": 26542
+ },
+ {
+ "epoch": 72.3242506811989,
+ "grad_norm": 1.0570135116577148,
+ "learning_rate": 3.7553488048966346e-06,
+ "loss": 0.0122,
+ "step": 26543
+ },
+ {
+ "epoch": 72.32697547683924,
+ "grad_norm": 1.047953486442566,
+ "learning_rate": 3.75465955439936e-06,
+ "loss": 0.0143,
+ "step": 26544
+ },
+ {
+ "epoch": 72.32970027247957,
+ "grad_norm": 1.532833456993103,
+ "learning_rate": 3.7539703525405925e-06,
+ "loss": 0.0159,
+ "step": 26545
+ },
+ {
+ "epoch": 72.33242506811989,
+ "grad_norm": 0.8685709834098816,
+ "learning_rate": 3.7532811993257056e-06,
+ "loss": 0.0094,
+ "step": 26546
+ },
+ {
+ "epoch": 72.33514986376022,
+ "grad_norm": 1.509178876876831,
+ "learning_rate": 3.752592094760065e-06,
+ "loss": 0.0122,
+ "step": 26547
+ },
+ {
+ "epoch": 72.33787465940054,
+ "grad_norm": 1.7254077196121216,
+ "learning_rate": 3.751903038849034e-06,
+ "loss": 0.0184,
+ "step": 26548
+ },
+ {
+ "epoch": 72.34059945504087,
+ "grad_norm": 2.1536285877227783,
+ "learning_rate": 3.7512140315979794e-06,
+ "loss": 0.0941,
+ "step": 26549
+ },
+ {
+ "epoch": 72.34332425068119,
+ "grad_norm": 1.0312435626983643,
+ "learning_rate": 3.750525073012271e-06,
+ "loss": 0.1452,
+ "step": 26550
+ },
+ {
+ "epoch": 72.34604904632153,
+ "grad_norm": 1.7155730724334717,
+ "learning_rate": 3.7498361630972733e-06,
+ "loss": 0.0518,
+ "step": 26551
+ },
+ {
+ "epoch": 72.34877384196186,
+ "grad_norm": 1.2558894157409668,
+ "learning_rate": 3.7491473018583503e-06,
+ "loss": 0.0097,
+ "step": 26552
+ },
+ {
+ "epoch": 72.35149863760218,
+ "grad_norm": 0.9303576946258545,
+ "learning_rate": 3.7484584893008634e-06,
+ "loss": 0.0093,
+ "step": 26553
+ },
+ {
+ "epoch": 72.35422343324251,
+ "grad_norm": 0.936884343624115,
+ "learning_rate": 3.7477697254301836e-06,
+ "loss": 0.0094,
+ "step": 26554
+ },
+ {
+ "epoch": 72.35694822888283,
+ "grad_norm": 1.837959885597229,
+ "learning_rate": 3.7470810102516717e-06,
+ "loss": 0.0345,
+ "step": 26555
+ },
+ {
+ "epoch": 72.35967302452316,
+ "grad_norm": 1.1815286874771118,
+ "learning_rate": 3.7463923437706896e-06,
+ "loss": 0.0155,
+ "step": 26556
+ },
+ {
+ "epoch": 72.3623978201635,
+ "grad_norm": 1.0979021787643433,
+ "learning_rate": 3.7457037259926043e-06,
+ "loss": 0.0223,
+ "step": 26557
+ },
+ {
+ "epoch": 72.36512261580381,
+ "grad_norm": 2.836784839630127,
+ "learning_rate": 3.7450151569227743e-06,
+ "loss": 0.064,
+ "step": 26558
+ },
+ {
+ "epoch": 72.36784741144415,
+ "grad_norm": 0.8565800786018372,
+ "learning_rate": 3.744326636566569e-06,
+ "loss": 0.0121,
+ "step": 26559
+ },
+ {
+ "epoch": 72.37057220708446,
+ "grad_norm": 2.1051628589630127,
+ "learning_rate": 3.7436381649293463e-06,
+ "loss": 0.0157,
+ "step": 26560
+ },
+ {
+ "epoch": 72.3732970027248,
+ "grad_norm": 2.001040458679199,
+ "learning_rate": 3.7429497420164638e-06,
+ "loss": 0.0643,
+ "step": 26561
+ },
+ {
+ "epoch": 72.37602179836512,
+ "grad_norm": 1.4737789630889893,
+ "learning_rate": 3.742261367833292e-06,
+ "loss": 0.0226,
+ "step": 26562
+ },
+ {
+ "epoch": 72.37874659400545,
+ "grad_norm": 1.520230770111084,
+ "learning_rate": 3.741573042385187e-06,
+ "loss": 0.0169,
+ "step": 26563
+ },
+ {
+ "epoch": 72.38147138964578,
+ "grad_norm": 1.526981234550476,
+ "learning_rate": 3.740884765677509e-06,
+ "loss": 0.0228,
+ "step": 26564
+ },
+ {
+ "epoch": 72.3841961852861,
+ "grad_norm": 1.1340967416763306,
+ "learning_rate": 3.740196537715617e-06,
+ "loss": 0.0224,
+ "step": 26565
+ },
+ {
+ "epoch": 72.38692098092643,
+ "grad_norm": 1.0492087602615356,
+ "learning_rate": 3.7395083585048752e-06,
+ "loss": 0.0189,
+ "step": 26566
+ },
+ {
+ "epoch": 72.38964577656675,
+ "grad_norm": 1.4685566425323486,
+ "learning_rate": 3.7388202280506413e-06,
+ "loss": 0.0355,
+ "step": 26567
+ },
+ {
+ "epoch": 72.39237057220708,
+ "grad_norm": 1.5902657508850098,
+ "learning_rate": 3.738132146358273e-06,
+ "loss": 0.0905,
+ "step": 26568
+ },
+ {
+ "epoch": 72.39509536784742,
+ "grad_norm": 2.041111469268799,
+ "learning_rate": 3.7374441134331276e-06,
+ "loss": 0.1413,
+ "step": 26569
+ },
+ {
+ "epoch": 72.39782016348774,
+ "grad_norm": 1.7205959558486938,
+ "learning_rate": 3.7367561292805686e-06,
+ "loss": 0.0825,
+ "step": 26570
+ },
+ {
+ "epoch": 72.40054495912807,
+ "grad_norm": 1.6211968660354614,
+ "learning_rate": 3.736068193905952e-06,
+ "loss": 0.1419,
+ "step": 26571
+ },
+ {
+ "epoch": 72.40326975476839,
+ "grad_norm": 0.9341147541999817,
+ "learning_rate": 3.7353803073146343e-06,
+ "loss": 0.0082,
+ "step": 26572
+ },
+ {
+ "epoch": 72.40599455040872,
+ "grad_norm": 1.100446105003357,
+ "learning_rate": 3.73469246951197e-06,
+ "loss": 0.0108,
+ "step": 26573
+ },
+ {
+ "epoch": 72.40871934604904,
+ "grad_norm": 1.4797812700271606,
+ "learning_rate": 3.7340046805033224e-06,
+ "loss": 0.0547,
+ "step": 26574
+ },
+ {
+ "epoch": 72.41144414168937,
+ "grad_norm": 1.8613848686218262,
+ "learning_rate": 3.7333169402940407e-06,
+ "loss": 0.1752,
+ "step": 26575
+ },
+ {
+ "epoch": 72.4141689373297,
+ "grad_norm": 1.2197439670562744,
+ "learning_rate": 3.732629248889489e-06,
+ "loss": 0.0846,
+ "step": 26576
+ },
+ {
+ "epoch": 72.41689373297002,
+ "grad_norm": 1.2194530963897705,
+ "learning_rate": 3.7319416062950154e-06,
+ "loss": 0.0993,
+ "step": 26577
+ },
+ {
+ "epoch": 72.41961852861036,
+ "grad_norm": 1.159663438796997,
+ "learning_rate": 3.731254012515982e-06,
+ "loss": 0.0125,
+ "step": 26578
+ },
+ {
+ "epoch": 72.42234332425068,
+ "grad_norm": 3.736027717590332,
+ "learning_rate": 3.7305664675577415e-06,
+ "loss": 0.0205,
+ "step": 26579
+ },
+ {
+ "epoch": 72.42506811989101,
+ "grad_norm": 1.4624404907226562,
+ "learning_rate": 3.729878971425647e-06,
+ "loss": 0.0305,
+ "step": 26580
+ },
+ {
+ "epoch": 72.42779291553134,
+ "grad_norm": 0.8597292304039001,
+ "learning_rate": 3.72919152412505e-06,
+ "loss": 0.0073,
+ "step": 26581
+ },
+ {
+ "epoch": 72.43051771117166,
+ "grad_norm": 1.285150170326233,
+ "learning_rate": 3.7285041256613095e-06,
+ "loss": 0.0177,
+ "step": 26582
+ },
+ {
+ "epoch": 72.433242506812,
+ "grad_norm": 1.566025733947754,
+ "learning_rate": 3.7278167760397788e-06,
+ "loss": 0.0378,
+ "step": 26583
+ },
+ {
+ "epoch": 72.43596730245231,
+ "grad_norm": 1.2927839756011963,
+ "learning_rate": 3.7271294752658093e-06,
+ "loss": 0.0205,
+ "step": 26584
+ },
+ {
+ "epoch": 72.43869209809264,
+ "grad_norm": 1.3510619401931763,
+ "learning_rate": 3.7264422233447496e-06,
+ "loss": 0.0191,
+ "step": 26585
+ },
+ {
+ "epoch": 72.44141689373296,
+ "grad_norm": 1.9315438270568848,
+ "learning_rate": 3.7257550202819593e-06,
+ "loss": 0.233,
+ "step": 26586
+ },
+ {
+ "epoch": 72.4441416893733,
+ "grad_norm": 1.1730116605758667,
+ "learning_rate": 3.7250678660827867e-06,
+ "loss": 0.0126,
+ "step": 26587
+ },
+ {
+ "epoch": 72.44686648501363,
+ "grad_norm": 1.337045431137085,
+ "learning_rate": 3.724380760752584e-06,
+ "loss": 0.0185,
+ "step": 26588
+ },
+ {
+ "epoch": 72.44959128065395,
+ "grad_norm": 1.371043086051941,
+ "learning_rate": 3.723693704296698e-06,
+ "loss": 0.0362,
+ "step": 26589
+ },
+ {
+ "epoch": 72.45231607629428,
+ "grad_norm": 5.724576473236084,
+ "learning_rate": 3.723006696720487e-06,
+ "loss": 0.062,
+ "step": 26590
+ },
+ {
+ "epoch": 72.4550408719346,
+ "grad_norm": 1.0939958095550537,
+ "learning_rate": 3.7223197380292986e-06,
+ "loss": 0.0144,
+ "step": 26591
+ },
+ {
+ "epoch": 72.45776566757493,
+ "grad_norm": 1.6812222003936768,
+ "learning_rate": 3.7216328282284776e-06,
+ "loss": 0.1397,
+ "step": 26592
+ },
+ {
+ "epoch": 72.46049046321527,
+ "grad_norm": 1.5629929304122925,
+ "learning_rate": 3.7209459673233817e-06,
+ "loss": 0.0104,
+ "step": 26593
+ },
+ {
+ "epoch": 72.46321525885558,
+ "grad_norm": 1.1139012575149536,
+ "learning_rate": 3.7202591553193534e-06,
+ "loss": 0.0156,
+ "step": 26594
+ },
+ {
+ "epoch": 72.46594005449592,
+ "grad_norm": 1.5821703672409058,
+ "learning_rate": 3.7195723922217487e-06,
+ "loss": 0.018,
+ "step": 26595
+ },
+ {
+ "epoch": 72.46866485013624,
+ "grad_norm": 2.312894344329834,
+ "learning_rate": 3.7188856780359117e-06,
+ "loss": 0.0131,
+ "step": 26596
+ },
+ {
+ "epoch": 72.47138964577657,
+ "grad_norm": 7.5756425857543945,
+ "learning_rate": 3.718199012767186e-06,
+ "loss": 0.2418,
+ "step": 26597
+ },
+ {
+ "epoch": 72.47411444141689,
+ "grad_norm": 1.1819591522216797,
+ "learning_rate": 3.7175123964209293e-06,
+ "loss": 0.0163,
+ "step": 26598
+ },
+ {
+ "epoch": 72.47683923705722,
+ "grad_norm": 1.370083212852478,
+ "learning_rate": 3.716825829002484e-06,
+ "loss": 0.0225,
+ "step": 26599
+ },
+ {
+ "epoch": 72.47956403269755,
+ "grad_norm": 1.8114451169967651,
+ "learning_rate": 3.716139310517196e-06,
+ "loss": 0.0445,
+ "step": 26600
+ },
+ {
+ "epoch": 72.48228882833787,
+ "grad_norm": 1.0451472997665405,
+ "learning_rate": 3.7154528409704095e-06,
+ "loss": 0.0831,
+ "step": 26601
+ },
+ {
+ "epoch": 72.4850136239782,
+ "grad_norm": 1.3100852966308594,
+ "learning_rate": 3.714766420367478e-06,
+ "loss": 0.0159,
+ "step": 26602
+ },
+ {
+ "epoch": 72.48773841961852,
+ "grad_norm": 1.7052286863327026,
+ "learning_rate": 3.7140800487137428e-06,
+ "loss": 0.03,
+ "step": 26603
+ },
+ {
+ "epoch": 72.49046321525886,
+ "grad_norm": 1.0998280048370361,
+ "learning_rate": 3.71339372601455e-06,
+ "loss": 0.0134,
+ "step": 26604
+ },
+ {
+ "epoch": 72.49318801089919,
+ "grad_norm": 1.3751780986785889,
+ "learning_rate": 3.7127074522752416e-06,
+ "loss": 0.0418,
+ "step": 26605
+ },
+ {
+ "epoch": 72.49591280653951,
+ "grad_norm": 1.7006580829620361,
+ "learning_rate": 3.712021227501168e-06,
+ "loss": 0.0244,
+ "step": 26606
+ },
+ {
+ "epoch": 72.49863760217984,
+ "grad_norm": 2.216249465942383,
+ "learning_rate": 3.711335051697671e-06,
+ "loss": 0.0737,
+ "step": 26607
+ },
+ {
+ "epoch": 72.50136239782016,
+ "grad_norm": 1.5671714544296265,
+ "learning_rate": 3.710648924870094e-06,
+ "loss": 0.0191,
+ "step": 26608
+ },
+ {
+ "epoch": 72.50408719346049,
+ "grad_norm": 1.2457261085510254,
+ "learning_rate": 3.709962847023777e-06,
+ "loss": 0.0574,
+ "step": 26609
+ },
+ {
+ "epoch": 72.50681198910081,
+ "grad_norm": 1.5593684911727905,
+ "learning_rate": 3.709276818164067e-06,
+ "loss": 0.1644,
+ "step": 26610
+ },
+ {
+ "epoch": 72.50953678474114,
+ "grad_norm": 1.0883910655975342,
+ "learning_rate": 3.708590838296311e-06,
+ "loss": 0.0094,
+ "step": 26611
+ },
+ {
+ "epoch": 72.51226158038148,
+ "grad_norm": 1.0730687379837036,
+ "learning_rate": 3.7079049074258465e-06,
+ "loss": 0.021,
+ "step": 26612
+ },
+ {
+ "epoch": 72.5149863760218,
+ "grad_norm": 0.7400323152542114,
+ "learning_rate": 3.7072190255580153e-06,
+ "loss": 0.0099,
+ "step": 26613
+ },
+ {
+ "epoch": 72.51771117166213,
+ "grad_norm": 1.7304209470748901,
+ "learning_rate": 3.7065331926981573e-06,
+ "loss": 0.0153,
+ "step": 26614
+ },
+ {
+ "epoch": 72.52043596730245,
+ "grad_norm": 2.0706264972686768,
+ "learning_rate": 3.7058474088516194e-06,
+ "loss": 0.0093,
+ "step": 26615
+ },
+ {
+ "epoch": 72.52316076294278,
+ "grad_norm": 1.4974820613861084,
+ "learning_rate": 3.7051616740237396e-06,
+ "loss": 0.0355,
+ "step": 26616
+ },
+ {
+ "epoch": 72.52588555858311,
+ "grad_norm": 1.0954111814498901,
+ "learning_rate": 3.7044759882198576e-06,
+ "loss": 0.016,
+ "step": 26617
+ },
+ {
+ "epoch": 72.52861035422343,
+ "grad_norm": 1.4915812015533447,
+ "learning_rate": 3.703790351445311e-06,
+ "loss": 0.0677,
+ "step": 26618
+ },
+ {
+ "epoch": 72.53133514986376,
+ "grad_norm": 1.2662367820739746,
+ "learning_rate": 3.703104763705446e-06,
+ "loss": 0.0177,
+ "step": 26619
+ },
+ {
+ "epoch": 72.53405994550408,
+ "grad_norm": 1.9477989673614502,
+ "learning_rate": 3.702419225005598e-06,
+ "loss": 0.1033,
+ "step": 26620
+ },
+ {
+ "epoch": 72.53678474114442,
+ "grad_norm": 1.6914021968841553,
+ "learning_rate": 3.701733735351103e-06,
+ "loss": 0.0215,
+ "step": 26621
+ },
+ {
+ "epoch": 72.53950953678473,
+ "grad_norm": 0.8301833271980286,
+ "learning_rate": 3.701048294747307e-06,
+ "loss": 0.0094,
+ "step": 26622
+ },
+ {
+ "epoch": 72.54223433242507,
+ "grad_norm": 2.1200616359710693,
+ "learning_rate": 3.7003629031995436e-06,
+ "loss": 0.086,
+ "step": 26623
+ },
+ {
+ "epoch": 72.5449591280654,
+ "grad_norm": 1.4668095111846924,
+ "learning_rate": 3.6996775607131507e-06,
+ "loss": 0.0137,
+ "step": 26624
+ },
+ {
+ "epoch": 72.54768392370572,
+ "grad_norm": 1.603123664855957,
+ "learning_rate": 3.698992267293464e-06,
+ "loss": 0.0506,
+ "step": 26625
+ },
+ {
+ "epoch": 72.55040871934605,
+ "grad_norm": 1.2159594297409058,
+ "learning_rate": 3.6983070229458216e-06,
+ "loss": 0.0148,
+ "step": 26626
+ },
+ {
+ "epoch": 72.55313351498637,
+ "grad_norm": 1.153887152671814,
+ "learning_rate": 3.6976218276755647e-06,
+ "loss": 0.013,
+ "step": 26627
+ },
+ {
+ "epoch": 72.5558583106267,
+ "grad_norm": 0.7884747982025146,
+ "learning_rate": 3.696936681488027e-06,
+ "loss": 0.0081,
+ "step": 26628
+ },
+ {
+ "epoch": 72.55858310626704,
+ "grad_norm": 1.1397624015808105,
+ "learning_rate": 3.696251584388544e-06,
+ "loss": 0.0087,
+ "step": 26629
+ },
+ {
+ "epoch": 72.56130790190736,
+ "grad_norm": 2.1059980392456055,
+ "learning_rate": 3.6955665363824455e-06,
+ "loss": 0.043,
+ "step": 26630
+ },
+ {
+ "epoch": 72.56403269754769,
+ "grad_norm": 1.2217087745666504,
+ "learning_rate": 3.694881537475077e-06,
+ "loss": 0.0143,
+ "step": 26631
+ },
+ {
+ "epoch": 72.566757493188,
+ "grad_norm": 1.2820243835449219,
+ "learning_rate": 3.6941965876717667e-06,
+ "loss": 0.0178,
+ "step": 26632
+ },
+ {
+ "epoch": 72.56948228882834,
+ "grad_norm": 1.5134896039962769,
+ "learning_rate": 3.693511686977852e-06,
+ "loss": 0.0359,
+ "step": 26633
+ },
+ {
+ "epoch": 72.57220708446866,
+ "grad_norm": 1.502055287361145,
+ "learning_rate": 3.692826835398661e-06,
+ "loss": 0.0624,
+ "step": 26634
+ },
+ {
+ "epoch": 72.57493188010899,
+ "grad_norm": 1.6316510438919067,
+ "learning_rate": 3.692142032939535e-06,
+ "loss": 0.0898,
+ "step": 26635
+ },
+ {
+ "epoch": 72.57765667574932,
+ "grad_norm": 0.809296190738678,
+ "learning_rate": 3.6914572796058033e-06,
+ "loss": 0.0042,
+ "step": 26636
+ },
+ {
+ "epoch": 72.58038147138964,
+ "grad_norm": 1.202850580215454,
+ "learning_rate": 3.6907725754028e-06,
+ "loss": 0.0145,
+ "step": 26637
+ },
+ {
+ "epoch": 72.58310626702998,
+ "grad_norm": 1.5116832256317139,
+ "learning_rate": 3.6900879203358543e-06,
+ "loss": 0.0256,
+ "step": 26638
+ },
+ {
+ "epoch": 72.5858310626703,
+ "grad_norm": 2.499264717102051,
+ "learning_rate": 3.6894033144103025e-06,
+ "loss": 0.164,
+ "step": 26639
+ },
+ {
+ "epoch": 72.58855585831063,
+ "grad_norm": 1.1471951007843018,
+ "learning_rate": 3.6887187576314755e-06,
+ "loss": 0.0114,
+ "step": 26640
+ },
+ {
+ "epoch": 72.59128065395096,
+ "grad_norm": 0.9753807783126831,
+ "learning_rate": 3.688034250004704e-06,
+ "loss": 0.01,
+ "step": 26641
+ },
+ {
+ "epoch": 72.59400544959128,
+ "grad_norm": 1.4688806533813477,
+ "learning_rate": 3.687349791535314e-06,
+ "loss": 0.0154,
+ "step": 26642
+ },
+ {
+ "epoch": 72.59673024523161,
+ "grad_norm": 1.6745823621749878,
+ "learning_rate": 3.6866653822286457e-06,
+ "loss": 0.0256,
+ "step": 26643
+ },
+ {
+ "epoch": 72.59945504087193,
+ "grad_norm": 1.0864595174789429,
+ "learning_rate": 3.685981022090019e-06,
+ "loss": 0.0125,
+ "step": 26644
+ },
+ {
+ "epoch": 72.60217983651226,
+ "grad_norm": 0.8755474090576172,
+ "learning_rate": 3.685296711124774e-06,
+ "loss": 0.0122,
+ "step": 26645
+ },
+ {
+ "epoch": 72.60490463215258,
+ "grad_norm": 1.9600934982299805,
+ "learning_rate": 3.684612449338231e-06,
+ "loss": 0.0707,
+ "step": 26646
+ },
+ {
+ "epoch": 72.60762942779292,
+ "grad_norm": 1.440081000328064,
+ "learning_rate": 3.6839282367357256e-06,
+ "loss": 0.0221,
+ "step": 26647
+ },
+ {
+ "epoch": 72.61035422343325,
+ "grad_norm": 1.6023520231246948,
+ "learning_rate": 3.683244073322584e-06,
+ "loss": 0.0183,
+ "step": 26648
+ },
+ {
+ "epoch": 72.61307901907357,
+ "grad_norm": 1.788483738899231,
+ "learning_rate": 3.682559959104135e-06,
+ "loss": 0.0154,
+ "step": 26649
+ },
+ {
+ "epoch": 72.6158038147139,
+ "grad_norm": 1.112752914428711,
+ "learning_rate": 3.6818758940857014e-06,
+ "loss": 0.0097,
+ "step": 26650
+ },
+ {
+ "epoch": 72.61852861035422,
+ "grad_norm": 1.1088306903839111,
+ "learning_rate": 3.6811918782726197e-06,
+ "loss": 0.0192,
+ "step": 26651
+ },
+ {
+ "epoch": 72.62125340599455,
+ "grad_norm": 1.1990773677825928,
+ "learning_rate": 3.680507911670211e-06,
+ "loss": 0.0119,
+ "step": 26652
+ },
+ {
+ "epoch": 72.62397820163488,
+ "grad_norm": 2.2689359188079834,
+ "learning_rate": 3.6798239942838042e-06,
+ "loss": 0.015,
+ "step": 26653
+ },
+ {
+ "epoch": 72.6267029972752,
+ "grad_norm": 1.8849736452102661,
+ "learning_rate": 3.679140126118721e-06,
+ "loss": 0.0221,
+ "step": 26654
+ },
+ {
+ "epoch": 72.62942779291554,
+ "grad_norm": 1.1412566900253296,
+ "learning_rate": 3.6784563071802935e-06,
+ "loss": 0.0205,
+ "step": 26655
+ },
+ {
+ "epoch": 72.63215258855585,
+ "grad_norm": 1.2242122888565063,
+ "learning_rate": 3.677772537473846e-06,
+ "loss": 0.0159,
+ "step": 26656
+ },
+ {
+ "epoch": 72.63487738419619,
+ "grad_norm": 1.4627066850662231,
+ "learning_rate": 3.6770888170047026e-06,
+ "loss": 0.0199,
+ "step": 26657
+ },
+ {
+ "epoch": 72.6376021798365,
+ "grad_norm": 0.9093347787857056,
+ "learning_rate": 3.6764051457781836e-06,
+ "loss": 0.0095,
+ "step": 26658
+ },
+ {
+ "epoch": 72.64032697547684,
+ "grad_norm": 0.929216206073761,
+ "learning_rate": 3.6757215237996225e-06,
+ "loss": 0.0109,
+ "step": 26659
+ },
+ {
+ "epoch": 72.64305177111717,
+ "grad_norm": 1.9227310419082642,
+ "learning_rate": 3.6750379510743374e-06,
+ "loss": 0.0179,
+ "step": 26660
+ },
+ {
+ "epoch": 72.64577656675749,
+ "grad_norm": 1.3188846111297607,
+ "learning_rate": 3.6743544276076503e-06,
+ "loss": 0.0221,
+ "step": 26661
+ },
+ {
+ "epoch": 72.64850136239782,
+ "grad_norm": 1.806088924407959,
+ "learning_rate": 3.6736709534048877e-06,
+ "loss": 0.0608,
+ "step": 26662
+ },
+ {
+ "epoch": 72.65122615803814,
+ "grad_norm": 1.8725361824035645,
+ "learning_rate": 3.6729875284713746e-06,
+ "loss": 0.1025,
+ "step": 26663
+ },
+ {
+ "epoch": 72.65395095367847,
+ "grad_norm": 1.2106900215148926,
+ "learning_rate": 3.6723041528124314e-06,
+ "loss": 0.0536,
+ "step": 26664
+ },
+ {
+ "epoch": 72.65667574931881,
+ "grad_norm": 1.4202828407287598,
+ "learning_rate": 3.67162082643338e-06,
+ "loss": 0.0202,
+ "step": 26665
+ },
+ {
+ "epoch": 72.65940054495913,
+ "grad_norm": 1.1170728206634521,
+ "learning_rate": 3.6709375493395383e-06,
+ "loss": 0.0767,
+ "step": 26666
+ },
+ {
+ "epoch": 72.66212534059946,
+ "grad_norm": 0.8995675444602966,
+ "learning_rate": 3.670254321536235e-06,
+ "loss": 0.0093,
+ "step": 26667
+ },
+ {
+ "epoch": 72.66485013623978,
+ "grad_norm": 1.548746943473816,
+ "learning_rate": 3.6695711430287873e-06,
+ "loss": 0.0165,
+ "step": 26668
+ },
+ {
+ "epoch": 72.66757493188011,
+ "grad_norm": 1.1536574363708496,
+ "learning_rate": 3.6688880138225158e-06,
+ "loss": 0.0178,
+ "step": 26669
+ },
+ {
+ "epoch": 72.67029972752043,
+ "grad_norm": 1.2492936849594116,
+ "learning_rate": 3.6682049339227364e-06,
+ "loss": 0.0412,
+ "step": 26670
+ },
+ {
+ "epoch": 72.67302452316076,
+ "grad_norm": 1.486220359802246,
+ "learning_rate": 3.6675219033347775e-06,
+ "loss": 0.0519,
+ "step": 26671
+ },
+ {
+ "epoch": 72.6757493188011,
+ "grad_norm": 2.091867685317993,
+ "learning_rate": 3.6668389220639545e-06,
+ "loss": 0.0369,
+ "step": 26672
+ },
+ {
+ "epoch": 72.67847411444141,
+ "grad_norm": 1.0814287662506104,
+ "learning_rate": 3.6661559901155852e-06,
+ "loss": 0.0173,
+ "step": 26673
+ },
+ {
+ "epoch": 72.68119891008175,
+ "grad_norm": 1.3792240619659424,
+ "learning_rate": 3.6654731074949856e-06,
+ "loss": 0.0237,
+ "step": 26674
+ },
+ {
+ "epoch": 72.68392370572207,
+ "grad_norm": 1.9631794691085815,
+ "learning_rate": 3.6647902742074825e-06,
+ "loss": 0.0865,
+ "step": 26675
+ },
+ {
+ "epoch": 72.6866485013624,
+ "grad_norm": 0.5989972949028015,
+ "learning_rate": 3.6641074902583875e-06,
+ "loss": 0.0046,
+ "step": 26676
+ },
+ {
+ "epoch": 72.68937329700273,
+ "grad_norm": 1.2921044826507568,
+ "learning_rate": 3.66342475565302e-06,
+ "loss": 0.0229,
+ "step": 26677
+ },
+ {
+ "epoch": 72.69209809264305,
+ "grad_norm": 1.2495017051696777,
+ "learning_rate": 3.662742070396693e-06,
+ "loss": 0.0182,
+ "step": 26678
+ },
+ {
+ "epoch": 72.69482288828338,
+ "grad_norm": 1.665451169013977,
+ "learning_rate": 3.662059434494727e-06,
+ "loss": 0.0336,
+ "step": 26679
+ },
+ {
+ "epoch": 72.6975476839237,
+ "grad_norm": 1.4212273359298706,
+ "learning_rate": 3.6613768479524413e-06,
+ "loss": 0.1002,
+ "step": 26680
+ },
+ {
+ "epoch": 72.70027247956403,
+ "grad_norm": 4.091477394104004,
+ "learning_rate": 3.660694310775148e-06,
+ "loss": 0.089,
+ "step": 26681
+ },
+ {
+ "epoch": 72.70299727520435,
+ "grad_norm": 1.1736485958099365,
+ "learning_rate": 3.6600118229681646e-06,
+ "loss": 0.0149,
+ "step": 26682
+ },
+ {
+ "epoch": 72.70572207084469,
+ "grad_norm": 1.158926010131836,
+ "learning_rate": 3.6593293845368018e-06,
+ "loss": 0.0126,
+ "step": 26683
+ },
+ {
+ "epoch": 72.70844686648502,
+ "grad_norm": 1.9123491048812866,
+ "learning_rate": 3.65864699548638e-06,
+ "loss": 0.0483,
+ "step": 26684
+ },
+ {
+ "epoch": 72.71117166212534,
+ "grad_norm": 1.705935001373291,
+ "learning_rate": 3.6579646558222117e-06,
+ "loss": 0.0408,
+ "step": 26685
+ },
+ {
+ "epoch": 72.71389645776567,
+ "grad_norm": 1.1045829057693481,
+ "learning_rate": 3.657282365549607e-06,
+ "loss": 0.1388,
+ "step": 26686
+ },
+ {
+ "epoch": 72.71662125340599,
+ "grad_norm": 1.5668622255325317,
+ "learning_rate": 3.656600124673887e-06,
+ "loss": 0.0181,
+ "step": 26687
+ },
+ {
+ "epoch": 72.71934604904632,
+ "grad_norm": 1.198077917098999,
+ "learning_rate": 3.6559179332003603e-06,
+ "loss": 0.0159,
+ "step": 26688
+ },
+ {
+ "epoch": 72.72207084468666,
+ "grad_norm": 1.2297080755233765,
+ "learning_rate": 3.6552357911343416e-06,
+ "loss": 0.0959,
+ "step": 26689
+ },
+ {
+ "epoch": 72.72479564032697,
+ "grad_norm": 1.5481609106063843,
+ "learning_rate": 3.654553698481138e-06,
+ "loss": 0.0373,
+ "step": 26690
+ },
+ {
+ "epoch": 72.7275204359673,
+ "grad_norm": 1.1751612424850464,
+ "learning_rate": 3.6538716552460708e-06,
+ "loss": 0.0104,
+ "step": 26691
+ },
+ {
+ "epoch": 72.73024523160763,
+ "grad_norm": 1.5627214908599854,
+ "learning_rate": 3.6531896614344454e-06,
+ "loss": 0.0675,
+ "step": 26692
+ },
+ {
+ "epoch": 72.73297002724796,
+ "grad_norm": 1.2000919580459595,
+ "learning_rate": 3.6525077170515755e-06,
+ "loss": 0.0222,
+ "step": 26693
+ },
+ {
+ "epoch": 72.73569482288828,
+ "grad_norm": 1.4805686473846436,
+ "learning_rate": 3.651825822102767e-06,
+ "loss": 0.0218,
+ "step": 26694
+ },
+ {
+ "epoch": 72.73841961852861,
+ "grad_norm": 1.269689917564392,
+ "learning_rate": 3.6511439765933355e-06,
+ "loss": 0.0134,
+ "step": 26695
+ },
+ {
+ "epoch": 72.74114441416894,
+ "grad_norm": 1.8185218572616577,
+ "learning_rate": 3.6504621805285935e-06,
+ "loss": 0.0181,
+ "step": 26696
+ },
+ {
+ "epoch": 72.74386920980926,
+ "grad_norm": 8.144083023071289,
+ "learning_rate": 3.6497804339138463e-06,
+ "loss": 0.0175,
+ "step": 26697
+ },
+ {
+ "epoch": 72.7465940054496,
+ "grad_norm": 1.6092777252197266,
+ "learning_rate": 3.649098736754406e-06,
+ "loss": 0.1561,
+ "step": 26698
+ },
+ {
+ "epoch": 72.74931880108991,
+ "grad_norm": 1.543379545211792,
+ "learning_rate": 3.648417089055577e-06,
+ "loss": 0.0237,
+ "step": 26699
+ },
+ {
+ "epoch": 72.75204359673025,
+ "grad_norm": 1.2310620546340942,
+ "learning_rate": 3.647735490822675e-06,
+ "loss": 0.0324,
+ "step": 26700
+ },
+ {
+ "epoch": 72.75476839237058,
+ "grad_norm": 1.6527758836746216,
+ "learning_rate": 3.647053942061004e-06,
+ "loss": 0.0389,
+ "step": 26701
+ },
+ {
+ "epoch": 72.7574931880109,
+ "grad_norm": 2.1396985054016113,
+ "learning_rate": 3.646372442775873e-06,
+ "loss": 0.016,
+ "step": 26702
+ },
+ {
+ "epoch": 72.76021798365123,
+ "grad_norm": 1.0753381252288818,
+ "learning_rate": 3.645690992972585e-06,
+ "loss": 0.013,
+ "step": 26703
+ },
+ {
+ "epoch": 72.76294277929155,
+ "grad_norm": 1.118537187576294,
+ "learning_rate": 3.6450095926564542e-06,
+ "loss": 0.113,
+ "step": 26704
+ },
+ {
+ "epoch": 72.76566757493188,
+ "grad_norm": 1.7880208492279053,
+ "learning_rate": 3.6443282418327853e-06,
+ "loss": 0.0671,
+ "step": 26705
+ },
+ {
+ "epoch": 72.7683923705722,
+ "grad_norm": 1.126508355140686,
+ "learning_rate": 3.6436469405068787e-06,
+ "loss": 0.0184,
+ "step": 26706
+ },
+ {
+ "epoch": 72.77111716621253,
+ "grad_norm": 1.9400789737701416,
+ "learning_rate": 3.642965688684049e-06,
+ "loss": 0.0531,
+ "step": 26707
+ },
+ {
+ "epoch": 72.77384196185287,
+ "grad_norm": 1.711106300354004,
+ "learning_rate": 3.6422844863695983e-06,
+ "loss": 0.0165,
+ "step": 26708
+ },
+ {
+ "epoch": 72.77656675749319,
+ "grad_norm": 1.3909517526626587,
+ "learning_rate": 3.6416033335688306e-06,
+ "loss": 0.024,
+ "step": 26709
+ },
+ {
+ "epoch": 72.77929155313352,
+ "grad_norm": 2.0400516986846924,
+ "learning_rate": 3.6409222302870494e-06,
+ "loss": 0.0555,
+ "step": 26710
+ },
+ {
+ "epoch": 72.78201634877384,
+ "grad_norm": 1.0134116411209106,
+ "learning_rate": 3.640241176529563e-06,
+ "loss": 0.0098,
+ "step": 26711
+ },
+ {
+ "epoch": 72.78474114441417,
+ "grad_norm": 1.4371888637542725,
+ "learning_rate": 3.6395601723016714e-06,
+ "loss": 0.0221,
+ "step": 26712
+ },
+ {
+ "epoch": 72.7874659400545,
+ "grad_norm": 1.0813937187194824,
+ "learning_rate": 3.638879217608684e-06,
+ "loss": 0.0127,
+ "step": 26713
+ },
+ {
+ "epoch": 72.79019073569482,
+ "grad_norm": 1.232323408126831,
+ "learning_rate": 3.6381983124558994e-06,
+ "loss": 0.0179,
+ "step": 26714
+ },
+ {
+ "epoch": 72.79291553133515,
+ "grad_norm": 1.3353933095932007,
+ "learning_rate": 3.637517456848619e-06,
+ "loss": 0.0164,
+ "step": 26715
+ },
+ {
+ "epoch": 72.79564032697547,
+ "grad_norm": 5.775277614593506,
+ "learning_rate": 3.6368366507921514e-06,
+ "loss": 0.023,
+ "step": 26716
+ },
+ {
+ "epoch": 72.7983651226158,
+ "grad_norm": 0.9398760199546814,
+ "learning_rate": 3.636155894291795e-06,
+ "loss": 0.0113,
+ "step": 26717
+ },
+ {
+ "epoch": 72.80108991825612,
+ "grad_norm": 0.8311593532562256,
+ "learning_rate": 3.6354751873528515e-06,
+ "loss": 0.0169,
+ "step": 26718
+ },
+ {
+ "epoch": 72.80381471389646,
+ "grad_norm": 0.9447697997093201,
+ "learning_rate": 3.634794529980619e-06,
+ "loss": 0.0134,
+ "step": 26719
+ },
+ {
+ "epoch": 72.80653950953679,
+ "grad_norm": 1.8910884857177734,
+ "learning_rate": 3.6341139221804057e-06,
+ "loss": 0.017,
+ "step": 26720
+ },
+ {
+ "epoch": 72.80926430517711,
+ "grad_norm": 0.8611125349998474,
+ "learning_rate": 3.633433363957507e-06,
+ "loss": 0.013,
+ "step": 26721
+ },
+ {
+ "epoch": 72.81198910081744,
+ "grad_norm": 1.6512869596481323,
+ "learning_rate": 3.6327528553172252e-06,
+ "loss": 0.0175,
+ "step": 26722
+ },
+ {
+ "epoch": 72.81471389645776,
+ "grad_norm": 1.2397080659866333,
+ "learning_rate": 3.6320723962648564e-06,
+ "loss": 0.0389,
+ "step": 26723
+ },
+ {
+ "epoch": 72.8174386920981,
+ "grad_norm": 1.5890799760818481,
+ "learning_rate": 3.631391986805706e-06,
+ "loss": 0.0349,
+ "step": 26724
+ },
+ {
+ "epoch": 72.82016348773843,
+ "grad_norm": 1.9727102518081665,
+ "learning_rate": 3.6307116269450692e-06,
+ "loss": 0.0473,
+ "step": 26725
+ },
+ {
+ "epoch": 72.82288828337875,
+ "grad_norm": 1.607129454612732,
+ "learning_rate": 3.630031316688245e-06,
+ "loss": 0.0206,
+ "step": 26726
+ },
+ {
+ "epoch": 72.82561307901908,
+ "grad_norm": 1.03040611743927,
+ "learning_rate": 3.62935105604053e-06,
+ "loss": 0.0241,
+ "step": 26727
+ },
+ {
+ "epoch": 72.8283378746594,
+ "grad_norm": 1.703018069267273,
+ "learning_rate": 3.628670845007226e-06,
+ "loss": 0.0728,
+ "step": 26728
+ },
+ {
+ "epoch": 72.83106267029973,
+ "grad_norm": 1.1430740356445312,
+ "learning_rate": 3.6279906835936287e-06,
+ "loss": 0.0146,
+ "step": 26729
+ },
+ {
+ "epoch": 72.83378746594005,
+ "grad_norm": 1.589750051498413,
+ "learning_rate": 3.627310571805032e-06,
+ "loss": 0.0268,
+ "step": 26730
+ },
+ {
+ "epoch": 72.83651226158038,
+ "grad_norm": 1.3222160339355469,
+ "learning_rate": 3.626630509646735e-06,
+ "loss": 0.038,
+ "step": 26731
+ },
+ {
+ "epoch": 72.83923705722071,
+ "grad_norm": 1.2597460746765137,
+ "learning_rate": 3.625950497124038e-06,
+ "loss": 0.0114,
+ "step": 26732
+ },
+ {
+ "epoch": 72.84196185286103,
+ "grad_norm": 1.3058810234069824,
+ "learning_rate": 3.6252705342422335e-06,
+ "loss": 0.0228,
+ "step": 26733
+ },
+ {
+ "epoch": 72.84468664850137,
+ "grad_norm": 1.8116321563720703,
+ "learning_rate": 3.6245906210066163e-06,
+ "loss": 0.017,
+ "step": 26734
+ },
+ {
+ "epoch": 72.84741144414168,
+ "grad_norm": 1.9625300168991089,
+ "learning_rate": 3.6239107574224785e-06,
+ "loss": 0.0262,
+ "step": 26735
+ },
+ {
+ "epoch": 72.85013623978202,
+ "grad_norm": 1.6400517225265503,
+ "learning_rate": 3.623230943495123e-06,
+ "loss": 0.0237,
+ "step": 26736
+ },
+ {
+ "epoch": 72.85286103542235,
+ "grad_norm": 1.6487494707107544,
+ "learning_rate": 3.6225511792298397e-06,
+ "loss": 0.0296,
+ "step": 26737
+ },
+ {
+ "epoch": 72.85558583106267,
+ "grad_norm": 1.4482122659683228,
+ "learning_rate": 3.621871464631922e-06,
+ "loss": 0.0275,
+ "step": 26738
+ },
+ {
+ "epoch": 72.858310626703,
+ "grad_norm": 1.3421964645385742,
+ "learning_rate": 3.6211917997066603e-06,
+ "loss": 0.0156,
+ "step": 26739
+ },
+ {
+ "epoch": 72.86103542234332,
+ "grad_norm": 1.6533498764038086,
+ "learning_rate": 3.620512184459356e-06,
+ "loss": 0.0306,
+ "step": 26740
+ },
+ {
+ "epoch": 72.86376021798365,
+ "grad_norm": 1.5000501871109009,
+ "learning_rate": 3.619832618895297e-06,
+ "loss": 0.1072,
+ "step": 26741
+ },
+ {
+ "epoch": 72.86648501362397,
+ "grad_norm": 2.551185369491577,
+ "learning_rate": 3.619153103019776e-06,
+ "loss": 0.132,
+ "step": 26742
+ },
+ {
+ "epoch": 72.8692098092643,
+ "grad_norm": 1.2805503606796265,
+ "learning_rate": 3.6184736368380813e-06,
+ "loss": 0.1171,
+ "step": 26743
+ },
+ {
+ "epoch": 72.87193460490464,
+ "grad_norm": 1.9286251068115234,
+ "learning_rate": 3.6177942203555125e-06,
+ "loss": 0.0263,
+ "step": 26744
+ },
+ {
+ "epoch": 72.87465940054496,
+ "grad_norm": 1.1950383186340332,
+ "learning_rate": 3.617114853577357e-06,
+ "loss": 0.0222,
+ "step": 26745
+ },
+ {
+ "epoch": 72.87738419618529,
+ "grad_norm": 1.3556121587753296,
+ "learning_rate": 3.6164355365089067e-06,
+ "loss": 0.0124,
+ "step": 26746
+ },
+ {
+ "epoch": 72.88010899182561,
+ "grad_norm": 0.72063148021698,
+ "learning_rate": 3.615756269155446e-06,
+ "loss": 0.0072,
+ "step": 26747
+ },
+ {
+ "epoch": 72.88283378746594,
+ "grad_norm": 1.322452187538147,
+ "learning_rate": 3.61507705152227e-06,
+ "loss": 0.0568,
+ "step": 26748
+ },
+ {
+ "epoch": 72.88555858310627,
+ "grad_norm": 1.1244663000106812,
+ "learning_rate": 3.6143978836146733e-06,
+ "loss": 0.0112,
+ "step": 26749
+ },
+ {
+ "epoch": 72.88828337874659,
+ "grad_norm": 1.4606208801269531,
+ "learning_rate": 3.613718765437939e-06,
+ "loss": 0.0259,
+ "step": 26750
+ },
+ {
+ "epoch": 72.89100817438693,
+ "grad_norm": 1.935965895652771,
+ "learning_rate": 3.6130396969973546e-06,
+ "loss": 0.0586,
+ "step": 26751
+ },
+ {
+ "epoch": 72.89373297002724,
+ "grad_norm": 1.4016751050949097,
+ "learning_rate": 3.6123606782982147e-06,
+ "loss": 0.0208,
+ "step": 26752
+ },
+ {
+ "epoch": 72.89645776566758,
+ "grad_norm": 1.440484881401062,
+ "learning_rate": 3.6116817093458034e-06,
+ "loss": 0.0158,
+ "step": 26753
+ },
+ {
+ "epoch": 72.8991825613079,
+ "grad_norm": 1.9753018617630005,
+ "learning_rate": 3.61100279014541e-06,
+ "loss": 0.0434,
+ "step": 26754
+ },
+ {
+ "epoch": 72.90190735694823,
+ "grad_norm": 1.474440097808838,
+ "learning_rate": 3.610323920702319e-06,
+ "loss": 0.0252,
+ "step": 26755
+ },
+ {
+ "epoch": 72.90463215258856,
+ "grad_norm": 1.6556968688964844,
+ "learning_rate": 3.6096451010218214e-06,
+ "loss": 0.1264,
+ "step": 26756
+ },
+ {
+ "epoch": 72.90735694822888,
+ "grad_norm": 4.656721115112305,
+ "learning_rate": 3.6089663311092025e-06,
+ "loss": 0.0276,
+ "step": 26757
+ },
+ {
+ "epoch": 72.91008174386921,
+ "grad_norm": 1.3128917217254639,
+ "learning_rate": 3.6082876109697485e-06,
+ "loss": 0.0158,
+ "step": 26758
+ },
+ {
+ "epoch": 72.91280653950953,
+ "grad_norm": 0.9220163226127625,
+ "learning_rate": 3.607608940608741e-06,
+ "loss": 0.0085,
+ "step": 26759
+ },
+ {
+ "epoch": 72.91553133514986,
+ "grad_norm": 1.270558476448059,
+ "learning_rate": 3.606930320031474e-06,
+ "loss": 0.0158,
+ "step": 26760
+ },
+ {
+ "epoch": 72.9182561307902,
+ "grad_norm": 1.4885302782058716,
+ "learning_rate": 3.606251749243227e-06,
+ "loss": 0.0236,
+ "step": 26761
+ },
+ {
+ "epoch": 72.92098092643052,
+ "grad_norm": 1.0729608535766602,
+ "learning_rate": 3.605573228249285e-06,
+ "loss": 0.0098,
+ "step": 26762
+ },
+ {
+ "epoch": 72.92370572207085,
+ "grad_norm": 0.9788365364074707,
+ "learning_rate": 3.604894757054931e-06,
+ "loss": 0.0222,
+ "step": 26763
+ },
+ {
+ "epoch": 72.92643051771117,
+ "grad_norm": 2.167494773864746,
+ "learning_rate": 3.604216335665449e-06,
+ "loss": 0.0157,
+ "step": 26764
+ },
+ {
+ "epoch": 72.9291553133515,
+ "grad_norm": 1.376555323600769,
+ "learning_rate": 3.603537964086129e-06,
+ "loss": 0.0538,
+ "step": 26765
+ },
+ {
+ "epoch": 72.93188010899182,
+ "grad_norm": 1.5772875547409058,
+ "learning_rate": 3.6028596423222484e-06,
+ "loss": 0.0304,
+ "step": 26766
+ },
+ {
+ "epoch": 72.93460490463215,
+ "grad_norm": 1.7784985303878784,
+ "learning_rate": 3.602181370379092e-06,
+ "loss": 0.0203,
+ "step": 26767
+ },
+ {
+ "epoch": 72.93732970027249,
+ "grad_norm": 1.0974540710449219,
+ "learning_rate": 3.601503148261938e-06,
+ "loss": 0.0094,
+ "step": 26768
+ },
+ {
+ "epoch": 72.9400544959128,
+ "grad_norm": 1.1879462003707886,
+ "learning_rate": 3.6008249759760737e-06,
+ "loss": 0.0103,
+ "step": 26769
+ },
+ {
+ "epoch": 72.94277929155314,
+ "grad_norm": 2.351264238357544,
+ "learning_rate": 3.600146853526779e-06,
+ "loss": 0.0207,
+ "step": 26770
+ },
+ {
+ "epoch": 72.94550408719346,
+ "grad_norm": 1.4934035539627075,
+ "learning_rate": 3.5994687809193317e-06,
+ "loss": 0.1521,
+ "step": 26771
+ },
+ {
+ "epoch": 72.94822888283379,
+ "grad_norm": 1.1168859004974365,
+ "learning_rate": 3.598790758159019e-06,
+ "loss": 0.0474,
+ "step": 26772
+ },
+ {
+ "epoch": 72.95095367847412,
+ "grad_norm": 1.3432073593139648,
+ "learning_rate": 3.5981127852511178e-06,
+ "loss": 0.1372,
+ "step": 26773
+ },
+ {
+ "epoch": 72.95367847411444,
+ "grad_norm": 1.5664137601852417,
+ "learning_rate": 3.5974348622009083e-06,
+ "loss": 0.0161,
+ "step": 26774
+ },
+ {
+ "epoch": 72.95640326975477,
+ "grad_norm": 1.3135292530059814,
+ "learning_rate": 3.5967569890136654e-06,
+ "loss": 0.0165,
+ "step": 26775
+ },
+ {
+ "epoch": 72.95912806539509,
+ "grad_norm": 2.234971761703491,
+ "learning_rate": 3.596079165694677e-06,
+ "loss": 0.0288,
+ "step": 26776
+ },
+ {
+ "epoch": 72.96185286103542,
+ "grad_norm": 1.3805809020996094,
+ "learning_rate": 3.595401392249218e-06,
+ "loss": 0.0147,
+ "step": 26777
+ },
+ {
+ "epoch": 72.96457765667574,
+ "grad_norm": 1.5882090330123901,
+ "learning_rate": 3.5947236686825683e-06,
+ "loss": 0.023,
+ "step": 26778
+ },
+ {
+ "epoch": 72.96730245231608,
+ "grad_norm": 1.4024906158447266,
+ "learning_rate": 3.5940459949999995e-06,
+ "loss": 0.0339,
+ "step": 26779
+ },
+ {
+ "epoch": 72.97002724795641,
+ "grad_norm": 1.6416181325912476,
+ "learning_rate": 3.593368371206798e-06,
+ "loss": 0.0916,
+ "step": 26780
+ },
+ {
+ "epoch": 72.97275204359673,
+ "grad_norm": 1.3068419694900513,
+ "learning_rate": 3.5926907973082335e-06,
+ "loss": 0.0269,
+ "step": 26781
+ },
+ {
+ "epoch": 72.97547683923706,
+ "grad_norm": 1.8523505926132202,
+ "learning_rate": 3.5920132733095912e-06,
+ "loss": 0.1476,
+ "step": 26782
+ },
+ {
+ "epoch": 72.97820163487738,
+ "grad_norm": 2.7124927043914795,
+ "learning_rate": 3.5913357992161422e-06,
+ "loss": 0.0962,
+ "step": 26783
+ },
+ {
+ "epoch": 72.98092643051771,
+ "grad_norm": 1.5436311960220337,
+ "learning_rate": 3.5906583750331614e-06,
+ "loss": 0.0848,
+ "step": 26784
+ },
+ {
+ "epoch": 72.98365122615803,
+ "grad_norm": 1.8496211767196655,
+ "learning_rate": 3.5899810007659296e-06,
+ "loss": 0.0258,
+ "step": 26785
+ },
+ {
+ "epoch": 72.98637602179836,
+ "grad_norm": 1.7183705568313599,
+ "learning_rate": 3.589303676419721e-06,
+ "loss": 0.0729,
+ "step": 26786
+ },
+ {
+ "epoch": 72.9891008174387,
+ "grad_norm": 1.4156242609024048,
+ "learning_rate": 3.5886264019998074e-06,
+ "loss": 0.0105,
+ "step": 26787
+ },
+ {
+ "epoch": 72.99182561307902,
+ "grad_norm": 2.9948313236236572,
+ "learning_rate": 3.5879491775114618e-06,
+ "loss": 0.0265,
+ "step": 26788
+ },
+ {
+ "epoch": 72.99455040871935,
+ "grad_norm": 0.9391252994537354,
+ "learning_rate": 3.587272002959965e-06,
+ "loss": 0.0169,
+ "step": 26789
+ },
+ {
+ "epoch": 72.99727520435967,
+ "grad_norm": 1.6095744371414185,
+ "learning_rate": 3.5865948783505877e-06,
+ "loss": 0.0474,
+ "step": 26790
+ },
+ {
+ "epoch": 73.0,
+ "grad_norm": 1.001164436340332,
+ "learning_rate": 3.585917803688603e-06,
+ "loss": 0.0216,
+ "step": 26791
+ },
+ {
+ "epoch": 73.00272479564033,
+ "grad_norm": 1.5885982513427734,
+ "learning_rate": 3.585240778979281e-06,
+ "loss": 0.1787,
+ "step": 26792
+ },
+ {
+ "epoch": 73.00544959128065,
+ "grad_norm": 1.2591066360473633,
+ "learning_rate": 3.584563804227901e-06,
+ "loss": 0.0467,
+ "step": 26793
+ },
+ {
+ "epoch": 73.00817438692098,
+ "grad_norm": 0.9301493763923645,
+ "learning_rate": 3.583886879439731e-06,
+ "loss": 0.0181,
+ "step": 26794
+ },
+ {
+ "epoch": 73.0108991825613,
+ "grad_norm": 1.630492925643921,
+ "learning_rate": 3.58321000462004e-06,
+ "loss": 0.1736,
+ "step": 26795
+ },
+ {
+ "epoch": 73.01362397820164,
+ "grad_norm": 1.6202369928359985,
+ "learning_rate": 3.582533179774107e-06,
+ "loss": 0.0842,
+ "step": 26796
+ },
+ {
+ "epoch": 73.01634877384195,
+ "grad_norm": 0.8104017376899719,
+ "learning_rate": 3.581856404907198e-06,
+ "loss": 0.0102,
+ "step": 26797
+ },
+ {
+ "epoch": 73.01907356948229,
+ "grad_norm": 1.6932439804077148,
+ "learning_rate": 3.5811796800245857e-06,
+ "loss": 0.0985,
+ "step": 26798
+ },
+ {
+ "epoch": 73.02179836512262,
+ "grad_norm": 0.7468889951705933,
+ "learning_rate": 3.580503005131536e-06,
+ "loss": 0.0083,
+ "step": 26799
+ },
+ {
+ "epoch": 73.02452316076294,
+ "grad_norm": 1.8475871086120605,
+ "learning_rate": 3.579826380233322e-06,
+ "loss": 0.1002,
+ "step": 26800
+ },
+ {
+ "epoch": 73.02724795640327,
+ "grad_norm": 1.2656397819519043,
+ "learning_rate": 3.579149805335217e-06,
+ "loss": 0.0123,
+ "step": 26801
+ },
+ {
+ "epoch": 73.02997275204359,
+ "grad_norm": 1.5699411630630493,
+ "learning_rate": 3.578473280442486e-06,
+ "loss": 0.0567,
+ "step": 26802
+ },
+ {
+ "epoch": 73.03269754768392,
+ "grad_norm": 1.5518677234649658,
+ "learning_rate": 3.5777968055603996e-06,
+ "loss": 0.019,
+ "step": 26803
+ },
+ {
+ "epoch": 73.03542234332426,
+ "grad_norm": 1.6804999113082886,
+ "learning_rate": 3.57712038069422e-06,
+ "loss": 0.0189,
+ "step": 26804
+ },
+ {
+ "epoch": 73.03814713896458,
+ "grad_norm": 1.6616747379302979,
+ "learning_rate": 3.5764440058492255e-06,
+ "loss": 0.0384,
+ "step": 26805
+ },
+ {
+ "epoch": 73.04087193460491,
+ "grad_norm": 1.7160879373550415,
+ "learning_rate": 3.5757676810306775e-06,
+ "loss": 0.0151,
+ "step": 26806
+ },
+ {
+ "epoch": 73.04359673024523,
+ "grad_norm": 1.424336314201355,
+ "learning_rate": 3.5750914062438437e-06,
+ "loss": 0.0117,
+ "step": 26807
+ },
+ {
+ "epoch": 73.04632152588556,
+ "grad_norm": 1.1326876878738403,
+ "learning_rate": 3.5744151814939874e-06,
+ "loss": 0.0231,
+ "step": 26808
+ },
+ {
+ "epoch": 73.04904632152588,
+ "grad_norm": 1.7301476001739502,
+ "learning_rate": 3.5737390067863833e-06,
+ "loss": 0.0337,
+ "step": 26809
+ },
+ {
+ "epoch": 73.05177111716621,
+ "grad_norm": 1.522494912147522,
+ "learning_rate": 3.5730628821262926e-06,
+ "loss": 0.0331,
+ "step": 26810
+ },
+ {
+ "epoch": 73.05449591280654,
+ "grad_norm": 1.4967694282531738,
+ "learning_rate": 3.5723868075189817e-06,
+ "loss": 0.0577,
+ "step": 26811
+ },
+ {
+ "epoch": 73.05722070844686,
+ "grad_norm": 1.3473355770111084,
+ "learning_rate": 3.5717107829697117e-06,
+ "loss": 0.0128,
+ "step": 26812
+ },
+ {
+ "epoch": 73.0599455040872,
+ "grad_norm": 0.6300485134124756,
+ "learning_rate": 3.5710348084837543e-06,
+ "loss": 0.0074,
+ "step": 26813
+ },
+ {
+ "epoch": 73.06267029972751,
+ "grad_norm": 0.8418791890144348,
+ "learning_rate": 3.5703588840663706e-06,
+ "loss": 0.0118,
+ "step": 26814
+ },
+ {
+ "epoch": 73.06539509536785,
+ "grad_norm": 0.9723634123802185,
+ "learning_rate": 3.569683009722826e-06,
+ "loss": 0.0125,
+ "step": 26815
+ },
+ {
+ "epoch": 73.06811989100818,
+ "grad_norm": 1.5083587169647217,
+ "learning_rate": 3.5690071854583763e-06,
+ "loss": 0.0174,
+ "step": 26816
+ },
+ {
+ "epoch": 73.0708446866485,
+ "grad_norm": 1.0181983709335327,
+ "learning_rate": 3.5683314112782972e-06,
+ "loss": 0.0089,
+ "step": 26817
+ },
+ {
+ "epoch": 73.07356948228883,
+ "grad_norm": 1.4106874465942383,
+ "learning_rate": 3.567655687187848e-06,
+ "loss": 0.0175,
+ "step": 26818
+ },
+ {
+ "epoch": 73.07629427792915,
+ "grad_norm": 1.8960984945297241,
+ "learning_rate": 3.566980013192288e-06,
+ "loss": 0.035,
+ "step": 26819
+ },
+ {
+ "epoch": 73.07901907356948,
+ "grad_norm": 0.4772966206073761,
+ "learning_rate": 3.566304389296877e-06,
+ "loss": 0.0043,
+ "step": 26820
+ },
+ {
+ "epoch": 73.0817438692098,
+ "grad_norm": 1.466974139213562,
+ "learning_rate": 3.5656288155068842e-06,
+ "loss": 0.0235,
+ "step": 26821
+ },
+ {
+ "epoch": 73.08446866485014,
+ "grad_norm": 1.1632007360458374,
+ "learning_rate": 3.5649532918275675e-06,
+ "loss": 0.0146,
+ "step": 26822
+ },
+ {
+ "epoch": 73.08719346049047,
+ "grad_norm": 1.8220267295837402,
+ "learning_rate": 3.5642778182641867e-06,
+ "loss": 0.0567,
+ "step": 26823
+ },
+ {
+ "epoch": 73.08991825613079,
+ "grad_norm": 1.174680471420288,
+ "learning_rate": 3.563602394822e-06,
+ "loss": 0.0175,
+ "step": 26824
+ },
+ {
+ "epoch": 73.09264305177112,
+ "grad_norm": 1.0014585256576538,
+ "learning_rate": 3.562927021506275e-06,
+ "loss": 0.0116,
+ "step": 26825
+ },
+ {
+ "epoch": 73.09536784741144,
+ "grad_norm": 1.4355638027191162,
+ "learning_rate": 3.5622516983222656e-06,
+ "loss": 0.0157,
+ "step": 26826
+ },
+ {
+ "epoch": 73.09809264305177,
+ "grad_norm": 1.5533653497695923,
+ "learning_rate": 3.561576425275234e-06,
+ "loss": 0.0981,
+ "step": 26827
+ },
+ {
+ "epoch": 73.1008174386921,
+ "grad_norm": 2.506122350692749,
+ "learning_rate": 3.5609012023704338e-06,
+ "loss": 0.0379,
+ "step": 26828
+ },
+ {
+ "epoch": 73.10354223433242,
+ "grad_norm": 1.4850932359695435,
+ "learning_rate": 3.560226029613132e-06,
+ "loss": 0.0162,
+ "step": 26829
+ },
+ {
+ "epoch": 73.10626702997276,
+ "grad_norm": 1.083440899848938,
+ "learning_rate": 3.5595509070085822e-06,
+ "loss": 0.0174,
+ "step": 26830
+ },
+ {
+ "epoch": 73.10899182561307,
+ "grad_norm": 2.2017297744750977,
+ "learning_rate": 3.558875834562042e-06,
+ "loss": 0.021,
+ "step": 26831
+ },
+ {
+ "epoch": 73.11171662125341,
+ "grad_norm": 1.9987988471984863,
+ "learning_rate": 3.558200812278767e-06,
+ "loss": 0.0239,
+ "step": 26832
+ },
+ {
+ "epoch": 73.11444141689373,
+ "grad_norm": 1.390193223953247,
+ "learning_rate": 3.557525840164017e-06,
+ "loss": 0.0179,
+ "step": 26833
+ },
+ {
+ "epoch": 73.11716621253406,
+ "grad_norm": 1.4627598524093628,
+ "learning_rate": 3.5568509182230503e-06,
+ "loss": 0.0323,
+ "step": 26834
+ },
+ {
+ "epoch": 73.11989100817439,
+ "grad_norm": 1.6351338624954224,
+ "learning_rate": 3.5561760464611227e-06,
+ "loss": 0.096,
+ "step": 26835
+ },
+ {
+ "epoch": 73.12261580381471,
+ "grad_norm": 0.9832163453102112,
+ "learning_rate": 3.555501224883484e-06,
+ "loss": 0.0156,
+ "step": 26836
+ },
+ {
+ "epoch": 73.12534059945504,
+ "grad_norm": 1.2897690534591675,
+ "learning_rate": 3.554826453495399e-06,
+ "loss": 0.0781,
+ "step": 26837
+ },
+ {
+ "epoch": 73.12806539509536,
+ "grad_norm": 2.421398401260376,
+ "learning_rate": 3.554151732302118e-06,
+ "loss": 0.1746,
+ "step": 26838
+ },
+ {
+ "epoch": 73.1307901907357,
+ "grad_norm": 1.2675755023956299,
+ "learning_rate": 3.553477061308895e-06,
+ "loss": 0.0151,
+ "step": 26839
+ },
+ {
+ "epoch": 73.13351498637603,
+ "grad_norm": 1.2903728485107422,
+ "learning_rate": 3.5528024405209825e-06,
+ "loss": 0.0938,
+ "step": 26840
+ },
+ {
+ "epoch": 73.13623978201635,
+ "grad_norm": 2.23583722114563,
+ "learning_rate": 3.55212786994364e-06,
+ "loss": 0.086,
+ "step": 26841
+ },
+ {
+ "epoch": 73.13896457765668,
+ "grad_norm": 1.1194401979446411,
+ "learning_rate": 3.551453349582119e-06,
+ "loss": 0.0132,
+ "step": 26842
+ },
+ {
+ "epoch": 73.141689373297,
+ "grad_norm": 1.9038807153701782,
+ "learning_rate": 3.550778879441671e-06,
+ "loss": 0.0336,
+ "step": 26843
+ },
+ {
+ "epoch": 73.14441416893733,
+ "grad_norm": 1.2172648906707764,
+ "learning_rate": 3.550104459527546e-06,
+ "loss": 0.0168,
+ "step": 26844
+ },
+ {
+ "epoch": 73.14713896457765,
+ "grad_norm": 1.1823198795318604,
+ "learning_rate": 3.549430089845004e-06,
+ "loss": 0.0153,
+ "step": 26845
+ },
+ {
+ "epoch": 73.14986376021798,
+ "grad_norm": 0.7558379173278809,
+ "learning_rate": 3.5487557703992927e-06,
+ "loss": 0.0112,
+ "step": 26846
+ },
+ {
+ "epoch": 73.15258855585832,
+ "grad_norm": 1.3384181261062622,
+ "learning_rate": 3.5480815011956626e-06,
+ "loss": 0.0652,
+ "step": 26847
+ },
+ {
+ "epoch": 73.15531335149863,
+ "grad_norm": 1.1708089113235474,
+ "learning_rate": 3.5474072822393635e-06,
+ "loss": 0.0134,
+ "step": 26848
+ },
+ {
+ "epoch": 73.15803814713897,
+ "grad_norm": 1.4616800546646118,
+ "learning_rate": 3.546733113535653e-06,
+ "loss": 0.0118,
+ "step": 26849
+ },
+ {
+ "epoch": 73.16076294277929,
+ "grad_norm": 1.707106351852417,
+ "learning_rate": 3.5460589950897726e-06,
+ "loss": 0.0567,
+ "step": 26850
+ },
+ {
+ "epoch": 73.16348773841962,
+ "grad_norm": 1.7232030630111694,
+ "learning_rate": 3.545384926906982e-06,
+ "loss": 0.0232,
+ "step": 26851
+ },
+ {
+ "epoch": 73.16621253405995,
+ "grad_norm": 1.36274254322052,
+ "learning_rate": 3.5447109089925247e-06,
+ "loss": 0.021,
+ "step": 26852
+ },
+ {
+ "epoch": 73.16893732970027,
+ "grad_norm": 1.2128552198410034,
+ "learning_rate": 3.5440369413516474e-06,
+ "loss": 0.0117,
+ "step": 26853
+ },
+ {
+ "epoch": 73.1716621253406,
+ "grad_norm": 1.4983010292053223,
+ "learning_rate": 3.5433630239896064e-06,
+ "loss": 0.0281,
+ "step": 26854
+ },
+ {
+ "epoch": 73.17438692098092,
+ "grad_norm": 1.182462453842163,
+ "learning_rate": 3.542689156911647e-06,
+ "loss": 0.0181,
+ "step": 26855
+ },
+ {
+ "epoch": 73.17711171662125,
+ "grad_norm": 1.6042964458465576,
+ "learning_rate": 3.542015340123016e-06,
+ "loss": 0.0274,
+ "step": 26856
+ },
+ {
+ "epoch": 73.17983651226157,
+ "grad_norm": 1.889832615852356,
+ "learning_rate": 3.541341573628958e-06,
+ "loss": 0.0748,
+ "step": 26857
+ },
+ {
+ "epoch": 73.1825613079019,
+ "grad_norm": 0.7693272233009338,
+ "learning_rate": 3.5406678574347287e-06,
+ "loss": 0.01,
+ "step": 26858
+ },
+ {
+ "epoch": 73.18528610354224,
+ "grad_norm": 1.4291727542877197,
+ "learning_rate": 3.5399941915455695e-06,
+ "loss": 0.0515,
+ "step": 26859
+ },
+ {
+ "epoch": 73.18801089918256,
+ "grad_norm": 1.8926098346710205,
+ "learning_rate": 3.5393205759667237e-06,
+ "loss": 0.1696,
+ "step": 26860
+ },
+ {
+ "epoch": 73.19073569482289,
+ "grad_norm": 1.614081621170044,
+ "learning_rate": 3.5386470107034443e-06,
+ "loss": 0.0807,
+ "step": 26861
+ },
+ {
+ "epoch": 73.19346049046321,
+ "grad_norm": 1.3774642944335938,
+ "learning_rate": 3.5379734957609748e-06,
+ "loss": 0.0155,
+ "step": 26862
+ },
+ {
+ "epoch": 73.19618528610354,
+ "grad_norm": 2.329394817352295,
+ "learning_rate": 3.5373000311445604e-06,
+ "loss": 0.0505,
+ "step": 26863
+ },
+ {
+ "epoch": 73.19891008174388,
+ "grad_norm": 1.318786859512329,
+ "learning_rate": 3.536626616859441e-06,
+ "loss": 0.0188,
+ "step": 26864
+ },
+ {
+ "epoch": 73.2016348773842,
+ "grad_norm": 1.4323241710662842,
+ "learning_rate": 3.5359532529108687e-06,
+ "loss": 0.0645,
+ "step": 26865
+ },
+ {
+ "epoch": 73.20435967302453,
+ "grad_norm": 1.2406256198883057,
+ "learning_rate": 3.535279939304085e-06,
+ "loss": 0.0539,
+ "step": 26866
+ },
+ {
+ "epoch": 73.20708446866485,
+ "grad_norm": 1.4622066020965576,
+ "learning_rate": 3.53460667604433e-06,
+ "loss": 0.1248,
+ "step": 26867
+ },
+ {
+ "epoch": 73.20980926430518,
+ "grad_norm": 0.6628721952438354,
+ "learning_rate": 3.5339334631368527e-06,
+ "loss": 0.0087,
+ "step": 26868
+ },
+ {
+ "epoch": 73.2125340599455,
+ "grad_norm": 1.544886589050293,
+ "learning_rate": 3.5332603005868906e-06,
+ "loss": 0.141,
+ "step": 26869
+ },
+ {
+ "epoch": 73.21525885558583,
+ "grad_norm": 1.522874355316162,
+ "learning_rate": 3.532587188399692e-06,
+ "loss": 0.0308,
+ "step": 26870
+ },
+ {
+ "epoch": 73.21798365122616,
+ "grad_norm": 1.6858489513397217,
+ "learning_rate": 3.531914126580497e-06,
+ "loss": 0.0259,
+ "step": 26871
+ },
+ {
+ "epoch": 73.22070844686648,
+ "grad_norm": 0.8736989498138428,
+ "learning_rate": 3.531241115134546e-06,
+ "loss": 0.0094,
+ "step": 26872
+ },
+ {
+ "epoch": 73.22343324250681,
+ "grad_norm": 1.0737591981887817,
+ "learning_rate": 3.530568154067078e-06,
+ "loss": 0.0111,
+ "step": 26873
+ },
+ {
+ "epoch": 73.22615803814713,
+ "grad_norm": 1.20064377784729,
+ "learning_rate": 3.52989524338334e-06,
+ "loss": 0.0166,
+ "step": 26874
+ },
+ {
+ "epoch": 73.22888283378747,
+ "grad_norm": 1.9041961431503296,
+ "learning_rate": 3.5292223830885696e-06,
+ "loss": 0.0383,
+ "step": 26875
+ },
+ {
+ "epoch": 73.2316076294278,
+ "grad_norm": 1.0177842378616333,
+ "learning_rate": 3.528549573188007e-06,
+ "loss": 0.0097,
+ "step": 26876
+ },
+ {
+ "epoch": 73.23433242506812,
+ "grad_norm": 1.8369476795196533,
+ "learning_rate": 3.527876813686889e-06,
+ "loss": 0.0429,
+ "step": 26877
+ },
+ {
+ "epoch": 73.23705722070845,
+ "grad_norm": 1.001833438873291,
+ "learning_rate": 3.5272041045904615e-06,
+ "loss": 0.0119,
+ "step": 26878
+ },
+ {
+ "epoch": 73.23978201634877,
+ "grad_norm": 1.5908348560333252,
+ "learning_rate": 3.52653144590396e-06,
+ "loss": 0.0406,
+ "step": 26879
+ },
+ {
+ "epoch": 73.2425068119891,
+ "grad_norm": 1.5699548721313477,
+ "learning_rate": 3.5258588376326198e-06,
+ "loss": 0.0382,
+ "step": 26880
+ },
+ {
+ "epoch": 73.24523160762942,
+ "grad_norm": 1.4443424940109253,
+ "learning_rate": 3.525186279781686e-06,
+ "loss": 0.0547,
+ "step": 26881
+ },
+ {
+ "epoch": 73.24795640326975,
+ "grad_norm": 1.555729866027832,
+ "learning_rate": 3.524513772356393e-06,
+ "loss": 0.0143,
+ "step": 26882
+ },
+ {
+ "epoch": 73.25068119891009,
+ "grad_norm": 1.3544296026229858,
+ "learning_rate": 3.5238413153619788e-06,
+ "loss": 0.0128,
+ "step": 26883
+ },
+ {
+ "epoch": 73.2534059945504,
+ "grad_norm": 1.1722815036773682,
+ "learning_rate": 3.5231689088036757e-06,
+ "loss": 0.0094,
+ "step": 26884
+ },
+ {
+ "epoch": 73.25613079019074,
+ "grad_norm": 1.1991057395935059,
+ "learning_rate": 3.522496552686725e-06,
+ "loss": 0.0277,
+ "step": 26885
+ },
+ {
+ "epoch": 73.25885558583106,
+ "grad_norm": 0.7400229573249817,
+ "learning_rate": 3.5218242470163667e-06,
+ "loss": 0.0059,
+ "step": 26886
+ },
+ {
+ "epoch": 73.26158038147139,
+ "grad_norm": 1.4573155641555786,
+ "learning_rate": 3.521151991797832e-06,
+ "loss": 0.0189,
+ "step": 26887
+ },
+ {
+ "epoch": 73.26430517711172,
+ "grad_norm": 1.5594916343688965,
+ "learning_rate": 3.5204797870363573e-06,
+ "loss": 0.0211,
+ "step": 26888
+ },
+ {
+ "epoch": 73.26702997275204,
+ "grad_norm": 1.6998999118804932,
+ "learning_rate": 3.5198076327371744e-06,
+ "loss": 0.023,
+ "step": 26889
+ },
+ {
+ "epoch": 73.26975476839237,
+ "grad_norm": 1.184098243713379,
+ "learning_rate": 3.519135528905524e-06,
+ "loss": 0.0941,
+ "step": 26890
+ },
+ {
+ "epoch": 73.2724795640327,
+ "grad_norm": 0.8400129079818726,
+ "learning_rate": 3.5184634755466374e-06,
+ "loss": 0.009,
+ "step": 26891
+ },
+ {
+ "epoch": 73.27520435967303,
+ "grad_norm": 1.0153467655181885,
+ "learning_rate": 3.5177914726657493e-06,
+ "loss": 0.0089,
+ "step": 26892
+ },
+ {
+ "epoch": 73.27792915531334,
+ "grad_norm": 0.8667426705360413,
+ "learning_rate": 3.5171195202680887e-06,
+ "loss": 0.0096,
+ "step": 26893
+ },
+ {
+ "epoch": 73.28065395095368,
+ "grad_norm": 0.9230378270149231,
+ "learning_rate": 3.516447618358896e-06,
+ "loss": 0.013,
+ "step": 26894
+ },
+ {
+ "epoch": 73.28337874659401,
+ "grad_norm": 1.5612901449203491,
+ "learning_rate": 3.5157757669434e-06,
+ "loss": 0.0656,
+ "step": 26895
+ },
+ {
+ "epoch": 73.28610354223433,
+ "grad_norm": 2.0320656299591064,
+ "learning_rate": 3.515103966026835e-06,
+ "loss": 0.0423,
+ "step": 26896
+ },
+ {
+ "epoch": 73.28882833787466,
+ "grad_norm": 1.3622171878814697,
+ "learning_rate": 3.5144322156144274e-06,
+ "loss": 0.0138,
+ "step": 26897
+ },
+ {
+ "epoch": 73.29155313351498,
+ "grad_norm": 0.8692339062690735,
+ "learning_rate": 3.5137605157114164e-06,
+ "loss": 0.0089,
+ "step": 26898
+ },
+ {
+ "epoch": 73.29427792915531,
+ "grad_norm": 1.384666919708252,
+ "learning_rate": 3.51308886632303e-06,
+ "loss": 0.0107,
+ "step": 26899
+ },
+ {
+ "epoch": 73.29700272479565,
+ "grad_norm": 2.936274528503418,
+ "learning_rate": 3.512417267454499e-06,
+ "loss": 0.0204,
+ "step": 26900
+ },
+ {
+ "epoch": 73.29972752043597,
+ "grad_norm": 0.9594311118125916,
+ "learning_rate": 3.5117457191110495e-06,
+ "loss": 0.0168,
+ "step": 26901
+ },
+ {
+ "epoch": 73.3024523160763,
+ "grad_norm": 1.9423656463623047,
+ "learning_rate": 3.5110742212979156e-06,
+ "loss": 0.0137,
+ "step": 26902
+ },
+ {
+ "epoch": 73.30517711171662,
+ "grad_norm": 0.8901655673980713,
+ "learning_rate": 3.5104027740203305e-06,
+ "loss": 0.0098,
+ "step": 26903
+ },
+ {
+ "epoch": 73.30790190735695,
+ "grad_norm": 1.2303189039230347,
+ "learning_rate": 3.5097313772835195e-06,
+ "loss": 0.0111,
+ "step": 26904
+ },
+ {
+ "epoch": 73.31062670299727,
+ "grad_norm": 1.3265072107315063,
+ "learning_rate": 3.509060031092708e-06,
+ "loss": 0.0192,
+ "step": 26905
+ },
+ {
+ "epoch": 73.3133514986376,
+ "grad_norm": 1.4859164953231812,
+ "learning_rate": 3.508388735453132e-06,
+ "loss": 0.0171,
+ "step": 26906
+ },
+ {
+ "epoch": 73.31607629427793,
+ "grad_norm": 0.8774119019508362,
+ "learning_rate": 3.507717490370015e-06,
+ "loss": 0.0069,
+ "step": 26907
+ },
+ {
+ "epoch": 73.31880108991825,
+ "grad_norm": 1.0680253505706787,
+ "learning_rate": 3.507046295848585e-06,
+ "loss": 0.0111,
+ "step": 26908
+ },
+ {
+ "epoch": 73.32152588555859,
+ "grad_norm": 1.6243782043457031,
+ "learning_rate": 3.506375151894067e-06,
+ "loss": 0.0178,
+ "step": 26909
+ },
+ {
+ "epoch": 73.3242506811989,
+ "grad_norm": 1.4997704029083252,
+ "learning_rate": 3.505704058511694e-06,
+ "loss": 0.0145,
+ "step": 26910
+ },
+ {
+ "epoch": 73.32697547683924,
+ "grad_norm": 0.9676305651664734,
+ "learning_rate": 3.505033015706688e-06,
+ "loss": 0.0942,
+ "step": 26911
+ },
+ {
+ "epoch": 73.32970027247957,
+ "grad_norm": 1.9199048280715942,
+ "learning_rate": 3.5043620234842757e-06,
+ "loss": 0.0248,
+ "step": 26912
+ },
+ {
+ "epoch": 73.33242506811989,
+ "grad_norm": 1.0797439813613892,
+ "learning_rate": 3.5036910818496792e-06,
+ "loss": 0.0098,
+ "step": 26913
+ },
+ {
+ "epoch": 73.33514986376022,
+ "grad_norm": 1.5246152877807617,
+ "learning_rate": 3.5030201908081316e-06,
+ "loss": 0.1549,
+ "step": 26914
+ },
+ {
+ "epoch": 73.33787465940054,
+ "grad_norm": 1.2112512588500977,
+ "learning_rate": 3.5023493503648543e-06,
+ "loss": 0.0204,
+ "step": 26915
+ },
+ {
+ "epoch": 73.34059945504087,
+ "grad_norm": 1.4717096090316772,
+ "learning_rate": 3.5016785605250702e-06,
+ "loss": 0.026,
+ "step": 26916
+ },
+ {
+ "epoch": 73.34332425068119,
+ "grad_norm": 1.2076141834259033,
+ "learning_rate": 3.5010078212940012e-06,
+ "loss": 0.016,
+ "step": 26917
+ },
+ {
+ "epoch": 73.34604904632153,
+ "grad_norm": 1.1581727266311646,
+ "learning_rate": 3.5003371326768778e-06,
+ "loss": 0.0312,
+ "step": 26918
+ },
+ {
+ "epoch": 73.34877384196186,
+ "grad_norm": 1.2806150913238525,
+ "learning_rate": 3.499666494678915e-06,
+ "loss": 0.0252,
+ "step": 26919
+ },
+ {
+ "epoch": 73.35149863760218,
+ "grad_norm": 1.4886395931243896,
+ "learning_rate": 3.4989959073053447e-06,
+ "loss": 0.016,
+ "step": 26920
+ },
+ {
+ "epoch": 73.35422343324251,
+ "grad_norm": 1.67676842212677,
+ "learning_rate": 3.498325370561385e-06,
+ "loss": 0.0197,
+ "step": 26921
+ },
+ {
+ "epoch": 73.35694822888283,
+ "grad_norm": 1.633744478225708,
+ "learning_rate": 3.497654884452254e-06,
+ "loss": 0.032,
+ "step": 26922
+ },
+ {
+ "epoch": 73.35967302452316,
+ "grad_norm": 1.2924302816390991,
+ "learning_rate": 3.4969844489831817e-06,
+ "loss": 0.0232,
+ "step": 26923
+ },
+ {
+ "epoch": 73.3623978201635,
+ "grad_norm": 2.057108163833618,
+ "learning_rate": 3.496314064159385e-06,
+ "loss": 0.0171,
+ "step": 26924
+ },
+ {
+ "epoch": 73.36512261580381,
+ "grad_norm": 2.496807098388672,
+ "learning_rate": 3.495643729986081e-06,
+ "loss": 0.0646,
+ "step": 26925
+ },
+ {
+ "epoch": 73.36784741144415,
+ "grad_norm": 1.612742304801941,
+ "learning_rate": 3.494973446468499e-06,
+ "loss": 0.084,
+ "step": 26926
+ },
+ {
+ "epoch": 73.37057220708446,
+ "grad_norm": 1.6425955295562744,
+ "learning_rate": 3.4943032136118537e-06,
+ "loss": 0.0928,
+ "step": 26927
+ },
+ {
+ "epoch": 73.3732970027248,
+ "grad_norm": 1.2437413930892944,
+ "learning_rate": 3.493633031421365e-06,
+ "loss": 0.0112,
+ "step": 26928
+ },
+ {
+ "epoch": 73.37602179836512,
+ "grad_norm": 1.0460067987442017,
+ "learning_rate": 3.4929628999022502e-06,
+ "loss": 0.0122,
+ "step": 26929
+ },
+ {
+ "epoch": 73.37874659400545,
+ "grad_norm": 1.2761969566345215,
+ "learning_rate": 3.492292819059734e-06,
+ "loss": 0.0191,
+ "step": 26930
+ },
+ {
+ "epoch": 73.38147138964578,
+ "grad_norm": 1.3297553062438965,
+ "learning_rate": 3.491622788899033e-06,
+ "loss": 0.015,
+ "step": 26931
+ },
+ {
+ "epoch": 73.3841961852861,
+ "grad_norm": 1.5916402339935303,
+ "learning_rate": 3.490952809425363e-06,
+ "loss": 0.0326,
+ "step": 26932
+ },
+ {
+ "epoch": 73.38692098092643,
+ "grad_norm": 1.8644955158233643,
+ "learning_rate": 3.4902828806439394e-06,
+ "loss": 0.0501,
+ "step": 26933
+ },
+ {
+ "epoch": 73.38964577656675,
+ "grad_norm": 1.6113057136535645,
+ "learning_rate": 3.4896130025599874e-06,
+ "loss": 0.0314,
+ "step": 26934
+ },
+ {
+ "epoch": 73.39237057220708,
+ "grad_norm": 1.633375883102417,
+ "learning_rate": 3.4889431751787195e-06,
+ "loss": 0.0174,
+ "step": 26935
+ },
+ {
+ "epoch": 73.39509536784742,
+ "grad_norm": 1.1995936632156372,
+ "learning_rate": 3.4882733985053487e-06,
+ "loss": 0.015,
+ "step": 26936
+ },
+ {
+ "epoch": 73.39782016348774,
+ "grad_norm": 1.530089020729065,
+ "learning_rate": 3.4876036725450992e-06,
+ "loss": 0.0179,
+ "step": 26937
+ },
+ {
+ "epoch": 73.40054495912807,
+ "grad_norm": 1.159950613975525,
+ "learning_rate": 3.486933997303179e-06,
+ "loss": 0.0164,
+ "step": 26938
+ },
+ {
+ "epoch": 73.40326975476839,
+ "grad_norm": 6.812144756317139,
+ "learning_rate": 3.486264372784811e-06,
+ "loss": 0.0492,
+ "step": 26939
+ },
+ {
+ "epoch": 73.40599455040872,
+ "grad_norm": 0.8591850996017456,
+ "learning_rate": 3.4855947989952054e-06,
+ "loss": 0.0114,
+ "step": 26940
+ },
+ {
+ "epoch": 73.40871934604904,
+ "grad_norm": 1.151686191558838,
+ "learning_rate": 3.484925275939579e-06,
+ "loss": 0.0285,
+ "step": 26941
+ },
+ {
+ "epoch": 73.41144414168937,
+ "grad_norm": 1.2089855670928955,
+ "learning_rate": 3.484255803623141e-06,
+ "loss": 0.0234,
+ "step": 26942
+ },
+ {
+ "epoch": 73.4141689373297,
+ "grad_norm": 1.0293521881103516,
+ "learning_rate": 3.483586382051113e-06,
+ "loss": 0.1515,
+ "step": 26943
+ },
+ {
+ "epoch": 73.41689373297002,
+ "grad_norm": 1.2582935094833374,
+ "learning_rate": 3.4829170112287037e-06,
+ "loss": 0.0102,
+ "step": 26944
+ },
+ {
+ "epoch": 73.41961852861036,
+ "grad_norm": 1.1811808347702026,
+ "learning_rate": 3.4822476911611236e-06,
+ "loss": 0.022,
+ "step": 26945
+ },
+ {
+ "epoch": 73.42234332425068,
+ "grad_norm": 1.134524941444397,
+ "learning_rate": 3.481578421853593e-06,
+ "loss": 0.1584,
+ "step": 26946
+ },
+ {
+ "epoch": 73.42506811989101,
+ "grad_norm": 1.6197205781936646,
+ "learning_rate": 3.480909203311319e-06,
+ "loss": 0.1851,
+ "step": 26947
+ },
+ {
+ "epoch": 73.42779291553134,
+ "grad_norm": 1.115896224975586,
+ "learning_rate": 3.480240035539515e-06,
+ "loss": 0.133,
+ "step": 26948
+ },
+ {
+ "epoch": 73.43051771117166,
+ "grad_norm": 1.3599300384521484,
+ "learning_rate": 3.479570918543388e-06,
+ "loss": 0.0301,
+ "step": 26949
+ },
+ {
+ "epoch": 73.433242506812,
+ "grad_norm": 2.0571916103363037,
+ "learning_rate": 3.478901852328156e-06,
+ "loss": 0.0296,
+ "step": 26950
+ },
+ {
+ "epoch": 73.43596730245231,
+ "grad_norm": 1.7245001792907715,
+ "learning_rate": 3.478232836899026e-06,
+ "loss": 0.149,
+ "step": 26951
+ },
+ {
+ "epoch": 73.43869209809264,
+ "grad_norm": 1.2692022323608398,
+ "learning_rate": 3.4775638722612095e-06,
+ "loss": 0.0188,
+ "step": 26952
+ },
+ {
+ "epoch": 73.44141689373296,
+ "grad_norm": 1.8034781217575073,
+ "learning_rate": 3.4768949584199118e-06,
+ "loss": 0.0498,
+ "step": 26953
+ },
+ {
+ "epoch": 73.4441416893733,
+ "grad_norm": 1.5726354122161865,
+ "learning_rate": 3.476226095380347e-06,
+ "loss": 0.0163,
+ "step": 26954
+ },
+ {
+ "epoch": 73.44686648501363,
+ "grad_norm": 1.3773224353790283,
+ "learning_rate": 3.4755572831477267e-06,
+ "loss": 0.0171,
+ "step": 26955
+ },
+ {
+ "epoch": 73.44959128065395,
+ "grad_norm": 2.5415568351745605,
+ "learning_rate": 3.4748885217272554e-06,
+ "loss": 0.0228,
+ "step": 26956
+ },
+ {
+ "epoch": 73.45231607629428,
+ "grad_norm": 1.5418081283569336,
+ "learning_rate": 3.4742198111241434e-06,
+ "loss": 0.0684,
+ "step": 26957
+ },
+ {
+ "epoch": 73.4550408719346,
+ "grad_norm": 1.0943989753723145,
+ "learning_rate": 3.473551151343594e-06,
+ "loss": 0.0143,
+ "step": 26958
+ },
+ {
+ "epoch": 73.45776566757493,
+ "grad_norm": 1.3076106309890747,
+ "learning_rate": 3.472882542390821e-06,
+ "loss": 0.0083,
+ "step": 26959
+ },
+ {
+ "epoch": 73.46049046321527,
+ "grad_norm": 1.2178289890289307,
+ "learning_rate": 3.472213984271028e-06,
+ "loss": 0.0115,
+ "step": 26960
+ },
+ {
+ "epoch": 73.46321525885558,
+ "grad_norm": 1.9858829975128174,
+ "learning_rate": 3.471545476989423e-06,
+ "loss": 0.014,
+ "step": 26961
+ },
+ {
+ "epoch": 73.46594005449592,
+ "grad_norm": 0.9908836483955383,
+ "learning_rate": 3.470877020551209e-06,
+ "loss": 0.0137,
+ "step": 26962
+ },
+ {
+ "epoch": 73.46866485013624,
+ "grad_norm": 1.671607255935669,
+ "learning_rate": 3.4702086149615975e-06,
+ "loss": 0.0171,
+ "step": 26963
+ },
+ {
+ "epoch": 73.47138964577657,
+ "grad_norm": 1.3025988340377808,
+ "learning_rate": 3.469540260225791e-06,
+ "loss": 0.0102,
+ "step": 26964
+ },
+ {
+ "epoch": 73.47411444141689,
+ "grad_norm": 1.2731163501739502,
+ "learning_rate": 3.468871956348995e-06,
+ "loss": 0.0143,
+ "step": 26965
+ },
+ {
+ "epoch": 73.47683923705722,
+ "grad_norm": 0.6905955076217651,
+ "learning_rate": 3.4682037033364115e-06,
+ "loss": 0.0046,
+ "step": 26966
+ },
+ {
+ "epoch": 73.47956403269755,
+ "grad_norm": 1.6803486347198486,
+ "learning_rate": 3.4675355011932486e-06,
+ "loss": 0.0203,
+ "step": 26967
+ },
+ {
+ "epoch": 73.48228882833787,
+ "grad_norm": 0.9458033442497253,
+ "learning_rate": 3.4668673499247108e-06,
+ "loss": 0.0106,
+ "step": 26968
+ },
+ {
+ "epoch": 73.4850136239782,
+ "grad_norm": 1.40166175365448,
+ "learning_rate": 3.4661992495359953e-06,
+ "loss": 0.0269,
+ "step": 26969
+ },
+ {
+ "epoch": 73.48773841961852,
+ "grad_norm": 0.9959577918052673,
+ "learning_rate": 3.4655312000323137e-06,
+ "loss": 0.0129,
+ "step": 26970
+ },
+ {
+ "epoch": 73.49046321525886,
+ "grad_norm": 1.1458752155303955,
+ "learning_rate": 3.464863201418861e-06,
+ "loss": 0.0228,
+ "step": 26971
+ },
+ {
+ "epoch": 73.49318801089919,
+ "grad_norm": 1.3995692729949951,
+ "learning_rate": 3.464195253700847e-06,
+ "loss": 0.0169,
+ "step": 26972
+ },
+ {
+ "epoch": 73.49591280653951,
+ "grad_norm": 1.6041685342788696,
+ "learning_rate": 3.4635273568834703e-06,
+ "loss": 0.0865,
+ "step": 26973
+ },
+ {
+ "epoch": 73.49863760217984,
+ "grad_norm": 1.1411067247390747,
+ "learning_rate": 3.4628595109719286e-06,
+ "loss": 0.0122,
+ "step": 26974
+ },
+ {
+ "epoch": 73.50136239782016,
+ "grad_norm": 1.0841937065124512,
+ "learning_rate": 3.46219171597143e-06,
+ "loss": 0.013,
+ "step": 26975
+ },
+ {
+ "epoch": 73.50408719346049,
+ "grad_norm": 1.8443586826324463,
+ "learning_rate": 3.4615239718871718e-06,
+ "loss": 0.0253,
+ "step": 26976
+ },
+ {
+ "epoch": 73.50681198910081,
+ "grad_norm": 1.4812121391296387,
+ "learning_rate": 3.460856278724355e-06,
+ "loss": 0.0634,
+ "step": 26977
+ },
+ {
+ "epoch": 73.50953678474114,
+ "grad_norm": 1.2648818492889404,
+ "learning_rate": 3.4601886364881754e-06,
+ "loss": 0.0218,
+ "step": 26978
+ },
+ {
+ "epoch": 73.51226158038148,
+ "grad_norm": 0.8423240780830383,
+ "learning_rate": 3.459521045183839e-06,
+ "loss": 0.0111,
+ "step": 26979
+ },
+ {
+ "epoch": 73.5149863760218,
+ "grad_norm": 1.2344391345977783,
+ "learning_rate": 3.4588535048165427e-06,
+ "loss": 0.07,
+ "step": 26980
+ },
+ {
+ "epoch": 73.51771117166213,
+ "grad_norm": 1.07499361038208,
+ "learning_rate": 3.4581860153914846e-06,
+ "loss": 0.0143,
+ "step": 26981
+ },
+ {
+ "epoch": 73.52043596730245,
+ "grad_norm": 2.2153923511505127,
+ "learning_rate": 3.4575185769138596e-06,
+ "loss": 0.0216,
+ "step": 26982
+ },
+ {
+ "epoch": 73.52316076294278,
+ "grad_norm": 1.4215158224105835,
+ "learning_rate": 3.4568511893888734e-06,
+ "loss": 0.0213,
+ "step": 26983
+ },
+ {
+ "epoch": 73.52588555858311,
+ "grad_norm": 1.8205430507659912,
+ "learning_rate": 3.4561838528217197e-06,
+ "loss": 0.0868,
+ "step": 26984
+ },
+ {
+ "epoch": 73.52861035422343,
+ "grad_norm": 2.0617756843566895,
+ "learning_rate": 3.455516567217595e-06,
+ "loss": 0.0333,
+ "step": 26985
+ },
+ {
+ "epoch": 73.53133514986376,
+ "grad_norm": 1.2313079833984375,
+ "learning_rate": 3.4548493325816933e-06,
+ "loss": 0.0257,
+ "step": 26986
+ },
+ {
+ "epoch": 73.53405994550408,
+ "grad_norm": 1.6946884393692017,
+ "learning_rate": 3.4541821489192175e-06,
+ "loss": 0.0079,
+ "step": 26987
+ },
+ {
+ "epoch": 73.53678474114442,
+ "grad_norm": 1.2124449014663696,
+ "learning_rate": 3.453515016235357e-06,
+ "loss": 0.0267,
+ "step": 26988
+ },
+ {
+ "epoch": 73.53950953678473,
+ "grad_norm": 1.324511170387268,
+ "learning_rate": 3.452847934535315e-06,
+ "loss": 0.0187,
+ "step": 26989
+ },
+ {
+ "epoch": 73.54223433242507,
+ "grad_norm": 1.5790374279022217,
+ "learning_rate": 3.452180903824279e-06,
+ "loss": 0.0383,
+ "step": 26990
+ },
+ {
+ "epoch": 73.5449591280654,
+ "grad_norm": 0.902313768863678,
+ "learning_rate": 3.4515139241074504e-06,
+ "loss": 0.0085,
+ "step": 26991
+ },
+ {
+ "epoch": 73.54768392370572,
+ "grad_norm": 0.9854360222816467,
+ "learning_rate": 3.4508469953900205e-06,
+ "loss": 0.0103,
+ "step": 26992
+ },
+ {
+ "epoch": 73.55040871934605,
+ "grad_norm": 1.4254298210144043,
+ "learning_rate": 3.4501801176771844e-06,
+ "loss": 0.0152,
+ "step": 26993
+ },
+ {
+ "epoch": 73.55313351498637,
+ "grad_norm": 1.7312108278274536,
+ "learning_rate": 3.44951329097413e-06,
+ "loss": 0.1078,
+ "step": 26994
+ },
+ {
+ "epoch": 73.5558583106267,
+ "grad_norm": 1.3564379215240479,
+ "learning_rate": 3.4488465152860593e-06,
+ "loss": 0.0129,
+ "step": 26995
+ },
+ {
+ "epoch": 73.55858310626704,
+ "grad_norm": 1.3131502866744995,
+ "learning_rate": 3.4481797906181613e-06,
+ "loss": 0.024,
+ "step": 26996
+ },
+ {
+ "epoch": 73.56130790190736,
+ "grad_norm": 1.6431033611297607,
+ "learning_rate": 3.4475131169756284e-06,
+ "loss": 0.0599,
+ "step": 26997
+ },
+ {
+ "epoch": 73.56403269754769,
+ "grad_norm": 3.182102918624878,
+ "learning_rate": 3.4468464943636483e-06,
+ "loss": 0.0505,
+ "step": 26998
+ },
+ {
+ "epoch": 73.566757493188,
+ "grad_norm": 1.2117812633514404,
+ "learning_rate": 3.4461799227874204e-06,
+ "loss": 0.0159,
+ "step": 26999
+ },
+ {
+ "epoch": 73.56948228882834,
+ "grad_norm": 1.011383295059204,
+ "learning_rate": 3.445513402252132e-06,
+ "loss": 0.0167,
+ "step": 27000
+ },
+ {
+ "epoch": 73.57220708446866,
+ "grad_norm": 1.4938158988952637,
+ "learning_rate": 3.444846932762975e-06,
+ "loss": 0.0401,
+ "step": 27001
+ },
+ {
+ "epoch": 73.57493188010899,
+ "grad_norm": 0.6678892374038696,
+ "learning_rate": 3.444180514325135e-06,
+ "loss": 0.0067,
+ "step": 27002
+ },
+ {
+ "epoch": 73.57765667574932,
+ "grad_norm": 1.4049758911132812,
+ "learning_rate": 3.4435141469438103e-06,
+ "loss": 0.0279,
+ "step": 27003
+ },
+ {
+ "epoch": 73.58038147138964,
+ "grad_norm": 1.630142092704773,
+ "learning_rate": 3.442847830624186e-06,
+ "loss": 0.0344,
+ "step": 27004
+ },
+ {
+ "epoch": 73.58310626702998,
+ "grad_norm": 1.5676254034042358,
+ "learning_rate": 3.4421815653714487e-06,
+ "loss": 0.2174,
+ "step": 27005
+ },
+ {
+ "epoch": 73.5858310626703,
+ "grad_norm": 1.7734506130218506,
+ "learning_rate": 3.4415153511907928e-06,
+ "loss": 0.0338,
+ "step": 27006
+ },
+ {
+ "epoch": 73.58855585831063,
+ "grad_norm": 1.23410964012146,
+ "learning_rate": 3.440849188087402e-06,
+ "loss": 0.0108,
+ "step": 27007
+ },
+ {
+ "epoch": 73.59128065395096,
+ "grad_norm": 1.7040746212005615,
+ "learning_rate": 3.440183076066469e-06,
+ "loss": 0.0173,
+ "step": 27008
+ },
+ {
+ "epoch": 73.59400544959128,
+ "grad_norm": 1.1833890676498413,
+ "learning_rate": 3.43951701513318e-06,
+ "loss": 0.0149,
+ "step": 27009
+ },
+ {
+ "epoch": 73.59673024523161,
+ "grad_norm": 1.549433708190918,
+ "learning_rate": 3.438851005292717e-06,
+ "loss": 0.0406,
+ "step": 27010
+ },
+ {
+ "epoch": 73.59945504087193,
+ "grad_norm": 1.7668654918670654,
+ "learning_rate": 3.438185046550275e-06,
+ "loss": 0.0971,
+ "step": 27011
+ },
+ {
+ "epoch": 73.60217983651226,
+ "grad_norm": 0.7129538655281067,
+ "learning_rate": 3.4375191389110376e-06,
+ "loss": 0.0087,
+ "step": 27012
+ },
+ {
+ "epoch": 73.60490463215258,
+ "grad_norm": 0.8599877953529358,
+ "learning_rate": 3.4368532823801893e-06,
+ "loss": 0.009,
+ "step": 27013
+ },
+ {
+ "epoch": 73.60762942779292,
+ "grad_norm": 1.584233045578003,
+ "learning_rate": 3.4361874769629123e-06,
+ "loss": 0.148,
+ "step": 27014
+ },
+ {
+ "epoch": 73.61035422343325,
+ "grad_norm": 1.4430599212646484,
+ "learning_rate": 3.4355217226644e-06,
+ "loss": 0.0183,
+ "step": 27015
+ },
+ {
+ "epoch": 73.61307901907357,
+ "grad_norm": 1.1536380052566528,
+ "learning_rate": 3.434856019489833e-06,
+ "loss": 0.008,
+ "step": 27016
+ },
+ {
+ "epoch": 73.6158038147139,
+ "grad_norm": 1.5872461795806885,
+ "learning_rate": 3.4341903674443967e-06,
+ "loss": 0.0134,
+ "step": 27017
+ },
+ {
+ "epoch": 73.61852861035422,
+ "grad_norm": 1.2278321981430054,
+ "learning_rate": 3.4335247665332704e-06,
+ "loss": 0.0343,
+ "step": 27018
+ },
+ {
+ "epoch": 73.62125340599455,
+ "grad_norm": 2.094120502471924,
+ "learning_rate": 3.432859216761645e-06,
+ "loss": 0.0404,
+ "step": 27019
+ },
+ {
+ "epoch": 73.62397820163488,
+ "grad_norm": 2.328496217727661,
+ "learning_rate": 3.432193718134701e-06,
+ "loss": 0.0253,
+ "step": 27020
+ },
+ {
+ "epoch": 73.6267029972752,
+ "grad_norm": 1.5310523509979248,
+ "learning_rate": 3.4315282706576215e-06,
+ "loss": 0.0128,
+ "step": 27021
+ },
+ {
+ "epoch": 73.62942779291554,
+ "grad_norm": 1.0742685794830322,
+ "learning_rate": 3.430862874335584e-06,
+ "loss": 0.0116,
+ "step": 27022
+ },
+ {
+ "epoch": 73.63215258855585,
+ "grad_norm": 1.0453150272369385,
+ "learning_rate": 3.4301975291737756e-06,
+ "loss": 0.0159,
+ "step": 27023
+ },
+ {
+ "epoch": 73.63487738419619,
+ "grad_norm": 0.8986366391181946,
+ "learning_rate": 3.4295322351773807e-06,
+ "loss": 0.041,
+ "step": 27024
+ },
+ {
+ "epoch": 73.6376021798365,
+ "grad_norm": 1.615122675895691,
+ "learning_rate": 3.4288669923515784e-06,
+ "loss": 0.0471,
+ "step": 27025
+ },
+ {
+ "epoch": 73.64032697547684,
+ "grad_norm": 1.103169322013855,
+ "learning_rate": 3.4282018007015472e-06,
+ "loss": 0.01,
+ "step": 27026
+ },
+ {
+ "epoch": 73.64305177111717,
+ "grad_norm": 2.5222537517547607,
+ "learning_rate": 3.4275366602324654e-06,
+ "loss": 0.0519,
+ "step": 27027
+ },
+ {
+ "epoch": 73.64577656675749,
+ "grad_norm": 0.9681851863861084,
+ "learning_rate": 3.4268715709495203e-06,
+ "loss": 0.0249,
+ "step": 27028
+ },
+ {
+ "epoch": 73.64850136239782,
+ "grad_norm": 1.9016525745391846,
+ "learning_rate": 3.4262065328578876e-06,
+ "loss": 0.0843,
+ "step": 27029
+ },
+ {
+ "epoch": 73.65122615803814,
+ "grad_norm": 1.414978265762329,
+ "learning_rate": 3.4255415459627473e-06,
+ "loss": 0.0634,
+ "step": 27030
+ },
+ {
+ "epoch": 73.65395095367847,
+ "grad_norm": 1.223704218864441,
+ "learning_rate": 3.4248766102692744e-06,
+ "loss": 0.0214,
+ "step": 27031
+ },
+ {
+ "epoch": 73.65667574931881,
+ "grad_norm": 1.2470992803573608,
+ "learning_rate": 3.4242117257826536e-06,
+ "loss": 0.0153,
+ "step": 27032
+ },
+ {
+ "epoch": 73.65940054495913,
+ "grad_norm": 1.3128334283828735,
+ "learning_rate": 3.4235468925080607e-06,
+ "loss": 0.0146,
+ "step": 27033
+ },
+ {
+ "epoch": 73.66212534059946,
+ "grad_norm": 2.1788957118988037,
+ "learning_rate": 3.4228821104506694e-06,
+ "loss": 0.1021,
+ "step": 27034
+ },
+ {
+ "epoch": 73.66485013623978,
+ "grad_norm": 1.7303118705749512,
+ "learning_rate": 3.4222173796156643e-06,
+ "loss": 0.2087,
+ "step": 27035
+ },
+ {
+ "epoch": 73.66757493188011,
+ "grad_norm": 0.5222562551498413,
+ "learning_rate": 3.4215527000082183e-06,
+ "loss": 0.0054,
+ "step": 27036
+ },
+ {
+ "epoch": 73.67029972752043,
+ "grad_norm": 1.8246971368789673,
+ "learning_rate": 3.420888071633508e-06,
+ "loss": 0.0624,
+ "step": 27037
+ },
+ {
+ "epoch": 73.67302452316076,
+ "grad_norm": 1.9763917922973633,
+ "learning_rate": 3.4202234944967062e-06,
+ "loss": 0.0427,
+ "step": 27038
+ },
+ {
+ "epoch": 73.6757493188011,
+ "grad_norm": 1.571960210800171,
+ "learning_rate": 3.4195589686029927e-06,
+ "loss": 0.0255,
+ "step": 27039
+ },
+ {
+ "epoch": 73.67847411444141,
+ "grad_norm": 1.1790450811386108,
+ "learning_rate": 3.418894493957545e-06,
+ "loss": 0.0551,
+ "step": 27040
+ },
+ {
+ "epoch": 73.68119891008175,
+ "grad_norm": 0.9328339099884033,
+ "learning_rate": 3.418230070565535e-06,
+ "loss": 0.0149,
+ "step": 27041
+ },
+ {
+ "epoch": 73.68392370572207,
+ "grad_norm": 1.5206252336502075,
+ "learning_rate": 3.417565698432138e-06,
+ "loss": 0.0146,
+ "step": 27042
+ },
+ {
+ "epoch": 73.6866485013624,
+ "grad_norm": 1.0707656145095825,
+ "learning_rate": 3.4169013775625235e-06,
+ "loss": 0.0124,
+ "step": 27043
+ },
+ {
+ "epoch": 73.68937329700273,
+ "grad_norm": 1.6653907299041748,
+ "learning_rate": 3.416237107961873e-06,
+ "loss": 0.0482,
+ "step": 27044
+ },
+ {
+ "epoch": 73.69209809264305,
+ "grad_norm": 1.3715310096740723,
+ "learning_rate": 3.415572889635356e-06,
+ "loss": 0.0184,
+ "step": 27045
+ },
+ {
+ "epoch": 73.69482288828338,
+ "grad_norm": 1.69326913356781,
+ "learning_rate": 3.414908722588145e-06,
+ "loss": 0.0293,
+ "step": 27046
+ },
+ {
+ "epoch": 73.6975476839237,
+ "grad_norm": 1.2013230323791504,
+ "learning_rate": 3.41424460682541e-06,
+ "loss": 0.0318,
+ "step": 27047
+ },
+ {
+ "epoch": 73.70027247956403,
+ "grad_norm": 1.0970932245254517,
+ "learning_rate": 3.413580542352328e-06,
+ "loss": 0.0205,
+ "step": 27048
+ },
+ {
+ "epoch": 73.70299727520435,
+ "grad_norm": 0.9455246925354004,
+ "learning_rate": 3.412916529174071e-06,
+ "loss": 0.0154,
+ "step": 27049
+ },
+ {
+ "epoch": 73.70572207084469,
+ "grad_norm": 1.3729888200759888,
+ "learning_rate": 3.4122525672958063e-06,
+ "loss": 0.1537,
+ "step": 27050
+ },
+ {
+ "epoch": 73.70844686648502,
+ "grad_norm": 1.4600746631622314,
+ "learning_rate": 3.411588656722704e-06,
+ "loss": 0.049,
+ "step": 27051
+ },
+ {
+ "epoch": 73.71117166212534,
+ "grad_norm": 1.902613639831543,
+ "learning_rate": 3.4109247974599413e-06,
+ "loss": 0.0217,
+ "step": 27052
+ },
+ {
+ "epoch": 73.71389645776567,
+ "grad_norm": 1.4964348077774048,
+ "learning_rate": 3.4102609895126824e-06,
+ "loss": 0.1237,
+ "step": 27053
+ },
+ {
+ "epoch": 73.71662125340599,
+ "grad_norm": 1.6304417848587036,
+ "learning_rate": 3.4095972328860973e-06,
+ "loss": 0.0201,
+ "step": 27054
+ },
+ {
+ "epoch": 73.71934604904632,
+ "grad_norm": 3.4002416133880615,
+ "learning_rate": 3.4089335275853585e-06,
+ "loss": 0.0131,
+ "step": 27055
+ },
+ {
+ "epoch": 73.72207084468666,
+ "grad_norm": 2.62733793258667,
+ "learning_rate": 3.408269873615635e-06,
+ "loss": 0.0628,
+ "step": 27056
+ },
+ {
+ "epoch": 73.72479564032697,
+ "grad_norm": 1.453933835029602,
+ "learning_rate": 3.407606270982089e-06,
+ "loss": 0.0216,
+ "step": 27057
+ },
+ {
+ "epoch": 73.7275204359673,
+ "grad_norm": 0.4906359016895294,
+ "learning_rate": 3.4069427196898973e-06,
+ "loss": 0.0052,
+ "step": 27058
+ },
+ {
+ "epoch": 73.73024523160763,
+ "grad_norm": 1.3630353212356567,
+ "learning_rate": 3.4062792197442197e-06,
+ "loss": 0.0677,
+ "step": 27059
+ },
+ {
+ "epoch": 73.73297002724796,
+ "grad_norm": 1.9271149635314941,
+ "learning_rate": 3.4056157711502304e-06,
+ "loss": 0.0313,
+ "step": 27060
+ },
+ {
+ "epoch": 73.73569482288828,
+ "grad_norm": 2.103055477142334,
+ "learning_rate": 3.404952373913093e-06,
+ "loss": 0.0794,
+ "step": 27061
+ },
+ {
+ "epoch": 73.73841961852861,
+ "grad_norm": 1.3908498287200928,
+ "learning_rate": 3.4042890280379748e-06,
+ "loss": 0.033,
+ "step": 27062
+ },
+ {
+ "epoch": 73.74114441416894,
+ "grad_norm": 0.8874065279960632,
+ "learning_rate": 3.403625733530038e-06,
+ "loss": 0.0114,
+ "step": 27063
+ },
+ {
+ "epoch": 73.74386920980926,
+ "grad_norm": 1.505640983581543,
+ "learning_rate": 3.402962490394456e-06,
+ "loss": 0.025,
+ "step": 27064
+ },
+ {
+ "epoch": 73.7465940054496,
+ "grad_norm": 1.2127387523651123,
+ "learning_rate": 3.402299298636388e-06,
+ "loss": 0.0396,
+ "step": 27065
+ },
+ {
+ "epoch": 73.74931880108991,
+ "grad_norm": 1.1046415567398071,
+ "learning_rate": 3.401636158261001e-06,
+ "loss": 0.0131,
+ "step": 27066
+ },
+ {
+ "epoch": 73.75204359673025,
+ "grad_norm": 1.2509124279022217,
+ "learning_rate": 3.4009730692734576e-06,
+ "loss": 0.0118,
+ "step": 27067
+ },
+ {
+ "epoch": 73.75476839237058,
+ "grad_norm": 1.2557088136672974,
+ "learning_rate": 3.4003100316789252e-06,
+ "loss": 0.0377,
+ "step": 27068
+ },
+ {
+ "epoch": 73.7574931880109,
+ "grad_norm": 1.066432237625122,
+ "learning_rate": 3.399647045482567e-06,
+ "loss": 0.0123,
+ "step": 27069
+ },
+ {
+ "epoch": 73.76021798365123,
+ "grad_norm": 1.3042656183242798,
+ "learning_rate": 3.398984110689544e-06,
+ "loss": 0.0182,
+ "step": 27070
+ },
+ {
+ "epoch": 73.76294277929155,
+ "grad_norm": 1.0550824403762817,
+ "learning_rate": 3.398321227305018e-06,
+ "loss": 0.018,
+ "step": 27071
+ },
+ {
+ "epoch": 73.76566757493188,
+ "grad_norm": 1.3152227401733398,
+ "learning_rate": 3.3976583953341567e-06,
+ "loss": 0.0111,
+ "step": 27072
+ },
+ {
+ "epoch": 73.7683923705722,
+ "grad_norm": 1.1756469011306763,
+ "learning_rate": 3.396995614782119e-06,
+ "loss": 0.0124,
+ "step": 27073
+ },
+ {
+ "epoch": 73.77111716621253,
+ "grad_norm": 1.497994065284729,
+ "learning_rate": 3.3963328856540633e-06,
+ "loss": 0.0702,
+ "step": 27074
+ },
+ {
+ "epoch": 73.77384196185287,
+ "grad_norm": 1.3820991516113281,
+ "learning_rate": 3.3956702079551563e-06,
+ "loss": 0.014,
+ "step": 27075
+ },
+ {
+ "epoch": 73.77656675749319,
+ "grad_norm": 1.1511873006820679,
+ "learning_rate": 3.3950075816905593e-06,
+ "loss": 0.0145,
+ "step": 27076
+ },
+ {
+ "epoch": 73.77929155313352,
+ "grad_norm": 0.8708515167236328,
+ "learning_rate": 3.3943450068654303e-06,
+ "loss": 0.008,
+ "step": 27077
+ },
+ {
+ "epoch": 73.78201634877384,
+ "grad_norm": 0.5857051610946655,
+ "learning_rate": 3.39368248348493e-06,
+ "loss": 0.0064,
+ "step": 27078
+ },
+ {
+ "epoch": 73.78474114441417,
+ "grad_norm": 1.7718372344970703,
+ "learning_rate": 3.3930200115542143e-06,
+ "loss": 0.0667,
+ "step": 27079
+ },
+ {
+ "epoch": 73.7874659400545,
+ "grad_norm": 1.5601534843444824,
+ "learning_rate": 3.39235759107845e-06,
+ "loss": 0.0678,
+ "step": 27080
+ },
+ {
+ "epoch": 73.79019073569482,
+ "grad_norm": 1.0120494365692139,
+ "learning_rate": 3.391695222062793e-06,
+ "loss": 0.0373,
+ "step": 27081
+ },
+ {
+ "epoch": 73.79291553133515,
+ "grad_norm": 2.082101345062256,
+ "learning_rate": 3.391032904512399e-06,
+ "loss": 0.0792,
+ "step": 27082
+ },
+ {
+ "epoch": 73.79564032697547,
+ "grad_norm": 1.4649040699005127,
+ "learning_rate": 3.3903706384324266e-06,
+ "loss": 0.0128,
+ "step": 27083
+ },
+ {
+ "epoch": 73.7983651226158,
+ "grad_norm": 0.9048721194267273,
+ "learning_rate": 3.3897084238280366e-06,
+ "loss": 0.0112,
+ "step": 27084
+ },
+ {
+ "epoch": 73.80108991825612,
+ "grad_norm": 1.1667832136154175,
+ "learning_rate": 3.389046260704385e-06,
+ "loss": 0.0124,
+ "step": 27085
+ },
+ {
+ "epoch": 73.80381471389646,
+ "grad_norm": 1.4792734384536743,
+ "learning_rate": 3.3883841490666293e-06,
+ "loss": 0.0639,
+ "step": 27086
+ },
+ {
+ "epoch": 73.80653950953679,
+ "grad_norm": 2.1483705043792725,
+ "learning_rate": 3.3877220889199213e-06,
+ "loss": 0.0555,
+ "step": 27087
+ },
+ {
+ "epoch": 73.80926430517711,
+ "grad_norm": 1.681509256362915,
+ "learning_rate": 3.387060080269423e-06,
+ "loss": 0.0234,
+ "step": 27088
+ },
+ {
+ "epoch": 73.81198910081744,
+ "grad_norm": 1.3172128200531006,
+ "learning_rate": 3.3863981231202892e-06,
+ "loss": 0.0087,
+ "step": 27089
+ },
+ {
+ "epoch": 73.81471389645776,
+ "grad_norm": 2.9946939945220947,
+ "learning_rate": 3.385736217477674e-06,
+ "loss": 0.0241,
+ "step": 27090
+ },
+ {
+ "epoch": 73.8174386920981,
+ "grad_norm": 0.9227899312973022,
+ "learning_rate": 3.3850743633467286e-06,
+ "loss": 0.0148,
+ "step": 27091
+ },
+ {
+ "epoch": 73.82016348773843,
+ "grad_norm": 1.84152090549469,
+ "learning_rate": 3.384412560732612e-06,
+ "loss": 0.0701,
+ "step": 27092
+ },
+ {
+ "epoch": 73.82288828337875,
+ "grad_norm": 1.4081165790557861,
+ "learning_rate": 3.383750809640479e-06,
+ "loss": 0.0655,
+ "step": 27093
+ },
+ {
+ "epoch": 73.82561307901908,
+ "grad_norm": 1.0578581094741821,
+ "learning_rate": 3.3830891100754837e-06,
+ "loss": 0.0925,
+ "step": 27094
+ },
+ {
+ "epoch": 73.8283378746594,
+ "grad_norm": 1.2003380060195923,
+ "learning_rate": 3.382427462042772e-06,
+ "loss": 0.0152,
+ "step": 27095
+ },
+ {
+ "epoch": 73.83106267029973,
+ "grad_norm": 1.394333839416504,
+ "learning_rate": 3.3817658655475074e-06,
+ "loss": 0.0275,
+ "step": 27096
+ },
+ {
+ "epoch": 73.83378746594005,
+ "grad_norm": 1.4145927429199219,
+ "learning_rate": 3.3811043205948366e-06,
+ "loss": 0.018,
+ "step": 27097
+ },
+ {
+ "epoch": 73.83651226158038,
+ "grad_norm": 1.2169650793075562,
+ "learning_rate": 3.3804428271899127e-06,
+ "loss": 0.0544,
+ "step": 27098
+ },
+ {
+ "epoch": 73.83923705722071,
+ "grad_norm": 1.1471956968307495,
+ "learning_rate": 3.3797813853378827e-06,
+ "loss": 0.0108,
+ "step": 27099
+ },
+ {
+ "epoch": 73.84196185286103,
+ "grad_norm": 1.2252599000930786,
+ "learning_rate": 3.3791199950439057e-06,
+ "loss": 0.0292,
+ "step": 27100
+ },
+ {
+ "epoch": 73.84468664850137,
+ "grad_norm": 1.9532212018966675,
+ "learning_rate": 3.3784586563131293e-06,
+ "loss": 0.0156,
+ "step": 27101
+ },
+ {
+ "epoch": 73.84741144414168,
+ "grad_norm": 1.1664880514144897,
+ "learning_rate": 3.3777973691507047e-06,
+ "loss": 0.0959,
+ "step": 27102
+ },
+ {
+ "epoch": 73.85013623978202,
+ "grad_norm": 1.0578514337539673,
+ "learning_rate": 3.3771361335617767e-06,
+ "loss": 0.0232,
+ "step": 27103
+ },
+ {
+ "epoch": 73.85286103542235,
+ "grad_norm": 1.418690800666809,
+ "learning_rate": 3.376474949551504e-06,
+ "loss": 0.0121,
+ "step": 27104
+ },
+ {
+ "epoch": 73.85558583106267,
+ "grad_norm": 0.7879897952079773,
+ "learning_rate": 3.375813817125029e-06,
+ "loss": 0.0073,
+ "step": 27105
+ },
+ {
+ "epoch": 73.858310626703,
+ "grad_norm": 0.8028835654258728,
+ "learning_rate": 3.3751527362875045e-06,
+ "loss": 0.0075,
+ "step": 27106
+ },
+ {
+ "epoch": 73.86103542234332,
+ "grad_norm": 1.9838515520095825,
+ "learning_rate": 3.374491707044073e-06,
+ "loss": 0.0211,
+ "step": 27107
+ },
+ {
+ "epoch": 73.86376021798365,
+ "grad_norm": 1.228370189666748,
+ "learning_rate": 3.3738307293998874e-06,
+ "loss": 0.0069,
+ "step": 27108
+ },
+ {
+ "epoch": 73.86648501362397,
+ "grad_norm": 0.8573243021965027,
+ "learning_rate": 3.373169803360098e-06,
+ "loss": 0.0082,
+ "step": 27109
+ },
+ {
+ "epoch": 73.8692098092643,
+ "grad_norm": 0.958381175994873,
+ "learning_rate": 3.3725089289298483e-06,
+ "loss": 0.0827,
+ "step": 27110
+ },
+ {
+ "epoch": 73.87193460490464,
+ "grad_norm": 1.28127920627594,
+ "learning_rate": 3.371848106114287e-06,
+ "loss": 0.0184,
+ "step": 27111
+ },
+ {
+ "epoch": 73.87465940054496,
+ "grad_norm": 1.0589410066604614,
+ "learning_rate": 3.371187334918555e-06,
+ "loss": 0.0311,
+ "step": 27112
+ },
+ {
+ "epoch": 73.87738419618529,
+ "grad_norm": 1.6659777164459229,
+ "learning_rate": 3.370526615347807e-06,
+ "loss": 0.0256,
+ "step": 27113
+ },
+ {
+ "epoch": 73.88010899182561,
+ "grad_norm": 1.0318173170089722,
+ "learning_rate": 3.369865947407184e-06,
+ "loss": 0.0104,
+ "step": 27114
+ },
+ {
+ "epoch": 73.88283378746594,
+ "grad_norm": 0.7486786842346191,
+ "learning_rate": 3.36920533110183e-06,
+ "loss": 0.0074,
+ "step": 27115
+ },
+ {
+ "epoch": 73.88555858310627,
+ "grad_norm": 0.9363235235214233,
+ "learning_rate": 3.36854476643689e-06,
+ "loss": 0.0131,
+ "step": 27116
+ },
+ {
+ "epoch": 73.88828337874659,
+ "grad_norm": 1.7827433347702026,
+ "learning_rate": 3.3678842534175126e-06,
+ "loss": 0.0474,
+ "step": 27117
+ },
+ {
+ "epoch": 73.89100817438693,
+ "grad_norm": 1.2526127099990845,
+ "learning_rate": 3.367223792048839e-06,
+ "loss": 0.0213,
+ "step": 27118
+ },
+ {
+ "epoch": 73.89373297002724,
+ "grad_norm": 0.9504752159118652,
+ "learning_rate": 3.36656338233601e-06,
+ "loss": 0.0182,
+ "step": 27119
+ },
+ {
+ "epoch": 73.89645776566758,
+ "grad_norm": 1.525449275970459,
+ "learning_rate": 3.365903024284175e-06,
+ "loss": 0.0261,
+ "step": 27120
+ },
+ {
+ "epoch": 73.8991825613079,
+ "grad_norm": 2.271817207336426,
+ "learning_rate": 3.3652427178984748e-06,
+ "loss": 0.0223,
+ "step": 27121
+ },
+ {
+ "epoch": 73.90190735694823,
+ "grad_norm": 1.6483286619186401,
+ "learning_rate": 3.3645824631840496e-06,
+ "loss": 0.0485,
+ "step": 27122
+ },
+ {
+ "epoch": 73.90463215258856,
+ "grad_norm": 1.1620711088180542,
+ "learning_rate": 3.36392226014604e-06,
+ "loss": 0.0098,
+ "step": 27123
+ },
+ {
+ "epoch": 73.90735694822888,
+ "grad_norm": 1.2472716569900513,
+ "learning_rate": 3.363262108789592e-06,
+ "loss": 0.0137,
+ "step": 27124
+ },
+ {
+ "epoch": 73.91008174386921,
+ "grad_norm": 1.8262978792190552,
+ "learning_rate": 3.3626020091198443e-06,
+ "loss": 0.0332,
+ "step": 27125
+ },
+ {
+ "epoch": 73.91280653950953,
+ "grad_norm": 1.1375837326049805,
+ "learning_rate": 3.3619419611419403e-06,
+ "loss": 0.0086,
+ "step": 27126
+ },
+ {
+ "epoch": 73.91553133514986,
+ "grad_norm": 1.1428331136703491,
+ "learning_rate": 3.3612819648610184e-06,
+ "loss": 0.0101,
+ "step": 27127
+ },
+ {
+ "epoch": 73.9182561307902,
+ "grad_norm": 1.1327046155929565,
+ "learning_rate": 3.360622020282216e-06,
+ "loss": 0.0132,
+ "step": 27128
+ },
+ {
+ "epoch": 73.92098092643052,
+ "grad_norm": 1.142771601676941,
+ "learning_rate": 3.359962127410679e-06,
+ "loss": 0.0119,
+ "step": 27129
+ },
+ {
+ "epoch": 73.92370572207085,
+ "grad_norm": 0.961137592792511,
+ "learning_rate": 3.3593022862515436e-06,
+ "loss": 0.0108,
+ "step": 27130
+ },
+ {
+ "epoch": 73.92643051771117,
+ "grad_norm": 0.8834181427955627,
+ "learning_rate": 3.358642496809947e-06,
+ "loss": 0.0091,
+ "step": 27131
+ },
+ {
+ "epoch": 73.9291553133515,
+ "grad_norm": 2.2710890769958496,
+ "learning_rate": 3.3579827590910274e-06,
+ "loss": 0.015,
+ "step": 27132
+ },
+ {
+ "epoch": 73.93188010899182,
+ "grad_norm": 1.0252459049224854,
+ "learning_rate": 3.3573230730999264e-06,
+ "loss": 0.0255,
+ "step": 27133
+ },
+ {
+ "epoch": 73.93460490463215,
+ "grad_norm": 1.5288622379302979,
+ "learning_rate": 3.3566634388417798e-06,
+ "loss": 0.0289,
+ "step": 27134
+ },
+ {
+ "epoch": 73.93732970027249,
+ "grad_norm": 1.3431135416030884,
+ "learning_rate": 3.3560038563217244e-06,
+ "loss": 0.0214,
+ "step": 27135
+ },
+ {
+ "epoch": 73.9400544959128,
+ "grad_norm": 2.04990553855896,
+ "learning_rate": 3.355344325544894e-06,
+ "loss": 0.0772,
+ "step": 27136
+ },
+ {
+ "epoch": 73.94277929155314,
+ "grad_norm": 1.0189299583435059,
+ "learning_rate": 3.354684846516432e-06,
+ "loss": 0.0089,
+ "step": 27137
+ },
+ {
+ "epoch": 73.94550408719346,
+ "grad_norm": 1.2143123149871826,
+ "learning_rate": 3.354025419241469e-06,
+ "loss": 0.0195,
+ "step": 27138
+ },
+ {
+ "epoch": 73.94822888283379,
+ "grad_norm": 2.093122959136963,
+ "learning_rate": 3.3533660437251438e-06,
+ "loss": 0.0846,
+ "step": 27139
+ },
+ {
+ "epoch": 73.95095367847412,
+ "grad_norm": 1.0623290538787842,
+ "learning_rate": 3.352706719972585e-06,
+ "loss": 0.0213,
+ "step": 27140
+ },
+ {
+ "epoch": 73.95367847411444,
+ "grad_norm": 4.883448123931885,
+ "learning_rate": 3.3520474479889365e-06,
+ "loss": 0.0168,
+ "step": 27141
+ },
+ {
+ "epoch": 73.95640326975477,
+ "grad_norm": 2.352402925491333,
+ "learning_rate": 3.3513882277793287e-06,
+ "loss": 0.0153,
+ "step": 27142
+ },
+ {
+ "epoch": 73.95912806539509,
+ "grad_norm": 2.053966522216797,
+ "learning_rate": 3.3507290593488916e-06,
+ "loss": 0.0436,
+ "step": 27143
+ },
+ {
+ "epoch": 73.96185286103542,
+ "grad_norm": 1.1868761777877808,
+ "learning_rate": 3.350069942702763e-06,
+ "loss": 0.0827,
+ "step": 27144
+ },
+ {
+ "epoch": 73.96457765667574,
+ "grad_norm": 0.8770736455917358,
+ "learning_rate": 3.349410877846079e-06,
+ "loss": 0.0112,
+ "step": 27145
+ },
+ {
+ "epoch": 73.96730245231608,
+ "grad_norm": 1.2542357444763184,
+ "learning_rate": 3.348751864783969e-06,
+ "loss": 0.0361,
+ "step": 27146
+ },
+ {
+ "epoch": 73.97002724795641,
+ "grad_norm": 1.7650741338729858,
+ "learning_rate": 3.3480929035215647e-06,
+ "loss": 0.1417,
+ "step": 27147
+ },
+ {
+ "epoch": 73.97275204359673,
+ "grad_norm": 0.7903884649276733,
+ "learning_rate": 3.3474339940639965e-06,
+ "loss": 0.009,
+ "step": 27148
+ },
+ {
+ "epoch": 73.97547683923706,
+ "grad_norm": 1.1674679517745972,
+ "learning_rate": 3.3467751364164013e-06,
+ "loss": 0.0291,
+ "step": 27149
+ },
+ {
+ "epoch": 73.97820163487738,
+ "grad_norm": 0.7200133204460144,
+ "learning_rate": 3.3461163305839074e-06,
+ "loss": 0.0105,
+ "step": 27150
+ },
+ {
+ "epoch": 73.98092643051771,
+ "grad_norm": 1.5155832767486572,
+ "learning_rate": 3.3454575765716446e-06,
+ "loss": 0.0142,
+ "step": 27151
+ },
+ {
+ "epoch": 73.98365122615803,
+ "grad_norm": 1.0488229990005493,
+ "learning_rate": 3.3447988743847402e-06,
+ "loss": 0.0193,
+ "step": 27152
+ },
+ {
+ "epoch": 73.98637602179836,
+ "grad_norm": 1.3862053155899048,
+ "learning_rate": 3.3441402240283325e-06,
+ "loss": 0.1047,
+ "step": 27153
+ },
+ {
+ "epoch": 73.9891008174387,
+ "grad_norm": 1.3631024360656738,
+ "learning_rate": 3.3434816255075464e-06,
+ "loss": 0.0258,
+ "step": 27154
+ },
+ {
+ "epoch": 73.99182561307902,
+ "grad_norm": 1.3165650367736816,
+ "learning_rate": 3.3428230788275106e-06,
+ "loss": 0.018,
+ "step": 27155
+ },
+ {
+ "epoch": 73.99455040871935,
+ "grad_norm": 1.1858011484146118,
+ "learning_rate": 3.3421645839933514e-06,
+ "loss": 0.0191,
+ "step": 27156
+ },
+ {
+ "epoch": 73.99727520435967,
+ "grad_norm": 1.7435907125473022,
+ "learning_rate": 3.3415061410102036e-06,
+ "loss": 0.0424,
+ "step": 27157
+ },
+ {
+ "epoch": 74.0,
+ "grad_norm": 1.7488386631011963,
+ "learning_rate": 3.3408477498831917e-06,
+ "loss": 0.1021,
+ "step": 27158
+ },
+ {
+ "epoch": 74.00272479564033,
+ "grad_norm": 1.226082682609558,
+ "learning_rate": 3.3401894106174427e-06,
+ "loss": 0.0121,
+ "step": 27159
+ },
+ {
+ "epoch": 74.00544959128065,
+ "grad_norm": 1.0151363611221313,
+ "learning_rate": 3.3395311232180783e-06,
+ "loss": 0.0137,
+ "step": 27160
+ },
+ {
+ "epoch": 74.00817438692098,
+ "grad_norm": 1.541085124015808,
+ "learning_rate": 3.3388728876902378e-06,
+ "loss": 0.0166,
+ "step": 27161
+ },
+ {
+ "epoch": 74.0108991825613,
+ "grad_norm": 0.9510148763656616,
+ "learning_rate": 3.3382147040390402e-06,
+ "loss": 0.0102,
+ "step": 27162
+ },
+ {
+ "epoch": 74.01362397820164,
+ "grad_norm": 2.2769365310668945,
+ "learning_rate": 3.3375565722696123e-06,
+ "loss": 0.0369,
+ "step": 27163
+ },
+ {
+ "epoch": 74.01634877384195,
+ "grad_norm": 1.881523847579956,
+ "learning_rate": 3.336898492387076e-06,
+ "loss": 0.0694,
+ "step": 27164
+ },
+ {
+ "epoch": 74.01907356948229,
+ "grad_norm": 0.9031161069869995,
+ "learning_rate": 3.3362404643965628e-06,
+ "loss": 0.0119,
+ "step": 27165
+ },
+ {
+ "epoch": 74.02179836512262,
+ "grad_norm": 1.1670020818710327,
+ "learning_rate": 3.335582488303195e-06,
+ "loss": 0.0146,
+ "step": 27166
+ },
+ {
+ "epoch": 74.02452316076294,
+ "grad_norm": 0.950164794921875,
+ "learning_rate": 3.3349245641120965e-06,
+ "loss": 0.0108,
+ "step": 27167
+ },
+ {
+ "epoch": 74.02724795640327,
+ "grad_norm": 1.1582589149475098,
+ "learning_rate": 3.334266691828387e-06,
+ "loss": 0.0092,
+ "step": 27168
+ },
+ {
+ "epoch": 74.02997275204359,
+ "grad_norm": 1.5667476654052734,
+ "learning_rate": 3.3336088714571977e-06,
+ "loss": 0.0601,
+ "step": 27169
+ },
+ {
+ "epoch": 74.03269754768392,
+ "grad_norm": 1.1961901187896729,
+ "learning_rate": 3.3329511030036476e-06,
+ "loss": 0.0136,
+ "step": 27170
+ },
+ {
+ "epoch": 74.03542234332426,
+ "grad_norm": 1.7885844707489014,
+ "learning_rate": 3.33229338647286e-06,
+ "loss": 0.1256,
+ "step": 27171
+ },
+ {
+ "epoch": 74.03814713896458,
+ "grad_norm": 1.5784988403320312,
+ "learning_rate": 3.3316357218699535e-06,
+ "loss": 0.0631,
+ "step": 27172
+ },
+ {
+ "epoch": 74.04087193460491,
+ "grad_norm": 1.7235804796218872,
+ "learning_rate": 3.330978109200057e-06,
+ "loss": 0.0643,
+ "step": 27173
+ },
+ {
+ "epoch": 74.04359673024523,
+ "grad_norm": 1.4703317880630493,
+ "learning_rate": 3.3303205484682878e-06,
+ "loss": 0.0228,
+ "step": 27174
+ },
+ {
+ "epoch": 74.04632152588556,
+ "grad_norm": 1.2474223375320435,
+ "learning_rate": 3.329663039679768e-06,
+ "loss": 0.1489,
+ "step": 27175
+ },
+ {
+ "epoch": 74.04904632152588,
+ "grad_norm": 0.7761075496673584,
+ "learning_rate": 3.329005582839614e-06,
+ "loss": 0.0064,
+ "step": 27176
+ },
+ {
+ "epoch": 74.05177111716621,
+ "grad_norm": 0.9903594255447388,
+ "learning_rate": 3.3283481779529503e-06,
+ "loss": 0.0123,
+ "step": 27177
+ },
+ {
+ "epoch": 74.05449591280654,
+ "grad_norm": 0.7939561009407043,
+ "learning_rate": 3.3276908250249e-06,
+ "loss": 0.015,
+ "step": 27178
+ },
+ {
+ "epoch": 74.05722070844686,
+ "grad_norm": 1.3965879678726196,
+ "learning_rate": 3.327033524060578e-06,
+ "loss": 0.0212,
+ "step": 27179
+ },
+ {
+ "epoch": 74.0599455040872,
+ "grad_norm": 1.0578590631484985,
+ "learning_rate": 3.3263762750651043e-06,
+ "loss": 0.014,
+ "step": 27180
+ },
+ {
+ "epoch": 74.06267029972751,
+ "grad_norm": 1.259132742881775,
+ "learning_rate": 3.325719078043593e-06,
+ "loss": 0.0149,
+ "step": 27181
+ },
+ {
+ "epoch": 74.06539509536785,
+ "grad_norm": 0.8515031933784485,
+ "learning_rate": 3.3250619330011715e-06,
+ "loss": 0.009,
+ "step": 27182
+ },
+ {
+ "epoch": 74.06811989100818,
+ "grad_norm": 2.363999366760254,
+ "learning_rate": 3.324404839942953e-06,
+ "loss": 0.0194,
+ "step": 27183
+ },
+ {
+ "epoch": 74.0708446866485,
+ "grad_norm": 0.743614912033081,
+ "learning_rate": 3.3237477988740508e-06,
+ "loss": 0.008,
+ "step": 27184
+ },
+ {
+ "epoch": 74.07356948228883,
+ "grad_norm": 2.2227108478546143,
+ "learning_rate": 3.3230908097995885e-06,
+ "loss": 0.0221,
+ "step": 27185
+ },
+ {
+ "epoch": 74.07629427792915,
+ "grad_norm": 1.6224907636642456,
+ "learning_rate": 3.322433872724681e-06,
+ "loss": 0.0219,
+ "step": 27186
+ },
+ {
+ "epoch": 74.07901907356948,
+ "grad_norm": 1.103265404701233,
+ "learning_rate": 3.3217769876544437e-06,
+ "loss": 0.0128,
+ "step": 27187
+ },
+ {
+ "epoch": 74.0817438692098,
+ "grad_norm": 1.4066033363342285,
+ "learning_rate": 3.3211201545939885e-06,
+ "loss": 0.0141,
+ "step": 27188
+ },
+ {
+ "epoch": 74.08446866485014,
+ "grad_norm": 1.0578380823135376,
+ "learning_rate": 3.320463373548438e-06,
+ "loss": 0.0125,
+ "step": 27189
+ },
+ {
+ "epoch": 74.08719346049047,
+ "grad_norm": 1.7841511964797974,
+ "learning_rate": 3.319806644522904e-06,
+ "loss": 0.0192,
+ "step": 27190
+ },
+ {
+ "epoch": 74.08991825613079,
+ "grad_norm": 2.154327392578125,
+ "learning_rate": 3.3191499675225e-06,
+ "loss": 0.0247,
+ "step": 27191
+ },
+ {
+ "epoch": 74.09264305177112,
+ "grad_norm": 1.3046305179595947,
+ "learning_rate": 3.318493342552339e-06,
+ "loss": 0.0126,
+ "step": 27192
+ },
+ {
+ "epoch": 74.09536784741144,
+ "grad_norm": 0.9774122834205627,
+ "learning_rate": 3.3178367696175396e-06,
+ "loss": 0.038,
+ "step": 27193
+ },
+ {
+ "epoch": 74.09809264305177,
+ "grad_norm": 0.9358721971511841,
+ "learning_rate": 3.3171802487232087e-06,
+ "loss": 0.0734,
+ "step": 27194
+ },
+ {
+ "epoch": 74.1008174386921,
+ "grad_norm": 1.2919687032699585,
+ "learning_rate": 3.3165237798744665e-06,
+ "loss": 0.0502,
+ "step": 27195
+ },
+ {
+ "epoch": 74.10354223433242,
+ "grad_norm": 0.8663638234138489,
+ "learning_rate": 3.3158673630764216e-06,
+ "loss": 0.012,
+ "step": 27196
+ },
+ {
+ "epoch": 74.10626702997276,
+ "grad_norm": 0.8830791711807251,
+ "learning_rate": 3.315210998334184e-06,
+ "loss": 0.0075,
+ "step": 27197
+ },
+ {
+ "epoch": 74.10899182561307,
+ "grad_norm": 1.413435935974121,
+ "learning_rate": 3.314554685652871e-06,
+ "loss": 0.0149,
+ "step": 27198
+ },
+ {
+ "epoch": 74.11171662125341,
+ "grad_norm": 0.9758617281913757,
+ "learning_rate": 3.3138984250375903e-06,
+ "loss": 0.015,
+ "step": 27199
+ },
+ {
+ "epoch": 74.11444141689373,
+ "grad_norm": 3.92372989654541,
+ "learning_rate": 3.313242216493454e-06,
+ "loss": 0.0317,
+ "step": 27200
+ },
+ {
+ "epoch": 74.11716621253406,
+ "grad_norm": 1.6892954111099243,
+ "learning_rate": 3.3125860600255686e-06,
+ "loss": 0.0204,
+ "step": 27201
+ },
+ {
+ "epoch": 74.11989100817439,
+ "grad_norm": 1.1319447755813599,
+ "learning_rate": 3.3119299556390516e-06,
+ "loss": 0.0177,
+ "step": 27202
+ },
+ {
+ "epoch": 74.12261580381471,
+ "grad_norm": 1.1394643783569336,
+ "learning_rate": 3.3112739033390083e-06,
+ "loss": 0.0079,
+ "step": 27203
+ },
+ {
+ "epoch": 74.12534059945504,
+ "grad_norm": 0.908420205116272,
+ "learning_rate": 3.3106179031305497e-06,
+ "loss": 0.0074,
+ "step": 27204
+ },
+ {
+ "epoch": 74.12806539509536,
+ "grad_norm": 3.0366342067718506,
+ "learning_rate": 3.309961955018779e-06,
+ "loss": 0.0793,
+ "step": 27205
+ },
+ {
+ "epoch": 74.1307901907357,
+ "grad_norm": 1.4616966247558594,
+ "learning_rate": 3.3093060590088132e-06,
+ "loss": 0.0567,
+ "step": 27206
+ },
+ {
+ "epoch": 74.13351498637603,
+ "grad_norm": 1.2583696842193604,
+ "learning_rate": 3.3086502151057564e-06,
+ "loss": 0.019,
+ "step": 27207
+ },
+ {
+ "epoch": 74.13623978201635,
+ "grad_norm": 2.1740121841430664,
+ "learning_rate": 3.3079944233147123e-06,
+ "loss": 0.0258,
+ "step": 27208
+ },
+ {
+ "epoch": 74.13896457765668,
+ "grad_norm": 1.4115062952041626,
+ "learning_rate": 3.307338683640796e-06,
+ "loss": 0.051,
+ "step": 27209
+ },
+ {
+ "epoch": 74.141689373297,
+ "grad_norm": 1.3158820867538452,
+ "learning_rate": 3.306682996089111e-06,
+ "loss": 0.0322,
+ "step": 27210
+ },
+ {
+ "epoch": 74.14441416893733,
+ "grad_norm": 1.3312309980392456,
+ "learning_rate": 3.30602736066476e-06,
+ "loss": 0.0857,
+ "step": 27211
+ },
+ {
+ "epoch": 74.14713896457765,
+ "grad_norm": 1.51815664768219,
+ "learning_rate": 3.3053717773728556e-06,
+ "loss": 0.1752,
+ "step": 27212
+ },
+ {
+ "epoch": 74.14986376021798,
+ "grad_norm": 1.0957505702972412,
+ "learning_rate": 3.3047162462184965e-06,
+ "loss": 0.0135,
+ "step": 27213
+ },
+ {
+ "epoch": 74.15258855585832,
+ "grad_norm": 2.867177724838257,
+ "learning_rate": 3.304060767206796e-06,
+ "loss": 0.0197,
+ "step": 27214
+ },
+ {
+ "epoch": 74.15531335149863,
+ "grad_norm": 1.0817116498947144,
+ "learning_rate": 3.3034053403428547e-06,
+ "loss": 0.0143,
+ "step": 27215
+ },
+ {
+ "epoch": 74.15803814713897,
+ "grad_norm": 2.0533595085144043,
+ "learning_rate": 3.3027499656317773e-06,
+ "loss": 0.0734,
+ "step": 27216
+ },
+ {
+ "epoch": 74.16076294277929,
+ "grad_norm": 1.7424595355987549,
+ "learning_rate": 3.302094643078664e-06,
+ "loss": 0.0148,
+ "step": 27217
+ },
+ {
+ "epoch": 74.16348773841962,
+ "grad_norm": 2.1501986980438232,
+ "learning_rate": 3.301439372688624e-06,
+ "loss": 0.0877,
+ "step": 27218
+ },
+ {
+ "epoch": 74.16621253405995,
+ "grad_norm": 0.8062586188316345,
+ "learning_rate": 3.3007841544667606e-06,
+ "loss": 0.0058,
+ "step": 27219
+ },
+ {
+ "epoch": 74.16893732970027,
+ "grad_norm": 0.816270649433136,
+ "learning_rate": 3.300128988418173e-06,
+ "loss": 0.0091,
+ "step": 27220
+ },
+ {
+ "epoch": 74.1716621253406,
+ "grad_norm": 1.2384343147277832,
+ "learning_rate": 3.2994738745479628e-06,
+ "loss": 0.0148,
+ "step": 27221
+ },
+ {
+ "epoch": 74.17438692098092,
+ "grad_norm": 2.3011608123779297,
+ "learning_rate": 3.2988188128612374e-06,
+ "loss": 0.0221,
+ "step": 27222
+ },
+ {
+ "epoch": 74.17711171662125,
+ "grad_norm": 0.5819554924964905,
+ "learning_rate": 3.2981638033630956e-06,
+ "loss": 0.006,
+ "step": 27223
+ },
+ {
+ "epoch": 74.17983651226157,
+ "grad_norm": 1.2201111316680908,
+ "learning_rate": 3.297508846058639e-06,
+ "loss": 0.0273,
+ "step": 27224
+ },
+ {
+ "epoch": 74.1825613079019,
+ "grad_norm": 0.7510902285575867,
+ "learning_rate": 3.2968539409529632e-06,
+ "loss": 0.0071,
+ "step": 27225
+ },
+ {
+ "epoch": 74.18528610354224,
+ "grad_norm": 1.3798680305480957,
+ "learning_rate": 3.2961990880511773e-06,
+ "loss": 0.032,
+ "step": 27226
+ },
+ {
+ "epoch": 74.18801089918256,
+ "grad_norm": 1.507961392402649,
+ "learning_rate": 3.2955442873583765e-06,
+ "loss": 0.026,
+ "step": 27227
+ },
+ {
+ "epoch": 74.19073569482289,
+ "grad_norm": 1.1863131523132324,
+ "learning_rate": 3.2948895388796585e-06,
+ "loss": 0.0086,
+ "step": 27228
+ },
+ {
+ "epoch": 74.19346049046321,
+ "grad_norm": 1.1975879669189453,
+ "learning_rate": 3.2942348426201244e-06,
+ "loss": 0.0116,
+ "step": 27229
+ },
+ {
+ "epoch": 74.19618528610354,
+ "grad_norm": 0.8374161124229431,
+ "learning_rate": 3.293580198584877e-06,
+ "loss": 0.0098,
+ "step": 27230
+ },
+ {
+ "epoch": 74.19891008174388,
+ "grad_norm": 1.0511962175369263,
+ "learning_rate": 3.2929256067790117e-06,
+ "loss": 0.0171,
+ "step": 27231
+ },
+ {
+ "epoch": 74.2016348773842,
+ "grad_norm": 1.9075944423675537,
+ "learning_rate": 3.292271067207625e-06,
+ "loss": 0.1481,
+ "step": 27232
+ },
+ {
+ "epoch": 74.20435967302453,
+ "grad_norm": 3.2185630798339844,
+ "learning_rate": 3.2916165798758127e-06,
+ "loss": 0.0461,
+ "step": 27233
+ },
+ {
+ "epoch": 74.20708446866485,
+ "grad_norm": 1.3018298149108887,
+ "learning_rate": 3.2909621447886773e-06,
+ "loss": 0.0105,
+ "step": 27234
+ },
+ {
+ "epoch": 74.20980926430518,
+ "grad_norm": 2.0988664627075195,
+ "learning_rate": 3.2903077619513135e-06,
+ "loss": 0.0287,
+ "step": 27235
+ },
+ {
+ "epoch": 74.2125340599455,
+ "grad_norm": 1.4232304096221924,
+ "learning_rate": 3.2896534313688165e-06,
+ "loss": 0.0269,
+ "step": 27236
+ },
+ {
+ "epoch": 74.21525885558583,
+ "grad_norm": 1.4132493734359741,
+ "learning_rate": 3.2889991530462793e-06,
+ "loss": 0.0335,
+ "step": 27237
+ },
+ {
+ "epoch": 74.21798365122616,
+ "grad_norm": 1.2567343711853027,
+ "learning_rate": 3.2883449269888034e-06,
+ "loss": 0.0178,
+ "step": 27238
+ },
+ {
+ "epoch": 74.22070844686648,
+ "grad_norm": 1.379630446434021,
+ "learning_rate": 3.287690753201482e-06,
+ "loss": 0.0169,
+ "step": 27239
+ },
+ {
+ "epoch": 74.22343324250681,
+ "grad_norm": 1.505889654159546,
+ "learning_rate": 3.2870366316894086e-06,
+ "loss": 0.0164,
+ "step": 27240
+ },
+ {
+ "epoch": 74.22615803814713,
+ "grad_norm": 1.34681236743927,
+ "learning_rate": 3.2863825624576752e-06,
+ "loss": 0.0709,
+ "step": 27241
+ },
+ {
+ "epoch": 74.22888283378747,
+ "grad_norm": 2.3469390869140625,
+ "learning_rate": 3.2857285455113807e-06,
+ "loss": 0.065,
+ "step": 27242
+ },
+ {
+ "epoch": 74.2316076294278,
+ "grad_norm": 0.6288487315177917,
+ "learning_rate": 3.2850745808556174e-06,
+ "loss": 0.0067,
+ "step": 27243
+ },
+ {
+ "epoch": 74.23433242506812,
+ "grad_norm": 1.3096333742141724,
+ "learning_rate": 3.284420668495476e-06,
+ "loss": 0.0347,
+ "step": 27244
+ },
+ {
+ "epoch": 74.23705722070845,
+ "grad_norm": 1.459423542022705,
+ "learning_rate": 3.2837668084360476e-06,
+ "loss": 0.0275,
+ "step": 27245
+ },
+ {
+ "epoch": 74.23978201634877,
+ "grad_norm": 1.1493809223175049,
+ "learning_rate": 3.2831130006824274e-06,
+ "loss": 0.0189,
+ "step": 27246
+ },
+ {
+ "epoch": 74.2425068119891,
+ "grad_norm": 1.3017375469207764,
+ "learning_rate": 3.2824592452397097e-06,
+ "loss": 0.0198,
+ "step": 27247
+ },
+ {
+ "epoch": 74.24523160762942,
+ "grad_norm": 2.1431422233581543,
+ "learning_rate": 3.2818055421129837e-06,
+ "loss": 0.0276,
+ "step": 27248
+ },
+ {
+ "epoch": 74.24795640326975,
+ "grad_norm": 1.9879058599472046,
+ "learning_rate": 3.281151891307336e-06,
+ "loss": 0.0143,
+ "step": 27249
+ },
+ {
+ "epoch": 74.25068119891009,
+ "grad_norm": 0.8970801830291748,
+ "learning_rate": 3.280498292827864e-06,
+ "loss": 0.0107,
+ "step": 27250
+ },
+ {
+ "epoch": 74.2534059945504,
+ "grad_norm": 1.4814649820327759,
+ "learning_rate": 3.2798447466796557e-06,
+ "loss": 0.0353,
+ "step": 27251
+ },
+ {
+ "epoch": 74.25613079019074,
+ "grad_norm": 1.4669604301452637,
+ "learning_rate": 3.2791912528678004e-06,
+ "loss": 0.0614,
+ "step": 27252
+ },
+ {
+ "epoch": 74.25885558583106,
+ "grad_norm": 0.849877655506134,
+ "learning_rate": 3.2785378113973843e-06,
+ "loss": 0.0124,
+ "step": 27253
+ },
+ {
+ "epoch": 74.26158038147139,
+ "grad_norm": 0.8832796812057495,
+ "learning_rate": 3.277884422273502e-06,
+ "loss": 0.0154,
+ "step": 27254
+ },
+ {
+ "epoch": 74.26430517711172,
+ "grad_norm": 1.3905080556869507,
+ "learning_rate": 3.277231085501239e-06,
+ "loss": 0.0212,
+ "step": 27255
+ },
+ {
+ "epoch": 74.26702997275204,
+ "grad_norm": 0.5823195576667786,
+ "learning_rate": 3.2765778010856854e-06,
+ "loss": 0.0055,
+ "step": 27256
+ },
+ {
+ "epoch": 74.26975476839237,
+ "grad_norm": 0.7499882578849792,
+ "learning_rate": 3.275924569031923e-06,
+ "loss": 0.0089,
+ "step": 27257
+ },
+ {
+ "epoch": 74.2724795640327,
+ "grad_norm": 1.2342901229858398,
+ "learning_rate": 3.275271389345047e-06,
+ "loss": 0.0167,
+ "step": 27258
+ },
+ {
+ "epoch": 74.27520435967303,
+ "grad_norm": 1.3229867219924927,
+ "learning_rate": 3.2746182620301416e-06,
+ "loss": 0.1421,
+ "step": 27259
+ },
+ {
+ "epoch": 74.27792915531334,
+ "grad_norm": 1.5674673318862915,
+ "learning_rate": 3.2739651870922926e-06,
+ "loss": 0.0153,
+ "step": 27260
+ },
+ {
+ "epoch": 74.28065395095368,
+ "grad_norm": 1.4340795278549194,
+ "learning_rate": 3.273312164536583e-06,
+ "loss": 0.0178,
+ "step": 27261
+ },
+ {
+ "epoch": 74.28337874659401,
+ "grad_norm": 1.4173429012298584,
+ "learning_rate": 3.272659194368104e-06,
+ "loss": 0.0355,
+ "step": 27262
+ },
+ {
+ "epoch": 74.28610354223433,
+ "grad_norm": 1.1087369918823242,
+ "learning_rate": 3.2720062765919368e-06,
+ "loss": 0.0169,
+ "step": 27263
+ },
+ {
+ "epoch": 74.28882833787466,
+ "grad_norm": 1.1180269718170166,
+ "learning_rate": 3.2713534112131705e-06,
+ "loss": 0.0104,
+ "step": 27264
+ },
+ {
+ "epoch": 74.29155313351498,
+ "grad_norm": 2.1938979625701904,
+ "learning_rate": 3.2707005982368877e-06,
+ "loss": 0.0512,
+ "step": 27265
+ },
+ {
+ "epoch": 74.29427792915531,
+ "grad_norm": 0.9733609557151794,
+ "learning_rate": 3.2700478376681686e-06,
+ "loss": 0.0099,
+ "step": 27266
+ },
+ {
+ "epoch": 74.29700272479565,
+ "grad_norm": 1.0881295204162598,
+ "learning_rate": 3.269395129512104e-06,
+ "loss": 0.0301,
+ "step": 27267
+ },
+ {
+ "epoch": 74.29972752043597,
+ "grad_norm": 1.8399423360824585,
+ "learning_rate": 3.268742473773774e-06,
+ "loss": 0.0234,
+ "step": 27268
+ },
+ {
+ "epoch": 74.3024523160763,
+ "grad_norm": 1.6228519678115845,
+ "learning_rate": 3.2680898704582562e-06,
+ "loss": 0.0351,
+ "step": 27269
+ },
+ {
+ "epoch": 74.30517711171662,
+ "grad_norm": 1.4450457096099854,
+ "learning_rate": 3.267437319570642e-06,
+ "loss": 0.0353,
+ "step": 27270
+ },
+ {
+ "epoch": 74.30790190735695,
+ "grad_norm": 1.7596248388290405,
+ "learning_rate": 3.2667848211160093e-06,
+ "loss": 0.0247,
+ "step": 27271
+ },
+ {
+ "epoch": 74.31062670299727,
+ "grad_norm": 1.3378651142120361,
+ "learning_rate": 3.26613237509944e-06,
+ "loss": 0.0272,
+ "step": 27272
+ },
+ {
+ "epoch": 74.3133514986376,
+ "grad_norm": 1.3025684356689453,
+ "learning_rate": 3.265479981526011e-06,
+ "loss": 0.0727,
+ "step": 27273
+ },
+ {
+ "epoch": 74.31607629427793,
+ "grad_norm": 0.9805086255073547,
+ "learning_rate": 3.2648276404008104e-06,
+ "loss": 0.057,
+ "step": 27274
+ },
+ {
+ "epoch": 74.31880108991825,
+ "grad_norm": 0.9274822473526001,
+ "learning_rate": 3.2641753517289164e-06,
+ "loss": 0.0086,
+ "step": 27275
+ },
+ {
+ "epoch": 74.32152588555859,
+ "grad_norm": 2.4538514614105225,
+ "learning_rate": 3.2635231155154067e-06,
+ "loss": 0.0485,
+ "step": 27276
+ },
+ {
+ "epoch": 74.3242506811989,
+ "grad_norm": 1.4736027717590332,
+ "learning_rate": 3.2628709317653593e-06,
+ "loss": 0.0243,
+ "step": 27277
+ },
+ {
+ "epoch": 74.32697547683924,
+ "grad_norm": 1.5016143321990967,
+ "learning_rate": 3.2622188004838594e-06,
+ "loss": 0.0186,
+ "step": 27278
+ },
+ {
+ "epoch": 74.32970027247957,
+ "grad_norm": 1.0337475538253784,
+ "learning_rate": 3.2615667216759837e-06,
+ "loss": 0.013,
+ "step": 27279
+ },
+ {
+ "epoch": 74.33242506811989,
+ "grad_norm": 14.256217002868652,
+ "learning_rate": 3.260914695346805e-06,
+ "loss": 0.0413,
+ "step": 27280
+ },
+ {
+ "epoch": 74.33514986376022,
+ "grad_norm": 0.9772511124610901,
+ "learning_rate": 3.2602627215014093e-06,
+ "loss": 0.0192,
+ "step": 27281
+ },
+ {
+ "epoch": 74.33787465940054,
+ "grad_norm": 1.5535460710525513,
+ "learning_rate": 3.259610800144868e-06,
+ "loss": 0.0188,
+ "step": 27282
+ },
+ {
+ "epoch": 74.34059945504087,
+ "grad_norm": 1.5439765453338623,
+ "learning_rate": 3.2589589312822634e-06,
+ "loss": 0.1624,
+ "step": 27283
+ },
+ {
+ "epoch": 74.34332425068119,
+ "grad_norm": 3.706313133239746,
+ "learning_rate": 3.2583071149186695e-06,
+ "loss": 0.0708,
+ "step": 27284
+ },
+ {
+ "epoch": 74.34604904632153,
+ "grad_norm": 1.3510853052139282,
+ "learning_rate": 3.2576553510591636e-06,
+ "loss": 0.0515,
+ "step": 27285
+ },
+ {
+ "epoch": 74.34877384196186,
+ "grad_norm": 0.9124401807785034,
+ "learning_rate": 3.257003639708818e-06,
+ "loss": 0.0837,
+ "step": 27286
+ },
+ {
+ "epoch": 74.35149863760218,
+ "grad_norm": 1.2421891689300537,
+ "learning_rate": 3.256351980872713e-06,
+ "loss": 0.0451,
+ "step": 27287
+ },
+ {
+ "epoch": 74.35422343324251,
+ "grad_norm": 1.8306083679199219,
+ "learning_rate": 3.2557003745559224e-06,
+ "loss": 0.0135,
+ "step": 27288
+ },
+ {
+ "epoch": 74.35694822888283,
+ "grad_norm": 1.3683935403823853,
+ "learning_rate": 3.25504882076352e-06,
+ "loss": 0.0334,
+ "step": 27289
+ },
+ {
+ "epoch": 74.35967302452316,
+ "grad_norm": 1.6018978357315063,
+ "learning_rate": 3.2543973195005764e-06,
+ "loss": 0.0345,
+ "step": 27290
+ },
+ {
+ "epoch": 74.3623978201635,
+ "grad_norm": 0.627051591873169,
+ "learning_rate": 3.2537458707721735e-06,
+ "loss": 0.0062,
+ "step": 27291
+ },
+ {
+ "epoch": 74.36512261580381,
+ "grad_norm": 1.496442198753357,
+ "learning_rate": 3.2530944745833803e-06,
+ "loss": 0.0754,
+ "step": 27292
+ },
+ {
+ "epoch": 74.36784741144415,
+ "grad_norm": 1.3514983654022217,
+ "learning_rate": 3.2524431309392667e-06,
+ "loss": 0.0182,
+ "step": 27293
+ },
+ {
+ "epoch": 74.37057220708446,
+ "grad_norm": 1.2077058553695679,
+ "learning_rate": 3.2517918398449133e-06,
+ "loss": 0.0129,
+ "step": 27294
+ },
+ {
+ "epoch": 74.3732970027248,
+ "grad_norm": 0.7439311742782593,
+ "learning_rate": 3.251140601305387e-06,
+ "loss": 0.0082,
+ "step": 27295
+ },
+ {
+ "epoch": 74.37602179836512,
+ "grad_norm": 1.0166934728622437,
+ "learning_rate": 3.2504894153257603e-06,
+ "loss": 0.1399,
+ "step": 27296
+ },
+ {
+ "epoch": 74.37874659400545,
+ "grad_norm": 0.8364735841751099,
+ "learning_rate": 3.2498382819111017e-06,
+ "loss": 0.0106,
+ "step": 27297
+ },
+ {
+ "epoch": 74.38147138964578,
+ "grad_norm": 1.8350789546966553,
+ "learning_rate": 3.2491872010664847e-06,
+ "loss": 0.1541,
+ "step": 27298
+ },
+ {
+ "epoch": 74.3841961852861,
+ "grad_norm": 1.7329390048980713,
+ "learning_rate": 3.2485361727969846e-06,
+ "loss": 0.0423,
+ "step": 27299
+ },
+ {
+ "epoch": 74.38692098092643,
+ "grad_norm": 0.986598789691925,
+ "learning_rate": 3.2478851971076675e-06,
+ "loss": 0.0769,
+ "step": 27300
+ },
+ {
+ "epoch": 74.38964577656675,
+ "grad_norm": 1.2879257202148438,
+ "learning_rate": 3.2472342740036033e-06,
+ "loss": 0.0227,
+ "step": 27301
+ },
+ {
+ "epoch": 74.39237057220708,
+ "grad_norm": 1.9901185035705566,
+ "learning_rate": 3.246583403489857e-06,
+ "loss": 0.0244,
+ "step": 27302
+ },
+ {
+ "epoch": 74.39509536784742,
+ "grad_norm": 0.7263370156288147,
+ "learning_rate": 3.2459325855715063e-06,
+ "loss": 0.0076,
+ "step": 27303
+ },
+ {
+ "epoch": 74.39782016348774,
+ "grad_norm": 0.7219961881637573,
+ "learning_rate": 3.245281820253614e-06,
+ "loss": 0.0085,
+ "step": 27304
+ },
+ {
+ "epoch": 74.40054495912807,
+ "grad_norm": 0.9923708438873291,
+ "learning_rate": 3.2446311075412508e-06,
+ "loss": 0.0324,
+ "step": 27305
+ },
+ {
+ "epoch": 74.40326975476839,
+ "grad_norm": 1.9160230159759521,
+ "learning_rate": 3.243980447439479e-06,
+ "loss": 0.0306,
+ "step": 27306
+ },
+ {
+ "epoch": 74.40599455040872,
+ "grad_norm": 1.433785319328308,
+ "learning_rate": 3.243329839953373e-06,
+ "loss": 0.0174,
+ "step": 27307
+ },
+ {
+ "epoch": 74.40871934604904,
+ "grad_norm": 1.1386616230010986,
+ "learning_rate": 3.2426792850879974e-06,
+ "loss": 0.0096,
+ "step": 27308
+ },
+ {
+ "epoch": 74.41144414168937,
+ "grad_norm": 1.543028473854065,
+ "learning_rate": 3.2420287828484175e-06,
+ "loss": 0.0261,
+ "step": 27309
+ },
+ {
+ "epoch": 74.4141689373297,
+ "grad_norm": 1.1622967720031738,
+ "learning_rate": 3.2413783332396977e-06,
+ "loss": 0.0587,
+ "step": 27310
+ },
+ {
+ "epoch": 74.41689373297002,
+ "grad_norm": 1.2146718502044678,
+ "learning_rate": 3.240727936266909e-06,
+ "loss": 0.0176,
+ "step": 27311
+ },
+ {
+ "epoch": 74.41961852861036,
+ "grad_norm": 0.8269326686859131,
+ "learning_rate": 3.2400775919351123e-06,
+ "loss": 0.0099,
+ "step": 27312
+ },
+ {
+ "epoch": 74.42234332425068,
+ "grad_norm": 1.3574824333190918,
+ "learning_rate": 3.2394273002493747e-06,
+ "loss": 0.0304,
+ "step": 27313
+ },
+ {
+ "epoch": 74.42506811989101,
+ "grad_norm": 1.7794851064682007,
+ "learning_rate": 3.2387770612147562e-06,
+ "loss": 0.0178,
+ "step": 27314
+ },
+ {
+ "epoch": 74.42779291553134,
+ "grad_norm": 1.6390048265457153,
+ "learning_rate": 3.238126874836325e-06,
+ "loss": 0.0199,
+ "step": 27315
+ },
+ {
+ "epoch": 74.43051771117166,
+ "grad_norm": 1.2468630075454712,
+ "learning_rate": 3.2374767411191468e-06,
+ "loss": 0.0088,
+ "step": 27316
+ },
+ {
+ "epoch": 74.433242506812,
+ "grad_norm": 0.8565782904624939,
+ "learning_rate": 3.2368266600682817e-06,
+ "loss": 0.0084,
+ "step": 27317
+ },
+ {
+ "epoch": 74.43596730245231,
+ "grad_norm": 0.9933147430419922,
+ "learning_rate": 3.23617663168879e-06,
+ "loss": 0.0112,
+ "step": 27318
+ },
+ {
+ "epoch": 74.43869209809264,
+ "grad_norm": 1.0590054988861084,
+ "learning_rate": 3.23552665598574e-06,
+ "loss": 0.0457,
+ "step": 27319
+ },
+ {
+ "epoch": 74.44141689373296,
+ "grad_norm": 1.7256277799606323,
+ "learning_rate": 3.2348767329641908e-06,
+ "loss": 0.029,
+ "step": 27320
+ },
+ {
+ "epoch": 74.4441416893733,
+ "grad_norm": 1.2401717901229858,
+ "learning_rate": 3.2342268626292037e-06,
+ "loss": 0.0631,
+ "step": 27321
+ },
+ {
+ "epoch": 74.44686648501363,
+ "grad_norm": 1.1524362564086914,
+ "learning_rate": 3.233577044985837e-06,
+ "loss": 0.0136,
+ "step": 27322
+ },
+ {
+ "epoch": 74.44959128065395,
+ "grad_norm": 1.4244015216827393,
+ "learning_rate": 3.2329272800391586e-06,
+ "loss": 0.0501,
+ "step": 27323
+ },
+ {
+ "epoch": 74.45231607629428,
+ "grad_norm": 1.5028743743896484,
+ "learning_rate": 3.232277567794223e-06,
+ "loss": 0.0168,
+ "step": 27324
+ },
+ {
+ "epoch": 74.4550408719346,
+ "grad_norm": 2.8605661392211914,
+ "learning_rate": 3.231627908256093e-06,
+ "loss": 0.0339,
+ "step": 27325
+ },
+ {
+ "epoch": 74.45776566757493,
+ "grad_norm": 1.340378761291504,
+ "learning_rate": 3.2309783014298235e-06,
+ "loss": 0.0156,
+ "step": 27326
+ },
+ {
+ "epoch": 74.46049046321527,
+ "grad_norm": 1.3536059856414795,
+ "learning_rate": 3.2303287473204813e-06,
+ "loss": 0.0185,
+ "step": 27327
+ },
+ {
+ "epoch": 74.46321525885558,
+ "grad_norm": 1.9340413808822632,
+ "learning_rate": 3.2296792459331196e-06,
+ "loss": 0.0187,
+ "step": 27328
+ },
+ {
+ "epoch": 74.46594005449592,
+ "grad_norm": 1.9248932600021362,
+ "learning_rate": 3.229029797272799e-06,
+ "loss": 0.1159,
+ "step": 27329
+ },
+ {
+ "epoch": 74.46866485013624,
+ "grad_norm": 0.8824494481086731,
+ "learning_rate": 3.228380401344573e-06,
+ "loss": 0.0116,
+ "step": 27330
+ },
+ {
+ "epoch": 74.47138964577657,
+ "grad_norm": 1.1317598819732666,
+ "learning_rate": 3.2277310581535046e-06,
+ "loss": 0.0186,
+ "step": 27331
+ },
+ {
+ "epoch": 74.47411444141689,
+ "grad_norm": 1.757906436920166,
+ "learning_rate": 3.2270817677046463e-06,
+ "loss": 0.122,
+ "step": 27332
+ },
+ {
+ "epoch": 74.47683923705722,
+ "grad_norm": 1.1530004739761353,
+ "learning_rate": 3.226432530003061e-06,
+ "loss": 0.0117,
+ "step": 27333
+ },
+ {
+ "epoch": 74.47956403269755,
+ "grad_norm": 1.1146302223205566,
+ "learning_rate": 3.225783345053797e-06,
+ "loss": 0.0094,
+ "step": 27334
+ },
+ {
+ "epoch": 74.48228882833787,
+ "grad_norm": 1.725182294845581,
+ "learning_rate": 3.2251342128619177e-06,
+ "loss": 0.0117,
+ "step": 27335
+ },
+ {
+ "epoch": 74.4850136239782,
+ "grad_norm": 1.039455533027649,
+ "learning_rate": 3.2244851334324755e-06,
+ "loss": 0.0088,
+ "step": 27336
+ },
+ {
+ "epoch": 74.48773841961852,
+ "grad_norm": 0.49563169479370117,
+ "learning_rate": 3.223836106770525e-06,
+ "loss": 0.0053,
+ "step": 27337
+ },
+ {
+ "epoch": 74.49046321525886,
+ "grad_norm": 0.971458375453949,
+ "learning_rate": 3.2231871328811182e-06,
+ "loss": 0.0133,
+ "step": 27338
+ },
+ {
+ "epoch": 74.49318801089919,
+ "grad_norm": 1.0585170984268188,
+ "learning_rate": 3.2225382117693137e-06,
+ "loss": 0.0101,
+ "step": 27339
+ },
+ {
+ "epoch": 74.49591280653951,
+ "grad_norm": 1.1118857860565186,
+ "learning_rate": 3.2218893434401645e-06,
+ "loss": 0.0164,
+ "step": 27340
+ },
+ {
+ "epoch": 74.49863760217984,
+ "grad_norm": 1.599461555480957,
+ "learning_rate": 3.2212405278987226e-06,
+ "loss": 0.0243,
+ "step": 27341
+ },
+ {
+ "epoch": 74.50136239782016,
+ "grad_norm": 0.8198654651641846,
+ "learning_rate": 3.2205917651500373e-06,
+ "loss": 0.015,
+ "step": 27342
+ },
+ {
+ "epoch": 74.50408719346049,
+ "grad_norm": 1.4109526872634888,
+ "learning_rate": 3.2199430551991695e-06,
+ "loss": 0.0332,
+ "step": 27343
+ },
+ {
+ "epoch": 74.50681198910081,
+ "grad_norm": 2.0997023582458496,
+ "learning_rate": 3.219294398051167e-06,
+ "loss": 0.0677,
+ "step": 27344
+ },
+ {
+ "epoch": 74.50953678474114,
+ "grad_norm": 1.8743031024932861,
+ "learning_rate": 3.2186457937110803e-06,
+ "loss": 0.0134,
+ "step": 27345
+ },
+ {
+ "epoch": 74.51226158038148,
+ "grad_norm": 1.3797036409378052,
+ "learning_rate": 3.21799724218396e-06,
+ "loss": 0.0234,
+ "step": 27346
+ },
+ {
+ "epoch": 74.5149863760218,
+ "grad_norm": 0.9664953351020813,
+ "learning_rate": 3.217348743474862e-06,
+ "loss": 0.0245,
+ "step": 27347
+ },
+ {
+ "epoch": 74.51771117166213,
+ "grad_norm": 1.2342463731765747,
+ "learning_rate": 3.216700297588833e-06,
+ "loss": 0.0468,
+ "step": 27348
+ },
+ {
+ "epoch": 74.52043596730245,
+ "grad_norm": 1.2307884693145752,
+ "learning_rate": 3.2160519045309203e-06,
+ "loss": 0.0975,
+ "step": 27349
+ },
+ {
+ "epoch": 74.52316076294278,
+ "grad_norm": 2.0814666748046875,
+ "learning_rate": 3.2154035643061808e-06,
+ "loss": 0.1773,
+ "step": 27350
+ },
+ {
+ "epoch": 74.52588555858311,
+ "grad_norm": 1.345716953277588,
+ "learning_rate": 3.214755276919657e-06,
+ "loss": 0.0284,
+ "step": 27351
+ },
+ {
+ "epoch": 74.52861035422343,
+ "grad_norm": 1.0225416421890259,
+ "learning_rate": 3.2141070423764043e-06,
+ "loss": 0.0131,
+ "step": 27352
+ },
+ {
+ "epoch": 74.53133514986376,
+ "grad_norm": 1.1590853929519653,
+ "learning_rate": 3.2134588606814665e-06,
+ "loss": 0.0185,
+ "step": 27353
+ },
+ {
+ "epoch": 74.53405994550408,
+ "grad_norm": 1.1200270652770996,
+ "learning_rate": 3.2128107318398925e-06,
+ "loss": 0.0118,
+ "step": 27354
+ },
+ {
+ "epoch": 74.53678474114442,
+ "grad_norm": 0.8220010995864868,
+ "learning_rate": 3.2121626558567277e-06,
+ "loss": 0.0081,
+ "step": 27355
+ },
+ {
+ "epoch": 74.53950953678473,
+ "grad_norm": 1.5838385820388794,
+ "learning_rate": 3.211514632737025e-06,
+ "loss": 0.0167,
+ "step": 27356
+ },
+ {
+ "epoch": 74.54223433242507,
+ "grad_norm": 1.259786605834961,
+ "learning_rate": 3.2108666624858276e-06,
+ "loss": 0.01,
+ "step": 27357
+ },
+ {
+ "epoch": 74.5449591280654,
+ "grad_norm": 0.9607710838317871,
+ "learning_rate": 3.210218745108179e-06,
+ "loss": 0.0091,
+ "step": 27358
+ },
+ {
+ "epoch": 74.54768392370572,
+ "grad_norm": 0.7903945446014404,
+ "learning_rate": 3.2095708806091306e-06,
+ "loss": 0.0107,
+ "step": 27359
+ },
+ {
+ "epoch": 74.55040871934605,
+ "grad_norm": 1.2220563888549805,
+ "learning_rate": 3.208923068993727e-06,
+ "loss": 0.0147,
+ "step": 27360
+ },
+ {
+ "epoch": 74.55313351498637,
+ "grad_norm": 1.1452851295471191,
+ "learning_rate": 3.2082753102670105e-06,
+ "loss": 0.0313,
+ "step": 27361
+ },
+ {
+ "epoch": 74.5558583106267,
+ "grad_norm": 1.2209229469299316,
+ "learning_rate": 3.207627604434025e-06,
+ "loss": 0.0261,
+ "step": 27362
+ },
+ {
+ "epoch": 74.55858310626704,
+ "grad_norm": 1.8207201957702637,
+ "learning_rate": 3.20697995149982e-06,
+ "loss": 0.0128,
+ "step": 27363
+ },
+ {
+ "epoch": 74.56130790190736,
+ "grad_norm": 1.2522846460342407,
+ "learning_rate": 3.2063323514694365e-06,
+ "loss": 0.0161,
+ "step": 27364
+ },
+ {
+ "epoch": 74.56403269754769,
+ "grad_norm": 1.3213590383529663,
+ "learning_rate": 3.2056848043479184e-06,
+ "loss": 0.0384,
+ "step": 27365
+ },
+ {
+ "epoch": 74.566757493188,
+ "grad_norm": 1.9171991348266602,
+ "learning_rate": 3.2050373101403053e-06,
+ "loss": 0.0943,
+ "step": 27366
+ },
+ {
+ "epoch": 74.56948228882834,
+ "grad_norm": 1.4820101261138916,
+ "learning_rate": 3.2043898688516428e-06,
+ "loss": 0.0252,
+ "step": 27367
+ },
+ {
+ "epoch": 74.57220708446866,
+ "grad_norm": 1.1522531509399414,
+ "learning_rate": 3.2037424804869765e-06,
+ "loss": 0.0123,
+ "step": 27368
+ },
+ {
+ "epoch": 74.57493188010899,
+ "grad_norm": 0.6279683709144592,
+ "learning_rate": 3.203095145051345e-06,
+ "loss": 0.0074,
+ "step": 27369
+ },
+ {
+ "epoch": 74.57765667574932,
+ "grad_norm": 1.0701682567596436,
+ "learning_rate": 3.202447862549789e-06,
+ "loss": 0.0139,
+ "step": 27370
+ },
+ {
+ "epoch": 74.58038147138964,
+ "grad_norm": 1.8164750337600708,
+ "learning_rate": 3.2018006329873475e-06,
+ "loss": 0.045,
+ "step": 27371
+ },
+ {
+ "epoch": 74.58310626702998,
+ "grad_norm": 1.3156073093414307,
+ "learning_rate": 3.201153456369067e-06,
+ "loss": 0.016,
+ "step": 27372
+ },
+ {
+ "epoch": 74.5858310626703,
+ "grad_norm": 1.6521493196487427,
+ "learning_rate": 3.2005063326999853e-06,
+ "loss": 0.0412,
+ "step": 27373
+ },
+ {
+ "epoch": 74.58855585831063,
+ "grad_norm": 0.8164673447608948,
+ "learning_rate": 3.199859261985141e-06,
+ "loss": 0.0088,
+ "step": 27374
+ },
+ {
+ "epoch": 74.59128065395096,
+ "grad_norm": 1.0036506652832031,
+ "learning_rate": 3.1992122442295702e-06,
+ "loss": 0.012,
+ "step": 27375
+ },
+ {
+ "epoch": 74.59400544959128,
+ "grad_norm": 1.596022129058838,
+ "learning_rate": 3.1985652794383193e-06,
+ "loss": 0.0609,
+ "step": 27376
+ },
+ {
+ "epoch": 74.59673024523161,
+ "grad_norm": 1.6595988273620605,
+ "learning_rate": 3.197918367616423e-06,
+ "loss": 0.024,
+ "step": 27377
+ },
+ {
+ "epoch": 74.59945504087193,
+ "grad_norm": 1.1594094038009644,
+ "learning_rate": 3.197271508768919e-06,
+ "loss": 0.0076,
+ "step": 27378
+ },
+ {
+ "epoch": 74.60217983651226,
+ "grad_norm": 0.5168235898017883,
+ "learning_rate": 3.1966247029008433e-06,
+ "loss": 0.0069,
+ "step": 27379
+ },
+ {
+ "epoch": 74.60490463215258,
+ "grad_norm": 0.7911115884780884,
+ "learning_rate": 3.1959779500172382e-06,
+ "loss": 0.0081,
+ "step": 27380
+ },
+ {
+ "epoch": 74.60762942779292,
+ "grad_norm": 1.6182332038879395,
+ "learning_rate": 3.195331250123138e-06,
+ "loss": 0.0469,
+ "step": 27381
+ },
+ {
+ "epoch": 74.61035422343325,
+ "grad_norm": 1.7980672121047974,
+ "learning_rate": 3.1946846032235754e-06,
+ "loss": 0.0159,
+ "step": 27382
+ },
+ {
+ "epoch": 74.61307901907357,
+ "grad_norm": 1.5025063753128052,
+ "learning_rate": 3.1940380093235932e-06,
+ "loss": 0.0311,
+ "step": 27383
+ },
+ {
+ "epoch": 74.6158038147139,
+ "grad_norm": 1.2742695808410645,
+ "learning_rate": 3.1933914684282206e-06,
+ "loss": 0.019,
+ "step": 27384
+ },
+ {
+ "epoch": 74.61852861035422,
+ "grad_norm": 1.58059561252594,
+ "learning_rate": 3.1927449805425005e-06,
+ "loss": 0.0218,
+ "step": 27385
+ },
+ {
+ "epoch": 74.62125340599455,
+ "grad_norm": 1.0802963972091675,
+ "learning_rate": 3.1920985456714625e-06,
+ "loss": 0.0113,
+ "step": 27386
+ },
+ {
+ "epoch": 74.62397820163488,
+ "grad_norm": 1.7387642860412598,
+ "learning_rate": 3.191452163820139e-06,
+ "loss": 0.0556,
+ "step": 27387
+ },
+ {
+ "epoch": 74.6267029972752,
+ "grad_norm": 1.5668971538543701,
+ "learning_rate": 3.19080583499357e-06,
+ "loss": 0.0193,
+ "step": 27388
+ },
+ {
+ "epoch": 74.62942779291554,
+ "grad_norm": 0.8838499188423157,
+ "learning_rate": 3.190159559196786e-06,
+ "loss": 0.0102,
+ "step": 27389
+ },
+ {
+ "epoch": 74.63215258855585,
+ "grad_norm": 0.745925784111023,
+ "learning_rate": 3.1895133364348208e-06,
+ "loss": 0.0073,
+ "step": 27390
+ },
+ {
+ "epoch": 74.63487738419619,
+ "grad_norm": 1.088210940361023,
+ "learning_rate": 3.1888671667127026e-06,
+ "loss": 0.06,
+ "step": 27391
+ },
+ {
+ "epoch": 74.6376021798365,
+ "grad_norm": 1.19389808177948,
+ "learning_rate": 3.1882210500354705e-06,
+ "loss": 0.0572,
+ "step": 27392
+ },
+ {
+ "epoch": 74.64032697547684,
+ "grad_norm": 1.5817097425460815,
+ "learning_rate": 3.187574986408155e-06,
+ "loss": 0.0649,
+ "step": 27393
+ },
+ {
+ "epoch": 74.64305177111717,
+ "grad_norm": 1.5925174951553345,
+ "learning_rate": 3.1869289758357845e-06,
+ "loss": 0.039,
+ "step": 27394
+ },
+ {
+ "epoch": 74.64577656675749,
+ "grad_norm": 1.2844396829605103,
+ "learning_rate": 3.1862830183233907e-06,
+ "loss": 0.0111,
+ "step": 27395
+ },
+ {
+ "epoch": 74.64850136239782,
+ "grad_norm": 0.9070179462432861,
+ "learning_rate": 3.185637113876007e-06,
+ "loss": 0.0118,
+ "step": 27396
+ },
+ {
+ "epoch": 74.65122615803814,
+ "grad_norm": 1.2016223669052124,
+ "learning_rate": 3.184991262498662e-06,
+ "loss": 0.0123,
+ "step": 27397
+ },
+ {
+ "epoch": 74.65395095367847,
+ "grad_norm": 1.2516803741455078,
+ "learning_rate": 3.1843454641963866e-06,
+ "loss": 0.0778,
+ "step": 27398
+ },
+ {
+ "epoch": 74.65667574931881,
+ "grad_norm": 0.9432477355003357,
+ "learning_rate": 3.183699718974207e-06,
+ "loss": 0.0106,
+ "step": 27399
+ },
+ {
+ "epoch": 74.65940054495913,
+ "grad_norm": 2.258064031600952,
+ "learning_rate": 3.183054026837157e-06,
+ "loss": 0.029,
+ "step": 27400
+ },
+ {
+ "epoch": 74.66212534059946,
+ "grad_norm": 1.404349684715271,
+ "learning_rate": 3.182408387790259e-06,
+ "loss": 0.0194,
+ "step": 27401
+ },
+ {
+ "epoch": 74.66485013623978,
+ "grad_norm": 1.4032467603683472,
+ "learning_rate": 3.181762801838548e-06,
+ "loss": 0.0127,
+ "step": 27402
+ },
+ {
+ "epoch": 74.66757493188011,
+ "grad_norm": 1.2700711488723755,
+ "learning_rate": 3.181117268987046e-06,
+ "loss": 0.0086,
+ "step": 27403
+ },
+ {
+ "epoch": 74.67029972752043,
+ "grad_norm": 2.34871768951416,
+ "learning_rate": 3.1804717892407877e-06,
+ "loss": 0.0641,
+ "step": 27404
+ },
+ {
+ "epoch": 74.67302452316076,
+ "grad_norm": 1.2661776542663574,
+ "learning_rate": 3.1798263626047944e-06,
+ "loss": 0.0954,
+ "step": 27405
+ },
+ {
+ "epoch": 74.6757493188011,
+ "grad_norm": 1.7813700437545776,
+ "learning_rate": 3.1791809890840942e-06,
+ "loss": 0.0342,
+ "step": 27406
+ },
+ {
+ "epoch": 74.67847411444141,
+ "grad_norm": 0.8637391924858093,
+ "learning_rate": 3.1785356686837097e-06,
+ "loss": 0.0947,
+ "step": 27407
+ },
+ {
+ "epoch": 74.68119891008175,
+ "grad_norm": 1.5954509973526,
+ "learning_rate": 3.177890401408673e-06,
+ "loss": 0.0129,
+ "step": 27408
+ },
+ {
+ "epoch": 74.68392370572207,
+ "grad_norm": 1.3425986766815186,
+ "learning_rate": 3.1772451872640066e-06,
+ "loss": 0.0183,
+ "step": 27409
+ },
+ {
+ "epoch": 74.6866485013624,
+ "grad_norm": 1.6594383716583252,
+ "learning_rate": 3.1766000262547347e-06,
+ "loss": 0.0655,
+ "step": 27410
+ },
+ {
+ "epoch": 74.68937329700273,
+ "grad_norm": 1.5662589073181152,
+ "learning_rate": 3.1759549183858797e-06,
+ "loss": 0.0701,
+ "step": 27411
+ },
+ {
+ "epoch": 74.69209809264305,
+ "grad_norm": 1.0678707361221313,
+ "learning_rate": 3.17530986366247e-06,
+ "loss": 0.0122,
+ "step": 27412
+ },
+ {
+ "epoch": 74.69482288828338,
+ "grad_norm": 1.6372933387756348,
+ "learning_rate": 3.174664862089528e-06,
+ "loss": 0.0134,
+ "step": 27413
+ },
+ {
+ "epoch": 74.6975476839237,
+ "grad_norm": 1.6506811380386353,
+ "learning_rate": 3.174019913672077e-06,
+ "loss": 0.0853,
+ "step": 27414
+ },
+ {
+ "epoch": 74.70027247956403,
+ "grad_norm": 1.329553484916687,
+ "learning_rate": 3.1733750184151357e-06,
+ "loss": 0.02,
+ "step": 27415
+ },
+ {
+ "epoch": 74.70299727520435,
+ "grad_norm": 1.3629258871078491,
+ "learning_rate": 3.172730176323733e-06,
+ "loss": 0.0151,
+ "step": 27416
+ },
+ {
+ "epoch": 74.70572207084469,
+ "grad_norm": 1.3079556226730347,
+ "learning_rate": 3.172085387402888e-06,
+ "loss": 0.0176,
+ "step": 27417
+ },
+ {
+ "epoch": 74.70844686648502,
+ "grad_norm": 1.2978333234786987,
+ "learning_rate": 3.171440651657619e-06,
+ "loss": 0.0184,
+ "step": 27418
+ },
+ {
+ "epoch": 74.71117166212534,
+ "grad_norm": 1.7797549962997437,
+ "learning_rate": 3.1707959690929545e-06,
+ "loss": 0.0152,
+ "step": 27419
+ },
+ {
+ "epoch": 74.71389645776567,
+ "grad_norm": 1.155711054801941,
+ "learning_rate": 3.1701513397139072e-06,
+ "loss": 0.0093,
+ "step": 27420
+ },
+ {
+ "epoch": 74.71662125340599,
+ "grad_norm": 1.7703969478607178,
+ "learning_rate": 3.169506763525505e-06,
+ "loss": 0.0193,
+ "step": 27421
+ },
+ {
+ "epoch": 74.71934604904632,
+ "grad_norm": 1.1283836364746094,
+ "learning_rate": 3.168862240532764e-06,
+ "loss": 0.0233,
+ "step": 27422
+ },
+ {
+ "epoch": 74.72207084468666,
+ "grad_norm": 1.7711827754974365,
+ "learning_rate": 3.1682177707407e-06,
+ "loss": 0.1007,
+ "step": 27423
+ },
+ {
+ "epoch": 74.72479564032697,
+ "grad_norm": 2.743496894836426,
+ "learning_rate": 3.167573354154341e-06,
+ "loss": 0.0243,
+ "step": 27424
+ },
+ {
+ "epoch": 74.7275204359673,
+ "grad_norm": 1.0688656568527222,
+ "learning_rate": 3.1669289907786994e-06,
+ "loss": 0.0096,
+ "step": 27425
+ },
+ {
+ "epoch": 74.73024523160763,
+ "grad_norm": 2.0037007331848145,
+ "learning_rate": 3.1662846806187952e-06,
+ "loss": 0.0743,
+ "step": 27426
+ },
+ {
+ "epoch": 74.73297002724796,
+ "grad_norm": 1.9070940017700195,
+ "learning_rate": 3.1656404236796436e-06,
+ "loss": 0.0168,
+ "step": 27427
+ },
+ {
+ "epoch": 74.73569482288828,
+ "grad_norm": 1.5756051540374756,
+ "learning_rate": 3.1649962199662665e-06,
+ "loss": 0.0879,
+ "step": 27428
+ },
+ {
+ "epoch": 74.73841961852861,
+ "grad_norm": 1.3854097127914429,
+ "learning_rate": 3.1643520694836793e-06,
+ "loss": 0.1571,
+ "step": 27429
+ },
+ {
+ "epoch": 74.74114441416894,
+ "grad_norm": 1.2564258575439453,
+ "learning_rate": 3.163707972236899e-06,
+ "loss": 0.0205,
+ "step": 27430
+ },
+ {
+ "epoch": 74.74386920980926,
+ "grad_norm": 1.7083858251571655,
+ "learning_rate": 3.163063928230937e-06,
+ "loss": 0.0605,
+ "step": 27431
+ },
+ {
+ "epoch": 74.7465940054496,
+ "grad_norm": 2.4970788955688477,
+ "learning_rate": 3.1624199374708163e-06,
+ "loss": 0.1067,
+ "step": 27432
+ },
+ {
+ "epoch": 74.74931880108991,
+ "grad_norm": 1.0723369121551514,
+ "learning_rate": 3.1617759999615495e-06,
+ "loss": 0.0115,
+ "step": 27433
+ },
+ {
+ "epoch": 74.75204359673025,
+ "grad_norm": 1.3572447299957275,
+ "learning_rate": 3.161132115708151e-06,
+ "loss": 0.032,
+ "step": 27434
+ },
+ {
+ "epoch": 74.75476839237058,
+ "grad_norm": 1.9044092893600464,
+ "learning_rate": 3.1604882847156314e-06,
+ "loss": 0.1104,
+ "step": 27435
+ },
+ {
+ "epoch": 74.7574931880109,
+ "grad_norm": 1.7311437129974365,
+ "learning_rate": 3.1598445069890093e-06,
+ "loss": 0.0178,
+ "step": 27436
+ },
+ {
+ "epoch": 74.76021798365123,
+ "grad_norm": 1.2545503377914429,
+ "learning_rate": 3.1592007825333015e-06,
+ "loss": 0.0075,
+ "step": 27437
+ },
+ {
+ "epoch": 74.76294277929155,
+ "grad_norm": 1.251584768295288,
+ "learning_rate": 3.1585571113535186e-06,
+ "loss": 0.0322,
+ "step": 27438
+ },
+ {
+ "epoch": 74.76566757493188,
+ "grad_norm": 1.3679919242858887,
+ "learning_rate": 3.1579134934546716e-06,
+ "loss": 0.0461,
+ "step": 27439
+ },
+ {
+ "epoch": 74.7683923705722,
+ "grad_norm": 1.2608850002288818,
+ "learning_rate": 3.1572699288417708e-06,
+ "loss": 0.0787,
+ "step": 27440
+ },
+ {
+ "epoch": 74.77111716621253,
+ "grad_norm": 1.2901504039764404,
+ "learning_rate": 3.156626417519836e-06,
+ "loss": 0.0177,
+ "step": 27441
+ },
+ {
+ "epoch": 74.77384196185287,
+ "grad_norm": 1.692442774772644,
+ "learning_rate": 3.1559829594938728e-06,
+ "loss": 0.2082,
+ "step": 27442
+ },
+ {
+ "epoch": 74.77656675749319,
+ "grad_norm": 1.580064058303833,
+ "learning_rate": 3.1553395547688914e-06,
+ "loss": 0.0486,
+ "step": 27443
+ },
+ {
+ "epoch": 74.77929155313352,
+ "grad_norm": 1.4037350416183472,
+ "learning_rate": 3.1546962033499084e-06,
+ "loss": 0.0322,
+ "step": 27444
+ },
+ {
+ "epoch": 74.78201634877384,
+ "grad_norm": 1.4034113883972168,
+ "learning_rate": 3.1540529052419323e-06,
+ "loss": 0.0134,
+ "step": 27445
+ },
+ {
+ "epoch": 74.78474114441417,
+ "grad_norm": 2.0111045837402344,
+ "learning_rate": 3.153409660449971e-06,
+ "loss": 0.1774,
+ "step": 27446
+ },
+ {
+ "epoch": 74.7874659400545,
+ "grad_norm": 1.398220419883728,
+ "learning_rate": 3.1527664689790316e-06,
+ "loss": 0.0176,
+ "step": 27447
+ },
+ {
+ "epoch": 74.79019073569482,
+ "grad_norm": 1.0099724531173706,
+ "learning_rate": 3.1521233308341292e-06,
+ "loss": 0.0094,
+ "step": 27448
+ },
+ {
+ "epoch": 74.79291553133515,
+ "grad_norm": 5.58784294128418,
+ "learning_rate": 3.1514802460202707e-06,
+ "loss": 0.0308,
+ "step": 27449
+ },
+ {
+ "epoch": 74.79564032697547,
+ "grad_norm": 1.5878221988677979,
+ "learning_rate": 3.1508372145424637e-06,
+ "loss": 0.0138,
+ "step": 27450
+ },
+ {
+ "epoch": 74.7983651226158,
+ "grad_norm": 1.4374638795852661,
+ "learning_rate": 3.1501942364057116e-06,
+ "loss": 0.0349,
+ "step": 27451
+ },
+ {
+ "epoch": 74.80108991825612,
+ "grad_norm": 1.1660943031311035,
+ "learning_rate": 3.1495513116150276e-06,
+ "loss": 0.0202,
+ "step": 27452
+ },
+ {
+ "epoch": 74.80381471389646,
+ "grad_norm": 1.9635212421417236,
+ "learning_rate": 3.14890844017542e-06,
+ "loss": 0.0324,
+ "step": 27453
+ },
+ {
+ "epoch": 74.80653950953679,
+ "grad_norm": 1.2503153085708618,
+ "learning_rate": 3.1482656220918938e-06,
+ "loss": 0.0141,
+ "step": 27454
+ },
+ {
+ "epoch": 74.80926430517711,
+ "grad_norm": 1.123518943786621,
+ "learning_rate": 3.1476228573694533e-06,
+ "loss": 0.0196,
+ "step": 27455
+ },
+ {
+ "epoch": 74.81198910081744,
+ "grad_norm": 1.5508253574371338,
+ "learning_rate": 3.1469801460131035e-06,
+ "loss": 0.1039,
+ "step": 27456
+ },
+ {
+ "epoch": 74.81471389645776,
+ "grad_norm": 1.2114272117614746,
+ "learning_rate": 3.146337488027854e-06,
+ "loss": 0.0149,
+ "step": 27457
+ },
+ {
+ "epoch": 74.8174386920981,
+ "grad_norm": 0.6122527122497559,
+ "learning_rate": 3.1456948834187083e-06,
+ "loss": 0.0067,
+ "step": 27458
+ },
+ {
+ "epoch": 74.82016348773843,
+ "grad_norm": 1.0828717947006226,
+ "learning_rate": 3.14505233219067e-06,
+ "loss": 0.0152,
+ "step": 27459
+ },
+ {
+ "epoch": 74.82288828337875,
+ "grad_norm": 1.8870315551757812,
+ "learning_rate": 3.1444098343487394e-06,
+ "loss": 0.113,
+ "step": 27460
+ },
+ {
+ "epoch": 74.82561307901908,
+ "grad_norm": 0.9535418748855591,
+ "learning_rate": 3.1437673898979282e-06,
+ "loss": 0.0105,
+ "step": 27461
+ },
+ {
+ "epoch": 74.8283378746594,
+ "grad_norm": 1.1225186586380005,
+ "learning_rate": 3.143124998843237e-06,
+ "loss": 0.0088,
+ "step": 27462
+ },
+ {
+ "epoch": 74.83106267029973,
+ "grad_norm": 1.8334614038467407,
+ "learning_rate": 3.1424826611896664e-06,
+ "loss": 0.155,
+ "step": 27463
+ },
+ {
+ "epoch": 74.83378746594005,
+ "grad_norm": 1.1888242959976196,
+ "learning_rate": 3.1418403769422166e-06,
+ "loss": 0.0097,
+ "step": 27464
+ },
+ {
+ "epoch": 74.83651226158038,
+ "grad_norm": 2.7455644607543945,
+ "learning_rate": 3.1411981461058973e-06,
+ "loss": 0.054,
+ "step": 27465
+ },
+ {
+ "epoch": 74.83923705722071,
+ "grad_norm": 3.777496576309204,
+ "learning_rate": 3.1405559686857058e-06,
+ "loss": 0.0089,
+ "step": 27466
+ },
+ {
+ "epoch": 74.84196185286103,
+ "grad_norm": 1.1415283679962158,
+ "learning_rate": 3.1399138446866405e-06,
+ "loss": 0.0149,
+ "step": 27467
+ },
+ {
+ "epoch": 74.84468664850137,
+ "grad_norm": 1.6729930639266968,
+ "learning_rate": 3.139271774113708e-06,
+ "loss": 0.0276,
+ "step": 27468
+ },
+ {
+ "epoch": 74.84741144414168,
+ "grad_norm": 1.2726584672927856,
+ "learning_rate": 3.138629756971907e-06,
+ "loss": 0.0106,
+ "step": 27469
+ },
+ {
+ "epoch": 74.85013623978202,
+ "grad_norm": 2.1425118446350098,
+ "learning_rate": 3.137987793266233e-06,
+ "loss": 0.0361,
+ "step": 27470
+ },
+ {
+ "epoch": 74.85286103542235,
+ "grad_norm": 1.6409001350402832,
+ "learning_rate": 3.1373458830016925e-06,
+ "loss": 0.0306,
+ "step": 27471
+ },
+ {
+ "epoch": 74.85558583106267,
+ "grad_norm": 1.3039836883544922,
+ "learning_rate": 3.136704026183278e-06,
+ "loss": 0.0152,
+ "step": 27472
+ },
+ {
+ "epoch": 74.858310626703,
+ "grad_norm": 0.8707895874977112,
+ "learning_rate": 3.136062222815994e-06,
+ "loss": 0.0058,
+ "step": 27473
+ },
+ {
+ "epoch": 74.86103542234332,
+ "grad_norm": 1.8951053619384766,
+ "learning_rate": 3.135420472904839e-06,
+ "loss": 0.1432,
+ "step": 27474
+ },
+ {
+ "epoch": 74.86376021798365,
+ "grad_norm": 1.660677433013916,
+ "learning_rate": 3.1347787764548064e-06,
+ "loss": 0.0267,
+ "step": 27475
+ },
+ {
+ "epoch": 74.86648501362397,
+ "grad_norm": 1.49832022190094,
+ "learning_rate": 3.1341371334708924e-06,
+ "loss": 0.0277,
+ "step": 27476
+ },
+ {
+ "epoch": 74.8692098092643,
+ "grad_norm": 1.6999804973602295,
+ "learning_rate": 3.1334955439581016e-06,
+ "loss": 0.0342,
+ "step": 27477
+ },
+ {
+ "epoch": 74.87193460490464,
+ "grad_norm": 1.8410056829452515,
+ "learning_rate": 3.1328540079214264e-06,
+ "loss": 0.0237,
+ "step": 27478
+ },
+ {
+ "epoch": 74.87465940054496,
+ "grad_norm": 1.388145923614502,
+ "learning_rate": 3.132212525365863e-06,
+ "loss": 0.0338,
+ "step": 27479
+ },
+ {
+ "epoch": 74.87738419618529,
+ "grad_norm": 1.3137457370758057,
+ "learning_rate": 3.131571096296404e-06,
+ "loss": 0.1031,
+ "step": 27480
+ },
+ {
+ "epoch": 74.88010899182561,
+ "grad_norm": 1.2762166261672974,
+ "learning_rate": 3.130929720718051e-06,
+ "loss": 0.0222,
+ "step": 27481
+ },
+ {
+ "epoch": 74.88283378746594,
+ "grad_norm": 1.0012710094451904,
+ "learning_rate": 3.130288398635797e-06,
+ "loss": 0.0107,
+ "step": 27482
+ },
+ {
+ "epoch": 74.88555858310627,
+ "grad_norm": 4.520012855529785,
+ "learning_rate": 3.129647130054636e-06,
+ "loss": 0.0575,
+ "step": 27483
+ },
+ {
+ "epoch": 74.88828337874659,
+ "grad_norm": 1.4592812061309814,
+ "learning_rate": 3.129005914979558e-06,
+ "loss": 0.0175,
+ "step": 27484
+ },
+ {
+ "epoch": 74.89100817438693,
+ "grad_norm": 1.62852942943573,
+ "learning_rate": 3.128364753415565e-06,
+ "loss": 0.0318,
+ "step": 27485
+ },
+ {
+ "epoch": 74.89373297002724,
+ "grad_norm": 1.5148123502731323,
+ "learning_rate": 3.127723645367645e-06,
+ "loss": 0.0671,
+ "step": 27486
+ },
+ {
+ "epoch": 74.89645776566758,
+ "grad_norm": 1.3158811330795288,
+ "learning_rate": 3.1270825908407897e-06,
+ "loss": 0.0421,
+ "step": 27487
+ },
+ {
+ "epoch": 74.8991825613079,
+ "grad_norm": 0.9996064305305481,
+ "learning_rate": 3.1264415898399935e-06,
+ "loss": 0.0089,
+ "step": 27488
+ },
+ {
+ "epoch": 74.90190735694823,
+ "grad_norm": 1.0669364929199219,
+ "learning_rate": 3.1258006423702536e-06,
+ "loss": 0.0114,
+ "step": 27489
+ },
+ {
+ "epoch": 74.90463215258856,
+ "grad_norm": 1.6519755125045776,
+ "learning_rate": 3.125159748436556e-06,
+ "loss": 0.1045,
+ "step": 27490
+ },
+ {
+ "epoch": 74.90735694822888,
+ "grad_norm": 1.05765962600708,
+ "learning_rate": 3.124518908043892e-06,
+ "loss": 0.0094,
+ "step": 27491
+ },
+ {
+ "epoch": 74.91008174386921,
+ "grad_norm": 0.9751759767532349,
+ "learning_rate": 3.123878121197251e-06,
+ "loss": 0.0106,
+ "step": 27492
+ },
+ {
+ "epoch": 74.91280653950953,
+ "grad_norm": 1.0797268152236938,
+ "learning_rate": 3.12323738790163e-06,
+ "loss": 0.0349,
+ "step": 27493
+ },
+ {
+ "epoch": 74.91553133514986,
+ "grad_norm": 1.1697791814804077,
+ "learning_rate": 3.1225967081620136e-06,
+ "loss": 0.0082,
+ "step": 27494
+ },
+ {
+ "epoch": 74.9182561307902,
+ "grad_norm": 0.6268850564956665,
+ "learning_rate": 3.121956081983394e-06,
+ "loss": 0.0061,
+ "step": 27495
+ },
+ {
+ "epoch": 74.92098092643052,
+ "grad_norm": 1.7336231470108032,
+ "learning_rate": 3.1213155093707547e-06,
+ "loss": 0.0189,
+ "step": 27496
+ },
+ {
+ "epoch": 74.92370572207085,
+ "grad_norm": 1.7240456342697144,
+ "learning_rate": 3.1206749903290924e-06,
+ "loss": 0.1462,
+ "step": 27497
+ },
+ {
+ "epoch": 74.92643051771117,
+ "grad_norm": 1.6494721174240112,
+ "learning_rate": 3.120034524863392e-06,
+ "loss": 0.0133,
+ "step": 27498
+ },
+ {
+ "epoch": 74.9291553133515,
+ "grad_norm": 1.8886693716049194,
+ "learning_rate": 3.119394112978641e-06,
+ "loss": 0.0572,
+ "step": 27499
+ },
+ {
+ "epoch": 74.93188010899182,
+ "grad_norm": 1.1115940809249878,
+ "learning_rate": 3.1187537546798243e-06,
+ "loss": 0.0119,
+ "step": 27500
+ },
+ {
+ "epoch": 74.93460490463215,
+ "grad_norm": 1.3049099445343018,
+ "learning_rate": 3.1181134499719346e-06,
+ "loss": 0.007,
+ "step": 27501
+ },
+ {
+ "epoch": 74.93732970027249,
+ "grad_norm": 1.202776312828064,
+ "learning_rate": 3.117473198859957e-06,
+ "loss": 0.0234,
+ "step": 27502
+ },
+ {
+ "epoch": 74.9400544959128,
+ "grad_norm": 1.9578888416290283,
+ "learning_rate": 3.1168330013488756e-06,
+ "loss": 0.0666,
+ "step": 27503
+ },
+ {
+ "epoch": 74.94277929155314,
+ "grad_norm": 1.296241044998169,
+ "learning_rate": 3.116192857443674e-06,
+ "loss": 0.1155,
+ "step": 27504
+ },
+ {
+ "epoch": 74.94550408719346,
+ "grad_norm": 1.2171552181243896,
+ "learning_rate": 3.1155527671493414e-06,
+ "loss": 0.0196,
+ "step": 27505
+ },
+ {
+ "epoch": 74.94822888283379,
+ "grad_norm": 1.9332282543182373,
+ "learning_rate": 3.114912730470865e-06,
+ "loss": 0.0083,
+ "step": 27506
+ },
+ {
+ "epoch": 74.95095367847412,
+ "grad_norm": 1.368833065032959,
+ "learning_rate": 3.1142727474132275e-06,
+ "loss": 0.0182,
+ "step": 27507
+ },
+ {
+ "epoch": 74.95367847411444,
+ "grad_norm": 1.2285470962524414,
+ "learning_rate": 3.113632817981408e-06,
+ "loss": 0.0132,
+ "step": 27508
+ },
+ {
+ "epoch": 74.95640326975477,
+ "grad_norm": 1.4123634099960327,
+ "learning_rate": 3.1129929421803982e-06,
+ "loss": 0.0177,
+ "step": 27509
+ },
+ {
+ "epoch": 74.95912806539509,
+ "grad_norm": 1.6903724670410156,
+ "learning_rate": 3.112353120015178e-06,
+ "loss": 0.02,
+ "step": 27510
+ },
+ {
+ "epoch": 74.96185286103542,
+ "grad_norm": 1.1647722721099854,
+ "learning_rate": 3.1117133514907294e-06,
+ "loss": 0.0247,
+ "step": 27511
+ },
+ {
+ "epoch": 74.96457765667574,
+ "grad_norm": 1.5225493907928467,
+ "learning_rate": 3.111073636612032e-06,
+ "loss": 0.0433,
+ "step": 27512
+ },
+ {
+ "epoch": 74.96730245231608,
+ "grad_norm": 1.1477662324905396,
+ "learning_rate": 3.1104339753840752e-06,
+ "loss": 0.0117,
+ "step": 27513
+ },
+ {
+ "epoch": 74.97002724795641,
+ "grad_norm": 2.336057424545288,
+ "learning_rate": 3.109794367811837e-06,
+ "loss": 0.1558,
+ "step": 27514
+ },
+ {
+ "epoch": 74.97275204359673,
+ "grad_norm": 1.3191337585449219,
+ "learning_rate": 3.109154813900298e-06,
+ "loss": 0.021,
+ "step": 27515
+ },
+ {
+ "epoch": 74.97547683923706,
+ "grad_norm": 1.2672045230865479,
+ "learning_rate": 3.108515313654438e-06,
+ "loss": 0.0455,
+ "step": 27516
+ },
+ {
+ "epoch": 74.97820163487738,
+ "grad_norm": 1.866294026374817,
+ "learning_rate": 3.1078758670792397e-06,
+ "loss": 0.0833,
+ "step": 27517
+ },
+ {
+ "epoch": 74.98092643051771,
+ "grad_norm": 1.2771520614624023,
+ "learning_rate": 3.1072364741796833e-06,
+ "loss": 0.0125,
+ "step": 27518
+ },
+ {
+ "epoch": 74.98365122615803,
+ "grad_norm": 1.5041974782943726,
+ "learning_rate": 3.1065971349607484e-06,
+ "loss": 0.0207,
+ "step": 27519
+ },
+ {
+ "epoch": 74.98637602179836,
+ "grad_norm": 1.589084506034851,
+ "learning_rate": 3.105957849427409e-06,
+ "loss": 0.0654,
+ "step": 27520
+ },
+ {
+ "epoch": 74.9891008174387,
+ "grad_norm": 1.3187096118927002,
+ "learning_rate": 3.1053186175846474e-06,
+ "loss": 0.0117,
+ "step": 27521
+ },
+ {
+ "epoch": 74.99182561307902,
+ "grad_norm": 1.5569778680801392,
+ "learning_rate": 3.104679439437447e-06,
+ "loss": 0.0744,
+ "step": 27522
+ },
+ {
+ "epoch": 74.99455040871935,
+ "grad_norm": 1.961440920829773,
+ "learning_rate": 3.1040403149907805e-06,
+ "loss": 0.0464,
+ "step": 27523
+ },
+ {
+ "epoch": 74.99727520435967,
+ "grad_norm": 0.9020395278930664,
+ "learning_rate": 3.1034012442496253e-06,
+ "loss": 0.0235,
+ "step": 27524
+ },
+ {
+ "epoch": 75.0,
+ "grad_norm": 1.4378288984298706,
+ "learning_rate": 3.1027622272189572e-06,
+ "loss": 0.0636,
+ "step": 27525
+ },
+ {
+ "epoch": 75.00272479564033,
+ "grad_norm": 2.0751328468322754,
+ "learning_rate": 3.102123263903758e-06,
+ "loss": 0.0283,
+ "step": 27526
+ },
+ {
+ "epoch": 75.00544959128065,
+ "grad_norm": 0.9939257502555847,
+ "learning_rate": 3.1014843543090002e-06,
+ "loss": 0.0098,
+ "step": 27527
+ },
+ {
+ "epoch": 75.00817438692098,
+ "grad_norm": 1.890942096710205,
+ "learning_rate": 3.1008454984396606e-06,
+ "loss": 0.0106,
+ "step": 27528
+ },
+ {
+ "epoch": 75.0108991825613,
+ "grad_norm": 1.1635282039642334,
+ "learning_rate": 3.1002066963007116e-06,
+ "loss": 0.0158,
+ "step": 27529
+ },
+ {
+ "epoch": 75.01362397820164,
+ "grad_norm": 1.1752938032150269,
+ "learning_rate": 3.099567947897133e-06,
+ "loss": 0.0177,
+ "step": 27530
+ },
+ {
+ "epoch": 75.01634877384195,
+ "grad_norm": 1.206947684288025,
+ "learning_rate": 3.0989292532338978e-06,
+ "loss": 0.0106,
+ "step": 27531
+ },
+ {
+ "epoch": 75.01907356948229,
+ "grad_norm": 1.4310277700424194,
+ "learning_rate": 3.098290612315976e-06,
+ "loss": 0.013,
+ "step": 27532
+ },
+ {
+ "epoch": 75.02179836512262,
+ "grad_norm": 1.5363856554031372,
+ "learning_rate": 3.0976520251483476e-06,
+ "loss": 0.0327,
+ "step": 27533
+ },
+ {
+ "epoch": 75.02452316076294,
+ "grad_norm": 1.1797982454299927,
+ "learning_rate": 3.097013491735984e-06,
+ "loss": 0.0149,
+ "step": 27534
+ },
+ {
+ "epoch": 75.02724795640327,
+ "grad_norm": 1.177031397819519,
+ "learning_rate": 3.0963750120838564e-06,
+ "loss": 0.0304,
+ "step": 27535
+ },
+ {
+ "epoch": 75.02997275204359,
+ "grad_norm": 1.5873252153396606,
+ "learning_rate": 3.0957365861969345e-06,
+ "loss": 0.0892,
+ "step": 27536
+ },
+ {
+ "epoch": 75.03269754768392,
+ "grad_norm": 1.2105565071105957,
+ "learning_rate": 3.095098214080198e-06,
+ "loss": 0.0136,
+ "step": 27537
+ },
+ {
+ "epoch": 75.03542234332426,
+ "grad_norm": 5.084540843963623,
+ "learning_rate": 3.0944598957386107e-06,
+ "loss": 0.0659,
+ "step": 27538
+ },
+ {
+ "epoch": 75.03814713896458,
+ "grad_norm": 1.0625745058059692,
+ "learning_rate": 3.0938216311771507e-06,
+ "loss": 0.0076,
+ "step": 27539
+ },
+ {
+ "epoch": 75.04087193460491,
+ "grad_norm": 7.943860054016113,
+ "learning_rate": 3.0931834204007862e-06,
+ "loss": 0.0118,
+ "step": 27540
+ },
+ {
+ "epoch": 75.04359673024523,
+ "grad_norm": 1.1555078029632568,
+ "learning_rate": 3.0925452634144825e-06,
+ "loss": 0.0221,
+ "step": 27541
+ },
+ {
+ "epoch": 75.04632152588556,
+ "grad_norm": 1.1678205728530884,
+ "learning_rate": 3.0919071602232175e-06,
+ "loss": 0.0059,
+ "step": 27542
+ },
+ {
+ "epoch": 75.04904632152588,
+ "grad_norm": 1.54570472240448,
+ "learning_rate": 3.0912691108319582e-06,
+ "loss": 0.0144,
+ "step": 27543
+ },
+ {
+ "epoch": 75.05177111716621,
+ "grad_norm": 1.6142523288726807,
+ "learning_rate": 3.0906311152456713e-06,
+ "loss": 0.0296,
+ "step": 27544
+ },
+ {
+ "epoch": 75.05449591280654,
+ "grad_norm": 3.930079698562622,
+ "learning_rate": 3.0899931734693246e-06,
+ "loss": 0.082,
+ "step": 27545
+ },
+ {
+ "epoch": 75.05722070844686,
+ "grad_norm": 1.5910385847091675,
+ "learning_rate": 3.089355285507891e-06,
+ "loss": 0.0208,
+ "step": 27546
+ },
+ {
+ "epoch": 75.0599455040872,
+ "grad_norm": 2.3911609649658203,
+ "learning_rate": 3.0887174513663376e-06,
+ "loss": 0.176,
+ "step": 27547
+ },
+ {
+ "epoch": 75.06267029972751,
+ "grad_norm": 0.8715260028839111,
+ "learning_rate": 3.0880796710496287e-06,
+ "loss": 0.0079,
+ "step": 27548
+ },
+ {
+ "epoch": 75.06539509536785,
+ "grad_norm": 1.5324138402938843,
+ "learning_rate": 3.08744194456273e-06,
+ "loss": 0.2101,
+ "step": 27549
+ },
+ {
+ "epoch": 75.06811989100818,
+ "grad_norm": 0.7639043927192688,
+ "learning_rate": 3.0868042719106152e-06,
+ "loss": 0.0073,
+ "step": 27550
+ },
+ {
+ "epoch": 75.0708446866485,
+ "grad_norm": 1.352965235710144,
+ "learning_rate": 3.086166653098246e-06,
+ "loss": 0.0324,
+ "step": 27551
+ },
+ {
+ "epoch": 75.07356948228883,
+ "grad_norm": 1.5362211465835571,
+ "learning_rate": 3.0855290881305845e-06,
+ "loss": 0.0854,
+ "step": 27552
+ },
+ {
+ "epoch": 75.07629427792915,
+ "grad_norm": 1.2697525024414062,
+ "learning_rate": 3.084891577012603e-06,
+ "loss": 0.0927,
+ "step": 27553
+ },
+ {
+ "epoch": 75.07901907356948,
+ "grad_norm": 1.5335156917572021,
+ "learning_rate": 3.0842541197492637e-06,
+ "loss": 0.061,
+ "step": 27554
+ },
+ {
+ "epoch": 75.0817438692098,
+ "grad_norm": 1.539293646812439,
+ "learning_rate": 3.08361671634553e-06,
+ "loss": 0.0189,
+ "step": 27555
+ },
+ {
+ "epoch": 75.08446866485014,
+ "grad_norm": 1.7086803913116455,
+ "learning_rate": 3.0829793668063646e-06,
+ "loss": 0.0295,
+ "step": 27556
+ },
+ {
+ "epoch": 75.08719346049047,
+ "grad_norm": 1.7906955480575562,
+ "learning_rate": 3.0823420711367322e-06,
+ "loss": 0.0227,
+ "step": 27557
+ },
+ {
+ "epoch": 75.08991825613079,
+ "grad_norm": 1.8543671369552612,
+ "learning_rate": 3.0817048293416017e-06,
+ "loss": 0.1127,
+ "step": 27558
+ },
+ {
+ "epoch": 75.09264305177112,
+ "grad_norm": 1.1484310626983643,
+ "learning_rate": 3.08106764142593e-06,
+ "loss": 0.0132,
+ "step": 27559
+ },
+ {
+ "epoch": 75.09536784741144,
+ "grad_norm": 1.3721463680267334,
+ "learning_rate": 3.0804305073946818e-06,
+ "loss": 0.0663,
+ "step": 27560
+ },
+ {
+ "epoch": 75.09809264305177,
+ "grad_norm": 2.2059903144836426,
+ "learning_rate": 3.079793427252814e-06,
+ "loss": 0.1231,
+ "step": 27561
+ },
+ {
+ "epoch": 75.1008174386921,
+ "grad_norm": 0.8661786913871765,
+ "learning_rate": 3.0791564010052967e-06,
+ "loss": 0.0197,
+ "step": 27562
+ },
+ {
+ "epoch": 75.10354223433242,
+ "grad_norm": 1.6376103162765503,
+ "learning_rate": 3.0785194286570854e-06,
+ "loss": 0.0309,
+ "step": 27563
+ },
+ {
+ "epoch": 75.10626702997276,
+ "grad_norm": 1.1779892444610596,
+ "learning_rate": 3.077882510213143e-06,
+ "loss": 0.0131,
+ "step": 27564
+ },
+ {
+ "epoch": 75.10899182561307,
+ "grad_norm": 1.3371515274047852,
+ "learning_rate": 3.077245645678425e-06,
+ "loss": 0.0295,
+ "step": 27565
+ },
+ {
+ "epoch": 75.11171662125341,
+ "grad_norm": 3.286377191543579,
+ "learning_rate": 3.0766088350578984e-06,
+ "loss": 0.137,
+ "step": 27566
+ },
+ {
+ "epoch": 75.11444141689373,
+ "grad_norm": 1.3241621255874634,
+ "learning_rate": 3.0759720783565196e-06,
+ "loss": 0.0648,
+ "step": 27567
+ },
+ {
+ "epoch": 75.11716621253406,
+ "grad_norm": 1.1998711824417114,
+ "learning_rate": 3.075335375579247e-06,
+ "loss": 0.0157,
+ "step": 27568
+ },
+ {
+ "epoch": 75.11989100817439,
+ "grad_norm": 1.561583161354065,
+ "learning_rate": 3.074698726731036e-06,
+ "loss": 0.1579,
+ "step": 27569
+ },
+ {
+ "epoch": 75.12261580381471,
+ "grad_norm": 1.3251947164535522,
+ "learning_rate": 3.074062131816853e-06,
+ "loss": 0.0189,
+ "step": 27570
+ },
+ {
+ "epoch": 75.12534059945504,
+ "grad_norm": 1.6682759523391724,
+ "learning_rate": 3.07342559084165e-06,
+ "loss": 0.0734,
+ "step": 27571
+ },
+ {
+ "epoch": 75.12806539509536,
+ "grad_norm": 1.3432854413986206,
+ "learning_rate": 3.0727891038103865e-06,
+ "loss": 0.0172,
+ "step": 27572
+ },
+ {
+ "epoch": 75.1307901907357,
+ "grad_norm": 1.827602744102478,
+ "learning_rate": 3.0721526707280113e-06,
+ "loss": 0.0278,
+ "step": 27573
+ },
+ {
+ "epoch": 75.13351498637603,
+ "grad_norm": 0.6298525333404541,
+ "learning_rate": 3.0715162915994944e-06,
+ "loss": 0.0061,
+ "step": 27574
+ },
+ {
+ "epoch": 75.13623978201635,
+ "grad_norm": 1.6733486652374268,
+ "learning_rate": 3.0708799664297863e-06,
+ "loss": 0.0579,
+ "step": 27575
+ },
+ {
+ "epoch": 75.13896457765668,
+ "grad_norm": 1.5303202867507935,
+ "learning_rate": 3.0702436952238412e-06,
+ "loss": 0.015,
+ "step": 27576
+ },
+ {
+ "epoch": 75.141689373297,
+ "grad_norm": 1.3141244649887085,
+ "learning_rate": 3.0696074779866127e-06,
+ "loss": 0.0451,
+ "step": 27577
+ },
+ {
+ "epoch": 75.14441416893733,
+ "grad_norm": 3.938678741455078,
+ "learning_rate": 3.0689713147230617e-06,
+ "loss": 0.0169,
+ "step": 27578
+ },
+ {
+ "epoch": 75.14713896457765,
+ "grad_norm": 1.1459320783615112,
+ "learning_rate": 3.068335205438139e-06,
+ "loss": 0.0078,
+ "step": 27579
+ },
+ {
+ "epoch": 75.14986376021798,
+ "grad_norm": 0.8923695087432861,
+ "learning_rate": 3.067699150136797e-06,
+ "loss": 0.0071,
+ "step": 27580
+ },
+ {
+ "epoch": 75.15258855585832,
+ "grad_norm": 1.423708438873291,
+ "learning_rate": 3.067063148823989e-06,
+ "loss": 0.0294,
+ "step": 27581
+ },
+ {
+ "epoch": 75.15531335149863,
+ "grad_norm": 2.079232931137085,
+ "learning_rate": 3.0664272015046735e-06,
+ "loss": 0.0341,
+ "step": 27582
+ },
+ {
+ "epoch": 75.15803814713897,
+ "grad_norm": 1.6459378004074097,
+ "learning_rate": 3.0657913081837987e-06,
+ "loss": 0.0226,
+ "step": 27583
+ },
+ {
+ "epoch": 75.16076294277929,
+ "grad_norm": 1.2412631511688232,
+ "learning_rate": 3.0651554688663188e-06,
+ "loss": 0.0131,
+ "step": 27584
+ },
+ {
+ "epoch": 75.16348773841962,
+ "grad_norm": 1.368913173675537,
+ "learning_rate": 3.064519683557182e-06,
+ "loss": 0.0161,
+ "step": 27585
+ },
+ {
+ "epoch": 75.16621253405995,
+ "grad_norm": 1.7198784351348877,
+ "learning_rate": 3.063883952261345e-06,
+ "loss": 0.0831,
+ "step": 27586
+ },
+ {
+ "epoch": 75.16893732970027,
+ "grad_norm": 1.9078258275985718,
+ "learning_rate": 3.0632482749837578e-06,
+ "loss": 0.054,
+ "step": 27587
+ },
+ {
+ "epoch": 75.1716621253406,
+ "grad_norm": 1.3036761283874512,
+ "learning_rate": 3.0626126517293686e-06,
+ "loss": 0.0758,
+ "step": 27588
+ },
+ {
+ "epoch": 75.17438692098092,
+ "grad_norm": 0.634172260761261,
+ "learning_rate": 3.061977082503126e-06,
+ "loss": 0.0066,
+ "step": 27589
+ },
+ {
+ "epoch": 75.17711171662125,
+ "grad_norm": 1.0060681104660034,
+ "learning_rate": 3.0613415673099822e-06,
+ "loss": 0.0085,
+ "step": 27590
+ },
+ {
+ "epoch": 75.17983651226157,
+ "grad_norm": 8.527623176574707,
+ "learning_rate": 3.0607061061548902e-06,
+ "loss": 0.0571,
+ "step": 27591
+ },
+ {
+ "epoch": 75.1825613079019,
+ "grad_norm": 1.8377901315689087,
+ "learning_rate": 3.0600706990427966e-06,
+ "loss": 0.0198,
+ "step": 27592
+ },
+ {
+ "epoch": 75.18528610354224,
+ "grad_norm": 1.6648131608963013,
+ "learning_rate": 3.0594353459786484e-06,
+ "loss": 0.0927,
+ "step": 27593
+ },
+ {
+ "epoch": 75.18801089918256,
+ "grad_norm": 1.333404779434204,
+ "learning_rate": 3.058800046967392e-06,
+ "loss": 0.0173,
+ "step": 27594
+ },
+ {
+ "epoch": 75.19073569482289,
+ "grad_norm": 1.2382725477218628,
+ "learning_rate": 3.058164802013979e-06,
+ "loss": 0.0396,
+ "step": 27595
+ },
+ {
+ "epoch": 75.19346049046321,
+ "grad_norm": 1.5945600271224976,
+ "learning_rate": 3.0575296111233567e-06,
+ "loss": 0.0621,
+ "step": 27596
+ },
+ {
+ "epoch": 75.19618528610354,
+ "grad_norm": 1.8017343282699585,
+ "learning_rate": 3.056894474300466e-06,
+ "loss": 0.0264,
+ "step": 27597
+ },
+ {
+ "epoch": 75.19891008174388,
+ "grad_norm": 1.6345990896224976,
+ "learning_rate": 3.0562593915502614e-06,
+ "loss": 0.0303,
+ "step": 27598
+ },
+ {
+ "epoch": 75.2016348773842,
+ "grad_norm": 2.9357407093048096,
+ "learning_rate": 3.055624362877685e-06,
+ "loss": 0.0179,
+ "step": 27599
+ },
+ {
+ "epoch": 75.20435967302453,
+ "grad_norm": 0.8606889843940735,
+ "learning_rate": 3.0549893882876825e-06,
+ "loss": 0.008,
+ "step": 27600
+ },
+ {
+ "epoch": 75.20708446866485,
+ "grad_norm": 0.7665416598320007,
+ "learning_rate": 3.0543544677851953e-06,
+ "loss": 0.0069,
+ "step": 27601
+ },
+ {
+ "epoch": 75.20980926430518,
+ "grad_norm": 1.1909406185150146,
+ "learning_rate": 3.0537196013751767e-06,
+ "loss": 0.0462,
+ "step": 27602
+ },
+ {
+ "epoch": 75.2125340599455,
+ "grad_norm": 1.091231346130371,
+ "learning_rate": 3.0530847890625646e-06,
+ "loss": 0.0093,
+ "step": 27603
+ },
+ {
+ "epoch": 75.21525885558583,
+ "grad_norm": 1.3041582107543945,
+ "learning_rate": 3.0524500308523064e-06,
+ "loss": 0.0142,
+ "step": 27604
+ },
+ {
+ "epoch": 75.21798365122616,
+ "grad_norm": 10.89159870147705,
+ "learning_rate": 3.0518153267493388e-06,
+ "loss": 0.106,
+ "step": 27605
+ },
+ {
+ "epoch": 75.22070844686648,
+ "grad_norm": 0.7839284539222717,
+ "learning_rate": 3.051180676758614e-06,
+ "loss": 0.0131,
+ "step": 27606
+ },
+ {
+ "epoch": 75.22343324250681,
+ "grad_norm": 1.094604730606079,
+ "learning_rate": 3.0505460808850674e-06,
+ "loss": 0.0137,
+ "step": 27607
+ },
+ {
+ "epoch": 75.22615803814713,
+ "grad_norm": 0.8729565143585205,
+ "learning_rate": 3.0499115391336464e-06,
+ "loss": 0.0097,
+ "step": 27608
+ },
+ {
+ "epoch": 75.22888283378747,
+ "grad_norm": 1.3784348964691162,
+ "learning_rate": 3.0492770515092917e-06,
+ "loss": 0.0249,
+ "step": 27609
+ },
+ {
+ "epoch": 75.2316076294278,
+ "grad_norm": 0.8839079141616821,
+ "learning_rate": 3.0486426180169404e-06,
+ "loss": 0.009,
+ "step": 27610
+ },
+ {
+ "epoch": 75.23433242506812,
+ "grad_norm": 1.132037878036499,
+ "learning_rate": 3.0480082386615394e-06,
+ "loss": 0.0079,
+ "step": 27611
+ },
+ {
+ "epoch": 75.23705722070845,
+ "grad_norm": 7.305002689361572,
+ "learning_rate": 3.047373913448026e-06,
+ "loss": 0.0638,
+ "step": 27612
+ },
+ {
+ "epoch": 75.23978201634877,
+ "grad_norm": 4.531543731689453,
+ "learning_rate": 3.0467396423813423e-06,
+ "loss": 0.0157,
+ "step": 27613
+ },
+ {
+ "epoch": 75.2425068119891,
+ "grad_norm": 1.6003336906433105,
+ "learning_rate": 3.0461054254664225e-06,
+ "loss": 0.0211,
+ "step": 27614
+ },
+ {
+ "epoch": 75.24523160762942,
+ "grad_norm": 1.6623220443725586,
+ "learning_rate": 3.0454712627082128e-06,
+ "loss": 0.0306,
+ "step": 27615
+ },
+ {
+ "epoch": 75.24795640326975,
+ "grad_norm": 1.1363540887832642,
+ "learning_rate": 3.044837154111648e-06,
+ "loss": 0.0162,
+ "step": 27616
+ },
+ {
+ "epoch": 75.25068119891009,
+ "grad_norm": 1.3334792852401733,
+ "learning_rate": 3.044203099681666e-06,
+ "loss": 0.0656,
+ "step": 27617
+ },
+ {
+ "epoch": 75.2534059945504,
+ "grad_norm": 1.1810948848724365,
+ "learning_rate": 3.043569099423209e-06,
+ "loss": 0.0232,
+ "step": 27618
+ },
+ {
+ "epoch": 75.25613079019074,
+ "grad_norm": 1.8607136011123657,
+ "learning_rate": 3.042935153341212e-06,
+ "loss": 0.0367,
+ "step": 27619
+ },
+ {
+ "epoch": 75.25885558583106,
+ "grad_norm": 1.6991349458694458,
+ "learning_rate": 3.042301261440612e-06,
+ "loss": 0.0523,
+ "step": 27620
+ },
+ {
+ "epoch": 75.26158038147139,
+ "grad_norm": 1.3721791505813599,
+ "learning_rate": 3.041667423726343e-06,
+ "loss": 0.0116,
+ "step": 27621
+ },
+ {
+ "epoch": 75.26430517711172,
+ "grad_norm": 0.8887773752212524,
+ "learning_rate": 3.041033640203348e-06,
+ "loss": 0.0094,
+ "step": 27622
+ },
+ {
+ "epoch": 75.26702997275204,
+ "grad_norm": 1.1052641868591309,
+ "learning_rate": 3.040399910876558e-06,
+ "loss": 0.0141,
+ "step": 27623
+ },
+ {
+ "epoch": 75.26975476839237,
+ "grad_norm": 1.8345931768417358,
+ "learning_rate": 3.039766235750907e-06,
+ "loss": 0.0338,
+ "step": 27624
+ },
+ {
+ "epoch": 75.2724795640327,
+ "grad_norm": 1.634987235069275,
+ "learning_rate": 3.039132614831335e-06,
+ "loss": 0.0148,
+ "step": 27625
+ },
+ {
+ "epoch": 75.27520435967303,
+ "grad_norm": 1.0546456575393677,
+ "learning_rate": 3.0384990481227716e-06,
+ "loss": 0.0059,
+ "step": 27626
+ },
+ {
+ "epoch": 75.27792915531334,
+ "grad_norm": 1.4037747383117676,
+ "learning_rate": 3.0378655356301554e-06,
+ "loss": 0.0774,
+ "step": 27627
+ },
+ {
+ "epoch": 75.28065395095368,
+ "grad_norm": 2.2012064456939697,
+ "learning_rate": 3.0372320773584196e-06,
+ "loss": 0.0818,
+ "step": 27628
+ },
+ {
+ "epoch": 75.28337874659401,
+ "grad_norm": 1.5193716287612915,
+ "learning_rate": 3.0365986733124963e-06,
+ "loss": 0.0356,
+ "step": 27629
+ },
+ {
+ "epoch": 75.28610354223433,
+ "grad_norm": 1.6531883478164673,
+ "learning_rate": 3.0359653234973154e-06,
+ "loss": 0.0566,
+ "step": 27630
+ },
+ {
+ "epoch": 75.28882833787466,
+ "grad_norm": 0.8980324268341064,
+ "learning_rate": 3.0353320279178146e-06,
+ "loss": 0.0081,
+ "step": 27631
+ },
+ {
+ "epoch": 75.29155313351498,
+ "grad_norm": 1.0823092460632324,
+ "learning_rate": 3.0346987865789245e-06,
+ "loss": 0.0114,
+ "step": 27632
+ },
+ {
+ "epoch": 75.29427792915531,
+ "grad_norm": 0.4555099904537201,
+ "learning_rate": 3.034065599485576e-06,
+ "loss": 0.0039,
+ "step": 27633
+ },
+ {
+ "epoch": 75.29700272479565,
+ "grad_norm": 1.13321053981781,
+ "learning_rate": 3.033432466642697e-06,
+ "loss": 0.0095,
+ "step": 27634
+ },
+ {
+ "epoch": 75.29972752043597,
+ "grad_norm": 1.4640742540359497,
+ "learning_rate": 3.032799388055225e-06,
+ "loss": 0.0198,
+ "step": 27635
+ },
+ {
+ "epoch": 75.3024523160763,
+ "grad_norm": 1.1089651584625244,
+ "learning_rate": 3.0321663637280875e-06,
+ "loss": 0.0918,
+ "step": 27636
+ },
+ {
+ "epoch": 75.30517711171662,
+ "grad_norm": 1.4315311908721924,
+ "learning_rate": 3.031533393666214e-06,
+ "loss": 0.006,
+ "step": 27637
+ },
+ {
+ "epoch": 75.30790190735695,
+ "grad_norm": 1.469812273979187,
+ "learning_rate": 3.030900477874531e-06,
+ "loss": 0.0315,
+ "step": 27638
+ },
+ {
+ "epoch": 75.31062670299727,
+ "grad_norm": 1.6635708808898926,
+ "learning_rate": 3.0302676163579738e-06,
+ "loss": 0.021,
+ "step": 27639
+ },
+ {
+ "epoch": 75.3133514986376,
+ "grad_norm": 1.2103321552276611,
+ "learning_rate": 3.0296348091214676e-06,
+ "loss": 0.0218,
+ "step": 27640
+ },
+ {
+ "epoch": 75.31607629427793,
+ "grad_norm": 1.080502986907959,
+ "learning_rate": 3.0290020561699385e-06,
+ "loss": 0.0218,
+ "step": 27641
+ },
+ {
+ "epoch": 75.31880108991825,
+ "grad_norm": 1.4318712949752808,
+ "learning_rate": 3.0283693575083174e-06,
+ "loss": 0.0195,
+ "step": 27642
+ },
+ {
+ "epoch": 75.32152588555859,
+ "grad_norm": 0.9373258352279663,
+ "learning_rate": 3.0277367131415336e-06,
+ "loss": 0.0135,
+ "step": 27643
+ },
+ {
+ "epoch": 75.3242506811989,
+ "grad_norm": 1.5181000232696533,
+ "learning_rate": 3.0271041230745122e-06,
+ "loss": 0.0185,
+ "step": 27644
+ },
+ {
+ "epoch": 75.32697547683924,
+ "grad_norm": 1.3251280784606934,
+ "learning_rate": 3.0264715873121796e-06,
+ "loss": 0.0926,
+ "step": 27645
+ },
+ {
+ "epoch": 75.32970027247957,
+ "grad_norm": 0.9150679707527161,
+ "learning_rate": 3.0258391058594595e-06,
+ "loss": 0.0097,
+ "step": 27646
+ },
+ {
+ "epoch": 75.33242506811989,
+ "grad_norm": 1.0589977502822876,
+ "learning_rate": 3.0252066787212817e-06,
+ "loss": 0.0089,
+ "step": 27647
+ },
+ {
+ "epoch": 75.33514986376022,
+ "grad_norm": 0.8742445111274719,
+ "learning_rate": 3.0245743059025713e-06,
+ "loss": 0.0084,
+ "step": 27648
+ },
+ {
+ "epoch": 75.33787465940054,
+ "grad_norm": 1.318638563156128,
+ "learning_rate": 3.023941987408251e-06,
+ "loss": 0.0292,
+ "step": 27649
+ },
+ {
+ "epoch": 75.34059945504087,
+ "grad_norm": 1.464058518409729,
+ "learning_rate": 3.0233097232432417e-06,
+ "loss": 0.0149,
+ "step": 27650
+ },
+ {
+ "epoch": 75.34332425068119,
+ "grad_norm": 1.44194757938385,
+ "learning_rate": 3.022677513412476e-06,
+ "loss": 0.0132,
+ "step": 27651
+ },
+ {
+ "epoch": 75.34604904632153,
+ "grad_norm": 1.4129916429519653,
+ "learning_rate": 3.022045357920873e-06,
+ "loss": 0.0122,
+ "step": 27652
+ },
+ {
+ "epoch": 75.34877384196186,
+ "grad_norm": 0.8213490843772888,
+ "learning_rate": 3.0214132567733565e-06,
+ "loss": 0.0125,
+ "step": 27653
+ },
+ {
+ "epoch": 75.35149863760218,
+ "grad_norm": 1.7387304306030273,
+ "learning_rate": 3.020781209974846e-06,
+ "loss": 0.0376,
+ "step": 27654
+ },
+ {
+ "epoch": 75.35422343324251,
+ "grad_norm": 0.8845280408859253,
+ "learning_rate": 3.0201492175302695e-06,
+ "loss": 0.0119,
+ "step": 27655
+ },
+ {
+ "epoch": 75.35694822888283,
+ "grad_norm": 1.0125471353530884,
+ "learning_rate": 3.019517279444546e-06,
+ "loss": 0.0121,
+ "step": 27656
+ },
+ {
+ "epoch": 75.35967302452316,
+ "grad_norm": 1.2774945497512817,
+ "learning_rate": 3.0188853957225984e-06,
+ "loss": 0.1133,
+ "step": 27657
+ },
+ {
+ "epoch": 75.3623978201635,
+ "grad_norm": 1.5990276336669922,
+ "learning_rate": 3.0182535663693422e-06,
+ "loss": 0.0259,
+ "step": 27658
+ },
+ {
+ "epoch": 75.36512261580381,
+ "grad_norm": 1.2625235319137573,
+ "learning_rate": 3.0176217913897034e-06,
+ "loss": 0.0107,
+ "step": 27659
+ },
+ {
+ "epoch": 75.36784741144415,
+ "grad_norm": 0.8324484825134277,
+ "learning_rate": 3.0169900707886036e-06,
+ "loss": 0.0055,
+ "step": 27660
+ },
+ {
+ "epoch": 75.37057220708446,
+ "grad_norm": 1.1653814315795898,
+ "learning_rate": 3.0163584045709615e-06,
+ "loss": 0.0114,
+ "step": 27661
+ },
+ {
+ "epoch": 75.3732970027248,
+ "grad_norm": 2.6133341789245605,
+ "learning_rate": 3.0157267927416912e-06,
+ "loss": 0.0382,
+ "step": 27662
+ },
+ {
+ "epoch": 75.37602179836512,
+ "grad_norm": 0.895386815071106,
+ "learning_rate": 3.0150952353057195e-06,
+ "loss": 0.01,
+ "step": 27663
+ },
+ {
+ "epoch": 75.37874659400545,
+ "grad_norm": 1.6218366622924805,
+ "learning_rate": 3.0144637322679617e-06,
+ "loss": 0.0679,
+ "step": 27664
+ },
+ {
+ "epoch": 75.38147138964578,
+ "grad_norm": 1.3215076923370361,
+ "learning_rate": 3.0138322836333343e-06,
+ "loss": 0.0257,
+ "step": 27665
+ },
+ {
+ "epoch": 75.3841961852861,
+ "grad_norm": 1.120356559753418,
+ "learning_rate": 3.0132008894067545e-06,
+ "loss": 0.0238,
+ "step": 27666
+ },
+ {
+ "epoch": 75.38692098092643,
+ "grad_norm": 1.2436186075210571,
+ "learning_rate": 3.012569549593143e-06,
+ "loss": 0.017,
+ "step": 27667
+ },
+ {
+ "epoch": 75.38964577656675,
+ "grad_norm": 1.2562730312347412,
+ "learning_rate": 3.011938264197416e-06,
+ "loss": 0.0175,
+ "step": 27668
+ },
+ {
+ "epoch": 75.39237057220708,
+ "grad_norm": 1.265891432762146,
+ "learning_rate": 3.0113070332244886e-06,
+ "loss": 0.0087,
+ "step": 27669
+ },
+ {
+ "epoch": 75.39509536784742,
+ "grad_norm": 1.2205818891525269,
+ "learning_rate": 3.0106758566792737e-06,
+ "loss": 0.1146,
+ "step": 27670
+ },
+ {
+ "epoch": 75.39782016348774,
+ "grad_norm": 1.7082029581069946,
+ "learning_rate": 3.0100447345666937e-06,
+ "loss": 0.0332,
+ "step": 27671
+ },
+ {
+ "epoch": 75.40054495912807,
+ "grad_norm": 1.1759909391403198,
+ "learning_rate": 3.0094136668916607e-06,
+ "loss": 0.0181,
+ "step": 27672
+ },
+ {
+ "epoch": 75.40326975476839,
+ "grad_norm": 1.0334360599517822,
+ "learning_rate": 3.008782653659089e-06,
+ "loss": 0.0148,
+ "step": 27673
+ },
+ {
+ "epoch": 75.40599455040872,
+ "grad_norm": 1.4435553550720215,
+ "learning_rate": 3.0081516948738887e-06,
+ "loss": 0.0284,
+ "step": 27674
+ },
+ {
+ "epoch": 75.40871934604904,
+ "grad_norm": 9.671640396118164,
+ "learning_rate": 3.007520790540982e-06,
+ "loss": 0.0237,
+ "step": 27675
+ },
+ {
+ "epoch": 75.41144414168937,
+ "grad_norm": 1.154584288597107,
+ "learning_rate": 3.0068899406652752e-06,
+ "loss": 0.0135,
+ "step": 27676
+ },
+ {
+ "epoch": 75.4141689373297,
+ "grad_norm": 1.3449468612670898,
+ "learning_rate": 3.0062591452516876e-06,
+ "loss": 0.0218,
+ "step": 27677
+ },
+ {
+ "epoch": 75.41689373297002,
+ "grad_norm": 2.2479984760284424,
+ "learning_rate": 3.0056284043051286e-06,
+ "loss": 0.0448,
+ "step": 27678
+ },
+ {
+ "epoch": 75.41961852861036,
+ "grad_norm": 1.116487979888916,
+ "learning_rate": 3.004997717830508e-06,
+ "loss": 0.016,
+ "step": 27679
+ },
+ {
+ "epoch": 75.42234332425068,
+ "grad_norm": 1.7257364988327026,
+ "learning_rate": 3.004367085832742e-06,
+ "loss": 0.0229,
+ "step": 27680
+ },
+ {
+ "epoch": 75.42506811989101,
+ "grad_norm": 1.0740619897842407,
+ "learning_rate": 3.0037365083167404e-06,
+ "loss": 0.0151,
+ "step": 27681
+ },
+ {
+ "epoch": 75.42779291553134,
+ "grad_norm": 1.6861770153045654,
+ "learning_rate": 3.003105985287411e-06,
+ "loss": 0.0431,
+ "step": 27682
+ },
+ {
+ "epoch": 75.43051771117166,
+ "grad_norm": 1.040895700454712,
+ "learning_rate": 3.00247551674967e-06,
+ "loss": 0.0423,
+ "step": 27683
+ },
+ {
+ "epoch": 75.433242506812,
+ "grad_norm": 1.5244276523590088,
+ "learning_rate": 3.0018451027084238e-06,
+ "loss": 0.0543,
+ "step": 27684
+ },
+ {
+ "epoch": 75.43596730245231,
+ "grad_norm": 1.30898118019104,
+ "learning_rate": 3.0012147431685832e-06,
+ "loss": 0.0115,
+ "step": 27685
+ },
+ {
+ "epoch": 75.43869209809264,
+ "grad_norm": 1.2146837711334229,
+ "learning_rate": 3.0005844381350545e-06,
+ "loss": 0.0476,
+ "step": 27686
+ },
+ {
+ "epoch": 75.44141689373296,
+ "grad_norm": 1.5903191566467285,
+ "learning_rate": 2.9999541876127513e-06,
+ "loss": 0.0192,
+ "step": 27687
+ },
+ {
+ "epoch": 75.4441416893733,
+ "grad_norm": 2.442235231399536,
+ "learning_rate": 2.99932399160658e-06,
+ "loss": 0.1242,
+ "step": 27688
+ },
+ {
+ "epoch": 75.44686648501363,
+ "grad_norm": 1.643560528755188,
+ "learning_rate": 2.998693850121447e-06,
+ "loss": 0.0611,
+ "step": 27689
+ },
+ {
+ "epoch": 75.44959128065395,
+ "grad_norm": 0.8367815017700195,
+ "learning_rate": 2.998063763162259e-06,
+ "loss": 0.0087,
+ "step": 27690
+ },
+ {
+ "epoch": 75.45231607629428,
+ "grad_norm": 0.8780845403671265,
+ "learning_rate": 2.9974337307339274e-06,
+ "loss": 0.0107,
+ "step": 27691
+ },
+ {
+ "epoch": 75.4550408719346,
+ "grad_norm": 1.4149683713912964,
+ "learning_rate": 2.9968037528413573e-06,
+ "loss": 0.0163,
+ "step": 27692
+ },
+ {
+ "epoch": 75.45776566757493,
+ "grad_norm": 1.172974705696106,
+ "learning_rate": 2.9961738294894504e-06,
+ "loss": 0.0136,
+ "step": 27693
+ },
+ {
+ "epoch": 75.46049046321527,
+ "grad_norm": 1.3822104930877686,
+ "learning_rate": 2.9955439606831195e-06,
+ "loss": 0.1016,
+ "step": 27694
+ },
+ {
+ "epoch": 75.46321525885558,
+ "grad_norm": 1.5110291242599487,
+ "learning_rate": 2.9949141464272635e-06,
+ "loss": 0.0122,
+ "step": 27695
+ },
+ {
+ "epoch": 75.46594005449592,
+ "grad_norm": 1.1650700569152832,
+ "learning_rate": 2.9942843867267947e-06,
+ "loss": 0.0143,
+ "step": 27696
+ },
+ {
+ "epoch": 75.46866485013624,
+ "grad_norm": 2.4542038440704346,
+ "learning_rate": 2.993654681586613e-06,
+ "loss": 0.0269,
+ "step": 27697
+ },
+ {
+ "epoch": 75.47138964577657,
+ "grad_norm": 0.9141539335250854,
+ "learning_rate": 2.993025031011623e-06,
+ "loss": 0.0096,
+ "step": 27698
+ },
+ {
+ "epoch": 75.47411444141689,
+ "grad_norm": 1.3401267528533936,
+ "learning_rate": 2.9923954350067263e-06,
+ "loss": 0.1349,
+ "step": 27699
+ },
+ {
+ "epoch": 75.47683923705722,
+ "grad_norm": 2.6330459117889404,
+ "learning_rate": 2.99176589357683e-06,
+ "loss": 0.0389,
+ "step": 27700
+ },
+ {
+ "epoch": 75.47956403269755,
+ "grad_norm": 1.2071197032928467,
+ "learning_rate": 2.9911364067268366e-06,
+ "loss": 0.0092,
+ "step": 27701
+ },
+ {
+ "epoch": 75.48228882833787,
+ "grad_norm": 1.6567225456237793,
+ "learning_rate": 2.9905069744616466e-06,
+ "loss": 0.0134,
+ "step": 27702
+ },
+ {
+ "epoch": 75.4850136239782,
+ "grad_norm": 1.0608042478561401,
+ "learning_rate": 2.9898775967861605e-06,
+ "loss": 0.0117,
+ "step": 27703
+ },
+ {
+ "epoch": 75.48773841961852,
+ "grad_norm": 1.0679219961166382,
+ "learning_rate": 2.9892482737052843e-06,
+ "loss": 0.0092,
+ "step": 27704
+ },
+ {
+ "epoch": 75.49046321525886,
+ "grad_norm": 1.0825906991958618,
+ "learning_rate": 2.9886190052239173e-06,
+ "loss": 0.0102,
+ "step": 27705
+ },
+ {
+ "epoch": 75.49318801089919,
+ "grad_norm": 1.5488547086715698,
+ "learning_rate": 2.987989791346957e-06,
+ "loss": 0.0593,
+ "step": 27706
+ },
+ {
+ "epoch": 75.49591280653951,
+ "grad_norm": 1.0094207525253296,
+ "learning_rate": 2.987360632079308e-06,
+ "loss": 0.006,
+ "step": 27707
+ },
+ {
+ "epoch": 75.49863760217984,
+ "grad_norm": 1.3956680297851562,
+ "learning_rate": 2.98673152742587e-06,
+ "loss": 0.0115,
+ "step": 27708
+ },
+ {
+ "epoch": 75.50136239782016,
+ "grad_norm": 0.9691541194915771,
+ "learning_rate": 2.9861024773915415e-06,
+ "loss": 0.009,
+ "step": 27709
+ },
+ {
+ "epoch": 75.50408719346049,
+ "grad_norm": 1.2206989526748657,
+ "learning_rate": 2.985473481981218e-06,
+ "loss": 0.0111,
+ "step": 27710
+ },
+ {
+ "epoch": 75.50681198910081,
+ "grad_norm": 0.9768041968345642,
+ "learning_rate": 2.9848445411998006e-06,
+ "loss": 0.0096,
+ "step": 27711
+ },
+ {
+ "epoch": 75.50953678474114,
+ "grad_norm": 1.5485786199569702,
+ "learning_rate": 2.9842156550521915e-06,
+ "loss": 0.167,
+ "step": 27712
+ },
+ {
+ "epoch": 75.51226158038148,
+ "grad_norm": 1.8043142557144165,
+ "learning_rate": 2.983586823543285e-06,
+ "loss": 0.0298,
+ "step": 27713
+ },
+ {
+ "epoch": 75.5149863760218,
+ "grad_norm": 2.1543307304382324,
+ "learning_rate": 2.9829580466779784e-06,
+ "loss": 0.0886,
+ "step": 27714
+ },
+ {
+ "epoch": 75.51771117166213,
+ "grad_norm": 0.9581639170646667,
+ "learning_rate": 2.982329324461166e-06,
+ "loss": 0.0115,
+ "step": 27715
+ },
+ {
+ "epoch": 75.52043596730245,
+ "grad_norm": 1.6290799379348755,
+ "learning_rate": 2.981700656897749e-06,
+ "loss": 0.0457,
+ "step": 27716
+ },
+ {
+ "epoch": 75.52316076294278,
+ "grad_norm": 1.6131707429885864,
+ "learning_rate": 2.981072043992623e-06,
+ "loss": 0.0473,
+ "step": 27717
+ },
+ {
+ "epoch": 75.52588555858311,
+ "grad_norm": 0.9393165707588196,
+ "learning_rate": 2.9804434857506816e-06,
+ "loss": 0.0113,
+ "step": 27718
+ },
+ {
+ "epoch": 75.52861035422343,
+ "grad_norm": 1.3834254741668701,
+ "learning_rate": 2.979814982176816e-06,
+ "loss": 0.0343,
+ "step": 27719
+ },
+ {
+ "epoch": 75.53133514986376,
+ "grad_norm": 1.8039356470108032,
+ "learning_rate": 2.9791865332759294e-06,
+ "loss": 0.0773,
+ "step": 27720
+ },
+ {
+ "epoch": 75.53405994550408,
+ "grad_norm": 1.3575114011764526,
+ "learning_rate": 2.978558139052913e-06,
+ "loss": 0.0288,
+ "step": 27721
+ },
+ {
+ "epoch": 75.53678474114442,
+ "grad_norm": 1.354455828666687,
+ "learning_rate": 2.9779297995126587e-06,
+ "loss": 0.1111,
+ "step": 27722
+ },
+ {
+ "epoch": 75.53950953678473,
+ "grad_norm": 1.009718894958496,
+ "learning_rate": 2.977301514660058e-06,
+ "loss": 0.0149,
+ "step": 27723
+ },
+ {
+ "epoch": 75.54223433242507,
+ "grad_norm": 1.0304501056671143,
+ "learning_rate": 2.97667328450001e-06,
+ "loss": 0.0102,
+ "step": 27724
+ },
+ {
+ "epoch": 75.5449591280654,
+ "grad_norm": 1.7803616523742676,
+ "learning_rate": 2.9760451090374054e-06,
+ "loss": 0.0691,
+ "step": 27725
+ },
+ {
+ "epoch": 75.54768392370572,
+ "grad_norm": 0.66324383020401,
+ "learning_rate": 2.9754169882771324e-06,
+ "loss": 0.0054,
+ "step": 27726
+ },
+ {
+ "epoch": 75.55040871934605,
+ "grad_norm": 1.6474698781967163,
+ "learning_rate": 2.9747889222240878e-06,
+ "loss": 0.0745,
+ "step": 27727
+ },
+ {
+ "epoch": 75.55313351498637,
+ "grad_norm": 1.2844666242599487,
+ "learning_rate": 2.974160910883158e-06,
+ "loss": 0.0174,
+ "step": 27728
+ },
+ {
+ "epoch": 75.5558583106267,
+ "grad_norm": 1.221523404121399,
+ "learning_rate": 2.9735329542592407e-06,
+ "loss": 0.0149,
+ "step": 27729
+ },
+ {
+ "epoch": 75.55858310626704,
+ "grad_norm": 1.333221435546875,
+ "learning_rate": 2.9729050523572222e-06,
+ "loss": 0.022,
+ "step": 27730
+ },
+ {
+ "epoch": 75.56130790190736,
+ "grad_norm": 1.3610492944717407,
+ "learning_rate": 2.9722772051819902e-06,
+ "loss": 0.0332,
+ "step": 27731
+ },
+ {
+ "epoch": 75.56403269754769,
+ "grad_norm": 1.445989966392517,
+ "learning_rate": 2.9716494127384397e-06,
+ "loss": 0.1431,
+ "step": 27732
+ },
+ {
+ "epoch": 75.566757493188,
+ "grad_norm": 1.6327879428863525,
+ "learning_rate": 2.971021675031458e-06,
+ "loss": 0.0139,
+ "step": 27733
+ },
+ {
+ "epoch": 75.56948228882834,
+ "grad_norm": 2.840414524078369,
+ "learning_rate": 2.9703939920659332e-06,
+ "loss": 0.0221,
+ "step": 27734
+ },
+ {
+ "epoch": 75.57220708446866,
+ "grad_norm": 1.6742137670516968,
+ "learning_rate": 2.96976636384675e-06,
+ "loss": 0.0758,
+ "step": 27735
+ },
+ {
+ "epoch": 75.57493188010899,
+ "grad_norm": 1.2541292905807495,
+ "learning_rate": 2.9691387903788037e-06,
+ "loss": 0.0145,
+ "step": 27736
+ },
+ {
+ "epoch": 75.57765667574932,
+ "grad_norm": 1.2509669065475464,
+ "learning_rate": 2.9685112716669783e-06,
+ "loss": 0.0155,
+ "step": 27737
+ },
+ {
+ "epoch": 75.58038147138964,
+ "grad_norm": 1.6606510877609253,
+ "learning_rate": 2.967883807716161e-06,
+ "loss": 0.1222,
+ "step": 27738
+ },
+ {
+ "epoch": 75.58310626702998,
+ "grad_norm": 1.6316217184066772,
+ "learning_rate": 2.967256398531234e-06,
+ "loss": 0.0341,
+ "step": 27739
+ },
+ {
+ "epoch": 75.5858310626703,
+ "grad_norm": 1.715712547302246,
+ "learning_rate": 2.966629044117092e-06,
+ "loss": 0.0815,
+ "step": 27740
+ },
+ {
+ "epoch": 75.58855585831063,
+ "grad_norm": 1.1423470973968506,
+ "learning_rate": 2.966001744478616e-06,
+ "loss": 0.0202,
+ "step": 27741
+ },
+ {
+ "epoch": 75.59128065395096,
+ "grad_norm": 1.5261846780776978,
+ "learning_rate": 2.9653744996206926e-06,
+ "loss": 0.0123,
+ "step": 27742
+ },
+ {
+ "epoch": 75.59400544959128,
+ "grad_norm": 0.8339619040489197,
+ "learning_rate": 2.964747309548203e-06,
+ "loss": 0.0088,
+ "step": 27743
+ },
+ {
+ "epoch": 75.59673024523161,
+ "grad_norm": 1.1903103590011597,
+ "learning_rate": 2.9641201742660376e-06,
+ "loss": 0.0625,
+ "step": 27744
+ },
+ {
+ "epoch": 75.59945504087193,
+ "grad_norm": 2.5251262187957764,
+ "learning_rate": 2.963493093779075e-06,
+ "loss": 0.0512,
+ "step": 27745
+ },
+ {
+ "epoch": 75.60217983651226,
+ "grad_norm": 1.3323771953582764,
+ "learning_rate": 2.962866068092205e-06,
+ "loss": 0.0134,
+ "step": 27746
+ },
+ {
+ "epoch": 75.60490463215258,
+ "grad_norm": 3.6148688793182373,
+ "learning_rate": 2.9622390972103043e-06,
+ "loss": 0.0252,
+ "step": 27747
+ },
+ {
+ "epoch": 75.60762942779292,
+ "grad_norm": 1.4337372779846191,
+ "learning_rate": 2.9616121811382626e-06,
+ "loss": 0.015,
+ "step": 27748
+ },
+ {
+ "epoch": 75.61035422343325,
+ "grad_norm": 1.3859630823135376,
+ "learning_rate": 2.9609853198809578e-06,
+ "loss": 0.019,
+ "step": 27749
+ },
+ {
+ "epoch": 75.61307901907357,
+ "grad_norm": 1.6680941581726074,
+ "learning_rate": 2.9603585134432733e-06,
+ "loss": 0.1171,
+ "step": 27750
+ },
+ {
+ "epoch": 75.6158038147139,
+ "grad_norm": 1.6860862970352173,
+ "learning_rate": 2.959731761830087e-06,
+ "loss": 0.029,
+ "step": 27751
+ },
+ {
+ "epoch": 75.61852861035422,
+ "grad_norm": 0.7857493758201599,
+ "learning_rate": 2.959105065046286e-06,
+ "loss": 0.0121,
+ "step": 27752
+ },
+ {
+ "epoch": 75.62125340599455,
+ "grad_norm": 0.9498242139816284,
+ "learning_rate": 2.958478423096748e-06,
+ "loss": 0.0139,
+ "step": 27753
+ },
+ {
+ "epoch": 75.62397820163488,
+ "grad_norm": 1.0825029611587524,
+ "learning_rate": 2.9578518359863537e-06,
+ "loss": 0.0157,
+ "step": 27754
+ },
+ {
+ "epoch": 75.6267029972752,
+ "grad_norm": 1.3059550523757935,
+ "learning_rate": 2.9572253037199795e-06,
+ "loss": 0.0146,
+ "step": 27755
+ },
+ {
+ "epoch": 75.62942779291554,
+ "grad_norm": 1.4083844423294067,
+ "learning_rate": 2.9565988263025114e-06,
+ "loss": 0.0525,
+ "step": 27756
+ },
+ {
+ "epoch": 75.63215258855585,
+ "grad_norm": 0.7895293235778809,
+ "learning_rate": 2.955972403738825e-06,
+ "loss": 0.0088,
+ "step": 27757
+ },
+ {
+ "epoch": 75.63487738419619,
+ "grad_norm": 2.543710231781006,
+ "learning_rate": 2.9553460360337983e-06,
+ "loss": 0.0088,
+ "step": 27758
+ },
+ {
+ "epoch": 75.6376021798365,
+ "grad_norm": 1.649006962776184,
+ "learning_rate": 2.9547197231923065e-06,
+ "loss": 0.0126,
+ "step": 27759
+ },
+ {
+ "epoch": 75.64032697547684,
+ "grad_norm": 1.698600172996521,
+ "learning_rate": 2.9540934652192343e-06,
+ "loss": 0.0103,
+ "step": 27760
+ },
+ {
+ "epoch": 75.64305177111717,
+ "grad_norm": 244.89488220214844,
+ "learning_rate": 2.953467262119456e-06,
+ "loss": 0.0506,
+ "step": 27761
+ },
+ {
+ "epoch": 75.64577656675749,
+ "grad_norm": 0.9647437930107117,
+ "learning_rate": 2.952841113897844e-06,
+ "loss": 0.0122,
+ "step": 27762
+ },
+ {
+ "epoch": 75.64850136239782,
+ "grad_norm": 1.0354119539260864,
+ "learning_rate": 2.9522150205592825e-06,
+ "loss": 0.0182,
+ "step": 27763
+ },
+ {
+ "epoch": 75.65122615803814,
+ "grad_norm": 1.032551646232605,
+ "learning_rate": 2.9515889821086395e-06,
+ "loss": 0.0123,
+ "step": 27764
+ },
+ {
+ "epoch": 75.65395095367847,
+ "grad_norm": 0.9115464091300964,
+ "learning_rate": 2.950962998550798e-06,
+ "loss": 0.008,
+ "step": 27765
+ },
+ {
+ "epoch": 75.65667574931881,
+ "grad_norm": 0.7988799214363098,
+ "learning_rate": 2.9503370698906298e-06,
+ "loss": 0.0083,
+ "step": 27766
+ },
+ {
+ "epoch": 75.65940054495913,
+ "grad_norm": 1.333451509475708,
+ "learning_rate": 2.9497111961330103e-06,
+ "loss": 0.0132,
+ "step": 27767
+ },
+ {
+ "epoch": 75.66212534059946,
+ "grad_norm": 0.9003843069076538,
+ "learning_rate": 2.94908537728281e-06,
+ "loss": 0.0095,
+ "step": 27768
+ },
+ {
+ "epoch": 75.66485013623978,
+ "grad_norm": 1.8514866828918457,
+ "learning_rate": 2.9484596133449086e-06,
+ "loss": 0.1122,
+ "step": 27769
+ },
+ {
+ "epoch": 75.66757493188011,
+ "grad_norm": 1.9140628576278687,
+ "learning_rate": 2.9478339043241753e-06,
+ "loss": 0.0599,
+ "step": 27770
+ },
+ {
+ "epoch": 75.67029972752043,
+ "grad_norm": 1.4675928354263306,
+ "learning_rate": 2.947208250225484e-06,
+ "loss": 0.1658,
+ "step": 27771
+ },
+ {
+ "epoch": 75.67302452316076,
+ "grad_norm": 1.1084649562835693,
+ "learning_rate": 2.946582651053709e-06,
+ "loss": 0.0176,
+ "step": 27772
+ },
+ {
+ "epoch": 75.6757493188011,
+ "grad_norm": 1.8510009050369263,
+ "learning_rate": 2.945957106813723e-06,
+ "loss": 0.121,
+ "step": 27773
+ },
+ {
+ "epoch": 75.67847411444141,
+ "grad_norm": 1.8380153179168701,
+ "learning_rate": 2.945331617510394e-06,
+ "loss": 0.1064,
+ "step": 27774
+ },
+ {
+ "epoch": 75.68119891008175,
+ "grad_norm": 2.114459276199341,
+ "learning_rate": 2.9447061831485933e-06,
+ "loss": 0.0221,
+ "step": 27775
+ },
+ {
+ "epoch": 75.68392370572207,
+ "grad_norm": 1.3526815176010132,
+ "learning_rate": 2.944080803733197e-06,
+ "loss": 0.013,
+ "step": 27776
+ },
+ {
+ "epoch": 75.6866485013624,
+ "grad_norm": 1.9415526390075684,
+ "learning_rate": 2.943455479269073e-06,
+ "loss": 0.051,
+ "step": 27777
+ },
+ {
+ "epoch": 75.68937329700273,
+ "grad_norm": 6.786689281463623,
+ "learning_rate": 2.9428302097610894e-06,
+ "loss": 0.0237,
+ "step": 27778
+ },
+ {
+ "epoch": 75.69209809264305,
+ "grad_norm": 2.586905002593994,
+ "learning_rate": 2.9422049952141142e-06,
+ "loss": 0.0521,
+ "step": 27779
+ },
+ {
+ "epoch": 75.69482288828338,
+ "grad_norm": 1.8695236444473267,
+ "learning_rate": 2.941579835633019e-06,
+ "loss": 0.0121,
+ "step": 27780
+ },
+ {
+ "epoch": 75.6975476839237,
+ "grad_norm": 2.1375274658203125,
+ "learning_rate": 2.9409547310226773e-06,
+ "loss": 0.1528,
+ "step": 27781
+ },
+ {
+ "epoch": 75.70027247956403,
+ "grad_norm": 1.297439455986023,
+ "learning_rate": 2.9403296813879513e-06,
+ "loss": 0.0188,
+ "step": 27782
+ },
+ {
+ "epoch": 75.70299727520435,
+ "grad_norm": 1.0777246952056885,
+ "learning_rate": 2.939704686733711e-06,
+ "loss": 0.0098,
+ "step": 27783
+ },
+ {
+ "epoch": 75.70572207084469,
+ "grad_norm": 1.0678895711898804,
+ "learning_rate": 2.939079747064821e-06,
+ "loss": 0.0212,
+ "step": 27784
+ },
+ {
+ "epoch": 75.70844686648502,
+ "grad_norm": 1.6430068016052246,
+ "learning_rate": 2.9384548623861533e-06,
+ "loss": 0.1595,
+ "step": 27785
+ },
+ {
+ "epoch": 75.71117166212534,
+ "grad_norm": 1.2554768323898315,
+ "learning_rate": 2.937830032702572e-06,
+ "loss": 0.0158,
+ "step": 27786
+ },
+ {
+ "epoch": 75.71389645776567,
+ "grad_norm": 2.056936025619507,
+ "learning_rate": 2.9372052580189424e-06,
+ "loss": 0.0251,
+ "step": 27787
+ },
+ {
+ "epoch": 75.71662125340599,
+ "grad_norm": 1.972566843032837,
+ "learning_rate": 2.936580538340129e-06,
+ "loss": 0.0244,
+ "step": 27788
+ },
+ {
+ "epoch": 75.71934604904632,
+ "grad_norm": 1.2815428972244263,
+ "learning_rate": 2.9359558736710016e-06,
+ "loss": 0.0116,
+ "step": 27789
+ },
+ {
+ "epoch": 75.72207084468666,
+ "grad_norm": 2.1485788822174072,
+ "learning_rate": 2.9353312640164223e-06,
+ "loss": 0.0885,
+ "step": 27790
+ },
+ {
+ "epoch": 75.72479564032697,
+ "grad_norm": 1.8150371313095093,
+ "learning_rate": 2.934706709381253e-06,
+ "loss": 0.1304,
+ "step": 27791
+ },
+ {
+ "epoch": 75.7275204359673,
+ "grad_norm": 1.2494217157363892,
+ "learning_rate": 2.9340822097703626e-06,
+ "loss": 0.017,
+ "step": 27792
+ },
+ {
+ "epoch": 75.73024523160763,
+ "grad_norm": 2.459643602371216,
+ "learning_rate": 2.933457765188613e-06,
+ "loss": 0.0285,
+ "step": 27793
+ },
+ {
+ "epoch": 75.73297002724796,
+ "grad_norm": 2.1643970012664795,
+ "learning_rate": 2.932833375640868e-06,
+ "loss": 0.0777,
+ "step": 27794
+ },
+ {
+ "epoch": 75.73569482288828,
+ "grad_norm": 1.8657094240188599,
+ "learning_rate": 2.932209041131985e-06,
+ "loss": 0.0297,
+ "step": 27795
+ },
+ {
+ "epoch": 75.73841961852861,
+ "grad_norm": 1.609315276145935,
+ "learning_rate": 2.9315847616668315e-06,
+ "loss": 0.0359,
+ "step": 27796
+ },
+ {
+ "epoch": 75.74114441416894,
+ "grad_norm": 1.020013689994812,
+ "learning_rate": 2.93096053725027e-06,
+ "loss": 0.009,
+ "step": 27797
+ },
+ {
+ "epoch": 75.74386920980926,
+ "grad_norm": 1.6114304065704346,
+ "learning_rate": 2.930336367887161e-06,
+ "loss": 0.0296,
+ "step": 27798
+ },
+ {
+ "epoch": 75.7465940054496,
+ "grad_norm": 1.8334778547286987,
+ "learning_rate": 2.929712253582365e-06,
+ "loss": 0.027,
+ "step": 27799
+ },
+ {
+ "epoch": 75.74931880108991,
+ "grad_norm": 2.514263153076172,
+ "learning_rate": 2.929088194340739e-06,
+ "loss": 0.0488,
+ "step": 27800
+ },
+ {
+ "epoch": 75.75204359673025,
+ "grad_norm": 2.014267921447754,
+ "learning_rate": 2.928464190167151e-06,
+ "loss": 0.1026,
+ "step": 27801
+ },
+ {
+ "epoch": 75.75476839237058,
+ "grad_norm": 2.0178263187408447,
+ "learning_rate": 2.9278402410664543e-06,
+ "loss": 0.0223,
+ "step": 27802
+ },
+ {
+ "epoch": 75.7574931880109,
+ "grad_norm": 1.4705806970596313,
+ "learning_rate": 2.927216347043511e-06,
+ "loss": 0.0137,
+ "step": 27803
+ },
+ {
+ "epoch": 75.76021798365123,
+ "grad_norm": 1.415264368057251,
+ "learning_rate": 2.9265925081031767e-06,
+ "loss": 0.0205,
+ "step": 27804
+ },
+ {
+ "epoch": 75.76294277929155,
+ "grad_norm": 1.4342734813690186,
+ "learning_rate": 2.9259687242503144e-06,
+ "loss": 0.0196,
+ "step": 27805
+ },
+ {
+ "epoch": 75.76566757493188,
+ "grad_norm": 1.3472650051116943,
+ "learning_rate": 2.9253449954897807e-06,
+ "loss": 0.0081,
+ "step": 27806
+ },
+ {
+ "epoch": 75.7683923705722,
+ "grad_norm": 2.62469482421875,
+ "learning_rate": 2.9247213218264326e-06,
+ "loss": 0.0405,
+ "step": 27807
+ },
+ {
+ "epoch": 75.77111716621253,
+ "grad_norm": 3.1182966232299805,
+ "learning_rate": 2.924097703265123e-06,
+ "loss": 0.0588,
+ "step": 27808
+ },
+ {
+ "epoch": 75.77384196185287,
+ "grad_norm": 1.9819788932800293,
+ "learning_rate": 2.9234741398107168e-06,
+ "loss": 0.0843,
+ "step": 27809
+ },
+ {
+ "epoch": 75.77656675749319,
+ "grad_norm": 1.3575072288513184,
+ "learning_rate": 2.922850631468066e-06,
+ "loss": 0.022,
+ "step": 27810
+ },
+ {
+ "epoch": 75.77929155313352,
+ "grad_norm": 1.8924345970153809,
+ "learning_rate": 2.9222271782420264e-06,
+ "loss": 0.0175,
+ "step": 27811
+ },
+ {
+ "epoch": 75.78201634877384,
+ "grad_norm": 1.7712459564208984,
+ "learning_rate": 2.921603780137451e-06,
+ "loss": 0.0485,
+ "step": 27812
+ },
+ {
+ "epoch": 75.78474114441417,
+ "grad_norm": 2.2250161170959473,
+ "learning_rate": 2.9209804371592e-06,
+ "loss": 0.0202,
+ "step": 27813
+ },
+ {
+ "epoch": 75.7874659400545,
+ "grad_norm": 1.745689034461975,
+ "learning_rate": 2.920357149312123e-06,
+ "loss": 0.0335,
+ "step": 27814
+ },
+ {
+ "epoch": 75.79019073569482,
+ "grad_norm": 2.4256277084350586,
+ "learning_rate": 2.9197339166010795e-06,
+ "loss": 0.0235,
+ "step": 27815
+ },
+ {
+ "epoch": 75.79291553133515,
+ "grad_norm": 3.918071985244751,
+ "learning_rate": 2.919110739030918e-06,
+ "loss": 0.0189,
+ "step": 27816
+ },
+ {
+ "epoch": 75.79564032697547,
+ "grad_norm": 1.5287482738494873,
+ "learning_rate": 2.918487616606497e-06,
+ "loss": 0.0222,
+ "step": 27817
+ },
+ {
+ "epoch": 75.7983651226158,
+ "grad_norm": 1.9324537515640259,
+ "learning_rate": 2.9178645493326664e-06,
+ "loss": 0.0268,
+ "step": 27818
+ },
+ {
+ "epoch": 75.80108991825612,
+ "grad_norm": 2.2831215858459473,
+ "learning_rate": 2.917241537214278e-06,
+ "loss": 0.0173,
+ "step": 27819
+ },
+ {
+ "epoch": 75.80381471389646,
+ "grad_norm": 1.5520331859588623,
+ "learning_rate": 2.916618580256182e-06,
+ "loss": 0.0202,
+ "step": 27820
+ },
+ {
+ "epoch": 75.80653950953679,
+ "grad_norm": 2.2655866146087646,
+ "learning_rate": 2.915995678463236e-06,
+ "loss": 0.0313,
+ "step": 27821
+ },
+ {
+ "epoch": 75.80926430517711,
+ "grad_norm": 1.5626039505004883,
+ "learning_rate": 2.9153728318402875e-06,
+ "loss": 0.0136,
+ "step": 27822
+ },
+ {
+ "epoch": 75.81198910081744,
+ "grad_norm": 2.466392993927002,
+ "learning_rate": 2.914750040392187e-06,
+ "loss": 0.0126,
+ "step": 27823
+ },
+ {
+ "epoch": 75.81471389645776,
+ "grad_norm": 1.7367984056472778,
+ "learning_rate": 2.9141273041237817e-06,
+ "loss": 0.018,
+ "step": 27824
+ },
+ {
+ "epoch": 75.8174386920981,
+ "grad_norm": 2.92641282081604,
+ "learning_rate": 2.9135046230399288e-06,
+ "loss": 0.0153,
+ "step": 27825
+ },
+ {
+ "epoch": 75.82016348773843,
+ "grad_norm": 1.6077091693878174,
+ "learning_rate": 2.9128819971454736e-06,
+ "loss": 0.0144,
+ "step": 27826
+ },
+ {
+ "epoch": 75.82288828337875,
+ "grad_norm": 0.9457487463951111,
+ "learning_rate": 2.9122594264452654e-06,
+ "loss": 0.0093,
+ "step": 27827
+ },
+ {
+ "epoch": 75.82561307901908,
+ "grad_norm": 1.7380335330963135,
+ "learning_rate": 2.9116369109441487e-06,
+ "loss": 0.0991,
+ "step": 27828
+ },
+ {
+ "epoch": 75.8283378746594,
+ "grad_norm": 2.2889487743377686,
+ "learning_rate": 2.9110144506469806e-06,
+ "loss": 0.0313,
+ "step": 27829
+ },
+ {
+ "epoch": 75.83106267029973,
+ "grad_norm": 1.797940969467163,
+ "learning_rate": 2.910392045558602e-06,
+ "loss": 0.1403,
+ "step": 27830
+ },
+ {
+ "epoch": 75.83378746594005,
+ "grad_norm": 1.2993146181106567,
+ "learning_rate": 2.9097696956838596e-06,
+ "loss": 0.0142,
+ "step": 27831
+ },
+ {
+ "epoch": 75.83651226158038,
+ "grad_norm": 0.8207955360412598,
+ "learning_rate": 2.9091474010276024e-06,
+ "loss": 0.0074,
+ "step": 27832
+ },
+ {
+ "epoch": 75.83923705722071,
+ "grad_norm": 1.208017349243164,
+ "learning_rate": 2.908525161594681e-06,
+ "loss": 0.0135,
+ "step": 27833
+ },
+ {
+ "epoch": 75.84196185286103,
+ "grad_norm": 2.043787956237793,
+ "learning_rate": 2.9079029773899357e-06,
+ "loss": 0.0107,
+ "step": 27834
+ },
+ {
+ "epoch": 75.84468664850137,
+ "grad_norm": 0.772265613079071,
+ "learning_rate": 2.9072808484182146e-06,
+ "loss": 0.0082,
+ "step": 27835
+ },
+ {
+ "epoch": 75.84741144414168,
+ "grad_norm": 1.5053569078445435,
+ "learning_rate": 2.90665877468436e-06,
+ "loss": 0.0207,
+ "step": 27836
+ },
+ {
+ "epoch": 75.85013623978202,
+ "grad_norm": 1.5652811527252197,
+ "learning_rate": 2.9060367561932202e-06,
+ "loss": 0.021,
+ "step": 27837
+ },
+ {
+ "epoch": 75.85286103542235,
+ "grad_norm": 1.4396601915359497,
+ "learning_rate": 2.905414792949638e-06,
+ "loss": 0.0203,
+ "step": 27838
+ },
+ {
+ "epoch": 75.85558583106267,
+ "grad_norm": 1.7830204963684082,
+ "learning_rate": 2.9047928849584573e-06,
+ "loss": 0.0176,
+ "step": 27839
+ },
+ {
+ "epoch": 75.858310626703,
+ "grad_norm": 1.26956307888031,
+ "learning_rate": 2.904171032224519e-06,
+ "loss": 0.1546,
+ "step": 27840
+ },
+ {
+ "epoch": 75.86103542234332,
+ "grad_norm": 1.8014453649520874,
+ "learning_rate": 2.9035492347526704e-06,
+ "loss": 0.0595,
+ "step": 27841
+ },
+ {
+ "epoch": 75.86376021798365,
+ "grad_norm": 1.983191728591919,
+ "learning_rate": 2.902927492547751e-06,
+ "loss": 0.085,
+ "step": 27842
+ },
+ {
+ "epoch": 75.86648501362397,
+ "grad_norm": 1.2622153759002686,
+ "learning_rate": 2.9023058056146058e-06,
+ "loss": 0.012,
+ "step": 27843
+ },
+ {
+ "epoch": 75.8692098092643,
+ "grad_norm": 1.3572947978973389,
+ "learning_rate": 2.9016841739580703e-06,
+ "loss": 0.0121,
+ "step": 27844
+ },
+ {
+ "epoch": 75.87193460490464,
+ "grad_norm": 1.4660645723342896,
+ "learning_rate": 2.9010625975829944e-06,
+ "loss": 0.0336,
+ "step": 27845
+ },
+ {
+ "epoch": 75.87465940054496,
+ "grad_norm": 1.4196664094924927,
+ "learning_rate": 2.9004410764942125e-06,
+ "loss": 0.0323,
+ "step": 27846
+ },
+ {
+ "epoch": 75.87738419618529,
+ "grad_norm": 2.0645651817321777,
+ "learning_rate": 2.8998196106965693e-06,
+ "loss": 0.0197,
+ "step": 27847
+ },
+ {
+ "epoch": 75.88010899182561,
+ "grad_norm": 1.6155366897583008,
+ "learning_rate": 2.899198200194897e-06,
+ "loss": 0.0173,
+ "step": 27848
+ },
+ {
+ "epoch": 75.88283378746594,
+ "grad_norm": 0.6952047944068909,
+ "learning_rate": 2.8985768449940422e-06,
+ "loss": 0.0084,
+ "step": 27849
+ },
+ {
+ "epoch": 75.88555858310627,
+ "grad_norm": 1.0769628286361694,
+ "learning_rate": 2.897955545098845e-06,
+ "loss": 0.0108,
+ "step": 27850
+ },
+ {
+ "epoch": 75.88828337874659,
+ "grad_norm": 1.219139575958252,
+ "learning_rate": 2.897334300514142e-06,
+ "loss": 0.0258,
+ "step": 27851
+ },
+ {
+ "epoch": 75.89100817438693,
+ "grad_norm": 1.6170605421066284,
+ "learning_rate": 2.89671311124477e-06,
+ "loss": 0.0304,
+ "step": 27852
+ },
+ {
+ "epoch": 75.89373297002724,
+ "grad_norm": 1.0490398406982422,
+ "learning_rate": 2.8960919772955642e-06,
+ "loss": 0.0102,
+ "step": 27853
+ },
+ {
+ "epoch": 75.89645776566758,
+ "grad_norm": 2.010769844055176,
+ "learning_rate": 2.89547089867137e-06,
+ "loss": 0.0335,
+ "step": 27854
+ },
+ {
+ "epoch": 75.8991825613079,
+ "grad_norm": 1.1166859865188599,
+ "learning_rate": 2.894849875377018e-06,
+ "loss": 0.0537,
+ "step": 27855
+ },
+ {
+ "epoch": 75.90190735694823,
+ "grad_norm": 1.4109340906143188,
+ "learning_rate": 2.8942289074173447e-06,
+ "loss": 0.0419,
+ "step": 27856
+ },
+ {
+ "epoch": 75.90463215258856,
+ "grad_norm": 1.496193766593933,
+ "learning_rate": 2.8936079947971907e-06,
+ "loss": 0.0735,
+ "step": 27857
+ },
+ {
+ "epoch": 75.90735694822888,
+ "grad_norm": 1.6891813278198242,
+ "learning_rate": 2.8929871375213893e-06,
+ "loss": 0.0158,
+ "step": 27858
+ },
+ {
+ "epoch": 75.91008174386921,
+ "grad_norm": 1.1619486808776855,
+ "learning_rate": 2.892366335594775e-06,
+ "loss": 0.0116,
+ "step": 27859
+ },
+ {
+ "epoch": 75.91280653950953,
+ "grad_norm": 1.0610116720199585,
+ "learning_rate": 2.8917455890221803e-06,
+ "loss": 0.007,
+ "step": 27860
+ },
+ {
+ "epoch": 75.91553133514986,
+ "grad_norm": 1.0908043384552002,
+ "learning_rate": 2.891124897808445e-06,
+ "loss": 0.0184,
+ "step": 27861
+ },
+ {
+ "epoch": 75.9182561307902,
+ "grad_norm": 1.219840168952942,
+ "learning_rate": 2.890504261958399e-06,
+ "loss": 0.0105,
+ "step": 27862
+ },
+ {
+ "epoch": 75.92098092643052,
+ "grad_norm": 1.0547221899032593,
+ "learning_rate": 2.8898836814768783e-06,
+ "loss": 0.0303,
+ "step": 27863
+ },
+ {
+ "epoch": 75.92370572207085,
+ "grad_norm": 2.1760973930358887,
+ "learning_rate": 2.889263156368711e-06,
+ "loss": 0.1016,
+ "step": 27864
+ },
+ {
+ "epoch": 75.92643051771117,
+ "grad_norm": 2.945624828338623,
+ "learning_rate": 2.8886426866387342e-06,
+ "loss": 0.0265,
+ "step": 27865
+ },
+ {
+ "epoch": 75.9291553133515,
+ "grad_norm": 1.424875020980835,
+ "learning_rate": 2.8880222722917817e-06,
+ "loss": 0.025,
+ "step": 27866
+ },
+ {
+ "epoch": 75.93188010899182,
+ "grad_norm": 1.536433219909668,
+ "learning_rate": 2.8874019133326825e-06,
+ "loss": 0.0088,
+ "step": 27867
+ },
+ {
+ "epoch": 75.93460490463215,
+ "grad_norm": 1.580129623413086,
+ "learning_rate": 2.8867816097662693e-06,
+ "loss": 0.0295,
+ "step": 27868
+ },
+ {
+ "epoch": 75.93732970027249,
+ "grad_norm": 1.4072116613388062,
+ "learning_rate": 2.886161361597368e-06,
+ "loss": 0.0424,
+ "step": 27869
+ },
+ {
+ "epoch": 75.9400544959128,
+ "grad_norm": 1.8397270441055298,
+ "learning_rate": 2.8855411688308165e-06,
+ "loss": 0.0138,
+ "step": 27870
+ },
+ {
+ "epoch": 75.94277929155314,
+ "grad_norm": 1.106646180152893,
+ "learning_rate": 2.8849210314714414e-06,
+ "loss": 0.0237,
+ "step": 27871
+ },
+ {
+ "epoch": 75.94550408719346,
+ "grad_norm": 5.691196918487549,
+ "learning_rate": 2.8843009495240724e-06,
+ "loss": 0.0451,
+ "step": 27872
+ },
+ {
+ "epoch": 75.94822888283379,
+ "grad_norm": 2.1056454181671143,
+ "learning_rate": 2.883680922993536e-06,
+ "loss": 0.0872,
+ "step": 27873
+ },
+ {
+ "epoch": 75.95095367847412,
+ "grad_norm": 1.038583517074585,
+ "learning_rate": 2.8830609518846652e-06,
+ "loss": 0.0106,
+ "step": 27874
+ },
+ {
+ "epoch": 75.95367847411444,
+ "grad_norm": 2.0853703022003174,
+ "learning_rate": 2.882441036202288e-06,
+ "loss": 0.0474,
+ "step": 27875
+ },
+ {
+ "epoch": 75.95640326975477,
+ "grad_norm": 1.3650623559951782,
+ "learning_rate": 2.8818211759512306e-06,
+ "loss": 0.0103,
+ "step": 27876
+ },
+ {
+ "epoch": 75.95912806539509,
+ "grad_norm": 1.2288780212402344,
+ "learning_rate": 2.8812013711363173e-06,
+ "loss": 0.0224,
+ "step": 27877
+ },
+ {
+ "epoch": 75.96185286103542,
+ "grad_norm": 1.644055724143982,
+ "learning_rate": 2.8805816217623815e-06,
+ "loss": 0.0211,
+ "step": 27878
+ },
+ {
+ "epoch": 75.96457765667574,
+ "grad_norm": 2.267949104309082,
+ "learning_rate": 2.8799619278342474e-06,
+ "loss": 0.1125,
+ "step": 27879
+ },
+ {
+ "epoch": 75.96730245231608,
+ "grad_norm": 1.436737060546875,
+ "learning_rate": 2.879342289356737e-06,
+ "loss": 0.0629,
+ "step": 27880
+ },
+ {
+ "epoch": 75.97002724795641,
+ "grad_norm": 2.739429473876953,
+ "learning_rate": 2.8787227063346835e-06,
+ "loss": 0.1077,
+ "step": 27881
+ },
+ {
+ "epoch": 75.97275204359673,
+ "grad_norm": 1.5493152141571045,
+ "learning_rate": 2.878103178772905e-06,
+ "loss": 0.0133,
+ "step": 27882
+ },
+ {
+ "epoch": 75.97547683923706,
+ "grad_norm": 1.314499020576477,
+ "learning_rate": 2.877483706676233e-06,
+ "loss": 0.0199,
+ "step": 27883
+ },
+ {
+ "epoch": 75.97820163487738,
+ "grad_norm": 1.2717937231063843,
+ "learning_rate": 2.876864290049488e-06,
+ "loss": 0.0085,
+ "step": 27884
+ },
+ {
+ "epoch": 75.98092643051771,
+ "grad_norm": 1.375518798828125,
+ "learning_rate": 2.8762449288974924e-06,
+ "loss": 0.0126,
+ "step": 27885
+ },
+ {
+ "epoch": 75.98365122615803,
+ "grad_norm": 1.2818557024002075,
+ "learning_rate": 2.8756256232250744e-06,
+ "loss": 0.0244,
+ "step": 27886
+ },
+ {
+ "epoch": 75.98637602179836,
+ "grad_norm": 1.8098798990249634,
+ "learning_rate": 2.875006373037055e-06,
+ "loss": 0.0154,
+ "step": 27887
+ },
+ {
+ "epoch": 75.9891008174387,
+ "grad_norm": 1.383561134338379,
+ "learning_rate": 2.874387178338257e-06,
+ "loss": 0.0211,
+ "step": 27888
+ },
+ {
+ "epoch": 75.99182561307902,
+ "grad_norm": 1.5735911130905151,
+ "learning_rate": 2.873768039133499e-06,
+ "loss": 0.0369,
+ "step": 27889
+ },
+ {
+ "epoch": 75.99455040871935,
+ "grad_norm": 1.764664649963379,
+ "learning_rate": 2.8731489554276092e-06,
+ "loss": 0.0197,
+ "step": 27890
+ },
+ {
+ "epoch": 75.99727520435967,
+ "grad_norm": 1.5158990621566772,
+ "learning_rate": 2.872529927225406e-06,
+ "loss": 0.0307,
+ "step": 27891
+ },
+ {
+ "epoch": 76.0,
+ "grad_norm": 1.2362256050109863,
+ "learning_rate": 2.8719109545317102e-06,
+ "loss": 0.0163,
+ "step": 27892
+ },
+ {
+ "epoch": 76.00272479564033,
+ "grad_norm": 1.3123928308486938,
+ "learning_rate": 2.871292037351339e-06,
+ "loss": 0.0226,
+ "step": 27893
+ },
+ {
+ "epoch": 76.00544959128065,
+ "grad_norm": 1.7610481977462769,
+ "learning_rate": 2.8706731756891194e-06,
+ "loss": 0.033,
+ "step": 27894
+ },
+ {
+ "epoch": 76.00817438692098,
+ "grad_norm": 0.641431987285614,
+ "learning_rate": 2.870054369549868e-06,
+ "loss": 0.0082,
+ "step": 27895
+ },
+ {
+ "epoch": 76.0108991825613,
+ "grad_norm": 1.2448641061782837,
+ "learning_rate": 2.869435618938403e-06,
+ "loss": 0.0195,
+ "step": 27896
+ },
+ {
+ "epoch": 76.01362397820164,
+ "grad_norm": 1.8529397249221802,
+ "learning_rate": 2.868816923859542e-06,
+ "loss": 0.0294,
+ "step": 27897
+ },
+ {
+ "epoch": 76.01634877384195,
+ "grad_norm": 1.372269868850708,
+ "learning_rate": 2.868198284318108e-06,
+ "loss": 0.015,
+ "step": 27898
+ },
+ {
+ "epoch": 76.01907356948229,
+ "grad_norm": 1.5136228799819946,
+ "learning_rate": 2.8675797003189155e-06,
+ "loss": 0.027,
+ "step": 27899
+ },
+ {
+ "epoch": 76.02179836512262,
+ "grad_norm": 1.3496731519699097,
+ "learning_rate": 2.86696117186678e-06,
+ "loss": 0.019,
+ "step": 27900
+ },
+ {
+ "epoch": 76.02452316076294,
+ "grad_norm": 2.0297837257385254,
+ "learning_rate": 2.866342698966521e-06,
+ "loss": 0.0663,
+ "step": 27901
+ },
+ {
+ "epoch": 76.02724795640327,
+ "grad_norm": 11.191564559936523,
+ "learning_rate": 2.865724281622959e-06,
+ "loss": 0.0321,
+ "step": 27902
+ },
+ {
+ "epoch": 76.02997275204359,
+ "grad_norm": 1.058618187904358,
+ "learning_rate": 2.865105919840906e-06,
+ "loss": 0.0098,
+ "step": 27903
+ },
+ {
+ "epoch": 76.03269754768392,
+ "grad_norm": 1.8595688343048096,
+ "learning_rate": 2.86448761362518e-06,
+ "loss": 0.0737,
+ "step": 27904
+ },
+ {
+ "epoch": 76.03542234332426,
+ "grad_norm": 1.996392011642456,
+ "learning_rate": 2.8638693629805904e-06,
+ "loss": 0.071,
+ "step": 27905
+ },
+ {
+ "epoch": 76.03814713896458,
+ "grad_norm": 1.2503070831298828,
+ "learning_rate": 2.8632511679119603e-06,
+ "loss": 0.0102,
+ "step": 27906
+ },
+ {
+ "epoch": 76.04087193460491,
+ "grad_norm": 1.8309611082077026,
+ "learning_rate": 2.8626330284241e-06,
+ "loss": 0.0264,
+ "step": 27907
+ },
+ {
+ "epoch": 76.04359673024523,
+ "grad_norm": 1.5734673738479614,
+ "learning_rate": 2.862014944521824e-06,
+ "loss": 0.0398,
+ "step": 27908
+ },
+ {
+ "epoch": 76.04632152588556,
+ "grad_norm": 1.502317190170288,
+ "learning_rate": 2.861396916209942e-06,
+ "loss": 0.0389,
+ "step": 27909
+ },
+ {
+ "epoch": 76.04904632152588,
+ "grad_norm": 1.1744232177734375,
+ "learning_rate": 2.8607789434932753e-06,
+ "loss": 0.0152,
+ "step": 27910
+ },
+ {
+ "epoch": 76.05177111716621,
+ "grad_norm": 1.3827869892120361,
+ "learning_rate": 2.8601610263766324e-06,
+ "loss": 0.0163,
+ "step": 27911
+ },
+ {
+ "epoch": 76.05449591280654,
+ "grad_norm": 0.9813432097434998,
+ "learning_rate": 2.8595431648648255e-06,
+ "loss": 0.0112,
+ "step": 27912
+ },
+ {
+ "epoch": 76.05722070844686,
+ "grad_norm": 1.4780770540237427,
+ "learning_rate": 2.8589253589626633e-06,
+ "loss": 0.1623,
+ "step": 27913
+ },
+ {
+ "epoch": 76.0599455040872,
+ "grad_norm": 0.97819983959198,
+ "learning_rate": 2.8583076086749627e-06,
+ "loss": 0.0139,
+ "step": 27914
+ },
+ {
+ "epoch": 76.06267029972751,
+ "grad_norm": 1.7825461626052856,
+ "learning_rate": 2.857689914006534e-06,
+ "loss": 0.0301,
+ "step": 27915
+ },
+ {
+ "epoch": 76.06539509536785,
+ "grad_norm": 1.4545681476593018,
+ "learning_rate": 2.857072274962186e-06,
+ "loss": 0.0252,
+ "step": 27916
+ },
+ {
+ "epoch": 76.06811989100818,
+ "grad_norm": 1.486774206161499,
+ "learning_rate": 2.856454691546726e-06,
+ "loss": 0.0147,
+ "step": 27917
+ },
+ {
+ "epoch": 76.0708446866485,
+ "grad_norm": 2.446328639984131,
+ "learning_rate": 2.8558371637649674e-06,
+ "loss": 0.0377,
+ "step": 27918
+ },
+ {
+ "epoch": 76.07356948228883,
+ "grad_norm": 1.5603291988372803,
+ "learning_rate": 2.855219691621721e-06,
+ "loss": 0.0279,
+ "step": 27919
+ },
+ {
+ "epoch": 76.07629427792915,
+ "grad_norm": 1.4714776277542114,
+ "learning_rate": 2.854602275121795e-06,
+ "loss": 0.0169,
+ "step": 27920
+ },
+ {
+ "epoch": 76.07901907356948,
+ "grad_norm": 2.0403144359588623,
+ "learning_rate": 2.853984914269993e-06,
+ "loss": 0.1162,
+ "step": 27921
+ },
+ {
+ "epoch": 76.0817438692098,
+ "grad_norm": 1.0365897417068481,
+ "learning_rate": 2.8533676090711294e-06,
+ "loss": 0.0144,
+ "step": 27922
+ },
+ {
+ "epoch": 76.08446866485014,
+ "grad_norm": 1.7293626070022583,
+ "learning_rate": 2.852750359530009e-06,
+ "loss": 0.0599,
+ "step": 27923
+ },
+ {
+ "epoch": 76.08719346049047,
+ "grad_norm": 1.930860996246338,
+ "learning_rate": 2.8521331656514383e-06,
+ "loss": 0.1101,
+ "step": 27924
+ },
+ {
+ "epoch": 76.08991825613079,
+ "grad_norm": 1.7979450225830078,
+ "learning_rate": 2.851516027440222e-06,
+ "loss": 0.0126,
+ "step": 27925
+ },
+ {
+ "epoch": 76.09264305177112,
+ "grad_norm": 2.3357677459716797,
+ "learning_rate": 2.8508989449011716e-06,
+ "loss": 0.1078,
+ "step": 27926
+ },
+ {
+ "epoch": 76.09536784741144,
+ "grad_norm": 1.6531572341918945,
+ "learning_rate": 2.8502819180390905e-06,
+ "loss": 0.0265,
+ "step": 27927
+ },
+ {
+ "epoch": 76.09809264305177,
+ "grad_norm": 1.113015055656433,
+ "learning_rate": 2.8496649468587835e-06,
+ "loss": 0.0299,
+ "step": 27928
+ },
+ {
+ "epoch": 76.1008174386921,
+ "grad_norm": 1.3125141859054565,
+ "learning_rate": 2.8490480313650526e-06,
+ "loss": 0.0195,
+ "step": 27929
+ },
+ {
+ "epoch": 76.10354223433242,
+ "grad_norm": 1.2828595638275146,
+ "learning_rate": 2.848431171562708e-06,
+ "loss": 0.0117,
+ "step": 27930
+ },
+ {
+ "epoch": 76.10626702997276,
+ "grad_norm": 1.6184757947921753,
+ "learning_rate": 2.847814367456552e-06,
+ "loss": 0.0101,
+ "step": 27931
+ },
+ {
+ "epoch": 76.10899182561307,
+ "grad_norm": 1.5514315366744995,
+ "learning_rate": 2.8471976190513873e-06,
+ "loss": 0.0547,
+ "step": 27932
+ },
+ {
+ "epoch": 76.11171662125341,
+ "grad_norm": 0.9960105419158936,
+ "learning_rate": 2.846580926352014e-06,
+ "loss": 0.01,
+ "step": 27933
+ },
+ {
+ "epoch": 76.11444141689373,
+ "grad_norm": 1.2160007953643799,
+ "learning_rate": 2.845964289363239e-06,
+ "loss": 0.0331,
+ "step": 27934
+ },
+ {
+ "epoch": 76.11716621253406,
+ "grad_norm": 1.1956968307495117,
+ "learning_rate": 2.8453477080898663e-06,
+ "loss": 0.0121,
+ "step": 27935
+ },
+ {
+ "epoch": 76.11989100817439,
+ "grad_norm": 1.7723227739334106,
+ "learning_rate": 2.844731182536695e-06,
+ "loss": 0.0145,
+ "step": 27936
+ },
+ {
+ "epoch": 76.12261580381471,
+ "grad_norm": 1.7302359342575073,
+ "learning_rate": 2.844114712708528e-06,
+ "loss": 0.0244,
+ "step": 27937
+ },
+ {
+ "epoch": 76.12534059945504,
+ "grad_norm": 1.1293996572494507,
+ "learning_rate": 2.843498298610161e-06,
+ "loss": 0.0186,
+ "step": 27938
+ },
+ {
+ "epoch": 76.12806539509536,
+ "grad_norm": 1.7877978086471558,
+ "learning_rate": 2.8428819402464024e-06,
+ "loss": 0.0247,
+ "step": 27939
+ },
+ {
+ "epoch": 76.1307901907357,
+ "grad_norm": 21.500886917114258,
+ "learning_rate": 2.842265637622049e-06,
+ "loss": 0.0147,
+ "step": 27940
+ },
+ {
+ "epoch": 76.13351498637603,
+ "grad_norm": 1.2234078645706177,
+ "learning_rate": 2.8416493907418986e-06,
+ "loss": 0.031,
+ "step": 27941
+ },
+ {
+ "epoch": 76.13623978201635,
+ "grad_norm": 2.5453624725341797,
+ "learning_rate": 2.841033199610754e-06,
+ "loss": 0.076,
+ "step": 27942
+ },
+ {
+ "epoch": 76.13896457765668,
+ "grad_norm": 1.778342604637146,
+ "learning_rate": 2.8404170642334126e-06,
+ "loss": 0.0759,
+ "step": 27943
+ },
+ {
+ "epoch": 76.141689373297,
+ "grad_norm": 1.0074070692062378,
+ "learning_rate": 2.8398009846146733e-06,
+ "loss": 0.0127,
+ "step": 27944
+ },
+ {
+ "epoch": 76.14441416893733,
+ "grad_norm": 1.6974185705184937,
+ "learning_rate": 2.839184960759329e-06,
+ "loss": 0.0184,
+ "step": 27945
+ },
+ {
+ "epoch": 76.14713896457765,
+ "grad_norm": 1.233082890510559,
+ "learning_rate": 2.838568992672186e-06,
+ "loss": 0.0183,
+ "step": 27946
+ },
+ {
+ "epoch": 76.14986376021798,
+ "grad_norm": 0.762740433216095,
+ "learning_rate": 2.837953080358038e-06,
+ "loss": 0.0094,
+ "step": 27947
+ },
+ {
+ "epoch": 76.15258855585832,
+ "grad_norm": 1.5841423273086548,
+ "learning_rate": 2.837337223821679e-06,
+ "loss": 0.0207,
+ "step": 27948
+ },
+ {
+ "epoch": 76.15531335149863,
+ "grad_norm": 1.4041695594787598,
+ "learning_rate": 2.836721423067905e-06,
+ "loss": 0.012,
+ "step": 27949
+ },
+ {
+ "epoch": 76.15803814713897,
+ "grad_norm": 1.0608031749725342,
+ "learning_rate": 2.836105678101517e-06,
+ "loss": 0.0124,
+ "step": 27950
+ },
+ {
+ "epoch": 76.16076294277929,
+ "grad_norm": 1.2244086265563965,
+ "learning_rate": 2.8354899889273045e-06,
+ "loss": 0.021,
+ "step": 27951
+ },
+ {
+ "epoch": 76.16348773841962,
+ "grad_norm": 1.1880172491073608,
+ "learning_rate": 2.834874355550069e-06,
+ "loss": 0.0335,
+ "step": 27952
+ },
+ {
+ "epoch": 76.16621253405995,
+ "grad_norm": 0.7647704482078552,
+ "learning_rate": 2.8342587779746e-06,
+ "loss": 0.0121,
+ "step": 27953
+ },
+ {
+ "epoch": 76.16893732970027,
+ "grad_norm": 2.5330891609191895,
+ "learning_rate": 2.8336432562056914e-06,
+ "loss": 0.0365,
+ "step": 27954
+ },
+ {
+ "epoch": 76.1716621253406,
+ "grad_norm": 2.312046527862549,
+ "learning_rate": 2.833027790248141e-06,
+ "loss": 0.0233,
+ "step": 27955
+ },
+ {
+ "epoch": 76.17438692098092,
+ "grad_norm": 18.70983123779297,
+ "learning_rate": 2.8324123801067395e-06,
+ "loss": 0.1279,
+ "step": 27956
+ },
+ {
+ "epoch": 76.17711171662125,
+ "grad_norm": 1.349021077156067,
+ "learning_rate": 2.8317970257862793e-06,
+ "loss": 0.0639,
+ "step": 27957
+ },
+ {
+ "epoch": 76.17983651226157,
+ "grad_norm": 1.1312898397445679,
+ "learning_rate": 2.83118172729155e-06,
+ "loss": 0.0185,
+ "step": 27958
+ },
+ {
+ "epoch": 76.1825613079019,
+ "grad_norm": 1.2927168607711792,
+ "learning_rate": 2.8305664846273507e-06,
+ "loss": 0.0697,
+ "step": 27959
+ },
+ {
+ "epoch": 76.18528610354224,
+ "grad_norm": 3.086885690689087,
+ "learning_rate": 2.8299512977984677e-06,
+ "loss": 0.0164,
+ "step": 27960
+ },
+ {
+ "epoch": 76.18801089918256,
+ "grad_norm": 1.6537905931472778,
+ "learning_rate": 2.8293361668096943e-06,
+ "loss": 0.0565,
+ "step": 27961
+ },
+ {
+ "epoch": 76.19073569482289,
+ "grad_norm": 1.0054715871810913,
+ "learning_rate": 2.828721091665816e-06,
+ "loss": 0.0156,
+ "step": 27962
+ },
+ {
+ "epoch": 76.19346049046321,
+ "grad_norm": 1.3250573873519897,
+ "learning_rate": 2.8281060723716303e-06,
+ "loss": 0.0531,
+ "step": 27963
+ },
+ {
+ "epoch": 76.19618528610354,
+ "grad_norm": 2.018580913543701,
+ "learning_rate": 2.8274911089319247e-06,
+ "loss": 0.1041,
+ "step": 27964
+ },
+ {
+ "epoch": 76.19891008174388,
+ "grad_norm": 1.3123900890350342,
+ "learning_rate": 2.8268762013514826e-06,
+ "loss": 0.0816,
+ "step": 27965
+ },
+ {
+ "epoch": 76.2016348773842,
+ "grad_norm": 1.1279557943344116,
+ "learning_rate": 2.826261349635102e-06,
+ "loss": 0.0723,
+ "step": 27966
+ },
+ {
+ "epoch": 76.20435967302453,
+ "grad_norm": 0.8099760413169861,
+ "learning_rate": 2.8256465537875677e-06,
+ "loss": 0.013,
+ "step": 27967
+ },
+ {
+ "epoch": 76.20708446866485,
+ "grad_norm": 1.3209840059280396,
+ "learning_rate": 2.8250318138136655e-06,
+ "loss": 0.0297,
+ "step": 27968
+ },
+ {
+ "epoch": 76.20980926430518,
+ "grad_norm": 0.9188111424446106,
+ "learning_rate": 2.8244171297181834e-06,
+ "loss": 0.0088,
+ "step": 27969
+ },
+ {
+ "epoch": 76.2125340599455,
+ "grad_norm": 2.0197434425354004,
+ "learning_rate": 2.823802501505909e-06,
+ "loss": 0.0235,
+ "step": 27970
+ },
+ {
+ "epoch": 76.21525885558583,
+ "grad_norm": 1.457369089126587,
+ "learning_rate": 2.8231879291816323e-06,
+ "loss": 0.1137,
+ "step": 27971
+ },
+ {
+ "epoch": 76.21798365122616,
+ "grad_norm": 0.6080678105354309,
+ "learning_rate": 2.822573412750137e-06,
+ "loss": 0.007,
+ "step": 27972
+ },
+ {
+ "epoch": 76.22070844686648,
+ "grad_norm": 1.4967117309570312,
+ "learning_rate": 2.8219589522162094e-06,
+ "loss": 0.0214,
+ "step": 27973
+ },
+ {
+ "epoch": 76.22343324250681,
+ "grad_norm": 2.3407700061798096,
+ "learning_rate": 2.8213445475846314e-06,
+ "loss": 0.0328,
+ "step": 27974
+ },
+ {
+ "epoch": 76.22615803814713,
+ "grad_norm": 1.3002939224243164,
+ "learning_rate": 2.820730198860194e-06,
+ "loss": 0.0105,
+ "step": 27975
+ },
+ {
+ "epoch": 76.22888283378747,
+ "grad_norm": 46.56536865234375,
+ "learning_rate": 2.820115906047679e-06,
+ "loss": 0.0243,
+ "step": 27976
+ },
+ {
+ "epoch": 76.2316076294278,
+ "grad_norm": 0.9461358785629272,
+ "learning_rate": 2.8195016691518708e-06,
+ "loss": 0.0106,
+ "step": 27977
+ },
+ {
+ "epoch": 76.23433242506812,
+ "grad_norm": 1.8614747524261475,
+ "learning_rate": 2.8188874881775484e-06,
+ "loss": 0.0822,
+ "step": 27978
+ },
+ {
+ "epoch": 76.23705722070845,
+ "grad_norm": 1.2888171672821045,
+ "learning_rate": 2.8182733631295023e-06,
+ "loss": 0.0141,
+ "step": 27979
+ },
+ {
+ "epoch": 76.23978201634877,
+ "grad_norm": 1.8779618740081787,
+ "learning_rate": 2.8176592940125124e-06,
+ "loss": 0.0142,
+ "step": 27980
+ },
+ {
+ "epoch": 76.2425068119891,
+ "grad_norm": 1.2570730447769165,
+ "learning_rate": 2.8170452808313597e-06,
+ "loss": 0.0149,
+ "step": 27981
+ },
+ {
+ "epoch": 76.24523160762942,
+ "grad_norm": 1.0657998323440552,
+ "learning_rate": 2.816431323590826e-06,
+ "loss": 0.0128,
+ "step": 27982
+ },
+ {
+ "epoch": 76.24795640326975,
+ "grad_norm": 0.9279531240463257,
+ "learning_rate": 2.815817422295696e-06,
+ "loss": 0.0136,
+ "step": 27983
+ },
+ {
+ "epoch": 76.25068119891009,
+ "grad_norm": 1.3604198694229126,
+ "learning_rate": 2.815203576950748e-06,
+ "loss": 0.0843,
+ "step": 27984
+ },
+ {
+ "epoch": 76.2534059945504,
+ "grad_norm": 1.4834064245224,
+ "learning_rate": 2.8145897875607642e-06,
+ "loss": 0.0267,
+ "step": 27985
+ },
+ {
+ "epoch": 76.25613079019074,
+ "grad_norm": 0.7950231432914734,
+ "learning_rate": 2.813976054130517e-06,
+ "loss": 0.0093,
+ "step": 27986
+ },
+ {
+ "epoch": 76.25885558583106,
+ "grad_norm": 1.332221508026123,
+ "learning_rate": 2.8133623766648e-06,
+ "loss": 0.017,
+ "step": 27987
+ },
+ {
+ "epoch": 76.26158038147139,
+ "grad_norm": 1.1544386148452759,
+ "learning_rate": 2.812748755168385e-06,
+ "loss": 0.0137,
+ "step": 27988
+ },
+ {
+ "epoch": 76.26430517711172,
+ "grad_norm": 0.7946781516075134,
+ "learning_rate": 2.8121351896460526e-06,
+ "loss": 0.0101,
+ "step": 27989
+ },
+ {
+ "epoch": 76.26702997275204,
+ "grad_norm": 1.2883729934692383,
+ "learning_rate": 2.8115216801025757e-06,
+ "loss": 0.0165,
+ "step": 27990
+ },
+ {
+ "epoch": 76.26975476839237,
+ "grad_norm": 1.1367970705032349,
+ "learning_rate": 2.81090822654274e-06,
+ "loss": 0.0197,
+ "step": 27991
+ },
+ {
+ "epoch": 76.2724795640327,
+ "grad_norm": 1.7043821811676025,
+ "learning_rate": 2.8102948289713192e-06,
+ "loss": 0.0336,
+ "step": 27992
+ },
+ {
+ "epoch": 76.27520435967303,
+ "grad_norm": 0.7587525248527527,
+ "learning_rate": 2.8096814873930913e-06,
+ "loss": 0.0087,
+ "step": 27993
+ },
+ {
+ "epoch": 76.27792915531334,
+ "grad_norm": 1.2880679368972778,
+ "learning_rate": 2.809068201812829e-06,
+ "loss": 0.0194,
+ "step": 27994
+ },
+ {
+ "epoch": 76.28065395095368,
+ "grad_norm": 1.2088193893432617,
+ "learning_rate": 2.808454972235316e-06,
+ "loss": 0.0157,
+ "step": 27995
+ },
+ {
+ "epoch": 76.28337874659401,
+ "grad_norm": 2.0625696182250977,
+ "learning_rate": 2.807841798665324e-06,
+ "loss": 0.0299,
+ "step": 27996
+ },
+ {
+ "epoch": 76.28610354223433,
+ "grad_norm": 2.25197434425354,
+ "learning_rate": 2.8072286811076276e-06,
+ "loss": 0.044,
+ "step": 27997
+ },
+ {
+ "epoch": 76.28882833787466,
+ "grad_norm": 0.832569420337677,
+ "learning_rate": 2.8066156195670013e-06,
+ "loss": 0.0132,
+ "step": 27998
+ },
+ {
+ "epoch": 76.29155313351498,
+ "grad_norm": 1.913678765296936,
+ "learning_rate": 2.806002614048223e-06,
+ "loss": 0.1057,
+ "step": 27999
+ },
+ {
+ "epoch": 76.29427792915531,
+ "grad_norm": 1.887197494506836,
+ "learning_rate": 2.8053896645560653e-06,
+ "loss": 0.0329,
+ "step": 28000
+ },
+ {
+ "epoch": 76.29700272479565,
+ "grad_norm": 1.3742221593856812,
+ "learning_rate": 2.8047767710953013e-06,
+ "loss": 0.0214,
+ "step": 28001
+ },
+ {
+ "epoch": 76.29972752043597,
+ "grad_norm": 1.522825002670288,
+ "learning_rate": 2.8041639336707007e-06,
+ "loss": 0.017,
+ "step": 28002
+ },
+ {
+ "epoch": 76.3024523160763,
+ "grad_norm": 1.4680309295654297,
+ "learning_rate": 2.8035511522870395e-06,
+ "loss": 0.0867,
+ "step": 28003
+ },
+ {
+ "epoch": 76.30517711171662,
+ "grad_norm": 1.0959111452102661,
+ "learning_rate": 2.8029384269490946e-06,
+ "loss": 0.0192,
+ "step": 28004
+ },
+ {
+ "epoch": 76.30790190735695,
+ "grad_norm": 3.007317304611206,
+ "learning_rate": 2.8023257576616327e-06,
+ "loss": 0.0573,
+ "step": 28005
+ },
+ {
+ "epoch": 76.31062670299727,
+ "grad_norm": 1.7449681758880615,
+ "learning_rate": 2.8017131444294233e-06,
+ "loss": 0.0534,
+ "step": 28006
+ },
+ {
+ "epoch": 76.3133514986376,
+ "grad_norm": 1.5740036964416504,
+ "learning_rate": 2.801100587257243e-06,
+ "loss": 0.0314,
+ "step": 28007
+ },
+ {
+ "epoch": 76.31607629427793,
+ "grad_norm": 1.1445226669311523,
+ "learning_rate": 2.8004880861498595e-06,
+ "loss": 0.0088,
+ "step": 28008
+ },
+ {
+ "epoch": 76.31880108991825,
+ "grad_norm": 1.3484776020050049,
+ "learning_rate": 2.7998756411120444e-06,
+ "loss": 0.0149,
+ "step": 28009
+ },
+ {
+ "epoch": 76.32152588555859,
+ "grad_norm": 0.7448385953903198,
+ "learning_rate": 2.7992632521485617e-06,
+ "loss": 0.0077,
+ "step": 28010
+ },
+ {
+ "epoch": 76.3242506811989,
+ "grad_norm": 1.6342824697494507,
+ "learning_rate": 2.7986509192641887e-06,
+ "loss": 0.1596,
+ "step": 28011
+ },
+ {
+ "epoch": 76.32697547683924,
+ "grad_norm": 1.9447791576385498,
+ "learning_rate": 2.7980386424636895e-06,
+ "loss": 0.0254,
+ "step": 28012
+ },
+ {
+ "epoch": 76.32970027247957,
+ "grad_norm": 1.1232341527938843,
+ "learning_rate": 2.7974264217518355e-06,
+ "loss": 0.0155,
+ "step": 28013
+ },
+ {
+ "epoch": 76.33242506811989,
+ "grad_norm": 1.0135912895202637,
+ "learning_rate": 2.7968142571333878e-06,
+ "loss": 0.0339,
+ "step": 28014
+ },
+ {
+ "epoch": 76.33514986376022,
+ "grad_norm": 2.0896167755126953,
+ "learning_rate": 2.796202148613123e-06,
+ "loss": 0.0188,
+ "step": 28015
+ },
+ {
+ "epoch": 76.33787465940054,
+ "grad_norm": 1.5786594152450562,
+ "learning_rate": 2.795590096195804e-06,
+ "loss": 0.0161,
+ "step": 28016
+ },
+ {
+ "epoch": 76.34059945504087,
+ "grad_norm": 1.5696293115615845,
+ "learning_rate": 2.7949780998861966e-06,
+ "loss": 0.0151,
+ "step": 28017
+ },
+ {
+ "epoch": 76.34332425068119,
+ "grad_norm": 1.7931654453277588,
+ "learning_rate": 2.794366159689066e-06,
+ "loss": 0.0186,
+ "step": 28018
+ },
+ {
+ "epoch": 76.34604904632153,
+ "grad_norm": 1.80446195602417,
+ "learning_rate": 2.793754275609182e-06,
+ "loss": 0.0262,
+ "step": 28019
+ },
+ {
+ "epoch": 76.34877384196186,
+ "grad_norm": 1.4524002075195312,
+ "learning_rate": 2.7931424476513047e-06,
+ "loss": 0.0098,
+ "step": 28020
+ },
+ {
+ "epoch": 76.35149863760218,
+ "grad_norm": 0.7434000968933105,
+ "learning_rate": 2.7925306758202052e-06,
+ "loss": 0.0094,
+ "step": 28021
+ },
+ {
+ "epoch": 76.35422343324251,
+ "grad_norm": 1.9530599117279053,
+ "learning_rate": 2.7919189601206455e-06,
+ "loss": 0.0203,
+ "step": 28022
+ },
+ {
+ "epoch": 76.35694822888283,
+ "grad_norm": 1.257136344909668,
+ "learning_rate": 2.7913073005573854e-06,
+ "loss": 0.0342,
+ "step": 28023
+ },
+ {
+ "epoch": 76.35967302452316,
+ "grad_norm": 1.0801293849945068,
+ "learning_rate": 2.790695697135195e-06,
+ "loss": 0.018,
+ "step": 28024
+ },
+ {
+ "epoch": 76.3623978201635,
+ "grad_norm": 1.8550621271133423,
+ "learning_rate": 2.790084149858835e-06,
+ "loss": 0.0187,
+ "step": 28025
+ },
+ {
+ "epoch": 76.36512261580381,
+ "grad_norm": 1.084545612335205,
+ "learning_rate": 2.7894726587330666e-06,
+ "loss": 0.0148,
+ "step": 28026
+ },
+ {
+ "epoch": 76.36784741144415,
+ "grad_norm": 1.68168306350708,
+ "learning_rate": 2.7888612237626512e-06,
+ "loss": 0.0888,
+ "step": 28027
+ },
+ {
+ "epoch": 76.37057220708446,
+ "grad_norm": 1.5331144332885742,
+ "learning_rate": 2.788249844952354e-06,
+ "loss": 0.0551,
+ "step": 28028
+ },
+ {
+ "epoch": 76.3732970027248,
+ "grad_norm": 1.6444979906082153,
+ "learning_rate": 2.787638522306936e-06,
+ "loss": 0.0199,
+ "step": 28029
+ },
+ {
+ "epoch": 76.37602179836512,
+ "grad_norm": 1.990762710571289,
+ "learning_rate": 2.787027255831154e-06,
+ "loss": 0.1004,
+ "step": 28030
+ },
+ {
+ "epoch": 76.37874659400545,
+ "grad_norm": 1.2225631475448608,
+ "learning_rate": 2.7864160455297753e-06,
+ "loss": 0.0307,
+ "step": 28031
+ },
+ {
+ "epoch": 76.38147138964578,
+ "grad_norm": 0.9137357473373413,
+ "learning_rate": 2.7858048914075543e-06,
+ "loss": 0.0089,
+ "step": 28032
+ },
+ {
+ "epoch": 76.3841961852861,
+ "grad_norm": 1.0720516443252563,
+ "learning_rate": 2.785193793469254e-06,
+ "loss": 0.0134,
+ "step": 28033
+ },
+ {
+ "epoch": 76.38692098092643,
+ "grad_norm": 2.607311487197876,
+ "learning_rate": 2.784582751719629e-06,
+ "loss": 0.029,
+ "step": 28034
+ },
+ {
+ "epoch": 76.38964577656675,
+ "grad_norm": 1.2081210613250732,
+ "learning_rate": 2.7839717661634446e-06,
+ "loss": 0.0137,
+ "step": 28035
+ },
+ {
+ "epoch": 76.39237057220708,
+ "grad_norm": 1.3490606546401978,
+ "learning_rate": 2.7833608368054556e-06,
+ "loss": 0.03,
+ "step": 28036
+ },
+ {
+ "epoch": 76.39509536784742,
+ "grad_norm": 1.0246927738189697,
+ "learning_rate": 2.7827499636504164e-06,
+ "loss": 0.0126,
+ "step": 28037
+ },
+ {
+ "epoch": 76.39782016348774,
+ "grad_norm": 1.1422265768051147,
+ "learning_rate": 2.7821391467030923e-06,
+ "loss": 0.0205,
+ "step": 28038
+ },
+ {
+ "epoch": 76.40054495912807,
+ "grad_norm": 1.603773832321167,
+ "learning_rate": 2.781528385968233e-06,
+ "loss": 0.0178,
+ "step": 28039
+ },
+ {
+ "epoch": 76.40326975476839,
+ "grad_norm": 3.1484625339508057,
+ "learning_rate": 2.780917681450601e-06,
+ "loss": 0.057,
+ "step": 28040
+ },
+ {
+ "epoch": 76.40599455040872,
+ "grad_norm": 0.9101327657699585,
+ "learning_rate": 2.7803070331549497e-06,
+ "loss": 0.0105,
+ "step": 28041
+ },
+ {
+ "epoch": 76.40871934604904,
+ "grad_norm": 1.2554194927215576,
+ "learning_rate": 2.7796964410860348e-06,
+ "loss": 0.0472,
+ "step": 28042
+ },
+ {
+ "epoch": 76.41144414168937,
+ "grad_norm": 0.9782923460006714,
+ "learning_rate": 2.7790859052486086e-06,
+ "loss": 0.0153,
+ "step": 28043
+ },
+ {
+ "epoch": 76.4141689373297,
+ "grad_norm": 1.1057664155960083,
+ "learning_rate": 2.778475425647432e-06,
+ "loss": 0.008,
+ "step": 28044
+ },
+ {
+ "epoch": 76.41689373297002,
+ "grad_norm": 1.1977900266647339,
+ "learning_rate": 2.777865002287257e-06,
+ "loss": 0.0086,
+ "step": 28045
+ },
+ {
+ "epoch": 76.41961852861036,
+ "grad_norm": 1.450848937034607,
+ "learning_rate": 2.777254635172837e-06,
+ "loss": 0.0458,
+ "step": 28046
+ },
+ {
+ "epoch": 76.42234332425068,
+ "grad_norm": 1.9279662370681763,
+ "learning_rate": 2.7766443243089215e-06,
+ "loss": 0.0184,
+ "step": 28047
+ },
+ {
+ "epoch": 76.42506811989101,
+ "grad_norm": 1.562779188156128,
+ "learning_rate": 2.776034069700271e-06,
+ "loss": 0.0224,
+ "step": 28048
+ },
+ {
+ "epoch": 76.42779291553134,
+ "grad_norm": 0.9556747078895569,
+ "learning_rate": 2.7754238713516347e-06,
+ "loss": 0.008,
+ "step": 28049
+ },
+ {
+ "epoch": 76.43051771117166,
+ "grad_norm": 1.3891842365264893,
+ "learning_rate": 2.7748137292677647e-06,
+ "loss": 0.0243,
+ "step": 28050
+ },
+ {
+ "epoch": 76.433242506812,
+ "grad_norm": 1.7976990938186646,
+ "learning_rate": 2.7742036434534094e-06,
+ "loss": 0.015,
+ "step": 28051
+ },
+ {
+ "epoch": 76.43596730245231,
+ "grad_norm": 1.4768253564834595,
+ "learning_rate": 2.773593613913327e-06,
+ "loss": 0.0386,
+ "step": 28052
+ },
+ {
+ "epoch": 76.43869209809264,
+ "grad_norm": 1.071175217628479,
+ "learning_rate": 2.772983640652265e-06,
+ "loss": 0.013,
+ "step": 28053
+ },
+ {
+ "epoch": 76.44141689373296,
+ "grad_norm": 1.9193652868270874,
+ "learning_rate": 2.7723737236749705e-06,
+ "loss": 0.0364,
+ "step": 28054
+ },
+ {
+ "epoch": 76.4441416893733,
+ "grad_norm": 1.279037594795227,
+ "learning_rate": 2.7717638629861965e-06,
+ "loss": 0.0581,
+ "step": 28055
+ },
+ {
+ "epoch": 76.44686648501363,
+ "grad_norm": 2.0172791481018066,
+ "learning_rate": 2.771154058590697e-06,
+ "loss": 0.0377,
+ "step": 28056
+ },
+ {
+ "epoch": 76.44959128065395,
+ "grad_norm": 0.8266141414642334,
+ "learning_rate": 2.7705443104932163e-06,
+ "loss": 0.0167,
+ "step": 28057
+ },
+ {
+ "epoch": 76.45231607629428,
+ "grad_norm": 1.0469287633895874,
+ "learning_rate": 2.7699346186985044e-06,
+ "loss": 0.0125,
+ "step": 28058
+ },
+ {
+ "epoch": 76.4550408719346,
+ "grad_norm": 4.007407188415527,
+ "learning_rate": 2.7693249832113055e-06,
+ "loss": 0.0205,
+ "step": 28059
+ },
+ {
+ "epoch": 76.45776566757493,
+ "grad_norm": 1.5792434215545654,
+ "learning_rate": 2.768715404036374e-06,
+ "loss": 0.0165,
+ "step": 28060
+ },
+ {
+ "epoch": 76.46049046321527,
+ "grad_norm": 1.10784912109375,
+ "learning_rate": 2.7681058811784545e-06,
+ "loss": 0.0129,
+ "step": 28061
+ },
+ {
+ "epoch": 76.46321525885558,
+ "grad_norm": 0.5141517519950867,
+ "learning_rate": 2.7674964146422933e-06,
+ "loss": 0.0051,
+ "step": 28062
+ },
+ {
+ "epoch": 76.46594005449592,
+ "grad_norm": 0.8441340327262878,
+ "learning_rate": 2.7668870044326335e-06,
+ "loss": 0.0098,
+ "step": 28063
+ },
+ {
+ "epoch": 76.46866485013624,
+ "grad_norm": 1.3589928150177002,
+ "learning_rate": 2.7662776505542286e-06,
+ "loss": 0.0179,
+ "step": 28064
+ },
+ {
+ "epoch": 76.47138964577657,
+ "grad_norm": 2.0451114177703857,
+ "learning_rate": 2.765668353011821e-06,
+ "loss": 0.039,
+ "step": 28065
+ },
+ {
+ "epoch": 76.47411444141689,
+ "grad_norm": 0.8782913684844971,
+ "learning_rate": 2.7650591118101557e-06,
+ "loss": 0.0057,
+ "step": 28066
+ },
+ {
+ "epoch": 76.47683923705722,
+ "grad_norm": 1.017835021018982,
+ "learning_rate": 2.7644499269539728e-06,
+ "loss": 0.0098,
+ "step": 28067
+ },
+ {
+ "epoch": 76.47956403269755,
+ "grad_norm": 1.4559831619262695,
+ "learning_rate": 2.7638407984480255e-06,
+ "loss": 0.0149,
+ "step": 28068
+ },
+ {
+ "epoch": 76.48228882833787,
+ "grad_norm": 1.934987187385559,
+ "learning_rate": 2.763231726297052e-06,
+ "loss": 0.1418,
+ "step": 28069
+ },
+ {
+ "epoch": 76.4850136239782,
+ "grad_norm": 1.1193591356277466,
+ "learning_rate": 2.762622710505797e-06,
+ "loss": 0.0114,
+ "step": 28070
+ },
+ {
+ "epoch": 76.48773841961852,
+ "grad_norm": 1.9656730890274048,
+ "learning_rate": 2.762013751079001e-06,
+ "loss": 0.0689,
+ "step": 28071
+ },
+ {
+ "epoch": 76.49046321525886,
+ "grad_norm": 2.0219478607177734,
+ "learning_rate": 2.7614048480214085e-06,
+ "loss": 0.0665,
+ "step": 28072
+ },
+ {
+ "epoch": 76.49318801089919,
+ "grad_norm": 1.0010136365890503,
+ "learning_rate": 2.7607960013377655e-06,
+ "loss": 0.0106,
+ "step": 28073
+ },
+ {
+ "epoch": 76.49591280653951,
+ "grad_norm": 1.6797635555267334,
+ "learning_rate": 2.76018721103281e-06,
+ "loss": 0.0814,
+ "step": 28074
+ },
+ {
+ "epoch": 76.49863760217984,
+ "grad_norm": 1.160410761833191,
+ "learning_rate": 2.7595784771112798e-06,
+ "loss": 0.0135,
+ "step": 28075
+ },
+ {
+ "epoch": 76.50136239782016,
+ "grad_norm": 0.8656134605407715,
+ "learning_rate": 2.758969799577923e-06,
+ "loss": 0.01,
+ "step": 28076
+ },
+ {
+ "epoch": 76.50408719346049,
+ "grad_norm": 1.6366562843322754,
+ "learning_rate": 2.7583611784374766e-06,
+ "loss": 0.0135,
+ "step": 28077
+ },
+ {
+ "epoch": 76.50681198910081,
+ "grad_norm": 1.0495240688323975,
+ "learning_rate": 2.7577526136946797e-06,
+ "loss": 0.0088,
+ "step": 28078
+ },
+ {
+ "epoch": 76.50953678474114,
+ "grad_norm": 2.012136936187744,
+ "learning_rate": 2.7571441053542703e-06,
+ "loss": 0.0157,
+ "step": 28079
+ },
+ {
+ "epoch": 76.51226158038148,
+ "grad_norm": 1.750071406364441,
+ "learning_rate": 2.756535653420992e-06,
+ "loss": 0.0122,
+ "step": 28080
+ },
+ {
+ "epoch": 76.5149863760218,
+ "grad_norm": 2.455294609069824,
+ "learning_rate": 2.755927257899581e-06,
+ "loss": 0.0922,
+ "step": 28081
+ },
+ {
+ "epoch": 76.51771117166213,
+ "grad_norm": 1.3797082901000977,
+ "learning_rate": 2.7553189187947747e-06,
+ "loss": 0.0127,
+ "step": 28082
+ },
+ {
+ "epoch": 76.52043596730245,
+ "grad_norm": 1.633833885192871,
+ "learning_rate": 2.7547106361113087e-06,
+ "loss": 0.0179,
+ "step": 28083
+ },
+ {
+ "epoch": 76.52316076294278,
+ "grad_norm": 1.0623663663864136,
+ "learning_rate": 2.754102409853927e-06,
+ "loss": 0.0083,
+ "step": 28084
+ },
+ {
+ "epoch": 76.52588555858311,
+ "grad_norm": 1.5212554931640625,
+ "learning_rate": 2.753494240027361e-06,
+ "loss": 0.0172,
+ "step": 28085
+ },
+ {
+ "epoch": 76.52861035422343,
+ "grad_norm": 1.8453058004379272,
+ "learning_rate": 2.7528861266363507e-06,
+ "loss": 0.0314,
+ "step": 28086
+ },
+ {
+ "epoch": 76.53133514986376,
+ "grad_norm": 3.6332762241363525,
+ "learning_rate": 2.7522780696856256e-06,
+ "loss": 0.0229,
+ "step": 28087
+ },
+ {
+ "epoch": 76.53405994550408,
+ "grad_norm": 1.0190356969833374,
+ "learning_rate": 2.751670069179928e-06,
+ "loss": 0.0235,
+ "step": 28088
+ },
+ {
+ "epoch": 76.53678474114442,
+ "grad_norm": 1.023364782333374,
+ "learning_rate": 2.751062125123989e-06,
+ "loss": 0.0118,
+ "step": 28089
+ },
+ {
+ "epoch": 76.53950953678473,
+ "grad_norm": 1.364981770515442,
+ "learning_rate": 2.7504542375225474e-06,
+ "loss": 0.0162,
+ "step": 28090
+ },
+ {
+ "epoch": 76.54223433242507,
+ "grad_norm": 2.0199155807495117,
+ "learning_rate": 2.7498464063803342e-06,
+ "loss": 0.0132,
+ "step": 28091
+ },
+ {
+ "epoch": 76.5449591280654,
+ "grad_norm": 1.2104275226593018,
+ "learning_rate": 2.7492386317020814e-06,
+ "loss": 0.0206,
+ "step": 28092
+ },
+ {
+ "epoch": 76.54768392370572,
+ "grad_norm": 0.8685348629951477,
+ "learning_rate": 2.748630913492528e-06,
+ "loss": 0.0103,
+ "step": 28093
+ },
+ {
+ "epoch": 76.55040871934605,
+ "grad_norm": 1.5394840240478516,
+ "learning_rate": 2.7480232517564033e-06,
+ "loss": 0.0702,
+ "step": 28094
+ },
+ {
+ "epoch": 76.55313351498637,
+ "grad_norm": 1.2181142568588257,
+ "learning_rate": 2.747415646498438e-06,
+ "loss": 0.0158,
+ "step": 28095
+ },
+ {
+ "epoch": 76.5558583106267,
+ "grad_norm": 1.6132469177246094,
+ "learning_rate": 2.746808097723368e-06,
+ "loss": 0.0979,
+ "step": 28096
+ },
+ {
+ "epoch": 76.55858310626704,
+ "grad_norm": 1.6030633449554443,
+ "learning_rate": 2.7462006054359234e-06,
+ "loss": 0.0403,
+ "step": 28097
+ },
+ {
+ "epoch": 76.56130790190736,
+ "grad_norm": 1.0047781467437744,
+ "learning_rate": 2.7455931696408356e-06,
+ "loss": 0.0074,
+ "step": 28098
+ },
+ {
+ "epoch": 76.56403269754769,
+ "grad_norm": 1.6863240003585815,
+ "learning_rate": 2.7449857903428314e-06,
+ "loss": 0.0482,
+ "step": 28099
+ },
+ {
+ "epoch": 76.566757493188,
+ "grad_norm": 0.9918360710144043,
+ "learning_rate": 2.7443784675466465e-06,
+ "loss": 0.0095,
+ "step": 28100
+ },
+ {
+ "epoch": 76.56948228882834,
+ "grad_norm": 1.4127418994903564,
+ "learning_rate": 2.7437712012570105e-06,
+ "loss": 0.0216,
+ "step": 28101
+ },
+ {
+ "epoch": 76.57220708446866,
+ "grad_norm": 1.1814707517623901,
+ "learning_rate": 2.7431639914786488e-06,
+ "loss": 0.12,
+ "step": 28102
+ },
+ {
+ "epoch": 76.57493188010899,
+ "grad_norm": 0.8436428308486938,
+ "learning_rate": 2.7425568382162904e-06,
+ "loss": 0.0086,
+ "step": 28103
+ },
+ {
+ "epoch": 76.57765667574932,
+ "grad_norm": 1.3575894832611084,
+ "learning_rate": 2.7419497414746687e-06,
+ "loss": 0.0224,
+ "step": 28104
+ },
+ {
+ "epoch": 76.58038147138964,
+ "grad_norm": 1.5184028148651123,
+ "learning_rate": 2.741342701258509e-06,
+ "loss": 0.0237,
+ "step": 28105
+ },
+ {
+ "epoch": 76.58310626702998,
+ "grad_norm": 1.1486246585845947,
+ "learning_rate": 2.7407357175725356e-06,
+ "loss": 0.0111,
+ "step": 28106
+ },
+ {
+ "epoch": 76.5858310626703,
+ "grad_norm": 2.32938289642334,
+ "learning_rate": 2.7401287904214813e-06,
+ "loss": 0.0872,
+ "step": 28107
+ },
+ {
+ "epoch": 76.58855585831063,
+ "grad_norm": 1.6810985803604126,
+ "learning_rate": 2.7395219198100674e-06,
+ "loss": 0.082,
+ "step": 28108
+ },
+ {
+ "epoch": 76.59128065395096,
+ "grad_norm": 1.2044918537139893,
+ "learning_rate": 2.7389151057430272e-06,
+ "loss": 0.0196,
+ "step": 28109
+ },
+ {
+ "epoch": 76.59400544959128,
+ "grad_norm": 0.881751537322998,
+ "learning_rate": 2.7383083482250815e-06,
+ "loss": 0.022,
+ "step": 28110
+ },
+ {
+ "epoch": 76.59673024523161,
+ "grad_norm": 0.9326409101486206,
+ "learning_rate": 2.7377016472609562e-06,
+ "loss": 0.0147,
+ "step": 28111
+ },
+ {
+ "epoch": 76.59945504087193,
+ "grad_norm": 0.8419937491416931,
+ "learning_rate": 2.7370950028553744e-06,
+ "loss": 0.0127,
+ "step": 28112
+ },
+ {
+ "epoch": 76.60217983651226,
+ "grad_norm": 1.572833776473999,
+ "learning_rate": 2.7364884150130667e-06,
+ "loss": 0.0254,
+ "step": 28113
+ },
+ {
+ "epoch": 76.60490463215258,
+ "grad_norm": 2.0955581665039062,
+ "learning_rate": 2.7358818837387525e-06,
+ "loss": 0.0232,
+ "step": 28114
+ },
+ {
+ "epoch": 76.60762942779292,
+ "grad_norm": 1.1870237588882446,
+ "learning_rate": 2.7352754090371538e-06,
+ "loss": 0.0108,
+ "step": 28115
+ },
+ {
+ "epoch": 76.61035422343325,
+ "grad_norm": 2.9489495754241943,
+ "learning_rate": 2.734668990912999e-06,
+ "loss": 0.0147,
+ "step": 28116
+ },
+ {
+ "epoch": 76.61307901907357,
+ "grad_norm": 1.6577178239822388,
+ "learning_rate": 2.734062629371008e-06,
+ "loss": 0.044,
+ "step": 28117
+ },
+ {
+ "epoch": 76.6158038147139,
+ "grad_norm": 1.044960618019104,
+ "learning_rate": 2.733456324415904e-06,
+ "loss": 0.0111,
+ "step": 28118
+ },
+ {
+ "epoch": 76.61852861035422,
+ "grad_norm": 1.246148705482483,
+ "learning_rate": 2.732850076052406e-06,
+ "loss": 0.0117,
+ "step": 28119
+ },
+ {
+ "epoch": 76.62125340599455,
+ "grad_norm": 1.1743625402450562,
+ "learning_rate": 2.7322438842852394e-06,
+ "loss": 0.01,
+ "step": 28120
+ },
+ {
+ "epoch": 76.62397820163488,
+ "grad_norm": 1.2764866352081299,
+ "learning_rate": 2.7316377491191236e-06,
+ "loss": 0.0172,
+ "step": 28121
+ },
+ {
+ "epoch": 76.6267029972752,
+ "grad_norm": 1.6200811862945557,
+ "learning_rate": 2.7310316705587793e-06,
+ "loss": 0.0904,
+ "step": 28122
+ },
+ {
+ "epoch": 76.62942779291554,
+ "grad_norm": 1.1799904108047485,
+ "learning_rate": 2.730425648608923e-06,
+ "loss": 0.016,
+ "step": 28123
+ },
+ {
+ "epoch": 76.63215258855585,
+ "grad_norm": 2.0518510341644287,
+ "learning_rate": 2.7298196832742786e-06,
+ "loss": 0.1674,
+ "step": 28124
+ },
+ {
+ "epoch": 76.63487738419619,
+ "grad_norm": 3.029378652572632,
+ "learning_rate": 2.729213774559567e-06,
+ "loss": 0.0118,
+ "step": 28125
+ },
+ {
+ "epoch": 76.6376021798365,
+ "grad_norm": 1.067708134651184,
+ "learning_rate": 2.728607922469505e-06,
+ "loss": 0.0366,
+ "step": 28126
+ },
+ {
+ "epoch": 76.64032697547684,
+ "grad_norm": 1.661458969116211,
+ "learning_rate": 2.728002127008811e-06,
+ "loss": 0.0126,
+ "step": 28127
+ },
+ {
+ "epoch": 76.64305177111717,
+ "grad_norm": 1.4130337238311768,
+ "learning_rate": 2.7273963881821984e-06,
+ "loss": 0.0748,
+ "step": 28128
+ },
+ {
+ "epoch": 76.64577656675749,
+ "grad_norm": 1.5655823945999146,
+ "learning_rate": 2.726790705994391e-06,
+ "loss": 0.0536,
+ "step": 28129
+ },
+ {
+ "epoch": 76.64850136239782,
+ "grad_norm": 1.0338757038116455,
+ "learning_rate": 2.7261850804501043e-06,
+ "loss": 0.0142,
+ "step": 28130
+ },
+ {
+ "epoch": 76.65122615803814,
+ "grad_norm": 1.6296470165252686,
+ "learning_rate": 2.725579511554055e-06,
+ "loss": 0.0895,
+ "step": 28131
+ },
+ {
+ "epoch": 76.65395095367847,
+ "grad_norm": 1.534921407699585,
+ "learning_rate": 2.724973999310954e-06,
+ "loss": 0.0305,
+ "step": 28132
+ },
+ {
+ "epoch": 76.65667574931881,
+ "grad_norm": 1.79728102684021,
+ "learning_rate": 2.7243685437255253e-06,
+ "loss": 0.1335,
+ "step": 28133
+ },
+ {
+ "epoch": 76.65940054495913,
+ "grad_norm": 1.2536066770553589,
+ "learning_rate": 2.7237631448024792e-06,
+ "loss": 0.1207,
+ "step": 28134
+ },
+ {
+ "epoch": 76.66212534059946,
+ "grad_norm": 0.6848405003547668,
+ "learning_rate": 2.723157802546532e-06,
+ "loss": 0.0078,
+ "step": 28135
+ },
+ {
+ "epoch": 76.66485013623978,
+ "grad_norm": 1.0660815238952637,
+ "learning_rate": 2.7225525169623947e-06,
+ "loss": 0.0874,
+ "step": 28136
+ },
+ {
+ "epoch": 76.66757493188011,
+ "grad_norm": 2.1440320014953613,
+ "learning_rate": 2.721947288054787e-06,
+ "loss": 0.0263,
+ "step": 28137
+ },
+ {
+ "epoch": 76.67029972752043,
+ "grad_norm": 1.3180049657821655,
+ "learning_rate": 2.7213421158284203e-06,
+ "loss": 0.0538,
+ "step": 28138
+ },
+ {
+ "epoch": 76.67302452316076,
+ "grad_norm": 1.4414622783660889,
+ "learning_rate": 2.720737000288003e-06,
+ "loss": 0.0493,
+ "step": 28139
+ },
+ {
+ "epoch": 76.6757493188011,
+ "grad_norm": 1.006507396697998,
+ "learning_rate": 2.7201319414382554e-06,
+ "loss": 0.0424,
+ "step": 28140
+ },
+ {
+ "epoch": 76.67847411444141,
+ "grad_norm": 1.050891637802124,
+ "learning_rate": 2.7195269392838818e-06,
+ "loss": 0.0111,
+ "step": 28141
+ },
+ {
+ "epoch": 76.68119891008175,
+ "grad_norm": 1.0072542428970337,
+ "learning_rate": 2.718921993829603e-06,
+ "loss": 0.013,
+ "step": 28142
+ },
+ {
+ "epoch": 76.68392370572207,
+ "grad_norm": 2.3462789058685303,
+ "learning_rate": 2.718317105080124e-06,
+ "loss": 0.0669,
+ "step": 28143
+ },
+ {
+ "epoch": 76.6866485013624,
+ "grad_norm": 1.3570564985275269,
+ "learning_rate": 2.717712273040154e-06,
+ "loss": 0.0113,
+ "step": 28144
+ },
+ {
+ "epoch": 76.68937329700273,
+ "grad_norm": 0.7255685925483704,
+ "learning_rate": 2.71710749771441e-06,
+ "loss": 0.0123,
+ "step": 28145
+ },
+ {
+ "epoch": 76.69209809264305,
+ "grad_norm": 1.404859185218811,
+ "learning_rate": 2.7165027791075984e-06,
+ "loss": 0.0615,
+ "step": 28146
+ },
+ {
+ "epoch": 76.69482288828338,
+ "grad_norm": 1.336265206336975,
+ "learning_rate": 2.715898117224428e-06,
+ "loss": 0.0206,
+ "step": 28147
+ },
+ {
+ "epoch": 76.6975476839237,
+ "grad_norm": 1.5865826606750488,
+ "learning_rate": 2.7152935120696056e-06,
+ "loss": 0.0308,
+ "step": 28148
+ },
+ {
+ "epoch": 76.70027247956403,
+ "grad_norm": 1.440368413925171,
+ "learning_rate": 2.7146889636478467e-06,
+ "loss": 0.0212,
+ "step": 28149
+ },
+ {
+ "epoch": 76.70299727520435,
+ "grad_norm": 1.0538734197616577,
+ "learning_rate": 2.714084471963855e-06,
+ "loss": 0.1459,
+ "step": 28150
+ },
+ {
+ "epoch": 76.70572207084469,
+ "grad_norm": 1.0240758657455444,
+ "learning_rate": 2.713480037022339e-06,
+ "loss": 0.0084,
+ "step": 28151
+ },
+ {
+ "epoch": 76.70844686648502,
+ "grad_norm": 1.1455206871032715,
+ "learning_rate": 2.712875658828004e-06,
+ "loss": 0.0141,
+ "step": 28152
+ },
+ {
+ "epoch": 76.71117166212534,
+ "grad_norm": 1.5490230321884155,
+ "learning_rate": 2.7122713373855605e-06,
+ "loss": 0.0272,
+ "step": 28153
+ },
+ {
+ "epoch": 76.71389645776567,
+ "grad_norm": 1.0592633485794067,
+ "learning_rate": 2.7116670726997142e-06,
+ "loss": 0.0224,
+ "step": 28154
+ },
+ {
+ "epoch": 76.71662125340599,
+ "grad_norm": 1.9334080219268799,
+ "learning_rate": 2.7110628647751704e-06,
+ "loss": 0.0747,
+ "step": 28155
+ },
+ {
+ "epoch": 76.71934604904632,
+ "grad_norm": 0.5046020150184631,
+ "learning_rate": 2.7104587136166304e-06,
+ "loss": 0.0053,
+ "step": 28156
+ },
+ {
+ "epoch": 76.72207084468666,
+ "grad_norm": 1.4300954341888428,
+ "learning_rate": 2.709854619228807e-06,
+ "loss": 0.0185,
+ "step": 28157
+ },
+ {
+ "epoch": 76.72479564032697,
+ "grad_norm": 0.9573366045951843,
+ "learning_rate": 2.709250581616397e-06,
+ "loss": 0.0102,
+ "step": 28158
+ },
+ {
+ "epoch": 76.7275204359673,
+ "grad_norm": 1.576087236404419,
+ "learning_rate": 2.7086466007841126e-06,
+ "loss": 0.079,
+ "step": 28159
+ },
+ {
+ "epoch": 76.73024523160763,
+ "grad_norm": 1.0755395889282227,
+ "learning_rate": 2.708042676736651e-06,
+ "loss": 0.0105,
+ "step": 28160
+ },
+ {
+ "epoch": 76.73297002724796,
+ "grad_norm": 2.5001885890960693,
+ "learning_rate": 2.707438809478722e-06,
+ "loss": 0.0203,
+ "step": 28161
+ },
+ {
+ "epoch": 76.73569482288828,
+ "grad_norm": 1.4444077014923096,
+ "learning_rate": 2.7068349990150235e-06,
+ "loss": 0.0114,
+ "step": 28162
+ },
+ {
+ "epoch": 76.73841961852861,
+ "grad_norm": 1.570514440536499,
+ "learning_rate": 2.7062312453502606e-06,
+ "loss": 0.0811,
+ "step": 28163
+ },
+ {
+ "epoch": 76.74114441416894,
+ "grad_norm": 1.1235564947128296,
+ "learning_rate": 2.70562754848913e-06,
+ "loss": 0.0343,
+ "step": 28164
+ },
+ {
+ "epoch": 76.74386920980926,
+ "grad_norm": 1.9828791618347168,
+ "learning_rate": 2.7050239084363404e-06,
+ "loss": 0.0961,
+ "step": 28165
+ },
+ {
+ "epoch": 76.7465940054496,
+ "grad_norm": 1.4147969484329224,
+ "learning_rate": 2.7044203251965907e-06,
+ "loss": 0.0285,
+ "step": 28166
+ },
+ {
+ "epoch": 76.74931880108991,
+ "grad_norm": 0.7231345772743225,
+ "learning_rate": 2.703816798774579e-06,
+ "loss": 0.0103,
+ "step": 28167
+ },
+ {
+ "epoch": 76.75204359673025,
+ "grad_norm": 1.8331745862960815,
+ "learning_rate": 2.703213329175006e-06,
+ "loss": 0.0819,
+ "step": 28168
+ },
+ {
+ "epoch": 76.75476839237058,
+ "grad_norm": 1.7541906833648682,
+ "learning_rate": 2.702609916402574e-06,
+ "loss": 0.0905,
+ "step": 28169
+ },
+ {
+ "epoch": 76.7574931880109,
+ "grad_norm": 1.5670959949493408,
+ "learning_rate": 2.7020065604619826e-06,
+ "loss": 0.1105,
+ "step": 28170
+ },
+ {
+ "epoch": 76.76021798365123,
+ "grad_norm": 1.8286930322647095,
+ "learning_rate": 2.701403261357929e-06,
+ "loss": 0.0788,
+ "step": 28171
+ },
+ {
+ "epoch": 76.76294277929155,
+ "grad_norm": 1.3882129192352295,
+ "learning_rate": 2.700800019095109e-06,
+ "loss": 0.0207,
+ "step": 28172
+ },
+ {
+ "epoch": 76.76566757493188,
+ "grad_norm": 1.5177198648452759,
+ "learning_rate": 2.700196833678226e-06,
+ "loss": 0.0338,
+ "step": 28173
+ },
+ {
+ "epoch": 76.7683923705722,
+ "grad_norm": 1.3804713487625122,
+ "learning_rate": 2.6995937051119747e-06,
+ "loss": 0.0138,
+ "step": 28174
+ },
+ {
+ "epoch": 76.77111716621253,
+ "grad_norm": 1.446550965309143,
+ "learning_rate": 2.6989906334010507e-06,
+ "loss": 0.022,
+ "step": 28175
+ },
+ {
+ "epoch": 76.77384196185287,
+ "grad_norm": 1.313765048980713,
+ "learning_rate": 2.6983876185501556e-06,
+ "loss": 0.051,
+ "step": 28176
+ },
+ {
+ "epoch": 76.77656675749319,
+ "grad_norm": 1.0687540769577026,
+ "learning_rate": 2.697784660563979e-06,
+ "loss": 0.0179,
+ "step": 28177
+ },
+ {
+ "epoch": 76.77929155313352,
+ "grad_norm": 1.6766928434371948,
+ "learning_rate": 2.697181759447224e-06,
+ "loss": 0.0256,
+ "step": 28178
+ },
+ {
+ "epoch": 76.78201634877384,
+ "grad_norm": 1.1846141815185547,
+ "learning_rate": 2.6965789152045818e-06,
+ "loss": 0.0162,
+ "step": 28179
+ },
+ {
+ "epoch": 76.78474114441417,
+ "grad_norm": 0.9085476398468018,
+ "learning_rate": 2.695976127840745e-06,
+ "loss": 0.013,
+ "step": 28180
+ },
+ {
+ "epoch": 76.7874659400545,
+ "grad_norm": 0.9671163558959961,
+ "learning_rate": 2.6953733973604147e-06,
+ "loss": 0.0082,
+ "step": 28181
+ },
+ {
+ "epoch": 76.79019073569482,
+ "grad_norm": 3.2547340393066406,
+ "learning_rate": 2.6947707237682807e-06,
+ "loss": 0.0192,
+ "step": 28182
+ },
+ {
+ "epoch": 76.79291553133515,
+ "grad_norm": 1.812705159187317,
+ "learning_rate": 2.6941681070690374e-06,
+ "loss": 0.0626,
+ "step": 28183
+ },
+ {
+ "epoch": 76.79564032697547,
+ "grad_norm": 1.5735973119735718,
+ "learning_rate": 2.693565547267375e-06,
+ "loss": 0.0269,
+ "step": 28184
+ },
+ {
+ "epoch": 76.7983651226158,
+ "grad_norm": 1.8227083683013916,
+ "learning_rate": 2.6929630443679923e-06,
+ "loss": 0.0228,
+ "step": 28185
+ },
+ {
+ "epoch": 76.80108991825612,
+ "grad_norm": 0.8673218488693237,
+ "learning_rate": 2.6923605983755772e-06,
+ "loss": 0.009,
+ "step": 28186
+ },
+ {
+ "epoch": 76.80381471389646,
+ "grad_norm": 1.5926611423492432,
+ "learning_rate": 2.6917582092948235e-06,
+ "loss": 0.0138,
+ "step": 28187
+ },
+ {
+ "epoch": 76.80653950953679,
+ "grad_norm": 1.2671302556991577,
+ "learning_rate": 2.691155877130418e-06,
+ "loss": 0.041,
+ "step": 28188
+ },
+ {
+ "epoch": 76.80926430517711,
+ "grad_norm": 1.0843186378479004,
+ "learning_rate": 2.690553601887058e-06,
+ "loss": 0.0145,
+ "step": 28189
+ },
+ {
+ "epoch": 76.81198910081744,
+ "grad_norm": 1.4607402086257935,
+ "learning_rate": 2.6899513835694323e-06,
+ "loss": 0.0123,
+ "step": 28190
+ },
+ {
+ "epoch": 76.81471389645776,
+ "grad_norm": 1.3549250364303589,
+ "learning_rate": 2.6893492221822293e-06,
+ "loss": 0.0188,
+ "step": 28191
+ },
+ {
+ "epoch": 76.8174386920981,
+ "grad_norm": 1.5549042224884033,
+ "learning_rate": 2.688747117730136e-06,
+ "loss": 0.0347,
+ "step": 28192
+ },
+ {
+ "epoch": 76.82016348773843,
+ "grad_norm": 1.609731912612915,
+ "learning_rate": 2.688145070217846e-06,
+ "loss": 0.0126,
+ "step": 28193
+ },
+ {
+ "epoch": 76.82288828337875,
+ "grad_norm": 1.1972508430480957,
+ "learning_rate": 2.687543079650049e-06,
+ "loss": 0.0117,
+ "step": 28194
+ },
+ {
+ "epoch": 76.82561307901908,
+ "grad_norm": 1.4685089588165283,
+ "learning_rate": 2.686941146031431e-06,
+ "loss": 0.0391,
+ "step": 28195
+ },
+ {
+ "epoch": 76.8283378746594,
+ "grad_norm": 1.5102473497390747,
+ "learning_rate": 2.68633926936668e-06,
+ "loss": 0.0193,
+ "step": 28196
+ },
+ {
+ "epoch": 76.83106267029973,
+ "grad_norm": 0.7348581552505493,
+ "learning_rate": 2.6857374496604805e-06,
+ "loss": 0.0079,
+ "step": 28197
+ },
+ {
+ "epoch": 76.83378746594005,
+ "grad_norm": 2.6924781799316406,
+ "learning_rate": 2.685135686917526e-06,
+ "loss": 0.0481,
+ "step": 28198
+ },
+ {
+ "epoch": 76.83651226158038,
+ "grad_norm": 1.5106842517852783,
+ "learning_rate": 2.6845339811424987e-06,
+ "loss": 0.0691,
+ "step": 28199
+ },
+ {
+ "epoch": 76.83923705722071,
+ "grad_norm": 1.684155821800232,
+ "learning_rate": 2.6839323323400856e-06,
+ "loss": 0.0487,
+ "step": 28200
+ },
+ {
+ "epoch": 76.84196185286103,
+ "grad_norm": 1.195516586303711,
+ "learning_rate": 2.683330740514969e-06,
+ "loss": 0.0893,
+ "step": 28201
+ },
+ {
+ "epoch": 76.84468664850137,
+ "grad_norm": 1.3210761547088623,
+ "learning_rate": 2.682729205671839e-06,
+ "loss": 0.0138,
+ "step": 28202
+ },
+ {
+ "epoch": 76.84741144414168,
+ "grad_norm": 1.421964406967163,
+ "learning_rate": 2.68212772781538e-06,
+ "loss": 0.0284,
+ "step": 28203
+ },
+ {
+ "epoch": 76.85013623978202,
+ "grad_norm": 1.2663403749465942,
+ "learning_rate": 2.6815263069502706e-06,
+ "loss": 0.1044,
+ "step": 28204
+ },
+ {
+ "epoch": 76.85286103542235,
+ "grad_norm": 1.3462859392166138,
+ "learning_rate": 2.6809249430812025e-06,
+ "loss": 0.0476,
+ "step": 28205
+ },
+ {
+ "epoch": 76.85558583106267,
+ "grad_norm": 1.1011691093444824,
+ "learning_rate": 2.680323636212856e-06,
+ "loss": 0.0091,
+ "step": 28206
+ },
+ {
+ "epoch": 76.858310626703,
+ "grad_norm": 1.3363831043243408,
+ "learning_rate": 2.6797223863499123e-06,
+ "loss": 0.0097,
+ "step": 28207
+ },
+ {
+ "epoch": 76.86103542234332,
+ "grad_norm": 1.3249553442001343,
+ "learning_rate": 2.6791211934970516e-06,
+ "loss": 0.0455,
+ "step": 28208
+ },
+ {
+ "epoch": 76.86376021798365,
+ "grad_norm": 0.977685809135437,
+ "learning_rate": 2.678520057658961e-06,
+ "loss": 0.1375,
+ "step": 28209
+ },
+ {
+ "epoch": 76.86648501362397,
+ "grad_norm": 1.4834145307540894,
+ "learning_rate": 2.6779189788403226e-06,
+ "loss": 0.0395,
+ "step": 28210
+ },
+ {
+ "epoch": 76.8692098092643,
+ "grad_norm": 1.8827553987503052,
+ "learning_rate": 2.677317957045815e-06,
+ "loss": 0.0187,
+ "step": 28211
+ },
+ {
+ "epoch": 76.87193460490464,
+ "grad_norm": 1.180516004562378,
+ "learning_rate": 2.6767169922801205e-06,
+ "loss": 0.0126,
+ "step": 28212
+ },
+ {
+ "epoch": 76.87465940054496,
+ "grad_norm": 1.1578350067138672,
+ "learning_rate": 2.676116084547915e-06,
+ "loss": 0.0219,
+ "step": 28213
+ },
+ {
+ "epoch": 76.87738419618529,
+ "grad_norm": 1.1765459775924683,
+ "learning_rate": 2.6755152338538847e-06,
+ "loss": 0.0071,
+ "step": 28214
+ },
+ {
+ "epoch": 76.88010899182561,
+ "grad_norm": 0.9590155482292175,
+ "learning_rate": 2.674914440202706e-06,
+ "loss": 0.011,
+ "step": 28215
+ },
+ {
+ "epoch": 76.88283378746594,
+ "grad_norm": 0.8253056406974792,
+ "learning_rate": 2.6743137035990583e-06,
+ "loss": 0.0108,
+ "step": 28216
+ },
+ {
+ "epoch": 76.88555858310627,
+ "grad_norm": 1.5930607318878174,
+ "learning_rate": 2.673713024047616e-06,
+ "loss": 0.0237,
+ "step": 28217
+ },
+ {
+ "epoch": 76.88828337874659,
+ "grad_norm": 0.8235417604446411,
+ "learning_rate": 2.673112401553064e-06,
+ "loss": 0.0073,
+ "step": 28218
+ },
+ {
+ "epoch": 76.89100817438693,
+ "grad_norm": 1.695634126663208,
+ "learning_rate": 2.6725118361200775e-06,
+ "loss": 0.0529,
+ "step": 28219
+ },
+ {
+ "epoch": 76.89373297002724,
+ "grad_norm": 1.0900770425796509,
+ "learning_rate": 2.6719113277533328e-06,
+ "loss": 0.0079,
+ "step": 28220
+ },
+ {
+ "epoch": 76.89645776566758,
+ "grad_norm": 1.6831319332122803,
+ "learning_rate": 2.6713108764575035e-06,
+ "loss": 0.1438,
+ "step": 28221
+ },
+ {
+ "epoch": 76.8991825613079,
+ "grad_norm": 1.4197547435760498,
+ "learning_rate": 2.670710482237272e-06,
+ "loss": 0.0292,
+ "step": 28222
+ },
+ {
+ "epoch": 76.90190735694823,
+ "grad_norm": 1.7072805166244507,
+ "learning_rate": 2.670110145097311e-06,
+ "loss": 0.0172,
+ "step": 28223
+ },
+ {
+ "epoch": 76.90463215258856,
+ "grad_norm": 1.2194559574127197,
+ "learning_rate": 2.6695098650422967e-06,
+ "loss": 0.0139,
+ "step": 28224
+ },
+ {
+ "epoch": 76.90735694822888,
+ "grad_norm": 0.6437077522277832,
+ "learning_rate": 2.6689096420769013e-06,
+ "loss": 0.0067,
+ "step": 28225
+ },
+ {
+ "epoch": 76.91008174386921,
+ "grad_norm": 1.7153979539871216,
+ "learning_rate": 2.668309476205805e-06,
+ "loss": 0.0373,
+ "step": 28226
+ },
+ {
+ "epoch": 76.91280653950953,
+ "grad_norm": 1.7908681631088257,
+ "learning_rate": 2.667709367433674e-06,
+ "loss": 0.0688,
+ "step": 28227
+ },
+ {
+ "epoch": 76.91553133514986,
+ "grad_norm": 1.6188724040985107,
+ "learning_rate": 2.6671093157651908e-06,
+ "loss": 0.0218,
+ "step": 28228
+ },
+ {
+ "epoch": 76.9182561307902,
+ "grad_norm": 2.115924835205078,
+ "learning_rate": 2.6665093212050197e-06,
+ "loss": 0.0862,
+ "step": 28229
+ },
+ {
+ "epoch": 76.92098092643052,
+ "grad_norm": 1.8953993320465088,
+ "learning_rate": 2.665909383757842e-06,
+ "loss": 0.0302,
+ "step": 28230
+ },
+ {
+ "epoch": 76.92370572207085,
+ "grad_norm": 0.6524559855461121,
+ "learning_rate": 2.665309503428326e-06,
+ "loss": 0.0076,
+ "step": 28231
+ },
+ {
+ "epoch": 76.92643051771117,
+ "grad_norm": 1.0371900796890259,
+ "learning_rate": 2.664709680221144e-06,
+ "loss": 0.0121,
+ "step": 28232
+ },
+ {
+ "epoch": 76.9291553133515,
+ "grad_norm": 1.2330613136291504,
+ "learning_rate": 2.664109914140963e-06,
+ "loss": 0.0088,
+ "step": 28233
+ },
+ {
+ "epoch": 76.93188010899182,
+ "grad_norm": 1.6966668367385864,
+ "learning_rate": 2.663510205192461e-06,
+ "loss": 0.1169,
+ "step": 28234
+ },
+ {
+ "epoch": 76.93460490463215,
+ "grad_norm": 1.1086643934249878,
+ "learning_rate": 2.6629105533803055e-06,
+ "loss": 0.0094,
+ "step": 28235
+ },
+ {
+ "epoch": 76.93732970027249,
+ "grad_norm": 5.844547748565674,
+ "learning_rate": 2.6623109587091665e-06,
+ "loss": 0.0223,
+ "step": 28236
+ },
+ {
+ "epoch": 76.9400544959128,
+ "grad_norm": 0.70591801404953,
+ "learning_rate": 2.66171142118371e-06,
+ "loss": 0.0065,
+ "step": 28237
+ },
+ {
+ "epoch": 76.94277929155314,
+ "grad_norm": 1.2910691499710083,
+ "learning_rate": 2.6611119408086127e-06,
+ "loss": 0.0916,
+ "step": 28238
+ },
+ {
+ "epoch": 76.94550408719346,
+ "grad_norm": 1.0701959133148193,
+ "learning_rate": 2.6605125175885394e-06,
+ "loss": 0.0149,
+ "step": 28239
+ },
+ {
+ "epoch": 76.94822888283379,
+ "grad_norm": 1.2754325866699219,
+ "learning_rate": 2.6599131515281586e-06,
+ "loss": 0.0147,
+ "step": 28240
+ },
+ {
+ "epoch": 76.95095367847412,
+ "grad_norm": 1.568213701248169,
+ "learning_rate": 2.659313842632134e-06,
+ "loss": 0.0311,
+ "step": 28241
+ },
+ {
+ "epoch": 76.95367847411444,
+ "grad_norm": 0.6164776086807251,
+ "learning_rate": 2.6587145909051405e-06,
+ "loss": 0.0063,
+ "step": 28242
+ },
+ {
+ "epoch": 76.95640326975477,
+ "grad_norm": 1.545177698135376,
+ "learning_rate": 2.6581153963518413e-06,
+ "loss": 0.0142,
+ "step": 28243
+ },
+ {
+ "epoch": 76.95912806539509,
+ "grad_norm": 1.7088850736618042,
+ "learning_rate": 2.6575162589769e-06,
+ "loss": 0.021,
+ "step": 28244
+ },
+ {
+ "epoch": 76.96185286103542,
+ "grad_norm": 2.255427122116089,
+ "learning_rate": 2.6569171787849867e-06,
+ "loss": 0.0688,
+ "step": 28245
+ },
+ {
+ "epoch": 76.96457765667574,
+ "grad_norm": 1.0382461547851562,
+ "learning_rate": 2.656318155780768e-06,
+ "loss": 0.0116,
+ "step": 28246
+ },
+ {
+ "epoch": 76.96730245231608,
+ "grad_norm": 1.336068034172058,
+ "learning_rate": 2.6557191899689073e-06,
+ "loss": 0.0163,
+ "step": 28247
+ },
+ {
+ "epoch": 76.97002724795641,
+ "grad_norm": 0.9587085247039795,
+ "learning_rate": 2.6551202813540686e-06,
+ "loss": 0.0134,
+ "step": 28248
+ },
+ {
+ "epoch": 76.97275204359673,
+ "grad_norm": 1.6065973043441772,
+ "learning_rate": 2.654521429940915e-06,
+ "loss": 0.0194,
+ "step": 28249
+ },
+ {
+ "epoch": 76.97547683923706,
+ "grad_norm": 1.0228276252746582,
+ "learning_rate": 2.653922635734115e-06,
+ "loss": 0.0136,
+ "step": 28250
+ },
+ {
+ "epoch": 76.97820163487738,
+ "grad_norm": 0.9182363748550415,
+ "learning_rate": 2.6533238987383283e-06,
+ "loss": 0.0165,
+ "step": 28251
+ },
+ {
+ "epoch": 76.98092643051771,
+ "grad_norm": 1.835250973701477,
+ "learning_rate": 2.6527252189582185e-06,
+ "loss": 0.1572,
+ "step": 28252
+ },
+ {
+ "epoch": 76.98365122615803,
+ "grad_norm": 1.76443350315094,
+ "learning_rate": 2.6521265963984467e-06,
+ "loss": 0.1199,
+ "step": 28253
+ },
+ {
+ "epoch": 76.98637602179836,
+ "grad_norm": 1.4700560569763184,
+ "learning_rate": 2.6515280310636783e-06,
+ "loss": 0.1072,
+ "step": 28254
+ },
+ {
+ "epoch": 76.9891008174387,
+ "grad_norm": 1.3883213996887207,
+ "learning_rate": 2.650929522958573e-06,
+ "loss": 0.0363,
+ "step": 28255
+ },
+ {
+ "epoch": 76.99182561307902,
+ "grad_norm": 1.5448646545410156,
+ "learning_rate": 2.6503310720877928e-06,
+ "loss": 0.0334,
+ "step": 28256
+ },
+ {
+ "epoch": 76.99455040871935,
+ "grad_norm": 1.3592387437820435,
+ "learning_rate": 2.6497326784559942e-06,
+ "loss": 0.0363,
+ "step": 28257
+ },
+ {
+ "epoch": 76.99727520435967,
+ "grad_norm": 1.4166369438171387,
+ "learning_rate": 2.649134342067844e-06,
+ "loss": 0.0434,
+ "step": 28258
+ },
+ {
+ "epoch": 77.0,
+ "grad_norm": 0.995343029499054,
+ "learning_rate": 2.648536062927999e-06,
+ "loss": 0.0141,
+ "step": 28259
+ },
+ {
+ "epoch": 77.00272479564033,
+ "grad_norm": 1.3665308952331543,
+ "learning_rate": 2.6479378410411184e-06,
+ "loss": 0.0147,
+ "step": 28260
+ },
+ {
+ "epoch": 77.00544959128065,
+ "grad_norm": 1.3179700374603271,
+ "learning_rate": 2.6473396764118575e-06,
+ "loss": 0.0262,
+ "step": 28261
+ },
+ {
+ "epoch": 77.00817438692098,
+ "grad_norm": 1.6415045261383057,
+ "learning_rate": 2.6467415690448795e-06,
+ "loss": 0.0803,
+ "step": 28262
+ },
+ {
+ "epoch": 77.0108991825613,
+ "grad_norm": 1.2363967895507812,
+ "learning_rate": 2.646143518944845e-06,
+ "loss": 0.0179,
+ "step": 28263
+ },
+ {
+ "epoch": 77.01362397820164,
+ "grad_norm": 2.3792879581451416,
+ "learning_rate": 2.6455455261164066e-06,
+ "loss": 0.1448,
+ "step": 28264
+ },
+ {
+ "epoch": 77.01634877384195,
+ "grad_norm": 0.643627941608429,
+ "learning_rate": 2.6449475905642242e-06,
+ "loss": 0.0053,
+ "step": 28265
+ },
+ {
+ "epoch": 77.01907356948229,
+ "grad_norm": 1.036365270614624,
+ "learning_rate": 2.6443497122929495e-06,
+ "loss": 0.0146,
+ "step": 28266
+ },
+ {
+ "epoch": 77.02179836512262,
+ "grad_norm": 1.2809556722640991,
+ "learning_rate": 2.6437518913072456e-06,
+ "loss": 0.0099,
+ "step": 28267
+ },
+ {
+ "epoch": 77.02452316076294,
+ "grad_norm": 1.2885847091674805,
+ "learning_rate": 2.6431541276117654e-06,
+ "loss": 0.0259,
+ "step": 28268
+ },
+ {
+ "epoch": 77.02724795640327,
+ "grad_norm": 0.8792701959609985,
+ "learning_rate": 2.6425564212111608e-06,
+ "loss": 0.007,
+ "step": 28269
+ },
+ {
+ "epoch": 77.02997275204359,
+ "grad_norm": 1.1695506572723389,
+ "learning_rate": 2.6419587721100938e-06,
+ "loss": 0.022,
+ "step": 28270
+ },
+ {
+ "epoch": 77.03269754768392,
+ "grad_norm": 1.626538872718811,
+ "learning_rate": 2.6413611803132146e-06,
+ "loss": 0.0344,
+ "step": 28271
+ },
+ {
+ "epoch": 77.03542234332426,
+ "grad_norm": 1.0024203062057495,
+ "learning_rate": 2.6407636458251773e-06,
+ "loss": 0.0099,
+ "step": 28272
+ },
+ {
+ "epoch": 77.03814713896458,
+ "grad_norm": 1.257720947265625,
+ "learning_rate": 2.640166168650633e-06,
+ "loss": 0.1317,
+ "step": 28273
+ },
+ {
+ "epoch": 77.04087193460491,
+ "grad_norm": 1.0053826570510864,
+ "learning_rate": 2.639568748794241e-06,
+ "loss": 0.0125,
+ "step": 28274
+ },
+ {
+ "epoch": 77.04359673024523,
+ "grad_norm": 1.0327574014663696,
+ "learning_rate": 2.6389713862606513e-06,
+ "loss": 0.006,
+ "step": 28275
+ },
+ {
+ "epoch": 77.04632152588556,
+ "grad_norm": 1.1551077365875244,
+ "learning_rate": 2.6383740810545145e-06,
+ "loss": 0.0132,
+ "step": 28276
+ },
+ {
+ "epoch": 77.04904632152588,
+ "grad_norm": 1.7814418077468872,
+ "learning_rate": 2.637776833180481e-06,
+ "loss": 0.0636,
+ "step": 28277
+ },
+ {
+ "epoch": 77.05177111716621,
+ "grad_norm": 1.1868869066238403,
+ "learning_rate": 2.6371796426432038e-06,
+ "loss": 0.0168,
+ "step": 28278
+ },
+ {
+ "epoch": 77.05449591280654,
+ "grad_norm": 1.154863715171814,
+ "learning_rate": 2.6365825094473387e-06,
+ "loss": 0.0329,
+ "step": 28279
+ },
+ {
+ "epoch": 77.05722070844686,
+ "grad_norm": 0.8842360973358154,
+ "learning_rate": 2.635985433597531e-06,
+ "loss": 0.0153,
+ "step": 28280
+ },
+ {
+ "epoch": 77.0599455040872,
+ "grad_norm": 1.8870865106582642,
+ "learning_rate": 2.6353884150984322e-06,
+ "loss": 0.0092,
+ "step": 28281
+ },
+ {
+ "epoch": 77.06267029972751,
+ "grad_norm": 1.3910572528839111,
+ "learning_rate": 2.6347914539546883e-06,
+ "loss": 0.0293,
+ "step": 28282
+ },
+ {
+ "epoch": 77.06539509536785,
+ "grad_norm": 1.136670470237732,
+ "learning_rate": 2.634194550170954e-06,
+ "loss": 0.0112,
+ "step": 28283
+ },
+ {
+ "epoch": 77.06811989100818,
+ "grad_norm": 1.324263572692871,
+ "learning_rate": 2.6335977037518767e-06,
+ "loss": 0.0166,
+ "step": 28284
+ },
+ {
+ "epoch": 77.0708446866485,
+ "grad_norm": 1.7185922861099243,
+ "learning_rate": 2.633000914702103e-06,
+ "loss": 0.0166,
+ "step": 28285
+ },
+ {
+ "epoch": 77.07356948228883,
+ "grad_norm": 1.7691971063613892,
+ "learning_rate": 2.632404183026277e-06,
+ "loss": 0.021,
+ "step": 28286
+ },
+ {
+ "epoch": 77.07629427792915,
+ "grad_norm": 1.0304774045944214,
+ "learning_rate": 2.631807508729054e-06,
+ "loss": 0.0208,
+ "step": 28287
+ },
+ {
+ "epoch": 77.07901907356948,
+ "grad_norm": 2.0733180046081543,
+ "learning_rate": 2.631210891815077e-06,
+ "loss": 0.0781,
+ "step": 28288
+ },
+ {
+ "epoch": 77.0817438692098,
+ "grad_norm": 0.941791832447052,
+ "learning_rate": 2.63061433228899e-06,
+ "loss": 0.0105,
+ "step": 28289
+ },
+ {
+ "epoch": 77.08446866485014,
+ "grad_norm": 1.9343498945236206,
+ "learning_rate": 2.6300178301554436e-06,
+ "loss": 0.0147,
+ "step": 28290
+ },
+ {
+ "epoch": 77.08719346049047,
+ "grad_norm": 1.5021721124649048,
+ "learning_rate": 2.6294213854190807e-06,
+ "loss": 0.0221,
+ "step": 28291
+ },
+ {
+ "epoch": 77.08991825613079,
+ "grad_norm": 1.397935152053833,
+ "learning_rate": 2.628824998084547e-06,
+ "loss": 0.0197,
+ "step": 28292
+ },
+ {
+ "epoch": 77.09264305177112,
+ "grad_norm": 1.6435590982437134,
+ "learning_rate": 2.628228668156485e-06,
+ "loss": 0.0383,
+ "step": 28293
+ },
+ {
+ "epoch": 77.09536784741144,
+ "grad_norm": 1.0033490657806396,
+ "learning_rate": 2.6276323956395433e-06,
+ "loss": 0.0094,
+ "step": 28294
+ },
+ {
+ "epoch": 77.09809264305177,
+ "grad_norm": 0.8655833601951599,
+ "learning_rate": 2.6270361805383593e-06,
+ "loss": 0.0136,
+ "step": 28295
+ },
+ {
+ "epoch": 77.1008174386921,
+ "grad_norm": 1.1399705410003662,
+ "learning_rate": 2.626440022857585e-06,
+ "loss": 0.0449,
+ "step": 28296
+ },
+ {
+ "epoch": 77.10354223433242,
+ "grad_norm": 1.2097729444503784,
+ "learning_rate": 2.6258439226018574e-06,
+ "loss": 0.0153,
+ "step": 28297
+ },
+ {
+ "epoch": 77.10626702997276,
+ "grad_norm": 1.8401597738265991,
+ "learning_rate": 2.625247879775816e-06,
+ "loss": 0.0331,
+ "step": 28298
+ },
+ {
+ "epoch": 77.10899182561307,
+ "grad_norm": 0.8291199803352356,
+ "learning_rate": 2.6246518943841115e-06,
+ "loss": 0.0074,
+ "step": 28299
+ },
+ {
+ "epoch": 77.11171662125341,
+ "grad_norm": 1.9757962226867676,
+ "learning_rate": 2.6240559664313805e-06,
+ "loss": 0.0762,
+ "step": 28300
+ },
+ {
+ "epoch": 77.11444141689373,
+ "grad_norm": 0.8132350444793701,
+ "learning_rate": 2.623460095922263e-06,
+ "loss": 0.016,
+ "step": 28301
+ },
+ {
+ "epoch": 77.11716621253406,
+ "grad_norm": 1.035691499710083,
+ "learning_rate": 2.622864282861399e-06,
+ "loss": 0.01,
+ "step": 28302
+ },
+ {
+ "epoch": 77.11989100817439,
+ "grad_norm": 1.2482669353485107,
+ "learning_rate": 2.6222685272534333e-06,
+ "loss": 0.0282,
+ "step": 28303
+ },
+ {
+ "epoch": 77.12261580381471,
+ "grad_norm": 1.823455572128296,
+ "learning_rate": 2.6216728291030024e-06,
+ "loss": 0.1156,
+ "step": 28304
+ },
+ {
+ "epoch": 77.12534059945504,
+ "grad_norm": 0.9157131314277649,
+ "learning_rate": 2.6210771884147468e-06,
+ "loss": 0.0106,
+ "step": 28305
+ },
+ {
+ "epoch": 77.12806539509536,
+ "grad_norm": 0.8032504320144653,
+ "learning_rate": 2.620481605193301e-06,
+ "loss": 0.0088,
+ "step": 28306
+ },
+ {
+ "epoch": 77.1307901907357,
+ "grad_norm": 36.295623779296875,
+ "learning_rate": 2.61988607944331e-06,
+ "loss": 0.0552,
+ "step": 28307
+ },
+ {
+ "epoch": 77.13351498637603,
+ "grad_norm": 1.0597295761108398,
+ "learning_rate": 2.6192906111694095e-06,
+ "loss": 0.0099,
+ "step": 28308
+ },
+ {
+ "epoch": 77.13623978201635,
+ "grad_norm": 2.885143995285034,
+ "learning_rate": 2.6186952003762355e-06,
+ "loss": 0.0108,
+ "step": 28309
+ },
+ {
+ "epoch": 77.13896457765668,
+ "grad_norm": 1.4677215814590454,
+ "learning_rate": 2.618099847068423e-06,
+ "loss": 0.032,
+ "step": 28310
+ },
+ {
+ "epoch": 77.141689373297,
+ "grad_norm": 1.145763635635376,
+ "learning_rate": 2.6175045512506138e-06,
+ "loss": 0.0106,
+ "step": 28311
+ },
+ {
+ "epoch": 77.14441416893733,
+ "grad_norm": 0.3928843140602112,
+ "learning_rate": 2.6169093129274416e-06,
+ "loss": 0.0054,
+ "step": 28312
+ },
+ {
+ "epoch": 77.14713896457765,
+ "grad_norm": 0.8904657959938049,
+ "learning_rate": 2.616314132103539e-06,
+ "loss": 0.0135,
+ "step": 28313
+ },
+ {
+ "epoch": 77.14986376021798,
+ "grad_norm": 1.3779032230377197,
+ "learning_rate": 2.6157190087835437e-06,
+ "loss": 0.0117,
+ "step": 28314
+ },
+ {
+ "epoch": 77.15258855585832,
+ "grad_norm": 0.8864033818244934,
+ "learning_rate": 2.6151239429720952e-06,
+ "loss": 0.0122,
+ "step": 28315
+ },
+ {
+ "epoch": 77.15531335149863,
+ "grad_norm": 1.4849931001663208,
+ "learning_rate": 2.6145289346738234e-06,
+ "loss": 0.0187,
+ "step": 28316
+ },
+ {
+ "epoch": 77.15803814713897,
+ "grad_norm": 1.3596138954162598,
+ "learning_rate": 2.613933983893363e-06,
+ "loss": 0.0153,
+ "step": 28317
+ },
+ {
+ "epoch": 77.16076294277929,
+ "grad_norm": 1.3404934406280518,
+ "learning_rate": 2.613339090635343e-06,
+ "loss": 0.0134,
+ "step": 28318
+ },
+ {
+ "epoch": 77.16348773841962,
+ "grad_norm": 1.784064769744873,
+ "learning_rate": 2.612744254904405e-06,
+ "loss": 0.0503,
+ "step": 28319
+ },
+ {
+ "epoch": 77.16621253405995,
+ "grad_norm": 1.0123845338821411,
+ "learning_rate": 2.612149476705176e-06,
+ "loss": 0.0137,
+ "step": 28320
+ },
+ {
+ "epoch": 77.16893732970027,
+ "grad_norm": 1.0028377771377563,
+ "learning_rate": 2.6115547560422884e-06,
+ "loss": 0.0123,
+ "step": 28321
+ },
+ {
+ "epoch": 77.1716621253406,
+ "grad_norm": 1.2194416522979736,
+ "learning_rate": 2.6109600929203726e-06,
+ "loss": 0.0157,
+ "step": 28322
+ },
+ {
+ "epoch": 77.17438692098092,
+ "grad_norm": 1.759289026260376,
+ "learning_rate": 2.6103654873440642e-06,
+ "loss": 0.0645,
+ "step": 28323
+ },
+ {
+ "epoch": 77.17711171662125,
+ "grad_norm": 1.1455191373825073,
+ "learning_rate": 2.6097709393179915e-06,
+ "loss": 0.0827,
+ "step": 28324
+ },
+ {
+ "epoch": 77.17983651226157,
+ "grad_norm": 0.5860495567321777,
+ "learning_rate": 2.6091764488467852e-06,
+ "loss": 0.0065,
+ "step": 28325
+ },
+ {
+ "epoch": 77.1825613079019,
+ "grad_norm": 1.310246229171753,
+ "learning_rate": 2.608582015935072e-06,
+ "loss": 0.0218,
+ "step": 28326
+ },
+ {
+ "epoch": 77.18528610354224,
+ "grad_norm": 1.2631738185882568,
+ "learning_rate": 2.607987640587487e-06,
+ "loss": 0.0152,
+ "step": 28327
+ },
+ {
+ "epoch": 77.18801089918256,
+ "grad_norm": 0.8456602096557617,
+ "learning_rate": 2.6073933228086568e-06,
+ "loss": 0.0472,
+ "step": 28328
+ },
+ {
+ "epoch": 77.19073569482289,
+ "grad_norm": 0.7935953736305237,
+ "learning_rate": 2.606799062603208e-06,
+ "loss": 0.0102,
+ "step": 28329
+ },
+ {
+ "epoch": 77.19346049046321,
+ "grad_norm": 1.4803807735443115,
+ "learning_rate": 2.6062048599757685e-06,
+ "loss": 0.0239,
+ "step": 28330
+ },
+ {
+ "epoch": 77.19618528610354,
+ "grad_norm": 1.641829013824463,
+ "learning_rate": 2.6056107149309673e-06,
+ "loss": 0.0463,
+ "step": 28331
+ },
+ {
+ "epoch": 77.19891008174388,
+ "grad_norm": 1.7584834098815918,
+ "learning_rate": 2.6050166274734346e-06,
+ "loss": 0.0774,
+ "step": 28332
+ },
+ {
+ "epoch": 77.2016348773842,
+ "grad_norm": 2.027272939682007,
+ "learning_rate": 2.6044225976077954e-06,
+ "loss": 0.0141,
+ "step": 28333
+ },
+ {
+ "epoch": 77.20435967302453,
+ "grad_norm": 1.3498399257659912,
+ "learning_rate": 2.6038286253386704e-06,
+ "loss": 0.0172,
+ "step": 28334
+ },
+ {
+ "epoch": 77.20708446866485,
+ "grad_norm": 1.6989136934280396,
+ "learning_rate": 2.6032347106706945e-06,
+ "loss": 0.0192,
+ "step": 28335
+ },
+ {
+ "epoch": 77.20980926430518,
+ "grad_norm": 1.0467491149902344,
+ "learning_rate": 2.602640853608488e-06,
+ "loss": 0.0113,
+ "step": 28336
+ },
+ {
+ "epoch": 77.2125340599455,
+ "grad_norm": 1.9834702014923096,
+ "learning_rate": 2.6020470541566757e-06,
+ "loss": 0.0803,
+ "step": 28337
+ },
+ {
+ "epoch": 77.21525885558583,
+ "grad_norm": 1.1559035778045654,
+ "learning_rate": 2.6014533123198803e-06,
+ "loss": 0.0277,
+ "step": 28338
+ },
+ {
+ "epoch": 77.21798365122616,
+ "grad_norm": 1.2252689599990845,
+ "learning_rate": 2.600859628102732e-06,
+ "loss": 0.0175,
+ "step": 28339
+ },
+ {
+ "epoch": 77.22070844686648,
+ "grad_norm": 1.341543197631836,
+ "learning_rate": 2.60026600150985e-06,
+ "loss": 0.0215,
+ "step": 28340
+ },
+ {
+ "epoch": 77.22343324250681,
+ "grad_norm": 1.191077709197998,
+ "learning_rate": 2.5996724325458588e-06,
+ "loss": 0.0207,
+ "step": 28341
+ },
+ {
+ "epoch": 77.22615803814713,
+ "grad_norm": 1.1904683113098145,
+ "learning_rate": 2.599078921215378e-06,
+ "loss": 0.0197,
+ "step": 28342
+ },
+ {
+ "epoch": 77.22888283378747,
+ "grad_norm": 1.7583578824996948,
+ "learning_rate": 2.598485467523034e-06,
+ "loss": 0.0415,
+ "step": 28343
+ },
+ {
+ "epoch": 77.2316076294278,
+ "grad_norm": 1.2809875011444092,
+ "learning_rate": 2.597892071473448e-06,
+ "loss": 0.0372,
+ "step": 28344
+ },
+ {
+ "epoch": 77.23433242506812,
+ "grad_norm": 6.8636088371276855,
+ "learning_rate": 2.5972987330712398e-06,
+ "loss": 0.017,
+ "step": 28345
+ },
+ {
+ "epoch": 77.23705722070845,
+ "grad_norm": 1.9313197135925293,
+ "learning_rate": 2.596705452321028e-06,
+ "loss": 0.0236,
+ "step": 28346
+ },
+ {
+ "epoch": 77.23978201634877,
+ "grad_norm": 0.7329443693161011,
+ "learning_rate": 2.5961122292274353e-06,
+ "loss": 0.0134,
+ "step": 28347
+ },
+ {
+ "epoch": 77.2425068119891,
+ "grad_norm": 1.9830704927444458,
+ "learning_rate": 2.5955190637950855e-06,
+ "loss": 0.0481,
+ "step": 28348
+ },
+ {
+ "epoch": 77.24523160762942,
+ "grad_norm": 1.5107353925704956,
+ "learning_rate": 2.5949259560285955e-06,
+ "loss": 0.0367,
+ "step": 28349
+ },
+ {
+ "epoch": 77.24795640326975,
+ "grad_norm": 1.4063695669174194,
+ "learning_rate": 2.594332905932584e-06,
+ "loss": 0.0576,
+ "step": 28350
+ },
+ {
+ "epoch": 77.25068119891009,
+ "grad_norm": 1.0135422945022583,
+ "learning_rate": 2.593739913511666e-06,
+ "loss": 0.0139,
+ "step": 28351
+ },
+ {
+ "epoch": 77.2534059945504,
+ "grad_norm": 1.2662116289138794,
+ "learning_rate": 2.5931469787704656e-06,
+ "loss": 0.0114,
+ "step": 28352
+ },
+ {
+ "epoch": 77.25613079019074,
+ "grad_norm": 0.9588803052902222,
+ "learning_rate": 2.5925541017135992e-06,
+ "loss": 0.0129,
+ "step": 28353
+ },
+ {
+ "epoch": 77.25885558583106,
+ "grad_norm": 1.0456790924072266,
+ "learning_rate": 2.5919612823456795e-06,
+ "loss": 0.0081,
+ "step": 28354
+ },
+ {
+ "epoch": 77.26158038147139,
+ "grad_norm": 2.3028934001922607,
+ "learning_rate": 2.59136852067133e-06,
+ "loss": 0.1122,
+ "step": 28355
+ },
+ {
+ "epoch": 77.26430517711172,
+ "grad_norm": 0.5589127540588379,
+ "learning_rate": 2.5907758166951635e-06,
+ "loss": 0.0058,
+ "step": 28356
+ },
+ {
+ "epoch": 77.26702997275204,
+ "grad_norm": 0.701241672039032,
+ "learning_rate": 2.5901831704217973e-06,
+ "loss": 0.0064,
+ "step": 28357
+ },
+ {
+ "epoch": 77.26975476839237,
+ "grad_norm": 1.4254391193389893,
+ "learning_rate": 2.589590581855843e-06,
+ "loss": 0.0677,
+ "step": 28358
+ },
+ {
+ "epoch": 77.2724795640327,
+ "grad_norm": 2.1804823875427246,
+ "learning_rate": 2.588998051001921e-06,
+ "loss": 0.0131,
+ "step": 28359
+ },
+ {
+ "epoch": 77.27520435967303,
+ "grad_norm": 1.1599946022033691,
+ "learning_rate": 2.5884055778646443e-06,
+ "loss": 0.0372,
+ "step": 28360
+ },
+ {
+ "epoch": 77.27792915531334,
+ "grad_norm": 0.9603344202041626,
+ "learning_rate": 2.5878131624486256e-06,
+ "loss": 0.0114,
+ "step": 28361
+ },
+ {
+ "epoch": 77.28065395095368,
+ "grad_norm": 0.5599574446678162,
+ "learning_rate": 2.587220804758478e-06,
+ "loss": 0.0063,
+ "step": 28362
+ },
+ {
+ "epoch": 77.28337874659401,
+ "grad_norm": 1.6303467750549316,
+ "learning_rate": 2.586628504798818e-06,
+ "loss": 0.0163,
+ "step": 28363
+ },
+ {
+ "epoch": 77.28610354223433,
+ "grad_norm": 2.0026779174804688,
+ "learning_rate": 2.586036262574254e-06,
+ "loss": 0.0287,
+ "step": 28364
+ },
+ {
+ "epoch": 77.28882833787466,
+ "grad_norm": 2.1538403034210205,
+ "learning_rate": 2.585444078089404e-06,
+ "loss": 0.063,
+ "step": 28365
+ },
+ {
+ "epoch": 77.29155313351498,
+ "grad_norm": 2.951711893081665,
+ "learning_rate": 2.5848519513488767e-06,
+ "loss": 0.0824,
+ "step": 28366
+ },
+ {
+ "epoch": 77.29427792915531,
+ "grad_norm": 31.554025650024414,
+ "learning_rate": 2.5842598823572805e-06,
+ "loss": 0.1215,
+ "step": 28367
+ },
+ {
+ "epoch": 77.29700272479565,
+ "grad_norm": 1.321960687637329,
+ "learning_rate": 2.583667871119233e-06,
+ "loss": 0.0371,
+ "step": 28368
+ },
+ {
+ "epoch": 77.29972752043597,
+ "grad_norm": 1.1168304681777954,
+ "learning_rate": 2.583075917639342e-06,
+ "loss": 0.0276,
+ "step": 28369
+ },
+ {
+ "epoch": 77.3024523160763,
+ "grad_norm": 1.172541618347168,
+ "learning_rate": 2.582484021922216e-06,
+ "loss": 0.0168,
+ "step": 28370
+ },
+ {
+ "epoch": 77.30517711171662,
+ "grad_norm": 1.404685139656067,
+ "learning_rate": 2.581892183972463e-06,
+ "loss": 0.0165,
+ "step": 28371
+ },
+ {
+ "epoch": 77.30790190735695,
+ "grad_norm": 1.0314340591430664,
+ "learning_rate": 2.5813004037946986e-06,
+ "loss": 0.0124,
+ "step": 28372
+ },
+ {
+ "epoch": 77.31062670299727,
+ "grad_norm": 1.2930805683135986,
+ "learning_rate": 2.5807086813935266e-06,
+ "loss": 0.0092,
+ "step": 28373
+ },
+ {
+ "epoch": 77.3133514986376,
+ "grad_norm": 1.0264593362808228,
+ "learning_rate": 2.580117016773558e-06,
+ "loss": 0.0155,
+ "step": 28374
+ },
+ {
+ "epoch": 77.31607629427793,
+ "grad_norm": 1.6720755100250244,
+ "learning_rate": 2.5795254099393953e-06,
+ "loss": 0.1349,
+ "step": 28375
+ },
+ {
+ "epoch": 77.31880108991825,
+ "grad_norm": 1.7847172021865845,
+ "learning_rate": 2.5789338608956526e-06,
+ "loss": 0.0266,
+ "step": 28376
+ },
+ {
+ "epoch": 77.32152588555859,
+ "grad_norm": 1.1309348344802856,
+ "learning_rate": 2.5783423696469356e-06,
+ "loss": 0.0475,
+ "step": 28377
+ },
+ {
+ "epoch": 77.3242506811989,
+ "grad_norm": 0.7719748020172119,
+ "learning_rate": 2.577750936197845e-06,
+ "loss": 0.0079,
+ "step": 28378
+ },
+ {
+ "epoch": 77.32697547683924,
+ "grad_norm": 1.7392284870147705,
+ "learning_rate": 2.577159560552995e-06,
+ "loss": 0.1114,
+ "step": 28379
+ },
+ {
+ "epoch": 77.32970027247957,
+ "grad_norm": 1.3602070808410645,
+ "learning_rate": 2.576568242716987e-06,
+ "loss": 0.0603,
+ "step": 28380
+ },
+ {
+ "epoch": 77.33242506811989,
+ "grad_norm": 0.8588678240776062,
+ "learning_rate": 2.5759769826944236e-06,
+ "loss": 0.0075,
+ "step": 28381
+ },
+ {
+ "epoch": 77.33514986376022,
+ "grad_norm": 1.03635573387146,
+ "learning_rate": 2.5753857804899163e-06,
+ "loss": 0.018,
+ "step": 28382
+ },
+ {
+ "epoch": 77.33787465940054,
+ "grad_norm": 0.6924264430999756,
+ "learning_rate": 2.574794636108061e-06,
+ "loss": 0.0083,
+ "step": 28383
+ },
+ {
+ "epoch": 77.34059945504087,
+ "grad_norm": 1.5028642416000366,
+ "learning_rate": 2.574203549553471e-06,
+ "loss": 0.0749,
+ "step": 28384
+ },
+ {
+ "epoch": 77.34332425068119,
+ "grad_norm": 1.6814162731170654,
+ "learning_rate": 2.573612520830744e-06,
+ "loss": 0.0223,
+ "step": 28385
+ },
+ {
+ "epoch": 77.34604904632153,
+ "grad_norm": 1.673851728439331,
+ "learning_rate": 2.573021549944483e-06,
+ "loss": 0.0328,
+ "step": 28386
+ },
+ {
+ "epoch": 77.34877384196186,
+ "grad_norm": 1.2040696144104004,
+ "learning_rate": 2.5724306368992893e-06,
+ "loss": 0.0116,
+ "step": 28387
+ },
+ {
+ "epoch": 77.35149863760218,
+ "grad_norm": 1.499186635017395,
+ "learning_rate": 2.5718397816997685e-06,
+ "loss": 0.0119,
+ "step": 28388
+ },
+ {
+ "epoch": 77.35422343324251,
+ "grad_norm": 1.0670570135116577,
+ "learning_rate": 2.5712489843505207e-06,
+ "loss": 0.0145,
+ "step": 28389
+ },
+ {
+ "epoch": 77.35694822888283,
+ "grad_norm": 1.299221396446228,
+ "learning_rate": 2.570658244856147e-06,
+ "loss": 0.0617,
+ "step": 28390
+ },
+ {
+ "epoch": 77.35967302452316,
+ "grad_norm": 1.5552022457122803,
+ "learning_rate": 2.5700675632212445e-06,
+ "loss": 0.0122,
+ "step": 28391
+ },
+ {
+ "epoch": 77.3623978201635,
+ "grad_norm": 1.0851459503173828,
+ "learning_rate": 2.5694769394504204e-06,
+ "loss": 0.0101,
+ "step": 28392
+ },
+ {
+ "epoch": 77.36512261580381,
+ "grad_norm": 1.0115090608596802,
+ "learning_rate": 2.5688863735482706e-06,
+ "loss": 0.0094,
+ "step": 28393
+ },
+ {
+ "epoch": 77.36784741144415,
+ "grad_norm": 1.6111884117126465,
+ "learning_rate": 2.5682958655193936e-06,
+ "loss": 0.0255,
+ "step": 28394
+ },
+ {
+ "epoch": 77.37057220708446,
+ "grad_norm": 1.0086733102798462,
+ "learning_rate": 2.5677054153683868e-06,
+ "loss": 0.0097,
+ "step": 28395
+ },
+ {
+ "epoch": 77.3732970027248,
+ "grad_norm": 1.1015410423278809,
+ "learning_rate": 2.567115023099854e-06,
+ "loss": 0.009,
+ "step": 28396
+ },
+ {
+ "epoch": 77.37602179836512,
+ "grad_norm": 1.7188280820846558,
+ "learning_rate": 2.5665246887183903e-06,
+ "loss": 0.0166,
+ "step": 28397
+ },
+ {
+ "epoch": 77.37874659400545,
+ "grad_norm": 0.748724102973938,
+ "learning_rate": 2.5659344122285902e-06,
+ "loss": 0.0096,
+ "step": 28398
+ },
+ {
+ "epoch": 77.38147138964578,
+ "grad_norm": 1.7762447595596313,
+ "learning_rate": 2.565344193635053e-06,
+ "loss": 0.0182,
+ "step": 28399
+ },
+ {
+ "epoch": 77.3841961852861,
+ "grad_norm": 0.9827097058296204,
+ "learning_rate": 2.56475403294238e-06,
+ "loss": 0.0345,
+ "step": 28400
+ },
+ {
+ "epoch": 77.38692098092643,
+ "grad_norm": 2.311241626739502,
+ "learning_rate": 2.5641639301551622e-06,
+ "loss": 0.0123,
+ "step": 28401
+ },
+ {
+ "epoch": 77.38964577656675,
+ "grad_norm": 1.2013903856277466,
+ "learning_rate": 2.563573885277998e-06,
+ "loss": 0.0116,
+ "step": 28402
+ },
+ {
+ "epoch": 77.39237057220708,
+ "grad_norm": 1.110144019126892,
+ "learning_rate": 2.5629838983154763e-06,
+ "loss": 0.0223,
+ "step": 28403
+ },
+ {
+ "epoch": 77.39509536784742,
+ "grad_norm": 0.6681897640228271,
+ "learning_rate": 2.5623939692722e-06,
+ "loss": 0.0096,
+ "step": 28404
+ },
+ {
+ "epoch": 77.39782016348774,
+ "grad_norm": 1.3152425289154053,
+ "learning_rate": 2.5618040981527603e-06,
+ "loss": 0.0211,
+ "step": 28405
+ },
+ {
+ "epoch": 77.40054495912807,
+ "grad_norm": 1.2244751453399658,
+ "learning_rate": 2.5612142849617516e-06,
+ "loss": 0.1501,
+ "step": 28406
+ },
+ {
+ "epoch": 77.40326975476839,
+ "grad_norm": 0.5591700077056885,
+ "learning_rate": 2.5606245297037624e-06,
+ "loss": 0.0051,
+ "step": 28407
+ },
+ {
+ "epoch": 77.40599455040872,
+ "grad_norm": 1.452684998512268,
+ "learning_rate": 2.5600348323833936e-06,
+ "loss": 0.0427,
+ "step": 28408
+ },
+ {
+ "epoch": 77.40871934604904,
+ "grad_norm": 1.119106411933899,
+ "learning_rate": 2.5594451930052346e-06,
+ "loss": 0.0987,
+ "step": 28409
+ },
+ {
+ "epoch": 77.41144414168937,
+ "grad_norm": 1.5575311183929443,
+ "learning_rate": 2.558855611573876e-06,
+ "loss": 0.0254,
+ "step": 28410
+ },
+ {
+ "epoch": 77.4141689373297,
+ "grad_norm": 1.3647706508636475,
+ "learning_rate": 2.558266088093907e-06,
+ "loss": 0.0181,
+ "step": 28411
+ },
+ {
+ "epoch": 77.41689373297002,
+ "grad_norm": 1.1064517498016357,
+ "learning_rate": 2.5576766225699267e-06,
+ "loss": 0.0108,
+ "step": 28412
+ },
+ {
+ "epoch": 77.41961852861036,
+ "grad_norm": 1.303978681564331,
+ "learning_rate": 2.557087215006521e-06,
+ "loss": 0.0123,
+ "step": 28413
+ },
+ {
+ "epoch": 77.42234332425068,
+ "grad_norm": 1.394464373588562,
+ "learning_rate": 2.5564978654082794e-06,
+ "loss": 0.0202,
+ "step": 28414
+ },
+ {
+ "epoch": 77.42506811989101,
+ "grad_norm": 1.2279926538467407,
+ "learning_rate": 2.5559085737797904e-06,
+ "loss": 0.03,
+ "step": 28415
+ },
+ {
+ "epoch": 77.42779291553134,
+ "grad_norm": 1.2653650045394897,
+ "learning_rate": 2.555319340125646e-06,
+ "loss": 0.0399,
+ "step": 28416
+ },
+ {
+ "epoch": 77.43051771117166,
+ "grad_norm": 0.9541388750076294,
+ "learning_rate": 2.5547301644504386e-06,
+ "loss": 0.0431,
+ "step": 28417
+ },
+ {
+ "epoch": 77.433242506812,
+ "grad_norm": 1.269261121749878,
+ "learning_rate": 2.5541410467587512e-06,
+ "loss": 0.0095,
+ "step": 28418
+ },
+ {
+ "epoch": 77.43596730245231,
+ "grad_norm": 1.6504299640655518,
+ "learning_rate": 2.5535519870551728e-06,
+ "loss": 0.0176,
+ "step": 28419
+ },
+ {
+ "epoch": 77.43869209809264,
+ "grad_norm": 1.247025489807129,
+ "learning_rate": 2.5529629853442937e-06,
+ "loss": 0.0751,
+ "step": 28420
+ },
+ {
+ "epoch": 77.44141689373296,
+ "grad_norm": 1.1968302726745605,
+ "learning_rate": 2.5523740416306995e-06,
+ "loss": 0.0159,
+ "step": 28421
+ },
+ {
+ "epoch": 77.4441416893733,
+ "grad_norm": 0.9239555597305298,
+ "learning_rate": 2.5517851559189764e-06,
+ "loss": 0.0203,
+ "step": 28422
+ },
+ {
+ "epoch": 77.44686648501363,
+ "grad_norm": 1.3938989639282227,
+ "learning_rate": 2.5511963282137077e-06,
+ "loss": 0.0149,
+ "step": 28423
+ },
+ {
+ "epoch": 77.44959128065395,
+ "grad_norm": 1.6789318323135376,
+ "learning_rate": 2.5506075585194856e-06,
+ "loss": 0.0164,
+ "step": 28424
+ },
+ {
+ "epoch": 77.45231607629428,
+ "grad_norm": 1.7233327627182007,
+ "learning_rate": 2.5500188468408917e-06,
+ "loss": 0.0244,
+ "step": 28425
+ },
+ {
+ "epoch": 77.4550408719346,
+ "grad_norm": 1.3254518508911133,
+ "learning_rate": 2.5494301931825117e-06,
+ "loss": 0.0304,
+ "step": 28426
+ },
+ {
+ "epoch": 77.45776566757493,
+ "grad_norm": 0.731105387210846,
+ "learning_rate": 2.5488415975489266e-06,
+ "loss": 0.0453,
+ "step": 28427
+ },
+ {
+ "epoch": 77.46049046321527,
+ "grad_norm": 0.6347128748893738,
+ "learning_rate": 2.5482530599447254e-06,
+ "loss": 0.0066,
+ "step": 28428
+ },
+ {
+ "epoch": 77.46321525885558,
+ "grad_norm": 1.1899099349975586,
+ "learning_rate": 2.5476645803744902e-06,
+ "loss": 0.0245,
+ "step": 28429
+ },
+ {
+ "epoch": 77.46594005449592,
+ "grad_norm": 2.080303907394409,
+ "learning_rate": 2.547076158842804e-06,
+ "loss": 0.017,
+ "step": 28430
+ },
+ {
+ "epoch": 77.46866485013624,
+ "grad_norm": 1.5735183954238892,
+ "learning_rate": 2.5464877953542457e-06,
+ "loss": 0.0214,
+ "step": 28431
+ },
+ {
+ "epoch": 77.47138964577657,
+ "grad_norm": 1.0655101537704468,
+ "learning_rate": 2.5458994899134024e-06,
+ "loss": 0.0107,
+ "step": 28432
+ },
+ {
+ "epoch": 77.47411444141689,
+ "grad_norm": 1.4285541772842407,
+ "learning_rate": 2.5453112425248516e-06,
+ "loss": 0.0211,
+ "step": 28433
+ },
+ {
+ "epoch": 77.47683923705722,
+ "grad_norm": 1.2241300344467163,
+ "learning_rate": 2.54472305319318e-06,
+ "loss": 0.0127,
+ "step": 28434
+ },
+ {
+ "epoch": 77.47956403269755,
+ "grad_norm": 2.36315655708313,
+ "learning_rate": 2.544134921922965e-06,
+ "loss": 0.0158,
+ "step": 28435
+ },
+ {
+ "epoch": 77.48228882833787,
+ "grad_norm": 1.217905044555664,
+ "learning_rate": 2.543546848718784e-06,
+ "loss": 0.0103,
+ "step": 28436
+ },
+ {
+ "epoch": 77.4850136239782,
+ "grad_norm": 0.9398493766784668,
+ "learning_rate": 2.542958833585224e-06,
+ "loss": 0.0103,
+ "step": 28437
+ },
+ {
+ "epoch": 77.48773841961852,
+ "grad_norm": 1.099652647972107,
+ "learning_rate": 2.5423708765268607e-06,
+ "loss": 0.0785,
+ "step": 28438
+ },
+ {
+ "epoch": 77.49046321525886,
+ "grad_norm": 1.9019792079925537,
+ "learning_rate": 2.541782977548273e-06,
+ "loss": 0.1764,
+ "step": 28439
+ },
+ {
+ "epoch": 77.49318801089919,
+ "grad_norm": 1.3450182676315308,
+ "learning_rate": 2.5411951366540357e-06,
+ "loss": 0.0184,
+ "step": 28440
+ },
+ {
+ "epoch": 77.49591280653951,
+ "grad_norm": 1.0524235963821411,
+ "learning_rate": 2.540607353848733e-06,
+ "loss": 0.0119,
+ "step": 28441
+ },
+ {
+ "epoch": 77.49863760217984,
+ "grad_norm": 1.7365305423736572,
+ "learning_rate": 2.5400196291369417e-06,
+ "loss": 0.0348,
+ "step": 28442
+ },
+ {
+ "epoch": 77.50136239782016,
+ "grad_norm": 1.1338505744934082,
+ "learning_rate": 2.5394319625232333e-06,
+ "loss": 0.0214,
+ "step": 28443
+ },
+ {
+ "epoch": 77.50408719346049,
+ "grad_norm": 1.0851263999938965,
+ "learning_rate": 2.538844354012192e-06,
+ "loss": 0.0214,
+ "step": 28444
+ },
+ {
+ "epoch": 77.50681198910081,
+ "grad_norm": 1.0563945770263672,
+ "learning_rate": 2.538256803608391e-06,
+ "loss": 0.0513,
+ "step": 28445
+ },
+ {
+ "epoch": 77.50953678474114,
+ "grad_norm": 1.8756686449050903,
+ "learning_rate": 2.5376693113164062e-06,
+ "loss": 0.0116,
+ "step": 28446
+ },
+ {
+ "epoch": 77.51226158038148,
+ "grad_norm": 1.8892192840576172,
+ "learning_rate": 2.537081877140809e-06,
+ "loss": 0.0321,
+ "step": 28447
+ },
+ {
+ "epoch": 77.5149863760218,
+ "grad_norm": 1.233333945274353,
+ "learning_rate": 2.536494501086181e-06,
+ "loss": 0.0168,
+ "step": 28448
+ },
+ {
+ "epoch": 77.51771117166213,
+ "grad_norm": 1.2233725786209106,
+ "learning_rate": 2.5359071831570937e-06,
+ "loss": 0.0459,
+ "step": 28449
+ },
+ {
+ "epoch": 77.52043596730245,
+ "grad_norm": 1.2178244590759277,
+ "learning_rate": 2.5353199233581184e-06,
+ "loss": 0.0075,
+ "step": 28450
+ },
+ {
+ "epoch": 77.52316076294278,
+ "grad_norm": 0.5974621772766113,
+ "learning_rate": 2.534732721693833e-06,
+ "loss": 0.0061,
+ "step": 28451
+ },
+ {
+ "epoch": 77.52588555858311,
+ "grad_norm": 1.0097196102142334,
+ "learning_rate": 2.5341455781688073e-06,
+ "loss": 0.02,
+ "step": 28452
+ },
+ {
+ "epoch": 77.52861035422343,
+ "grad_norm": 1.2226929664611816,
+ "learning_rate": 2.5335584927876178e-06,
+ "loss": 0.0647,
+ "step": 28453
+ },
+ {
+ "epoch": 77.53133514986376,
+ "grad_norm": 1.394325852394104,
+ "learning_rate": 2.5329714655548352e-06,
+ "loss": 0.0521,
+ "step": 28454
+ },
+ {
+ "epoch": 77.53405994550408,
+ "grad_norm": 2.1591954231262207,
+ "learning_rate": 2.53238449647503e-06,
+ "loss": 0.0647,
+ "step": 28455
+ },
+ {
+ "epoch": 77.53678474114442,
+ "grad_norm": 1.5315495729446411,
+ "learning_rate": 2.531797585552771e-06,
+ "loss": 0.086,
+ "step": 28456
+ },
+ {
+ "epoch": 77.53950953678473,
+ "grad_norm": 0.7409439086914062,
+ "learning_rate": 2.5312107327926348e-06,
+ "loss": 0.0065,
+ "step": 28457
+ },
+ {
+ "epoch": 77.54223433242507,
+ "grad_norm": 1.2204574346542358,
+ "learning_rate": 2.5306239381991893e-06,
+ "loss": 0.013,
+ "step": 28458
+ },
+ {
+ "epoch": 77.5449591280654,
+ "grad_norm": 1.2806973457336426,
+ "learning_rate": 2.5300372017770046e-06,
+ "loss": 0.0095,
+ "step": 28459
+ },
+ {
+ "epoch": 77.54768392370572,
+ "grad_norm": 1.4891000986099243,
+ "learning_rate": 2.5294505235306465e-06,
+ "loss": 0.0917,
+ "step": 28460
+ },
+ {
+ "epoch": 77.55040871934605,
+ "grad_norm": 1.3198299407958984,
+ "learning_rate": 2.52886390346469e-06,
+ "loss": 0.0457,
+ "step": 28461
+ },
+ {
+ "epoch": 77.55313351498637,
+ "grad_norm": 0.9516324996948242,
+ "learning_rate": 2.5282773415837013e-06,
+ "loss": 0.0113,
+ "step": 28462
+ },
+ {
+ "epoch": 77.5558583106267,
+ "grad_norm": 0.6138918995857239,
+ "learning_rate": 2.527690837892245e-06,
+ "loss": 0.0071,
+ "step": 28463
+ },
+ {
+ "epoch": 77.55858310626704,
+ "grad_norm": 1.1899765729904175,
+ "learning_rate": 2.5271043923948946e-06,
+ "loss": 0.0419,
+ "step": 28464
+ },
+ {
+ "epoch": 77.56130790190736,
+ "grad_norm": 1.1741408109664917,
+ "learning_rate": 2.5265180050962147e-06,
+ "loss": 0.0147,
+ "step": 28465
+ },
+ {
+ "epoch": 77.56403269754769,
+ "grad_norm": 1.6063344478607178,
+ "learning_rate": 2.5259316760007724e-06,
+ "loss": 0.0125,
+ "step": 28466
+ },
+ {
+ "epoch": 77.566757493188,
+ "grad_norm": 1.2378618717193604,
+ "learning_rate": 2.525345405113131e-06,
+ "loss": 0.029,
+ "step": 28467
+ },
+ {
+ "epoch": 77.56948228882834,
+ "grad_norm": 1.5965771675109863,
+ "learning_rate": 2.5247591924378577e-06,
+ "loss": 0.0296,
+ "step": 28468
+ },
+ {
+ "epoch": 77.57220708446866,
+ "grad_norm": 0.9742435216903687,
+ "learning_rate": 2.5241730379795235e-06,
+ "loss": 0.0094,
+ "step": 28469
+ },
+ {
+ "epoch": 77.57493188010899,
+ "grad_norm": 1.2109049558639526,
+ "learning_rate": 2.5235869417426885e-06,
+ "loss": 0.0362,
+ "step": 28470
+ },
+ {
+ "epoch": 77.57765667574932,
+ "grad_norm": 0.9880409240722656,
+ "learning_rate": 2.523000903731918e-06,
+ "loss": 0.013,
+ "step": 28471
+ },
+ {
+ "epoch": 77.58038147138964,
+ "grad_norm": 1.1982656717300415,
+ "learning_rate": 2.522414923951774e-06,
+ "loss": 0.0376,
+ "step": 28472
+ },
+ {
+ "epoch": 77.58310626702998,
+ "grad_norm": 1.7204793691635132,
+ "learning_rate": 2.5218290024068224e-06,
+ "loss": 0.0296,
+ "step": 28473
+ },
+ {
+ "epoch": 77.5858310626703,
+ "grad_norm": 1.1170623302459717,
+ "learning_rate": 2.521243139101628e-06,
+ "loss": 0.0118,
+ "step": 28474
+ },
+ {
+ "epoch": 77.58855585831063,
+ "grad_norm": 0.7520933151245117,
+ "learning_rate": 2.52065733404075e-06,
+ "loss": 0.008,
+ "step": 28475
+ },
+ {
+ "epoch": 77.59128065395096,
+ "grad_norm": 1.4327635765075684,
+ "learning_rate": 2.5200715872287495e-06,
+ "loss": 0.02,
+ "step": 28476
+ },
+ {
+ "epoch": 77.59400544959128,
+ "grad_norm": 1.1227959394454956,
+ "learning_rate": 2.5194858986701933e-06,
+ "loss": 0.0145,
+ "step": 28477
+ },
+ {
+ "epoch": 77.59673024523161,
+ "grad_norm": 1.2522543668746948,
+ "learning_rate": 2.5189002683696408e-06,
+ "loss": 0.0136,
+ "step": 28478
+ },
+ {
+ "epoch": 77.59945504087193,
+ "grad_norm": 1.1126556396484375,
+ "learning_rate": 2.518314696331652e-06,
+ "loss": 0.0339,
+ "step": 28479
+ },
+ {
+ "epoch": 77.60217983651226,
+ "grad_norm": 2.1843295097351074,
+ "learning_rate": 2.5177291825607843e-06,
+ "loss": 0.0247,
+ "step": 28480
+ },
+ {
+ "epoch": 77.60490463215258,
+ "grad_norm": 1.4458611011505127,
+ "learning_rate": 2.5171437270616038e-06,
+ "loss": 0.0424,
+ "step": 28481
+ },
+ {
+ "epoch": 77.60762942779292,
+ "grad_norm": 1.3683078289031982,
+ "learning_rate": 2.5165583298386673e-06,
+ "loss": 0.0955,
+ "step": 28482
+ },
+ {
+ "epoch": 77.61035422343325,
+ "grad_norm": 2.7313833236694336,
+ "learning_rate": 2.5159729908965346e-06,
+ "loss": 0.1665,
+ "step": 28483
+ },
+ {
+ "epoch": 77.61307901907357,
+ "grad_norm": 0.9046844840049744,
+ "learning_rate": 2.5153877102397596e-06,
+ "loss": 0.1152,
+ "step": 28484
+ },
+ {
+ "epoch": 77.6158038147139,
+ "grad_norm": 1.8886390924453735,
+ "learning_rate": 2.5148024878729037e-06,
+ "loss": 0.0636,
+ "step": 28485
+ },
+ {
+ "epoch": 77.61852861035422,
+ "grad_norm": 2.8659508228302,
+ "learning_rate": 2.5142173238005286e-06,
+ "loss": 0.0762,
+ "step": 28486
+ },
+ {
+ "epoch": 77.62125340599455,
+ "grad_norm": 1.2791568040847778,
+ "learning_rate": 2.5136322180271886e-06,
+ "loss": 0.012,
+ "step": 28487
+ },
+ {
+ "epoch": 77.62397820163488,
+ "grad_norm": 2.297523260116577,
+ "learning_rate": 2.5130471705574367e-06,
+ "loss": 0.0249,
+ "step": 28488
+ },
+ {
+ "epoch": 77.6267029972752,
+ "grad_norm": 1.1691434383392334,
+ "learning_rate": 2.512462181395835e-06,
+ "loss": 0.0115,
+ "step": 28489
+ },
+ {
+ "epoch": 77.62942779291554,
+ "grad_norm": 1.1786670684814453,
+ "learning_rate": 2.5118772505469367e-06,
+ "loss": 0.0136,
+ "step": 28490
+ },
+ {
+ "epoch": 77.63215258855585,
+ "grad_norm": 1.687559962272644,
+ "learning_rate": 2.5112923780152975e-06,
+ "loss": 0.018,
+ "step": 28491
+ },
+ {
+ "epoch": 77.63487738419619,
+ "grad_norm": 3.3714489936828613,
+ "learning_rate": 2.5107075638054703e-06,
+ "loss": 0.0832,
+ "step": 28492
+ },
+ {
+ "epoch": 77.6376021798365,
+ "grad_norm": 1.5481958389282227,
+ "learning_rate": 2.510122807922013e-06,
+ "loss": 0.0417,
+ "step": 28493
+ },
+ {
+ "epoch": 77.64032697547684,
+ "grad_norm": 2.067981243133545,
+ "learning_rate": 2.5095381103694794e-06,
+ "loss": 0.1675,
+ "step": 28494
+ },
+ {
+ "epoch": 77.64305177111717,
+ "grad_norm": 1.006509780883789,
+ "learning_rate": 2.508953471152421e-06,
+ "loss": 0.0175,
+ "step": 28495
+ },
+ {
+ "epoch": 77.64577656675749,
+ "grad_norm": 1.3883905410766602,
+ "learning_rate": 2.508368890275389e-06,
+ "loss": 0.0332,
+ "step": 28496
+ },
+ {
+ "epoch": 77.64850136239782,
+ "grad_norm": 0.8750317692756653,
+ "learning_rate": 2.5077843677429424e-06,
+ "loss": 0.008,
+ "step": 28497
+ },
+ {
+ "epoch": 77.65122615803814,
+ "grad_norm": 0.991430938243866,
+ "learning_rate": 2.5071999035596297e-06,
+ "loss": 0.0093,
+ "step": 28498
+ },
+ {
+ "epoch": 77.65395095367847,
+ "grad_norm": 1.4937375783920288,
+ "learning_rate": 2.5066154977300027e-06,
+ "loss": 0.0259,
+ "step": 28499
+ },
+ {
+ "epoch": 77.65667574931881,
+ "grad_norm": 1.0546661615371704,
+ "learning_rate": 2.506031150258611e-06,
+ "loss": 0.019,
+ "step": 28500
+ },
+ {
+ "epoch": 77.65940054495913,
+ "grad_norm": 1.1323041915893555,
+ "learning_rate": 2.5054468611500106e-06,
+ "loss": 0.0803,
+ "step": 28501
+ },
+ {
+ "epoch": 77.66212534059946,
+ "grad_norm": 1.0369411706924438,
+ "learning_rate": 2.5048626304087443e-06,
+ "loss": 0.0175,
+ "step": 28502
+ },
+ {
+ "epoch": 77.66485013623978,
+ "grad_norm": 1.302817463874817,
+ "learning_rate": 2.504278458039371e-06,
+ "loss": 0.0247,
+ "step": 28503
+ },
+ {
+ "epoch": 77.66757493188011,
+ "grad_norm": 1.7331663370132446,
+ "learning_rate": 2.5036943440464323e-06,
+ "loss": 0.0179,
+ "step": 28504
+ },
+ {
+ "epoch": 77.67029972752043,
+ "grad_norm": 1.3330585956573486,
+ "learning_rate": 2.5031102884344837e-06,
+ "loss": 0.0634,
+ "step": 28505
+ },
+ {
+ "epoch": 77.67302452316076,
+ "grad_norm": 1.0312634706497192,
+ "learning_rate": 2.5025262912080715e-06,
+ "loss": 0.024,
+ "step": 28506
+ },
+ {
+ "epoch": 77.6757493188011,
+ "grad_norm": 1.0994726419448853,
+ "learning_rate": 2.5019423523717425e-06,
+ "loss": 0.1034,
+ "step": 28507
+ },
+ {
+ "epoch": 77.67847411444141,
+ "grad_norm": 1.7417466640472412,
+ "learning_rate": 2.5013584719300423e-06,
+ "loss": 0.0292,
+ "step": 28508
+ },
+ {
+ "epoch": 77.68119891008175,
+ "grad_norm": 1.3509434461593628,
+ "learning_rate": 2.500774649887524e-06,
+ "loss": 0.0191,
+ "step": 28509
+ },
+ {
+ "epoch": 77.68392370572207,
+ "grad_norm": 1.9049501419067383,
+ "learning_rate": 2.500190886248731e-06,
+ "loss": 0.0162,
+ "step": 28510
+ },
+ {
+ "epoch": 77.6866485013624,
+ "grad_norm": 1.2243982553482056,
+ "learning_rate": 2.499607181018211e-06,
+ "loss": 0.0539,
+ "step": 28511
+ },
+ {
+ "epoch": 77.68937329700273,
+ "grad_norm": 1.3654323816299438,
+ "learning_rate": 2.4990235342005056e-06,
+ "loss": 0.011,
+ "step": 28512
+ },
+ {
+ "epoch": 77.69209809264305,
+ "grad_norm": 2.0325076580047607,
+ "learning_rate": 2.4984399458001662e-06,
+ "loss": 0.0638,
+ "step": 28513
+ },
+ {
+ "epoch": 77.69482288828338,
+ "grad_norm": 1.376541018486023,
+ "learning_rate": 2.4978564158217344e-06,
+ "loss": 0.0326,
+ "step": 28514
+ },
+ {
+ "epoch": 77.6975476839237,
+ "grad_norm": 1.2637457847595215,
+ "learning_rate": 2.497272944269756e-06,
+ "loss": 0.0253,
+ "step": 28515
+ },
+ {
+ "epoch": 77.70027247956403,
+ "grad_norm": 1.4951692819595337,
+ "learning_rate": 2.4966895311487716e-06,
+ "loss": 0.0164,
+ "step": 28516
+ },
+ {
+ "epoch": 77.70299727520435,
+ "grad_norm": 1.720140814781189,
+ "learning_rate": 2.496106176463331e-06,
+ "loss": 0.0989,
+ "step": 28517
+ },
+ {
+ "epoch": 77.70572207084469,
+ "grad_norm": 1.3829479217529297,
+ "learning_rate": 2.495522880217973e-06,
+ "loss": 0.0949,
+ "step": 28518
+ },
+ {
+ "epoch": 77.70844686648502,
+ "grad_norm": 2.2204859256744385,
+ "learning_rate": 2.494939642417239e-06,
+ "loss": 0.036,
+ "step": 28519
+ },
+ {
+ "epoch": 77.71117166212534,
+ "grad_norm": 1.7868378162384033,
+ "learning_rate": 2.494356463065676e-06,
+ "loss": 0.0652,
+ "step": 28520
+ },
+ {
+ "epoch": 77.71389645776567,
+ "grad_norm": 1.2097266912460327,
+ "learning_rate": 2.493773342167821e-06,
+ "loss": 0.0155,
+ "step": 28521
+ },
+ {
+ "epoch": 77.71662125340599,
+ "grad_norm": 0.9633457660675049,
+ "learning_rate": 2.49319027972822e-06,
+ "loss": 0.0092,
+ "step": 28522
+ },
+ {
+ "epoch": 77.71934604904632,
+ "grad_norm": 1.1738078594207764,
+ "learning_rate": 2.492607275751412e-06,
+ "loss": 0.0226,
+ "step": 28523
+ },
+ {
+ "epoch": 77.72207084468666,
+ "grad_norm": 1.3710618019104004,
+ "learning_rate": 2.4920243302419367e-06,
+ "loss": 0.0878,
+ "step": 28524
+ },
+ {
+ "epoch": 77.72479564032697,
+ "grad_norm": 1.3574637174606323,
+ "learning_rate": 2.491441443204331e-06,
+ "loss": 0.0653,
+ "step": 28525
+ },
+ {
+ "epoch": 77.7275204359673,
+ "grad_norm": 1.3406649827957153,
+ "learning_rate": 2.4908586146431413e-06,
+ "loss": 0.0224,
+ "step": 28526
+ },
+ {
+ "epoch": 77.73024523160763,
+ "grad_norm": 1.017437219619751,
+ "learning_rate": 2.4902758445629027e-06,
+ "loss": 0.0206,
+ "step": 28527
+ },
+ {
+ "epoch": 77.73297002724796,
+ "grad_norm": 1.4136055707931519,
+ "learning_rate": 2.4896931329681508e-06,
+ "loss": 0.0273,
+ "step": 28528
+ },
+ {
+ "epoch": 77.73569482288828,
+ "grad_norm": 1.315792441368103,
+ "learning_rate": 2.4891104798634303e-06,
+ "loss": 0.028,
+ "step": 28529
+ },
+ {
+ "epoch": 77.73841961852861,
+ "grad_norm": 0.8927386403083801,
+ "learning_rate": 2.4885278852532758e-06,
+ "loss": 0.0151,
+ "step": 28530
+ },
+ {
+ "epoch": 77.74114441416894,
+ "grad_norm": 1.1702433824539185,
+ "learning_rate": 2.4879453491422236e-06,
+ "loss": 0.0318,
+ "step": 28531
+ },
+ {
+ "epoch": 77.74386920980926,
+ "grad_norm": 1.7061023712158203,
+ "learning_rate": 2.4873628715348088e-06,
+ "loss": 0.0374,
+ "step": 28532
+ },
+ {
+ "epoch": 77.7465940054496,
+ "grad_norm": 0.6325781941413879,
+ "learning_rate": 2.486780452435573e-06,
+ "loss": 0.0082,
+ "step": 28533
+ },
+ {
+ "epoch": 77.74931880108991,
+ "grad_norm": 0.911754310131073,
+ "learning_rate": 2.48619809184905e-06,
+ "loss": 0.0072,
+ "step": 28534
+ },
+ {
+ "epoch": 77.75204359673025,
+ "grad_norm": 0.919879674911499,
+ "learning_rate": 2.485615789779774e-06,
+ "loss": 0.0107,
+ "step": 28535
+ },
+ {
+ "epoch": 77.75476839237058,
+ "grad_norm": 1.03573477268219,
+ "learning_rate": 2.485033546232277e-06,
+ "loss": 0.0197,
+ "step": 28536
+ },
+ {
+ "epoch": 77.7574931880109,
+ "grad_norm": 1.518622636795044,
+ "learning_rate": 2.4844513612110977e-06,
+ "loss": 0.015,
+ "step": 28537
+ },
+ {
+ "epoch": 77.76021798365123,
+ "grad_norm": 1.4869656562805176,
+ "learning_rate": 2.4838692347207717e-06,
+ "loss": 0.0134,
+ "step": 28538
+ },
+ {
+ "epoch": 77.76294277929155,
+ "grad_norm": 1.8292478322982788,
+ "learning_rate": 2.48328716676583e-06,
+ "loss": 0.0543,
+ "step": 28539
+ },
+ {
+ "epoch": 77.76566757493188,
+ "grad_norm": 1.8011804819107056,
+ "learning_rate": 2.4827051573508067e-06,
+ "loss": 0.0642,
+ "step": 28540
+ },
+ {
+ "epoch": 77.7683923705722,
+ "grad_norm": 1.4563730955123901,
+ "learning_rate": 2.482123206480229e-06,
+ "loss": 0.0511,
+ "step": 28541
+ },
+ {
+ "epoch": 77.77111716621253,
+ "grad_norm": 1.6633377075195312,
+ "learning_rate": 2.4815413141586387e-06,
+ "loss": 0.1703,
+ "step": 28542
+ },
+ {
+ "epoch": 77.77384196185287,
+ "grad_norm": 0.9758829474449158,
+ "learning_rate": 2.4809594803905614e-06,
+ "loss": 0.011,
+ "step": 28543
+ },
+ {
+ "epoch": 77.77656675749319,
+ "grad_norm": 1.5255473852157593,
+ "learning_rate": 2.48037770518053e-06,
+ "loss": 0.1079,
+ "step": 28544
+ },
+ {
+ "epoch": 77.77929155313352,
+ "grad_norm": 1.290862798690796,
+ "learning_rate": 2.4797959885330724e-06,
+ "loss": 0.0159,
+ "step": 28545
+ },
+ {
+ "epoch": 77.78201634877384,
+ "grad_norm": 1.7354440689086914,
+ "learning_rate": 2.479214330452725e-06,
+ "loss": 0.1825,
+ "step": 28546
+ },
+ {
+ "epoch": 77.78474114441417,
+ "grad_norm": 2.191549777984619,
+ "learning_rate": 2.4786327309440138e-06,
+ "loss": 0.0331,
+ "step": 28547
+ },
+ {
+ "epoch": 77.7874659400545,
+ "grad_norm": 0.9159541726112366,
+ "learning_rate": 2.4780511900114678e-06,
+ "loss": 0.0095,
+ "step": 28548
+ },
+ {
+ "epoch": 77.79019073569482,
+ "grad_norm": 0.6110228300094604,
+ "learning_rate": 2.4774697076596154e-06,
+ "loss": 0.0061,
+ "step": 28549
+ },
+ {
+ "epoch": 77.79291553133515,
+ "grad_norm": 1.71611750125885,
+ "learning_rate": 2.47688828389299e-06,
+ "loss": 0.0104,
+ "step": 28550
+ },
+ {
+ "epoch": 77.79564032697547,
+ "grad_norm": 1.4782685041427612,
+ "learning_rate": 2.4763069187161158e-06,
+ "loss": 0.0544,
+ "step": 28551
+ },
+ {
+ "epoch": 77.7983651226158,
+ "grad_norm": 0.9935230612754822,
+ "learning_rate": 2.4757256121335182e-06,
+ "loss": 0.0105,
+ "step": 28552
+ },
+ {
+ "epoch": 77.80108991825612,
+ "grad_norm": 5.451310634613037,
+ "learning_rate": 2.4751443641497307e-06,
+ "loss": 0.0333,
+ "step": 28553
+ },
+ {
+ "epoch": 77.80381471389646,
+ "grad_norm": 1.8358381986618042,
+ "learning_rate": 2.4745631747692732e-06,
+ "loss": 0.0873,
+ "step": 28554
+ },
+ {
+ "epoch": 77.80653950953679,
+ "grad_norm": 1.1178480386734009,
+ "learning_rate": 2.473982043996679e-06,
+ "loss": 0.0088,
+ "step": 28555
+ },
+ {
+ "epoch": 77.80926430517711,
+ "grad_norm": 1.0089017152786255,
+ "learning_rate": 2.4734009718364693e-06,
+ "loss": 0.0149,
+ "step": 28556
+ },
+ {
+ "epoch": 77.81198910081744,
+ "grad_norm": 0.7431792616844177,
+ "learning_rate": 2.4728199582931687e-06,
+ "loss": 0.012,
+ "step": 28557
+ },
+ {
+ "epoch": 77.81471389645776,
+ "grad_norm": 1.4766119718551636,
+ "learning_rate": 2.4722390033713074e-06,
+ "loss": 0.0655,
+ "step": 28558
+ },
+ {
+ "epoch": 77.8174386920981,
+ "grad_norm": 1.4701077938079834,
+ "learning_rate": 2.471658107075405e-06,
+ "loss": 0.0504,
+ "step": 28559
+ },
+ {
+ "epoch": 77.82016348773843,
+ "grad_norm": 1.26137375831604,
+ "learning_rate": 2.4710772694099885e-06,
+ "loss": 0.0131,
+ "step": 28560
+ },
+ {
+ "epoch": 77.82288828337875,
+ "grad_norm": 1.2513620853424072,
+ "learning_rate": 2.470496490379576e-06,
+ "loss": 0.1751,
+ "step": 28561
+ },
+ {
+ "epoch": 77.82561307901908,
+ "grad_norm": 0.7543807029724121,
+ "learning_rate": 2.4699157699886977e-06,
+ "loss": 0.0087,
+ "step": 28562
+ },
+ {
+ "epoch": 77.8283378746594,
+ "grad_norm": 2.013458490371704,
+ "learning_rate": 2.4693351082418725e-06,
+ "loss": 0.0505,
+ "step": 28563
+ },
+ {
+ "epoch": 77.83106267029973,
+ "grad_norm": 1.39830482006073,
+ "learning_rate": 2.4687545051436233e-06,
+ "loss": 0.05,
+ "step": 28564
+ },
+ {
+ "epoch": 77.83378746594005,
+ "grad_norm": 1.3784246444702148,
+ "learning_rate": 2.46817396069847e-06,
+ "loss": 0.1511,
+ "step": 28565
+ },
+ {
+ "epoch": 77.83651226158038,
+ "grad_norm": 0.986489474773407,
+ "learning_rate": 2.4675934749109375e-06,
+ "loss": 0.0142,
+ "step": 28566
+ },
+ {
+ "epoch": 77.83923705722071,
+ "grad_norm": 1.4245661497116089,
+ "learning_rate": 2.4670130477855437e-06,
+ "loss": 0.0217,
+ "step": 28567
+ },
+ {
+ "epoch": 77.84196185286103,
+ "grad_norm": 1.5985379219055176,
+ "learning_rate": 2.466432679326811e-06,
+ "loss": 0.0397,
+ "step": 28568
+ },
+ {
+ "epoch": 77.84468664850137,
+ "grad_norm": 2.1296427249908447,
+ "learning_rate": 2.4658523695392545e-06,
+ "loss": 0.0839,
+ "step": 28569
+ },
+ {
+ "epoch": 77.84741144414168,
+ "grad_norm": 1.6536705493927002,
+ "learning_rate": 2.4652721184274008e-06,
+ "loss": 0.0461,
+ "step": 28570
+ },
+ {
+ "epoch": 77.85013623978202,
+ "grad_norm": 6.243826389312744,
+ "learning_rate": 2.4646919259957623e-06,
+ "loss": 0.0227,
+ "step": 28571
+ },
+ {
+ "epoch": 77.85286103542235,
+ "grad_norm": 1.511678695678711,
+ "learning_rate": 2.464111792248863e-06,
+ "loss": 0.0096,
+ "step": 28572
+ },
+ {
+ "epoch": 77.85558583106267,
+ "grad_norm": 1.482251524925232,
+ "learning_rate": 2.4635317171912164e-06,
+ "loss": 0.025,
+ "step": 28573
+ },
+ {
+ "epoch": 77.858310626703,
+ "grad_norm": 0.9841312766075134,
+ "learning_rate": 2.462951700827344e-06,
+ "loss": 0.0142,
+ "step": 28574
+ },
+ {
+ "epoch": 77.86103542234332,
+ "grad_norm": 2.423558235168457,
+ "learning_rate": 2.462371743161762e-06,
+ "loss": 0.0793,
+ "step": 28575
+ },
+ {
+ "epoch": 77.86376021798365,
+ "grad_norm": 1.3421781063079834,
+ "learning_rate": 2.461791844198986e-06,
+ "loss": 0.0081,
+ "step": 28576
+ },
+ {
+ "epoch": 77.86648501362397,
+ "grad_norm": 1.7166131734848022,
+ "learning_rate": 2.4612120039435306e-06,
+ "loss": 0.0176,
+ "step": 28577
+ },
+ {
+ "epoch": 77.8692098092643,
+ "grad_norm": 1.532470464706421,
+ "learning_rate": 2.4606322223999147e-06,
+ "loss": 0.0124,
+ "step": 28578
+ },
+ {
+ "epoch": 77.87193460490464,
+ "grad_norm": 1.0385793447494507,
+ "learning_rate": 2.4600524995726537e-06,
+ "loss": 0.0173,
+ "step": 28579
+ },
+ {
+ "epoch": 77.87465940054496,
+ "grad_norm": 1.4260272979736328,
+ "learning_rate": 2.45947283546626e-06,
+ "loss": 0.0207,
+ "step": 28580
+ },
+ {
+ "epoch": 77.87738419618529,
+ "grad_norm": 2.211726188659668,
+ "learning_rate": 2.4588932300852476e-06,
+ "loss": 0.0217,
+ "step": 28581
+ },
+ {
+ "epoch": 77.88010899182561,
+ "grad_norm": 0.8332365155220032,
+ "learning_rate": 2.4583136834341336e-06,
+ "loss": 0.0119,
+ "step": 28582
+ },
+ {
+ "epoch": 77.88283378746594,
+ "grad_norm": 0.998366117477417,
+ "learning_rate": 2.4577341955174307e-06,
+ "loss": 0.0116,
+ "step": 28583
+ },
+ {
+ "epoch": 77.88555858310627,
+ "grad_norm": 2.4745092391967773,
+ "learning_rate": 2.4571547663396513e-06,
+ "loss": 0.0457,
+ "step": 28584
+ },
+ {
+ "epoch": 77.88828337874659,
+ "grad_norm": 0.7166932225227356,
+ "learning_rate": 2.456575395905304e-06,
+ "loss": 0.0101,
+ "step": 28585
+ },
+ {
+ "epoch": 77.89100817438693,
+ "grad_norm": 1.0878939628601074,
+ "learning_rate": 2.455996084218909e-06,
+ "loss": 0.0117,
+ "step": 28586
+ },
+ {
+ "epoch": 77.89373297002724,
+ "grad_norm": 1.075600266456604,
+ "learning_rate": 2.4554168312849724e-06,
+ "loss": 0.0365,
+ "step": 28587
+ },
+ {
+ "epoch": 77.89645776566758,
+ "grad_norm": 1.0185284614562988,
+ "learning_rate": 2.454837637108004e-06,
+ "loss": 0.0095,
+ "step": 28588
+ },
+ {
+ "epoch": 77.8991825613079,
+ "grad_norm": 0.7724354863166809,
+ "learning_rate": 2.4542585016925215e-06,
+ "loss": 0.0177,
+ "step": 28589
+ },
+ {
+ "epoch": 77.90190735694823,
+ "grad_norm": 0.4744369089603424,
+ "learning_rate": 2.4536794250430264e-06,
+ "loss": 0.0036,
+ "step": 28590
+ },
+ {
+ "epoch": 77.90463215258856,
+ "grad_norm": 1.1371771097183228,
+ "learning_rate": 2.453100407164036e-06,
+ "loss": 0.0747,
+ "step": 28591
+ },
+ {
+ "epoch": 77.90735694822888,
+ "grad_norm": 1.0116015672683716,
+ "learning_rate": 2.4525214480600577e-06,
+ "loss": 0.0137,
+ "step": 28592
+ },
+ {
+ "epoch": 77.91008174386921,
+ "grad_norm": 0.8686796426773071,
+ "learning_rate": 2.451942547735596e-06,
+ "loss": 0.0078,
+ "step": 28593
+ },
+ {
+ "epoch": 77.91280653950953,
+ "grad_norm": 1.3699194192886353,
+ "learning_rate": 2.4513637061951655e-06,
+ "loss": 0.0098,
+ "step": 28594
+ },
+ {
+ "epoch": 77.91553133514986,
+ "grad_norm": 1.1539247035980225,
+ "learning_rate": 2.450784923443272e-06,
+ "loss": 0.0215,
+ "step": 28595
+ },
+ {
+ "epoch": 77.9182561307902,
+ "grad_norm": 1.2668583393096924,
+ "learning_rate": 2.450206199484423e-06,
+ "loss": 0.0193,
+ "step": 28596
+ },
+ {
+ "epoch": 77.92098092643052,
+ "grad_norm": 1.1273236274719238,
+ "learning_rate": 2.4496275343231214e-06,
+ "loss": 0.0105,
+ "step": 28597
+ },
+ {
+ "epoch": 77.92370572207085,
+ "grad_norm": 0.8807132244110107,
+ "learning_rate": 2.4490489279638806e-06,
+ "loss": 0.0095,
+ "step": 28598
+ },
+ {
+ "epoch": 77.92643051771117,
+ "grad_norm": 1.3914614915847778,
+ "learning_rate": 2.448470380411204e-06,
+ "loss": 0.0195,
+ "step": 28599
+ },
+ {
+ "epoch": 77.9291553133515,
+ "grad_norm": 0.5585650205612183,
+ "learning_rate": 2.447891891669596e-06,
+ "loss": 0.0059,
+ "step": 28600
+ },
+ {
+ "epoch": 77.93188010899182,
+ "grad_norm": 1.782913327217102,
+ "learning_rate": 2.4473134617435614e-06,
+ "loss": 0.0172,
+ "step": 28601
+ },
+ {
+ "epoch": 77.93460490463215,
+ "grad_norm": 1.3978779315948486,
+ "learning_rate": 2.4467350906376085e-06,
+ "loss": 0.0495,
+ "step": 28602
+ },
+ {
+ "epoch": 77.93732970027249,
+ "grad_norm": 1.4516502618789673,
+ "learning_rate": 2.44615677835624e-06,
+ "loss": 0.0523,
+ "step": 28603
+ },
+ {
+ "epoch": 77.9400544959128,
+ "grad_norm": 1.2107065916061401,
+ "learning_rate": 2.44557852490396e-06,
+ "loss": 0.0245,
+ "step": 28604
+ },
+ {
+ "epoch": 77.94277929155314,
+ "grad_norm": 1.4852216243743896,
+ "learning_rate": 2.4450003302852663e-06,
+ "loss": 0.0345,
+ "step": 28605
+ },
+ {
+ "epoch": 77.94550408719346,
+ "grad_norm": 0.9350071549415588,
+ "learning_rate": 2.4444221945046687e-06,
+ "loss": 0.0081,
+ "step": 28606
+ },
+ {
+ "epoch": 77.94822888283379,
+ "grad_norm": 1.3178482055664062,
+ "learning_rate": 2.4438441175666706e-06,
+ "loss": 0.0216,
+ "step": 28607
+ },
+ {
+ "epoch": 77.95095367847412,
+ "grad_norm": 1.3953204154968262,
+ "learning_rate": 2.4432660994757707e-06,
+ "loss": 0.022,
+ "step": 28608
+ },
+ {
+ "epoch": 77.95367847411444,
+ "grad_norm": 1.866542100906372,
+ "learning_rate": 2.4426881402364722e-06,
+ "loss": 0.0126,
+ "step": 28609
+ },
+ {
+ "epoch": 77.95640326975477,
+ "grad_norm": 1.0069276094436646,
+ "learning_rate": 2.4421102398532716e-06,
+ "loss": 0.0068,
+ "step": 28610
+ },
+ {
+ "epoch": 77.95912806539509,
+ "grad_norm": 1.3102562427520752,
+ "learning_rate": 2.441532398330676e-06,
+ "loss": 0.0992,
+ "step": 28611
+ },
+ {
+ "epoch": 77.96185286103542,
+ "grad_norm": 2.566638946533203,
+ "learning_rate": 2.4409546156731835e-06,
+ "loss": 0.0096,
+ "step": 28612
+ },
+ {
+ "epoch": 77.96457765667574,
+ "grad_norm": 0.8740595579147339,
+ "learning_rate": 2.4403768918852923e-06,
+ "loss": 0.0138,
+ "step": 28613
+ },
+ {
+ "epoch": 77.96730245231608,
+ "grad_norm": 1.638161063194275,
+ "learning_rate": 2.4397992269714996e-06,
+ "loss": 0.0233,
+ "step": 28614
+ },
+ {
+ "epoch": 77.97002724795641,
+ "grad_norm": 1.5065536499023438,
+ "learning_rate": 2.4392216209363106e-06,
+ "loss": 0.009,
+ "step": 28615
+ },
+ {
+ "epoch": 77.97275204359673,
+ "grad_norm": 0.573079526424408,
+ "learning_rate": 2.438644073784221e-06,
+ "loss": 0.0043,
+ "step": 28616
+ },
+ {
+ "epoch": 77.97547683923706,
+ "grad_norm": 1.2304097414016724,
+ "learning_rate": 2.438066585519724e-06,
+ "loss": 0.0155,
+ "step": 28617
+ },
+ {
+ "epoch": 77.97820163487738,
+ "grad_norm": 3.6435399055480957,
+ "learning_rate": 2.4374891561473236e-06,
+ "loss": 0.0245,
+ "step": 28618
+ },
+ {
+ "epoch": 77.98092643051771,
+ "grad_norm": 1.1146737337112427,
+ "learning_rate": 2.4369117856715152e-06,
+ "loss": 0.0935,
+ "step": 28619
+ },
+ {
+ "epoch": 77.98365122615803,
+ "grad_norm": 1.4462915658950806,
+ "learning_rate": 2.436334474096794e-06,
+ "loss": 0.2042,
+ "step": 28620
+ },
+ {
+ "epoch": 77.98637602179836,
+ "grad_norm": 1.61564040184021,
+ "learning_rate": 2.4357572214276536e-06,
+ "loss": 0.0582,
+ "step": 28621
+ },
+ {
+ "epoch": 77.9891008174387,
+ "grad_norm": 1.6922175884246826,
+ "learning_rate": 2.4351800276685932e-06,
+ "loss": 0.0261,
+ "step": 28622
+ },
+ {
+ "epoch": 77.99182561307902,
+ "grad_norm": 1.0890395641326904,
+ "learning_rate": 2.4346028928241095e-06,
+ "loss": 0.0109,
+ "step": 28623
+ },
+ {
+ "epoch": 77.99455040871935,
+ "grad_norm": 1.1419399976730347,
+ "learning_rate": 2.434025816898695e-06,
+ "loss": 0.0125,
+ "step": 28624
+ },
+ {
+ "epoch": 77.99727520435967,
+ "grad_norm": 1.2620596885681152,
+ "learning_rate": 2.4334487998968435e-06,
+ "loss": 0.0118,
+ "step": 28625
+ },
+ {
+ "epoch": 78.0,
+ "grad_norm": 2.040494918823242,
+ "learning_rate": 2.432871841823047e-06,
+ "loss": 0.018,
+ "step": 28626
+ },
+ {
+ "epoch": 78.00272479564033,
+ "grad_norm": 1.5461397171020508,
+ "learning_rate": 2.432294942681803e-06,
+ "loss": 0.0253,
+ "step": 28627
+ },
+ {
+ "epoch": 78.00544959128065,
+ "grad_norm": 2.0027787685394287,
+ "learning_rate": 2.431718102477604e-06,
+ "loss": 0.0269,
+ "step": 28628
+ },
+ {
+ "epoch": 78.00817438692098,
+ "grad_norm": 1.3239442110061646,
+ "learning_rate": 2.4311413212149395e-06,
+ "loss": 0.0346,
+ "step": 28629
+ },
+ {
+ "epoch": 78.0108991825613,
+ "grad_norm": 2.0311148166656494,
+ "learning_rate": 2.4305645988983007e-06,
+ "loss": 0.0434,
+ "step": 28630
+ },
+ {
+ "epoch": 78.01362397820164,
+ "grad_norm": 1.0320303440093994,
+ "learning_rate": 2.4299879355321844e-06,
+ "loss": 0.0108,
+ "step": 28631
+ },
+ {
+ "epoch": 78.01634877384195,
+ "grad_norm": 0.8456431031227112,
+ "learning_rate": 2.4294113311210775e-06,
+ "loss": 0.0114,
+ "step": 28632
+ },
+ {
+ "epoch": 78.01907356948229,
+ "grad_norm": 0.9742415547370911,
+ "learning_rate": 2.4288347856694718e-06,
+ "loss": 0.012,
+ "step": 28633
+ },
+ {
+ "epoch": 78.02179836512262,
+ "grad_norm": 0.8024862408638,
+ "learning_rate": 2.428258299181854e-06,
+ "loss": 0.0074,
+ "step": 28634
+ },
+ {
+ "epoch": 78.02452316076294,
+ "grad_norm": 2.0074849128723145,
+ "learning_rate": 2.4276818716627195e-06,
+ "loss": 0.0236,
+ "step": 28635
+ },
+ {
+ "epoch": 78.02724795640327,
+ "grad_norm": 1.559186339378357,
+ "learning_rate": 2.4271055031165557e-06,
+ "loss": 0.047,
+ "step": 28636
+ },
+ {
+ "epoch": 78.02997275204359,
+ "grad_norm": 1.1818230152130127,
+ "learning_rate": 2.4265291935478472e-06,
+ "loss": 0.0784,
+ "step": 28637
+ },
+ {
+ "epoch": 78.03269754768392,
+ "grad_norm": 1.5839229822158813,
+ "learning_rate": 2.4259529429610884e-06,
+ "loss": 0.0155,
+ "step": 28638
+ },
+ {
+ "epoch": 78.03542234332426,
+ "grad_norm": 1.0253630876541138,
+ "learning_rate": 2.425376751360764e-06,
+ "loss": 0.0399,
+ "step": 28639
+ },
+ {
+ "epoch": 78.03814713896458,
+ "grad_norm": 0.8354984521865845,
+ "learning_rate": 2.4248006187513597e-06,
+ "loss": 0.0101,
+ "step": 28640
+ },
+ {
+ "epoch": 78.04087193460491,
+ "grad_norm": 1.3447368144989014,
+ "learning_rate": 2.4242245451373667e-06,
+ "loss": 0.017,
+ "step": 28641
+ },
+ {
+ "epoch": 78.04359673024523,
+ "grad_norm": 1.7040327787399292,
+ "learning_rate": 2.423648530523266e-06,
+ "loss": 0.0504,
+ "step": 28642
+ },
+ {
+ "epoch": 78.04632152588556,
+ "grad_norm": 1.4163553714752197,
+ "learning_rate": 2.4230725749135496e-06,
+ "loss": 0.0255,
+ "step": 28643
+ },
+ {
+ "epoch": 78.04904632152588,
+ "grad_norm": 0.5501015186309814,
+ "learning_rate": 2.4224966783127014e-06,
+ "loss": 0.0052,
+ "step": 28644
+ },
+ {
+ "epoch": 78.05177111716621,
+ "grad_norm": 1.5670114755630493,
+ "learning_rate": 2.4219208407252036e-06,
+ "loss": 0.0423,
+ "step": 28645
+ },
+ {
+ "epoch": 78.05449591280654,
+ "grad_norm": 1.483088731765747,
+ "learning_rate": 2.42134506215554e-06,
+ "loss": 0.0132,
+ "step": 28646
+ },
+ {
+ "epoch": 78.05722070844686,
+ "grad_norm": 1.426088809967041,
+ "learning_rate": 2.4207693426082003e-06,
+ "loss": 0.0133,
+ "step": 28647
+ },
+ {
+ "epoch": 78.0599455040872,
+ "grad_norm": 3.7164173126220703,
+ "learning_rate": 2.420193682087665e-06,
+ "loss": 0.0979,
+ "step": 28648
+ },
+ {
+ "epoch": 78.06267029972751,
+ "grad_norm": 1.315002202987671,
+ "learning_rate": 2.419618080598417e-06,
+ "loss": 0.0114,
+ "step": 28649
+ },
+ {
+ "epoch": 78.06539509536785,
+ "grad_norm": 0.9589597582817078,
+ "learning_rate": 2.4190425381449366e-06,
+ "loss": 0.0068,
+ "step": 28650
+ },
+ {
+ "epoch": 78.06811989100818,
+ "grad_norm": 4.518345832824707,
+ "learning_rate": 2.4184670547317123e-06,
+ "loss": 0.0528,
+ "step": 28651
+ },
+ {
+ "epoch": 78.0708446866485,
+ "grad_norm": 0.7694119215011597,
+ "learning_rate": 2.417891630363222e-06,
+ "loss": 0.0104,
+ "step": 28652
+ },
+ {
+ "epoch": 78.07356948228883,
+ "grad_norm": 0.5944347977638245,
+ "learning_rate": 2.4173162650439485e-06,
+ "loss": 0.0059,
+ "step": 28653
+ },
+ {
+ "epoch": 78.07629427792915,
+ "grad_norm": 1.7117589712142944,
+ "learning_rate": 2.4167409587783676e-06,
+ "loss": 0.0479,
+ "step": 28654
+ },
+ {
+ "epoch": 78.07901907356948,
+ "grad_norm": 1.4379005432128906,
+ "learning_rate": 2.416165711570968e-06,
+ "loss": 0.0696,
+ "step": 28655
+ },
+ {
+ "epoch": 78.0817438692098,
+ "grad_norm": 1.5576567649841309,
+ "learning_rate": 2.4155905234262255e-06,
+ "loss": 0.0984,
+ "step": 28656
+ },
+ {
+ "epoch": 78.08446866485014,
+ "grad_norm": 0.9570499062538147,
+ "learning_rate": 2.4150153943486166e-06,
+ "loss": 0.0105,
+ "step": 28657
+ },
+ {
+ "epoch": 78.08719346049047,
+ "grad_norm": 0.9662322402000427,
+ "learning_rate": 2.414440324342624e-06,
+ "loss": 0.0073,
+ "step": 28658
+ },
+ {
+ "epoch": 78.08991825613079,
+ "grad_norm": 0.6851658821105957,
+ "learning_rate": 2.413865313412729e-06,
+ "loss": 0.0084,
+ "step": 28659
+ },
+ {
+ "epoch": 78.09264305177112,
+ "grad_norm": 0.7891389727592468,
+ "learning_rate": 2.4132903615634073e-06,
+ "loss": 0.0106,
+ "step": 28660
+ },
+ {
+ "epoch": 78.09536784741144,
+ "grad_norm": 0.7398011088371277,
+ "learning_rate": 2.412715468799135e-06,
+ "loss": 0.008,
+ "step": 28661
+ },
+ {
+ "epoch": 78.09809264305177,
+ "grad_norm": 1.0885188579559326,
+ "learning_rate": 2.412140635124388e-06,
+ "loss": 0.0108,
+ "step": 28662
+ },
+ {
+ "epoch": 78.1008174386921,
+ "grad_norm": 1.9275920391082764,
+ "learning_rate": 2.4115658605436488e-06,
+ "loss": 0.0208,
+ "step": 28663
+ },
+ {
+ "epoch": 78.10354223433242,
+ "grad_norm": 2.3398075103759766,
+ "learning_rate": 2.41099114506139e-06,
+ "loss": 0.139,
+ "step": 28664
+ },
+ {
+ "epoch": 78.10626702997276,
+ "grad_norm": 1.7694885730743408,
+ "learning_rate": 2.410416488682088e-06,
+ "loss": 0.0214,
+ "step": 28665
+ },
+ {
+ "epoch": 78.10899182561307,
+ "grad_norm": 0.48717716336250305,
+ "learning_rate": 2.409841891410215e-06,
+ "loss": 0.0055,
+ "step": 28666
+ },
+ {
+ "epoch": 78.11171662125341,
+ "grad_norm": 0.5093296766281128,
+ "learning_rate": 2.409267353250252e-06,
+ "loss": 0.006,
+ "step": 28667
+ },
+ {
+ "epoch": 78.11444141689373,
+ "grad_norm": 1.6139194965362549,
+ "learning_rate": 2.408692874206672e-06,
+ "loss": 0.0357,
+ "step": 28668
+ },
+ {
+ "epoch": 78.11716621253406,
+ "grad_norm": 1.1410198211669922,
+ "learning_rate": 2.4081184542839455e-06,
+ "loss": 0.0139,
+ "step": 28669
+ },
+ {
+ "epoch": 78.11989100817439,
+ "grad_norm": 0.9302891492843628,
+ "learning_rate": 2.407544093486546e-06,
+ "loss": 0.011,
+ "step": 28670
+ },
+ {
+ "epoch": 78.12261580381471,
+ "grad_norm": 0.6880380511283875,
+ "learning_rate": 2.4069697918189515e-06,
+ "loss": 0.0058,
+ "step": 28671
+ },
+ {
+ "epoch": 78.12534059945504,
+ "grad_norm": 1.3299278020858765,
+ "learning_rate": 2.406395549285633e-06,
+ "loss": 0.0658,
+ "step": 28672
+ },
+ {
+ "epoch": 78.12806539509536,
+ "grad_norm": 1.20512056350708,
+ "learning_rate": 2.4058213658910602e-06,
+ "loss": 0.0325,
+ "step": 28673
+ },
+ {
+ "epoch": 78.1307901907357,
+ "grad_norm": 38.32707595825195,
+ "learning_rate": 2.4052472416397034e-06,
+ "loss": 0.0182,
+ "step": 28674
+ },
+ {
+ "epoch": 78.13351498637603,
+ "grad_norm": 0.6823643445968628,
+ "learning_rate": 2.4046731765360377e-06,
+ "loss": 0.0055,
+ "step": 28675
+ },
+ {
+ "epoch": 78.13623978201635,
+ "grad_norm": 1.1748605966567993,
+ "learning_rate": 2.404099170584534e-06,
+ "loss": 0.131,
+ "step": 28676
+ },
+ {
+ "epoch": 78.13896457765668,
+ "grad_norm": 0.548704206943512,
+ "learning_rate": 2.4035252237896635e-06,
+ "loss": 0.0052,
+ "step": 28677
+ },
+ {
+ "epoch": 78.141689373297,
+ "grad_norm": 1.0614607334136963,
+ "learning_rate": 2.40295133615589e-06,
+ "loss": 0.0134,
+ "step": 28678
+ },
+ {
+ "epoch": 78.14441416893733,
+ "grad_norm": 0.9564207792282104,
+ "learning_rate": 2.40237750768769e-06,
+ "loss": 0.0231,
+ "step": 28679
+ },
+ {
+ "epoch": 78.14713896457765,
+ "grad_norm": 1.3409656286239624,
+ "learning_rate": 2.4018037383895287e-06,
+ "loss": 0.0233,
+ "step": 28680
+ },
+ {
+ "epoch": 78.14986376021798,
+ "grad_norm": 1.7423663139343262,
+ "learning_rate": 2.401230028265876e-06,
+ "loss": 0.0116,
+ "step": 28681
+ },
+ {
+ "epoch": 78.15258855585832,
+ "grad_norm": 1.263335108757019,
+ "learning_rate": 2.400656377321197e-06,
+ "loss": 0.0097,
+ "step": 28682
+ },
+ {
+ "epoch": 78.15531335149863,
+ "grad_norm": 1.4432626962661743,
+ "learning_rate": 2.400082785559964e-06,
+ "loss": 0.0623,
+ "step": 28683
+ },
+ {
+ "epoch": 78.15803814713897,
+ "grad_norm": 1.2948222160339355,
+ "learning_rate": 2.3995092529866415e-06,
+ "loss": 0.0124,
+ "step": 28684
+ },
+ {
+ "epoch": 78.16076294277929,
+ "grad_norm": 1.0139796733856201,
+ "learning_rate": 2.3989357796056966e-06,
+ "loss": 0.0161,
+ "step": 28685
+ },
+ {
+ "epoch": 78.16348773841962,
+ "grad_norm": 1.1718569993972778,
+ "learning_rate": 2.398362365421593e-06,
+ "loss": 0.0914,
+ "step": 28686
+ },
+ {
+ "epoch": 78.16621253405995,
+ "grad_norm": 1.8131288290023804,
+ "learning_rate": 2.3977890104388002e-06,
+ "loss": 0.1039,
+ "step": 28687
+ },
+ {
+ "epoch": 78.16893732970027,
+ "grad_norm": 1.769735336303711,
+ "learning_rate": 2.3972157146617823e-06,
+ "loss": 0.0218,
+ "step": 28688
+ },
+ {
+ "epoch": 78.1716621253406,
+ "grad_norm": 1.5432149171829224,
+ "learning_rate": 2.396642478095004e-06,
+ "loss": 0.0507,
+ "step": 28689
+ },
+ {
+ "epoch": 78.17438692098092,
+ "grad_norm": 1.0786222219467163,
+ "learning_rate": 2.396069300742926e-06,
+ "loss": 0.0086,
+ "step": 28690
+ },
+ {
+ "epoch": 78.17711171662125,
+ "grad_norm": 1.0141032934188843,
+ "learning_rate": 2.3954961826100154e-06,
+ "loss": 0.0131,
+ "step": 28691
+ },
+ {
+ "epoch": 78.17983651226157,
+ "grad_norm": 1.5855261087417603,
+ "learning_rate": 2.3949231237007396e-06,
+ "loss": 0.0303,
+ "step": 28692
+ },
+ {
+ "epoch": 78.1825613079019,
+ "grad_norm": 0.9417167901992798,
+ "learning_rate": 2.3943501240195565e-06,
+ "loss": 0.012,
+ "step": 28693
+ },
+ {
+ "epoch": 78.18528610354224,
+ "grad_norm": 1.2879786491394043,
+ "learning_rate": 2.3937771835709313e-06,
+ "loss": 0.0235,
+ "step": 28694
+ },
+ {
+ "epoch": 78.18801089918256,
+ "grad_norm": 0.6982269287109375,
+ "learning_rate": 2.393204302359321e-06,
+ "loss": 0.0069,
+ "step": 28695
+ },
+ {
+ "epoch": 78.19073569482289,
+ "grad_norm": 1.440417766571045,
+ "learning_rate": 2.392631480389194e-06,
+ "loss": 0.0451,
+ "step": 28696
+ },
+ {
+ "epoch": 78.19346049046321,
+ "grad_norm": 1.0936203002929688,
+ "learning_rate": 2.3920587176650077e-06,
+ "loss": 0.0198,
+ "step": 28697
+ },
+ {
+ "epoch": 78.19618528610354,
+ "grad_norm": 1.1594690084457397,
+ "learning_rate": 2.3914860141912234e-06,
+ "loss": 0.0091,
+ "step": 28698
+ },
+ {
+ "epoch": 78.19891008174388,
+ "grad_norm": 1.279130458831787,
+ "learning_rate": 2.3909133699722976e-06,
+ "loss": 0.1107,
+ "step": 28699
+ },
+ {
+ "epoch": 78.2016348773842,
+ "grad_norm": 1.5968025922775269,
+ "learning_rate": 2.390340785012697e-06,
+ "loss": 0.025,
+ "step": 28700
+ },
+ {
+ "epoch": 78.20435967302453,
+ "grad_norm": 1.5692977905273438,
+ "learning_rate": 2.389768259316878e-06,
+ "loss": 0.0315,
+ "step": 28701
+ },
+ {
+ "epoch": 78.20708446866485,
+ "grad_norm": 0.9425067901611328,
+ "learning_rate": 2.389195792889295e-06,
+ "loss": 0.0135,
+ "step": 28702
+ },
+ {
+ "epoch": 78.20980926430518,
+ "grad_norm": 1.1145944595336914,
+ "learning_rate": 2.3886233857344143e-06,
+ "loss": 0.0116,
+ "step": 28703
+ },
+ {
+ "epoch": 78.2125340599455,
+ "grad_norm": 1.381462812423706,
+ "learning_rate": 2.388051037856689e-06,
+ "loss": 0.0112,
+ "step": 28704
+ },
+ {
+ "epoch": 78.21525885558583,
+ "grad_norm": 1.1474213600158691,
+ "learning_rate": 2.3874787492605765e-06,
+ "loss": 0.0139,
+ "step": 28705
+ },
+ {
+ "epoch": 78.21798365122616,
+ "grad_norm": 1.4305301904678345,
+ "learning_rate": 2.3869065199505328e-06,
+ "loss": 0.0327,
+ "step": 28706
+ },
+ {
+ "epoch": 78.22070844686648,
+ "grad_norm": 1.5046948194503784,
+ "learning_rate": 2.3863343499310187e-06,
+ "loss": 0.0207,
+ "step": 28707
+ },
+ {
+ "epoch": 78.22343324250681,
+ "grad_norm": 1.2201722860336304,
+ "learning_rate": 2.385762239206485e-06,
+ "loss": 0.011,
+ "step": 28708
+ },
+ {
+ "epoch": 78.22615803814713,
+ "grad_norm": 0.7149650454521179,
+ "learning_rate": 2.3851901877813923e-06,
+ "loss": 0.0075,
+ "step": 28709
+ },
+ {
+ "epoch": 78.22888283378747,
+ "grad_norm": 1.6609026193618774,
+ "learning_rate": 2.384618195660193e-06,
+ "loss": 0.069,
+ "step": 28710
+ },
+ {
+ "epoch": 78.2316076294278,
+ "grad_norm": 2.5391385555267334,
+ "learning_rate": 2.3840462628473403e-06,
+ "loss": 0.0778,
+ "step": 28711
+ },
+ {
+ "epoch": 78.23433242506812,
+ "grad_norm": 1.0471131801605225,
+ "learning_rate": 2.3834743893472923e-06,
+ "loss": 0.1128,
+ "step": 28712
+ },
+ {
+ "epoch": 78.23705722070845,
+ "grad_norm": 0.9498384594917297,
+ "learning_rate": 2.382902575164501e-06,
+ "loss": 0.0166,
+ "step": 28713
+ },
+ {
+ "epoch": 78.23978201634877,
+ "grad_norm": 0.9129235148429871,
+ "learning_rate": 2.382330820303419e-06,
+ "loss": 0.0113,
+ "step": 28714
+ },
+ {
+ "epoch": 78.2425068119891,
+ "grad_norm": 1.1446059942245483,
+ "learning_rate": 2.381759124768496e-06,
+ "loss": 0.0201,
+ "step": 28715
+ },
+ {
+ "epoch": 78.24523160762942,
+ "grad_norm": 1.5660603046417236,
+ "learning_rate": 2.381187488564192e-06,
+ "loss": 0.0408,
+ "step": 28716
+ },
+ {
+ "epoch": 78.24795640326975,
+ "grad_norm": 1.368711233139038,
+ "learning_rate": 2.3806159116949536e-06,
+ "loss": 0.0528,
+ "step": 28717
+ },
+ {
+ "epoch": 78.25068119891009,
+ "grad_norm": 1.4201512336730957,
+ "learning_rate": 2.380044394165233e-06,
+ "loss": 0.0187,
+ "step": 28718
+ },
+ {
+ "epoch": 78.2534059945504,
+ "grad_norm": 0.8563976287841797,
+ "learning_rate": 2.379472935979479e-06,
+ "loss": 0.011,
+ "step": 28719
+ },
+ {
+ "epoch": 78.25613079019074,
+ "grad_norm": 1.1560932397842407,
+ "learning_rate": 2.3789015371421476e-06,
+ "loss": 0.009,
+ "step": 28720
+ },
+ {
+ "epoch": 78.25885558583106,
+ "grad_norm": 1.5711486339569092,
+ "learning_rate": 2.3783301976576854e-06,
+ "loss": 0.0681,
+ "step": 28721
+ },
+ {
+ "epoch": 78.26158038147139,
+ "grad_norm": 1.6381486654281616,
+ "learning_rate": 2.3777589175305428e-06,
+ "loss": 0.0173,
+ "step": 28722
+ },
+ {
+ "epoch": 78.26430517711172,
+ "grad_norm": 1.1441494226455688,
+ "learning_rate": 2.3771876967651652e-06,
+ "loss": 0.0062,
+ "step": 28723
+ },
+ {
+ "epoch": 78.26702997275204,
+ "grad_norm": 0.8858271241188049,
+ "learning_rate": 2.3766165353660065e-06,
+ "loss": 0.0072,
+ "step": 28724
+ },
+ {
+ "epoch": 78.26975476839237,
+ "grad_norm": 1.47085440158844,
+ "learning_rate": 2.376045433337514e-06,
+ "loss": 0.031,
+ "step": 28725
+ },
+ {
+ "epoch": 78.2724795640327,
+ "grad_norm": 1.0769915580749512,
+ "learning_rate": 2.3754743906841316e-06,
+ "loss": 0.0084,
+ "step": 28726
+ },
+ {
+ "epoch": 78.27520435967303,
+ "grad_norm": 1.504031777381897,
+ "learning_rate": 2.3749034074103084e-06,
+ "loss": 0.0158,
+ "step": 28727
+ },
+ {
+ "epoch": 78.27792915531334,
+ "grad_norm": 1.1036574840545654,
+ "learning_rate": 2.374332483520495e-06,
+ "loss": 0.0191,
+ "step": 28728
+ },
+ {
+ "epoch": 78.28065395095368,
+ "grad_norm": 1.1570953130722046,
+ "learning_rate": 2.373761619019136e-06,
+ "loss": 0.0092,
+ "step": 28729
+ },
+ {
+ "epoch": 78.28337874659401,
+ "grad_norm": 1.2523099184036255,
+ "learning_rate": 2.373190813910674e-06,
+ "loss": 0.067,
+ "step": 28730
+ },
+ {
+ "epoch": 78.28610354223433,
+ "grad_norm": 1.2379200458526611,
+ "learning_rate": 2.372620068199555e-06,
+ "loss": 0.0124,
+ "step": 28731
+ },
+ {
+ "epoch": 78.28882833787466,
+ "grad_norm": 0.9862879514694214,
+ "learning_rate": 2.372049381890228e-06,
+ "loss": 0.009,
+ "step": 28732
+ },
+ {
+ "epoch": 78.29155313351498,
+ "grad_norm": 1.6621119976043701,
+ "learning_rate": 2.3714787549871355e-06,
+ "loss": 0.1282,
+ "step": 28733
+ },
+ {
+ "epoch": 78.29427792915531,
+ "grad_norm": 2.1842401027679443,
+ "learning_rate": 2.3709081874947204e-06,
+ "loss": 0.1212,
+ "step": 28734
+ },
+ {
+ "epoch": 78.29700272479565,
+ "grad_norm": 1.5539965629577637,
+ "learning_rate": 2.3703376794174237e-06,
+ "loss": 0.0318,
+ "step": 28735
+ },
+ {
+ "epoch": 78.29972752043597,
+ "grad_norm": 1.381049633026123,
+ "learning_rate": 2.3697672307596955e-06,
+ "loss": 0.0468,
+ "step": 28736
+ },
+ {
+ "epoch": 78.3024523160763,
+ "grad_norm": 1.4358950853347778,
+ "learning_rate": 2.3691968415259727e-06,
+ "loss": 0.0532,
+ "step": 28737
+ },
+ {
+ "epoch": 78.30517711171662,
+ "grad_norm": 2.2460551261901855,
+ "learning_rate": 2.3686265117207007e-06,
+ "loss": 0.028,
+ "step": 28738
+ },
+ {
+ "epoch": 78.30790190735695,
+ "grad_norm": 1.1320031881332397,
+ "learning_rate": 2.3680562413483167e-06,
+ "loss": 0.0279,
+ "step": 28739
+ },
+ {
+ "epoch": 78.31062670299727,
+ "grad_norm": 1.420065999031067,
+ "learning_rate": 2.3674860304132675e-06,
+ "loss": 0.0114,
+ "step": 28740
+ },
+ {
+ "epoch": 78.3133514986376,
+ "grad_norm": 0.8532713651657104,
+ "learning_rate": 2.366915878919992e-06,
+ "loss": 0.0071,
+ "step": 28741
+ },
+ {
+ "epoch": 78.31607629427793,
+ "grad_norm": 0.9858006834983826,
+ "learning_rate": 2.366345786872929e-06,
+ "loss": 0.013,
+ "step": 28742
+ },
+ {
+ "epoch": 78.31880108991825,
+ "grad_norm": 1.233660340309143,
+ "learning_rate": 2.3657757542765145e-06,
+ "loss": 0.0105,
+ "step": 28743
+ },
+ {
+ "epoch": 78.32152588555859,
+ "grad_norm": 1.0425481796264648,
+ "learning_rate": 2.3652057811351982e-06,
+ "loss": 0.015,
+ "step": 28744
+ },
+ {
+ "epoch": 78.3242506811989,
+ "grad_norm": 1.4932760000228882,
+ "learning_rate": 2.364635867453414e-06,
+ "loss": 0.0466,
+ "step": 28745
+ },
+ {
+ "epoch": 78.32697547683924,
+ "grad_norm": 0.7747379541397095,
+ "learning_rate": 2.3640660132356e-06,
+ "loss": 0.0055,
+ "step": 28746
+ },
+ {
+ "epoch": 78.32970027247957,
+ "grad_norm": 1.2304967641830444,
+ "learning_rate": 2.3634962184861897e-06,
+ "loss": 0.0131,
+ "step": 28747
+ },
+ {
+ "epoch": 78.33242506811989,
+ "grad_norm": 1.701802372932434,
+ "learning_rate": 2.36292648320963e-06,
+ "loss": 0.0192,
+ "step": 28748
+ },
+ {
+ "epoch": 78.33514986376022,
+ "grad_norm": 0.9200157523155212,
+ "learning_rate": 2.362356807410352e-06,
+ "loss": 0.0104,
+ "step": 28749
+ },
+ {
+ "epoch": 78.33787465940054,
+ "grad_norm": 0.9791919589042664,
+ "learning_rate": 2.3617871910927927e-06,
+ "loss": 0.0371,
+ "step": 28750
+ },
+ {
+ "epoch": 78.34059945504087,
+ "grad_norm": 1.0727890729904175,
+ "learning_rate": 2.361217634261387e-06,
+ "loss": 0.047,
+ "step": 28751
+ },
+ {
+ "epoch": 78.34332425068119,
+ "grad_norm": 1.2894471883773804,
+ "learning_rate": 2.360648136920575e-06,
+ "loss": 0.016,
+ "step": 28752
+ },
+ {
+ "epoch": 78.34604904632153,
+ "grad_norm": 1.653097152709961,
+ "learning_rate": 2.360078699074789e-06,
+ "loss": 0.0197,
+ "step": 28753
+ },
+ {
+ "epoch": 78.34877384196186,
+ "grad_norm": 1.929317831993103,
+ "learning_rate": 2.3595093207284657e-06,
+ "loss": 0.1652,
+ "step": 28754
+ },
+ {
+ "epoch": 78.35149863760218,
+ "grad_norm": 2.124781847000122,
+ "learning_rate": 2.3589400018860343e-06,
+ "loss": 0.0427,
+ "step": 28755
+ },
+ {
+ "epoch": 78.35422343324251,
+ "grad_norm": 1.305857539176941,
+ "learning_rate": 2.3583707425519344e-06,
+ "loss": 0.0107,
+ "step": 28756
+ },
+ {
+ "epoch": 78.35694822888283,
+ "grad_norm": 1.1359374523162842,
+ "learning_rate": 2.3578015427305977e-06,
+ "loss": 0.0144,
+ "step": 28757
+ },
+ {
+ "epoch": 78.35967302452316,
+ "grad_norm": 0.7304384708404541,
+ "learning_rate": 2.357232402426457e-06,
+ "loss": 0.0104,
+ "step": 28758
+ },
+ {
+ "epoch": 78.3623978201635,
+ "grad_norm": 1.3150269985198975,
+ "learning_rate": 2.3566633216439416e-06,
+ "loss": 0.0739,
+ "step": 28759
+ },
+ {
+ "epoch": 78.36512261580381,
+ "grad_norm": 0.7757869362831116,
+ "learning_rate": 2.356094300387486e-06,
+ "loss": 0.011,
+ "step": 28760
+ },
+ {
+ "epoch": 78.36784741144415,
+ "grad_norm": 1.0649356842041016,
+ "learning_rate": 2.355525338661524e-06,
+ "loss": 0.0112,
+ "step": 28761
+ },
+ {
+ "epoch": 78.37057220708446,
+ "grad_norm": 1.369959831237793,
+ "learning_rate": 2.3549564364704856e-06,
+ "loss": 0.0177,
+ "step": 28762
+ },
+ {
+ "epoch": 78.3732970027248,
+ "grad_norm": 1.3365883827209473,
+ "learning_rate": 2.3543875938187997e-06,
+ "loss": 0.0451,
+ "step": 28763
+ },
+ {
+ "epoch": 78.37602179836512,
+ "grad_norm": 1.0648020505905151,
+ "learning_rate": 2.353818810710895e-06,
+ "loss": 0.0797,
+ "step": 28764
+ },
+ {
+ "epoch": 78.37874659400545,
+ "grad_norm": 1.0715960264205933,
+ "learning_rate": 2.3532500871512054e-06,
+ "loss": 0.011,
+ "step": 28765
+ },
+ {
+ "epoch": 78.38147138964578,
+ "grad_norm": 0.8343406319618225,
+ "learning_rate": 2.3526814231441586e-06,
+ "loss": 0.0082,
+ "step": 28766
+ },
+ {
+ "epoch": 78.3841961852861,
+ "grad_norm": 2.236469268798828,
+ "learning_rate": 2.35211281869418e-06,
+ "loss": 0.0212,
+ "step": 28767
+ },
+ {
+ "epoch": 78.38692098092643,
+ "grad_norm": 1.335731863975525,
+ "learning_rate": 2.3515442738057027e-06,
+ "loss": 0.0135,
+ "step": 28768
+ },
+ {
+ "epoch": 78.38964577656675,
+ "grad_norm": 1.35103440284729,
+ "learning_rate": 2.350975788483153e-06,
+ "loss": 0.0194,
+ "step": 28769
+ },
+ {
+ "epoch": 78.39237057220708,
+ "grad_norm": 1.19074285030365,
+ "learning_rate": 2.3504073627309575e-06,
+ "loss": 0.0117,
+ "step": 28770
+ },
+ {
+ "epoch": 78.39509536784742,
+ "grad_norm": 1.1494446992874146,
+ "learning_rate": 2.3498389965535406e-06,
+ "loss": 0.0916,
+ "step": 28771
+ },
+ {
+ "epoch": 78.39782016348774,
+ "grad_norm": 1.7297815084457397,
+ "learning_rate": 2.3492706899553354e-06,
+ "loss": 0.0651,
+ "step": 28772
+ },
+ {
+ "epoch": 78.40054495912807,
+ "grad_norm": 1.3030585050582886,
+ "learning_rate": 2.3487024429407625e-06,
+ "loss": 0.0794,
+ "step": 28773
+ },
+ {
+ "epoch": 78.40326975476839,
+ "grad_norm": 1.6087900400161743,
+ "learning_rate": 2.3481342555142484e-06,
+ "loss": 0.0291,
+ "step": 28774
+ },
+ {
+ "epoch": 78.40599455040872,
+ "grad_norm": 0.9299991130828857,
+ "learning_rate": 2.3475661276802164e-06,
+ "loss": 0.0081,
+ "step": 28775
+ },
+ {
+ "epoch": 78.40871934604904,
+ "grad_norm": 1.3829702138900757,
+ "learning_rate": 2.346998059443095e-06,
+ "loss": 0.0187,
+ "step": 28776
+ },
+ {
+ "epoch": 78.41144414168937,
+ "grad_norm": 1.4929834604263306,
+ "learning_rate": 2.3464300508073046e-06,
+ "loss": 0.0882,
+ "step": 28777
+ },
+ {
+ "epoch": 78.4141689373297,
+ "grad_norm": 1.171833872795105,
+ "learning_rate": 2.3458621017772733e-06,
+ "loss": 0.046,
+ "step": 28778
+ },
+ {
+ "epoch": 78.41689373297002,
+ "grad_norm": 1.1838059425354004,
+ "learning_rate": 2.3452942123574208e-06,
+ "loss": 0.0429,
+ "step": 28779
+ },
+ {
+ "epoch": 78.41961852861036,
+ "grad_norm": 1.0845739841461182,
+ "learning_rate": 2.344726382552168e-06,
+ "loss": 0.0211,
+ "step": 28780
+ },
+ {
+ "epoch": 78.42234332425068,
+ "grad_norm": 1.8859978914260864,
+ "learning_rate": 2.3441586123659434e-06,
+ "loss": 0.2686,
+ "step": 28781
+ },
+ {
+ "epoch": 78.42506811989101,
+ "grad_norm": 1.26189124584198,
+ "learning_rate": 2.3435909018031645e-06,
+ "loss": 0.0102,
+ "step": 28782
+ },
+ {
+ "epoch": 78.42779291553134,
+ "grad_norm": 1.180721640586853,
+ "learning_rate": 2.3430232508682525e-06,
+ "loss": 0.0167,
+ "step": 28783
+ },
+ {
+ "epoch": 78.43051771117166,
+ "grad_norm": 0.9322836399078369,
+ "learning_rate": 2.3424556595656276e-06,
+ "loss": 0.0073,
+ "step": 28784
+ },
+ {
+ "epoch": 78.433242506812,
+ "grad_norm": 0.5605153441429138,
+ "learning_rate": 2.341888127899714e-06,
+ "loss": 0.0049,
+ "step": 28785
+ },
+ {
+ "epoch": 78.43596730245231,
+ "grad_norm": 1.282295823097229,
+ "learning_rate": 2.341320655874929e-06,
+ "loss": 0.0149,
+ "step": 28786
+ },
+ {
+ "epoch": 78.43869209809264,
+ "grad_norm": 2.2128491401672363,
+ "learning_rate": 2.3407532434956913e-06,
+ "loss": 0.0218,
+ "step": 28787
+ },
+ {
+ "epoch": 78.44141689373296,
+ "grad_norm": 1.4866783618927002,
+ "learning_rate": 2.34018589076642e-06,
+ "loss": 0.0646,
+ "step": 28788
+ },
+ {
+ "epoch": 78.4441416893733,
+ "grad_norm": 1.3248445987701416,
+ "learning_rate": 2.3396185976915355e-06,
+ "loss": 0.0069,
+ "step": 28789
+ },
+ {
+ "epoch": 78.44686648501363,
+ "grad_norm": 1.4655470848083496,
+ "learning_rate": 2.3390513642754555e-06,
+ "loss": 0.0402,
+ "step": 28790
+ },
+ {
+ "epoch": 78.44959128065395,
+ "grad_norm": 0.97535240650177,
+ "learning_rate": 2.338484190522594e-06,
+ "loss": 0.0126,
+ "step": 28791
+ },
+ {
+ "epoch": 78.45231607629428,
+ "grad_norm": 1.3333735466003418,
+ "learning_rate": 2.3379170764373737e-06,
+ "loss": 0.0255,
+ "step": 28792
+ },
+ {
+ "epoch": 78.4550408719346,
+ "grad_norm": 0.9310436248779297,
+ "learning_rate": 2.3373500220242095e-06,
+ "loss": 0.0106,
+ "step": 28793
+ },
+ {
+ "epoch": 78.45776566757493,
+ "grad_norm": 1.3890258073806763,
+ "learning_rate": 2.3367830272875136e-06,
+ "loss": 0.047,
+ "step": 28794
+ },
+ {
+ "epoch": 78.46049046321527,
+ "grad_norm": 1.1955673694610596,
+ "learning_rate": 2.336216092231707e-06,
+ "loss": 0.0451,
+ "step": 28795
+ },
+ {
+ "epoch": 78.46321525885558,
+ "grad_norm": 1.145203948020935,
+ "learning_rate": 2.3356492168612e-06,
+ "loss": 0.0644,
+ "step": 28796
+ },
+ {
+ "epoch": 78.46594005449592,
+ "grad_norm": 1.1864780187606812,
+ "learning_rate": 2.335082401180413e-06,
+ "loss": 0.0146,
+ "step": 28797
+ },
+ {
+ "epoch": 78.46866485013624,
+ "grad_norm": 0.8296866416931152,
+ "learning_rate": 2.3345156451937577e-06,
+ "loss": 0.0081,
+ "step": 28798
+ },
+ {
+ "epoch": 78.47138964577657,
+ "grad_norm": 1.4621171951293945,
+ "learning_rate": 2.333948948905649e-06,
+ "loss": 0.0162,
+ "step": 28799
+ },
+ {
+ "epoch": 78.47411444141689,
+ "grad_norm": 1.1291446685791016,
+ "learning_rate": 2.3333823123204946e-06,
+ "loss": 0.0267,
+ "step": 28800
+ },
+ {
+ "epoch": 78.47683923705722,
+ "grad_norm": 1.7875186204910278,
+ "learning_rate": 2.332815735442716e-06,
+ "loss": 0.0542,
+ "step": 28801
+ },
+ {
+ "epoch": 78.47956403269755,
+ "grad_norm": 1.428162693977356,
+ "learning_rate": 2.332249218276721e-06,
+ "loss": 0.1495,
+ "step": 28802
+ },
+ {
+ "epoch": 78.48228882833787,
+ "grad_norm": 0.7631807923316956,
+ "learning_rate": 2.3316827608269233e-06,
+ "loss": 0.008,
+ "step": 28803
+ },
+ {
+ "epoch": 78.4850136239782,
+ "grad_norm": 0.9667528867721558,
+ "learning_rate": 2.3311163630977296e-06,
+ "loss": 0.0094,
+ "step": 28804
+ },
+ {
+ "epoch": 78.48773841961852,
+ "grad_norm": 2.044530153274536,
+ "learning_rate": 2.3305500250935574e-06,
+ "loss": 0.0352,
+ "step": 28805
+ },
+ {
+ "epoch": 78.49046321525886,
+ "grad_norm": 6.567902088165283,
+ "learning_rate": 2.3299837468188156e-06,
+ "loss": 0.026,
+ "step": 28806
+ },
+ {
+ "epoch": 78.49318801089919,
+ "grad_norm": 1.169472336769104,
+ "learning_rate": 2.329417528277914e-06,
+ "loss": 0.011,
+ "step": 28807
+ },
+ {
+ "epoch": 78.49591280653951,
+ "grad_norm": 1.433454990386963,
+ "learning_rate": 2.328851369475258e-06,
+ "loss": 0.0357,
+ "step": 28808
+ },
+ {
+ "epoch": 78.49863760217984,
+ "grad_norm": 1.336125135421753,
+ "learning_rate": 2.3282852704152638e-06,
+ "loss": 0.0124,
+ "step": 28809
+ },
+ {
+ "epoch": 78.50136239782016,
+ "grad_norm": 1.2574992179870605,
+ "learning_rate": 2.3277192311023366e-06,
+ "loss": 0.0296,
+ "step": 28810
+ },
+ {
+ "epoch": 78.50408719346049,
+ "grad_norm": 1.3061425685882568,
+ "learning_rate": 2.3271532515408825e-06,
+ "loss": 0.022,
+ "step": 28811
+ },
+ {
+ "epoch": 78.50681198910081,
+ "grad_norm": 2.125537872314453,
+ "learning_rate": 2.326587331735313e-06,
+ "loss": 0.1682,
+ "step": 28812
+ },
+ {
+ "epoch": 78.50953678474114,
+ "grad_norm": 1.7484337091445923,
+ "learning_rate": 2.3260214716900353e-06,
+ "loss": 0.0494,
+ "step": 28813
+ },
+ {
+ "epoch": 78.51226158038148,
+ "grad_norm": 1.3639811277389526,
+ "learning_rate": 2.325455671409457e-06,
+ "loss": 0.0633,
+ "step": 28814
+ },
+ {
+ "epoch": 78.5149863760218,
+ "grad_norm": 1.2075384855270386,
+ "learning_rate": 2.324889930897982e-06,
+ "loss": 0.0754,
+ "step": 28815
+ },
+ {
+ "epoch": 78.51771117166213,
+ "grad_norm": 1.734252691268921,
+ "learning_rate": 2.324324250160015e-06,
+ "loss": 0.1731,
+ "step": 28816
+ },
+ {
+ "epoch": 78.52043596730245,
+ "grad_norm": 0.9992340803146362,
+ "learning_rate": 2.3237586291999657e-06,
+ "loss": 0.0115,
+ "step": 28817
+ },
+ {
+ "epoch": 78.52316076294278,
+ "grad_norm": 1.5954134464263916,
+ "learning_rate": 2.323193068022238e-06,
+ "loss": 0.0274,
+ "step": 28818
+ },
+ {
+ "epoch": 78.52588555858311,
+ "grad_norm": 1.412981629371643,
+ "learning_rate": 2.3226275666312346e-06,
+ "loss": 0.0393,
+ "step": 28819
+ },
+ {
+ "epoch": 78.52861035422343,
+ "grad_norm": 1.2547751665115356,
+ "learning_rate": 2.322062125031359e-06,
+ "loss": 0.0439,
+ "step": 28820
+ },
+ {
+ "epoch": 78.53133514986376,
+ "grad_norm": 1.156510829925537,
+ "learning_rate": 2.3214967432270187e-06,
+ "loss": 0.0159,
+ "step": 28821
+ },
+ {
+ "epoch": 78.53405994550408,
+ "grad_norm": 1.8707702159881592,
+ "learning_rate": 2.3209314212226143e-06,
+ "loss": 0.0326,
+ "step": 28822
+ },
+ {
+ "epoch": 78.53678474114442,
+ "grad_norm": 1.3266671895980835,
+ "learning_rate": 2.320366159022549e-06,
+ "loss": 0.0211,
+ "step": 28823
+ },
+ {
+ "epoch": 78.53950953678473,
+ "grad_norm": 1.6427714824676514,
+ "learning_rate": 2.319800956631222e-06,
+ "loss": 0.1295,
+ "step": 28824
+ },
+ {
+ "epoch": 78.54223433242507,
+ "grad_norm": 1.410858154296875,
+ "learning_rate": 2.3192358140530415e-06,
+ "loss": 0.0684,
+ "step": 28825
+ },
+ {
+ "epoch": 78.5449591280654,
+ "grad_norm": 0.9732717871665955,
+ "learning_rate": 2.3186707312924053e-06,
+ "loss": 0.007,
+ "step": 28826
+ },
+ {
+ "epoch": 78.54768392370572,
+ "grad_norm": 1.286903738975525,
+ "learning_rate": 2.3181057083537127e-06,
+ "loss": 0.0624,
+ "step": 28827
+ },
+ {
+ "epoch": 78.55040871934605,
+ "grad_norm": 1.8721026182174683,
+ "learning_rate": 2.3175407452413647e-06,
+ "loss": 0.0342,
+ "step": 28828
+ },
+ {
+ "epoch": 78.55313351498637,
+ "grad_norm": 1.6674082279205322,
+ "learning_rate": 2.316975841959761e-06,
+ "loss": 0.0172,
+ "step": 28829
+ },
+ {
+ "epoch": 78.5558583106267,
+ "grad_norm": 1.3566277027130127,
+ "learning_rate": 2.3164109985133043e-06,
+ "loss": 0.0171,
+ "step": 28830
+ },
+ {
+ "epoch": 78.55858310626704,
+ "grad_norm": 0.8031501770019531,
+ "learning_rate": 2.315846214906393e-06,
+ "loss": 0.007,
+ "step": 28831
+ },
+ {
+ "epoch": 78.56130790190736,
+ "grad_norm": 1.3548743724822998,
+ "learning_rate": 2.315281491143421e-06,
+ "loss": 0.1313,
+ "step": 28832
+ },
+ {
+ "epoch": 78.56403269754769,
+ "grad_norm": 1.0401756763458252,
+ "learning_rate": 2.314716827228791e-06,
+ "loss": 0.0871,
+ "step": 28833
+ },
+ {
+ "epoch": 78.566757493188,
+ "grad_norm": 0.8505842089653015,
+ "learning_rate": 2.3141522231669e-06,
+ "loss": 0.0109,
+ "step": 28834
+ },
+ {
+ "epoch": 78.56948228882834,
+ "grad_norm": 1.0356285572052002,
+ "learning_rate": 2.3135876789621436e-06,
+ "loss": 0.0257,
+ "step": 28835
+ },
+ {
+ "epoch": 78.57220708446866,
+ "grad_norm": 1.868502140045166,
+ "learning_rate": 2.313023194618916e-06,
+ "loss": 0.1556,
+ "step": 28836
+ },
+ {
+ "epoch": 78.57493188010899,
+ "grad_norm": 0.88411945104599,
+ "learning_rate": 2.31245877014162e-06,
+ "loss": 0.0106,
+ "step": 28837
+ },
+ {
+ "epoch": 78.57765667574932,
+ "grad_norm": 0.5505541563034058,
+ "learning_rate": 2.3118944055346482e-06,
+ "loss": 0.0057,
+ "step": 28838
+ },
+ {
+ "epoch": 78.58038147138964,
+ "grad_norm": 1.0303598642349243,
+ "learning_rate": 2.3113301008023936e-06,
+ "loss": 0.0196,
+ "step": 28839
+ },
+ {
+ "epoch": 78.58310626702998,
+ "grad_norm": 1.7676630020141602,
+ "learning_rate": 2.3107658559492506e-06,
+ "loss": 0.0277,
+ "step": 28840
+ },
+ {
+ "epoch": 78.5858310626703,
+ "grad_norm": 2.368727445602417,
+ "learning_rate": 2.3102016709796193e-06,
+ "loss": 0.0157,
+ "step": 28841
+ },
+ {
+ "epoch": 78.58855585831063,
+ "grad_norm": 1.6289421319961548,
+ "learning_rate": 2.3096375458978892e-06,
+ "loss": 0.0264,
+ "step": 28842
+ },
+ {
+ "epoch": 78.59128065395096,
+ "grad_norm": 1.2216933965682983,
+ "learning_rate": 2.3090734807084545e-06,
+ "loss": 0.0151,
+ "step": 28843
+ },
+ {
+ "epoch": 78.59400544959128,
+ "grad_norm": 1.0389516353607178,
+ "learning_rate": 2.3085094754157045e-06,
+ "loss": 0.0067,
+ "step": 28844
+ },
+ {
+ "epoch": 78.59673024523161,
+ "grad_norm": 0.7801070809364319,
+ "learning_rate": 2.3079455300240394e-06,
+ "loss": 0.0097,
+ "step": 28845
+ },
+ {
+ "epoch": 78.59945504087193,
+ "grad_norm": 1.525168776512146,
+ "learning_rate": 2.3073816445378427e-06,
+ "loss": 0.0191,
+ "step": 28846
+ },
+ {
+ "epoch": 78.60217983651226,
+ "grad_norm": 1.0332738161087036,
+ "learning_rate": 2.3068178189615142e-06,
+ "loss": 0.008,
+ "step": 28847
+ },
+ {
+ "epoch": 78.60490463215258,
+ "grad_norm": 1.3900439739227295,
+ "learning_rate": 2.306254053299439e-06,
+ "loss": 0.0229,
+ "step": 28848
+ },
+ {
+ "epoch": 78.60762942779292,
+ "grad_norm": 4.495482921600342,
+ "learning_rate": 2.305690347556009e-06,
+ "loss": 0.0175,
+ "step": 28849
+ },
+ {
+ "epoch": 78.61035422343325,
+ "grad_norm": 0.9766920804977417,
+ "learning_rate": 2.305126701735617e-06,
+ "loss": 0.0208,
+ "step": 28850
+ },
+ {
+ "epoch": 78.61307901907357,
+ "grad_norm": 1.474504828453064,
+ "learning_rate": 2.304563115842651e-06,
+ "loss": 0.0209,
+ "step": 28851
+ },
+ {
+ "epoch": 78.6158038147139,
+ "grad_norm": 1.6041401624679565,
+ "learning_rate": 2.303999589881496e-06,
+ "loss": 0.0836,
+ "step": 28852
+ },
+ {
+ "epoch": 78.61852861035422,
+ "grad_norm": 1.0266506671905518,
+ "learning_rate": 2.3034361238565474e-06,
+ "loss": 0.008,
+ "step": 28853
+ },
+ {
+ "epoch": 78.62125340599455,
+ "grad_norm": 0.6486322283744812,
+ "learning_rate": 2.302872717772191e-06,
+ "loss": 0.0055,
+ "step": 28854
+ },
+ {
+ "epoch": 78.62397820163488,
+ "grad_norm": 1.1400014162063599,
+ "learning_rate": 2.302309371632814e-06,
+ "loss": 0.0121,
+ "step": 28855
+ },
+ {
+ "epoch": 78.6267029972752,
+ "grad_norm": 1.5153236389160156,
+ "learning_rate": 2.301746085442801e-06,
+ "loss": 0.0672,
+ "step": 28856
+ },
+ {
+ "epoch": 78.62942779291554,
+ "grad_norm": 1.4883477687835693,
+ "learning_rate": 2.301182859206544e-06,
+ "loss": 0.029,
+ "step": 28857
+ },
+ {
+ "epoch": 78.63215258855585,
+ "grad_norm": 1.1234055757522583,
+ "learning_rate": 2.3006196929284276e-06,
+ "loss": 0.0085,
+ "step": 28858
+ },
+ {
+ "epoch": 78.63487738419619,
+ "grad_norm": 1.491485834121704,
+ "learning_rate": 2.3000565866128377e-06,
+ "loss": 0.0841,
+ "step": 28859
+ },
+ {
+ "epoch": 78.6376021798365,
+ "grad_norm": 1.450780987739563,
+ "learning_rate": 2.299493540264156e-06,
+ "loss": 0.0693,
+ "step": 28860
+ },
+ {
+ "epoch": 78.64032697547684,
+ "grad_norm": 1.3937666416168213,
+ "learning_rate": 2.2989305538867735e-06,
+ "loss": 0.0712,
+ "step": 28861
+ },
+ {
+ "epoch": 78.64305177111717,
+ "grad_norm": 0.9559317827224731,
+ "learning_rate": 2.298367627485072e-06,
+ "loss": 0.0141,
+ "step": 28862
+ },
+ {
+ "epoch": 78.64577656675749,
+ "grad_norm": 0.7494531273841858,
+ "learning_rate": 2.2978047610634335e-06,
+ "loss": 0.0112,
+ "step": 28863
+ },
+ {
+ "epoch": 78.64850136239782,
+ "grad_norm": 1.2708728313446045,
+ "learning_rate": 2.297241954626246e-06,
+ "loss": 0.047,
+ "step": 28864
+ },
+ {
+ "epoch": 78.65122615803814,
+ "grad_norm": 0.8313508629798889,
+ "learning_rate": 2.2966792081778866e-06,
+ "loss": 0.0057,
+ "step": 28865
+ },
+ {
+ "epoch": 78.65395095367847,
+ "grad_norm": 1.2288633584976196,
+ "learning_rate": 2.2961165217227453e-06,
+ "loss": 0.0197,
+ "step": 28866
+ },
+ {
+ "epoch": 78.65667574931881,
+ "grad_norm": 1.4859033823013306,
+ "learning_rate": 2.2955538952652e-06,
+ "loss": 0.0341,
+ "step": 28867
+ },
+ {
+ "epoch": 78.65940054495913,
+ "grad_norm": 1.0702627897262573,
+ "learning_rate": 2.294991328809634e-06,
+ "loss": 0.0455,
+ "step": 28868
+ },
+ {
+ "epoch": 78.66212534059946,
+ "grad_norm": 1.8875101804733276,
+ "learning_rate": 2.2944288223604237e-06,
+ "loss": 0.1525,
+ "step": 28869
+ },
+ {
+ "epoch": 78.66485013623978,
+ "grad_norm": 1.2802953720092773,
+ "learning_rate": 2.293866375921957e-06,
+ "loss": 0.0115,
+ "step": 28870
+ },
+ {
+ "epoch": 78.66757493188011,
+ "grad_norm": 1.8941344022750854,
+ "learning_rate": 2.2933039894986107e-06,
+ "loss": 0.0287,
+ "step": 28871
+ },
+ {
+ "epoch": 78.67029972752043,
+ "grad_norm": 0.9614849090576172,
+ "learning_rate": 2.2927416630947653e-06,
+ "loss": 0.0081,
+ "step": 28872
+ },
+ {
+ "epoch": 78.67302452316076,
+ "grad_norm": 0.520854651927948,
+ "learning_rate": 2.2921793967147963e-06,
+ "loss": 0.006,
+ "step": 28873
+ },
+ {
+ "epoch": 78.6757493188011,
+ "grad_norm": 1.4203989505767822,
+ "learning_rate": 2.2916171903630892e-06,
+ "loss": 0.0355,
+ "step": 28874
+ },
+ {
+ "epoch": 78.67847411444141,
+ "grad_norm": 1.0675092935562134,
+ "learning_rate": 2.2910550440440194e-06,
+ "loss": 0.0167,
+ "step": 28875
+ },
+ {
+ "epoch": 78.68119891008175,
+ "grad_norm": 1.1139178276062012,
+ "learning_rate": 2.290492957761963e-06,
+ "loss": 0.0212,
+ "step": 28876
+ },
+ {
+ "epoch": 78.68392370572207,
+ "grad_norm": 0.9674838185310364,
+ "learning_rate": 2.2899309315213005e-06,
+ "loss": 0.0122,
+ "step": 28877
+ },
+ {
+ "epoch": 78.6866485013624,
+ "grad_norm": 1.3287419080734253,
+ "learning_rate": 2.2893689653264084e-06,
+ "loss": 0.0086,
+ "step": 28878
+ },
+ {
+ "epoch": 78.68937329700273,
+ "grad_norm": 1.4632946252822876,
+ "learning_rate": 2.2888070591816626e-06,
+ "loss": 0.0271,
+ "step": 28879
+ },
+ {
+ "epoch": 78.69209809264305,
+ "grad_norm": 1.2715636491775513,
+ "learning_rate": 2.2882452130914367e-06,
+ "loss": 0.0139,
+ "step": 28880
+ },
+ {
+ "epoch": 78.69482288828338,
+ "grad_norm": 1.194767951965332,
+ "learning_rate": 2.287683427060109e-06,
+ "loss": 0.0196,
+ "step": 28881
+ },
+ {
+ "epoch": 78.6975476839237,
+ "grad_norm": 0.805637776851654,
+ "learning_rate": 2.2871217010920567e-06,
+ "loss": 0.0086,
+ "step": 28882
+ },
+ {
+ "epoch": 78.70027247956403,
+ "grad_norm": 1.6509860754013062,
+ "learning_rate": 2.286560035191653e-06,
+ "loss": 0.0516,
+ "step": 28883
+ },
+ {
+ "epoch": 78.70299727520435,
+ "grad_norm": 1.0889664888381958,
+ "learning_rate": 2.2859984293632708e-06,
+ "loss": 0.0087,
+ "step": 28884
+ },
+ {
+ "epoch": 78.70572207084469,
+ "grad_norm": 2.5832245349884033,
+ "learning_rate": 2.285436883611282e-06,
+ "loss": 0.0243,
+ "step": 28885
+ },
+ {
+ "epoch": 78.70844686648502,
+ "grad_norm": 1.596606969833374,
+ "learning_rate": 2.284875397940065e-06,
+ "loss": 0.0423,
+ "step": 28886
+ },
+ {
+ "epoch": 78.71117166212534,
+ "grad_norm": 0.7888264656066895,
+ "learning_rate": 2.284313972353991e-06,
+ "loss": 0.0098,
+ "step": 28887
+ },
+ {
+ "epoch": 78.71389645776567,
+ "grad_norm": 1.1381028890609741,
+ "learning_rate": 2.28375260685743e-06,
+ "loss": 0.0114,
+ "step": 28888
+ },
+ {
+ "epoch": 78.71662125340599,
+ "grad_norm": 0.9487293362617493,
+ "learning_rate": 2.2831913014547535e-06,
+ "loss": 0.0125,
+ "step": 28889
+ },
+ {
+ "epoch": 78.71934604904632,
+ "grad_norm": 0.6364315748214722,
+ "learning_rate": 2.282630056150338e-06,
+ "loss": 0.0071,
+ "step": 28890
+ },
+ {
+ "epoch": 78.72207084468666,
+ "grad_norm": 1.3301420211791992,
+ "learning_rate": 2.2820688709485504e-06,
+ "loss": 0.0825,
+ "step": 28891
+ },
+ {
+ "epoch": 78.72479564032697,
+ "grad_norm": 2.3045144081115723,
+ "learning_rate": 2.2815077458537628e-06,
+ "loss": 0.016,
+ "step": 28892
+ },
+ {
+ "epoch": 78.7275204359673,
+ "grad_norm": 1.6152242422103882,
+ "learning_rate": 2.2809466808703416e-06,
+ "loss": 0.024,
+ "step": 28893
+ },
+ {
+ "epoch": 78.73024523160763,
+ "grad_norm": 1.2252408266067505,
+ "learning_rate": 2.280385676002661e-06,
+ "loss": 0.02,
+ "step": 28894
+ },
+ {
+ "epoch": 78.73297002724796,
+ "grad_norm": 1.4443467855453491,
+ "learning_rate": 2.2798247312550893e-06,
+ "loss": 0.0168,
+ "step": 28895
+ },
+ {
+ "epoch": 78.73569482288828,
+ "grad_norm": 1.1882686614990234,
+ "learning_rate": 2.279263846631994e-06,
+ "loss": 0.0188,
+ "step": 28896
+ },
+ {
+ "epoch": 78.73841961852861,
+ "grad_norm": 0.5348740816116333,
+ "learning_rate": 2.2787030221377403e-06,
+ "loss": 0.0077,
+ "step": 28897
+ },
+ {
+ "epoch": 78.74114441416894,
+ "grad_norm": 1.3443660736083984,
+ "learning_rate": 2.278142257776699e-06,
+ "loss": 0.0105,
+ "step": 28898
+ },
+ {
+ "epoch": 78.74386920980926,
+ "grad_norm": 0.950590193271637,
+ "learning_rate": 2.27758155355324e-06,
+ "loss": 0.0118,
+ "step": 28899
+ },
+ {
+ "epoch": 78.7465940054496,
+ "grad_norm": 1.5659716129302979,
+ "learning_rate": 2.277020909471728e-06,
+ "loss": 0.0183,
+ "step": 28900
+ },
+ {
+ "epoch": 78.74931880108991,
+ "grad_norm": 1.4661813974380493,
+ "learning_rate": 2.2764603255365248e-06,
+ "loss": 0.0301,
+ "step": 28901
+ },
+ {
+ "epoch": 78.75204359673025,
+ "grad_norm": 1.3192716836929321,
+ "learning_rate": 2.2758998017520027e-06,
+ "loss": 0.0313,
+ "step": 28902
+ },
+ {
+ "epoch": 78.75476839237058,
+ "grad_norm": 1.1705331802368164,
+ "learning_rate": 2.2753393381225252e-06,
+ "loss": 0.0899,
+ "step": 28903
+ },
+ {
+ "epoch": 78.7574931880109,
+ "grad_norm": 1.1819164752960205,
+ "learning_rate": 2.274778934652456e-06,
+ "loss": 0.0375,
+ "step": 28904
+ },
+ {
+ "epoch": 78.76021798365123,
+ "grad_norm": 0.8967912793159485,
+ "learning_rate": 2.274218591346157e-06,
+ "loss": 0.0112,
+ "step": 28905
+ },
+ {
+ "epoch": 78.76294277929155,
+ "grad_norm": 1.73661470413208,
+ "learning_rate": 2.2736583082079977e-06,
+ "loss": 0.0173,
+ "step": 28906
+ },
+ {
+ "epoch": 78.76566757493188,
+ "grad_norm": 1.7975637912750244,
+ "learning_rate": 2.273098085242339e-06,
+ "loss": 0.0605,
+ "step": 28907
+ },
+ {
+ "epoch": 78.7683923705722,
+ "grad_norm": 1.7741202116012573,
+ "learning_rate": 2.2725379224535438e-06,
+ "loss": 0.0186,
+ "step": 28908
+ },
+ {
+ "epoch": 78.77111716621253,
+ "grad_norm": 1.3489271402359009,
+ "learning_rate": 2.271977819845971e-06,
+ "loss": 0.0236,
+ "step": 28909
+ },
+ {
+ "epoch": 78.77384196185287,
+ "grad_norm": 1.8662540912628174,
+ "learning_rate": 2.27141777742399e-06,
+ "loss": 0.0324,
+ "step": 28910
+ },
+ {
+ "epoch": 78.77656675749319,
+ "grad_norm": 1.3593990802764893,
+ "learning_rate": 2.2708577951919585e-06,
+ "loss": 0.0947,
+ "step": 28911
+ },
+ {
+ "epoch": 78.77929155313352,
+ "grad_norm": 0.7788009643554688,
+ "learning_rate": 2.270297873154237e-06,
+ "loss": 0.008,
+ "step": 28912
+ },
+ {
+ "epoch": 78.78201634877384,
+ "grad_norm": 0.8813377618789673,
+ "learning_rate": 2.2697380113151848e-06,
+ "loss": 0.0098,
+ "step": 28913
+ },
+ {
+ "epoch": 78.78474114441417,
+ "grad_norm": 0.6486082673072815,
+ "learning_rate": 2.269178209679166e-06,
+ "loss": 0.0046,
+ "step": 28914
+ },
+ {
+ "epoch": 78.7874659400545,
+ "grad_norm": 0.6653639674186707,
+ "learning_rate": 2.2686184682505365e-06,
+ "loss": 0.0078,
+ "step": 28915
+ },
+ {
+ "epoch": 78.79019073569482,
+ "grad_norm": 1.124166488647461,
+ "learning_rate": 2.2680587870336603e-06,
+ "loss": 0.0691,
+ "step": 28916
+ },
+ {
+ "epoch": 78.79291553133515,
+ "grad_norm": 0.9699380397796631,
+ "learning_rate": 2.2674991660328903e-06,
+ "loss": 0.0138,
+ "step": 28917
+ },
+ {
+ "epoch": 78.79564032697547,
+ "grad_norm": 1.4146751165390015,
+ "learning_rate": 2.2669396052525914e-06,
+ "loss": 0.0072,
+ "step": 28918
+ },
+ {
+ "epoch": 78.7983651226158,
+ "grad_norm": 1.8181840181350708,
+ "learning_rate": 2.266380104697117e-06,
+ "loss": 0.2666,
+ "step": 28919
+ },
+ {
+ "epoch": 78.80108991825612,
+ "grad_norm": 2.0094406604766846,
+ "learning_rate": 2.265820664370826e-06,
+ "loss": 0.0911,
+ "step": 28920
+ },
+ {
+ "epoch": 78.80381471389646,
+ "grad_norm": 0.9393265843391418,
+ "learning_rate": 2.265261284278072e-06,
+ "loss": 0.0101,
+ "step": 28921
+ },
+ {
+ "epoch": 78.80653950953679,
+ "grad_norm": 1.3905245065689087,
+ "learning_rate": 2.264701964423217e-06,
+ "loss": 0.0196,
+ "step": 28922
+ },
+ {
+ "epoch": 78.80926430517711,
+ "grad_norm": 1.2685353755950928,
+ "learning_rate": 2.2641427048106135e-06,
+ "loss": 0.0254,
+ "step": 28923
+ },
+ {
+ "epoch": 78.81198910081744,
+ "grad_norm": 1.560920238494873,
+ "learning_rate": 2.2635835054446186e-06,
+ "loss": 0.0296,
+ "step": 28924
+ },
+ {
+ "epoch": 78.81471389645776,
+ "grad_norm": 1.5693854093551636,
+ "learning_rate": 2.2630243663295825e-06,
+ "loss": 0.0204,
+ "step": 28925
+ },
+ {
+ "epoch": 78.8174386920981,
+ "grad_norm": 0.7532402276992798,
+ "learning_rate": 2.2624652874698673e-06,
+ "loss": 0.0091,
+ "step": 28926
+ },
+ {
+ "epoch": 78.82016348773843,
+ "grad_norm": 1.3853952884674072,
+ "learning_rate": 2.2619062688698234e-06,
+ "loss": 0.0916,
+ "step": 28927
+ },
+ {
+ "epoch": 78.82288828337875,
+ "grad_norm": 1.0213408470153809,
+ "learning_rate": 2.2613473105338046e-06,
+ "loss": 0.0098,
+ "step": 28928
+ },
+ {
+ "epoch": 78.82561307901908,
+ "grad_norm": 0.8888218402862549,
+ "learning_rate": 2.2607884124661607e-06,
+ "loss": 0.0101,
+ "step": 28929
+ },
+ {
+ "epoch": 78.8283378746594,
+ "grad_norm": 1.4393377304077148,
+ "learning_rate": 2.2602295746712498e-06,
+ "loss": 0.0431,
+ "step": 28930
+ },
+ {
+ "epoch": 78.83106267029973,
+ "grad_norm": 1.5098282098770142,
+ "learning_rate": 2.2596707971534226e-06,
+ "loss": 0.0251,
+ "step": 28931
+ },
+ {
+ "epoch": 78.83378746594005,
+ "grad_norm": 1.6417872905731201,
+ "learning_rate": 2.2591120799170262e-06,
+ "loss": 0.0531,
+ "step": 28932
+ },
+ {
+ "epoch": 78.83651226158038,
+ "grad_norm": 1.3029823303222656,
+ "learning_rate": 2.25855342296642e-06,
+ "loss": 0.0113,
+ "step": 28933
+ },
+ {
+ "epoch": 78.83923705722071,
+ "grad_norm": 0.6250361800193787,
+ "learning_rate": 2.257994826305947e-06,
+ "loss": 0.0063,
+ "step": 28934
+ },
+ {
+ "epoch": 78.84196185286103,
+ "grad_norm": 1.1511660814285278,
+ "learning_rate": 2.2574362899399648e-06,
+ "loss": 0.0158,
+ "step": 28935
+ },
+ {
+ "epoch": 78.84468664850137,
+ "grad_norm": 0.9283928871154785,
+ "learning_rate": 2.2568778138728186e-06,
+ "loss": 0.0188,
+ "step": 28936
+ },
+ {
+ "epoch": 78.84741144414168,
+ "grad_norm": 0.8640484809875488,
+ "learning_rate": 2.256319398108859e-06,
+ "loss": 0.0085,
+ "step": 28937
+ },
+ {
+ "epoch": 78.85013623978202,
+ "grad_norm": 0.9172985553741455,
+ "learning_rate": 2.255761042652432e-06,
+ "loss": 0.0067,
+ "step": 28938
+ },
+ {
+ "epoch": 78.85286103542235,
+ "grad_norm": 1.0558632612228394,
+ "learning_rate": 2.255202747507892e-06,
+ "loss": 0.0162,
+ "step": 28939
+ },
+ {
+ "epoch": 78.85558583106267,
+ "grad_norm": 0.795630693435669,
+ "learning_rate": 2.2546445126795822e-06,
+ "loss": 0.0784,
+ "step": 28940
+ },
+ {
+ "epoch": 78.858310626703,
+ "grad_norm": 1.426708698272705,
+ "learning_rate": 2.2540863381718502e-06,
+ "loss": 0.0178,
+ "step": 28941
+ },
+ {
+ "epoch": 78.86103542234332,
+ "grad_norm": 1.7380928993225098,
+ "learning_rate": 2.253528223989048e-06,
+ "loss": 0.0498,
+ "step": 28942
+ },
+ {
+ "epoch": 78.86376021798365,
+ "grad_norm": 0.5654027462005615,
+ "learning_rate": 2.2529701701355167e-06,
+ "loss": 0.0056,
+ "step": 28943
+ },
+ {
+ "epoch": 78.86648501362397,
+ "grad_norm": 0.7257049679756165,
+ "learning_rate": 2.252412176615606e-06,
+ "loss": 0.0088,
+ "step": 28944
+ },
+ {
+ "epoch": 78.8692098092643,
+ "grad_norm": 1.0333832502365112,
+ "learning_rate": 2.2518542434336564e-06,
+ "loss": 0.0066,
+ "step": 28945
+ },
+ {
+ "epoch": 78.87193460490464,
+ "grad_norm": 1.1716374158859253,
+ "learning_rate": 2.2512963705940192e-06,
+ "loss": 0.0144,
+ "step": 28946
+ },
+ {
+ "epoch": 78.87465940054496,
+ "grad_norm": 0.5830609798431396,
+ "learning_rate": 2.2507385581010364e-06,
+ "loss": 0.005,
+ "step": 28947
+ },
+ {
+ "epoch": 78.87738419618529,
+ "grad_norm": 1.4811080694198608,
+ "learning_rate": 2.2501808059590537e-06,
+ "loss": 0.0225,
+ "step": 28948
+ },
+ {
+ "epoch": 78.88010899182561,
+ "grad_norm": 0.8598942756652832,
+ "learning_rate": 2.249623114172409e-06,
+ "loss": 0.0097,
+ "step": 28949
+ },
+ {
+ "epoch": 78.88283378746594,
+ "grad_norm": 2.9221017360687256,
+ "learning_rate": 2.2490654827454505e-06,
+ "loss": 0.0513,
+ "step": 28950
+ },
+ {
+ "epoch": 78.88555858310627,
+ "grad_norm": 0.646652102470398,
+ "learning_rate": 2.2485079116825236e-06,
+ "loss": 0.0052,
+ "step": 28951
+ },
+ {
+ "epoch": 78.88828337874659,
+ "grad_norm": 0.7954646944999695,
+ "learning_rate": 2.247950400987968e-06,
+ "loss": 0.0085,
+ "step": 28952
+ },
+ {
+ "epoch": 78.89100817438693,
+ "grad_norm": 2.3646461963653564,
+ "learning_rate": 2.247392950666124e-06,
+ "loss": 0.0566,
+ "step": 28953
+ },
+ {
+ "epoch": 78.89373297002724,
+ "grad_norm": 1.0446374416351318,
+ "learning_rate": 2.246835560721332e-06,
+ "loss": 0.0194,
+ "step": 28954
+ },
+ {
+ "epoch": 78.89645776566758,
+ "grad_norm": 1.5412089824676514,
+ "learning_rate": 2.246278231157939e-06,
+ "loss": 0.0265,
+ "step": 28955
+ },
+ {
+ "epoch": 78.8991825613079,
+ "grad_norm": 0.9640472531318665,
+ "learning_rate": 2.24572096198028e-06,
+ "loss": 0.0104,
+ "step": 28956
+ },
+ {
+ "epoch": 78.90190735694823,
+ "grad_norm": 1.059377670288086,
+ "learning_rate": 2.2451637531926973e-06,
+ "loss": 0.0086,
+ "step": 28957
+ },
+ {
+ "epoch": 78.90463215258856,
+ "grad_norm": 1.544297695159912,
+ "learning_rate": 2.244606604799526e-06,
+ "loss": 0.0535,
+ "step": 28958
+ },
+ {
+ "epoch": 78.90735694822888,
+ "grad_norm": 0.7658765316009521,
+ "learning_rate": 2.2440495168051134e-06,
+ "loss": 0.0096,
+ "step": 28959
+ },
+ {
+ "epoch": 78.91008174386921,
+ "grad_norm": 1.2799015045166016,
+ "learning_rate": 2.2434924892137923e-06,
+ "loss": 0.0174,
+ "step": 28960
+ },
+ {
+ "epoch": 78.91280653950953,
+ "grad_norm": 1.2162201404571533,
+ "learning_rate": 2.2429355220299e-06,
+ "loss": 0.0143,
+ "step": 28961
+ },
+ {
+ "epoch": 78.91553133514986,
+ "grad_norm": 1.2999601364135742,
+ "learning_rate": 2.2423786152577788e-06,
+ "loss": 0.0095,
+ "step": 28962
+ },
+ {
+ "epoch": 78.9182561307902,
+ "grad_norm": 1.0502122640609741,
+ "learning_rate": 2.241821768901762e-06,
+ "loss": 0.0809,
+ "step": 28963
+ },
+ {
+ "epoch": 78.92098092643052,
+ "grad_norm": 1.2565703392028809,
+ "learning_rate": 2.241264982966189e-06,
+ "loss": 0.015,
+ "step": 28964
+ },
+ {
+ "epoch": 78.92370572207085,
+ "grad_norm": 1.4066243171691895,
+ "learning_rate": 2.2407082574553905e-06,
+ "loss": 0.0164,
+ "step": 28965
+ },
+ {
+ "epoch": 78.92643051771117,
+ "grad_norm": 0.7988731861114502,
+ "learning_rate": 2.2401515923737083e-06,
+ "loss": 0.0082,
+ "step": 28966
+ },
+ {
+ "epoch": 78.9291553133515,
+ "grad_norm": 1.490052580833435,
+ "learning_rate": 2.2395949877254776e-06,
+ "loss": 0.0261,
+ "step": 28967
+ },
+ {
+ "epoch": 78.93188010899182,
+ "grad_norm": 1.2269814014434814,
+ "learning_rate": 2.23903844351503e-06,
+ "loss": 0.0244,
+ "step": 28968
+ },
+ {
+ "epoch": 78.93460490463215,
+ "grad_norm": 0.9568396806716919,
+ "learning_rate": 2.2384819597467033e-06,
+ "loss": 0.0077,
+ "step": 28969
+ },
+ {
+ "epoch": 78.93732970027249,
+ "grad_norm": 2.4934279918670654,
+ "learning_rate": 2.2379255364248252e-06,
+ "loss": 0.0165,
+ "step": 28970
+ },
+ {
+ "epoch": 78.9400544959128,
+ "grad_norm": 1.3784831762313843,
+ "learning_rate": 2.237369173553736e-06,
+ "loss": 0.0372,
+ "step": 28971
+ },
+ {
+ "epoch": 78.94277929155314,
+ "grad_norm": 1.0410429239273071,
+ "learning_rate": 2.2368128711377666e-06,
+ "loss": 0.0152,
+ "step": 28972
+ },
+ {
+ "epoch": 78.94550408719346,
+ "grad_norm": 0.6309718489646912,
+ "learning_rate": 2.2362566291812483e-06,
+ "loss": 0.0053,
+ "step": 28973
+ },
+ {
+ "epoch": 78.94822888283379,
+ "grad_norm": 2.894662618637085,
+ "learning_rate": 2.2357004476885112e-06,
+ "loss": 0.0177,
+ "step": 28974
+ },
+ {
+ "epoch": 78.95095367847412,
+ "grad_norm": 1.3251374959945679,
+ "learning_rate": 2.2351443266638918e-06,
+ "loss": 0.0434,
+ "step": 28975
+ },
+ {
+ "epoch": 78.95367847411444,
+ "grad_norm": 1.5782231092453003,
+ "learning_rate": 2.234588266111718e-06,
+ "loss": 0.0212,
+ "step": 28976
+ },
+ {
+ "epoch": 78.95640326975477,
+ "grad_norm": 1.83824622631073,
+ "learning_rate": 2.2340322660363213e-06,
+ "loss": 0.0421,
+ "step": 28977
+ },
+ {
+ "epoch": 78.95912806539509,
+ "grad_norm": 0.4737642705440521,
+ "learning_rate": 2.2334763264420277e-06,
+ "loss": 0.005,
+ "step": 28978
+ },
+ {
+ "epoch": 78.96185286103542,
+ "grad_norm": 0.809288740158081,
+ "learning_rate": 2.2329204473331745e-06,
+ "loss": 0.0092,
+ "step": 28979
+ },
+ {
+ "epoch": 78.96457765667574,
+ "grad_norm": 1.38800847530365,
+ "learning_rate": 2.2323646287140856e-06,
+ "loss": 0.0833,
+ "step": 28980
+ },
+ {
+ "epoch": 78.96730245231608,
+ "grad_norm": 0.7282382845878601,
+ "learning_rate": 2.2318088705890917e-06,
+ "loss": 0.0078,
+ "step": 28981
+ },
+ {
+ "epoch": 78.97002724795641,
+ "grad_norm": 0.9793028831481934,
+ "learning_rate": 2.2312531729625175e-06,
+ "loss": 0.0124,
+ "step": 28982
+ },
+ {
+ "epoch": 78.97275204359673,
+ "grad_norm": 1.151275873184204,
+ "learning_rate": 2.230697535838696e-06,
+ "loss": 0.0415,
+ "step": 28983
+ },
+ {
+ "epoch": 78.97547683923706,
+ "grad_norm": 1.4420721530914307,
+ "learning_rate": 2.2301419592219496e-06,
+ "loss": 0.0232,
+ "step": 28984
+ },
+ {
+ "epoch": 78.97820163487738,
+ "grad_norm": 1.665091872215271,
+ "learning_rate": 2.2295864431166105e-06,
+ "loss": 0.0285,
+ "step": 28985
+ },
+ {
+ "epoch": 78.98092643051771,
+ "grad_norm": 0.9037884473800659,
+ "learning_rate": 2.229030987526999e-06,
+ "loss": 0.0116,
+ "step": 28986
+ },
+ {
+ "epoch": 78.98365122615803,
+ "grad_norm": 1.2875901460647583,
+ "learning_rate": 2.2284755924574477e-06,
+ "loss": 0.0273,
+ "step": 28987
+ },
+ {
+ "epoch": 78.98637602179836,
+ "grad_norm": 1.5435775518417358,
+ "learning_rate": 2.2279202579122773e-06,
+ "loss": 0.0256,
+ "step": 28988
+ },
+ {
+ "epoch": 78.9891008174387,
+ "grad_norm": 1.5144495964050293,
+ "learning_rate": 2.2273649838958133e-06,
+ "loss": 0.0219,
+ "step": 28989
+ },
+ {
+ "epoch": 78.99182561307902,
+ "grad_norm": 1.7367302179336548,
+ "learning_rate": 2.2268097704123793e-06,
+ "loss": 0.023,
+ "step": 28990
+ },
+ {
+ "epoch": 78.99455040871935,
+ "grad_norm": 1.0641164779663086,
+ "learning_rate": 2.2262546174663024e-06,
+ "loss": 0.0192,
+ "step": 28991
+ },
+ {
+ "epoch": 78.99727520435967,
+ "grad_norm": 0.5521531701087952,
+ "learning_rate": 2.225699525061905e-06,
+ "loss": 0.0064,
+ "step": 28992
+ },
+ {
+ "epoch": 79.0,
+ "grad_norm": 0.7567386627197266,
+ "learning_rate": 2.2251444932035094e-06,
+ "loss": 0.0108,
+ "step": 28993
+ },
+ {
+ "epoch": 79.00272479564033,
+ "grad_norm": 1.0937917232513428,
+ "learning_rate": 2.224589521895435e-06,
+ "loss": 0.1006,
+ "step": 28994
+ },
+ {
+ "epoch": 79.00544959128065,
+ "grad_norm": 1.9542877674102783,
+ "learning_rate": 2.22403461114201e-06,
+ "loss": 0.0443,
+ "step": 28995
+ },
+ {
+ "epoch": 79.00817438692098,
+ "grad_norm": 1.4629085063934326,
+ "learning_rate": 2.2234797609475536e-06,
+ "loss": 0.0277,
+ "step": 28996
+ },
+ {
+ "epoch": 79.0108991825613,
+ "grad_norm": 0.8321093916893005,
+ "learning_rate": 2.222924971316386e-06,
+ "loss": 0.01,
+ "step": 28997
+ },
+ {
+ "epoch": 79.01362397820164,
+ "grad_norm": 0.986552357673645,
+ "learning_rate": 2.222370242252826e-06,
+ "loss": 0.0169,
+ "step": 28998
+ },
+ {
+ "epoch": 79.01634877384195,
+ "grad_norm": 0.9042943120002747,
+ "learning_rate": 2.221815573761199e-06,
+ "loss": 0.0122,
+ "step": 28999
+ },
+ {
+ "epoch": 79.01907356948229,
+ "grad_norm": 1.1249909400939941,
+ "learning_rate": 2.2212609658458217e-06,
+ "loss": 0.0104,
+ "step": 29000
+ },
+ {
+ "epoch": 79.02179836512262,
+ "grad_norm": 1.1653852462768555,
+ "learning_rate": 2.2207064185110107e-06,
+ "loss": 0.0249,
+ "step": 29001
+ },
+ {
+ "epoch": 79.02452316076294,
+ "grad_norm": 1.3209730386734009,
+ "learning_rate": 2.220151931761091e-06,
+ "loss": 0.0599,
+ "step": 29002
+ },
+ {
+ "epoch": 79.02724795640327,
+ "grad_norm": 0.5065374374389648,
+ "learning_rate": 2.219597505600375e-06,
+ "loss": 0.0058,
+ "step": 29003
+ },
+ {
+ "epoch": 79.02997275204359,
+ "grad_norm": 1.4551349878311157,
+ "learning_rate": 2.219043140033186e-06,
+ "loss": 0.0444,
+ "step": 29004
+ },
+ {
+ "epoch": 79.03269754768392,
+ "grad_norm": 1.6630839109420776,
+ "learning_rate": 2.2184888350638378e-06,
+ "loss": 0.1057,
+ "step": 29005
+ },
+ {
+ "epoch": 79.03542234332426,
+ "grad_norm": 1.3556360006332397,
+ "learning_rate": 2.2179345906966464e-06,
+ "loss": 0.0126,
+ "step": 29006
+ },
+ {
+ "epoch": 79.03814713896458,
+ "grad_norm": 1.2819921970367432,
+ "learning_rate": 2.2173804069359316e-06,
+ "loss": 0.054,
+ "step": 29007
+ },
+ {
+ "epoch": 79.04087193460491,
+ "grad_norm": 0.6702831983566284,
+ "learning_rate": 2.2168262837860087e-06,
+ "loss": 0.0075,
+ "step": 29008
+ },
+ {
+ "epoch": 79.04359673024523,
+ "grad_norm": 1.0582998991012573,
+ "learning_rate": 2.216272221251192e-06,
+ "loss": 0.0076,
+ "step": 29009
+ },
+ {
+ "epoch": 79.04632152588556,
+ "grad_norm": 1.5840171575546265,
+ "learning_rate": 2.215718219335794e-06,
+ "loss": 0.0116,
+ "step": 29010
+ },
+ {
+ "epoch": 79.04904632152588,
+ "grad_norm": 0.6970599889755249,
+ "learning_rate": 2.215164278044134e-06,
+ "loss": 0.0075,
+ "step": 29011
+ },
+ {
+ "epoch": 79.05177111716621,
+ "grad_norm": 1.119528889656067,
+ "learning_rate": 2.2146103973805243e-06,
+ "loss": 0.0103,
+ "step": 29012
+ },
+ {
+ "epoch": 79.05449591280654,
+ "grad_norm": 1.0247236490249634,
+ "learning_rate": 2.2140565773492783e-06,
+ "loss": 0.0505,
+ "step": 29013
+ },
+ {
+ "epoch": 79.05722070844686,
+ "grad_norm": 1.153326153755188,
+ "learning_rate": 2.213502817954707e-06,
+ "loss": 0.0277,
+ "step": 29014
+ },
+ {
+ "epoch": 79.0599455040872,
+ "grad_norm": 1.3724230527877808,
+ "learning_rate": 2.2129491192011276e-06,
+ "loss": 0.01,
+ "step": 29015
+ },
+ {
+ "epoch": 79.06267029972751,
+ "grad_norm": 1.6338582038879395,
+ "learning_rate": 2.2123954810928494e-06,
+ "loss": 0.0194,
+ "step": 29016
+ },
+ {
+ "epoch": 79.06539509536785,
+ "grad_norm": 1.2594175338745117,
+ "learning_rate": 2.2118419036341852e-06,
+ "loss": 0.0309,
+ "step": 29017
+ },
+ {
+ "epoch": 79.06811989100818,
+ "grad_norm": 1.2292157411575317,
+ "learning_rate": 2.211288386829442e-06,
+ "loss": 0.0121,
+ "step": 29018
+ },
+ {
+ "epoch": 79.0708446866485,
+ "grad_norm": 0.7911825180053711,
+ "learning_rate": 2.2107349306829353e-06,
+ "loss": 0.0059,
+ "step": 29019
+ },
+ {
+ "epoch": 79.07356948228883,
+ "grad_norm": 0.8029688000679016,
+ "learning_rate": 2.2101815351989766e-06,
+ "loss": 0.0094,
+ "step": 29020
+ },
+ {
+ "epoch": 79.07629427792915,
+ "grad_norm": 1.525253415107727,
+ "learning_rate": 2.2096282003818727e-06,
+ "loss": 0.0528,
+ "step": 29021
+ },
+ {
+ "epoch": 79.07901907356948,
+ "grad_norm": 0.8280035853385925,
+ "learning_rate": 2.2090749262359335e-06,
+ "loss": 0.0101,
+ "step": 29022
+ },
+ {
+ "epoch": 79.0817438692098,
+ "grad_norm": 0.5020565390586853,
+ "learning_rate": 2.2085217127654667e-06,
+ "loss": 0.0045,
+ "step": 29023
+ },
+ {
+ "epoch": 79.08446866485014,
+ "grad_norm": 0.7323960065841675,
+ "learning_rate": 2.2079685599747836e-06,
+ "loss": 0.0102,
+ "step": 29024
+ },
+ {
+ "epoch": 79.08719346049047,
+ "grad_norm": 1.161496877670288,
+ "learning_rate": 2.2074154678681915e-06,
+ "loss": 0.0088,
+ "step": 29025
+ },
+ {
+ "epoch": 79.08991825613079,
+ "grad_norm": 0.7610247731208801,
+ "learning_rate": 2.2068624364499923e-06,
+ "loss": 0.0079,
+ "step": 29026
+ },
+ {
+ "epoch": 79.09264305177112,
+ "grad_norm": 1.2842580080032349,
+ "learning_rate": 2.2063094657245023e-06,
+ "loss": 0.0718,
+ "step": 29027
+ },
+ {
+ "epoch": 79.09536784741144,
+ "grad_norm": 1.1980974674224854,
+ "learning_rate": 2.2057565556960224e-06,
+ "loss": 0.0056,
+ "step": 29028
+ },
+ {
+ "epoch": 79.09809264305177,
+ "grad_norm": 1.2599010467529297,
+ "learning_rate": 2.20520370636886e-06,
+ "loss": 0.0165,
+ "step": 29029
+ },
+ {
+ "epoch": 79.1008174386921,
+ "grad_norm": 0.723651647567749,
+ "learning_rate": 2.2046509177473172e-06,
+ "loss": 0.0074,
+ "step": 29030
+ },
+ {
+ "epoch": 79.10354223433242,
+ "grad_norm": 1.2506153583526611,
+ "learning_rate": 2.204098189835705e-06,
+ "loss": 0.0129,
+ "step": 29031
+ },
+ {
+ "epoch": 79.10626702997276,
+ "grad_norm": 0.5673043727874756,
+ "learning_rate": 2.203545522638326e-06,
+ "loss": 0.0075,
+ "step": 29032
+ },
+ {
+ "epoch": 79.10899182561307,
+ "grad_norm": 0.8231469988822937,
+ "learning_rate": 2.2029929161594832e-06,
+ "loss": 0.0086,
+ "step": 29033
+ },
+ {
+ "epoch": 79.11171662125341,
+ "grad_norm": 1.0721300840377808,
+ "learning_rate": 2.202440370403477e-06,
+ "loss": 0.0088,
+ "step": 29034
+ },
+ {
+ "epoch": 79.11444141689373,
+ "grad_norm": 0.7113640904426575,
+ "learning_rate": 2.201887885374615e-06,
+ "loss": 0.01,
+ "step": 29035
+ },
+ {
+ "epoch": 79.11716621253406,
+ "grad_norm": 1.155202031135559,
+ "learning_rate": 2.201335461077201e-06,
+ "loss": 0.012,
+ "step": 29036
+ },
+ {
+ "epoch": 79.11989100817439,
+ "grad_norm": 0.9222427606582642,
+ "learning_rate": 2.2007830975155366e-06,
+ "loss": 0.008,
+ "step": 29037
+ },
+ {
+ "epoch": 79.12261580381471,
+ "grad_norm": 1.4269224405288696,
+ "learning_rate": 2.200230794693922e-06,
+ "loss": 0.0172,
+ "step": 29038
+ },
+ {
+ "epoch": 79.12534059945504,
+ "grad_norm": 1.2918378114700317,
+ "learning_rate": 2.1996785526166565e-06,
+ "loss": 0.0305,
+ "step": 29039
+ },
+ {
+ "epoch": 79.12806539509536,
+ "grad_norm": 1.4371978044509888,
+ "learning_rate": 2.199126371288045e-06,
+ "loss": 0.0188,
+ "step": 29040
+ },
+ {
+ "epoch": 79.1307901907357,
+ "grad_norm": 0.6200845241546631,
+ "learning_rate": 2.1985742507123873e-06,
+ "loss": 0.0069,
+ "step": 29041
+ },
+ {
+ "epoch": 79.13351498637603,
+ "grad_norm": 1.1617746353149414,
+ "learning_rate": 2.1980221908939812e-06,
+ "loss": 0.0101,
+ "step": 29042
+ },
+ {
+ "epoch": 79.13623978201635,
+ "grad_norm": 2.344871759414673,
+ "learning_rate": 2.1974701918371253e-06,
+ "loss": 0.0619,
+ "step": 29043
+ },
+ {
+ "epoch": 79.13896457765668,
+ "grad_norm": 2.1115097999572754,
+ "learning_rate": 2.196918253546123e-06,
+ "loss": 0.0141,
+ "step": 29044
+ },
+ {
+ "epoch": 79.141689373297,
+ "grad_norm": 1.1288162469863892,
+ "learning_rate": 2.1963663760252706e-06,
+ "loss": 0.0126,
+ "step": 29045
+ },
+ {
+ "epoch": 79.14441416893733,
+ "grad_norm": 0.8687530755996704,
+ "learning_rate": 2.195814559278866e-06,
+ "loss": 0.0169,
+ "step": 29046
+ },
+ {
+ "epoch": 79.14713896457765,
+ "grad_norm": 1.9921205043792725,
+ "learning_rate": 2.1952628033112023e-06,
+ "loss": 0.0269,
+ "step": 29047
+ },
+ {
+ "epoch": 79.14986376021798,
+ "grad_norm": 1.4977518320083618,
+ "learning_rate": 2.194711108126585e-06,
+ "loss": 0.0314,
+ "step": 29048
+ },
+ {
+ "epoch": 79.15258855585832,
+ "grad_norm": 1.171364665031433,
+ "learning_rate": 2.1941594737293046e-06,
+ "loss": 0.0102,
+ "step": 29049
+ },
+ {
+ "epoch": 79.15531335149863,
+ "grad_norm": 1.5378938913345337,
+ "learning_rate": 2.193607900123658e-06,
+ "loss": 0.0094,
+ "step": 29050
+ },
+ {
+ "epoch": 79.15803814713897,
+ "grad_norm": 1.2819361686706543,
+ "learning_rate": 2.193056387313943e-06,
+ "loss": 0.036,
+ "step": 29051
+ },
+ {
+ "epoch": 79.16076294277929,
+ "grad_norm": 0.5913614630699158,
+ "learning_rate": 2.1925049353044516e-06,
+ "loss": 0.0116,
+ "step": 29052
+ },
+ {
+ "epoch": 79.16348773841962,
+ "grad_norm": 1.407529592514038,
+ "learning_rate": 2.191953544099483e-06,
+ "loss": 0.0535,
+ "step": 29053
+ },
+ {
+ "epoch": 79.16621253405995,
+ "grad_norm": 1.4180570840835571,
+ "learning_rate": 2.1914022137033298e-06,
+ "loss": 0.0639,
+ "step": 29054
+ },
+ {
+ "epoch": 79.16893732970027,
+ "grad_norm": 0.967038094997406,
+ "learning_rate": 2.1908509441202808e-06,
+ "loss": 0.0156,
+ "step": 29055
+ },
+ {
+ "epoch": 79.1716621253406,
+ "grad_norm": 1.4900633096694946,
+ "learning_rate": 2.1902997353546375e-06,
+ "loss": 0.1728,
+ "step": 29056
+ },
+ {
+ "epoch": 79.17438692098092,
+ "grad_norm": 1.0234529972076416,
+ "learning_rate": 2.189748587410687e-06,
+ "loss": 0.0266,
+ "step": 29057
+ },
+ {
+ "epoch": 79.17711171662125,
+ "grad_norm": 1.081392526626587,
+ "learning_rate": 2.1891975002927235e-06,
+ "loss": 0.0135,
+ "step": 29058
+ },
+ {
+ "epoch": 79.17983651226157,
+ "grad_norm": 0.701111912727356,
+ "learning_rate": 2.188646474005036e-06,
+ "loss": 0.0059,
+ "step": 29059
+ },
+ {
+ "epoch": 79.1825613079019,
+ "grad_norm": 1.2708685398101807,
+ "learning_rate": 2.1880955085519217e-06,
+ "loss": 0.024,
+ "step": 29060
+ },
+ {
+ "epoch": 79.18528610354224,
+ "grad_norm": 1.2297734022140503,
+ "learning_rate": 2.187544603937668e-06,
+ "loss": 0.0195,
+ "step": 29061
+ },
+ {
+ "epoch": 79.18801089918256,
+ "grad_norm": 1.4035158157348633,
+ "learning_rate": 2.1869937601665646e-06,
+ "loss": 0.0116,
+ "step": 29062
+ },
+ {
+ "epoch": 79.19073569482289,
+ "grad_norm": 1.5636730194091797,
+ "learning_rate": 2.1864429772428997e-06,
+ "loss": 0.0171,
+ "step": 29063
+ },
+ {
+ "epoch": 79.19346049046321,
+ "grad_norm": 1.2463241815567017,
+ "learning_rate": 2.1858922551709694e-06,
+ "loss": 0.0235,
+ "step": 29064
+ },
+ {
+ "epoch": 79.19618528610354,
+ "grad_norm": 1.2827446460723877,
+ "learning_rate": 2.1853415939550572e-06,
+ "loss": 0.0198,
+ "step": 29065
+ },
+ {
+ "epoch": 79.19891008174388,
+ "grad_norm": 2.4486327171325684,
+ "learning_rate": 2.1847909935994538e-06,
+ "loss": 0.1443,
+ "step": 29066
+ },
+ {
+ "epoch": 79.2016348773842,
+ "grad_norm": 1.7004660367965698,
+ "learning_rate": 2.1842404541084437e-06,
+ "loss": 0.0334,
+ "step": 29067
+ },
+ {
+ "epoch": 79.20435967302453,
+ "grad_norm": 0.8871360421180725,
+ "learning_rate": 2.1836899754863207e-06,
+ "loss": 0.0085,
+ "step": 29068
+ },
+ {
+ "epoch": 79.20708446866485,
+ "grad_norm": 1.7397288084030151,
+ "learning_rate": 2.183139557737368e-06,
+ "loss": 0.1248,
+ "step": 29069
+ },
+ {
+ "epoch": 79.20980926430518,
+ "grad_norm": 1.4672861099243164,
+ "learning_rate": 2.18258920086587e-06,
+ "loss": 0.0361,
+ "step": 29070
+ },
+ {
+ "epoch": 79.2125340599455,
+ "grad_norm": 0.9448686838150024,
+ "learning_rate": 2.1820389048761172e-06,
+ "loss": 0.0108,
+ "step": 29071
+ },
+ {
+ "epoch": 79.21525885558583,
+ "grad_norm": 0.9146217703819275,
+ "learning_rate": 2.1814886697723946e-06,
+ "loss": 0.0102,
+ "step": 29072
+ },
+ {
+ "epoch": 79.21798365122616,
+ "grad_norm": 0.8742615580558777,
+ "learning_rate": 2.1809384955589873e-06,
+ "loss": 0.0087,
+ "step": 29073
+ },
+ {
+ "epoch": 79.22070844686648,
+ "grad_norm": 1.2590808868408203,
+ "learning_rate": 2.1803883822401795e-06,
+ "loss": 0.0139,
+ "step": 29074
+ },
+ {
+ "epoch": 79.22343324250681,
+ "grad_norm": 4.14694356918335,
+ "learning_rate": 2.1798383298202518e-06,
+ "loss": 0.0286,
+ "step": 29075
+ },
+ {
+ "epoch": 79.22615803814713,
+ "grad_norm": 0.46457627415657043,
+ "learning_rate": 2.1792883383034956e-06,
+ "loss": 0.0048,
+ "step": 29076
+ },
+ {
+ "epoch": 79.22888283378747,
+ "grad_norm": 1.0129966735839844,
+ "learning_rate": 2.1787384076941897e-06,
+ "loss": 0.0916,
+ "step": 29077
+ },
+ {
+ "epoch": 79.2316076294278,
+ "grad_norm": 1.299119472503662,
+ "learning_rate": 2.1781885379966173e-06,
+ "loss": 0.0173,
+ "step": 29078
+ },
+ {
+ "epoch": 79.23433242506812,
+ "grad_norm": 1.8402504920959473,
+ "learning_rate": 2.177638729215057e-06,
+ "loss": 0.0275,
+ "step": 29079
+ },
+ {
+ "epoch": 79.23705722070845,
+ "grad_norm": 1.0213285684585571,
+ "learning_rate": 2.177088981353799e-06,
+ "loss": 0.0157,
+ "step": 29080
+ },
+ {
+ "epoch": 79.23978201634877,
+ "grad_norm": 1.0866577625274658,
+ "learning_rate": 2.1765392944171183e-06,
+ "loss": 0.0139,
+ "step": 29081
+ },
+ {
+ "epoch": 79.2425068119891,
+ "grad_norm": 1.3527318239212036,
+ "learning_rate": 2.1759896684092997e-06,
+ "loss": 0.0178,
+ "step": 29082
+ },
+ {
+ "epoch": 79.24523160762942,
+ "grad_norm": 1.7735931873321533,
+ "learning_rate": 2.1754401033346172e-06,
+ "loss": 0.0666,
+ "step": 29083
+ },
+ {
+ "epoch": 79.24795640326975,
+ "grad_norm": 1.2562059164047241,
+ "learning_rate": 2.174890599197359e-06,
+ "loss": 0.0376,
+ "step": 29084
+ },
+ {
+ "epoch": 79.25068119891009,
+ "grad_norm": 1.1960961818695068,
+ "learning_rate": 2.174341156001801e-06,
+ "loss": 0.0114,
+ "step": 29085
+ },
+ {
+ "epoch": 79.2534059945504,
+ "grad_norm": 1.1211775541305542,
+ "learning_rate": 2.173791773752223e-06,
+ "loss": 0.0119,
+ "step": 29086
+ },
+ {
+ "epoch": 79.25613079019074,
+ "grad_norm": 1.5489630699157715,
+ "learning_rate": 2.1732424524528984e-06,
+ "loss": 0.0334,
+ "step": 29087
+ },
+ {
+ "epoch": 79.25885558583106,
+ "grad_norm": 1.1199517250061035,
+ "learning_rate": 2.172693192108111e-06,
+ "loss": 0.0227,
+ "step": 29088
+ },
+ {
+ "epoch": 79.26158038147139,
+ "grad_norm": 0.5915191173553467,
+ "learning_rate": 2.1721439927221386e-06,
+ "loss": 0.0076,
+ "step": 29089
+ },
+ {
+ "epoch": 79.26430517711172,
+ "grad_norm": 1.1195778846740723,
+ "learning_rate": 2.1715948542992583e-06,
+ "loss": 0.0066,
+ "step": 29090
+ },
+ {
+ "epoch": 79.26702997275204,
+ "grad_norm": 1.0413169860839844,
+ "learning_rate": 2.171045776843742e-06,
+ "loss": 0.014,
+ "step": 29091
+ },
+ {
+ "epoch": 79.26975476839237,
+ "grad_norm": 0.8878083825111389,
+ "learning_rate": 2.1704967603598716e-06,
+ "loss": 0.0472,
+ "step": 29092
+ },
+ {
+ "epoch": 79.2724795640327,
+ "grad_norm": 1.9274837970733643,
+ "learning_rate": 2.1699478048519207e-06,
+ "loss": 0.01,
+ "step": 29093
+ },
+ {
+ "epoch": 79.27520435967303,
+ "grad_norm": 0.8874469995498657,
+ "learning_rate": 2.169398910324164e-06,
+ "loss": 0.0102,
+ "step": 29094
+ },
+ {
+ "epoch": 79.27792915531334,
+ "grad_norm": 0.8275299072265625,
+ "learning_rate": 2.168850076780874e-06,
+ "loss": 0.0092,
+ "step": 29095
+ },
+ {
+ "epoch": 79.28065395095368,
+ "grad_norm": 1.2529503107070923,
+ "learning_rate": 2.1683013042263303e-06,
+ "loss": 0.0359,
+ "step": 29096
+ },
+ {
+ "epoch": 79.28337874659401,
+ "grad_norm": 1.1828376054763794,
+ "learning_rate": 2.1677525926648046e-06,
+ "loss": 0.0328,
+ "step": 29097
+ },
+ {
+ "epoch": 79.28610354223433,
+ "grad_norm": 1.5413779020309448,
+ "learning_rate": 2.1672039421005685e-06,
+ "loss": 0.0214,
+ "step": 29098
+ },
+ {
+ "epoch": 79.28882833787466,
+ "grad_norm": 1.1786149740219116,
+ "learning_rate": 2.166655352537894e-06,
+ "loss": 0.0141,
+ "step": 29099
+ },
+ {
+ "epoch": 79.29155313351498,
+ "grad_norm": 1.677643060684204,
+ "learning_rate": 2.166106823981058e-06,
+ "loss": 0.0402,
+ "step": 29100
+ },
+ {
+ "epoch": 79.29427792915531,
+ "grad_norm": 1.2385532855987549,
+ "learning_rate": 2.1655583564343295e-06,
+ "loss": 0.0052,
+ "step": 29101
+ },
+ {
+ "epoch": 79.29700272479565,
+ "grad_norm": 1.7901291847229004,
+ "learning_rate": 2.1650099499019817e-06,
+ "loss": 0.0339,
+ "step": 29102
+ },
+ {
+ "epoch": 79.29972752043597,
+ "grad_norm": 1.7087721824645996,
+ "learning_rate": 2.1644616043882806e-06,
+ "loss": 0.0338,
+ "step": 29103
+ },
+ {
+ "epoch": 79.3024523160763,
+ "grad_norm": 1.3210787773132324,
+ "learning_rate": 2.1639133198975003e-06,
+ "loss": 0.1009,
+ "step": 29104
+ },
+ {
+ "epoch": 79.30517711171662,
+ "grad_norm": 0.9216951727867126,
+ "learning_rate": 2.1633650964339137e-06,
+ "loss": 0.0102,
+ "step": 29105
+ },
+ {
+ "epoch": 79.30790190735695,
+ "grad_norm": 0.8876740336418152,
+ "learning_rate": 2.162816934001787e-06,
+ "loss": 0.0102,
+ "step": 29106
+ },
+ {
+ "epoch": 79.31062670299727,
+ "grad_norm": 1.2489911317825317,
+ "learning_rate": 2.162268832605391e-06,
+ "loss": 0.0903,
+ "step": 29107
+ },
+ {
+ "epoch": 79.3133514986376,
+ "grad_norm": 0.4597548544406891,
+ "learning_rate": 2.161720792248989e-06,
+ "loss": 0.004,
+ "step": 29108
+ },
+ {
+ "epoch": 79.31607629427793,
+ "grad_norm": 1.0715570449829102,
+ "learning_rate": 2.161172812936857e-06,
+ "loss": 0.0091,
+ "step": 29109
+ },
+ {
+ "epoch": 79.31880108991825,
+ "grad_norm": 1.4390521049499512,
+ "learning_rate": 2.1606248946732578e-06,
+ "loss": 0.0595,
+ "step": 29110
+ },
+ {
+ "epoch": 79.32152588555859,
+ "grad_norm": 1.4166854619979858,
+ "learning_rate": 2.1600770374624604e-06,
+ "loss": 0.0532,
+ "step": 29111
+ },
+ {
+ "epoch": 79.3242506811989,
+ "grad_norm": 1.040635585784912,
+ "learning_rate": 2.1595292413087277e-06,
+ "loss": 0.0872,
+ "step": 29112
+ },
+ {
+ "epoch": 79.32697547683924,
+ "grad_norm": 1.8025219440460205,
+ "learning_rate": 2.1589815062163323e-06,
+ "loss": 0.0214,
+ "step": 29113
+ },
+ {
+ "epoch": 79.32970027247957,
+ "grad_norm": 0.8911259174346924,
+ "learning_rate": 2.1584338321895358e-06,
+ "loss": 0.0104,
+ "step": 29114
+ },
+ {
+ "epoch": 79.33242506811989,
+ "grad_norm": 1.13921320438385,
+ "learning_rate": 2.157886219232601e-06,
+ "loss": 0.0753,
+ "step": 29115
+ },
+ {
+ "epoch": 79.33514986376022,
+ "grad_norm": 1.9557658433914185,
+ "learning_rate": 2.1573386673498e-06,
+ "loss": 0.0181,
+ "step": 29116
+ },
+ {
+ "epoch": 79.33787465940054,
+ "grad_norm": 1.1037222146987915,
+ "learning_rate": 2.156791176545392e-06,
+ "loss": 0.0091,
+ "step": 29117
+ },
+ {
+ "epoch": 79.34059945504087,
+ "grad_norm": 1.450122356414795,
+ "learning_rate": 2.156243746823643e-06,
+ "loss": 0.0603,
+ "step": 29118
+ },
+ {
+ "epoch": 79.34332425068119,
+ "grad_norm": 0.8364536166191101,
+ "learning_rate": 2.155696378188812e-06,
+ "loss": 0.0132,
+ "step": 29119
+ },
+ {
+ "epoch": 79.34604904632153,
+ "grad_norm": 0.8436452150344849,
+ "learning_rate": 2.155149070645167e-06,
+ "loss": 0.0143,
+ "step": 29120
+ },
+ {
+ "epoch": 79.34877384196186,
+ "grad_norm": 0.8105842471122742,
+ "learning_rate": 2.154601824196967e-06,
+ "loss": 0.0994,
+ "step": 29121
+ },
+ {
+ "epoch": 79.35149863760218,
+ "grad_norm": 0.747840404510498,
+ "learning_rate": 2.1540546388484774e-06,
+ "loss": 0.0086,
+ "step": 29122
+ },
+ {
+ "epoch": 79.35422343324251,
+ "grad_norm": 1.5960206985473633,
+ "learning_rate": 2.1535075146039575e-06,
+ "loss": 0.0798,
+ "step": 29123
+ },
+ {
+ "epoch": 79.35694822888283,
+ "grad_norm": 0.8038755059242249,
+ "learning_rate": 2.152960451467667e-06,
+ "loss": 0.0153,
+ "step": 29124
+ },
+ {
+ "epoch": 79.35967302452316,
+ "grad_norm": 1.5522053241729736,
+ "learning_rate": 2.1524134494438697e-06,
+ "loss": 0.0262,
+ "step": 29125
+ },
+ {
+ "epoch": 79.3623978201635,
+ "grad_norm": 1.3262959718704224,
+ "learning_rate": 2.151866508536824e-06,
+ "loss": 0.0183,
+ "step": 29126
+ },
+ {
+ "epoch": 79.36512261580381,
+ "grad_norm": 1.6586555242538452,
+ "learning_rate": 2.1513196287507898e-06,
+ "loss": 0.0867,
+ "step": 29127
+ },
+ {
+ "epoch": 79.36784741144415,
+ "grad_norm": 1.5258349180221558,
+ "learning_rate": 2.1507728100900227e-06,
+ "loss": 0.0121,
+ "step": 29128
+ },
+ {
+ "epoch": 79.37057220708446,
+ "grad_norm": 1.0126173496246338,
+ "learning_rate": 2.150226052558787e-06,
+ "loss": 0.0192,
+ "step": 29129
+ },
+ {
+ "epoch": 79.3732970027248,
+ "grad_norm": 0.769167959690094,
+ "learning_rate": 2.149679356161338e-06,
+ "loss": 0.0087,
+ "step": 29130
+ },
+ {
+ "epoch": 79.37602179836512,
+ "grad_norm": 1.4727541208267212,
+ "learning_rate": 2.1491327209019342e-06,
+ "loss": 0.0236,
+ "step": 29131
+ },
+ {
+ "epoch": 79.37874659400545,
+ "grad_norm": 1.072219729423523,
+ "learning_rate": 2.148586146784829e-06,
+ "loss": 0.0095,
+ "step": 29132
+ },
+ {
+ "epoch": 79.38147138964578,
+ "grad_norm": 1.851108193397522,
+ "learning_rate": 2.148039633814285e-06,
+ "loss": 0.0181,
+ "step": 29133
+ },
+ {
+ "epoch": 79.3841961852861,
+ "grad_norm": 1.5865823030471802,
+ "learning_rate": 2.1474931819945555e-06,
+ "loss": 0.0438,
+ "step": 29134
+ },
+ {
+ "epoch": 79.38692098092643,
+ "grad_norm": 0.8885623216629028,
+ "learning_rate": 2.1469467913298937e-06,
+ "loss": 0.0186,
+ "step": 29135
+ },
+ {
+ "epoch": 79.38964577656675,
+ "grad_norm": 1.7100576162338257,
+ "learning_rate": 2.1464004618245605e-06,
+ "loss": 0.0208,
+ "step": 29136
+ },
+ {
+ "epoch": 79.39237057220708,
+ "grad_norm": 0.8645238876342773,
+ "learning_rate": 2.145854193482807e-06,
+ "loss": 0.1137,
+ "step": 29137
+ },
+ {
+ "epoch": 79.39509536784742,
+ "grad_norm": 0.3470247983932495,
+ "learning_rate": 2.1453079863088888e-06,
+ "loss": 0.0031,
+ "step": 29138
+ },
+ {
+ "epoch": 79.39782016348774,
+ "grad_norm": 1.7699092626571655,
+ "learning_rate": 2.144761840307057e-06,
+ "loss": 0.0275,
+ "step": 29139
+ },
+ {
+ "epoch": 79.40054495912807,
+ "grad_norm": 1.0067129135131836,
+ "learning_rate": 2.1442157554815656e-06,
+ "loss": 0.0082,
+ "step": 29140
+ },
+ {
+ "epoch": 79.40326975476839,
+ "grad_norm": 1.4103070497512817,
+ "learning_rate": 2.143669731836673e-06,
+ "loss": 0.0127,
+ "step": 29141
+ },
+ {
+ "epoch": 79.40599455040872,
+ "grad_norm": 1.262904167175293,
+ "learning_rate": 2.1431237693766273e-06,
+ "loss": 0.0147,
+ "step": 29142
+ },
+ {
+ "epoch": 79.40871934604904,
+ "grad_norm": 1.0061589479446411,
+ "learning_rate": 2.1425778681056807e-06,
+ "loss": 0.0102,
+ "step": 29143
+ },
+ {
+ "epoch": 79.41144414168937,
+ "grad_norm": 1.7624493837356567,
+ "learning_rate": 2.1420320280280827e-06,
+ "loss": 0.0233,
+ "step": 29144
+ },
+ {
+ "epoch": 79.4141689373297,
+ "grad_norm": 1.7629679441452026,
+ "learning_rate": 2.141486249148089e-06,
+ "loss": 0.0929,
+ "step": 29145
+ },
+ {
+ "epoch": 79.41689373297002,
+ "grad_norm": 2.7435855865478516,
+ "learning_rate": 2.1409405314699473e-06,
+ "loss": 0.0214,
+ "step": 29146
+ },
+ {
+ "epoch": 79.41961852861036,
+ "grad_norm": 1.0670863389968872,
+ "learning_rate": 2.1403948749979074e-06,
+ "loss": 0.1207,
+ "step": 29147
+ },
+ {
+ "epoch": 79.42234332425068,
+ "grad_norm": 2.3514249324798584,
+ "learning_rate": 2.139849279736217e-06,
+ "loss": 0.0181,
+ "step": 29148
+ },
+ {
+ "epoch": 79.42506811989101,
+ "grad_norm": 1.9218058586120605,
+ "learning_rate": 2.13930374568913e-06,
+ "loss": 0.0231,
+ "step": 29149
+ },
+ {
+ "epoch": 79.42779291553134,
+ "grad_norm": 1.793965220451355,
+ "learning_rate": 2.1387582728608913e-06,
+ "loss": 0.1503,
+ "step": 29150
+ },
+ {
+ "epoch": 79.43051771117166,
+ "grad_norm": 1.0104447603225708,
+ "learning_rate": 2.138212861255752e-06,
+ "loss": 0.0111,
+ "step": 29151
+ },
+ {
+ "epoch": 79.433242506812,
+ "grad_norm": 1.0365288257598877,
+ "learning_rate": 2.1376675108779543e-06,
+ "loss": 0.0108,
+ "step": 29152
+ },
+ {
+ "epoch": 79.43596730245231,
+ "grad_norm": 1.4258074760437012,
+ "learning_rate": 2.137122221731751e-06,
+ "loss": 0.0088,
+ "step": 29153
+ },
+ {
+ "epoch": 79.43869209809264,
+ "grad_norm": 1.3162626028060913,
+ "learning_rate": 2.1365769938213876e-06,
+ "loss": 0.0248,
+ "step": 29154
+ },
+ {
+ "epoch": 79.44141689373296,
+ "grad_norm": 0.8278371691703796,
+ "learning_rate": 2.1360318271511093e-06,
+ "loss": 0.0093,
+ "step": 29155
+ },
+ {
+ "epoch": 79.4441416893733,
+ "grad_norm": 0.582617998123169,
+ "learning_rate": 2.1354867217251573e-06,
+ "loss": 0.0052,
+ "step": 29156
+ },
+ {
+ "epoch": 79.44686648501363,
+ "grad_norm": 1.5022403001785278,
+ "learning_rate": 2.134941677547786e-06,
+ "loss": 0.1545,
+ "step": 29157
+ },
+ {
+ "epoch": 79.44959128065395,
+ "grad_norm": 1.5485198497772217,
+ "learning_rate": 2.1343966946232366e-06,
+ "loss": 0.039,
+ "step": 29158
+ },
+ {
+ "epoch": 79.45231607629428,
+ "grad_norm": 1.113293170928955,
+ "learning_rate": 2.1338517729557517e-06,
+ "loss": 0.02,
+ "step": 29159
+ },
+ {
+ "epoch": 79.4550408719346,
+ "grad_norm": 0.9797650575637817,
+ "learning_rate": 2.133306912549574e-06,
+ "loss": 0.0132,
+ "step": 29160
+ },
+ {
+ "epoch": 79.45776566757493,
+ "grad_norm": 1.0517603158950806,
+ "learning_rate": 2.1327621134089516e-06,
+ "loss": 0.0186,
+ "step": 29161
+ },
+ {
+ "epoch": 79.46049046321527,
+ "grad_norm": 1.100361704826355,
+ "learning_rate": 2.132217375538125e-06,
+ "loss": 0.0165,
+ "step": 29162
+ },
+ {
+ "epoch": 79.46321525885558,
+ "grad_norm": 1.2454136610031128,
+ "learning_rate": 2.131672698941335e-06,
+ "loss": 0.0165,
+ "step": 29163
+ },
+ {
+ "epoch": 79.46594005449592,
+ "grad_norm": 0.84520423412323,
+ "learning_rate": 2.131128083622823e-06,
+ "loss": 0.0763,
+ "step": 29164
+ },
+ {
+ "epoch": 79.46866485013624,
+ "grad_norm": 1.1456295251846313,
+ "learning_rate": 2.130583529586835e-06,
+ "loss": 0.0225,
+ "step": 29165
+ },
+ {
+ "epoch": 79.47138964577657,
+ "grad_norm": 0.6974464058876038,
+ "learning_rate": 2.130039036837609e-06,
+ "loss": 0.0128,
+ "step": 29166
+ },
+ {
+ "epoch": 79.47411444141689,
+ "grad_norm": 1.4286943674087524,
+ "learning_rate": 2.129494605379385e-06,
+ "loss": 0.0136,
+ "step": 29167
+ },
+ {
+ "epoch": 79.47683923705722,
+ "grad_norm": 1.199409008026123,
+ "learning_rate": 2.1289502352164015e-06,
+ "loss": 0.0151,
+ "step": 29168
+ },
+ {
+ "epoch": 79.47956403269755,
+ "grad_norm": 1.5763298273086548,
+ "learning_rate": 2.1284059263529035e-06,
+ "loss": 0.0652,
+ "step": 29169
+ },
+ {
+ "epoch": 79.48228882833787,
+ "grad_norm": 1.4098020792007446,
+ "learning_rate": 2.127861678793126e-06,
+ "loss": 0.0327,
+ "step": 29170
+ },
+ {
+ "epoch": 79.4850136239782,
+ "grad_norm": 0.7872903943061829,
+ "learning_rate": 2.1273174925413086e-06,
+ "loss": 0.0084,
+ "step": 29171
+ },
+ {
+ "epoch": 79.48773841961852,
+ "grad_norm": 0.9917256236076355,
+ "learning_rate": 2.126773367601688e-06,
+ "loss": 0.0068,
+ "step": 29172
+ },
+ {
+ "epoch": 79.49046321525886,
+ "grad_norm": 0.8632399439811707,
+ "learning_rate": 2.126229303978501e-06,
+ "loss": 0.0086,
+ "step": 29173
+ },
+ {
+ "epoch": 79.49318801089919,
+ "grad_norm": 1.583006501197815,
+ "learning_rate": 2.1256853016759905e-06,
+ "loss": 0.0119,
+ "step": 29174
+ },
+ {
+ "epoch": 79.49591280653951,
+ "grad_norm": 1.1189228296279907,
+ "learning_rate": 2.1251413606983883e-06,
+ "loss": 0.0109,
+ "step": 29175
+ },
+ {
+ "epoch": 79.49863760217984,
+ "grad_norm": 1.112747073173523,
+ "learning_rate": 2.1245974810499326e-06,
+ "loss": 0.0205,
+ "step": 29176
+ },
+ {
+ "epoch": 79.50136239782016,
+ "grad_norm": 0.8969922065734863,
+ "learning_rate": 2.1240536627348553e-06,
+ "loss": 0.009,
+ "step": 29177
+ },
+ {
+ "epoch": 79.50408719346049,
+ "grad_norm": 1.2624197006225586,
+ "learning_rate": 2.1235099057573973e-06,
+ "loss": 0.0149,
+ "step": 29178
+ },
+ {
+ "epoch": 79.50681198910081,
+ "grad_norm": 1.6586860418319702,
+ "learning_rate": 2.122966210121791e-06,
+ "loss": 0.0125,
+ "step": 29179
+ },
+ {
+ "epoch": 79.50953678474114,
+ "grad_norm": 0.8992807269096375,
+ "learning_rate": 2.122422575832267e-06,
+ "loss": 0.0314,
+ "step": 29180
+ },
+ {
+ "epoch": 79.51226158038148,
+ "grad_norm": 1.4808218479156494,
+ "learning_rate": 2.1218790028930656e-06,
+ "loss": 0.0109,
+ "step": 29181
+ },
+ {
+ "epoch": 79.5149863760218,
+ "grad_norm": 1.7333450317382812,
+ "learning_rate": 2.121335491308417e-06,
+ "loss": 0.1676,
+ "step": 29182
+ },
+ {
+ "epoch": 79.51771117166213,
+ "grad_norm": 2.2695858478546143,
+ "learning_rate": 2.1207920410825543e-06,
+ "loss": 0.1686,
+ "step": 29183
+ },
+ {
+ "epoch": 79.52043596730245,
+ "grad_norm": 0.9860207438468933,
+ "learning_rate": 2.1202486522197063e-06,
+ "loss": 0.0115,
+ "step": 29184
+ },
+ {
+ "epoch": 79.52316076294278,
+ "grad_norm": 1.2998666763305664,
+ "learning_rate": 2.1197053247241116e-06,
+ "loss": 0.0415,
+ "step": 29185
+ },
+ {
+ "epoch": 79.52588555858311,
+ "grad_norm": 0.8609174489974976,
+ "learning_rate": 2.1191620585999973e-06,
+ "loss": 0.0107,
+ "step": 29186
+ },
+ {
+ "epoch": 79.52861035422343,
+ "grad_norm": 1.243141770362854,
+ "learning_rate": 2.1186188538515964e-06,
+ "loss": 0.0185,
+ "step": 29187
+ },
+ {
+ "epoch": 79.53133514986376,
+ "grad_norm": 0.9053519368171692,
+ "learning_rate": 2.118075710483134e-06,
+ "loss": 0.0112,
+ "step": 29188
+ },
+ {
+ "epoch": 79.53405994550408,
+ "grad_norm": 0.9603801369667053,
+ "learning_rate": 2.117532628498847e-06,
+ "loss": 0.0147,
+ "step": 29189
+ },
+ {
+ "epoch": 79.53678474114442,
+ "grad_norm": 2.2664780616760254,
+ "learning_rate": 2.1169896079029607e-06,
+ "loss": 0.0377,
+ "step": 29190
+ },
+ {
+ "epoch": 79.53950953678473,
+ "grad_norm": 0.821271538734436,
+ "learning_rate": 2.116446648699708e-06,
+ "loss": 0.0077,
+ "step": 29191
+ },
+ {
+ "epoch": 79.54223433242507,
+ "grad_norm": 1.4841398000717163,
+ "learning_rate": 2.1159037508933154e-06,
+ "loss": 0.0419,
+ "step": 29192
+ },
+ {
+ "epoch": 79.5449591280654,
+ "grad_norm": 2.0914313793182373,
+ "learning_rate": 2.115360914488007e-06,
+ "loss": 0.1872,
+ "step": 29193
+ },
+ {
+ "epoch": 79.54768392370572,
+ "grad_norm": 1.1163691282272339,
+ "learning_rate": 2.1148181394880175e-06,
+ "loss": 0.0561,
+ "step": 29194
+ },
+ {
+ "epoch": 79.55040871934605,
+ "grad_norm": 1.580741047859192,
+ "learning_rate": 2.1142754258975707e-06,
+ "loss": 0.0116,
+ "step": 29195
+ },
+ {
+ "epoch": 79.55313351498637,
+ "grad_norm": 0.7964393496513367,
+ "learning_rate": 2.1137327737208923e-06,
+ "loss": 0.0117,
+ "step": 29196
+ },
+ {
+ "epoch": 79.5558583106267,
+ "grad_norm": 1.3050127029418945,
+ "learning_rate": 2.113190182962207e-06,
+ "loss": 0.0441,
+ "step": 29197
+ },
+ {
+ "epoch": 79.55858310626704,
+ "grad_norm": 1.1055047512054443,
+ "learning_rate": 2.1126476536257446e-06,
+ "loss": 0.0454,
+ "step": 29198
+ },
+ {
+ "epoch": 79.56130790190736,
+ "grad_norm": 1.1803534030914307,
+ "learning_rate": 2.1121051857157294e-06,
+ "loss": 0.0179,
+ "step": 29199
+ },
+ {
+ "epoch": 79.56403269754769,
+ "grad_norm": 0.9861465096473694,
+ "learning_rate": 2.1115627792363823e-06,
+ "loss": 0.0091,
+ "step": 29200
+ },
+ {
+ "epoch": 79.566757493188,
+ "grad_norm": 1.3692988157272339,
+ "learning_rate": 2.1110204341919326e-06,
+ "loss": 0.01,
+ "step": 29201
+ },
+ {
+ "epoch": 79.56948228882834,
+ "grad_norm": 0.7749727368354797,
+ "learning_rate": 2.110478150586602e-06,
+ "loss": 0.0082,
+ "step": 29202
+ },
+ {
+ "epoch": 79.57220708446866,
+ "grad_norm": 0.927131712436676,
+ "learning_rate": 2.1099359284246144e-06,
+ "loss": 0.0797,
+ "step": 29203
+ },
+ {
+ "epoch": 79.57493188010899,
+ "grad_norm": 0.8538650274276733,
+ "learning_rate": 2.1093937677101884e-06,
+ "loss": 0.011,
+ "step": 29204
+ },
+ {
+ "epoch": 79.57765667574932,
+ "grad_norm": 1.304757833480835,
+ "learning_rate": 2.1088516684475526e-06,
+ "loss": 0.0191,
+ "step": 29205
+ },
+ {
+ "epoch": 79.58038147138964,
+ "grad_norm": 0.9125247597694397,
+ "learning_rate": 2.1083096306409255e-06,
+ "loss": 0.0163,
+ "step": 29206
+ },
+ {
+ "epoch": 79.58310626702998,
+ "grad_norm": 1.452116847038269,
+ "learning_rate": 2.107767654294528e-06,
+ "loss": 0.0493,
+ "step": 29207
+ },
+ {
+ "epoch": 79.5858310626703,
+ "grad_norm": 1.5297305583953857,
+ "learning_rate": 2.1072257394125828e-06,
+ "loss": 0.0088,
+ "step": 29208
+ },
+ {
+ "epoch": 79.58855585831063,
+ "grad_norm": 0.7883374691009521,
+ "learning_rate": 2.106683885999308e-06,
+ "loss": 0.0078,
+ "step": 29209
+ },
+ {
+ "epoch": 79.59128065395096,
+ "grad_norm": 1.293935775756836,
+ "learning_rate": 2.1061420940589273e-06,
+ "loss": 0.0843,
+ "step": 29210
+ },
+ {
+ "epoch": 79.59400544959128,
+ "grad_norm": 1.442724585533142,
+ "learning_rate": 2.105600363595658e-06,
+ "loss": 0.0822,
+ "step": 29211
+ },
+ {
+ "epoch": 79.59673024523161,
+ "grad_norm": 1.3485560417175293,
+ "learning_rate": 2.1050586946137184e-06,
+ "loss": 0.0211,
+ "step": 29212
+ },
+ {
+ "epoch": 79.59945504087193,
+ "grad_norm": 0.7431215643882751,
+ "learning_rate": 2.104517087117326e-06,
+ "loss": 0.0085,
+ "step": 29213
+ },
+ {
+ "epoch": 79.60217983651226,
+ "grad_norm": 1.149628758430481,
+ "learning_rate": 2.1039755411107023e-06,
+ "loss": 0.0165,
+ "step": 29214
+ },
+ {
+ "epoch": 79.60490463215258,
+ "grad_norm": 0.8642511367797852,
+ "learning_rate": 2.1034340565980636e-06,
+ "loss": 0.0096,
+ "step": 29215
+ },
+ {
+ "epoch": 79.60762942779292,
+ "grad_norm": 1.6082831621170044,
+ "learning_rate": 2.1028926335836253e-06,
+ "loss": 0.0174,
+ "step": 29216
+ },
+ {
+ "epoch": 79.61035422343325,
+ "grad_norm": 1.20810866355896,
+ "learning_rate": 2.1023512720716033e-06,
+ "loss": 0.0142,
+ "step": 29217
+ },
+ {
+ "epoch": 79.61307901907357,
+ "grad_norm": 1.5729997158050537,
+ "learning_rate": 2.1018099720662178e-06,
+ "loss": 0.0269,
+ "step": 29218
+ },
+ {
+ "epoch": 79.6158038147139,
+ "grad_norm": 1.0707792043685913,
+ "learning_rate": 2.101268733571682e-06,
+ "loss": 0.0856,
+ "step": 29219
+ },
+ {
+ "epoch": 79.61852861035422,
+ "grad_norm": 1.004106879234314,
+ "learning_rate": 2.100727556592211e-06,
+ "loss": 0.033,
+ "step": 29220
+ },
+ {
+ "epoch": 79.62125340599455,
+ "grad_norm": 0.8927044868469238,
+ "learning_rate": 2.1001864411320174e-06,
+ "loss": 0.0106,
+ "step": 29221
+ },
+ {
+ "epoch": 79.62397820163488,
+ "grad_norm": 1.3542925119400024,
+ "learning_rate": 2.0996453871953192e-06,
+ "loss": 0.0698,
+ "step": 29222
+ },
+ {
+ "epoch": 79.6267029972752,
+ "grad_norm": 0.9084672331809998,
+ "learning_rate": 2.0991043947863286e-06,
+ "loss": 0.0781,
+ "step": 29223
+ },
+ {
+ "epoch": 79.62942779291554,
+ "grad_norm": 0.7567002177238464,
+ "learning_rate": 2.0985634639092567e-06,
+ "loss": 0.0066,
+ "step": 29224
+ },
+ {
+ "epoch": 79.63215258855585,
+ "grad_norm": 0.7281543612480164,
+ "learning_rate": 2.0980225945683185e-06,
+ "loss": 0.0087,
+ "step": 29225
+ },
+ {
+ "epoch": 79.63487738419619,
+ "grad_norm": 2.042844533920288,
+ "learning_rate": 2.097481786767728e-06,
+ "loss": 0.1305,
+ "step": 29226
+ },
+ {
+ "epoch": 79.6376021798365,
+ "grad_norm": 0.29793015122413635,
+ "learning_rate": 2.096941040511695e-06,
+ "loss": 0.0034,
+ "step": 29227
+ },
+ {
+ "epoch": 79.64032697547684,
+ "grad_norm": 2.5971953868865967,
+ "learning_rate": 2.096400355804431e-06,
+ "loss": 0.0187,
+ "step": 29228
+ },
+ {
+ "epoch": 79.64305177111717,
+ "grad_norm": 1.7773102521896362,
+ "learning_rate": 2.095859732650144e-06,
+ "loss": 0.0824,
+ "step": 29229
+ },
+ {
+ "epoch": 79.64577656675749,
+ "grad_norm": 4.644961357116699,
+ "learning_rate": 2.0953191710530495e-06,
+ "loss": 0.1822,
+ "step": 29230
+ },
+ {
+ "epoch": 79.64850136239782,
+ "grad_norm": 0.8256014585494995,
+ "learning_rate": 2.0947786710173545e-06,
+ "loss": 0.0162,
+ "step": 29231
+ },
+ {
+ "epoch": 79.65122615803814,
+ "grad_norm": 1.1618348360061646,
+ "learning_rate": 2.09423823254727e-06,
+ "loss": 0.0099,
+ "step": 29232
+ },
+ {
+ "epoch": 79.65395095367847,
+ "grad_norm": 1.0794894695281982,
+ "learning_rate": 2.093697855647e-06,
+ "loss": 0.0389,
+ "step": 29233
+ },
+ {
+ "epoch": 79.65667574931881,
+ "grad_norm": 1.0620204210281372,
+ "learning_rate": 2.0931575403207594e-06,
+ "loss": 0.0319,
+ "step": 29234
+ },
+ {
+ "epoch": 79.65940054495913,
+ "grad_norm": 1.1903190612792969,
+ "learning_rate": 2.0926172865727534e-06,
+ "loss": 0.0097,
+ "step": 29235
+ },
+ {
+ "epoch": 79.66212534059946,
+ "grad_norm": 1.7892152070999146,
+ "learning_rate": 2.09207709440719e-06,
+ "loss": 0.0853,
+ "step": 29236
+ },
+ {
+ "epoch": 79.66485013623978,
+ "grad_norm": 0.9870406985282898,
+ "learning_rate": 2.091536963828272e-06,
+ "loss": 0.0066,
+ "step": 29237
+ },
+ {
+ "epoch": 79.66757493188011,
+ "grad_norm": 0.5809057950973511,
+ "learning_rate": 2.090996894840214e-06,
+ "loss": 0.0053,
+ "step": 29238
+ },
+ {
+ "epoch": 79.67029972752043,
+ "grad_norm": 1.020043134689331,
+ "learning_rate": 2.0904568874472166e-06,
+ "loss": 0.0093,
+ "step": 29239
+ },
+ {
+ "epoch": 79.67302452316076,
+ "grad_norm": 1.5048930644989014,
+ "learning_rate": 2.089916941653486e-06,
+ "loss": 0.0195,
+ "step": 29240
+ },
+ {
+ "epoch": 79.6757493188011,
+ "grad_norm": 0.902747392654419,
+ "learning_rate": 2.0893770574632253e-06,
+ "loss": 0.0115,
+ "step": 29241
+ },
+ {
+ "epoch": 79.67847411444141,
+ "grad_norm": 1.2804278135299683,
+ "learning_rate": 2.088837234880642e-06,
+ "loss": 0.0147,
+ "step": 29242
+ },
+ {
+ "epoch": 79.68119891008175,
+ "grad_norm": 1.2059317827224731,
+ "learning_rate": 2.0882974739099417e-06,
+ "loss": 0.0218,
+ "step": 29243
+ },
+ {
+ "epoch": 79.68392370572207,
+ "grad_norm": 1.2573162317276,
+ "learning_rate": 2.087757774555326e-06,
+ "loss": 0.0312,
+ "step": 29244
+ },
+ {
+ "epoch": 79.6866485013624,
+ "grad_norm": 0.38992828130722046,
+ "learning_rate": 2.0872181368209954e-06,
+ "loss": 0.0042,
+ "step": 29245
+ },
+ {
+ "epoch": 79.68937329700273,
+ "grad_norm": 1.9131171703338623,
+ "learning_rate": 2.0866785607111583e-06,
+ "loss": 0.0184,
+ "step": 29246
+ },
+ {
+ "epoch": 79.69209809264305,
+ "grad_norm": 1.4200972318649292,
+ "learning_rate": 2.0861390462300137e-06,
+ "loss": 0.0315,
+ "step": 29247
+ },
+ {
+ "epoch": 79.69482288828338,
+ "grad_norm": 1.394679069519043,
+ "learning_rate": 2.085599593381764e-06,
+ "loss": 0.0186,
+ "step": 29248
+ },
+ {
+ "epoch": 79.6975476839237,
+ "grad_norm": 1.1278467178344727,
+ "learning_rate": 2.085060202170607e-06,
+ "loss": 0.0253,
+ "step": 29249
+ },
+ {
+ "epoch": 79.70027247956403,
+ "grad_norm": 1.1667983531951904,
+ "learning_rate": 2.0845208726007484e-06,
+ "loss": 0.0202,
+ "step": 29250
+ },
+ {
+ "epoch": 79.70299727520435,
+ "grad_norm": 0.9689109921455383,
+ "learning_rate": 2.083981604676387e-06,
+ "loss": 0.011,
+ "step": 29251
+ },
+ {
+ "epoch": 79.70572207084469,
+ "grad_norm": 1.295931100845337,
+ "learning_rate": 2.083442398401723e-06,
+ "loss": 0.0204,
+ "step": 29252
+ },
+ {
+ "epoch": 79.70844686648502,
+ "grad_norm": 0.840076208114624,
+ "learning_rate": 2.082903253780951e-06,
+ "loss": 0.0052,
+ "step": 29253
+ },
+ {
+ "epoch": 79.71117166212534,
+ "grad_norm": 0.9254581928253174,
+ "learning_rate": 2.082364170818276e-06,
+ "loss": 0.0143,
+ "step": 29254
+ },
+ {
+ "epoch": 79.71389645776567,
+ "grad_norm": 1.2732149362564087,
+ "learning_rate": 2.081825149517895e-06,
+ "loss": 0.0187,
+ "step": 29255
+ },
+ {
+ "epoch": 79.71662125340599,
+ "grad_norm": 0.7980197668075562,
+ "learning_rate": 2.0812861898840043e-06,
+ "loss": 0.0097,
+ "step": 29256
+ },
+ {
+ "epoch": 79.71934604904632,
+ "grad_norm": 1.3658030033111572,
+ "learning_rate": 2.0807472919207993e-06,
+ "loss": 0.0436,
+ "step": 29257
+ },
+ {
+ "epoch": 79.72207084468666,
+ "grad_norm": 1.74403977394104,
+ "learning_rate": 2.080208455632482e-06,
+ "loss": 0.0521,
+ "step": 29258
+ },
+ {
+ "epoch": 79.72479564032697,
+ "grad_norm": 0.6979853510856628,
+ "learning_rate": 2.079669681023244e-06,
+ "loss": 0.0063,
+ "step": 29259
+ },
+ {
+ "epoch": 79.7275204359673,
+ "grad_norm": 0.7376928925514221,
+ "learning_rate": 2.079130968097287e-06,
+ "loss": 0.0093,
+ "step": 29260
+ },
+ {
+ "epoch": 79.73024523160763,
+ "grad_norm": 1.1796188354492188,
+ "learning_rate": 2.078592316858802e-06,
+ "loss": 0.0077,
+ "step": 29261
+ },
+ {
+ "epoch": 79.73297002724796,
+ "grad_norm": 0.9848873019218445,
+ "learning_rate": 2.0780537273119826e-06,
+ "loss": 0.0092,
+ "step": 29262
+ },
+ {
+ "epoch": 79.73569482288828,
+ "grad_norm": 0.7650244832038879,
+ "learning_rate": 2.077515199461029e-06,
+ "loss": 0.0055,
+ "step": 29263
+ },
+ {
+ "epoch": 79.73841961852861,
+ "grad_norm": 1.372969627380371,
+ "learning_rate": 2.076976733310131e-06,
+ "loss": 0.0518,
+ "step": 29264
+ },
+ {
+ "epoch": 79.74114441416894,
+ "grad_norm": 1.371038794517517,
+ "learning_rate": 2.0764383288634813e-06,
+ "loss": 0.0281,
+ "step": 29265
+ },
+ {
+ "epoch": 79.74386920980926,
+ "grad_norm": 1.316696286201477,
+ "learning_rate": 2.0758999861252783e-06,
+ "loss": 0.0093,
+ "step": 29266
+ },
+ {
+ "epoch": 79.7465940054496,
+ "grad_norm": 1.0388160943984985,
+ "learning_rate": 2.0753617050997097e-06,
+ "loss": 0.0097,
+ "step": 29267
+ },
+ {
+ "epoch": 79.74931880108991,
+ "grad_norm": 0.8972904682159424,
+ "learning_rate": 2.07482348579097e-06,
+ "loss": 0.0074,
+ "step": 29268
+ },
+ {
+ "epoch": 79.75204359673025,
+ "grad_norm": 0.9584097266197205,
+ "learning_rate": 2.074285328203247e-06,
+ "loss": 0.0129,
+ "step": 29269
+ },
+ {
+ "epoch": 79.75476839237058,
+ "grad_norm": 2.444329023361206,
+ "learning_rate": 2.073747232340737e-06,
+ "loss": 0.0518,
+ "step": 29270
+ },
+ {
+ "epoch": 79.7574931880109,
+ "grad_norm": 1.3425185680389404,
+ "learning_rate": 2.0732091982076286e-06,
+ "loss": 0.0268,
+ "step": 29271
+ },
+ {
+ "epoch": 79.76021798365123,
+ "grad_norm": 1.3352030515670776,
+ "learning_rate": 2.072671225808113e-06,
+ "loss": 0.02,
+ "step": 29272
+ },
+ {
+ "epoch": 79.76294277929155,
+ "grad_norm": 1.6382094621658325,
+ "learning_rate": 2.072133315146375e-06,
+ "loss": 0.0484,
+ "step": 29273
+ },
+ {
+ "epoch": 79.76566757493188,
+ "grad_norm": 1.3390671014785767,
+ "learning_rate": 2.07159546622661e-06,
+ "loss": 0.0797,
+ "step": 29274
+ },
+ {
+ "epoch": 79.7683923705722,
+ "grad_norm": 1.3787811994552612,
+ "learning_rate": 2.0710576790530046e-06,
+ "loss": 0.1009,
+ "step": 29275
+ },
+ {
+ "epoch": 79.77111716621253,
+ "grad_norm": 1.5386314392089844,
+ "learning_rate": 2.070519953629745e-06,
+ "loss": 0.0197,
+ "step": 29276
+ },
+ {
+ "epoch": 79.77384196185287,
+ "grad_norm": 1.2648661136627197,
+ "learning_rate": 2.069982289961022e-06,
+ "loss": 0.0175,
+ "step": 29277
+ },
+ {
+ "epoch": 79.77656675749319,
+ "grad_norm": 1.5869289636611938,
+ "learning_rate": 2.06944468805102e-06,
+ "loss": 0.0913,
+ "step": 29278
+ },
+ {
+ "epoch": 79.77929155313352,
+ "grad_norm": 1.7687273025512695,
+ "learning_rate": 2.0689071479039303e-06,
+ "loss": 0.0637,
+ "step": 29279
+ },
+ {
+ "epoch": 79.78201634877384,
+ "grad_norm": 1.0824594497680664,
+ "learning_rate": 2.0683696695239364e-06,
+ "loss": 0.019,
+ "step": 29280
+ },
+ {
+ "epoch": 79.78474114441417,
+ "grad_norm": 1.4007899761199951,
+ "learning_rate": 2.0678322529152238e-06,
+ "loss": 0.0132,
+ "step": 29281
+ },
+ {
+ "epoch": 79.7874659400545,
+ "grad_norm": 0.7989504337310791,
+ "learning_rate": 2.067294898081975e-06,
+ "loss": 0.0075,
+ "step": 29282
+ },
+ {
+ "epoch": 79.79019073569482,
+ "grad_norm": 1.5593643188476562,
+ "learning_rate": 2.066757605028382e-06,
+ "loss": 0.0133,
+ "step": 29283
+ },
+ {
+ "epoch": 79.79291553133515,
+ "grad_norm": 1.430261254310608,
+ "learning_rate": 2.066220373758625e-06,
+ "loss": 0.0198,
+ "step": 29284
+ },
+ {
+ "epoch": 79.79564032697547,
+ "grad_norm": 1.7103241682052612,
+ "learning_rate": 2.065683204276889e-06,
+ "loss": 0.038,
+ "step": 29285
+ },
+ {
+ "epoch": 79.7983651226158,
+ "grad_norm": 2.0240578651428223,
+ "learning_rate": 2.0651460965873537e-06,
+ "loss": 0.0381,
+ "step": 29286
+ },
+ {
+ "epoch": 79.80108991825612,
+ "grad_norm": 1.6694669723510742,
+ "learning_rate": 2.0646090506942084e-06,
+ "loss": 0.0896,
+ "step": 29287
+ },
+ {
+ "epoch": 79.80381471389646,
+ "grad_norm": 1.1031310558319092,
+ "learning_rate": 2.0640720666016314e-06,
+ "loss": 0.0112,
+ "step": 29288
+ },
+ {
+ "epoch": 79.80653950953679,
+ "grad_norm": 0.9820802807807922,
+ "learning_rate": 2.063535144313804e-06,
+ "loss": 0.0164,
+ "step": 29289
+ },
+ {
+ "epoch": 79.80926430517711,
+ "grad_norm": 0.9107992649078369,
+ "learning_rate": 2.0629982838349117e-06,
+ "loss": 0.0132,
+ "step": 29290
+ },
+ {
+ "epoch": 79.81198910081744,
+ "grad_norm": 1.295307993888855,
+ "learning_rate": 2.0624614851691327e-06,
+ "loss": 0.0267,
+ "step": 29291
+ },
+ {
+ "epoch": 79.81471389645776,
+ "grad_norm": 1.1955751180648804,
+ "learning_rate": 2.0619247483206495e-06,
+ "loss": 0.0274,
+ "step": 29292
+ },
+ {
+ "epoch": 79.8174386920981,
+ "grad_norm": 0.6758440136909485,
+ "learning_rate": 2.0613880732936376e-06,
+ "loss": 0.0087,
+ "step": 29293
+ },
+ {
+ "epoch": 79.82016348773843,
+ "grad_norm": 1.005573034286499,
+ "learning_rate": 2.0608514600922793e-06,
+ "loss": 0.0145,
+ "step": 29294
+ },
+ {
+ "epoch": 79.82288828337875,
+ "grad_norm": 1.0521366596221924,
+ "learning_rate": 2.0603149087207576e-06,
+ "loss": 0.0147,
+ "step": 29295
+ },
+ {
+ "epoch": 79.82561307901908,
+ "grad_norm": 1.2656145095825195,
+ "learning_rate": 2.0597784191832483e-06,
+ "loss": 0.0957,
+ "step": 29296
+ },
+ {
+ "epoch": 79.8283378746594,
+ "grad_norm": 1.4980597496032715,
+ "learning_rate": 2.059241991483928e-06,
+ "loss": 0.0997,
+ "step": 29297
+ },
+ {
+ "epoch": 79.83106267029973,
+ "grad_norm": 1.2861464023590088,
+ "learning_rate": 2.058705625626973e-06,
+ "loss": 0.0454,
+ "step": 29298
+ },
+ {
+ "epoch": 79.83378746594005,
+ "grad_norm": 1.3388468027114868,
+ "learning_rate": 2.0581693216165668e-06,
+ "loss": 0.0102,
+ "step": 29299
+ },
+ {
+ "epoch": 79.83651226158038,
+ "grad_norm": 0.834946870803833,
+ "learning_rate": 2.057633079456881e-06,
+ "loss": 0.0099,
+ "step": 29300
+ },
+ {
+ "epoch": 79.83923705722071,
+ "grad_norm": 1.164642095565796,
+ "learning_rate": 2.0570968991520933e-06,
+ "loss": 0.0118,
+ "step": 29301
+ },
+ {
+ "epoch": 79.84196185286103,
+ "grad_norm": 1.1183366775512695,
+ "learning_rate": 2.056560780706377e-06,
+ "loss": 0.0256,
+ "step": 29302
+ },
+ {
+ "epoch": 79.84468664850137,
+ "grad_norm": 0.6914745569229126,
+ "learning_rate": 2.0560247241239127e-06,
+ "loss": 0.011,
+ "step": 29303
+ },
+ {
+ "epoch": 79.84741144414168,
+ "grad_norm": 4.168025493621826,
+ "learning_rate": 2.0554887294088723e-06,
+ "loss": 0.0314,
+ "step": 29304
+ },
+ {
+ "epoch": 79.85013623978202,
+ "grad_norm": 1.1977612972259521,
+ "learning_rate": 2.0549527965654293e-06,
+ "loss": 0.0455,
+ "step": 29305
+ },
+ {
+ "epoch": 79.85286103542235,
+ "grad_norm": 0.741605281829834,
+ "learning_rate": 2.0544169255977563e-06,
+ "loss": 0.0084,
+ "step": 29306
+ },
+ {
+ "epoch": 79.85558583106267,
+ "grad_norm": 1.0516674518585205,
+ "learning_rate": 2.0538811165100303e-06,
+ "loss": 0.0269,
+ "step": 29307
+ },
+ {
+ "epoch": 79.858310626703,
+ "grad_norm": 1.0597370862960815,
+ "learning_rate": 2.053345369306423e-06,
+ "loss": 0.1313,
+ "step": 29308
+ },
+ {
+ "epoch": 79.86103542234332,
+ "grad_norm": 1.8599299192428589,
+ "learning_rate": 2.052809683991103e-06,
+ "loss": 0.0546,
+ "step": 29309
+ },
+ {
+ "epoch": 79.86376021798365,
+ "grad_norm": 1.4881922006607056,
+ "learning_rate": 2.052274060568249e-06,
+ "loss": 0.0091,
+ "step": 29310
+ },
+ {
+ "epoch": 79.86648501362397,
+ "grad_norm": 1.1372556686401367,
+ "learning_rate": 2.0517384990420253e-06,
+ "loss": 0.0086,
+ "step": 29311
+ },
+ {
+ "epoch": 79.8692098092643,
+ "grad_norm": 1.198376178741455,
+ "learning_rate": 2.0512029994166093e-06,
+ "loss": 0.0159,
+ "step": 29312
+ },
+ {
+ "epoch": 79.87193460490464,
+ "grad_norm": 1.4751152992248535,
+ "learning_rate": 2.050667561696168e-06,
+ "loss": 0.0243,
+ "step": 29313
+ },
+ {
+ "epoch": 79.87465940054496,
+ "grad_norm": 0.9092456102371216,
+ "learning_rate": 2.05013218588487e-06,
+ "loss": 0.0124,
+ "step": 29314
+ },
+ {
+ "epoch": 79.87738419618529,
+ "grad_norm": 0.9928335547447205,
+ "learning_rate": 2.0495968719868895e-06,
+ "loss": 0.0118,
+ "step": 29315
+ },
+ {
+ "epoch": 79.88010899182561,
+ "grad_norm": 1.0496717691421509,
+ "learning_rate": 2.049061620006392e-06,
+ "loss": 0.1196,
+ "step": 29316
+ },
+ {
+ "epoch": 79.88283378746594,
+ "grad_norm": 0.978094756603241,
+ "learning_rate": 2.048526429947547e-06,
+ "loss": 0.0155,
+ "step": 29317
+ },
+ {
+ "epoch": 79.88555858310627,
+ "grad_norm": 1.3715877532958984,
+ "learning_rate": 2.0479913018145194e-06,
+ "loss": 0.0069,
+ "step": 29318
+ },
+ {
+ "epoch": 79.88828337874659,
+ "grad_norm": 0.9483415484428406,
+ "learning_rate": 2.0474562356114824e-06,
+ "loss": 0.0131,
+ "step": 29319
+ },
+ {
+ "epoch": 79.89100817438693,
+ "grad_norm": 1.4756592512130737,
+ "learning_rate": 2.0469212313426e-06,
+ "loss": 0.0202,
+ "step": 29320
+ },
+ {
+ "epoch": 79.89373297002724,
+ "grad_norm": 2.021817922592163,
+ "learning_rate": 2.04638628901204e-06,
+ "loss": 0.0482,
+ "step": 29321
+ },
+ {
+ "epoch": 79.89645776566758,
+ "grad_norm": 0.698173463344574,
+ "learning_rate": 2.0458514086239644e-06,
+ "loss": 0.0095,
+ "step": 29322
+ },
+ {
+ "epoch": 79.8991825613079,
+ "grad_norm": 1.6584523916244507,
+ "learning_rate": 2.0453165901825445e-06,
+ "loss": 0.0585,
+ "step": 29323
+ },
+ {
+ "epoch": 79.90190735694823,
+ "grad_norm": 1.5343987941741943,
+ "learning_rate": 2.0447818336919433e-06,
+ "loss": 0.0263,
+ "step": 29324
+ },
+ {
+ "epoch": 79.90463215258856,
+ "grad_norm": 0.6557533144950867,
+ "learning_rate": 2.0442471391563244e-06,
+ "loss": 0.008,
+ "step": 29325
+ },
+ {
+ "epoch": 79.90735694822888,
+ "grad_norm": 0.7059116363525391,
+ "learning_rate": 2.0437125065798515e-06,
+ "loss": 0.0073,
+ "step": 29326
+ },
+ {
+ "epoch": 79.91008174386921,
+ "grad_norm": 1.1087794303894043,
+ "learning_rate": 2.04317793596669e-06,
+ "loss": 0.012,
+ "step": 29327
+ },
+ {
+ "epoch": 79.91280653950953,
+ "grad_norm": 1.2966820001602173,
+ "learning_rate": 2.0426434273210016e-06,
+ "loss": 0.0112,
+ "step": 29328
+ },
+ {
+ "epoch": 79.91553133514986,
+ "grad_norm": 1.2646733522415161,
+ "learning_rate": 2.042108980646953e-06,
+ "loss": 0.1978,
+ "step": 29329
+ },
+ {
+ "epoch": 79.9182561307902,
+ "grad_norm": 1.367485761642456,
+ "learning_rate": 2.0415745959486997e-06,
+ "loss": 0.0102,
+ "step": 29330
+ },
+ {
+ "epoch": 79.92098092643052,
+ "grad_norm": 1.3735036849975586,
+ "learning_rate": 2.0410402732304115e-06,
+ "loss": 0.02,
+ "step": 29331
+ },
+ {
+ "epoch": 79.92370572207085,
+ "grad_norm": 1.1617002487182617,
+ "learning_rate": 2.040506012496244e-06,
+ "loss": 0.0496,
+ "step": 29332
+ },
+ {
+ "epoch": 79.92643051771117,
+ "grad_norm": 1.1325618028640747,
+ "learning_rate": 2.03997181375036e-06,
+ "loss": 0.014,
+ "step": 29333
+ },
+ {
+ "epoch": 79.9291553133515,
+ "grad_norm": 0.7500057816505432,
+ "learning_rate": 2.039437676996917e-06,
+ "loss": 0.009,
+ "step": 29334
+ },
+ {
+ "epoch": 79.93188010899182,
+ "grad_norm": 2.184562921524048,
+ "learning_rate": 2.0389036022400786e-06,
+ "loss": 0.1192,
+ "step": 29335
+ },
+ {
+ "epoch": 79.93460490463215,
+ "grad_norm": 1.3050280809402466,
+ "learning_rate": 2.038369589484004e-06,
+ "loss": 0.0264,
+ "step": 29336
+ },
+ {
+ "epoch": 79.93732970027249,
+ "grad_norm": 0.6077647805213928,
+ "learning_rate": 2.03783563873285e-06,
+ "loss": 0.0088,
+ "step": 29337
+ },
+ {
+ "epoch": 79.9400544959128,
+ "grad_norm": 1.5563544034957886,
+ "learning_rate": 2.0373017499907743e-06,
+ "loss": 0.0456,
+ "step": 29338
+ },
+ {
+ "epoch": 79.94277929155314,
+ "grad_norm": 1.1363805532455444,
+ "learning_rate": 2.0367679232619385e-06,
+ "loss": 0.0287,
+ "step": 29339
+ },
+ {
+ "epoch": 79.94550408719346,
+ "grad_norm": 1.7199121713638306,
+ "learning_rate": 2.0362341585504974e-06,
+ "loss": 0.0677,
+ "step": 29340
+ },
+ {
+ "epoch": 79.94822888283379,
+ "grad_norm": 1.016003131866455,
+ "learning_rate": 2.0357004558606087e-06,
+ "loss": 0.0104,
+ "step": 29341
+ },
+ {
+ "epoch": 79.95095367847412,
+ "grad_norm": 0.65450119972229,
+ "learning_rate": 2.0351668151964253e-06,
+ "loss": 0.0059,
+ "step": 29342
+ },
+ {
+ "epoch": 79.95367847411444,
+ "grad_norm": 1.4426316022872925,
+ "learning_rate": 2.034633236562109e-06,
+ "loss": 0.0274,
+ "step": 29343
+ },
+ {
+ "epoch": 79.95640326975477,
+ "grad_norm": 0.6666722297668457,
+ "learning_rate": 2.0340997199618128e-06,
+ "loss": 0.0054,
+ "step": 29344
+ },
+ {
+ "epoch": 79.95912806539509,
+ "grad_norm": 1.1145573854446411,
+ "learning_rate": 2.03356626539969e-06,
+ "loss": 0.0209,
+ "step": 29345
+ },
+ {
+ "epoch": 79.96185286103542,
+ "grad_norm": 0.7060474157333374,
+ "learning_rate": 2.0330328728798988e-06,
+ "loss": 0.0068,
+ "step": 29346
+ },
+ {
+ "epoch": 79.96457765667574,
+ "grad_norm": 0.9081828594207764,
+ "learning_rate": 2.0324995424065884e-06,
+ "loss": 0.0079,
+ "step": 29347
+ },
+ {
+ "epoch": 79.96730245231608,
+ "grad_norm": 1.3175088167190552,
+ "learning_rate": 2.0319662739839174e-06,
+ "loss": 0.0325,
+ "step": 29348
+ },
+ {
+ "epoch": 79.97002724795641,
+ "grad_norm": 0.8860103487968445,
+ "learning_rate": 2.0314330676160364e-06,
+ "loss": 0.0082,
+ "step": 29349
+ },
+ {
+ "epoch": 79.97275204359673,
+ "grad_norm": 0.9626463651657104,
+ "learning_rate": 2.030899923307097e-06,
+ "loss": 0.0076,
+ "step": 29350
+ },
+ {
+ "epoch": 79.97547683923706,
+ "grad_norm": 1.900956153869629,
+ "learning_rate": 2.0303668410612544e-06,
+ "loss": 0.0759,
+ "step": 29351
+ },
+ {
+ "epoch": 79.97820163487738,
+ "grad_norm": 1.6359031200408936,
+ "learning_rate": 2.0298338208826574e-06,
+ "loss": 0.0787,
+ "step": 29352
+ },
+ {
+ "epoch": 79.98092643051771,
+ "grad_norm": 1.5763801336288452,
+ "learning_rate": 2.0293008627754597e-06,
+ "loss": 0.0776,
+ "step": 29353
+ },
+ {
+ "epoch": 79.98365122615803,
+ "grad_norm": 1.2358347177505493,
+ "learning_rate": 2.0287679667438063e-06,
+ "loss": 0.0302,
+ "step": 29354
+ },
+ {
+ "epoch": 79.98637602179836,
+ "grad_norm": 1.5034798383712769,
+ "learning_rate": 2.028235132791855e-06,
+ "loss": 0.017,
+ "step": 29355
+ },
+ {
+ "epoch": 79.9891008174387,
+ "grad_norm": 1.2307544946670532,
+ "learning_rate": 2.027702360923751e-06,
+ "loss": 0.0115,
+ "step": 29356
+ },
+ {
+ "epoch": 79.99182561307902,
+ "grad_norm": 1.0317033529281616,
+ "learning_rate": 2.0271696511436457e-06,
+ "loss": 0.0182,
+ "step": 29357
+ },
+ {
+ "epoch": 79.99455040871935,
+ "grad_norm": 1.573950171470642,
+ "learning_rate": 2.0266370034556824e-06,
+ "loss": 0.0632,
+ "step": 29358
+ },
+ {
+ "epoch": 79.99727520435967,
+ "grad_norm": 1.1530307531356812,
+ "learning_rate": 2.0261044178640166e-06,
+ "loss": 0.0442,
+ "step": 29359
+ },
+ {
+ "epoch": 80.0,
+ "grad_norm": 2.015204906463623,
+ "learning_rate": 2.025571894372794e-06,
+ "loss": 0.1319,
+ "step": 29360
+ },
+ {
+ "epoch": 80.00272479564033,
+ "grad_norm": 1.5586072206497192,
+ "learning_rate": 2.02503943298616e-06,
+ "loss": 0.0546,
+ "step": 29361
+ },
+ {
+ "epoch": 80.00544959128065,
+ "grad_norm": 0.6502701044082642,
+ "learning_rate": 2.0245070337082596e-06,
+ "loss": 0.0086,
+ "step": 29362
+ },
+ {
+ "epoch": 80.00817438692098,
+ "grad_norm": 1.1211776733398438,
+ "learning_rate": 2.0239746965432417e-06,
+ "loss": 0.0105,
+ "step": 29363
+ },
+ {
+ "epoch": 80.0108991825613,
+ "grad_norm": 3.7588419914245605,
+ "learning_rate": 2.023442421495255e-06,
+ "loss": 0.0688,
+ "step": 29364
+ },
+ {
+ "epoch": 80.01362397820164,
+ "grad_norm": 1.0296452045440674,
+ "learning_rate": 2.0229102085684416e-06,
+ "loss": 0.1068,
+ "step": 29365
+ },
+ {
+ "epoch": 80.01634877384195,
+ "grad_norm": 0.6821953058242798,
+ "learning_rate": 2.022378057766947e-06,
+ "loss": 0.012,
+ "step": 29366
+ },
+ {
+ "epoch": 80.01907356948229,
+ "grad_norm": 1.3459632396697998,
+ "learning_rate": 2.021845969094913e-06,
+ "loss": 0.0172,
+ "step": 29367
+ },
+ {
+ "epoch": 80.02179836512262,
+ "grad_norm": 1.0372462272644043,
+ "learning_rate": 2.0213139425564888e-06,
+ "loss": 0.016,
+ "step": 29368
+ },
+ {
+ "epoch": 80.02452316076294,
+ "grad_norm": 1.862581729888916,
+ "learning_rate": 2.0207819781558137e-06,
+ "loss": 0.0459,
+ "step": 29369
+ },
+ {
+ "epoch": 80.02724795640327,
+ "grad_norm": 0.5169583559036255,
+ "learning_rate": 2.0202500758970323e-06,
+ "loss": 0.0052,
+ "step": 29370
+ },
+ {
+ "epoch": 80.02997275204359,
+ "grad_norm": 1.1533674001693726,
+ "learning_rate": 2.0197182357842836e-06,
+ "loss": 0.0147,
+ "step": 29371
+ },
+ {
+ "epoch": 80.03269754768392,
+ "grad_norm": 0.8773316144943237,
+ "learning_rate": 2.0191864578217155e-06,
+ "loss": 0.0116,
+ "step": 29372
+ },
+ {
+ "epoch": 80.03542234332426,
+ "grad_norm": 1.3246209621429443,
+ "learning_rate": 2.0186547420134663e-06,
+ "loss": 0.0103,
+ "step": 29373
+ },
+ {
+ "epoch": 80.03814713896458,
+ "grad_norm": 0.8049681186676025,
+ "learning_rate": 2.018123088363675e-06,
+ "loss": 0.009,
+ "step": 29374
+ },
+ {
+ "epoch": 80.04087193460491,
+ "grad_norm": 1.389654278755188,
+ "learning_rate": 2.017591496876485e-06,
+ "loss": 0.0256,
+ "step": 29375
+ },
+ {
+ "epoch": 80.04359673024523,
+ "grad_norm": 0.8661041855812073,
+ "learning_rate": 2.0170599675560376e-06,
+ "loss": 0.01,
+ "step": 29376
+ },
+ {
+ "epoch": 80.04632152588556,
+ "grad_norm": 0.9318989515304565,
+ "learning_rate": 2.0165285004064692e-06,
+ "loss": 0.0094,
+ "step": 29377
+ },
+ {
+ "epoch": 80.04904632152588,
+ "grad_norm": 1.074893593788147,
+ "learning_rate": 2.0159970954319173e-06,
+ "loss": 0.0075,
+ "step": 29378
+ },
+ {
+ "epoch": 80.05177111716621,
+ "grad_norm": 1.1771177053451538,
+ "learning_rate": 2.0154657526365227e-06,
+ "loss": 0.0108,
+ "step": 29379
+ },
+ {
+ "epoch": 80.05449591280654,
+ "grad_norm": 0.9986112713813782,
+ "learning_rate": 2.014934472024427e-06,
+ "loss": 0.0115,
+ "step": 29380
+ },
+ {
+ "epoch": 80.05722070844686,
+ "grad_norm": 1.2946001291275024,
+ "learning_rate": 2.014403253599765e-06,
+ "loss": 0.1359,
+ "step": 29381
+ },
+ {
+ "epoch": 80.0599455040872,
+ "grad_norm": 0.661898672580719,
+ "learning_rate": 2.0138720973666727e-06,
+ "loss": 0.0069,
+ "step": 29382
+ },
+ {
+ "epoch": 80.06267029972751,
+ "grad_norm": 0.46940088272094727,
+ "learning_rate": 2.0133410033292854e-06,
+ "loss": 0.007,
+ "step": 29383
+ },
+ {
+ "epoch": 80.06539509536785,
+ "grad_norm": 1.1467759609222412,
+ "learning_rate": 2.0128099714917436e-06,
+ "loss": 0.048,
+ "step": 29384
+ },
+ {
+ "epoch": 80.06811989100818,
+ "grad_norm": 1.3775314092636108,
+ "learning_rate": 2.0122790018581805e-06,
+ "loss": 0.0839,
+ "step": 29385
+ },
+ {
+ "epoch": 80.0708446866485,
+ "grad_norm": 1.1462275981903076,
+ "learning_rate": 2.0117480944327317e-06,
+ "loss": 0.0187,
+ "step": 29386
+ },
+ {
+ "epoch": 80.07356948228883,
+ "grad_norm": 0.46274101734161377,
+ "learning_rate": 2.011217249219529e-06,
+ "loss": 0.0031,
+ "step": 29387
+ },
+ {
+ "epoch": 80.07629427792915,
+ "grad_norm": 1.6515474319458008,
+ "learning_rate": 2.010686466222712e-06,
+ "loss": 0.035,
+ "step": 29388
+ },
+ {
+ "epoch": 80.07901907356948,
+ "grad_norm": 1.345141887664795,
+ "learning_rate": 2.0101557454464115e-06,
+ "loss": 0.0463,
+ "step": 29389
+ },
+ {
+ "epoch": 80.0817438692098,
+ "grad_norm": 1.3571465015411377,
+ "learning_rate": 2.009625086894761e-06,
+ "loss": 0.015,
+ "step": 29390
+ },
+ {
+ "epoch": 80.08446866485014,
+ "grad_norm": 1.5761598348617554,
+ "learning_rate": 2.0090944905718913e-06,
+ "loss": 0.0778,
+ "step": 29391
+ },
+ {
+ "epoch": 80.08719346049047,
+ "grad_norm": 1.3790258169174194,
+ "learning_rate": 2.0085639564819383e-06,
+ "loss": 0.0436,
+ "step": 29392
+ },
+ {
+ "epoch": 80.08991825613079,
+ "grad_norm": 1.2920981645584106,
+ "learning_rate": 2.0080334846290317e-06,
+ "loss": 0.0189,
+ "step": 29393
+ },
+ {
+ "epoch": 80.09264305177112,
+ "grad_norm": 0.9131345152854919,
+ "learning_rate": 2.007503075017303e-06,
+ "loss": 0.0147,
+ "step": 29394
+ },
+ {
+ "epoch": 80.09536784741144,
+ "grad_norm": 1.0548608303070068,
+ "learning_rate": 2.006972727650881e-06,
+ "loss": 0.0098,
+ "step": 29395
+ },
+ {
+ "epoch": 80.09809264305177,
+ "grad_norm": 1.4403491020202637,
+ "learning_rate": 2.0064424425339014e-06,
+ "loss": 0.0223,
+ "step": 29396
+ },
+ {
+ "epoch": 80.1008174386921,
+ "grad_norm": 1.2053241729736328,
+ "learning_rate": 2.0059122196704872e-06,
+ "loss": 0.0119,
+ "step": 29397
+ },
+ {
+ "epoch": 80.10354223433242,
+ "grad_norm": 1.7920397520065308,
+ "learning_rate": 2.0053820590647744e-06,
+ "loss": 0.0308,
+ "step": 29398
+ },
+ {
+ "epoch": 80.10626702997276,
+ "grad_norm": 1.0753004550933838,
+ "learning_rate": 2.004851960720886e-06,
+ "loss": 0.0155,
+ "step": 29399
+ },
+ {
+ "epoch": 80.10899182561307,
+ "grad_norm": 1.685126543045044,
+ "learning_rate": 2.0043219246429555e-06,
+ "loss": 0.0167,
+ "step": 29400
+ },
+ {
+ "epoch": 80.11171662125341,
+ "grad_norm": 1.503054141998291,
+ "learning_rate": 2.003791950835109e-06,
+ "loss": 0.0155,
+ "step": 29401
+ },
+ {
+ "epoch": 80.11444141689373,
+ "grad_norm": 2.2649338245391846,
+ "learning_rate": 2.0032620393014734e-06,
+ "loss": 0.0166,
+ "step": 29402
+ },
+ {
+ "epoch": 80.11716621253406,
+ "grad_norm": 0.796226978302002,
+ "learning_rate": 2.0027321900461726e-06,
+ "loss": 0.0069,
+ "step": 29403
+ },
+ {
+ "epoch": 80.11989100817439,
+ "grad_norm": 1.6397318840026855,
+ "learning_rate": 2.002202403073339e-06,
+ "loss": 0.0236,
+ "step": 29404
+ },
+ {
+ "epoch": 80.12261580381471,
+ "grad_norm": 2.1915242671966553,
+ "learning_rate": 2.0016726783870966e-06,
+ "loss": 0.06,
+ "step": 29405
+ },
+ {
+ "epoch": 80.12534059945504,
+ "grad_norm": 1.6962716579437256,
+ "learning_rate": 2.001143015991569e-06,
+ "loss": 0.0325,
+ "step": 29406
+ },
+ {
+ "epoch": 80.12806539509536,
+ "grad_norm": 1.7790743112564087,
+ "learning_rate": 2.0006134158908796e-06,
+ "loss": 0.0853,
+ "step": 29407
+ },
+ {
+ "epoch": 80.1307901907357,
+ "grad_norm": 1.4656898975372314,
+ "learning_rate": 2.0000838780891573e-06,
+ "loss": 0.0154,
+ "step": 29408
+ },
+ {
+ "epoch": 80.13351498637603,
+ "grad_norm": 1.032019853591919,
+ "learning_rate": 1.9995544025905254e-06,
+ "loss": 0.0133,
+ "step": 29409
+ },
+ {
+ "epoch": 80.13623978201635,
+ "grad_norm": 1.7807748317718506,
+ "learning_rate": 1.9990249893991064e-06,
+ "loss": 0.0173,
+ "step": 29410
+ },
+ {
+ "epoch": 80.13896457765668,
+ "grad_norm": 1.5537704229354858,
+ "learning_rate": 1.9984956385190204e-06,
+ "loss": 0.0413,
+ "step": 29411
+ },
+ {
+ "epoch": 80.141689373297,
+ "grad_norm": 1.315130591392517,
+ "learning_rate": 1.9979663499543956e-06,
+ "loss": 0.013,
+ "step": 29412
+ },
+ {
+ "epoch": 80.14441416893733,
+ "grad_norm": 1.3815863132476807,
+ "learning_rate": 1.9974371237093505e-06,
+ "loss": 0.0114,
+ "step": 29413
+ },
+ {
+ "epoch": 80.14713896457765,
+ "grad_norm": 1.5789473056793213,
+ "learning_rate": 1.996907959788006e-06,
+ "loss": 0.0539,
+ "step": 29414
+ },
+ {
+ "epoch": 80.14986376021798,
+ "grad_norm": 1.7675691843032837,
+ "learning_rate": 1.9963788581944832e-06,
+ "loss": 0.0226,
+ "step": 29415
+ },
+ {
+ "epoch": 80.15258855585832,
+ "grad_norm": 1.0113333463668823,
+ "learning_rate": 1.995849818932908e-06,
+ "loss": 0.0062,
+ "step": 29416
+ },
+ {
+ "epoch": 80.15531335149863,
+ "grad_norm": 1.1554268598556519,
+ "learning_rate": 1.9953208420073965e-06,
+ "loss": 0.0079,
+ "step": 29417
+ },
+ {
+ "epoch": 80.15803814713897,
+ "grad_norm": 0.7278926372528076,
+ "learning_rate": 1.9947919274220693e-06,
+ "loss": 0.0092,
+ "step": 29418
+ },
+ {
+ "epoch": 80.16076294277929,
+ "grad_norm": 1.06137216091156,
+ "learning_rate": 1.994263075181042e-06,
+ "loss": 0.0132,
+ "step": 29419
+ },
+ {
+ "epoch": 80.16348773841962,
+ "grad_norm": 1.3164128065109253,
+ "learning_rate": 1.993734285288439e-06,
+ "loss": 0.0446,
+ "step": 29420
+ },
+ {
+ "epoch": 80.16621253405995,
+ "grad_norm": 1.4118679761886597,
+ "learning_rate": 1.9932055577483743e-06,
+ "loss": 0.0202,
+ "step": 29421
+ },
+ {
+ "epoch": 80.16893732970027,
+ "grad_norm": 1.4656741619110107,
+ "learning_rate": 1.9926768925649685e-06,
+ "loss": 0.0496,
+ "step": 29422
+ },
+ {
+ "epoch": 80.1716621253406,
+ "grad_norm": 1.4894801378250122,
+ "learning_rate": 1.992148289742334e-06,
+ "loss": 0.0476,
+ "step": 29423
+ },
+ {
+ "epoch": 80.17438692098092,
+ "grad_norm": 1.2925283908843994,
+ "learning_rate": 1.9916197492845947e-06,
+ "loss": 0.0095,
+ "step": 29424
+ },
+ {
+ "epoch": 80.17711171662125,
+ "grad_norm": 1.2885818481445312,
+ "learning_rate": 1.991091271195862e-06,
+ "loss": 0.0171,
+ "step": 29425
+ },
+ {
+ "epoch": 80.17983651226157,
+ "grad_norm": 1.3824102878570557,
+ "learning_rate": 1.9905628554802525e-06,
+ "loss": 0.0234,
+ "step": 29426
+ },
+ {
+ "epoch": 80.1825613079019,
+ "grad_norm": 0.9008615612983704,
+ "learning_rate": 1.9900345021418797e-06,
+ "loss": 0.0088,
+ "step": 29427
+ },
+ {
+ "epoch": 80.18528610354224,
+ "grad_norm": 2.2998502254486084,
+ "learning_rate": 1.9895062111848628e-06,
+ "loss": 0.0147,
+ "step": 29428
+ },
+ {
+ "epoch": 80.18801089918256,
+ "grad_norm": 1.0889054536819458,
+ "learning_rate": 1.9889779826133136e-06,
+ "loss": 0.0129,
+ "step": 29429
+ },
+ {
+ "epoch": 80.19073569482289,
+ "grad_norm": 1.4353342056274414,
+ "learning_rate": 1.9884498164313458e-06,
+ "loss": 0.0426,
+ "step": 29430
+ },
+ {
+ "epoch": 80.19346049046321,
+ "grad_norm": 0.7716575860977173,
+ "learning_rate": 1.9879217126430704e-06,
+ "loss": 0.0093,
+ "step": 29431
+ },
+ {
+ "epoch": 80.19618528610354,
+ "grad_norm": 1.5977674722671509,
+ "learning_rate": 1.9873936712526033e-06,
+ "loss": 0.0225,
+ "step": 29432
+ },
+ {
+ "epoch": 80.19891008174388,
+ "grad_norm": 0.6355977654457092,
+ "learning_rate": 1.9868656922640596e-06,
+ "loss": 0.0064,
+ "step": 29433
+ },
+ {
+ "epoch": 80.2016348773842,
+ "grad_norm": 1.9424351453781128,
+ "learning_rate": 1.986337775681546e-06,
+ "loss": 0.0155,
+ "step": 29434
+ },
+ {
+ "epoch": 80.20435967302453,
+ "grad_norm": 2.3832321166992188,
+ "learning_rate": 1.9858099215091773e-06,
+ "loss": 0.0504,
+ "step": 29435
+ },
+ {
+ "epoch": 80.20708446866485,
+ "grad_norm": 1.2908880710601807,
+ "learning_rate": 1.9852821297510605e-06,
+ "loss": 0.0134,
+ "step": 29436
+ },
+ {
+ "epoch": 80.20980926430518,
+ "grad_norm": 1.6614229679107666,
+ "learning_rate": 1.9847544004113105e-06,
+ "loss": 0.0692,
+ "step": 29437
+ },
+ {
+ "epoch": 80.2125340599455,
+ "grad_norm": 0.9882975220680237,
+ "learning_rate": 1.9842267334940357e-06,
+ "loss": 0.0802,
+ "step": 29438
+ },
+ {
+ "epoch": 80.21525885558583,
+ "grad_norm": 1.2916947603225708,
+ "learning_rate": 1.9836991290033427e-06,
+ "loss": 0.0558,
+ "step": 29439
+ },
+ {
+ "epoch": 80.21798365122616,
+ "grad_norm": 1.3554904460906982,
+ "learning_rate": 1.9831715869433456e-06,
+ "loss": 0.0418,
+ "step": 29440
+ },
+ {
+ "epoch": 80.22070844686648,
+ "grad_norm": 1.1599124670028687,
+ "learning_rate": 1.982644107318149e-06,
+ "loss": 0.0268,
+ "step": 29441
+ },
+ {
+ "epoch": 80.22343324250681,
+ "grad_norm": 1.671295166015625,
+ "learning_rate": 1.9821166901318634e-06,
+ "loss": 0.0468,
+ "step": 29442
+ },
+ {
+ "epoch": 80.22615803814713,
+ "grad_norm": 0.9318678379058838,
+ "learning_rate": 1.981589335388592e-06,
+ "loss": 0.0103,
+ "step": 29443
+ },
+ {
+ "epoch": 80.22888283378747,
+ "grad_norm": 1.2868931293487549,
+ "learning_rate": 1.9810620430924465e-06,
+ "loss": 0.0192,
+ "step": 29444
+ },
+ {
+ "epoch": 80.2316076294278,
+ "grad_norm": 0.9344775676727295,
+ "learning_rate": 1.980534813247532e-06,
+ "loss": 0.0073,
+ "step": 29445
+ },
+ {
+ "epoch": 80.23433242506812,
+ "grad_norm": 1.2206652164459229,
+ "learning_rate": 1.9800076458579552e-06,
+ "loss": 0.0218,
+ "step": 29446
+ },
+ {
+ "epoch": 80.23705722070845,
+ "grad_norm": 1.3317277431488037,
+ "learning_rate": 1.9794805409278174e-06,
+ "loss": 0.0254,
+ "step": 29447
+ },
+ {
+ "epoch": 80.23978201634877,
+ "grad_norm": 1.0683892965316772,
+ "learning_rate": 1.978953498461227e-06,
+ "loss": 0.0183,
+ "step": 29448
+ },
+ {
+ "epoch": 80.2425068119891,
+ "grad_norm": 0.8019749522209167,
+ "learning_rate": 1.978426518462291e-06,
+ "loss": 0.0088,
+ "step": 29449
+ },
+ {
+ "epoch": 80.24523160762942,
+ "grad_norm": 1.0316002368927002,
+ "learning_rate": 1.9778996009351116e-06,
+ "loss": 0.0108,
+ "step": 29450
+ },
+ {
+ "epoch": 80.24795640326975,
+ "grad_norm": 0.6009722352027893,
+ "learning_rate": 1.977372745883792e-06,
+ "loss": 0.0048,
+ "step": 29451
+ },
+ {
+ "epoch": 80.25068119891009,
+ "grad_norm": 1.0645695924758911,
+ "learning_rate": 1.976845953312432e-06,
+ "loss": 0.0138,
+ "step": 29452
+ },
+ {
+ "epoch": 80.2534059945504,
+ "grad_norm": 1.070552945137024,
+ "learning_rate": 1.9763192232251405e-06,
+ "loss": 0.0099,
+ "step": 29453
+ },
+ {
+ "epoch": 80.25613079019074,
+ "grad_norm": 1.081357479095459,
+ "learning_rate": 1.975792555626017e-06,
+ "loss": 0.0333,
+ "step": 29454
+ },
+ {
+ "epoch": 80.25885558583106,
+ "grad_norm": 1.2194374799728394,
+ "learning_rate": 1.975265950519163e-06,
+ "loss": 0.0161,
+ "step": 29455
+ },
+ {
+ "epoch": 80.26158038147139,
+ "grad_norm": 1.1556252241134644,
+ "learning_rate": 1.9747394079086756e-06,
+ "loss": 0.0396,
+ "step": 29456
+ },
+ {
+ "epoch": 80.26430517711172,
+ "grad_norm": 1.172325849533081,
+ "learning_rate": 1.974212927798663e-06,
+ "loss": 0.0756,
+ "step": 29457
+ },
+ {
+ "epoch": 80.26702997275204,
+ "grad_norm": 1.8135839700698853,
+ "learning_rate": 1.973686510193221e-06,
+ "loss": 0.1025,
+ "step": 29458
+ },
+ {
+ "epoch": 80.26975476839237,
+ "grad_norm": 2.289238214492798,
+ "learning_rate": 1.973160155096452e-06,
+ "loss": 0.0129,
+ "step": 29459
+ },
+ {
+ "epoch": 80.2724795640327,
+ "grad_norm": 0.9529938697814941,
+ "learning_rate": 1.9726338625124488e-06,
+ "loss": 0.039,
+ "step": 29460
+ },
+ {
+ "epoch": 80.27520435967303,
+ "grad_norm": 1.25319242477417,
+ "learning_rate": 1.972107632445318e-06,
+ "loss": 0.1412,
+ "step": 29461
+ },
+ {
+ "epoch": 80.27792915531334,
+ "grad_norm": 1.7547961473464966,
+ "learning_rate": 1.971581464899155e-06,
+ "loss": 0.0129,
+ "step": 29462
+ },
+ {
+ "epoch": 80.28065395095368,
+ "grad_norm": 0.9633354544639587,
+ "learning_rate": 1.9710553598780534e-06,
+ "loss": 0.0089,
+ "step": 29463
+ },
+ {
+ "epoch": 80.28337874659401,
+ "grad_norm": 1.8397653102874756,
+ "learning_rate": 1.9705293173861182e-06,
+ "loss": 0.0293,
+ "step": 29464
+ },
+ {
+ "epoch": 80.28610354223433,
+ "grad_norm": 1.2050485610961914,
+ "learning_rate": 1.9700033374274386e-06,
+ "loss": 0.0078,
+ "step": 29465
+ },
+ {
+ "epoch": 80.28882833787466,
+ "grad_norm": 1.6132885217666626,
+ "learning_rate": 1.9694774200061175e-06,
+ "loss": 0.0184,
+ "step": 29466
+ },
+ {
+ "epoch": 80.29155313351498,
+ "grad_norm": 1.0858160257339478,
+ "learning_rate": 1.9689515651262482e-06,
+ "loss": 0.0134,
+ "step": 29467
+ },
+ {
+ "epoch": 80.29427792915531,
+ "grad_norm": 1.4718116521835327,
+ "learning_rate": 1.9684257727919233e-06,
+ "loss": 0.0474,
+ "step": 29468
+ },
+ {
+ "epoch": 80.29700272479565,
+ "grad_norm": 0.7789562344551086,
+ "learning_rate": 1.9679000430072414e-06,
+ "loss": 0.0095,
+ "step": 29469
+ },
+ {
+ "epoch": 80.29972752043597,
+ "grad_norm": 1.5051437616348267,
+ "learning_rate": 1.9673743757762963e-06,
+ "loss": 0.0115,
+ "step": 29470
+ },
+ {
+ "epoch": 80.3024523160763,
+ "grad_norm": 0.5387822985649109,
+ "learning_rate": 1.9668487711031804e-06,
+ "loss": 0.0043,
+ "step": 29471
+ },
+ {
+ "epoch": 80.30517711171662,
+ "grad_norm": 1.7437795400619507,
+ "learning_rate": 1.9663232289919854e-06,
+ "loss": 0.094,
+ "step": 29472
+ },
+ {
+ "epoch": 80.30790190735695,
+ "grad_norm": 1.329782485961914,
+ "learning_rate": 1.9657977494468095e-06,
+ "loss": 0.0159,
+ "step": 29473
+ },
+ {
+ "epoch": 80.31062670299727,
+ "grad_norm": 1.4808303117752075,
+ "learning_rate": 1.9652723324717415e-06,
+ "loss": 0.0255,
+ "step": 29474
+ },
+ {
+ "epoch": 80.3133514986376,
+ "grad_norm": 1.1540828943252563,
+ "learning_rate": 1.964746978070875e-06,
+ "loss": 0.0148,
+ "step": 29475
+ },
+ {
+ "epoch": 80.31607629427793,
+ "grad_norm": 1.472664475440979,
+ "learning_rate": 1.964221686248298e-06,
+ "loss": 0.0173,
+ "step": 29476
+ },
+ {
+ "epoch": 80.31880108991825,
+ "grad_norm": 1.5507525205612183,
+ "learning_rate": 1.9636964570081054e-06,
+ "loss": 0.0165,
+ "step": 29477
+ },
+ {
+ "epoch": 80.32152588555859,
+ "grad_norm": 0.7999559640884399,
+ "learning_rate": 1.9631712903543854e-06,
+ "loss": 0.0087,
+ "step": 29478
+ },
+ {
+ "epoch": 80.3242506811989,
+ "grad_norm": 1.6818009614944458,
+ "learning_rate": 1.9626461862912304e-06,
+ "loss": 0.0538,
+ "step": 29479
+ },
+ {
+ "epoch": 80.32697547683924,
+ "grad_norm": 1.2700049877166748,
+ "learning_rate": 1.962121144822725e-06,
+ "loss": 0.0209,
+ "step": 29480
+ },
+ {
+ "epoch": 80.32970027247957,
+ "grad_norm": 1.2607086896896362,
+ "learning_rate": 1.9615961659529637e-06,
+ "loss": 0.0276,
+ "step": 29481
+ },
+ {
+ "epoch": 80.33242506811989,
+ "grad_norm": 4.096279621124268,
+ "learning_rate": 1.961071249686033e-06,
+ "loss": 0.0125,
+ "step": 29482
+ },
+ {
+ "epoch": 80.33514986376022,
+ "grad_norm": 0.8601272702217102,
+ "learning_rate": 1.9605463960260186e-06,
+ "loss": 0.1358,
+ "step": 29483
+ },
+ {
+ "epoch": 80.33787465940054,
+ "grad_norm": 0.6869960427284241,
+ "learning_rate": 1.960021604977009e-06,
+ "loss": 0.0088,
+ "step": 29484
+ },
+ {
+ "epoch": 80.34059945504087,
+ "grad_norm": 1.457580804824829,
+ "learning_rate": 1.9594968765430956e-06,
+ "loss": 0.0883,
+ "step": 29485
+ },
+ {
+ "epoch": 80.34332425068119,
+ "grad_norm": 0.48318397998809814,
+ "learning_rate": 1.9589722107283628e-06,
+ "loss": 0.0055,
+ "step": 29486
+ },
+ {
+ "epoch": 80.34604904632153,
+ "grad_norm": 1.1197354793548584,
+ "learning_rate": 1.958447607536894e-06,
+ "loss": 0.0169,
+ "step": 29487
+ },
+ {
+ "epoch": 80.34877384196186,
+ "grad_norm": 0.7464203238487244,
+ "learning_rate": 1.9579230669727745e-06,
+ "loss": 0.0058,
+ "step": 29488
+ },
+ {
+ "epoch": 80.35149863760218,
+ "grad_norm": 1.2876944541931152,
+ "learning_rate": 1.957398589040094e-06,
+ "loss": 0.0177,
+ "step": 29489
+ },
+ {
+ "epoch": 80.35422343324251,
+ "grad_norm": 1.233993411064148,
+ "learning_rate": 1.9568741737429344e-06,
+ "loss": 0.027,
+ "step": 29490
+ },
+ {
+ "epoch": 80.35694822888283,
+ "grad_norm": 2.914808511734009,
+ "learning_rate": 1.9563498210853792e-06,
+ "loss": 0.0452,
+ "step": 29491
+ },
+ {
+ "epoch": 80.35967302452316,
+ "grad_norm": 0.7809935212135315,
+ "learning_rate": 1.9558255310715103e-06,
+ "loss": 0.0073,
+ "step": 29492
+ },
+ {
+ "epoch": 80.3623978201635,
+ "grad_norm": 1.2498260736465454,
+ "learning_rate": 1.955301303705417e-06,
+ "loss": 0.0185,
+ "step": 29493
+ },
+ {
+ "epoch": 80.36512261580381,
+ "grad_norm": 1.8946287631988525,
+ "learning_rate": 1.9547771389911774e-06,
+ "loss": 0.0244,
+ "step": 29494
+ },
+ {
+ "epoch": 80.36784741144415,
+ "grad_norm": 0.9807755947113037,
+ "learning_rate": 1.9542530369328738e-06,
+ "loss": 0.007,
+ "step": 29495
+ },
+ {
+ "epoch": 80.37057220708446,
+ "grad_norm": 1.291832685470581,
+ "learning_rate": 1.9537289975345876e-06,
+ "loss": 0.0189,
+ "step": 29496
+ },
+ {
+ "epoch": 80.3732970027248,
+ "grad_norm": 1.3317344188690186,
+ "learning_rate": 1.9532050208004014e-06,
+ "loss": 0.0134,
+ "step": 29497
+ },
+ {
+ "epoch": 80.37602179836512,
+ "grad_norm": 0.573392391204834,
+ "learning_rate": 1.9526811067343977e-06,
+ "loss": 0.0066,
+ "step": 29498
+ },
+ {
+ "epoch": 80.37874659400545,
+ "grad_norm": 0.9292529225349426,
+ "learning_rate": 1.952157255340653e-06,
+ "loss": 0.0436,
+ "step": 29499
+ },
+ {
+ "epoch": 80.38147138964578,
+ "grad_norm": 1.3828158378601074,
+ "learning_rate": 1.9516334666232474e-06,
+ "loss": 0.1266,
+ "step": 29500
+ },
+ {
+ "epoch": 80.3841961852861,
+ "grad_norm": 1.5283503532409668,
+ "learning_rate": 1.95110974058626e-06,
+ "loss": 0.1108,
+ "step": 29501
+ },
+ {
+ "epoch": 80.38692098092643,
+ "grad_norm": 0.3738432228565216,
+ "learning_rate": 1.950586077233775e-06,
+ "loss": 0.0037,
+ "step": 29502
+ },
+ {
+ "epoch": 80.38964577656675,
+ "grad_norm": 1.243774652481079,
+ "learning_rate": 1.950062476569865e-06,
+ "loss": 0.011,
+ "step": 29503
+ },
+ {
+ "epoch": 80.39237057220708,
+ "grad_norm": 0.8820761442184448,
+ "learning_rate": 1.9495389385986076e-06,
+ "loss": 0.008,
+ "step": 29504
+ },
+ {
+ "epoch": 80.39509536784742,
+ "grad_norm": 0.7300599813461304,
+ "learning_rate": 1.9490154633240843e-06,
+ "loss": 0.007,
+ "step": 29505
+ },
+ {
+ "epoch": 80.39782016348774,
+ "grad_norm": 0.9548833966255188,
+ "learning_rate": 1.948492050750369e-06,
+ "loss": 0.0133,
+ "step": 29506
+ },
+ {
+ "epoch": 80.40054495912807,
+ "grad_norm": 1.5015803575515747,
+ "learning_rate": 1.9479687008815396e-06,
+ "loss": 0.017,
+ "step": 29507
+ },
+ {
+ "epoch": 80.40326975476839,
+ "grad_norm": 0.831342875957489,
+ "learning_rate": 1.947445413721667e-06,
+ "loss": 0.0131,
+ "step": 29508
+ },
+ {
+ "epoch": 80.40599455040872,
+ "grad_norm": 1.1130962371826172,
+ "learning_rate": 1.9469221892748337e-06,
+ "loss": 0.0448,
+ "step": 29509
+ },
+ {
+ "epoch": 80.40871934604904,
+ "grad_norm": 0.6033104658126831,
+ "learning_rate": 1.9463990275451107e-06,
+ "loss": 0.006,
+ "step": 29510
+ },
+ {
+ "epoch": 80.41144414168937,
+ "grad_norm": 1.047935128211975,
+ "learning_rate": 1.9458759285365723e-06,
+ "loss": 0.018,
+ "step": 29511
+ },
+ {
+ "epoch": 80.4141689373297,
+ "grad_norm": 1.8864437341690063,
+ "learning_rate": 1.9453528922532915e-06,
+ "loss": 0.1297,
+ "step": 29512
+ },
+ {
+ "epoch": 80.41689373297002,
+ "grad_norm": 1.0734527111053467,
+ "learning_rate": 1.944829918699346e-06,
+ "loss": 0.0128,
+ "step": 29513
+ },
+ {
+ "epoch": 80.41961852861036,
+ "grad_norm": 1.5648527145385742,
+ "learning_rate": 1.944307007878805e-06,
+ "loss": 0.0373,
+ "step": 29514
+ },
+ {
+ "epoch": 80.42234332425068,
+ "grad_norm": 0.6713213920593262,
+ "learning_rate": 1.943784159795742e-06,
+ "loss": 0.0063,
+ "step": 29515
+ },
+ {
+ "epoch": 80.42506811989101,
+ "grad_norm": 0.9228100180625916,
+ "learning_rate": 1.943261374454226e-06,
+ "loss": 0.0122,
+ "step": 29516
+ },
+ {
+ "epoch": 80.42779291553134,
+ "grad_norm": 0.733084499835968,
+ "learning_rate": 1.9427386518583313e-06,
+ "loss": 0.0087,
+ "step": 29517
+ },
+ {
+ "epoch": 80.43051771117166,
+ "grad_norm": 0.9572983980178833,
+ "learning_rate": 1.942215992012131e-06,
+ "loss": 0.0202,
+ "step": 29518
+ },
+ {
+ "epoch": 80.433242506812,
+ "grad_norm": 1.1350630521774292,
+ "learning_rate": 1.9416933949196926e-06,
+ "loss": 0.0112,
+ "step": 29519
+ },
+ {
+ "epoch": 80.43596730245231,
+ "grad_norm": 1.469810128211975,
+ "learning_rate": 1.9411708605850875e-06,
+ "loss": 0.0517,
+ "step": 29520
+ },
+ {
+ "epoch": 80.43869209809264,
+ "grad_norm": 1.2307252883911133,
+ "learning_rate": 1.9406483890123817e-06,
+ "loss": 0.0239,
+ "step": 29521
+ },
+ {
+ "epoch": 80.44141689373296,
+ "grad_norm": 1.034891963005066,
+ "learning_rate": 1.9401259802056495e-06,
+ "loss": 0.0147,
+ "step": 29522
+ },
+ {
+ "epoch": 80.4441416893733,
+ "grad_norm": 1.3464069366455078,
+ "learning_rate": 1.939603634168957e-06,
+ "loss": 0.0137,
+ "step": 29523
+ },
+ {
+ "epoch": 80.44686648501363,
+ "grad_norm": 0.528106153011322,
+ "learning_rate": 1.939081350906369e-06,
+ "loss": 0.0044,
+ "step": 29524
+ },
+ {
+ "epoch": 80.44959128065395,
+ "grad_norm": 1.2346034049987793,
+ "learning_rate": 1.9385591304219585e-06,
+ "loss": 0.0159,
+ "step": 29525
+ },
+ {
+ "epoch": 80.45231607629428,
+ "grad_norm": 0.59989333152771,
+ "learning_rate": 1.93803697271979e-06,
+ "loss": 0.0043,
+ "step": 29526
+ },
+ {
+ "epoch": 80.4550408719346,
+ "grad_norm": 0.49359390139579773,
+ "learning_rate": 1.937514877803931e-06,
+ "loss": 0.0044,
+ "step": 29527
+ },
+ {
+ "epoch": 80.45776566757493,
+ "grad_norm": 0.7795090079307556,
+ "learning_rate": 1.9369928456784425e-06,
+ "loss": 0.0094,
+ "step": 29528
+ },
+ {
+ "epoch": 80.46049046321527,
+ "grad_norm": 0.9974470734596252,
+ "learning_rate": 1.936470876347398e-06,
+ "loss": 0.0824,
+ "step": 29529
+ },
+ {
+ "epoch": 80.46321525885558,
+ "grad_norm": 1.0245137214660645,
+ "learning_rate": 1.935948969814858e-06,
+ "loss": 0.0084,
+ "step": 29530
+ },
+ {
+ "epoch": 80.46594005449592,
+ "grad_norm": 1.7068684101104736,
+ "learning_rate": 1.935427126084889e-06,
+ "loss": 0.0695,
+ "step": 29531
+ },
+ {
+ "epoch": 80.46866485013624,
+ "grad_norm": 0.7351219058036804,
+ "learning_rate": 1.934905345161551e-06,
+ "loss": 0.0088,
+ "step": 29532
+ },
+ {
+ "epoch": 80.47138964577657,
+ "grad_norm": 1.169875144958496,
+ "learning_rate": 1.9343836270489124e-06,
+ "loss": 0.0146,
+ "step": 29533
+ },
+ {
+ "epoch": 80.47411444141689,
+ "grad_norm": 2.6123838424682617,
+ "learning_rate": 1.933861971751032e-06,
+ "loss": 0.0472,
+ "step": 29534
+ },
+ {
+ "epoch": 80.47683923705722,
+ "grad_norm": 1.0780197381973267,
+ "learning_rate": 1.9333403792719783e-06,
+ "loss": 0.0184,
+ "step": 29535
+ },
+ {
+ "epoch": 80.47956403269755,
+ "grad_norm": 1.461605429649353,
+ "learning_rate": 1.9328188496158095e-06,
+ "loss": 0.0051,
+ "step": 29536
+ },
+ {
+ "epoch": 80.48228882833787,
+ "grad_norm": 0.9368560314178467,
+ "learning_rate": 1.9322973827865854e-06,
+ "loss": 0.0128,
+ "step": 29537
+ },
+ {
+ "epoch": 80.4850136239782,
+ "grad_norm": 1.4425424337387085,
+ "learning_rate": 1.9317759787883727e-06,
+ "loss": 0.018,
+ "step": 29538
+ },
+ {
+ "epoch": 80.48773841961852,
+ "grad_norm": 1.0747528076171875,
+ "learning_rate": 1.931254637625228e-06,
+ "loss": 0.0106,
+ "step": 29539
+ },
+ {
+ "epoch": 80.49046321525886,
+ "grad_norm": 1.1557859182357788,
+ "learning_rate": 1.9307333593012134e-06,
+ "loss": 0.0197,
+ "step": 29540
+ },
+ {
+ "epoch": 80.49318801089919,
+ "grad_norm": 1.5089144706726074,
+ "learning_rate": 1.9302121438203847e-06,
+ "loss": 0.1568,
+ "step": 29541
+ },
+ {
+ "epoch": 80.49591280653951,
+ "grad_norm": 1.6050474643707275,
+ "learning_rate": 1.929690991186808e-06,
+ "loss": 0.0151,
+ "step": 29542
+ },
+ {
+ "epoch": 80.49863760217984,
+ "grad_norm": 1.7120879888534546,
+ "learning_rate": 1.9291699014045362e-06,
+ "loss": 0.0521,
+ "step": 29543
+ },
+ {
+ "epoch": 80.50136239782016,
+ "grad_norm": 1.5400235652923584,
+ "learning_rate": 1.92864887447763e-06,
+ "loss": 0.0151,
+ "step": 29544
+ },
+ {
+ "epoch": 80.50408719346049,
+ "grad_norm": 1.0958738327026367,
+ "learning_rate": 1.928127910410145e-06,
+ "loss": 0.0741,
+ "step": 29545
+ },
+ {
+ "epoch": 80.50681198910081,
+ "grad_norm": 1.3744903802871704,
+ "learning_rate": 1.9276070092061414e-06,
+ "loss": 0.0215,
+ "step": 29546
+ },
+ {
+ "epoch": 80.50953678474114,
+ "grad_norm": 0.8080568313598633,
+ "learning_rate": 1.927086170869675e-06,
+ "loss": 0.0135,
+ "step": 29547
+ },
+ {
+ "epoch": 80.51226158038148,
+ "grad_norm": 1.190894603729248,
+ "learning_rate": 1.926565395404799e-06,
+ "loss": 0.013,
+ "step": 29548
+ },
+ {
+ "epoch": 80.5149863760218,
+ "grad_norm": 2.0751702785491943,
+ "learning_rate": 1.9260446828155745e-06,
+ "loss": 0.1844,
+ "step": 29549
+ },
+ {
+ "epoch": 80.51771117166213,
+ "grad_norm": 1.804749846458435,
+ "learning_rate": 1.9255240331060543e-06,
+ "loss": 0.0313,
+ "step": 29550
+ },
+ {
+ "epoch": 80.52043596730245,
+ "grad_norm": 1.0255616903305054,
+ "learning_rate": 1.92500344628029e-06,
+ "loss": 0.0219,
+ "step": 29551
+ },
+ {
+ "epoch": 80.52316076294278,
+ "grad_norm": 0.8044003248214722,
+ "learning_rate": 1.9244829223423424e-06,
+ "loss": 0.0104,
+ "step": 29552
+ },
+ {
+ "epoch": 80.52588555858311,
+ "grad_norm": 2.3606209754943848,
+ "learning_rate": 1.9239624612962593e-06,
+ "loss": 0.2448,
+ "step": 29553
+ },
+ {
+ "epoch": 80.52861035422343,
+ "grad_norm": 1.1554163694381714,
+ "learning_rate": 1.9234420631460982e-06,
+ "loss": 0.0111,
+ "step": 29554
+ },
+ {
+ "epoch": 80.53133514986376,
+ "grad_norm": 1.4020954370498657,
+ "learning_rate": 1.922921727895911e-06,
+ "loss": 0.0387,
+ "step": 29555
+ },
+ {
+ "epoch": 80.53405994550408,
+ "grad_norm": 1.1771619319915771,
+ "learning_rate": 1.9224014555497493e-06,
+ "loss": 0.0106,
+ "step": 29556
+ },
+ {
+ "epoch": 80.53678474114442,
+ "grad_norm": 1.5815478563308716,
+ "learning_rate": 1.921881246111662e-06,
+ "loss": 0.021,
+ "step": 29557
+ },
+ {
+ "epoch": 80.53950953678473,
+ "grad_norm": 0.8157501220703125,
+ "learning_rate": 1.9213610995857067e-06,
+ "loss": 0.0067,
+ "step": 29558
+ },
+ {
+ "epoch": 80.54223433242507,
+ "grad_norm": 1.242624044418335,
+ "learning_rate": 1.9208410159759316e-06,
+ "loss": 0.0125,
+ "step": 29559
+ },
+ {
+ "epoch": 80.5449591280654,
+ "grad_norm": 0.9923348426818848,
+ "learning_rate": 1.9203209952863866e-06,
+ "loss": 0.0079,
+ "step": 29560
+ },
+ {
+ "epoch": 80.54768392370572,
+ "grad_norm": 1.056434988975525,
+ "learning_rate": 1.9198010375211185e-06,
+ "loss": 0.0086,
+ "step": 29561
+ },
+ {
+ "epoch": 80.55040871934605,
+ "grad_norm": 1.147315502166748,
+ "learning_rate": 1.919281142684183e-06,
+ "loss": 0.0478,
+ "step": 29562
+ },
+ {
+ "epoch": 80.55313351498637,
+ "grad_norm": 1.0247108936309814,
+ "learning_rate": 1.918761310779624e-06,
+ "loss": 0.028,
+ "step": 29563
+ },
+ {
+ "epoch": 80.5558583106267,
+ "grad_norm": 1.2467284202575684,
+ "learning_rate": 1.9182415418114943e-06,
+ "loss": 0.1155,
+ "step": 29564
+ },
+ {
+ "epoch": 80.55858310626704,
+ "grad_norm": 1.5565032958984375,
+ "learning_rate": 1.917721835783836e-06,
+ "loss": 0.0795,
+ "step": 29565
+ },
+ {
+ "epoch": 80.56130790190736,
+ "grad_norm": 1.1681501865386963,
+ "learning_rate": 1.9172021927007024e-06,
+ "loss": 0.0195,
+ "step": 29566
+ },
+ {
+ "epoch": 80.56403269754769,
+ "grad_norm": 1.159302830696106,
+ "learning_rate": 1.9166826125661376e-06,
+ "loss": 0.0854,
+ "step": 29567
+ },
+ {
+ "epoch": 80.566757493188,
+ "grad_norm": 1.5844395160675049,
+ "learning_rate": 1.916163095384188e-06,
+ "loss": 0.0196,
+ "step": 29568
+ },
+ {
+ "epoch": 80.56948228882834,
+ "grad_norm": 1.3760770559310913,
+ "learning_rate": 1.9156436411588963e-06,
+ "loss": 0.0815,
+ "step": 29569
+ },
+ {
+ "epoch": 80.57220708446866,
+ "grad_norm": 1.5622652769088745,
+ "learning_rate": 1.9151242498943155e-06,
+ "loss": 0.0741,
+ "step": 29570
+ },
+ {
+ "epoch": 80.57493188010899,
+ "grad_norm": 1.4578578472137451,
+ "learning_rate": 1.9146049215944886e-06,
+ "loss": 0.0177,
+ "step": 29571
+ },
+ {
+ "epoch": 80.57765667574932,
+ "grad_norm": 1.0933475494384766,
+ "learning_rate": 1.914085656263457e-06,
+ "loss": 0.0238,
+ "step": 29572
+ },
+ {
+ "epoch": 80.58038147138964,
+ "grad_norm": 1.493249773979187,
+ "learning_rate": 1.9135664539052634e-06,
+ "loss": 0.0356,
+ "step": 29573
+ },
+ {
+ "epoch": 80.58310626702998,
+ "grad_norm": 1.3290793895721436,
+ "learning_rate": 1.913047314523957e-06,
+ "loss": 0.0234,
+ "step": 29574
+ },
+ {
+ "epoch": 80.5858310626703,
+ "grad_norm": 1.4508335590362549,
+ "learning_rate": 1.9125282381235775e-06,
+ "loss": 0.0158,
+ "step": 29575
+ },
+ {
+ "epoch": 80.58855585831063,
+ "grad_norm": 1.4976669549942017,
+ "learning_rate": 1.912009224708168e-06,
+ "loss": 0.0112,
+ "step": 29576
+ },
+ {
+ "epoch": 80.59128065395096,
+ "grad_norm": 1.0363578796386719,
+ "learning_rate": 1.9114902742817675e-06,
+ "loss": 0.0119,
+ "step": 29577
+ },
+ {
+ "epoch": 80.59400544959128,
+ "grad_norm": 1.7021323442459106,
+ "learning_rate": 1.9109713868484235e-06,
+ "loss": 0.016,
+ "step": 29578
+ },
+ {
+ "epoch": 80.59673024523161,
+ "grad_norm": 1.2553988695144653,
+ "learning_rate": 1.910452562412174e-06,
+ "loss": 0.0183,
+ "step": 29579
+ },
+ {
+ "epoch": 80.59945504087193,
+ "grad_norm": 1.7016758918762207,
+ "learning_rate": 1.9099338009770596e-06,
+ "loss": 0.1067,
+ "step": 29580
+ },
+ {
+ "epoch": 80.60217983651226,
+ "grad_norm": 1.1293045282363892,
+ "learning_rate": 1.9094151025471175e-06,
+ "loss": 0.012,
+ "step": 29581
+ },
+ {
+ "epoch": 80.60490463215258,
+ "grad_norm": 1.7180309295654297,
+ "learning_rate": 1.908896467126392e-06,
+ "loss": 0.022,
+ "step": 29582
+ },
+ {
+ "epoch": 80.60762942779292,
+ "grad_norm": 1.060294270515442,
+ "learning_rate": 1.908377894718921e-06,
+ "loss": 0.025,
+ "step": 29583
+ },
+ {
+ "epoch": 80.61035422343325,
+ "grad_norm": 1.5813850164413452,
+ "learning_rate": 1.907859385328743e-06,
+ "loss": 0.0403,
+ "step": 29584
+ },
+ {
+ "epoch": 80.61307901907357,
+ "grad_norm": 20.134777069091797,
+ "learning_rate": 1.9073409389598929e-06,
+ "loss": 0.0288,
+ "step": 29585
+ },
+ {
+ "epoch": 80.6158038147139,
+ "grad_norm": 0.8781540989875793,
+ "learning_rate": 1.9068225556164112e-06,
+ "loss": 0.0077,
+ "step": 29586
+ },
+ {
+ "epoch": 80.61852861035422,
+ "grad_norm": 0.8698044419288635,
+ "learning_rate": 1.9063042353023375e-06,
+ "loss": 0.0798,
+ "step": 29587
+ },
+ {
+ "epoch": 80.62125340599455,
+ "grad_norm": 1.8008531332015991,
+ "learning_rate": 1.9057859780217058e-06,
+ "loss": 0.0157,
+ "step": 29588
+ },
+ {
+ "epoch": 80.62397820163488,
+ "grad_norm": 1.1597702503204346,
+ "learning_rate": 1.9052677837785505e-06,
+ "loss": 0.0367,
+ "step": 29589
+ },
+ {
+ "epoch": 80.6267029972752,
+ "grad_norm": 1.3006651401519775,
+ "learning_rate": 1.9047496525769104e-06,
+ "loss": 0.0121,
+ "step": 29590
+ },
+ {
+ "epoch": 80.62942779291554,
+ "grad_norm": 1.9400322437286377,
+ "learning_rate": 1.9042315844208203e-06,
+ "loss": 0.0386,
+ "step": 29591
+ },
+ {
+ "epoch": 80.63215258855585,
+ "grad_norm": 0.8100321292877197,
+ "learning_rate": 1.9037135793143135e-06,
+ "loss": 0.0077,
+ "step": 29592
+ },
+ {
+ "epoch": 80.63487738419619,
+ "grad_norm": 1.1698096990585327,
+ "learning_rate": 1.9031956372614236e-06,
+ "loss": 0.0228,
+ "step": 29593
+ },
+ {
+ "epoch": 80.6376021798365,
+ "grad_norm": 0.9394975900650024,
+ "learning_rate": 1.902677758266186e-06,
+ "loss": 0.1425,
+ "step": 29594
+ },
+ {
+ "epoch": 80.64032697547684,
+ "grad_norm": 1.2282851934432983,
+ "learning_rate": 1.9021599423326353e-06,
+ "loss": 0.0074,
+ "step": 29595
+ },
+ {
+ "epoch": 80.64305177111717,
+ "grad_norm": 0.49083977937698364,
+ "learning_rate": 1.9016421894648006e-06,
+ "loss": 0.0042,
+ "step": 29596
+ },
+ {
+ "epoch": 80.64577656675749,
+ "grad_norm": 3.3358771800994873,
+ "learning_rate": 1.9011244996667144e-06,
+ "loss": 0.078,
+ "step": 29597
+ },
+ {
+ "epoch": 80.64850136239782,
+ "grad_norm": 0.7738671898841858,
+ "learning_rate": 1.900606872942413e-06,
+ "loss": 0.0132,
+ "step": 29598
+ },
+ {
+ "epoch": 80.65122615803814,
+ "grad_norm": 1.1502002477645874,
+ "learning_rate": 1.9000893092959238e-06,
+ "loss": 0.0103,
+ "step": 29599
+ },
+ {
+ "epoch": 80.65395095367847,
+ "grad_norm": 1.7116354703903198,
+ "learning_rate": 1.8995718087312787e-06,
+ "loss": 0.1255,
+ "step": 29600
+ },
+ {
+ "epoch": 80.65667574931881,
+ "grad_norm": 0.6969296336174011,
+ "learning_rate": 1.8990543712525045e-06,
+ "loss": 0.0078,
+ "step": 29601
+ },
+ {
+ "epoch": 80.65940054495913,
+ "grad_norm": 0.8149996399879456,
+ "learning_rate": 1.898536996863637e-06,
+ "loss": 0.0091,
+ "step": 29602
+ },
+ {
+ "epoch": 80.66212534059946,
+ "grad_norm": 1.1093348264694214,
+ "learning_rate": 1.8980196855687005e-06,
+ "loss": 0.0097,
+ "step": 29603
+ },
+ {
+ "epoch": 80.66485013623978,
+ "grad_norm": 0.8719304800033569,
+ "learning_rate": 1.8975024373717277e-06,
+ "loss": 0.1237,
+ "step": 29604
+ },
+ {
+ "epoch": 80.66757493188011,
+ "grad_norm": 0.9900907278060913,
+ "learning_rate": 1.8969852522767452e-06,
+ "loss": 0.0148,
+ "step": 29605
+ },
+ {
+ "epoch": 80.67029972752043,
+ "grad_norm": 1.5257720947265625,
+ "learning_rate": 1.8964681302877786e-06,
+ "loss": 0.0239,
+ "step": 29606
+ },
+ {
+ "epoch": 80.67302452316076,
+ "grad_norm": 2.068896770477295,
+ "learning_rate": 1.8959510714088603e-06,
+ "loss": 0.024,
+ "step": 29607
+ },
+ {
+ "epoch": 80.6757493188011,
+ "grad_norm": 1.4974676370620728,
+ "learning_rate": 1.8954340756440137e-06,
+ "loss": 0.023,
+ "step": 29608
+ },
+ {
+ "epoch": 80.67847411444141,
+ "grad_norm": 2.287141799926758,
+ "learning_rate": 1.8949171429972657e-06,
+ "loss": 0.0396,
+ "step": 29609
+ },
+ {
+ "epoch": 80.68119891008175,
+ "grad_norm": 1.3596906661987305,
+ "learning_rate": 1.8944002734726396e-06,
+ "loss": 0.0202,
+ "step": 29610
+ },
+ {
+ "epoch": 80.68392370572207,
+ "grad_norm": 0.4906667470932007,
+ "learning_rate": 1.8938834670741656e-06,
+ "loss": 0.0048,
+ "step": 29611
+ },
+ {
+ "epoch": 80.6866485013624,
+ "grad_norm": 1.3333882093429565,
+ "learning_rate": 1.8933667238058672e-06,
+ "loss": 0.0188,
+ "step": 29612
+ },
+ {
+ "epoch": 80.68937329700273,
+ "grad_norm": 0.8931892514228821,
+ "learning_rate": 1.8928500436717645e-06,
+ "loss": 0.0092,
+ "step": 29613
+ },
+ {
+ "epoch": 80.69209809264305,
+ "grad_norm": 1.3758031129837036,
+ "learning_rate": 1.8923334266758874e-06,
+ "loss": 0.0374,
+ "step": 29614
+ },
+ {
+ "epoch": 80.69482288828338,
+ "grad_norm": 1.2122210264205933,
+ "learning_rate": 1.8918168728222564e-06,
+ "loss": 0.0153,
+ "step": 29615
+ },
+ {
+ "epoch": 80.6975476839237,
+ "grad_norm": 1.4292452335357666,
+ "learning_rate": 1.8913003821148957e-06,
+ "loss": 0.0128,
+ "step": 29616
+ },
+ {
+ "epoch": 80.70027247956403,
+ "grad_norm": 0.8433777093887329,
+ "learning_rate": 1.8907839545578233e-06,
+ "loss": 0.0113,
+ "step": 29617
+ },
+ {
+ "epoch": 80.70299727520435,
+ "grad_norm": 1.7929044961929321,
+ "learning_rate": 1.890267590155066e-06,
+ "loss": 0.033,
+ "step": 29618
+ },
+ {
+ "epoch": 80.70572207084469,
+ "grad_norm": 1.8189384937286377,
+ "learning_rate": 1.8897512889106451e-06,
+ "loss": 0.0586,
+ "step": 29619
+ },
+ {
+ "epoch": 80.70844686648502,
+ "grad_norm": 0.9171208143234253,
+ "learning_rate": 1.8892350508285773e-06,
+ "loss": 0.0144,
+ "step": 29620
+ },
+ {
+ "epoch": 80.71117166212534,
+ "grad_norm": 1.7676657438278198,
+ "learning_rate": 1.8887188759128883e-06,
+ "loss": 0.0204,
+ "step": 29621
+ },
+ {
+ "epoch": 80.71389645776567,
+ "grad_norm": 1.0773626565933228,
+ "learning_rate": 1.8882027641675927e-06,
+ "loss": 0.0105,
+ "step": 29622
+ },
+ {
+ "epoch": 80.71662125340599,
+ "grad_norm": 1.5741899013519287,
+ "learning_rate": 1.887686715596716e-06,
+ "loss": 0.0428,
+ "step": 29623
+ },
+ {
+ "epoch": 80.71934604904632,
+ "grad_norm": 0.8020696640014648,
+ "learning_rate": 1.887170730204274e-06,
+ "loss": 0.0114,
+ "step": 29624
+ },
+ {
+ "epoch": 80.72207084468666,
+ "grad_norm": 1.2048730850219727,
+ "learning_rate": 1.886654807994286e-06,
+ "loss": 0.0584,
+ "step": 29625
+ },
+ {
+ "epoch": 80.72479564032697,
+ "grad_norm": 1.3806627988815308,
+ "learning_rate": 1.8861389489707648e-06,
+ "loss": 0.0601,
+ "step": 29626
+ },
+ {
+ "epoch": 80.7275204359673,
+ "grad_norm": 1.3640732765197754,
+ "learning_rate": 1.8856231531377367e-06,
+ "loss": 0.0151,
+ "step": 29627
+ },
+ {
+ "epoch": 80.73024523160763,
+ "grad_norm": 1.2394742965698242,
+ "learning_rate": 1.8851074204992137e-06,
+ "loss": 0.0153,
+ "step": 29628
+ },
+ {
+ "epoch": 80.73297002724796,
+ "grad_norm": 1.8326958417892456,
+ "learning_rate": 1.8845917510592126e-06,
+ "loss": 0.041,
+ "step": 29629
+ },
+ {
+ "epoch": 80.73569482288828,
+ "grad_norm": 0.9675979614257812,
+ "learning_rate": 1.884076144821747e-06,
+ "loss": 0.0864,
+ "step": 29630
+ },
+ {
+ "epoch": 80.73841961852861,
+ "grad_norm": 0.8943597674369812,
+ "learning_rate": 1.8835606017908382e-06,
+ "loss": 0.0086,
+ "step": 29631
+ },
+ {
+ "epoch": 80.74114441416894,
+ "grad_norm": 1.7657307386398315,
+ "learning_rate": 1.883045121970498e-06,
+ "loss": 0.018,
+ "step": 29632
+ },
+ {
+ "epoch": 80.74386920980926,
+ "grad_norm": 3.3651163578033447,
+ "learning_rate": 1.8825297053647418e-06,
+ "loss": 0.0211,
+ "step": 29633
+ },
+ {
+ "epoch": 80.7465940054496,
+ "grad_norm": 1.2447524070739746,
+ "learning_rate": 1.8820143519775792e-06,
+ "loss": 0.0216,
+ "step": 29634
+ },
+ {
+ "epoch": 80.74931880108991,
+ "grad_norm": 1.5326813459396362,
+ "learning_rate": 1.8814990618130313e-06,
+ "loss": 0.0411,
+ "step": 29635
+ },
+ {
+ "epoch": 80.75204359673025,
+ "grad_norm": 1.1106290817260742,
+ "learning_rate": 1.8809838348751065e-06,
+ "loss": 0.0154,
+ "step": 29636
+ },
+ {
+ "epoch": 80.75476839237058,
+ "grad_norm": 1.5790318250656128,
+ "learning_rate": 1.8804686711678167e-06,
+ "loss": 0.035,
+ "step": 29637
+ },
+ {
+ "epoch": 80.7574931880109,
+ "grad_norm": 1.1379880905151367,
+ "learning_rate": 1.8799535706951743e-06,
+ "loss": 0.0166,
+ "step": 29638
+ },
+ {
+ "epoch": 80.76021798365123,
+ "grad_norm": 0.9501550197601318,
+ "learning_rate": 1.8794385334611953e-06,
+ "loss": 0.0064,
+ "step": 29639
+ },
+ {
+ "epoch": 80.76294277929155,
+ "grad_norm": 0.9548057317733765,
+ "learning_rate": 1.8789235594698885e-06,
+ "loss": 0.007,
+ "step": 29640
+ },
+ {
+ "epoch": 80.76566757493188,
+ "grad_norm": 1.2675931453704834,
+ "learning_rate": 1.8784086487252628e-06,
+ "loss": 0.0462,
+ "step": 29641
+ },
+ {
+ "epoch": 80.7683923705722,
+ "grad_norm": 1.6175315380096436,
+ "learning_rate": 1.8778938012313263e-06,
+ "loss": 0.0505,
+ "step": 29642
+ },
+ {
+ "epoch": 80.77111716621253,
+ "grad_norm": 1.4293577671051025,
+ "learning_rate": 1.8773790169920947e-06,
+ "loss": 0.0291,
+ "step": 29643
+ },
+ {
+ "epoch": 80.77384196185287,
+ "grad_norm": 1.0141725540161133,
+ "learning_rate": 1.8768642960115735e-06,
+ "loss": 0.016,
+ "step": 29644
+ },
+ {
+ "epoch": 80.77656675749319,
+ "grad_norm": 0.9711900353431702,
+ "learning_rate": 1.8763496382937718e-06,
+ "loss": 0.0091,
+ "step": 29645
+ },
+ {
+ "epoch": 80.77929155313352,
+ "grad_norm": 1.1674789190292358,
+ "learning_rate": 1.8758350438426954e-06,
+ "loss": 0.0203,
+ "step": 29646
+ },
+ {
+ "epoch": 80.78201634877384,
+ "grad_norm": 2.113799571990967,
+ "learning_rate": 1.8753205126623564e-06,
+ "loss": 0.0783,
+ "step": 29647
+ },
+ {
+ "epoch": 80.78474114441417,
+ "grad_norm": 1.8627899885177612,
+ "learning_rate": 1.8748060447567595e-06,
+ "loss": 0.0738,
+ "step": 29648
+ },
+ {
+ "epoch": 80.7874659400545,
+ "grad_norm": 0.7619850039482117,
+ "learning_rate": 1.8742916401299128e-06,
+ "loss": 0.0055,
+ "step": 29649
+ },
+ {
+ "epoch": 80.79019073569482,
+ "grad_norm": 0.7915595769882202,
+ "learning_rate": 1.8737772987858172e-06,
+ "loss": 0.0163,
+ "step": 29650
+ },
+ {
+ "epoch": 80.79291553133515,
+ "grad_norm": 2.9928109645843506,
+ "learning_rate": 1.8732630207284863e-06,
+ "loss": 0.0454,
+ "step": 29651
+ },
+ {
+ "epoch": 80.79564032697547,
+ "grad_norm": 1.4186183214187622,
+ "learning_rate": 1.8727488059619203e-06,
+ "loss": 0.0514,
+ "step": 29652
+ },
+ {
+ "epoch": 80.7983651226158,
+ "grad_norm": 1.9611786603927612,
+ "learning_rate": 1.872234654490126e-06,
+ "loss": 0.0603,
+ "step": 29653
+ },
+ {
+ "epoch": 80.80108991825612,
+ "grad_norm": 0.9325669407844543,
+ "learning_rate": 1.8717205663171034e-06,
+ "loss": 0.0069,
+ "step": 29654
+ },
+ {
+ "epoch": 80.80381471389646,
+ "grad_norm": 1.1028732061386108,
+ "learning_rate": 1.8712065414468595e-06,
+ "loss": 0.0139,
+ "step": 29655
+ },
+ {
+ "epoch": 80.80653950953679,
+ "grad_norm": 1.6721081733703613,
+ "learning_rate": 1.8706925798833997e-06,
+ "loss": 0.0188,
+ "step": 29656
+ },
+ {
+ "epoch": 80.80926430517711,
+ "grad_norm": 1.2263832092285156,
+ "learning_rate": 1.8701786816307255e-06,
+ "loss": 0.0147,
+ "step": 29657
+ },
+ {
+ "epoch": 80.81198910081744,
+ "grad_norm": 1.5620321035385132,
+ "learning_rate": 1.8696648466928347e-06,
+ "loss": 0.1434,
+ "step": 29658
+ },
+ {
+ "epoch": 80.81471389645776,
+ "grad_norm": 0.6795973181724548,
+ "learning_rate": 1.8691510750737352e-06,
+ "loss": 0.006,
+ "step": 29659
+ },
+ {
+ "epoch": 80.8174386920981,
+ "grad_norm": 0.9721299409866333,
+ "learning_rate": 1.8686373667774248e-06,
+ "loss": 0.0115,
+ "step": 29660
+ },
+ {
+ "epoch": 80.82016348773843,
+ "grad_norm": 1.2019383907318115,
+ "learning_rate": 1.8681237218079063e-06,
+ "loss": 0.0147,
+ "step": 29661
+ },
+ {
+ "epoch": 80.82288828337875,
+ "grad_norm": 1.5381453037261963,
+ "learning_rate": 1.8676101401691748e-06,
+ "loss": 0.0346,
+ "step": 29662
+ },
+ {
+ "epoch": 80.82561307901908,
+ "grad_norm": 1.2664680480957031,
+ "learning_rate": 1.8670966218652365e-06,
+ "loss": 0.0606,
+ "step": 29663
+ },
+ {
+ "epoch": 80.8283378746594,
+ "grad_norm": 0.9388695955276489,
+ "learning_rate": 1.866583166900089e-06,
+ "loss": 0.012,
+ "step": 29664
+ },
+ {
+ "epoch": 80.83106267029973,
+ "grad_norm": 1.5581873655319214,
+ "learning_rate": 1.866069775277729e-06,
+ "loss": 0.0119,
+ "step": 29665
+ },
+ {
+ "epoch": 80.83378746594005,
+ "grad_norm": 0.5034410953521729,
+ "learning_rate": 1.8655564470021526e-06,
+ "loss": 0.005,
+ "step": 29666
+ },
+ {
+ "epoch": 80.83651226158038,
+ "grad_norm": 0.9123667478561401,
+ "learning_rate": 1.8650431820773651e-06,
+ "loss": 0.0201,
+ "step": 29667
+ },
+ {
+ "epoch": 80.83923705722071,
+ "grad_norm": 1.4766902923583984,
+ "learning_rate": 1.864529980507358e-06,
+ "loss": 0.0107,
+ "step": 29668
+ },
+ {
+ "epoch": 80.84196185286103,
+ "grad_norm": 0.8661112785339355,
+ "learning_rate": 1.8640168422961303e-06,
+ "loss": 0.0083,
+ "step": 29669
+ },
+ {
+ "epoch": 80.84468664850137,
+ "grad_norm": 1.3767850399017334,
+ "learning_rate": 1.8635037674476752e-06,
+ "loss": 0.096,
+ "step": 29670
+ },
+ {
+ "epoch": 80.84741144414168,
+ "grad_norm": 1.073331594467163,
+ "learning_rate": 1.8629907559659922e-06,
+ "loss": 0.0139,
+ "step": 29671
+ },
+ {
+ "epoch": 80.85013623978202,
+ "grad_norm": 1.2165812253952026,
+ "learning_rate": 1.8624778078550743e-06,
+ "loss": 0.0076,
+ "step": 29672
+ },
+ {
+ "epoch": 80.85286103542235,
+ "grad_norm": 0.9277904033660889,
+ "learning_rate": 1.8619649231189186e-06,
+ "loss": 0.0096,
+ "step": 29673
+ },
+ {
+ "epoch": 80.85558583106267,
+ "grad_norm": 0.8144540190696716,
+ "learning_rate": 1.8614521017615196e-06,
+ "loss": 0.0096,
+ "step": 29674
+ },
+ {
+ "epoch": 80.858310626703,
+ "grad_norm": 1.1384704113006592,
+ "learning_rate": 1.8609393437868662e-06,
+ "loss": 0.0852,
+ "step": 29675
+ },
+ {
+ "epoch": 80.86103542234332,
+ "grad_norm": 0.8645658493041992,
+ "learning_rate": 1.8604266491989575e-06,
+ "loss": 0.0091,
+ "step": 29676
+ },
+ {
+ "epoch": 80.86376021798365,
+ "grad_norm": 1.5463740825653076,
+ "learning_rate": 1.859914018001785e-06,
+ "loss": 0.0404,
+ "step": 29677
+ },
+ {
+ "epoch": 80.86648501362397,
+ "grad_norm": 1.5631438493728638,
+ "learning_rate": 1.8594014501993362e-06,
+ "loss": 0.0594,
+ "step": 29678
+ },
+ {
+ "epoch": 80.8692098092643,
+ "grad_norm": 0.9687581062316895,
+ "learning_rate": 1.8588889457956106e-06,
+ "loss": 0.0095,
+ "step": 29679
+ },
+ {
+ "epoch": 80.87193460490464,
+ "grad_norm": 1.0158239603042603,
+ "learning_rate": 1.8583765047945956e-06,
+ "loss": 0.0071,
+ "step": 29680
+ },
+ {
+ "epoch": 80.87465940054496,
+ "grad_norm": 1.2642829418182373,
+ "learning_rate": 1.8578641272002818e-06,
+ "loss": 0.1285,
+ "step": 29681
+ },
+ {
+ "epoch": 80.87738419618529,
+ "grad_norm": 1.3806689977645874,
+ "learning_rate": 1.857351813016659e-06,
+ "loss": 0.0201,
+ "step": 29682
+ },
+ {
+ "epoch": 80.88010899182561,
+ "grad_norm": 0.7610867619514465,
+ "learning_rate": 1.8568395622477198e-06,
+ "loss": 0.0059,
+ "step": 29683
+ },
+ {
+ "epoch": 80.88283378746594,
+ "grad_norm": 1.248455286026001,
+ "learning_rate": 1.856327374897452e-06,
+ "loss": 0.0361,
+ "step": 29684
+ },
+ {
+ "epoch": 80.88555858310627,
+ "grad_norm": 1.5848110914230347,
+ "learning_rate": 1.8558152509698446e-06,
+ "loss": 0.0123,
+ "step": 29685
+ },
+ {
+ "epoch": 80.88828337874659,
+ "grad_norm": 1.5311391353607178,
+ "learning_rate": 1.8553031904688833e-06,
+ "loss": 0.0211,
+ "step": 29686
+ },
+ {
+ "epoch": 80.89100817438693,
+ "grad_norm": 1.685337781906128,
+ "learning_rate": 1.8547911933985607e-06,
+ "loss": 0.1688,
+ "step": 29687
+ },
+ {
+ "epoch": 80.89373297002724,
+ "grad_norm": 1.0977756977081299,
+ "learning_rate": 1.8542792597628622e-06,
+ "loss": 0.0124,
+ "step": 29688
+ },
+ {
+ "epoch": 80.89645776566758,
+ "grad_norm": 1.0834137201309204,
+ "learning_rate": 1.8537673895657726e-06,
+ "loss": 0.009,
+ "step": 29689
+ },
+ {
+ "epoch": 80.8991825613079,
+ "grad_norm": 1.6296892166137695,
+ "learning_rate": 1.853255582811283e-06,
+ "loss": 0.009,
+ "step": 29690
+ },
+ {
+ "epoch": 80.90190735694823,
+ "grad_norm": 1.1439818143844604,
+ "learning_rate": 1.8527438395033747e-06,
+ "loss": 0.0142,
+ "step": 29691
+ },
+ {
+ "epoch": 80.90463215258856,
+ "grad_norm": 1.6660680770874023,
+ "learning_rate": 1.8522321596460368e-06,
+ "loss": 0.0375,
+ "step": 29692
+ },
+ {
+ "epoch": 80.90735694822888,
+ "grad_norm": 1.1486763954162598,
+ "learning_rate": 1.8517205432432527e-06,
+ "loss": 0.0375,
+ "step": 29693
+ },
+ {
+ "epoch": 80.91008174386921,
+ "grad_norm": 0.9677402377128601,
+ "learning_rate": 1.851208990299007e-06,
+ "loss": 0.0124,
+ "step": 29694
+ },
+ {
+ "epoch": 80.91280653950953,
+ "grad_norm": 1.5494641065597534,
+ "learning_rate": 1.850697500817281e-06,
+ "loss": 0.0305,
+ "step": 29695
+ },
+ {
+ "epoch": 80.91553133514986,
+ "grad_norm": 2.0505926609039307,
+ "learning_rate": 1.8501860748020629e-06,
+ "loss": 0.0218,
+ "step": 29696
+ },
+ {
+ "epoch": 80.9182561307902,
+ "grad_norm": 1.2098982334136963,
+ "learning_rate": 1.8496747122573333e-06,
+ "loss": 0.0273,
+ "step": 29697
+ },
+ {
+ "epoch": 80.92098092643052,
+ "grad_norm": 1.0656585693359375,
+ "learning_rate": 1.8491634131870716e-06,
+ "loss": 0.1192,
+ "step": 29698
+ },
+ {
+ "epoch": 80.92370572207085,
+ "grad_norm": 1.5306059122085571,
+ "learning_rate": 1.848652177595267e-06,
+ "loss": 0.018,
+ "step": 29699
+ },
+ {
+ "epoch": 80.92643051771117,
+ "grad_norm": 1.280849575996399,
+ "learning_rate": 1.848141005485895e-06,
+ "loss": 0.0114,
+ "step": 29700
+ },
+ {
+ "epoch": 80.9291553133515,
+ "grad_norm": 0.8554812669754028,
+ "learning_rate": 1.8476298968629401e-06,
+ "loss": 0.0068,
+ "step": 29701
+ },
+ {
+ "epoch": 80.93188010899182,
+ "grad_norm": 0.72218918800354,
+ "learning_rate": 1.8471188517303774e-06,
+ "loss": 0.0072,
+ "step": 29702
+ },
+ {
+ "epoch": 80.93460490463215,
+ "grad_norm": 1.0131006240844727,
+ "learning_rate": 1.846607870092194e-06,
+ "loss": 0.0441,
+ "step": 29703
+ },
+ {
+ "epoch": 80.93732970027249,
+ "grad_norm": 0.8276346921920776,
+ "learning_rate": 1.8460969519523664e-06,
+ "loss": 0.0108,
+ "step": 29704
+ },
+ {
+ "epoch": 80.9400544959128,
+ "grad_norm": 1.2714301347732544,
+ "learning_rate": 1.845586097314872e-06,
+ "loss": 0.0155,
+ "step": 29705
+ },
+ {
+ "epoch": 80.94277929155314,
+ "grad_norm": 0.8658757209777832,
+ "learning_rate": 1.845075306183689e-06,
+ "loss": 0.0117,
+ "step": 29706
+ },
+ {
+ "epoch": 80.94550408719346,
+ "grad_norm": 1.312825083732605,
+ "learning_rate": 1.8445645785627974e-06,
+ "loss": 0.0774,
+ "step": 29707
+ },
+ {
+ "epoch": 80.94822888283379,
+ "grad_norm": 1.6391254663467407,
+ "learning_rate": 1.8440539144561763e-06,
+ "loss": 0.0205,
+ "step": 29708
+ },
+ {
+ "epoch": 80.95095367847412,
+ "grad_norm": 1.81781005859375,
+ "learning_rate": 1.8435433138678017e-06,
+ "loss": 0.0628,
+ "step": 29709
+ },
+ {
+ "epoch": 80.95367847411444,
+ "grad_norm": 1.2340580224990845,
+ "learning_rate": 1.8430327768016475e-06,
+ "loss": 0.0109,
+ "step": 29710
+ },
+ {
+ "epoch": 80.95640326975477,
+ "grad_norm": 1.3692010641098022,
+ "learning_rate": 1.84252230326169e-06,
+ "loss": 0.0332,
+ "step": 29711
+ },
+ {
+ "epoch": 80.95912806539509,
+ "grad_norm": 1.3921781778335571,
+ "learning_rate": 1.8420118932519092e-06,
+ "loss": 0.0193,
+ "step": 29712
+ },
+ {
+ "epoch": 80.96185286103542,
+ "grad_norm": 1.3547475337982178,
+ "learning_rate": 1.8415015467762764e-06,
+ "loss": 0.0325,
+ "step": 29713
+ },
+ {
+ "epoch": 80.96457765667574,
+ "grad_norm": 1.0463752746582031,
+ "learning_rate": 1.8409912638387684e-06,
+ "loss": 0.0223,
+ "step": 29714
+ },
+ {
+ "epoch": 80.96730245231608,
+ "grad_norm": 0.7958125472068787,
+ "learning_rate": 1.840481044443354e-06,
+ "loss": 0.0085,
+ "step": 29715
+ },
+ {
+ "epoch": 80.97002724795641,
+ "grad_norm": 1.3546994924545288,
+ "learning_rate": 1.8399708885940136e-06,
+ "loss": 0.0186,
+ "step": 29716
+ },
+ {
+ "epoch": 80.97275204359673,
+ "grad_norm": 1.107937216758728,
+ "learning_rate": 1.8394607962947164e-06,
+ "loss": 0.024,
+ "step": 29717
+ },
+ {
+ "epoch": 80.97547683923706,
+ "grad_norm": 0.6469739675521851,
+ "learning_rate": 1.8389507675494367e-06,
+ "loss": 0.005,
+ "step": 29718
+ },
+ {
+ "epoch": 80.97820163487738,
+ "grad_norm": 1.620415449142456,
+ "learning_rate": 1.8384408023621437e-06,
+ "loss": 0.0805,
+ "step": 29719
+ },
+ {
+ "epoch": 80.98092643051771,
+ "grad_norm": 0.9630799889564514,
+ "learning_rate": 1.8379309007368129e-06,
+ "loss": 0.0887,
+ "step": 29720
+ },
+ {
+ "epoch": 80.98365122615803,
+ "grad_norm": 0.972714900970459,
+ "learning_rate": 1.8374210626774137e-06,
+ "loss": 0.0131,
+ "step": 29721
+ },
+ {
+ "epoch": 80.98637602179836,
+ "grad_norm": 1.2293802499771118,
+ "learning_rate": 1.8369112881879136e-06,
+ "loss": 0.0283,
+ "step": 29722
+ },
+ {
+ "epoch": 80.9891008174387,
+ "grad_norm": 0.8549829721450806,
+ "learning_rate": 1.8364015772722887e-06,
+ "loss": 0.0088,
+ "step": 29723
+ },
+ {
+ "epoch": 80.99182561307902,
+ "grad_norm": 1.1241443157196045,
+ "learning_rate": 1.8358919299345024e-06,
+ "loss": 0.0143,
+ "step": 29724
+ },
+ {
+ "epoch": 80.99455040871935,
+ "grad_norm": 1.1251850128173828,
+ "learning_rate": 1.8353823461785304e-06,
+ "loss": 0.0147,
+ "step": 29725
+ },
+ {
+ "epoch": 80.99727520435967,
+ "grad_norm": 0.8581736087799072,
+ "learning_rate": 1.834872826008338e-06,
+ "loss": 0.0077,
+ "step": 29726
+ },
+ {
+ "epoch": 81.0,
+ "grad_norm": 1.1833873987197876,
+ "learning_rate": 1.8343633694278895e-06,
+ "loss": 0.0142,
+ "step": 29727
+ },
+ {
+ "epoch": 81.00272479564033,
+ "grad_norm": 0.7314872741699219,
+ "learning_rate": 1.83385397644116e-06,
+ "loss": 0.0089,
+ "step": 29728
+ },
+ {
+ "epoch": 81.00544959128065,
+ "grad_norm": 1.5644571781158447,
+ "learning_rate": 1.8333446470521122e-06,
+ "loss": 0.0229,
+ "step": 29729
+ },
+ {
+ "epoch": 81.00817438692098,
+ "grad_norm": 1.1427139043807983,
+ "learning_rate": 1.8328353812647136e-06,
+ "loss": 0.0146,
+ "step": 29730
+ },
+ {
+ "epoch": 81.0108991825613,
+ "grad_norm": 1.4622725248336792,
+ "learning_rate": 1.8323261790829283e-06,
+ "loss": 0.0889,
+ "step": 29731
+ },
+ {
+ "epoch": 81.01362397820164,
+ "grad_norm": 1.0350446701049805,
+ "learning_rate": 1.831817040510726e-06,
+ "loss": 0.0124,
+ "step": 29732
+ },
+ {
+ "epoch": 81.01634877384195,
+ "grad_norm": 1.2746223211288452,
+ "learning_rate": 1.8313079655520693e-06,
+ "loss": 0.0113,
+ "step": 29733
+ },
+ {
+ "epoch": 81.01907356948229,
+ "grad_norm": 0.8746063113212585,
+ "learning_rate": 1.8307989542109238e-06,
+ "loss": 0.0328,
+ "step": 29734
+ },
+ {
+ "epoch": 81.02179836512262,
+ "grad_norm": 0.7549441456794739,
+ "learning_rate": 1.83029000649125e-06,
+ "loss": 0.0091,
+ "step": 29735
+ },
+ {
+ "epoch": 81.02452316076294,
+ "grad_norm": 1.5021424293518066,
+ "learning_rate": 1.8297811223970174e-06,
+ "loss": 0.0093,
+ "step": 29736
+ },
+ {
+ "epoch": 81.02724795640327,
+ "grad_norm": 0.8744590282440186,
+ "learning_rate": 1.8292723019321867e-06,
+ "loss": 0.0173,
+ "step": 29737
+ },
+ {
+ "epoch": 81.02997275204359,
+ "grad_norm": 1.5767401456832886,
+ "learning_rate": 1.8287635451007202e-06,
+ "loss": 0.0117,
+ "step": 29738
+ },
+ {
+ "epoch": 81.03269754768392,
+ "grad_norm": 1.6331926584243774,
+ "learning_rate": 1.8282548519065778e-06,
+ "loss": 0.1109,
+ "step": 29739
+ },
+ {
+ "epoch": 81.03542234332426,
+ "grad_norm": 0.6168925762176514,
+ "learning_rate": 1.8277462223537257e-06,
+ "loss": 0.0043,
+ "step": 29740
+ },
+ {
+ "epoch": 81.03814713896458,
+ "grad_norm": 0.6873663067817688,
+ "learning_rate": 1.8272376564461203e-06,
+ "loss": 0.007,
+ "step": 29741
+ },
+ {
+ "epoch": 81.04087193460491,
+ "grad_norm": 0.9462354183197021,
+ "learning_rate": 1.8267291541877287e-06,
+ "loss": 0.0093,
+ "step": 29742
+ },
+ {
+ "epoch": 81.04359673024523,
+ "grad_norm": 1.4677972793579102,
+ "learning_rate": 1.826220715582504e-06,
+ "loss": 0.0831,
+ "step": 29743
+ },
+ {
+ "epoch": 81.04632152588556,
+ "grad_norm": 1.3759363889694214,
+ "learning_rate": 1.8257123406344125e-06,
+ "loss": 0.024,
+ "step": 29744
+ },
+ {
+ "epoch": 81.04904632152588,
+ "grad_norm": 1.453350305557251,
+ "learning_rate": 1.8252040293474093e-06,
+ "loss": 0.01,
+ "step": 29745
+ },
+ {
+ "epoch": 81.05177111716621,
+ "grad_norm": 0.7757698893547058,
+ "learning_rate": 1.824695781725454e-06,
+ "loss": 0.0071,
+ "step": 29746
+ },
+ {
+ "epoch": 81.05449591280654,
+ "grad_norm": 0.6550415754318237,
+ "learning_rate": 1.8241875977725032e-06,
+ "loss": 0.0068,
+ "step": 29747
+ },
+ {
+ "epoch": 81.05722070844686,
+ "grad_norm": 0.9942609667778015,
+ "learning_rate": 1.823679477492518e-06,
+ "loss": 0.017,
+ "step": 29748
+ },
+ {
+ "epoch": 81.0599455040872,
+ "grad_norm": 1.6925396919250488,
+ "learning_rate": 1.8231714208894547e-06,
+ "loss": 0.2206,
+ "step": 29749
+ },
+ {
+ "epoch": 81.06267029972751,
+ "grad_norm": 1.0608975887298584,
+ "learning_rate": 1.8226634279672684e-06,
+ "loss": 0.0098,
+ "step": 29750
+ },
+ {
+ "epoch": 81.06539509536785,
+ "grad_norm": 0.49298837780952454,
+ "learning_rate": 1.822155498729914e-06,
+ "loss": 0.0049,
+ "step": 29751
+ },
+ {
+ "epoch": 81.06811989100818,
+ "grad_norm": 1.4216399192810059,
+ "learning_rate": 1.8216476331813516e-06,
+ "loss": 0.1204,
+ "step": 29752
+ },
+ {
+ "epoch": 81.0708446866485,
+ "grad_norm": 1.3625584840774536,
+ "learning_rate": 1.8211398313255347e-06,
+ "loss": 0.0612,
+ "step": 29753
+ },
+ {
+ "epoch": 81.07356948228883,
+ "grad_norm": 1.339282751083374,
+ "learning_rate": 1.8206320931664168e-06,
+ "loss": 0.0163,
+ "step": 29754
+ },
+ {
+ "epoch": 81.07629427792915,
+ "grad_norm": 1.1065369844436646,
+ "learning_rate": 1.8201244187079514e-06,
+ "loss": 0.0263,
+ "step": 29755
+ },
+ {
+ "epoch": 81.07901907356948,
+ "grad_norm": 1.1992741823196411,
+ "learning_rate": 1.8196168079540954e-06,
+ "loss": 0.0614,
+ "step": 29756
+ },
+ {
+ "epoch": 81.0817438692098,
+ "grad_norm": 1.1646250486373901,
+ "learning_rate": 1.8191092609088e-06,
+ "loss": 0.0183,
+ "step": 29757
+ },
+ {
+ "epoch": 81.08446866485014,
+ "grad_norm": 1.2568418979644775,
+ "learning_rate": 1.8186017775760167e-06,
+ "loss": 0.02,
+ "step": 29758
+ },
+ {
+ "epoch": 81.08719346049047,
+ "grad_norm": 1.336759090423584,
+ "learning_rate": 1.818094357959702e-06,
+ "loss": 0.0161,
+ "step": 29759
+ },
+ {
+ "epoch": 81.08991825613079,
+ "grad_norm": 1.5284219980239868,
+ "learning_rate": 1.8175870020638032e-06,
+ "loss": 0.0557,
+ "step": 29760
+ },
+ {
+ "epoch": 81.09264305177112,
+ "grad_norm": 1.2305400371551514,
+ "learning_rate": 1.8170797098922744e-06,
+ "loss": 0.016,
+ "step": 29761
+ },
+ {
+ "epoch": 81.09536784741144,
+ "grad_norm": 1.7819459438323975,
+ "learning_rate": 1.8165724814490671e-06,
+ "loss": 0.0334,
+ "step": 29762
+ },
+ {
+ "epoch": 81.09809264305177,
+ "grad_norm": 1.8965933322906494,
+ "learning_rate": 1.816065316738126e-06,
+ "loss": 0.0152,
+ "step": 29763
+ },
+ {
+ "epoch": 81.1008174386921,
+ "grad_norm": 1.4672491550445557,
+ "learning_rate": 1.8155582157634077e-06,
+ "loss": 0.0294,
+ "step": 29764
+ },
+ {
+ "epoch": 81.10354223433242,
+ "grad_norm": 1.0470800399780273,
+ "learning_rate": 1.8150511785288593e-06,
+ "loss": 0.0232,
+ "step": 29765
+ },
+ {
+ "epoch": 81.10626702997276,
+ "grad_norm": 1.4963328838348389,
+ "learning_rate": 1.8145442050384287e-06,
+ "loss": 0.0601,
+ "step": 29766
+ },
+ {
+ "epoch": 81.10899182561307,
+ "grad_norm": 0.9253717064857483,
+ "learning_rate": 1.8140372952960617e-06,
+ "loss": 0.0161,
+ "step": 29767
+ },
+ {
+ "epoch": 81.11171662125341,
+ "grad_norm": 1.4379554986953735,
+ "learning_rate": 1.8135304493057104e-06,
+ "loss": 0.0689,
+ "step": 29768
+ },
+ {
+ "epoch": 81.11444141689373,
+ "grad_norm": 1.0343574285507202,
+ "learning_rate": 1.813023667071322e-06,
+ "loss": 0.0138,
+ "step": 29769
+ },
+ {
+ "epoch": 81.11716621253406,
+ "grad_norm": 2.415585994720459,
+ "learning_rate": 1.8125169485968409e-06,
+ "loss": 0.0241,
+ "step": 29770
+ },
+ {
+ "epoch": 81.11989100817439,
+ "grad_norm": 1.2764850854873657,
+ "learning_rate": 1.8120102938862115e-06,
+ "loss": 0.0253,
+ "step": 29771
+ },
+ {
+ "epoch": 81.12261580381471,
+ "grad_norm": 1.1048951148986816,
+ "learning_rate": 1.8115037029433857e-06,
+ "loss": 0.0172,
+ "step": 29772
+ },
+ {
+ "epoch": 81.12534059945504,
+ "grad_norm": 0.666232705116272,
+ "learning_rate": 1.8109971757723055e-06,
+ "loss": 0.0065,
+ "step": 29773
+ },
+ {
+ "epoch": 81.12806539509536,
+ "grad_norm": 1.2019109725952148,
+ "learning_rate": 1.8104907123769155e-06,
+ "loss": 0.0159,
+ "step": 29774
+ },
+ {
+ "epoch": 81.1307901907357,
+ "grad_norm": 1.565062165260315,
+ "learning_rate": 1.8099843127611572e-06,
+ "loss": 0.0389,
+ "step": 29775
+ },
+ {
+ "epoch": 81.13351498637603,
+ "grad_norm": 0.767196536064148,
+ "learning_rate": 1.8094779769289782e-06,
+ "loss": 0.0068,
+ "step": 29776
+ },
+ {
+ "epoch": 81.13623978201635,
+ "grad_norm": 1.5010464191436768,
+ "learning_rate": 1.8089717048843236e-06,
+ "loss": 0.1316,
+ "step": 29777
+ },
+ {
+ "epoch": 81.13896457765668,
+ "grad_norm": 0.944647490978241,
+ "learning_rate": 1.8084654966311333e-06,
+ "loss": 0.0095,
+ "step": 29778
+ },
+ {
+ "epoch": 81.141689373297,
+ "grad_norm": 0.948310136795044,
+ "learning_rate": 1.8079593521733496e-06,
+ "loss": 0.0092,
+ "step": 29779
+ },
+ {
+ "epoch": 81.14441416893733,
+ "grad_norm": 0.7673332691192627,
+ "learning_rate": 1.8074532715149129e-06,
+ "loss": 0.0076,
+ "step": 29780
+ },
+ {
+ "epoch": 81.14713896457765,
+ "grad_norm": 1.8861192464828491,
+ "learning_rate": 1.806947254659769e-06,
+ "loss": 0.0173,
+ "step": 29781
+ },
+ {
+ "epoch": 81.14986376021798,
+ "grad_norm": 0.9863997101783752,
+ "learning_rate": 1.8064413016118566e-06,
+ "loss": 0.0625,
+ "step": 29782
+ },
+ {
+ "epoch": 81.15258855585832,
+ "grad_norm": 0.6693442463874817,
+ "learning_rate": 1.805935412375115e-06,
+ "loss": 0.0108,
+ "step": 29783
+ },
+ {
+ "epoch": 81.15531335149863,
+ "grad_norm": 0.8920835852622986,
+ "learning_rate": 1.8054295869534822e-06,
+ "loss": 0.0099,
+ "step": 29784
+ },
+ {
+ "epoch": 81.15803814713897,
+ "grad_norm": 1.3090935945510864,
+ "learning_rate": 1.8049238253509027e-06,
+ "loss": 0.0271,
+ "step": 29785
+ },
+ {
+ "epoch": 81.16076294277929,
+ "grad_norm": 1.5238908529281616,
+ "learning_rate": 1.8044181275713136e-06,
+ "loss": 0.0169,
+ "step": 29786
+ },
+ {
+ "epoch": 81.16348773841962,
+ "grad_norm": 3.5892138481140137,
+ "learning_rate": 1.803912493618648e-06,
+ "loss": 0.1367,
+ "step": 29787
+ },
+ {
+ "epoch": 81.16621253405995,
+ "grad_norm": 1.0096304416656494,
+ "learning_rate": 1.8034069234968521e-06,
+ "loss": 0.0196,
+ "step": 29788
+ },
+ {
+ "epoch": 81.16893732970027,
+ "grad_norm": 1.7012392282485962,
+ "learning_rate": 1.802901417209859e-06,
+ "loss": 0.0247,
+ "step": 29789
+ },
+ {
+ "epoch": 81.1716621253406,
+ "grad_norm": 0.4579229950904846,
+ "learning_rate": 1.802395974761605e-06,
+ "loss": 0.0057,
+ "step": 29790
+ },
+ {
+ "epoch": 81.17438692098092,
+ "grad_norm": 1.7174330949783325,
+ "learning_rate": 1.8018905961560252e-06,
+ "loss": 0.1156,
+ "step": 29791
+ },
+ {
+ "epoch": 81.17711171662125,
+ "grad_norm": 1.3226938247680664,
+ "learning_rate": 1.8013852813970578e-06,
+ "loss": 0.0256,
+ "step": 29792
+ },
+ {
+ "epoch": 81.17983651226157,
+ "grad_norm": 0.7744078040122986,
+ "learning_rate": 1.800880030488641e-06,
+ "loss": 0.0072,
+ "step": 29793
+ },
+ {
+ "epoch": 81.1825613079019,
+ "grad_norm": 0.7510553002357483,
+ "learning_rate": 1.8003748434347056e-06,
+ "loss": 0.0086,
+ "step": 29794
+ },
+ {
+ "epoch": 81.18528610354224,
+ "grad_norm": 0.5958682298660278,
+ "learning_rate": 1.7998697202391879e-06,
+ "loss": 0.0057,
+ "step": 29795
+ },
+ {
+ "epoch": 81.18801089918256,
+ "grad_norm": 1.2440439462661743,
+ "learning_rate": 1.7993646609060178e-06,
+ "loss": 0.0806,
+ "step": 29796
+ },
+ {
+ "epoch": 81.19073569482289,
+ "grad_norm": 1.457468032836914,
+ "learning_rate": 1.7988596654391344e-06,
+ "loss": 0.0104,
+ "step": 29797
+ },
+ {
+ "epoch": 81.19346049046321,
+ "grad_norm": 0.8475886583328247,
+ "learning_rate": 1.7983547338424679e-06,
+ "loss": 0.0271,
+ "step": 29798
+ },
+ {
+ "epoch": 81.19618528610354,
+ "grad_norm": 1.0457533597946167,
+ "learning_rate": 1.7978498661199517e-06,
+ "loss": 0.012,
+ "step": 29799
+ },
+ {
+ "epoch": 81.19891008174388,
+ "grad_norm": 1.4002381563186646,
+ "learning_rate": 1.7973450622755128e-06,
+ "loss": 0.0236,
+ "step": 29800
+ },
+ {
+ "epoch": 81.2016348773842,
+ "grad_norm": 1.2613555192947388,
+ "learning_rate": 1.7968403223130903e-06,
+ "loss": 0.0108,
+ "step": 29801
+ },
+ {
+ "epoch": 81.20435967302453,
+ "grad_norm": 0.7430508136749268,
+ "learning_rate": 1.79633564623661e-06,
+ "loss": 0.0058,
+ "step": 29802
+ },
+ {
+ "epoch": 81.20708446866485,
+ "grad_norm": 1.0540673732757568,
+ "learning_rate": 1.7958310340500052e-06,
+ "loss": 0.0117,
+ "step": 29803
+ },
+ {
+ "epoch": 81.20980926430518,
+ "grad_norm": 0.8126279711723328,
+ "learning_rate": 1.7953264857572007e-06,
+ "loss": 0.0072,
+ "step": 29804
+ },
+ {
+ "epoch": 81.2125340599455,
+ "grad_norm": 0.6868245601654053,
+ "learning_rate": 1.794822001362132e-06,
+ "loss": 0.0085,
+ "step": 29805
+ },
+ {
+ "epoch": 81.21525885558583,
+ "grad_norm": 0.5646125674247742,
+ "learning_rate": 1.7943175808687263e-06,
+ "loss": 0.0058,
+ "step": 29806
+ },
+ {
+ "epoch": 81.21798365122616,
+ "grad_norm": 1.8460376262664795,
+ "learning_rate": 1.793813224280908e-06,
+ "loss": 0.0343,
+ "step": 29807
+ },
+ {
+ "epoch": 81.22070844686648,
+ "grad_norm": 1.8080641031265259,
+ "learning_rate": 1.793308931602611e-06,
+ "loss": 0.0324,
+ "step": 29808
+ },
+ {
+ "epoch": 81.22343324250681,
+ "grad_norm": 1.4297146797180176,
+ "learning_rate": 1.7928047028377593e-06,
+ "loss": 0.0485,
+ "step": 29809
+ },
+ {
+ "epoch": 81.22615803814713,
+ "grad_norm": 0.9782441854476929,
+ "learning_rate": 1.792300537990278e-06,
+ "loss": 0.0178,
+ "step": 29810
+ },
+ {
+ "epoch": 81.22888283378747,
+ "grad_norm": 0.899713397026062,
+ "learning_rate": 1.7917964370640995e-06,
+ "loss": 0.0221,
+ "step": 29811
+ },
+ {
+ "epoch": 81.2316076294278,
+ "grad_norm": 1.2871118783950806,
+ "learning_rate": 1.7912924000631426e-06,
+ "loss": 0.014,
+ "step": 29812
+ },
+ {
+ "epoch": 81.23433242506812,
+ "grad_norm": 1.22330904006958,
+ "learning_rate": 1.790788426991339e-06,
+ "loss": 0.0196,
+ "step": 29813
+ },
+ {
+ "epoch": 81.23705722070845,
+ "grad_norm": 0.943957507610321,
+ "learning_rate": 1.7902845178526108e-06,
+ "loss": 0.0156,
+ "step": 29814
+ },
+ {
+ "epoch": 81.23978201634877,
+ "grad_norm": 0.7742528319358826,
+ "learning_rate": 1.7897806726508825e-06,
+ "loss": 0.0082,
+ "step": 29815
+ },
+ {
+ "epoch": 81.2425068119891,
+ "grad_norm": 0.5986607074737549,
+ "learning_rate": 1.789276891390076e-06,
+ "loss": 0.0056,
+ "step": 29816
+ },
+ {
+ "epoch": 81.24523160762942,
+ "grad_norm": 1.6101137399673462,
+ "learning_rate": 1.7887731740741199e-06,
+ "loss": 0.0637,
+ "step": 29817
+ },
+ {
+ "epoch": 81.24795640326975,
+ "grad_norm": 1.105849027633667,
+ "learning_rate": 1.7882695207069333e-06,
+ "loss": 0.0151,
+ "step": 29818
+ },
+ {
+ "epoch": 81.25068119891009,
+ "grad_norm": 1.1144073009490967,
+ "learning_rate": 1.787765931292439e-06,
+ "loss": 0.0079,
+ "step": 29819
+ },
+ {
+ "epoch": 81.2534059945504,
+ "grad_norm": 1.4825087785720825,
+ "learning_rate": 1.7872624058345578e-06,
+ "loss": 0.1084,
+ "step": 29820
+ },
+ {
+ "epoch": 81.25613079019074,
+ "grad_norm": 1.3467247486114502,
+ "learning_rate": 1.7867589443372145e-06,
+ "loss": 0.0533,
+ "step": 29821
+ },
+ {
+ "epoch": 81.25885558583106,
+ "grad_norm": 1.3456811904907227,
+ "learning_rate": 1.7862555468043286e-06,
+ "loss": 0.0789,
+ "step": 29822
+ },
+ {
+ "epoch": 81.26158038147139,
+ "grad_norm": 1.9234548807144165,
+ "learning_rate": 1.7857522132398197e-06,
+ "loss": 0.0193,
+ "step": 29823
+ },
+ {
+ "epoch": 81.26430517711172,
+ "grad_norm": 2.068068742752075,
+ "learning_rate": 1.785248943647605e-06,
+ "loss": 0.0269,
+ "step": 29824
+ },
+ {
+ "epoch": 81.26702997275204,
+ "grad_norm": 0.8173772692680359,
+ "learning_rate": 1.7847457380316114e-06,
+ "loss": 0.0058,
+ "step": 29825
+ },
+ {
+ "epoch": 81.26975476839237,
+ "grad_norm": 1.511385202407837,
+ "learning_rate": 1.7842425963957522e-06,
+ "loss": 0.0923,
+ "step": 29826
+ },
+ {
+ "epoch": 81.2724795640327,
+ "grad_norm": 1.194757103919983,
+ "learning_rate": 1.7837395187439454e-06,
+ "loss": 0.0121,
+ "step": 29827
+ },
+ {
+ "epoch": 81.27520435967303,
+ "grad_norm": 0.9505067467689514,
+ "learning_rate": 1.7832365050801104e-06,
+ "loss": 0.0077,
+ "step": 29828
+ },
+ {
+ "epoch": 81.27792915531334,
+ "grad_norm": 1.2452930212020874,
+ "learning_rate": 1.7827335554081671e-06,
+ "loss": 0.1086,
+ "step": 29829
+ },
+ {
+ "epoch": 81.28065395095368,
+ "grad_norm": 1.1334327459335327,
+ "learning_rate": 1.7822306697320313e-06,
+ "loss": 0.0503,
+ "step": 29830
+ },
+ {
+ "epoch": 81.28337874659401,
+ "grad_norm": 0.49408119916915894,
+ "learning_rate": 1.7817278480556188e-06,
+ "loss": 0.0042,
+ "step": 29831
+ },
+ {
+ "epoch": 81.28610354223433,
+ "grad_norm": 0.7817517518997192,
+ "learning_rate": 1.7812250903828421e-06,
+ "loss": 0.0066,
+ "step": 29832
+ },
+ {
+ "epoch": 81.28882833787466,
+ "grad_norm": 1.2831109762191772,
+ "learning_rate": 1.7807223967176223e-06,
+ "loss": 0.012,
+ "step": 29833
+ },
+ {
+ "epoch": 81.29155313351498,
+ "grad_norm": 0.785581111907959,
+ "learning_rate": 1.7802197670638722e-06,
+ "loss": 0.0086,
+ "step": 29834
+ },
+ {
+ "epoch": 81.29427792915531,
+ "grad_norm": 1.1404376029968262,
+ "learning_rate": 1.779717201425506e-06,
+ "loss": 0.0112,
+ "step": 29835
+ },
+ {
+ "epoch": 81.29700272479565,
+ "grad_norm": 0.8013392090797424,
+ "learning_rate": 1.7792146998064352e-06,
+ "loss": 0.0054,
+ "step": 29836
+ },
+ {
+ "epoch": 81.29972752043597,
+ "grad_norm": 0.9239001870155334,
+ "learning_rate": 1.778712262210578e-06,
+ "loss": 0.1393,
+ "step": 29837
+ },
+ {
+ "epoch": 81.3024523160763,
+ "grad_norm": 1.209990382194519,
+ "learning_rate": 1.7782098886418452e-06,
+ "loss": 0.0121,
+ "step": 29838
+ },
+ {
+ "epoch": 81.30517711171662,
+ "grad_norm": 1.0591483116149902,
+ "learning_rate": 1.77770757910415e-06,
+ "loss": 0.0125,
+ "step": 29839
+ },
+ {
+ "epoch": 81.30790190735695,
+ "grad_norm": 1.1180896759033203,
+ "learning_rate": 1.7772053336013994e-06,
+ "loss": 0.0781,
+ "step": 29840
+ },
+ {
+ "epoch": 81.31062670299727,
+ "grad_norm": 1.0803391933441162,
+ "learning_rate": 1.7767031521375133e-06,
+ "loss": 0.0276,
+ "step": 29841
+ },
+ {
+ "epoch": 81.3133514986376,
+ "grad_norm": 0.970026433467865,
+ "learning_rate": 1.7762010347163971e-06,
+ "loss": 0.0115,
+ "step": 29842
+ },
+ {
+ "epoch": 81.31607629427793,
+ "grad_norm": 1.5301984548568726,
+ "learning_rate": 1.7756989813419633e-06,
+ "loss": 0.0136,
+ "step": 29843
+ },
+ {
+ "epoch": 81.31880108991825,
+ "grad_norm": 0.925940215587616,
+ "learning_rate": 1.7751969920181179e-06,
+ "loss": 0.0109,
+ "step": 29844
+ },
+ {
+ "epoch": 81.32152588555859,
+ "grad_norm": 1.99742591381073,
+ "learning_rate": 1.774695066748774e-06,
+ "loss": 0.0208,
+ "step": 29845
+ },
+ {
+ "epoch": 81.3242506811989,
+ "grad_norm": 2.024193525314331,
+ "learning_rate": 1.774193205537843e-06,
+ "loss": 0.0156,
+ "step": 29846
+ },
+ {
+ "epoch": 81.32697547683924,
+ "grad_norm": 1.294040560722351,
+ "learning_rate": 1.7736914083892299e-06,
+ "loss": 0.0256,
+ "step": 29847
+ },
+ {
+ "epoch": 81.32970027247957,
+ "grad_norm": 1.0300416946411133,
+ "learning_rate": 1.7731896753068444e-06,
+ "loss": 0.0486,
+ "step": 29848
+ },
+ {
+ "epoch": 81.33242506811989,
+ "grad_norm": 1.0704145431518555,
+ "learning_rate": 1.7726880062945895e-06,
+ "loss": 0.0124,
+ "step": 29849
+ },
+ {
+ "epoch": 81.33514986376022,
+ "grad_norm": 1.9306104183197021,
+ "learning_rate": 1.7721864013563784e-06,
+ "loss": 0.091,
+ "step": 29850
+ },
+ {
+ "epoch": 81.33787465940054,
+ "grad_norm": 1.395646572113037,
+ "learning_rate": 1.7716848604961144e-06,
+ "loss": 0.0339,
+ "step": 29851
+ },
+ {
+ "epoch": 81.34059945504087,
+ "grad_norm": 1.1610318422317505,
+ "learning_rate": 1.7711833837177017e-06,
+ "loss": 0.0074,
+ "step": 29852
+ },
+ {
+ "epoch": 81.34332425068119,
+ "grad_norm": 1.8823410272598267,
+ "learning_rate": 1.7706819710250512e-06,
+ "loss": 0.0177,
+ "step": 29853
+ },
+ {
+ "epoch": 81.34604904632153,
+ "grad_norm": 1.064893126487732,
+ "learning_rate": 1.770180622422063e-06,
+ "loss": 0.0195,
+ "step": 29854
+ },
+ {
+ "epoch": 81.34877384196186,
+ "grad_norm": 0.8161922097206116,
+ "learning_rate": 1.7696793379126442e-06,
+ "loss": 0.143,
+ "step": 29855
+ },
+ {
+ "epoch": 81.35149863760218,
+ "grad_norm": 1.543809413909912,
+ "learning_rate": 1.769178117500696e-06,
+ "loss": 0.0701,
+ "step": 29856
+ },
+ {
+ "epoch": 81.35422343324251,
+ "grad_norm": 0.9607840180397034,
+ "learning_rate": 1.7686769611901255e-06,
+ "loss": 0.0098,
+ "step": 29857
+ },
+ {
+ "epoch": 81.35694822888283,
+ "grad_norm": 1.0600049495697021,
+ "learning_rate": 1.7681758689848338e-06,
+ "loss": 0.0338,
+ "step": 29858
+ },
+ {
+ "epoch": 81.35967302452316,
+ "grad_norm": 1.2097487449645996,
+ "learning_rate": 1.7676748408887235e-06,
+ "loss": 0.0248,
+ "step": 29859
+ },
+ {
+ "epoch": 81.3623978201635,
+ "grad_norm": 1.5173133611679077,
+ "learning_rate": 1.7671738769056945e-06,
+ "loss": 0.087,
+ "step": 29860
+ },
+ {
+ "epoch": 81.36512261580381,
+ "grad_norm": 1.2027069330215454,
+ "learning_rate": 1.7666729770396495e-06,
+ "loss": 0.0097,
+ "step": 29861
+ },
+ {
+ "epoch": 81.36784741144415,
+ "grad_norm": 14.19427490234375,
+ "learning_rate": 1.7661721412944943e-06,
+ "loss": 0.028,
+ "step": 29862
+ },
+ {
+ "epoch": 81.37057220708446,
+ "grad_norm": 1.316371202468872,
+ "learning_rate": 1.7656713696741246e-06,
+ "loss": 0.0204,
+ "step": 29863
+ },
+ {
+ "epoch": 81.3732970027248,
+ "grad_norm": 0.9014247059822083,
+ "learning_rate": 1.7651706621824405e-06,
+ "loss": 0.0208,
+ "step": 29864
+ },
+ {
+ "epoch": 81.37602179836512,
+ "grad_norm": 0.5979772806167603,
+ "learning_rate": 1.7646700188233411e-06,
+ "loss": 0.0053,
+ "step": 29865
+ },
+ {
+ "epoch": 81.37874659400545,
+ "grad_norm": 1.0005215406417847,
+ "learning_rate": 1.764169439600728e-06,
+ "loss": 0.073,
+ "step": 29866
+ },
+ {
+ "epoch": 81.38147138964578,
+ "grad_norm": 0.8646894693374634,
+ "learning_rate": 1.763668924518499e-06,
+ "loss": 0.0068,
+ "step": 29867
+ },
+ {
+ "epoch": 81.3841961852861,
+ "grad_norm": 1.1899572610855103,
+ "learning_rate": 1.763168473580551e-06,
+ "loss": 0.0316,
+ "step": 29868
+ },
+ {
+ "epoch": 81.38692098092643,
+ "grad_norm": 0.8330519795417786,
+ "learning_rate": 1.7626680867907787e-06,
+ "loss": 0.0148,
+ "step": 29869
+ },
+ {
+ "epoch": 81.38964577656675,
+ "grad_norm": 1.6386346817016602,
+ "learning_rate": 1.7621677641530855e-06,
+ "loss": 0.2198,
+ "step": 29870
+ },
+ {
+ "epoch": 81.39237057220708,
+ "grad_norm": 1.0470025539398193,
+ "learning_rate": 1.761667505671365e-06,
+ "loss": 0.0164,
+ "step": 29871
+ },
+ {
+ "epoch": 81.39509536784742,
+ "grad_norm": 0.6703610420227051,
+ "learning_rate": 1.7611673113495098e-06,
+ "loss": 0.008,
+ "step": 29872
+ },
+ {
+ "epoch": 81.39782016348774,
+ "grad_norm": 1.228548288345337,
+ "learning_rate": 1.7606671811914212e-06,
+ "loss": 0.0082,
+ "step": 29873
+ },
+ {
+ "epoch": 81.40054495912807,
+ "grad_norm": 0.7772295475006104,
+ "learning_rate": 1.7601671152009903e-06,
+ "loss": 0.0071,
+ "step": 29874
+ },
+ {
+ "epoch": 81.40326975476839,
+ "grad_norm": 0.7885545492172241,
+ "learning_rate": 1.7596671133821142e-06,
+ "loss": 0.012,
+ "step": 29875
+ },
+ {
+ "epoch": 81.40599455040872,
+ "grad_norm": 0.6203483939170837,
+ "learning_rate": 1.7591671757386818e-06,
+ "loss": 0.0062,
+ "step": 29876
+ },
+ {
+ "epoch": 81.40871934604904,
+ "grad_norm": 1.0607985258102417,
+ "learning_rate": 1.7586673022745936e-06,
+ "loss": 0.0648,
+ "step": 29877
+ },
+ {
+ "epoch": 81.41144414168937,
+ "grad_norm": 0.9789583683013916,
+ "learning_rate": 1.7581674929937354e-06,
+ "loss": 0.0079,
+ "step": 29878
+ },
+ {
+ "epoch": 81.4141689373297,
+ "grad_norm": 0.6505737900733948,
+ "learning_rate": 1.7576677479000072e-06,
+ "loss": 0.0054,
+ "step": 29879
+ },
+ {
+ "epoch": 81.41689373297002,
+ "grad_norm": 1.2228714227676392,
+ "learning_rate": 1.7571680669972969e-06,
+ "loss": 0.0651,
+ "step": 29880
+ },
+ {
+ "epoch": 81.41961852861036,
+ "grad_norm": 0.8164642453193665,
+ "learning_rate": 1.7566684502894938e-06,
+ "loss": 0.0117,
+ "step": 29881
+ },
+ {
+ "epoch": 81.42234332425068,
+ "grad_norm": 1.0326436758041382,
+ "learning_rate": 1.7561688977804946e-06,
+ "loss": 0.009,
+ "step": 29882
+ },
+ {
+ "epoch": 81.42506811989101,
+ "grad_norm": 1.0266211032867432,
+ "learning_rate": 1.7556694094741867e-06,
+ "loss": 0.009,
+ "step": 29883
+ },
+ {
+ "epoch": 81.42779291553134,
+ "grad_norm": 1.1048088073730469,
+ "learning_rate": 1.7551699853744587e-06,
+ "loss": 0.0197,
+ "step": 29884
+ },
+ {
+ "epoch": 81.43051771117166,
+ "grad_norm": 1.4133096933364868,
+ "learning_rate": 1.7546706254852008e-06,
+ "loss": 0.0203,
+ "step": 29885
+ },
+ {
+ "epoch": 81.433242506812,
+ "grad_norm": 1.2474074363708496,
+ "learning_rate": 1.7541713298103036e-06,
+ "loss": 0.0115,
+ "step": 29886
+ },
+ {
+ "epoch": 81.43596730245231,
+ "grad_norm": 1.4261817932128906,
+ "learning_rate": 1.753672098353656e-06,
+ "loss": 0.0221,
+ "step": 29887
+ },
+ {
+ "epoch": 81.43869209809264,
+ "grad_norm": 1.2089245319366455,
+ "learning_rate": 1.7531729311191436e-06,
+ "loss": 0.0204,
+ "step": 29888
+ },
+ {
+ "epoch": 81.44141689373296,
+ "grad_norm": 1.131346344947815,
+ "learning_rate": 1.7526738281106537e-06,
+ "loss": 0.0324,
+ "step": 29889
+ },
+ {
+ "epoch": 81.4441416893733,
+ "grad_norm": 1.1939404010772705,
+ "learning_rate": 1.7521747893320762e-06,
+ "loss": 0.0251,
+ "step": 29890
+ },
+ {
+ "epoch": 81.44686648501363,
+ "grad_norm": 1.0166103839874268,
+ "learning_rate": 1.7516758147872958e-06,
+ "loss": 0.0205,
+ "step": 29891
+ },
+ {
+ "epoch": 81.44959128065395,
+ "grad_norm": 1.0733625888824463,
+ "learning_rate": 1.7511769044801996e-06,
+ "loss": 0.0107,
+ "step": 29892
+ },
+ {
+ "epoch": 81.45231607629428,
+ "grad_norm": 1.3097907304763794,
+ "learning_rate": 1.7506780584146688e-06,
+ "loss": 0.0437,
+ "step": 29893
+ },
+ {
+ "epoch": 81.4550408719346,
+ "grad_norm": 1.6683440208435059,
+ "learning_rate": 1.7501792765945947e-06,
+ "loss": 0.0636,
+ "step": 29894
+ },
+ {
+ "epoch": 81.45776566757493,
+ "grad_norm": 1.447029948234558,
+ "learning_rate": 1.7496805590238586e-06,
+ "loss": 0.0151,
+ "step": 29895
+ },
+ {
+ "epoch": 81.46049046321527,
+ "grad_norm": 1.0474745035171509,
+ "learning_rate": 1.7491819057063419e-06,
+ "loss": 0.0153,
+ "step": 29896
+ },
+ {
+ "epoch": 81.46321525885558,
+ "grad_norm": 1.7265455722808838,
+ "learning_rate": 1.7486833166459316e-06,
+ "loss": 0.0188,
+ "step": 29897
+ },
+ {
+ "epoch": 81.46594005449592,
+ "grad_norm": 0.8863412141799927,
+ "learning_rate": 1.7481847918465133e-06,
+ "loss": 0.0058,
+ "step": 29898
+ },
+ {
+ "epoch": 81.46866485013624,
+ "grad_norm": 0.864849328994751,
+ "learning_rate": 1.7476863313119653e-06,
+ "loss": 0.0306,
+ "step": 29899
+ },
+ {
+ "epoch": 81.47138964577657,
+ "grad_norm": 0.5415080189704895,
+ "learning_rate": 1.7471879350461706e-06,
+ "loss": 0.0066,
+ "step": 29900
+ },
+ {
+ "epoch": 81.47411444141689,
+ "grad_norm": 1.357985019683838,
+ "learning_rate": 1.7466896030530078e-06,
+ "loss": 0.0139,
+ "step": 29901
+ },
+ {
+ "epoch": 81.47683923705722,
+ "grad_norm": 0.8897461891174316,
+ "learning_rate": 1.7461913353363635e-06,
+ "loss": 0.0173,
+ "step": 29902
+ },
+ {
+ "epoch": 81.47956403269755,
+ "grad_norm": 1.1827187538146973,
+ "learning_rate": 1.745693131900116e-06,
+ "loss": 0.0291,
+ "step": 29903
+ },
+ {
+ "epoch": 81.48228882833787,
+ "grad_norm": 28.36092185974121,
+ "learning_rate": 1.745194992748145e-06,
+ "loss": 0.0098,
+ "step": 29904
+ },
+ {
+ "epoch": 81.4850136239782,
+ "grad_norm": 0.6548804640769958,
+ "learning_rate": 1.7446969178843265e-06,
+ "loss": 0.0065,
+ "step": 29905
+ },
+ {
+ "epoch": 81.48773841961852,
+ "grad_norm": 0.8392534255981445,
+ "learning_rate": 1.7441989073125453e-06,
+ "loss": 0.011,
+ "step": 29906
+ },
+ {
+ "epoch": 81.49046321525886,
+ "grad_norm": 1.0972822904586792,
+ "learning_rate": 1.7437009610366773e-06,
+ "loss": 0.0129,
+ "step": 29907
+ },
+ {
+ "epoch": 81.49318801089919,
+ "grad_norm": 0.8448237776756287,
+ "learning_rate": 1.7432030790606003e-06,
+ "loss": 0.0114,
+ "step": 29908
+ },
+ {
+ "epoch": 81.49591280653951,
+ "grad_norm": 1.386980414390564,
+ "learning_rate": 1.742705261388189e-06,
+ "loss": 0.0107,
+ "step": 29909
+ },
+ {
+ "epoch": 81.49863760217984,
+ "grad_norm": 1.218396544456482,
+ "learning_rate": 1.742207508023327e-06,
+ "loss": 0.0284,
+ "step": 29910
+ },
+ {
+ "epoch": 81.50136239782016,
+ "grad_norm": 0.5941455960273743,
+ "learning_rate": 1.7417098189698866e-06,
+ "loss": 0.0097,
+ "step": 29911
+ },
+ {
+ "epoch": 81.50408719346049,
+ "grad_norm": 1.5838618278503418,
+ "learning_rate": 1.741212194231744e-06,
+ "loss": 0.0647,
+ "step": 29912
+ },
+ {
+ "epoch": 81.50681198910081,
+ "grad_norm": 0.8697401285171509,
+ "learning_rate": 1.7407146338127713e-06,
+ "loss": 0.0117,
+ "step": 29913
+ },
+ {
+ "epoch": 81.50953678474114,
+ "grad_norm": 0.8789445757865906,
+ "learning_rate": 1.7402171377168509e-06,
+ "loss": 0.009,
+ "step": 29914
+ },
+ {
+ "epoch": 81.51226158038148,
+ "grad_norm": 1.3755091428756714,
+ "learning_rate": 1.7397197059478543e-06,
+ "loss": 0.0417,
+ "step": 29915
+ },
+ {
+ "epoch": 81.5149863760218,
+ "grad_norm": 1.0537010431289673,
+ "learning_rate": 1.7392223385096542e-06,
+ "loss": 0.0145,
+ "step": 29916
+ },
+ {
+ "epoch": 81.51771117166213,
+ "grad_norm": 0.9021462202072144,
+ "learning_rate": 1.7387250354061225e-06,
+ "loss": 0.0114,
+ "step": 29917
+ },
+ {
+ "epoch": 81.52043596730245,
+ "grad_norm": 1.3797096014022827,
+ "learning_rate": 1.7382277966411366e-06,
+ "loss": 0.0169,
+ "step": 29918
+ },
+ {
+ "epoch": 81.52316076294278,
+ "grad_norm": 0.6377888917922974,
+ "learning_rate": 1.7377306222185663e-06,
+ "loss": 0.01,
+ "step": 29919
+ },
+ {
+ "epoch": 81.52588555858311,
+ "grad_norm": 1.1186705827713013,
+ "learning_rate": 1.7372335121422857e-06,
+ "loss": 0.0133,
+ "step": 29920
+ },
+ {
+ "epoch": 81.52861035422343,
+ "grad_norm": 1.1760271787643433,
+ "learning_rate": 1.7367364664161602e-06,
+ "loss": 0.0205,
+ "step": 29921
+ },
+ {
+ "epoch": 81.53133514986376,
+ "grad_norm": 1.1381725072860718,
+ "learning_rate": 1.736239485044069e-06,
+ "loss": 0.0391,
+ "step": 29922
+ },
+ {
+ "epoch": 81.53405994550408,
+ "grad_norm": 0.8846547603607178,
+ "learning_rate": 1.7357425680298778e-06,
+ "loss": 0.0094,
+ "step": 29923
+ },
+ {
+ "epoch": 81.53678474114442,
+ "grad_norm": 1.6590447425842285,
+ "learning_rate": 1.7352457153774583e-06,
+ "loss": 0.0448,
+ "step": 29924
+ },
+ {
+ "epoch": 81.53950953678473,
+ "grad_norm": 0.7003193497657776,
+ "learning_rate": 1.734748927090676e-06,
+ "loss": 0.0076,
+ "step": 29925
+ },
+ {
+ "epoch": 81.54223433242507,
+ "grad_norm": 1.2451996803283691,
+ "learning_rate": 1.7342522031734054e-06,
+ "loss": 0.0118,
+ "step": 29926
+ },
+ {
+ "epoch": 81.5449591280654,
+ "grad_norm": 1.444868564605713,
+ "learning_rate": 1.7337555436295128e-06,
+ "loss": 0.015,
+ "step": 29927
+ },
+ {
+ "epoch": 81.54768392370572,
+ "grad_norm": 1.2185842990875244,
+ "learning_rate": 1.733258948462866e-06,
+ "loss": 0.0163,
+ "step": 29928
+ },
+ {
+ "epoch": 81.55040871934605,
+ "grad_norm": 0.8273036479949951,
+ "learning_rate": 1.7327624176773306e-06,
+ "loss": 0.0095,
+ "step": 29929
+ },
+ {
+ "epoch": 81.55313351498637,
+ "grad_norm": 1.296770691871643,
+ "learning_rate": 1.7322659512767748e-06,
+ "loss": 0.0701,
+ "step": 29930
+ },
+ {
+ "epoch": 81.5558583106267,
+ "grad_norm": 1.7962931394577026,
+ "learning_rate": 1.7317695492650676e-06,
+ "loss": 0.0254,
+ "step": 29931
+ },
+ {
+ "epoch": 81.55858310626704,
+ "grad_norm": 1.8406153917312622,
+ "learning_rate": 1.731273211646074e-06,
+ "loss": 0.0133,
+ "step": 29932
+ },
+ {
+ "epoch": 81.56130790190736,
+ "grad_norm": 1.0126819610595703,
+ "learning_rate": 1.7307769384236584e-06,
+ "loss": 0.0301,
+ "step": 29933
+ },
+ {
+ "epoch": 81.56403269754769,
+ "grad_norm": 1.9395678043365479,
+ "learning_rate": 1.7302807296016833e-06,
+ "loss": 0.0359,
+ "step": 29934
+ },
+ {
+ "epoch": 81.566757493188,
+ "grad_norm": 1.2593821287155151,
+ "learning_rate": 1.7297845851840177e-06,
+ "loss": 0.048,
+ "step": 29935
+ },
+ {
+ "epoch": 81.56948228882834,
+ "grad_norm": 1.1943647861480713,
+ "learning_rate": 1.7292885051745246e-06,
+ "loss": 0.0259,
+ "step": 29936
+ },
+ {
+ "epoch": 81.57220708446866,
+ "grad_norm": 1.215572476387024,
+ "learning_rate": 1.7287924895770624e-06,
+ "loss": 0.0112,
+ "step": 29937
+ },
+ {
+ "epoch": 81.57493188010899,
+ "grad_norm": 1.3051522970199585,
+ "learning_rate": 1.7282965383955019e-06,
+ "loss": 0.0094,
+ "step": 29938
+ },
+ {
+ "epoch": 81.57765667574932,
+ "grad_norm": 1.687208652496338,
+ "learning_rate": 1.7278006516337008e-06,
+ "loss": 0.0124,
+ "step": 29939
+ },
+ {
+ "epoch": 81.58038147138964,
+ "grad_norm": 1.1389355659484863,
+ "learning_rate": 1.727304829295523e-06,
+ "loss": 0.0912,
+ "step": 29940
+ },
+ {
+ "epoch": 81.58310626702998,
+ "grad_norm": 1.7097781896591187,
+ "learning_rate": 1.7268090713848262e-06,
+ "loss": 0.0285,
+ "step": 29941
+ },
+ {
+ "epoch": 81.5858310626703,
+ "grad_norm": 1.2220464944839478,
+ "learning_rate": 1.7263133779054763e-06,
+ "loss": 0.0079,
+ "step": 29942
+ },
+ {
+ "epoch": 81.58855585831063,
+ "grad_norm": 0.6210678219795227,
+ "learning_rate": 1.7258177488613316e-06,
+ "loss": 0.0044,
+ "step": 29943
+ },
+ {
+ "epoch": 81.59128065395096,
+ "grad_norm": 1.0382039546966553,
+ "learning_rate": 1.7253221842562518e-06,
+ "loss": 0.0092,
+ "step": 29944
+ },
+ {
+ "epoch": 81.59400544959128,
+ "grad_norm": 1.4768733978271484,
+ "learning_rate": 1.7248266840940941e-06,
+ "loss": 0.0126,
+ "step": 29945
+ },
+ {
+ "epoch": 81.59673024523161,
+ "grad_norm": 1.7909175157546997,
+ "learning_rate": 1.7243312483787222e-06,
+ "loss": 0.018,
+ "step": 29946
+ },
+ {
+ "epoch": 81.59945504087193,
+ "grad_norm": 0.4352174401283264,
+ "learning_rate": 1.7238358771139896e-06,
+ "loss": 0.0043,
+ "step": 29947
+ },
+ {
+ "epoch": 81.60217983651226,
+ "grad_norm": 0.9727744460105896,
+ "learning_rate": 1.7233405703037599e-06,
+ "loss": 0.0028,
+ "step": 29948
+ },
+ {
+ "epoch": 81.60490463215258,
+ "grad_norm": 1.1287450790405273,
+ "learning_rate": 1.7228453279518863e-06,
+ "loss": 0.0505,
+ "step": 29949
+ },
+ {
+ "epoch": 81.60762942779292,
+ "grad_norm": 2.8113198280334473,
+ "learning_rate": 1.7223501500622252e-06,
+ "loss": 0.0439,
+ "step": 29950
+ },
+ {
+ "epoch": 81.61035422343325,
+ "grad_norm": 1.312502145767212,
+ "learning_rate": 1.7218550366386378e-06,
+ "loss": 0.0265,
+ "step": 29951
+ },
+ {
+ "epoch": 81.61307901907357,
+ "grad_norm": 0.9728286266326904,
+ "learning_rate": 1.7213599876849762e-06,
+ "loss": 0.0134,
+ "step": 29952
+ },
+ {
+ "epoch": 81.6158038147139,
+ "grad_norm": 0.9771829843521118,
+ "learning_rate": 1.7208650032050966e-06,
+ "loss": 0.0147,
+ "step": 29953
+ },
+ {
+ "epoch": 81.61852861035422,
+ "grad_norm": 1.303297996520996,
+ "learning_rate": 1.7203700832028513e-06,
+ "loss": 0.0183,
+ "step": 29954
+ },
+ {
+ "epoch": 81.62125340599455,
+ "grad_norm": 0.6815423965454102,
+ "learning_rate": 1.7198752276821007e-06,
+ "loss": 0.0066,
+ "step": 29955
+ },
+ {
+ "epoch": 81.62397820163488,
+ "grad_norm": 1.3491597175598145,
+ "learning_rate": 1.7193804366466948e-06,
+ "loss": 0.0504,
+ "step": 29956
+ },
+ {
+ "epoch": 81.6267029972752,
+ "grad_norm": 1.4391828775405884,
+ "learning_rate": 1.7188857101004873e-06,
+ "loss": 0.0427,
+ "step": 29957
+ },
+ {
+ "epoch": 81.62942779291554,
+ "grad_norm": 0.8916184902191162,
+ "learning_rate": 1.7183910480473287e-06,
+ "loss": 0.0884,
+ "step": 29958
+ },
+ {
+ "epoch": 81.63215258855585,
+ "grad_norm": 1.8263884782791138,
+ "learning_rate": 1.7178964504910766e-06,
+ "loss": 0.012,
+ "step": 29959
+ },
+ {
+ "epoch": 81.63487738419619,
+ "grad_norm": 0.8211814165115356,
+ "learning_rate": 1.7174019174355805e-06,
+ "loss": 0.0171,
+ "step": 29960
+ },
+ {
+ "epoch": 81.6376021798365,
+ "grad_norm": 1.0665431022644043,
+ "learning_rate": 1.7169074488846905e-06,
+ "loss": 0.0165,
+ "step": 29961
+ },
+ {
+ "epoch": 81.64032697547684,
+ "grad_norm": 0.8527252674102783,
+ "learning_rate": 1.716413044842259e-06,
+ "loss": 0.0123,
+ "step": 29962
+ },
+ {
+ "epoch": 81.64305177111717,
+ "grad_norm": 1.6678776741027832,
+ "learning_rate": 1.7159187053121373e-06,
+ "loss": 0.0525,
+ "step": 29963
+ },
+ {
+ "epoch": 81.64577656675749,
+ "grad_norm": 1.2479768991470337,
+ "learning_rate": 1.7154244302981715e-06,
+ "loss": 0.0357,
+ "step": 29964
+ },
+ {
+ "epoch": 81.64850136239782,
+ "grad_norm": 1.3206435441970825,
+ "learning_rate": 1.7149302198042162e-06,
+ "loss": 0.0259,
+ "step": 29965
+ },
+ {
+ "epoch": 81.65122615803814,
+ "grad_norm": 1.4872303009033203,
+ "learning_rate": 1.7144360738341148e-06,
+ "loss": 0.0425,
+ "step": 29966
+ },
+ {
+ "epoch": 81.65395095367847,
+ "grad_norm": 1.0423409938812256,
+ "learning_rate": 1.713941992391721e-06,
+ "loss": 0.0121,
+ "step": 29967
+ },
+ {
+ "epoch": 81.65667574931881,
+ "grad_norm": 1.3756407499313354,
+ "learning_rate": 1.7134479754808808e-06,
+ "loss": 0.0209,
+ "step": 29968
+ },
+ {
+ "epoch": 81.65940054495913,
+ "grad_norm": 1.6636056900024414,
+ "learning_rate": 1.712954023105441e-06,
+ "loss": 0.0286,
+ "step": 29969
+ },
+ {
+ "epoch": 81.66212534059946,
+ "grad_norm": 0.7374190092086792,
+ "learning_rate": 1.712460135269246e-06,
+ "loss": 0.0072,
+ "step": 29970
+ },
+ {
+ "epoch": 81.66485013623978,
+ "grad_norm": 0.7958109974861145,
+ "learning_rate": 1.7119663119761476e-06,
+ "loss": 0.0071,
+ "step": 29971
+ },
+ {
+ "epoch": 81.66757493188011,
+ "grad_norm": 1.1904085874557495,
+ "learning_rate": 1.711472553229988e-06,
+ "loss": 0.0162,
+ "step": 29972
+ },
+ {
+ "epoch": 81.67029972752043,
+ "grad_norm": 0.7558897137641907,
+ "learning_rate": 1.7109788590346132e-06,
+ "loss": 0.0058,
+ "step": 29973
+ },
+ {
+ "epoch": 81.67302452316076,
+ "grad_norm": 1.294895052909851,
+ "learning_rate": 1.7104852293938668e-06,
+ "loss": 0.0109,
+ "step": 29974
+ },
+ {
+ "epoch": 81.6757493188011,
+ "grad_norm": 1.8263660669326782,
+ "learning_rate": 1.7099916643115965e-06,
+ "loss": 0.0736,
+ "step": 29975
+ },
+ {
+ "epoch": 81.67847411444141,
+ "grad_norm": 1.2798954248428345,
+ "learning_rate": 1.709498163791644e-06,
+ "loss": 0.0209,
+ "step": 29976
+ },
+ {
+ "epoch": 81.68119891008175,
+ "grad_norm": 0.7412692904472351,
+ "learning_rate": 1.709004727837854e-06,
+ "loss": 0.0146,
+ "step": 29977
+ },
+ {
+ "epoch": 81.68392370572207,
+ "grad_norm": 2.3632123470306396,
+ "learning_rate": 1.7085113564540657e-06,
+ "loss": 0.1662,
+ "step": 29978
+ },
+ {
+ "epoch": 81.6866485013624,
+ "grad_norm": 0.5678695440292358,
+ "learning_rate": 1.7080180496441257e-06,
+ "loss": 0.0053,
+ "step": 29979
+ },
+ {
+ "epoch": 81.68937329700273,
+ "grad_norm": 1.1573899984359741,
+ "learning_rate": 1.7075248074118745e-06,
+ "loss": 0.0125,
+ "step": 29980
+ },
+ {
+ "epoch": 81.69209809264305,
+ "grad_norm": 1.015529990196228,
+ "learning_rate": 1.7070316297611512e-06,
+ "loss": 0.0143,
+ "step": 29981
+ },
+ {
+ "epoch": 81.69482288828338,
+ "grad_norm": 1.1002167463302612,
+ "learning_rate": 1.7065385166957982e-06,
+ "loss": 0.0124,
+ "step": 29982
+ },
+ {
+ "epoch": 81.6975476839237,
+ "grad_norm": 0.9726895689964294,
+ "learning_rate": 1.7060454682196592e-06,
+ "loss": 0.0191,
+ "step": 29983
+ },
+ {
+ "epoch": 81.70027247956403,
+ "grad_norm": 0.800695538520813,
+ "learning_rate": 1.7055524843365711e-06,
+ "loss": 0.0079,
+ "step": 29984
+ },
+ {
+ "epoch": 81.70299727520435,
+ "grad_norm": 1.4920068979263306,
+ "learning_rate": 1.7050595650503732e-06,
+ "loss": 0.0857,
+ "step": 29985
+ },
+ {
+ "epoch": 81.70572207084469,
+ "grad_norm": 0.7367912530899048,
+ "learning_rate": 1.7045667103649021e-06,
+ "loss": 0.0092,
+ "step": 29986
+ },
+ {
+ "epoch": 81.70844686648502,
+ "grad_norm": 2.2714953422546387,
+ "learning_rate": 1.7040739202840006e-06,
+ "loss": 0.1287,
+ "step": 29987
+ },
+ {
+ "epoch": 81.71117166212534,
+ "grad_norm": 1.9381186962127686,
+ "learning_rate": 1.7035811948115044e-06,
+ "loss": 0.0131,
+ "step": 29988
+ },
+ {
+ "epoch": 81.71389645776567,
+ "grad_norm": 0.7585522532463074,
+ "learning_rate": 1.7030885339512505e-06,
+ "loss": 0.016,
+ "step": 29989
+ },
+ {
+ "epoch": 81.71662125340599,
+ "grad_norm": 1.4214805364608765,
+ "learning_rate": 1.7025959377070745e-06,
+ "loss": 0.0137,
+ "step": 29990
+ },
+ {
+ "epoch": 81.71934604904632,
+ "grad_norm": 1.0640227794647217,
+ "learning_rate": 1.7021034060828157e-06,
+ "loss": 0.0069,
+ "step": 29991
+ },
+ {
+ "epoch": 81.72207084468666,
+ "grad_norm": 1.9587284326553345,
+ "learning_rate": 1.70161093908231e-06,
+ "loss": 0.0777,
+ "step": 29992
+ },
+ {
+ "epoch": 81.72479564032697,
+ "grad_norm": 2.119757890701294,
+ "learning_rate": 1.7011185367093897e-06,
+ "loss": 0.0426,
+ "step": 29993
+ },
+ {
+ "epoch": 81.7275204359673,
+ "grad_norm": 1.2096681594848633,
+ "learning_rate": 1.7006261989678895e-06,
+ "loss": 0.0177,
+ "step": 29994
+ },
+ {
+ "epoch": 81.73024523160763,
+ "grad_norm": 2.297252655029297,
+ "learning_rate": 1.7001339258616478e-06,
+ "loss": 0.028,
+ "step": 29995
+ },
+ {
+ "epoch": 81.73297002724796,
+ "grad_norm": 0.9516564011573792,
+ "learning_rate": 1.6996417173944957e-06,
+ "loss": 0.0147,
+ "step": 29996
+ },
+ {
+ "epoch": 81.73569482288828,
+ "grad_norm": 0.9134554862976074,
+ "learning_rate": 1.6991495735702667e-06,
+ "loss": 0.0776,
+ "step": 29997
+ },
+ {
+ "epoch": 81.73841961852861,
+ "grad_norm": 1.2586597204208374,
+ "learning_rate": 1.6986574943927913e-06,
+ "loss": 0.0213,
+ "step": 29998
+ },
+ {
+ "epoch": 81.74114441416894,
+ "grad_norm": 1.4809881448745728,
+ "learning_rate": 1.6981654798659041e-06,
+ "loss": 0.0119,
+ "step": 29999
+ },
+ {
+ "epoch": 81.74386920980926,
+ "grad_norm": 0.911523163318634,
+ "learning_rate": 1.69767352999344e-06,
+ "loss": 0.0109,
+ "step": 30000
+ },
+ {
+ "epoch": 81.7465940054496,
+ "grad_norm": 1.1354886293411255,
+ "learning_rate": 1.6971816447792266e-06,
+ "loss": 0.0971,
+ "step": 30001
+ },
+ {
+ "epoch": 81.74931880108991,
+ "grad_norm": 0.9044008255004883,
+ "learning_rate": 1.6966898242270935e-06,
+ "loss": 0.0068,
+ "step": 30002
+ },
+ {
+ "epoch": 81.75204359673025,
+ "grad_norm": 1.071486473083496,
+ "learning_rate": 1.6961980683408742e-06,
+ "loss": 0.0212,
+ "step": 30003
+ },
+ {
+ "epoch": 81.75476839237058,
+ "grad_norm": 1.0784000158309937,
+ "learning_rate": 1.6957063771243977e-06,
+ "loss": 0.0124,
+ "step": 30004
+ },
+ {
+ "epoch": 81.7574931880109,
+ "grad_norm": 1.4698678255081177,
+ "learning_rate": 1.695214750581493e-06,
+ "loss": 0.0097,
+ "step": 30005
+ },
+ {
+ "epoch": 81.76021798365123,
+ "grad_norm": 0.9475911259651184,
+ "learning_rate": 1.6947231887159854e-06,
+ "loss": 0.0105,
+ "step": 30006
+ },
+ {
+ "epoch": 81.76294277929155,
+ "grad_norm": 1.5377850532531738,
+ "learning_rate": 1.6942316915317091e-06,
+ "loss": 0.0578,
+ "step": 30007
+ },
+ {
+ "epoch": 81.76566757493188,
+ "grad_norm": 0.5196578502655029,
+ "learning_rate": 1.6937402590324892e-06,
+ "loss": 0.0049,
+ "step": 30008
+ },
+ {
+ "epoch": 81.7683923705722,
+ "grad_norm": 1.8021284341812134,
+ "learning_rate": 1.6932488912221523e-06,
+ "loss": 0.0184,
+ "step": 30009
+ },
+ {
+ "epoch": 81.77111716621253,
+ "grad_norm": 0.9065690636634827,
+ "learning_rate": 1.6927575881045243e-06,
+ "loss": 0.0105,
+ "step": 30010
+ },
+ {
+ "epoch": 81.77384196185287,
+ "grad_norm": 1.4030945301055908,
+ "learning_rate": 1.6922663496834346e-06,
+ "loss": 0.045,
+ "step": 30011
+ },
+ {
+ "epoch": 81.77656675749319,
+ "grad_norm": 1.873130440711975,
+ "learning_rate": 1.6917751759627066e-06,
+ "loss": 0.086,
+ "step": 30012
+ },
+ {
+ "epoch": 81.77929155313352,
+ "grad_norm": 1.010912299156189,
+ "learning_rate": 1.6912840669461672e-06,
+ "loss": 0.0171,
+ "step": 30013
+ },
+ {
+ "epoch": 81.78201634877384,
+ "grad_norm": 1.4127763509750366,
+ "learning_rate": 1.6907930226376367e-06,
+ "loss": 0.0507,
+ "step": 30014
+ },
+ {
+ "epoch": 81.78474114441417,
+ "grad_norm": 0.8484329581260681,
+ "learning_rate": 1.6903020430409456e-06,
+ "loss": 0.0105,
+ "step": 30015
+ },
+ {
+ "epoch": 81.7874659400545,
+ "grad_norm": 1.7571585178375244,
+ "learning_rate": 1.6898111281599129e-06,
+ "loss": 0.0317,
+ "step": 30016
+ },
+ {
+ "epoch": 81.79019073569482,
+ "grad_norm": 3.4716835021972656,
+ "learning_rate": 1.6893202779983654e-06,
+ "loss": 0.0186,
+ "step": 30017
+ },
+ {
+ "epoch": 81.79291553133515,
+ "grad_norm": 0.6748204827308655,
+ "learning_rate": 1.6888294925601235e-06,
+ "loss": 0.0072,
+ "step": 30018
+ },
+ {
+ "epoch": 81.79564032697547,
+ "grad_norm": 1.16807222366333,
+ "learning_rate": 1.6883387718490086e-06,
+ "loss": 0.1379,
+ "step": 30019
+ },
+ {
+ "epoch": 81.7983651226158,
+ "grad_norm": 1.0325136184692383,
+ "learning_rate": 1.6878481158688454e-06,
+ "loss": 0.1435,
+ "step": 30020
+ },
+ {
+ "epoch": 81.80108991825612,
+ "grad_norm": 1.1663093566894531,
+ "learning_rate": 1.6873575246234542e-06,
+ "loss": 0.0125,
+ "step": 30021
+ },
+ {
+ "epoch": 81.80381471389646,
+ "grad_norm": 1.2948620319366455,
+ "learning_rate": 1.6868669981166553e-06,
+ "loss": 0.103,
+ "step": 30022
+ },
+ {
+ "epoch": 81.80653950953679,
+ "grad_norm": 1.1987512111663818,
+ "learning_rate": 1.6863765363522654e-06,
+ "loss": 0.007,
+ "step": 30023
+ },
+ {
+ "epoch": 81.80926430517711,
+ "grad_norm": 1.1392887830734253,
+ "learning_rate": 1.685886139334111e-06,
+ "loss": 0.0103,
+ "step": 30024
+ },
+ {
+ "epoch": 81.81198910081744,
+ "grad_norm": 0.6347302794456482,
+ "learning_rate": 1.685395807066007e-06,
+ "loss": 0.0057,
+ "step": 30025
+ },
+ {
+ "epoch": 81.81471389645776,
+ "grad_norm": 1.560043454170227,
+ "learning_rate": 1.68490553955177e-06,
+ "loss": 0.052,
+ "step": 30026
+ },
+ {
+ "epoch": 81.8174386920981,
+ "grad_norm": 1.1465222835540771,
+ "learning_rate": 1.6844153367952242e-06,
+ "loss": 0.0151,
+ "step": 30027
+ },
+ {
+ "epoch": 81.82016348773843,
+ "grad_norm": 0.8741336464881897,
+ "learning_rate": 1.6839251988001838e-06,
+ "loss": 0.0152,
+ "step": 30028
+ },
+ {
+ "epoch": 81.82288828337875,
+ "grad_norm": 0.6355151534080505,
+ "learning_rate": 1.6834351255704673e-06,
+ "loss": 0.0039,
+ "step": 30029
+ },
+ {
+ "epoch": 81.82561307901908,
+ "grad_norm": 1.5448594093322754,
+ "learning_rate": 1.6829451171098865e-06,
+ "loss": 0.0468,
+ "step": 30030
+ },
+ {
+ "epoch": 81.8283378746594,
+ "grad_norm": 1.836379051208496,
+ "learning_rate": 1.6824551734222649e-06,
+ "loss": 0.05,
+ "step": 30031
+ },
+ {
+ "epoch": 81.83106267029973,
+ "grad_norm": 0.6769028902053833,
+ "learning_rate": 1.681965294511414e-06,
+ "loss": 0.0069,
+ "step": 30032
+ },
+ {
+ "epoch": 81.83378746594005,
+ "grad_norm": 1.3211177587509155,
+ "learning_rate": 1.6814754803811474e-06,
+ "loss": 0.0462,
+ "step": 30033
+ },
+ {
+ "epoch": 81.83651226158038,
+ "grad_norm": 0.7192840576171875,
+ "learning_rate": 1.6809857310352852e-06,
+ "loss": 0.0057,
+ "step": 30034
+ },
+ {
+ "epoch": 81.83923705722071,
+ "grad_norm": 0.7326059937477112,
+ "learning_rate": 1.6804960464776344e-06,
+ "loss": 0.0054,
+ "step": 30035
+ },
+ {
+ "epoch": 81.84196185286103,
+ "grad_norm": 1.424622893333435,
+ "learning_rate": 1.6800064267120163e-06,
+ "loss": 0.039,
+ "step": 30036
+ },
+ {
+ "epoch": 81.84468664850137,
+ "grad_norm": 1.7116928100585938,
+ "learning_rate": 1.6795168717422404e-06,
+ "loss": 0.0144,
+ "step": 30037
+ },
+ {
+ "epoch": 81.84741144414168,
+ "grad_norm": 0.7746756076812744,
+ "learning_rate": 1.6790273815721191e-06,
+ "loss": 0.0053,
+ "step": 30038
+ },
+ {
+ "epoch": 81.85013623978202,
+ "grad_norm": 0.9554222226142883,
+ "learning_rate": 1.6785379562054615e-06,
+ "loss": 0.0148,
+ "step": 30039
+ },
+ {
+ "epoch": 81.85286103542235,
+ "grad_norm": 1.6457794904708862,
+ "learning_rate": 1.6780485956460857e-06,
+ "loss": 0.0418,
+ "step": 30040
+ },
+ {
+ "epoch": 81.85558583106267,
+ "grad_norm": 1.261488914489746,
+ "learning_rate": 1.6775592998977995e-06,
+ "loss": 0.06,
+ "step": 30041
+ },
+ {
+ "epoch": 81.858310626703,
+ "grad_norm": 1.1652650833129883,
+ "learning_rate": 1.6770700689644138e-06,
+ "loss": 0.0339,
+ "step": 30042
+ },
+ {
+ "epoch": 81.86103542234332,
+ "grad_norm": 0.8180764317512512,
+ "learning_rate": 1.676580902849736e-06,
+ "loss": 0.0069,
+ "step": 30043
+ },
+ {
+ "epoch": 81.86376021798365,
+ "grad_norm": 1.8339259624481201,
+ "learning_rate": 1.67609180155758e-06,
+ "loss": 0.0941,
+ "step": 30044
+ },
+ {
+ "epoch": 81.86648501362397,
+ "grad_norm": 0.9967746138572693,
+ "learning_rate": 1.6756027650917539e-06,
+ "loss": 0.0232,
+ "step": 30045
+ },
+ {
+ "epoch": 81.8692098092643,
+ "grad_norm": 3.081847906112671,
+ "learning_rate": 1.6751137934560635e-06,
+ "loss": 0.0214,
+ "step": 30046
+ },
+ {
+ "epoch": 81.87193460490464,
+ "grad_norm": 1.4285300970077515,
+ "learning_rate": 1.6746248866543201e-06,
+ "loss": 0.0282,
+ "step": 30047
+ },
+ {
+ "epoch": 81.87465940054496,
+ "grad_norm": 2.6369221210479736,
+ "learning_rate": 1.6741360446903309e-06,
+ "loss": 0.1037,
+ "step": 30048
+ },
+ {
+ "epoch": 81.87738419618529,
+ "grad_norm": 1.8950984477996826,
+ "learning_rate": 1.6736472675679028e-06,
+ "loss": 0.0804,
+ "step": 30049
+ },
+ {
+ "epoch": 81.88010899182561,
+ "grad_norm": 0.9257028102874756,
+ "learning_rate": 1.6731585552908381e-06,
+ "loss": 0.0092,
+ "step": 30050
+ },
+ {
+ "epoch": 81.88283378746594,
+ "grad_norm": 0.6872841119766235,
+ "learning_rate": 1.672669907862947e-06,
+ "loss": 0.0097,
+ "step": 30051
+ },
+ {
+ "epoch": 81.88555858310627,
+ "grad_norm": 0.3960573077201843,
+ "learning_rate": 1.672181325288037e-06,
+ "loss": 0.0039,
+ "step": 30052
+ },
+ {
+ "epoch": 81.88828337874659,
+ "grad_norm": 0.458296000957489,
+ "learning_rate": 1.6716928075699112e-06,
+ "loss": 0.0055,
+ "step": 30053
+ },
+ {
+ "epoch": 81.89100817438693,
+ "grad_norm": 1.3301618099212646,
+ "learning_rate": 1.6712043547123735e-06,
+ "loss": 0.104,
+ "step": 30054
+ },
+ {
+ "epoch": 81.89373297002724,
+ "grad_norm": 1.250206470489502,
+ "learning_rate": 1.6707159667192263e-06,
+ "loss": 0.0451,
+ "step": 30055
+ },
+ {
+ "epoch": 81.89645776566758,
+ "grad_norm": 1.1338279247283936,
+ "learning_rate": 1.6702276435942776e-06,
+ "loss": 0.0201,
+ "step": 30056
+ },
+ {
+ "epoch": 81.8991825613079,
+ "grad_norm": 1.321953296661377,
+ "learning_rate": 1.6697393853413269e-06,
+ "loss": 0.0209,
+ "step": 30057
+ },
+ {
+ "epoch": 81.90190735694823,
+ "grad_norm": 0.8693960309028625,
+ "learning_rate": 1.6692511919641786e-06,
+ "loss": 0.0087,
+ "step": 30058
+ },
+ {
+ "epoch": 81.90463215258856,
+ "grad_norm": 0.7306152582168579,
+ "learning_rate": 1.6687630634666308e-06,
+ "loss": 0.0068,
+ "step": 30059
+ },
+ {
+ "epoch": 81.90735694822888,
+ "grad_norm": 1.7227165699005127,
+ "learning_rate": 1.6682749998524917e-06,
+ "loss": 0.0366,
+ "step": 30060
+ },
+ {
+ "epoch": 81.91008174386921,
+ "grad_norm": 1.2479939460754395,
+ "learning_rate": 1.667787001125557e-06,
+ "loss": 0.0115,
+ "step": 30061
+ },
+ {
+ "epoch": 81.91280653950953,
+ "grad_norm": 1.0959688425064087,
+ "learning_rate": 1.6672990672896304e-06,
+ "loss": 0.0225,
+ "step": 30062
+ },
+ {
+ "epoch": 81.91553133514986,
+ "grad_norm": 1.244680643081665,
+ "learning_rate": 1.666811198348508e-06,
+ "loss": 0.0923,
+ "step": 30063
+ },
+ {
+ "epoch": 81.9182561307902,
+ "grad_norm": 1.6484622955322266,
+ "learning_rate": 1.6663233943059932e-06,
+ "loss": 0.0164,
+ "step": 30064
+ },
+ {
+ "epoch": 81.92098092643052,
+ "grad_norm": 0.823180079460144,
+ "learning_rate": 1.6658356551658838e-06,
+ "loss": 0.0076,
+ "step": 30065
+ },
+ {
+ "epoch": 81.92370572207085,
+ "grad_norm": 0.8261279463768005,
+ "learning_rate": 1.6653479809319795e-06,
+ "loss": 0.0045,
+ "step": 30066
+ },
+ {
+ "epoch": 81.92643051771117,
+ "grad_norm": 1.0794459581375122,
+ "learning_rate": 1.6648603716080725e-06,
+ "loss": 0.0087,
+ "step": 30067
+ },
+ {
+ "epoch": 81.9291553133515,
+ "grad_norm": 1.2903634309768677,
+ "learning_rate": 1.6643728271979665e-06,
+ "loss": 0.0124,
+ "step": 30068
+ },
+ {
+ "epoch": 81.93188010899182,
+ "grad_norm": 1.3662464618682861,
+ "learning_rate": 1.6638853477054573e-06,
+ "loss": 0.0119,
+ "step": 30069
+ },
+ {
+ "epoch": 81.93460490463215,
+ "grad_norm": 1.2574185132980347,
+ "learning_rate": 1.663397933134342e-06,
+ "loss": 0.1312,
+ "step": 30070
+ },
+ {
+ "epoch": 81.93732970027249,
+ "grad_norm": 0.8762077689170837,
+ "learning_rate": 1.6629105834884119e-06,
+ "loss": 0.0081,
+ "step": 30071
+ },
+ {
+ "epoch": 81.9400544959128,
+ "grad_norm": 0.6338369846343994,
+ "learning_rate": 1.6624232987714695e-06,
+ "loss": 0.0069,
+ "step": 30072
+ },
+ {
+ "epoch": 81.94277929155314,
+ "grad_norm": 1.689451813697815,
+ "learning_rate": 1.6619360789873051e-06,
+ "loss": 0.0772,
+ "step": 30073
+ },
+ {
+ "epoch": 81.94550408719346,
+ "grad_norm": 0.7298003435134888,
+ "learning_rate": 1.6614489241397148e-06,
+ "loss": 0.0094,
+ "step": 30074
+ },
+ {
+ "epoch": 81.94822888283379,
+ "grad_norm": 0.7243173718452454,
+ "learning_rate": 1.6609618342324884e-06,
+ "loss": 0.0072,
+ "step": 30075
+ },
+ {
+ "epoch": 81.95095367847412,
+ "grad_norm": 1.06134033203125,
+ "learning_rate": 1.6604748092694268e-06,
+ "loss": 0.083,
+ "step": 30076
+ },
+ {
+ "epoch": 81.95367847411444,
+ "grad_norm": 1.4295382499694824,
+ "learning_rate": 1.6599878492543175e-06,
+ "loss": 0.0158,
+ "step": 30077
+ },
+ {
+ "epoch": 81.95640326975477,
+ "grad_norm": 1.0759676694869995,
+ "learning_rate": 1.6595009541909558e-06,
+ "loss": 0.0141,
+ "step": 30078
+ },
+ {
+ "epoch": 81.95912806539509,
+ "grad_norm": 0.9659489393234253,
+ "learning_rate": 1.6590141240831281e-06,
+ "loss": 0.0073,
+ "step": 30079
+ },
+ {
+ "epoch": 81.96185286103542,
+ "grad_norm": 1.3621917963027954,
+ "learning_rate": 1.658527358934634e-06,
+ "loss": 0.0147,
+ "step": 30080
+ },
+ {
+ "epoch": 81.96457765667574,
+ "grad_norm": 1.0149317979812622,
+ "learning_rate": 1.6580406587492592e-06,
+ "loss": 0.0091,
+ "step": 30081
+ },
+ {
+ "epoch": 81.96730245231608,
+ "grad_norm": 1.3488472700119019,
+ "learning_rate": 1.6575540235307964e-06,
+ "loss": 0.0252,
+ "step": 30082
+ },
+ {
+ "epoch": 81.97002724795641,
+ "grad_norm": 1.372117042541504,
+ "learning_rate": 1.6570674532830312e-06,
+ "loss": 0.0579,
+ "step": 30083
+ },
+ {
+ "epoch": 81.97275204359673,
+ "grad_norm": 1.4567408561706543,
+ "learning_rate": 1.6565809480097583e-06,
+ "loss": 0.1334,
+ "step": 30084
+ },
+ {
+ "epoch": 81.97547683923706,
+ "grad_norm": 1.187963843345642,
+ "learning_rate": 1.6560945077147628e-06,
+ "loss": 0.0268,
+ "step": 30085
+ },
+ {
+ "epoch": 81.97820163487738,
+ "grad_norm": 0.9162863492965698,
+ "learning_rate": 1.655608132401837e-06,
+ "loss": 0.0083,
+ "step": 30086
+ },
+ {
+ "epoch": 81.98092643051771,
+ "grad_norm": 1.0554120540618896,
+ "learning_rate": 1.6551218220747634e-06,
+ "loss": 0.0146,
+ "step": 30087
+ },
+ {
+ "epoch": 81.98365122615803,
+ "grad_norm": 1.8134150505065918,
+ "learning_rate": 1.6546355767373357e-06,
+ "loss": 0.0266,
+ "step": 30088
+ },
+ {
+ "epoch": 81.98637602179836,
+ "grad_norm": 1.3121217489242554,
+ "learning_rate": 1.6541493963933375e-06,
+ "loss": 0.0154,
+ "step": 30089
+ },
+ {
+ "epoch": 81.9891008174387,
+ "grad_norm": 0.9532355070114136,
+ "learning_rate": 1.6536632810465547e-06,
+ "loss": 0.0123,
+ "step": 30090
+ },
+ {
+ "epoch": 81.99182561307902,
+ "grad_norm": 0.7915785908699036,
+ "learning_rate": 1.653177230700771e-06,
+ "loss": 0.0091,
+ "step": 30091
+ },
+ {
+ "epoch": 81.99455040871935,
+ "grad_norm": 1.389866828918457,
+ "learning_rate": 1.6526912453597776e-06,
+ "loss": 0.0671,
+ "step": 30092
+ },
+ {
+ "epoch": 81.99727520435967,
+ "grad_norm": 0.7845536470413208,
+ "learning_rate": 1.6522053250273551e-06,
+ "loss": 0.0111,
+ "step": 30093
+ },
+ {
+ "epoch": 82.0,
+ "grad_norm": 1.2198553085327148,
+ "learning_rate": 1.6517194697072903e-06,
+ "loss": 0.0207,
+ "step": 30094
+ },
+ {
+ "epoch": 82.00272479564033,
+ "grad_norm": 3.188927173614502,
+ "learning_rate": 1.6512336794033622e-06,
+ "loss": 0.0144,
+ "step": 30095
+ },
+ {
+ "epoch": 82.00544959128065,
+ "grad_norm": 1.5179226398468018,
+ "learning_rate": 1.6507479541193595e-06,
+ "loss": 0.0175,
+ "step": 30096
+ },
+ {
+ "epoch": 82.00817438692098,
+ "grad_norm": 1.2778687477111816,
+ "learning_rate": 1.650262293859064e-06,
+ "loss": 0.11,
+ "step": 30097
+ },
+ {
+ "epoch": 82.0108991825613,
+ "grad_norm": 1.0118077993392944,
+ "learning_rate": 1.6497766986262564e-06,
+ "loss": 0.0085,
+ "step": 30098
+ },
+ {
+ "epoch": 82.01362397820164,
+ "grad_norm": 1.9388890266418457,
+ "learning_rate": 1.6492911684247171e-06,
+ "loss": 0.1177,
+ "step": 30099
+ },
+ {
+ "epoch": 82.01634877384195,
+ "grad_norm": 1.2852920293807983,
+ "learning_rate": 1.6488057032582316e-06,
+ "loss": 0.0119,
+ "step": 30100
+ },
+ {
+ "epoch": 82.01907356948229,
+ "grad_norm": 0.7121536135673523,
+ "learning_rate": 1.6483203031305784e-06,
+ "loss": 0.0086,
+ "step": 30101
+ },
+ {
+ "epoch": 82.02179836512262,
+ "grad_norm": 0.8186589479446411,
+ "learning_rate": 1.6478349680455351e-06,
+ "loss": 0.0102,
+ "step": 30102
+ },
+ {
+ "epoch": 82.02452316076294,
+ "grad_norm": 0.39556604623794556,
+ "learning_rate": 1.6473496980068882e-06,
+ "loss": 0.0043,
+ "step": 30103
+ },
+ {
+ "epoch": 82.02724795640327,
+ "grad_norm": 2.508228302001953,
+ "learning_rate": 1.6468644930184097e-06,
+ "loss": 0.0238,
+ "step": 30104
+ },
+ {
+ "epoch": 82.02997275204359,
+ "grad_norm": 1.1084777116775513,
+ "learning_rate": 1.6463793530838834e-06,
+ "loss": 0.0259,
+ "step": 30105
+ },
+ {
+ "epoch": 82.03269754768392,
+ "grad_norm": 1.3652406930923462,
+ "learning_rate": 1.6458942782070864e-06,
+ "loss": 0.0074,
+ "step": 30106
+ },
+ {
+ "epoch": 82.03542234332426,
+ "grad_norm": 0.8981074690818787,
+ "learning_rate": 1.6454092683917954e-06,
+ "loss": 0.0253,
+ "step": 30107
+ },
+ {
+ "epoch": 82.03814713896458,
+ "grad_norm": 2.0052361488342285,
+ "learning_rate": 1.6449243236417866e-06,
+ "loss": 0.0812,
+ "step": 30108
+ },
+ {
+ "epoch": 82.04087193460491,
+ "grad_norm": 1.3904856443405151,
+ "learning_rate": 1.6444394439608402e-06,
+ "loss": 0.0119,
+ "step": 30109
+ },
+ {
+ "epoch": 82.04359673024523,
+ "grad_norm": 1.130008578300476,
+ "learning_rate": 1.6439546293527309e-06,
+ "loss": 0.0158,
+ "step": 30110
+ },
+ {
+ "epoch": 82.04632152588556,
+ "grad_norm": 1.085877776145935,
+ "learning_rate": 1.6434698798212312e-06,
+ "loss": 0.0861,
+ "step": 30111
+ },
+ {
+ "epoch": 82.04904632152588,
+ "grad_norm": 1.538804531097412,
+ "learning_rate": 1.6429851953701214e-06,
+ "loss": 0.0248,
+ "step": 30112
+ },
+ {
+ "epoch": 82.05177111716621,
+ "grad_norm": 0.5452132225036621,
+ "learning_rate": 1.6425005760031742e-06,
+ "loss": 0.0052,
+ "step": 30113
+ },
+ {
+ "epoch": 82.05449591280654,
+ "grad_norm": 1.470569133758545,
+ "learning_rate": 1.642016021724163e-06,
+ "loss": 0.0199,
+ "step": 30114
+ },
+ {
+ "epoch": 82.05722070844686,
+ "grad_norm": 1.2086175680160522,
+ "learning_rate": 1.6415315325368608e-06,
+ "loss": 0.0156,
+ "step": 30115
+ },
+ {
+ "epoch": 82.0599455040872,
+ "grad_norm": 1.2150112390518188,
+ "learning_rate": 1.641047108445044e-06,
+ "loss": 0.0471,
+ "step": 30116
+ },
+ {
+ "epoch": 82.06267029972751,
+ "grad_norm": 1.5707359313964844,
+ "learning_rate": 1.6405627494524823e-06,
+ "loss": 0.1657,
+ "step": 30117
+ },
+ {
+ "epoch": 82.06539509536785,
+ "grad_norm": 0.8169738054275513,
+ "learning_rate": 1.64007845556295e-06,
+ "loss": 0.0117,
+ "step": 30118
+ },
+ {
+ "epoch": 82.06811989100818,
+ "grad_norm": 0.9129536151885986,
+ "learning_rate": 1.6395942267802145e-06,
+ "loss": 0.0072,
+ "step": 30119
+ },
+ {
+ "epoch": 82.0708446866485,
+ "grad_norm": 1.6486246585845947,
+ "learning_rate": 1.6391100631080504e-06,
+ "loss": 0.0262,
+ "step": 30120
+ },
+ {
+ "epoch": 82.07356948228883,
+ "grad_norm": 1.4208093881607056,
+ "learning_rate": 1.63862596455023e-06,
+ "loss": 0.0136,
+ "step": 30121
+ },
+ {
+ "epoch": 82.07629427792915,
+ "grad_norm": 1.0638511180877686,
+ "learning_rate": 1.6381419311105217e-06,
+ "loss": 0.0111,
+ "step": 30122
+ },
+ {
+ "epoch": 82.07901907356948,
+ "grad_norm": 1.861801266670227,
+ "learning_rate": 1.6376579627926958e-06,
+ "loss": 0.03,
+ "step": 30123
+ },
+ {
+ "epoch": 82.0817438692098,
+ "grad_norm": 1.4968713521957397,
+ "learning_rate": 1.637174059600517e-06,
+ "loss": 0.0239,
+ "step": 30124
+ },
+ {
+ "epoch": 82.08446866485014,
+ "grad_norm": 1.4951095581054688,
+ "learning_rate": 1.6366902215377612e-06,
+ "loss": 0.0179,
+ "step": 30125
+ },
+ {
+ "epoch": 82.08719346049047,
+ "grad_norm": 0.7292322516441345,
+ "learning_rate": 1.6362064486081908e-06,
+ "loss": 0.0058,
+ "step": 30126
+ },
+ {
+ "epoch": 82.08991825613079,
+ "grad_norm": 1.23666512966156,
+ "learning_rate": 1.6357227408155773e-06,
+ "loss": 0.015,
+ "step": 30127
+ },
+ {
+ "epoch": 82.09264305177112,
+ "grad_norm": 1.0522898435592651,
+ "learning_rate": 1.6352390981636823e-06,
+ "loss": 0.0323,
+ "step": 30128
+ },
+ {
+ "epoch": 82.09536784741144,
+ "grad_norm": 0.44787538051605225,
+ "learning_rate": 1.6347555206562792e-06,
+ "loss": 0.0042,
+ "step": 30129
+ },
+ {
+ "epoch": 82.09809264305177,
+ "grad_norm": 2.8151934146881104,
+ "learning_rate": 1.6342720082971297e-06,
+ "loss": 0.0235,
+ "step": 30130
+ },
+ {
+ "epoch": 82.1008174386921,
+ "grad_norm": 1.4800801277160645,
+ "learning_rate": 1.6337885610900017e-06,
+ "loss": 0.0572,
+ "step": 30131
+ },
+ {
+ "epoch": 82.10354223433242,
+ "grad_norm": 2.13505482673645,
+ "learning_rate": 1.6333051790386556e-06,
+ "loss": 0.0448,
+ "step": 30132
+ },
+ {
+ "epoch": 82.10626702997276,
+ "grad_norm": 0.849990963935852,
+ "learning_rate": 1.6328218621468628e-06,
+ "loss": 0.0109,
+ "step": 30133
+ },
+ {
+ "epoch": 82.10899182561307,
+ "grad_norm": 0.8356029987335205,
+ "learning_rate": 1.6323386104183836e-06,
+ "loss": 0.0089,
+ "step": 30134
+ },
+ {
+ "epoch": 82.11171662125341,
+ "grad_norm": 0.7750582695007324,
+ "learning_rate": 1.6318554238569807e-06,
+ "loss": 0.0058,
+ "step": 30135
+ },
+ {
+ "epoch": 82.11444141689373,
+ "grad_norm": 0.9882745146751404,
+ "learning_rate": 1.6313723024664175e-06,
+ "loss": 0.0105,
+ "step": 30136
+ },
+ {
+ "epoch": 82.11716621253406,
+ "grad_norm": 1.1588122844696045,
+ "learning_rate": 1.6308892462504599e-06,
+ "loss": 0.0087,
+ "step": 30137
+ },
+ {
+ "epoch": 82.11989100817439,
+ "grad_norm": 1.0091924667358398,
+ "learning_rate": 1.6304062552128674e-06,
+ "loss": 0.022,
+ "step": 30138
+ },
+ {
+ "epoch": 82.12261580381471,
+ "grad_norm": 1.482271671295166,
+ "learning_rate": 1.6299233293574012e-06,
+ "loss": 0.0194,
+ "step": 30139
+ },
+ {
+ "epoch": 82.12534059945504,
+ "grad_norm": 0.8233141303062439,
+ "learning_rate": 1.6294404686878207e-06,
+ "loss": 0.0096,
+ "step": 30140
+ },
+ {
+ "epoch": 82.12806539509536,
+ "grad_norm": 1.022288203239441,
+ "learning_rate": 1.6289576732078916e-06,
+ "loss": 0.0082,
+ "step": 30141
+ },
+ {
+ "epoch": 82.1307901907357,
+ "grad_norm": 0.5397799611091614,
+ "learning_rate": 1.6284749429213697e-06,
+ "loss": 0.0064,
+ "step": 30142
+ },
+ {
+ "epoch": 82.13351498637603,
+ "grad_norm": 1.2828294038772583,
+ "learning_rate": 1.6279922778320156e-06,
+ "loss": 0.095,
+ "step": 30143
+ },
+ {
+ "epoch": 82.13623978201635,
+ "grad_norm": 0.7367492318153381,
+ "learning_rate": 1.6275096779435862e-06,
+ "loss": 0.0075,
+ "step": 30144
+ },
+ {
+ "epoch": 82.13896457765668,
+ "grad_norm": 1.0684360265731812,
+ "learning_rate": 1.6270271432598438e-06,
+ "loss": 0.0299,
+ "step": 30145
+ },
+ {
+ "epoch": 82.141689373297,
+ "grad_norm": 0.9778770804405212,
+ "learning_rate": 1.6265446737845458e-06,
+ "loss": 0.0078,
+ "step": 30146
+ },
+ {
+ "epoch": 82.14441416893733,
+ "grad_norm": 1.3151588439941406,
+ "learning_rate": 1.626062269521448e-06,
+ "loss": 0.0264,
+ "step": 30147
+ },
+ {
+ "epoch": 82.14713896457765,
+ "grad_norm": 0.6394616365432739,
+ "learning_rate": 1.6255799304743048e-06,
+ "loss": 0.0062,
+ "step": 30148
+ },
+ {
+ "epoch": 82.14986376021798,
+ "grad_norm": 0.7759509086608887,
+ "learning_rate": 1.6250976566468779e-06,
+ "loss": 0.0097,
+ "step": 30149
+ },
+ {
+ "epoch": 82.15258855585832,
+ "grad_norm": 0.9575676321983337,
+ "learning_rate": 1.624615448042921e-06,
+ "loss": 0.0059,
+ "step": 30150
+ },
+ {
+ "epoch": 82.15531335149863,
+ "grad_norm": 1.145212173461914,
+ "learning_rate": 1.6241333046661888e-06,
+ "loss": 0.0469,
+ "step": 30151
+ },
+ {
+ "epoch": 82.15803814713897,
+ "grad_norm": 0.4940405488014221,
+ "learning_rate": 1.6236512265204352e-06,
+ "loss": 0.0067,
+ "step": 30152
+ },
+ {
+ "epoch": 82.16076294277929,
+ "grad_norm": 0.9451754093170166,
+ "learning_rate": 1.623169213609419e-06,
+ "loss": 0.0097,
+ "step": 30153
+ },
+ {
+ "epoch": 82.16348773841962,
+ "grad_norm": 1.433803915977478,
+ "learning_rate": 1.6226872659368885e-06,
+ "loss": 0.0137,
+ "step": 30154
+ },
+ {
+ "epoch": 82.16621253405995,
+ "grad_norm": 0.9813557863235474,
+ "learning_rate": 1.622205383506602e-06,
+ "loss": 0.0162,
+ "step": 30155
+ },
+ {
+ "epoch": 82.16893732970027,
+ "grad_norm": 1.2337979078292847,
+ "learning_rate": 1.6217235663223085e-06,
+ "loss": 0.0299,
+ "step": 30156
+ },
+ {
+ "epoch": 82.1716621253406,
+ "grad_norm": 1.4243110418319702,
+ "learning_rate": 1.621241814387764e-06,
+ "loss": 0.0222,
+ "step": 30157
+ },
+ {
+ "epoch": 82.17438692098092,
+ "grad_norm": 1.2291576862335205,
+ "learning_rate": 1.6207601277067197e-06,
+ "loss": 0.0109,
+ "step": 30158
+ },
+ {
+ "epoch": 82.17711171662125,
+ "grad_norm": 1.187192440032959,
+ "learning_rate": 1.6202785062829252e-06,
+ "loss": 0.0197,
+ "step": 30159
+ },
+ {
+ "epoch": 82.17983651226157,
+ "grad_norm": 0.7339444756507874,
+ "learning_rate": 1.6197969501201295e-06,
+ "loss": 0.0055,
+ "step": 30160
+ },
+ {
+ "epoch": 82.1825613079019,
+ "grad_norm": 1.3780763149261475,
+ "learning_rate": 1.6193154592220871e-06,
+ "loss": 0.0357,
+ "step": 30161
+ },
+ {
+ "epoch": 82.18528610354224,
+ "grad_norm": 1.1085867881774902,
+ "learning_rate": 1.6188340335925478e-06,
+ "loss": 0.0205,
+ "step": 30162
+ },
+ {
+ "epoch": 82.18801089918256,
+ "grad_norm": 0.9985409379005432,
+ "learning_rate": 1.6183526732352583e-06,
+ "loss": 0.0121,
+ "step": 30163
+ },
+ {
+ "epoch": 82.19073569482289,
+ "grad_norm": 1.3492578268051147,
+ "learning_rate": 1.6178713781539658e-06,
+ "loss": 0.0133,
+ "step": 30164
+ },
+ {
+ "epoch": 82.19346049046321,
+ "grad_norm": 0.9891548752784729,
+ "learning_rate": 1.6173901483524246e-06,
+ "loss": 0.0732,
+ "step": 30165
+ },
+ {
+ "epoch": 82.19618528610354,
+ "grad_norm": 1.0871168375015259,
+ "learning_rate": 1.6169089838343776e-06,
+ "loss": 0.0259,
+ "step": 30166
+ },
+ {
+ "epoch": 82.19891008174388,
+ "grad_norm": 1.0995230674743652,
+ "learning_rate": 1.6164278846035752e-06,
+ "loss": 0.0371,
+ "step": 30167
+ },
+ {
+ "epoch": 82.2016348773842,
+ "grad_norm": 1.026180386543274,
+ "learning_rate": 1.61594685066376e-06,
+ "loss": 0.008,
+ "step": 30168
+ },
+ {
+ "epoch": 82.20435967302453,
+ "grad_norm": 2.69376802444458,
+ "learning_rate": 1.6154658820186819e-06,
+ "loss": 0.0439,
+ "step": 30169
+ },
+ {
+ "epoch": 82.20708446866485,
+ "grad_norm": 1.4062000513076782,
+ "learning_rate": 1.6149849786720873e-06,
+ "loss": 0.0579,
+ "step": 30170
+ },
+ {
+ "epoch": 82.20980926430518,
+ "grad_norm": 1.410487174987793,
+ "learning_rate": 1.6145041406277162e-06,
+ "loss": 0.0189,
+ "step": 30171
+ },
+ {
+ "epoch": 82.2125340599455,
+ "grad_norm": 1.1724950075149536,
+ "learning_rate": 1.6140233678893203e-06,
+ "loss": 0.013,
+ "step": 30172
+ },
+ {
+ "epoch": 82.21525885558583,
+ "grad_norm": 1.533851981163025,
+ "learning_rate": 1.6135426604606375e-06,
+ "loss": 0.0324,
+ "step": 30173
+ },
+ {
+ "epoch": 82.21798365122616,
+ "grad_norm": 1.1366310119628906,
+ "learning_rate": 1.613062018345417e-06,
+ "loss": 0.0463,
+ "step": 30174
+ },
+ {
+ "epoch": 82.22070844686648,
+ "grad_norm": 1.1689008474349976,
+ "learning_rate": 1.6125814415473994e-06,
+ "loss": 0.0148,
+ "step": 30175
+ },
+ {
+ "epoch": 82.22343324250681,
+ "grad_norm": 1.6861674785614014,
+ "learning_rate": 1.6121009300703262e-06,
+ "loss": 0.0339,
+ "step": 30176
+ },
+ {
+ "epoch": 82.22615803814713,
+ "grad_norm": 0.7651405930519104,
+ "learning_rate": 1.6116204839179428e-06,
+ "loss": 0.0093,
+ "step": 30177
+ },
+ {
+ "epoch": 82.22888283378747,
+ "grad_norm": 1.2753280401229858,
+ "learning_rate": 1.611140103093989e-06,
+ "loss": 0.0207,
+ "step": 30178
+ },
+ {
+ "epoch": 82.2316076294278,
+ "grad_norm": 1.0871697664260864,
+ "learning_rate": 1.6106597876022057e-06,
+ "loss": 0.0078,
+ "step": 30179
+ },
+ {
+ "epoch": 82.23433242506812,
+ "grad_norm": 1.2178541421890259,
+ "learning_rate": 1.6101795374463314e-06,
+ "loss": 0.0103,
+ "step": 30180
+ },
+ {
+ "epoch": 82.23705722070845,
+ "grad_norm": 1.2519407272338867,
+ "learning_rate": 1.6096993526301118e-06,
+ "loss": 0.0431,
+ "step": 30181
+ },
+ {
+ "epoch": 82.23978201634877,
+ "grad_norm": 1.8292075395584106,
+ "learning_rate": 1.6092192331572843e-06,
+ "loss": 0.0124,
+ "step": 30182
+ },
+ {
+ "epoch": 82.2425068119891,
+ "grad_norm": 0.7711525559425354,
+ "learning_rate": 1.6087391790315866e-06,
+ "loss": 0.0096,
+ "step": 30183
+ },
+ {
+ "epoch": 82.24523160762942,
+ "grad_norm": 1.3814290761947632,
+ "learning_rate": 1.6082591902567557e-06,
+ "loss": 0.0623,
+ "step": 30184
+ },
+ {
+ "epoch": 82.24795640326975,
+ "grad_norm": 1.0573571920394897,
+ "learning_rate": 1.6077792668365344e-06,
+ "loss": 0.1466,
+ "step": 30185
+ },
+ {
+ "epoch": 82.25068119891009,
+ "grad_norm": 0.7536629438400269,
+ "learning_rate": 1.6072994087746585e-06,
+ "loss": 0.0116,
+ "step": 30186
+ },
+ {
+ "epoch": 82.2534059945504,
+ "grad_norm": 1.1137065887451172,
+ "learning_rate": 1.606819616074864e-06,
+ "loss": 0.0516,
+ "step": 30187
+ },
+ {
+ "epoch": 82.25613079019074,
+ "grad_norm": 0.679571807384491,
+ "learning_rate": 1.6063398887408866e-06,
+ "loss": 0.0064,
+ "step": 30188
+ },
+ {
+ "epoch": 82.25885558583106,
+ "grad_norm": 1.298797369003296,
+ "learning_rate": 1.6058602267764633e-06,
+ "loss": 0.0366,
+ "step": 30189
+ },
+ {
+ "epoch": 82.26158038147139,
+ "grad_norm": 1.0821623802185059,
+ "learning_rate": 1.6053806301853337e-06,
+ "loss": 0.0112,
+ "step": 30190
+ },
+ {
+ "epoch": 82.26430517711172,
+ "grad_norm": 2.1788785457611084,
+ "learning_rate": 1.6049010989712288e-06,
+ "loss": 0.1758,
+ "step": 30191
+ },
+ {
+ "epoch": 82.26702997275204,
+ "grad_norm": 1.3369942903518677,
+ "learning_rate": 1.6044216331378837e-06,
+ "loss": 0.0537,
+ "step": 30192
+ },
+ {
+ "epoch": 82.26975476839237,
+ "grad_norm": 1.5992311239242554,
+ "learning_rate": 1.603942232689031e-06,
+ "loss": 0.0107,
+ "step": 30193
+ },
+ {
+ "epoch": 82.2724795640327,
+ "grad_norm": 0.8057867288589478,
+ "learning_rate": 1.6034628976284084e-06,
+ "loss": 0.0095,
+ "step": 30194
+ },
+ {
+ "epoch": 82.27520435967303,
+ "grad_norm": 0.8561340570449829,
+ "learning_rate": 1.6029836279597455e-06,
+ "loss": 0.0118,
+ "step": 30195
+ },
+ {
+ "epoch": 82.27792915531334,
+ "grad_norm": 1.128212332725525,
+ "learning_rate": 1.602504423686775e-06,
+ "loss": 0.0159,
+ "step": 30196
+ },
+ {
+ "epoch": 82.28065395095368,
+ "grad_norm": 0.9873716235160828,
+ "learning_rate": 1.6020252848132311e-06,
+ "loss": 0.0089,
+ "step": 30197
+ },
+ {
+ "epoch": 82.28337874659401,
+ "grad_norm": 1.1147462129592896,
+ "learning_rate": 1.6015462113428436e-06,
+ "loss": 0.0144,
+ "step": 30198
+ },
+ {
+ "epoch": 82.28610354223433,
+ "grad_norm": 1.1257060766220093,
+ "learning_rate": 1.601067203279345e-06,
+ "loss": 0.0151,
+ "step": 30199
+ },
+ {
+ "epoch": 82.28882833787466,
+ "grad_norm": 0.8763598203659058,
+ "learning_rate": 1.6005882606264621e-06,
+ "loss": 0.0085,
+ "step": 30200
+ },
+ {
+ "epoch": 82.29155313351498,
+ "grad_norm": 0.8107163310050964,
+ "learning_rate": 1.6001093833879288e-06,
+ "loss": 0.0794,
+ "step": 30201
+ },
+ {
+ "epoch": 82.29427792915531,
+ "grad_norm": 1.7783206701278687,
+ "learning_rate": 1.5996305715674743e-06,
+ "loss": 0.0184,
+ "step": 30202
+ },
+ {
+ "epoch": 82.29700272479565,
+ "grad_norm": 1.1771018505096436,
+ "learning_rate": 1.5991518251688265e-06,
+ "loss": 0.0073,
+ "step": 30203
+ },
+ {
+ "epoch": 82.29972752043597,
+ "grad_norm": 1.0886545181274414,
+ "learning_rate": 1.5986731441957114e-06,
+ "loss": 0.0096,
+ "step": 30204
+ },
+ {
+ "epoch": 82.3024523160763,
+ "grad_norm": 1.4083064794540405,
+ "learning_rate": 1.5981945286518597e-06,
+ "loss": 0.0284,
+ "step": 30205
+ },
+ {
+ "epoch": 82.30517711171662,
+ "grad_norm": 0.7510116100311279,
+ "learning_rate": 1.597715978541e-06,
+ "loss": 0.0212,
+ "step": 30206
+ },
+ {
+ "epoch": 82.30790190735695,
+ "grad_norm": 1.2354168891906738,
+ "learning_rate": 1.597237493866859e-06,
+ "loss": 0.0592,
+ "step": 30207
+ },
+ {
+ "epoch": 82.31062670299727,
+ "grad_norm": 1.007689356803894,
+ "learning_rate": 1.5967590746331619e-06,
+ "loss": 0.0088,
+ "step": 30208
+ },
+ {
+ "epoch": 82.3133514986376,
+ "grad_norm": 0.831365704536438,
+ "learning_rate": 1.5962807208436315e-06,
+ "loss": 0.0093,
+ "step": 30209
+ },
+ {
+ "epoch": 82.31607629427793,
+ "grad_norm": 0.8191099762916565,
+ "learning_rate": 1.5958024325019993e-06,
+ "loss": 0.0107,
+ "step": 30210
+ },
+ {
+ "epoch": 82.31880108991825,
+ "grad_norm": 1.353762149810791,
+ "learning_rate": 1.595324209611987e-06,
+ "loss": 0.0137,
+ "step": 30211
+ },
+ {
+ "epoch": 82.32152588555859,
+ "grad_norm": 1.7847286462783813,
+ "learning_rate": 1.594846052177319e-06,
+ "loss": 0.1699,
+ "step": 30212
+ },
+ {
+ "epoch": 82.3242506811989,
+ "grad_norm": 1.1001160144805908,
+ "learning_rate": 1.594367960201718e-06,
+ "loss": 0.0294,
+ "step": 30213
+ },
+ {
+ "epoch": 82.32697547683924,
+ "grad_norm": 1.4401648044586182,
+ "learning_rate": 1.5938899336889102e-06,
+ "loss": 0.0569,
+ "step": 30214
+ },
+ {
+ "epoch": 82.32970027247957,
+ "grad_norm": 0.7301326990127563,
+ "learning_rate": 1.5934119726426177e-06,
+ "loss": 0.0059,
+ "step": 30215
+ },
+ {
+ "epoch": 82.33242506811989,
+ "grad_norm": 1.1818958520889282,
+ "learning_rate": 1.5929340770665614e-06,
+ "loss": 0.0829,
+ "step": 30216
+ },
+ {
+ "epoch": 82.33514986376022,
+ "grad_norm": 1.1443928480148315,
+ "learning_rate": 1.5924562469644623e-06,
+ "loss": 0.0156,
+ "step": 30217
+ },
+ {
+ "epoch": 82.33787465940054,
+ "grad_norm": 1.4045501947402954,
+ "learning_rate": 1.5919784823400452e-06,
+ "loss": 0.0872,
+ "step": 30218
+ },
+ {
+ "epoch": 82.34059945504087,
+ "grad_norm": 1.3216784000396729,
+ "learning_rate": 1.5915007831970297e-06,
+ "loss": 0.0112,
+ "step": 30219
+ },
+ {
+ "epoch": 82.34332425068119,
+ "grad_norm": 1.6931450366973877,
+ "learning_rate": 1.5910231495391326e-06,
+ "loss": 0.1412,
+ "step": 30220
+ },
+ {
+ "epoch": 82.34604904632153,
+ "grad_norm": 1.4502661228179932,
+ "learning_rate": 1.5905455813700788e-06,
+ "loss": 0.0211,
+ "step": 30221
+ },
+ {
+ "epoch": 82.34877384196186,
+ "grad_norm": 1.5760444402694702,
+ "learning_rate": 1.590068078693584e-06,
+ "loss": 0.0557,
+ "step": 30222
+ },
+ {
+ "epoch": 82.35149863760218,
+ "grad_norm": 1.468763828277588,
+ "learning_rate": 1.58959064151337e-06,
+ "loss": 0.0145,
+ "step": 30223
+ },
+ {
+ "epoch": 82.35422343324251,
+ "grad_norm": 1.1536672115325928,
+ "learning_rate": 1.5891132698331536e-06,
+ "loss": 0.0153,
+ "step": 30224
+ },
+ {
+ "epoch": 82.35694822888283,
+ "grad_norm": 0.8121574521064758,
+ "learning_rate": 1.5886359636566494e-06,
+ "loss": 0.0064,
+ "step": 30225
+ },
+ {
+ "epoch": 82.35967302452316,
+ "grad_norm": 1.0651297569274902,
+ "learning_rate": 1.5881587229875806e-06,
+ "loss": 0.0178,
+ "step": 30226
+ },
+ {
+ "epoch": 82.3623978201635,
+ "grad_norm": 1.471063256263733,
+ "learning_rate": 1.5876815478296602e-06,
+ "loss": 0.0421,
+ "step": 30227
+ },
+ {
+ "epoch": 82.36512261580381,
+ "grad_norm": 1.686028242111206,
+ "learning_rate": 1.5872044381866058e-06,
+ "loss": 0.0066,
+ "step": 30228
+ },
+ {
+ "epoch": 82.36784741144415,
+ "grad_norm": 0.976803183555603,
+ "learning_rate": 1.5867273940621297e-06,
+ "loss": 0.0114,
+ "step": 30229
+ },
+ {
+ "epoch": 82.37057220708446,
+ "grad_norm": 0.7657887935638428,
+ "learning_rate": 1.5862504154599522e-06,
+ "loss": 0.0074,
+ "step": 30230
+ },
+ {
+ "epoch": 82.3732970027248,
+ "grad_norm": 1.0851670503616333,
+ "learning_rate": 1.5857735023837861e-06,
+ "loss": 0.0079,
+ "step": 30231
+ },
+ {
+ "epoch": 82.37602179836512,
+ "grad_norm": 1.2051239013671875,
+ "learning_rate": 1.585296654837345e-06,
+ "loss": 0.0339,
+ "step": 30232
+ },
+ {
+ "epoch": 82.37874659400545,
+ "grad_norm": 2.277061700820923,
+ "learning_rate": 1.5848198728243402e-06,
+ "loss": 0.0079,
+ "step": 30233
+ },
+ {
+ "epoch": 82.38147138964578,
+ "grad_norm": 0.6977469325065613,
+ "learning_rate": 1.5843431563484902e-06,
+ "loss": 0.0064,
+ "step": 30234
+ },
+ {
+ "epoch": 82.3841961852861,
+ "grad_norm": 1.187865972518921,
+ "learning_rate": 1.583866505413505e-06,
+ "loss": 0.0368,
+ "step": 30235
+ },
+ {
+ "epoch": 82.38692098092643,
+ "grad_norm": 0.9003686904907227,
+ "learning_rate": 1.5833899200230963e-06,
+ "loss": 0.0136,
+ "step": 30236
+ },
+ {
+ "epoch": 82.38964577656675,
+ "grad_norm": 1.8803967237472534,
+ "learning_rate": 1.582913400180973e-06,
+ "loss": 0.0603,
+ "step": 30237
+ },
+ {
+ "epoch": 82.39237057220708,
+ "grad_norm": 2.471881628036499,
+ "learning_rate": 1.5824369458908518e-06,
+ "loss": 0.0563,
+ "step": 30238
+ },
+ {
+ "epoch": 82.39509536784742,
+ "grad_norm": 0.8147236108779907,
+ "learning_rate": 1.581960557156441e-06,
+ "loss": 0.0116,
+ "step": 30239
+ },
+ {
+ "epoch": 82.39782016348774,
+ "grad_norm": 1.1764111518859863,
+ "learning_rate": 1.5814842339814485e-06,
+ "loss": 0.0415,
+ "step": 30240
+ },
+ {
+ "epoch": 82.40054495912807,
+ "grad_norm": 0.35829365253448486,
+ "learning_rate": 1.5810079763695852e-06,
+ "loss": 0.003,
+ "step": 30241
+ },
+ {
+ "epoch": 82.40326975476839,
+ "grad_norm": 0.5702731609344482,
+ "learning_rate": 1.5805317843245638e-06,
+ "loss": 0.0042,
+ "step": 30242
+ },
+ {
+ "epoch": 82.40599455040872,
+ "grad_norm": 1.0712209939956665,
+ "learning_rate": 1.5800556578500891e-06,
+ "loss": 0.0103,
+ "step": 30243
+ },
+ {
+ "epoch": 82.40871934604904,
+ "grad_norm": 1.7824128866195679,
+ "learning_rate": 1.5795795969498706e-06,
+ "loss": 0.0266,
+ "step": 30244
+ },
+ {
+ "epoch": 82.41144414168937,
+ "grad_norm": 1.3930704593658447,
+ "learning_rate": 1.579103601627613e-06,
+ "loss": 0.0945,
+ "step": 30245
+ },
+ {
+ "epoch": 82.4141689373297,
+ "grad_norm": 1.1186679601669312,
+ "learning_rate": 1.5786276718870274e-06,
+ "loss": 0.0161,
+ "step": 30246
+ },
+ {
+ "epoch": 82.41689373297002,
+ "grad_norm": 1.6332776546478271,
+ "learning_rate": 1.578151807731818e-06,
+ "loss": 0.0163,
+ "step": 30247
+ },
+ {
+ "epoch": 82.41961852861036,
+ "grad_norm": 1.6748617887496948,
+ "learning_rate": 1.5776760091656918e-06,
+ "loss": 0.0434,
+ "step": 30248
+ },
+ {
+ "epoch": 82.42234332425068,
+ "grad_norm": 0.8203908205032349,
+ "learning_rate": 1.577200276192351e-06,
+ "loss": 0.0095,
+ "step": 30249
+ },
+ {
+ "epoch": 82.42506811989101,
+ "grad_norm": 1.3263330459594727,
+ "learning_rate": 1.5767246088155064e-06,
+ "loss": 0.0346,
+ "step": 30250
+ },
+ {
+ "epoch": 82.42779291553134,
+ "grad_norm": 0.9073047637939453,
+ "learning_rate": 1.5762490070388592e-06,
+ "loss": 0.0129,
+ "step": 30251
+ },
+ {
+ "epoch": 82.43051771117166,
+ "grad_norm": 0.6081103086471558,
+ "learning_rate": 1.5757734708661144e-06,
+ "loss": 0.0098,
+ "step": 30252
+ },
+ {
+ "epoch": 82.433242506812,
+ "grad_norm": 0.7481346130371094,
+ "learning_rate": 1.575298000300971e-06,
+ "loss": 0.0088,
+ "step": 30253
+ },
+ {
+ "epoch": 82.43596730245231,
+ "grad_norm": 1.2430821657180786,
+ "learning_rate": 1.5748225953471385e-06,
+ "loss": 0.04,
+ "step": 30254
+ },
+ {
+ "epoch": 82.43869209809264,
+ "grad_norm": 1.408116102218628,
+ "learning_rate": 1.5743472560083162e-06,
+ "loss": 0.0829,
+ "step": 30255
+ },
+ {
+ "epoch": 82.44141689373296,
+ "grad_norm": 1.315536618232727,
+ "learning_rate": 1.5738719822882064e-06,
+ "loss": 0.0527,
+ "step": 30256
+ },
+ {
+ "epoch": 82.4441416893733,
+ "grad_norm": 1.3409425020217896,
+ "learning_rate": 1.5733967741905087e-06,
+ "loss": 0.0791,
+ "step": 30257
+ },
+ {
+ "epoch": 82.44686648501363,
+ "grad_norm": 1.185915470123291,
+ "learning_rate": 1.5729216317189245e-06,
+ "loss": 0.0303,
+ "step": 30258
+ },
+ {
+ "epoch": 82.44959128065395,
+ "grad_norm": 1.3899260759353638,
+ "learning_rate": 1.5724465548771584e-06,
+ "loss": 0.0225,
+ "step": 30259
+ },
+ {
+ "epoch": 82.45231607629428,
+ "grad_norm": 0.8174042701721191,
+ "learning_rate": 1.5719715436689076e-06,
+ "loss": 0.006,
+ "step": 30260
+ },
+ {
+ "epoch": 82.4550408719346,
+ "grad_norm": 1.255523681640625,
+ "learning_rate": 1.5714965980978692e-06,
+ "loss": 0.044,
+ "step": 30261
+ },
+ {
+ "epoch": 82.45776566757493,
+ "grad_norm": 0.7740725874900818,
+ "learning_rate": 1.5710217181677456e-06,
+ "loss": 0.0057,
+ "step": 30262
+ },
+ {
+ "epoch": 82.46049046321527,
+ "grad_norm": 1.1154687404632568,
+ "learning_rate": 1.570546903882233e-06,
+ "loss": 0.0118,
+ "step": 30263
+ },
+ {
+ "epoch": 82.46321525885558,
+ "grad_norm": 1.4423139095306396,
+ "learning_rate": 1.5700721552450315e-06,
+ "loss": 0.019,
+ "step": 30264
+ },
+ {
+ "epoch": 82.46594005449592,
+ "grad_norm": 1.1747256517410278,
+ "learning_rate": 1.5695974722598328e-06,
+ "loss": 0.0097,
+ "step": 30265
+ },
+ {
+ "epoch": 82.46866485013624,
+ "grad_norm": 1.01933753490448,
+ "learning_rate": 1.5691228549303406e-06,
+ "loss": 0.0146,
+ "step": 30266
+ },
+ {
+ "epoch": 82.47138964577657,
+ "grad_norm": 0.7103509902954102,
+ "learning_rate": 1.5686483032602484e-06,
+ "loss": 0.0075,
+ "step": 30267
+ },
+ {
+ "epoch": 82.47411444141689,
+ "grad_norm": 0.9798891544342041,
+ "learning_rate": 1.5681738172532513e-06,
+ "loss": 0.0103,
+ "step": 30268
+ },
+ {
+ "epoch": 82.47683923705722,
+ "grad_norm": 0.7145289182662964,
+ "learning_rate": 1.5676993969130439e-06,
+ "loss": 0.0054,
+ "step": 30269
+ },
+ {
+ "epoch": 82.47956403269755,
+ "grad_norm": 1.0392165184020996,
+ "learning_rate": 1.5672250422433233e-06,
+ "loss": 0.0136,
+ "step": 30270
+ },
+ {
+ "epoch": 82.48228882833787,
+ "grad_norm": 0.6566623449325562,
+ "learning_rate": 1.5667507532477833e-06,
+ "loss": 0.006,
+ "step": 30271
+ },
+ {
+ "epoch": 82.4850136239782,
+ "grad_norm": 1.1566523313522339,
+ "learning_rate": 1.5662765299301174e-06,
+ "loss": 0.0216,
+ "step": 30272
+ },
+ {
+ "epoch": 82.48773841961852,
+ "grad_norm": 1.8386319875717163,
+ "learning_rate": 1.5658023722940151e-06,
+ "loss": 0.1067,
+ "step": 30273
+ },
+ {
+ "epoch": 82.49046321525886,
+ "grad_norm": 1.2908538579940796,
+ "learning_rate": 1.5653282803431734e-06,
+ "loss": 0.014,
+ "step": 30274
+ },
+ {
+ "epoch": 82.49318801089919,
+ "grad_norm": 1.0049371719360352,
+ "learning_rate": 1.564854254081285e-06,
+ "loss": 0.0157,
+ "step": 30275
+ },
+ {
+ "epoch": 82.49591280653951,
+ "grad_norm": 0.6827918887138367,
+ "learning_rate": 1.5643802935120411e-06,
+ "loss": 0.0068,
+ "step": 30276
+ },
+ {
+ "epoch": 82.49863760217984,
+ "grad_norm": 1.9792038202285767,
+ "learning_rate": 1.5639063986391322e-06,
+ "loss": 0.0959,
+ "step": 30277
+ },
+ {
+ "epoch": 82.50136239782016,
+ "grad_norm": 1.0599373579025269,
+ "learning_rate": 1.5634325694662455e-06,
+ "loss": 0.0115,
+ "step": 30278
+ },
+ {
+ "epoch": 82.50408719346049,
+ "grad_norm": 1.0010462999343872,
+ "learning_rate": 1.5629588059970769e-06,
+ "loss": 0.0122,
+ "step": 30279
+ },
+ {
+ "epoch": 82.50681198910081,
+ "grad_norm": 1.0634328126907349,
+ "learning_rate": 1.5624851082353143e-06,
+ "loss": 0.0846,
+ "step": 30280
+ },
+ {
+ "epoch": 82.50953678474114,
+ "grad_norm": 0.9855024218559265,
+ "learning_rate": 1.562011476184645e-06,
+ "loss": 0.0094,
+ "step": 30281
+ },
+ {
+ "epoch": 82.51226158038148,
+ "grad_norm": 0.9271959662437439,
+ "learning_rate": 1.561537909848757e-06,
+ "loss": 0.0099,
+ "step": 30282
+ },
+ {
+ "epoch": 82.5149863760218,
+ "grad_norm": 1.2329994440078735,
+ "learning_rate": 1.561064409231342e-06,
+ "loss": 0.0177,
+ "step": 30283
+ },
+ {
+ "epoch": 82.51771117166213,
+ "grad_norm": 1.4154409170150757,
+ "learning_rate": 1.560590974336087e-06,
+ "loss": 0.0134,
+ "step": 30284
+ },
+ {
+ "epoch": 82.52043596730245,
+ "grad_norm": 0.9274729490280151,
+ "learning_rate": 1.5601176051666744e-06,
+ "loss": 0.0107,
+ "step": 30285
+ },
+ {
+ "epoch": 82.52316076294278,
+ "grad_norm": 1.243600845336914,
+ "learning_rate": 1.559644301726797e-06,
+ "loss": 0.0127,
+ "step": 30286
+ },
+ {
+ "epoch": 82.52588555858311,
+ "grad_norm": 0.612587034702301,
+ "learning_rate": 1.5591710640201385e-06,
+ "loss": 0.006,
+ "step": 30287
+ },
+ {
+ "epoch": 82.52861035422343,
+ "grad_norm": 2.815392255783081,
+ "learning_rate": 1.5586978920503836e-06,
+ "loss": 0.0121,
+ "step": 30288
+ },
+ {
+ "epoch": 82.53133514986376,
+ "grad_norm": 0.8032522797584534,
+ "learning_rate": 1.5582247858212152e-06,
+ "loss": 0.0081,
+ "step": 30289
+ },
+ {
+ "epoch": 82.53405994550408,
+ "grad_norm": 0.8341848254203796,
+ "learning_rate": 1.5577517453363234e-06,
+ "loss": 0.0087,
+ "step": 30290
+ },
+ {
+ "epoch": 82.53678474114442,
+ "grad_norm": 1.327301263809204,
+ "learning_rate": 1.5572787705993863e-06,
+ "loss": 0.0112,
+ "step": 30291
+ },
+ {
+ "epoch": 82.53950953678473,
+ "grad_norm": 1.1023961305618286,
+ "learning_rate": 1.5568058616140934e-06,
+ "loss": 0.0078,
+ "step": 30292
+ },
+ {
+ "epoch": 82.54223433242507,
+ "grad_norm": 0.8884566426277161,
+ "learning_rate": 1.5563330183841252e-06,
+ "loss": 0.0149,
+ "step": 30293
+ },
+ {
+ "epoch": 82.5449591280654,
+ "grad_norm": 1.339106559753418,
+ "learning_rate": 1.5558602409131606e-06,
+ "loss": 0.0227,
+ "step": 30294
+ },
+ {
+ "epoch": 82.54768392370572,
+ "grad_norm": 1.5092456340789795,
+ "learning_rate": 1.555387529204887e-06,
+ "loss": 0.0196,
+ "step": 30295
+ },
+ {
+ "epoch": 82.55040871934605,
+ "grad_norm": 1.3302956819534302,
+ "learning_rate": 1.5549148832629834e-06,
+ "loss": 0.074,
+ "step": 30296
+ },
+ {
+ "epoch": 82.55313351498637,
+ "grad_norm": 1.6532145738601685,
+ "learning_rate": 1.5544423030911316e-06,
+ "loss": 0.011,
+ "step": 30297
+ },
+ {
+ "epoch": 82.5558583106267,
+ "grad_norm": 33.550086975097656,
+ "learning_rate": 1.5539697886930082e-06,
+ "loss": 0.0221,
+ "step": 30298
+ },
+ {
+ "epoch": 82.55858310626704,
+ "grad_norm": 0.74742591381073,
+ "learning_rate": 1.5534973400722986e-06,
+ "loss": 0.0063,
+ "step": 30299
+ },
+ {
+ "epoch": 82.56130790190736,
+ "grad_norm": 0.9917572140693665,
+ "learning_rate": 1.5530249572326806e-06,
+ "loss": 0.008,
+ "step": 30300
+ },
+ {
+ "epoch": 82.56403269754769,
+ "grad_norm": 1.978385329246521,
+ "learning_rate": 1.5525526401778313e-06,
+ "loss": 0.0306,
+ "step": 30301
+ },
+ {
+ "epoch": 82.566757493188,
+ "grad_norm": 2.1984241008758545,
+ "learning_rate": 1.5520803889114288e-06,
+ "loss": 0.1351,
+ "step": 30302
+ },
+ {
+ "epoch": 82.56948228882834,
+ "grad_norm": 1.4542068243026733,
+ "learning_rate": 1.5516082034371538e-06,
+ "loss": 0.0515,
+ "step": 30303
+ },
+ {
+ "epoch": 82.57220708446866,
+ "grad_norm": 2.218402624130249,
+ "learning_rate": 1.5511360837586832e-06,
+ "loss": 0.1194,
+ "step": 30304
+ },
+ {
+ "epoch": 82.57493188010899,
+ "grad_norm": 18.61517906188965,
+ "learning_rate": 1.5506640298796927e-06,
+ "loss": 0.0233,
+ "step": 30305
+ },
+ {
+ "epoch": 82.57765667574932,
+ "grad_norm": 0.8993164896965027,
+ "learning_rate": 1.5501920418038553e-06,
+ "loss": 0.0099,
+ "step": 30306
+ },
+ {
+ "epoch": 82.58038147138964,
+ "grad_norm": 1.231819987297058,
+ "learning_rate": 1.5497201195348532e-06,
+ "loss": 0.0238,
+ "step": 30307
+ },
+ {
+ "epoch": 82.58310626702998,
+ "grad_norm": 0.8344288468360901,
+ "learning_rate": 1.5492482630763594e-06,
+ "loss": 0.009,
+ "step": 30308
+ },
+ {
+ "epoch": 82.5858310626703,
+ "grad_norm": 1.3531464338302612,
+ "learning_rate": 1.5487764724320453e-06,
+ "loss": 0.0235,
+ "step": 30309
+ },
+ {
+ "epoch": 82.58855585831063,
+ "grad_norm": 1.5282865762710571,
+ "learning_rate": 1.548304747605589e-06,
+ "loss": 0.0217,
+ "step": 30310
+ },
+ {
+ "epoch": 82.59128065395096,
+ "grad_norm": 0.6697331666946411,
+ "learning_rate": 1.5478330886006643e-06,
+ "loss": 0.0069,
+ "step": 30311
+ },
+ {
+ "epoch": 82.59400544959128,
+ "grad_norm": 1.315079927444458,
+ "learning_rate": 1.5473614954209449e-06,
+ "loss": 0.0087,
+ "step": 30312
+ },
+ {
+ "epoch": 82.59673024523161,
+ "grad_norm": 1.5574427843093872,
+ "learning_rate": 1.546889968070101e-06,
+ "loss": 0.071,
+ "step": 30313
+ },
+ {
+ "epoch": 82.59945504087193,
+ "grad_norm": 0.9694766402244568,
+ "learning_rate": 1.5464185065518034e-06,
+ "loss": 0.0125,
+ "step": 30314
+ },
+ {
+ "epoch": 82.60217983651226,
+ "grad_norm": 1.336678385734558,
+ "learning_rate": 1.54594711086973e-06,
+ "loss": 0.0296,
+ "step": 30315
+ },
+ {
+ "epoch": 82.60490463215258,
+ "grad_norm": 2.154788017272949,
+ "learning_rate": 1.5454757810275468e-06,
+ "loss": 0.0142,
+ "step": 30316
+ },
+ {
+ "epoch": 82.60762942779292,
+ "grad_norm": 1.0403449535369873,
+ "learning_rate": 1.5450045170289275e-06,
+ "loss": 0.0763,
+ "step": 30317
+ },
+ {
+ "epoch": 82.61035422343325,
+ "grad_norm": 1.9575085639953613,
+ "learning_rate": 1.544533318877538e-06,
+ "loss": 0.0725,
+ "step": 30318
+ },
+ {
+ "epoch": 82.61307901907357,
+ "grad_norm": 0.9285496473312378,
+ "learning_rate": 1.5440621865770533e-06,
+ "loss": 0.0088,
+ "step": 30319
+ },
+ {
+ "epoch": 82.6158038147139,
+ "grad_norm": 1.2453596591949463,
+ "learning_rate": 1.5435911201311405e-06,
+ "loss": 0.0162,
+ "step": 30320
+ },
+ {
+ "epoch": 82.61852861035422,
+ "grad_norm": 0.8245920538902283,
+ "learning_rate": 1.5431201195434664e-06,
+ "loss": 0.0143,
+ "step": 30321
+ },
+ {
+ "epoch": 82.62125340599455,
+ "grad_norm": 1.018281102180481,
+ "learning_rate": 1.5426491848176995e-06,
+ "loss": 0.0842,
+ "step": 30322
+ },
+ {
+ "epoch": 82.62397820163488,
+ "grad_norm": 1.3900829553604126,
+ "learning_rate": 1.542178315957511e-06,
+ "loss": 0.0277,
+ "step": 30323
+ },
+ {
+ "epoch": 82.6267029972752,
+ "grad_norm": 0.8926966786384583,
+ "learning_rate": 1.541707512966566e-06,
+ "loss": 0.0074,
+ "step": 30324
+ },
+ {
+ "epoch": 82.62942779291554,
+ "grad_norm": 0.9605621099472046,
+ "learning_rate": 1.541236775848529e-06,
+ "loss": 0.1181,
+ "step": 30325
+ },
+ {
+ "epoch": 82.63215258855585,
+ "grad_norm": 1.0315896272659302,
+ "learning_rate": 1.5407661046070643e-06,
+ "loss": 0.0125,
+ "step": 30326
+ },
+ {
+ "epoch": 82.63487738419619,
+ "grad_norm": 0.5422201752662659,
+ "learning_rate": 1.5402954992458451e-06,
+ "loss": 0.0058,
+ "step": 30327
+ },
+ {
+ "epoch": 82.6376021798365,
+ "grad_norm": 1.6920164823532104,
+ "learning_rate": 1.5398249597685334e-06,
+ "loss": 0.2,
+ "step": 30328
+ },
+ {
+ "epoch": 82.64032697547684,
+ "grad_norm": 1.3774725198745728,
+ "learning_rate": 1.5393544861787912e-06,
+ "loss": 0.019,
+ "step": 30329
+ },
+ {
+ "epoch": 82.64305177111717,
+ "grad_norm": 1.0956847667694092,
+ "learning_rate": 1.5388840784802817e-06,
+ "loss": 0.0211,
+ "step": 30330
+ },
+ {
+ "epoch": 82.64577656675749,
+ "grad_norm": 1.75157630443573,
+ "learning_rate": 1.5384137366766739e-06,
+ "loss": 0.0125,
+ "step": 30331
+ },
+ {
+ "epoch": 82.64850136239782,
+ "grad_norm": 1.2931731939315796,
+ "learning_rate": 1.5379434607716271e-06,
+ "loss": 0.0267,
+ "step": 30332
+ },
+ {
+ "epoch": 82.65122615803814,
+ "grad_norm": 1.1469416618347168,
+ "learning_rate": 1.537473250768804e-06,
+ "loss": 0.0125,
+ "step": 30333
+ },
+ {
+ "epoch": 82.65395095367847,
+ "grad_norm": 0.4095413386821747,
+ "learning_rate": 1.537003106671865e-06,
+ "loss": 0.0041,
+ "step": 30334
+ },
+ {
+ "epoch": 82.65667574931881,
+ "grad_norm": 1.3085955381393433,
+ "learning_rate": 1.536533028484476e-06,
+ "loss": 0.0448,
+ "step": 30335
+ },
+ {
+ "epoch": 82.65940054495913,
+ "grad_norm": 0.8682668209075928,
+ "learning_rate": 1.5360630162102953e-06,
+ "loss": 0.0111,
+ "step": 30336
+ },
+ {
+ "epoch": 82.66212534059946,
+ "grad_norm": 1.0296149253845215,
+ "learning_rate": 1.535593069852983e-06,
+ "loss": 0.0066,
+ "step": 30337
+ },
+ {
+ "epoch": 82.66485013623978,
+ "grad_norm": 0.8641826510429382,
+ "learning_rate": 1.5351231894161967e-06,
+ "loss": 0.0142,
+ "step": 30338
+ },
+ {
+ "epoch": 82.66757493188011,
+ "grad_norm": 1.5388976335525513,
+ "learning_rate": 1.534653374903602e-06,
+ "loss": 0.054,
+ "step": 30339
+ },
+ {
+ "epoch": 82.67029972752043,
+ "grad_norm": 0.7611644864082336,
+ "learning_rate": 1.5341836263188537e-06,
+ "loss": 0.0087,
+ "step": 30340
+ },
+ {
+ "epoch": 82.67302452316076,
+ "grad_norm": 1.6213856935501099,
+ "learning_rate": 1.53371394366561e-06,
+ "loss": 0.0219,
+ "step": 30341
+ },
+ {
+ "epoch": 82.6757493188011,
+ "grad_norm": 1.6586408615112305,
+ "learning_rate": 1.5332443269475284e-06,
+ "loss": 0.0833,
+ "step": 30342
+ },
+ {
+ "epoch": 82.67847411444141,
+ "grad_norm": 1.0605303049087524,
+ "learning_rate": 1.5327747761682677e-06,
+ "loss": 0.1122,
+ "step": 30343
+ },
+ {
+ "epoch": 82.68119891008175,
+ "grad_norm": 0.9497774243354797,
+ "learning_rate": 1.5323052913314863e-06,
+ "loss": 0.0109,
+ "step": 30344
+ },
+ {
+ "epoch": 82.68392370572207,
+ "grad_norm": 1.3344265222549438,
+ "learning_rate": 1.531835872440839e-06,
+ "loss": 0.0387,
+ "step": 30345
+ },
+ {
+ "epoch": 82.6866485013624,
+ "grad_norm": 1.0698401927947998,
+ "learning_rate": 1.5313665194999816e-06,
+ "loss": 0.0166,
+ "step": 30346
+ },
+ {
+ "epoch": 82.68937329700273,
+ "grad_norm": 1.8302791118621826,
+ "learning_rate": 1.530897232512566e-06,
+ "loss": 0.0128,
+ "step": 30347
+ },
+ {
+ "epoch": 82.69209809264305,
+ "grad_norm": 1.1016271114349365,
+ "learning_rate": 1.5304280114822534e-06,
+ "loss": 0.0111,
+ "step": 30348
+ },
+ {
+ "epoch": 82.69482288828338,
+ "grad_norm": 1.1083097457885742,
+ "learning_rate": 1.5299588564126944e-06,
+ "loss": 0.0072,
+ "step": 30349
+ },
+ {
+ "epoch": 82.6975476839237,
+ "grad_norm": 1.4023267030715942,
+ "learning_rate": 1.5294897673075405e-06,
+ "loss": 0.0112,
+ "step": 30350
+ },
+ {
+ "epoch": 82.70027247956403,
+ "grad_norm": 0.8714505434036255,
+ "learning_rate": 1.5290207441704507e-06,
+ "loss": 0.0122,
+ "step": 30351
+ },
+ {
+ "epoch": 82.70299727520435,
+ "grad_norm": 3.0341644287109375,
+ "learning_rate": 1.5285517870050748e-06,
+ "loss": 0.0995,
+ "step": 30352
+ },
+ {
+ "epoch": 82.70572207084469,
+ "grad_norm": 1.322498083114624,
+ "learning_rate": 1.528082895815065e-06,
+ "loss": 0.0346,
+ "step": 30353
+ },
+ {
+ "epoch": 82.70844686648502,
+ "grad_norm": 1.1280895471572876,
+ "learning_rate": 1.5276140706040698e-06,
+ "loss": 0.0157,
+ "step": 30354
+ },
+ {
+ "epoch": 82.71117166212534,
+ "grad_norm": 0.6083174347877502,
+ "learning_rate": 1.5271453113757462e-06,
+ "loss": 0.0068,
+ "step": 30355
+ },
+ {
+ "epoch": 82.71389645776567,
+ "grad_norm": 1.3163414001464844,
+ "learning_rate": 1.5266766181337423e-06,
+ "loss": 0.0168,
+ "step": 30356
+ },
+ {
+ "epoch": 82.71662125340599,
+ "grad_norm": 0.813116729259491,
+ "learning_rate": 1.5262079908817084e-06,
+ "loss": 0.0091,
+ "step": 30357
+ },
+ {
+ "epoch": 82.71934604904632,
+ "grad_norm": 1.0755376815795898,
+ "learning_rate": 1.5257394296232907e-06,
+ "loss": 0.0411,
+ "step": 30358
+ },
+ {
+ "epoch": 82.72207084468666,
+ "grad_norm": 0.5506759881973267,
+ "learning_rate": 1.5252709343621452e-06,
+ "loss": 0.0065,
+ "step": 30359
+ },
+ {
+ "epoch": 82.72479564032697,
+ "grad_norm": 1.6442257165908813,
+ "learning_rate": 1.5248025051019145e-06,
+ "loss": 0.0295,
+ "step": 30360
+ },
+ {
+ "epoch": 82.7275204359673,
+ "grad_norm": 1.3700740337371826,
+ "learning_rate": 1.524334141846251e-06,
+ "loss": 0.0263,
+ "step": 30361
+ },
+ {
+ "epoch": 82.73024523160763,
+ "grad_norm": 1.6617544889450073,
+ "learning_rate": 1.5238658445987997e-06,
+ "loss": 0.0155,
+ "step": 30362
+ },
+ {
+ "epoch": 82.73297002724796,
+ "grad_norm": 1.6179925203323364,
+ "learning_rate": 1.5233976133632067e-06,
+ "loss": 0.0548,
+ "step": 30363
+ },
+ {
+ "epoch": 82.73569482288828,
+ "grad_norm": 1.1714732646942139,
+ "learning_rate": 1.5229294481431233e-06,
+ "loss": 0.0414,
+ "step": 30364
+ },
+ {
+ "epoch": 82.73841961852861,
+ "grad_norm": 1.1730802059173584,
+ "learning_rate": 1.5224613489421914e-06,
+ "loss": 0.0104,
+ "step": 30365
+ },
+ {
+ "epoch": 82.74114441416894,
+ "grad_norm": 2.412360429763794,
+ "learning_rate": 1.5219933157640576e-06,
+ "loss": 0.029,
+ "step": 30366
+ },
+ {
+ "epoch": 82.74386920980926,
+ "grad_norm": 1.9855279922485352,
+ "learning_rate": 1.521525348612366e-06,
+ "loss": 0.082,
+ "step": 30367
+ },
+ {
+ "epoch": 82.7465940054496,
+ "grad_norm": 1.5001089572906494,
+ "learning_rate": 1.5210574474907635e-06,
+ "loss": 0.0312,
+ "step": 30368
+ },
+ {
+ "epoch": 82.74931880108991,
+ "grad_norm": 1.6758549213409424,
+ "learning_rate": 1.520589612402893e-06,
+ "loss": 0.0097,
+ "step": 30369
+ },
+ {
+ "epoch": 82.75204359673025,
+ "grad_norm": 1.5167903900146484,
+ "learning_rate": 1.5201218433523945e-06,
+ "loss": 0.0125,
+ "step": 30370
+ },
+ {
+ "epoch": 82.75476839237058,
+ "grad_norm": 0.7610389590263367,
+ "learning_rate": 1.5196541403429176e-06,
+ "loss": 0.0083,
+ "step": 30371
+ },
+ {
+ "epoch": 82.7574931880109,
+ "grad_norm": 1.7201954126358032,
+ "learning_rate": 1.5191865033781005e-06,
+ "loss": 0.0253,
+ "step": 30372
+ },
+ {
+ "epoch": 82.76021798365123,
+ "grad_norm": 1.159612774848938,
+ "learning_rate": 1.518718932461587e-06,
+ "loss": 0.0337,
+ "step": 30373
+ },
+ {
+ "epoch": 82.76294277929155,
+ "grad_norm": 1.8809396028518677,
+ "learning_rate": 1.518251427597014e-06,
+ "loss": 0.0519,
+ "step": 30374
+ },
+ {
+ "epoch": 82.76566757493188,
+ "grad_norm": 0.8215601444244385,
+ "learning_rate": 1.5177839887880287e-06,
+ "loss": 0.0105,
+ "step": 30375
+ },
+ {
+ "epoch": 82.7683923705722,
+ "grad_norm": 0.7616230845451355,
+ "learning_rate": 1.5173166160382692e-06,
+ "loss": 0.0082,
+ "step": 30376
+ },
+ {
+ "epoch": 82.77111716621253,
+ "grad_norm": 1.4525693655014038,
+ "learning_rate": 1.5168493093513714e-06,
+ "loss": 0.0184,
+ "step": 30377
+ },
+ {
+ "epoch": 82.77384196185287,
+ "grad_norm": 1.4212034940719604,
+ "learning_rate": 1.5163820687309816e-06,
+ "loss": 0.0187,
+ "step": 30378
+ },
+ {
+ "epoch": 82.77656675749319,
+ "grad_norm": 0.7611475586891174,
+ "learning_rate": 1.5159148941807323e-06,
+ "loss": 0.0112,
+ "step": 30379
+ },
+ {
+ "epoch": 82.77929155313352,
+ "grad_norm": 1.336260199546814,
+ "learning_rate": 1.5154477857042672e-06,
+ "loss": 0.0116,
+ "step": 30380
+ },
+ {
+ "epoch": 82.78201634877384,
+ "grad_norm": 0.8926466107368469,
+ "learning_rate": 1.5149807433052222e-06,
+ "loss": 0.1321,
+ "step": 30381
+ },
+ {
+ "epoch": 82.78474114441417,
+ "grad_norm": 0.5892461538314819,
+ "learning_rate": 1.5145137669872334e-06,
+ "loss": 0.0055,
+ "step": 30382
+ },
+ {
+ "epoch": 82.7874659400545,
+ "grad_norm": 1.7304357290267944,
+ "learning_rate": 1.5140468567539357e-06,
+ "loss": 0.0451,
+ "step": 30383
+ },
+ {
+ "epoch": 82.79019073569482,
+ "grad_norm": 0.9694221019744873,
+ "learning_rate": 1.5135800126089707e-06,
+ "loss": 0.008,
+ "step": 30384
+ },
+ {
+ "epoch": 82.79291553133515,
+ "grad_norm": 0.8297336101531982,
+ "learning_rate": 1.5131132345559718e-06,
+ "loss": 0.0086,
+ "step": 30385
+ },
+ {
+ "epoch": 82.79564032697547,
+ "grad_norm": 1.791976809501648,
+ "learning_rate": 1.5126465225985731e-06,
+ "loss": 0.0251,
+ "step": 30386
+ },
+ {
+ "epoch": 82.7983651226158,
+ "grad_norm": 0.6654614806175232,
+ "learning_rate": 1.5121798767404084e-06,
+ "loss": 0.0063,
+ "step": 30387
+ },
+ {
+ "epoch": 82.80108991825612,
+ "grad_norm": 0.8295830488204956,
+ "learning_rate": 1.5117132969851156e-06,
+ "loss": 0.0138,
+ "step": 30388
+ },
+ {
+ "epoch": 82.80381471389646,
+ "grad_norm": 1.759433388710022,
+ "learning_rate": 1.5112467833363265e-06,
+ "loss": 0.031,
+ "step": 30389
+ },
+ {
+ "epoch": 82.80653950953679,
+ "grad_norm": 8.696706771850586,
+ "learning_rate": 1.5107803357976748e-06,
+ "loss": 0.0604,
+ "step": 30390
+ },
+ {
+ "epoch": 82.80926430517711,
+ "grad_norm": 0.5397960543632507,
+ "learning_rate": 1.5103139543727895e-06,
+ "loss": 0.0063,
+ "step": 30391
+ },
+ {
+ "epoch": 82.81198910081744,
+ "grad_norm": 1.0598856210708618,
+ "learning_rate": 1.5098476390653082e-06,
+ "loss": 0.0335,
+ "step": 30392
+ },
+ {
+ "epoch": 82.81471389645776,
+ "grad_norm": 1.0988715887069702,
+ "learning_rate": 1.509381389878861e-06,
+ "loss": 0.0112,
+ "step": 30393
+ },
+ {
+ "epoch": 82.8174386920981,
+ "grad_norm": 0.9154739379882812,
+ "learning_rate": 1.5089152068170755e-06,
+ "loss": 0.0096,
+ "step": 30394
+ },
+ {
+ "epoch": 82.82016348773843,
+ "grad_norm": 0.9101088643074036,
+ "learning_rate": 1.5084490898835857e-06,
+ "loss": 0.0121,
+ "step": 30395
+ },
+ {
+ "epoch": 82.82288828337875,
+ "grad_norm": 1.7229012250900269,
+ "learning_rate": 1.5079830390820226e-06,
+ "loss": 0.0097,
+ "step": 30396
+ },
+ {
+ "epoch": 82.82561307901908,
+ "grad_norm": 1.2907911539077759,
+ "learning_rate": 1.5075170544160145e-06,
+ "loss": 0.0183,
+ "step": 30397
+ },
+ {
+ "epoch": 82.8283378746594,
+ "grad_norm": 0.7992650866508484,
+ "learning_rate": 1.5070511358891904e-06,
+ "loss": 0.0053,
+ "step": 30398
+ },
+ {
+ "epoch": 82.83106267029973,
+ "grad_norm": 0.8763000965118408,
+ "learning_rate": 1.5065852835051764e-06,
+ "loss": 0.0084,
+ "step": 30399
+ },
+ {
+ "epoch": 82.83378746594005,
+ "grad_norm": 1.9122391939163208,
+ "learning_rate": 1.5061194972676041e-06,
+ "loss": 0.0331,
+ "step": 30400
+ },
+ {
+ "epoch": 82.83651226158038,
+ "grad_norm": 0.9131175875663757,
+ "learning_rate": 1.5056537771801016e-06,
+ "loss": 0.0127,
+ "step": 30401
+ },
+ {
+ "epoch": 82.83923705722071,
+ "grad_norm": 1.0901801586151123,
+ "learning_rate": 1.5051881232462928e-06,
+ "loss": 0.0126,
+ "step": 30402
+ },
+ {
+ "epoch": 82.84196185286103,
+ "grad_norm": 1.7211683988571167,
+ "learning_rate": 1.5047225354698037e-06,
+ "loss": 0.0157,
+ "step": 30403
+ },
+ {
+ "epoch": 82.84468664850137,
+ "grad_norm": 0.616275429725647,
+ "learning_rate": 1.5042570138542645e-06,
+ "loss": 0.0064,
+ "step": 30404
+ },
+ {
+ "epoch": 82.84741144414168,
+ "grad_norm": 1.2513952255249023,
+ "learning_rate": 1.503791558403298e-06,
+ "loss": 0.0143,
+ "step": 30405
+ },
+ {
+ "epoch": 82.85013623978202,
+ "grad_norm": 2.0991077423095703,
+ "learning_rate": 1.5033261691205292e-06,
+ "loss": 0.1641,
+ "step": 30406
+ },
+ {
+ "epoch": 82.85286103542235,
+ "grad_norm": 1.39664626121521,
+ "learning_rate": 1.5028608460095807e-06,
+ "loss": 0.0119,
+ "step": 30407
+ },
+ {
+ "epoch": 82.85558583106267,
+ "grad_norm": 0.8301781415939331,
+ "learning_rate": 1.5023955890740804e-06,
+ "loss": 0.0107,
+ "step": 30408
+ },
+ {
+ "epoch": 82.858310626703,
+ "grad_norm": 1.1213196516036987,
+ "learning_rate": 1.5019303983176504e-06,
+ "loss": 0.0104,
+ "step": 30409
+ },
+ {
+ "epoch": 82.86103542234332,
+ "grad_norm": 2.6810450553894043,
+ "learning_rate": 1.501465273743913e-06,
+ "loss": 0.012,
+ "step": 30410
+ },
+ {
+ "epoch": 82.86376021798365,
+ "grad_norm": 1.806121826171875,
+ "learning_rate": 1.5010002153564873e-06,
+ "loss": 0.027,
+ "step": 30411
+ },
+ {
+ "epoch": 82.86648501362397,
+ "grad_norm": 3.031928777694702,
+ "learning_rate": 1.5005352231589987e-06,
+ "loss": 0.0124,
+ "step": 30412
+ },
+ {
+ "epoch": 82.8692098092643,
+ "grad_norm": 0.42970430850982666,
+ "learning_rate": 1.5000702971550707e-06,
+ "loss": 0.0045,
+ "step": 30413
+ },
+ {
+ "epoch": 82.87193460490464,
+ "grad_norm": 1.1897904872894287,
+ "learning_rate": 1.4996054373483215e-06,
+ "loss": 0.1113,
+ "step": 30414
+ },
+ {
+ "epoch": 82.87465940054496,
+ "grad_norm": 1.04651939868927,
+ "learning_rate": 1.4991406437423694e-06,
+ "loss": 0.0625,
+ "step": 30415
+ },
+ {
+ "epoch": 82.87738419618529,
+ "grad_norm": 1.17790687084198,
+ "learning_rate": 1.4986759163408393e-06,
+ "loss": 0.0162,
+ "step": 30416
+ },
+ {
+ "epoch": 82.88010899182561,
+ "grad_norm": 1.4981619119644165,
+ "learning_rate": 1.498211255147347e-06,
+ "loss": 0.0565,
+ "step": 30417
+ },
+ {
+ "epoch": 82.88283378746594,
+ "grad_norm": 1.0692217350006104,
+ "learning_rate": 1.497746660165511e-06,
+ "loss": 0.0561,
+ "step": 30418
+ },
+ {
+ "epoch": 82.88555858310627,
+ "grad_norm": 1.3957242965698242,
+ "learning_rate": 1.4972821313989495e-06,
+ "loss": 0.0197,
+ "step": 30419
+ },
+ {
+ "epoch": 82.88828337874659,
+ "grad_norm": 0.5802691578865051,
+ "learning_rate": 1.4968176688512824e-06,
+ "loss": 0.0047,
+ "step": 30420
+ },
+ {
+ "epoch": 82.89100817438693,
+ "grad_norm": 1.0447227954864502,
+ "learning_rate": 1.4963532725261266e-06,
+ "loss": 0.0184,
+ "step": 30421
+ },
+ {
+ "epoch": 82.89373297002724,
+ "grad_norm": 1.1170533895492554,
+ "learning_rate": 1.495888942427096e-06,
+ "loss": 0.0174,
+ "step": 30422
+ },
+ {
+ "epoch": 82.89645776566758,
+ "grad_norm": 1.0016272068023682,
+ "learning_rate": 1.4954246785578085e-06,
+ "loss": 0.028,
+ "step": 30423
+ },
+ {
+ "epoch": 82.8991825613079,
+ "grad_norm": 1.4130759239196777,
+ "learning_rate": 1.4949604809218809e-06,
+ "loss": 0.009,
+ "step": 30424
+ },
+ {
+ "epoch": 82.90190735694823,
+ "grad_norm": 1.8887778520584106,
+ "learning_rate": 1.4944963495229281e-06,
+ "loss": 0.0288,
+ "step": 30425
+ },
+ {
+ "epoch": 82.90463215258856,
+ "grad_norm": 0.81565922498703,
+ "learning_rate": 1.4940322843645628e-06,
+ "loss": 0.0066,
+ "step": 30426
+ },
+ {
+ "epoch": 82.90735694822888,
+ "grad_norm": 1.1799578666687012,
+ "learning_rate": 1.4935682854503986e-06,
+ "loss": 0.0123,
+ "step": 30427
+ },
+ {
+ "epoch": 82.91008174386921,
+ "grad_norm": 1.1660012006759644,
+ "learning_rate": 1.4931043527840538e-06,
+ "loss": 0.1204,
+ "step": 30428
+ },
+ {
+ "epoch": 82.91280653950953,
+ "grad_norm": 1.2111999988555908,
+ "learning_rate": 1.4926404863691357e-06,
+ "loss": 0.0286,
+ "step": 30429
+ },
+ {
+ "epoch": 82.91553133514986,
+ "grad_norm": 1.5117849111557007,
+ "learning_rate": 1.4921766862092612e-06,
+ "loss": 0.0125,
+ "step": 30430
+ },
+ {
+ "epoch": 82.9182561307902,
+ "grad_norm": 0.8401332497596741,
+ "learning_rate": 1.491712952308042e-06,
+ "loss": 0.0124,
+ "step": 30431
+ },
+ {
+ "epoch": 82.92098092643052,
+ "grad_norm": 1.496330738067627,
+ "learning_rate": 1.4912492846690862e-06,
+ "loss": 0.0235,
+ "step": 30432
+ },
+ {
+ "epoch": 82.92370572207085,
+ "grad_norm": 1.2777401208877563,
+ "learning_rate": 1.4907856832960088e-06,
+ "loss": 0.0231,
+ "step": 30433
+ },
+ {
+ "epoch": 82.92643051771117,
+ "grad_norm": 1.5825244188308716,
+ "learning_rate": 1.490322148192419e-06,
+ "loss": 0.0765,
+ "step": 30434
+ },
+ {
+ "epoch": 82.9291553133515,
+ "grad_norm": 0.9996124505996704,
+ "learning_rate": 1.489858679361923e-06,
+ "loss": 0.0114,
+ "step": 30435
+ },
+ {
+ "epoch": 82.93188010899182,
+ "grad_norm": 0.8020808100700378,
+ "learning_rate": 1.4893952768081377e-06,
+ "loss": 0.0072,
+ "step": 30436
+ },
+ {
+ "epoch": 82.93460490463215,
+ "grad_norm": 1.6248327493667603,
+ "learning_rate": 1.488931940534667e-06,
+ "loss": 0.1513,
+ "step": 30437
+ },
+ {
+ "epoch": 82.93732970027249,
+ "grad_norm": 1.7258591651916504,
+ "learning_rate": 1.4884686705451213e-06,
+ "loss": 0.1219,
+ "step": 30438
+ },
+ {
+ "epoch": 82.9400544959128,
+ "grad_norm": 1.6514081954956055,
+ "learning_rate": 1.4880054668431043e-06,
+ "loss": 0.0201,
+ "step": 30439
+ },
+ {
+ "epoch": 82.94277929155314,
+ "grad_norm": 0.960905134677887,
+ "learning_rate": 1.48754232943223e-06,
+ "loss": 0.0077,
+ "step": 30440
+ },
+ {
+ "epoch": 82.94550408719346,
+ "grad_norm": 1.7237696647644043,
+ "learning_rate": 1.487079258316101e-06,
+ "loss": 0.0862,
+ "step": 30441
+ },
+ {
+ "epoch": 82.94822888283379,
+ "grad_norm": 0.8534479737281799,
+ "learning_rate": 1.4866162534983252e-06,
+ "loss": 0.0094,
+ "step": 30442
+ },
+ {
+ "epoch": 82.95095367847412,
+ "grad_norm": 1.4426360130310059,
+ "learning_rate": 1.4861533149825058e-06,
+ "loss": 0.0173,
+ "step": 30443
+ },
+ {
+ "epoch": 82.95367847411444,
+ "grad_norm": 0.6997561454772949,
+ "learning_rate": 1.485690442772253e-06,
+ "loss": 0.0054,
+ "step": 30444
+ },
+ {
+ "epoch": 82.95640326975477,
+ "grad_norm": 1.2758557796478271,
+ "learning_rate": 1.4852276368711694e-06,
+ "loss": 0.0539,
+ "step": 30445
+ },
+ {
+ "epoch": 82.95912806539509,
+ "grad_norm": 0.873546302318573,
+ "learning_rate": 1.4847648972828554e-06,
+ "loss": 0.0076,
+ "step": 30446
+ },
+ {
+ "epoch": 82.96185286103542,
+ "grad_norm": 0.613328754901886,
+ "learning_rate": 1.4843022240109218e-06,
+ "loss": 0.0058,
+ "step": 30447
+ },
+ {
+ "epoch": 82.96457765667574,
+ "grad_norm": 1.198778748512268,
+ "learning_rate": 1.4838396170589653e-06,
+ "loss": 0.0086,
+ "step": 30448
+ },
+ {
+ "epoch": 82.96730245231608,
+ "grad_norm": 1.4945597648620605,
+ "learning_rate": 1.4833770764305944e-06,
+ "loss": 0.0327,
+ "step": 30449
+ },
+ {
+ "epoch": 82.97002724795641,
+ "grad_norm": 1.0880647897720337,
+ "learning_rate": 1.4829146021294083e-06,
+ "loss": 0.0097,
+ "step": 30450
+ },
+ {
+ "epoch": 82.97275204359673,
+ "grad_norm": 1.0555146932601929,
+ "learning_rate": 1.4824521941590097e-06,
+ "loss": 0.0089,
+ "step": 30451
+ },
+ {
+ "epoch": 82.97547683923706,
+ "grad_norm": 0.924129843711853,
+ "learning_rate": 1.4819898525229958e-06,
+ "loss": 0.0127,
+ "step": 30452
+ },
+ {
+ "epoch": 82.97820163487738,
+ "grad_norm": 1.3516842126846313,
+ "learning_rate": 1.4815275772249748e-06,
+ "loss": 0.0183,
+ "step": 30453
+ },
+ {
+ "epoch": 82.98092643051771,
+ "grad_norm": 1.4632296562194824,
+ "learning_rate": 1.4810653682685417e-06,
+ "loss": 0.0271,
+ "step": 30454
+ },
+ {
+ "epoch": 82.98365122615803,
+ "grad_norm": 1.288290023803711,
+ "learning_rate": 1.4806032256572978e-06,
+ "loss": 0.0756,
+ "step": 30455
+ },
+ {
+ "epoch": 82.98637602179836,
+ "grad_norm": 1.3794457912445068,
+ "learning_rate": 1.4801411493948392e-06,
+ "loss": 0.0479,
+ "step": 30456
+ },
+ {
+ "epoch": 82.9891008174387,
+ "grad_norm": 0.4818294644355774,
+ "learning_rate": 1.47967913948477e-06,
+ "loss": 0.0053,
+ "step": 30457
+ },
+ {
+ "epoch": 82.99182561307902,
+ "grad_norm": 1.6229397058486938,
+ "learning_rate": 1.4792171959306844e-06,
+ "loss": 0.0314,
+ "step": 30458
+ },
+ {
+ "epoch": 82.99455040871935,
+ "grad_norm": 0.9205620884895325,
+ "learning_rate": 1.4787553187361793e-06,
+ "loss": 0.0101,
+ "step": 30459
+ },
+ {
+ "epoch": 82.99727520435967,
+ "grad_norm": 1.422849416732788,
+ "learning_rate": 1.4782935079048567e-06,
+ "loss": 0.0405,
+ "step": 30460
+ },
+ {
+ "epoch": 83.0,
+ "grad_norm": 1.9244393110275269,
+ "learning_rate": 1.4778317634403082e-06,
+ "loss": 0.0132,
+ "step": 30461
+ },
+ {
+ "epoch": 83.00272479564033,
+ "grad_norm": 1.2337408065795898,
+ "learning_rate": 1.4773700853461337e-06,
+ "loss": 0.0241,
+ "step": 30462
+ },
+ {
+ "epoch": 83.00544959128065,
+ "grad_norm": 0.7322710752487183,
+ "learning_rate": 1.4769084736259232e-06,
+ "loss": 0.0079,
+ "step": 30463
+ },
+ {
+ "epoch": 83.00817438692098,
+ "grad_norm": 1.3760126829147339,
+ "learning_rate": 1.4764469282832749e-06,
+ "loss": 0.0194,
+ "step": 30464
+ },
+ {
+ "epoch": 83.0108991825613,
+ "grad_norm": 1.0027879476547241,
+ "learning_rate": 1.4759854493217863e-06,
+ "loss": 0.0077,
+ "step": 30465
+ },
+ {
+ "epoch": 83.01362397820164,
+ "grad_norm": 0.5840253233909607,
+ "learning_rate": 1.4755240367450485e-06,
+ "loss": 0.0058,
+ "step": 30466
+ },
+ {
+ "epoch": 83.01634877384195,
+ "grad_norm": 0.3388805687427521,
+ "learning_rate": 1.4750626905566557e-06,
+ "loss": 0.0039,
+ "step": 30467
+ },
+ {
+ "epoch": 83.01907356948229,
+ "grad_norm": 1.1585227251052856,
+ "learning_rate": 1.4746014107601969e-06,
+ "loss": 0.0493,
+ "step": 30468
+ },
+ {
+ "epoch": 83.02179836512262,
+ "grad_norm": 1.7236264944076538,
+ "learning_rate": 1.4741401973592718e-06,
+ "loss": 0.0231,
+ "step": 30469
+ },
+ {
+ "epoch": 83.02452316076294,
+ "grad_norm": 1.3799790143966675,
+ "learning_rate": 1.4736790503574672e-06,
+ "loss": 0.0082,
+ "step": 30470
+ },
+ {
+ "epoch": 83.02724795640327,
+ "grad_norm": 1.3145793676376343,
+ "learning_rate": 1.473217969758376e-06,
+ "loss": 0.0239,
+ "step": 30471
+ },
+ {
+ "epoch": 83.02997275204359,
+ "grad_norm": 1.396007776260376,
+ "learning_rate": 1.4727569555655863e-06,
+ "loss": 0.0424,
+ "step": 30472
+ },
+ {
+ "epoch": 83.03269754768392,
+ "grad_norm": 1.246602177619934,
+ "learning_rate": 1.4722960077826941e-06,
+ "loss": 0.1134,
+ "step": 30473
+ },
+ {
+ "epoch": 83.03542234332426,
+ "grad_norm": 0.5653808116912842,
+ "learning_rate": 1.4718351264132858e-06,
+ "loss": 0.0064,
+ "step": 30474
+ },
+ {
+ "epoch": 83.03814713896458,
+ "grad_norm": 0.6748356819152832,
+ "learning_rate": 1.4713743114609501e-06,
+ "loss": 0.0098,
+ "step": 30475
+ },
+ {
+ "epoch": 83.04087193460491,
+ "grad_norm": 1.9163353443145752,
+ "learning_rate": 1.4709135629292759e-06,
+ "loss": 0.0188,
+ "step": 30476
+ },
+ {
+ "epoch": 83.04359673024523,
+ "grad_norm": 1.4702714681625366,
+ "learning_rate": 1.4704528808218544e-06,
+ "loss": 0.0678,
+ "step": 30477
+ },
+ {
+ "epoch": 83.04632152588556,
+ "grad_norm": 1.171187400817871,
+ "learning_rate": 1.4699922651422705e-06,
+ "loss": 0.1674,
+ "step": 30478
+ },
+ {
+ "epoch": 83.04904632152588,
+ "grad_norm": 1.2275354862213135,
+ "learning_rate": 1.4695317158941126e-06,
+ "loss": 0.0101,
+ "step": 30479
+ },
+ {
+ "epoch": 83.05177111716621,
+ "grad_norm": 0.9208979606628418,
+ "learning_rate": 1.4690712330809653e-06,
+ "loss": 0.009,
+ "step": 30480
+ },
+ {
+ "epoch": 83.05449591280654,
+ "grad_norm": 2.0512964725494385,
+ "learning_rate": 1.4686108167064162e-06,
+ "loss": 0.1239,
+ "step": 30481
+ },
+ {
+ "epoch": 83.05722070844686,
+ "grad_norm": 3.5405900478363037,
+ "learning_rate": 1.4681504667740544e-06,
+ "loss": 0.0544,
+ "step": 30482
+ },
+ {
+ "epoch": 83.0599455040872,
+ "grad_norm": 0.9799007177352905,
+ "learning_rate": 1.4676901832874612e-06,
+ "loss": 0.0192,
+ "step": 30483
+ },
+ {
+ "epoch": 83.06267029972751,
+ "grad_norm": 0.9926896095275879,
+ "learning_rate": 1.467229966250221e-06,
+ "loss": 0.0194,
+ "step": 30484
+ },
+ {
+ "epoch": 83.06539509536785,
+ "grad_norm": 1.2072227001190186,
+ "learning_rate": 1.4667698156659205e-06,
+ "loss": 0.0161,
+ "step": 30485
+ },
+ {
+ "epoch": 83.06811989100818,
+ "grad_norm": 1.1208873987197876,
+ "learning_rate": 1.4663097315381426e-06,
+ "loss": 0.0353,
+ "step": 30486
+ },
+ {
+ "epoch": 83.0708446866485,
+ "grad_norm": 0.8649042248725891,
+ "learning_rate": 1.4658497138704697e-06,
+ "loss": 0.0104,
+ "step": 30487
+ },
+ {
+ "epoch": 83.07356948228883,
+ "grad_norm": 0.9429376721382141,
+ "learning_rate": 1.4653897626664825e-06,
+ "loss": 0.0775,
+ "step": 30488
+ },
+ {
+ "epoch": 83.07629427792915,
+ "grad_norm": 1.3533298969268799,
+ "learning_rate": 1.4649298779297671e-06,
+ "loss": 0.0151,
+ "step": 30489
+ },
+ {
+ "epoch": 83.07901907356948,
+ "grad_norm": 0.8481666445732117,
+ "learning_rate": 1.4644700596639028e-06,
+ "loss": 0.0085,
+ "step": 30490
+ },
+ {
+ "epoch": 83.0817438692098,
+ "grad_norm": 1.6667215824127197,
+ "learning_rate": 1.4640103078724722e-06,
+ "loss": 0.0814,
+ "step": 30491
+ },
+ {
+ "epoch": 83.08446866485014,
+ "grad_norm": 2.9963479042053223,
+ "learning_rate": 1.4635506225590511e-06,
+ "loss": 0.0158,
+ "step": 30492
+ },
+ {
+ "epoch": 83.08719346049047,
+ "grad_norm": 1.0372586250305176,
+ "learning_rate": 1.4630910037272262e-06,
+ "loss": 0.0113,
+ "step": 30493
+ },
+ {
+ "epoch": 83.08991825613079,
+ "grad_norm": 1.4380046129226685,
+ "learning_rate": 1.4626314513805728e-06,
+ "loss": 0.0188,
+ "step": 30494
+ },
+ {
+ "epoch": 83.09264305177112,
+ "grad_norm": 0.5643282532691956,
+ "learning_rate": 1.4621719655226718e-06,
+ "loss": 0.0041,
+ "step": 30495
+ },
+ {
+ "epoch": 83.09536784741144,
+ "grad_norm": 2.0453972816467285,
+ "learning_rate": 1.461712546157098e-06,
+ "loss": 0.0728,
+ "step": 30496
+ },
+ {
+ "epoch": 83.09809264305177,
+ "grad_norm": 1.2031112909317017,
+ "learning_rate": 1.461253193287434e-06,
+ "loss": 0.0423,
+ "step": 30497
+ },
+ {
+ "epoch": 83.1008174386921,
+ "grad_norm": 1.2288151979446411,
+ "learning_rate": 1.4607939069172539e-06,
+ "loss": 0.0184,
+ "step": 30498
+ },
+ {
+ "epoch": 83.10354223433242,
+ "grad_norm": 1.4761908054351807,
+ "learning_rate": 1.4603346870501377e-06,
+ "loss": 0.0271,
+ "step": 30499
+ },
+ {
+ "epoch": 83.10626702997276,
+ "grad_norm": 0.9836786985397339,
+ "learning_rate": 1.4598755336896575e-06,
+ "loss": 0.0093,
+ "step": 30500
+ },
+ {
+ "epoch": 83.10899182561307,
+ "grad_norm": 1.3904423713684082,
+ "learning_rate": 1.4594164468393945e-06,
+ "loss": 0.0095,
+ "step": 30501
+ },
+ {
+ "epoch": 83.11171662125341,
+ "grad_norm": 0.9789084792137146,
+ "learning_rate": 1.4589574265029216e-06,
+ "loss": 0.0269,
+ "step": 30502
+ },
+ {
+ "epoch": 83.11444141689373,
+ "grad_norm": 0.7557938694953918,
+ "learning_rate": 1.4584984726838136e-06,
+ "loss": 0.0067,
+ "step": 30503
+ },
+ {
+ "epoch": 83.11716621253406,
+ "grad_norm": 0.9836606979370117,
+ "learning_rate": 1.4580395853856412e-06,
+ "loss": 0.0095,
+ "step": 30504
+ },
+ {
+ "epoch": 83.11989100817439,
+ "grad_norm": 1.6951881647109985,
+ "learning_rate": 1.4575807646119844e-06,
+ "loss": 0.0755,
+ "step": 30505
+ },
+ {
+ "epoch": 83.12261580381471,
+ "grad_norm": 1.242132306098938,
+ "learning_rate": 1.4571220103664142e-06,
+ "loss": 0.0816,
+ "step": 30506
+ },
+ {
+ "epoch": 83.12534059945504,
+ "grad_norm": 1.4365614652633667,
+ "learning_rate": 1.4566633226525028e-06,
+ "loss": 0.0855,
+ "step": 30507
+ },
+ {
+ "epoch": 83.12806539509536,
+ "grad_norm": 0.6092231273651123,
+ "learning_rate": 1.456204701473819e-06,
+ "loss": 0.0064,
+ "step": 30508
+ },
+ {
+ "epoch": 83.1307901907357,
+ "grad_norm": 1.2027146816253662,
+ "learning_rate": 1.4557461468339418e-06,
+ "loss": 0.0103,
+ "step": 30509
+ },
+ {
+ "epoch": 83.13351498637603,
+ "grad_norm": 1.1049292087554932,
+ "learning_rate": 1.4552876587364373e-06,
+ "loss": 0.0121,
+ "step": 30510
+ },
+ {
+ "epoch": 83.13623978201635,
+ "grad_norm": 1.5205433368682861,
+ "learning_rate": 1.4548292371848782e-06,
+ "loss": 0.0504,
+ "step": 30511
+ },
+ {
+ "epoch": 83.13896457765668,
+ "grad_norm": 1.2204991579055786,
+ "learning_rate": 1.4543708821828317e-06,
+ "loss": 0.005,
+ "step": 30512
+ },
+ {
+ "epoch": 83.141689373297,
+ "grad_norm": 1.284879207611084,
+ "learning_rate": 1.4539125937338727e-06,
+ "loss": 0.0102,
+ "step": 30513
+ },
+ {
+ "epoch": 83.14441416893733,
+ "grad_norm": 1.9000215530395508,
+ "learning_rate": 1.453454371841566e-06,
+ "loss": 0.0424,
+ "step": 30514
+ },
+ {
+ "epoch": 83.14713896457765,
+ "grad_norm": 0.9576534628868103,
+ "learning_rate": 1.4529962165094803e-06,
+ "loss": 0.0234,
+ "step": 30515
+ },
+ {
+ "epoch": 83.14986376021798,
+ "grad_norm": 1.4848943948745728,
+ "learning_rate": 1.4525381277411864e-06,
+ "loss": 0.0156,
+ "step": 30516
+ },
+ {
+ "epoch": 83.15258855585832,
+ "grad_norm": 1.4787318706512451,
+ "learning_rate": 1.452080105540249e-06,
+ "loss": 0.012,
+ "step": 30517
+ },
+ {
+ "epoch": 83.15531335149863,
+ "grad_norm": 0.8263862729072571,
+ "learning_rate": 1.451622149910239e-06,
+ "loss": 0.0092,
+ "step": 30518
+ },
+ {
+ "epoch": 83.15803814713897,
+ "grad_norm": 1.6707484722137451,
+ "learning_rate": 1.4511642608547195e-06,
+ "loss": 0.1459,
+ "step": 30519
+ },
+ {
+ "epoch": 83.16076294277929,
+ "grad_norm": 1.4901715517044067,
+ "learning_rate": 1.4507064383772585e-06,
+ "loss": 0.0278,
+ "step": 30520
+ },
+ {
+ "epoch": 83.16348773841962,
+ "grad_norm": 1.115020513534546,
+ "learning_rate": 1.4502486824814176e-06,
+ "loss": 0.0438,
+ "step": 30521
+ },
+ {
+ "epoch": 83.16621253405995,
+ "grad_norm": 1.7271740436553955,
+ "learning_rate": 1.4497909931707677e-06,
+ "loss": 0.0255,
+ "step": 30522
+ },
+ {
+ "epoch": 83.16893732970027,
+ "grad_norm": 0.40085500478744507,
+ "learning_rate": 1.4493333704488698e-06,
+ "loss": 0.0047,
+ "step": 30523
+ },
+ {
+ "epoch": 83.1716621253406,
+ "grad_norm": 0.7222799062728882,
+ "learning_rate": 1.4488758143192871e-06,
+ "loss": 0.0069,
+ "step": 30524
+ },
+ {
+ "epoch": 83.17438692098092,
+ "grad_norm": 1.4328985214233398,
+ "learning_rate": 1.4484183247855853e-06,
+ "loss": 0.0572,
+ "step": 30525
+ },
+ {
+ "epoch": 83.17711171662125,
+ "grad_norm": 3.9997217655181885,
+ "learning_rate": 1.4479609018513275e-06,
+ "loss": 0.0277,
+ "step": 30526
+ },
+ {
+ "epoch": 83.17983651226157,
+ "grad_norm": 1.0311381816864014,
+ "learning_rate": 1.4475035455200748e-06,
+ "loss": 0.0162,
+ "step": 30527
+ },
+ {
+ "epoch": 83.1825613079019,
+ "grad_norm": 0.7858263850212097,
+ "learning_rate": 1.4470462557953869e-06,
+ "loss": 0.0087,
+ "step": 30528
+ },
+ {
+ "epoch": 83.18528610354224,
+ "grad_norm": 1.1314302682876587,
+ "learning_rate": 1.4465890326808297e-06,
+ "loss": 0.0148,
+ "step": 30529
+ },
+ {
+ "epoch": 83.18801089918256,
+ "grad_norm": 0.894782543182373,
+ "learning_rate": 1.4461318761799615e-06,
+ "loss": 0.0057,
+ "step": 30530
+ },
+ {
+ "epoch": 83.19073569482289,
+ "grad_norm": 1.3311378955841064,
+ "learning_rate": 1.4456747862963427e-06,
+ "loss": 0.0183,
+ "step": 30531
+ },
+ {
+ "epoch": 83.19346049046321,
+ "grad_norm": 1.6061079502105713,
+ "learning_rate": 1.4452177630335318e-06,
+ "loss": 0.0269,
+ "step": 30532
+ },
+ {
+ "epoch": 83.19618528610354,
+ "grad_norm": 1.2109295129776,
+ "learning_rate": 1.444760806395089e-06,
+ "loss": 0.0149,
+ "step": 30533
+ },
+ {
+ "epoch": 83.19891008174388,
+ "grad_norm": 0.8822950720787048,
+ "learning_rate": 1.4443039163845761e-06,
+ "loss": 0.0081,
+ "step": 30534
+ },
+ {
+ "epoch": 83.2016348773842,
+ "grad_norm": 1.1302238702774048,
+ "learning_rate": 1.4438470930055492e-06,
+ "loss": 0.018,
+ "step": 30535
+ },
+ {
+ "epoch": 83.20435967302453,
+ "grad_norm": 1.0256437063217163,
+ "learning_rate": 1.4433903362615654e-06,
+ "loss": 0.0217,
+ "step": 30536
+ },
+ {
+ "epoch": 83.20708446866485,
+ "grad_norm": 1.2418361902236938,
+ "learning_rate": 1.4429336461561804e-06,
+ "loss": 0.0783,
+ "step": 30537
+ },
+ {
+ "epoch": 83.20980926430518,
+ "grad_norm": 0.858366847038269,
+ "learning_rate": 1.4424770226929551e-06,
+ "loss": 0.0087,
+ "step": 30538
+ },
+ {
+ "epoch": 83.2125340599455,
+ "grad_norm": 1.6130478382110596,
+ "learning_rate": 1.4420204658754432e-06,
+ "loss": 0.0327,
+ "step": 30539
+ },
+ {
+ "epoch": 83.21525885558583,
+ "grad_norm": 1.2198225259780884,
+ "learning_rate": 1.4415639757071998e-06,
+ "loss": 0.0835,
+ "step": 30540
+ },
+ {
+ "epoch": 83.21798365122616,
+ "grad_norm": 0.7292611002922058,
+ "learning_rate": 1.4411075521917794e-06,
+ "loss": 0.008,
+ "step": 30541
+ },
+ {
+ "epoch": 83.22070844686648,
+ "grad_norm": 1.216036081314087,
+ "learning_rate": 1.4406511953327397e-06,
+ "loss": 0.0343,
+ "step": 30542
+ },
+ {
+ "epoch": 83.22343324250681,
+ "grad_norm": 1.224653959274292,
+ "learning_rate": 1.4401949051336328e-06,
+ "loss": 0.0786,
+ "step": 30543
+ },
+ {
+ "epoch": 83.22615803814713,
+ "grad_norm": 1.9465508460998535,
+ "learning_rate": 1.4397386815980107e-06,
+ "loss": 0.058,
+ "step": 30544
+ },
+ {
+ "epoch": 83.22888283378747,
+ "grad_norm": 1.231035590171814,
+ "learning_rate": 1.4392825247294307e-06,
+ "loss": 0.031,
+ "step": 30545
+ },
+ {
+ "epoch": 83.2316076294278,
+ "grad_norm": 0.7885746955871582,
+ "learning_rate": 1.4388264345314419e-06,
+ "loss": 0.008,
+ "step": 30546
+ },
+ {
+ "epoch": 83.23433242506812,
+ "grad_norm": 0.9534540772438049,
+ "learning_rate": 1.438370411007598e-06,
+ "loss": 0.0182,
+ "step": 30547
+ },
+ {
+ "epoch": 83.23705722070845,
+ "grad_norm": 0.8718841075897217,
+ "learning_rate": 1.4379144541614476e-06,
+ "loss": 0.0086,
+ "step": 30548
+ },
+ {
+ "epoch": 83.23978201634877,
+ "grad_norm": 1.317224383354187,
+ "learning_rate": 1.4374585639965433e-06,
+ "loss": 0.0205,
+ "step": 30549
+ },
+ {
+ "epoch": 83.2425068119891,
+ "grad_norm": 1.4954283237457275,
+ "learning_rate": 1.437002740516439e-06,
+ "loss": 0.0236,
+ "step": 30550
+ },
+ {
+ "epoch": 83.24523160762942,
+ "grad_norm": 1.2485885620117188,
+ "learning_rate": 1.4365469837246816e-06,
+ "loss": 0.0153,
+ "step": 30551
+ },
+ {
+ "epoch": 83.24795640326975,
+ "grad_norm": 1.2396241426467896,
+ "learning_rate": 1.4360912936248206e-06,
+ "loss": 0.0081,
+ "step": 30552
+ },
+ {
+ "epoch": 83.25068119891009,
+ "grad_norm": 1.4365724325180054,
+ "learning_rate": 1.4356356702204033e-06,
+ "loss": 0.0189,
+ "step": 30553
+ },
+ {
+ "epoch": 83.2534059945504,
+ "grad_norm": 2.847155809402466,
+ "learning_rate": 1.4351801135149812e-06,
+ "loss": 0.0266,
+ "step": 30554
+ },
+ {
+ "epoch": 83.25613079019074,
+ "grad_norm": 1.3297019004821777,
+ "learning_rate": 1.4347246235121003e-06,
+ "loss": 0.0429,
+ "step": 30555
+ },
+ {
+ "epoch": 83.25885558583106,
+ "grad_norm": 1.464576005935669,
+ "learning_rate": 1.434269200215309e-06,
+ "loss": 0.1603,
+ "step": 30556
+ },
+ {
+ "epoch": 83.26158038147139,
+ "grad_norm": 0.9378790855407715,
+ "learning_rate": 1.4338138436281512e-06,
+ "loss": 0.0095,
+ "step": 30557
+ },
+ {
+ "epoch": 83.26430517711172,
+ "grad_norm": 0.5424307584762573,
+ "learning_rate": 1.4333585537541782e-06,
+ "loss": 0.0048,
+ "step": 30558
+ },
+ {
+ "epoch": 83.26702997275204,
+ "grad_norm": 1.0061924457550049,
+ "learning_rate": 1.432903330596933e-06,
+ "loss": 0.0537,
+ "step": 30559
+ },
+ {
+ "epoch": 83.26975476839237,
+ "grad_norm": 0.7608798146247864,
+ "learning_rate": 1.4324481741599606e-06,
+ "loss": 0.0096,
+ "step": 30560
+ },
+ {
+ "epoch": 83.2724795640327,
+ "grad_norm": 1.2940722703933716,
+ "learning_rate": 1.4319930844468032e-06,
+ "loss": 0.0166,
+ "step": 30561
+ },
+ {
+ "epoch": 83.27520435967303,
+ "grad_norm": 2.1042635440826416,
+ "learning_rate": 1.4315380614610109e-06,
+ "loss": 0.0679,
+ "step": 30562
+ },
+ {
+ "epoch": 83.27792915531334,
+ "grad_norm": 0.8083535432815552,
+ "learning_rate": 1.4310831052061248e-06,
+ "loss": 0.0117,
+ "step": 30563
+ },
+ {
+ "epoch": 83.28065395095368,
+ "grad_norm": 1.5949532985687256,
+ "learning_rate": 1.4306282156856866e-06,
+ "loss": 0.0504,
+ "step": 30564
+ },
+ {
+ "epoch": 83.28337874659401,
+ "grad_norm": 1.218855619430542,
+ "learning_rate": 1.430173392903238e-06,
+ "loss": 0.0165,
+ "step": 30565
+ },
+ {
+ "epoch": 83.28610354223433,
+ "grad_norm": 0.6062264442443848,
+ "learning_rate": 1.4297186368623262e-06,
+ "loss": 0.0049,
+ "step": 30566
+ },
+ {
+ "epoch": 83.28882833787466,
+ "grad_norm": 1.2433676719665527,
+ "learning_rate": 1.4292639475664872e-06,
+ "loss": 0.0534,
+ "step": 30567
+ },
+ {
+ "epoch": 83.29155313351498,
+ "grad_norm": 1.2115095853805542,
+ "learning_rate": 1.4288093250192669e-06,
+ "loss": 0.1164,
+ "step": 30568
+ },
+ {
+ "epoch": 83.29427792915531,
+ "grad_norm": 0.9426988959312439,
+ "learning_rate": 1.4283547692242017e-06,
+ "loss": 0.0125,
+ "step": 30569
+ },
+ {
+ "epoch": 83.29700272479565,
+ "grad_norm": 2.1717283725738525,
+ "learning_rate": 1.4279002801848352e-06,
+ "loss": 0.0174,
+ "step": 30570
+ },
+ {
+ "epoch": 83.29972752043597,
+ "grad_norm": 0.7269127368927002,
+ "learning_rate": 1.4274458579047056e-06,
+ "loss": 0.0064,
+ "step": 30571
+ },
+ {
+ "epoch": 83.3024523160763,
+ "grad_norm": 1.4889981746673584,
+ "learning_rate": 1.4269915023873526e-06,
+ "loss": 0.0253,
+ "step": 30572
+ },
+ {
+ "epoch": 83.30517711171662,
+ "grad_norm": 1.4597381353378296,
+ "learning_rate": 1.4265372136363109e-06,
+ "loss": 0.0123,
+ "step": 30573
+ },
+ {
+ "epoch": 83.30790190735695,
+ "grad_norm": 0.9008587002754211,
+ "learning_rate": 1.4260829916551245e-06,
+ "loss": 0.0072,
+ "step": 30574
+ },
+ {
+ "epoch": 83.31062670299727,
+ "grad_norm": 0.7763868570327759,
+ "learning_rate": 1.425628836447327e-06,
+ "loss": 0.0077,
+ "step": 30575
+ },
+ {
+ "epoch": 83.3133514986376,
+ "grad_norm": 0.8953559398651123,
+ "learning_rate": 1.4251747480164557e-06,
+ "loss": 0.0204,
+ "step": 30576
+ },
+ {
+ "epoch": 83.31607629427793,
+ "grad_norm": 1.7143878936767578,
+ "learning_rate": 1.4247207263660467e-06,
+ "loss": 0.0541,
+ "step": 30577
+ },
+ {
+ "epoch": 83.31880108991825,
+ "grad_norm": 1.1666144132614136,
+ "learning_rate": 1.4242667714996371e-06,
+ "loss": 0.0302,
+ "step": 30578
+ },
+ {
+ "epoch": 83.32152588555859,
+ "grad_norm": 1.0364078283309937,
+ "learning_rate": 1.4238128834207633e-06,
+ "loss": 0.0105,
+ "step": 30579
+ },
+ {
+ "epoch": 83.3242506811989,
+ "grad_norm": 2.2579712867736816,
+ "learning_rate": 1.4233590621329585e-06,
+ "loss": 0.0248,
+ "step": 30580
+ },
+ {
+ "epoch": 83.32697547683924,
+ "grad_norm": 1.0954582691192627,
+ "learning_rate": 1.4229053076397547e-06,
+ "loss": 0.0482,
+ "step": 30581
+ },
+ {
+ "epoch": 83.32970027247957,
+ "grad_norm": 1.4485355615615845,
+ "learning_rate": 1.4224516199446915e-06,
+ "loss": 0.0424,
+ "step": 30582
+ },
+ {
+ "epoch": 83.33242506811989,
+ "grad_norm": 0.6962015628814697,
+ "learning_rate": 1.4219979990512977e-06,
+ "loss": 0.0067,
+ "step": 30583
+ },
+ {
+ "epoch": 83.33514986376022,
+ "grad_norm": 1.4085817337036133,
+ "learning_rate": 1.4215444449631054e-06,
+ "loss": 0.0406,
+ "step": 30584
+ },
+ {
+ "epoch": 83.33787465940054,
+ "grad_norm": 0.7302588224411011,
+ "learning_rate": 1.4210909576836496e-06,
+ "loss": 0.0115,
+ "step": 30585
+ },
+ {
+ "epoch": 83.34059945504087,
+ "grad_norm": 1.0565569400787354,
+ "learning_rate": 1.4206375372164628e-06,
+ "loss": 0.0182,
+ "step": 30586
+ },
+ {
+ "epoch": 83.34332425068119,
+ "grad_norm": 1.2028887271881104,
+ "learning_rate": 1.4201841835650755e-06,
+ "loss": 0.137,
+ "step": 30587
+ },
+ {
+ "epoch": 83.34604904632153,
+ "grad_norm": 0.23960934579372406,
+ "learning_rate": 1.419730896733017e-06,
+ "loss": 0.0029,
+ "step": 30588
+ },
+ {
+ "epoch": 83.34877384196186,
+ "grad_norm": 2.5155680179595947,
+ "learning_rate": 1.419277676723816e-06,
+ "loss": 0.0474,
+ "step": 30589
+ },
+ {
+ "epoch": 83.35149863760218,
+ "grad_norm": 0.8853926658630371,
+ "learning_rate": 1.418824523541007e-06,
+ "loss": 0.0103,
+ "step": 30590
+ },
+ {
+ "epoch": 83.35422343324251,
+ "grad_norm": 1.303849458694458,
+ "learning_rate": 1.4183714371881164e-06,
+ "loss": 0.0241,
+ "step": 30591
+ },
+ {
+ "epoch": 83.35694822888283,
+ "grad_norm": 1.883477807044983,
+ "learning_rate": 1.4179184176686722e-06,
+ "loss": 0.0785,
+ "step": 30592
+ },
+ {
+ "epoch": 83.35967302452316,
+ "grad_norm": 0.743253767490387,
+ "learning_rate": 1.4174654649862007e-06,
+ "loss": 0.0055,
+ "step": 30593
+ },
+ {
+ "epoch": 83.3623978201635,
+ "grad_norm": 1.3586562871932983,
+ "learning_rate": 1.4170125791442346e-06,
+ "loss": 0.0121,
+ "step": 30594
+ },
+ {
+ "epoch": 83.36512261580381,
+ "grad_norm": 1.1491588354110718,
+ "learning_rate": 1.4165597601462987e-06,
+ "loss": 0.0103,
+ "step": 30595
+ },
+ {
+ "epoch": 83.36784741144415,
+ "grad_norm": 1.5295398235321045,
+ "learning_rate": 1.4161070079959193e-06,
+ "loss": 0.0168,
+ "step": 30596
+ },
+ {
+ "epoch": 83.37057220708446,
+ "grad_norm": 0.8028443455696106,
+ "learning_rate": 1.4156543226966191e-06,
+ "loss": 0.0105,
+ "step": 30597
+ },
+ {
+ "epoch": 83.3732970027248,
+ "grad_norm": 1.0419270992279053,
+ "learning_rate": 1.4152017042519295e-06,
+ "loss": 0.0192,
+ "step": 30598
+ },
+ {
+ "epoch": 83.37602179836512,
+ "grad_norm": 0.9783596396446228,
+ "learning_rate": 1.4147491526653723e-06,
+ "loss": 0.0218,
+ "step": 30599
+ },
+ {
+ "epoch": 83.37874659400545,
+ "grad_norm": 1.376021146774292,
+ "learning_rate": 1.4142966679404724e-06,
+ "loss": 0.0106,
+ "step": 30600
+ },
+ {
+ "epoch": 83.38147138964578,
+ "grad_norm": 0.7631276845932007,
+ "learning_rate": 1.4138442500807524e-06,
+ "loss": 0.0111,
+ "step": 30601
+ },
+ {
+ "epoch": 83.3841961852861,
+ "grad_norm": 0.5510170459747314,
+ "learning_rate": 1.4133918990897366e-06,
+ "loss": 0.0057,
+ "step": 30602
+ },
+ {
+ "epoch": 83.38692098092643,
+ "grad_norm": 1.2901009321212769,
+ "learning_rate": 1.4129396149709496e-06,
+ "loss": 0.0247,
+ "step": 30603
+ },
+ {
+ "epoch": 83.38964577656675,
+ "grad_norm": 1.1238586902618408,
+ "learning_rate": 1.412487397727914e-06,
+ "loss": 0.0176,
+ "step": 30604
+ },
+ {
+ "epoch": 83.39237057220708,
+ "grad_norm": 1.1321161985397339,
+ "learning_rate": 1.4120352473641497e-06,
+ "loss": 0.0176,
+ "step": 30605
+ },
+ {
+ "epoch": 83.39509536784742,
+ "grad_norm": 3.6607682704925537,
+ "learning_rate": 1.4115831638831756e-06,
+ "loss": 0.0132,
+ "step": 30606
+ },
+ {
+ "epoch": 83.39782016348774,
+ "grad_norm": 1.5752092599868774,
+ "learning_rate": 1.411131147288518e-06,
+ "loss": 0.0364,
+ "step": 30607
+ },
+ {
+ "epoch": 83.40054495912807,
+ "grad_norm": 1.0988633632659912,
+ "learning_rate": 1.4106791975836941e-06,
+ "loss": 0.0104,
+ "step": 30608
+ },
+ {
+ "epoch": 83.40326975476839,
+ "grad_norm": 0.8492132425308228,
+ "learning_rate": 1.4102273147722235e-06,
+ "loss": 0.0054,
+ "step": 30609
+ },
+ {
+ "epoch": 83.40599455040872,
+ "grad_norm": 1.0630353689193726,
+ "learning_rate": 1.4097754988576273e-06,
+ "loss": 0.0117,
+ "step": 30610
+ },
+ {
+ "epoch": 83.40871934604904,
+ "grad_norm": 0.6179193258285522,
+ "learning_rate": 1.4093237498434232e-06,
+ "loss": 0.0052,
+ "step": 30611
+ },
+ {
+ "epoch": 83.41144414168937,
+ "grad_norm": 0.6706658601760864,
+ "learning_rate": 1.408872067733128e-06,
+ "loss": 0.0095,
+ "step": 30612
+ },
+ {
+ "epoch": 83.4141689373297,
+ "grad_norm": 1.2895252704620361,
+ "learning_rate": 1.4084204525302602e-06,
+ "loss": 0.0335,
+ "step": 30613
+ },
+ {
+ "epoch": 83.41689373297002,
+ "grad_norm": 1.331107258796692,
+ "learning_rate": 1.4079689042383394e-06,
+ "loss": 0.0204,
+ "step": 30614
+ },
+ {
+ "epoch": 83.41961852861036,
+ "grad_norm": 1.4770256280899048,
+ "learning_rate": 1.407517422860879e-06,
+ "loss": 0.0125,
+ "step": 30615
+ },
+ {
+ "epoch": 83.42234332425068,
+ "grad_norm": 0.9371821880340576,
+ "learning_rate": 1.4070660084013977e-06,
+ "loss": 0.0175,
+ "step": 30616
+ },
+ {
+ "epoch": 83.42506811989101,
+ "grad_norm": 1.1315960884094238,
+ "learning_rate": 1.4066146608634068e-06,
+ "loss": 0.0105,
+ "step": 30617
+ },
+ {
+ "epoch": 83.42779291553134,
+ "grad_norm": 1.2609113454818726,
+ "learning_rate": 1.4061633802504237e-06,
+ "loss": 0.0116,
+ "step": 30618
+ },
+ {
+ "epoch": 83.43051771117166,
+ "grad_norm": 1.0493215322494507,
+ "learning_rate": 1.4057121665659667e-06,
+ "loss": 0.0168,
+ "step": 30619
+ },
+ {
+ "epoch": 83.433242506812,
+ "grad_norm": 0.40471965074539185,
+ "learning_rate": 1.4052610198135475e-06,
+ "loss": 0.0039,
+ "step": 30620
+ },
+ {
+ "epoch": 83.43596730245231,
+ "grad_norm": 1.1904324293136597,
+ "learning_rate": 1.4048099399966774e-06,
+ "loss": 0.0426,
+ "step": 30621
+ },
+ {
+ "epoch": 83.43869209809264,
+ "grad_norm": 1.2808705568313599,
+ "learning_rate": 1.4043589271188695e-06,
+ "loss": 0.0138,
+ "step": 30622
+ },
+ {
+ "epoch": 83.44141689373296,
+ "grad_norm": 0.8344308137893677,
+ "learning_rate": 1.4039079811836398e-06,
+ "loss": 0.0078,
+ "step": 30623
+ },
+ {
+ "epoch": 83.4441416893733,
+ "grad_norm": 1.123140811920166,
+ "learning_rate": 1.4034571021944976e-06,
+ "loss": 0.0943,
+ "step": 30624
+ },
+ {
+ "epoch": 83.44686648501363,
+ "grad_norm": 1.2470723390579224,
+ "learning_rate": 1.4030062901549557e-06,
+ "loss": 0.0137,
+ "step": 30625
+ },
+ {
+ "epoch": 83.44959128065395,
+ "grad_norm": 1.5684654712677002,
+ "learning_rate": 1.4025555450685213e-06,
+ "loss": 0.0082,
+ "step": 30626
+ },
+ {
+ "epoch": 83.45231607629428,
+ "grad_norm": 0.9526797533035278,
+ "learning_rate": 1.4021048669387094e-06,
+ "loss": 0.0102,
+ "step": 30627
+ },
+ {
+ "epoch": 83.4550408719346,
+ "grad_norm": 0.5674511194229126,
+ "learning_rate": 1.4016542557690282e-06,
+ "loss": 0.0062,
+ "step": 30628
+ },
+ {
+ "epoch": 83.45776566757493,
+ "grad_norm": 1.4819567203521729,
+ "learning_rate": 1.4012037115629873e-06,
+ "loss": 0.0244,
+ "step": 30629
+ },
+ {
+ "epoch": 83.46049046321527,
+ "grad_norm": 1.1792335510253906,
+ "learning_rate": 1.4007532343240936e-06,
+ "loss": 0.016,
+ "step": 30630
+ },
+ {
+ "epoch": 83.46321525885558,
+ "grad_norm": 1.103869915008545,
+ "learning_rate": 1.400302824055858e-06,
+ "loss": 0.0085,
+ "step": 30631
+ },
+ {
+ "epoch": 83.46594005449592,
+ "grad_norm": 0.45725321769714355,
+ "learning_rate": 1.3998524807617875e-06,
+ "loss": 0.0053,
+ "step": 30632
+ },
+ {
+ "epoch": 83.46866485013624,
+ "grad_norm": 0.6155889630317688,
+ "learning_rate": 1.399402204445387e-06,
+ "loss": 0.0046,
+ "step": 30633
+ },
+ {
+ "epoch": 83.47138964577657,
+ "grad_norm": 1.8144148588180542,
+ "learning_rate": 1.3989519951101682e-06,
+ "loss": 0.0127,
+ "step": 30634
+ },
+ {
+ "epoch": 83.47411444141689,
+ "grad_norm": 1.408462405204773,
+ "learning_rate": 1.3985018527596316e-06,
+ "loss": 0.0139,
+ "step": 30635
+ },
+ {
+ "epoch": 83.47683923705722,
+ "grad_norm": 2.386399269104004,
+ "learning_rate": 1.398051777397288e-06,
+ "loss": 0.018,
+ "step": 30636
+ },
+ {
+ "epoch": 83.47956403269755,
+ "grad_norm": 1.0543404817581177,
+ "learning_rate": 1.3976017690266397e-06,
+ "loss": 0.0163,
+ "step": 30637
+ },
+ {
+ "epoch": 83.48228882833787,
+ "grad_norm": 0.6442644000053406,
+ "learning_rate": 1.3971518276511909e-06,
+ "loss": 0.0096,
+ "step": 30638
+ },
+ {
+ "epoch": 83.4850136239782,
+ "grad_norm": 1.3273049592971802,
+ "learning_rate": 1.3967019532744475e-06,
+ "loss": 0.0152,
+ "step": 30639
+ },
+ {
+ "epoch": 83.48773841961852,
+ "grad_norm": 1.2922780513763428,
+ "learning_rate": 1.3962521458999135e-06,
+ "loss": 0.0241,
+ "step": 30640
+ },
+ {
+ "epoch": 83.49046321525886,
+ "grad_norm": 0.9594244360923767,
+ "learning_rate": 1.3958024055310903e-06,
+ "loss": 0.0069,
+ "step": 30641
+ },
+ {
+ "epoch": 83.49318801089919,
+ "grad_norm": 1.3037254810333252,
+ "learning_rate": 1.39535273217148e-06,
+ "loss": 0.0458,
+ "step": 30642
+ },
+ {
+ "epoch": 83.49591280653951,
+ "grad_norm": 0.9800367951393127,
+ "learning_rate": 1.3949031258245872e-06,
+ "loss": 0.0159,
+ "step": 30643
+ },
+ {
+ "epoch": 83.49863760217984,
+ "grad_norm": 1.4062539339065552,
+ "learning_rate": 1.3944535864939112e-06,
+ "loss": 0.0147,
+ "step": 30644
+ },
+ {
+ "epoch": 83.50136239782016,
+ "grad_norm": 1.7982226610183716,
+ "learning_rate": 1.394004114182954e-06,
+ "loss": 0.0515,
+ "step": 30645
+ },
+ {
+ "epoch": 83.50408719346049,
+ "grad_norm": 1.1153644323349,
+ "learning_rate": 1.393554708895214e-06,
+ "loss": 0.0187,
+ "step": 30646
+ },
+ {
+ "epoch": 83.50681198910081,
+ "grad_norm": 1.228195071220398,
+ "learning_rate": 1.3931053706341946e-06,
+ "loss": 0.01,
+ "step": 30647
+ },
+ {
+ "epoch": 83.50953678474114,
+ "grad_norm": 1.4667587280273438,
+ "learning_rate": 1.3926560994033945e-06,
+ "loss": 0.0112,
+ "step": 30648
+ },
+ {
+ "epoch": 83.51226158038148,
+ "grad_norm": 0.566287100315094,
+ "learning_rate": 1.3922068952063107e-06,
+ "loss": 0.0094,
+ "step": 30649
+ },
+ {
+ "epoch": 83.5149863760218,
+ "grad_norm": 1.0660319328308105,
+ "learning_rate": 1.3917577580464414e-06,
+ "loss": 0.0374,
+ "step": 30650
+ },
+ {
+ "epoch": 83.51771117166213,
+ "grad_norm": 1.3219547271728516,
+ "learning_rate": 1.3913086879272874e-06,
+ "loss": 0.0702,
+ "step": 30651
+ },
+ {
+ "epoch": 83.52043596730245,
+ "grad_norm": 0.8276838660240173,
+ "learning_rate": 1.3908596848523436e-06,
+ "loss": 0.0104,
+ "step": 30652
+ },
+ {
+ "epoch": 83.52316076294278,
+ "grad_norm": 0.9595957398414612,
+ "learning_rate": 1.390410748825106e-06,
+ "loss": 0.0153,
+ "step": 30653
+ },
+ {
+ "epoch": 83.52588555858311,
+ "grad_norm": 1.0775631666183472,
+ "learning_rate": 1.389961879849072e-06,
+ "loss": 0.0596,
+ "step": 30654
+ },
+ {
+ "epoch": 83.52861035422343,
+ "grad_norm": 1.1107059717178345,
+ "learning_rate": 1.3895130779277398e-06,
+ "loss": 0.0219,
+ "step": 30655
+ },
+ {
+ "epoch": 83.53133514986376,
+ "grad_norm": 1.4137237071990967,
+ "learning_rate": 1.3890643430646033e-06,
+ "loss": 0.0225,
+ "step": 30656
+ },
+ {
+ "epoch": 83.53405994550408,
+ "grad_norm": 0.9403820037841797,
+ "learning_rate": 1.3886156752631563e-06,
+ "loss": 0.1006,
+ "step": 30657
+ },
+ {
+ "epoch": 83.53678474114442,
+ "grad_norm": 0.9737577438354492,
+ "learning_rate": 1.3881670745268904e-06,
+ "loss": 0.0493,
+ "step": 30658
+ },
+ {
+ "epoch": 83.53950953678473,
+ "grad_norm": 0.7082639932632446,
+ "learning_rate": 1.3877185408593052e-06,
+ "loss": 0.0081,
+ "step": 30659
+ },
+ {
+ "epoch": 83.54223433242507,
+ "grad_norm": 1.5191152095794678,
+ "learning_rate": 1.38727007426389e-06,
+ "loss": 0.0297,
+ "step": 30660
+ },
+ {
+ "epoch": 83.5449591280654,
+ "grad_norm": 0.8532187938690186,
+ "learning_rate": 1.3868216747441388e-06,
+ "loss": 0.0118,
+ "step": 30661
+ },
+ {
+ "epoch": 83.54768392370572,
+ "grad_norm": 1.1931378841400146,
+ "learning_rate": 1.386373342303542e-06,
+ "loss": 0.0279,
+ "step": 30662
+ },
+ {
+ "epoch": 83.55040871934605,
+ "grad_norm": 1.08328378200531,
+ "learning_rate": 1.3859250769455933e-06,
+ "loss": 0.012,
+ "step": 30663
+ },
+ {
+ "epoch": 83.55313351498637,
+ "grad_norm": 1.4465779066085815,
+ "learning_rate": 1.3854768786737826e-06,
+ "loss": 0.0074,
+ "step": 30664
+ },
+ {
+ "epoch": 83.5558583106267,
+ "grad_norm": 1.2843246459960938,
+ "learning_rate": 1.3850287474916013e-06,
+ "loss": 0.0164,
+ "step": 30665
+ },
+ {
+ "epoch": 83.55858310626704,
+ "grad_norm": 0.5580465197563171,
+ "learning_rate": 1.3845806834025365e-06,
+ "loss": 0.0055,
+ "step": 30666
+ },
+ {
+ "epoch": 83.56130790190736,
+ "grad_norm": 1.7980715036392212,
+ "learning_rate": 1.3841326864100823e-06,
+ "loss": 0.0223,
+ "step": 30667
+ },
+ {
+ "epoch": 83.56403269754769,
+ "grad_norm": 1.155578851699829,
+ "learning_rate": 1.3836847565177258e-06,
+ "loss": 0.0079,
+ "step": 30668
+ },
+ {
+ "epoch": 83.566757493188,
+ "grad_norm": 1.865431308746338,
+ "learning_rate": 1.3832368937289543e-06,
+ "loss": 0.0121,
+ "step": 30669
+ },
+ {
+ "epoch": 83.56948228882834,
+ "grad_norm": 3.0540099143981934,
+ "learning_rate": 1.3827890980472547e-06,
+ "loss": 0.0288,
+ "step": 30670
+ },
+ {
+ "epoch": 83.57220708446866,
+ "grad_norm": 1.5792523622512817,
+ "learning_rate": 1.3823413694761167e-06,
+ "loss": 0.024,
+ "step": 30671
+ },
+ {
+ "epoch": 83.57493188010899,
+ "grad_norm": 0.8009580373764038,
+ "learning_rate": 1.3818937080190287e-06,
+ "loss": 0.015,
+ "step": 30672
+ },
+ {
+ "epoch": 83.57765667574932,
+ "grad_norm": 1.4639389514923096,
+ "learning_rate": 1.3814461136794755e-06,
+ "loss": 0.0454,
+ "step": 30673
+ },
+ {
+ "epoch": 83.58038147138964,
+ "grad_norm": 1.7265641689300537,
+ "learning_rate": 1.3809985864609398e-06,
+ "loss": 0.0118,
+ "step": 30674
+ },
+ {
+ "epoch": 83.58310626702998,
+ "grad_norm": 1.4376230239868164,
+ "learning_rate": 1.3805511263669113e-06,
+ "loss": 0.0939,
+ "step": 30675
+ },
+ {
+ "epoch": 83.5858310626703,
+ "grad_norm": 1.314435362815857,
+ "learning_rate": 1.3801037334008738e-06,
+ "loss": 0.0187,
+ "step": 30676
+ },
+ {
+ "epoch": 83.58855585831063,
+ "grad_norm": 1.10556960105896,
+ "learning_rate": 1.379656407566311e-06,
+ "loss": 0.1274,
+ "step": 30677
+ },
+ {
+ "epoch": 83.59128065395096,
+ "grad_norm": 5.317464351654053,
+ "learning_rate": 1.3792091488667047e-06,
+ "loss": 0.014,
+ "step": 30678
+ },
+ {
+ "epoch": 83.59400544959128,
+ "grad_norm": 1.398157000541687,
+ "learning_rate": 1.3787619573055421e-06,
+ "loss": 0.0227,
+ "step": 30679
+ },
+ {
+ "epoch": 83.59673024523161,
+ "grad_norm": 0.8680464625358582,
+ "learning_rate": 1.378314832886305e-06,
+ "loss": 0.0092,
+ "step": 30680
+ },
+ {
+ "epoch": 83.59945504087193,
+ "grad_norm": 1.827495813369751,
+ "learning_rate": 1.3778677756124724e-06,
+ "loss": 0.064,
+ "step": 30681
+ },
+ {
+ "epoch": 83.60217983651226,
+ "grad_norm": 5.892185211181641,
+ "learning_rate": 1.3774207854875277e-06,
+ "loss": 0.0845,
+ "step": 30682
+ },
+ {
+ "epoch": 83.60490463215258,
+ "grad_norm": 1.2546207904815674,
+ "learning_rate": 1.3769738625149532e-06,
+ "loss": 0.1384,
+ "step": 30683
+ },
+ {
+ "epoch": 83.60762942779292,
+ "grad_norm": 0.7290654182434082,
+ "learning_rate": 1.3765270066982294e-06,
+ "loss": 0.0072,
+ "step": 30684
+ },
+ {
+ "epoch": 83.61035422343325,
+ "grad_norm": 1.1675654649734497,
+ "learning_rate": 1.3760802180408362e-06,
+ "loss": 0.0096,
+ "step": 30685
+ },
+ {
+ "epoch": 83.61307901907357,
+ "grad_norm": 1.0840063095092773,
+ "learning_rate": 1.3756334965462502e-06,
+ "loss": 0.0223,
+ "step": 30686
+ },
+ {
+ "epoch": 83.6158038147139,
+ "grad_norm": 1.1363621950149536,
+ "learning_rate": 1.3751868422179527e-06,
+ "loss": 0.0111,
+ "step": 30687
+ },
+ {
+ "epoch": 83.61852861035422,
+ "grad_norm": 1.3878223896026611,
+ "learning_rate": 1.3747402550594246e-06,
+ "loss": 0.0364,
+ "step": 30688
+ },
+ {
+ "epoch": 83.62125340599455,
+ "grad_norm": 0.3959164023399353,
+ "learning_rate": 1.3742937350741414e-06,
+ "loss": 0.0038,
+ "step": 30689
+ },
+ {
+ "epoch": 83.62397820163488,
+ "grad_norm": 0.5204312801361084,
+ "learning_rate": 1.3738472822655824e-06,
+ "loss": 0.0052,
+ "step": 30690
+ },
+ {
+ "epoch": 83.6267029972752,
+ "grad_norm": 0.8150025606155396,
+ "learning_rate": 1.3734008966372192e-06,
+ "loss": 0.0092,
+ "step": 30691
+ },
+ {
+ "epoch": 83.62942779291554,
+ "grad_norm": 0.9421283602714539,
+ "learning_rate": 1.3729545781925357e-06,
+ "loss": 0.0085,
+ "step": 30692
+ },
+ {
+ "epoch": 83.63215258855585,
+ "grad_norm": 1.3234155178070068,
+ "learning_rate": 1.3725083269350038e-06,
+ "loss": 0.0274,
+ "step": 30693
+ },
+ {
+ "epoch": 83.63487738419619,
+ "grad_norm": 0.5683311820030212,
+ "learning_rate": 1.3720621428680991e-06,
+ "loss": 0.0059,
+ "step": 30694
+ },
+ {
+ "epoch": 83.6376021798365,
+ "grad_norm": 1.9808571338653564,
+ "learning_rate": 1.3716160259952948e-06,
+ "loss": 0.0482,
+ "step": 30695
+ },
+ {
+ "epoch": 83.64032697547684,
+ "grad_norm": 0.6266232132911682,
+ "learning_rate": 1.3711699763200692e-06,
+ "loss": 0.0035,
+ "step": 30696
+ },
+ {
+ "epoch": 83.64305177111717,
+ "grad_norm": 1.3238823413848877,
+ "learning_rate": 1.370723993845895e-06,
+ "loss": 0.1169,
+ "step": 30697
+ },
+ {
+ "epoch": 83.64577656675749,
+ "grad_norm": 0.9889927506446838,
+ "learning_rate": 1.3702780785762415e-06,
+ "loss": 0.0102,
+ "step": 30698
+ },
+ {
+ "epoch": 83.64850136239782,
+ "grad_norm": 0.9844257235527039,
+ "learning_rate": 1.3698322305145871e-06,
+ "loss": 0.0141,
+ "step": 30699
+ },
+ {
+ "epoch": 83.65122615803814,
+ "grad_norm": 1.4150974750518799,
+ "learning_rate": 1.3693864496644017e-06,
+ "loss": 0.0444,
+ "step": 30700
+ },
+ {
+ "epoch": 83.65395095367847,
+ "grad_norm": 0.9742388129234314,
+ "learning_rate": 1.3689407360291562e-06,
+ "loss": 0.0243,
+ "step": 30701
+ },
+ {
+ "epoch": 83.65667574931881,
+ "grad_norm": 2.4040985107421875,
+ "learning_rate": 1.3684950896123205e-06,
+ "loss": 0.1593,
+ "step": 30702
+ },
+ {
+ "epoch": 83.65940054495913,
+ "grad_norm": 0.9630717039108276,
+ "learning_rate": 1.3680495104173697e-06,
+ "loss": 0.082,
+ "step": 30703
+ },
+ {
+ "epoch": 83.66212534059946,
+ "grad_norm": 0.8549289703369141,
+ "learning_rate": 1.3676039984477684e-06,
+ "loss": 0.0106,
+ "step": 30704
+ },
+ {
+ "epoch": 83.66485013623978,
+ "grad_norm": 0.6711997389793396,
+ "learning_rate": 1.3671585537069931e-06,
+ "loss": 0.0084,
+ "step": 30705
+ },
+ {
+ "epoch": 83.66757493188011,
+ "grad_norm": 0.9243277311325073,
+ "learning_rate": 1.3667131761985076e-06,
+ "loss": 0.0124,
+ "step": 30706
+ },
+ {
+ "epoch": 83.67029972752043,
+ "grad_norm": 1.0384763479232788,
+ "learning_rate": 1.3662678659257811e-06,
+ "loss": 0.0138,
+ "step": 30707
+ },
+ {
+ "epoch": 83.67302452316076,
+ "grad_norm": 1.0279492139816284,
+ "learning_rate": 1.3658226228922844e-06,
+ "loss": 0.0356,
+ "step": 30708
+ },
+ {
+ "epoch": 83.6757493188011,
+ "grad_norm": 1.2607074975967407,
+ "learning_rate": 1.3653774471014825e-06,
+ "loss": 0.0227,
+ "step": 30709
+ },
+ {
+ "epoch": 83.67847411444141,
+ "grad_norm": 0.4489002525806427,
+ "learning_rate": 1.3649323385568435e-06,
+ "loss": 0.0048,
+ "step": 30710
+ },
+ {
+ "epoch": 83.68119891008175,
+ "grad_norm": 0.49523648619651794,
+ "learning_rate": 1.3644872972618316e-06,
+ "loss": 0.0044,
+ "step": 30711
+ },
+ {
+ "epoch": 83.68392370572207,
+ "grad_norm": 1.292623519897461,
+ "learning_rate": 1.3640423232199174e-06,
+ "loss": 0.0828,
+ "step": 30712
+ },
+ {
+ "epoch": 83.6866485013624,
+ "grad_norm": 1.6316665410995483,
+ "learning_rate": 1.3635974164345634e-06,
+ "loss": 0.0157,
+ "step": 30713
+ },
+ {
+ "epoch": 83.68937329700273,
+ "grad_norm": 1.0687549114227295,
+ "learning_rate": 1.3631525769092336e-06,
+ "loss": 0.0115,
+ "step": 30714
+ },
+ {
+ "epoch": 83.69209809264305,
+ "grad_norm": 1.1130869388580322,
+ "learning_rate": 1.362707804647393e-06,
+ "loss": 0.0715,
+ "step": 30715
+ },
+ {
+ "epoch": 83.69482288828338,
+ "grad_norm": 1.2543542385101318,
+ "learning_rate": 1.3622630996525067e-06,
+ "loss": 0.0711,
+ "step": 30716
+ },
+ {
+ "epoch": 83.6975476839237,
+ "grad_norm": 1.2879440784454346,
+ "learning_rate": 1.3618184619280383e-06,
+ "loss": 0.063,
+ "step": 30717
+ },
+ {
+ "epoch": 83.70027247956403,
+ "grad_norm": 0.8302642107009888,
+ "learning_rate": 1.3613738914774477e-06,
+ "loss": 0.0163,
+ "step": 30718
+ },
+ {
+ "epoch": 83.70299727520435,
+ "grad_norm": 1.2733657360076904,
+ "learning_rate": 1.3609293883042007e-06,
+ "loss": 0.0233,
+ "step": 30719
+ },
+ {
+ "epoch": 83.70572207084469,
+ "grad_norm": 1.526957392692566,
+ "learning_rate": 1.360484952411758e-06,
+ "loss": 0.0413,
+ "step": 30720
+ },
+ {
+ "epoch": 83.70844686648502,
+ "grad_norm": 1.2614229917526245,
+ "learning_rate": 1.3600405838035779e-06,
+ "loss": 0.0103,
+ "step": 30721
+ },
+ {
+ "epoch": 83.71117166212534,
+ "grad_norm": 0.9663424491882324,
+ "learning_rate": 1.3595962824831255e-06,
+ "loss": 0.0536,
+ "step": 30722
+ },
+ {
+ "epoch": 83.71389645776567,
+ "grad_norm": 1.214935302734375,
+ "learning_rate": 1.3591520484538578e-06,
+ "loss": 0.015,
+ "step": 30723
+ },
+ {
+ "epoch": 83.71662125340599,
+ "grad_norm": 1.0371240377426147,
+ "learning_rate": 1.3587078817192378e-06,
+ "loss": 0.0176,
+ "step": 30724
+ },
+ {
+ "epoch": 83.71934604904632,
+ "grad_norm": 1.5638933181762695,
+ "learning_rate": 1.3582637822827228e-06,
+ "loss": 0.032,
+ "step": 30725
+ },
+ {
+ "epoch": 83.72207084468666,
+ "grad_norm": 0.4691667854785919,
+ "learning_rate": 1.3578197501477708e-06,
+ "loss": 0.004,
+ "step": 30726
+ },
+ {
+ "epoch": 83.72479564032697,
+ "grad_norm": 1.880297064781189,
+ "learning_rate": 1.3573757853178392e-06,
+ "loss": 0.0346,
+ "step": 30727
+ },
+ {
+ "epoch": 83.7275204359673,
+ "grad_norm": 2.0071115493774414,
+ "learning_rate": 1.3569318877963878e-06,
+ "loss": 0.1105,
+ "step": 30728
+ },
+ {
+ "epoch": 83.73024523160763,
+ "grad_norm": 0.9602733850479126,
+ "learning_rate": 1.3564880575868734e-06,
+ "loss": 0.0111,
+ "step": 30729
+ },
+ {
+ "epoch": 83.73297002724796,
+ "grad_norm": 0.6192739605903625,
+ "learning_rate": 1.3560442946927522e-06,
+ "loss": 0.007,
+ "step": 30730
+ },
+ {
+ "epoch": 83.73569482288828,
+ "grad_norm": 1.7615811824798584,
+ "learning_rate": 1.355600599117477e-06,
+ "loss": 0.0147,
+ "step": 30731
+ },
+ {
+ "epoch": 83.73841961852861,
+ "grad_norm": 1.0219354629516602,
+ "learning_rate": 1.3551569708645084e-06,
+ "loss": 0.0717,
+ "step": 30732
+ },
+ {
+ "epoch": 83.74114441416894,
+ "grad_norm": 0.8655136227607727,
+ "learning_rate": 1.3547134099372993e-06,
+ "loss": 0.0131,
+ "step": 30733
+ },
+ {
+ "epoch": 83.74386920980926,
+ "grad_norm": 0.9533057808876038,
+ "learning_rate": 1.3542699163393036e-06,
+ "loss": 0.0113,
+ "step": 30734
+ },
+ {
+ "epoch": 83.7465940054496,
+ "grad_norm": 1.660105586051941,
+ "learning_rate": 1.353826490073974e-06,
+ "loss": 0.0174,
+ "step": 30735
+ },
+ {
+ "epoch": 83.74931880108991,
+ "grad_norm": 1.1927332878112793,
+ "learning_rate": 1.3533831311447665e-06,
+ "loss": 0.0096,
+ "step": 30736
+ },
+ {
+ "epoch": 83.75204359673025,
+ "grad_norm": 0.8393914699554443,
+ "learning_rate": 1.3529398395551342e-06,
+ "loss": 0.0052,
+ "step": 30737
+ },
+ {
+ "epoch": 83.75476839237058,
+ "grad_norm": 1.301459789276123,
+ "learning_rate": 1.3524966153085272e-06,
+ "loss": 0.0241,
+ "step": 30738
+ },
+ {
+ "epoch": 83.7574931880109,
+ "grad_norm": 1.0570746660232544,
+ "learning_rate": 1.3520534584083932e-06,
+ "loss": 0.0089,
+ "step": 30739
+ },
+ {
+ "epoch": 83.76021798365123,
+ "grad_norm": 1.1070303916931152,
+ "learning_rate": 1.351610368858194e-06,
+ "loss": 0.0129,
+ "step": 30740
+ },
+ {
+ "epoch": 83.76294277929155,
+ "grad_norm": 1.2401663064956665,
+ "learning_rate": 1.351167346661375e-06,
+ "loss": 0.0199,
+ "step": 30741
+ },
+ {
+ "epoch": 83.76566757493188,
+ "grad_norm": 0.68219393491745,
+ "learning_rate": 1.3507243918213864e-06,
+ "loss": 0.0124,
+ "step": 30742
+ },
+ {
+ "epoch": 83.7683923705722,
+ "grad_norm": 1.7290319204330444,
+ "learning_rate": 1.3502815043416751e-06,
+ "loss": 0.0426,
+ "step": 30743
+ },
+ {
+ "epoch": 83.77111716621253,
+ "grad_norm": 1.3349156379699707,
+ "learning_rate": 1.349838684225695e-06,
+ "loss": 0.0315,
+ "step": 30744
+ },
+ {
+ "epoch": 83.77384196185287,
+ "grad_norm": 1.287969946861267,
+ "learning_rate": 1.3493959314768934e-06,
+ "loss": 0.0629,
+ "step": 30745
+ },
+ {
+ "epoch": 83.77656675749319,
+ "grad_norm": 0.8955380916595459,
+ "learning_rate": 1.3489532460987176e-06,
+ "loss": 0.0107,
+ "step": 30746
+ },
+ {
+ "epoch": 83.77929155313352,
+ "grad_norm": 0.6854398846626282,
+ "learning_rate": 1.3485106280946137e-06,
+ "loss": 0.0058,
+ "step": 30747
+ },
+ {
+ "epoch": 83.78201634877384,
+ "grad_norm": 1.401458501815796,
+ "learning_rate": 1.348068077468032e-06,
+ "loss": 0.0197,
+ "step": 30748
+ },
+ {
+ "epoch": 83.78474114441417,
+ "grad_norm": 1.6973650455474854,
+ "learning_rate": 1.3476255942224193e-06,
+ "loss": 0.1111,
+ "step": 30749
+ },
+ {
+ "epoch": 83.7874659400545,
+ "grad_norm": 0.41217395663261414,
+ "learning_rate": 1.3471831783612189e-06,
+ "loss": 0.0048,
+ "step": 30750
+ },
+ {
+ "epoch": 83.79019073569482,
+ "grad_norm": 0.8036642074584961,
+ "learning_rate": 1.3467408298878747e-06,
+ "loss": 0.0085,
+ "step": 30751
+ },
+ {
+ "epoch": 83.79291553133515,
+ "grad_norm": 1.2861453294754028,
+ "learning_rate": 1.3462985488058366e-06,
+ "loss": 0.0585,
+ "step": 30752
+ },
+ {
+ "epoch": 83.79564032697547,
+ "grad_norm": 1.5455809831619263,
+ "learning_rate": 1.3458563351185482e-06,
+ "loss": 0.023,
+ "step": 30753
+ },
+ {
+ "epoch": 83.7983651226158,
+ "grad_norm": 0.58194899559021,
+ "learning_rate": 1.3454141888294514e-06,
+ "loss": 0.0083,
+ "step": 30754
+ },
+ {
+ "epoch": 83.80108991825612,
+ "grad_norm": 0.8223621845245361,
+ "learning_rate": 1.3449721099419876e-06,
+ "loss": 0.0101,
+ "step": 30755
+ },
+ {
+ "epoch": 83.80381471389646,
+ "grad_norm": 1.4943281412124634,
+ "learning_rate": 1.3445300984596032e-06,
+ "loss": 0.0103,
+ "step": 30756
+ },
+ {
+ "epoch": 83.80653950953679,
+ "grad_norm": 1.4161689281463623,
+ "learning_rate": 1.344088154385741e-06,
+ "loss": 0.016,
+ "step": 30757
+ },
+ {
+ "epoch": 83.80926430517711,
+ "grad_norm": 0.6644172668457031,
+ "learning_rate": 1.3436462777238424e-06,
+ "loss": 0.008,
+ "step": 30758
+ },
+ {
+ "epoch": 83.81198910081744,
+ "grad_norm": 1.276590347290039,
+ "learning_rate": 1.343204468477346e-06,
+ "loss": 0.0653,
+ "step": 30759
+ },
+ {
+ "epoch": 83.81471389645776,
+ "grad_norm": 1.527084231376648,
+ "learning_rate": 1.3427627266496967e-06,
+ "loss": 0.0309,
+ "step": 30760
+ },
+ {
+ "epoch": 83.8174386920981,
+ "grad_norm": 1.4429482221603394,
+ "learning_rate": 1.342321052244333e-06,
+ "loss": 0.0146,
+ "step": 30761
+ },
+ {
+ "epoch": 83.82016348773843,
+ "grad_norm": 2.0829193592071533,
+ "learning_rate": 1.3418794452646932e-06,
+ "loss": 0.0099,
+ "step": 30762
+ },
+ {
+ "epoch": 83.82288828337875,
+ "grad_norm": 0.9206532835960388,
+ "learning_rate": 1.3414379057142158e-06,
+ "loss": 0.0964,
+ "step": 30763
+ },
+ {
+ "epoch": 83.82561307901908,
+ "grad_norm": 0.8828400373458862,
+ "learning_rate": 1.3409964335963443e-06,
+ "loss": 0.0063,
+ "step": 30764
+ },
+ {
+ "epoch": 83.8283378746594,
+ "grad_norm": 1.2684894800186157,
+ "learning_rate": 1.3405550289145131e-06,
+ "loss": 0.0279,
+ "step": 30765
+ },
+ {
+ "epoch": 83.83106267029973,
+ "grad_norm": 3.3268215656280518,
+ "learning_rate": 1.3401136916721614e-06,
+ "loss": 0.0221,
+ "step": 30766
+ },
+ {
+ "epoch": 83.83378746594005,
+ "grad_norm": 0.5751216411590576,
+ "learning_rate": 1.339672421872722e-06,
+ "loss": 0.0053,
+ "step": 30767
+ },
+ {
+ "epoch": 83.83651226158038,
+ "grad_norm": 1.380770206451416,
+ "learning_rate": 1.339231219519639e-06,
+ "loss": 0.0588,
+ "step": 30768
+ },
+ {
+ "epoch": 83.83923705722071,
+ "grad_norm": 1.761250376701355,
+ "learning_rate": 1.3387900846163427e-06,
+ "loss": 0.0847,
+ "step": 30769
+ },
+ {
+ "epoch": 83.84196185286103,
+ "grad_norm": 1.1569592952728271,
+ "learning_rate": 1.3383490171662717e-06,
+ "loss": 0.018,
+ "step": 30770
+ },
+ {
+ "epoch": 83.84468664850137,
+ "grad_norm": 1.0305577516555786,
+ "learning_rate": 1.3379080171728575e-06,
+ "loss": 0.0175,
+ "step": 30771
+ },
+ {
+ "epoch": 83.84741144414168,
+ "grad_norm": 0.7519474625587463,
+ "learning_rate": 1.3374670846395388e-06,
+ "loss": 0.0074,
+ "step": 30772
+ },
+ {
+ "epoch": 83.85013623978202,
+ "grad_norm": 0.9935907125473022,
+ "learning_rate": 1.3370262195697449e-06,
+ "loss": 0.0203,
+ "step": 30773
+ },
+ {
+ "epoch": 83.85286103542235,
+ "grad_norm": 0.9290697574615479,
+ "learning_rate": 1.336585421966915e-06,
+ "loss": 0.0099,
+ "step": 30774
+ },
+ {
+ "epoch": 83.85558583106267,
+ "grad_norm": 0.9526328444480896,
+ "learning_rate": 1.3361446918344789e-06,
+ "loss": 0.0089,
+ "step": 30775
+ },
+ {
+ "epoch": 83.858310626703,
+ "grad_norm": 1.209380030632019,
+ "learning_rate": 1.3357040291758661e-06,
+ "loss": 0.0066,
+ "step": 30776
+ },
+ {
+ "epoch": 83.86103542234332,
+ "grad_norm": 1.391644835472107,
+ "learning_rate": 1.3352634339945147e-06,
+ "loss": 0.0319,
+ "step": 30777
+ },
+ {
+ "epoch": 83.86376021798365,
+ "grad_norm": 1.551396369934082,
+ "learning_rate": 1.3348229062938522e-06,
+ "loss": 0.0103,
+ "step": 30778
+ },
+ {
+ "epoch": 83.86648501362397,
+ "grad_norm": 1.6448683738708496,
+ "learning_rate": 1.3343824460773103e-06,
+ "loss": 0.0675,
+ "step": 30779
+ },
+ {
+ "epoch": 83.8692098092643,
+ "grad_norm": 1.1304501295089722,
+ "learning_rate": 1.3339420533483171e-06,
+ "loss": 0.0111,
+ "step": 30780
+ },
+ {
+ "epoch": 83.87193460490464,
+ "grad_norm": 1.4436684846878052,
+ "learning_rate": 1.3335017281103059e-06,
+ "loss": 0.0722,
+ "step": 30781
+ },
+ {
+ "epoch": 83.87465940054496,
+ "grad_norm": 0.7314038872718811,
+ "learning_rate": 1.3330614703667045e-06,
+ "loss": 0.0123,
+ "step": 30782
+ },
+ {
+ "epoch": 83.87738419618529,
+ "grad_norm": 1.2300742864608765,
+ "learning_rate": 1.3326212801209392e-06,
+ "loss": 0.01,
+ "step": 30783
+ },
+ {
+ "epoch": 83.88010899182561,
+ "grad_norm": 0.8397956490516663,
+ "learning_rate": 1.332181157376442e-06,
+ "loss": 0.0094,
+ "step": 30784
+ },
+ {
+ "epoch": 83.88283378746594,
+ "grad_norm": 1.1643950939178467,
+ "learning_rate": 1.33174110213664e-06,
+ "loss": 0.0222,
+ "step": 30785
+ },
+ {
+ "epoch": 83.88555858310627,
+ "grad_norm": 0.998520016670227,
+ "learning_rate": 1.3313011144049593e-06,
+ "loss": 0.0153,
+ "step": 30786
+ },
+ {
+ "epoch": 83.88828337874659,
+ "grad_norm": 0.9324813485145569,
+ "learning_rate": 1.3308611941848237e-06,
+ "loss": 0.0097,
+ "step": 30787
+ },
+ {
+ "epoch": 83.89100817438693,
+ "grad_norm": 0.845969021320343,
+ "learning_rate": 1.330421341479664e-06,
+ "loss": 0.0101,
+ "step": 30788
+ },
+ {
+ "epoch": 83.89373297002724,
+ "grad_norm": 0.7029719948768616,
+ "learning_rate": 1.3299815562929052e-06,
+ "loss": 0.0082,
+ "step": 30789
+ },
+ {
+ "epoch": 83.89645776566758,
+ "grad_norm": 0.7671523690223694,
+ "learning_rate": 1.3295418386279679e-06,
+ "loss": 0.008,
+ "step": 30790
+ },
+ {
+ "epoch": 83.8991825613079,
+ "grad_norm": 0.7680314183235168,
+ "learning_rate": 1.3291021884882815e-06,
+ "loss": 0.0049,
+ "step": 30791
+ },
+ {
+ "epoch": 83.90190735694823,
+ "grad_norm": 1.3751614093780518,
+ "learning_rate": 1.3286626058772668e-06,
+ "loss": 0.008,
+ "step": 30792
+ },
+ {
+ "epoch": 83.90463215258856,
+ "grad_norm": 1.13052237033844,
+ "learning_rate": 1.3282230907983507e-06,
+ "loss": 0.0447,
+ "step": 30793
+ },
+ {
+ "epoch": 83.90735694822888,
+ "grad_norm": 1.0547924041748047,
+ "learning_rate": 1.327783643254954e-06,
+ "loss": 0.1079,
+ "step": 30794
+ },
+ {
+ "epoch": 83.91008174386921,
+ "grad_norm": 1.002150297164917,
+ "learning_rate": 1.3273442632504985e-06,
+ "loss": 0.029,
+ "step": 30795
+ },
+ {
+ "epoch": 83.91280653950953,
+ "grad_norm": 1.9914507865905762,
+ "learning_rate": 1.3269049507884057e-06,
+ "loss": 0.1097,
+ "step": 30796
+ },
+ {
+ "epoch": 83.91553133514986,
+ "grad_norm": 1.3559038639068604,
+ "learning_rate": 1.3264657058720998e-06,
+ "loss": 0.054,
+ "step": 30797
+ },
+ {
+ "epoch": 83.9182561307902,
+ "grad_norm": 1.453914761543274,
+ "learning_rate": 1.326026528504999e-06,
+ "loss": 0.1135,
+ "step": 30798
+ },
+ {
+ "epoch": 83.92098092643052,
+ "grad_norm": 1.4384552240371704,
+ "learning_rate": 1.325587418690525e-06,
+ "loss": 0.0549,
+ "step": 30799
+ },
+ {
+ "epoch": 83.92370572207085,
+ "grad_norm": 1.2425554990768433,
+ "learning_rate": 1.325148376432095e-06,
+ "loss": 0.0226,
+ "step": 30800
+ },
+ {
+ "epoch": 83.92643051771117,
+ "grad_norm": 1.112921118736267,
+ "learning_rate": 1.324709401733133e-06,
+ "loss": 0.0046,
+ "step": 30801
+ },
+ {
+ "epoch": 83.9291553133515,
+ "grad_norm": 1.094912052154541,
+ "learning_rate": 1.3242704945970541e-06,
+ "loss": 0.0055,
+ "step": 30802
+ },
+ {
+ "epoch": 83.93188010899182,
+ "grad_norm": 2.1306326389312744,
+ "learning_rate": 1.3238316550272767e-06,
+ "loss": 0.015,
+ "step": 30803
+ },
+ {
+ "epoch": 83.93460490463215,
+ "grad_norm": 1.1349595785140991,
+ "learning_rate": 1.323392883027218e-06,
+ "loss": 0.0119,
+ "step": 30804
+ },
+ {
+ "epoch": 83.93732970027249,
+ "grad_norm": 1.4025019407272339,
+ "learning_rate": 1.3229541786002987e-06,
+ "loss": 0.0555,
+ "step": 30805
+ },
+ {
+ "epoch": 83.9400544959128,
+ "grad_norm": 1.275890827178955,
+ "learning_rate": 1.3225155417499325e-06,
+ "loss": 0.0563,
+ "step": 30806
+ },
+ {
+ "epoch": 83.94277929155314,
+ "grad_norm": 1.2282865047454834,
+ "learning_rate": 1.3220769724795334e-06,
+ "loss": 0.0276,
+ "step": 30807
+ },
+ {
+ "epoch": 83.94550408719346,
+ "grad_norm": 1.456722617149353,
+ "learning_rate": 1.32163847079252e-06,
+ "loss": 0.0775,
+ "step": 30808
+ },
+ {
+ "epoch": 83.94822888283379,
+ "grad_norm": 1.4049814939498901,
+ "learning_rate": 1.3212000366923095e-06,
+ "loss": 0.0304,
+ "step": 30809
+ },
+ {
+ "epoch": 83.95095367847412,
+ "grad_norm": 0.8106567859649658,
+ "learning_rate": 1.3207616701823135e-06,
+ "loss": 0.0079,
+ "step": 30810
+ },
+ {
+ "epoch": 83.95367847411444,
+ "grad_norm": 2.161699056625366,
+ "learning_rate": 1.3203233712659457e-06,
+ "loss": 0.0392,
+ "step": 30811
+ },
+ {
+ "epoch": 83.95640326975477,
+ "grad_norm": 1.9799714088439941,
+ "learning_rate": 1.3198851399466184e-06,
+ "loss": 0.0444,
+ "step": 30812
+ },
+ {
+ "epoch": 83.95912806539509,
+ "grad_norm": 1.245556116104126,
+ "learning_rate": 1.3194469762277485e-06,
+ "loss": 0.0228,
+ "step": 30813
+ },
+ {
+ "epoch": 83.96185286103542,
+ "grad_norm": 1.1889595985412598,
+ "learning_rate": 1.3190088801127454e-06,
+ "loss": 0.0152,
+ "step": 30814
+ },
+ {
+ "epoch": 83.96457765667574,
+ "grad_norm": 0.6649832129478455,
+ "learning_rate": 1.3185708516050233e-06,
+ "loss": 0.007,
+ "step": 30815
+ },
+ {
+ "epoch": 83.96730245231608,
+ "grad_norm": 0.9640118479728699,
+ "learning_rate": 1.3181328907079883e-06,
+ "loss": 0.0077,
+ "step": 30816
+ },
+ {
+ "epoch": 83.97002724795641,
+ "grad_norm": 0.8255684971809387,
+ "learning_rate": 1.3176949974250575e-06,
+ "loss": 0.0059,
+ "step": 30817
+ },
+ {
+ "epoch": 83.97275204359673,
+ "grad_norm": 0.952425479888916,
+ "learning_rate": 1.3172571717596393e-06,
+ "loss": 0.0146,
+ "step": 30818
+ },
+ {
+ "epoch": 83.97547683923706,
+ "grad_norm": 0.7477033138275146,
+ "learning_rate": 1.3168194137151423e-06,
+ "loss": 0.0064,
+ "step": 30819
+ },
+ {
+ "epoch": 83.97820163487738,
+ "grad_norm": 1.600587010383606,
+ "learning_rate": 1.3163817232949738e-06,
+ "loss": 0.0141,
+ "step": 30820
+ },
+ {
+ "epoch": 83.98092643051771,
+ "grad_norm": 1.521441102027893,
+ "learning_rate": 1.3159441005025475e-06,
+ "loss": 0.0448,
+ "step": 30821
+ },
+ {
+ "epoch": 83.98365122615803,
+ "grad_norm": 1.350489854812622,
+ "learning_rate": 1.3155065453412696e-06,
+ "loss": 0.0369,
+ "step": 30822
+ },
+ {
+ "epoch": 83.98637602179836,
+ "grad_norm": 1.6784510612487793,
+ "learning_rate": 1.3150690578145465e-06,
+ "loss": 0.0706,
+ "step": 30823
+ },
+ {
+ "epoch": 83.9891008174387,
+ "grad_norm": 1.0165907144546509,
+ "learning_rate": 1.314631637925784e-06,
+ "loss": 0.0114,
+ "step": 30824
+ },
+ {
+ "epoch": 83.99182561307902,
+ "grad_norm": 1.3109904527664185,
+ "learning_rate": 1.3141942856783908e-06,
+ "loss": 0.0198,
+ "step": 30825
+ },
+ {
+ "epoch": 83.99455040871935,
+ "grad_norm": 0.8300231695175171,
+ "learning_rate": 1.313757001075775e-06,
+ "loss": 0.0616,
+ "step": 30826
+ },
+ {
+ "epoch": 83.99727520435967,
+ "grad_norm": 1.2711458206176758,
+ "learning_rate": 1.3133197841213396e-06,
+ "loss": 0.1147,
+ "step": 30827
+ },
+ {
+ "epoch": 84.0,
+ "grad_norm": 1.170265793800354,
+ "learning_rate": 1.3128826348184886e-06,
+ "loss": 0.0216,
+ "step": 30828
+ },
+ {
+ "epoch": 84.00272479564033,
+ "grad_norm": 1.1386185884475708,
+ "learning_rate": 1.3124455531706304e-06,
+ "loss": 0.0223,
+ "step": 30829
+ },
+ {
+ "epoch": 84.00544959128065,
+ "grad_norm": 1.56293523311615,
+ "learning_rate": 1.3120085391811666e-06,
+ "loss": 0.0112,
+ "step": 30830
+ },
+ {
+ "epoch": 84.00817438692098,
+ "grad_norm": 0.9350024461746216,
+ "learning_rate": 1.3115715928535e-06,
+ "loss": 0.0083,
+ "step": 30831
+ },
+ {
+ "epoch": 84.0108991825613,
+ "grad_norm": 0.9045727252960205,
+ "learning_rate": 1.3111347141910324e-06,
+ "loss": 0.0088,
+ "step": 30832
+ },
+ {
+ "epoch": 84.01362397820164,
+ "grad_norm": 1.501177191734314,
+ "learning_rate": 1.3106979031971701e-06,
+ "loss": 0.015,
+ "step": 30833
+ },
+ {
+ "epoch": 84.01634877384195,
+ "grad_norm": 0.9568030834197998,
+ "learning_rate": 1.3102611598753135e-06,
+ "loss": 0.1471,
+ "step": 30834
+ },
+ {
+ "epoch": 84.01907356948229,
+ "grad_norm": 0.8061081767082214,
+ "learning_rate": 1.3098244842288622e-06,
+ "loss": 0.0064,
+ "step": 30835
+ },
+ {
+ "epoch": 84.02179836512262,
+ "grad_norm": 1.6839494705200195,
+ "learning_rate": 1.3093878762612167e-06,
+ "loss": 0.0351,
+ "step": 30836
+ },
+ {
+ "epoch": 84.02452316076294,
+ "grad_norm": 1.5558909177780151,
+ "learning_rate": 1.308951335975781e-06,
+ "loss": 0.031,
+ "step": 30837
+ },
+ {
+ "epoch": 84.02724795640327,
+ "grad_norm": 8.156891822814941,
+ "learning_rate": 1.3085148633759525e-06,
+ "loss": 0.0624,
+ "step": 30838
+ },
+ {
+ "epoch": 84.02997275204359,
+ "grad_norm": 0.9118248224258423,
+ "learning_rate": 1.3080784584651308e-06,
+ "loss": 0.0079,
+ "step": 30839
+ },
+ {
+ "epoch": 84.03269754768392,
+ "grad_norm": 1.1634347438812256,
+ "learning_rate": 1.3076421212467117e-06,
+ "loss": 0.0598,
+ "step": 30840
+ },
+ {
+ "epoch": 84.03542234332426,
+ "grad_norm": 1.6082981824874878,
+ "learning_rate": 1.307205851724097e-06,
+ "loss": 0.0241,
+ "step": 30841
+ },
+ {
+ "epoch": 84.03814713896458,
+ "grad_norm": 0.8647699952125549,
+ "learning_rate": 1.3067696499006832e-06,
+ "loss": 0.0097,
+ "step": 30842
+ },
+ {
+ "epoch": 84.04087193460491,
+ "grad_norm": 1.3919141292572021,
+ "learning_rate": 1.3063335157798685e-06,
+ "loss": 0.017,
+ "step": 30843
+ },
+ {
+ "epoch": 84.04359673024523,
+ "grad_norm": 1.9519574642181396,
+ "learning_rate": 1.305897449365049e-06,
+ "loss": 0.0235,
+ "step": 30844
+ },
+ {
+ "epoch": 84.04632152588556,
+ "grad_norm": 1.0781115293502808,
+ "learning_rate": 1.3054614506596186e-06,
+ "loss": 0.0078,
+ "step": 30845
+ },
+ {
+ "epoch": 84.04904632152588,
+ "grad_norm": 0.8580396175384521,
+ "learning_rate": 1.305025519666976e-06,
+ "loss": 0.007,
+ "step": 30846
+ },
+ {
+ "epoch": 84.05177111716621,
+ "grad_norm": 1.3840922117233276,
+ "learning_rate": 1.3045896563905147e-06,
+ "loss": 0.051,
+ "step": 30847
+ },
+ {
+ "epoch": 84.05449591280654,
+ "grad_norm": 1.1533639430999756,
+ "learning_rate": 1.3041538608336279e-06,
+ "loss": 0.0261,
+ "step": 30848
+ },
+ {
+ "epoch": 84.05722070844686,
+ "grad_norm": 1.7809356451034546,
+ "learning_rate": 1.3037181329997128e-06,
+ "loss": 0.0634,
+ "step": 30849
+ },
+ {
+ "epoch": 84.0599455040872,
+ "grad_norm": 0.6493775844573975,
+ "learning_rate": 1.3032824728921612e-06,
+ "loss": 0.0087,
+ "step": 30850
+ },
+ {
+ "epoch": 84.06267029972751,
+ "grad_norm": 1.6429719924926758,
+ "learning_rate": 1.3028468805143657e-06,
+ "loss": 0.0292,
+ "step": 30851
+ },
+ {
+ "epoch": 84.06539509536785,
+ "grad_norm": 1.271653175354004,
+ "learning_rate": 1.302411355869716e-06,
+ "loss": 0.0104,
+ "step": 30852
+ },
+ {
+ "epoch": 84.06811989100818,
+ "grad_norm": 1.938011884689331,
+ "learning_rate": 1.3019758989616094e-06,
+ "loss": 0.1023,
+ "step": 30853
+ },
+ {
+ "epoch": 84.0708446866485,
+ "grad_norm": 0.9869240522384644,
+ "learning_rate": 1.3015405097934342e-06,
+ "loss": 0.0128,
+ "step": 30854
+ },
+ {
+ "epoch": 84.07356948228883,
+ "grad_norm": 1.992308497428894,
+ "learning_rate": 1.301105188368582e-06,
+ "loss": 0.0413,
+ "step": 30855
+ },
+ {
+ "epoch": 84.07629427792915,
+ "grad_norm": 1.357407808303833,
+ "learning_rate": 1.3006699346904406e-06,
+ "loss": 0.0228,
+ "step": 30856
+ },
+ {
+ "epoch": 84.07901907356948,
+ "grad_norm": 1.598831057548523,
+ "learning_rate": 1.3002347487624035e-06,
+ "loss": 0.0534,
+ "step": 30857
+ },
+ {
+ "epoch": 84.0817438692098,
+ "grad_norm": 0.8016869425773621,
+ "learning_rate": 1.299799630587858e-06,
+ "loss": 0.0107,
+ "step": 30858
+ },
+ {
+ "epoch": 84.08446866485014,
+ "grad_norm": 1.0035064220428467,
+ "learning_rate": 1.2993645801701914e-06,
+ "loss": 0.0128,
+ "step": 30859
+ },
+ {
+ "epoch": 84.08719346049047,
+ "grad_norm": 0.8671107292175293,
+ "learning_rate": 1.2989295975127957e-06,
+ "loss": 0.0133,
+ "step": 30860
+ },
+ {
+ "epoch": 84.08991825613079,
+ "grad_norm": 1.4894543886184692,
+ "learning_rate": 1.2984946826190537e-06,
+ "loss": 0.0103,
+ "step": 30861
+ },
+ {
+ "epoch": 84.09264305177112,
+ "grad_norm": 1.2364081144332886,
+ "learning_rate": 1.2980598354923567e-06,
+ "loss": 0.0668,
+ "step": 30862
+ },
+ {
+ "epoch": 84.09536784741144,
+ "grad_norm": 0.6284877061843872,
+ "learning_rate": 1.2976250561360904e-06,
+ "loss": 0.0049,
+ "step": 30863
+ },
+ {
+ "epoch": 84.09809264305177,
+ "grad_norm": 1.705036997795105,
+ "learning_rate": 1.2971903445536393e-06,
+ "loss": 0.0133,
+ "step": 30864
+ },
+ {
+ "epoch": 84.1008174386921,
+ "grad_norm": 1.356521725654602,
+ "learning_rate": 1.2967557007483888e-06,
+ "loss": 0.0186,
+ "step": 30865
+ },
+ {
+ "epoch": 84.10354223433242,
+ "grad_norm": 1.3084828853607178,
+ "learning_rate": 1.296321124723726e-06,
+ "loss": 0.0643,
+ "step": 30866
+ },
+ {
+ "epoch": 84.10626702997276,
+ "grad_norm": 0.6864467263221741,
+ "learning_rate": 1.2958866164830342e-06,
+ "loss": 0.0128,
+ "step": 30867
+ },
+ {
+ "epoch": 84.10899182561307,
+ "grad_norm": 0.9470438957214355,
+ "learning_rate": 1.295452176029698e-06,
+ "loss": 0.0096,
+ "step": 30868
+ },
+ {
+ "epoch": 84.11171662125341,
+ "grad_norm": 0.8977245092391968,
+ "learning_rate": 1.2950178033670969e-06,
+ "loss": 0.0076,
+ "step": 30869
+ },
+ {
+ "epoch": 84.11444141689373,
+ "grad_norm": 1.441697120666504,
+ "learning_rate": 1.2945834984986194e-06,
+ "loss": 0.0203,
+ "step": 30870
+ },
+ {
+ "epoch": 84.11716621253406,
+ "grad_norm": 0.9635302424430847,
+ "learning_rate": 1.2941492614276463e-06,
+ "loss": 0.0074,
+ "step": 30871
+ },
+ {
+ "epoch": 84.11989100817439,
+ "grad_norm": 0.9363664388656616,
+ "learning_rate": 1.2937150921575548e-06,
+ "loss": 0.0099,
+ "step": 30872
+ },
+ {
+ "epoch": 84.12261580381471,
+ "grad_norm": 0.7006563544273376,
+ "learning_rate": 1.293280990691733e-06,
+ "loss": 0.0076,
+ "step": 30873
+ },
+ {
+ "epoch": 84.12534059945504,
+ "grad_norm": 1.097943902015686,
+ "learning_rate": 1.292846957033559e-06,
+ "loss": 0.015,
+ "step": 30874
+ },
+ {
+ "epoch": 84.12806539509536,
+ "grad_norm": 0.7385596036911011,
+ "learning_rate": 1.2924129911864126e-06,
+ "loss": 0.0051,
+ "step": 30875
+ },
+ {
+ "epoch": 84.1307901907357,
+ "grad_norm": 1.1102268695831299,
+ "learning_rate": 1.2919790931536714e-06,
+ "loss": 0.0789,
+ "step": 30876
+ },
+ {
+ "epoch": 84.13351498637603,
+ "grad_norm": 1.029744029045105,
+ "learning_rate": 1.291545262938716e-06,
+ "loss": 0.0115,
+ "step": 30877
+ },
+ {
+ "epoch": 84.13623978201635,
+ "grad_norm": 1.1543246507644653,
+ "learning_rate": 1.291111500544928e-06,
+ "loss": 0.0648,
+ "step": 30878
+ },
+ {
+ "epoch": 84.13896457765668,
+ "grad_norm": 1.1939020156860352,
+ "learning_rate": 1.2906778059756842e-06,
+ "loss": 0.0465,
+ "step": 30879
+ },
+ {
+ "epoch": 84.141689373297,
+ "grad_norm": 1.1612849235534668,
+ "learning_rate": 1.2902441792343611e-06,
+ "loss": 0.0178,
+ "step": 30880
+ },
+ {
+ "epoch": 84.14441416893733,
+ "grad_norm": 1.218149185180664,
+ "learning_rate": 1.2898106203243343e-06,
+ "loss": 0.0961,
+ "step": 30881
+ },
+ {
+ "epoch": 84.14713896457765,
+ "grad_norm": 0.9170424342155457,
+ "learning_rate": 1.2893771292489833e-06,
+ "loss": 0.1413,
+ "step": 30882
+ },
+ {
+ "epoch": 84.14986376021798,
+ "grad_norm": 0.7212207317352295,
+ "learning_rate": 1.2889437060116827e-06,
+ "loss": 0.0043,
+ "step": 30883
+ },
+ {
+ "epoch": 84.15258855585832,
+ "grad_norm": 16.58095932006836,
+ "learning_rate": 1.2885103506158091e-06,
+ "loss": 0.0247,
+ "step": 30884
+ },
+ {
+ "epoch": 84.15531335149863,
+ "grad_norm": 0.5110080242156982,
+ "learning_rate": 1.2880770630647344e-06,
+ "loss": 0.0058,
+ "step": 30885
+ },
+ {
+ "epoch": 84.15803814713897,
+ "grad_norm": 1.5506778955459595,
+ "learning_rate": 1.2876438433618365e-06,
+ "loss": 0.0386,
+ "step": 30886
+ },
+ {
+ "epoch": 84.16076294277929,
+ "grad_norm": 0.976514458656311,
+ "learning_rate": 1.2872106915104876e-06,
+ "loss": 0.1354,
+ "step": 30887
+ },
+ {
+ "epoch": 84.16348773841962,
+ "grad_norm": 1.5439789295196533,
+ "learning_rate": 1.2867776075140625e-06,
+ "loss": 0.0286,
+ "step": 30888
+ },
+ {
+ "epoch": 84.16621253405995,
+ "grad_norm": 0.8742840886116028,
+ "learning_rate": 1.2863445913759299e-06,
+ "loss": 0.0074,
+ "step": 30889
+ },
+ {
+ "epoch": 84.16893732970027,
+ "grad_norm": 1.521973729133606,
+ "learning_rate": 1.2859116430994677e-06,
+ "loss": 0.0147,
+ "step": 30890
+ },
+ {
+ "epoch": 84.1716621253406,
+ "grad_norm": 0.6180062890052795,
+ "learning_rate": 1.285478762688046e-06,
+ "loss": 0.005,
+ "step": 30891
+ },
+ {
+ "epoch": 84.17438692098092,
+ "grad_norm": 1.3309669494628906,
+ "learning_rate": 1.285045950145032e-06,
+ "loss": 0.0306,
+ "step": 30892
+ },
+ {
+ "epoch": 84.17711171662125,
+ "grad_norm": 1.0353224277496338,
+ "learning_rate": 1.2846132054738026e-06,
+ "loss": 0.0097,
+ "step": 30893
+ },
+ {
+ "epoch": 84.17983651226157,
+ "grad_norm": 1.408420443534851,
+ "learning_rate": 1.284180528677722e-06,
+ "loss": 0.0169,
+ "step": 30894
+ },
+ {
+ "epoch": 84.1825613079019,
+ "grad_norm": 0.574731171131134,
+ "learning_rate": 1.2837479197601654e-06,
+ "loss": 0.0074,
+ "step": 30895
+ },
+ {
+ "epoch": 84.18528610354224,
+ "grad_norm": 0.951782763004303,
+ "learning_rate": 1.2833153787245e-06,
+ "loss": 0.0176,
+ "step": 30896
+ },
+ {
+ "epoch": 84.18801089918256,
+ "grad_norm": 0.5778957605361938,
+ "learning_rate": 1.282882905574091e-06,
+ "loss": 0.0042,
+ "step": 30897
+ },
+ {
+ "epoch": 84.19073569482289,
+ "grad_norm": 0.8312801718711853,
+ "learning_rate": 1.2824505003123123e-06,
+ "loss": 0.0078,
+ "step": 30898
+ },
+ {
+ "epoch": 84.19346049046321,
+ "grad_norm": 1.3025245666503906,
+ "learning_rate": 1.2820181629425276e-06,
+ "loss": 0.0171,
+ "step": 30899
+ },
+ {
+ "epoch": 84.19618528610354,
+ "grad_norm": 1.07806396484375,
+ "learning_rate": 1.2815858934681059e-06,
+ "loss": 0.0105,
+ "step": 30900
+ },
+ {
+ "epoch": 84.19891008174388,
+ "grad_norm": 0.9065499901771545,
+ "learning_rate": 1.2811536918924095e-06,
+ "loss": 0.0122,
+ "step": 30901
+ },
+ {
+ "epoch": 84.2016348773842,
+ "grad_norm": 1.6495314836502075,
+ "learning_rate": 1.2807215582188104e-06,
+ "loss": 0.0779,
+ "step": 30902
+ },
+ {
+ "epoch": 84.20435967302453,
+ "grad_norm": 0.770395815372467,
+ "learning_rate": 1.2802894924506703e-06,
+ "loss": 0.0102,
+ "step": 30903
+ },
+ {
+ "epoch": 84.20708446866485,
+ "grad_norm": 1.0663467645645142,
+ "learning_rate": 1.2798574945913556e-06,
+ "loss": 0.0053,
+ "step": 30904
+ },
+ {
+ "epoch": 84.20980926430518,
+ "grad_norm": 1.3362900018692017,
+ "learning_rate": 1.2794255646442278e-06,
+ "loss": 0.0188,
+ "step": 30905
+ },
+ {
+ "epoch": 84.2125340599455,
+ "grad_norm": 1.7197734117507935,
+ "learning_rate": 1.2789937026126553e-06,
+ "loss": 0.0611,
+ "step": 30906
+ },
+ {
+ "epoch": 84.21525885558583,
+ "grad_norm": 0.7288493514060974,
+ "learning_rate": 1.278561908499999e-06,
+ "loss": 0.0076,
+ "step": 30907
+ },
+ {
+ "epoch": 84.21798365122616,
+ "grad_norm": 1.37356698513031,
+ "learning_rate": 1.2781301823096226e-06,
+ "loss": 0.0237,
+ "step": 30908
+ },
+ {
+ "epoch": 84.22070844686648,
+ "grad_norm": 1.3076438903808594,
+ "learning_rate": 1.2776985240448847e-06,
+ "loss": 0.032,
+ "step": 30909
+ },
+ {
+ "epoch": 84.22343324250681,
+ "grad_norm": 1.3179932832717896,
+ "learning_rate": 1.2772669337091525e-06,
+ "loss": 0.0118,
+ "step": 30910
+ },
+ {
+ "epoch": 84.22615803814713,
+ "grad_norm": 1.2349909543991089,
+ "learning_rate": 1.2768354113057823e-06,
+ "loss": 0.0998,
+ "step": 30911
+ },
+ {
+ "epoch": 84.22888283378747,
+ "grad_norm": 1.2138094902038574,
+ "learning_rate": 1.2764039568381392e-06,
+ "loss": 0.1074,
+ "step": 30912
+ },
+ {
+ "epoch": 84.2316076294278,
+ "grad_norm": 1.5805373191833496,
+ "learning_rate": 1.2759725703095793e-06,
+ "loss": 0.01,
+ "step": 30913
+ },
+ {
+ "epoch": 84.23433242506812,
+ "grad_norm": 0.9298591017723083,
+ "learning_rate": 1.2755412517234667e-06,
+ "loss": 0.0075,
+ "step": 30914
+ },
+ {
+ "epoch": 84.23705722070845,
+ "grad_norm": 1.1317012310028076,
+ "learning_rate": 1.2751100010831575e-06,
+ "loss": 0.0508,
+ "step": 30915
+ },
+ {
+ "epoch": 84.23978201634877,
+ "grad_norm": 0.9155967235565186,
+ "learning_rate": 1.2746788183920112e-06,
+ "loss": 0.0077,
+ "step": 30916
+ },
+ {
+ "epoch": 84.2425068119891,
+ "grad_norm": 1.0215331315994263,
+ "learning_rate": 1.274247703653383e-06,
+ "loss": 0.0059,
+ "step": 30917
+ },
+ {
+ "epoch": 84.24523160762942,
+ "grad_norm": 1.1356420516967773,
+ "learning_rate": 1.2738166568706345e-06,
+ "loss": 0.0091,
+ "step": 30918
+ },
+ {
+ "epoch": 84.24795640326975,
+ "grad_norm": 1.1966699361801147,
+ "learning_rate": 1.273385678047122e-06,
+ "loss": 0.024,
+ "step": 30919
+ },
+ {
+ "epoch": 84.25068119891009,
+ "grad_norm": 1.2291289567947388,
+ "learning_rate": 1.2729547671862008e-06,
+ "loss": 0.0066,
+ "step": 30920
+ },
+ {
+ "epoch": 84.2534059945504,
+ "grad_norm": 0.7897740006446838,
+ "learning_rate": 1.2725239242912246e-06,
+ "loss": 0.0089,
+ "step": 30921
+ },
+ {
+ "epoch": 84.25613079019074,
+ "grad_norm": 1.1408342123031616,
+ "learning_rate": 1.272093149365553e-06,
+ "loss": 0.0103,
+ "step": 30922
+ },
+ {
+ "epoch": 84.25885558583106,
+ "grad_norm": 0.880445659160614,
+ "learning_rate": 1.2716624424125402e-06,
+ "loss": 0.0109,
+ "step": 30923
+ },
+ {
+ "epoch": 84.26158038147139,
+ "grad_norm": 2.4871788024902344,
+ "learning_rate": 1.2712318034355387e-06,
+ "loss": 0.0121,
+ "step": 30924
+ },
+ {
+ "epoch": 84.26430517711172,
+ "grad_norm": 0.4779399335384369,
+ "learning_rate": 1.2708012324379004e-06,
+ "loss": 0.0048,
+ "step": 30925
+ },
+ {
+ "epoch": 84.26702997275204,
+ "grad_norm": 1.3041024208068848,
+ "learning_rate": 1.2703707294229839e-06,
+ "loss": 0.0835,
+ "step": 30926
+ },
+ {
+ "epoch": 84.26975476839237,
+ "grad_norm": 0.9080114364624023,
+ "learning_rate": 1.2699402943941386e-06,
+ "loss": 0.0112,
+ "step": 30927
+ },
+ {
+ "epoch": 84.2724795640327,
+ "grad_norm": 0.7495811581611633,
+ "learning_rate": 1.2695099273547152e-06,
+ "loss": 0.079,
+ "step": 30928
+ },
+ {
+ "epoch": 84.27520435967303,
+ "grad_norm": 1.1578171253204346,
+ "learning_rate": 1.2690796283080698e-06,
+ "loss": 0.0439,
+ "step": 30929
+ },
+ {
+ "epoch": 84.27792915531334,
+ "grad_norm": 0.949432909488678,
+ "learning_rate": 1.2686493972575498e-06,
+ "loss": 0.0095,
+ "step": 30930
+ },
+ {
+ "epoch": 84.28065395095368,
+ "grad_norm": 1.4928089380264282,
+ "learning_rate": 1.2682192342065093e-06,
+ "loss": 0.0151,
+ "step": 30931
+ },
+ {
+ "epoch": 84.28337874659401,
+ "grad_norm": 0.578228771686554,
+ "learning_rate": 1.2677891391582953e-06,
+ "loss": 0.0052,
+ "step": 30932
+ },
+ {
+ "epoch": 84.28610354223433,
+ "grad_norm": 2.411561965942383,
+ "learning_rate": 1.2673591121162577e-06,
+ "loss": 0.0145,
+ "step": 30933
+ },
+ {
+ "epoch": 84.28882833787466,
+ "grad_norm": 1.3042559623718262,
+ "learning_rate": 1.2669291530837469e-06,
+ "loss": 0.0258,
+ "step": 30934
+ },
+ {
+ "epoch": 84.29155313351498,
+ "grad_norm": 1.1056838035583496,
+ "learning_rate": 1.2664992620641114e-06,
+ "loss": 0.017,
+ "step": 30935
+ },
+ {
+ "epoch": 84.29427792915531,
+ "grad_norm": 1.3093079328536987,
+ "learning_rate": 1.2660694390606986e-06,
+ "loss": 0.0148,
+ "step": 30936
+ },
+ {
+ "epoch": 84.29700272479565,
+ "grad_norm": 0.8971964120864868,
+ "learning_rate": 1.2656396840768549e-06,
+ "loss": 0.0099,
+ "step": 30937
+ },
+ {
+ "epoch": 84.29972752043597,
+ "grad_norm": 1.3406469821929932,
+ "learning_rate": 1.2652099971159293e-06,
+ "loss": 0.0096,
+ "step": 30938
+ },
+ {
+ "epoch": 84.3024523160763,
+ "grad_norm": 1.7135660648345947,
+ "learning_rate": 1.2647803781812673e-06,
+ "loss": 0.0097,
+ "step": 30939
+ },
+ {
+ "epoch": 84.30517711171662,
+ "grad_norm": 0.6533663868904114,
+ "learning_rate": 1.264350827276214e-06,
+ "loss": 0.0134,
+ "step": 30940
+ },
+ {
+ "epoch": 84.30790190735695,
+ "grad_norm": 1.417508840560913,
+ "learning_rate": 1.2639213444041133e-06,
+ "loss": 0.0799,
+ "step": 30941
+ },
+ {
+ "epoch": 84.31062670299727,
+ "grad_norm": 1.221024990081787,
+ "learning_rate": 1.2634919295683145e-06,
+ "loss": 0.0087,
+ "step": 30942
+ },
+ {
+ "epoch": 84.3133514986376,
+ "grad_norm": 1.3641762733459473,
+ "learning_rate": 1.26306258277216e-06,
+ "loss": 0.0093,
+ "step": 30943
+ },
+ {
+ "epoch": 84.31607629427793,
+ "grad_norm": 1.075429916381836,
+ "learning_rate": 1.262633304018992e-06,
+ "loss": 0.0387,
+ "step": 30944
+ },
+ {
+ "epoch": 84.31880108991825,
+ "grad_norm": 1.5480092763900757,
+ "learning_rate": 1.2622040933121527e-06,
+ "loss": 0.0281,
+ "step": 30945
+ },
+ {
+ "epoch": 84.32152588555859,
+ "grad_norm": 1.580188512802124,
+ "learning_rate": 1.261774950654986e-06,
+ "loss": 0.0087,
+ "step": 30946
+ },
+ {
+ "epoch": 84.3242506811989,
+ "grad_norm": 0.9341733455657959,
+ "learning_rate": 1.261345876050838e-06,
+ "loss": 0.0078,
+ "step": 30947
+ },
+ {
+ "epoch": 84.32697547683924,
+ "grad_norm": 0.9618826508522034,
+ "learning_rate": 1.2609168695030471e-06,
+ "loss": 0.0196,
+ "step": 30948
+ },
+ {
+ "epoch": 84.32970027247957,
+ "grad_norm": 1.2589759826660156,
+ "learning_rate": 1.260487931014953e-06,
+ "loss": 0.0167,
+ "step": 30949
+ },
+ {
+ "epoch": 84.33242506811989,
+ "grad_norm": 1.1577026844024658,
+ "learning_rate": 1.2600590605898966e-06,
+ "loss": 0.0135,
+ "step": 30950
+ },
+ {
+ "epoch": 84.33514986376022,
+ "grad_norm": 1.554538607597351,
+ "learning_rate": 1.2596302582312214e-06,
+ "loss": 0.0419,
+ "step": 30951
+ },
+ {
+ "epoch": 84.33787465940054,
+ "grad_norm": 1.4635593891143799,
+ "learning_rate": 1.2592015239422639e-06,
+ "loss": 0.0162,
+ "step": 30952
+ },
+ {
+ "epoch": 84.34059945504087,
+ "grad_norm": 1.4487899541854858,
+ "learning_rate": 1.2587728577263636e-06,
+ "loss": 0.0683,
+ "step": 30953
+ },
+ {
+ "epoch": 84.34332425068119,
+ "grad_norm": 1.4335434436798096,
+ "learning_rate": 1.2583442595868566e-06,
+ "loss": 0.021,
+ "step": 30954
+ },
+ {
+ "epoch": 84.34604904632153,
+ "grad_norm": 1.50652277469635,
+ "learning_rate": 1.257915729527085e-06,
+ "loss": 0.0165,
+ "step": 30955
+ },
+ {
+ "epoch": 84.34877384196186,
+ "grad_norm": 1.5402899980545044,
+ "learning_rate": 1.2574872675503858e-06,
+ "loss": 0.0405,
+ "step": 30956
+ },
+ {
+ "epoch": 84.35149863760218,
+ "grad_norm": 0.8946382403373718,
+ "learning_rate": 1.2570588736600909e-06,
+ "loss": 0.005,
+ "step": 30957
+ },
+ {
+ "epoch": 84.35422343324251,
+ "grad_norm": 1.8874146938323975,
+ "learning_rate": 1.2566305478595431e-06,
+ "loss": 0.0369,
+ "step": 30958
+ },
+ {
+ "epoch": 84.35694822888283,
+ "grad_norm": 1.1926295757293701,
+ "learning_rate": 1.2562022901520744e-06,
+ "loss": 0.0197,
+ "step": 30959
+ },
+ {
+ "epoch": 84.35967302452316,
+ "grad_norm": 0.773543655872345,
+ "learning_rate": 1.2557741005410207e-06,
+ "loss": 0.0069,
+ "step": 30960
+ },
+ {
+ "epoch": 84.3623978201635,
+ "grad_norm": 1.09900963306427,
+ "learning_rate": 1.255345979029715e-06,
+ "loss": 0.0105,
+ "step": 30961
+ },
+ {
+ "epoch": 84.36512261580381,
+ "grad_norm": 0.7539656758308411,
+ "learning_rate": 1.2549179256214939e-06,
+ "loss": 0.0124,
+ "step": 30962
+ },
+ {
+ "epoch": 84.36784741144415,
+ "grad_norm": 1.2824788093566895,
+ "learning_rate": 1.254489940319692e-06,
+ "loss": 0.0121,
+ "step": 30963
+ },
+ {
+ "epoch": 84.37057220708446,
+ "grad_norm": 0.9134441018104553,
+ "learning_rate": 1.2540620231276423e-06,
+ "loss": 0.0106,
+ "step": 30964
+ },
+ {
+ "epoch": 84.3732970027248,
+ "grad_norm": 1.6827211380004883,
+ "learning_rate": 1.2536341740486746e-06,
+ "loss": 0.1633,
+ "step": 30965
+ },
+ {
+ "epoch": 84.37602179836512,
+ "grad_norm": 0.8139957785606384,
+ "learning_rate": 1.2532063930861205e-06,
+ "loss": 0.011,
+ "step": 30966
+ },
+ {
+ "epoch": 84.37874659400545,
+ "grad_norm": 1.884419560432434,
+ "learning_rate": 1.2527786802433162e-06,
+ "loss": 0.1066,
+ "step": 30967
+ },
+ {
+ "epoch": 84.38147138964578,
+ "grad_norm": 0.8597909808158875,
+ "learning_rate": 1.2523510355235902e-06,
+ "loss": 0.013,
+ "step": 30968
+ },
+ {
+ "epoch": 84.3841961852861,
+ "grad_norm": 1.0543917417526245,
+ "learning_rate": 1.251923458930272e-06,
+ "loss": 0.0369,
+ "step": 30969
+ },
+ {
+ "epoch": 84.38692098092643,
+ "grad_norm": 0.6379631757736206,
+ "learning_rate": 1.25149595046669e-06,
+ "loss": 0.0054,
+ "step": 30970
+ },
+ {
+ "epoch": 84.38964577656675,
+ "grad_norm": 1.2614576816558838,
+ "learning_rate": 1.2510685101361797e-06,
+ "loss": 0.0309,
+ "step": 30971
+ },
+ {
+ "epoch": 84.39237057220708,
+ "grad_norm": 0.7008721232414246,
+ "learning_rate": 1.2506411379420647e-06,
+ "loss": 0.0068,
+ "step": 30972
+ },
+ {
+ "epoch": 84.39509536784742,
+ "grad_norm": 1.3988360166549683,
+ "learning_rate": 1.2502138338876757e-06,
+ "loss": 0.0502,
+ "step": 30973
+ },
+ {
+ "epoch": 84.39782016348774,
+ "grad_norm": 0.7076833844184875,
+ "learning_rate": 1.2497865979763368e-06,
+ "loss": 0.0087,
+ "step": 30974
+ },
+ {
+ "epoch": 84.40054495912807,
+ "grad_norm": 0.7922095060348511,
+ "learning_rate": 1.249359430211381e-06,
+ "loss": 0.0123,
+ "step": 30975
+ },
+ {
+ "epoch": 84.40326975476839,
+ "grad_norm": 1.2521989345550537,
+ "learning_rate": 1.248932330596132e-06,
+ "loss": 0.0104,
+ "step": 30976
+ },
+ {
+ "epoch": 84.40599455040872,
+ "grad_norm": 0.5371987819671631,
+ "learning_rate": 1.2485052991339174e-06,
+ "loss": 0.0055,
+ "step": 30977
+ },
+ {
+ "epoch": 84.40871934604904,
+ "grad_norm": 1.24932861328125,
+ "learning_rate": 1.248078335828059e-06,
+ "loss": 0.0071,
+ "step": 30978
+ },
+ {
+ "epoch": 84.41144414168937,
+ "grad_norm": 0.7221924662590027,
+ "learning_rate": 1.2476514406818863e-06,
+ "loss": 0.0072,
+ "step": 30979
+ },
+ {
+ "epoch": 84.4141689373297,
+ "grad_norm": 2.032341718673706,
+ "learning_rate": 1.247224613698721e-06,
+ "loss": 0.0087,
+ "step": 30980
+ },
+ {
+ "epoch": 84.41689373297002,
+ "grad_norm": 1.3459172248840332,
+ "learning_rate": 1.2467978548818914e-06,
+ "loss": 0.0647,
+ "step": 30981
+ },
+ {
+ "epoch": 84.41961852861036,
+ "grad_norm": 1.046241044998169,
+ "learning_rate": 1.2463711642347154e-06,
+ "loss": 0.1115,
+ "step": 30982
+ },
+ {
+ "epoch": 84.42234332425068,
+ "grad_norm": 0.8622056841850281,
+ "learning_rate": 1.245944541760522e-06,
+ "loss": 0.0109,
+ "step": 30983
+ },
+ {
+ "epoch": 84.42506811989101,
+ "grad_norm": 0.9158808588981628,
+ "learning_rate": 1.24551798746263e-06,
+ "loss": 0.0065,
+ "step": 30984
+ },
+ {
+ "epoch": 84.42779291553134,
+ "grad_norm": 0.8635522127151489,
+ "learning_rate": 1.245091501344362e-06,
+ "loss": 0.0113,
+ "step": 30985
+ },
+ {
+ "epoch": 84.43051771117166,
+ "grad_norm": 1.0754350423812866,
+ "learning_rate": 1.244665083409038e-06,
+ "loss": 0.0152,
+ "step": 30986
+ },
+ {
+ "epoch": 84.433242506812,
+ "grad_norm": 0.9609194397926331,
+ "learning_rate": 1.2442387336599825e-06,
+ "loss": 0.0085,
+ "step": 30987
+ },
+ {
+ "epoch": 84.43596730245231,
+ "grad_norm": 0.6632782220840454,
+ "learning_rate": 1.2438124521005135e-06,
+ "loss": 0.0064,
+ "step": 30988
+ },
+ {
+ "epoch": 84.43869209809264,
+ "grad_norm": 1.563328504562378,
+ "learning_rate": 1.2433862387339524e-06,
+ "loss": 0.0085,
+ "step": 30989
+ },
+ {
+ "epoch": 84.44141689373296,
+ "grad_norm": 1.0791025161743164,
+ "learning_rate": 1.2429600935636144e-06,
+ "loss": 0.0124,
+ "step": 30990
+ },
+ {
+ "epoch": 84.4441416893733,
+ "grad_norm": 0.9499502778053284,
+ "learning_rate": 1.2425340165928234e-06,
+ "loss": 0.0117,
+ "step": 30991
+ },
+ {
+ "epoch": 84.44686648501363,
+ "grad_norm": 0.968917965888977,
+ "learning_rate": 1.242108007824896e-06,
+ "loss": 0.0063,
+ "step": 30992
+ },
+ {
+ "epoch": 84.44959128065395,
+ "grad_norm": 1.1551958322525024,
+ "learning_rate": 1.2416820672631502e-06,
+ "loss": 0.0121,
+ "step": 30993
+ },
+ {
+ "epoch": 84.45231607629428,
+ "grad_norm": 1.5345706939697266,
+ "learning_rate": 1.2412561949108992e-06,
+ "loss": 0.0841,
+ "step": 30994
+ },
+ {
+ "epoch": 84.4550408719346,
+ "grad_norm": 0.8423279523849487,
+ "learning_rate": 1.240830390771466e-06,
+ "loss": 0.0085,
+ "step": 30995
+ },
+ {
+ "epoch": 84.45776566757493,
+ "grad_norm": 1.5047060251235962,
+ "learning_rate": 1.2404046548481641e-06,
+ "loss": 0.0093,
+ "step": 30996
+ },
+ {
+ "epoch": 84.46049046321527,
+ "grad_norm": 3.5814027786254883,
+ "learning_rate": 1.2399789871443057e-06,
+ "loss": 0.0392,
+ "step": 30997
+ },
+ {
+ "epoch": 84.46321525885558,
+ "grad_norm": 0.9732081294059753,
+ "learning_rate": 1.2395533876632094e-06,
+ "loss": 0.0082,
+ "step": 30998
+ },
+ {
+ "epoch": 84.46594005449592,
+ "grad_norm": 1.2685030698776245,
+ "learning_rate": 1.2391278564081921e-06,
+ "loss": 0.0524,
+ "step": 30999
+ },
+ {
+ "epoch": 84.46866485013624,
+ "grad_norm": 0.8983018398284912,
+ "learning_rate": 1.2387023933825637e-06,
+ "loss": 0.0122,
+ "step": 31000
+ },
+ {
+ "epoch": 84.47138964577657,
+ "grad_norm": 1.608008623123169,
+ "learning_rate": 1.2382769985896405e-06,
+ "loss": 0.0207,
+ "step": 31001
+ },
+ {
+ "epoch": 84.47411444141689,
+ "grad_norm": 1.2213331460952759,
+ "learning_rate": 1.2378516720327315e-06,
+ "loss": 0.0329,
+ "step": 31002
+ },
+ {
+ "epoch": 84.47683923705722,
+ "grad_norm": 0.7914794087409973,
+ "learning_rate": 1.2374264137151537e-06,
+ "loss": 0.009,
+ "step": 31003
+ },
+ {
+ "epoch": 84.47956403269755,
+ "grad_norm": 0.990422248840332,
+ "learning_rate": 1.2370012236402162e-06,
+ "loss": 0.0297,
+ "step": 31004
+ },
+ {
+ "epoch": 84.48228882833787,
+ "grad_norm": 1.1205251216888428,
+ "learning_rate": 1.236576101811232e-06,
+ "loss": 0.035,
+ "step": 31005
+ },
+ {
+ "epoch": 84.4850136239782,
+ "grad_norm": 1.4302164316177368,
+ "learning_rate": 1.2361510482315087e-06,
+ "loss": 0.1782,
+ "step": 31006
+ },
+ {
+ "epoch": 84.48773841961852,
+ "grad_norm": 1.452676773071289,
+ "learning_rate": 1.235726062904361e-06,
+ "loss": 0.0212,
+ "step": 31007
+ },
+ {
+ "epoch": 84.49046321525886,
+ "grad_norm": 1.2781407833099365,
+ "learning_rate": 1.2353011458330976e-06,
+ "loss": 0.0284,
+ "step": 31008
+ },
+ {
+ "epoch": 84.49318801089919,
+ "grad_norm": 0.7584709525108337,
+ "learning_rate": 1.234876297021026e-06,
+ "loss": 0.0067,
+ "step": 31009
+ },
+ {
+ "epoch": 84.49591280653951,
+ "grad_norm": 0.6916016936302185,
+ "learning_rate": 1.2344515164714531e-06,
+ "loss": 0.006,
+ "step": 31010
+ },
+ {
+ "epoch": 84.49863760217984,
+ "grad_norm": 1.0762795209884644,
+ "learning_rate": 1.2340268041876923e-06,
+ "loss": 0.0132,
+ "step": 31011
+ },
+ {
+ "epoch": 84.50136239782016,
+ "grad_norm": 1.0385866165161133,
+ "learning_rate": 1.2336021601730485e-06,
+ "loss": 0.0112,
+ "step": 31012
+ },
+ {
+ "epoch": 84.50408719346049,
+ "grad_norm": 0.7600952982902527,
+ "learning_rate": 1.2331775844308292e-06,
+ "loss": 0.0084,
+ "step": 31013
+ },
+ {
+ "epoch": 84.50681198910081,
+ "grad_norm": 1.239766240119934,
+ "learning_rate": 1.232753076964338e-06,
+ "loss": 0.0238,
+ "step": 31014
+ },
+ {
+ "epoch": 84.50953678474114,
+ "grad_norm": 1.4178870916366577,
+ "learning_rate": 1.2323286377768839e-06,
+ "loss": 0.0126,
+ "step": 31015
+ },
+ {
+ "epoch": 84.51226158038148,
+ "grad_norm": 2.111236810684204,
+ "learning_rate": 1.231904266871775e-06,
+ "loss": 0.0083,
+ "step": 31016
+ },
+ {
+ "epoch": 84.5149863760218,
+ "grad_norm": 0.893808126449585,
+ "learning_rate": 1.2314799642523135e-06,
+ "loss": 0.0202,
+ "step": 31017
+ },
+ {
+ "epoch": 84.51771117166213,
+ "grad_norm": 0.683675229549408,
+ "learning_rate": 1.2310557299218029e-06,
+ "loss": 0.0067,
+ "step": 31018
+ },
+ {
+ "epoch": 84.52043596730245,
+ "grad_norm": 1.1370407342910767,
+ "learning_rate": 1.2306315638835464e-06,
+ "loss": 0.0181,
+ "step": 31019
+ },
+ {
+ "epoch": 84.52316076294278,
+ "grad_norm": 1.1777008771896362,
+ "learning_rate": 1.2302074661408502e-06,
+ "loss": 0.0121,
+ "step": 31020
+ },
+ {
+ "epoch": 84.52588555858311,
+ "grad_norm": 1.5798588991165161,
+ "learning_rate": 1.2297834366970174e-06,
+ "loss": 0.0377,
+ "step": 31021
+ },
+ {
+ "epoch": 84.52861035422343,
+ "grad_norm": 1.4152531623840332,
+ "learning_rate": 1.2293594755553462e-06,
+ "loss": 0.1534,
+ "step": 31022
+ },
+ {
+ "epoch": 84.53133514986376,
+ "grad_norm": 1.8922615051269531,
+ "learning_rate": 1.228935582719143e-06,
+ "loss": 0.0128,
+ "step": 31023
+ },
+ {
+ "epoch": 84.53405994550408,
+ "grad_norm": 2.564849853515625,
+ "learning_rate": 1.2285117581917084e-06,
+ "loss": 0.0712,
+ "step": 31024
+ },
+ {
+ "epoch": 84.53678474114442,
+ "grad_norm": 0.7368241548538208,
+ "learning_rate": 1.2280880019763407e-06,
+ "loss": 0.0044,
+ "step": 31025
+ },
+ {
+ "epoch": 84.53950953678473,
+ "grad_norm": 1.3271645307540894,
+ "learning_rate": 1.227664314076339e-06,
+ "loss": 0.0399,
+ "step": 31026
+ },
+ {
+ "epoch": 84.54223433242507,
+ "grad_norm": 0.7734759449958801,
+ "learning_rate": 1.227240694495009e-06,
+ "loss": 0.0048,
+ "step": 31027
+ },
+ {
+ "epoch": 84.5449591280654,
+ "grad_norm": 1.4214001893997192,
+ "learning_rate": 1.2268171432356447e-06,
+ "loss": 0.1004,
+ "step": 31028
+ },
+ {
+ "epoch": 84.54768392370572,
+ "grad_norm": 1.186771273612976,
+ "learning_rate": 1.2263936603015457e-06,
+ "loss": 0.0187,
+ "step": 31029
+ },
+ {
+ "epoch": 84.55040871934605,
+ "grad_norm": 2.0800251960754395,
+ "learning_rate": 1.2259702456960099e-06,
+ "loss": 0.0593,
+ "step": 31030
+ },
+ {
+ "epoch": 84.55313351498637,
+ "grad_norm": 1.4638558626174927,
+ "learning_rate": 1.2255468994223342e-06,
+ "loss": 0.0631,
+ "step": 31031
+ },
+ {
+ "epoch": 84.5558583106267,
+ "grad_norm": 1.5258677005767822,
+ "learning_rate": 1.225123621483819e-06,
+ "loss": 0.0177,
+ "step": 31032
+ },
+ {
+ "epoch": 84.55858310626704,
+ "grad_norm": 0.8751928806304932,
+ "learning_rate": 1.2247004118837603e-06,
+ "loss": 0.0785,
+ "step": 31033
+ },
+ {
+ "epoch": 84.56130790190736,
+ "grad_norm": 1.1115351915359497,
+ "learning_rate": 1.2242772706254502e-06,
+ "loss": 0.011,
+ "step": 31034
+ },
+ {
+ "epoch": 84.56403269754769,
+ "grad_norm": 1.1006813049316406,
+ "learning_rate": 1.2238541977121853e-06,
+ "loss": 0.0677,
+ "step": 31035
+ },
+ {
+ "epoch": 84.566757493188,
+ "grad_norm": 1.276435136795044,
+ "learning_rate": 1.223431193147263e-06,
+ "loss": 0.0962,
+ "step": 31036
+ },
+ {
+ "epoch": 84.56948228882834,
+ "grad_norm": 1.0032027959823608,
+ "learning_rate": 1.2230082569339773e-06,
+ "loss": 0.0064,
+ "step": 31037
+ },
+ {
+ "epoch": 84.57220708446866,
+ "grad_norm": 0.8589203953742981,
+ "learning_rate": 1.222585389075619e-06,
+ "loss": 0.0055,
+ "step": 31038
+ },
+ {
+ "epoch": 84.57493188010899,
+ "grad_norm": 0.8181492686271667,
+ "learning_rate": 1.222162589575483e-06,
+ "loss": 0.0099,
+ "step": 31039
+ },
+ {
+ "epoch": 84.57765667574932,
+ "grad_norm": 1.4376132488250732,
+ "learning_rate": 1.2217398584368634e-06,
+ "loss": 0.0657,
+ "step": 31040
+ },
+ {
+ "epoch": 84.58038147138964,
+ "grad_norm": 0.939508855342865,
+ "learning_rate": 1.221317195663051e-06,
+ "loss": 0.0824,
+ "step": 31041
+ },
+ {
+ "epoch": 84.58310626702998,
+ "grad_norm": 1.0582079887390137,
+ "learning_rate": 1.2208946012573352e-06,
+ "loss": 0.0272,
+ "step": 31042
+ },
+ {
+ "epoch": 84.5858310626703,
+ "grad_norm": 1.7792601585388184,
+ "learning_rate": 1.2204720752230126e-06,
+ "loss": 0.0124,
+ "step": 31043
+ },
+ {
+ "epoch": 84.58855585831063,
+ "grad_norm": 1.487618327140808,
+ "learning_rate": 1.2200496175633714e-06,
+ "loss": 0.1152,
+ "step": 31044
+ },
+ {
+ "epoch": 84.59128065395096,
+ "grad_norm": 0.9967483282089233,
+ "learning_rate": 1.2196272282817e-06,
+ "loss": 0.0144,
+ "step": 31045
+ },
+ {
+ "epoch": 84.59400544959128,
+ "grad_norm": 1.0604125261306763,
+ "learning_rate": 1.2192049073812872e-06,
+ "loss": 0.0313,
+ "step": 31046
+ },
+ {
+ "epoch": 84.59673024523161,
+ "grad_norm": 0.5896845459938049,
+ "learning_rate": 1.218782654865427e-06,
+ "loss": 0.0055,
+ "step": 31047
+ },
+ {
+ "epoch": 84.59945504087193,
+ "grad_norm": 0.8661300539970398,
+ "learning_rate": 1.218360470737402e-06,
+ "loss": 0.0097,
+ "step": 31048
+ },
+ {
+ "epoch": 84.60217983651226,
+ "grad_norm": 0.9228329062461853,
+ "learning_rate": 1.2179383550005052e-06,
+ "loss": 0.0067,
+ "step": 31049
+ },
+ {
+ "epoch": 84.60490463215258,
+ "grad_norm": 0.8962114453315735,
+ "learning_rate": 1.2175163076580222e-06,
+ "loss": 0.0084,
+ "step": 31050
+ },
+ {
+ "epoch": 84.60762942779292,
+ "grad_norm": 1.537314772605896,
+ "learning_rate": 1.2170943287132375e-06,
+ "loss": 0.0842,
+ "step": 31051
+ },
+ {
+ "epoch": 84.61035422343325,
+ "grad_norm": 0.982888400554657,
+ "learning_rate": 1.2166724181694411e-06,
+ "loss": 0.0425,
+ "step": 31052
+ },
+ {
+ "epoch": 84.61307901907357,
+ "grad_norm": 0.3580421805381775,
+ "learning_rate": 1.2162505760299182e-06,
+ "loss": 0.0041,
+ "step": 31053
+ },
+ {
+ "epoch": 84.6158038147139,
+ "grad_norm": 1.4357779026031494,
+ "learning_rate": 1.2158288022979514e-06,
+ "loss": 0.0466,
+ "step": 31054
+ },
+ {
+ "epoch": 84.61852861035422,
+ "grad_norm": 1.8163148164749146,
+ "learning_rate": 1.215407096976826e-06,
+ "loss": 0.0314,
+ "step": 31055
+ },
+ {
+ "epoch": 84.62125340599455,
+ "grad_norm": 0.9437087178230286,
+ "learning_rate": 1.2149854600698296e-06,
+ "loss": 0.0125,
+ "step": 31056
+ },
+ {
+ "epoch": 84.62397820163488,
+ "grad_norm": 1.143755555152893,
+ "learning_rate": 1.2145638915802438e-06,
+ "loss": 0.0077,
+ "step": 31057
+ },
+ {
+ "epoch": 84.6267029972752,
+ "grad_norm": 1.2656954526901245,
+ "learning_rate": 1.2141423915113515e-06,
+ "loss": 0.0067,
+ "step": 31058
+ },
+ {
+ "epoch": 84.62942779291554,
+ "grad_norm": 1.0464705228805542,
+ "learning_rate": 1.2137209598664334e-06,
+ "loss": 0.0247,
+ "step": 31059
+ },
+ {
+ "epoch": 84.63215258855585,
+ "grad_norm": 1.1943631172180176,
+ "learning_rate": 1.213299596648776e-06,
+ "loss": 0.0595,
+ "step": 31060
+ },
+ {
+ "epoch": 84.63487738419619,
+ "grad_norm": 1.5149438381195068,
+ "learning_rate": 1.2128783018616585e-06,
+ "loss": 0.0815,
+ "step": 31061
+ },
+ {
+ "epoch": 84.6376021798365,
+ "grad_norm": 1.932174801826477,
+ "learning_rate": 1.212457075508362e-06,
+ "loss": 0.0123,
+ "step": 31062
+ },
+ {
+ "epoch": 84.64032697547684,
+ "grad_norm": 1.7638304233551025,
+ "learning_rate": 1.2120359175921636e-06,
+ "loss": 0.2105,
+ "step": 31063
+ },
+ {
+ "epoch": 84.64305177111717,
+ "grad_norm": 1.0161144733428955,
+ "learning_rate": 1.2116148281163497e-06,
+ "loss": 0.0206,
+ "step": 31064
+ },
+ {
+ "epoch": 84.64577656675749,
+ "grad_norm": 1.0749468803405762,
+ "learning_rate": 1.2111938070841979e-06,
+ "loss": 0.0234,
+ "step": 31065
+ },
+ {
+ "epoch": 84.64850136239782,
+ "grad_norm": 0.5190554857254028,
+ "learning_rate": 1.2107728544989817e-06,
+ "loss": 0.0052,
+ "step": 31066
+ },
+ {
+ "epoch": 84.65122615803814,
+ "grad_norm": 1.7976651191711426,
+ "learning_rate": 1.2103519703639844e-06,
+ "loss": 0.0592,
+ "step": 31067
+ },
+ {
+ "epoch": 84.65395095367847,
+ "grad_norm": 0.7870824337005615,
+ "learning_rate": 1.2099311546824853e-06,
+ "loss": 0.0097,
+ "step": 31068
+ },
+ {
+ "epoch": 84.65667574931881,
+ "grad_norm": 0.8843756318092346,
+ "learning_rate": 1.20951040745776e-06,
+ "loss": 0.0093,
+ "step": 31069
+ },
+ {
+ "epoch": 84.65940054495913,
+ "grad_norm": 1.3531910181045532,
+ "learning_rate": 1.2090897286930837e-06,
+ "loss": 0.0246,
+ "step": 31070
+ },
+ {
+ "epoch": 84.66212534059946,
+ "grad_norm": 1.2606308460235596,
+ "learning_rate": 1.2086691183917322e-06,
+ "loss": 0.0138,
+ "step": 31071
+ },
+ {
+ "epoch": 84.66485013623978,
+ "grad_norm": 1.2851914167404175,
+ "learning_rate": 1.208248576556985e-06,
+ "loss": 0.0175,
+ "step": 31072
+ },
+ {
+ "epoch": 84.66757493188011,
+ "grad_norm": 2.2620177268981934,
+ "learning_rate": 1.2078281031921146e-06,
+ "loss": 0.054,
+ "step": 31073
+ },
+ {
+ "epoch": 84.67029972752043,
+ "grad_norm": 1.30450439453125,
+ "learning_rate": 1.2074076983003956e-06,
+ "loss": 0.0133,
+ "step": 31074
+ },
+ {
+ "epoch": 84.67302452316076,
+ "grad_norm": 0.7186424732208252,
+ "learning_rate": 1.2069873618851001e-06,
+ "loss": 0.0097,
+ "step": 31075
+ },
+ {
+ "epoch": 84.6757493188011,
+ "grad_norm": 1.2797297239303589,
+ "learning_rate": 1.2065670939495068e-06,
+ "loss": 0.0187,
+ "step": 31076
+ },
+ {
+ "epoch": 84.67847411444141,
+ "grad_norm": 0.9114944338798523,
+ "learning_rate": 1.2061468944968857e-06,
+ "loss": 0.0056,
+ "step": 31077
+ },
+ {
+ "epoch": 84.68119891008175,
+ "grad_norm": 1.1178172826766968,
+ "learning_rate": 1.2057267635305102e-06,
+ "loss": 0.019,
+ "step": 31078
+ },
+ {
+ "epoch": 84.68392370572207,
+ "grad_norm": 2.3389182090759277,
+ "learning_rate": 1.205306701053649e-06,
+ "loss": 0.02,
+ "step": 31079
+ },
+ {
+ "epoch": 84.6866485013624,
+ "grad_norm": 0.6999226808547974,
+ "learning_rate": 1.204886707069578e-06,
+ "loss": 0.0102,
+ "step": 31080
+ },
+ {
+ "epoch": 84.68937329700273,
+ "grad_norm": 1.5120255947113037,
+ "learning_rate": 1.2044667815815668e-06,
+ "loss": 0.0243,
+ "step": 31081
+ },
+ {
+ "epoch": 84.69209809264305,
+ "grad_norm": 0.6595015525817871,
+ "learning_rate": 1.2040469245928843e-06,
+ "loss": 0.0069,
+ "step": 31082
+ },
+ {
+ "epoch": 84.69482288828338,
+ "grad_norm": 1.0801547765731812,
+ "learning_rate": 1.2036271361067996e-06,
+ "loss": 0.014,
+ "step": 31083
+ },
+ {
+ "epoch": 84.6975476839237,
+ "grad_norm": 0.7901831269264221,
+ "learning_rate": 1.203207416126584e-06,
+ "loss": 0.0071,
+ "step": 31084
+ },
+ {
+ "epoch": 84.70027247956403,
+ "grad_norm": 1.1059235334396362,
+ "learning_rate": 1.2027877646555087e-06,
+ "loss": 0.0122,
+ "step": 31085
+ },
+ {
+ "epoch": 84.70299727520435,
+ "grad_norm": 1.9493991136550903,
+ "learning_rate": 1.202368181696838e-06,
+ "loss": 0.0199,
+ "step": 31086
+ },
+ {
+ "epoch": 84.70572207084469,
+ "grad_norm": 0.676081657409668,
+ "learning_rate": 1.2019486672538383e-06,
+ "loss": 0.006,
+ "step": 31087
+ },
+ {
+ "epoch": 84.70844686648502,
+ "grad_norm": 1.1252377033233643,
+ "learning_rate": 1.2015292213297824e-06,
+ "loss": 0.0218,
+ "step": 31088
+ },
+ {
+ "epoch": 84.71117166212534,
+ "grad_norm": 1.01111900806427,
+ "learning_rate": 1.2011098439279333e-06,
+ "loss": 0.0189,
+ "step": 31089
+ },
+ {
+ "epoch": 84.71389645776567,
+ "grad_norm": 1.1973373889923096,
+ "learning_rate": 1.2006905350515575e-06,
+ "loss": 0.0431,
+ "step": 31090
+ },
+ {
+ "epoch": 84.71662125340599,
+ "grad_norm": 0.7323651313781738,
+ "learning_rate": 1.2002712947039185e-06,
+ "loss": 0.0062,
+ "step": 31091
+ },
+ {
+ "epoch": 84.71934604904632,
+ "grad_norm": 1.0636111497879028,
+ "learning_rate": 1.1998521228882853e-06,
+ "loss": 0.0097,
+ "step": 31092
+ },
+ {
+ "epoch": 84.72207084468666,
+ "grad_norm": 1.0034586191177368,
+ "learning_rate": 1.1994330196079208e-06,
+ "loss": 0.012,
+ "step": 31093
+ },
+ {
+ "epoch": 84.72479564032697,
+ "grad_norm": 1.1617721319198608,
+ "learning_rate": 1.1990139848660886e-06,
+ "loss": 0.0264,
+ "step": 31094
+ },
+ {
+ "epoch": 84.7275204359673,
+ "grad_norm": 1.3357515335083008,
+ "learning_rate": 1.198595018666051e-06,
+ "loss": 0.0484,
+ "step": 31095
+ },
+ {
+ "epoch": 84.73024523160763,
+ "grad_norm": 0.8876215815544128,
+ "learning_rate": 1.1981761210110732e-06,
+ "loss": 0.0116,
+ "step": 31096
+ },
+ {
+ "epoch": 84.73297002724796,
+ "grad_norm": 0.9185503125190735,
+ "learning_rate": 1.1977572919044166e-06,
+ "loss": 0.0068,
+ "step": 31097
+ },
+ {
+ "epoch": 84.73569482288828,
+ "grad_norm": 2.542173147201538,
+ "learning_rate": 1.1973385313493436e-06,
+ "loss": 0.0116,
+ "step": 31098
+ },
+ {
+ "epoch": 84.73841961852861,
+ "grad_norm": 1.6313633918762207,
+ "learning_rate": 1.1969198393491123e-06,
+ "loss": 0.023,
+ "step": 31099
+ },
+ {
+ "epoch": 84.74114441416894,
+ "grad_norm": 1.1778494119644165,
+ "learning_rate": 1.1965012159069855e-06,
+ "loss": 0.0252,
+ "step": 31100
+ },
+ {
+ "epoch": 84.74386920980926,
+ "grad_norm": 1.3852213621139526,
+ "learning_rate": 1.1960826610262266e-06,
+ "loss": 0.0121,
+ "step": 31101
+ },
+ {
+ "epoch": 84.7465940054496,
+ "grad_norm": 1.3121076822280884,
+ "learning_rate": 1.1956641747100918e-06,
+ "loss": 0.0097,
+ "step": 31102
+ },
+ {
+ "epoch": 84.74931880108991,
+ "grad_norm": 0.6582732796669006,
+ "learning_rate": 1.1952457569618415e-06,
+ "loss": 0.0036,
+ "step": 31103
+ },
+ {
+ "epoch": 84.75204359673025,
+ "grad_norm": 1.517339825630188,
+ "learning_rate": 1.1948274077847322e-06,
+ "loss": 0.0066,
+ "step": 31104
+ },
+ {
+ "epoch": 84.75476839237058,
+ "grad_norm": 1.2597436904907227,
+ "learning_rate": 1.1944091271820258e-06,
+ "loss": 0.0151,
+ "step": 31105
+ },
+ {
+ "epoch": 84.7574931880109,
+ "grad_norm": 1.2146379947662354,
+ "learning_rate": 1.1939909151569773e-06,
+ "loss": 0.0113,
+ "step": 31106
+ },
+ {
+ "epoch": 84.76021798365123,
+ "grad_norm": 0.6767370700836182,
+ "learning_rate": 1.193572771712842e-06,
+ "loss": 0.0113,
+ "step": 31107
+ },
+ {
+ "epoch": 84.76294277929155,
+ "grad_norm": 1.4760773181915283,
+ "learning_rate": 1.1931546968528807e-06,
+ "loss": 0.0173,
+ "step": 31108
+ },
+ {
+ "epoch": 84.76566757493188,
+ "grad_norm": 1.0263220071792603,
+ "learning_rate": 1.192736690580346e-06,
+ "loss": 0.0522,
+ "step": 31109
+ },
+ {
+ "epoch": 84.7683923705722,
+ "grad_norm": 1.0797466039657593,
+ "learning_rate": 1.1923187528984958e-06,
+ "loss": 0.0582,
+ "step": 31110
+ },
+ {
+ "epoch": 84.77111716621253,
+ "grad_norm": 0.9102641940116882,
+ "learning_rate": 1.1919008838105816e-06,
+ "loss": 0.0091,
+ "step": 31111
+ },
+ {
+ "epoch": 84.77384196185287,
+ "grad_norm": 1.2479532957077026,
+ "learning_rate": 1.191483083319862e-06,
+ "loss": 0.1224,
+ "step": 31112
+ },
+ {
+ "epoch": 84.77656675749319,
+ "grad_norm": 1.3013873100280762,
+ "learning_rate": 1.191065351429589e-06,
+ "loss": 0.0521,
+ "step": 31113
+ },
+ {
+ "epoch": 84.77929155313352,
+ "grad_norm": 1.0773556232452393,
+ "learning_rate": 1.190647688143014e-06,
+ "loss": 0.0746,
+ "step": 31114
+ },
+ {
+ "epoch": 84.78201634877384,
+ "grad_norm": 0.8431684970855713,
+ "learning_rate": 1.1902300934633904e-06,
+ "loss": 0.0072,
+ "step": 31115
+ },
+ {
+ "epoch": 84.78474114441417,
+ "grad_norm": 1.3483566045761108,
+ "learning_rate": 1.189812567393973e-06,
+ "loss": 0.0161,
+ "step": 31116
+ },
+ {
+ "epoch": 84.7874659400545,
+ "grad_norm": 0.6405753493309021,
+ "learning_rate": 1.1893951099380097e-06,
+ "loss": 0.0053,
+ "step": 31117
+ },
+ {
+ "epoch": 84.79019073569482,
+ "grad_norm": 1.1433300971984863,
+ "learning_rate": 1.1889777210987564e-06,
+ "loss": 0.0132,
+ "step": 31118
+ },
+ {
+ "epoch": 84.79291553133515,
+ "grad_norm": 1.0742815732955933,
+ "learning_rate": 1.1885604008794604e-06,
+ "loss": 0.0076,
+ "step": 31119
+ },
+ {
+ "epoch": 84.79564032697547,
+ "grad_norm": 1.0717556476593018,
+ "learning_rate": 1.1881431492833694e-06,
+ "loss": 0.0122,
+ "step": 31120
+ },
+ {
+ "epoch": 84.7983651226158,
+ "grad_norm": 0.970238447189331,
+ "learning_rate": 1.1877259663137398e-06,
+ "loss": 0.009,
+ "step": 31121
+ },
+ {
+ "epoch": 84.80108991825612,
+ "grad_norm": 0.8420488834381104,
+ "learning_rate": 1.1873088519738163e-06,
+ "loss": 0.0069,
+ "step": 31122
+ },
+ {
+ "epoch": 84.80381471389646,
+ "grad_norm": 1.1317652463912964,
+ "learning_rate": 1.1868918062668467e-06,
+ "loss": 0.0069,
+ "step": 31123
+ },
+ {
+ "epoch": 84.80653950953679,
+ "grad_norm": 1.6906763315200806,
+ "learning_rate": 1.1864748291960793e-06,
+ "loss": 0.0156,
+ "step": 31124
+ },
+ {
+ "epoch": 84.80926430517711,
+ "grad_norm": 0.9849642515182495,
+ "learning_rate": 1.1860579207647638e-06,
+ "loss": 0.0523,
+ "step": 31125
+ },
+ {
+ "epoch": 84.81198910081744,
+ "grad_norm": 1.0466686487197876,
+ "learning_rate": 1.1856410809761454e-06,
+ "loss": 0.01,
+ "step": 31126
+ },
+ {
+ "epoch": 84.81471389645776,
+ "grad_norm": 1.2862639427185059,
+ "learning_rate": 1.1852243098334703e-06,
+ "loss": 0.0137,
+ "step": 31127
+ },
+ {
+ "epoch": 84.8174386920981,
+ "grad_norm": 0.7948804497718811,
+ "learning_rate": 1.1848076073399838e-06,
+ "loss": 0.0086,
+ "step": 31128
+ },
+ {
+ "epoch": 84.82016348773843,
+ "grad_norm": 1.191856026649475,
+ "learning_rate": 1.1843909734989323e-06,
+ "loss": 0.0131,
+ "step": 31129
+ },
+ {
+ "epoch": 84.82288828337875,
+ "grad_norm": 1.4012558460235596,
+ "learning_rate": 1.1839744083135618e-06,
+ "loss": 0.0437,
+ "step": 31130
+ },
+ {
+ "epoch": 84.82561307901908,
+ "grad_norm": 1.9934625625610352,
+ "learning_rate": 1.183557911787112e-06,
+ "loss": 0.0538,
+ "step": 31131
+ },
+ {
+ "epoch": 84.8283378746594,
+ "grad_norm": 0.7403671145439148,
+ "learning_rate": 1.1831414839228318e-06,
+ "loss": 0.0069,
+ "step": 31132
+ },
+ {
+ "epoch": 84.83106267029973,
+ "grad_norm": 1.7167210578918457,
+ "learning_rate": 1.1827251247239614e-06,
+ "loss": 0.0283,
+ "step": 31133
+ },
+ {
+ "epoch": 84.83378746594005,
+ "grad_norm": 0.3965725004673004,
+ "learning_rate": 1.1823088341937416e-06,
+ "loss": 0.0039,
+ "step": 31134
+ },
+ {
+ "epoch": 84.83651226158038,
+ "grad_norm": 1.0466104745864868,
+ "learning_rate": 1.1818926123354201e-06,
+ "loss": 0.0479,
+ "step": 31135
+ },
+ {
+ "epoch": 84.83923705722071,
+ "grad_norm": 0.7949053645133972,
+ "learning_rate": 1.181476459152232e-06,
+ "loss": 0.0086,
+ "step": 31136
+ },
+ {
+ "epoch": 84.84196185286103,
+ "grad_norm": 0.8172470927238464,
+ "learning_rate": 1.1810603746474235e-06,
+ "loss": 0.006,
+ "step": 31137
+ },
+ {
+ "epoch": 84.84468664850137,
+ "grad_norm": 1.8121659755706787,
+ "learning_rate": 1.1806443588242333e-06,
+ "loss": 0.0727,
+ "step": 31138
+ },
+ {
+ "epoch": 84.84741144414168,
+ "grad_norm": 1.0158679485321045,
+ "learning_rate": 1.1802284116859008e-06,
+ "loss": 0.0769,
+ "step": 31139
+ },
+ {
+ "epoch": 84.85013623978202,
+ "grad_norm": 1.2075536251068115,
+ "learning_rate": 1.1798125332356646e-06,
+ "loss": 0.0488,
+ "step": 31140
+ },
+ {
+ "epoch": 84.85286103542235,
+ "grad_norm": 0.6784634590148926,
+ "learning_rate": 1.1793967234767656e-06,
+ "loss": 0.0079,
+ "step": 31141
+ },
+ {
+ "epoch": 84.85558583106267,
+ "grad_norm": 1.3443244695663452,
+ "learning_rate": 1.178980982412441e-06,
+ "loss": 0.0626,
+ "step": 31142
+ },
+ {
+ "epoch": 84.858310626703,
+ "grad_norm": 0.8007664084434509,
+ "learning_rate": 1.1785653100459282e-06,
+ "loss": 0.0099,
+ "step": 31143
+ },
+ {
+ "epoch": 84.86103542234332,
+ "grad_norm": 1.568280816078186,
+ "learning_rate": 1.1781497063804636e-06,
+ "loss": 0.0289,
+ "step": 31144
+ },
+ {
+ "epoch": 84.86376021798365,
+ "grad_norm": 0.9246842861175537,
+ "learning_rate": 1.177734171419287e-06,
+ "loss": 0.0226,
+ "step": 31145
+ },
+ {
+ "epoch": 84.86648501362397,
+ "grad_norm": 1.6642388105392456,
+ "learning_rate": 1.1773187051656332e-06,
+ "loss": 0.071,
+ "step": 31146
+ },
+ {
+ "epoch": 84.8692098092643,
+ "grad_norm": 1.0559399127960205,
+ "learning_rate": 1.1769033076227364e-06,
+ "loss": 0.0221,
+ "step": 31147
+ },
+ {
+ "epoch": 84.87193460490464,
+ "grad_norm": 0.9060742259025574,
+ "learning_rate": 1.1764879787938311e-06,
+ "loss": 0.0074,
+ "step": 31148
+ },
+ {
+ "epoch": 84.87465940054496,
+ "grad_norm": 1.2987346649169922,
+ "learning_rate": 1.1760727186821552e-06,
+ "loss": 0.0174,
+ "step": 31149
+ },
+ {
+ "epoch": 84.87738419618529,
+ "grad_norm": 0.8579551577568054,
+ "learning_rate": 1.175657527290941e-06,
+ "loss": 0.0169,
+ "step": 31150
+ },
+ {
+ "epoch": 84.88010899182561,
+ "grad_norm": 0.9618338942527771,
+ "learning_rate": 1.1752424046234223e-06,
+ "loss": 0.0136,
+ "step": 31151
+ },
+ {
+ "epoch": 84.88283378746594,
+ "grad_norm": 0.52611243724823,
+ "learning_rate": 1.1748273506828278e-06,
+ "loss": 0.0059,
+ "step": 31152
+ },
+ {
+ "epoch": 84.88555858310627,
+ "grad_norm": 1.8299561738967896,
+ "learning_rate": 1.174412365472397e-06,
+ "loss": 0.0174,
+ "step": 31153
+ },
+ {
+ "epoch": 84.88828337874659,
+ "grad_norm": 1.0032140016555786,
+ "learning_rate": 1.1739974489953598e-06,
+ "loss": 0.0098,
+ "step": 31154
+ },
+ {
+ "epoch": 84.89100817438693,
+ "grad_norm": 0.9496662020683289,
+ "learning_rate": 1.1735826012549457e-06,
+ "loss": 0.009,
+ "step": 31155
+ },
+ {
+ "epoch": 84.89373297002724,
+ "grad_norm": 2.1865246295928955,
+ "learning_rate": 1.1731678222543841e-06,
+ "loss": 0.0193,
+ "step": 31156
+ },
+ {
+ "epoch": 84.89645776566758,
+ "grad_norm": 1.7434521913528442,
+ "learning_rate": 1.1727531119969104e-06,
+ "loss": 0.0258,
+ "step": 31157
+ },
+ {
+ "epoch": 84.8991825613079,
+ "grad_norm": 0.477472722530365,
+ "learning_rate": 1.172338470485751e-06,
+ "loss": 0.0042,
+ "step": 31158
+ },
+ {
+ "epoch": 84.90190735694823,
+ "grad_norm": 1.5167977809906006,
+ "learning_rate": 1.1719238977241343e-06,
+ "loss": 0.0438,
+ "step": 31159
+ },
+ {
+ "epoch": 84.90463215258856,
+ "grad_norm": 1.4333313703536987,
+ "learning_rate": 1.1715093937152888e-06,
+ "loss": 0.0538,
+ "step": 31160
+ },
+ {
+ "epoch": 84.90735694822888,
+ "grad_norm": 1.3746347427368164,
+ "learning_rate": 1.1710949584624466e-06,
+ "loss": 0.0091,
+ "step": 31161
+ },
+ {
+ "epoch": 84.91008174386921,
+ "grad_norm": 0.7741885781288147,
+ "learning_rate": 1.1706805919688323e-06,
+ "loss": 0.0074,
+ "step": 31162
+ },
+ {
+ "epoch": 84.91280653950953,
+ "grad_norm": 1.8143165111541748,
+ "learning_rate": 1.1702662942376742e-06,
+ "loss": 0.0357,
+ "step": 31163
+ },
+ {
+ "epoch": 84.91553133514986,
+ "grad_norm": 1.6816905736923218,
+ "learning_rate": 1.1698520652721946e-06,
+ "loss": 0.0601,
+ "step": 31164
+ },
+ {
+ "epoch": 84.9182561307902,
+ "grad_norm": 1.8474605083465576,
+ "learning_rate": 1.1694379050756256e-06,
+ "loss": 0.032,
+ "step": 31165
+ },
+ {
+ "epoch": 84.92098092643052,
+ "grad_norm": 0.7626786828041077,
+ "learning_rate": 1.16902381365119e-06,
+ "loss": 0.0074,
+ "step": 31166
+ },
+ {
+ "epoch": 84.92370572207085,
+ "grad_norm": 0.7633988857269287,
+ "learning_rate": 1.168609791002112e-06,
+ "loss": 0.0095,
+ "step": 31167
+ },
+ {
+ "epoch": 84.92643051771117,
+ "grad_norm": 0.86881023645401,
+ "learning_rate": 1.168195837131616e-06,
+ "loss": 0.0157,
+ "step": 31168
+ },
+ {
+ "epoch": 84.9291553133515,
+ "grad_norm": 1.6492201089859009,
+ "learning_rate": 1.1677819520429258e-06,
+ "loss": 0.0407,
+ "step": 31169
+ },
+ {
+ "epoch": 84.93188010899182,
+ "grad_norm": 1.5903904438018799,
+ "learning_rate": 1.1673681357392674e-06,
+ "loss": 0.1054,
+ "step": 31170
+ },
+ {
+ "epoch": 84.93460490463215,
+ "grad_norm": 0.89569091796875,
+ "learning_rate": 1.166954388223862e-06,
+ "loss": 0.0313,
+ "step": 31171
+ },
+ {
+ "epoch": 84.93732970027249,
+ "grad_norm": 0.6418753862380981,
+ "learning_rate": 1.1665407094999292e-06,
+ "loss": 0.0069,
+ "step": 31172
+ },
+ {
+ "epoch": 84.9400544959128,
+ "grad_norm": 1.4064894914627075,
+ "learning_rate": 1.1661270995706953e-06,
+ "loss": 0.1093,
+ "step": 31173
+ },
+ {
+ "epoch": 84.94277929155314,
+ "grad_norm": 0.7610468864440918,
+ "learning_rate": 1.1657135584393797e-06,
+ "loss": 0.0067,
+ "step": 31174
+ },
+ {
+ "epoch": 84.94550408719346,
+ "grad_norm": 1.04702627658844,
+ "learning_rate": 1.1653000861092012e-06,
+ "loss": 0.0146,
+ "step": 31175
+ },
+ {
+ "epoch": 84.94822888283379,
+ "grad_norm": 0.9159866571426392,
+ "learning_rate": 1.1648866825833804e-06,
+ "loss": 0.1774,
+ "step": 31176
+ },
+ {
+ "epoch": 84.95095367847412,
+ "grad_norm": 0.9258944988250732,
+ "learning_rate": 1.1644733478651394e-06,
+ "loss": 0.007,
+ "step": 31177
+ },
+ {
+ "epoch": 84.95367847411444,
+ "grad_norm": 1.139818549156189,
+ "learning_rate": 1.1640600819576965e-06,
+ "loss": 0.0096,
+ "step": 31178
+ },
+ {
+ "epoch": 84.95640326975477,
+ "grad_norm": 0.8216949105262756,
+ "learning_rate": 1.163646884864268e-06,
+ "loss": 0.0154,
+ "step": 31179
+ },
+ {
+ "epoch": 84.95912806539509,
+ "grad_norm": 1.4497795104980469,
+ "learning_rate": 1.1632337565880715e-06,
+ "loss": 0.0092,
+ "step": 31180
+ },
+ {
+ "epoch": 84.96185286103542,
+ "grad_norm": 0.7397109866142273,
+ "learning_rate": 1.1628206971323287e-06,
+ "loss": 0.006,
+ "step": 31181
+ },
+ {
+ "epoch": 84.96457765667574,
+ "grad_norm": 1.3853650093078613,
+ "learning_rate": 1.1624077065002537e-06,
+ "loss": 0.0274,
+ "step": 31182
+ },
+ {
+ "epoch": 84.96730245231608,
+ "grad_norm": 0.7777881622314453,
+ "learning_rate": 1.161994784695063e-06,
+ "loss": 0.0068,
+ "step": 31183
+ },
+ {
+ "epoch": 84.97002724795641,
+ "grad_norm": 0.40646815299987793,
+ "learning_rate": 1.1615819317199696e-06,
+ "loss": 0.0045,
+ "step": 31184
+ },
+ {
+ "epoch": 84.97275204359673,
+ "grad_norm": 1.221697211265564,
+ "learning_rate": 1.161169147578195e-06,
+ "loss": 0.0139,
+ "step": 31185
+ },
+ {
+ "epoch": 84.97547683923706,
+ "grad_norm": 0.8579499125480652,
+ "learning_rate": 1.160756432272947e-06,
+ "loss": 0.1324,
+ "step": 31186
+ },
+ {
+ "epoch": 84.97820163487738,
+ "grad_norm": 1.0202383995056152,
+ "learning_rate": 1.1603437858074463e-06,
+ "loss": 0.0135,
+ "step": 31187
+ },
+ {
+ "epoch": 84.98092643051771,
+ "grad_norm": 1.2611113786697388,
+ "learning_rate": 1.1599312081849046e-06,
+ "loss": 0.0277,
+ "step": 31188
+ },
+ {
+ "epoch": 84.98365122615803,
+ "grad_norm": 0.7578103542327881,
+ "learning_rate": 1.1595186994085306e-06,
+ "loss": 0.0116,
+ "step": 31189
+ },
+ {
+ "epoch": 84.98637602179836,
+ "grad_norm": 1.4286226034164429,
+ "learning_rate": 1.159106259481544e-06,
+ "loss": 0.0555,
+ "step": 31190
+ },
+ {
+ "epoch": 84.9891008174387,
+ "grad_norm": 0.892857015132904,
+ "learning_rate": 1.1586938884071519e-06,
+ "loss": 0.018,
+ "step": 31191
+ },
+ {
+ "epoch": 84.99182561307902,
+ "grad_norm": 1.4298133850097656,
+ "learning_rate": 1.1582815861885676e-06,
+ "loss": 0.0697,
+ "step": 31192
+ },
+ {
+ "epoch": 84.99455040871935,
+ "grad_norm": 1.539121389389038,
+ "learning_rate": 1.1578693528289998e-06,
+ "loss": 0.0506,
+ "step": 31193
+ },
+ {
+ "epoch": 84.99727520435967,
+ "grad_norm": 1.2559641599655151,
+ "learning_rate": 1.1574571883316632e-06,
+ "loss": 0.0344,
+ "step": 31194
+ },
+ {
+ "epoch": 85.0,
+ "grad_norm": 0.6562543511390686,
+ "learning_rate": 1.1570450926997657e-06,
+ "loss": 0.0074,
+ "step": 31195
+ },
+ {
+ "epoch": 85.00272479564033,
+ "grad_norm": 1.5147266387939453,
+ "learning_rate": 1.1566330659365134e-06,
+ "loss": 0.0377,
+ "step": 31196
+ },
+ {
+ "epoch": 85.00544959128065,
+ "grad_norm": 0.559665322303772,
+ "learning_rate": 1.1562211080451201e-06,
+ "loss": 0.0086,
+ "step": 31197
+ },
+ {
+ "epoch": 85.00817438692098,
+ "grad_norm": 0.8267061114311218,
+ "learning_rate": 1.155809219028794e-06,
+ "loss": 0.0068,
+ "step": 31198
+ },
+ {
+ "epoch": 85.0108991825613,
+ "grad_norm": 0.9955828785896301,
+ "learning_rate": 1.1553973988907397e-06,
+ "loss": 0.0163,
+ "step": 31199
+ },
+ {
+ "epoch": 85.01362397820164,
+ "grad_norm": 1.646280288696289,
+ "learning_rate": 1.1549856476341637e-06,
+ "loss": 0.0171,
+ "step": 31200
+ },
+ {
+ "epoch": 85.01634877384195,
+ "grad_norm": 1.2356786727905273,
+ "learning_rate": 1.1545739652622768e-06,
+ "loss": 0.0254,
+ "step": 31201
+ },
+ {
+ "epoch": 85.01907356948229,
+ "grad_norm": 1.1800451278686523,
+ "learning_rate": 1.154162351778283e-06,
+ "loss": 0.0119,
+ "step": 31202
+ },
+ {
+ "epoch": 85.02179836512262,
+ "grad_norm": 1.1486557722091675,
+ "learning_rate": 1.1537508071853865e-06,
+ "loss": 0.0085,
+ "step": 31203
+ },
+ {
+ "epoch": 85.02452316076294,
+ "grad_norm": 1.4382379055023193,
+ "learning_rate": 1.1533393314867959e-06,
+ "loss": 0.056,
+ "step": 31204
+ },
+ {
+ "epoch": 85.02724795640327,
+ "grad_norm": 1.2685670852661133,
+ "learning_rate": 1.1529279246857129e-06,
+ "loss": 0.136,
+ "step": 31205
+ },
+ {
+ "epoch": 85.02997275204359,
+ "grad_norm": 1.72891104221344,
+ "learning_rate": 1.1525165867853427e-06,
+ "loss": 0.0512,
+ "step": 31206
+ },
+ {
+ "epoch": 85.03269754768392,
+ "grad_norm": 1.6946566104888916,
+ "learning_rate": 1.1521053177888897e-06,
+ "loss": 0.0126,
+ "step": 31207
+ },
+ {
+ "epoch": 85.03542234332426,
+ "grad_norm": 1.971754789352417,
+ "learning_rate": 1.1516941176995555e-06,
+ "loss": 0.0197,
+ "step": 31208
+ },
+ {
+ "epoch": 85.03814713896458,
+ "grad_norm": 1.3832604885101318,
+ "learning_rate": 1.1512829865205399e-06,
+ "loss": 0.0173,
+ "step": 31209
+ },
+ {
+ "epoch": 85.04087193460491,
+ "grad_norm": 2.0269651412963867,
+ "learning_rate": 1.1508719242550504e-06,
+ "loss": 0.0755,
+ "step": 31210
+ },
+ {
+ "epoch": 85.04359673024523,
+ "grad_norm": 0.495909720659256,
+ "learning_rate": 1.1504609309062865e-06,
+ "loss": 0.0049,
+ "step": 31211
+ },
+ {
+ "epoch": 85.04632152588556,
+ "grad_norm": 1.2400718927383423,
+ "learning_rate": 1.1500500064774468e-06,
+ "loss": 0.0163,
+ "step": 31212
+ },
+ {
+ "epoch": 85.04904632152588,
+ "grad_norm": 0.8926257491111755,
+ "learning_rate": 1.149639150971731e-06,
+ "loss": 0.0138,
+ "step": 31213
+ },
+ {
+ "epoch": 85.05177111716621,
+ "grad_norm": 1.0885889530181885,
+ "learning_rate": 1.149228364392343e-06,
+ "loss": 0.0138,
+ "step": 31214
+ },
+ {
+ "epoch": 85.05449591280654,
+ "grad_norm": 0.8240970373153687,
+ "learning_rate": 1.1488176467424783e-06,
+ "loss": 0.0126,
+ "step": 31215
+ },
+ {
+ "epoch": 85.05722070844686,
+ "grad_norm": 0.869529664516449,
+ "learning_rate": 1.1484069980253365e-06,
+ "loss": 0.0098,
+ "step": 31216
+ },
+ {
+ "epoch": 85.0599455040872,
+ "grad_norm": 1.5607807636260986,
+ "learning_rate": 1.1479964182441172e-06,
+ "loss": 0.0795,
+ "step": 31217
+ },
+ {
+ "epoch": 85.06267029972751,
+ "grad_norm": 0.878281831741333,
+ "learning_rate": 1.147585907402018e-06,
+ "loss": 0.0171,
+ "step": 31218
+ },
+ {
+ "epoch": 85.06539509536785,
+ "grad_norm": 1.4919626712799072,
+ "learning_rate": 1.1471754655022338e-06,
+ "loss": 0.0595,
+ "step": 31219
+ },
+ {
+ "epoch": 85.06811989100818,
+ "grad_norm": 0.8982182145118713,
+ "learning_rate": 1.14676509254796e-06,
+ "loss": 0.0111,
+ "step": 31220
+ },
+ {
+ "epoch": 85.0708446866485,
+ "grad_norm": 4.877945899963379,
+ "learning_rate": 1.1463547885423953e-06,
+ "loss": 0.0101,
+ "step": 31221
+ },
+ {
+ "epoch": 85.07356948228883,
+ "grad_norm": 1.2318229675292969,
+ "learning_rate": 1.145944553488736e-06,
+ "loss": 0.1608,
+ "step": 31222
+ },
+ {
+ "epoch": 85.07629427792915,
+ "grad_norm": 1.4625179767608643,
+ "learning_rate": 1.1455343873901758e-06,
+ "loss": 0.1091,
+ "step": 31223
+ },
+ {
+ "epoch": 85.07901907356948,
+ "grad_norm": 1.722320795059204,
+ "learning_rate": 1.1451242902499093e-06,
+ "loss": 0.0709,
+ "step": 31224
+ },
+ {
+ "epoch": 85.0817438692098,
+ "grad_norm": 0.9054408669471741,
+ "learning_rate": 1.1447142620711282e-06,
+ "loss": 0.0245,
+ "step": 31225
+ },
+ {
+ "epoch": 85.08446866485014,
+ "grad_norm": 1.2324929237365723,
+ "learning_rate": 1.1443043028570277e-06,
+ "loss": 0.0825,
+ "step": 31226
+ },
+ {
+ "epoch": 85.08719346049047,
+ "grad_norm": 1.0026990175247192,
+ "learning_rate": 1.143894412610802e-06,
+ "loss": 0.0155,
+ "step": 31227
+ },
+ {
+ "epoch": 85.08991825613079,
+ "grad_norm": 1.062003493309021,
+ "learning_rate": 1.1434845913356407e-06,
+ "loss": 0.0108,
+ "step": 31228
+ },
+ {
+ "epoch": 85.09264305177112,
+ "grad_norm": 1.5287526845932007,
+ "learning_rate": 1.1430748390347345e-06,
+ "loss": 0.0225,
+ "step": 31229
+ },
+ {
+ "epoch": 85.09536784741144,
+ "grad_norm": 0.9297279119491577,
+ "learning_rate": 1.142665155711279e-06,
+ "loss": 0.0093,
+ "step": 31230
+ },
+ {
+ "epoch": 85.09809264305177,
+ "grad_norm": 0.9198514819145203,
+ "learning_rate": 1.142255541368461e-06,
+ "loss": 0.012,
+ "step": 31231
+ },
+ {
+ "epoch": 85.1008174386921,
+ "grad_norm": 1.7017067670822144,
+ "learning_rate": 1.141845996009472e-06,
+ "loss": 0.0618,
+ "step": 31232
+ },
+ {
+ "epoch": 85.10354223433242,
+ "grad_norm": 1.3498274087905884,
+ "learning_rate": 1.1414365196374999e-06,
+ "loss": 0.0197,
+ "step": 31233
+ },
+ {
+ "epoch": 85.10626702997276,
+ "grad_norm": 1.046231746673584,
+ "learning_rate": 1.141027112255737e-06,
+ "loss": 0.0059,
+ "step": 31234
+ },
+ {
+ "epoch": 85.10899182561307,
+ "grad_norm": 1.024294376373291,
+ "learning_rate": 1.1406177738673697e-06,
+ "loss": 0.0288,
+ "step": 31235
+ },
+ {
+ "epoch": 85.11171662125341,
+ "grad_norm": 0.9585621356964111,
+ "learning_rate": 1.1402085044755862e-06,
+ "loss": 0.0086,
+ "step": 31236
+ },
+ {
+ "epoch": 85.11444141689373,
+ "grad_norm": 1.2749395370483398,
+ "learning_rate": 1.139799304083571e-06,
+ "loss": 0.0255,
+ "step": 31237
+ },
+ {
+ "epoch": 85.11716621253406,
+ "grad_norm": 0.9385104775428772,
+ "learning_rate": 1.1393901726945145e-06,
+ "loss": 0.0158,
+ "step": 31238
+ },
+ {
+ "epoch": 85.11989100817439,
+ "grad_norm": 1.044340968132019,
+ "learning_rate": 1.1389811103116032e-06,
+ "loss": 0.0161,
+ "step": 31239
+ },
+ {
+ "epoch": 85.12261580381471,
+ "grad_norm": 1.3397059440612793,
+ "learning_rate": 1.1385721169380226e-06,
+ "loss": 0.1078,
+ "step": 31240
+ },
+ {
+ "epoch": 85.12534059945504,
+ "grad_norm": 0.6321120262145996,
+ "learning_rate": 1.1381631925769555e-06,
+ "loss": 0.0056,
+ "step": 31241
+ },
+ {
+ "epoch": 85.12806539509536,
+ "grad_norm": 0.858486533164978,
+ "learning_rate": 1.1377543372315892e-06,
+ "loss": 0.0796,
+ "step": 31242
+ },
+ {
+ "epoch": 85.1307901907357,
+ "grad_norm": 0.9462739825248718,
+ "learning_rate": 1.1373455509051078e-06,
+ "loss": 0.0062,
+ "step": 31243
+ },
+ {
+ "epoch": 85.13351498637603,
+ "grad_norm": 0.709806501865387,
+ "learning_rate": 1.1369368336006937e-06,
+ "loss": 0.008,
+ "step": 31244
+ },
+ {
+ "epoch": 85.13623978201635,
+ "grad_norm": 2.75942325592041,
+ "learning_rate": 1.136528185321528e-06,
+ "loss": 0.1047,
+ "step": 31245
+ },
+ {
+ "epoch": 85.13896457765668,
+ "grad_norm": 0.8314111828804016,
+ "learning_rate": 1.136119606070798e-06,
+ "loss": 0.0059,
+ "step": 31246
+ },
+ {
+ "epoch": 85.141689373297,
+ "grad_norm": 0.5960561633110046,
+ "learning_rate": 1.1357110958516826e-06,
+ "loss": 0.0047,
+ "step": 31247
+ },
+ {
+ "epoch": 85.14441416893733,
+ "grad_norm": 0.7847698330879211,
+ "learning_rate": 1.135302654667364e-06,
+ "loss": 0.0707,
+ "step": 31248
+ },
+ {
+ "epoch": 85.14713896457765,
+ "grad_norm": 0.7571911215782166,
+ "learning_rate": 1.1348942825210218e-06,
+ "loss": 0.0055,
+ "step": 31249
+ },
+ {
+ "epoch": 85.14986376021798,
+ "grad_norm": 1.88889479637146,
+ "learning_rate": 1.1344859794158391e-06,
+ "loss": 0.1152,
+ "step": 31250
+ },
+ {
+ "epoch": 85.15258855585832,
+ "grad_norm": 1.2619813680648804,
+ "learning_rate": 1.1340777453549944e-06,
+ "loss": 0.034,
+ "step": 31251
+ },
+ {
+ "epoch": 85.15531335149863,
+ "grad_norm": 1.628435492515564,
+ "learning_rate": 1.1336695803416664e-06,
+ "loss": 0.0179,
+ "step": 31252
+ },
+ {
+ "epoch": 85.15803814713897,
+ "grad_norm": 1.3770197629928589,
+ "learning_rate": 1.1332614843790325e-06,
+ "loss": 0.0055,
+ "step": 31253
+ },
+ {
+ "epoch": 85.16076294277929,
+ "grad_norm": 1.8619600534439087,
+ "learning_rate": 1.1328534574702755e-06,
+ "loss": 0.0364,
+ "step": 31254
+ },
+ {
+ "epoch": 85.16348773841962,
+ "grad_norm": 1.2890480756759644,
+ "learning_rate": 1.1324454996185673e-06,
+ "loss": 0.0335,
+ "step": 31255
+ },
+ {
+ "epoch": 85.16621253405995,
+ "grad_norm": 0.8751192092895508,
+ "learning_rate": 1.1320376108270904e-06,
+ "loss": 0.0077,
+ "step": 31256
+ },
+ {
+ "epoch": 85.16893732970027,
+ "grad_norm": 0.7375828623771667,
+ "learning_rate": 1.1316297910990203e-06,
+ "loss": 0.0076,
+ "step": 31257
+ },
+ {
+ "epoch": 85.1716621253406,
+ "grad_norm": 0.724890410900116,
+ "learning_rate": 1.1312220404375285e-06,
+ "loss": 0.0067,
+ "step": 31258
+ },
+ {
+ "epoch": 85.17438692098092,
+ "grad_norm": 1.526516318321228,
+ "learning_rate": 1.1308143588457965e-06,
+ "loss": 0.1292,
+ "step": 31259
+ },
+ {
+ "epoch": 85.17711171662125,
+ "grad_norm": 0.9521564245223999,
+ "learning_rate": 1.1304067463269974e-06,
+ "loss": 0.0191,
+ "step": 31260
+ },
+ {
+ "epoch": 85.17983651226157,
+ "grad_norm": 0.8176760077476501,
+ "learning_rate": 1.129999202884302e-06,
+ "loss": 0.007,
+ "step": 31261
+ },
+ {
+ "epoch": 85.1825613079019,
+ "grad_norm": 1.4293429851531982,
+ "learning_rate": 1.1295917285208901e-06,
+ "loss": 0.0144,
+ "step": 31262
+ },
+ {
+ "epoch": 85.18528610354224,
+ "grad_norm": 1.7097827196121216,
+ "learning_rate": 1.1291843232399324e-06,
+ "loss": 0.0139,
+ "step": 31263
+ },
+ {
+ "epoch": 85.18801089918256,
+ "grad_norm": 2.4034557342529297,
+ "learning_rate": 1.128776987044602e-06,
+ "loss": 0.0208,
+ "step": 31264
+ },
+ {
+ "epoch": 85.19073569482289,
+ "grad_norm": 1.518538236618042,
+ "learning_rate": 1.1283697199380683e-06,
+ "loss": 0.0143,
+ "step": 31265
+ },
+ {
+ "epoch": 85.19346049046321,
+ "grad_norm": 1.222566843032837,
+ "learning_rate": 1.1279625219235068e-06,
+ "loss": 0.0825,
+ "step": 31266
+ },
+ {
+ "epoch": 85.19618528610354,
+ "grad_norm": 0.5043854117393494,
+ "learning_rate": 1.1275553930040894e-06,
+ "loss": 0.0064,
+ "step": 31267
+ },
+ {
+ "epoch": 85.19891008174388,
+ "grad_norm": 0.5467373728752136,
+ "learning_rate": 1.1271483331829835e-06,
+ "loss": 0.005,
+ "step": 31268
+ },
+ {
+ "epoch": 85.2016348773842,
+ "grad_norm": 0.6758986115455627,
+ "learning_rate": 1.12674134246336e-06,
+ "loss": 0.0074,
+ "step": 31269
+ },
+ {
+ "epoch": 85.20435967302453,
+ "grad_norm": 1.555763602256775,
+ "learning_rate": 1.1263344208483906e-06,
+ "loss": 0.0268,
+ "step": 31270
+ },
+ {
+ "epoch": 85.20708446866485,
+ "grad_norm": 0.5402599573135376,
+ "learning_rate": 1.125927568341244e-06,
+ "loss": 0.0045,
+ "step": 31271
+ },
+ {
+ "epoch": 85.20980926430518,
+ "grad_norm": 0.6453306078910828,
+ "learning_rate": 1.1255207849450845e-06,
+ "loss": 0.0057,
+ "step": 31272
+ },
+ {
+ "epoch": 85.2125340599455,
+ "grad_norm": 0.9196082353591919,
+ "learning_rate": 1.125114070663087e-06,
+ "loss": 0.0403,
+ "step": 31273
+ },
+ {
+ "epoch": 85.21525885558583,
+ "grad_norm": 0.6898474097251892,
+ "learning_rate": 1.1247074254984126e-06,
+ "loss": 0.0052,
+ "step": 31274
+ },
+ {
+ "epoch": 85.21798365122616,
+ "grad_norm": 0.8540601134300232,
+ "learning_rate": 1.1243008494542328e-06,
+ "loss": 0.0049,
+ "step": 31275
+ },
+ {
+ "epoch": 85.22070844686648,
+ "grad_norm": 2.6517152786254883,
+ "learning_rate": 1.1238943425337135e-06,
+ "loss": 0.0209,
+ "step": 31276
+ },
+ {
+ "epoch": 85.22343324250681,
+ "grad_norm": 1.4674535989761353,
+ "learning_rate": 1.1234879047400183e-06,
+ "loss": 0.1304,
+ "step": 31277
+ },
+ {
+ "epoch": 85.22615803814713,
+ "grad_norm": 0.7983243465423584,
+ "learning_rate": 1.1230815360763126e-06,
+ "loss": 0.0122,
+ "step": 31278
+ },
+ {
+ "epoch": 85.22888283378747,
+ "grad_norm": 1.6707452535629272,
+ "learning_rate": 1.1226752365457626e-06,
+ "loss": 0.0143,
+ "step": 31279
+ },
+ {
+ "epoch": 85.2316076294278,
+ "grad_norm": 0.9845320582389832,
+ "learning_rate": 1.122269006151534e-06,
+ "loss": 0.0203,
+ "step": 31280
+ },
+ {
+ "epoch": 85.23433242506812,
+ "grad_norm": 0.9067668318748474,
+ "learning_rate": 1.121862844896786e-06,
+ "loss": 0.0106,
+ "step": 31281
+ },
+ {
+ "epoch": 85.23705722070845,
+ "grad_norm": 1.4346855878829956,
+ "learning_rate": 1.1214567527846854e-06,
+ "loss": 0.0827,
+ "step": 31282
+ },
+ {
+ "epoch": 85.23978201634877,
+ "grad_norm": 1.3215043544769287,
+ "learning_rate": 1.1210507298183948e-06,
+ "loss": 0.0182,
+ "step": 31283
+ },
+ {
+ "epoch": 85.2425068119891,
+ "grad_norm": 1.4627442359924316,
+ "learning_rate": 1.1206447760010752e-06,
+ "loss": 0.0445,
+ "step": 31284
+ },
+ {
+ "epoch": 85.24523160762942,
+ "grad_norm": 1.3294557332992554,
+ "learning_rate": 1.1202388913358863e-06,
+ "loss": 0.011,
+ "step": 31285
+ },
+ {
+ "epoch": 85.24795640326975,
+ "grad_norm": 0.9747486710548401,
+ "learning_rate": 1.1198330758259934e-06,
+ "loss": 0.0147,
+ "step": 31286
+ },
+ {
+ "epoch": 85.25068119891009,
+ "grad_norm": 0.908835768699646,
+ "learning_rate": 1.119427329474555e-06,
+ "loss": 0.0132,
+ "step": 31287
+ },
+ {
+ "epoch": 85.2534059945504,
+ "grad_norm": 1.5156389474868774,
+ "learning_rate": 1.1190216522847297e-06,
+ "loss": 0.0268,
+ "step": 31288
+ },
+ {
+ "epoch": 85.25613079019074,
+ "grad_norm": 1.1198335886001587,
+ "learning_rate": 1.118616044259676e-06,
+ "loss": 0.0221,
+ "step": 31289
+ },
+ {
+ "epoch": 85.25885558583106,
+ "grad_norm": 1.1143964529037476,
+ "learning_rate": 1.118210505402556e-06,
+ "loss": 0.0164,
+ "step": 31290
+ },
+ {
+ "epoch": 85.26158038147139,
+ "grad_norm": 1.5469872951507568,
+ "learning_rate": 1.117805035716527e-06,
+ "loss": 0.0127,
+ "step": 31291
+ },
+ {
+ "epoch": 85.26430517711172,
+ "grad_norm": 0.8624209761619568,
+ "learning_rate": 1.117399635204748e-06,
+ "loss": 0.0091,
+ "step": 31292
+ },
+ {
+ "epoch": 85.26702997275204,
+ "grad_norm": 0.633934736251831,
+ "learning_rate": 1.1169943038703734e-06,
+ "loss": 0.0058,
+ "step": 31293
+ },
+ {
+ "epoch": 85.26975476839237,
+ "grad_norm": 0.4474845230579376,
+ "learning_rate": 1.1165890417165604e-06,
+ "loss": 0.0051,
+ "step": 31294
+ },
+ {
+ "epoch": 85.2724795640327,
+ "grad_norm": 1.06611168384552,
+ "learning_rate": 1.1161838487464682e-06,
+ "loss": 0.0119,
+ "step": 31295
+ },
+ {
+ "epoch": 85.27520435967303,
+ "grad_norm": 1.0876250267028809,
+ "learning_rate": 1.115778724963249e-06,
+ "loss": 0.0241,
+ "step": 31296
+ },
+ {
+ "epoch": 85.27792915531334,
+ "grad_norm": 0.9268489480018616,
+ "learning_rate": 1.1153736703700602e-06,
+ "loss": 0.0083,
+ "step": 31297
+ },
+ {
+ "epoch": 85.28065395095368,
+ "grad_norm": 0.9602741599082947,
+ "learning_rate": 1.1149686849700524e-06,
+ "loss": 0.0098,
+ "step": 31298
+ },
+ {
+ "epoch": 85.28337874659401,
+ "grad_norm": 1.0003422498703003,
+ "learning_rate": 1.1145637687663856e-06,
+ "loss": 0.0118,
+ "step": 31299
+ },
+ {
+ "epoch": 85.28610354223433,
+ "grad_norm": 1.2877683639526367,
+ "learning_rate": 1.1141589217622096e-06,
+ "loss": 0.0098,
+ "step": 31300
+ },
+ {
+ "epoch": 85.28882833787466,
+ "grad_norm": 1.282645344734192,
+ "learning_rate": 1.1137541439606781e-06,
+ "loss": 0.0736,
+ "step": 31301
+ },
+ {
+ "epoch": 85.29155313351498,
+ "grad_norm": 0.7351586222648621,
+ "learning_rate": 1.113349435364941e-06,
+ "loss": 0.0077,
+ "step": 31302
+ },
+ {
+ "epoch": 85.29427792915531,
+ "grad_norm": 1.0556094646453857,
+ "learning_rate": 1.1129447959781546e-06,
+ "loss": 0.0054,
+ "step": 31303
+ },
+ {
+ "epoch": 85.29700272479565,
+ "grad_norm": 1.2971222400665283,
+ "learning_rate": 1.112540225803468e-06,
+ "loss": 0.0235,
+ "step": 31304
+ },
+ {
+ "epoch": 85.29972752043597,
+ "grad_norm": 0.6883981227874756,
+ "learning_rate": 1.1121357248440301e-06,
+ "loss": 0.0062,
+ "step": 31305
+ },
+ {
+ "epoch": 85.3024523160763,
+ "grad_norm": 0.9038981199264526,
+ "learning_rate": 1.1117312931029933e-06,
+ "loss": 0.0054,
+ "step": 31306
+ },
+ {
+ "epoch": 85.30517711171662,
+ "grad_norm": 0.9057982563972473,
+ "learning_rate": 1.1113269305835095e-06,
+ "loss": 0.1452,
+ "step": 31307
+ },
+ {
+ "epoch": 85.30790190735695,
+ "grad_norm": 0.7290861010551453,
+ "learning_rate": 1.110922637288725e-06,
+ "loss": 0.007,
+ "step": 31308
+ },
+ {
+ "epoch": 85.31062670299727,
+ "grad_norm": 0.941266655921936,
+ "learning_rate": 1.1105184132217894e-06,
+ "loss": 0.0098,
+ "step": 31309
+ },
+ {
+ "epoch": 85.3133514986376,
+ "grad_norm": 1.39170503616333,
+ "learning_rate": 1.110114258385847e-06,
+ "loss": 0.0517,
+ "step": 31310
+ },
+ {
+ "epoch": 85.31607629427793,
+ "grad_norm": 1.1438125371932983,
+ "learning_rate": 1.109710172784052e-06,
+ "loss": 0.0376,
+ "step": 31311
+ },
+ {
+ "epoch": 85.31880108991825,
+ "grad_norm": 1.6600253582000732,
+ "learning_rate": 1.1093061564195473e-06,
+ "loss": 0.1351,
+ "step": 31312
+ },
+ {
+ "epoch": 85.32152588555859,
+ "grad_norm": 1.1361488103866577,
+ "learning_rate": 1.1089022092954804e-06,
+ "loss": 0.0398,
+ "step": 31313
+ },
+ {
+ "epoch": 85.3242506811989,
+ "grad_norm": 0.5304917097091675,
+ "learning_rate": 1.1084983314149956e-06,
+ "loss": 0.0053,
+ "step": 31314
+ },
+ {
+ "epoch": 85.32697547683924,
+ "grad_norm": 1.7508974075317383,
+ "learning_rate": 1.1080945227812401e-06,
+ "loss": 0.0206,
+ "step": 31315
+ },
+ {
+ "epoch": 85.32970027247957,
+ "grad_norm": 1.2379913330078125,
+ "learning_rate": 1.1076907833973594e-06,
+ "loss": 0.0088,
+ "step": 31316
+ },
+ {
+ "epoch": 85.33242506811989,
+ "grad_norm": 1.1726566553115845,
+ "learning_rate": 1.1072871132664964e-06,
+ "loss": 0.0082,
+ "step": 31317
+ },
+ {
+ "epoch": 85.33514986376022,
+ "grad_norm": 0.8424919247627258,
+ "learning_rate": 1.1068835123917932e-06,
+ "loss": 0.0085,
+ "step": 31318
+ },
+ {
+ "epoch": 85.33787465940054,
+ "grad_norm": 1.6758794784545898,
+ "learning_rate": 1.106479980776397e-06,
+ "loss": 0.0359,
+ "step": 31319
+ },
+ {
+ "epoch": 85.34059945504087,
+ "grad_norm": 1.0206469297409058,
+ "learning_rate": 1.106076518423449e-06,
+ "loss": 0.0189,
+ "step": 31320
+ },
+ {
+ "epoch": 85.34332425068119,
+ "grad_norm": 0.5040510892868042,
+ "learning_rate": 1.1056731253360896e-06,
+ "loss": 0.0068,
+ "step": 31321
+ },
+ {
+ "epoch": 85.34604904632153,
+ "grad_norm": 1.2238179445266724,
+ "learning_rate": 1.1052698015174601e-06,
+ "loss": 0.0202,
+ "step": 31322
+ },
+ {
+ "epoch": 85.34877384196186,
+ "grad_norm": 1.0197378396987915,
+ "learning_rate": 1.1048665469707055e-06,
+ "loss": 0.038,
+ "step": 31323
+ },
+ {
+ "epoch": 85.35149863760218,
+ "grad_norm": 0.9196767807006836,
+ "learning_rate": 1.104463361698962e-06,
+ "loss": 0.0198,
+ "step": 31324
+ },
+ {
+ "epoch": 85.35422343324251,
+ "grad_norm": 1.1197291612625122,
+ "learning_rate": 1.1040602457053729e-06,
+ "loss": 0.0212,
+ "step": 31325
+ },
+ {
+ "epoch": 85.35694822888283,
+ "grad_norm": 1.1385339498519897,
+ "learning_rate": 1.1036571989930744e-06,
+ "loss": 0.0128,
+ "step": 31326
+ },
+ {
+ "epoch": 85.35967302452316,
+ "grad_norm": 1.5205588340759277,
+ "learning_rate": 1.1032542215652098e-06,
+ "loss": 0.0335,
+ "step": 31327
+ },
+ {
+ "epoch": 85.3623978201635,
+ "grad_norm": 0.6602927446365356,
+ "learning_rate": 1.1028513134249153e-06,
+ "loss": 0.0046,
+ "step": 31328
+ },
+ {
+ "epoch": 85.36512261580381,
+ "grad_norm": 1.087900996208191,
+ "learning_rate": 1.1024484745753271e-06,
+ "loss": 0.0088,
+ "step": 31329
+ },
+ {
+ "epoch": 85.36784741144415,
+ "grad_norm": 0.8521842360496521,
+ "learning_rate": 1.1020457050195833e-06,
+ "loss": 0.0072,
+ "step": 31330
+ },
+ {
+ "epoch": 85.37057220708446,
+ "grad_norm": 1.2879387140274048,
+ "learning_rate": 1.1016430047608218e-06,
+ "loss": 0.0938,
+ "step": 31331
+ },
+ {
+ "epoch": 85.3732970027248,
+ "grad_norm": 0.6345471739768982,
+ "learning_rate": 1.1012403738021793e-06,
+ "loss": 0.0068,
+ "step": 31332
+ },
+ {
+ "epoch": 85.37602179836512,
+ "grad_norm": 1.2401373386383057,
+ "learning_rate": 1.10083781214679e-06,
+ "loss": 0.0484,
+ "step": 31333
+ },
+ {
+ "epoch": 85.37874659400545,
+ "grad_norm": 0.817585825920105,
+ "learning_rate": 1.1004353197977869e-06,
+ "loss": 0.0067,
+ "step": 31334
+ },
+ {
+ "epoch": 85.38147138964578,
+ "grad_norm": 0.9332095980644226,
+ "learning_rate": 1.1000328967583096e-06,
+ "loss": 0.0104,
+ "step": 31335
+ },
+ {
+ "epoch": 85.3841961852861,
+ "grad_norm": 1.1839679479599,
+ "learning_rate": 1.099630543031489e-06,
+ "loss": 0.0827,
+ "step": 31336
+ },
+ {
+ "epoch": 85.38692098092643,
+ "grad_norm": 0.9901526570320129,
+ "learning_rate": 1.0992282586204605e-06,
+ "loss": 0.0075,
+ "step": 31337
+ },
+ {
+ "epoch": 85.38964577656675,
+ "grad_norm": 1.014120101928711,
+ "learning_rate": 1.0988260435283526e-06,
+ "loss": 0.0087,
+ "step": 31338
+ },
+ {
+ "epoch": 85.39237057220708,
+ "grad_norm": 1.4698179960250854,
+ "learning_rate": 1.098423897758304e-06,
+ "loss": 0.0072,
+ "step": 31339
+ },
+ {
+ "epoch": 85.39509536784742,
+ "grad_norm": 1.1160107851028442,
+ "learning_rate": 1.098021821313443e-06,
+ "loss": 0.0191,
+ "step": 31340
+ },
+ {
+ "epoch": 85.39782016348774,
+ "grad_norm": 0.9701451063156128,
+ "learning_rate": 1.0976198141968998e-06,
+ "loss": 0.0438,
+ "step": 31341
+ },
+ {
+ "epoch": 85.40054495912807,
+ "grad_norm": 0.7426602244377136,
+ "learning_rate": 1.097217876411808e-06,
+ "loss": 0.0122,
+ "step": 31342
+ },
+ {
+ "epoch": 85.40326975476839,
+ "grad_norm": 2.132826089859009,
+ "learning_rate": 1.0968160079612966e-06,
+ "loss": 0.0421,
+ "step": 31343
+ },
+ {
+ "epoch": 85.40599455040872,
+ "grad_norm": 0.8677119612693787,
+ "learning_rate": 1.0964142088484963e-06,
+ "loss": 0.0089,
+ "step": 31344
+ },
+ {
+ "epoch": 85.40871934604904,
+ "grad_norm": 1.820024847984314,
+ "learning_rate": 1.0960124790765358e-06,
+ "loss": 0.0144,
+ "step": 31345
+ },
+ {
+ "epoch": 85.41144414168937,
+ "grad_norm": 0.5429341197013855,
+ "learning_rate": 1.0956108186485414e-06,
+ "loss": 0.0069,
+ "step": 31346
+ },
+ {
+ "epoch": 85.4141689373297,
+ "grad_norm": 1.4624090194702148,
+ "learning_rate": 1.0952092275676461e-06,
+ "loss": 0.057,
+ "step": 31347
+ },
+ {
+ "epoch": 85.41689373297002,
+ "grad_norm": 1.5363203287124634,
+ "learning_rate": 1.0948077058369744e-06,
+ "loss": 0.0101,
+ "step": 31348
+ },
+ {
+ "epoch": 85.41961852861036,
+ "grad_norm": 0.8766846060752869,
+ "learning_rate": 1.0944062534596534e-06,
+ "loss": 0.0084,
+ "step": 31349
+ },
+ {
+ "epoch": 85.42234332425068,
+ "grad_norm": 2.5367066860198975,
+ "learning_rate": 1.0940048704388085e-06,
+ "loss": 0.0492,
+ "step": 31350
+ },
+ {
+ "epoch": 85.42506811989101,
+ "grad_norm": 1.2901103496551514,
+ "learning_rate": 1.0936035567775682e-06,
+ "loss": 0.0146,
+ "step": 31351
+ },
+ {
+ "epoch": 85.42779291553134,
+ "grad_norm": 0.9336767792701721,
+ "learning_rate": 1.093202312479057e-06,
+ "loss": 0.012,
+ "step": 31352
+ },
+ {
+ "epoch": 85.43051771117166,
+ "grad_norm": 1.2381517887115479,
+ "learning_rate": 1.0928011375464e-06,
+ "loss": 0.0175,
+ "step": 31353
+ },
+ {
+ "epoch": 85.433242506812,
+ "grad_norm": 0.8661911487579346,
+ "learning_rate": 1.0924000319827189e-06,
+ "loss": 0.0082,
+ "step": 31354
+ },
+ {
+ "epoch": 85.43596730245231,
+ "grad_norm": 0.9827131628990173,
+ "learning_rate": 1.0919989957911414e-06,
+ "loss": 0.0491,
+ "step": 31355
+ },
+ {
+ "epoch": 85.43869209809264,
+ "grad_norm": 1.440603256225586,
+ "learning_rate": 1.0915980289747885e-06,
+ "loss": 0.0384,
+ "step": 31356
+ },
+ {
+ "epoch": 85.44141689373296,
+ "grad_norm": 0.6387438774108887,
+ "learning_rate": 1.0911971315367842e-06,
+ "loss": 0.0069,
+ "step": 31357
+ },
+ {
+ "epoch": 85.4441416893733,
+ "grad_norm": 1.1313254833221436,
+ "learning_rate": 1.0907963034802472e-06,
+ "loss": 0.0091,
+ "step": 31358
+ },
+ {
+ "epoch": 85.44686648501363,
+ "grad_norm": 0.7624623775482178,
+ "learning_rate": 1.0903955448083025e-06,
+ "loss": 0.0099,
+ "step": 31359
+ },
+ {
+ "epoch": 85.44959128065395,
+ "grad_norm": 0.9544327259063721,
+ "learning_rate": 1.0899948555240724e-06,
+ "loss": 0.0327,
+ "step": 31360
+ },
+ {
+ "epoch": 85.45231607629428,
+ "grad_norm": 1.6015321016311646,
+ "learning_rate": 1.0895942356306754e-06,
+ "loss": 0.0144,
+ "step": 31361
+ },
+ {
+ "epoch": 85.4550408719346,
+ "grad_norm": 1.674625039100647,
+ "learning_rate": 1.0891936851312325e-06,
+ "loss": 0.0265,
+ "step": 31362
+ },
+ {
+ "epoch": 85.45776566757493,
+ "grad_norm": 1.0801703929901123,
+ "learning_rate": 1.0887932040288597e-06,
+ "loss": 0.0166,
+ "step": 31363
+ },
+ {
+ "epoch": 85.46049046321527,
+ "grad_norm": 1.5174745321273804,
+ "learning_rate": 1.0883927923266813e-06,
+ "loss": 0.0666,
+ "step": 31364
+ },
+ {
+ "epoch": 85.46321525885558,
+ "grad_norm": 1.575509786605835,
+ "learning_rate": 1.0879924500278116e-06,
+ "loss": 0.0136,
+ "step": 31365
+ },
+ {
+ "epoch": 85.46594005449592,
+ "grad_norm": 0.9654614925384521,
+ "learning_rate": 1.0875921771353714e-06,
+ "loss": 0.007,
+ "step": 31366
+ },
+ {
+ "epoch": 85.46866485013624,
+ "grad_norm": 2.5991885662078857,
+ "learning_rate": 1.0871919736524739e-06,
+ "loss": 0.0144,
+ "step": 31367
+ },
+ {
+ "epoch": 85.47138964577657,
+ "grad_norm": 0.7285441160202026,
+ "learning_rate": 1.0867918395822397e-06,
+ "loss": 0.0073,
+ "step": 31368
+ },
+ {
+ "epoch": 85.47411444141689,
+ "grad_norm": 0.9438849091529846,
+ "learning_rate": 1.0863917749277852e-06,
+ "loss": 0.0836,
+ "step": 31369
+ },
+ {
+ "epoch": 85.47683923705722,
+ "grad_norm": 1.4366567134857178,
+ "learning_rate": 1.0859917796922216e-06,
+ "loss": 0.0257,
+ "step": 31370
+ },
+ {
+ "epoch": 85.47956403269755,
+ "grad_norm": 1.1012587547302246,
+ "learning_rate": 1.0855918538786692e-06,
+ "loss": 0.0048,
+ "step": 31371
+ },
+ {
+ "epoch": 85.48228882833787,
+ "grad_norm": 1.2474981546401978,
+ "learning_rate": 1.0851919974902403e-06,
+ "loss": 0.009,
+ "step": 31372
+ },
+ {
+ "epoch": 85.4850136239782,
+ "grad_norm": 1.08516526222229,
+ "learning_rate": 1.0847922105300502e-06,
+ "loss": 0.0136,
+ "step": 31373
+ },
+ {
+ "epoch": 85.48773841961852,
+ "grad_norm": 1.2749546766281128,
+ "learning_rate": 1.0843924930012084e-06,
+ "loss": 0.0651,
+ "step": 31374
+ },
+ {
+ "epoch": 85.49046321525886,
+ "grad_norm": 1.1379833221435547,
+ "learning_rate": 1.0839928449068316e-06,
+ "loss": 0.0099,
+ "step": 31375
+ },
+ {
+ "epoch": 85.49318801089919,
+ "grad_norm": 0.8127520680427551,
+ "learning_rate": 1.0835932662500338e-06,
+ "loss": 0.009,
+ "step": 31376
+ },
+ {
+ "epoch": 85.49591280653951,
+ "grad_norm": 1.1080358028411865,
+ "learning_rate": 1.0831937570339247e-06,
+ "loss": 0.0083,
+ "step": 31377
+ },
+ {
+ "epoch": 85.49863760217984,
+ "grad_norm": 1.05815589427948,
+ "learning_rate": 1.082794317261615e-06,
+ "loss": 0.0519,
+ "step": 31378
+ },
+ {
+ "epoch": 85.50136239782016,
+ "grad_norm": 1.1020056009292603,
+ "learning_rate": 1.0823949469362138e-06,
+ "loss": 0.0594,
+ "step": 31379
+ },
+ {
+ "epoch": 85.50408719346049,
+ "grad_norm": 1.1328026056289673,
+ "learning_rate": 1.081995646060836e-06,
+ "loss": 0.012,
+ "step": 31380
+ },
+ {
+ "epoch": 85.50681198910081,
+ "grad_norm": 1.3704164028167725,
+ "learning_rate": 1.0815964146385894e-06,
+ "loss": 0.0158,
+ "step": 31381
+ },
+ {
+ "epoch": 85.50953678474114,
+ "grad_norm": 0.857265830039978,
+ "learning_rate": 1.0811972526725834e-06,
+ "loss": 0.0085,
+ "step": 31382
+ },
+ {
+ "epoch": 85.51226158038148,
+ "grad_norm": 1.463172197341919,
+ "learning_rate": 1.0807981601659234e-06,
+ "loss": 0.0269,
+ "step": 31383
+ },
+ {
+ "epoch": 85.5149863760218,
+ "grad_norm": 1.4180221557617188,
+ "learning_rate": 1.0803991371217227e-06,
+ "loss": 0.0096,
+ "step": 31384
+ },
+ {
+ "epoch": 85.51771117166213,
+ "grad_norm": 1.4399727582931519,
+ "learning_rate": 1.0800001835430861e-06,
+ "loss": 0.0301,
+ "step": 31385
+ },
+ {
+ "epoch": 85.52043596730245,
+ "grad_norm": 0.9841528534889221,
+ "learning_rate": 1.0796012994331207e-06,
+ "loss": 0.0209,
+ "step": 31386
+ },
+ {
+ "epoch": 85.52316076294278,
+ "grad_norm": 0.6076793670654297,
+ "learning_rate": 1.0792024847949323e-06,
+ "loss": 0.0079,
+ "step": 31387
+ },
+ {
+ "epoch": 85.52588555858311,
+ "grad_norm": 0.6537055969238281,
+ "learning_rate": 1.0788037396316287e-06,
+ "loss": 0.0053,
+ "step": 31388
+ },
+ {
+ "epoch": 85.52861035422343,
+ "grad_norm": 1.7559527158737183,
+ "learning_rate": 1.0784050639463151e-06,
+ "loss": 0.0137,
+ "step": 31389
+ },
+ {
+ "epoch": 85.53133514986376,
+ "grad_norm": 0.6575019359588623,
+ "learning_rate": 1.0780064577420935e-06,
+ "loss": 0.0046,
+ "step": 31390
+ },
+ {
+ "epoch": 85.53405994550408,
+ "grad_norm": 4.817774295806885,
+ "learning_rate": 1.0776079210220725e-06,
+ "loss": 0.0592,
+ "step": 31391
+ },
+ {
+ "epoch": 85.53678474114442,
+ "grad_norm": 0.7338019013404846,
+ "learning_rate": 1.0772094537893518e-06,
+ "loss": 0.0116,
+ "step": 31392
+ },
+ {
+ "epoch": 85.53950953678473,
+ "grad_norm": 1.134956955909729,
+ "learning_rate": 1.076811056047039e-06,
+ "loss": 0.0177,
+ "step": 31393
+ },
+ {
+ "epoch": 85.54223433242507,
+ "grad_norm": 1.7928904294967651,
+ "learning_rate": 1.076412727798234e-06,
+ "loss": 0.0813,
+ "step": 31394
+ },
+ {
+ "epoch": 85.5449591280654,
+ "grad_norm": 0.8251391649246216,
+ "learning_rate": 1.0760144690460384e-06,
+ "loss": 0.0331,
+ "step": 31395
+ },
+ {
+ "epoch": 85.54768392370572,
+ "grad_norm": 1.6246507167816162,
+ "learning_rate": 1.0756162797935555e-06,
+ "loss": 0.0732,
+ "step": 31396
+ },
+ {
+ "epoch": 85.55040871934605,
+ "grad_norm": 1.0550060272216797,
+ "learning_rate": 1.0752181600438872e-06,
+ "loss": 0.1485,
+ "step": 31397
+ },
+ {
+ "epoch": 85.55313351498637,
+ "grad_norm": 1.2187035083770752,
+ "learning_rate": 1.0748201098001322e-06,
+ "loss": 0.0134,
+ "step": 31398
+ },
+ {
+ "epoch": 85.5558583106267,
+ "grad_norm": 0.9677852988243103,
+ "learning_rate": 1.0744221290653877e-06,
+ "loss": 0.1242,
+ "step": 31399
+ },
+ {
+ "epoch": 85.55858310626704,
+ "grad_norm": 0.7046143412590027,
+ "learning_rate": 1.0740242178427595e-06,
+ "loss": 0.0044,
+ "step": 31400
+ },
+ {
+ "epoch": 85.56130790190736,
+ "grad_norm": 1.1405011415481567,
+ "learning_rate": 1.0736263761353437e-06,
+ "loss": 0.0421,
+ "step": 31401
+ },
+ {
+ "epoch": 85.56403269754769,
+ "grad_norm": 1.5311459302902222,
+ "learning_rate": 1.0732286039462381e-06,
+ "loss": 0.0646,
+ "step": 31402
+ },
+ {
+ "epoch": 85.566757493188,
+ "grad_norm": 0.9846664667129517,
+ "learning_rate": 1.0728309012785386e-06,
+ "loss": 0.0074,
+ "step": 31403
+ },
+ {
+ "epoch": 85.56948228882834,
+ "grad_norm": 1.7310078144073486,
+ "learning_rate": 1.0724332681353466e-06,
+ "loss": 0.0208,
+ "step": 31404
+ },
+ {
+ "epoch": 85.57220708446866,
+ "grad_norm": 3.406576633453369,
+ "learning_rate": 1.0720357045197572e-06,
+ "loss": 0.0106,
+ "step": 31405
+ },
+ {
+ "epoch": 85.57493188010899,
+ "grad_norm": 0.8771328330039978,
+ "learning_rate": 1.0716382104348667e-06,
+ "loss": 0.0132,
+ "step": 31406
+ },
+ {
+ "epoch": 85.57765667574932,
+ "grad_norm": 0.9722668528556824,
+ "learning_rate": 1.0712407858837682e-06,
+ "loss": 0.0344,
+ "step": 31407
+ },
+ {
+ "epoch": 85.58038147138964,
+ "grad_norm": 1.1241613626480103,
+ "learning_rate": 1.0708434308695604e-06,
+ "loss": 0.0265,
+ "step": 31408
+ },
+ {
+ "epoch": 85.58310626702998,
+ "grad_norm": 1.1643272638320923,
+ "learning_rate": 1.0704461453953373e-06,
+ "loss": 0.0479,
+ "step": 31409
+ },
+ {
+ "epoch": 85.5858310626703,
+ "grad_norm": 0.9776894450187683,
+ "learning_rate": 1.0700489294641902e-06,
+ "loss": 0.008,
+ "step": 31410
+ },
+ {
+ "epoch": 85.58855585831063,
+ "grad_norm": 0.9264628291130066,
+ "learning_rate": 1.0696517830792141e-06,
+ "loss": 0.0111,
+ "step": 31411
+ },
+ {
+ "epoch": 85.59128065395096,
+ "grad_norm": 0.37359619140625,
+ "learning_rate": 1.0692547062435056e-06,
+ "loss": 0.0041,
+ "step": 31412
+ },
+ {
+ "epoch": 85.59400544959128,
+ "grad_norm": 1.0998663902282715,
+ "learning_rate": 1.0688576989601528e-06,
+ "loss": 0.0166,
+ "step": 31413
+ },
+ {
+ "epoch": 85.59673024523161,
+ "grad_norm": 1.1344866752624512,
+ "learning_rate": 1.0684607612322495e-06,
+ "loss": 0.0117,
+ "step": 31414
+ },
+ {
+ "epoch": 85.59945504087193,
+ "grad_norm": 1.0739136934280396,
+ "learning_rate": 1.068063893062884e-06,
+ "loss": 0.0546,
+ "step": 31415
+ },
+ {
+ "epoch": 85.60217983651226,
+ "grad_norm": 1.5363625288009644,
+ "learning_rate": 1.0676670944551515e-06,
+ "loss": 0.0144,
+ "step": 31416
+ },
+ {
+ "epoch": 85.60490463215258,
+ "grad_norm": 1.5547716617584229,
+ "learning_rate": 1.0672703654121407e-06,
+ "loss": 0.0402,
+ "step": 31417
+ },
+ {
+ "epoch": 85.60762942779292,
+ "grad_norm": 0.8766106963157654,
+ "learning_rate": 1.0668737059369405e-06,
+ "loss": 0.0066,
+ "step": 31418
+ },
+ {
+ "epoch": 85.61035422343325,
+ "grad_norm": 1.3148001432418823,
+ "learning_rate": 1.0664771160326382e-06,
+ "loss": 0.0157,
+ "step": 31419
+ },
+ {
+ "epoch": 85.61307901907357,
+ "grad_norm": 1.4404197931289673,
+ "learning_rate": 1.0660805957023268e-06,
+ "loss": 0.0081,
+ "step": 31420
+ },
+ {
+ "epoch": 85.6158038147139,
+ "grad_norm": 0.9159865379333496,
+ "learning_rate": 1.0656841449490918e-06,
+ "loss": 0.0085,
+ "step": 31421
+ },
+ {
+ "epoch": 85.61852861035422,
+ "grad_norm": 0.8944001197814941,
+ "learning_rate": 1.0652877637760217e-06,
+ "loss": 0.0067,
+ "step": 31422
+ },
+ {
+ "epoch": 85.62125340599455,
+ "grad_norm": 1.0089422464370728,
+ "learning_rate": 1.0648914521862008e-06,
+ "loss": 0.0104,
+ "step": 31423
+ },
+ {
+ "epoch": 85.62397820163488,
+ "grad_norm": 1.1512165069580078,
+ "learning_rate": 1.064495210182719e-06,
+ "loss": 0.0177,
+ "step": 31424
+ },
+ {
+ "epoch": 85.6267029972752,
+ "grad_norm": 0.9117131233215332,
+ "learning_rate": 1.0640990377686611e-06,
+ "loss": 0.0112,
+ "step": 31425
+ },
+ {
+ "epoch": 85.62942779291554,
+ "grad_norm": 1.3069169521331787,
+ "learning_rate": 1.0637029349471128e-06,
+ "loss": 0.018,
+ "step": 31426
+ },
+ {
+ "epoch": 85.63215258855585,
+ "grad_norm": 0.8338037133216858,
+ "learning_rate": 1.0633069017211562e-06,
+ "loss": 0.0147,
+ "step": 31427
+ },
+ {
+ "epoch": 85.63487738419619,
+ "grad_norm": 1.1697087287902832,
+ "learning_rate": 1.0629109380938785e-06,
+ "loss": 0.0158,
+ "step": 31428
+ },
+ {
+ "epoch": 85.6376021798365,
+ "grad_norm": 0.8257501125335693,
+ "learning_rate": 1.062515044068364e-06,
+ "loss": 0.006,
+ "step": 31429
+ },
+ {
+ "epoch": 85.64032697547684,
+ "grad_norm": 1.2585846185684204,
+ "learning_rate": 1.0621192196476949e-06,
+ "loss": 0.015,
+ "step": 31430
+ },
+ {
+ "epoch": 85.64305177111717,
+ "grad_norm": 0.7280923128128052,
+ "learning_rate": 1.0617234648349506e-06,
+ "loss": 0.0153,
+ "step": 31431
+ },
+ {
+ "epoch": 85.64577656675749,
+ "grad_norm": 1.3031646013259888,
+ "learning_rate": 1.06132777963322e-06,
+ "loss": 0.0187,
+ "step": 31432
+ },
+ {
+ "epoch": 85.64850136239782,
+ "grad_norm": 1.0076415538787842,
+ "learning_rate": 1.0609321640455793e-06,
+ "loss": 0.0173,
+ "step": 31433
+ },
+ {
+ "epoch": 85.65122615803814,
+ "grad_norm": 0.6655715107917786,
+ "learning_rate": 1.060536618075113e-06,
+ "loss": 0.0058,
+ "step": 31434
+ },
+ {
+ "epoch": 85.65395095367847,
+ "grad_norm": 1.399652361869812,
+ "learning_rate": 1.0601411417248964e-06,
+ "loss": 0.0165,
+ "step": 31435
+ },
+ {
+ "epoch": 85.65667574931881,
+ "grad_norm": 1.1415435075759888,
+ "learning_rate": 1.0597457349980156e-06,
+ "loss": 0.0169,
+ "step": 31436
+ },
+ {
+ "epoch": 85.65940054495913,
+ "grad_norm": 0.9732296466827393,
+ "learning_rate": 1.0593503978975472e-06,
+ "loss": 0.0244,
+ "step": 31437
+ },
+ {
+ "epoch": 85.66212534059946,
+ "grad_norm": 0.7024714350700378,
+ "learning_rate": 1.05895513042657e-06,
+ "loss": 0.0112,
+ "step": 31438
+ },
+ {
+ "epoch": 85.66485013623978,
+ "grad_norm": 0.6480536460876465,
+ "learning_rate": 1.0585599325881602e-06,
+ "loss": 0.007,
+ "step": 31439
+ },
+ {
+ "epoch": 85.66757493188011,
+ "grad_norm": 0.788438081741333,
+ "learning_rate": 1.0581648043853999e-06,
+ "loss": 0.0091,
+ "step": 31440
+ },
+ {
+ "epoch": 85.67029972752043,
+ "grad_norm": 0.7240451574325562,
+ "learning_rate": 1.0577697458213654e-06,
+ "loss": 0.011,
+ "step": 31441
+ },
+ {
+ "epoch": 85.67302452316076,
+ "grad_norm": 1.8407646417617798,
+ "learning_rate": 1.0573747568991311e-06,
+ "loss": 0.0213,
+ "step": 31442
+ },
+ {
+ "epoch": 85.6757493188011,
+ "grad_norm": 0.9531087279319763,
+ "learning_rate": 1.0569798376217722e-06,
+ "loss": 0.1388,
+ "step": 31443
+ },
+ {
+ "epoch": 85.67847411444141,
+ "grad_norm": 0.509494423866272,
+ "learning_rate": 1.0565849879923673e-06,
+ "loss": 0.0057,
+ "step": 31444
+ },
+ {
+ "epoch": 85.68119891008175,
+ "grad_norm": 1.1784523725509644,
+ "learning_rate": 1.056190208013993e-06,
+ "loss": 0.0397,
+ "step": 31445
+ },
+ {
+ "epoch": 85.68392370572207,
+ "grad_norm": 0.9302405118942261,
+ "learning_rate": 1.05579549768972e-06,
+ "loss": 0.0135,
+ "step": 31446
+ },
+ {
+ "epoch": 85.6866485013624,
+ "grad_norm": 0.7948907017707825,
+ "learning_rate": 1.0554008570226248e-06,
+ "loss": 0.0434,
+ "step": 31447
+ },
+ {
+ "epoch": 85.68937329700273,
+ "grad_norm": 0.9181851744651794,
+ "learning_rate": 1.0550062860157783e-06,
+ "loss": 0.0166,
+ "step": 31448
+ },
+ {
+ "epoch": 85.69209809264305,
+ "grad_norm": 1.363887906074524,
+ "learning_rate": 1.054611784672256e-06,
+ "loss": 0.0223,
+ "step": 31449
+ },
+ {
+ "epoch": 85.69482288828338,
+ "grad_norm": 1.3371474742889404,
+ "learning_rate": 1.0542173529951293e-06,
+ "loss": 0.0295,
+ "step": 31450
+ },
+ {
+ "epoch": 85.6975476839237,
+ "grad_norm": 2.003787040710449,
+ "learning_rate": 1.053822990987471e-06,
+ "loss": 0.1245,
+ "step": 31451
+ },
+ {
+ "epoch": 85.70027247956403,
+ "grad_norm": 1.1479905843734741,
+ "learning_rate": 1.053428698652349e-06,
+ "loss": 0.0125,
+ "step": 31452
+ },
+ {
+ "epoch": 85.70299727520435,
+ "grad_norm": 1.0084344148635864,
+ "learning_rate": 1.0530344759928369e-06,
+ "loss": 0.0058,
+ "step": 31453
+ },
+ {
+ "epoch": 85.70572207084469,
+ "grad_norm": 0.8427611589431763,
+ "learning_rate": 1.0526403230120064e-06,
+ "loss": 0.0112,
+ "step": 31454
+ },
+ {
+ "epoch": 85.70844686648502,
+ "grad_norm": 0.7903887033462524,
+ "learning_rate": 1.0522462397129218e-06,
+ "loss": 0.0051,
+ "step": 31455
+ },
+ {
+ "epoch": 85.71117166212534,
+ "grad_norm": 1.518173336982727,
+ "learning_rate": 1.0518522260986575e-06,
+ "loss": 0.0077,
+ "step": 31456
+ },
+ {
+ "epoch": 85.71389645776567,
+ "grad_norm": 1.2908216714859009,
+ "learning_rate": 1.0514582821722807e-06,
+ "loss": 0.0846,
+ "step": 31457
+ },
+ {
+ "epoch": 85.71662125340599,
+ "grad_norm": 0.826878011226654,
+ "learning_rate": 1.0510644079368582e-06,
+ "loss": 0.0076,
+ "step": 31458
+ },
+ {
+ "epoch": 85.71934604904632,
+ "grad_norm": 0.8074092864990234,
+ "learning_rate": 1.0506706033954561e-06,
+ "loss": 0.0067,
+ "step": 31459
+ },
+ {
+ "epoch": 85.72207084468666,
+ "grad_norm": 1.2539403438568115,
+ "learning_rate": 1.0502768685511456e-06,
+ "loss": 0.021,
+ "step": 31460
+ },
+ {
+ "epoch": 85.72479564032697,
+ "grad_norm": 0.5878051519393921,
+ "learning_rate": 1.0498832034069884e-06,
+ "loss": 0.0046,
+ "step": 31461
+ },
+ {
+ "epoch": 85.7275204359673,
+ "grad_norm": 0.7396246194839478,
+ "learning_rate": 1.0494896079660554e-06,
+ "loss": 0.0069,
+ "step": 31462
+ },
+ {
+ "epoch": 85.73024523160763,
+ "grad_norm": 1.0443394184112549,
+ "learning_rate": 1.049096082231409e-06,
+ "loss": 0.0202,
+ "step": 31463
+ },
+ {
+ "epoch": 85.73297002724796,
+ "grad_norm": 0.6267593502998352,
+ "learning_rate": 1.0487026262061117e-06,
+ "loss": 0.0047,
+ "step": 31464
+ },
+ {
+ "epoch": 85.73569482288828,
+ "grad_norm": 4.909656524658203,
+ "learning_rate": 1.0483092398932326e-06,
+ "loss": 0.0429,
+ "step": 31465
+ },
+ {
+ "epoch": 85.73841961852861,
+ "grad_norm": 0.8071418404579163,
+ "learning_rate": 1.0479159232958337e-06,
+ "loss": 0.1065,
+ "step": 31466
+ },
+ {
+ "epoch": 85.74114441416894,
+ "grad_norm": 2.2030701637268066,
+ "learning_rate": 1.0475226764169766e-06,
+ "loss": 0.0223,
+ "step": 31467
+ },
+ {
+ "epoch": 85.74386920980926,
+ "grad_norm": 0.5813393592834473,
+ "learning_rate": 1.0471294992597237e-06,
+ "loss": 0.0054,
+ "step": 31468
+ },
+ {
+ "epoch": 85.7465940054496,
+ "grad_norm": 0.7720832228660583,
+ "learning_rate": 1.046736391827139e-06,
+ "loss": 0.0066,
+ "step": 31469
+ },
+ {
+ "epoch": 85.74931880108991,
+ "grad_norm": 2.358271360397339,
+ "learning_rate": 1.0463433541222834e-06,
+ "loss": 0.0288,
+ "step": 31470
+ },
+ {
+ "epoch": 85.75204359673025,
+ "grad_norm": 1.404632329940796,
+ "learning_rate": 1.0459503861482179e-06,
+ "loss": 0.0139,
+ "step": 31471
+ },
+ {
+ "epoch": 85.75476839237058,
+ "grad_norm": 0.9746274352073669,
+ "learning_rate": 1.0455574879079999e-06,
+ "loss": 0.0149,
+ "step": 31472
+ },
+ {
+ "epoch": 85.7574931880109,
+ "grad_norm": 1.225077748298645,
+ "learning_rate": 1.045164659404695e-06,
+ "loss": 0.0249,
+ "step": 31473
+ },
+ {
+ "epoch": 85.76021798365123,
+ "grad_norm": 1.0654075145721436,
+ "learning_rate": 1.0447719006413592e-06,
+ "loss": 0.0294,
+ "step": 31474
+ },
+ {
+ "epoch": 85.76294277929155,
+ "grad_norm": 1.1040514707565308,
+ "learning_rate": 1.0443792116210515e-06,
+ "loss": 0.0119,
+ "step": 31475
+ },
+ {
+ "epoch": 85.76566757493188,
+ "grad_norm": 0.7977851629257202,
+ "learning_rate": 1.0439865923468284e-06,
+ "loss": 0.0086,
+ "step": 31476
+ },
+ {
+ "epoch": 85.7683923705722,
+ "grad_norm": 1.1109875440597534,
+ "learning_rate": 1.0435940428217516e-06,
+ "loss": 0.0126,
+ "step": 31477
+ },
+ {
+ "epoch": 85.77111716621253,
+ "grad_norm": 1.108444094657898,
+ "learning_rate": 1.0432015630488756e-06,
+ "loss": 0.0518,
+ "step": 31478
+ },
+ {
+ "epoch": 85.77384196185287,
+ "grad_norm": 1.268106460571289,
+ "learning_rate": 1.0428091530312556e-06,
+ "loss": 0.0122,
+ "step": 31479
+ },
+ {
+ "epoch": 85.77656675749319,
+ "grad_norm": 2.081590175628662,
+ "learning_rate": 1.0424168127719502e-06,
+ "loss": 0.1572,
+ "step": 31480
+ },
+ {
+ "epoch": 85.77929155313352,
+ "grad_norm": 1.024396538734436,
+ "learning_rate": 1.042024542274016e-06,
+ "loss": 0.0091,
+ "step": 31481
+ },
+ {
+ "epoch": 85.78201634877384,
+ "grad_norm": 1.1975317001342773,
+ "learning_rate": 1.0416323415405061e-06,
+ "loss": 0.1046,
+ "step": 31482
+ },
+ {
+ "epoch": 85.78474114441417,
+ "grad_norm": 1.3260650634765625,
+ "learning_rate": 1.0412402105744756e-06,
+ "loss": 0.0096,
+ "step": 31483
+ },
+ {
+ "epoch": 85.7874659400545,
+ "grad_norm": 1.229056477546692,
+ "learning_rate": 1.0408481493789758e-06,
+ "loss": 0.0616,
+ "step": 31484
+ },
+ {
+ "epoch": 85.79019073569482,
+ "grad_norm": 1.1463392972946167,
+ "learning_rate": 1.040456157957065e-06,
+ "loss": 0.0073,
+ "step": 31485
+ },
+ {
+ "epoch": 85.79291553133515,
+ "grad_norm": 3.232442617416382,
+ "learning_rate": 1.0400642363117918e-06,
+ "loss": 0.0126,
+ "step": 31486
+ },
+ {
+ "epoch": 85.79564032697547,
+ "grad_norm": 0.6024051308631897,
+ "learning_rate": 1.039672384446211e-06,
+ "loss": 0.0069,
+ "step": 31487
+ },
+ {
+ "epoch": 85.7983651226158,
+ "grad_norm": 1.3646776676177979,
+ "learning_rate": 1.0392806023633717e-06,
+ "loss": 0.0173,
+ "step": 31488
+ },
+ {
+ "epoch": 85.80108991825612,
+ "grad_norm": 1.1334049701690674,
+ "learning_rate": 1.0388888900663274e-06,
+ "loss": 0.1016,
+ "step": 31489
+ },
+ {
+ "epoch": 85.80381471389646,
+ "grad_norm": 1.1729451417922974,
+ "learning_rate": 1.038497247558129e-06,
+ "loss": 0.0084,
+ "step": 31490
+ },
+ {
+ "epoch": 85.80653950953679,
+ "grad_norm": 1.3898309469223022,
+ "learning_rate": 1.038105674841825e-06,
+ "loss": 0.0243,
+ "step": 31491
+ },
+ {
+ "epoch": 85.80926430517711,
+ "grad_norm": 1.6650149822235107,
+ "learning_rate": 1.0377141719204642e-06,
+ "loss": 0.1511,
+ "step": 31492
+ },
+ {
+ "epoch": 85.81198910081744,
+ "grad_norm": 0.6353366374969482,
+ "learning_rate": 1.0373227387970986e-06,
+ "loss": 0.0058,
+ "step": 31493
+ },
+ {
+ "epoch": 85.81471389645776,
+ "grad_norm": 1.6170036792755127,
+ "learning_rate": 1.0369313754747744e-06,
+ "loss": 0.0209,
+ "step": 31494
+ },
+ {
+ "epoch": 85.8174386920981,
+ "grad_norm": 1.2113056182861328,
+ "learning_rate": 1.0365400819565407e-06,
+ "loss": 0.0132,
+ "step": 31495
+ },
+ {
+ "epoch": 85.82016348773843,
+ "grad_norm": 1.1956473588943481,
+ "learning_rate": 1.0361488582454404e-06,
+ "loss": 0.032,
+ "step": 31496
+ },
+ {
+ "epoch": 85.82288828337875,
+ "grad_norm": 1.1701991558074951,
+ "learning_rate": 1.0357577043445287e-06,
+ "loss": 0.0097,
+ "step": 31497
+ },
+ {
+ "epoch": 85.82561307901908,
+ "grad_norm": 0.6444112658500671,
+ "learning_rate": 1.0353666202568468e-06,
+ "loss": 0.0066,
+ "step": 31498
+ },
+ {
+ "epoch": 85.8283378746594,
+ "grad_norm": 1.3697189092636108,
+ "learning_rate": 1.0349756059854421e-06,
+ "loss": 0.0419,
+ "step": 31499
+ },
+ {
+ "epoch": 85.83106267029973,
+ "grad_norm": 0.8953949809074402,
+ "learning_rate": 1.0345846615333565e-06,
+ "loss": 0.0094,
+ "step": 31500
+ },
+ {
+ "epoch": 85.83378746594005,
+ "grad_norm": 1.8621463775634766,
+ "learning_rate": 1.034193786903639e-06,
+ "loss": 0.1024,
+ "step": 31501
+ },
+ {
+ "epoch": 85.83651226158038,
+ "grad_norm": 1.3461792469024658,
+ "learning_rate": 1.033802982099331e-06,
+ "loss": 0.025,
+ "step": 31502
+ },
+ {
+ "epoch": 85.83923705722071,
+ "grad_norm": 1.8028372526168823,
+ "learning_rate": 1.0334122471234775e-06,
+ "loss": 0.0807,
+ "step": 31503
+ },
+ {
+ "epoch": 85.84196185286103,
+ "grad_norm": 1.0706132650375366,
+ "learning_rate": 1.033021581979118e-06,
+ "loss": 0.0309,
+ "step": 31504
+ },
+ {
+ "epoch": 85.84468664850137,
+ "grad_norm": 1.4863368272781372,
+ "learning_rate": 1.0326309866693007e-06,
+ "loss": 0.1021,
+ "step": 31505
+ },
+ {
+ "epoch": 85.84741144414168,
+ "grad_norm": 0.9284864664077759,
+ "learning_rate": 1.032240461197065e-06,
+ "loss": 0.0063,
+ "step": 31506
+ },
+ {
+ "epoch": 85.85013623978202,
+ "grad_norm": 0.8061333298683167,
+ "learning_rate": 1.0318500055654513e-06,
+ "loss": 0.0086,
+ "step": 31507
+ },
+ {
+ "epoch": 85.85286103542235,
+ "grad_norm": 1.2789382934570312,
+ "learning_rate": 1.0314596197774985e-06,
+ "loss": 0.0369,
+ "step": 31508
+ },
+ {
+ "epoch": 85.85558583106267,
+ "grad_norm": 1.0718834400177002,
+ "learning_rate": 1.0310693038362518e-06,
+ "loss": 0.0142,
+ "step": 31509
+ },
+ {
+ "epoch": 85.858310626703,
+ "grad_norm": 1.7509987354278564,
+ "learning_rate": 1.03067905774475e-06,
+ "loss": 0.0107,
+ "step": 31510
+ },
+ {
+ "epoch": 85.86103542234332,
+ "grad_norm": 1.5223422050476074,
+ "learning_rate": 1.0302888815060297e-06,
+ "loss": 0.0396,
+ "step": 31511
+ },
+ {
+ "epoch": 85.86376021798365,
+ "grad_norm": 0.7160951495170593,
+ "learning_rate": 1.0298987751231293e-06,
+ "loss": 0.0071,
+ "step": 31512
+ },
+ {
+ "epoch": 85.86648501362397,
+ "grad_norm": 1.673683524131775,
+ "learning_rate": 1.0295087385990888e-06,
+ "loss": 0.1485,
+ "step": 31513
+ },
+ {
+ "epoch": 85.8692098092643,
+ "grad_norm": 1.6162478923797607,
+ "learning_rate": 1.0291187719369466e-06,
+ "loss": 0.0115,
+ "step": 31514
+ },
+ {
+ "epoch": 85.87193460490464,
+ "grad_norm": 1.0814405679702759,
+ "learning_rate": 1.0287288751397395e-06,
+ "loss": 0.0151,
+ "step": 31515
+ },
+ {
+ "epoch": 85.87465940054496,
+ "grad_norm": 1.5862916707992554,
+ "learning_rate": 1.0283390482105037e-06,
+ "loss": 0.0096,
+ "step": 31516
+ },
+ {
+ "epoch": 85.87738419618529,
+ "grad_norm": 0.8689847588539124,
+ "learning_rate": 1.0279492911522714e-06,
+ "loss": 0.0107,
+ "step": 31517
+ },
+ {
+ "epoch": 85.88010899182561,
+ "grad_norm": 0.9142242074012756,
+ "learning_rate": 1.0275596039680835e-06,
+ "loss": 0.0107,
+ "step": 31518
+ },
+ {
+ "epoch": 85.88283378746594,
+ "grad_norm": 1.278174638748169,
+ "learning_rate": 1.027169986660973e-06,
+ "loss": 0.01,
+ "step": 31519
+ },
+ {
+ "epoch": 85.88555858310627,
+ "grad_norm": 0.7877478003501892,
+ "learning_rate": 1.026780439233972e-06,
+ "loss": 0.0116,
+ "step": 31520
+ },
+ {
+ "epoch": 85.88828337874659,
+ "grad_norm": 0.560499370098114,
+ "learning_rate": 1.026390961690118e-06,
+ "loss": 0.0039,
+ "step": 31521
+ },
+ {
+ "epoch": 85.89100817438693,
+ "grad_norm": 1.0667134523391724,
+ "learning_rate": 1.026001554032443e-06,
+ "loss": 0.0154,
+ "step": 31522
+ },
+ {
+ "epoch": 85.89373297002724,
+ "grad_norm": 1.3615150451660156,
+ "learning_rate": 1.025612216263978e-06,
+ "loss": 0.018,
+ "step": 31523
+ },
+ {
+ "epoch": 85.89645776566758,
+ "grad_norm": 0.7733775973320007,
+ "learning_rate": 1.0252229483877551e-06,
+ "loss": 0.0076,
+ "step": 31524
+ },
+ {
+ "epoch": 85.8991825613079,
+ "grad_norm": 0.9764891862869263,
+ "learning_rate": 1.0248337504068085e-06,
+ "loss": 0.0133,
+ "step": 31525
+ },
+ {
+ "epoch": 85.90190735694823,
+ "grad_norm": 0.8878510594367981,
+ "learning_rate": 1.0244446223241677e-06,
+ "loss": 0.0173,
+ "step": 31526
+ },
+ {
+ "epoch": 85.90463215258856,
+ "grad_norm": 0.3766728639602661,
+ "learning_rate": 1.024055564142863e-06,
+ "loss": 0.0036,
+ "step": 31527
+ },
+ {
+ "epoch": 85.90735694822888,
+ "grad_norm": 1.3693338632583618,
+ "learning_rate": 1.0236665758659237e-06,
+ "loss": 0.0252,
+ "step": 31528
+ },
+ {
+ "epoch": 85.91008174386921,
+ "grad_norm": 0.5933798551559448,
+ "learning_rate": 1.023277657496381e-06,
+ "loss": 0.0075,
+ "step": 31529
+ },
+ {
+ "epoch": 85.91280653950953,
+ "grad_norm": 0.8660727143287659,
+ "learning_rate": 1.0228888090372613e-06,
+ "loss": 0.0059,
+ "step": 31530
+ },
+ {
+ "epoch": 85.91553133514986,
+ "grad_norm": 1.4373294115066528,
+ "learning_rate": 1.0225000304915977e-06,
+ "loss": 0.0135,
+ "step": 31531
+ },
+ {
+ "epoch": 85.9182561307902,
+ "grad_norm": 1.6974674463272095,
+ "learning_rate": 1.0221113218624134e-06,
+ "loss": 0.0609,
+ "step": 31532
+ },
+ {
+ "epoch": 85.92098092643052,
+ "grad_norm": 0.8808004260063171,
+ "learning_rate": 1.021722683152736e-06,
+ "loss": 0.0123,
+ "step": 31533
+ },
+ {
+ "epoch": 85.92370572207085,
+ "grad_norm": 0.7785600423812866,
+ "learning_rate": 1.0213341143655952e-06,
+ "loss": 0.0083,
+ "step": 31534
+ },
+ {
+ "epoch": 85.92643051771117,
+ "grad_norm": 1.6208388805389404,
+ "learning_rate": 1.0209456155040166e-06,
+ "loss": 0.0975,
+ "step": 31535
+ },
+ {
+ "epoch": 85.9291553133515,
+ "grad_norm": 1.3021880388259888,
+ "learning_rate": 1.020557186571023e-06,
+ "loss": 0.0131,
+ "step": 31536
+ },
+ {
+ "epoch": 85.93188010899182,
+ "grad_norm": 3.1269917488098145,
+ "learning_rate": 1.0201688275696397e-06,
+ "loss": 0.026,
+ "step": 31537
+ },
+ {
+ "epoch": 85.93460490463215,
+ "grad_norm": 1.0592831373214722,
+ "learning_rate": 1.0197805385028935e-06,
+ "loss": 0.0087,
+ "step": 31538
+ },
+ {
+ "epoch": 85.93732970027249,
+ "grad_norm": 1.14744234085083,
+ "learning_rate": 1.0193923193738087e-06,
+ "loss": 0.0198,
+ "step": 31539
+ },
+ {
+ "epoch": 85.9400544959128,
+ "grad_norm": 0.6289823055267334,
+ "learning_rate": 1.019004170185407e-06,
+ "loss": 0.0068,
+ "step": 31540
+ },
+ {
+ "epoch": 85.94277929155314,
+ "grad_norm": 0.8241512179374695,
+ "learning_rate": 1.0186160909407095e-06,
+ "loss": 0.0066,
+ "step": 31541
+ },
+ {
+ "epoch": 85.94550408719346,
+ "grad_norm": 1.7259780168533325,
+ "learning_rate": 1.0182280816427436e-06,
+ "loss": 0.0875,
+ "step": 31542
+ },
+ {
+ "epoch": 85.94822888283379,
+ "grad_norm": 0.6579462885856628,
+ "learning_rate": 1.0178401422945273e-06,
+ "loss": 0.0109,
+ "step": 31543
+ },
+ {
+ "epoch": 85.95095367847412,
+ "grad_norm": 1.4044580459594727,
+ "learning_rate": 1.0174522728990821e-06,
+ "loss": 0.0141,
+ "step": 31544
+ },
+ {
+ "epoch": 85.95367847411444,
+ "grad_norm": 1.5247597694396973,
+ "learning_rate": 1.0170644734594304e-06,
+ "loss": 0.0188,
+ "step": 31545
+ },
+ {
+ "epoch": 85.95640326975477,
+ "grad_norm": 0.9020019173622131,
+ "learning_rate": 1.0166767439785917e-06,
+ "loss": 0.0104,
+ "step": 31546
+ },
+ {
+ "epoch": 85.95912806539509,
+ "grad_norm": 0.7545179724693298,
+ "learning_rate": 1.016289084459584e-06,
+ "loss": 0.0078,
+ "step": 31547
+ },
+ {
+ "epoch": 85.96185286103542,
+ "grad_norm": 0.706710159778595,
+ "learning_rate": 1.015901494905429e-06,
+ "loss": 0.0068,
+ "step": 31548
+ },
+ {
+ "epoch": 85.96457765667574,
+ "grad_norm": 1.2231459617614746,
+ "learning_rate": 1.0155139753191424e-06,
+ "loss": 0.0207,
+ "step": 31549
+ },
+ {
+ "epoch": 85.96730245231608,
+ "grad_norm": 1.113112449645996,
+ "learning_rate": 1.0151265257037457e-06,
+ "loss": 0.012,
+ "step": 31550
+ },
+ {
+ "epoch": 85.97002724795641,
+ "grad_norm": 1.2088648080825806,
+ "learning_rate": 1.0147391460622547e-06,
+ "loss": 0.048,
+ "step": 31551
+ },
+ {
+ "epoch": 85.97275204359673,
+ "grad_norm": 1.3080893754959106,
+ "learning_rate": 1.0143518363976857e-06,
+ "loss": 0.0085,
+ "step": 31552
+ },
+ {
+ "epoch": 85.97547683923706,
+ "grad_norm": 1.5290980339050293,
+ "learning_rate": 1.013964596713053e-06,
+ "loss": 0.0667,
+ "step": 31553
+ },
+ {
+ "epoch": 85.97820163487738,
+ "grad_norm": 0.7983057498931885,
+ "learning_rate": 1.0135774270113774e-06,
+ "loss": 0.0106,
+ "step": 31554
+ },
+ {
+ "epoch": 85.98092643051771,
+ "grad_norm": 0.9996146559715271,
+ "learning_rate": 1.0131903272956722e-06,
+ "loss": 0.0279,
+ "step": 31555
+ },
+ {
+ "epoch": 85.98365122615803,
+ "grad_norm": 2.1755058765411377,
+ "learning_rate": 1.0128032975689506e-06,
+ "loss": 0.0044,
+ "step": 31556
+ },
+ {
+ "epoch": 85.98637602179836,
+ "grad_norm": 0.6995307207107544,
+ "learning_rate": 1.0124163378342255e-06,
+ "loss": 0.0061,
+ "step": 31557
+ },
+ {
+ "epoch": 85.9891008174387,
+ "grad_norm": 1.796732783317566,
+ "learning_rate": 1.0120294480945148e-06,
+ "loss": 0.0361,
+ "step": 31558
+ },
+ {
+ "epoch": 85.99182561307902,
+ "grad_norm": 0.7854394912719727,
+ "learning_rate": 1.0116426283528301e-06,
+ "loss": 0.0058,
+ "step": 31559
+ },
+ {
+ "epoch": 85.99455040871935,
+ "grad_norm": 1.6787447929382324,
+ "learning_rate": 1.0112558786121829e-06,
+ "loss": 0.0524,
+ "step": 31560
+ },
+ {
+ "epoch": 85.99727520435967,
+ "grad_norm": 1.3914892673492432,
+ "learning_rate": 1.0108691988755837e-06,
+ "loss": 0.0817,
+ "step": 31561
+ },
+ {
+ "epoch": 86.0,
+ "grad_norm": 1.4071968793869019,
+ "learning_rate": 1.010482589146048e-06,
+ "loss": 0.0146,
+ "step": 31562
+ },
+ {
+ "epoch": 86.00272479564033,
+ "grad_norm": 1.5323563814163208,
+ "learning_rate": 1.0100960494265843e-06,
+ "loss": 0.059,
+ "step": 31563
+ },
+ {
+ "epoch": 86.00544959128065,
+ "grad_norm": 0.9737720489501953,
+ "learning_rate": 1.0097095797202016e-06,
+ "loss": 0.005,
+ "step": 31564
+ },
+ {
+ "epoch": 86.00817438692098,
+ "grad_norm": 1.0652945041656494,
+ "learning_rate": 1.0093231800299109e-06,
+ "loss": 0.0204,
+ "step": 31565
+ },
+ {
+ "epoch": 86.0108991825613,
+ "grad_norm": 0.5872132778167725,
+ "learning_rate": 1.0089368503587238e-06,
+ "loss": 0.0057,
+ "step": 31566
+ },
+ {
+ "epoch": 86.01362397820164,
+ "grad_norm": 0.7006679177284241,
+ "learning_rate": 1.008550590709647e-06,
+ "loss": 0.0089,
+ "step": 31567
+ },
+ {
+ "epoch": 86.01634877384195,
+ "grad_norm": 1.067752480506897,
+ "learning_rate": 1.0081644010856895e-06,
+ "loss": 0.0126,
+ "step": 31568
+ },
+ {
+ "epoch": 86.01907356948229,
+ "grad_norm": 0.7289900183677673,
+ "learning_rate": 1.0077782814898563e-06,
+ "loss": 0.0095,
+ "step": 31569
+ },
+ {
+ "epoch": 86.02179836512262,
+ "grad_norm": 1.6430131196975708,
+ "learning_rate": 1.0073922319251584e-06,
+ "loss": 0.0063,
+ "step": 31570
+ },
+ {
+ "epoch": 86.02452316076294,
+ "grad_norm": 0.930815577507019,
+ "learning_rate": 1.0070062523946001e-06,
+ "loss": 0.048,
+ "step": 31571
+ },
+ {
+ "epoch": 86.02724795640327,
+ "grad_norm": 1.009628415107727,
+ "learning_rate": 1.006620342901188e-06,
+ "loss": 0.0294,
+ "step": 31572
+ },
+ {
+ "epoch": 86.02997275204359,
+ "grad_norm": 0.9648259878158569,
+ "learning_rate": 1.0062345034479236e-06,
+ "loss": 0.0166,
+ "step": 31573
+ },
+ {
+ "epoch": 86.03269754768392,
+ "grad_norm": 1.2402757406234741,
+ "learning_rate": 1.0058487340378187e-06,
+ "loss": 0.038,
+ "step": 31574
+ },
+ {
+ "epoch": 86.03542234332426,
+ "grad_norm": 0.8314828276634216,
+ "learning_rate": 1.0054630346738747e-06,
+ "loss": 0.0079,
+ "step": 31575
+ },
+ {
+ "epoch": 86.03814713896458,
+ "grad_norm": 1.186190128326416,
+ "learning_rate": 1.005077405359095e-06,
+ "loss": 0.0131,
+ "step": 31576
+ },
+ {
+ "epoch": 86.04087193460491,
+ "grad_norm": 1.0899765491485596,
+ "learning_rate": 1.0046918460964806e-06,
+ "loss": 0.0118,
+ "step": 31577
+ },
+ {
+ "epoch": 86.04359673024523,
+ "grad_norm": 1.090238332748413,
+ "learning_rate": 1.0043063568890388e-06,
+ "loss": 0.0342,
+ "step": 31578
+ },
+ {
+ "epoch": 86.04632152588556,
+ "grad_norm": 1.0378495454788208,
+ "learning_rate": 1.0039209377397696e-06,
+ "loss": 0.0183,
+ "step": 31579
+ },
+ {
+ "epoch": 86.04904632152588,
+ "grad_norm": 0.5388764142990112,
+ "learning_rate": 1.003535588651674e-06,
+ "loss": 0.0054,
+ "step": 31580
+ },
+ {
+ "epoch": 86.05177111716621,
+ "grad_norm": 1.3910185098648071,
+ "learning_rate": 1.0031503096277529e-06,
+ "loss": 0.0116,
+ "step": 31581
+ },
+ {
+ "epoch": 86.05449591280654,
+ "grad_norm": 1.0995912551879883,
+ "learning_rate": 1.002765100671007e-06,
+ "loss": 0.0163,
+ "step": 31582
+ },
+ {
+ "epoch": 86.05722070844686,
+ "grad_norm": 1.7880595922470093,
+ "learning_rate": 1.0023799617844387e-06,
+ "loss": 0.0355,
+ "step": 31583
+ },
+ {
+ "epoch": 86.0599455040872,
+ "grad_norm": 1.6269091367721558,
+ "learning_rate": 1.0019948929710456e-06,
+ "loss": 0.0186,
+ "step": 31584
+ },
+ {
+ "epoch": 86.06267029972751,
+ "grad_norm": 1.5159621238708496,
+ "learning_rate": 1.0016098942338249e-06,
+ "loss": 0.0902,
+ "step": 31585
+ },
+ {
+ "epoch": 86.06539509536785,
+ "grad_norm": 1.224748134613037,
+ "learning_rate": 1.0012249655757788e-06,
+ "loss": 0.0123,
+ "step": 31586
+ },
+ {
+ "epoch": 86.06811989100818,
+ "grad_norm": 0.6267834901809692,
+ "learning_rate": 1.000840106999903e-06,
+ "loss": 0.0051,
+ "step": 31587
+ },
+ {
+ "epoch": 86.0708446866485,
+ "grad_norm": 0.6367396712303162,
+ "learning_rate": 1.0004553185091948e-06,
+ "loss": 0.0064,
+ "step": 31588
+ },
+ {
+ "epoch": 86.07356948228883,
+ "grad_norm": 1.7287262678146362,
+ "learning_rate": 1.0000706001066485e-06,
+ "loss": 0.1655,
+ "step": 31589
+ },
+ {
+ "epoch": 86.07629427792915,
+ "grad_norm": 0.8106772303581238,
+ "learning_rate": 9.996859517952652e-07,
+ "loss": 0.0057,
+ "step": 31590
+ },
+ {
+ "epoch": 86.07901907356948,
+ "grad_norm": 1.165492296218872,
+ "learning_rate": 9.993013735780376e-07,
+ "loss": 0.0408,
+ "step": 31591
+ },
+ {
+ "epoch": 86.0817438692098,
+ "grad_norm": 0.9448391795158386,
+ "learning_rate": 9.989168654579617e-07,
+ "loss": 0.0109,
+ "step": 31592
+ },
+ {
+ "epoch": 86.08446866485014,
+ "grad_norm": 0.9323025345802307,
+ "learning_rate": 9.985324274380292e-07,
+ "loss": 0.0123,
+ "step": 31593
+ },
+ {
+ "epoch": 86.08719346049047,
+ "grad_norm": 1.700677514076233,
+ "learning_rate": 9.981480595212389e-07,
+ "loss": 0.0613,
+ "step": 31594
+ },
+ {
+ "epoch": 86.08991825613079,
+ "grad_norm": 0.8293555378913879,
+ "learning_rate": 9.977637617105818e-07,
+ "loss": 0.0131,
+ "step": 31595
+ },
+ {
+ "epoch": 86.09264305177112,
+ "grad_norm": 1.50361168384552,
+ "learning_rate": 9.973795340090498e-07,
+ "loss": 0.1848,
+ "step": 31596
+ },
+ {
+ "epoch": 86.09536784741144,
+ "grad_norm": 0.4151199162006378,
+ "learning_rate": 9.96995376419635e-07,
+ "loss": 0.0042,
+ "step": 31597
+ },
+ {
+ "epoch": 86.09809264305177,
+ "grad_norm": 1.5517070293426514,
+ "learning_rate": 9.966112889453328e-07,
+ "loss": 0.0266,
+ "step": 31598
+ },
+ {
+ "epoch": 86.1008174386921,
+ "grad_norm": 1.1450873613357544,
+ "learning_rate": 9.962272715891296e-07,
+ "loss": 0.0187,
+ "step": 31599
+ },
+ {
+ "epoch": 86.10354223433242,
+ "grad_norm": 6.508196830749512,
+ "learning_rate": 9.958433243540211e-07,
+ "loss": 0.0523,
+ "step": 31600
+ },
+ {
+ "epoch": 86.10626702997276,
+ "grad_norm": 1.1688334941864014,
+ "learning_rate": 9.954594472429947e-07,
+ "loss": 0.0085,
+ "step": 31601
+ },
+ {
+ "epoch": 86.10899182561307,
+ "grad_norm": 1.1058293581008911,
+ "learning_rate": 9.950756402590379e-07,
+ "loss": 0.0184,
+ "step": 31602
+ },
+ {
+ "epoch": 86.11171662125341,
+ "grad_norm": 0.8338466286659241,
+ "learning_rate": 9.94691903405145e-07,
+ "loss": 0.0088,
+ "step": 31603
+ },
+ {
+ "epoch": 86.11444141689373,
+ "grad_norm": 0.7426790595054626,
+ "learning_rate": 9.943082366843005e-07,
+ "loss": 0.0743,
+ "step": 31604
+ },
+ {
+ "epoch": 86.11716621253406,
+ "grad_norm": 1.1147410869598389,
+ "learning_rate": 9.939246400994929e-07,
+ "loss": 0.0798,
+ "step": 31605
+ },
+ {
+ "epoch": 86.11989100817439,
+ "grad_norm": 1.1825190782546997,
+ "learning_rate": 9.935411136537121e-07,
+ "loss": 0.0194,
+ "step": 31606
+ },
+ {
+ "epoch": 86.12261580381471,
+ "grad_norm": 0.6131374835968018,
+ "learning_rate": 9.931576573499425e-07,
+ "loss": 0.0054,
+ "step": 31607
+ },
+ {
+ "epoch": 86.12534059945504,
+ "grad_norm": 1.312126874923706,
+ "learning_rate": 9.927742711911715e-07,
+ "loss": 0.0541,
+ "step": 31608
+ },
+ {
+ "epoch": 86.12806539509536,
+ "grad_norm": 0.8253557085990906,
+ "learning_rate": 9.923909551803824e-07,
+ "loss": 0.0157,
+ "step": 31609
+ },
+ {
+ "epoch": 86.1307901907357,
+ "grad_norm": 0.6359311938285828,
+ "learning_rate": 9.920077093205648e-07,
+ "loss": 0.0066,
+ "step": 31610
+ },
+ {
+ "epoch": 86.13351498637603,
+ "grad_norm": 1.252207636833191,
+ "learning_rate": 9.916245336147023e-07,
+ "loss": 0.0124,
+ "step": 31611
+ },
+ {
+ "epoch": 86.13623978201635,
+ "grad_norm": 0.8132942914962769,
+ "learning_rate": 9.912414280657766e-07,
+ "loss": 0.0122,
+ "step": 31612
+ },
+ {
+ "epoch": 86.13896457765668,
+ "grad_norm": 1.6108932495117188,
+ "learning_rate": 9.908583926767712e-07,
+ "loss": 0.0584,
+ "step": 31613
+ },
+ {
+ "epoch": 86.141689373297,
+ "grad_norm": 1.4678210020065308,
+ "learning_rate": 9.904754274506723e-07,
+ "loss": 0.0143,
+ "step": 31614
+ },
+ {
+ "epoch": 86.14441416893733,
+ "grad_norm": 1.2679916620254517,
+ "learning_rate": 9.90092532390462e-07,
+ "loss": 0.0871,
+ "step": 31615
+ },
+ {
+ "epoch": 86.14713896457765,
+ "grad_norm": 0.9603802561759949,
+ "learning_rate": 9.89709707499118e-07,
+ "loss": 0.0094,
+ "step": 31616
+ },
+ {
+ "epoch": 86.14986376021798,
+ "grad_norm": 1.3849260807037354,
+ "learning_rate": 9.893269527796268e-07,
+ "loss": 0.0167,
+ "step": 31617
+ },
+ {
+ "epoch": 86.15258855585832,
+ "grad_norm": 1.2371352910995483,
+ "learning_rate": 9.88944268234966e-07,
+ "loss": 0.0343,
+ "step": 31618
+ },
+ {
+ "epoch": 86.15531335149863,
+ "grad_norm": 0.7720845937728882,
+ "learning_rate": 9.885616538681186e-07,
+ "loss": 0.0172,
+ "step": 31619
+ },
+ {
+ "epoch": 86.15803814713897,
+ "grad_norm": 0.7756643891334534,
+ "learning_rate": 9.881791096820625e-07,
+ "loss": 0.0051,
+ "step": 31620
+ },
+ {
+ "epoch": 86.16076294277929,
+ "grad_norm": 0.8182379603385925,
+ "learning_rate": 9.877966356797785e-07,
+ "loss": 0.0071,
+ "step": 31621
+ },
+ {
+ "epoch": 86.16348773841962,
+ "grad_norm": 0.47679901123046875,
+ "learning_rate": 9.87414231864242e-07,
+ "loss": 0.0039,
+ "step": 31622
+ },
+ {
+ "epoch": 86.16621253405995,
+ "grad_norm": 0.9018247127532959,
+ "learning_rate": 9.870318982384352e-07,
+ "loss": 0.0096,
+ "step": 31623
+ },
+ {
+ "epoch": 86.16893732970027,
+ "grad_norm": 0.9483987092971802,
+ "learning_rate": 9.866496348053334e-07,
+ "loss": 0.0317,
+ "step": 31624
+ },
+ {
+ "epoch": 86.1716621253406,
+ "grad_norm": 0.5834227204322815,
+ "learning_rate": 9.862674415679152e-07,
+ "loss": 0.0059,
+ "step": 31625
+ },
+ {
+ "epoch": 86.17438692098092,
+ "grad_norm": 1.1430580615997314,
+ "learning_rate": 9.858853185291538e-07,
+ "loss": 0.0125,
+ "step": 31626
+ },
+ {
+ "epoch": 86.17711171662125,
+ "grad_norm": 0.5160636901855469,
+ "learning_rate": 9.855032656920283e-07,
+ "loss": 0.0054,
+ "step": 31627
+ },
+ {
+ "epoch": 86.17983651226157,
+ "grad_norm": 0.9391559958457947,
+ "learning_rate": 9.851212830595148e-07,
+ "loss": 0.0107,
+ "step": 31628
+ },
+ {
+ "epoch": 86.1825613079019,
+ "grad_norm": 0.31911495327949524,
+ "learning_rate": 9.847393706345843e-07,
+ "loss": 0.0033,
+ "step": 31629
+ },
+ {
+ "epoch": 86.18528610354224,
+ "grad_norm": 3.0158846378326416,
+ "learning_rate": 9.843575284202145e-07,
+ "loss": 0.015,
+ "step": 31630
+ },
+ {
+ "epoch": 86.18801089918256,
+ "grad_norm": 3.738466501235962,
+ "learning_rate": 9.839757564193787e-07,
+ "loss": 0.0404,
+ "step": 31631
+ },
+ {
+ "epoch": 86.19073569482289,
+ "grad_norm": 0.9008553624153137,
+ "learning_rate": 9.835940546350498e-07,
+ "loss": 0.0209,
+ "step": 31632
+ },
+ {
+ "epoch": 86.19346049046321,
+ "grad_norm": 2.1773102283477783,
+ "learning_rate": 9.832124230701979e-07,
+ "loss": 0.0144,
+ "step": 31633
+ },
+ {
+ "epoch": 86.19618528610354,
+ "grad_norm": 0.5926921367645264,
+ "learning_rate": 9.828308617277971e-07,
+ "loss": 0.0045,
+ "step": 31634
+ },
+ {
+ "epoch": 86.19891008174388,
+ "grad_norm": 1.0006972551345825,
+ "learning_rate": 9.824493706108218e-07,
+ "loss": 0.0192,
+ "step": 31635
+ },
+ {
+ "epoch": 86.2016348773842,
+ "grad_norm": 0.5832185745239258,
+ "learning_rate": 9.820679497222396e-07,
+ "loss": 0.0073,
+ "step": 31636
+ },
+ {
+ "epoch": 86.20435967302453,
+ "grad_norm": 1.9972734451293945,
+ "learning_rate": 9.816865990650238e-07,
+ "loss": 0.0264,
+ "step": 31637
+ },
+ {
+ "epoch": 86.20708446866485,
+ "grad_norm": 0.557236909866333,
+ "learning_rate": 9.813053186421395e-07,
+ "loss": 0.0049,
+ "step": 31638
+ },
+ {
+ "epoch": 86.20980926430518,
+ "grad_norm": 1.0789930820465088,
+ "learning_rate": 9.8092410845656e-07,
+ "loss": 0.0839,
+ "step": 31639
+ },
+ {
+ "epoch": 86.2125340599455,
+ "grad_norm": 1.500464677810669,
+ "learning_rate": 9.805429685112555e-07,
+ "loss": 0.1221,
+ "step": 31640
+ },
+ {
+ "epoch": 86.21525885558583,
+ "grad_norm": 1.073128342628479,
+ "learning_rate": 9.80161898809191e-07,
+ "loss": 0.0885,
+ "step": 31641
+ },
+ {
+ "epoch": 86.21798365122616,
+ "grad_norm": 1.1016218662261963,
+ "learning_rate": 9.79780899353333e-07,
+ "loss": 0.0082,
+ "step": 31642
+ },
+ {
+ "epoch": 86.22070844686648,
+ "grad_norm": 0.9596053957939148,
+ "learning_rate": 9.793999701466527e-07,
+ "loss": 0.1025,
+ "step": 31643
+ },
+ {
+ "epoch": 86.22343324250681,
+ "grad_norm": 1.0771135091781616,
+ "learning_rate": 9.790191111921166e-07,
+ "loss": 0.0656,
+ "step": 31644
+ },
+ {
+ "epoch": 86.22615803814713,
+ "grad_norm": 1.3088247776031494,
+ "learning_rate": 9.786383224926877e-07,
+ "loss": 0.0212,
+ "step": 31645
+ },
+ {
+ "epoch": 86.22888283378747,
+ "grad_norm": 1.3626776933670044,
+ "learning_rate": 9.782576040513325e-07,
+ "loss": 0.0118,
+ "step": 31646
+ },
+ {
+ "epoch": 86.2316076294278,
+ "grad_norm": 1.2022695541381836,
+ "learning_rate": 9.778769558710177e-07,
+ "loss": 0.0157,
+ "step": 31647
+ },
+ {
+ "epoch": 86.23433242506812,
+ "grad_norm": 1.3768020868301392,
+ "learning_rate": 9.774963779547076e-07,
+ "loss": 0.018,
+ "step": 31648
+ },
+ {
+ "epoch": 86.23705722070845,
+ "grad_norm": 1.1252870559692383,
+ "learning_rate": 9.771158703053652e-07,
+ "loss": 0.0107,
+ "step": 31649
+ },
+ {
+ "epoch": 86.23978201634877,
+ "grad_norm": 1.3053357601165771,
+ "learning_rate": 9.767354329259515e-07,
+ "loss": 0.0201,
+ "step": 31650
+ },
+ {
+ "epoch": 86.2425068119891,
+ "grad_norm": 1.0553585290908813,
+ "learning_rate": 9.76355065819432e-07,
+ "loss": 0.013,
+ "step": 31651
+ },
+ {
+ "epoch": 86.24523160762942,
+ "grad_norm": 1.56803297996521,
+ "learning_rate": 9.75974768988771e-07,
+ "loss": 0.0484,
+ "step": 31652
+ },
+ {
+ "epoch": 86.24795640326975,
+ "grad_norm": 1.2955487966537476,
+ "learning_rate": 9.755945424369283e-07,
+ "loss": 0.0148,
+ "step": 31653
+ },
+ {
+ "epoch": 86.25068119891009,
+ "grad_norm": 1.214418888092041,
+ "learning_rate": 9.752143861668628e-07,
+ "loss": 0.01,
+ "step": 31654
+ },
+ {
+ "epoch": 86.2534059945504,
+ "grad_norm": 0.7450932860374451,
+ "learning_rate": 9.748343001815385e-07,
+ "loss": 0.0068,
+ "step": 31655
+ },
+ {
+ "epoch": 86.25613079019074,
+ "grad_norm": 0.9981510043144226,
+ "learning_rate": 9.744542844839145e-07,
+ "loss": 0.0313,
+ "step": 31656
+ },
+ {
+ "epoch": 86.25885558583106,
+ "grad_norm": 1.2041618824005127,
+ "learning_rate": 9.740743390769503e-07,
+ "loss": 0.0106,
+ "step": 31657
+ },
+ {
+ "epoch": 86.26158038147139,
+ "grad_norm": 1.8787336349487305,
+ "learning_rate": 9.736944639636025e-07,
+ "loss": 0.0197,
+ "step": 31658
+ },
+ {
+ "epoch": 86.26430517711172,
+ "grad_norm": 1.3535383939743042,
+ "learning_rate": 9.733146591468334e-07,
+ "loss": 0.0169,
+ "step": 31659
+ },
+ {
+ "epoch": 86.26702997275204,
+ "grad_norm": 0.841386616230011,
+ "learning_rate": 9.729349246296005e-07,
+ "loss": 0.0056,
+ "step": 31660
+ },
+ {
+ "epoch": 86.26975476839237,
+ "grad_norm": 1.5876892805099487,
+ "learning_rate": 9.72555260414858e-07,
+ "loss": 0.1402,
+ "step": 31661
+ },
+ {
+ "epoch": 86.2724795640327,
+ "grad_norm": 1.2483134269714355,
+ "learning_rate": 9.721756665055637e-07,
+ "loss": 0.0587,
+ "step": 31662
+ },
+ {
+ "epoch": 86.27520435967303,
+ "grad_norm": 1.0733387470245361,
+ "learning_rate": 9.717961429046773e-07,
+ "loss": 0.017,
+ "step": 31663
+ },
+ {
+ "epoch": 86.27792915531334,
+ "grad_norm": 1.052677869796753,
+ "learning_rate": 9.714166896151511e-07,
+ "loss": 0.0103,
+ "step": 31664
+ },
+ {
+ "epoch": 86.28065395095368,
+ "grad_norm": 0.8182767033576965,
+ "learning_rate": 9.710373066399413e-07,
+ "loss": 0.0095,
+ "step": 31665
+ },
+ {
+ "epoch": 86.28337874659401,
+ "grad_norm": 1.6031588315963745,
+ "learning_rate": 9.706579939820005e-07,
+ "loss": 0.0543,
+ "step": 31666
+ },
+ {
+ "epoch": 86.28610354223433,
+ "grad_norm": 1.0976805686950684,
+ "learning_rate": 9.702787516442858e-07,
+ "loss": 0.0135,
+ "step": 31667
+ },
+ {
+ "epoch": 86.28882833787466,
+ "grad_norm": 1.3117749691009521,
+ "learning_rate": 9.698995796297483e-07,
+ "loss": 0.0778,
+ "step": 31668
+ },
+ {
+ "epoch": 86.29155313351498,
+ "grad_norm": 1.4770926237106323,
+ "learning_rate": 9.695204779413448e-07,
+ "loss": 0.0977,
+ "step": 31669
+ },
+ {
+ "epoch": 86.29427792915531,
+ "grad_norm": 0.9313795566558838,
+ "learning_rate": 9.691414465820226e-07,
+ "loss": 0.0078,
+ "step": 31670
+ },
+ {
+ "epoch": 86.29700272479565,
+ "grad_norm": 1.802208662033081,
+ "learning_rate": 9.687624855547372e-07,
+ "loss": 0.0142,
+ "step": 31671
+ },
+ {
+ "epoch": 86.29972752043597,
+ "grad_norm": 1.0581640005111694,
+ "learning_rate": 9.683835948624397e-07,
+ "loss": 0.0168,
+ "step": 31672
+ },
+ {
+ "epoch": 86.3024523160763,
+ "grad_norm": 1.5312882661819458,
+ "learning_rate": 9.680047745080811e-07,
+ "loss": 0.1202,
+ "step": 31673
+ },
+ {
+ "epoch": 86.30517711171662,
+ "grad_norm": 1.747002124786377,
+ "learning_rate": 9.676260244946078e-07,
+ "loss": 0.0251,
+ "step": 31674
+ },
+ {
+ "epoch": 86.30790190735695,
+ "grad_norm": 0.9866296052932739,
+ "learning_rate": 9.672473448249741e-07,
+ "loss": 0.0897,
+ "step": 31675
+ },
+ {
+ "epoch": 86.31062670299727,
+ "grad_norm": 2.647376537322998,
+ "learning_rate": 9.668687355021266e-07,
+ "loss": 0.0144,
+ "step": 31676
+ },
+ {
+ "epoch": 86.3133514986376,
+ "grad_norm": 0.7629173994064331,
+ "learning_rate": 9.664901965290163e-07,
+ "loss": 0.01,
+ "step": 31677
+ },
+ {
+ "epoch": 86.31607629427793,
+ "grad_norm": 0.7708972096443176,
+ "learning_rate": 9.661117279085862e-07,
+ "loss": 0.0129,
+ "step": 31678
+ },
+ {
+ "epoch": 86.31880108991825,
+ "grad_norm": 1.3274891376495361,
+ "learning_rate": 9.6573332964379e-07,
+ "loss": 0.0215,
+ "step": 31679
+ },
+ {
+ "epoch": 86.32152588555859,
+ "grad_norm": 1.4905266761779785,
+ "learning_rate": 9.653550017375724e-07,
+ "loss": 0.0151,
+ "step": 31680
+ },
+ {
+ "epoch": 86.3242506811989,
+ "grad_norm": 1.5888495445251465,
+ "learning_rate": 9.64976744192878e-07,
+ "loss": 0.0419,
+ "step": 31681
+ },
+ {
+ "epoch": 86.32697547683924,
+ "grad_norm": 0.8860918283462524,
+ "learning_rate": 9.645985570126526e-07,
+ "loss": 0.0123,
+ "step": 31682
+ },
+ {
+ "epoch": 86.32970027247957,
+ "grad_norm": 0.7956109046936035,
+ "learning_rate": 9.642204401998446e-07,
+ "loss": 0.009,
+ "step": 31683
+ },
+ {
+ "epoch": 86.33242506811989,
+ "grad_norm": 1.3225890398025513,
+ "learning_rate": 9.63842393757397e-07,
+ "loss": 0.0101,
+ "step": 31684
+ },
+ {
+ "epoch": 86.33514986376022,
+ "grad_norm": 1.550040602684021,
+ "learning_rate": 9.634644176882523e-07,
+ "loss": 0.1033,
+ "step": 31685
+ },
+ {
+ "epoch": 86.33787465940054,
+ "grad_norm": 0.8937479853630066,
+ "learning_rate": 9.63086511995358e-07,
+ "loss": 0.0147,
+ "step": 31686
+ },
+ {
+ "epoch": 86.34059945504087,
+ "grad_norm": 1.5008089542388916,
+ "learning_rate": 9.62708676681653e-07,
+ "loss": 0.0103,
+ "step": 31687
+ },
+ {
+ "epoch": 86.34332425068119,
+ "grad_norm": 0.6812260746955872,
+ "learning_rate": 9.623309117500833e-07,
+ "loss": 0.0085,
+ "step": 31688
+ },
+ {
+ "epoch": 86.34604904632153,
+ "grad_norm": 1.7042449712753296,
+ "learning_rate": 9.619532172035906e-07,
+ "loss": 0.1276,
+ "step": 31689
+ },
+ {
+ "epoch": 86.34877384196186,
+ "grad_norm": 1.4824339151382446,
+ "learning_rate": 9.615755930451144e-07,
+ "loss": 0.0173,
+ "step": 31690
+ },
+ {
+ "epoch": 86.35149863760218,
+ "grad_norm": 1.1914188861846924,
+ "learning_rate": 9.611980392775956e-07,
+ "loss": 0.0094,
+ "step": 31691
+ },
+ {
+ "epoch": 86.35422343324251,
+ "grad_norm": 1.727779746055603,
+ "learning_rate": 9.608205559039763e-07,
+ "loss": 0.008,
+ "step": 31692
+ },
+ {
+ "epoch": 86.35694822888283,
+ "grad_norm": 1.4455915689468384,
+ "learning_rate": 9.604431429271965e-07,
+ "loss": 0.0201,
+ "step": 31693
+ },
+ {
+ "epoch": 86.35967302452316,
+ "grad_norm": 0.7335498929023743,
+ "learning_rate": 9.600658003501928e-07,
+ "loss": 0.0077,
+ "step": 31694
+ },
+ {
+ "epoch": 86.3623978201635,
+ "grad_norm": 1.420154333114624,
+ "learning_rate": 9.596885281759071e-07,
+ "loss": 0.0264,
+ "step": 31695
+ },
+ {
+ "epoch": 86.36512261580381,
+ "grad_norm": 0.8386861681938171,
+ "learning_rate": 9.593113264072762e-07,
+ "loss": 0.0085,
+ "step": 31696
+ },
+ {
+ "epoch": 86.36784741144415,
+ "grad_norm": 6.142862796783447,
+ "learning_rate": 9.589341950472374e-07,
+ "loss": 0.0106,
+ "step": 31697
+ },
+ {
+ "epoch": 86.37057220708446,
+ "grad_norm": 1.301869511604309,
+ "learning_rate": 9.585571340987265e-07,
+ "loss": 0.0408,
+ "step": 31698
+ },
+ {
+ "epoch": 86.3732970027248,
+ "grad_norm": 1.181685209274292,
+ "learning_rate": 9.581801435646843e-07,
+ "loss": 0.0225,
+ "step": 31699
+ },
+ {
+ "epoch": 86.37602179836512,
+ "grad_norm": 0.8655313849449158,
+ "learning_rate": 9.578032234480428e-07,
+ "loss": 0.0076,
+ "step": 31700
+ },
+ {
+ "epoch": 86.37874659400545,
+ "grad_norm": 1.5062205791473389,
+ "learning_rate": 9.574263737517398e-07,
+ "loss": 0.0458,
+ "step": 31701
+ },
+ {
+ "epoch": 86.38147138964578,
+ "grad_norm": 1.5973970890045166,
+ "learning_rate": 9.570495944787062e-07,
+ "loss": 0.0481,
+ "step": 31702
+ },
+ {
+ "epoch": 86.3841961852861,
+ "grad_norm": 1.0277140140533447,
+ "learning_rate": 9.566728856318807e-07,
+ "loss": 0.0112,
+ "step": 31703
+ },
+ {
+ "epoch": 86.38692098092643,
+ "grad_norm": 1.0472464561462402,
+ "learning_rate": 9.562962472141967e-07,
+ "loss": 0.009,
+ "step": 31704
+ },
+ {
+ "epoch": 86.38964577656675,
+ "grad_norm": 0.8840304017066956,
+ "learning_rate": 9.559196792285864e-07,
+ "loss": 0.0114,
+ "step": 31705
+ },
+ {
+ "epoch": 86.39237057220708,
+ "grad_norm": 0.9617149233818054,
+ "learning_rate": 9.555431816779836e-07,
+ "loss": 0.0099,
+ "step": 31706
+ },
+ {
+ "epoch": 86.39509536784742,
+ "grad_norm": 1.6969274282455444,
+ "learning_rate": 9.551667545653164e-07,
+ "loss": 0.0162,
+ "step": 31707
+ },
+ {
+ "epoch": 86.39782016348774,
+ "grad_norm": 1.450893759727478,
+ "learning_rate": 9.547903978935214e-07,
+ "loss": 0.0152,
+ "step": 31708
+ },
+ {
+ "epoch": 86.40054495912807,
+ "grad_norm": 0.8438453078269958,
+ "learning_rate": 9.544141116655282e-07,
+ "loss": 0.01,
+ "step": 31709
+ },
+ {
+ "epoch": 86.40326975476839,
+ "grad_norm": 1.1785690784454346,
+ "learning_rate": 9.540378958842666e-07,
+ "loss": 0.0187,
+ "step": 31710
+ },
+ {
+ "epoch": 86.40599455040872,
+ "grad_norm": 1.8430098295211792,
+ "learning_rate": 9.536617505526647e-07,
+ "loss": 0.0093,
+ "step": 31711
+ },
+ {
+ "epoch": 86.40871934604904,
+ "grad_norm": 1.4047727584838867,
+ "learning_rate": 9.532856756736552e-07,
+ "loss": 0.0201,
+ "step": 31712
+ },
+ {
+ "epoch": 86.41144414168937,
+ "grad_norm": 0.8366448879241943,
+ "learning_rate": 9.529096712501662e-07,
+ "loss": 0.0116,
+ "step": 31713
+ },
+ {
+ "epoch": 86.4141689373297,
+ "grad_norm": 1.5940827131271362,
+ "learning_rate": 9.525337372851262e-07,
+ "loss": 0.0142,
+ "step": 31714
+ },
+ {
+ "epoch": 86.41689373297002,
+ "grad_norm": 0.9567275047302246,
+ "learning_rate": 9.521578737814596e-07,
+ "loss": 0.0106,
+ "step": 31715
+ },
+ {
+ "epoch": 86.41961852861036,
+ "grad_norm": 1.3814548254013062,
+ "learning_rate": 9.517820807420985e-07,
+ "loss": 0.0789,
+ "step": 31716
+ },
+ {
+ "epoch": 86.42234332425068,
+ "grad_norm": 1.346566915512085,
+ "learning_rate": 9.514063581699673e-07,
+ "loss": 0.0161,
+ "step": 31717
+ },
+ {
+ "epoch": 86.42506811989101,
+ "grad_norm": 1.7836711406707764,
+ "learning_rate": 9.510307060679902e-07,
+ "loss": 0.0251,
+ "step": 31718
+ },
+ {
+ "epoch": 86.42779291553134,
+ "grad_norm": 0.6884857416152954,
+ "learning_rate": 9.506551244390949e-07,
+ "loss": 0.0047,
+ "step": 31719
+ },
+ {
+ "epoch": 86.43051771117166,
+ "grad_norm": 1.5858205556869507,
+ "learning_rate": 9.50279613286208e-07,
+ "loss": 0.0138,
+ "step": 31720
+ },
+ {
+ "epoch": 86.433242506812,
+ "grad_norm": 1.1737993955612183,
+ "learning_rate": 9.499041726122515e-07,
+ "loss": 0.0204,
+ "step": 31721
+ },
+ {
+ "epoch": 86.43596730245231,
+ "grad_norm": 1.0412678718566895,
+ "learning_rate": 9.49528802420151e-07,
+ "loss": 0.0119,
+ "step": 31722
+ },
+ {
+ "epoch": 86.43869209809264,
+ "grad_norm": 0.9816436171531677,
+ "learning_rate": 9.491535027128262e-07,
+ "loss": 0.0133,
+ "step": 31723
+ },
+ {
+ "epoch": 86.44141689373296,
+ "grad_norm": 1.4075291156768799,
+ "learning_rate": 9.487782734932049e-07,
+ "loss": 0.0205,
+ "step": 31724
+ },
+ {
+ "epoch": 86.4441416893733,
+ "grad_norm": 0.8826952576637268,
+ "learning_rate": 9.48403114764207e-07,
+ "loss": 0.0537,
+ "step": 31725
+ },
+ {
+ "epoch": 86.44686648501363,
+ "grad_norm": 1.3252325057983398,
+ "learning_rate": 9.480280265287545e-07,
+ "loss": 0.0255,
+ "step": 31726
+ },
+ {
+ "epoch": 86.44959128065395,
+ "grad_norm": 0.8038133978843689,
+ "learning_rate": 9.476530087897673e-07,
+ "loss": 0.0071,
+ "step": 31727
+ },
+ {
+ "epoch": 86.45231607629428,
+ "grad_norm": 1.0776559114456177,
+ "learning_rate": 9.472780615501675e-07,
+ "loss": 0.0088,
+ "step": 31728
+ },
+ {
+ "epoch": 86.4550408719346,
+ "grad_norm": 2.35263991355896,
+ "learning_rate": 9.469031848128762e-07,
+ "loss": 0.0384,
+ "step": 31729
+ },
+ {
+ "epoch": 86.45776566757493,
+ "grad_norm": 1.7373228073120117,
+ "learning_rate": 9.4652837858081e-07,
+ "loss": 0.0456,
+ "step": 31730
+ },
+ {
+ "epoch": 86.46049046321527,
+ "grad_norm": 0.7442569732666016,
+ "learning_rate": 9.461536428568885e-07,
+ "loss": 0.007,
+ "step": 31731
+ },
+ {
+ "epoch": 86.46321525885558,
+ "grad_norm": 1.0651520490646362,
+ "learning_rate": 9.457789776440329e-07,
+ "loss": 0.0128,
+ "step": 31732
+ },
+ {
+ "epoch": 86.46594005449592,
+ "grad_norm": 0.9867625832557678,
+ "learning_rate": 9.454043829451586e-07,
+ "loss": 0.1466,
+ "step": 31733
+ },
+ {
+ "epoch": 86.46866485013624,
+ "grad_norm": 1.005980372428894,
+ "learning_rate": 9.450298587631846e-07,
+ "loss": 0.0215,
+ "step": 31734
+ },
+ {
+ "epoch": 86.47138964577657,
+ "grad_norm": 0.881462037563324,
+ "learning_rate": 9.446554051010238e-07,
+ "loss": 0.0084,
+ "step": 31735
+ },
+ {
+ "epoch": 86.47411444141689,
+ "grad_norm": 1.3231884241104126,
+ "learning_rate": 9.442810219615984e-07,
+ "loss": 0.0137,
+ "step": 31736
+ },
+ {
+ "epoch": 86.47683923705722,
+ "grad_norm": 1.0050363540649414,
+ "learning_rate": 9.439067093478171e-07,
+ "loss": 0.0078,
+ "step": 31737
+ },
+ {
+ "epoch": 86.47956403269755,
+ "grad_norm": 1.6052521467208862,
+ "learning_rate": 9.435324672626023e-07,
+ "loss": 0.0383,
+ "step": 31738
+ },
+ {
+ "epoch": 86.48228882833787,
+ "grad_norm": 1.312638282775879,
+ "learning_rate": 9.431582957088625e-07,
+ "loss": 0.0146,
+ "step": 31739
+ },
+ {
+ "epoch": 86.4850136239782,
+ "grad_norm": 1.8167940378189087,
+ "learning_rate": 9.427841946895166e-07,
+ "loss": 0.0101,
+ "step": 31740
+ },
+ {
+ "epoch": 86.48773841961852,
+ "grad_norm": 0.7791603207588196,
+ "learning_rate": 9.424101642074756e-07,
+ "loss": 0.0083,
+ "step": 31741
+ },
+ {
+ "epoch": 86.49046321525886,
+ "grad_norm": 1.3846913576126099,
+ "learning_rate": 9.420362042656539e-07,
+ "loss": 0.0204,
+ "step": 31742
+ },
+ {
+ "epoch": 86.49318801089919,
+ "grad_norm": 0.8258998990058899,
+ "learning_rate": 9.41662314866959e-07,
+ "loss": 0.0091,
+ "step": 31743
+ },
+ {
+ "epoch": 86.49591280653951,
+ "grad_norm": 0.7241672873497009,
+ "learning_rate": 9.412884960143099e-07,
+ "loss": 0.0061,
+ "step": 31744
+ },
+ {
+ "epoch": 86.49863760217984,
+ "grad_norm": 1.4985216856002808,
+ "learning_rate": 9.409147477106139e-07,
+ "loss": 0.0317,
+ "step": 31745
+ },
+ {
+ "epoch": 86.50136239782016,
+ "grad_norm": 1.3719069957733154,
+ "learning_rate": 9.405410699587825e-07,
+ "loss": 0.0358,
+ "step": 31746
+ },
+ {
+ "epoch": 86.50408719346049,
+ "grad_norm": 5.605031490325928,
+ "learning_rate": 9.40167462761723e-07,
+ "loss": 0.0214,
+ "step": 31747
+ },
+ {
+ "epoch": 86.50681198910081,
+ "grad_norm": 0.5910335779190063,
+ "learning_rate": 9.397939261223499e-07,
+ "loss": 0.0057,
+ "step": 31748
+ },
+ {
+ "epoch": 86.50953678474114,
+ "grad_norm": 0.5433288812637329,
+ "learning_rate": 9.394204600435708e-07,
+ "loss": 0.0056,
+ "step": 31749
+ },
+ {
+ "epoch": 86.51226158038148,
+ "grad_norm": 1.183218002319336,
+ "learning_rate": 9.390470645282923e-07,
+ "loss": 0.0408,
+ "step": 31750
+ },
+ {
+ "epoch": 86.5149863760218,
+ "grad_norm": 0.9915559887886047,
+ "learning_rate": 9.386737395794232e-07,
+ "loss": 0.0171,
+ "step": 31751
+ },
+ {
+ "epoch": 86.51771117166213,
+ "grad_norm": 1.4848408699035645,
+ "learning_rate": 9.383004851998723e-07,
+ "loss": 0.0218,
+ "step": 31752
+ },
+ {
+ "epoch": 86.52043596730245,
+ "grad_norm": 1.5620253086090088,
+ "learning_rate": 9.379273013925449e-07,
+ "loss": 0.0359,
+ "step": 31753
+ },
+ {
+ "epoch": 86.52316076294278,
+ "grad_norm": 1.315170168876648,
+ "learning_rate": 9.375541881603479e-07,
+ "loss": 0.0092,
+ "step": 31754
+ },
+ {
+ "epoch": 86.52588555858311,
+ "grad_norm": 1.176330327987671,
+ "learning_rate": 9.371811455061874e-07,
+ "loss": 0.0939,
+ "step": 31755
+ },
+ {
+ "epoch": 86.52861035422343,
+ "grad_norm": 1.037104845046997,
+ "learning_rate": 9.368081734329682e-07,
+ "loss": 0.0199,
+ "step": 31756
+ },
+ {
+ "epoch": 86.53133514986376,
+ "grad_norm": 1.5526808500289917,
+ "learning_rate": 9.364352719435965e-07,
+ "loss": 0.0214,
+ "step": 31757
+ },
+ {
+ "epoch": 86.53405994550408,
+ "grad_norm": 0.9965370297431946,
+ "learning_rate": 9.360624410409746e-07,
+ "loss": 0.0148,
+ "step": 31758
+ },
+ {
+ "epoch": 86.53678474114442,
+ "grad_norm": 0.9649010300636292,
+ "learning_rate": 9.356896807280047e-07,
+ "loss": 0.019,
+ "step": 31759
+ },
+ {
+ "epoch": 86.53950953678473,
+ "grad_norm": 1.6520518064498901,
+ "learning_rate": 9.353169910075944e-07,
+ "loss": 0.1042,
+ "step": 31760
+ },
+ {
+ "epoch": 86.54223433242507,
+ "grad_norm": 1.2040461301803589,
+ "learning_rate": 9.349443718826434e-07,
+ "loss": 0.1437,
+ "step": 31761
+ },
+ {
+ "epoch": 86.5449591280654,
+ "grad_norm": 0.6428322792053223,
+ "learning_rate": 9.345718233560541e-07,
+ "loss": 0.0069,
+ "step": 31762
+ },
+ {
+ "epoch": 86.54768392370572,
+ "grad_norm": 0.567793071269989,
+ "learning_rate": 9.341993454307252e-07,
+ "loss": 0.0069,
+ "step": 31763
+ },
+ {
+ "epoch": 86.55040871934605,
+ "grad_norm": 1.0385023355484009,
+ "learning_rate": 9.338269381095632e-07,
+ "loss": 0.019,
+ "step": 31764
+ },
+ {
+ "epoch": 86.55313351498637,
+ "grad_norm": 1.0378992557525635,
+ "learning_rate": 9.334546013954637e-07,
+ "loss": 0.0095,
+ "step": 31765
+ },
+ {
+ "epoch": 86.5558583106267,
+ "grad_norm": 1.2682908773422241,
+ "learning_rate": 9.330823352913287e-07,
+ "loss": 0.0893,
+ "step": 31766
+ },
+ {
+ "epoch": 86.55858310626704,
+ "grad_norm": 0.5341607928276062,
+ "learning_rate": 9.327101398000548e-07,
+ "loss": 0.004,
+ "step": 31767
+ },
+ {
+ "epoch": 86.56130790190736,
+ "grad_norm": 1.0017958879470825,
+ "learning_rate": 9.323380149245442e-07,
+ "loss": 0.0088,
+ "step": 31768
+ },
+ {
+ "epoch": 86.56403269754769,
+ "grad_norm": 0.9108811020851135,
+ "learning_rate": 9.319659606676934e-07,
+ "loss": 0.0123,
+ "step": 31769
+ },
+ {
+ "epoch": 86.566757493188,
+ "grad_norm": 0.6273127198219299,
+ "learning_rate": 9.315939770324e-07,
+ "loss": 0.0075,
+ "step": 31770
+ },
+ {
+ "epoch": 86.56948228882834,
+ "grad_norm": 1.325376033782959,
+ "learning_rate": 9.312220640215585e-07,
+ "loss": 0.0137,
+ "step": 31771
+ },
+ {
+ "epoch": 86.57220708446866,
+ "grad_norm": 1.0197821855545044,
+ "learning_rate": 9.308502216380677e-07,
+ "loss": 0.0134,
+ "step": 31772
+ },
+ {
+ "epoch": 86.57493188010899,
+ "grad_norm": 1.2415226697921753,
+ "learning_rate": 9.304784498848263e-07,
+ "loss": 0.0191,
+ "step": 31773
+ },
+ {
+ "epoch": 86.57765667574932,
+ "grad_norm": 1.5476901531219482,
+ "learning_rate": 9.301067487647253e-07,
+ "loss": 0.011,
+ "step": 31774
+ },
+ {
+ "epoch": 86.58038147138964,
+ "grad_norm": 1.4753687381744385,
+ "learning_rate": 9.297351182806624e-07,
+ "loss": 0.0162,
+ "step": 31775
+ },
+ {
+ "epoch": 86.58310626702998,
+ "grad_norm": 1.3553292751312256,
+ "learning_rate": 9.293635584355276e-07,
+ "loss": 0.0104,
+ "step": 31776
+ },
+ {
+ "epoch": 86.5858310626703,
+ "grad_norm": 1.632495403289795,
+ "learning_rate": 9.289920692322196e-07,
+ "loss": 0.0652,
+ "step": 31777
+ },
+ {
+ "epoch": 86.58855585831063,
+ "grad_norm": 1.2517584562301636,
+ "learning_rate": 9.286206506736306e-07,
+ "loss": 0.0493,
+ "step": 31778
+ },
+ {
+ "epoch": 86.59128065395096,
+ "grad_norm": 1.0041871070861816,
+ "learning_rate": 9.282493027626482e-07,
+ "loss": 0.0093,
+ "step": 31779
+ },
+ {
+ "epoch": 86.59400544959128,
+ "grad_norm": 0.6610626578330994,
+ "learning_rate": 9.278780255021713e-07,
+ "loss": 0.0091,
+ "step": 31780
+ },
+ {
+ "epoch": 86.59673024523161,
+ "grad_norm": 1.337950587272644,
+ "learning_rate": 9.275068188950887e-07,
+ "loss": 0.0174,
+ "step": 31781
+ },
+ {
+ "epoch": 86.59945504087193,
+ "grad_norm": 2.0397284030914307,
+ "learning_rate": 9.271356829442902e-07,
+ "loss": 0.1483,
+ "step": 31782
+ },
+ {
+ "epoch": 86.60217983651226,
+ "grad_norm": 0.6674907803535461,
+ "learning_rate": 9.267646176526657e-07,
+ "loss": 0.0061,
+ "step": 31783
+ },
+ {
+ "epoch": 86.60490463215258,
+ "grad_norm": 1.1412439346313477,
+ "learning_rate": 9.263936230231075e-07,
+ "loss": 0.007,
+ "step": 31784
+ },
+ {
+ "epoch": 86.60762942779292,
+ "grad_norm": 1.5798169374465942,
+ "learning_rate": 9.260226990585042e-07,
+ "loss": 0.0094,
+ "step": 31785
+ },
+ {
+ "epoch": 86.61035422343325,
+ "grad_norm": 1.4391030073165894,
+ "learning_rate": 9.256518457617436e-07,
+ "loss": 0.0102,
+ "step": 31786
+ },
+ {
+ "epoch": 86.61307901907357,
+ "grad_norm": 1.2102261781692505,
+ "learning_rate": 9.252810631357134e-07,
+ "loss": 0.008,
+ "step": 31787
+ },
+ {
+ "epoch": 86.6158038147139,
+ "grad_norm": 1.7400888204574585,
+ "learning_rate": 9.249103511833013e-07,
+ "loss": 0.0895,
+ "step": 31788
+ },
+ {
+ "epoch": 86.61852861035422,
+ "grad_norm": 0.5372892618179321,
+ "learning_rate": 9.245397099073971e-07,
+ "loss": 0.0079,
+ "step": 31789
+ },
+ {
+ "epoch": 86.62125340599455,
+ "grad_norm": 1.2694039344787598,
+ "learning_rate": 9.241691393108864e-07,
+ "loss": 0.0154,
+ "step": 31790
+ },
+ {
+ "epoch": 86.62397820163488,
+ "grad_norm": 0.7383987307548523,
+ "learning_rate": 9.237986393966536e-07,
+ "loss": 0.0074,
+ "step": 31791
+ },
+ {
+ "epoch": 86.6267029972752,
+ "grad_norm": 0.7332767844200134,
+ "learning_rate": 9.23428210167584e-07,
+ "loss": 0.0061,
+ "step": 31792
+ },
+ {
+ "epoch": 86.62942779291554,
+ "grad_norm": 0.9634053111076355,
+ "learning_rate": 9.230578516265642e-07,
+ "loss": 0.011,
+ "step": 31793
+ },
+ {
+ "epoch": 86.63215258855585,
+ "grad_norm": 1.328743577003479,
+ "learning_rate": 9.226875637764788e-07,
+ "loss": 0.0175,
+ "step": 31794
+ },
+ {
+ "epoch": 86.63487738419619,
+ "grad_norm": 1.5092451572418213,
+ "learning_rate": 9.223173466202096e-07,
+ "loss": 0.0568,
+ "step": 31795
+ },
+ {
+ "epoch": 86.6376021798365,
+ "grad_norm": 0.8378072381019592,
+ "learning_rate": 9.219472001606389e-07,
+ "loss": 0.0128,
+ "step": 31796
+ },
+ {
+ "epoch": 86.64032697547684,
+ "grad_norm": 1.131913185119629,
+ "learning_rate": 9.215771244006533e-07,
+ "loss": 0.0595,
+ "step": 31797
+ },
+ {
+ "epoch": 86.64305177111717,
+ "grad_norm": 1.0797710418701172,
+ "learning_rate": 9.212071193431327e-07,
+ "loss": 0.0696,
+ "step": 31798
+ },
+ {
+ "epoch": 86.64577656675749,
+ "grad_norm": 0.7710082530975342,
+ "learning_rate": 9.208371849909592e-07,
+ "loss": 0.0078,
+ "step": 31799
+ },
+ {
+ "epoch": 86.64850136239782,
+ "grad_norm": 1.2474015951156616,
+ "learning_rate": 9.204673213470105e-07,
+ "loss": 0.0099,
+ "step": 31800
+ },
+ {
+ "epoch": 86.65122615803814,
+ "grad_norm": 0.9566617012023926,
+ "learning_rate": 9.200975284141733e-07,
+ "loss": 0.0075,
+ "step": 31801
+ },
+ {
+ "epoch": 86.65395095367847,
+ "grad_norm": 1.1158151626586914,
+ "learning_rate": 9.197278061953241e-07,
+ "loss": 0.0407,
+ "step": 31802
+ },
+ {
+ "epoch": 86.65667574931881,
+ "grad_norm": 1.2230209112167358,
+ "learning_rate": 9.193581546933406e-07,
+ "loss": 0.0195,
+ "step": 31803
+ },
+ {
+ "epoch": 86.65940054495913,
+ "grad_norm": 0.979611873626709,
+ "learning_rate": 9.189885739111048e-07,
+ "loss": 0.0124,
+ "step": 31804
+ },
+ {
+ "epoch": 86.66212534059946,
+ "grad_norm": 1.3539385795593262,
+ "learning_rate": 9.186190638514925e-07,
+ "loss": 0.0756,
+ "step": 31805
+ },
+ {
+ "epoch": 86.66485013623978,
+ "grad_norm": 0.6500447392463684,
+ "learning_rate": 9.182496245173844e-07,
+ "loss": 0.0075,
+ "step": 31806
+ },
+ {
+ "epoch": 86.66757493188011,
+ "grad_norm": 1.4245851039886475,
+ "learning_rate": 9.178802559116551e-07,
+ "loss": 0.0617,
+ "step": 31807
+ },
+ {
+ "epoch": 86.67029972752043,
+ "grad_norm": 0.6658335328102112,
+ "learning_rate": 9.175109580371811e-07,
+ "loss": 0.006,
+ "step": 31808
+ },
+ {
+ "epoch": 86.67302452316076,
+ "grad_norm": 1.5009146928787231,
+ "learning_rate": 9.171417308968411e-07,
+ "loss": 0.0179,
+ "step": 31809
+ },
+ {
+ "epoch": 86.6757493188011,
+ "grad_norm": 0.7043866515159607,
+ "learning_rate": 9.167725744935097e-07,
+ "loss": 0.0064,
+ "step": 31810
+ },
+ {
+ "epoch": 86.67847411444141,
+ "grad_norm": 1.3424817323684692,
+ "learning_rate": 9.1640348883006e-07,
+ "loss": 0.1083,
+ "step": 31811
+ },
+ {
+ "epoch": 86.68119891008175,
+ "grad_norm": 1.3294485807418823,
+ "learning_rate": 9.160344739093663e-07,
+ "loss": 0.0408,
+ "step": 31812
+ },
+ {
+ "epoch": 86.68392370572207,
+ "grad_norm": 0.6738141775131226,
+ "learning_rate": 9.156655297343053e-07,
+ "loss": 0.0066,
+ "step": 31813
+ },
+ {
+ "epoch": 86.6866485013624,
+ "grad_norm": 1.0465840101242065,
+ "learning_rate": 9.152966563077492e-07,
+ "loss": 0.0142,
+ "step": 31814
+ },
+ {
+ "epoch": 86.68937329700273,
+ "grad_norm": 1.1368612051010132,
+ "learning_rate": 9.149278536325701e-07,
+ "loss": 0.0091,
+ "step": 31815
+ },
+ {
+ "epoch": 86.69209809264305,
+ "grad_norm": 1.3886549472808838,
+ "learning_rate": 9.145591217116378e-07,
+ "loss": 0.0482,
+ "step": 31816
+ },
+ {
+ "epoch": 86.69482288828338,
+ "grad_norm": 1.2640502452850342,
+ "learning_rate": 9.141904605478291e-07,
+ "loss": 0.0268,
+ "step": 31817
+ },
+ {
+ "epoch": 86.6975476839237,
+ "grad_norm": 0.9433082342147827,
+ "learning_rate": 9.138218701440138e-07,
+ "loss": 0.0116,
+ "step": 31818
+ },
+ {
+ "epoch": 86.70027247956403,
+ "grad_norm": 0.617490828037262,
+ "learning_rate": 9.134533505030596e-07,
+ "loss": 0.0053,
+ "step": 31819
+ },
+ {
+ "epoch": 86.70299727520435,
+ "grad_norm": 1.7367496490478516,
+ "learning_rate": 9.130849016278376e-07,
+ "loss": 0.0579,
+ "step": 31820
+ },
+ {
+ "epoch": 86.70572207084469,
+ "grad_norm": 1.1469571590423584,
+ "learning_rate": 9.127165235212199e-07,
+ "loss": 0.0242,
+ "step": 31821
+ },
+ {
+ "epoch": 86.70844686648502,
+ "grad_norm": 1.8616083860397339,
+ "learning_rate": 9.12348216186072e-07,
+ "loss": 0.015,
+ "step": 31822
+ },
+ {
+ "epoch": 86.71117166212534,
+ "grad_norm": 1.1538841724395752,
+ "learning_rate": 9.119799796252626e-07,
+ "loss": 0.013,
+ "step": 31823
+ },
+ {
+ "epoch": 86.71389645776567,
+ "grad_norm": 1.5582326650619507,
+ "learning_rate": 9.116118138416619e-07,
+ "loss": 0.0197,
+ "step": 31824
+ },
+ {
+ "epoch": 86.71662125340599,
+ "grad_norm": 1.099698543548584,
+ "learning_rate": 9.112437188381362e-07,
+ "loss": 0.0151,
+ "step": 31825
+ },
+ {
+ "epoch": 86.71934604904632,
+ "grad_norm": 1.268071174621582,
+ "learning_rate": 9.108756946175523e-07,
+ "loss": 0.0441,
+ "step": 31826
+ },
+ {
+ "epoch": 86.72207084468666,
+ "grad_norm": 1.6268820762634277,
+ "learning_rate": 9.105077411827757e-07,
+ "loss": 0.1009,
+ "step": 31827
+ },
+ {
+ "epoch": 86.72479564032697,
+ "grad_norm": 1.3175957202911377,
+ "learning_rate": 9.101398585366705e-07,
+ "loss": 0.0602,
+ "step": 31828
+ },
+ {
+ "epoch": 86.7275204359673,
+ "grad_norm": 0.9765293002128601,
+ "learning_rate": 9.097720466821059e-07,
+ "loss": 0.0194,
+ "step": 31829
+ },
+ {
+ "epoch": 86.73024523160763,
+ "grad_norm": 2.5108861923217773,
+ "learning_rate": 9.094043056219437e-07,
+ "loss": 0.0079,
+ "step": 31830
+ },
+ {
+ "epoch": 86.73297002724796,
+ "grad_norm": 0.9313309192657471,
+ "learning_rate": 9.090366353590485e-07,
+ "loss": 0.0153,
+ "step": 31831
+ },
+ {
+ "epoch": 86.73569482288828,
+ "grad_norm": 1.6624470949172974,
+ "learning_rate": 9.086690358962813e-07,
+ "loss": 0.0181,
+ "step": 31832
+ },
+ {
+ "epoch": 86.73841961852861,
+ "grad_norm": 1.0148653984069824,
+ "learning_rate": 9.083015072365087e-07,
+ "loss": 0.0102,
+ "step": 31833
+ },
+ {
+ "epoch": 86.74114441416894,
+ "grad_norm": 2.0661098957061768,
+ "learning_rate": 9.079340493825916e-07,
+ "loss": 0.0298,
+ "step": 31834
+ },
+ {
+ "epoch": 86.74386920980926,
+ "grad_norm": 1.2472600936889648,
+ "learning_rate": 9.075666623373913e-07,
+ "loss": 0.0874,
+ "step": 31835
+ },
+ {
+ "epoch": 86.7465940054496,
+ "grad_norm": 1.0020045042037964,
+ "learning_rate": 9.071993461037676e-07,
+ "loss": 0.0245,
+ "step": 31836
+ },
+ {
+ "epoch": 86.74931880108991,
+ "grad_norm": 1.1759446859359741,
+ "learning_rate": 9.068321006845837e-07,
+ "loss": 0.0432,
+ "step": 31837
+ },
+ {
+ "epoch": 86.75204359673025,
+ "grad_norm": 0.714281439781189,
+ "learning_rate": 9.064649260826997e-07,
+ "loss": 0.01,
+ "step": 31838
+ },
+ {
+ "epoch": 86.75476839237058,
+ "grad_norm": 1.332667589187622,
+ "learning_rate": 9.060978223009742e-07,
+ "loss": 0.0149,
+ "step": 31839
+ },
+ {
+ "epoch": 86.7574931880109,
+ "grad_norm": 1.530187964439392,
+ "learning_rate": 9.05730789342264e-07,
+ "loss": 0.0254,
+ "step": 31840
+ },
+ {
+ "epoch": 86.76021798365123,
+ "grad_norm": 1.1519681215286255,
+ "learning_rate": 9.053638272094301e-07,
+ "loss": 0.0159,
+ "step": 31841
+ },
+ {
+ "epoch": 86.76294277929155,
+ "grad_norm": 0.6790101528167725,
+ "learning_rate": 9.049969359053312e-07,
+ "loss": 0.0096,
+ "step": 31842
+ },
+ {
+ "epoch": 86.76566757493188,
+ "grad_norm": 1.1410365104675293,
+ "learning_rate": 9.046301154328241e-07,
+ "loss": 0.0735,
+ "step": 31843
+ },
+ {
+ "epoch": 86.7683923705722,
+ "grad_norm": 1.0868351459503174,
+ "learning_rate": 9.04263365794762e-07,
+ "loss": 0.0114,
+ "step": 31844
+ },
+ {
+ "epoch": 86.77111716621253,
+ "grad_norm": 1.4164373874664307,
+ "learning_rate": 9.038966869940069e-07,
+ "loss": 0.0126,
+ "step": 31845
+ },
+ {
+ "epoch": 86.77384196185287,
+ "grad_norm": 1.1406253576278687,
+ "learning_rate": 9.035300790334112e-07,
+ "loss": 0.041,
+ "step": 31846
+ },
+ {
+ "epoch": 86.77656675749319,
+ "grad_norm": 1.1731103658676147,
+ "learning_rate": 9.031635419158314e-07,
+ "loss": 0.0205,
+ "step": 31847
+ },
+ {
+ "epoch": 86.77929155313352,
+ "grad_norm": 0.8510398864746094,
+ "learning_rate": 9.027970756441184e-07,
+ "loss": 0.0086,
+ "step": 31848
+ },
+ {
+ "epoch": 86.78201634877384,
+ "grad_norm": 1.2029154300689697,
+ "learning_rate": 9.024306802211313e-07,
+ "loss": 0.0138,
+ "step": 31849
+ },
+ {
+ "epoch": 86.78474114441417,
+ "grad_norm": 1.029610514640808,
+ "learning_rate": 9.020643556497211e-07,
+ "loss": 0.0131,
+ "step": 31850
+ },
+ {
+ "epoch": 86.7874659400545,
+ "grad_norm": 1.2202783823013306,
+ "learning_rate": 9.016981019327397e-07,
+ "loss": 0.022,
+ "step": 31851
+ },
+ {
+ "epoch": 86.79019073569482,
+ "grad_norm": 0.9652754068374634,
+ "learning_rate": 9.013319190730407e-07,
+ "loss": 0.0078,
+ "step": 31852
+ },
+ {
+ "epoch": 86.79291553133515,
+ "grad_norm": 1.609299898147583,
+ "learning_rate": 9.009658070734762e-07,
+ "loss": 0.1415,
+ "step": 31853
+ },
+ {
+ "epoch": 86.79564032697547,
+ "grad_norm": 0.38093629479408264,
+ "learning_rate": 9.005997659368982e-07,
+ "loss": 0.0029,
+ "step": 31854
+ },
+ {
+ "epoch": 86.7983651226158,
+ "grad_norm": 1.1814299821853638,
+ "learning_rate": 9.002337956661556e-07,
+ "loss": 0.0259,
+ "step": 31855
+ },
+ {
+ "epoch": 86.80108991825612,
+ "grad_norm": 0.9230058193206787,
+ "learning_rate": 8.998678962640983e-07,
+ "loss": 0.0127,
+ "step": 31856
+ },
+ {
+ "epoch": 86.80381471389646,
+ "grad_norm": 1.3133575916290283,
+ "learning_rate": 8.995020677335764e-07,
+ "loss": 0.0114,
+ "step": 31857
+ },
+ {
+ "epoch": 86.80653950953679,
+ "grad_norm": 1.1202949285507202,
+ "learning_rate": 8.99136310077442e-07,
+ "loss": 0.078,
+ "step": 31858
+ },
+ {
+ "epoch": 86.80926430517711,
+ "grad_norm": 1.1732218265533447,
+ "learning_rate": 8.987706232985405e-07,
+ "loss": 0.0126,
+ "step": 31859
+ },
+ {
+ "epoch": 86.81198910081744,
+ "grad_norm": 1.281707763671875,
+ "learning_rate": 8.984050073997197e-07,
+ "loss": 0.0196,
+ "step": 31860
+ },
+ {
+ "epoch": 86.81471389645776,
+ "grad_norm": 0.8080025911331177,
+ "learning_rate": 8.980394623838273e-07,
+ "loss": 0.0156,
+ "step": 31861
+ },
+ {
+ "epoch": 86.8174386920981,
+ "grad_norm": 1.5645309686660767,
+ "learning_rate": 8.976739882537122e-07,
+ "loss": 0.0805,
+ "step": 31862
+ },
+ {
+ "epoch": 86.82016348773843,
+ "grad_norm": 1.5215656757354736,
+ "learning_rate": 8.973085850122187e-07,
+ "loss": 0.0174,
+ "step": 31863
+ },
+ {
+ "epoch": 86.82288828337875,
+ "grad_norm": 1.1225882768630981,
+ "learning_rate": 8.969432526621924e-07,
+ "loss": 0.0145,
+ "step": 31864
+ },
+ {
+ "epoch": 86.82561307901908,
+ "grad_norm": 1.707106113433838,
+ "learning_rate": 8.965779912064787e-07,
+ "loss": 0.0234,
+ "step": 31865
+ },
+ {
+ "epoch": 86.8283378746594,
+ "grad_norm": 0.9600009918212891,
+ "learning_rate": 8.962128006479232e-07,
+ "loss": 0.0106,
+ "step": 31866
+ },
+ {
+ "epoch": 86.83106267029973,
+ "grad_norm": 1.0464403629302979,
+ "learning_rate": 8.958476809893702e-07,
+ "loss": 0.0092,
+ "step": 31867
+ },
+ {
+ "epoch": 86.83378746594005,
+ "grad_norm": 1.4683841466903687,
+ "learning_rate": 8.954826322336596e-07,
+ "loss": 0.0331,
+ "step": 31868
+ },
+ {
+ "epoch": 86.83651226158038,
+ "grad_norm": 1.2202757596969604,
+ "learning_rate": 8.951176543836404e-07,
+ "loss": 0.0169,
+ "step": 31869
+ },
+ {
+ "epoch": 86.83923705722071,
+ "grad_norm": 0.571311354637146,
+ "learning_rate": 8.947527474421513e-07,
+ "loss": 0.0047,
+ "step": 31870
+ },
+ {
+ "epoch": 86.84196185286103,
+ "grad_norm": 1.029624104499817,
+ "learning_rate": 8.943879114120347e-07,
+ "loss": 0.0436,
+ "step": 31871
+ },
+ {
+ "epoch": 86.84468664850137,
+ "grad_norm": 1.143439769744873,
+ "learning_rate": 8.940231462961302e-07,
+ "loss": 0.0127,
+ "step": 31872
+ },
+ {
+ "epoch": 86.84741144414168,
+ "grad_norm": 0.9800959825515747,
+ "learning_rate": 8.936584520972824e-07,
+ "loss": 0.0493,
+ "step": 31873
+ },
+ {
+ "epoch": 86.85013623978202,
+ "grad_norm": 1.063007116317749,
+ "learning_rate": 8.932938288183268e-07,
+ "loss": 0.03,
+ "step": 31874
+ },
+ {
+ "epoch": 86.85286103542235,
+ "grad_norm": 0.7182239890098572,
+ "learning_rate": 8.929292764621078e-07,
+ "loss": 0.0074,
+ "step": 31875
+ },
+ {
+ "epoch": 86.85558583106267,
+ "grad_norm": 1.1285161972045898,
+ "learning_rate": 8.925647950314631e-07,
+ "loss": 0.0096,
+ "step": 31876
+ },
+ {
+ "epoch": 86.858310626703,
+ "grad_norm": 1.3380343914031982,
+ "learning_rate": 8.922003845292293e-07,
+ "loss": 0.0143,
+ "step": 31877
+ },
+ {
+ "epoch": 86.86103542234332,
+ "grad_norm": 1.2307947874069214,
+ "learning_rate": 8.918360449582464e-07,
+ "loss": 0.0201,
+ "step": 31878
+ },
+ {
+ "epoch": 86.86376021798365,
+ "grad_norm": 1.2448254823684692,
+ "learning_rate": 8.914717763213521e-07,
+ "loss": 0.0087,
+ "step": 31879
+ },
+ {
+ "epoch": 86.86648501362397,
+ "grad_norm": 2.0500776767730713,
+ "learning_rate": 8.911075786213819e-07,
+ "loss": 0.0857,
+ "step": 31880
+ },
+ {
+ "epoch": 86.8692098092643,
+ "grad_norm": 0.952014684677124,
+ "learning_rate": 8.907434518611713e-07,
+ "loss": 0.0121,
+ "step": 31881
+ },
+ {
+ "epoch": 86.87193460490464,
+ "grad_norm": 1.50927734375,
+ "learning_rate": 8.903793960435592e-07,
+ "loss": 0.0976,
+ "step": 31882
+ },
+ {
+ "epoch": 86.87465940054496,
+ "grad_norm": 1.1840922832489014,
+ "learning_rate": 8.900154111713787e-07,
+ "loss": 0.0136,
+ "step": 31883
+ },
+ {
+ "epoch": 86.87738419618529,
+ "grad_norm": 0.39408785104751587,
+ "learning_rate": 8.896514972474657e-07,
+ "loss": 0.0046,
+ "step": 31884
+ },
+ {
+ "epoch": 86.88010899182561,
+ "grad_norm": 1.2891830205917358,
+ "learning_rate": 8.892876542746509e-07,
+ "loss": 0.0427,
+ "step": 31885
+ },
+ {
+ "epoch": 86.88283378746594,
+ "grad_norm": 1.8999143838882446,
+ "learning_rate": 8.889238822557733e-07,
+ "loss": 0.0206,
+ "step": 31886
+ },
+ {
+ "epoch": 86.88555858310627,
+ "grad_norm": 1.4140346050262451,
+ "learning_rate": 8.885601811936628e-07,
+ "loss": 0.0798,
+ "step": 31887
+ },
+ {
+ "epoch": 86.88828337874659,
+ "grad_norm": 0.8552857041358948,
+ "learning_rate": 8.881965510911506e-07,
+ "loss": 0.0082,
+ "step": 31888
+ },
+ {
+ "epoch": 86.89100817438693,
+ "grad_norm": 0.841987133026123,
+ "learning_rate": 8.87832991951072e-07,
+ "loss": 0.0074,
+ "step": 31889
+ },
+ {
+ "epoch": 86.89373297002724,
+ "grad_norm": 1.516995906829834,
+ "learning_rate": 8.874695037762571e-07,
+ "loss": 0.0561,
+ "step": 31890
+ },
+ {
+ "epoch": 86.89645776566758,
+ "grad_norm": 1.3121954202651978,
+ "learning_rate": 8.871060865695347e-07,
+ "loss": 0.0114,
+ "step": 31891
+ },
+ {
+ "epoch": 86.8991825613079,
+ "grad_norm": 1.1445640325546265,
+ "learning_rate": 8.867427403337392e-07,
+ "loss": 0.0068,
+ "step": 31892
+ },
+ {
+ "epoch": 86.90190735694823,
+ "grad_norm": 0.9304394125938416,
+ "learning_rate": 8.86379465071695e-07,
+ "loss": 0.0076,
+ "step": 31893
+ },
+ {
+ "epoch": 86.90463215258856,
+ "grad_norm": 0.8625255227088928,
+ "learning_rate": 8.860162607862366e-07,
+ "loss": 0.0089,
+ "step": 31894
+ },
+ {
+ "epoch": 86.90735694822888,
+ "grad_norm": 0.678864061832428,
+ "learning_rate": 8.856531274801904e-07,
+ "loss": 0.0751,
+ "step": 31895
+ },
+ {
+ "epoch": 86.91008174386921,
+ "grad_norm": 0.7345802187919617,
+ "learning_rate": 8.852900651563833e-07,
+ "loss": 0.0061,
+ "step": 31896
+ },
+ {
+ "epoch": 86.91280653950953,
+ "grad_norm": 1.2723158597946167,
+ "learning_rate": 8.84927073817643e-07,
+ "loss": 0.0225,
+ "step": 31897
+ },
+ {
+ "epoch": 86.91553133514986,
+ "grad_norm": 0.7054772973060608,
+ "learning_rate": 8.845641534667993e-07,
+ "loss": 0.0054,
+ "step": 31898
+ },
+ {
+ "epoch": 86.9182561307902,
+ "grad_norm": 1.3271297216415405,
+ "learning_rate": 8.842013041066754e-07,
+ "loss": 0.0504,
+ "step": 31899
+ },
+ {
+ "epoch": 86.92098092643052,
+ "grad_norm": 1.4712973833084106,
+ "learning_rate": 8.838385257400994e-07,
+ "loss": 0.0124,
+ "step": 31900
+ },
+ {
+ "epoch": 86.92370572207085,
+ "grad_norm": 1.3929105997085571,
+ "learning_rate": 8.834758183698933e-07,
+ "loss": 0.0094,
+ "step": 31901
+ },
+ {
+ "epoch": 86.92643051771117,
+ "grad_norm": 1.8474277257919312,
+ "learning_rate": 8.831131819988859e-07,
+ "loss": 0.0134,
+ "step": 31902
+ },
+ {
+ "epoch": 86.9291553133515,
+ "grad_norm": 1.2334834337234497,
+ "learning_rate": 8.827506166298994e-07,
+ "loss": 0.0721,
+ "step": 31903
+ },
+ {
+ "epoch": 86.93188010899182,
+ "grad_norm": 1.0671985149383545,
+ "learning_rate": 8.823881222657582e-07,
+ "loss": 0.0463,
+ "step": 31904
+ },
+ {
+ "epoch": 86.93460490463215,
+ "grad_norm": 1.4273265600204468,
+ "learning_rate": 8.820256989092824e-07,
+ "loss": 0.0214,
+ "step": 31905
+ },
+ {
+ "epoch": 86.93732970027249,
+ "grad_norm": 1.4143832921981812,
+ "learning_rate": 8.816633465632996e-07,
+ "loss": 0.0763,
+ "step": 31906
+ },
+ {
+ "epoch": 86.9400544959128,
+ "grad_norm": 1.1067320108413696,
+ "learning_rate": 8.813010652306286e-07,
+ "loss": 0.0067,
+ "step": 31907
+ },
+ {
+ "epoch": 86.94277929155314,
+ "grad_norm": 0.7655547857284546,
+ "learning_rate": 8.809388549140918e-07,
+ "loss": 0.0086,
+ "step": 31908
+ },
+ {
+ "epoch": 86.94550408719346,
+ "grad_norm": 0.6154427528381348,
+ "learning_rate": 8.805767156165057e-07,
+ "loss": 0.0048,
+ "step": 31909
+ },
+ {
+ "epoch": 86.94822888283379,
+ "grad_norm": 1.3427239656448364,
+ "learning_rate": 8.802146473406991e-07,
+ "loss": 0.0376,
+ "step": 31910
+ },
+ {
+ "epoch": 86.95095367847412,
+ "grad_norm": 0.8470531105995178,
+ "learning_rate": 8.798526500894866e-07,
+ "loss": 0.0091,
+ "step": 31911
+ },
+ {
+ "epoch": 86.95367847411444,
+ "grad_norm": 1.0821675062179565,
+ "learning_rate": 8.79490723865688e-07,
+ "loss": 0.0076,
+ "step": 31912
+ },
+ {
+ "epoch": 86.95640326975477,
+ "grad_norm": 1.6674662828445435,
+ "learning_rate": 8.7912886867212e-07,
+ "loss": 0.0225,
+ "step": 31913
+ },
+ {
+ "epoch": 86.95912806539509,
+ "grad_norm": 1.0136361122131348,
+ "learning_rate": 8.787670845116059e-07,
+ "loss": 0.054,
+ "step": 31914
+ },
+ {
+ "epoch": 86.96185286103542,
+ "grad_norm": 0.9572783708572388,
+ "learning_rate": 8.784053713869589e-07,
+ "loss": 0.0136,
+ "step": 31915
+ },
+ {
+ "epoch": 86.96457765667574,
+ "grad_norm": 0.9930816888809204,
+ "learning_rate": 8.78043729300998e-07,
+ "loss": 0.0168,
+ "step": 31916
+ },
+ {
+ "epoch": 86.96730245231608,
+ "grad_norm": 0.9399051666259766,
+ "learning_rate": 8.776821582565365e-07,
+ "loss": 0.0091,
+ "step": 31917
+ },
+ {
+ "epoch": 86.97002724795641,
+ "grad_norm": 0.9285911321640015,
+ "learning_rate": 8.773206582563953e-07,
+ "loss": 0.0071,
+ "step": 31918
+ },
+ {
+ "epoch": 86.97275204359673,
+ "grad_norm": 0.6618587970733643,
+ "learning_rate": 8.769592293033868e-07,
+ "loss": 0.0079,
+ "step": 31919
+ },
+ {
+ "epoch": 86.97547683923706,
+ "grad_norm": 0.6305716037750244,
+ "learning_rate": 8.765978714003265e-07,
+ "loss": 0.0061,
+ "step": 31920
+ },
+ {
+ "epoch": 86.97820163487738,
+ "grad_norm": 1.3354281187057495,
+ "learning_rate": 8.762365845500265e-07,
+ "loss": 0.0143,
+ "step": 31921
+ },
+ {
+ "epoch": 86.98092643051771,
+ "grad_norm": 1.1019819974899292,
+ "learning_rate": 8.758753687553035e-07,
+ "loss": 0.1152,
+ "step": 31922
+ },
+ {
+ "epoch": 86.98365122615803,
+ "grad_norm": 0.9511229991912842,
+ "learning_rate": 8.755142240189706e-07,
+ "loss": 0.0059,
+ "step": 31923
+ },
+ {
+ "epoch": 86.98637602179836,
+ "grad_norm": 0.5197628140449524,
+ "learning_rate": 8.751531503438382e-07,
+ "loss": 0.0054,
+ "step": 31924
+ },
+ {
+ "epoch": 86.9891008174387,
+ "grad_norm": 1.1228231191635132,
+ "learning_rate": 8.747921477327182e-07,
+ "loss": 0.0748,
+ "step": 31925
+ },
+ {
+ "epoch": 86.99182561307902,
+ "grad_norm": 1.0610140562057495,
+ "learning_rate": 8.744312161884228e-07,
+ "loss": 0.017,
+ "step": 31926
+ },
+ {
+ "epoch": 86.99455040871935,
+ "grad_norm": 1.2571550607681274,
+ "learning_rate": 8.740703557137653e-07,
+ "loss": 0.0113,
+ "step": 31927
+ },
+ {
+ "epoch": 86.99727520435967,
+ "grad_norm": 1.1422632932662964,
+ "learning_rate": 8.737095663115547e-07,
+ "loss": 0.1005,
+ "step": 31928
+ },
+ {
+ "epoch": 87.0,
+ "grad_norm": 0.9703148603439331,
+ "learning_rate": 8.733488479845997e-07,
+ "loss": 0.0133,
+ "step": 31929
+ },
+ {
+ "epoch": 87.00272479564033,
+ "grad_norm": 1.501866340637207,
+ "learning_rate": 8.729882007357083e-07,
+ "loss": 0.0108,
+ "step": 31930
+ },
+ {
+ "epoch": 87.00544959128065,
+ "grad_norm": 1.4724425077438354,
+ "learning_rate": 8.726276245676934e-07,
+ "loss": 0.0869,
+ "step": 31931
+ },
+ {
+ "epoch": 87.00817438692098,
+ "grad_norm": 0.7004421949386597,
+ "learning_rate": 8.722671194833598e-07,
+ "loss": 0.0088,
+ "step": 31932
+ },
+ {
+ "epoch": 87.0108991825613,
+ "grad_norm": 1.958540678024292,
+ "learning_rate": 8.719066854855151e-07,
+ "loss": 0.0675,
+ "step": 31933
+ },
+ {
+ "epoch": 87.01362397820164,
+ "grad_norm": 2.762939214706421,
+ "learning_rate": 8.715463225769693e-07,
+ "loss": 0.0166,
+ "step": 31934
+ },
+ {
+ "epoch": 87.01634877384195,
+ "grad_norm": 2.1067450046539307,
+ "learning_rate": 8.711860307605257e-07,
+ "loss": 0.0509,
+ "step": 31935
+ },
+ {
+ "epoch": 87.01907356948229,
+ "grad_norm": 0.44561323523521423,
+ "learning_rate": 8.708258100389933e-07,
+ "loss": 0.0048,
+ "step": 31936
+ },
+ {
+ "epoch": 87.02179836512262,
+ "grad_norm": 0.3438386619091034,
+ "learning_rate": 8.704656604151729e-07,
+ "loss": 0.0037,
+ "step": 31937
+ },
+ {
+ "epoch": 87.02452316076294,
+ "grad_norm": 0.4144197106361389,
+ "learning_rate": 8.701055818918736e-07,
+ "loss": 0.0051,
+ "step": 31938
+ },
+ {
+ "epoch": 87.02724795640327,
+ "grad_norm": 4.078123092651367,
+ "learning_rate": 8.697455744718985e-07,
+ "loss": 0.0459,
+ "step": 31939
+ },
+ {
+ "epoch": 87.02997275204359,
+ "grad_norm": 1.0164738893508911,
+ "learning_rate": 8.693856381580512e-07,
+ "loss": 0.077,
+ "step": 31940
+ },
+ {
+ "epoch": 87.03269754768392,
+ "grad_norm": 1.267040491104126,
+ "learning_rate": 8.690257729531326e-07,
+ "loss": 0.0096,
+ "step": 31941
+ },
+ {
+ "epoch": 87.03542234332426,
+ "grad_norm": 0.8147222995758057,
+ "learning_rate": 8.686659788599494e-07,
+ "loss": 0.0086,
+ "step": 31942
+ },
+ {
+ "epoch": 87.03814713896458,
+ "grad_norm": 1.0424702167510986,
+ "learning_rate": 8.683062558812994e-07,
+ "loss": 0.0122,
+ "step": 31943
+ },
+ {
+ "epoch": 87.04087193460491,
+ "grad_norm": 0.9953773021697998,
+ "learning_rate": 8.679466040199891e-07,
+ "loss": 0.0181,
+ "step": 31944
+ },
+ {
+ "epoch": 87.04359673024523,
+ "grad_norm": 1.2302438020706177,
+ "learning_rate": 8.675870232788164e-07,
+ "loss": 0.0935,
+ "step": 31945
+ },
+ {
+ "epoch": 87.04632152588556,
+ "grad_norm": 1.745718240737915,
+ "learning_rate": 8.672275136605801e-07,
+ "loss": 0.0626,
+ "step": 31946
+ },
+ {
+ "epoch": 87.04904632152588,
+ "grad_norm": 0.6896252036094666,
+ "learning_rate": 8.668680751680836e-07,
+ "loss": 0.0087,
+ "step": 31947
+ },
+ {
+ "epoch": 87.05177111716621,
+ "grad_norm": 0.7153891324996948,
+ "learning_rate": 8.665087078041245e-07,
+ "loss": 0.0049,
+ "step": 31948
+ },
+ {
+ "epoch": 87.05449591280654,
+ "grad_norm": 0.7339609265327454,
+ "learning_rate": 8.661494115715019e-07,
+ "loss": 0.0052,
+ "step": 31949
+ },
+ {
+ "epoch": 87.05722070844686,
+ "grad_norm": 1.944807529449463,
+ "learning_rate": 8.657901864730112e-07,
+ "loss": 0.0196,
+ "step": 31950
+ },
+ {
+ "epoch": 87.0599455040872,
+ "grad_norm": 0.8994581699371338,
+ "learning_rate": 8.654310325114557e-07,
+ "loss": 0.0232,
+ "step": 31951
+ },
+ {
+ "epoch": 87.06267029972751,
+ "grad_norm": 0.8814605474472046,
+ "learning_rate": 8.650719496896287e-07,
+ "loss": 0.0295,
+ "step": 31952
+ },
+ {
+ "epoch": 87.06539509536785,
+ "grad_norm": 1.0013184547424316,
+ "learning_rate": 8.647129380103248e-07,
+ "loss": 0.0051,
+ "step": 31953
+ },
+ {
+ "epoch": 87.06811989100818,
+ "grad_norm": 1.0830539464950562,
+ "learning_rate": 8.643539974763449e-07,
+ "loss": 0.0334,
+ "step": 31954
+ },
+ {
+ "epoch": 87.0708446866485,
+ "grad_norm": 1.0241308212280273,
+ "learning_rate": 8.639951280904824e-07,
+ "loss": 0.0788,
+ "step": 31955
+ },
+ {
+ "epoch": 87.07356948228883,
+ "grad_norm": 0.5294018983840942,
+ "learning_rate": 8.636363298555317e-07,
+ "loss": 0.0073,
+ "step": 31956
+ },
+ {
+ "epoch": 87.07629427792915,
+ "grad_norm": 1.3754050731658936,
+ "learning_rate": 8.632776027742851e-07,
+ "loss": 0.0512,
+ "step": 31957
+ },
+ {
+ "epoch": 87.07901907356948,
+ "grad_norm": 0.7760536670684814,
+ "learning_rate": 8.629189468495402e-07,
+ "loss": 0.0086,
+ "step": 31958
+ },
+ {
+ "epoch": 87.0817438692098,
+ "grad_norm": 1.025914192199707,
+ "learning_rate": 8.625603620840894e-07,
+ "loss": 0.0076,
+ "step": 31959
+ },
+ {
+ "epoch": 87.08446866485014,
+ "grad_norm": 0.6028929352760315,
+ "learning_rate": 8.622018484807215e-07,
+ "loss": 0.0065,
+ "step": 31960
+ },
+ {
+ "epoch": 87.08719346049047,
+ "grad_norm": 1.1817529201507568,
+ "learning_rate": 8.618434060422331e-07,
+ "loss": 0.0174,
+ "step": 31961
+ },
+ {
+ "epoch": 87.08991825613079,
+ "grad_norm": 0.9813389182090759,
+ "learning_rate": 8.614850347714132e-07,
+ "loss": 0.0157,
+ "step": 31962
+ },
+ {
+ "epoch": 87.09264305177112,
+ "grad_norm": 1.551021695137024,
+ "learning_rate": 8.61126734671055e-07,
+ "loss": 0.0145,
+ "step": 31963
+ },
+ {
+ "epoch": 87.09536784741144,
+ "grad_norm": 0.9496782422065735,
+ "learning_rate": 8.607685057439486e-07,
+ "loss": 0.009,
+ "step": 31964
+ },
+ {
+ "epoch": 87.09809264305177,
+ "grad_norm": 0.8716671466827393,
+ "learning_rate": 8.604103479928815e-07,
+ "loss": 0.0076,
+ "step": 31965
+ },
+ {
+ "epoch": 87.1008174386921,
+ "grad_norm": 1.2491669654846191,
+ "learning_rate": 8.600522614206442e-07,
+ "loss": 0.1247,
+ "step": 31966
+ },
+ {
+ "epoch": 87.10354223433242,
+ "grad_norm": 1.1687482595443726,
+ "learning_rate": 8.596942460300261e-07,
+ "loss": 0.0211,
+ "step": 31967
+ },
+ {
+ "epoch": 87.10626702997276,
+ "grad_norm": 2.1867589950561523,
+ "learning_rate": 8.593363018238154e-07,
+ "loss": 0.0396,
+ "step": 31968
+ },
+ {
+ "epoch": 87.10899182561307,
+ "grad_norm": 1.4376840591430664,
+ "learning_rate": 8.589784288047997e-07,
+ "loss": 0.011,
+ "step": 31969
+ },
+ {
+ "epoch": 87.11171662125341,
+ "grad_norm": 1.5938578844070435,
+ "learning_rate": 8.586206269757647e-07,
+ "loss": 0.0326,
+ "step": 31970
+ },
+ {
+ "epoch": 87.11444141689373,
+ "grad_norm": 1.079949975013733,
+ "learning_rate": 8.58262896339499e-07,
+ "loss": 0.0088,
+ "step": 31971
+ },
+ {
+ "epoch": 87.11716621253406,
+ "grad_norm": 1.090402364730835,
+ "learning_rate": 8.579052368987872e-07,
+ "loss": 0.0125,
+ "step": 31972
+ },
+ {
+ "epoch": 87.11989100817439,
+ "grad_norm": 5.2678632736206055,
+ "learning_rate": 8.57547648656416e-07,
+ "loss": 0.1024,
+ "step": 31973
+ },
+ {
+ "epoch": 87.12261580381471,
+ "grad_norm": 1.1838059425354004,
+ "learning_rate": 8.571901316151676e-07,
+ "loss": 0.0097,
+ "step": 31974
+ },
+ {
+ "epoch": 87.12534059945504,
+ "grad_norm": 0.8550049662590027,
+ "learning_rate": 8.568326857778297e-07,
+ "loss": 0.005,
+ "step": 31975
+ },
+ {
+ "epoch": 87.12806539509536,
+ "grad_norm": 0.8888531923294067,
+ "learning_rate": 8.564753111471847e-07,
+ "loss": 0.0078,
+ "step": 31976
+ },
+ {
+ "epoch": 87.1307901907357,
+ "grad_norm": 1.0694398880004883,
+ "learning_rate": 8.561180077260134e-07,
+ "loss": 0.0104,
+ "step": 31977
+ },
+ {
+ "epoch": 87.13351498637603,
+ "grad_norm": 1.0144646167755127,
+ "learning_rate": 8.557607755171016e-07,
+ "loss": 0.0076,
+ "step": 31978
+ },
+ {
+ "epoch": 87.13623978201635,
+ "grad_norm": 1.3750020265579224,
+ "learning_rate": 8.554036145232314e-07,
+ "loss": 0.0119,
+ "step": 31979
+ },
+ {
+ "epoch": 87.13896457765668,
+ "grad_norm": 1.860777735710144,
+ "learning_rate": 8.550465247471839e-07,
+ "loss": 0.0159,
+ "step": 31980
+ },
+ {
+ "epoch": 87.141689373297,
+ "grad_norm": 1.211183786392212,
+ "learning_rate": 8.546895061917404e-07,
+ "loss": 0.0315,
+ "step": 31981
+ },
+ {
+ "epoch": 87.14441416893733,
+ "grad_norm": 1.2319164276123047,
+ "learning_rate": 8.543325588596785e-07,
+ "loss": 0.0055,
+ "step": 31982
+ },
+ {
+ "epoch": 87.14713896457765,
+ "grad_norm": 1.4667755365371704,
+ "learning_rate": 8.539756827537815e-07,
+ "loss": 0.0158,
+ "step": 31983
+ },
+ {
+ "epoch": 87.14986376021798,
+ "grad_norm": 1.062248945236206,
+ "learning_rate": 8.536188778768284e-07,
+ "loss": 0.0088,
+ "step": 31984
+ },
+ {
+ "epoch": 87.15258855585832,
+ "grad_norm": 0.7409974336624146,
+ "learning_rate": 8.53262144231597e-07,
+ "loss": 0.0119,
+ "step": 31985
+ },
+ {
+ "epoch": 87.15531335149863,
+ "grad_norm": 0.8657655715942383,
+ "learning_rate": 8.529054818208638e-07,
+ "loss": 0.0336,
+ "step": 31986
+ },
+ {
+ "epoch": 87.15803814713897,
+ "grad_norm": 1.4252712726593018,
+ "learning_rate": 8.525488906474089e-07,
+ "loss": 0.0717,
+ "step": 31987
+ },
+ {
+ "epoch": 87.16076294277929,
+ "grad_norm": 0.9692824482917786,
+ "learning_rate": 8.521923707140101e-07,
+ "loss": 0.007,
+ "step": 31988
+ },
+ {
+ "epoch": 87.16348773841962,
+ "grad_norm": 1.0053894519805908,
+ "learning_rate": 8.518359220234429e-07,
+ "loss": 0.0142,
+ "step": 31989
+ },
+ {
+ "epoch": 87.16621253405995,
+ "grad_norm": 1.4012635946273804,
+ "learning_rate": 8.514795445784807e-07,
+ "loss": 0.0139,
+ "step": 31990
+ },
+ {
+ "epoch": 87.16893732970027,
+ "grad_norm": 1.481851577758789,
+ "learning_rate": 8.511232383819024e-07,
+ "loss": 0.1052,
+ "step": 31991
+ },
+ {
+ "epoch": 87.1716621253406,
+ "grad_norm": 1.242518424987793,
+ "learning_rate": 8.507670034364834e-07,
+ "loss": 0.0345,
+ "step": 31992
+ },
+ {
+ "epoch": 87.17438692098092,
+ "grad_norm": 1.4043967723846436,
+ "learning_rate": 8.504108397449962e-07,
+ "loss": 0.0081,
+ "step": 31993
+ },
+ {
+ "epoch": 87.17711171662125,
+ "grad_norm": 1.1592986583709717,
+ "learning_rate": 8.500547473102116e-07,
+ "loss": 0.0133,
+ "step": 31994
+ },
+ {
+ "epoch": 87.17983651226157,
+ "grad_norm": 2.222552537918091,
+ "learning_rate": 8.496987261349077e-07,
+ "loss": 0.012,
+ "step": 31995
+ },
+ {
+ "epoch": 87.1825613079019,
+ "grad_norm": 0.8621810078620911,
+ "learning_rate": 8.493427762218575e-07,
+ "loss": 0.0095,
+ "step": 31996
+ },
+ {
+ "epoch": 87.18528610354224,
+ "grad_norm": 1.3656619787216187,
+ "learning_rate": 8.489868975738303e-07,
+ "loss": 0.018,
+ "step": 31997
+ },
+ {
+ "epoch": 87.18801089918256,
+ "grad_norm": 0.7458478808403015,
+ "learning_rate": 8.486310901935968e-07,
+ "loss": 0.0096,
+ "step": 31998
+ },
+ {
+ "epoch": 87.19073569482289,
+ "grad_norm": 0.5114712715148926,
+ "learning_rate": 8.482753540839328e-07,
+ "loss": 0.0045,
+ "step": 31999
+ },
+ {
+ "epoch": 87.19346049046321,
+ "grad_norm": 1.1399693489074707,
+ "learning_rate": 8.479196892476049e-07,
+ "loss": 0.021,
+ "step": 32000
+ },
+ {
+ "epoch": 87.19618528610354,
+ "grad_norm": 1.055837869644165,
+ "learning_rate": 8.475640956873843e-07,
+ "loss": 0.0217,
+ "step": 32001
+ },
+ {
+ "epoch": 87.19891008174388,
+ "grad_norm": 0.9309377670288086,
+ "learning_rate": 8.472085734060376e-07,
+ "loss": 0.0278,
+ "step": 32002
+ },
+ {
+ "epoch": 87.2016348773842,
+ "grad_norm": 0.6696506142616272,
+ "learning_rate": 8.468531224063381e-07,
+ "loss": 0.0085,
+ "step": 32003
+ },
+ {
+ "epoch": 87.20435967302453,
+ "grad_norm": 3.322969675064087,
+ "learning_rate": 8.464977426910514e-07,
+ "loss": 0.0085,
+ "step": 32004
+ },
+ {
+ "epoch": 87.20708446866485,
+ "grad_norm": 0.8461806774139404,
+ "learning_rate": 8.461424342629465e-07,
+ "loss": 0.0093,
+ "step": 32005
+ },
+ {
+ "epoch": 87.20980926430518,
+ "grad_norm": 1.176608681678772,
+ "learning_rate": 8.457871971247877e-07,
+ "loss": 0.0397,
+ "step": 32006
+ },
+ {
+ "epoch": 87.2125340599455,
+ "grad_norm": 0.8143172860145569,
+ "learning_rate": 8.454320312793451e-07,
+ "loss": 0.0064,
+ "step": 32007
+ },
+ {
+ "epoch": 87.21525885558583,
+ "grad_norm": 1.4184406995773315,
+ "learning_rate": 8.450769367293832e-07,
+ "loss": 0.0096,
+ "step": 32008
+ },
+ {
+ "epoch": 87.21798365122616,
+ "grad_norm": 0.7233802080154419,
+ "learning_rate": 8.447219134776674e-07,
+ "loss": 0.0063,
+ "step": 32009
+ },
+ {
+ "epoch": 87.22070844686648,
+ "grad_norm": 0.998618483543396,
+ "learning_rate": 8.443669615269611e-07,
+ "loss": 0.0819,
+ "step": 32010
+ },
+ {
+ "epoch": 87.22343324250681,
+ "grad_norm": 0.9382662177085876,
+ "learning_rate": 8.440120808800323e-07,
+ "loss": 0.016,
+ "step": 32011
+ },
+ {
+ "epoch": 87.22615803814713,
+ "grad_norm": 1.3288559913635254,
+ "learning_rate": 8.436572715396407e-07,
+ "loss": 0.042,
+ "step": 32012
+ },
+ {
+ "epoch": 87.22888283378747,
+ "grad_norm": 0.8337856531143188,
+ "learning_rate": 8.433025335085532e-07,
+ "loss": 0.0093,
+ "step": 32013
+ },
+ {
+ "epoch": 87.2316076294278,
+ "grad_norm": 0.7378425002098083,
+ "learning_rate": 8.429478667895307e-07,
+ "loss": 0.0047,
+ "step": 32014
+ },
+ {
+ "epoch": 87.23433242506812,
+ "grad_norm": 1.6069865226745605,
+ "learning_rate": 8.425932713853346e-07,
+ "loss": 0.0276,
+ "step": 32015
+ },
+ {
+ "epoch": 87.23705722070845,
+ "grad_norm": 0.7862610816955566,
+ "learning_rate": 8.422387472987281e-07,
+ "loss": 0.0077,
+ "step": 32016
+ },
+ {
+ "epoch": 87.23978201634877,
+ "grad_norm": 1.0770453214645386,
+ "learning_rate": 8.418842945324724e-07,
+ "loss": 0.0077,
+ "step": 32017
+ },
+ {
+ "epoch": 87.2425068119891,
+ "grad_norm": 1.4871577024459839,
+ "learning_rate": 8.415299130893251e-07,
+ "loss": 0.013,
+ "step": 32018
+ },
+ {
+ "epoch": 87.24523160762942,
+ "grad_norm": 1.0151830911636353,
+ "learning_rate": 8.411756029720497e-07,
+ "loss": 0.0157,
+ "step": 32019
+ },
+ {
+ "epoch": 87.24795640326975,
+ "grad_norm": 0.8940179944038391,
+ "learning_rate": 8.408213641834029e-07,
+ "loss": 0.0084,
+ "step": 32020
+ },
+ {
+ "epoch": 87.25068119891009,
+ "grad_norm": 1.1225570440292358,
+ "learning_rate": 8.404671967261457e-07,
+ "loss": 0.0319,
+ "step": 32021
+ },
+ {
+ "epoch": 87.2534059945504,
+ "grad_norm": 1.0083731412887573,
+ "learning_rate": 8.401131006030327e-07,
+ "loss": 0.0135,
+ "step": 32022
+ },
+ {
+ "epoch": 87.25613079019074,
+ "grad_norm": 1.527470588684082,
+ "learning_rate": 8.397590758168272e-07,
+ "loss": 0.0463,
+ "step": 32023
+ },
+ {
+ "epoch": 87.25885558583106,
+ "grad_norm": 0.9080622792243958,
+ "learning_rate": 8.394051223702815e-07,
+ "loss": 0.0141,
+ "step": 32024
+ },
+ {
+ "epoch": 87.26158038147139,
+ "grad_norm": 2.0018808841705322,
+ "learning_rate": 8.390512402661555e-07,
+ "loss": 0.1126,
+ "step": 32025
+ },
+ {
+ "epoch": 87.26430517711172,
+ "grad_norm": 0.9326978325843811,
+ "learning_rate": 8.386974295072004e-07,
+ "loss": 0.0077,
+ "step": 32026
+ },
+ {
+ "epoch": 87.26702997275204,
+ "grad_norm": 0.7758012413978577,
+ "learning_rate": 8.383436900961772e-07,
+ "loss": 0.0184,
+ "step": 32027
+ },
+ {
+ "epoch": 87.26975476839237,
+ "grad_norm": 0.8474690318107605,
+ "learning_rate": 8.379900220358384e-07,
+ "loss": 0.0106,
+ "step": 32028
+ },
+ {
+ "epoch": 87.2724795640327,
+ "grad_norm": 1.6296659708023071,
+ "learning_rate": 8.376364253289371e-07,
+ "loss": 0.1103,
+ "step": 32029
+ },
+ {
+ "epoch": 87.27520435967303,
+ "grad_norm": 0.8888805508613586,
+ "learning_rate": 8.372828999782301e-07,
+ "loss": 0.0089,
+ "step": 32030
+ },
+ {
+ "epoch": 87.27792915531334,
+ "grad_norm": 1.3581703901290894,
+ "learning_rate": 8.369294459864663e-07,
+ "loss": 0.0077,
+ "step": 32031
+ },
+ {
+ "epoch": 87.28065395095368,
+ "grad_norm": 0.9423989653587341,
+ "learning_rate": 8.365760633564035e-07,
+ "loss": 0.0487,
+ "step": 32032
+ },
+ {
+ "epoch": 87.28337874659401,
+ "grad_norm": 0.6789316534996033,
+ "learning_rate": 8.362227520907918e-07,
+ "loss": 0.0073,
+ "step": 32033
+ },
+ {
+ "epoch": 87.28610354223433,
+ "grad_norm": 1.2746474742889404,
+ "learning_rate": 8.35869512192381e-07,
+ "loss": 0.0123,
+ "step": 32034
+ },
+ {
+ "epoch": 87.28882833787466,
+ "grad_norm": 0.5904275178909302,
+ "learning_rate": 8.355163436639236e-07,
+ "loss": 0.0057,
+ "step": 32035
+ },
+ {
+ "epoch": 87.29155313351498,
+ "grad_norm": 1.0901923179626465,
+ "learning_rate": 8.351632465081705e-07,
+ "loss": 0.1771,
+ "step": 32036
+ },
+ {
+ "epoch": 87.29427792915531,
+ "grad_norm": 1.4311213493347168,
+ "learning_rate": 8.348102207278708e-07,
+ "loss": 0.0616,
+ "step": 32037
+ },
+ {
+ "epoch": 87.29700272479565,
+ "grad_norm": 0.6182129979133606,
+ "learning_rate": 8.344572663257755e-07,
+ "loss": 0.0107,
+ "step": 32038
+ },
+ {
+ "epoch": 87.29972752043597,
+ "grad_norm": 0.5768386125564575,
+ "learning_rate": 8.341043833046292e-07,
+ "loss": 0.0064,
+ "step": 32039
+ },
+ {
+ "epoch": 87.3024523160763,
+ "grad_norm": 1.4387633800506592,
+ "learning_rate": 8.337515716671851e-07,
+ "loss": 0.0603,
+ "step": 32040
+ },
+ {
+ "epoch": 87.30517711171662,
+ "grad_norm": 0.6899411678314209,
+ "learning_rate": 8.333988314161879e-07,
+ "loss": 0.0095,
+ "step": 32041
+ },
+ {
+ "epoch": 87.30790190735695,
+ "grad_norm": 0.9664255976676941,
+ "learning_rate": 8.33046162554385e-07,
+ "loss": 0.021,
+ "step": 32042
+ },
+ {
+ "epoch": 87.31062670299727,
+ "grad_norm": 1.145666241645813,
+ "learning_rate": 8.326935650845247e-07,
+ "loss": 0.0382,
+ "step": 32043
+ },
+ {
+ "epoch": 87.3133514986376,
+ "grad_norm": 1.2922641038894653,
+ "learning_rate": 8.323410390093523e-07,
+ "loss": 0.0285,
+ "step": 32044
+ },
+ {
+ "epoch": 87.31607629427793,
+ "grad_norm": 1.389336109161377,
+ "learning_rate": 8.319885843316123e-07,
+ "loss": 0.0148,
+ "step": 32045
+ },
+ {
+ "epoch": 87.31880108991825,
+ "grad_norm": 1.2120839357376099,
+ "learning_rate": 8.316362010540479e-07,
+ "loss": 0.0594,
+ "step": 32046
+ },
+ {
+ "epoch": 87.32152588555859,
+ "grad_norm": 1.5538225173950195,
+ "learning_rate": 8.312838891794073e-07,
+ "loss": 0.0314,
+ "step": 32047
+ },
+ {
+ "epoch": 87.3242506811989,
+ "grad_norm": 2.12937331199646,
+ "learning_rate": 8.309316487104335e-07,
+ "loss": 0.0372,
+ "step": 32048
+ },
+ {
+ "epoch": 87.32697547683924,
+ "grad_norm": 1.4293590784072876,
+ "learning_rate": 8.305794796498689e-07,
+ "loss": 0.0129,
+ "step": 32049
+ },
+ {
+ "epoch": 87.32970027247957,
+ "grad_norm": 2.4253861904144287,
+ "learning_rate": 8.302273820004558e-07,
+ "loss": 0.0389,
+ "step": 32050
+ },
+ {
+ "epoch": 87.33242506811989,
+ "grad_norm": 1.602457880973816,
+ "learning_rate": 8.298753557649353e-07,
+ "loss": 0.0241,
+ "step": 32051
+ },
+ {
+ "epoch": 87.33514986376022,
+ "grad_norm": 0.6233178377151489,
+ "learning_rate": 8.295234009460529e-07,
+ "loss": 0.0057,
+ "step": 32052
+ },
+ {
+ "epoch": 87.33787465940054,
+ "grad_norm": 0.7218292355537415,
+ "learning_rate": 8.291715175465453e-07,
+ "loss": 0.0064,
+ "step": 32053
+ },
+ {
+ "epoch": 87.34059945504087,
+ "grad_norm": 1.9182507991790771,
+ "learning_rate": 8.288197055691561e-07,
+ "loss": 0.0287,
+ "step": 32054
+ },
+ {
+ "epoch": 87.34332425068119,
+ "grad_norm": 1.7296518087387085,
+ "learning_rate": 8.284679650166227e-07,
+ "loss": 0.0344,
+ "step": 32055
+ },
+ {
+ "epoch": 87.34604904632153,
+ "grad_norm": 2.2030909061431885,
+ "learning_rate": 8.281162958916866e-07,
+ "loss": 0.0524,
+ "step": 32056
+ },
+ {
+ "epoch": 87.34877384196186,
+ "grad_norm": 0.7025398015975952,
+ "learning_rate": 8.277646981970855e-07,
+ "loss": 0.074,
+ "step": 32057
+ },
+ {
+ "epoch": 87.35149863760218,
+ "grad_norm": 0.7927071452140808,
+ "learning_rate": 8.274131719355571e-07,
+ "loss": 0.0091,
+ "step": 32058
+ },
+ {
+ "epoch": 87.35422343324251,
+ "grad_norm": 1.6472606658935547,
+ "learning_rate": 8.270617171098383e-07,
+ "loss": 0.0139,
+ "step": 32059
+ },
+ {
+ "epoch": 87.35694822888283,
+ "grad_norm": 0.8431602716445923,
+ "learning_rate": 8.2671033372267e-07,
+ "loss": 0.0143,
+ "step": 32060
+ },
+ {
+ "epoch": 87.35967302452316,
+ "grad_norm": 1.6197830438613892,
+ "learning_rate": 8.263590217767858e-07,
+ "loss": 0.0737,
+ "step": 32061
+ },
+ {
+ "epoch": 87.3623978201635,
+ "grad_norm": 1.3791521787643433,
+ "learning_rate": 8.260077812749212e-07,
+ "loss": 0.0981,
+ "step": 32062
+ },
+ {
+ "epoch": 87.36512261580381,
+ "grad_norm": 1.038432002067566,
+ "learning_rate": 8.25656612219814e-07,
+ "loss": 0.0707,
+ "step": 32063
+ },
+ {
+ "epoch": 87.36784741144415,
+ "grad_norm": 0.9706127047538757,
+ "learning_rate": 8.253055146141964e-07,
+ "loss": 0.0216,
+ "step": 32064
+ },
+ {
+ "epoch": 87.37057220708446,
+ "grad_norm": 1.6339167356491089,
+ "learning_rate": 8.249544884608052e-07,
+ "loss": 0.0813,
+ "step": 32065
+ },
+ {
+ "epoch": 87.3732970027248,
+ "grad_norm": 1.3107435703277588,
+ "learning_rate": 8.246035337623737e-07,
+ "loss": 0.0085,
+ "step": 32066
+ },
+ {
+ "epoch": 87.37602179836512,
+ "grad_norm": 0.7459743618965149,
+ "learning_rate": 8.24252650521633e-07,
+ "loss": 0.0064,
+ "step": 32067
+ },
+ {
+ "epoch": 87.37874659400545,
+ "grad_norm": 0.8802080154418945,
+ "learning_rate": 8.23901838741319e-07,
+ "loss": 0.0046,
+ "step": 32068
+ },
+ {
+ "epoch": 87.38147138964578,
+ "grad_norm": 0.939276933670044,
+ "learning_rate": 8.235510984241623e-07,
+ "loss": 0.0061,
+ "step": 32069
+ },
+ {
+ "epoch": 87.3841961852861,
+ "grad_norm": 2.09948468208313,
+ "learning_rate": 8.232004295728935e-07,
+ "loss": 0.0229,
+ "step": 32070
+ },
+ {
+ "epoch": 87.38692098092643,
+ "grad_norm": 1.2472474575042725,
+ "learning_rate": 8.228498321902434e-07,
+ "loss": 0.1538,
+ "step": 32071
+ },
+ {
+ "epoch": 87.38964577656675,
+ "grad_norm": 1.216773509979248,
+ "learning_rate": 8.224993062789443e-07,
+ "loss": 0.0057,
+ "step": 32072
+ },
+ {
+ "epoch": 87.39237057220708,
+ "grad_norm": 0.8972787261009216,
+ "learning_rate": 8.221488518417264e-07,
+ "loss": 0.009,
+ "step": 32073
+ },
+ {
+ "epoch": 87.39509536784742,
+ "grad_norm": 1.0032449960708618,
+ "learning_rate": 8.217984688813174e-07,
+ "loss": 0.0179,
+ "step": 32074
+ },
+ {
+ "epoch": 87.39782016348774,
+ "grad_norm": 0.7540227174758911,
+ "learning_rate": 8.21448157400444e-07,
+ "loss": 0.0066,
+ "step": 32075
+ },
+ {
+ "epoch": 87.40054495912807,
+ "grad_norm": 0.9412367343902588,
+ "learning_rate": 8.210979174018397e-07,
+ "loss": 0.0114,
+ "step": 32076
+ },
+ {
+ "epoch": 87.40326975476839,
+ "grad_norm": 0.668241024017334,
+ "learning_rate": 8.207477488882287e-07,
+ "loss": 0.004,
+ "step": 32077
+ },
+ {
+ "epoch": 87.40599455040872,
+ "grad_norm": 0.9649755358695984,
+ "learning_rate": 8.203976518623402e-07,
+ "loss": 0.0105,
+ "step": 32078
+ },
+ {
+ "epoch": 87.40871934604904,
+ "grad_norm": 1.3208906650543213,
+ "learning_rate": 8.200476263268964e-07,
+ "loss": 0.057,
+ "step": 32079
+ },
+ {
+ "epoch": 87.41144414168937,
+ "grad_norm": 1.353971242904663,
+ "learning_rate": 8.196976722846273e-07,
+ "loss": 0.0374,
+ "step": 32080
+ },
+ {
+ "epoch": 87.4141689373297,
+ "grad_norm": 1.12264084815979,
+ "learning_rate": 8.193477897382574e-07,
+ "loss": 0.0104,
+ "step": 32081
+ },
+ {
+ "epoch": 87.41689373297002,
+ "grad_norm": 1.4434669017791748,
+ "learning_rate": 8.189979786905111e-07,
+ "loss": 0.0457,
+ "step": 32082
+ },
+ {
+ "epoch": 87.41961852861036,
+ "grad_norm": 1.5692682266235352,
+ "learning_rate": 8.186482391441131e-07,
+ "loss": 0.0203,
+ "step": 32083
+ },
+ {
+ "epoch": 87.42234332425068,
+ "grad_norm": 0.8861784338951111,
+ "learning_rate": 8.182985711017877e-07,
+ "loss": 0.0124,
+ "step": 32084
+ },
+ {
+ "epoch": 87.42506811989101,
+ "grad_norm": 1.1286817789077759,
+ "learning_rate": 8.179489745662583e-07,
+ "loss": 0.0131,
+ "step": 32085
+ },
+ {
+ "epoch": 87.42779291553134,
+ "grad_norm": 1.1912060976028442,
+ "learning_rate": 8.175994495402462e-07,
+ "loss": 0.0169,
+ "step": 32086
+ },
+ {
+ "epoch": 87.43051771117166,
+ "grad_norm": 1.1940748691558838,
+ "learning_rate": 8.172499960264724e-07,
+ "loss": 0.0164,
+ "step": 32087
+ },
+ {
+ "epoch": 87.433242506812,
+ "grad_norm": 0.6790863871574402,
+ "learning_rate": 8.169006140276614e-07,
+ "loss": 0.0061,
+ "step": 32088
+ },
+ {
+ "epoch": 87.43596730245231,
+ "grad_norm": 0.8698200583457947,
+ "learning_rate": 8.165513035465345e-07,
+ "loss": 0.0081,
+ "step": 32089
+ },
+ {
+ "epoch": 87.43869209809264,
+ "grad_norm": 0.8816965818405151,
+ "learning_rate": 8.162020645858093e-07,
+ "loss": 0.0079,
+ "step": 32090
+ },
+ {
+ "epoch": 87.44141689373296,
+ "grad_norm": 1.0560836791992188,
+ "learning_rate": 8.15852897148205e-07,
+ "loss": 0.0099,
+ "step": 32091
+ },
+ {
+ "epoch": 87.4441416893733,
+ "grad_norm": 0.9454711675643921,
+ "learning_rate": 8.155038012364447e-07,
+ "loss": 0.0139,
+ "step": 32092
+ },
+ {
+ "epoch": 87.44686648501363,
+ "grad_norm": 0.9691860675811768,
+ "learning_rate": 8.151547768532453e-07,
+ "loss": 0.0083,
+ "step": 32093
+ },
+ {
+ "epoch": 87.44959128065395,
+ "grad_norm": 0.5820713043212891,
+ "learning_rate": 8.148058240013246e-07,
+ "loss": 0.0077,
+ "step": 32094
+ },
+ {
+ "epoch": 87.45231607629428,
+ "grad_norm": 1.6098521947860718,
+ "learning_rate": 8.144569426833981e-07,
+ "loss": 0.0159,
+ "step": 32095
+ },
+ {
+ "epoch": 87.4550408719346,
+ "grad_norm": 1.217653512954712,
+ "learning_rate": 8.141081329021883e-07,
+ "loss": 0.0157,
+ "step": 32096
+ },
+ {
+ "epoch": 87.45776566757493,
+ "grad_norm": 0.4197142422199249,
+ "learning_rate": 8.137593946604083e-07,
+ "loss": 0.0045,
+ "step": 32097
+ },
+ {
+ "epoch": 87.46049046321527,
+ "grad_norm": 0.736491858959198,
+ "learning_rate": 8.134107279607728e-07,
+ "loss": 0.0086,
+ "step": 32098
+ },
+ {
+ "epoch": 87.46321525885558,
+ "grad_norm": 1.0904163122177124,
+ "learning_rate": 8.130621328060006e-07,
+ "loss": 0.0183,
+ "step": 32099
+ },
+ {
+ "epoch": 87.46594005449592,
+ "grad_norm": 0.8125515580177307,
+ "learning_rate": 8.127136091988031e-07,
+ "loss": 0.0101,
+ "step": 32100
+ },
+ {
+ "epoch": 87.46866485013624,
+ "grad_norm": 0.7140673995018005,
+ "learning_rate": 8.123651571418978e-07,
+ "loss": 0.0075,
+ "step": 32101
+ },
+ {
+ "epoch": 87.47138964577657,
+ "grad_norm": 1.9722809791564941,
+ "learning_rate": 8.120167766379972e-07,
+ "loss": 0.0101,
+ "step": 32102
+ },
+ {
+ "epoch": 87.47411444141689,
+ "grad_norm": 1.2279298305511475,
+ "learning_rate": 8.116684676898134e-07,
+ "loss": 0.0214,
+ "step": 32103
+ },
+ {
+ "epoch": 87.47683923705722,
+ "grad_norm": 0.8558847308158875,
+ "learning_rate": 8.113202303000589e-07,
+ "loss": 0.0056,
+ "step": 32104
+ },
+ {
+ "epoch": 87.47956403269755,
+ "grad_norm": 0.5442972183227539,
+ "learning_rate": 8.109720644714492e-07,
+ "loss": 0.0039,
+ "step": 32105
+ },
+ {
+ "epoch": 87.48228882833787,
+ "grad_norm": 1.0584895610809326,
+ "learning_rate": 8.106239702066921e-07,
+ "loss": 0.0163,
+ "step": 32106
+ },
+ {
+ "epoch": 87.4850136239782,
+ "grad_norm": 1.3764569759368896,
+ "learning_rate": 8.102759475084986e-07,
+ "loss": 0.0129,
+ "step": 32107
+ },
+ {
+ "epoch": 87.48773841961852,
+ "grad_norm": 1.4816011190414429,
+ "learning_rate": 8.099279963795814e-07,
+ "loss": 0.0099,
+ "step": 32108
+ },
+ {
+ "epoch": 87.49046321525886,
+ "grad_norm": 1.069483995437622,
+ "learning_rate": 8.095801168226503e-07,
+ "loss": 0.0211,
+ "step": 32109
+ },
+ {
+ "epoch": 87.49318801089919,
+ "grad_norm": 0.8815926909446716,
+ "learning_rate": 8.092323088404119e-07,
+ "loss": 0.008,
+ "step": 32110
+ },
+ {
+ "epoch": 87.49591280653951,
+ "grad_norm": 3.3414628505706787,
+ "learning_rate": 8.088845724355765e-07,
+ "loss": 0.0161,
+ "step": 32111
+ },
+ {
+ "epoch": 87.49863760217984,
+ "grad_norm": 0.8070042133331299,
+ "learning_rate": 8.085369076108529e-07,
+ "loss": 0.0105,
+ "step": 32112
+ },
+ {
+ "epoch": 87.50136239782016,
+ "grad_norm": 1.1761560440063477,
+ "learning_rate": 8.081893143689479e-07,
+ "loss": 0.0085,
+ "step": 32113
+ },
+ {
+ "epoch": 87.50408719346049,
+ "grad_norm": 0.4878067076206207,
+ "learning_rate": 8.078417927125692e-07,
+ "loss": 0.0076,
+ "step": 32114
+ },
+ {
+ "epoch": 87.50681198910081,
+ "grad_norm": 1.3532030582427979,
+ "learning_rate": 8.074943426444204e-07,
+ "loss": 0.0185,
+ "step": 32115
+ },
+ {
+ "epoch": 87.50953678474114,
+ "grad_norm": 3.4627585411071777,
+ "learning_rate": 8.071469641672103e-07,
+ "loss": 0.0399,
+ "step": 32116
+ },
+ {
+ "epoch": 87.51226158038148,
+ "grad_norm": 1.2762348651885986,
+ "learning_rate": 8.067996572836456e-07,
+ "loss": 0.0213,
+ "step": 32117
+ },
+ {
+ "epoch": 87.5149863760218,
+ "grad_norm": 0.7896509766578674,
+ "learning_rate": 8.064524219964298e-07,
+ "loss": 0.0095,
+ "step": 32118
+ },
+ {
+ "epoch": 87.51771117166213,
+ "grad_norm": 2.0790493488311768,
+ "learning_rate": 8.061052583082652e-07,
+ "loss": 0.0103,
+ "step": 32119
+ },
+ {
+ "epoch": 87.52043596730245,
+ "grad_norm": 1.3665964603424072,
+ "learning_rate": 8.057581662218561e-07,
+ "loss": 0.0216,
+ "step": 32120
+ },
+ {
+ "epoch": 87.52316076294278,
+ "grad_norm": 0.9510179162025452,
+ "learning_rate": 8.054111457399083e-07,
+ "loss": 0.0112,
+ "step": 32121
+ },
+ {
+ "epoch": 87.52588555858311,
+ "grad_norm": 0.7162869572639465,
+ "learning_rate": 8.050641968651218e-07,
+ "loss": 0.0103,
+ "step": 32122
+ },
+ {
+ "epoch": 87.52861035422343,
+ "grad_norm": 0.6479834318161011,
+ "learning_rate": 8.047173196002001e-07,
+ "loss": 0.0072,
+ "step": 32123
+ },
+ {
+ "epoch": 87.53133514986376,
+ "grad_norm": 1.0495021343231201,
+ "learning_rate": 8.043705139478431e-07,
+ "loss": 0.0093,
+ "step": 32124
+ },
+ {
+ "epoch": 87.53405994550408,
+ "grad_norm": 1.3376322984695435,
+ "learning_rate": 8.040237799107531e-07,
+ "loss": 0.0104,
+ "step": 32125
+ },
+ {
+ "epoch": 87.53678474114442,
+ "grad_norm": 1.5171504020690918,
+ "learning_rate": 8.036771174916302e-07,
+ "loss": 0.0738,
+ "step": 32126
+ },
+ {
+ "epoch": 87.53950953678473,
+ "grad_norm": 4.3180060386657715,
+ "learning_rate": 8.033305266931712e-07,
+ "loss": 0.022,
+ "step": 32127
+ },
+ {
+ "epoch": 87.54223433242507,
+ "grad_norm": 1.6409718990325928,
+ "learning_rate": 8.029840075180817e-07,
+ "loss": 0.0391,
+ "step": 32128
+ },
+ {
+ "epoch": 87.5449591280654,
+ "grad_norm": 1.1160104274749756,
+ "learning_rate": 8.026375599690551e-07,
+ "loss": 0.016,
+ "step": 32129
+ },
+ {
+ "epoch": 87.54768392370572,
+ "grad_norm": 1.2582231760025024,
+ "learning_rate": 8.022911840487924e-07,
+ "loss": 0.0087,
+ "step": 32130
+ },
+ {
+ "epoch": 87.55040871934605,
+ "grad_norm": 1.237484097480774,
+ "learning_rate": 8.019448797599871e-07,
+ "loss": 0.0147,
+ "step": 32131
+ },
+ {
+ "epoch": 87.55313351498637,
+ "grad_norm": 0.7724403142929077,
+ "learning_rate": 8.015986471053394e-07,
+ "loss": 0.008,
+ "step": 32132
+ },
+ {
+ "epoch": 87.5558583106267,
+ "grad_norm": 1.2741060256958008,
+ "learning_rate": 8.012524860875481e-07,
+ "loss": 0.0136,
+ "step": 32133
+ },
+ {
+ "epoch": 87.55858310626704,
+ "grad_norm": 0.8016225099563599,
+ "learning_rate": 8.009063967093055e-07,
+ "loss": 0.0086,
+ "step": 32134
+ },
+ {
+ "epoch": 87.56130790190736,
+ "grad_norm": 1.6211912631988525,
+ "learning_rate": 8.005603789733074e-07,
+ "loss": 0.0143,
+ "step": 32135
+ },
+ {
+ "epoch": 87.56403269754769,
+ "grad_norm": 0.9683616161346436,
+ "learning_rate": 8.00214432882248e-07,
+ "loss": 0.0307,
+ "step": 32136
+ },
+ {
+ "epoch": 87.566757493188,
+ "grad_norm": 1.0918328762054443,
+ "learning_rate": 7.998685584388244e-07,
+ "loss": 0.011,
+ "step": 32137
+ },
+ {
+ "epoch": 87.56948228882834,
+ "grad_norm": 1.8252322673797607,
+ "learning_rate": 7.995227556457274e-07,
+ "loss": 0.0706,
+ "step": 32138
+ },
+ {
+ "epoch": 87.57220708446866,
+ "grad_norm": 0.7485754489898682,
+ "learning_rate": 7.991770245056519e-07,
+ "loss": 0.0085,
+ "step": 32139
+ },
+ {
+ "epoch": 87.57493188010899,
+ "grad_norm": 0.7632095813751221,
+ "learning_rate": 7.988313650212876e-07,
+ "loss": 0.0101,
+ "step": 32140
+ },
+ {
+ "epoch": 87.57765667574932,
+ "grad_norm": 1.171895980834961,
+ "learning_rate": 7.984857771953303e-07,
+ "loss": 0.0304,
+ "step": 32141
+ },
+ {
+ "epoch": 87.58038147138964,
+ "grad_norm": 0.9695442914962769,
+ "learning_rate": 7.981402610304689e-07,
+ "loss": 0.014,
+ "step": 32142
+ },
+ {
+ "epoch": 87.58310626702998,
+ "grad_norm": 1.4247373342514038,
+ "learning_rate": 7.977948165293959e-07,
+ "loss": 0.0057,
+ "step": 32143
+ },
+ {
+ "epoch": 87.5858310626703,
+ "grad_norm": 1.1452957391738892,
+ "learning_rate": 7.974494436947989e-07,
+ "loss": 0.087,
+ "step": 32144
+ },
+ {
+ "epoch": 87.58855585831063,
+ "grad_norm": 2.292343854904175,
+ "learning_rate": 7.971041425293713e-07,
+ "loss": 0.0388,
+ "step": 32145
+ },
+ {
+ "epoch": 87.59128065395096,
+ "grad_norm": 0.6404055953025818,
+ "learning_rate": 7.967589130358e-07,
+ "loss": 0.0079,
+ "step": 32146
+ },
+ {
+ "epoch": 87.59400544959128,
+ "grad_norm": 1.1179028749465942,
+ "learning_rate": 7.964137552167739e-07,
+ "loss": 0.0167,
+ "step": 32147
+ },
+ {
+ "epoch": 87.59673024523161,
+ "grad_norm": 0.4309660494327545,
+ "learning_rate": 7.960686690749797e-07,
+ "loss": 0.0051,
+ "step": 32148
+ },
+ {
+ "epoch": 87.59945504087193,
+ "grad_norm": 0.594758152961731,
+ "learning_rate": 7.957236546131086e-07,
+ "loss": 0.0055,
+ "step": 32149
+ },
+ {
+ "epoch": 87.60217983651226,
+ "grad_norm": 0.5982861518859863,
+ "learning_rate": 7.95378711833843e-07,
+ "loss": 0.0098,
+ "step": 32150
+ },
+ {
+ "epoch": 87.60490463215258,
+ "grad_norm": 1.4403671026229858,
+ "learning_rate": 7.950338407398739e-07,
+ "loss": 0.032,
+ "step": 32151
+ },
+ {
+ "epoch": 87.60762942779292,
+ "grad_norm": 1.0718027353286743,
+ "learning_rate": 7.946890413338837e-07,
+ "loss": 0.1389,
+ "step": 32152
+ },
+ {
+ "epoch": 87.61035422343325,
+ "grad_norm": 1.8121916055679321,
+ "learning_rate": 7.943443136185602e-07,
+ "loss": 0.0132,
+ "step": 32153
+ },
+ {
+ "epoch": 87.61307901907357,
+ "grad_norm": 1.3891035318374634,
+ "learning_rate": 7.93999657596587e-07,
+ "loss": 0.008,
+ "step": 32154
+ },
+ {
+ "epoch": 87.6158038147139,
+ "grad_norm": 1.3965266942977905,
+ "learning_rate": 7.936550732706483e-07,
+ "loss": 0.0483,
+ "step": 32155
+ },
+ {
+ "epoch": 87.61852861035422,
+ "grad_norm": 1.4146392345428467,
+ "learning_rate": 7.933105606434255e-07,
+ "loss": 0.0401,
+ "step": 32156
+ },
+ {
+ "epoch": 87.62125340599455,
+ "grad_norm": 1.1010395288467407,
+ "learning_rate": 7.929661197176052e-07,
+ "loss": 0.0387,
+ "step": 32157
+ },
+ {
+ "epoch": 87.62397820163488,
+ "grad_norm": 1.307008981704712,
+ "learning_rate": 7.926217504958688e-07,
+ "loss": 0.0704,
+ "step": 32158
+ },
+ {
+ "epoch": 87.6267029972752,
+ "grad_norm": 1.1203536987304688,
+ "learning_rate": 7.922774529808985e-07,
+ "loss": 0.1142,
+ "step": 32159
+ },
+ {
+ "epoch": 87.62942779291554,
+ "grad_norm": 1.3178730010986328,
+ "learning_rate": 7.919332271753722e-07,
+ "loss": 0.0757,
+ "step": 32160
+ },
+ {
+ "epoch": 87.63215258855585,
+ "grad_norm": 1.167089581489563,
+ "learning_rate": 7.915890730819764e-07,
+ "loss": 0.0121,
+ "step": 32161
+ },
+ {
+ "epoch": 87.63487738419619,
+ "grad_norm": 0.536034345626831,
+ "learning_rate": 7.912449907033881e-07,
+ "loss": 0.0083,
+ "step": 32162
+ },
+ {
+ "epoch": 87.6376021798365,
+ "grad_norm": 1.202487587928772,
+ "learning_rate": 7.909009800422874e-07,
+ "loss": 0.0066,
+ "step": 32163
+ },
+ {
+ "epoch": 87.64032697547684,
+ "grad_norm": 0.8351073265075684,
+ "learning_rate": 7.905570411013519e-07,
+ "loss": 0.0074,
+ "step": 32164
+ },
+ {
+ "epoch": 87.64305177111717,
+ "grad_norm": 1.2280465364456177,
+ "learning_rate": 7.90213173883263e-07,
+ "loss": 0.0128,
+ "step": 32165
+ },
+ {
+ "epoch": 87.64577656675749,
+ "grad_norm": 1.738390326499939,
+ "learning_rate": 7.898693783906975e-07,
+ "loss": 0.016,
+ "step": 32166
+ },
+ {
+ "epoch": 87.64850136239782,
+ "grad_norm": 1.0352765321731567,
+ "learning_rate": 7.895256546263319e-07,
+ "loss": 0.0186,
+ "step": 32167
+ },
+ {
+ "epoch": 87.65122615803814,
+ "grad_norm": 0.9183186888694763,
+ "learning_rate": 7.891820025928432e-07,
+ "loss": 0.012,
+ "step": 32168
+ },
+ {
+ "epoch": 87.65395095367847,
+ "grad_norm": 1.4173932075500488,
+ "learning_rate": 7.888384222929101e-07,
+ "loss": 0.1127,
+ "step": 32169
+ },
+ {
+ "epoch": 87.65667574931881,
+ "grad_norm": 0.8623688220977783,
+ "learning_rate": 7.884949137292075e-07,
+ "loss": 0.0162,
+ "step": 32170
+ },
+ {
+ "epoch": 87.65940054495913,
+ "grad_norm": 1.7067315578460693,
+ "learning_rate": 7.881514769044107e-07,
+ "loss": 0.0777,
+ "step": 32171
+ },
+ {
+ "epoch": 87.66212534059946,
+ "grad_norm": 0.8345460295677185,
+ "learning_rate": 7.878081118211911e-07,
+ "loss": 0.0087,
+ "step": 32172
+ },
+ {
+ "epoch": 87.66485013623978,
+ "grad_norm": 0.7641470432281494,
+ "learning_rate": 7.874648184822275e-07,
+ "loss": 0.016,
+ "step": 32173
+ },
+ {
+ "epoch": 87.66757493188011,
+ "grad_norm": 0.42884400486946106,
+ "learning_rate": 7.871215968901913e-07,
+ "loss": 0.0036,
+ "step": 32174
+ },
+ {
+ "epoch": 87.67029972752043,
+ "grad_norm": 1.4422305822372437,
+ "learning_rate": 7.867784470477557e-07,
+ "loss": 0.0667,
+ "step": 32175
+ },
+ {
+ "epoch": 87.67302452316076,
+ "grad_norm": 0.519700288772583,
+ "learning_rate": 7.864353689575909e-07,
+ "loss": 0.0049,
+ "step": 32176
+ },
+ {
+ "epoch": 87.6757493188011,
+ "grad_norm": 1.0632493495941162,
+ "learning_rate": 7.860923626223726e-07,
+ "loss": 0.0159,
+ "step": 32177
+ },
+ {
+ "epoch": 87.67847411444141,
+ "grad_norm": 1.430753231048584,
+ "learning_rate": 7.857494280447709e-07,
+ "loss": 0.1151,
+ "step": 32178
+ },
+ {
+ "epoch": 87.68119891008175,
+ "grad_norm": 1.7531042098999023,
+ "learning_rate": 7.854065652274556e-07,
+ "loss": 0.0193,
+ "step": 32179
+ },
+ {
+ "epoch": 87.68392370572207,
+ "grad_norm": 1.1061627864837646,
+ "learning_rate": 7.850637741730959e-07,
+ "loss": 0.0596,
+ "step": 32180
+ },
+ {
+ "epoch": 87.6866485013624,
+ "grad_norm": 0.94151371717453,
+ "learning_rate": 7.847210548843653e-07,
+ "loss": 0.0163,
+ "step": 32181
+ },
+ {
+ "epoch": 87.68937329700273,
+ "grad_norm": 1.9788652658462524,
+ "learning_rate": 7.843784073639304e-07,
+ "loss": 0.0271,
+ "step": 32182
+ },
+ {
+ "epoch": 87.69209809264305,
+ "grad_norm": 1.4276583194732666,
+ "learning_rate": 7.840358316144592e-07,
+ "loss": 0.0063,
+ "step": 32183
+ },
+ {
+ "epoch": 87.69482288828338,
+ "grad_norm": 1.7247270345687866,
+ "learning_rate": 7.836933276386183e-07,
+ "loss": 0.0748,
+ "step": 32184
+ },
+ {
+ "epoch": 87.6975476839237,
+ "grad_norm": 0.6716213226318359,
+ "learning_rate": 7.833508954390778e-07,
+ "loss": 0.0063,
+ "step": 32185
+ },
+ {
+ "epoch": 87.70027247956403,
+ "grad_norm": 1.0905117988586426,
+ "learning_rate": 7.830085350185057e-07,
+ "loss": 0.0352,
+ "step": 32186
+ },
+ {
+ "epoch": 87.70299727520435,
+ "grad_norm": 1.293830394744873,
+ "learning_rate": 7.826662463795665e-07,
+ "loss": 0.1148,
+ "step": 32187
+ },
+ {
+ "epoch": 87.70572207084469,
+ "grad_norm": 0.8074221611022949,
+ "learning_rate": 7.823240295249257e-07,
+ "loss": 0.0106,
+ "step": 32188
+ },
+ {
+ "epoch": 87.70844686648502,
+ "grad_norm": 1.19817054271698,
+ "learning_rate": 7.819818844572468e-07,
+ "loss": 0.0206,
+ "step": 32189
+ },
+ {
+ "epoch": 87.71117166212534,
+ "grad_norm": 1.2344199419021606,
+ "learning_rate": 7.81639811179199e-07,
+ "loss": 0.0143,
+ "step": 32190
+ },
+ {
+ "epoch": 87.71389645776567,
+ "grad_norm": 1.1137604713439941,
+ "learning_rate": 7.812978096934421e-07,
+ "loss": 0.0739,
+ "step": 32191
+ },
+ {
+ "epoch": 87.71662125340599,
+ "grad_norm": 1.4494208097457886,
+ "learning_rate": 7.809558800026407e-07,
+ "loss": 0.0397,
+ "step": 32192
+ },
+ {
+ "epoch": 87.71934604904632,
+ "grad_norm": 1.4090254306793213,
+ "learning_rate": 7.806140221094593e-07,
+ "loss": 0.008,
+ "step": 32193
+ },
+ {
+ "epoch": 87.72207084468666,
+ "grad_norm": 0.7060765624046326,
+ "learning_rate": 7.802722360165593e-07,
+ "loss": 0.0117,
+ "step": 32194
+ },
+ {
+ "epoch": 87.72479564032697,
+ "grad_norm": 0.8848668336868286,
+ "learning_rate": 7.799305217266018e-07,
+ "loss": 0.0081,
+ "step": 32195
+ },
+ {
+ "epoch": 87.7275204359673,
+ "grad_norm": 1.0197608470916748,
+ "learning_rate": 7.795888792422479e-07,
+ "loss": 0.018,
+ "step": 32196
+ },
+ {
+ "epoch": 87.73024523160763,
+ "grad_norm": 0.7803107500076294,
+ "learning_rate": 7.792473085661601e-07,
+ "loss": 0.0128,
+ "step": 32197
+ },
+ {
+ "epoch": 87.73297002724796,
+ "grad_norm": 0.6989822387695312,
+ "learning_rate": 7.789058097009983e-07,
+ "loss": 0.0071,
+ "step": 32198
+ },
+ {
+ "epoch": 87.73569482288828,
+ "grad_norm": 0.6510767936706543,
+ "learning_rate": 7.785643826494205e-07,
+ "loss": 0.0081,
+ "step": 32199
+ },
+ {
+ "epoch": 87.73841961852861,
+ "grad_norm": 1.3960398435592651,
+ "learning_rate": 7.782230274140845e-07,
+ "loss": 0.0341,
+ "step": 32200
+ },
+ {
+ "epoch": 87.74114441416894,
+ "grad_norm": 1.839763879776001,
+ "learning_rate": 7.778817439976515e-07,
+ "loss": 0.0183,
+ "step": 32201
+ },
+ {
+ "epoch": 87.74386920980926,
+ "grad_norm": 1.1008251905441284,
+ "learning_rate": 7.775405324027807e-07,
+ "loss": 0.0061,
+ "step": 32202
+ },
+ {
+ "epoch": 87.7465940054496,
+ "grad_norm": 0.7233414649963379,
+ "learning_rate": 7.771993926321264e-07,
+ "loss": 0.007,
+ "step": 32203
+ },
+ {
+ "epoch": 87.74931880108991,
+ "grad_norm": 1.1401537656784058,
+ "learning_rate": 7.768583246883476e-07,
+ "loss": 0.0867,
+ "step": 32204
+ },
+ {
+ "epoch": 87.75204359673025,
+ "grad_norm": 0.9629846811294556,
+ "learning_rate": 7.765173285740968e-07,
+ "loss": 0.0083,
+ "step": 32205
+ },
+ {
+ "epoch": 87.75476839237058,
+ "grad_norm": 0.8468570709228516,
+ "learning_rate": 7.761764042920339e-07,
+ "loss": 0.0086,
+ "step": 32206
+ },
+ {
+ "epoch": 87.7574931880109,
+ "grad_norm": 1.4239524602890015,
+ "learning_rate": 7.758355518448124e-07,
+ "loss": 0.0304,
+ "step": 32207
+ },
+ {
+ "epoch": 87.76021798365123,
+ "grad_norm": 1.8680729866027832,
+ "learning_rate": 7.754947712350869e-07,
+ "loss": 0.0589,
+ "step": 32208
+ },
+ {
+ "epoch": 87.76294277929155,
+ "grad_norm": 0.7368519306182861,
+ "learning_rate": 7.751540624655085e-07,
+ "loss": 0.0062,
+ "step": 32209
+ },
+ {
+ "epoch": 87.76566757493188,
+ "grad_norm": 1.0460662841796875,
+ "learning_rate": 7.748134255387363e-07,
+ "loss": 0.0204,
+ "step": 32210
+ },
+ {
+ "epoch": 87.7683923705722,
+ "grad_norm": 0.7418779730796814,
+ "learning_rate": 7.744728604574203e-07,
+ "loss": 0.0079,
+ "step": 32211
+ },
+ {
+ "epoch": 87.77111716621253,
+ "grad_norm": 1.6284927129745483,
+ "learning_rate": 7.741323672242118e-07,
+ "loss": 0.0755,
+ "step": 32212
+ },
+ {
+ "epoch": 87.77384196185287,
+ "grad_norm": 1.5841522216796875,
+ "learning_rate": 7.73791945841762e-07,
+ "loss": 0.0825,
+ "step": 32213
+ },
+ {
+ "epoch": 87.77656675749319,
+ "grad_norm": 1.3508226871490479,
+ "learning_rate": 7.734515963127254e-07,
+ "loss": 0.0162,
+ "step": 32214
+ },
+ {
+ "epoch": 87.77929155313352,
+ "grad_norm": 0.84182208776474,
+ "learning_rate": 7.73111318639751e-07,
+ "loss": 0.0139,
+ "step": 32215
+ },
+ {
+ "epoch": 87.78201634877384,
+ "grad_norm": 1.0916779041290283,
+ "learning_rate": 7.727711128254878e-07,
+ "loss": 0.0221,
+ "step": 32216
+ },
+ {
+ "epoch": 87.78474114441417,
+ "grad_norm": 1.6537209749221802,
+ "learning_rate": 7.72430978872587e-07,
+ "loss": 0.02,
+ "step": 32217
+ },
+ {
+ "epoch": 87.7874659400545,
+ "grad_norm": 0.5724300742149353,
+ "learning_rate": 7.720909167836943e-07,
+ "loss": 0.0048,
+ "step": 32218
+ },
+ {
+ "epoch": 87.79019073569482,
+ "grad_norm": 1.2912054061889648,
+ "learning_rate": 7.717509265614631e-07,
+ "loss": 0.0102,
+ "step": 32219
+ },
+ {
+ "epoch": 87.79291553133515,
+ "grad_norm": 1.5868207216262817,
+ "learning_rate": 7.714110082085391e-07,
+ "loss": 0.0306,
+ "step": 32220
+ },
+ {
+ "epoch": 87.79564032697547,
+ "grad_norm": 1.852681040763855,
+ "learning_rate": 7.710711617275668e-07,
+ "loss": 0.0629,
+ "step": 32221
+ },
+ {
+ "epoch": 87.7983651226158,
+ "grad_norm": 1.053434133529663,
+ "learning_rate": 7.707313871211974e-07,
+ "loss": 0.0567,
+ "step": 32222
+ },
+ {
+ "epoch": 87.80108991825612,
+ "grad_norm": 1.0068387985229492,
+ "learning_rate": 7.703916843920745e-07,
+ "loss": 0.0119,
+ "step": 32223
+ },
+ {
+ "epoch": 87.80381471389646,
+ "grad_norm": 0.9954105019569397,
+ "learning_rate": 7.700520535428435e-07,
+ "loss": 0.009,
+ "step": 32224
+ },
+ {
+ "epoch": 87.80653950953679,
+ "grad_norm": 1.6694270372390747,
+ "learning_rate": 7.697124945761492e-07,
+ "loss": 0.0539,
+ "step": 32225
+ },
+ {
+ "epoch": 87.80926430517711,
+ "grad_norm": 1.3102492094039917,
+ "learning_rate": 7.693730074946393e-07,
+ "loss": 0.0792,
+ "step": 32226
+ },
+ {
+ "epoch": 87.81198910081744,
+ "grad_norm": 0.8872539401054382,
+ "learning_rate": 7.690335923009539e-07,
+ "loss": 0.008,
+ "step": 32227
+ },
+ {
+ "epoch": 87.81471389645776,
+ "grad_norm": 0.9785916209220886,
+ "learning_rate": 7.686942489977389e-07,
+ "loss": 0.0281,
+ "step": 32228
+ },
+ {
+ "epoch": 87.8174386920981,
+ "grad_norm": 0.6210910677909851,
+ "learning_rate": 7.683549775876331e-07,
+ "loss": 0.0063,
+ "step": 32229
+ },
+ {
+ "epoch": 87.82016348773843,
+ "grad_norm": 1.0942493677139282,
+ "learning_rate": 7.680157780732844e-07,
+ "loss": 0.0067,
+ "step": 32230
+ },
+ {
+ "epoch": 87.82288828337875,
+ "grad_norm": 1.0064902305603027,
+ "learning_rate": 7.676766504573319e-07,
+ "loss": 0.0095,
+ "step": 32231
+ },
+ {
+ "epoch": 87.82561307901908,
+ "grad_norm": 0.9102238416671753,
+ "learning_rate": 7.673375947424155e-07,
+ "loss": 0.0102,
+ "step": 32232
+ },
+ {
+ "epoch": 87.8283378746594,
+ "grad_norm": 1.5147305727005005,
+ "learning_rate": 7.669986109311756e-07,
+ "loss": 0.0823,
+ "step": 32233
+ },
+ {
+ "epoch": 87.83106267029973,
+ "grad_norm": 1.752812147140503,
+ "learning_rate": 7.666596990262542e-07,
+ "loss": 0.1067,
+ "step": 32234
+ },
+ {
+ "epoch": 87.83378746594005,
+ "grad_norm": 1.1500290632247925,
+ "learning_rate": 7.663208590302907e-07,
+ "loss": 0.013,
+ "step": 32235
+ },
+ {
+ "epoch": 87.83651226158038,
+ "grad_norm": 0.8300559520721436,
+ "learning_rate": 7.659820909459215e-07,
+ "loss": 0.0068,
+ "step": 32236
+ },
+ {
+ "epoch": 87.83923705722071,
+ "grad_norm": 1.9470148086547852,
+ "learning_rate": 7.656433947757869e-07,
+ "loss": 0.0876,
+ "step": 32237
+ },
+ {
+ "epoch": 87.84196185286103,
+ "grad_norm": 1.1169192790985107,
+ "learning_rate": 7.653047705225258e-07,
+ "loss": 0.0144,
+ "step": 32238
+ },
+ {
+ "epoch": 87.84468664850137,
+ "grad_norm": 0.7491866946220398,
+ "learning_rate": 7.649662181887729e-07,
+ "loss": 0.0088,
+ "step": 32239
+ },
+ {
+ "epoch": 87.84741144414168,
+ "grad_norm": 1.0520657300949097,
+ "learning_rate": 7.646277377771671e-07,
+ "loss": 0.0216,
+ "step": 32240
+ },
+ {
+ "epoch": 87.85013623978202,
+ "grad_norm": 0.8389083743095398,
+ "learning_rate": 7.642893292903408e-07,
+ "loss": 0.0087,
+ "step": 32241
+ },
+ {
+ "epoch": 87.85286103542235,
+ "grad_norm": 0.3658880889415741,
+ "learning_rate": 7.639509927309341e-07,
+ "loss": 0.0039,
+ "step": 32242
+ },
+ {
+ "epoch": 87.85558583106267,
+ "grad_norm": 0.8242413997650146,
+ "learning_rate": 7.636127281015804e-07,
+ "loss": 0.0106,
+ "step": 32243
+ },
+ {
+ "epoch": 87.858310626703,
+ "grad_norm": 0.8247904181480408,
+ "learning_rate": 7.63274535404912e-07,
+ "loss": 0.0101,
+ "step": 32244
+ },
+ {
+ "epoch": 87.86103542234332,
+ "grad_norm": 0.7819012403488159,
+ "learning_rate": 7.629364146435636e-07,
+ "loss": 0.0067,
+ "step": 32245
+ },
+ {
+ "epoch": 87.86376021798365,
+ "grad_norm": 0.8528632521629333,
+ "learning_rate": 7.625983658201708e-07,
+ "loss": 0.0063,
+ "step": 32246
+ },
+ {
+ "epoch": 87.86648501362397,
+ "grad_norm": 0.9586229920387268,
+ "learning_rate": 7.622603889373648e-07,
+ "loss": 0.0163,
+ "step": 32247
+ },
+ {
+ "epoch": 87.8692098092643,
+ "grad_norm": 1.0154750347137451,
+ "learning_rate": 7.619224839977767e-07,
+ "loss": 0.0486,
+ "step": 32248
+ },
+ {
+ "epoch": 87.87193460490464,
+ "grad_norm": 0.6868213415145874,
+ "learning_rate": 7.615846510040381e-07,
+ "loss": 0.0042,
+ "step": 32249
+ },
+ {
+ "epoch": 87.87465940054496,
+ "grad_norm": 8.729660987854004,
+ "learning_rate": 7.612468899587822e-07,
+ "loss": 0.0762,
+ "step": 32250
+ },
+ {
+ "epoch": 87.87738419618529,
+ "grad_norm": 0.7769232988357544,
+ "learning_rate": 7.609092008646379e-07,
+ "loss": 0.0243,
+ "step": 32251
+ },
+ {
+ "epoch": 87.88010899182561,
+ "grad_norm": 0.7368071675300598,
+ "learning_rate": 7.605715837242355e-07,
+ "loss": 0.0116,
+ "step": 32252
+ },
+ {
+ "epoch": 87.88283378746594,
+ "grad_norm": 1.4869650602340698,
+ "learning_rate": 7.60234038540203e-07,
+ "loss": 0.0222,
+ "step": 32253
+ },
+ {
+ "epoch": 87.88555858310627,
+ "grad_norm": 1.2598609924316406,
+ "learning_rate": 7.598965653151702e-07,
+ "loss": 0.0225,
+ "step": 32254
+ },
+ {
+ "epoch": 87.88828337874659,
+ "grad_norm": 0.8570705652236938,
+ "learning_rate": 7.595591640517663e-07,
+ "loss": 0.012,
+ "step": 32255
+ },
+ {
+ "epoch": 87.89100817438693,
+ "grad_norm": 1.3913484811782837,
+ "learning_rate": 7.592218347526192e-07,
+ "loss": 0.0389,
+ "step": 32256
+ },
+ {
+ "epoch": 87.89373297002724,
+ "grad_norm": 0.9564818739891052,
+ "learning_rate": 7.588845774203535e-07,
+ "loss": 0.0314,
+ "step": 32257
+ },
+ {
+ "epoch": 87.89645776566758,
+ "grad_norm": 1.337197184562683,
+ "learning_rate": 7.585473920575992e-07,
+ "loss": 0.2231,
+ "step": 32258
+ },
+ {
+ "epoch": 87.8991825613079,
+ "grad_norm": 0.9515216946601868,
+ "learning_rate": 7.582102786669798e-07,
+ "loss": 0.0101,
+ "step": 32259
+ },
+ {
+ "epoch": 87.90190735694823,
+ "grad_norm": 1.3905712366104126,
+ "learning_rate": 7.57873237251121e-07,
+ "loss": 0.0306,
+ "step": 32260
+ },
+ {
+ "epoch": 87.90463215258856,
+ "grad_norm": 1.0962551832199097,
+ "learning_rate": 7.575362678126474e-07,
+ "loss": 0.0288,
+ "step": 32261
+ },
+ {
+ "epoch": 87.90735694822888,
+ "grad_norm": 1.1749762296676636,
+ "learning_rate": 7.571993703541847e-07,
+ "loss": 0.0079,
+ "step": 32262
+ },
+ {
+ "epoch": 87.91008174386921,
+ "grad_norm": 1.4922362565994263,
+ "learning_rate": 7.568625448783562e-07,
+ "loss": 0.0269,
+ "step": 32263
+ },
+ {
+ "epoch": 87.91280653950953,
+ "grad_norm": 2.719937324523926,
+ "learning_rate": 7.565257913877843e-07,
+ "loss": 0.0226,
+ "step": 32264
+ },
+ {
+ "epoch": 87.91553133514986,
+ "grad_norm": 0.7428253889083862,
+ "learning_rate": 7.561891098850904e-07,
+ "loss": 0.0052,
+ "step": 32265
+ },
+ {
+ "epoch": 87.9182561307902,
+ "grad_norm": 0.9162585139274597,
+ "learning_rate": 7.558525003729e-07,
+ "loss": 0.0091,
+ "step": 32266
+ },
+ {
+ "epoch": 87.92098092643052,
+ "grad_norm": 1.493939995765686,
+ "learning_rate": 7.555159628538322e-07,
+ "loss": 0.0326,
+ "step": 32267
+ },
+ {
+ "epoch": 87.92370572207085,
+ "grad_norm": 1.3230786323547363,
+ "learning_rate": 7.551794973305093e-07,
+ "loss": 0.0756,
+ "step": 32268
+ },
+ {
+ "epoch": 87.92643051771117,
+ "grad_norm": 1.0407882928848267,
+ "learning_rate": 7.548431038055492e-07,
+ "loss": 0.0131,
+ "step": 32269
+ },
+ {
+ "epoch": 87.9291553133515,
+ "grad_norm": 0.7926427721977234,
+ "learning_rate": 7.545067822815733e-07,
+ "loss": 0.0089,
+ "step": 32270
+ },
+ {
+ "epoch": 87.93188010899182,
+ "grad_norm": 0.9069899916648865,
+ "learning_rate": 7.541705327612037e-07,
+ "loss": 0.0067,
+ "step": 32271
+ },
+ {
+ "epoch": 87.93460490463215,
+ "grad_norm": 1.5166010856628418,
+ "learning_rate": 7.53834355247055e-07,
+ "loss": 0.038,
+ "step": 32272
+ },
+ {
+ "epoch": 87.93732970027249,
+ "grad_norm": 0.4023211896419525,
+ "learning_rate": 7.534982497417476e-07,
+ "loss": 0.0041,
+ "step": 32273
+ },
+ {
+ "epoch": 87.9400544959128,
+ "grad_norm": 0.5464200973510742,
+ "learning_rate": 7.531622162478957e-07,
+ "loss": 0.0066,
+ "step": 32274
+ },
+ {
+ "epoch": 87.94277929155314,
+ "grad_norm": 0.7725834250450134,
+ "learning_rate": 7.528262547681209e-07,
+ "loss": 0.007,
+ "step": 32275
+ },
+ {
+ "epoch": 87.94550408719346,
+ "grad_norm": 1.0801151990890503,
+ "learning_rate": 7.524903653050386e-07,
+ "loss": 0.0088,
+ "step": 32276
+ },
+ {
+ "epoch": 87.94822888283379,
+ "grad_norm": 1.6720669269561768,
+ "learning_rate": 7.521545478612624e-07,
+ "loss": 0.0305,
+ "step": 32277
+ },
+ {
+ "epoch": 87.95095367847412,
+ "grad_norm": 1.424151062965393,
+ "learning_rate": 7.518188024394079e-07,
+ "loss": 0.0741,
+ "step": 32278
+ },
+ {
+ "epoch": 87.95367847411444,
+ "grad_norm": 0.8747496008872986,
+ "learning_rate": 7.514831290420921e-07,
+ "loss": 0.0149,
+ "step": 32279
+ },
+ {
+ "epoch": 87.95640326975477,
+ "grad_norm": 0.8342167735099792,
+ "learning_rate": 7.511475276719282e-07,
+ "loss": 0.0091,
+ "step": 32280
+ },
+ {
+ "epoch": 87.95912806539509,
+ "grad_norm": 1.8590582609176636,
+ "learning_rate": 7.508119983315288e-07,
+ "loss": 0.0747,
+ "step": 32281
+ },
+ {
+ "epoch": 87.96185286103542,
+ "grad_norm": 0.7957608103752136,
+ "learning_rate": 7.504765410235082e-07,
+ "loss": 0.0845,
+ "step": 32282
+ },
+ {
+ "epoch": 87.96457765667574,
+ "grad_norm": 1.4569305181503296,
+ "learning_rate": 7.501411557504801e-07,
+ "loss": 0.0111,
+ "step": 32283
+ },
+ {
+ "epoch": 87.96730245231608,
+ "grad_norm": 1.6300373077392578,
+ "learning_rate": 7.498058425150534e-07,
+ "loss": 0.0115,
+ "step": 32284
+ },
+ {
+ "epoch": 87.97002724795641,
+ "grad_norm": 1.4663770198822021,
+ "learning_rate": 7.494706013198406e-07,
+ "loss": 0.0222,
+ "step": 32285
+ },
+ {
+ "epoch": 87.97275204359673,
+ "grad_norm": 0.9172735810279846,
+ "learning_rate": 7.491354321674538e-07,
+ "loss": 0.0264,
+ "step": 32286
+ },
+ {
+ "epoch": 87.97547683923706,
+ "grad_norm": 1.16066312789917,
+ "learning_rate": 7.488003350605022e-07,
+ "loss": 0.0094,
+ "step": 32287
+ },
+ {
+ "epoch": 87.97820163487738,
+ "grad_norm": 1.1271787881851196,
+ "learning_rate": 7.48465310001596e-07,
+ "loss": 0.0119,
+ "step": 32288
+ },
+ {
+ "epoch": 87.98092643051771,
+ "grad_norm": 0.4977739453315735,
+ "learning_rate": 7.481303569933452e-07,
+ "loss": 0.0046,
+ "step": 32289
+ },
+ {
+ "epoch": 87.98365122615803,
+ "grad_norm": 0.9360405802726746,
+ "learning_rate": 7.477954760383554e-07,
+ "loss": 0.0171,
+ "step": 32290
+ },
+ {
+ "epoch": 87.98637602179836,
+ "grad_norm": 1.3794825077056885,
+ "learning_rate": 7.474606671392381e-07,
+ "loss": 0.0777,
+ "step": 32291
+ },
+ {
+ "epoch": 87.9891008174387,
+ "grad_norm": 1.2230182886123657,
+ "learning_rate": 7.471259302986e-07,
+ "loss": 0.1872,
+ "step": 32292
+ },
+ {
+ "epoch": 87.99182561307902,
+ "grad_norm": 1.1504573822021484,
+ "learning_rate": 7.467912655190468e-07,
+ "loss": 0.0295,
+ "step": 32293
+ },
+ {
+ "epoch": 87.99455040871935,
+ "grad_norm": 0.8149654269218445,
+ "learning_rate": 7.46456672803183e-07,
+ "loss": 0.0058,
+ "step": 32294
+ },
+ {
+ "epoch": 87.99727520435967,
+ "grad_norm": 1.2667367458343506,
+ "learning_rate": 7.461221521536199e-07,
+ "loss": 0.068,
+ "step": 32295
+ },
+ {
+ "epoch": 88.0,
+ "grad_norm": 0.6845406293869019,
+ "learning_rate": 7.457877035729588e-07,
+ "loss": 0.0066,
+ "step": 32296
+ },
+ {
+ "epoch": 88.00272479564033,
+ "grad_norm": 0.5321373343467712,
+ "learning_rate": 7.454533270638054e-07,
+ "loss": 0.0061,
+ "step": 32297
+ },
+ {
+ "epoch": 88.00544959128065,
+ "grad_norm": 1.668205976486206,
+ "learning_rate": 7.451190226287619e-07,
+ "loss": 0.0151,
+ "step": 32298
+ },
+ {
+ "epoch": 88.00817438692098,
+ "grad_norm": 1.514703631401062,
+ "learning_rate": 7.447847902704364e-07,
+ "loss": 0.0191,
+ "step": 32299
+ },
+ {
+ "epoch": 88.0108991825613,
+ "grad_norm": 0.7576146721839905,
+ "learning_rate": 7.444506299914278e-07,
+ "loss": 0.0846,
+ "step": 32300
+ },
+ {
+ "epoch": 88.01362397820164,
+ "grad_norm": 1.3126094341278076,
+ "learning_rate": 7.441165417943386e-07,
+ "loss": 0.0489,
+ "step": 32301
+ },
+ {
+ "epoch": 88.01634877384195,
+ "grad_norm": 1.2383345365524292,
+ "learning_rate": 7.437825256817732e-07,
+ "loss": 0.0263,
+ "step": 32302
+ },
+ {
+ "epoch": 88.01907356948229,
+ "grad_norm": 1.2545228004455566,
+ "learning_rate": 7.434485816563319e-07,
+ "loss": 0.0305,
+ "step": 32303
+ },
+ {
+ "epoch": 88.02179836512262,
+ "grad_norm": 0.7593337297439575,
+ "learning_rate": 7.431147097206126e-07,
+ "loss": 0.0091,
+ "step": 32304
+ },
+ {
+ "epoch": 88.02452316076294,
+ "grad_norm": 0.4529589116573334,
+ "learning_rate": 7.42780909877221e-07,
+ "loss": 0.0043,
+ "step": 32305
+ },
+ {
+ "epoch": 88.02724795640327,
+ "grad_norm": 0.9096177220344543,
+ "learning_rate": 7.424471821287516e-07,
+ "loss": 0.0062,
+ "step": 32306
+ },
+ {
+ "epoch": 88.02997275204359,
+ "grad_norm": 0.44406619668006897,
+ "learning_rate": 7.421135264778067e-07,
+ "loss": 0.0047,
+ "step": 32307
+ },
+ {
+ "epoch": 88.03269754768392,
+ "grad_norm": 0.9156663417816162,
+ "learning_rate": 7.417799429269835e-07,
+ "loss": 0.0068,
+ "step": 32308
+ },
+ {
+ "epoch": 88.03542234332426,
+ "grad_norm": 3.045671224594116,
+ "learning_rate": 7.414464314788816e-07,
+ "loss": 0.012,
+ "step": 32309
+ },
+ {
+ "epoch": 88.03814713896458,
+ "grad_norm": 1.0240148305892944,
+ "learning_rate": 7.411129921360937e-07,
+ "loss": 0.0187,
+ "step": 32310
+ },
+ {
+ "epoch": 88.04087193460491,
+ "grad_norm": 0.8861274123191833,
+ "learning_rate": 7.407796249012223e-07,
+ "loss": 0.0712,
+ "step": 32311
+ },
+ {
+ "epoch": 88.04359673024523,
+ "grad_norm": 1.2645601034164429,
+ "learning_rate": 7.404463297768616e-07,
+ "loss": 0.1133,
+ "step": 32312
+ },
+ {
+ "epoch": 88.04632152588556,
+ "grad_norm": 1.4435468912124634,
+ "learning_rate": 7.401131067656054e-07,
+ "loss": 0.1398,
+ "step": 32313
+ },
+ {
+ "epoch": 88.04904632152588,
+ "grad_norm": 1.1481436491012573,
+ "learning_rate": 7.397799558700502e-07,
+ "loss": 0.1024,
+ "step": 32314
+ },
+ {
+ "epoch": 88.05177111716621,
+ "grad_norm": 0.8060486912727356,
+ "learning_rate": 7.394468770927909e-07,
+ "loss": 0.1219,
+ "step": 32315
+ },
+ {
+ "epoch": 88.05449591280654,
+ "grad_norm": 1.308603048324585,
+ "learning_rate": 7.39113870436422e-07,
+ "loss": 0.0473,
+ "step": 32316
+ },
+ {
+ "epoch": 88.05722070844686,
+ "grad_norm": 0.8349003195762634,
+ "learning_rate": 7.387809359035369e-07,
+ "loss": 0.01,
+ "step": 32317
+ },
+ {
+ "epoch": 88.0599455040872,
+ "grad_norm": 0.8540116548538208,
+ "learning_rate": 7.384480734967248e-07,
+ "loss": 0.0071,
+ "step": 32318
+ },
+ {
+ "epoch": 88.06267029972751,
+ "grad_norm": 1.5727182626724243,
+ "learning_rate": 7.381152832185833e-07,
+ "loss": 0.009,
+ "step": 32319
+ },
+ {
+ "epoch": 88.06539509536785,
+ "grad_norm": 1.4504870176315308,
+ "learning_rate": 7.377825650717029e-07,
+ "loss": 0.0208,
+ "step": 32320
+ },
+ {
+ "epoch": 88.06811989100818,
+ "grad_norm": 1.179781436920166,
+ "learning_rate": 7.374499190586725e-07,
+ "loss": 0.0109,
+ "step": 32321
+ },
+ {
+ "epoch": 88.0708446866485,
+ "grad_norm": 1.2214266061782837,
+ "learning_rate": 7.371173451820812e-07,
+ "loss": 0.0073,
+ "step": 32322
+ },
+ {
+ "epoch": 88.07356948228883,
+ "grad_norm": 1.435700535774231,
+ "learning_rate": 7.367848434445246e-07,
+ "loss": 0.0188,
+ "step": 32323
+ },
+ {
+ "epoch": 88.07629427792915,
+ "grad_norm": 1.0993831157684326,
+ "learning_rate": 7.364524138485896e-07,
+ "loss": 0.0285,
+ "step": 32324
+ },
+ {
+ "epoch": 88.07901907356948,
+ "grad_norm": 1.0190867185592651,
+ "learning_rate": 7.361200563968662e-07,
+ "loss": 0.0138,
+ "step": 32325
+ },
+ {
+ "epoch": 88.0817438692098,
+ "grad_norm": 0.7319145202636719,
+ "learning_rate": 7.357877710919392e-07,
+ "loss": 0.0138,
+ "step": 32326
+ },
+ {
+ "epoch": 88.08446866485014,
+ "grad_norm": 0.5817360877990723,
+ "learning_rate": 7.354555579363998e-07,
+ "loss": 0.0069,
+ "step": 32327
+ },
+ {
+ "epoch": 88.08719346049047,
+ "grad_norm": 0.5780275464057922,
+ "learning_rate": 7.351234169328359e-07,
+ "loss": 0.0048,
+ "step": 32328
+ },
+ {
+ "epoch": 88.08991825613079,
+ "grad_norm": 1.1613845825195312,
+ "learning_rate": 7.34791348083831e-07,
+ "loss": 0.0085,
+ "step": 32329
+ },
+ {
+ "epoch": 88.09264305177112,
+ "grad_norm": 0.7122410535812378,
+ "learning_rate": 7.344593513919718e-07,
+ "loss": 0.0075,
+ "step": 32330
+ },
+ {
+ "epoch": 88.09536784741144,
+ "grad_norm": 1.503135085105896,
+ "learning_rate": 7.341274268598476e-07,
+ "loss": 0.1493,
+ "step": 32331
+ },
+ {
+ "epoch": 88.09809264305177,
+ "grad_norm": 0.9116985201835632,
+ "learning_rate": 7.337955744900394e-07,
+ "loss": 0.0111,
+ "step": 32332
+ },
+ {
+ "epoch": 88.1008174386921,
+ "grad_norm": 0.9681580066680908,
+ "learning_rate": 7.334637942851342e-07,
+ "loss": 0.0193,
+ "step": 32333
+ },
+ {
+ "epoch": 88.10354223433242,
+ "grad_norm": 1.3105651140213013,
+ "learning_rate": 7.331320862477121e-07,
+ "loss": 0.0265,
+ "step": 32334
+ },
+ {
+ "epoch": 88.10626702997276,
+ "grad_norm": 0.8215150237083435,
+ "learning_rate": 7.328004503803609e-07,
+ "loss": 0.0076,
+ "step": 32335
+ },
+ {
+ "epoch": 88.10899182561307,
+ "grad_norm": 0.8611791729927063,
+ "learning_rate": 7.324688866856621e-07,
+ "loss": 0.0064,
+ "step": 32336
+ },
+ {
+ "epoch": 88.11171662125341,
+ "grad_norm": 0.9168757200241089,
+ "learning_rate": 7.321373951661969e-07,
+ "loss": 0.0159,
+ "step": 32337
+ },
+ {
+ "epoch": 88.11444141689373,
+ "grad_norm": 0.796818196773529,
+ "learning_rate": 7.318059758245455e-07,
+ "loss": 0.023,
+ "step": 32338
+ },
+ {
+ "epoch": 88.11716621253406,
+ "grad_norm": 1.072625994682312,
+ "learning_rate": 7.314746286632913e-07,
+ "loss": 0.0103,
+ "step": 32339
+ },
+ {
+ "epoch": 88.11989100817439,
+ "grad_norm": 0.5793470740318298,
+ "learning_rate": 7.311433536850165e-07,
+ "loss": 0.0051,
+ "step": 32340
+ },
+ {
+ "epoch": 88.12261580381471,
+ "grad_norm": 1.0612388849258423,
+ "learning_rate": 7.308121508922983e-07,
+ "loss": 0.0042,
+ "step": 32341
+ },
+ {
+ "epoch": 88.12534059945504,
+ "grad_norm": 0.8427013754844666,
+ "learning_rate": 7.304810202877144e-07,
+ "loss": 0.0136,
+ "step": 32342
+ },
+ {
+ "epoch": 88.12806539509536,
+ "grad_norm": 1.1014618873596191,
+ "learning_rate": 7.301499618738484e-07,
+ "loss": 0.0124,
+ "step": 32343
+ },
+ {
+ "epoch": 88.1307901907357,
+ "grad_norm": 0.49778488278388977,
+ "learning_rate": 7.29818975653277e-07,
+ "loss": 0.0046,
+ "step": 32344
+ },
+ {
+ "epoch": 88.13351498637603,
+ "grad_norm": 0.4955245554447174,
+ "learning_rate": 7.29488061628576e-07,
+ "loss": 0.0056,
+ "step": 32345
+ },
+ {
+ "epoch": 88.13623978201635,
+ "grad_norm": 1.03538179397583,
+ "learning_rate": 7.291572198023223e-07,
+ "loss": 0.0126,
+ "step": 32346
+ },
+ {
+ "epoch": 88.13896457765668,
+ "grad_norm": 1.4988065958023071,
+ "learning_rate": 7.288264501770947e-07,
+ "loss": 0.0461,
+ "step": 32347
+ },
+ {
+ "epoch": 88.141689373297,
+ "grad_norm": 1.4023349285125732,
+ "learning_rate": 7.284957527554693e-07,
+ "loss": 0.0421,
+ "step": 32348
+ },
+ {
+ "epoch": 88.14441416893733,
+ "grad_norm": 0.6524114608764648,
+ "learning_rate": 7.281651275400203e-07,
+ "loss": 0.0051,
+ "step": 32349
+ },
+ {
+ "epoch": 88.14713896457765,
+ "grad_norm": 1.2888778448104858,
+ "learning_rate": 7.278345745333204e-07,
+ "loss": 0.0327,
+ "step": 32350
+ },
+ {
+ "epoch": 88.14986376021798,
+ "grad_norm": 0.5471346378326416,
+ "learning_rate": 7.275040937379485e-07,
+ "loss": 0.0052,
+ "step": 32351
+ },
+ {
+ "epoch": 88.15258855585832,
+ "grad_norm": 0.8345967531204224,
+ "learning_rate": 7.27173685156477e-07,
+ "loss": 0.0131,
+ "step": 32352
+ },
+ {
+ "epoch": 88.15531335149863,
+ "grad_norm": 1.3571503162384033,
+ "learning_rate": 7.268433487914784e-07,
+ "loss": 0.0755,
+ "step": 32353
+ },
+ {
+ "epoch": 88.15803814713897,
+ "grad_norm": 0.9906461238861084,
+ "learning_rate": 7.265130846455226e-07,
+ "loss": 0.0126,
+ "step": 32354
+ },
+ {
+ "epoch": 88.16076294277929,
+ "grad_norm": 1.6007304191589355,
+ "learning_rate": 7.261828927211867e-07,
+ "loss": 0.0492,
+ "step": 32355
+ },
+ {
+ "epoch": 88.16348773841962,
+ "grad_norm": 1.129335880279541,
+ "learning_rate": 7.258527730210385e-07,
+ "loss": 0.0263,
+ "step": 32356
+ },
+ {
+ "epoch": 88.16621253405995,
+ "grad_norm": 1.786434292793274,
+ "learning_rate": 7.255227255476516e-07,
+ "loss": 0.1025,
+ "step": 32357
+ },
+ {
+ "epoch": 88.16893732970027,
+ "grad_norm": 0.9673599600791931,
+ "learning_rate": 7.25192750303596e-07,
+ "loss": 0.0104,
+ "step": 32358
+ },
+ {
+ "epoch": 88.1716621253406,
+ "grad_norm": 1.228432297706604,
+ "learning_rate": 7.248628472914388e-07,
+ "loss": 0.1144,
+ "step": 32359
+ },
+ {
+ "epoch": 88.17438692098092,
+ "grad_norm": 2.6414666175842285,
+ "learning_rate": 7.245330165137521e-07,
+ "loss": 0.0269,
+ "step": 32360
+ },
+ {
+ "epoch": 88.17711171662125,
+ "grad_norm": 1.284377098083496,
+ "learning_rate": 7.24203257973104e-07,
+ "loss": 0.0208,
+ "step": 32361
+ },
+ {
+ "epoch": 88.17983651226157,
+ "grad_norm": 0.9169119000434875,
+ "learning_rate": 7.238735716720635e-07,
+ "loss": 0.0114,
+ "step": 32362
+ },
+ {
+ "epoch": 88.1825613079019,
+ "grad_norm": 0.5177499055862427,
+ "learning_rate": 7.235439576131942e-07,
+ "loss": 0.0043,
+ "step": 32363
+ },
+ {
+ "epoch": 88.18528610354224,
+ "grad_norm": 0.7183284163475037,
+ "learning_rate": 7.232144157990683e-07,
+ "loss": 0.0062,
+ "step": 32364
+ },
+ {
+ "epoch": 88.18801089918256,
+ "grad_norm": 1.116137981414795,
+ "learning_rate": 7.228849462322496e-07,
+ "loss": 0.0185,
+ "step": 32365
+ },
+ {
+ "epoch": 88.19073569482289,
+ "grad_norm": 1.1816461086273193,
+ "learning_rate": 7.225555489153024e-07,
+ "loss": 0.0287,
+ "step": 32366
+ },
+ {
+ "epoch": 88.19346049046321,
+ "grad_norm": 1.434565544128418,
+ "learning_rate": 7.22226223850796e-07,
+ "loss": 0.1179,
+ "step": 32367
+ },
+ {
+ "epoch": 88.19618528610354,
+ "grad_norm": 0.9229934215545654,
+ "learning_rate": 7.218969710412938e-07,
+ "loss": 0.0194,
+ "step": 32368
+ },
+ {
+ "epoch": 88.19891008174388,
+ "grad_norm": 1.2228593826293945,
+ "learning_rate": 7.215677904893593e-07,
+ "loss": 0.0796,
+ "step": 32369
+ },
+ {
+ "epoch": 88.2016348773842,
+ "grad_norm": 1.040808081626892,
+ "learning_rate": 7.212386821975537e-07,
+ "loss": 0.0153,
+ "step": 32370
+ },
+ {
+ "epoch": 88.20435967302453,
+ "grad_norm": 1.1550230979919434,
+ "learning_rate": 7.209096461684451e-07,
+ "loss": 0.0217,
+ "step": 32371
+ },
+ {
+ "epoch": 88.20708446866485,
+ "grad_norm": 0.7656853795051575,
+ "learning_rate": 7.205806824045924e-07,
+ "loss": 0.0066,
+ "step": 32372
+ },
+ {
+ "epoch": 88.20980926430518,
+ "grad_norm": 1.22402024269104,
+ "learning_rate": 7.202517909085583e-07,
+ "loss": 0.0481,
+ "step": 32373
+ },
+ {
+ "epoch": 88.2125340599455,
+ "grad_norm": 1.1728020906448364,
+ "learning_rate": 7.199229716829048e-07,
+ "loss": 0.01,
+ "step": 32374
+ },
+ {
+ "epoch": 88.21525885558583,
+ "grad_norm": 1.5886129140853882,
+ "learning_rate": 7.195942247301923e-07,
+ "loss": 0.0297,
+ "step": 32375
+ },
+ {
+ "epoch": 88.21798365122616,
+ "grad_norm": 6.2917585372924805,
+ "learning_rate": 7.192655500529821e-07,
+ "loss": 0.0119,
+ "step": 32376
+ },
+ {
+ "epoch": 88.22070844686648,
+ "grad_norm": 0.9772661328315735,
+ "learning_rate": 7.189369476538333e-07,
+ "loss": 0.0138,
+ "step": 32377
+ },
+ {
+ "epoch": 88.22343324250681,
+ "grad_norm": 0.9263694882392883,
+ "learning_rate": 7.186084175353047e-07,
+ "loss": 0.0067,
+ "step": 32378
+ },
+ {
+ "epoch": 88.22615803814713,
+ "grad_norm": 0.7094914317131042,
+ "learning_rate": 7.182799596999534e-07,
+ "loss": 0.0087,
+ "step": 32379
+ },
+ {
+ "epoch": 88.22888283378747,
+ "grad_norm": 0.8742457628250122,
+ "learning_rate": 7.179515741503418e-07,
+ "loss": 0.0065,
+ "step": 32380
+ },
+ {
+ "epoch": 88.2316076294278,
+ "grad_norm": 0.9961302876472473,
+ "learning_rate": 7.176232608890232e-07,
+ "loss": 0.0073,
+ "step": 32381
+ },
+ {
+ "epoch": 88.23433242506812,
+ "grad_norm": 1.0967893600463867,
+ "learning_rate": 7.172950199185569e-07,
+ "loss": 0.0069,
+ "step": 32382
+ },
+ {
+ "epoch": 88.23705722070845,
+ "grad_norm": 1.0302423238754272,
+ "learning_rate": 7.169668512414951e-07,
+ "loss": 0.0112,
+ "step": 32383
+ },
+ {
+ "epoch": 88.23978201634877,
+ "grad_norm": 0.757252037525177,
+ "learning_rate": 7.166387548603993e-07,
+ "loss": 0.0086,
+ "step": 32384
+ },
+ {
+ "epoch": 88.2425068119891,
+ "grad_norm": 1.1823132038116455,
+ "learning_rate": 7.163107307778227e-07,
+ "loss": 0.0259,
+ "step": 32385
+ },
+ {
+ "epoch": 88.24523160762942,
+ "grad_norm": 1.1264976263046265,
+ "learning_rate": 7.159827789963181e-07,
+ "loss": 0.0075,
+ "step": 32386
+ },
+ {
+ "epoch": 88.24795640326975,
+ "grad_norm": 0.5220778584480286,
+ "learning_rate": 7.156548995184398e-07,
+ "loss": 0.0052,
+ "step": 32387
+ },
+ {
+ "epoch": 88.25068119891009,
+ "grad_norm": 1.4959135055541992,
+ "learning_rate": 7.153270923467437e-07,
+ "loss": 0.0392,
+ "step": 32388
+ },
+ {
+ "epoch": 88.2534059945504,
+ "grad_norm": 0.5625965595245361,
+ "learning_rate": 7.149993574837822e-07,
+ "loss": 0.0078,
+ "step": 32389
+ },
+ {
+ "epoch": 88.25613079019074,
+ "grad_norm": 1.004814624786377,
+ "learning_rate": 7.146716949321042e-07,
+ "loss": 0.0143,
+ "step": 32390
+ },
+ {
+ "epoch": 88.25885558583106,
+ "grad_norm": 1.803884506225586,
+ "learning_rate": 7.143441046942645e-07,
+ "loss": 0.0146,
+ "step": 32391
+ },
+ {
+ "epoch": 88.26158038147139,
+ "grad_norm": 0.934225857257843,
+ "learning_rate": 7.140165867728166e-07,
+ "loss": 0.0095,
+ "step": 32392
+ },
+ {
+ "epoch": 88.26430517711172,
+ "grad_norm": 0.8516022562980652,
+ "learning_rate": 7.136891411703073e-07,
+ "loss": 0.0092,
+ "step": 32393
+ },
+ {
+ "epoch": 88.26702997275204,
+ "grad_norm": 1.3735359907150269,
+ "learning_rate": 7.13361767889289e-07,
+ "loss": 0.0179,
+ "step": 32394
+ },
+ {
+ "epoch": 88.26975476839237,
+ "grad_norm": 1.2124453783035278,
+ "learning_rate": 7.130344669323086e-07,
+ "loss": 0.0215,
+ "step": 32395
+ },
+ {
+ "epoch": 88.2724795640327,
+ "grad_norm": 0.7687087655067444,
+ "learning_rate": 7.127072383019185e-07,
+ "loss": 0.0072,
+ "step": 32396
+ },
+ {
+ "epoch": 88.27520435967303,
+ "grad_norm": 0.8406593203544617,
+ "learning_rate": 7.123800820006643e-07,
+ "loss": 0.0308,
+ "step": 32397
+ },
+ {
+ "epoch": 88.27792915531334,
+ "grad_norm": 0.5331346988677979,
+ "learning_rate": 7.120529980310964e-07,
+ "loss": 0.0048,
+ "step": 32398
+ },
+ {
+ "epoch": 88.28065395095368,
+ "grad_norm": 1.1469411849975586,
+ "learning_rate": 7.117259863957593e-07,
+ "loss": 0.0169,
+ "step": 32399
+ },
+ {
+ "epoch": 88.28337874659401,
+ "grad_norm": 0.6620889902114868,
+ "learning_rate": 7.113990470972021e-07,
+ "loss": 0.0056,
+ "step": 32400
+ },
+ {
+ "epoch": 88.28610354223433,
+ "grad_norm": 1.2000864744186401,
+ "learning_rate": 7.110721801379705e-07,
+ "loss": 0.0073,
+ "step": 32401
+ },
+ {
+ "epoch": 88.28882833787466,
+ "grad_norm": 0.6067875623703003,
+ "learning_rate": 7.107453855206103e-07,
+ "loss": 0.0049,
+ "step": 32402
+ },
+ {
+ "epoch": 88.29155313351498,
+ "grad_norm": 1.0333188772201538,
+ "learning_rate": 7.104186632476639e-07,
+ "loss": 0.0191,
+ "step": 32403
+ },
+ {
+ "epoch": 88.29427792915531,
+ "grad_norm": 1.1295621395111084,
+ "learning_rate": 7.100920133216793e-07,
+ "loss": 0.0111,
+ "step": 32404
+ },
+ {
+ "epoch": 88.29700272479565,
+ "grad_norm": 1.2730379104614258,
+ "learning_rate": 7.097654357451999e-07,
+ "loss": 0.0128,
+ "step": 32405
+ },
+ {
+ "epoch": 88.29972752043597,
+ "grad_norm": 0.9196459054946899,
+ "learning_rate": 7.094389305207683e-07,
+ "loss": 0.0152,
+ "step": 32406
+ },
+ {
+ "epoch": 88.3024523160763,
+ "grad_norm": 0.8327659368515015,
+ "learning_rate": 7.091124976509245e-07,
+ "loss": 0.0111,
+ "step": 32407
+ },
+ {
+ "epoch": 88.30517711171662,
+ "grad_norm": 1.1799390316009521,
+ "learning_rate": 7.087861371382143e-07,
+ "loss": 0.0219,
+ "step": 32408
+ },
+ {
+ "epoch": 88.30790190735695,
+ "grad_norm": 1.1664032936096191,
+ "learning_rate": 7.084598489851813e-07,
+ "loss": 0.0736,
+ "step": 32409
+ },
+ {
+ "epoch": 88.31062670299727,
+ "grad_norm": 0.7837077379226685,
+ "learning_rate": 7.081336331943623e-07,
+ "loss": 0.0091,
+ "step": 32410
+ },
+ {
+ "epoch": 88.3133514986376,
+ "grad_norm": 0.7269379496574402,
+ "learning_rate": 7.078074897682985e-07,
+ "loss": 0.0127,
+ "step": 32411
+ },
+ {
+ "epoch": 88.31607629427793,
+ "grad_norm": 1.1005768775939941,
+ "learning_rate": 7.074814187095324e-07,
+ "loss": 0.0303,
+ "step": 32412
+ },
+ {
+ "epoch": 88.31880108991825,
+ "grad_norm": 1.188597321510315,
+ "learning_rate": 7.071554200206021e-07,
+ "loss": 0.0151,
+ "step": 32413
+ },
+ {
+ "epoch": 88.32152588555859,
+ "grad_norm": 0.9218148589134216,
+ "learning_rate": 7.068294937040465e-07,
+ "loss": 0.1127,
+ "step": 32414
+ },
+ {
+ "epoch": 88.3242506811989,
+ "grad_norm": 0.5721375346183777,
+ "learning_rate": 7.065036397624014e-07,
+ "loss": 0.0041,
+ "step": 32415
+ },
+ {
+ "epoch": 88.32697547683924,
+ "grad_norm": 0.8904895782470703,
+ "learning_rate": 7.061778581982093e-07,
+ "loss": 0.0125,
+ "step": 32416
+ },
+ {
+ "epoch": 88.32970027247957,
+ "grad_norm": 0.7308708429336548,
+ "learning_rate": 7.058521490140047e-07,
+ "loss": 0.008,
+ "step": 32417
+ },
+ {
+ "epoch": 88.33242506811989,
+ "grad_norm": 0.7259790897369385,
+ "learning_rate": 7.055265122123245e-07,
+ "loss": 0.0074,
+ "step": 32418
+ },
+ {
+ "epoch": 88.33514986376022,
+ "grad_norm": 0.828073263168335,
+ "learning_rate": 7.052009477957022e-07,
+ "loss": 0.0063,
+ "step": 32419
+ },
+ {
+ "epoch": 88.33787465940054,
+ "grad_norm": 0.6372078657150269,
+ "learning_rate": 7.048754557666781e-07,
+ "loss": 0.0057,
+ "step": 32420
+ },
+ {
+ "epoch": 88.34059945504087,
+ "grad_norm": 1.2552013397216797,
+ "learning_rate": 7.045500361277846e-07,
+ "loss": 0.0165,
+ "step": 32421
+ },
+ {
+ "epoch": 88.34332425068119,
+ "grad_norm": 1.857148289680481,
+ "learning_rate": 7.042246888815563e-07,
+ "loss": 0.0459,
+ "step": 32422
+ },
+ {
+ "epoch": 88.34604904632153,
+ "grad_norm": 0.9404576420783997,
+ "learning_rate": 7.038994140305255e-07,
+ "loss": 0.0058,
+ "step": 32423
+ },
+ {
+ "epoch": 88.34877384196186,
+ "grad_norm": 0.6070570945739746,
+ "learning_rate": 7.035742115772281e-07,
+ "loss": 0.0073,
+ "step": 32424
+ },
+ {
+ "epoch": 88.35149863760218,
+ "grad_norm": 1.4632357358932495,
+ "learning_rate": 7.032490815241943e-07,
+ "loss": 0.0318,
+ "step": 32425
+ },
+ {
+ "epoch": 88.35422343324251,
+ "grad_norm": 1.0554046630859375,
+ "learning_rate": 7.029240238739587e-07,
+ "loss": 0.0385,
+ "step": 32426
+ },
+ {
+ "epoch": 88.35694822888283,
+ "grad_norm": 1.18092679977417,
+ "learning_rate": 7.025990386290527e-07,
+ "loss": 0.087,
+ "step": 32427
+ },
+ {
+ "epoch": 88.35967302452316,
+ "grad_norm": 1.4152570962905884,
+ "learning_rate": 7.022741257920029e-07,
+ "loss": 0.0213,
+ "step": 32428
+ },
+ {
+ "epoch": 88.3623978201635,
+ "grad_norm": 1.0414149761199951,
+ "learning_rate": 7.019492853653454e-07,
+ "loss": 0.0132,
+ "step": 32429
+ },
+ {
+ "epoch": 88.36512261580381,
+ "grad_norm": 2.578324317932129,
+ "learning_rate": 7.016245173516079e-07,
+ "loss": 0.0584,
+ "step": 32430
+ },
+ {
+ "epoch": 88.36784741144415,
+ "grad_norm": 1.3893414735794067,
+ "learning_rate": 7.012998217533174e-07,
+ "loss": 0.1467,
+ "step": 32431
+ },
+ {
+ "epoch": 88.37057220708446,
+ "grad_norm": 0.9474067687988281,
+ "learning_rate": 7.009751985730062e-07,
+ "loss": 0.0168,
+ "step": 32432
+ },
+ {
+ "epoch": 88.3732970027248,
+ "grad_norm": 0.7133904695510864,
+ "learning_rate": 7.006506478132003e-07,
+ "loss": 0.0072,
+ "step": 32433
+ },
+ {
+ "epoch": 88.37602179836512,
+ "grad_norm": 1.3905742168426514,
+ "learning_rate": 7.003261694764286e-07,
+ "loss": 0.0086,
+ "step": 32434
+ },
+ {
+ "epoch": 88.37874659400545,
+ "grad_norm": 0.76395183801651,
+ "learning_rate": 7.000017635652157e-07,
+ "loss": 0.0078,
+ "step": 32435
+ },
+ {
+ "epoch": 88.38147138964578,
+ "grad_norm": 1.784542441368103,
+ "learning_rate": 6.996774300820919e-07,
+ "loss": 0.0384,
+ "step": 32436
+ },
+ {
+ "epoch": 88.3841961852861,
+ "grad_norm": 0.6089380383491516,
+ "learning_rate": 6.993531690295796e-07,
+ "loss": 0.0073,
+ "step": 32437
+ },
+ {
+ "epoch": 88.38692098092643,
+ "grad_norm": 2.191131591796875,
+ "learning_rate": 6.990289804102068e-07,
+ "loss": 0.1771,
+ "step": 32438
+ },
+ {
+ "epoch": 88.38964577656675,
+ "grad_norm": 0.5078168511390686,
+ "learning_rate": 6.987048642264938e-07,
+ "loss": 0.004,
+ "step": 32439
+ },
+ {
+ "epoch": 88.39237057220708,
+ "grad_norm": 0.6841101050376892,
+ "learning_rate": 6.983808204809706e-07,
+ "loss": 0.0072,
+ "step": 32440
+ },
+ {
+ "epoch": 88.39509536784742,
+ "grad_norm": 0.9162114262580872,
+ "learning_rate": 6.980568491761586e-07,
+ "loss": 0.1347,
+ "step": 32441
+ },
+ {
+ "epoch": 88.39782016348774,
+ "grad_norm": 1.1738747358322144,
+ "learning_rate": 6.97732950314578e-07,
+ "loss": 0.0113,
+ "step": 32442
+ },
+ {
+ "epoch": 88.40054495912807,
+ "grad_norm": 1.038608431816101,
+ "learning_rate": 6.974091238987557e-07,
+ "loss": 0.0778,
+ "step": 32443
+ },
+ {
+ "epoch": 88.40326975476839,
+ "grad_norm": 0.9673500657081604,
+ "learning_rate": 6.970853699312108e-07,
+ "loss": 0.0115,
+ "step": 32444
+ },
+ {
+ "epoch": 88.40599455040872,
+ "grad_norm": 1.5304644107818604,
+ "learning_rate": 6.967616884144668e-07,
+ "loss": 0.0077,
+ "step": 32445
+ },
+ {
+ "epoch": 88.40871934604904,
+ "grad_norm": 1.1385674476623535,
+ "learning_rate": 6.964380793510428e-07,
+ "loss": 0.0169,
+ "step": 32446
+ },
+ {
+ "epoch": 88.41144414168937,
+ "grad_norm": 0.6433848142623901,
+ "learning_rate": 6.961145427434601e-07,
+ "loss": 0.0073,
+ "step": 32447
+ },
+ {
+ "epoch": 88.4141689373297,
+ "grad_norm": 1.918460726737976,
+ "learning_rate": 6.957910785942357e-07,
+ "loss": 0.1319,
+ "step": 32448
+ },
+ {
+ "epoch": 88.41689373297002,
+ "grad_norm": 0.9632396697998047,
+ "learning_rate": 6.95467686905893e-07,
+ "loss": 0.015,
+ "step": 32449
+ },
+ {
+ "epoch": 88.41961852861036,
+ "grad_norm": 0.9392176866531372,
+ "learning_rate": 6.951443676809489e-07,
+ "loss": 0.0294,
+ "step": 32450
+ },
+ {
+ "epoch": 88.42234332425068,
+ "grad_norm": 0.7236586809158325,
+ "learning_rate": 6.948211209219191e-07,
+ "loss": 0.0097,
+ "step": 32451
+ },
+ {
+ "epoch": 88.42506811989101,
+ "grad_norm": 1.1634674072265625,
+ "learning_rate": 6.94497946631324e-07,
+ "loss": 0.0315,
+ "step": 32452
+ },
+ {
+ "epoch": 88.42779291553134,
+ "grad_norm": 1.0488587617874146,
+ "learning_rate": 6.941748448116803e-07,
+ "loss": 0.0331,
+ "step": 32453
+ },
+ {
+ "epoch": 88.43051771117166,
+ "grad_norm": 0.7121601104736328,
+ "learning_rate": 6.938518154655027e-07,
+ "loss": 0.006,
+ "step": 32454
+ },
+ {
+ "epoch": 88.433242506812,
+ "grad_norm": 1.0011286735534668,
+ "learning_rate": 6.93528858595306e-07,
+ "loss": 0.0112,
+ "step": 32455
+ },
+ {
+ "epoch": 88.43596730245231,
+ "grad_norm": 0.9078178405761719,
+ "learning_rate": 6.93205974203609e-07,
+ "loss": 0.0054,
+ "step": 32456
+ },
+ {
+ "epoch": 88.43869209809264,
+ "grad_norm": 0.8956049084663391,
+ "learning_rate": 6.928831622929244e-07,
+ "loss": 0.0112,
+ "step": 32457
+ },
+ {
+ "epoch": 88.44141689373296,
+ "grad_norm": 0.5952017307281494,
+ "learning_rate": 6.925604228657657e-07,
+ "loss": 0.0036,
+ "step": 32458
+ },
+ {
+ "epoch": 88.4441416893733,
+ "grad_norm": 1.0305442810058594,
+ "learning_rate": 6.922377559246462e-07,
+ "loss": 0.0138,
+ "step": 32459
+ },
+ {
+ "epoch": 88.44686648501363,
+ "grad_norm": 0.5812694430351257,
+ "learning_rate": 6.919151614720787e-07,
+ "loss": 0.0056,
+ "step": 32460
+ },
+ {
+ "epoch": 88.44959128065395,
+ "grad_norm": 2.165811538696289,
+ "learning_rate": 6.915926395105776e-07,
+ "loss": 0.0838,
+ "step": 32461
+ },
+ {
+ "epoch": 88.45231607629428,
+ "grad_norm": 0.9405644536018372,
+ "learning_rate": 6.912701900426544e-07,
+ "loss": 0.0145,
+ "step": 32462
+ },
+ {
+ "epoch": 88.4550408719346,
+ "grad_norm": 0.5374742150306702,
+ "learning_rate": 6.909478130708192e-07,
+ "loss": 0.0058,
+ "step": 32463
+ },
+ {
+ "epoch": 88.45776566757493,
+ "grad_norm": 1.6932237148284912,
+ "learning_rate": 6.906255085975799e-07,
+ "loss": 0.0779,
+ "step": 32464
+ },
+ {
+ "epoch": 88.46049046321527,
+ "grad_norm": 1.8270277976989746,
+ "learning_rate": 6.903032766254525e-07,
+ "loss": 0.0188,
+ "step": 32465
+ },
+ {
+ "epoch": 88.46321525885558,
+ "grad_norm": 1.3156033754348755,
+ "learning_rate": 6.899811171569437e-07,
+ "loss": 0.0654,
+ "step": 32466
+ },
+ {
+ "epoch": 88.46594005449592,
+ "grad_norm": 1.1352930068969727,
+ "learning_rate": 6.896590301945616e-07,
+ "loss": 0.0187,
+ "step": 32467
+ },
+ {
+ "epoch": 88.46866485013624,
+ "grad_norm": 1.0334529876708984,
+ "learning_rate": 6.893370157408141e-07,
+ "loss": 0.0139,
+ "step": 32468
+ },
+ {
+ "epoch": 88.47138964577657,
+ "grad_norm": 0.9468151926994324,
+ "learning_rate": 6.890150737982115e-07,
+ "loss": 0.0125,
+ "step": 32469
+ },
+ {
+ "epoch": 88.47411444141689,
+ "grad_norm": 0.5985283851623535,
+ "learning_rate": 6.886932043692596e-07,
+ "loss": 0.0066,
+ "step": 32470
+ },
+ {
+ "epoch": 88.47683923705722,
+ "grad_norm": 1.0171117782592773,
+ "learning_rate": 6.883714074564663e-07,
+ "loss": 0.0615,
+ "step": 32471
+ },
+ {
+ "epoch": 88.47956403269755,
+ "grad_norm": 0.8388184309005737,
+ "learning_rate": 6.88049683062334e-07,
+ "loss": 0.0098,
+ "step": 32472
+ },
+ {
+ "epoch": 88.48228882833787,
+ "grad_norm": 1.535872220993042,
+ "learning_rate": 6.87728031189373e-07,
+ "loss": 0.0489,
+ "step": 32473
+ },
+ {
+ "epoch": 88.4850136239782,
+ "grad_norm": 0.8846063613891602,
+ "learning_rate": 6.874064518400869e-07,
+ "loss": 0.0107,
+ "step": 32474
+ },
+ {
+ "epoch": 88.48773841961852,
+ "grad_norm": 0.6561586856842041,
+ "learning_rate": 6.870849450169781e-07,
+ "loss": 0.006,
+ "step": 32475
+ },
+ {
+ "epoch": 88.49046321525886,
+ "grad_norm": 2.757720708847046,
+ "learning_rate": 6.867635107225512e-07,
+ "loss": 0.0369,
+ "step": 32476
+ },
+ {
+ "epoch": 88.49318801089919,
+ "grad_norm": 0.6703823208808899,
+ "learning_rate": 6.864421489593132e-07,
+ "loss": 0.0054,
+ "step": 32477
+ },
+ {
+ "epoch": 88.49591280653951,
+ "grad_norm": 1.0841474533081055,
+ "learning_rate": 6.861208597297631e-07,
+ "loss": 0.0177,
+ "step": 32478
+ },
+ {
+ "epoch": 88.49863760217984,
+ "grad_norm": 0.9819075465202332,
+ "learning_rate": 6.857996430364045e-07,
+ "loss": 0.0146,
+ "step": 32479
+ },
+ {
+ "epoch": 88.50136239782016,
+ "grad_norm": 2.2619194984436035,
+ "learning_rate": 6.854784988817365e-07,
+ "loss": 0.0089,
+ "step": 32480
+ },
+ {
+ "epoch": 88.50408719346049,
+ "grad_norm": 0.8117721080780029,
+ "learning_rate": 6.851574272682638e-07,
+ "loss": 0.1066,
+ "step": 32481
+ },
+ {
+ "epoch": 88.50681198910081,
+ "grad_norm": 0.584650993347168,
+ "learning_rate": 6.848364281984843e-07,
+ "loss": 0.006,
+ "step": 32482
+ },
+ {
+ "epoch": 88.50953678474114,
+ "grad_norm": 0.8923815488815308,
+ "learning_rate": 6.845155016749006e-07,
+ "loss": 0.0123,
+ "step": 32483
+ },
+ {
+ "epoch": 88.51226158038148,
+ "grad_norm": 1.3140207529067993,
+ "learning_rate": 6.841946477000072e-07,
+ "loss": 0.0152,
+ "step": 32484
+ },
+ {
+ "epoch": 88.5149863760218,
+ "grad_norm": 1.3117071390151978,
+ "learning_rate": 6.838738662763089e-07,
+ "loss": 0.0334,
+ "step": 32485
+ },
+ {
+ "epoch": 88.51771117166213,
+ "grad_norm": 1.024350881576538,
+ "learning_rate": 6.835531574063003e-07,
+ "loss": 0.0299,
+ "step": 32486
+ },
+ {
+ "epoch": 88.52043596730245,
+ "grad_norm": 1.19551682472229,
+ "learning_rate": 6.832325210924806e-07,
+ "loss": 0.0116,
+ "step": 32487
+ },
+ {
+ "epoch": 88.52316076294278,
+ "grad_norm": 2.441592216491699,
+ "learning_rate": 6.829119573373443e-07,
+ "loss": 0.0738,
+ "step": 32488
+ },
+ {
+ "epoch": 88.52588555858311,
+ "grad_norm": 1.324022889137268,
+ "learning_rate": 6.825914661433908e-07,
+ "loss": 0.0598,
+ "step": 32489
+ },
+ {
+ "epoch": 88.52861035422343,
+ "grad_norm": 0.9229509830474854,
+ "learning_rate": 6.822710475131155e-07,
+ "loss": 0.013,
+ "step": 32490
+ },
+ {
+ "epoch": 88.53133514986376,
+ "grad_norm": 1.1370232105255127,
+ "learning_rate": 6.819507014490134e-07,
+ "loss": 0.0419,
+ "step": 32491
+ },
+ {
+ "epoch": 88.53405994550408,
+ "grad_norm": 0.9004497528076172,
+ "learning_rate": 6.81630427953578e-07,
+ "loss": 0.0113,
+ "step": 32492
+ },
+ {
+ "epoch": 88.53678474114442,
+ "grad_norm": 1.5294321775436401,
+ "learning_rate": 6.81310227029307e-07,
+ "loss": 0.0275,
+ "step": 32493
+ },
+ {
+ "epoch": 88.53950953678473,
+ "grad_norm": 1.1788089275360107,
+ "learning_rate": 6.809900986786889e-07,
+ "loss": 0.084,
+ "step": 32494
+ },
+ {
+ "epoch": 88.54223433242507,
+ "grad_norm": 1.0744454860687256,
+ "learning_rate": 6.806700429042223e-07,
+ "loss": 0.0066,
+ "step": 32495
+ },
+ {
+ "epoch": 88.5449591280654,
+ "grad_norm": 1.9215654134750366,
+ "learning_rate": 6.803500597083967e-07,
+ "loss": 0.0381,
+ "step": 32496
+ },
+ {
+ "epoch": 88.54768392370572,
+ "grad_norm": 0.9686970114707947,
+ "learning_rate": 6.800301490937056e-07,
+ "loss": 0.0128,
+ "step": 32497
+ },
+ {
+ "epoch": 88.55040871934605,
+ "grad_norm": 1.5334423780441284,
+ "learning_rate": 6.797103110626402e-07,
+ "loss": 0.0122,
+ "step": 32498
+ },
+ {
+ "epoch": 88.55313351498637,
+ "grad_norm": 1.6648393869400024,
+ "learning_rate": 6.793905456176909e-07,
+ "loss": 0.0264,
+ "step": 32499
+ },
+ {
+ "epoch": 88.5558583106267,
+ "grad_norm": 1.1064209938049316,
+ "learning_rate": 6.790708527613454e-07,
+ "loss": 0.0258,
+ "step": 32500
+ },
+ {
+ "epoch": 88.55858310626704,
+ "grad_norm": 0.9832415580749512,
+ "learning_rate": 6.787512324960987e-07,
+ "loss": 0.0111,
+ "step": 32501
+ },
+ {
+ "epoch": 88.56130790190736,
+ "grad_norm": 0.5715138912200928,
+ "learning_rate": 6.784316848244377e-07,
+ "loss": 0.0063,
+ "step": 32502
+ },
+ {
+ "epoch": 88.56403269754769,
+ "grad_norm": 0.49376380443573,
+ "learning_rate": 6.781122097488501e-07,
+ "loss": 0.004,
+ "step": 32503
+ },
+ {
+ "epoch": 88.566757493188,
+ "grad_norm": 1.5788899660110474,
+ "learning_rate": 6.777928072718232e-07,
+ "loss": 0.0389,
+ "step": 32504
+ },
+ {
+ "epoch": 88.56948228882834,
+ "grad_norm": 1.874849557876587,
+ "learning_rate": 6.774734773958469e-07,
+ "loss": 0.032,
+ "step": 32505
+ },
+ {
+ "epoch": 88.57220708446866,
+ "grad_norm": 1.149145483970642,
+ "learning_rate": 6.771542201234071e-07,
+ "loss": 0.0397,
+ "step": 32506
+ },
+ {
+ "epoch": 88.57493188010899,
+ "grad_norm": 0.3929470181465149,
+ "learning_rate": 6.768350354569909e-07,
+ "loss": 0.0038,
+ "step": 32507
+ },
+ {
+ "epoch": 88.57765667574932,
+ "grad_norm": 0.7169041633605957,
+ "learning_rate": 6.765159233990814e-07,
+ "loss": 0.0061,
+ "step": 32508
+ },
+ {
+ "epoch": 88.58038147138964,
+ "grad_norm": 0.923984169960022,
+ "learning_rate": 6.761968839521682e-07,
+ "loss": 0.0096,
+ "step": 32509
+ },
+ {
+ "epoch": 88.58310626702998,
+ "grad_norm": 0.7973461151123047,
+ "learning_rate": 6.758779171187324e-07,
+ "loss": 0.0082,
+ "step": 32510
+ },
+ {
+ "epoch": 88.5858310626703,
+ "grad_norm": 1.1363245248794556,
+ "learning_rate": 6.755590229012588e-07,
+ "loss": 0.0265,
+ "step": 32511
+ },
+ {
+ "epoch": 88.58855585831063,
+ "grad_norm": 1.5156110525131226,
+ "learning_rate": 6.752402013022319e-07,
+ "loss": 0.0619,
+ "step": 32512
+ },
+ {
+ "epoch": 88.59128065395096,
+ "grad_norm": 0.9433643817901611,
+ "learning_rate": 6.749214523241332e-07,
+ "loss": 0.0166,
+ "step": 32513
+ },
+ {
+ "epoch": 88.59400544959128,
+ "grad_norm": 1.5971944332122803,
+ "learning_rate": 6.746027759694474e-07,
+ "loss": 0.1094,
+ "step": 32514
+ },
+ {
+ "epoch": 88.59673024523161,
+ "grad_norm": 0.7449738383293152,
+ "learning_rate": 6.742841722406556e-07,
+ "loss": 0.0048,
+ "step": 32515
+ },
+ {
+ "epoch": 88.59945504087193,
+ "grad_norm": 1.5191032886505127,
+ "learning_rate": 6.739656411402363e-07,
+ "loss": 0.0261,
+ "step": 32516
+ },
+ {
+ "epoch": 88.60217983651226,
+ "grad_norm": 0.5808982253074646,
+ "learning_rate": 6.736471826706747e-07,
+ "loss": 0.0037,
+ "step": 32517
+ },
+ {
+ "epoch": 88.60490463215258,
+ "grad_norm": 1.2449208498001099,
+ "learning_rate": 6.733287968344482e-07,
+ "loss": 0.0262,
+ "step": 32518
+ },
+ {
+ "epoch": 88.60762942779292,
+ "grad_norm": 0.8039193749427795,
+ "learning_rate": 6.730104836340367e-07,
+ "loss": 0.0103,
+ "step": 32519
+ },
+ {
+ "epoch": 88.61035422343325,
+ "grad_norm": 1.5093945264816284,
+ "learning_rate": 6.726922430719185e-07,
+ "loss": 0.0237,
+ "step": 32520
+ },
+ {
+ "epoch": 88.61307901907357,
+ "grad_norm": 1.625341534614563,
+ "learning_rate": 6.723740751505736e-07,
+ "loss": 0.022,
+ "step": 32521
+ },
+ {
+ "epoch": 88.6158038147139,
+ "grad_norm": 1.1639204025268555,
+ "learning_rate": 6.720559798724801e-07,
+ "loss": 0.0085,
+ "step": 32522
+ },
+ {
+ "epoch": 88.61852861035422,
+ "grad_norm": 0.9657247066497803,
+ "learning_rate": 6.717379572401139e-07,
+ "loss": 0.0124,
+ "step": 32523
+ },
+ {
+ "epoch": 88.62125340599455,
+ "grad_norm": 1.9073123931884766,
+ "learning_rate": 6.714200072559507e-07,
+ "loss": 0.0148,
+ "step": 32524
+ },
+ {
+ "epoch": 88.62397820163488,
+ "grad_norm": 1.1114286184310913,
+ "learning_rate": 6.711021299224696e-07,
+ "loss": 0.0076,
+ "step": 32525
+ },
+ {
+ "epoch": 88.6267029972752,
+ "grad_norm": 1.1664174795150757,
+ "learning_rate": 6.707843252421453e-07,
+ "loss": 0.0709,
+ "step": 32526
+ },
+ {
+ "epoch": 88.62942779291554,
+ "grad_norm": 1.1559926271438599,
+ "learning_rate": 6.704665932174525e-07,
+ "loss": 0.0504,
+ "step": 32527
+ },
+ {
+ "epoch": 88.63215258855585,
+ "grad_norm": 0.7560103535652161,
+ "learning_rate": 6.701489338508638e-07,
+ "loss": 0.0099,
+ "step": 32528
+ },
+ {
+ "epoch": 88.63487738419619,
+ "grad_norm": 1.0639030933380127,
+ "learning_rate": 6.698313471448547e-07,
+ "loss": 0.0503,
+ "step": 32529
+ },
+ {
+ "epoch": 88.6376021798365,
+ "grad_norm": 1.0624189376831055,
+ "learning_rate": 6.695138331019002e-07,
+ "loss": 0.1018,
+ "step": 32530
+ },
+ {
+ "epoch": 88.64032697547684,
+ "grad_norm": 1.503642201423645,
+ "learning_rate": 6.691963917244726e-07,
+ "loss": 0.0291,
+ "step": 32531
+ },
+ {
+ "epoch": 88.64305177111717,
+ "grad_norm": 0.8536162972450256,
+ "learning_rate": 6.688790230150421e-07,
+ "loss": 0.0083,
+ "step": 32532
+ },
+ {
+ "epoch": 88.64577656675749,
+ "grad_norm": 1.0311760902404785,
+ "learning_rate": 6.685617269760802e-07,
+ "loss": 0.0084,
+ "step": 32533
+ },
+ {
+ "epoch": 88.64850136239782,
+ "grad_norm": 1.3529002666473389,
+ "learning_rate": 6.682445036100616e-07,
+ "loss": 0.0227,
+ "step": 32534
+ },
+ {
+ "epoch": 88.65122615803814,
+ "grad_norm": 0.5649893879890442,
+ "learning_rate": 6.679273529194541e-07,
+ "loss": 0.0057,
+ "step": 32535
+ },
+ {
+ "epoch": 88.65395095367847,
+ "grad_norm": 1.3361003398895264,
+ "learning_rate": 6.67610274906727e-07,
+ "loss": 0.0279,
+ "step": 32536
+ },
+ {
+ "epoch": 88.65667574931881,
+ "grad_norm": 1.4460997581481934,
+ "learning_rate": 6.672932695743506e-07,
+ "loss": 0.009,
+ "step": 32537
+ },
+ {
+ "epoch": 88.65940054495913,
+ "grad_norm": 1.1239690780639648,
+ "learning_rate": 6.669763369247939e-07,
+ "loss": 0.0212,
+ "step": 32538
+ },
+ {
+ "epoch": 88.66212534059946,
+ "grad_norm": 0.3101179599761963,
+ "learning_rate": 6.666594769605261e-07,
+ "loss": 0.0035,
+ "step": 32539
+ },
+ {
+ "epoch": 88.66485013623978,
+ "grad_norm": 0.9236528873443604,
+ "learning_rate": 6.66342689684012e-07,
+ "loss": 0.0097,
+ "step": 32540
+ },
+ {
+ "epoch": 88.66757493188011,
+ "grad_norm": 0.6529943346977234,
+ "learning_rate": 6.660259750977227e-07,
+ "loss": 0.006,
+ "step": 32541
+ },
+ {
+ "epoch": 88.67029972752043,
+ "grad_norm": 1.2716357707977295,
+ "learning_rate": 6.65709333204122e-07,
+ "loss": 0.0133,
+ "step": 32542
+ },
+ {
+ "epoch": 88.67302452316076,
+ "grad_norm": 1.2861262559890747,
+ "learning_rate": 6.653927640056779e-07,
+ "loss": 0.0136,
+ "step": 32543
+ },
+ {
+ "epoch": 88.6757493188011,
+ "grad_norm": 1.5994800329208374,
+ "learning_rate": 6.650762675048516e-07,
+ "loss": 0.0184,
+ "step": 32544
+ },
+ {
+ "epoch": 88.67847411444141,
+ "grad_norm": 0.8209812045097351,
+ "learning_rate": 6.647598437041114e-07,
+ "loss": 0.0078,
+ "step": 32545
+ },
+ {
+ "epoch": 88.68119891008175,
+ "grad_norm": 0.9815989136695862,
+ "learning_rate": 6.644434926059228e-07,
+ "loss": 0.0097,
+ "step": 32546
+ },
+ {
+ "epoch": 88.68392370572207,
+ "grad_norm": 1.2249196767807007,
+ "learning_rate": 6.641272142127475e-07,
+ "loss": 0.0158,
+ "step": 32547
+ },
+ {
+ "epoch": 88.6866485013624,
+ "grad_norm": 0.5456929206848145,
+ "learning_rate": 6.638110085270488e-07,
+ "loss": 0.0047,
+ "step": 32548
+ },
+ {
+ "epoch": 88.68937329700273,
+ "grad_norm": 0.8484465479850769,
+ "learning_rate": 6.634948755512882e-07,
+ "loss": 0.011,
+ "step": 32549
+ },
+ {
+ "epoch": 88.69209809264305,
+ "grad_norm": 1.5140070915222168,
+ "learning_rate": 6.631788152879304e-07,
+ "loss": 0.0408,
+ "step": 32550
+ },
+ {
+ "epoch": 88.69482288828338,
+ "grad_norm": 1.4155182838439941,
+ "learning_rate": 6.628628277394356e-07,
+ "loss": 0.0186,
+ "step": 32551
+ },
+ {
+ "epoch": 88.6975476839237,
+ "grad_norm": 2.3508574962615967,
+ "learning_rate": 6.625469129082651e-07,
+ "loss": 0.0555,
+ "step": 32552
+ },
+ {
+ "epoch": 88.70027247956403,
+ "grad_norm": 0.8693823218345642,
+ "learning_rate": 6.62231070796876e-07,
+ "loss": 0.0377,
+ "step": 32553
+ },
+ {
+ "epoch": 88.70299727520435,
+ "grad_norm": 1.1615219116210938,
+ "learning_rate": 6.619153014077328e-07,
+ "loss": 0.0158,
+ "step": 32554
+ },
+ {
+ "epoch": 88.70572207084469,
+ "grad_norm": 2.125788688659668,
+ "learning_rate": 6.615996047432937e-07,
+ "loss": 0.0349,
+ "step": 32555
+ },
+ {
+ "epoch": 88.70844686648502,
+ "grad_norm": 1.3906364440917969,
+ "learning_rate": 6.612839808060146e-07,
+ "loss": 0.0188,
+ "step": 32556
+ },
+ {
+ "epoch": 88.71117166212534,
+ "grad_norm": 1.1571823358535767,
+ "learning_rate": 6.609684295983553e-07,
+ "loss": 0.0118,
+ "step": 32557
+ },
+ {
+ "epoch": 88.71389645776567,
+ "grad_norm": 0.8745749592781067,
+ "learning_rate": 6.606529511227743e-07,
+ "loss": 0.0152,
+ "step": 32558
+ },
+ {
+ "epoch": 88.71662125340599,
+ "grad_norm": 1.4224164485931396,
+ "learning_rate": 6.603375453817273e-07,
+ "loss": 0.0258,
+ "step": 32559
+ },
+ {
+ "epoch": 88.71934604904632,
+ "grad_norm": 0.8090235590934753,
+ "learning_rate": 6.600222123776712e-07,
+ "loss": 0.0093,
+ "step": 32560
+ },
+ {
+ "epoch": 88.72207084468666,
+ "grad_norm": 1.2053598165512085,
+ "learning_rate": 6.597069521130594e-07,
+ "loss": 0.0191,
+ "step": 32561
+ },
+ {
+ "epoch": 88.72479564032697,
+ "grad_norm": 1.0805208683013916,
+ "learning_rate": 6.593917645903491e-07,
+ "loss": 0.0356,
+ "step": 32562
+ },
+ {
+ "epoch": 88.7275204359673,
+ "grad_norm": 0.8014301657676697,
+ "learning_rate": 6.590766498119983e-07,
+ "loss": 0.0057,
+ "step": 32563
+ },
+ {
+ "epoch": 88.73024523160763,
+ "grad_norm": 1.3241361379623413,
+ "learning_rate": 6.58761607780456e-07,
+ "loss": 0.0191,
+ "step": 32564
+ },
+ {
+ "epoch": 88.73297002724796,
+ "grad_norm": 0.5379200577735901,
+ "learning_rate": 6.584466384981769e-07,
+ "loss": 0.0045,
+ "step": 32565
+ },
+ {
+ "epoch": 88.73569482288828,
+ "grad_norm": 1.6858173608779907,
+ "learning_rate": 6.58131741967617e-07,
+ "loss": 0.0566,
+ "step": 32566
+ },
+ {
+ "epoch": 88.73841961852861,
+ "grad_norm": 0.8219602704048157,
+ "learning_rate": 6.578169181912252e-07,
+ "loss": 0.0086,
+ "step": 32567
+ },
+ {
+ "epoch": 88.74114441416894,
+ "grad_norm": 0.6108104586601257,
+ "learning_rate": 6.575021671714554e-07,
+ "loss": 0.0065,
+ "step": 32568
+ },
+ {
+ "epoch": 88.74386920980926,
+ "grad_norm": 0.8748952746391296,
+ "learning_rate": 6.571874889107554e-07,
+ "loss": 0.0061,
+ "step": 32569
+ },
+ {
+ "epoch": 88.7465940054496,
+ "grad_norm": 1.345271110534668,
+ "learning_rate": 6.568728834115812e-07,
+ "loss": 0.0098,
+ "step": 32570
+ },
+ {
+ "epoch": 88.74931880108991,
+ "grad_norm": 1.0289160013198853,
+ "learning_rate": 6.565583506763795e-07,
+ "loss": 0.0095,
+ "step": 32571
+ },
+ {
+ "epoch": 88.75204359673025,
+ "grad_norm": 1.1375980377197266,
+ "learning_rate": 6.562438907076018e-07,
+ "loss": 0.0117,
+ "step": 32572
+ },
+ {
+ "epoch": 88.75476839237058,
+ "grad_norm": 1.1102759838104248,
+ "learning_rate": 6.559295035076929e-07,
+ "loss": 0.0158,
+ "step": 32573
+ },
+ {
+ "epoch": 88.7574931880109,
+ "grad_norm": 1.4043065309524536,
+ "learning_rate": 6.556151890791073e-07,
+ "loss": 0.0108,
+ "step": 32574
+ },
+ {
+ "epoch": 88.76021798365123,
+ "grad_norm": 0.8455353379249573,
+ "learning_rate": 6.553009474242888e-07,
+ "loss": 0.0069,
+ "step": 32575
+ },
+ {
+ "epoch": 88.76294277929155,
+ "grad_norm": 0.8111148476600647,
+ "learning_rate": 6.549867785456864e-07,
+ "loss": 0.0752,
+ "step": 32576
+ },
+ {
+ "epoch": 88.76566757493188,
+ "grad_norm": 1.6576833724975586,
+ "learning_rate": 6.546726824457438e-07,
+ "loss": 0.0313,
+ "step": 32577
+ },
+ {
+ "epoch": 88.7683923705722,
+ "grad_norm": 1.166200876235962,
+ "learning_rate": 6.543586591269124e-07,
+ "loss": 0.0596,
+ "step": 32578
+ },
+ {
+ "epoch": 88.77111716621253,
+ "grad_norm": 0.93400639295578,
+ "learning_rate": 6.540447085916346e-07,
+ "loss": 0.0169,
+ "step": 32579
+ },
+ {
+ "epoch": 88.77384196185287,
+ "grad_norm": 0.8460935950279236,
+ "learning_rate": 6.537308308423529e-07,
+ "loss": 0.0101,
+ "step": 32580
+ },
+ {
+ "epoch": 88.77656675749319,
+ "grad_norm": 1.4068222045898438,
+ "learning_rate": 6.534170258815165e-07,
+ "loss": 0.0911,
+ "step": 32581
+ },
+ {
+ "epoch": 88.77929155313352,
+ "grad_norm": 1.4459940195083618,
+ "learning_rate": 6.531032937115677e-07,
+ "loss": 0.0083,
+ "step": 32582
+ },
+ {
+ "epoch": 88.78201634877384,
+ "grad_norm": 1.315310001373291,
+ "learning_rate": 6.527896343349505e-07,
+ "loss": 0.1833,
+ "step": 32583
+ },
+ {
+ "epoch": 88.78474114441417,
+ "grad_norm": 0.5295960307121277,
+ "learning_rate": 6.52476047754107e-07,
+ "loss": 0.0045,
+ "step": 32584
+ },
+ {
+ "epoch": 88.7874659400545,
+ "grad_norm": 1.5018444061279297,
+ "learning_rate": 6.521625339714765e-07,
+ "loss": 0.0593,
+ "step": 32585
+ },
+ {
+ "epoch": 88.79019073569482,
+ "grad_norm": 0.990424633026123,
+ "learning_rate": 6.51849092989506e-07,
+ "loss": 0.0146,
+ "step": 32586
+ },
+ {
+ "epoch": 88.79291553133515,
+ "grad_norm": 1.4484387636184692,
+ "learning_rate": 6.515357248106325e-07,
+ "loss": 0.0958,
+ "step": 32587
+ },
+ {
+ "epoch": 88.79564032697547,
+ "grad_norm": 0.8569030165672302,
+ "learning_rate": 6.512224294372993e-07,
+ "loss": 0.0083,
+ "step": 32588
+ },
+ {
+ "epoch": 88.7983651226158,
+ "grad_norm": 1.0415432453155518,
+ "learning_rate": 6.509092068719425e-07,
+ "loss": 0.0074,
+ "step": 32589
+ },
+ {
+ "epoch": 88.80108991825612,
+ "grad_norm": 1.3801058530807495,
+ "learning_rate": 6.505960571170056e-07,
+ "loss": 0.0154,
+ "step": 32590
+ },
+ {
+ "epoch": 88.80381471389646,
+ "grad_norm": 1.3749128580093384,
+ "learning_rate": 6.502829801749266e-07,
+ "loss": 0.0467,
+ "step": 32591
+ },
+ {
+ "epoch": 88.80653950953679,
+ "grad_norm": 1.4173424243927002,
+ "learning_rate": 6.499699760481415e-07,
+ "loss": 0.0235,
+ "step": 32592
+ },
+ {
+ "epoch": 88.80926430517711,
+ "grad_norm": 0.6067011952400208,
+ "learning_rate": 6.496570447390893e-07,
+ "loss": 0.0049,
+ "step": 32593
+ },
+ {
+ "epoch": 88.81198910081744,
+ "grad_norm": 0.883727490901947,
+ "learning_rate": 6.49344186250207e-07,
+ "loss": 0.0064,
+ "step": 32594
+ },
+ {
+ "epoch": 88.81471389645776,
+ "grad_norm": 1.0084599256515503,
+ "learning_rate": 6.490314005839327e-07,
+ "loss": 0.0327,
+ "step": 32595
+ },
+ {
+ "epoch": 88.8174386920981,
+ "grad_norm": 1.0601624250411987,
+ "learning_rate": 6.487186877426998e-07,
+ "loss": 0.0113,
+ "step": 32596
+ },
+ {
+ "epoch": 88.82016348773843,
+ "grad_norm": 0.5102577805519104,
+ "learning_rate": 6.484060477289433e-07,
+ "loss": 0.0051,
+ "step": 32597
+ },
+ {
+ "epoch": 88.82288828337875,
+ "grad_norm": 0.887415885925293,
+ "learning_rate": 6.480934805451001e-07,
+ "loss": 0.016,
+ "step": 32598
+ },
+ {
+ "epoch": 88.82561307901908,
+ "grad_norm": 0.7867955565452576,
+ "learning_rate": 6.477809861936047e-07,
+ "loss": 0.0072,
+ "step": 32599
+ },
+ {
+ "epoch": 88.8283378746594,
+ "grad_norm": 1.2387006282806396,
+ "learning_rate": 6.474685646768908e-07,
+ "loss": 0.0492,
+ "step": 32600
+ },
+ {
+ "epoch": 88.83106267029973,
+ "grad_norm": 0.5061392188072205,
+ "learning_rate": 6.471562159973899e-07,
+ "loss": 0.0056,
+ "step": 32601
+ },
+ {
+ "epoch": 88.83378746594005,
+ "grad_norm": 2.193077802658081,
+ "learning_rate": 6.468439401575344e-07,
+ "loss": 0.0625,
+ "step": 32602
+ },
+ {
+ "epoch": 88.83651226158038,
+ "grad_norm": 0.7096750140190125,
+ "learning_rate": 6.46531737159759e-07,
+ "loss": 0.0059,
+ "step": 32603
+ },
+ {
+ "epoch": 88.83923705722071,
+ "grad_norm": 1.327993631362915,
+ "learning_rate": 6.46219607006493e-07,
+ "loss": 0.0121,
+ "step": 32604
+ },
+ {
+ "epoch": 88.84196185286103,
+ "grad_norm": 1.225622534751892,
+ "learning_rate": 6.459075497001655e-07,
+ "loss": 0.0125,
+ "step": 32605
+ },
+ {
+ "epoch": 88.84468664850137,
+ "grad_norm": 0.39478856325149536,
+ "learning_rate": 6.455955652432111e-07,
+ "loss": 0.0041,
+ "step": 32606
+ },
+ {
+ "epoch": 88.84741144414168,
+ "grad_norm": 1.039612054824829,
+ "learning_rate": 6.45283653638058e-07,
+ "loss": 0.0158,
+ "step": 32607
+ },
+ {
+ "epoch": 88.85013623978202,
+ "grad_norm": 1.4335986375808716,
+ "learning_rate": 6.449718148871342e-07,
+ "loss": 0.0848,
+ "step": 32608
+ },
+ {
+ "epoch": 88.85286103542235,
+ "grad_norm": 1.3314348459243774,
+ "learning_rate": 6.446600489928667e-07,
+ "loss": 0.0221,
+ "step": 32609
+ },
+ {
+ "epoch": 88.85558583106267,
+ "grad_norm": 0.8160584568977356,
+ "learning_rate": 6.443483559576879e-07,
+ "loss": 0.0111,
+ "step": 32610
+ },
+ {
+ "epoch": 88.858310626703,
+ "grad_norm": 0.9039756059646606,
+ "learning_rate": 6.440367357840228e-07,
+ "loss": 0.0076,
+ "step": 32611
+ },
+ {
+ "epoch": 88.86103542234332,
+ "grad_norm": 1.2229604721069336,
+ "learning_rate": 6.43725188474299e-07,
+ "loss": 0.0099,
+ "step": 32612
+ },
+ {
+ "epoch": 88.86376021798365,
+ "grad_norm": 0.8536512851715088,
+ "learning_rate": 6.434137140309393e-07,
+ "loss": 0.0356,
+ "step": 32613
+ },
+ {
+ "epoch": 88.86648501362397,
+ "grad_norm": 1.0948132276535034,
+ "learning_rate": 6.43102312456374e-07,
+ "loss": 0.0112,
+ "step": 32614
+ },
+ {
+ "epoch": 88.8692098092643,
+ "grad_norm": 0.5389711260795593,
+ "learning_rate": 6.427909837530278e-07,
+ "loss": 0.0054,
+ "step": 32615
+ },
+ {
+ "epoch": 88.87193460490464,
+ "grad_norm": 2.113151788711548,
+ "learning_rate": 6.424797279233242e-07,
+ "loss": 0.0179,
+ "step": 32616
+ },
+ {
+ "epoch": 88.87465940054496,
+ "grad_norm": 0.8863828182220459,
+ "learning_rate": 6.421685449696868e-07,
+ "loss": 0.0084,
+ "step": 32617
+ },
+ {
+ "epoch": 88.87738419618529,
+ "grad_norm": 0.942398190498352,
+ "learning_rate": 6.418574348945384e-07,
+ "loss": 0.0126,
+ "step": 32618
+ },
+ {
+ "epoch": 88.88010899182561,
+ "grad_norm": 1.5410829782485962,
+ "learning_rate": 6.415463977003045e-07,
+ "loss": 0.0761,
+ "step": 32619
+ },
+ {
+ "epoch": 88.88283378746594,
+ "grad_norm": 1.1717145442962646,
+ "learning_rate": 6.412354333894067e-07,
+ "loss": 0.007,
+ "step": 32620
+ },
+ {
+ "epoch": 88.88555858310627,
+ "grad_norm": 0.7939753532409668,
+ "learning_rate": 6.409245419642663e-07,
+ "loss": 0.0048,
+ "step": 32621
+ },
+ {
+ "epoch": 88.88828337874659,
+ "grad_norm": 0.8811405897140503,
+ "learning_rate": 6.406137234273013e-07,
+ "loss": 0.0287,
+ "step": 32622
+ },
+ {
+ "epoch": 88.89100817438693,
+ "grad_norm": 1.3336745500564575,
+ "learning_rate": 6.403029777809377e-07,
+ "loss": 0.039,
+ "step": 32623
+ },
+ {
+ "epoch": 88.89373297002724,
+ "grad_norm": 1.2989461421966553,
+ "learning_rate": 6.399923050275935e-07,
+ "loss": 0.0078,
+ "step": 32624
+ },
+ {
+ "epoch": 88.89645776566758,
+ "grad_norm": 2.077906370162964,
+ "learning_rate": 6.396817051696869e-07,
+ "loss": 0.0292,
+ "step": 32625
+ },
+ {
+ "epoch": 88.8991825613079,
+ "grad_norm": 0.5503972172737122,
+ "learning_rate": 6.39371178209639e-07,
+ "loss": 0.0079,
+ "step": 32626
+ },
+ {
+ "epoch": 88.90190735694823,
+ "grad_norm": 1.385947823524475,
+ "learning_rate": 6.39060724149867e-07,
+ "loss": 0.0219,
+ "step": 32627
+ },
+ {
+ "epoch": 88.90463215258856,
+ "grad_norm": 1.2694889307022095,
+ "learning_rate": 6.387503429927888e-07,
+ "loss": 0.0572,
+ "step": 32628
+ },
+ {
+ "epoch": 88.90735694822888,
+ "grad_norm": 0.9886922240257263,
+ "learning_rate": 6.384400347408204e-07,
+ "loss": 0.0075,
+ "step": 32629
+ },
+ {
+ "epoch": 88.91008174386921,
+ "grad_norm": 1.2462801933288574,
+ "learning_rate": 6.381297993963809e-07,
+ "loss": 0.0649,
+ "step": 32630
+ },
+ {
+ "epoch": 88.91280653950953,
+ "grad_norm": 0.4829780161380768,
+ "learning_rate": 6.378196369618828e-07,
+ "loss": 0.0088,
+ "step": 32631
+ },
+ {
+ "epoch": 88.91553133514986,
+ "grad_norm": 1.4472674131393433,
+ "learning_rate": 6.375095474397475e-07,
+ "loss": 0.0387,
+ "step": 32632
+ },
+ {
+ "epoch": 88.9182561307902,
+ "grad_norm": 1.0596401691436768,
+ "learning_rate": 6.371995308323853e-07,
+ "loss": 0.0332,
+ "step": 32633
+ },
+ {
+ "epoch": 88.92098092643052,
+ "grad_norm": 0.4849154055118561,
+ "learning_rate": 6.368895871422099e-07,
+ "loss": 0.0054,
+ "step": 32634
+ },
+ {
+ "epoch": 88.92370572207085,
+ "grad_norm": 1.5603132247924805,
+ "learning_rate": 6.365797163716392e-07,
+ "loss": 0.0643,
+ "step": 32635
+ },
+ {
+ "epoch": 88.92643051771117,
+ "grad_norm": 0.9288502931594849,
+ "learning_rate": 6.362699185230848e-07,
+ "loss": 0.0105,
+ "step": 32636
+ },
+ {
+ "epoch": 88.9291553133515,
+ "grad_norm": 1.3014543056488037,
+ "learning_rate": 6.359601935989579e-07,
+ "loss": 0.0216,
+ "step": 32637
+ },
+ {
+ "epoch": 88.93188010899182,
+ "grad_norm": 0.7887872457504272,
+ "learning_rate": 6.356505416016712e-07,
+ "loss": 0.0078,
+ "step": 32638
+ },
+ {
+ "epoch": 88.93460490463215,
+ "grad_norm": 1.3281413316726685,
+ "learning_rate": 6.353409625336371e-07,
+ "loss": 0.087,
+ "step": 32639
+ },
+ {
+ "epoch": 88.93732970027249,
+ "grad_norm": 0.9513595700263977,
+ "learning_rate": 6.350314563972671e-07,
+ "loss": 0.133,
+ "step": 32640
+ },
+ {
+ "epoch": 88.9400544959128,
+ "grad_norm": 0.7080846428871155,
+ "learning_rate": 6.347220231949702e-07,
+ "loss": 0.0143,
+ "step": 32641
+ },
+ {
+ "epoch": 88.94277929155314,
+ "grad_norm": 1.349782943725586,
+ "learning_rate": 6.344126629291559e-07,
+ "loss": 0.0896,
+ "step": 32642
+ },
+ {
+ "epoch": 88.94550408719346,
+ "grad_norm": 3.53540301322937,
+ "learning_rate": 6.341033756022352e-07,
+ "loss": 0.0878,
+ "step": 32643
+ },
+ {
+ "epoch": 88.94822888283379,
+ "grad_norm": 0.5549563765525818,
+ "learning_rate": 6.337941612166154e-07,
+ "loss": 0.006,
+ "step": 32644
+ },
+ {
+ "epoch": 88.95095367847412,
+ "grad_norm": 0.7886529564857483,
+ "learning_rate": 6.334850197747066e-07,
+ "loss": 0.0072,
+ "step": 32645
+ },
+ {
+ "epoch": 88.95367847411444,
+ "grad_norm": 2.206639051437378,
+ "learning_rate": 6.331759512789126e-07,
+ "loss": 0.0389,
+ "step": 32646
+ },
+ {
+ "epoch": 88.95640326975477,
+ "grad_norm": 0.7893012762069702,
+ "learning_rate": 6.328669557316447e-07,
+ "loss": 0.0137,
+ "step": 32647
+ },
+ {
+ "epoch": 88.95912806539509,
+ "grad_norm": 0.635030210018158,
+ "learning_rate": 6.325580331353065e-07,
+ "loss": 0.0078,
+ "step": 32648
+ },
+ {
+ "epoch": 88.96185286103542,
+ "grad_norm": 0.9105276465415955,
+ "learning_rate": 6.32249183492305e-07,
+ "loss": 0.0067,
+ "step": 32649
+ },
+ {
+ "epoch": 88.96457765667574,
+ "grad_norm": 1.0956720113754272,
+ "learning_rate": 6.319404068050439e-07,
+ "loss": 0.0186,
+ "step": 32650
+ },
+ {
+ "epoch": 88.96730245231608,
+ "grad_norm": 1.3061813116073608,
+ "learning_rate": 6.316317030759322e-07,
+ "loss": 0.0192,
+ "step": 32651
+ },
+ {
+ "epoch": 88.97002724795641,
+ "grad_norm": 0.8634379506111145,
+ "learning_rate": 6.313230723073704e-07,
+ "loss": 0.0106,
+ "step": 32652
+ },
+ {
+ "epoch": 88.97275204359673,
+ "grad_norm": 1.3323620557785034,
+ "learning_rate": 6.310145145017633e-07,
+ "loss": 0.0124,
+ "step": 32653
+ },
+ {
+ "epoch": 88.97547683923706,
+ "grad_norm": 1.1698510646820068,
+ "learning_rate": 6.30706029661512e-07,
+ "loss": 0.0099,
+ "step": 32654
+ },
+ {
+ "epoch": 88.97820163487738,
+ "grad_norm": 1.0633735656738281,
+ "learning_rate": 6.303976177890214e-07,
+ "loss": 0.0189,
+ "step": 32655
+ },
+ {
+ "epoch": 88.98092643051771,
+ "grad_norm": 1.2730908393859863,
+ "learning_rate": 6.300892788866942e-07,
+ "loss": 0.0166,
+ "step": 32656
+ },
+ {
+ "epoch": 88.98365122615803,
+ "grad_norm": 0.9520397782325745,
+ "learning_rate": 6.297810129569282e-07,
+ "loss": 0.0179,
+ "step": 32657
+ },
+ {
+ "epoch": 88.98637602179836,
+ "grad_norm": 0.4920234978199005,
+ "learning_rate": 6.294728200021261e-07,
+ "loss": 0.0045,
+ "step": 32658
+ },
+ {
+ "epoch": 88.9891008174387,
+ "grad_norm": 0.9644466042518616,
+ "learning_rate": 6.291647000246892e-07,
+ "loss": 0.015,
+ "step": 32659
+ },
+ {
+ "epoch": 88.99182561307902,
+ "grad_norm": 1.4101142883300781,
+ "learning_rate": 6.288566530270157e-07,
+ "loss": 0.0543,
+ "step": 32660
+ },
+ {
+ "epoch": 88.99455040871935,
+ "grad_norm": 1.024294137954712,
+ "learning_rate": 6.285486790115047e-07,
+ "loss": 0.009,
+ "step": 32661
+ },
+ {
+ "epoch": 88.99727520435967,
+ "grad_norm": 0.899195671081543,
+ "learning_rate": 6.282407779805544e-07,
+ "loss": 0.1435,
+ "step": 32662
+ },
+ {
+ "epoch": 89.0,
+ "grad_norm": 1.0124510526657104,
+ "learning_rate": 6.279329499365649e-07,
+ "loss": 0.0091,
+ "step": 32663
+ },
+ {
+ "epoch": 89.00272479564033,
+ "grad_norm": 1.1268147230148315,
+ "learning_rate": 6.276251948819312e-07,
+ "loss": 0.1212,
+ "step": 32664
+ },
+ {
+ "epoch": 89.00544959128065,
+ "grad_norm": 1.1304576396942139,
+ "learning_rate": 6.273175128190512e-07,
+ "loss": 0.073,
+ "step": 32665
+ },
+ {
+ "epoch": 89.00817438692098,
+ "grad_norm": 1.3921688795089722,
+ "learning_rate": 6.270099037503185e-07,
+ "loss": 0.0123,
+ "step": 32666
+ },
+ {
+ "epoch": 89.0108991825613,
+ "grad_norm": 1.9878745079040527,
+ "learning_rate": 6.267023676781314e-07,
+ "loss": 0.0172,
+ "step": 32667
+ },
+ {
+ "epoch": 89.01362397820164,
+ "grad_norm": 1.0848573446273804,
+ "learning_rate": 6.263949046048868e-07,
+ "loss": 0.1093,
+ "step": 32668
+ },
+ {
+ "epoch": 89.01634877384195,
+ "grad_norm": 1.252785563468933,
+ "learning_rate": 6.260875145329759e-07,
+ "loss": 0.0103,
+ "step": 32669
+ },
+ {
+ "epoch": 89.01907356948229,
+ "grad_norm": 0.7286538481712341,
+ "learning_rate": 6.257801974647937e-07,
+ "loss": 0.0106,
+ "step": 32670
+ },
+ {
+ "epoch": 89.02179836512262,
+ "grad_norm": 1.3837857246398926,
+ "learning_rate": 6.254729534027337e-07,
+ "loss": 0.0484,
+ "step": 32671
+ },
+ {
+ "epoch": 89.02452316076294,
+ "grad_norm": 1.2221788167953491,
+ "learning_rate": 6.251657823491897e-07,
+ "loss": 0.0157,
+ "step": 32672
+ },
+ {
+ "epoch": 89.02724795640327,
+ "grad_norm": 1.295993685722351,
+ "learning_rate": 6.248586843065518e-07,
+ "loss": 0.0108,
+ "step": 32673
+ },
+ {
+ "epoch": 89.02997275204359,
+ "grad_norm": 0.9001659750938416,
+ "learning_rate": 6.245516592772116e-07,
+ "loss": 0.0067,
+ "step": 32674
+ },
+ {
+ "epoch": 89.03269754768392,
+ "grad_norm": 1.9949092864990234,
+ "learning_rate": 6.242447072635626e-07,
+ "loss": 0.0727,
+ "step": 32675
+ },
+ {
+ "epoch": 89.03542234332426,
+ "grad_norm": 0.9971086978912354,
+ "learning_rate": 6.239378282679953e-07,
+ "loss": 0.0074,
+ "step": 32676
+ },
+ {
+ "epoch": 89.03814713896458,
+ "grad_norm": 0.761946976184845,
+ "learning_rate": 6.236310222928977e-07,
+ "loss": 0.0055,
+ "step": 32677
+ },
+ {
+ "epoch": 89.04087193460491,
+ "grad_norm": 0.8279271721839905,
+ "learning_rate": 6.233242893406577e-07,
+ "loss": 0.0085,
+ "step": 32678
+ },
+ {
+ "epoch": 89.04359673024523,
+ "grad_norm": 0.9189581274986267,
+ "learning_rate": 6.23017629413668e-07,
+ "loss": 0.0121,
+ "step": 32679
+ },
+ {
+ "epoch": 89.04632152588556,
+ "grad_norm": 1.2925187349319458,
+ "learning_rate": 6.227110425143157e-07,
+ "loss": 0.0136,
+ "step": 32680
+ },
+ {
+ "epoch": 89.04904632152588,
+ "grad_norm": 0.6309067606925964,
+ "learning_rate": 6.224045286449887e-07,
+ "loss": 0.0078,
+ "step": 32681
+ },
+ {
+ "epoch": 89.05177111716621,
+ "grad_norm": 0.7074931263923645,
+ "learning_rate": 6.220980878080707e-07,
+ "loss": 0.0112,
+ "step": 32682
+ },
+ {
+ "epoch": 89.05449591280654,
+ "grad_norm": 2.1272318363189697,
+ "learning_rate": 6.217917200059509e-07,
+ "loss": 0.0431,
+ "step": 32683
+ },
+ {
+ "epoch": 89.05722070844686,
+ "grad_norm": 1.3618755340576172,
+ "learning_rate": 6.214854252410174e-07,
+ "loss": 0.0106,
+ "step": 32684
+ },
+ {
+ "epoch": 89.0599455040872,
+ "grad_norm": 0.9296203255653381,
+ "learning_rate": 6.211792035156539e-07,
+ "loss": 0.0071,
+ "step": 32685
+ },
+ {
+ "epoch": 89.06267029972751,
+ "grad_norm": 1.0491554737091064,
+ "learning_rate": 6.208730548322439e-07,
+ "loss": 0.0061,
+ "step": 32686
+ },
+ {
+ "epoch": 89.06539509536785,
+ "grad_norm": 0.9420356750488281,
+ "learning_rate": 6.205669791931723e-07,
+ "loss": 0.0082,
+ "step": 32687
+ },
+ {
+ "epoch": 89.06811989100818,
+ "grad_norm": 1.9627712965011597,
+ "learning_rate": 6.202609766008239e-07,
+ "loss": 0.0112,
+ "step": 32688
+ },
+ {
+ "epoch": 89.0708446866485,
+ "grad_norm": 1.0849237442016602,
+ "learning_rate": 6.199550470575821e-07,
+ "loss": 0.0157,
+ "step": 32689
+ },
+ {
+ "epoch": 89.07356948228883,
+ "grad_norm": 1.838118314743042,
+ "learning_rate": 6.196491905658264e-07,
+ "loss": 0.0596,
+ "step": 32690
+ },
+ {
+ "epoch": 89.07629427792915,
+ "grad_norm": 1.2947349548339844,
+ "learning_rate": 6.193434071279414e-07,
+ "loss": 0.0338,
+ "step": 32691
+ },
+ {
+ "epoch": 89.07901907356948,
+ "grad_norm": 1.490253210067749,
+ "learning_rate": 6.190376967463097e-07,
+ "loss": 0.0087,
+ "step": 32692
+ },
+ {
+ "epoch": 89.0817438692098,
+ "grad_norm": 1.0575201511383057,
+ "learning_rate": 6.187320594233093e-07,
+ "loss": 0.0067,
+ "step": 32693
+ },
+ {
+ "epoch": 89.08446866485014,
+ "grad_norm": 1.1602590084075928,
+ "learning_rate": 6.184264951613206e-07,
+ "loss": 0.0832,
+ "step": 32694
+ },
+ {
+ "epoch": 89.08719346049047,
+ "grad_norm": 1.5207042694091797,
+ "learning_rate": 6.181210039627261e-07,
+ "loss": 0.0757,
+ "step": 32695
+ },
+ {
+ "epoch": 89.08991825613079,
+ "grad_norm": 1.4399749040603638,
+ "learning_rate": 6.178155858299029e-07,
+ "loss": 0.0289,
+ "step": 32696
+ },
+ {
+ "epoch": 89.09264305177112,
+ "grad_norm": 1.5929343700408936,
+ "learning_rate": 6.1751024076523e-07,
+ "loss": 0.03,
+ "step": 32697
+ },
+ {
+ "epoch": 89.09536784741144,
+ "grad_norm": 0.9374881386756897,
+ "learning_rate": 6.172049687710846e-07,
+ "loss": 0.0133,
+ "step": 32698
+ },
+ {
+ "epoch": 89.09809264305177,
+ "grad_norm": 1.3496907949447632,
+ "learning_rate": 6.168997698498458e-07,
+ "loss": 0.017,
+ "step": 32699
+ },
+ {
+ "epoch": 89.1008174386921,
+ "grad_norm": 1.1359084844589233,
+ "learning_rate": 6.165946440038883e-07,
+ "loss": 0.0229,
+ "step": 32700
+ },
+ {
+ "epoch": 89.10354223433242,
+ "grad_norm": 1.0794864892959595,
+ "learning_rate": 6.162895912355904e-07,
+ "loss": 0.0107,
+ "step": 32701
+ },
+ {
+ "epoch": 89.10626702997276,
+ "grad_norm": 1.5401904582977295,
+ "learning_rate": 6.159846115473278e-07,
+ "loss": 0.0085,
+ "step": 32702
+ },
+ {
+ "epoch": 89.10899182561307,
+ "grad_norm": 0.5907041430473328,
+ "learning_rate": 6.156797049414742e-07,
+ "loss": 0.0061,
+ "step": 32703
+ },
+ {
+ "epoch": 89.11171662125341,
+ "grad_norm": 1.429768443107605,
+ "learning_rate": 6.153748714204055e-07,
+ "loss": 0.0758,
+ "step": 32704
+ },
+ {
+ "epoch": 89.11444141689373,
+ "grad_norm": 0.7910343408584595,
+ "learning_rate": 6.150701109864954e-07,
+ "loss": 0.0081,
+ "step": 32705
+ },
+ {
+ "epoch": 89.11716621253406,
+ "grad_norm": 1.0485939979553223,
+ "learning_rate": 6.147654236421174e-07,
+ "loss": 0.0177,
+ "step": 32706
+ },
+ {
+ "epoch": 89.11989100817439,
+ "grad_norm": 0.5084940195083618,
+ "learning_rate": 6.144608093896431e-07,
+ "loss": 0.0049,
+ "step": 32707
+ },
+ {
+ "epoch": 89.12261580381471,
+ "grad_norm": 1.0640316009521484,
+ "learning_rate": 6.141562682314473e-07,
+ "loss": 0.0445,
+ "step": 32708
+ },
+ {
+ "epoch": 89.12534059945504,
+ "grad_norm": 1.4240599870681763,
+ "learning_rate": 6.138518001699001e-07,
+ "loss": 0.0088,
+ "step": 32709
+ },
+ {
+ "epoch": 89.12806539509536,
+ "grad_norm": 0.4838639795780182,
+ "learning_rate": 6.135474052073731e-07,
+ "loss": 0.0048,
+ "step": 32710
+ },
+ {
+ "epoch": 89.1307901907357,
+ "grad_norm": 0.8588528633117676,
+ "learning_rate": 6.132430833462344e-07,
+ "loss": 0.0109,
+ "step": 32711
+ },
+ {
+ "epoch": 89.13351498637603,
+ "grad_norm": 1.1029281616210938,
+ "learning_rate": 6.129388345888598e-07,
+ "loss": 0.0161,
+ "step": 32712
+ },
+ {
+ "epoch": 89.13623978201635,
+ "grad_norm": 0.27080830931663513,
+ "learning_rate": 6.126346589376142e-07,
+ "loss": 0.0035,
+ "step": 32713
+ },
+ {
+ "epoch": 89.13896457765668,
+ "grad_norm": 1.2841107845306396,
+ "learning_rate": 6.123305563948667e-07,
+ "loss": 0.0363,
+ "step": 32714
+ },
+ {
+ "epoch": 89.141689373297,
+ "grad_norm": 1.4910169839859009,
+ "learning_rate": 6.120265269629877e-07,
+ "loss": 0.0607,
+ "step": 32715
+ },
+ {
+ "epoch": 89.14441416893733,
+ "grad_norm": 1.4531099796295166,
+ "learning_rate": 6.117225706443441e-07,
+ "loss": 0.056,
+ "step": 32716
+ },
+ {
+ "epoch": 89.14713896457765,
+ "grad_norm": 1.0038620233535767,
+ "learning_rate": 6.11418687441302e-07,
+ "loss": 0.0153,
+ "step": 32717
+ },
+ {
+ "epoch": 89.14986376021798,
+ "grad_norm": 0.9106032252311707,
+ "learning_rate": 6.111148773562303e-07,
+ "loss": 0.0236,
+ "step": 32718
+ },
+ {
+ "epoch": 89.15258855585832,
+ "grad_norm": 1.119272232055664,
+ "learning_rate": 6.108111403914919e-07,
+ "loss": 0.0159,
+ "step": 32719
+ },
+ {
+ "epoch": 89.15531335149863,
+ "grad_norm": 0.546339213848114,
+ "learning_rate": 6.105074765494556e-07,
+ "loss": 0.0066,
+ "step": 32720
+ },
+ {
+ "epoch": 89.15803814713897,
+ "grad_norm": 0.9608066082000732,
+ "learning_rate": 6.102038858324855e-07,
+ "loss": 0.0111,
+ "step": 32721
+ },
+ {
+ "epoch": 89.16076294277929,
+ "grad_norm": 0.8395963311195374,
+ "learning_rate": 6.09900368242945e-07,
+ "loss": 0.0101,
+ "step": 32722
+ },
+ {
+ "epoch": 89.16348773841962,
+ "grad_norm": 1.2945839166641235,
+ "learning_rate": 6.095969237831956e-07,
+ "loss": 0.0289,
+ "step": 32723
+ },
+ {
+ "epoch": 89.16621253405995,
+ "grad_norm": 0.9817522168159485,
+ "learning_rate": 6.092935524556065e-07,
+ "loss": 0.0238,
+ "step": 32724
+ },
+ {
+ "epoch": 89.16893732970027,
+ "grad_norm": 1.0552119016647339,
+ "learning_rate": 6.089902542625359e-07,
+ "loss": 0.0194,
+ "step": 32725
+ },
+ {
+ "epoch": 89.1716621253406,
+ "grad_norm": 0.7462309002876282,
+ "learning_rate": 6.086870292063473e-07,
+ "loss": 0.0113,
+ "step": 32726
+ },
+ {
+ "epoch": 89.17438692098092,
+ "grad_norm": 1.2648727893829346,
+ "learning_rate": 6.083838772894002e-07,
+ "loss": 0.0145,
+ "step": 32727
+ },
+ {
+ "epoch": 89.17711171662125,
+ "grad_norm": 1.0240609645843506,
+ "learning_rate": 6.080807985140591e-07,
+ "loss": 0.0108,
+ "step": 32728
+ },
+ {
+ "epoch": 89.17983651226157,
+ "grad_norm": 0.602739691734314,
+ "learning_rate": 6.077777928826811e-07,
+ "loss": 0.0053,
+ "step": 32729
+ },
+ {
+ "epoch": 89.1825613079019,
+ "grad_norm": 1.1342841386795044,
+ "learning_rate": 6.074748603976288e-07,
+ "loss": 0.0167,
+ "step": 32730
+ },
+ {
+ "epoch": 89.18528610354224,
+ "grad_norm": 1.5034358501434326,
+ "learning_rate": 6.07172001061258e-07,
+ "loss": 0.081,
+ "step": 32731
+ },
+ {
+ "epoch": 89.18801089918256,
+ "grad_norm": 0.9884399771690369,
+ "learning_rate": 6.068692148759314e-07,
+ "loss": 0.0172,
+ "step": 32732
+ },
+ {
+ "epoch": 89.19073569482289,
+ "grad_norm": 0.8873487114906311,
+ "learning_rate": 6.065665018440037e-07,
+ "loss": 0.007,
+ "step": 32733
+ },
+ {
+ "epoch": 89.19346049046321,
+ "grad_norm": 1.0102190971374512,
+ "learning_rate": 6.062638619678352e-07,
+ "loss": 0.0092,
+ "step": 32734
+ },
+ {
+ "epoch": 89.19618528610354,
+ "grad_norm": 1.4013372659683228,
+ "learning_rate": 6.059612952497773e-07,
+ "loss": 0.0117,
+ "step": 32735
+ },
+ {
+ "epoch": 89.19891008174388,
+ "grad_norm": 0.975582480430603,
+ "learning_rate": 6.056588016921949e-07,
+ "loss": 0.0113,
+ "step": 32736
+ },
+ {
+ "epoch": 89.2016348773842,
+ "grad_norm": 1.0417664051055908,
+ "learning_rate": 6.053563812974384e-07,
+ "loss": 0.0098,
+ "step": 32737
+ },
+ {
+ "epoch": 89.20435967302453,
+ "grad_norm": 1.1693636178970337,
+ "learning_rate": 6.050540340678635e-07,
+ "loss": 0.01,
+ "step": 32738
+ },
+ {
+ "epoch": 89.20708446866485,
+ "grad_norm": 1.5689011812210083,
+ "learning_rate": 6.04751760005825e-07,
+ "loss": 0.0149,
+ "step": 32739
+ },
+ {
+ "epoch": 89.20980926430518,
+ "grad_norm": 0.7742889523506165,
+ "learning_rate": 6.044495591136778e-07,
+ "loss": 0.0062,
+ "step": 32740
+ },
+ {
+ "epoch": 89.2125340599455,
+ "grad_norm": 0.7730068564414978,
+ "learning_rate": 6.041474313937756e-07,
+ "loss": 0.0161,
+ "step": 32741
+ },
+ {
+ "epoch": 89.21525885558583,
+ "grad_norm": 1.4011759757995605,
+ "learning_rate": 6.038453768484709e-07,
+ "loss": 0.0123,
+ "step": 32742
+ },
+ {
+ "epoch": 89.21798365122616,
+ "grad_norm": 1.48442804813385,
+ "learning_rate": 6.035433954801151e-07,
+ "loss": 0.039,
+ "step": 32743
+ },
+ {
+ "epoch": 89.22070844686648,
+ "grad_norm": 0.5736701488494873,
+ "learning_rate": 6.03241487291062e-07,
+ "loss": 0.0067,
+ "step": 32744
+ },
+ {
+ "epoch": 89.22343324250681,
+ "grad_norm": 0.7867155075073242,
+ "learning_rate": 6.029396522836617e-07,
+ "loss": 0.0058,
+ "step": 32745
+ },
+ {
+ "epoch": 89.22615803814713,
+ "grad_norm": 0.8613691329956055,
+ "learning_rate": 6.026378904602648e-07,
+ "loss": 0.012,
+ "step": 32746
+ },
+ {
+ "epoch": 89.22888283378747,
+ "grad_norm": 0.9706292748451233,
+ "learning_rate": 6.023362018232193e-07,
+ "loss": 0.0135,
+ "step": 32747
+ },
+ {
+ "epoch": 89.2316076294278,
+ "grad_norm": 0.8072986006736755,
+ "learning_rate": 6.0203458637488e-07,
+ "loss": 0.0142,
+ "step": 32748
+ },
+ {
+ "epoch": 89.23433242506812,
+ "grad_norm": 0.9964858889579773,
+ "learning_rate": 6.017330441175917e-07,
+ "loss": 0.0083,
+ "step": 32749
+ },
+ {
+ "epoch": 89.23705722070845,
+ "grad_norm": 1.0410569906234741,
+ "learning_rate": 6.014315750537048e-07,
+ "loss": 0.0112,
+ "step": 32750
+ },
+ {
+ "epoch": 89.23978201634877,
+ "grad_norm": 1.30545973777771,
+ "learning_rate": 6.011301791855639e-07,
+ "loss": 0.134,
+ "step": 32751
+ },
+ {
+ "epoch": 89.2425068119891,
+ "grad_norm": 1.159164309501648,
+ "learning_rate": 6.008288565155196e-07,
+ "loss": 0.0125,
+ "step": 32752
+ },
+ {
+ "epoch": 89.24523160762942,
+ "grad_norm": 0.8676906228065491,
+ "learning_rate": 6.005276070459187e-07,
+ "loss": 0.0113,
+ "step": 32753
+ },
+ {
+ "epoch": 89.24795640326975,
+ "grad_norm": 1.3680224418640137,
+ "learning_rate": 6.002264307791061e-07,
+ "loss": 0.023,
+ "step": 32754
+ },
+ {
+ "epoch": 89.25068119891009,
+ "grad_norm": 1.1615803241729736,
+ "learning_rate": 5.999253277174255e-07,
+ "loss": 0.0243,
+ "step": 32755
+ },
+ {
+ "epoch": 89.2534059945504,
+ "grad_norm": 0.8551202416419983,
+ "learning_rate": 5.996242978632271e-07,
+ "loss": 0.009,
+ "step": 32756
+ },
+ {
+ "epoch": 89.25613079019074,
+ "grad_norm": 0.8115959167480469,
+ "learning_rate": 5.993233412188504e-07,
+ "loss": 0.0096,
+ "step": 32757
+ },
+ {
+ "epoch": 89.25885558583106,
+ "grad_norm": 1.3359906673431396,
+ "learning_rate": 5.990224577866422e-07,
+ "loss": 0.0214,
+ "step": 32758
+ },
+ {
+ "epoch": 89.26158038147139,
+ "grad_norm": 1.1843032836914062,
+ "learning_rate": 5.987216475689428e-07,
+ "loss": 0.0119,
+ "step": 32759
+ },
+ {
+ "epoch": 89.26430517711172,
+ "grad_norm": 1.1827421188354492,
+ "learning_rate": 5.984209105680972e-07,
+ "loss": 0.0167,
+ "step": 32760
+ },
+ {
+ "epoch": 89.26702997275204,
+ "grad_norm": 0.722416877746582,
+ "learning_rate": 5.981202467864478e-07,
+ "loss": 0.0075,
+ "step": 32761
+ },
+ {
+ "epoch": 89.26975476839237,
+ "grad_norm": 0.8984020352363586,
+ "learning_rate": 5.97819656226335e-07,
+ "loss": 0.0073,
+ "step": 32762
+ },
+ {
+ "epoch": 89.2724795640327,
+ "grad_norm": 1.5581332445144653,
+ "learning_rate": 5.975191388900991e-07,
+ "loss": 0.0181,
+ "step": 32763
+ },
+ {
+ "epoch": 89.27520435967303,
+ "grad_norm": 0.4901202917098999,
+ "learning_rate": 5.972186947800829e-07,
+ "loss": 0.0052,
+ "step": 32764
+ },
+ {
+ "epoch": 89.27792915531334,
+ "grad_norm": 0.988463282585144,
+ "learning_rate": 5.969183238986243e-07,
+ "loss": 0.0347,
+ "step": 32765
+ },
+ {
+ "epoch": 89.28065395095368,
+ "grad_norm": 0.9315866231918335,
+ "learning_rate": 5.966180262480636e-07,
+ "loss": 0.0156,
+ "step": 32766
+ },
+ {
+ "epoch": 89.28337874659401,
+ "grad_norm": 0.9524497985839844,
+ "learning_rate": 5.96317801830737e-07,
+ "loss": 0.0194,
+ "step": 32767
+ },
+ {
+ "epoch": 89.28610354223433,
+ "grad_norm": 3.201395034790039,
+ "learning_rate": 5.960176506489868e-07,
+ "loss": 0.0236,
+ "step": 32768
+ },
+ {
+ "epoch": 89.28882833787466,
+ "grad_norm": 1.4689843654632568,
+ "learning_rate": 5.957175727051467e-07,
+ "loss": 0.0196,
+ "step": 32769
+ },
+ {
+ "epoch": 89.29155313351498,
+ "grad_norm": 1.4222344160079956,
+ "learning_rate": 5.954175680015572e-07,
+ "loss": 0.0162,
+ "step": 32770
+ },
+ {
+ "epoch": 89.29427792915531,
+ "grad_norm": 1.4878612756729126,
+ "learning_rate": 5.951176365405531e-07,
+ "loss": 0.0531,
+ "step": 32771
+ },
+ {
+ "epoch": 89.29700272479565,
+ "grad_norm": 0.8698878288269043,
+ "learning_rate": 5.948177783244691e-07,
+ "loss": 0.007,
+ "step": 32772
+ },
+ {
+ "epoch": 89.29972752043597,
+ "grad_norm": 1.1524243354797363,
+ "learning_rate": 5.945179933556422e-07,
+ "loss": 0.0427,
+ "step": 32773
+ },
+ {
+ "epoch": 89.3024523160763,
+ "grad_norm": 1.5679150819778442,
+ "learning_rate": 5.942182816364061e-07,
+ "loss": 0.0104,
+ "step": 32774
+ },
+ {
+ "epoch": 89.30517711171662,
+ "grad_norm": 1.1779695749282837,
+ "learning_rate": 5.939186431690957e-07,
+ "loss": 0.0185,
+ "step": 32775
+ },
+ {
+ "epoch": 89.30790190735695,
+ "grad_norm": 1.3977652788162231,
+ "learning_rate": 5.936190779560435e-07,
+ "loss": 0.0497,
+ "step": 32776
+ },
+ {
+ "epoch": 89.31062670299727,
+ "grad_norm": 1.1816238164901733,
+ "learning_rate": 5.933195859995833e-07,
+ "loss": 0.0337,
+ "step": 32777
+ },
+ {
+ "epoch": 89.3133514986376,
+ "grad_norm": 1.4802186489105225,
+ "learning_rate": 5.930201673020486e-07,
+ "loss": 0.0527,
+ "step": 32778
+ },
+ {
+ "epoch": 89.31607629427793,
+ "grad_norm": 1.2400912046432495,
+ "learning_rate": 5.927208218657676e-07,
+ "loss": 0.052,
+ "step": 32779
+ },
+ {
+ "epoch": 89.31880108991825,
+ "grad_norm": 1.927915334701538,
+ "learning_rate": 5.924215496930763e-07,
+ "loss": 0.0146,
+ "step": 32780
+ },
+ {
+ "epoch": 89.32152588555859,
+ "grad_norm": 0.6282851099967957,
+ "learning_rate": 5.921223507863028e-07,
+ "loss": 0.0055,
+ "step": 32781
+ },
+ {
+ "epoch": 89.3242506811989,
+ "grad_norm": 1.4469561576843262,
+ "learning_rate": 5.918232251477784e-07,
+ "loss": 0.0475,
+ "step": 32782
+ },
+ {
+ "epoch": 89.32697547683924,
+ "grad_norm": 1.767168641090393,
+ "learning_rate": 5.915241727798293e-07,
+ "loss": 0.012,
+ "step": 32783
+ },
+ {
+ "epoch": 89.32970027247957,
+ "grad_norm": 0.8327605724334717,
+ "learning_rate": 5.91225193684789e-07,
+ "loss": 0.0067,
+ "step": 32784
+ },
+ {
+ "epoch": 89.33242506811989,
+ "grad_norm": 0.45202621817588806,
+ "learning_rate": 5.909262878649846e-07,
+ "loss": 0.0052,
+ "step": 32785
+ },
+ {
+ "epoch": 89.33514986376022,
+ "grad_norm": 1.3248730897903442,
+ "learning_rate": 5.90627455322742e-07,
+ "loss": 0.0101,
+ "step": 32786
+ },
+ {
+ "epoch": 89.33787465940054,
+ "grad_norm": 1.0057116746902466,
+ "learning_rate": 5.903286960603905e-07,
+ "loss": 0.0138,
+ "step": 32787
+ },
+ {
+ "epoch": 89.34059945504087,
+ "grad_norm": 0.6179301142692566,
+ "learning_rate": 5.900300100802558e-07,
+ "loss": 0.0059,
+ "step": 32788
+ },
+ {
+ "epoch": 89.34332425068119,
+ "grad_norm": 1.19826340675354,
+ "learning_rate": 5.897313973846652e-07,
+ "loss": 0.0444,
+ "step": 32789
+ },
+ {
+ "epoch": 89.34604904632153,
+ "grad_norm": 0.7656465768814087,
+ "learning_rate": 5.894328579759434e-07,
+ "loss": 0.0351,
+ "step": 32790
+ },
+ {
+ "epoch": 89.34877384196186,
+ "grad_norm": 4.070415496826172,
+ "learning_rate": 5.891343918564163e-07,
+ "loss": 0.0752,
+ "step": 32791
+ },
+ {
+ "epoch": 89.35149863760218,
+ "grad_norm": 0.874921977519989,
+ "learning_rate": 5.888359990284065e-07,
+ "loss": 0.0087,
+ "step": 32792
+ },
+ {
+ "epoch": 89.35422343324251,
+ "grad_norm": 0.4561637341976166,
+ "learning_rate": 5.885376794942399e-07,
+ "loss": 0.0033,
+ "step": 32793
+ },
+ {
+ "epoch": 89.35694822888283,
+ "grad_norm": 1.1725455522537231,
+ "learning_rate": 5.882394332562392e-07,
+ "loss": 0.011,
+ "step": 32794
+ },
+ {
+ "epoch": 89.35967302452316,
+ "grad_norm": 1.0804979801177979,
+ "learning_rate": 5.879412603167267e-07,
+ "loss": 0.044,
+ "step": 32795
+ },
+ {
+ "epoch": 89.3623978201635,
+ "grad_norm": 1.2329307794570923,
+ "learning_rate": 5.87643160678023e-07,
+ "loss": 0.0453,
+ "step": 32796
+ },
+ {
+ "epoch": 89.36512261580381,
+ "grad_norm": 0.9602465629577637,
+ "learning_rate": 5.87345134342453e-07,
+ "loss": 0.0082,
+ "step": 32797
+ },
+ {
+ "epoch": 89.36784741144415,
+ "grad_norm": 1.3761242628097534,
+ "learning_rate": 5.870471813123368e-07,
+ "loss": 0.0553,
+ "step": 32798
+ },
+ {
+ "epoch": 89.37057220708446,
+ "grad_norm": 1.3837095499038696,
+ "learning_rate": 5.867493015899916e-07,
+ "loss": 0.017,
+ "step": 32799
+ },
+ {
+ "epoch": 89.3732970027248,
+ "grad_norm": 1.2645758390426636,
+ "learning_rate": 5.864514951777423e-07,
+ "loss": 0.0241,
+ "step": 32800
+ },
+ {
+ "epoch": 89.37602179836512,
+ "grad_norm": 1.8790591955184937,
+ "learning_rate": 5.861537620779056e-07,
+ "loss": 0.0112,
+ "step": 32801
+ },
+ {
+ "epoch": 89.37874659400545,
+ "grad_norm": 1.2418513298034668,
+ "learning_rate": 5.858561022928e-07,
+ "loss": 0.0289,
+ "step": 32802
+ },
+ {
+ "epoch": 89.38147138964578,
+ "grad_norm": 0.9674936532974243,
+ "learning_rate": 5.855585158247434e-07,
+ "loss": 0.0105,
+ "step": 32803
+ },
+ {
+ "epoch": 89.3841961852861,
+ "grad_norm": 0.740847110748291,
+ "learning_rate": 5.852610026760541e-07,
+ "loss": 0.0076,
+ "step": 32804
+ },
+ {
+ "epoch": 89.38692098092643,
+ "grad_norm": 1.2572518587112427,
+ "learning_rate": 5.849635628490502e-07,
+ "loss": 0.0267,
+ "step": 32805
+ },
+ {
+ "epoch": 89.38964577656675,
+ "grad_norm": 1.1196528673171997,
+ "learning_rate": 5.846661963460465e-07,
+ "loss": 0.0879,
+ "step": 32806
+ },
+ {
+ "epoch": 89.39237057220708,
+ "grad_norm": 0.7269967794418335,
+ "learning_rate": 5.8436890316936e-07,
+ "loss": 0.0075,
+ "step": 32807
+ },
+ {
+ "epoch": 89.39509536784742,
+ "grad_norm": 0.7152775526046753,
+ "learning_rate": 5.840716833213033e-07,
+ "loss": 0.0102,
+ "step": 32808
+ },
+ {
+ "epoch": 89.39782016348774,
+ "grad_norm": 0.7728998064994812,
+ "learning_rate": 5.837745368041958e-07,
+ "loss": 0.0073,
+ "step": 32809
+ },
+ {
+ "epoch": 89.40054495912807,
+ "grad_norm": 0.9484930634498596,
+ "learning_rate": 5.834774636203488e-07,
+ "loss": 0.0149,
+ "step": 32810
+ },
+ {
+ "epoch": 89.40326975476839,
+ "grad_norm": 0.8391991853713989,
+ "learning_rate": 5.831804637720762e-07,
+ "loss": 0.0076,
+ "step": 32811
+ },
+ {
+ "epoch": 89.40599455040872,
+ "grad_norm": 0.9776426553726196,
+ "learning_rate": 5.828835372616892e-07,
+ "loss": 0.0157,
+ "step": 32812
+ },
+ {
+ "epoch": 89.40871934604904,
+ "grad_norm": 1.0557631254196167,
+ "learning_rate": 5.82586684091504e-07,
+ "loss": 0.035,
+ "step": 32813
+ },
+ {
+ "epoch": 89.41144414168937,
+ "grad_norm": 1.1177924871444702,
+ "learning_rate": 5.822899042638297e-07,
+ "loss": 0.0919,
+ "step": 32814
+ },
+ {
+ "epoch": 89.4141689373297,
+ "grad_norm": 1.2595441341400146,
+ "learning_rate": 5.8199319778098e-07,
+ "loss": 0.0221,
+ "step": 32815
+ },
+ {
+ "epoch": 89.41689373297002,
+ "grad_norm": 2.135347843170166,
+ "learning_rate": 5.816965646452622e-07,
+ "loss": 0.0358,
+ "step": 32816
+ },
+ {
+ "epoch": 89.41961852861036,
+ "grad_norm": 1.2536766529083252,
+ "learning_rate": 5.814000048589896e-07,
+ "loss": 0.0239,
+ "step": 32817
+ },
+ {
+ "epoch": 89.42234332425068,
+ "grad_norm": 2.1363258361816406,
+ "learning_rate": 5.811035184244706e-07,
+ "loss": 0.0375,
+ "step": 32818
+ },
+ {
+ "epoch": 89.42506811989101,
+ "grad_norm": 1.2458350658416748,
+ "learning_rate": 5.808071053440145e-07,
+ "loss": 0.0465,
+ "step": 32819
+ },
+ {
+ "epoch": 89.42779291553134,
+ "grad_norm": 1.037219524383545,
+ "learning_rate": 5.805107656199272e-07,
+ "loss": 0.0077,
+ "step": 32820
+ },
+ {
+ "epoch": 89.43051771117166,
+ "grad_norm": 0.6293460726737976,
+ "learning_rate": 5.8021449925452e-07,
+ "loss": 0.0044,
+ "step": 32821
+ },
+ {
+ "epoch": 89.433242506812,
+ "grad_norm": 1.7547701597213745,
+ "learning_rate": 5.799183062501001e-07,
+ "loss": 0.0253,
+ "step": 32822
+ },
+ {
+ "epoch": 89.43596730245231,
+ "grad_norm": 0.759885847568512,
+ "learning_rate": 5.796221866089735e-07,
+ "loss": 0.0097,
+ "step": 32823
+ },
+ {
+ "epoch": 89.43869209809264,
+ "grad_norm": 0.9413685202598572,
+ "learning_rate": 5.79326140333445e-07,
+ "loss": 0.0092,
+ "step": 32824
+ },
+ {
+ "epoch": 89.44141689373296,
+ "grad_norm": 0.9885408282279968,
+ "learning_rate": 5.790301674258214e-07,
+ "loss": 0.0168,
+ "step": 32825
+ },
+ {
+ "epoch": 89.4441416893733,
+ "grad_norm": 1.3179137706756592,
+ "learning_rate": 5.787342678884089e-07,
+ "loss": 0.0971,
+ "step": 32826
+ },
+ {
+ "epoch": 89.44686648501363,
+ "grad_norm": 0.7733421921730042,
+ "learning_rate": 5.784384417235101e-07,
+ "loss": 0.0072,
+ "step": 32827
+ },
+ {
+ "epoch": 89.44959128065395,
+ "grad_norm": 0.8119680881500244,
+ "learning_rate": 5.781426889334285e-07,
+ "loss": 0.038,
+ "step": 32828
+ },
+ {
+ "epoch": 89.45231607629428,
+ "grad_norm": 1.4509254693984985,
+ "learning_rate": 5.778470095204691e-07,
+ "loss": 0.0199,
+ "step": 32829
+ },
+ {
+ "epoch": 89.4550408719346,
+ "grad_norm": 1.0733753442764282,
+ "learning_rate": 5.775514034869345e-07,
+ "loss": 0.0276,
+ "step": 32830
+ },
+ {
+ "epoch": 89.45776566757493,
+ "grad_norm": 0.8708061575889587,
+ "learning_rate": 5.772558708351262e-07,
+ "loss": 0.0143,
+ "step": 32831
+ },
+ {
+ "epoch": 89.46049046321527,
+ "grad_norm": 0.5269947648048401,
+ "learning_rate": 5.769604115673444e-07,
+ "loss": 0.0069,
+ "step": 32832
+ },
+ {
+ "epoch": 89.46321525885558,
+ "grad_norm": 0.44941914081573486,
+ "learning_rate": 5.76665025685893e-07,
+ "loss": 0.0043,
+ "step": 32833
+ },
+ {
+ "epoch": 89.46594005449592,
+ "grad_norm": 1.3281902074813843,
+ "learning_rate": 5.763697131930712e-07,
+ "loss": 0.0137,
+ "step": 32834
+ },
+ {
+ "epoch": 89.46866485013624,
+ "grad_norm": 0.6803986430168152,
+ "learning_rate": 5.760744740911794e-07,
+ "loss": 0.0052,
+ "step": 32835
+ },
+ {
+ "epoch": 89.47138964577657,
+ "grad_norm": 1.8806133270263672,
+ "learning_rate": 5.757793083825147e-07,
+ "loss": 0.0164,
+ "step": 32836
+ },
+ {
+ "epoch": 89.47411444141689,
+ "grad_norm": 2.1495068073272705,
+ "learning_rate": 5.754842160693785e-07,
+ "loss": 0.0418,
+ "step": 32837
+ },
+ {
+ "epoch": 89.47683923705722,
+ "grad_norm": 1.1903669834136963,
+ "learning_rate": 5.751891971540658e-07,
+ "loss": 0.0174,
+ "step": 32838
+ },
+ {
+ "epoch": 89.47956403269755,
+ "grad_norm": 1.0953869819641113,
+ "learning_rate": 5.748942516388789e-07,
+ "loss": 0.0242,
+ "step": 32839
+ },
+ {
+ "epoch": 89.48228882833787,
+ "grad_norm": 0.5618864297866821,
+ "learning_rate": 5.745993795261096e-07,
+ "loss": 0.0044,
+ "step": 32840
+ },
+ {
+ "epoch": 89.4850136239782,
+ "grad_norm": 1.164741039276123,
+ "learning_rate": 5.743045808180592e-07,
+ "loss": 0.0126,
+ "step": 32841
+ },
+ {
+ "epoch": 89.48773841961852,
+ "grad_norm": 1.5689786672592163,
+ "learning_rate": 5.740098555170215e-07,
+ "loss": 0.0445,
+ "step": 32842
+ },
+ {
+ "epoch": 89.49046321525886,
+ "grad_norm": 0.8203811049461365,
+ "learning_rate": 5.737152036252914e-07,
+ "loss": 0.1423,
+ "step": 32843
+ },
+ {
+ "epoch": 89.49318801089919,
+ "grad_norm": 1.016513466835022,
+ "learning_rate": 5.734206251451624e-07,
+ "loss": 0.0125,
+ "step": 32844
+ },
+ {
+ "epoch": 89.49591280653951,
+ "grad_norm": 0.9336903095245361,
+ "learning_rate": 5.731261200789318e-07,
+ "loss": 0.0121,
+ "step": 32845
+ },
+ {
+ "epoch": 89.49863760217984,
+ "grad_norm": 1.471013069152832,
+ "learning_rate": 5.728316884288909e-07,
+ "loss": 0.0073,
+ "step": 32846
+ },
+ {
+ "epoch": 89.50136239782016,
+ "grad_norm": 0.7257657647132874,
+ "learning_rate": 5.725373301973336e-07,
+ "loss": 0.0096,
+ "step": 32847
+ },
+ {
+ "epoch": 89.50408719346049,
+ "grad_norm": 1.0760234594345093,
+ "learning_rate": 5.722430453865513e-07,
+ "loss": 0.0091,
+ "step": 32848
+ },
+ {
+ "epoch": 89.50681198910081,
+ "grad_norm": 0.5498788356781006,
+ "learning_rate": 5.719488339988378e-07,
+ "loss": 0.005,
+ "step": 32849
+ },
+ {
+ "epoch": 89.50953678474114,
+ "grad_norm": 1.3494290113449097,
+ "learning_rate": 5.716546960364844e-07,
+ "loss": 0.0332,
+ "step": 32850
+ },
+ {
+ "epoch": 89.51226158038148,
+ "grad_norm": 1.0040202140808105,
+ "learning_rate": 5.713606315017794e-07,
+ "loss": 0.0128,
+ "step": 32851
+ },
+ {
+ "epoch": 89.5149863760218,
+ "grad_norm": 1.0470376014709473,
+ "learning_rate": 5.710666403970133e-07,
+ "loss": 0.0222,
+ "step": 32852
+ },
+ {
+ "epoch": 89.51771117166213,
+ "grad_norm": 0.6106497645378113,
+ "learning_rate": 5.707727227244786e-07,
+ "loss": 0.0043,
+ "step": 32853
+ },
+ {
+ "epoch": 89.52043596730245,
+ "grad_norm": 1.022108793258667,
+ "learning_rate": 5.704788784864612e-07,
+ "loss": 0.0983,
+ "step": 32854
+ },
+ {
+ "epoch": 89.52316076294278,
+ "grad_norm": 1.3689988851547241,
+ "learning_rate": 5.701851076852504e-07,
+ "loss": 0.0302,
+ "step": 32855
+ },
+ {
+ "epoch": 89.52588555858311,
+ "grad_norm": 0.4319162666797638,
+ "learning_rate": 5.698914103231357e-07,
+ "loss": 0.0038,
+ "step": 32856
+ },
+ {
+ "epoch": 89.52861035422343,
+ "grad_norm": 1.5258710384368896,
+ "learning_rate": 5.695977864024016e-07,
+ "loss": 0.1184,
+ "step": 32857
+ },
+ {
+ "epoch": 89.53133514986376,
+ "grad_norm": 0.9578184485435486,
+ "learning_rate": 5.693042359253387e-07,
+ "loss": 0.0097,
+ "step": 32858
+ },
+ {
+ "epoch": 89.53405994550408,
+ "grad_norm": 0.9783326983451843,
+ "learning_rate": 5.690107588942306e-07,
+ "loss": 0.0692,
+ "step": 32859
+ },
+ {
+ "epoch": 89.53678474114442,
+ "grad_norm": 0.9382436275482178,
+ "learning_rate": 5.687173553113623e-07,
+ "loss": 0.009,
+ "step": 32860
+ },
+ {
+ "epoch": 89.53950953678473,
+ "grad_norm": 0.9739310145378113,
+ "learning_rate": 5.684240251790186e-07,
+ "loss": 0.0085,
+ "step": 32861
+ },
+ {
+ "epoch": 89.54223433242507,
+ "grad_norm": 1.3089765310287476,
+ "learning_rate": 5.681307684994874e-07,
+ "loss": 0.0144,
+ "step": 32862
+ },
+ {
+ "epoch": 89.5449591280654,
+ "grad_norm": 1.7433899641036987,
+ "learning_rate": 5.678375852750496e-07,
+ "loss": 0.0692,
+ "step": 32863
+ },
+ {
+ "epoch": 89.54768392370572,
+ "grad_norm": 0.7490487694740295,
+ "learning_rate": 5.675444755079873e-07,
+ "loss": 0.0125,
+ "step": 32864
+ },
+ {
+ "epoch": 89.55040871934605,
+ "grad_norm": 1.3122334480285645,
+ "learning_rate": 5.672514392005868e-07,
+ "loss": 0.1548,
+ "step": 32865
+ },
+ {
+ "epoch": 89.55313351498637,
+ "grad_norm": 1.1022037267684937,
+ "learning_rate": 5.669584763551284e-07,
+ "loss": 0.0433,
+ "step": 32866
+ },
+ {
+ "epoch": 89.5558583106267,
+ "grad_norm": 1.0253276824951172,
+ "learning_rate": 5.666655869738935e-07,
+ "loss": 0.0386,
+ "step": 32867
+ },
+ {
+ "epoch": 89.55858310626704,
+ "grad_norm": 1.1596813201904297,
+ "learning_rate": 5.663727710591627e-07,
+ "loss": 0.0321,
+ "step": 32868
+ },
+ {
+ "epoch": 89.56130790190736,
+ "grad_norm": 1.6475889682769775,
+ "learning_rate": 5.660800286132173e-07,
+ "loss": 0.148,
+ "step": 32869
+ },
+ {
+ "epoch": 89.56403269754769,
+ "grad_norm": 1.4931083917617798,
+ "learning_rate": 5.657873596383378e-07,
+ "loss": 0.0914,
+ "step": 32870
+ },
+ {
+ "epoch": 89.566757493188,
+ "grad_norm": 0.7620217204093933,
+ "learning_rate": 5.654947641368036e-07,
+ "loss": 0.008,
+ "step": 32871
+ },
+ {
+ "epoch": 89.56948228882834,
+ "grad_norm": 1.1156176328659058,
+ "learning_rate": 5.652022421108894e-07,
+ "loss": 0.0076,
+ "step": 32872
+ },
+ {
+ "epoch": 89.57220708446866,
+ "grad_norm": 1.9792946577072144,
+ "learning_rate": 5.649097935628778e-07,
+ "loss": 0.0275,
+ "step": 32873
+ },
+ {
+ "epoch": 89.57493188010899,
+ "grad_norm": 1.2065457105636597,
+ "learning_rate": 5.646174184950459e-07,
+ "loss": 0.0502,
+ "step": 32874
+ },
+ {
+ "epoch": 89.57765667574932,
+ "grad_norm": 1.2579890489578247,
+ "learning_rate": 5.643251169096698e-07,
+ "loss": 0.0633,
+ "step": 32875
+ },
+ {
+ "epoch": 89.58038147138964,
+ "grad_norm": 0.7529277801513672,
+ "learning_rate": 5.640328888090263e-07,
+ "loss": 0.0061,
+ "step": 32876
+ },
+ {
+ "epoch": 89.58310626702998,
+ "grad_norm": 1.0485402345657349,
+ "learning_rate": 5.637407341953893e-07,
+ "loss": 0.0203,
+ "step": 32877
+ },
+ {
+ "epoch": 89.5858310626703,
+ "grad_norm": 1.5010977983474731,
+ "learning_rate": 5.634486530710382e-07,
+ "loss": 0.0176,
+ "step": 32878
+ },
+ {
+ "epoch": 89.58855585831063,
+ "grad_norm": 1.039919376373291,
+ "learning_rate": 5.631566454382442e-07,
+ "loss": 0.0067,
+ "step": 32879
+ },
+ {
+ "epoch": 89.59128065395096,
+ "grad_norm": 0.6787399649620056,
+ "learning_rate": 5.628647112992824e-07,
+ "loss": 0.0077,
+ "step": 32880
+ },
+ {
+ "epoch": 89.59400544959128,
+ "grad_norm": 1.7648745775222778,
+ "learning_rate": 5.625728506564254e-07,
+ "loss": 0.0149,
+ "step": 32881
+ },
+ {
+ "epoch": 89.59673024523161,
+ "grad_norm": 1.5413260459899902,
+ "learning_rate": 5.622810635119491e-07,
+ "loss": 0.0866,
+ "step": 32882
+ },
+ {
+ "epoch": 89.59945504087193,
+ "grad_norm": 0.973796546459198,
+ "learning_rate": 5.61989349868124e-07,
+ "loss": 0.0378,
+ "step": 32883
+ },
+ {
+ "epoch": 89.60217983651226,
+ "grad_norm": 0.9773348569869995,
+ "learning_rate": 5.616977097272214e-07,
+ "loss": 0.0087,
+ "step": 32884
+ },
+ {
+ "epoch": 89.60490463215258,
+ "grad_norm": 0.6920629143714905,
+ "learning_rate": 5.614061430915118e-07,
+ "loss": 0.0074,
+ "step": 32885
+ },
+ {
+ "epoch": 89.60762942779292,
+ "grad_norm": 1.0105397701263428,
+ "learning_rate": 5.611146499632692e-07,
+ "loss": 0.0081,
+ "step": 32886
+ },
+ {
+ "epoch": 89.61035422343325,
+ "grad_norm": 0.6906039118766785,
+ "learning_rate": 5.608232303447625e-07,
+ "loss": 0.0211,
+ "step": 32887
+ },
+ {
+ "epoch": 89.61307901907357,
+ "grad_norm": 1.5245766639709473,
+ "learning_rate": 5.605318842382589e-07,
+ "loss": 0.0226,
+ "step": 32888
+ },
+ {
+ "epoch": 89.6158038147139,
+ "grad_norm": 0.9050509929656982,
+ "learning_rate": 5.60240611646029e-07,
+ "loss": 0.006,
+ "step": 32889
+ },
+ {
+ "epoch": 89.61852861035422,
+ "grad_norm": 1.722063422203064,
+ "learning_rate": 5.599494125703431e-07,
+ "loss": 0.0323,
+ "step": 32890
+ },
+ {
+ "epoch": 89.62125340599455,
+ "grad_norm": 1.9566655158996582,
+ "learning_rate": 5.59658287013467e-07,
+ "loss": 0.0143,
+ "step": 32891
+ },
+ {
+ "epoch": 89.62397820163488,
+ "grad_norm": 1.0794888734817505,
+ "learning_rate": 5.59367234977668e-07,
+ "loss": 0.0116,
+ "step": 32892
+ },
+ {
+ "epoch": 89.6267029972752,
+ "grad_norm": 1.1479851007461548,
+ "learning_rate": 5.590762564652119e-07,
+ "loss": 0.0835,
+ "step": 32893
+ },
+ {
+ "epoch": 89.62942779291554,
+ "grad_norm": 0.9062066078186035,
+ "learning_rate": 5.587853514783681e-07,
+ "loss": 0.008,
+ "step": 32894
+ },
+ {
+ "epoch": 89.63215258855585,
+ "grad_norm": 0.7929849624633789,
+ "learning_rate": 5.584945200193992e-07,
+ "loss": 0.0095,
+ "step": 32895
+ },
+ {
+ "epoch": 89.63487738419619,
+ "grad_norm": 1.1152139902114868,
+ "learning_rate": 5.582037620905723e-07,
+ "loss": 0.0081,
+ "step": 32896
+ },
+ {
+ "epoch": 89.6376021798365,
+ "grad_norm": 0.9957127571105957,
+ "learning_rate": 5.579130776941488e-07,
+ "loss": 0.0095,
+ "step": 32897
+ },
+ {
+ "epoch": 89.64032697547684,
+ "grad_norm": 0.6185359358787537,
+ "learning_rate": 5.576224668323949e-07,
+ "loss": 0.0054,
+ "step": 32898
+ },
+ {
+ "epoch": 89.64305177111717,
+ "grad_norm": 1.370137095451355,
+ "learning_rate": 5.57331929507574e-07,
+ "loss": 0.0756,
+ "step": 32899
+ },
+ {
+ "epoch": 89.64577656675749,
+ "grad_norm": 0.9716392159461975,
+ "learning_rate": 5.570414657219481e-07,
+ "loss": 0.0054,
+ "step": 32900
+ },
+ {
+ "epoch": 89.64850136239782,
+ "grad_norm": 0.8531251549720764,
+ "learning_rate": 5.567510754777782e-07,
+ "loss": 0.0067,
+ "step": 32901
+ },
+ {
+ "epoch": 89.65122615803814,
+ "grad_norm": 0.7087875604629517,
+ "learning_rate": 5.564607587773274e-07,
+ "loss": 0.0076,
+ "step": 32902
+ },
+ {
+ "epoch": 89.65395095367847,
+ "grad_norm": 0.9113971590995789,
+ "learning_rate": 5.561705156228569e-07,
+ "loss": 0.008,
+ "step": 32903
+ },
+ {
+ "epoch": 89.65667574931881,
+ "grad_norm": 0.5847072005271912,
+ "learning_rate": 5.558803460166274e-07,
+ "loss": 0.0049,
+ "step": 32904
+ },
+ {
+ "epoch": 89.65940054495913,
+ "grad_norm": 1.3642884492874146,
+ "learning_rate": 5.555902499608956e-07,
+ "loss": 0.018,
+ "step": 32905
+ },
+ {
+ "epoch": 89.66212534059946,
+ "grad_norm": 1.17289400100708,
+ "learning_rate": 5.553002274579244e-07,
+ "loss": 0.0127,
+ "step": 32906
+ },
+ {
+ "epoch": 89.66485013623978,
+ "grad_norm": 1.4282149076461792,
+ "learning_rate": 5.550102785099687e-07,
+ "loss": 0.0241,
+ "step": 32907
+ },
+ {
+ "epoch": 89.66757493188011,
+ "grad_norm": 1.292185664176941,
+ "learning_rate": 5.547204031192921e-07,
+ "loss": 0.0862,
+ "step": 32908
+ },
+ {
+ "epoch": 89.67029972752043,
+ "grad_norm": 1.235490083694458,
+ "learning_rate": 5.544306012881462e-07,
+ "loss": 0.1911,
+ "step": 32909
+ },
+ {
+ "epoch": 89.67302452316076,
+ "grad_norm": 1.481964111328125,
+ "learning_rate": 5.541408730187925e-07,
+ "loss": 0.0167,
+ "step": 32910
+ },
+ {
+ "epoch": 89.6757493188011,
+ "grad_norm": 1.287192702293396,
+ "learning_rate": 5.538512183134859e-07,
+ "loss": 0.0641,
+ "step": 32911
+ },
+ {
+ "epoch": 89.67847411444141,
+ "grad_norm": 1.0809624195098877,
+ "learning_rate": 5.535616371744823e-07,
+ "loss": 0.0254,
+ "step": 32912
+ },
+ {
+ "epoch": 89.68119891008175,
+ "grad_norm": 0.8205623626708984,
+ "learning_rate": 5.532721296040344e-07,
+ "loss": 0.0355,
+ "step": 32913
+ },
+ {
+ "epoch": 89.68392370572207,
+ "grad_norm": 1.1392544507980347,
+ "learning_rate": 5.529826956044016e-07,
+ "loss": 0.0413,
+ "step": 32914
+ },
+ {
+ "epoch": 89.6866485013624,
+ "grad_norm": 0.6311061382293701,
+ "learning_rate": 5.526933351778352e-07,
+ "loss": 0.0082,
+ "step": 32915
+ },
+ {
+ "epoch": 89.68937329700273,
+ "grad_norm": 0.8290489912033081,
+ "learning_rate": 5.52404048326588e-07,
+ "loss": 0.0073,
+ "step": 32916
+ },
+ {
+ "epoch": 89.69209809264305,
+ "grad_norm": 0.59209805727005,
+ "learning_rate": 5.521148350529137e-07,
+ "loss": 0.0058,
+ "step": 32917
+ },
+ {
+ "epoch": 89.69482288828338,
+ "grad_norm": 1.0988407135009766,
+ "learning_rate": 5.518256953590661e-07,
+ "loss": 0.0454,
+ "step": 32918
+ },
+ {
+ "epoch": 89.6975476839237,
+ "grad_norm": 1.2256556749343872,
+ "learning_rate": 5.515366292472956e-07,
+ "loss": 0.0396,
+ "step": 32919
+ },
+ {
+ "epoch": 89.70027247956403,
+ "grad_norm": 1.8204495906829834,
+ "learning_rate": 5.512476367198538e-07,
+ "loss": 0.0093,
+ "step": 32920
+ },
+ {
+ "epoch": 89.70299727520435,
+ "grad_norm": 1.32970130443573,
+ "learning_rate": 5.509587177789911e-07,
+ "loss": 0.0188,
+ "step": 32921
+ },
+ {
+ "epoch": 89.70572207084469,
+ "grad_norm": 0.8919068574905396,
+ "learning_rate": 5.506698724269577e-07,
+ "loss": 0.011,
+ "step": 32922
+ },
+ {
+ "epoch": 89.70844686648502,
+ "grad_norm": 1.0556234121322632,
+ "learning_rate": 5.503811006660032e-07,
+ "loss": 0.1242,
+ "step": 32923
+ },
+ {
+ "epoch": 89.71117166212534,
+ "grad_norm": 0.5446197986602783,
+ "learning_rate": 5.500924024983756e-07,
+ "loss": 0.0065,
+ "step": 32924
+ },
+ {
+ "epoch": 89.71389645776567,
+ "grad_norm": 1.4084696769714355,
+ "learning_rate": 5.498037779263265e-07,
+ "loss": 0.1032,
+ "step": 32925
+ },
+ {
+ "epoch": 89.71662125340599,
+ "grad_norm": 1.5163460969924927,
+ "learning_rate": 5.495152269520987e-07,
+ "loss": 0.0485,
+ "step": 32926
+ },
+ {
+ "epoch": 89.71934604904632,
+ "grad_norm": 1.0121526718139648,
+ "learning_rate": 5.492267495779447e-07,
+ "loss": 0.017,
+ "step": 32927
+ },
+ {
+ "epoch": 89.72207084468666,
+ "grad_norm": 1.4447565078735352,
+ "learning_rate": 5.489383458061071e-07,
+ "loss": 0.0369,
+ "step": 32928
+ },
+ {
+ "epoch": 89.72479564032697,
+ "grad_norm": 1.0571105480194092,
+ "learning_rate": 5.48650015638833e-07,
+ "loss": 0.0129,
+ "step": 32929
+ },
+ {
+ "epoch": 89.7275204359673,
+ "grad_norm": 1.105916976928711,
+ "learning_rate": 5.483617590783696e-07,
+ "loss": 0.0549,
+ "step": 32930
+ },
+ {
+ "epoch": 89.73024523160763,
+ "grad_norm": 1.3987189531326294,
+ "learning_rate": 5.480735761269607e-07,
+ "loss": 0.0479,
+ "step": 32931
+ },
+ {
+ "epoch": 89.73297002724796,
+ "grad_norm": 4.564671039581299,
+ "learning_rate": 5.477854667868509e-07,
+ "loss": 0.0569,
+ "step": 32932
+ },
+ {
+ "epoch": 89.73569482288828,
+ "grad_norm": 0.5230503678321838,
+ "learning_rate": 5.474974310602821e-07,
+ "loss": 0.0048,
+ "step": 32933
+ },
+ {
+ "epoch": 89.73841961852861,
+ "grad_norm": 0.9951581954956055,
+ "learning_rate": 5.472094689495e-07,
+ "loss": 0.0097,
+ "step": 32934
+ },
+ {
+ "epoch": 89.74114441416894,
+ "grad_norm": 0.8574295043945312,
+ "learning_rate": 5.469215804567474e-07,
+ "loss": 0.0086,
+ "step": 32935
+ },
+ {
+ "epoch": 89.74386920980926,
+ "grad_norm": 0.7504445314407349,
+ "learning_rate": 5.466337655842647e-07,
+ "loss": 0.0083,
+ "step": 32936
+ },
+ {
+ "epoch": 89.7465940054496,
+ "grad_norm": 0.9419887065887451,
+ "learning_rate": 5.463460243342922e-07,
+ "loss": 0.0068,
+ "step": 32937
+ },
+ {
+ "epoch": 89.74931880108991,
+ "grad_norm": 0.8361653685569763,
+ "learning_rate": 5.460583567090738e-07,
+ "loss": 0.0149,
+ "step": 32938
+ },
+ {
+ "epoch": 89.75204359673025,
+ "grad_norm": 1.0283327102661133,
+ "learning_rate": 5.457707627108489e-07,
+ "loss": 0.0166,
+ "step": 32939
+ },
+ {
+ "epoch": 89.75476839237058,
+ "grad_norm": 1.4998918771743774,
+ "learning_rate": 5.454832423418577e-07,
+ "loss": 0.0932,
+ "step": 32940
+ },
+ {
+ "epoch": 89.7574931880109,
+ "grad_norm": 0.6857601404190063,
+ "learning_rate": 5.451957956043363e-07,
+ "loss": 0.006,
+ "step": 32941
+ },
+ {
+ "epoch": 89.76021798365123,
+ "grad_norm": 0.9103477001190186,
+ "learning_rate": 5.449084225005263e-07,
+ "loss": 0.0076,
+ "step": 32942
+ },
+ {
+ "epoch": 89.76294277929155,
+ "grad_norm": 3.280553102493286,
+ "learning_rate": 5.44621123032667e-07,
+ "loss": 0.1643,
+ "step": 32943
+ },
+ {
+ "epoch": 89.76566757493188,
+ "grad_norm": 1.16701340675354,
+ "learning_rate": 5.443338972029932e-07,
+ "loss": 0.0075,
+ "step": 32944
+ },
+ {
+ "epoch": 89.7683923705722,
+ "grad_norm": 5.575774669647217,
+ "learning_rate": 5.440467450137432e-07,
+ "loss": 0.0462,
+ "step": 32945
+ },
+ {
+ "epoch": 89.77111716621253,
+ "grad_norm": 1.064772605895996,
+ "learning_rate": 5.437596664671518e-07,
+ "loss": 0.0143,
+ "step": 32946
+ },
+ {
+ "epoch": 89.77384196185287,
+ "grad_norm": 0.9619712233543396,
+ "learning_rate": 5.434726615654562e-07,
+ "loss": 0.0661,
+ "step": 32947
+ },
+ {
+ "epoch": 89.77656675749319,
+ "grad_norm": 1.00767982006073,
+ "learning_rate": 5.431857303108923e-07,
+ "loss": 0.0148,
+ "step": 32948
+ },
+ {
+ "epoch": 89.77929155313352,
+ "grad_norm": 0.9132387638092041,
+ "learning_rate": 5.428988727056928e-07,
+ "loss": 0.0195,
+ "step": 32949
+ },
+ {
+ "epoch": 89.78201634877384,
+ "grad_norm": 1.6407126188278198,
+ "learning_rate": 5.426120887520914e-07,
+ "loss": 0.0313,
+ "step": 32950
+ },
+ {
+ "epoch": 89.78474114441417,
+ "grad_norm": 0.9640188813209534,
+ "learning_rate": 5.423253784523241e-07,
+ "loss": 0.0095,
+ "step": 32951
+ },
+ {
+ "epoch": 89.7874659400545,
+ "grad_norm": 0.7560170292854309,
+ "learning_rate": 5.420387418086215e-07,
+ "loss": 0.0063,
+ "step": 32952
+ },
+ {
+ "epoch": 89.79019073569482,
+ "grad_norm": 0.9145221710205078,
+ "learning_rate": 5.417521788232161e-07,
+ "loss": 0.0106,
+ "step": 32953
+ },
+ {
+ "epoch": 89.79291553133515,
+ "grad_norm": 1.1202096939086914,
+ "learning_rate": 5.414656894983416e-07,
+ "loss": 0.0112,
+ "step": 32954
+ },
+ {
+ "epoch": 89.79564032697547,
+ "grad_norm": 1.4867974519729614,
+ "learning_rate": 5.411792738362275e-07,
+ "loss": 0.041,
+ "step": 32955
+ },
+ {
+ "epoch": 89.7983651226158,
+ "grad_norm": 1.0216246843338013,
+ "learning_rate": 5.408929318391054e-07,
+ "loss": 0.0178,
+ "step": 32956
+ },
+ {
+ "epoch": 89.80108991825612,
+ "grad_norm": 0.8651511669158936,
+ "learning_rate": 5.406066635092034e-07,
+ "loss": 0.0074,
+ "step": 32957
+ },
+ {
+ "epoch": 89.80381471389646,
+ "grad_norm": 1.0882511138916016,
+ "learning_rate": 5.403204688487517e-07,
+ "loss": 0.0096,
+ "step": 32958
+ },
+ {
+ "epoch": 89.80653950953679,
+ "grad_norm": 1.3925551176071167,
+ "learning_rate": 5.400343478599812e-07,
+ "loss": 0.0106,
+ "step": 32959
+ },
+ {
+ "epoch": 89.80926430517711,
+ "grad_norm": 1.3650050163269043,
+ "learning_rate": 5.397483005451199e-07,
+ "loss": 0.0192,
+ "step": 32960
+ },
+ {
+ "epoch": 89.81198910081744,
+ "grad_norm": 0.6145510673522949,
+ "learning_rate": 5.394623269063926e-07,
+ "loss": 0.0047,
+ "step": 32961
+ },
+ {
+ "epoch": 89.81471389645776,
+ "grad_norm": 1.2875546216964722,
+ "learning_rate": 5.391764269460276e-07,
+ "loss": 0.0707,
+ "step": 32962
+ },
+ {
+ "epoch": 89.8174386920981,
+ "grad_norm": 1.0641230344772339,
+ "learning_rate": 5.388906006662542e-07,
+ "loss": 0.0063,
+ "step": 32963
+ },
+ {
+ "epoch": 89.82016348773843,
+ "grad_norm": 1.4162644147872925,
+ "learning_rate": 5.386048480692952e-07,
+ "loss": 0.0222,
+ "step": 32964
+ },
+ {
+ "epoch": 89.82288828337875,
+ "grad_norm": 0.6221365928649902,
+ "learning_rate": 5.383191691573775e-07,
+ "loss": 0.0065,
+ "step": 32965
+ },
+ {
+ "epoch": 89.82561307901908,
+ "grad_norm": 0.6882986426353455,
+ "learning_rate": 5.380335639327238e-07,
+ "loss": 0.0081,
+ "step": 32966
+ },
+ {
+ "epoch": 89.8283378746594,
+ "grad_norm": 1.1901817321777344,
+ "learning_rate": 5.377480323975626e-07,
+ "loss": 0.055,
+ "step": 32967
+ },
+ {
+ "epoch": 89.83106267029973,
+ "grad_norm": 1.557247519493103,
+ "learning_rate": 5.37462574554114e-07,
+ "loss": 0.052,
+ "step": 32968
+ },
+ {
+ "epoch": 89.83378746594005,
+ "grad_norm": 1.1979358196258545,
+ "learning_rate": 5.37177190404603e-07,
+ "loss": 0.0065,
+ "step": 32969
+ },
+ {
+ "epoch": 89.83651226158038,
+ "grad_norm": 1.177612066268921,
+ "learning_rate": 5.368918799512502e-07,
+ "loss": 0.0218,
+ "step": 32970
+ },
+ {
+ "epoch": 89.83923705722071,
+ "grad_norm": 1.010812520980835,
+ "learning_rate": 5.366066431962802e-07,
+ "loss": 0.0838,
+ "step": 32971
+ },
+ {
+ "epoch": 89.84196185286103,
+ "grad_norm": 1.363764762878418,
+ "learning_rate": 5.363214801419126e-07,
+ "loss": 0.1406,
+ "step": 32972
+ },
+ {
+ "epoch": 89.84468664850137,
+ "grad_norm": 1.4778685569763184,
+ "learning_rate": 5.360363907903676e-07,
+ "loss": 0.0368,
+ "step": 32973
+ },
+ {
+ "epoch": 89.84741144414168,
+ "grad_norm": 1.2788587808609009,
+ "learning_rate": 5.357513751438682e-07,
+ "loss": 0.0223,
+ "step": 32974
+ },
+ {
+ "epoch": 89.85013623978202,
+ "grad_norm": 0.8828480839729309,
+ "learning_rate": 5.354664332046311e-07,
+ "loss": 0.009,
+ "step": 32975
+ },
+ {
+ "epoch": 89.85286103542235,
+ "grad_norm": 0.7271072268486023,
+ "learning_rate": 5.351815649748782e-07,
+ "loss": 0.0062,
+ "step": 32976
+ },
+ {
+ "epoch": 89.85558583106267,
+ "grad_norm": 0.7748902440071106,
+ "learning_rate": 5.348967704568265e-07,
+ "loss": 0.0102,
+ "step": 32977
+ },
+ {
+ "epoch": 89.858310626703,
+ "grad_norm": 1.0003039836883545,
+ "learning_rate": 5.34612049652693e-07,
+ "loss": 0.0783,
+ "step": 32978
+ },
+ {
+ "epoch": 89.86103542234332,
+ "grad_norm": 1.3289248943328857,
+ "learning_rate": 5.343274025646972e-07,
+ "loss": 0.0088,
+ "step": 32979
+ },
+ {
+ "epoch": 89.86376021798365,
+ "grad_norm": 0.9725673198699951,
+ "learning_rate": 5.34042829195055e-07,
+ "loss": 0.0179,
+ "step": 32980
+ },
+ {
+ "epoch": 89.86648501362397,
+ "grad_norm": 1.3651012182235718,
+ "learning_rate": 5.337583295459825e-07,
+ "loss": 0.0763,
+ "step": 32981
+ },
+ {
+ "epoch": 89.8692098092643,
+ "grad_norm": 0.668754518032074,
+ "learning_rate": 5.334739036196945e-07,
+ "loss": 0.0093,
+ "step": 32982
+ },
+ {
+ "epoch": 89.87193460490464,
+ "grad_norm": 0.3638509511947632,
+ "learning_rate": 5.331895514184083e-07,
+ "loss": 0.0039,
+ "step": 32983
+ },
+ {
+ "epoch": 89.87465940054496,
+ "grad_norm": 0.8588996529579163,
+ "learning_rate": 5.329052729443363e-07,
+ "loss": 0.0059,
+ "step": 32984
+ },
+ {
+ "epoch": 89.87738419618529,
+ "grad_norm": 0.8250793218612671,
+ "learning_rate": 5.326210681996947e-07,
+ "loss": 0.0139,
+ "step": 32985
+ },
+ {
+ "epoch": 89.88010899182561,
+ "grad_norm": 1.3221673965454102,
+ "learning_rate": 5.323369371866926e-07,
+ "loss": 0.0135,
+ "step": 32986
+ },
+ {
+ "epoch": 89.88283378746594,
+ "grad_norm": 0.33950167894363403,
+ "learning_rate": 5.320528799075486e-07,
+ "loss": 0.0034,
+ "step": 32987
+ },
+ {
+ "epoch": 89.88555858310627,
+ "grad_norm": 1.3884479999542236,
+ "learning_rate": 5.317688963644718e-07,
+ "loss": 0.0951,
+ "step": 32988
+ },
+ {
+ "epoch": 89.88828337874659,
+ "grad_norm": 1.0589570999145508,
+ "learning_rate": 5.314849865596738e-07,
+ "loss": 0.012,
+ "step": 32989
+ },
+ {
+ "epoch": 89.89100817438693,
+ "grad_norm": 1.1022061109542847,
+ "learning_rate": 5.312011504953641e-07,
+ "loss": 0.0166,
+ "step": 32990
+ },
+ {
+ "epoch": 89.89373297002724,
+ "grad_norm": 0.7353715300559998,
+ "learning_rate": 5.309173881737573e-07,
+ "loss": 0.0068,
+ "step": 32991
+ },
+ {
+ "epoch": 89.89645776566758,
+ "grad_norm": 0.671091616153717,
+ "learning_rate": 5.306336995970607e-07,
+ "loss": 0.0062,
+ "step": 32992
+ },
+ {
+ "epoch": 89.8991825613079,
+ "grad_norm": 0.9259076714515686,
+ "learning_rate": 5.303500847674825e-07,
+ "loss": 0.1392,
+ "step": 32993
+ },
+ {
+ "epoch": 89.90190735694823,
+ "grad_norm": 1.1121870279312134,
+ "learning_rate": 5.300665436872332e-07,
+ "loss": 0.0233,
+ "step": 32994
+ },
+ {
+ "epoch": 89.90463215258856,
+ "grad_norm": 1.2188806533813477,
+ "learning_rate": 5.297830763585221e-07,
+ "loss": 0.0211,
+ "step": 32995
+ },
+ {
+ "epoch": 89.90735694822888,
+ "grad_norm": 0.42307665944099426,
+ "learning_rate": 5.294996827835564e-07,
+ "loss": 0.005,
+ "step": 32996
+ },
+ {
+ "epoch": 89.91008174386921,
+ "grad_norm": 0.48339200019836426,
+ "learning_rate": 5.292163629645419e-07,
+ "loss": 0.0049,
+ "step": 32997
+ },
+ {
+ "epoch": 89.91280653950953,
+ "grad_norm": 1.3164972066879272,
+ "learning_rate": 5.289331169036838e-07,
+ "loss": 0.0099,
+ "step": 32998
+ },
+ {
+ "epoch": 89.91553133514986,
+ "grad_norm": 0.6441060900688171,
+ "learning_rate": 5.286499446031912e-07,
+ "loss": 0.0046,
+ "step": 32999
+ },
+ {
+ "epoch": 89.9182561307902,
+ "grad_norm": 1.093172550201416,
+ "learning_rate": 5.28366846065268e-07,
+ "loss": 0.0468,
+ "step": 33000
+ },
+ {
+ "epoch": 89.92098092643052,
+ "grad_norm": 0.9331873059272766,
+ "learning_rate": 5.280838212921192e-07,
+ "loss": 0.0148,
+ "step": 33001
+ },
+ {
+ "epoch": 89.92370572207085,
+ "grad_norm": 1.3999714851379395,
+ "learning_rate": 5.278008702859471e-07,
+ "loss": 0.0125,
+ "step": 33002
+ },
+ {
+ "epoch": 89.92643051771117,
+ "grad_norm": 1.1328109502792358,
+ "learning_rate": 5.275179930489593e-07,
+ "loss": 0.0176,
+ "step": 33003
+ },
+ {
+ "epoch": 89.9291553133515,
+ "grad_norm": 1.7585058212280273,
+ "learning_rate": 5.272351895833561e-07,
+ "loss": 0.0873,
+ "step": 33004
+ },
+ {
+ "epoch": 89.93188010899182,
+ "grad_norm": 0.44578614830970764,
+ "learning_rate": 5.2695245989134e-07,
+ "loss": 0.0049,
+ "step": 33005
+ },
+ {
+ "epoch": 89.93460490463215,
+ "grad_norm": 0.6228973865509033,
+ "learning_rate": 5.266698039751117e-07,
+ "loss": 0.0032,
+ "step": 33006
+ },
+ {
+ "epoch": 89.93732970027249,
+ "grad_norm": 0.4320741295814514,
+ "learning_rate": 5.263872218368759e-07,
+ "loss": 0.0051,
+ "step": 33007
+ },
+ {
+ "epoch": 89.9400544959128,
+ "grad_norm": 0.6771335601806641,
+ "learning_rate": 5.261047134788311e-07,
+ "loss": 0.0096,
+ "step": 33008
+ },
+ {
+ "epoch": 89.94277929155314,
+ "grad_norm": 0.5931823253631592,
+ "learning_rate": 5.258222789031775e-07,
+ "loss": 0.0066,
+ "step": 33009
+ },
+ {
+ "epoch": 89.94550408719346,
+ "grad_norm": 1.442533016204834,
+ "learning_rate": 5.255399181121146e-07,
+ "loss": 0.0173,
+ "step": 33010
+ },
+ {
+ "epoch": 89.94822888283379,
+ "grad_norm": 1.2267720699310303,
+ "learning_rate": 5.252576311078417e-07,
+ "loss": 0.0206,
+ "step": 33011
+ },
+ {
+ "epoch": 89.95095367847412,
+ "grad_norm": 0.5429529547691345,
+ "learning_rate": 5.249754178925581e-07,
+ "loss": 0.0053,
+ "step": 33012
+ },
+ {
+ "epoch": 89.95367847411444,
+ "grad_norm": 1.1633669137954712,
+ "learning_rate": 5.246932784684622e-07,
+ "loss": 0.0137,
+ "step": 33013
+ },
+ {
+ "epoch": 89.95640326975477,
+ "grad_norm": 0.7436643838882446,
+ "learning_rate": 5.244112128377477e-07,
+ "loss": 0.0105,
+ "step": 33014
+ },
+ {
+ "epoch": 89.95912806539509,
+ "grad_norm": 1.3171460628509521,
+ "learning_rate": 5.241292210026161e-07,
+ "loss": 0.1149,
+ "step": 33015
+ },
+ {
+ "epoch": 89.96185286103542,
+ "grad_norm": 0.9404233694076538,
+ "learning_rate": 5.238473029652602e-07,
+ "loss": 0.0057,
+ "step": 33016
+ },
+ {
+ "epoch": 89.96457765667574,
+ "grad_norm": 0.8409035801887512,
+ "learning_rate": 5.235654587278771e-07,
+ "loss": 0.0093,
+ "step": 33017
+ },
+ {
+ "epoch": 89.96730245231608,
+ "grad_norm": 0.9187071323394775,
+ "learning_rate": 5.232836882926596e-07,
+ "loss": 0.0111,
+ "step": 33018
+ },
+ {
+ "epoch": 89.97002724795641,
+ "grad_norm": 1.2913093566894531,
+ "learning_rate": 5.230019916618046e-07,
+ "loss": 0.0205,
+ "step": 33019
+ },
+ {
+ "epoch": 89.97275204359673,
+ "grad_norm": 1.0855611562728882,
+ "learning_rate": 5.227203688375049e-07,
+ "loss": 0.0787,
+ "step": 33020
+ },
+ {
+ "epoch": 89.97547683923706,
+ "grad_norm": 0.49704375863075256,
+ "learning_rate": 5.224388198219554e-07,
+ "loss": 0.0042,
+ "step": 33021
+ },
+ {
+ "epoch": 89.97820163487738,
+ "grad_norm": 0.9584676027297974,
+ "learning_rate": 5.221573446173444e-07,
+ "loss": 0.0171,
+ "step": 33022
+ },
+ {
+ "epoch": 89.98092643051771,
+ "grad_norm": 1.9802955389022827,
+ "learning_rate": 5.218759432258691e-07,
+ "loss": 0.0144,
+ "step": 33023
+ },
+ {
+ "epoch": 89.98365122615803,
+ "grad_norm": 0.7261468768119812,
+ "learning_rate": 5.215946156497187e-07,
+ "loss": 0.0054,
+ "step": 33024
+ },
+ {
+ "epoch": 89.98637602179836,
+ "grad_norm": 0.9996471405029297,
+ "learning_rate": 5.213133618910849e-07,
+ "loss": 0.0157,
+ "step": 33025
+ },
+ {
+ "epoch": 89.9891008174387,
+ "grad_norm": 0.8657174706459045,
+ "learning_rate": 5.210321819521558e-07,
+ "loss": 0.008,
+ "step": 33026
+ },
+ {
+ "epoch": 89.99182561307902,
+ "grad_norm": 0.9337591528892517,
+ "learning_rate": 5.207510758351231e-07,
+ "loss": 0.0088,
+ "step": 33027
+ },
+ {
+ "epoch": 89.99455040871935,
+ "grad_norm": 0.37716302275657654,
+ "learning_rate": 5.204700435421772e-07,
+ "loss": 0.0028,
+ "step": 33028
+ },
+ {
+ "epoch": 89.99727520435967,
+ "grad_norm": 0.6254814267158508,
+ "learning_rate": 5.201890850755054e-07,
+ "loss": 0.011,
+ "step": 33029
+ },
+ {
+ "epoch": 90.0,
+ "grad_norm": 1.4187566041946411,
+ "learning_rate": 5.199082004372958e-07,
+ "loss": 0.0099,
+ "step": 33030
+ },
+ {
+ "epoch": 90.00272479564033,
+ "grad_norm": 0.925563633441925,
+ "learning_rate": 5.196273896297354e-07,
+ "loss": 0.0164,
+ "step": 33031
+ },
+ {
+ "epoch": 90.00544959128065,
+ "grad_norm": 1.4483182430267334,
+ "learning_rate": 5.193466526550117e-07,
+ "loss": 0.0109,
+ "step": 33032
+ },
+ {
+ "epoch": 90.00817438692098,
+ "grad_norm": 1.548980474472046,
+ "learning_rate": 5.190659895153127e-07,
+ "loss": 0.0226,
+ "step": 33033
+ },
+ {
+ "epoch": 90.0108991825613,
+ "grad_norm": 1.5153781175613403,
+ "learning_rate": 5.187854002128223e-07,
+ "loss": 0.0126,
+ "step": 33034
+ },
+ {
+ "epoch": 90.01362397820164,
+ "grad_norm": 0.9039390087127686,
+ "learning_rate": 5.185048847497242e-07,
+ "loss": 0.0101,
+ "step": 33035
+ },
+ {
+ "epoch": 90.01634877384195,
+ "grad_norm": 0.5790703296661377,
+ "learning_rate": 5.182244431282058e-07,
+ "loss": 0.005,
+ "step": 33036
+ },
+ {
+ "epoch": 90.01907356948229,
+ "grad_norm": 0.9822517037391663,
+ "learning_rate": 5.179440753504517e-07,
+ "loss": 0.0111,
+ "step": 33037
+ },
+ {
+ "epoch": 90.02179836512262,
+ "grad_norm": 1.3503514528274536,
+ "learning_rate": 5.176637814186414e-07,
+ "loss": 0.0103,
+ "step": 33038
+ },
+ {
+ "epoch": 90.02452316076294,
+ "grad_norm": 0.7494741678237915,
+ "learning_rate": 5.17383561334962e-07,
+ "loss": 0.0085,
+ "step": 33039
+ },
+ {
+ "epoch": 90.02724795640327,
+ "grad_norm": 0.9891656041145325,
+ "learning_rate": 5.171034151015941e-07,
+ "loss": 0.0104,
+ "step": 33040
+ },
+ {
+ "epoch": 90.02997275204359,
+ "grad_norm": 0.7627041935920715,
+ "learning_rate": 5.168233427207204e-07,
+ "loss": 0.0448,
+ "step": 33041
+ },
+ {
+ "epoch": 90.03269754768392,
+ "grad_norm": 1.0562723875045776,
+ "learning_rate": 5.165433441945189e-07,
+ "loss": 0.0398,
+ "step": 33042
+ },
+ {
+ "epoch": 90.03542234332426,
+ "grad_norm": 0.47078797221183777,
+ "learning_rate": 5.162634195251748e-07,
+ "loss": 0.0041,
+ "step": 33043
+ },
+ {
+ "epoch": 90.03814713896458,
+ "grad_norm": 0.6099017262458801,
+ "learning_rate": 5.15983568714864e-07,
+ "loss": 0.0062,
+ "step": 33044
+ },
+ {
+ "epoch": 90.04087193460491,
+ "grad_norm": 0.9222371578216553,
+ "learning_rate": 5.157037917657692e-07,
+ "loss": 0.0062,
+ "step": 33045
+ },
+ {
+ "epoch": 90.04359673024523,
+ "grad_norm": 1.1705721616744995,
+ "learning_rate": 5.154240886800688e-07,
+ "loss": 0.0111,
+ "step": 33046
+ },
+ {
+ "epoch": 90.04632152588556,
+ "grad_norm": 0.5432872176170349,
+ "learning_rate": 5.151444594599386e-07,
+ "loss": 0.0082,
+ "step": 33047
+ },
+ {
+ "epoch": 90.04904632152588,
+ "grad_norm": 0.6005358099937439,
+ "learning_rate": 5.148649041075593e-07,
+ "loss": 0.005,
+ "step": 33048
+ },
+ {
+ "epoch": 90.05177111716621,
+ "grad_norm": 1.286275863647461,
+ "learning_rate": 5.145854226251068e-07,
+ "loss": 0.0169,
+ "step": 33049
+ },
+ {
+ "epoch": 90.05449591280654,
+ "grad_norm": 0.8032387495040894,
+ "learning_rate": 5.143060150147583e-07,
+ "loss": 0.0068,
+ "step": 33050
+ },
+ {
+ "epoch": 90.05722070844686,
+ "grad_norm": 0.8737594485282898,
+ "learning_rate": 5.140266812786865e-07,
+ "loss": 0.0084,
+ "step": 33051
+ },
+ {
+ "epoch": 90.0599455040872,
+ "grad_norm": 3.6475412845611572,
+ "learning_rate": 5.137474214190719e-07,
+ "loss": 0.0534,
+ "step": 33052
+ },
+ {
+ "epoch": 90.06267029972751,
+ "grad_norm": 0.9473435282707214,
+ "learning_rate": 5.134682354380871e-07,
+ "loss": 0.0105,
+ "step": 33053
+ },
+ {
+ "epoch": 90.06539509536785,
+ "grad_norm": 0.9933578968048096,
+ "learning_rate": 5.131891233379071e-07,
+ "loss": 0.0196,
+ "step": 33054
+ },
+ {
+ "epoch": 90.06811989100818,
+ "grad_norm": 2.0395312309265137,
+ "learning_rate": 5.129100851207025e-07,
+ "loss": 0.2192,
+ "step": 33055
+ },
+ {
+ "epoch": 90.0708446866485,
+ "grad_norm": 1.2871400117874146,
+ "learning_rate": 5.126311207886503e-07,
+ "loss": 0.101,
+ "step": 33056
+ },
+ {
+ "epoch": 90.07356948228883,
+ "grad_norm": 1.651120901107788,
+ "learning_rate": 5.123522303439222e-07,
+ "loss": 0.0174,
+ "step": 33057
+ },
+ {
+ "epoch": 90.07629427792915,
+ "grad_norm": 0.9830305576324463,
+ "learning_rate": 5.120734137886884e-07,
+ "loss": 0.0077,
+ "step": 33058
+ },
+ {
+ "epoch": 90.07901907356948,
+ "grad_norm": 1.3652156591415405,
+ "learning_rate": 5.117946711251209e-07,
+ "loss": 0.0219,
+ "step": 33059
+ },
+ {
+ "epoch": 90.0817438692098,
+ "grad_norm": 1.952669620513916,
+ "learning_rate": 5.115160023553922e-07,
+ "loss": 0.0961,
+ "step": 33060
+ },
+ {
+ "epoch": 90.08446866485014,
+ "grad_norm": 1.3449866771697998,
+ "learning_rate": 5.112374074816706e-07,
+ "loss": 0.0196,
+ "step": 33061
+ },
+ {
+ "epoch": 90.08719346049047,
+ "grad_norm": 0.8451568484306335,
+ "learning_rate": 5.109588865061277e-07,
+ "loss": 0.01,
+ "step": 33062
+ },
+ {
+ "epoch": 90.08991825613079,
+ "grad_norm": 0.4195702373981476,
+ "learning_rate": 5.106804394309306e-07,
+ "loss": 0.0052,
+ "step": 33063
+ },
+ {
+ "epoch": 90.09264305177112,
+ "grad_norm": 0.938915491104126,
+ "learning_rate": 5.104020662582499e-07,
+ "loss": 0.0087,
+ "step": 33064
+ },
+ {
+ "epoch": 90.09536784741144,
+ "grad_norm": 1.590753197669983,
+ "learning_rate": 5.101237669902526e-07,
+ "loss": 0.15,
+ "step": 33065
+ },
+ {
+ "epoch": 90.09809264305177,
+ "grad_norm": 1.154801845550537,
+ "learning_rate": 5.09845541629106e-07,
+ "loss": 0.0166,
+ "step": 33066
+ },
+ {
+ "epoch": 90.1008174386921,
+ "grad_norm": 1.1423015594482422,
+ "learning_rate": 5.095673901769749e-07,
+ "loss": 0.0668,
+ "step": 33067
+ },
+ {
+ "epoch": 90.10354223433242,
+ "grad_norm": 0.8521717190742493,
+ "learning_rate": 5.092893126360288e-07,
+ "loss": 0.0372,
+ "step": 33068
+ },
+ {
+ "epoch": 90.10626702997276,
+ "grad_norm": 1.3382155895233154,
+ "learning_rate": 5.090113090084325e-07,
+ "loss": 0.0274,
+ "step": 33069
+ },
+ {
+ "epoch": 90.10899182561307,
+ "grad_norm": 0.781175971031189,
+ "learning_rate": 5.087333792963512e-07,
+ "loss": 0.0086,
+ "step": 33070
+ },
+ {
+ "epoch": 90.11171662125341,
+ "grad_norm": 1.3478001356124878,
+ "learning_rate": 5.084555235019473e-07,
+ "loss": 0.0398,
+ "step": 33071
+ },
+ {
+ "epoch": 90.11444141689373,
+ "grad_norm": 0.9458490610122681,
+ "learning_rate": 5.081777416273869e-07,
+ "loss": 0.0096,
+ "step": 33072
+ },
+ {
+ "epoch": 90.11716621253406,
+ "grad_norm": 0.5571717619895935,
+ "learning_rate": 5.079000336748341e-07,
+ "loss": 0.0056,
+ "step": 33073
+ },
+ {
+ "epoch": 90.11989100817439,
+ "grad_norm": 0.7890389561653137,
+ "learning_rate": 5.076223996464491e-07,
+ "loss": 0.0082,
+ "step": 33074
+ },
+ {
+ "epoch": 90.12261580381471,
+ "grad_norm": 2.856292963027954,
+ "learning_rate": 5.073448395443948e-07,
+ "loss": 0.0625,
+ "step": 33075
+ },
+ {
+ "epoch": 90.12534059945504,
+ "grad_norm": 0.8279886245727539,
+ "learning_rate": 5.070673533708337e-07,
+ "loss": 0.013,
+ "step": 33076
+ },
+ {
+ "epoch": 90.12806539509536,
+ "grad_norm": 1.0462496280670166,
+ "learning_rate": 5.067899411279276e-07,
+ "loss": 0.0127,
+ "step": 33077
+ },
+ {
+ "epoch": 90.1307901907357,
+ "grad_norm": 1.5251785516738892,
+ "learning_rate": 5.065126028178357e-07,
+ "loss": 0.0387,
+ "step": 33078
+ },
+ {
+ "epoch": 90.13351498637603,
+ "grad_norm": 1.3685859441757202,
+ "learning_rate": 5.062353384427144e-07,
+ "loss": 0.0239,
+ "step": 33079
+ },
+ {
+ "epoch": 90.13623978201635,
+ "grad_norm": 1.3901315927505493,
+ "learning_rate": 5.059581480047304e-07,
+ "loss": 0.0445,
+ "step": 33080
+ },
+ {
+ "epoch": 90.13896457765668,
+ "grad_norm": 2.376812219619751,
+ "learning_rate": 5.05681031506039e-07,
+ "loss": 0.026,
+ "step": 33081
+ },
+ {
+ "epoch": 90.141689373297,
+ "grad_norm": 1.150416612625122,
+ "learning_rate": 5.054039889487972e-07,
+ "loss": 0.0431,
+ "step": 33082
+ },
+ {
+ "epoch": 90.14441416893733,
+ "grad_norm": 4.112997055053711,
+ "learning_rate": 5.051270203351632e-07,
+ "loss": 0.0055,
+ "step": 33083
+ },
+ {
+ "epoch": 90.14713896457765,
+ "grad_norm": 1.238270878791809,
+ "learning_rate": 5.048501256672955e-07,
+ "loss": 0.1491,
+ "step": 33084
+ },
+ {
+ "epoch": 90.14986376021798,
+ "grad_norm": 1.2997932434082031,
+ "learning_rate": 5.045733049473489e-07,
+ "loss": 0.0402,
+ "step": 33085
+ },
+ {
+ "epoch": 90.15258855585832,
+ "grad_norm": 1.1296288967132568,
+ "learning_rate": 5.042965581774794e-07,
+ "loss": 0.0792,
+ "step": 33086
+ },
+ {
+ "epoch": 90.15531335149863,
+ "grad_norm": 1.2433183193206787,
+ "learning_rate": 5.040198853598421e-07,
+ "loss": 0.0491,
+ "step": 33087
+ },
+ {
+ "epoch": 90.15803814713897,
+ "grad_norm": 0.6505377888679504,
+ "learning_rate": 5.037432864965941e-07,
+ "loss": 0.005,
+ "step": 33088
+ },
+ {
+ "epoch": 90.16076294277929,
+ "grad_norm": 2.5210673809051514,
+ "learning_rate": 5.034667615898869e-07,
+ "loss": 0.031,
+ "step": 33089
+ },
+ {
+ "epoch": 90.16348773841962,
+ "grad_norm": 1.4890064001083374,
+ "learning_rate": 5.031903106418756e-07,
+ "loss": 0.0236,
+ "step": 33090
+ },
+ {
+ "epoch": 90.16621253405995,
+ "grad_norm": 0.8299844264984131,
+ "learning_rate": 5.029139336547106e-07,
+ "loss": 0.0605,
+ "step": 33091
+ },
+ {
+ "epoch": 90.16893732970027,
+ "grad_norm": 1.1252647638320923,
+ "learning_rate": 5.02637630630548e-07,
+ "loss": 0.024,
+ "step": 33092
+ },
+ {
+ "epoch": 90.1716621253406,
+ "grad_norm": 0.9544477462768555,
+ "learning_rate": 5.023614015715372e-07,
+ "loss": 0.0351,
+ "step": 33093
+ },
+ {
+ "epoch": 90.17438692098092,
+ "grad_norm": 0.7989944219589233,
+ "learning_rate": 5.020852464798309e-07,
+ "loss": 0.0124,
+ "step": 33094
+ },
+ {
+ "epoch": 90.17711171662125,
+ "grad_norm": 0.36710843443870544,
+ "learning_rate": 5.018091653575774e-07,
+ "loss": 0.0036,
+ "step": 33095
+ },
+ {
+ "epoch": 90.17983651226157,
+ "grad_norm": 1.3575284481048584,
+ "learning_rate": 5.015331582069282e-07,
+ "loss": 0.0118,
+ "step": 33096
+ },
+ {
+ "epoch": 90.1825613079019,
+ "grad_norm": 1.9591059684753418,
+ "learning_rate": 5.01257225030034e-07,
+ "loss": 0.0376,
+ "step": 33097
+ },
+ {
+ "epoch": 90.18528610354224,
+ "grad_norm": 0.9154413938522339,
+ "learning_rate": 5.009813658290441e-07,
+ "loss": 0.0136,
+ "step": 33098
+ },
+ {
+ "epoch": 90.18801089918256,
+ "grad_norm": 0.4420170485973358,
+ "learning_rate": 5.007055806061045e-07,
+ "loss": 0.0048,
+ "step": 33099
+ },
+ {
+ "epoch": 90.19073569482289,
+ "grad_norm": 1.0771362781524658,
+ "learning_rate": 5.004298693633625e-07,
+ "loss": 0.0117,
+ "step": 33100
+ },
+ {
+ "epoch": 90.19346049046321,
+ "grad_norm": 0.909327507019043,
+ "learning_rate": 5.001542321029695e-07,
+ "loss": 0.0335,
+ "step": 33101
+ },
+ {
+ "epoch": 90.19618528610354,
+ "grad_norm": 0.598884642124176,
+ "learning_rate": 4.998786688270685e-07,
+ "loss": 0.0071,
+ "step": 33102
+ },
+ {
+ "epoch": 90.19891008174388,
+ "grad_norm": 0.7449193596839905,
+ "learning_rate": 4.996031795378065e-07,
+ "loss": 0.0085,
+ "step": 33103
+ },
+ {
+ "epoch": 90.2016348773842,
+ "grad_norm": 1.0967881679534912,
+ "learning_rate": 4.993277642373296e-07,
+ "loss": 0.1377,
+ "step": 33104
+ },
+ {
+ "epoch": 90.20435967302453,
+ "grad_norm": 1.1062391996383667,
+ "learning_rate": 4.990524229277826e-07,
+ "loss": 0.014,
+ "step": 33105
+ },
+ {
+ "epoch": 90.20708446866485,
+ "grad_norm": 1.2727203369140625,
+ "learning_rate": 4.987771556113096e-07,
+ "loss": 0.0133,
+ "step": 33106
+ },
+ {
+ "epoch": 90.20980926430518,
+ "grad_norm": 1.406898856163025,
+ "learning_rate": 4.985019622900522e-07,
+ "loss": 0.0573,
+ "step": 33107
+ },
+ {
+ "epoch": 90.2125340599455,
+ "grad_norm": 0.8195104598999023,
+ "learning_rate": 4.982268429661575e-07,
+ "loss": 0.0084,
+ "step": 33108
+ },
+ {
+ "epoch": 90.21525885558583,
+ "grad_norm": 1.1116247177124023,
+ "learning_rate": 4.979517976417669e-07,
+ "loss": 0.0118,
+ "step": 33109
+ },
+ {
+ "epoch": 90.21798365122616,
+ "grad_norm": 0.7842905521392822,
+ "learning_rate": 4.976768263190212e-07,
+ "loss": 0.0075,
+ "step": 33110
+ },
+ {
+ "epoch": 90.22070844686648,
+ "grad_norm": 1.425728678703308,
+ "learning_rate": 4.97401929000062e-07,
+ "loss": 0.0508,
+ "step": 33111
+ },
+ {
+ "epoch": 90.22343324250681,
+ "grad_norm": 1.5794662237167358,
+ "learning_rate": 4.971271056870308e-07,
+ "loss": 0.0518,
+ "step": 33112
+ },
+ {
+ "epoch": 90.22615803814713,
+ "grad_norm": 0.5257372856140137,
+ "learning_rate": 4.968523563820682e-07,
+ "loss": 0.007,
+ "step": 33113
+ },
+ {
+ "epoch": 90.22888283378747,
+ "grad_norm": 0.9364306926727295,
+ "learning_rate": 4.965776810873135e-07,
+ "loss": 0.0142,
+ "step": 33114
+ },
+ {
+ "epoch": 90.2316076294278,
+ "grad_norm": 1.2492127418518066,
+ "learning_rate": 4.963030798049063e-07,
+ "loss": 0.0064,
+ "step": 33115
+ },
+ {
+ "epoch": 90.23433242506812,
+ "grad_norm": 0.7311891317367554,
+ "learning_rate": 4.960285525369834e-07,
+ "loss": 0.007,
+ "step": 33116
+ },
+ {
+ "epoch": 90.23705722070845,
+ "grad_norm": 1.8515942096710205,
+ "learning_rate": 4.957540992856869e-07,
+ "loss": 0.0603,
+ "step": 33117
+ },
+ {
+ "epoch": 90.23978201634877,
+ "grad_norm": 0.9356517195701599,
+ "learning_rate": 4.954797200531502e-07,
+ "loss": 0.0159,
+ "step": 33118
+ },
+ {
+ "epoch": 90.2425068119891,
+ "grad_norm": 0.8691231608390808,
+ "learning_rate": 4.95205414841512e-07,
+ "loss": 0.0114,
+ "step": 33119
+ },
+ {
+ "epoch": 90.24523160762942,
+ "grad_norm": 1.1187107563018799,
+ "learning_rate": 4.949311836529059e-07,
+ "loss": 0.0118,
+ "step": 33120
+ },
+ {
+ "epoch": 90.24795640326975,
+ "grad_norm": 0.9724262356758118,
+ "learning_rate": 4.946570264894724e-07,
+ "loss": 0.024,
+ "step": 33121
+ },
+ {
+ "epoch": 90.25068119891009,
+ "grad_norm": 2.114525079727173,
+ "learning_rate": 4.943829433533431e-07,
+ "loss": 0.0696,
+ "step": 33122
+ },
+ {
+ "epoch": 90.2534059945504,
+ "grad_norm": 0.7640067338943481,
+ "learning_rate": 4.941089342466543e-07,
+ "loss": 0.0065,
+ "step": 33123
+ },
+ {
+ "epoch": 90.25613079019074,
+ "grad_norm": 1.1178407669067383,
+ "learning_rate": 4.938349991715375e-07,
+ "loss": 0.0297,
+ "step": 33124
+ },
+ {
+ "epoch": 90.25885558583106,
+ "grad_norm": 0.8121495842933655,
+ "learning_rate": 4.935611381301298e-07,
+ "loss": 0.0051,
+ "step": 33125
+ },
+ {
+ "epoch": 90.26158038147139,
+ "grad_norm": 0.6097666621208191,
+ "learning_rate": 4.932873511245606e-07,
+ "loss": 0.0069,
+ "step": 33126
+ },
+ {
+ "epoch": 90.26430517711172,
+ "grad_norm": 1.0923436880111694,
+ "learning_rate": 4.930136381569628e-07,
+ "loss": 0.0584,
+ "step": 33127
+ },
+ {
+ "epoch": 90.26702997275204,
+ "grad_norm": 0.9428333044052124,
+ "learning_rate": 4.927399992294701e-07,
+ "loss": 0.119,
+ "step": 33128
+ },
+ {
+ "epoch": 90.26975476839237,
+ "grad_norm": 0.8203036785125732,
+ "learning_rate": 4.924664343442132e-07,
+ "loss": 0.019,
+ "step": 33129
+ },
+ {
+ "epoch": 90.2724795640327,
+ "grad_norm": 1.1886911392211914,
+ "learning_rate": 4.92192943503319e-07,
+ "loss": 0.0364,
+ "step": 33130
+ },
+ {
+ "epoch": 90.27520435967303,
+ "grad_norm": 1.4938709735870361,
+ "learning_rate": 4.919195267089227e-07,
+ "loss": 0.0089,
+ "step": 33131
+ },
+ {
+ "epoch": 90.27792915531334,
+ "grad_norm": 1.2831131219863892,
+ "learning_rate": 4.916461839631492e-07,
+ "loss": 0.1057,
+ "step": 33132
+ },
+ {
+ "epoch": 90.28065395095368,
+ "grad_norm": 2.694809675216675,
+ "learning_rate": 4.913729152681302e-07,
+ "loss": 0.0379,
+ "step": 33133
+ },
+ {
+ "epoch": 90.28337874659401,
+ "grad_norm": 0.6072165966033936,
+ "learning_rate": 4.910997206259927e-07,
+ "loss": 0.0067,
+ "step": 33134
+ },
+ {
+ "epoch": 90.28610354223433,
+ "grad_norm": 0.8132880330085754,
+ "learning_rate": 4.908266000388651e-07,
+ "loss": 0.0071,
+ "step": 33135
+ },
+ {
+ "epoch": 90.28882833787466,
+ "grad_norm": 0.8146174550056458,
+ "learning_rate": 4.905535535088724e-07,
+ "loss": 0.0082,
+ "step": 33136
+ },
+ {
+ "epoch": 90.29155313351498,
+ "grad_norm": 1.022404670715332,
+ "learning_rate": 4.902805810381439e-07,
+ "loss": 0.0209,
+ "step": 33137
+ },
+ {
+ "epoch": 90.29427792915531,
+ "grad_norm": 0.5461866855621338,
+ "learning_rate": 4.900076826288047e-07,
+ "loss": 0.0057,
+ "step": 33138
+ },
+ {
+ "epoch": 90.29700272479565,
+ "grad_norm": 1.258934736251831,
+ "learning_rate": 4.897348582829797e-07,
+ "loss": 0.01,
+ "step": 33139
+ },
+ {
+ "epoch": 90.29972752043597,
+ "grad_norm": 1.060585856437683,
+ "learning_rate": 4.894621080027906e-07,
+ "loss": 0.011,
+ "step": 33140
+ },
+ {
+ "epoch": 90.3024523160763,
+ "grad_norm": 1.7655590772628784,
+ "learning_rate": 4.891894317903678e-07,
+ "loss": 0.0733,
+ "step": 33141
+ },
+ {
+ "epoch": 90.30517711171662,
+ "grad_norm": 0.8850334286689758,
+ "learning_rate": 4.889168296478309e-07,
+ "loss": 0.0401,
+ "step": 33142
+ },
+ {
+ "epoch": 90.30790190735695,
+ "grad_norm": 1.4663782119750977,
+ "learning_rate": 4.886443015773023e-07,
+ "loss": 0.0105,
+ "step": 33143
+ },
+ {
+ "epoch": 90.31062670299727,
+ "grad_norm": 1.4113191366195679,
+ "learning_rate": 4.883718475809063e-07,
+ "loss": 0.0207,
+ "step": 33144
+ },
+ {
+ "epoch": 90.3133514986376,
+ "grad_norm": 1.272804856300354,
+ "learning_rate": 4.880994676607642e-07,
+ "loss": 0.0285,
+ "step": 33145
+ },
+ {
+ "epoch": 90.31607629427793,
+ "grad_norm": 0.5573623180389404,
+ "learning_rate": 4.878271618189978e-07,
+ "loss": 0.0063,
+ "step": 33146
+ },
+ {
+ "epoch": 90.31880108991825,
+ "grad_norm": 0.9040772318840027,
+ "learning_rate": 4.875549300577264e-07,
+ "loss": 0.0141,
+ "step": 33147
+ },
+ {
+ "epoch": 90.32152588555859,
+ "grad_norm": 1.0779584646224976,
+ "learning_rate": 4.872827723790707e-07,
+ "loss": 0.0341,
+ "step": 33148
+ },
+ {
+ "epoch": 90.3242506811989,
+ "grad_norm": 0.8653183579444885,
+ "learning_rate": 4.870106887851522e-07,
+ "loss": 0.0142,
+ "step": 33149
+ },
+ {
+ "epoch": 90.32697547683924,
+ "grad_norm": 1.0723000764846802,
+ "learning_rate": 4.867386792780882e-07,
+ "loss": 0.0163,
+ "step": 33150
+ },
+ {
+ "epoch": 90.32970027247957,
+ "grad_norm": 1.9742013216018677,
+ "learning_rate": 4.864667438599968e-07,
+ "loss": 0.0202,
+ "step": 33151
+ },
+ {
+ "epoch": 90.33242506811989,
+ "grad_norm": 1.1269820928573608,
+ "learning_rate": 4.861948825329954e-07,
+ "loss": 0.0471,
+ "step": 33152
+ },
+ {
+ "epoch": 90.33514986376022,
+ "grad_norm": 0.3897784650325775,
+ "learning_rate": 4.859230952992022e-07,
+ "loss": 0.0046,
+ "step": 33153
+ },
+ {
+ "epoch": 90.33787465940054,
+ "grad_norm": 0.9914458990097046,
+ "learning_rate": 4.856513821607356e-07,
+ "loss": 0.0748,
+ "step": 33154
+ },
+ {
+ "epoch": 90.34059945504087,
+ "grad_norm": 1.2801849842071533,
+ "learning_rate": 4.853797431197082e-07,
+ "loss": 0.0651,
+ "step": 33155
+ },
+ {
+ "epoch": 90.34332425068119,
+ "grad_norm": 0.9061378836631775,
+ "learning_rate": 4.851081781782363e-07,
+ "loss": 0.0065,
+ "step": 33156
+ },
+ {
+ "epoch": 90.34604904632153,
+ "grad_norm": 1.0149480104446411,
+ "learning_rate": 4.84836687338438e-07,
+ "loss": 0.0166,
+ "step": 33157
+ },
+ {
+ "epoch": 90.34877384196186,
+ "grad_norm": 1.5226136445999146,
+ "learning_rate": 4.84565270602424e-07,
+ "loss": 0.1374,
+ "step": 33158
+ },
+ {
+ "epoch": 90.35149863760218,
+ "grad_norm": 1.2342023849487305,
+ "learning_rate": 4.842939279723102e-07,
+ "loss": 0.0312,
+ "step": 33159
+ },
+ {
+ "epoch": 90.35422343324251,
+ "grad_norm": 0.949762761592865,
+ "learning_rate": 4.840226594502073e-07,
+ "loss": 0.0625,
+ "step": 33160
+ },
+ {
+ "epoch": 90.35694822888283,
+ "grad_norm": 1.735934853553772,
+ "learning_rate": 4.837514650382313e-07,
+ "loss": 0.0204,
+ "step": 33161
+ },
+ {
+ "epoch": 90.35967302452316,
+ "grad_norm": 0.6767263412475586,
+ "learning_rate": 4.834803447384917e-07,
+ "loss": 0.0064,
+ "step": 33162
+ },
+ {
+ "epoch": 90.3623978201635,
+ "grad_norm": 1.3365209102630615,
+ "learning_rate": 4.832092985531023e-07,
+ "loss": 0.0512,
+ "step": 33163
+ },
+ {
+ "epoch": 90.36512261580381,
+ "grad_norm": 0.7649092674255371,
+ "learning_rate": 4.829383264841703e-07,
+ "loss": 0.0075,
+ "step": 33164
+ },
+ {
+ "epoch": 90.36784741144415,
+ "grad_norm": 1.1028088331222534,
+ "learning_rate": 4.826674285338085e-07,
+ "loss": 0.0115,
+ "step": 33165
+ },
+ {
+ "epoch": 90.37057220708446,
+ "grad_norm": 0.8845653533935547,
+ "learning_rate": 4.823966047041273e-07,
+ "loss": 0.0083,
+ "step": 33166
+ },
+ {
+ "epoch": 90.3732970027248,
+ "grad_norm": 0.7646281123161316,
+ "learning_rate": 4.821258549972363e-07,
+ "loss": 0.007,
+ "step": 33167
+ },
+ {
+ "epoch": 90.37602179836512,
+ "grad_norm": 0.980556070804596,
+ "learning_rate": 4.818551794152404e-07,
+ "loss": 0.0063,
+ "step": 33168
+ },
+ {
+ "epoch": 90.37874659400545,
+ "grad_norm": 1.3218331336975098,
+ "learning_rate": 4.815845779602512e-07,
+ "loss": 0.0103,
+ "step": 33169
+ },
+ {
+ "epoch": 90.38147138964578,
+ "grad_norm": 0.5130805969238281,
+ "learning_rate": 4.813140506343761e-07,
+ "loss": 0.0034,
+ "step": 33170
+ },
+ {
+ "epoch": 90.3841961852861,
+ "grad_norm": 0.7781091928482056,
+ "learning_rate": 4.810435974397199e-07,
+ "loss": 0.0074,
+ "step": 33171
+ },
+ {
+ "epoch": 90.38692098092643,
+ "grad_norm": 0.7644485831260681,
+ "learning_rate": 4.807732183783886e-07,
+ "loss": 0.0066,
+ "step": 33172
+ },
+ {
+ "epoch": 90.38964577656675,
+ "grad_norm": 0.6645479798316956,
+ "learning_rate": 4.805029134524908e-07,
+ "loss": 0.0079,
+ "step": 33173
+ },
+ {
+ "epoch": 90.39237057220708,
+ "grad_norm": 0.5616403222084045,
+ "learning_rate": 4.80232682664129e-07,
+ "loss": 0.005,
+ "step": 33174
+ },
+ {
+ "epoch": 90.39509536784742,
+ "grad_norm": 1.429247260093689,
+ "learning_rate": 4.799625260154095e-07,
+ "loss": 0.0173,
+ "step": 33175
+ },
+ {
+ "epoch": 90.39782016348774,
+ "grad_norm": 1.5918858051300049,
+ "learning_rate": 4.796924435084338e-07,
+ "loss": 0.0872,
+ "step": 33176
+ },
+ {
+ "epoch": 90.40054495912807,
+ "grad_norm": 2.290346622467041,
+ "learning_rate": 4.794224351453092e-07,
+ "loss": 0.0096,
+ "step": 33177
+ },
+ {
+ "epoch": 90.40326975476839,
+ "grad_norm": 1.178679347038269,
+ "learning_rate": 4.79152500928135e-07,
+ "loss": 0.009,
+ "step": 33178
+ },
+ {
+ "epoch": 90.40599455040872,
+ "grad_norm": 1.062418818473816,
+ "learning_rate": 4.788826408590153e-07,
+ "loss": 0.0567,
+ "step": 33179
+ },
+ {
+ "epoch": 90.40871934604904,
+ "grad_norm": 1.3147791624069214,
+ "learning_rate": 4.786128549400492e-07,
+ "loss": 0.0551,
+ "step": 33180
+ },
+ {
+ "epoch": 90.41144414168937,
+ "grad_norm": 1.1485360860824585,
+ "learning_rate": 4.783431431733421e-07,
+ "loss": 0.0119,
+ "step": 33181
+ },
+ {
+ "epoch": 90.4141689373297,
+ "grad_norm": 0.5061028003692627,
+ "learning_rate": 4.780735055609908e-07,
+ "loss": 0.0057,
+ "step": 33182
+ },
+ {
+ "epoch": 90.41689373297002,
+ "grad_norm": 1.3124412298202515,
+ "learning_rate": 4.778039421050973e-07,
+ "loss": 0.0125,
+ "step": 33183
+ },
+ {
+ "epoch": 90.41961852861036,
+ "grad_norm": 1.631659984588623,
+ "learning_rate": 4.775344528077596e-07,
+ "loss": 0.05,
+ "step": 33184
+ },
+ {
+ "epoch": 90.42234332425068,
+ "grad_norm": 1.1348059177398682,
+ "learning_rate": 4.772650376710775e-07,
+ "loss": 0.0112,
+ "step": 33185
+ },
+ {
+ "epoch": 90.42506811989101,
+ "grad_norm": 0.6202444434165955,
+ "learning_rate": 4.76995696697149e-07,
+ "loss": 0.0108,
+ "step": 33186
+ },
+ {
+ "epoch": 90.42779291553134,
+ "grad_norm": 0.5259019732475281,
+ "learning_rate": 4.767264298880714e-07,
+ "loss": 0.0041,
+ "step": 33187
+ },
+ {
+ "epoch": 90.43051771117166,
+ "grad_norm": 1.4508682489395142,
+ "learning_rate": 4.7645723724594084e-07,
+ "loss": 0.0064,
+ "step": 33188
+ },
+ {
+ "epoch": 90.433242506812,
+ "grad_norm": 2.735836982727051,
+ "learning_rate": 4.7618811877285566e-07,
+ "loss": 0.0494,
+ "step": 33189
+ },
+ {
+ "epoch": 90.43596730245231,
+ "grad_norm": 0.9914137721061707,
+ "learning_rate": 4.759190744709119e-07,
+ "loss": 0.0092,
+ "step": 33190
+ },
+ {
+ "epoch": 90.43869209809264,
+ "grad_norm": 0.6747142672538757,
+ "learning_rate": 4.756501043422024e-07,
+ "loss": 0.0067,
+ "step": 33191
+ },
+ {
+ "epoch": 90.44141689373296,
+ "grad_norm": 1.2695550918579102,
+ "learning_rate": 4.7538120838882317e-07,
+ "loss": 0.019,
+ "step": 33192
+ },
+ {
+ "epoch": 90.4441416893733,
+ "grad_norm": 0.5513871312141418,
+ "learning_rate": 4.751123866128693e-07,
+ "loss": 0.006,
+ "step": 33193
+ },
+ {
+ "epoch": 90.44686648501363,
+ "grad_norm": 0.5350810289382935,
+ "learning_rate": 4.748436390164346e-07,
+ "loss": 0.0039,
+ "step": 33194
+ },
+ {
+ "epoch": 90.44959128065395,
+ "grad_norm": 0.9850096106529236,
+ "learning_rate": 4.7457496560160966e-07,
+ "loss": 0.0061,
+ "step": 33195
+ },
+ {
+ "epoch": 90.45231607629428,
+ "grad_norm": 1.5938515663146973,
+ "learning_rate": 4.743063663704883e-07,
+ "loss": 0.0095,
+ "step": 33196
+ },
+ {
+ "epoch": 90.4550408719346,
+ "grad_norm": 0.7272943258285522,
+ "learning_rate": 4.740378413251634e-07,
+ "loss": 0.0064,
+ "step": 33197
+ },
+ {
+ "epoch": 90.45776566757493,
+ "grad_norm": 0.960972249507904,
+ "learning_rate": 4.737693904677243e-07,
+ "loss": 0.0111,
+ "step": 33198
+ },
+ {
+ "epoch": 90.46049046321527,
+ "grad_norm": 0.47570717334747314,
+ "learning_rate": 4.735010138002616e-07,
+ "loss": 0.0058,
+ "step": 33199
+ },
+ {
+ "epoch": 90.46321525885558,
+ "grad_norm": 0.9442270398139954,
+ "learning_rate": 4.7323271132486805e-07,
+ "loss": 0.0169,
+ "step": 33200
+ },
+ {
+ "epoch": 90.46594005449592,
+ "grad_norm": 1.1216280460357666,
+ "learning_rate": 4.7296448304363086e-07,
+ "loss": 0.0991,
+ "step": 33201
+ },
+ {
+ "epoch": 90.46866485013624,
+ "grad_norm": 1.3952564001083374,
+ "learning_rate": 4.726963289586395e-07,
+ "loss": 0.0067,
+ "step": 33202
+ },
+ {
+ "epoch": 90.47138964577657,
+ "grad_norm": 0.7878663539886475,
+ "learning_rate": 4.7242824907198335e-07,
+ "loss": 0.0106,
+ "step": 33203
+ },
+ {
+ "epoch": 90.47411444141689,
+ "grad_norm": 0.7300451993942261,
+ "learning_rate": 4.721602433857497e-07,
+ "loss": 0.0084,
+ "step": 33204
+ },
+ {
+ "epoch": 90.47683923705722,
+ "grad_norm": 0.7912105917930603,
+ "learning_rate": 4.7189231190202353e-07,
+ "loss": 0.0107,
+ "step": 33205
+ },
+ {
+ "epoch": 90.47956403269755,
+ "grad_norm": 1.379963994026184,
+ "learning_rate": 4.716244546228943e-07,
+ "loss": 0.0415,
+ "step": 33206
+ },
+ {
+ "epoch": 90.48228882833787,
+ "grad_norm": 1.5496551990509033,
+ "learning_rate": 4.7135667155044807e-07,
+ "loss": 0.0143,
+ "step": 33207
+ },
+ {
+ "epoch": 90.4850136239782,
+ "grad_norm": 0.5928764939308167,
+ "learning_rate": 4.710889626867687e-07,
+ "loss": 0.0073,
+ "step": 33208
+ },
+ {
+ "epoch": 90.48773841961852,
+ "grad_norm": 2.403338670730591,
+ "learning_rate": 4.7082132803394133e-07,
+ "loss": 0.1208,
+ "step": 33209
+ },
+ {
+ "epoch": 90.49046321525886,
+ "grad_norm": 1.133947730064392,
+ "learning_rate": 4.7055376759405083e-07,
+ "loss": 0.0099,
+ "step": 33210
+ },
+ {
+ "epoch": 90.49318801089919,
+ "grad_norm": 0.7943406105041504,
+ "learning_rate": 4.702862813691822e-07,
+ "loss": 0.0098,
+ "step": 33211
+ },
+ {
+ "epoch": 90.49591280653951,
+ "grad_norm": 0.780269205570221,
+ "learning_rate": 4.700188693614149e-07,
+ "loss": 0.0082,
+ "step": 33212
+ },
+ {
+ "epoch": 90.49863760217984,
+ "grad_norm": 0.3194793462753296,
+ "learning_rate": 4.6975153157283624e-07,
+ "loss": 0.0035,
+ "step": 33213
+ },
+ {
+ "epoch": 90.50136239782016,
+ "grad_norm": 1.5277498960494995,
+ "learning_rate": 4.6948426800552447e-07,
+ "loss": 0.0077,
+ "step": 33214
+ },
+ {
+ "epoch": 90.50408719346049,
+ "grad_norm": 0.6761731505393982,
+ "learning_rate": 4.692170786615635e-07,
+ "loss": 0.0059,
+ "step": 33215
+ },
+ {
+ "epoch": 90.50681198910081,
+ "grad_norm": 0.36271947622299194,
+ "learning_rate": 4.689499635430317e-07,
+ "loss": 0.0042,
+ "step": 33216
+ },
+ {
+ "epoch": 90.50953678474114,
+ "grad_norm": 1.4222813844680786,
+ "learning_rate": 4.6868292265201176e-07,
+ "loss": 0.0627,
+ "step": 33217
+ },
+ {
+ "epoch": 90.51226158038148,
+ "grad_norm": 1.4326664209365845,
+ "learning_rate": 4.684159559905821e-07,
+ "loss": 0.0422,
+ "step": 33218
+ },
+ {
+ "epoch": 90.5149863760218,
+ "grad_norm": 0.7001932859420776,
+ "learning_rate": 4.681490635608232e-07,
+ "loss": 0.0107,
+ "step": 33219
+ },
+ {
+ "epoch": 90.51771117166213,
+ "grad_norm": 1.7279783487319946,
+ "learning_rate": 4.678822453648124e-07,
+ "loss": 0.0408,
+ "step": 33220
+ },
+ {
+ "epoch": 90.52043596730245,
+ "grad_norm": 1.766376256942749,
+ "learning_rate": 4.676155014046269e-07,
+ "loss": 0.0386,
+ "step": 33221
+ },
+ {
+ "epoch": 90.52316076294278,
+ "grad_norm": 0.7891702651977539,
+ "learning_rate": 4.6734883168234603e-07,
+ "loss": 0.0081,
+ "step": 33222
+ },
+ {
+ "epoch": 90.52588555858311,
+ "grad_norm": 1.8012150526046753,
+ "learning_rate": 4.67082236200046e-07,
+ "loss": 0.1412,
+ "step": 33223
+ },
+ {
+ "epoch": 90.52861035422343,
+ "grad_norm": 1.1256985664367676,
+ "learning_rate": 4.6681571495980184e-07,
+ "loss": 0.0112,
+ "step": 33224
+ },
+ {
+ "epoch": 90.53133514986376,
+ "grad_norm": 2.671830654144287,
+ "learning_rate": 4.665492679636896e-07,
+ "loss": 0.0749,
+ "step": 33225
+ },
+ {
+ "epoch": 90.53405994550408,
+ "grad_norm": 0.4976314902305603,
+ "learning_rate": 4.6628289521378543e-07,
+ "loss": 0.0051,
+ "step": 33226
+ },
+ {
+ "epoch": 90.53678474114442,
+ "grad_norm": 0.49103236198425293,
+ "learning_rate": 4.6601659671216325e-07,
+ "loss": 0.0038,
+ "step": 33227
+ },
+ {
+ "epoch": 90.53950953678473,
+ "grad_norm": 0.6824777722358704,
+ "learning_rate": 4.6575037246089807e-07,
+ "loss": 0.0073,
+ "step": 33228
+ },
+ {
+ "epoch": 90.54223433242507,
+ "grad_norm": 0.9129568934440613,
+ "learning_rate": 4.6548422246206036e-07,
+ "loss": 0.0109,
+ "step": 33229
+ },
+ {
+ "epoch": 90.5449591280654,
+ "grad_norm": 1.0983541011810303,
+ "learning_rate": 4.6521814671772525e-07,
+ "loss": 0.0152,
+ "step": 33230
+ },
+ {
+ "epoch": 90.54768392370572,
+ "grad_norm": 0.77965247631073,
+ "learning_rate": 4.6495214522996435e-07,
+ "loss": 0.01,
+ "step": 33231
+ },
+ {
+ "epoch": 90.55040871934605,
+ "grad_norm": 1.1012898683547974,
+ "learning_rate": 4.6468621800085043e-07,
+ "loss": 0.0156,
+ "step": 33232
+ },
+ {
+ "epoch": 90.55313351498637,
+ "grad_norm": 1.0979722738265991,
+ "learning_rate": 4.644203650324508e-07,
+ "loss": 0.0105,
+ "step": 33233
+ },
+ {
+ "epoch": 90.5558583106267,
+ "grad_norm": 0.6855060458183289,
+ "learning_rate": 4.641545863268393e-07,
+ "loss": 0.011,
+ "step": 33234
+ },
+ {
+ "epoch": 90.55858310626704,
+ "grad_norm": 1.361122965812683,
+ "learning_rate": 4.6388888188608537e-07,
+ "loss": 0.1361,
+ "step": 33235
+ },
+ {
+ "epoch": 90.56130790190736,
+ "grad_norm": 1.2910090684890747,
+ "learning_rate": 4.636232517122585e-07,
+ "loss": 0.0361,
+ "step": 33236
+ },
+ {
+ "epoch": 90.56403269754769,
+ "grad_norm": 1.1310073137283325,
+ "learning_rate": 4.6335769580742594e-07,
+ "loss": 0.0449,
+ "step": 33237
+ },
+ {
+ "epoch": 90.566757493188,
+ "grad_norm": 0.8746498823165894,
+ "learning_rate": 4.6309221417365826e-07,
+ "loss": 0.0133,
+ "step": 33238
+ },
+ {
+ "epoch": 90.56948228882834,
+ "grad_norm": 1.1589735746383667,
+ "learning_rate": 4.6282680681302037e-07,
+ "loss": 0.0188,
+ "step": 33239
+ },
+ {
+ "epoch": 90.57220708446866,
+ "grad_norm": 0.7377759218215942,
+ "learning_rate": 4.6256147372758074e-07,
+ "loss": 0.0068,
+ "step": 33240
+ },
+ {
+ "epoch": 90.57493188010899,
+ "grad_norm": 0.5758037567138672,
+ "learning_rate": 4.622962149194044e-07,
+ "loss": 0.0061,
+ "step": 33241
+ },
+ {
+ "epoch": 90.57765667574932,
+ "grad_norm": 0.8269096612930298,
+ "learning_rate": 4.620310303905584e-07,
+ "loss": 0.0079,
+ "step": 33242
+ },
+ {
+ "epoch": 90.58038147138964,
+ "grad_norm": 0.9122932553291321,
+ "learning_rate": 4.617659201431091e-07,
+ "loss": 0.0393,
+ "step": 33243
+ },
+ {
+ "epoch": 90.58310626702998,
+ "grad_norm": 0.886144757270813,
+ "learning_rate": 4.6150088417911907e-07,
+ "loss": 0.0135,
+ "step": 33244
+ },
+ {
+ "epoch": 90.5858310626703,
+ "grad_norm": 1.279776930809021,
+ "learning_rate": 4.612359225006524e-07,
+ "loss": 0.0135,
+ "step": 33245
+ },
+ {
+ "epoch": 90.58855585831063,
+ "grad_norm": 0.9966426491737366,
+ "learning_rate": 4.60971035109774e-07,
+ "loss": 0.0063,
+ "step": 33246
+ },
+ {
+ "epoch": 90.59128065395096,
+ "grad_norm": 0.5143934488296509,
+ "learning_rate": 4.6070622200854673e-07,
+ "loss": 0.0076,
+ "step": 33247
+ },
+ {
+ "epoch": 90.59400544959128,
+ "grad_norm": 0.4997609257698059,
+ "learning_rate": 4.6044148319903225e-07,
+ "loss": 0.0044,
+ "step": 33248
+ },
+ {
+ "epoch": 90.59673024523161,
+ "grad_norm": 0.6383733153343201,
+ "learning_rate": 4.6017681868329114e-07,
+ "loss": 0.075,
+ "step": 33249
+ },
+ {
+ "epoch": 90.59945504087193,
+ "grad_norm": 2.2054145336151123,
+ "learning_rate": 4.5991222846338724e-07,
+ "loss": 0.0784,
+ "step": 33250
+ },
+ {
+ "epoch": 90.60217983651226,
+ "grad_norm": 2.2438161373138428,
+ "learning_rate": 4.5964771254137785e-07,
+ "loss": 0.012,
+ "step": 33251
+ },
+ {
+ "epoch": 90.60490463215258,
+ "grad_norm": 1.731163740158081,
+ "learning_rate": 4.5938327091932687e-07,
+ "loss": 0.0975,
+ "step": 33252
+ },
+ {
+ "epoch": 90.60762942779292,
+ "grad_norm": 1.0728963613510132,
+ "learning_rate": 4.591189035992916e-07,
+ "loss": 0.0467,
+ "step": 33253
+ },
+ {
+ "epoch": 90.61035422343325,
+ "grad_norm": 1.2540615797042847,
+ "learning_rate": 4.5885461058333137e-07,
+ "loss": 0.0256,
+ "step": 33254
+ },
+ {
+ "epoch": 90.61307901907357,
+ "grad_norm": 1.2978813648223877,
+ "learning_rate": 4.585903918735046e-07,
+ "loss": 0.0133,
+ "step": 33255
+ },
+ {
+ "epoch": 90.6158038147139,
+ "grad_norm": 1.0762839317321777,
+ "learning_rate": 4.583262474718686e-07,
+ "loss": 0.0119,
+ "step": 33256
+ },
+ {
+ "epoch": 90.61852861035422,
+ "grad_norm": 0.4427762031555176,
+ "learning_rate": 4.5806217738047944e-07,
+ "loss": 0.0038,
+ "step": 33257
+ },
+ {
+ "epoch": 90.62125340599455,
+ "grad_norm": 3.041553497314453,
+ "learning_rate": 4.5779818160139657e-07,
+ "loss": 0.0147,
+ "step": 33258
+ },
+ {
+ "epoch": 90.62397820163488,
+ "grad_norm": 1.6969051361083984,
+ "learning_rate": 4.575342601366739e-07,
+ "loss": 0.0486,
+ "step": 33259
+ },
+ {
+ "epoch": 90.6267029972752,
+ "grad_norm": 0.9959744811058044,
+ "learning_rate": 4.572704129883676e-07,
+ "loss": 0.0687,
+ "step": 33260
+ },
+ {
+ "epoch": 90.62942779291554,
+ "grad_norm": 0.6243171095848083,
+ "learning_rate": 4.5700664015853045e-07,
+ "loss": 0.0104,
+ "step": 33261
+ },
+ {
+ "epoch": 90.63215258855585,
+ "grad_norm": 1.0806987285614014,
+ "learning_rate": 4.567429416492208e-07,
+ "loss": 0.009,
+ "step": 33262
+ },
+ {
+ "epoch": 90.63487738419619,
+ "grad_norm": 1.2914866209030151,
+ "learning_rate": 4.564793174624893e-07,
+ "loss": 0.0124,
+ "step": 33263
+ },
+ {
+ "epoch": 90.6376021798365,
+ "grad_norm": 0.3743385672569275,
+ "learning_rate": 4.5621576760038975e-07,
+ "loss": 0.004,
+ "step": 33264
+ },
+ {
+ "epoch": 90.64032697547684,
+ "grad_norm": 2.138580560684204,
+ "learning_rate": 4.5595229206497283e-07,
+ "loss": 0.0128,
+ "step": 33265
+ },
+ {
+ "epoch": 90.64305177111717,
+ "grad_norm": 1.0664558410644531,
+ "learning_rate": 4.556888908582946e-07,
+ "loss": 0.0242,
+ "step": 33266
+ },
+ {
+ "epoch": 90.64577656675749,
+ "grad_norm": 1.0539733171463013,
+ "learning_rate": 4.5542556398240347e-07,
+ "loss": 0.017,
+ "step": 33267
+ },
+ {
+ "epoch": 90.64850136239782,
+ "grad_norm": 1.545487642288208,
+ "learning_rate": 4.5516231143935e-07,
+ "loss": 0.0118,
+ "step": 33268
+ },
+ {
+ "epoch": 90.65122615803814,
+ "grad_norm": 0.671704113483429,
+ "learning_rate": 4.5489913323118585e-07,
+ "loss": 0.0078,
+ "step": 33269
+ },
+ {
+ "epoch": 90.65395095367847,
+ "grad_norm": 0.550736665725708,
+ "learning_rate": 4.546360293599594e-07,
+ "loss": 0.0047,
+ "step": 33270
+ },
+ {
+ "epoch": 90.65667574931881,
+ "grad_norm": 0.6587178111076355,
+ "learning_rate": 4.543729998277213e-07,
+ "loss": 0.0076,
+ "step": 33271
+ },
+ {
+ "epoch": 90.65940054495913,
+ "grad_norm": 1.2468892335891724,
+ "learning_rate": 4.5411004463651877e-07,
+ "loss": 0.0711,
+ "step": 33272
+ },
+ {
+ "epoch": 90.66212534059946,
+ "grad_norm": 1.497235655784607,
+ "learning_rate": 4.538471637884001e-07,
+ "loss": 0.0123,
+ "step": 33273
+ },
+ {
+ "epoch": 90.66485013623978,
+ "grad_norm": 1.722974419593811,
+ "learning_rate": 4.535843572854115e-07,
+ "loss": 0.0169,
+ "step": 33274
+ },
+ {
+ "epoch": 90.66757493188011,
+ "grad_norm": 0.785627543926239,
+ "learning_rate": 4.5332162512960245e-07,
+ "loss": 0.0131,
+ "step": 33275
+ },
+ {
+ "epoch": 90.67029972752043,
+ "grad_norm": 1.5262354612350464,
+ "learning_rate": 4.5305896732301566e-07,
+ "loss": 0.0352,
+ "step": 33276
+ },
+ {
+ "epoch": 90.67302452316076,
+ "grad_norm": 1.3490524291992188,
+ "learning_rate": 4.527963838676985e-07,
+ "loss": 0.0084,
+ "step": 33277
+ },
+ {
+ "epoch": 90.6757493188011,
+ "grad_norm": 0.4542495608329773,
+ "learning_rate": 4.525338747656971e-07,
+ "loss": 0.0037,
+ "step": 33278
+ },
+ {
+ "epoch": 90.67847411444141,
+ "grad_norm": 1.1983743906021118,
+ "learning_rate": 4.522714400190542e-07,
+ "loss": 0.1179,
+ "step": 33279
+ },
+ {
+ "epoch": 90.68119891008175,
+ "grad_norm": 0.6972528696060181,
+ "learning_rate": 4.5200907962981376e-07,
+ "loss": 0.006,
+ "step": 33280
+ },
+ {
+ "epoch": 90.68392370572207,
+ "grad_norm": 1.151095986366272,
+ "learning_rate": 4.517467936000186e-07,
+ "loss": 0.0155,
+ "step": 33281
+ },
+ {
+ "epoch": 90.6866485013624,
+ "grad_norm": 1.2587600946426392,
+ "learning_rate": 4.5148458193171264e-07,
+ "loss": 0.0111,
+ "step": 33282
+ },
+ {
+ "epoch": 90.68937329700273,
+ "grad_norm": 1.2754952907562256,
+ "learning_rate": 4.5122244462693865e-07,
+ "loss": 0.022,
+ "step": 33283
+ },
+ {
+ "epoch": 90.69209809264305,
+ "grad_norm": 2.021860122680664,
+ "learning_rate": 4.5096038168773613e-07,
+ "loss": 0.0619,
+ "step": 33284
+ },
+ {
+ "epoch": 90.69482288828338,
+ "grad_norm": 1.285725712776184,
+ "learning_rate": 4.506983931161457e-07,
+ "loss": 0.0578,
+ "step": 33285
+ },
+ {
+ "epoch": 90.6975476839237,
+ "grad_norm": 1.4170883893966675,
+ "learning_rate": 4.5043647891420794e-07,
+ "loss": 0.0072,
+ "step": 33286
+ },
+ {
+ "epoch": 90.70027247956403,
+ "grad_norm": 0.4114612936973572,
+ "learning_rate": 4.5017463908396564e-07,
+ "loss": 0.003,
+ "step": 33287
+ },
+ {
+ "epoch": 90.70299727520435,
+ "grad_norm": 0.7143770456314087,
+ "learning_rate": 4.4991287362745604e-07,
+ "loss": 0.0127,
+ "step": 33288
+ },
+ {
+ "epoch": 90.70572207084469,
+ "grad_norm": 2.485661029815674,
+ "learning_rate": 4.496511825467176e-07,
+ "loss": 0.0154,
+ "step": 33289
+ },
+ {
+ "epoch": 90.70844686648502,
+ "grad_norm": 0.7212743163108826,
+ "learning_rate": 4.493895658437875e-07,
+ "loss": 0.0092,
+ "step": 33290
+ },
+ {
+ "epoch": 90.71117166212534,
+ "grad_norm": 1.2386162281036377,
+ "learning_rate": 4.4912802352070406e-07,
+ "loss": 0.0113,
+ "step": 33291
+ },
+ {
+ "epoch": 90.71389645776567,
+ "grad_norm": 0.9937238693237305,
+ "learning_rate": 4.4886655557950576e-07,
+ "loss": 0.0062,
+ "step": 33292
+ },
+ {
+ "epoch": 90.71662125340599,
+ "grad_norm": 1.2071659564971924,
+ "learning_rate": 4.486051620222265e-07,
+ "loss": 0.0127,
+ "step": 33293
+ },
+ {
+ "epoch": 90.71934604904632,
+ "grad_norm": 0.769974410533905,
+ "learning_rate": 4.4834384285090126e-07,
+ "loss": 0.0096,
+ "step": 33294
+ },
+ {
+ "epoch": 90.72207084468666,
+ "grad_norm": 0.9442760348320007,
+ "learning_rate": 4.480825980675685e-07,
+ "loss": 0.0103,
+ "step": 33295
+ },
+ {
+ "epoch": 90.72479564032697,
+ "grad_norm": 1.5918372869491577,
+ "learning_rate": 4.47821427674261e-07,
+ "loss": 0.1714,
+ "step": 33296
+ },
+ {
+ "epoch": 90.7275204359673,
+ "grad_norm": 0.6704189777374268,
+ "learning_rate": 4.475603316730115e-07,
+ "loss": 0.0093,
+ "step": 33297
+ },
+ {
+ "epoch": 90.73024523160763,
+ "grad_norm": 0.5558213591575623,
+ "learning_rate": 4.4729931006585514e-07,
+ "loss": 0.0051,
+ "step": 33298
+ },
+ {
+ "epoch": 90.73297002724796,
+ "grad_norm": 1.5154848098754883,
+ "learning_rate": 4.470383628548247e-07,
+ "loss": 0.0404,
+ "step": 33299
+ },
+ {
+ "epoch": 90.73569482288828,
+ "grad_norm": 1.008408784866333,
+ "learning_rate": 4.4677749004195305e-07,
+ "loss": 0.0054,
+ "step": 33300
+ },
+ {
+ "epoch": 90.73841961852861,
+ "grad_norm": 1.203877568244934,
+ "learning_rate": 4.4651669162926847e-07,
+ "loss": 0.0081,
+ "step": 33301
+ },
+ {
+ "epoch": 90.74114441416894,
+ "grad_norm": 1.206678867340088,
+ "learning_rate": 4.4625596761880496e-07,
+ "loss": 0.009,
+ "step": 33302
+ },
+ {
+ "epoch": 90.74386920980926,
+ "grad_norm": 1.1642204523086548,
+ "learning_rate": 4.459953180125931e-07,
+ "loss": 0.0168,
+ "step": 33303
+ },
+ {
+ "epoch": 90.7465940054496,
+ "grad_norm": 1.2771822214126587,
+ "learning_rate": 4.457347428126635e-07,
+ "loss": 0.0099,
+ "step": 33304
+ },
+ {
+ "epoch": 90.74931880108991,
+ "grad_norm": 1.0146735906600952,
+ "learning_rate": 4.454742420210434e-07,
+ "loss": 0.0086,
+ "step": 33305
+ },
+ {
+ "epoch": 90.75204359673025,
+ "grad_norm": 0.8619017601013184,
+ "learning_rate": 4.452138156397612e-07,
+ "loss": 0.0073,
+ "step": 33306
+ },
+ {
+ "epoch": 90.75476839237058,
+ "grad_norm": 1.16953444480896,
+ "learning_rate": 4.449534636708475e-07,
+ "loss": 0.0659,
+ "step": 33307
+ },
+ {
+ "epoch": 90.7574931880109,
+ "grad_norm": 0.9940509796142578,
+ "learning_rate": 4.446931861163284e-07,
+ "loss": 0.0075,
+ "step": 33308
+ },
+ {
+ "epoch": 90.76021798365123,
+ "grad_norm": 1.1055049896240234,
+ "learning_rate": 4.4443298297823124e-07,
+ "loss": 0.0247,
+ "step": 33309
+ },
+ {
+ "epoch": 90.76294277929155,
+ "grad_norm": 1.109468936920166,
+ "learning_rate": 4.4417285425858213e-07,
+ "loss": 0.0223,
+ "step": 33310
+ },
+ {
+ "epoch": 90.76566757493188,
+ "grad_norm": 0.8615556955337524,
+ "learning_rate": 4.4391279995940727e-07,
+ "loss": 0.1038,
+ "step": 33311
+ },
+ {
+ "epoch": 90.7683923705722,
+ "grad_norm": 1.5570197105407715,
+ "learning_rate": 4.436528200827328e-07,
+ "loss": 0.0224,
+ "step": 33312
+ },
+ {
+ "epoch": 90.77111716621253,
+ "grad_norm": 1.4017330408096313,
+ "learning_rate": 4.4339291463058266e-07,
+ "loss": 0.0235,
+ "step": 33313
+ },
+ {
+ "epoch": 90.77384196185287,
+ "grad_norm": 0.9108633995056152,
+ "learning_rate": 4.4313308360497855e-07,
+ "loss": 0.0096,
+ "step": 33314
+ },
+ {
+ "epoch": 90.77656675749319,
+ "grad_norm": 1.3933829069137573,
+ "learning_rate": 4.428733270079488e-07,
+ "loss": 0.0176,
+ "step": 33315
+ },
+ {
+ "epoch": 90.77929155313352,
+ "grad_norm": 1.10630202293396,
+ "learning_rate": 4.4261364484151416e-07,
+ "loss": 0.0135,
+ "step": 33316
+ },
+ {
+ "epoch": 90.78201634877384,
+ "grad_norm": 0.9096522927284241,
+ "learning_rate": 4.423540371076962e-07,
+ "loss": 0.0096,
+ "step": 33317
+ },
+ {
+ "epoch": 90.78474114441417,
+ "grad_norm": 0.8447855114936829,
+ "learning_rate": 4.420945038085156e-07,
+ "loss": 0.1179,
+ "step": 33318
+ },
+ {
+ "epoch": 90.7874659400545,
+ "grad_norm": 0.6463069915771484,
+ "learning_rate": 4.418350449459974e-07,
+ "loss": 0.0087,
+ "step": 33319
+ },
+ {
+ "epoch": 90.79019073569482,
+ "grad_norm": 0.9261990785598755,
+ "learning_rate": 4.4157566052215885e-07,
+ "loss": 0.0099,
+ "step": 33320
+ },
+ {
+ "epoch": 90.79291553133515,
+ "grad_norm": 0.6789969801902771,
+ "learning_rate": 4.413163505390228e-07,
+ "loss": 0.0104,
+ "step": 33321
+ },
+ {
+ "epoch": 90.79564032697547,
+ "grad_norm": 0.8385263681411743,
+ "learning_rate": 4.410571149986065e-07,
+ "loss": 0.0112,
+ "step": 33322
+ },
+ {
+ "epoch": 90.7983651226158,
+ "grad_norm": 1.1740902662277222,
+ "learning_rate": 4.4079795390293057e-07,
+ "loss": 0.0171,
+ "step": 33323
+ },
+ {
+ "epoch": 90.80108991825612,
+ "grad_norm": 1.6696385145187378,
+ "learning_rate": 4.405388672540123e-07,
+ "loss": 0.1005,
+ "step": 33324
+ },
+ {
+ "epoch": 90.80381471389646,
+ "grad_norm": 1.7983770370483398,
+ "learning_rate": 4.4027985505387005e-07,
+ "loss": 0.0224,
+ "step": 33325
+ },
+ {
+ "epoch": 90.80653950953679,
+ "grad_norm": 1.1733636856079102,
+ "learning_rate": 4.400209173045189e-07,
+ "loss": 0.0162,
+ "step": 33326
+ },
+ {
+ "epoch": 90.80926430517711,
+ "grad_norm": 1.0538158416748047,
+ "learning_rate": 4.397620540079794e-07,
+ "loss": 0.0812,
+ "step": 33327
+ },
+ {
+ "epoch": 90.81198910081744,
+ "grad_norm": 1.2226991653442383,
+ "learning_rate": 4.395032651662645e-07,
+ "loss": 0.0465,
+ "step": 33328
+ },
+ {
+ "epoch": 90.81471389645776,
+ "grad_norm": 1.268640160560608,
+ "learning_rate": 4.3924455078139027e-07,
+ "loss": 0.0141,
+ "step": 33329
+ },
+ {
+ "epoch": 90.8174386920981,
+ "grad_norm": 1.115167260169983,
+ "learning_rate": 4.389859108553707e-07,
+ "loss": 0.013,
+ "step": 33330
+ },
+ {
+ "epoch": 90.82016348773843,
+ "grad_norm": 1.192569375038147,
+ "learning_rate": 4.38727345390223e-07,
+ "loss": 0.0572,
+ "step": 33331
+ },
+ {
+ "epoch": 90.82288828337875,
+ "grad_norm": 1.354540467262268,
+ "learning_rate": 4.384688543879578e-07,
+ "loss": 0.0317,
+ "step": 33332
+ },
+ {
+ "epoch": 90.82561307901908,
+ "grad_norm": 0.8279160857200623,
+ "learning_rate": 4.382104378505903e-07,
+ "loss": 0.0095,
+ "step": 33333
+ },
+ {
+ "epoch": 90.8283378746594,
+ "grad_norm": 1.1926524639129639,
+ "learning_rate": 4.3795209578013085e-07,
+ "loss": 0.0267,
+ "step": 33334
+ },
+ {
+ "epoch": 90.83106267029973,
+ "grad_norm": 1.080073595046997,
+ "learning_rate": 4.3769382817859363e-07,
+ "loss": 0.1074,
+ "step": 33335
+ },
+ {
+ "epoch": 90.83378746594005,
+ "grad_norm": 1.2128134965896606,
+ "learning_rate": 4.3743563504799027e-07,
+ "loss": 0.0142,
+ "step": 33336
+ },
+ {
+ "epoch": 90.83651226158038,
+ "grad_norm": 1.1451507806777954,
+ "learning_rate": 4.37177516390328e-07,
+ "loss": 0.0166,
+ "step": 33337
+ },
+ {
+ "epoch": 90.83923705722071,
+ "grad_norm": 1.1326994895935059,
+ "learning_rate": 4.3691947220762087e-07,
+ "loss": 0.0147,
+ "step": 33338
+ },
+ {
+ "epoch": 90.84196185286103,
+ "grad_norm": 1.0493181943893433,
+ "learning_rate": 4.366615025018761e-07,
+ "loss": 0.0094,
+ "step": 33339
+ },
+ {
+ "epoch": 90.84468664850137,
+ "grad_norm": 0.6914105415344238,
+ "learning_rate": 4.364036072751043e-07,
+ "loss": 0.0079,
+ "step": 33340
+ },
+ {
+ "epoch": 90.84741144414168,
+ "grad_norm": 1.3739972114562988,
+ "learning_rate": 4.361457865293139e-07,
+ "loss": 0.0319,
+ "step": 33341
+ },
+ {
+ "epoch": 90.85013623978202,
+ "grad_norm": 2.375411033630371,
+ "learning_rate": 4.358880402665111e-07,
+ "loss": 0.0401,
+ "step": 33342
+ },
+ {
+ "epoch": 90.85286103542235,
+ "grad_norm": 1.0268595218658447,
+ "learning_rate": 4.3563036848870535e-07,
+ "loss": 0.0248,
+ "step": 33343
+ },
+ {
+ "epoch": 90.85558583106267,
+ "grad_norm": 1.1378893852233887,
+ "learning_rate": 4.353727711979028e-07,
+ "loss": 0.0119,
+ "step": 33344
+ },
+ {
+ "epoch": 90.858310626703,
+ "grad_norm": 0.9894020557403564,
+ "learning_rate": 4.351152483961096e-07,
+ "loss": 0.0109,
+ "step": 33345
+ },
+ {
+ "epoch": 90.86103542234332,
+ "grad_norm": 1.7857165336608887,
+ "learning_rate": 4.3485780008532987e-07,
+ "loss": 0.1874,
+ "step": 33346
+ },
+ {
+ "epoch": 90.86376021798365,
+ "grad_norm": 0.40092331171035767,
+ "learning_rate": 4.346004262675707e-07,
+ "loss": 0.0039,
+ "step": 33347
+ },
+ {
+ "epoch": 90.86648501362397,
+ "grad_norm": 1.5286319255828857,
+ "learning_rate": 4.3434312694483504e-07,
+ "loss": 0.0971,
+ "step": 33348
+ },
+ {
+ "epoch": 90.8692098092643,
+ "grad_norm": 0.6185675859451294,
+ "learning_rate": 4.3408590211912793e-07,
+ "loss": 0.007,
+ "step": 33349
+ },
+ {
+ "epoch": 90.87193460490464,
+ "grad_norm": 1.3057000637054443,
+ "learning_rate": 4.3382875179244997e-07,
+ "loss": 0.0227,
+ "step": 33350
+ },
+ {
+ "epoch": 90.87465940054496,
+ "grad_norm": 1.3880324363708496,
+ "learning_rate": 4.335716759668074e-07,
+ "loss": 0.0159,
+ "step": 33351
+ },
+ {
+ "epoch": 90.87738419618529,
+ "grad_norm": 0.7951704859733582,
+ "learning_rate": 4.3331467464420076e-07,
+ "loss": 0.0084,
+ "step": 33352
+ },
+ {
+ "epoch": 90.88010899182561,
+ "grad_norm": 0.5318107008934021,
+ "learning_rate": 4.330577478266318e-07,
+ "loss": 0.007,
+ "step": 33353
+ },
+ {
+ "epoch": 90.88283378746594,
+ "grad_norm": 0.9598751068115234,
+ "learning_rate": 4.3280089551609895e-07,
+ "loss": 0.0082,
+ "step": 33354
+ },
+ {
+ "epoch": 90.88555858310627,
+ "grad_norm": 0.7412259578704834,
+ "learning_rate": 4.3254411771460616e-07,
+ "loss": 0.0062,
+ "step": 33355
+ },
+ {
+ "epoch": 90.88828337874659,
+ "grad_norm": 0.9785472750663757,
+ "learning_rate": 4.322874144241518e-07,
+ "loss": 0.0235,
+ "step": 33356
+ },
+ {
+ "epoch": 90.89100817438693,
+ "grad_norm": 0.7086524963378906,
+ "learning_rate": 4.320307856467365e-07,
+ "loss": 0.0098,
+ "step": 33357
+ },
+ {
+ "epoch": 90.89373297002724,
+ "grad_norm": 0.5601330995559692,
+ "learning_rate": 4.317742313843565e-07,
+ "loss": 0.0061,
+ "step": 33358
+ },
+ {
+ "epoch": 90.89645776566758,
+ "grad_norm": 0.7577287554740906,
+ "learning_rate": 4.315177516390101e-07,
+ "loss": 0.0149,
+ "step": 33359
+ },
+ {
+ "epoch": 90.8991825613079,
+ "grad_norm": 0.4376874566078186,
+ "learning_rate": 4.312613464126958e-07,
+ "loss": 0.0054,
+ "step": 33360
+ },
+ {
+ "epoch": 90.90190735694823,
+ "grad_norm": 1.8768008947372437,
+ "learning_rate": 4.310050157074108e-07,
+ "loss": 0.0364,
+ "step": 33361
+ },
+ {
+ "epoch": 90.90463215258856,
+ "grad_norm": 0.7834678292274475,
+ "learning_rate": 4.307487595251503e-07,
+ "loss": 0.007,
+ "step": 33362
+ },
+ {
+ "epoch": 90.90735694822888,
+ "grad_norm": 1.496172547340393,
+ "learning_rate": 4.304925778679103e-07,
+ "loss": 0.0136,
+ "step": 33363
+ },
+ {
+ "epoch": 90.91008174386921,
+ "grad_norm": 1.0922050476074219,
+ "learning_rate": 4.3023647073768716e-07,
+ "loss": 0.0401,
+ "step": 33364
+ },
+ {
+ "epoch": 90.91280653950953,
+ "grad_norm": 1.097405195236206,
+ "learning_rate": 4.299804381364736e-07,
+ "loss": 0.0214,
+ "step": 33365
+ },
+ {
+ "epoch": 90.91553133514986,
+ "grad_norm": 1.56802237033844,
+ "learning_rate": 4.2972448006626366e-07,
+ "loss": 0.0286,
+ "step": 33366
+ },
+ {
+ "epoch": 90.9182561307902,
+ "grad_norm": 1.046995759010315,
+ "learning_rate": 4.294685965290535e-07,
+ "loss": 0.0073,
+ "step": 33367
+ },
+ {
+ "epoch": 90.92098092643052,
+ "grad_norm": 1.1386696100234985,
+ "learning_rate": 4.292127875268326e-07,
+ "loss": 0.0216,
+ "step": 33368
+ },
+ {
+ "epoch": 90.92370572207085,
+ "grad_norm": 1.52900230884552,
+ "learning_rate": 4.2895705306159497e-07,
+ "loss": 0.0554,
+ "step": 33369
+ },
+ {
+ "epoch": 90.92643051771117,
+ "grad_norm": 1.4722065925598145,
+ "learning_rate": 4.2870139313533013e-07,
+ "loss": 0.0119,
+ "step": 33370
+ },
+ {
+ "epoch": 90.9291553133515,
+ "grad_norm": 0.6934711933135986,
+ "learning_rate": 4.284458077500309e-07,
+ "loss": 0.0076,
+ "step": 33371
+ },
+ {
+ "epoch": 90.93188010899182,
+ "grad_norm": 0.9682675004005432,
+ "learning_rate": 4.2819029690768896e-07,
+ "loss": 0.0096,
+ "step": 33372
+ },
+ {
+ "epoch": 90.93460490463215,
+ "grad_norm": 0.94873046875,
+ "learning_rate": 4.27934860610294e-07,
+ "loss": 0.011,
+ "step": 33373
+ },
+ {
+ "epoch": 90.93732970027249,
+ "grad_norm": 0.8113616108894348,
+ "learning_rate": 4.2767949885983316e-07,
+ "loss": 0.014,
+ "step": 33374
+ },
+ {
+ "epoch": 90.9400544959128,
+ "grad_norm": 1.2795168161392212,
+ "learning_rate": 4.27424211658295e-07,
+ "loss": 0.0199,
+ "step": 33375
+ },
+ {
+ "epoch": 90.94277929155314,
+ "grad_norm": 1.3506745100021362,
+ "learning_rate": 4.2716899900766994e-07,
+ "loss": 0.0213,
+ "step": 33376
+ },
+ {
+ "epoch": 90.94550408719346,
+ "grad_norm": 1.2691428661346436,
+ "learning_rate": 4.2691386090994545e-07,
+ "loss": 0.0351,
+ "step": 33377
+ },
+ {
+ "epoch": 90.94822888283379,
+ "grad_norm": 1.0262072086334229,
+ "learning_rate": 4.266587973671077e-07,
+ "loss": 0.013,
+ "step": 33378
+ },
+ {
+ "epoch": 90.95095367847412,
+ "grad_norm": 1.5117613077163696,
+ "learning_rate": 4.264038083811417e-07,
+ "loss": 0.0143,
+ "step": 33379
+ },
+ {
+ "epoch": 90.95367847411444,
+ "grad_norm": 1.2588376998901367,
+ "learning_rate": 4.261488939540348e-07,
+ "loss": 0.0068,
+ "step": 33380
+ },
+ {
+ "epoch": 90.95640326975477,
+ "grad_norm": 0.9713168144226074,
+ "learning_rate": 4.2589405408777426e-07,
+ "loss": 0.0133,
+ "step": 33381
+ },
+ {
+ "epoch": 90.95912806539509,
+ "grad_norm": 1.8134044408798218,
+ "learning_rate": 4.256392887843408e-07,
+ "loss": 0.0084,
+ "step": 33382
+ },
+ {
+ "epoch": 90.96185286103542,
+ "grad_norm": 1.5474984645843506,
+ "learning_rate": 4.2538459804572053e-07,
+ "loss": 0.0239,
+ "step": 33383
+ },
+ {
+ "epoch": 90.96457765667574,
+ "grad_norm": 1.2071853876113892,
+ "learning_rate": 4.251299818738974e-07,
+ "loss": 0.042,
+ "step": 33384
+ },
+ {
+ "epoch": 90.96730245231608,
+ "grad_norm": 0.31101930141448975,
+ "learning_rate": 4.2487544027085435e-07,
+ "loss": 0.0034,
+ "step": 33385
+ },
+ {
+ "epoch": 90.97002724795641,
+ "grad_norm": 0.9977463483810425,
+ "learning_rate": 4.246209732385709e-07,
+ "loss": 0.0091,
+ "step": 33386
+ },
+ {
+ "epoch": 90.97275204359673,
+ "grad_norm": 1.0609400272369385,
+ "learning_rate": 4.2436658077903316e-07,
+ "loss": 0.0103,
+ "step": 33387
+ },
+ {
+ "epoch": 90.97547683923706,
+ "grad_norm": 1.089484691619873,
+ "learning_rate": 4.2411226289421846e-07,
+ "loss": 0.0256,
+ "step": 33388
+ },
+ {
+ "epoch": 90.97820163487738,
+ "grad_norm": 0.6012098789215088,
+ "learning_rate": 4.2385801958611196e-07,
+ "loss": 0.0046,
+ "step": 33389
+ },
+ {
+ "epoch": 90.98092643051771,
+ "grad_norm": 1.0651785135269165,
+ "learning_rate": 4.2360385085668976e-07,
+ "loss": 0.0293,
+ "step": 33390
+ },
+ {
+ "epoch": 90.98365122615803,
+ "grad_norm": 1.0710227489471436,
+ "learning_rate": 4.2334975670793145e-07,
+ "loss": 0.0594,
+ "step": 33391
+ },
+ {
+ "epoch": 90.98637602179836,
+ "grad_norm": 0.9992578625679016,
+ "learning_rate": 4.2309573714181763e-07,
+ "loss": 0.0141,
+ "step": 33392
+ },
+ {
+ "epoch": 90.9891008174387,
+ "grad_norm": 0.7971827983856201,
+ "learning_rate": 4.2284179216032675e-07,
+ "loss": 0.0074,
+ "step": 33393
+ },
+ {
+ "epoch": 90.99182561307902,
+ "grad_norm": 0.8805732727050781,
+ "learning_rate": 4.2258792176543604e-07,
+ "loss": 0.0063,
+ "step": 33394
+ },
+ {
+ "epoch": 90.99455040871935,
+ "grad_norm": 1.3528186082839966,
+ "learning_rate": 4.223341259591207e-07,
+ "loss": 0.0435,
+ "step": 33395
+ },
+ {
+ "epoch": 90.99727520435967,
+ "grad_norm": 0.8409598469734192,
+ "learning_rate": 4.220804047433602e-07,
+ "loss": 0.0085,
+ "step": 33396
+ },
+ {
+ "epoch": 91.0,
+ "grad_norm": 1.1178979873657227,
+ "learning_rate": 4.218267581201296e-07,
+ "loss": 0.0193,
+ "step": 33397
+ },
+ {
+ "epoch": 91.00272479564033,
+ "grad_norm": 0.4104558229446411,
+ "learning_rate": 4.2157318609140407e-07,
+ "loss": 0.0028,
+ "step": 33398
+ },
+ {
+ "epoch": 91.00544959128065,
+ "grad_norm": 1.0744565725326538,
+ "learning_rate": 4.213196886591564e-07,
+ "loss": 0.0187,
+ "step": 33399
+ },
+ {
+ "epoch": 91.00817438692098,
+ "grad_norm": 1.0413529872894287,
+ "learning_rate": 4.2106626582536394e-07,
+ "loss": 0.0198,
+ "step": 33400
+ },
+ {
+ "epoch": 91.0108991825613,
+ "grad_norm": 0.876501739025116,
+ "learning_rate": 4.208129175919995e-07,
+ "loss": 0.0092,
+ "step": 33401
+ },
+ {
+ "epoch": 91.01362397820164,
+ "grad_norm": 0.5843108296394348,
+ "learning_rate": 4.205596439610349e-07,
+ "loss": 0.0054,
+ "step": 33402
+ },
+ {
+ "epoch": 91.01634877384195,
+ "grad_norm": 0.9228523373603821,
+ "learning_rate": 4.2030644493444294e-07,
+ "loss": 0.0194,
+ "step": 33403
+ },
+ {
+ "epoch": 91.01907356948229,
+ "grad_norm": 0.7975098490715027,
+ "learning_rate": 4.200533205141977e-07,
+ "loss": 0.0064,
+ "step": 33404
+ },
+ {
+ "epoch": 91.02179836512262,
+ "grad_norm": 1.039247989654541,
+ "learning_rate": 4.1980027070226857e-07,
+ "loss": 0.0774,
+ "step": 33405
+ },
+ {
+ "epoch": 91.02452316076294,
+ "grad_norm": 1.6228551864624023,
+ "learning_rate": 4.195472955006252e-07,
+ "loss": 0.0356,
+ "step": 33406
+ },
+ {
+ "epoch": 91.02724795640327,
+ "grad_norm": 1.1001535654067993,
+ "learning_rate": 4.1929439491123936e-07,
+ "loss": 0.0492,
+ "step": 33407
+ },
+ {
+ "epoch": 91.02997275204359,
+ "grad_norm": 0.9938335418701172,
+ "learning_rate": 4.190415689360816e-07,
+ "loss": 0.0273,
+ "step": 33408
+ },
+ {
+ "epoch": 91.03269754768392,
+ "grad_norm": 1.2078840732574463,
+ "learning_rate": 4.187888175771193e-07,
+ "loss": 0.09,
+ "step": 33409
+ },
+ {
+ "epoch": 91.03542234332426,
+ "grad_norm": 1.0317790508270264,
+ "learning_rate": 4.1853614083632197e-07,
+ "loss": 0.0626,
+ "step": 33410
+ },
+ {
+ "epoch": 91.03814713896458,
+ "grad_norm": 1.0393762588500977,
+ "learning_rate": 4.1828353871565473e-07,
+ "loss": 0.0111,
+ "step": 33411
+ },
+ {
+ "epoch": 91.04087193460491,
+ "grad_norm": 1.3813141584396362,
+ "learning_rate": 4.180310112170882e-07,
+ "loss": 0.0566,
+ "step": 33412
+ },
+ {
+ "epoch": 91.04359673024523,
+ "grad_norm": 0.8822026252746582,
+ "learning_rate": 4.177785583425886e-07,
+ "loss": 0.006,
+ "step": 33413
+ },
+ {
+ "epoch": 91.04632152588556,
+ "grad_norm": 1.213881015777588,
+ "learning_rate": 4.1752618009412105e-07,
+ "loss": 0.1031,
+ "step": 33414
+ },
+ {
+ "epoch": 91.04904632152588,
+ "grad_norm": 0.7107138633728027,
+ "learning_rate": 4.1727387647364945e-07,
+ "loss": 0.011,
+ "step": 33415
+ },
+ {
+ "epoch": 91.05177111716621,
+ "grad_norm": 0.7287269830703735,
+ "learning_rate": 4.1702164748314236e-07,
+ "loss": 0.0066,
+ "step": 33416
+ },
+ {
+ "epoch": 91.05449591280654,
+ "grad_norm": 0.9083626866340637,
+ "learning_rate": 4.167694931245614e-07,
+ "loss": 0.0121,
+ "step": 33417
+ },
+ {
+ "epoch": 91.05722070844686,
+ "grad_norm": 0.702420175075531,
+ "learning_rate": 4.165174133998717e-07,
+ "loss": 0.0087,
+ "step": 33418
+ },
+ {
+ "epoch": 91.0599455040872,
+ "grad_norm": 0.9924845695495605,
+ "learning_rate": 4.1626540831103513e-07,
+ "loss": 0.1408,
+ "step": 33419
+ },
+ {
+ "epoch": 91.06267029972751,
+ "grad_norm": 0.9040300250053406,
+ "learning_rate": 4.1601347786001557e-07,
+ "loss": 0.0123,
+ "step": 33420
+ },
+ {
+ "epoch": 91.06539509536785,
+ "grad_norm": 0.6057246923446655,
+ "learning_rate": 4.1576162204877477e-07,
+ "loss": 0.0086,
+ "step": 33421
+ },
+ {
+ "epoch": 91.06811989100818,
+ "grad_norm": 1.277686357498169,
+ "learning_rate": 4.155098408792735e-07,
+ "loss": 0.0084,
+ "step": 33422
+ },
+ {
+ "epoch": 91.0708446866485,
+ "grad_norm": 0.969914436340332,
+ "learning_rate": 4.152581343534734e-07,
+ "loss": 0.006,
+ "step": 33423
+ },
+ {
+ "epoch": 91.07356948228883,
+ "grad_norm": 1.1470130681991577,
+ "learning_rate": 4.1500650247333294e-07,
+ "loss": 0.0504,
+ "step": 33424
+ },
+ {
+ "epoch": 91.07629427792915,
+ "grad_norm": 1.1389007568359375,
+ "learning_rate": 4.1475494524081504e-07,
+ "loss": 0.1146,
+ "step": 33425
+ },
+ {
+ "epoch": 91.07901907356948,
+ "grad_norm": 1.5579596757888794,
+ "learning_rate": 4.14503462657877e-07,
+ "loss": 0.0083,
+ "step": 33426
+ },
+ {
+ "epoch": 91.0817438692098,
+ "grad_norm": 0.9181563258171082,
+ "learning_rate": 4.142520547264772e-07,
+ "loss": 0.0623,
+ "step": 33427
+ },
+ {
+ "epoch": 91.08446866485014,
+ "grad_norm": 0.5640206336975098,
+ "learning_rate": 4.140007214485742e-07,
+ "loss": 0.005,
+ "step": 33428
+ },
+ {
+ "epoch": 91.08719346049047,
+ "grad_norm": 1.0247015953063965,
+ "learning_rate": 4.1374946282612517e-07,
+ "loss": 0.0086,
+ "step": 33429
+ },
+ {
+ "epoch": 91.08991825613079,
+ "grad_norm": 1.3333468437194824,
+ "learning_rate": 4.134982788610875e-07,
+ "loss": 0.0549,
+ "step": 33430
+ },
+ {
+ "epoch": 91.09264305177112,
+ "grad_norm": 0.8582410216331482,
+ "learning_rate": 4.132471695554141e-07,
+ "loss": 0.0143,
+ "step": 33431
+ },
+ {
+ "epoch": 91.09536784741144,
+ "grad_norm": 0.9757174253463745,
+ "learning_rate": 4.1299613491106563e-07,
+ "loss": 0.0268,
+ "step": 33432
+ },
+ {
+ "epoch": 91.09809264305177,
+ "grad_norm": 1.247473120689392,
+ "learning_rate": 4.127451749299938e-07,
+ "loss": 0.0336,
+ "step": 33433
+ },
+ {
+ "epoch": 91.1008174386921,
+ "grad_norm": 0.40233278274536133,
+ "learning_rate": 4.1249428961415485e-07,
+ "loss": 0.0039,
+ "step": 33434
+ },
+ {
+ "epoch": 91.10354223433242,
+ "grad_norm": 1.249998927116394,
+ "learning_rate": 4.122434789655005e-07,
+ "loss": 0.0154,
+ "step": 33435
+ },
+ {
+ "epoch": 91.10626702997276,
+ "grad_norm": 1.1949518918991089,
+ "learning_rate": 4.11992742985986e-07,
+ "loss": 0.0484,
+ "step": 33436
+ },
+ {
+ "epoch": 91.10899182561307,
+ "grad_norm": 1.1788967847824097,
+ "learning_rate": 4.1174208167756414e-07,
+ "loss": 0.0088,
+ "step": 33437
+ },
+ {
+ "epoch": 91.11171662125341,
+ "grad_norm": 1.1904077529907227,
+ "learning_rate": 4.1149149504218553e-07,
+ "loss": 0.0153,
+ "step": 33438
+ },
+ {
+ "epoch": 91.11444141689373,
+ "grad_norm": 1.2112261056900024,
+ "learning_rate": 4.1124098308180204e-07,
+ "loss": 0.0492,
+ "step": 33439
+ },
+ {
+ "epoch": 91.11716621253406,
+ "grad_norm": 1.511974811553955,
+ "learning_rate": 4.109905457983643e-07,
+ "loss": 0.0184,
+ "step": 33440
+ },
+ {
+ "epoch": 91.11989100817439,
+ "grad_norm": 1.2359580993652344,
+ "learning_rate": 4.1074018319382513e-07,
+ "loss": 0.0054,
+ "step": 33441
+ },
+ {
+ "epoch": 91.12261580381471,
+ "grad_norm": 0.9134237766265869,
+ "learning_rate": 4.1048989527013306e-07,
+ "loss": 0.0064,
+ "step": 33442
+ },
+ {
+ "epoch": 91.12534059945504,
+ "grad_norm": 0.7421379089355469,
+ "learning_rate": 4.102396820292376e-07,
+ "loss": 0.0087,
+ "step": 33443
+ },
+ {
+ "epoch": 91.12806539509536,
+ "grad_norm": 0.8784224390983582,
+ "learning_rate": 4.0998954347308496e-07,
+ "loss": 0.0172,
+ "step": 33444
+ },
+ {
+ "epoch": 91.1307901907357,
+ "grad_norm": 1.127231240272522,
+ "learning_rate": 4.0973947960362583e-07,
+ "loss": 0.0487,
+ "step": 33445
+ },
+ {
+ "epoch": 91.13351498637603,
+ "grad_norm": 0.7883209586143494,
+ "learning_rate": 4.094894904228075e-07,
+ "loss": 0.0058,
+ "step": 33446
+ },
+ {
+ "epoch": 91.13623978201635,
+ "grad_norm": 1.4822059869766235,
+ "learning_rate": 4.0923957593257626e-07,
+ "loss": 0.116,
+ "step": 33447
+ },
+ {
+ "epoch": 91.13896457765668,
+ "grad_norm": 1.4443351030349731,
+ "learning_rate": 4.0898973613487714e-07,
+ "loss": 0.0163,
+ "step": 33448
+ },
+ {
+ "epoch": 91.141689373297,
+ "grad_norm": 0.9251310229301453,
+ "learning_rate": 4.0873997103165976e-07,
+ "loss": 0.0087,
+ "step": 33449
+ },
+ {
+ "epoch": 91.14441416893733,
+ "grad_norm": 1.0791784524917603,
+ "learning_rate": 4.084902806248647e-07,
+ "loss": 0.012,
+ "step": 33450
+ },
+ {
+ "epoch": 91.14713896457765,
+ "grad_norm": 1.0705912113189697,
+ "learning_rate": 4.0824066491643944e-07,
+ "loss": 0.0071,
+ "step": 33451
+ },
+ {
+ "epoch": 91.14986376021798,
+ "grad_norm": 1.180453896522522,
+ "learning_rate": 4.079911239083267e-07,
+ "loss": 0.0086,
+ "step": 33452
+ },
+ {
+ "epoch": 91.15258855585832,
+ "grad_norm": 2.5612995624542236,
+ "learning_rate": 4.0774165760247177e-07,
+ "loss": 0.0175,
+ "step": 33453
+ },
+ {
+ "epoch": 91.15531335149863,
+ "grad_norm": 1.1194065809249878,
+ "learning_rate": 4.0749226600081513e-07,
+ "loss": 0.0354,
+ "step": 33454
+ },
+ {
+ "epoch": 91.15803814713897,
+ "grad_norm": 1.2864845991134644,
+ "learning_rate": 4.0724294910529873e-07,
+ "loss": 0.0248,
+ "step": 33455
+ },
+ {
+ "epoch": 91.16076294277929,
+ "grad_norm": 1.653692364692688,
+ "learning_rate": 4.069937069178675e-07,
+ "loss": 0.0161,
+ "step": 33456
+ },
+ {
+ "epoch": 91.16348773841962,
+ "grad_norm": 4.366382598876953,
+ "learning_rate": 4.06744539440459e-07,
+ "loss": 0.0096,
+ "step": 33457
+ },
+ {
+ "epoch": 91.16621253405995,
+ "grad_norm": 0.7484425902366638,
+ "learning_rate": 4.0649544667501595e-07,
+ "loss": 0.0053,
+ "step": 33458
+ },
+ {
+ "epoch": 91.16893732970027,
+ "grad_norm": 1.8122172355651855,
+ "learning_rate": 4.0624642862347796e-07,
+ "loss": 0.0673,
+ "step": 33459
+ },
+ {
+ "epoch": 91.1716621253406,
+ "grad_norm": 0.8682962656021118,
+ "learning_rate": 4.059974852877824e-07,
+ "loss": 0.006,
+ "step": 33460
+ },
+ {
+ "epoch": 91.17438692098092,
+ "grad_norm": 1.6025928258895874,
+ "learning_rate": 4.057486166698721e-07,
+ "loss": 0.0214,
+ "step": 33461
+ },
+ {
+ "epoch": 91.17711171662125,
+ "grad_norm": 1.4229069948196411,
+ "learning_rate": 4.054998227716822e-07,
+ "loss": 0.0086,
+ "step": 33462
+ },
+ {
+ "epoch": 91.17983651226157,
+ "grad_norm": 1.8004448413848877,
+ "learning_rate": 4.0525110359515005e-07,
+ "loss": 0.0356,
+ "step": 33463
+ },
+ {
+ "epoch": 91.1825613079019,
+ "grad_norm": 0.8500205874443054,
+ "learning_rate": 4.0500245914221303e-07,
+ "loss": 0.0138,
+ "step": 33464
+ },
+ {
+ "epoch": 91.18528610354224,
+ "grad_norm": 0.8952769041061401,
+ "learning_rate": 4.0475388941480955e-07,
+ "loss": 0.0363,
+ "step": 33465
+ },
+ {
+ "epoch": 91.18801089918256,
+ "grad_norm": 1.449439287185669,
+ "learning_rate": 4.045053944148736e-07,
+ "loss": 0.062,
+ "step": 33466
+ },
+ {
+ "epoch": 91.19073569482289,
+ "grad_norm": 0.9832485318183899,
+ "learning_rate": 4.0425697414434027e-07,
+ "loss": 0.0132,
+ "step": 33467
+ },
+ {
+ "epoch": 91.19346049046321,
+ "grad_norm": 0.9745160937309265,
+ "learning_rate": 4.0400862860514366e-07,
+ "loss": 0.0159,
+ "step": 33468
+ },
+ {
+ "epoch": 91.19618528610354,
+ "grad_norm": 0.9293946623802185,
+ "learning_rate": 4.037603577992211e-07,
+ "loss": 0.0108,
+ "step": 33469
+ },
+ {
+ "epoch": 91.19891008174388,
+ "grad_norm": 1.4092068672180176,
+ "learning_rate": 4.035121617285032e-07,
+ "loss": 0.1164,
+ "step": 33470
+ },
+ {
+ "epoch": 91.2016348773842,
+ "grad_norm": 0.5300266742706299,
+ "learning_rate": 4.0326404039492283e-07,
+ "loss": 0.0053,
+ "step": 33471
+ },
+ {
+ "epoch": 91.20435967302453,
+ "grad_norm": 0.985741913318634,
+ "learning_rate": 4.030159938004141e-07,
+ "loss": 0.0132,
+ "step": 33472
+ },
+ {
+ "epoch": 91.20708446866485,
+ "grad_norm": 3.6153371334075928,
+ "learning_rate": 4.027680219469088e-07,
+ "loss": 0.0116,
+ "step": 33473
+ },
+ {
+ "epoch": 91.20980926430518,
+ "grad_norm": 1.5514990091323853,
+ "learning_rate": 4.025201248363353e-07,
+ "loss": 0.0253,
+ "step": 33474
+ },
+ {
+ "epoch": 91.2125340599455,
+ "grad_norm": 1.0500550270080566,
+ "learning_rate": 4.0227230247062765e-07,
+ "loss": 0.0222,
+ "step": 33475
+ },
+ {
+ "epoch": 91.21525885558583,
+ "grad_norm": 0.8476006984710693,
+ "learning_rate": 4.020245548517132e-07,
+ "loss": 0.0106,
+ "step": 33476
+ },
+ {
+ "epoch": 91.21798365122616,
+ "grad_norm": 0.7531718611717224,
+ "learning_rate": 4.017768819815238e-07,
+ "loss": 0.0052,
+ "step": 33477
+ },
+ {
+ "epoch": 91.22070844686648,
+ "grad_norm": 1.4597316980361938,
+ "learning_rate": 4.0152928386198775e-07,
+ "loss": 0.0377,
+ "step": 33478
+ },
+ {
+ "epoch": 91.22343324250681,
+ "grad_norm": 0.7313485741615295,
+ "learning_rate": 4.0128176049503145e-07,
+ "loss": 0.0057,
+ "step": 33479
+ },
+ {
+ "epoch": 91.22615803814713,
+ "grad_norm": 0.7825775146484375,
+ "learning_rate": 4.0103431188258324e-07,
+ "loss": 0.0117,
+ "step": 33480
+ },
+ {
+ "epoch": 91.22888283378747,
+ "grad_norm": 0.6493341326713562,
+ "learning_rate": 4.0078693802657273e-07,
+ "loss": 0.0069,
+ "step": 33481
+ },
+ {
+ "epoch": 91.2316076294278,
+ "grad_norm": 0.6419547200202942,
+ "learning_rate": 4.00539638928924e-07,
+ "loss": 0.0073,
+ "step": 33482
+ },
+ {
+ "epoch": 91.23433242506812,
+ "grad_norm": 0.9720820784568787,
+ "learning_rate": 4.002924145915632e-07,
+ "loss": 0.0093,
+ "step": 33483
+ },
+ {
+ "epoch": 91.23705722070845,
+ "grad_norm": 1.227913737297058,
+ "learning_rate": 4.000452650164155e-07,
+ "loss": 0.0117,
+ "step": 33484
+ },
+ {
+ "epoch": 91.23978201634877,
+ "grad_norm": 1.001293659210205,
+ "learning_rate": 3.997981902054071e-07,
+ "loss": 0.0273,
+ "step": 33485
+ },
+ {
+ "epoch": 91.2425068119891,
+ "grad_norm": 0.8048069477081299,
+ "learning_rate": 3.9955119016046206e-07,
+ "loss": 0.0114,
+ "step": 33486
+ },
+ {
+ "epoch": 91.24523160762942,
+ "grad_norm": 0.7145854830741882,
+ "learning_rate": 3.9930426488350215e-07,
+ "loss": 0.009,
+ "step": 33487
+ },
+ {
+ "epoch": 91.24795640326975,
+ "grad_norm": 1.0419774055480957,
+ "learning_rate": 3.990574143764514e-07,
+ "loss": 0.0255,
+ "step": 33488
+ },
+ {
+ "epoch": 91.25068119891009,
+ "grad_norm": 1.3436946868896484,
+ "learning_rate": 3.988106386412338e-07,
+ "loss": 0.1324,
+ "step": 33489
+ },
+ {
+ "epoch": 91.2534059945504,
+ "grad_norm": 0.9511478543281555,
+ "learning_rate": 3.9856393767976896e-07,
+ "loss": 0.0163,
+ "step": 33490
+ },
+ {
+ "epoch": 91.25613079019074,
+ "grad_norm": 0.8115828037261963,
+ "learning_rate": 3.983173114939798e-07,
+ "loss": 0.1162,
+ "step": 33491
+ },
+ {
+ "epoch": 91.25885558583106,
+ "grad_norm": 1.0396617650985718,
+ "learning_rate": 3.980707600857847e-07,
+ "loss": 0.0425,
+ "step": 33492
+ },
+ {
+ "epoch": 91.26158038147139,
+ "grad_norm": 1.080506682395935,
+ "learning_rate": 3.9782428345710665e-07,
+ "loss": 0.0155,
+ "step": 33493
+ },
+ {
+ "epoch": 91.26430517711172,
+ "grad_norm": 0.8016747832298279,
+ "learning_rate": 3.9757788160986524e-07,
+ "loss": 0.0046,
+ "step": 33494
+ },
+ {
+ "epoch": 91.26702997275204,
+ "grad_norm": 0.6809645891189575,
+ "learning_rate": 3.9733155454597774e-07,
+ "loss": 0.0106,
+ "step": 33495
+ },
+ {
+ "epoch": 91.26975476839237,
+ "grad_norm": 1.6735765933990479,
+ "learning_rate": 3.970853022673615e-07,
+ "loss": 0.0186,
+ "step": 33496
+ },
+ {
+ "epoch": 91.2724795640327,
+ "grad_norm": 1.2794408798217773,
+ "learning_rate": 3.9683912477593734e-07,
+ "loss": 0.0407,
+ "step": 33497
+ },
+ {
+ "epoch": 91.27520435967303,
+ "grad_norm": 1.2309173345565796,
+ "learning_rate": 3.965930220736203e-07,
+ "loss": 0.0249,
+ "step": 33498
+ },
+ {
+ "epoch": 91.27792915531334,
+ "grad_norm": 1.2911275625228882,
+ "learning_rate": 3.963469941623288e-07,
+ "loss": 0.0101,
+ "step": 33499
+ },
+ {
+ "epoch": 91.28065395095368,
+ "grad_norm": 1.01865816116333,
+ "learning_rate": 3.9610104104397586e-07,
+ "loss": 0.0129,
+ "step": 33500
+ },
+ {
+ "epoch": 91.28337874659401,
+ "grad_norm": 1.4247344732284546,
+ "learning_rate": 3.9585516272047984e-07,
+ "loss": 0.0154,
+ "step": 33501
+ },
+ {
+ "epoch": 91.28610354223433,
+ "grad_norm": 0.4860742390155792,
+ "learning_rate": 3.9560935919375597e-07,
+ "loss": 0.0041,
+ "step": 33502
+ },
+ {
+ "epoch": 91.28882833787466,
+ "grad_norm": 0.8528375029563904,
+ "learning_rate": 3.95363630465716e-07,
+ "loss": 0.0085,
+ "step": 33503
+ },
+ {
+ "epoch": 91.29155313351498,
+ "grad_norm": 1.2869775295257568,
+ "learning_rate": 3.95117976538274e-07,
+ "loss": 0.0051,
+ "step": 33504
+ },
+ {
+ "epoch": 91.29427792915531,
+ "grad_norm": 1.01718008518219,
+ "learning_rate": 3.9487239741334615e-07,
+ "loss": 0.0184,
+ "step": 33505
+ },
+ {
+ "epoch": 91.29700272479565,
+ "grad_norm": 0.807528018951416,
+ "learning_rate": 3.946268930928421e-07,
+ "loss": 0.0079,
+ "step": 33506
+ },
+ {
+ "epoch": 91.29972752043597,
+ "grad_norm": 0.8977676630020142,
+ "learning_rate": 3.9438146357867467e-07,
+ "loss": 0.0109,
+ "step": 33507
+ },
+ {
+ "epoch": 91.3024523160763,
+ "grad_norm": 4.233188629150391,
+ "learning_rate": 3.9413610887275466e-07,
+ "loss": 0.0206,
+ "step": 33508
+ },
+ {
+ "epoch": 91.30517711171662,
+ "grad_norm": 0.9372884631156921,
+ "learning_rate": 3.938908289769927e-07,
+ "loss": 0.0429,
+ "step": 33509
+ },
+ {
+ "epoch": 91.30790190735695,
+ "grad_norm": 0.6719347834587097,
+ "learning_rate": 3.9364562389330174e-07,
+ "loss": 0.0093,
+ "step": 33510
+ },
+ {
+ "epoch": 91.31062670299727,
+ "grad_norm": 1.383008599281311,
+ "learning_rate": 3.934004936235891e-07,
+ "loss": 0.055,
+ "step": 33511
+ },
+ {
+ "epoch": 91.3133514986376,
+ "grad_norm": 0.910950779914856,
+ "learning_rate": 3.9315543816976444e-07,
+ "loss": 0.054,
+ "step": 33512
+ },
+ {
+ "epoch": 91.31607629427793,
+ "grad_norm": 0.9888265132904053,
+ "learning_rate": 3.92910457533735e-07,
+ "loss": 0.0108,
+ "step": 33513
+ },
+ {
+ "epoch": 91.31880108991825,
+ "grad_norm": 1.1300674676895142,
+ "learning_rate": 3.926655517174105e-07,
+ "loss": 0.023,
+ "step": 33514
+ },
+ {
+ "epoch": 91.32152588555859,
+ "grad_norm": 1.120875597000122,
+ "learning_rate": 3.924207207226971e-07,
+ "loss": 0.015,
+ "step": 33515
+ },
+ {
+ "epoch": 91.3242506811989,
+ "grad_norm": 0.9217361211776733,
+ "learning_rate": 3.9217596455150107e-07,
+ "loss": 0.0123,
+ "step": 33516
+ },
+ {
+ "epoch": 91.32697547683924,
+ "grad_norm": 1.1585452556610107,
+ "learning_rate": 3.9193128320573095e-07,
+ "loss": 0.0092,
+ "step": 33517
+ },
+ {
+ "epoch": 91.32970027247957,
+ "grad_norm": 1.6630866527557373,
+ "learning_rate": 3.916866766872907e-07,
+ "loss": 0.0185,
+ "step": 33518
+ },
+ {
+ "epoch": 91.33242506811989,
+ "grad_norm": 2.000761032104492,
+ "learning_rate": 3.914421449980854e-07,
+ "loss": 0.0452,
+ "step": 33519
+ },
+ {
+ "epoch": 91.33514986376022,
+ "grad_norm": 0.6191617846488953,
+ "learning_rate": 3.9119768814001813e-07,
+ "loss": 0.0045,
+ "step": 33520
+ },
+ {
+ "epoch": 91.33787465940054,
+ "grad_norm": 0.8229928612709045,
+ "learning_rate": 3.9095330611499506e-07,
+ "loss": 0.0092,
+ "step": 33521
+ },
+ {
+ "epoch": 91.34059945504087,
+ "grad_norm": 0.7808035016059875,
+ "learning_rate": 3.90708998924918e-07,
+ "loss": 0.0079,
+ "step": 33522
+ },
+ {
+ "epoch": 91.34332425068119,
+ "grad_norm": 0.774378776550293,
+ "learning_rate": 3.9046476657168985e-07,
+ "loss": 0.0053,
+ "step": 33523
+ },
+ {
+ "epoch": 91.34604904632153,
+ "grad_norm": 0.9986081719398499,
+ "learning_rate": 3.9022060905721246e-07,
+ "loss": 0.121,
+ "step": 33524
+ },
+ {
+ "epoch": 91.34877384196186,
+ "grad_norm": 1.1169133186340332,
+ "learning_rate": 3.8997652638338876e-07,
+ "loss": 0.013,
+ "step": 33525
+ },
+ {
+ "epoch": 91.35149863760218,
+ "grad_norm": 0.896031379699707,
+ "learning_rate": 3.8973251855211726e-07,
+ "loss": 0.0331,
+ "step": 33526
+ },
+ {
+ "epoch": 91.35422343324251,
+ "grad_norm": 1.2542587518692017,
+ "learning_rate": 3.8948858556530077e-07,
+ "loss": 0.0883,
+ "step": 33527
+ },
+ {
+ "epoch": 91.35694822888283,
+ "grad_norm": 0.48640063405036926,
+ "learning_rate": 3.8924472742483785e-07,
+ "loss": 0.0041,
+ "step": 33528
+ },
+ {
+ "epoch": 91.35967302452316,
+ "grad_norm": 1.263288974761963,
+ "learning_rate": 3.890009441326259e-07,
+ "loss": 0.0987,
+ "step": 33529
+ },
+ {
+ "epoch": 91.3623978201635,
+ "grad_norm": 1.0591069459915161,
+ "learning_rate": 3.887572356905678e-07,
+ "loss": 0.0641,
+ "step": 33530
+ },
+ {
+ "epoch": 91.36512261580381,
+ "grad_norm": 0.8181588649749756,
+ "learning_rate": 3.8851360210055755e-07,
+ "loss": 0.0085,
+ "step": 33531
+ },
+ {
+ "epoch": 91.36784741144415,
+ "grad_norm": 0.9466190338134766,
+ "learning_rate": 3.882700433644948e-07,
+ "loss": 0.0136,
+ "step": 33532
+ },
+ {
+ "epoch": 91.37057220708446,
+ "grad_norm": 1.1365543603897095,
+ "learning_rate": 3.880265594842747e-07,
+ "loss": 0.0129,
+ "step": 33533
+ },
+ {
+ "epoch": 91.3732970027248,
+ "grad_norm": 0.9922269582748413,
+ "learning_rate": 3.877831504617946e-07,
+ "loss": 0.0142,
+ "step": 33534
+ },
+ {
+ "epoch": 91.37602179836512,
+ "grad_norm": 1.0767585039138794,
+ "learning_rate": 3.8753981629895076e-07,
+ "loss": 0.0236,
+ "step": 33535
+ },
+ {
+ "epoch": 91.37874659400545,
+ "grad_norm": 0.930620014667511,
+ "learning_rate": 3.8729655699763613e-07,
+ "loss": 0.0499,
+ "step": 33536
+ },
+ {
+ "epoch": 91.38147138964578,
+ "grad_norm": 1.10387122631073,
+ "learning_rate": 3.8705337255974697e-07,
+ "loss": 0.0076,
+ "step": 33537
+ },
+ {
+ "epoch": 91.3841961852861,
+ "grad_norm": 0.9443877339363098,
+ "learning_rate": 3.8681026298717836e-07,
+ "loss": 0.0058,
+ "step": 33538
+ },
+ {
+ "epoch": 91.38692098092643,
+ "grad_norm": 3.2628121376037598,
+ "learning_rate": 3.8656722828182115e-07,
+ "loss": 0.0091,
+ "step": 33539
+ },
+ {
+ "epoch": 91.38964577656675,
+ "grad_norm": 0.9235361218452454,
+ "learning_rate": 3.86324268445567e-07,
+ "loss": 0.0058,
+ "step": 33540
+ },
+ {
+ "epoch": 91.39237057220708,
+ "grad_norm": 0.7104455232620239,
+ "learning_rate": 3.860813834803123e-07,
+ "loss": 0.0084,
+ "step": 33541
+ },
+ {
+ "epoch": 91.39509536784742,
+ "grad_norm": 0.8874745965003967,
+ "learning_rate": 3.858385733879455e-07,
+ "loss": 0.0072,
+ "step": 33542
+ },
+ {
+ "epoch": 91.39782016348774,
+ "grad_norm": 0.8060672879219055,
+ "learning_rate": 3.855958381703584e-07,
+ "loss": 0.0075,
+ "step": 33543
+ },
+ {
+ "epoch": 91.40054495912807,
+ "grad_norm": 0.9021698832511902,
+ "learning_rate": 3.853531778294417e-07,
+ "loss": 0.0057,
+ "step": 33544
+ },
+ {
+ "epoch": 91.40326975476839,
+ "grad_norm": 1.533369541168213,
+ "learning_rate": 3.851105923670839e-07,
+ "loss": 0.0086,
+ "step": 33545
+ },
+ {
+ "epoch": 91.40599455040872,
+ "grad_norm": 0.37478283047676086,
+ "learning_rate": 3.848680817851769e-07,
+ "loss": 0.0032,
+ "step": 33546
+ },
+ {
+ "epoch": 91.40871934604904,
+ "grad_norm": 1.350470781326294,
+ "learning_rate": 3.8462564608560793e-07,
+ "loss": 0.0211,
+ "step": 33547
+ },
+ {
+ "epoch": 91.41144414168937,
+ "grad_norm": 1.3947505950927734,
+ "learning_rate": 3.8438328527026445e-07,
+ "loss": 0.0753,
+ "step": 33548
+ },
+ {
+ "epoch": 91.4141689373297,
+ "grad_norm": 1.3809478282928467,
+ "learning_rate": 3.8414099934103386e-07,
+ "loss": 0.0555,
+ "step": 33549
+ },
+ {
+ "epoch": 91.41689373297002,
+ "grad_norm": 0.572896420955658,
+ "learning_rate": 3.8389878829980465e-07,
+ "loss": 0.0078,
+ "step": 33550
+ },
+ {
+ "epoch": 91.41961852861036,
+ "grad_norm": 0.7747504115104675,
+ "learning_rate": 3.836566521484619e-07,
+ "loss": 0.0101,
+ "step": 33551
+ },
+ {
+ "epoch": 91.42234332425068,
+ "grad_norm": 1.486462116241455,
+ "learning_rate": 3.83414590888892e-07,
+ "loss": 0.0263,
+ "step": 33552
+ },
+ {
+ "epoch": 91.42506811989101,
+ "grad_norm": 1.0553442239761353,
+ "learning_rate": 3.8317260452297776e-07,
+ "loss": 0.016,
+ "step": 33553
+ },
+ {
+ "epoch": 91.42779291553134,
+ "grad_norm": 0.7874895930290222,
+ "learning_rate": 3.8293069305260776e-07,
+ "loss": 0.0104,
+ "step": 33554
+ },
+ {
+ "epoch": 91.43051771117166,
+ "grad_norm": 1.050277590751648,
+ "learning_rate": 3.826888564796638e-07,
+ "loss": 0.0412,
+ "step": 33555
+ },
+ {
+ "epoch": 91.433242506812,
+ "grad_norm": 0.6901969909667969,
+ "learning_rate": 3.824470948060299e-07,
+ "loss": 0.0053,
+ "step": 33556
+ },
+ {
+ "epoch": 91.43596730245231,
+ "grad_norm": 0.7721924781799316,
+ "learning_rate": 3.8220540803358687e-07,
+ "loss": 0.0114,
+ "step": 33557
+ },
+ {
+ "epoch": 91.43869209809264,
+ "grad_norm": 1.9121111631393433,
+ "learning_rate": 3.8196379616421975e-07,
+ "loss": 0.0871,
+ "step": 33558
+ },
+ {
+ "epoch": 91.44141689373296,
+ "grad_norm": 1.0173461437225342,
+ "learning_rate": 3.817222591998093e-07,
+ "loss": 0.015,
+ "step": 33559
+ },
+ {
+ "epoch": 91.4441416893733,
+ "grad_norm": 0.5539981722831726,
+ "learning_rate": 3.8148079714223407e-07,
+ "loss": 0.005,
+ "step": 33560
+ },
+ {
+ "epoch": 91.44686648501363,
+ "grad_norm": 1.5124554634094238,
+ "learning_rate": 3.812394099933781e-07,
+ "loss": 0.0206,
+ "step": 33561
+ },
+ {
+ "epoch": 91.44959128065395,
+ "grad_norm": 1.0527797937393188,
+ "learning_rate": 3.809980977551209e-07,
+ "loss": 0.0123,
+ "step": 33562
+ },
+ {
+ "epoch": 91.45231607629428,
+ "grad_norm": 0.4206392467021942,
+ "learning_rate": 3.80756860429341e-07,
+ "loss": 0.0042,
+ "step": 33563
+ },
+ {
+ "epoch": 91.4550408719346,
+ "grad_norm": 0.7454065680503845,
+ "learning_rate": 3.80515698017917e-07,
+ "loss": 0.1372,
+ "step": 33564
+ },
+ {
+ "epoch": 91.45776566757493,
+ "grad_norm": 1.1127605438232422,
+ "learning_rate": 3.802746105227251e-07,
+ "loss": 0.0074,
+ "step": 33565
+ },
+ {
+ "epoch": 91.46049046321527,
+ "grad_norm": 0.7901374101638794,
+ "learning_rate": 3.800335979456471e-07,
+ "loss": 0.0081,
+ "step": 33566
+ },
+ {
+ "epoch": 91.46321525885558,
+ "grad_norm": 1.0543724298477173,
+ "learning_rate": 3.797926602885582e-07,
+ "loss": 0.0082,
+ "step": 33567
+ },
+ {
+ "epoch": 91.46594005449592,
+ "grad_norm": 1.0314801931381226,
+ "learning_rate": 3.795517975533336e-07,
+ "loss": 0.0133,
+ "step": 33568
+ },
+ {
+ "epoch": 91.46866485013624,
+ "grad_norm": 0.9203073382377625,
+ "learning_rate": 3.7931100974184843e-07,
+ "loss": 0.0071,
+ "step": 33569
+ },
+ {
+ "epoch": 91.47138964577657,
+ "grad_norm": 0.8464831709861755,
+ "learning_rate": 3.790702968559812e-07,
+ "loss": 0.0148,
+ "step": 33570
+ },
+ {
+ "epoch": 91.47411444141689,
+ "grad_norm": 1.2529977560043335,
+ "learning_rate": 3.788296588976048e-07,
+ "loss": 0.0325,
+ "step": 33571
+ },
+ {
+ "epoch": 91.47683923705722,
+ "grad_norm": 0.5883649587631226,
+ "learning_rate": 3.785890958685945e-07,
+ "loss": 0.0052,
+ "step": 33572
+ },
+ {
+ "epoch": 91.47956403269755,
+ "grad_norm": 0.9162911176681519,
+ "learning_rate": 3.783486077708198e-07,
+ "loss": 0.0082,
+ "step": 33573
+ },
+ {
+ "epoch": 91.48228882833787,
+ "grad_norm": 0.9750089645385742,
+ "learning_rate": 3.781081946061582e-07,
+ "loss": 0.0203,
+ "step": 33574
+ },
+ {
+ "epoch": 91.4850136239782,
+ "grad_norm": 0.9026535749435425,
+ "learning_rate": 3.778678563764804e-07,
+ "loss": 0.0093,
+ "step": 33575
+ },
+ {
+ "epoch": 91.48773841961852,
+ "grad_norm": 0.9216136336326599,
+ "learning_rate": 3.7762759308365814e-07,
+ "loss": 0.0089,
+ "step": 33576
+ },
+ {
+ "epoch": 91.49046321525886,
+ "grad_norm": 1.422658920288086,
+ "learning_rate": 3.7738740472956115e-07,
+ "loss": 0.0171,
+ "step": 33577
+ },
+ {
+ "epoch": 91.49318801089919,
+ "grad_norm": 1.405768632888794,
+ "learning_rate": 3.771472913160612e-07,
+ "loss": 0.0352,
+ "step": 33578
+ },
+ {
+ "epoch": 91.49591280653951,
+ "grad_norm": 1.2883291244506836,
+ "learning_rate": 3.769072528450302e-07,
+ "loss": 0.0535,
+ "step": 33579
+ },
+ {
+ "epoch": 91.49863760217984,
+ "grad_norm": 1.0081130266189575,
+ "learning_rate": 3.766672893183354e-07,
+ "loss": 0.0086,
+ "step": 33580
+ },
+ {
+ "epoch": 91.50136239782016,
+ "grad_norm": 0.39544668793678284,
+ "learning_rate": 3.7642740073784435e-07,
+ "loss": 0.0033,
+ "step": 33581
+ },
+ {
+ "epoch": 91.50408719346049,
+ "grad_norm": 0.8374701142311096,
+ "learning_rate": 3.761875871054288e-07,
+ "loss": 0.0102,
+ "step": 33582
+ },
+ {
+ "epoch": 91.50681198910081,
+ "grad_norm": 1.5071488618850708,
+ "learning_rate": 3.7594784842295505e-07,
+ "loss": 0.1176,
+ "step": 33583
+ },
+ {
+ "epoch": 91.50953678474114,
+ "grad_norm": 1.9059754610061646,
+ "learning_rate": 3.7570818469228834e-07,
+ "loss": 0.0465,
+ "step": 33584
+ },
+ {
+ "epoch": 91.51226158038148,
+ "grad_norm": 1.7414790391921997,
+ "learning_rate": 3.754685959152971e-07,
+ "loss": 0.0497,
+ "step": 33585
+ },
+ {
+ "epoch": 91.5149863760218,
+ "grad_norm": 1.2218924760818481,
+ "learning_rate": 3.7522908209384644e-07,
+ "loss": 0.1204,
+ "step": 33586
+ },
+ {
+ "epoch": 91.51771117166213,
+ "grad_norm": 1.1852672100067139,
+ "learning_rate": 3.749896432298017e-07,
+ "loss": 0.0411,
+ "step": 33587
+ },
+ {
+ "epoch": 91.52043596730245,
+ "grad_norm": 1.6549838781356812,
+ "learning_rate": 3.7475027932502904e-07,
+ "loss": 0.0739,
+ "step": 33588
+ },
+ {
+ "epoch": 91.52316076294278,
+ "grad_norm": 0.8813589215278625,
+ "learning_rate": 3.7451099038138926e-07,
+ "loss": 0.0195,
+ "step": 33589
+ },
+ {
+ "epoch": 91.52588555858311,
+ "grad_norm": 0.5838894844055176,
+ "learning_rate": 3.742717764007497e-07,
+ "loss": 0.0065,
+ "step": 33590
+ },
+ {
+ "epoch": 91.52861035422343,
+ "grad_norm": 0.948276937007904,
+ "learning_rate": 3.740326373849723e-07,
+ "loss": 0.0097,
+ "step": 33591
+ },
+ {
+ "epoch": 91.53133514986376,
+ "grad_norm": 1.2779974937438965,
+ "learning_rate": 3.737935733359177e-07,
+ "loss": 0.0154,
+ "step": 33592
+ },
+ {
+ "epoch": 91.53405994550408,
+ "grad_norm": 1.4494024515151978,
+ "learning_rate": 3.735545842554489e-07,
+ "loss": 0.1136,
+ "step": 33593
+ },
+ {
+ "epoch": 91.53678474114442,
+ "grad_norm": 0.9177722930908203,
+ "learning_rate": 3.733156701454277e-07,
+ "loss": 0.0078,
+ "step": 33594
+ },
+ {
+ "epoch": 91.53950953678473,
+ "grad_norm": 0.662607729434967,
+ "learning_rate": 3.7307683100771377e-07,
+ "loss": 0.0061,
+ "step": 33595
+ },
+ {
+ "epoch": 91.54223433242507,
+ "grad_norm": 0.9369434118270874,
+ "learning_rate": 3.7283806684416777e-07,
+ "loss": 0.0124,
+ "step": 33596
+ },
+ {
+ "epoch": 91.5449591280654,
+ "grad_norm": 1.5256778001785278,
+ "learning_rate": 3.7259937765664943e-07,
+ "loss": 0.0555,
+ "step": 33597
+ },
+ {
+ "epoch": 91.54768392370572,
+ "grad_norm": 0.9503446221351624,
+ "learning_rate": 3.7236076344701613e-07,
+ "loss": 0.0146,
+ "step": 33598
+ },
+ {
+ "epoch": 91.55040871934605,
+ "grad_norm": 1.5170420408248901,
+ "learning_rate": 3.721222242171285e-07,
+ "loss": 0.0987,
+ "step": 33599
+ },
+ {
+ "epoch": 91.55313351498637,
+ "grad_norm": 1.092363953590393,
+ "learning_rate": 3.7188375996884295e-07,
+ "loss": 0.0084,
+ "step": 33600
+ },
+ {
+ "epoch": 91.5558583106267,
+ "grad_norm": 1.0555522441864014,
+ "learning_rate": 3.7164537070401573e-07,
+ "loss": 0.0111,
+ "step": 33601
+ },
+ {
+ "epoch": 91.55858310626704,
+ "grad_norm": 0.8921197652816772,
+ "learning_rate": 3.7140705642450647e-07,
+ "loss": 0.0087,
+ "step": 33602
+ },
+ {
+ "epoch": 91.56130790190736,
+ "grad_norm": 1.3143115043640137,
+ "learning_rate": 3.711688171321681e-07,
+ "loss": 0.0178,
+ "step": 33603
+ },
+ {
+ "epoch": 91.56403269754769,
+ "grad_norm": 1.6107945442199707,
+ "learning_rate": 3.709306528288581e-07,
+ "loss": 0.1128,
+ "step": 33604
+ },
+ {
+ "epoch": 91.566757493188,
+ "grad_norm": 1.0836811065673828,
+ "learning_rate": 3.7069256351642823e-07,
+ "loss": 0.0089,
+ "step": 33605
+ },
+ {
+ "epoch": 91.56948228882834,
+ "grad_norm": 1.3081395626068115,
+ "learning_rate": 3.7045454919673597e-07,
+ "loss": 0.01,
+ "step": 33606
+ },
+ {
+ "epoch": 91.57220708446866,
+ "grad_norm": 1.404384970664978,
+ "learning_rate": 3.702166098716342e-07,
+ "loss": 0.0197,
+ "step": 33607
+ },
+ {
+ "epoch": 91.57493188010899,
+ "grad_norm": 1.1749217510223389,
+ "learning_rate": 3.6997874554297484e-07,
+ "loss": 0.0075,
+ "step": 33608
+ },
+ {
+ "epoch": 91.57765667574932,
+ "grad_norm": 0.9953048825263977,
+ "learning_rate": 3.697409562126109e-07,
+ "loss": 0.0281,
+ "step": 33609
+ },
+ {
+ "epoch": 91.58038147138964,
+ "grad_norm": 1.7686827182769775,
+ "learning_rate": 3.6950324188239517e-07,
+ "loss": 0.0286,
+ "step": 33610
+ },
+ {
+ "epoch": 91.58310626702998,
+ "grad_norm": 2.15368914604187,
+ "learning_rate": 3.6926560255417745e-07,
+ "loss": 0.0152,
+ "step": 33611
+ },
+ {
+ "epoch": 91.5858310626703,
+ "grad_norm": 0.9534210562705994,
+ "learning_rate": 3.690280382298095e-07,
+ "loss": 0.0687,
+ "step": 33612
+ },
+ {
+ "epoch": 91.58855585831063,
+ "grad_norm": 2.6903445720672607,
+ "learning_rate": 3.68790548911141e-07,
+ "loss": 0.0638,
+ "step": 33613
+ },
+ {
+ "epoch": 91.59128065395096,
+ "grad_norm": 1.9204344749450684,
+ "learning_rate": 3.685531346000215e-07,
+ "loss": 0.0736,
+ "step": 33614
+ },
+ {
+ "epoch": 91.59400544959128,
+ "grad_norm": 1.0449788570404053,
+ "learning_rate": 3.6831579529830076e-07,
+ "loss": 0.031,
+ "step": 33615
+ },
+ {
+ "epoch": 91.59673024523161,
+ "grad_norm": 1.1115394830703735,
+ "learning_rate": 3.680785310078272e-07,
+ "loss": 0.0337,
+ "step": 33616
+ },
+ {
+ "epoch": 91.59945504087193,
+ "grad_norm": 1.6180737018585205,
+ "learning_rate": 3.6784134173044716e-07,
+ "loss": 0.108,
+ "step": 33617
+ },
+ {
+ "epoch": 91.60217983651226,
+ "grad_norm": 0.9056680202484131,
+ "learning_rate": 3.6760422746800806e-07,
+ "loss": 0.0109,
+ "step": 33618
+ },
+ {
+ "epoch": 91.60490463215258,
+ "grad_norm": 0.9498978853225708,
+ "learning_rate": 3.673671882223595e-07,
+ "loss": 0.0091,
+ "step": 33619
+ },
+ {
+ "epoch": 91.60762942779292,
+ "grad_norm": 0.44555845856666565,
+ "learning_rate": 3.671302239953434e-07,
+ "loss": 0.0059,
+ "step": 33620
+ },
+ {
+ "epoch": 91.61035422343325,
+ "grad_norm": 1.4525202512741089,
+ "learning_rate": 3.6689333478880816e-07,
+ "loss": 0.0208,
+ "step": 33621
+ },
+ {
+ "epoch": 91.61307901907357,
+ "grad_norm": 1.0808353424072266,
+ "learning_rate": 3.6665652060459576e-07,
+ "loss": 0.0111,
+ "step": 33622
+ },
+ {
+ "epoch": 91.6158038147139,
+ "grad_norm": 0.8177077770233154,
+ "learning_rate": 3.664197814445547e-07,
+ "loss": 0.0061,
+ "step": 33623
+ },
+ {
+ "epoch": 91.61852861035422,
+ "grad_norm": 1.1995034217834473,
+ "learning_rate": 3.661831173105257e-07,
+ "loss": 0.0244,
+ "step": 33624
+ },
+ {
+ "epoch": 91.62125340599455,
+ "grad_norm": 0.8160623908042908,
+ "learning_rate": 3.6594652820435064e-07,
+ "loss": 0.0122,
+ "step": 33625
+ },
+ {
+ "epoch": 91.62397820163488,
+ "grad_norm": 1.701219081878662,
+ "learning_rate": 3.657100141278758e-07,
+ "loss": 0.0313,
+ "step": 33626
+ },
+ {
+ "epoch": 91.6267029972752,
+ "grad_norm": 0.7860073447227478,
+ "learning_rate": 3.654735750829408e-07,
+ "loss": 0.0061,
+ "step": 33627
+ },
+ {
+ "epoch": 91.62942779291554,
+ "grad_norm": 0.7979462742805481,
+ "learning_rate": 3.652372110713887e-07,
+ "loss": 0.009,
+ "step": 33628
+ },
+ {
+ "epoch": 91.63215258855585,
+ "grad_norm": 1.5036976337432861,
+ "learning_rate": 3.650009220950568e-07,
+ "loss": 0.0093,
+ "step": 33629
+ },
+ {
+ "epoch": 91.63487738419619,
+ "grad_norm": 1.0333654880523682,
+ "learning_rate": 3.647647081557881e-07,
+ "loss": 0.022,
+ "step": 33630
+ },
+ {
+ "epoch": 91.6376021798365,
+ "grad_norm": 2.0305354595184326,
+ "learning_rate": 3.6452856925542235e-07,
+ "loss": 0.1251,
+ "step": 33631
+ },
+ {
+ "epoch": 91.64032697547684,
+ "grad_norm": 1.010170817375183,
+ "learning_rate": 3.642925053957991e-07,
+ "loss": 0.0198,
+ "step": 33632
+ },
+ {
+ "epoch": 91.64305177111717,
+ "grad_norm": 1.349159836769104,
+ "learning_rate": 3.6405651657875464e-07,
+ "loss": 0.0803,
+ "step": 33633
+ },
+ {
+ "epoch": 91.64577656675749,
+ "grad_norm": 0.7518176436424255,
+ "learning_rate": 3.638206028061264e-07,
+ "loss": 0.0067,
+ "step": 33634
+ },
+ {
+ "epoch": 91.64850136239782,
+ "grad_norm": 1.2805321216583252,
+ "learning_rate": 3.635847640797552e-07,
+ "loss": 0.0152,
+ "step": 33635
+ },
+ {
+ "epoch": 91.65122615803814,
+ "grad_norm": 1.0955942869186401,
+ "learning_rate": 3.6334900040147615e-07,
+ "loss": 0.0126,
+ "step": 33636
+ },
+ {
+ "epoch": 91.65395095367847,
+ "grad_norm": 1.1592689752578735,
+ "learning_rate": 3.6311331177312447e-07,
+ "loss": 0.013,
+ "step": 33637
+ },
+ {
+ "epoch": 91.65667574931881,
+ "grad_norm": 1.07011079788208,
+ "learning_rate": 3.628776981965343e-07,
+ "loss": 0.0064,
+ "step": 33638
+ },
+ {
+ "epoch": 91.65940054495913,
+ "grad_norm": 1.476989507675171,
+ "learning_rate": 3.626421596735441e-07,
+ "loss": 0.0633,
+ "step": 33639
+ },
+ {
+ "epoch": 91.66212534059946,
+ "grad_norm": 1.0836615562438965,
+ "learning_rate": 3.624066962059869e-07,
+ "loss": 0.0082,
+ "step": 33640
+ },
+ {
+ "epoch": 91.66485013623978,
+ "grad_norm": 0.9394398927688599,
+ "learning_rate": 3.6217130779569565e-07,
+ "loss": 0.0928,
+ "step": 33641
+ },
+ {
+ "epoch": 91.66757493188011,
+ "grad_norm": 0.936298131942749,
+ "learning_rate": 3.6193599444450335e-07,
+ "loss": 0.0265,
+ "step": 33642
+ },
+ {
+ "epoch": 91.67029972752043,
+ "grad_norm": 2.1657185554504395,
+ "learning_rate": 3.6170075615424407e-07,
+ "loss": 0.0271,
+ "step": 33643
+ },
+ {
+ "epoch": 91.67302452316076,
+ "grad_norm": 0.8522036075592041,
+ "learning_rate": 3.6146559292674967e-07,
+ "loss": 0.0121,
+ "step": 33644
+ },
+ {
+ "epoch": 91.6757493188011,
+ "grad_norm": 1.1212341785430908,
+ "learning_rate": 3.612305047638487e-07,
+ "loss": 0.0238,
+ "step": 33645
+ },
+ {
+ "epoch": 91.67847411444141,
+ "grad_norm": 0.7341930270195007,
+ "learning_rate": 3.6099549166737527e-07,
+ "loss": 0.0079,
+ "step": 33646
+ },
+ {
+ "epoch": 91.68119891008175,
+ "grad_norm": 0.8441574573516846,
+ "learning_rate": 3.6076055363916005e-07,
+ "loss": 0.008,
+ "step": 33647
+ },
+ {
+ "epoch": 91.68392370572207,
+ "grad_norm": 0.9025126099586487,
+ "learning_rate": 3.605256906810306e-07,
+ "loss": 0.0207,
+ "step": 33648
+ },
+ {
+ "epoch": 91.6866485013624,
+ "grad_norm": 1.0722522735595703,
+ "learning_rate": 3.602909027948176e-07,
+ "loss": 0.0201,
+ "step": 33649
+ },
+ {
+ "epoch": 91.68937329700273,
+ "grad_norm": 0.7561191320419312,
+ "learning_rate": 3.6005618998234736e-07,
+ "loss": 0.0073,
+ "step": 33650
+ },
+ {
+ "epoch": 91.69209809264305,
+ "grad_norm": 1.8081878423690796,
+ "learning_rate": 3.598215522454507e-07,
+ "loss": 0.0068,
+ "step": 33651
+ },
+ {
+ "epoch": 91.69482288828338,
+ "grad_norm": 1.1850672960281372,
+ "learning_rate": 3.595869895859527e-07,
+ "loss": 0.0548,
+ "step": 33652
+ },
+ {
+ "epoch": 91.6975476839237,
+ "grad_norm": 0.9707894921302795,
+ "learning_rate": 3.593525020056821e-07,
+ "loss": 0.0223,
+ "step": 33653
+ },
+ {
+ "epoch": 91.70027247956403,
+ "grad_norm": 1.1374857425689697,
+ "learning_rate": 3.5911808950646165e-07,
+ "loss": 0.0084,
+ "step": 33654
+ },
+ {
+ "epoch": 91.70299727520435,
+ "grad_norm": 1.4190822839736938,
+ "learning_rate": 3.588837520901212e-07,
+ "loss": 0.0419,
+ "step": 33655
+ },
+ {
+ "epoch": 91.70572207084469,
+ "grad_norm": 2.172971487045288,
+ "learning_rate": 3.586494897584836e-07,
+ "loss": 0.0413,
+ "step": 33656
+ },
+ {
+ "epoch": 91.70844686648502,
+ "grad_norm": 2.409306764602661,
+ "learning_rate": 3.5841530251337297e-07,
+ "loss": 0.0097,
+ "step": 33657
+ },
+ {
+ "epoch": 91.71117166212534,
+ "grad_norm": 1.8143998384475708,
+ "learning_rate": 3.581811903566135e-07,
+ "loss": 0.0167,
+ "step": 33658
+ },
+ {
+ "epoch": 91.71389645776567,
+ "grad_norm": 0.654679000377655,
+ "learning_rate": 3.5794715329002805e-07,
+ "loss": 0.006,
+ "step": 33659
+ },
+ {
+ "epoch": 91.71662125340599,
+ "grad_norm": 1.1440250873565674,
+ "learning_rate": 3.577131913154419e-07,
+ "loss": 0.0124,
+ "step": 33660
+ },
+ {
+ "epoch": 91.71934604904632,
+ "grad_norm": 1.1876823902130127,
+ "learning_rate": 3.574793044346736e-07,
+ "loss": 0.0132,
+ "step": 33661
+ },
+ {
+ "epoch": 91.72207084468666,
+ "grad_norm": 0.8591668605804443,
+ "learning_rate": 3.57245492649545e-07,
+ "loss": 0.0105,
+ "step": 33662
+ },
+ {
+ "epoch": 91.72479564032697,
+ "grad_norm": 1.8021485805511475,
+ "learning_rate": 3.570117559618802e-07,
+ "loss": 0.0125,
+ "step": 33663
+ },
+ {
+ "epoch": 91.7275204359673,
+ "grad_norm": 0.9828913807868958,
+ "learning_rate": 3.5677809437349554e-07,
+ "loss": 0.0269,
+ "step": 33664
+ },
+ {
+ "epoch": 91.73024523160763,
+ "grad_norm": 1.2146265506744385,
+ "learning_rate": 3.565445078862151e-07,
+ "loss": 0.0241,
+ "step": 33665
+ },
+ {
+ "epoch": 91.73297002724796,
+ "grad_norm": 1.1057407855987549,
+ "learning_rate": 3.5631099650185294e-07,
+ "loss": 0.0769,
+ "step": 33666
+ },
+ {
+ "epoch": 91.73569482288828,
+ "grad_norm": 0.5804660320281982,
+ "learning_rate": 3.5607756022223216e-07,
+ "loss": 0.0078,
+ "step": 33667
+ },
+ {
+ "epoch": 91.73841961852861,
+ "grad_norm": 1.3818237781524658,
+ "learning_rate": 3.55844199049169e-07,
+ "loss": 0.0062,
+ "step": 33668
+ },
+ {
+ "epoch": 91.74114441416894,
+ "grad_norm": 1.2838189601898193,
+ "learning_rate": 3.556109129844809e-07,
+ "loss": 0.0217,
+ "step": 33669
+ },
+ {
+ "epoch": 91.74386920980926,
+ "grad_norm": 1.3427329063415527,
+ "learning_rate": 3.5537770202998314e-07,
+ "loss": 0.0301,
+ "step": 33670
+ },
+ {
+ "epoch": 91.7465940054496,
+ "grad_norm": 1.5605268478393555,
+ "learning_rate": 3.551445661874953e-07,
+ "loss": 0.009,
+ "step": 33671
+ },
+ {
+ "epoch": 91.74931880108991,
+ "grad_norm": 1.4097014665603638,
+ "learning_rate": 3.5491150545883034e-07,
+ "loss": 0.0142,
+ "step": 33672
+ },
+ {
+ "epoch": 91.75204359673025,
+ "grad_norm": 0.7754060626029968,
+ "learning_rate": 3.546785198458047e-07,
+ "loss": 0.008,
+ "step": 33673
+ },
+ {
+ "epoch": 91.75476839237058,
+ "grad_norm": 0.9917222857475281,
+ "learning_rate": 3.544456093502313e-07,
+ "loss": 0.0194,
+ "step": 33674
+ },
+ {
+ "epoch": 91.7574931880109,
+ "grad_norm": 0.7817190289497375,
+ "learning_rate": 3.542127739739254e-07,
+ "loss": 0.1306,
+ "step": 33675
+ },
+ {
+ "epoch": 91.76021798365123,
+ "grad_norm": 0.8521759510040283,
+ "learning_rate": 3.5398001371870104e-07,
+ "loss": 0.1305,
+ "step": 33676
+ },
+ {
+ "epoch": 91.76294277929155,
+ "grad_norm": 0.9198516607284546,
+ "learning_rate": 3.537473285863691e-07,
+ "loss": 0.0102,
+ "step": 33677
+ },
+ {
+ "epoch": 91.76566757493188,
+ "grad_norm": 0.5687785744667053,
+ "learning_rate": 3.535147185787413e-07,
+ "loss": 0.0062,
+ "step": 33678
+ },
+ {
+ "epoch": 91.7683923705722,
+ "grad_norm": 0.4819241464138031,
+ "learning_rate": 3.532821836976319e-07,
+ "loss": 0.0051,
+ "step": 33679
+ },
+ {
+ "epoch": 91.77111716621253,
+ "grad_norm": 1.0541948080062866,
+ "learning_rate": 3.530497239448505e-07,
+ "loss": 0.0132,
+ "step": 33680
+ },
+ {
+ "epoch": 91.77384196185287,
+ "grad_norm": 1.320935606956482,
+ "learning_rate": 3.5281733932220566e-07,
+ "loss": 0.0243,
+ "step": 33681
+ },
+ {
+ "epoch": 91.77656675749319,
+ "grad_norm": 1.2654852867126465,
+ "learning_rate": 3.525850298315103e-07,
+ "loss": 0.0238,
+ "step": 33682
+ },
+ {
+ "epoch": 91.77929155313352,
+ "grad_norm": 1.4136899709701538,
+ "learning_rate": 3.523527954745709e-07,
+ "loss": 0.036,
+ "step": 33683
+ },
+ {
+ "epoch": 91.78201634877384,
+ "grad_norm": 1.6132502555847168,
+ "learning_rate": 3.5212063625319925e-07,
+ "loss": 0.051,
+ "step": 33684
+ },
+ {
+ "epoch": 91.78474114441417,
+ "grad_norm": 1.3203603029251099,
+ "learning_rate": 3.518885521692006e-07,
+ "loss": 0.0436,
+ "step": 33685
+ },
+ {
+ "epoch": 91.7874659400545,
+ "grad_norm": 0.6300475597381592,
+ "learning_rate": 3.516565432243824e-07,
+ "loss": 0.0082,
+ "step": 33686
+ },
+ {
+ "epoch": 91.79019073569482,
+ "grad_norm": 0.7689418196678162,
+ "learning_rate": 3.5142460942055424e-07,
+ "loss": 0.0072,
+ "step": 33687
+ },
+ {
+ "epoch": 91.79291553133515,
+ "grad_norm": 0.7229488492012024,
+ "learning_rate": 3.511927507595203e-07,
+ "loss": 0.0059,
+ "step": 33688
+ },
+ {
+ "epoch": 91.79564032697547,
+ "grad_norm": 0.8578216433525085,
+ "learning_rate": 3.5096096724308694e-07,
+ "loss": 0.0591,
+ "step": 33689
+ },
+ {
+ "epoch": 91.7983651226158,
+ "grad_norm": 1.2056118249893188,
+ "learning_rate": 3.507292588730582e-07,
+ "loss": 0.0139,
+ "step": 33690
+ },
+ {
+ "epoch": 91.80108991825612,
+ "grad_norm": 1.218881368637085,
+ "learning_rate": 3.5049762565123936e-07,
+ "loss": 0.0309,
+ "step": 33691
+ },
+ {
+ "epoch": 91.80381471389646,
+ "grad_norm": 1.1752355098724365,
+ "learning_rate": 3.502660675794356e-07,
+ "loss": 0.012,
+ "step": 33692
+ },
+ {
+ "epoch": 91.80653950953679,
+ "grad_norm": 0.7736294865608215,
+ "learning_rate": 3.5003458465944884e-07,
+ "loss": 0.0089,
+ "step": 33693
+ },
+ {
+ "epoch": 91.80926430517711,
+ "grad_norm": 0.7485520243644714,
+ "learning_rate": 3.4980317689308094e-07,
+ "loss": 0.0173,
+ "step": 33694
+ },
+ {
+ "epoch": 91.81198910081744,
+ "grad_norm": 0.4236682653427124,
+ "learning_rate": 3.495718442821361e-07,
+ "loss": 0.0059,
+ "step": 33695
+ },
+ {
+ "epoch": 91.81471389645776,
+ "grad_norm": 1.258766531944275,
+ "learning_rate": 3.493405868284161e-07,
+ "loss": 0.019,
+ "step": 33696
+ },
+ {
+ "epoch": 91.8174386920981,
+ "grad_norm": 1.0589830875396729,
+ "learning_rate": 3.491094045337196e-07,
+ "loss": 0.0167,
+ "step": 33697
+ },
+ {
+ "epoch": 91.82016348773843,
+ "grad_norm": 1.2251675128936768,
+ "learning_rate": 3.4887829739984836e-07,
+ "loss": 0.0199,
+ "step": 33698
+ },
+ {
+ "epoch": 91.82288828337875,
+ "grad_norm": 0.8820494413375854,
+ "learning_rate": 3.486472654286022e-07,
+ "loss": 0.0065,
+ "step": 33699
+ },
+ {
+ "epoch": 91.82561307901908,
+ "grad_norm": 0.798437774181366,
+ "learning_rate": 3.484163086217818e-07,
+ "loss": 0.0112,
+ "step": 33700
+ },
+ {
+ "epoch": 91.8283378746594,
+ "grad_norm": 0.8551892042160034,
+ "learning_rate": 3.481854269811835e-07,
+ "loss": 0.0097,
+ "step": 33701
+ },
+ {
+ "epoch": 91.83106267029973,
+ "grad_norm": 1.0640287399291992,
+ "learning_rate": 3.4795462050860704e-07,
+ "loss": 0.0181,
+ "step": 33702
+ },
+ {
+ "epoch": 91.83378746594005,
+ "grad_norm": 0.9016211628913879,
+ "learning_rate": 3.4772388920584877e-07,
+ "loss": 0.0118,
+ "step": 33703
+ },
+ {
+ "epoch": 91.83651226158038,
+ "grad_norm": 0.5478224754333496,
+ "learning_rate": 3.4749323307470717e-07,
+ "loss": 0.0048,
+ "step": 33704
+ },
+ {
+ "epoch": 91.83923705722071,
+ "grad_norm": 0.5448170900344849,
+ "learning_rate": 3.472626521169775e-07,
+ "loss": 0.005,
+ "step": 33705
+ },
+ {
+ "epoch": 91.84196185286103,
+ "grad_norm": 0.7888721823692322,
+ "learning_rate": 3.4703214633445504e-07,
+ "loss": 0.0071,
+ "step": 33706
+ },
+ {
+ "epoch": 91.84468664850137,
+ "grad_norm": 2.0440728664398193,
+ "learning_rate": 3.4680171572893494e-07,
+ "loss": 0.0373,
+ "step": 33707
+ },
+ {
+ "epoch": 91.84741144414168,
+ "grad_norm": 0.6920275688171387,
+ "learning_rate": 3.465713603022136e-07,
+ "loss": 0.0063,
+ "step": 33708
+ },
+ {
+ "epoch": 91.85013623978202,
+ "grad_norm": 0.7758582830429077,
+ "learning_rate": 3.463410800560829e-07,
+ "loss": 0.0109,
+ "step": 33709
+ },
+ {
+ "epoch": 91.85286103542235,
+ "grad_norm": 0.37378478050231934,
+ "learning_rate": 3.4611087499233696e-07,
+ "loss": 0.0027,
+ "step": 33710
+ },
+ {
+ "epoch": 91.85558583106267,
+ "grad_norm": 1.472303867340088,
+ "learning_rate": 3.458807451127699e-07,
+ "loss": 0.083,
+ "step": 33711
+ },
+ {
+ "epoch": 91.858310626703,
+ "grad_norm": 0.9583253264427185,
+ "learning_rate": 3.4565069041917256e-07,
+ "loss": 0.0109,
+ "step": 33712
+ },
+ {
+ "epoch": 91.86103542234332,
+ "grad_norm": 1.011045217514038,
+ "learning_rate": 3.4542071091333674e-07,
+ "loss": 0.0393,
+ "step": 33713
+ },
+ {
+ "epoch": 91.86376021798365,
+ "grad_norm": 0.786953866481781,
+ "learning_rate": 3.451908065970533e-07,
+ "loss": 0.0045,
+ "step": 33714
+ },
+ {
+ "epoch": 91.86648501362397,
+ "grad_norm": 0.9548712372779846,
+ "learning_rate": 3.44960977472113e-07,
+ "loss": 0.0086,
+ "step": 33715
+ },
+ {
+ "epoch": 91.8692098092643,
+ "grad_norm": 1.4418838024139404,
+ "learning_rate": 3.4473122354030663e-07,
+ "loss": 0.0523,
+ "step": 33716
+ },
+ {
+ "epoch": 91.87193460490464,
+ "grad_norm": 0.9688159227371216,
+ "learning_rate": 3.445015448034228e-07,
+ "loss": 0.0141,
+ "step": 33717
+ },
+ {
+ "epoch": 91.87465940054496,
+ "grad_norm": 1.1900972127914429,
+ "learning_rate": 3.4427194126325006e-07,
+ "loss": 0.0106,
+ "step": 33718
+ },
+ {
+ "epoch": 91.87738419618529,
+ "grad_norm": 1.3135223388671875,
+ "learning_rate": 3.4404241292157583e-07,
+ "loss": 0.0199,
+ "step": 33719
+ },
+ {
+ "epoch": 91.88010899182561,
+ "grad_norm": 1.2103418111801147,
+ "learning_rate": 3.438129597801898e-07,
+ "loss": 0.0117,
+ "step": 33720
+ },
+ {
+ "epoch": 91.88283378746594,
+ "grad_norm": 0.49731212854385376,
+ "learning_rate": 3.4358358184087727e-07,
+ "loss": 0.0042,
+ "step": 33721
+ },
+ {
+ "epoch": 91.88555858310627,
+ "grad_norm": 0.8556985855102539,
+ "learning_rate": 3.433542791054245e-07,
+ "loss": 0.013,
+ "step": 33722
+ },
+ {
+ "epoch": 91.88828337874659,
+ "grad_norm": 1.5518488883972168,
+ "learning_rate": 3.4312505157561794e-07,
+ "loss": 0.0269,
+ "step": 33723
+ },
+ {
+ "epoch": 91.89100817438693,
+ "grad_norm": 1.093356728553772,
+ "learning_rate": 3.428958992532427e-07,
+ "loss": 0.0115,
+ "step": 33724
+ },
+ {
+ "epoch": 91.89373297002724,
+ "grad_norm": 1.2238081693649292,
+ "learning_rate": 3.4266682214008416e-07,
+ "loss": 0.0197,
+ "step": 33725
+ },
+ {
+ "epoch": 91.89645776566758,
+ "grad_norm": 1.1182043552398682,
+ "learning_rate": 3.424378202379253e-07,
+ "loss": 0.0884,
+ "step": 33726
+ },
+ {
+ "epoch": 91.8991825613079,
+ "grad_norm": 0.8178949952125549,
+ "learning_rate": 3.422088935485479e-07,
+ "loss": 0.0145,
+ "step": 33727
+ },
+ {
+ "epoch": 91.90190735694823,
+ "grad_norm": 0.47963136434555054,
+ "learning_rate": 3.419800420737396e-07,
+ "loss": 0.005,
+ "step": 33728
+ },
+ {
+ "epoch": 91.90463215258856,
+ "grad_norm": 0.8027451634407043,
+ "learning_rate": 3.4175126581527883e-07,
+ "loss": 0.0191,
+ "step": 33729
+ },
+ {
+ "epoch": 91.90735694822888,
+ "grad_norm": 0.9669674038887024,
+ "learning_rate": 3.4152256477494874e-07,
+ "loss": 0.0126,
+ "step": 33730
+ },
+ {
+ "epoch": 91.91008174386921,
+ "grad_norm": 0.4880145192146301,
+ "learning_rate": 3.412939389545289e-07,
+ "loss": 0.0046,
+ "step": 33731
+ },
+ {
+ "epoch": 91.91280653950953,
+ "grad_norm": 0.8657566905021667,
+ "learning_rate": 3.410653883558002e-07,
+ "loss": 0.0155,
+ "step": 33732
+ },
+ {
+ "epoch": 91.91553133514986,
+ "grad_norm": 1.481467366218567,
+ "learning_rate": 3.4083691298054556e-07,
+ "loss": 0.0133,
+ "step": 33733
+ },
+ {
+ "epoch": 91.9182561307902,
+ "grad_norm": 1.5545343160629272,
+ "learning_rate": 3.406085128305414e-07,
+ "loss": 0.0792,
+ "step": 33734
+ },
+ {
+ "epoch": 91.92098092643052,
+ "grad_norm": 1.4520232677459717,
+ "learning_rate": 3.4038018790756634e-07,
+ "loss": 0.0058,
+ "step": 33735
+ },
+ {
+ "epoch": 91.92370572207085,
+ "grad_norm": 0.8401662111282349,
+ "learning_rate": 3.401519382134011e-07,
+ "loss": 0.0059,
+ "step": 33736
+ },
+ {
+ "epoch": 91.92643051771117,
+ "grad_norm": 0.4064794182777405,
+ "learning_rate": 3.3992376374982097e-07,
+ "loss": 0.0047,
+ "step": 33737
+ },
+ {
+ "epoch": 91.9291553133515,
+ "grad_norm": 1.1017109155654907,
+ "learning_rate": 3.396956645186045e-07,
+ "loss": 0.0607,
+ "step": 33738
+ },
+ {
+ "epoch": 91.93188010899182,
+ "grad_norm": 0.7714987397193909,
+ "learning_rate": 3.3946764052152583e-07,
+ "loss": 0.0094,
+ "step": 33739
+ },
+ {
+ "epoch": 91.93460490463215,
+ "grad_norm": 1.1254154443740845,
+ "learning_rate": 3.392396917603624e-07,
+ "loss": 0.0419,
+ "step": 33740
+ },
+ {
+ "epoch": 91.93732970027249,
+ "grad_norm": 1.4996196031570435,
+ "learning_rate": 3.3901181823689065e-07,
+ "loss": 0.0447,
+ "step": 33741
+ },
+ {
+ "epoch": 91.9400544959128,
+ "grad_norm": 1.450362205505371,
+ "learning_rate": 3.3878401995288357e-07,
+ "loss": 0.022,
+ "step": 33742
+ },
+ {
+ "epoch": 91.94277929155314,
+ "grad_norm": 2.494549036026001,
+ "learning_rate": 3.385562969101142e-07,
+ "loss": 0.0238,
+ "step": 33743
+ },
+ {
+ "epoch": 91.94550408719346,
+ "grad_norm": 1.1640267372131348,
+ "learning_rate": 3.3832864911035877e-07,
+ "loss": 0.0187,
+ "step": 33744
+ },
+ {
+ "epoch": 91.94822888283379,
+ "grad_norm": 1.3125386238098145,
+ "learning_rate": 3.3810107655538826e-07,
+ "loss": 0.0118,
+ "step": 33745
+ },
+ {
+ "epoch": 91.95095367847412,
+ "grad_norm": 1.2716859579086304,
+ "learning_rate": 3.378735792469756e-07,
+ "loss": 0.0641,
+ "step": 33746
+ },
+ {
+ "epoch": 91.95367847411444,
+ "grad_norm": 0.5508002638816833,
+ "learning_rate": 3.3764615718689273e-07,
+ "loss": 0.0067,
+ "step": 33747
+ },
+ {
+ "epoch": 91.95640326975477,
+ "grad_norm": 1.218878984451294,
+ "learning_rate": 3.374188103769105e-07,
+ "loss": 0.0212,
+ "step": 33748
+ },
+ {
+ "epoch": 91.95912806539509,
+ "grad_norm": 0.6011059880256653,
+ "learning_rate": 3.3719153881879963e-07,
+ "loss": 0.0045,
+ "step": 33749
+ },
+ {
+ "epoch": 91.96185286103542,
+ "grad_norm": 1.0385249853134155,
+ "learning_rate": 3.3696434251432984e-07,
+ "loss": 0.013,
+ "step": 33750
+ },
+ {
+ "epoch": 91.96457765667574,
+ "grad_norm": 0.5629593729972839,
+ "learning_rate": 3.3673722146526976e-07,
+ "loss": 0.0049,
+ "step": 33751
+ },
+ {
+ "epoch": 91.96730245231608,
+ "grad_norm": 1.3833887577056885,
+ "learning_rate": 3.365101756733913e-07,
+ "loss": 0.0859,
+ "step": 33752
+ },
+ {
+ "epoch": 91.97002724795641,
+ "grad_norm": 0.42499950528144836,
+ "learning_rate": 3.3628320514045963e-07,
+ "loss": 0.0035,
+ "step": 33753
+ },
+ {
+ "epoch": 91.97275204359673,
+ "grad_norm": 0.6881384253501892,
+ "learning_rate": 3.3605630986824346e-07,
+ "loss": 0.0078,
+ "step": 33754
+ },
+ {
+ "epoch": 91.97547683923706,
+ "grad_norm": 1.0689847469329834,
+ "learning_rate": 3.3582948985850904e-07,
+ "loss": 0.026,
+ "step": 33755
+ },
+ {
+ "epoch": 91.97820163487738,
+ "grad_norm": 0.9165669083595276,
+ "learning_rate": 3.35602745113025e-07,
+ "loss": 0.018,
+ "step": 33756
+ },
+ {
+ "epoch": 91.98092643051771,
+ "grad_norm": 1.2907904386520386,
+ "learning_rate": 3.3537607563355447e-07,
+ "loss": 0.0592,
+ "step": 33757
+ },
+ {
+ "epoch": 91.98365122615803,
+ "grad_norm": 1.1743125915527344,
+ "learning_rate": 3.3514948142186476e-07,
+ "loss": 0.0235,
+ "step": 33758
+ },
+ {
+ "epoch": 91.98637602179836,
+ "grad_norm": 1.4441217184066772,
+ "learning_rate": 3.34922962479719e-07,
+ "loss": 0.0081,
+ "step": 33759
+ },
+ {
+ "epoch": 91.9891008174387,
+ "grad_norm": 1.8144505023956299,
+ "learning_rate": 3.3469651880888245e-07,
+ "loss": 0.0302,
+ "step": 33760
+ },
+ {
+ "epoch": 91.99182561307902,
+ "grad_norm": 0.83965003490448,
+ "learning_rate": 3.344701504111192e-07,
+ "loss": 0.0079,
+ "step": 33761
+ },
+ {
+ "epoch": 91.99455040871935,
+ "grad_norm": 0.9813574552536011,
+ "learning_rate": 3.342438572881912e-07,
+ "loss": 0.01,
+ "step": 33762
+ },
+ {
+ "epoch": 91.99727520435967,
+ "grad_norm": 1.154587984085083,
+ "learning_rate": 3.3401763944185927e-07,
+ "loss": 0.0106,
+ "step": 33763
+ },
+ {
+ "epoch": 92.0,
+ "grad_norm": 0.6444743275642395,
+ "learning_rate": 3.3379149687388866e-07,
+ "loss": 0.0063,
+ "step": 33764
+ },
+ {
+ "epoch": 92.00272479564033,
+ "grad_norm": 0.6209907531738281,
+ "learning_rate": 3.33565429586038e-07,
+ "loss": 0.0098,
+ "step": 33765
+ },
+ {
+ "epoch": 92.00544959128065,
+ "grad_norm": 0.5015397667884827,
+ "learning_rate": 3.3333943758006914e-07,
+ "loss": 0.005,
+ "step": 33766
+ },
+ {
+ "epoch": 92.00817438692098,
+ "grad_norm": 0.9352481365203857,
+ "learning_rate": 3.331135208577407e-07,
+ "loss": 0.0057,
+ "step": 33767
+ },
+ {
+ "epoch": 92.0108991825613,
+ "grad_norm": 0.92298424243927,
+ "learning_rate": 3.3288767942081244e-07,
+ "loss": 0.0145,
+ "step": 33768
+ },
+ {
+ "epoch": 92.01362397820164,
+ "grad_norm": 1.3550920486450195,
+ "learning_rate": 3.326619132710451e-07,
+ "loss": 0.0419,
+ "step": 33769
+ },
+ {
+ "epoch": 92.01634877384195,
+ "grad_norm": 1.207899808883667,
+ "learning_rate": 3.3243622241019623e-07,
+ "loss": 0.0099,
+ "step": 33770
+ },
+ {
+ "epoch": 92.01907356948229,
+ "grad_norm": 0.6840733289718628,
+ "learning_rate": 3.3221060684002217e-07,
+ "loss": 0.0082,
+ "step": 33771
+ },
+ {
+ "epoch": 92.02179836512262,
+ "grad_norm": 1.630222201347351,
+ "learning_rate": 3.3198506656228037e-07,
+ "loss": 0.0163,
+ "step": 33772
+ },
+ {
+ "epoch": 92.02452316076294,
+ "grad_norm": 1.6675463914871216,
+ "learning_rate": 3.317596015787283e-07,
+ "loss": 0.0945,
+ "step": 33773
+ },
+ {
+ "epoch": 92.02724795640327,
+ "grad_norm": 1.897666335105896,
+ "learning_rate": 3.3153421189112136e-07,
+ "loss": 0.0056,
+ "step": 33774
+ },
+ {
+ "epoch": 92.02997275204359,
+ "grad_norm": 0.978567361831665,
+ "learning_rate": 3.313088975012135e-07,
+ "loss": 0.0166,
+ "step": 33775
+ },
+ {
+ "epoch": 92.03269754768392,
+ "grad_norm": 1.4647853374481201,
+ "learning_rate": 3.310836584107624e-07,
+ "loss": 0.0142,
+ "step": 33776
+ },
+ {
+ "epoch": 92.03542234332426,
+ "grad_norm": 1.0876308679580688,
+ "learning_rate": 3.308584946215199e-07,
+ "loss": 0.0719,
+ "step": 33777
+ },
+ {
+ "epoch": 92.03814713896458,
+ "grad_norm": 0.9052349925041199,
+ "learning_rate": 3.306334061352401e-07,
+ "loss": 0.0053,
+ "step": 33778
+ },
+ {
+ "epoch": 92.04087193460491,
+ "grad_norm": 1.154747724533081,
+ "learning_rate": 3.304083929536761e-07,
+ "loss": 0.0198,
+ "step": 33779
+ },
+ {
+ "epoch": 92.04359673024523,
+ "grad_norm": 1.311823844909668,
+ "learning_rate": 3.3018345507858094e-07,
+ "loss": 0.0151,
+ "step": 33780
+ },
+ {
+ "epoch": 92.04632152588556,
+ "grad_norm": 0.8830193281173706,
+ "learning_rate": 3.2995859251170547e-07,
+ "loss": 0.0081,
+ "step": 33781
+ },
+ {
+ "epoch": 92.04904632152588,
+ "grad_norm": 1.6688169240951538,
+ "learning_rate": 3.297338052548016e-07,
+ "loss": 0.0057,
+ "step": 33782
+ },
+ {
+ "epoch": 92.05177111716621,
+ "grad_norm": 1.1398242712020874,
+ "learning_rate": 3.295090933096179e-07,
+ "loss": 0.0109,
+ "step": 33783
+ },
+ {
+ "epoch": 92.05449591280654,
+ "grad_norm": 1.014640212059021,
+ "learning_rate": 3.292844566779063e-07,
+ "loss": 0.0061,
+ "step": 33784
+ },
+ {
+ "epoch": 92.05722070844686,
+ "grad_norm": 0.9392091631889343,
+ "learning_rate": 3.290598953614177e-07,
+ "loss": 0.0071,
+ "step": 33785
+ },
+ {
+ "epoch": 92.0599455040872,
+ "grad_norm": 1.3392583131790161,
+ "learning_rate": 3.288354093618995e-07,
+ "loss": 0.013,
+ "step": 33786
+ },
+ {
+ "epoch": 92.06267029972751,
+ "grad_norm": 0.7962372899055481,
+ "learning_rate": 3.2861099868109924e-07,
+ "loss": 0.0148,
+ "step": 33787
+ },
+ {
+ "epoch": 92.06539509536785,
+ "grad_norm": 1.496867060661316,
+ "learning_rate": 3.283866633207644e-07,
+ "loss": 0.0201,
+ "step": 33788
+ },
+ {
+ "epoch": 92.06811989100818,
+ "grad_norm": 1.4594770669937134,
+ "learning_rate": 3.281624032826447e-07,
+ "loss": 0.0816,
+ "step": 33789
+ },
+ {
+ "epoch": 92.0708446866485,
+ "grad_norm": 1.2912046909332275,
+ "learning_rate": 3.279382185684843e-07,
+ "loss": 0.0483,
+ "step": 33790
+ },
+ {
+ "epoch": 92.07356948228883,
+ "grad_norm": 1.4418247938156128,
+ "learning_rate": 3.2771410918002954e-07,
+ "loss": 0.1094,
+ "step": 33791
+ },
+ {
+ "epoch": 92.07629427792915,
+ "grad_norm": 19.147193908691406,
+ "learning_rate": 3.2749007511902463e-07,
+ "loss": 0.0597,
+ "step": 33792
+ },
+ {
+ "epoch": 92.07901907356948,
+ "grad_norm": 0.9866797924041748,
+ "learning_rate": 3.272661163872182e-07,
+ "loss": 0.0212,
+ "step": 33793
+ },
+ {
+ "epoch": 92.0817438692098,
+ "grad_norm": 1.0206027030944824,
+ "learning_rate": 3.2704223298635096e-07,
+ "loss": 0.0101,
+ "step": 33794
+ },
+ {
+ "epoch": 92.08446866485014,
+ "grad_norm": 0.7577354907989502,
+ "learning_rate": 3.2681842491816716e-07,
+ "loss": 0.0063,
+ "step": 33795
+ },
+ {
+ "epoch": 92.08719346049047,
+ "grad_norm": 1.4983100891113281,
+ "learning_rate": 3.2659469218440986e-07,
+ "loss": 0.0123,
+ "step": 33796
+ },
+ {
+ "epoch": 92.08991825613079,
+ "grad_norm": 1.2969175577163696,
+ "learning_rate": 3.263710347868221e-07,
+ "loss": 0.0161,
+ "step": 33797
+ },
+ {
+ "epoch": 92.09264305177112,
+ "grad_norm": 1.7584086656570435,
+ "learning_rate": 3.2614745272714577e-07,
+ "loss": 0.1596,
+ "step": 33798
+ },
+ {
+ "epoch": 92.09536784741144,
+ "grad_norm": 1.0752599239349365,
+ "learning_rate": 3.2592394600712065e-07,
+ "loss": 0.02,
+ "step": 33799
+ },
+ {
+ "epoch": 92.09809264305177,
+ "grad_norm": 1.8060872554779053,
+ "learning_rate": 3.2570051462848973e-07,
+ "loss": 0.0426,
+ "step": 33800
+ },
+ {
+ "epoch": 92.1008174386921,
+ "grad_norm": 1.054193139076233,
+ "learning_rate": 3.2547715859299055e-07,
+ "loss": 0.0176,
+ "step": 33801
+ },
+ {
+ "epoch": 92.10354223433242,
+ "grad_norm": 1.4565768241882324,
+ "learning_rate": 3.2525387790236507e-07,
+ "loss": 0.1138,
+ "step": 33802
+ },
+ {
+ "epoch": 92.10626702997276,
+ "grad_norm": 1.8134818077087402,
+ "learning_rate": 3.250306725583507e-07,
+ "loss": 0.0322,
+ "step": 33803
+ },
+ {
+ "epoch": 92.10899182561307,
+ "grad_norm": 1.0628961324691772,
+ "learning_rate": 3.248075425626851e-07,
+ "loss": 0.0199,
+ "step": 33804
+ },
+ {
+ "epoch": 92.11171662125341,
+ "grad_norm": 0.7619818449020386,
+ "learning_rate": 3.2458448791710786e-07,
+ "loss": 0.0128,
+ "step": 33805
+ },
+ {
+ "epoch": 92.11444141689373,
+ "grad_norm": 1.7229937314987183,
+ "learning_rate": 3.2436150862335536e-07,
+ "loss": 0.0519,
+ "step": 33806
+ },
+ {
+ "epoch": 92.11716621253406,
+ "grad_norm": 0.790875256061554,
+ "learning_rate": 3.2413860468316517e-07,
+ "loss": 0.0061,
+ "step": 33807
+ },
+ {
+ "epoch": 92.11989100817439,
+ "grad_norm": 1.7854721546173096,
+ "learning_rate": 3.239157760982692e-07,
+ "loss": 0.0114,
+ "step": 33808
+ },
+ {
+ "epoch": 92.12261580381471,
+ "grad_norm": 0.7216194868087769,
+ "learning_rate": 3.2369302287040826e-07,
+ "loss": 0.0038,
+ "step": 33809
+ },
+ {
+ "epoch": 92.12534059945504,
+ "grad_norm": 0.7238480448722839,
+ "learning_rate": 3.234703450013132e-07,
+ "loss": 0.0062,
+ "step": 33810
+ },
+ {
+ "epoch": 92.12806539509536,
+ "grad_norm": 0.6191335916519165,
+ "learning_rate": 3.232477424927205e-07,
+ "loss": 0.0051,
+ "step": 33811
+ },
+ {
+ "epoch": 92.1307901907357,
+ "grad_norm": 1.5931757688522339,
+ "learning_rate": 3.23025215346362e-07,
+ "loss": 0.0436,
+ "step": 33812
+ },
+ {
+ "epoch": 92.13351498637603,
+ "grad_norm": 0.6730411052703857,
+ "learning_rate": 3.22802763563973e-07,
+ "loss": 0.0052,
+ "step": 33813
+ },
+ {
+ "epoch": 92.13623978201635,
+ "grad_norm": 0.7010020017623901,
+ "learning_rate": 3.2258038714728436e-07,
+ "loss": 0.0073,
+ "step": 33814
+ },
+ {
+ "epoch": 92.13896457765668,
+ "grad_norm": 1.234999656677246,
+ "learning_rate": 3.22358086098028e-07,
+ "loss": 0.018,
+ "step": 33815
+ },
+ {
+ "epoch": 92.141689373297,
+ "grad_norm": 0.7892544269561768,
+ "learning_rate": 3.2213586041793364e-07,
+ "loss": 0.011,
+ "step": 33816
+ },
+ {
+ "epoch": 92.14441416893733,
+ "grad_norm": 1.2499886751174927,
+ "learning_rate": 3.219137101087355e-07,
+ "loss": 0.0443,
+ "step": 33817
+ },
+ {
+ "epoch": 92.14713896457765,
+ "grad_norm": 1.0164977312088013,
+ "learning_rate": 3.2169163517216217e-07,
+ "loss": 0.038,
+ "step": 33818
+ },
+ {
+ "epoch": 92.14986376021798,
+ "grad_norm": 0.8888025879859924,
+ "learning_rate": 3.2146963560994226e-07,
+ "loss": 0.0104,
+ "step": 33819
+ },
+ {
+ "epoch": 92.15258855585832,
+ "grad_norm": 1.1394600868225098,
+ "learning_rate": 3.212477114238044e-07,
+ "loss": 0.0446,
+ "step": 33820
+ },
+ {
+ "epoch": 92.15531335149863,
+ "grad_norm": 0.5795865058898926,
+ "learning_rate": 3.210258626154794e-07,
+ "loss": 0.0102,
+ "step": 33821
+ },
+ {
+ "epoch": 92.15803814713897,
+ "grad_norm": 1.4301459789276123,
+ "learning_rate": 3.208040891866926e-07,
+ "loss": 0.0326,
+ "step": 33822
+ },
+ {
+ "epoch": 92.16076294277929,
+ "grad_norm": 0.881291389465332,
+ "learning_rate": 3.2058239113917256e-07,
+ "loss": 0.0067,
+ "step": 33823
+ },
+ {
+ "epoch": 92.16348773841962,
+ "grad_norm": 1.8473505973815918,
+ "learning_rate": 3.2036076847464457e-07,
+ "loss": 0.0157,
+ "step": 33824
+ },
+ {
+ "epoch": 92.16621253405995,
+ "grad_norm": 0.9793704152107239,
+ "learning_rate": 3.201392211948362e-07,
+ "loss": 0.0723,
+ "step": 33825
+ },
+ {
+ "epoch": 92.16893732970027,
+ "grad_norm": 0.5820945501327515,
+ "learning_rate": 3.199177493014716e-07,
+ "loss": 0.0046,
+ "step": 33826
+ },
+ {
+ "epoch": 92.1716621253406,
+ "grad_norm": 1.0331205129623413,
+ "learning_rate": 3.196963527962771e-07,
+ "loss": 0.0155,
+ "step": 33827
+ },
+ {
+ "epoch": 92.17438692098092,
+ "grad_norm": 1.1756877899169922,
+ "learning_rate": 3.194750316809736e-07,
+ "loss": 0.0168,
+ "step": 33828
+ },
+ {
+ "epoch": 92.17711171662125,
+ "grad_norm": 1.3448628187179565,
+ "learning_rate": 3.192537859572886e-07,
+ "loss": 0.0155,
+ "step": 33829
+ },
+ {
+ "epoch": 92.17983651226157,
+ "grad_norm": 0.6684107184410095,
+ "learning_rate": 3.190326156269441e-07,
+ "loss": 0.0055,
+ "step": 33830
+ },
+ {
+ "epoch": 92.1825613079019,
+ "grad_norm": 0.729377806186676,
+ "learning_rate": 3.188115206916609e-07,
+ "loss": 0.0082,
+ "step": 33831
+ },
+ {
+ "epoch": 92.18528610354224,
+ "grad_norm": 1.1843340396881104,
+ "learning_rate": 3.1859050115316205e-07,
+ "loss": 0.0333,
+ "step": 33832
+ },
+ {
+ "epoch": 92.18801089918256,
+ "grad_norm": 1.1218959093093872,
+ "learning_rate": 3.183695570131695e-07,
+ "loss": 0.0082,
+ "step": 33833
+ },
+ {
+ "epoch": 92.19073569482289,
+ "grad_norm": 1.475399136543274,
+ "learning_rate": 3.1814868827340194e-07,
+ "loss": 0.0118,
+ "step": 33834
+ },
+ {
+ "epoch": 92.19346049046321,
+ "grad_norm": 1.046160101890564,
+ "learning_rate": 3.179278949355824e-07,
+ "loss": 0.0283,
+ "step": 33835
+ },
+ {
+ "epoch": 92.19618528610354,
+ "grad_norm": 1.1685609817504883,
+ "learning_rate": 3.177071770014273e-07,
+ "loss": 0.0202,
+ "step": 33836
+ },
+ {
+ "epoch": 92.19891008174388,
+ "grad_norm": 1.6927623748779297,
+ "learning_rate": 3.1748653447265634e-07,
+ "loss": 0.1114,
+ "step": 33837
+ },
+ {
+ "epoch": 92.2016348773842,
+ "grad_norm": 0.7218412756919861,
+ "learning_rate": 3.1726596735099035e-07,
+ "loss": 0.0079,
+ "step": 33838
+ },
+ {
+ "epoch": 92.20435967302453,
+ "grad_norm": 1.172763466835022,
+ "learning_rate": 3.1704547563814583e-07,
+ "loss": 0.0191,
+ "step": 33839
+ },
+ {
+ "epoch": 92.20708446866485,
+ "grad_norm": 0.8740037083625793,
+ "learning_rate": 3.1682505933583796e-07,
+ "loss": 0.014,
+ "step": 33840
+ },
+ {
+ "epoch": 92.20980926430518,
+ "grad_norm": 0.49438726902008057,
+ "learning_rate": 3.166047184457854e-07,
+ "loss": 0.0035,
+ "step": 33841
+ },
+ {
+ "epoch": 92.2125340599455,
+ "grad_norm": 1.2492921352386475,
+ "learning_rate": 3.1638445296970463e-07,
+ "loss": 0.0153,
+ "step": 33842
+ },
+ {
+ "epoch": 92.21525885558583,
+ "grad_norm": 2.10772442817688,
+ "learning_rate": 3.1616426290930977e-07,
+ "loss": 0.0247,
+ "step": 33843
+ },
+ {
+ "epoch": 92.21798365122616,
+ "grad_norm": 0.6308934688568115,
+ "learning_rate": 3.159441482663139e-07,
+ "loss": 0.0076,
+ "step": 33844
+ },
+ {
+ "epoch": 92.22070844686648,
+ "grad_norm": 0.8357388377189636,
+ "learning_rate": 3.1572410904243566e-07,
+ "loss": 0.0062,
+ "step": 33845
+ },
+ {
+ "epoch": 92.22343324250681,
+ "grad_norm": 0.6922870874404907,
+ "learning_rate": 3.155041452393859e-07,
+ "loss": 0.0067,
+ "step": 33846
+ },
+ {
+ "epoch": 92.22615803814713,
+ "grad_norm": 1.2071850299835205,
+ "learning_rate": 3.1528425685887874e-07,
+ "loss": 0.042,
+ "step": 33847
+ },
+ {
+ "epoch": 92.22888283378747,
+ "grad_norm": 0.7584984302520752,
+ "learning_rate": 3.15064443902624e-07,
+ "loss": 0.0103,
+ "step": 33848
+ },
+ {
+ "epoch": 92.2316076294278,
+ "grad_norm": 0.6427136659622192,
+ "learning_rate": 3.14844706372337e-07,
+ "loss": 0.0056,
+ "step": 33849
+ },
+ {
+ "epoch": 92.23433242506812,
+ "grad_norm": 1.1847141981124878,
+ "learning_rate": 3.146250442697285e-07,
+ "loss": 0.0699,
+ "step": 33850
+ },
+ {
+ "epoch": 92.23705722070845,
+ "grad_norm": 1.646498680114746,
+ "learning_rate": 3.144054575965072e-07,
+ "loss": 0.0808,
+ "step": 33851
+ },
+ {
+ "epoch": 92.23978201634877,
+ "grad_norm": 1.7201144695281982,
+ "learning_rate": 3.141859463543839e-07,
+ "loss": 0.005,
+ "step": 33852
+ },
+ {
+ "epoch": 92.2425068119891,
+ "grad_norm": 1.0422029495239258,
+ "learning_rate": 3.1396651054506956e-07,
+ "loss": 0.0183,
+ "step": 33853
+ },
+ {
+ "epoch": 92.24523160762942,
+ "grad_norm": 1.6642436981201172,
+ "learning_rate": 3.137471501702727e-07,
+ "loss": 0.1248,
+ "step": 33854
+ },
+ {
+ "epoch": 92.24795640326975,
+ "grad_norm": 0.9662118554115295,
+ "learning_rate": 3.1352786523170085e-07,
+ "loss": 0.0127,
+ "step": 33855
+ },
+ {
+ "epoch": 92.25068119891009,
+ "grad_norm": 1.5360617637634277,
+ "learning_rate": 3.1330865573106274e-07,
+ "loss": 0.0718,
+ "step": 33856
+ },
+ {
+ "epoch": 92.2534059945504,
+ "grad_norm": 1.0927133560180664,
+ "learning_rate": 3.130895216700636e-07,
+ "loss": 0.0097,
+ "step": 33857
+ },
+ {
+ "epoch": 92.25613079019074,
+ "grad_norm": 1.3534842729568481,
+ "learning_rate": 3.1287046305041313e-07,
+ "loss": 0.0167,
+ "step": 33858
+ },
+ {
+ "epoch": 92.25885558583106,
+ "grad_norm": 0.8092286586761475,
+ "learning_rate": 3.126514798738156e-07,
+ "loss": 0.0122,
+ "step": 33859
+ },
+ {
+ "epoch": 92.26158038147139,
+ "grad_norm": 1.8035444021224976,
+ "learning_rate": 3.124325721419752e-07,
+ "loss": 0.095,
+ "step": 33860
+ },
+ {
+ "epoch": 92.26430517711172,
+ "grad_norm": 1.2876704931259155,
+ "learning_rate": 3.122137398565994e-07,
+ "loss": 0.01,
+ "step": 33861
+ },
+ {
+ "epoch": 92.26702997275204,
+ "grad_norm": 0.9556859135627747,
+ "learning_rate": 3.119949830193902e-07,
+ "loss": 0.0111,
+ "step": 33862
+ },
+ {
+ "epoch": 92.26975476839237,
+ "grad_norm": 0.5078927278518677,
+ "learning_rate": 3.1177630163205406e-07,
+ "loss": 0.0041,
+ "step": 33863
+ },
+ {
+ "epoch": 92.2724795640327,
+ "grad_norm": 0.48346251249313354,
+ "learning_rate": 3.115576956962896e-07,
+ "loss": 0.0041,
+ "step": 33864
+ },
+ {
+ "epoch": 92.27520435967303,
+ "grad_norm": 0.8472157716751099,
+ "learning_rate": 3.1133916521380427e-07,
+ "loss": 0.0093,
+ "step": 33865
+ },
+ {
+ "epoch": 92.27792915531334,
+ "grad_norm": 0.827942430973053,
+ "learning_rate": 3.111207101862968e-07,
+ "loss": 0.1138,
+ "step": 33866
+ },
+ {
+ "epoch": 92.28065395095368,
+ "grad_norm": 0.7452948689460754,
+ "learning_rate": 3.1090233061547017e-07,
+ "loss": 0.0125,
+ "step": 33867
+ },
+ {
+ "epoch": 92.28337874659401,
+ "grad_norm": 0.6881251931190491,
+ "learning_rate": 3.106840265030231e-07,
+ "loss": 0.0065,
+ "step": 33868
+ },
+ {
+ "epoch": 92.28610354223433,
+ "grad_norm": 1.0659189224243164,
+ "learning_rate": 3.1046579785065756e-07,
+ "loss": 0.0061,
+ "step": 33869
+ },
+ {
+ "epoch": 92.28882833787466,
+ "grad_norm": 1.034164547920227,
+ "learning_rate": 3.1024764466007106e-07,
+ "loss": 0.1209,
+ "step": 33870
+ },
+ {
+ "epoch": 92.29155313351498,
+ "grad_norm": 0.4200597405433655,
+ "learning_rate": 3.100295669329667e-07,
+ "loss": 0.0026,
+ "step": 33871
+ },
+ {
+ "epoch": 92.29427792915531,
+ "grad_norm": 1.4467817544937134,
+ "learning_rate": 3.0981156467103867e-07,
+ "loss": 0.0638,
+ "step": 33872
+ },
+ {
+ "epoch": 92.29700272479565,
+ "grad_norm": 1.2791281938552856,
+ "learning_rate": 3.095936378759856e-07,
+ "loss": 0.0132,
+ "step": 33873
+ },
+ {
+ "epoch": 92.29972752043597,
+ "grad_norm": 1.2298080921173096,
+ "learning_rate": 3.0937578654950727e-07,
+ "loss": 0.0076,
+ "step": 33874
+ },
+ {
+ "epoch": 92.3024523160763,
+ "grad_norm": 1.2244395017623901,
+ "learning_rate": 3.091580106932979e-07,
+ "loss": 0.0069,
+ "step": 33875
+ },
+ {
+ "epoch": 92.30517711171662,
+ "grad_norm": 0.8025880455970764,
+ "learning_rate": 3.089403103090538e-07,
+ "loss": 0.0341,
+ "step": 33876
+ },
+ {
+ "epoch": 92.30790190735695,
+ "grad_norm": 0.8709864020347595,
+ "learning_rate": 3.0872268539846926e-07,
+ "loss": 0.0107,
+ "step": 33877
+ },
+ {
+ "epoch": 92.31062670299727,
+ "grad_norm": 1.4883372783660889,
+ "learning_rate": 3.0850513596324294e-07,
+ "loss": 0.0186,
+ "step": 33878
+ },
+ {
+ "epoch": 92.3133514986376,
+ "grad_norm": 1.2974320650100708,
+ "learning_rate": 3.0828766200506563e-07,
+ "loss": 0.0079,
+ "step": 33879
+ },
+ {
+ "epoch": 92.31607629427793,
+ "grad_norm": 1.158897876739502,
+ "learning_rate": 3.0807026352563273e-07,
+ "loss": 0.0398,
+ "step": 33880
+ },
+ {
+ "epoch": 92.31880108991825,
+ "grad_norm": 1.2469861507415771,
+ "learning_rate": 3.078529405266351e-07,
+ "loss": 0.0256,
+ "step": 33881
+ },
+ {
+ "epoch": 92.32152588555859,
+ "grad_norm": 0.8509884476661682,
+ "learning_rate": 3.0763569300976794e-07,
+ "loss": 0.016,
+ "step": 33882
+ },
+ {
+ "epoch": 92.3242506811989,
+ "grad_norm": 0.7714977860450745,
+ "learning_rate": 3.074185209767222e-07,
+ "loss": 0.0065,
+ "step": 33883
+ },
+ {
+ "epoch": 92.32697547683924,
+ "grad_norm": 0.8207780122756958,
+ "learning_rate": 3.0720142442918765e-07,
+ "loss": 0.0091,
+ "step": 33884
+ },
+ {
+ "epoch": 92.32970027247957,
+ "grad_norm": 0.4661642014980316,
+ "learning_rate": 3.069844033688574e-07,
+ "loss": 0.004,
+ "step": 33885
+ },
+ {
+ "epoch": 92.33242506811989,
+ "grad_norm": 1.1099629402160645,
+ "learning_rate": 3.0676745779742114e-07,
+ "loss": 0.0191,
+ "step": 33886
+ },
+ {
+ "epoch": 92.33514986376022,
+ "grad_norm": 0.6219572424888611,
+ "learning_rate": 3.0655058771656755e-07,
+ "loss": 0.0059,
+ "step": 33887
+ },
+ {
+ "epoch": 92.33787465940054,
+ "grad_norm": 1.1818000078201294,
+ "learning_rate": 3.0633379312798637e-07,
+ "loss": 0.0305,
+ "step": 33888
+ },
+ {
+ "epoch": 92.34059945504087,
+ "grad_norm": 1.0469355583190918,
+ "learning_rate": 3.061170740333641e-07,
+ "loss": 0.0116,
+ "step": 33889
+ },
+ {
+ "epoch": 92.34332425068119,
+ "grad_norm": 1.3827615976333618,
+ "learning_rate": 3.0590043043439156e-07,
+ "loss": 0.0237,
+ "step": 33890
+ },
+ {
+ "epoch": 92.34604904632153,
+ "grad_norm": 1.105055332183838,
+ "learning_rate": 3.0568386233275404e-07,
+ "loss": 0.0132,
+ "step": 33891
+ },
+ {
+ "epoch": 92.34877384196186,
+ "grad_norm": 0.6753857731819153,
+ "learning_rate": 3.0546736973013915e-07,
+ "loss": 0.0078,
+ "step": 33892
+ },
+ {
+ "epoch": 92.35149863760218,
+ "grad_norm": 1.416237711906433,
+ "learning_rate": 3.05250952628231e-07,
+ "loss": 0.0295,
+ "step": 33893
+ },
+ {
+ "epoch": 92.35422343324251,
+ "grad_norm": 1.029259204864502,
+ "learning_rate": 3.0503461102871836e-07,
+ "loss": 0.0081,
+ "step": 33894
+ },
+ {
+ "epoch": 92.35694822888283,
+ "grad_norm": 1.02763831615448,
+ "learning_rate": 3.048183449332831e-07,
+ "loss": 0.0869,
+ "step": 33895
+ },
+ {
+ "epoch": 92.35967302452316,
+ "grad_norm": 0.7746995687484741,
+ "learning_rate": 3.046021543436106e-07,
+ "loss": 0.0064,
+ "step": 33896
+ },
+ {
+ "epoch": 92.3623978201635,
+ "grad_norm": 0.6067550182342529,
+ "learning_rate": 3.04386039261384e-07,
+ "loss": 0.0053,
+ "step": 33897
+ },
+ {
+ "epoch": 92.36512261580381,
+ "grad_norm": 0.8986235857009888,
+ "learning_rate": 3.0416999968828743e-07,
+ "loss": 0.0105,
+ "step": 33898
+ },
+ {
+ "epoch": 92.36784741144415,
+ "grad_norm": 1.2879085540771484,
+ "learning_rate": 3.039540356260029e-07,
+ "loss": 0.176,
+ "step": 33899
+ },
+ {
+ "epoch": 92.37057220708446,
+ "grad_norm": 0.6998643279075623,
+ "learning_rate": 3.037381470762124e-07,
+ "loss": 0.0071,
+ "step": 33900
+ },
+ {
+ "epoch": 92.3732970027248,
+ "grad_norm": 0.592660129070282,
+ "learning_rate": 3.035223340405957e-07,
+ "loss": 0.0077,
+ "step": 33901
+ },
+ {
+ "epoch": 92.37602179836512,
+ "grad_norm": 0.5025231242179871,
+ "learning_rate": 3.033065965208359e-07,
+ "loss": 0.0039,
+ "step": 33902
+ },
+ {
+ "epoch": 92.37874659400545,
+ "grad_norm": 1.6738100051879883,
+ "learning_rate": 3.030909345186128e-07,
+ "loss": 0.0342,
+ "step": 33903
+ },
+ {
+ "epoch": 92.38147138964578,
+ "grad_norm": 0.8989980816841125,
+ "learning_rate": 3.028753480356061e-07,
+ "loss": 0.0101,
+ "step": 33904
+ },
+ {
+ "epoch": 92.3841961852861,
+ "grad_norm": 1.0017898082733154,
+ "learning_rate": 3.026598370734912e-07,
+ "loss": 0.0222,
+ "step": 33905
+ },
+ {
+ "epoch": 92.38692098092643,
+ "grad_norm": 1.1279137134552002,
+ "learning_rate": 3.0244440163395115e-07,
+ "loss": 0.0073,
+ "step": 33906
+ },
+ {
+ "epoch": 92.38964577656675,
+ "grad_norm": 1.2833013534545898,
+ "learning_rate": 3.022290417186624e-07,
+ "loss": 0.0139,
+ "step": 33907
+ },
+ {
+ "epoch": 92.39237057220708,
+ "grad_norm": 0.5399102568626404,
+ "learning_rate": 3.020137573293014e-07,
+ "loss": 0.0055,
+ "step": 33908
+ },
+ {
+ "epoch": 92.39509536784742,
+ "grad_norm": 1.187691569328308,
+ "learning_rate": 3.0179854846754455e-07,
+ "loss": 0.144,
+ "step": 33909
+ },
+ {
+ "epoch": 92.39782016348774,
+ "grad_norm": 1.0239044427871704,
+ "learning_rate": 3.015834151350694e-07,
+ "loss": 0.041,
+ "step": 33910
+ },
+ {
+ "epoch": 92.40054495912807,
+ "grad_norm": 1.2796515226364136,
+ "learning_rate": 3.0136835733355017e-07,
+ "loss": 0.0736,
+ "step": 33911
+ },
+ {
+ "epoch": 92.40326975476839,
+ "grad_norm": 0.7347439527511597,
+ "learning_rate": 3.0115337506466226e-07,
+ "loss": 0.0082,
+ "step": 33912
+ },
+ {
+ "epoch": 92.40599455040872,
+ "grad_norm": 1.557692289352417,
+ "learning_rate": 3.009384683300787e-07,
+ "loss": 0.0108,
+ "step": 33913
+ },
+ {
+ "epoch": 92.40871934604904,
+ "grad_norm": 1.072334885597229,
+ "learning_rate": 3.007236371314748e-07,
+ "loss": 0.0228,
+ "step": 33914
+ },
+ {
+ "epoch": 92.41144414168937,
+ "grad_norm": 0.9456065893173218,
+ "learning_rate": 3.0050888147052373e-07,
+ "loss": 0.0165,
+ "step": 33915
+ },
+ {
+ "epoch": 92.4141689373297,
+ "grad_norm": 0.968695342540741,
+ "learning_rate": 3.002942013488974e-07,
+ "loss": 0.0664,
+ "step": 33916
+ },
+ {
+ "epoch": 92.41689373297002,
+ "grad_norm": 0.92174232006073,
+ "learning_rate": 3.000795967682657e-07,
+ "loss": 0.0119,
+ "step": 33917
+ },
+ {
+ "epoch": 92.41961852861036,
+ "grad_norm": 0.5307268500328064,
+ "learning_rate": 2.9986506773030386e-07,
+ "loss": 0.0047,
+ "step": 33918
+ },
+ {
+ "epoch": 92.42234332425068,
+ "grad_norm": 1.2574512958526611,
+ "learning_rate": 2.996506142366795e-07,
+ "loss": 0.0098,
+ "step": 33919
+ },
+ {
+ "epoch": 92.42506811989101,
+ "grad_norm": 1.224897861480713,
+ "learning_rate": 2.994362362890657e-07,
+ "loss": 0.0433,
+ "step": 33920
+ },
+ {
+ "epoch": 92.42779291553134,
+ "grad_norm": 0.9889746308326721,
+ "learning_rate": 2.992219338891278e-07,
+ "loss": 0.0126,
+ "step": 33921
+ },
+ {
+ "epoch": 92.43051771117166,
+ "grad_norm": 1.3134145736694336,
+ "learning_rate": 2.9900770703853774e-07,
+ "loss": 0.0897,
+ "step": 33922
+ },
+ {
+ "epoch": 92.433242506812,
+ "grad_norm": 0.7447843551635742,
+ "learning_rate": 2.9879355573896427e-07,
+ "loss": 0.0051,
+ "step": 33923
+ },
+ {
+ "epoch": 92.43596730245231,
+ "grad_norm": 1.3304386138916016,
+ "learning_rate": 2.9857947999207493e-07,
+ "loss": 0.0816,
+ "step": 33924
+ },
+ {
+ "epoch": 92.43869209809264,
+ "grad_norm": 1.3272194862365723,
+ "learning_rate": 2.98365479799535e-07,
+ "loss": 0.0751,
+ "step": 33925
+ },
+ {
+ "epoch": 92.44141689373296,
+ "grad_norm": 1.247362732887268,
+ "learning_rate": 2.981515551630132e-07,
+ "loss": 0.1011,
+ "step": 33926
+ },
+ {
+ "epoch": 92.4441416893733,
+ "grad_norm": 1.028810739517212,
+ "learning_rate": 2.979377060841748e-07,
+ "loss": 0.0314,
+ "step": 33927
+ },
+ {
+ "epoch": 92.44686648501363,
+ "grad_norm": 0.634709358215332,
+ "learning_rate": 2.9772393256468634e-07,
+ "loss": 0.0047,
+ "step": 33928
+ },
+ {
+ "epoch": 92.44959128065395,
+ "grad_norm": 1.9204241037368774,
+ "learning_rate": 2.9751023460620975e-07,
+ "loss": 0.0118,
+ "step": 33929
+ },
+ {
+ "epoch": 92.45231607629428,
+ "grad_norm": 1.109150767326355,
+ "learning_rate": 2.972966122104115e-07,
+ "loss": 0.008,
+ "step": 33930
+ },
+ {
+ "epoch": 92.4550408719346,
+ "grad_norm": 1.1206895112991333,
+ "learning_rate": 2.9708306537895583e-07,
+ "loss": 0.013,
+ "step": 33931
+ },
+ {
+ "epoch": 92.45776566757493,
+ "grad_norm": 1.0416812896728516,
+ "learning_rate": 2.968695941135047e-07,
+ "loss": 0.0156,
+ "step": 33932
+ },
+ {
+ "epoch": 92.46049046321527,
+ "grad_norm": 0.9177446365356445,
+ "learning_rate": 2.9665619841572015e-07,
+ "loss": 0.0196,
+ "step": 33933
+ },
+ {
+ "epoch": 92.46321525885558,
+ "grad_norm": 0.907960057258606,
+ "learning_rate": 2.964428782872664e-07,
+ "loss": 0.0065,
+ "step": 33934
+ },
+ {
+ "epoch": 92.46594005449592,
+ "grad_norm": 1.2127174139022827,
+ "learning_rate": 2.9622963372980204e-07,
+ "loss": 0.0615,
+ "step": 33935
+ },
+ {
+ "epoch": 92.46866485013624,
+ "grad_norm": 0.8162363767623901,
+ "learning_rate": 2.9601646474498926e-07,
+ "loss": 0.0192,
+ "step": 33936
+ },
+ {
+ "epoch": 92.47138964577657,
+ "grad_norm": 0.8205770254135132,
+ "learning_rate": 2.958033713344877e-07,
+ "loss": 0.0114,
+ "step": 33937
+ },
+ {
+ "epoch": 92.47411444141689,
+ "grad_norm": 0.28497329354286194,
+ "learning_rate": 2.9559035349995715e-07,
+ "loss": 0.0026,
+ "step": 33938
+ },
+ {
+ "epoch": 92.47683923705722,
+ "grad_norm": 1.0583525896072388,
+ "learning_rate": 2.953774112430563e-07,
+ "loss": 0.0602,
+ "step": 33939
+ },
+ {
+ "epoch": 92.47956403269755,
+ "grad_norm": 1.339512586593628,
+ "learning_rate": 2.9516454456544386e-07,
+ "loss": 0.0107,
+ "step": 33940
+ },
+ {
+ "epoch": 92.48228882833787,
+ "grad_norm": 0.6908200979232788,
+ "learning_rate": 2.9495175346877846e-07,
+ "loss": 0.0091,
+ "step": 33941
+ },
+ {
+ "epoch": 92.4850136239782,
+ "grad_norm": 1.6842223405838013,
+ "learning_rate": 2.9473903795471546e-07,
+ "loss": 0.0338,
+ "step": 33942
+ },
+ {
+ "epoch": 92.48773841961852,
+ "grad_norm": 1.4820711612701416,
+ "learning_rate": 2.9452639802491356e-07,
+ "loss": 0.0718,
+ "step": 33943
+ },
+ {
+ "epoch": 92.49046321525886,
+ "grad_norm": 0.7071810960769653,
+ "learning_rate": 2.9431383368102696e-07,
+ "loss": 0.0107,
+ "step": 33944
+ },
+ {
+ "epoch": 92.49318801089919,
+ "grad_norm": 1.1152889728546143,
+ "learning_rate": 2.941013449247132e-07,
+ "loss": 0.0315,
+ "step": 33945
+ },
+ {
+ "epoch": 92.49591280653951,
+ "grad_norm": 1.829584002494812,
+ "learning_rate": 2.9388893175762325e-07,
+ "loss": 0.0755,
+ "step": 33946
+ },
+ {
+ "epoch": 92.49863760217984,
+ "grad_norm": 1.4387803077697754,
+ "learning_rate": 2.936765941814157e-07,
+ "loss": 0.1058,
+ "step": 33947
+ },
+ {
+ "epoch": 92.50136239782016,
+ "grad_norm": 0.917516827583313,
+ "learning_rate": 2.934643321977426e-07,
+ "loss": 0.0129,
+ "step": 33948
+ },
+ {
+ "epoch": 92.50408719346049,
+ "grad_norm": 1.0029677152633667,
+ "learning_rate": 2.932521458082549e-07,
+ "loss": 0.0446,
+ "step": 33949
+ },
+ {
+ "epoch": 92.50681198910081,
+ "grad_norm": 0.45350703597068787,
+ "learning_rate": 2.93040035014609e-07,
+ "loss": 0.0039,
+ "step": 33950
+ },
+ {
+ "epoch": 92.50953678474114,
+ "grad_norm": 0.4705452024936676,
+ "learning_rate": 2.9282799981845464e-07,
+ "loss": 0.0046,
+ "step": 33951
+ },
+ {
+ "epoch": 92.51226158038148,
+ "grad_norm": 0.6593430042266846,
+ "learning_rate": 2.9261604022144397e-07,
+ "loss": 0.008,
+ "step": 33952
+ },
+ {
+ "epoch": 92.5149863760218,
+ "grad_norm": 0.531963050365448,
+ "learning_rate": 2.9240415622522557e-07,
+ "loss": 0.0046,
+ "step": 33953
+ },
+ {
+ "epoch": 92.51771117166213,
+ "grad_norm": 0.7758480310440063,
+ "learning_rate": 2.9219234783145147e-07,
+ "loss": 0.0078,
+ "step": 33954
+ },
+ {
+ "epoch": 92.52043596730245,
+ "grad_norm": 0.630588948726654,
+ "learning_rate": 2.9198061504177144e-07,
+ "loss": 0.0062,
+ "step": 33955
+ },
+ {
+ "epoch": 92.52316076294278,
+ "grad_norm": 1.2851837873458862,
+ "learning_rate": 2.917689578578331e-07,
+ "loss": 0.014,
+ "step": 33956
+ },
+ {
+ "epoch": 92.52588555858311,
+ "grad_norm": 0.8895322680473328,
+ "learning_rate": 2.915573762812873e-07,
+ "loss": 0.0049,
+ "step": 33957
+ },
+ {
+ "epoch": 92.52861035422343,
+ "grad_norm": 1.6794418096542358,
+ "learning_rate": 2.9134587031377835e-07,
+ "loss": 0.0121,
+ "step": 33958
+ },
+ {
+ "epoch": 92.53133514986376,
+ "grad_norm": 0.9976317286491394,
+ "learning_rate": 2.91134439956956e-07,
+ "loss": 0.0125,
+ "step": 33959
+ },
+ {
+ "epoch": 92.53405994550408,
+ "grad_norm": 1.3423060178756714,
+ "learning_rate": 2.909230852124667e-07,
+ "loss": 0.022,
+ "step": 33960
+ },
+ {
+ "epoch": 92.53678474114442,
+ "grad_norm": 1.2505449056625366,
+ "learning_rate": 2.907118060819569e-07,
+ "loss": 0.0161,
+ "step": 33961
+ },
+ {
+ "epoch": 92.53950953678473,
+ "grad_norm": 1.4436038732528687,
+ "learning_rate": 2.905006025670687e-07,
+ "loss": 0.015,
+ "step": 33962
+ },
+ {
+ "epoch": 92.54223433242507,
+ "grad_norm": 0.8681697249412537,
+ "learning_rate": 2.902894746694507e-07,
+ "loss": 0.0061,
+ "step": 33963
+ },
+ {
+ "epoch": 92.5449591280654,
+ "grad_norm": 1.149079442024231,
+ "learning_rate": 2.90078422390746e-07,
+ "loss": 0.0467,
+ "step": 33964
+ },
+ {
+ "epoch": 92.54768392370572,
+ "grad_norm": 1.076175570487976,
+ "learning_rate": 2.898674457325967e-07,
+ "loss": 0.0077,
+ "step": 33965
+ },
+ {
+ "epoch": 92.55040871934605,
+ "grad_norm": 1.2203747034072876,
+ "learning_rate": 2.8965654469664815e-07,
+ "loss": 0.0123,
+ "step": 33966
+ },
+ {
+ "epoch": 92.55313351498637,
+ "grad_norm": 1.2456499338150024,
+ "learning_rate": 2.8944571928454126e-07,
+ "loss": 0.0066,
+ "step": 33967
+ },
+ {
+ "epoch": 92.5558583106267,
+ "grad_norm": 0.8995123505592346,
+ "learning_rate": 2.892349694979202e-07,
+ "loss": 0.0282,
+ "step": 33968
+ },
+ {
+ "epoch": 92.55858310626704,
+ "grad_norm": 1.0337812900543213,
+ "learning_rate": 2.890242953384237e-07,
+ "loss": 0.0397,
+ "step": 33969
+ },
+ {
+ "epoch": 92.56130790190736,
+ "grad_norm": 1.1962248086929321,
+ "learning_rate": 2.888136968076927e-07,
+ "loss": 0.0234,
+ "step": 33970
+ },
+ {
+ "epoch": 92.56403269754769,
+ "grad_norm": 1.1268762350082397,
+ "learning_rate": 2.88603173907368e-07,
+ "loss": 0.0097,
+ "step": 33971
+ },
+ {
+ "epoch": 92.566757493188,
+ "grad_norm": 1.3689600229263306,
+ "learning_rate": 2.883927266390907e-07,
+ "loss": 0.063,
+ "step": 33972
+ },
+ {
+ "epoch": 92.56948228882834,
+ "grad_norm": 0.5250450968742371,
+ "learning_rate": 2.88182355004496e-07,
+ "loss": 0.0053,
+ "step": 33973
+ },
+ {
+ "epoch": 92.57220708446866,
+ "grad_norm": 0.9409488439559937,
+ "learning_rate": 2.8797205900522594e-07,
+ "loss": 0.0135,
+ "step": 33974
+ },
+ {
+ "epoch": 92.57493188010899,
+ "grad_norm": 0.44566628336906433,
+ "learning_rate": 2.877618386429171e-07,
+ "loss": 0.0044,
+ "step": 33975
+ },
+ {
+ "epoch": 92.57765667574932,
+ "grad_norm": 1.6837193965911865,
+ "learning_rate": 2.875516939192069e-07,
+ "loss": 0.0411,
+ "step": 33976
+ },
+ {
+ "epoch": 92.58038147138964,
+ "grad_norm": 0.7415374517440796,
+ "learning_rate": 2.873416248357308e-07,
+ "loss": 0.0087,
+ "step": 33977
+ },
+ {
+ "epoch": 92.58310626702998,
+ "grad_norm": 1.28081476688385,
+ "learning_rate": 2.8713163139412527e-07,
+ "loss": 0.0422,
+ "step": 33978
+ },
+ {
+ "epoch": 92.5858310626703,
+ "grad_norm": 0.8171629905700684,
+ "learning_rate": 2.869217135960267e-07,
+ "loss": 0.0062,
+ "step": 33979
+ },
+ {
+ "epoch": 92.58855585831063,
+ "grad_norm": 0.8560984134674072,
+ "learning_rate": 2.8671187144306834e-07,
+ "loss": 0.0959,
+ "step": 33980
+ },
+ {
+ "epoch": 92.59128065395096,
+ "grad_norm": 1.0430504083633423,
+ "learning_rate": 2.8650210493688656e-07,
+ "loss": 0.0091,
+ "step": 33981
+ },
+ {
+ "epoch": 92.59400544959128,
+ "grad_norm": 0.8616706132888794,
+ "learning_rate": 2.862924140791123e-07,
+ "loss": 0.0085,
+ "step": 33982
+ },
+ {
+ "epoch": 92.59673024523161,
+ "grad_norm": 0.9433708786964417,
+ "learning_rate": 2.8608279887138103e-07,
+ "loss": 0.0599,
+ "step": 33983
+ },
+ {
+ "epoch": 92.59945504087193,
+ "grad_norm": 1.0898241996765137,
+ "learning_rate": 2.858732593153246e-07,
+ "loss": 0.0353,
+ "step": 33984
+ },
+ {
+ "epoch": 92.60217983651226,
+ "grad_norm": 0.9720243215560913,
+ "learning_rate": 2.8566379541257406e-07,
+ "loss": 0.0175,
+ "step": 33985
+ },
+ {
+ "epoch": 92.60490463215258,
+ "grad_norm": 1.9089182615280151,
+ "learning_rate": 2.8545440716476025e-07,
+ "loss": 0.0119,
+ "step": 33986
+ },
+ {
+ "epoch": 92.60762942779292,
+ "grad_norm": 1.5685304403305054,
+ "learning_rate": 2.8524509457351635e-07,
+ "loss": 0.099,
+ "step": 33987
+ },
+ {
+ "epoch": 92.61035422343325,
+ "grad_norm": 0.9736130833625793,
+ "learning_rate": 2.85035857640471e-07,
+ "loss": 0.0194,
+ "step": 33988
+ },
+ {
+ "epoch": 92.61307901907357,
+ "grad_norm": 0.7703962326049805,
+ "learning_rate": 2.84826696367253e-07,
+ "loss": 0.0074,
+ "step": 33989
+ },
+ {
+ "epoch": 92.6158038147139,
+ "grad_norm": 1.1764954328536987,
+ "learning_rate": 2.846176107554921e-07,
+ "loss": 0.0299,
+ "step": 33990
+ },
+ {
+ "epoch": 92.61852861035422,
+ "grad_norm": 1.0237224102020264,
+ "learning_rate": 2.8440860080681587e-07,
+ "loss": 0.0179,
+ "step": 33991
+ },
+ {
+ "epoch": 92.62125340599455,
+ "grad_norm": 0.7398515343666077,
+ "learning_rate": 2.8419966652285303e-07,
+ "loss": 0.0155,
+ "step": 33992
+ },
+ {
+ "epoch": 92.62397820163488,
+ "grad_norm": 1.047119379043579,
+ "learning_rate": 2.8399080790523227e-07,
+ "loss": 0.0217,
+ "step": 33993
+ },
+ {
+ "epoch": 92.6267029972752,
+ "grad_norm": 0.8056607842445374,
+ "learning_rate": 2.837820249555756e-07,
+ "loss": 0.0127,
+ "step": 33994
+ },
+ {
+ "epoch": 92.62942779291554,
+ "grad_norm": 1.2474956512451172,
+ "learning_rate": 2.8357331767551397e-07,
+ "loss": 0.093,
+ "step": 33995
+ },
+ {
+ "epoch": 92.63215258855585,
+ "grad_norm": 0.735967218875885,
+ "learning_rate": 2.8336468606667057e-07,
+ "loss": 0.0068,
+ "step": 33996
+ },
+ {
+ "epoch": 92.63487738419619,
+ "grad_norm": 1.0064735412597656,
+ "learning_rate": 2.8315613013066954e-07,
+ "loss": 0.0218,
+ "step": 33997
+ },
+ {
+ "epoch": 92.6376021798365,
+ "grad_norm": 1.3311595916748047,
+ "learning_rate": 2.8294764986913524e-07,
+ "loss": 0.0242,
+ "step": 33998
+ },
+ {
+ "epoch": 92.64032697547684,
+ "grad_norm": 0.8334822058677673,
+ "learning_rate": 2.82739245283693e-07,
+ "loss": 0.0083,
+ "step": 33999
+ },
+ {
+ "epoch": 92.64305177111717,
+ "grad_norm": 0.5948672294616699,
+ "learning_rate": 2.8253091637596485e-07,
+ "loss": 0.0065,
+ "step": 34000
+ },
+ {
+ "epoch": 92.64577656675749,
+ "grad_norm": 0.7195226550102234,
+ "learning_rate": 2.8232266314757286e-07,
+ "loss": 0.0061,
+ "step": 34001
+ },
+ {
+ "epoch": 92.64850136239782,
+ "grad_norm": 1.026427984237671,
+ "learning_rate": 2.8211448560013787e-07,
+ "loss": 0.008,
+ "step": 34002
+ },
+ {
+ "epoch": 92.65122615803814,
+ "grad_norm": 0.9751279354095459,
+ "learning_rate": 2.8190638373528423e-07,
+ "loss": 0.1246,
+ "step": 34003
+ },
+ {
+ "epoch": 92.65395095367847,
+ "grad_norm": 1.0564044713974,
+ "learning_rate": 2.816983575546306e-07,
+ "loss": 0.0417,
+ "step": 34004
+ },
+ {
+ "epoch": 92.65667574931881,
+ "grad_norm": 2.3748745918273926,
+ "learning_rate": 2.814904070597979e-07,
+ "loss": 0.0453,
+ "step": 34005
+ },
+ {
+ "epoch": 92.65940054495913,
+ "grad_norm": 0.8862770795822144,
+ "learning_rate": 2.8128253225240376e-07,
+ "loss": 0.015,
+ "step": 34006
+ },
+ {
+ "epoch": 92.66212534059946,
+ "grad_norm": 0.8816129565238953,
+ "learning_rate": 2.810747331340691e-07,
+ "loss": 0.0078,
+ "step": 34007
+ },
+ {
+ "epoch": 92.66485013623978,
+ "grad_norm": 1.1335464715957642,
+ "learning_rate": 2.8086700970641034e-07,
+ "loss": 0.0206,
+ "step": 34008
+ },
+ {
+ "epoch": 92.66757493188011,
+ "grad_norm": 1.1507577896118164,
+ "learning_rate": 2.806593619710485e-07,
+ "loss": 0.0407,
+ "step": 34009
+ },
+ {
+ "epoch": 92.67029972752043,
+ "grad_norm": 1.3220223188400269,
+ "learning_rate": 2.804517899295978e-07,
+ "loss": 0.0114,
+ "step": 34010
+ },
+ {
+ "epoch": 92.67302452316076,
+ "grad_norm": 1.2072594165802002,
+ "learning_rate": 2.802442935836758e-07,
+ "loss": 0.0163,
+ "step": 34011
+ },
+ {
+ "epoch": 92.6757493188011,
+ "grad_norm": 1.4605786800384521,
+ "learning_rate": 2.800368729348979e-07,
+ "loss": 0.021,
+ "step": 34012
+ },
+ {
+ "epoch": 92.67847411444141,
+ "grad_norm": 0.8624869585037231,
+ "learning_rate": 2.798295279848817e-07,
+ "loss": 0.0336,
+ "step": 34013
+ },
+ {
+ "epoch": 92.68119891008175,
+ "grad_norm": 0.5233227014541626,
+ "learning_rate": 2.796222587352382e-07,
+ "loss": 0.0078,
+ "step": 34014
+ },
+ {
+ "epoch": 92.68392370572207,
+ "grad_norm": 0.7184723615646362,
+ "learning_rate": 2.7941506518758486e-07,
+ "loss": 0.0463,
+ "step": 34015
+ },
+ {
+ "epoch": 92.6866485013624,
+ "grad_norm": 1.5253820419311523,
+ "learning_rate": 2.7920794734353384e-07,
+ "loss": 0.0799,
+ "step": 34016
+ },
+ {
+ "epoch": 92.68937329700273,
+ "grad_norm": 0.47415798902511597,
+ "learning_rate": 2.790009052046994e-07,
+ "loss": 0.0041,
+ "step": 34017
+ },
+ {
+ "epoch": 92.69209809264305,
+ "grad_norm": 1.9702260494232178,
+ "learning_rate": 2.787939387726912e-07,
+ "loss": 0.0609,
+ "step": 34018
+ },
+ {
+ "epoch": 92.69482288828338,
+ "grad_norm": 1.1195886135101318,
+ "learning_rate": 2.7858704804912483e-07,
+ "loss": 0.0233,
+ "step": 34019
+ },
+ {
+ "epoch": 92.6975476839237,
+ "grad_norm": 0.4756803810596466,
+ "learning_rate": 2.7838023303560893e-07,
+ "loss": 0.0049,
+ "step": 34020
+ },
+ {
+ "epoch": 92.70027247956403,
+ "grad_norm": 1.279557704925537,
+ "learning_rate": 2.7817349373375545e-07,
+ "loss": 0.0318,
+ "step": 34021
+ },
+ {
+ "epoch": 92.70299727520435,
+ "grad_norm": 0.9025135636329651,
+ "learning_rate": 2.779668301451721e-07,
+ "loss": 0.0753,
+ "step": 34022
+ },
+ {
+ "epoch": 92.70572207084469,
+ "grad_norm": 1.2179107666015625,
+ "learning_rate": 2.7776024227147093e-07,
+ "loss": 0.0347,
+ "step": 34023
+ },
+ {
+ "epoch": 92.70844686648502,
+ "grad_norm": 0.8794276714324951,
+ "learning_rate": 2.775537301142617e-07,
+ "loss": 0.0102,
+ "step": 34024
+ },
+ {
+ "epoch": 92.71117166212534,
+ "grad_norm": 0.8789601922035217,
+ "learning_rate": 2.773472936751487e-07,
+ "loss": 0.0097,
+ "step": 34025
+ },
+ {
+ "epoch": 92.71389645776567,
+ "grad_norm": 1.059244155883789,
+ "learning_rate": 2.7714093295574285e-07,
+ "loss": 0.0155,
+ "step": 34026
+ },
+ {
+ "epoch": 92.71662125340599,
+ "grad_norm": 0.9304068088531494,
+ "learning_rate": 2.769346479576507e-07,
+ "loss": 0.0066,
+ "step": 34027
+ },
+ {
+ "epoch": 92.71934604904632,
+ "grad_norm": 1.1130577325820923,
+ "learning_rate": 2.7672843868247865e-07,
+ "loss": 0.0502,
+ "step": 34028
+ },
+ {
+ "epoch": 92.72207084468666,
+ "grad_norm": 1.1083345413208008,
+ "learning_rate": 2.765223051318333e-07,
+ "loss": 0.0427,
+ "step": 34029
+ },
+ {
+ "epoch": 92.72479564032697,
+ "grad_norm": 0.8988147974014282,
+ "learning_rate": 2.7631624730731887e-07,
+ "loss": 0.0202,
+ "step": 34030
+ },
+ {
+ "epoch": 92.7275204359673,
+ "grad_norm": 0.8107754588127136,
+ "learning_rate": 2.7611026521053963e-07,
+ "loss": 0.0087,
+ "step": 34031
+ },
+ {
+ "epoch": 92.73024523160763,
+ "grad_norm": 1.5289849042892456,
+ "learning_rate": 2.75904358843101e-07,
+ "loss": 0.1508,
+ "step": 34032
+ },
+ {
+ "epoch": 92.73297002724796,
+ "grad_norm": 2.258798360824585,
+ "learning_rate": 2.7569852820660713e-07,
+ "loss": 0.0108,
+ "step": 34033
+ },
+ {
+ "epoch": 92.73569482288828,
+ "grad_norm": 2.01765513420105,
+ "learning_rate": 2.754927733026591e-07,
+ "loss": 0.038,
+ "step": 34034
+ },
+ {
+ "epoch": 92.73841961852861,
+ "grad_norm": 1.2749356031417847,
+ "learning_rate": 2.7528709413286116e-07,
+ "loss": 0.0208,
+ "step": 34035
+ },
+ {
+ "epoch": 92.74114441416894,
+ "grad_norm": 1.5671273469924927,
+ "learning_rate": 2.750814906988142e-07,
+ "loss": 0.0392,
+ "step": 34036
+ },
+ {
+ "epoch": 92.74386920980926,
+ "grad_norm": 1.1500232219696045,
+ "learning_rate": 2.748759630021203e-07,
+ "loss": 0.0196,
+ "step": 34037
+ },
+ {
+ "epoch": 92.7465940054496,
+ "grad_norm": 0.7993881106376648,
+ "learning_rate": 2.74670511044377e-07,
+ "loss": 0.0063,
+ "step": 34038
+ },
+ {
+ "epoch": 92.74931880108991,
+ "grad_norm": 1.7025288343429565,
+ "learning_rate": 2.7446513482718874e-07,
+ "loss": 0.0408,
+ "step": 34039
+ },
+ {
+ "epoch": 92.75204359673025,
+ "grad_norm": 1.08448326587677,
+ "learning_rate": 2.74259834352153e-07,
+ "loss": 0.0296,
+ "step": 34040
+ },
+ {
+ "epoch": 92.75476839237058,
+ "grad_norm": 1.5050565004348755,
+ "learning_rate": 2.740546096208685e-07,
+ "loss": 0.0166,
+ "step": 34041
+ },
+ {
+ "epoch": 92.7574931880109,
+ "grad_norm": 1.4759631156921387,
+ "learning_rate": 2.738494606349329e-07,
+ "loss": 0.0607,
+ "step": 34042
+ },
+ {
+ "epoch": 92.76021798365123,
+ "grad_norm": 1.1877226829528809,
+ "learning_rate": 2.736443873959449e-07,
+ "loss": 0.0263,
+ "step": 34043
+ },
+ {
+ "epoch": 92.76294277929155,
+ "grad_norm": 1.2834104299545288,
+ "learning_rate": 2.73439389905501e-07,
+ "loss": 0.1102,
+ "step": 34044
+ },
+ {
+ "epoch": 92.76566757493188,
+ "grad_norm": 0.7748650312423706,
+ "learning_rate": 2.7323446816519993e-07,
+ "loss": 0.0179,
+ "step": 34045
+ },
+ {
+ "epoch": 92.7683923705722,
+ "grad_norm": 1.4086123704910278,
+ "learning_rate": 2.7302962217663486e-07,
+ "loss": 0.0282,
+ "step": 34046
+ },
+ {
+ "epoch": 92.77111716621253,
+ "grad_norm": 1.0793348550796509,
+ "learning_rate": 2.728248519414001e-07,
+ "loss": 0.0191,
+ "step": 34047
+ },
+ {
+ "epoch": 92.77384196185287,
+ "grad_norm": 0.5681210160255432,
+ "learning_rate": 2.7262015746109425e-07,
+ "loss": 0.0048,
+ "step": 34048
+ },
+ {
+ "epoch": 92.77656675749319,
+ "grad_norm": 0.8253543972969055,
+ "learning_rate": 2.7241553873730953e-07,
+ "loss": 0.0109,
+ "step": 34049
+ },
+ {
+ "epoch": 92.77929155313352,
+ "grad_norm": 0.7411318421363831,
+ "learning_rate": 2.722109957716401e-07,
+ "loss": 0.0095,
+ "step": 34050
+ },
+ {
+ "epoch": 92.78201634877384,
+ "grad_norm": 0.7358511686325073,
+ "learning_rate": 2.72006528565677e-07,
+ "loss": 0.0081,
+ "step": 34051
+ },
+ {
+ "epoch": 92.78474114441417,
+ "grad_norm": 0.9968487024307251,
+ "learning_rate": 2.7180213712101444e-07,
+ "loss": 0.0145,
+ "step": 34052
+ },
+ {
+ "epoch": 92.7874659400545,
+ "grad_norm": 1.1542505025863647,
+ "learning_rate": 2.715978214392434e-07,
+ "loss": 0.0462,
+ "step": 34053
+ },
+ {
+ "epoch": 92.79019073569482,
+ "grad_norm": 1.4069948196411133,
+ "learning_rate": 2.7139358152195707e-07,
+ "loss": 0.0239,
+ "step": 34054
+ },
+ {
+ "epoch": 92.79291553133515,
+ "grad_norm": 0.8566818237304688,
+ "learning_rate": 2.711894173707419e-07,
+ "loss": 0.0224,
+ "step": 34055
+ },
+ {
+ "epoch": 92.79564032697547,
+ "grad_norm": 0.8985498547554016,
+ "learning_rate": 2.709853289871922e-07,
+ "loss": 0.0107,
+ "step": 34056
+ },
+ {
+ "epoch": 92.7983651226158,
+ "grad_norm": 1.9708691835403442,
+ "learning_rate": 2.707813163728945e-07,
+ "loss": 0.0095,
+ "step": 34057
+ },
+ {
+ "epoch": 92.80108991825612,
+ "grad_norm": 1.1173791885375977,
+ "learning_rate": 2.7057737952943863e-07,
+ "loss": 0.0189,
+ "step": 34058
+ },
+ {
+ "epoch": 92.80381471389646,
+ "grad_norm": 1.596221923828125,
+ "learning_rate": 2.703735184584133e-07,
+ "loss": 0.0632,
+ "step": 34059
+ },
+ {
+ "epoch": 92.80653950953679,
+ "grad_norm": 1.178827166557312,
+ "learning_rate": 2.701697331614062e-07,
+ "loss": 0.0125,
+ "step": 34060
+ },
+ {
+ "epoch": 92.80926430517711,
+ "grad_norm": 0.5239414572715759,
+ "learning_rate": 2.699660236400048e-07,
+ "loss": 0.0052,
+ "step": 34061
+ },
+ {
+ "epoch": 92.81198910081744,
+ "grad_norm": 0.6079800128936768,
+ "learning_rate": 2.697623898957946e-07,
+ "loss": 0.008,
+ "step": 34062
+ },
+ {
+ "epoch": 92.81471389645776,
+ "grad_norm": 0.7180619835853577,
+ "learning_rate": 2.695588319303599e-07,
+ "loss": 0.0062,
+ "step": 34063
+ },
+ {
+ "epoch": 92.8174386920981,
+ "grad_norm": 1.159970760345459,
+ "learning_rate": 2.693553497452905e-07,
+ "loss": 0.0198,
+ "step": 34064
+ },
+ {
+ "epoch": 92.82016348773843,
+ "grad_norm": 1.0249528884887695,
+ "learning_rate": 2.6915194334216743e-07,
+ "loss": 0.0576,
+ "step": 34065
+ },
+ {
+ "epoch": 92.82288828337875,
+ "grad_norm": 1.6945968866348267,
+ "learning_rate": 2.68948612722576e-07,
+ "loss": 0.0228,
+ "step": 34066
+ },
+ {
+ "epoch": 92.82561307901908,
+ "grad_norm": 0.9408406615257263,
+ "learning_rate": 2.687453578880983e-07,
+ "loss": 0.0112,
+ "step": 34067
+ },
+ {
+ "epoch": 92.8283378746594,
+ "grad_norm": 0.8788657188415527,
+ "learning_rate": 2.6854217884031974e-07,
+ "loss": 0.0108,
+ "step": 34068
+ },
+ {
+ "epoch": 92.83106267029973,
+ "grad_norm": 0.7991819977760315,
+ "learning_rate": 2.683390755808213e-07,
+ "loss": 0.0099,
+ "step": 34069
+ },
+ {
+ "epoch": 92.83378746594005,
+ "grad_norm": 0.8724727630615234,
+ "learning_rate": 2.68136048111185e-07,
+ "loss": 0.0121,
+ "step": 34070
+ },
+ {
+ "epoch": 92.83651226158038,
+ "grad_norm": 1.2786370515823364,
+ "learning_rate": 2.679330964329907e-07,
+ "loss": 0.0215,
+ "step": 34071
+ },
+ {
+ "epoch": 92.83923705722071,
+ "grad_norm": 0.6778982877731323,
+ "learning_rate": 2.677302205478216e-07,
+ "loss": 0.007,
+ "step": 34072
+ },
+ {
+ "epoch": 92.84196185286103,
+ "grad_norm": 0.6529743075370789,
+ "learning_rate": 2.675274204572564e-07,
+ "loss": 0.0071,
+ "step": 34073
+ },
+ {
+ "epoch": 92.84468664850137,
+ "grad_norm": 1.3437191247940063,
+ "learning_rate": 2.6732469616287393e-07,
+ "loss": 0.0164,
+ "step": 34074
+ },
+ {
+ "epoch": 92.84741144414168,
+ "grad_norm": 0.9631590843200684,
+ "learning_rate": 2.671220476662528e-07,
+ "loss": 0.0063,
+ "step": 34075
+ },
+ {
+ "epoch": 92.85013623978202,
+ "grad_norm": 1.0865522623062134,
+ "learning_rate": 2.66919474968973e-07,
+ "loss": 0.0125,
+ "step": 34076
+ },
+ {
+ "epoch": 92.85286103542235,
+ "grad_norm": 1.0264983177185059,
+ "learning_rate": 2.667169780726109e-07,
+ "loss": 0.0112,
+ "step": 34077
+ },
+ {
+ "epoch": 92.85558583106267,
+ "grad_norm": 0.7998203039169312,
+ "learning_rate": 2.6651455697874306e-07,
+ "loss": 0.0073,
+ "step": 34078
+ },
+ {
+ "epoch": 92.858310626703,
+ "grad_norm": 1.4524152278900146,
+ "learning_rate": 2.6631221168894603e-07,
+ "loss": 0.0267,
+ "step": 34079
+ },
+ {
+ "epoch": 92.86103542234332,
+ "grad_norm": 0.8326719999313354,
+ "learning_rate": 2.661099422047986e-07,
+ "loss": 0.0104,
+ "step": 34080
+ },
+ {
+ "epoch": 92.86376021798365,
+ "grad_norm": 1.820332646369934,
+ "learning_rate": 2.659077485278716e-07,
+ "loss": 0.0597,
+ "step": 34081
+ },
+ {
+ "epoch": 92.86648501362397,
+ "grad_norm": 1.2510004043579102,
+ "learning_rate": 2.6570563065974273e-07,
+ "loss": 0.0067,
+ "step": 34082
+ },
+ {
+ "epoch": 92.8692098092643,
+ "grad_norm": 1.1902060508728027,
+ "learning_rate": 2.655035886019841e-07,
+ "loss": 0.0291,
+ "step": 34083
+ },
+ {
+ "epoch": 92.87193460490464,
+ "grad_norm": 0.9243100881576538,
+ "learning_rate": 2.653016223561722e-07,
+ "loss": 0.0072,
+ "step": 34084
+ },
+ {
+ "epoch": 92.87465940054496,
+ "grad_norm": 0.5186290144920349,
+ "learning_rate": 2.650997319238768e-07,
+ "loss": 0.0068,
+ "step": 34085
+ },
+ {
+ "epoch": 92.87738419618529,
+ "grad_norm": 0.7031328678131104,
+ "learning_rate": 2.6489791730667125e-07,
+ "loss": 0.0083,
+ "step": 34086
+ },
+ {
+ "epoch": 92.88010899182561,
+ "grad_norm": 0.3745608627796173,
+ "learning_rate": 2.6469617850612637e-07,
+ "loss": 0.0034,
+ "step": 34087
+ },
+ {
+ "epoch": 92.88283378746594,
+ "grad_norm": 1.5017414093017578,
+ "learning_rate": 2.6449451552381653e-07,
+ "loss": 0.0114,
+ "step": 34088
+ },
+ {
+ "epoch": 92.88555858310627,
+ "grad_norm": 0.896996259689331,
+ "learning_rate": 2.642929283613094e-07,
+ "loss": 0.0084,
+ "step": 34089
+ },
+ {
+ "epoch": 92.88828337874659,
+ "grad_norm": 0.9588186740875244,
+ "learning_rate": 2.640914170201747e-07,
+ "loss": 0.0096,
+ "step": 34090
+ },
+ {
+ "epoch": 92.89100817438693,
+ "grad_norm": 0.7000792622566223,
+ "learning_rate": 2.6388998150198243e-07,
+ "loss": 0.0076,
+ "step": 34091
+ },
+ {
+ "epoch": 92.89373297002724,
+ "grad_norm": 0.6537090539932251,
+ "learning_rate": 2.6368862180830234e-07,
+ "loss": 0.0066,
+ "step": 34092
+ },
+ {
+ "epoch": 92.89645776566758,
+ "grad_norm": 0.6651500463485718,
+ "learning_rate": 2.6348733794070215e-07,
+ "loss": 0.0073,
+ "step": 34093
+ },
+ {
+ "epoch": 92.8991825613079,
+ "grad_norm": 0.49570974707603455,
+ "learning_rate": 2.632861299007483e-07,
+ "loss": 0.0039,
+ "step": 34094
+ },
+ {
+ "epoch": 92.90190735694823,
+ "grad_norm": 1.9343160390853882,
+ "learning_rate": 2.6308499769000963e-07,
+ "loss": 0.1378,
+ "step": 34095
+ },
+ {
+ "epoch": 92.90463215258856,
+ "grad_norm": 0.8064333200454712,
+ "learning_rate": 2.6288394131005147e-07,
+ "loss": 0.0093,
+ "step": 34096
+ },
+ {
+ "epoch": 92.90735694822888,
+ "grad_norm": 1.3533408641815186,
+ "learning_rate": 2.6268296076243925e-07,
+ "loss": 0.0136,
+ "step": 34097
+ },
+ {
+ "epoch": 92.91008174386921,
+ "grad_norm": 0.5536149144172668,
+ "learning_rate": 2.6248205604873955e-07,
+ "loss": 0.0059,
+ "step": 34098
+ },
+ {
+ "epoch": 92.91280653950953,
+ "grad_norm": 1.0264559984207153,
+ "learning_rate": 2.622812271705155e-07,
+ "loss": 0.0493,
+ "step": 34099
+ },
+ {
+ "epoch": 92.91553133514986,
+ "grad_norm": 1.3560484647750854,
+ "learning_rate": 2.6208047412933145e-07,
+ "loss": 0.0207,
+ "step": 34100
+ },
+ {
+ "epoch": 92.9182561307902,
+ "grad_norm": 1.9466440677642822,
+ "learning_rate": 2.618797969267528e-07,
+ "loss": 0.1012,
+ "step": 34101
+ },
+ {
+ "epoch": 92.92098092643052,
+ "grad_norm": 1.0792534351348877,
+ "learning_rate": 2.616791955643405e-07,
+ "loss": 0.0124,
+ "step": 34102
+ },
+ {
+ "epoch": 92.92370572207085,
+ "grad_norm": 0.9237636923789978,
+ "learning_rate": 2.614786700436556e-07,
+ "loss": 0.0177,
+ "step": 34103
+ },
+ {
+ "epoch": 92.92643051771117,
+ "grad_norm": 0.821398138999939,
+ "learning_rate": 2.6127822036626337e-07,
+ "loss": 0.0091,
+ "step": 34104
+ },
+ {
+ "epoch": 92.9291553133515,
+ "grad_norm": 1.5208579301834106,
+ "learning_rate": 2.610778465337216e-07,
+ "loss": 0.0187,
+ "step": 34105
+ },
+ {
+ "epoch": 92.93188010899182,
+ "grad_norm": 0.8855971693992615,
+ "learning_rate": 2.608775485475934e-07,
+ "loss": 0.0115,
+ "step": 34106
+ },
+ {
+ "epoch": 92.93460490463215,
+ "grad_norm": 0.6391450762748718,
+ "learning_rate": 2.606773264094353e-07,
+ "loss": 0.0078,
+ "step": 34107
+ },
+ {
+ "epoch": 92.93732970027249,
+ "grad_norm": 0.5215526819229126,
+ "learning_rate": 2.6047718012081056e-07,
+ "loss": 0.005,
+ "step": 34108
+ },
+ {
+ "epoch": 92.9400544959128,
+ "grad_norm": 0.40193575620651245,
+ "learning_rate": 2.602771096832757e-07,
+ "loss": 0.0037,
+ "step": 34109
+ },
+ {
+ "epoch": 92.94277929155314,
+ "grad_norm": 0.7241118550300598,
+ "learning_rate": 2.600771150983894e-07,
+ "loss": 0.0092,
+ "step": 34110
+ },
+ {
+ "epoch": 92.94550408719346,
+ "grad_norm": 0.6278384327888489,
+ "learning_rate": 2.5987719636770824e-07,
+ "loss": 0.0054,
+ "step": 34111
+ },
+ {
+ "epoch": 92.94822888283379,
+ "grad_norm": 1.139971137046814,
+ "learning_rate": 2.5967735349278877e-07,
+ "loss": 0.0318,
+ "step": 34112
+ },
+ {
+ "epoch": 92.95095367847412,
+ "grad_norm": 0.8007215857505798,
+ "learning_rate": 2.5947758647519085e-07,
+ "loss": 0.0125,
+ "step": 34113
+ },
+ {
+ "epoch": 92.95367847411444,
+ "grad_norm": 0.8602543473243713,
+ "learning_rate": 2.592778953164676e-07,
+ "loss": 0.0109,
+ "step": 34114
+ },
+ {
+ "epoch": 92.95640326975477,
+ "grad_norm": 1.0175364017486572,
+ "learning_rate": 2.590782800181746e-07,
+ "loss": 0.0056,
+ "step": 34115
+ },
+ {
+ "epoch": 92.95912806539509,
+ "grad_norm": 1.1251448392868042,
+ "learning_rate": 2.5887874058186604e-07,
+ "loss": 0.0416,
+ "step": 34116
+ },
+ {
+ "epoch": 92.96185286103542,
+ "grad_norm": 1.184984803199768,
+ "learning_rate": 2.5867927700909624e-07,
+ "loss": 0.0126,
+ "step": 34117
+ },
+ {
+ "epoch": 92.96457765667574,
+ "grad_norm": 0.6696310043334961,
+ "learning_rate": 2.584798893014195e-07,
+ "loss": 0.0068,
+ "step": 34118
+ },
+ {
+ "epoch": 92.96730245231608,
+ "grad_norm": 0.9103220701217651,
+ "learning_rate": 2.58280577460388e-07,
+ "loss": 0.0111,
+ "step": 34119
+ },
+ {
+ "epoch": 92.97002724795641,
+ "grad_norm": 0.7974044680595398,
+ "learning_rate": 2.580813414875516e-07,
+ "loss": 0.0447,
+ "step": 34120
+ },
+ {
+ "epoch": 92.97275204359673,
+ "grad_norm": 0.8990626335144043,
+ "learning_rate": 2.578821813844667e-07,
+ "loss": 0.0114,
+ "step": 34121
+ },
+ {
+ "epoch": 92.97547683923706,
+ "grad_norm": 2.251157283782959,
+ "learning_rate": 2.5768309715268116e-07,
+ "loss": 0.0257,
+ "step": 34122
+ },
+ {
+ "epoch": 92.97820163487738,
+ "grad_norm": 1.0099703073501587,
+ "learning_rate": 2.5748408879374463e-07,
+ "loss": 0.009,
+ "step": 34123
+ },
+ {
+ "epoch": 92.98092643051771,
+ "grad_norm": 1.1400960683822632,
+ "learning_rate": 2.572851563092105e-07,
+ "loss": 0.0169,
+ "step": 34124
+ },
+ {
+ "epoch": 92.98365122615803,
+ "grad_norm": 0.2285366803407669,
+ "learning_rate": 2.5708629970062516e-07,
+ "loss": 0.0031,
+ "step": 34125
+ },
+ {
+ "epoch": 92.98637602179836,
+ "grad_norm": 1.0356969833374023,
+ "learning_rate": 2.5688751896953854e-07,
+ "loss": 0.0109,
+ "step": 34126
+ },
+ {
+ "epoch": 92.9891008174387,
+ "grad_norm": 1.739640235900879,
+ "learning_rate": 2.566888141174972e-07,
+ "loss": 0.0159,
+ "step": 34127
+ },
+ {
+ "epoch": 92.99182561307902,
+ "grad_norm": 1.1138309240341187,
+ "learning_rate": 2.5649018514604995e-07,
+ "loss": 0.0167,
+ "step": 34128
+ },
+ {
+ "epoch": 92.99455040871935,
+ "grad_norm": 1.086129903793335,
+ "learning_rate": 2.5629163205674434e-07,
+ "loss": 0.0124,
+ "step": 34129
+ },
+ {
+ "epoch": 92.99727520435967,
+ "grad_norm": 1.3995270729064941,
+ "learning_rate": 2.5609315485112583e-07,
+ "loss": 0.0072,
+ "step": 34130
+ },
+ {
+ "epoch": 93.0,
+ "grad_norm": 0.969954788684845,
+ "learning_rate": 2.5589475353073987e-07,
+ "loss": 0.0081,
+ "step": 34131
+ },
+ {
+ "epoch": 93.00272479564033,
+ "grad_norm": 0.9096276164054871,
+ "learning_rate": 2.556964280971319e-07,
+ "loss": 0.0068,
+ "step": 34132
+ },
+ {
+ "epoch": 93.00544959128065,
+ "grad_norm": 1.1075332164764404,
+ "learning_rate": 2.5549817855184624e-07,
+ "loss": 0.0139,
+ "step": 34133
+ },
+ {
+ "epoch": 93.00817438692098,
+ "grad_norm": 0.9137442708015442,
+ "learning_rate": 2.5530000489642823e-07,
+ "loss": 0.0098,
+ "step": 34134
+ },
+ {
+ "epoch": 93.0108991825613,
+ "grad_norm": 1.1585898399353027,
+ "learning_rate": 2.5510190713242014e-07,
+ "loss": 0.0108,
+ "step": 34135
+ },
+ {
+ "epoch": 93.01362397820164,
+ "grad_norm": 0.9409946203231812,
+ "learning_rate": 2.5490388526136387e-07,
+ "loss": 0.0253,
+ "step": 34136
+ },
+ {
+ "epoch": 93.01634877384195,
+ "grad_norm": 1.240315318107605,
+ "learning_rate": 2.547059392848028e-07,
+ "loss": 0.038,
+ "step": 34137
+ },
+ {
+ "epoch": 93.01907356948229,
+ "grad_norm": 0.45061227679252625,
+ "learning_rate": 2.5450806920427783e-07,
+ "loss": 0.0046,
+ "step": 34138
+ },
+ {
+ "epoch": 93.02179836512262,
+ "grad_norm": 1.1712549924850464,
+ "learning_rate": 2.543102750213311e-07,
+ "loss": 0.0401,
+ "step": 34139
+ },
+ {
+ "epoch": 93.02452316076294,
+ "grad_norm": 1.021704912185669,
+ "learning_rate": 2.5411255673750137e-07,
+ "loss": 0.0803,
+ "step": 34140
+ },
+ {
+ "epoch": 93.02724795640327,
+ "grad_norm": 1.0986099243164062,
+ "learning_rate": 2.539149143543307e-07,
+ "loss": 0.0469,
+ "step": 34141
+ },
+ {
+ "epoch": 93.02997275204359,
+ "grad_norm": 0.6707022786140442,
+ "learning_rate": 2.5371734787335676e-07,
+ "loss": 0.0084,
+ "step": 34142
+ },
+ {
+ "epoch": 93.03269754768392,
+ "grad_norm": 0.8581982851028442,
+ "learning_rate": 2.535198572961173e-07,
+ "loss": 0.0725,
+ "step": 34143
+ },
+ {
+ "epoch": 93.03542234332426,
+ "grad_norm": 0.8209960460662842,
+ "learning_rate": 2.5332244262415207e-07,
+ "loss": 0.106,
+ "step": 34144
+ },
+ {
+ "epoch": 93.03814713896458,
+ "grad_norm": 1.3507858514785767,
+ "learning_rate": 2.5312510385899767e-07,
+ "loss": 0.0257,
+ "step": 34145
+ },
+ {
+ "epoch": 93.04087193460491,
+ "grad_norm": 0.7516080141067505,
+ "learning_rate": 2.5292784100219184e-07,
+ "loss": 0.0095,
+ "step": 34146
+ },
+ {
+ "epoch": 93.04359673024523,
+ "grad_norm": 0.8144369721412659,
+ "learning_rate": 2.5273065405526986e-07,
+ "loss": 0.0091,
+ "step": 34147
+ },
+ {
+ "epoch": 93.04632152588556,
+ "grad_norm": 1.0431199073791504,
+ "learning_rate": 2.525335430197673e-07,
+ "loss": 0.0155,
+ "step": 34148
+ },
+ {
+ "epoch": 93.04904632152588,
+ "grad_norm": 1.2244585752487183,
+ "learning_rate": 2.5233650789722063e-07,
+ "loss": 0.0079,
+ "step": 34149
+ },
+ {
+ "epoch": 93.05177111716621,
+ "grad_norm": 1.1061230897903442,
+ "learning_rate": 2.5213954868916425e-07,
+ "loss": 0.0077,
+ "step": 34150
+ },
+ {
+ "epoch": 93.05449591280654,
+ "grad_norm": 1.4477334022521973,
+ "learning_rate": 2.519426653971302e-07,
+ "loss": 0.0281,
+ "step": 34151
+ },
+ {
+ "epoch": 93.05722070844686,
+ "grad_norm": 1.1586636304855347,
+ "learning_rate": 2.517458580226528e-07,
+ "loss": 0.0396,
+ "step": 34152
+ },
+ {
+ "epoch": 93.0599455040872,
+ "grad_norm": 0.8863892555236816,
+ "learning_rate": 2.515491265672654e-07,
+ "loss": 0.0075,
+ "step": 34153
+ },
+ {
+ "epoch": 93.06267029972751,
+ "grad_norm": 0.7873084545135498,
+ "learning_rate": 2.513524710324999e-07,
+ "loss": 0.0091,
+ "step": 34154
+ },
+ {
+ "epoch": 93.06539509536785,
+ "grad_norm": 1.0906192064285278,
+ "learning_rate": 2.511558914198875e-07,
+ "loss": 0.0959,
+ "step": 34155
+ },
+ {
+ "epoch": 93.06811989100818,
+ "grad_norm": 0.7586634159088135,
+ "learning_rate": 2.5095938773095906e-07,
+ "loss": 0.0103,
+ "step": 34156
+ },
+ {
+ "epoch": 93.0708446866485,
+ "grad_norm": 1.3597358465194702,
+ "learning_rate": 2.5076295996724454e-07,
+ "loss": 0.009,
+ "step": 34157
+ },
+ {
+ "epoch": 93.07356948228883,
+ "grad_norm": 0.9446816444396973,
+ "learning_rate": 2.5056660813027603e-07,
+ "loss": 0.0151,
+ "step": 34158
+ },
+ {
+ "epoch": 93.07629427792915,
+ "grad_norm": 1.9231733083724976,
+ "learning_rate": 2.5037033222158003e-07,
+ "loss": 0.0436,
+ "step": 34159
+ },
+ {
+ "epoch": 93.07901907356948,
+ "grad_norm": 1.0943351984024048,
+ "learning_rate": 2.501741322426854e-07,
+ "loss": 0.0175,
+ "step": 34160
+ },
+ {
+ "epoch": 93.0817438692098,
+ "grad_norm": 1.9915615320205688,
+ "learning_rate": 2.4997800819512086e-07,
+ "loss": 0.0307,
+ "step": 34161
+ },
+ {
+ "epoch": 93.08446866485014,
+ "grad_norm": 0.5901363492012024,
+ "learning_rate": 2.497819600804152e-07,
+ "loss": 0.0045,
+ "step": 34162
+ },
+ {
+ "epoch": 93.08719346049047,
+ "grad_norm": 1.0423262119293213,
+ "learning_rate": 2.495859879000917e-07,
+ "loss": 0.0164,
+ "step": 34163
+ },
+ {
+ "epoch": 93.08991825613079,
+ "grad_norm": 0.8715385794639587,
+ "learning_rate": 2.4939009165567904e-07,
+ "loss": 0.1217,
+ "step": 34164
+ },
+ {
+ "epoch": 93.09264305177112,
+ "grad_norm": 0.9078057408332825,
+ "learning_rate": 2.491942713487039e-07,
+ "loss": 0.0183,
+ "step": 34165
+ },
+ {
+ "epoch": 93.09536784741144,
+ "grad_norm": 1.2822377681732178,
+ "learning_rate": 2.4899852698068937e-07,
+ "loss": 0.0074,
+ "step": 34166
+ },
+ {
+ "epoch": 93.09809264305177,
+ "grad_norm": 1.3165682554244995,
+ "learning_rate": 2.48802858553161e-07,
+ "loss": 0.0164,
+ "step": 34167
+ },
+ {
+ "epoch": 93.1008174386921,
+ "grad_norm": 1.077413558959961,
+ "learning_rate": 2.4860726606764086e-07,
+ "loss": 0.0846,
+ "step": 34168
+ },
+ {
+ "epoch": 93.10354223433242,
+ "grad_norm": 1.2389057874679565,
+ "learning_rate": 2.484117495256544e-07,
+ "loss": 0.0857,
+ "step": 34169
+ },
+ {
+ "epoch": 93.10626702997276,
+ "grad_norm": 0.4773244261741638,
+ "learning_rate": 2.4821630892872375e-07,
+ "loss": 0.0077,
+ "step": 34170
+ },
+ {
+ "epoch": 93.10899182561307,
+ "grad_norm": 0.9528753161430359,
+ "learning_rate": 2.48020944278371e-07,
+ "loss": 0.0131,
+ "step": 34171
+ },
+ {
+ "epoch": 93.11171662125341,
+ "grad_norm": 1.7629848718643188,
+ "learning_rate": 2.4782565557611494e-07,
+ "loss": 0.0141,
+ "step": 34172
+ },
+ {
+ "epoch": 93.11444141689373,
+ "grad_norm": 1.4175232648849487,
+ "learning_rate": 2.4763044282348105e-07,
+ "loss": 0.0506,
+ "step": 34173
+ },
+ {
+ "epoch": 93.11716621253406,
+ "grad_norm": 0.7821018695831299,
+ "learning_rate": 2.47435306021987e-07,
+ "loss": 0.0234,
+ "step": 34174
+ },
+ {
+ "epoch": 93.11989100817439,
+ "grad_norm": 1.2067729234695435,
+ "learning_rate": 2.472402451731526e-07,
+ "loss": 0.0263,
+ "step": 34175
+ },
+ {
+ "epoch": 93.12261580381471,
+ "grad_norm": 1.2662323713302612,
+ "learning_rate": 2.4704526027849675e-07,
+ "loss": 0.011,
+ "step": 34176
+ },
+ {
+ "epoch": 93.12534059945504,
+ "grad_norm": 1.5528093576431274,
+ "learning_rate": 2.4685035133953926e-07,
+ "loss": 0.0592,
+ "step": 34177
+ },
+ {
+ "epoch": 93.12806539509536,
+ "grad_norm": 1.9372915029525757,
+ "learning_rate": 2.466555183577968e-07,
+ "loss": 0.0116,
+ "step": 34178
+ },
+ {
+ "epoch": 93.1307901907357,
+ "grad_norm": 2.1289453506469727,
+ "learning_rate": 2.4646076133478803e-07,
+ "loss": 0.0197,
+ "step": 34179
+ },
+ {
+ "epoch": 93.13351498637603,
+ "grad_norm": 0.645927906036377,
+ "learning_rate": 2.4626608027202735e-07,
+ "loss": 0.0052,
+ "step": 34180
+ },
+ {
+ "epoch": 93.13623978201635,
+ "grad_norm": 0.783437967300415,
+ "learning_rate": 2.4607147517103356e-07,
+ "loss": 0.0096,
+ "step": 34181
+ },
+ {
+ "epoch": 93.13896457765668,
+ "grad_norm": 1.2901921272277832,
+ "learning_rate": 2.45876946033321e-07,
+ "loss": 0.0339,
+ "step": 34182
+ },
+ {
+ "epoch": 93.141689373297,
+ "grad_norm": 1.2975738048553467,
+ "learning_rate": 2.456824928604051e-07,
+ "loss": 0.0785,
+ "step": 34183
+ },
+ {
+ "epoch": 93.14441416893733,
+ "grad_norm": 1.4615509510040283,
+ "learning_rate": 2.454881156538003e-07,
+ "loss": 0.0196,
+ "step": 34184
+ },
+ {
+ "epoch": 93.14713896457765,
+ "grad_norm": 1.057643175125122,
+ "learning_rate": 2.4529381441501963e-07,
+ "loss": 0.0079,
+ "step": 34185
+ },
+ {
+ "epoch": 93.14986376021798,
+ "grad_norm": 0.9048334956169128,
+ "learning_rate": 2.450995891455765e-07,
+ "loss": 0.0095,
+ "step": 34186
+ },
+ {
+ "epoch": 93.15258855585832,
+ "grad_norm": 0.6493059396743774,
+ "learning_rate": 2.449054398469852e-07,
+ "loss": 0.0085,
+ "step": 34187
+ },
+ {
+ "epoch": 93.15531335149863,
+ "grad_norm": 0.4467495381832123,
+ "learning_rate": 2.4471136652075454e-07,
+ "loss": 0.004,
+ "step": 34188
+ },
+ {
+ "epoch": 93.15803814713897,
+ "grad_norm": 0.49320292472839355,
+ "learning_rate": 2.4451736916839884e-07,
+ "loss": 0.0044,
+ "step": 34189
+ },
+ {
+ "epoch": 93.16076294277929,
+ "grad_norm": 0.818321704864502,
+ "learning_rate": 2.44323447791428e-07,
+ "loss": 0.0103,
+ "step": 34190
+ },
+ {
+ "epoch": 93.16348773841962,
+ "grad_norm": 0.7792367339134216,
+ "learning_rate": 2.44129602391352e-07,
+ "loss": 0.0097,
+ "step": 34191
+ },
+ {
+ "epoch": 93.16621253405995,
+ "grad_norm": 0.9812141060829163,
+ "learning_rate": 2.439358329696795e-07,
+ "loss": 0.0081,
+ "step": 34192
+ },
+ {
+ "epoch": 93.16893732970027,
+ "grad_norm": 1.2588212490081787,
+ "learning_rate": 2.437421395279216e-07,
+ "loss": 0.0148,
+ "step": 34193
+ },
+ {
+ "epoch": 93.1716621253406,
+ "grad_norm": 1.2222744226455688,
+ "learning_rate": 2.4354852206758705e-07,
+ "loss": 0.0551,
+ "step": 34194
+ },
+ {
+ "epoch": 93.17438692098092,
+ "grad_norm": 0.9932199716567993,
+ "learning_rate": 2.4335498059018135e-07,
+ "loss": 0.0069,
+ "step": 34195
+ },
+ {
+ "epoch": 93.17711171662125,
+ "grad_norm": 0.5628964304924011,
+ "learning_rate": 2.431615150972133e-07,
+ "loss": 0.0051,
+ "step": 34196
+ },
+ {
+ "epoch": 93.17983651226157,
+ "grad_norm": 0.9807910919189453,
+ "learning_rate": 2.4296812559018835e-07,
+ "loss": 0.0095,
+ "step": 34197
+ },
+ {
+ "epoch": 93.1825613079019,
+ "grad_norm": 1.6497738361358643,
+ "learning_rate": 2.4277481207061527e-07,
+ "loss": 0.0246,
+ "step": 34198
+ },
+ {
+ "epoch": 93.18528610354224,
+ "grad_norm": 1.2041940689086914,
+ "learning_rate": 2.425815745399973e-07,
+ "loss": 0.0179,
+ "step": 34199
+ },
+ {
+ "epoch": 93.18801089918256,
+ "grad_norm": 0.9098929762840271,
+ "learning_rate": 2.4238841299983997e-07,
+ "loss": 0.0356,
+ "step": 34200
+ },
+ {
+ "epoch": 93.19073569482289,
+ "grad_norm": 1.6684560775756836,
+ "learning_rate": 2.421953274516464e-07,
+ "loss": 0.0202,
+ "step": 34201
+ },
+ {
+ "epoch": 93.19346049046321,
+ "grad_norm": 0.5783078670501709,
+ "learning_rate": 2.420023178969233e-07,
+ "loss": 0.0054,
+ "step": 34202
+ },
+ {
+ "epoch": 93.19618528610354,
+ "grad_norm": 1.1017045974731445,
+ "learning_rate": 2.418093843371716e-07,
+ "loss": 0.005,
+ "step": 34203
+ },
+ {
+ "epoch": 93.19891008174388,
+ "grad_norm": 0.9904557466506958,
+ "learning_rate": 2.416165267738946e-07,
+ "loss": 0.0131,
+ "step": 34204
+ },
+ {
+ "epoch": 93.2016348773842,
+ "grad_norm": 1.3563578128814697,
+ "learning_rate": 2.414237452085932e-07,
+ "loss": 0.1512,
+ "step": 34205
+ },
+ {
+ "epoch": 93.20435967302453,
+ "grad_norm": 0.9119817614555359,
+ "learning_rate": 2.4123103964277085e-07,
+ "loss": 0.0094,
+ "step": 34206
+ },
+ {
+ "epoch": 93.20708446866485,
+ "grad_norm": 0.6174470782279968,
+ "learning_rate": 2.410384100779262e-07,
+ "loss": 0.0069,
+ "step": 34207
+ },
+ {
+ "epoch": 93.20980926430518,
+ "grad_norm": 1.1788578033447266,
+ "learning_rate": 2.4084585651556025e-07,
+ "loss": 0.084,
+ "step": 34208
+ },
+ {
+ "epoch": 93.2125340599455,
+ "grad_norm": 0.8315308690071106,
+ "learning_rate": 2.40653378957173e-07,
+ "loss": 0.009,
+ "step": 34209
+ },
+ {
+ "epoch": 93.21525885558583,
+ "grad_norm": 0.7026253938674927,
+ "learning_rate": 2.404609774042632e-07,
+ "loss": 0.0061,
+ "step": 34210
+ },
+ {
+ "epoch": 93.21798365122616,
+ "grad_norm": 1.0459672212600708,
+ "learning_rate": 2.402686518583297e-07,
+ "loss": 0.0656,
+ "step": 34211
+ },
+ {
+ "epoch": 93.22070844686648,
+ "grad_norm": 0.4655736982822418,
+ "learning_rate": 2.4007640232086793e-07,
+ "loss": 0.0041,
+ "step": 34212
+ },
+ {
+ "epoch": 93.22343324250681,
+ "grad_norm": 1.179086446762085,
+ "learning_rate": 2.398842287933789e-07,
+ "loss": 0.0493,
+ "step": 34213
+ },
+ {
+ "epoch": 93.22615803814713,
+ "grad_norm": 1.1668204069137573,
+ "learning_rate": 2.396921312773559e-07,
+ "loss": 0.018,
+ "step": 34214
+ },
+ {
+ "epoch": 93.22888283378747,
+ "grad_norm": 0.7294105887413025,
+ "learning_rate": 2.395001097742977e-07,
+ "loss": 0.0068,
+ "step": 34215
+ },
+ {
+ "epoch": 93.2316076294278,
+ "grad_norm": 0.3847408890724182,
+ "learning_rate": 2.393081642856976e-07,
+ "loss": 0.0042,
+ "step": 34216
+ },
+ {
+ "epoch": 93.23433242506812,
+ "grad_norm": 0.8578014969825745,
+ "learning_rate": 2.39116294813051e-07,
+ "loss": 0.0123,
+ "step": 34217
+ },
+ {
+ "epoch": 93.23705722070845,
+ "grad_norm": 1.2481307983398438,
+ "learning_rate": 2.3892450135785447e-07,
+ "loss": 0.0675,
+ "step": 34218
+ },
+ {
+ "epoch": 93.23978201634877,
+ "grad_norm": 0.7961851358413696,
+ "learning_rate": 2.3873278392159806e-07,
+ "loss": 0.0085,
+ "step": 34219
+ },
+ {
+ "epoch": 93.2425068119891,
+ "grad_norm": 1.0871185064315796,
+ "learning_rate": 2.385411425057782e-07,
+ "loss": 0.0119,
+ "step": 34220
+ },
+ {
+ "epoch": 93.24523160762942,
+ "grad_norm": 0.41980114579200745,
+ "learning_rate": 2.3834957711188378e-07,
+ "loss": 0.0065,
+ "step": 34221
+ },
+ {
+ "epoch": 93.24795640326975,
+ "grad_norm": 1.0079010725021362,
+ "learning_rate": 2.3815808774141025e-07,
+ "loss": 0.0118,
+ "step": 34222
+ },
+ {
+ "epoch": 93.25068119891009,
+ "grad_norm": 0.7724305987358093,
+ "learning_rate": 2.379666743958464e-07,
+ "loss": 0.0089,
+ "step": 34223
+ },
+ {
+ "epoch": 93.2534059945504,
+ "grad_norm": 2.7323899269104004,
+ "learning_rate": 2.3777533707668444e-07,
+ "loss": 0.0714,
+ "step": 34224
+ },
+ {
+ "epoch": 93.25613079019074,
+ "grad_norm": 1.4223284721374512,
+ "learning_rate": 2.375840757854131e-07,
+ "loss": 0.0399,
+ "step": 34225
+ },
+ {
+ "epoch": 93.25885558583106,
+ "grad_norm": 0.8273661732673645,
+ "learning_rate": 2.3739289052352344e-07,
+ "loss": 0.0115,
+ "step": 34226
+ },
+ {
+ "epoch": 93.26158038147139,
+ "grad_norm": 1.028301477432251,
+ "learning_rate": 2.3720178129250425e-07,
+ "loss": 0.0735,
+ "step": 34227
+ },
+ {
+ "epoch": 93.26430517711172,
+ "grad_norm": 1.3300575017929077,
+ "learning_rate": 2.3701074809384327e-07,
+ "loss": 0.0376,
+ "step": 34228
+ },
+ {
+ "epoch": 93.26702997275204,
+ "grad_norm": 1.2902735471725464,
+ "learning_rate": 2.3681979092902706e-07,
+ "loss": 0.0263,
+ "step": 34229
+ },
+ {
+ "epoch": 93.26975476839237,
+ "grad_norm": 0.9758832454681396,
+ "learning_rate": 2.3662890979954556e-07,
+ "loss": 0.0053,
+ "step": 34230
+ },
+ {
+ "epoch": 93.2724795640327,
+ "grad_norm": 0.7044846415519714,
+ "learning_rate": 2.36438104706882e-07,
+ "loss": 0.0071,
+ "step": 34231
+ },
+ {
+ "epoch": 93.27520435967303,
+ "grad_norm": 0.9053559303283691,
+ "learning_rate": 2.362473756525263e-07,
+ "loss": 0.0075,
+ "step": 34232
+ },
+ {
+ "epoch": 93.27792915531334,
+ "grad_norm": 0.7184256911277771,
+ "learning_rate": 2.3605672263796066e-07,
+ "loss": 0.0086,
+ "step": 34233
+ },
+ {
+ "epoch": 93.28065395095368,
+ "grad_norm": 1.7119741439819336,
+ "learning_rate": 2.358661456646716e-07,
+ "loss": 0.0396,
+ "step": 34234
+ },
+ {
+ "epoch": 93.28337874659401,
+ "grad_norm": 0.752848207950592,
+ "learning_rate": 2.3567564473414239e-07,
+ "loss": 0.0068,
+ "step": 34235
+ },
+ {
+ "epoch": 93.28610354223433,
+ "grad_norm": 1.837430477142334,
+ "learning_rate": 2.3548521984785745e-07,
+ "loss": 0.0103,
+ "step": 34236
+ },
+ {
+ "epoch": 93.28882833787466,
+ "grad_norm": 1.7726030349731445,
+ "learning_rate": 2.3529487100729774e-07,
+ "loss": 0.0402,
+ "step": 34237
+ },
+ {
+ "epoch": 93.29155313351498,
+ "grad_norm": 0.9521014094352722,
+ "learning_rate": 2.3510459821394883e-07,
+ "loss": 0.0067,
+ "step": 34238
+ },
+ {
+ "epoch": 93.29427792915531,
+ "grad_norm": 1.5670232772827148,
+ "learning_rate": 2.3491440146929168e-07,
+ "loss": 0.0753,
+ "step": 34239
+ },
+ {
+ "epoch": 93.29700272479565,
+ "grad_norm": 2.6013031005859375,
+ "learning_rate": 2.3472428077480625e-07,
+ "loss": 0.0235,
+ "step": 34240
+ },
+ {
+ "epoch": 93.29972752043597,
+ "grad_norm": 1.053096055984497,
+ "learning_rate": 2.3453423613197245e-07,
+ "loss": 0.0078,
+ "step": 34241
+ },
+ {
+ "epoch": 93.3024523160763,
+ "grad_norm": 0.971339225769043,
+ "learning_rate": 2.3434426754227358e-07,
+ "loss": 0.0186,
+ "step": 34242
+ },
+ {
+ "epoch": 93.30517711171662,
+ "grad_norm": 2.4354350566864014,
+ "learning_rate": 2.3415437500718618e-07,
+ "loss": 0.0184,
+ "step": 34243
+ },
+ {
+ "epoch": 93.30790190735695,
+ "grad_norm": 1.3257049322128296,
+ "learning_rate": 2.3396455852819133e-07,
+ "loss": 0.0583,
+ "step": 34244
+ },
+ {
+ "epoch": 93.31062670299727,
+ "grad_norm": 0.6171401739120483,
+ "learning_rate": 2.337748181067645e-07,
+ "loss": 0.0101,
+ "step": 34245
+ },
+ {
+ "epoch": 93.3133514986376,
+ "grad_norm": 1.3234282732009888,
+ "learning_rate": 2.335851537443856e-07,
+ "loss": 0.1002,
+ "step": 34246
+ },
+ {
+ "epoch": 93.31607629427793,
+ "grad_norm": 1.1273252964019775,
+ "learning_rate": 2.3339556544253238e-07,
+ "loss": 0.0224,
+ "step": 34247
+ },
+ {
+ "epoch": 93.31880108991825,
+ "grad_norm": 1.621555209159851,
+ "learning_rate": 2.332060532026792e-07,
+ "loss": 0.048,
+ "step": 34248
+ },
+ {
+ "epoch": 93.32152588555859,
+ "grad_norm": 0.8923760652542114,
+ "learning_rate": 2.3301661702630153e-07,
+ "loss": 0.0067,
+ "step": 34249
+ },
+ {
+ "epoch": 93.3242506811989,
+ "grad_norm": 0.865920901298523,
+ "learning_rate": 2.3282725691487817e-07,
+ "loss": 0.0071,
+ "step": 34250
+ },
+ {
+ "epoch": 93.32697547683924,
+ "grad_norm": 0.7809281349182129,
+ "learning_rate": 2.3263797286988132e-07,
+ "loss": 0.0083,
+ "step": 34251
+ },
+ {
+ "epoch": 93.32970027247957,
+ "grad_norm": 0.7757065296173096,
+ "learning_rate": 2.3244876489278534e-07,
+ "loss": 0.012,
+ "step": 34252
+ },
+ {
+ "epoch": 93.33242506811989,
+ "grad_norm": 0.9234596490859985,
+ "learning_rate": 2.3225963298506348e-07,
+ "loss": 0.0138,
+ "step": 34253
+ },
+ {
+ "epoch": 93.33514986376022,
+ "grad_norm": 0.6880778074264526,
+ "learning_rate": 2.3207057714819015e-07,
+ "loss": 0.006,
+ "step": 34254
+ },
+ {
+ "epoch": 93.33787465940054,
+ "grad_norm": 0.7625666856765747,
+ "learning_rate": 2.3188159738363637e-07,
+ "loss": 0.0103,
+ "step": 34255
+ },
+ {
+ "epoch": 93.34059945504087,
+ "grad_norm": 0.8392622470855713,
+ "learning_rate": 2.316926936928754e-07,
+ "loss": 0.0252,
+ "step": 34256
+ },
+ {
+ "epoch": 93.34332425068119,
+ "grad_norm": 0.49502187967300415,
+ "learning_rate": 2.3150386607737497e-07,
+ "loss": 0.0042,
+ "step": 34257
+ },
+ {
+ "epoch": 93.34604904632153,
+ "grad_norm": 0.5480369925498962,
+ "learning_rate": 2.3131511453861056e-07,
+ "loss": 0.005,
+ "step": 34258
+ },
+ {
+ "epoch": 93.34877384196186,
+ "grad_norm": 1.2849671840667725,
+ "learning_rate": 2.3112643907804877e-07,
+ "loss": 0.0087,
+ "step": 34259
+ },
+ {
+ "epoch": 93.35149863760218,
+ "grad_norm": 0.5738608241081238,
+ "learning_rate": 2.3093783969715954e-07,
+ "loss": 0.0049,
+ "step": 34260
+ },
+ {
+ "epoch": 93.35422343324251,
+ "grad_norm": 1.627978801727295,
+ "learning_rate": 2.3074931639741172e-07,
+ "loss": 0.0178,
+ "step": 34261
+ },
+ {
+ "epoch": 93.35694822888283,
+ "grad_norm": 0.5682874917984009,
+ "learning_rate": 2.305608691802741e-07,
+ "loss": 0.0046,
+ "step": 34262
+ },
+ {
+ "epoch": 93.35967302452316,
+ "grad_norm": 0.46626681089401245,
+ "learning_rate": 2.303724980472144e-07,
+ "loss": 0.0026,
+ "step": 34263
+ },
+ {
+ "epoch": 93.3623978201635,
+ "grad_norm": 0.7525758147239685,
+ "learning_rate": 2.3018420299969924e-07,
+ "loss": 0.0091,
+ "step": 34264
+ },
+ {
+ "epoch": 93.36512261580381,
+ "grad_norm": 0.6278699040412903,
+ "learning_rate": 2.299959840391941e-07,
+ "loss": 0.007,
+ "step": 34265
+ },
+ {
+ "epoch": 93.36784741144415,
+ "grad_norm": 0.817343533039093,
+ "learning_rate": 2.298078411671656e-07,
+ "loss": 0.0155,
+ "step": 34266
+ },
+ {
+ "epoch": 93.37057220708446,
+ "grad_norm": 0.7387863993644714,
+ "learning_rate": 2.2961977438508033e-07,
+ "loss": 0.0344,
+ "step": 34267
+ },
+ {
+ "epoch": 93.3732970027248,
+ "grad_norm": 1.9696147441864014,
+ "learning_rate": 2.2943178369440155e-07,
+ "loss": 0.0981,
+ "step": 34268
+ },
+ {
+ "epoch": 93.37602179836512,
+ "grad_norm": 1.2381324768066406,
+ "learning_rate": 2.292438690965937e-07,
+ "loss": 0.0665,
+ "step": 34269
+ },
+ {
+ "epoch": 93.37874659400545,
+ "grad_norm": 1.163172960281372,
+ "learning_rate": 2.2905603059311998e-07,
+ "loss": 0.0164,
+ "step": 34270
+ },
+ {
+ "epoch": 93.38147138964578,
+ "grad_norm": 0.8330952525138855,
+ "learning_rate": 2.2886826818544262e-07,
+ "loss": 0.0078,
+ "step": 34271
+ },
+ {
+ "epoch": 93.3841961852861,
+ "grad_norm": 0.3713056743144989,
+ "learning_rate": 2.2868058187502595e-07,
+ "loss": 0.0034,
+ "step": 34272
+ },
+ {
+ "epoch": 93.38692098092643,
+ "grad_norm": 0.6103315949440002,
+ "learning_rate": 2.2849297166332996e-07,
+ "loss": 0.0068,
+ "step": 34273
+ },
+ {
+ "epoch": 93.38964577656675,
+ "grad_norm": 1.6876039505004883,
+ "learning_rate": 2.2830543755181566e-07,
+ "loss": 0.0856,
+ "step": 34274
+ },
+ {
+ "epoch": 93.39237057220708,
+ "grad_norm": 0.7601882815361023,
+ "learning_rate": 2.2811797954194527e-07,
+ "loss": 0.0068,
+ "step": 34275
+ },
+ {
+ "epoch": 93.39509536784742,
+ "grad_norm": 1.1490607261657715,
+ "learning_rate": 2.2793059763517756e-07,
+ "loss": 0.0207,
+ "step": 34276
+ },
+ {
+ "epoch": 93.39782016348774,
+ "grad_norm": 1.6166050434112549,
+ "learning_rate": 2.277432918329703e-07,
+ "loss": 0.0285,
+ "step": 34277
+ },
+ {
+ "epoch": 93.40054495912807,
+ "grad_norm": 1.0487051010131836,
+ "learning_rate": 2.2755606213678448e-07,
+ "loss": 0.0149,
+ "step": 34278
+ },
+ {
+ "epoch": 93.40326975476839,
+ "grad_norm": 0.7804040312767029,
+ "learning_rate": 2.2736890854807792e-07,
+ "loss": 0.0068,
+ "step": 34279
+ },
+ {
+ "epoch": 93.40599455040872,
+ "grad_norm": 1.4695430994033813,
+ "learning_rate": 2.2718183106830826e-07,
+ "loss": 0.0527,
+ "step": 34280
+ },
+ {
+ "epoch": 93.40871934604904,
+ "grad_norm": 1.9331098794937134,
+ "learning_rate": 2.2699482969892994e-07,
+ "loss": 0.0095,
+ "step": 34281
+ },
+ {
+ "epoch": 93.41144414168937,
+ "grad_norm": 0.8509560823440552,
+ "learning_rate": 2.2680790444140287e-07,
+ "loss": 0.0736,
+ "step": 34282
+ },
+ {
+ "epoch": 93.4141689373297,
+ "grad_norm": 1.0328383445739746,
+ "learning_rate": 2.2662105529718036e-07,
+ "loss": 0.0083,
+ "step": 34283
+ },
+ {
+ "epoch": 93.41689373297002,
+ "grad_norm": 1.4631651639938354,
+ "learning_rate": 2.2643428226772013e-07,
+ "loss": 0.0151,
+ "step": 34284
+ },
+ {
+ "epoch": 93.41961852861036,
+ "grad_norm": 0.5373004674911499,
+ "learning_rate": 2.2624758535447432e-07,
+ "loss": 0.0055,
+ "step": 34285
+ },
+ {
+ "epoch": 93.42234332425068,
+ "grad_norm": 2.9186387062072754,
+ "learning_rate": 2.2606096455889736e-07,
+ "loss": 0.0651,
+ "step": 34286
+ },
+ {
+ "epoch": 93.42506811989101,
+ "grad_norm": 0.7010611295700073,
+ "learning_rate": 2.2587441988244362e-07,
+ "loss": 0.0074,
+ "step": 34287
+ },
+ {
+ "epoch": 93.42779291553134,
+ "grad_norm": 0.9967067837715149,
+ "learning_rate": 2.256879513265653e-07,
+ "loss": 0.0081,
+ "step": 34288
+ },
+ {
+ "epoch": 93.43051771117166,
+ "grad_norm": 0.6058615446090698,
+ "learning_rate": 2.2550155889271453e-07,
+ "loss": 0.0099,
+ "step": 34289
+ },
+ {
+ "epoch": 93.433242506812,
+ "grad_norm": 0.6318084597587585,
+ "learning_rate": 2.2531524258234238e-07,
+ "loss": 0.0087,
+ "step": 34290
+ },
+ {
+ "epoch": 93.43596730245231,
+ "grad_norm": 1.476667881011963,
+ "learning_rate": 2.2512900239690104e-07,
+ "loss": 0.0748,
+ "step": 34291
+ },
+ {
+ "epoch": 93.43869209809264,
+ "grad_norm": 0.9258168935775757,
+ "learning_rate": 2.2494283833784047e-07,
+ "loss": 0.0049,
+ "step": 34292
+ },
+ {
+ "epoch": 93.44141689373296,
+ "grad_norm": 2.0368564128875732,
+ "learning_rate": 2.247567504066106e-07,
+ "loss": 0.0295,
+ "step": 34293
+ },
+ {
+ "epoch": 93.4441416893733,
+ "grad_norm": 1.3794611692428589,
+ "learning_rate": 2.2457073860466028e-07,
+ "loss": 0.0158,
+ "step": 34294
+ },
+ {
+ "epoch": 93.44686648501363,
+ "grad_norm": 0.5099110007286072,
+ "learning_rate": 2.2438480293343833e-07,
+ "loss": 0.0047,
+ "step": 34295
+ },
+ {
+ "epoch": 93.44959128065395,
+ "grad_norm": 0.8633345365524292,
+ "learning_rate": 2.2419894339439364e-07,
+ "loss": 0.0367,
+ "step": 34296
+ },
+ {
+ "epoch": 93.45231607629428,
+ "grad_norm": 1.4345427751541138,
+ "learning_rate": 2.2401315998897166e-07,
+ "loss": 0.0134,
+ "step": 34297
+ },
+ {
+ "epoch": 93.4550408719346,
+ "grad_norm": 0.6072905659675598,
+ "learning_rate": 2.2382745271862128e-07,
+ "loss": 0.0072,
+ "step": 34298
+ },
+ {
+ "epoch": 93.45776566757493,
+ "grad_norm": 1.942396879196167,
+ "learning_rate": 2.2364182158478797e-07,
+ "loss": 0.0622,
+ "step": 34299
+ },
+ {
+ "epoch": 93.46049046321527,
+ "grad_norm": 1.0904335975646973,
+ "learning_rate": 2.2345626658891728e-07,
+ "loss": 0.0136,
+ "step": 34300
+ },
+ {
+ "epoch": 93.46321525885558,
+ "grad_norm": 1.370275855064392,
+ "learning_rate": 2.232707877324547e-07,
+ "loss": 0.1147,
+ "step": 34301
+ },
+ {
+ "epoch": 93.46594005449592,
+ "grad_norm": 1.7252029180526733,
+ "learning_rate": 2.2308538501684464e-07,
+ "loss": 0.009,
+ "step": 34302
+ },
+ {
+ "epoch": 93.46866485013624,
+ "grad_norm": 0.7405182123184204,
+ "learning_rate": 2.2290005844353145e-07,
+ "loss": 0.0068,
+ "step": 34303
+ },
+ {
+ "epoch": 93.47138964577657,
+ "grad_norm": 2.0520033836364746,
+ "learning_rate": 2.227148080139574e-07,
+ "loss": 0.0126,
+ "step": 34304
+ },
+ {
+ "epoch": 93.47411444141689,
+ "grad_norm": 0.9709004759788513,
+ "learning_rate": 2.2252963372956572e-07,
+ "loss": 0.0127,
+ "step": 34305
+ },
+ {
+ "epoch": 93.47683923705722,
+ "grad_norm": 1.8223763704299927,
+ "learning_rate": 2.2234453559179858e-07,
+ "loss": 0.0204,
+ "step": 34306
+ },
+ {
+ "epoch": 93.47956403269755,
+ "grad_norm": 1.0468248128890991,
+ "learning_rate": 2.221595136020971e-07,
+ "loss": 0.018,
+ "step": 34307
+ },
+ {
+ "epoch": 93.48228882833787,
+ "grad_norm": 1.2463645935058594,
+ "learning_rate": 2.219745677619034e-07,
+ "loss": 0.0172,
+ "step": 34308
+ },
+ {
+ "epoch": 93.4850136239782,
+ "grad_norm": 1.6375960111618042,
+ "learning_rate": 2.217896980726575e-07,
+ "loss": 0.0298,
+ "step": 34309
+ },
+ {
+ "epoch": 93.48773841961852,
+ "grad_norm": 0.5957671999931335,
+ "learning_rate": 2.2160490453579707e-07,
+ "loss": 0.007,
+ "step": 34310
+ },
+ {
+ "epoch": 93.49046321525886,
+ "grad_norm": 0.946688711643219,
+ "learning_rate": 2.2142018715276437e-07,
+ "loss": 0.0304,
+ "step": 34311
+ },
+ {
+ "epoch": 93.49318801089919,
+ "grad_norm": 0.9985805749893188,
+ "learning_rate": 2.2123554592499598e-07,
+ "loss": 0.0309,
+ "step": 34312
+ },
+ {
+ "epoch": 93.49591280653951,
+ "grad_norm": 1.2689807415008545,
+ "learning_rate": 2.2105098085393073e-07,
+ "loss": 0.0084,
+ "step": 34313
+ },
+ {
+ "epoch": 93.49863760217984,
+ "grad_norm": 1.141046166419983,
+ "learning_rate": 2.208664919410053e-07,
+ "loss": 0.012,
+ "step": 34314
+ },
+ {
+ "epoch": 93.50136239782016,
+ "grad_norm": 0.30538612604141235,
+ "learning_rate": 2.2068207918765738e-07,
+ "loss": 0.004,
+ "step": 34315
+ },
+ {
+ "epoch": 93.50408719346049,
+ "grad_norm": 1.2148454189300537,
+ "learning_rate": 2.2049774259532253e-07,
+ "loss": 0.0316,
+ "step": 34316
+ },
+ {
+ "epoch": 93.50681198910081,
+ "grad_norm": 0.8840599656105042,
+ "learning_rate": 2.2031348216543514e-07,
+ "loss": 0.0302,
+ "step": 34317
+ },
+ {
+ "epoch": 93.50953678474114,
+ "grad_norm": 1.636589765548706,
+ "learning_rate": 2.2012929789943183e-07,
+ "loss": 0.0061,
+ "step": 34318
+ },
+ {
+ "epoch": 93.51226158038148,
+ "grad_norm": 0.7291244268417358,
+ "learning_rate": 2.1994518979874812e-07,
+ "loss": 0.0085,
+ "step": 34319
+ },
+ {
+ "epoch": 93.5149863760218,
+ "grad_norm": 0.9252595901489258,
+ "learning_rate": 2.197611578648162e-07,
+ "loss": 0.0062,
+ "step": 34320
+ },
+ {
+ "epoch": 93.51771117166213,
+ "grad_norm": 0.6836085319519043,
+ "learning_rate": 2.195772020990705e-07,
+ "loss": 0.0088,
+ "step": 34321
+ },
+ {
+ "epoch": 93.52043596730245,
+ "grad_norm": 0.7615414261817932,
+ "learning_rate": 2.19393322502941e-07,
+ "loss": 0.0063,
+ "step": 34322
+ },
+ {
+ "epoch": 93.52316076294278,
+ "grad_norm": 2.2111921310424805,
+ "learning_rate": 2.1920951907786204e-07,
+ "loss": 0.1153,
+ "step": 34323
+ },
+ {
+ "epoch": 93.52588555858311,
+ "grad_norm": 1.792403221130371,
+ "learning_rate": 2.1902579182526584e-07,
+ "loss": 0.0074,
+ "step": 34324
+ },
+ {
+ "epoch": 93.52861035422343,
+ "grad_norm": 1.31396484375,
+ "learning_rate": 2.1884214074658127e-07,
+ "loss": 0.0183,
+ "step": 34325
+ },
+ {
+ "epoch": 93.53133514986376,
+ "grad_norm": 1.1603330373764038,
+ "learning_rate": 2.186585658432383e-07,
+ "loss": 0.0897,
+ "step": 34326
+ },
+ {
+ "epoch": 93.53405994550408,
+ "grad_norm": 0.853361964225769,
+ "learning_rate": 2.18475067116668e-07,
+ "loss": 0.0129,
+ "step": 34327
+ },
+ {
+ "epoch": 93.53678474114442,
+ "grad_norm": 0.5597491264343262,
+ "learning_rate": 2.1829164456829922e-07,
+ "loss": 0.0058,
+ "step": 34328
+ },
+ {
+ "epoch": 93.53950953678473,
+ "grad_norm": 1.169508457183838,
+ "learning_rate": 2.181082981995608e-07,
+ "loss": 0.0073,
+ "step": 34329
+ },
+ {
+ "epoch": 93.54223433242507,
+ "grad_norm": 1.3149635791778564,
+ "learning_rate": 2.1792502801187944e-07,
+ "loss": 0.0085,
+ "step": 34330
+ },
+ {
+ "epoch": 93.5449591280654,
+ "grad_norm": 0.5225382447242737,
+ "learning_rate": 2.177418340066839e-07,
+ "loss": 0.0076,
+ "step": 34331
+ },
+ {
+ "epoch": 93.54768392370572,
+ "grad_norm": 0.7100327014923096,
+ "learning_rate": 2.1755871618539981e-07,
+ "loss": 0.0081,
+ "step": 34332
+ },
+ {
+ "epoch": 93.55040871934605,
+ "grad_norm": 1.0554592609405518,
+ "learning_rate": 2.1737567454945374e-07,
+ "loss": 0.0095,
+ "step": 34333
+ },
+ {
+ "epoch": 93.55313351498637,
+ "grad_norm": 0.6501619815826416,
+ "learning_rate": 2.1719270910027013e-07,
+ "loss": 0.0109,
+ "step": 34334
+ },
+ {
+ "epoch": 93.5558583106267,
+ "grad_norm": 1.623820424079895,
+ "learning_rate": 2.1700981983927448e-07,
+ "loss": 0.0161,
+ "step": 34335
+ },
+ {
+ "epoch": 93.55858310626704,
+ "grad_norm": 0.7059155702590942,
+ "learning_rate": 2.1682700676789236e-07,
+ "loss": 0.0069,
+ "step": 34336
+ },
+ {
+ "epoch": 93.56130790190736,
+ "grad_norm": 1.1910474300384521,
+ "learning_rate": 2.1664426988754705e-07,
+ "loss": 0.029,
+ "step": 34337
+ },
+ {
+ "epoch": 93.56403269754769,
+ "grad_norm": 1.4068326950073242,
+ "learning_rate": 2.1646160919966076e-07,
+ "loss": 0.0491,
+ "step": 34338
+ },
+ {
+ "epoch": 93.566757493188,
+ "grad_norm": 1.3425408601760864,
+ "learning_rate": 2.1627902470565675e-07,
+ "loss": 0.1251,
+ "step": 34339
+ },
+ {
+ "epoch": 93.56948228882834,
+ "grad_norm": 0.7984750270843506,
+ "learning_rate": 2.1609651640695727e-07,
+ "loss": 0.0103,
+ "step": 34340
+ },
+ {
+ "epoch": 93.57220708446866,
+ "grad_norm": 5.13101863861084,
+ "learning_rate": 2.159140843049834e-07,
+ "loss": 0.0151,
+ "step": 34341
+ },
+ {
+ "epoch": 93.57493188010899,
+ "grad_norm": 0.6806535720825195,
+ "learning_rate": 2.1573172840115397e-07,
+ "loss": 0.0055,
+ "step": 34342
+ },
+ {
+ "epoch": 93.57765667574932,
+ "grad_norm": 1.1963005065917969,
+ "learning_rate": 2.155494486968923e-07,
+ "loss": 0.0294,
+ "step": 34343
+ },
+ {
+ "epoch": 93.58038147138964,
+ "grad_norm": 0.8902276754379272,
+ "learning_rate": 2.153672451936173e-07,
+ "loss": 0.1397,
+ "step": 34344
+ },
+ {
+ "epoch": 93.58310626702998,
+ "grad_norm": 0.9232585430145264,
+ "learning_rate": 2.1518511789274666e-07,
+ "loss": 0.0105,
+ "step": 34345
+ },
+ {
+ "epoch": 93.5858310626703,
+ "grad_norm": 3.7436256408691406,
+ "learning_rate": 2.1500306679569816e-07,
+ "loss": 0.011,
+ "step": 34346
+ },
+ {
+ "epoch": 93.58855585831063,
+ "grad_norm": 1.1941624879837036,
+ "learning_rate": 2.1482109190389178e-07,
+ "loss": 0.0199,
+ "step": 34347
+ },
+ {
+ "epoch": 93.59128065395096,
+ "grad_norm": 1.2657442092895508,
+ "learning_rate": 2.1463919321874415e-07,
+ "loss": 0.0496,
+ "step": 34348
+ },
+ {
+ "epoch": 93.59400544959128,
+ "grad_norm": 1.19292151927948,
+ "learning_rate": 2.1445737074167194e-07,
+ "loss": 0.0523,
+ "step": 34349
+ },
+ {
+ "epoch": 93.59673024523161,
+ "grad_norm": 1.3414533138275146,
+ "learning_rate": 2.1427562447408957e-07,
+ "loss": 0.0329,
+ "step": 34350
+ },
+ {
+ "epoch": 93.59945504087193,
+ "grad_norm": 0.7344908118247986,
+ "learning_rate": 2.1409395441741477e-07,
+ "loss": 0.0078,
+ "step": 34351
+ },
+ {
+ "epoch": 93.60217983651226,
+ "grad_norm": 1.1706393957138062,
+ "learning_rate": 2.1391236057305974e-07,
+ "loss": 0.0184,
+ "step": 34352
+ },
+ {
+ "epoch": 93.60490463215258,
+ "grad_norm": 1.3383913040161133,
+ "learning_rate": 2.1373084294244118e-07,
+ "loss": 0.0116,
+ "step": 34353
+ },
+ {
+ "epoch": 93.60762942779292,
+ "grad_norm": 0.7016202211380005,
+ "learning_rate": 2.1354940152697234e-07,
+ "loss": 0.0054,
+ "step": 34354
+ },
+ {
+ "epoch": 93.61035422343325,
+ "grad_norm": 0.6102537512779236,
+ "learning_rate": 2.1336803632806436e-07,
+ "loss": 0.0074,
+ "step": 34355
+ },
+ {
+ "epoch": 93.61307901907357,
+ "grad_norm": 0.8468412160873413,
+ "learning_rate": 2.1318674734713273e-07,
+ "loss": 0.0479,
+ "step": 34356
+ },
+ {
+ "epoch": 93.6158038147139,
+ "grad_norm": 1.824342966079712,
+ "learning_rate": 2.130055345855875e-07,
+ "loss": 0.0611,
+ "step": 34357
+ },
+ {
+ "epoch": 93.61852861035422,
+ "grad_norm": 0.5660526156425476,
+ "learning_rate": 2.1282439804483966e-07,
+ "loss": 0.0051,
+ "step": 34358
+ },
+ {
+ "epoch": 93.62125340599455,
+ "grad_norm": 0.9291568994522095,
+ "learning_rate": 2.126433377263004e-07,
+ "loss": 0.0787,
+ "step": 34359
+ },
+ {
+ "epoch": 93.62397820163488,
+ "grad_norm": 0.7242770791053772,
+ "learning_rate": 2.1246235363137968e-07,
+ "loss": 0.0062,
+ "step": 34360
+ },
+ {
+ "epoch": 93.6267029972752,
+ "grad_norm": 1.3262603282928467,
+ "learning_rate": 2.1228144576148857e-07,
+ "loss": 0.0177,
+ "step": 34361
+ },
+ {
+ "epoch": 93.62942779291554,
+ "grad_norm": 1.0279886722564697,
+ "learning_rate": 2.1210061411803262e-07,
+ "loss": 0.0153,
+ "step": 34362
+ },
+ {
+ "epoch": 93.63215258855585,
+ "grad_norm": 0.9240914583206177,
+ "learning_rate": 2.1191985870242293e-07,
+ "loss": 0.0136,
+ "step": 34363
+ },
+ {
+ "epoch": 93.63487738419619,
+ "grad_norm": 0.6163058876991272,
+ "learning_rate": 2.1173917951606615e-07,
+ "loss": 0.0031,
+ "step": 34364
+ },
+ {
+ "epoch": 93.6376021798365,
+ "grad_norm": 1.031180500984192,
+ "learning_rate": 2.1155857656037004e-07,
+ "loss": 0.0075,
+ "step": 34365
+ },
+ {
+ "epoch": 93.64032697547684,
+ "grad_norm": 0.5382005572319031,
+ "learning_rate": 2.1137804983674014e-07,
+ "loss": 0.0044,
+ "step": 34366
+ },
+ {
+ "epoch": 93.64305177111717,
+ "grad_norm": 1.4077738523483276,
+ "learning_rate": 2.1119759934658313e-07,
+ "loss": 0.027,
+ "step": 34367
+ },
+ {
+ "epoch": 93.64577656675749,
+ "grad_norm": 1.1224299669265747,
+ "learning_rate": 2.110172250913045e-07,
+ "loss": 0.007,
+ "step": 34368
+ },
+ {
+ "epoch": 93.64850136239782,
+ "grad_norm": 1.5455721616744995,
+ "learning_rate": 2.1083692707230763e-07,
+ "loss": 0.0295,
+ "step": 34369
+ },
+ {
+ "epoch": 93.65122615803814,
+ "grad_norm": 0.5472944974899292,
+ "learning_rate": 2.1065670529099914e-07,
+ "loss": 0.0053,
+ "step": 34370
+ },
+ {
+ "epoch": 93.65395095367847,
+ "grad_norm": 0.704287052154541,
+ "learning_rate": 2.1047655974878012e-07,
+ "loss": 0.0052,
+ "step": 34371
+ },
+ {
+ "epoch": 93.65667574931881,
+ "grad_norm": 1.4617345333099365,
+ "learning_rate": 2.10296490447055e-07,
+ "loss": 0.0415,
+ "step": 34372
+ },
+ {
+ "epoch": 93.65940054495913,
+ "grad_norm": 0.9363269209861755,
+ "learning_rate": 2.1011649738722606e-07,
+ "loss": 0.0099,
+ "step": 34373
+ },
+ {
+ "epoch": 93.66212534059946,
+ "grad_norm": 1.4094462394714355,
+ "learning_rate": 2.0993658057069432e-07,
+ "loss": 0.0114,
+ "step": 34374
+ },
+ {
+ "epoch": 93.66485013623978,
+ "grad_norm": 1.3266812562942505,
+ "learning_rate": 2.0975673999886093e-07,
+ "loss": 0.0116,
+ "step": 34375
+ },
+ {
+ "epoch": 93.66757493188011,
+ "grad_norm": 0.6909528374671936,
+ "learning_rate": 2.0957697567312807e-07,
+ "loss": 0.0103,
+ "step": 34376
+ },
+ {
+ "epoch": 93.67029972752043,
+ "grad_norm": 0.7040731310844421,
+ "learning_rate": 2.0939728759489463e-07,
+ "loss": 0.0199,
+ "step": 34377
+ },
+ {
+ "epoch": 93.67302452316076,
+ "grad_norm": 0.8520128130912781,
+ "learning_rate": 2.0921767576555952e-07,
+ "loss": 0.0664,
+ "step": 34378
+ },
+ {
+ "epoch": 93.6757493188011,
+ "grad_norm": 0.8714601993560791,
+ "learning_rate": 2.0903814018652047e-07,
+ "loss": 0.0054,
+ "step": 34379
+ },
+ {
+ "epoch": 93.67847411444141,
+ "grad_norm": 2.1422839164733887,
+ "learning_rate": 2.088586808591797e-07,
+ "loss": 0.1433,
+ "step": 34380
+ },
+ {
+ "epoch": 93.68119891008175,
+ "grad_norm": 1.0607444047927856,
+ "learning_rate": 2.0867929778493167e-07,
+ "loss": 0.012,
+ "step": 34381
+ },
+ {
+ "epoch": 93.68392370572207,
+ "grad_norm": 1.1867141723632812,
+ "learning_rate": 2.0849999096517304e-07,
+ "loss": 0.0218,
+ "step": 34382
+ },
+ {
+ "epoch": 93.6866485013624,
+ "grad_norm": 1.392859935760498,
+ "learning_rate": 2.0832076040130267e-07,
+ "loss": 0.0293,
+ "step": 34383
+ },
+ {
+ "epoch": 93.68937329700273,
+ "grad_norm": 0.8555455207824707,
+ "learning_rate": 2.08141606094715e-07,
+ "loss": 0.0141,
+ "step": 34384
+ },
+ {
+ "epoch": 93.69209809264305,
+ "grad_norm": 1.1048709154129028,
+ "learning_rate": 2.079625280468045e-07,
+ "loss": 0.0272,
+ "step": 34385
+ },
+ {
+ "epoch": 93.69482288828338,
+ "grad_norm": 0.2646663784980774,
+ "learning_rate": 2.077835262589667e-07,
+ "loss": 0.0026,
+ "step": 34386
+ },
+ {
+ "epoch": 93.6975476839237,
+ "grad_norm": 2.2425262928009033,
+ "learning_rate": 2.0760460073259603e-07,
+ "loss": 0.0198,
+ "step": 34387
+ },
+ {
+ "epoch": 93.70027247956403,
+ "grad_norm": 1.2147655487060547,
+ "learning_rate": 2.0742575146908583e-07,
+ "loss": 0.0347,
+ "step": 34388
+ },
+ {
+ "epoch": 93.70299727520435,
+ "grad_norm": 1.3868567943572998,
+ "learning_rate": 2.0724697846982832e-07,
+ "loss": 0.1487,
+ "step": 34389
+ },
+ {
+ "epoch": 93.70572207084469,
+ "grad_norm": 1.4634689092636108,
+ "learning_rate": 2.0706828173621684e-07,
+ "loss": 0.0649,
+ "step": 34390
+ },
+ {
+ "epoch": 93.70844686648502,
+ "grad_norm": 0.6837377548217773,
+ "learning_rate": 2.0688966126964138e-07,
+ "loss": 0.0073,
+ "step": 34391
+ },
+ {
+ "epoch": 93.71117166212534,
+ "grad_norm": 1.363095760345459,
+ "learning_rate": 2.0671111707149416e-07,
+ "loss": 0.0116,
+ "step": 34392
+ },
+ {
+ "epoch": 93.71389645776567,
+ "grad_norm": 1.357659101486206,
+ "learning_rate": 2.065326491431663e-07,
+ "loss": 0.0664,
+ "step": 34393
+ },
+ {
+ "epoch": 93.71662125340599,
+ "grad_norm": 0.7929434776306152,
+ "learning_rate": 2.0635425748604664e-07,
+ "loss": 0.0105,
+ "step": 34394
+ },
+ {
+ "epoch": 93.71934604904632,
+ "grad_norm": 0.7304992079734802,
+ "learning_rate": 2.0617594210152413e-07,
+ "loss": 0.0074,
+ "step": 34395
+ },
+ {
+ "epoch": 93.72207084468666,
+ "grad_norm": 1.3681288957595825,
+ "learning_rate": 2.0599770299098987e-07,
+ "loss": 0.0568,
+ "step": 34396
+ },
+ {
+ "epoch": 93.72479564032697,
+ "grad_norm": 0.701920747756958,
+ "learning_rate": 2.058195401558294e-07,
+ "loss": 0.0066,
+ "step": 34397
+ },
+ {
+ "epoch": 93.7275204359673,
+ "grad_norm": 1.781754970550537,
+ "learning_rate": 2.0564145359743048e-07,
+ "loss": 0.0172,
+ "step": 34398
+ },
+ {
+ "epoch": 93.73024523160763,
+ "grad_norm": 0.9166093468666077,
+ "learning_rate": 2.0546344331718092e-07,
+ "loss": 0.0114,
+ "step": 34399
+ },
+ {
+ "epoch": 93.73297002724796,
+ "grad_norm": 2.2859325408935547,
+ "learning_rate": 2.052855093164674e-07,
+ "loss": 0.0146,
+ "step": 34400
+ },
+ {
+ "epoch": 93.73569482288828,
+ "grad_norm": 1.1182284355163574,
+ "learning_rate": 2.0510765159667546e-07,
+ "loss": 0.0812,
+ "step": 34401
+ },
+ {
+ "epoch": 93.73841961852861,
+ "grad_norm": 3.5883982181549072,
+ "learning_rate": 2.049298701591884e-07,
+ "loss": 0.0185,
+ "step": 34402
+ },
+ {
+ "epoch": 93.74114441416894,
+ "grad_norm": 0.6543452739715576,
+ "learning_rate": 2.0475216500539298e-07,
+ "loss": 0.0069,
+ "step": 34403
+ },
+ {
+ "epoch": 93.74386920980926,
+ "grad_norm": 1.9529204368591309,
+ "learning_rate": 2.0457453613667134e-07,
+ "loss": 0.0813,
+ "step": 34404
+ },
+ {
+ "epoch": 93.7465940054496,
+ "grad_norm": 1.208975911140442,
+ "learning_rate": 2.043969835544102e-07,
+ "loss": 0.0075,
+ "step": 34405
+ },
+ {
+ "epoch": 93.74931880108991,
+ "grad_norm": 0.45874229073524475,
+ "learning_rate": 2.042195072599884e-07,
+ "loss": 0.0039,
+ "step": 34406
+ },
+ {
+ "epoch": 93.75204359673025,
+ "grad_norm": 0.7624847888946533,
+ "learning_rate": 2.0404210725479046e-07,
+ "loss": 0.0056,
+ "step": 34407
+ },
+ {
+ "epoch": 93.75476839237058,
+ "grad_norm": 1.2371668815612793,
+ "learning_rate": 2.0386478354019745e-07,
+ "loss": 0.0338,
+ "step": 34408
+ },
+ {
+ "epoch": 93.7574931880109,
+ "grad_norm": 0.9922500252723694,
+ "learning_rate": 2.0368753611759052e-07,
+ "loss": 0.0131,
+ "step": 34409
+ },
+ {
+ "epoch": 93.76021798365123,
+ "grad_norm": 1.985095739364624,
+ "learning_rate": 2.0351036498834965e-07,
+ "loss": 0.0953,
+ "step": 34410
+ },
+ {
+ "epoch": 93.76294277929155,
+ "grad_norm": 0.8566673398017883,
+ "learning_rate": 2.0333327015385373e-07,
+ "loss": 0.0099,
+ "step": 34411
+ },
+ {
+ "epoch": 93.76566757493188,
+ "grad_norm": 1.1298238039016724,
+ "learning_rate": 2.0315625161548392e-07,
+ "loss": 0.0359,
+ "step": 34412
+ },
+ {
+ "epoch": 93.7683923705722,
+ "grad_norm": 0.9418379068374634,
+ "learning_rate": 2.0297930937461685e-07,
+ "loss": 0.0124,
+ "step": 34413
+ },
+ {
+ "epoch": 93.77111716621253,
+ "grad_norm": 0.7858708500862122,
+ "learning_rate": 2.0280244343263366e-07,
+ "loss": 0.0081,
+ "step": 34414
+ },
+ {
+ "epoch": 93.77384196185287,
+ "grad_norm": 0.9715008735656738,
+ "learning_rate": 2.026256537909077e-07,
+ "loss": 0.0099,
+ "step": 34415
+ },
+ {
+ "epoch": 93.77656675749319,
+ "grad_norm": 0.6079581379890442,
+ "learning_rate": 2.0244894045081898e-07,
+ "loss": 0.0066,
+ "step": 34416
+ },
+ {
+ "epoch": 93.77929155313352,
+ "grad_norm": 1.9896315336227417,
+ "learning_rate": 2.0227230341374194e-07,
+ "loss": 0.0125,
+ "step": 34417
+ },
+ {
+ "epoch": 93.78201634877384,
+ "grad_norm": 1.335133671760559,
+ "learning_rate": 2.0209574268105324e-07,
+ "loss": 0.014,
+ "step": 34418
+ },
+ {
+ "epoch": 93.78474114441417,
+ "grad_norm": 0.6670296788215637,
+ "learning_rate": 2.0191925825412739e-07,
+ "loss": 0.0052,
+ "step": 34419
+ },
+ {
+ "epoch": 93.7874659400545,
+ "grad_norm": 1.5582481622695923,
+ "learning_rate": 2.017428501343388e-07,
+ "loss": 0.0114,
+ "step": 34420
+ },
+ {
+ "epoch": 93.79019073569482,
+ "grad_norm": 0.9027248024940491,
+ "learning_rate": 2.0156651832306085e-07,
+ "loss": 0.0165,
+ "step": 34421
+ },
+ {
+ "epoch": 93.79291553133515,
+ "grad_norm": 0.6555768847465515,
+ "learning_rate": 2.0139026282166908e-07,
+ "loss": 0.0066,
+ "step": 34422
+ },
+ {
+ "epoch": 93.79564032697547,
+ "grad_norm": 0.7662385106086731,
+ "learning_rate": 2.012140836315335e-07,
+ "loss": 0.0085,
+ "step": 34423
+ },
+ {
+ "epoch": 93.7983651226158,
+ "grad_norm": 1.25184166431427,
+ "learning_rate": 2.0103798075402747e-07,
+ "loss": 0.0687,
+ "step": 34424
+ },
+ {
+ "epoch": 93.80108991825612,
+ "grad_norm": 1.418323278427124,
+ "learning_rate": 2.0086195419052323e-07,
+ "loss": 0.0154,
+ "step": 34425
+ },
+ {
+ "epoch": 93.80381471389646,
+ "grad_norm": 1.9502592086791992,
+ "learning_rate": 2.0068600394238968e-07,
+ "loss": 0.0137,
+ "step": 34426
+ },
+ {
+ "epoch": 93.80653950953679,
+ "grad_norm": 1.0277847051620483,
+ "learning_rate": 2.0051013001099794e-07,
+ "loss": 0.0371,
+ "step": 34427
+ },
+ {
+ "epoch": 93.80926430517711,
+ "grad_norm": 0.9257871508598328,
+ "learning_rate": 2.0033433239771803e-07,
+ "loss": 0.0099,
+ "step": 34428
+ },
+ {
+ "epoch": 93.81198910081744,
+ "grad_norm": 2.2898807525634766,
+ "learning_rate": 2.0015861110392e-07,
+ "loss": 0.0172,
+ "step": 34429
+ },
+ {
+ "epoch": 93.81471389645776,
+ "grad_norm": 1.666783332824707,
+ "learning_rate": 1.9998296613097046e-07,
+ "loss": 0.0097,
+ "step": 34430
+ },
+ {
+ "epoch": 93.8174386920981,
+ "grad_norm": 1.2388150691986084,
+ "learning_rate": 1.9980739748023837e-07,
+ "loss": 0.0079,
+ "step": 34431
+ },
+ {
+ "epoch": 93.82016348773843,
+ "grad_norm": 0.629598081111908,
+ "learning_rate": 1.9963190515309038e-07,
+ "loss": 0.0071,
+ "step": 34432
+ },
+ {
+ "epoch": 93.82288828337875,
+ "grad_norm": 0.880767822265625,
+ "learning_rate": 1.9945648915089432e-07,
+ "loss": 0.0199,
+ "step": 34433
+ },
+ {
+ "epoch": 93.82561307901908,
+ "grad_norm": 5.981553077697754,
+ "learning_rate": 1.9928114947501577e-07,
+ "loss": 0.0088,
+ "step": 34434
+ },
+ {
+ "epoch": 93.8283378746594,
+ "grad_norm": 0.7362450361251831,
+ "learning_rate": 1.9910588612681914e-07,
+ "loss": 0.0069,
+ "step": 34435
+ },
+ {
+ "epoch": 93.83106267029973,
+ "grad_norm": 2.0255024433135986,
+ "learning_rate": 1.9893069910767117e-07,
+ "loss": 0.0099,
+ "step": 34436
+ },
+ {
+ "epoch": 93.83378746594005,
+ "grad_norm": 0.8889692425727844,
+ "learning_rate": 1.9875558841893627e-07,
+ "loss": 0.0125,
+ "step": 34437
+ },
+ {
+ "epoch": 93.83651226158038,
+ "grad_norm": 2.952483892440796,
+ "learning_rate": 1.9858055406197562e-07,
+ "loss": 0.0483,
+ "step": 34438
+ },
+ {
+ "epoch": 93.83923705722071,
+ "grad_norm": 0.5079165101051331,
+ "learning_rate": 1.984055960381559e-07,
+ "loss": 0.0058,
+ "step": 34439
+ },
+ {
+ "epoch": 93.84196185286103,
+ "grad_norm": 1.3724334239959717,
+ "learning_rate": 1.98230714348836e-07,
+ "loss": 0.0158,
+ "step": 34440
+ },
+ {
+ "epoch": 93.84468664850137,
+ "grad_norm": 0.9440301060676575,
+ "learning_rate": 1.9805590899538152e-07,
+ "loss": 0.0067,
+ "step": 34441
+ },
+ {
+ "epoch": 93.84741144414168,
+ "grad_norm": 3.522355794906616,
+ "learning_rate": 1.9788117997915136e-07,
+ "loss": 0.0698,
+ "step": 34442
+ },
+ {
+ "epoch": 93.85013623978202,
+ "grad_norm": 0.790513277053833,
+ "learning_rate": 1.9770652730150775e-07,
+ "loss": 0.0045,
+ "step": 34443
+ },
+ {
+ "epoch": 93.85286103542235,
+ "grad_norm": 1.599066138267517,
+ "learning_rate": 1.9753195096380962e-07,
+ "loss": 0.031,
+ "step": 34444
+ },
+ {
+ "epoch": 93.85558583106267,
+ "grad_norm": 0.9913542866706848,
+ "learning_rate": 1.973574509674181e-07,
+ "loss": 0.0208,
+ "step": 34445
+ },
+ {
+ "epoch": 93.858310626703,
+ "grad_norm": 1.7547615766525269,
+ "learning_rate": 1.9718302731368988e-07,
+ "loss": 0.0202,
+ "step": 34446
+ },
+ {
+ "epoch": 93.86103542234332,
+ "grad_norm": 0.8412469029426575,
+ "learning_rate": 1.97008680003985e-07,
+ "loss": 0.0161,
+ "step": 34447
+ },
+ {
+ "epoch": 93.86376021798365,
+ "grad_norm": 3.017685890197754,
+ "learning_rate": 1.9683440903966123e-07,
+ "loss": 0.0173,
+ "step": 34448
+ },
+ {
+ "epoch": 93.86648501362397,
+ "grad_norm": 0.8435469269752502,
+ "learning_rate": 1.9666021442207638e-07,
+ "loss": 0.0068,
+ "step": 34449
+ },
+ {
+ "epoch": 93.8692098092643,
+ "grad_norm": 1.156922459602356,
+ "learning_rate": 1.9648609615258606e-07,
+ "loss": 0.0302,
+ "step": 34450
+ },
+ {
+ "epoch": 93.87193460490464,
+ "grad_norm": 1.5767648220062256,
+ "learning_rate": 1.9631205423254586e-07,
+ "loss": 0.0435,
+ "step": 34451
+ },
+ {
+ "epoch": 93.87465940054496,
+ "grad_norm": 2.050903797149658,
+ "learning_rate": 1.961380886633113e-07,
+ "loss": 0.0323,
+ "step": 34452
+ },
+ {
+ "epoch": 93.87738419618529,
+ "grad_norm": 1.2632384300231934,
+ "learning_rate": 1.959641994462391e-07,
+ "loss": 0.0422,
+ "step": 34453
+ },
+ {
+ "epoch": 93.88010899182561,
+ "grad_norm": 0.9873860478401184,
+ "learning_rate": 1.9579038658268157e-07,
+ "loss": 0.0176,
+ "step": 34454
+ },
+ {
+ "epoch": 93.88283378746594,
+ "grad_norm": 1.2090240716934204,
+ "learning_rate": 1.9561665007399312e-07,
+ "loss": 0.0281,
+ "step": 34455
+ },
+ {
+ "epoch": 93.88555858310627,
+ "grad_norm": 0.87428879737854,
+ "learning_rate": 1.95442989921526e-07,
+ "loss": 0.0091,
+ "step": 34456
+ },
+ {
+ "epoch": 93.88828337874659,
+ "grad_norm": 1.0274615287780762,
+ "learning_rate": 1.9526940612663359e-07,
+ "loss": 0.0149,
+ "step": 34457
+ },
+ {
+ "epoch": 93.89100817438693,
+ "grad_norm": 0.5071557760238647,
+ "learning_rate": 1.9509589869066814e-07,
+ "loss": 0.005,
+ "step": 34458
+ },
+ {
+ "epoch": 93.89373297002724,
+ "grad_norm": 1.093205451965332,
+ "learning_rate": 1.9492246761497967e-07,
+ "loss": 0.0164,
+ "step": 34459
+ },
+ {
+ "epoch": 93.89645776566758,
+ "grad_norm": 0.9071512222290039,
+ "learning_rate": 1.9474911290091936e-07,
+ "loss": 0.0081,
+ "step": 34460
+ },
+ {
+ "epoch": 93.8991825613079,
+ "grad_norm": 0.5314107537269592,
+ "learning_rate": 1.9457583454983719e-07,
+ "loss": 0.0058,
+ "step": 34461
+ },
+ {
+ "epoch": 93.90190735694823,
+ "grad_norm": 0.8921732902526855,
+ "learning_rate": 1.9440263256308433e-07,
+ "loss": 0.011,
+ "step": 34462
+ },
+ {
+ "epoch": 93.90463215258856,
+ "grad_norm": 2.417358875274658,
+ "learning_rate": 1.942295069420075e-07,
+ "loss": 0.014,
+ "step": 34463
+ },
+ {
+ "epoch": 93.90735694822888,
+ "grad_norm": 1.5206549167633057,
+ "learning_rate": 1.9405645768795446e-07,
+ "loss": 0.0222,
+ "step": 34464
+ },
+ {
+ "epoch": 93.91008174386921,
+ "grad_norm": 1.4672776460647583,
+ "learning_rate": 1.9388348480227525e-07,
+ "loss": 0.0091,
+ "step": 34465
+ },
+ {
+ "epoch": 93.91280653950953,
+ "grad_norm": 2.0015108585357666,
+ "learning_rate": 1.937105882863155e-07,
+ "loss": 0.1468,
+ "step": 34466
+ },
+ {
+ "epoch": 93.91553133514986,
+ "grad_norm": 1.4130357503890991,
+ "learning_rate": 1.935377681414219e-07,
+ "loss": 0.0247,
+ "step": 34467
+ },
+ {
+ "epoch": 93.9182561307902,
+ "grad_norm": 0.5847865343093872,
+ "learning_rate": 1.933650243689411e-07,
+ "loss": 0.0066,
+ "step": 34468
+ },
+ {
+ "epoch": 93.92098092643052,
+ "grad_norm": 1.1894843578338623,
+ "learning_rate": 1.9319235697021766e-07,
+ "loss": 0.0397,
+ "step": 34469
+ },
+ {
+ "epoch": 93.92370572207085,
+ "grad_norm": 1.5131866931915283,
+ "learning_rate": 1.93019765946596e-07,
+ "loss": 0.0111,
+ "step": 34470
+ },
+ {
+ "epoch": 93.92643051771117,
+ "grad_norm": 1.6262925863265991,
+ "learning_rate": 1.928472512994206e-07,
+ "loss": 0.1159,
+ "step": 34471
+ },
+ {
+ "epoch": 93.9291553133515,
+ "grad_norm": 1.87568998336792,
+ "learning_rate": 1.9267481303003487e-07,
+ "loss": 0.0277,
+ "step": 34472
+ },
+ {
+ "epoch": 93.93188010899182,
+ "grad_norm": 0.9719721674919128,
+ "learning_rate": 1.9250245113978329e-07,
+ "loss": 0.0416,
+ "step": 34473
+ },
+ {
+ "epoch": 93.93460490463215,
+ "grad_norm": 1.062349796295166,
+ "learning_rate": 1.9233016563000695e-07,
+ "loss": 0.0737,
+ "step": 34474
+ },
+ {
+ "epoch": 93.93732970027249,
+ "grad_norm": 2.3163950443267822,
+ "learning_rate": 1.9215795650204817e-07,
+ "loss": 0.0137,
+ "step": 34475
+ },
+ {
+ "epoch": 93.9400544959128,
+ "grad_norm": 1.1111204624176025,
+ "learning_rate": 1.9198582375724584e-07,
+ "loss": 0.0165,
+ "step": 34476
+ },
+ {
+ "epoch": 93.94277929155314,
+ "grad_norm": 1.386568307876587,
+ "learning_rate": 1.9181376739694336e-07,
+ "loss": 0.0509,
+ "step": 34477
+ },
+ {
+ "epoch": 93.94550408719346,
+ "grad_norm": 0.7509040236473083,
+ "learning_rate": 1.9164178742247964e-07,
+ "loss": 0.0053,
+ "step": 34478
+ },
+ {
+ "epoch": 93.94822888283379,
+ "grad_norm": 1.7769138813018799,
+ "learning_rate": 1.914698838351936e-07,
+ "loss": 0.0397,
+ "step": 34479
+ },
+ {
+ "epoch": 93.95095367847412,
+ "grad_norm": 0.9588366746902466,
+ "learning_rate": 1.912980566364231e-07,
+ "loss": 0.0313,
+ "step": 34480
+ },
+ {
+ "epoch": 93.95367847411444,
+ "grad_norm": 0.5403172373771667,
+ "learning_rate": 1.9112630582750924e-07,
+ "loss": 0.0081,
+ "step": 34481
+ },
+ {
+ "epoch": 93.95640326975477,
+ "grad_norm": 1.2868295907974243,
+ "learning_rate": 1.9095463140978876e-07,
+ "loss": 0.0481,
+ "step": 34482
+ },
+ {
+ "epoch": 93.95912806539509,
+ "grad_norm": 1.056473970413208,
+ "learning_rate": 1.9078303338459725e-07,
+ "loss": 0.0153,
+ "step": 34483
+ },
+ {
+ "epoch": 93.96185286103542,
+ "grad_norm": 0.9531393647193909,
+ "learning_rate": 1.906115117532703e-07,
+ "loss": 0.0133,
+ "step": 34484
+ },
+ {
+ "epoch": 93.96457765667574,
+ "grad_norm": 1.2717576026916504,
+ "learning_rate": 1.9044006651714686e-07,
+ "loss": 0.0932,
+ "step": 34485
+ },
+ {
+ "epoch": 93.96730245231608,
+ "grad_norm": 0.7556722164154053,
+ "learning_rate": 1.902686976775603e-07,
+ "loss": 0.009,
+ "step": 34486
+ },
+ {
+ "epoch": 93.97002724795641,
+ "grad_norm": 1.327765941619873,
+ "learning_rate": 1.9009740523584507e-07,
+ "loss": 0.0109,
+ "step": 34487
+ },
+ {
+ "epoch": 93.97275204359673,
+ "grad_norm": 1.0047179460525513,
+ "learning_rate": 1.899261891933346e-07,
+ "loss": 0.0317,
+ "step": 34488
+ },
+ {
+ "epoch": 93.97547683923706,
+ "grad_norm": 1.104649543762207,
+ "learning_rate": 1.897550495513645e-07,
+ "loss": 0.1161,
+ "step": 34489
+ },
+ {
+ "epoch": 93.97820163487738,
+ "grad_norm": 0.6957873702049255,
+ "learning_rate": 1.8958398631126586e-07,
+ "loss": 0.0086,
+ "step": 34490
+ },
+ {
+ "epoch": 93.98092643051771,
+ "grad_norm": 0.7012134790420532,
+ "learning_rate": 1.8941299947437098e-07,
+ "loss": 0.0049,
+ "step": 34491
+ },
+ {
+ "epoch": 93.98365122615803,
+ "grad_norm": 1.6795061826705933,
+ "learning_rate": 1.8924208904201215e-07,
+ "loss": 0.0948,
+ "step": 34492
+ },
+ {
+ "epoch": 93.98637602179836,
+ "grad_norm": 0.8546541929244995,
+ "learning_rate": 1.890712550155205e-07,
+ "loss": 0.0089,
+ "step": 34493
+ },
+ {
+ "epoch": 93.9891008174387,
+ "grad_norm": 1.2096797227859497,
+ "learning_rate": 1.8890049739622606e-07,
+ "loss": 0.0303,
+ "step": 34494
+ },
+ {
+ "epoch": 93.99182561307902,
+ "grad_norm": 1.8440830707550049,
+ "learning_rate": 1.8872981618545894e-07,
+ "loss": 0.0097,
+ "step": 34495
+ },
+ {
+ "epoch": 93.99455040871935,
+ "grad_norm": 0.6233428120613098,
+ "learning_rate": 1.8855921138454802e-07,
+ "loss": 0.0046,
+ "step": 34496
+ },
+ {
+ "epoch": 93.99727520435967,
+ "grad_norm": 1.0193965435028076,
+ "learning_rate": 1.8838868299482227e-07,
+ "loss": 0.022,
+ "step": 34497
+ },
+ {
+ "epoch": 94.0,
+ "grad_norm": 0.479464590549469,
+ "learning_rate": 1.8821823101760949e-07,
+ "loss": 0.0049,
+ "step": 34498
+ },
+ {
+ "epoch": 94.00272479564033,
+ "grad_norm": 0.9867615699768066,
+ "learning_rate": 1.8804785545423753e-07,
+ "loss": 0.016,
+ "step": 34499
+ },
+ {
+ "epoch": 94.00544959128065,
+ "grad_norm": 1.1888459920883179,
+ "learning_rate": 1.878775563060331e-07,
+ "loss": 0.0161,
+ "step": 34500
+ },
+ {
+ "epoch": 94.00817438692098,
+ "grad_norm": 1.214171051979065,
+ "learning_rate": 1.8770733357432293e-07,
+ "loss": 0.0344,
+ "step": 34501
+ },
+ {
+ "epoch": 94.0108991825613,
+ "grad_norm": 1.1748430728912354,
+ "learning_rate": 1.8753718726043146e-07,
+ "loss": 0.023,
+ "step": 34502
+ },
+ {
+ "epoch": 94.01362397820164,
+ "grad_norm": 1.2197591066360474,
+ "learning_rate": 1.8736711736568547e-07,
+ "loss": 0.0168,
+ "step": 34503
+ },
+ {
+ "epoch": 94.01634877384195,
+ "grad_norm": 0.9922661781311035,
+ "learning_rate": 1.871971238914072e-07,
+ "loss": 0.013,
+ "step": 34504
+ },
+ {
+ "epoch": 94.01907356948229,
+ "grad_norm": 0.8372026085853577,
+ "learning_rate": 1.870272068389223e-07,
+ "loss": 0.0945,
+ "step": 34505
+ },
+ {
+ "epoch": 94.02179836512262,
+ "grad_norm": 0.990118682384491,
+ "learning_rate": 1.868573662095541e-07,
+ "loss": 0.014,
+ "step": 34506
+ },
+ {
+ "epoch": 94.02452316076294,
+ "grad_norm": 2.081666946411133,
+ "learning_rate": 1.8668760200462488e-07,
+ "loss": 0.0653,
+ "step": 34507
+ },
+ {
+ "epoch": 94.02724795640327,
+ "grad_norm": 0.6701085567474365,
+ "learning_rate": 1.8651791422545697e-07,
+ "loss": 0.0048,
+ "step": 34508
+ },
+ {
+ "epoch": 94.02997275204359,
+ "grad_norm": 0.5549752116203308,
+ "learning_rate": 1.8634830287337036e-07,
+ "loss": 0.0061,
+ "step": 34509
+ },
+ {
+ "epoch": 94.03269754768392,
+ "grad_norm": 0.6764034032821655,
+ "learning_rate": 1.861787679496885e-07,
+ "loss": 0.0049,
+ "step": 34510
+ },
+ {
+ "epoch": 94.03542234332426,
+ "grad_norm": 0.9285191893577576,
+ "learning_rate": 1.8600930945573027e-07,
+ "loss": 0.0924,
+ "step": 34511
+ },
+ {
+ "epoch": 94.03814713896458,
+ "grad_norm": 0.5586862564086914,
+ "learning_rate": 1.8583992739281577e-07,
+ "loss": 0.0049,
+ "step": 34512
+ },
+ {
+ "epoch": 94.04087193460491,
+ "grad_norm": 0.7204278707504272,
+ "learning_rate": 1.8567062176226393e-07,
+ "loss": 0.0072,
+ "step": 34513
+ },
+ {
+ "epoch": 94.04359673024523,
+ "grad_norm": 1.000265121459961,
+ "learning_rate": 1.8550139256539368e-07,
+ "loss": 0.0064,
+ "step": 34514
+ },
+ {
+ "epoch": 94.04632152588556,
+ "grad_norm": 1.770232081413269,
+ "learning_rate": 1.8533223980352288e-07,
+ "loss": 0.0641,
+ "step": 34515
+ },
+ {
+ "epoch": 94.04904632152588,
+ "grad_norm": 1.5886955261230469,
+ "learning_rate": 1.8516316347796826e-07,
+ "loss": 0.0414,
+ "step": 34516
+ },
+ {
+ "epoch": 94.05177111716621,
+ "grad_norm": 1.4909799098968506,
+ "learning_rate": 1.8499416359004763e-07,
+ "loss": 0.0979,
+ "step": 34517
+ },
+ {
+ "epoch": 94.05449591280654,
+ "grad_norm": 1.2594705820083618,
+ "learning_rate": 1.848252401410755e-07,
+ "loss": 0.1073,
+ "step": 34518
+ },
+ {
+ "epoch": 94.05722070844686,
+ "grad_norm": 0.932488203048706,
+ "learning_rate": 1.846563931323697e-07,
+ "loss": 0.0071,
+ "step": 34519
+ },
+ {
+ "epoch": 94.0599455040872,
+ "grad_norm": 1.2251955270767212,
+ "learning_rate": 1.8448762256524256e-07,
+ "loss": 0.0133,
+ "step": 34520
+ },
+ {
+ "epoch": 94.06267029972751,
+ "grad_norm": 1.1250877380371094,
+ "learning_rate": 1.8431892844101183e-07,
+ "loss": 0.0191,
+ "step": 34521
+ },
+ {
+ "epoch": 94.06539509536785,
+ "grad_norm": 0.43762701749801636,
+ "learning_rate": 1.8415031076098767e-07,
+ "loss": 0.0041,
+ "step": 34522
+ },
+ {
+ "epoch": 94.06811989100818,
+ "grad_norm": 0.602100670337677,
+ "learning_rate": 1.8398176952648672e-07,
+ "loss": 0.0056,
+ "step": 34523
+ },
+ {
+ "epoch": 94.0708446866485,
+ "grad_norm": 0.9755491614341736,
+ "learning_rate": 1.8381330473881797e-07,
+ "loss": 0.0101,
+ "step": 34524
+ },
+ {
+ "epoch": 94.07356948228883,
+ "grad_norm": 0.841698169708252,
+ "learning_rate": 1.836449163992948e-07,
+ "loss": 0.012,
+ "step": 34525
+ },
+ {
+ "epoch": 94.07629427792915,
+ "grad_norm": 0.8499727249145508,
+ "learning_rate": 1.8347660450923066e-07,
+ "loss": 0.0079,
+ "step": 34526
+ },
+ {
+ "epoch": 94.07901907356948,
+ "grad_norm": 1.080688238143921,
+ "learning_rate": 1.8330836906993443e-07,
+ "loss": 0.0257,
+ "step": 34527
+ },
+ {
+ "epoch": 94.0817438692098,
+ "grad_norm": 1.2099391222000122,
+ "learning_rate": 1.8314021008271732e-07,
+ "loss": 0.0286,
+ "step": 34528
+ },
+ {
+ "epoch": 94.08446866485014,
+ "grad_norm": 1.3834881782531738,
+ "learning_rate": 1.8297212754888716e-07,
+ "loss": 0.0068,
+ "step": 34529
+ },
+ {
+ "epoch": 94.08719346049047,
+ "grad_norm": 0.9782344102859497,
+ "learning_rate": 1.8280412146975512e-07,
+ "loss": 0.0154,
+ "step": 34530
+ },
+ {
+ "epoch": 94.08991825613079,
+ "grad_norm": 1.0306553840637207,
+ "learning_rate": 1.8263619184662796e-07,
+ "loss": 0.0127,
+ "step": 34531
+ },
+ {
+ "epoch": 94.09264305177112,
+ "grad_norm": 1.1728399991989136,
+ "learning_rate": 1.8246833868081348e-07,
+ "loss": 0.025,
+ "step": 34532
+ },
+ {
+ "epoch": 94.09536784741144,
+ "grad_norm": 0.8565760254859924,
+ "learning_rate": 1.8230056197362068e-07,
+ "loss": 0.0079,
+ "step": 34533
+ },
+ {
+ "epoch": 94.09809264305177,
+ "grad_norm": 1.4755572080612183,
+ "learning_rate": 1.8213286172635514e-07,
+ "loss": 0.0254,
+ "step": 34534
+ },
+ {
+ "epoch": 94.1008174386921,
+ "grad_norm": 0.6154555082321167,
+ "learning_rate": 1.819652379403225e-07,
+ "loss": 0.0093,
+ "step": 34535
+ },
+ {
+ "epoch": 94.10354223433242,
+ "grad_norm": 1.1415910720825195,
+ "learning_rate": 1.8179769061682838e-07,
+ "loss": 0.0061,
+ "step": 34536
+ },
+ {
+ "epoch": 94.10626702997276,
+ "grad_norm": 0.9470281600952148,
+ "learning_rate": 1.8163021975717841e-07,
+ "loss": 0.0175,
+ "step": 34537
+ },
+ {
+ "epoch": 94.10899182561307,
+ "grad_norm": 0.6014063358306885,
+ "learning_rate": 1.8146282536267602e-07,
+ "loss": 0.006,
+ "step": 34538
+ },
+ {
+ "epoch": 94.11171662125341,
+ "grad_norm": 0.4788425862789154,
+ "learning_rate": 1.8129550743462565e-07,
+ "loss": 0.0058,
+ "step": 34539
+ },
+ {
+ "epoch": 94.11444141689373,
+ "grad_norm": 0.5906739830970764,
+ "learning_rate": 1.8112826597432854e-07,
+ "loss": 0.0059,
+ "step": 34540
+ },
+ {
+ "epoch": 94.11716621253406,
+ "grad_norm": 1.4477248191833496,
+ "learning_rate": 1.8096110098308917e-07,
+ "loss": 0.0203,
+ "step": 34541
+ },
+ {
+ "epoch": 94.11989100817439,
+ "grad_norm": 1.036250352859497,
+ "learning_rate": 1.8079401246220874e-07,
+ "loss": 0.0122,
+ "step": 34542
+ },
+ {
+ "epoch": 94.12261580381471,
+ "grad_norm": 1.2453988790512085,
+ "learning_rate": 1.8062700041298952e-07,
+ "loss": 0.1386,
+ "step": 34543
+ },
+ {
+ "epoch": 94.12534059945504,
+ "grad_norm": 1.034147024154663,
+ "learning_rate": 1.8046006483673052e-07,
+ "loss": 0.0161,
+ "step": 34544
+ },
+ {
+ "epoch": 94.12806539509536,
+ "grad_norm": 1.1676909923553467,
+ "learning_rate": 1.8029320573473063e-07,
+ "loss": 0.0543,
+ "step": 34545
+ },
+ {
+ "epoch": 94.1307901907357,
+ "grad_norm": 0.8245187997817993,
+ "learning_rate": 1.801264231082933e-07,
+ "loss": 0.0063,
+ "step": 34546
+ },
+ {
+ "epoch": 94.13351498637603,
+ "grad_norm": 0.9339087605476379,
+ "learning_rate": 1.7995971695871528e-07,
+ "loss": 0.016,
+ "step": 34547
+ },
+ {
+ "epoch": 94.13623978201635,
+ "grad_norm": 1.3371790647506714,
+ "learning_rate": 1.7979308728729438e-07,
+ "loss": 0.0102,
+ "step": 34548
+ },
+ {
+ "epoch": 94.13896457765668,
+ "grad_norm": 0.9769082069396973,
+ "learning_rate": 1.7962653409532848e-07,
+ "loss": 0.0128,
+ "step": 34549
+ },
+ {
+ "epoch": 94.141689373297,
+ "grad_norm": 0.7209330797195435,
+ "learning_rate": 1.7946005738411654e-07,
+ "loss": 0.0112,
+ "step": 34550
+ },
+ {
+ "epoch": 94.14441416893733,
+ "grad_norm": 1.1587783098220825,
+ "learning_rate": 1.7929365715495305e-07,
+ "loss": 0.0771,
+ "step": 34551
+ },
+ {
+ "epoch": 94.14713896457765,
+ "grad_norm": 0.9125856161117554,
+ "learning_rate": 1.791273334091337e-07,
+ "loss": 0.0065,
+ "step": 34552
+ },
+ {
+ "epoch": 94.14986376021798,
+ "grad_norm": 0.6746422052383423,
+ "learning_rate": 1.7896108614795406e-07,
+ "loss": 0.0072,
+ "step": 34553
+ },
+ {
+ "epoch": 94.15258855585832,
+ "grad_norm": 1.2365517616271973,
+ "learning_rate": 1.7879491537271087e-07,
+ "loss": 0.0156,
+ "step": 34554
+ },
+ {
+ "epoch": 94.15531335149863,
+ "grad_norm": 1.1208075284957886,
+ "learning_rate": 1.7862882108469648e-07,
+ "loss": 0.0151,
+ "step": 34555
+ },
+ {
+ "epoch": 94.15803814713897,
+ "grad_norm": 1.7578378915786743,
+ "learning_rate": 1.7846280328520428e-07,
+ "loss": 0.0178,
+ "step": 34556
+ },
+ {
+ "epoch": 94.16076294277929,
+ "grad_norm": 0.7246118187904358,
+ "learning_rate": 1.7829686197552766e-07,
+ "loss": 0.0097,
+ "step": 34557
+ },
+ {
+ "epoch": 94.16348773841962,
+ "grad_norm": 1.8333921432495117,
+ "learning_rate": 1.7813099715695892e-07,
+ "loss": 0.0702,
+ "step": 34558
+ },
+ {
+ "epoch": 94.16621253405995,
+ "grad_norm": 1.0656296014785767,
+ "learning_rate": 1.7796520883079038e-07,
+ "loss": 0.0092,
+ "step": 34559
+ },
+ {
+ "epoch": 94.16893732970027,
+ "grad_norm": 0.8622037172317505,
+ "learning_rate": 1.7779949699831212e-07,
+ "loss": 0.0062,
+ "step": 34560
+ },
+ {
+ "epoch": 94.1716621253406,
+ "grad_norm": 0.7190074920654297,
+ "learning_rate": 1.776338616608153e-07,
+ "loss": 0.0076,
+ "step": 34561
+ },
+ {
+ "epoch": 94.17438692098092,
+ "grad_norm": 0.809807300567627,
+ "learning_rate": 1.7746830281959005e-07,
+ "loss": 0.0068,
+ "step": 34562
+ },
+ {
+ "epoch": 94.17711171662125,
+ "grad_norm": 0.9540445804595947,
+ "learning_rate": 1.7730282047592528e-07,
+ "loss": 0.0072,
+ "step": 34563
+ },
+ {
+ "epoch": 94.17983651226157,
+ "grad_norm": 0.6000019907951355,
+ "learning_rate": 1.7713741463111112e-07,
+ "loss": 0.0073,
+ "step": 34564
+ },
+ {
+ "epoch": 94.1825613079019,
+ "grad_norm": 0.9813573360443115,
+ "learning_rate": 1.7697208528643316e-07,
+ "loss": 0.0713,
+ "step": 34565
+ },
+ {
+ "epoch": 94.18528610354224,
+ "grad_norm": 0.46836772561073303,
+ "learning_rate": 1.7680683244318154e-07,
+ "loss": 0.0034,
+ "step": 34566
+ },
+ {
+ "epoch": 94.18801089918256,
+ "grad_norm": 0.47356468439102173,
+ "learning_rate": 1.7664165610264184e-07,
+ "loss": 0.0044,
+ "step": 34567
+ },
+ {
+ "epoch": 94.19073569482289,
+ "grad_norm": 0.9575560688972473,
+ "learning_rate": 1.7647655626610083e-07,
+ "loss": 0.0449,
+ "step": 34568
+ },
+ {
+ "epoch": 94.19346049046321,
+ "grad_norm": 0.8025112748146057,
+ "learning_rate": 1.7631153293484305e-07,
+ "loss": 0.1046,
+ "step": 34569
+ },
+ {
+ "epoch": 94.19618528610354,
+ "grad_norm": 0.6761380434036255,
+ "learning_rate": 1.7614658611015635e-07,
+ "loss": 0.0043,
+ "step": 34570
+ },
+ {
+ "epoch": 94.19891008174388,
+ "grad_norm": 0.9144987463951111,
+ "learning_rate": 1.75981715793323e-07,
+ "loss": 0.015,
+ "step": 34571
+ },
+ {
+ "epoch": 94.2016348773842,
+ "grad_norm": 1.4522408246994019,
+ "learning_rate": 1.758169219856287e-07,
+ "loss": 0.0209,
+ "step": 34572
+ },
+ {
+ "epoch": 94.20435967302453,
+ "grad_norm": 0.9567128419876099,
+ "learning_rate": 1.7565220468835465e-07,
+ "loss": 0.0158,
+ "step": 34573
+ },
+ {
+ "epoch": 94.20708446866485,
+ "grad_norm": 0.48269718885421753,
+ "learning_rate": 1.7548756390278642e-07,
+ "loss": 0.0035,
+ "step": 34574
+ },
+ {
+ "epoch": 94.20980926430518,
+ "grad_norm": 1.4634939432144165,
+ "learning_rate": 1.7532299963020417e-07,
+ "loss": 0.05,
+ "step": 34575
+ },
+ {
+ "epoch": 94.2125340599455,
+ "grad_norm": 1.1917026042938232,
+ "learning_rate": 1.7515851187189015e-07,
+ "loss": 0.0247,
+ "step": 34576
+ },
+ {
+ "epoch": 94.21525885558583,
+ "grad_norm": 0.6156890392303467,
+ "learning_rate": 1.749941006291245e-07,
+ "loss": 0.0046,
+ "step": 34577
+ },
+ {
+ "epoch": 94.21798365122616,
+ "grad_norm": 0.9607864618301392,
+ "learning_rate": 1.7482976590319058e-07,
+ "loss": 0.0104,
+ "step": 34578
+ },
+ {
+ "epoch": 94.22070844686648,
+ "grad_norm": 1.2469208240509033,
+ "learning_rate": 1.7466550769536516e-07,
+ "loss": 0.0658,
+ "step": 34579
+ },
+ {
+ "epoch": 94.22343324250681,
+ "grad_norm": 1.3975019454956055,
+ "learning_rate": 1.7450132600692947e-07,
+ "loss": 0.0702,
+ "step": 34580
+ },
+ {
+ "epoch": 94.22615803814713,
+ "grad_norm": 1.3541306257247925,
+ "learning_rate": 1.7433722083915915e-07,
+ "loss": 0.0059,
+ "step": 34581
+ },
+ {
+ "epoch": 94.22888283378747,
+ "grad_norm": 0.7542895078659058,
+ "learning_rate": 1.741731921933365e-07,
+ "loss": 0.0139,
+ "step": 34582
+ },
+ {
+ "epoch": 94.2316076294278,
+ "grad_norm": 0.8464952111244202,
+ "learning_rate": 1.7400924007073604e-07,
+ "loss": 0.0077,
+ "step": 34583
+ },
+ {
+ "epoch": 94.23433242506812,
+ "grad_norm": 1.4036213159561157,
+ "learning_rate": 1.7384536447263457e-07,
+ "loss": 0.0082,
+ "step": 34584
+ },
+ {
+ "epoch": 94.23705722070845,
+ "grad_norm": 1.08400297164917,
+ "learning_rate": 1.736815654003099e-07,
+ "loss": 0.0095,
+ "step": 34585
+ },
+ {
+ "epoch": 94.23978201634877,
+ "grad_norm": 1.124982237815857,
+ "learning_rate": 1.7351784285503658e-07,
+ "loss": 0.0166,
+ "step": 34586
+ },
+ {
+ "epoch": 94.2425068119891,
+ "grad_norm": 1.5421245098114014,
+ "learning_rate": 1.733541968380914e-07,
+ "loss": 0.0333,
+ "step": 34587
+ },
+ {
+ "epoch": 94.24523160762942,
+ "grad_norm": 0.5166683197021484,
+ "learning_rate": 1.7319062735074666e-07,
+ "loss": 0.0058,
+ "step": 34588
+ },
+ {
+ "epoch": 94.24795640326975,
+ "grad_norm": 0.24971330165863037,
+ "learning_rate": 1.7302713439427577e-07,
+ "loss": 0.0029,
+ "step": 34589
+ },
+ {
+ "epoch": 94.25068119891009,
+ "grad_norm": 0.669484555721283,
+ "learning_rate": 1.7286371796995438e-07,
+ "loss": 0.0083,
+ "step": 34590
+ },
+ {
+ "epoch": 94.2534059945504,
+ "grad_norm": 1.906599521636963,
+ "learning_rate": 1.7270037807905372e-07,
+ "loss": 0.0221,
+ "step": 34591
+ },
+ {
+ "epoch": 94.25613079019074,
+ "grad_norm": 1.2946703433990479,
+ "learning_rate": 1.7253711472284717e-07,
+ "loss": 0.0138,
+ "step": 34592
+ },
+ {
+ "epoch": 94.25885558583106,
+ "grad_norm": 0.8839141726493835,
+ "learning_rate": 1.7237392790260377e-07,
+ "loss": 0.0101,
+ "step": 34593
+ },
+ {
+ "epoch": 94.26158038147139,
+ "grad_norm": 1.3647944927215576,
+ "learning_rate": 1.722108176195958e-07,
+ "loss": 0.0177,
+ "step": 34594
+ },
+ {
+ "epoch": 94.26430517711172,
+ "grad_norm": 1.849166750907898,
+ "learning_rate": 1.7204778387509559e-07,
+ "loss": 0.0506,
+ "step": 34595
+ },
+ {
+ "epoch": 94.26702997275204,
+ "grad_norm": 0.865872323513031,
+ "learning_rate": 1.7188482667036986e-07,
+ "loss": 0.01,
+ "step": 34596
+ },
+ {
+ "epoch": 94.26975476839237,
+ "grad_norm": 1.0484185218811035,
+ "learning_rate": 1.7172194600668767e-07,
+ "loss": 0.0285,
+ "step": 34597
+ },
+ {
+ "epoch": 94.2724795640327,
+ "grad_norm": 0.693323016166687,
+ "learning_rate": 1.715591418853202e-07,
+ "loss": 0.0057,
+ "step": 34598
+ },
+ {
+ "epoch": 94.27520435967303,
+ "grad_norm": 0.5978161096572876,
+ "learning_rate": 1.7139641430753419e-07,
+ "loss": 0.0046,
+ "step": 34599
+ },
+ {
+ "epoch": 94.27792915531334,
+ "grad_norm": 0.6116475462913513,
+ "learning_rate": 1.712337632745964e-07,
+ "loss": 0.0066,
+ "step": 34600
+ },
+ {
+ "epoch": 94.28065395095368,
+ "grad_norm": 0.6318194270133972,
+ "learning_rate": 1.7107118878777251e-07,
+ "loss": 0.0071,
+ "step": 34601
+ },
+ {
+ "epoch": 94.28337874659401,
+ "grad_norm": 1.0522629022598267,
+ "learning_rate": 1.709086908483304e-07,
+ "loss": 0.0617,
+ "step": 34602
+ },
+ {
+ "epoch": 94.28610354223433,
+ "grad_norm": 1.8258591890335083,
+ "learning_rate": 1.707462694575357e-07,
+ "loss": 0.0398,
+ "step": 34603
+ },
+ {
+ "epoch": 94.28882833787466,
+ "grad_norm": 1.2065497636795044,
+ "learning_rate": 1.7058392461665185e-07,
+ "loss": 0.0147,
+ "step": 34604
+ },
+ {
+ "epoch": 94.29155313351498,
+ "grad_norm": 0.6162532567977905,
+ "learning_rate": 1.704216563269434e-07,
+ "loss": 0.0091,
+ "step": 34605
+ },
+ {
+ "epoch": 94.29427792915531,
+ "grad_norm": 0.985260009765625,
+ "learning_rate": 1.7025946458967602e-07,
+ "loss": 0.0109,
+ "step": 34606
+ },
+ {
+ "epoch": 94.29700272479565,
+ "grad_norm": 0.7734326720237732,
+ "learning_rate": 1.70097349406112e-07,
+ "loss": 0.0076,
+ "step": 34607
+ },
+ {
+ "epoch": 94.29972752043597,
+ "grad_norm": 0.9597383737564087,
+ "learning_rate": 1.6993531077751256e-07,
+ "loss": 0.0112,
+ "step": 34608
+ },
+ {
+ "epoch": 94.3024523160763,
+ "grad_norm": 1.045339584350586,
+ "learning_rate": 1.6977334870514007e-07,
+ "loss": 0.0161,
+ "step": 34609
+ },
+ {
+ "epoch": 94.30517711171662,
+ "grad_norm": 1.3966044187545776,
+ "learning_rate": 1.696114631902568e-07,
+ "loss": 0.0528,
+ "step": 34610
+ },
+ {
+ "epoch": 94.30790190735695,
+ "grad_norm": 1.1035420894622803,
+ "learning_rate": 1.6944965423412286e-07,
+ "loss": 0.0665,
+ "step": 34611
+ },
+ {
+ "epoch": 94.31062670299727,
+ "grad_norm": 0.6106114983558655,
+ "learning_rate": 1.6928792183799948e-07,
+ "loss": 0.0053,
+ "step": 34612
+ },
+ {
+ "epoch": 94.3133514986376,
+ "grad_norm": 0.9366812109947205,
+ "learning_rate": 1.6912626600314452e-07,
+ "loss": 0.0128,
+ "step": 34613
+ },
+ {
+ "epoch": 94.31607629427793,
+ "grad_norm": 0.8993019461631775,
+ "learning_rate": 1.6896468673081812e-07,
+ "loss": 0.0125,
+ "step": 34614
+ },
+ {
+ "epoch": 94.31880108991825,
+ "grad_norm": 0.9624632596969604,
+ "learning_rate": 1.6880318402227814e-07,
+ "loss": 0.009,
+ "step": 34615
+ },
+ {
+ "epoch": 94.32152588555859,
+ "grad_norm": 0.8675649166107178,
+ "learning_rate": 1.6864175787878244e-07,
+ "loss": 0.0069,
+ "step": 34616
+ },
+ {
+ "epoch": 94.3242506811989,
+ "grad_norm": 0.8733184933662415,
+ "learning_rate": 1.6848040830158895e-07,
+ "loss": 0.0045,
+ "step": 34617
+ },
+ {
+ "epoch": 94.32697547683924,
+ "grad_norm": 0.8444889783859253,
+ "learning_rate": 1.683191352919533e-07,
+ "loss": 0.0123,
+ "step": 34618
+ },
+ {
+ "epoch": 94.32970027247957,
+ "grad_norm": 0.8552581667900085,
+ "learning_rate": 1.681579388511323e-07,
+ "loss": 0.04,
+ "step": 34619
+ },
+ {
+ "epoch": 94.33242506811989,
+ "grad_norm": 0.8590198159217834,
+ "learning_rate": 1.6799681898038044e-07,
+ "loss": 0.0119,
+ "step": 34620
+ },
+ {
+ "epoch": 94.33514986376022,
+ "grad_norm": 0.909363865852356,
+ "learning_rate": 1.6783577568095123e-07,
+ "loss": 0.0716,
+ "step": 34621
+ },
+ {
+ "epoch": 94.33787465940054,
+ "grad_norm": 1.1217552423477173,
+ "learning_rate": 1.6767480895410248e-07,
+ "loss": 0.0129,
+ "step": 34622
+ },
+ {
+ "epoch": 94.34059945504087,
+ "grad_norm": 0.5792556405067444,
+ "learning_rate": 1.6751391880108548e-07,
+ "loss": 0.0046,
+ "step": 34623
+ },
+ {
+ "epoch": 94.34332425068119,
+ "grad_norm": 1.0021330118179321,
+ "learning_rate": 1.6735310522315473e-07,
+ "loss": 0.0175,
+ "step": 34624
+ },
+ {
+ "epoch": 94.34604904632153,
+ "grad_norm": 1.0581334829330444,
+ "learning_rate": 1.6719236822155927e-07,
+ "loss": 0.0287,
+ "step": 34625
+ },
+ {
+ "epoch": 94.34877384196186,
+ "grad_norm": 0.9769737720489502,
+ "learning_rate": 1.6703170779755474e-07,
+ "loss": 0.0138,
+ "step": 34626
+ },
+ {
+ "epoch": 94.35149863760218,
+ "grad_norm": 0.7524864077568054,
+ "learning_rate": 1.6687112395239014e-07,
+ "loss": 0.0093,
+ "step": 34627
+ },
+ {
+ "epoch": 94.35422343324251,
+ "grad_norm": 0.8354538679122925,
+ "learning_rate": 1.667106166873167e-07,
+ "loss": 0.014,
+ "step": 34628
+ },
+ {
+ "epoch": 94.35694822888283,
+ "grad_norm": 0.9513828754425049,
+ "learning_rate": 1.6655018600358452e-07,
+ "loss": 0.0068,
+ "step": 34629
+ },
+ {
+ "epoch": 94.35967302452316,
+ "grad_norm": 1.0525519847869873,
+ "learning_rate": 1.6638983190244374e-07,
+ "loss": 0.0087,
+ "step": 34630
+ },
+ {
+ "epoch": 94.3623978201635,
+ "grad_norm": 1.0697859525680542,
+ "learning_rate": 1.6622955438514111e-07,
+ "loss": 0.0186,
+ "step": 34631
+ },
+ {
+ "epoch": 94.36512261580381,
+ "grad_norm": 0.9621445536613464,
+ "learning_rate": 1.6606935345292785e-07,
+ "loss": 0.0107,
+ "step": 34632
+ },
+ {
+ "epoch": 94.36784741144415,
+ "grad_norm": 0.6756777167320251,
+ "learning_rate": 1.6590922910704966e-07,
+ "loss": 0.0052,
+ "step": 34633
+ },
+ {
+ "epoch": 94.37057220708446,
+ "grad_norm": 1.0942288637161255,
+ "learning_rate": 1.6574918134875216e-07,
+ "loss": 0.0423,
+ "step": 34634
+ },
+ {
+ "epoch": 94.3732970027248,
+ "grad_norm": 0.8555592894554138,
+ "learning_rate": 1.6558921017928554e-07,
+ "loss": 0.0088,
+ "step": 34635
+ },
+ {
+ "epoch": 94.37602179836512,
+ "grad_norm": 1.8032376766204834,
+ "learning_rate": 1.6542931559989316e-07,
+ "loss": 0.0225,
+ "step": 34636
+ },
+ {
+ "epoch": 94.37874659400545,
+ "grad_norm": 0.5382828116416931,
+ "learning_rate": 1.6526949761182076e-07,
+ "loss": 0.0058,
+ "step": 34637
+ },
+ {
+ "epoch": 94.38147138964578,
+ "grad_norm": 0.7938833236694336,
+ "learning_rate": 1.6510975621631177e-07,
+ "loss": 0.0071,
+ "step": 34638
+ },
+ {
+ "epoch": 94.3841961852861,
+ "grad_norm": 1.2240774631500244,
+ "learning_rate": 1.6495009141461293e-07,
+ "loss": 0.0207,
+ "step": 34639
+ },
+ {
+ "epoch": 94.38692098092643,
+ "grad_norm": 0.987959086894989,
+ "learning_rate": 1.6479050320796552e-07,
+ "loss": 0.008,
+ "step": 34640
+ },
+ {
+ "epoch": 94.38964577656675,
+ "grad_norm": 0.8958355188369751,
+ "learning_rate": 1.646309915976141e-07,
+ "loss": 0.0167,
+ "step": 34641
+ },
+ {
+ "epoch": 94.39237057220708,
+ "grad_norm": 0.6661410927772522,
+ "learning_rate": 1.6447155658479763e-07,
+ "loss": 0.0083,
+ "step": 34642
+ },
+ {
+ "epoch": 94.39509536784742,
+ "grad_norm": 1.2834250926971436,
+ "learning_rate": 1.6431219817076182e-07,
+ "loss": 0.031,
+ "step": 34643
+ },
+ {
+ "epoch": 94.39782016348774,
+ "grad_norm": 0.49183350801467896,
+ "learning_rate": 1.6415291635674567e-07,
+ "loss": 0.0041,
+ "step": 34644
+ },
+ {
+ "epoch": 94.40054495912807,
+ "grad_norm": 1.0330954790115356,
+ "learning_rate": 1.6399371114399043e-07,
+ "loss": 0.0102,
+ "step": 34645
+ },
+ {
+ "epoch": 94.40326975476839,
+ "grad_norm": 1.0745831727981567,
+ "learning_rate": 1.6383458253373508e-07,
+ "loss": 0.0922,
+ "step": 34646
+ },
+ {
+ "epoch": 94.40599455040872,
+ "grad_norm": 0.8755220174789429,
+ "learning_rate": 1.6367553052721973e-07,
+ "loss": 0.0122,
+ "step": 34647
+ },
+ {
+ "epoch": 94.40871934604904,
+ "grad_norm": 1.235651969909668,
+ "learning_rate": 1.6351655512568343e-07,
+ "loss": 0.0117,
+ "step": 34648
+ },
+ {
+ "epoch": 94.41144414168937,
+ "grad_norm": 0.9135094881057739,
+ "learning_rate": 1.6335765633036294e-07,
+ "loss": 0.0101,
+ "step": 34649
+ },
+ {
+ "epoch": 94.4141689373297,
+ "grad_norm": 1.5865031480789185,
+ "learning_rate": 1.6319883414249615e-07,
+ "loss": 0.0489,
+ "step": 34650
+ },
+ {
+ "epoch": 94.41689373297002,
+ "grad_norm": 0.7293276786804199,
+ "learning_rate": 1.63040088563321e-07,
+ "loss": 0.0123,
+ "step": 34651
+ },
+ {
+ "epoch": 94.41961852861036,
+ "grad_norm": 0.8953284025192261,
+ "learning_rate": 1.62881419594072e-07,
+ "loss": 0.1272,
+ "step": 34652
+ },
+ {
+ "epoch": 94.42234332425068,
+ "grad_norm": 1.3466522693634033,
+ "learning_rate": 1.6272282723598709e-07,
+ "loss": 0.0317,
+ "step": 34653
+ },
+ {
+ "epoch": 94.42506811989101,
+ "grad_norm": 0.790352463722229,
+ "learning_rate": 1.625643114902986e-07,
+ "loss": 0.0067,
+ "step": 34654
+ },
+ {
+ "epoch": 94.42779291553134,
+ "grad_norm": 1.1475447416305542,
+ "learning_rate": 1.6240587235824445e-07,
+ "loss": 0.037,
+ "step": 34655
+ },
+ {
+ "epoch": 94.43051771117166,
+ "grad_norm": 0.5465062856674194,
+ "learning_rate": 1.6224750984105586e-07,
+ "loss": 0.0043,
+ "step": 34656
+ },
+ {
+ "epoch": 94.433242506812,
+ "grad_norm": 0.999060869216919,
+ "learning_rate": 1.620892239399674e-07,
+ "loss": 0.0468,
+ "step": 34657
+ },
+ {
+ "epoch": 94.43596730245231,
+ "grad_norm": 0.39915335178375244,
+ "learning_rate": 1.6193101465621031e-07,
+ "loss": 0.0041,
+ "step": 34658
+ },
+ {
+ "epoch": 94.43869209809264,
+ "grad_norm": 0.34470877051353455,
+ "learning_rate": 1.6177288199101915e-07,
+ "loss": 0.0034,
+ "step": 34659
+ },
+ {
+ "epoch": 94.44141689373296,
+ "grad_norm": 0.9003914594650269,
+ "learning_rate": 1.6161482594562404e-07,
+ "loss": 0.0158,
+ "step": 34660
+ },
+ {
+ "epoch": 94.4441416893733,
+ "grad_norm": 0.5063498020172119,
+ "learning_rate": 1.6145684652125625e-07,
+ "loss": 0.0044,
+ "step": 34661
+ },
+ {
+ "epoch": 94.44686648501363,
+ "grad_norm": 1.0354562997817993,
+ "learning_rate": 1.6129894371914368e-07,
+ "loss": 0.0989,
+ "step": 34662
+ },
+ {
+ "epoch": 94.44959128065395,
+ "grad_norm": 0.6949871182441711,
+ "learning_rate": 1.6114111754051976e-07,
+ "loss": 0.0077,
+ "step": 34663
+ },
+ {
+ "epoch": 94.45231607629428,
+ "grad_norm": 1.315464735031128,
+ "learning_rate": 1.609833679866124e-07,
+ "loss": 0.0175,
+ "step": 34664
+ },
+ {
+ "epoch": 94.4550408719346,
+ "grad_norm": 1.191664695739746,
+ "learning_rate": 1.6082569505865064e-07,
+ "loss": 0.0539,
+ "step": 34665
+ },
+ {
+ "epoch": 94.45776566757493,
+ "grad_norm": 0.8655499219894409,
+ "learning_rate": 1.6066809875786016e-07,
+ "loss": 0.0076,
+ "step": 34666
+ },
+ {
+ "epoch": 94.46049046321527,
+ "grad_norm": 1.0011757612228394,
+ "learning_rate": 1.6051057908546997e-07,
+ "loss": 0.0097,
+ "step": 34667
+ },
+ {
+ "epoch": 94.46321525885558,
+ "grad_norm": 0.5609661340713501,
+ "learning_rate": 1.6035313604270798e-07,
+ "loss": 0.0054,
+ "step": 34668
+ },
+ {
+ "epoch": 94.46594005449592,
+ "grad_norm": 1.2939646244049072,
+ "learning_rate": 1.6019576963079875e-07,
+ "loss": 0.0471,
+ "step": 34669
+ },
+ {
+ "epoch": 94.46866485013624,
+ "grad_norm": 0.9147465229034424,
+ "learning_rate": 1.6003847985096687e-07,
+ "loss": 0.0682,
+ "step": 34670
+ },
+ {
+ "epoch": 94.47138964577657,
+ "grad_norm": 0.8248966336250305,
+ "learning_rate": 1.5988126670444026e-07,
+ "loss": 0.01,
+ "step": 34671
+ },
+ {
+ "epoch": 94.47411444141689,
+ "grad_norm": 1.0835182666778564,
+ "learning_rate": 1.5972413019244014e-07,
+ "loss": 0.0299,
+ "step": 34672
+ },
+ {
+ "epoch": 94.47683923705722,
+ "grad_norm": 5.2142229080200195,
+ "learning_rate": 1.5956707031619333e-07,
+ "loss": 0.081,
+ "step": 34673
+ },
+ {
+ "epoch": 94.47956403269755,
+ "grad_norm": 0.8066415190696716,
+ "learning_rate": 1.5941008707691996e-07,
+ "loss": 0.0084,
+ "step": 34674
+ },
+ {
+ "epoch": 94.48228882833787,
+ "grad_norm": 1.0197324752807617,
+ "learning_rate": 1.592531804758446e-07,
+ "loss": 0.0181,
+ "step": 34675
+ },
+ {
+ "epoch": 94.4850136239782,
+ "grad_norm": 1.143264889717102,
+ "learning_rate": 1.590963505141896e-07,
+ "loss": 0.0396,
+ "step": 34676
+ },
+ {
+ "epoch": 94.48773841961852,
+ "grad_norm": 1.5827010869979858,
+ "learning_rate": 1.589395971931751e-07,
+ "loss": 0.0948,
+ "step": 34677
+ },
+ {
+ "epoch": 94.49046321525886,
+ "grad_norm": 1.5002622604370117,
+ "learning_rate": 1.5878292051402123e-07,
+ "loss": 0.0448,
+ "step": 34678
+ },
+ {
+ "epoch": 94.49318801089919,
+ "grad_norm": 1.1096930503845215,
+ "learning_rate": 1.5862632047795033e-07,
+ "loss": 0.0081,
+ "step": 34679
+ },
+ {
+ "epoch": 94.49591280653951,
+ "grad_norm": 0.966236412525177,
+ "learning_rate": 1.5846979708618038e-07,
+ "loss": 0.0314,
+ "step": 34680
+ },
+ {
+ "epoch": 94.49863760217984,
+ "grad_norm": 1.1464959383010864,
+ "learning_rate": 1.5831335033993144e-07,
+ "loss": 0.009,
+ "step": 34681
+ },
+ {
+ "epoch": 94.50136239782016,
+ "grad_norm": 0.9455300569534302,
+ "learning_rate": 1.5815698024042147e-07,
+ "loss": 0.0299,
+ "step": 34682
+ },
+ {
+ "epoch": 94.50408719346049,
+ "grad_norm": 1.8014466762542725,
+ "learning_rate": 1.5800068678886726e-07,
+ "loss": 0.0169,
+ "step": 34683
+ },
+ {
+ "epoch": 94.50681198910081,
+ "grad_norm": 0.862085223197937,
+ "learning_rate": 1.5784446998648784e-07,
+ "loss": 0.0102,
+ "step": 34684
+ },
+ {
+ "epoch": 94.50953678474114,
+ "grad_norm": 1.3346914052963257,
+ "learning_rate": 1.5768832983449888e-07,
+ "loss": 0.0132,
+ "step": 34685
+ },
+ {
+ "epoch": 94.51226158038148,
+ "grad_norm": 1.2748260498046875,
+ "learning_rate": 1.57532266334115e-07,
+ "loss": 0.056,
+ "step": 34686
+ },
+ {
+ "epoch": 94.5149863760218,
+ "grad_norm": 1.4471477270126343,
+ "learning_rate": 1.573762794865541e-07,
+ "loss": 0.2224,
+ "step": 34687
+ },
+ {
+ "epoch": 94.51771117166213,
+ "grad_norm": 0.9408209919929504,
+ "learning_rate": 1.5722036929303075e-07,
+ "loss": 0.0154,
+ "step": 34688
+ },
+ {
+ "epoch": 94.52043596730245,
+ "grad_norm": 0.7485821843147278,
+ "learning_rate": 1.5706453575475733e-07,
+ "loss": 0.0067,
+ "step": 34689
+ },
+ {
+ "epoch": 94.52316076294278,
+ "grad_norm": 1.3281381130218506,
+ "learning_rate": 1.5690877887294843e-07,
+ "loss": 0.0161,
+ "step": 34690
+ },
+ {
+ "epoch": 94.52588555858311,
+ "grad_norm": 1.2304705381393433,
+ "learning_rate": 1.567530986488186e-07,
+ "loss": 0.0123,
+ "step": 34691
+ },
+ {
+ "epoch": 94.52861035422343,
+ "grad_norm": 0.7088093757629395,
+ "learning_rate": 1.5659749508357802e-07,
+ "loss": 0.0063,
+ "step": 34692
+ },
+ {
+ "epoch": 94.53133514986376,
+ "grad_norm": 1.0977038145065308,
+ "learning_rate": 1.5644196817843904e-07,
+ "loss": 0.011,
+ "step": 34693
+ },
+ {
+ "epoch": 94.53405994550408,
+ "grad_norm": 1.1076685190200806,
+ "learning_rate": 1.5628651793461292e-07,
+ "loss": 0.0407,
+ "step": 34694
+ },
+ {
+ "epoch": 94.53678474114442,
+ "grad_norm": 1.2256685495376587,
+ "learning_rate": 1.561311443533109e-07,
+ "loss": 0.0077,
+ "step": 34695
+ },
+ {
+ "epoch": 94.53950953678473,
+ "grad_norm": 1.7200984954833984,
+ "learning_rate": 1.55975847435742e-07,
+ "loss": 0.0751,
+ "step": 34696
+ },
+ {
+ "epoch": 94.54223433242507,
+ "grad_norm": 1.5315364599227905,
+ "learning_rate": 1.558206271831175e-07,
+ "loss": 0.1397,
+ "step": 34697
+ },
+ {
+ "epoch": 94.5449591280654,
+ "grad_norm": 0.818249523639679,
+ "learning_rate": 1.5566548359664423e-07,
+ "loss": 0.0067,
+ "step": 34698
+ },
+ {
+ "epoch": 94.54768392370572,
+ "grad_norm": 0.9281209707260132,
+ "learning_rate": 1.555104166775323e-07,
+ "loss": 0.0876,
+ "step": 34699
+ },
+ {
+ "epoch": 94.55040871934605,
+ "grad_norm": 0.9715766906738281,
+ "learning_rate": 1.5535542642698742e-07,
+ "loss": 0.0125,
+ "step": 34700
+ },
+ {
+ "epoch": 94.55313351498637,
+ "grad_norm": 0.9694222807884216,
+ "learning_rate": 1.5520051284621862e-07,
+ "loss": 0.0129,
+ "step": 34701
+ },
+ {
+ "epoch": 94.5558583106267,
+ "grad_norm": 1.0762524604797363,
+ "learning_rate": 1.550456759364316e-07,
+ "loss": 0.014,
+ "step": 34702
+ },
+ {
+ "epoch": 94.55858310626704,
+ "grad_norm": 0.8768687844276428,
+ "learning_rate": 1.5489091569882986e-07,
+ "loss": 0.0126,
+ "step": 34703
+ },
+ {
+ "epoch": 94.56130790190736,
+ "grad_norm": 0.9233958721160889,
+ "learning_rate": 1.547362321346224e-07,
+ "loss": 0.0155,
+ "step": 34704
+ },
+ {
+ "epoch": 94.56403269754769,
+ "grad_norm": 0.9540348052978516,
+ "learning_rate": 1.5458162524501274e-07,
+ "loss": 0.0113,
+ "step": 34705
+ },
+ {
+ "epoch": 94.566757493188,
+ "grad_norm": 1.1224007606506348,
+ "learning_rate": 1.5442709503120322e-07,
+ "loss": 0.0192,
+ "step": 34706
+ },
+ {
+ "epoch": 94.56948228882834,
+ "grad_norm": 0.9168362617492676,
+ "learning_rate": 1.5427264149439957e-07,
+ "loss": 0.0176,
+ "step": 34707
+ },
+ {
+ "epoch": 94.57220708446866,
+ "grad_norm": 0.519738495349884,
+ "learning_rate": 1.5411826463580414e-07,
+ "loss": 0.0065,
+ "step": 34708
+ },
+ {
+ "epoch": 94.57493188010899,
+ "grad_norm": 1.051548957824707,
+ "learning_rate": 1.5396396445661822e-07,
+ "loss": 0.0097,
+ "step": 34709
+ },
+ {
+ "epoch": 94.57765667574932,
+ "grad_norm": 0.9522570371627808,
+ "learning_rate": 1.5380974095804303e-07,
+ "loss": 0.0117,
+ "step": 34710
+ },
+ {
+ "epoch": 94.58038147138964,
+ "grad_norm": 1.1404117345809937,
+ "learning_rate": 1.5365559414128207e-07,
+ "loss": 0.0094,
+ "step": 34711
+ },
+ {
+ "epoch": 94.58310626702998,
+ "grad_norm": 1.5010104179382324,
+ "learning_rate": 1.5350152400753326e-07,
+ "loss": 0.0462,
+ "step": 34712
+ },
+ {
+ "epoch": 94.5858310626703,
+ "grad_norm": 1.5538623332977295,
+ "learning_rate": 1.533475305579979e-07,
+ "loss": 0.1161,
+ "step": 34713
+ },
+ {
+ "epoch": 94.58855585831063,
+ "grad_norm": 0.8732542991638184,
+ "learning_rate": 1.5319361379387609e-07,
+ "loss": 0.0074,
+ "step": 34714
+ },
+ {
+ "epoch": 94.59128065395096,
+ "grad_norm": 1.079404592514038,
+ "learning_rate": 1.530397737163647e-07,
+ "loss": 0.1226,
+ "step": 34715
+ },
+ {
+ "epoch": 94.59400544959128,
+ "grad_norm": 1.2000950574874878,
+ "learning_rate": 1.5288601032666272e-07,
+ "loss": 0.011,
+ "step": 34716
+ },
+ {
+ "epoch": 94.59673024523161,
+ "grad_norm": 0.9785675406455994,
+ "learning_rate": 1.5273232362596812e-07,
+ "loss": 0.0091,
+ "step": 34717
+ },
+ {
+ "epoch": 94.59945504087193,
+ "grad_norm": 2.5373196601867676,
+ "learning_rate": 1.525787136154766e-07,
+ "loss": 0.0086,
+ "step": 34718
+ },
+ {
+ "epoch": 94.60217983651226,
+ "grad_norm": 1.0426654815673828,
+ "learning_rate": 1.52425180296385e-07,
+ "loss": 0.0179,
+ "step": 34719
+ },
+ {
+ "epoch": 94.60490463215258,
+ "grad_norm": 1.5384323596954346,
+ "learning_rate": 1.5227172366989006e-07,
+ "loss": 0.0123,
+ "step": 34720
+ },
+ {
+ "epoch": 94.60762942779292,
+ "grad_norm": 0.8291800022125244,
+ "learning_rate": 1.5211834373718536e-07,
+ "loss": 0.0077,
+ "step": 34721
+ },
+ {
+ "epoch": 94.61035422343325,
+ "grad_norm": 0.86922687292099,
+ "learning_rate": 1.519650404994666e-07,
+ "loss": 0.0103,
+ "step": 34722
+ },
+ {
+ "epoch": 94.61307901907357,
+ "grad_norm": 1.748979091644287,
+ "learning_rate": 1.518118139579261e-07,
+ "loss": 0.0199,
+ "step": 34723
+ },
+ {
+ "epoch": 94.6158038147139,
+ "grad_norm": 1.1613229513168335,
+ "learning_rate": 1.516586641137585e-07,
+ "loss": 0.0238,
+ "step": 34724
+ },
+ {
+ "epoch": 94.61852861035422,
+ "grad_norm": 1.705859899520874,
+ "learning_rate": 1.515055909681562e-07,
+ "loss": 0.0197,
+ "step": 34725
+ },
+ {
+ "epoch": 94.62125340599455,
+ "grad_norm": 0.9748759269714355,
+ "learning_rate": 1.5135259452231265e-07,
+ "loss": 0.0048,
+ "step": 34726
+ },
+ {
+ "epoch": 94.62397820163488,
+ "grad_norm": 0.6729390621185303,
+ "learning_rate": 1.5119967477741694e-07,
+ "loss": 0.0065,
+ "step": 34727
+ },
+ {
+ "epoch": 94.6267029972752,
+ "grad_norm": 1.5791082382202148,
+ "learning_rate": 1.5104683173466139e-07,
+ "loss": 0.0371,
+ "step": 34728
+ },
+ {
+ "epoch": 94.62942779291554,
+ "grad_norm": 1.2764322757720947,
+ "learning_rate": 1.508940653952362e-07,
+ "loss": 0.0169,
+ "step": 34729
+ },
+ {
+ "epoch": 94.63215258855585,
+ "grad_norm": 0.6947923898696899,
+ "learning_rate": 1.5074137576033044e-07,
+ "loss": 0.0095,
+ "step": 34730
+ },
+ {
+ "epoch": 94.63487738419619,
+ "grad_norm": 0.7777339816093445,
+ "learning_rate": 1.5058876283113423e-07,
+ "loss": 0.0212,
+ "step": 34731
+ },
+ {
+ "epoch": 94.6376021798365,
+ "grad_norm": 0.6861275434494019,
+ "learning_rate": 1.504362266088355e-07,
+ "loss": 0.0106,
+ "step": 34732
+ },
+ {
+ "epoch": 94.64032697547684,
+ "grad_norm": 1.3703460693359375,
+ "learning_rate": 1.5028376709462335e-07,
+ "loss": 0.0433,
+ "step": 34733
+ },
+ {
+ "epoch": 94.64305177111717,
+ "grad_norm": 0.9836257696151733,
+ "learning_rate": 1.5013138428968344e-07,
+ "loss": 0.0147,
+ "step": 34734
+ },
+ {
+ "epoch": 94.64577656675749,
+ "grad_norm": 0.7723977565765381,
+ "learning_rate": 1.4997907819520373e-07,
+ "loss": 0.0156,
+ "step": 34735
+ },
+ {
+ "epoch": 94.64850136239782,
+ "grad_norm": 0.9940891265869141,
+ "learning_rate": 1.4982684881236996e-07,
+ "loss": 0.0384,
+ "step": 34736
+ },
+ {
+ "epoch": 94.65122615803814,
+ "grad_norm": 0.7798472046852112,
+ "learning_rate": 1.4967469614236673e-07,
+ "loss": 0.0049,
+ "step": 34737
+ },
+ {
+ "epoch": 94.65395095367847,
+ "grad_norm": 0.6167153120040894,
+ "learning_rate": 1.4952262018638087e-07,
+ "loss": 0.0052,
+ "step": 34738
+ },
+ {
+ "epoch": 94.65667574931881,
+ "grad_norm": 0.8525026440620422,
+ "learning_rate": 1.4937062094559584e-07,
+ "loss": 0.0105,
+ "step": 34739
+ },
+ {
+ "epoch": 94.65940054495913,
+ "grad_norm": 0.736781656742096,
+ "learning_rate": 1.4921869842119518e-07,
+ "loss": 0.0051,
+ "step": 34740
+ },
+ {
+ "epoch": 94.66212534059946,
+ "grad_norm": 1.3187028169631958,
+ "learning_rate": 1.4906685261436238e-07,
+ "loss": 0.1109,
+ "step": 34741
+ },
+ {
+ "epoch": 94.66485013623978,
+ "grad_norm": 0.8964061141014099,
+ "learning_rate": 1.4891508352627982e-07,
+ "loss": 0.0175,
+ "step": 34742
+ },
+ {
+ "epoch": 94.66757493188011,
+ "grad_norm": 1.2164716720581055,
+ "learning_rate": 1.4876339115812878e-07,
+ "loss": 0.023,
+ "step": 34743
+ },
+ {
+ "epoch": 94.67029972752043,
+ "grad_norm": 1.1761411428451538,
+ "learning_rate": 1.4861177551109273e-07,
+ "loss": 0.0177,
+ "step": 34744
+ },
+ {
+ "epoch": 94.67302452316076,
+ "grad_norm": 0.8780534863471985,
+ "learning_rate": 1.4846023658634967e-07,
+ "loss": 0.0124,
+ "step": 34745
+ },
+ {
+ "epoch": 94.6757493188011,
+ "grad_norm": 1.0542972087860107,
+ "learning_rate": 1.4830877438508308e-07,
+ "loss": 0.0236,
+ "step": 34746
+ },
+ {
+ "epoch": 94.67847411444141,
+ "grad_norm": 1.2261340618133545,
+ "learning_rate": 1.4815738890846864e-07,
+ "loss": 0.0474,
+ "step": 34747
+ },
+ {
+ "epoch": 94.68119891008175,
+ "grad_norm": 1.2100470066070557,
+ "learning_rate": 1.4800608015768768e-07,
+ "loss": 0.055,
+ "step": 34748
+ },
+ {
+ "epoch": 94.68392370572207,
+ "grad_norm": 0.6837366223335266,
+ "learning_rate": 1.4785484813391927e-07,
+ "loss": 0.0069,
+ "step": 34749
+ },
+ {
+ "epoch": 94.6866485013624,
+ "grad_norm": 0.6747808456420898,
+ "learning_rate": 1.477036928383402e-07,
+ "loss": 0.0064,
+ "step": 34750
+ },
+ {
+ "epoch": 94.68937329700273,
+ "grad_norm": 1.5947809219360352,
+ "learning_rate": 1.475526142721262e-07,
+ "loss": 0.0408,
+ "step": 34751
+ },
+ {
+ "epoch": 94.69209809264305,
+ "grad_norm": 1.304197907447815,
+ "learning_rate": 1.4740161243645634e-07,
+ "loss": 0.054,
+ "step": 34752
+ },
+ {
+ "epoch": 94.69482288828338,
+ "grad_norm": 1.2686907052993774,
+ "learning_rate": 1.4725068733250636e-07,
+ "loss": 0.0167,
+ "step": 34753
+ },
+ {
+ "epoch": 94.6975476839237,
+ "grad_norm": 1.159557580947876,
+ "learning_rate": 1.4709983896145085e-07,
+ "loss": 0.0138,
+ "step": 34754
+ },
+ {
+ "epoch": 94.70027247956403,
+ "grad_norm": 0.8277232646942139,
+ "learning_rate": 1.4694906732446334e-07,
+ "loss": 0.015,
+ "step": 34755
+ },
+ {
+ "epoch": 94.70299727520435,
+ "grad_norm": 0.6746438145637512,
+ "learning_rate": 1.4679837242272065e-07,
+ "loss": 0.01,
+ "step": 34756
+ },
+ {
+ "epoch": 94.70572207084469,
+ "grad_norm": 1.5117599964141846,
+ "learning_rate": 1.4664775425739408e-07,
+ "loss": 0.0616,
+ "step": 34757
+ },
+ {
+ "epoch": 94.70844686648502,
+ "grad_norm": 0.8943847417831421,
+ "learning_rate": 1.4649721282965934e-07,
+ "loss": 0.0066,
+ "step": 34758
+ },
+ {
+ "epoch": 94.71117166212534,
+ "grad_norm": 0.9968817830085754,
+ "learning_rate": 1.463467481406855e-07,
+ "loss": 0.0481,
+ "step": 34759
+ },
+ {
+ "epoch": 94.71389645776567,
+ "grad_norm": 1.0107324123382568,
+ "learning_rate": 1.4619636019164608e-07,
+ "loss": 0.0084,
+ "step": 34760
+ },
+ {
+ "epoch": 94.71662125340599,
+ "grad_norm": 1.2163920402526855,
+ "learning_rate": 1.4604604898371344e-07,
+ "loss": 0.0587,
+ "step": 34761
+ },
+ {
+ "epoch": 94.71934604904632,
+ "grad_norm": 0.7295968532562256,
+ "learning_rate": 1.4589581451805666e-07,
+ "loss": 0.0083,
+ "step": 34762
+ },
+ {
+ "epoch": 94.72207084468666,
+ "grad_norm": 0.8953152894973755,
+ "learning_rate": 1.4574565679584485e-07,
+ "loss": 0.0074,
+ "step": 34763
+ },
+ {
+ "epoch": 94.72479564032697,
+ "grad_norm": 1.0156559944152832,
+ "learning_rate": 1.4559557581825034e-07,
+ "loss": 0.0189,
+ "step": 34764
+ },
+ {
+ "epoch": 94.7275204359673,
+ "grad_norm": 1.8064318895339966,
+ "learning_rate": 1.4544557158643892e-07,
+ "loss": 0.0931,
+ "step": 34765
+ },
+ {
+ "epoch": 94.73024523160763,
+ "grad_norm": 0.949104905128479,
+ "learning_rate": 1.452956441015807e-07,
+ "loss": 0.0086,
+ "step": 34766
+ },
+ {
+ "epoch": 94.73297002724796,
+ "grad_norm": 1.137424349784851,
+ "learning_rate": 1.4514579336484368e-07,
+ "loss": 0.0831,
+ "step": 34767
+ },
+ {
+ "epoch": 94.73569482288828,
+ "grad_norm": 1.4664150476455688,
+ "learning_rate": 1.4499601937739249e-07,
+ "loss": 0.0074,
+ "step": 34768
+ },
+ {
+ "epoch": 94.73841961852861,
+ "grad_norm": 0.9410414695739746,
+ "learning_rate": 1.448463221403962e-07,
+ "loss": 0.0207,
+ "step": 34769
+ },
+ {
+ "epoch": 94.74114441416894,
+ "grad_norm": 1.1493085622787476,
+ "learning_rate": 1.4469670165502047e-07,
+ "loss": 0.069,
+ "step": 34770
+ },
+ {
+ "epoch": 94.74386920980926,
+ "grad_norm": 0.855169415473938,
+ "learning_rate": 1.445471579224278e-07,
+ "loss": 0.0081,
+ "step": 34771
+ },
+ {
+ "epoch": 94.7465940054496,
+ "grad_norm": 1.4348429441452026,
+ "learning_rate": 1.4439769094378498e-07,
+ "loss": 0.0579,
+ "step": 34772
+ },
+ {
+ "epoch": 94.74931880108991,
+ "grad_norm": 1.2679741382598877,
+ "learning_rate": 1.4424830072025664e-07,
+ "loss": 0.0168,
+ "step": 34773
+ },
+ {
+ "epoch": 94.75204359673025,
+ "grad_norm": 0.4310300350189209,
+ "learning_rate": 1.440989872530052e-07,
+ "loss": 0.0045,
+ "step": 34774
+ },
+ {
+ "epoch": 94.75476839237058,
+ "grad_norm": 1.3812497854232788,
+ "learning_rate": 1.4394975054319305e-07,
+ "loss": 0.0196,
+ "step": 34775
+ },
+ {
+ "epoch": 94.7574931880109,
+ "grad_norm": 1.4298144578933716,
+ "learning_rate": 1.438005905919837e-07,
+ "loss": 0.0474,
+ "step": 34776
+ },
+ {
+ "epoch": 94.76021798365123,
+ "grad_norm": 0.9622344970703125,
+ "learning_rate": 1.4365150740053735e-07,
+ "loss": 0.0215,
+ "step": 34777
+ },
+ {
+ "epoch": 94.76294277929155,
+ "grad_norm": 2.3926310539245605,
+ "learning_rate": 1.435025009700175e-07,
+ "loss": 0.0102,
+ "step": 34778
+ },
+ {
+ "epoch": 94.76566757493188,
+ "grad_norm": 0.6342161893844604,
+ "learning_rate": 1.43353571301581e-07,
+ "loss": 0.0069,
+ "step": 34779
+ },
+ {
+ "epoch": 94.7683923705722,
+ "grad_norm": 0.844117283821106,
+ "learning_rate": 1.4320471839639027e-07,
+ "loss": 0.0636,
+ "step": 34780
+ },
+ {
+ "epoch": 94.77111716621253,
+ "grad_norm": 0.8272580504417419,
+ "learning_rate": 1.4305594225560438e-07,
+ "loss": 0.0111,
+ "step": 34781
+ },
+ {
+ "epoch": 94.77384196185287,
+ "grad_norm": 1.036089301109314,
+ "learning_rate": 1.4290724288038016e-07,
+ "loss": 0.0131,
+ "step": 34782
+ },
+ {
+ "epoch": 94.77656675749319,
+ "grad_norm": 0.775672972202301,
+ "learning_rate": 1.4275862027187893e-07,
+ "loss": 0.0087,
+ "step": 34783
+ },
+ {
+ "epoch": 94.77929155313352,
+ "grad_norm": 1.2521331310272217,
+ "learning_rate": 1.4261007443125418e-07,
+ "loss": 0.0285,
+ "step": 34784
+ },
+ {
+ "epoch": 94.78201634877384,
+ "grad_norm": 0.7728263139724731,
+ "learning_rate": 1.4246160535966725e-07,
+ "loss": 0.0065,
+ "step": 34785
+ },
+ {
+ "epoch": 94.78474114441417,
+ "grad_norm": 1.1436433792114258,
+ "learning_rate": 1.423132130582705e-07,
+ "loss": 0.0321,
+ "step": 34786
+ },
+ {
+ "epoch": 94.7874659400545,
+ "grad_norm": 1.1660547256469727,
+ "learning_rate": 1.4216489752822192e-07,
+ "loss": 0.0876,
+ "step": 34787
+ },
+ {
+ "epoch": 94.79019073569482,
+ "grad_norm": 0.7602578401565552,
+ "learning_rate": 1.4201665877067615e-07,
+ "loss": 0.0068,
+ "step": 34788
+ },
+ {
+ "epoch": 94.79291553133515,
+ "grad_norm": 0.7650889158248901,
+ "learning_rate": 1.418684967867867e-07,
+ "loss": 0.0183,
+ "step": 34789
+ },
+ {
+ "epoch": 94.79564032697547,
+ "grad_norm": 1.3863164186477661,
+ "learning_rate": 1.4172041157770822e-07,
+ "loss": 0.0291,
+ "step": 34790
+ },
+ {
+ "epoch": 94.7983651226158,
+ "grad_norm": 1.1983598470687866,
+ "learning_rate": 1.415724031445942e-07,
+ "loss": 0.0081,
+ "step": 34791
+ },
+ {
+ "epoch": 94.80108991825612,
+ "grad_norm": 0.8047984838485718,
+ "learning_rate": 1.4142447148859706e-07,
+ "loss": 0.0038,
+ "step": 34792
+ },
+ {
+ "epoch": 94.80381471389646,
+ "grad_norm": 0.8945087790489197,
+ "learning_rate": 1.412766166108681e-07,
+ "loss": 0.0174,
+ "step": 34793
+ },
+ {
+ "epoch": 94.80653950953679,
+ "grad_norm": 1.074729561805725,
+ "learning_rate": 1.4112883851256088e-07,
+ "loss": 0.0113,
+ "step": 34794
+ },
+ {
+ "epoch": 94.80926430517711,
+ "grad_norm": 1.3682342767715454,
+ "learning_rate": 1.4098113719482333e-07,
+ "loss": 0.0258,
+ "step": 34795
+ },
+ {
+ "epoch": 94.81198910081744,
+ "grad_norm": 0.6620335578918457,
+ "learning_rate": 1.4083351265880786e-07,
+ "loss": 0.007,
+ "step": 34796
+ },
+ {
+ "epoch": 94.81471389645776,
+ "grad_norm": 0.68452388048172,
+ "learning_rate": 1.4068596490566467e-07,
+ "loss": 0.0071,
+ "step": 34797
+ },
+ {
+ "epoch": 94.8174386920981,
+ "grad_norm": 1.1961950063705444,
+ "learning_rate": 1.405384939365406e-07,
+ "loss": 0.0794,
+ "step": 34798
+ },
+ {
+ "epoch": 94.82016348773843,
+ "grad_norm": 1.3106296062469482,
+ "learning_rate": 1.403910997525859e-07,
+ "loss": 0.0259,
+ "step": 34799
+ },
+ {
+ "epoch": 94.82288828337875,
+ "grad_norm": 1.1097543239593506,
+ "learning_rate": 1.402437823549474e-07,
+ "loss": 0.0264,
+ "step": 34800
+ },
+ {
+ "epoch": 94.82561307901908,
+ "grad_norm": 0.9677045941352844,
+ "learning_rate": 1.4009654174477417e-07,
+ "loss": 0.0129,
+ "step": 34801
+ },
+ {
+ "epoch": 94.8283378746594,
+ "grad_norm": 1.2904528379440308,
+ "learning_rate": 1.3994937792321085e-07,
+ "loss": 0.0133,
+ "step": 34802
+ },
+ {
+ "epoch": 94.83106267029973,
+ "grad_norm": 0.4911992847919464,
+ "learning_rate": 1.398022908914054e-07,
+ "loss": 0.0053,
+ "step": 34803
+ },
+ {
+ "epoch": 94.83378746594005,
+ "grad_norm": 1.2375723123550415,
+ "learning_rate": 1.3965528065050137e-07,
+ "loss": 0.0106,
+ "step": 34804
+ },
+ {
+ "epoch": 94.83651226158038,
+ "grad_norm": 0.7456478476524353,
+ "learning_rate": 1.3950834720164453e-07,
+ "loss": 0.0069,
+ "step": 34805
+ },
+ {
+ "epoch": 94.83923705722071,
+ "grad_norm": 1.509171962738037,
+ "learning_rate": 1.3936149054598058e-07,
+ "loss": 0.1613,
+ "step": 34806
+ },
+ {
+ "epoch": 94.84196185286103,
+ "grad_norm": 1.675751805305481,
+ "learning_rate": 1.3921471068465085e-07,
+ "loss": 0.0111,
+ "step": 34807
+ },
+ {
+ "epoch": 94.84468664850137,
+ "grad_norm": 1.1365553140640259,
+ "learning_rate": 1.390680076188e-07,
+ "loss": 0.0055,
+ "step": 34808
+ },
+ {
+ "epoch": 94.84741144414168,
+ "grad_norm": 1.7103031873703003,
+ "learning_rate": 1.3892138134957045e-07,
+ "loss": 0.0153,
+ "step": 34809
+ },
+ {
+ "epoch": 94.85013623978202,
+ "grad_norm": 0.8294908404350281,
+ "learning_rate": 1.3877483187810348e-07,
+ "loss": 0.0057,
+ "step": 34810
+ },
+ {
+ "epoch": 94.85286103542235,
+ "grad_norm": 0.6550543308258057,
+ "learning_rate": 1.386283592055404e-07,
+ "loss": 0.0087,
+ "step": 34811
+ },
+ {
+ "epoch": 94.85558583106267,
+ "grad_norm": 0.876032292842865,
+ "learning_rate": 1.3848196333302255e-07,
+ "loss": 0.0056,
+ "step": 34812
+ },
+ {
+ "epoch": 94.858310626703,
+ "grad_norm": 0.7443567514419556,
+ "learning_rate": 1.383356442616901e-07,
+ "loss": 0.0096,
+ "step": 34813
+ },
+ {
+ "epoch": 94.86103542234332,
+ "grad_norm": 1.2821893692016602,
+ "learning_rate": 1.3818940199268216e-07,
+ "loss": 0.0158,
+ "step": 34814
+ },
+ {
+ "epoch": 94.86376021798365,
+ "grad_norm": 1.1505436897277832,
+ "learning_rate": 1.380432365271378e-07,
+ "loss": 0.0368,
+ "step": 34815
+ },
+ {
+ "epoch": 94.86648501362397,
+ "grad_norm": 1.2051005363464355,
+ "learning_rate": 1.378971478661939e-07,
+ "loss": 0.023,
+ "step": 34816
+ },
+ {
+ "epoch": 94.8692098092643,
+ "grad_norm": 1.2443115711212158,
+ "learning_rate": 1.3775113601099067e-07,
+ "loss": 0.0753,
+ "step": 34817
+ },
+ {
+ "epoch": 94.87193460490464,
+ "grad_norm": 1.3176660537719727,
+ "learning_rate": 1.3760520096266494e-07,
+ "loss": 0.0306,
+ "step": 34818
+ },
+ {
+ "epoch": 94.87465940054496,
+ "grad_norm": 1.4987415075302124,
+ "learning_rate": 1.374593427223514e-07,
+ "loss": 0.0837,
+ "step": 34819
+ },
+ {
+ "epoch": 94.87738419618529,
+ "grad_norm": 1.0276464223861694,
+ "learning_rate": 1.373135612911869e-07,
+ "loss": 0.0106,
+ "step": 34820
+ },
+ {
+ "epoch": 94.88010899182561,
+ "grad_norm": 0.8084632158279419,
+ "learning_rate": 1.371678566703083e-07,
+ "loss": 0.0108,
+ "step": 34821
+ },
+ {
+ "epoch": 94.88283378746594,
+ "grad_norm": 1.3334424495697021,
+ "learning_rate": 1.3702222886084803e-07,
+ "loss": 0.1078,
+ "step": 34822
+ },
+ {
+ "epoch": 94.88555858310627,
+ "grad_norm": 0.32680371403694153,
+ "learning_rate": 1.3687667786394188e-07,
+ "loss": 0.0032,
+ "step": 34823
+ },
+ {
+ "epoch": 94.88828337874659,
+ "grad_norm": 0.5410128235816956,
+ "learning_rate": 1.3673120368072224e-07,
+ "loss": 0.0058,
+ "step": 34824
+ },
+ {
+ "epoch": 94.89100817438693,
+ "grad_norm": 1.3396881818771362,
+ "learning_rate": 1.3658580631232266e-07,
+ "loss": 0.02,
+ "step": 34825
+ },
+ {
+ "epoch": 94.89373297002724,
+ "grad_norm": 2.440096378326416,
+ "learning_rate": 1.3644048575987557e-07,
+ "loss": 0.0103,
+ "step": 34826
+ },
+ {
+ "epoch": 94.89645776566758,
+ "grad_norm": 0.7502179741859436,
+ "learning_rate": 1.3629524202451228e-07,
+ "loss": 0.0079,
+ "step": 34827
+ },
+ {
+ "epoch": 94.8991825613079,
+ "grad_norm": 1.375289797782898,
+ "learning_rate": 1.36150075107363e-07,
+ "loss": 0.0085,
+ "step": 34828
+ },
+ {
+ "epoch": 94.90190735694823,
+ "grad_norm": 1.0794705152511597,
+ "learning_rate": 1.3600498500956129e-07,
+ "loss": 0.0231,
+ "step": 34829
+ },
+ {
+ "epoch": 94.90463215258856,
+ "grad_norm": 0.7906955480575562,
+ "learning_rate": 1.3585997173223396e-07,
+ "loss": 0.0081,
+ "step": 34830
+ },
+ {
+ "epoch": 94.90735694822888,
+ "grad_norm": 1.2408907413482666,
+ "learning_rate": 1.3571503527651352e-07,
+ "loss": 0.0128,
+ "step": 34831
+ },
+ {
+ "epoch": 94.91008174386921,
+ "grad_norm": 1.6451284885406494,
+ "learning_rate": 1.355701756435246e-07,
+ "loss": 0.0337,
+ "step": 34832
+ },
+ {
+ "epoch": 94.91280653950953,
+ "grad_norm": 0.6979427337646484,
+ "learning_rate": 1.3542539283439959e-07,
+ "loss": 0.0055,
+ "step": 34833
+ },
+ {
+ "epoch": 94.91553133514986,
+ "grad_norm": 0.9326317310333252,
+ "learning_rate": 1.3528068685026208e-07,
+ "loss": 0.0095,
+ "step": 34834
+ },
+ {
+ "epoch": 94.9182561307902,
+ "grad_norm": 0.9797750115394592,
+ "learning_rate": 1.3513605769224225e-07,
+ "loss": 0.0355,
+ "step": 34835
+ },
+ {
+ "epoch": 94.92098092643052,
+ "grad_norm": 1.0745896100997925,
+ "learning_rate": 1.3499150536146588e-07,
+ "loss": 0.0665,
+ "step": 34836
+ },
+ {
+ "epoch": 94.92370572207085,
+ "grad_norm": 0.7395225167274475,
+ "learning_rate": 1.3484702985905763e-07,
+ "loss": 0.0055,
+ "step": 34837
+ },
+ {
+ "epoch": 94.92643051771117,
+ "grad_norm": 0.4829704463481903,
+ "learning_rate": 1.3470263118614324e-07,
+ "loss": 0.0045,
+ "step": 34838
+ },
+ {
+ "epoch": 94.9291553133515,
+ "grad_norm": 0.9965943098068237,
+ "learning_rate": 1.345583093438474e-07,
+ "loss": 0.012,
+ "step": 34839
+ },
+ {
+ "epoch": 94.93188010899182,
+ "grad_norm": 1.1925249099731445,
+ "learning_rate": 1.3441406433329362e-07,
+ "loss": 0.0087,
+ "step": 34840
+ },
+ {
+ "epoch": 94.93460490463215,
+ "grad_norm": 0.8737728595733643,
+ "learning_rate": 1.3426989615560548e-07,
+ "loss": 0.0284,
+ "step": 34841
+ },
+ {
+ "epoch": 94.93732970027249,
+ "grad_norm": 0.6489598751068115,
+ "learning_rate": 1.3412580481190651e-07,
+ "loss": 0.0062,
+ "step": 34842
+ },
+ {
+ "epoch": 94.9400544959128,
+ "grad_norm": 1.095517873764038,
+ "learning_rate": 1.3398179030331805e-07,
+ "loss": 0.0982,
+ "step": 34843
+ },
+ {
+ "epoch": 94.94277929155314,
+ "grad_norm": 1.2769657373428345,
+ "learning_rate": 1.3383785263096138e-07,
+ "loss": 0.0471,
+ "step": 34844
+ },
+ {
+ "epoch": 94.94550408719346,
+ "grad_norm": 2.484668731689453,
+ "learning_rate": 1.3369399179595787e-07,
+ "loss": 0.0224,
+ "step": 34845
+ },
+ {
+ "epoch": 94.94822888283379,
+ "grad_norm": 3.4520227909088135,
+ "learning_rate": 1.3355020779942883e-07,
+ "loss": 0.0106,
+ "step": 34846
+ },
+ {
+ "epoch": 94.95095367847412,
+ "grad_norm": 1.2012732028961182,
+ "learning_rate": 1.3340650064249337e-07,
+ "loss": 0.0147,
+ "step": 34847
+ },
+ {
+ "epoch": 94.95367847411444,
+ "grad_norm": 1.3484669923782349,
+ "learning_rate": 1.3326287032626838e-07,
+ "loss": 0.1685,
+ "step": 34848
+ },
+ {
+ "epoch": 94.95640326975477,
+ "grad_norm": 0.7204333543777466,
+ "learning_rate": 1.331193168518774e-07,
+ "loss": 0.0066,
+ "step": 34849
+ },
+ {
+ "epoch": 94.95912806539509,
+ "grad_norm": 1.223780632019043,
+ "learning_rate": 1.3297584022043396e-07,
+ "loss": 0.0103,
+ "step": 34850
+ },
+ {
+ "epoch": 94.96185286103542,
+ "grad_norm": 0.6735231280326843,
+ "learning_rate": 1.328324404330572e-07,
+ "loss": 0.0056,
+ "step": 34851
+ },
+ {
+ "epoch": 94.96457765667574,
+ "grad_norm": 1.6547911167144775,
+ "learning_rate": 1.32689117490864e-07,
+ "loss": 0.0338,
+ "step": 34852
+ },
+ {
+ "epoch": 94.96730245231608,
+ "grad_norm": 0.7004072666168213,
+ "learning_rate": 1.325458713949701e-07,
+ "loss": 0.0112,
+ "step": 34853
+ },
+ {
+ "epoch": 94.97002724795641,
+ "grad_norm": 1.205323576927185,
+ "learning_rate": 1.3240270214649242e-07,
+ "loss": 0.0175,
+ "step": 34854
+ },
+ {
+ "epoch": 94.97275204359673,
+ "grad_norm": 1.5960785150527954,
+ "learning_rate": 1.322596097465445e-07,
+ "loss": 0.0493,
+ "step": 34855
+ },
+ {
+ "epoch": 94.97547683923706,
+ "grad_norm": 0.7014725208282471,
+ "learning_rate": 1.3211659419624102e-07,
+ "loss": 0.0075,
+ "step": 34856
+ },
+ {
+ "epoch": 94.97820163487738,
+ "grad_norm": 0.7987989783287048,
+ "learning_rate": 1.319736554966955e-07,
+ "loss": 0.0078,
+ "step": 34857
+ },
+ {
+ "epoch": 94.98092643051771,
+ "grad_norm": 1.5453985929489136,
+ "learning_rate": 1.3183079364902153e-07,
+ "loss": 0.0102,
+ "step": 34858
+ },
+ {
+ "epoch": 94.98365122615803,
+ "grad_norm": 0.8482023477554321,
+ "learning_rate": 1.3168800865433262e-07,
+ "loss": 0.0162,
+ "step": 34859
+ },
+ {
+ "epoch": 94.98637602179836,
+ "grad_norm": 1.2591350078582764,
+ "learning_rate": 1.3154530051373904e-07,
+ "loss": 0.0155,
+ "step": 34860
+ },
+ {
+ "epoch": 94.9891008174387,
+ "grad_norm": 2.176868438720703,
+ "learning_rate": 1.3140266922835322e-07,
+ "loss": 0.0453,
+ "step": 34861
+ },
+ {
+ "epoch": 94.99182561307902,
+ "grad_norm": 0.6126188039779663,
+ "learning_rate": 1.3126011479928646e-07,
+ "loss": 0.0068,
+ "step": 34862
+ },
+ {
+ "epoch": 94.99455040871935,
+ "grad_norm": 0.8187693953514099,
+ "learning_rate": 1.3111763722764792e-07,
+ "loss": 0.0076,
+ "step": 34863
+ },
+ {
+ "epoch": 94.99727520435967,
+ "grad_norm": 0.9478728175163269,
+ "learning_rate": 1.3097523651454668e-07,
+ "loss": 0.0122,
+ "step": 34864
+ },
+ {
+ "epoch": 95.0,
+ "grad_norm": 1.0832983255386353,
+ "learning_rate": 1.30832912661093e-07,
+ "loss": 0.01,
+ "step": 34865
+ },
+ {
+ "epoch": 95.00272479564033,
+ "grad_norm": 1.7438350915908813,
+ "learning_rate": 1.3069066566839594e-07,
+ "loss": 0.0607,
+ "step": 34866
+ },
+ {
+ "epoch": 95.00544959128065,
+ "grad_norm": 1.2963237762451172,
+ "learning_rate": 1.3054849553756243e-07,
+ "loss": 0.012,
+ "step": 34867
+ },
+ {
+ "epoch": 95.00817438692098,
+ "grad_norm": 1.0145472288131714,
+ "learning_rate": 1.3040640226969824e-07,
+ "loss": 0.0125,
+ "step": 34868
+ },
+ {
+ "epoch": 95.0108991825613,
+ "grad_norm": 1.2883012294769287,
+ "learning_rate": 1.3026438586591138e-07,
+ "loss": 0.0255,
+ "step": 34869
+ },
+ {
+ "epoch": 95.01362397820164,
+ "grad_norm": 1.8783804178237915,
+ "learning_rate": 1.3012244632730876e-07,
+ "loss": 0.0228,
+ "step": 34870
+ },
+ {
+ "epoch": 95.01634877384195,
+ "grad_norm": 1.1169592142105103,
+ "learning_rate": 1.29980583654995e-07,
+ "loss": 0.0504,
+ "step": 34871
+ },
+ {
+ "epoch": 95.01907356948229,
+ "grad_norm": 0.5929177403450012,
+ "learning_rate": 1.298387978500737e-07,
+ "loss": 0.0043,
+ "step": 34872
+ },
+ {
+ "epoch": 95.02179836512262,
+ "grad_norm": 0.6045824885368347,
+ "learning_rate": 1.2969708891365063e-07,
+ "loss": 0.0099,
+ "step": 34873
+ },
+ {
+ "epoch": 95.02452316076294,
+ "grad_norm": 0.8695037961006165,
+ "learning_rate": 1.2955545684682935e-07,
+ "loss": 0.0113,
+ "step": 34874
+ },
+ {
+ "epoch": 95.02724795640327,
+ "grad_norm": 0.46376410126686096,
+ "learning_rate": 1.2941390165071232e-07,
+ "loss": 0.004,
+ "step": 34875
+ },
+ {
+ "epoch": 95.02997275204359,
+ "grad_norm": 1.0929380655288696,
+ "learning_rate": 1.292724233264031e-07,
+ "loss": 0.0096,
+ "step": 34876
+ },
+ {
+ "epoch": 95.03269754768392,
+ "grad_norm": 0.44876348972320557,
+ "learning_rate": 1.2913102187500083e-07,
+ "loss": 0.0045,
+ "step": 34877
+ },
+ {
+ "epoch": 95.03542234332426,
+ "grad_norm": 3.0275933742523193,
+ "learning_rate": 1.2898969729760903e-07,
+ "loss": 0.1636,
+ "step": 34878
+ },
+ {
+ "epoch": 95.03814713896458,
+ "grad_norm": 0.8979824185371399,
+ "learning_rate": 1.2884844959532795e-07,
+ "loss": 0.0079,
+ "step": 34879
+ },
+ {
+ "epoch": 95.04087193460491,
+ "grad_norm": 1.1750354766845703,
+ "learning_rate": 1.287072787692567e-07,
+ "loss": 0.0655,
+ "step": 34880
+ },
+ {
+ "epoch": 95.04359673024523,
+ "grad_norm": 1.3134732246398926,
+ "learning_rate": 1.2856618482049664e-07,
+ "loss": 0.0915,
+ "step": 34881
+ },
+ {
+ "epoch": 95.04632152588556,
+ "grad_norm": 0.9228324294090271,
+ "learning_rate": 1.284251677501458e-07,
+ "loss": 0.0069,
+ "step": 34882
+ },
+ {
+ "epoch": 95.04904632152588,
+ "grad_norm": 0.8272329568862915,
+ "learning_rate": 1.2828422755930103e-07,
+ "loss": 0.0131,
+ "step": 34883
+ },
+ {
+ "epoch": 95.05177111716621,
+ "grad_norm": 1.265885829925537,
+ "learning_rate": 1.2814336424906038e-07,
+ "loss": 0.0201,
+ "step": 34884
+ },
+ {
+ "epoch": 95.05449591280654,
+ "grad_norm": 1.1938129663467407,
+ "learning_rate": 1.2800257782052183e-07,
+ "loss": 0.0259,
+ "step": 34885
+ },
+ {
+ "epoch": 95.05722070844686,
+ "grad_norm": 0.9718248844146729,
+ "learning_rate": 1.2786186827478231e-07,
+ "loss": 0.006,
+ "step": 34886
+ },
+ {
+ "epoch": 95.0599455040872,
+ "grad_norm": 0.8364661335945129,
+ "learning_rate": 1.2772123561293648e-07,
+ "loss": 0.005,
+ "step": 34887
+ },
+ {
+ "epoch": 95.06267029972751,
+ "grad_norm": 0.8991084694862366,
+ "learning_rate": 1.2758067983608013e-07,
+ "loss": 0.0682,
+ "step": 34888
+ },
+ {
+ "epoch": 95.06539509536785,
+ "grad_norm": 1.6104437112808228,
+ "learning_rate": 1.2744020094530684e-07,
+ "loss": 0.0298,
+ "step": 34889
+ },
+ {
+ "epoch": 95.06811989100818,
+ "grad_norm": 1.5108712911605835,
+ "learning_rate": 1.272997989417124e-07,
+ "loss": 0.0326,
+ "step": 34890
+ },
+ {
+ "epoch": 95.0708446866485,
+ "grad_norm": 0.9809255003929138,
+ "learning_rate": 1.2715947382638927e-07,
+ "loss": 0.0103,
+ "step": 34891
+ },
+ {
+ "epoch": 95.07356948228883,
+ "grad_norm": 2.160085678100586,
+ "learning_rate": 1.27019225600431e-07,
+ "loss": 0.039,
+ "step": 34892
+ },
+ {
+ "epoch": 95.07629427792915,
+ "grad_norm": 1.3197431564331055,
+ "learning_rate": 1.2687905426492786e-07,
+ "loss": 0.1233,
+ "step": 34893
+ },
+ {
+ "epoch": 95.07901907356948,
+ "grad_norm": 1.3987102508544922,
+ "learning_rate": 1.267389598209734e-07,
+ "loss": 0.0341,
+ "step": 34894
+ },
+ {
+ "epoch": 95.0817438692098,
+ "grad_norm": 1.429640531539917,
+ "learning_rate": 1.2659894226965896e-07,
+ "loss": 0.0123,
+ "step": 34895
+ },
+ {
+ "epoch": 95.08446866485014,
+ "grad_norm": 1.4893227815628052,
+ "learning_rate": 1.264590016120737e-07,
+ "loss": 0.0752,
+ "step": 34896
+ },
+ {
+ "epoch": 95.08719346049047,
+ "grad_norm": 0.8330496549606323,
+ "learning_rate": 1.2631913784930673e-07,
+ "loss": 0.0092,
+ "step": 34897
+ },
+ {
+ "epoch": 95.08991825613079,
+ "grad_norm": 1.3296852111816406,
+ "learning_rate": 1.261793509824505e-07,
+ "loss": 0.0361,
+ "step": 34898
+ },
+ {
+ "epoch": 95.09264305177112,
+ "grad_norm": 0.9310983419418335,
+ "learning_rate": 1.260396410125908e-07,
+ "loss": 0.0122,
+ "step": 34899
+ },
+ {
+ "epoch": 95.09536784741144,
+ "grad_norm": 1.9924851655960083,
+ "learning_rate": 1.2590000794081681e-07,
+ "loss": 0.0309,
+ "step": 34900
+ },
+ {
+ "epoch": 95.09809264305177,
+ "grad_norm": 1.3644193410873413,
+ "learning_rate": 1.2576045176821429e-07,
+ "loss": 0.0282,
+ "step": 34901
+ },
+ {
+ "epoch": 95.1008174386921,
+ "grad_norm": 1.0366102457046509,
+ "learning_rate": 1.2562097249587235e-07,
+ "loss": 0.0146,
+ "step": 34902
+ },
+ {
+ "epoch": 95.10354223433242,
+ "grad_norm": 1.4036293029785156,
+ "learning_rate": 1.2548157012487684e-07,
+ "loss": 0.0118,
+ "step": 34903
+ },
+ {
+ "epoch": 95.10626702997276,
+ "grad_norm": 1.7031922340393066,
+ "learning_rate": 1.253422446563135e-07,
+ "loss": 0.0425,
+ "step": 34904
+ },
+ {
+ "epoch": 95.10899182561307,
+ "grad_norm": 1.1380873918533325,
+ "learning_rate": 1.2520299609126597e-07,
+ "loss": 0.062,
+ "step": 34905
+ },
+ {
+ "epoch": 95.11171662125341,
+ "grad_norm": 0.5005462765693665,
+ "learning_rate": 1.250638244308211e-07,
+ "loss": 0.0075,
+ "step": 34906
+ },
+ {
+ "epoch": 95.11444141689373,
+ "grad_norm": 1.2038224935531616,
+ "learning_rate": 1.2492472967606028e-07,
+ "loss": 0.0101,
+ "step": 34907
+ },
+ {
+ "epoch": 95.11716621253406,
+ "grad_norm": 0.7755666971206665,
+ "learning_rate": 1.247857118280682e-07,
+ "loss": 0.008,
+ "step": 34908
+ },
+ {
+ "epoch": 95.11989100817439,
+ "grad_norm": 0.5514383912086487,
+ "learning_rate": 1.246467708879262e-07,
+ "loss": 0.0058,
+ "step": 34909
+ },
+ {
+ "epoch": 95.12261580381471,
+ "grad_norm": 1.286802053451538,
+ "learning_rate": 1.2450790685671898e-07,
+ "loss": 0.0144,
+ "step": 34910
+ },
+ {
+ "epoch": 95.12534059945504,
+ "grad_norm": 0.3285485804080963,
+ "learning_rate": 1.2436911973552567e-07,
+ "loss": 0.003,
+ "step": 34911
+ },
+ {
+ "epoch": 95.12806539509536,
+ "grad_norm": 0.8725586533546448,
+ "learning_rate": 1.2423040952542764e-07,
+ "loss": 0.037,
+ "step": 34912
+ },
+ {
+ "epoch": 95.1307901907357,
+ "grad_norm": 1.64523184299469,
+ "learning_rate": 1.2409177622750511e-07,
+ "loss": 0.0139,
+ "step": 34913
+ },
+ {
+ "epoch": 95.13351498637603,
+ "grad_norm": 0.6452071070671082,
+ "learning_rate": 1.2395321984283836e-07,
+ "loss": 0.0062,
+ "step": 34914
+ },
+ {
+ "epoch": 95.13623978201635,
+ "grad_norm": 1.2600250244140625,
+ "learning_rate": 1.238147403725054e-07,
+ "loss": 0.016,
+ "step": 34915
+ },
+ {
+ "epoch": 95.13896457765668,
+ "grad_norm": 0.8971394896507263,
+ "learning_rate": 1.236763378175865e-07,
+ "loss": 0.0114,
+ "step": 34916
+ },
+ {
+ "epoch": 95.141689373297,
+ "grad_norm": 1.2799994945526123,
+ "learning_rate": 1.235380121791574e-07,
+ "loss": 0.0093,
+ "step": 34917
+ },
+ {
+ "epoch": 95.14441416893733,
+ "grad_norm": 1.4970656633377075,
+ "learning_rate": 1.233997634582973e-07,
+ "loss": 0.0185,
+ "step": 34918
+ },
+ {
+ "epoch": 95.14713896457765,
+ "grad_norm": 2.1435868740081787,
+ "learning_rate": 1.2326159165608086e-07,
+ "loss": 0.0205,
+ "step": 34919
+ },
+ {
+ "epoch": 95.14986376021798,
+ "grad_norm": 1.5739686489105225,
+ "learning_rate": 1.2312349677358615e-07,
+ "loss": 0.0242,
+ "step": 34920
+ },
+ {
+ "epoch": 95.15258855585832,
+ "grad_norm": 0.3478407561779022,
+ "learning_rate": 1.2298547881188672e-07,
+ "loss": 0.0029,
+ "step": 34921
+ },
+ {
+ "epoch": 95.15531335149863,
+ "grad_norm": 0.7020912170410156,
+ "learning_rate": 1.228475377720595e-07,
+ "loss": 0.0099,
+ "step": 34922
+ },
+ {
+ "epoch": 95.15803814713897,
+ "grad_norm": 0.41511955857276917,
+ "learning_rate": 1.2270967365517694e-07,
+ "loss": 0.0061,
+ "step": 34923
+ },
+ {
+ "epoch": 95.16076294277929,
+ "grad_norm": 1.9592636823654175,
+ "learning_rate": 1.2257188646231487e-07,
+ "loss": 0.0223,
+ "step": 34924
+ },
+ {
+ "epoch": 95.16348773841962,
+ "grad_norm": 1.2062309980392456,
+ "learning_rate": 1.2243417619454356e-07,
+ "loss": 0.0116,
+ "step": 34925
+ },
+ {
+ "epoch": 95.16621253405995,
+ "grad_norm": 0.7851657271385193,
+ "learning_rate": 1.2229654285293767e-07,
+ "loss": 0.0058,
+ "step": 34926
+ },
+ {
+ "epoch": 95.16893732970027,
+ "grad_norm": 1.226921558380127,
+ "learning_rate": 1.221589864385686e-07,
+ "loss": 0.0313,
+ "step": 34927
+ },
+ {
+ "epoch": 95.1716621253406,
+ "grad_norm": 0.4852854907512665,
+ "learning_rate": 1.220215069525077e-07,
+ "loss": 0.0051,
+ "step": 34928
+ },
+ {
+ "epoch": 95.17438692098092,
+ "grad_norm": 0.801765501499176,
+ "learning_rate": 1.218841043958241e-07,
+ "loss": 0.0122,
+ "step": 34929
+ },
+ {
+ "epoch": 95.17711171662125,
+ "grad_norm": 1.030411720275879,
+ "learning_rate": 1.217467787695903e-07,
+ "loss": 0.011,
+ "step": 34930
+ },
+ {
+ "epoch": 95.17983651226157,
+ "grad_norm": 1.0711168050765991,
+ "learning_rate": 1.216095300748743e-07,
+ "loss": 0.0991,
+ "step": 34931
+ },
+ {
+ "epoch": 95.1825613079019,
+ "grad_norm": 0.6399945616722107,
+ "learning_rate": 1.214723583127464e-07,
+ "loss": 0.0065,
+ "step": 34932
+ },
+ {
+ "epoch": 95.18528610354224,
+ "grad_norm": 0.860160768032074,
+ "learning_rate": 1.2133526348427237e-07,
+ "loss": 0.109,
+ "step": 34933
+ },
+ {
+ "epoch": 95.18801089918256,
+ "grad_norm": 0.8961932063102722,
+ "learning_rate": 1.211982455905214e-07,
+ "loss": 0.0164,
+ "step": 34934
+ },
+ {
+ "epoch": 95.19073569482289,
+ "grad_norm": 1.1050232648849487,
+ "learning_rate": 1.210613046325615e-07,
+ "loss": 0.01,
+ "step": 34935
+ },
+ {
+ "epoch": 95.19346049046321,
+ "grad_norm": 1.1119617223739624,
+ "learning_rate": 1.2092444061145737e-07,
+ "loss": 0.0175,
+ "step": 34936
+ },
+ {
+ "epoch": 95.19618528610354,
+ "grad_norm": 1.0995550155639648,
+ "learning_rate": 1.2078765352827591e-07,
+ "loss": 0.008,
+ "step": 34937
+ },
+ {
+ "epoch": 95.19891008174388,
+ "grad_norm": 0.5077326893806458,
+ "learning_rate": 1.206509433840819e-07,
+ "loss": 0.0048,
+ "step": 34938
+ },
+ {
+ "epoch": 95.2016348773842,
+ "grad_norm": 0.55118328332901,
+ "learning_rate": 1.205143101799411e-07,
+ "loss": 0.0054,
+ "step": 34939
+ },
+ {
+ "epoch": 95.20435967302453,
+ "grad_norm": 0.8015845417976379,
+ "learning_rate": 1.2037775391691708e-07,
+ "loss": 0.0095,
+ "step": 34940
+ },
+ {
+ "epoch": 95.20708446866485,
+ "grad_norm": 0.46727582812309265,
+ "learning_rate": 1.2024127459607237e-07,
+ "loss": 0.0047,
+ "step": 34941
+ },
+ {
+ "epoch": 95.20980926430518,
+ "grad_norm": 0.7487692832946777,
+ "learning_rate": 1.2010487221847056e-07,
+ "loss": 0.0116,
+ "step": 34942
+ },
+ {
+ "epoch": 95.2125340599455,
+ "grad_norm": 1.5140056610107422,
+ "learning_rate": 1.199685467851741e-07,
+ "loss": 0.0498,
+ "step": 34943
+ },
+ {
+ "epoch": 95.21525885558583,
+ "grad_norm": 1.1921229362487793,
+ "learning_rate": 1.1983229829724552e-07,
+ "loss": 0.0318,
+ "step": 34944
+ },
+ {
+ "epoch": 95.21798365122616,
+ "grad_norm": 0.7899995446205139,
+ "learning_rate": 1.1969612675574394e-07,
+ "loss": 0.0072,
+ "step": 34945
+ },
+ {
+ "epoch": 95.22070844686648,
+ "grad_norm": 1.180734634399414,
+ "learning_rate": 1.1956003216173184e-07,
+ "loss": 0.024,
+ "step": 34946
+ },
+ {
+ "epoch": 95.22343324250681,
+ "grad_norm": 0.4678342640399933,
+ "learning_rate": 1.194240145162673e-07,
+ "loss": 0.0037,
+ "step": 34947
+ },
+ {
+ "epoch": 95.22615803814713,
+ "grad_norm": 0.5270805954933167,
+ "learning_rate": 1.1928807382041163e-07,
+ "loss": 0.004,
+ "step": 34948
+ },
+ {
+ "epoch": 95.22888283378747,
+ "grad_norm": 0.6820994019508362,
+ "learning_rate": 1.1915221007522182e-07,
+ "loss": 0.0055,
+ "step": 34949
+ },
+ {
+ "epoch": 95.2316076294278,
+ "grad_norm": 1.458154320716858,
+ "learning_rate": 1.1901642328175589e-07,
+ "loss": 0.1087,
+ "step": 34950
+ },
+ {
+ "epoch": 95.23433242506812,
+ "grad_norm": 0.7114889025688171,
+ "learning_rate": 1.1888071344107299e-07,
+ "loss": 0.0074,
+ "step": 34951
+ },
+ {
+ "epoch": 95.23705722070845,
+ "grad_norm": 0.3171512186527252,
+ "learning_rate": 1.1874508055422895e-07,
+ "loss": 0.0029,
+ "step": 34952
+ },
+ {
+ "epoch": 95.23978201634877,
+ "grad_norm": 1.4815175533294678,
+ "learning_rate": 1.1860952462227959e-07,
+ "loss": 0.1133,
+ "step": 34953
+ },
+ {
+ "epoch": 95.2425068119891,
+ "grad_norm": 1.4018956422805786,
+ "learning_rate": 1.1847404564628185e-07,
+ "loss": 0.0975,
+ "step": 34954
+ },
+ {
+ "epoch": 95.24523160762942,
+ "grad_norm": 1.2722430229187012,
+ "learning_rate": 1.1833864362729042e-07,
+ "loss": 0.0752,
+ "step": 34955
+ },
+ {
+ "epoch": 95.24795640326975,
+ "grad_norm": 1.4619126319885254,
+ "learning_rate": 1.1820331856635891e-07,
+ "loss": 0.0099,
+ "step": 34956
+ },
+ {
+ "epoch": 95.25068119891009,
+ "grad_norm": 0.6927023530006409,
+ "learning_rate": 1.1806807046454205e-07,
+ "loss": 0.0067,
+ "step": 34957
+ },
+ {
+ "epoch": 95.2534059945504,
+ "grad_norm": 0.9558382034301758,
+ "learning_rate": 1.1793289932289343e-07,
+ "loss": 0.0067,
+ "step": 34958
+ },
+ {
+ "epoch": 95.25613079019074,
+ "grad_norm": 1.061163306236267,
+ "learning_rate": 1.1779780514246442e-07,
+ "loss": 0.0095,
+ "step": 34959
+ },
+ {
+ "epoch": 95.25885558583106,
+ "grad_norm": 0.8181156516075134,
+ "learning_rate": 1.1766278792430974e-07,
+ "loss": 0.0059,
+ "step": 34960
+ },
+ {
+ "epoch": 95.26158038147139,
+ "grad_norm": 1.5060408115386963,
+ "learning_rate": 1.1752784766947744e-07,
+ "loss": 0.035,
+ "step": 34961
+ },
+ {
+ "epoch": 95.26430517711172,
+ "grad_norm": 0.7602459192276001,
+ "learning_rate": 1.1739298437902114e-07,
+ "loss": 0.0092,
+ "step": 34962
+ },
+ {
+ "epoch": 95.26702997275204,
+ "grad_norm": 1.799902319908142,
+ "learning_rate": 1.1725819805398997e-07,
+ "loss": 0.0325,
+ "step": 34963
+ },
+ {
+ "epoch": 95.26975476839237,
+ "grad_norm": 0.577941358089447,
+ "learning_rate": 1.1712348869543421e-07,
+ "loss": 0.0059,
+ "step": 34964
+ },
+ {
+ "epoch": 95.2724795640327,
+ "grad_norm": 1.1205368041992188,
+ "learning_rate": 1.1698885630440193e-07,
+ "loss": 0.077,
+ "step": 34965
+ },
+ {
+ "epoch": 95.27520435967303,
+ "grad_norm": 1.3362846374511719,
+ "learning_rate": 1.1685430088194227e-07,
+ "loss": 0.0205,
+ "step": 34966
+ },
+ {
+ "epoch": 95.27792915531334,
+ "grad_norm": 0.6850632429122925,
+ "learning_rate": 1.167198224291044e-07,
+ "loss": 0.0042,
+ "step": 34967
+ },
+ {
+ "epoch": 95.28065395095368,
+ "grad_norm": 1.4848781824111938,
+ "learning_rate": 1.1658542094693415e-07,
+ "loss": 0.0624,
+ "step": 34968
+ },
+ {
+ "epoch": 95.28337874659401,
+ "grad_norm": 1.077387809753418,
+ "learning_rate": 1.1645109643647734e-07,
+ "loss": 0.0093,
+ "step": 34969
+ },
+ {
+ "epoch": 95.28610354223433,
+ "grad_norm": 0.7033839225769043,
+ "learning_rate": 1.1631684889878203e-07,
+ "loss": 0.129,
+ "step": 34970
+ },
+ {
+ "epoch": 95.28882833787466,
+ "grad_norm": 1.062102198600769,
+ "learning_rate": 1.1618267833489294e-07,
+ "loss": 0.0088,
+ "step": 34971
+ },
+ {
+ "epoch": 95.29155313351498,
+ "grad_norm": 0.7232000231742859,
+ "learning_rate": 1.160485847458559e-07,
+ "loss": 0.0111,
+ "step": 34972
+ },
+ {
+ "epoch": 95.29427792915531,
+ "grad_norm": 0.993803083896637,
+ "learning_rate": 1.159145681327134e-07,
+ "loss": 0.0089,
+ "step": 34973
+ },
+ {
+ "epoch": 95.29700272479565,
+ "grad_norm": 1.0047231912612915,
+ "learning_rate": 1.1578062849651017e-07,
+ "loss": 0.0366,
+ "step": 34974
+ },
+ {
+ "epoch": 95.29972752043597,
+ "grad_norm": 1.9305890798568726,
+ "learning_rate": 1.1564676583828982e-07,
+ "loss": 0.0097,
+ "step": 34975
+ },
+ {
+ "epoch": 95.3024523160763,
+ "grad_norm": 1.215661644935608,
+ "learning_rate": 1.1551298015909373e-07,
+ "loss": 0.0162,
+ "step": 34976
+ },
+ {
+ "epoch": 95.30517711171662,
+ "grad_norm": 1.0597659349441528,
+ "learning_rate": 1.1537927145996441e-07,
+ "loss": 0.0073,
+ "step": 34977
+ },
+ {
+ "epoch": 95.30790190735695,
+ "grad_norm": 1.253554344177246,
+ "learning_rate": 1.1524563974194325e-07,
+ "loss": 0.0244,
+ "step": 34978
+ },
+ {
+ "epoch": 95.31062670299727,
+ "grad_norm": 1.70692777633667,
+ "learning_rate": 1.1511208500607163e-07,
+ "loss": 0.01,
+ "step": 34979
+ },
+ {
+ "epoch": 95.3133514986376,
+ "grad_norm": 0.9226534962654114,
+ "learning_rate": 1.1497860725338872e-07,
+ "loss": 0.007,
+ "step": 34980
+ },
+ {
+ "epoch": 95.31607629427793,
+ "grad_norm": 0.9367064237594604,
+ "learning_rate": 1.148452064849337e-07,
+ "loss": 0.0131,
+ "step": 34981
+ },
+ {
+ "epoch": 95.31880108991825,
+ "grad_norm": 0.4153384268283844,
+ "learning_rate": 1.1471188270174571e-07,
+ "loss": 0.0041,
+ "step": 34982
+ },
+ {
+ "epoch": 95.32152588555859,
+ "grad_norm": 1.2190041542053223,
+ "learning_rate": 1.1457863590486507e-07,
+ "loss": 0.015,
+ "step": 34983
+ },
+ {
+ "epoch": 95.3242506811989,
+ "grad_norm": 0.7176473140716553,
+ "learning_rate": 1.1444546609532648e-07,
+ "loss": 0.011,
+ "step": 34984
+ },
+ {
+ "epoch": 95.32697547683924,
+ "grad_norm": 0.8766396641731262,
+ "learning_rate": 1.1431237327416911e-07,
+ "loss": 0.0145,
+ "step": 34985
+ },
+ {
+ "epoch": 95.32970027247957,
+ "grad_norm": 3.3743462562561035,
+ "learning_rate": 1.1417935744242769e-07,
+ "loss": 0.1411,
+ "step": 34986
+ },
+ {
+ "epoch": 95.33242506811989,
+ "grad_norm": 1.3439784049987793,
+ "learning_rate": 1.140464186011403e-07,
+ "loss": 0.0243,
+ "step": 34987
+ },
+ {
+ "epoch": 95.33514986376022,
+ "grad_norm": 1.7881370782852173,
+ "learning_rate": 1.1391355675134164e-07,
+ "loss": 0.0374,
+ "step": 34988
+ },
+ {
+ "epoch": 95.33787465940054,
+ "grad_norm": 0.5216455459594727,
+ "learning_rate": 1.1378077189406422e-07,
+ "loss": 0.005,
+ "step": 34989
+ },
+ {
+ "epoch": 95.34059945504087,
+ "grad_norm": 0.9562675952911377,
+ "learning_rate": 1.13648064030345e-07,
+ "loss": 0.0074,
+ "step": 34990
+ },
+ {
+ "epoch": 95.34332425068119,
+ "grad_norm": 0.9546094536781311,
+ "learning_rate": 1.135154331612165e-07,
+ "loss": 0.0098,
+ "step": 34991
+ },
+ {
+ "epoch": 95.34604904632153,
+ "grad_norm": 2.0339183807373047,
+ "learning_rate": 1.133828792877123e-07,
+ "loss": 0.0748,
+ "step": 34992
+ },
+ {
+ "epoch": 95.34877384196186,
+ "grad_norm": 1.6323388814926147,
+ "learning_rate": 1.1325040241086383e-07,
+ "loss": 0.045,
+ "step": 34993
+ },
+ {
+ "epoch": 95.35149863760218,
+ "grad_norm": 0.6254098415374756,
+ "learning_rate": 1.1311800253170246e-07,
+ "loss": 0.0048,
+ "step": 34994
+ },
+ {
+ "epoch": 95.35422343324251,
+ "grad_norm": 0.5589385032653809,
+ "learning_rate": 1.1298567965125962e-07,
+ "loss": 0.0045,
+ "step": 34995
+ },
+ {
+ "epoch": 95.35694822888283,
+ "grad_norm": 0.9251401424407959,
+ "learning_rate": 1.1285343377056668e-07,
+ "loss": 0.0069,
+ "step": 34996
+ },
+ {
+ "epoch": 95.35967302452316,
+ "grad_norm": 2.2934441566467285,
+ "learning_rate": 1.1272126489065283e-07,
+ "loss": 0.0341,
+ "step": 34997
+ },
+ {
+ "epoch": 95.3623978201635,
+ "grad_norm": 1.2697057723999023,
+ "learning_rate": 1.1258917301254724e-07,
+ "loss": 0.024,
+ "step": 34998
+ },
+ {
+ "epoch": 95.36512261580381,
+ "grad_norm": 1.3993206024169922,
+ "learning_rate": 1.1245715813727908e-07,
+ "loss": 0.0077,
+ "step": 34999
+ },
+ {
+ "epoch": 95.36784741144415,
+ "grad_norm": 1.6604740619659424,
+ "learning_rate": 1.1232522026587755e-07,
+ "loss": 0.0552,
+ "step": 35000
+ },
+ {
+ "epoch": 95.37057220708446,
+ "grad_norm": 1.0495904684066772,
+ "learning_rate": 1.1219335939936738e-07,
+ "loss": 0.1104,
+ "step": 35001
+ },
+ {
+ "epoch": 95.3732970027248,
+ "grad_norm": 1.2672878503799438,
+ "learning_rate": 1.1206157553877772e-07,
+ "loss": 0.0084,
+ "step": 35002
+ },
+ {
+ "epoch": 95.37602179836512,
+ "grad_norm": 0.9481752514839172,
+ "learning_rate": 1.1192986868513445e-07,
+ "loss": 0.0102,
+ "step": 35003
+ },
+ {
+ "epoch": 95.37874659400545,
+ "grad_norm": 0.8525968790054321,
+ "learning_rate": 1.1179823883946228e-07,
+ "loss": 0.0123,
+ "step": 35004
+ },
+ {
+ "epoch": 95.38147138964578,
+ "grad_norm": 1.0388330221176147,
+ "learning_rate": 1.1166668600278818e-07,
+ "loss": 0.0654,
+ "step": 35005
+ },
+ {
+ "epoch": 95.3841961852861,
+ "grad_norm": 1.8215935230255127,
+ "learning_rate": 1.1153521017613467e-07,
+ "loss": 0.0296,
+ "step": 35006
+ },
+ {
+ "epoch": 95.38692098092643,
+ "grad_norm": 1.0371050834655762,
+ "learning_rate": 1.1140381136052758e-07,
+ "loss": 0.0126,
+ "step": 35007
+ },
+ {
+ "epoch": 95.38964577656675,
+ "grad_norm": 0.4566120207309723,
+ "learning_rate": 1.1127248955698833e-07,
+ "loss": 0.0041,
+ "step": 35008
+ },
+ {
+ "epoch": 95.39237057220708,
+ "grad_norm": 0.9659622311592102,
+ "learning_rate": 1.1114124476654164e-07,
+ "loss": 0.008,
+ "step": 35009
+ },
+ {
+ "epoch": 95.39509536784742,
+ "grad_norm": 0.9155789613723755,
+ "learning_rate": 1.1101007699020782e-07,
+ "loss": 0.0076,
+ "step": 35010
+ },
+ {
+ "epoch": 95.39782016348774,
+ "grad_norm": 1.0106359720230103,
+ "learning_rate": 1.108789862290105e-07,
+ "loss": 0.0145,
+ "step": 35011
+ },
+ {
+ "epoch": 95.40054495912807,
+ "grad_norm": 1.2943533658981323,
+ "learning_rate": 1.1074797248396884e-07,
+ "loss": 0.009,
+ "step": 35012
+ },
+ {
+ "epoch": 95.40326975476839,
+ "grad_norm": 1.7146551609039307,
+ "learning_rate": 1.1061703575610316e-07,
+ "loss": 0.0119,
+ "step": 35013
+ },
+ {
+ "epoch": 95.40599455040872,
+ "grad_norm": 1.6536866426467896,
+ "learning_rate": 1.1048617604643374e-07,
+ "loss": 0.024,
+ "step": 35014
+ },
+ {
+ "epoch": 95.40871934604904,
+ "grad_norm": 0.6211981177330017,
+ "learning_rate": 1.1035539335598089e-07,
+ "loss": 0.0057,
+ "step": 35015
+ },
+ {
+ "epoch": 95.41144414168937,
+ "grad_norm": 1.1823080778121948,
+ "learning_rate": 1.1022468768576044e-07,
+ "loss": 0.0093,
+ "step": 35016
+ },
+ {
+ "epoch": 95.4141689373297,
+ "grad_norm": 1.2031886577606201,
+ "learning_rate": 1.1009405903679271e-07,
+ "loss": 0.0138,
+ "step": 35017
+ },
+ {
+ "epoch": 95.41689373297002,
+ "grad_norm": 0.7622225284576416,
+ "learning_rate": 1.0996350741009355e-07,
+ "loss": 0.01,
+ "step": 35018
+ },
+ {
+ "epoch": 95.41961852861036,
+ "grad_norm": 1.3986165523529053,
+ "learning_rate": 1.0983303280668101e-07,
+ "loss": 0.0241,
+ "step": 35019
+ },
+ {
+ "epoch": 95.42234332425068,
+ "grad_norm": 0.8269719481468201,
+ "learning_rate": 1.0970263522756985e-07,
+ "loss": 0.012,
+ "step": 35020
+ },
+ {
+ "epoch": 95.42506811989101,
+ "grad_norm": 1.9675264358520508,
+ "learning_rate": 1.0957231467377704e-07,
+ "loss": 0.0234,
+ "step": 35021
+ },
+ {
+ "epoch": 95.42779291553134,
+ "grad_norm": 1.2975666522979736,
+ "learning_rate": 1.0944207114631622e-07,
+ "loss": 0.0084,
+ "step": 35022
+ },
+ {
+ "epoch": 95.43051771117166,
+ "grad_norm": 1.2558917999267578,
+ "learning_rate": 1.09311904646201e-07,
+ "loss": 0.0104,
+ "step": 35023
+ },
+ {
+ "epoch": 95.433242506812,
+ "grad_norm": 0.6752201318740845,
+ "learning_rate": 1.0918181517444836e-07,
+ "loss": 0.0096,
+ "step": 35024
+ },
+ {
+ "epoch": 95.43596730245231,
+ "grad_norm": 1.4039174318313599,
+ "learning_rate": 1.0905180273206862e-07,
+ "loss": 0.0252,
+ "step": 35025
+ },
+ {
+ "epoch": 95.43869209809264,
+ "grad_norm": 1.2395610809326172,
+ "learning_rate": 1.0892186732007538e-07,
+ "loss": 0.1204,
+ "step": 35026
+ },
+ {
+ "epoch": 95.44141689373296,
+ "grad_norm": 0.6201266050338745,
+ "learning_rate": 1.0879200893947895e-07,
+ "loss": 0.0072,
+ "step": 35027
+ },
+ {
+ "epoch": 95.4441416893733,
+ "grad_norm": 2.093418598175049,
+ "learning_rate": 1.0866222759129296e-07,
+ "loss": 0.0567,
+ "step": 35028
+ },
+ {
+ "epoch": 95.44686648501363,
+ "grad_norm": 1.2504923343658447,
+ "learning_rate": 1.0853252327652775e-07,
+ "loss": 0.0219,
+ "step": 35029
+ },
+ {
+ "epoch": 95.44959128065395,
+ "grad_norm": 0.6469532251358032,
+ "learning_rate": 1.0840289599619247e-07,
+ "loss": 0.0051,
+ "step": 35030
+ },
+ {
+ "epoch": 95.45231607629428,
+ "grad_norm": 1.1269532442092896,
+ "learning_rate": 1.0827334575129633e-07,
+ "loss": 0.0227,
+ "step": 35031
+ },
+ {
+ "epoch": 95.4550408719346,
+ "grad_norm": 1.2411683797836304,
+ "learning_rate": 1.0814387254284964e-07,
+ "loss": 0.0338,
+ "step": 35032
+ },
+ {
+ "epoch": 95.45776566757493,
+ "grad_norm": 0.6215547919273376,
+ "learning_rate": 1.0801447637186047e-07,
+ "loss": 0.0061,
+ "step": 35033
+ },
+ {
+ "epoch": 95.46049046321527,
+ "grad_norm": 1.1590352058410645,
+ "learning_rate": 1.0788515723933579e-07,
+ "loss": 0.0101,
+ "step": 35034
+ },
+ {
+ "epoch": 95.46321525885558,
+ "grad_norm": 1.0848896503448486,
+ "learning_rate": 1.077559151462837e-07,
+ "loss": 0.0659,
+ "step": 35035
+ },
+ {
+ "epoch": 95.46594005449592,
+ "grad_norm": 0.6615561842918396,
+ "learning_rate": 1.0762675009370893e-07,
+ "loss": 0.0051,
+ "step": 35036
+ },
+ {
+ "epoch": 95.46866485013624,
+ "grad_norm": 1.2013782262802124,
+ "learning_rate": 1.0749766208261959e-07,
+ "loss": 0.0151,
+ "step": 35037
+ },
+ {
+ "epoch": 95.47138964577657,
+ "grad_norm": 2.015084743499756,
+ "learning_rate": 1.073686511140204e-07,
+ "loss": 0.0086,
+ "step": 35038
+ },
+ {
+ "epoch": 95.47411444141689,
+ "grad_norm": 0.8192855715751648,
+ "learning_rate": 1.0723971718891502e-07,
+ "loss": 0.0129,
+ "step": 35039
+ },
+ {
+ "epoch": 95.47683923705722,
+ "grad_norm": 1.3002387285232544,
+ "learning_rate": 1.071108603083082e-07,
+ "loss": 0.0076,
+ "step": 35040
+ },
+ {
+ "epoch": 95.47956403269755,
+ "grad_norm": 1.2112407684326172,
+ "learning_rate": 1.0698208047320358e-07,
+ "loss": 0.0108,
+ "step": 35041
+ },
+ {
+ "epoch": 95.48228882833787,
+ "grad_norm": 1.524543046951294,
+ "learning_rate": 1.068533776846048e-07,
+ "loss": 0.1478,
+ "step": 35042
+ },
+ {
+ "epoch": 95.4850136239782,
+ "grad_norm": 1.3990323543548584,
+ "learning_rate": 1.0672475194351328e-07,
+ "loss": 0.0732,
+ "step": 35043
+ },
+ {
+ "epoch": 95.48773841961852,
+ "grad_norm": 2.063148021697998,
+ "learning_rate": 1.0659620325093045e-07,
+ "loss": 0.0701,
+ "step": 35044
+ },
+ {
+ "epoch": 95.49046321525886,
+ "grad_norm": 0.935492753982544,
+ "learning_rate": 1.0646773160785884e-07,
+ "loss": 0.0559,
+ "step": 35045
+ },
+ {
+ "epoch": 95.49318801089919,
+ "grad_norm": 0.40584254264831543,
+ "learning_rate": 1.0633933701529764e-07,
+ "loss": 0.0045,
+ "step": 35046
+ },
+ {
+ "epoch": 95.49591280653951,
+ "grad_norm": 1.4706298112869263,
+ "learning_rate": 1.0621101947424717e-07,
+ "loss": 0.0281,
+ "step": 35047
+ },
+ {
+ "epoch": 95.49863760217984,
+ "grad_norm": 1.5442341566085815,
+ "learning_rate": 1.0608277898570663e-07,
+ "loss": 0.068,
+ "step": 35048
+ },
+ {
+ "epoch": 95.50136239782016,
+ "grad_norm": 1.5092254877090454,
+ "learning_rate": 1.0595461555067521e-07,
+ "loss": 0.0501,
+ "step": 35049
+ },
+ {
+ "epoch": 95.50408719346049,
+ "grad_norm": 0.3586072325706482,
+ "learning_rate": 1.0582652917015101e-07,
+ "loss": 0.0031,
+ "step": 35050
+ },
+ {
+ "epoch": 95.50681198910081,
+ "grad_norm": 0.9421473145484924,
+ "learning_rate": 1.0569851984513102e-07,
+ "loss": 0.0123,
+ "step": 35051
+ },
+ {
+ "epoch": 95.50953678474114,
+ "grad_norm": 1.364957332611084,
+ "learning_rate": 1.0557058757661331e-07,
+ "loss": 0.0385,
+ "step": 35052
+ },
+ {
+ "epoch": 95.51226158038148,
+ "grad_norm": 0.5814760327339172,
+ "learning_rate": 1.0544273236559266e-07,
+ "loss": 0.006,
+ "step": 35053
+ },
+ {
+ "epoch": 95.5149863760218,
+ "grad_norm": 0.7359285950660706,
+ "learning_rate": 1.0531495421306492e-07,
+ "loss": 0.0718,
+ "step": 35054
+ },
+ {
+ "epoch": 95.51771117166213,
+ "grad_norm": 2.1071038246154785,
+ "learning_rate": 1.0518725312002709e-07,
+ "loss": 0.028,
+ "step": 35055
+ },
+ {
+ "epoch": 95.52043596730245,
+ "grad_norm": 1.2403911352157593,
+ "learning_rate": 1.0505962908747281e-07,
+ "loss": 0.0122,
+ "step": 35056
+ },
+ {
+ "epoch": 95.52316076294278,
+ "grad_norm": 0.9468063116073608,
+ "learning_rate": 1.0493208211639349e-07,
+ "loss": 0.0118,
+ "step": 35057
+ },
+ {
+ "epoch": 95.52588555858311,
+ "grad_norm": 0.6799874305725098,
+ "learning_rate": 1.0480461220778725e-07,
+ "loss": 0.0154,
+ "step": 35058
+ },
+ {
+ "epoch": 95.52861035422343,
+ "grad_norm": 0.6844435334205627,
+ "learning_rate": 1.0467721936264219e-07,
+ "loss": 0.006,
+ "step": 35059
+ },
+ {
+ "epoch": 95.53133514986376,
+ "grad_norm": 1.0528020858764648,
+ "learning_rate": 1.0454990358195305e-07,
+ "loss": 0.0077,
+ "step": 35060
+ },
+ {
+ "epoch": 95.53405994550408,
+ "grad_norm": 0.8855552077293396,
+ "learning_rate": 1.0442266486671126e-07,
+ "loss": 0.0142,
+ "step": 35061
+ },
+ {
+ "epoch": 95.53678474114442,
+ "grad_norm": 0.75606769323349,
+ "learning_rate": 1.0429550321790716e-07,
+ "loss": 0.0141,
+ "step": 35062
+ },
+ {
+ "epoch": 95.53950953678473,
+ "grad_norm": 1.2715234756469727,
+ "learning_rate": 1.0416841863653105e-07,
+ "loss": 0.0153,
+ "step": 35063
+ },
+ {
+ "epoch": 95.54223433242507,
+ "grad_norm": 0.5106908679008484,
+ "learning_rate": 1.0404141112357325e-07,
+ "loss": 0.0042,
+ "step": 35064
+ },
+ {
+ "epoch": 95.5449591280654,
+ "grad_norm": 0.8285999298095703,
+ "learning_rate": 1.0391448068002185e-07,
+ "loss": 0.0059,
+ "step": 35065
+ },
+ {
+ "epoch": 95.54768392370572,
+ "grad_norm": 1.5004093647003174,
+ "learning_rate": 1.0378762730686609e-07,
+ "loss": 0.0079,
+ "step": 35066
+ },
+ {
+ "epoch": 95.55040871934605,
+ "grad_norm": 1.2233668565750122,
+ "learning_rate": 1.0366085100509405e-07,
+ "loss": 0.0233,
+ "step": 35067
+ },
+ {
+ "epoch": 95.55313351498637,
+ "grad_norm": 1.127862572669983,
+ "learning_rate": 1.0353415177569271e-07,
+ "loss": 0.0074,
+ "step": 35068
+ },
+ {
+ "epoch": 95.5558583106267,
+ "grad_norm": 1.0615273714065552,
+ "learning_rate": 1.0340752961964795e-07,
+ "loss": 0.073,
+ "step": 35069
+ },
+ {
+ "epoch": 95.55858310626704,
+ "grad_norm": 3.2094717025756836,
+ "learning_rate": 1.0328098453794788e-07,
+ "loss": 0.0461,
+ "step": 35070
+ },
+ {
+ "epoch": 95.56130790190736,
+ "grad_norm": 1.580406904220581,
+ "learning_rate": 1.0315451653157615e-07,
+ "loss": 0.0204,
+ "step": 35071
+ },
+ {
+ "epoch": 95.56403269754769,
+ "grad_norm": 1.0764014720916748,
+ "learning_rate": 1.0302812560151976e-07,
+ "loss": 0.0246,
+ "step": 35072
+ },
+ {
+ "epoch": 95.566757493188,
+ "grad_norm": 1.0062681436538696,
+ "learning_rate": 1.0290181174876124e-07,
+ "loss": 0.1298,
+ "step": 35073
+ },
+ {
+ "epoch": 95.56948228882834,
+ "grad_norm": 0.8724644780158997,
+ "learning_rate": 1.0277557497428536e-07,
+ "loss": 0.0081,
+ "step": 35074
+ },
+ {
+ "epoch": 95.57220708446866,
+ "grad_norm": 0.753361701965332,
+ "learning_rate": 1.0264941527907247e-07,
+ "loss": 0.0822,
+ "step": 35075
+ },
+ {
+ "epoch": 95.57493188010899,
+ "grad_norm": 1.3160866498947144,
+ "learning_rate": 1.0252333266410952e-07,
+ "loss": 0.0164,
+ "step": 35076
+ },
+ {
+ "epoch": 95.57765667574932,
+ "grad_norm": 0.8282985687255859,
+ "learning_rate": 1.0239732713037576e-07,
+ "loss": 0.0083,
+ "step": 35077
+ },
+ {
+ "epoch": 95.58038147138964,
+ "grad_norm": 0.8805135488510132,
+ "learning_rate": 1.0227139867885371e-07,
+ "loss": 0.0189,
+ "step": 35078
+ },
+ {
+ "epoch": 95.58310626702998,
+ "grad_norm": 0.6166729927062988,
+ "learning_rate": 1.0214554731052261e-07,
+ "loss": 0.0073,
+ "step": 35079
+ },
+ {
+ "epoch": 95.5858310626703,
+ "grad_norm": 0.397935152053833,
+ "learning_rate": 1.020197730263639e-07,
+ "loss": 0.0036,
+ "step": 35080
+ },
+ {
+ "epoch": 95.58855585831063,
+ "grad_norm": 1.0118930339813232,
+ "learning_rate": 1.0189407582735566e-07,
+ "loss": 0.0156,
+ "step": 35081
+ },
+ {
+ "epoch": 95.59128065395096,
+ "grad_norm": 1.0041221380233765,
+ "learning_rate": 1.0176845571447936e-07,
+ "loss": 0.0397,
+ "step": 35082
+ },
+ {
+ "epoch": 95.59400544959128,
+ "grad_norm": 1.2550594806671143,
+ "learning_rate": 1.0164291268870974e-07,
+ "loss": 0.0314,
+ "step": 35083
+ },
+ {
+ "epoch": 95.59673024523161,
+ "grad_norm": 0.4702872633934021,
+ "learning_rate": 1.0151744675102826e-07,
+ "loss": 0.0048,
+ "step": 35084
+ },
+ {
+ "epoch": 95.59945504087193,
+ "grad_norm": 0.6220273375511169,
+ "learning_rate": 1.0139205790240969e-07,
+ "loss": 0.008,
+ "step": 35085
+ },
+ {
+ "epoch": 95.60217983651226,
+ "grad_norm": 0.9133254289627075,
+ "learning_rate": 1.0126674614383214e-07,
+ "loss": 0.0451,
+ "step": 35086
+ },
+ {
+ "epoch": 95.60490463215258,
+ "grad_norm": 1.3303638696670532,
+ "learning_rate": 1.0114151147626927e-07,
+ "loss": 0.0507,
+ "step": 35087
+ },
+ {
+ "epoch": 95.60762942779292,
+ "grad_norm": 1.0126844644546509,
+ "learning_rate": 1.0101635390069809e-07,
+ "loss": 0.008,
+ "step": 35088
+ },
+ {
+ "epoch": 95.61035422343325,
+ "grad_norm": 0.835628867149353,
+ "learning_rate": 1.0089127341809336e-07,
+ "loss": 0.0066,
+ "step": 35089
+ },
+ {
+ "epoch": 95.61307901907357,
+ "grad_norm": 0.6456746459007263,
+ "learning_rate": 1.0076627002942873e-07,
+ "loss": 0.0091,
+ "step": 35090
+ },
+ {
+ "epoch": 95.6158038147139,
+ "grad_norm": 2.196284055709839,
+ "learning_rate": 1.006413437356768e-07,
+ "loss": 0.0219,
+ "step": 35091
+ },
+ {
+ "epoch": 95.61852861035422,
+ "grad_norm": 1.03243887424469,
+ "learning_rate": 1.0051649453781231e-07,
+ "loss": 0.0104,
+ "step": 35092
+ },
+ {
+ "epoch": 95.62125340599455,
+ "grad_norm": 1.3203699588775635,
+ "learning_rate": 1.0039172243680673e-07,
+ "loss": 0.0114,
+ "step": 35093
+ },
+ {
+ "epoch": 95.62397820163488,
+ "grad_norm": 0.7029433250427246,
+ "learning_rate": 1.0026702743363148e-07,
+ "loss": 0.0096,
+ "step": 35094
+ },
+ {
+ "epoch": 95.6267029972752,
+ "grad_norm": 1.2847658395767212,
+ "learning_rate": 1.0014240952925803e-07,
+ "loss": 0.016,
+ "step": 35095
+ },
+ {
+ "epoch": 95.62942779291554,
+ "grad_norm": 16.1458683013916,
+ "learning_rate": 1.000178687246578e-07,
+ "loss": 0.0465,
+ "step": 35096
+ },
+ {
+ "epoch": 95.63215258855585,
+ "grad_norm": 1.1985818147659302,
+ "learning_rate": 9.989340502079891e-08,
+ "loss": 0.0289,
+ "step": 35097
+ },
+ {
+ "epoch": 95.63487738419619,
+ "grad_norm": 1.5475257635116577,
+ "learning_rate": 9.976901841865172e-08,
+ "loss": 0.0285,
+ "step": 35098
+ },
+ {
+ "epoch": 95.6376021798365,
+ "grad_norm": 1.5097841024398804,
+ "learning_rate": 9.964470891918431e-08,
+ "loss": 0.0376,
+ "step": 35099
+ },
+ {
+ "epoch": 95.64032697547684,
+ "grad_norm": 1.4729397296905518,
+ "learning_rate": 9.952047652336594e-08,
+ "loss": 0.0588,
+ "step": 35100
+ },
+ {
+ "epoch": 95.64305177111717,
+ "grad_norm": 2.7066256999969482,
+ "learning_rate": 9.939632123216359e-08,
+ "loss": 0.0644,
+ "step": 35101
+ },
+ {
+ "epoch": 95.64577656675749,
+ "grad_norm": 1.2870478630065918,
+ "learning_rate": 9.927224304654315e-08,
+ "loss": 0.1026,
+ "step": 35102
+ },
+ {
+ "epoch": 95.64850136239782,
+ "grad_norm": 0.5905954241752625,
+ "learning_rate": 9.914824196747274e-08,
+ "loss": 0.0058,
+ "step": 35103
+ },
+ {
+ "epoch": 95.65122615803814,
+ "grad_norm": 0.848738431930542,
+ "learning_rate": 9.902431799591605e-08,
+ "loss": 0.0097,
+ "step": 35104
+ },
+ {
+ "epoch": 95.65395095367847,
+ "grad_norm": 0.6020531058311462,
+ "learning_rate": 9.890047113284118e-08,
+ "loss": 0.0067,
+ "step": 35105
+ },
+ {
+ "epoch": 95.65667574931881,
+ "grad_norm": 1.507232666015625,
+ "learning_rate": 9.877670137920958e-08,
+ "loss": 0.069,
+ "step": 35106
+ },
+ {
+ "epoch": 95.65940054495913,
+ "grad_norm": 0.5216037631034851,
+ "learning_rate": 9.865300873598605e-08,
+ "loss": 0.0037,
+ "step": 35107
+ },
+ {
+ "epoch": 95.66212534059946,
+ "grad_norm": 0.6456120610237122,
+ "learning_rate": 9.852939320413424e-08,
+ "loss": 0.006,
+ "step": 35108
+ },
+ {
+ "epoch": 95.66485013623978,
+ "grad_norm": 1.2622965574264526,
+ "learning_rate": 9.840585478461673e-08,
+ "loss": 0.0089,
+ "step": 35109
+ },
+ {
+ "epoch": 95.66757493188011,
+ "grad_norm": 0.8570942282676697,
+ "learning_rate": 9.828239347839607e-08,
+ "loss": 0.0067,
+ "step": 35110
+ },
+ {
+ "epoch": 95.67029972752043,
+ "grad_norm": 0.7509015798568726,
+ "learning_rate": 9.815900928643374e-08,
+ "loss": 0.0117,
+ "step": 35111
+ },
+ {
+ "epoch": 95.67302452316076,
+ "grad_norm": 1.0251152515411377,
+ "learning_rate": 9.803570220969006e-08,
+ "loss": 0.0084,
+ "step": 35112
+ },
+ {
+ "epoch": 95.6757493188011,
+ "grad_norm": 0.7553905844688416,
+ "learning_rate": 9.791247224912647e-08,
+ "loss": 0.0129,
+ "step": 35113
+ },
+ {
+ "epoch": 95.67847411444141,
+ "grad_norm": 0.6505278944969177,
+ "learning_rate": 9.778931940570113e-08,
+ "loss": 0.0061,
+ "step": 35114
+ },
+ {
+ "epoch": 95.68119891008175,
+ "grad_norm": 0.5725364685058594,
+ "learning_rate": 9.766624368037436e-08,
+ "loss": 0.0061,
+ "step": 35115
+ },
+ {
+ "epoch": 95.68392370572207,
+ "grad_norm": 1.037086009979248,
+ "learning_rate": 9.754324507410317e-08,
+ "loss": 0.0161,
+ "step": 35116
+ },
+ {
+ "epoch": 95.6866485013624,
+ "grad_norm": 1.643896460533142,
+ "learning_rate": 9.742032358784681e-08,
+ "loss": 0.0423,
+ "step": 35117
+ },
+ {
+ "epoch": 95.68937329700273,
+ "grad_norm": 0.7849664092063904,
+ "learning_rate": 9.729747922256228e-08,
+ "loss": 0.1313,
+ "step": 35118
+ },
+ {
+ "epoch": 95.69209809264305,
+ "grad_norm": 1.206495761871338,
+ "learning_rate": 9.717471197920658e-08,
+ "loss": 0.0137,
+ "step": 35119
+ },
+ {
+ "epoch": 95.69482288828338,
+ "grad_norm": 0.7267259359359741,
+ "learning_rate": 9.705202185873563e-08,
+ "loss": 0.005,
+ "step": 35120
+ },
+ {
+ "epoch": 95.6975476839237,
+ "grad_norm": 0.8728659749031067,
+ "learning_rate": 9.692940886210422e-08,
+ "loss": 0.0146,
+ "step": 35121
+ },
+ {
+ "epoch": 95.70027247956403,
+ "grad_norm": 0.918586015701294,
+ "learning_rate": 9.680687299026826e-08,
+ "loss": 0.0105,
+ "step": 35122
+ },
+ {
+ "epoch": 95.70299727520435,
+ "grad_norm": 0.9888033866882324,
+ "learning_rate": 9.66844142441814e-08,
+ "loss": 0.0114,
+ "step": 35123
+ },
+ {
+ "epoch": 95.70572207084469,
+ "grad_norm": 0.963701605796814,
+ "learning_rate": 9.656203262479735e-08,
+ "loss": 0.0102,
+ "step": 35124
+ },
+ {
+ "epoch": 95.70844686648502,
+ "grad_norm": 0.8138212561607361,
+ "learning_rate": 9.64397281330709e-08,
+ "loss": 0.0067,
+ "step": 35125
+ },
+ {
+ "epoch": 95.71117166212534,
+ "grad_norm": 1.756137490272522,
+ "learning_rate": 9.631750076995128e-08,
+ "loss": 0.0156,
+ "step": 35126
+ },
+ {
+ "epoch": 95.71389645776567,
+ "grad_norm": 0.6387712359428406,
+ "learning_rate": 9.619535053639328e-08,
+ "loss": 0.0065,
+ "step": 35127
+ },
+ {
+ "epoch": 95.71662125340599,
+ "grad_norm": 0.6469250917434692,
+ "learning_rate": 9.607327743334616e-08,
+ "loss": 0.0063,
+ "step": 35128
+ },
+ {
+ "epoch": 95.71934604904632,
+ "grad_norm": 1.4857604503631592,
+ "learning_rate": 9.595128146176247e-08,
+ "loss": 0.0767,
+ "step": 35129
+ },
+ {
+ "epoch": 95.72207084468666,
+ "grad_norm": 0.8822433352470398,
+ "learning_rate": 9.582936262259146e-08,
+ "loss": 0.0304,
+ "step": 35130
+ },
+ {
+ "epoch": 95.72479564032697,
+ "grad_norm": 1.220104694366455,
+ "learning_rate": 9.570752091678237e-08,
+ "loss": 0.1061,
+ "step": 35131
+ },
+ {
+ "epoch": 95.7275204359673,
+ "grad_norm": 0.8622046113014221,
+ "learning_rate": 9.558575634528444e-08,
+ "loss": 0.0278,
+ "step": 35132
+ },
+ {
+ "epoch": 95.73024523160763,
+ "grad_norm": 0.8049647212028503,
+ "learning_rate": 9.54640689090458e-08,
+ "loss": 0.0136,
+ "step": 35133
+ },
+ {
+ "epoch": 95.73297002724796,
+ "grad_norm": 1.1933703422546387,
+ "learning_rate": 9.534245860901347e-08,
+ "loss": 0.0128,
+ "step": 35134
+ },
+ {
+ "epoch": 95.73569482288828,
+ "grad_norm": 1.1257661581039429,
+ "learning_rate": 9.522092544613671e-08,
+ "loss": 0.0315,
+ "step": 35135
+ },
+ {
+ "epoch": 95.73841961852861,
+ "grad_norm": 1.1646466255187988,
+ "learning_rate": 9.509946942135916e-08,
+ "loss": 0.0185,
+ "step": 35136
+ },
+ {
+ "epoch": 95.74114441416894,
+ "grad_norm": 1.445011854171753,
+ "learning_rate": 9.497809053562901e-08,
+ "loss": 0.1664,
+ "step": 35137
+ },
+ {
+ "epoch": 95.74386920980926,
+ "grad_norm": 0.9146530628204346,
+ "learning_rate": 9.485678878989102e-08,
+ "loss": 0.0239,
+ "step": 35138
+ },
+ {
+ "epoch": 95.7465940054496,
+ "grad_norm": 1.3840464353561401,
+ "learning_rate": 9.473556418508889e-08,
+ "loss": 0.0404,
+ "step": 35139
+ },
+ {
+ "epoch": 95.74931880108991,
+ "grad_norm": 1.8611255884170532,
+ "learning_rate": 9.461441672216743e-08,
+ "loss": 0.0362,
+ "step": 35140
+ },
+ {
+ "epoch": 95.75204359673025,
+ "grad_norm": 0.4933811128139496,
+ "learning_rate": 9.449334640207031e-08,
+ "loss": 0.0041,
+ "step": 35141
+ },
+ {
+ "epoch": 95.75476839237058,
+ "grad_norm": 0.9907389879226685,
+ "learning_rate": 9.437235322574123e-08,
+ "loss": 0.0134,
+ "step": 35142
+ },
+ {
+ "epoch": 95.7574931880109,
+ "grad_norm": 0.6383262872695923,
+ "learning_rate": 9.425143719411945e-08,
+ "loss": 0.0072,
+ "step": 35143
+ },
+ {
+ "epoch": 95.76021798365123,
+ "grad_norm": 1.2686415910720825,
+ "learning_rate": 9.413059830814975e-08,
+ "loss": 0.0125,
+ "step": 35144
+ },
+ {
+ "epoch": 95.76294277929155,
+ "grad_norm": 0.9071741104125977,
+ "learning_rate": 9.400983656877138e-08,
+ "loss": 0.0156,
+ "step": 35145
+ },
+ {
+ "epoch": 95.76566757493188,
+ "grad_norm": 1.1017247438430786,
+ "learning_rate": 9.388915197692694e-08,
+ "loss": 0.01,
+ "step": 35146
+ },
+ {
+ "epoch": 95.7683923705722,
+ "grad_norm": 1.3484774827957153,
+ "learning_rate": 9.376854453355455e-08,
+ "loss": 0.0164,
+ "step": 35147
+ },
+ {
+ "epoch": 95.77111716621253,
+ "grad_norm": 0.9241007566452026,
+ "learning_rate": 9.364801423959235e-08,
+ "loss": 0.0086,
+ "step": 35148
+ },
+ {
+ "epoch": 95.77384196185287,
+ "grad_norm": 1.2938823699951172,
+ "learning_rate": 9.352756109598183e-08,
+ "loss": 0.0147,
+ "step": 35149
+ },
+ {
+ "epoch": 95.77656675749319,
+ "grad_norm": 1.1741771697998047,
+ "learning_rate": 9.340718510365998e-08,
+ "loss": 0.0084,
+ "step": 35150
+ },
+ {
+ "epoch": 95.77929155313352,
+ "grad_norm": 1.2122564315795898,
+ "learning_rate": 9.328688626356497e-08,
+ "loss": 0.008,
+ "step": 35151
+ },
+ {
+ "epoch": 95.78201634877384,
+ "grad_norm": 1.2467875480651855,
+ "learning_rate": 9.316666457663048e-08,
+ "loss": 0.0826,
+ "step": 35152
+ },
+ {
+ "epoch": 95.78474114441417,
+ "grad_norm": 1.521092414855957,
+ "learning_rate": 9.304652004379689e-08,
+ "loss": 0.1045,
+ "step": 35153
+ },
+ {
+ "epoch": 95.7874659400545,
+ "grad_norm": 0.3217719793319702,
+ "learning_rate": 9.292645266599787e-08,
+ "loss": 0.0025,
+ "step": 35154
+ },
+ {
+ "epoch": 95.79019073569482,
+ "grad_norm": 1.0359758138656616,
+ "learning_rate": 9.280646244416936e-08,
+ "loss": 0.0071,
+ "step": 35155
+ },
+ {
+ "epoch": 95.79291553133515,
+ "grad_norm": 0.4713518023490906,
+ "learning_rate": 9.268654937924393e-08,
+ "loss": 0.0043,
+ "step": 35156
+ },
+ {
+ "epoch": 95.79564032697547,
+ "grad_norm": 0.5315970182418823,
+ "learning_rate": 9.256671347215751e-08,
+ "loss": 0.0055,
+ "step": 35157
+ },
+ {
+ "epoch": 95.7983651226158,
+ "grad_norm": 1.467827558517456,
+ "learning_rate": 9.244695472384268e-08,
+ "loss": 0.1093,
+ "step": 35158
+ },
+ {
+ "epoch": 95.80108991825612,
+ "grad_norm": 1.1100431680679321,
+ "learning_rate": 9.232727313523204e-08,
+ "loss": 0.016,
+ "step": 35159
+ },
+ {
+ "epoch": 95.80381471389646,
+ "grad_norm": 1.0142109394073486,
+ "learning_rate": 9.220766870725705e-08,
+ "loss": 0.0165,
+ "step": 35160
+ },
+ {
+ "epoch": 95.80653950953679,
+ "grad_norm": 0.64913409948349,
+ "learning_rate": 9.208814144085032e-08,
+ "loss": 0.0058,
+ "step": 35161
+ },
+ {
+ "epoch": 95.80926430517711,
+ "grad_norm": 0.7215105891227722,
+ "learning_rate": 9.19686913369422e-08,
+ "loss": 0.0114,
+ "step": 35162
+ },
+ {
+ "epoch": 95.81198910081744,
+ "grad_norm": 1.1708970069885254,
+ "learning_rate": 9.184931839646417e-08,
+ "loss": 0.0559,
+ "step": 35163
+ },
+ {
+ "epoch": 95.81471389645776,
+ "grad_norm": 0.5743633508682251,
+ "learning_rate": 9.173002262034326e-08,
+ "loss": 0.0051,
+ "step": 35164
+ },
+ {
+ "epoch": 95.8174386920981,
+ "grad_norm": 1.61489999294281,
+ "learning_rate": 9.161080400951095e-08,
+ "loss": 0.0079,
+ "step": 35165
+ },
+ {
+ "epoch": 95.82016348773843,
+ "grad_norm": 0.8482322692871094,
+ "learning_rate": 9.149166256489428e-08,
+ "loss": 0.0071,
+ "step": 35166
+ },
+ {
+ "epoch": 95.82288828337875,
+ "grad_norm": 1.2507797479629517,
+ "learning_rate": 9.13725982874214e-08,
+ "loss": 0.0085,
+ "step": 35167
+ },
+ {
+ "epoch": 95.82561307901908,
+ "grad_norm": 0.9326859712600708,
+ "learning_rate": 9.125361117802045e-08,
+ "loss": 0.0075,
+ "step": 35168
+ },
+ {
+ "epoch": 95.8283378746594,
+ "grad_norm": 1.5856267213821411,
+ "learning_rate": 9.113470123761736e-08,
+ "loss": 0.0531,
+ "step": 35169
+ },
+ {
+ "epoch": 95.83106267029973,
+ "grad_norm": 0.6874185800552368,
+ "learning_rate": 9.101586846713806e-08,
+ "loss": 0.0091,
+ "step": 35170
+ },
+ {
+ "epoch": 95.83378746594005,
+ "grad_norm": 1.2491592168807983,
+ "learning_rate": 9.089711286750847e-08,
+ "loss": 0.0211,
+ "step": 35171
+ },
+ {
+ "epoch": 95.83651226158038,
+ "grad_norm": 1.2672135829925537,
+ "learning_rate": 9.077843443965229e-08,
+ "loss": 0.0119,
+ "step": 35172
+ },
+ {
+ "epoch": 95.83923705722071,
+ "grad_norm": 1.3756386041641235,
+ "learning_rate": 9.065983318449656e-08,
+ "loss": 0.0234,
+ "step": 35173
+ },
+ {
+ "epoch": 95.84196185286103,
+ "grad_norm": 0.8064942955970764,
+ "learning_rate": 9.054130910296166e-08,
+ "loss": 0.0138,
+ "step": 35174
+ },
+ {
+ "epoch": 95.84468664850137,
+ "grad_norm": 1.2887816429138184,
+ "learning_rate": 9.042286219597352e-08,
+ "loss": 0.0185,
+ "step": 35175
+ },
+ {
+ "epoch": 95.84741144414168,
+ "grad_norm": 0.791183590888977,
+ "learning_rate": 9.030449246445139e-08,
+ "loss": 0.0125,
+ "step": 35176
+ },
+ {
+ "epoch": 95.85013623978202,
+ "grad_norm": 0.7754648923873901,
+ "learning_rate": 9.018619990932009e-08,
+ "loss": 0.0269,
+ "step": 35177
+ },
+ {
+ "epoch": 95.85286103542235,
+ "grad_norm": 0.8426916003227234,
+ "learning_rate": 9.00679845314989e-08,
+ "loss": 0.0122,
+ "step": 35178
+ },
+ {
+ "epoch": 95.85558583106267,
+ "grad_norm": 0.9070789217948914,
+ "learning_rate": 8.99498463319104e-08,
+ "loss": 0.0079,
+ "step": 35179
+ },
+ {
+ "epoch": 95.858310626703,
+ "grad_norm": 1.6002840995788574,
+ "learning_rate": 8.983178531147274e-08,
+ "loss": 0.0117,
+ "step": 35180
+ },
+ {
+ "epoch": 95.86103542234332,
+ "grad_norm": 0.6175892949104309,
+ "learning_rate": 8.971380147110632e-08,
+ "loss": 0.0075,
+ "step": 35181
+ },
+ {
+ "epoch": 95.86376021798365,
+ "grad_norm": 0.7879091501235962,
+ "learning_rate": 8.959589481172925e-08,
+ "loss": 0.0075,
+ "step": 35182
+ },
+ {
+ "epoch": 95.86648501362397,
+ "grad_norm": 0.9778711199760437,
+ "learning_rate": 8.947806533426195e-08,
+ "loss": 0.0077,
+ "step": 35183
+ },
+ {
+ "epoch": 95.8692098092643,
+ "grad_norm": 1.0600863695144653,
+ "learning_rate": 8.936031303961922e-08,
+ "loss": 0.0094,
+ "step": 35184
+ },
+ {
+ "epoch": 95.87193460490464,
+ "grad_norm": 2.5290064811706543,
+ "learning_rate": 8.924263792872035e-08,
+ "loss": 0.0214,
+ "step": 35185
+ },
+ {
+ "epoch": 95.87465940054496,
+ "grad_norm": 1.0881497859954834,
+ "learning_rate": 8.912504000248012e-08,
+ "loss": 0.0104,
+ "step": 35186
+ },
+ {
+ "epoch": 95.87738419618529,
+ "grad_norm": 0.9492342472076416,
+ "learning_rate": 8.900751926181561e-08,
+ "loss": 0.0158,
+ "step": 35187
+ },
+ {
+ "epoch": 95.88010899182561,
+ "grad_norm": 1.266540288925171,
+ "learning_rate": 8.889007570764052e-08,
+ "loss": 0.0177,
+ "step": 35188
+ },
+ {
+ "epoch": 95.88283378746594,
+ "grad_norm": 1.3277277946472168,
+ "learning_rate": 8.87727093408708e-08,
+ "loss": 0.0154,
+ "step": 35189
+ },
+ {
+ "epoch": 95.88555858310627,
+ "grad_norm": 0.8719479441642761,
+ "learning_rate": 8.865542016242124e-08,
+ "loss": 0.0088,
+ "step": 35190
+ },
+ {
+ "epoch": 95.88828337874659,
+ "grad_norm": 1.217974066734314,
+ "learning_rate": 8.853820817320447e-08,
+ "loss": 0.1015,
+ "step": 35191
+ },
+ {
+ "epoch": 95.89100817438693,
+ "grad_norm": 0.6730855703353882,
+ "learning_rate": 8.842107337413197e-08,
+ "loss": 0.0061,
+ "step": 35192
+ },
+ {
+ "epoch": 95.89373297002724,
+ "grad_norm": 1.1727237701416016,
+ "learning_rate": 8.830401576611747e-08,
+ "loss": 0.0172,
+ "step": 35193
+ },
+ {
+ "epoch": 95.89645776566758,
+ "grad_norm": 0.3563900887966156,
+ "learning_rate": 8.818703535007245e-08,
+ "loss": 0.0036,
+ "step": 35194
+ },
+ {
+ "epoch": 95.8991825613079,
+ "grad_norm": 0.7911906242370605,
+ "learning_rate": 8.807013212690729e-08,
+ "loss": 0.012,
+ "step": 35195
+ },
+ {
+ "epoch": 95.90190735694823,
+ "grad_norm": 1.2310336828231812,
+ "learning_rate": 8.795330609753349e-08,
+ "loss": 0.0373,
+ "step": 35196
+ },
+ {
+ "epoch": 95.90463215258856,
+ "grad_norm": 0.5083425641059875,
+ "learning_rate": 8.783655726285922e-08,
+ "loss": 0.0068,
+ "step": 35197
+ },
+ {
+ "epoch": 95.90735694822888,
+ "grad_norm": 0.766724705696106,
+ "learning_rate": 8.771988562379596e-08,
+ "loss": 0.0039,
+ "step": 35198
+ },
+ {
+ "epoch": 95.91008174386921,
+ "grad_norm": 1.2027606964111328,
+ "learning_rate": 8.760329118125189e-08,
+ "loss": 0.0194,
+ "step": 35199
+ },
+ {
+ "epoch": 95.91280653950953,
+ "grad_norm": 10.317850112915039,
+ "learning_rate": 8.748677393613292e-08,
+ "loss": 0.0253,
+ "step": 35200
+ },
+ {
+ "epoch": 95.91553133514986,
+ "grad_norm": 1.0208263397216797,
+ "learning_rate": 8.737033388934834e-08,
+ "loss": 0.0234,
+ "step": 35201
+ },
+ {
+ "epoch": 95.9182561307902,
+ "grad_norm": 1.097761631011963,
+ "learning_rate": 8.725397104180521e-08,
+ "loss": 0.0574,
+ "step": 35202
+ },
+ {
+ "epoch": 95.92098092643052,
+ "grad_norm": 0.7590011358261108,
+ "learning_rate": 8.713768539440836e-08,
+ "loss": 0.0055,
+ "step": 35203
+ },
+ {
+ "epoch": 95.92370572207085,
+ "grad_norm": 0.8184288740158081,
+ "learning_rate": 8.702147694806484e-08,
+ "loss": 0.0111,
+ "step": 35204
+ },
+ {
+ "epoch": 95.92643051771117,
+ "grad_norm": 0.8684128522872925,
+ "learning_rate": 8.690534570367837e-08,
+ "loss": 0.0083,
+ "step": 35205
+ },
+ {
+ "epoch": 95.9291553133515,
+ "grad_norm": 0.9797582030296326,
+ "learning_rate": 8.678929166215377e-08,
+ "loss": 0.0072,
+ "step": 35206
+ },
+ {
+ "epoch": 95.93188010899182,
+ "grad_norm": 0.75872403383255,
+ "learning_rate": 8.66733148243959e-08,
+ "loss": 0.0151,
+ "step": 35207
+ },
+ {
+ "epoch": 95.93460490463215,
+ "grad_norm": 0.3262571394443512,
+ "learning_rate": 8.655741519130622e-08,
+ "loss": 0.0035,
+ "step": 35208
+ },
+ {
+ "epoch": 95.93732970027249,
+ "grad_norm": 0.408328115940094,
+ "learning_rate": 8.644159276378961e-08,
+ "loss": 0.0045,
+ "step": 35209
+ },
+ {
+ "epoch": 95.9400544959128,
+ "grad_norm": 1.807011604309082,
+ "learning_rate": 8.632584754274642e-08,
+ "loss": 0.0217,
+ "step": 35210
+ },
+ {
+ "epoch": 95.94277929155314,
+ "grad_norm": 1.3421812057495117,
+ "learning_rate": 8.621017952907817e-08,
+ "loss": 0.0266,
+ "step": 35211
+ },
+ {
+ "epoch": 95.94550408719346,
+ "grad_norm": 0.9516888856887817,
+ "learning_rate": 8.609458872368526e-08,
+ "loss": 0.0271,
+ "step": 35212
+ },
+ {
+ "epoch": 95.94822888283379,
+ "grad_norm": 0.7672529816627502,
+ "learning_rate": 8.597907512746917e-08,
+ "loss": 0.0082,
+ "step": 35213
+ },
+ {
+ "epoch": 95.95095367847412,
+ "grad_norm": 0.5311959385871887,
+ "learning_rate": 8.58636387413292e-08,
+ "loss": 0.0058,
+ "step": 35214
+ },
+ {
+ "epoch": 95.95367847411444,
+ "grad_norm": 1.0211808681488037,
+ "learning_rate": 8.57482795661646e-08,
+ "loss": 0.0072,
+ "step": 35215
+ },
+ {
+ "epoch": 95.95640326975477,
+ "grad_norm": 1.5541502237319946,
+ "learning_rate": 8.563299760287358e-08,
+ "loss": 0.1707,
+ "step": 35216
+ },
+ {
+ "epoch": 95.95912806539509,
+ "grad_norm": 0.9182538986206055,
+ "learning_rate": 8.551779285235206e-08,
+ "loss": 0.0279,
+ "step": 35217
+ },
+ {
+ "epoch": 95.96185286103542,
+ "grad_norm": 0.8082046508789062,
+ "learning_rate": 8.540266531549934e-08,
+ "loss": 0.0081,
+ "step": 35218
+ },
+ {
+ "epoch": 95.96457765667574,
+ "grad_norm": 1.5798070430755615,
+ "learning_rate": 8.528761499321247e-08,
+ "loss": 0.0573,
+ "step": 35219
+ },
+ {
+ "epoch": 95.96730245231608,
+ "grad_norm": 0.589583158493042,
+ "learning_rate": 8.51726418863863e-08,
+ "loss": 0.0045,
+ "step": 35220
+ },
+ {
+ "epoch": 95.97002724795641,
+ "grad_norm": 1.2458869218826294,
+ "learning_rate": 8.505774599591566e-08,
+ "loss": 0.0432,
+ "step": 35221
+ },
+ {
+ "epoch": 95.97275204359673,
+ "grad_norm": 0.6286764144897461,
+ "learning_rate": 8.49429273226976e-08,
+ "loss": 0.0077,
+ "step": 35222
+ },
+ {
+ "epoch": 95.97547683923706,
+ "grad_norm": 1.0071245431900024,
+ "learning_rate": 8.482818586762365e-08,
+ "loss": 0.0481,
+ "step": 35223
+ },
+ {
+ "epoch": 95.97820163487738,
+ "grad_norm": 0.9049698710441589,
+ "learning_rate": 8.471352163158974e-08,
+ "loss": 0.0091,
+ "step": 35224
+ },
+ {
+ "epoch": 95.98092643051771,
+ "grad_norm": 1.1033416986465454,
+ "learning_rate": 8.45989346154863e-08,
+ "loss": 0.0137,
+ "step": 35225
+ },
+ {
+ "epoch": 95.98365122615803,
+ "grad_norm": 0.9366199374198914,
+ "learning_rate": 8.448442482020814e-08,
+ "loss": 0.0089,
+ "step": 35226
+ },
+ {
+ "epoch": 95.98637602179836,
+ "grad_norm": 1.2914425134658813,
+ "learning_rate": 8.436999224664677e-08,
+ "loss": 0.0753,
+ "step": 35227
+ },
+ {
+ "epoch": 95.9891008174387,
+ "grad_norm": 1.0404072999954224,
+ "learning_rate": 8.42556368956915e-08,
+ "loss": 0.009,
+ "step": 35228
+ },
+ {
+ "epoch": 95.99182561307902,
+ "grad_norm": 1.1364814043045044,
+ "learning_rate": 8.414135876823382e-08,
+ "loss": 0.0081,
+ "step": 35229
+ },
+ {
+ "epoch": 95.99455040871935,
+ "grad_norm": 1.8272546529769897,
+ "learning_rate": 8.402715786516524e-08,
+ "loss": 0.0192,
+ "step": 35230
+ },
+ {
+ "epoch": 95.99727520435967,
+ "grad_norm": 1.49300217628479,
+ "learning_rate": 8.391303418737395e-08,
+ "loss": 0.089,
+ "step": 35231
+ },
+ {
+ "epoch": 96.0,
+ "grad_norm": 0.9200458526611328,
+ "learning_rate": 8.379898773574924e-08,
+ "loss": 0.0233,
+ "step": 35232
+ },
+ {
+ "epoch": 96.00272479564033,
+ "grad_norm": 0.8763314485549927,
+ "learning_rate": 8.368501851117816e-08,
+ "loss": 0.0068,
+ "step": 35233
+ },
+ {
+ "epoch": 96.00544959128065,
+ "grad_norm": 0.989841878414154,
+ "learning_rate": 8.357112651454891e-08,
+ "loss": 0.0154,
+ "step": 35234
+ },
+ {
+ "epoch": 96.00817438692098,
+ "grad_norm": 1.7306077480316162,
+ "learning_rate": 8.345731174674854e-08,
+ "loss": 0.0179,
+ "step": 35235
+ },
+ {
+ "epoch": 96.0108991825613,
+ "grad_norm": 1.4381941556930542,
+ "learning_rate": 8.334357420866413e-08,
+ "loss": 0.0443,
+ "step": 35236
+ },
+ {
+ "epoch": 96.01362397820164,
+ "grad_norm": 1.0018935203552246,
+ "learning_rate": 8.322991390117941e-08,
+ "loss": 0.0175,
+ "step": 35237
+ },
+ {
+ "epoch": 96.01634877384195,
+ "grad_norm": 1.4569989442825317,
+ "learning_rate": 8.311633082518256e-08,
+ "loss": 0.0193,
+ "step": 35238
+ },
+ {
+ "epoch": 96.01907356948229,
+ "grad_norm": 1.066101312637329,
+ "learning_rate": 8.300282498155621e-08,
+ "loss": 0.0093,
+ "step": 35239
+ },
+ {
+ "epoch": 96.02179836512262,
+ "grad_norm": 0.749472439289093,
+ "learning_rate": 8.28893963711841e-08,
+ "loss": 0.0087,
+ "step": 35240
+ },
+ {
+ "epoch": 96.02452316076294,
+ "grad_norm": 0.796978235244751,
+ "learning_rate": 8.277604499494995e-08,
+ "loss": 0.0115,
+ "step": 35241
+ },
+ {
+ "epoch": 96.02724795640327,
+ "grad_norm": 0.9569693207740784,
+ "learning_rate": 8.266277085373753e-08,
+ "loss": 0.0056,
+ "step": 35242
+ },
+ {
+ "epoch": 96.02997275204359,
+ "grad_norm": 0.887015700340271,
+ "learning_rate": 8.254957394842833e-08,
+ "loss": 0.0076,
+ "step": 35243
+ },
+ {
+ "epoch": 96.03269754768392,
+ "grad_norm": 1.2845730781555176,
+ "learning_rate": 8.243645427990388e-08,
+ "loss": 0.0101,
+ "step": 35244
+ },
+ {
+ "epoch": 96.03542234332426,
+ "grad_norm": 0.47020238637924194,
+ "learning_rate": 8.232341184904458e-08,
+ "loss": 0.0039,
+ "step": 35245
+ },
+ {
+ "epoch": 96.03814713896458,
+ "grad_norm": 1.5733766555786133,
+ "learning_rate": 8.221044665673194e-08,
+ "loss": 0.0221,
+ "step": 35246
+ },
+ {
+ "epoch": 96.04087193460491,
+ "grad_norm": 1.2627713680267334,
+ "learning_rate": 8.209755870384417e-08,
+ "loss": 0.1459,
+ "step": 35247
+ },
+ {
+ "epoch": 96.04359673024523,
+ "grad_norm": 0.8329108357429504,
+ "learning_rate": 8.198474799126167e-08,
+ "loss": 0.0067,
+ "step": 35248
+ },
+ {
+ "epoch": 96.04632152588556,
+ "grad_norm": 0.6323956251144409,
+ "learning_rate": 8.18720145198626e-08,
+ "loss": 0.0084,
+ "step": 35249
+ },
+ {
+ "epoch": 96.04904632152588,
+ "grad_norm": 0.78331458568573,
+ "learning_rate": 8.175935829052628e-08,
+ "loss": 0.0056,
+ "step": 35250
+ },
+ {
+ "epoch": 96.05177111716621,
+ "grad_norm": 1.1879396438598633,
+ "learning_rate": 8.164677930412867e-08,
+ "loss": 0.0127,
+ "step": 35251
+ },
+ {
+ "epoch": 96.05449591280654,
+ "grad_norm": 0.7648675441741943,
+ "learning_rate": 8.153427756154575e-08,
+ "loss": 0.0231,
+ "step": 35252
+ },
+ {
+ "epoch": 96.05722070844686,
+ "grad_norm": 1.3489032983779907,
+ "learning_rate": 8.142185306365457e-08,
+ "loss": 0.1437,
+ "step": 35253
+ },
+ {
+ "epoch": 96.0599455040872,
+ "grad_norm": 1.2307409048080444,
+ "learning_rate": 8.130950581133113e-08,
+ "loss": 0.0216,
+ "step": 35254
+ },
+ {
+ "epoch": 96.06267029972751,
+ "grad_norm": 0.36908891797065735,
+ "learning_rate": 8.119723580545136e-08,
+ "loss": 0.004,
+ "step": 35255
+ },
+ {
+ "epoch": 96.06539509536785,
+ "grad_norm": 1.3528039455413818,
+ "learning_rate": 8.10850430468868e-08,
+ "loss": 0.0862,
+ "step": 35256
+ },
+ {
+ "epoch": 96.06811989100818,
+ "grad_norm": 0.6304677724838257,
+ "learning_rate": 8.097292753651232e-08,
+ "loss": 0.004,
+ "step": 35257
+ },
+ {
+ "epoch": 96.0708446866485,
+ "grad_norm": 0.9997826218605042,
+ "learning_rate": 8.086088927520274e-08,
+ "loss": 0.007,
+ "step": 35258
+ },
+ {
+ "epoch": 96.07356948228883,
+ "grad_norm": 2.0083374977111816,
+ "learning_rate": 8.074892826382852e-08,
+ "loss": 0.095,
+ "step": 35259
+ },
+ {
+ "epoch": 96.07629427792915,
+ "grad_norm": 0.823188304901123,
+ "learning_rate": 8.063704450326227e-08,
+ "loss": 0.0107,
+ "step": 35260
+ },
+ {
+ "epoch": 96.07901907356948,
+ "grad_norm": 0.8370969891548157,
+ "learning_rate": 8.052523799437551e-08,
+ "loss": 0.0119,
+ "step": 35261
+ },
+ {
+ "epoch": 96.0817438692098,
+ "grad_norm": 1.7621291875839233,
+ "learning_rate": 8.041350873803978e-08,
+ "loss": 0.0689,
+ "step": 35262
+ },
+ {
+ "epoch": 96.08446866485014,
+ "grad_norm": 0.815436840057373,
+ "learning_rate": 8.030185673512325e-08,
+ "loss": 0.0097,
+ "step": 35263
+ },
+ {
+ "epoch": 96.08719346049047,
+ "grad_norm": 0.7175476551055908,
+ "learning_rate": 8.019028198649747e-08,
+ "loss": 0.023,
+ "step": 35264
+ },
+ {
+ "epoch": 96.08991825613079,
+ "grad_norm": 0.5164682269096375,
+ "learning_rate": 8.007878449303064e-08,
+ "loss": 0.0053,
+ "step": 35265
+ },
+ {
+ "epoch": 96.09264305177112,
+ "grad_norm": 0.9006514549255371,
+ "learning_rate": 7.996736425558982e-08,
+ "loss": 0.0121,
+ "step": 35266
+ },
+ {
+ "epoch": 96.09536784741144,
+ "grad_norm": 0.788565456867218,
+ "learning_rate": 7.985602127504432e-08,
+ "loss": 0.0093,
+ "step": 35267
+ },
+ {
+ "epoch": 96.09809264305177,
+ "grad_norm": 1.2750164270401,
+ "learning_rate": 7.974475555226125e-08,
+ "loss": 0.0282,
+ "step": 35268
+ },
+ {
+ "epoch": 96.1008174386921,
+ "grad_norm": 0.7884176969528198,
+ "learning_rate": 7.963356708810654e-08,
+ "loss": 0.0072,
+ "step": 35269
+ },
+ {
+ "epoch": 96.10354223433242,
+ "grad_norm": 1.3492348194122314,
+ "learning_rate": 7.95224558834462e-08,
+ "loss": 0.0361,
+ "step": 35270
+ },
+ {
+ "epoch": 96.10626702997276,
+ "grad_norm": 1.1908611059188843,
+ "learning_rate": 7.941142193914619e-08,
+ "loss": 0.0221,
+ "step": 35271
+ },
+ {
+ "epoch": 96.10899182561307,
+ "grad_norm": 1.0446041822433472,
+ "learning_rate": 7.930046525607027e-08,
+ "loss": 0.0108,
+ "step": 35272
+ },
+ {
+ "epoch": 96.11171662125341,
+ "grad_norm": 1.3537211418151855,
+ "learning_rate": 7.91895858350833e-08,
+ "loss": 0.0403,
+ "step": 35273
+ },
+ {
+ "epoch": 96.11444141689373,
+ "grad_norm": 0.7542659640312195,
+ "learning_rate": 7.907878367704791e-08,
+ "loss": 0.0047,
+ "step": 35274
+ },
+ {
+ "epoch": 96.11716621253406,
+ "grad_norm": 1.0322134494781494,
+ "learning_rate": 7.896805878282898e-08,
+ "loss": 0.0263,
+ "step": 35275
+ },
+ {
+ "epoch": 96.11989100817439,
+ "grad_norm": 1.0721451044082642,
+ "learning_rate": 7.885741115328583e-08,
+ "loss": 0.0816,
+ "step": 35276
+ },
+ {
+ "epoch": 96.12261580381471,
+ "grad_norm": 0.8566046953201294,
+ "learning_rate": 7.87468407892833e-08,
+ "loss": 0.0076,
+ "step": 35277
+ },
+ {
+ "epoch": 96.12534059945504,
+ "grad_norm": 0.85439133644104,
+ "learning_rate": 7.86363476916796e-08,
+ "loss": 0.0149,
+ "step": 35278
+ },
+ {
+ "epoch": 96.12806539509536,
+ "grad_norm": 0.956926167011261,
+ "learning_rate": 7.85259318613385e-08,
+ "loss": 0.0109,
+ "step": 35279
+ },
+ {
+ "epoch": 96.1307901907357,
+ "grad_norm": 1.3024933338165283,
+ "learning_rate": 7.841559329911707e-08,
+ "loss": 0.0595,
+ "step": 35280
+ },
+ {
+ "epoch": 96.13351498637603,
+ "grad_norm": 0.9283532500267029,
+ "learning_rate": 7.830533200587576e-08,
+ "loss": 0.0086,
+ "step": 35281
+ },
+ {
+ "epoch": 96.13623978201635,
+ "grad_norm": 2.594912052154541,
+ "learning_rate": 7.819514798247274e-08,
+ "loss": 0.109,
+ "step": 35282
+ },
+ {
+ "epoch": 96.13896457765668,
+ "grad_norm": 1.704879641532898,
+ "learning_rate": 7.808504122976624e-08,
+ "loss": 0.0078,
+ "step": 35283
+ },
+ {
+ "epoch": 96.141689373297,
+ "grad_norm": 1.9496246576309204,
+ "learning_rate": 7.797501174861555e-08,
+ "loss": 0.031,
+ "step": 35284
+ },
+ {
+ "epoch": 96.14441416893733,
+ "grad_norm": 1.4647735357284546,
+ "learning_rate": 7.786505953987555e-08,
+ "loss": 0.0176,
+ "step": 35285
+ },
+ {
+ "epoch": 96.14713896457765,
+ "grad_norm": 0.9755575060844421,
+ "learning_rate": 7.775518460440223e-08,
+ "loss": 0.0208,
+ "step": 35286
+ },
+ {
+ "epoch": 96.14986376021798,
+ "grad_norm": 0.8531626462936401,
+ "learning_rate": 7.76453869430538e-08,
+ "loss": 0.0141,
+ "step": 35287
+ },
+ {
+ "epoch": 96.15258855585832,
+ "grad_norm": 1.569745659828186,
+ "learning_rate": 7.753566655668287e-08,
+ "loss": 0.0445,
+ "step": 35288
+ },
+ {
+ "epoch": 96.15531335149863,
+ "grad_norm": 1.111611247062683,
+ "learning_rate": 7.742602344614547e-08,
+ "loss": 0.0089,
+ "step": 35289
+ },
+ {
+ "epoch": 96.15803814713897,
+ "grad_norm": 0.5949771404266357,
+ "learning_rate": 7.731645761229311e-08,
+ "loss": 0.0056,
+ "step": 35290
+ },
+ {
+ "epoch": 96.16076294277929,
+ "grad_norm": 1.0812345743179321,
+ "learning_rate": 7.72069690559829e-08,
+ "loss": 0.0206,
+ "step": 35291
+ },
+ {
+ "epoch": 96.16348773841962,
+ "grad_norm": 0.7164740562438965,
+ "learning_rate": 7.709755777806416e-08,
+ "loss": 0.0045,
+ "step": 35292
+ },
+ {
+ "epoch": 96.16621253405995,
+ "grad_norm": 1.4342808723449707,
+ "learning_rate": 7.698822377939063e-08,
+ "loss": 0.0232,
+ "step": 35293
+ },
+ {
+ "epoch": 96.16893732970027,
+ "grad_norm": 0.7603983283042908,
+ "learning_rate": 7.687896706081388e-08,
+ "loss": 0.0058,
+ "step": 35294
+ },
+ {
+ "epoch": 96.1716621253406,
+ "grad_norm": 0.6402683258056641,
+ "learning_rate": 7.676978762318321e-08,
+ "loss": 0.007,
+ "step": 35295
+ },
+ {
+ "epoch": 96.17438692098092,
+ "grad_norm": 0.8324275016784668,
+ "learning_rate": 7.666068546735129e-08,
+ "loss": 0.012,
+ "step": 35296
+ },
+ {
+ "epoch": 96.17711171662125,
+ "grad_norm": 0.6949520707130432,
+ "learning_rate": 7.655166059416519e-08,
+ "loss": 0.0111,
+ "step": 35297
+ },
+ {
+ "epoch": 96.17983651226157,
+ "grad_norm": 0.704329788684845,
+ "learning_rate": 7.644271300447647e-08,
+ "loss": 0.0106,
+ "step": 35298
+ },
+ {
+ "epoch": 96.1825613079019,
+ "grad_norm": 0.9878809452056885,
+ "learning_rate": 7.633384269913224e-08,
+ "loss": 0.0082,
+ "step": 35299
+ },
+ {
+ "epoch": 96.18528610354224,
+ "grad_norm": 1.1737680435180664,
+ "learning_rate": 7.62250496789807e-08,
+ "loss": 0.0458,
+ "step": 35300
+ },
+ {
+ "epoch": 96.18801089918256,
+ "grad_norm": 1.4191699028015137,
+ "learning_rate": 7.611633394487005e-08,
+ "loss": 0.1047,
+ "step": 35301
+ },
+ {
+ "epoch": 96.19073569482289,
+ "grad_norm": 0.5077221393585205,
+ "learning_rate": 7.600769549764408e-08,
+ "loss": 0.0041,
+ "step": 35302
+ },
+ {
+ "epoch": 96.19346049046321,
+ "grad_norm": 1.1246440410614014,
+ "learning_rate": 7.58991343381521e-08,
+ "loss": 0.0063,
+ "step": 35303
+ },
+ {
+ "epoch": 96.19618528610354,
+ "grad_norm": 1.0835527181625366,
+ "learning_rate": 7.579065046723789e-08,
+ "loss": 0.012,
+ "step": 35304
+ },
+ {
+ "epoch": 96.19891008174388,
+ "grad_norm": 0.8698011040687561,
+ "learning_rate": 7.56822438857463e-08,
+ "loss": 0.0503,
+ "step": 35305
+ },
+ {
+ "epoch": 96.2016348773842,
+ "grad_norm": 0.481824666261673,
+ "learning_rate": 7.557391459452224e-08,
+ "loss": 0.0035,
+ "step": 35306
+ },
+ {
+ "epoch": 96.20435967302453,
+ "grad_norm": 1.2697802782058716,
+ "learning_rate": 7.546566259440946e-08,
+ "loss": 0.0116,
+ "step": 35307
+ },
+ {
+ "epoch": 96.20708446866485,
+ "grad_norm": 1.612202525138855,
+ "learning_rate": 7.535748788625064e-08,
+ "loss": 0.0131,
+ "step": 35308
+ },
+ {
+ "epoch": 96.20980926430518,
+ "grad_norm": 0.798570990562439,
+ "learning_rate": 7.524939047088842e-08,
+ "loss": 0.0146,
+ "step": 35309
+ },
+ {
+ "epoch": 96.2125340599455,
+ "grad_norm": 0.8485632538795471,
+ "learning_rate": 7.514137034916325e-08,
+ "loss": 0.082,
+ "step": 35310
+ },
+ {
+ "epoch": 96.21525885558583,
+ "grad_norm": 1.414610743522644,
+ "learning_rate": 7.503342752191889e-08,
+ "loss": 0.0268,
+ "step": 35311
+ },
+ {
+ "epoch": 96.21798365122616,
+ "grad_norm": 1.5406595468521118,
+ "learning_rate": 7.492556198999468e-08,
+ "loss": 0.0096,
+ "step": 35312
+ },
+ {
+ "epoch": 96.22070844686648,
+ "grad_norm": 0.9355694055557251,
+ "learning_rate": 7.481777375423105e-08,
+ "loss": 0.0099,
+ "step": 35313
+ },
+ {
+ "epoch": 96.22343324250681,
+ "grad_norm": 1.0451769828796387,
+ "learning_rate": 7.471006281546622e-08,
+ "loss": 0.0062,
+ "step": 35314
+ },
+ {
+ "epoch": 96.22615803814713,
+ "grad_norm": 0.6799387335777283,
+ "learning_rate": 7.460242917453952e-08,
+ "loss": 0.0068,
+ "step": 35315
+ },
+ {
+ "epoch": 96.22888283378747,
+ "grad_norm": 0.3932533860206604,
+ "learning_rate": 7.44948728322914e-08,
+ "loss": 0.005,
+ "step": 35316
+ },
+ {
+ "epoch": 96.2316076294278,
+ "grad_norm": 0.7780214548110962,
+ "learning_rate": 7.438739378955673e-08,
+ "loss": 0.0072,
+ "step": 35317
+ },
+ {
+ "epoch": 96.23433242506812,
+ "grad_norm": 2.3527355194091797,
+ "learning_rate": 7.427999204717374e-08,
+ "loss": 0.0857,
+ "step": 35318
+ },
+ {
+ "epoch": 96.23705722070845,
+ "grad_norm": 0.9330821633338928,
+ "learning_rate": 7.417266760597842e-08,
+ "loss": 0.009,
+ "step": 35319
+ },
+ {
+ "epoch": 96.23978201634877,
+ "grad_norm": 0.9175005555152893,
+ "learning_rate": 7.406542046680787e-08,
+ "loss": 0.0076,
+ "step": 35320
+ },
+ {
+ "epoch": 96.2425068119891,
+ "grad_norm": 0.9811730980873108,
+ "learning_rate": 7.395825063049588e-08,
+ "loss": 0.0061,
+ "step": 35321
+ },
+ {
+ "epoch": 96.24523160762942,
+ "grad_norm": 0.8991129398345947,
+ "learning_rate": 7.385115809787624e-08,
+ "loss": 0.013,
+ "step": 35322
+ },
+ {
+ "epoch": 96.24795640326975,
+ "grad_norm": 0.9068847298622131,
+ "learning_rate": 7.374414286978604e-08,
+ "loss": 0.0129,
+ "step": 35323
+ },
+ {
+ "epoch": 96.25068119891009,
+ "grad_norm": 1.3970781564712524,
+ "learning_rate": 7.363720494705684e-08,
+ "loss": 0.0545,
+ "step": 35324
+ },
+ {
+ "epoch": 96.2534059945504,
+ "grad_norm": 0.7774384021759033,
+ "learning_rate": 7.353034433052131e-08,
+ "loss": 0.0065,
+ "step": 35325
+ },
+ {
+ "epoch": 96.25613079019074,
+ "grad_norm": 1.112571120262146,
+ "learning_rate": 7.3423561021011e-08,
+ "loss": 0.1009,
+ "step": 35326
+ },
+ {
+ "epoch": 96.25885558583106,
+ "grad_norm": 1.4499166011810303,
+ "learning_rate": 7.33168550193597e-08,
+ "loss": 0.024,
+ "step": 35327
+ },
+ {
+ "epoch": 96.26158038147139,
+ "grad_norm": 1.2300962209701538,
+ "learning_rate": 7.321022632639674e-08,
+ "loss": 0.0122,
+ "step": 35328
+ },
+ {
+ "epoch": 96.26430517711172,
+ "grad_norm": 1.251291275024414,
+ "learning_rate": 7.310367494295367e-08,
+ "loss": 0.0115,
+ "step": 35329
+ },
+ {
+ "epoch": 96.26702997275204,
+ "grad_norm": 0.8060489296913147,
+ "learning_rate": 7.299720086985872e-08,
+ "loss": 0.0129,
+ "step": 35330
+ },
+ {
+ "epoch": 96.26975476839237,
+ "grad_norm": 1.0377684831619263,
+ "learning_rate": 7.289080410794236e-08,
+ "loss": 0.0109,
+ "step": 35331
+ },
+ {
+ "epoch": 96.2724795640327,
+ "grad_norm": 1.0229837894439697,
+ "learning_rate": 7.27844846580339e-08,
+ "loss": 0.0124,
+ "step": 35332
+ },
+ {
+ "epoch": 96.27520435967303,
+ "grad_norm": 1.2199400663375854,
+ "learning_rate": 7.267824252095823e-08,
+ "loss": 0.0251,
+ "step": 35333
+ },
+ {
+ "epoch": 96.27792915531334,
+ "grad_norm": 0.8657593727111816,
+ "learning_rate": 7.257207769754693e-08,
+ "loss": 0.0124,
+ "step": 35334
+ },
+ {
+ "epoch": 96.28065395095368,
+ "grad_norm": 0.8097676038742065,
+ "learning_rate": 7.246599018862377e-08,
+ "loss": 0.1277,
+ "step": 35335
+ },
+ {
+ "epoch": 96.28337874659401,
+ "grad_norm": 1.855539321899414,
+ "learning_rate": 7.235997999501587e-08,
+ "loss": 0.0295,
+ "step": 35336
+ },
+ {
+ "epoch": 96.28610354223433,
+ "grad_norm": 0.6716867089271545,
+ "learning_rate": 7.225404711755035e-08,
+ "loss": 0.0088,
+ "step": 35337
+ },
+ {
+ "epoch": 96.28882833787466,
+ "grad_norm": 0.9804553389549255,
+ "learning_rate": 7.214819155704878e-08,
+ "loss": 0.0614,
+ "step": 35338
+ },
+ {
+ "epoch": 96.29155313351498,
+ "grad_norm": 0.6539981961250305,
+ "learning_rate": 7.204241331433936e-08,
+ "loss": 0.005,
+ "step": 35339
+ },
+ {
+ "epoch": 96.29427792915531,
+ "grad_norm": 0.7259693145751953,
+ "learning_rate": 7.193671239024258e-08,
+ "loss": 0.0076,
+ "step": 35340
+ },
+ {
+ "epoch": 96.29700272479565,
+ "grad_norm": 1.4798357486724854,
+ "learning_rate": 7.183108878558443e-08,
+ "loss": 0.0444,
+ "step": 35341
+ },
+ {
+ "epoch": 96.29972752043597,
+ "grad_norm": 1.2626726627349854,
+ "learning_rate": 7.172554250118535e-08,
+ "loss": 0.0248,
+ "step": 35342
+ },
+ {
+ "epoch": 96.3024523160763,
+ "grad_norm": 1.1842796802520752,
+ "learning_rate": 7.162007353786915e-08,
+ "loss": 0.0232,
+ "step": 35343
+ },
+ {
+ "epoch": 96.30517711171662,
+ "grad_norm": 0.8027030229568481,
+ "learning_rate": 7.151468189645516e-08,
+ "loss": 0.0073,
+ "step": 35344
+ },
+ {
+ "epoch": 96.30790190735695,
+ "grad_norm": 0.7987467050552368,
+ "learning_rate": 7.140936757776607e-08,
+ "loss": 0.0111,
+ "step": 35345
+ },
+ {
+ "epoch": 96.31062670299727,
+ "grad_norm": 1.2186411619186401,
+ "learning_rate": 7.13041305826201e-08,
+ "loss": 0.0401,
+ "step": 35346
+ },
+ {
+ "epoch": 96.3133514986376,
+ "grad_norm": 0.5742531418800354,
+ "learning_rate": 7.119897091183991e-08,
+ "loss": 0.0065,
+ "step": 35347
+ },
+ {
+ "epoch": 96.31607629427793,
+ "grad_norm": 0.949908435344696,
+ "learning_rate": 7.109388856624156e-08,
+ "loss": 0.0209,
+ "step": 35348
+ },
+ {
+ "epoch": 96.31880108991825,
+ "grad_norm": 0.9327294826507568,
+ "learning_rate": 7.098888354664435e-08,
+ "loss": 0.048,
+ "step": 35349
+ },
+ {
+ "epoch": 96.32152588555859,
+ "grad_norm": 1.1320171356201172,
+ "learning_rate": 7.088395585386543e-08,
+ "loss": 0.014,
+ "step": 35350
+ },
+ {
+ "epoch": 96.3242506811989,
+ "grad_norm": 1.0687507390975952,
+ "learning_rate": 7.077910548872303e-08,
+ "loss": 0.0775,
+ "step": 35351
+ },
+ {
+ "epoch": 96.32697547683924,
+ "grad_norm": 1.2502411603927612,
+ "learning_rate": 7.067433245203426e-08,
+ "loss": 0.0155,
+ "step": 35352
+ },
+ {
+ "epoch": 96.32970027247957,
+ "grad_norm": 0.9892243146896362,
+ "learning_rate": 7.056963674461403e-08,
+ "loss": 0.0317,
+ "step": 35353
+ },
+ {
+ "epoch": 96.33242506811989,
+ "grad_norm": 1.4796222448349,
+ "learning_rate": 7.046501836727837e-08,
+ "loss": 0.0186,
+ "step": 35354
+ },
+ {
+ "epoch": 96.33514986376022,
+ "grad_norm": 0.5572302341461182,
+ "learning_rate": 7.036047732083995e-08,
+ "loss": 0.0066,
+ "step": 35355
+ },
+ {
+ "epoch": 96.33787465940054,
+ "grad_norm": 1.5197350978851318,
+ "learning_rate": 7.025601360611589e-08,
+ "loss": 0.0161,
+ "step": 35356
+ },
+ {
+ "epoch": 96.34059945504087,
+ "grad_norm": 2.6596696376800537,
+ "learning_rate": 7.015162722391777e-08,
+ "loss": 0.0085,
+ "step": 35357
+ },
+ {
+ "epoch": 96.34332425068119,
+ "grad_norm": 1.1613879203796387,
+ "learning_rate": 7.004731817505939e-08,
+ "loss": 0.0108,
+ "step": 35358
+ },
+ {
+ "epoch": 96.34604904632153,
+ "grad_norm": 1.0376052856445312,
+ "learning_rate": 6.994308646035342e-08,
+ "loss": 0.01,
+ "step": 35359
+ },
+ {
+ "epoch": 96.34877384196186,
+ "grad_norm": 1.0147429704666138,
+ "learning_rate": 6.983893208061032e-08,
+ "loss": 0.0088,
+ "step": 35360
+ },
+ {
+ "epoch": 96.35149863760218,
+ "grad_norm": 1.6167724132537842,
+ "learning_rate": 6.973485503664279e-08,
+ "loss": 0.0319,
+ "step": 35361
+ },
+ {
+ "epoch": 96.35422343324251,
+ "grad_norm": 0.6665457487106323,
+ "learning_rate": 6.963085532925906e-08,
+ "loss": 0.0061,
+ "step": 35362
+ },
+ {
+ "epoch": 96.35694822888283,
+ "grad_norm": 1.0524688959121704,
+ "learning_rate": 6.952693295927182e-08,
+ "loss": 0.0123,
+ "step": 35363
+ },
+ {
+ "epoch": 96.35967302452316,
+ "grad_norm": 0.9946007132530212,
+ "learning_rate": 6.942308792748931e-08,
+ "loss": 0.0128,
+ "step": 35364
+ },
+ {
+ "epoch": 96.3623978201635,
+ "grad_norm": 0.8363449573516846,
+ "learning_rate": 6.931932023471976e-08,
+ "loss": 0.0155,
+ "step": 35365
+ },
+ {
+ "epoch": 96.36512261580381,
+ "grad_norm": 1.2431470155715942,
+ "learning_rate": 6.921562988177144e-08,
+ "loss": 0.0821,
+ "step": 35366
+ },
+ {
+ "epoch": 96.36784741144415,
+ "grad_norm": 1.536660075187683,
+ "learning_rate": 6.911201686945257e-08,
+ "loss": 0.0204,
+ "step": 35367
+ },
+ {
+ "epoch": 96.37057220708446,
+ "grad_norm": 1.416346788406372,
+ "learning_rate": 6.90084811985703e-08,
+ "loss": 0.1268,
+ "step": 35368
+ },
+ {
+ "epoch": 96.3732970027248,
+ "grad_norm": 0.8056458234786987,
+ "learning_rate": 6.89050228699295e-08,
+ "loss": 0.0096,
+ "step": 35369
+ },
+ {
+ "epoch": 96.37602179836512,
+ "grad_norm": 0.7209925055503845,
+ "learning_rate": 6.880164188433735e-08,
+ "loss": 0.0102,
+ "step": 35370
+ },
+ {
+ "epoch": 96.37874659400545,
+ "grad_norm": 2.9088354110717773,
+ "learning_rate": 6.869833824259764e-08,
+ "loss": 0.007,
+ "step": 35371
+ },
+ {
+ "epoch": 96.38147138964578,
+ "grad_norm": 1.8868597745895386,
+ "learning_rate": 6.859511194551637e-08,
+ "loss": 0.1031,
+ "step": 35372
+ },
+ {
+ "epoch": 96.3841961852861,
+ "grad_norm": 0.5518752336502075,
+ "learning_rate": 6.849196299389627e-08,
+ "loss": 0.0044,
+ "step": 35373
+ },
+ {
+ "epoch": 96.38692098092643,
+ "grad_norm": 1.2854071855545044,
+ "learning_rate": 6.838889138854221e-08,
+ "loss": 0.017,
+ "step": 35374
+ },
+ {
+ "epoch": 96.38964577656675,
+ "grad_norm": 0.35395389795303345,
+ "learning_rate": 6.828589713025358e-08,
+ "loss": 0.0041,
+ "step": 35375
+ },
+ {
+ "epoch": 96.39237057220708,
+ "grad_norm": 0.6987147927284241,
+ "learning_rate": 6.818298021983638e-08,
+ "loss": 0.006,
+ "step": 35376
+ },
+ {
+ "epoch": 96.39509536784742,
+ "grad_norm": 0.7957742810249329,
+ "learning_rate": 6.808014065808998e-08,
+ "loss": 0.0106,
+ "step": 35377
+ },
+ {
+ "epoch": 96.39782016348774,
+ "grad_norm": 1.0435928106307983,
+ "learning_rate": 6.797737844581709e-08,
+ "loss": 0.0406,
+ "step": 35378
+ },
+ {
+ "epoch": 96.40054495912807,
+ "grad_norm": 0.7807527184486389,
+ "learning_rate": 6.787469358381483e-08,
+ "loss": 0.0078,
+ "step": 35379
+ },
+ {
+ "epoch": 96.40326975476839,
+ "grad_norm": 1.1360260248184204,
+ "learning_rate": 6.777208607288588e-08,
+ "loss": 0.0063,
+ "step": 35380
+ },
+ {
+ "epoch": 96.40599455040872,
+ "grad_norm": 1.3248192071914673,
+ "learning_rate": 6.76695559138274e-08,
+ "loss": 0.0254,
+ "step": 35381
+ },
+ {
+ "epoch": 96.40871934604904,
+ "grad_norm": 0.8698750734329224,
+ "learning_rate": 6.756710310743875e-08,
+ "loss": 0.0111,
+ "step": 35382
+ },
+ {
+ "epoch": 96.41144414168937,
+ "grad_norm": 0.5257014632225037,
+ "learning_rate": 6.746472765451928e-08,
+ "loss": 0.0054,
+ "step": 35383
+ },
+ {
+ "epoch": 96.4141689373297,
+ "grad_norm": 1.1118470430374146,
+ "learning_rate": 6.73624295558628e-08,
+ "loss": 0.0221,
+ "step": 35384
+ },
+ {
+ "epoch": 96.41689373297002,
+ "grad_norm": 0.657621443271637,
+ "learning_rate": 6.72602088122698e-08,
+ "loss": 0.0044,
+ "step": 35385
+ },
+ {
+ "epoch": 96.41961852861036,
+ "grad_norm": 0.8082538843154907,
+ "learning_rate": 6.715806542453407e-08,
+ "loss": 0.1076,
+ "step": 35386
+ },
+ {
+ "epoch": 96.42234332425068,
+ "grad_norm": 1.5939807891845703,
+ "learning_rate": 6.705599939345053e-08,
+ "loss": 0.0241,
+ "step": 35387
+ },
+ {
+ "epoch": 96.42506811989101,
+ "grad_norm": 1.101244568824768,
+ "learning_rate": 6.695401071981633e-08,
+ "loss": 0.05,
+ "step": 35388
+ },
+ {
+ "epoch": 96.42779291553134,
+ "grad_norm": 1.7461330890655518,
+ "learning_rate": 6.685209940442527e-08,
+ "loss": 0.0074,
+ "step": 35389
+ },
+ {
+ "epoch": 96.43051771117166,
+ "grad_norm": 1.2128634452819824,
+ "learning_rate": 6.675026544806895e-08,
+ "loss": 0.0194,
+ "step": 35390
+ },
+ {
+ "epoch": 96.433242506812,
+ "grad_norm": 5.596292495727539,
+ "learning_rate": 6.664850885154228e-08,
+ "loss": 0.0199,
+ "step": 35391
+ },
+ {
+ "epoch": 96.43596730245231,
+ "grad_norm": 1.5452741384506226,
+ "learning_rate": 6.654682961563797e-08,
+ "loss": 0.0401,
+ "step": 35392
+ },
+ {
+ "epoch": 96.43869209809264,
+ "grad_norm": 1.3662312030792236,
+ "learning_rate": 6.644522774114648e-08,
+ "loss": 0.0295,
+ "step": 35393
+ },
+ {
+ "epoch": 96.44141689373296,
+ "grad_norm": 0.7721708416938782,
+ "learning_rate": 6.634370322886052e-08,
+ "loss": 0.0099,
+ "step": 35394
+ },
+ {
+ "epoch": 96.4441416893733,
+ "grad_norm": 0.9421107172966003,
+ "learning_rate": 6.624225607956946e-08,
+ "loss": 0.0076,
+ "step": 35395
+ },
+ {
+ "epoch": 96.44686648501363,
+ "grad_norm": 1.1543314456939697,
+ "learning_rate": 6.614088629406379e-08,
+ "loss": 0.0215,
+ "step": 35396
+ },
+ {
+ "epoch": 96.44959128065395,
+ "grad_norm": 1.6893932819366455,
+ "learning_rate": 6.603959387313397e-08,
+ "loss": 0.0117,
+ "step": 35397
+ },
+ {
+ "epoch": 96.45231607629428,
+ "grad_norm": 0.6677010655403137,
+ "learning_rate": 6.593837881756826e-08,
+ "loss": 0.0081,
+ "step": 35398
+ },
+ {
+ "epoch": 96.4550408719346,
+ "grad_norm": 0.8486539125442505,
+ "learning_rate": 6.583724112815382e-08,
+ "loss": 0.009,
+ "step": 35399
+ },
+ {
+ "epoch": 96.45776566757493,
+ "grad_norm": 1.1725249290466309,
+ "learning_rate": 6.57361808056789e-08,
+ "loss": 0.0124,
+ "step": 35400
+ },
+ {
+ "epoch": 96.46049046321527,
+ "grad_norm": 0.7833046913146973,
+ "learning_rate": 6.563519785093175e-08,
+ "loss": 0.0067,
+ "step": 35401
+ },
+ {
+ "epoch": 96.46321525885558,
+ "grad_norm": 1.425215721130371,
+ "learning_rate": 6.553429226469732e-08,
+ "loss": 0.0654,
+ "step": 35402
+ },
+ {
+ "epoch": 96.46594005449592,
+ "grad_norm": 2.8748433589935303,
+ "learning_rate": 6.543346404776274e-08,
+ "loss": 0.0603,
+ "step": 35403
+ },
+ {
+ "epoch": 96.46866485013624,
+ "grad_norm": 1.7367169857025146,
+ "learning_rate": 6.533271320091183e-08,
+ "loss": 0.0456,
+ "step": 35404
+ },
+ {
+ "epoch": 96.47138964577657,
+ "grad_norm": 0.9711940288543701,
+ "learning_rate": 6.523203972493064e-08,
+ "loss": 0.0148,
+ "step": 35405
+ },
+ {
+ "epoch": 96.47411444141689,
+ "grad_norm": 1.2097513675689697,
+ "learning_rate": 6.513144362060297e-08,
+ "loss": 0.0203,
+ "step": 35406
+ },
+ {
+ "epoch": 96.47683923705722,
+ "grad_norm": 0.7398169040679932,
+ "learning_rate": 6.503092488871043e-08,
+ "loss": 0.0084,
+ "step": 35407
+ },
+ {
+ "epoch": 96.47956403269755,
+ "grad_norm": 1.5178656578063965,
+ "learning_rate": 6.493048353003905e-08,
+ "loss": 0.0086,
+ "step": 35408
+ },
+ {
+ "epoch": 96.48228882833787,
+ "grad_norm": 0.7791979312896729,
+ "learning_rate": 6.483011954536823e-08,
+ "loss": 0.0106,
+ "step": 35409
+ },
+ {
+ "epoch": 96.4850136239782,
+ "grad_norm": 1.6248571872711182,
+ "learning_rate": 6.472983293548063e-08,
+ "loss": 0.0428,
+ "step": 35410
+ },
+ {
+ "epoch": 96.48773841961852,
+ "grad_norm": 0.6717859506607056,
+ "learning_rate": 6.462962370115789e-08,
+ "loss": 0.0106,
+ "step": 35411
+ },
+ {
+ "epoch": 96.49046321525886,
+ "grad_norm": 0.761789083480835,
+ "learning_rate": 6.452949184317936e-08,
+ "loss": 0.0059,
+ "step": 35412
+ },
+ {
+ "epoch": 96.49318801089919,
+ "grad_norm": 1.1767845153808594,
+ "learning_rate": 6.442943736232554e-08,
+ "loss": 0.0128,
+ "step": 35413
+ },
+ {
+ "epoch": 96.49591280653951,
+ "grad_norm": 1.1814922094345093,
+ "learning_rate": 6.43294602593758e-08,
+ "loss": 0.0124,
+ "step": 35414
+ },
+ {
+ "epoch": 96.49863760217984,
+ "grad_norm": 0.6013202667236328,
+ "learning_rate": 6.422956053510842e-08,
+ "loss": 0.0048,
+ "step": 35415
+ },
+ {
+ "epoch": 96.50136239782016,
+ "grad_norm": 0.9613073468208313,
+ "learning_rate": 6.412973819030054e-08,
+ "loss": 0.0075,
+ "step": 35416
+ },
+ {
+ "epoch": 96.50408719346049,
+ "grad_norm": 0.604301929473877,
+ "learning_rate": 6.402999322573155e-08,
+ "loss": 0.0041,
+ "step": 35417
+ },
+ {
+ "epoch": 96.50681198910081,
+ "grad_norm": 1.8740332126617432,
+ "learning_rate": 6.393032564217749e-08,
+ "loss": 0.0945,
+ "step": 35418
+ },
+ {
+ "epoch": 96.50953678474114,
+ "grad_norm": 1.0526384115219116,
+ "learning_rate": 6.383073544041329e-08,
+ "loss": 0.0074,
+ "step": 35419
+ },
+ {
+ "epoch": 96.51226158038148,
+ "grad_norm": 1.1650419235229492,
+ "learning_rate": 6.373122262121501e-08,
+ "loss": 0.0273,
+ "step": 35420
+ },
+ {
+ "epoch": 96.5149863760218,
+ "grad_norm": 0.7087242603302002,
+ "learning_rate": 6.363178718535867e-08,
+ "loss": 0.0087,
+ "step": 35421
+ },
+ {
+ "epoch": 96.51771117166213,
+ "grad_norm": 1.2978801727294922,
+ "learning_rate": 6.353242913361923e-08,
+ "loss": 0.0105,
+ "step": 35422
+ },
+ {
+ "epoch": 96.52043596730245,
+ "grad_norm": 0.8758021593093872,
+ "learning_rate": 6.34331484667683e-08,
+ "loss": 0.0129,
+ "step": 35423
+ },
+ {
+ "epoch": 96.52316076294278,
+ "grad_norm": 0.809105634689331,
+ "learning_rate": 6.333394518558078e-08,
+ "loss": 0.0069,
+ "step": 35424
+ },
+ {
+ "epoch": 96.52588555858311,
+ "grad_norm": 0.8128492832183838,
+ "learning_rate": 6.323481929082831e-08,
+ "loss": 0.0051,
+ "step": 35425
+ },
+ {
+ "epoch": 96.52861035422343,
+ "grad_norm": 0.7639485001564026,
+ "learning_rate": 6.313577078328358e-08,
+ "loss": 0.0207,
+ "step": 35426
+ },
+ {
+ "epoch": 96.53133514986376,
+ "grad_norm": 0.7525876760482788,
+ "learning_rate": 6.303679966371712e-08,
+ "loss": 0.0086,
+ "step": 35427
+ },
+ {
+ "epoch": 96.53405994550408,
+ "grad_norm": 1.034570574760437,
+ "learning_rate": 6.29379059329005e-08,
+ "loss": 0.023,
+ "step": 35428
+ },
+ {
+ "epoch": 96.53678474114442,
+ "grad_norm": 0.6846426129341125,
+ "learning_rate": 6.283908959160424e-08,
+ "loss": 0.1011,
+ "step": 35429
+ },
+ {
+ "epoch": 96.53950953678473,
+ "grad_norm": 1.3742680549621582,
+ "learning_rate": 6.274035064059658e-08,
+ "loss": 0.0211,
+ "step": 35430
+ },
+ {
+ "epoch": 96.54223433242507,
+ "grad_norm": 1.3201643228530884,
+ "learning_rate": 6.264168908064694e-08,
+ "loss": 0.0704,
+ "step": 35431
+ },
+ {
+ "epoch": 96.5449591280654,
+ "grad_norm": 0.930867612361908,
+ "learning_rate": 6.254310491252469e-08,
+ "loss": 0.0378,
+ "step": 35432
+ },
+ {
+ "epoch": 96.54768392370572,
+ "grad_norm": 1.0854929685592651,
+ "learning_rate": 6.244459813699699e-08,
+ "loss": 0.0431,
+ "step": 35433
+ },
+ {
+ "epoch": 96.55040871934605,
+ "grad_norm": 0.7931585311889648,
+ "learning_rate": 6.234616875483101e-08,
+ "loss": 0.0081,
+ "step": 35434
+ },
+ {
+ "epoch": 96.55313351498637,
+ "grad_norm": 0.9414231777191162,
+ "learning_rate": 6.224781676679281e-08,
+ "loss": 0.0119,
+ "step": 35435
+ },
+ {
+ "epoch": 96.5558583106267,
+ "grad_norm": 1.446812391281128,
+ "learning_rate": 6.214954217364844e-08,
+ "loss": 0.0466,
+ "step": 35436
+ },
+ {
+ "epoch": 96.55858310626704,
+ "grad_norm": 1.0654802322387695,
+ "learning_rate": 6.205134497616394e-08,
+ "loss": 0.0095,
+ "step": 35437
+ },
+ {
+ "epoch": 96.56130790190736,
+ "grad_norm": 0.5225703716278076,
+ "learning_rate": 6.195322517510427e-08,
+ "loss": 0.0063,
+ "step": 35438
+ },
+ {
+ "epoch": 96.56403269754769,
+ "grad_norm": 0.5785564184188843,
+ "learning_rate": 6.185518277123215e-08,
+ "loss": 0.0055,
+ "step": 35439
+ },
+ {
+ "epoch": 96.566757493188,
+ "grad_norm": 0.5429187417030334,
+ "learning_rate": 6.175721776531251e-08,
+ "loss": 0.0051,
+ "step": 35440
+ },
+ {
+ "epoch": 96.56948228882834,
+ "grad_norm": 1.5912915468215942,
+ "learning_rate": 6.16593301581081e-08,
+ "loss": 0.0676,
+ "step": 35441
+ },
+ {
+ "epoch": 96.57220708446866,
+ "grad_norm": 0.8758354783058167,
+ "learning_rate": 6.156151995038052e-08,
+ "loss": 0.0154,
+ "step": 35442
+ },
+ {
+ "epoch": 96.57493188010899,
+ "grad_norm": 2.171773910522461,
+ "learning_rate": 6.146378714289136e-08,
+ "loss": 0.188,
+ "step": 35443
+ },
+ {
+ "epoch": 96.57765667574932,
+ "grad_norm": 0.9925405383110046,
+ "learning_rate": 6.136613173640338e-08,
+ "loss": 0.0179,
+ "step": 35444
+ },
+ {
+ "epoch": 96.58038147138964,
+ "grad_norm": 0.3850635886192322,
+ "learning_rate": 6.126855373167595e-08,
+ "loss": 0.0049,
+ "step": 35445
+ },
+ {
+ "epoch": 96.58310626702998,
+ "grad_norm": 0.8047220706939697,
+ "learning_rate": 6.117105312946848e-08,
+ "loss": 0.0144,
+ "step": 35446
+ },
+ {
+ "epoch": 96.5858310626703,
+ "grad_norm": 0.5075055956840515,
+ "learning_rate": 6.107362993054145e-08,
+ "loss": 0.0044,
+ "step": 35447
+ },
+ {
+ "epoch": 96.58855585831063,
+ "grad_norm": 1.511101245880127,
+ "learning_rate": 6.097628413565315e-08,
+ "loss": 0.0538,
+ "step": 35448
+ },
+ {
+ "epoch": 96.59128065395096,
+ "grad_norm": 0.6451207399368286,
+ "learning_rate": 6.087901574556188e-08,
+ "loss": 0.0052,
+ "step": 35449
+ },
+ {
+ "epoch": 96.59400544959128,
+ "grad_norm": 0.8259966373443604,
+ "learning_rate": 6.078182476102479e-08,
+ "loss": 0.006,
+ "step": 35450
+ },
+ {
+ "epoch": 96.59673024523161,
+ "grad_norm": 1.456793189048767,
+ "learning_rate": 6.068471118279796e-08,
+ "loss": 0.0205,
+ "step": 35451
+ },
+ {
+ "epoch": 96.59945504087193,
+ "grad_norm": 0.9304360151290894,
+ "learning_rate": 6.058767501163965e-08,
+ "loss": 0.0078,
+ "step": 35452
+ },
+ {
+ "epoch": 96.60217983651226,
+ "grad_norm": 1.0804176330566406,
+ "learning_rate": 6.049071624830371e-08,
+ "loss": 0.0618,
+ "step": 35453
+ },
+ {
+ "epoch": 96.60490463215258,
+ "grad_norm": 1.4847033023834229,
+ "learning_rate": 6.039383489354734e-08,
+ "loss": 0.0109,
+ "step": 35454
+ },
+ {
+ "epoch": 96.60762942779292,
+ "grad_norm": 1.090707778930664,
+ "learning_rate": 6.029703094812433e-08,
+ "loss": 0.0319,
+ "step": 35455
+ },
+ {
+ "epoch": 96.61035422343325,
+ "grad_norm": 1.5866479873657227,
+ "learning_rate": 6.020030441278635e-08,
+ "loss": 0.1306,
+ "step": 35456
+ },
+ {
+ "epoch": 96.61307901907357,
+ "grad_norm": 1.0959781408309937,
+ "learning_rate": 6.010365528828944e-08,
+ "loss": 0.0114,
+ "step": 35457
+ },
+ {
+ "epoch": 96.6158038147139,
+ "grad_norm": 0.7162144780158997,
+ "learning_rate": 6.00070835753852e-08,
+ "loss": 0.0127,
+ "step": 35458
+ },
+ {
+ "epoch": 96.61852861035422,
+ "grad_norm": 1.09774649143219,
+ "learning_rate": 5.991058927482641e-08,
+ "loss": 0.0307,
+ "step": 35459
+ },
+ {
+ "epoch": 96.62125340599455,
+ "grad_norm": 1.1278702020645142,
+ "learning_rate": 5.981417238736353e-08,
+ "loss": 0.0424,
+ "step": 35460
+ },
+ {
+ "epoch": 96.62397820163488,
+ "grad_norm": 1.0822484493255615,
+ "learning_rate": 5.971783291374822e-08,
+ "loss": 0.0333,
+ "step": 35461
+ },
+ {
+ "epoch": 96.6267029972752,
+ "grad_norm": 0.935177743434906,
+ "learning_rate": 5.962157085473097e-08,
+ "loss": 0.0737,
+ "step": 35462
+ },
+ {
+ "epoch": 96.62942779291554,
+ "grad_norm": 1.4999643564224243,
+ "learning_rate": 5.952538621106008e-08,
+ "loss": 0.0544,
+ "step": 35463
+ },
+ {
+ "epoch": 96.63215258855585,
+ "grad_norm": 1.1627200841903687,
+ "learning_rate": 5.942927898348494e-08,
+ "loss": 0.0179,
+ "step": 35464
+ },
+ {
+ "epoch": 96.63487738419619,
+ "grad_norm": 1.058876872062683,
+ "learning_rate": 5.933324917275607e-08,
+ "loss": 0.0087,
+ "step": 35465
+ },
+ {
+ "epoch": 96.6376021798365,
+ "grad_norm": 0.5869283080101013,
+ "learning_rate": 5.923729677961954e-08,
+ "loss": 0.0058,
+ "step": 35466
+ },
+ {
+ "epoch": 96.64032697547684,
+ "grad_norm": 1.019718885421753,
+ "learning_rate": 5.914142180482252e-08,
+ "loss": 0.0157,
+ "step": 35467
+ },
+ {
+ "epoch": 96.64305177111717,
+ "grad_norm": 0.8783105611801147,
+ "learning_rate": 5.90456242491122e-08,
+ "loss": 0.0107,
+ "step": 35468
+ },
+ {
+ "epoch": 96.64577656675749,
+ "grad_norm": 1.45699942111969,
+ "learning_rate": 5.8949904113235756e-08,
+ "loss": 0.0392,
+ "step": 35469
+ },
+ {
+ "epoch": 96.64850136239782,
+ "grad_norm": 1.6824450492858887,
+ "learning_rate": 5.885426139793593e-08,
+ "loss": 0.0902,
+ "step": 35470
+ },
+ {
+ "epoch": 96.65122615803814,
+ "grad_norm": 1.3906660079956055,
+ "learning_rate": 5.87586961039599e-08,
+ "loss": 0.0378,
+ "step": 35471
+ },
+ {
+ "epoch": 96.65395095367847,
+ "grad_norm": 0.48098742961883545,
+ "learning_rate": 5.866320823205152e-08,
+ "loss": 0.0054,
+ "step": 35472
+ },
+ {
+ "epoch": 96.65667574931881,
+ "grad_norm": 1.051723837852478,
+ "learning_rate": 5.856779778295352e-08,
+ "loss": 0.0226,
+ "step": 35473
+ },
+ {
+ "epoch": 96.65940054495913,
+ "grad_norm": 1.532332420349121,
+ "learning_rate": 5.847246475740975e-08,
+ "loss": 0.0264,
+ "step": 35474
+ },
+ {
+ "epoch": 96.66212534059946,
+ "grad_norm": 1.2175512313842773,
+ "learning_rate": 5.837720915616296e-08,
+ "loss": 0.026,
+ "step": 35475
+ },
+ {
+ "epoch": 96.66485013623978,
+ "grad_norm": 1.018885850906372,
+ "learning_rate": 5.828203097995366e-08,
+ "loss": 0.024,
+ "step": 35476
+ },
+ {
+ "epoch": 96.66757493188011,
+ "grad_norm": 0.9965648055076599,
+ "learning_rate": 5.8186930229524595e-08,
+ "loss": 0.0993,
+ "step": 35477
+ },
+ {
+ "epoch": 96.67029972752043,
+ "grad_norm": 1.429558515548706,
+ "learning_rate": 5.809190690561517e-08,
+ "loss": 0.0242,
+ "step": 35478
+ },
+ {
+ "epoch": 96.67302452316076,
+ "grad_norm": 0.6842082142829895,
+ "learning_rate": 5.799696100896701e-08,
+ "loss": 0.0106,
+ "step": 35479
+ },
+ {
+ "epoch": 96.6757493188011,
+ "grad_norm": 0.8472544550895691,
+ "learning_rate": 5.790209254031731e-08,
+ "loss": 0.0054,
+ "step": 35480
+ },
+ {
+ "epoch": 96.67847411444141,
+ "grad_norm": 0.9687547087669373,
+ "learning_rate": 5.780730150040659e-08,
+ "loss": 0.062,
+ "step": 35481
+ },
+ {
+ "epoch": 96.68119891008175,
+ "grad_norm": 1.1783868074417114,
+ "learning_rate": 5.7712587889973136e-08,
+ "loss": 0.0233,
+ "step": 35482
+ },
+ {
+ "epoch": 96.68392370572207,
+ "grad_norm": 1.2142536640167236,
+ "learning_rate": 5.761795170975304e-08,
+ "loss": 0.0138,
+ "step": 35483
+ },
+ {
+ "epoch": 96.6866485013624,
+ "grad_norm": 0.5116913914680481,
+ "learning_rate": 5.752339296048459e-08,
+ "loss": 0.0053,
+ "step": 35484
+ },
+ {
+ "epoch": 96.68937329700273,
+ "grad_norm": 1.5990698337554932,
+ "learning_rate": 5.742891164290498e-08,
+ "loss": 0.0151,
+ "step": 35485
+ },
+ {
+ "epoch": 96.69209809264305,
+ "grad_norm": 0.4173421263694763,
+ "learning_rate": 5.733450775774807e-08,
+ "loss": 0.005,
+ "step": 35486
+ },
+ {
+ "epoch": 96.69482288828338,
+ "grad_norm": 1.1129423379898071,
+ "learning_rate": 5.724018130574993e-08,
+ "loss": 0.0112,
+ "step": 35487
+ },
+ {
+ "epoch": 96.6975476839237,
+ "grad_norm": 0.9756750464439392,
+ "learning_rate": 5.714593228764553e-08,
+ "loss": 0.014,
+ "step": 35488
+ },
+ {
+ "epoch": 96.70027247956403,
+ "grad_norm": 0.7570908665657043,
+ "learning_rate": 5.7051760704168734e-08,
+ "loss": 0.0102,
+ "step": 35489
+ },
+ {
+ "epoch": 96.70299727520435,
+ "grad_norm": 0.447642058134079,
+ "learning_rate": 5.695766655605228e-08,
+ "loss": 0.0039,
+ "step": 35490
+ },
+ {
+ "epoch": 96.70572207084469,
+ "grad_norm": 1.4075931310653687,
+ "learning_rate": 5.686364984403003e-08,
+ "loss": 0.012,
+ "step": 35491
+ },
+ {
+ "epoch": 96.70844686648502,
+ "grad_norm": 1.0832757949829102,
+ "learning_rate": 5.676971056883362e-08,
+ "loss": 0.0656,
+ "step": 35492
+ },
+ {
+ "epoch": 96.71117166212534,
+ "grad_norm": 0.7135277390480042,
+ "learning_rate": 5.6675848731194695e-08,
+ "loss": 0.0066,
+ "step": 35493
+ },
+ {
+ "epoch": 96.71389645776567,
+ "grad_norm": 0.6995453834533691,
+ "learning_rate": 5.658206433184377e-08,
+ "loss": 0.0084,
+ "step": 35494
+ },
+ {
+ "epoch": 96.71662125340599,
+ "grad_norm": 0.7278634309768677,
+ "learning_rate": 5.648835737151248e-08,
+ "loss": 0.0227,
+ "step": 35495
+ },
+ {
+ "epoch": 96.71934604904632,
+ "grad_norm": 1.2676811218261719,
+ "learning_rate": 5.6394727850929145e-08,
+ "loss": 0.0181,
+ "step": 35496
+ },
+ {
+ "epoch": 96.72207084468666,
+ "grad_norm": 1.683194875717163,
+ "learning_rate": 5.6301175770824277e-08,
+ "loss": 0.0144,
+ "step": 35497
+ },
+ {
+ "epoch": 96.72479564032697,
+ "grad_norm": 1.0766547918319702,
+ "learning_rate": 5.620770113192509e-08,
+ "loss": 0.01,
+ "step": 35498
+ },
+ {
+ "epoch": 96.7275204359673,
+ "grad_norm": 1.448333740234375,
+ "learning_rate": 5.611430393496098e-08,
+ "loss": 0.0116,
+ "step": 35499
+ },
+ {
+ "epoch": 96.73024523160763,
+ "grad_norm": 1.2021136283874512,
+ "learning_rate": 5.602098418065804e-08,
+ "loss": 0.0099,
+ "step": 35500
+ },
+ {
+ "epoch": 96.73297002724796,
+ "grad_norm": 1.0989521741867065,
+ "learning_rate": 5.592774186974348e-08,
+ "loss": 0.0069,
+ "step": 35501
+ },
+ {
+ "epoch": 96.73569482288828,
+ "grad_norm": 0.8747550249099731,
+ "learning_rate": 5.5834577002944476e-08,
+ "loss": 0.0128,
+ "step": 35502
+ },
+ {
+ "epoch": 96.73841961852861,
+ "grad_norm": 1.1980881690979004,
+ "learning_rate": 5.574148958098491e-08,
+ "loss": 0.0603,
+ "step": 35503
+ },
+ {
+ "epoch": 96.74114441416894,
+ "grad_norm": 0.7756946086883545,
+ "learning_rate": 5.5648479604590856e-08,
+ "loss": 0.0084,
+ "step": 35504
+ },
+ {
+ "epoch": 96.74386920980926,
+ "grad_norm": 1.4742728471755981,
+ "learning_rate": 5.555554707448618e-08,
+ "loss": 0.0446,
+ "step": 35505
+ },
+ {
+ "epoch": 96.7465940054496,
+ "grad_norm": 0.9043451547622681,
+ "learning_rate": 5.5462691991395865e-08,
+ "loss": 0.0145,
+ "step": 35506
+ },
+ {
+ "epoch": 96.74931880108991,
+ "grad_norm": 1.2957626581192017,
+ "learning_rate": 5.5369914356041556e-08,
+ "loss": 0.0597,
+ "step": 35507
+ },
+ {
+ "epoch": 96.75204359673025,
+ "grad_norm": 1.0076942443847656,
+ "learning_rate": 5.527721416914489e-08,
+ "loss": 0.0147,
+ "step": 35508
+ },
+ {
+ "epoch": 96.75476839237058,
+ "grad_norm": 0.591376006603241,
+ "learning_rate": 5.5184591431430845e-08,
+ "loss": 0.0067,
+ "step": 35509
+ },
+ {
+ "epoch": 96.7574931880109,
+ "grad_norm": 1.5566169023513794,
+ "learning_rate": 5.5092046143618846e-08,
+ "loss": 0.0812,
+ "step": 35510
+ },
+ {
+ "epoch": 96.76021798365123,
+ "grad_norm": 0.5774223208427429,
+ "learning_rate": 5.4999578306429436e-08,
+ "loss": 0.0073,
+ "step": 35511
+ },
+ {
+ "epoch": 96.76294277929155,
+ "grad_norm": 1.2997885942459106,
+ "learning_rate": 5.490718792058314e-08,
+ "loss": 0.0482,
+ "step": 35512
+ },
+ {
+ "epoch": 96.76566757493188,
+ "grad_norm": 0.5321596264839172,
+ "learning_rate": 5.4814874986800495e-08,
+ "loss": 0.0056,
+ "step": 35513
+ },
+ {
+ "epoch": 96.7683923705722,
+ "grad_norm": 1.0160164833068848,
+ "learning_rate": 5.4722639505798704e-08,
+ "loss": 0.0133,
+ "step": 35514
+ },
+ {
+ "epoch": 96.77111716621253,
+ "grad_norm": 1.371077299118042,
+ "learning_rate": 5.4630481478297196e-08,
+ "loss": 0.0113,
+ "step": 35515
+ },
+ {
+ "epoch": 96.77384196185287,
+ "grad_norm": 1.393301248550415,
+ "learning_rate": 5.453840090501317e-08,
+ "loss": 0.02,
+ "step": 35516
+ },
+ {
+ "epoch": 96.77656675749319,
+ "grad_norm": 1.0892360210418701,
+ "learning_rate": 5.4446397786664936e-08,
+ "loss": 0.0863,
+ "step": 35517
+ },
+ {
+ "epoch": 96.77929155313352,
+ "grad_norm": 0.9813014268875122,
+ "learning_rate": 5.43544721239675e-08,
+ "loss": 0.0646,
+ "step": 35518
+ },
+ {
+ "epoch": 96.78201634877384,
+ "grad_norm": 1.0421407222747803,
+ "learning_rate": 5.4262623917636925e-08,
+ "loss": 0.0135,
+ "step": 35519
+ },
+ {
+ "epoch": 96.78474114441417,
+ "grad_norm": 0.7690547704696655,
+ "learning_rate": 5.417085316838933e-08,
+ "loss": 0.0102,
+ "step": 35520
+ },
+ {
+ "epoch": 96.7874659400545,
+ "grad_norm": 1.5118962526321411,
+ "learning_rate": 5.407915987693857e-08,
+ "loss": 0.0118,
+ "step": 35521
+ },
+ {
+ "epoch": 96.79019073569482,
+ "grad_norm": 1.9835567474365234,
+ "learning_rate": 5.398754404399964e-08,
+ "loss": 0.028,
+ "step": 35522
+ },
+ {
+ "epoch": 96.79291553133515,
+ "grad_norm": 0.9991610646247864,
+ "learning_rate": 5.38960056702853e-08,
+ "loss": 0.0525,
+ "step": 35523
+ },
+ {
+ "epoch": 96.79564032697547,
+ "grad_norm": 0.6667013764381409,
+ "learning_rate": 5.380454475650942e-08,
+ "loss": 0.0072,
+ "step": 35524
+ },
+ {
+ "epoch": 96.7983651226158,
+ "grad_norm": 1.360778570175171,
+ "learning_rate": 5.371316130338255e-08,
+ "loss": 0.0501,
+ "step": 35525
+ },
+ {
+ "epoch": 96.80108991825612,
+ "grad_norm": 0.9800916314125061,
+ "learning_rate": 5.362185531161746e-08,
+ "loss": 0.0289,
+ "step": 35526
+ },
+ {
+ "epoch": 96.80381471389646,
+ "grad_norm": 1.2056695222854614,
+ "learning_rate": 5.353062678192689e-08,
+ "loss": 0.0158,
+ "step": 35527
+ },
+ {
+ "epoch": 96.80653950953679,
+ "grad_norm": 0.8053821325302124,
+ "learning_rate": 5.343947571501806e-08,
+ "loss": 0.0092,
+ "step": 35528
+ },
+ {
+ "epoch": 96.80926430517711,
+ "grad_norm": 1.009867787361145,
+ "learning_rate": 5.334840211160264e-08,
+ "loss": 0.0095,
+ "step": 35529
+ },
+ {
+ "epoch": 96.81198910081744,
+ "grad_norm": 1.8366364240646362,
+ "learning_rate": 5.3257405972390043e-08,
+ "loss": 0.0095,
+ "step": 35530
+ },
+ {
+ "epoch": 96.81471389645776,
+ "grad_norm": 1.6652898788452148,
+ "learning_rate": 5.316648729808971e-08,
+ "loss": 0.0162,
+ "step": 35531
+ },
+ {
+ "epoch": 96.8174386920981,
+ "grad_norm": 1.0500088930130005,
+ "learning_rate": 5.307564608940663e-08,
+ "loss": 0.0157,
+ "step": 35532
+ },
+ {
+ "epoch": 96.82016348773843,
+ "grad_norm": 1.4442493915557861,
+ "learning_rate": 5.2984882347051345e-08,
+ "loss": 0.0059,
+ "step": 35533
+ },
+ {
+ "epoch": 96.82288828337875,
+ "grad_norm": 0.9895848631858826,
+ "learning_rate": 5.2894196071729966e-08,
+ "loss": 0.0124,
+ "step": 35534
+ },
+ {
+ "epoch": 96.82561307901908,
+ "grad_norm": 1.2646143436431885,
+ "learning_rate": 5.280358726414858e-08,
+ "loss": 0.0073,
+ "step": 35535
+ },
+ {
+ "epoch": 96.8283378746594,
+ "grad_norm": 0.8848221898078918,
+ "learning_rate": 5.271305592501108e-08,
+ "loss": 0.015,
+ "step": 35536
+ },
+ {
+ "epoch": 96.83106267029973,
+ "grad_norm": 1.3608940839767456,
+ "learning_rate": 5.262260205502578e-08,
+ "loss": 0.0425,
+ "step": 35537
+ },
+ {
+ "epoch": 96.83378746594005,
+ "grad_norm": 1.6552237272262573,
+ "learning_rate": 5.253222565489546e-08,
+ "loss": 0.0367,
+ "step": 35538
+ },
+ {
+ "epoch": 96.83651226158038,
+ "grad_norm": 1.5954221487045288,
+ "learning_rate": 5.244192672532289e-08,
+ "loss": 0.1259,
+ "step": 35539
+ },
+ {
+ "epoch": 96.83923705722071,
+ "grad_norm": 1.1544461250305176,
+ "learning_rate": 5.2351705267013055e-08,
+ "loss": 0.0127,
+ "step": 35540
+ },
+ {
+ "epoch": 96.84196185286103,
+ "grad_norm": 1.8766486644744873,
+ "learning_rate": 5.2261561280667614e-08,
+ "loss": 0.0473,
+ "step": 35541
+ },
+ {
+ "epoch": 96.84468664850137,
+ "grad_norm": 0.8398651480674744,
+ "learning_rate": 5.217149476698935e-08,
+ "loss": 0.0358,
+ "step": 35542
+ },
+ {
+ "epoch": 96.84741144414168,
+ "grad_norm": 1.4228144884109497,
+ "learning_rate": 5.2081505726679916e-08,
+ "loss": 0.0142,
+ "step": 35543
+ },
+ {
+ "epoch": 96.85013623978202,
+ "grad_norm": 1.0204805135726929,
+ "learning_rate": 5.199159416043875e-08,
+ "loss": 0.0069,
+ "step": 35544
+ },
+ {
+ "epoch": 96.85286103542235,
+ "grad_norm": 1.2904127836227417,
+ "learning_rate": 5.190176006896641e-08,
+ "loss": 0.0241,
+ "step": 35545
+ },
+ {
+ "epoch": 96.85558583106267,
+ "grad_norm": 1.2785520553588867,
+ "learning_rate": 5.181200345296455e-08,
+ "loss": 0.0131,
+ "step": 35546
+ },
+ {
+ "epoch": 96.858310626703,
+ "grad_norm": 0.542849063873291,
+ "learning_rate": 5.172232431312929e-08,
+ "loss": 0.0054,
+ "step": 35547
+ },
+ {
+ "epoch": 96.86103542234332,
+ "grad_norm": 0.8196302652359009,
+ "learning_rate": 5.163272265016117e-08,
+ "loss": 0.0134,
+ "step": 35548
+ },
+ {
+ "epoch": 96.86376021798365,
+ "grad_norm": 0.6417269706726074,
+ "learning_rate": 5.154319846475631e-08,
+ "loss": 0.0057,
+ "step": 35549
+ },
+ {
+ "epoch": 96.86648501362397,
+ "grad_norm": 0.7701423764228821,
+ "learning_rate": 5.1453751757614136e-08,
+ "loss": 0.0063,
+ "step": 35550
+ },
+ {
+ "epoch": 96.8692098092643,
+ "grad_norm": 0.569082498550415,
+ "learning_rate": 5.1364382529428545e-08,
+ "loss": 0.0047,
+ "step": 35551
+ },
+ {
+ "epoch": 96.87193460490464,
+ "grad_norm": 1.675007700920105,
+ "learning_rate": 5.1275090780897875e-08,
+ "loss": 0.0176,
+ "step": 35552
+ },
+ {
+ "epoch": 96.87465940054496,
+ "grad_norm": 0.41582930088043213,
+ "learning_rate": 5.1185876512716004e-08,
+ "loss": 0.0043,
+ "step": 35553
+ },
+ {
+ "epoch": 96.87738419618529,
+ "grad_norm": 0.8605927228927612,
+ "learning_rate": 5.109673972557905e-08,
+ "loss": 0.0069,
+ "step": 35554
+ },
+ {
+ "epoch": 96.88010899182561,
+ "grad_norm": 1.0190485715866089,
+ "learning_rate": 5.1007680420179786e-08,
+ "loss": 0.0098,
+ "step": 35555
+ },
+ {
+ "epoch": 96.88283378746594,
+ "grad_norm": 1.2118732929229736,
+ "learning_rate": 5.0918698597212104e-08,
+ "loss": 0.1006,
+ "step": 35556
+ },
+ {
+ "epoch": 96.88555858310627,
+ "grad_norm": 1.185126543045044,
+ "learning_rate": 5.082979425736989e-08,
+ "loss": 0.0546,
+ "step": 35557
+ },
+ {
+ "epoch": 96.88828337874659,
+ "grad_norm": 0.9090191721916199,
+ "learning_rate": 5.074096740134593e-08,
+ "loss": 0.0094,
+ "step": 35558
+ },
+ {
+ "epoch": 96.89100817438693,
+ "grad_norm": 1.4923572540283203,
+ "learning_rate": 5.065221802983078e-08,
+ "loss": 0.0359,
+ "step": 35559
+ },
+ {
+ "epoch": 96.89373297002724,
+ "grad_norm": 1.006446361541748,
+ "learning_rate": 5.0563546143515e-08,
+ "loss": 0.0778,
+ "step": 35560
+ },
+ {
+ "epoch": 96.89645776566758,
+ "grad_norm": 0.8379726409912109,
+ "learning_rate": 5.047495174309025e-08,
+ "loss": 0.0103,
+ "step": 35561
+ },
+ {
+ "epoch": 96.8991825613079,
+ "grad_norm": 0.7311747074127197,
+ "learning_rate": 5.03864348292471e-08,
+ "loss": 0.0079,
+ "step": 35562
+ },
+ {
+ "epoch": 96.90190735694823,
+ "grad_norm": 0.6814349889755249,
+ "learning_rate": 5.0297995402673884e-08,
+ "loss": 0.0073,
+ "step": 35563
+ },
+ {
+ "epoch": 96.90463215258856,
+ "grad_norm": 0.7348998785018921,
+ "learning_rate": 5.0209633464060046e-08,
+ "loss": 0.0042,
+ "step": 35564
+ },
+ {
+ "epoch": 96.90735694822888,
+ "grad_norm": 1.1728848218917847,
+ "learning_rate": 5.0121349014091715e-08,
+ "loss": 0.0645,
+ "step": 35565
+ },
+ {
+ "epoch": 96.91008174386921,
+ "grad_norm": 1.2327337265014648,
+ "learning_rate": 5.003314205345944e-08,
+ "loss": 0.0483,
+ "step": 35566
+ },
+ {
+ "epoch": 96.91280653950953,
+ "grad_norm": 1.1757078170776367,
+ "learning_rate": 4.994501258284934e-08,
+ "loss": 0.1012,
+ "step": 35567
+ },
+ {
+ "epoch": 96.91553133514986,
+ "grad_norm": 0.6585571765899658,
+ "learning_rate": 4.985696060294642e-08,
+ "loss": 0.0066,
+ "step": 35568
+ },
+ {
+ "epoch": 96.9182561307902,
+ "grad_norm": 2.361471176147461,
+ "learning_rate": 4.9768986114436814e-08,
+ "loss": 0.0339,
+ "step": 35569
+ },
+ {
+ "epoch": 96.92098092643052,
+ "grad_norm": 1.6934252977371216,
+ "learning_rate": 4.968108911800662e-08,
+ "loss": 0.0172,
+ "step": 35570
+ },
+ {
+ "epoch": 96.92370572207085,
+ "grad_norm": 0.6433117985725403,
+ "learning_rate": 4.9593269614339744e-08,
+ "loss": 0.0077,
+ "step": 35571
+ },
+ {
+ "epoch": 96.92643051771117,
+ "grad_norm": 0.719902753829956,
+ "learning_rate": 4.9505527604121194e-08,
+ "loss": 0.0058,
+ "step": 35572
+ },
+ {
+ "epoch": 96.9291553133515,
+ "grad_norm": 1.3307214975357056,
+ "learning_rate": 4.941786308803154e-08,
+ "loss": 0.0461,
+ "step": 35573
+ },
+ {
+ "epoch": 96.93188010899182,
+ "grad_norm": 1.1628620624542236,
+ "learning_rate": 4.933027606675689e-08,
+ "loss": 0.0292,
+ "step": 35574
+ },
+ {
+ "epoch": 96.93460490463215,
+ "grad_norm": 0.37249767780303955,
+ "learning_rate": 4.924276654097671e-08,
+ "loss": 0.0047,
+ "step": 35575
+ },
+ {
+ "epoch": 96.93732970027249,
+ "grad_norm": 1.2477262020111084,
+ "learning_rate": 4.9155334511373775e-08,
+ "loss": 0.0516,
+ "step": 35576
+ },
+ {
+ "epoch": 96.9400544959128,
+ "grad_norm": 1.494789958000183,
+ "learning_rate": 4.9067979978629776e-08,
+ "loss": 0.1202,
+ "step": 35577
+ },
+ {
+ "epoch": 96.94277929155314,
+ "grad_norm": 0.3458223342895508,
+ "learning_rate": 4.898070294342305e-08,
+ "loss": 0.0039,
+ "step": 35578
+ },
+ {
+ "epoch": 96.94550408719346,
+ "grad_norm": 1.1049401760101318,
+ "learning_rate": 4.8893503406435285e-08,
+ "loss": 0.0099,
+ "step": 35579
+ },
+ {
+ "epoch": 96.94822888283379,
+ "grad_norm": 0.33557599782943726,
+ "learning_rate": 4.8806381368343705e-08,
+ "loss": 0.0031,
+ "step": 35580
+ },
+ {
+ "epoch": 96.95095367847412,
+ "grad_norm": 0.8286622166633606,
+ "learning_rate": 4.8719336829827765e-08,
+ "loss": 0.0784,
+ "step": 35581
+ },
+ {
+ "epoch": 96.95367847411444,
+ "grad_norm": 1.1306160688400269,
+ "learning_rate": 4.8632369791565825e-08,
+ "loss": 0.0166,
+ "step": 35582
+ },
+ {
+ "epoch": 96.95640326975477,
+ "grad_norm": 0.8637627959251404,
+ "learning_rate": 4.854548025423511e-08,
+ "loss": 0.0081,
+ "step": 35583
+ },
+ {
+ "epoch": 96.95912806539509,
+ "grad_norm": 1.5423208475112915,
+ "learning_rate": 4.845866821851064e-08,
+ "loss": 0.0562,
+ "step": 35584
+ },
+ {
+ "epoch": 96.96185286103542,
+ "grad_norm": 0.6251918077468872,
+ "learning_rate": 4.837193368507076e-08,
+ "loss": 0.007,
+ "step": 35585
+ },
+ {
+ "epoch": 96.96457765667574,
+ "grad_norm": 0.8687853217124939,
+ "learning_rate": 4.828527665458937e-08,
+ "loss": 0.0077,
+ "step": 35586
+ },
+ {
+ "epoch": 96.96730245231608,
+ "grad_norm": 1.2604279518127441,
+ "learning_rate": 4.8198697127742613e-08,
+ "loss": 0.0138,
+ "step": 35587
+ },
+ {
+ "epoch": 96.97002724795641,
+ "grad_norm": 1.0283077955245972,
+ "learning_rate": 4.811219510520437e-08,
+ "loss": 0.0057,
+ "step": 35588
+ },
+ {
+ "epoch": 96.97275204359673,
+ "grad_norm": 0.5833441615104675,
+ "learning_rate": 4.802577058764635e-08,
+ "loss": 0.0044,
+ "step": 35589
+ },
+ {
+ "epoch": 96.97547683923706,
+ "grad_norm": 0.8689666986465454,
+ "learning_rate": 4.793942357574466e-08,
+ "loss": 0.0106,
+ "step": 35590
+ },
+ {
+ "epoch": 96.97820163487738,
+ "grad_norm": 2.910576105117798,
+ "learning_rate": 4.7853154070169885e-08,
+ "loss": 0.0101,
+ "step": 35591
+ },
+ {
+ "epoch": 96.98092643051771,
+ "grad_norm": 2.8830645084381104,
+ "learning_rate": 4.776696207159481e-08,
+ "loss": 0.0078,
+ "step": 35592
+ },
+ {
+ "epoch": 96.98365122615803,
+ "grad_norm": 1.1054487228393555,
+ "learning_rate": 4.7680847580691136e-08,
+ "loss": 0.0779,
+ "step": 35593
+ },
+ {
+ "epoch": 96.98637602179836,
+ "grad_norm": 1.1755321025848389,
+ "learning_rate": 4.75948105981272e-08,
+ "loss": 0.0486,
+ "step": 35594
+ },
+ {
+ "epoch": 96.9891008174387,
+ "grad_norm": 0.8826853036880493,
+ "learning_rate": 4.7508851124574705e-08,
+ "loss": 0.0299,
+ "step": 35595
+ },
+ {
+ "epoch": 96.99182561307902,
+ "grad_norm": 1.003972053527832,
+ "learning_rate": 4.742296916070421e-08,
+ "loss": 0.0369,
+ "step": 35596
+ },
+ {
+ "epoch": 96.99455040871935,
+ "grad_norm": 0.6746710538864136,
+ "learning_rate": 4.733716470718186e-08,
+ "loss": 0.0072,
+ "step": 35597
+ },
+ {
+ "epoch": 96.99727520435967,
+ "grad_norm": 1.9137908220291138,
+ "learning_rate": 4.725143776467822e-08,
+ "loss": 0.0249,
+ "step": 35598
+ },
+ {
+ "epoch": 97.0,
+ "grad_norm": 0.8281910419464111,
+ "learning_rate": 4.716578833386054e-08,
+ "loss": 0.0057,
+ "step": 35599
+ },
+ {
+ "epoch": 97.00272479564033,
+ "grad_norm": 1.0648388862609863,
+ "learning_rate": 4.7080216415394954e-08,
+ "loss": 0.0162,
+ "step": 35600
+ },
+ {
+ "epoch": 97.00544959128065,
+ "grad_norm": 0.6285977959632874,
+ "learning_rate": 4.699472200994759e-08,
+ "loss": 0.0055,
+ "step": 35601
+ },
+ {
+ "epoch": 97.00817438692098,
+ "grad_norm": 0.8425440192222595,
+ "learning_rate": 4.69093051181857e-08,
+ "loss": 0.0146,
+ "step": 35602
+ },
+ {
+ "epoch": 97.0108991825613,
+ "grad_norm": 0.9732640981674194,
+ "learning_rate": 4.682396574077319e-08,
+ "loss": 0.0082,
+ "step": 35603
+ },
+ {
+ "epoch": 97.01362397820164,
+ "grad_norm": 0.8449105024337769,
+ "learning_rate": 4.6738703878376204e-08,
+ "loss": 0.0119,
+ "step": 35604
+ },
+ {
+ "epoch": 97.01634877384195,
+ "grad_norm": 0.6733137965202332,
+ "learning_rate": 4.665351953165642e-08,
+ "loss": 0.0071,
+ "step": 35605
+ },
+ {
+ "epoch": 97.01907356948229,
+ "grad_norm": 1.1684843301773071,
+ "learning_rate": 4.656841270127999e-08,
+ "loss": 0.0431,
+ "step": 35606
+ },
+ {
+ "epoch": 97.02179836512262,
+ "grad_norm": 1.6330671310424805,
+ "learning_rate": 4.6483383387908585e-08,
+ "loss": 0.0303,
+ "step": 35607
+ },
+ {
+ "epoch": 97.02452316076294,
+ "grad_norm": 1.4565246105194092,
+ "learning_rate": 4.6398431592202805e-08,
+ "loss": 0.017,
+ "step": 35608
+ },
+ {
+ "epoch": 97.02724795640327,
+ "grad_norm": 0.7698649764060974,
+ "learning_rate": 4.6313557314826564e-08,
+ "loss": 0.0098,
+ "step": 35609
+ },
+ {
+ "epoch": 97.02997275204359,
+ "grad_norm": 0.9091979265213013,
+ "learning_rate": 4.622876055643932e-08,
+ "loss": 0.0064,
+ "step": 35610
+ },
+ {
+ "epoch": 97.03269754768392,
+ "grad_norm": 0.6510867476463318,
+ "learning_rate": 4.614404131770389e-08,
+ "loss": 0.0067,
+ "step": 35611
+ },
+ {
+ "epoch": 97.03542234332426,
+ "grad_norm": 1.1431992053985596,
+ "learning_rate": 4.605939959927641e-08,
+ "loss": 0.0151,
+ "step": 35612
+ },
+ {
+ "epoch": 97.03814713896458,
+ "grad_norm": 0.7004616856575012,
+ "learning_rate": 4.597483540181968e-08,
+ "loss": 0.006,
+ "step": 35613
+ },
+ {
+ "epoch": 97.04087193460491,
+ "grad_norm": 1.091273307800293,
+ "learning_rate": 4.5890348725988744e-08,
+ "loss": 0.007,
+ "step": 35614
+ },
+ {
+ "epoch": 97.04359673024523,
+ "grad_norm": 0.43244725465774536,
+ "learning_rate": 4.5805939572444166e-08,
+ "loss": 0.0046,
+ "step": 35615
+ },
+ {
+ "epoch": 97.04632152588556,
+ "grad_norm": 1.2420799732208252,
+ "learning_rate": 4.572160794184322e-08,
+ "loss": 0.0116,
+ "step": 35616
+ },
+ {
+ "epoch": 97.04904632152588,
+ "grad_norm": 0.831074059009552,
+ "learning_rate": 4.563735383484091e-08,
+ "loss": 0.0085,
+ "step": 35617
+ },
+ {
+ "epoch": 97.05177111716621,
+ "grad_norm": 1.4915581941604614,
+ "learning_rate": 4.5553177252095624e-08,
+ "loss": 0.0155,
+ "step": 35618
+ },
+ {
+ "epoch": 97.05449591280654,
+ "grad_norm": 0.7203131318092346,
+ "learning_rate": 4.5469078194261276e-08,
+ "loss": 0.0073,
+ "step": 35619
+ },
+ {
+ "epoch": 97.05722070844686,
+ "grad_norm": 0.424826979637146,
+ "learning_rate": 4.5385056661992886e-08,
+ "loss": 0.0051,
+ "step": 35620
+ },
+ {
+ "epoch": 97.0599455040872,
+ "grad_norm": 0.4368210732936859,
+ "learning_rate": 4.53011126559455e-08,
+ "loss": 0.0039,
+ "step": 35621
+ },
+ {
+ "epoch": 97.06267029972751,
+ "grad_norm": 0.3822444677352905,
+ "learning_rate": 4.5217246176773036e-08,
+ "loss": 0.0039,
+ "step": 35622
+ },
+ {
+ "epoch": 97.06539509536785,
+ "grad_norm": 0.9697496294975281,
+ "learning_rate": 4.5133457225128296e-08,
+ "loss": 0.0077,
+ "step": 35623
+ },
+ {
+ "epoch": 97.06811989100818,
+ "grad_norm": 1.2951271533966064,
+ "learning_rate": 4.50497458016641e-08,
+ "loss": 0.0245,
+ "step": 35624
+ },
+ {
+ "epoch": 97.0708446866485,
+ "grad_norm": 0.7218852043151855,
+ "learning_rate": 4.496611190703104e-08,
+ "loss": 0.0113,
+ "step": 35625
+ },
+ {
+ "epoch": 97.07356948228883,
+ "grad_norm": 1.1558761596679688,
+ "learning_rate": 4.488255554188192e-08,
+ "loss": 0.0675,
+ "step": 35626
+ },
+ {
+ "epoch": 97.07629427792915,
+ "grad_norm": 1.1268943548202515,
+ "learning_rate": 4.479907670686734e-08,
+ "loss": 0.0102,
+ "step": 35627
+ },
+ {
+ "epoch": 97.07901907356948,
+ "grad_norm": 0.8783280849456787,
+ "learning_rate": 4.4715675402636774e-08,
+ "loss": 0.0119,
+ "step": 35628
+ },
+ {
+ "epoch": 97.0817438692098,
+ "grad_norm": 1.4471755027770996,
+ "learning_rate": 4.463235162984081e-08,
+ "loss": 0.0101,
+ "step": 35629
+ },
+ {
+ "epoch": 97.08446866485014,
+ "grad_norm": 0.5426023602485657,
+ "learning_rate": 4.454910538912671e-08,
+ "loss": 0.0057,
+ "step": 35630
+ },
+ {
+ "epoch": 97.08719346049047,
+ "grad_norm": 0.9726139307022095,
+ "learning_rate": 4.446593668114396e-08,
+ "loss": 0.0407,
+ "step": 35631
+ },
+ {
+ "epoch": 97.08991825613079,
+ "grad_norm": 1.0226190090179443,
+ "learning_rate": 4.438284550654093e-08,
+ "loss": 0.0113,
+ "step": 35632
+ },
+ {
+ "epoch": 97.09264305177112,
+ "grad_norm": 1.2721867561340332,
+ "learning_rate": 4.4299831865962653e-08,
+ "loss": 0.081,
+ "step": 35633
+ },
+ {
+ "epoch": 97.09536784741144,
+ "grad_norm": 2.58241605758667,
+ "learning_rate": 4.42168957600575e-08,
+ "loss": 0.0481,
+ "step": 35634
+ },
+ {
+ "epoch": 97.09809264305177,
+ "grad_norm": 1.2095149755477905,
+ "learning_rate": 4.413403718947162e-08,
+ "loss": 0.0162,
+ "step": 35635
+ },
+ {
+ "epoch": 97.1008174386921,
+ "grad_norm": 0.8570220470428467,
+ "learning_rate": 4.4051256154847844e-08,
+ "loss": 0.0051,
+ "step": 35636
+ },
+ {
+ "epoch": 97.10354223433242,
+ "grad_norm": 0.7268704175949097,
+ "learning_rate": 4.396855265683342e-08,
+ "loss": 0.0064,
+ "step": 35637
+ },
+ {
+ "epoch": 97.10626702997276,
+ "grad_norm": 0.503399133682251,
+ "learning_rate": 4.388592669607117e-08,
+ "loss": 0.0051,
+ "step": 35638
+ },
+ {
+ "epoch": 97.10899182561307,
+ "grad_norm": 0.6108579635620117,
+ "learning_rate": 4.3803378273205024e-08,
+ "loss": 0.0051,
+ "step": 35639
+ },
+ {
+ "epoch": 97.11171662125341,
+ "grad_norm": 2.4232873916625977,
+ "learning_rate": 4.37209073888778e-08,
+ "loss": 0.0966,
+ "step": 35640
+ },
+ {
+ "epoch": 97.11444141689373,
+ "grad_norm": 1.0452427864074707,
+ "learning_rate": 4.363851404373121e-08,
+ "loss": 0.012,
+ "step": 35641
+ },
+ {
+ "epoch": 97.11716621253406,
+ "grad_norm": 1.1466513872146606,
+ "learning_rate": 4.3556198238408064e-08,
+ "loss": 0.0182,
+ "step": 35642
+ },
+ {
+ "epoch": 97.11989100817439,
+ "grad_norm": 1.1299593448638916,
+ "learning_rate": 4.3473959973547866e-08,
+ "loss": 0.0208,
+ "step": 35643
+ },
+ {
+ "epoch": 97.12261580381471,
+ "grad_norm": 1.167195200920105,
+ "learning_rate": 4.339179924979342e-08,
+ "loss": 0.0456,
+ "step": 35644
+ },
+ {
+ "epoch": 97.12534059945504,
+ "grad_norm": 1.0740149021148682,
+ "learning_rate": 4.3309716067782e-08,
+ "loss": 0.0825,
+ "step": 35645
+ },
+ {
+ "epoch": 97.12806539509536,
+ "grad_norm": 0.8909195065498352,
+ "learning_rate": 4.322771042815421e-08,
+ "loss": 0.0056,
+ "step": 35646
+ },
+ {
+ "epoch": 97.1307901907357,
+ "grad_norm": 0.5913050174713135,
+ "learning_rate": 4.314578233154953e-08,
+ "loss": 0.0059,
+ "step": 35647
+ },
+ {
+ "epoch": 97.13351498637603,
+ "grad_norm": 0.6551040410995483,
+ "learning_rate": 4.306393177860413e-08,
+ "loss": 0.0078,
+ "step": 35648
+ },
+ {
+ "epoch": 97.13623978201635,
+ "grad_norm": 1.3295245170593262,
+ "learning_rate": 4.2982158769956375e-08,
+ "loss": 0.1144,
+ "step": 35649
+ },
+ {
+ "epoch": 97.13896457765668,
+ "grad_norm": 1.092057466506958,
+ "learning_rate": 4.290046330624354e-08,
+ "loss": 0.0152,
+ "step": 35650
+ },
+ {
+ "epoch": 97.141689373297,
+ "grad_norm": 1.2517433166503906,
+ "learning_rate": 4.281884538810066e-08,
+ "loss": 0.0264,
+ "step": 35651
+ },
+ {
+ "epoch": 97.14441416893733,
+ "grad_norm": 0.560115396976471,
+ "learning_rate": 4.2737305016165024e-08,
+ "loss": 0.0045,
+ "step": 35652
+ },
+ {
+ "epoch": 97.14713896457765,
+ "grad_norm": 0.5696465373039246,
+ "learning_rate": 4.265584219107055e-08,
+ "loss": 0.0048,
+ "step": 35653
+ },
+ {
+ "epoch": 97.14986376021798,
+ "grad_norm": 0.5463761687278748,
+ "learning_rate": 4.25744569134523e-08,
+ "loss": 0.0048,
+ "step": 35654
+ },
+ {
+ "epoch": 97.15258855585832,
+ "grad_norm": 0.6900113821029663,
+ "learning_rate": 4.24931491839431e-08,
+ "loss": 0.0065,
+ "step": 35655
+ },
+ {
+ "epoch": 97.15531335149863,
+ "grad_norm": 1.488874077796936,
+ "learning_rate": 4.241191900317687e-08,
+ "loss": 0.021,
+ "step": 35656
+ },
+ {
+ "epoch": 97.15803814713897,
+ "grad_norm": 0.8127095103263855,
+ "learning_rate": 4.233076637178646e-08,
+ "loss": 0.0117,
+ "step": 35657
+ },
+ {
+ "epoch": 97.16076294277929,
+ "grad_norm": 1.19450843334198,
+ "learning_rate": 4.224969129040246e-08,
+ "loss": 0.0218,
+ "step": 35658
+ },
+ {
+ "epoch": 97.16348773841962,
+ "grad_norm": 0.8194655776023865,
+ "learning_rate": 4.2168693759658815e-08,
+ "loss": 0.0166,
+ "step": 35659
+ },
+ {
+ "epoch": 97.16621253405995,
+ "grad_norm": 1.973394751548767,
+ "learning_rate": 4.208777378018392e-08,
+ "loss": 0.0192,
+ "step": 35660
+ },
+ {
+ "epoch": 97.16893732970027,
+ "grad_norm": 0.8356120586395264,
+ "learning_rate": 4.200693135260836e-08,
+ "loss": 0.0281,
+ "step": 35661
+ },
+ {
+ "epoch": 97.1716621253406,
+ "grad_norm": 1.4111748933792114,
+ "learning_rate": 4.192616647756387e-08,
+ "loss": 0.0176,
+ "step": 35662
+ },
+ {
+ "epoch": 97.17438692098092,
+ "grad_norm": 0.7878537774085999,
+ "learning_rate": 4.184547915567661e-08,
+ "loss": 0.0046,
+ "step": 35663
+ },
+ {
+ "epoch": 97.17711171662125,
+ "grad_norm": 1.1223273277282715,
+ "learning_rate": 4.1764869387577175e-08,
+ "loss": 0.0094,
+ "step": 35664
+ },
+ {
+ "epoch": 97.17983651226157,
+ "grad_norm": 0.8749145269393921,
+ "learning_rate": 4.168433717389286e-08,
+ "loss": 0.0073,
+ "step": 35665
+ },
+ {
+ "epoch": 97.1825613079019,
+ "grad_norm": 0.7196384072303772,
+ "learning_rate": 4.1603882515248714e-08,
+ "loss": 0.01,
+ "step": 35666
+ },
+ {
+ "epoch": 97.18528610354224,
+ "grad_norm": 0.7312146425247192,
+ "learning_rate": 4.1523505412274235e-08,
+ "loss": 0.0082,
+ "step": 35667
+ },
+ {
+ "epoch": 97.18801089918256,
+ "grad_norm": 0.7384331226348877,
+ "learning_rate": 4.144320586559447e-08,
+ "loss": 0.0143,
+ "step": 35668
+ },
+ {
+ "epoch": 97.19073569482289,
+ "grad_norm": 0.7273850440979004,
+ "learning_rate": 4.1362983875833373e-08,
+ "loss": 0.0056,
+ "step": 35669
+ },
+ {
+ "epoch": 97.19346049046321,
+ "grad_norm": 1.6312512159347534,
+ "learning_rate": 4.1282839443617106e-08,
+ "loss": 0.0402,
+ "step": 35670
+ },
+ {
+ "epoch": 97.19618528610354,
+ "grad_norm": 1.3212982416152954,
+ "learning_rate": 4.1202772569570724e-08,
+ "loss": 0.0351,
+ "step": 35671
+ },
+ {
+ "epoch": 97.19891008174388,
+ "grad_norm": 0.849904477596283,
+ "learning_rate": 4.1122783254315955e-08,
+ "loss": 0.0102,
+ "step": 35672
+ },
+ {
+ "epoch": 97.2016348773842,
+ "grad_norm": 0.9240012764930725,
+ "learning_rate": 4.1042871498476746e-08,
+ "loss": 0.0069,
+ "step": 35673
+ },
+ {
+ "epoch": 97.20435967302453,
+ "grad_norm": 6.102384090423584,
+ "learning_rate": 4.09630373026737e-08,
+ "loss": 0.0549,
+ "step": 35674
+ },
+ {
+ "epoch": 97.20708446866485,
+ "grad_norm": 1.1992857456207275,
+ "learning_rate": 4.088328066753078e-08,
+ "loss": 0.0667,
+ "step": 35675
+ },
+ {
+ "epoch": 97.20980926430518,
+ "grad_norm": 0.8886632323265076,
+ "learning_rate": 4.0803601593668586e-08,
+ "loss": 0.0121,
+ "step": 35676
+ },
+ {
+ "epoch": 97.2125340599455,
+ "grad_norm": 1.2456198930740356,
+ "learning_rate": 4.072400008170774e-08,
+ "loss": 0.0242,
+ "step": 35677
+ },
+ {
+ "epoch": 97.21525885558583,
+ "grad_norm": 0.8257033824920654,
+ "learning_rate": 4.064447613226774e-08,
+ "loss": 0.0113,
+ "step": 35678
+ },
+ {
+ "epoch": 97.21798365122616,
+ "grad_norm": 0.38485434651374817,
+ "learning_rate": 4.0565029745966987e-08,
+ "loss": 0.0032,
+ "step": 35679
+ },
+ {
+ "epoch": 97.22070844686648,
+ "grad_norm": 0.8328494429588318,
+ "learning_rate": 4.0485660923426095e-08,
+ "loss": 0.0064,
+ "step": 35680
+ },
+ {
+ "epoch": 97.22343324250681,
+ "grad_norm": 0.8663792610168457,
+ "learning_rate": 4.0406369665262343e-08,
+ "loss": 0.0108,
+ "step": 35681
+ },
+ {
+ "epoch": 97.22615803814713,
+ "grad_norm": 1.0921297073364258,
+ "learning_rate": 4.032715597209302e-08,
+ "loss": 0.0136,
+ "step": 35682
+ },
+ {
+ "epoch": 97.22888283378747,
+ "grad_norm": 1.8528507947921753,
+ "learning_rate": 4.024801984453541e-08,
+ "loss": 0.0333,
+ "step": 35683
+ },
+ {
+ "epoch": 97.2316076294278,
+ "grad_norm": 0.3028218746185303,
+ "learning_rate": 4.016896128320569e-08,
+ "loss": 0.0032,
+ "step": 35684
+ },
+ {
+ "epoch": 97.23433242506812,
+ "grad_norm": 1.2419084310531616,
+ "learning_rate": 4.008998028872002e-08,
+ "loss": 0.0218,
+ "step": 35685
+ },
+ {
+ "epoch": 97.23705722070845,
+ "grad_norm": 1.376476764678955,
+ "learning_rate": 4.0011076861691254e-08,
+ "loss": 0.0285,
+ "step": 35686
+ },
+ {
+ "epoch": 97.23978201634877,
+ "grad_norm": 2.314342498779297,
+ "learning_rate": 3.993225100273779e-08,
+ "loss": 0.1085,
+ "step": 35687
+ },
+ {
+ "epoch": 97.2425068119891,
+ "grad_norm": 1.4309449195861816,
+ "learning_rate": 3.985350271247024e-08,
+ "loss": 0.0418,
+ "step": 35688
+ },
+ {
+ "epoch": 97.24523160762942,
+ "grad_norm": 2.1002848148345947,
+ "learning_rate": 3.9774831991503674e-08,
+ "loss": 0.0198,
+ "step": 35689
+ },
+ {
+ "epoch": 97.24795640326975,
+ "grad_norm": 1.1871228218078613,
+ "learning_rate": 3.9696238840448705e-08,
+ "loss": 0.0119,
+ "step": 35690
+ },
+ {
+ "epoch": 97.25068119891009,
+ "grad_norm": 1.3384345769882202,
+ "learning_rate": 3.9617723259920417e-08,
+ "loss": 0.0084,
+ "step": 35691
+ },
+ {
+ "epoch": 97.2534059945504,
+ "grad_norm": 2.1313235759735107,
+ "learning_rate": 3.95392852505283e-08,
+ "loss": 0.0375,
+ "step": 35692
+ },
+ {
+ "epoch": 97.25613079019074,
+ "grad_norm": 1.0231026411056519,
+ "learning_rate": 3.9460924812884107e-08,
+ "loss": 0.0212,
+ "step": 35693
+ },
+ {
+ "epoch": 97.25885558583106,
+ "grad_norm": 0.9126567840576172,
+ "learning_rate": 3.938264194759622e-08,
+ "loss": 0.0679,
+ "step": 35694
+ },
+ {
+ "epoch": 97.26158038147139,
+ "grad_norm": 1.2276010513305664,
+ "learning_rate": 3.9304436655276393e-08,
+ "loss": 0.0213,
+ "step": 35695
+ },
+ {
+ "epoch": 97.26430517711172,
+ "grad_norm": 2.9656124114990234,
+ "learning_rate": 3.922630893653301e-08,
+ "loss": 0.0105,
+ "step": 35696
+ },
+ {
+ "epoch": 97.26702997275204,
+ "grad_norm": 0.8621907830238342,
+ "learning_rate": 3.914825879197559e-08,
+ "loss": 0.1236,
+ "step": 35697
+ },
+ {
+ "epoch": 97.26975476839237,
+ "grad_norm": 1.2348284721374512,
+ "learning_rate": 3.9070286222210316e-08,
+ "loss": 0.0185,
+ "step": 35698
+ },
+ {
+ "epoch": 97.2724795640327,
+ "grad_norm": 1.73397696018219,
+ "learning_rate": 3.899239122784448e-08,
+ "loss": 0.0151,
+ "step": 35699
+ },
+ {
+ "epoch": 97.27520435967303,
+ "grad_norm": 1.1967445611953735,
+ "learning_rate": 3.891457380948538e-08,
+ "loss": 0.0116,
+ "step": 35700
+ },
+ {
+ "epoch": 97.27792915531334,
+ "grad_norm": 0.9062530994415283,
+ "learning_rate": 3.883683396773919e-08,
+ "loss": 0.0061,
+ "step": 35701
+ },
+ {
+ "epoch": 97.28065395095368,
+ "grad_norm": 1.5762012004852295,
+ "learning_rate": 3.875917170321208e-08,
+ "loss": 0.0832,
+ "step": 35702
+ },
+ {
+ "epoch": 97.28337874659401,
+ "grad_norm": 1.0671169757843018,
+ "learning_rate": 3.8681587016506924e-08,
+ "loss": 0.0133,
+ "step": 35703
+ },
+ {
+ "epoch": 97.28610354223433,
+ "grad_norm": 1.0753633975982666,
+ "learning_rate": 3.860407990822879e-08,
+ "loss": 0.0104,
+ "step": 35704
+ },
+ {
+ "epoch": 97.28882833787466,
+ "grad_norm": 1.1674185991287231,
+ "learning_rate": 3.8526650378982736e-08,
+ "loss": 0.0209,
+ "step": 35705
+ },
+ {
+ "epoch": 97.29155313351498,
+ "grad_norm": 1.7842401266098022,
+ "learning_rate": 3.8449298429369397e-08,
+ "loss": 0.0214,
+ "step": 35706
+ },
+ {
+ "epoch": 97.29427792915531,
+ "grad_norm": 0.5116434097290039,
+ "learning_rate": 3.8372024059991633e-08,
+ "loss": 0.0048,
+ "step": 35707
+ },
+ {
+ "epoch": 97.29700272479565,
+ "grad_norm": 0.8006274104118347,
+ "learning_rate": 3.8294827271453394e-08,
+ "loss": 0.0117,
+ "step": 35708
+ },
+ {
+ "epoch": 97.29972752043597,
+ "grad_norm": 1.4666790962219238,
+ "learning_rate": 3.8217708064354206e-08,
+ "loss": 0.0277,
+ "step": 35709
+ },
+ {
+ "epoch": 97.3024523160763,
+ "grad_norm": 1.0368918180465698,
+ "learning_rate": 3.81406664392936e-08,
+ "loss": 0.0153,
+ "step": 35710
+ },
+ {
+ "epoch": 97.30517711171662,
+ "grad_norm": 0.9145317077636719,
+ "learning_rate": 3.80637023968744e-08,
+ "loss": 0.0058,
+ "step": 35711
+ },
+ {
+ "epoch": 97.30790190735695,
+ "grad_norm": 1.6203583478927612,
+ "learning_rate": 3.798681593769393e-08,
+ "loss": 0.0412,
+ "step": 35712
+ },
+ {
+ "epoch": 97.31062670299727,
+ "grad_norm": 0.7598527073860168,
+ "learning_rate": 3.791000706235171e-08,
+ "loss": 0.0154,
+ "step": 35713
+ },
+ {
+ "epoch": 97.3133514986376,
+ "grad_norm": 0.8451588749885559,
+ "learning_rate": 3.783327577144502e-08,
+ "loss": 0.0098,
+ "step": 35714
+ },
+ {
+ "epoch": 97.31607629427793,
+ "grad_norm": 1.4935237169265747,
+ "learning_rate": 3.775662206557229e-08,
+ "loss": 0.0142,
+ "step": 35715
+ },
+ {
+ "epoch": 97.31880108991825,
+ "grad_norm": 0.6524025201797485,
+ "learning_rate": 3.768004594533081e-08,
+ "loss": 0.0059,
+ "step": 35716
+ },
+ {
+ "epoch": 97.32152588555859,
+ "grad_norm": 0.7521406412124634,
+ "learning_rate": 3.760354741131678e-08,
+ "loss": 0.0068,
+ "step": 35717
+ },
+ {
+ "epoch": 97.3242506811989,
+ "grad_norm": 1.8233006000518799,
+ "learning_rate": 3.752712646412526e-08,
+ "loss": 0.0795,
+ "step": 35718
+ },
+ {
+ "epoch": 97.32697547683924,
+ "grad_norm": 1.155234694480896,
+ "learning_rate": 3.745078310435135e-08,
+ "loss": 0.0215,
+ "step": 35719
+ },
+ {
+ "epoch": 97.32970027247957,
+ "grad_norm": 1.0763806104660034,
+ "learning_rate": 3.737451733259123e-08,
+ "loss": 0.0184,
+ "step": 35720
+ },
+ {
+ "epoch": 97.33242506811989,
+ "grad_norm": 1.5135457515716553,
+ "learning_rate": 3.729832914943665e-08,
+ "loss": 0.0476,
+ "step": 35721
+ },
+ {
+ "epoch": 97.33514986376022,
+ "grad_norm": 0.7992256283760071,
+ "learning_rate": 3.722221855548269e-08,
+ "loss": 0.0056,
+ "step": 35722
+ },
+ {
+ "epoch": 97.33787465940054,
+ "grad_norm": 1.168269157409668,
+ "learning_rate": 3.71461855513211e-08,
+ "loss": 0.0539,
+ "step": 35723
+ },
+ {
+ "epoch": 97.34059945504087,
+ "grad_norm": 0.5168452262878418,
+ "learning_rate": 3.707023013754474e-08,
+ "loss": 0.0046,
+ "step": 35724
+ },
+ {
+ "epoch": 97.34332425068119,
+ "grad_norm": 1.1928743124008179,
+ "learning_rate": 3.699435231474424e-08,
+ "loss": 0.0073,
+ "step": 35725
+ },
+ {
+ "epoch": 97.34604904632153,
+ "grad_norm": 1.388916254043579,
+ "learning_rate": 3.691855208351136e-08,
+ "loss": 0.0625,
+ "step": 35726
+ },
+ {
+ "epoch": 97.34877384196186,
+ "grad_norm": 0.7200934886932373,
+ "learning_rate": 3.6842829444436734e-08,
+ "loss": 0.0048,
+ "step": 35727
+ },
+ {
+ "epoch": 97.35149863760218,
+ "grad_norm": 1.0587372779846191,
+ "learning_rate": 3.676718439810767e-08,
+ "loss": 0.0134,
+ "step": 35728
+ },
+ {
+ "epoch": 97.35422343324251,
+ "grad_norm": 2.421874761581421,
+ "learning_rate": 3.6691616945117025e-08,
+ "loss": 0.0202,
+ "step": 35729
+ },
+ {
+ "epoch": 97.35694822888283,
+ "grad_norm": 0.547734260559082,
+ "learning_rate": 3.6616127086051e-08,
+ "loss": 0.0066,
+ "step": 35730
+ },
+ {
+ "epoch": 97.35967302452316,
+ "grad_norm": 1.1388540267944336,
+ "learning_rate": 3.65407148214969e-08,
+ "loss": 0.0121,
+ "step": 35731
+ },
+ {
+ "epoch": 97.3623978201635,
+ "grad_norm": 1.1520246267318726,
+ "learning_rate": 3.646538015204426e-08,
+ "loss": 0.0073,
+ "step": 35732
+ },
+ {
+ "epoch": 97.36512261580381,
+ "grad_norm": 1.473513126373291,
+ "learning_rate": 3.639012307827816e-08,
+ "loss": 0.0465,
+ "step": 35733
+ },
+ {
+ "epoch": 97.36784741144415,
+ "grad_norm": 1.0183950662612915,
+ "learning_rate": 3.63149436007848e-08,
+ "loss": 0.0543,
+ "step": 35734
+ },
+ {
+ "epoch": 97.37057220708446,
+ "grad_norm": 0.9344877004623413,
+ "learning_rate": 3.623984172015038e-08,
+ "loss": 0.0358,
+ "step": 35735
+ },
+ {
+ "epoch": 97.3732970027248,
+ "grad_norm": 0.8537148833274841,
+ "learning_rate": 3.616481743695888e-08,
+ "loss": 0.0088,
+ "step": 35736
+ },
+ {
+ "epoch": 97.37602179836512,
+ "grad_norm": 0.699112057685852,
+ "learning_rate": 3.6089870751795374e-08,
+ "loss": 0.0062,
+ "step": 35737
+ },
+ {
+ "epoch": 97.37874659400545,
+ "grad_norm": 1.6251168251037598,
+ "learning_rate": 3.6015001665243853e-08,
+ "loss": 0.0523,
+ "step": 35738
+ },
+ {
+ "epoch": 97.38147138964578,
+ "grad_norm": 0.550065279006958,
+ "learning_rate": 3.594021017788607e-08,
+ "loss": 0.0049,
+ "step": 35739
+ },
+ {
+ "epoch": 97.3841961852861,
+ "grad_norm": 1.3086715936660767,
+ "learning_rate": 3.586549629030489e-08,
+ "loss": 0.0557,
+ "step": 35740
+ },
+ {
+ "epoch": 97.38692098092643,
+ "grad_norm": 0.8626537322998047,
+ "learning_rate": 3.579086000308318e-08,
+ "loss": 0.0096,
+ "step": 35741
+ },
+ {
+ "epoch": 97.38964577656675,
+ "grad_norm": 1.2443716526031494,
+ "learning_rate": 3.571630131680159e-08,
+ "loss": 0.0208,
+ "step": 35742
+ },
+ {
+ "epoch": 97.39237057220708,
+ "grad_norm": 0.6437604427337646,
+ "learning_rate": 3.564182023204077e-08,
+ "loss": 0.0098,
+ "step": 35743
+ },
+ {
+ "epoch": 97.39509536784742,
+ "grad_norm": 1.3036613464355469,
+ "learning_rate": 3.556741674938024e-08,
+ "loss": 0.0281,
+ "step": 35744
+ },
+ {
+ "epoch": 97.39782016348774,
+ "grad_norm": 1.0074255466461182,
+ "learning_rate": 3.549309086940067e-08,
+ "loss": 0.0087,
+ "step": 35745
+ },
+ {
+ "epoch": 97.40054495912807,
+ "grad_norm": 0.7562723755836487,
+ "learning_rate": 3.541884259267936e-08,
+ "loss": 0.0053,
+ "step": 35746
+ },
+ {
+ "epoch": 97.40326975476839,
+ "grad_norm": 1.2716466188430786,
+ "learning_rate": 3.5344671919795844e-08,
+ "loss": 0.0239,
+ "step": 35747
+ },
+ {
+ "epoch": 97.40599455040872,
+ "grad_norm": 0.5196647047996521,
+ "learning_rate": 3.5270578851327454e-08,
+ "loss": 0.0058,
+ "step": 35748
+ },
+ {
+ "epoch": 97.40871934604904,
+ "grad_norm": 1.0398705005645752,
+ "learning_rate": 3.519656338785038e-08,
+ "loss": 0.0173,
+ "step": 35749
+ },
+ {
+ "epoch": 97.41144414168937,
+ "grad_norm": 1.6451892852783203,
+ "learning_rate": 3.512262552994305e-08,
+ "loss": 0.0101,
+ "step": 35750
+ },
+ {
+ "epoch": 97.4141689373297,
+ "grad_norm": 0.5752031803131104,
+ "learning_rate": 3.504876527817946e-08,
+ "loss": 0.0069,
+ "step": 35751
+ },
+ {
+ "epoch": 97.41689373297002,
+ "grad_norm": 3.1299257278442383,
+ "learning_rate": 3.49749826331347e-08,
+ "loss": 0.0473,
+ "step": 35752
+ },
+ {
+ "epoch": 97.41961852861036,
+ "grad_norm": 1.094840168952942,
+ "learning_rate": 3.490127759538497e-08,
+ "loss": 0.0221,
+ "step": 35753
+ },
+ {
+ "epoch": 97.42234332425068,
+ "grad_norm": 1.3069614171981812,
+ "learning_rate": 3.482765016550316e-08,
+ "loss": 0.0343,
+ "step": 35754
+ },
+ {
+ "epoch": 97.42506811989101,
+ "grad_norm": 1.02880859375,
+ "learning_rate": 3.475410034406212e-08,
+ "loss": 0.0067,
+ "step": 35755
+ },
+ {
+ "epoch": 97.42779291553134,
+ "grad_norm": 0.4646350145339966,
+ "learning_rate": 3.468062813163586e-08,
+ "loss": 0.0034,
+ "step": 35756
+ },
+ {
+ "epoch": 97.43051771117166,
+ "grad_norm": 0.5119317173957825,
+ "learning_rate": 3.4607233528796136e-08,
+ "loss": 0.004,
+ "step": 35757
+ },
+ {
+ "epoch": 97.433242506812,
+ "grad_norm": 1.2723958492279053,
+ "learning_rate": 3.453391653611471e-08,
+ "loss": 0.0787,
+ "step": 35758
+ },
+ {
+ "epoch": 97.43596730245231,
+ "grad_norm": 0.9807150959968567,
+ "learning_rate": 3.446067715416224e-08,
+ "loss": 0.0143,
+ "step": 35759
+ },
+ {
+ "epoch": 97.43869209809264,
+ "grad_norm": 0.5286177396774292,
+ "learning_rate": 3.438751538350937e-08,
+ "loss": 0.0062,
+ "step": 35760
+ },
+ {
+ "epoch": 97.44141689373296,
+ "grad_norm": 1.8826327323913574,
+ "learning_rate": 3.431443122472566e-08,
+ "loss": 0.0208,
+ "step": 35761
+ },
+ {
+ "epoch": 97.4441416893733,
+ "grad_norm": 1.2199695110321045,
+ "learning_rate": 3.424142467838065e-08,
+ "loss": 0.0802,
+ "step": 35762
+ },
+ {
+ "epoch": 97.44686648501363,
+ "grad_norm": 0.8878253698348999,
+ "learning_rate": 3.4168495745042776e-08,
+ "loss": 0.0138,
+ "step": 35763
+ },
+ {
+ "epoch": 97.44959128065395,
+ "grad_norm": 1.7570720911026,
+ "learning_rate": 3.409564442527935e-08,
+ "loss": 0.0543,
+ "step": 35764
+ },
+ {
+ "epoch": 97.45231607629428,
+ "grad_norm": 0.6195181608200073,
+ "learning_rate": 3.402287071965882e-08,
+ "loss": 0.0077,
+ "step": 35765
+ },
+ {
+ "epoch": 97.4550408719346,
+ "grad_norm": 0.99741530418396,
+ "learning_rate": 3.395017462874739e-08,
+ "loss": 0.015,
+ "step": 35766
+ },
+ {
+ "epoch": 97.45776566757493,
+ "grad_norm": 0.8652946352958679,
+ "learning_rate": 3.387755615311128e-08,
+ "loss": 0.0084,
+ "step": 35767
+ },
+ {
+ "epoch": 97.46049046321527,
+ "grad_norm": 0.8596867322921753,
+ "learning_rate": 3.380501529331559e-08,
+ "loss": 0.0108,
+ "step": 35768
+ },
+ {
+ "epoch": 97.46321525885558,
+ "grad_norm": 1.9369438886642456,
+ "learning_rate": 3.3732552049925423e-08,
+ "loss": 0.1313,
+ "step": 35769
+ },
+ {
+ "epoch": 97.46594005449592,
+ "grad_norm": 0.6045882701873779,
+ "learning_rate": 3.366016642350589e-08,
+ "loss": 0.0063,
+ "step": 35770
+ },
+ {
+ "epoch": 97.46866485013624,
+ "grad_norm": 1.5127960443496704,
+ "learning_rate": 3.358785841462098e-08,
+ "loss": 0.057,
+ "step": 35771
+ },
+ {
+ "epoch": 97.47138964577657,
+ "grad_norm": 0.5709162950515747,
+ "learning_rate": 3.3515628023831346e-08,
+ "loss": 0.0074,
+ "step": 35772
+ },
+ {
+ "epoch": 97.47411444141689,
+ "grad_norm": 1.083299994468689,
+ "learning_rate": 3.344347525170322e-08,
+ "loss": 0.1157,
+ "step": 35773
+ },
+ {
+ "epoch": 97.47683923705722,
+ "grad_norm": 0.4636647403240204,
+ "learning_rate": 3.337140009879503e-08,
+ "loss": 0.0036,
+ "step": 35774
+ },
+ {
+ "epoch": 97.47956403269755,
+ "grad_norm": 0.8109275102615356,
+ "learning_rate": 3.329940256566966e-08,
+ "loss": 0.007,
+ "step": 35775
+ },
+ {
+ "epoch": 97.48228882833787,
+ "grad_norm": 0.7956485748291016,
+ "learning_rate": 3.3227482652888885e-08,
+ "loss": 0.0132,
+ "step": 35776
+ },
+ {
+ "epoch": 97.4850136239782,
+ "grad_norm": 1.0223912000656128,
+ "learning_rate": 3.315564036101115e-08,
+ "loss": 0.0104,
+ "step": 35777
+ },
+ {
+ "epoch": 97.48773841961852,
+ "grad_norm": 1.2079901695251465,
+ "learning_rate": 3.308387569059601e-08,
+ "loss": 0.0812,
+ "step": 35778
+ },
+ {
+ "epoch": 97.49046321525886,
+ "grad_norm": 1.2122223377227783,
+ "learning_rate": 3.3012188642203015e-08,
+ "loss": 0.008,
+ "step": 35779
+ },
+ {
+ "epoch": 97.49318801089919,
+ "grad_norm": 1.3515764474868774,
+ "learning_rate": 3.29405792163906e-08,
+ "loss": 0.0274,
+ "step": 35780
+ },
+ {
+ "epoch": 97.49591280653951,
+ "grad_norm": 1.0832830667495728,
+ "learning_rate": 3.2869047413716104e-08,
+ "loss": 0.0632,
+ "step": 35781
+ },
+ {
+ "epoch": 97.49863760217984,
+ "grad_norm": 0.7450041770935059,
+ "learning_rate": 3.2797593234735745e-08,
+ "loss": 0.007,
+ "step": 35782
+ },
+ {
+ "epoch": 97.50136239782016,
+ "grad_norm": 0.9359983205795288,
+ "learning_rate": 3.272621668000797e-08,
+ "loss": 0.014,
+ "step": 35783
+ },
+ {
+ "epoch": 97.50408719346049,
+ "grad_norm": 1.0504759550094604,
+ "learning_rate": 3.265491775008789e-08,
+ "loss": 0.0464,
+ "step": 35784
+ },
+ {
+ "epoch": 97.50681198910081,
+ "grad_norm": 1.2818557024002075,
+ "learning_rate": 3.25836964455295e-08,
+ "loss": 0.1311,
+ "step": 35785
+ },
+ {
+ "epoch": 97.50953678474114,
+ "grad_norm": 0.3302555978298187,
+ "learning_rate": 3.251255276688903e-08,
+ "loss": 0.0048,
+ "step": 35786
+ },
+ {
+ "epoch": 97.51226158038148,
+ "grad_norm": 1.3982675075531006,
+ "learning_rate": 3.2441486714720474e-08,
+ "loss": 0.1441,
+ "step": 35787
+ },
+ {
+ "epoch": 97.5149863760218,
+ "grad_norm": 0.8035872578620911,
+ "learning_rate": 3.2370498289576727e-08,
+ "loss": 0.0082,
+ "step": 35788
+ },
+ {
+ "epoch": 97.51771117166213,
+ "grad_norm": 0.8281283974647522,
+ "learning_rate": 3.2299587492009565e-08,
+ "loss": 0.0073,
+ "step": 35789
+ },
+ {
+ "epoch": 97.52043596730245,
+ "grad_norm": 1.7936248779296875,
+ "learning_rate": 3.2228754322574106e-08,
+ "loss": 0.0604,
+ "step": 35790
+ },
+ {
+ "epoch": 97.52316076294278,
+ "grad_norm": 0.7171409130096436,
+ "learning_rate": 3.21579987818188e-08,
+ "loss": 0.004,
+ "step": 35791
+ },
+ {
+ "epoch": 97.52588555858311,
+ "grad_norm": 1.7810548543930054,
+ "learning_rate": 3.208732087029653e-08,
+ "loss": 0.0176,
+ "step": 35792
+ },
+ {
+ "epoch": 97.52861035422343,
+ "grad_norm": 1.3396354913711548,
+ "learning_rate": 3.201672058855798e-08,
+ "loss": 0.0526,
+ "step": 35793
+ },
+ {
+ "epoch": 97.53133514986376,
+ "grad_norm": 0.5930657982826233,
+ "learning_rate": 3.19461979371527e-08,
+ "loss": 0.0092,
+ "step": 35794
+ },
+ {
+ "epoch": 97.53405994550408,
+ "grad_norm": 0.981349527835846,
+ "learning_rate": 3.1875752916628034e-08,
+ "loss": 0.0133,
+ "step": 35795
+ },
+ {
+ "epoch": 97.53678474114442,
+ "grad_norm": 1.6805877685546875,
+ "learning_rate": 3.180538552753576e-08,
+ "loss": 0.0968,
+ "step": 35796
+ },
+ {
+ "epoch": 97.53950953678473,
+ "grad_norm": 0.5359209775924683,
+ "learning_rate": 3.1735095770421e-08,
+ "loss": 0.0032,
+ "step": 35797
+ },
+ {
+ "epoch": 97.54223433242507,
+ "grad_norm": 0.9081090092658997,
+ "learning_rate": 3.166488364583331e-08,
+ "loss": 0.0151,
+ "step": 35798
+ },
+ {
+ "epoch": 97.5449591280654,
+ "grad_norm": 0.4267675280570984,
+ "learning_rate": 3.159474915431893e-08,
+ "loss": 0.005,
+ "step": 35799
+ },
+ {
+ "epoch": 97.54768392370572,
+ "grad_norm": 0.8493682146072388,
+ "learning_rate": 3.1524692296422966e-08,
+ "loss": 0.0425,
+ "step": 35800
+ },
+ {
+ "epoch": 97.55040871934605,
+ "grad_norm": 0.958163321018219,
+ "learning_rate": 3.145471307269277e-08,
+ "loss": 0.09,
+ "step": 35801
+ },
+ {
+ "epoch": 97.55313351498637,
+ "grad_norm": 0.9516919255256653,
+ "learning_rate": 3.138481148367123e-08,
+ "loss": 0.0192,
+ "step": 35802
+ },
+ {
+ "epoch": 97.5558583106267,
+ "grad_norm": 1.0044877529144287,
+ "learning_rate": 3.1314987529905696e-08,
+ "loss": 0.0077,
+ "step": 35803
+ },
+ {
+ "epoch": 97.55858310626704,
+ "grad_norm": 1.507880449295044,
+ "learning_rate": 3.124524121193684e-08,
+ "loss": 0.1083,
+ "step": 35804
+ },
+ {
+ "epoch": 97.56130790190736,
+ "grad_norm": 1.286005973815918,
+ "learning_rate": 3.117557253030978e-08,
+ "loss": 0.1956,
+ "step": 35805
+ },
+ {
+ "epoch": 97.56403269754769,
+ "grad_norm": 0.9606514573097229,
+ "learning_rate": 3.1105981485566314e-08,
+ "loss": 0.0149,
+ "step": 35806
+ },
+ {
+ "epoch": 97.566757493188,
+ "grad_norm": 1.4330432415008545,
+ "learning_rate": 3.1036468078248226e-08,
+ "loss": 0.0339,
+ "step": 35807
+ },
+ {
+ "epoch": 97.56948228882834,
+ "grad_norm": 0.5926778316497803,
+ "learning_rate": 3.0967032308897304e-08,
+ "loss": 0.0036,
+ "step": 35808
+ },
+ {
+ "epoch": 97.57220708446866,
+ "grad_norm": 0.7168121337890625,
+ "learning_rate": 3.089767417805534e-08,
+ "loss": 0.0061,
+ "step": 35809
+ },
+ {
+ "epoch": 97.57493188010899,
+ "grad_norm": 1.078332781791687,
+ "learning_rate": 3.082839368626078e-08,
+ "loss": 0.1213,
+ "step": 35810
+ },
+ {
+ "epoch": 97.57765667574932,
+ "grad_norm": 0.9653197526931763,
+ "learning_rate": 3.075919083405321e-08,
+ "loss": 0.0151,
+ "step": 35811
+ },
+ {
+ "epoch": 97.58038147138964,
+ "grad_norm": 0.950004518032074,
+ "learning_rate": 3.06900656219733e-08,
+ "loss": 0.0181,
+ "step": 35812
+ },
+ {
+ "epoch": 97.58310626702998,
+ "grad_norm": 1.0147298574447632,
+ "learning_rate": 3.0621018050557285e-08,
+ "loss": 0.0116,
+ "step": 35813
+ },
+ {
+ "epoch": 97.5858310626703,
+ "grad_norm": 0.8502739071846008,
+ "learning_rate": 3.0552048120343625e-08,
+ "loss": 0.0095,
+ "step": 35814
+ },
+ {
+ "epoch": 97.58855585831063,
+ "grad_norm": 1.2039647102355957,
+ "learning_rate": 3.048315583187078e-08,
+ "loss": 0.0433,
+ "step": 35815
+ },
+ {
+ "epoch": 97.59128065395096,
+ "grad_norm": 1.2399312257766724,
+ "learning_rate": 3.041434118567388e-08,
+ "loss": 0.0098,
+ "step": 35816
+ },
+ {
+ "epoch": 97.59400544959128,
+ "grad_norm": 0.5439571738243103,
+ "learning_rate": 3.034560418228916e-08,
+ "loss": 0.0042,
+ "step": 35817
+ },
+ {
+ "epoch": 97.59673024523161,
+ "grad_norm": 1.2876406908035278,
+ "learning_rate": 3.027694482225285e-08,
+ "loss": 0.0332,
+ "step": 35818
+ },
+ {
+ "epoch": 97.59945504087193,
+ "grad_norm": 1.1082900762557983,
+ "learning_rate": 3.020836310609787e-08,
+ "loss": 0.0118,
+ "step": 35819
+ },
+ {
+ "epoch": 97.60217983651226,
+ "grad_norm": 0.6044007539749146,
+ "learning_rate": 3.0139859034359344e-08,
+ "loss": 0.0045,
+ "step": 35820
+ },
+ {
+ "epoch": 97.60490463215258,
+ "grad_norm": 0.8190687298774719,
+ "learning_rate": 3.0071432607571284e-08,
+ "loss": 0.0105,
+ "step": 35821
+ },
+ {
+ "epoch": 97.60762942779292,
+ "grad_norm": 1.3098201751708984,
+ "learning_rate": 3.000308382626549e-08,
+ "loss": 0.0082,
+ "step": 35822
+ },
+ {
+ "epoch": 97.61035422343325,
+ "grad_norm": 1.04621160030365,
+ "learning_rate": 2.9934812690974865e-08,
+ "loss": 0.025,
+ "step": 35823
+ },
+ {
+ "epoch": 97.61307901907357,
+ "grad_norm": 1.0620018243789673,
+ "learning_rate": 2.986661920223122e-08,
+ "loss": 0.0179,
+ "step": 35824
+ },
+ {
+ "epoch": 97.6158038147139,
+ "grad_norm": 1.0106998682022095,
+ "learning_rate": 2.9798503360565225e-08,
+ "loss": 0.0362,
+ "step": 35825
+ },
+ {
+ "epoch": 97.61852861035422,
+ "grad_norm": 1.2642900943756104,
+ "learning_rate": 2.9730465166508683e-08,
+ "loss": 0.0688,
+ "step": 35826
+ },
+ {
+ "epoch": 97.62125340599455,
+ "grad_norm": 1.1032878160476685,
+ "learning_rate": 2.966250462058895e-08,
+ "loss": 0.0138,
+ "step": 35827
+ },
+ {
+ "epoch": 97.62397820163488,
+ "grad_norm": 1.7931972742080688,
+ "learning_rate": 2.9594621723336715e-08,
+ "loss": 0.0278,
+ "step": 35828
+ },
+ {
+ "epoch": 97.6267029972752,
+ "grad_norm": 0.931978166103363,
+ "learning_rate": 2.952681647528155e-08,
+ "loss": 0.01,
+ "step": 35829
+ },
+ {
+ "epoch": 97.62942779291554,
+ "grad_norm": 1.2940152883529663,
+ "learning_rate": 2.9459088876950814e-08,
+ "loss": 0.014,
+ "step": 35830
+ },
+ {
+ "epoch": 97.63215258855585,
+ "grad_norm": 0.6399456858634949,
+ "learning_rate": 2.939143892887075e-08,
+ "loss": 0.0067,
+ "step": 35831
+ },
+ {
+ "epoch": 97.63487738419619,
+ "grad_norm": 1.918784499168396,
+ "learning_rate": 2.932386663156983e-08,
+ "loss": 0.0077,
+ "step": 35832
+ },
+ {
+ "epoch": 97.6376021798365,
+ "grad_norm": 1.3241512775421143,
+ "learning_rate": 2.9256371985573183e-08,
+ "loss": 0.0153,
+ "step": 35833
+ },
+ {
+ "epoch": 97.64032697547684,
+ "grad_norm": 0.7549949884414673,
+ "learning_rate": 2.918895499140817e-08,
+ "loss": 0.0067,
+ "step": 35834
+ },
+ {
+ "epoch": 97.64305177111717,
+ "grad_norm": 1.8538486957550049,
+ "learning_rate": 2.91216156495977e-08,
+ "loss": 0.033,
+ "step": 35835
+ },
+ {
+ "epoch": 97.64577656675749,
+ "grad_norm": 0.8119707107543945,
+ "learning_rate": 2.9054353960666914e-08,
+ "loss": 0.0078,
+ "step": 35836
+ },
+ {
+ "epoch": 97.64850136239782,
+ "grad_norm": 1.213109016418457,
+ "learning_rate": 2.8987169925140947e-08,
+ "loss": 0.0102,
+ "step": 35837
+ },
+ {
+ "epoch": 97.65122615803814,
+ "grad_norm": 0.9270329475402832,
+ "learning_rate": 2.89200635435416e-08,
+ "loss": 0.0285,
+ "step": 35838
+ },
+ {
+ "epoch": 97.65395095367847,
+ "grad_norm": 1.0564634799957275,
+ "learning_rate": 2.885303481639179e-08,
+ "loss": 0.0085,
+ "step": 35839
+ },
+ {
+ "epoch": 97.65667574931881,
+ "grad_norm": 0.8375817537307739,
+ "learning_rate": 2.8786083744213322e-08,
+ "loss": 0.0062,
+ "step": 35840
+ },
+ {
+ "epoch": 97.65940054495913,
+ "grad_norm": 0.9991700053215027,
+ "learning_rate": 2.8719210327528e-08,
+ "loss": 0.0432,
+ "step": 35841
+ },
+ {
+ "epoch": 97.66212534059946,
+ "grad_norm": 0.504315972328186,
+ "learning_rate": 2.8652414566857633e-08,
+ "loss": 0.0061,
+ "step": 35842
+ },
+ {
+ "epoch": 97.66485013623978,
+ "grad_norm": 0.7569475173950195,
+ "learning_rate": 2.858569646272069e-08,
+ "loss": 0.0089,
+ "step": 35843
+ },
+ {
+ "epoch": 97.66757493188011,
+ "grad_norm": 1.5563697814941406,
+ "learning_rate": 2.8519056015636758e-08,
+ "loss": 0.1167,
+ "step": 35844
+ },
+ {
+ "epoch": 97.67029972752043,
+ "grad_norm": 1.245442509651184,
+ "learning_rate": 2.845249322612542e-08,
+ "loss": 0.0115,
+ "step": 35845
+ },
+ {
+ "epoch": 97.67302452316076,
+ "grad_norm": 1.0664957761764526,
+ "learning_rate": 2.8386008094706264e-08,
+ "loss": 0.0121,
+ "step": 35846
+ },
+ {
+ "epoch": 97.6757493188011,
+ "grad_norm": 1.0964927673339844,
+ "learning_rate": 2.8319600621895537e-08,
+ "loss": 0.0069,
+ "step": 35847
+ },
+ {
+ "epoch": 97.67847411444141,
+ "grad_norm": 1.3458775281906128,
+ "learning_rate": 2.8253270808209498e-08,
+ "loss": 0.0272,
+ "step": 35848
+ },
+ {
+ "epoch": 97.68119891008175,
+ "grad_norm": 1.0376285314559937,
+ "learning_rate": 2.818701865416773e-08,
+ "loss": 0.0055,
+ "step": 35849
+ },
+ {
+ "epoch": 97.68392370572207,
+ "grad_norm": 0.7763722538948059,
+ "learning_rate": 2.8120844160283155e-08,
+ "loss": 0.0075,
+ "step": 35850
+ },
+ {
+ "epoch": 97.6866485013624,
+ "grad_norm": 1.1301106214523315,
+ "learning_rate": 2.805474732707425e-08,
+ "loss": 0.0284,
+ "step": 35851
+ },
+ {
+ "epoch": 97.68937329700273,
+ "grad_norm": 1.1538116931915283,
+ "learning_rate": 2.7988728155052825e-08,
+ "loss": 0.0078,
+ "step": 35852
+ },
+ {
+ "epoch": 97.69209809264305,
+ "grad_norm": 1.0972274541854858,
+ "learning_rate": 2.7922786644734023e-08,
+ "loss": 0.0157,
+ "step": 35853
+ },
+ {
+ "epoch": 97.69482288828338,
+ "grad_norm": 0.7736707329750061,
+ "learning_rate": 2.785692279663188e-08,
+ "loss": 0.01,
+ "step": 35854
+ },
+ {
+ "epoch": 97.6975476839237,
+ "grad_norm": 0.8762786388397217,
+ "learning_rate": 2.779113661125932e-08,
+ "loss": 0.0094,
+ "step": 35855
+ },
+ {
+ "epoch": 97.70027247956403,
+ "grad_norm": 0.930557370185852,
+ "learning_rate": 2.7725428089128148e-08,
+ "loss": 0.0202,
+ "step": 35856
+ },
+ {
+ "epoch": 97.70299727520435,
+ "grad_norm": 0.9098291397094727,
+ "learning_rate": 2.7659797230751295e-08,
+ "loss": 0.0774,
+ "step": 35857
+ },
+ {
+ "epoch": 97.70572207084469,
+ "grad_norm": 1.4422376155853271,
+ "learning_rate": 2.759424403663835e-08,
+ "loss": 0.083,
+ "step": 35858
+ },
+ {
+ "epoch": 97.70844686648502,
+ "grad_norm": 1.769151210784912,
+ "learning_rate": 2.752876850730002e-08,
+ "loss": 0.018,
+ "step": 35859
+ },
+ {
+ "epoch": 97.71117166212534,
+ "grad_norm": 1.3018429279327393,
+ "learning_rate": 2.7463370643247e-08,
+ "loss": 0.0369,
+ "step": 35860
+ },
+ {
+ "epoch": 97.71389645776567,
+ "grad_norm": 0.9258415699005127,
+ "learning_rate": 2.7398050444987777e-08,
+ "loss": 0.0135,
+ "step": 35861
+ },
+ {
+ "epoch": 97.71662125340599,
+ "grad_norm": 0.8797904849052429,
+ "learning_rate": 2.7332807913033055e-08,
+ "loss": 0.0962,
+ "step": 35862
+ },
+ {
+ "epoch": 97.71934604904632,
+ "grad_norm": 1.1532164812088013,
+ "learning_rate": 2.7267643047887982e-08,
+ "loss": 0.0144,
+ "step": 35863
+ },
+ {
+ "epoch": 97.72207084468666,
+ "grad_norm": 0.7577705383300781,
+ "learning_rate": 2.720255585006215e-08,
+ "loss": 0.0123,
+ "step": 35864
+ },
+ {
+ "epoch": 97.72479564032697,
+ "grad_norm": 1.1345601081848145,
+ "learning_rate": 2.7137546320061824e-08,
+ "loss": 0.02,
+ "step": 35865
+ },
+ {
+ "epoch": 97.7275204359673,
+ "grad_norm": 0.850476861000061,
+ "learning_rate": 2.7072614458393263e-08,
+ "loss": 0.0087,
+ "step": 35866
+ },
+ {
+ "epoch": 97.73024523160763,
+ "grad_norm": 0.7155185341835022,
+ "learning_rate": 2.7007760265561623e-08,
+ "loss": 0.0065,
+ "step": 35867
+ },
+ {
+ "epoch": 97.73297002724796,
+ "grad_norm": 1.0401415824890137,
+ "learning_rate": 2.694298374207316e-08,
+ "loss": 0.0079,
+ "step": 35868
+ },
+ {
+ "epoch": 97.73569482288828,
+ "grad_norm": 1.3925045728683472,
+ "learning_rate": 2.6878284888431916e-08,
+ "loss": 0.0221,
+ "step": 35869
+ },
+ {
+ "epoch": 97.73841961852861,
+ "grad_norm": 1.0106996297836304,
+ "learning_rate": 2.6813663705141935e-08,
+ "loss": 0.0428,
+ "step": 35870
+ },
+ {
+ "epoch": 97.74114441416894,
+ "grad_norm": 1.4314485788345337,
+ "learning_rate": 2.6749120192705037e-08,
+ "loss": 0.0668,
+ "step": 35871
+ },
+ {
+ "epoch": 97.74386920980926,
+ "grad_norm": 1.1417917013168335,
+ "learning_rate": 2.668465435162637e-08,
+ "loss": 0.0364,
+ "step": 35872
+ },
+ {
+ "epoch": 97.7465940054496,
+ "grad_norm": 1.5113611221313477,
+ "learning_rate": 2.6620266182405542e-08,
+ "loss": 0.1279,
+ "step": 35873
+ },
+ {
+ "epoch": 97.74931880108991,
+ "grad_norm": 0.6438184976577759,
+ "learning_rate": 2.6555955685545476e-08,
+ "loss": 0.0062,
+ "step": 35874
+ },
+ {
+ "epoch": 97.75204359673025,
+ "grad_norm": 1.351394534111023,
+ "learning_rate": 2.6491722861546887e-08,
+ "loss": 0.0164,
+ "step": 35875
+ },
+ {
+ "epoch": 97.75476839237058,
+ "grad_norm": 0.5696614980697632,
+ "learning_rate": 2.6427567710909374e-08,
+ "loss": 0.0054,
+ "step": 35876
+ },
+ {
+ "epoch": 97.7574931880109,
+ "grad_norm": 1.1289310455322266,
+ "learning_rate": 2.636349023413254e-08,
+ "loss": 0.052,
+ "step": 35877
+ },
+ {
+ "epoch": 97.76021798365123,
+ "grad_norm": 0.9116040468215942,
+ "learning_rate": 2.629949043171709e-08,
+ "loss": 0.0099,
+ "step": 35878
+ },
+ {
+ "epoch": 97.76294277929155,
+ "grad_norm": 1.3081592321395874,
+ "learning_rate": 2.62355683041593e-08,
+ "loss": 0.0103,
+ "step": 35879
+ },
+ {
+ "epoch": 97.76566757493188,
+ "grad_norm": 0.9398660659790039,
+ "learning_rate": 2.6171723851957652e-08,
+ "loss": 0.0106,
+ "step": 35880
+ },
+ {
+ "epoch": 97.7683923705722,
+ "grad_norm": 1.0888280868530273,
+ "learning_rate": 2.6107957075609537e-08,
+ "loss": 0.0073,
+ "step": 35881
+ },
+ {
+ "epoch": 97.77111716621253,
+ "grad_norm": 0.4549747705459595,
+ "learning_rate": 2.6044267975612324e-08,
+ "loss": 0.0063,
+ "step": 35882
+ },
+ {
+ "epoch": 97.77384196185287,
+ "grad_norm": 0.787796676158905,
+ "learning_rate": 2.5980656552461182e-08,
+ "loss": 0.1362,
+ "step": 35883
+ },
+ {
+ "epoch": 97.77656675749319,
+ "grad_norm": 1.238281488418579,
+ "learning_rate": 2.5917122806652374e-08,
+ "loss": 0.0363,
+ "step": 35884
+ },
+ {
+ "epoch": 97.77929155313352,
+ "grad_norm": 1.2996712923049927,
+ "learning_rate": 2.585366673867884e-08,
+ "loss": 0.0695,
+ "step": 35885
+ },
+ {
+ "epoch": 97.78201634877384,
+ "grad_norm": 1.4734375476837158,
+ "learning_rate": 2.5790288349035742e-08,
+ "loss": 0.0246,
+ "step": 35886
+ },
+ {
+ "epoch": 97.78474114441417,
+ "grad_norm": 0.7510788440704346,
+ "learning_rate": 2.5726987638218238e-08,
+ "loss": 0.0093,
+ "step": 35887
+ },
+ {
+ "epoch": 97.7874659400545,
+ "grad_norm": 0.8438241481781006,
+ "learning_rate": 2.5663764606717046e-08,
+ "loss": 0.0191,
+ "step": 35888
+ },
+ {
+ "epoch": 97.79019073569482,
+ "grad_norm": 2.530409097671509,
+ "learning_rate": 2.5600619255025106e-08,
+ "loss": 0.0529,
+ "step": 35889
+ },
+ {
+ "epoch": 97.79291553133515,
+ "grad_norm": 0.44092896580696106,
+ "learning_rate": 2.5537551583635356e-08,
+ "loss": 0.0035,
+ "step": 35890
+ },
+ {
+ "epoch": 97.79564032697547,
+ "grad_norm": 0.74306321144104,
+ "learning_rate": 2.5474561593037406e-08,
+ "loss": 0.0061,
+ "step": 35891
+ },
+ {
+ "epoch": 97.7983651226158,
+ "grad_norm": 1.8056942224502563,
+ "learning_rate": 2.5411649283723083e-08,
+ "loss": 0.0128,
+ "step": 35892
+ },
+ {
+ "epoch": 97.80108991825612,
+ "grad_norm": 0.7659159302711487,
+ "learning_rate": 2.5348814656181996e-08,
+ "loss": 0.01,
+ "step": 35893
+ },
+ {
+ "epoch": 97.80381471389646,
+ "grad_norm": 0.9672598838806152,
+ "learning_rate": 2.5286057710902647e-08,
+ "loss": 0.0773,
+ "step": 35894
+ },
+ {
+ "epoch": 97.80653950953679,
+ "grad_norm": 0.8100597262382507,
+ "learning_rate": 2.5223378448374636e-08,
+ "loss": 0.0585,
+ "step": 35895
+ },
+ {
+ "epoch": 97.80926430517711,
+ "grad_norm": 1.592249870300293,
+ "learning_rate": 2.5160776869086467e-08,
+ "loss": 0.1051,
+ "step": 35896
+ },
+ {
+ "epoch": 97.81198910081744,
+ "grad_norm": 1.0789308547973633,
+ "learning_rate": 2.509825297352442e-08,
+ "loss": 0.0088,
+ "step": 35897
+ },
+ {
+ "epoch": 97.81471389645776,
+ "grad_norm": 0.6677157878875732,
+ "learning_rate": 2.5035806762175874e-08,
+ "loss": 0.0064,
+ "step": 35898
+ },
+ {
+ "epoch": 97.8174386920981,
+ "grad_norm": 0.8017475605010986,
+ "learning_rate": 2.4973438235528223e-08,
+ "loss": 0.0638,
+ "step": 35899
+ },
+ {
+ "epoch": 97.82016348773843,
+ "grad_norm": 1.0050116777420044,
+ "learning_rate": 2.4911147394065526e-08,
+ "loss": 0.0423,
+ "step": 35900
+ },
+ {
+ "epoch": 97.82288828337875,
+ "grad_norm": 1.154062032699585,
+ "learning_rate": 2.484893423827406e-08,
+ "loss": 0.0246,
+ "step": 35901
+ },
+ {
+ "epoch": 97.82561307901908,
+ "grad_norm": 0.788696825504303,
+ "learning_rate": 2.4786798768638987e-08,
+ "loss": 0.0065,
+ "step": 35902
+ },
+ {
+ "epoch": 97.8283378746594,
+ "grad_norm": 0.9790076017379761,
+ "learning_rate": 2.4724740985642148e-08,
+ "loss": 0.0118,
+ "step": 35903
+ },
+ {
+ "epoch": 97.83106267029973,
+ "grad_norm": 0.9974740147590637,
+ "learning_rate": 2.466276088976871e-08,
+ "loss": 0.0088,
+ "step": 35904
+ },
+ {
+ "epoch": 97.83378746594005,
+ "grad_norm": 1.7928324937820435,
+ "learning_rate": 2.4600858481500512e-08,
+ "loss": 0.008,
+ "step": 35905
+ },
+ {
+ "epoch": 97.83651226158038,
+ "grad_norm": 1.2826201915740967,
+ "learning_rate": 2.4539033761319387e-08,
+ "loss": 0.0073,
+ "step": 35906
+ },
+ {
+ "epoch": 97.83923705722071,
+ "grad_norm": 1.435958981513977,
+ "learning_rate": 2.447728672970717e-08,
+ "loss": 0.019,
+ "step": 35907
+ },
+ {
+ "epoch": 97.84196185286103,
+ "grad_norm": 1.5450217723846436,
+ "learning_rate": 2.4415617387145708e-08,
+ "loss": 0.0257,
+ "step": 35908
+ },
+ {
+ "epoch": 97.84468664850137,
+ "grad_norm": 0.6918794512748718,
+ "learning_rate": 2.4354025734113495e-08,
+ "loss": 0.0058,
+ "step": 35909
+ },
+ {
+ "epoch": 97.84741144414168,
+ "grad_norm": 1.3826720714569092,
+ "learning_rate": 2.4292511771091266e-08,
+ "loss": 0.02,
+ "step": 35910
+ },
+ {
+ "epoch": 97.85013623978202,
+ "grad_norm": 0.8530563116073608,
+ "learning_rate": 2.4231075498557522e-08,
+ "loss": 0.0086,
+ "step": 35911
+ },
+ {
+ "epoch": 97.85286103542235,
+ "grad_norm": 0.7820173501968384,
+ "learning_rate": 2.416971691699188e-08,
+ "loss": 0.0335,
+ "step": 35912
+ },
+ {
+ "epoch": 97.85558583106267,
+ "grad_norm": 1.0041383504867554,
+ "learning_rate": 2.410843602687063e-08,
+ "loss": 0.0074,
+ "step": 35913
+ },
+ {
+ "epoch": 97.858310626703,
+ "grad_norm": 1.0719447135925293,
+ "learning_rate": 2.4047232828672272e-08,
+ "loss": 0.0408,
+ "step": 35914
+ },
+ {
+ "epoch": 97.86103542234332,
+ "grad_norm": 0.6736873984336853,
+ "learning_rate": 2.39861073228731e-08,
+ "loss": 0.0117,
+ "step": 35915
+ },
+ {
+ "epoch": 97.86376021798365,
+ "grad_norm": 1.739210844039917,
+ "learning_rate": 2.392505950994939e-08,
+ "loss": 0.0286,
+ "step": 35916
+ },
+ {
+ "epoch": 97.86648501362397,
+ "grad_norm": 1.4174292087554932,
+ "learning_rate": 2.386408939037521e-08,
+ "loss": 0.0805,
+ "step": 35917
+ },
+ {
+ "epoch": 97.8692098092643,
+ "grad_norm": 0.7907889485359192,
+ "learning_rate": 2.380319696462685e-08,
+ "loss": 0.0087,
+ "step": 35918
+ },
+ {
+ "epoch": 97.87193460490464,
+ "grad_norm": 0.9688663482666016,
+ "learning_rate": 2.3742382233178374e-08,
+ "loss": 0.0421,
+ "step": 35919
+ },
+ {
+ "epoch": 97.87465940054496,
+ "grad_norm": 0.6956574320793152,
+ "learning_rate": 2.3681645196502733e-08,
+ "loss": 0.0082,
+ "step": 35920
+ },
+ {
+ "epoch": 97.87738419618529,
+ "grad_norm": 1.0010639429092407,
+ "learning_rate": 2.3620985855072887e-08,
+ "loss": 0.0113,
+ "step": 35921
+ },
+ {
+ "epoch": 97.88010899182561,
+ "grad_norm": 1.2698616981506348,
+ "learning_rate": 2.3560404209362897e-08,
+ "loss": 0.1538,
+ "step": 35922
+ },
+ {
+ "epoch": 97.88283378746594,
+ "grad_norm": 1.1200437545776367,
+ "learning_rate": 2.349990025984239e-08,
+ "loss": 0.0227,
+ "step": 35923
+ },
+ {
+ "epoch": 97.88555858310627,
+ "grad_norm": 0.8120627999305725,
+ "learning_rate": 2.343947400698432e-08,
+ "loss": 0.0051,
+ "step": 35924
+ },
+ {
+ "epoch": 97.88828337874659,
+ "grad_norm": 1.836391568183899,
+ "learning_rate": 2.3379125451258312e-08,
+ "loss": 0.0706,
+ "step": 35925
+ },
+ {
+ "epoch": 97.89100817438693,
+ "grad_norm": 0.7885872721672058,
+ "learning_rate": 2.3318854593135098e-08,
+ "loss": 0.0069,
+ "step": 35926
+ },
+ {
+ "epoch": 97.89373297002724,
+ "grad_norm": 0.9640624523162842,
+ "learning_rate": 2.3258661433082087e-08,
+ "loss": 0.0308,
+ "step": 35927
+ },
+ {
+ "epoch": 97.89645776566758,
+ "grad_norm": 0.854282796382904,
+ "learning_rate": 2.3198545971571118e-08,
+ "loss": 0.0053,
+ "step": 35928
+ },
+ {
+ "epoch": 97.8991825613079,
+ "grad_norm": 1.0893661975860596,
+ "learning_rate": 2.3138508209067377e-08,
+ "loss": 0.0108,
+ "step": 35929
+ },
+ {
+ "epoch": 97.90190735694823,
+ "grad_norm": 0.6848377585411072,
+ "learning_rate": 2.307854814604049e-08,
+ "loss": 0.0052,
+ "step": 35930
+ },
+ {
+ "epoch": 97.90463215258856,
+ "grad_norm": 0.80499666929245,
+ "learning_rate": 2.301866578295675e-08,
+ "loss": 0.0129,
+ "step": 35931
+ },
+ {
+ "epoch": 97.90735694822888,
+ "grad_norm": 0.569067120552063,
+ "learning_rate": 2.2958861120283558e-08,
+ "loss": 0.0043,
+ "step": 35932
+ },
+ {
+ "epoch": 97.91008174386921,
+ "grad_norm": 0.7623683214187622,
+ "learning_rate": 2.2899134158483883e-08,
+ "loss": 0.0709,
+ "step": 35933
+ },
+ {
+ "epoch": 97.91280653950953,
+ "grad_norm": 0.8749887943267822,
+ "learning_rate": 2.283948489802512e-08,
+ "loss": 0.0066,
+ "step": 35934
+ },
+ {
+ "epoch": 97.91553133514986,
+ "grad_norm": 1.6114599704742432,
+ "learning_rate": 2.2779913339371352e-08,
+ "loss": 0.0163,
+ "step": 35935
+ },
+ {
+ "epoch": 97.9182561307902,
+ "grad_norm": 0.49862775206565857,
+ "learning_rate": 2.2720419482987754e-08,
+ "loss": 0.0043,
+ "step": 35936
+ },
+ {
+ "epoch": 97.92098092643052,
+ "grad_norm": 1.1544688940048218,
+ "learning_rate": 2.2661003329335073e-08,
+ "loss": 0.0145,
+ "step": 35937
+ },
+ {
+ "epoch": 97.92370572207085,
+ "grad_norm": 1.0432337522506714,
+ "learning_rate": 2.2601664878877384e-08,
+ "loss": 0.0072,
+ "step": 35938
+ },
+ {
+ "epoch": 97.92643051771117,
+ "grad_norm": 1.6393215656280518,
+ "learning_rate": 2.2542404132077643e-08,
+ "loss": 0.0457,
+ "step": 35939
+ },
+ {
+ "epoch": 97.9291553133515,
+ "grad_norm": 1.8384164571762085,
+ "learning_rate": 2.248322108939549e-08,
+ "loss": 0.0349,
+ "step": 35940
+ },
+ {
+ "epoch": 97.93188010899182,
+ "grad_norm": 1.1330687999725342,
+ "learning_rate": 2.2424115751293885e-08,
+ "loss": 0.0134,
+ "step": 35941
+ },
+ {
+ "epoch": 97.93460490463215,
+ "grad_norm": 0.8935698866844177,
+ "learning_rate": 2.2365088118231347e-08,
+ "loss": 0.0144,
+ "step": 35942
+ },
+ {
+ "epoch": 97.93732970027249,
+ "grad_norm": 1.342336654663086,
+ "learning_rate": 2.2306138190668624e-08,
+ "loss": 0.0081,
+ "step": 35943
+ },
+ {
+ "epoch": 97.9400544959128,
+ "grad_norm": 1.2579725980758667,
+ "learning_rate": 2.2247265969065347e-08,
+ "loss": 0.013,
+ "step": 35944
+ },
+ {
+ "epoch": 97.94277929155314,
+ "grad_norm": 0.8384612798690796,
+ "learning_rate": 2.2188471453877815e-08,
+ "loss": 0.0128,
+ "step": 35945
+ },
+ {
+ "epoch": 97.94550408719346,
+ "grad_norm": 0.9608295559883118,
+ "learning_rate": 2.212975464556677e-08,
+ "loss": 0.0269,
+ "step": 35946
+ },
+ {
+ "epoch": 97.94822888283379,
+ "grad_norm": 0.7374114990234375,
+ "learning_rate": 2.2071115544587406e-08,
+ "loss": 0.0046,
+ "step": 35947
+ },
+ {
+ "epoch": 97.95095367847412,
+ "grad_norm": 1.0704584121704102,
+ "learning_rate": 2.2012554151397137e-08,
+ "loss": 0.0105,
+ "step": 35948
+ },
+ {
+ "epoch": 97.95367847411444,
+ "grad_norm": 0.42471882700920105,
+ "learning_rate": 2.195407046645115e-08,
+ "loss": 0.0039,
+ "step": 35949
+ },
+ {
+ "epoch": 97.95640326975477,
+ "grad_norm": 0.8685757517814636,
+ "learning_rate": 2.1895664490206857e-08,
+ "loss": 0.0083,
+ "step": 35950
+ },
+ {
+ "epoch": 97.95912806539509,
+ "grad_norm": 1.0190463066101074,
+ "learning_rate": 2.1837336223117233e-08,
+ "loss": 0.0102,
+ "step": 35951
+ },
+ {
+ "epoch": 97.96185286103542,
+ "grad_norm": 0.959486722946167,
+ "learning_rate": 2.1779085665637467e-08,
+ "loss": 0.0129,
+ "step": 35952
+ },
+ {
+ "epoch": 97.96457765667574,
+ "grad_norm": 1.2240228652954102,
+ "learning_rate": 2.172091281822164e-08,
+ "loss": 0.0146,
+ "step": 35953
+ },
+ {
+ "epoch": 97.96730245231608,
+ "grad_norm": 1.1918801069259644,
+ "learning_rate": 2.1662817681321613e-08,
+ "loss": 0.0194,
+ "step": 35954
+ },
+ {
+ "epoch": 97.97002724795641,
+ "grad_norm": 1.0514777898788452,
+ "learning_rate": 2.1604800255390356e-08,
+ "loss": 0.0549,
+ "step": 35955
+ },
+ {
+ "epoch": 97.97275204359673,
+ "grad_norm": 0.6304622292518616,
+ "learning_rate": 2.1546860540879734e-08,
+ "loss": 0.0075,
+ "step": 35956
+ },
+ {
+ "epoch": 97.97547683923706,
+ "grad_norm": 0.7780354022979736,
+ "learning_rate": 2.14889985382416e-08,
+ "loss": 0.0089,
+ "step": 35957
+ },
+ {
+ "epoch": 97.97820163487738,
+ "grad_norm": 1.1221201419830322,
+ "learning_rate": 2.14312142479256e-08,
+ "loss": 0.087,
+ "step": 35958
+ },
+ {
+ "epoch": 97.98092643051771,
+ "grad_norm": 1.5960665941238403,
+ "learning_rate": 2.1373507670381376e-08,
+ "loss": 0.0221,
+ "step": 35959
+ },
+ {
+ "epoch": 97.98365122615803,
+ "grad_norm": 1.040752649307251,
+ "learning_rate": 2.1315878806060787e-08,
+ "loss": 0.0147,
+ "step": 35960
+ },
+ {
+ "epoch": 97.98637602179836,
+ "grad_norm": 0.6522642374038696,
+ "learning_rate": 2.125832765541014e-08,
+ "loss": 0.0061,
+ "step": 35961
+ },
+ {
+ "epoch": 97.9891008174387,
+ "grad_norm": 1.0407400131225586,
+ "learning_rate": 2.1200854218877964e-08,
+ "loss": 0.0075,
+ "step": 35962
+ },
+ {
+ "epoch": 97.99182561307902,
+ "grad_norm": 1.4510979652404785,
+ "learning_rate": 2.1143458496912795e-08,
+ "loss": 0.0047,
+ "step": 35963
+ },
+ {
+ "epoch": 97.99455040871935,
+ "grad_norm": 0.8309508562088013,
+ "learning_rate": 2.1086140489960937e-08,
+ "loss": 0.0071,
+ "step": 35964
+ },
+ {
+ "epoch": 97.99727520435967,
+ "grad_norm": 0.8030584454536438,
+ "learning_rate": 2.1028900198468704e-08,
+ "loss": 0.0059,
+ "step": 35965
+ },
+ {
+ "epoch": 98.0,
+ "grad_norm": 1.4675887823104858,
+ "learning_rate": 2.0971737622883515e-08,
+ "loss": 0.0164,
+ "step": 35966
+ },
+ {
+ "epoch": 98.00272479564033,
+ "grad_norm": 0.8461031913757324,
+ "learning_rate": 2.0914652763648348e-08,
+ "loss": 0.0084,
+ "step": 35967
+ },
+ {
+ "epoch": 98.00544959128065,
+ "grad_norm": 1.0063178539276123,
+ "learning_rate": 2.085764562120951e-08,
+ "loss": 0.0154,
+ "step": 35968
+ },
+ {
+ "epoch": 98.00817438692098,
+ "grad_norm": 2.860100746154785,
+ "learning_rate": 2.0800716196008874e-08,
+ "loss": 0.0301,
+ "step": 35969
+ },
+ {
+ "epoch": 98.0108991825613,
+ "grad_norm": 1.847880482673645,
+ "learning_rate": 2.074386448849164e-08,
+ "loss": 0.0194,
+ "step": 35970
+ },
+ {
+ "epoch": 98.01362397820164,
+ "grad_norm": 0.9811599850654602,
+ "learning_rate": 2.068709049910078e-08,
+ "loss": 0.0199,
+ "step": 35971
+ },
+ {
+ "epoch": 98.01634877384195,
+ "grad_norm": 1.2767016887664795,
+ "learning_rate": 2.0630394228277063e-08,
+ "loss": 0.0226,
+ "step": 35972
+ },
+ {
+ "epoch": 98.01907356948229,
+ "grad_norm": 1.0033702850341797,
+ "learning_rate": 2.057377567646235e-08,
+ "loss": 0.0089,
+ "step": 35973
+ },
+ {
+ "epoch": 98.02179836512262,
+ "grad_norm": 0.4695083498954773,
+ "learning_rate": 2.05172348440974e-08,
+ "loss": 0.006,
+ "step": 35974
+ },
+ {
+ "epoch": 98.02452316076294,
+ "grad_norm": 0.7618486881256104,
+ "learning_rate": 2.0460771731624086e-08,
+ "loss": 0.0064,
+ "step": 35975
+ },
+ {
+ "epoch": 98.02724795640327,
+ "grad_norm": 1.1579090356826782,
+ "learning_rate": 2.040438633948094e-08,
+ "loss": 0.0748,
+ "step": 35976
+ },
+ {
+ "epoch": 98.02997275204359,
+ "grad_norm": 1.2181732654571533,
+ "learning_rate": 2.034807866810762e-08,
+ "loss": 0.0392,
+ "step": 35977
+ },
+ {
+ "epoch": 98.03269754768392,
+ "grad_norm": 1.2335567474365234,
+ "learning_rate": 2.029184871794154e-08,
+ "loss": 0.1165,
+ "step": 35978
+ },
+ {
+ "epoch": 98.03542234332426,
+ "grad_norm": 1.1884949207305908,
+ "learning_rate": 2.0235696489421254e-08,
+ "loss": 0.1336,
+ "step": 35979
+ },
+ {
+ "epoch": 98.03814713896458,
+ "grad_norm": 1.0540610551834106,
+ "learning_rate": 2.0179621982984178e-08,
+ "loss": 0.062,
+ "step": 35980
+ },
+ {
+ "epoch": 98.04087193460491,
+ "grad_norm": 16.585844039916992,
+ "learning_rate": 2.0123625199067743e-08,
+ "loss": 0.0294,
+ "step": 35981
+ },
+ {
+ "epoch": 98.04359673024523,
+ "grad_norm": 0.8580868244171143,
+ "learning_rate": 2.0067706138107157e-08,
+ "loss": 0.0169,
+ "step": 35982
+ },
+ {
+ "epoch": 98.04632152588556,
+ "grad_norm": 1.2580316066741943,
+ "learning_rate": 2.001186480053763e-08,
+ "loss": 0.0335,
+ "step": 35983
+ },
+ {
+ "epoch": 98.04904632152588,
+ "grad_norm": 1.6449337005615234,
+ "learning_rate": 1.9956101186794362e-08,
+ "loss": 0.0156,
+ "step": 35984
+ },
+ {
+ "epoch": 98.05177111716621,
+ "grad_norm": 0.3561263084411621,
+ "learning_rate": 1.990041529731257e-08,
+ "loss": 0.0041,
+ "step": 35985
+ },
+ {
+ "epoch": 98.05449591280654,
+ "grad_norm": 1.1089496612548828,
+ "learning_rate": 1.9844807132524123e-08,
+ "loss": 0.0056,
+ "step": 35986
+ },
+ {
+ "epoch": 98.05722070844686,
+ "grad_norm": 1.162574052810669,
+ "learning_rate": 1.9789276692863125e-08,
+ "loss": 0.0113,
+ "step": 35987
+ },
+ {
+ "epoch": 98.0599455040872,
+ "grad_norm": 0.8880468606948853,
+ "learning_rate": 1.9733823978762555e-08,
+ "loss": 0.0123,
+ "step": 35988
+ },
+ {
+ "epoch": 98.06267029972751,
+ "grad_norm": 0.28870776295661926,
+ "learning_rate": 1.96784489906543e-08,
+ "loss": 0.0026,
+ "step": 35989
+ },
+ {
+ "epoch": 98.06539509536785,
+ "grad_norm": 0.7314019799232483,
+ "learning_rate": 1.9623151728968004e-08,
+ "loss": 0.0107,
+ "step": 35990
+ },
+ {
+ "epoch": 98.06811989100818,
+ "grad_norm": 1.5500030517578125,
+ "learning_rate": 1.9567932194135553e-08,
+ "loss": 0.0142,
+ "step": 35991
+ },
+ {
+ "epoch": 98.0708446866485,
+ "grad_norm": 0.7102097868919373,
+ "learning_rate": 1.951279038658771e-08,
+ "loss": 0.0051,
+ "step": 35992
+ },
+ {
+ "epoch": 98.07356948228883,
+ "grad_norm": 1.1077345609664917,
+ "learning_rate": 1.9457726306751913e-08,
+ "loss": 0.0568,
+ "step": 35993
+ },
+ {
+ "epoch": 98.07629427792915,
+ "grad_norm": 0.6840619444847107,
+ "learning_rate": 1.9402739955058925e-08,
+ "loss": 0.0083,
+ "step": 35994
+ },
+ {
+ "epoch": 98.07901907356948,
+ "grad_norm": 1.7611173391342163,
+ "learning_rate": 1.9347831331936186e-08,
+ "loss": 0.1681,
+ "step": 35995
+ },
+ {
+ "epoch": 98.0817438692098,
+ "grad_norm": 1.1905288696289062,
+ "learning_rate": 1.9293000437811127e-08,
+ "loss": 0.0284,
+ "step": 35996
+ },
+ {
+ "epoch": 98.08446866485014,
+ "grad_norm": 0.8527974486351013,
+ "learning_rate": 1.9238247273110077e-08,
+ "loss": 0.0055,
+ "step": 35997
+ },
+ {
+ "epoch": 98.08719346049047,
+ "grad_norm": 0.813072681427002,
+ "learning_rate": 1.9183571838261584e-08,
+ "loss": 0.0068,
+ "step": 35998
+ },
+ {
+ "epoch": 98.08991825613079,
+ "grad_norm": 0.87314373254776,
+ "learning_rate": 1.9128974133689748e-08,
+ "loss": 0.0071,
+ "step": 35999
+ },
+ {
+ "epoch": 98.09264305177112,
+ "grad_norm": 1.20131254196167,
+ "learning_rate": 1.907445415981979e-08,
+ "loss": 0.0124,
+ "step": 36000
+ },
+ {
+ "epoch": 98.09536784741144,
+ "grad_norm": 0.8736028075218201,
+ "learning_rate": 1.902001191707803e-08,
+ "loss": 0.0101,
+ "step": 36001
+ },
+ {
+ "epoch": 98.09809264305177,
+ "grad_norm": 1.0924359560012817,
+ "learning_rate": 1.8965647405885247e-08,
+ "loss": 0.013,
+ "step": 36002
+ },
+ {
+ "epoch": 98.1008174386921,
+ "grad_norm": 1.3925150632858276,
+ "learning_rate": 1.8911360626668874e-08,
+ "loss": 0.0078,
+ "step": 36003
+ },
+ {
+ "epoch": 98.10354223433242,
+ "grad_norm": 0.7674984335899353,
+ "learning_rate": 1.8857151579847465e-08,
+ "loss": 0.1305,
+ "step": 36004
+ },
+ {
+ "epoch": 98.10626702997276,
+ "grad_norm": 0.5960785746574402,
+ "learning_rate": 1.8803020265847348e-08,
+ "loss": 0.0067,
+ "step": 36005
+ },
+ {
+ "epoch": 98.10899182561307,
+ "grad_norm": 1.5204001665115356,
+ "learning_rate": 1.874896668508597e-08,
+ "loss": 0.0855,
+ "step": 36006
+ },
+ {
+ "epoch": 98.11171662125341,
+ "grad_norm": 1.8018078804016113,
+ "learning_rate": 1.8694990837988534e-08,
+ "loss": 0.007,
+ "step": 36007
+ },
+ {
+ "epoch": 98.11444141689373,
+ "grad_norm": 1.4303230047225952,
+ "learning_rate": 1.8641092724971388e-08,
+ "loss": 0.1605,
+ "step": 36008
+ },
+ {
+ "epoch": 98.11716621253406,
+ "grad_norm": 0.8031323552131653,
+ "learning_rate": 1.858727234645752e-08,
+ "loss": 0.0057,
+ "step": 36009
+ },
+ {
+ "epoch": 98.11989100817439,
+ "grad_norm": 0.3257509768009186,
+ "learning_rate": 1.8533529702864374e-08,
+ "loss": 0.004,
+ "step": 36010
+ },
+ {
+ "epoch": 98.12261580381471,
+ "grad_norm": 1.0443928241729736,
+ "learning_rate": 1.8479864794610502e-08,
+ "loss": 0.01,
+ "step": 36011
+ },
+ {
+ "epoch": 98.12534059945504,
+ "grad_norm": 0.7123222947120667,
+ "learning_rate": 1.8426277622114463e-08,
+ "loss": 0.0068,
+ "step": 36012
+ },
+ {
+ "epoch": 98.12806539509536,
+ "grad_norm": 0.4483824074268341,
+ "learning_rate": 1.83727681857937e-08,
+ "loss": 0.004,
+ "step": 36013
+ },
+ {
+ "epoch": 98.1307901907357,
+ "grad_norm": 0.7531833052635193,
+ "learning_rate": 1.8319336486063433e-08,
+ "loss": 0.0055,
+ "step": 36014
+ },
+ {
+ "epoch": 98.13351498637603,
+ "grad_norm": 1.2539390325546265,
+ "learning_rate": 1.826598252334222e-08,
+ "loss": 0.0394,
+ "step": 36015
+ },
+ {
+ "epoch": 98.13623978201635,
+ "grad_norm": 0.7289319634437561,
+ "learning_rate": 1.821270629804306e-08,
+ "loss": 0.0094,
+ "step": 36016
+ },
+ {
+ "epoch": 98.13896457765668,
+ "grad_norm": 0.9716967344284058,
+ "learning_rate": 1.81595078105834e-08,
+ "loss": 0.0066,
+ "step": 36017
+ },
+ {
+ "epoch": 98.141689373297,
+ "grad_norm": 1.3100593090057373,
+ "learning_rate": 1.810638706137513e-08,
+ "loss": 0.0713,
+ "step": 36018
+ },
+ {
+ "epoch": 98.14441416893733,
+ "grad_norm": 0.8560550212860107,
+ "learning_rate": 1.8053344050833478e-08,
+ "loss": 0.0091,
+ "step": 36019
+ },
+ {
+ "epoch": 98.14713896457765,
+ "grad_norm": 1.3468188047409058,
+ "learning_rate": 1.800037877937144e-08,
+ "loss": 0.0218,
+ "step": 36020
+ },
+ {
+ "epoch": 98.14986376021798,
+ "grad_norm": 1.2919073104858398,
+ "learning_rate": 1.7947491247399808e-08,
+ "loss": 0.0434,
+ "step": 36021
+ },
+ {
+ "epoch": 98.15258855585832,
+ "grad_norm": 1.298195719718933,
+ "learning_rate": 1.7894681455333794e-08,
+ "loss": 0.0134,
+ "step": 36022
+ },
+ {
+ "epoch": 98.15531335149863,
+ "grad_norm": 1.1114519834518433,
+ "learning_rate": 1.784194940358086e-08,
+ "loss": 0.0221,
+ "step": 36023
+ },
+ {
+ "epoch": 98.15803814713897,
+ "grad_norm": 1.162466049194336,
+ "learning_rate": 1.7789295092555114e-08,
+ "loss": 0.0173,
+ "step": 36024
+ },
+ {
+ "epoch": 98.16076294277929,
+ "grad_norm": 2.1714348793029785,
+ "learning_rate": 1.773671852266401e-08,
+ "loss": 0.0208,
+ "step": 36025
+ },
+ {
+ "epoch": 98.16348773841962,
+ "grad_norm": 0.5691305994987488,
+ "learning_rate": 1.7684219694318328e-08,
+ "loss": 0.0052,
+ "step": 36026
+ },
+ {
+ "epoch": 98.16621253405995,
+ "grad_norm": 1.4446914196014404,
+ "learning_rate": 1.763179860792663e-08,
+ "loss": 0.0309,
+ "step": 36027
+ },
+ {
+ "epoch": 98.16893732970027,
+ "grad_norm": 0.8341707587242126,
+ "learning_rate": 1.7579455263897483e-08,
+ "loss": 0.0143,
+ "step": 36028
+ },
+ {
+ "epoch": 98.1716621253406,
+ "grad_norm": 1.3411327600479126,
+ "learning_rate": 1.7527189662638334e-08,
+ "loss": 0.0293,
+ "step": 36029
+ },
+ {
+ "epoch": 98.17438692098092,
+ "grad_norm": 0.6561723351478577,
+ "learning_rate": 1.7475001804555525e-08,
+ "loss": 0.0085,
+ "step": 36030
+ },
+ {
+ "epoch": 98.17711171662125,
+ "grad_norm": 0.6526325941085815,
+ "learning_rate": 1.7422891690056508e-08,
+ "loss": 0.007,
+ "step": 36031
+ },
+ {
+ "epoch": 98.17983651226157,
+ "grad_norm": 0.814092218875885,
+ "learning_rate": 1.7370859319547628e-08,
+ "loss": 0.0114,
+ "step": 36032
+ },
+ {
+ "epoch": 98.1825613079019,
+ "grad_norm": 1.195733904838562,
+ "learning_rate": 1.7318904693431894e-08,
+ "loss": 0.0162,
+ "step": 36033
+ },
+ {
+ "epoch": 98.18528610354224,
+ "grad_norm": 1.526248574256897,
+ "learning_rate": 1.7267027812115646e-08,
+ "loss": 0.0682,
+ "step": 36034
+ },
+ {
+ "epoch": 98.18801089918256,
+ "grad_norm": 1.5487006902694702,
+ "learning_rate": 1.7215228676003005e-08,
+ "loss": 0.0963,
+ "step": 36035
+ },
+ {
+ "epoch": 98.19073569482289,
+ "grad_norm": 0.4467778205871582,
+ "learning_rate": 1.7163507285495872e-08,
+ "loss": 0.0056,
+ "step": 36036
+ },
+ {
+ "epoch": 98.19346049046321,
+ "grad_norm": 0.8157820701599121,
+ "learning_rate": 1.7111863640999483e-08,
+ "loss": 0.0737,
+ "step": 36037
+ },
+ {
+ "epoch": 98.19618528610354,
+ "grad_norm": 0.7000458836555481,
+ "learning_rate": 1.706029774291351e-08,
+ "loss": 0.0062,
+ "step": 36038
+ },
+ {
+ "epoch": 98.19891008174388,
+ "grad_norm": 0.9211087226867676,
+ "learning_rate": 1.7008809591639863e-08,
+ "loss": 0.0359,
+ "step": 36039
+ },
+ {
+ "epoch": 98.2016348773842,
+ "grad_norm": 1.089221715927124,
+ "learning_rate": 1.6957399187581547e-08,
+ "loss": 0.1206,
+ "step": 36040
+ },
+ {
+ "epoch": 98.20435967302453,
+ "grad_norm": 0.5728310942649841,
+ "learning_rate": 1.6906066531137132e-08,
+ "loss": 0.0057,
+ "step": 36041
+ },
+ {
+ "epoch": 98.20708446866485,
+ "grad_norm": 0.8405706882476807,
+ "learning_rate": 1.6854811622707412e-08,
+ "loss": 0.0161,
+ "step": 36042
+ },
+ {
+ "epoch": 98.20980926430518,
+ "grad_norm": 1.4197323322296143,
+ "learning_rate": 1.6803634462690955e-08,
+ "loss": 0.0338,
+ "step": 36043
+ },
+ {
+ "epoch": 98.2125340599455,
+ "grad_norm": 0.8037477731704712,
+ "learning_rate": 1.675253505148633e-08,
+ "loss": 0.0097,
+ "step": 36044
+ },
+ {
+ "epoch": 98.21525885558583,
+ "grad_norm": 0.7398695349693298,
+ "learning_rate": 1.6701513389492106e-08,
+ "loss": 0.0155,
+ "step": 36045
+ },
+ {
+ "epoch": 98.21798365122616,
+ "grad_norm": 0.4331390857696533,
+ "learning_rate": 1.6650569477104638e-08,
+ "loss": 0.0048,
+ "step": 36046
+ },
+ {
+ "epoch": 98.22070844686648,
+ "grad_norm": 0.6775155067443848,
+ "learning_rate": 1.6599703314721382e-08,
+ "loss": 0.0069,
+ "step": 36047
+ },
+ {
+ "epoch": 98.22343324250681,
+ "grad_norm": 0.6439900994300842,
+ "learning_rate": 1.6548914902738688e-08,
+ "loss": 0.0054,
+ "step": 36048
+ },
+ {
+ "epoch": 98.22615803814713,
+ "grad_norm": 0.6944945454597473,
+ "learning_rate": 1.6498204241551796e-08,
+ "loss": 0.0067,
+ "step": 36049
+ },
+ {
+ "epoch": 98.22888283378747,
+ "grad_norm": 1.0537477731704712,
+ "learning_rate": 1.6447571331554836e-08,
+ "loss": 0.0528,
+ "step": 36050
+ },
+ {
+ "epoch": 98.2316076294278,
+ "grad_norm": 0.4459365904331207,
+ "learning_rate": 1.6397016173143043e-08,
+ "loss": 0.0041,
+ "step": 36051
+ },
+ {
+ "epoch": 98.23433242506812,
+ "grad_norm": 0.9278878569602966,
+ "learning_rate": 1.6346538766710552e-08,
+ "loss": 0.0525,
+ "step": 36052
+ },
+ {
+ "epoch": 98.23705722070845,
+ "grad_norm": 1.3269041776657104,
+ "learning_rate": 1.6296139112649268e-08,
+ "loss": 0.0212,
+ "step": 36053
+ },
+ {
+ "epoch": 98.23978201634877,
+ "grad_norm": 1.3730762004852295,
+ "learning_rate": 1.6245817211353322e-08,
+ "loss": 0.0277,
+ "step": 36054
+ },
+ {
+ "epoch": 98.2425068119891,
+ "grad_norm": 0.719479501247406,
+ "learning_rate": 1.61955730632124e-08,
+ "loss": 0.0048,
+ "step": 36055
+ },
+ {
+ "epoch": 98.24523160762942,
+ "grad_norm": 0.6655589938163757,
+ "learning_rate": 1.6145406668619523e-08,
+ "loss": 0.0071,
+ "step": 36056
+ },
+ {
+ "epoch": 98.24795640326975,
+ "grad_norm": 1.3701386451721191,
+ "learning_rate": 1.609531802796549e-08,
+ "loss": 0.0108,
+ "step": 36057
+ },
+ {
+ "epoch": 98.25068119891009,
+ "grad_norm": 0.5831694602966309,
+ "learning_rate": 1.604530714163888e-08,
+ "loss": 0.0051,
+ "step": 36058
+ },
+ {
+ "epoch": 98.2534059945504,
+ "grad_norm": 1.8017206192016602,
+ "learning_rate": 1.5995374010030483e-08,
+ "loss": 0.1228,
+ "step": 36059
+ },
+ {
+ "epoch": 98.25613079019074,
+ "grad_norm": 0.937202513217926,
+ "learning_rate": 1.594551863352889e-08,
+ "loss": 0.0098,
+ "step": 36060
+ },
+ {
+ "epoch": 98.25885558583106,
+ "grad_norm": 1.1085753440856934,
+ "learning_rate": 1.5895741012521558e-08,
+ "loss": 0.0702,
+ "step": 36061
+ },
+ {
+ "epoch": 98.26158038147139,
+ "grad_norm": 1.7401248216629028,
+ "learning_rate": 1.5846041147397072e-08,
+ "loss": 0.0125,
+ "step": 36062
+ },
+ {
+ "epoch": 98.26430517711172,
+ "grad_norm": 0.5041373372077942,
+ "learning_rate": 1.5796419038542897e-08,
+ "loss": 0.004,
+ "step": 36063
+ },
+ {
+ "epoch": 98.26702997275204,
+ "grad_norm": 1.1402373313903809,
+ "learning_rate": 1.5746874686344284e-08,
+ "loss": 0.0067,
+ "step": 36064
+ },
+ {
+ "epoch": 98.26975476839237,
+ "grad_norm": 0.5209479928016663,
+ "learning_rate": 1.5697408091188694e-08,
+ "loss": 0.0059,
+ "step": 36065
+ },
+ {
+ "epoch": 98.2724795640327,
+ "grad_norm": 1.0057073831558228,
+ "learning_rate": 1.5648019253459156e-08,
+ "loss": 0.041,
+ "step": 36066
+ },
+ {
+ "epoch": 98.27520435967303,
+ "grad_norm": 0.59275221824646,
+ "learning_rate": 1.5598708173542032e-08,
+ "loss": 0.005,
+ "step": 36067
+ },
+ {
+ "epoch": 98.27792915531334,
+ "grad_norm": 0.9679164290428162,
+ "learning_rate": 1.5549474851820346e-08,
+ "loss": 0.0078,
+ "step": 36068
+ },
+ {
+ "epoch": 98.28065395095368,
+ "grad_norm": 0.9719621539115906,
+ "learning_rate": 1.5500319288678234e-08,
+ "loss": 0.0237,
+ "step": 36069
+ },
+ {
+ "epoch": 98.28337874659401,
+ "grad_norm": 2.094942808151245,
+ "learning_rate": 1.5451241484498724e-08,
+ "loss": 0.0318,
+ "step": 36070
+ },
+ {
+ "epoch": 98.28610354223433,
+ "grad_norm": 1.0368640422821045,
+ "learning_rate": 1.5402241439663736e-08,
+ "loss": 0.0325,
+ "step": 36071
+ },
+ {
+ "epoch": 98.28882833787466,
+ "grad_norm": 0.9645627737045288,
+ "learning_rate": 1.535331915455407e-08,
+ "loss": 0.0095,
+ "step": 36072
+ },
+ {
+ "epoch": 98.29155313351498,
+ "grad_norm": 0.48115041851997375,
+ "learning_rate": 1.5304474629551648e-08,
+ "loss": 0.0045,
+ "step": 36073
+ },
+ {
+ "epoch": 98.29427792915531,
+ "grad_norm": 1.397071123123169,
+ "learning_rate": 1.5255707865036163e-08,
+ "loss": 0.0203,
+ "step": 36074
+ },
+ {
+ "epoch": 98.29700272479565,
+ "grad_norm": 1.1873646974563599,
+ "learning_rate": 1.5207018861388423e-08,
+ "loss": 0.0152,
+ "step": 36075
+ },
+ {
+ "epoch": 98.29972752043597,
+ "grad_norm": 1.9191906452178955,
+ "learning_rate": 1.5158407618987014e-08,
+ "loss": 0.0154,
+ "step": 36076
+ },
+ {
+ "epoch": 98.3024523160763,
+ "grad_norm": 0.564508855342865,
+ "learning_rate": 1.5109874138210522e-08,
+ "loss": 0.005,
+ "step": 36077
+ },
+ {
+ "epoch": 98.30517711171662,
+ "grad_norm": 0.9539740681648254,
+ "learning_rate": 1.5061418419437536e-08,
+ "loss": 0.0192,
+ "step": 36078
+ },
+ {
+ "epoch": 98.30790190735695,
+ "grad_norm": 0.6698846220970154,
+ "learning_rate": 1.5013040463043305e-08,
+ "loss": 0.0102,
+ "step": 36079
+ },
+ {
+ "epoch": 98.31062670299727,
+ "grad_norm": 1.2009483575820923,
+ "learning_rate": 1.496474026940753e-08,
+ "loss": 0.0086,
+ "step": 36080
+ },
+ {
+ "epoch": 98.3133514986376,
+ "grad_norm": 0.6390812397003174,
+ "learning_rate": 1.4916517838904355e-08,
+ "loss": 0.0059,
+ "step": 36081
+ },
+ {
+ "epoch": 98.31607629427793,
+ "grad_norm": 0.9024732708930969,
+ "learning_rate": 1.4868373171910144e-08,
+ "loss": 0.0071,
+ "step": 36082
+ },
+ {
+ "epoch": 98.31880108991825,
+ "grad_norm": 0.6268677115440369,
+ "learning_rate": 1.4820306268800155e-08,
+ "loss": 0.0052,
+ "step": 36083
+ },
+ {
+ "epoch": 98.32152588555859,
+ "grad_norm": 0.7836298942565918,
+ "learning_rate": 1.4772317129947422e-08,
+ "loss": 0.0107,
+ "step": 36084
+ },
+ {
+ "epoch": 98.3242506811989,
+ "grad_norm": 1.5763474702835083,
+ "learning_rate": 1.47244057557272e-08,
+ "loss": 0.0736,
+ "step": 36085
+ },
+ {
+ "epoch": 98.32697547683924,
+ "grad_norm": 1.0511314868927002,
+ "learning_rate": 1.4676572146511414e-08,
+ "loss": 0.0383,
+ "step": 36086
+ },
+ {
+ "epoch": 98.32970027247957,
+ "grad_norm": 1.0720446109771729,
+ "learning_rate": 1.46288163026731e-08,
+ "loss": 0.0251,
+ "step": 36087
+ },
+ {
+ "epoch": 98.33242506811989,
+ "grad_norm": 1.2067469358444214,
+ "learning_rate": 1.458113822458529e-08,
+ "loss": 0.0181,
+ "step": 36088
+ },
+ {
+ "epoch": 98.33514986376022,
+ "grad_norm": 0.9606432318687439,
+ "learning_rate": 1.453353791261658e-08,
+ "loss": 0.0164,
+ "step": 36089
+ },
+ {
+ "epoch": 98.33787465940054,
+ "grad_norm": 1.2037771940231323,
+ "learning_rate": 1.4486015367140005e-08,
+ "loss": 0.0409,
+ "step": 36090
+ },
+ {
+ "epoch": 98.34059945504087,
+ "grad_norm": 0.7260562181472778,
+ "learning_rate": 1.443857058852527e-08,
+ "loss": 0.0058,
+ "step": 36091
+ },
+ {
+ "epoch": 98.34332425068119,
+ "grad_norm": 1.129482626914978,
+ "learning_rate": 1.4391203577140966e-08,
+ "loss": 0.0285,
+ "step": 36092
+ },
+ {
+ "epoch": 98.34604904632153,
+ "grad_norm": 1.3475263118743896,
+ "learning_rate": 1.4343914333356801e-08,
+ "loss": 0.0403,
+ "step": 36093
+ },
+ {
+ "epoch": 98.34877384196186,
+ "grad_norm": 1.079330563545227,
+ "learning_rate": 1.4296702857541367e-08,
+ "loss": 0.0081,
+ "step": 36094
+ },
+ {
+ "epoch": 98.35149863760218,
+ "grad_norm": 0.5882667899131775,
+ "learning_rate": 1.4249569150061037e-08,
+ "loss": 0.0058,
+ "step": 36095
+ },
+ {
+ "epoch": 98.35422343324251,
+ "grad_norm": 1.200168490409851,
+ "learning_rate": 1.4202513211284408e-08,
+ "loss": 0.1199,
+ "step": 36096
+ },
+ {
+ "epoch": 98.35694822888283,
+ "grad_norm": 1.8241533041000366,
+ "learning_rate": 1.415553504157674e-08,
+ "loss": 0.0255,
+ "step": 36097
+ },
+ {
+ "epoch": 98.35967302452316,
+ "grad_norm": 0.6314891576766968,
+ "learning_rate": 1.4108634641304409e-08,
+ "loss": 0.0053,
+ "step": 36098
+ },
+ {
+ "epoch": 98.3623978201635,
+ "grad_norm": 0.9289513826370239,
+ "learning_rate": 1.4061812010832676e-08,
+ "loss": 0.0134,
+ "step": 36099
+ },
+ {
+ "epoch": 98.36512261580381,
+ "grad_norm": 0.5743896961212158,
+ "learning_rate": 1.4015067150525696e-08,
+ "loss": 0.0055,
+ "step": 36100
+ },
+ {
+ "epoch": 98.36784741144415,
+ "grad_norm": 1.0927269458770752,
+ "learning_rate": 1.3968400060748733e-08,
+ "loss": 0.0115,
+ "step": 36101
+ },
+ {
+ "epoch": 98.37057220708446,
+ "grad_norm": 1.5928540229797363,
+ "learning_rate": 1.3921810741864828e-08,
+ "loss": 0.0083,
+ "step": 36102
+ },
+ {
+ "epoch": 98.3732970027248,
+ "grad_norm": 1.1474097967147827,
+ "learning_rate": 1.3875299194235914e-08,
+ "loss": 0.0466,
+ "step": 36103
+ },
+ {
+ "epoch": 98.37602179836512,
+ "grad_norm": 0.5688949227333069,
+ "learning_rate": 1.3828865418223925e-08,
+ "loss": 0.0055,
+ "step": 36104
+ },
+ {
+ "epoch": 98.37874659400545,
+ "grad_norm": 1.403441071510315,
+ "learning_rate": 1.3782509414191902e-08,
+ "loss": 0.0138,
+ "step": 36105
+ },
+ {
+ "epoch": 98.38147138964578,
+ "grad_norm": 0.7315671443939209,
+ "learning_rate": 1.373623118250067e-08,
+ "loss": 0.0077,
+ "step": 36106
+ },
+ {
+ "epoch": 98.3841961852861,
+ "grad_norm": 1.4600472450256348,
+ "learning_rate": 1.3690030723509939e-08,
+ "loss": 0.0191,
+ "step": 36107
+ },
+ {
+ "epoch": 98.38692098092643,
+ "grad_norm": 1.882390022277832,
+ "learning_rate": 1.3643908037580533e-08,
+ "loss": 0.1227,
+ "step": 36108
+ },
+ {
+ "epoch": 98.38964577656675,
+ "grad_norm": 1.2267812490463257,
+ "learning_rate": 1.3597863125069943e-08,
+ "loss": 0.013,
+ "step": 36109
+ },
+ {
+ "epoch": 98.39237057220708,
+ "grad_norm": 1.0773046016693115,
+ "learning_rate": 1.3551895986338993e-08,
+ "loss": 0.1083,
+ "step": 36110
+ },
+ {
+ "epoch": 98.39509536784742,
+ "grad_norm": 7.197941303253174,
+ "learning_rate": 1.3506006621742951e-08,
+ "loss": 0.0122,
+ "step": 36111
+ },
+ {
+ "epoch": 98.39782016348774,
+ "grad_norm": 2.170322895050049,
+ "learning_rate": 1.3460195031641532e-08,
+ "loss": 0.1307,
+ "step": 36112
+ },
+ {
+ "epoch": 98.40054495912807,
+ "grad_norm": 1.1133842468261719,
+ "learning_rate": 1.3414461216391117e-08,
+ "loss": 0.0237,
+ "step": 36113
+ },
+ {
+ "epoch": 98.40326975476839,
+ "grad_norm": 0.6707451939582825,
+ "learning_rate": 1.3368805176346977e-08,
+ "loss": 0.0062,
+ "step": 36114
+ },
+ {
+ "epoch": 98.40599455040872,
+ "grad_norm": 0.9167181849479675,
+ "learning_rate": 1.3323226911864385e-08,
+ "loss": 0.0693,
+ "step": 36115
+ },
+ {
+ "epoch": 98.40871934604904,
+ "grad_norm": 0.827379584312439,
+ "learning_rate": 1.327772642330083e-08,
+ "loss": 0.0087,
+ "step": 36116
+ },
+ {
+ "epoch": 98.41144414168937,
+ "grad_norm": 1.3252384662628174,
+ "learning_rate": 1.3232303711007143e-08,
+ "loss": 0.0244,
+ "step": 36117
+ },
+ {
+ "epoch": 98.4141689373297,
+ "grad_norm": 1.2402358055114746,
+ "learning_rate": 1.3186958775339709e-08,
+ "loss": 0.0123,
+ "step": 36118
+ },
+ {
+ "epoch": 98.41689373297002,
+ "grad_norm": 1.2246747016906738,
+ "learning_rate": 1.3141691616651575e-08,
+ "loss": 0.0511,
+ "step": 36119
+ },
+ {
+ "epoch": 98.41961852861036,
+ "grad_norm": 1.3156375885009766,
+ "learning_rate": 1.3096502235293573e-08,
+ "loss": 0.0732,
+ "step": 36120
+ },
+ {
+ "epoch": 98.42234332425068,
+ "grad_norm": 2.313568115234375,
+ "learning_rate": 1.3051390631618755e-08,
+ "loss": 0.066,
+ "step": 36121
+ },
+ {
+ "epoch": 98.42506811989101,
+ "grad_norm": 1.0806630849838257,
+ "learning_rate": 1.3006356805977949e-08,
+ "loss": 0.1181,
+ "step": 36122
+ },
+ {
+ "epoch": 98.42779291553134,
+ "grad_norm": 1.0052813291549683,
+ "learning_rate": 1.2961400758721987e-08,
+ "loss": 0.0105,
+ "step": 36123
+ },
+ {
+ "epoch": 98.43051771117166,
+ "grad_norm": 0.7112525701522827,
+ "learning_rate": 1.2916522490201699e-08,
+ "loss": 0.0088,
+ "step": 36124
+ },
+ {
+ "epoch": 98.433242506812,
+ "grad_norm": 0.6135408878326416,
+ "learning_rate": 1.2871722000766806e-08,
+ "loss": 0.0058,
+ "step": 36125
+ },
+ {
+ "epoch": 98.43596730245231,
+ "grad_norm": 1.3155241012573242,
+ "learning_rate": 1.2826999290764807e-08,
+ "loss": 0.0181,
+ "step": 36126
+ },
+ {
+ "epoch": 98.43869209809264,
+ "grad_norm": 1.5866032838821411,
+ "learning_rate": 1.2782354360544314e-08,
+ "loss": 0.0094,
+ "step": 36127
+ },
+ {
+ "epoch": 98.44141689373296,
+ "grad_norm": 1.0288199186325073,
+ "learning_rate": 1.2737787210453933e-08,
+ "loss": 0.0237,
+ "step": 36128
+ },
+ {
+ "epoch": 98.4441416893733,
+ "grad_norm": 1.1498011350631714,
+ "learning_rate": 1.2693297840840058e-08,
+ "loss": 0.0155,
+ "step": 36129
+ },
+ {
+ "epoch": 98.44686648501363,
+ "grad_norm": 0.9196683764457703,
+ "learning_rate": 1.2648886252050186e-08,
+ "loss": 0.0146,
+ "step": 36130
+ },
+ {
+ "epoch": 98.44959128065395,
+ "grad_norm": 0.9678031802177429,
+ "learning_rate": 1.2604552444428487e-08,
+ "loss": 0.0125,
+ "step": 36131
+ },
+ {
+ "epoch": 98.45231607629428,
+ "grad_norm": 0.677754819393158,
+ "learning_rate": 1.2560296418320239e-08,
+ "loss": 0.0065,
+ "step": 36132
+ },
+ {
+ "epoch": 98.4550408719346,
+ "grad_norm": 0.4710390865802765,
+ "learning_rate": 1.2516118174071834e-08,
+ "loss": 0.0047,
+ "step": 36133
+ },
+ {
+ "epoch": 98.45776566757493,
+ "grad_norm": 0.7132671475410461,
+ "learning_rate": 1.2472017712027439e-08,
+ "loss": 0.007,
+ "step": 36134
+ },
+ {
+ "epoch": 98.46049046321527,
+ "grad_norm": 0.9714264273643494,
+ "learning_rate": 1.2427995032527895e-08,
+ "loss": 0.0086,
+ "step": 36135
+ },
+ {
+ "epoch": 98.46321525885558,
+ "grad_norm": 0.9833755493164062,
+ "learning_rate": 1.238405013591848e-08,
+ "loss": 0.0258,
+ "step": 36136
+ },
+ {
+ "epoch": 98.46594005449592,
+ "grad_norm": 0.7575590014457703,
+ "learning_rate": 1.2340183022541142e-08,
+ "loss": 0.0074,
+ "step": 36137
+ },
+ {
+ "epoch": 98.46866485013624,
+ "grad_norm": 0.4835565686225891,
+ "learning_rate": 1.2296393692735608e-08,
+ "loss": 0.0046,
+ "step": 36138
+ },
+ {
+ "epoch": 98.47138964577657,
+ "grad_norm": 0.8626381158828735,
+ "learning_rate": 1.225268214684494e-08,
+ "loss": 0.0327,
+ "step": 36139
+ },
+ {
+ "epoch": 98.47411444141689,
+ "grad_norm": 1.0745208263397217,
+ "learning_rate": 1.2209048385209977e-08,
+ "loss": 0.0379,
+ "step": 36140
+ },
+ {
+ "epoch": 98.47683923705722,
+ "grad_norm": 1.033342957496643,
+ "learning_rate": 1.2165492408168223e-08,
+ "loss": 0.0133,
+ "step": 36141
+ },
+ {
+ "epoch": 98.47956403269755,
+ "grad_norm": 1.2851531505584717,
+ "learning_rate": 1.2122014216060518e-08,
+ "loss": 0.0595,
+ "step": 36142
+ },
+ {
+ "epoch": 98.48228882833787,
+ "grad_norm": 1.4943434000015259,
+ "learning_rate": 1.2078613809225482e-08,
+ "loss": 0.0234,
+ "step": 36143
+ },
+ {
+ "epoch": 98.4850136239782,
+ "grad_norm": 0.9420099854469299,
+ "learning_rate": 1.2035291187999509e-08,
+ "loss": 0.009,
+ "step": 36144
+ },
+ {
+ "epoch": 98.48773841961852,
+ "grad_norm": 0.7680992484092712,
+ "learning_rate": 1.199204635272233e-08,
+ "loss": 0.004,
+ "step": 36145
+ },
+ {
+ "epoch": 98.49046321525886,
+ "grad_norm": 0.5444515347480774,
+ "learning_rate": 1.1948879303729233e-08,
+ "loss": 0.0033,
+ "step": 36146
+ },
+ {
+ "epoch": 98.49318801089919,
+ "grad_norm": 0.8455659747123718,
+ "learning_rate": 1.1905790041357724e-08,
+ "loss": 0.0083,
+ "step": 36147
+ },
+ {
+ "epoch": 98.49591280653951,
+ "grad_norm": 0.6218907833099365,
+ "learning_rate": 1.1862778565940869e-08,
+ "loss": 0.004,
+ "step": 36148
+ },
+ {
+ "epoch": 98.49863760217984,
+ "grad_norm": 0.8836853504180908,
+ "learning_rate": 1.1819844877816178e-08,
+ "loss": 0.0118,
+ "step": 36149
+ },
+ {
+ "epoch": 98.50136239782016,
+ "grad_norm": 3.733849048614502,
+ "learning_rate": 1.1776988977316717e-08,
+ "loss": 0.0171,
+ "step": 36150
+ },
+ {
+ "epoch": 98.50408719346049,
+ "grad_norm": 1.6196726560592651,
+ "learning_rate": 1.1734210864777773e-08,
+ "loss": 0.0995,
+ "step": 36151
+ },
+ {
+ "epoch": 98.50681198910081,
+ "grad_norm": 0.9799879193305969,
+ "learning_rate": 1.1691510540530193e-08,
+ "loss": 0.0198,
+ "step": 36152
+ },
+ {
+ "epoch": 98.50953678474114,
+ "grad_norm": 0.7036740183830261,
+ "learning_rate": 1.1648888004907043e-08,
+ "loss": 0.0043,
+ "step": 36153
+ },
+ {
+ "epoch": 98.51226158038148,
+ "grad_norm": 0.8473700284957886,
+ "learning_rate": 1.1606343258242503e-08,
+ "loss": 0.0846,
+ "step": 36154
+ },
+ {
+ "epoch": 98.5149863760218,
+ "grad_norm": 1.0729087591171265,
+ "learning_rate": 1.1563876300865195e-08,
+ "loss": 0.0119,
+ "step": 36155
+ },
+ {
+ "epoch": 98.51771117166213,
+ "grad_norm": 0.9033671021461487,
+ "learning_rate": 1.152148713310708e-08,
+ "loss": 0.0209,
+ "step": 36156
+ },
+ {
+ "epoch": 98.52043596730245,
+ "grad_norm": 0.8044594526290894,
+ "learning_rate": 1.1479175755297889e-08,
+ "loss": 0.0187,
+ "step": 36157
+ },
+ {
+ "epoch": 98.52316076294278,
+ "grad_norm": 1.1266852617263794,
+ "learning_rate": 1.1436942167768472e-08,
+ "loss": 0.0272,
+ "step": 36158
+ },
+ {
+ "epoch": 98.52588555858311,
+ "grad_norm": 1.1029819250106812,
+ "learning_rate": 1.1394786370845234e-08,
+ "loss": 0.0621,
+ "step": 36159
+ },
+ {
+ "epoch": 98.52861035422343,
+ "grad_norm": 0.5430566668510437,
+ "learning_rate": 1.1352708364859022e-08,
+ "loss": 0.0039,
+ "step": 36160
+ },
+ {
+ "epoch": 98.53133514986376,
+ "grad_norm": 0.6413061618804932,
+ "learning_rate": 1.131070815013624e-08,
+ "loss": 0.0056,
+ "step": 36161
+ },
+ {
+ "epoch": 98.53405994550408,
+ "grad_norm": 0.843063473701477,
+ "learning_rate": 1.1268785727004406e-08,
+ "loss": 0.1334,
+ "step": 36162
+ },
+ {
+ "epoch": 98.53678474114442,
+ "grad_norm": 0.8860489130020142,
+ "learning_rate": 1.1226941095788813e-08,
+ "loss": 0.012,
+ "step": 36163
+ },
+ {
+ "epoch": 98.53950953678473,
+ "grad_norm": 0.6258013248443604,
+ "learning_rate": 1.1185174256816978e-08,
+ "loss": 0.0047,
+ "step": 36164
+ },
+ {
+ "epoch": 98.54223433242507,
+ "grad_norm": 0.380521297454834,
+ "learning_rate": 1.1143485210413086e-08,
+ "loss": 0.0043,
+ "step": 36165
+ },
+ {
+ "epoch": 98.5449591280654,
+ "grad_norm": 0.9789325594902039,
+ "learning_rate": 1.1101873956902431e-08,
+ "loss": 0.021,
+ "step": 36166
+ },
+ {
+ "epoch": 98.54768392370572,
+ "grad_norm": 0.7188648581504822,
+ "learning_rate": 1.10603404966092e-08,
+ "loss": 0.0058,
+ "step": 36167
+ },
+ {
+ "epoch": 98.55040871934605,
+ "grad_norm": 0.9840901494026184,
+ "learning_rate": 1.101888482985647e-08,
+ "loss": 0.0137,
+ "step": 36168
+ },
+ {
+ "epoch": 98.55313351498637,
+ "grad_norm": 0.727048933506012,
+ "learning_rate": 1.0977506956967309e-08,
+ "loss": 0.0062,
+ "step": 36169
+ },
+ {
+ "epoch": 98.5558583106267,
+ "grad_norm": 0.659720778465271,
+ "learning_rate": 1.093620687826369e-08,
+ "loss": 0.0088,
+ "step": 36170
+ },
+ {
+ "epoch": 98.55858310626704,
+ "grad_norm": 1.043558120727539,
+ "learning_rate": 1.0894984594067571e-08,
+ "loss": 0.013,
+ "step": 36171
+ },
+ {
+ "epoch": 98.56130790190736,
+ "grad_norm": 0.9254286885261536,
+ "learning_rate": 1.0853840104698698e-08,
+ "loss": 0.0103,
+ "step": 36172
+ },
+ {
+ "epoch": 98.56403269754769,
+ "grad_norm": 1.436519742012024,
+ "learning_rate": 1.0812773410480148e-08,
+ "loss": 0.0131,
+ "step": 36173
+ },
+ {
+ "epoch": 98.566757493188,
+ "grad_norm": 1.1371173858642578,
+ "learning_rate": 1.0771784511729444e-08,
+ "loss": 0.0207,
+ "step": 36174
+ },
+ {
+ "epoch": 98.56948228882834,
+ "grad_norm": 0.7689312696456909,
+ "learning_rate": 1.0730873408766329e-08,
+ "loss": 0.0074,
+ "step": 36175
+ },
+ {
+ "epoch": 98.57220708446866,
+ "grad_norm": 1.797876000404358,
+ "learning_rate": 1.0690040101910549e-08,
+ "loss": 0.0326,
+ "step": 36176
+ },
+ {
+ "epoch": 98.57493188010899,
+ "grad_norm": 1.058950662612915,
+ "learning_rate": 1.0649284591478514e-08,
+ "loss": 0.0204,
+ "step": 36177
+ },
+ {
+ "epoch": 98.57765667574932,
+ "grad_norm": 1.281187891960144,
+ "learning_rate": 1.0608606877788863e-08,
+ "loss": 0.0373,
+ "step": 36178
+ },
+ {
+ "epoch": 98.58038147138964,
+ "grad_norm": 1.2735719680786133,
+ "learning_rate": 1.0568006961156895e-08,
+ "loss": 0.0088,
+ "step": 36179
+ },
+ {
+ "epoch": 98.58310626702998,
+ "grad_norm": 0.6863452792167664,
+ "learning_rate": 1.0527484841900137e-08,
+ "loss": 0.0073,
+ "step": 36180
+ },
+ {
+ "epoch": 98.5858310626703,
+ "grad_norm": 1.0522154569625854,
+ "learning_rate": 1.0487040520335e-08,
+ "loss": 0.0085,
+ "step": 36181
+ },
+ {
+ "epoch": 98.58855585831063,
+ "grad_norm": 0.8815805315971375,
+ "learning_rate": 1.044667399677457e-08,
+ "loss": 0.0074,
+ "step": 36182
+ },
+ {
+ "epoch": 98.59128065395096,
+ "grad_norm": 0.9726758003234863,
+ "learning_rate": 1.0406385271533037e-08,
+ "loss": 0.0311,
+ "step": 36183
+ },
+ {
+ "epoch": 98.59400544959128,
+ "grad_norm": 0.8971498012542725,
+ "learning_rate": 1.0366174344925705e-08,
+ "loss": 0.0133,
+ "step": 36184
+ },
+ {
+ "epoch": 98.59673024523161,
+ "grad_norm": 1.45903742313385,
+ "learning_rate": 1.0326041217265659e-08,
+ "loss": 0.1474,
+ "step": 36185
+ },
+ {
+ "epoch": 98.59945504087193,
+ "grad_norm": 0.6291977167129517,
+ "learning_rate": 1.0285985888863759e-08,
+ "loss": 0.0058,
+ "step": 36186
+ },
+ {
+ "epoch": 98.60217983651226,
+ "grad_norm": 0.9251425266265869,
+ "learning_rate": 1.0246008360034199e-08,
+ "loss": 0.0096,
+ "step": 36187
+ },
+ {
+ "epoch": 98.60490463215258,
+ "grad_norm": 1.0490931272506714,
+ "learning_rate": 1.020610863108673e-08,
+ "loss": 0.0251,
+ "step": 36188
+ },
+ {
+ "epoch": 98.60762942779292,
+ "grad_norm": 0.7021046280860901,
+ "learning_rate": 1.0166286702332218e-08,
+ "loss": 0.0062,
+ "step": 36189
+ },
+ {
+ "epoch": 98.61035422343325,
+ "grad_norm": 0.7184568047523499,
+ "learning_rate": 1.0126542574081521e-08,
+ "loss": 0.0067,
+ "step": 36190
+ },
+ {
+ "epoch": 98.61307901907357,
+ "grad_norm": 1.149901270866394,
+ "learning_rate": 1.0086876246644395e-08,
+ "loss": 0.0292,
+ "step": 36191
+ },
+ {
+ "epoch": 98.6158038147139,
+ "grad_norm": 0.6910189986228943,
+ "learning_rate": 1.0047287720328369e-08,
+ "loss": 0.0089,
+ "step": 36192
+ },
+ {
+ "epoch": 98.61852861035422,
+ "grad_norm": 1.4245702028274536,
+ "learning_rate": 1.0007776995442087e-08,
+ "loss": 0.0096,
+ "step": 36193
+ },
+ {
+ "epoch": 98.62125340599455,
+ "grad_norm": 0.9353606104850769,
+ "learning_rate": 9.96834407229419e-09,
+ "loss": 0.0109,
+ "step": 36194
+ },
+ {
+ "epoch": 98.62397820163488,
+ "grad_norm": 0.770659863948822,
+ "learning_rate": 9.928988951192209e-09,
+ "loss": 0.0043,
+ "step": 36195
+ },
+ {
+ "epoch": 98.6267029972752,
+ "grad_norm": 0.8701274991035461,
+ "learning_rate": 9.889711632440346e-09,
+ "loss": 0.0103,
+ "step": 36196
+ },
+ {
+ "epoch": 98.62942779291554,
+ "grad_norm": 1.539430856704712,
+ "learning_rate": 9.850512116347245e-09,
+ "loss": 0.0748,
+ "step": 36197
+ },
+ {
+ "epoch": 98.63215258855585,
+ "grad_norm": 0.8806151747703552,
+ "learning_rate": 9.811390403215992e-09,
+ "loss": 0.0162,
+ "step": 36198
+ },
+ {
+ "epoch": 98.63487738419619,
+ "grad_norm": 0.9791736006736755,
+ "learning_rate": 9.772346493351902e-09,
+ "loss": 0.0103,
+ "step": 36199
+ },
+ {
+ "epoch": 98.6376021798365,
+ "grad_norm": 1.6379611492156982,
+ "learning_rate": 9.733380387059177e-09,
+ "loss": 0.1025,
+ "step": 36200
+ },
+ {
+ "epoch": 98.64032697547684,
+ "grad_norm": 2.62658953666687,
+ "learning_rate": 9.694492084642016e-09,
+ "loss": 0.0162,
+ "step": 36201
+ },
+ {
+ "epoch": 98.64305177111717,
+ "grad_norm": 1.1547921895980835,
+ "learning_rate": 9.6556815864024e-09,
+ "loss": 0.0149,
+ "step": 36202
+ },
+ {
+ "epoch": 98.64577656675749,
+ "grad_norm": 1.69525945186615,
+ "learning_rate": 9.616948892642307e-09,
+ "loss": 0.0933,
+ "step": 36203
+ },
+ {
+ "epoch": 98.64850136239782,
+ "grad_norm": 1.3838359117507935,
+ "learning_rate": 9.578294003664835e-09,
+ "loss": 0.0109,
+ "step": 36204
+ },
+ {
+ "epoch": 98.65122615803814,
+ "grad_norm": 1.7098582983016968,
+ "learning_rate": 9.539716919768626e-09,
+ "loss": 0.0468,
+ "step": 36205
+ },
+ {
+ "epoch": 98.65395095367847,
+ "grad_norm": 1.3063849210739136,
+ "learning_rate": 9.501217641256776e-09,
+ "loss": 0.0346,
+ "step": 36206
+ },
+ {
+ "epoch": 98.65667574931881,
+ "grad_norm": 1.0582362413406372,
+ "learning_rate": 9.462796168427935e-09,
+ "loss": 0.0479,
+ "step": 36207
+ },
+ {
+ "epoch": 98.65940054495913,
+ "grad_norm": 0.5250722169876099,
+ "learning_rate": 9.42445250158075e-09,
+ "loss": 0.0049,
+ "step": 36208
+ },
+ {
+ "epoch": 98.66212534059946,
+ "grad_norm": 0.7396645545959473,
+ "learning_rate": 9.386186641013872e-09,
+ "loss": 0.0041,
+ "step": 36209
+ },
+ {
+ "epoch": 98.66485013623978,
+ "grad_norm": 0.9204198122024536,
+ "learning_rate": 9.347998587027063e-09,
+ "loss": 0.0074,
+ "step": 36210
+ },
+ {
+ "epoch": 98.66757493188011,
+ "grad_norm": 1.198790431022644,
+ "learning_rate": 9.30988833991564e-09,
+ "loss": 0.0293,
+ "step": 36211
+ },
+ {
+ "epoch": 98.67029972752043,
+ "grad_norm": 0.7092175483703613,
+ "learning_rate": 9.271855899977144e-09,
+ "loss": 0.0072,
+ "step": 36212
+ },
+ {
+ "epoch": 98.67302452316076,
+ "grad_norm": 1.1241538524627686,
+ "learning_rate": 9.233901267509115e-09,
+ "loss": 0.0449,
+ "step": 36213
+ },
+ {
+ "epoch": 98.6757493188011,
+ "grad_norm": 1.1154484748840332,
+ "learning_rate": 9.196024442804652e-09,
+ "loss": 0.0136,
+ "step": 36214
+ },
+ {
+ "epoch": 98.67847411444141,
+ "grad_norm": 0.8924081325531006,
+ "learning_rate": 9.158225426160183e-09,
+ "loss": 0.0206,
+ "step": 36215
+ },
+ {
+ "epoch": 98.68119891008175,
+ "grad_norm": 1.8350354433059692,
+ "learning_rate": 9.120504217871029e-09,
+ "loss": 0.1084,
+ "step": 36216
+ },
+ {
+ "epoch": 98.68392370572207,
+ "grad_norm": 1.140293002128601,
+ "learning_rate": 9.082860818228068e-09,
+ "loss": 0.0467,
+ "step": 36217
+ },
+ {
+ "epoch": 98.6866485013624,
+ "grad_norm": 0.8952414989471436,
+ "learning_rate": 9.045295227527728e-09,
+ "loss": 0.0111,
+ "step": 36218
+ },
+ {
+ "epoch": 98.68937329700273,
+ "grad_norm": 1.2352014780044556,
+ "learning_rate": 9.007807446059779e-09,
+ "loss": 0.0566,
+ "step": 36219
+ },
+ {
+ "epoch": 98.69209809264305,
+ "grad_norm": 1.0734472274780273,
+ "learning_rate": 8.970397474118431e-09,
+ "loss": 0.0189,
+ "step": 36220
+ },
+ {
+ "epoch": 98.69482288828338,
+ "grad_norm": 1.1852258443832397,
+ "learning_rate": 8.93306531199345e-09,
+ "loss": 0.0082,
+ "step": 36221
+ },
+ {
+ "epoch": 98.6975476839237,
+ "grad_norm": 0.6034201383590698,
+ "learning_rate": 8.895810959976825e-09,
+ "loss": 0.0073,
+ "step": 36222
+ },
+ {
+ "epoch": 98.70027247956403,
+ "grad_norm": 0.8924165964126587,
+ "learning_rate": 8.858634418358325e-09,
+ "loss": 0.0107,
+ "step": 36223
+ },
+ {
+ "epoch": 98.70299727520435,
+ "grad_norm": 0.6823277473449707,
+ "learning_rate": 8.821535687426609e-09,
+ "loss": 0.0071,
+ "step": 36224
+ },
+ {
+ "epoch": 98.70572207084469,
+ "grad_norm": 1.4606468677520752,
+ "learning_rate": 8.784514767470331e-09,
+ "loss": 0.0347,
+ "step": 36225
+ },
+ {
+ "epoch": 98.70844686648502,
+ "grad_norm": 1.3690017461776733,
+ "learning_rate": 8.747571658780373e-09,
+ "loss": 0.0079,
+ "step": 36226
+ },
+ {
+ "epoch": 98.71117166212534,
+ "grad_norm": 1.0299593210220337,
+ "learning_rate": 8.710706361642063e-09,
+ "loss": 0.0137,
+ "step": 36227
+ },
+ {
+ "epoch": 98.71389645776567,
+ "grad_norm": 0.9563325047492981,
+ "learning_rate": 8.673918876342947e-09,
+ "loss": 0.0337,
+ "step": 36228
+ },
+ {
+ "epoch": 98.71662125340599,
+ "grad_norm": 0.6014434695243835,
+ "learning_rate": 8.63720920316946e-09,
+ "loss": 0.0096,
+ "step": 36229
+ },
+ {
+ "epoch": 98.71934604904632,
+ "grad_norm": 0.893867015838623,
+ "learning_rate": 8.600577342408046e-09,
+ "loss": 0.0229,
+ "step": 36230
+ },
+ {
+ "epoch": 98.72207084468666,
+ "grad_norm": 1.2934709787368774,
+ "learning_rate": 8.564023294344026e-09,
+ "loss": 0.0162,
+ "step": 36231
+ },
+ {
+ "epoch": 98.72479564032697,
+ "grad_norm": 1.280702829360962,
+ "learning_rate": 8.527547059262731e-09,
+ "loss": 0.1052,
+ "step": 36232
+ },
+ {
+ "epoch": 98.7275204359673,
+ "grad_norm": 0.6869642734527588,
+ "learning_rate": 8.491148637446156e-09,
+ "loss": 0.0079,
+ "step": 36233
+ },
+ {
+ "epoch": 98.73024523160763,
+ "grad_norm": 1.4125590324401855,
+ "learning_rate": 8.454828029178518e-09,
+ "loss": 0.0411,
+ "step": 36234
+ },
+ {
+ "epoch": 98.73297002724796,
+ "grad_norm": 1.0688910484313965,
+ "learning_rate": 8.418585234744037e-09,
+ "loss": 0.056,
+ "step": 36235
+ },
+ {
+ "epoch": 98.73569482288828,
+ "grad_norm": 0.5110642910003662,
+ "learning_rate": 8.382420254424706e-09,
+ "loss": 0.0053,
+ "step": 36236
+ },
+ {
+ "epoch": 98.73841961852861,
+ "grad_norm": 1.8262989521026611,
+ "learning_rate": 8.346333088500303e-09,
+ "loss": 0.0192,
+ "step": 36237
+ },
+ {
+ "epoch": 98.74114441416894,
+ "grad_norm": 1.0947142839431763,
+ "learning_rate": 8.310323737252824e-09,
+ "loss": 0.0063,
+ "step": 36238
+ },
+ {
+ "epoch": 98.74386920980926,
+ "grad_norm": 0.7630906701087952,
+ "learning_rate": 8.274392200964265e-09,
+ "loss": 0.0086,
+ "step": 36239
+ },
+ {
+ "epoch": 98.7465940054496,
+ "grad_norm": 1.1678982973098755,
+ "learning_rate": 8.238538479912184e-09,
+ "loss": 0.1213,
+ "step": 36240
+ },
+ {
+ "epoch": 98.74931880108991,
+ "grad_norm": 0.7399142384529114,
+ "learning_rate": 8.202762574376355e-09,
+ "loss": 0.0073,
+ "step": 36241
+ },
+ {
+ "epoch": 98.75204359673025,
+ "grad_norm": 0.8911531567573547,
+ "learning_rate": 8.167064484636555e-09,
+ "loss": 0.0171,
+ "step": 36242
+ },
+ {
+ "epoch": 98.75476839237058,
+ "grad_norm": 0.9478684067726135,
+ "learning_rate": 8.13144421097034e-09,
+ "loss": 0.0067,
+ "step": 36243
+ },
+ {
+ "epoch": 98.7574931880109,
+ "grad_norm": 0.886153519153595,
+ "learning_rate": 8.095901753654156e-09,
+ "loss": 0.0067,
+ "step": 36244
+ },
+ {
+ "epoch": 98.76021798365123,
+ "grad_norm": 1.3267852067947388,
+ "learning_rate": 8.060437112965558e-09,
+ "loss": 0.0157,
+ "step": 36245
+ },
+ {
+ "epoch": 98.76294277929155,
+ "grad_norm": 1.0634489059448242,
+ "learning_rate": 8.02505028918099e-09,
+ "loss": 0.0167,
+ "step": 36246
+ },
+ {
+ "epoch": 98.76566757493188,
+ "grad_norm": 1.4418994188308716,
+ "learning_rate": 7.989741282575792e-09,
+ "loss": 0.0418,
+ "step": 36247
+ },
+ {
+ "epoch": 98.7683923705722,
+ "grad_norm": 1.1179955005645752,
+ "learning_rate": 7.954510093425294e-09,
+ "loss": 0.0293,
+ "step": 36248
+ },
+ {
+ "epoch": 98.77111716621253,
+ "grad_norm": 0.8936944007873535,
+ "learning_rate": 7.919356722002613e-09,
+ "loss": 0.0071,
+ "step": 36249
+ },
+ {
+ "epoch": 98.77384196185287,
+ "grad_norm": 0.406175434589386,
+ "learning_rate": 7.884281168583086e-09,
+ "loss": 0.0039,
+ "step": 36250
+ },
+ {
+ "epoch": 98.77656675749319,
+ "grad_norm": 0.9114712476730347,
+ "learning_rate": 7.849283433438714e-09,
+ "loss": 0.0186,
+ "step": 36251
+ },
+ {
+ "epoch": 98.77929155313352,
+ "grad_norm": 0.8577885031700134,
+ "learning_rate": 7.814363516842616e-09,
+ "loss": 0.0113,
+ "step": 36252
+ },
+ {
+ "epoch": 98.78201634877384,
+ "grad_norm": 0.8739956617355347,
+ "learning_rate": 7.779521419066793e-09,
+ "loss": 0.0075,
+ "step": 36253
+ },
+ {
+ "epoch": 98.78474114441417,
+ "grad_norm": 1.1459224224090576,
+ "learning_rate": 7.744757140382141e-09,
+ "loss": 0.0143,
+ "step": 36254
+ },
+ {
+ "epoch": 98.7874659400545,
+ "grad_norm": 3.2036566734313965,
+ "learning_rate": 7.710070681060666e-09,
+ "loss": 0.0106,
+ "step": 36255
+ },
+ {
+ "epoch": 98.79019073569482,
+ "grad_norm": 0.6775052547454834,
+ "learning_rate": 7.67546204137104e-09,
+ "loss": 0.0075,
+ "step": 36256
+ },
+ {
+ "epoch": 98.79291553133515,
+ "grad_norm": 1.0119221210479736,
+ "learning_rate": 7.640931221583048e-09,
+ "loss": 0.0109,
+ "step": 36257
+ },
+ {
+ "epoch": 98.79564032697547,
+ "grad_norm": 1.5220680236816406,
+ "learning_rate": 7.606478221965363e-09,
+ "loss": 0.0153,
+ "step": 36258
+ },
+ {
+ "epoch": 98.7983651226158,
+ "grad_norm": 1.0919644832611084,
+ "learning_rate": 7.572103042787771e-09,
+ "loss": 0.0191,
+ "step": 36259
+ },
+ {
+ "epoch": 98.80108991825612,
+ "grad_norm": 1.2510679960250854,
+ "learning_rate": 7.537805684316724e-09,
+ "loss": 0.0236,
+ "step": 36260
+ },
+ {
+ "epoch": 98.80381471389646,
+ "grad_norm": 0.6117183566093445,
+ "learning_rate": 7.503586146819785e-09,
+ "loss": 0.0056,
+ "step": 36261
+ },
+ {
+ "epoch": 98.80653950953679,
+ "grad_norm": 0.8787217140197754,
+ "learning_rate": 7.4694444305623e-09,
+ "loss": 0.0112,
+ "step": 36262
+ },
+ {
+ "epoch": 98.80926430517711,
+ "grad_norm": 1.4968916177749634,
+ "learning_rate": 7.4353805358129416e-09,
+ "loss": 0.0259,
+ "step": 36263
+ },
+ {
+ "epoch": 98.81198910081744,
+ "grad_norm": 0.7979059815406799,
+ "learning_rate": 7.401394462833722e-09,
+ "loss": 0.0074,
+ "step": 36264
+ },
+ {
+ "epoch": 98.81471389645776,
+ "grad_norm": 1.2332895994186401,
+ "learning_rate": 7.3674862118922055e-09,
+ "loss": 0.0186,
+ "step": 36265
+ },
+ {
+ "epoch": 98.8174386920981,
+ "grad_norm": 1.1442842483520508,
+ "learning_rate": 7.333655783249294e-09,
+ "loss": 0.0254,
+ "step": 36266
+ },
+ {
+ "epoch": 98.82016348773843,
+ "grad_norm": 1.402367353439331,
+ "learning_rate": 7.299903177171441e-09,
+ "loss": 0.0313,
+ "step": 36267
+ },
+ {
+ "epoch": 98.82288828337875,
+ "grad_norm": 1.0101616382598877,
+ "learning_rate": 7.2662283939206604e-09,
+ "loss": 0.0059,
+ "step": 36268
+ },
+ {
+ "epoch": 98.82561307901908,
+ "grad_norm": 0.8787945508956909,
+ "learning_rate": 7.232631433757853e-09,
+ "loss": 0.0081,
+ "step": 36269
+ },
+ {
+ "epoch": 98.8283378746594,
+ "grad_norm": 1.5436629056930542,
+ "learning_rate": 7.1991122969461425e-09,
+ "loss": 0.0172,
+ "step": 36270
+ },
+ {
+ "epoch": 98.83106267029973,
+ "grad_norm": 1.0108474493026733,
+ "learning_rate": 7.165670983746431e-09,
+ "loss": 0.0092,
+ "step": 36271
+ },
+ {
+ "epoch": 98.83378746594005,
+ "grad_norm": 1.1141397953033447,
+ "learning_rate": 7.132307494418511e-09,
+ "loss": 0.0092,
+ "step": 36272
+ },
+ {
+ "epoch": 98.83651226158038,
+ "grad_norm": 1.3692843914031982,
+ "learning_rate": 7.099021829223285e-09,
+ "loss": 0.0178,
+ "step": 36273
+ },
+ {
+ "epoch": 98.83923705722071,
+ "grad_norm": 0.9389157295227051,
+ "learning_rate": 7.065813988418324e-09,
+ "loss": 0.0142,
+ "step": 36274
+ },
+ {
+ "epoch": 98.84196185286103,
+ "grad_norm": 1.727347731590271,
+ "learning_rate": 7.032683972263421e-09,
+ "loss": 0.0132,
+ "step": 36275
+ },
+ {
+ "epoch": 98.84468664850137,
+ "grad_norm": 0.8405441045761108,
+ "learning_rate": 6.999631781017258e-09,
+ "loss": 0.0038,
+ "step": 36276
+ },
+ {
+ "epoch": 98.84741144414168,
+ "grad_norm": 0.7177353501319885,
+ "learning_rate": 6.966657414935185e-09,
+ "loss": 0.0099,
+ "step": 36277
+ },
+ {
+ "epoch": 98.85013623978202,
+ "grad_norm": 1.0587660074234009,
+ "learning_rate": 6.933760874274775e-09,
+ "loss": 0.0139,
+ "step": 36278
+ },
+ {
+ "epoch": 98.85286103542235,
+ "grad_norm": 0.7917664051055908,
+ "learning_rate": 6.9009421592936e-09,
+ "loss": 0.0066,
+ "step": 36279
+ },
+ {
+ "epoch": 98.85558583106267,
+ "grad_norm": 0.93915855884552,
+ "learning_rate": 6.8682012702459e-09,
+ "loss": 0.023,
+ "step": 36280
+ },
+ {
+ "epoch": 98.858310626703,
+ "grad_norm": 0.7826685309410095,
+ "learning_rate": 6.835538207387027e-09,
+ "loss": 0.0047,
+ "step": 36281
+ },
+ {
+ "epoch": 98.86103542234332,
+ "grad_norm": 0.6528874039649963,
+ "learning_rate": 6.802952970971222e-09,
+ "loss": 0.0081,
+ "step": 36282
+ },
+ {
+ "epoch": 98.86376021798365,
+ "grad_norm": 0.8234850764274597,
+ "learning_rate": 6.770445561252725e-09,
+ "loss": 0.0118,
+ "step": 36283
+ },
+ {
+ "epoch": 98.86648501362397,
+ "grad_norm": 1.1198320388793945,
+ "learning_rate": 6.738015978484669e-09,
+ "loss": 0.009,
+ "step": 36284
+ },
+ {
+ "epoch": 98.8692098092643,
+ "grad_norm": 1.4291325807571411,
+ "learning_rate": 6.705664222919072e-09,
+ "loss": 0.0204,
+ "step": 36285
+ },
+ {
+ "epoch": 98.87193460490464,
+ "grad_norm": 0.7126988172531128,
+ "learning_rate": 6.673390294807958e-09,
+ "loss": 0.0282,
+ "step": 36286
+ },
+ {
+ "epoch": 98.87465940054496,
+ "grad_norm": 0.7710235714912415,
+ "learning_rate": 6.641194194402234e-09,
+ "loss": 0.0065,
+ "step": 36287
+ },
+ {
+ "epoch": 98.87738419618529,
+ "grad_norm": 0.9626188278198242,
+ "learning_rate": 6.609075921953923e-09,
+ "loss": 0.0653,
+ "step": 36288
+ },
+ {
+ "epoch": 98.88010899182561,
+ "grad_norm": 0.7661487460136414,
+ "learning_rate": 6.577035477712823e-09,
+ "loss": 0.009,
+ "step": 36289
+ },
+ {
+ "epoch": 98.88283378746594,
+ "grad_norm": 1.2224106788635254,
+ "learning_rate": 6.545072861927626e-09,
+ "loss": 0.0218,
+ "step": 36290
+ },
+ {
+ "epoch": 98.88555858310627,
+ "grad_norm": 1.1829947233200073,
+ "learning_rate": 6.5131880748481314e-09,
+ "loss": 0.0506,
+ "step": 36291
+ },
+ {
+ "epoch": 98.88828337874659,
+ "grad_norm": 1.7935855388641357,
+ "learning_rate": 6.481381116721919e-09,
+ "loss": 0.0379,
+ "step": 36292
+ },
+ {
+ "epoch": 98.89100817438693,
+ "grad_norm": 0.7636347413063049,
+ "learning_rate": 6.449651987797678e-09,
+ "loss": 0.0102,
+ "step": 36293
+ },
+ {
+ "epoch": 98.89373297002724,
+ "grad_norm": 1.100306510925293,
+ "learning_rate": 6.418000688320769e-09,
+ "loss": 0.0124,
+ "step": 36294
+ },
+ {
+ "epoch": 98.89645776566758,
+ "grad_norm": 0.5303922295570374,
+ "learning_rate": 6.386427218539881e-09,
+ "loss": 0.006,
+ "step": 36295
+ },
+ {
+ "epoch": 98.8991825613079,
+ "grad_norm": 0.7615512013435364,
+ "learning_rate": 6.354931578699264e-09,
+ "loss": 0.0094,
+ "step": 36296
+ },
+ {
+ "epoch": 98.90190735694823,
+ "grad_norm": 0.6857603788375854,
+ "learning_rate": 6.323513769045386e-09,
+ "loss": 0.0077,
+ "step": 36297
+ },
+ {
+ "epoch": 98.90463215258856,
+ "grad_norm": 2.89074969291687,
+ "learning_rate": 6.292173789821387e-09,
+ "loss": 0.0165,
+ "step": 36298
+ },
+ {
+ "epoch": 98.90735694822888,
+ "grad_norm": 0.7204524278640747,
+ "learning_rate": 6.260911641272626e-09,
+ "loss": 0.0159,
+ "step": 36299
+ },
+ {
+ "epoch": 98.91008174386921,
+ "grad_norm": 0.929356038570404,
+ "learning_rate": 6.229727323643353e-09,
+ "loss": 0.1005,
+ "step": 36300
+ },
+ {
+ "epoch": 98.91280653950953,
+ "grad_norm": 1.1123864650726318,
+ "learning_rate": 6.198620837174485e-09,
+ "loss": 0.0077,
+ "step": 36301
+ },
+ {
+ "epoch": 98.91553133514986,
+ "grad_norm": 2.401057481765747,
+ "learning_rate": 6.167592182108051e-09,
+ "loss": 0.0105,
+ "step": 36302
+ },
+ {
+ "epoch": 98.9182561307902,
+ "grad_norm": 2.0445261001586914,
+ "learning_rate": 6.1366413586883e-09,
+ "loss": 0.0639,
+ "step": 36303
+ },
+ {
+ "epoch": 98.92098092643052,
+ "grad_norm": 1.4909777641296387,
+ "learning_rate": 6.10576836715393e-09,
+ "loss": 0.0159,
+ "step": 36304
+ },
+ {
+ "epoch": 98.92370572207085,
+ "grad_norm": 0.7275319695472717,
+ "learning_rate": 6.07497320774697e-09,
+ "loss": 0.0116,
+ "step": 36305
+ },
+ {
+ "epoch": 98.92643051771117,
+ "grad_norm": 1.4214550256729126,
+ "learning_rate": 6.044255880705008e-09,
+ "loss": 0.0903,
+ "step": 36306
+ },
+ {
+ "epoch": 98.9291553133515,
+ "grad_norm": 0.41992610692977905,
+ "learning_rate": 6.013616386270071e-09,
+ "loss": 0.0032,
+ "step": 36307
+ },
+ {
+ "epoch": 98.93188010899182,
+ "grad_norm": 0.3194815218448639,
+ "learning_rate": 5.983054724679749e-09,
+ "loss": 0.0033,
+ "step": 36308
+ },
+ {
+ "epoch": 98.93460490463215,
+ "grad_norm": 2.5367431640625,
+ "learning_rate": 5.952570896170518e-09,
+ "loss": 0.0129,
+ "step": 36309
+ },
+ {
+ "epoch": 98.93732970027249,
+ "grad_norm": 1.0532137155532837,
+ "learning_rate": 5.9221649009810756e-09,
+ "loss": 0.0111,
+ "step": 36310
+ },
+ {
+ "epoch": 98.9400544959128,
+ "grad_norm": 1.4583449363708496,
+ "learning_rate": 5.891836739349011e-09,
+ "loss": 0.016,
+ "step": 36311
+ },
+ {
+ "epoch": 98.94277929155314,
+ "grad_norm": 1.2258121967315674,
+ "learning_rate": 5.86158641150969e-09,
+ "loss": 0.0086,
+ "step": 36312
+ },
+ {
+ "epoch": 98.94550408719346,
+ "grad_norm": 0.9719292521476746,
+ "learning_rate": 5.83141391769737e-09,
+ "loss": 0.0126,
+ "step": 36313
+ },
+ {
+ "epoch": 98.94822888283379,
+ "grad_norm": 0.3590724468231201,
+ "learning_rate": 5.80131925814964e-09,
+ "loss": 0.0036,
+ "step": 36314
+ },
+ {
+ "epoch": 98.95095367847412,
+ "grad_norm": 0.7638126611709595,
+ "learning_rate": 5.771302433098536e-09,
+ "loss": 0.0093,
+ "step": 36315
+ },
+ {
+ "epoch": 98.95367847411444,
+ "grad_norm": 1.1635639667510986,
+ "learning_rate": 5.741363442779424e-09,
+ "loss": 0.0223,
+ "step": 36316
+ },
+ {
+ "epoch": 98.95640326975477,
+ "grad_norm": 1.4873956441879272,
+ "learning_rate": 5.711502287425452e-09,
+ "loss": 0.0244,
+ "step": 36317
+ },
+ {
+ "epoch": 98.95912806539509,
+ "grad_norm": 0.7235299944877625,
+ "learning_rate": 5.6817189672675464e-09,
+ "loss": 0.0068,
+ "step": 36318
+ },
+ {
+ "epoch": 98.96185286103542,
+ "grad_norm": 0.9464858770370483,
+ "learning_rate": 5.652013482539964e-09,
+ "loss": 0.0596,
+ "step": 36319
+ },
+ {
+ "epoch": 98.96457765667574,
+ "grad_norm": 0.4427625834941864,
+ "learning_rate": 5.6223858334725215e-09,
+ "loss": 0.0051,
+ "step": 36320
+ },
+ {
+ "epoch": 98.96730245231608,
+ "grad_norm": 0.9278931617736816,
+ "learning_rate": 5.592836020295034e-09,
+ "loss": 0.0177,
+ "step": 36321
+ },
+ {
+ "epoch": 98.97002724795641,
+ "grad_norm": 0.7989128232002258,
+ "learning_rate": 5.563364043240649e-09,
+ "loss": 0.0065,
+ "step": 36322
+ },
+ {
+ "epoch": 98.97275204359673,
+ "grad_norm": 1.185469388961792,
+ "learning_rate": 5.533969902535852e-09,
+ "loss": 0.0782,
+ "step": 36323
+ },
+ {
+ "epoch": 98.97547683923706,
+ "grad_norm": 1.0109432935714722,
+ "learning_rate": 5.504653598411569e-09,
+ "loss": 0.0112,
+ "step": 36324
+ },
+ {
+ "epoch": 98.97820163487738,
+ "grad_norm": 0.8040210008621216,
+ "learning_rate": 5.475415131094286e-09,
+ "loss": 0.0074,
+ "step": 36325
+ },
+ {
+ "epoch": 98.98092643051771,
+ "grad_norm": 1.7476675510406494,
+ "learning_rate": 5.446254500813819e-09,
+ "loss": 0.0412,
+ "step": 36326
+ },
+ {
+ "epoch": 98.98365122615803,
+ "grad_norm": 1.0177794694900513,
+ "learning_rate": 5.417171707795544e-09,
+ "loss": 0.0184,
+ "step": 36327
+ },
+ {
+ "epoch": 98.98637602179836,
+ "grad_norm": 0.9335569143295288,
+ "learning_rate": 5.388166752267054e-09,
+ "loss": 0.0105,
+ "step": 36328
+ },
+ {
+ "epoch": 98.9891008174387,
+ "grad_norm": 1.6519578695297241,
+ "learning_rate": 5.359239634454838e-09,
+ "loss": 0.0125,
+ "step": 36329
+ },
+ {
+ "epoch": 98.99182561307902,
+ "grad_norm": 1.102130651473999,
+ "learning_rate": 5.330390354582049e-09,
+ "loss": 0.0571,
+ "step": 36330
+ },
+ {
+ "epoch": 98.99455040871935,
+ "grad_norm": 1.4858750104904175,
+ "learning_rate": 5.301618912874062e-09,
+ "loss": 0.0293,
+ "step": 36331
+ },
+ {
+ "epoch": 98.99727520435967,
+ "grad_norm": 2.1924631595611572,
+ "learning_rate": 5.2729253095562535e-09,
+ "loss": 0.1392,
+ "step": 36332
+ },
+ {
+ "epoch": 99.0,
+ "grad_norm": 0.5974382758140564,
+ "learning_rate": 5.2443095448506674e-09,
+ "loss": 0.0089,
+ "step": 36333
+ },
+ {
+ "epoch": 99.00272479564033,
+ "grad_norm": 1.055482029914856,
+ "learning_rate": 5.2157716189815685e-09,
+ "loss": 0.0077,
+ "step": 36334
+ },
+ {
+ "epoch": 99.00544959128065,
+ "grad_norm": 0.8504883646965027,
+ "learning_rate": 5.1873115321698916e-09,
+ "loss": 0.0069,
+ "step": 36335
+ },
+ {
+ "epoch": 99.00817438692098,
+ "grad_norm": 1.258363127708435,
+ "learning_rate": 5.158929284637682e-09,
+ "loss": 0.0064,
+ "step": 36336
+ },
+ {
+ "epoch": 99.0108991825613,
+ "grad_norm": 1.819697380065918,
+ "learning_rate": 5.130624876605872e-09,
+ "loss": 0.0388,
+ "step": 36337
+ },
+ {
+ "epoch": 99.01362397820164,
+ "grad_norm": 1.5627843141555786,
+ "learning_rate": 5.102398308295398e-09,
+ "loss": 0.0394,
+ "step": 36338
+ },
+ {
+ "epoch": 99.01634877384195,
+ "grad_norm": 0.917741596698761,
+ "learning_rate": 5.074249579924972e-09,
+ "loss": 0.0254,
+ "step": 36339
+ },
+ {
+ "epoch": 99.01907356948229,
+ "grad_norm": 0.5929439067840576,
+ "learning_rate": 5.046178691715531e-09,
+ "loss": 0.0066,
+ "step": 36340
+ },
+ {
+ "epoch": 99.02179836512262,
+ "grad_norm": 0.6242928504943848,
+ "learning_rate": 5.018185643884676e-09,
+ "loss": 0.0089,
+ "step": 36341
+ },
+ {
+ "epoch": 99.02452316076294,
+ "grad_norm": 0.6476610898971558,
+ "learning_rate": 4.9902704366500135e-09,
+ "loss": 0.0062,
+ "step": 36342
+ },
+ {
+ "epoch": 99.02724795640327,
+ "grad_norm": 0.944577693939209,
+ "learning_rate": 4.9624330702302544e-09,
+ "loss": 0.0129,
+ "step": 36343
+ },
+ {
+ "epoch": 99.02997275204359,
+ "grad_norm": 1.3255552053451538,
+ "learning_rate": 4.9346735448407845e-09,
+ "loss": 0.0146,
+ "step": 36344
+ },
+ {
+ "epoch": 99.03269754768392,
+ "grad_norm": 1.048995852470398,
+ "learning_rate": 4.906991860698096e-09,
+ "loss": 0.013,
+ "step": 36345
+ },
+ {
+ "epoch": 99.03542234332426,
+ "grad_norm": 0.6965101957321167,
+ "learning_rate": 4.879388018018683e-09,
+ "loss": 0.0061,
+ "step": 36346
+ },
+ {
+ "epoch": 99.03814713896458,
+ "grad_norm": 0.7452976107597351,
+ "learning_rate": 4.8518620170168175e-09,
+ "loss": 0.0062,
+ "step": 36347
+ },
+ {
+ "epoch": 99.04087193460491,
+ "grad_norm": 1.1007194519042969,
+ "learning_rate": 4.824413857906773e-09,
+ "loss": 0.0618,
+ "step": 36348
+ },
+ {
+ "epoch": 99.04359673024523,
+ "grad_norm": 2.5798962116241455,
+ "learning_rate": 4.7970435409017136e-09,
+ "loss": 0.0155,
+ "step": 36349
+ },
+ {
+ "epoch": 99.04632152588556,
+ "grad_norm": 1.0747082233428955,
+ "learning_rate": 4.769751066217021e-09,
+ "loss": 0.0109,
+ "step": 36350
+ },
+ {
+ "epoch": 99.04904632152588,
+ "grad_norm": 1.0944069623947144,
+ "learning_rate": 4.742536434062528e-09,
+ "loss": 0.0196,
+ "step": 36351
+ },
+ {
+ "epoch": 99.05177111716621,
+ "grad_norm": 1.1840583086013794,
+ "learning_rate": 4.715399644652508e-09,
+ "loss": 0.0147,
+ "step": 36352
+ },
+ {
+ "epoch": 99.05449591280654,
+ "grad_norm": 0.858555257320404,
+ "learning_rate": 4.6883406981967915e-09,
+ "loss": 0.0107,
+ "step": 36353
+ },
+ {
+ "epoch": 99.05722070844686,
+ "grad_norm": 0.9485266208648682,
+ "learning_rate": 4.661359594906323e-09,
+ "loss": 0.0339,
+ "step": 36354
+ },
+ {
+ "epoch": 99.0599455040872,
+ "grad_norm": 0.8904531598091125,
+ "learning_rate": 4.634456334992044e-09,
+ "loss": 0.0123,
+ "step": 36355
+ },
+ {
+ "epoch": 99.06267029972751,
+ "grad_norm": 1.0079540014266968,
+ "learning_rate": 4.607630918662676e-09,
+ "loss": 0.0287,
+ "step": 36356
+ },
+ {
+ "epoch": 99.06539509536785,
+ "grad_norm": 1.028802752494812,
+ "learning_rate": 4.580883346128051e-09,
+ "loss": 0.0077,
+ "step": 36357
+ },
+ {
+ "epoch": 99.06811989100818,
+ "grad_norm": 0.9676283597946167,
+ "learning_rate": 4.554213617594672e-09,
+ "loss": 0.011,
+ "step": 36358
+ },
+ {
+ "epoch": 99.0708446866485,
+ "grad_norm": 1.059435248374939,
+ "learning_rate": 4.527621733271259e-09,
+ "loss": 0.0613,
+ "step": 36359
+ },
+ {
+ "epoch": 99.07356948228883,
+ "grad_norm": 0.5957311987876892,
+ "learning_rate": 4.5011076933665356e-09,
+ "loss": 0.0051,
+ "step": 36360
+ },
+ {
+ "epoch": 99.07629427792915,
+ "grad_norm": 1.3893128633499146,
+ "learning_rate": 4.474671498083671e-09,
+ "loss": 0.02,
+ "step": 36361
+ },
+ {
+ "epoch": 99.07901907356948,
+ "grad_norm": 0.9166763424873352,
+ "learning_rate": 4.448313147631389e-09,
+ "loss": 0.0227,
+ "step": 36362
+ },
+ {
+ "epoch": 99.0817438692098,
+ "grad_norm": 3.4075543880462646,
+ "learning_rate": 4.422032642213969e-09,
+ "loss": 0.012,
+ "step": 36363
+ },
+ {
+ "epoch": 99.08446866485014,
+ "grad_norm": 0.9031820893287659,
+ "learning_rate": 4.395829982035693e-09,
+ "loss": 0.0182,
+ "step": 36364
+ },
+ {
+ "epoch": 99.08719346049047,
+ "grad_norm": 1.0828522443771362,
+ "learning_rate": 4.369705167301952e-09,
+ "loss": 0.0433,
+ "step": 36365
+ },
+ {
+ "epoch": 99.08991825613079,
+ "grad_norm": 0.6740212440490723,
+ "learning_rate": 4.343658198213696e-09,
+ "loss": 0.0094,
+ "step": 36366
+ },
+ {
+ "epoch": 99.09264305177112,
+ "grad_norm": 0.9212453365325928,
+ "learning_rate": 4.317689074976317e-09,
+ "loss": 0.0071,
+ "step": 36367
+ },
+ {
+ "epoch": 99.09536784741144,
+ "grad_norm": 0.938154399394989,
+ "learning_rate": 4.291797797790764e-09,
+ "loss": 0.0083,
+ "step": 36368
+ },
+ {
+ "epoch": 99.09809264305177,
+ "grad_norm": 2.3194282054901123,
+ "learning_rate": 4.2659843668591e-09,
+ "loss": 0.0508,
+ "step": 36369
+ },
+ {
+ "epoch": 99.1008174386921,
+ "grad_norm": 1.303553581237793,
+ "learning_rate": 4.240248782382272e-09,
+ "loss": 0.0232,
+ "step": 36370
+ },
+ {
+ "epoch": 99.10354223433242,
+ "grad_norm": 1.073468565940857,
+ "learning_rate": 4.214591044560123e-09,
+ "loss": 0.0121,
+ "step": 36371
+ },
+ {
+ "epoch": 99.10626702997276,
+ "grad_norm": 1.0172368288040161,
+ "learning_rate": 4.189011153593603e-09,
+ "loss": 0.0175,
+ "step": 36372
+ },
+ {
+ "epoch": 99.10899182561307,
+ "grad_norm": 1.8318792581558228,
+ "learning_rate": 4.163509109681441e-09,
+ "loss": 0.0098,
+ "step": 36373
+ },
+ {
+ "epoch": 99.11171662125341,
+ "grad_norm": 1.0358262062072754,
+ "learning_rate": 4.1380849130223666e-09,
+ "loss": 0.0252,
+ "step": 36374
+ },
+ {
+ "epoch": 99.11444141689373,
+ "grad_norm": 0.9365017414093018,
+ "learning_rate": 4.112738563814e-09,
+ "loss": 0.0658,
+ "step": 36375
+ },
+ {
+ "epoch": 99.11716621253406,
+ "grad_norm": 1.2956022024154663,
+ "learning_rate": 4.087470062252852e-09,
+ "loss": 0.0229,
+ "step": 36376
+ },
+ {
+ "epoch": 99.11989100817439,
+ "grad_norm": 1.111101508140564,
+ "learning_rate": 4.06227940853765e-09,
+ "loss": 0.0358,
+ "step": 36377
+ },
+ {
+ "epoch": 99.12261580381471,
+ "grad_norm": 1.3386849164962769,
+ "learning_rate": 4.0371666028637954e-09,
+ "loss": 0.0342,
+ "step": 36378
+ },
+ {
+ "epoch": 99.12534059945504,
+ "grad_norm": 0.4273928999900818,
+ "learning_rate": 4.012131645426687e-09,
+ "loss": 0.0043,
+ "step": 36379
+ },
+ {
+ "epoch": 99.12806539509536,
+ "grad_norm": 0.6269893646240234,
+ "learning_rate": 3.987174536421723e-09,
+ "loss": 0.0066,
+ "step": 36380
+ },
+ {
+ "epoch": 99.1307901907357,
+ "grad_norm": 1.2650542259216309,
+ "learning_rate": 3.962295276042083e-09,
+ "loss": 0.0149,
+ "step": 36381
+ },
+ {
+ "epoch": 99.13351498637603,
+ "grad_norm": 2.433002233505249,
+ "learning_rate": 3.937493864483166e-09,
+ "loss": 0.0162,
+ "step": 36382
+ },
+ {
+ "epoch": 99.13623978201635,
+ "grad_norm": 1.2623467445373535,
+ "learning_rate": 3.912770301937041e-09,
+ "loss": 0.0167,
+ "step": 36383
+ },
+ {
+ "epoch": 99.13896457765668,
+ "grad_norm": 0.5302137732505798,
+ "learning_rate": 3.8881245885957765e-09,
+ "loss": 0.0053,
+ "step": 36384
+ },
+ {
+ "epoch": 99.141689373297,
+ "grad_norm": 1.3124817609786987,
+ "learning_rate": 3.863556724652551e-09,
+ "loss": 0.0067,
+ "step": 36385
+ },
+ {
+ "epoch": 99.14441416893733,
+ "grad_norm": 0.992249071598053,
+ "learning_rate": 3.839066710298323e-09,
+ "loss": 0.0165,
+ "step": 36386
+ },
+ {
+ "epoch": 99.14713896457765,
+ "grad_norm": 0.7802175879478455,
+ "learning_rate": 3.81465454572405e-09,
+ "loss": 0.0112,
+ "step": 36387
+ },
+ {
+ "epoch": 99.14986376021798,
+ "grad_norm": 0.7151644229888916,
+ "learning_rate": 3.790320231119582e-09,
+ "loss": 0.0096,
+ "step": 36388
+ },
+ {
+ "epoch": 99.15258855585832,
+ "grad_norm": 1.4612313508987427,
+ "learning_rate": 3.766063766673656e-09,
+ "loss": 0.0206,
+ "step": 36389
+ },
+ {
+ "epoch": 99.15531335149863,
+ "grad_norm": 0.5027884244918823,
+ "learning_rate": 3.7418851525772296e-09,
+ "loss": 0.0033,
+ "step": 36390
+ },
+ {
+ "epoch": 99.15803814713897,
+ "grad_norm": 0.9636223316192627,
+ "learning_rate": 3.7177843890157108e-09,
+ "loss": 0.01,
+ "step": 36391
+ },
+ {
+ "epoch": 99.16076294277929,
+ "grad_norm": 0.7698834538459778,
+ "learning_rate": 3.6937614761789474e-09,
+ "loss": 0.0042,
+ "step": 36392
+ },
+ {
+ "epoch": 99.16348773841962,
+ "grad_norm": 0.8375442028045654,
+ "learning_rate": 3.6698164142534574e-09,
+ "loss": 0.0174,
+ "step": 36393
+ },
+ {
+ "epoch": 99.16621253405995,
+ "grad_norm": 0.817392110824585,
+ "learning_rate": 3.645949203425758e-09,
+ "loss": 0.0091,
+ "step": 36394
+ },
+ {
+ "epoch": 99.16893732970027,
+ "grad_norm": 0.4181765913963318,
+ "learning_rate": 3.6221598438812567e-09,
+ "loss": 0.0036,
+ "step": 36395
+ },
+ {
+ "epoch": 99.1716621253406,
+ "grad_norm": 0.5365371704101562,
+ "learning_rate": 3.5984483358064703e-09,
+ "loss": 0.0039,
+ "step": 36396
+ },
+ {
+ "epoch": 99.17438692098092,
+ "grad_norm": 0.7159658670425415,
+ "learning_rate": 3.574814679384586e-09,
+ "loss": 0.0073,
+ "step": 36397
+ },
+ {
+ "epoch": 99.17711171662125,
+ "grad_norm": 0.9685971736907959,
+ "learning_rate": 3.5512588748010114e-09,
+ "loss": 0.0733,
+ "step": 36398
+ },
+ {
+ "epoch": 99.17983651226157,
+ "grad_norm": 0.9732572436332703,
+ "learning_rate": 3.527780922237822e-09,
+ "loss": 0.0144,
+ "step": 36399
+ },
+ {
+ "epoch": 99.1825613079019,
+ "grad_norm": 0.7113685011863708,
+ "learning_rate": 3.504380821878206e-09,
+ "loss": 0.0062,
+ "step": 36400
+ },
+ {
+ "epoch": 99.18528610354224,
+ "grad_norm": 0.9187856316566467,
+ "learning_rate": 3.4810585739053492e-09,
+ "loss": 0.0173,
+ "step": 36401
+ },
+ {
+ "epoch": 99.18801089918256,
+ "grad_norm": 1.2252247333526611,
+ "learning_rate": 3.4578141785002182e-09,
+ "loss": 0.0153,
+ "step": 36402
+ },
+ {
+ "epoch": 99.19073569482289,
+ "grad_norm": 0.9905200600624084,
+ "learning_rate": 3.43464763584378e-09,
+ "loss": 0.0574,
+ "step": 36403
+ },
+ {
+ "epoch": 99.19346049046321,
+ "grad_norm": 0.7125391364097595,
+ "learning_rate": 3.4115589461169994e-09,
+ "loss": 0.0145,
+ "step": 36404
+ },
+ {
+ "epoch": 99.19618528610354,
+ "grad_norm": 0.8735656142234802,
+ "learning_rate": 3.388548109498624e-09,
+ "loss": 0.0341,
+ "step": 36405
+ },
+ {
+ "epoch": 99.19891008174388,
+ "grad_norm": 1.295193076133728,
+ "learning_rate": 3.3656151261696192e-09,
+ "loss": 0.1462,
+ "step": 36406
+ },
+ {
+ "epoch": 99.2016348773842,
+ "grad_norm": 1.4889827966690063,
+ "learning_rate": 3.3427599963065105e-09,
+ "loss": 0.0129,
+ "step": 36407
+ },
+ {
+ "epoch": 99.20435967302453,
+ "grad_norm": 0.9651525020599365,
+ "learning_rate": 3.319982720088044e-09,
+ "loss": 0.01,
+ "step": 36408
+ },
+ {
+ "epoch": 99.20708446866485,
+ "grad_norm": 1.1160513162612915,
+ "learning_rate": 3.2972832976918557e-09,
+ "loss": 0.0214,
+ "step": 36409
+ },
+ {
+ "epoch": 99.20980926430518,
+ "grad_norm": 1.3868825435638428,
+ "learning_rate": 3.274661729295581e-09,
+ "loss": 0.04,
+ "step": 36410
+ },
+ {
+ "epoch": 99.2125340599455,
+ "grad_norm": 0.45743802189826965,
+ "learning_rate": 3.2521180150735244e-09,
+ "loss": 0.0048,
+ "step": 36411
+ },
+ {
+ "epoch": 99.21525885558583,
+ "grad_norm": 1.1315032243728638,
+ "learning_rate": 3.2296521552033223e-09,
+ "loss": 0.0863,
+ "step": 36412
+ },
+ {
+ "epoch": 99.21798365122616,
+ "grad_norm": 0.7977487444877625,
+ "learning_rate": 3.207264149857059e-09,
+ "loss": 0.0097,
+ "step": 36413
+ },
+ {
+ "epoch": 99.22070844686648,
+ "grad_norm": 0.598854124546051,
+ "learning_rate": 3.18495399921237e-09,
+ "loss": 0.0064,
+ "step": 36414
+ },
+ {
+ "epoch": 99.22343324250681,
+ "grad_norm": 0.6429373025894165,
+ "learning_rate": 3.1627217034413403e-09,
+ "loss": 0.0081,
+ "step": 36415
+ },
+ {
+ "epoch": 99.22615803814713,
+ "grad_norm": 0.8168427348136902,
+ "learning_rate": 3.1405672627171645e-09,
+ "loss": 0.0058,
+ "step": 36416
+ },
+ {
+ "epoch": 99.22888283378747,
+ "grad_norm": 0.8430668711662292,
+ "learning_rate": 3.118490677211927e-09,
+ "loss": 0.0091,
+ "step": 36417
+ },
+ {
+ "epoch": 99.2316076294278,
+ "grad_norm": 0.9313650131225586,
+ "learning_rate": 3.096491947098823e-09,
+ "loss": 0.0355,
+ "step": 36418
+ },
+ {
+ "epoch": 99.23433242506812,
+ "grad_norm": 1.9717377424240112,
+ "learning_rate": 3.0745710725477164e-09,
+ "loss": 0.0195,
+ "step": 36419
+ },
+ {
+ "epoch": 99.23705722070845,
+ "grad_norm": 0.9548336863517761,
+ "learning_rate": 3.0527280537306915e-09,
+ "loss": 0.01,
+ "step": 36420
+ },
+ {
+ "epoch": 99.23978201634877,
+ "grad_norm": 0.4401150643825531,
+ "learning_rate": 3.0309628908176125e-09,
+ "loss": 0.0048,
+ "step": 36421
+ },
+ {
+ "epoch": 99.2425068119891,
+ "grad_norm": 1.546948790550232,
+ "learning_rate": 3.0092755839772336e-09,
+ "loss": 0.0104,
+ "step": 36422
+ },
+ {
+ "epoch": 99.24523160762942,
+ "grad_norm": 0.8673144578933716,
+ "learning_rate": 2.9876661333783084e-09,
+ "loss": 0.0102,
+ "step": 36423
+ },
+ {
+ "epoch": 99.24795640326975,
+ "grad_norm": 1.663370966911316,
+ "learning_rate": 2.9661345391907013e-09,
+ "loss": 0.1889,
+ "step": 36424
+ },
+ {
+ "epoch": 99.25068119891009,
+ "grad_norm": 0.9231250882148743,
+ "learning_rate": 2.9446808015809457e-09,
+ "loss": 0.0385,
+ "step": 36425
+ },
+ {
+ "epoch": 99.2534059945504,
+ "grad_norm": 0.7667209506034851,
+ "learning_rate": 2.923304920715575e-09,
+ "loss": 0.0119,
+ "step": 36426
+ },
+ {
+ "epoch": 99.25613079019074,
+ "grad_norm": 0.7251611948013306,
+ "learning_rate": 2.902006896762233e-09,
+ "loss": 0.0067,
+ "step": 36427
+ },
+ {
+ "epoch": 99.25885558583106,
+ "grad_norm": 0.7317537069320679,
+ "learning_rate": 2.8807867298863425e-09,
+ "loss": 0.0064,
+ "step": 36428
+ },
+ {
+ "epoch": 99.26158038147139,
+ "grad_norm": 1.0643892288208008,
+ "learning_rate": 2.8596444202522166e-09,
+ "loss": 0.052,
+ "step": 36429
+ },
+ {
+ "epoch": 99.26430517711172,
+ "grad_norm": 1.085923671722412,
+ "learning_rate": 2.8385799680263892e-09,
+ "loss": 0.1369,
+ "step": 36430
+ },
+ {
+ "epoch": 99.26702997275204,
+ "grad_norm": 1.112084150314331,
+ "learning_rate": 2.817593373372063e-09,
+ "loss": 0.0066,
+ "step": 36431
+ },
+ {
+ "epoch": 99.26975476839237,
+ "grad_norm": 1.0392259359359741,
+ "learning_rate": 2.79668463645244e-09,
+ "loss": 0.0283,
+ "step": 36432
+ },
+ {
+ "epoch": 99.2724795640327,
+ "grad_norm": 0.7541884183883667,
+ "learning_rate": 2.775853757429614e-09,
+ "loss": 0.0061,
+ "step": 36433
+ },
+ {
+ "epoch": 99.27520435967303,
+ "grad_norm": 1.3034611940383911,
+ "learning_rate": 2.755100736467897e-09,
+ "loss": 0.0632,
+ "step": 36434
+ },
+ {
+ "epoch": 99.27792915531334,
+ "grad_norm": 0.7877656817436218,
+ "learning_rate": 2.734425573727162e-09,
+ "loss": 0.006,
+ "step": 36435
+ },
+ {
+ "epoch": 99.28065395095368,
+ "grad_norm": 1.5813179016113281,
+ "learning_rate": 2.7138282693683904e-09,
+ "loss": 0.0278,
+ "step": 36436
+ },
+ {
+ "epoch": 99.28337874659401,
+ "grad_norm": 0.7188247442245483,
+ "learning_rate": 2.6933088235536752e-09,
+ "loss": 0.0107,
+ "step": 36437
+ },
+ {
+ "epoch": 99.28610354223433,
+ "grad_norm": 0.6535064578056335,
+ "learning_rate": 2.672867236440668e-09,
+ "loss": 0.0066,
+ "step": 36438
+ },
+ {
+ "epoch": 99.28882833787466,
+ "grad_norm": 0.5746757388114929,
+ "learning_rate": 2.6525035081903516e-09,
+ "loss": 0.0055,
+ "step": 36439
+ },
+ {
+ "epoch": 99.29155313351498,
+ "grad_norm": 0.5492826700210571,
+ "learning_rate": 2.632217638960377e-09,
+ "loss": 0.0043,
+ "step": 36440
+ },
+ {
+ "epoch": 99.29427792915531,
+ "grad_norm": 1.4315780401229858,
+ "learning_rate": 2.6120096289083964e-09,
+ "loss": 0.039,
+ "step": 36441
+ },
+ {
+ "epoch": 99.29700272479565,
+ "grad_norm": 1.033063292503357,
+ "learning_rate": 2.5918794781931713e-09,
+ "loss": 0.0602,
+ "step": 36442
+ },
+ {
+ "epoch": 99.29972752043597,
+ "grad_norm": 1.113369107246399,
+ "learning_rate": 2.571827186970133e-09,
+ "loss": 0.0296,
+ "step": 36443
+ },
+ {
+ "epoch": 99.3024523160763,
+ "grad_norm": 1.9665552377700806,
+ "learning_rate": 2.551852755395823e-09,
+ "loss": 0.0349,
+ "step": 36444
+ },
+ {
+ "epoch": 99.30517711171662,
+ "grad_norm": 0.3016335964202881,
+ "learning_rate": 2.5319561836267824e-09,
+ "loss": 0.0034,
+ "step": 36445
+ },
+ {
+ "epoch": 99.30790190735695,
+ "grad_norm": 1.0003637075424194,
+ "learning_rate": 2.5121374718162228e-09,
+ "loss": 0.0102,
+ "step": 36446
+ },
+ {
+ "epoch": 99.31062670299727,
+ "grad_norm": 1.0929118394851685,
+ "learning_rate": 2.4923966201195748e-09,
+ "loss": 0.009,
+ "step": 36447
+ },
+ {
+ "epoch": 99.3133514986376,
+ "grad_norm": 2.9143753051757812,
+ "learning_rate": 2.4727336286900493e-09,
+ "loss": 0.1013,
+ "step": 36448
+ },
+ {
+ "epoch": 99.31607629427793,
+ "grad_norm": 0.893912672996521,
+ "learning_rate": 2.4531484976819677e-09,
+ "loss": 0.015,
+ "step": 36449
+ },
+ {
+ "epoch": 99.31880108991825,
+ "grad_norm": 0.4229692220687866,
+ "learning_rate": 2.4336412272463197e-09,
+ "loss": 0.0049,
+ "step": 36450
+ },
+ {
+ "epoch": 99.32152588555859,
+ "grad_norm": 2.2412381172180176,
+ "learning_rate": 2.414211817535206e-09,
+ "loss": 0.011,
+ "step": 36451
+ },
+ {
+ "epoch": 99.3242506811989,
+ "grad_norm": 0.5776176452636719,
+ "learning_rate": 2.3948602687018375e-09,
+ "loss": 0.0041,
+ "step": 36452
+ },
+ {
+ "epoch": 99.32697547683924,
+ "grad_norm": 1.1002448797225952,
+ "learning_rate": 2.375586580893874e-09,
+ "loss": 0.0105,
+ "step": 36453
+ },
+ {
+ "epoch": 99.32970027247957,
+ "grad_norm": 0.8868784308433533,
+ "learning_rate": 2.3563907542645257e-09,
+ "loss": 0.0848,
+ "step": 36454
+ },
+ {
+ "epoch": 99.33242506811989,
+ "grad_norm": 1.048520803451538,
+ "learning_rate": 2.337272788961453e-09,
+ "loss": 0.0098,
+ "step": 36455
+ },
+ {
+ "epoch": 99.33514986376022,
+ "grad_norm": 0.7354152798652649,
+ "learning_rate": 2.318232685133426e-09,
+ "loss": 0.0292,
+ "step": 36456
+ },
+ {
+ "epoch": 99.33787465940054,
+ "grad_norm": 0.5923061370849609,
+ "learning_rate": 2.2992704429292136e-09,
+ "loss": 0.0056,
+ "step": 36457
+ },
+ {
+ "epoch": 99.34059945504087,
+ "grad_norm": 0.46030065417289734,
+ "learning_rate": 2.280386062497586e-09,
+ "loss": 0.0038,
+ "step": 36458
+ },
+ {
+ "epoch": 99.34332425068119,
+ "grad_norm": 0.34683528542518616,
+ "learning_rate": 2.2615795439839826e-09,
+ "loss": 0.0048,
+ "step": 36459
+ },
+ {
+ "epoch": 99.34604904632153,
+ "grad_norm": 0.9430139660835266,
+ "learning_rate": 2.242850887534953e-09,
+ "loss": 0.0742,
+ "step": 36460
+ },
+ {
+ "epoch": 99.34877384196186,
+ "grad_norm": 1.4432446956634521,
+ "learning_rate": 2.2242000932970464e-09,
+ "loss": 0.0098,
+ "step": 36461
+ },
+ {
+ "epoch": 99.35149863760218,
+ "grad_norm": 0.7602928280830383,
+ "learning_rate": 2.205627161415702e-09,
+ "loss": 0.0093,
+ "step": 36462
+ },
+ {
+ "epoch": 99.35422343324251,
+ "grad_norm": 1.5255515575408936,
+ "learning_rate": 2.187132092034139e-09,
+ "loss": 0.0104,
+ "step": 36463
+ },
+ {
+ "epoch": 99.35694822888283,
+ "grad_norm": 1.2301828861236572,
+ "learning_rate": 2.168714885298906e-09,
+ "loss": 0.0068,
+ "step": 36464
+ },
+ {
+ "epoch": 99.35967302452316,
+ "grad_norm": 40.75705337524414,
+ "learning_rate": 2.150375541351002e-09,
+ "loss": 0.0228,
+ "step": 36465
+ },
+ {
+ "epoch": 99.3623978201635,
+ "grad_norm": 0.612281322479248,
+ "learning_rate": 2.1321140603336452e-09,
+ "loss": 0.0057,
+ "step": 36466
+ },
+ {
+ "epoch": 99.36512261580381,
+ "grad_norm": 0.916857123374939,
+ "learning_rate": 2.113930442390055e-09,
+ "loss": 0.0065,
+ "step": 36467
+ },
+ {
+ "epoch": 99.36784741144415,
+ "grad_norm": 1.3068941831588745,
+ "learning_rate": 2.09582468766234e-09,
+ "loss": 0.0214,
+ "step": 36468
+ },
+ {
+ "epoch": 99.37057220708446,
+ "grad_norm": 1.4073234796524048,
+ "learning_rate": 2.0777967962892775e-09,
+ "loss": 0.0292,
+ "step": 36469
+ },
+ {
+ "epoch": 99.3732970027248,
+ "grad_norm": 0.7052296996116638,
+ "learning_rate": 2.059846768412976e-09,
+ "loss": 0.0058,
+ "step": 36470
+ },
+ {
+ "epoch": 99.37602179836512,
+ "grad_norm": 1.0046499967575073,
+ "learning_rate": 2.0419746041733247e-09,
+ "loss": 0.017,
+ "step": 36471
+ },
+ {
+ "epoch": 99.37874659400545,
+ "grad_norm": 0.9084253311157227,
+ "learning_rate": 2.0241803037091002e-09,
+ "loss": 0.0138,
+ "step": 36472
+ },
+ {
+ "epoch": 99.38147138964578,
+ "grad_norm": 0.941601574420929,
+ "learning_rate": 2.0064638671579706e-09,
+ "loss": 0.0924,
+ "step": 36473
+ },
+ {
+ "epoch": 99.3841961852861,
+ "grad_norm": 1.2844842672348022,
+ "learning_rate": 1.988825294658714e-09,
+ "loss": 0.0869,
+ "step": 36474
+ },
+ {
+ "epoch": 99.38692098092643,
+ "grad_norm": 1.244869351387024,
+ "learning_rate": 1.9712645863489977e-09,
+ "loss": 0.0223,
+ "step": 36475
+ },
+ {
+ "epoch": 99.38964577656675,
+ "grad_norm": 1.0714658498764038,
+ "learning_rate": 1.9537817423653794e-09,
+ "loss": 0.0106,
+ "step": 36476
+ },
+ {
+ "epoch": 99.39237057220708,
+ "grad_norm": 0.8116857409477234,
+ "learning_rate": 1.9363767628444164e-09,
+ "loss": 0.0222,
+ "step": 36477
+ },
+ {
+ "epoch": 99.39509536784742,
+ "grad_norm": 1.2919832468032837,
+ "learning_rate": 1.919049647920446e-09,
+ "loss": 0.0134,
+ "step": 36478
+ },
+ {
+ "epoch": 99.39782016348774,
+ "grad_norm": 0.9644562602043152,
+ "learning_rate": 1.9018003977289147e-09,
+ "loss": 0.0059,
+ "step": 36479
+ },
+ {
+ "epoch": 99.40054495912807,
+ "grad_norm": 0.8708236217498779,
+ "learning_rate": 1.8846290124052703e-09,
+ "loss": 0.0053,
+ "step": 36480
+ },
+ {
+ "epoch": 99.40326975476839,
+ "grad_norm": 1.122618556022644,
+ "learning_rate": 1.8675354920816293e-09,
+ "loss": 0.0417,
+ "step": 36481
+ },
+ {
+ "epoch": 99.40599455040872,
+ "grad_norm": 0.5786561369895935,
+ "learning_rate": 1.8505198368923282e-09,
+ "loss": 0.0067,
+ "step": 36482
+ },
+ {
+ "epoch": 99.40871934604904,
+ "grad_norm": 0.5623153448104858,
+ "learning_rate": 1.833582046968374e-09,
+ "loss": 0.0058,
+ "step": 36483
+ },
+ {
+ "epoch": 99.41144414168937,
+ "grad_norm": 1.179062843322754,
+ "learning_rate": 1.8167221224441035e-09,
+ "loss": 0.0413,
+ "step": 36484
+ },
+ {
+ "epoch": 99.4141689373297,
+ "grad_norm": 0.9757093191146851,
+ "learning_rate": 1.799940063448302e-09,
+ "loss": 0.0205,
+ "step": 36485
+ },
+ {
+ "epoch": 99.41689373297002,
+ "grad_norm": 0.6642643213272095,
+ "learning_rate": 1.783235870113087e-09,
+ "loss": 0.0071,
+ "step": 36486
+ },
+ {
+ "epoch": 99.41961852861036,
+ "grad_norm": 0.978675127029419,
+ "learning_rate": 1.7666095425683538e-09,
+ "loss": 0.0112,
+ "step": 36487
+ },
+ {
+ "epoch": 99.42234332425068,
+ "grad_norm": 1.0850191116333008,
+ "learning_rate": 1.7500610809439989e-09,
+ "loss": 0.0079,
+ "step": 36488
+ },
+ {
+ "epoch": 99.42506811989101,
+ "grad_norm": 1.113951563835144,
+ "learning_rate": 1.7335904853676977e-09,
+ "loss": 0.0091,
+ "step": 36489
+ },
+ {
+ "epoch": 99.42779291553134,
+ "grad_norm": 0.8939641118049622,
+ "learning_rate": 1.7171977559682363e-09,
+ "loss": 0.0107,
+ "step": 36490
+ },
+ {
+ "epoch": 99.43051771117166,
+ "grad_norm": 0.9832046031951904,
+ "learning_rate": 1.70088289287329e-09,
+ "loss": 0.0141,
+ "step": 36491
+ },
+ {
+ "epoch": 99.433242506812,
+ "grad_norm": 1.3743202686309814,
+ "learning_rate": 1.684645896210535e-09,
+ "loss": 0.0112,
+ "step": 36492
+ },
+ {
+ "epoch": 99.43596730245231,
+ "grad_norm": 1.1549997329711914,
+ "learning_rate": 1.6684867661065363e-09,
+ "loss": 0.0337,
+ "step": 36493
+ },
+ {
+ "epoch": 99.43869209809264,
+ "grad_norm": 1.5246202945709229,
+ "learning_rate": 1.6524055026856388e-09,
+ "loss": 0.0496,
+ "step": 36494
+ },
+ {
+ "epoch": 99.44141689373296,
+ "grad_norm": 1.0420753955841064,
+ "learning_rate": 1.6364021060744084e-09,
+ "loss": 0.0131,
+ "step": 36495
+ },
+ {
+ "epoch": 99.4441416893733,
+ "grad_norm": 0.8744569420814514,
+ "learning_rate": 1.6204765763971897e-09,
+ "loss": 0.012,
+ "step": 36496
+ },
+ {
+ "epoch": 99.44686648501363,
+ "grad_norm": 0.7852693796157837,
+ "learning_rate": 1.6046289137783277e-09,
+ "loss": 0.011,
+ "step": 36497
+ },
+ {
+ "epoch": 99.44959128065395,
+ "grad_norm": 1.2899314165115356,
+ "learning_rate": 1.5888591183399472e-09,
+ "loss": 0.0486,
+ "step": 36498
+ },
+ {
+ "epoch": 99.45231607629428,
+ "grad_norm": 2.411069869995117,
+ "learning_rate": 1.5731671902063927e-09,
+ "loss": 0.0344,
+ "step": 36499
+ },
+ {
+ "epoch": 99.4550408719346,
+ "grad_norm": 0.5415963530540466,
+ "learning_rate": 1.557553129498679e-09,
+ "loss": 0.0067,
+ "step": 36500
+ },
+ {
+ "epoch": 99.45776566757493,
+ "grad_norm": 0.8462944626808167,
+ "learning_rate": 1.5420169363389303e-09,
+ "loss": 0.0088,
+ "step": 36501
+ },
+ {
+ "epoch": 99.46049046321527,
+ "grad_norm": 1.041366457939148,
+ "learning_rate": 1.5265586108481612e-09,
+ "loss": 0.1089,
+ "step": 36502
+ },
+ {
+ "epoch": 99.46321525885558,
+ "grad_norm": 1.4284944534301758,
+ "learning_rate": 1.5111781531473856e-09,
+ "loss": 0.0116,
+ "step": 36503
+ },
+ {
+ "epoch": 99.46594005449592,
+ "grad_norm": 0.4902544319629669,
+ "learning_rate": 1.4958755633542876e-09,
+ "loss": 0.0044,
+ "step": 36504
+ },
+ {
+ "epoch": 99.46866485013624,
+ "grad_norm": 1.0226242542266846,
+ "learning_rate": 1.4806508415909914e-09,
+ "loss": 0.0508,
+ "step": 36505
+ },
+ {
+ "epoch": 99.47138964577657,
+ "grad_norm": 1.4559136629104614,
+ "learning_rate": 1.4655039879740706e-09,
+ "loss": 0.1603,
+ "step": 36506
+ },
+ {
+ "epoch": 99.47411444141689,
+ "grad_norm": 1.3760749101638794,
+ "learning_rate": 1.4504350026223192e-09,
+ "loss": 0.0147,
+ "step": 36507
+ },
+ {
+ "epoch": 99.47683923705722,
+ "grad_norm": 0.24618315696716309,
+ "learning_rate": 1.43544388565231e-09,
+ "loss": 0.0025,
+ "step": 36508
+ },
+ {
+ "epoch": 99.47956403269755,
+ "grad_norm": 0.5106381177902222,
+ "learning_rate": 1.4205306371817274e-09,
+ "loss": 0.005,
+ "step": 36509
+ },
+ {
+ "epoch": 99.48228882833787,
+ "grad_norm": 1.5095163583755493,
+ "learning_rate": 1.405695257326034e-09,
+ "loss": 0.0718,
+ "step": 36510
+ },
+ {
+ "epoch": 99.4850136239782,
+ "grad_norm": 0.8864157199859619,
+ "learning_rate": 1.3909377462006933e-09,
+ "loss": 0.0862,
+ "step": 36511
+ },
+ {
+ "epoch": 99.48773841961852,
+ "grad_norm": 0.9940311312675476,
+ "learning_rate": 1.3762581039211686e-09,
+ "loss": 0.025,
+ "step": 36512
+ },
+ {
+ "epoch": 99.49046321525886,
+ "grad_norm": 0.8760824203491211,
+ "learning_rate": 1.3616563306018126e-09,
+ "loss": 0.0108,
+ "step": 36513
+ },
+ {
+ "epoch": 99.49318801089919,
+ "grad_norm": 0.5413751006126404,
+ "learning_rate": 1.3471324263558684e-09,
+ "loss": 0.0057,
+ "step": 36514
+ },
+ {
+ "epoch": 99.49591280653951,
+ "grad_norm": 1.460724949836731,
+ "learning_rate": 1.3326863912965782e-09,
+ "loss": 0.0376,
+ "step": 36515
+ },
+ {
+ "epoch": 99.49863760217984,
+ "grad_norm": 1.1126245260238647,
+ "learning_rate": 1.318318225536075e-09,
+ "loss": 0.0232,
+ "step": 36516
+ },
+ {
+ "epoch": 99.50136239782016,
+ "grad_norm": 1.644063115119934,
+ "learning_rate": 1.3040279291876013e-09,
+ "loss": 0.0263,
+ "step": 36517
+ },
+ {
+ "epoch": 99.50408719346049,
+ "grad_norm": 1.8117982149124146,
+ "learning_rate": 1.2898155023610692e-09,
+ "loss": 0.0777,
+ "step": 36518
+ },
+ {
+ "epoch": 99.50681198910081,
+ "grad_norm": 1.0528934001922607,
+ "learning_rate": 1.2756809451675012e-09,
+ "loss": 0.0203,
+ "step": 36519
+ },
+ {
+ "epoch": 99.50953678474114,
+ "grad_norm": 0.7698866724967957,
+ "learning_rate": 1.2616242577168092e-09,
+ "loss": 0.006,
+ "step": 36520
+ },
+ {
+ "epoch": 99.51226158038148,
+ "grad_norm": 1.3500726222991943,
+ "learning_rate": 1.2476454401200155e-09,
+ "loss": 0.0382,
+ "step": 36521
+ },
+ {
+ "epoch": 99.5149863760218,
+ "grad_norm": 0.9559323191642761,
+ "learning_rate": 1.2337444924837017e-09,
+ "loss": 0.0365,
+ "step": 36522
+ },
+ {
+ "epoch": 99.51771117166213,
+ "grad_norm": 0.9203585982322693,
+ "learning_rate": 1.2199214149177795e-09,
+ "loss": 0.0336,
+ "step": 36523
+ },
+ {
+ "epoch": 99.52043596730245,
+ "grad_norm": 1.3936797380447388,
+ "learning_rate": 1.2061762075288307e-09,
+ "loss": 0.0176,
+ "step": 36524
+ },
+ {
+ "epoch": 99.52316076294278,
+ "grad_norm": 0.5366933941841125,
+ "learning_rate": 1.192508870424547e-09,
+ "loss": 0.0058,
+ "step": 36525
+ },
+ {
+ "epoch": 99.52588555858311,
+ "grad_norm": 1.0354785919189453,
+ "learning_rate": 1.1789194037115093e-09,
+ "loss": 0.0487,
+ "step": 36526
+ },
+ {
+ "epoch": 99.52861035422343,
+ "grad_norm": 0.9724841117858887,
+ "learning_rate": 1.1654078074940788e-09,
+ "loss": 0.0298,
+ "step": 36527
+ },
+ {
+ "epoch": 99.53133514986376,
+ "grad_norm": 2.4224467277526855,
+ "learning_rate": 1.1519740818799473e-09,
+ "loss": 0.0128,
+ "step": 36528
+ },
+ {
+ "epoch": 99.53405994550408,
+ "grad_norm": 0.8343061804771423,
+ "learning_rate": 1.1386182269723656e-09,
+ "loss": 0.0061,
+ "step": 36529
+ },
+ {
+ "epoch": 99.53678474114442,
+ "grad_norm": 1.0792675018310547,
+ "learning_rate": 1.1253402428745841e-09,
+ "loss": 0.0078,
+ "step": 36530
+ },
+ {
+ "epoch": 99.53950953678473,
+ "grad_norm": 0.7293926477432251,
+ "learning_rate": 1.1121401296909639e-09,
+ "loss": 0.0776,
+ "step": 36531
+ },
+ {
+ "epoch": 99.54223433242507,
+ "grad_norm": 0.850073516368866,
+ "learning_rate": 1.099017887524756e-09,
+ "loss": 0.009,
+ "step": 36532
+ },
+ {
+ "epoch": 99.5449591280654,
+ "grad_norm": 0.8132539987564087,
+ "learning_rate": 1.0859735164769903e-09,
+ "loss": 0.0078,
+ "step": 36533
+ },
+ {
+ "epoch": 99.54768392370572,
+ "grad_norm": 1.7840840816497803,
+ "learning_rate": 1.0730070166509177e-09,
+ "loss": 0.0612,
+ "step": 36534
+ },
+ {
+ "epoch": 99.55040871934605,
+ "grad_norm": 1.184390902519226,
+ "learning_rate": 1.060118388145348e-09,
+ "loss": 0.0109,
+ "step": 36535
+ },
+ {
+ "epoch": 99.55313351498637,
+ "grad_norm": 1.1258374452590942,
+ "learning_rate": 1.0473076310624218e-09,
+ "loss": 0.0128,
+ "step": 36536
+ },
+ {
+ "epoch": 99.5558583106267,
+ "grad_norm": 0.8875135183334351,
+ "learning_rate": 1.034574745500949e-09,
+ "loss": 0.0482,
+ "step": 36537
+ },
+ {
+ "epoch": 99.55858310626704,
+ "grad_norm": 1.3262945413589478,
+ "learning_rate": 1.0219197315597396e-09,
+ "loss": 0.0078,
+ "step": 36538
+ },
+ {
+ "epoch": 99.56130790190736,
+ "grad_norm": 2.649268627166748,
+ "learning_rate": 1.0093425893387133e-09,
+ "loss": 0.0347,
+ "step": 36539
+ },
+ {
+ "epoch": 99.56403269754769,
+ "grad_norm": 2.1651105880737305,
+ "learning_rate": 9.968433189355697e-10,
+ "loss": 0.2215,
+ "step": 36540
+ },
+ {
+ "epoch": 99.566757493188,
+ "grad_norm": 1.2350386381149292,
+ "learning_rate": 9.844219204468985e-10,
+ "loss": 0.0204,
+ "step": 36541
+ },
+ {
+ "epoch": 99.56948228882834,
+ "grad_norm": 0.6300219893455505,
+ "learning_rate": 9.720783939692891e-10,
+ "loss": 0.0087,
+ "step": 36542
+ },
+ {
+ "epoch": 99.57220708446866,
+ "grad_norm": 0.8912532925605774,
+ "learning_rate": 9.598127395993307e-10,
+ "loss": 0.007,
+ "step": 36543
+ },
+ {
+ "epoch": 99.57493188010899,
+ "grad_norm": 0.7070656418800354,
+ "learning_rate": 9.476249574325026e-10,
+ "loss": 0.0122,
+ "step": 36544
+ },
+ {
+ "epoch": 99.57765667574932,
+ "grad_norm": 0.7108834385871887,
+ "learning_rate": 9.355150475642837e-10,
+ "loss": 0.0063,
+ "step": 36545
+ },
+ {
+ "epoch": 99.58038147138964,
+ "grad_norm": 1.0778053998947144,
+ "learning_rate": 9.234830100879333e-10,
+ "loss": 0.0786,
+ "step": 36546
+ },
+ {
+ "epoch": 99.58310626702998,
+ "grad_norm": 1.446662187576294,
+ "learning_rate": 9.115288450978199e-10,
+ "loss": 0.1051,
+ "step": 36547
+ },
+ {
+ "epoch": 99.5858310626703,
+ "grad_norm": 0.4672255218029022,
+ "learning_rate": 8.996525526872024e-10,
+ "loss": 0.0035,
+ "step": 36548
+ },
+ {
+ "epoch": 99.58855585831063,
+ "grad_norm": 0.9260085821151733,
+ "learning_rate": 8.878541329482293e-10,
+ "loss": 0.006,
+ "step": 36549
+ },
+ {
+ "epoch": 99.59128065395096,
+ "grad_norm": 0.4919837415218353,
+ "learning_rate": 8.761335859730491e-10,
+ "loss": 0.0061,
+ "step": 36550
+ },
+ {
+ "epoch": 99.59400544959128,
+ "grad_norm": 0.9059943556785583,
+ "learning_rate": 8.644909118515898e-10,
+ "loss": 0.0082,
+ "step": 36551
+ },
+ {
+ "epoch": 99.59673024523161,
+ "grad_norm": 0.46500882506370544,
+ "learning_rate": 8.529261106771103e-10,
+ "loss": 0.0081,
+ "step": 36552
+ },
+ {
+ "epoch": 99.59945504087193,
+ "grad_norm": 1.45890474319458,
+ "learning_rate": 8.41439182537318e-10,
+ "loss": 0.0102,
+ "step": 36553
+ },
+ {
+ "epoch": 99.60217983651226,
+ "grad_norm": 0.9883960485458374,
+ "learning_rate": 8.300301275221412e-10,
+ "loss": 0.0086,
+ "step": 36554
+ },
+ {
+ "epoch": 99.60490463215258,
+ "grad_norm": 1.0567998886108398,
+ "learning_rate": 8.186989457215078e-10,
+ "loss": 0.0211,
+ "step": 36555
+ },
+ {
+ "epoch": 99.60762942779292,
+ "grad_norm": 1.1310794353485107,
+ "learning_rate": 8.074456372231254e-10,
+ "loss": 0.0234,
+ "step": 36556
+ },
+ {
+ "epoch": 99.61035422343325,
+ "grad_norm": 0.627931535243988,
+ "learning_rate": 7.962702021147018e-10,
+ "loss": 0.0047,
+ "step": 36557
+ },
+ {
+ "epoch": 99.61307901907357,
+ "grad_norm": 1.3540805578231812,
+ "learning_rate": 7.851726404828342e-10,
+ "loss": 0.0202,
+ "step": 36558
+ },
+ {
+ "epoch": 99.6158038147139,
+ "grad_norm": 1.3373403549194336,
+ "learning_rate": 7.741529524141201e-10,
+ "loss": 0.0088,
+ "step": 36559
+ },
+ {
+ "epoch": 99.61852861035422,
+ "grad_norm": 0.7662239670753479,
+ "learning_rate": 7.632111379951568e-10,
+ "loss": 0.008,
+ "step": 36560
+ },
+ {
+ "epoch": 99.62125340599455,
+ "grad_norm": 0.6587647199630737,
+ "learning_rate": 7.523471973092112e-10,
+ "loss": 0.0053,
+ "step": 36561
+ },
+ {
+ "epoch": 99.62397820163488,
+ "grad_norm": 1.272156834602356,
+ "learning_rate": 7.415611304428805e-10,
+ "loss": 0.0337,
+ "step": 36562
+ },
+ {
+ "epoch": 99.6267029972752,
+ "grad_norm": 2.4060840606689453,
+ "learning_rate": 7.308529374794316e-10,
+ "loss": 0.031,
+ "step": 36563
+ },
+ {
+ "epoch": 99.62942779291554,
+ "grad_norm": 0.9832499027252197,
+ "learning_rate": 7.20222618502131e-10,
+ "loss": 0.0247,
+ "step": 36564
+ },
+ {
+ "epoch": 99.63215258855585,
+ "grad_norm": 0.5286083817481995,
+ "learning_rate": 7.096701735942457e-10,
+ "loss": 0.005,
+ "step": 36565
+ },
+ {
+ "epoch": 99.63487738419619,
+ "grad_norm": 0.7461893558502197,
+ "learning_rate": 6.991956028379321e-10,
+ "loss": 0.0129,
+ "step": 36566
+ },
+ {
+ "epoch": 99.6376021798365,
+ "grad_norm": 0.8546726107597351,
+ "learning_rate": 6.887989063131262e-10,
+ "loss": 0.0093,
+ "step": 36567
+ },
+ {
+ "epoch": 99.64032697547684,
+ "grad_norm": 1.026053786277771,
+ "learning_rate": 6.784800841030948e-10,
+ "loss": 0.0443,
+ "step": 36568
+ },
+ {
+ "epoch": 99.64305177111717,
+ "grad_norm": 0.991606593132019,
+ "learning_rate": 6.682391362877738e-10,
+ "loss": 0.0058,
+ "step": 36569
+ },
+ {
+ "epoch": 99.64577656675749,
+ "grad_norm": 0.34491023421287537,
+ "learning_rate": 6.58076062944879e-10,
+ "loss": 0.0034,
+ "step": 36570
+ },
+ {
+ "epoch": 99.64850136239782,
+ "grad_norm": 0.3711836338043213,
+ "learning_rate": 6.479908641565668e-10,
+ "loss": 0.0047,
+ "step": 36571
+ },
+ {
+ "epoch": 99.65122615803814,
+ "grad_norm": 0.8016419410705566,
+ "learning_rate": 6.379835399983326e-10,
+ "loss": 0.0058,
+ "step": 36572
+ },
+ {
+ "epoch": 99.65395095367847,
+ "grad_norm": 0.7077916860580444,
+ "learning_rate": 6.280540905512222e-10,
+ "loss": 0.0069,
+ "step": 36573
+ },
+ {
+ "epoch": 99.65667574931881,
+ "grad_norm": 0.5742380619049072,
+ "learning_rate": 6.18202515889621e-10,
+ "loss": 0.0065,
+ "step": 36574
+ },
+ {
+ "epoch": 99.65940054495913,
+ "grad_norm": 0.6638831496238708,
+ "learning_rate": 6.084288160923546e-10,
+ "loss": 0.0075,
+ "step": 36575
+ },
+ {
+ "epoch": 99.66212534059946,
+ "grad_norm": 0.6625300049781799,
+ "learning_rate": 5.987329912349182e-10,
+ "loss": 0.005,
+ "step": 36576
+ },
+ {
+ "epoch": 99.66485013623978,
+ "grad_norm": 1.1085331439971924,
+ "learning_rate": 5.891150413928071e-10,
+ "loss": 0.0101,
+ "step": 36577
+ },
+ {
+ "epoch": 99.66757493188011,
+ "grad_norm": 1.2990927696228027,
+ "learning_rate": 5.795749666404061e-10,
+ "loss": 0.0635,
+ "step": 36578
+ },
+ {
+ "epoch": 99.67029972752043,
+ "grad_norm": 0.8111783862113953,
+ "learning_rate": 5.701127670532103e-10,
+ "loss": 0.0087,
+ "step": 36579
+ },
+ {
+ "epoch": 99.67302452316076,
+ "grad_norm": 0.4371224045753479,
+ "learning_rate": 5.607284427033843e-10,
+ "loss": 0.0043,
+ "step": 36580
+ },
+ {
+ "epoch": 99.6757493188011,
+ "grad_norm": 1.6787724494934082,
+ "learning_rate": 5.514219936642029e-10,
+ "loss": 0.0542,
+ "step": 36581
+ },
+ {
+ "epoch": 99.67847411444141,
+ "grad_norm": 1.5500749349594116,
+ "learning_rate": 5.42193420010051e-10,
+ "loss": 0.0994,
+ "step": 36582
+ },
+ {
+ "epoch": 99.68119891008175,
+ "grad_norm": 0.595932126045227,
+ "learning_rate": 5.330427218108725e-10,
+ "loss": 0.0059,
+ "step": 36583
+ },
+ {
+ "epoch": 99.68392370572207,
+ "grad_norm": 0.9983651041984558,
+ "learning_rate": 5.239698991388321e-10,
+ "loss": 0.0074,
+ "step": 36584
+ },
+ {
+ "epoch": 99.6866485013624,
+ "grad_norm": 0.85166335105896,
+ "learning_rate": 5.149749520638736e-10,
+ "loss": 0.0094,
+ "step": 36585
+ },
+ {
+ "epoch": 99.68937329700273,
+ "grad_norm": 1.1394339799880981,
+ "learning_rate": 5.060578806559413e-10,
+ "loss": 0.0216,
+ "step": 36586
+ },
+ {
+ "epoch": 99.69209809264305,
+ "grad_norm": 0.7544242143630981,
+ "learning_rate": 4.972186849860894e-10,
+ "loss": 0.0078,
+ "step": 36587
+ },
+ {
+ "epoch": 99.69482288828338,
+ "grad_norm": 0.5883969664573669,
+ "learning_rate": 4.884573651209313e-10,
+ "loss": 0.0068,
+ "step": 36588
+ },
+ {
+ "epoch": 99.6975476839237,
+ "grad_norm": 0.9329765439033508,
+ "learning_rate": 4.797739211304108e-10,
+ "loss": 0.0116,
+ "step": 36589
+ },
+ {
+ "epoch": 99.70027247956403,
+ "grad_norm": 2.682224988937378,
+ "learning_rate": 4.711683530822519e-10,
+ "loss": 0.0559,
+ "step": 36590
+ },
+ {
+ "epoch": 99.70299727520435,
+ "grad_norm": 1.261016845703125,
+ "learning_rate": 4.626406610419576e-10,
+ "loss": 0.02,
+ "step": 36591
+ },
+ {
+ "epoch": 99.70572207084469,
+ "grad_norm": 1.5486286878585815,
+ "learning_rate": 4.5419084507725143e-10,
+ "loss": 0.0508,
+ "step": 36592
+ },
+ {
+ "epoch": 99.70844686648502,
+ "grad_norm": 0.6405063271522522,
+ "learning_rate": 4.458189052536366e-10,
+ "loss": 0.0086,
+ "step": 36593
+ },
+ {
+ "epoch": 99.71117166212534,
+ "grad_norm": 1.2200278043746948,
+ "learning_rate": 4.37524841635506e-10,
+ "loss": 0.0091,
+ "step": 36594
+ },
+ {
+ "epoch": 99.71389645776567,
+ "grad_norm": 1.0259157419204712,
+ "learning_rate": 4.2930865428836286e-10,
+ "loss": 0.0147,
+ "step": 36595
+ },
+ {
+ "epoch": 99.71662125340599,
+ "grad_norm": 0.8574162721633911,
+ "learning_rate": 4.2117034327548987e-10,
+ "loss": 0.0078,
+ "step": 36596
+ },
+ {
+ "epoch": 99.71934604904632,
+ "grad_norm": 0.888594925403595,
+ "learning_rate": 4.1310990866128e-10,
+ "loss": 0.0109,
+ "step": 36597
+ },
+ {
+ "epoch": 99.72207084468666,
+ "grad_norm": 0.7313604950904846,
+ "learning_rate": 4.0512735050790566e-10,
+ "loss": 0.0089,
+ "step": 36598
+ },
+ {
+ "epoch": 99.72479564032697,
+ "grad_norm": 1.1689982414245605,
+ "learning_rate": 3.9722266887753933e-10,
+ "loss": 0.0435,
+ "step": 36599
+ },
+ {
+ "epoch": 99.7275204359673,
+ "grad_norm": 0.9045603275299072,
+ "learning_rate": 3.893958638312434e-10,
+ "loss": 0.0647,
+ "step": 36600
+ },
+ {
+ "epoch": 99.73024523160763,
+ "grad_norm": 1.0030831098556519,
+ "learning_rate": 3.8164693543119024e-10,
+ "loss": 0.007,
+ "step": 36601
+ },
+ {
+ "epoch": 99.73297002724796,
+ "grad_norm": 0.8626757860183716,
+ "learning_rate": 3.7397588373733196e-10,
+ "loss": 0.0114,
+ "step": 36602
+ },
+ {
+ "epoch": 99.73569482288828,
+ "grad_norm": 1.2405248880386353,
+ "learning_rate": 3.6638270880851034e-10,
+ "loss": 0.1328,
+ "step": 36603
+ },
+ {
+ "epoch": 99.73841961852861,
+ "grad_norm": 0.6727249622344971,
+ "learning_rate": 3.588674107035672e-10,
+ "loss": 0.0071,
+ "step": 36604
+ },
+ {
+ "epoch": 99.74114441416894,
+ "grad_norm": 1.1927236318588257,
+ "learning_rate": 3.514299894835649e-10,
+ "loss": 0.0062,
+ "step": 36605
+ },
+ {
+ "epoch": 99.74386920980926,
+ "grad_norm": 0.9313592314720154,
+ "learning_rate": 3.4407044520401443e-10,
+ "loss": 0.0079,
+ "step": 36606
+ },
+ {
+ "epoch": 99.7465940054496,
+ "grad_norm": 1.3259987831115723,
+ "learning_rate": 3.3678877792264754e-10,
+ "loss": 0.0205,
+ "step": 36607
+ },
+ {
+ "epoch": 99.74931880108991,
+ "grad_norm": 1.371843695640564,
+ "learning_rate": 3.295849876971957e-10,
+ "loss": 0.0627,
+ "step": 36608
+ },
+ {
+ "epoch": 99.75204359673025,
+ "grad_norm": 1.4672036170959473,
+ "learning_rate": 3.2245907458317015e-10,
+ "loss": 0.0097,
+ "step": 36609
+ },
+ {
+ "epoch": 99.75476839237058,
+ "grad_norm": 0.8919857740402222,
+ "learning_rate": 3.154110386349718e-10,
+ "loss": 0.0193,
+ "step": 36610
+ },
+ {
+ "epoch": 99.7574931880109,
+ "grad_norm": 0.8674205541610718,
+ "learning_rate": 3.08440879909222e-10,
+ "loss": 0.0077,
+ "step": 36611
+ },
+ {
+ "epoch": 99.76021798365123,
+ "grad_norm": 1.0443745851516724,
+ "learning_rate": 3.015485984603217e-10,
+ "loss": 0.0058,
+ "step": 36612
+ },
+ {
+ "epoch": 99.76294277929155,
+ "grad_norm": 1.8402045965194702,
+ "learning_rate": 2.9473419434045136e-10,
+ "loss": 0.1254,
+ "step": 36613
+ },
+ {
+ "epoch": 99.76566757493188,
+ "grad_norm": 1.7670795917510986,
+ "learning_rate": 2.879976676029017e-10,
+ "loss": 0.1845,
+ "step": 36614
+ },
+ {
+ "epoch": 99.7683923705722,
+ "grad_norm": 2.3225512504577637,
+ "learning_rate": 2.8133901830096343e-10,
+ "loss": 0.0092,
+ "step": 36615
+ },
+ {
+ "epoch": 99.77111716621253,
+ "grad_norm": 0.8132645487785339,
+ "learning_rate": 2.7475824648570683e-10,
+ "loss": 0.0101,
+ "step": 36616
+ },
+ {
+ "epoch": 99.77384196185287,
+ "grad_norm": 1.002020239830017,
+ "learning_rate": 2.6825535220931233e-10,
+ "loss": 0.0103,
+ "step": 36617
+ },
+ {
+ "epoch": 99.77656675749319,
+ "grad_norm": 1.1251944303512573,
+ "learning_rate": 2.6183033552174e-10,
+ "loss": 0.0301,
+ "step": 36618
+ },
+ {
+ "epoch": 99.77929155313352,
+ "grad_norm": 0.6965276598930359,
+ "learning_rate": 2.554831964729498e-10,
+ "loss": 0.0057,
+ "step": 36619
+ },
+ {
+ "epoch": 99.78201634877384,
+ "grad_norm": 1.0351601839065552,
+ "learning_rate": 2.4921393511290195e-10,
+ "loss": 0.0064,
+ "step": 36620
+ },
+ {
+ "epoch": 99.78474114441417,
+ "grad_norm": 0.999735414981842,
+ "learning_rate": 2.4302255149044607e-10,
+ "loss": 0.0668,
+ "step": 36621
+ },
+ {
+ "epoch": 99.7874659400545,
+ "grad_norm": 0.6967249512672424,
+ "learning_rate": 2.369090456533218e-10,
+ "loss": 0.0128,
+ "step": 36622
+ },
+ {
+ "epoch": 99.79019073569482,
+ "grad_norm": 0.8213074803352356,
+ "learning_rate": 2.3087341764926885e-10,
+ "loss": 0.0139,
+ "step": 36623
+ },
+ {
+ "epoch": 99.79291553133515,
+ "grad_norm": 1.3058196306228638,
+ "learning_rate": 2.2491566752491646e-10,
+ "loss": 0.0817,
+ "step": 36624
+ },
+ {
+ "epoch": 99.79564032697547,
+ "grad_norm": 0.6410449743270874,
+ "learning_rate": 2.19035795326894e-10,
+ "loss": 0.0086,
+ "step": 36625
+ },
+ {
+ "epoch": 99.7983651226158,
+ "grad_norm": 0.9345778822898865,
+ "learning_rate": 2.1323380110183088e-10,
+ "loss": 0.0049,
+ "step": 36626
+ },
+ {
+ "epoch": 99.80108991825612,
+ "grad_norm": 0.9602949619293213,
+ "learning_rate": 2.0750968489413604e-10,
+ "loss": 0.0249,
+ "step": 36627
+ },
+ {
+ "epoch": 99.80381471389646,
+ "grad_norm": 3.0153098106384277,
+ "learning_rate": 2.0186344674821834e-10,
+ "loss": 0.0369,
+ "step": 36628
+ },
+ {
+ "epoch": 99.80653950953679,
+ "grad_norm": 1.2782790660858154,
+ "learning_rate": 1.9629508670848674e-10,
+ "loss": 0.0388,
+ "step": 36629
+ },
+ {
+ "epoch": 99.80926430517711,
+ "grad_norm": 1.3569929599761963,
+ "learning_rate": 1.9080460481823993e-10,
+ "loss": 0.0335,
+ "step": 36630
+ },
+ {
+ "epoch": 99.81198910081744,
+ "grad_norm": 0.7588345408439636,
+ "learning_rate": 1.853920011207766e-10,
+ "loss": 0.0097,
+ "step": 36631
+ },
+ {
+ "epoch": 99.81471389645776,
+ "grad_norm": 1.469595193862915,
+ "learning_rate": 1.8005727565717502e-10,
+ "loss": 0.0124,
+ "step": 36632
+ },
+ {
+ "epoch": 99.8174386920981,
+ "grad_norm": 1.6229714155197144,
+ "learning_rate": 1.7480042846851341e-10,
+ "loss": 0.0599,
+ "step": 36633
+ },
+ {
+ "epoch": 99.82016348773843,
+ "grad_norm": 0.884316086769104,
+ "learning_rate": 1.696214595980905e-10,
+ "loss": 0.0152,
+ "step": 36634
+ },
+ {
+ "epoch": 99.82288828337875,
+ "grad_norm": 1.372717261314392,
+ "learning_rate": 1.6452036908365388e-10,
+ "loss": 0.0331,
+ "step": 36635
+ },
+ {
+ "epoch": 99.82561307901908,
+ "grad_norm": 1.0641883611679077,
+ "learning_rate": 1.59497156967392e-10,
+ "loss": 0.0115,
+ "step": 36636
+ },
+ {
+ "epoch": 99.8283378746594,
+ "grad_norm": 1.1394352912902832,
+ "learning_rate": 1.545518232859422e-10,
+ "loss": 0.0539,
+ "step": 36637
+ },
+ {
+ "epoch": 99.83106267029973,
+ "grad_norm": 1.340164065361023,
+ "learning_rate": 1.4968436807927255e-10,
+ "loss": 0.0109,
+ "step": 36638
+ },
+ {
+ "epoch": 99.83378746594005,
+ "grad_norm": 0.874573826789856,
+ "learning_rate": 1.4489479138513062e-10,
+ "loss": 0.0094,
+ "step": 36639
+ },
+ {
+ "epoch": 99.83651226158038,
+ "grad_norm": 1.0564724206924438,
+ "learning_rate": 1.40183093241264e-10,
+ "loss": 0.0828,
+ "step": 36640
+ },
+ {
+ "epoch": 99.83923705722071,
+ "grad_norm": 1.2262564897537231,
+ "learning_rate": 1.3554927368319982e-10,
+ "loss": 0.0461,
+ "step": 36641
+ },
+ {
+ "epoch": 99.84196185286103,
+ "grad_norm": 1.3921183347702026,
+ "learning_rate": 1.3099333274757543e-10,
+ "loss": 0.0394,
+ "step": 36642
+ },
+ {
+ "epoch": 99.84468664850137,
+ "grad_norm": 1.0243321657180786,
+ "learning_rate": 1.2651527046991796e-10,
+ "loss": 0.0152,
+ "step": 36643
+ },
+ {
+ "epoch": 99.84741144414168,
+ "grad_norm": 0.9394124150276184,
+ "learning_rate": 1.2211508688464434e-10,
+ "loss": 0.0115,
+ "step": 36644
+ },
+ {
+ "epoch": 99.85013623978202,
+ "grad_norm": 2.6445891857147217,
+ "learning_rate": 1.1779278202728174e-10,
+ "loss": 0.0149,
+ "step": 36645
+ },
+ {
+ "epoch": 99.85286103542235,
+ "grad_norm": 1.0469396114349365,
+ "learning_rate": 1.1354835593113678e-10,
+ "loss": 0.013,
+ "step": 36646
+ },
+ {
+ "epoch": 99.85558583106267,
+ "grad_norm": 1.7961831092834473,
+ "learning_rate": 1.0938180862729575e-10,
+ "loss": 0.0323,
+ "step": 36647
+ },
+ {
+ "epoch": 99.858310626703,
+ "grad_norm": 0.9601427912712097,
+ "learning_rate": 1.0529314015128577e-10,
+ "loss": 0.0094,
+ "step": 36648
+ },
+ {
+ "epoch": 99.86103542234332,
+ "grad_norm": 0.5433586239814758,
+ "learning_rate": 1.0128235053308289e-10,
+ "loss": 0.0035,
+ "step": 36649
+ },
+ {
+ "epoch": 99.86376021798365,
+ "grad_norm": 1.0919102430343628,
+ "learning_rate": 9.734943980377331e-11,
+ "loss": 0.0093,
+ "step": 36650
+ },
+ {
+ "epoch": 99.86648501362397,
+ "grad_norm": 1.046339750289917,
+ "learning_rate": 9.349440799444331e-11,
+ "loss": 0.0082,
+ "step": 36651
+ },
+ {
+ "epoch": 99.8692098092643,
+ "grad_norm": 0.7733861804008484,
+ "learning_rate": 8.97172551350689e-11,
+ "loss": 0.1263,
+ "step": 36652
+ },
+ {
+ "epoch": 99.87193460490464,
+ "grad_norm": 0.7991666197776794,
+ "learning_rate": 8.60179812556261e-11,
+ "loss": 0.006,
+ "step": 36653
+ },
+ {
+ "epoch": 99.87465940054496,
+ "grad_norm": 1.0193151235580444,
+ "learning_rate": 8.23965863838705e-11,
+ "loss": 0.1004,
+ "step": 36654
+ },
+ {
+ "epoch": 99.87738419618529,
+ "grad_norm": 1.2108503580093384,
+ "learning_rate": 7.885307054866786e-11,
+ "loss": 0.0138,
+ "step": 36655
+ },
+ {
+ "epoch": 99.88010899182561,
+ "grad_norm": 0.5539959073066711,
+ "learning_rate": 7.53874337777738e-11,
+ "loss": 0.0056,
+ "step": 36656
+ },
+ {
+ "epoch": 99.88283378746594,
+ "grad_norm": 1.0490033626556396,
+ "learning_rate": 7.199967609783365e-11,
+ "loss": 0.0214,
+ "step": 36657
+ },
+ {
+ "epoch": 99.88555858310627,
+ "grad_norm": 0.9285327792167664,
+ "learning_rate": 6.868979753549277e-11,
+ "loss": 0.0142,
+ "step": 36658
+ },
+ {
+ "epoch": 99.88828337874659,
+ "grad_norm": 0.9224516153335571,
+ "learning_rate": 6.545779811628627e-11,
+ "loss": 0.0077,
+ "step": 36659
+ },
+ {
+ "epoch": 99.89100817438693,
+ "grad_norm": 0.8336037397384644,
+ "learning_rate": 6.230367786574931e-11,
+ "loss": 0.0143,
+ "step": 36660
+ },
+ {
+ "epoch": 99.89373297002724,
+ "grad_norm": 0.28376656770706177,
+ "learning_rate": 5.922743680719655e-11,
+ "loss": 0.0024,
+ "step": 36661
+ },
+ {
+ "epoch": 99.89645776566758,
+ "grad_norm": 0.36154839396476746,
+ "learning_rate": 5.622907496616314e-11,
+ "loss": 0.0047,
+ "step": 36662
+ },
+ {
+ "epoch": 99.8991825613079,
+ "grad_norm": 0.8392691612243652,
+ "learning_rate": 5.330859236596375e-11,
+ "loss": 0.0506,
+ "step": 36663
+ },
+ {
+ "epoch": 99.90190735694823,
+ "grad_norm": 1.67721688747406,
+ "learning_rate": 5.046598902769262e-11,
+ "loss": 0.139,
+ "step": 36664
+ },
+ {
+ "epoch": 99.90463215258856,
+ "grad_norm": 2.5148797035217285,
+ "learning_rate": 4.770126497466443e-11,
+ "loss": 0.0086,
+ "step": 36665
+ },
+ {
+ "epoch": 99.90735694822888,
+ "grad_norm": 0.8229836225509644,
+ "learning_rate": 4.5014420229083645e-11,
+ "loss": 0.0095,
+ "step": 36666
+ },
+ {
+ "epoch": 99.91008174386921,
+ "grad_norm": 1.0967562198638916,
+ "learning_rate": 4.240545481093428e-11,
+ "loss": 0.0162,
+ "step": 36667
+ },
+ {
+ "epoch": 99.91280653950953,
+ "grad_norm": 1.783069372177124,
+ "learning_rate": 3.987436874020034e-11,
+ "loss": 0.0223,
+ "step": 36668
+ },
+ {
+ "epoch": 99.91553133514986,
+ "grad_norm": 1.6673619747161865,
+ "learning_rate": 3.7421162036865854e-11,
+ "loss": 0.0673,
+ "step": 36669
+ },
+ {
+ "epoch": 99.9182561307902,
+ "grad_norm": 1.022261381149292,
+ "learning_rate": 3.504583472091483e-11,
+ "loss": 0.0148,
+ "step": 36670
+ },
+ {
+ "epoch": 99.92098092643052,
+ "grad_norm": 1.4416850805282593,
+ "learning_rate": 3.274838681011083e-11,
+ "loss": 0.054,
+ "step": 36671
+ },
+ {
+ "epoch": 99.92370572207085,
+ "grad_norm": 0.9525250792503357,
+ "learning_rate": 3.0528818322217435e-11,
+ "loss": 0.0102,
+ "step": 36672
+ },
+ {
+ "epoch": 99.92643051771117,
+ "grad_norm": 0.9363698959350586,
+ "learning_rate": 2.838712927388798e-11,
+ "loss": 0.0095,
+ "step": 36673
+ },
+ {
+ "epoch": 99.9291553133515,
+ "grad_norm": 1.0197443962097168,
+ "learning_rate": 2.6323319683996262e-11,
+ "loss": 0.0104,
+ "step": 36674
+ },
+ {
+ "epoch": 99.93188010899182,
+ "grad_norm": 1.0397732257843018,
+ "learning_rate": 2.4337389565864956e-11,
+ "loss": 0.136,
+ "step": 36675
+ },
+ {
+ "epoch": 99.93460490463215,
+ "grad_norm": 1.2098485231399536,
+ "learning_rate": 2.2429338937257627e-11,
+ "loss": 0.0102,
+ "step": 36676
+ },
+ {
+ "epoch": 99.93732970027249,
+ "grad_norm": 0.8727114200592041,
+ "learning_rate": 2.059916781260718e-11,
+ "loss": 0.0125,
+ "step": 36677
+ },
+ {
+ "epoch": 99.9400544959128,
+ "grad_norm": 0.7807077765464783,
+ "learning_rate": 1.8846876205236287e-11,
+ "loss": 0.0093,
+ "step": 36678
+ },
+ {
+ "epoch": 99.94277929155314,
+ "grad_norm": 1.5374289751052856,
+ "learning_rate": 1.7172464128467626e-11,
+ "loss": 0.0481,
+ "step": 36679
+ },
+ {
+ "epoch": 99.94550408719346,
+ "grad_norm": 0.725468099117279,
+ "learning_rate": 1.5575931596734094e-11,
+ "loss": 0.0061,
+ "step": 36680
+ },
+ {
+ "epoch": 99.94822888283379,
+ "grad_norm": 0.6850793361663818,
+ "learning_rate": 1.4057278622248149e-11,
+ "loss": 0.0064,
+ "step": 36681
+ },
+ {
+ "epoch": 99.95095367847412,
+ "grad_norm": 0.8419177532196045,
+ "learning_rate": 1.2616505216112019e-11,
+ "loss": 0.0074,
+ "step": 36682
+ },
+ {
+ "epoch": 99.95367847411444,
+ "grad_norm": 1.2669341564178467,
+ "learning_rate": 1.1253611389427931e-11,
+ "loss": 0.0349,
+ "step": 36683
+ },
+ {
+ "epoch": 99.95640326975477,
+ "grad_norm": 0.9836031794548035,
+ "learning_rate": 9.968597153298121e-12,
+ "loss": 0.0067,
+ "step": 36684
+ },
+ {
+ "epoch": 99.95912806539509,
+ "grad_norm": 1.4341623783111572,
+ "learning_rate": 8.761462518824815e-12,
+ "loss": 0.0585,
+ "step": 36685
+ },
+ {
+ "epoch": 99.96185286103542,
+ "grad_norm": 1.1942996978759766,
+ "learning_rate": 7.632207492669353e-12,
+ "loss": 0.0097,
+ "step": 36686
+ },
+ {
+ "epoch": 99.96457765667574,
+ "grad_norm": 0.7035285830497742,
+ "learning_rate": 6.580832085933964e-12,
+ "loss": 0.0051,
+ "step": 36687
+ },
+ {
+ "epoch": 99.96730245231608,
+ "grad_norm": 1.511474847793579,
+ "learning_rate": 5.6073363063902094e-12,
+ "loss": 0.0193,
+ "step": 36688
+ },
+ {
+ "epoch": 99.97002724795641,
+ "grad_norm": 0.72588711977005,
+ "learning_rate": 4.711720160699429e-12,
+ "loss": 0.0068,
+ "step": 36689
+ },
+ {
+ "epoch": 99.97275204359673,
+ "grad_norm": 0.5077633261680603,
+ "learning_rate": 3.893983657743405e-12,
+ "loss": 0.0057,
+ "step": 36690
+ },
+ {
+ "epoch": 99.97547683923706,
+ "grad_norm": 0.8793197274208069,
+ "learning_rate": 3.154126800852808e-12,
+ "loss": 0.0086,
+ "step": 36691
+ },
+ {
+ "epoch": 99.97820163487738,
+ "grad_norm": 0.9360848069190979,
+ "learning_rate": 2.492149598909421e-12,
+ "loss": 0.0181,
+ "step": 36692
+ },
+ {
+ "epoch": 99.98092643051771,
+ "grad_norm": 1.2109025716781616,
+ "learning_rate": 1.908052055243914e-12,
+ "loss": 0.0331,
+ "step": 36693
+ },
+ {
+ "epoch": 99.98365122615803,
+ "grad_norm": 2.383509635925293,
+ "learning_rate": 1.4018341754074017e-12,
+ "loss": 0.0604,
+ "step": 36694
+ },
+ {
+ "epoch": 99.98637602179836,
+ "grad_norm": 0.4946274161338806,
+ "learning_rate": 9.7349596162033e-13,
+ "loss": 0.0051,
+ "step": 36695
+ },
+ {
+ "epoch": 99.9891008174387,
+ "grad_norm": 0.9729773998260498,
+ "learning_rate": 6.23037419433814e-13,
+ "loss": 0.0109,
+ "step": 36696
+ },
+ {
+ "epoch": 99.99182561307902,
+ "grad_norm": 0.9471953511238098,
+ "learning_rate": 3.5045854995807705e-13,
+ "loss": 0.0456,
+ "step": 36697
+ },
+ {
+ "epoch": 99.99455040871935,
+ "grad_norm": 0.6186493635177612,
+ "learning_rate": 1.5575935652378804e-13,
+ "loss": 0.0057,
+ "step": 36698
+ },
+ {
+ "epoch": 99.99727520435967,
+ "grad_norm": 1.7595164775848389,
+ "learning_rate": 3.893983913094701e-14,
+ "loss": 0.0154,
+ "step": 36699
+ },
+ {
+ "epoch": 100.0,
+ "grad_norm": 0.48446109890937805,
+ "learning_rate": 0.0,
+ "loss": 0.0047,
+ "step": 36700
+ },
+ {
+ "epoch": 100.0,
+ "step": 36700,
+ "total_flos": 8.57029458526208e+16,
+ "train_loss": 0.36938278674226693,
+ "train_runtime": 20903.7359,
+ "train_samples_per_second": 56.09,
+ "train_steps_per_second": 1.756
+ }
+ ],
+ "logging_steps": 1.0,
+ "max_steps": 36700,
+ "num_input_tokens_seen": 0,
+ "num_train_epochs": 100,
+ "save_steps": 50000,
+ "stateful_callbacks": {
+ "TrainerControl": {
+ "args": {
+ "should_epoch_stop": false,
+ "should_evaluate": false,
+ "should_log": false,
+ "should_save": true,
+ "should_training_stop": true
+ },
+ "attributes": {}
+ }
+ },
+ "total_flos": 8.57029458526208e+16,
+ "train_batch_size": 2,
+ "trial_name": null,
+ "trial_params": null
+}
diff --git a/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/config.json b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..8379e80e7fe6666e07d7c1fbaeb1adf8de83e580
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/config.json
@@ -0,0 +1,239 @@
+{
+ "_name_or_path": "microsoft/Florence-2-large",
+ "architectures": [
+ "Florence2ForConditionalGeneration"
+ ],
+ "auto_map": {
+ "AutoConfig": "microsoft/Florence-2-large--configuration_florence2.Florence2Config",
+ "AutoModelForCausalLM": "microsoft/Florence-2-large--modeling_florence2.Florence2ForConditionalGeneration"
+ },
+ "bos_token_id": 0,
+ "eos_token_id": 2,
+ "hidden_size": 1024,
+ "ignore_index": -100,
+ "is_encoder_decoder": true,
+ "model_type": "florence2",
+ "pad_token_id": 1,
+ "projection_dim": 1024,
+ "response_format": "florence2",
+ "text_config": {
+ "_name_or_path": "",
+ "activation_dropout": 0.1,
+ "activation_function": "gelu",
+ "add_bias_logits": false,
+ "add_cross_attention": false,
+ "add_final_layer_norm": false,
+ "architectures": null,
+ "attention_dropout": 0.1,
+ "bad_words_ids": null,
+ "begin_suppress_tokens": null,
+ "bos_token_id": 0,
+ "chunk_size_feed_forward": 0,
+ "classif_dropout": 0.1,
+ "classifier_dropout": 0.0,
+ "cross_attention_hidden_size": null,
+ "d_model": 1024,
+ "decoder_attention_heads": 16,
+ "decoder_ffn_dim": 4096,
+ "decoder_layerdrop": 0.0,
+ "decoder_layers": 12,
+ "decoder_start_token_id": 2,
+ "diversity_penalty": 0.0,
+ "do_sample": false,
+ "dropout": 0.1,
+ "early_stopping": true,
+ "encoder_attention_heads": 16,
+ "encoder_ffn_dim": 4096,
+ "encoder_layerdrop": 0.0,
+ "encoder_layers": 12,
+ "encoder_no_repeat_ngram_size": 0,
+ "eos_token_id": 2,
+ "exponential_decay_length_penalty": null,
+ "finetuning_task": null,
+ "forced_bos_token_id": 0,
+ "forced_eos_token_id": 2,
+ "gradient_checkpointing": false,
+ "id2label": {
+ "0": "LABEL_0",
+ "1": "LABEL_1",
+ "2": "LABEL_2"
+ },
+ "init_std": 0.02,
+ "is_decoder": false,
+ "is_encoder_decoder": true,
+ "label2id": {
+ "LABEL_0": 0,
+ "LABEL_1": 1,
+ "LABEL_2": 2
+ },
+ "length_penalty": 1.0,
+ "max_length": 20,
+ "max_position_embeddings": 2048,
+ "min_length": 0,
+ "model_type": "florence2_language",
+ "no_repeat_ngram_size": 3,
+ "normalize_before": false,
+ "num_beam_groups": 1,
+ "num_beams": 3,
+ "num_hidden_layers": 12,
+ "num_return_sequences": 1,
+ "output_attentions": false,
+ "output_hidden_states": false,
+ "output_scores": false,
+ "pad_token_id": 1,
+ "prefix": null,
+ "problem_type": null,
+ "pruned_heads": {},
+ "remove_invalid_values": false,
+ "repetition_penalty": 1.0,
+ "return_dict": true,
+ "return_dict_in_generate": false,
+ "scale_embedding": false,
+ "sep_token_id": null,
+ "suppress_tokens": null,
+ "task_specific_params": null,
+ "temperature": 1.0,
+ "tf_legacy_loss": false,
+ "tie_encoder_decoder": false,
+ "tie_word_embeddings": true,
+ "tokenizer_class": null,
+ "top_k": 50,
+ "top_p": 1.0,
+ "torch_dtype": null,
+ "torchscript": false,
+ "typical_p": 1.0,
+ "use_bfloat16": false,
+ "use_cache": true,
+ "vocab_size": 51289
+ },
+ "torch_dtype": "bfloat16",
+ "transformers_version": "4.45.0",
+ "vision_config": {
+ "_name_or_path": "",
+ "add_cross_attention": false,
+ "architectures": null,
+ "bad_words_ids": null,
+ "begin_suppress_tokens": null,
+ "bos_token_id": null,
+ "chunk_size_feed_forward": 0,
+ "cross_attention_hidden_size": null,
+ "decoder_start_token_id": null,
+ "depths": [
+ 1,
+ 1,
+ 9,
+ 1
+ ],
+ "dim_embed": [
+ 256,
+ 512,
+ 1024,
+ 2048
+ ],
+ "diversity_penalty": 0.0,
+ "do_sample": false,
+ "drop_path_rate": 0.1,
+ "early_stopping": false,
+ "enable_checkpoint": false,
+ "encoder_no_repeat_ngram_size": 0,
+ "eos_token_id": null,
+ "exponential_decay_length_penalty": null,
+ "finetuning_task": null,
+ "forced_bos_token_id": null,
+ "forced_eos_token_id": null,
+ "id2label": {
+ "0": "LABEL_0",
+ "1": "LABEL_1"
+ },
+ "image_feature_source": [
+ "spatial_avg_pool",
+ "temporal_avg_pool"
+ ],
+ "image_pos_embed": {
+ "max_pos_embeddings": 50,
+ "type": "learned_abs_2d"
+ },
+ "is_decoder": false,
+ "is_encoder_decoder": false,
+ "label2id": {
+ "LABEL_0": 0,
+ "LABEL_1": 1
+ },
+ "length_penalty": 1.0,
+ "max_length": 20,
+ "min_length": 0,
+ "model_type": "davit",
+ "no_repeat_ngram_size": 0,
+ "num_beam_groups": 1,
+ "num_beams": 1,
+ "num_groups": [
+ 8,
+ 16,
+ 32,
+ 64
+ ],
+ "num_heads": [
+ 8,
+ 16,
+ 32,
+ 64
+ ],
+ "num_return_sequences": 1,
+ "output_attentions": false,
+ "output_hidden_states": false,
+ "output_scores": false,
+ "pad_token_id": null,
+ "patch_padding": [
+ 3,
+ 1,
+ 1,
+ 1
+ ],
+ "patch_prenorm": [
+ false,
+ true,
+ true,
+ true
+ ],
+ "patch_size": [
+ 7,
+ 3,
+ 3,
+ 3
+ ],
+ "patch_stride": [
+ 4,
+ 2,
+ 2,
+ 2
+ ],
+ "prefix": null,
+ "problem_type": null,
+ "projection_dim": 1024,
+ "pruned_heads": {},
+ "remove_invalid_values": false,
+ "repetition_penalty": 1.0,
+ "return_dict": true,
+ "return_dict_in_generate": false,
+ "sep_token_id": null,
+ "suppress_tokens": null,
+ "task_specific_params": null,
+ "temperature": 1.0,
+ "tf_legacy_loss": false,
+ "tie_encoder_decoder": false,
+ "tie_word_embeddings": true,
+ "tokenizer_class": null,
+ "top_k": 50,
+ "top_p": 1.0,
+ "torch_dtype": null,
+ "torchscript": false,
+ "typical_p": 1.0,
+ "use_bfloat16": false,
+ "visual_temporal_embedding": {
+ "max_temporal_embeddings": 100,
+ "type": "COSINE"
+ },
+ "window_size": 12
+ },
+ "vocab_size": 51289
+}
diff --git a/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/1282.xml b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/1282.xml
new file mode 100644
index 0000000000000000000000000000000000000000..35328f31117768f83bc564cfd64511503e9f7158
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/1282.xml
@@ -0,0 +1 @@
+
\ No newline at end of file
diff --git a/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/1424.xml b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/1424.xml
new file mode 100644
index 0000000000000000000000000000000000000000..b234e06907a8107c4bbb241e0c5056f0a744d8a5
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/1424.xml
@@ -0,0 +1 @@
+
\ No newline at end of file
diff --git a/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/1561.xml b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/1561.xml
new file mode 100644
index 0000000000000000000000000000000000000000..a44b9ed9ba5ae37a07ca586fe665ead8121353f9
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/1561.xml
@@ -0,0 +1 @@
+
\ No newline at end of file
diff --git a/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/1729.xml b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/1729.xml
new file mode 100644
index 0000000000000000000000000000000000000000..3133e9918ffc74e56113ab03394491be9a190d53
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/1729.xml
@@ -0,0 +1 @@
+
\ No newline at end of file
diff --git a/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/1852.xml b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/1852.xml
new file mode 100644
index 0000000000000000000000000000000000000000..f845fcaba377995a5cdee9a9e4e6989f4381257c
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/1852.xml
@@ -0,0 +1 @@
+
\ No newline at end of file
diff --git a/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/2020.xml b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/2020.xml
new file mode 100644
index 0000000000000000000000000000000000000000..7a344f79da742130d4be5707da2c91ab6c8d82a4
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/2020.xml
@@ -0,0 +1 @@
+
\ No newline at end of file
diff --git a/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/2074.xml b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/2074.xml
new file mode 100644
index 0000000000000000000000000000000000000000..243e3ef296e494e8b1bbebbb65f28849ace6db6a
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/2074.xml
@@ -0,0 +1 @@
+
\ No newline at end of file
diff --git a/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/2157.xml b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/2157.xml
new file mode 100644
index 0000000000000000000000000000000000000000..e078183f908333d8642e95ffa3349fecbdfb7e80
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/2157.xml
@@ -0,0 +1 @@
+
\ No newline at end of file
diff --git a/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/2333.xml b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/2333.xml
new file mode 100644
index 0000000000000000000000000000000000000000..f2f6bee634d1aeadde37a5c0cd2a0e29b14d9e56
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/2333.xml
@@ -0,0 +1 @@
+
\ No newline at end of file
diff --git a/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/2403.xml b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/2403.xml
new file mode 100644
index 0000000000000000000000000000000000000000..f9e830974d34737fdb793fb525d3c20aa7666940
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/2403.xml
@@ -0,0 +1 @@
+
\ No newline at end of file
diff --git a/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/2582.xml b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/2582.xml
new file mode 100644
index 0000000000000000000000000000000000000000..9814424ad3b89421fbde20cbac72b7f1d2d844ca
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/2582.xml
@@ -0,0 +1 @@
+
\ No newline at end of file
diff --git a/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/3078.xml b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/3078.xml
new file mode 100644
index 0000000000000000000000000000000000000000..a62b1c23784670513d096a81383561036b58a0a3
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/3078.xml
@@ -0,0 +1 @@
+
\ No newline at end of file
diff --git a/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/3164.xml b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/3164.xml
new file mode 100644
index 0000000000000000000000000000000000000000..8d19f2d0c9770ebd9c6de5b2c4c298057e5b5e51
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/3164.xml
@@ -0,0 +1 @@
+
\ No newline at end of file
diff --git a/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/3197.xml b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/3197.xml
new file mode 100644
index 0000000000000000000000000000000000000000..1752c5d3eda9c56c3ff4bb2c8f15a0f03d3d943b
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/3197.xml
@@ -0,0 +1 @@
+
\ No newline at end of file
diff --git a/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/326.xml b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/326.xml
new file mode 100644
index 0000000000000000000000000000000000000000..47b8abaa2c68fb0a3466004e8504a14751741423
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/326.xml
@@ -0,0 +1 @@
+
\ No newline at end of file
diff --git a/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/3268.xml b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/3268.xml
new file mode 100644
index 0000000000000000000000000000000000000000..87cc92bcc8cc0f19bebce19ad453f31f5f4b4115
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/3268.xml
@@ -0,0 +1 @@
+
\ No newline at end of file
diff --git a/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/3314.xml b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/3314.xml
new file mode 100644
index 0000000000000000000000000000000000000000..cf9347edeebb41e8fc95749cbc031402e6438d03
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/3314.xml
@@ -0,0 +1 @@
+
\ No newline at end of file
diff --git a/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/3641.xml b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/3641.xml
new file mode 100644
index 0000000000000000000000000000000000000000..2520023b52c317d2ffaba53a80d8be5f7f5e13b3
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/3641.xml
@@ -0,0 +1 @@
+
\ No newline at end of file
diff --git a/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/3655.xml b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/3655.xml
new file mode 100644
index 0000000000000000000000000000000000000000..5c16271715a2510d10e64f12ac5e17158544b09c
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/3655.xml
@@ -0,0 +1 @@
+
\ No newline at end of file
diff --git a/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/3762.xml b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/3762.xml
new file mode 100644
index 0000000000000000000000000000000000000000..0e7291a2199b9198c0588e55988053b226bdec77
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/3762.xml
@@ -0,0 +1 @@
+
\ No newline at end of file
diff --git a/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/3922.xml b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/3922.xml
new file mode 100644
index 0000000000000000000000000000000000000000..ee7a44ccaea9ec723e280b73d3d72067138b3034
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/3922.xml
@@ -0,0 +1 @@
+
\ No newline at end of file
diff --git a/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/4176.xml b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/4176.xml
new file mode 100644
index 0000000000000000000000000000000000000000..f276ad70850a6f4b603cd99bcb56cc6247815542
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/4176.xml
@@ -0,0 +1 @@
+
\ No newline at end of file
diff --git a/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/4269.xml b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/4269.xml
new file mode 100644
index 0000000000000000000000000000000000000000..317f01a569c4b34415e3c5f6d609b35bc145e5a1
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/4269.xml
@@ -0,0 +1 @@
+
\ No newline at end of file
diff --git a/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/4573.xml b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/4573.xml
new file mode 100644
index 0000000000000000000000000000000000000000..8c8e1d160946d31aa6f8c1a0784822dfc50662f6
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/4573.xml
@@ -0,0 +1 @@
+
\ No newline at end of file
diff --git a/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/4623.xml b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/4623.xml
new file mode 100644
index 0000000000000000000000000000000000000000..e67f90f195a36584688ec5c259d9917243da6490
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/4623.xml
@@ -0,0 +1 @@
+
\ No newline at end of file
diff --git a/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/4722.xml b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/4722.xml
new file mode 100644
index 0000000000000000000000000000000000000000..fc0b94606d19adc3f5635c89874e87b85960d92d
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/4722.xml
@@ -0,0 +1 @@
+
\ No newline at end of file
diff --git a/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/4842.xml b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/4842.xml
new file mode 100644
index 0000000000000000000000000000000000000000..436e8d748728382b4520100de54dba0f39b00177
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/4842.xml
@@ -0,0 +1 @@
+
\ No newline at end of file
diff --git a/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/4901.xml b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/4901.xml
new file mode 100644
index 0000000000000000000000000000000000000000..e581cc3cce7326a7124889b318e4a2a6009a8d9a
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/4901.xml
@@ -0,0 +1 @@
+
\ No newline at end of file
diff --git a/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/5904.xml b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/5904.xml
new file mode 100644
index 0000000000000000000000000000000000000000..e1f5aa7a9bff08bbd123180662a90220b6047e82
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/5904.xml
@@ -0,0 +1 @@
+
\ No newline at end of file
diff --git a/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/5962.xml b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/5962.xml
new file mode 100644
index 0000000000000000000000000000000000000000..62455b1f8a09d176112f3957c0297fdea3c3b56a
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/5962.xml
@@ -0,0 +1 @@
+
\ No newline at end of file
diff --git a/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/6311.xml b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/6311.xml
new file mode 100644
index 0000000000000000000000000000000000000000..335708886437eda46471ef7af98388ad02c28a0b
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/6311.xml
@@ -0,0 +1 @@
+
\ No newline at end of file
diff --git a/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/6341.xml b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/6341.xml
new file mode 100644
index 0000000000000000000000000000000000000000..8d1a061214d557c12d278774a350af9c41df83a6
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/6341.xml
@@ -0,0 +1 @@
+
\ No newline at end of file
diff --git a/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/6528.xml b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/6528.xml
new file mode 100644
index 0000000000000000000000000000000000000000..66aa6e6075c8f7249a080c9048e580879422b3cc
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/6528.xml
@@ -0,0 +1 @@
+
\ No newline at end of file
diff --git a/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/6674.xml b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/6674.xml
new file mode 100644
index 0000000000000000000000000000000000000000..63853417c5c6b6a25e02d9caf292b58625950c0d
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/6674.xml
@@ -0,0 +1 @@
+
\ No newline at end of file
diff --git a/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/7076.xml b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/7076.xml
new file mode 100644
index 0000000000000000000000000000000000000000..c15d8f243ea0cd9283a15f74e003402d2284f6e6
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/7076.xml
@@ -0,0 +1 @@
+
\ No newline at end of file
diff --git a/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/7188.xml b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/7188.xml
new file mode 100644
index 0000000000000000000000000000000000000000..036264f40e4bf7e2b0876b5c22fa9c0e06d4e4f7
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/7188.xml
@@ -0,0 +1 @@
+
\ No newline at end of file
diff --git a/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/720.xml b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/720.xml
new file mode 100644
index 0000000000000000000000000000000000000000..eeee7f194a412f0fd86747e120aab46b094d17d3
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/720.xml
@@ -0,0 +1 @@
+
\ No newline at end of file
diff --git a/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/7498.xml b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/7498.xml
new file mode 100644
index 0000000000000000000000000000000000000000..e39ca461235f860c01f49f7fa86c52e0f0c4a426
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/7498.xml
@@ -0,0 +1 @@
+
\ No newline at end of file
diff --git a/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/7620.xml b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/7620.xml
new file mode 100644
index 0000000000000000000000000000000000000000..381470d3f589cf99d8dc7c3a8e91c72ebde3cc9c
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/7620.xml
@@ -0,0 +1 @@
+
\ No newline at end of file
diff --git a/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/7865.xml b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/7865.xml
new file mode 100644
index 0000000000000000000000000000000000000000..cf66b566f70ed272c7da1cfe9951602eb92d34e6
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/7865.xml
@@ -0,0 +1 @@
+
\ No newline at end of file
diff --git a/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/7894.xml b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/7894.xml
new file mode 100644
index 0000000000000000000000000000000000000000..e058a4130052b3c1d672fd77d7a0465acd2e8523
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/7894.xml
@@ -0,0 +1 @@
+
\ No newline at end of file
diff --git a/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/8088.xml b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/8088.xml
new file mode 100644
index 0000000000000000000000000000000000000000..ad4274c8c1ae75cf2ec9e2f498fc86a089c88c8e
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/8088.xml
@@ -0,0 +1 @@
+
\ No newline at end of file
diff --git a/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/910.xml b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/910.xml
new file mode 100644
index 0000000000000000000000000000000000000000..2ec7c7859b1d411a6c34315047b1490634b953e0
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/910.xml
@@ -0,0 +1 @@
+
\ No newline at end of file
diff --git a/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/939.xml b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/939.xml
new file mode 100644
index 0000000000000000000000000000000000000000..093e4b4bbbfd395280fd88be81b24e35134ed743
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/dota_Task1/939.xml
@@ -0,0 +1 @@
+
\ No newline at end of file
diff --git a/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/eval_liqingyun@SH-IDC1-10-140-37-31_device0_rank0.log b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/eval_liqingyun@SH-IDC1-10-140-37-31_device0_rank0.log
new file mode 100644
index 0000000000000000000000000000000000000000..9df991a79083fd37dbd3a4d2975122599a7ee998
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m/parsed_answers_test/eval_liqingyun@SH-IDC1-10-140-37-31_device0_rank0.log
@@ -0,0 +1,769 @@
+2025/01/24 19:46:12 - fair1m test Evaluation - INFO - --------------------
+2025/01/24 19:46:12 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:12 - fair1m test Evaluation - INFO - Fuzzy matched dam to dump truck
+2025/01/24 19:46:12 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:12 - fair1m test Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 19:46:12 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:12 - fair1m test Evaluation - INFO - Fuzzy matched traileriler to trailer
+2025/01/24 19:46:12 - fair1m test Evaluation - INFO - Fuzzy matched boing747 to boeing747
+2025/01/24 19:46:12 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:12 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:12 - fair1m test Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 19:46:12 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:12 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:12 - fair1m test Evaluation - INFO - Fuzzy matched expressway toll station to tractor
+2025/01/24 19:46:12 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:12 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:12 - fair1m test Evaluation - INFO - Fuzzy matched windmill to intersection
+2025/01/24 19:46:12 - fair1m test Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 19:46:12 - fair1m test Evaluation - INFO - Fuzzy matched 787 to boeing787
+2025/01/24 19:46:12 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:12 - fair1m test Evaluation - INFO - Fuzzy matched golffield to football field
+2025/01/24 19:46:12 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:12 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:12 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:12 - fair1m test Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 19:46:12 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:12 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:12 - fair1m test Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 19:46:12 - fair1m test Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 19:46:12 - fair1m test Evaluation - INFO - Fuzzy matched boing747 to boeing747
+2025/01/24 19:46:12 - fair1m test Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 19:46:12 - fair1m test Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 19:46:12 - fair1m test Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 19:46:13 - fair1m test Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 19:46:13 - fair1m test Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 19:46:13 - fair1m test Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 19:46:13 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:13 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:13 - fair1m test Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 19:46:13 - fair1m test Evaluation - INFO - Fuzzy matched large vehicle to trailer
+2025/01/24 19:46:13 - fair1m test Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 19:46:13 - fair1m test Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 19:46:13 - fair1m test Evaluation - INFO - Fuzzy matched golffield to football field
+2025/01/24 19:46:13 - fair1m test Evaluation - INFO - Fuzzy matched storagetank to van
+2025/01/24 19:46:13 - fair1m test Evaluation - INFO - Fuzzy matched golffield to football field
+2025/01/24 19:46:13 - fair1m test Evaluation - INFO - Fuzzy matched golffield to football field
+2025/01/24 19:46:13 - fair1m test Evaluation - INFO - Fuzzy matched dam to dump truck
+2025/01/24 19:46:13 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:13 - fair1m test Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 19:46:13 - fair1m test Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 19:46:13 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:13 - fair1m test Evaluation - INFO - Fuzzy matched overpass to passenger ship
+2025/01/24 19:46:13 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:13 - fair1m test Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 19:46:13 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:13 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:13 - fair1m test Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 19:46:13 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:13 - fair1m test Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 19:46:13 - fair1m test Evaluation - INFO - Fuzzy matched vehicle to van
+2025/01/24 19:46:13 - fair1m test Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 19:46:13 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:13 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:13 - fair1m test Evaluation - INFO - Fuzzy matched 787 to boeing787
+2025/01/24 19:46:13 - fair1m test Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 19:46:13 - fair1m test Evaluation - INFO - Fuzzy matched tractorractor to tractor
+2025/01/24 19:46:13 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:13 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:13 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:13 - fair1m test Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 19:46:13 - fair1m test Evaluation - INFO - Fuzzy matched intersectionsection to intersection
+2025/01/24 19:46:13 - fair1m test Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 19:46:13 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:14 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:14 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:14 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:14 - fair1m test Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 19:46:14 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:14 - fair1m test Evaluation - INFO - Fuzzy matched storagetank to van
+2025/01/24 19:46:14 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:14 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:14 - fair1m test Evaluation - INFO - Fuzzy matched 787 to boeing787
+2025/01/24 19:46:14 - fair1m test Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 19:46:14 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:14 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:14 - fair1m test Evaluation - INFO - Fuzzy matched golffield to football field
+2025/01/24 19:46:14 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:14 - fair1m test Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 19:46:14 - fair1m test Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 19:46:14 - fair1m test Evaluation - INFO - Fuzzy matched cellcell container to trailer
+2025/01/24 19:46:14 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:14 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:14 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:14 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:14 - fair1m test Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 19:46:14 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:14 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:14 - fair1m test Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 19:46:14 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:14 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:14 - fair1m test Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 19:46:14 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:14 - fair1m test Evaluation - INFO - Fuzzy matched storagetank to van
+2025/01/24 19:46:14 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:14 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:14 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:14 - fair1m test Evaluation - INFO - Fuzzy matched swimming pool to fishing boat
+2025/01/24 19:46:14 - fair1m test Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 19:46:14 - fair1m test Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 19:46:14 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:14 - fair1m test Evaluation - INFO - Fuzzy matched storage tank to van
+2025/01/24 19:46:14 - fair1m test Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 19:46:14 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:14 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:14 - fair1m test Evaluation - INFO - Fuzzy matched 330 to a330
+2025/01/24 19:46:14 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:15 - fair1m test Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 19:46:15 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:15 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:15 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:15 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:15 - fair1m test Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 19:46:15 - fair1m test Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 19:46:15 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:15 - fair1m test Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 19:46:15 - fair1m test Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 19:46:15 - fair1m test Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 19:46:15 - fair1m test Evaluation - INFO - Fuzzy matched vehicle to van
+2025/01/24 19:46:15 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:15 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:15 - fair1m test Evaluation - INFO - Fuzzy matched 330 to a330
+2025/01/24 19:46:15 - fair1m test Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 19:46:15 - fair1m test Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 19:46:15 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:15 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:15 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:15 - fair1m test Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 19:46:15 - fair1m test Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 19:46:15 - fair1m test Evaluation - INFO - Fuzzy matched busbus to bus
+2025/01/24 19:46:15 - fair1m test Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 19:46:15 - fair1m test Evaluation - INFO - Fuzzy matched dam to dump truck
+2025/01/24 19:46:15 - fair1m test Evaluation - INFO - Fuzzy matched swimming pool to fishing boat
+2025/01/24 19:46:15 - fair1m test Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 19:46:15 - fair1m test Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 19:46:15 - fair1m test Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 19:46:15 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:15 - fair1m test Evaluation - INFO - Fuzzy matched dam to dump truck
+2025/01/24 19:46:15 - fair1m test Evaluation - INFO - Fuzzy matched golffield to football field
+2025/01/24 19:46:15 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:15 - fair1m test Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 19:46:15 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:15 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:15 - fair1m test Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 19:46:15 - fair1m test Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 19:46:15 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:15 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:16 - fair1m test Evaluation - INFO - Fuzzy matched boing747 to boeing747
+2025/01/24 19:46:16 - fair1m test Evaluation - INFO - Fuzzy matched golffield to football field
+2025/01/24 19:46:16 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:16 - fair1m test Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 19:46:16 - fair1m test Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 19:46:16 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:16 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:16 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:16 - fair1m test Evaluation - INFO - Fuzzy matched warshiphip to warship
+2025/01/24 19:46:16 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:16 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:16 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:16 - fair1m test Evaluation - INFO - Fuzzy matched expressway toll station to tractor
+2025/01/24 19:46:16 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:16 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:16 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:16 - fair1m test Evaluation - INFO - Fuzzy matched boing747 to boeing747
+2025/01/24 19:46:16 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:16 - fair1m test Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 19:46:16 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:16 - fair1m test Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 19:46:16 - fair1m test Evaluation - INFO - Fuzzy matched dam to dump truck
+2025/01/24 19:46:16 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:16 - fair1m test Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 19:46:16 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:16 - fair1m test Evaluation - INFO - Fuzzy matched warshiphip to warship
+2025/01/24 19:46:16 - fair1m test Evaluation - INFO - Fuzzy matched baseball diamond to baseball field
+2025/01/24 19:46:16 - fair1m test Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 19:46:16 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:16 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:16 - fair1m test Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 19:46:16 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:16 - fair1m test Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 19:46:16 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:16 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:16 - fair1m test Evaluation - INFO - Fuzzy matched boe747 to boeing747
+2025/01/24 19:46:16 - fair1m test Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 19:46:16 - fair1m test Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 19:46:16 - fair1m test Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 19:46:16 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:16 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:16 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:16 - fair1m test Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 19:46:17 - fair1m test Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 19:46:17 - fair1m test Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 19:46:17 - fair1m test Evaluation - INFO - Fuzzy matched boe777 to boeing777
+2025/01/24 19:46:17 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:17 - fair1m test Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 19:46:17 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:17 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:17 - fair1m test Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 19:46:17 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:17 - fair1m test Evaluation - INFO - Fuzzy matched boing747 to boeing747
+2025/01/24 19:46:17 - fair1m test Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 19:46:17 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:17 - fair1m test Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 19:46:17 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:17 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:17 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:17 - fair1m test Evaluation - INFO - Fuzzy matched trainstation to trailer
+2025/01/24 19:46:17 - fair1m test Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 19:46:17 - fair1m test Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 19:46:17 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:17 - fair1m test Evaluation - INFO - Fuzzy matched golffield to football field
+2025/01/24 19:46:17 - fair1m test Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 19:46:17 - fair1m test Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 19:46:17 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched expressway toll station to tractor
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched dam to dump truck
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched intersectionsectionsection to intersection
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched helicopter to tennis court
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched expressway toll station to tractor
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched boe777 to boeing777
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched 787 to boeing787
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched dam to dump truck
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched boe to boeing737
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched boe ship to passenger ship
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched boeing73737 to boeing737
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched expressway toll station to tractor
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched airport to cargo truck
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched expressway toll station to tractor
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:18 - fair1m test Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched golffield to football field
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched storage tank to van
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched dam to dump truck
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched chimney to intersection
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched fishing boat boat to fishing boat
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched soccer ball field to baseball field
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched football field field to football field
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched 747 to boeing747
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched boing747 to boeing747
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched boe to boeing737
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched storagetank to van
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:19 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:20 - fair1m test Evaluation - INFO - Fuzzy matched 787 to boeing787
+2025/01/24 19:46:20 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:20 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:20 - fair1m test Evaluation - INFO - Fuzzy matched boeim787 to boeing787
+2025/01/24 19:46:20 - fair1m test Evaluation - INFO - Fuzzy matched baseball diamond to baseball field
+2025/01/24 19:46:20 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:20 - fair1m test Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 19:46:20 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:20 - fair1m test Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 19:46:20 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:20 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:20 - fair1m test Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 19:46:20 - fair1m test Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 19:46:20 - fair1m test Evaluation - INFO - Fuzzy matched boing747 to boeing747
+2025/01/24 19:46:20 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:20 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:20 - fair1m test Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 19:46:20 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:20 - fair1m test Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 19:46:20 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:20 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:20 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:20 - fair1m test Evaluation - INFO - Fuzzy matched swimming pool to fishing boat
+2025/01/24 19:46:20 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:20 - fair1m test Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 19:46:20 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:20 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:20 - fair1m test Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 19:46:20 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:20 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:20 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:20 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:20 - fair1m test Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 19:46:20 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:20 - fair1m test Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 19:46:20 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:20 - fair1m test Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 19:46:20 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:20 - fair1m test Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 19:46:20 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:20 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:20 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:20 - fair1m test Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 19:46:20 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:20 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched overpass to passenger ship
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched storagetank to van
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched boeing321 to boeing737
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched 787 to boeing787
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched 787 to boeing787
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched expressway toll station to tractor
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched storage tank to van
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched dam to dump truck
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched swimming pool to fishing boat
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched dredger to bridge
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched 747 to boeing747
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched expressway service area to arj21
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched boeing73737 to boeing737
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched golffield to football field
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched ground track field to dump truck
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched 787 to boeing787
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched golffield to football field
+2025/01/24 19:46:21 - fair1m test Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 19:46:22 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:22 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:22 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:22 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:22 - fair1m test Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 19:46:22 - fair1m test Evaluation - INFO - Fuzzy matched swimming pool to fishing boat
+2025/01/24 19:46:22 - fair1m test Evaluation - INFO - Fuzzy matched boeing73737 to boeing737
+2025/01/24 19:46:22 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:22 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:22 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:22 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:22 - fair1m test Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 19:46:22 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:22 - fair1m test Evaluation - INFO - Fuzzy matched windmill to intersection
+2025/01/24 19:46:22 - fair1m test Evaluation - INFO - Fuzzy matched windmill to intersection
+2025/01/24 19:46:22 - fair1m test Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 19:46:22 - fair1m test Evaluation - INFO - Fuzzy matched boing747 to boeing747
+2025/01/24 19:46:22 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:22 - fair1m test Evaluation - INFO - Fuzzy matched golffield to football field
+2025/01/24 19:46:22 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:22 - fair1m test Evaluation - INFO - Fuzzy matched airport to cargo truck
+2025/01/24 19:46:22 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:22 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:22 - fair1m test Evaluation - INFO - Fuzzy matched golffield to football field
+2025/01/24 19:46:22 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:22 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:22 - fair1m test Evaluation - INFO - Fuzzy matched golffield to football field
+2025/01/24 19:46:22 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:22 - fair1m test Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 19:46:22 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:22 - fair1m test Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 19:46:22 - fair1m test Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 19:46:22 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:22 - fair1m test Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 19:46:22 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:22 - fair1m test Evaluation - INFO - Fuzzy matched ground track field to dump truck
+2025/01/24 19:46:22 - fair1m test Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 19:46:22 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:22 - fair1m test Evaluation - INFO - Fuzzy matched boing747 to boeing747
+2025/01/24 19:46:22 - fair1m test Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 19:46:23 - fair1m test Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 19:46:23 - fair1m test Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 19:46:23 - fair1m test Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 19:46:23 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:23 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:23 - fair1m test Evaluation - INFO - Fuzzy matched container to trailer
+2025/01/24 19:46:23 - fair1m test Evaluation - INFO - Fuzzy matched storage tank to van
+2025/01/24 19:46:23 - fair1m test Evaluation - INFO - Fuzzy matched ground track field to dump truck
+2025/01/24 19:46:23 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:23 - fair1m test Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 19:46:23 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:23 - fair1m test Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 19:46:23 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:23 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:23 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:23 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:23 - fair1m test Evaluation - INFO - Fuzzy matched borbor to motorboat
+2025/01/24 19:46:23 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:23 - fair1m test Evaluation - INFO - Fuzzy matched dam to dump truck
+2025/01/24 19:46:23 - fair1m test Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 19:46:23 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:23 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:23 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:23 - fair1m test Evaluation - INFO - Fuzzy matched 787 to boeing787
+2025/01/24 19:46:23 - fair1m test Evaluation - INFO - Fuzzy matched storagetank to van
+2025/01/24 19:46:23 - fair1m test Evaluation - INFO - Fuzzy matched ground track field to dump truck
+2025/01/24 19:46:23 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:23 - fair1m test Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 19:46:23 - fair1m test Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 19:46:23 - fair1m test Evaluation - INFO - Fuzzy matched ground track field to dump truck
+2025/01/24 19:46:23 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:23 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:23 - fair1m test Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 19:46:23 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:23 - fair1m test Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 19:46:23 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:23 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:23 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:23 - fair1m test Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 19:46:23 - fair1m test Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 19:46:23 - fair1m test Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched dam to dump truck
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched expressway toll station to tractor
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched helicopter to tennis court
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched golffield to football field
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched expressway toll station to tractor
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched boing747 to boeing747
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched warshiphip to warship
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched boe to boeing737
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched ground track field to dump truck
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched boing747 to boeing747
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched golffield to football field
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched tractor tractor tractor tractor to tractor
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched storagetank to van
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched storage tank to van
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched vehicle to van
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched boing747 to boeing747
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched dam to dump truck
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched eing777 to boeing777
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:24 - fair1m test Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 19:46:25 - fair1m test Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 19:46:25 - fair1m test Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 19:46:25 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:25 - fair1m test Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 19:46:25 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:25 - fair1m test Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 19:46:25 - fair1m test Evaluation - INFO - Fuzzy matched swimming pool to fishing boat
+2025/01/24 19:46:25 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:25 - fair1m test Evaluation - INFO - Fuzzy matched airport to cargo truck
+2025/01/24 19:46:25 - fair1m test Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 19:46:25 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:25 - fair1m test Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 19:46:25 - fair1m test Evaluation - INFO - Fuzzy matched windmill to intersection
+2025/01/24 19:46:25 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:25 - fair1m test Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 19:46:25 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:25 - fair1m test Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 19:46:25 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:25 - fair1m test Evaluation - INFO - Fuzzy matched storagetank to van
+2025/01/24 19:46:25 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:25 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:25 - fair1m test Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 19:46:25 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:25 - fair1m test Evaluation - INFO - Fuzzy matched dam to dump truck
+2025/01/24 19:46:25 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:25 - fair1m test Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 19:46:25 - fair1m test Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 19:46:25 - fair1m test Evaluation - INFO - Fuzzy matched boe to boeing737
+2025/01/24 19:46:25 - fair1m test Evaluation - INFO - Fuzzy matched boe track field to football field
+2025/01/24 19:46:25 - fair1m test Evaluation - INFO - Fuzzy matched boe to boeing737
+2025/01/24 19:46:25 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:25 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:25 - fair1m test Evaluation - INFO - Fuzzy matched dam to dump truck
+2025/01/24 19:46:25 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:25 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:25 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:25 - fair1m test Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 19:46:25 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:25 - fair1m test Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 19:46:25 - fair1m test Evaluation - INFO - Fuzzy matched boeugboat to tugboat
+2025/01/24 19:46:25 - fair1m test Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 19:46:26 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:26 - fair1m test Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 19:46:26 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:26 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:26 - fair1m test Evaluation - INFO - Fuzzy matched vanicle to van
+2025/01/24 19:46:26 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:26 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:26 - fair1m test Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 19:46:26 - fair1m test Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 19:46:26 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:26 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:26 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:26 - fair1m test Evaluation - INFO - Fuzzy matched dam to dump truck
+2025/01/24 19:46:26 - fair1m test Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 19:46:26 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:26 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:26 - fair1m test Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 19:46:26 - fair1m test Evaluation - INFO - Fuzzy matched ground track field to dump truck
+2025/01/24 19:46:26 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:26 - fair1m test Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 19:46:26 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:26 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:26 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:26 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:26 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:26 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:26 - fair1m test Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 19:46:26 - fair1m test Evaluation - INFO - Fuzzy matched borbor to motorboat
+2025/01/24 19:46:26 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:26 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:26 - fair1m test Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 19:46:26 - fair1m test Evaluation - INFO - Fuzzy matched boing747 to boeing747
+2025/01/24 19:46:26 - fair1m test Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 19:46:26 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:26 - fair1m test Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 19:46:26 - fair1m test Evaluation - INFO - Fuzzy matched 330 to a330
+2025/01/24 19:46:26 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:26 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:26 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:26 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:26 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:26 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:26 - fair1m test Evaluation - INFO - Fuzzy matched airplane to van
+2025/01/24 19:46:26 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:26 - fair1m test Evaluation - INFO - Fuzzy matched airport to cargo truck
+2025/01/24 19:46:26 - fair1m test Evaluation - INFO - Fuzzy matched dam to dump truck
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched storage tank to van
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched swimming pool to fishing boat
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched dredger to bridge
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched busbusbus to bus
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched ground track field to dump truck
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched boe777 to boeing777
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched expressway toll station to tractor
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched boe777 to boeing777
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched cargo ship to dry cargo ship
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched dam to dump truck
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched baseball diamond to baseball field
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched baseball diamond to baseball field
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched boe747 to boeing747
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched 787 to boeing787
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched container to trailer
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:27 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:28 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:28 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:28 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:28 - fair1m test Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 19:46:28 - fair1m test Evaluation - INFO - Fuzzy matched tractorractor to tractor
+2025/01/24 19:46:28 - fair1m test Evaluation - INFO - Fuzzy matched soccer ball field to baseball field
+2025/01/24 19:46:28 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:28 - fair1m test Evaluation - INFO - Fuzzy matched eing787 to boeing787
+2025/01/24 19:46:28 - fair1m test Evaluation - INFO - Fuzzy matched golffield to football field
+2025/01/24 19:46:28 - fair1m test Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 19:46:28 - fair1m test Evaluation - INFO - Fuzzy matched expressway toll station to tractor
+2025/01/24 19:46:28 - fair1m test Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 19:46:28 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:28 - fair1m test Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 19:46:28 - fair1m test Evaluation - INFO - Fuzzy matched dam to dump truck
+2025/01/24 19:46:28 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:28 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:28 - fair1m test Evaluation - INFO - Fuzzy matched dam to dump truck
+2025/01/24 19:46:28 - fair1m test Evaluation - INFO - Fuzzy matched cargo ship to dry cargo ship
+2025/01/24 19:46:28 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:28 - fair1m test Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 19:46:28 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:28 - fair1m test Evaluation - INFO - Fuzzy matched boeug787 to boeing787
+2025/01/24 19:46:28 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:28 - fair1m test Evaluation - INFO - Fuzzy matched storagetank to van
+2025/01/24 19:46:28 - fair1m test Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 19:46:28 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:28 - fair1m test Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 19:46:28 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:28 - fair1m test Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 19:46:28 - fair1m test Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 19:46:28 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:28 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:28 - fair1m test Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 19:46:28 - fair1m test Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 19:46:28 - fair1m test Evaluation - INFO - Fuzzy matched dam to dump truck
+2025/01/24 19:46:28 - fair1m test Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 19:46:28 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:46:28 - fair1m test Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 19:46:28 - fair1m test Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 19:47:15 - fair1m test Evaluation - INFO - The submission file save at checkpoints/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2_2/eval_fair1m/parsed_answers_test/dota_Task1/dota_Task1.zip
+2025/01/24 19:47:15 - fair1m test Evaluation - INFO - --------------------
+
+2025/01/24 19:47:20 - fair1m test Evaluation - INFO - Dumped results to checkpoints/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2_2/eval_fair1m/parsed_answers_test/output.pkl
diff --git a/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m_2/eval_liqingyun@SH-IDC1-10-140-37-61_device0_rank0.log b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m_2/eval_liqingyun@SH-IDC1-10-140-37-61_device0_rank0.log
new file mode 100644
index 0000000000000000000000000000000000000000..8779722085fa6c1b524c1bce0ff339e2a02b8fab
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_fair1m_2/eval_liqingyun@SH-IDC1-10-140-37-61_device0_rank0.log
@@ -0,0 +1,1463 @@
+2025/01/24 20:27:42 - fair1m_2.0_train val Evaluation - INFO - --------------------
+2025/01/24 20:27:42 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 20:27:42 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:27:42 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:42 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:42 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:42 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 20:27:42 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:42 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:27:42 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 20:27:42 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 20:27:42 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:42 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched golffield to football field
+2025/01/24 20:27:42 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:42 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:42 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched aircraft to truck tractor
+2025/01/24 20:27:42 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe747 to boeing747
+2025/01/24 20:27:42 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:42 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:27:42 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:42 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:42 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:42 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:42 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:27:42 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:42 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:42 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:42 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:43 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:27:43 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:43 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:43 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:43 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:43 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:43 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched overpass to passenger ship
+2025/01/24 20:27:43 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:43 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:43 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:43 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:43 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 20:27:43 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:43 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:43 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched airplane to van
+2025/01/24 20:27:43 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 20:27:43 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:43 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:27:43 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 20:27:43 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:43 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing747 to boeing747
+2025/01/24 20:27:43 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:27:43 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:43 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:43 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:43 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:43 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 20:27:43 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:43 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 20:27:43 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:43 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched tractorractorractor to tractor
+2025/01/24 20:27:43 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched storage tank to van
+2025/01/24 20:27:43 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:43 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:43 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe747 to boeing747
+2025/01/24 20:27:43 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched car to liquid cargo ship
+2025/01/24 20:27:43 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:27:43 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:43 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:43 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 20:27:43 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:43 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:43 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:43 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:44 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:44 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:44 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:44 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:44 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ground track field to dump truck
+2025/01/24 20:27:44 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:44 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:44 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:44 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:44 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:44 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:27:44 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:44 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeway787 to boeing787
+2025/01/24 20:27:44 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:44 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 20:27:44 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing747 to boeing747
+2025/01/24 20:27:44 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:44 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched overpass to passenger ship
+2025/01/24 20:27:44 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:44 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:44 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 20:27:44 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:44 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeway787 to boeing787
+2025/01/24 20:27:44 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:27:44 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched swimming pool to fishing boat
+2025/01/24 20:27:44 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:44 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:44 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched eing777 to boeing777
+2025/01/24 20:27:44 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:27:44 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:44 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched airport to cargo truck
+2025/01/24 20:27:44 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:44 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:44 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:44 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:45 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:45 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 20:27:45 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:45 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched chimney to intersection
+2025/01/24 20:27:45 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:45 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:27:45 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:45 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:27:45 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:27:45 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:45 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe747 to boeing747
+2025/01/24 20:27:45 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:45 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:45 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:45 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 20:27:45 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched intersectionintersection to intersection
+2025/01/24 20:27:45 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing747 to boeing747
+2025/01/24 20:27:45 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 20:27:45 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:45 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched golffield to football field
+2025/01/24 20:27:45 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:27:45 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:27:45 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 20:27:45 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:45 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:45 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:45 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:45 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:45 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:45 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:45 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:45 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:45 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:45 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:45 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 20:27:45 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 20:27:45 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:46 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:46 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:46 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:46 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched golffield to football field
+2025/01/24 20:27:46 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched dredger to bridge
+2025/01/24 20:27:46 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched airport to cargo truck
+2025/01/24 20:27:46 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched storage tank to van
+2025/01/24 20:27:46 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 20:27:46 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:27:46 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:46 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:27:46 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:46 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:27:46 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:27:46 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched soccer ball field to baseball field
+2025/01/24 20:27:46 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 20:27:46 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched 787 to boeing787
+2025/01/24 20:27:46 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:46 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:27:46 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 20:27:46 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:46 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 20:27:46 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:27:46 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe777 to boeing777
+2025/01/24 20:27:46 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 20:27:46 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:46 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:46 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:46 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 20:27:46 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:46 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:46 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:46 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 20:27:47 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 20:27:47 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 20:27:47 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:47 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched 777 to boeing777
+2025/01/24 20:27:47 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:47 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:47 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:47 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched airplane to van
+2025/01/24 20:27:47 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched dam to dump truck
+2025/01/24 20:27:47 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:27:47 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeug787 to boeing787
+2025/01/24 20:27:47 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:47 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 20:27:47 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 20:27:47 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:47 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:47 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:47 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:27:47 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:47 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:27:47 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:27:47 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:47 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched dam to dump truck
+2025/01/24 20:27:47 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:27:47 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:27:47 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:47 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:27:47 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched dredger to bridge
+2025/01/24 20:27:47 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:47 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:47 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched expressway toll station to tractor
+2025/01/24 20:27:47 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:27:47 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe777 to boeing777
+2025/01/24 20:27:47 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:47 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe777 to boeing777
+2025/01/24 20:27:47 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:48 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:48 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:48 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:48 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:48 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 20:27:48 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe747 to boeing747
+2025/01/24 20:27:48 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:27:48 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:48 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:48 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:27:48 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 20:27:48 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:48 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:48 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:48 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:48 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:27:48 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:27:48 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:48 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:27:48 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:49 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 20:27:49 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe747 to boeing747
+2025/01/24 20:27:49 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:49 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:49 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched swimming pool to fishing boat
+2025/01/24 20:27:49 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:27:49 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:49 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 20:27:49 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:49 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:49 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:27:49 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:49 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:49 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:49 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:49 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:49 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:49 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 20:27:49 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:27:49 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeway777 to boeing777
+2025/01/24 20:27:49 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ground track field to dump truck
+2025/01/24 20:27:49 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:49 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched 330 to a330
+2025/01/24 20:27:49 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:49 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:27:49 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched airplane to van
+2025/01/24 20:27:49 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:49 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:49 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:49 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:49 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:49 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:49 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:27:49 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:49 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched 787 to boeing787
+2025/01/24 20:27:50 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:50 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:27:50 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:50 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:50 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 20:27:50 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:50 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:27:50 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:50 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:50 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing747 to boeing747
+2025/01/24 20:27:50 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched airplane to van
+2025/01/24 20:27:50 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:50 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:50 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:50 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched dam to dump truck
+2025/01/24 20:27:50 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 20:27:50 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:50 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched 330 to a330
+2025/01/24 20:27:50 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:27:50 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:50 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:50 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:50 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched airplane to van
+2025/01/24 20:27:50 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:50 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:50 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:50 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched engineering pool to engineering ship
+2025/01/24 20:27:50 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched golffield to football field
+2025/01/24 20:27:50 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 20:27:50 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:27:50 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:50 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:27:50 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:50 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:50 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:50 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:27:51 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:51 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:51 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:51 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:51 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:51 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:51 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:51 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 20:27:51 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched swimming pool to fishing boat
+2025/01/24 20:27:51 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:51 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:51 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 20:27:51 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:51 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:27:51 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 20:27:51 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:51 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:27:51 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:51 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched dam to dump truck
+2025/01/24 20:27:51 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 20:27:51 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:51 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched soccer ball field to baseball field
+2025/01/24 20:27:51 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched dredger to bridge
+2025/01/24 20:27:51 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched soccer ball field to baseball field
+2025/01/24 20:27:51 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:27:51 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:51 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:27:51 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched dredger to bridge
+2025/01/24 20:27:51 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:51 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:27:51 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:27:52 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:52 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:52 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 20:27:52 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 20:27:52 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:52 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:52 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:52 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:52 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 20:27:52 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:52 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:27:52 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:52 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe to boeing737
+2025/01/24 20:27:52 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:52 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 20:27:52 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:52 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:52 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched eing777 to boeing777
+2025/01/24 20:27:52 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:52 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:27:52 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 20:27:52 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:52 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing747 to boeing747
+2025/01/24 20:27:52 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:52 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:52 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched container to trailer
+2025/01/24 20:27:52 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:52 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:52 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:27:52 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:52 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched eing777 to boeing777
+2025/01/24 20:27:52 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:52 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:52 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched airport to cargo truck
+2025/01/24 20:27:52 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:52 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:52 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 20:27:52 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:27:53 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:53 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:53 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:53 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:53 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched airport to cargo truck
+2025/01/24 20:27:53 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing747 to boeing747
+2025/01/24 20:27:53 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:53 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 20:27:53 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:53 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:53 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:53 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:53 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:53 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:53 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched windmill to intersection
+2025/01/24 20:27:53 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing747 to boeing747
+2025/01/24 20:27:53 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:53 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched fishing truck to fishing boat
+2025/01/24 20:27:53 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 20:27:53 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched 787 to boeing787
+2025/01/24 20:27:53 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:53 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched golffield to football field
+2025/01/24 20:27:53 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched stadium to bus
+2025/01/24 20:27:53 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:27:53 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched swimming pool to fishing boat
+2025/01/24 20:27:53 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:53 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:53 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:53 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:27:53 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched golffield to football field
+2025/01/24 20:27:53 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:53 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:53 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 20:27:53 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:53 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:53 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing747 to boeing747
+2025/01/24 20:27:53 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:54 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:54 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 20:27:54 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:54 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:54 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched dam to dump truck
+2025/01/24 20:27:54 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:54 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched eing787 to boeing787
+2025/01/24 20:27:54 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:54 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe747 to boeing747
+2025/01/24 20:27:54 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched dam to dump truck
+2025/01/24 20:27:54 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:27:54 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeway777 to boeing777
+2025/01/24 20:27:54 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched swimming pool to fishing boat
+2025/01/24 20:27:54 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:54 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:54 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched vanicle to van
+2025/01/24 20:27:54 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:54 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 20:27:54 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:54 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:54 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:54 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:27:54 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:54 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched dam to dump truck
+2025/01/24 20:27:54 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:54 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:27:54 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:27:54 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:54 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:54 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:54 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:54 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:54 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:54 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:54 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:54 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:54 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched traileriler to trailer
+2025/01/24 20:27:54 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:54 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:55 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched airport to cargo truck
+2025/01/24 20:27:55 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:55 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:55 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:27:55 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:55 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 20:27:55 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 20:27:55 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:55 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:27:55 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeugboat to tugboat
+2025/01/24 20:27:55 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:55 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:55 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched storage tank to van
+2025/01/24 20:27:55 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ground track field to dump truck
+2025/01/24 20:27:55 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:55 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 20:27:55 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 20:27:55 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:55 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:55 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 20:27:55 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:55 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:55 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:27:55 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched airplane to van
+2025/01/24 20:27:55 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe747 to boeing747
+2025/01/24 20:27:55 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:55 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:55 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 20:27:55 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:55 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:55 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched windmill to intersection
+2025/01/24 20:27:55 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched airplane to van
+2025/01/24 20:27:55 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:55 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:56 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 20:27:56 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched airplane to van
+2025/01/24 20:27:56 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:56 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched golffield to football field
+2025/01/24 20:27:56 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched 350 to a350
+2025/01/24 20:27:56 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:56 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:56 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:56 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:56 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched dam to dump truck
+2025/01/24 20:27:56 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:56 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 20:27:56 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:27:56 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:27:56 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe to boeing737
+2025/01/24 20:27:56 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe ship to passenger ship
+2025/01/24 20:27:56 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched swimming pool to fishing boat
+2025/01/24 20:27:56 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:56 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:56 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:27:56 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 20:27:56 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:56 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:56 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:56 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 20:27:56 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:56 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 20:27:56 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 20:27:56 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:56 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:27:56 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe ship to passenger ship
+2025/01/24 20:27:56 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 20:27:56 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched airplane to van
+2025/01/24 20:27:56 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:56 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched airplane to van
+2025/01/24 20:27:56 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 20:27:56 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:56 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:57 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:57 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:57 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:57 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched busbus to bus
+2025/01/24 20:27:57 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched 350 to a350
+2025/01/24 20:27:57 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched storagetank to van
+2025/01/24 20:27:57 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:57 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:57 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:57 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:27:57 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 20:27:57 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched dam to dump truck
+2025/01/24 20:27:57 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:27:57 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:27:57 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 20:27:57 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:57 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:57 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:27:57 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:57 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 20:27:57 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:57 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 20:27:57 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:57 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:57 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:57 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:57 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 20:27:57 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 20:27:57 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:57 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:27:57 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:57 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:57 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:57 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 20:27:58 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:58 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:58 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 20:27:58 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:58 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:58 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:58 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:58 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched container to trailer
+2025/01/24 20:27:58 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched dam to dump truck
+2025/01/24 20:27:58 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:58 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:58 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:58 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:58 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:58 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:58 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:58 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:58 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:58 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:27:58 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:58 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched airplane to van
+2025/01/24 20:27:58 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:58 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:58 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:58 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:58 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:27:58 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:58 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched liquid cargo cargo ship to liquid cargo ship
+2025/01/24 20:27:58 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:58 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:58 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:58 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:27:58 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:58 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 20:27:58 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:58 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 20:27:58 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 20:27:59 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:59 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:59 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:59 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched dam to dump truck
+2025/01/24 20:27:59 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:59 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:59 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:59 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 20:27:59 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:59 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched dam to dump truck
+2025/01/24 20:27:59 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:27:59 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched golffield to football field
+2025/01/24 20:27:59 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:59 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched airplane to van
+2025/01/24 20:27:59 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:59 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:27:59 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:59 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched stadium to bus
+2025/01/24 20:27:59 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched golffield to football field
+2025/01/24 20:27:59 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:59 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe747 to boeing747
+2025/01/24 20:27:59 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:59 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:59 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:59 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:59 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:59 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:27:59 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:59 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 20:27:59 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:27:59 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:27:59 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 20:27:59 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:27:59 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:59 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:59 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:59 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing747 to boeing747
+2025/01/24 20:27:59 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:59 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 20:27:59 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:59 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:59 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched overpass to passenger ship
+2025/01/24 20:27:59 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:27:59 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 20:27:59 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:27:59 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:28:00 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:00 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:00 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched 330 to a330
+2025/01/24 20:28:00 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:00 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:00 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 20:28:00 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:28:00 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:28:00 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:00 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:00 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:00 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:28:00 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:00 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:00 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:00 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:28:00 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:28:00 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe777 to boeing777
+2025/01/24 20:28:00 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:00 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:28:00 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:01 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:01 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:01 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:28:01 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 20:28:01 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:01 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:01 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched dam to dump truck
+2025/01/24 20:28:01 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:01 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:28:01 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe777 to boeing777
+2025/01/24 20:28:01 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:01 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing747 to boeing747
+2025/01/24 20:28:01 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:28:01 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:01 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched dredger to bridge
+2025/01/24 20:28:01 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:28:01 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched storage tank to van
+2025/01/24 20:28:01 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:01 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched golffield to football field
+2025/01/24 20:28:01 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:01 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ground track field to dump truck
+2025/01/24 20:28:01 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:01 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched 777 to boeing777
+2025/01/24 20:28:01 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:01 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:01 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:01 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:28:01 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:01 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:01 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 20:28:01 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:28:01 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:01 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 20:28:01 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:01 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:01 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:01 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched airport to cargo truck
+2025/01/24 20:28:01 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 20:28:01 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:28:01 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 20:28:01 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:02 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:02 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 20:28:02 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:02 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:02 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:02 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched dam to dump truck
+2025/01/24 20:28:02 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:02 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:28:02 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:28:02 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched airport to cargo truck
+2025/01/24 20:28:02 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:02 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:28:02 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:02 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:28:02 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:28:02 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe to boeing737
+2025/01/24 20:28:02 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:02 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing73737 to boeing737
+2025/01/24 20:28:02 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 20:28:02 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:02 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:28:02 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:02 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:02 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 20:28:02 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:02 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched golffield to football field
+2025/01/24 20:28:02 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:02 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:02 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:28:02 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:28:03 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:28:03 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:28:03 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:28:03 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:03 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:03 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 20:28:03 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:28:03 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:03 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 20:28:03 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:03 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:03 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched airplane to van
+2025/01/24 20:28:03 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:03 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched busbus to bus
+2025/01/24 20:28:03 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:28:03 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:03 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched airplane to van
+2025/01/24 20:28:03 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched eing777 to boeing777
+2025/01/24 20:28:03 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:03 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:03 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched overpass to passenger ship
+2025/01/24 20:28:03 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:28:03 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:28:03 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 20:28:03 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:03 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:03 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:03 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:03 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched airport to cargo truck
+2025/01/24 20:28:03 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:04 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:04 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched airplane to van
+2025/01/24 20:28:04 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:04 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe747 to boeing747
+2025/01/24 20:28:04 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 20:28:04 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched 747 to boeing747
+2025/01/24 20:28:04 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:04 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:04 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:04 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 20:28:04 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:04 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched car to liquid cargo ship
+2025/01/24 20:28:04 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:04 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:04 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:28:04 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:28:04 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:28:04 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:04 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:04 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:28:04 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 20:28:04 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:04 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:28:04 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:28:04 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeug777 to boeing777
+2025/01/24 20:28:04 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:04 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:28:04 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:04 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:28:04 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched airplane to van
+2025/01/24 20:28:04 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:28:04 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:04 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched fishing truck to fishing boat
+2025/01/24 20:28:04 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 20:28:04 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:28:04 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:04 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:04 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:04 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:04 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 20:28:04 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:04 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 20:28:04 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 20:28:04 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:04 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 20:28:05 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:05 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:05 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:05 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:05 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:28:05 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeway777 to boeing777
+2025/01/24 20:28:05 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:05 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched swimming pool to fishing boat
+2025/01/24 20:28:05 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:05 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeway787 to boeing787
+2025/01/24 20:28:05 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:05 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeim787 to boeing787
+2025/01/24 20:28:05 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched 777 to boeing777
+2025/01/24 20:28:05 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:05 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:05 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:05 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:05 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:05 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:28:05 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:05 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:05 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched dam to dump truck
+2025/01/24 20:28:05 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:05 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:05 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:05 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:05 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:05 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:05 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:05 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:05 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched trainstation to trailer
+2025/01/24 20:28:05 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched 777 to boeing777
+2025/01/24 20:28:05 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:28:05 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:05 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched airplane to van
+2025/01/24 20:28:05 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 20:28:05 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:05 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:28:05 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeim777 to boeing777
+2025/01/24 20:28:05 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:05 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:05 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched airport to cargo truck
+2025/01/24 20:28:05 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:05 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched overpass to passenger ship
+2025/01/24 20:28:05 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:05 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:05 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:05 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:06 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched 350 to a350
+2025/01/24 20:28:06 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:06 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing747 to boeing747
+2025/01/24 20:28:06 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:06 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 20:28:06 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 20:28:06 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched airplane to van
+2025/01/24 20:28:06 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched airport to cargo truck
+2025/01/24 20:28:06 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:06 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:06 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:06 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:06 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:28:06 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:06 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:06 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched aircraft to truck tractor
+2025/01/24 20:28:06 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:06 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:06 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:06 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:06 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:28:06 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:28:06 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:06 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 20:28:06 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:06 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 20:28:06 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:28:06 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 20:28:06 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched vanicle to van
+2025/01/24 20:28:06 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:06 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:06 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:06 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:28:06 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:28:06 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:28:06 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:06 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:06 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:06 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:07 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:28:07 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeim777 to boeing777
+2025/01/24 20:28:07 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeway787 to boeing787
+2025/01/24 20:28:07 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:28:07 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:28:07 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:07 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:07 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:28:07 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched storagetank to van
+2025/01/24 20:28:07 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:28:07 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:07 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:07 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:07 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:07 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe747 to boeing747
+2025/01/24 20:28:07 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:07 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 20:28:07 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched airplane to van
+2025/01/24 20:28:07 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:28:07 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe to boeing737
+2025/01/24 20:28:07 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:07 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 20:28:07 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:28:07 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 20:28:07 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:07 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched container to trailer
+2025/01/24 20:28:07 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:07 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe to boeing737
+2025/01/24 20:28:07 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe to boeing737
+2025/01/24 20:28:07 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched airport to cargo truck
+2025/01/24 20:28:07 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 20:28:07 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:07 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:28:07 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe777 to boeing777
+2025/01/24 20:28:08 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:08 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched container to trailer
+2025/01/24 20:28:08 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:08 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 20:28:08 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:28:08 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeway777 to boeing777
+2025/01/24 20:28:08 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:08 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:08 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:08 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:28:08 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:08 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:08 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched fishing pool to fishing boat
+2025/01/24 20:28:08 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:08 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:08 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:08 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 20:28:08 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:08 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:08 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:08 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 20:28:08 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:28:08 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:08 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:08 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 20:28:08 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched airplane to van
+2025/01/24 20:28:08 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched trainstation to trailer
+2025/01/24 20:28:08 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:08 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:28:08 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ground track field to dump truck
+2025/01/24 20:28:08 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:08 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:08 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:08 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:09 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:28:09 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe777 to boeing777
+2025/01/24 20:28:09 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched bo to boeing737
+2025/01/24 20:28:09 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:28:09 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing747 to boeing747
+2025/01/24 20:28:09 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:09 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeway toll station to tractor
+2025/01/24 20:28:09 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe to boeing737
+2025/01/24 20:28:09 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeimimotorboat to motorboat
+2025/01/24 20:28:09 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:09 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:09 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 20:28:09 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:09 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched aircraft to truck tractor
+2025/01/24 20:28:09 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:09 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:28:09 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 20:28:09 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched container to trailer
+2025/01/24 20:28:09 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:28:09 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:09 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:28:09 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:09 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing73737 to boeing737
+2025/01/24 20:28:09 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:09 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:09 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ground track field to dump truck
+2025/01/24 20:28:09 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:09 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:28:09 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:09 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:28:09 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:09 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched 330 to a330
+2025/01/24 20:28:09 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:10 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:28:10 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:10 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:10 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:10 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched aircraft to truck tractor
+2025/01/24 20:28:10 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:10 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:10 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:10 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched airport to cargo truck
+2025/01/24 20:28:10 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:10 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:10 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:28:10 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:10 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:10 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:10 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:28:10 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing747 to boeing747
+2025/01/24 20:28:10 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:10 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:10 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:10 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe747 to boeing747
+2025/01/24 20:28:10 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:10 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:10 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:11 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 20:28:11 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 20:28:11 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:28:11 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:11 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched airport to cargo truck
+2025/01/24 20:28:11 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:28:11 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 20:28:11 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:11 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:11 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:11 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:11 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:11 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched 330 to a330
+2025/01/24 20:28:11 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:11 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 20:28:11 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:11 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:11 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:11 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched golffield to football field
+2025/01/24 20:28:11 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched soccer ball field to baseball field
+2025/01/24 20:28:11 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:11 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:11 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 20:28:11 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:11 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:28:11 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:28:11 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:11 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 20:28:11 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched fishing pool to fishing boat
+2025/01/24 20:28:11 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:11 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:11 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:11 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:11 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 20:28:11 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:12 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:12 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:12 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:12 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:12 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 20:28:12 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched overpass to passenger ship
+2025/01/24 20:28:12 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 20:28:12 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:12 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 20:28:12 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:28:12 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:28:12 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe777 to boeing777
+2025/01/24 20:28:12 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched 747 to boeing747
+2025/01/24 20:28:12 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:12 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:28:12 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:28:12 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:12 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:12 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:12 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:12 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:12 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:28:12 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:12 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:12 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:28:12 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:12 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 20:28:12 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:13 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:28:13 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeway777 to boeing777
+2025/01/24 20:28:13 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:13 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:28:13 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched airplane to van
+2025/01/24 20:28:13 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:13 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:13 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing747 to boeing747
+2025/01/24 20:28:13 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:13 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched warshiphiphip to warship
+2025/01/24 20:28:13 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 20:28:13 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched airport to cargo truck
+2025/01/24 20:28:13 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:13 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:13 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeug787 to boeing787
+2025/01/24 20:28:13 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 20:28:13 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:28:13 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing747 to boeing747
+2025/01/24 20:28:13 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:13 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 20:28:13 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:13 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched large vehicle to trailer
+2025/01/24 20:28:13 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:13 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched golffield to football field
+2025/01/24 20:28:13 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched soccer ball field to baseball field
+2025/01/24 20:28:13 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:13 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:13 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:13 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe to boeing737
+2025/01/24 20:28:13 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 20:28:13 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 20:28:13 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:13 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing747 to boeing747
+2025/01/24 20:28:13 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:13 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 20:28:13 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:13 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 20:28:13 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:13 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:13 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:13 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:28:13 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:14 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:14 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:14 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:14 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:14 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:14 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:14 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:14 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:14 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeway737 to boeing737
+2025/01/24 20:28:14 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:14 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:14 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:14 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 20:28:14 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:14 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:28:14 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:28:14 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 20:28:14 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:28:14 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:28:14 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:14 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:28:14 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:28:15 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:15 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 20:28:15 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:15 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 20:28:15 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:15 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched golffield to football field
+2025/01/24 20:28:15 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:15 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:28:15 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched airplane to van
+2025/01/24 20:28:15 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:15 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:15 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched baseball field field field to baseball field
+2025/01/24 20:28:15 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 20:28:15 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:15 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:15 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe747 to boeing747
+2025/01/24 20:28:15 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:15 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:28:15 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe777 to boeing777
+2025/01/24 20:28:15 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:15 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:28:15 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:15 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:15 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:15 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:15 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:28:16 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:16 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:16 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:16 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:16 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:16 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:16 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched golffield to football field
+2025/01/24 20:28:16 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:16 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched swimming pool to fishing boat
+2025/01/24 20:28:16 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:16 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:16 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:16 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:16 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:16 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:16 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe747 to boeing747
+2025/01/24 20:28:16 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:16 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 20:28:16 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:28:16 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe777 to boeing777
+2025/01/24 20:28:16 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:16 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:16 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 20:28:16 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched 787 to boeing787
+2025/01/24 20:28:16 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:16 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:16 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 20:28:16 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:16 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched 747 to boeing747
+2025/01/24 20:28:16 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:16 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:16 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:16 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 20:28:16 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:28:16 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:16 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:16 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:17 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe777 to boeing777
+2025/01/24 20:28:17 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 20:28:17 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:28:17 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:17 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:17 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:17 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 20:28:17 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:17 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe777 to boeing777
+2025/01/24 20:28:17 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 20:28:17 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:17 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:17 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 20:28:17 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:28:17 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:17 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:17 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:17 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 20:28:17 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:17 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:17 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:17 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:17 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 20:28:17 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched eing777 to boeing777
+2025/01/24 20:28:17 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:28:17 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:28:17 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:28:17 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeim777 to boeing777
+2025/01/24 20:28:17 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:28:17 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:17 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:17 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:17 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 20:28:17 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:17 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:17 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:17 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:28:17 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 20:28:18 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:28:18 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched 777 to boeing777
+2025/01/24 20:28:18 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:18 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched 330 to a330
+2025/01/24 20:28:18 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:28:18 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 20:28:18 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:18 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:18 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:18 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:18 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe to boeing737
+2025/01/24 20:28:18 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 20:28:18 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:28:18 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:28:18 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:18 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched windmill to intersection
+2025/01/24 20:28:18 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:28:18 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:18 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:18 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 20:28:18 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:18 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:18 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:18 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:28:18 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:18 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:28:18 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing747 to boeing747
+2025/01/24 20:28:18 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:18 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:28:18 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:18 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:18 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:18 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 20:28:18 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:19 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:19 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 20:28:19 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:19 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 20:28:19 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:19 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 20:28:19 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:19 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:28:19 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:19 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe777 to boeing777
+2025/01/24 20:28:19 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:19 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:19 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:19 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:19 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:28:19 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:19 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:28:19 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:19 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:19 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:19 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:28:19 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:19 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:19 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:19 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:19 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:19 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:19 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:19 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:19 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:19 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:19 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:28:19 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeway777 to boeing777
+2025/01/24 20:28:19 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:19 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:19 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:19 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeim787 to boeing787
+2025/01/24 20:28:19 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched container to trailer
+2025/01/24 20:28:19 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:19 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 20:28:19 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:19 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 20:28:19 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:19 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 20:28:19 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched plane to van
+2025/01/24 20:28:19 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:19 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:20 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:20 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 20:28:20 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing747 to boeing747
+2025/01/24 20:28:20 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:20 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:28:20 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:20 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:20 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ground track field to dump truck
+2025/01/24 20:28:20 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:20 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:20 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:20 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:20 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:20 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:20 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:20 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:20 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:20 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:28:20 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:20 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 20:28:20 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:28:20 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:20 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 20:28:20 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched dam to dump truck
+2025/01/24 20:28:20 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:20 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:20 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 20:28:20 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched airplane to van
+2025/01/24 20:28:20 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:28:20 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeway777 to boeing777
+2025/01/24 20:28:20 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe to boeing737
+2025/01/24 20:28:20 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeimimney to boeing737
+2025/01/24 20:28:20 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:20 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched dry cargo cargo ship to dry cargo ship
+2025/01/24 20:28:20 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:20 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:20 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:20 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:20 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:28:20 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched aircraft to truck tractor
+2025/01/24 20:28:20 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing747 to boeing747
+2025/01/24 20:28:21 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched dredger to bridge
+2025/01/24 20:28:21 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:28:21 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:21 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:21 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:21 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:21 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:21 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:21 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:21 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 20:28:21 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:21 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 20:28:21 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:21 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched harbor to excavator
+2025/01/24 20:28:21 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched fishing truck to fishing boat
+2025/01/24 20:28:21 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:21 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:21 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:21 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:21 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:21 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:21 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:21 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:21 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:21 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing747 to boeing747
+2025/01/24 20:28:21 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeruck tractor to tractor
+2025/01/24 20:28:21 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:21 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:21 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:21 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched airplane to van
+2025/01/24 20:28:21 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:21 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:21 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched dam to dump truck
+2025/01/24 20:28:21 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 20:28:21 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:21 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:21 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:22 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:22 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boe787 to boeing787
+2025/01/24 20:28:22 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:22 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing787 to boeing787
+2025/01/24 20:28:22 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing777 to boeing777
+2025/01/24 20:28:22 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:22 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched 350 to a350
+2025/01/24 20:28:22 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boing777 to boeing777
+2025/01/24 20:28:22 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeing7373737 to boeing737
+2025/01/24 20:28:22 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:22 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched boeeing787 to boeing787
+2025/01/24 20:28:22 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:22 - fair1m_2.0_train val Evaluation - INFO - Fuzzy matched ship to passenger ship
+2025/01/24 20:28:26 - fair1m_2.0_train val Evaluation - INFO -
+---------------iou_thr: 0.5---------------
+2025/01/24 20:29:53 - fair1m_2.0_train val Evaluation - INFO -
++-------------------+--------+--------+--------+-------+
+| class | gts | dets | recall | ap |
++-------------------+--------+--------+--------+-------+
+| Boeing737 | 3616 | 2828 | 0.254 | 0.123 |
+| Boeing747 | 1595 | 1530 | 0.800 | 0.685 |
+| Boeing777 | 588 | 1039 | 0.240 | 0.043 |
+| Boeing787 | 1265 | 985 | 0.349 | 0.183 |
+| C919 | 51 | 68 | 0.235 | 0.060 |
+| A220 | 4159 | 7146 | 0.700 | 0.334 |
+| A321 | 2058 | 2229 | 0.600 | 0.318 |
+| A330 | 1123 | 1274 | 0.577 | 0.326 |
+| A350 | 677 | 600 | 0.644 | 0.498 |
+| ARJ21 | 264 | 361 | 0.235 | 0.077 |
+| Passenger Ship | 2293 | 1450 | 0.200 | 0.087 |
+| Motorboat | 13483 | 14175 | 0.549 | 0.296 |
+| Fishing Boat | 7098 | 9795 | 0.474 | 0.174 |
+| Tugboat | 783 | 1200 | 0.396 | 0.165 |
+| Engineering Ship | 4447 | 2296 | 0.286 | 0.198 |
+| Liquid Cargo Ship | 1686 | 2239 | 0.457 | 0.176 |
+| Dry Cargo Ship | 14234 | 16268 | 0.629 | 0.398 |
+| Warship | 981 | 801 | 0.502 | 0.384 |
+| Small Car | 112470 | 116999 | 0.723 | 0.535 |
+| Bus | 1179 | 1367 | 0.500 | 0.264 |
+| Cargo Truck | 12789 | 11551 | 0.526 | 0.342 |
+| Dump Truck | 14587 | 14047 | 0.568 | 0.328 |
+| Van | 121631 | 115675 | 0.686 | 0.472 |
+| Trailer | 1044 | 1081 | 0.245 | 0.088 |
+| Tractor | 66 | 133 | 0.061 | 0.030 |
+| Excavator | 708 | 672 | 0.429 | 0.259 |
+| Truck Tractor | 707 | 194 | 0.168 | 0.126 |
+| Basketball Court | 765 | 1031 | 0.655 | 0.329 |
+| Tennis Court | 2562 | 2569 | 0.835 | 0.700 |
+| Football Field | 835 | 1596 | 0.601 | 0.202 |
+| Baseball Field | 798 | 798 | 0.865 | 0.728 |
+| Intersection | 6805 | 8549 | 0.519 | 0.229 |
+| Roundabout | 471 | 555 | 0.682 | 0.421 |
+| Bridge | 1192 | 2755 | 0.492 | 0.102 |
++-------------------+--------+--------+--------+-------+
+| mAP | | | | 0.285 |
++-------------------+--------+--------+--------+-------+
+2025/01/24 20:29:53 - fair1m_2.0_train val Evaluation - INFO - --------------------
+
+2025/01/24 20:30:04 - fair1m_2.0_train val Evaluation - INFO - Dumped results to checkpoints/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2_2/eval_fair1m_2/output.pkl
diff --git a/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_rsar/parsed_answers_test/eval_liqingyun@SH-IDC1-10-140-37-115_device0_rank0.log b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_rsar/parsed_answers_test/eval_liqingyun@SH-IDC1-10-140-37-115_device0_rank0.log
new file mode 100644
index 0000000000000000000000000000000000000000..b066449e2dffc742ae1b0fd64e30acf8194c65ce
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_rsar/parsed_answers_test/eval_liqingyun@SH-IDC1-10-140-37-115_device0_rank0.log
@@ -0,0 +1,22 @@
+2025/01/25 00:29:41 - rsar test Evaluation - INFO - --------------------
+2025/01/25 00:29:43 - rsar test Evaluation - INFO - Fuzzy matched trainstation to tank
+2025/01/25 00:29:43 - rsar test Evaluation - INFO - Fuzzy matched baseball field to car
+2025/01/25 00:29:46 - rsar test Evaluation - INFO - Fuzzy matched trainstation to tank
+2025/01/25 00:29:51 - rsar test Evaluation - INFO -
+---------------iou_thr: 0.5---------------
+2025/01/25 00:30:30 - rsar test Evaluation - INFO -
++----------+-------+-------+--------+-------+
+| class | gts | dets | recall | ap |
++----------+-------+-------+--------+-------+
+| ship | 21394 | 21534 | 0.871 | 0.722 |
+| aircraft | 988 | 945 | 0.745 | 0.602 |
+| car | 2462 | 2413 | 0.926 | 0.868 |
+| tank | 1026 | 1006 | 0.783 | 0.619 |
+| bridge | 3388 | 3354 | 0.625 | 0.422 |
+| harbor | 399 | 399 | 0.762 | 0.615 |
++----------+-------+-------+--------+-------+
+| mAP | | | | 0.641 |
++----------+-------+-------+--------+-------+
+2025/01/25 00:30:30 - rsar test Evaluation - INFO - --------------------
+
+2025/01/25 00:30:36 - rsar test Evaluation - INFO - Dumped results to checkpoints/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2_2/eval_rsar/parsed_answers_test/output.pkl
diff --git a/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_srsdd/parsed_answers_test.json b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_srsdd/parsed_answers_test.json
new file mode 100644
index 0000000000000000000000000000000000000000..487d0b6f2079417bc9e81d80d5e5646bcade69c5
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_srsdd/parsed_answers_test.json
@@ -0,0 +1,6840 @@
+{
+ "L634.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 316.9280090332031,
+ 857.6000366210938,
+ 212.48001098632812,
+ 985.6000366210938,
+ 246.2720184326172,
+ 1009.1520385742188,
+ 348.6720275878906,
+ 881.1520385742188
+ ]
+ ]
+ ],
+ "labels": [
+ "ore oil"
+ ]
+ }
+ },
+ "L656.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 392.7040100097656,
+ 270.8480224609375,
+ 316.9280090332031,
+ 346.6240234375,
+ 336.3840026855469,
+ 366.08001708984375,
+ 412.1600341796875,
+ 291.3280029296875
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L600.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 909.8240356445312,
+ 835.072021484375,
+ 905.72802734375,
+ 849.4080200195312,
+ 945.6640625,
+ 859.6480712890625,
+ 950.7840576171875,
+ 845.31201171875
+ ],
+ [
+ 906.7520141601562,
+ 851.4560546875,
+ 903.6800537109375,
+ 861.696044921875,
+ 943.6160278320312,
+ 872.9600219726562,
+ 946.6880493164062,
+ 861.696044921875
+ ],
+ [
+ 902.6560668945312,
+ 862.7200317382812,
+ 899.5840454101562,
+ 872.9600219726562,
+ 941.5680541992188,
+ 884.2240600585938,
+ 944.6400146484375,
+ 872.9600219726562
+ ],
+ [
+ 897.5360717773438,
+ 873.9840698242188,
+ 894.4640502929688,
+ 884.2240600585938,
+ 934.4000244140625,
+ 895.488037109375,
+ 937.4720458984375,
+ 885.248046875
+ ],
+ [
+ 892.416015625,
+ 884.2240600585938,
+ 888.320068359375,
+ 895.488037109375,
+ 926.2080688476562,
+ 906.7520141601562,
+ 929.280029296875,
+ 895.488037109375
+ ],
+ [
+ 884.2240600585938,
+ 886.2720336914062,
+ 881.1520385742188,
+ 895.488037109375,
+ 920.0640258789062,
+ 906.7520141601562,
+ 923.1360473632812,
+ 896.5120239257812
+ ],
+ [
+ 884.2240600585938,
+ 896.5120239257812,
+ 881.1520385742188,
+ 906.7520141601562,
+ 918.0160522460938,
+ 918.0160522460938,
+ 921.0880737304688,
+ 907.7760620117188
+ ],
+ [
+ 838.14404296875,
+ 907.7760620117188,
+ 835.072021484375,
+ 918.0160522460938,
+ 870.9120483398438,
+ 929.280029296875,
+ 872.9600219726562,
+ 919.0400390625
+ ],
+ [
+ 341.5040283203125,
+ 945.6640625,
+ 333.31201171875,
+ 962.0480346679688,
+ 387.5840148925781,
+ 991.7440185546875,
+ 395.7760314941406,
+ 976.384033203125
+ ],
+ [
+ 425.4720153808594,
+ 994.8160400390625,
+ 417.280029296875,
+ 1010.176025390625,
+ 458.2400207519531,
+ 1023.488037109375,
+ 465.40802001953125,
+ 1017.3440551757812
+ ]
+ ],
+ [
+ [
+ 171.52000427246094,
+ 742.9120483398438,
+ 167.42401123046875,
+ 756.2240600585938,
+ 211.4560089111328,
+ 771.5840454101562,
+ 215.55201721191406,
+ 759.2960205078125
+ ],
+ [
+ 84.4800033569336,
+ 825.8560180664062,
+ 82.4320068359375,
+ 840.1920166015625,
+ 148.99200439453125,
+ 850.4320678710938,
+ 151.04000854492188,
+ 836.0960693359375
+ ],
+ [
+ 85.5040054321289,
+ 840.1920166015625,
+ 84.4800033569336,
+ 852.4800415039062,
+ 142.84800720214844,
+ 861.696044921875,
+ 144.89601135253906,
+ 850.4320678710938
+ ],
+ [
+ 84.4800033569336,
+ 850.4320678710938,
+ 82.4320068359375,
+ 861.696044921875,
+ 156.16000366210938,
+ 872.9600219726562,
+ 158.2080078125,
+ 859.6480712890625
+ ],
+ [
+ 79.36000061035156,
+ 859.6480712890625,
+ 76.28800201416016,
+ 870.9120483398438,
+ 139.7760009765625,
+ 883.2000122070312,
+ 141.82400512695312,
+ 870.9120483398438
+ ],
+ [
+ 73.21600341796875,
+ 870.9120483398438,
+ 70.14400482177734,
+ 883.2000122070312,
+ 136.70401000976562,
+ 894.4640502929688,
+ 138.75201416015625,
+ 881.1520385742188
+ ],
+ [
+ 67.07200622558594,
+ 882.176025390625,
+ 65.02400207519531,
+ 893.4400634765625,
+ 130.5600128173828,
+ 905.72802734375,
+ 132.60800170898438,
+ 893.4400634765625
+ ],
+ [
+ 18.944000244140625,
+ 905.72802734375,
+ 0.5120000243186951,
+ 1023.488037109375,
+ 19.968000411987305,
+ 1023.488037109375,
+ 40.448001861572266,
+ 909.8240356445312
+ ]
+ ]
+ ],
+ "labels": [
+ "container",
+ "fishing boat boat boat"
+ ]
+ }
+ },
+ "L617.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 952.83203125,
+ 602.6240234375,
+ 936.4480590820312,
+ 612.864013671875,
+ 969.216064453125,
+ 662.0160522460938,
+ 984.5760498046875,
+ 651.7760009765625
+ ],
+ [
+ 928.2560424804688,
+ 613.8880004882812,
+ 911.8720703125,
+ 624.1280517578125,
+ 940.5440673828125,
+ 665.0880126953125,
+ 955.904052734375,
+ 653.8240356445312
+ ],
+ [
+ 900.6080322265625,
+ 620.0320434570312,
+ 884.2240600585938,
+ 631.2960205078125,
+ 910.8480224609375,
+ 672.2560424804688,
+ 928.2560424804688,
+ 660.9920043945312
+ ],
+ [
+ 879.1040649414062,
+ 633.3440551757812,
+ 861.696044921875,
+ 644.6080322265625,
+ 886.2720336914062,
+ 677.3760375976562,
+ 902.6560668945312,
+ 665.0880126953125
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L620.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 617.9840087890625,
+ 932.35205078125,
+ 586.2400512695312,
+ 940.5440673828125,
+ 612.864013671875,
+ 1023.488037109375,
+ 644.6080322265625,
+ 1019.3920288085938
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L557.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 279.0400085449219,
+ 774.6560668945312,
+ 194.04800415039062,
+ 804.35205078125,
+ 201.21600341796875,
+ 824.83203125,
+ 286.2080078125,
+ 796.1600341796875
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L604.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 945.6640625,
+ 0.5120000243186951,
+ 748.0320434570312,
+ 51.71200180053711,
+ 760.3200073242188,
+ 85.5040054321289,
+ 957.9520263671875,
+ 15.872000694274902
+ ]
+ ]
+ ],
+ "labels": [
+ "ore oil"
+ ]
+ }
+ },
+ "L574.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 711.1680297851562,
+ 115.20000457763672,
+ 695.8080444335938,
+ 131.58399963378906,
+ 760.3200073242188,
+ 187.9040069580078,
+ 774.6560668945312,
+ 170.49600219726562
+ ]
+ ],
+ [
+ [
+ 933.3760375976562,
+ 503.2960205078125,
+ 864.76806640625,
+ 529.9200439453125,
+ 873.9840698242188,
+ 554.4960327148438,
+ 941.5680541992188,
+ 525.8240356445312
+ ],
+ [
+ 833.0240478515625,
+ 538.1119995117188,
+ 763.3920288085938,
+ 554.4960327148438,
+ 769.5360107421875,
+ 583.1680297851562,
+ 840.1920166015625,
+ 564.7360229492188
+ ]
+ ]
+ ],
+ "labels": [
+ "container",
+ "lawenforce"
+ ]
+ }
+ },
+ "L654.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 672.2560424804688,
+ 0.5120000243186951,
+ 620.0320434570312,
+ 35.3280029296875,
+ 634.3680419921875,
+ 51.71200180053711,
+ 686.592041015625,
+ 14.848000526428223
+ ],
+ [
+ 836.0960693359375,
+ 45.56800079345703,
+ 781.8240356445312,
+ 110.08000183105469,
+ 804.35205078125,
+ 128.5120086669922,
+ 857.6000366210938,
+ 62.97600173950195
+ ],
+ [
+ 549.3760375976562,
+ 101.88800811767578,
+ 510.4640197753906,
+ 179.71200561523438,
+ 544.2560424804688,
+ 197.12001037597656,
+ 582.14404296875,
+ 119.29600524902344
+ ],
+ [
+ 879.1040649414062,
+ 253.4400177001953,
+ 826.8800659179688,
+ 326.1440124511719,
+ 852.4800415039062,
+ 343.552001953125,
+ 903.6800537109375,
+ 269.8240051269531
+ ],
+ [
+ 497.15203857421875,
+ 283.1360168457031,
+ 439.8080139160156,
+ 357.8880310058594,
+ 467.4560241699219,
+ 378.3680114746094,
+ 523.7760009765625,
+ 303.61602783203125
+ ],
+ [
+ 528.89599609375,
+ 346.6240234375,
+ 475.64801025390625,
+ 421.3760070800781,
+ 502.27203369140625,
+ 438.7840270996094,
+ 554.4960327148438,
+ 365.0560302734375
+ ],
+ [
+ 378.3680114746094,
+ 421.3760070800781,
+ 329.21600341796875,
+ 486.9120178222656,
+ 353.7920227050781,
+ 506.3680114746094,
+ 403.968017578125,
+ 439.8080139160156
+ ],
+ [
+ 218.62400817871094,
+ 565.760009765625,
+ 166.40000915527344,
+ 645.6320190429688,
+ 192.00001525878906,
+ 662.0160522460938,
+ 244.22401428222656,
+ 583.1680297851562
+ ],
+ [
+ 300.54400634765625,
+ 674.3040161132812,
+ 244.22401428222656,
+ 758.2720336914062,
+ 270.8480224609375,
+ 775.6800537109375,
+ 328.1920166015625,
+ 691.7120361328125
+ ]
+ ],
+ [
+ [
+ 667.1360473632812,
+ 0.5120000243186951,
+ 619.008056640625,
+ 46.592002868652344,
+ 640.5120239257812,
+ 66.04800415039062,
+ 686.592041015625,
+ 10.75200080871582
+ ]
+ ]
+ ],
+ "labels": [
+ "container",
+ "dredgergerger"
+ ]
+ }
+ },
+ "L632.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 996.8640747070312,
+ 75.26399993896484,
+ 970.2400512695312,
+ 87.552001953125,
+ 1023.488037109375,
+ 201.21600341796875,
+ 1023.488037109375,
+ 187.9040069580078
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L564.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 251.3920135498047,
+ 554.4960327148438,
+ 251.3920135498047,
+ 664.0640258789062,
+ 284.1600036621094,
+ 664.0640258789062,
+ 284.1600036621094,
+ 554.4960327148438
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L555.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 34.30400085449219,
+ 299.52001953125,
+ 0.5120000243186951,
+ 326.1440124511719,
+ 144.89601135253906,
+ 559.6160278320312,
+ 185.8560028076172,
+ 534.0160522460938
+ ]
+ ]
+ ],
+ "labels": [
+ "ore oil"
+ ]
+ }
+ },
+ "L538.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 699.904052734375,
+ 120.32000732421875,
+ 679.4240112304688,
+ 205.31201171875,
+ 702.9760131835938,
+ 210.4320068359375,
+ 724.4800415039062,
+ 126.46400451660156
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L579.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 465.40802001953125,
+ 433.6640319824219,
+ 389.63201904296875,
+ 461.31201171875,
+ 398.8480224609375,
+ 398.8480224609375,
+ 473.60003662109375,
+ 457.2160339355469
+ ],
+ [
+ 469.5040283203125,
+ 456.1920166015625,
+ 394.75201416015625,
+ 482.8160095214844,
+ 401.9200134277344,
+ 506.3680114746094,
+ 477.6960144042969,
+ 477.6960144042969
+ ],
+ [
+ 336.3840026855469,
+ 631.2960205078125,
+ 318.97601318359375,
+ 640.5120239257812,
+ 338.4320068359375,
+ 695.8080444335938,
+ 354.8160095214844,
+ 687.6160278320312
+ ],
+ [
+ 97.79200744628906,
+ 743.93603515625,
+ 68.09600067138672,
+ 768.5120239257812,
+ 98.81600189208984,
+ 803.3280639648438,
+ 127.48800659179688,
+ 778.7520141601562
+ ],
+ [
+ 342.52801513671875,
+ 912.8960571289062,
+ 333.31201171875,
+ 932.35205078125,
+ 381.4400329589844,
+ 956.9280395507812,
+ 390.656005859375,
+ 938.4960327148438
+ ],
+ [
+ 258.55999755859375,
+ 919.0400390625,
+ 249.34400939941406,
+ 936.4480590820312,
+ 288.2560119628906,
+ 961.0240478515625,
+ 298.4960021972656,
+ 943.6160278320312
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L613.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 147.96800231933594,
+ 473.60003662109375,
+ 473.60003662109375,
+ 515.5840454101562,
+ 58.880001068115234,
+ 539.1360473632812,
+ 159.2320098876953,
+ 497.15203857421875
+ ],
+ [
+ 432.6400146484375,
+ 827.904052734375,
+ 298.4960021972656,
+ 882.176025390625,
+ 309.760009765625,
+ 908.800048828125,
+ 444.9280090332031,
+ 855.5520629882812
+ ]
+ ]
+ ],
+ "labels": [
+ "cell container"
+ ]
+ }
+ },
+ "L642.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 441.85601806640625,
+ 133.6320037841797,
+ 296.4480285644531,
+ 193.0240020751953,
+ 305.66400146484375,
+ 216.5760040283203,
+ 451.072021484375,
+ 158.2080078125
+ ]
+ ]
+ ],
+ "labels": [
+ "ore oil"
+ ]
+ }
+ },
+ "L585.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 146.94400024414062,
+ 583.1680297851562,
+ 74.24000549316406,
+ 626.176025390625,
+ 85.5040054321289,
+ 648.7040405273438,
+ 159.2320098876953,
+ 605.696044921875
+ ],
+ [
+ 388.6080322265625,
+ 587.2640380859375,
+ 331.2640075683594,
+ 630.2720336914062,
+ 347.64801025390625,
+ 650.7520141601562,
+ 404.99200439453125,
+ 605.696044921875
+ ],
+ [
+ 326.1440124511719,
+ 844.2880249023438,
+ 227.8400115966797,
+ 895.488037109375,
+ 240.1280059814453,
+ 923.1360473632812,
+ 339.4560241699219,
+ 870.9120483398438
+ ]
+ ]
+ ],
+ "labels": [
+ "dredger"
+ ]
+ }
+ },
+ "L631.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 825.8560180664062,
+ 426.49603271484375,
+ 819.7120361328125,
+ 441.85601806640625,
+ 898.56005859375,
+ 470.52801513671875,
+ 904.7040405273438,
+ 455.16802978515625
+ ],
+ [
+ 818.6880493164062,
+ 440.83203125,
+ 812.5440673828125,
+ 457.2160339355469,
+ 891.3920288085938,
+ 485.8880310058594,
+ 895.488037109375,
+ 469.5040283203125
+ ],
+ [
+ 935.424072265625,
+ 471.5520324707031,
+ 929.280029296875,
+ 487.93603515625,
+ 1006.080078125,
+ 516.6080322265625,
+ 1013.248046875,
+ 501.2480163574219
+ ],
+ [
+ 929.280029296875,
+ 482.8160095214844,
+ 923.1360473632812,
+ 499.20001220703125,
+ 1005.0560302734375,
+ 527.8720092773438,
+ 1012.2240600585938,
+ 511.488037109375
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L589.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 909.8240356445312,
+ 252.416015625,
+ 840.1920166015625,
+ 278.0160217285156,
+ 852.4800415039062,
+ 306.6880187988281,
+ 921.0880737304688,
+ 280.06402587890625
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L583.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 526.8480224609375,
+ 372.2240295410156,
+ 504.3200378417969,
+ 426.49603271484375,
+ 521.72802734375,
+ 432.6400146484375,
+ 543.2320556640625,
+ 378.3680114746094
+ ]
+ ],
+ [
+ [
+ 1016.320068359375,
+ 92.67200469970703,
+ 972.2880249023438,
+ 157.1840057373047,
+ 990.7200317382812,
+ 174.59201049804688,
+ 1023.488037109375,
+ 109.0560073852539
+ ],
+ [
+ 923.1360473632812,
+ 631.2960205078125,
+ 880.1280517578125,
+ 699.904052734375,
+ 896.5120239257812,
+ 712.1920166015625,
+ 939.52001953125,
+ 642.56005859375
+ ]
+ ]
+ ],
+ "labels": [
+ "cell container",
+ "container"
+ ]
+ }
+ },
+ "L662.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 80.38400268554688,
+ 248.32000732421875,
+ 0.5120000243186951,
+ 282.11199951171875,
+ 6.656000137329102,
+ 306.6880187988281,
+ 89.60000610351562,
+ 272.8960266113281
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L536.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 28.160001754760742,
+ 0.5120000243186951,
+ 27.136001586914062,
+ 15.872000694274902,
+ 95.74400329589844,
+ 22.016000747680664,
+ 97.79200744628906,
+ 3.5840001106262207
+ ]
+ ],
+ [
+ [
+ 809.4720458984375,
+ 146.94400024414062,
+ 694.7840576171875,
+ 184.83201599121094,
+ 702.9760131835938,
+ 213.50401306152344,
+ 817.6640625,
+ 175.6160125732422
+ ]
+ ]
+ ],
+ "labels": [
+ "container",
+ "dredger"
+ ]
+ }
+ },
+ "L556.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 94.72000122070312,
+ 62.97600173950195,
+ 9.72800064086914,
+ 85.5040054321289,
+ 15.872000694274902,
+ 112.12800598144531,
+ 100.86400604248047,
+ 88.57600402832031
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L599.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 981.5040283203125,
+ 372.2240295410156,
+ 981.5040283203125,
+ 390.656005859375,
+ 1023.488037109375,
+ 390.656005859375,
+ 1023.488037109375,
+ 372.2240295410156
+ ],
+ [
+ 974.3360595703125,
+ 384.51202392578125,
+ 920.0640258789062,
+ 388.6080322265625,
+ 921.0880737304688,
+ 401.9200134277344,
+ 975.3600463867188,
+ 397.8240051269531
+ ],
+ [
+ 813.5680541992188,
+ 412.1600341796875,
+ 770.56005859375,
+ 415.2320251464844,
+ 771.5840454101562,
+ 430.5920104980469,
+ 814.592041015625,
+ 427.52001953125
+ ],
+ [
+ 817.6640625,
+ 542.2080078125,
+ 816.6400146484375,
+ 557.5680541992188,
+ 873.9840698242188,
+ 561.6640014648438,
+ 875.008056640625,
+ 545.280029296875
+ ],
+ [
+ 936.4480590820312,
+ 542.2080078125,
+ 935.424072265625,
+ 560.6400146484375,
+ 1020.4160766601562,
+ 566.7839965820312,
+ 1021.4400634765625,
+ 548.35205078125
+ ],
+ [
+ 814.592041015625,
+ 550.4000244140625,
+ 813.5680541992188,
+ 567.8080444335938,
+ 873.9840698242188,
+ 571.904052734375,
+ 875.008056640625,
+ 554.4960327148438
+ ],
+ [
+ 935.424072265625,
+ 550.4000244140625,
+ 934.4000244140625,
+ 568.83203125,
+ 1016.320068359375,
+ 573.9520263671875,
+ 1018.3680419921875,
+ 554.4960327148438
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L605.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 593.4080200195312,
+ 262.656005859375,
+ 578.0480346679688,
+ 275.968017578125,
+ 617.9840087890625,
+ 323.072021484375,
+ 633.3440551757812,
+ 309.760009765625
+ ],
+ [
+ 290.30401611328125,
+ 457.2160339355469,
+ 274.9440002441406,
+ 466.4320373535156,
+ 310.7840270996094,
+ 514.5599975585938,
+ 326.1440124511719,
+ 505.3440246582031
+ ],
+ [
+ 312.8320007324219,
+ 457.2160339355469,
+ 298.4960021972656,
+ 466.4320373535156,
+ 331.2640075683594,
+ 518.656005859375,
+ 344.5760192871094,
+ 509.4400329589844
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L595.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 780.800048828125,
+ 215.55201721191406,
+ 772.6080322265625,
+ 241.15200805664062,
+ 817.6640625,
+ 257.5360107421875,
+ 825.8560180664062,
+ 232.9600067138672
+ ],
+ [
+ 164.3520050048828,
+ 237.05601501464844,
+ 143.87200927734375,
+ 268.8000183105469,
+ 161.28001403808594,
+ 281.0880126953125,
+ 180.7360076904297,
+ 247.29600524902344
+ ],
+ [
+ 140.8000030517578,
+ 290.30401611328125,
+ 127.48800659179688,
+ 305.66400146484375,
+ 140.8000030517578,
+ 315.9040222167969,
+ 154.1120147705078,
+ 298.4960021972656
+ ]
+ ],
+ [
+ [
+ 674.3040161132812,
+ 762.3680419921875,
+ 607.7440185546875,
+ 357.8880310058594,
+ 635.3920288085938,
+ 887.2960205078125,
+ 701.9520263671875,
+ 778.7520141601562
+ ]
+ ]
+ ],
+ "labels": [
+ "container",
+ "dredger"
+ ]
+ }
+ },
+ "L630.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 247.29600524902344,
+ 860.6720581054688,
+ 205.31201171875,
+ 928.2560424804688,
+ 223.74400329589844,
+ 941.5680541992188,
+ 266.75201416015625,
+ 873.9840698242188
+ ],
+ [
+ 585.2160034179688,
+ 937.4720458984375,
+ 543.2320556640625,
+ 1023.488037109375,
+ 566.7839965820312,
+ 1023.488037109375,
+ 609.7920532226562,
+ 948.7360229492188
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L660.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 310.7840270996094,
+ 293.3760070800781,
+ 217.60000610351562,
+ 303.61602783203125,
+ 220.67201232910156,
+ 329.21600341796875,
+ 313.85601806640625,
+ 317.9520263671875
+ ]
+ ],
+ [
+ [
+ 310.7840270996094,
+ 293.3760070800781,
+ 222.7200164794922,
+ 303.61602783203125,
+ 225.79200744628906,
+ 327.16802978515625,
+ 313.85601806640625,
+ 317.9520263671875
+ ]
+ ]
+ ],
+ "labels": [
+ "cell container",
+ "container"
+ ]
+ }
+ },
+ "L581.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 418.30401611328125,
+ 454.1440124511719,
+ 228.864013671875,
+ 519.6800537109375,
+ 241.15200805664062,
+ 559.6160278320312,
+ 429.5680236816406,
+ 492.0320129394531
+ ]
+ ]
+ ],
+ "labels": [
+ "cell container"
+ ]
+ }
+ },
+ "L568.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 713.2160034179688,
+ 35.3280029296875,
+ 695.8080444335938,
+ 51.71200180053711,
+ 773.6320190429688,
+ 118.27200317382812,
+ 788.9920654296875,
+ 99.84000396728516
+ ],
+ [
+ 232.9600067138672,
+ 587.2640380859375,
+ 215.55201721191406,
+ 602.6240234375,
+ 256.51202392578125,
+ 643.5840454101562,
+ 272.8960266113281,
+ 628.2240600585938
+ ],
+ [
+ 280.06402587890625,
+ 624.1280517578125,
+ 263.6800231933594,
+ 638.4640502929688,
+ 300.54400634765625,
+ 682.4960327148438,
+ 316.9280090332031,
+ 668.1600341796875
+ ],
+ [
+ 357.8880310058594,
+ 676.35205078125,
+ 341.5040283203125,
+ 692.7360229492188,
+ 401.9200134277344,
+ 745.9840087890625,
+ 418.30401611328125,
+ 728.5760498046875
+ ],
+ [
+ 341.5040283203125,
+ 692.7360229492188,
+ 323.072021484375,
+ 709.1200561523438,
+ 381.4400329589844,
+ 760.3200073242188,
+ 397.8240051269531,
+ 742.9120483398438
+ ],
+ [
+ 735.7440185546875,
+ 758.2720336914062,
+ 718.3360595703125,
+ 774.6560668945312,
+ 765.4400634765625,
+ 815.6160278320312,
+ 780.800048828125,
+ 799.2320556640625
+ ],
+ [
+ 502.27203369140625,
+ 787.968017578125,
+ 479.7440185546875,
+ 811.52001953125,
+ 541.1840209960938,
+ 866.8160400390625,
+ 563.7120361328125,
+ 842.2400512695312
+ ],
+ [
+ 821.760009765625,
+ 813.5680541992188,
+ 803.3280639648438,
+ 834.0480346679688,
+ 852.4800415039062,
+ 877.0560302734375,
+ 870.9120483398438,
+ 854.5280151367188
+ ],
+ [
+ 594.4320068359375,
+ 858.6240234375,
+ 573.9520263671875,
+ 879.1040649414062,
+ 623.10400390625,
+ 928.2560424804688,
+ 642.56005859375,
+ 906.7520141601562
+ ],
+ [
+ 877.0560302734375,
+ 855.5520629882812,
+ 860.6720581054688,
+ 873.9840698242188,
+ 908.800048828125,
+ 914.9440307617188,
+ 925.1840209960938,
+ 896.5120239257812
+ ],
+ [
+ 942.592041015625,
+ 901.6320190429688,
+ 927.2320556640625,
+ 920.0640258789062,
+ 975.3600463867188,
+ 961.0240478515625,
+ 990.7200317382812,
+ 942.592041015625
+ ],
+ [
+ 679.4240112304688,
+ 914.9440307617188,
+ 653.8240356445312,
+ 940.5440673828125,
+ 738.8160400390625,
+ 1019.3920288085938,
+ 763.3920288085938,
+ 993.7920532226562
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L609.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 410.1120300292969,
+ 102.91200256347656,
+ 350.72003173828125,
+ 131.58399963378906,
+ 361.9840087890625,
+ 155.13600158691406,
+ 421.3760070800781,
+ 126.46400451660156
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L546.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 370.176025390625,
+ 602.6240234375,
+ 353.7920227050781,
+ 607.7440185546875,
+ 368.1280212402344,
+ 640.5120239257812,
+ 383.4880065917969,
+ 634.3680419921875
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L661.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 333.31201171875,
+ 243.20001220703125,
+ 256.51202392578125,
+ 267.7760009765625,
+ 264.7040100097656,
+ 291.3280029296875,
+ 341.5040283203125,
+ 264.7040100097656
+ ],
+ [
+ 219.64801025390625,
+ 544.2560424804688,
+ 117.24800872802734,
+ 608.7680053710938,
+ 127.48800659179688,
+ 609.7920532226562,
+ 227.8400115966797,
+ 544.2560424804688
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L611.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 853.5040283203125,
+ 199.1680145263672,
+ 840.1920166015625,
+ 226.81600952148438,
+ 911.8720703125,
+ 261.63201904296875,
+ 926.2080688476562,
+ 233.9840087890625
+ ],
+ [
+ 836.0960693359375,
+ 233.9840087890625,
+ 823.8080444335938,
+ 257.5360107421875,
+ 897.5360717773438,
+ 296.4480285644531,
+ 909.8240356445312,
+ 272.8960266113281
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L655.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 410.1120300292969,
+ 253.4400177001953,
+ 385.5360107421875,
+ 276.99200439453125,
+ 455.16802978515625,
+ 346.6240234375,
+ 478.72003173828125,
+ 323.072021484375
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L549.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 171.52000427246094,
+ 818.6880493164062,
+ 115.20000457763672,
+ 838.14404296875,
+ 121.34400939941406,
+ 852.4800415039062,
+ 176.6400146484375,
+ 834.0480346679688
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L532.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 84.4800033569336,
+ 85.5040054321289,
+ 84.4800033569336,
+ 97.79200744628906,
+ 116.22400665283203,
+ 100.86400604248047,
+ 117.24800872802734,
+ 89.60000610351562
+ ],
+ [
+ 700.9280395507812,
+ 820.7360229492188,
+ 621.0560302734375,
+ 842.2400512695312,
+ 628.2240600585938,
+ 873.9840698242188,
+ 709.1200561523438,
+ 853.5040283203125
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L645.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 8.704000473022461,
+ 474.6240234375,
+ 0.5120000243186951,
+ 486.9120178222656,
+ 47.61600112915039,
+ 559.6160278320312,
+ 61.952003479003906,
+ 549.3760375976562
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L596.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 150.01600646972656,
+ 255.48800659179688,
+ 95.74400329589844,
+ 282.11199951171875,
+ 110.08000183105469,
+ 301.5680236816406,
+ 162.3040008544922,
+ 274.9440002441406
+ ],
+ [
+ 173.56800842285156,
+ 287.2320251464844,
+ 117.24800872802734,
+ 312.8320007324219,
+ 129.5360107421875,
+ 332.28802490234375,
+ 185.8560028076172,
+ 305.66400146484375
+ ],
+ [
+ 201.21600341796875,
+ 333.31201171875,
+ 142.84800720214844,
+ 356.864013671875,
+ 154.1120147705078,
+ 378.3680114746094,
+ 212.48001098632812,
+ 351.7440185546875
+ ],
+ [
+ 218.62400817871094,
+ 364.0320129394531,
+ 160.25601196289062,
+ 389.63201904296875,
+ 171.52000427246094,
+ 410.1120300292969,
+ 230.91201782226562,
+ 383.4880065917969
+ ],
+ [
+ 235.0080108642578,
+ 387.5840148925781,
+ 178.68800354003906,
+ 418.30401611328125,
+ 190.97601318359375,
+ 438.7840270996094,
+ 247.29600524902344,
+ 410.1120300292969
+ ],
+ [
+ 279.0400085449219,
+ 397.8240051269531,
+ 223.74400329589844,
+ 426.49603271484375,
+ 235.0080108642578,
+ 446.97601318359375,
+ 291.3280029296875,
+ 418.30401611328125
+ ],
+ [
+ 293.3760070800781,
+ 426.49603271484375,
+ 242.17601013183594,
+ 450.04803466796875,
+ 253.4400177001953,
+ 471.5520324707031,
+ 303.61602783203125,
+ 446.97601318359375
+ ],
+ [
+ 316.9280090332031,
+ 448.0000305175781,
+ 261.63201904296875,
+ 475.64801025390625,
+ 271.87200927734375,
+ 494.08001708984375,
+ 326.1440124511719,
+ 467.4560241699219
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L548.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 453.1200256347656,
+ 10.75200080871582,
+ 413.18402099609375,
+ 99.84000396728516,
+ 435.7120056152344,
+ 110.08000183105469,
+ 474.6240234375,
+ 22.016000747680664
+ ],
+ [
+ 860.6720581054688,
+ 163.3280029296875,
+ 817.6640625,
+ 238.08001708984375,
+ 840.1920166015625,
+ 249.34400939941406,
+ 883.2000122070312,
+ 174.59201049804688
+ ],
+ [
+ 507.39202880859375,
+ 281.0880126953125,
+ 471.5520324707031,
+ 356.864013671875,
+ 496.1280212402344,
+ 366.08001708984375,
+ 530.9440307617188,
+ 290.30401611328125
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L644.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 616.9600219726562,
+ 530.9440307617188,
+ 584.1920166015625,
+ 565.760009765625,
+ 597.5040283203125,
+ 583.1680297851562,
+ 633.3440551757812,
+ 548.35205078125
+ ]
+ ],
+ [
+ [
+ 382.4640197753906,
+ 576.0,
+ 335.3600158691406,
+ 605.696044921875,
+ 463.3600158691406,
+ 799.2320556640625,
+ 509.4400329589844,
+ 768.5120239257812
+ ]
+ ]
+ ],
+ "labels": [
+ "container",
+ "ore oil"
+ ]
+ }
+ },
+ "L550.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 127.48800659179688,
+ 81.40800476074219,
+ 75.26399993896484,
+ 117.24800872802734,
+ 86.52800750732422,
+ 133.6320037841797,
+ 139.7760009765625,
+ 97.79200744628906
+ ],
+ [
+ 354.8160095214844,
+ 216.5760040283203,
+ 296.4480285644531,
+ 255.48800659179688,
+ 306.6880187988281,
+ 278.0160217285156,
+ 366.08001708984375,
+ 238.08001708984375
+ ],
+ [
+ 354.8160095214844,
+ 919.0400390625,
+ 346.6240234375,
+ 943.6160278320312,
+ 395.7760314941406,
+ 960.0000610351562,
+ 403.968017578125,
+ 509.4400329589844
+ ]
+ ],
+ [
+ [
+ 592.384033203125,
+ 397.8240051269531,
+ 516.6080322265625,
+ 424.4480285644531,
+ 524.800048828125,
+ 449.0240173339844,
+ 599.552001953125,
+ 422.4000244140625
+ ],
+ [
+ 1010.176025390625,
+ 429.5680236816406,
+ 947.7120361328125,
+ 467.4560241699219,
+ 958.97607421875,
+ 487.93603515625,
+ 1023.488037109375,
+ 450.04803466796875
+ ]
+ ]
+ ],
+ "labels": [
+ "container",
+ "dredger"
+ ]
+ }
+ },
+ "L608.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 374.2720031738281,
+ 530.9440307617188,
+ 364.0320129394531,
+ 569.8560180664062,
+ 402.94403076171875,
+ 587.2640380859375,
+ 413.18402099609375,
+ 549.3760375976562
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L616.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 711.1680297851562,
+ 763.3920288085938,
+ 648.7040405273438,
+ 807.4240112304688,
+ 659.968017578125,
+ 821.760009765625,
+ 723.4560546875,
+ 776.7040405273438
+ ],
+ [
+ 725.5040283203125,
+ 778.7520141601562,
+ 665.0880126953125,
+ 821.760009765625,
+ 676.35205078125,
+ 836.0960693359375,
+ 735.7440185546875,
+ 794.112060546875
+ ],
+ [
+ 742.9120483398438,
+ 795.1360473632812,
+ 682.4960327148438,
+ 837.1200561523438,
+ 693.760009765625,
+ 851.4560546875,
+ 754.176025390625,
+ 809.4720458984375
+ ],
+ [
+ 756.2240600585938,
+ 814.592041015625,
+ 695.8080444335938,
+ 854.5280151367188,
+ 707.072021484375,
+ 868.864013671875,
+ 767.488037109375,
+ 829.9520263671875
+ ],
+ [
+ 813.5680541992188,
+ 865.7920532226562,
+ 755.2000122070312,
+ 908.800048828125,
+ 766.4640502929688,
+ 925.1840209960938,
+ 825.8560180664062,
+ 882.176025390625
+ ],
+ [
+ 835.072021484375,
+ 880.1280517578125,
+ 774.6560668945312,
+ 922.112060546875,
+ 785.9200439453125,
+ 936.4480590820312,
+ 846.3360595703125,
+ 894.4640502929688
+ ],
+ [
+ 857.6000366210938,
+ 904.7040405273438,
+ 793.0880126953125,
+ 938.4960327148438,
+ 801.280029296875,
+ 954.8800659179688,
+ 865.7920532226562,
+ 922.112060546875
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L621.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 483.84002685546875,
+ 101.88800811767578,
+ 449.0240173339844,
+ 168.44801330566406,
+ 475.64801025390625,
+ 182.7840118408203,
+ 511.488037109375,
+ 117.24800872802734
+ ],
+ [
+ 751.1040649414062,
+ 366.08001708984375,
+ 716.2880249023438,
+ 431.61602783203125,
+ 742.9120483398438,
+ 446.97601318359375,
+ 777.72802734375,
+ 381.4400329589844
+ ],
+ [
+ 482.8160095214844,
+ 734.7200317382812,
+ 476.6720275878906,
+ 767.488037109375,
+ 597.5040283203125,
+ 794.112060546875,
+ 602.6240234375,
+ 761.3440551757812
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L597.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 573.9520263671875,
+ 318.97601318359375,
+ 509.4400329589844,
+ 353.7920227050781,
+ 517.6320190429688,
+ 374.2720031738281,
+ 583.1680297851562,
+ 338.4320068359375
+ ],
+ [
+ 633.3440551757812,
+ 336.3840026855469,
+ 523.7760009765625,
+ 357.8880310058594,
+ 528.89599609375,
+ 377.3440246582031,
+ 638.4640502929688,
+ 357.8880310058594
+ ],
+ [
+ 215.55201721191406,
+ 470.52801513671875,
+ 25.088001251220703,
+ 560.6400146484375,
+ 40.448001861572266,
+ 588.2880249023438,
+ 229.8880157470703,
+ 499.20001220703125
+ ],
+ [
+ 743.93603515625,
+ 668.1600341796875,
+ 668.1600341796875,
+ 726.5280151367188,
+ 683.52001953125,
+ 747.008056640625,
+ 760.3200073242188,
+ 689.6640625
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L638.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 653.8240356445312,
+ 260.6080017089844,
+ 632.3200073242188,
+ 275.968017578125,
+ 672.2560424804688,
+ 333.31201171875,
+ 693.760009765625,
+ 317.9520263671875
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L653.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 40.448001861572266,
+ 105.9840087890625,
+ 20.992000579833984,
+ 131.58399963378906,
+ 108.0320053100586,
+ 194.04800415039062,
+ 127.48800659179688,
+ 169.47201538085938
+ ]
+ ]
+ ],
+ "labels": [
+ "dredger"
+ ]
+ }
+ },
+ "L650.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 634.3680419921875,
+ 153.0880126953125,
+ 614.9120483398438,
+ 262.656005859375,
+ 651.7760009765625,
+ 269.8240051269531,
+ 672.2560424804688,
+ 160.25601196289062
+ ],
+ [
+ 729.6000366210938,
+ 167.42401123046875,
+ 713.2160034179688,
+ 264.7040100097656,
+ 741.8880615234375,
+ 269.8240051269531,
+ 759.2960205078125,
+ 172.54400634765625
+ ],
+ [
+ 235.0080108642578,
+ 409.0880126953125,
+ 186.8800048828125,
+ 446.97601318359375,
+ 303.61602783203125,
+ 541.1840209960938,
+ 349.6960144042969,
+ 502.27203369140625
+ ],
+ [
+ 651.7760009765625,
+ 637.4400024414062,
+ 638.4640502929688,
+ 699.904052734375,
+ 704.0000610351562,
+ 710.14404296875,
+ 718.3360595703125,
+ 647.6800537109375
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L570.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 335.3600158691406,
+ 75.26399993896484,
+ 233.9840087890625,
+ 127.48800659179688,
+ 247.29600524902344,
+ 150.01600646972656,
+ 346.6240234375,
+ 97.79200744628906
+ ],
+ [
+ 10.75200080871582,
+ 538.1119995117188,
+ 4.6080002784729,
+ 574.9760131835938,
+ 135.6800079345703,
+ 596.4800415039062,
+ 140.8000030517578,
+ 558.592041015625
+ ]
+ ]
+ ],
+ "labels": [
+ "dredger"
+ ]
+ }
+ },
+ "L591.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 339.4560241699219,
+ 425.4720153808594,
+ 328.1920166015625,
+ 437.760009765625,
+ 370.176025390625,
+ 474.6240234375,
+ 381.4400329589844,
+ 462.3360290527344
+ ],
+ [
+ 387.5840148925781,
+ 461.31201171875,
+ 377.3440246582031,
+ 472.5760192871094,
+ 418.30401611328125,
+ 511.488037109375,
+ 429.5680236816406,
+ 498.176025390625
+ ],
+ [
+ 463.3600158691406,
+ 517.6320190429688,
+ 448.0000305175781,
+ 535.0400390625,
+ 527.8720092773438,
+ 590.3359985351562,
+ 542.2080078125,
+ 572.9280395507812
+ ],
+ [
+ 455.16802978515625,
+ 527.8720092773438,
+ 438.7840270996094,
+ 545.280029296875,
+ 499.20001220703125,
+ 591.3600463867188,
+ 512.5120239257812,
+ 573.9520263671875
+ ],
+ [
+ 803.3280639648438,
+ 552.447998046875,
+ 795.1360473632812,
+ 561.6640014648438,
+ 836.0960693359375,
+ 597.5040283203125,
+ 844.2880249023438,
+ 587.2640380859375
+ ],
+ [
+ 796.1600341796875,
+ 558.592041015625,
+ 785.9200439453125,
+ 568.83203125,
+ 825.8560180664062,
+ 603.6480102539062,
+ 835.072021484375,
+ 594.4320068359375
+ ],
+ [
+ 845.31201171875,
+ 601.6000366210938,
+ 837.1200561523438,
+ 609.7920532226562,
+ 877.0560302734375,
+ 641.5360107421875,
+ 884.2240600585938,
+ 634.3680419921875
+ ],
+ [
+ 837.1200561523438,
+ 610.8160400390625,
+ 828.9280395507812,
+ 619.008056640625,
+ 866.8160400390625,
+ 650.7520141601562,
+ 875.008056640625,
+ 642.56005859375
+ ],
+ [
+ 623.10400390625,
+ 635.3920288085938,
+ 612.864013671875,
+ 645.6320190429688,
+ 651.7760009765625,
+ 682.4960327148438,
+ 662.0160522460938,
+ 672.2560424804688
+ ],
+ [
+ 611.8400268554688,
+ 643.5840454101562,
+ 600.5760498046875,
+ 655.8720092773438,
+ 642.56005859375,
+ 692.7360229492188,
+ 653.8240356445312,
+ 681.4720458984375
+ ],
+ [
+ 939.52001953125,
+ 655.8720092773438,
+ 932.35205078125,
+ 665.0880126953125,
+ 969.216064453125,
+ 697.8560180664062,
+ 978.4320678710938,
+ 688.6400146484375
+ ],
+ [
+ 929.280029296875,
+ 664.0640258789062,
+ 920.0640258789062,
+ 674.3040161132812,
+ 958.97607421875,
+ 710.14404296875,
+ 968.1920166015625,
+ 699.904052734375
+ ],
+ [
+ 990.7200317382812,
+ 696.83203125,
+ 982.528076171875,
+ 707.072021484375,
+ 1021.4400634765625,
+ 742.9120483398438,
+ 1023.488037109375,
+ 732.6720581054688
+ ],
+ [
+ 984.5760498046875,
+ 709.1200561523438,
+ 977.4080200195312,
+ 719.3600463867188,
+ 1014.2720336914062,
+ 754.176025390625,
+ 1022.4640502929688,
+ 743.93603515625
+ ],
+ [
+ 823.8080444335938,
+ 781.8240356445312,
+ 812.5440673828125,
+ 794.112060546875,
+ 864.76806640625,
+ 840.1920166015625,
+ 876.0320434570312,
+ 827.904052734375
+ ],
+ [
+ 808.4480590820312,
+ 796.1600341796875,
+ 797.1840209960938,
+ 808.4480590820312,
+ 849.4080200195312,
+ 853.5040283203125,
+ 859.6480712890625,
+ 842.2400512695312
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L664.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 226.81600952148438,
+ 353.7920227050781,
+ 199.1680145263672,
+ 366.08001708984375,
+ 242.17601013183594,
+ 464.384033203125,
+ 268.8000183105469,
+ 451.072021484375
+ ],
+ [
+ 105.9840087890625,
+ 608.7680053710938,
+ 74.24000549316406,
+ 745.9840087890625,
+ 118.27200317382812,
+ 755.2000122070312,
+ 150.01600646972656,
+ 617.9840087890625
+ ]
+ ],
+ [
+ [
+ 551.4240112304688,
+ 702.9760131835938,
+ 427.52001953125,
+ 737.7920532226562,
+ 434.6880187988281,
+ 764.416015625,
+ 558.592041015625,
+ 727.5520629882812
+ ],
+ [
+ 90.6240005493164,
+ 929.280029296875,
+ 51.71200180053711,
+ 1005.0560302734375,
+ 69.12000274658203,
+ 1013.248046875,
+ 108.0320053100586,
+ 938.4960327148438
+ ]
+ ]
+ ],
+ "labels": [
+ "cell container",
+ "container"
+ ]
+ }
+ },
+ "L593.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 84.4800033569336,
+ 413.18402099609375,
+ 55.80800247192383,
+ 532.9920043945312,
+ 80.38400268554688,
+ 539.1360473632812,
+ 108.0320053100586,
+ 419.3280334472656
+ ],
+ [
+ 377.3440246582031,
+ 543.2320556640625,
+ 337.40802001953125,
+ 691.7120361328125,
+ 364.0320129394531,
+ 699.904052734375,
+ 402.94403076171875,
+ 552.447998046875
+ ],
+ [
+ 625.1520385742188,
+ 767.488037109375,
+ 540.1600341796875,
+ 859.6480712890625,
+ 566.7839965820312,
+ 881.1520385742188,
+ 653.8240356445312,
+ 788.9920654296875
+ ],
+ [
+ 197.12001037597656,
+ 856.5760498046875,
+ 162.3040008544922,
+ 975.3600463867188,
+ 187.9040069580078,
+ 982.528076171875,
+ 221.69601440429688,
+ 863.7440185546875
+ ]
+ ],
+ [
+ [
+ 899.5840454101562,
+ 456.1920166015625,
+ 843.2640380859375,
+ 528.89599609375,
+ 864.76806640625,
+ 546.3040161132812,
+ 922.112060546875,
+ 471.5520324707031
+ ]
+ ]
+ ],
+ "labels": [
+ "cell container",
+ "dredger"
+ ]
+ }
+ },
+ "L657.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 706.0480346679688,
+ 430.5920104980469,
+ 699.904052734375,
+ 446.97601318359375,
+ 770.56005859375,
+ 470.52801513671875,
+ 776.7040405273438,
+ 454.1440124511719
+ ],
+ [
+ 684.5440063476562,
+ 526.8480224609375,
+ 678.4000244140625,
+ 582.14404296875,
+ 690.6880493164062,
+ 584.1920166015625,
+ 698.8800048828125,
+ 528.89599609375
+ ],
+ [
+ 694.7840576171875,
+ 532.9920043945312,
+ 688.6400146484375,
+ 587.2640380859375,
+ 701.9520263671875,
+ 589.31201171875,
+ 709.1200561523438,
+ 535.0400390625
+ ]
+ ],
+ [
+ [
+ 958.97607421875,
+ 842.2400512695312,
+ 800.2560424804688,
+ 1023.488037109375,
+ 835.072021484375,
+ 1023.488037109375,
+ 993.7920532226562,
+ 868.864013671875
+ ]
+ ]
+ ],
+ "labels": [
+ "container",
+ "ore oil"
+ ]
+ }
+ },
+ "L537.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 901.6320190429688,
+ 257.5360107421875,
+ 900.6080322265625,
+ 281.0880126953125,
+ 945.6640625,
+ 283.1360168457031,
+ 946.6880493164062,
+ 260.6080017089844
+ ],
+ [
+ 901.6320190429688,
+ 397.8240051269531,
+ 900.6080322265625,
+ 420.3520202636719,
+ 945.6640625,
+ 422.4000244140625,
+ 946.6880493164062,
+ 398.8480224609375
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L592.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 626.176025390625,
+ 375.2960205078125,
+ 615.93603515625,
+ 432.6400146484375,
+ 734.7200317382812,
+ 454.1440124511719,
+ 744.9600219726562,
+ 395.7760314941406
+ ],
+ [
+ 835.072021484375,
+ 592.384033203125,
+ 712.1920166015625,
+ 614.9120483398438,
+ 718.3360595703125,
+ 654.8480224609375,
+ 842.2400512695312,
+ 632.3200073242188
+ ],
+ [
+ 597.5040283203125,
+ 639.488037109375,
+ 526.8480224609375,
+ 660.9920043945312,
+ 536.0640258789062,
+ 701.9520263671875,
+ 606.7200317382812,
+ 680.4480590820312
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L625.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 871.93603515625,
+ 102.91200256347656,
+ 763.3920288085938,
+ 127.48800659179688,
+ 772.6080322265625,
+ 173.56800842285156,
+ 882.176025390625,
+ 148.99200439453125
+ ]
+ ]
+ ],
+ "labels": [
+ "dredger"
+ ]
+ }
+ },
+ "L647.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 435.7120056152344,
+ 94.72000122070312,
+ 431.61602783203125,
+ 123.39200592041016,
+ 597.5040283203125,
+ 142.84800720214844,
+ 599.552001953125,
+ 114.1760025024414
+ ],
+ [
+ 83.45600128173828,
+ 111.10400390625,
+ 79.36000061035156,
+ 139.7760009765625,
+ 244.22401428222656,
+ 161.28001403808594,
+ 247.29600524902344,
+ 131.58399963378906
+ ]
+ ]
+ ],
+ "labels": [
+ "cell container"
+ ]
+ }
+ },
+ "L558.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 466.4320373535156,
+ 85.5040054321289,
+ 426.49603271484375,
+ 94.72000122070312,
+ 429.5680236816406,
+ 118.27200317382812,
+ 469.5040283203125,
+ 109.0560073852539
+ ],
+ [
+ 606.7200317382812,
+ 88.57600402832031,
+ 526.8480224609375,
+ 92.67200469970703,
+ 527.8720092773438,
+ 112.12800598144531,
+ 607.7440185546875,
+ 108.0320053100586
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L646.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 53.76000213623047,
+ 377.3440246582031,
+ 0.5120000243186951,
+ 551.4240112304688,
+ 12.800000190734863,
+ 560.6400146484375,
+ 80.38400268554688,
+ 386.5600280761719
+ ]
+ ]
+ ],
+ "labels": [
+ "cell container"
+ ]
+ }
+ },
+ "L544.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 657.9200439453125,
+ 657.9200439453125,
+ 657.9200439453125,
+ 691.7120361328125,
+ 733.696044921875,
+ 691.7120361328125,
+ 733.696044921875,
+ 657.9200439453125
+ ],
+ [
+ 374.2720031738281,
+ 675.3280029296875,
+ 372.2240295410156,
+ 692.7360229492188,
+ 439.8080139160156,
+ 698.8800048828125,
+ 441.85601806640625,
+ 681.4720458984375
+ ]
+ ],
+ [
+ [
+ 581.1200561523438,
+ 541.1840209960938,
+ 463.3600158691406,
+ 549.3760375976562,
+ 464.384033203125,
+ 570.8800048828125,
+ 582.14404296875,
+ 562.6880493164062
+ ]
+ ],
+ [
+ [
+ 945.6640625,
+ 519.6800537109375,
+ 775.6800537109375,
+ 534.0160522460938,
+ 777.72802734375,
+ 559.6160278320312,
+ 947.7120361328125,
+ 546.3040161132812
+ ]
+ ]
+ ],
+ "labels": [
+ "container",
+ "dredger",
+ "ore oil"
+ ]
+ }
+ },
+ "L652.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 654.8480224609375,
+ 5.63200044631958,
+ 646.656005859375,
+ 14.848000526428223,
+ 677.3760375976562,
+ 40.448001861572266,
+ 684.5440063476562,
+ 30.2080020904541
+ ],
+ [
+ 647.6800537109375,
+ 16.895999908447266,
+ 640.5120239257812,
+ 25.088001251220703,
+ 669.1840209960938,
+ 50.68800354003906,
+ 677.3760375976562,
+ 42.496002197265625
+ ],
+ [
+ 639.488037109375,
+ 26.112001419067383,
+ 631.2960205078125,
+ 35.3280029296875,
+ 658.9440307617188,
+ 58.880001068115234,
+ 668.1600341796875,
+ 49.66400146484375
+ ],
+ [
+ 630.2720336914062,
+ 35.3280029296875,
+ 622.0800170898438,
+ 43.52000045776367,
+ 650.7520141601562,
+ 68.09600067138672,
+ 658.9440307617188,
+ 58.880001068115234
+ ],
+ [
+ 678.4000244140625,
+ 42.496002197265625,
+ 671.2320556640625,
+ 50.68800354003906,
+ 698.8800048828125,
+ 73.21600341796875,
+ 707.072021484375,
+ 65.02400207519531
+ ],
+ [
+ 670.2080078125,
+ 48.6400032043457,
+ 663.0400390625,
+ 58.880001068115234,
+ 692.7360229492188,
+ 83.45600128173828,
+ 700.9280395507812,
+ 73.21600341796875
+ ],
+ [
+ 726.5280151367188,
+ 58.880001068115234,
+ 720.384033203125,
+ 68.09600067138672,
+ 748.0320434570312,
+ 91.64800262451172,
+ 755.2000122070312,
+ 82.4320068359375
+ ],
+ [
+ 718.3360595703125,
+ 71.16800689697266,
+ 710.14404296875,
+ 80.38400268554688,
+ 738.8160400390625,
+ 102.91200256347656,
+ 745.9840087890625,
+ 93.69600677490234
+ ],
+ [
+ 710.14404296875,
+ 80.38400268554688,
+ 701.9520263671875,
+ 88.57600402832031,
+ 728.5760498046875,
+ 113.15200805664062,
+ 736.7680053710938,
+ 103.93600463867188
+ ],
+ [
+ 756.2240600585938,
+ 90.6240005493164,
+ 748.0320434570312,
+ 99.84000396728516,
+ 774.6560668945312,
+ 125.44000244140625,
+ 782.8480224609375,
+ 116.22400665283203
+ ],
+ [
+ 744.9600219726562,
+ 99.84000396728516,
+ 736.7680053710938,
+ 109.0560073852539,
+ 766.4640502929688,
+ 134.656005859375,
+ 774.6560668945312,
+ 126.46400451660156
+ ],
+ [
+ 807.4240112304688,
+ 115.20000457763672,
+ 799.2320556640625,
+ 125.44000244140625,
+ 826.8800659179688,
+ 148.99200439453125,
+ 835.072021484375,
+ 139.7760009765625
+ ],
+ [
+ 797.1840209960938,
+ 126.46400451660156,
+ 788.9920654296875,
+ 135.6800079345703,
+ 817.6640625,
+ 161.28001403808594,
+ 825.8560180664062,
+ 151.04000854492188
+ ],
+ [
+ 784.8960571289062,
+ 135.6800079345703,
+ 776.7040405273438,
+ 144.89601135253906,
+ 806.4000244140625,
+ 170.49600219726562,
+ 814.592041015625,
+ 161.28001403808594
+ ],
+ [
+ 773.6320190429688,
+ 144.89601135253906,
+ 765.4400634765625,
+ 154.1120147705078,
+ 795.1360473632812,
+ 179.71200561523438,
+ 803.3280639648438,
+ 170.49600219726562
+ ],
+ [
+ 765.4400634765625,
+ 151.04000854492188,
+ 756.2240600585938,
+ 160.25601196289062,
+ 786.9440307617188,
+ 185.8560028076172,
+ 795.1360473632812,
+ 175.6160125732422
+ ],
+ [
+ 835.072021484375,
+ 137.72801208496094,
+ 826.8800659179688,
+ 146.94400024414062,
+ 854.5280151367188,
+ 170.49600219726562,
+ 861.696044921875,
+ 161.28001403808594
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L588.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 513.5360107421875,
+ 407.0400085449219,
+ 477.6960144042969,
+ 422.4000244140625,
+ 535.0400390625,
+ 566.7839965820312,
+ 571.904052734375,
+ 551.4240112304688
+ ]
+ ]
+ ],
+ "labels": [
+ "cell container"
+ ]
+ }
+ },
+ "L540.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 333.31201171875,
+ 0.5120000243186951,
+ 325.1200256347656,
+ 24.064001083374023,
+ 371.20001220703125,
+ 36.35200119018555,
+ 378.3680114746094,
+ 0.5120000243186951
+ ],
+ [
+ 681.4720458984375,
+ 0.5120000243186951,
+ 612.864013671875,
+ 116.22400665283203,
+ 656.8960571289062,
+ 138.75201416015625,
+ 725.5040283203125,
+ 15.872000694274902
+ ],
+ [
+ 445.9520263671875,
+ 8.704000473022461,
+ 438.7840270996094,
+ 25.088001251220703,
+ 479.7440185546875,
+ 38.400001525878906,
+ 485.8880310058594,
+ 23.040000915527344
+ ],
+ [
+ 649.72802734375,
+ 17.920000076293945,
+ 643.5840454101562,
+ 33.28000259399414,
+ 688.6400146484375,
+ 47.61600112915039,
+ 693.760009765625,
+ 33.28000259399414
+ ],
+ [
+ 639.488037109375,
+ 33.28000259399414,
+ 633.3440551757812,
+ 45.56800079345703,
+ 679.4240112304688,
+ 62.97600173950195,
+ 683.52001953125,
+ 47.61600112915039
+ ],
+ [
+ 634.3680419921875,
+ 52.73600387573242,
+ 628.2240600585938,
+ 66.04800415039062,
+ 674.3040161132812,
+ 81.40800476074219,
+ 678.4000244140625,
+ 68.09600067138672
+ ],
+ [
+ 85.5040054321289,
+ 60.928001403808594,
+ 83.45600128173828,
+ 83.45600128173828,
+ 148.99200439453125,
+ 91.64800262451172,
+ 152.0640106201172,
+ 70.14400482177734
+ ],
+ [
+ 912.8960571289062,
+ 66.04800415039062,
+ 907.7760620117188,
+ 81.40800476074219,
+ 945.6640625,
+ 91.64800262451172,
+ 950.7840576171875,
+ 78.33600616455078
+ ],
+ [
+ 83.45600128173828,
+ 81.40800476074219,
+ 80.38400268554688,
+ 99.84000396728516,
+ 142.84800720214844,
+ 108.0320053100586,
+ 145.92001342773438,
+ 88.57600402832031
+ ],
+ [
+ 906.7520141601562,
+ 84.4800033569336,
+ 902.6560668945312,
+ 96.76800537109375,
+ 941.5680541992188,
+ 109.0560073852539,
+ 945.6640625,
+ 94.72000122070312
+ ],
+ [
+ 88.57600402832031,
+ 97.79200744628906,
+ 85.5040054321289,
+ 114.1760025024414,
+ 139.7760009765625,
+ 123.39200592041016,
+ 142.84800720214844,
+ 107.00800323486328
+ ],
+ [
+ 902.6560668945312,
+ 97.79200744628906,
+ 898.56005859375,
+ 111.10400390625,
+ 936.4480590820312,
+ 123.39200592041016,
+ 939.52001953125,
+ 109.0560073852539
+ ],
+ [
+ 73.21600341796875,
+ 102.91200256347656,
+ 70.14400482177734,
+ 122.36800384521484,
+ 130.5600128173828,
+ 130.5600128173828,
+ 133.6320037841797,
+ 113.15200805664062
+ ],
+ [
+ 891.3920288085938,
+ 108.0320053100586,
+ 887.2960205078125,
+ 121.34400939941406,
+ 926.2080688476562,
+ 133.6320037841797,
+ 930.3040161132812,
+ 120.32000732421875
+ ],
+ [
+ 884.2240600585938,
+ 118.27200317382812,
+ 880.1280517578125,
+ 131.58399963378906,
+ 918.0160522460938,
+ 143.87200927734375,
+ 922.112060546875,
+ 129.5360107421875
+ ],
+ [
+ 838.14404296875,
+ 139.7760009765625,
+ 834.0480346679688,
+ 151.04000854492188,
+ 870.9120483398438,
+ 163.3280029296875,
+ 875.008056640625,
+ 152.0640106201172
+ ],
+ [
+ 341.5040283203125,
+ 178.68800354003906,
+ 333.31201171875,
+ 196.09600830078125,
+ 387.5840148925781,
+ 223.74400329589844,
+ 396.8000183105469,
+ 207.36001586914062
+ ],
+ [
+ 559.6160278320312,
+ 223.74400329589844,
+ 553.4720458984375,
+ 238.08001708984375,
+ 593.4080200195312,
+ 255.48800659179688,
+ 598.5280151367188,
+ 242.17601013183594
+ ],
+ [
+ 425.4720153808594,
+ 225.79200744628906,
+ 416.2560119628906,
+ 242.17601013183594,
+ 456.1920166015625,
+ 263.6800231933594,
+ 465.40802001953125,
+ 249.34400939941406
+ ],
+ [
+ 550.4000244140625,
+ 240.1280059814453,
+ 542.2080078125,
+ 255.48800659179688,
+ 578.0480346679688,
+ 273.9200134277344,
+ 586.2400512695312,
+ 258.55999755859375
+ ],
+ [
+ 689.6640625,
+ 305.66400146484375,
+ 681.4720458984375,
+ 318.97601318359375,
+ 717.31201171875,
+ 349.6960144042969,
+ 726.5280151367188,
+ 338.4320068359375
+ ],
+ [
+ 793.0880126953125,
+ 336.3840026855469,
+ 784.8960571289062,
+ 348.6720275878906,
+ 823.8080444335938,
+ 372.2240295410156,
+ 832.0000610351562,
+ 358.9120178222656
+ ],
+ [
+ 915.968017578125,
+ 378.3680114746094,
+ 906.7520141601562,
+ 389.63201904296875,
+ 941.5680541992188,
+ 418.30401611328125,
+ 950.7840576171875,
+ 406.0160217285156
+ ],
+ [
+ 669.1840209960938,
+ 418.30401611328125,
+ 663.0400390625,
+ 431.61602783203125,
+ 690.6880493164062,
+ 445.9520263671875,
+ 696.83203125,
+ 431.61602783203125
+ ],
+ [
+ 700.9280395507812,
+ 458.2400207519531,
+ 660.9920043945312,
+ 469.5040283203125,
+ 665.0880126953125,
+ 481.7920227050781,
+ 705.0240478515625,
+ 469.5040283203125
+ ],
+ [
+ 723.4560546875,
+ 498.176025390625,
+ 709.1200561523438,
+ 509.4400329589844,
+ 732.6720581054688,
+ 547.3280029296875,
+ 748.0320434570312,
+ 536.0640258789062
+ ],
+ [
+ 778.7520141601562,
+ 725.5040283203125,
+ 760.3200073242188,
+ 742.9120483398438,
+ 801.280029296875,
+ 782.8480224609375,
+ 818.6880493164062,
+ 765.4400634765625
+ ],
+ [
+ 711.1680297851562,
+ 859.6480712890625,
+ 701.9520263671875,
+ 871.93603515625,
+ 762.3680419921875,
+ 918.0160522460938,
+ 771.5840454101562,
+ 905.72802734375
+ ],
+ [
+ 699.904052734375,
+ 873.9840698242188,
+ 688.6400146484375,
+ 888.320068359375,
+ 748.0320434570312,
+ 934.4000244140625,
+ 759.2960205078125,
+ 921.0880737304688
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L577.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 862.7200317382812,
+ 554.4960327148438,
+ 854.5280151367188,
+ 574.9760131835938,
+ 911.8720703125,
+ 598.5280151367188,
+ 921.0880737304688,
+ 577.0240478515625
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L628.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 141.82400512695312,
+ 41.47200012207031,
+ 141.82400512695312,
+ 68.09600067138672,
+ 232.9600067138672,
+ 68.09600067138672,
+ 232.9600067138672,
+ 41.47200012207031
+ ]
+ ]
+ ],
+ "labels": [
+ "cell container"
+ ]
+ }
+ },
+ "L560.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 676.35205078125,
+ 281.0880126953125,
+ 669.1840209960938,
+ 296.4480285644531,
+ 712.1920166015625,
+ 309.760009765625,
+ 718.3360595703125,
+ 294.4000244140625
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L590.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 138.75201416015625,
+ 268.8000183105469,
+ 124.41600799560547,
+ 349.6960144042969,
+ 152.0640106201172,
+ 354.8160095214844,
+ 165.37600708007812,
+ 274.9440002441406
+ ]
+ ],
+ [
+ [
+ 231.93600463867188,
+ 807.4240112304688,
+ 135.6800079345703,
+ 810.4960327148438,
+ 136.70401000976562,
+ 835.072021484375,
+ 232.9600067138672,
+ 830.9760131835938
+ ]
+ ]
+ ],
+ "labels": [
+ "cell container",
+ "container"
+ ]
+ }
+ },
+ "L567.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 373.2480163574219,
+ 293.3760070800781,
+ 363.0080261230469,
+ 318.97601318359375,
+ 457.2160339355469,
+ 357.8880310058594,
+ 468.4800109863281,
+ 333.31201171875
+ ],
+ [
+ 747.008056640625,
+ 821.760009765625,
+ 717.31201171875,
+ 858.6240234375,
+ 728.5760498046875,
+ 868.864013671875,
+ 759.2960205078125,
+ 832.0000610351562
+ ],
+ [
+ 689.6640625,
+ 893.4400634765625,
+ 675.3280029296875,
+ 920.0640258789062,
+ 686.592041015625,
+ 926.2080688476562,
+ 701.9520263671875,
+ 898.56005859375
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L547.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 853.5040283203125,
+ 849.4080200195312,
+ 848.384033203125,
+ 867.8400268554688,
+ 910.8480224609375,
+ 885.248046875,
+ 916.9920654296875,
+ 865.7920532226562
+ ],
+ [
+ 844.2880249023438,
+ 867.8400268554688,
+ 838.14404296875,
+ 887.2960205078125,
+ 900.6080322265625,
+ 905.72802734375,
+ 906.7520141601562,
+ 887.2960205078125
+ ],
+ [
+ 835.072021484375,
+ 890.3680419921875,
+ 828.9280395507812,
+ 908.800048828125,
+ 892.416015625,
+ 928.2560424804688,
+ 897.5360717773438,
+ 908.800048828125
+ ]
+ ],
+ [
+ [
+ 913.9200439453125,
+ 653.8240356445312,
+ 735.7440185546875,
+ 719.3600463867188,
+ 750.0800170898438,
+ 750.0800170898438,
+ 928.2560424804688,
+ 685.5680541992188
+ ]
+ ]
+ ],
+ "labels": [
+ "container",
+ "ore oil"
+ ]
+ }
+ },
+ "L618.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 696.83203125,
+ 105.9840087890625,
+ 633.3440551757812,
+ 162.3040008544922,
+ 887.2960205078125,
+ 396.8000183105469,
+ 950.7840576171875,
+ 341.5040283203125
+ ]
+ ]
+ ],
+ "labels": [
+ "ore oil"
+ ]
+ }
+ },
+ "L623.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 855.5520629882812,
+ 879.1040649414062,
+ 849.4080200195312,
+ 905.72802734375,
+ 1011.2000732421875,
+ 930.3040161132812,
+ 1018.3680419921875,
+ 901.6320190429688
+ ]
+ ],
+ [
+ [
+ 666.112060546875,
+ 896.5120239257812,
+ 643.5840454101562,
+ 908.800048828125,
+ 654.8480224609375,
+ 947.7120361328125,
+ 677.3760375976562,
+ 937.4720458984375
+ ]
+ ]
+ ],
+ "labels": [
+ "cell container",
+ "container"
+ ]
+ }
+ },
+ "L641.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 942.592041015625,
+ 375.2960205078125,
+ 930.3040161132812,
+ 446.97601318359375,
+ 946.6880493164062,
+ 451.072021484375,
+ 960.0000610351562,
+ 379.39202880859375
+ ],
+ [
+ 446.97601318359375,
+ 602.6240234375,
+ 433.6640319824219,
+ 681.4720458984375,
+ 451.072021484375,
+ 683.52001953125,
+ 463.3600158691406,
+ 605.696044921875
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L572.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 683.52001953125,
+ 30.2080020904541,
+ 510.4640197753906,
+ 99.84000396728516,
+ 524.800048828125,
+ 139.7760009765625,
+ 698.8800048828125,
+ 70.14400482177734
+ ]
+ ]
+ ],
+ "labels": [
+ "ore oil"
+ ]
+ }
+ },
+ "L643.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 382.4640197753906,
+ 0.5120000243186951,
+ 369.1520080566406,
+ 18.944000244140625,
+ 563.7120361328125,
+ 156.16000366210938,
+ 578.0480346679688,
+ 129.5360107421875
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L610.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 669.1840209960938,
+ 748.0320434570312,
+ 638.4640502929688,
+ 768.5120239257812,
+ 707.072021484375,
+ 869.8880615234375,
+ 738.8160400390625,
+ 848.384033203125
+ ]
+ ]
+ ],
+ "labels": [
+ "dredger"
+ ]
+ }
+ },
+ "L594.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 49.66400146484375,
+ 460.28802490234375,
+ 35.3280029296875,
+ 473.60003662109375,
+ 79.36000061035156,
+ 511.488037109375,
+ 92.67200469970703,
+ 498.176025390625
+ ],
+ [
+ 134.656005859375,
+ 535.0400390625,
+ 122.36800384521484,
+ 549.3760375976562,
+ 160.25601196289062,
+ 587.2640380859375,
+ 172.54400634765625,
+ 572.9280395507812
+ ],
+ [
+ 459.2640075683594,
+ 817.6640625,
+ 443.9040222167969,
+ 842.2400512695312,
+ 562.6880493164062,
+ 916.9920654296875,
+ 578.0480346679688,
+ 894.4640502929688
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L582.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 80.38400268554688,
+ 225.79200744628906,
+ 78.33600616455078,
+ 252.416015625,
+ 84.4800033569336,
+ 253.4400177001953,
+ 88.57600402832031,
+ 226.81600952148438
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L561.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 706.0480346679688,
+ 0.5120000243186951,
+ 696.83203125,
+ 0.5120000243186951,
+ 742.9120483398438,
+ 34.30400085449219,
+ 752.1280517578125,
+ 24.064001083374023
+ ],
+ [
+ 799.2320556640625,
+ 80.38400268554688,
+ 788.9920654296875,
+ 85.5040054321289,
+ 811.52001953125,
+ 122.36800384521484,
+ 821.760009765625,
+ 116.22400665283203
+ ],
+ [
+ 787.968017578125,
+ 87.552001953125,
+ 777.72802734375,
+ 93.69600677490234,
+ 801.280029296875,
+ 129.5360107421875,
+ 810.4960327148438,
+ 124.41600799560547
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L575.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 228.864013671875,
+ 495.1040344238281,
+ 121.34400939941406,
+ 675.3280029296875,
+ 159.2320098876953,
+ 693.760009765625,
+ 265.72802734375,
+ 514.5599975585938
+ ]
+ ]
+ ],
+ "labels": [
+ "ore oil"
+ ]
+ }
+ },
+ "L615.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 865.7920532226562,
+ 28.160001754760742,
+ 851.4560546875,
+ 67.07200622558594,
+ 867.8400268554688,
+ 74.24000549316406,
+ 884.2240600585938,
+ 35.3280029296875
+ ],
+ [
+ 996.8640747070312,
+ 69.12000274658203,
+ 987.6480712890625,
+ 86.52800750732422,
+ 1023.488037109375,
+ 105.9840087890625,
+ 1023.488037109375,
+ 88.57600402832031
+ ],
+ [
+ 799.2320556640625,
+ 109.0560073852539,
+ 781.8240356445312,
+ 147.96800231933594,
+ 799.2320556640625,
+ 155.13600158691406,
+ 815.6160278320312,
+ 117.24800872802734
+ ],
+ [
+ 947.7120361328125,
+ 129.5360107421875,
+ 940.5440673828125,
+ 143.87200927734375,
+ 969.216064453125,
+ 157.1840057373047,
+ 976.384033203125,
+ 143.87200927734375
+ ],
+ [
+ 937.4720458984375,
+ 148.99200439453125,
+ 930.3040161132812,
+ 163.3280029296875,
+ 957.9520263671875,
+ 175.6160125732422,
+ 965.1200561523438,
+ 162.3040008544922
+ ],
+ [
+ 920.0640258789062,
+ 170.49600219726562,
+ 912.8960571289062,
+ 183.80801391601562,
+ 940.5440673828125,
+ 197.12001037597656,
+ 947.7120361328125,
+ 182.7840118408203
+ ],
+ [
+ 179.71200561523438,
+ 452.09600830078125,
+ 155.13600158691406,
+ 489.9840087890625,
+ 165.37600708007812,
+ 497.15203857421875,
+ 190.97601318359375,
+ 459.2640075683594
+ ],
+ [
+ 200.1920166015625,
+ 465.40802001953125,
+ 167.42401123046875,
+ 503.2960205078125,
+ 177.66400146484375,
+ 511.488037109375,
+ 210.4320068359375,
+ 473.60003662109375
+ ],
+ [
+ 869.8880615234375,
+ 669.1840209960938,
+ 844.2880249023438,
+ 679.4240112304688,
+ 847.3600463867188,
+ 689.6640625,
+ 873.9840698242188,
+ 679.4240112304688
+ ],
+ [
+ 775.6800537109375,
+ 650.7520141601562,
+ 748.0320434570312,
+ 662.0160522460938,
+ 751.1040649414062,
+ 672.2560424804688,
+ 779.7760620117188,
+ 660.9920043945312
+ ],
+ [
+ 745.9840087890625,
+ 660.9920043945312,
+ 719.3600463867188,
+ 673.280029296875,
+ 723.4560546875,
+ 682.4960327148438,
+ 750.0800170898438,
+ 671.2320556640625
+ ],
+ [
+ 671.2320556640625,
+ 671.2320556640625,
+ 639.488037109375,
+ 682.4960327148438,
+ 642.56005859375,
+ 691.7120361328125,
+ 674.3040161132812,
+ 681.4720458984375
+ ],
+ [
+ 673.280029296875,
+ 681.4720458984375,
+ 642.56005859375,
+ 692.7360229492188,
+ 646.656005859375,
+ 701.9520263671875,
+ 677.3760375976562,
+ 689.6640625
+ ],
+ [
+ 678.4000244140625,
+ 685.5680541992188,
+ 648.7040405273438,
+ 697.8560180664062,
+ 652.800048828125,
+ 707.072021484375,
+ 681.4720458984375,
+ 694.7840576171875
+ ],
+ [
+ 762.3680419921875,
+ 693.760009765625,
+ 734.7200317382812,
+ 706.0480346679688,
+ 738.8160400390625,
+ 715.2640380859375,
+ 766.4640502929688,
+ 702.9760131835938
+ ],
+ [
+ 847.3600463867188,
+ 732.6720581054688,
+ 838.14404296875,
+ 741.8880615234375,
+ 860.6720581054688,
+ 767.488037109375,
+ 869.8880615234375,
+ 759.2960205078125
+ ],
+ [
+ 839.1680297851562,
+ 741.8880615234375,
+ 830.9760131835938,
+ 750.0800170898438,
+ 854.5280151367188,
+ 773.6320190429688,
+ 862.7200317382812,
+ 765.4400634765625
+ ],
+ [
+ 828.9280395507812,
+ 751.1040649414062,
+ 819.7120361328125,
+ 759.2960205078125,
+ 842.2400512695312,
+ 781.8240356445312,
+ 851.4560546875,
+ 773.6320190429688
+ ],
+ [
+ 880.1280517578125,
+ 758.2720336914062,
+ 871.93603515625,
+ 766.4640502929688,
+ 894.4640502929688,
+ 790.0160522460938,
+ 902.6560668945312,
+ 781.8240356445312
+ ],
+ [
+ 849.4080200195312,
+ 761.3440551757812,
+ 841.216064453125,
+ 769.5360107421875,
+ 863.7440185546875,
+ 793.0880126953125,
+ 871.93603515625,
+ 783.8720092773438
+ ],
+ [
+ 886.2720336914062,
+ 761.3440551757812,
+ 877.0560302734375,
+ 769.5360107421875,
+ 898.56005859375,
+ 792.0640258789062,
+ 906.7520141601562,
+ 783.8720092773438
+ ],
+ [
+ 936.4480590820312,
+ 776.7040405273438,
+ 929.280029296875,
+ 784.8960571289062,
+ 951.8080444335938,
+ 808.4480590820312,
+ 958.97607421875,
+ 800.2560424804688
+ ],
+ [
+ 928.2560424804688,
+ 784.8960571289062,
+ 920.0640258789062,
+ 794.112060546875,
+ 940.5440673828125,
+ 817.6640625,
+ 948.7360229492188,
+ 810.4960327148438
+ ],
+ [
+ 919.0400390625,
+ 792.0640258789062,
+ 910.8480224609375,
+ 801.280029296875,
+ 932.35205078125,
+ 824.83203125,
+ 940.5440673828125,
+ 815.6160278320312
+ ]
+ ],
+ [
+ [
+ 184.83201599121094,
+ 375.2960205078125,
+ 95.74400329589844,
+ 470.52801513671875,
+ 114.1760025024414,
+ 487.93603515625,
+ 203.26400756835938,
+ 391.6800231933594
+ ]
+ ]
+ ],
+ "labels": [
+ "container",
+ "ore oil"
+ ]
+ }
+ },
+ "L627.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 667.1360473632812,
+ 815.6160278320312,
+ 666.112060546875,
+ 842.2400512695312,
+ 807.4240112304688,
+ 848.384033203125,
+ 808.4480590820312,
+ 820.7360229492188
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L554.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 16.895999908447266,
+ 990.7200317382812,
+ 5.63200044631958,
+ 1008.1280517578125,
+ 47.61600112915039,
+ 1023.488037109375,
+ 56.83200454711914,
+ 1010.176025390625
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L535.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 1001.9840698242188,
+ 414.2080078125,
+ 956.9280395507812,
+ 425.4720153808594,
+ 961.0240478515625,
+ 443.9040222167969,
+ 1006.080078125,
+ 432.6400146484375
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L562.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 288.2560119628906,
+ 774.6560668945312,
+ 282.11199951171875,
+ 783.8720092773438,
+ 315.9040222167969,
+ 809.4720458984375,
+ 323.072021484375,
+ 800.2560424804688
+ ],
+ [
+ 281.0880126953125,
+ 783.8720092773438,
+ 272.8960266113281,
+ 792.0640258789062,
+ 307.7120056152344,
+ 817.6640625,
+ 314.8800048828125,
+ 809.4720458984375
+ ],
+ [
+ 269.8240051269531,
+ 792.0640258789062,
+ 262.656005859375,
+ 801.280029296875,
+ 298.4960021972656,
+ 826.8800659179688,
+ 305.66400146484375,
+ 817.6640625
+ ],
+ [
+ 255.48800659179688,
+ 800.2560424804688,
+ 249.34400939941406,
+ 810.4960327148438,
+ 283.1360168457031,
+ 833.0240478515625,
+ 290.30401611328125,
+ 823.8080444335938
+ ],
+ [
+ 246.2720184326172,
+ 810.4960327148438,
+ 240.1280059814453,
+ 819.7120361328125,
+ 271.87200927734375,
+ 842.2400512695312,
+ 279.0400085449219,
+ 833.0240478515625
+ ],
+ [
+ 241.15200805664062,
+ 820.7360229492188,
+ 235.0080108642578,
+ 830.9760131835938,
+ 265.72802734375,
+ 854.5280151367188,
+ 272.8960266113281,
+ 845.31201171875
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L659.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 341.5040283203125,
+ 426.49603271484375,
+ 330.2400207519531,
+ 539.1360473632812,
+ 360.96002197265625,
+ 543.2320556640625,
+ 374.2720031738281,
+ 430.5920104980469
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L569.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 195.07200622558594,
+ 22.016000747680664,
+ 180.7360076904297,
+ 47.61600112915039,
+ 306.6880187988281,
+ 107.00800323486328,
+ 318.97601318359375,
+ 83.45600128173828
+ ]
+ ]
+ ],
+ "labels": [
+ "ore oil"
+ ]
+ }
+ },
+ "L658.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 849.4080200195312,
+ 170.49600219726562,
+ 747.008056640625,
+ 293.3760070800781,
+ 774.6560668945312,
+ 314.8800048828125,
+ 877.0560302734375,
+ 192.00001525878906
+ ]
+ ]
+ ],
+ "labels": [
+ "ore oil"
+ ]
+ }
+ },
+ "L603.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 798.2080078125,
+ 0.5120000243186951,
+ 796.1600341796875,
+ 15.872000694274902,
+ 861.696044921875,
+ 22.016000747680664,
+ 862.7200317382812,
+ 3.5840001106262207
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L651.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 97.79200744628906,
+ 29.184001922607422,
+ 35.3280029296875,
+ 49.66400146484375,
+ 42.496002197265625,
+ 77.31200408935547,
+ 105.9840087890625,
+ 55.80800247192383
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L640.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 304.6400146484375,
+ 0.5120000243186951,
+ 242.17601013183594,
+ 38.400001525878906,
+ 469.5040283203125,
+ 273.9200134277344,
+ 532.9920043945312,
+ 224.76800537109375
+ ]
+ ]
+ ],
+ "labels": [
+ "ore oil"
+ ]
+ }
+ },
+ "L533.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 348.6720275878906,
+ 510.4640197753906,
+ 334.3360290527344,
+ 527.8720092773438,
+ 372.2240295410156,
+ 563.7120361328125,
+ 387.5840148925781,
+ 545.280029296875
+ ]
+ ],
+ [
+ [
+ 60.928001403808594,
+ 95.74400329589844,
+ 0.5120000243186951,
+ 139.7760009765625,
+ 0.5120000243186951,
+ 157.1840057373047,
+ 72.19200134277344,
+ 114.1760025024414
+ ],
+ [
+ 67.07200622558594,
+ 112.12800598144531,
+ 6.656000137329102,
+ 154.1120147705078,
+ 17.920000076293945,
+ 170.49600219726562,
+ 78.33600616455078,
+ 127.48800659179688
+ ],
+ [
+ 86.52800750732422,
+ 136.70401000976562,
+ 24.064001083374023,
+ 170.49600219726562,
+ 33.28000259399414,
+ 186.8800048828125,
+ 95.74400329589844,
+ 154.1120147705078
+ ]
+ ]
+ ],
+ "labels": [
+ "container",
+ "ore oil"
+ ]
+ }
+ },
+ "L637.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 780.800048828125,
+ 990.7200317382812,
+ 772.6080322265625,
+ 1010.176025390625,
+ 817.6640625,
+ 1023.488037109375,
+ 825.8560180664062,
+ 1007.1040649414062
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L571.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 551.4240112304688,
+ 181.760009765625,
+ 509.4400329589844,
+ 326.1440124511719,
+ 557.5680541992188,
+ 338.4320068359375,
+ 597.5040283203125,
+ 192.00001525878906
+ ],
+ [
+ 264.7040100097656,
+ 742.9120483398438,
+ 225.79200744628906,
+ 909.8240356445312,
+ 280.06402587890625,
+ 920.0640258789062,
+ 318.97601318359375,
+ 753.1520385742188
+ ]
+ ]
+ ],
+ "labels": [
+ "cell container"
+ ]
+ }
+ },
+ "L584.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 348.6720275878906,
+ 118.27200317382812,
+ 263.6800231933594,
+ 127.48800659179688,
+ 265.72802734375,
+ 144.89601135253906,
+ 350.72003173828125,
+ 136.70401000976562
+ ],
+ [
+ 349.6960144042969,
+ 134.656005859375,
+ 271.87200927734375,
+ 147.96800231933594,
+ 272.8960266113281,
+ 159.2320098876953,
+ 350.72003173828125,
+ 144.89601135253906
+ ],
+ [
+ 162.3040008544922,
+ 175.6160125732422,
+ 86.52800750732422,
+ 186.8800048828125,
+ 88.57600402832031,
+ 199.1680145263672,
+ 164.3520050048828,
+ 187.9040069580078
+ ],
+ [
+ 165.37600708007812,
+ 189.95201110839844,
+ 96.76800537109375,
+ 198.14401245117188,
+ 97.79200744628906,
+ 209.4080047607422,
+ 166.40000915527344,
+ 201.21600341796875
+ ],
+ [
+ 171.52000427246094,
+ 201.21600341796875,
+ 100.86400604248047,
+ 213.50401306152344,
+ 102.91200256347656,
+ 227.8400115966797,
+ 173.56800842285156,
+ 214.52801513671875
+ ],
+ [
+ 173.56800842285156,
+ 218.62400817871094,
+ 104.96000671386719,
+ 231.93600463867188,
+ 107.00800323486328,
+ 243.20001220703125,
+ 175.6160125732422,
+ 230.91201782226562
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L559.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 904.7040405273438,
+ 14.848000526428223,
+ 835.072021484375,
+ 47.61600112915039,
+ 842.2400512695312,
+ 66.04800415039062,
+ 912.8960571289062,
+ 34.30400085449219
+ ],
+ [
+ 341.5040283203125,
+ 100.86400604248047,
+ 310.7840270996094,
+ 107.00800323486328,
+ 323.072021484375,
+ 182.7840118408203,
+ 353.7920227050781,
+ 177.66400146484375
+ ],
+ [
+ 624.1280517578125,
+ 280.06402587890625,
+ 590.3359985351562,
+ 361.9840087890625,
+ 616.9600219726562,
+ 371.20001220703125,
+ 650.7520141601562,
+ 289.2799987792969
+ ],
+ [
+ 892.416015625,
+ 538.1119995117188,
+ 814.592041015625,
+ 567.8080444335938,
+ 821.760009765625,
+ 588.2880249023438,
+ 898.56005859375,
+ 557.5680541992188
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L586.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 374.2720031738281,
+ 27.136001586914062,
+ 316.9280090332031,
+ 104.96000671386719,
+ 336.3840026855469,
+ 121.34400939941406,
+ 393.72802734375,
+ 43.52000045776367
+ ],
+ [
+ 38.400001525878906,
+ 219.64801025390625,
+ 32.25600051879883,
+ 227.8400115966797,
+ 56.83200454711914,
+ 245.24801635742188,
+ 62.97600173950195,
+ 237.05601501464844
+ ]
+ ],
+ [
+ [
+ 761.3440551757812,
+ 403.968017578125,
+ 655.8720092773438,
+ 526.8480224609375,
+ 682.4960327148438,
+ 546.3040161132812,
+ 786.9440307617188,
+ 424.4480285644531
+ ]
+ ]
+ ],
+ "labels": [
+ "container",
+ "ore oil"
+ ]
+ }
+ },
+ "L534.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 710.14404296875,
+ 856.5760498046875,
+ 637.4400024414062,
+ 902.6560668945312,
+ 649.72802734375,
+ 925.1840209960938,
+ 723.4560546875,
+ 880.1280517578125
+ ]
+ ]
+ ],
+ "labels": [
+ "dredger"
+ ]
+ }
+ },
+ "L542.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 525.8240356445312,
+ 487.93603515625,
+ 510.4640197753906,
+ 535.0400390625,
+ 529.9200439453125,
+ 540.1600341796875,
+ 544.2560424804688,
+ 494.08001708984375
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L624.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 127.48800659179688,
+ 550.4000244140625,
+ 121.34400939941406,
+ 560.6400146484375,
+ 147.96800231933594,
+ 574.9760131835938,
+ 154.1120147705078,
+ 563.7120361328125
+ ],
+ [
+ 116.22400665283203,
+ 568.83203125,
+ 109.0560073852539,
+ 581.1200561523438,
+ 134.656005859375,
+ 594.4320068359375,
+ 140.8000030517578,
+ 581.1200561523438
+ ],
+ [
+ 90.6240005493164,
+ 597.5040283203125,
+ 85.5040054321289,
+ 608.7680053710938,
+ 109.0560073852539,
+ 621.0560302734375,
+ 115.20000457763672,
+ 609.7920532226562
+ ],
+ [
+ 77.31200408935547,
+ 612.864013671875,
+ 70.14400482177734,
+ 625.1520385742188,
+ 95.74400329589844,
+ 637.4400024414062,
+ 101.88800811767578,
+ 625.1520385742188
+ ],
+ [
+ 67.07200622558594,
+ 630.2720336914062,
+ 60.928001403808594,
+ 642.56005859375,
+ 85.5040054321289,
+ 653.8240356445312,
+ 91.64800262451172,
+ 641.5360107421875
+ ],
+ [
+ 90.6240005493164,
+ 803.3280639648438,
+ 65.02400207519531,
+ 808.4480590820312,
+ 67.07200622558594,
+ 817.6640625,
+ 92.67200469970703,
+ 812.5440673828125
+ ],
+ [
+ 45.56800079345703,
+ 822.7840576171875,
+ 19.968000411987305,
+ 827.904052734375,
+ 22.016000747680664,
+ 838.14404296875,
+ 47.61600112915039,
+ 833.0240478515625
+ ]
+ ],
+ [
+ [
+ 658.9440307617188,
+ 638.4640502929688,
+ 558.592041015625,
+ 738.8160400390625,
+ 584.1920166015625,
+ 761.3440551757812,
+ 683.52001953125,
+ 662.0160522460938
+ ],
+ [
+ 544.2560424804688,
+ 755.2000122070312,
+ 483.84002685546875,
+ 853.5040283203125,
+ 509.4400329589844,
+ 873.9840698242188,
+ 568.83203125,
+ 775.6800537109375
+ ],
+ [
+ 462.3360290527344,
+ 862.7200317382812,
+ 399.87200927734375,
+ 931.3280639648438,
+ 426.49603271484375,
+ 953.8560180664062,
+ 487.93603515625,
+ 886.2720336914062
+ ]
+ ]
+ ],
+ "labels": [
+ "container",
+ "dredger"
+ ]
+ }
+ },
+ "L580.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 971.2640380859375,
+ 78.33600616455078,
+ 931.3280639648438,
+ 199.1680145263672,
+ 960.0000610351562,
+ 208.38400268554688,
+ 999.93603515625,
+ 87.552001953125
+ ]
+ ],
+ [
+ [
+ 669.1840209960938,
+ 51.71200180053711,
+ 657.9200439453125,
+ 87.552001953125,
+ 683.52001953125,
+ 93.69600677490234,
+ 694.7840576171875,
+ 58.880001068115234
+ ]
+ ]
+ ],
+ "labels": [
+ "cell container",
+ "container"
+ ]
+ }
+ },
+ "L551.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 857.6000366210938,
+ 58.880001068115234,
+ 855.5520629882812,
+ 71.16800689697266,
+ 918.0160522460938,
+ 83.45600128173828,
+ 920.0640258789062,
+ 70.14400482177734
+ ],
+ [
+ 854.5280151367188,
+ 71.16800689697266,
+ 853.5040283203125,
+ 84.4800033569336,
+ 911.8720703125,
+ 93.69600677490234,
+ 913.9200439453125,
+ 82.4320068359375
+ ],
+ [
+ 854.5280151367188,
+ 81.40800476074219,
+ 853.5040283203125,
+ 93.69600677490234,
+ 910.8480224609375,
+ 104.96000671386719,
+ 912.8960571289062,
+ 92.67200469970703
+ ],
+ [
+ 854.5280151367188,
+ 93.69600677490234,
+ 853.5040283203125,
+ 104.96000671386719,
+ 921.0880737304688,
+ 116.22400665283203,
+ 922.112060546875,
+ 103.93600463867188
+ ],
+ [
+ 854.5280151367188,
+ 104.96000671386719,
+ 853.5040283203125,
+ 117.24800872802734,
+ 909.8240356445312,
+ 127.48800659179688,
+ 911.8720703125,
+ 117.24800872802734
+ ],
+ [
+ 854.5280151367188,
+ 117.24800872802734,
+ 853.5040283203125,
+ 128.5120086669922,
+ 911.8720703125,
+ 139.7760009765625,
+ 913.9200439453125,
+ 128.5120086669922
+ ]
+ ],
+ [
+ [
+ 383.4880065917969,
+ 47.61600112915039,
+ 382.4640197753906,
+ 59.90400314331055,
+ 424.4480285644531,
+ 64.0,
+ 425.4720153808594,
+ 51.71200180053711
+ ],
+ [
+ 511.488037109375,
+ 50.68800354003906,
+ 510.4640197753906,
+ 62.97600173950195,
+ 535.0400390625,
+ 68.09600067138672,
+ 537.0880126953125,
+ 55.80800247192383
+ ],
+ [
+ 543.2320556640625,
+ 56.83200454711914,
+ 542.2080078125,
+ 69.12000274658203,
+ 578.0480346679688,
+ 73.21600341796875,
+ 579.072021484375,
+ 60.928001403808594
+ ],
+ [
+ 672.2560424804688,
+ 55.80800247192383,
+ 667.1360473632812,
+ 100.86400604248047,
+ 675.3280029296875,
+ 101.88800811767578,
+ 681.4720458984375,
+ 56.83200454711914
+ ],
+ [
+ 686.592041015625,
+ 55.80800247192383,
+ 681.4720458984375,
+ 101.88800811767578,
+ 689.6640625,
+ 103.93600463867188,
+ 694.7840576171875,
+ 56.83200454711914
+ ],
+ [
+ 700.9280395507812,
+ 58.880001068115234,
+ 694.7840576171875,
+ 103.93600463867188,
+ 704.0000610351562,
+ 104.96000671386719,
+ 710.14404296875,
+ 59.90400314331055
+ ],
+ [
+ 715.2640380859375,
+ 61.952003479003906,
+ 709.1200561523438,
+ 107.00800323486328,
+ 718.3360595703125,
+ 108.0320053100586,
+ 725.5040283203125,
+ 62.97600173950195
+ ],
+ [
+ 728.5760498046875,
+ 62.97600173950195,
+ 722.4320068359375,
+ 108.0320053100586,
+ 730.6240234375,
+ 110.08000183105469,
+ 737.7920532226562,
+ 64.0
+ ],
+ [
+ 740.864013671875,
+ 73.21600341796875,
+ 734.7200317382812,
+ 116.22400665283203,
+ 743.93603515625,
+ 118.27200317382812,
+ 750.0800170898438,
+ 75.26399993896484
+ ],
+ [
+ 205.31201171875,
+ 69.12000274658203,
+ 204.2880096435547,
+ 81.40800476074219,
+ 240.1280059814453,
+ 85.5040054321289,
+ 241.15200805664062,
+ 73.21600341796875
+ ],
+ [
+ 200.1920166015625,
+ 81.40800476074219,
+ 198.14401245117188,
+ 92.67200469970703,
+ 240.1280059814453,
+ 96.76800537109375,
+ 241.15200805664062,
+ 85.5040054321289
+ ],
+ [
+ 197.12001037597656,
+ 90.6240005493164,
+ 196.09600830078125,
+ 102.91200256347656,
+ 245.24801635742188,
+ 108.0320053100586,
+ 246.2720184326172,
+ 96.76800537109375
+ ],
+ [
+ 242.17601013183594,
+ 98.81600189208984,
+ 200.1920166015625,
+ 102.91200256347656,
+ 201.21600341796875,
+ 113.15200805664062,
+ 243.20001220703125,
+ 109.0560073852539
+ ]
+ ]
+ ],
+ "labels": [
+ "container",
+ "lawenforce"
+ ]
+ }
+ },
+ "L636.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 252.416015625,
+ 293.3760070800781,
+ 212.48001098632812,
+ 322.0480041503906,
+ 220.67201232910156,
+ 335.3600158691406,
+ 261.63201904296875,
+ 305.66400146484375
+ ],
+ [
+ 260.6080017089844,
+ 310.7840270996094,
+ 224.76800537109375,
+ 342.52801513671875,
+ 233.9840087890625,
+ 354.8160095214844,
+ 269.8240051269531,
+ 323.072021484375
+ ],
+ [
+ 294.4000244140625,
+ 328.1920166015625,
+ 256.51202392578125,
+ 357.8880310058594,
+ 264.7040100097656,
+ 371.20001220703125,
+ 302.5920104980469,
+ 340.4800109863281
+ ],
+ [
+ 308.73602294921875,
+ 346.6240234375,
+ 269.8240051269531,
+ 377.3440246582031,
+ 278.0160217285156,
+ 389.63201904296875,
+ 316.9280090332031,
+ 359.9360046386719
+ ],
+ [
+ 334.3360290527344,
+ 372.2240295410156,
+ 296.4480285644531,
+ 399.87200927734375,
+ 303.61602783203125,
+ 414.2080078125,
+ 342.52801513671875,
+ 385.5360107421875
+ ],
+ [
+ 345.6000061035156,
+ 389.63201904296875,
+ 308.73602294921875,
+ 418.30401611328125,
+ 316.9280090332031,
+ 431.61602783203125,
+ 354.8160095214844,
+ 403.968017578125
+ ],
+ [
+ 372.2240295410156,
+ 402.94403076171875,
+ 334.3360290527344,
+ 432.6400146484375,
+ 342.52801513671875,
+ 445.9520263671875,
+ 381.4400329589844,
+ 417.280029296875
+ ],
+ [
+ 383.4880065917969,
+ 411.1360168457031,
+ 345.6000061035156,
+ 442.8800354003906,
+ 354.8160095214844,
+ 457.2160339355469,
+ 393.72802734375,
+ 425.4720153808594
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L629.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 624.1280517578125,
+ 849.4080200195312,
+ 512.5120239257812,
+ 892.416015625,
+ 521.72802734375,
+ 918.0160522460938,
+ 633.3440551757812,
+ 875.008056640625
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L663.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 105.9840087890625,
+ 519.6800537109375,
+ 98.81600189208984,
+ 530.9440307617188,
+ 122.36800384521484,
+ 551.4240112304688,
+ 130.5600128173828,
+ 539.1360473632812
+ ],
+ [
+ 91.64800262451172,
+ 537.0880126953125,
+ 84.4800033569336,
+ 549.3760375976562,
+ 109.0560073852539,
+ 568.83203125,
+ 117.24800872802734,
+ 557.5680541992188
+ ],
+ [
+ 81.40800476074219,
+ 552.447998046875,
+ 74.24000549316406,
+ 563.7120361328125,
+ 97.79200744628906,
+ 582.14404296875,
+ 104.96000671386719,
+ 570.8800048828125
+ ],
+ [
+ 67.07200622558594,
+ 567.8080444335938,
+ 59.90400314331055,
+ 579.072021484375,
+ 85.5040054321289,
+ 596.4800415039062,
+ 92.67200469970703,
+ 585.2160034179688
+ ],
+ [
+ 54.78400421142578,
+ 581.1200561523438,
+ 47.61600112915039,
+ 592.384033203125,
+ 74.24000549316406,
+ 610.8160400390625,
+ 81.40800476074219,
+ 598.5280151367188
+ ],
+ [
+ 40.448001861572266,
+ 599.552001953125,
+ 34.30400085449219,
+ 610.8160400390625,
+ 60.928001403808594,
+ 629.248046875,
+ 68.09600067138672,
+ 617.9840087890625
+ ],
+ [
+ 29.184001922607422,
+ 615.93603515625,
+ 23.040000915527344,
+ 627.2000122070312,
+ 47.61600112915039,
+ 644.6080322265625,
+ 54.78400421142578,
+ 633.3440551757812
+ ],
+ [
+ 159.2320098876953,
+ 738.8160400390625,
+ 120.32000732421875,
+ 749.0560302734375,
+ 123.39200592041016,
+ 762.3680419921875,
+ 161.28001403808594,
+ 752.1280517578125
+ ],
+ [
+ 157.1840057373047,
+ 754.176025390625,
+ 121.34400939941406,
+ 764.416015625,
+ 124.41600799560547,
+ 775.6800537109375,
+ 160.25601196289062,
+ 766.4640502929688
+ ],
+ [
+ 105.9840087890625,
+ 756.2240600585938,
+ 64.0,
+ 765.4400634765625,
+ 66.04800415039062,
+ 777.72802734375,
+ 108.0320053100586,
+ 768.5120239257812
+ ],
+ [
+ 111.10400390625,
+ 766.4640502929688,
+ 69.12000274658203,
+ 775.6800537109375,
+ 71.16800689697266,
+ 787.968017578125,
+ 114.1760025024414,
+ 778.7520141601562
+ ],
+ [
+ 28.160001754760742,
+ 771.5840454101562,
+ 0.5120000243186951,
+ 781.8240356445312,
+ 0.5120000243186951,
+ 794.112060546875,
+ 31.23200225830078,
+ 783.8720092773438
+ ],
+ [
+ 28.160001754760742,
+ 781.8240356445312,
+ 0.5120000243186951,
+ 792.0640258789062,
+ 0.5120000243186951,
+ 803.3280639648438,
+ 31.23200225830078,
+ 792.0640258789062
+ ],
+ [
+ 9.72800064086914,
+ 928.2560424804688,
+ 0.5120000243186951,
+ 941.5680541992188,
+ 23.040000915527344,
+ 963.072021484375,
+ 34.30400085449219,
+ 948.7360229492188
+ ],
+ [
+ 36.35200119018555,
+ 955.904052734375,
+ 28.160001754760742,
+ 965.1200561523438,
+ 51.71200180053711,
+ 984.5760498046875,
+ 59.90400314331055,
+ 975.3600463867188
+ ],
+ [
+ 70.14400482177734,
+ 981.5040283203125,
+ 61.952003479003906,
+ 990.7200317382812,
+ 85.5040054321289,
+ 1011.2000732421875,
+ 92.67200469970703,
+ 1001.9840698242188
+ ]
+ ],
+ [
+ [
+ 595.4560546875,
+ 623.10400390625,
+ 540.1600341796875,
+ 698.8800048828125,
+ 568.83203125,
+ 717.31201171875,
+ 623.10400390625,
+ 641.5360107421875
+ ],
+ [
+ 523.7760009765625,
+ 725.5040283203125,
+ 466.4320373535156,
+ 802.3040161132812,
+ 494.08001708984375,
+ 819.7120361328125,
+ 551.4240112304688,
+ 742.9120483398438
+ ],
+ [
+ 440.83203125,
+ 826.8800659179688,
+ 383.4880065917969,
+ 901.6320190429688,
+ 413.18402099609375,
+ 922.112060546875,
+ 469.5040283203125,
+ 847.3600463867188
+ ],
+ [
+ 377.3440246582031,
+ 908.800048828125,
+ 315.9040222167969,
+ 981.5040283203125,
+ 341.5040283203125,
+ 1001.9840698242188,
+ 401.9200134277344,
+ 928.2560424804688
+ ]
+ ]
+ ],
+ "labels": [
+ "container",
+ "dredger"
+ ]
+ }
+ },
+ "L606.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 358.9120178222656,
+ 153.0880126953125,
+ 350.72003173828125,
+ 169.47201538085938,
+ 387.5840148925781,
+ 187.9040069580078,
+ 394.75201416015625,
+ 170.49600219726562
+ ],
+ [
+ 354.8160095214844,
+ 155.13600158691406,
+ 346.6240234375,
+ 170.49600219726562,
+ 384.51202392578125,
+ 187.9040069580078,
+ 390.656005859375,
+ 171.52000427246094
+ ],
+ [
+ 990.7200317382812,
+ 558.592041015625,
+ 982.528076171875,
+ 577.0240478515625,
+ 1021.4400634765625,
+ 591.3600463867188,
+ 1023.488037109375,
+ 573.9520263671875
+ ],
+ [
+ 0.5120000243186951,
+ 704.0000610351562,
+ 0.5120000243186951,
+ 736.7680053710938,
+ 62.97600173950195,
+ 754.176025390625,
+ 71.16800689697266,
+ 720.384033203125
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L563.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 456.1920166015625,
+ 652.800048828125,
+ 402.94403076171875,
+ 700.9280395507812,
+ 420.3520202636719,
+ 717.31201171875,
+ 471.5520324707031,
+ 669.1840209960938
+ ],
+ [
+ 805.3760375976562,
+ 689.6640625,
+ 728.5760498046875,
+ 736.7680053710938,
+ 740.864013671875,
+ 755.2000122070312,
+ 815.6160278320312,
+ 709.1200561523438
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L614.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 936.4480590820312,
+ 81.40800476074219,
+ 877.0560302734375,
+ 85.5040054321289,
+ 878.0800170898438,
+ 103.93600463867188,
+ 937.4720458984375,
+ 99.84000396728516
+ ],
+ [
+ 958.97607421875,
+ 119.29600524902344,
+ 899.5840454101562,
+ 123.39200592041016,
+ 900.6080322265625,
+ 141.82400512695312,
+ 960.0000610351562,
+ 137.72801208496094
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L665.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 696.83203125,
+ 875.008056640625,
+ 633.3440551757812,
+ 926.2080688476562,
+ 762.3680419921875,
+ 1023.488037109375,
+ 910.8480224609375,
+ 1023.488037109375
+ ]
+ ]
+ ],
+ "labels": [
+ "ore oil"
+ ]
+ }
+ },
+ "L553.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 568.83203125,
+ 345.6000061035156,
+ 541.1840209960938,
+ 373.2480163574219,
+ 552.447998046875,
+ 387.5840148925781,
+ 582.14404296875,
+ 359.9360046386719
+ ],
+ [
+ 754.176025390625,
+ 590.3359985351562,
+ 712.1920166015625,
+ 602.6240234375,
+ 718.3360595703125,
+ 629.248046875,
+ 760.3200073242188,
+ 615.93603515625
+ ],
+ [
+ 904.7040405273438,
+ 783.8720092773438,
+ 834.0480346679688,
+ 814.592041015625,
+ 842.2400512695312,
+ 834.0480346679688,
+ 912.8960571289062,
+ 801.280029296875
+ ],
+ [
+ 341.5040283203125,
+ 870.9120483398438,
+ 312.8320007324219,
+ 875.008056640625,
+ 323.072021484375,
+ 953.8560180664062,
+ 350.72003173828125,
+ 948.7360229492188
+ ]
+ ],
+ [
+ [
+ 926.2080688476562,
+ 493.0560302734375,
+ 845.31201171875,
+ 514.5599975585938,
+ 849.4080200195312,
+ 532.9920043945312,
+ 931.3280639648438,
+ 511.488037109375
+ ]
+ ]
+ ],
+ "labels": [
+ "container",
+ "dredger"
+ ]
+ }
+ },
+ "L635.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 561.6640014648438,
+ 424.4480285644531,
+ 561.6640014648438,
+ 476.6720275878906,
+ 635.3920288085938,
+ 476.6720275878906,
+ 635.3920288085938,
+ 424.4480285644531
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L578.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 977.4080200195312,
+ 620.0320434570312,
+ 896.5120239257812,
+ 653.8240356445312,
+ 905.72802734375,
+ 674.3040161132812,
+ 985.6000366210938,
+ 640.5120239257812
+ ],
+ [
+ 767.488037109375,
+ 724.4800415039062,
+ 637.4400024414062,
+ 775.6800537109375,
+ 647.6800537109375,
+ 806.4000244140625,
+ 778.7520141601562,
+ 754.176025390625
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L619.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 543.2320556640625,
+ 364.0320129394531,
+ 535.0400390625,
+ 380.416015625,
+ 569.8560180664062,
+ 393.72802734375,
+ 577.0240478515625,
+ 377.3440246582031
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L633.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 341.5040283203125,
+ 526.8480224609375,
+ 284.1600036621094,
+ 550.4000244140625,
+ 291.3280029296875,
+ 568.83203125,
+ 348.6720275878906,
+ 546.3040161132812
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L543.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 25.088001251220703,
+ 919.0400390625,
+ 0.5120000243186951,
+ 943.6160278320312,
+ 78.33600616455078,
+ 1023.488037109375,
+ 102.91200256347656,
+ 1000.9600219726562
+ ]
+ ],
+ [
+ [
+ 781.8240356445312,
+ 810.4960327148438,
+ 758.2720336914062,
+ 834.0480346679688,
+ 857.6000366210938,
+ 935.424072265625,
+ 881.1520385742188,
+ 908.800048828125
+ ],
+ [
+ 212.48001098632812,
+ 347.64801025390625,
+ 194.04800415039062,
+ 366.08001708984375,
+ 283.1360168457031,
+ 457.2160339355469,
+ 301.5680236816406,
+ 435.7120056152344
+ ]
+ ]
+ ],
+ "labels": [
+ "cell container",
+ "dredger"
+ ]
+ }
+ },
+ "L573.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 113.15200805664062,
+ 422.4000244140625,
+ 15.872000694274902,
+ 564.7360229492188,
+ 62.97600173950195,
+ 594.4320068359375,
+ 160.25601196289062,
+ 451.072021484375
+ ]
+ ]
+ ],
+ "labels": [
+ "ore oil"
+ ]
+ }
+ },
+ "L541.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 31.23200225830078,
+ 115.20000457763672,
+ 20.992000579833984,
+ 126.46400451660156,
+ 67.07200622558594,
+ 170.49600219726562,
+ 78.33600616455078,
+ 159.2320098876953
+ ],
+ [
+ 19.968000411987305,
+ 125.44000244140625,
+ 9.72800064086914,
+ 135.6800079345703,
+ 51.71200180053711,
+ 177.66400146484375,
+ 61.952003479003906,
+ 167.42401123046875
+ ],
+ [
+ 9.72800064086914,
+ 133.6320037841797,
+ 0.5120000243186951,
+ 145.92001342773438,
+ 40.448001861572266,
+ 185.8560028076172,
+ 49.66400146484375,
+ 173.56800842285156
+ ],
+ [
+ 0.5120000243186951,
+ 145.92001342773438,
+ 0.5120000243186951,
+ 156.16000366210938,
+ 40.448001861572266,
+ 195.07200622558594,
+ 49.66400146484375,
+ 184.83201599121094
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L602.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 877.0560302734375,
+ 614.9120483398438,
+ 842.2400512695312,
+ 743.93603515625,
+ 877.0560302734375,
+ 754.176025390625,
+ 911.8720703125,
+ 624.1280517578125
+ ]
+ ],
+ [
+ [
+ 247.29600524902344,
+ 90.6240005493164,
+ 205.31201171875,
+ 158.2080078125,
+ 223.74400329589844,
+ 170.49600219726562,
+ 266.75201416015625,
+ 103.93600463867188
+ ],
+ [
+ 585.2160034179688,
+ 168.44801330566406,
+ 534.0160522460938,
+ 528.89599609375,
+ 567.8080444335938,
+ 542.2080078125,
+ 619.008056640625,
+ 180.7360076904297
+ ],
+ [
+ 155.13600158691406,
+ 631.2960205078125,
+ 99.84000396728516,
+ 699.904052734375,
+ 127.48800659179688,
+ 721.4080200195312,
+ 180.7360076904297,
+ 650.7520141601562
+ ],
+ [
+ 454.1440124511719,
+ 767.488037109375,
+ 409.0880126953125,
+ 865.7920532226562,
+ 434.6880187988281,
+ 880.1280517578125,
+ 481.7920227050781,
+ 781.8240356445312
+ ],
+ [
+ 856.5760498046875,
+ 929.280029296875,
+ 815.6160278320312,
+ 1023.488037109375,
+ 840.1920166015625,
+ 1023.488037109375,
+ 883.2000122070312,
+ 941.5680541992188
+ ]
+ ]
+ ],
+ "labels": [
+ "cell container",
+ "container"
+ ]
+ }
+ },
+ "L639.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 842.2400512695312,
+ 227.8400115966797,
+ 836.0960693359375,
+ 235.0080108642578,
+ 866.8160400390625,
+ 261.63201904296875,
+ 872.9600219726562,
+ 255.48800659179688
+ ],
+ [
+ 833.0240478515625,
+ 233.9840087890625,
+ 826.8800659179688,
+ 241.15200805664062,
+ 857.6000366210938,
+ 267.7760009765625,
+ 863.7440185546875,
+ 260.6080017089844
+ ]
+ ]
+ ],
+ "labels": [
+ "fishing"
+ ]
+ }
+ },
+ "L565.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 25.088001251220703,
+ 219.64801025390625,
+ 0.5120000243186951,
+ 237.05601501464844,
+ 5.63200044631958,
+ 264.7040100097656,
+ 34.30400085449219,
+ 249.34400939941406
+ ],
+ [
+ 38.400001525878906,
+ 233.9840087890625,
+ 0.5120000243186951,
+ 250.36801147460938,
+ 2.56000018119812,
+ 275.968017578125,
+ 46.592002868652344,
+ 258.55999755859375
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L576.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 772.6080322265625,
+ 170.49600219726562,
+ 695.8080444335938,
+ 224.76800537109375,
+ 716.2880249023438,
+ 250.36801147460938,
+ 793.0880126953125,
+ 195.07200622558594
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L649.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 886.2720336914062,
+ 484.864013671875,
+ 868.864013671875,
+ 494.08001708984375,
+ 899.5840454101562,
+ 554.4960327148438,
+ 915.968017578125,
+ 545.280029296875
+ ],
+ [
+ 867.8400268554688,
+ 494.08001708984375,
+ 851.4560546875,
+ 503.2960205078125,
+ 879.1040649414062,
+ 562.6880493164062,
+ 895.488037109375,
+ 554.4960327148438
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L626.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 373.2480163574219,
+ 129.5360107421875,
+ 356.864013671875,
+ 157.1840057373047,
+ 373.2480163574219,
+ 168.44801330566406,
+ 389.63201904296875,
+ 138.75201416015625
+ ],
+ [
+ 243.20001220703125,
+ 151.04000854492188,
+ 229.8880157470703,
+ 168.44801330566406,
+ 240.1280059814453,
+ 176.6400146484375,
+ 254.46400451660156,
+ 159.2320098876953
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L545.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 765.4400634765625,
+ 805.3760375976562,
+ 676.35205078125,
+ 830.9760131835938,
+ 682.4960327148438,
+ 855.5520629882812,
+ 771.5840454101562,
+ 829.9520263671875
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L612.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 305.66400146484375,
+ 564.7360229492188,
+ 297.4720153808594,
+ 616.9600219726562,
+ 311.8080139160156,
+ 621.0560302734375,
+ 322.0480041503906,
+ 566.7839965820312
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L598.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 133.6320037841797,
+ 1.5360000133514404,
+ 128.5120086669922,
+ 7.680000305175781,
+ 156.16000366210938,
+ 30.2080020904541,
+ 161.28001403808594,
+ 24.064001083374023
+ ],
+ [
+ 125.44000244140625,
+ 8.704000473022461,
+ 121.34400939941406,
+ 14.848000526428223,
+ 146.94400024414062,
+ 37.37600326538086,
+ 152.0640106201172,
+ 31.23200225830078
+ ],
+ [
+ 121.34400939941406,
+ 15.872000694274902,
+ 116.22400665283203,
+ 22.016000747680664,
+ 142.84800720214844,
+ 43.52000045776367,
+ 146.94400024414062,
+ 38.400001525878906
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L552.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 904.7040405273438,
+ 41.47200012207031,
+ 904.7040405273438,
+ 66.04800415039062,
+ 1000.9600219726562,
+ 66.04800415039062,
+ 1000.9600219726562,
+ 41.47200012207031
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L566.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 700.9280395507812,
+ 54.78400421142578,
+ 624.1280517578125,
+ 85.5040054321289,
+ 637.4400024414062,
+ 112.12800598144531,
+ 714.2400512695312,
+ 82.4320068359375
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L601.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 654.8480224609375,
+ 832.0000610351562,
+ 511.488037109375,
+ 854.5280151367188,
+ 516.6080322265625,
+ 884.2240600585938,
+ 658.9440307617188,
+ 860.6720581054688
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L607.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 430.5920104980469,
+ 256.51202392578125,
+ 377.3440246582031,
+ 282.11199951171875,
+ 387.5840148925781,
+ 301.5680236816406,
+ 440.83203125,
+ 275.968017578125
+ ],
+ [
+ 27.136001586914062,
+ 431.61602783203125,
+ 0.5120000243186951,
+ 532.9920043945312,
+ 25.088001251220703,
+ 544.2560424804688,
+ 59.90400314331055,
+ 444.9280090332031
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L587.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 549.3760375976562,
+ 0.5120000243186951,
+ 540.1600341796875,
+ 0.5120000243186951,
+ 562.6880493164062,
+ 18.944000244140625,
+ 571.904052734375,
+ 7.680000305175781
+ ],
+ [
+ 536.0640258789062,
+ 9.72800064086914,
+ 526.8480224609375,
+ 20.992000579833984,
+ 551.4240112304688,
+ 40.448001861572266,
+ 558.592041015625,
+ 29.184001922607422
+ ],
+ [
+ 523.7760009765625,
+ 30.2080020904541,
+ 514.5599975585938,
+ 40.448001861572266,
+ 537.0880126953125,
+ 58.880001068115234,
+ 546.3040161132812,
+ 47.61600112915039
+ ],
+ [
+ 915.968017578125,
+ 32.25600051879883,
+ 903.6800537109375,
+ 49.66400146484375,
+ 938.4960327148438,
+ 74.24000549316406,
+ 950.7840576171875,
+ 55.80800247192383
+ ],
+ [
+ 741.8880615234375,
+ 62.97600173950195,
+ 721.4080200195312,
+ 86.52800750732422,
+ 734.7200317382812,
+ 97.79200744628906,
+ 756.2240600585938,
+ 74.24000549316406
+ ],
+ [
+ 895.488037109375,
+ 71.16800689697266,
+ 889.3440551757812,
+ 80.38400268554688,
+ 910.8480224609375,
+ 96.76800537109375,
+ 918.0160522460938,
+ 86.52800750732422
+ ],
+ [
+ 885.248046875,
+ 82.4320068359375,
+ 875.008056640625,
+ 93.69600677490234,
+ 898.56005859375,
+ 111.10400390625,
+ 908.800048828125,
+ 98.81600189208984
+ ],
+ [
+ 886.2720336914062,
+ 85.5040054321289,
+ 877.0560302734375,
+ 96.76800537109375,
+ 896.5120239257812,
+ 111.10400390625,
+ 905.72802734375,
+ 98.81600189208984
+ ],
+ [
+ 710.14404296875,
+ 93.69600677490234,
+ 688.6400146484375,
+ 122.36800384521484,
+ 702.9760131835938,
+ 132.60800170898438,
+ 724.4800415039062,
+ 104.96000671386719
+ ],
+ [
+ 866.8160400390625,
+ 100.86400604248047,
+ 859.6480712890625,
+ 111.10400390625,
+ 883.2000122070312,
+ 127.48800659179688,
+ 890.3680419921875,
+ 117.24800872802734
+ ],
+ [
+ 858.6240234375,
+ 121.34400939941406,
+ 852.4800415039062,
+ 130.5600128173828,
+ 873.9840698242188,
+ 146.94400024414062,
+ 881.1520385742188,
+ 136.70401000976562
+ ],
+ [
+ 684.5440063476562,
+ 131.58399963378906,
+ 664.0640258789062,
+ 164.3520050048828,
+ 680.4480590820312,
+ 174.59201049804688,
+ 699.904052734375,
+ 141.82400512695312
+ ],
+ [
+ 849.4080200195312,
+ 134.656005859375,
+ 842.2400512695312,
+ 144.89601135253906,
+ 863.7440185546875,
+ 162.3040008544922,
+ 870.9120483398438,
+ 151.04000854492188
+ ],
+ [
+ 827.904052734375,
+ 161.28001403808594,
+ 820.7360229492188,
+ 170.49600219726562,
+ 842.2400512695312,
+ 186.8800048828125,
+ 849.4080200195312,
+ 176.6400146484375
+ ],
+ [
+ 949.7600708007812,
+ 296.4480285644531,
+ 920.0640258789062,
+ 305.66400146484375,
+ 922.112060546875,
+ 311.8080139160156,
+ 952.83203125,
+ 301.5680236816406
+ ],
+ [
+ 952.83203125,
+ 304.6400146484375,
+ 924.1600341796875,
+ 314.8800048828125,
+ 926.2080688476562,
+ 321.0240173339844,
+ 954.8800659179688,
+ 310.7840270996094
+ ],
+ [
+ 941.5680541992188,
+ 304.6400146484375,
+ 921.0880737304688,
+ 313.85601806640625,
+ 924.1600341796875,
+ 320.0,
+ 944.6400146484375,
+ 310.7840270996094
+ ],
+ [
+ 900.6080322265625,
+ 311.8080139160156,
+ 872.9600219726562,
+ 322.0480041503906,
+ 875.008056640625,
+ 328.1920166015625,
+ 902.6560668945312,
+ 318.97601318359375
+ ],
+ [
+ 902.6560668945312,
+ 323.072021484375,
+ 875.008056640625,
+ 332.28802490234375,
+ 877.0560302734375,
+ 338.4320068359375,
+ 904.7040405273438,
+ 328.1920166015625
+ ],
+ [
+ 856.5760498046875,
+ 323.072021484375,
+ 828.9280395507812,
+ 333.31201171875,
+ 830.9760131835938,
+ 339.4560241699219,
+ 859.6480712890625,
+ 329.21600341796875
+ ],
+ [
+ 903.6800537109375,
+ 326.1440124511719,
+ 877.0560302734375,
+ 336.3840026855469,
+ 879.1040649414062,
+ 341.5040283203125,
+ 906.7520141601562,
+ 332.28802490234375
+ ],
+ [
+ 905.72802734375,
+ 327.16802978515625,
+ 880.1280517578125,
+ 337.40802001953125,
+ 882.176025390625,
+ 342.52801513671875,
+ 907.7760620117188,
+ 333.31201171875
+ ],
+ [
+ 810.4960327148438,
+ 338.4320068359375,
+ 780.800048828125,
+ 347.64801025390625,
+ 782.8480224609375,
+ 353.7920227050781,
+ 812.5440673828125,
+ 343.552001953125
+ ],
+ [
+ 680.4480590820312,
+ 376.32000732421875,
+ 652.800048828125,
+ 385.5360107421875,
+ 655.8720092773438,
+ 391.6800231933594,
+ 682.4960327148438,
+ 382.4640197753906
+ ],
+ [
+ 680.4480590820312,
+ 385.5360107421875,
+ 653.8240356445312,
+ 395.7760314941406,
+ 656.8960571289062,
+ 401.9200134277344,
+ 682.4960327148438,
+ 391.6800231933594
+ ],
+ [
+ 500.2240295410156,
+ 459.2640075683594,
+ 467.4560241699219,
+ 470.52801513671875,
+ 469.5040283203125,
+ 477.6960144042969,
+ 503.2960205078125,
+ 467.4560241699219
+ ],
+ [
+ 458.2400207519531,
+ 487.93603515625,
+ 429.5680236816406,
+ 498.176025390625,
+ 431.61602783203125,
+ 504.3200378417969,
+ 460.28802490234375,
+ 494.08001708984375
+ ],
+ [
+ 412.1600341796875,
+ 495.1040344238281,
+ 383.4880065917969,
+ 506.3680114746094,
+ 385.5360107421875,
+ 512.5120239257812,
+ 416.2560119628906,
+ 502.27203369140625
+ ],
+ [
+ 460.28802490234375,
+ 502.27203369140625,
+ 433.6640319824219,
+ 511.488037109375,
+ 436.73602294921875,
+ 517.6320190429688,
+ 463.3600158691406,
+ 507.39202880859375
+ ],
+ [
+ 463.3600158691406,
+ 504.3200378417969,
+ 438.7840270996094,
+ 513.5360107421875,
+ 441.85601806640625,
+ 519.6800537109375,
+ 466.4320373535156,
+ 510.4640197753906
+ ],
+ [
+ 377.3440246582031,
+ 511.488037109375,
+ 346.6240234375,
+ 522.7520141601562,
+ 349.6960144042969,
+ 528.89599609375,
+ 379.39202880859375,
+ 517.6320190429688
+ ],
+ [
+ 315.9040222167969,
+ 518.656005859375,
+ 284.1600036621094,
+ 529.9200439453125,
+ 287.2320251464844,
+ 537.0880126953125,
+ 318.97601318359375,
+ 525.8240356445312
+ ],
+ [
+ 377.3440246582031,
+ 524.800048828125,
+ 346.6240234375,
+ 535.0400390625,
+ 349.6960144042969,
+ 542.2080078125,
+ 379.39202880859375,
+ 530.9440307617188
+ ],
+ [
+ 712.1920166015625,
+ 524.800048828125,
+ 686.592041015625,
+ 539.1360473632812,
+ 690.6880493164062,
+ 546.3040161132812,
+ 717.31201171875,
+ 531.968017578125
+ ],
+ [
+ 321.0240173339844,
+ 530.9440307617188,
+ 290.30401611328125,
+ 542.2080078125,
+ 293.3760070800781,
+ 549.3760375976562,
+ 324.09600830078125,
+ 539.1360473632812
+ ],
+ [
+ 752.1280517578125,
+ 530.9440307617188,
+ 745.9840087890625,
+ 538.1119995117188,
+ 770.56005859375,
+ 554.4960327148438,
+ 775.6800537109375,
+ 547.3280029296875
+ ],
+ [
+ 747.008056640625,
+ 538.1119995117188,
+ 740.864013671875,
+ 545.280029296875,
+ 767.488037109375,
+ 563.7120361328125,
+ 772.6080322265625,
+ 556.5440063476562
+ ],
+ [
+ 330.2400207519531,
+ 542.2080078125,
+ 299.52001953125,
+ 553.4720458984375,
+ 301.5680236816406,
+ 559.6160278320312,
+ 333.31201171875,
+ 549.3760375976562
+ ],
+ [
+ 738.8160400390625,
+ 546.3040161132812,
+ 732.6720581054688,
+ 553.4720458984375,
+ 759.2960205078125,
+ 572.9280395507812,
+ 765.4400634765625,
+ 564.7360229492188
+ ],
+ [
+ 730.6240234375,
+ 549.3760375976562,
+ 725.5040283203125,
+ 555.52001953125,
+ 752.1280517578125,
+ 577.0240478515625,
+ 758.2720336914062,
+ 569.8560180664062
+ ],
+ [
+ 723.4560546875,
+ 552.447998046875,
+ 717.31201171875,
+ 558.592041015625,
+ 744.9600219726562,
+ 580.0960083007812,
+ 750.0800170898438,
+ 572.9280395507812
+ ],
+ [
+ 231.93600463867188,
+ 558.592041015625,
+ 204.2880096435547,
+ 569.8560180664062,
+ 207.36001586914062,
+ 577.0240478515625,
+ 235.0080108642578,
+ 565.760009765625
+ ],
+ [
+ 991.7440185546875,
+ 600.5760498046875,
+ 984.5760498046875,
+ 608.7680053710938,
+ 1011.2000732421875,
+ 632.3200073242188,
+ 1018.3680419921875,
+ 624.1280517578125
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L539.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 321.0240173339844,
+ 342.52801513671875,
+ 315.9040222167969,
+ 381.4400329589844,
+ 332.28802490234375,
+ 383.4880065917969,
+ 337.40802001953125,
+ 345.6000061035156
+ ],
+ [
+ 554.4960327148438,
+ 513.5360107421875,
+ 548.35205078125,
+ 525.8240356445312,
+ 578.0480346679688,
+ 540.1600341796875,
+ 584.1920166015625,
+ 528.89599609375
+ ],
+ [
+ 543.2320556640625,
+ 532.9920043945312,
+ 537.0880126953125,
+ 544.2560424804688,
+ 565.760009765625,
+ 559.6160278320312,
+ 571.904052734375,
+ 548.35205078125
+ ],
+ [
+ 511.488037109375,
+ 568.83203125,
+ 506.3680114746094,
+ 579.072021484375,
+ 532.9920043945312,
+ 595.4560546875,
+ 538.1119995117188,
+ 585.2160034179688
+ ],
+ [
+ 498.176025390625,
+ 589.31201171875,
+ 492.0320129394531,
+ 600.5760498046875,
+ 518.656005859375,
+ 616.9600219726562,
+ 524.800048828125,
+ 604.6720581054688
+ ],
+ [
+ 484.864013671875,
+ 608.7680053710938,
+ 478.72003173828125,
+ 621.0560302734375,
+ 507.39202880859375,
+ 636.416015625,
+ 513.5360107421875,
+ 624.1280517578125
+ ],
+ [
+ 886.2720336914062,
+ 608.7680053710938,
+ 877.0560302734375,
+ 621.0560302734375,
+ 908.800048828125,
+ 642.56005859375,
+ 916.9920654296875,
+ 631.2960205078125
+ ],
+ [
+ 855.5520629882812,
+ 643.5840454101562,
+ 848.384033203125,
+ 655.8720092773438,
+ 881.1520385742188,
+ 678.4000244140625,
+ 889.3440551757812,
+ 667.1360473632812
+ ],
+ [
+ 847.3600463867188,
+ 656.8960571289062,
+ 840.1920166015625,
+ 668.1600341796875,
+ 871.93603515625,
+ 687.6160278320312,
+ 880.1280517578125,
+ 677.3760375976562
+ ],
+ [
+ 836.0960693359375,
+ 674.3040161132812,
+ 828.9280395507812,
+ 685.5680541992188,
+ 860.6720581054688,
+ 705.0240478515625,
+ 867.8400268554688,
+ 692.7360229492188
+ ],
+ [
+ 824.83203125,
+ 689.6640625,
+ 817.6640625,
+ 700.9280395507812,
+ 850.4320678710938,
+ 722.4320068359375,
+ 856.5760498046875,
+ 711.1680297851562
+ ],
+ [
+ 813.5680541992188,
+ 704.0000610351562,
+ 807.4240112304688,
+ 715.2640380859375,
+ 838.14404296875,
+ 734.7200317382812,
+ 845.31201171875,
+ 724.4800415039062
+ ],
+ [
+ 807.4240112304688,
+ 718.3360595703125,
+ 800.2560424804688,
+ 728.5760498046875,
+ 829.9520263671875,
+ 749.0560302734375,
+ 837.1200561523438,
+ 737.7920532226562
+ ],
+ [
+ 797.1840209960938,
+ 731.6480102539062,
+ 790.0160522460938,
+ 743.93603515625,
+ 819.7120361328125,
+ 764.416015625,
+ 827.904052734375,
+ 752.1280517578125
+ ],
+ [
+ 905.72802734375,
+ 858.6240234375,
+ 873.9840698242188,
+ 873.9840698242188,
+ 879.1040649414062,
+ 884.2240600585938,
+ 910.8480224609375,
+ 869.8880615234375
+ ],
+ [
+ 913.9200439453125,
+ 860.6720581054688,
+ 882.176025390625,
+ 876.0320434570312,
+ 886.2720336914062,
+ 886.2720336914062,
+ 919.0400390625,
+ 870.9120483398438
+ ],
+ [
+ 919.0400390625,
+ 869.8880615234375,
+ 887.2960205078125,
+ 885.248046875,
+ 891.3920288085938,
+ 894.4640502929688,
+ 923.1360473632812,
+ 880.1280517578125
+ ],
+ [
+ 863.7440185546875,
+ 872.9600219726562,
+ 830.9760131835938,
+ 888.320068359375,
+ 834.0480346679688,
+ 896.5120239257812,
+ 867.8400268554688,
+ 881.1520385742188
+ ],
+ [
+ 926.2080688476562,
+ 882.176025390625,
+ 893.4400634765625,
+ 895.488037109375,
+ 896.5120239257812,
+ 904.7040405273438,
+ 930.3040161132812,
+ 891.3920288085938
+ ],
+ [
+ 866.8160400390625,
+ 883.2000122070312,
+ 834.0480346679688,
+ 898.56005859375,
+ 838.14404296875,
+ 907.7760620117188,
+ 870.9120483398438,
+ 892.416015625
+ ],
+ [
+ 813.5680541992188,
+ 893.4400634765625,
+ 780.800048828125,
+ 905.72802734375,
+ 785.9200439453125,
+ 916.9920654296875,
+ 818.6880493164062,
+ 903.6800537109375
+ ],
+ [
+ 727.5520629882812,
+ 914.9440307617188,
+ 694.7840576171875,
+ 930.3040161132812,
+ 699.904052734375,
+ 940.5440673828125,
+ 731.6480102539062,
+ 925.1840209960938
+ ],
+ [
+ 680.4480590820312,
+ 930.3040161132812,
+ 649.72802734375,
+ 943.6160278320312,
+ 653.8240356445312,
+ 952.83203125,
+ 683.52001953125,
+ 939.52001953125
+ ],
+ [
+ 643.5840454101562,
+ 939.52001953125,
+ 610.8160400390625,
+ 954.8800659179688,
+ 613.8880004882812,
+ 964.0960693359375,
+ 646.656005859375,
+ 948.7360229492188
+ ],
+ [
+ 645.6320190429688,
+ 951.8080444335938,
+ 612.864013671875,
+ 966.14404296875,
+ 616.9600219726562,
+ 975.3600463867188,
+ 648.7040405273438,
+ 961.0240478515625
+ ],
+ [
+ 648.7040405273438,
+ 961.0240478515625,
+ 616.9600219726562,
+ 975.3600463867188,
+ 621.0560302734375,
+ 984.5760498046875,
+ 652.800048828125,
+ 971.2640380859375
+ ],
+ [
+ 654.8480224609375,
+ 969.216064453125,
+ 623.10400390625,
+ 982.528076171875,
+ 627.2000122070312,
+ 991.7440185546875,
+ 657.9200439453125,
+ 978.4320678710938
+ ],
+ [
+ 658.9440307617188,
+ 975.3600463867188,
+ 627.2000122070312,
+ 989.696044921875,
+ 631.2960205078125,
+ 999.93603515625,
+ 663.0400390625,
+ 984.5760498046875
+ ],
+ [
+ 540.1600341796875,
+ 980.4800415039062,
+ 508.416015625,
+ 994.8160400390625,
+ 511.488037109375,
+ 1005.0560302734375,
+ 545.280029296875,
+ 989.696044921875
+ ],
+ [
+ 664.0640258789062,
+ 982.528076171875,
+ 631.2960205078125,
+ 997.8880615234375,
+ 635.3920288085938,
+ 1008.1280517578125,
+ 668.1600341796875,
+ 992.76806640625
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ },
+ "L622.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 961.0240478515625,
+ 22.016000747680664,
+ 948.7360229492188,
+ 37.37600326538086,
+ 1009.1520385742188,
+ 85.5040054321289,
+ 1020.4160766601562,
+ 71.16800689697266
+ ],
+ [
+ 171.52000427246094,
+ 498.176025390625,
+ 93.69600677490234,
+ 526.8480224609375,
+ 103.93600463867188,
+ 554.4960327148438,
+ 181.760009765625,
+ 523.7760009765625
+ ],
+ [
+ 62.97600173950195,
+ 534.0160522460938,
+ 5.63200044631958,
+ 554.4960327148438,
+ 12.800000190734863,
+ 572.9280395507812,
+ 70.14400482177734,
+ 554.4960327148438
+ ],
+ [
+ 64.0,
+ 547.3280029296875,
+ 9.72800064086914,
+ 568.83203125,
+ 17.920000076293945,
+ 588.2880249023438,
+ 72.19200134277344,
+ 566.7839965820312
+ ],
+ [
+ 909.8240356445312,
+ 891.3920288085938,
+ 887.2960205078125,
+ 913.9200439453125,
+ 939.52001953125,
+ 965.1200561523438,
+ 962.0480346679688,
+ 941.5680541992188
+ ]
+ ],
+ [
+ [
+ 429.5680236816406,
+ 528.89599609375,
+ 414.2080078125,
+ 547.3280029296875,
+ 466.4320373535156,
+ 586.2400512695312,
+ 479.7440185546875,
+ 566.7839965820312
+ ],
+ [
+ 484.864013671875,
+ 570.8800048828125,
+ 469.5040283203125,
+ 588.2880249023438,
+ 511.488037109375,
+ 624.1280517578125,
+ 526.8480224609375,
+ 606.7200317382812
+ ],
+ [
+ 883.2000122070312,
+ 668.1600341796875,
+ 860.6720581054688,
+ 689.6640625,
+ 913.9200439453125,
+ 729.6000366210938,
+ 934.4000244140625,
+ 709.1200561523438
+ ]
+ ]
+ ],
+ "labels": [
+ "container",
+ "dredger"
+ ]
+ }
+ },
+ "L648.png": {
+ "": {
+ "polygons": [
+ [
+ [
+ 132.60800170898438,
+ 14.848000526428223,
+ 62.97600173950195,
+ 47.61600112915039,
+ 70.14400482177734,
+ 66.04800415039062,
+ 140.8000030517578,
+ 34.30400085449219
+ ],
+ [
+ 116.22400665283203,
+ 541.1840209960938,
+ 46.592002868652344,
+ 565.760009765625,
+ 51.71200180053711,
+ 586.2400512695312,
+ 122.36800384521484,
+ 559.6160278320312
+ ],
+ [
+ 585.2160034179688,
+ 554.4960327148438,
+ 584.1920166015625,
+ 571.904052734375,
+ 699.904052734375,
+ 586.2400512695312,
+ 700.9280395507812,
+ 568.83203125
+ ]
+ ]
+ ],
+ "labels": [
+ "container"
+ ]
+ }
+ }
+}
\ No newline at end of file
diff --git a/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_srsdd/parsed_answers_test/eval_liqingyun@SH-IDC1-10-140-37-115_device0_rank0.log b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_srsdd/parsed_answers_test/eval_liqingyun@SH-IDC1-10-140-37-115_device0_rank0.log
new file mode 100644
index 0000000000000000000000000000000000000000..dcb02d38ca6e5e2fd7b4490b7191daf5acb815c3
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/eval_srsdd/parsed_answers_test/eval_liqingyun@SH-IDC1-10-140-37-115_device0_rank0.log
@@ -0,0 +1,21 @@
+2025/01/25 00:22:09 - srsdd test Evaluation - INFO - --------------------
+2025/01/25 00:22:09 - srsdd test Evaluation - INFO - Fuzzy matched fishing boat boat boat to fishing
+2025/01/25 00:22:09 - srsdd test Evaluation - INFO - Fuzzy matched dredgergerger to dredger
+2025/01/25 00:22:09 - srsdd test Evaluation - INFO -
+---------------iou_thr: 0.5---------------
+2025/01/25 00:22:55 - srsdd test Evaluation - INFO -
++----------------+-----+------+--------+-------+
+| class | gts | dets | recall | ap |
++----------------+-----+------+--------+-------+
+| Cell-Container | 22 | 23 | 0.636 | 0.393 |
+| Container | 450 | 430 | 0.502 | 0.315 |
+| Dredger | 46 | 29 | 0.457 | 0.405 |
+| Fishing | 82 | 10 | 0.098 | 0.091 |
+| LawEnforce | 5 | 15 | 0.000 | 0.000 |
+| ore-oil | 34 | 21 | 0.382 | 0.275 |
++----------------+-----+------+--------+-------+
+| mAP | | | | 0.246 |
++----------------+-----+------+--------+-------+
+2025/01/25 00:22:55 - srsdd test Evaluation - INFO - --------------------
+
+2025/01/25 00:22:55 - srsdd test Evaluation - INFO - Dumped results to checkpoints/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2_2/eval_srsdd/parsed_answers_test/output.pkl
diff --git a/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/generation_config.json b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/generation_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..69b5fa5daf6edc982b3b5144a842b85fb917ee2d
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/generation_config.json
@@ -0,0 +1,5 @@
+{
+ "attn_implementation": "flash_attention_2",
+ "num_beams": 3,
+ "transformers_version": "4.45.0"
+}
diff --git a/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/special_tokens_map.json b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/special_tokens_map.json
new file mode 100644
index 0000000000000000000000000000000000000000..d748fe272a81974cb735e2d1d6821bb294566849
--- /dev/null
+++ b/florence-2-l_vis1024-lang2048_dota1-dior-fair1m1-srsdd1-rsar-v2-concat_b2x16-100e-slurm-zero2/special_tokens_map.json
@@ -0,0 +1,7185 @@
+{
+ "additional_special_tokens": [
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "",
+ "lstrip": false,
+ "normalized": false,
+ "rstrip": false,
+ "single_word": false
+ },
+ {
+ "content": "