diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,3905 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.999096657633243,
+  "eval_steps": 500,
+  "global_step": 553,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0018066847335140017,
+      "grad_norm": 2.8337757951637657,
+      "learning_rate": 0.0,
+      "loss": 0.8251,
+      "step": 1
+    },
+    {
+      "epoch": 0.0036133694670280035,
+      "grad_norm": 2.8815146894381005,
+      "learning_rate": 1.7857142857142858e-07,
+      "loss": 0.8284,
+      "step": 2
+    },
+    {
+      "epoch": 0.005420054200542005,
+      "grad_norm": 2.888366333138549,
+      "learning_rate": 3.5714285714285716e-07,
+      "loss": 0.8432,
+      "step": 3
+    },
+    {
+      "epoch": 0.007226738934056007,
+      "grad_norm": 2.8554157896663708,
+      "learning_rate": 5.357142857142857e-07,
+      "loss": 0.8398,
+      "step": 4
+    },
+    {
+      "epoch": 0.009033423667570008,
+      "grad_norm": 2.7996302090063954,
+      "learning_rate": 7.142857142857143e-07,
+      "loss": 0.8303,
+      "step": 5
+    },
+    {
+      "epoch": 0.01084010840108401,
+      "grad_norm": 2.758540000006854,
+      "learning_rate": 8.928571428571429e-07,
+      "loss": 0.8306,
+      "step": 6
+    },
+    {
+      "epoch": 0.012646793134598013,
+      "grad_norm": 2.7440712027687453,
+      "learning_rate": 1.0714285714285714e-06,
+      "loss": 0.8129,
+      "step": 7
+    },
+    {
+      "epoch": 0.014453477868112014,
+      "grad_norm": 2.789670841276354,
+      "learning_rate": 1.25e-06,
+      "loss": 0.8217,
+      "step": 8
+    },
+    {
+      "epoch": 0.016260162601626018,
+      "grad_norm": 2.637271873271262,
+      "learning_rate": 1.4285714285714286e-06,
+      "loss": 0.8192,
+      "step": 9
+    },
+    {
+      "epoch": 0.018066847335140017,
+      "grad_norm": 2.6274284830887247,
+      "learning_rate": 1.6071428571428574e-06,
+      "loss": 0.8146,
+      "step": 10
+    },
+    {
+      "epoch": 0.01987353206865402,
+      "grad_norm": 2.2614743320388992,
+      "learning_rate": 1.7857142857142859e-06,
+      "loss": 0.8022,
+      "step": 11
+    },
+    {
+      "epoch": 0.02168021680216802,
+      "grad_norm": 2.1218891806221882,
+      "learning_rate": 1.9642857142857144e-06,
+      "loss": 0.8131,
+      "step": 12
+    },
+    {
+      "epoch": 0.023486901535682024,
+      "grad_norm": 2.1362019297809636,
+      "learning_rate": 2.1428571428571427e-06,
+      "loss": 0.8047,
+      "step": 13
+    },
+    {
+      "epoch": 0.025293586269196026,
+      "grad_norm": 2.043181941163786,
+      "learning_rate": 2.321428571428572e-06,
+      "loss": 0.8006,
+      "step": 14
+    },
+    {
+      "epoch": 0.02710027100271003,
+      "grad_norm": 1.4654153096579285,
+      "learning_rate": 2.5e-06,
+      "loss": 0.7863,
+      "step": 15
+    },
+    {
+      "epoch": 0.028906955736224028,
+      "grad_norm": 1.424950643183691,
+      "learning_rate": 2.6785714285714285e-06,
+      "loss": 0.7794,
+      "step": 16
+    },
+    {
+      "epoch": 0.03071364046973803,
+      "grad_norm": 1.371800886806853,
+      "learning_rate": 2.8571428571428573e-06,
+      "loss": 0.7871,
+      "step": 17
+    },
+    {
+      "epoch": 0.032520325203252036,
+      "grad_norm": 1.2910993566284936,
+      "learning_rate": 3.0357142857142856e-06,
+      "loss": 0.7792,
+      "step": 18
+    },
+    {
+      "epoch": 0.03432700993676603,
+      "grad_norm": 1.2099658145135412,
+      "learning_rate": 3.2142857142857147e-06,
+      "loss": 0.7728,
+      "step": 19
+    },
+    {
+      "epoch": 0.036133694670280034,
+      "grad_norm": 1.0479474035735314,
+      "learning_rate": 3.3928571428571435e-06,
+      "loss": 0.743,
+      "step": 20
+    },
+    {
+      "epoch": 0.037940379403794036,
+      "grad_norm": 1.112276534792924,
+      "learning_rate": 3.5714285714285718e-06,
+      "loss": 0.7403,
+      "step": 21
+    },
+    {
+      "epoch": 0.03974706413730804,
+      "grad_norm": 1.3035077079579298,
+      "learning_rate": 3.7500000000000005e-06,
+      "loss": 0.7296,
+      "step": 22
+    },
+    {
+      "epoch": 0.04155374887082204,
+      "grad_norm": 1.3695661288387202,
+      "learning_rate": 3.928571428571429e-06,
+      "loss": 0.7302,
+      "step": 23
+    },
+    {
+      "epoch": 0.04336043360433604,
+      "grad_norm": 1.2631605637070566,
+      "learning_rate": 4.107142857142857e-06,
+      "loss": 0.7359,
+      "step": 24
+    },
+    {
+      "epoch": 0.045167118337850046,
+      "grad_norm": 0.9577414796019008,
+      "learning_rate": 4.2857142857142855e-06,
+      "loss": 0.713,
+      "step": 25
+    },
+    {
+      "epoch": 0.04697380307136405,
+      "grad_norm": 0.7548293526053346,
+      "learning_rate": 4.464285714285715e-06,
+      "loss": 0.7275,
+      "step": 26
+    },
+    {
+      "epoch": 0.04878048780487805,
+      "grad_norm": 0.6708116039946823,
+      "learning_rate": 4.642857142857144e-06,
+      "loss": 0.7021,
+      "step": 27
+    },
+    {
+      "epoch": 0.05058717253839205,
+      "grad_norm": 0.7960055478682195,
+      "learning_rate": 4.821428571428572e-06,
+      "loss": 0.6998,
+      "step": 28
+    },
+    {
+      "epoch": 0.052393857271906055,
+      "grad_norm": 0.8790784871024999,
+      "learning_rate": 5e-06,
+      "loss": 0.7086,
+      "step": 29
+    },
+    {
+      "epoch": 0.05420054200542006,
+      "grad_norm": 0.9579979234005123,
+      "learning_rate": 4.999955240022903e-06,
+      "loss": 0.7065,
+      "step": 30
+    },
+    {
+      "epoch": 0.05600722673893405,
+      "grad_norm": 1.003933265230055,
+      "learning_rate": 4.999820961694372e-06,
+      "loss": 0.6937,
+      "step": 31
+    },
+    {
+      "epoch": 0.057813911472448055,
+      "grad_norm": 0.6710805038763018,
+      "learning_rate": 4.999597169822646e-06,
+      "loss": 0.6812,
+      "step": 32
+    },
+    {
+      "epoch": 0.05962059620596206,
+      "grad_norm": 0.5001839964623817,
+      "learning_rate": 4.999283872421259e-06,
+      "loss": 0.6911,
+      "step": 33
+    },
+    {
+      "epoch": 0.06142728093947606,
+      "grad_norm": 0.5681434362593133,
+      "learning_rate": 4.998881080708759e-06,
+      "loss": 0.6766,
+      "step": 34
+    },
+    {
+      "epoch": 0.06323396567299007,
+      "grad_norm": 0.5752663504650961,
+      "learning_rate": 4.998388809108304e-06,
+      "loss": 0.6816,
+      "step": 35
+    },
+    {
+      "epoch": 0.06504065040650407,
+      "grad_norm": 0.4990120714898939,
+      "learning_rate": 4.997807075247147e-06,
+      "loss": 0.6754,
+      "step": 36
+    },
+    {
+      "epoch": 0.06684733514001806,
+      "grad_norm": 0.4974519304946596,
+      "learning_rate": 4.997135899956002e-06,
+      "loss": 0.6782,
+      "step": 37
+    },
+    {
+      "epoch": 0.06865401987353206,
+      "grad_norm": 0.4349874019011085,
+      "learning_rate": 4.996375307268303e-06,
+      "loss": 0.6709,
+      "step": 38
+    },
+    {
+      "epoch": 0.07046070460704607,
+      "grad_norm": 0.48312199739205014,
+      "learning_rate": 4.995525324419338e-06,
+      "loss": 0.6707,
+      "step": 39
+    },
+    {
+      "epoch": 0.07226738934056007,
+      "grad_norm": 0.3760455090466998,
+      "learning_rate": 4.994585981845278e-06,
+      "loss": 0.6592,
+      "step": 40
+    },
+    {
+      "epoch": 0.07407407407407407,
+      "grad_norm": 0.362711899175025,
+      "learning_rate": 4.993557313182086e-06,
+      "loss": 0.6666,
+      "step": 41
+    },
+    {
+      "epoch": 0.07588075880758807,
+      "grad_norm": 0.3527864788229306,
+      "learning_rate": 4.992439355264308e-06,
+      "loss": 0.656,
+      "step": 42
+    },
+    {
+      "epoch": 0.07768744354110207,
+      "grad_norm": 0.37039474310454246,
+      "learning_rate": 4.9912321481237616e-06,
+      "loss": 0.6567,
+      "step": 43
+    },
+    {
+      "epoch": 0.07949412827461608,
+      "grad_norm": 0.3381704985127657,
+      "learning_rate": 4.989935734988098e-06,
+      "loss": 0.6492,
+      "step": 44
+    },
+    {
+      "epoch": 0.08130081300813008,
+      "grad_norm": 0.3368799245864057,
+      "learning_rate": 4.988550162279255e-06,
+      "loss": 0.6558,
+      "step": 45
+    },
+    {
+      "epoch": 0.08310749774164408,
+      "grad_norm": 0.33217201284358455,
+      "learning_rate": 4.9870754796117956e-06,
+      "loss": 0.658,
+      "step": 46
+    },
+    {
+      "epoch": 0.08491418247515808,
+      "grad_norm": 0.3203999210132069,
+      "learning_rate": 4.985511739791129e-06,
+      "loss": 0.6596,
+      "step": 47
+    },
+    {
+      "epoch": 0.08672086720867209,
+      "grad_norm": 0.3108831280349791,
+      "learning_rate": 4.983858998811622e-06,
+      "loss": 0.6465,
+      "step": 48
+    },
+    {
+      "epoch": 0.08852755194218609,
+      "grad_norm": 0.28944557461245024,
+      "learning_rate": 4.982117315854594e-06,
+      "loss": 0.651,
+      "step": 49
+    },
+    {
+      "epoch": 0.09033423667570009,
+      "grad_norm": 0.2724500694923891,
+      "learning_rate": 4.980286753286196e-06,
+      "loss": 0.6493,
+      "step": 50
+    },
+    {
+      "epoch": 0.0921409214092141,
+      "grad_norm": 0.29233908216498683,
+      "learning_rate": 4.978367376655177e-06,
+      "loss": 0.6391,
+      "step": 51
+    },
+    {
+      "epoch": 0.0939476061427281,
+      "grad_norm": 0.29050327837368234,
+      "learning_rate": 4.976359254690543e-06,
+      "loss": 0.6577,
+      "step": 52
+    },
+    {
+      "epoch": 0.0957542908762421,
+      "grad_norm": 0.26277310012199956,
+      "learning_rate": 4.974262459299088e-06,
+      "loss": 0.641,
+      "step": 53
+    },
+    {
+      "epoch": 0.0975609756097561,
+      "grad_norm": 0.2694317681957786,
+      "learning_rate": 4.9720770655628216e-06,
+      "loss": 0.6322,
+      "step": 54
+    },
+    {
+      "epoch": 0.0993676603432701,
+      "grad_norm": 0.27643434259676186,
+      "learning_rate": 4.969803151736285e-06,
+      "loss": 0.6446,
+      "step": 55
+    },
+    {
+      "epoch": 0.1011743450767841,
+      "grad_norm": 0.27938135134555137,
+      "learning_rate": 4.967440799243739e-06,
+      "loss": 0.6505,
+      "step": 56
+    },
+    {
+      "epoch": 0.10298102981029811,
+      "grad_norm": 0.2723254558437683,
+      "learning_rate": 4.964990092676263e-06,
+      "loss": 0.6358,
+      "step": 57
+    },
+    {
+      "epoch": 0.10478771454381211,
+      "grad_norm": 0.2476211836425744,
+      "learning_rate": 4.962451119788709e-06,
+      "loss": 0.6435,
+      "step": 58
+    },
+    {
+      "epoch": 0.10659439927732611,
+      "grad_norm": 0.2385367032602417,
+      "learning_rate": 4.959823971496575e-06,
+      "loss": 0.6289,
+      "step": 59
+    },
+    {
+      "epoch": 0.10840108401084012,
+      "grad_norm": 0.24895994998658083,
+      "learning_rate": 4.957108741872736e-06,
+      "loss": 0.6299,
+      "step": 60
+    },
+    {
+      "epoch": 0.1102077687443541,
+      "grad_norm": 0.28129654057596637,
+      "learning_rate": 4.954305528144085e-06,
+      "loss": 0.6339,
+      "step": 61
+    },
+    {
+      "epoch": 0.1120144534778681,
+      "grad_norm": 0.25373425749505557,
+      "learning_rate": 4.9514144306880506e-06,
+      "loss": 0.6393,
+      "step": 62
+    },
+    {
+      "epoch": 0.11382113821138211,
+      "grad_norm": 0.2583299024885213,
+      "learning_rate": 4.948435553028994e-06,
+      "loss": 0.6383,
+      "step": 63
+    },
+    {
+      "epoch": 0.11562782294489611,
+      "grad_norm": 0.24423529044993955,
+      "learning_rate": 4.9453690018345144e-06,
+      "loss": 0.6359,
+      "step": 64
+    },
+    {
+      "epoch": 0.11743450767841011,
+      "grad_norm": 0.24605558051142096,
+      "learning_rate": 4.942214886911619e-06,
+      "loss": 0.631,
+      "step": 65
+    },
+    {
+      "epoch": 0.11924119241192412,
+      "grad_norm": 0.2548336350045181,
+      "learning_rate": 4.938973321202799e-06,
+      "loss": 0.6323,
+      "step": 66
+    },
+    {
+      "epoch": 0.12104787714543812,
+      "grad_norm": 0.32845268505362657,
+      "learning_rate": 4.935644420781978e-06,
+      "loss": 0.6229,
+      "step": 67
+    },
+    {
+      "epoch": 0.12285456187895212,
+      "grad_norm": 0.25203625601717533,
+      "learning_rate": 4.932228304850363e-06,
+      "loss": 0.6386,
+      "step": 68
+    },
+    {
+      "epoch": 0.12466124661246612,
+      "grad_norm": 0.2467239132093227,
+      "learning_rate": 4.9287250957321685e-06,
+      "loss": 0.6401,
+      "step": 69
+    },
+    {
+      "epoch": 0.12646793134598014,
+      "grad_norm": 0.28444822854929847,
+      "learning_rate": 4.925134918870245e-06,
+      "loss": 0.646,
+      "step": 70
+    },
+    {
+      "epoch": 0.12827461607949414,
+      "grad_norm": 0.23105312295557145,
+      "learning_rate": 4.921457902821578e-06,
+      "loss": 0.6164,
+      "step": 71
+    },
+    {
+      "epoch": 0.13008130081300814,
+      "grad_norm": 0.23048405537510025,
+      "learning_rate": 4.917694179252692e-06,
+      "loss": 0.6342,
+      "step": 72
+    },
+    {
+      "epoch": 0.13188798554652212,
+      "grad_norm": 0.24150602333720247,
+      "learning_rate": 4.9138438829349296e-06,
+      "loss": 0.6332,
+      "step": 73
+    },
+    {
+      "epoch": 0.13369467028003612,
+      "grad_norm": 0.2462886765781574,
+      "learning_rate": 4.909907151739634e-06,
+      "loss": 0.6246,
+      "step": 74
+    },
+    {
+      "epoch": 0.13550135501355012,
+      "grad_norm": 0.2821908512766841,
+      "learning_rate": 4.9058841266332005e-06,
+      "loss": 0.6227,
+      "step": 75
+    },
+    {
+      "epoch": 0.13730803974706413,
+      "grad_norm": 0.25172603846639535,
+      "learning_rate": 4.901774951672041e-06,
+      "loss": 0.6181,
+      "step": 76
+    },
+    {
+      "epoch": 0.13911472448057813,
+      "grad_norm": 0.23478283786140364,
+      "learning_rate": 4.897579773997415e-06,
+      "loss": 0.6123,
+      "step": 77
+    },
+    {
+      "epoch": 0.14092140921409213,
+      "grad_norm": 0.24524464756556244,
+      "learning_rate": 4.893298743830168e-06,
+      "loss": 0.625,
+      "step": 78
+    },
+    {
+      "epoch": 0.14272809394760613,
+      "grad_norm": 0.2557156779275681,
+      "learning_rate": 4.8889320144653525e-06,
+      "loss": 0.6347,
+      "step": 79
+    },
+    {
+      "epoch": 0.14453477868112014,
+      "grad_norm": 0.2501383477274707,
+      "learning_rate": 4.884479742266731e-06,
+      "loss": 0.6435,
+      "step": 80
+    },
+    {
+      "epoch": 0.14634146341463414,
+      "grad_norm": 0.2777905831842133,
+      "learning_rate": 4.879942086661185e-06,
+      "loss": 0.6249,
+      "step": 81
+    },
+    {
+      "epoch": 0.14814814814814814,
+      "grad_norm": 0.2311051834333071,
+      "learning_rate": 4.875319210133004e-06,
+      "loss": 0.6186,
+      "step": 82
+    },
+    {
+      "epoch": 0.14995483288166214,
+      "grad_norm": 0.24510662627866567,
+      "learning_rate": 4.870611278218066e-06,
+      "loss": 0.6301,
+      "step": 83
+    },
+    {
+      "epoch": 0.15176151761517614,
+      "grad_norm": 0.2457506431539522,
+      "learning_rate": 4.865818459497911e-06,
+      "loss": 0.6221,
+      "step": 84
+    },
+    {
+      "epoch": 0.15356820234869015,
+      "grad_norm": 0.23750506338642635,
+      "learning_rate": 4.860940925593703e-06,
+      "loss": 0.6238,
+      "step": 85
+    },
+    {
+      "epoch": 0.15537488708220415,
+      "grad_norm": 0.24583995522041222,
+      "learning_rate": 4.855978851160088e-06,
+      "loss": 0.6245,
+      "step": 86
+    },
+    {
+      "epoch": 0.15718157181571815,
+      "grad_norm": 0.24412234277510458,
+      "learning_rate": 4.850932413878934e-06,
+      "loss": 0.6266,
+      "step": 87
+    },
+    {
+      "epoch": 0.15898825654923215,
+      "grad_norm": 0.23692543465028104,
+      "learning_rate": 4.845801794452978e-06,
+      "loss": 0.6307,
+      "step": 88
+    },
+    {
+      "epoch": 0.16079494128274616,
+      "grad_norm": 0.24055583410087153,
+      "learning_rate": 4.8405871765993435e-06,
+      "loss": 0.6258,
+      "step": 89
+    },
+    {
+      "epoch": 0.16260162601626016,
+      "grad_norm": 0.25682681695753845,
+      "learning_rate": 4.8352887470429726e-06,
+      "loss": 0.6193,
+      "step": 90
+    },
+    {
+      "epoch": 0.16440831074977416,
+      "grad_norm": 0.2451987238773762,
+      "learning_rate": 4.829906695509934e-06,
+      "loss": 0.6175,
+      "step": 91
+    },
+    {
+      "epoch": 0.16621499548328816,
+      "grad_norm": 0.33308910526145086,
+      "learning_rate": 4.824441214720629e-06,
+      "loss": 0.6207,
+      "step": 92
+    },
+    {
+      "epoch": 0.16802168021680217,
+      "grad_norm": 0.2599636527935745,
+      "learning_rate": 4.8188925003828945e-06,
+      "loss": 0.6275,
+      "step": 93
+    },
+    {
+      "epoch": 0.16982836495031617,
+      "grad_norm": 0.24262017713570547,
+      "learning_rate": 4.813260751184992e-06,
+      "loss": 0.6255,
+      "step": 94
+    },
+    {
+      "epoch": 0.17163504968383017,
+      "grad_norm": 0.24155086390799096,
+      "learning_rate": 4.8075461687884935e-06,
+      "loss": 0.6184,
+      "step": 95
+    },
+    {
+      "epoch": 0.17344173441734417,
+      "grad_norm": 0.2399226609650259,
+      "learning_rate": 4.801748957821061e-06,
+      "loss": 0.6164,
+      "step": 96
+    },
+    {
+      "epoch": 0.17524841915085818,
+      "grad_norm": 0.22903596103655702,
+      "learning_rate": 4.795869325869117e-06,
+      "loss": 0.6229,
+      "step": 97
+    },
+    {
+      "epoch": 0.17705510388437218,
+      "grad_norm": 0.2307755782529314,
+      "learning_rate": 4.7899074834704165e-06,
+      "loss": 0.6173,
+      "step": 98
+    },
+    {
+      "epoch": 0.17886178861788618,
+      "grad_norm": 0.24823755288051938,
+      "learning_rate": 4.783863644106502e-06,
+      "loss": 0.6292,
+      "step": 99
+    },
+    {
+      "epoch": 0.18066847335140018,
+      "grad_norm": 0.21658543875130712,
+      "learning_rate": 4.777738024195065e-06,
+      "loss": 0.6032,
+      "step": 100
+    },
+    {
+      "epoch": 0.18247515808491419,
+      "grad_norm": 0.24020742361139732,
+      "learning_rate": 4.771530843082187e-06,
+      "loss": 0.6184,
+      "step": 101
+    },
+    {
+      "epoch": 0.1842818428184282,
+      "grad_norm": 0.23250154874953444,
+      "learning_rate": 4.765242323034498e-06,
+      "loss": 0.6373,
+      "step": 102
+    },
+    {
+      "epoch": 0.1860885275519422,
+      "grad_norm": 0.2515794157481243,
+      "learning_rate": 4.7588726892312085e-06,
+      "loss": 0.6162,
+      "step": 103
+    },
+    {
+      "epoch": 0.1878952122854562,
+      "grad_norm": 0.2388452484683859,
+      "learning_rate": 4.752422169756048e-06,
+      "loss": 0.6147,
+      "step": 104
+    },
+    {
+      "epoch": 0.1897018970189702,
+      "grad_norm": 0.23451396083233195,
+      "learning_rate": 4.7458909955891015e-06,
+      "loss": 0.6066,
+      "step": 105
+    },
+    {
+      "epoch": 0.1915085817524842,
+      "grad_norm": 0.23211345055411792,
+      "learning_rate": 4.7392794005985324e-06,
+      "loss": 0.6076,
+      "step": 106
+    },
+    {
+      "epoch": 0.1933152664859982,
+      "grad_norm": 0.2405071836183994,
+      "learning_rate": 4.732587621532214e-06,
+      "loss": 0.6139,
+      "step": 107
+    },
+    {
+      "epoch": 0.1951219512195122,
+      "grad_norm": 0.238214343050957,
+      "learning_rate": 4.7258158980092475e-06,
+      "loss": 0.6125,
+      "step": 108
+    },
+    {
+      "epoch": 0.1969286359530262,
+      "grad_norm": 0.23118377153943317,
+      "learning_rate": 4.718964472511386e-06,
+      "loss": 0.6224,
+      "step": 109
+    },
+    {
+      "epoch": 0.1987353206865402,
+      "grad_norm": 0.4227293348441868,
+      "learning_rate": 4.712033590374346e-06,
+      "loss": 0.604,
+      "step": 110
+    },
+    {
+      "epoch": 0.2005420054200542,
+      "grad_norm": 0.22706445039114057,
+      "learning_rate": 4.705023499779031e-06,
+      "loss": 0.6231,
+      "step": 111
+    },
+    {
+      "epoch": 0.2023486901535682,
+      "grad_norm": 0.21914294121010527,
+      "learning_rate": 4.6979344517426345e-06,
+      "loss": 0.6157,
+      "step": 112
+    },
+    {
+      "epoch": 0.2041553748870822,
+      "grad_norm": 0.2160022263814738,
+      "learning_rate": 4.690766700109659e-06,
+      "loss": 0.6193,
+      "step": 113
+    },
+    {
+      "epoch": 0.20596205962059622,
+      "grad_norm": 0.2304707253911858,
+      "learning_rate": 4.683520501542825e-06,
+      "loss": 0.6021,
+      "step": 114
+    },
+    {
+      "epoch": 0.20776874435411022,
+      "grad_norm": 0.24127382191408644,
+      "learning_rate": 4.676196115513876e-06,
+      "loss": 0.6186,
+      "step": 115
+    },
+    {
+      "epoch": 0.20957542908762422,
+      "grad_norm": 0.2406722048256909,
+      "learning_rate": 4.668793804294294e-06,
+      "loss": 0.6129,
+      "step": 116
+    },
+    {
+      "epoch": 0.21138211382113822,
+      "grad_norm": 0.24481652156673936,
+      "learning_rate": 4.661313832945904e-06,
+      "loss": 0.6228,
+      "step": 117
+    },
+    {
+      "epoch": 0.21318879855465223,
+      "grad_norm": 0.2283120728906989,
+      "learning_rate": 4.653756469311381e-06,
+      "loss": 0.6104,
+      "step": 118
+    },
+    {
+      "epoch": 0.21499548328816623,
+      "grad_norm": 0.2191260930120605,
+      "learning_rate": 4.646121984004666e-06,
+      "loss": 0.6015,
+      "step": 119
+    },
+    {
+      "epoch": 0.21680216802168023,
+      "grad_norm": 0.24306778049828862,
+      "learning_rate": 4.638410650401267e-06,
+      "loss": 0.6193,
+      "step": 120
+    },
+    {
+      "epoch": 0.2186088527551942,
+      "grad_norm": 0.2269775698273598,
+      "learning_rate": 4.630622744628478e-06,
+      "loss": 0.6157,
+      "step": 121
+    },
+    {
+      "epoch": 0.2204155374887082,
+      "grad_norm": 0.22557834203825844,
+      "learning_rate": 4.622758545555485e-06,
+      "loss": 0.6205,
+      "step": 122
+    },
+    {
+      "epoch": 0.2222222222222222,
+      "grad_norm": 0.2388627744876031,
+      "learning_rate": 4.614818334783384e-06,
+      "loss": 0.6117,
+      "step": 123
+    },
+    {
+      "epoch": 0.2240289069557362,
+      "grad_norm": 0.33690846185965356,
+      "learning_rate": 4.606802396635098e-06,
+      "loss": 0.5973,
+      "step": 124
+    },
+    {
+      "epoch": 0.22583559168925021,
+      "grad_norm": 0.2559933256711994,
+      "learning_rate": 4.598711018145193e-06,
+      "loss": 0.6035,
+      "step": 125
+    },
+    {
+      "epoch": 0.22764227642276422,
+      "grad_norm": 0.23041365241908557,
+      "learning_rate": 4.590544489049602e-06,
+      "loss": 0.6167,
+      "step": 126
+    },
+    {
+      "epoch": 0.22944896115627822,
+      "grad_norm": 0.2386416733431471,
+      "learning_rate": 4.582303101775249e-06,
+      "loss": 0.6081,
+      "step": 127
+    },
+    {
+      "epoch": 0.23125564588979222,
+      "grad_norm": 0.2253681313258335,
+      "learning_rate": 4.573987151429579e-06,
+      "loss": 0.6075,
+      "step": 128
+    },
+    {
+      "epoch": 0.23306233062330622,
+      "grad_norm": 0.21952470403470545,
+      "learning_rate": 4.565596935789987e-06,
+      "loss": 0.6031,
+      "step": 129
+    },
+    {
+      "epoch": 0.23486901535682023,
+      "grad_norm": 0.24331315011786012,
+      "learning_rate": 4.557132755293164e-06,
+      "loss": 0.6219,
+      "step": 130
+    },
+    {
+      "epoch": 0.23667570009033423,
+      "grad_norm": 0.24835834109503124,
+      "learning_rate": 4.54859491302433e-06,
+      "loss": 0.6154,
+      "step": 131
+    },
+    {
+      "epoch": 0.23848238482384823,
+      "grad_norm": 0.22732916433505923,
+      "learning_rate": 4.539983714706383e-06,
+      "loss": 0.5945,
+      "step": 132
+    },
+    {
+      "epoch": 0.24028906955736223,
+      "grad_norm": 0.2181618946471109,
+      "learning_rate": 4.531299468688956e-06,
+      "loss": 0.61,
+      "step": 133
+    },
+    {
+      "epoch": 0.24209575429087624,
+      "grad_norm": 0.2259645647372355,
+      "learning_rate": 4.522542485937369e-06,
+      "loss": 0.6105,
+      "step": 134
+    },
+    {
+      "epoch": 0.24390243902439024,
+      "grad_norm": 0.2176244229453355,
+      "learning_rate": 4.5137130800215025e-06,
+      "loss": 0.6069,
+      "step": 135
+    },
+    {
+      "epoch": 0.24570912375790424,
+      "grad_norm": 0.2552195449531324,
+      "learning_rate": 4.50481156710456e-06,
+      "loss": 0.6113,
+      "step": 136
+    },
+    {
+      "epoch": 0.24751580849141824,
+      "grad_norm": 0.23421861438168898,
+      "learning_rate": 4.495838265931754e-06,
+      "loss": 0.5952,
+      "step": 137
+    },
+    {
+      "epoch": 0.24932249322493225,
+      "grad_norm": 0.21810921222578047,
+      "learning_rate": 4.486793497818889e-06,
+      "loss": 0.6121,
+      "step": 138
+    },
+    {
+      "epoch": 0.25112917795844625,
+      "grad_norm": 0.22773114029207175,
+      "learning_rate": 4.477677586640854e-06,
+      "loss": 0.6156,
+      "step": 139
+    },
+    {
+      "epoch": 0.2529358626919603,
+      "grad_norm": 0.23705077325515508,
+      "learning_rate": 4.4684908588200305e-06,
+      "loss": 0.6156,
+      "step": 140
+    },
+    {
+      "epoch": 0.25474254742547425,
+      "grad_norm": 0.24780177201776643,
+      "learning_rate": 4.4592336433146e-06,
+      "loss": 0.5953,
+      "step": 141
+    },
+    {
+      "epoch": 0.2565492321589883,
+      "grad_norm": 0.22862755103918672,
+      "learning_rate": 4.449906271606766e-06,
+      "loss": 0.616,
+      "step": 142
+    },
+    {
+      "epoch": 0.25835591689250226,
+      "grad_norm": 0.2404243241200089,
+      "learning_rate": 4.440509077690883e-06,
+      "loss": 0.607,
+      "step": 143
+    },
+    {
+      "epoch": 0.2601626016260163,
+      "grad_norm": 0.22293770252772532,
+      "learning_rate": 4.431042398061499e-06,
+      "loss": 0.5965,
+      "step": 144
+    },
+    {
+      "epoch": 0.26196928635953026,
+      "grad_norm": 0.20972193775216136,
+      "learning_rate": 4.421506571701305e-06,
+      "loss": 0.6093,
+      "step": 145
+    },
+    {
+      "epoch": 0.26377597109304424,
+      "grad_norm": 0.23518600200676912,
+      "learning_rate": 4.411901940068997e-06,
+      "loss": 0.6112,
+      "step": 146
+    },
+    {
+      "epoch": 0.26558265582655827,
+      "grad_norm": 0.22746108225760642,
+      "learning_rate": 4.402228847087046e-06,
+      "loss": 0.6037,
+      "step": 147
+    },
+    {
+      "epoch": 0.26738934056007224,
+      "grad_norm": 0.24903963512865804,
+      "learning_rate": 4.3924876391293915e-06,
+      "loss": 0.6048,
+      "step": 148
+    },
+    {
+      "epoch": 0.26919602529358627,
+      "grad_norm": 0.23472947723463866,
+      "learning_rate": 4.382678665009028e-06,
+      "loss": 0.6016,
+      "step": 149
+    },
+    {
+      "epoch": 0.27100271002710025,
+      "grad_norm": 0.23431207097830484,
+      "learning_rate": 4.372802275965521e-06,
+      "loss": 0.61,
+      "step": 150
+    },
+    {
+      "epoch": 0.2728093947606143,
+      "grad_norm": 0.22422800322111083,
+      "learning_rate": 4.362858825652428e-06,
+      "loss": 0.5949,
+      "step": 151
+    },
+    {
+      "epoch": 0.27461607949412825,
+      "grad_norm": 0.23561711836643753,
+      "learning_rate": 4.352848670124637e-06,
+      "loss": 0.5996,
+      "step": 152
+    },
+    {
+      "epoch": 0.2764227642276423,
+      "grad_norm": 0.21652376026883982,
+      "learning_rate": 4.342772167825613e-06,
+      "loss": 0.6009,
+      "step": 153
+    },
+    {
+      "epoch": 0.27822944896115626,
+      "grad_norm": 0.22742532543566119,
+      "learning_rate": 4.332629679574566e-06,
+      "loss": 0.6063,
+      "step": 154
+    },
+    {
+      "epoch": 0.2800361336946703,
+      "grad_norm": 0.3993451377830147,
+      "learning_rate": 4.322421568553529e-06,
+      "loss": 0.5961,
+      "step": 155
+    },
+    {
+      "epoch": 0.28184281842818426,
+      "grad_norm": 0.21468028516819,
+      "learning_rate": 4.312148200294355e-06,
+      "loss": 0.598,
+      "step": 156
+    },
+    {
+      "epoch": 0.2836495031616983,
+      "grad_norm": 0.2790705925726011,
+      "learning_rate": 4.3018099426656255e-06,
+      "loss": 0.6108,
+      "step": 157
+    },
+    {
+      "epoch": 0.28545618789521227,
+      "grad_norm": 0.22328718703580291,
+      "learning_rate": 4.291407165859481e-06,
+      "loss": 0.6073,
+      "step": 158
+    },
+    {
+      "epoch": 0.2872628726287263,
+      "grad_norm": 0.24079715236297133,
+      "learning_rate": 4.280940242378363e-06,
+      "loss": 0.5892,
+      "step": 159
+    },
+    {
+      "epoch": 0.28906955736224027,
+      "grad_norm": 0.3349816071529816,
+      "learning_rate": 4.2704095470216745e-06,
+      "loss": 0.5985,
+      "step": 160
+    },
+    {
+      "epoch": 0.2908762420957543,
+      "grad_norm": 0.2351483485898414,
+      "learning_rate": 4.259815456872363e-06,
+      "loss": 0.6041,
+      "step": 161
+    },
+    {
+      "epoch": 0.2926829268292683,
+      "grad_norm": 0.23385513145667092,
+      "learning_rate": 4.249158351283414e-06,
+      "loss": 0.5903,
+      "step": 162
+    },
+    {
+      "epoch": 0.2944896115627823,
+      "grad_norm": 0.22962794810239154,
+      "learning_rate": 4.2384386118642696e-06,
+      "loss": 0.6006,
+      "step": 163
+    },
+    {
+      "epoch": 0.2962962962962963,
+      "grad_norm": 0.227041646781334,
+      "learning_rate": 4.227656622467162e-06,
+      "loss": 0.609,
+      "step": 164
+    },
+    {
+      "epoch": 0.2981029810298103,
+      "grad_norm": 0.23339515173861125,
+      "learning_rate": 4.216812769173371e-06,
+      "loss": 0.6008,
+      "step": 165
+    },
+    {
+      "epoch": 0.2999096657633243,
+      "grad_norm": 0.2279892093631049,
+      "learning_rate": 4.205907440279395e-06,
+      "loss": 0.6067,
+      "step": 166
+    },
+    {
+      "epoch": 0.3017163504968383,
+      "grad_norm": 0.24349945347421037,
+      "learning_rate": 4.194941026283053e-06,
+      "loss": 0.6064,
+      "step": 167
+    },
+    {
+      "epoch": 0.3035230352303523,
+      "grad_norm": 0.2191845217404836,
+      "learning_rate": 4.183913919869495e-06,
+      "loss": 0.6039,
+      "step": 168
+    },
+    {
+      "epoch": 0.3053297199638663,
+      "grad_norm": 0.2180478057517194,
+      "learning_rate": 4.172826515897146e-06,
+      "loss": 0.597,
+      "step": 169
+    },
+    {
+      "epoch": 0.3071364046973803,
+      "grad_norm": 0.22148092498016186,
+      "learning_rate": 4.161679211383565e-06,
+      "loss": 0.6116,
+      "step": 170
+    },
+    {
+      "epoch": 0.3089430894308943,
+      "grad_norm": 0.21976711391650572,
+      "learning_rate": 4.150472405491226e-06,
+      "loss": 0.5937,
+      "step": 171
+    },
+    {
+      "epoch": 0.3107497741644083,
+      "grad_norm": 0.2359272043525047,
+      "learning_rate": 4.139206499513231e-06,
+      "loss": 0.6002,
+      "step": 172
+    },
+    {
+      "epoch": 0.31255645889792233,
+      "grad_norm": 0.20743057651505534,
+      "learning_rate": 4.127881896858934e-06,
+      "loss": 0.6036,
+      "step": 173
+    },
+    {
+      "epoch": 0.3143631436314363,
+      "grad_norm": 0.2185478761510027,
+      "learning_rate": 4.116499003039499e-06,
+      "loss": 0.5947,
+      "step": 174
+    },
+    {
+      "epoch": 0.31616982836495033,
+      "grad_norm": 0.24479063305435536,
+      "learning_rate": 4.105058225653381e-06,
+      "loss": 0.6,
+      "step": 175
+    },
+    {
+      "epoch": 0.3179765130984643,
+      "grad_norm": 0.22541665254858562,
+      "learning_rate": 4.093559974371725e-06,
+      "loss": 0.6137,
+      "step": 176
+    },
+    {
+      "epoch": 0.31978319783197834,
+      "grad_norm": 0.23354779603092837,
+      "learning_rate": 4.0820046609237026e-06,
+      "loss": 0.5881,
+      "step": 177
+    },
+    {
+      "epoch": 0.3215898825654923,
+      "grad_norm": 0.22004453141007924,
+      "learning_rate": 4.070392699081767e-06,
+      "loss": 0.5948,
+      "step": 178
+    },
+    {
+      "epoch": 0.32339656729900634,
+      "grad_norm": 0.22301964541218125,
+      "learning_rate": 4.058724504646834e-06,
+      "loss": 0.6054,
+      "step": 179
+    },
+    {
+      "epoch": 0.3252032520325203,
+      "grad_norm": 0.23125154097212539,
+      "learning_rate": 4.047000495433397e-06,
+      "loss": 0.6135,
+      "step": 180
+    },
+    {
+      "epoch": 0.32700993676603435,
+      "grad_norm": 0.2194814968746533,
+      "learning_rate": 4.035221091254563e-06,
+      "loss": 0.5902,
+      "step": 181
+    },
+    {
+      "epoch": 0.3288166214995483,
+      "grad_norm": 0.22257729627885203,
+      "learning_rate": 4.023386713907021e-06,
+      "loss": 0.6005,
+      "step": 182
+    },
+    {
+      "epoch": 0.33062330623306235,
+      "grad_norm": 0.22264840616047507,
+      "learning_rate": 4.011497787155938e-06,
+      "loss": 0.6123,
+      "step": 183
+    },
+    {
+      "epoch": 0.3324299909665763,
+      "grad_norm": 0.22891861604269678,
+      "learning_rate": 3.999554736719785e-06,
+      "loss": 0.6093,
+      "step": 184
+    },
+    {
+      "epoch": 0.33423667570009036,
+      "grad_norm": 0.2368591129219394,
+      "learning_rate": 3.987557990255093e-06,
+      "loss": 0.6054,
+      "step": 185
+    },
+    {
+      "epoch": 0.33604336043360433,
+      "grad_norm": 0.2483359395611164,
+      "learning_rate": 3.975507977341141e-06,
+      "loss": 0.6011,
+      "step": 186
+    },
+    {
+      "epoch": 0.33785004516711836,
+      "grad_norm": 0.20257786753801169,
+      "learning_rate": 3.963405129464569e-06,
+      "loss": 0.5983,
+      "step": 187
+    },
+    {
+      "epoch": 0.33965672990063234,
+      "grad_norm": 0.2186759693689697,
+      "learning_rate": 3.951249880003934e-06,
+      "loss": 0.5973,
+      "step": 188
+    },
+    {
+      "epoch": 0.34146341463414637,
+      "grad_norm": 0.21876513682058207,
+      "learning_rate": 3.939042664214185e-06,
+      "loss": 0.6052,
+      "step": 189
+    },
+    {
+      "epoch": 0.34327009936766034,
+      "grad_norm": 0.2099916515113305,
+      "learning_rate": 3.92678391921108e-06,
+      "loss": 0.5936,
+      "step": 190
+    },
+    {
+      "epoch": 0.34507678410117437,
+      "grad_norm": 0.2343120602972887,
+      "learning_rate": 3.914474083955537e-06,
+      "loss": 0.592,
+      "step": 191
+    },
+    {
+      "epoch": 0.34688346883468835,
+      "grad_norm": 0.20255138476243217,
+      "learning_rate": 3.902113599237911e-06,
+      "loss": 0.5986,
+      "step": 192
+    },
+    {
+      "epoch": 0.3486901535682023,
+      "grad_norm": 0.20455659875517468,
+      "learning_rate": 3.889702907662212e-06,
+      "loss": 0.6009,
+      "step": 193
+    },
+    {
+      "epoch": 0.35049683830171635,
+      "grad_norm": 0.22151992117493785,
+      "learning_rate": 3.8772424536302565e-06,
+      "loss": 0.6022,
+      "step": 194
+    },
+    {
+      "epoch": 0.3523035230352303,
+      "grad_norm": 0.23833014531149102,
+      "learning_rate": 3.864732683325755e-06,
+      "loss": 0.5856,
+      "step": 195
+    },
+    {
+      "epoch": 0.35411020776874436,
+      "grad_norm": 0.2139415037008631,
+      "learning_rate": 3.852174044698333e-06,
+      "loss": 0.5989,
+      "step": 196
+    },
+    {
+      "epoch": 0.35591689250225833,
+      "grad_norm": 0.2194749540116677,
+      "learning_rate": 3.839566987447492e-06,
+      "loss": 0.598,
+      "step": 197
+    },
+    {
+      "epoch": 0.35772357723577236,
+      "grad_norm": 0.2381302592790127,
+      "learning_rate": 3.826911963006508e-06,
+      "loss": 0.5752,
+      "step": 198
+    },
+    {
+      "epoch": 0.35953026196928634,
+      "grad_norm": 0.23075119894415708,
+      "learning_rate": 3.8142094245262617e-06,
+      "loss": 0.5978,
+      "step": 199
+    },
+    {
+      "epoch": 0.36133694670280037,
+      "grad_norm": 0.23286158873769425,
+      "learning_rate": 3.801459826859022e-06,
+      "loss": 0.5877,
+      "step": 200
+    },
+    {
+      "epoch": 0.36314363143631434,
+      "grad_norm": 0.22495213943890388,
+      "learning_rate": 3.788663626542146e-06,
+      "loss": 0.5822,
+      "step": 201
+    },
+    {
+      "epoch": 0.36495031616982837,
+      "grad_norm": 0.22147751444154135,
+      "learning_rate": 3.7758212817817406e-06,
+      "loss": 0.6091,
+      "step": 202
+    },
+    {
+      "epoch": 0.36675700090334235,
+      "grad_norm": 0.23184305404944502,
+      "learning_rate": 3.7629332524362532e-06,
+      "loss": 0.5962,
+      "step": 203
+    },
+    {
+      "epoch": 0.3685636856368564,
+      "grad_norm": 0.23109985129728863,
+      "learning_rate": 3.7500000000000005e-06,
+      "loss": 0.6088,
+      "step": 204
+    },
+    {
+      "epoch": 0.37037037037037035,
+      "grad_norm": 0.22131095632923808,
+      "learning_rate": 3.7370219875866497e-06,
+      "loss": 0.5985,
+      "step": 205
+    },
+    {
+      "epoch": 0.3721770551038844,
+      "grad_norm": 0.22292160877295167,
+      "learning_rate": 3.7239996799126315e-06,
+      "loss": 0.5862,
+      "step": 206
+    },
+    {
+      "epoch": 0.37398373983739835,
+      "grad_norm": 0.23049662951148336,
+      "learning_rate": 3.7109335432805006e-06,
+      "loss": 0.5943,
+      "step": 207
+    },
+    {
+      "epoch": 0.3757904245709124,
+      "grad_norm": 0.22653281643581213,
+      "learning_rate": 3.697824045562238e-06,
+      "loss": 0.5807,
+      "step": 208
+    },
+    {
+      "epoch": 0.37759710930442636,
+      "grad_norm": 0.2261566142221293,
+      "learning_rate": 3.684671656182497e-06,
+      "loss": 0.6027,
+      "step": 209
+    },
+    {
+      "epoch": 0.3794037940379404,
+      "grad_norm": 0.20736585757802733,
+      "learning_rate": 3.671476846101797e-06,
+      "loss": 0.5975,
+      "step": 210
+    },
+    {
+      "epoch": 0.38121047877145436,
+      "grad_norm": 0.26687108119675534,
+      "learning_rate": 3.658240087799655e-06,
+      "loss": 0.5995,
+      "step": 211
+    },
+    {
+      "epoch": 0.3830171635049684,
+      "grad_norm": 0.2701090110685354,
+      "learning_rate": 3.644961855257669e-06,
+      "loss": 0.608,
+      "step": 212
+    },
+    {
+      "epoch": 0.38482384823848237,
+      "grad_norm": 0.23937840184912948,
+      "learning_rate": 3.6316426239425484e-06,
+      "loss": 0.6063,
+      "step": 213
+    },
+    {
+      "epoch": 0.3866305329719964,
+      "grad_norm": 0.26506285623579445,
+      "learning_rate": 3.6182828707890816e-06,
+      "loss": 0.5925,
+      "step": 214
+    },
+    {
+      "epoch": 0.3884372177055104,
+      "grad_norm": 0.23169437753599226,
+      "learning_rate": 3.6048830741830678e-06,
+      "loss": 0.6193,
+      "step": 215
+    },
+    {
+      "epoch": 0.3902439024390244,
+      "grad_norm": 0.24107575514470772,
+      "learning_rate": 3.5914437139441754e-06,
+      "loss": 0.6084,
+      "step": 216
+    },
+    {
+      "epoch": 0.3920505871725384,
+      "grad_norm": 0.24497545746975022,
+      "learning_rate": 3.5779652713087717e-06,
+      "loss": 0.6081,
+      "step": 217
+    },
+    {
+      "epoch": 0.3938572719060524,
+      "grad_norm": 0.22341956530539367,
+      "learning_rate": 3.564448228912682e-06,
+      "loss": 0.6001,
+      "step": 218
+    },
+    {
+      "epoch": 0.3956639566395664,
+      "grad_norm": 0.23686461833528502,
+      "learning_rate": 3.5508930707739143e-06,
+      "loss": 0.5942,
+      "step": 219
+    },
+    {
+      "epoch": 0.3974706413730804,
+      "grad_norm": 0.22220451962583762,
+      "learning_rate": 3.5373002822753217e-06,
+      "loss": 0.5982,
+      "step": 220
+    },
+    {
+      "epoch": 0.3992773261065944,
+      "grad_norm": 0.20991512264600337,
+      "learning_rate": 3.523670350147227e-06,
+      "loss": 0.586,
+      "step": 221
+    },
+    {
+      "epoch": 0.4010840108401084,
+      "grad_norm": 0.2597272978289921,
+      "learning_rate": 3.510003762449988e-06,
+      "loss": 0.6031,
+      "step": 222
+    },
+    {
+      "epoch": 0.4028906955736224,
+      "grad_norm": 0.23319091020778485,
+      "learning_rate": 3.496301008556529e-06,
+      "loss": 0.5861,
+      "step": 223
+    },
+    {
+      "epoch": 0.4046973803071364,
+      "grad_norm": 0.21740514214818824,
+      "learning_rate": 3.4825625791348093e-06,
+      "loss": 0.5917,
+      "step": 224
+    },
+    {
+      "epoch": 0.4065040650406504,
+      "grad_norm": 0.22499201448021908,
+      "learning_rate": 3.4687889661302577e-06,
+      "loss": 0.5951,
+      "step": 225
+    },
+    {
+      "epoch": 0.4083107497741644,
+      "grad_norm": 0.21928333843378897,
+      "learning_rate": 3.454980662748156e-06,
+      "loss": 0.6022,
+      "step": 226
+    },
+    {
+      "epoch": 0.4101174345076784,
+      "grad_norm": 0.24366650749602084,
+      "learning_rate": 3.44113816343598e-06,
+      "loss": 0.6071,
+      "step": 227
+    },
+    {
+      "epoch": 0.41192411924119243,
+      "grad_norm": 0.2475981006923509,
+      "learning_rate": 3.4272619638656914e-06,
+      "loss": 0.5936,
+      "step": 228
+    },
+    {
+      "epoch": 0.4137308039747064,
+      "grad_norm": 0.2153288321867736,
+      "learning_rate": 3.4133525609159883e-06,
+      "loss": 0.5953,
+      "step": 229
+    },
+    {
+      "epoch": 0.41553748870822044,
+      "grad_norm": 0.22686236074382107,
+      "learning_rate": 3.399410452654518e-06,
+      "loss": 0.5977,
+      "step": 230
+    },
+    {
+      "epoch": 0.4173441734417344,
+      "grad_norm": 0.229068504220945,
+      "learning_rate": 3.3854361383200372e-06,
+      "loss": 0.5982,
+      "step": 231
+    },
+    {
+      "epoch": 0.41915085817524844,
+      "grad_norm": 0.22116222448509554,
+      "learning_rate": 3.3714301183045382e-06,
+      "loss": 0.6009,
+      "step": 232
+    },
+    {
+      "epoch": 0.4209575429087624,
+      "grad_norm": 0.21437564493080197,
+      "learning_rate": 3.357392894135329e-06,
+      "loss": 0.5817,
+      "step": 233
+    },
+    {
+      "epoch": 0.42276422764227645,
+      "grad_norm": 0.21046827286991526,
+      "learning_rate": 3.3433249684570757e-06,
+      "loss": 0.5983,
+      "step": 234
+    },
+    {
+      "epoch": 0.4245709123757904,
+      "grad_norm": 0.21841757461269223,
+      "learning_rate": 3.329226845013802e-06,
+      "loss": 0.593,
+      "step": 235
+    },
+    {
+      "epoch": 0.42637759710930445,
+      "grad_norm": 0.22078792972966105,
+      "learning_rate": 3.315099028630855e-06,
+      "loss": 0.6048,
+      "step": 236
+    },
+    {
+      "epoch": 0.4281842818428184,
+      "grad_norm": 0.2198180395201429,
+      "learning_rate": 3.3009420251968245e-06,
+      "loss": 0.603,
+      "step": 237
+    },
+    {
+      "epoch": 0.42999096657633246,
+      "grad_norm": 0.22974521057955885,
+      "learning_rate": 3.28675634164543e-06,
+      "loss": 0.6122,
+      "step": 238
+    },
+    {
+      "epoch": 0.43179765130984643,
+      "grad_norm": 0.24212813694550406,
+      "learning_rate": 3.272542485937369e-06,
+      "loss": 0.5942,
+      "step": 239
+    },
+    {
+      "epoch": 0.43360433604336046,
+      "grad_norm": 0.2151106216359987,
+      "learning_rate": 3.258300967042125e-06,
+      "loss": 0.5947,
+      "step": 240
+    },
+    {
+      "epoch": 0.43541102077687444,
+      "grad_norm": 0.2523287732959785,
+      "learning_rate": 3.2440322949197467e-06,
+      "loss": 0.5927,
+      "step": 241
+    },
+    {
+      "epoch": 0.4372177055103884,
+      "grad_norm": 0.21564961729801052,
+      "learning_rate": 3.229736980502584e-06,
+      "loss": 0.5934,
+      "step": 242
+    },
+    {
+      "epoch": 0.43902439024390244,
+      "grad_norm": 0.2242395897057702,
+      "learning_rate": 3.2154155356769922e-06,
+      "loss": 0.5973,
+      "step": 243
+    },
+    {
+      "epoch": 0.4408310749774164,
+      "grad_norm": 0.2232376839835937,
+      "learning_rate": 3.201068473265007e-06,
+      "loss": 0.6029,
+      "step": 244
+    },
+    {
+      "epoch": 0.44263775971093045,
+      "grad_norm": 0.2192077460257132,
+      "learning_rate": 3.186696307005976e-06,
+      "loss": 0.5962,
+      "step": 245
+    },
+    {
+      "epoch": 0.4444444444444444,
+      "grad_norm": 0.2222570172571517,
+      "learning_rate": 3.1722995515381644e-06,
+      "loss": 0.6026,
+      "step": 246
+    },
+    {
+      "epoch": 0.44625112917795845,
+      "grad_norm": 0.21527201790189038,
+      "learning_rate": 3.1578787223803296e-06,
+      "loss": 0.5855,
+      "step": 247
+    },
+    {
+      "epoch": 0.4480578139114724,
+      "grad_norm": 0.35449983881422265,
+      "learning_rate": 3.1434343359132565e-06,
+      "loss": 0.6037,
+      "step": 248
+    },
+    {
+      "epoch": 0.44986449864498645,
+      "grad_norm": 0.2062475229994345,
+      "learning_rate": 3.128966909361272e-06,
+      "loss": 0.5867,
+      "step": 249
+    },
+    {
+      "epoch": 0.45167118337850043,
+      "grad_norm": 0.21960764548448194,
+      "learning_rate": 3.1144769607737204e-06,
+      "loss": 0.6011,
+      "step": 250
+    },
+    {
+      "epoch": 0.45347786811201446,
+      "grad_norm": 0.22438900737589454,
+      "learning_rate": 3.099965009006415e-06,
+      "loss": 0.5894,
+      "step": 251
+    },
+    {
+      "epoch": 0.45528455284552843,
+      "grad_norm": 0.21781256304179047,
+      "learning_rate": 3.08543157370306e-06,
+      "loss": 0.5705,
+      "step": 252
+    },
+    {
+      "epoch": 0.45709123757904246,
+      "grad_norm": 0.23539403534261766,
+      "learning_rate": 3.0708771752766397e-06,
+      "loss": 0.5933,
+      "step": 253
+    },
+    {
+      "epoch": 0.45889792231255644,
+      "grad_norm": 0.2329715923142623,
+      "learning_rate": 3.056302334890786e-06,
+      "loss": 0.5919,
+      "step": 254
+    },
+    {
+      "epoch": 0.46070460704607047,
+      "grad_norm": 0.20731356483768837,
+      "learning_rate": 3.041707574441118e-06,
+      "loss": 0.5864,
+      "step": 255
+    },
+    {
+      "epoch": 0.46251129177958444,
+      "grad_norm": 0.23154663889322297,
+      "learning_rate": 3.027093416536548e-06,
+      "loss": 0.5868,
+      "step": 256
+    },
+    {
+      "epoch": 0.4643179765130985,
+      "grad_norm": 0.2206996661813995,
+      "learning_rate": 3.0124603844805767e-06,
+      "loss": 0.596,
+      "step": 257
+    },
+    {
+      "epoch": 0.46612466124661245,
+      "grad_norm": 0.2047968733465246,
+      "learning_rate": 2.9978090022525456e-06,
+      "loss": 0.5913,
+      "step": 258
+    },
+    {
+      "epoch": 0.4679313459801265,
+      "grad_norm": 0.21949609169081136,
+      "learning_rate": 2.9831397944888833e-06,
+      "loss": 0.5954,
+      "step": 259
+    },
+    {
+      "epoch": 0.46973803071364045,
+      "grad_norm": 0.2313444882097326,
+      "learning_rate": 2.9684532864643123e-06,
+      "loss": 0.6094,
+      "step": 260
+    },
+    {
+      "epoch": 0.4715447154471545,
+      "grad_norm": 0.23087971694617518,
+      "learning_rate": 2.953750004073041e-06,
+      "loss": 0.6008,
+      "step": 261
+    },
+    {
+      "epoch": 0.47335140018066846,
+      "grad_norm": 0.20279190342082626,
+      "learning_rate": 2.9390304738099385e-06,
+      "loss": 0.5897,
+      "step": 262
+    },
+    {
+      "epoch": 0.4751580849141825,
+      "grad_norm": 0.20638429907077857,
+      "learning_rate": 2.9242952227516726e-06,
+      "loss": 0.593,
+      "step": 263
+    },
+    {
+      "epoch": 0.47696476964769646,
+      "grad_norm": 0.22710491639329788,
+      "learning_rate": 2.9095447785378446e-06,
+      "loss": 0.6007,
+      "step": 264
+    },
+    {
+      "epoch": 0.4787714543812105,
+      "grad_norm": 0.21937325130622803,
+      "learning_rate": 2.89477966935209e-06,
+      "loss": 0.5811,
+      "step": 265
+    },
+    {
+      "epoch": 0.48057813911472447,
+      "grad_norm": 0.21143254384346213,
+      "learning_rate": 2.8800004239031687e-06,
+      "loss": 0.6052,
+      "step": 266
+    },
+    {
+      "epoch": 0.4823848238482385,
+      "grad_norm": 0.21519263645728862,
+      "learning_rate": 2.8652075714060296e-06,
+      "loss": 0.5895,
+      "step": 267
+    },
+    {
+      "epoch": 0.48419150858175247,
+      "grad_norm": 0.2241818383841725,
+      "learning_rate": 2.850401641562865e-06,
+      "loss": 0.5903,
+      "step": 268
+    },
+    {
+      "epoch": 0.4859981933152665,
+      "grad_norm": 0.2145645757032261,
+      "learning_rate": 2.835583164544139e-06,
+      "loss": 0.5958,
+      "step": 269
+    },
+    {
+      "epoch": 0.4878048780487805,
+      "grad_norm": 0.21634063420771613,
+      "learning_rate": 2.820752670969606e-06,
+      "loss": 0.5785,
+      "step": 270
+    },
+    {
+      "epoch": 0.4896115627822945,
+      "grad_norm": 0.22267426902820459,
+      "learning_rate": 2.805910691889307e-06,
+      "loss": 0.5996,
+      "step": 271
+    },
+    {
+      "epoch": 0.4914182475158085,
+      "grad_norm": 0.2209235755384397,
+      "learning_rate": 2.791057758764557e-06,
+      "loss": 0.5941,
+      "step": 272
+    },
+    {
+      "epoch": 0.4932249322493225,
+      "grad_norm": 0.21545901337093543,
+      "learning_rate": 2.776194403448915e-06,
+      "loss": 0.6022,
+      "step": 273
+    },
+    {
+      "epoch": 0.4950316169828365,
+      "grad_norm": 0.23880363865262091,
+      "learning_rate": 2.761321158169134e-06,
+      "loss": 0.5893,
+      "step": 274
+    },
+    {
+      "epoch": 0.4968383017163505,
+      "grad_norm": 0.20265741752620156,
+      "learning_rate": 2.7464385555061092e-06,
+      "loss": 0.5842,
+      "step": 275
+    },
+    {
+      "epoch": 0.4986449864498645,
+      "grad_norm": 0.19957842808608858,
+      "learning_rate": 2.731547128375804e-06,
+      "loss": 0.5765,
+      "step": 276
+    },
+    {
+      "epoch": 0.5004516711833785,
+      "grad_norm": 0.24313154793244834,
+      "learning_rate": 2.7166474100101676e-06,
+      "loss": 0.5889,
+      "step": 277
+    },
+    {
+      "epoch": 0.5022583559168925,
+      "grad_norm": 0.21578072503135698,
+      "learning_rate": 2.7017399339380435e-06,
+      "loss": 0.5752,
+      "step": 278
+    },
+    {
+      "epoch": 0.5040650406504065,
+      "grad_norm": 0.22695165367196,
+      "learning_rate": 2.686825233966061e-06,
+      "loss": 0.5901,
+      "step": 279
+    },
+    {
+      "epoch": 0.5058717253839206,
+      "grad_norm": 0.22124104537145198,
+      "learning_rate": 2.6719038441595236e-06,
+      "loss": 0.5865,
+      "step": 280
+    },
+    {
+      "epoch": 0.5076784101174345,
+      "grad_norm": 0.20876329166600804,
+      "learning_rate": 2.6569762988232838e-06,
+      "loss": 0.5875,
+      "step": 281
+    },
+    {
+      "epoch": 0.5094850948509485,
+      "grad_norm": 0.21423770322129848,
+      "learning_rate": 2.642043132482612e-06,
+      "loss": 0.5805,
+      "step": 282
+    },
+    {
+      "epoch": 0.5112917795844625,
+      "grad_norm": 0.21416739364725013,
+      "learning_rate": 2.6271048798640547e-06,
+      "loss": 0.5979,
+      "step": 283
+    },
+    {
+      "epoch": 0.5130984643179766,
+      "grad_norm": 0.20856984547372137,
+      "learning_rate": 2.6121620758762877e-06,
+      "loss": 0.5854,
+      "step": 284
+    },
+    {
+      "epoch": 0.5149051490514905,
+      "grad_norm": 0.20653823718614767,
+      "learning_rate": 2.5972152555909625e-06,
+      "loss": 0.5777,
+      "step": 285
+    },
+    {
+      "epoch": 0.5167118337850045,
+      "grad_norm": 0.20566248008855165,
+      "learning_rate": 2.5822649542235468e-06,
+      "loss": 0.5784,
+      "step": 286
+    },
+    {
+      "epoch": 0.5185185185185185,
+      "grad_norm": 0.2256337819483314,
+      "learning_rate": 2.5673117071141574e-06,
+      "loss": 0.6006,
+      "step": 287
+    },
+    {
+      "epoch": 0.5203252032520326,
+      "grad_norm": 0.2187486752392677,
+      "learning_rate": 2.5523560497083927e-06,
+      "loss": 0.5963,
+      "step": 288
+    },
+    {
+      "epoch": 0.5221318879855466,
+      "grad_norm": 0.2256640463012208,
+      "learning_rate": 2.5373985175381595e-06,
+      "loss": 0.5806,
+      "step": 289
+    },
+    {
+      "epoch": 0.5239385727190605,
+      "grad_norm": 0.20558830059411975,
+      "learning_rate": 2.522439646202495e-06,
+      "loss": 0.5927,
+      "step": 290
+    },
+    {
+      "epoch": 0.5257452574525745,
+      "grad_norm": 0.21091753896787113,
+      "learning_rate": 2.507479971348391e-06,
+      "loss": 0.5974,
+      "step": 291
+    },
+    {
+      "epoch": 0.5275519421860885,
+      "grad_norm": 0.20163177164774673,
+      "learning_rate": 2.49252002865161e-06,
+      "loss": 0.5999,
+      "step": 292
+    },
+    {
+      "epoch": 0.5293586269196026,
+      "grad_norm": 0.21301882714221618,
+      "learning_rate": 2.4775603537975055e-06,
+      "loss": 0.5943,
+      "step": 293
+    },
+    {
+      "epoch": 0.5311653116531165,
+      "grad_norm": 0.21486758628846855,
+      "learning_rate": 2.4626014824618418e-06,
+      "loss": 0.5855,
+      "step": 294
+    },
+    {
+      "epoch": 0.5329719963866305,
+      "grad_norm": 0.20752918302238268,
+      "learning_rate": 2.447643950291608e-06,
+      "loss": 0.5843,
+      "step": 295
+    },
+    {
+      "epoch": 0.5347786811201445,
+      "grad_norm": 0.21785997768890877,
+      "learning_rate": 2.4326882928858435e-06,
+      "loss": 0.5868,
+      "step": 296
+    },
+    {
+      "epoch": 0.5365853658536586,
+      "grad_norm": 0.21841154134063173,
+      "learning_rate": 2.417735045776453e-06,
+      "loss": 0.5935,
+      "step": 297
+    },
+    {
+      "epoch": 0.5383920505871725,
+      "grad_norm": 0.20318141655763375,
+      "learning_rate": 2.4027847444090384e-06,
+      "loss": 0.5873,
+      "step": 298
+    },
+    {
+      "epoch": 0.5401987353206865,
+      "grad_norm": 0.3832839441010807,
+      "learning_rate": 2.3878379241237136e-06,
+      "loss": 0.604,
+      "step": 299
+    },
+    {
+      "epoch": 0.5420054200542005,
+      "grad_norm": 0.2105425262709393,
+      "learning_rate": 2.372895120135946e-06,
+      "loss": 0.5917,
+      "step": 300
+    },
+    {
+      "epoch": 0.5438121047877146,
+      "grad_norm": 0.21755593789778102,
+      "learning_rate": 2.3579568675173894e-06,
+      "loss": 0.5917,
+      "step": 301
+    },
+    {
+      "epoch": 0.5456187895212286,
+      "grad_norm": 0.19989127666575213,
+      "learning_rate": 2.3430237011767166e-06,
+      "loss": 0.6047,
+      "step": 302
+    },
+    {
+      "epoch": 0.5474254742547425,
+      "grad_norm": 0.2055760488608315,
+      "learning_rate": 2.3280961558404773e-06,
+      "loss": 0.5859,
+      "step": 303
+    },
+    {
+      "epoch": 0.5492321589882565,
+      "grad_norm": 0.2121939170236,
+      "learning_rate": 2.3131747660339396e-06,
+      "loss": 0.581,
+      "step": 304
+    },
+    {
+      "epoch": 0.5510388437217706,
+      "grad_norm": 0.21164112959432857,
+      "learning_rate": 2.2982600660619574e-06,
+      "loss": 0.5872,
+      "step": 305
+    },
+    {
+      "epoch": 0.5528455284552846,
+      "grad_norm": 0.20304982313955747,
+      "learning_rate": 2.2833525899898324e-06,
+      "loss": 0.5851,
+      "step": 306
+    },
+    {
+      "epoch": 0.5546522131887985,
+      "grad_norm": 0.2144496646482213,
+      "learning_rate": 2.268452871624197e-06,
+      "loss": 0.6032,
+      "step": 307
+    },
+    {
+      "epoch": 0.5564588979223125,
+      "grad_norm": 0.20431153355938347,
+      "learning_rate": 2.253561444493891e-06,
+      "loss": 0.5824,
+      "step": 308
+    },
+    {
+      "epoch": 0.5582655826558266,
+      "grad_norm": 0.21664720003578983,
+      "learning_rate": 2.238678841830867e-06,
+      "loss": 0.5915,
+      "step": 309
+    },
+    {
+      "epoch": 0.5600722673893406,
+      "grad_norm": 0.20197470495761588,
+      "learning_rate": 2.2238055965510853e-06,
+      "loss": 0.5751,
+      "step": 310
+    },
+    {
+      "epoch": 0.5618789521228545,
+      "grad_norm": 0.4074972564492422,
+      "learning_rate": 2.2089422412354434e-06,
+      "loss": 0.5929,
+      "step": 311
+    },
+    {
+      "epoch": 0.5636856368563685,
+      "grad_norm": 0.19998974102467304,
+      "learning_rate": 2.1940893081106946e-06,
+      "loss": 0.5923,
+      "step": 312
+    },
+    {
+      "epoch": 0.5654923215898826,
+      "grad_norm": 0.2117759167471859,
+      "learning_rate": 2.179247329030395e-06,
+      "loss": 0.5946,
+      "step": 313
+    },
+    {
+      "epoch": 0.5672990063233966,
+      "grad_norm": 0.2017003806183738,
+      "learning_rate": 2.1644168354558623e-06,
+      "loss": 0.5859,
+      "step": 314
+    },
+    {
+      "epoch": 0.5691056910569106,
+      "grad_norm": 0.20263926965950277,
+      "learning_rate": 2.1495983584371354e-06,
+      "loss": 0.5855,
+      "step": 315
+    },
+    {
+      "epoch": 0.5709123757904245,
+      "grad_norm": 0.20684496610144393,
+      "learning_rate": 2.134792428593971e-06,
+      "loss": 0.5831,
+      "step": 316
+    },
+    {
+      "epoch": 0.5727190605239386,
+      "grad_norm": 0.2144432554994338,
+      "learning_rate": 2.119999576096832e-06,
+      "loss": 0.5774,
+      "step": 317
+    },
+    {
+      "epoch": 0.5745257452574526,
+      "grad_norm": 0.21247466919295185,
+      "learning_rate": 2.1052203306479108e-06,
+      "loss": 0.5896,
+      "step": 318
+    },
+    {
+      "epoch": 0.5763324299909666,
+      "grad_norm": 0.21422659294493748,
+      "learning_rate": 2.090455221462156e-06,
+      "loss": 0.5993,
+      "step": 319
+    },
+    {
+      "epoch": 0.5781391147244805,
+      "grad_norm": 0.22762142116566894,
+      "learning_rate": 2.0757047772483278e-06,
+      "loss": 0.583,
+      "step": 320
+    },
+    {
+      "epoch": 0.5799457994579946,
+      "grad_norm": 0.20743085860376967,
+      "learning_rate": 2.0609695261900624e-06,
+      "loss": 0.5784,
+      "step": 321
+    },
+    {
+      "epoch": 0.5817524841915086,
+      "grad_norm": 0.21691881173857994,
+      "learning_rate": 2.0462499959269596e-06,
+      "loss": 0.6002,
+      "step": 322
+    },
+    {
+      "epoch": 0.5835591689250226,
+      "grad_norm": 0.22483397993289836,
+      "learning_rate": 2.031546713535688e-06,
+      "loss": 0.5857,
+      "step": 323
+    },
+    {
+      "epoch": 0.5853658536585366,
+      "grad_norm": 0.1941832295922758,
+      "learning_rate": 2.0168602055111175e-06,
+      "loss": 0.5657,
+      "step": 324
+    },
+    {
+      "epoch": 0.5871725383920506,
+      "grad_norm": 0.2106447854766266,
+      "learning_rate": 2.0021909977474553e-06,
+      "loss": 0.5842,
+      "step": 325
+    },
+    {
+      "epoch": 0.5889792231255646,
+      "grad_norm": 0.20623748048616064,
+      "learning_rate": 1.987539615519424e-06,
+      "loss": 0.5928,
+      "step": 326
+    },
+    {
+      "epoch": 0.5907859078590786,
+      "grad_norm": 0.24434345791680295,
+      "learning_rate": 1.9729065834634533e-06,
+      "loss": 0.5808,
+      "step": 327
+    },
+    {
+      "epoch": 0.5925925925925926,
+      "grad_norm": 0.2091268010269944,
+      "learning_rate": 1.958292425558883e-06,
+      "loss": 0.5863,
+      "step": 328
+    },
+    {
+      "epoch": 0.5943992773261066,
+      "grad_norm": 0.2290855141568565,
+      "learning_rate": 1.9436976651092143e-06,
+      "loss": 0.5778,
+      "step": 329
+    },
+    {
+      "epoch": 0.5962059620596206,
+      "grad_norm": 0.2326519271370711,
+      "learning_rate": 1.9291228247233607e-06,
+      "loss": 0.5847,
+      "step": 330
+    },
+    {
+      "epoch": 0.5980126467931346,
+      "grad_norm": 0.21071088791420517,
+      "learning_rate": 1.9145684262969404e-06,
+      "loss": 0.5901,
+      "step": 331
+    },
+    {
+      "epoch": 0.5998193315266486,
+      "grad_norm": 0.19854103741318652,
+      "learning_rate": 1.9000349909935852e-06,
+      "loss": 0.5721,
+      "step": 332
+    },
+    {
+      "epoch": 0.6016260162601627,
+      "grad_norm": 0.19972103169844774,
+      "learning_rate": 1.8855230392262809e-06,
+      "loss": 0.5929,
+      "step": 333
+    },
+    {
+      "epoch": 0.6034327009936766,
+      "grad_norm": 0.20709957794921252,
+      "learning_rate": 1.8710330906387288e-06,
+      "loss": 0.6057,
+      "step": 334
+    },
+    {
+      "epoch": 0.6052393857271906,
+      "grad_norm": 0.21536606128137195,
+      "learning_rate": 1.8565656640867448e-06,
+      "loss": 0.5879,
+      "step": 335
+    },
+    {
+      "epoch": 0.6070460704607046,
+      "grad_norm": 0.20178356038933531,
+      "learning_rate": 1.8421212776196712e-06,
+      "loss": 0.5854,
+      "step": 336
+    },
+    {
+      "epoch": 0.6088527551942186,
+      "grad_norm": 0.2014990307576943,
+      "learning_rate": 1.827700448461836e-06,
+      "loss": 0.5863,
+      "step": 337
+    },
+    {
+      "epoch": 0.6106594399277326,
+      "grad_norm": 0.2136062417554571,
+      "learning_rate": 1.813303692994025e-06,
+      "loss": 0.5822,
+      "step": 338
+    },
+    {
+      "epoch": 0.6124661246612466,
+      "grad_norm": 0.2019958187136403,
+      "learning_rate": 1.7989315267349936e-06,
+      "loss": 0.591,
+      "step": 339
+    },
+    {
+      "epoch": 0.6142728093947606,
+      "grad_norm": 0.1964885288651704,
+      "learning_rate": 1.7845844643230086e-06,
+      "loss": 0.5865,
+      "step": 340
+    },
+    {
+      "epoch": 0.6160794941282746,
+      "grad_norm": 0.1981630352126773,
+      "learning_rate": 1.770263019497417e-06,
+      "loss": 0.5845,
+      "step": 341
+    },
+    {
+      "epoch": 0.6178861788617886,
+      "grad_norm": 0.2088515893373291,
+      "learning_rate": 1.7559677050802543e-06,
+      "loss": 0.5882,
+      "step": 342
+    },
+    {
+      "epoch": 0.6196928635953026,
+      "grad_norm": 0.20763167733521767,
+      "learning_rate": 1.7416990329578753e-06,
+      "loss": 0.5825,
+      "step": 343
+    },
+    {
+      "epoch": 0.6214995483288166,
+      "grad_norm": 0.22467403511881023,
+      "learning_rate": 1.7274575140626318e-06,
+      "loss": 0.572,
+      "step": 344
+    },
+    {
+      "epoch": 0.6233062330623306,
+      "grad_norm": 0.21203843083401738,
+      "learning_rate": 1.7132436583545703e-06,
+      "loss": 0.5866,
+      "step": 345
+    },
+    {
+      "epoch": 0.6251129177958447,
+      "grad_norm": 0.22023981214278937,
+      "learning_rate": 1.699057974803176e-06,
+      "loss": 0.5884,
+      "step": 346
+    },
+    {
+      "epoch": 0.6269196025293586,
+      "grad_norm": 0.20345704072148607,
+      "learning_rate": 1.6849009713691456e-06,
+      "loss": 0.5729,
+      "step": 347
+    },
+    {
+      "epoch": 0.6287262872628726,
+      "grad_norm": 0.2203981009854613,
+      "learning_rate": 1.670773154986199e-06,
+      "loss": 0.5741,
+      "step": 348
+    },
+    {
+      "epoch": 0.6305329719963866,
+      "grad_norm": 0.2124430311627528,
+      "learning_rate": 1.6566750315429254e-06,
+      "loss": 0.5764,
+      "step": 349
+    },
+    {
+      "epoch": 0.6323396567299007,
+      "grad_norm": 0.2167604699186445,
+      "learning_rate": 1.6426071058646718e-06,
+      "loss": 0.5892,
+      "step": 350
+    },
+    {
+      "epoch": 0.6341463414634146,
+      "grad_norm": 0.2068136183136913,
+      "learning_rate": 1.6285698816954626e-06,
+      "loss": 0.5795,
+      "step": 351
+    },
+    {
+      "epoch": 0.6359530261969286,
+      "grad_norm": 0.22139771703636216,
+      "learning_rate": 1.6145638616799636e-06,
+      "loss": 0.5933,
+      "step": 352
+    },
+    {
+      "epoch": 0.6377597109304426,
+      "grad_norm": 0.2169072872024213,
+      "learning_rate": 1.6005895473454836e-06,
+      "loss": 0.5931,
+      "step": 353
+    },
+    {
+      "epoch": 0.6395663956639567,
+      "grad_norm": 0.19987512033116334,
+      "learning_rate": 1.5866474390840126e-06,
+      "loss": 0.5885,
+      "step": 354
+    },
+    {
+      "epoch": 0.6413730803974707,
+      "grad_norm": 0.1976920115301846,
+      "learning_rate": 1.5727380361343103e-06,
+      "loss": 0.5778,
+      "step": 355
+    },
+    {
+      "epoch": 0.6431797651309846,
+      "grad_norm": 0.20307084136655387,
+      "learning_rate": 1.55886183656402e-06,
+      "loss": 0.5932,
+      "step": 356
+    },
+    {
+      "epoch": 0.6449864498644986,
+      "grad_norm": 0.22040332423078723,
+      "learning_rate": 1.545019337251844e-06,
+      "loss": 0.5935,
+      "step": 357
+    },
+    {
+      "epoch": 0.6467931345980127,
+      "grad_norm": 0.19534686963161454,
+      "learning_rate": 1.5312110338697427e-06,
+      "loss": 0.5989,
+      "step": 358
+    },
+    {
+      "epoch": 0.6485998193315267,
+      "grad_norm": 0.2071188241938468,
+      "learning_rate": 1.5174374208651913e-06,
+      "loss": 0.5713,
+      "step": 359
+    },
+    {
+      "epoch": 0.6504065040650406,
+      "grad_norm": 0.22023603109435608,
+      "learning_rate": 1.503698991443471e-06,
+      "loss": 0.5799,
+      "step": 360
+    },
+    {
+      "epoch": 0.6522131887985546,
+      "grad_norm": 0.23279333269648594,
+      "learning_rate": 1.489996237550012e-06,
+      "loss": 0.5864,
+      "step": 361
+    },
+    {
+      "epoch": 0.6540198735320687,
+      "grad_norm": 0.21130573335881214,
+      "learning_rate": 1.4763296498527744e-06,
+      "loss": 0.5975,
+      "step": 362
+    },
+    {
+      "epoch": 0.6558265582655827,
+      "grad_norm": 0.21591708899052672,
+      "learning_rate": 1.4626997177246787e-06,
+      "loss": 0.5878,
+      "step": 363
+    },
+    {
+      "epoch": 0.6576332429990966,
+      "grad_norm": 0.2371469411885916,
+      "learning_rate": 1.4491069292260867e-06,
+      "loss": 0.5862,
+      "step": 364
+    },
+    {
+      "epoch": 0.6594399277326106,
+      "grad_norm": 0.20206187850887294,
+      "learning_rate": 1.4355517710873184e-06,
+      "loss": 0.5784,
+      "step": 365
+    },
+    {
+      "epoch": 0.6612466124661247,
+      "grad_norm": 0.20548129735299142,
+      "learning_rate": 1.4220347286912296e-06,
+      "loss": 0.6009,
+      "step": 366
+    },
+    {
+      "epoch": 0.6630532971996387,
+      "grad_norm": 0.20462764510073306,
+      "learning_rate": 1.4085562860558256e-06,
+      "loss": 0.5818,
+      "step": 367
+    },
+    {
+      "epoch": 0.6648599819331527,
+      "grad_norm": 0.21559927754232822,
+      "learning_rate": 1.395116925816934e-06,
+      "loss": 0.5937,
+      "step": 368
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "grad_norm": 0.24435009358387672,
+      "learning_rate": 1.3817171292109182e-06,
+      "loss": 0.5945,
+      "step": 369
+    },
+    {
+      "epoch": 0.6684733514001807,
+      "grad_norm": 0.20734944097931568,
+      "learning_rate": 1.3683573760574526e-06,
+      "loss": 0.5931,
+      "step": 370
+    },
+    {
+      "epoch": 0.6702800361336947,
+      "grad_norm": 0.28137987727489505,
+      "learning_rate": 1.3550381447423317e-06,
+      "loss": 0.5813,
+      "step": 371
+    },
+    {
+      "epoch": 0.6720867208672087,
+      "grad_norm": 0.20129121777849085,
+      "learning_rate": 1.3417599122003464e-06,
+      "loss": 0.5992,
+      "step": 372
+    },
+    {
+      "epoch": 0.6738934056007226,
+      "grad_norm": 0.207337984886545,
+      "learning_rate": 1.3285231538982036e-06,
+      "loss": 0.5848,
+      "step": 373
+    },
+    {
+      "epoch": 0.6757000903342367,
+      "grad_norm": 0.19414987785618634,
+      "learning_rate": 1.3153283438175036e-06,
+      "loss": 0.5734,
+      "step": 374
+    },
+    {
+      "epoch": 0.6775067750677507,
+      "grad_norm": 0.19139015126778156,
+      "learning_rate": 1.3021759544377632e-06,
+      "loss": 0.5889,
+      "step": 375
+    },
+    {
+      "epoch": 0.6793134598012647,
+      "grad_norm": 0.19455499325414877,
+      "learning_rate": 1.2890664567195e-06,
+      "loss": 0.5735,
+      "step": 376
+    },
+    {
+      "epoch": 0.6811201445347786,
+      "grad_norm": 0.20663774065898713,
+      "learning_rate": 1.27600032008737e-06,
+      "loss": 0.589,
+      "step": 377
+    },
+    {
+      "epoch": 0.6829268292682927,
+      "grad_norm": 0.19788629496456012,
+      "learning_rate": 1.2629780124133511e-06,
+      "loss": 0.5903,
+      "step": 378
+    },
+    {
+      "epoch": 0.6847335140018067,
+      "grad_norm": 0.20175651716363546,
+      "learning_rate": 1.2500000000000007e-06,
+      "loss": 0.592,
+      "step": 379
+    },
+    {
+      "epoch": 0.6865401987353207,
+      "grad_norm": 0.18819797058808696,
+      "learning_rate": 1.2370667475637474e-06,
+      "loss": 0.5733,
+      "step": 380
+    },
+    {
+      "epoch": 0.6883468834688347,
+      "grad_norm": 0.20251745573990865,
+      "learning_rate": 1.2241787182182596e-06,
+      "loss": 0.5893,
+      "step": 381
+    },
+    {
+      "epoch": 0.6901535682023487,
+      "grad_norm": 0.20807680817859822,
+      "learning_rate": 1.2113363734578548e-06,
+      "loss": 0.5771,
+      "step": 382
+    },
+    {
+      "epoch": 0.6919602529358627,
+      "grad_norm": 0.19645518245442806,
+      "learning_rate": 1.1985401731409793e-06,
+      "loss": 0.5966,
+      "step": 383
+    },
+    {
+      "epoch": 0.6937669376693767,
+      "grad_norm": 0.2079159019479432,
+      "learning_rate": 1.185790575473738e-06,
+      "loss": 0.5951,
+      "step": 384
+    },
+    {
+      "epoch": 0.6955736224028907,
+      "grad_norm": 0.19274863118289084,
+      "learning_rate": 1.1730880369934933e-06,
+      "loss": 0.5769,
+      "step": 385
+    },
+    {
+      "epoch": 0.6973803071364046,
+      "grad_norm": 0.18698223784520404,
+      "learning_rate": 1.160433012552508e-06,
+      "loss": 0.5881,
+      "step": 386
+    },
+    {
+      "epoch": 0.6991869918699187,
+      "grad_norm": 0.20644762996969238,
+      "learning_rate": 1.1478259553016683e-06,
+      "loss": 0.5883,
+      "step": 387
+    },
+    {
+      "epoch": 0.7009936766034327,
+      "grad_norm": 0.20354407357487309,
+      "learning_rate": 1.1352673166742463e-06,
+      "loss": 0.5878,
+      "step": 388
+    },
+    {
+      "epoch": 0.7028003613369467,
+      "grad_norm": 0.21972699427814146,
+      "learning_rate": 1.122757546369744e-06,
+      "loss": 0.5915,
+      "step": 389
+    },
+    {
+      "epoch": 0.7046070460704607,
+      "grad_norm": 0.21740455290435515,
+      "learning_rate": 1.1102970923377893e-06,
+      "loss": 0.5837,
+      "step": 390
+    },
+    {
+      "epoch": 0.7064137308039747,
+      "grad_norm": 0.2024883299761826,
+      "learning_rate": 1.0978864007620896e-06,
+      "loss": 0.5852,
+      "step": 391
+    },
+    {
+      "epoch": 0.7082204155374887,
+      "grad_norm": 0.2187579171935915,
+      "learning_rate": 1.085525916044464e-06,
+      "loss": 0.591,
+      "step": 392
+    },
+    {
+      "epoch": 0.7100271002710027,
+      "grad_norm": 0.20784813623988108,
+      "learning_rate": 1.073216080788921e-06,
+      "loss": 0.5954,
+      "step": 393
+    },
+    {
+      "epoch": 0.7118337850045167,
+      "grad_norm": 0.19102198360909103,
+      "learning_rate": 1.0609573357858166e-06,
+      "loss": 0.5969,
+      "step": 394
+    },
+    {
+      "epoch": 0.7136404697380307,
+      "grad_norm": 0.22162192671998815,
+      "learning_rate": 1.048750119996066e-06,
+      "loss": 0.5937,
+      "step": 395
+    },
+    {
+      "epoch": 0.7154471544715447,
+      "grad_norm": 0.25432100082751324,
+      "learning_rate": 1.0365948705354309e-06,
+      "loss": 0.5863,
+      "step": 396
+    },
+    {
+      "epoch": 0.7172538392050587,
+      "grad_norm": 0.24853077784213276,
+      "learning_rate": 1.0244920226588599e-06,
+      "loss": 0.5853,
+      "step": 397
+    },
+    {
+      "epoch": 0.7190605239385727,
+      "grad_norm": 0.20385116203175707,
+      "learning_rate": 1.0124420097449077e-06,
+      "loss": 0.5871,
+      "step": 398
+    },
+    {
+      "epoch": 0.7208672086720868,
+      "grad_norm": 0.18532845669355508,
+      "learning_rate": 1.0004452632802158e-06,
+      "loss": 0.5916,
+      "step": 399
+    },
+    {
+      "epoch": 0.7226738934056007,
+      "grad_norm": 0.20423244886055075,
+      "learning_rate": 9.88502212844063e-07,
+      "loss": 0.5872,
+      "step": 400
+    },
+    {
+      "epoch": 0.7244805781391147,
+      "grad_norm": 0.19509657230241026,
+      "learning_rate": 9.7661328609298e-07,
+      "loss": 0.5898,
+      "step": 401
+    },
+    {
+      "epoch": 0.7262872628726287,
+      "grad_norm": 0.21353802006223266,
+      "learning_rate": 9.64778908745437e-07,
+      "loss": 0.5976,
+      "step": 402
+    },
+    {
+      "epoch": 0.7280939476061428,
+      "grad_norm": 0.20435405015728278,
+      "learning_rate": 9.529995045666041e-07,
+      "loss": 0.5788,
+      "step": 403
+    },
+    {
+      "epoch": 0.7299006323396567,
+      "grad_norm": 0.1984529066802218,
+      "learning_rate": 9.412754953531664e-07,
+      "loss": 0.5827,
+      "step": 404
+    },
+    {
+      "epoch": 0.7317073170731707,
+      "grad_norm": 0.2167858934267235,
+      "learning_rate": 9.296073009182341e-07,
+      "loss": 0.5821,
+      "step": 405
+    },
+    {
+      "epoch": 0.7335140018066847,
+      "grad_norm": 0.1987523835045179,
+      "learning_rate": 9.179953390762977e-07,
+      "loss": 0.5889,
+      "step": 406
+    },
+    {
+      "epoch": 0.7353206865401988,
+      "grad_norm": 0.20697438352924746,
+      "learning_rate": 9.064400256282757e-07,
+      "loss": 0.5806,
+      "step": 407
+    },
+    {
+      "epoch": 0.7371273712737128,
+      "grad_norm": 0.19116589279660615,
+      "learning_rate": 8.949417743466199e-07,
+      "loss": 0.5754,
+      "step": 408
+    },
+    {
+      "epoch": 0.7389340560072267,
+      "grad_norm": 0.211250643585691,
+      "learning_rate": 8.835009969605013e-07,
+      "loss": 0.5927,
+      "step": 409
+    },
+    {
+      "epoch": 0.7407407407407407,
+      "grad_norm": 0.20536890873705252,
+      "learning_rate": 8.721181031410661e-07,
+      "loss": 0.5949,
+      "step": 410
+    },
+    {
+      "epoch": 0.7425474254742548,
+      "grad_norm": 0.2093940066655208,
+      "learning_rate": 8.607935004867693e-07,
+      "loss": 0.5963,
+      "step": 411
+    },
+    {
+      "epoch": 0.7443541102077688,
+      "grad_norm": 0.20232355314663383,
+      "learning_rate": 8.495275945087744e-07,
+      "loss": 0.5694,
+      "step": 412
+    },
+    {
+      "epoch": 0.7461607949412827,
+      "grad_norm": 0.19499772252681724,
+      "learning_rate": 8.383207886164366e-07,
+      "loss": 0.5859,
+      "step": 413
+    },
+    {
+      "epoch": 0.7479674796747967,
+      "grad_norm": 0.24905986363655266,
+      "learning_rate": 8.271734841028553e-07,
+      "loss": 0.5732,
+      "step": 414
+    },
+    {
+      "epoch": 0.7497741644083108,
+      "grad_norm": 0.1976495120915159,
+      "learning_rate": 8.16086080130506e-07,
+      "loss": 0.595,
+      "step": 415
+    },
+    {
+      "epoch": 0.7515808491418248,
+      "grad_norm": 0.20289395576214606,
+      "learning_rate": 8.050589737169485e-07,
+      "loss": 0.5811,
+      "step": 416
+    },
+    {
+      "epoch": 0.7533875338753387,
+      "grad_norm": 0.2131733098150311,
+      "learning_rate": 7.940925597206053e-07,
+      "loss": 0.6034,
+      "step": 417
+    },
+    {
+      "epoch": 0.7551942186088527,
+      "grad_norm": 0.2043801754256928,
+      "learning_rate": 7.831872308266306e-07,
+      "loss": 0.5856,
+      "step": 418
+    },
+    {
+      "epoch": 0.7570009033423668,
+      "grad_norm": 0.1875418775915756,
+      "learning_rate": 7.723433775328385e-07,
+      "loss": 0.5857,
+      "step": 419
+    },
+    {
+      "epoch": 0.7588075880758808,
+      "grad_norm": 0.1976789303893716,
+      "learning_rate": 7.615613881357315e-07,
+      "loss": 0.6002,
+      "step": 420
+    },
+    {
+      "epoch": 0.7606142728093948,
+      "grad_norm": 0.1914286809513274,
+      "learning_rate": 7.508416487165862e-07,
+      "loss": 0.5805,
+      "step": 421
+    },
+    {
+      "epoch": 0.7624209575429087,
+      "grad_norm": 0.22294210796830027,
+      "learning_rate": 7.401845431276378e-07,
+      "loss": 0.5842,
+      "step": 422
+    },
+    {
+      "epoch": 0.7642276422764228,
+      "grad_norm": 0.22041399450512092,
+      "learning_rate": 7.295904529783265e-07,
+      "loss": 0.5935,
+      "step": 423
+    },
+    {
+      "epoch": 0.7660343270099368,
+      "grad_norm": 0.19814565637399076,
+      "learning_rate": 7.190597576216385e-07,
+      "loss": 0.5975,
+      "step": 424
+    },
+    {
+      "epoch": 0.7678410117434508,
+      "grad_norm": 0.20032853547884555,
+      "learning_rate": 7.085928341405193e-07,
+      "loss": 0.576,
+      "step": 425
+    },
+    {
+      "epoch": 0.7696476964769647,
+      "grad_norm": 0.19451615175607623,
+      "learning_rate": 6.98190057334375e-07,
+      "loss": 0.5963,
+      "step": 426
+    },
+    {
+      "epoch": 0.7714543812104788,
+      "grad_norm": 0.24329319131853208,
+      "learning_rate": 6.878517997056458e-07,
+      "loss": 0.568,
+      "step": 427
+    },
+    {
+      "epoch": 0.7732610659439928,
+      "grad_norm": 0.21219438141394756,
+      "learning_rate": 6.775784314464717e-07,
+      "loss": 0.5886,
+      "step": 428
+    },
+    {
+      "epoch": 0.7750677506775068,
+      "grad_norm": 0.19342036608461366,
+      "learning_rate": 6.673703204254348e-07,
+      "loss": 0.5827,
+      "step": 429
+    },
+    {
+      "epoch": 0.7768744354110207,
+      "grad_norm": 0.19432449104766572,
+      "learning_rate": 6.572278321743871e-07,
+      "loss": 0.5849,
+      "step": 430
+    },
+    {
+      "epoch": 0.7786811201445348,
+      "grad_norm": 0.20880674647287806,
+      "learning_rate": 6.471513298753634e-07,
+      "loss": 0.594,
+      "step": 431
+    },
+    {
+      "epoch": 0.7804878048780488,
+      "grad_norm": 0.1880528634390492,
+      "learning_rate": 6.371411743475717e-07,
+      "loss": 0.5885,
+      "step": 432
+    },
+    {
+      "epoch": 0.7822944896115628,
+      "grad_norm": 0.20738544996334468,
+      "learning_rate": 6.271977240344795e-07,
+      "loss": 0.5973,
+      "step": 433
+    },
+    {
+      "epoch": 0.7841011743450768,
+      "grad_norm": 0.19621576251799466,
+      "learning_rate": 6.17321334990973e-07,
+      "loss": 0.5963,
+      "step": 434
+    },
+    {
+      "epoch": 0.7859078590785907,
+      "grad_norm": 0.19592692847981952,
+      "learning_rate": 6.075123608706093e-07,
+      "loss": 0.5728,
+      "step": 435
+    },
+    {
+      "epoch": 0.7877145438121048,
+      "grad_norm": 0.1896894091910412,
+      "learning_rate": 5.97771152912954e-07,
+      "loss": 0.587,
+      "step": 436
+    },
+    {
+      "epoch": 0.7895212285456188,
+      "grad_norm": 0.19300077047150074,
+      "learning_rate": 5.880980599310041e-07,
+      "loss": 0.585,
+      "step": 437
+    },
+    {
+      "epoch": 0.7913279132791328,
+      "grad_norm": 0.20695396686165002,
+      "learning_rate": 5.784934282986956e-07,
+      "loss": 0.5773,
+      "step": 438
+    },
+    {
+      "epoch": 0.7931345980126467,
+      "grad_norm": 0.24669040853690524,
+      "learning_rate": 5.689576019385015e-07,
+      "loss": 0.5941,
+      "step": 439
+    },
+    {
+      "epoch": 0.7949412827461608,
+      "grad_norm": 0.2416648334177664,
+      "learning_rate": 5.59490922309118e-07,
+      "loss": 0.5861,
+      "step": 440
+    },
+    {
+      "epoch": 0.7967479674796748,
+      "grad_norm": 0.19023570934613537,
+      "learning_rate": 5.500937283932348e-07,
+      "loss": 0.57,
+      "step": 441
+    },
+    {
+      "epoch": 0.7985546522131888,
+      "grad_norm": 0.19597698226304042,
+      "learning_rate": 5.407663566854008e-07,
+      "loss": 0.5801,
+      "step": 442
+    },
+    {
+      "epoch": 0.8003613369467028,
+      "grad_norm": 0.2132598888952329,
+      "learning_rate": 5.3150914117997e-07,
+      "loss": 0.588,
+      "step": 443
+    },
+    {
+      "epoch": 0.8021680216802168,
+      "grad_norm": 0.19785474755013488,
+      "learning_rate": 5.223224133591475e-07,
+      "loss": 0.5938,
+      "step": 444
+    },
+    {
+      "epoch": 0.8039747064137308,
+      "grad_norm": 0.20562314317318503,
+      "learning_rate": 5.132065021811123e-07,
+      "loss": 0.5816,
+      "step": 445
+    },
+    {
+      "epoch": 0.8057813911472448,
+      "grad_norm": 0.19393590998324234,
+      "learning_rate": 5.041617340682467e-07,
+      "loss": 0.5785,
+      "step": 446
+    },
+    {
+      "epoch": 0.8075880758807588,
+      "grad_norm": 0.19561218339143732,
+      "learning_rate": 4.951884328954401e-07,
+      "loss": 0.5839,
+      "step": 447
+    },
+    {
+      "epoch": 0.8093947606142728,
+      "grad_norm": 0.19261102405426736,
+      "learning_rate": 4.862869199784984e-07,
+      "loss": 0.5864,
+      "step": 448
+    },
+    {
+      "epoch": 0.8112014453477868,
+      "grad_norm": 0.19722138182817417,
+      "learning_rate": 4.774575140626317e-07,
+      "loss": 0.582,
+      "step": 449
+    },
+    {
+      "epoch": 0.8130081300813008,
+      "grad_norm": 0.19905614668622285,
+      "learning_rate": 4.687005313110454e-07,
+      "loss": 0.5921,
+      "step": 450
+    },
+    {
+      "epoch": 0.8148148148148148,
+      "grad_norm": 0.2192978730560753,
+      "learning_rate": 4.600162852936171e-07,
+      "loss": 0.5947,
+      "step": 451
+    },
+    {
+      "epoch": 0.8166214995483289,
+      "grad_norm": 0.19931575730564813,
+      "learning_rate": 4.514050869756703e-07,
+      "loss": 0.5898,
+      "step": 452
+    },
+    {
+      "epoch": 0.8184281842818428,
+      "grad_norm": 0.19512336370042827,
+      "learning_rate": 4.4286724470683576e-07,
+      "loss": 0.5818,
+      "step": 453
+    },
+    {
+      "epoch": 0.8202348690153568,
+      "grad_norm": 0.19982404753283103,
+      "learning_rate": 4.344030642100133e-07,
+      "loss": 0.5664,
+      "step": 454
+    },
+    {
+      "epoch": 0.8220415537488708,
+      "grad_norm": 0.2075498736745964,
+      "learning_rate": 4.2601284857042263e-07,
+      "loss": 0.5905,
+      "step": 455
+    },
+    {
+      "epoch": 0.8238482384823849,
+      "grad_norm": 0.19838773426671638,
+      "learning_rate": 4.1769689822475147e-07,
+      "loss": 0.5819,
+      "step": 456
+    },
+    {
+      "epoch": 0.8256549232158988,
+      "grad_norm": 0.20470638453793452,
+      "learning_rate": 4.0945551095039837e-07,
+      "loss": 0.5965,
+      "step": 457
+    },
+    {
+      "epoch": 0.8274616079494128,
+      "grad_norm": 0.20178722345870734,
+      "learning_rate": 4.012889818548069e-07,
+      "loss": 0.5888,
+      "step": 458
+    },
+    {
+      "epoch": 0.8292682926829268,
+      "grad_norm": 0.18729573179360015,
+      "learning_rate": 3.931976033649021e-07,
+      "loss": 0.5782,
+      "step": 459
+    },
+    {
+      "epoch": 0.8310749774164409,
+      "grad_norm": 0.1986072934668412,
+      "learning_rate": 3.851816652166165e-07,
+      "loss": 0.5881,
+      "step": 460
+    },
+    {
+      "epoch": 0.8328816621499548,
+      "grad_norm": 0.1881841489141097,
+      "learning_rate": 3.772414544445163e-07,
+      "loss": 0.5756,
+      "step": 461
+    },
+    {
+      "epoch": 0.8346883468834688,
+      "grad_norm": 0.20510840467823352,
+      "learning_rate": 3.6937725537152277e-07,
+      "loss": 0.5831,
+      "step": 462
+    },
+    {
+      "epoch": 0.8364950316169828,
+      "grad_norm": 0.18584350117057763,
+      "learning_rate": 3.615893495987335e-07,
+      "loss": 0.5879,
+      "step": 463
+    },
+    {
+      "epoch": 0.8383017163504969,
+      "grad_norm": 0.19674763846332158,
+      "learning_rate": 3.538780159953348e-07,
+      "loss": 0.5902,
+      "step": 464
+    },
+    {
+      "epoch": 0.8401084010840109,
+      "grad_norm": 0.20468627530762953,
+      "learning_rate": 3.462435306886194e-07,
+      "loss": 0.5997,
+      "step": 465
+    },
+    {
+      "epoch": 0.8419150858175248,
+      "grad_norm": 0.20528912287709225,
+      "learning_rate": 3.3868616705409723e-07,
+      "loss": 0.5836,
+      "step": 466
+    },
+    {
+      "epoch": 0.8437217705510388,
+      "grad_norm": 0.27273003628867537,
+      "learning_rate": 3.312061957057061e-07,
+      "loss": 0.5799,
+      "step": 467
+    },
+    {
+      "epoch": 0.8455284552845529,
+      "grad_norm": 0.19053099912349022,
+      "learning_rate": 3.2380388448612437e-07,
+      "loss": 0.5905,
+      "step": 468
+    },
+    {
+      "epoch": 0.8473351400180669,
+      "grad_norm": 0.19085010418502601,
+      "learning_rate": 3.164794984571759e-07,
+      "loss": 0.5793,
+      "step": 469
+    },
+    {
+      "epoch": 0.8491418247515808,
+      "grad_norm": 0.20111905938166816,
+      "learning_rate": 3.092332998903416e-07,
+      "loss": 0.5749,
+      "step": 470
+    },
+    {
+      "epoch": 0.8509485094850948,
+      "grad_norm": 0.20358849876251972,
+      "learning_rate": 3.020655482573659e-07,
+      "loss": 0.6049,
+      "step": 471
+    },
+    {
+      "epoch": 0.8527551942186089,
+      "grad_norm": 0.2008916336129198,
+      "learning_rate": 2.949765002209698e-07,
+      "loss": 0.5761,
+      "step": 472
+    },
+    {
+      "epoch": 0.8545618789521229,
+      "grad_norm": 0.20435420909912377,
+      "learning_rate": 2.8796640962565374e-07,
+      "loss": 0.5955,
+      "step": 473
+    },
+    {
+      "epoch": 0.8563685636856369,
+      "grad_norm": 0.18287280066629497,
+      "learning_rate": 2.810355274886148e-07,
+      "loss": 0.5907,
+      "step": 474
+    },
+    {
+      "epoch": 0.8581752484191508,
+      "grad_norm": 0.20714481331888074,
+      "learning_rate": 2.7418410199075293e-07,
+      "loss": 0.5805,
+      "step": 475
+    },
+    {
+      "epoch": 0.8599819331526649,
+      "grad_norm": 0.19382788901021236,
+      "learning_rate": 2.674123784677868e-07,
+      "loss": 0.5874,
+      "step": 476
+    },
+    {
+      "epoch": 0.8617886178861789,
+      "grad_norm": 0.18523438801234704,
+      "learning_rate": 2.6072059940146775e-07,
+      "loss": 0.5836,
+      "step": 477
+    },
+    {
+      "epoch": 0.8635953026196929,
+      "grad_norm": 0.2028457298614708,
+      "learning_rate": 2.5410900441089905e-07,
+      "loss": 0.5953,
+      "step": 478
+    },
+    {
+      "epoch": 0.8654019873532068,
+      "grad_norm": 0.2078467522572559,
+      "learning_rate": 2.4757783024395244e-07,
+      "loss": 0.5816,
+      "step": 479
+    },
+    {
+      "epoch": 0.8672086720867209,
+      "grad_norm": 0.1873967534012521,
+      "learning_rate": 2.4112731076879254e-07,
+      "loss": 0.5798,
+      "step": 480
+    },
+    {
+      "epoch": 0.8690153568202349,
+      "grad_norm": 0.1933251862519231,
+      "learning_rate": 2.3475767696550327e-07,
+      "loss": 0.5894,
+      "step": 481
+    },
+    {
+      "epoch": 0.8708220415537489,
+      "grad_norm": 0.19064894480648312,
+      "learning_rate": 2.284691569178138e-07,
+      "loss": 0.5931,
+      "step": 482
+    },
+    {
+      "epoch": 0.8726287262872628,
+      "grad_norm": 0.2284657581088162,
+      "learning_rate": 2.2226197580493657e-07,
+      "loss": 0.5779,
+      "step": 483
+    },
+    {
+      "epoch": 0.8744354110207768,
+      "grad_norm": 0.19935241220998284,
+      "learning_rate": 2.1613635589349756e-07,
+      "loss": 0.5757,
+      "step": 484
+    },
+    {
+      "epoch": 0.8762420957542909,
+      "grad_norm": 0.2053137455836231,
+      "learning_rate": 2.100925165295839e-07,
+      "loss": 0.5848,
+      "step": 485
+    },
+    {
+      "epoch": 0.8780487804878049,
+      "grad_norm": 0.2010408803052494,
+      "learning_rate": 2.041306741308832e-07,
+      "loss": 0.5912,
+      "step": 486
+    },
+    {
+      "epoch": 0.8798554652213189,
+      "grad_norm": 0.19347960149543364,
+      "learning_rate": 1.982510421789402e-07,
+      "loss": 0.5787,
+      "step": 487
+    },
+    {
+      "epoch": 0.8816621499548328,
+      "grad_norm": 0.19351069706418417,
+      "learning_rate": 1.9245383121150678e-07,
+      "loss": 0.5782,
+      "step": 488
+    },
+    {
+      "epoch": 0.8834688346883469,
+      "grad_norm": 0.18822543167978809,
+      "learning_rate": 1.8673924881500826e-07,
+      "loss": 0.5875,
+      "step": 489
+    },
+    {
+      "epoch": 0.8852755194218609,
+      "grad_norm": 0.19257655016560987,
+      "learning_rate": 1.8110749961710582e-07,
+      "loss": 0.5843,
+      "step": 490
+    },
+    {
+      "epoch": 0.8870822041553749,
+      "grad_norm": 0.25925387426203955,
+      "learning_rate": 1.7555878527937164e-07,
+      "loss": 0.5762,
+      "step": 491
+    },
+    {
+      "epoch": 0.8888888888888888,
+      "grad_norm": 0.19553638949563604,
+      "learning_rate": 1.7009330449006712e-07,
+      "loss": 0.5781,
+      "step": 492
+    },
+    {
+      "epoch": 0.8906955736224029,
+      "grad_norm": 0.1945143019656124,
+      "learning_rate": 1.6471125295702773e-07,
+      "loss": 0.5721,
+      "step": 493
+    },
+    {
+      "epoch": 0.8925022583559169,
+      "grad_norm": 0.264819060676782,
+      "learning_rate": 1.59412823400657e-07,
+      "loss": 0.5894,
+      "step": 494
+    },
+    {
+      "epoch": 0.8943089430894309,
+      "grad_norm": 0.19083307342352715,
+      "learning_rate": 1.5419820554702313e-07,
+      "loss": 0.586,
+      "step": 495
+    },
+    {
+      "epoch": 0.8961156278229448,
+      "grad_norm": 0.32477290657900737,
+      "learning_rate": 1.4906758612106637e-07,
+      "loss": 0.599,
+      "step": 496
+    },
+    {
+      "epoch": 0.8979223125564589,
+      "grad_norm": 0.2446280764829226,
+      "learning_rate": 1.4402114883991318e-07,
+      "loss": 0.5858,
+      "step": 497
+    },
+    {
+      "epoch": 0.8997289972899729,
+      "grad_norm": 0.20615033714545855,
+      "learning_rate": 1.3905907440629752e-07,
+      "loss": 0.5847,
+      "step": 498
+    },
+    {
+      "epoch": 0.9015356820234869,
+      "grad_norm": 0.2087451994496376,
+      "learning_rate": 1.3418154050208937e-07,
+      "loss": 0.5948,
+      "step": 499
+    },
+    {
+      "epoch": 0.9033423667570009,
+      "grad_norm": 0.3050011108253672,
+      "learning_rate": 1.2938872178193395e-07,
+      "loss": 0.5923,
+      "step": 500
+    },
+    {
+      "epoch": 0.9051490514905149,
+      "grad_norm": 0.2832626023760989,
+      "learning_rate": 1.2468078986699633e-07,
+      "loss": 0.59,
+      "step": 501
+    },
+    {
+      "epoch": 0.9069557362240289,
+      "grad_norm": 0.1893248345236461,
+      "learning_rate": 1.200579133388155e-07,
+      "loss": 0.586,
+      "step": 502
+    },
+    {
+      "epoch": 0.9087624209575429,
+      "grad_norm": 0.18631095112826992,
+      "learning_rate": 1.1552025773327008e-07,
+      "loss": 0.5814,
+      "step": 503
+    },
+    {
+      "epoch": 0.9105691056910569,
+      "grad_norm": 0.1825244758133897,
+      "learning_rate": 1.1106798553464804e-07,
+      "loss": 0.5969,
+      "step": 504
+    },
+    {
+      "epoch": 0.912375790424571,
+      "grad_norm": 0.19272725378716518,
+      "learning_rate": 1.067012561698319e-07,
+      "loss": 0.5681,
+      "step": 505
+    },
+    {
+      "epoch": 0.9141824751580849,
+      "grad_norm": 0.18360862506223993,
+      "learning_rate": 1.0242022600258611e-07,
+      "loss": 0.5887,
+      "step": 506
+    },
+    {
+      "epoch": 0.9159891598915989,
+      "grad_norm": 0.21234881666515093,
+      "learning_rate": 9.822504832796037e-08,
+      "loss": 0.5876,
+      "step": 507
+    },
+    {
+      "epoch": 0.9177958446251129,
+      "grad_norm": 0.19196260234968993,
+      "learning_rate": 9.41158733667999e-08,
+      "loss": 0.58,
+      "step": 508
+    },
+    {
+      "epoch": 0.919602529358627,
+      "grad_norm": 0.1969153672664882,
+      "learning_rate": 9.00928482603669e-08,
+      "loss": 0.5997,
+      "step": 509
+    },
+    {
+      "epoch": 0.9214092140921409,
+      "grad_norm": 0.1945863429430772,
+      "learning_rate": 8.615611706507043e-08,
+      "loss": 0.5827,
+      "step": 510
+    },
+    {
+      "epoch": 0.9232158988256549,
+      "grad_norm": 0.23778461680575155,
+      "learning_rate": 8.230582074730902e-08,
+      "loss": 0.573,
+      "step": 511
+    },
+    {
+      "epoch": 0.9250225835591689,
+      "grad_norm": 0.20332757728437817,
+      "learning_rate": 7.854209717842231e-08,
+      "loss": 0.6005,
+      "step": 512
+    },
+    {
+      "epoch": 0.926829268292683,
+      "grad_norm": 0.19387134731372843,
+      "learning_rate": 7.486508112975549e-08,
+      "loss": 0.5833,
+      "step": 513
+    },
+    {
+      "epoch": 0.928635953026197,
+      "grad_norm": 0.19782012258453638,
+      "learning_rate": 7.127490426783124e-08,
+      "loss": 0.5797,
+      "step": 514
+    },
+    {
+      "epoch": 0.9304426377597109,
+      "grad_norm": 0.19286615430131876,
+      "learning_rate": 6.777169514963766e-08,
+      "loss": 0.5844,
+      "step": 515
+    },
+    {
+      "epoch": 0.9322493224932249,
+      "grad_norm": 0.19637869096966293,
+      "learning_rate": 6.435557921802254e-08,
+      "loss": 0.5924,
+      "step": 516
+    },
+    {
+      "epoch": 0.934056007226739,
+      "grad_norm": 0.188554049011414,
+      "learning_rate": 6.102667879720165e-08,
+      "loss": 0.5799,
+      "step": 517
+    },
+    {
+      "epoch": 0.935862691960253,
+      "grad_norm": 0.20883892103413773,
+      "learning_rate": 5.778511308838108e-08,
+      "loss": 0.5829,
+      "step": 518
+    },
+    {
+      "epoch": 0.9376693766937669,
+      "grad_norm": 0.20437240656846004,
+      "learning_rate": 5.463099816548578e-08,
+      "loss": 0.5717,
+      "step": 519
+    },
+    {
+      "epoch": 0.9394760614272809,
+      "grad_norm": 0.19852353971692513,
+      "learning_rate": 5.15644469710061e-08,
+      "loss": 0.582,
+      "step": 520
+    },
+    {
+      "epoch": 0.941282746160795,
+      "grad_norm": 0.18095323109737302,
+      "learning_rate": 4.8585569311949966e-08,
+      "loss": 0.5889,
+      "step": 521
+    },
+    {
+      "epoch": 0.943089430894309,
+      "grad_norm": 0.18174076650377002,
+      "learning_rate": 4.5694471855914915e-08,
+      "loss": 0.5838,
+      "step": 522
+    },
+    {
+      "epoch": 0.9448961156278229,
+      "grad_norm": 0.18479803288886815,
+      "learning_rate": 4.289125812726475e-08,
+      "loss": 0.5748,
+      "step": 523
+    },
+    {
+      "epoch": 0.9467028003613369,
+      "grad_norm": 0.19193590197592694,
+      "learning_rate": 4.017602850342584e-08,
+      "loss": 0.5859,
+      "step": 524
+    },
+    {
+      "epoch": 0.948509485094851,
+      "grad_norm": 0.19470238994455222,
+      "learning_rate": 3.754888021129083e-08,
+      "loss": 0.5788,
+      "step": 525
+    },
+    {
+      "epoch": 0.950316169828365,
+      "grad_norm": 0.18966253565903646,
+      "learning_rate": 3.5009907323737826e-08,
+      "loss": 0.5839,
+      "step": 526
+    },
+    {
+      "epoch": 0.952122854561879,
+      "grad_norm": 0.20275158347015088,
+      "learning_rate": 3.2559200756260845e-08,
+      "loss": 0.5832,
+      "step": 527
+    },
+    {
+      "epoch": 0.9539295392953929,
+      "grad_norm": 0.18676114511717182,
+      "learning_rate": 3.019684826371633e-08,
+      "loss": 0.5801,
+      "step": 528
+    },
+    {
+      "epoch": 0.955736224028907,
+      "grad_norm": 0.2163227342300819,
+      "learning_rate": 2.7922934437178695e-08,
+      "loss": 0.5883,
+      "step": 529
+    },
+    {
+      "epoch": 0.957542908762421,
+      "grad_norm": 0.21344582125626332,
+      "learning_rate": 2.5737540700912777e-08,
+      "loss": 0.5813,
+      "step": 530
+    },
+    {
+      "epoch": 0.959349593495935,
+      "grad_norm": 0.18607944951873648,
+      "learning_rate": 2.3640745309457258e-08,
+      "loss": 0.5923,
+      "step": 531
+    },
+    {
+      "epoch": 0.9611562782294489,
+      "grad_norm": 0.18510297216059776,
+      "learning_rate": 2.16326233448233e-08,
+      "loss": 0.5807,
+      "step": 532
+    },
+    {
+      "epoch": 0.9629629629629629,
+      "grad_norm": 0.1974938236004036,
+      "learning_rate": 1.9713246713805588e-08,
+      "loss": 0.5858,
+      "step": 533
+    },
+    {
+      "epoch": 0.964769647696477,
+      "grad_norm": 0.20144502512514245,
+      "learning_rate": 1.7882684145406616e-08,
+      "loss": 0.5825,
+      "step": 534
+    },
+    {
+      "epoch": 0.966576332429991,
+      "grad_norm": 0.1905225782300761,
+      "learning_rate": 1.6141001188378082e-08,
+      "loss": 0.5827,
+      "step": 535
+    },
+    {
+      "epoch": 0.9683830171635049,
+      "grad_norm": 0.19016394964817498,
+      "learning_rate": 1.4488260208871397e-08,
+      "loss": 0.5929,
+      "step": 536
+    },
+    {
+      "epoch": 0.9701897018970189,
+      "grad_norm": 0.18423294596460907,
+      "learning_rate": 1.2924520388204465e-08,
+      "loss": 0.5987,
+      "step": 537
+    },
+    {
+      "epoch": 0.971996386630533,
+      "grad_norm": 0.20028347781639222,
+      "learning_rate": 1.1449837720745039e-08,
+      "loss": 0.5817,
+      "step": 538
+    },
+    {
+      "epoch": 0.973803071364047,
+      "grad_norm": 0.19564562368726848,
+      "learning_rate": 1.006426501190233e-08,
+      "loss": 0.5784,
+      "step": 539
+    },
+    {
+      "epoch": 0.975609756097561,
+      "grad_norm": 0.18942355645645967,
+      "learning_rate": 8.767851876239075e-09,
+      "loss": 0.5796,
+      "step": 540
+    },
+    {
+      "epoch": 0.9774164408310749,
+      "grad_norm": 0.1872475487831209,
+      "learning_rate": 7.560644735692957e-09,
+      "loss": 0.5884,
+      "step": 541
+    },
+    {
+      "epoch": 0.979223125564589,
+      "grad_norm": 0.199499510224137,
+      "learning_rate": 6.442686817914878e-09,
+      "loss": 0.5805,
+      "step": 542
+    },
+    {
+      "epoch": 0.981029810298103,
+      "grad_norm": 0.18125890965998775,
+      "learning_rate": 5.414018154721868e-09,
+      "loss": 0.5935,
+      "step": 543
+    },
+    {
+      "epoch": 0.982836495031617,
+      "grad_norm": 0.21667359886815266,
+      "learning_rate": 4.474675580662113e-09,
+      "loss": 0.593,
+      "step": 544
+    },
+    {
+      "epoch": 0.9846431797651309,
+      "grad_norm": 0.19857918487394038,
+      "learning_rate": 3.6246927316976875e-09,
+      "loss": 0.5964,
+      "step": 545
+    },
+    {
+      "epoch": 0.986449864498645,
+      "grad_norm": 0.1929371434279703,
+      "learning_rate": 2.864100043998563e-09,
+      "loss": 0.5862,
+      "step": 546
+    },
+    {
+      "epoch": 0.988256549232159,
+      "grad_norm": 0.18917095075953314,
+      "learning_rate": 2.192924752854042e-09,
+      "loss": 0.5983,
+      "step": 547
+    },
+    {
+      "epoch": 0.990063233965673,
+      "grad_norm": 0.19052822059983585,
+      "learning_rate": 1.6111908916965902e-09,
+      "loss": 0.5948,
+      "step": 548
+    },
+    {
+      "epoch": 0.991869918699187,
+      "grad_norm": 0.20501303397618248,
+      "learning_rate": 1.1189192912416935e-09,
+      "loss": 0.5892,
+      "step": 549
+    },
+    {
+      "epoch": 0.993676603432701,
+      "grad_norm": 0.19753256710279912,
+      "learning_rate": 7.161275787415101e-10,
+      "loss": 0.5795,
+      "step": 550
+    },
+    {
+      "epoch": 0.995483288166215,
+      "grad_norm": 0.1882439395222658,
+      "learning_rate": 4.0283017735454066e-10,
+      "loss": 0.5983,
+      "step": 551
+    },
+    {
+      "epoch": 0.997289972899729,
+      "grad_norm": 0.19218850747751368,
+      "learning_rate": 1.790383056282652e-10,
+      "loss": 0.5806,
+      "step": 552
+    },
+    {
+      "epoch": 0.999096657633243,
+      "grad_norm": 0.19150393638935695,
+      "learning_rate": 4.4759977098074445e-11,
+      "loss": 0.5844,
+      "step": 553
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 553,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 100,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 641050695434240.0,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}