diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,7328 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.66688,
+  "eval_steps": 500,
+  "global_step": 1042,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.00064,
+      "grad_norm": 2.353584373604404,
+      "learning_rate": 0.0,
+      "loss": 0.5764,
+      "step": 1
+    },
+    {
+      "epoch": 0.00128,
+      "grad_norm": 2.5677274470706184,
+      "learning_rate": 3.79746835443038e-07,
+      "loss": 0.6154,
+      "step": 2
+    },
+    {
+      "epoch": 0.00192,
+      "grad_norm": 2.434815038353528,
+      "learning_rate": 7.59493670886076e-07,
+      "loss": 0.5762,
+      "step": 3
+    },
+    {
+      "epoch": 0.00256,
+      "grad_norm": 2.4265789349847133,
+      "learning_rate": 1.139240506329114e-06,
+      "loss": 0.5898,
+      "step": 4
+    },
+    {
+      "epoch": 0.0032,
+      "grad_norm": 2.4968640172692007,
+      "learning_rate": 1.518987341772152e-06,
+      "loss": 0.5854,
+      "step": 5
+    },
+    {
+      "epoch": 0.00384,
+      "grad_norm": 2.6270464110646,
+      "learning_rate": 1.8987341772151901e-06,
+      "loss": 0.6063,
+      "step": 6
+    },
+    {
+      "epoch": 0.00448,
+      "grad_norm": 2.3240861061539664,
+      "learning_rate": 2.278481012658228e-06,
+      "loss": 0.5713,
+      "step": 7
+    },
+    {
+      "epoch": 0.00512,
+      "grad_norm": 2.277096457178879,
+      "learning_rate": 2.6582278481012658e-06,
+      "loss": 0.5981,
+      "step": 8
+    },
+    {
+      "epoch": 0.00576,
+      "grad_norm": 2.0556904877285276,
+      "learning_rate": 3.037974683544304e-06,
+      "loss": 0.5783,
+      "step": 9
+    },
+    {
+      "epoch": 0.0064,
+      "grad_norm": 1.8883532047436715,
+      "learning_rate": 3.4177215189873417e-06,
+      "loss": 0.5626,
+      "step": 10
+    },
+    {
+      "epoch": 0.00704,
+      "grad_norm": 1.4054834462317158,
+      "learning_rate": 3.7974683544303802e-06,
+      "loss": 0.5762,
+      "step": 11
+    },
+    {
+      "epoch": 0.00768,
+      "grad_norm": 1.2672259034804543,
+      "learning_rate": 4.1772151898734175e-06,
+      "loss": 0.5447,
+      "step": 12
+    },
+    {
+      "epoch": 0.00832,
+      "grad_norm": 1.1315187874621562,
+      "learning_rate": 4.556962025316456e-06,
+      "loss": 0.5003,
+      "step": 13
+    },
+    {
+      "epoch": 0.00896,
+      "grad_norm": 1.2442008617105749,
+      "learning_rate": 4.936708860759494e-06,
+      "loss": 0.5235,
+      "step": 14
+    },
+    {
+      "epoch": 0.0096,
+      "grad_norm": 1.1708354130739331,
+      "learning_rate": 5.3164556962025316e-06,
+      "loss": 0.4659,
+      "step": 15
+    },
+    {
+      "epoch": 0.01024,
+      "grad_norm": 1.3068048804077694,
+      "learning_rate": 5.69620253164557e-06,
+      "loss": 0.4601,
+      "step": 16
+    },
+    {
+      "epoch": 0.01088,
+      "grad_norm": 1.487196594526998,
+      "learning_rate": 6.075949367088608e-06,
+      "loss": 0.487,
+      "step": 17
+    },
+    {
+      "epoch": 0.01152,
+      "grad_norm": 1.3702547576006188,
+      "learning_rate": 6.455696202531646e-06,
+      "loss": 0.4829,
+      "step": 18
+    },
+    {
+      "epoch": 0.01216,
+      "grad_norm": 0.985500593752308,
+      "learning_rate": 6.835443037974683e-06,
+      "loss": 0.4463,
+      "step": 19
+    },
+    {
+      "epoch": 0.0128,
+      "grad_norm": 0.9909916347896555,
+      "learning_rate": 7.215189873417722e-06,
+      "loss": 0.4554,
+      "step": 20
+    },
+    {
+      "epoch": 0.01344,
+      "grad_norm": 1.1838910835370198,
+      "learning_rate": 7.5949367088607605e-06,
+      "loss": 0.4622,
+      "step": 21
+    },
+    {
+      "epoch": 0.01408,
+      "grad_norm": 1.6876677020755413,
+      "learning_rate": 7.974683544303797e-06,
+      "loss": 0.3952,
+      "step": 22
+    },
+    {
+      "epoch": 0.01472,
+      "grad_norm": 1.0417104358020024,
+      "learning_rate": 8.354430379746835e-06,
+      "loss": 0.4105,
+      "step": 23
+    },
+    {
+      "epoch": 0.01536,
+      "grad_norm": 0.703419627437881,
+      "learning_rate": 8.734177215189873e-06,
+      "loss": 0.3797,
+      "step": 24
+    },
+    {
+      "epoch": 0.016,
+      "grad_norm": 0.7189072691985104,
+      "learning_rate": 9.113924050632912e-06,
+      "loss": 0.4206,
+      "step": 25
+    },
+    {
+      "epoch": 0.01664,
+      "grad_norm": 0.7280445280679197,
+      "learning_rate": 9.49367088607595e-06,
+      "loss": 0.3963,
+      "step": 26
+    },
+    {
+      "epoch": 0.01728,
+      "grad_norm": 0.5914861117650524,
+      "learning_rate": 9.873417721518988e-06,
+      "loss": 0.3928,
+      "step": 27
+    },
+    {
+      "epoch": 0.01792,
+      "grad_norm": 0.6824922709052414,
+      "learning_rate": 1.0253164556962025e-05,
+      "loss": 0.3909,
+      "step": 28
+    },
+    {
+      "epoch": 0.01856,
+      "grad_norm": 0.6157178643639211,
+      "learning_rate": 1.0632911392405063e-05,
+      "loss": 0.3778,
+      "step": 29
+    },
+    {
+      "epoch": 0.0192,
+      "grad_norm": 0.7040224694990468,
+      "learning_rate": 1.1012658227848103e-05,
+      "loss": 0.4034,
+      "step": 30
+    },
+    {
+      "epoch": 0.01984,
+      "grad_norm": 0.6042406871448358,
+      "learning_rate": 1.139240506329114e-05,
+      "loss": 0.3991,
+      "step": 31
+    },
+    {
+      "epoch": 0.02048,
+      "grad_norm": 0.5581052443719693,
+      "learning_rate": 1.1772151898734176e-05,
+      "loss": 0.4011,
+      "step": 32
+    },
+    {
+      "epoch": 0.02112,
+      "grad_norm": 0.4266761178757486,
+      "learning_rate": 1.2151898734177216e-05,
+      "loss": 0.3639,
+      "step": 33
+    },
+    {
+      "epoch": 0.02176,
+      "grad_norm": 0.49600505321594285,
+      "learning_rate": 1.2531645569620253e-05,
+      "loss": 0.3784,
+      "step": 34
+    },
+    {
+      "epoch": 0.0224,
+      "grad_norm": 0.45932753795907266,
+      "learning_rate": 1.2911392405063291e-05,
+      "loss": 0.3751,
+      "step": 35
+    },
+    {
+      "epoch": 0.02304,
+      "grad_norm": 0.5363867843128812,
+      "learning_rate": 1.3291139240506329e-05,
+      "loss": 0.3865,
+      "step": 36
+    },
+    {
+      "epoch": 0.02368,
+      "grad_norm": 0.5488419069165299,
+      "learning_rate": 1.3670886075949367e-05,
+      "loss": 0.3787,
+      "step": 37
+    },
+    {
+      "epoch": 0.02432,
+      "grad_norm": 0.48587043713463085,
+      "learning_rate": 1.4050632911392406e-05,
+      "loss": 0.3712,
+      "step": 38
+    },
+    {
+      "epoch": 0.02496,
+      "grad_norm": 0.4440990518273777,
+      "learning_rate": 1.4430379746835444e-05,
+      "loss": 0.3652,
+      "step": 39
+    },
+    {
+      "epoch": 0.0256,
+      "grad_norm": 0.3751498370487217,
+      "learning_rate": 1.4810126582278482e-05,
+      "loss": 0.3569,
+      "step": 40
+    },
+    {
+      "epoch": 0.02624,
+      "grad_norm": 0.43725343772159886,
+      "learning_rate": 1.5189873417721521e-05,
+      "loss": 0.3158,
+      "step": 41
+    },
+    {
+      "epoch": 0.02688,
+      "grad_norm": 0.4888026864502293,
+      "learning_rate": 1.5569620253164557e-05,
+      "loss": 0.3442,
+      "step": 42
+    },
+    {
+      "epoch": 0.02752,
+      "grad_norm": 0.44586123200596317,
+      "learning_rate": 1.5949367088607595e-05,
+      "loss": 0.3328,
+      "step": 43
+    },
+    {
+      "epoch": 0.02816,
+      "grad_norm": 0.49767798630217713,
+      "learning_rate": 1.6329113924050632e-05,
+      "loss": 0.351,
+      "step": 44
+    },
+    {
+      "epoch": 0.0288,
+      "grad_norm": 0.4336104979866255,
+      "learning_rate": 1.670886075949367e-05,
+      "loss": 0.3612,
+      "step": 45
+    },
+    {
+      "epoch": 0.02944,
+      "grad_norm": 0.4981279303284916,
+      "learning_rate": 1.708860759493671e-05,
+      "loss": 0.3543,
+      "step": 46
+    },
+    {
+      "epoch": 0.03008,
+      "grad_norm": 0.3991975649542343,
+      "learning_rate": 1.7468354430379746e-05,
+      "loss": 0.346,
+      "step": 47
+    },
+    {
+      "epoch": 0.03072,
+      "grad_norm": 0.4340035974965301,
+      "learning_rate": 1.7848101265822783e-05,
+      "loss": 0.3563,
+      "step": 48
+    },
+    {
+      "epoch": 0.03136,
+      "grad_norm": 0.40441725662186223,
+      "learning_rate": 1.8227848101265824e-05,
+      "loss": 0.3378,
+      "step": 49
+    },
+    {
+      "epoch": 0.032,
+      "grad_norm": 0.4015157764097668,
+      "learning_rate": 1.860759493670886e-05,
+      "loss": 0.3477,
+      "step": 50
+    },
+    {
+      "epoch": 0.03264,
+      "grad_norm": 0.3897529365297187,
+      "learning_rate": 1.89873417721519e-05,
+      "loss": 0.3341,
+      "step": 51
+    },
+    {
+      "epoch": 0.03328,
+      "grad_norm": 0.39864724195969276,
+      "learning_rate": 1.9367088607594938e-05,
+      "loss": 0.3322,
+      "step": 52
+    },
+    {
+      "epoch": 0.03392,
+      "grad_norm": 0.365402956200434,
+      "learning_rate": 1.9746835443037975e-05,
+      "loss": 0.297,
+      "step": 53
+    },
+    {
+      "epoch": 0.03456,
+      "grad_norm": 0.4449097185367681,
+      "learning_rate": 2.0126582278481013e-05,
+      "loss": 0.3424,
+      "step": 54
+    },
+    {
+      "epoch": 0.0352,
+      "grad_norm": 0.43181805843495463,
+      "learning_rate": 2.050632911392405e-05,
+      "loss": 0.3351,
+      "step": 55
+    },
+    {
+      "epoch": 0.03584,
+      "grad_norm": 0.392546267080926,
+      "learning_rate": 2.0886075949367092e-05,
+      "loss": 0.334,
+      "step": 56
+    },
+    {
+      "epoch": 0.03648,
+      "grad_norm": 0.38060689632546174,
+      "learning_rate": 2.1265822784810126e-05,
+      "loss": 0.3328,
+      "step": 57
+    },
+    {
+      "epoch": 0.03712,
+      "grad_norm": 0.389736897725578,
+      "learning_rate": 2.1645569620253164e-05,
+      "loss": 0.351,
+      "step": 58
+    },
+    {
+      "epoch": 0.03776,
+      "grad_norm": 0.42262406549661324,
+      "learning_rate": 2.2025316455696205e-05,
+      "loss": 0.3378,
+      "step": 59
+    },
+    {
+      "epoch": 0.0384,
+      "grad_norm": 0.43235384034762964,
+      "learning_rate": 2.240506329113924e-05,
+      "loss": 0.3308,
+      "step": 60
+    },
+    {
+      "epoch": 0.03904,
+      "grad_norm": 0.3945507271925106,
+      "learning_rate": 2.278481012658228e-05,
+      "loss": 0.3388,
+      "step": 61
+    },
+    {
+      "epoch": 0.03968,
+      "grad_norm": 0.40849904164663575,
+      "learning_rate": 2.3164556962025318e-05,
+      "loss": 0.3191,
+      "step": 62
+    },
+    {
+      "epoch": 0.04032,
+      "grad_norm": 0.4559537771952015,
+      "learning_rate": 2.3544303797468353e-05,
+      "loss": 0.3245,
+      "step": 63
+    },
+    {
+      "epoch": 0.04096,
+      "grad_norm": 0.40069809350325963,
+      "learning_rate": 2.3924050632911394e-05,
+      "loss": 0.3566,
+      "step": 64
+    },
+    {
+      "epoch": 0.0416,
+      "grad_norm": 0.37133926498283504,
+      "learning_rate": 2.430379746835443e-05,
+      "loss": 0.3147,
+      "step": 65
+    },
+    {
+      "epoch": 0.04224,
+      "grad_norm": 0.4628360444788038,
+      "learning_rate": 2.468354430379747e-05,
+      "loss": 0.3498,
+      "step": 66
+    },
+    {
+      "epoch": 0.04288,
+      "grad_norm": 0.38282967415212005,
+      "learning_rate": 2.5063291139240507e-05,
+      "loss": 0.3171,
+      "step": 67
+    },
+    {
+      "epoch": 0.04352,
+      "grad_norm": 0.4221420372731314,
+      "learning_rate": 2.5443037974683545e-05,
+      "loss": 0.3288,
+      "step": 68
+    },
+    {
+      "epoch": 0.04416,
+      "grad_norm": 0.4014634593404327,
+      "learning_rate": 2.5822784810126582e-05,
+      "loss": 0.2967,
+      "step": 69
+    },
+    {
+      "epoch": 0.0448,
+      "grad_norm": 0.4226874979052922,
+      "learning_rate": 2.620253164556962e-05,
+      "loss": 0.3311,
+      "step": 70
+    },
+    {
+      "epoch": 0.04544,
+      "grad_norm": 0.3691551141136815,
+      "learning_rate": 2.6582278481012658e-05,
+      "loss": 0.3154,
+      "step": 71
+    },
+    {
+      "epoch": 0.04608,
+      "grad_norm": 0.4930935215298506,
+      "learning_rate": 2.69620253164557e-05,
+      "loss": 0.3248,
+      "step": 72
+    },
+    {
+      "epoch": 0.04672,
+      "grad_norm": 0.40344283789369767,
+      "learning_rate": 2.7341772151898733e-05,
+      "loss": 0.3146,
+      "step": 73
+    },
+    {
+      "epoch": 0.04736,
+      "grad_norm": 0.40136750473831273,
+      "learning_rate": 2.7721518987341774e-05,
+      "loss": 0.3084,
+      "step": 74
+    },
+    {
+      "epoch": 0.048,
+      "grad_norm": 0.433002156207923,
+      "learning_rate": 2.8101265822784812e-05,
+      "loss": 0.2992,
+      "step": 75
+    },
+    {
+      "epoch": 0.04864,
+      "grad_norm": 0.45976406035512724,
+      "learning_rate": 2.8481012658227846e-05,
+      "loss": 0.3203,
+      "step": 76
+    },
+    {
+      "epoch": 0.04928,
+      "grad_norm": 0.46555202230017373,
+      "learning_rate": 2.8860759493670888e-05,
+      "loss": 0.326,
+      "step": 77
+    },
+    {
+      "epoch": 0.04992,
+      "grad_norm": 0.40061810400853354,
+      "learning_rate": 2.9240506329113925e-05,
+      "loss": 0.2902,
+      "step": 78
+    },
+    {
+      "epoch": 0.05056,
+      "grad_norm": 0.44228162263369347,
+      "learning_rate": 2.9620253164556963e-05,
+      "loss": 0.3235,
+      "step": 79
+    },
+    {
+      "epoch": 0.0512,
+      "grad_norm": 0.4587098467220033,
+      "learning_rate": 3e-05,
+      "loss": 0.3135,
+      "step": 80
+    },
+    {
+      "epoch": 0.05184,
+      "grad_norm": 0.42643590142876026,
+      "learning_rate": 2.9999966342756535e-05,
+      "loss": 0.3222,
+      "step": 81
+    },
+    {
+      "epoch": 0.05248,
+      "grad_norm": 0.4312879132975502,
+      "learning_rate": 2.9999865371177178e-05,
+      "loss": 0.3161,
+      "step": 82
+    },
+    {
+      "epoch": 0.05312,
+      "grad_norm": 0.44554806900527943,
+      "learning_rate": 2.9999697085715054e-05,
+      "loss": 0.3156,
+      "step": 83
+    },
+    {
+      "epoch": 0.05376,
+      "grad_norm": 0.4388949943183408,
+      "learning_rate": 2.9999461487125358e-05,
+      "loss": 0.3278,
+      "step": 84
+    },
+    {
+      "epoch": 0.0544,
+      "grad_norm": 0.4830420761548017,
+      "learning_rate": 2.999915857646538e-05,
+      "loss": 0.3242,
+      "step": 85
+    },
+    {
+      "epoch": 0.05504,
+      "grad_norm": 0.42290888303690005,
+      "learning_rate": 2.9998788355094472e-05,
+      "loss": 0.3051,
+      "step": 86
+    },
+    {
+      "epoch": 0.05568,
+      "grad_norm": 0.39931894559101605,
+      "learning_rate": 2.9998350824674046e-05,
+      "loss": 0.3174,
+      "step": 87
+    },
+    {
+      "epoch": 0.05632,
+      "grad_norm": 0.43885002522435895,
+      "learning_rate": 2.999784598716758e-05,
+      "loss": 0.3167,
+      "step": 88
+    },
+    {
+      "epoch": 0.05696,
+      "grad_norm": 0.4385666955310095,
+      "learning_rate": 2.9997273844840597e-05,
+      "loss": 0.32,
+      "step": 89
+    },
+    {
+      "epoch": 0.0576,
+      "grad_norm": 0.4241668561559285,
+      "learning_rate": 2.9996634400260665e-05,
+      "loss": 0.339,
+      "step": 90
+    },
+    {
+      "epoch": 0.05824,
+      "grad_norm": 0.49455156274720913,
+      "learning_rate": 2.9995927656297376e-05,
+      "loss": 0.3135,
+      "step": 91
+    },
+    {
+      "epoch": 0.05888,
+      "grad_norm": 0.3990302088629719,
+      "learning_rate": 2.9995153616122335e-05,
+      "loss": 0.3026,
+      "step": 92
+    },
+    {
+      "epoch": 0.05952,
+      "grad_norm": 0.38959449381416456,
+      "learning_rate": 2.9994312283209147e-05,
+      "loss": 0.2898,
+      "step": 93
+    },
+    {
+      "epoch": 0.06016,
+      "grad_norm": 0.42585557119259293,
+      "learning_rate": 2.9993403661333413e-05,
+      "loss": 0.3021,
+      "step": 94
+    },
+    {
+      "epoch": 0.0608,
+      "grad_norm": 0.37467889420763917,
+      "learning_rate": 2.9992427754572692e-05,
+      "loss": 0.273,
+      "step": 95
+    },
+    {
+      "epoch": 0.06144,
+      "grad_norm": 0.4466178048330342,
+      "learning_rate": 2.9991384567306485e-05,
+      "loss": 0.3116,
+      "step": 96
+    },
+    {
+      "epoch": 0.06208,
+      "grad_norm": 0.4133406435053229,
+      "learning_rate": 2.9990274104216246e-05,
+      "loss": 0.3066,
+      "step": 97
+    },
+    {
+      "epoch": 0.06272,
+      "grad_norm": 0.42703847116501487,
+      "learning_rate": 2.9989096370285314e-05,
+      "loss": 0.3029,
+      "step": 98
+    },
+    {
+      "epoch": 0.06336,
+      "grad_norm": 0.42408773456463106,
+      "learning_rate": 2.9987851370798936e-05,
+      "loss": 0.3302,
+      "step": 99
+    },
+    {
+      "epoch": 0.064,
+      "grad_norm": 0.36334101776279903,
+      "learning_rate": 2.998653911134421e-05,
+      "loss": 0.2749,
+      "step": 100
+    },
+    {
+      "epoch": 0.06464,
+      "grad_norm": 0.39914472077814156,
+      "learning_rate": 2.9985159597810067e-05,
+      "loss": 0.28,
+      "step": 101
+    },
+    {
+      "epoch": 0.06528,
+      "grad_norm": 0.4421338433902218,
+      "learning_rate": 2.9983712836387263e-05,
+      "loss": 0.3284,
+      "step": 102
+    },
+    {
+      "epoch": 0.06592,
+      "grad_norm": 0.4681114190701383,
+      "learning_rate": 2.9982198833568332e-05,
+      "loss": 0.3106,
+      "step": 103
+    },
+    {
+      "epoch": 0.06656,
+      "grad_norm": 0.43495699820223616,
+      "learning_rate": 2.998061759614756e-05,
+      "loss": 0.3034,
+      "step": 104
+    },
+    {
+      "epoch": 0.0672,
+      "grad_norm": 0.48523982047419434,
+      "learning_rate": 2.9978969131220956e-05,
+      "loss": 0.3199,
+      "step": 105
+    },
+    {
+      "epoch": 0.06784,
+      "grad_norm": 0.46970443054055944,
+      "learning_rate": 2.9977253446186236e-05,
+      "loss": 0.3145,
+      "step": 106
+    },
+    {
+      "epoch": 0.06848,
+      "grad_norm": 0.3762833085351204,
+      "learning_rate": 2.997547054874275e-05,
+      "loss": 0.2784,
+      "step": 107
+    },
+    {
+      "epoch": 0.06912,
+      "grad_norm": 0.4018479556059187,
+      "learning_rate": 2.9973620446891495e-05,
+      "loss": 0.29,
+      "step": 108
+    },
+    {
+      "epoch": 0.06976,
+      "grad_norm": 0.38557801552857196,
+      "learning_rate": 2.997170314893504e-05,
+      "loss": 0.2973,
+      "step": 109
+    },
+    {
+      "epoch": 0.0704,
+      "grad_norm": 0.42760504312755043,
+      "learning_rate": 2.9969718663477524e-05,
+      "loss": 0.3087,
+      "step": 110
+    },
+    {
+      "epoch": 0.07104,
+      "grad_norm": 0.4336880764497262,
+      "learning_rate": 2.9967666999424588e-05,
+      "loss": 0.3202,
+      "step": 111
+    },
+    {
+      "epoch": 0.07168,
+      "grad_norm": 0.4361380772072425,
+      "learning_rate": 2.9965548165983337e-05,
+      "loss": 0.2858,
+      "step": 112
+    },
+    {
+      "epoch": 0.07232,
+      "grad_norm": 0.36094868753422277,
+      "learning_rate": 2.9963362172662324e-05,
+      "loss": 0.2965,
+      "step": 113
+    },
+    {
+      "epoch": 0.07296,
+      "grad_norm": 0.5213407516270206,
+      "learning_rate": 2.9961109029271478e-05,
+      "loss": 0.2877,
+      "step": 114
+    },
+    {
+      "epoch": 0.0736,
+      "grad_norm": 0.39899966857979996,
+      "learning_rate": 2.9958788745922088e-05,
+      "loss": 0.2967,
+      "step": 115
+    },
+    {
+      "epoch": 0.07424,
+      "grad_norm": 0.4922285856357075,
+      "learning_rate": 2.995640133302672e-05,
+      "loss": 0.3135,
+      "step": 116
+    },
+    {
+      "epoch": 0.07488,
+      "grad_norm": 0.3840683744534582,
+      "learning_rate": 2.9953946801299213e-05,
+      "loss": 0.3021,
+      "step": 117
+    },
+    {
+      "epoch": 0.07552,
+      "grad_norm": 0.4183493778424685,
+      "learning_rate": 2.995142516175461e-05,
+      "loss": 0.307,
+      "step": 118
+    },
+    {
+      "epoch": 0.07616,
+      "grad_norm": 0.35220612158537656,
+      "learning_rate": 2.994883642570909e-05,
+      "loss": 0.2702,
+      "step": 119
+    },
+    {
+      "epoch": 0.0768,
+      "grad_norm": 0.3936594224655844,
+      "learning_rate": 2.9946180604779952e-05,
+      "loss": 0.2978,
+      "step": 120
+    },
+    {
+      "epoch": 0.07744,
+      "grad_norm": 0.4020931932900933,
+      "learning_rate": 2.9943457710885548e-05,
+      "loss": 0.3115,
+      "step": 121
+    },
+    {
+      "epoch": 0.07808,
+      "grad_norm": 0.413807873620709,
+      "learning_rate": 2.9940667756245218e-05,
+      "loss": 0.3084,
+      "step": 122
+    },
+    {
+      "epoch": 0.07872,
+      "grad_norm": 0.3861233473719098,
+      "learning_rate": 2.9937810753379266e-05,
+      "loss": 0.2844,
+      "step": 123
+    },
+    {
+      "epoch": 0.07936,
+      "grad_norm": 0.3386752443740303,
+      "learning_rate": 2.993488671510886e-05,
+      "loss": 0.3037,
+      "step": 124
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.3972212865125969,
+      "learning_rate": 2.993189565455601e-05,
+      "loss": 0.3137,
+      "step": 125
+    },
+    {
+      "epoch": 0.08064,
+      "grad_norm": 0.38002895692953814,
+      "learning_rate": 2.9928837585143497e-05,
+      "loss": 0.2873,
+      "step": 126
+    },
+    {
+      "epoch": 0.08128,
+      "grad_norm": 0.44072168693518854,
+      "learning_rate": 2.992571252059482e-05,
+      "loss": 0.2782,
+      "step": 127
+    },
+    {
+      "epoch": 0.08192,
+      "grad_norm": 0.3479281011658149,
+      "learning_rate": 2.992252047493411e-05,
+      "loss": 0.2799,
+      "step": 128
+    },
+    {
+      "epoch": 0.08256,
+      "grad_norm": 0.4142266632731221,
+      "learning_rate": 2.9919261462486098e-05,
+      "loss": 0.3176,
+      "step": 129
+    },
+    {
+      "epoch": 0.0832,
+      "grad_norm": 0.35944919396457825,
+      "learning_rate": 2.991593549787604e-05,
+      "loss": 0.2963,
+      "step": 130
+    },
+    {
+      "epoch": 0.08384,
+      "grad_norm": 0.3837171188201694,
+      "learning_rate": 2.9912542596029635e-05,
+      "loss": 0.3057,
+      "step": 131
+    },
+    {
+      "epoch": 0.08448,
+      "grad_norm": 0.4763114305033498,
+      "learning_rate": 2.990908277217298e-05,
+      "loss": 0.3224,
+      "step": 132
+    },
+    {
+      "epoch": 0.08512,
+      "grad_norm": 0.3834181145690126,
+      "learning_rate": 2.9905556041832494e-05,
+      "loss": 0.2841,
+      "step": 133
+    },
+    {
+      "epoch": 0.08576,
+      "grad_norm": 0.40814627377992635,
+      "learning_rate": 2.990196242083485e-05,
+      "loss": 0.2871,
+      "step": 134
+    },
+    {
+      "epoch": 0.0864,
+      "grad_norm": 0.444873139895432,
+      "learning_rate": 2.98983019253069e-05,
+      "loss": 0.29,
+      "step": 135
+    },
+    {
+      "epoch": 0.08704,
+      "grad_norm": 0.4015570039172884,
+      "learning_rate": 2.9894574571675593e-05,
+      "loss": 0.29,
+      "step": 136
+    },
+    {
+      "epoch": 0.08768,
+      "grad_norm": 0.4192576277736737,
+      "learning_rate": 2.989078037666793e-05,
+      "loss": 0.3187,
+      "step": 137
+    },
+    {
+      "epoch": 0.08832,
+      "grad_norm": 0.4701185999968217,
+      "learning_rate": 2.988691935731086e-05,
+      "loss": 0.3039,
+      "step": 138
+    },
+    {
+      "epoch": 0.08896,
+      "grad_norm": 0.331862366916375,
+      "learning_rate": 2.988299153093122e-05,
+      "loss": 0.2814,
+      "step": 139
+    },
+    {
+      "epoch": 0.0896,
+      "grad_norm": 0.4503832184907915,
+      "learning_rate": 2.987899691515565e-05,
+      "loss": 0.3084,
+      "step": 140
+    },
+    {
+      "epoch": 0.09024,
+      "grad_norm": 0.3734705387771823,
+      "learning_rate": 2.987493552791052e-05,
+      "loss": 0.2837,
+      "step": 141
+    },
+    {
+      "epoch": 0.09088,
+      "grad_norm": 0.3978970762252604,
+      "learning_rate": 2.9870807387421843e-05,
+      "loss": 0.3032,
+      "step": 142
+    },
+    {
+      "epoch": 0.09152,
+      "grad_norm": 0.39348398869038875,
+      "learning_rate": 2.986661251221519e-05,
+      "loss": 0.3062,
+      "step": 143
+    },
+    {
+      "epoch": 0.09216,
+      "grad_norm": 0.4493064184937299,
+      "learning_rate": 2.986235092111563e-05,
+      "loss": 0.3071,
+      "step": 144
+    },
+    {
+      "epoch": 0.0928,
+      "grad_norm": 0.44946841788948644,
+      "learning_rate": 2.985802263324761e-05,
+      "loss": 0.3025,
+      "step": 145
+    },
+    {
+      "epoch": 0.09344,
+      "grad_norm": 0.5016226955727763,
+      "learning_rate": 2.9853627668034898e-05,
+      "loss": 0.3246,
+      "step": 146
+    },
+    {
+      "epoch": 0.09408,
+      "grad_norm": 0.399973023082225,
+      "learning_rate": 2.9849166045200476e-05,
+      "loss": 0.2811,
+      "step": 147
+    },
+    {
+      "epoch": 0.09472,
+      "grad_norm": 0.3739130134478663,
+      "learning_rate": 2.9844637784766478e-05,
+      "loss": 0.2855,
+      "step": 148
+    },
+    {
+      "epoch": 0.09536,
+      "grad_norm": 0.43805848638708655,
+      "learning_rate": 2.9840042907054068e-05,
+      "loss": 0.2798,
+      "step": 149
+    },
+    {
+      "epoch": 0.096,
+      "grad_norm": 0.4402314818129774,
+      "learning_rate": 2.9835381432683363e-05,
+      "loss": 0.2847,
+      "step": 150
+    },
+    {
+      "epoch": 0.09664,
+      "grad_norm": 0.4047049034331425,
+      "learning_rate": 2.9830653382573358e-05,
+      "loss": 0.2861,
+      "step": 151
+    },
+    {
+      "epoch": 0.09728,
+      "grad_norm": 0.43978737589580585,
+      "learning_rate": 2.982585877794179e-05,
+      "loss": 0.3054,
+      "step": 152
+    },
+    {
+      "epoch": 0.09792,
+      "grad_norm": 0.4730334615845102,
+      "learning_rate": 2.9820997640305097e-05,
+      "loss": 0.2817,
+      "step": 153
+    },
+    {
+      "epoch": 0.09856,
+      "grad_norm": 0.3740570479644525,
+      "learning_rate": 2.981606999147827e-05,
+      "loss": 0.2608,
+      "step": 154
+    },
+    {
+      "epoch": 0.0992,
+      "grad_norm": 0.481760965756092,
+      "learning_rate": 2.9811075853574788e-05,
+      "loss": 0.3113,
+      "step": 155
+    },
+    {
+      "epoch": 0.09984,
+      "grad_norm": 0.3961997084965111,
+      "learning_rate": 2.98060152490065e-05,
+      "loss": 0.3088,
+      "step": 156
+    },
+    {
+      "epoch": 0.10048,
+      "grad_norm": 0.428730401496227,
+      "learning_rate": 2.9800888200483552e-05,
+      "loss": 0.3129,
+      "step": 157
+    },
+    {
+      "epoch": 0.10112,
+      "grad_norm": 0.4178705790625643,
+      "learning_rate": 2.979569473101424e-05,
+      "loss": 0.3284,
+      "step": 158
+    },
+    {
+      "epoch": 0.10176,
+      "grad_norm": 0.4463567693300283,
+      "learning_rate": 2.9790434863904957e-05,
+      "loss": 0.3044,
+      "step": 159
+    },
+    {
+      "epoch": 0.1024,
+      "grad_norm": 0.3939775657775284,
+      "learning_rate": 2.9785108622760045e-05,
+      "loss": 0.2894,
+      "step": 160
+    },
+    {
+      "epoch": 0.10304,
+      "grad_norm": 0.38174144831499374,
+      "learning_rate": 2.9779716031481717e-05,
+      "loss": 0.2779,
+      "step": 161
+    },
+    {
+      "epoch": 0.10368,
+      "grad_norm": 0.38589717775088483,
+      "learning_rate": 2.9774257114269955e-05,
+      "loss": 0.3123,
+      "step": 162
+    },
+    {
+      "epoch": 0.10432,
+      "grad_norm": 0.3673496875585954,
+      "learning_rate": 2.9768731895622355e-05,
+      "loss": 0.2771,
+      "step": 163
+    },
+    {
+      "epoch": 0.10496,
+      "grad_norm": 0.3571608186420396,
+      "learning_rate": 2.9763140400334072e-05,
+      "loss": 0.2897,
+      "step": 164
+    },
+    {
+      "epoch": 0.1056,
+      "grad_norm": 0.3939651393603113,
+      "learning_rate": 2.975748265349769e-05,
+      "loss": 0.2966,
+      "step": 165
+    },
+    {
+      "epoch": 0.10624,
+      "grad_norm": 0.3805663760796052,
+      "learning_rate": 2.975175868050309e-05,
+      "loss": 0.3156,
+      "step": 166
+    },
+    {
+      "epoch": 0.10688,
+      "grad_norm": 0.3972270203399388,
+      "learning_rate": 2.9745968507037356e-05,
+      "loss": 0.2963,
+      "step": 167
+    },
+    {
+      "epoch": 0.10752,
+      "grad_norm": 0.39226032551563944,
+      "learning_rate": 2.974011215908467e-05,
+      "loss": 0.2933,
+      "step": 168
+    },
+    {
+      "epoch": 0.10816,
+      "grad_norm": 0.35875066299615677,
+      "learning_rate": 2.9734189662926163e-05,
+      "loss": 0.2758,
+      "step": 169
+    },
+    {
+      "epoch": 0.1088,
+      "grad_norm": 0.3709380144753209,
+      "learning_rate": 2.9728201045139813e-05,
+      "loss": 0.2927,
+      "step": 170
+    },
+    {
+      "epoch": 0.10944,
+      "grad_norm": 0.370201710217779,
+      "learning_rate": 2.972214633260035e-05,
+      "loss": 0.2866,
+      "step": 171
+    },
+    {
+      "epoch": 0.11008,
+      "grad_norm": 0.3758060771260273,
+      "learning_rate": 2.9716025552479093e-05,
+      "loss": 0.2994,
+      "step": 172
+    },
+    {
+      "epoch": 0.11072,
+      "grad_norm": 0.32580329325564505,
+      "learning_rate": 2.9709838732243844e-05,
+      "loss": 0.2802,
+      "step": 173
+    },
+    {
+      "epoch": 0.11136,
+      "grad_norm": 0.38248606018972797,
+      "learning_rate": 2.970358589965879e-05,
+      "loss": 0.2739,
+      "step": 174
+    },
+    {
+      "epoch": 0.112,
+      "grad_norm": 0.36493445057406637,
+      "learning_rate": 2.9697267082784342e-05,
+      "loss": 0.2874,
+      "step": 175
+    },
+    {
+      "epoch": 0.11264,
+      "grad_norm": 0.6384082314576618,
+      "learning_rate": 2.969088230997703e-05,
+      "loss": 0.2835,
+      "step": 176
+    },
+    {
+      "epoch": 0.11328,
+      "grad_norm": 0.4046365021981396,
+      "learning_rate": 2.9684431609889365e-05,
+      "loss": 0.2964,
+      "step": 177
+    },
+    {
+      "epoch": 0.11392,
+      "grad_norm": 0.3777231337256365,
+      "learning_rate": 2.9677915011469717e-05,
+      "loss": 0.302,
+      "step": 178
+    },
+    {
+      "epoch": 0.11456,
+      "grad_norm": 0.3550577304426545,
+      "learning_rate": 2.9671332543962183e-05,
+      "loss": 0.2809,
+      "step": 179
+    },
+    {
+      "epoch": 0.1152,
+      "grad_norm": 0.34027303690053873,
+      "learning_rate": 2.9664684236906466e-05,
+      "loss": 0.2998,
+      "step": 180
+    },
+    {
+      "epoch": 0.11584,
+      "grad_norm": 0.3705269161905217,
+      "learning_rate": 2.965797012013772e-05,
+      "loss": 0.3187,
+      "step": 181
+    },
+    {
+      "epoch": 0.11648,
+      "grad_norm": 0.36000193066901365,
+      "learning_rate": 2.9651190223786427e-05,
+      "loss": 0.2838,
+      "step": 182
+    },
+    {
+      "epoch": 0.11712,
+      "grad_norm": 0.32994641779773276,
+      "learning_rate": 2.9644344578278284e-05,
+      "loss": 0.2791,
+      "step": 183
+    },
+    {
+      "epoch": 0.11776,
+      "grad_norm": 0.35570577433544603,
+      "learning_rate": 2.963743321433402e-05,
+      "loss": 0.2773,
+      "step": 184
+    },
+    {
+      "epoch": 0.1184,
+      "grad_norm": 0.3596282933620285,
+      "learning_rate": 2.9630456162969298e-05,
+      "loss": 0.2797,
+      "step": 185
+    },
+    {
+      "epoch": 0.11904,
+      "grad_norm": 0.3262807239848876,
+      "learning_rate": 2.9623413455494563e-05,
+      "loss": 0.288,
+      "step": 186
+    },
+    {
+      "epoch": 0.11968,
+      "grad_norm": 0.3565865179949324,
+      "learning_rate": 2.9616305123514897e-05,
+      "loss": 0.2887,
+      "step": 187
+    },
+    {
+      "epoch": 0.12032,
+      "grad_norm": 0.39532927846862465,
+      "learning_rate": 2.9609131198929884e-05,
+      "loss": 0.2782,
+      "step": 188
+    },
+    {
+      "epoch": 0.12096,
+      "grad_norm": 0.40888574656849996,
+      "learning_rate": 2.9601891713933457e-05,
+      "loss": 0.2718,
+      "step": 189
+    },
+    {
+      "epoch": 0.1216,
+      "grad_norm": 0.3494105157933286,
+      "learning_rate": 2.9594586701013765e-05,
+      "loss": 0.2925,
+      "step": 190
+    },
+    {
+      "epoch": 0.12224,
+      "grad_norm": 0.38169153601979233,
+      "learning_rate": 2.958721619295302e-05,
+      "loss": 0.2783,
+      "step": 191
+    },
+    {
+      "epoch": 0.12288,
+      "grad_norm": 0.3650159986813621,
+      "learning_rate": 2.9579780222827354e-05,
+      "loss": 0.2839,
+      "step": 192
+    },
+    {
+      "epoch": 0.12352,
+      "grad_norm": 0.32047234144052256,
+      "learning_rate": 2.957227882400667e-05,
+      "loss": 0.2781,
+      "step": 193
+    },
+    {
+      "epoch": 0.12416,
+      "grad_norm": 0.39912184054626987,
+      "learning_rate": 2.9564712030154486e-05,
+      "loss": 0.2992,
+      "step": 194
+    },
+    {
+      "epoch": 0.1248,
+      "grad_norm": 0.3589964401769852,
+      "learning_rate": 2.9557079875227795e-05,
+      "loss": 0.2991,
+      "step": 195
+    },
+    {
+      "epoch": 0.12544,
+      "grad_norm": 0.3095249869798657,
+      "learning_rate": 2.95493823934769e-05,
+      "loss": 0.2886,
+      "step": 196
+    },
+    {
+      "epoch": 0.12608,
+      "grad_norm": 0.33996550944422993,
+      "learning_rate": 2.954161961944527e-05,
+      "loss": 0.2705,
+      "step": 197
+    },
+    {
+      "epoch": 0.12672,
+      "grad_norm": 0.3423071479543653,
+      "learning_rate": 2.953379158796938e-05,
+      "loss": 0.2795,
+      "step": 198
+    },
+    {
+      "epoch": 0.12736,
+      "grad_norm": 0.36082529918938056,
+      "learning_rate": 2.9525898334178566e-05,
+      "loss": 0.2887,
+      "step": 199
+    },
+    {
+      "epoch": 0.128,
+      "grad_norm": 0.37640661693173727,
+      "learning_rate": 2.951793989349484e-05,
+      "loss": 0.282,
+      "step": 200
+    },
+    {
+      "epoch": 0.12864,
+      "grad_norm": 0.3023070350545028,
+      "learning_rate": 2.950991630163277e-05,
+      "loss": 0.254,
+      "step": 201
+    },
+    {
+      "epoch": 0.12928,
+      "grad_norm": 0.3746495017193779,
+      "learning_rate": 2.950182759459928e-05,
+      "loss": 0.2913,
+      "step": 202
+    },
+    {
+      "epoch": 0.12992,
+      "grad_norm": 0.38483323478343573,
+      "learning_rate": 2.949367380869351e-05,
+      "loss": 0.2913,
+      "step": 203
+    },
+    {
+      "epoch": 0.13056,
+      "grad_norm": 0.4213879707012684,
+      "learning_rate": 2.9485454980506663e-05,
+      "loss": 0.2841,
+      "step": 204
+    },
+    {
+      "epoch": 0.1312,
+      "grad_norm": 0.4131757338780886,
+      "learning_rate": 2.9477171146921816e-05,
+      "loss": 0.2981,
+      "step": 205
+    },
+    {
+      "epoch": 0.13184,
+      "grad_norm": 0.4415959043983817,
+      "learning_rate": 2.946882234511377e-05,
+      "loss": 0.2975,
+      "step": 206
+    },
+    {
+      "epoch": 0.13248,
+      "grad_norm": 0.37499387107669097,
+      "learning_rate": 2.9460408612548876e-05,
+      "loss": 0.2844,
+      "step": 207
+    },
+    {
+      "epoch": 0.13312,
+      "grad_norm": 0.41827519988058354,
+      "learning_rate": 2.9451929986984875e-05,
+      "loss": 0.2958,
+      "step": 208
+    },
+    {
+      "epoch": 0.13376,
+      "grad_norm": 0.33683714204034787,
+      "learning_rate": 2.9443386506470725e-05,
+      "loss": 0.2686,
+      "step": 209
+    },
+    {
+      "epoch": 0.1344,
+      "grad_norm": 0.38867413035222104,
+      "learning_rate": 2.9434778209346427e-05,
+      "loss": 0.2876,
+      "step": 210
+    },
+    {
+      "epoch": 0.13504,
+      "grad_norm": 0.39089889654238963,
+      "learning_rate": 2.942610513424285e-05,
+      "loss": 0.2832,
+      "step": 211
+    },
+    {
+      "epoch": 0.13568,
+      "grad_norm": 0.3384476658736123,
+      "learning_rate": 2.9417367320081567e-05,
+      "loss": 0.2604,
+      "step": 212
+    },
+    {
+      "epoch": 0.13632,
+      "grad_norm": 0.38210814875894833,
+      "learning_rate": 2.940856480607468e-05,
+      "loss": 0.2986,
+      "step": 213
+    },
+    {
+      "epoch": 0.13696,
+      "grad_norm": 0.38075475653821916,
+      "learning_rate": 2.9399697631724637e-05,
+      "loss": 0.2761,
+      "step": 214
+    },
+    {
+      "epoch": 0.1376,
+      "grad_norm": 0.36451096593797544,
+      "learning_rate": 2.9390765836824053e-05,
+      "loss": 0.3053,
+      "step": 215
+    },
+    {
+      "epoch": 0.13824,
+      "grad_norm": 0.36903700040818893,
+      "learning_rate": 2.938176946145555e-05,
+      "loss": 0.2776,
+      "step": 216
+    },
+    {
+      "epoch": 0.13888,
+      "grad_norm": 0.3858847022411217,
+      "learning_rate": 2.9372708545991542e-05,
+      "loss": 0.2682,
+      "step": 217
+    },
+    {
+      "epoch": 0.13952,
+      "grad_norm": 0.3664388884748071,
+      "learning_rate": 2.936358313109409e-05,
+      "loss": 0.263,
+      "step": 218
+    },
+    {
+      "epoch": 0.14016,
+      "grad_norm": 0.3833106914044559,
+      "learning_rate": 2.935439325771471e-05,
+      "loss": 0.284,
+      "step": 219
+    },
+    {
+      "epoch": 0.1408,
+      "grad_norm": 0.4366332113795893,
+      "learning_rate": 2.9345138967094174e-05,
+      "loss": 0.2827,
+      "step": 220
+    },
+    {
+      "epoch": 0.14144,
+      "grad_norm": 0.35644258927978634,
+      "learning_rate": 2.9335820300762334e-05,
+      "loss": 0.2763,
+      "step": 221
+    },
+    {
+      "epoch": 0.14208,
+      "grad_norm": 0.4086383954714273,
+      "learning_rate": 2.9326437300537937e-05,
+      "loss": 0.2731,
+      "step": 222
+    },
+    {
+      "epoch": 0.14272,
+      "grad_norm": 0.37566830211795793,
+      "learning_rate": 2.9316990008528446e-05,
+      "loss": 0.3035,
+      "step": 223
+    },
+    {
+      "epoch": 0.14336,
+      "grad_norm": 0.37673883341537595,
+      "learning_rate": 2.9307478467129827e-05,
+      "loss": 0.2807,
+      "step": 224
+    },
+    {
+      "epoch": 0.144,
+      "grad_norm": 0.3658815413836672,
+      "learning_rate": 2.9297902719026392e-05,
+      "loss": 0.2872,
+      "step": 225
+    },
+    {
+      "epoch": 0.14464,
+      "grad_norm": 0.3723900174142704,
+      "learning_rate": 2.928826280719057e-05,
+      "loss": 0.2628,
+      "step": 226
+    },
+    {
+      "epoch": 0.14528,
+      "grad_norm": 0.3583483332892453,
+      "learning_rate": 2.9278558774882748e-05,
+      "loss": 0.2757,
+      "step": 227
+    },
+    {
+      "epoch": 0.14592,
+      "grad_norm": 0.36256098670512743,
+      "learning_rate": 2.9268790665651053e-05,
+      "loss": 0.2795,
+      "step": 228
+    },
+    {
+      "epoch": 0.14656,
+      "grad_norm": 0.4447205295517405,
+      "learning_rate": 2.925895852333117e-05,
+      "loss": 0.2917,
+      "step": 229
+    },
+    {
+      "epoch": 0.1472,
+      "grad_norm": 0.33653314086590325,
+      "learning_rate": 2.924906239204614e-05,
+      "loss": 0.2856,
+      "step": 230
+    },
+    {
+      "epoch": 0.14784,
+      "grad_norm": 0.3875121699387206,
+      "learning_rate": 2.9239102316206166e-05,
+      "loss": 0.2756,
+      "step": 231
+    },
+    {
+      "epoch": 0.14848,
+      "grad_norm": 0.4191113393584937,
+      "learning_rate": 2.9229078340508404e-05,
+      "loss": 0.2947,
+      "step": 232
+    },
+    {
+      "epoch": 0.14912,
+      "grad_norm": 0.3866679704383513,
+      "learning_rate": 2.9218990509936774e-05,
+      "loss": 0.2948,
+      "step": 233
+    },
+    {
+      "epoch": 0.14976,
+      "grad_norm": 0.44210860261964957,
+      "learning_rate": 2.9208838869761756e-05,
+      "loss": 0.2823,
+      "step": 234
+    },
+    {
+      "epoch": 0.1504,
+      "grad_norm": 0.3569394913477398,
+      "learning_rate": 2.9198623465540172e-05,
+      "loss": 0.2896,
+      "step": 235
+    },
+    {
+      "epoch": 0.15104,
+      "grad_norm": 0.3582653155321637,
+      "learning_rate": 2.9188344343115005e-05,
+      "loss": 0.2639,
+      "step": 236
+    },
+    {
+      "epoch": 0.15168,
+      "grad_norm": 0.3632299874629792,
+      "learning_rate": 2.9178001548615176e-05,
+      "loss": 0.2752,
+      "step": 237
+    },
+    {
+      "epoch": 0.15232,
+      "grad_norm": 0.3627690799272646,
+      "learning_rate": 2.9167595128455357e-05,
+      "loss": 0.2697,
+      "step": 238
+    },
+    {
+      "epoch": 0.15296,
+      "grad_norm": 0.31681761773997114,
+      "learning_rate": 2.915712512933572e-05,
+      "loss": 0.2886,
+      "step": 239
+    },
+    {
+      "epoch": 0.1536,
+      "grad_norm": 0.3652202557075009,
+      "learning_rate": 2.914659159824178e-05,
+      "loss": 0.291,
+      "step": 240
+    },
+    {
+      "epoch": 0.15424,
+      "grad_norm": 0.32355380076781154,
+      "learning_rate": 2.913599458244416e-05,
+      "loss": 0.288,
+      "step": 241
+    },
+    {
+      "epoch": 0.15488,
+      "grad_norm": 0.3263439505506079,
+      "learning_rate": 2.9125334129498368e-05,
+      "loss": 0.2669,
+      "step": 242
+    },
+    {
+      "epoch": 0.15552,
+      "grad_norm": 0.34698579986117173,
+      "learning_rate": 2.911461028724459e-05,
+      "loss": 0.2623,
+      "step": 243
+    },
+    {
+      "epoch": 0.15616,
+      "grad_norm": 0.3560347825174995,
+      "learning_rate": 2.9103823103807503e-05,
+      "loss": 0.2779,
+      "step": 244
+    },
+    {
+      "epoch": 0.1568,
+      "grad_norm": 0.3829349365942763,
+      "learning_rate": 2.9092972627596012e-05,
+      "loss": 0.285,
+      "step": 245
+    },
+    {
+      "epoch": 0.15744,
+      "grad_norm": 0.35634808472353574,
+      "learning_rate": 2.9082058907303064e-05,
+      "loss": 0.2787,
+      "step": 246
+    },
+    {
+      "epoch": 0.15808,
+      "grad_norm": 0.3973522847788536,
+      "learning_rate": 2.9071081991905436e-05,
+      "loss": 0.2686,
+      "step": 247
+    },
+    {
+      "epoch": 0.15872,
+      "grad_norm": 0.3503037505655835,
+      "learning_rate": 2.9060041930663477e-05,
+      "loss": 0.2601,
+      "step": 248
+    },
+    {
+      "epoch": 0.15936,
+      "grad_norm": 0.42641830259468066,
+      "learning_rate": 2.9048938773120933e-05,
+      "loss": 0.2571,
+      "step": 249
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.33022262318082957,
+      "learning_rate": 2.903777256910469e-05,
+      "loss": 0.2539,
+      "step": 250
+    },
+    {
+      "epoch": 0.16064,
+      "grad_norm": 0.34806931009286257,
+      "learning_rate": 2.902654336872457e-05,
+      "loss": 0.2855,
+      "step": 251
+    },
+    {
+      "epoch": 0.16128,
+      "grad_norm": 0.3802291207724926,
+      "learning_rate": 2.9015251222373094e-05,
+      "loss": 0.2505,
+      "step": 252
+    },
+    {
+      "epoch": 0.16192,
+      "grad_norm": 0.4182370813509436,
+      "learning_rate": 2.9003896180725268e-05,
+      "loss": 0.2821,
+      "step": 253
+    },
+    {
+      "epoch": 0.16256,
+      "grad_norm": 0.45914515409569023,
+      "learning_rate": 2.8992478294738345e-05,
+      "loss": 0.2934,
+      "step": 254
+    },
+    {
+      "epoch": 0.1632,
+      "grad_norm": 0.3490846199737997,
+      "learning_rate": 2.8980997615651597e-05,
+      "loss": 0.2853,
+      "step": 255
+    },
+    {
+      "epoch": 0.16384,
+      "grad_norm": 0.3945607808525512,
+      "learning_rate": 2.8969454194986095e-05,
+      "loss": 0.2704,
+      "step": 256
+    },
+    {
+      "epoch": 0.16448,
+      "grad_norm": 0.4134287311468698,
+      "learning_rate": 2.8957848084544473e-05,
+      "loss": 0.2856,
+      "step": 257
+    },
+    {
+      "epoch": 0.16512,
+      "grad_norm": 0.388727953362913,
+      "learning_rate": 2.8946179336410682e-05,
+      "loss": 0.2707,
+      "step": 258
+    },
+    {
+      "epoch": 0.16576,
+      "grad_norm": 0.3707496935300037,
+      "learning_rate": 2.8934448002949775e-05,
+      "loss": 0.2736,
+      "step": 259
+    },
+    {
+      "epoch": 0.1664,
+      "grad_norm": 0.37509134347160566,
+      "learning_rate": 2.892265413680767e-05,
+      "loss": 0.2656,
+      "step": 260
+    },
+    {
+      "epoch": 0.16704,
+      "grad_norm": 0.37839278243532787,
+      "learning_rate": 2.8910797790910902e-05,
+      "loss": 0.2954,
+      "step": 261
+    },
+    {
+      "epoch": 0.16768,
+      "grad_norm": 0.3709934691909708,
+      "learning_rate": 2.889887901846639e-05,
+      "loss": 0.2771,
+      "step": 262
+    },
+    {
+      "epoch": 0.16832,
+      "grad_norm": 0.38441885239694634,
+      "learning_rate": 2.8886897872961203e-05,
+      "loss": 0.2872,
+      "step": 263
+    },
+    {
+      "epoch": 0.16896,
+      "grad_norm": 0.36199416231449505,
+      "learning_rate": 2.887485440816233e-05,
+      "loss": 0.2577,
+      "step": 264
+    },
+    {
+      "epoch": 0.1696,
+      "grad_norm": 0.35207892498047666,
+      "learning_rate": 2.8862748678116402e-05,
+      "loss": 0.2692,
+      "step": 265
+    },
+    {
+      "epoch": 0.17024,
+      "grad_norm": 0.3167681279515036,
+      "learning_rate": 2.885058073714949e-05,
+      "loss": 0.2517,
+      "step": 266
+    },
+    {
+      "epoch": 0.17088,
+      "grad_norm": 0.3836456596956298,
+      "learning_rate": 2.8838350639866843e-05,
+      "loss": 0.3038,
+      "step": 267
+    },
+    {
+      "epoch": 0.17152,
+      "grad_norm": 0.3604280243223731,
+      "learning_rate": 2.882605844115264e-05,
+      "loss": 0.2748,
+      "step": 268
+    },
+    {
+      "epoch": 0.17216,
+      "grad_norm": 0.33239517181664746,
+      "learning_rate": 2.8813704196169753e-05,
+      "loss": 0.2705,
+      "step": 269
+    },
+    {
+      "epoch": 0.1728,
+      "grad_norm": 0.4453367026441279,
+      "learning_rate": 2.8801287960359494e-05,
+      "loss": 0.2752,
+      "step": 270
+    },
+    {
+      "epoch": 0.17344,
+      "grad_norm": 0.3672385598637218,
+      "learning_rate": 2.8788809789441364e-05,
+      "loss": 0.2813,
+      "step": 271
+    },
+    {
+      "epoch": 0.17408,
+      "grad_norm": 0.42500271676682233,
+      "learning_rate": 2.8776269739412803e-05,
+      "loss": 0.2581,
+      "step": 272
+    },
+    {
+      "epoch": 0.17472,
+      "grad_norm": 0.3812194331681742,
+      "learning_rate": 2.8763667866548956e-05,
+      "loss": 0.2797,
+      "step": 273
+    },
+    {
+      "epoch": 0.17536,
+      "grad_norm": 0.398047671024292,
+      "learning_rate": 2.875100422740239e-05,
+      "loss": 0.2822,
+      "step": 274
+    },
+    {
+      "epoch": 0.176,
+      "grad_norm": 0.4382339198136674,
+      "learning_rate": 2.8738278878802865e-05,
+      "loss": 0.2951,
+      "step": 275
+    },
+    {
+      "epoch": 0.17664,
+      "grad_norm": 0.32914734501477555,
+      "learning_rate": 2.8725491877857073e-05,
+      "loss": 0.2505,
+      "step": 276
+    },
+    {
+      "epoch": 0.17728,
+      "grad_norm": 0.359221594407232,
+      "learning_rate": 2.8712643281948365e-05,
+      "loss": 0.2682,
+      "step": 277
+    },
+    {
+      "epoch": 0.17792,
+      "grad_norm": 0.3610201048162127,
+      "learning_rate": 2.8699733148736525e-05,
+      "loss": 0.2706,
+      "step": 278
+    },
+    {
+      "epoch": 0.17856,
+      "grad_norm": 0.3576575465005173,
+      "learning_rate": 2.868676153615748e-05,
+      "loss": 0.2725,
+      "step": 279
+    },
+    {
+      "epoch": 0.1792,
+      "grad_norm": 0.39648269977135403,
+      "learning_rate": 2.8673728502423067e-05,
+      "loss": 0.2638,
+      "step": 280
+    },
+    {
+      "epoch": 0.17984,
+      "grad_norm": 0.35937870544419476,
+      "learning_rate": 2.8660634106020747e-05,
+      "loss": 0.2753,
+      "step": 281
+    },
+    {
+      "epoch": 0.18048,
+      "grad_norm": 0.4065407489775043,
+      "learning_rate": 2.8647478405713355e-05,
+      "loss": 0.2814,
+      "step": 282
+    },
+    {
+      "epoch": 0.18112,
+      "grad_norm": 0.3401826548234812,
+      "learning_rate": 2.8634261460538845e-05,
+      "loss": 0.2734,
+      "step": 283
+    },
+    {
+      "epoch": 0.18176,
+      "grad_norm": 0.3871247999638719,
+      "learning_rate": 2.8620983329810004e-05,
+      "loss": 0.266,
+      "step": 284
+    },
+    {
+      "epoch": 0.1824,
+      "grad_norm": 0.40719492038320787,
+      "learning_rate": 2.860764407311421e-05,
+      "loss": 0.2727,
+      "step": 285
+    },
+    {
+      "epoch": 0.18304,
+      "grad_norm": 0.3399520864358679,
+      "learning_rate": 2.8594243750313137e-05,
+      "loss": 0.2689,
+      "step": 286
+    },
+    {
+      "epoch": 0.18368,
+      "grad_norm": 0.3953164307154916,
+      "learning_rate": 2.8580782421542514e-05,
+      "loss": 0.273,
+      "step": 287
+    },
+    {
+      "epoch": 0.18432,
+      "grad_norm": 0.4447124642066109,
+      "learning_rate": 2.8567260147211826e-05,
+      "loss": 0.282,
+      "step": 288
+    },
+    {
+      "epoch": 0.18496,
+      "grad_norm": 0.3479285885985358,
+      "learning_rate": 2.8553676988004083e-05,
+      "loss": 0.2846,
+      "step": 289
+    },
+    {
+      "epoch": 0.1856,
+      "grad_norm": 0.4427716710403484,
+      "learning_rate": 2.8540033004875506e-05,
+      "loss": 0.2731,
+      "step": 290
+    },
+    {
+      "epoch": 0.18624,
+      "grad_norm": 0.3389993255265989,
+      "learning_rate": 2.852632825905528e-05,
+      "loss": 0.2728,
+      "step": 291
+    },
+    {
+      "epoch": 0.18688,
+      "grad_norm": 0.36214801666992735,
+      "learning_rate": 2.851256281204526e-05,
+      "loss": 0.2963,
+      "step": 292
+    },
+    {
+      "epoch": 0.18752,
+      "grad_norm": 0.3225822277564517,
+      "learning_rate": 2.849873672561972e-05,
+      "loss": 0.256,
+      "step": 293
+    },
+    {
+      "epoch": 0.18816,
+      "grad_norm": 0.36090735754771,
+      "learning_rate": 2.8484850061825052e-05,
+      "loss": 0.274,
+      "step": 294
+    },
+    {
+      "epoch": 0.1888,
+      "grad_norm": 0.3619026443591473,
+      "learning_rate": 2.84709028829795e-05,
+      "loss": 0.2904,
+      "step": 295
+    },
+    {
+      "epoch": 0.18944,
+      "grad_norm": 0.3252751498399478,
+      "learning_rate": 2.8456895251672867e-05,
+      "loss": 0.2892,
+      "step": 296
+    },
+    {
+      "epoch": 0.19008,
+      "grad_norm": 0.3420708202146713,
+      "learning_rate": 2.8442827230766265e-05,
+      "loss": 0.2509,
+      "step": 297
+    },
+    {
+      "epoch": 0.19072,
+      "grad_norm": 0.3927376280930936,
+      "learning_rate": 2.8428698883391805e-05,
+      "loss": 0.2902,
+      "step": 298
+    },
+    {
+      "epoch": 0.19136,
+      "grad_norm": 0.33206871786295983,
+      "learning_rate": 2.8414510272952306e-05,
+      "loss": 0.2725,
+      "step": 299
+    },
+    {
+      "epoch": 0.192,
+      "grad_norm": 0.31541664358451077,
+      "learning_rate": 2.840026146312104e-05,
+      "loss": 0.2379,
+      "step": 300
+    },
+    {
+      "epoch": 0.19264,
+      "grad_norm": 0.3610931095304509,
+      "learning_rate": 2.8385952517841433e-05,
+      "loss": 0.2778,
+      "step": 301
+    },
+    {
+      "epoch": 0.19328,
+      "grad_norm": 0.32547764886975145,
+      "learning_rate": 2.837158350132677e-05,
+      "loss": 0.2786,
+      "step": 302
+    },
+    {
+      "epoch": 0.19392,
+      "grad_norm": 0.3428550126773416,
+      "learning_rate": 2.835715447805991e-05,
+      "loss": 0.2896,
+      "step": 303
+    },
+    {
+      "epoch": 0.19456,
+      "grad_norm": 0.34537169429710707,
+      "learning_rate": 2.8342665512793018e-05,
+      "loss": 0.2888,
+      "step": 304
+    },
+    {
+      "epoch": 0.1952,
+      "grad_norm": 0.3526668903699371,
+      "learning_rate": 2.8328116670547237e-05,
+      "loss": 0.2711,
+      "step": 305
+    },
+    {
+      "epoch": 0.19584,
+      "grad_norm": 0.3298129440685655,
+      "learning_rate": 2.8313508016612428e-05,
+      "loss": 0.2723,
+      "step": 306
+    },
+    {
+      "epoch": 0.19648,
+      "grad_norm": 0.3252403541009243,
+      "learning_rate": 2.8298839616546854e-05,
+      "loss": 0.2783,
+      "step": 307
+    },
+    {
+      "epoch": 0.19712,
+      "grad_norm": 0.31942049727047156,
+      "learning_rate": 2.8284111536176907e-05,
+      "loss": 0.3047,
+      "step": 308
+    },
+    {
+      "epoch": 0.19776,
+      "grad_norm": 0.28119673519843524,
+      "learning_rate": 2.8269323841596802e-05,
+      "loss": 0.2685,
+      "step": 309
+    },
+    {
+      "epoch": 0.1984,
+      "grad_norm": 0.3574212794376199,
+      "learning_rate": 2.825447659916827e-05,
+      "loss": 0.271,
+      "step": 310
+    },
+    {
+      "epoch": 0.19904,
+      "grad_norm": 0.30153919945084046,
+      "learning_rate": 2.823956987552028e-05,
+      "loss": 0.2529,
+      "step": 311
+    },
+    {
+      "epoch": 0.19968,
+      "grad_norm": 0.3494979054883366,
+      "learning_rate": 2.8224603737548737e-05,
+      "loss": 0.2886,
+      "step": 312
+    },
+    {
+      "epoch": 0.20032,
+      "grad_norm": 0.30740471706963013,
+      "learning_rate": 2.8209578252416162e-05,
+      "loss": 0.2876,
+      "step": 313
+    },
+    {
+      "epoch": 0.20096,
+      "grad_norm": 0.3279624895202358,
+      "learning_rate": 2.8194493487551402e-05,
+      "loss": 0.2946,
+      "step": 314
+    },
+    {
+      "epoch": 0.2016,
+      "grad_norm": 0.3430080164088275,
+      "learning_rate": 2.8179349510649354e-05,
+      "loss": 0.2891,
+      "step": 315
+    },
+    {
+      "epoch": 0.20224,
+      "grad_norm": 0.30496631367689814,
+      "learning_rate": 2.8164146389670605e-05,
+      "loss": 0.2842,
+      "step": 316
+    },
+    {
+      "epoch": 0.20288,
+      "grad_norm": 0.379409454373658,
+      "learning_rate": 2.8148884192841183e-05,
+      "loss": 0.294,
+      "step": 317
+    },
+    {
+      "epoch": 0.20352,
+      "grad_norm": 0.2948622980345085,
+      "learning_rate": 2.8133562988652218e-05,
+      "loss": 0.2584,
+      "step": 318
+    },
+    {
+      "epoch": 0.20416,
+      "grad_norm": 0.3443019691027711,
+      "learning_rate": 2.8118182845859636e-05,
+      "loss": 0.251,
+      "step": 319
+    },
+    {
+      "epoch": 0.2048,
+      "grad_norm": 0.36699154745018375,
+      "learning_rate": 2.810274383348387e-05,
+      "loss": 0.2952,
+      "step": 320
+    },
+    {
+      "epoch": 0.20544,
+      "grad_norm": 0.33967893078585726,
+      "learning_rate": 2.8087246020809536e-05,
+      "loss": 0.2796,
+      "step": 321
+    },
+    {
+      "epoch": 0.20608,
+      "grad_norm": 0.31170315997663467,
+      "learning_rate": 2.8071689477385117e-05,
+      "loss": 0.2721,
+      "step": 322
+    },
+    {
+      "epoch": 0.20672,
+      "grad_norm": 0.3242087780438831,
+      "learning_rate": 2.8056074273022666e-05,
+      "loss": 0.2774,
+      "step": 323
+    },
+    {
+      "epoch": 0.20736,
+      "grad_norm": 0.3555981636400892,
+      "learning_rate": 2.8040400477797476e-05,
+      "loss": 0.2797,
+      "step": 324
+    },
+    {
+      "epoch": 0.208,
+      "grad_norm": 0.309319481935765,
+      "learning_rate": 2.8024668162047783e-05,
+      "loss": 0.292,
+      "step": 325
+    },
+    {
+      "epoch": 0.20864,
+      "grad_norm": 0.38567954428634077,
+      "learning_rate": 2.8008877396374434e-05,
+      "loss": 0.2622,
+      "step": 326
+    },
+    {
+      "epoch": 0.20928,
+      "grad_norm": 0.3558299126479417,
+      "learning_rate": 2.799302825164058e-05,
+      "loss": 0.2741,
+      "step": 327
+    },
+    {
+      "epoch": 0.20992,
+      "grad_norm": 0.32961441916292467,
+      "learning_rate": 2.7977120798971374e-05,
+      "loss": 0.2669,
+      "step": 328
+    },
+    {
+      "epoch": 0.21056,
+      "grad_norm": 0.30783578854216875,
+      "learning_rate": 2.7961155109753596e-05,
+      "loss": 0.2645,
+      "step": 329
+    },
+    {
+      "epoch": 0.2112,
+      "grad_norm": 0.32966575581899893,
+      "learning_rate": 2.79451312556354e-05,
+      "loss": 0.2654,
+      "step": 330
+    },
+    {
+      "epoch": 0.21184,
+      "grad_norm": 0.31714659245735044,
+      "learning_rate": 2.7929049308525958e-05,
+      "loss": 0.2542,
+      "step": 331
+    },
+    {
+      "epoch": 0.21248,
+      "grad_norm": 0.3196236044042015,
+      "learning_rate": 2.7912909340595133e-05,
+      "loss": 0.2624,
+      "step": 332
+    },
+    {
+      "epoch": 0.21312,
+      "grad_norm": 0.3579856676787265,
+      "learning_rate": 2.7896711424273166e-05,
+      "loss": 0.272,
+      "step": 333
+    },
+    {
+      "epoch": 0.21376,
+      "grad_norm": 0.3327728411912822,
+      "learning_rate": 2.7880455632250358e-05,
+      "loss": 0.2588,
+      "step": 334
+    },
+    {
+      "epoch": 0.2144,
+      "grad_norm": 0.31027790499808855,
+      "learning_rate": 2.7864142037476727e-05,
+      "loss": 0.2728,
+      "step": 335
+    },
+    {
+      "epoch": 0.21504,
+      "grad_norm": 0.3202262124080976,
+      "learning_rate": 2.784777071316169e-05,
+      "loss": 0.2818,
+      "step": 336
+    },
+    {
+      "epoch": 0.21568,
+      "grad_norm": 0.29821110258387634,
+      "learning_rate": 2.7831341732773733e-05,
+      "loss": 0.2561,
+      "step": 337
+    },
+    {
+      "epoch": 0.21632,
+      "grad_norm": 0.33344963895852425,
+      "learning_rate": 2.7814855170040083e-05,
+      "loss": 0.2859,
+      "step": 338
+    },
+    {
+      "epoch": 0.21696,
+      "grad_norm": 0.3260713930863079,
+      "learning_rate": 2.7798311098946375e-05,
+      "loss": 0.2529,
+      "step": 339
+    },
+    {
+      "epoch": 0.2176,
+      "grad_norm": 0.3749841150106177,
+      "learning_rate": 2.7781709593736316e-05,
+      "loss": 0.2807,
+      "step": 340
+    },
+    {
+      "epoch": 0.21824,
+      "grad_norm": 0.34099250856014796,
+      "learning_rate": 2.7765050728911363e-05,
+      "loss": 0.2886,
+      "step": 341
+    },
+    {
+      "epoch": 0.21888,
+      "grad_norm": 0.33765459803209535,
+      "learning_rate": 2.7748334579230375e-05,
+      "loss": 0.2497,
+      "step": 342
+    },
+    {
+      "epoch": 0.21952,
+      "grad_norm": 0.2851983974846401,
+      "learning_rate": 2.773156121970929e-05,
+      "loss": 0.2401,
+      "step": 343
+    },
+    {
+      "epoch": 0.22016,
+      "grad_norm": 3.444718072231663,
+      "learning_rate": 2.7714730725620786e-05,
+      "loss": 0.2706,
+      "step": 344
+    },
+    {
+      "epoch": 0.2208,
+      "grad_norm": 0.41628559582587626,
+      "learning_rate": 2.7697843172493925e-05,
+      "loss": 0.2732,
+      "step": 345
+    },
+    {
+      "epoch": 0.22144,
+      "grad_norm": 0.33063598183417514,
+      "learning_rate": 2.7680898636113845e-05,
+      "loss": 0.2775,
+      "step": 346
+    },
+    {
+      "epoch": 0.22208,
+      "grad_norm": 0.3317549477105874,
+      "learning_rate": 2.7663897192521393e-05,
+      "loss": 0.2673,
+      "step": 347
+    },
+    {
+      "epoch": 0.22272,
+      "grad_norm": 0.39451049358358514,
+      "learning_rate": 2.764683891801281e-05,
+      "loss": 0.2853,
+      "step": 348
+    },
+    {
+      "epoch": 0.22336,
+      "grad_norm": 0.33825604933869896,
+      "learning_rate": 2.7629723889139345e-05,
+      "loss": 0.2836,
+      "step": 349
+    },
+    {
+      "epoch": 0.224,
+      "grad_norm": 0.3938596661138968,
+      "learning_rate": 2.7612552182706968e-05,
+      "loss": 0.2679,
+      "step": 350
+    },
+    {
+      "epoch": 0.22464,
+      "grad_norm": 0.3709101970004551,
+      "learning_rate": 2.759532387577599e-05,
+      "loss": 0.2819,
+      "step": 351
+    },
+    {
+      "epoch": 0.22528,
+      "grad_norm": 0.35417172495939264,
+      "learning_rate": 2.7578039045660713e-05,
+      "loss": 0.2639,
+      "step": 352
+    },
+    {
+      "epoch": 0.22592,
+      "grad_norm": 0.3327176351302671,
+      "learning_rate": 2.75606977699291e-05,
+      "loss": 0.2639,
+      "step": 353
+    },
+    {
+      "epoch": 0.22656,
+      "grad_norm": 0.3564326153038442,
+      "learning_rate": 2.754330012640243e-05,
+      "loss": 0.2556,
+      "step": 354
+    },
+    {
+      "epoch": 0.2272,
+      "grad_norm": 0.32512504523834834,
+      "learning_rate": 2.752584619315493e-05,
+      "loss": 0.2643,
+      "step": 355
+    },
+    {
+      "epoch": 0.22784,
+      "grad_norm": 0.3882343021011366,
+      "learning_rate": 2.7508336048513437e-05,
+      "loss": 0.2786,
+      "step": 356
+    },
+    {
+      "epoch": 0.22848,
+      "grad_norm": 0.3645455650133275,
+      "learning_rate": 2.7490769771057043e-05,
+      "loss": 0.2642,
+      "step": 357
+    },
+    {
+      "epoch": 0.22912,
+      "grad_norm": 0.36026260628073103,
+      "learning_rate": 2.747314743961675e-05,
+      "loss": 0.2639,
+      "step": 358
+    },
+    {
+      "epoch": 0.22976,
+      "grad_norm": 0.3515677194617887,
+      "learning_rate": 2.7455469133275095e-05,
+      "loss": 0.2814,
+      "step": 359
+    },
+    {
+      "epoch": 0.2304,
+      "grad_norm": 0.34391091885619035,
+      "learning_rate": 2.743773493136583e-05,
+      "loss": 0.2673,
+      "step": 360
+    },
+    {
+      "epoch": 0.23104,
+      "grad_norm": 0.3449879915659821,
+      "learning_rate": 2.7419944913473533e-05,
+      "loss": 0.2333,
+      "step": 361
+    },
+    {
+      "epoch": 0.23168,
+      "grad_norm": 0.31990363584739007,
+      "learning_rate": 2.7402099159433258e-05,
+      "loss": 0.2532,
+      "step": 362
+    },
+    {
+      "epoch": 0.23232,
+      "grad_norm": 0.3875818304841908,
+      "learning_rate": 2.73841977493302e-05,
+      "loss": 0.2811,
+      "step": 363
+    },
+    {
+      "epoch": 0.23296,
+      "grad_norm": 0.3566549850839257,
+      "learning_rate": 2.7366240763499302e-05,
+      "loss": 0.2507,
+      "step": 364
+    },
+    {
+      "epoch": 0.2336,
+      "grad_norm": 0.3080698897167787,
+      "learning_rate": 2.7348228282524916e-05,
+      "loss": 0.2603,
+      "step": 365
+    },
+    {
+      "epoch": 0.23424,
+      "grad_norm": 0.3357967412245159,
+      "learning_rate": 2.7330160387240443e-05,
+      "loss": 0.2451,
+      "step": 366
+    },
+    {
+      "epoch": 0.23488,
+      "grad_norm": 0.3527005868734707,
+      "learning_rate": 2.731203715872795e-05,
+      "loss": 0.275,
+      "step": 367
+    },
+    {
+      "epoch": 0.23552,
+      "grad_norm": 0.316985061769891,
+      "learning_rate": 2.729385867831783e-05,
+      "loss": 0.269,
+      "step": 368
+    },
+    {
+      "epoch": 0.23616,
+      "grad_norm": 0.33344202384537386,
+      "learning_rate": 2.7275625027588414e-05,
+      "loss": 0.2691,
+      "step": 369
+    },
+    {
+      "epoch": 0.2368,
+      "grad_norm": 0.34625417830177385,
+      "learning_rate": 2.7257336288365634e-05,
+      "loss": 0.2653,
+      "step": 370
+    },
+    {
+      "epoch": 0.23744,
+      "grad_norm": 0.3267922712231095,
+      "learning_rate": 2.7238992542722625e-05,
+      "loss": 0.2617,
+      "step": 371
+    },
+    {
+      "epoch": 0.23808,
+      "grad_norm": 0.330793248709517,
+      "learning_rate": 2.722059387297938e-05,
+      "loss": 0.2574,
+      "step": 372
+    },
+    {
+      "epoch": 0.23872,
+      "grad_norm": 0.3463369707262466,
+      "learning_rate": 2.720214036170236e-05,
+      "loss": 0.2627,
+      "step": 373
+    },
+    {
+      "epoch": 0.23936,
+      "grad_norm": 0.33043396116356993,
+      "learning_rate": 2.7183632091704143e-05,
+      "loss": 0.2708,
+      "step": 374
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.3485722629331789,
+      "learning_rate": 2.716506914604305e-05,
+      "loss": 0.2612,
+      "step": 375
+    },
+    {
+      "epoch": 0.24064,
+      "grad_norm": 0.3412635617995565,
+      "learning_rate": 2.7146451608022748e-05,
+      "loss": 0.2607,
+      "step": 376
+    },
+    {
+      "epoch": 0.24128,
+      "grad_norm": 0.35685994147745786,
+      "learning_rate": 2.7127779561191905e-05,
+      "loss": 0.2577,
+      "step": 377
+    },
+    {
+      "epoch": 0.24192,
+      "grad_norm": 0.3039380817502753,
+      "learning_rate": 2.7109053089343815e-05,
+      "loss": 0.2611,
+      "step": 378
+    },
+    {
+      "epoch": 0.24256,
+      "grad_norm": 0.43279102042200523,
+      "learning_rate": 2.7090272276515993e-05,
+      "loss": 0.2756,
+      "step": 379
+    },
+    {
+      "epoch": 0.2432,
+      "grad_norm": 0.31054351672072755,
+      "learning_rate": 2.707143720698983e-05,
+      "loss": 0.2708,
+      "step": 380
+    },
+    {
+      "epoch": 0.24384,
+      "grad_norm": 0.36851176929762747,
+      "learning_rate": 2.7052547965290186e-05,
+      "loss": 0.276,
+      "step": 381
+    },
+    {
+      "epoch": 0.24448,
+      "grad_norm": 0.3627663453356185,
+      "learning_rate": 2.703360463618504e-05,
+      "loss": 0.2732,
+      "step": 382
+    },
+    {
+      "epoch": 0.24512,
+      "grad_norm": 0.3576702828639981,
+      "learning_rate": 2.7014607304685096e-05,
+      "loss": 0.2625,
+      "step": 383
+    },
+    {
+      "epoch": 0.24576,
+      "grad_norm": 0.3437585046911205,
+      "learning_rate": 2.699555605604339e-05,
+      "loss": 0.2717,
+      "step": 384
+    },
+    {
+      "epoch": 0.2464,
+      "grad_norm": 0.3563941457915153,
+      "learning_rate": 2.6976450975754923e-05,
+      "loss": 0.2599,
+      "step": 385
+    },
+    {
+      "epoch": 0.24704,
+      "grad_norm": 0.32499802120658194,
+      "learning_rate": 2.6957292149556276e-05,
+      "loss": 0.2677,
+      "step": 386
+    },
+    {
+      "epoch": 0.24768,
+      "grad_norm": 0.33344191861008443,
+      "learning_rate": 2.6938079663425218e-05,
+      "loss": 0.2882,
+      "step": 387
+    },
+    {
+      "epoch": 0.24832,
+      "grad_norm": 0.3517570106726394,
+      "learning_rate": 2.691881360358033e-05,
+      "loss": 0.2754,
+      "step": 388
+    },
+    {
+      "epoch": 0.24896,
+      "grad_norm": 0.3237247905400656,
+      "learning_rate": 2.6899494056480596e-05,
+      "loss": 0.2545,
+      "step": 389
+    },
+    {
+      "epoch": 0.2496,
+      "grad_norm": 0.3605247711441668,
+      "learning_rate": 2.6880121108825056e-05,
+      "loss": 0.2858,
+      "step": 390
+    },
+    {
+      "epoch": 0.25024,
+      "grad_norm": 0.3227764776694619,
+      "learning_rate": 2.6860694847552374e-05,
+      "loss": 0.2588,
+      "step": 391
+    },
+    {
+      "epoch": 0.25088,
+      "grad_norm": 0.30914534952022565,
+      "learning_rate": 2.6841215359840468e-05,
+      "loss": 0.265,
+      "step": 392
+    },
+    {
+      "epoch": 0.25152,
+      "grad_norm": 0.3040468595020032,
+      "learning_rate": 2.682168273310612e-05,
+      "loss": 0.2572,
+      "step": 393
+    },
+    {
+      "epoch": 0.25216,
+      "grad_norm": 0.3582628170978083,
+      "learning_rate": 2.680209705500458e-05,
+      "loss": 0.2692,
+      "step": 394
+    },
+    {
+      "epoch": 0.2528,
+      "grad_norm": 0.3288877765942096,
+      "learning_rate": 2.6782458413429177e-05,
+      "loss": 0.2556,
+      "step": 395
+    },
+    {
+      "epoch": 0.25344,
+      "grad_norm": 0.3483940416687742,
+      "learning_rate": 2.6762766896510914e-05,
+      "loss": 0.2509,
+      "step": 396
+    },
+    {
+      "epoch": 0.25408,
+      "grad_norm": 0.4148522716152593,
+      "learning_rate": 2.6743022592618075e-05,
+      "loss": 0.2855,
+      "step": 397
+    },
+    {
+      "epoch": 0.25472,
+      "grad_norm": 0.3153140469461542,
+      "learning_rate": 2.6723225590355852e-05,
+      "loss": 0.2422,
+      "step": 398
+    },
+    {
+      "epoch": 0.25536,
+      "grad_norm": 0.3414987747573693,
+      "learning_rate": 2.6703375978565906e-05,
+      "loss": 0.2566,
+      "step": 399
+    },
+    {
+      "epoch": 0.256,
+      "grad_norm": 0.3082723258062477,
+      "learning_rate": 2.6683473846326002e-05,
+      "loss": 0.2692,
+      "step": 400
+    },
+    {
+      "epoch": 0.25664,
+      "grad_norm": 0.3120921043255892,
+      "learning_rate": 2.666351928294959e-05,
+      "loss": 0.2484,
+      "step": 401
+    },
+    {
+      "epoch": 0.25728,
+      "grad_norm": 0.34491023806967935,
+      "learning_rate": 2.6643512377985423e-05,
+      "loss": 0.2634,
+      "step": 402
+    },
+    {
+      "epoch": 0.25792,
+      "grad_norm": 1.227578859205086,
+      "learning_rate": 2.6623453221217137e-05,
+      "loss": 0.2681,
+      "step": 403
+    },
+    {
+      "epoch": 0.25856,
+      "grad_norm": 0.36139208492186886,
+      "learning_rate": 2.660334190266285e-05,
+      "loss": 0.2781,
+      "step": 404
+    },
+    {
+      "epoch": 0.2592,
+      "grad_norm": 0.3146578283750008,
+      "learning_rate": 2.658317851257477e-05,
+      "loss": 0.2661,
+      "step": 405
+    },
+    {
+      "epoch": 0.25984,
+      "grad_norm": 0.3241167195918667,
+      "learning_rate": 2.6562963141438783e-05,
+      "loss": 0.2747,
+      "step": 406
+    },
+    {
+      "epoch": 0.26048,
+      "grad_norm": 0.32983966170598866,
+      "learning_rate": 2.6542695879974044e-05,
+      "loss": 0.2821,
+      "step": 407
+    },
+    {
+      "epoch": 0.26112,
+      "grad_norm": 0.3302761405872256,
+      "learning_rate": 2.652237681913257e-05,
+      "loss": 0.2677,
+      "step": 408
+    },
+    {
+      "epoch": 0.26176,
+      "grad_norm": 0.33848152131881326,
+      "learning_rate": 2.6502006050098842e-05,
+      "loss": 0.2542,
+      "step": 409
+    },
+    {
+      "epoch": 0.2624,
+      "grad_norm": 0.3413235227280745,
+      "learning_rate": 2.648158366428938e-05,
+      "loss": 0.2664,
+      "step": 410
+    },
+    {
+      "epoch": 0.26304,
+      "grad_norm": 0.32916338812223606,
+      "learning_rate": 2.6461109753352355e-05,
+      "loss": 0.2616,
+      "step": 411
+    },
+    {
+      "epoch": 0.26368,
+      "grad_norm": 0.3053892800899407,
+      "learning_rate": 2.6440584409167144e-05,
+      "loss": 0.2583,
+      "step": 412
+    },
+    {
+      "epoch": 0.26432,
+      "grad_norm": 0.3385379902087113,
+      "learning_rate": 2.6420007723843952e-05,
+      "loss": 0.256,
+      "step": 413
+    },
+    {
+      "epoch": 0.26496,
+      "grad_norm": 0.37571936271303413,
+      "learning_rate": 2.639937978972338e-05,
+      "loss": 0.2503,
+      "step": 414
+    },
+    {
+      "epoch": 0.2656,
+      "grad_norm": 0.2854100162554339,
+      "learning_rate": 2.6378700699376015e-05,
+      "loss": 0.2415,
+      "step": 415
+    },
+    {
+      "epoch": 0.26624,
+      "grad_norm": 0.31271921873205044,
+      "learning_rate": 2.6357970545602014e-05,
+      "loss": 0.2475,
+      "step": 416
+    },
+    {
+      "epoch": 0.26688,
+      "grad_norm": 0.34030180044647645,
+      "learning_rate": 2.6337189421430688e-05,
+      "loss": 0.2379,
+      "step": 417
+    },
+    {
+      "epoch": 0.26752,
+      "grad_norm": 0.4075617302343357,
+      "learning_rate": 2.631635742012009e-05,
+      "loss": 0.2742,
+      "step": 418
+    },
+    {
+      "epoch": 0.26816,
+      "grad_norm": 0.3607203044257088,
+      "learning_rate": 2.629547463515657e-05,
+      "loss": 0.2694,
+      "step": 419
+    },
+    {
+      "epoch": 0.2688,
+      "grad_norm": 0.3244858888191209,
+      "learning_rate": 2.6274541160254405e-05,
+      "loss": 0.2633,
+      "step": 420
+    },
+    {
+      "epoch": 0.26944,
+      "grad_norm": 0.3940598188912196,
+      "learning_rate": 2.6253557089355333e-05,
+      "loss": 0.2466,
+      "step": 421
+    },
+    {
+      "epoch": 0.27008,
+      "grad_norm": 0.3011262776053056,
+      "learning_rate": 2.6232522516628153e-05,
+      "loss": 0.2727,
+      "step": 422
+    },
+    {
+      "epoch": 0.27072,
+      "grad_norm": 0.3739061295982007,
+      "learning_rate": 2.6211437536468292e-05,
+      "loss": 0.2671,
+      "step": 423
+    },
+    {
+      "epoch": 0.27136,
+      "grad_norm": 0.32945664324477797,
+      "learning_rate": 2.6190302243497396e-05,
+      "loss": 0.2708,
+      "step": 424
+    },
+    {
+      "epoch": 0.272,
+      "grad_norm": 0.36534937038877574,
+      "learning_rate": 2.6169116732562892e-05,
+      "loss": 0.2298,
+      "step": 425
+    },
+    {
+      "epoch": 0.27264,
+      "grad_norm": 0.33670531669484066,
+      "learning_rate": 2.6147881098737562e-05,
+      "loss": 0.2681,
+      "step": 426
+    },
+    {
+      "epoch": 0.27328,
+      "grad_norm": 0.35965800175602797,
+      "learning_rate": 2.612659543731913e-05,
+      "loss": 0.2675,
+      "step": 427
+    },
+    {
+      "epoch": 0.27392,
+      "grad_norm": 0.3509912783523724,
+      "learning_rate": 2.610525984382982e-05,
+      "loss": 0.2752,
+      "step": 428
+    },
+    {
+      "epoch": 0.27456,
+      "grad_norm": 0.3341800947139574,
+      "learning_rate": 2.6083874414015934e-05,
+      "loss": 0.2651,
+      "step": 429
+    },
+    {
+      "epoch": 0.2752,
+      "grad_norm": 0.3512265177853723,
+      "learning_rate": 2.606243924384742e-05,
+      "loss": 0.2746,
+      "step": 430
+    },
+    {
+      "epoch": 0.27584,
+      "grad_norm": 0.3314007433299324,
+      "learning_rate": 2.6040954429517442e-05,
+      "loss": 0.2406,
+      "step": 431
+    },
+    {
+      "epoch": 0.27648,
+      "grad_norm": 0.31623472637208033,
+      "learning_rate": 2.6019420067441958e-05,
+      "loss": 0.2432,
+      "step": 432
+    },
+    {
+      "epoch": 0.27712,
+      "grad_norm": 0.33629626691329856,
+      "learning_rate": 2.599783625425926e-05,
+      "loss": 0.2642,
+      "step": 433
+    },
+    {
+      "epoch": 0.27776,
+      "grad_norm": 0.32932652640189397,
+      "learning_rate": 2.597620308682957e-05,
+      "loss": 0.2722,
+      "step": 434
+    },
+    {
+      "epoch": 0.2784,
+      "grad_norm": 0.3162553853731888,
+      "learning_rate": 2.5954520662234602e-05,
+      "loss": 0.253,
+      "step": 435
+    },
+    {
+      "epoch": 0.27904,
+      "grad_norm": 0.32294208690580367,
+      "learning_rate": 2.5932789077777103e-05,
+      "loss": 0.2453,
+      "step": 436
+    },
+    {
+      "epoch": 0.27968,
+      "grad_norm": 0.2860688977348506,
+      "learning_rate": 2.591100843098043e-05,
+      "loss": 0.2583,
+      "step": 437
+    },
+    {
+      "epoch": 0.28032,
+      "grad_norm": 0.3382649825825897,
+      "learning_rate": 2.5889178819588134e-05,
+      "loss": 0.243,
+      "step": 438
+    },
+    {
+      "epoch": 0.28096,
+      "grad_norm": 0.3426287563355019,
+      "learning_rate": 2.5867300341563477e-05,
+      "loss": 0.2832,
+      "step": 439
+    },
+    {
+      "epoch": 0.2816,
+      "grad_norm": 0.3616023005679438,
+      "learning_rate": 2.5845373095089028e-05,
+      "loss": 0.2586,
+      "step": 440
+    },
+    {
+      "epoch": 0.28224,
+      "grad_norm": 0.3680942120537055,
+      "learning_rate": 2.5823397178566217e-05,
+      "loss": 0.2449,
+      "step": 441
+    },
+    {
+      "epoch": 0.28288,
+      "grad_norm": 0.3266840109358284,
+      "learning_rate": 2.580137269061488e-05,
+      "loss": 0.2806,
+      "step": 442
+    },
+    {
+      "epoch": 0.28352,
+      "grad_norm": 0.3268151383316364,
+      "learning_rate": 2.5779299730072815e-05,
+      "loss": 0.264,
+      "step": 443
+    },
+    {
+      "epoch": 0.28416,
+      "grad_norm": 0.32651398276400556,
+      "learning_rate": 2.5757178395995358e-05,
+      "loss": 0.2672,
+      "step": 444
+    },
+    {
+      "epoch": 0.2848,
+      "grad_norm": 0.35504343797167337,
+      "learning_rate": 2.5735008787654933e-05,
+      "loss": 0.2563,
+      "step": 445
+    },
+    {
+      "epoch": 0.28544,
+      "grad_norm": 0.3332900931306675,
+      "learning_rate": 2.5712791004540592e-05,
+      "loss": 0.2679,
+      "step": 446
+    },
+    {
+      "epoch": 0.28608,
+      "grad_norm": 0.32153395285613084,
+      "learning_rate": 2.5690525146357575e-05,
+      "loss": 0.2572,
+      "step": 447
+    },
+    {
+      "epoch": 0.28672,
+      "grad_norm": 0.33090800287138983,
+      "learning_rate": 2.566821131302688e-05,
+      "loss": 0.2834,
+      "step": 448
+    },
+    {
+      "epoch": 0.28736,
+      "grad_norm": 0.336200932174992,
+      "learning_rate": 2.5645849604684775e-05,
+      "loss": 0.248,
+      "step": 449
+    },
+    {
+      "epoch": 0.288,
+      "grad_norm": 0.32237186462753925,
+      "learning_rate": 2.56234401216824e-05,
+      "loss": 0.2665,
+      "step": 450
+    },
+    {
+      "epoch": 0.28864,
+      "grad_norm": 0.3195380119597909,
+      "learning_rate": 2.5600982964585272e-05,
+      "loss": 0.2508,
+      "step": 451
+    },
+    {
+      "epoch": 0.28928,
+      "grad_norm": 0.3475060876563337,
+      "learning_rate": 2.5578478234172865e-05,
+      "loss": 0.2594,
+      "step": 452
+    },
+    {
+      "epoch": 0.28992,
+      "grad_norm": 0.627047630421656,
+      "learning_rate": 2.5555926031438134e-05,
+      "loss": 0.2485,
+      "step": 453
+    },
+    {
+      "epoch": 0.29056,
+      "grad_norm": 0.3005632921400129,
+      "learning_rate": 2.5533326457587072e-05,
+      "loss": 0.2363,
+      "step": 454
+    },
+    {
+      "epoch": 0.2912,
+      "grad_norm": 0.3907484843122131,
+      "learning_rate": 2.551067961403827e-05,
+      "loss": 0.2635,
+      "step": 455
+    },
+    {
+      "epoch": 0.29184,
+      "grad_norm": 0.30910241081220835,
+      "learning_rate": 2.5487985602422425e-05,
+      "loss": 0.2554,
+      "step": 456
+    },
+    {
+      "epoch": 0.29248,
+      "grad_norm": 0.3316657790671481,
+      "learning_rate": 2.546524452458193e-05,
+      "loss": 0.2465,
+      "step": 457
+    },
+    {
+      "epoch": 0.29312,
+      "grad_norm": 0.33514519152514977,
+      "learning_rate": 2.5442456482570383e-05,
+      "loss": 0.2509,
+      "step": 458
+    },
+    {
+      "epoch": 0.29376,
+      "grad_norm": 0.3135722919118478,
+      "learning_rate": 2.5419621578652137e-05,
+      "loss": 0.2584,
+      "step": 459
+    },
+    {
+      "epoch": 0.2944,
+      "grad_norm": 0.8767929120277721,
+      "learning_rate": 2.539673991530185e-05,
+      "loss": 0.2649,
+      "step": 460
+    },
+    {
+      "epoch": 0.29504,
+      "grad_norm": 0.32933106915081395,
+      "learning_rate": 2.5373811595204022e-05,
+      "loss": 0.2727,
+      "step": 461
+    },
+    {
+      "epoch": 0.29568,
+      "grad_norm": 0.4082581717544081,
+      "learning_rate": 2.5350836721252525e-05,
+      "loss": 0.2569,
+      "step": 462
+    },
+    {
+      "epoch": 0.29632,
+      "grad_norm": 0.3412102305943323,
+      "learning_rate": 2.5327815396550142e-05,
+      "loss": 0.2639,
+      "step": 463
+    },
+    {
+      "epoch": 0.29696,
+      "grad_norm": 0.31780550108168804,
+      "learning_rate": 2.530474772440812e-05,
+      "loss": 0.2418,
+      "step": 464
+    },
+    {
+      "epoch": 0.2976,
+      "grad_norm": 0.29529324755348607,
+      "learning_rate": 2.5281633808345702e-05,
+      "loss": 0.2471,
+      "step": 465
+    },
+    {
+      "epoch": 0.29824,
+      "grad_norm": 0.33728330149501135,
+      "learning_rate": 2.5258473752089636e-05,
+      "loss": 0.2801,
+      "step": 466
+    },
+    {
+      "epoch": 0.29888,
+      "grad_norm": 0.3267488434365671,
+      "learning_rate": 2.5235267659573746e-05,
+      "loss": 0.2414,
+      "step": 467
+    },
+    {
+      "epoch": 0.29952,
+      "grad_norm": 0.29150985108582134,
+      "learning_rate": 2.521201563493845e-05,
+      "loss": 0.265,
+      "step": 468
+    },
+    {
+      "epoch": 0.30016,
+      "grad_norm": 0.30336323784858,
+      "learning_rate": 2.5188717782530292e-05,
+      "loss": 0.2593,
+      "step": 469
+    },
+    {
+      "epoch": 0.3008,
+      "grad_norm": 0.29926798408682637,
+      "learning_rate": 2.516537420690146e-05,
+      "loss": 0.2604,
+      "step": 470
+    },
+    {
+      "epoch": 0.30144,
+      "grad_norm": 0.3065859162009593,
+      "learning_rate": 2.514198501280934e-05,
+      "loss": 0.2629,
+      "step": 471
+    },
+    {
+      "epoch": 0.30208,
+      "grad_norm": 0.3049740162502653,
+      "learning_rate": 2.5118550305216054e-05,
+      "loss": 0.2754,
+      "step": 472
+    },
+    {
+      "epoch": 0.30272,
+      "grad_norm": 0.30876372307121774,
+      "learning_rate": 2.5095070189287944e-05,
+      "loss": 0.2668,
+      "step": 473
+    },
+    {
+      "epoch": 0.30336,
+      "grad_norm": 0.29273269101927585,
+      "learning_rate": 2.5071544770395143e-05,
+      "loss": 0.2558,
+      "step": 474
+    },
+    {
+      "epoch": 0.304,
+      "grad_norm": 0.3076653994547958,
+      "learning_rate": 2.5047974154111092e-05,
+      "loss": 0.2489,
+      "step": 475
+    },
+    {
+      "epoch": 0.30464,
+      "grad_norm": 0.3333719011093153,
+      "learning_rate": 2.5024358446212046e-05,
+      "loss": 0.2435,
+      "step": 476
+    },
+    {
+      "epoch": 0.30528,
+      "grad_norm": 0.38541240875825966,
+      "learning_rate": 2.5000697752676622e-05,
+      "loss": 0.2791,
+      "step": 477
+    },
+    {
+      "epoch": 0.30592,
+      "grad_norm": 0.31076311872010254,
+      "learning_rate": 2.4976992179685317e-05,
+      "loss": 0.2693,
+      "step": 478
+    },
+    {
+      "epoch": 0.30656,
+      "grad_norm": 0.3140598892806967,
+      "learning_rate": 2.4953241833620034e-05,
+      "loss": 0.2453,
+      "step": 479
+    },
+    {
+      "epoch": 0.3072,
+      "grad_norm": 0.30866549699497225,
+      "learning_rate": 2.4929446821063596e-05,
+      "loss": 0.2494,
+      "step": 480
+    },
+    {
+      "epoch": 0.30784,
+      "grad_norm": 0.3057358078354188,
+      "learning_rate": 2.4905607248799265e-05,
+      "loss": 0.2583,
+      "step": 481
+    },
+    {
+      "epoch": 0.30848,
+      "grad_norm": 0.2832731794280749,
+      "learning_rate": 2.4881723223810295e-05,
+      "loss": 0.2687,
+      "step": 482
+    },
+    {
+      "epoch": 0.30912,
+      "grad_norm": 0.3051553661664977,
+      "learning_rate": 2.4857794853279396e-05,
+      "loss": 0.2599,
+      "step": 483
+    },
+    {
+      "epoch": 0.30976,
+      "grad_norm": 0.2998378381048481,
+      "learning_rate": 2.4833822244588312e-05,
+      "loss": 0.2451,
+      "step": 484
+    },
+    {
+      "epoch": 0.3104,
+      "grad_norm": 0.3174813787668605,
+      "learning_rate": 2.4809805505317296e-05,
+      "loss": 0.264,
+      "step": 485
+    },
+    {
+      "epoch": 0.31104,
+      "grad_norm": 0.28982862061791764,
+      "learning_rate": 2.4785744743244644e-05,
+      "loss": 0.2478,
+      "step": 486
+    },
+    {
+      "epoch": 0.31168,
+      "grad_norm": 0.3147239609211115,
+      "learning_rate": 2.4761640066346217e-05,
+      "loss": 0.2372,
+      "step": 487
+    },
+    {
+      "epoch": 0.31232,
+      "grad_norm": 0.2898871152482304,
+      "learning_rate": 2.4737491582794945e-05,
+      "loss": 0.2638,
+      "step": 488
+    },
+    {
+      "epoch": 0.31296,
+      "grad_norm": 0.33890483931828097,
+      "learning_rate": 2.4713299400960342e-05,
+      "loss": 0.2444,
+      "step": 489
+    },
+    {
+      "epoch": 0.3136,
+      "grad_norm": 0.28327565279880107,
+      "learning_rate": 2.4689063629408034e-05,
+      "loss": 0.272,
+      "step": 490
+    },
+    {
+      "epoch": 0.31424,
+      "grad_norm": 0.3212172431491676,
+      "learning_rate": 2.4664784376899257e-05,
+      "loss": 0.2545,
+      "step": 491
+    },
+    {
+      "epoch": 0.31488,
+      "grad_norm": 0.3367903918403918,
+      "learning_rate": 2.4640461752390367e-05,
+      "loss": 0.2632,
+      "step": 492
+    },
+    {
+      "epoch": 0.31552,
+      "grad_norm": 0.3230413061531243,
+      "learning_rate": 2.4616095865032366e-05,
+      "loss": 0.2825,
+      "step": 493
+    },
+    {
+      "epoch": 0.31616,
+      "grad_norm": 0.34034207513600895,
+      "learning_rate": 2.459168682417041e-05,
+      "loss": 0.271,
+      "step": 494
+    },
+    {
+      "epoch": 0.3168,
+      "grad_norm": 0.32141809222820283,
+      "learning_rate": 2.4567234739343283e-05,
+      "loss": 0.2396,
+      "step": 495
+    },
+    {
+      "epoch": 0.31744,
+      "grad_norm": 0.2753406394259948,
+      "learning_rate": 2.454273972028297e-05,
+      "loss": 0.2404,
+      "step": 496
+    },
+    {
+      "epoch": 0.31808,
+      "grad_norm": 0.3351114179983899,
+      "learning_rate": 2.451820187691411e-05,
+      "loss": 0.2763,
+      "step": 497
+    },
+    {
+      "epoch": 0.31872,
+      "grad_norm": 0.3513219675385064,
+      "learning_rate": 2.4493621319353525e-05,
+      "loss": 0.2609,
+      "step": 498
+    },
+    {
+      "epoch": 0.31936,
+      "grad_norm": 0.3110180234096309,
+      "learning_rate": 2.4468998157909723e-05,
+      "loss": 0.2402,
+      "step": 499
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.2964890570904897,
+      "learning_rate": 2.444433250308241e-05,
+      "loss": 0.2659,
+      "step": 500
+    },
+    {
+      "epoch": 0.32064,
+      "grad_norm": 0.3579764939490037,
+      "learning_rate": 2.4419624465561964e-05,
+      "loss": 0.2514,
+      "step": 501
+    },
+    {
+      "epoch": 0.32128,
+      "grad_norm": 0.3144597113409246,
+      "learning_rate": 2.4394874156228988e-05,
+      "loss": 0.2608,
+      "step": 502
+    },
+    {
+      "epoch": 0.32192,
+      "grad_norm": 0.31609390625354067,
+      "learning_rate": 2.4370081686153767e-05,
+      "loss": 0.2443,
+      "step": 503
+    },
+    {
+      "epoch": 0.32256,
+      "grad_norm": 0.29746982935843747,
+      "learning_rate": 2.4345247166595803e-05,
+      "loss": 0.2377,
+      "step": 504
+    },
+    {
+      "epoch": 0.3232,
+      "grad_norm": 0.34126278199611504,
+      "learning_rate": 2.4320370709003284e-05,
+      "loss": 0.2554,
+      "step": 505
+    },
+    {
+      "epoch": 0.32384,
+      "grad_norm": 0.2990237684037871,
+      "learning_rate": 2.429545242501261e-05,
+      "loss": 0.2848,
+      "step": 506
+    },
+    {
+      "epoch": 0.32448,
+      "grad_norm": 0.32322206092320094,
+      "learning_rate": 2.4270492426447884e-05,
+      "loss": 0.2561,
+      "step": 507
+    },
+    {
+      "epoch": 0.32512,
+      "grad_norm": 0.3260943452277802,
+      "learning_rate": 2.424549082532041e-05,
+      "loss": 0.26,
+      "step": 508
+    },
+    {
+      "epoch": 0.32576,
+      "grad_norm": 0.4295758739365506,
+      "learning_rate": 2.422044773382817e-05,
+      "loss": 0.2513,
+      "step": 509
+    },
+    {
+      "epoch": 0.3264,
+      "grad_norm": 0.31835152340497586,
+      "learning_rate": 2.4195363264355365e-05,
+      "loss": 0.2505,
+      "step": 510
+    },
+    {
+      "epoch": 0.32704,
+      "grad_norm": 0.3531418180586244,
+      "learning_rate": 2.417023752947188e-05,
+      "loss": 0.2609,
+      "step": 511
+    },
+    {
+      "epoch": 0.32768,
+      "grad_norm": 0.28968278915332163,
+      "learning_rate": 2.4145070641932767e-05,
+      "loss": 0.2472,
+      "step": 512
+    },
+    {
+      "epoch": 0.32832,
+      "grad_norm": 0.3176579318606065,
+      "learning_rate": 2.4119862714677773e-05,
+      "loss": 0.2563,
+      "step": 513
+    },
+    {
+      "epoch": 0.32896,
+      "grad_norm": 0.3447755475779212,
+      "learning_rate": 2.4094613860830813e-05,
+      "loss": 0.299,
+      "step": 514
+    },
+    {
+      "epoch": 0.3296,
+      "grad_norm": 0.30246755944222914,
+      "learning_rate": 2.4069324193699453e-05,
+      "loss": 0.2285,
+      "step": 515
+    },
+    {
+      "epoch": 0.33024,
+      "grad_norm": 0.36003409699329864,
+      "learning_rate": 2.4043993826774433e-05,
+      "loss": 0.2473,
+      "step": 516
+    },
+    {
+      "epoch": 0.33088,
+      "grad_norm": 0.35159179010717023,
+      "learning_rate": 2.4018622873729136e-05,
+      "loss": 0.2654,
+      "step": 517
+    },
+    {
+      "epoch": 0.33152,
+      "grad_norm": 0.30073111767960253,
+      "learning_rate": 2.3993211448419055e-05,
+      "loss": 0.249,
+      "step": 518
+    },
+    {
+      "epoch": 0.33216,
+      "grad_norm": 0.3319898092071986,
+      "learning_rate": 2.3967759664881347e-05,
+      "loss": 0.2737,
+      "step": 519
+    },
+    {
+      "epoch": 0.3328,
+      "grad_norm": 0.4467760115064727,
+      "learning_rate": 2.394226763733425e-05,
+      "loss": 0.2585,
+      "step": 520
+    },
+    {
+      "epoch": 0.33344,
+      "grad_norm": 0.3041939438386619,
+      "learning_rate": 2.3916735480176618e-05,
+      "loss": 0.2266,
+      "step": 521
+    },
+    {
+      "epoch": 0.33408,
+      "grad_norm": 0.3026737354404247,
+      "learning_rate": 2.389116330798739e-05,
+      "loss": 0.2627,
+      "step": 522
+    },
+    {
+      "epoch": 0.33472,
+      "grad_norm": 0.36457095372526566,
+      "learning_rate": 2.3865551235525073e-05,
+      "loss": 0.2535,
+      "step": 523
+    },
+    {
+      "epoch": 0.33536,
+      "grad_norm": 0.3111554932975257,
+      "learning_rate": 2.3839899377727237e-05,
+      "loss": 0.249,
+      "step": 524
+    },
+    {
+      "epoch": 0.336,
+      "grad_norm": 0.2873311087186447,
+      "learning_rate": 2.381420784970999e-05,
+      "loss": 0.2808,
+      "step": 525
+    },
+    {
+      "epoch": 0.33664,
+      "grad_norm": 0.31448944336119605,
+      "learning_rate": 2.378847676676747e-05,
+      "loss": 0.2599,
+      "step": 526
+    },
+    {
+      "epoch": 0.33728,
+      "grad_norm": 0.3105232703780657,
+      "learning_rate": 2.3762706244371314e-05,
+      "loss": 0.2475,
+      "step": 527
+    },
+    {
+      "epoch": 0.33792,
+      "grad_norm": 0.2918281600818534,
+      "learning_rate": 2.3736896398170163e-05,
+      "loss": 0.2481,
+      "step": 528
+    },
+    {
+      "epoch": 0.33856,
+      "grad_norm": 0.3139402887967583,
+      "learning_rate": 2.3711047343989115e-05,
+      "loss": 0.2751,
+      "step": 529
+    },
+    {
+      "epoch": 0.3392,
+      "grad_norm": 0.2999435271292121,
+      "learning_rate": 2.368515919782923e-05,
+      "loss": 0.2414,
+      "step": 530
+    },
+    {
+      "epoch": 0.33984,
+      "grad_norm": 0.31432103076673396,
+      "learning_rate": 2.3659232075866982e-05,
+      "loss": 0.2571,
+      "step": 531
+    },
+    {
+      "epoch": 0.34048,
+      "grad_norm": 0.3208208970926117,
+      "learning_rate": 2.363326609445378e-05,
+      "loss": 0.2523,
+      "step": 532
+    },
+    {
+      "epoch": 0.34112,
+      "grad_norm": 0.3366034671455189,
+      "learning_rate": 2.3607261370115397e-05,
+      "loss": 0.2643,
+      "step": 533
+    },
+    {
+      "epoch": 0.34176,
+      "grad_norm": 0.30511352998629143,
+      "learning_rate": 2.3581218019551475e-05,
+      "loss": 0.2528,
+      "step": 534
+    },
+    {
+      "epoch": 0.3424,
+      "grad_norm": 0.2985896860812654,
+      "learning_rate": 2.3555136159635e-05,
+      "loss": 0.2657,
+      "step": 535
+    },
+    {
+      "epoch": 0.34304,
+      "grad_norm": 0.2900503900856783,
+      "learning_rate": 2.3529015907411782e-05,
+      "loss": 0.2503,
+      "step": 536
+    },
+    {
+      "epoch": 0.34368,
+      "grad_norm": 0.28173595548277097,
+      "learning_rate": 2.3502857380099907e-05,
+      "loss": 0.2681,
+      "step": 537
+    },
+    {
+      "epoch": 0.34432,
+      "grad_norm": 0.2998778456549225,
+      "learning_rate": 2.3476660695089232e-05,
+      "loss": 0.251,
+      "step": 538
+    },
+    {
+      "epoch": 0.34496,
+      "grad_norm": 0.30133046241449024,
+      "learning_rate": 2.345042596994085e-05,
+      "loss": 0.2683,
+      "step": 539
+    },
+    {
+      "epoch": 0.3456,
+      "grad_norm": 0.339364481773199,
+      "learning_rate": 2.3424153322386567e-05,
+      "loss": 0.2386,
+      "step": 540
+    },
+    {
+      "epoch": 0.34624,
+      "grad_norm": 0.30211757676321005,
+      "learning_rate": 2.3397842870328366e-05,
+      "loss": 0.2395,
+      "step": 541
+    },
+    {
+      "epoch": 0.34688,
+      "grad_norm": 0.28986596928919606,
+      "learning_rate": 2.3371494731837888e-05,
+      "loss": 0.2388,
+      "step": 542
+    },
+    {
+      "epoch": 0.34752,
+      "grad_norm": 0.2995992340705576,
+      "learning_rate": 2.334510902515589e-05,
+      "loss": 0.2653,
+      "step": 543
+    },
+    {
+      "epoch": 0.34816,
+      "grad_norm": 0.3101278774802601,
+      "learning_rate": 2.331868586869174e-05,
+      "loss": 0.2432,
+      "step": 544
+    },
+    {
+      "epoch": 0.3488,
+      "grad_norm": 0.3244592292118408,
+      "learning_rate": 2.329222538102284e-05,
+      "loss": 0.2542,
+      "step": 545
+    },
+    {
+      "epoch": 0.34944,
+      "grad_norm": 0.3283217812360253,
+      "learning_rate": 2.326572768089413e-05,
+      "loss": 0.2633,
+      "step": 546
+    },
+    {
+      "epoch": 0.35008,
+      "grad_norm": 0.29818778249713357,
+      "learning_rate": 2.3239192887217557e-05,
+      "loss": 0.2322,
+      "step": 547
+    },
+    {
+      "epoch": 0.35072,
+      "grad_norm": 0.30402576136874054,
+      "learning_rate": 2.3212621119071524e-05,
+      "loss": 0.2595,
+      "step": 548
+    },
+    {
+      "epoch": 0.35136,
+      "grad_norm": 0.3102857808050957,
+      "learning_rate": 2.318601249570035e-05,
+      "loss": 0.2395,
+      "step": 549
+    },
+    {
+      "epoch": 0.352,
+      "grad_norm": 0.3212647408602525,
+      "learning_rate": 2.315936713651378e-05,
+      "loss": 0.2465,
+      "step": 550
+    },
+    {
+      "epoch": 0.35264,
+      "grad_norm": 0.31569595485115426,
+      "learning_rate": 2.3132685161086363e-05,
+      "loss": 0.2534,
+      "step": 551
+    },
+    {
+      "epoch": 0.35328,
+      "grad_norm": 0.28971716623099575,
+      "learning_rate": 2.3105966689157016e-05,
+      "loss": 0.2561,
+      "step": 552
+    },
+    {
+      "epoch": 0.35392,
+      "grad_norm": 0.2801164703529546,
+      "learning_rate": 2.3079211840628424e-05,
+      "loss": 0.2464,
+      "step": 553
+    },
+    {
+      "epoch": 0.35456,
+      "grad_norm": 0.3186204827863453,
+      "learning_rate": 2.30524207355665e-05,
+      "loss": 0.2618,
+      "step": 554
+    },
+    {
+      "epoch": 0.3552,
+      "grad_norm": 0.32778556243620743,
+      "learning_rate": 2.3025593494199885e-05,
+      "loss": 0.2427,
+      "step": 555
+    },
+    {
+      "epoch": 0.35584,
+      "grad_norm": 0.2913558771027596,
+      "learning_rate": 2.299873023691938e-05,
+      "loss": 0.2353,
+      "step": 556
+    },
+    {
+      "epoch": 0.35648,
+      "grad_norm": 0.3031256464779152,
+      "learning_rate": 2.29718310842774e-05,
+      "loss": 0.2436,
+      "step": 557
+    },
+    {
+      "epoch": 0.35712,
+      "grad_norm": 0.33268039473458894,
+      "learning_rate": 2.294489615698747e-05,
+      "loss": 0.2592,
+      "step": 558
+    },
+    {
+      "epoch": 0.35776,
+      "grad_norm": 0.29522664892298234,
+      "learning_rate": 2.2917925575923635e-05,
+      "loss": 0.26,
+      "step": 559
+    },
+    {
+      "epoch": 0.3584,
+      "grad_norm": 0.3286262825442643,
+      "learning_rate": 2.289091946211995e-05,
+      "loss": 0.2605,
+      "step": 560
+    },
+    {
+      "epoch": 0.35904,
+      "grad_norm": 0.3246361703020228,
+      "learning_rate": 2.286387793676993e-05,
+      "loss": 0.261,
+      "step": 561
+    },
+    {
+      "epoch": 0.35968,
+      "grad_norm": 0.28944082962457157,
+      "learning_rate": 2.2836801121226008e-05,
+      "loss": 0.2328,
+      "step": 562
+    },
+    {
+      "epoch": 0.36032,
+      "grad_norm": 0.31103631358812533,
+      "learning_rate": 2.2809689136998966e-05,
+      "loss": 0.241,
+      "step": 563
+    },
+    {
+      "epoch": 0.36096,
+      "grad_norm": 0.3449549892583192,
+      "learning_rate": 2.278254210575744e-05,
+      "loss": 0.2445,
+      "step": 564
+    },
+    {
+      "epoch": 0.3616,
+      "grad_norm": 0.3157228500256909,
+      "learning_rate": 2.275536014932732e-05,
+      "loss": 0.2509,
+      "step": 565
+    },
+    {
+      "epoch": 0.36224,
+      "grad_norm": 0.30095023154019934,
+      "learning_rate": 2.272814338969124e-05,
+      "loss": 0.2487,
+      "step": 566
+    },
+    {
+      "epoch": 0.36288,
+      "grad_norm": 0.337477979450585,
+      "learning_rate": 2.2700891948988006e-05,
+      "loss": 0.2542,
+      "step": 567
+    },
+    {
+      "epoch": 0.36352,
+      "grad_norm": 0.2928546697145262,
+      "learning_rate": 2.2673605949512082e-05,
+      "loss": 0.2554,
+      "step": 568
+    },
+    {
+      "epoch": 0.36416,
+      "grad_norm": 0.32020331341086156,
+      "learning_rate": 2.264628551371299e-05,
+      "loss": 0.2709,
+      "step": 569
+    },
+    {
+      "epoch": 0.3648,
+      "grad_norm": 0.3289347732911309,
+      "learning_rate": 2.261893076419482e-05,
+      "loss": 0.2348,
+      "step": 570
+    },
+    {
+      "epoch": 0.36544,
+      "grad_norm": 0.3102608521013213,
+      "learning_rate": 2.2591541823715615e-05,
+      "loss": 0.2642,
+      "step": 571
+    },
+    {
+      "epoch": 0.36608,
+      "grad_norm": 0.34028430297253665,
+      "learning_rate": 2.256411881518689e-05,
+      "loss": 0.2463,
+      "step": 572
+    },
+    {
+      "epoch": 0.36672,
+      "grad_norm": 0.31129961600145073,
+      "learning_rate": 2.253666186167301e-05,
+      "loss": 0.2453,
+      "step": 573
+    },
+    {
+      "epoch": 0.36736,
+      "grad_norm": 0.2658006077381884,
+      "learning_rate": 2.2509171086390715e-05,
+      "loss": 0.2389,
+      "step": 574
+    },
+    {
+      "epoch": 0.368,
+      "grad_norm": 0.3792894727601205,
+      "learning_rate": 2.248164661270849e-05,
+      "loss": 0.2482,
+      "step": 575
+    },
+    {
+      "epoch": 0.36864,
+      "grad_norm": 0.27490046335554513,
+      "learning_rate": 2.245408856414605e-05,
+      "loss": 0.2694,
+      "step": 576
+    },
+    {
+      "epoch": 0.36928,
+      "grad_norm": 0.37579350044221205,
+      "learning_rate": 2.2426497064373797e-05,
+      "loss": 0.2243,
+      "step": 577
+    },
+    {
+      "epoch": 0.36992,
+      "grad_norm": 0.3261363751456689,
+      "learning_rate": 2.239887223721223e-05,
+      "loss": 0.2497,
+      "step": 578
+    },
+    {
+      "epoch": 0.37056,
+      "grad_norm": 0.34380133531131896,
+      "learning_rate": 2.2371214206631433e-05,
+      "loss": 0.2366,
+      "step": 579
+    },
+    {
+      "epoch": 0.3712,
+      "grad_norm": 0.3663930396216925,
+      "learning_rate": 2.234352309675048e-05,
+      "loss": 0.2659,
+      "step": 580
+    },
+    {
+      "epoch": 0.37184,
+      "grad_norm": 0.3144693208874203,
+      "learning_rate": 2.2315799031836887e-05,
+      "loss": 0.2289,
+      "step": 581
+    },
+    {
+      "epoch": 0.37248,
+      "grad_norm": 0.3393482246858389,
+      "learning_rate": 2.2288042136306076e-05,
+      "loss": 0.2695,
+      "step": 582
+    },
+    {
+      "epoch": 0.37312,
+      "grad_norm": 0.3749005999238712,
+      "learning_rate": 2.2260252534720783e-05,
+      "loss": 0.2457,
+      "step": 583
+    },
+    {
+      "epoch": 0.37376,
+      "grad_norm": 0.3150815215156016,
+      "learning_rate": 2.2232430351790533e-05,
+      "loss": 0.2593,
+      "step": 584
+    },
+    {
+      "epoch": 0.3744,
+      "grad_norm": 0.31855318577485014,
+      "learning_rate": 2.220457571237105e-05,
+      "loss": 0.254,
+      "step": 585
+    },
+    {
+      "epoch": 0.37504,
+      "grad_norm": 0.3492735691294561,
+      "learning_rate": 2.2176688741463732e-05,
+      "loss": 0.2602,
+      "step": 586
+    },
+    {
+      "epoch": 0.37568,
+      "grad_norm": 0.29164067568396596,
+      "learning_rate": 2.2148769564215045e-05,
+      "loss": 0.2761,
+      "step": 587
+    },
+    {
+      "epoch": 0.37632,
+      "grad_norm": 0.30406557246274263,
+      "learning_rate": 2.2120818305915997e-05,
+      "loss": 0.2447,
+      "step": 588
+    },
+    {
+      "epoch": 0.37696,
+      "grad_norm": 0.35536363697863144,
+      "learning_rate": 2.209283509200156e-05,
+      "loss": 0.2394,
+      "step": 589
+    },
+    {
+      "epoch": 0.3776,
+      "grad_norm": 0.2847948409258987,
+      "learning_rate": 2.2064820048050113e-05,
+      "loss": 0.2421,
+      "step": 590
+    },
+    {
+      "epoch": 0.37824,
+      "grad_norm": 0.3020189567244142,
+      "learning_rate": 2.203677329978288e-05,
+      "loss": 0.242,
+      "step": 591
+    },
+    {
+      "epoch": 0.37888,
+      "grad_norm": 0.31145773031624513,
+      "learning_rate": 2.200869497306336e-05,
+      "loss": 0.247,
+      "step": 592
+    },
+    {
+      "epoch": 0.37952,
+      "grad_norm": 0.3092351679740992,
+      "learning_rate": 2.198058519389676e-05,
+      "loss": 0.2481,
+      "step": 593
+    },
+    {
+      "epoch": 0.38016,
+      "grad_norm": 0.28783865305151457,
+      "learning_rate": 2.1952444088429444e-05,
+      "loss": 0.237,
+      "step": 594
+    },
+    {
+      "epoch": 0.3808,
+      "grad_norm": 0.3053618118372085,
+      "learning_rate": 2.1924271782948342e-05,
+      "loss": 0.2548,
+      "step": 595
+    },
+    {
+      "epoch": 0.38144,
+      "grad_norm": 0.3441276743502453,
+      "learning_rate": 2.1896068403880408e-05,
+      "loss": 0.2522,
+      "step": 596
+    },
+    {
+      "epoch": 0.38208,
+      "grad_norm": 0.28759508000199346,
+      "learning_rate": 2.1867834077792045e-05,
+      "loss": 0.2472,
+      "step": 597
+    },
+    {
+      "epoch": 0.38272,
+      "grad_norm": 0.32884139441178306,
+      "learning_rate": 2.183956893138852e-05,
+      "loss": 0.2377,
+      "step": 598
+    },
+    {
+      "epoch": 0.38336,
+      "grad_norm": 0.3184809395657912,
+      "learning_rate": 2.181127309151344e-05,
+      "loss": 0.2395,
+      "step": 599
+    },
+    {
+      "epoch": 0.384,
+      "grad_norm": 0.3579225369104567,
+      "learning_rate": 2.1782946685148126e-05,
+      "loss": 0.2362,
+      "step": 600
+    },
+    {
+      "epoch": 0.38464,
+      "grad_norm": 0.3668180890999166,
+      "learning_rate": 2.175458983941108e-05,
+      "loss": 0.275,
+      "step": 601
+    },
+    {
+      "epoch": 0.38528,
+      "grad_norm": 0.284016254548238,
+      "learning_rate": 2.1726202681557398e-05,
+      "loss": 0.2353,
+      "step": 602
+    },
+    {
+      "epoch": 0.38592,
+      "grad_norm": 0.3572757014239311,
+      "learning_rate": 2.1697785338978215e-05,
+      "loss": 0.2354,
+      "step": 603
+    },
+    {
+      "epoch": 0.38656,
+      "grad_norm": 0.2977040020384556,
+      "learning_rate": 2.166933793920012e-05,
+      "loss": 0.232,
+      "step": 604
+    },
+    {
+      "epoch": 0.3872,
+      "grad_norm": 0.30172127232859347,
+      "learning_rate": 2.1640860609884588e-05,
+      "loss": 0.2341,
+      "step": 605
+    },
+    {
+      "epoch": 0.38784,
+      "grad_norm": 0.34442418696488436,
+      "learning_rate": 2.161235347882741e-05,
+      "loss": 0.2507,
+      "step": 606
+    },
+    {
+      "epoch": 0.38848,
+      "grad_norm": 0.2948413370383127,
+      "learning_rate": 2.1583816673958104e-05,
+      "loss": 0.2499,
+      "step": 607
+    },
+    {
+      "epoch": 0.38912,
+      "grad_norm": 0.30004465284599646,
+      "learning_rate": 2.1555250323339368e-05,
+      "loss": 0.2437,
+      "step": 608
+    },
+    {
+      "epoch": 0.38976,
+      "grad_norm": 0.3099575774566211,
+      "learning_rate": 2.1526654555166484e-05,
+      "loss": 0.2511,
+      "step": 609
+    },
+    {
+      "epoch": 0.3904,
+      "grad_norm": 0.29571761737867036,
+      "learning_rate": 2.1498029497766744e-05,
+      "loss": 0.2552,
+      "step": 610
+    },
+    {
+      "epoch": 0.39104,
+      "grad_norm": 0.3029984741223696,
+      "learning_rate": 2.1469375279598898e-05,
+      "loss": 0.2479,
+      "step": 611
+    },
+    {
+      "epoch": 0.39168,
+      "grad_norm": 0.3182893728592652,
+      "learning_rate": 2.144069202925253e-05,
+      "loss": 0.2456,
+      "step": 612
+    },
+    {
+      "epoch": 0.39232,
+      "grad_norm": 0.3256336345330972,
+      "learning_rate": 2.1411979875447535e-05,
+      "loss": 0.2326,
+      "step": 613
+    },
+    {
+      "epoch": 0.39296,
+      "grad_norm": 0.3213948031280962,
+      "learning_rate": 2.1383238947033504e-05,
+      "loss": 0.255,
+      "step": 614
+    },
+    {
+      "epoch": 0.3936,
+      "grad_norm": 0.29013566998871226,
+      "learning_rate": 2.1354469372989158e-05,
+      "loss": 0.2663,
+      "step": 615
+    },
+    {
+      "epoch": 0.39424,
+      "grad_norm": 0.34420320849332475,
+      "learning_rate": 2.132567128242178e-05,
+      "loss": 0.2553,
+      "step": 616
+    },
+    {
+      "epoch": 0.39488,
+      "grad_norm": 0.3132567476522736,
+      "learning_rate": 2.12968448045666e-05,
+      "loss": 0.2573,
+      "step": 617
+    },
+    {
+      "epoch": 0.39552,
+      "grad_norm": 0.3090743818474423,
+      "learning_rate": 2.1267990068786278e-05,
+      "loss": 0.2382,
+      "step": 618
+    },
+    {
+      "epoch": 0.39616,
+      "grad_norm": 0.2985193819257022,
+      "learning_rate": 2.1239107204570245e-05,
+      "loss": 0.278,
+      "step": 619
+    },
+    {
+      "epoch": 0.3968,
+      "grad_norm": 0.3202824851156904,
+      "learning_rate": 2.1210196341534182e-05,
+      "loss": 0.2349,
+      "step": 620
+    },
+    {
+      "epoch": 0.39744,
+      "grad_norm": 0.28647763445799634,
+      "learning_rate": 2.1181257609419422e-05,
+      "loss": 0.2499,
+      "step": 621
+    },
+    {
+      "epoch": 0.39808,
+      "grad_norm": 0.3013665694615083,
+      "learning_rate": 2.115229113809236e-05,
+      "loss": 0.247,
+      "step": 622
+    },
+    {
+      "epoch": 0.39872,
+      "grad_norm": 0.2914815601166263,
+      "learning_rate": 2.1123297057543864e-05,
+      "loss": 0.2482,
+      "step": 623
+    },
+    {
+      "epoch": 0.39936,
+      "grad_norm": 0.27858576448210315,
+      "learning_rate": 2.109427549788872e-05,
+      "loss": 0.2403,
+      "step": 624
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.2811765312335858,
+      "learning_rate": 2.106522658936502e-05,
+      "loss": 0.2276,
+      "step": 625
+    },
+    {
+      "epoch": 0.40064,
+      "grad_norm": 0.2975257782496434,
+      "learning_rate": 2.1036150462333585e-05,
+      "loss": 0.2868,
+      "step": 626
+    },
+    {
+      "epoch": 0.40128,
+      "grad_norm": 0.3101473385119221,
+      "learning_rate": 2.100704724727739e-05,
+      "loss": 0.2394,
+      "step": 627
+    },
+    {
+      "epoch": 0.40192,
+      "grad_norm": 0.2753240305984353,
+      "learning_rate": 2.0977917074800966e-05,
+      "loss": 0.253,
+      "step": 628
+    },
+    {
+      "epoch": 0.40256,
+      "grad_norm": 0.33128143290767126,
+      "learning_rate": 2.0948760075629816e-05,
+      "loss": 0.2431,
+      "step": 629
+    },
+    {
+      "epoch": 0.4032,
+      "grad_norm": 0.29785787572122974,
+      "learning_rate": 2.0919576380609847e-05,
+      "loss": 0.2403,
+      "step": 630
+    },
+    {
+      "epoch": 0.40384,
+      "grad_norm": 0.31021479700099863,
+      "learning_rate": 2.0890366120706748e-05,
+      "loss": 0.2412,
+      "step": 631
+    },
+    {
+      "epoch": 0.40448,
+      "grad_norm": 0.32003709399269986,
+      "learning_rate": 2.086112942700543e-05,
+      "loss": 0.2428,
+      "step": 632
+    },
+    {
+      "epoch": 0.40512,
+      "grad_norm": 0.28802313751198955,
+      "learning_rate": 2.083186643070943e-05,
+      "loss": 0.2509,
+      "step": 633
+    },
+    {
+      "epoch": 0.40576,
+      "grad_norm": 0.30877621913191516,
+      "learning_rate": 2.0802577263140323e-05,
+      "loss": 0.2503,
+      "step": 634
+    },
+    {
+      "epoch": 0.4064,
+      "grad_norm": 0.2798340868739933,
+      "learning_rate": 2.0773262055737122e-05,
+      "loss": 0.2586,
+      "step": 635
+    },
+    {
+      "epoch": 0.40704,
+      "grad_norm": 0.29456576669510637,
+      "learning_rate": 2.074392094005571e-05,
+      "loss": 0.2425,
+      "step": 636
+    },
+    {
+      "epoch": 0.40768,
+      "grad_norm": 0.2954614927035877,
+      "learning_rate": 2.0714554047768224e-05,
+      "loss": 0.2516,
+      "step": 637
+    },
+    {
+      "epoch": 0.40832,
+      "grad_norm": 0.29091464782031107,
+      "learning_rate": 2.068516151066249e-05,
+      "loss": 0.2463,
+      "step": 638
+    },
+    {
+      "epoch": 0.40896,
+      "grad_norm": 0.2971818958989882,
+      "learning_rate": 2.0655743460641403e-05,
+      "loss": 0.2279,
+      "step": 639
+    },
+    {
+      "epoch": 0.4096,
+      "grad_norm": 0.26433254320966026,
+      "learning_rate": 2.062630002972237e-05,
+      "loss": 0.2363,
+      "step": 640
+    },
+    {
+      "epoch": 0.41024,
+      "grad_norm": 0.3354545785932358,
+      "learning_rate": 2.0596831350036674e-05,
+      "loss": 0.2418,
+      "step": 641
+    },
+    {
+      "epoch": 0.41088,
+      "grad_norm": 0.3109013845802427,
+      "learning_rate": 2.0567337553828935e-05,
+      "loss": 0.2524,
+      "step": 642
+    },
+    {
+      "epoch": 0.41152,
+      "grad_norm": 0.3402728181158717,
+      "learning_rate": 2.0537818773456458e-05,
+      "loss": 0.2274,
+      "step": 643
+    },
+    {
+      "epoch": 0.41216,
+      "grad_norm": 0.31342814288430576,
+      "learning_rate": 2.0508275141388684e-05,
+      "loss": 0.2761,
+      "step": 644
+    },
+    {
+      "epoch": 0.4128,
+      "grad_norm": 0.2942734303684883,
+      "learning_rate": 2.047870679020657e-05,
+      "loss": 0.2235,
+      "step": 645
+    },
+    {
+      "epoch": 0.41344,
+      "grad_norm": 0.33039525789141777,
+      "learning_rate": 2.044911385260202e-05,
+      "loss": 0.2335,
+      "step": 646
+    },
+    {
+      "epoch": 0.41408,
+      "grad_norm": 0.2767673737540941,
+      "learning_rate": 2.0419496461377253e-05,
+      "loss": 0.2438,
+      "step": 647
+    },
+    {
+      "epoch": 0.41472,
+      "grad_norm": 0.36279599663063367,
+      "learning_rate": 2.0389854749444232e-05,
+      "loss": 0.2528,
+      "step": 648
+    },
+    {
+      "epoch": 0.41536,
+      "grad_norm": 0.29068552338360093,
+      "learning_rate": 2.0360188849824076e-05,
+      "loss": 0.2687,
+      "step": 649
+    },
+    {
+      "epoch": 0.416,
+      "grad_norm": 0.300468767033639,
+      "learning_rate": 2.033049889564643e-05,
+      "loss": 0.2415,
+      "step": 650
+    },
+    {
+      "epoch": 0.41664,
+      "grad_norm": 0.31567351162652707,
+      "learning_rate": 2.03007850201489e-05,
+      "loss": 0.2412,
+      "step": 651
+    },
+    {
+      "epoch": 0.41728,
+      "grad_norm": 0.2960445310895459,
+      "learning_rate": 2.0271047356676448e-05,
+      "loss": 0.2672,
+      "step": 652
+    },
+    {
+      "epoch": 0.41792,
+      "grad_norm": 0.3137476648522256,
+      "learning_rate": 2.0241286038680756e-05,
+      "loss": 0.2505,
+      "step": 653
+    },
+    {
+      "epoch": 0.41856,
+      "grad_norm": 0.2903028208499828,
+      "learning_rate": 2.0211501199719704e-05,
+      "loss": 0.2328,
+      "step": 654
+    },
+    {
+      "epoch": 0.4192,
+      "grad_norm": 0.3042506515512182,
+      "learning_rate": 2.0181692973456686e-05,
+      "loss": 0.2537,
+      "step": 655
+    },
+    {
+      "epoch": 0.41984,
+      "grad_norm": 0.29549374179951227,
+      "learning_rate": 2.015186149366007e-05,
+      "loss": 0.2538,
+      "step": 656
+    },
+    {
+      "epoch": 0.42048,
+      "grad_norm": 0.3194068056008517,
+      "learning_rate": 2.0122006894202577e-05,
+      "loss": 0.2331,
+      "step": 657
+    },
+    {
+      "epoch": 0.42112,
+      "grad_norm": 0.2719676743833399,
+      "learning_rate": 2.0092129309060672e-05,
+      "loss": 0.2514,
+      "step": 658
+    },
+    {
+      "epoch": 0.42176,
+      "grad_norm": 0.3289205044581153,
+      "learning_rate": 2.0062228872313976e-05,
+      "loss": 0.2584,
+      "step": 659
+    },
+    {
+      "epoch": 0.4224,
+      "grad_norm": 0.2923593537514368,
+      "learning_rate": 2.0032305718144665e-05,
+      "loss": 0.2578,
+      "step": 660
+    },
+    {
+      "epoch": 0.42304,
+      "grad_norm": 0.29063317542597716,
+      "learning_rate": 2.0002359980836853e-05,
+      "loss": 0.2384,
+      "step": 661
+    },
+    {
+      "epoch": 0.42368,
+      "grad_norm": 0.28894947654548037,
+      "learning_rate": 1.9972391794776003e-05,
+      "loss": 0.2409,
+      "step": 662
+    },
+    {
+      "epoch": 0.42432,
+      "grad_norm": 0.3098536165538198,
+      "learning_rate": 1.994240129444832e-05,
+      "loss": 0.2311,
+      "step": 663
+    },
+    {
+      "epoch": 0.42496,
+      "grad_norm": 0.3060220488640442,
+      "learning_rate": 1.9912388614440146e-05,
+      "loss": 0.2317,
+      "step": 664
+    },
+    {
+      "epoch": 0.4256,
+      "grad_norm": 0.28385024763607003,
+      "learning_rate": 1.988235388943736e-05,
+      "loss": 0.2625,
+      "step": 665
+    },
+    {
+      "epoch": 0.42624,
+      "grad_norm": 0.3250332428858291,
+      "learning_rate": 1.985229725422477e-05,
+      "loss": 0.2509,
+      "step": 666
+    },
+    {
+      "epoch": 0.42688,
+      "grad_norm": 0.27609914066722663,
+      "learning_rate": 1.98222188436855e-05,
+      "loss": 0.2515,
+      "step": 667
+    },
+    {
+      "epoch": 0.42752,
+      "grad_norm": 0.3764195638951758,
+      "learning_rate": 1.979211879280041e-05,
+      "loss": 0.2384,
+      "step": 668
+    },
+    {
+      "epoch": 0.42816,
+      "grad_norm": 0.2882112064916445,
+      "learning_rate": 1.9761997236647466e-05,
+      "loss": 0.2486,
+      "step": 669
+    },
+    {
+      "epoch": 0.4288,
+      "grad_norm": 0.3031253323342361,
+      "learning_rate": 1.9731854310401133e-05,
+      "loss": 0.2465,
+      "step": 670
+    },
+    {
+      "epoch": 0.42944,
+      "grad_norm": 0.31461387253551104,
+      "learning_rate": 1.970169014933179e-05,
+      "loss": 0.2368,
+      "step": 671
+    },
+    {
+      "epoch": 0.43008,
+      "grad_norm": 0.321768024600862,
+      "learning_rate": 1.967150488880511e-05,
+      "loss": 0.2651,
+      "step": 672
+    },
+    {
+      "epoch": 0.43072,
+      "grad_norm": 0.31937360402288967,
+      "learning_rate": 1.964129866428144e-05,
+      "loss": 0.2658,
+      "step": 673
+    },
+    {
+      "epoch": 0.43136,
+      "grad_norm": 0.28909986387996445,
+      "learning_rate": 1.9611071611315225e-05,
+      "loss": 0.2382,
+      "step": 674
+    },
+    {
+      "epoch": 0.432,
+      "grad_norm": 0.29848411611660336,
+      "learning_rate": 1.9580823865554353e-05,
+      "loss": 0.2193,
+      "step": 675
+    },
+    {
+      "epoch": 0.43264,
+      "grad_norm": 0.3251801996349692,
+      "learning_rate": 1.95505555627396e-05,
+      "loss": 0.2633,
+      "step": 676
+    },
+    {
+      "epoch": 0.43328,
+      "grad_norm": 0.31621533846449545,
+      "learning_rate": 1.952026683870398e-05,
+      "loss": 0.2465,
+      "step": 677
+    },
+    {
+      "epoch": 0.43392,
+      "grad_norm": 0.2690978345819651,
+      "learning_rate": 1.9489957829372165e-05,
+      "loss": 0.2234,
+      "step": 678
+    },
+    {
+      "epoch": 0.43456,
+      "grad_norm": 0.2834458494566235,
+      "learning_rate": 1.9459628670759837e-05,
+      "loss": 0.2576,
+      "step": 679
+    },
+    {
+      "epoch": 0.4352,
+      "grad_norm": 0.29167897883221355,
+      "learning_rate": 1.9429279498973116e-05,
+      "loss": 0.2557,
+      "step": 680
+    },
+    {
+      "epoch": 0.43584,
+      "grad_norm": 0.30646443869067846,
+      "learning_rate": 1.939891045020793e-05,
+      "loss": 0.248,
+      "step": 681
+    },
+    {
+      "epoch": 0.43648,
+      "grad_norm": 0.27473410735634074,
+      "learning_rate": 1.936852166074941e-05,
+      "loss": 0.2543,
+      "step": 682
+    },
+    {
+      "epoch": 0.43712,
+      "grad_norm": 0.2639389373623012,
+      "learning_rate": 1.9338113266971276e-05,
+      "loss": 0.2197,
+      "step": 683
+    },
+    {
+      "epoch": 0.43776,
+      "grad_norm": 0.3075802635764548,
+      "learning_rate": 1.9307685405335217e-05,
+      "loss": 0.2433,
+      "step": 684
+    },
+    {
+      "epoch": 0.4384,
+      "grad_norm": 0.270502938103472,
+      "learning_rate": 1.9277238212390296e-05,
+      "loss": 0.2474,
+      "step": 685
+    },
+    {
+      "epoch": 0.43904,
+      "grad_norm": 0.2838293202736154,
+      "learning_rate": 1.9246771824772326e-05,
+      "loss": 0.2266,
+      "step": 686
+    },
+    {
+      "epoch": 0.43968,
+      "grad_norm": 0.2853706779734916,
+      "learning_rate": 1.9216286379203257e-05,
+      "loss": 0.2461,
+      "step": 687
+    },
+    {
+      "epoch": 0.44032,
+      "grad_norm": 0.29829572135331384,
+      "learning_rate": 1.9185782012490564e-05,
+      "loss": 0.22,
+      "step": 688
+    },
+    {
+      "epoch": 0.44096,
+      "grad_norm": 0.2766232059234513,
+      "learning_rate": 1.9155258861526627e-05,
+      "loss": 0.2315,
+      "step": 689
+    },
+    {
+      "epoch": 0.4416,
+      "grad_norm": 0.27212184486467605,
+      "learning_rate": 1.912471706328814e-05,
+      "loss": 0.2457,
+      "step": 690
+    },
+    {
+      "epoch": 0.44224,
+      "grad_norm": 0.2746235664697286,
+      "learning_rate": 1.9094156754835458e-05,
+      "loss": 0.2239,
+      "step": 691
+    },
+    {
+      "epoch": 0.44288,
+      "grad_norm": 0.30054140099920756,
+      "learning_rate": 1.9063578073312018e-05,
+      "loss": 0.2473,
+      "step": 692
+    },
+    {
+      "epoch": 0.44352,
+      "grad_norm": 0.30126883324495013,
+      "learning_rate": 1.9032981155943702e-05,
+      "loss": 0.2623,
+      "step": 693
+    },
+    {
+      "epoch": 0.44416,
+      "grad_norm": 0.2628697573720338,
+      "learning_rate": 1.9002366140038236e-05,
+      "loss": 0.248,
+      "step": 694
+    },
+    {
+      "epoch": 0.4448,
+      "grad_norm": 0.2606114154909525,
+      "learning_rate": 1.8971733162984552e-05,
+      "loss": 0.2494,
+      "step": 695
+    },
+    {
+      "epoch": 0.44544,
+      "grad_norm": 0.2564796649553452,
+      "learning_rate": 1.8941082362252198e-05,
+      "loss": 0.2332,
+      "step": 696
+    },
+    {
+      "epoch": 0.44608,
+      "grad_norm": 0.30845561489676177,
+      "learning_rate": 1.891041387539069e-05,
+      "loss": 0.2462,
+      "step": 697
+    },
+    {
+      "epoch": 0.44672,
+      "grad_norm": 0.27617590444390094,
+      "learning_rate": 1.887972784002894e-05,
+      "loss": 0.2197,
+      "step": 698
+    },
+    {
+      "epoch": 0.44736,
+      "grad_norm": 0.30470085842871797,
+      "learning_rate": 1.884902439387459e-05,
+      "loss": 0.2575,
+      "step": 699
+    },
+    {
+      "epoch": 0.448,
+      "grad_norm": 0.250257667186702,
+      "learning_rate": 1.8818303674713422e-05,
+      "loss": 0.2449,
+      "step": 700
+    },
+    {
+      "epoch": 0.44864,
+      "grad_norm": 0.2848134111806949,
+      "learning_rate": 1.8787565820408734e-05,
+      "loss": 0.2446,
+      "step": 701
+    },
+    {
+      "epoch": 0.44928,
+      "grad_norm": 0.2960785043913214,
+      "learning_rate": 1.875681096890072e-05,
+      "loss": 0.2354,
+      "step": 702
+    },
+    {
+      "epoch": 0.44992,
+      "grad_norm": 0.24645274331403022,
+      "learning_rate": 1.8726039258205844e-05,
+      "loss": 0.2587,
+      "step": 703
+    },
+    {
+      "epoch": 0.45056,
+      "grad_norm": 0.2887458899807905,
+      "learning_rate": 1.8695250826416237e-05,
+      "loss": 0.2198,
+      "step": 704
+    },
+    {
+      "epoch": 0.4512,
+      "grad_norm": 0.2781754228552816,
+      "learning_rate": 1.8664445811699063e-05,
+      "loss": 0.2476,
+      "step": 705
+    },
+    {
+      "epoch": 0.45184,
+      "grad_norm": 0.27395246203285356,
+      "learning_rate": 1.8633624352295914e-05,
+      "loss": 0.2252,
+      "step": 706
+    },
+    {
+      "epoch": 0.45248,
+      "grad_norm": 0.2725538240915513,
+      "learning_rate": 1.8602786586522163e-05,
+      "loss": 0.2324,
+      "step": 707
+    },
+    {
+      "epoch": 0.45312,
+      "grad_norm": 0.28094440947721694,
+      "learning_rate": 1.8571932652766376e-05,
+      "loss": 0.2325,
+      "step": 708
+    },
+    {
+      "epoch": 0.45376,
+      "grad_norm": 0.2923745301438137,
+      "learning_rate": 1.854106268948966e-05,
+      "loss": 0.2254,
+      "step": 709
+    },
+    {
+      "epoch": 0.4544,
+      "grad_norm": 0.2801508861752559,
+      "learning_rate": 1.8510176835225063e-05,
+      "loss": 0.2433,
+      "step": 710
+    },
+    {
+      "epoch": 0.45504,
+      "grad_norm": 0.27180662682502943,
+      "learning_rate": 1.847927522857696e-05,
+      "loss": 0.2311,
+      "step": 711
+    },
+    {
+      "epoch": 0.45568,
+      "grad_norm": 0.28702262741491485,
+      "learning_rate": 1.8448358008220393e-05,
+      "loss": 0.236,
+      "step": 712
+    },
+    {
+      "epoch": 0.45632,
+      "grad_norm": 0.2773135841625751,
+      "learning_rate": 1.8417425312900485e-05,
+      "loss": 0.2484,
+      "step": 713
+    },
+    {
+      "epoch": 0.45696,
+      "grad_norm": 0.2863331904071235,
+      "learning_rate": 1.8386477281431814e-05,
+      "loss": 0.2476,
+      "step": 714
+    },
+    {
+      "epoch": 0.4576,
+      "grad_norm": 0.2956391633466897,
+      "learning_rate": 1.8355514052697757e-05,
+      "loss": 0.2452,
+      "step": 715
+    },
+    {
+      "epoch": 0.45824,
+      "grad_norm": 0.2887659913133557,
+      "learning_rate": 1.832453576564991e-05,
+      "loss": 0.2448,
+      "step": 716
+    },
+    {
+      "epoch": 0.45888,
+      "grad_norm": 0.29739285828614076,
+      "learning_rate": 1.8293542559307438e-05,
+      "loss": 0.2334,
+      "step": 717
+    },
+    {
+      "epoch": 0.45952,
+      "grad_norm": 0.3094313748335245,
+      "learning_rate": 1.8262534572756462e-05,
+      "loss": 0.2586,
+      "step": 718
+    },
+    {
+      "epoch": 0.46016,
+      "grad_norm": 0.2793044911825085,
+      "learning_rate": 1.823151194514943e-05,
+      "loss": 0.2503,
+      "step": 719
+    },
+    {
+      "epoch": 0.4608,
+      "grad_norm": 0.2781664739633705,
+      "learning_rate": 1.820047481570449e-05,
+      "loss": 0.2539,
+      "step": 720
+    },
+    {
+      "epoch": 0.46144,
+      "grad_norm": 0.3041251930429675,
+      "learning_rate": 1.816942332370487e-05,
+      "loss": 0.249,
+      "step": 721
+    },
+    {
+      "epoch": 0.46208,
+      "grad_norm": 0.2997931472177727,
+      "learning_rate": 1.8138357608498255e-05,
+      "loss": 0.2467,
+      "step": 722
+    },
+    {
+      "epoch": 0.46272,
+      "grad_norm": 0.2795161631070679,
+      "learning_rate": 1.8107277809496163e-05,
+      "loss": 0.2387,
+      "step": 723
+    },
+    {
+      "epoch": 0.46336,
+      "grad_norm": 0.2796349459167825,
+      "learning_rate": 1.80761840661733e-05,
+      "loss": 0.2332,
+      "step": 724
+    },
+    {
+      "epoch": 0.464,
+      "grad_norm": 0.3141385412008042,
+      "learning_rate": 1.8045076518066966e-05,
+      "loss": 0.223,
+      "step": 725
+    },
+    {
+      "epoch": 0.46464,
+      "grad_norm": 0.3332215661467418,
+      "learning_rate": 1.80139553047764e-05,
+      "loss": 0.241,
+      "step": 726
+    },
+    {
+      "epoch": 0.46528,
+      "grad_norm": 0.3014159403594374,
+      "learning_rate": 1.798282056596217e-05,
+      "loss": 0.2509,
+      "step": 727
+    },
+    {
+      "epoch": 0.46592,
+      "grad_norm": 0.3008006533466029,
+      "learning_rate": 1.795167244134553e-05,
+      "loss": 0.223,
+      "step": 728
+    },
+    {
+      "epoch": 0.46656,
+      "grad_norm": 0.42627687451650975,
+      "learning_rate": 1.7920511070707833e-05,
+      "loss": 0.2536,
+      "step": 729
+    },
+    {
+      "epoch": 0.4672,
+      "grad_norm": 0.32225997058646555,
+      "learning_rate": 1.7889336593889846e-05,
+      "loss": 0.2792,
+      "step": 730
+    },
+    {
+      "epoch": 0.46784,
+      "grad_norm": 0.3214254990679866,
+      "learning_rate": 1.7858149150791162e-05,
+      "loss": 0.2621,
+      "step": 731
+    },
+    {
+      "epoch": 0.46848,
+      "grad_norm": 0.29296285930759747,
+      "learning_rate": 1.7826948881369578e-05,
+      "loss": 0.2557,
+      "step": 732
+    },
+    {
+      "epoch": 0.46912,
+      "grad_norm": 0.32495773394451755,
+      "learning_rate": 1.7795735925640416e-05,
+      "loss": 0.2695,
+      "step": 733
+    },
+    {
+      "epoch": 0.46976,
+      "grad_norm": 0.31407279664152,
+      "learning_rate": 1.7764510423675963e-05,
+      "loss": 0.248,
+      "step": 734
+    },
+    {
+      "epoch": 0.4704,
+      "grad_norm": 0.3109001212194977,
+      "learning_rate": 1.7733272515604783e-05,
+      "loss": 0.2483,
+      "step": 735
+    },
+    {
+      "epoch": 0.47104,
+      "grad_norm": 0.3131228629395079,
+      "learning_rate": 1.7702022341611137e-05,
+      "loss": 0.2597,
+      "step": 736
+    },
+    {
+      "epoch": 0.47168,
+      "grad_norm": 0.2960954497025586,
+      "learning_rate": 1.7670760041934322e-05,
+      "loss": 0.2562,
+      "step": 737
+    },
+    {
+      "epoch": 0.47232,
+      "grad_norm": 0.2781329801515648,
+      "learning_rate": 1.7639485756868043e-05,
+      "loss": 0.2169,
+      "step": 738
+    },
+    {
+      "epoch": 0.47296,
+      "grad_norm": 0.28424930879878024,
+      "learning_rate": 1.7608199626759796e-05,
+      "loss": 0.2614,
+      "step": 739
+    },
+    {
+      "epoch": 0.4736,
+      "grad_norm": 0.29420062292031857,
+      "learning_rate": 1.7576901792010236e-05,
+      "loss": 0.241,
+      "step": 740
+    },
+    {
+      "epoch": 0.47424,
+      "grad_norm": 0.30298761962774723,
+      "learning_rate": 1.7545592393072542e-05,
+      "loss": 0.2423,
+      "step": 741
+    },
+    {
+      "epoch": 0.47488,
+      "grad_norm": 0.3003875946079927,
+      "learning_rate": 1.751427157045179e-05,
+      "loss": 0.2364,
+      "step": 742
+    },
+    {
+      "epoch": 0.47552,
+      "grad_norm": 0.28771325149749016,
+      "learning_rate": 1.748293946470432e-05,
+      "loss": 0.2372,
+      "step": 743
+    },
+    {
+      "epoch": 0.47616,
+      "grad_norm": 0.31694239516805306,
+      "learning_rate": 1.745159621643711e-05,
+      "loss": 0.2414,
+      "step": 744
+    },
+    {
+      "epoch": 0.4768,
+      "grad_norm": 0.3313169785473572,
+      "learning_rate": 1.742024196630713e-05,
+      "loss": 0.2605,
+      "step": 745
+    },
+    {
+      "epoch": 0.47744,
+      "grad_norm": 0.32517473503518,
+      "learning_rate": 1.7388876855020735e-05,
+      "loss": 0.2618,
+      "step": 746
+    },
+    {
+      "epoch": 0.47808,
+      "grad_norm": 0.2854775316164675,
+      "learning_rate": 1.7357501023333017e-05,
+      "loss": 0.2487,
+      "step": 747
+    },
+    {
+      "epoch": 0.47872,
+      "grad_norm": 0.3480918768403915,
+      "learning_rate": 1.732611461204718e-05,
+      "loss": 0.2551,
+      "step": 748
+    },
+    {
+      "epoch": 0.47936,
+      "grad_norm": 0.27798931707993946,
+      "learning_rate": 1.7294717762013898e-05,
+      "loss": 0.2208,
+      "step": 749
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.30941965572051944,
+      "learning_rate": 1.7263310614130696e-05,
+      "loss": 0.2574,
+      "step": 750
+    },
+    {
+      "epoch": 0.48064,
+      "grad_norm": 0.274215437510584,
+      "learning_rate": 1.723189330934131e-05,
+      "loss": 0.2537,
+      "step": 751
+    },
+    {
+      "epoch": 0.48128,
+      "grad_norm": 0.27338644568214365,
+      "learning_rate": 1.7200465988635057e-05,
+      "loss": 0.2276,
+      "step": 752
+    },
+    {
+      "epoch": 0.48192,
+      "grad_norm": 0.30295123698091525,
+      "learning_rate": 1.7169028793046202e-05,
+      "loss": 0.241,
+      "step": 753
+    },
+    {
+      "epoch": 0.48256,
+      "grad_norm": 0.3023883382178243,
+      "learning_rate": 1.713758186365332e-05,
+      "loss": 0.2415,
+      "step": 754
+    },
+    {
+      "epoch": 0.4832,
+      "grad_norm": 0.29982309786504135,
+      "learning_rate": 1.710612534157868e-05,
+      "loss": 0.2502,
+      "step": 755
+    },
+    {
+      "epoch": 0.48384,
+      "grad_norm": 0.29405059504029574,
+      "learning_rate": 1.7074659367987588e-05,
+      "loss": 0.2504,
+      "step": 756
+    },
+    {
+      "epoch": 0.48448,
+      "grad_norm": 0.29717042313602116,
+      "learning_rate": 1.704318408408777e-05,
+      "loss": 0.2221,
+      "step": 757
+    },
+    {
+      "epoch": 0.48512,
+      "grad_norm": 0.28950952560619164,
+      "learning_rate": 1.7011699631128727e-05,
+      "loss": 0.2601,
+      "step": 758
+    },
+    {
+      "epoch": 0.48576,
+      "grad_norm": 0.27616868083933593,
+      "learning_rate": 1.6980206150401112e-05,
+      "loss": 0.2435,
+      "step": 759
+    },
+    {
+      "epoch": 0.4864,
+      "grad_norm": 0.3192248657906988,
+      "learning_rate": 1.6948703783236093e-05,
+      "loss": 0.2577,
+      "step": 760
+    },
+    {
+      "epoch": 0.48704,
+      "grad_norm": 0.261596366161892,
+      "learning_rate": 1.6917192671004725e-05,
+      "loss": 0.2385,
+      "step": 761
+    },
+    {
+      "epoch": 0.48768,
+      "grad_norm": 0.2908784165700428,
+      "learning_rate": 1.688567295511729e-05,
+      "loss": 0.2563,
+      "step": 762
+    },
+    {
+      "epoch": 0.48832,
+      "grad_norm": 0.30518493124643414,
+      "learning_rate": 1.6854144777022685e-05,
+      "loss": 0.2518,
+      "step": 763
+    },
+    {
+      "epoch": 0.48896,
+      "grad_norm": 0.3945838451684876,
+      "learning_rate": 1.682260827820779e-05,
+      "loss": 0.2453,
+      "step": 764
+    },
+    {
+      "epoch": 0.4896,
+      "grad_norm": 0.2971094396719173,
+      "learning_rate": 1.6791063600196818e-05,
+      "loss": 0.2395,
+      "step": 765
+    },
+    {
+      "epoch": 0.49024,
+      "grad_norm": 0.29418187047199595,
+      "learning_rate": 1.675951088455069e-05,
+      "loss": 0.2339,
+      "step": 766
+    },
+    {
+      "epoch": 0.49088,
+      "grad_norm": 0.31939862410372316,
+      "learning_rate": 1.6727950272866405e-05,
+      "loss": 0.2579,
+      "step": 767
+    },
+    {
+      "epoch": 0.49152,
+      "grad_norm": 0.2910294262479968,
+      "learning_rate": 1.6696381906776383e-05,
+      "loss": 0.2638,
+      "step": 768
+    },
+    {
+      "epoch": 0.49216,
+      "grad_norm": 0.2848802319438362,
+      "learning_rate": 1.666480592794784e-05,
+      "loss": 0.2495,
+      "step": 769
+    },
+    {
+      "epoch": 0.4928,
+      "grad_norm": 0.31358274735097835,
+      "learning_rate": 1.663322247808218e-05,
+      "loss": 0.2304,
+      "step": 770
+    },
+    {
+      "epoch": 0.49344,
+      "grad_norm": 0.27825339734876414,
+      "learning_rate": 1.6601631698914303e-05,
+      "loss": 0.2157,
+      "step": 771
+    },
+    {
+      "epoch": 0.49408,
+      "grad_norm": 0.29204963150572166,
+      "learning_rate": 1.657003373221202e-05,
+      "loss": 0.2289,
+      "step": 772
+    },
+    {
+      "epoch": 0.49472,
+      "grad_norm": 0.31293590502176905,
+      "learning_rate": 1.6538428719775402e-05,
+      "loss": 0.2345,
+      "step": 773
+    },
+    {
+      "epoch": 0.49536,
+      "grad_norm": 0.3006040607999008,
+      "learning_rate": 1.6506816803436117e-05,
+      "loss": 0.2571,
+      "step": 774
+    },
+    {
+      "epoch": 0.496,
+      "grad_norm": 0.26799679474754523,
+      "learning_rate": 1.6475198125056824e-05,
+      "loss": 0.252,
+      "step": 775
+    },
+    {
+      "epoch": 0.49664,
+      "grad_norm": 0.2821301194781917,
+      "learning_rate": 1.6443572826530545e-05,
+      "loss": 0.2331,
+      "step": 776
+    },
+    {
+      "epoch": 0.49728,
+      "grad_norm": 0.2967133869603498,
+      "learning_rate": 1.6411941049779984e-05,
+      "loss": 0.2526,
+      "step": 777
+    },
+    {
+      "epoch": 0.49792,
+      "grad_norm": 0.29653102428208794,
+      "learning_rate": 1.6380302936756933e-05,
+      "loss": 0.2407,
+      "step": 778
+    },
+    {
+      "epoch": 0.49856,
+      "grad_norm": 0.2712138400371119,
+      "learning_rate": 1.6348658629441618e-05,
+      "loss": 0.2687,
+      "step": 779
+    },
+    {
+      "epoch": 0.4992,
+      "grad_norm": 0.29223153618349473,
+      "learning_rate": 1.6317008269842056e-05,
+      "loss": 0.2603,
+      "step": 780
+    },
+    {
+      "epoch": 0.49984,
+      "grad_norm": 0.2895681083759542,
+      "learning_rate": 1.628535199999343e-05,
+      "loss": 0.2472,
+      "step": 781
+    },
+    {
+      "epoch": 0.50048,
+      "grad_norm": 0.2753546178449307,
+      "learning_rate": 1.6253689961957442e-05,
+      "loss": 0.2501,
+      "step": 782
+    },
+    {
+      "epoch": 0.50112,
+      "grad_norm": 0.2874283254253285,
+      "learning_rate": 1.6222022297821685e-05,
+      "loss": 0.2341,
+      "step": 783
+    },
+    {
+      "epoch": 0.50176,
+      "grad_norm": 0.25670956021391794,
+      "learning_rate": 1.6190349149698993e-05,
+      "loss": 0.2389,
+      "step": 784
+    },
+    {
+      "epoch": 0.5024,
+      "grad_norm": 0.3721981372004176,
+      "learning_rate": 1.6158670659726816e-05,
+      "loss": 0.2398,
+      "step": 785
+    },
+    {
+      "epoch": 0.50304,
+      "grad_norm": 0.26265623036956204,
+      "learning_rate": 1.612698697006657e-05,
+      "loss": 0.2449,
+      "step": 786
+    },
+    {
+      "epoch": 0.50368,
+      "grad_norm": 0.32648868955386057,
+      "learning_rate": 1.6095298222903018e-05,
+      "loss": 0.2425,
+      "step": 787
+    },
+    {
+      "epoch": 0.50432,
+      "grad_norm": 0.27240791695651867,
+      "learning_rate": 1.6063604560443602e-05,
+      "loss": 0.2608,
+      "step": 788
+    },
+    {
+      "epoch": 0.50496,
+      "grad_norm": 0.2752436970583074,
+      "learning_rate": 1.6031906124917835e-05,
+      "loss": 0.2533,
+      "step": 789
+    },
+    {
+      "epoch": 0.5056,
+      "grad_norm": 0.3240780608668842,
+      "learning_rate": 1.6000203058576643e-05,
+      "loss": 0.2328,
+      "step": 790
+    },
+    {
+      "epoch": 0.50624,
+      "grad_norm": 0.3125211914154205,
+      "learning_rate": 1.5968495503691738e-05,
+      "loss": 0.2499,
+      "step": 791
+    },
+    {
+      "epoch": 0.50688,
+      "grad_norm": 0.2814512806616593,
+      "learning_rate": 1.593678360255497e-05,
+      "loss": 0.2518,
+      "step": 792
+    },
+    {
+      "epoch": 0.50752,
+      "grad_norm": 0.26953227957530657,
+      "learning_rate": 1.5905067497477697e-05,
+      "loss": 0.2549,
+      "step": 793
+    },
+    {
+      "epoch": 0.50816,
+      "grad_norm": 0.27737474201549767,
+      "learning_rate": 1.587334733079014e-05,
+      "loss": 0.2262,
+      "step": 794
+    },
+    {
+      "epoch": 0.5088,
+      "grad_norm": 0.2999832225978434,
+      "learning_rate": 1.5841623244840756e-05,
+      "loss": 0.2303,
+      "step": 795
+    },
+    {
+      "epoch": 0.50944,
+      "grad_norm": 0.29629202724937315,
+      "learning_rate": 1.5809895381995575e-05,
+      "loss": 0.2728,
+      "step": 796
+    },
+    {
+      "epoch": 0.51008,
+      "grad_norm": 0.29824141391200387,
+      "learning_rate": 1.5778163884637584e-05,
+      "loss": 0.2376,
+      "step": 797
+    },
+    {
+      "epoch": 0.51072,
+      "grad_norm": 0.3117999804048647,
+      "learning_rate": 1.5746428895166088e-05,
+      "loss": 0.2509,
+      "step": 798
+    },
+    {
+      "epoch": 0.51136,
+      "grad_norm": 0.2868360103448722,
+      "learning_rate": 1.5714690555996048e-05,
+      "loss": 0.23,
+      "step": 799
+    },
+    {
+      "epoch": 0.512,
+      "grad_norm": 0.3102954368583538,
+      "learning_rate": 1.568294900955747e-05,
+      "loss": 0.2451,
+      "step": 800
+    },
+    {
+      "epoch": 0.51264,
+      "grad_norm": 0.32979715453781355,
+      "learning_rate": 1.565120439829474e-05,
+      "loss": 0.2576,
+      "step": 801
+    },
+    {
+      "epoch": 0.51328,
+      "grad_norm": 0.2570931740010072,
+      "learning_rate": 1.561945686466602e-05,
+      "loss": 0.2285,
+      "step": 802
+    },
+    {
+      "epoch": 0.51392,
+      "grad_norm": 0.29296845098959595,
+      "learning_rate": 1.5587706551142566e-05,
+      "loss": 0.2269,
+      "step": 803
+    },
+    {
+      "epoch": 0.51456,
+      "grad_norm": 0.34949753344114726,
+      "learning_rate": 1.5555953600208115e-05,
+      "loss": 0.2382,
+      "step": 804
+    },
+    {
+      "epoch": 0.5152,
+      "grad_norm": 0.31167228945070136,
+      "learning_rate": 1.5524198154358237e-05,
+      "loss": 0.2211,
+      "step": 805
+    },
+    {
+      "epoch": 0.51584,
+      "grad_norm": 0.26584493436752854,
+      "learning_rate": 1.5492440356099705e-05,
+      "loss": 0.2332,
+      "step": 806
+    },
+    {
+      "epoch": 0.51648,
+      "grad_norm": 0.3370974950051766,
+      "learning_rate": 1.5460680347949847e-05,
+      "loss": 0.2309,
+      "step": 807
+    },
+    {
+      "epoch": 0.51712,
+      "grad_norm": 0.3112492758911654,
+      "learning_rate": 1.5428918272435903e-05,
+      "loss": 0.2403,
+      "step": 808
+    },
+    {
+      "epoch": 0.51776,
+      "grad_norm": 0.29616924145650064,
+      "learning_rate": 1.5397154272094396e-05,
+      "loss": 0.2162,
+      "step": 809
+    },
+    {
+      "epoch": 0.5184,
+      "grad_norm": 0.32277865645191245,
+      "learning_rate": 1.5365388489470488e-05,
+      "loss": 0.2474,
+      "step": 810
+    },
+    {
+      "epoch": 0.51904,
+      "grad_norm": 0.2694955547783349,
+      "learning_rate": 1.5333621067117328e-05,
+      "loss": 0.243,
+      "step": 811
+    },
+    {
+      "epoch": 0.51968,
+      "grad_norm": 0.2805877678663554,
+      "learning_rate": 1.5301852147595442e-05,
+      "loss": 0.2545,
+      "step": 812
+    },
+    {
+      "epoch": 0.52032,
+      "grad_norm": 0.3359964195281161,
+      "learning_rate": 1.5270081873472057e-05,
+      "loss": 0.257,
+      "step": 813
+    },
+    {
+      "epoch": 0.52096,
+      "grad_norm": 0.29024170855725706,
+      "learning_rate": 1.523831038732049e-05,
+      "loss": 0.2381,
+      "step": 814
+    },
+    {
+      "epoch": 0.5216,
+      "grad_norm": 0.28870485110686106,
+      "learning_rate": 1.5206537831719492e-05,
+      "loss": 0.24,
+      "step": 815
+    },
+    {
+      "epoch": 0.52224,
+      "grad_norm": 0.3025657209936217,
+      "learning_rate": 1.5174764349252618e-05,
+      "loss": 0.224,
+      "step": 816
+    },
+    {
+      "epoch": 0.52288,
+      "grad_norm": 0.3138551823765259,
+      "learning_rate": 1.5142990082507577e-05,
+      "loss": 0.2442,
+      "step": 817
+    },
+    {
+      "epoch": 0.52352,
+      "grad_norm": 0.33587005223397937,
+      "learning_rate": 1.5111215174075599e-05,
+      "loss": 0.2339,
+      "step": 818
+    },
+    {
+      "epoch": 0.52416,
+      "grad_norm": 0.2870581549015684,
+      "learning_rate": 1.5079439766550794e-05,
+      "loss": 0.2164,
+      "step": 819
+    },
+    {
+      "epoch": 0.5248,
+      "grad_norm": 0.30441790714779565,
+      "learning_rate": 1.5047664002529515e-05,
+      "loss": 0.2253,
+      "step": 820
+    },
+    {
+      "epoch": 0.52544,
+      "grad_norm": 0.26117389090769383,
+      "learning_rate": 1.5015888024609712e-05,
+      "loss": 0.2329,
+      "step": 821
+    },
+    {
+      "epoch": 0.52608,
+      "grad_norm": 0.32251717028933996,
+      "learning_rate": 1.4984111975390292e-05,
+      "loss": 0.2393,
+      "step": 822
+    },
+    {
+      "epoch": 0.52672,
+      "grad_norm": 0.2882726784004254,
+      "learning_rate": 1.4952335997470491e-05,
+      "loss": 0.2453,
+      "step": 823
+    },
+    {
+      "epoch": 0.52736,
+      "grad_norm": 0.26989640657141045,
+      "learning_rate": 1.4920560233449207e-05,
+      "loss": 0.2346,
+      "step": 824
+    },
+    {
+      "epoch": 0.528,
+      "grad_norm": 0.2751823160442032,
+      "learning_rate": 1.4888784825924407e-05,
+      "loss": 0.2569,
+      "step": 825
+    },
+    {
+      "epoch": 0.52864,
+      "grad_norm": 0.29490569815536594,
+      "learning_rate": 1.4857009917492426e-05,
+      "loss": 0.2264,
+      "step": 826
+    },
+    {
+      "epoch": 0.52928,
+      "grad_norm": 0.27688984626798974,
+      "learning_rate": 1.4825235650747387e-05,
+      "loss": 0.2345,
+      "step": 827
+    },
+    {
+      "epoch": 0.52992,
+      "grad_norm": 0.28219958318446914,
+      "learning_rate": 1.4793462168280511e-05,
+      "loss": 0.2414,
+      "step": 828
+    },
+    {
+      "epoch": 0.53056,
+      "grad_norm": 0.2956396188477475,
+      "learning_rate": 1.476168961267951e-05,
+      "loss": 0.2285,
+      "step": 829
+    },
+    {
+      "epoch": 0.5312,
+      "grad_norm": 0.28322337068731906,
+      "learning_rate": 1.4729918126527947e-05,
+      "loss": 0.2407,
+      "step": 830
+    },
+    {
+      "epoch": 0.53184,
+      "grad_norm": 0.29936618373781043,
+      "learning_rate": 1.4698147852404558e-05,
+      "loss": 0.2702,
+      "step": 831
+    },
+    {
+      "epoch": 0.53248,
+      "grad_norm": 0.2885632951752394,
+      "learning_rate": 1.4666378932882675e-05,
+      "loss": 0.2453,
+      "step": 832
+    },
+    {
+      "epoch": 0.53312,
+      "grad_norm": 0.2824069984311681,
+      "learning_rate": 1.4634611510529518e-05,
+      "loss": 0.2872,
+      "step": 833
+    },
+    {
+      "epoch": 0.53376,
+      "grad_norm": 0.26481375162906423,
+      "learning_rate": 1.4602845727905607e-05,
+      "loss": 0.2368,
+      "step": 834
+    },
+    {
+      "epoch": 0.5344,
+      "grad_norm": 0.2982641248402793,
+      "learning_rate": 1.45710817275641e-05,
+      "loss": 0.2233,
+      "step": 835
+    },
+    {
+      "epoch": 0.53504,
+      "grad_norm": 0.2754410887201362,
+      "learning_rate": 1.4539319652050161e-05,
+      "loss": 0.2297,
+      "step": 836
+    },
+    {
+      "epoch": 0.53568,
+      "grad_norm": 0.30179074980368453,
+      "learning_rate": 1.4507559643900297e-05,
+      "loss": 0.245,
+      "step": 837
+    },
+    {
+      "epoch": 0.53632,
+      "grad_norm": 0.28782475725189016,
+      "learning_rate": 1.4475801845641769e-05,
+      "loss": 0.223,
+      "step": 838
+    },
+    {
+      "epoch": 0.53696,
+      "grad_norm": 0.27921358522056533,
+      "learning_rate": 1.4444046399791889e-05,
+      "loss": 0.2486,
+      "step": 839
+    },
+    {
+      "epoch": 0.5376,
+      "grad_norm": 0.25429071712312123,
+      "learning_rate": 1.4412293448857436e-05,
+      "loss": 0.2556,
+      "step": 840
+    },
+    {
+      "epoch": 0.53824,
+      "grad_norm": 0.33436275819647254,
+      "learning_rate": 1.4380543135333981e-05,
+      "loss": 0.2227,
+      "step": 841
+    },
+    {
+      "epoch": 0.53888,
+      "grad_norm": 0.25980186753049656,
+      "learning_rate": 1.4348795601705255e-05,
+      "loss": 0.2284,
+      "step": 842
+    },
+    {
+      "epoch": 0.53952,
+      "grad_norm": 0.2931541805752396,
+      "learning_rate": 1.4317050990442536e-05,
+      "loss": 0.2299,
+      "step": 843
+    },
+    {
+      "epoch": 0.54016,
+      "grad_norm": 0.27197020619538004,
+      "learning_rate": 1.4285309444003954e-05,
+      "loss": 0.2364,
+      "step": 844
+    },
+    {
+      "epoch": 0.5408,
+      "grad_norm": 0.32369548656685315,
+      "learning_rate": 1.4253571104833916e-05,
+      "loss": 0.2404,
+      "step": 845
+    },
+    {
+      "epoch": 0.54144,
+      "grad_norm": 0.24237700919520694,
+      "learning_rate": 1.4221836115362416e-05,
+      "loss": 0.2313,
+      "step": 846
+    },
+    {
+      "epoch": 0.54208,
+      "grad_norm": 0.24761816529676725,
+      "learning_rate": 1.4190104618004433e-05,
+      "loss": 0.2303,
+      "step": 847
+    },
+    {
+      "epoch": 0.54272,
+      "grad_norm": 0.2784023956156151,
+      "learning_rate": 1.4158376755159247e-05,
+      "loss": 0.2314,
+      "step": 848
+    },
+    {
+      "epoch": 0.54336,
+      "grad_norm": 0.31803120979153876,
+      "learning_rate": 1.4126652669209863e-05,
+      "loss": 0.2303,
+      "step": 849
+    },
+    {
+      "epoch": 0.544,
+      "grad_norm": 0.2929767473430786,
+      "learning_rate": 1.4094932502522303e-05,
+      "loss": 0.2122,
+      "step": 850
+    },
+    {
+      "epoch": 0.54464,
+      "grad_norm": 0.2953841454200276,
+      "learning_rate": 1.4063216397445033e-05,
+      "loss": 0.2329,
+      "step": 851
+    },
+    {
+      "epoch": 0.54528,
+      "grad_norm": 0.30878745217593195,
+      "learning_rate": 1.4031504496308266e-05,
+      "loss": 0.2399,
+      "step": 852
+    },
+    {
+      "epoch": 0.54592,
+      "grad_norm": 0.31428954382492774,
+      "learning_rate": 1.3999796941423358e-05,
+      "loss": 0.2499,
+      "step": 853
+    },
+    {
+      "epoch": 0.54656,
+      "grad_norm": 0.3060385149750599,
+      "learning_rate": 1.3968093875082169e-05,
+      "loss": 0.2389,
+      "step": 854
+    },
+    {
+      "epoch": 0.5472,
+      "grad_norm": 0.2781567367160064,
+      "learning_rate": 1.3936395439556398e-05,
+      "loss": 0.2346,
+      "step": 855
+    },
+    {
+      "epoch": 0.54784,
+      "grad_norm": 0.26401029057360725,
+      "learning_rate": 1.3904701777096986e-05,
+      "loss": 0.2527,
+      "step": 856
+    },
+    {
+      "epoch": 0.54848,
+      "grad_norm": 0.30906251660580814,
+      "learning_rate": 1.3873013029933431e-05,
+      "loss": 0.2272,
+      "step": 857
+    },
+    {
+      "epoch": 0.54912,
+      "grad_norm": 0.29563053269349876,
+      "learning_rate": 1.3841329340273188e-05,
+      "loss": 0.245,
+      "step": 858
+    },
+    {
+      "epoch": 0.54976,
+      "grad_norm": 0.28373950112106466,
+      "learning_rate": 1.3809650850301011e-05,
+      "loss": 0.2502,
+      "step": 859
+    },
+    {
+      "epoch": 0.5504,
+      "grad_norm": 0.27758705472251655,
+      "learning_rate": 1.3777977702178322e-05,
+      "loss": 0.2423,
+      "step": 860
+    },
+    {
+      "epoch": 0.55104,
+      "grad_norm": 0.27818143971961046,
+      "learning_rate": 1.3746310038042559e-05,
+      "loss": 0.2514,
+      "step": 861
+    },
+    {
+      "epoch": 0.55168,
+      "grad_norm": 0.2938599930393852,
+      "learning_rate": 1.3714648000006575e-05,
+      "loss": 0.2118,
+      "step": 862
+    },
+    {
+      "epoch": 0.55232,
+      "grad_norm": 0.28122086138442776,
+      "learning_rate": 1.3682991730157945e-05,
+      "loss": 0.2343,
+      "step": 863
+    },
+    {
+      "epoch": 0.55296,
+      "grad_norm": 0.2651049751908861,
+      "learning_rate": 1.3651341370558384e-05,
+      "loss": 0.2467,
+      "step": 864
+    },
+    {
+      "epoch": 0.5536,
+      "grad_norm": 0.5461359931247247,
+      "learning_rate": 1.3619697063243069e-05,
+      "loss": 0.2412,
+      "step": 865
+    },
+    {
+      "epoch": 0.55424,
+      "grad_norm": 0.2825582797542218,
+      "learning_rate": 1.3588058950220015e-05,
+      "loss": 0.228,
+      "step": 866
+    },
+    {
+      "epoch": 0.55488,
+      "grad_norm": 0.2892085558287696,
+      "learning_rate": 1.3556427173469458e-05,
+      "loss": 0.2249,
+      "step": 867
+    },
+    {
+      "epoch": 0.55552,
+      "grad_norm": 0.33273851653903624,
+      "learning_rate": 1.3524801874943172e-05,
+      "loss": 0.2323,
+      "step": 868
+    },
+    {
+      "epoch": 0.55616,
+      "grad_norm": 0.2862657843529436,
+      "learning_rate": 1.3493183196563889e-05,
+      "loss": 0.2505,
+      "step": 869
+    },
+    {
+      "epoch": 0.5568,
+      "grad_norm": 0.2808714505427905,
+      "learning_rate": 1.3461571280224602e-05,
+      "loss": 0.2354,
+      "step": 870
+    },
+    {
+      "epoch": 0.55744,
+      "grad_norm": 0.2893966418475647,
+      "learning_rate": 1.3429966267787981e-05,
+      "loss": 0.2466,
+      "step": 871
+    },
+    {
+      "epoch": 0.55808,
+      "grad_norm": 0.28389562050876976,
+      "learning_rate": 1.3398368301085699e-05,
+      "loss": 0.2407,
+      "step": 872
+    },
+    {
+      "epoch": 0.55872,
+      "grad_norm": 0.3099947400233259,
+      "learning_rate": 1.3366777521917827e-05,
+      "loss": 0.2149,
+      "step": 873
+    },
+    {
+      "epoch": 0.55936,
+      "grad_norm": 0.29887382629492687,
+      "learning_rate": 1.3335194072052159e-05,
+      "loss": 0.2202,
+      "step": 874
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.26967826249749416,
+      "learning_rate": 1.3303618093223625e-05,
+      "loss": 0.2391,
+      "step": 875
+    },
+    {
+      "epoch": 0.56064,
+      "grad_norm": 0.2465764334550036,
+      "learning_rate": 1.3272049727133599e-05,
+      "loss": 0.2203,
+      "step": 876
+    },
+    {
+      "epoch": 0.56128,
+      "grad_norm": 0.305343526383751,
+      "learning_rate": 1.3240489115449313e-05,
+      "loss": 0.2182,
+      "step": 877
+    },
+    {
+      "epoch": 0.56192,
+      "grad_norm": 0.28091661109697963,
+      "learning_rate": 1.3208936399803185e-05,
+      "loss": 0.2535,
+      "step": 878
+    },
+    {
+      "epoch": 0.56256,
+      "grad_norm": 0.31220903386699683,
+      "learning_rate": 1.3177391721792211e-05,
+      "loss": 0.2446,
+      "step": 879
+    },
+    {
+      "epoch": 0.5632,
+      "grad_norm": 0.2768392451822474,
+      "learning_rate": 1.314585522297732e-05,
+      "loss": 0.2161,
+      "step": 880
+    },
+    {
+      "epoch": 0.56384,
+      "grad_norm": 0.28782642159262545,
+      "learning_rate": 1.3114327044882714e-05,
+      "loss": 0.2282,
+      "step": 881
+    },
+    {
+      "epoch": 0.56448,
+      "grad_norm": 0.29064566337542225,
+      "learning_rate": 1.308280732899528e-05,
+      "loss": 0.2309,
+      "step": 882
+    },
+    {
+      "epoch": 0.56512,
+      "grad_norm": 0.2715775172810025,
+      "learning_rate": 1.3051296216763904e-05,
+      "loss": 0.2295,
+      "step": 883
+    },
+    {
+      "epoch": 0.56576,
+      "grad_norm": 0.29121770887438814,
+      "learning_rate": 1.3019793849598892e-05,
+      "loss": 0.2262,
+      "step": 884
+    },
+    {
+      "epoch": 0.5664,
+      "grad_norm": 0.297890794161894,
+      "learning_rate": 1.2988300368871277e-05,
+      "loss": 0.2561,
+      "step": 885
+    },
+    {
+      "epoch": 0.56704,
+      "grad_norm": 0.271128766397459,
+      "learning_rate": 1.2956815915912237e-05,
+      "loss": 0.2344,
+      "step": 886
+    },
+    {
+      "epoch": 0.56768,
+      "grad_norm": 0.2781072834762682,
+      "learning_rate": 1.2925340632012413e-05,
+      "loss": 0.2444,
+      "step": 887
+    },
+    {
+      "epoch": 0.56832,
+      "grad_norm": 0.2859546652650676,
+      "learning_rate": 1.2893874658421323e-05,
+      "loss": 0.2551,
+      "step": 888
+    },
+    {
+      "epoch": 0.56896,
+      "grad_norm": 0.2748465688510692,
+      "learning_rate": 1.2862418136346682e-05,
+      "loss": 0.2385,
+      "step": 889
+    },
+    {
+      "epoch": 0.5696,
+      "grad_norm": 0.2673167724605502,
+      "learning_rate": 1.2830971206953805e-05,
+      "loss": 0.2528,
+      "step": 890
+    },
+    {
+      "epoch": 0.57024,
+      "grad_norm": 0.26724844154208877,
+      "learning_rate": 1.2799534011364946e-05,
+      "loss": 0.2516,
+      "step": 891
+    },
+    {
+      "epoch": 0.57088,
+      "grad_norm": 0.2579234824641567,
+      "learning_rate": 1.2768106690658687e-05,
+      "loss": 0.2315,
+      "step": 892
+    },
+    {
+      "epoch": 0.57152,
+      "grad_norm": 0.2527831824097707,
+      "learning_rate": 1.2736689385869304e-05,
+      "loss": 0.2296,
+      "step": 893
+    },
+    {
+      "epoch": 0.57216,
+      "grad_norm": 0.2664685763787419,
+      "learning_rate": 1.2705282237986103e-05,
+      "loss": 0.236,
+      "step": 894
+    },
+    {
+      "epoch": 0.5728,
+      "grad_norm": 0.2887492093513382,
+      "learning_rate": 1.2673885387952824e-05,
+      "loss": 0.2407,
+      "step": 895
+    },
+    {
+      "epoch": 0.57344,
+      "grad_norm": 0.27142741542549464,
+      "learning_rate": 1.2642498976666982e-05,
+      "loss": 0.2513,
+      "step": 896
+    },
+    {
+      "epoch": 0.57408,
+      "grad_norm": 0.29898124747847593,
+      "learning_rate": 1.2611123144979269e-05,
+      "loss": 0.2434,
+      "step": 897
+    },
+    {
+      "epoch": 0.57472,
+      "grad_norm": 0.28771180626146736,
+      "learning_rate": 1.2579758033692873e-05,
+      "loss": 0.2339,
+      "step": 898
+    },
+    {
+      "epoch": 0.57536,
+      "grad_norm": 0.28998889567540337,
+      "learning_rate": 1.2548403783562896e-05,
+      "loss": 0.2118,
+      "step": 899
+    },
+    {
+      "epoch": 0.576,
+      "grad_norm": 0.2951782292199617,
+      "learning_rate": 1.2517060535295681e-05,
+      "loss": 0.2053,
+      "step": 900
+    },
+    {
+      "epoch": 0.57664,
+      "grad_norm": 0.3468538532182766,
+      "learning_rate": 1.2485728429548214e-05,
+      "loss": 0.2489,
+      "step": 901
+    },
+    {
+      "epoch": 0.57728,
+      "grad_norm": 0.2706265741912179,
+      "learning_rate": 1.245440760692746e-05,
+      "loss": 0.2301,
+      "step": 902
+    },
+    {
+      "epoch": 0.57792,
+      "grad_norm": 0.2798718644343022,
+      "learning_rate": 1.2423098207989763e-05,
+      "loss": 0.2214,
+      "step": 903
+    },
+    {
+      "epoch": 0.57856,
+      "grad_norm": 0.2976915427803377,
+      "learning_rate": 1.2391800373240205e-05,
+      "loss": 0.2403,
+      "step": 904
+    },
+    {
+      "epoch": 0.5792,
+      "grad_norm": 0.27049596723000147,
+      "learning_rate": 1.2360514243131959e-05,
+      "loss": 0.228,
+      "step": 905
+    },
+    {
+      "epoch": 0.57984,
+      "grad_norm": 0.2703696976439254,
+      "learning_rate": 1.2329239958065682e-05,
+      "loss": 0.2327,
+      "step": 906
+    },
+    {
+      "epoch": 0.58048,
+      "grad_norm": 0.2570600372926204,
+      "learning_rate": 1.2297977658388861e-05,
+      "loss": 0.2428,
+      "step": 907
+    },
+    {
+      "epoch": 0.58112,
+      "grad_norm": 0.24470082431115914,
+      "learning_rate": 1.226672748439522e-05,
+      "loss": 0.2388,
+      "step": 908
+    },
+    {
+      "epoch": 0.58176,
+      "grad_norm": 0.292744513244813,
+      "learning_rate": 1.2235489576324041e-05,
+      "loss": 0.2473,
+      "step": 909
+    },
+    {
+      "epoch": 0.5824,
+      "grad_norm": 0.2536473659907088,
+      "learning_rate": 1.2204264074359588e-05,
+      "loss": 0.2275,
+      "step": 910
+    },
+    {
+      "epoch": 0.58304,
+      "grad_norm": 0.263296518582845,
+      "learning_rate": 1.2173051118630425e-05,
+      "loss": 0.2175,
+      "step": 911
+    },
+    {
+      "epoch": 0.58368,
+      "grad_norm": 0.24857681795290157,
+      "learning_rate": 1.2141850849208837e-05,
+      "loss": 0.2322,
+      "step": 912
+    },
+    {
+      "epoch": 0.58432,
+      "grad_norm": 0.3260665654581631,
+      "learning_rate": 1.2110663406110158e-05,
+      "loss": 0.2434,
+      "step": 913
+    },
+    {
+      "epoch": 0.58496,
+      "grad_norm": 0.27474043074530696,
+      "learning_rate": 1.2079488929292173e-05,
+      "loss": 0.2377,
+      "step": 914
+    },
+    {
+      "epoch": 0.5856,
+      "grad_norm": 0.28245927837321005,
+      "learning_rate": 1.2048327558654472e-05,
+      "loss": 0.2517,
+      "step": 915
+    },
+    {
+      "epoch": 0.58624,
+      "grad_norm": 0.29411641561968144,
+      "learning_rate": 1.2017179434037834e-05,
+      "loss": 0.2581,
+      "step": 916
+    },
+    {
+      "epoch": 0.58688,
+      "grad_norm": 0.26848031902662794,
+      "learning_rate": 1.1986044695223602e-05,
+      "loss": 0.2244,
+      "step": 917
+    },
+    {
+      "epoch": 0.58752,
+      "grad_norm": 0.29005436513316146,
+      "learning_rate": 1.1954923481933034e-05,
+      "loss": 0.2437,
+      "step": 918
+    },
+    {
+      "epoch": 0.58816,
+      "grad_norm": 0.2895482712254696,
+      "learning_rate": 1.1923815933826701e-05,
+      "loss": 0.255,
+      "step": 919
+    },
+    {
+      "epoch": 0.5888,
+      "grad_norm": 0.28193500617004763,
+      "learning_rate": 1.1892722190503836e-05,
+      "loss": 0.2523,
+      "step": 920
+    },
+    {
+      "epoch": 0.58944,
+      "grad_norm": 0.2726047082654661,
+      "learning_rate": 1.1861642391501745e-05,
+      "loss": 0.2195,
+      "step": 921
+    },
+    {
+      "epoch": 0.59008,
+      "grad_norm": 0.24683767749203545,
+      "learning_rate": 1.1830576676295131e-05,
+      "loss": 0.2443,
+      "step": 922
+    },
+    {
+      "epoch": 0.59072,
+      "grad_norm": 0.27491734969798565,
+      "learning_rate": 1.1799525184295516e-05,
+      "loss": 0.238,
+      "step": 923
+    },
+    {
+      "epoch": 0.59136,
+      "grad_norm": 0.2989310502987566,
+      "learning_rate": 1.1768488054850575e-05,
+      "loss": 0.2168,
+      "step": 924
+    },
+    {
+      "epoch": 0.592,
+      "grad_norm": 0.2628836187369257,
+      "learning_rate": 1.1737465427243544e-05,
+      "loss": 0.233,
+      "step": 925
+    },
+    {
+      "epoch": 0.59264,
+      "grad_norm": 0.26752822061811826,
+      "learning_rate": 1.1706457440692565e-05,
+      "loss": 0.2263,
+      "step": 926
+    },
+    {
+      "epoch": 0.59328,
+      "grad_norm": 0.2786500979998503,
+      "learning_rate": 1.1675464234350096e-05,
+      "loss": 0.2104,
+      "step": 927
+    },
+    {
+      "epoch": 0.59392,
+      "grad_norm": 0.2616073295413085,
+      "learning_rate": 1.1644485947302247e-05,
+      "loss": 0.2094,
+      "step": 928
+    },
+    {
+      "epoch": 0.59456,
+      "grad_norm": 0.26094785406848103,
+      "learning_rate": 1.1613522718568188e-05,
+      "loss": 0.2277,
+      "step": 929
+    },
+    {
+      "epoch": 0.5952,
+      "grad_norm": 0.26956180157071696,
+      "learning_rate": 1.1582574687099515e-05,
+      "loss": 0.2421,
+      "step": 930
+    },
+    {
+      "epoch": 0.59584,
+      "grad_norm": 0.2534923867260137,
+      "learning_rate": 1.1551641991779607e-05,
+      "loss": 0.2095,
+      "step": 931
+    },
+    {
+      "epoch": 0.59648,
+      "grad_norm": 0.2613077590529168,
+      "learning_rate": 1.1520724771423045e-05,
+      "loss": 0.2442,
+      "step": 932
+    },
+    {
+      "epoch": 0.59712,
+      "grad_norm": 0.28565509842439857,
+      "learning_rate": 1.1489823164774935e-05,
+      "loss": 0.2191,
+      "step": 933
+    },
+    {
+      "epoch": 0.59776,
+      "grad_norm": 0.24145916856078634,
+      "learning_rate": 1.1458937310510346e-05,
+      "loss": 0.2404,
+      "step": 934
+    },
+    {
+      "epoch": 0.5984,
+      "grad_norm": 0.2675973415183905,
+      "learning_rate": 1.1428067347233628e-05,
+      "loss": 0.2161,
+      "step": 935
+    },
+    {
+      "epoch": 0.59904,
+      "grad_norm": 0.29913176403299363,
+      "learning_rate": 1.1397213413477841e-05,
+      "loss": 0.2412,
+      "step": 936
+    },
+    {
+      "epoch": 0.59968,
+      "grad_norm": 0.253172980422077,
+      "learning_rate": 1.136637564770409e-05,
+      "loss": 0.2407,
+      "step": 937
+    },
+    {
+      "epoch": 0.60032,
+      "grad_norm": 0.2933110102178001,
+      "learning_rate": 1.1335554188300941e-05,
+      "loss": 0.227,
+      "step": 938
+    },
+    {
+      "epoch": 0.60096,
+      "grad_norm": 0.2668181844305263,
+      "learning_rate": 1.1304749173583767e-05,
+      "loss": 0.2456,
+      "step": 939
+    },
+    {
+      "epoch": 0.6016,
+      "grad_norm": 0.30285003168122526,
+      "learning_rate": 1.1273960741794158e-05,
+      "loss": 0.2366,
+      "step": 940
+    },
+    {
+      "epoch": 0.60224,
+      "grad_norm": 0.2861300051951289,
+      "learning_rate": 1.1243189031099285e-05,
+      "loss": 0.2519,
+      "step": 941
+    },
+    {
+      "epoch": 0.60288,
+      "grad_norm": 0.2700232363591384,
+      "learning_rate": 1.1212434179591267e-05,
+      "loss": 0.2276,
+      "step": 942
+    },
+    {
+      "epoch": 0.60352,
+      "grad_norm": 0.28146618766930903,
+      "learning_rate": 1.1181696325286582e-05,
+      "loss": 0.2473,
+      "step": 943
+    },
+    {
+      "epoch": 0.60416,
+      "grad_norm": 0.2687027470794158,
+      "learning_rate": 1.115097560612541e-05,
+      "loss": 0.2516,
+      "step": 944
+    },
+    {
+      "epoch": 0.6048,
+      "grad_norm": 0.2675754902708525,
+      "learning_rate": 1.1120272159971065e-05,
+      "loss": 0.2241,
+      "step": 945
+    },
+    {
+      "epoch": 0.60544,
+      "grad_norm": 0.290279648929516,
+      "learning_rate": 1.1089586124609309e-05,
+      "loss": 0.239,
+      "step": 946
+    },
+    {
+      "epoch": 0.60608,
+      "grad_norm": 0.294057403686239,
+      "learning_rate": 1.105891763774781e-05,
+      "loss": 0.2318,
+      "step": 947
+    },
+    {
+      "epoch": 0.60672,
+      "grad_norm": 0.25256828227195943,
+      "learning_rate": 1.1028266837015452e-05,
+      "loss": 0.2284,
+      "step": 948
+    },
+    {
+      "epoch": 0.60736,
+      "grad_norm": 0.2758672356678394,
+      "learning_rate": 1.099763385996177e-05,
+      "loss": 0.2269,
+      "step": 949
+    },
+    {
+      "epoch": 0.608,
+      "grad_norm": 0.267987007000224,
+      "learning_rate": 1.0967018844056298e-05,
+      "loss": 0.2244,
+      "step": 950
+    },
+    {
+      "epoch": 0.60864,
+      "grad_norm": 0.2554232644912386,
+      "learning_rate": 1.0936421926687988e-05,
+      "loss": 0.2393,
+      "step": 951
+    },
+    {
+      "epoch": 0.60928,
+      "grad_norm": 0.2707325937678355,
+      "learning_rate": 1.0905843245164545e-05,
+      "loss": 0.2275,
+      "step": 952
+    },
+    {
+      "epoch": 0.60992,
+      "grad_norm": 0.27164108322281655,
+      "learning_rate": 1.0875282936711864e-05,
+      "loss": 0.2415,
+      "step": 953
+    },
+    {
+      "epoch": 0.61056,
+      "grad_norm": 0.26648401964107277,
+      "learning_rate": 1.0844741138473375e-05,
+      "loss": 0.2293,
+      "step": 954
+    },
+    {
+      "epoch": 0.6112,
+      "grad_norm": 0.2521340523387244,
+      "learning_rate": 1.0814217987509439e-05,
+      "loss": 0.2075,
+      "step": 955
+    },
+    {
+      "epoch": 0.61184,
+      "grad_norm": 0.2684643123842368,
+      "learning_rate": 1.0783713620796746e-05,
+      "loss": 0.2375,
+      "step": 956
+    },
+    {
+      "epoch": 0.61248,
+      "grad_norm": 0.27116694710494305,
+      "learning_rate": 1.0753228175227671e-05,
+      "loss": 0.2305,
+      "step": 957
+    },
+    {
+      "epoch": 0.61312,
+      "grad_norm": 0.2590575742273794,
+      "learning_rate": 1.0722761787609705e-05,
+      "loss": 0.2343,
+      "step": 958
+    },
+    {
+      "epoch": 0.61376,
+      "grad_norm": 0.26450915604012226,
+      "learning_rate": 1.0692314594664786e-05,
+      "loss": 0.231,
+      "step": 959
+    },
+    {
+      "epoch": 0.6144,
+      "grad_norm": 0.2772686588210157,
+      "learning_rate": 1.0661886733028727e-05,
+      "loss": 0.2195,
+      "step": 960
+    },
+    {
+      "epoch": 0.61504,
+      "grad_norm": 0.2526658352973938,
+      "learning_rate": 1.0631478339250591e-05,
+      "loss": 0.2493,
+      "step": 961
+    },
+    {
+      "epoch": 0.61568,
+      "grad_norm": 0.25957998768193064,
+      "learning_rate": 1.0601089549792075e-05,
+      "loss": 0.2225,
+      "step": 962
+    },
+    {
+      "epoch": 0.61632,
+      "grad_norm": 0.25769736790438313,
+      "learning_rate": 1.0570720501026887e-05,
+      "loss": 0.2136,
+      "step": 963
+    },
+    {
+      "epoch": 0.61696,
+      "grad_norm": 0.25528326261877526,
+      "learning_rate": 1.0540371329240169e-05,
+      "loss": 0.2372,
+      "step": 964
+    },
+    {
+      "epoch": 0.6176,
+      "grad_norm": 0.249689252596263,
+      "learning_rate": 1.0510042170627837e-05,
+      "loss": 0.2134,
+      "step": 965
+    },
+    {
+      "epoch": 0.61824,
+      "grad_norm": 0.2758362589138092,
+      "learning_rate": 1.0479733161296016e-05,
+      "loss": 0.2357,
+      "step": 966
+    },
+    {
+      "epoch": 0.61888,
+      "grad_norm": 0.28258470196272717,
+      "learning_rate": 1.0449444437260402e-05,
+      "loss": 0.2206,
+      "step": 967
+    },
+    {
+      "epoch": 0.61952,
+      "grad_norm": 0.27809718765019137,
+      "learning_rate": 1.0419176134445648e-05,
+      "loss": 0.246,
+      "step": 968
+    },
+    {
+      "epoch": 0.62016,
+      "grad_norm": 0.2381941918663544,
+      "learning_rate": 1.0388928388684781e-05,
+      "loss": 0.2284,
+      "step": 969
+    },
+    {
+      "epoch": 0.6208,
+      "grad_norm": 0.2841270455338671,
+      "learning_rate": 1.0358701335718555e-05,
+      "loss": 0.2138,
+      "step": 970
+    },
+    {
+      "epoch": 0.62144,
+      "grad_norm": 0.3024383331911469,
+      "learning_rate": 1.0328495111194891e-05,
+      "loss": 0.2184,
+      "step": 971
+    },
+    {
+      "epoch": 0.62208,
+      "grad_norm": 0.2663851401696372,
+      "learning_rate": 1.0298309850668208e-05,
+      "loss": 0.2254,
+      "step": 972
+    },
+    {
+      "epoch": 0.62272,
+      "grad_norm": 0.2854229308183058,
+      "learning_rate": 1.0268145689598871e-05,
+      "loss": 0.2012,
+      "step": 973
+    },
+    {
+      "epoch": 0.62336,
+      "grad_norm": 0.2782131277120352,
+      "learning_rate": 1.0238002763352538e-05,
+      "loss": 0.2275,
+      "step": 974
+    },
+    {
+      "epoch": 0.624,
+      "grad_norm": 0.2707133951459095,
+      "learning_rate": 1.0207881207199593e-05,
+      "loss": 0.2252,
+      "step": 975
+    },
+    {
+      "epoch": 0.62464,
+      "grad_norm": 0.26175738176839053,
+      "learning_rate": 1.01777811563145e-05,
+      "loss": 0.2355,
+      "step": 976
+    },
+    {
+      "epoch": 0.62528,
+      "grad_norm": 0.2850575026663318,
+      "learning_rate": 1.0147702745775237e-05,
+      "loss": 0.2531,
+      "step": 977
+    },
+    {
+      "epoch": 0.62592,
+      "grad_norm": 0.25643392894513545,
+      "learning_rate": 1.0117646110562643e-05,
+      "loss": 0.247,
+      "step": 978
+    },
+    {
+      "epoch": 0.62656,
+      "grad_norm": 0.2809794639097617,
+      "learning_rate": 1.0087611385559855e-05,
+      "loss": 0.2364,
+      "step": 979
+    },
+    {
+      "epoch": 0.6272,
+      "grad_norm": 0.273925223907152,
+      "learning_rate": 1.0057598705551683e-05,
+      "loss": 0.2457,
+      "step": 980
+    },
+    {
+      "epoch": 0.62784,
+      "grad_norm": 0.2611585139690243,
+      "learning_rate": 1.0027608205223997e-05,
+      "loss": 0.2135,
+      "step": 981
+    },
+    {
+      "epoch": 0.62848,
+      "grad_norm": 0.2757636790658412,
+      "learning_rate": 9.99764001916315e-06,
+      "loss": 0.234,
+      "step": 982
+    },
+    {
+      "epoch": 0.62912,
+      "grad_norm": 0.2621085325200055,
+      "learning_rate": 9.967694281855337e-06,
+      "loss": 0.2416,
+      "step": 983
+    },
+    {
+      "epoch": 0.62976,
+      "grad_norm": 0.2536738273794677,
+      "learning_rate": 9.937771127686025e-06,
+      "loss": 0.252,
+      "step": 984
+    },
+    {
+      "epoch": 0.6304,
+      "grad_norm": 0.274093486797772,
+      "learning_rate": 9.90787069093933e-06,
+      "loss": 0.2229,
+      "step": 985
+    },
+    {
+      "epoch": 0.63104,
+      "grad_norm": 0.301875864424414,
+      "learning_rate": 9.877993105797429e-06,
+      "loss": 0.2061,
+      "step": 986
+    },
+    {
+      "epoch": 0.63168,
+      "grad_norm": 0.26810505359168685,
+      "learning_rate": 9.848138506339932e-06,
+      "loss": 0.2208,
+      "step": 987
+    },
+    {
+      "epoch": 0.63232,
+      "grad_norm": 0.2673327996540591,
+      "learning_rate": 9.81830702654332e-06,
+      "loss": 0.2308,
+      "step": 988
+    },
+    {
+      "epoch": 0.63296,
+      "grad_norm": 0.2706612405650287,
+      "learning_rate": 9.788498800280297e-06,
+      "loss": 0.2359,
+      "step": 989
+    },
+    {
+      "epoch": 0.6336,
+      "grad_norm": 0.2811331772020439,
+      "learning_rate": 9.75871396131924e-06,
+      "loss": 0.2138,
+      "step": 990
+    },
+    {
+      "epoch": 0.63424,
+      "grad_norm": 0.27083688175583204,
+      "learning_rate": 9.728952643323558e-06,
+      "loss": 0.2266,
+      "step": 991
+    },
+    {
+      "epoch": 0.63488,
+      "grad_norm": 0.27787455753996565,
+      "learning_rate": 9.699214979851095e-06,
+      "loss": 0.2418,
+      "step": 992
+    },
+    {
+      "epoch": 0.63552,
+      "grad_norm": 0.277957464786416,
+      "learning_rate": 9.669501104353572e-06,
+      "loss": 0.2301,
+      "step": 993
+    },
+    {
+      "epoch": 0.63616,
+      "grad_norm": 0.2698851027191522,
+      "learning_rate": 9.639811150175925e-06,
+      "loss": 0.2403,
+      "step": 994
+    },
+    {
+      "epoch": 0.6368,
+      "grad_norm": 0.2685878765882483,
+      "learning_rate": 9.61014525055577e-06,
+      "loss": 0.2294,
+      "step": 995
+    },
+    {
+      "epoch": 0.63744,
+      "grad_norm": 0.2680664569299894,
+      "learning_rate": 9.580503538622753e-06,
+      "loss": 0.2528,
+      "step": 996
+    },
+    {
+      "epoch": 0.63808,
+      "grad_norm": 0.24690195463824632,
+      "learning_rate": 9.550886147397989e-06,
+      "loss": 0.2299,
+      "step": 997
+    },
+    {
+      "epoch": 0.63872,
+      "grad_norm": 0.2624707704490792,
+      "learning_rate": 9.521293209793431e-06,
+      "loss": 0.2316,
+      "step": 998
+    },
+    {
+      "epoch": 0.63936,
+      "grad_norm": 0.2928803573448213,
+      "learning_rate": 9.491724858611323e-06,
+      "loss": 0.248,
+      "step": 999
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.28532627144669237,
+      "learning_rate": 9.462181226543545e-06,
+      "loss": 0.2337,
+      "step": 1000
+    },
+    {
+      "epoch": 0.64064,
+      "grad_norm": 0.2874639246394616,
+      "learning_rate": 9.43266244617107e-06,
+      "loss": 0.2198,
+      "step": 1001
+    },
+    {
+      "epoch": 0.64128,
+      "grad_norm": 0.3005802763115371,
+      "learning_rate": 9.403168649963325e-06,
+      "loss": 0.2425,
+      "step": 1002
+    },
+    {
+      "epoch": 0.64192,
+      "grad_norm": 0.27021178999951345,
+      "learning_rate": 9.37369997027763e-06,
+      "loss": 0.2357,
+      "step": 1003
+    },
+    {
+      "epoch": 0.64256,
+      "grad_norm": 0.295266429086591,
+      "learning_rate": 9.344256539358598e-06,
+      "loss": 0.2179,
+      "step": 1004
+    },
+    {
+      "epoch": 0.6432,
+      "grad_norm": 0.268243598549565,
+      "learning_rate": 9.314838489337512e-06,
+      "loss": 0.2323,
+      "step": 1005
+    },
+    {
+      "epoch": 0.64384,
+      "grad_norm": 0.28382119167552344,
+      "learning_rate": 9.285445952231777e-06,
+      "loss": 0.2191,
+      "step": 1006
+    },
+    {
+      "epoch": 0.64448,
+      "grad_norm": 0.28583856240407873,
+      "learning_rate": 9.256079059944295e-06,
+      "loss": 0.236,
+      "step": 1007
+    },
+    {
+      "epoch": 0.64512,
+      "grad_norm": 0.2665670998720098,
+      "learning_rate": 9.226737944262879e-06,
+      "loss": 0.2404,
+      "step": 1008
+    },
+    {
+      "epoch": 0.64576,
+      "grad_norm": 0.2618331652239897,
+      "learning_rate": 9.197422736859681e-06,
+      "loss": 0.2515,
+      "step": 1009
+    },
+    {
+      "epoch": 0.6464,
+      "grad_norm": 0.2743575127281784,
+      "learning_rate": 9.168133569290574e-06,
+      "loss": 0.2366,
+      "step": 1010
+    },
+    {
+      "epoch": 0.64704,
+      "grad_norm": 0.27112251220328204,
+      "learning_rate": 9.138870572994573e-06,
+      "loss": 0.2441,
+      "step": 1011
+    },
+    {
+      "epoch": 0.64768,
+      "grad_norm": 0.259583608714577,
+      "learning_rate": 9.109633879293258e-06,
+      "loss": 0.2345,
+      "step": 1012
+    },
+    {
+      "epoch": 0.64832,
+      "grad_norm": 0.26936694813868906,
+      "learning_rate": 9.080423619390156e-06,
+      "loss": 0.2245,
+      "step": 1013
+    },
+    {
+      "epoch": 0.64896,
+      "grad_norm": 0.23709705994485894,
+      "learning_rate": 9.051239924370187e-06,
+      "loss": 0.2147,
+      "step": 1014
+    },
+    {
+      "epoch": 0.6496,
+      "grad_norm": 0.320565619860577,
+      "learning_rate": 9.02208292519904e-06,
+      "loss": 0.2686,
+      "step": 1015
+    },
+    {
+      "epoch": 0.65024,
+      "grad_norm": 0.23594177473213718,
+      "learning_rate": 8.992952752722612e-06,
+      "loss": 0.2373,
+      "step": 1016
+    },
+    {
+      "epoch": 0.65088,
+      "grad_norm": 0.2610455828430524,
+      "learning_rate": 8.963849537666416e-06,
+      "loss": 0.2298,
+      "step": 1017
+    },
+    {
+      "epoch": 0.65152,
+      "grad_norm": 0.27883308538071555,
+      "learning_rate": 8.934773410634979e-06,
+      "loss": 0.2284,
+      "step": 1018
+    },
+    {
+      "epoch": 0.65216,
+      "grad_norm": 0.243708170639781,
+      "learning_rate": 8.90572450211128e-06,
+      "loss": 0.2381,
+      "step": 1019
+    },
+    {
+      "epoch": 0.6528,
+      "grad_norm": 0.25214049350809076,
+      "learning_rate": 8.876702942456135e-06,
+      "loss": 0.2236,
+      "step": 1020
+    },
+    {
+      "epoch": 0.65344,
+      "grad_norm": 0.2695393266182502,
+      "learning_rate": 8.847708861907647e-06,
+      "loss": 0.2518,
+      "step": 1021
+    },
+    {
+      "epoch": 0.65408,
+      "grad_norm": 0.2894111851022704,
+      "learning_rate": 8.818742390580579e-06,
+      "loss": 0.2102,
+      "step": 1022
+    },
+    {
+      "epoch": 0.65472,
+      "grad_norm": 0.28025286701489427,
+      "learning_rate": 8.789803658465822e-06,
+      "loss": 0.2374,
+      "step": 1023
+    },
+    {
+      "epoch": 0.65536,
+      "grad_norm": 0.28195313787899245,
+      "learning_rate": 8.760892795429758e-06,
+      "loss": 0.2298,
+      "step": 1024
+    },
+    {
+      "epoch": 0.656,
+      "grad_norm": 0.26147501573586185,
+      "learning_rate": 8.732009931213728e-06,
+      "loss": 0.2374,
+      "step": 1025
+    },
+    {
+      "epoch": 0.65664,
+      "grad_norm": 0.26624563474494223,
+      "learning_rate": 8.703155195433398e-06,
+      "loss": 0.2136,
+      "step": 1026
+    },
+    {
+      "epoch": 0.65728,
+      "grad_norm": 0.29498065678389157,
+      "learning_rate": 8.674328717578224e-06,
+      "loss": 0.2542,
+      "step": 1027
+    },
+    {
+      "epoch": 0.65792,
+      "grad_norm": 0.2592368682687856,
+      "learning_rate": 8.645530627010841e-06,
+      "loss": 0.2289,
+      "step": 1028
+    },
+    {
+      "epoch": 0.65856,
+      "grad_norm": 0.5187956052357068,
+      "learning_rate": 8.616761052966497e-06,
+      "loss": 0.2092,
+      "step": 1029
+    },
+    {
+      "epoch": 0.6592,
+      "grad_norm": 0.2727221527101844,
+      "learning_rate": 8.588020124552466e-06,
+      "loss": 0.2427,
+      "step": 1030
+    },
+    {
+      "epoch": 0.65984,
+      "grad_norm": 0.25445411990157907,
+      "learning_rate": 8.559307970747471e-06,
+      "loss": 0.2211,
+      "step": 1031
+    },
+    {
+      "epoch": 0.66048,
+      "grad_norm": 0.24623299916346522,
+      "learning_rate": 8.530624720401107e-06,
+      "loss": 0.2155,
+      "step": 1032
+    },
+    {
+      "epoch": 0.66112,
+      "grad_norm": 0.24184315685116256,
+      "learning_rate": 8.501970502233253e-06,
+      "loss": 0.2348,
+      "step": 1033
+    },
+    {
+      "epoch": 0.66176,
+      "grad_norm": 0.2721264275062898,
+      "learning_rate": 8.473345444833518e-06,
+      "loss": 0.2321,
+      "step": 1034
+    },
+    {
+      "epoch": 0.6624,
+      "grad_norm": 0.2730712748831622,
+      "learning_rate": 8.444749676660633e-06,
+      "loss": 0.2418,
+      "step": 1035
+    },
+    {
+      "epoch": 0.66304,
+      "grad_norm": 0.2721775229972965,
+      "learning_rate": 8.416183326041903e-06,
+      "loss": 0.2245,
+      "step": 1036
+    },
+    {
+      "epoch": 0.66368,
+      "grad_norm": 0.2462298760618664,
+      "learning_rate": 8.387646521172593e-06,
+      "loss": 0.2356,
+      "step": 1037
+    },
+    {
+      "epoch": 0.66432,
+      "grad_norm": 0.2601849010444262,
+      "learning_rate": 8.359139390115416e-06,
+      "loss": 0.2481,
+      "step": 1038
+    },
+    {
+      "epoch": 0.66496,
+      "grad_norm": 0.26359632890879,
+      "learning_rate": 8.330662060799878e-06,
+      "loss": 0.2459,
+      "step": 1039
+    },
+    {
+      "epoch": 0.6656,
+      "grad_norm": 0.2394782595448413,
+      "learning_rate": 8.302214661021784e-06,
+      "loss": 0.2119,
+      "step": 1040
+    },
+    {
+      "epoch": 0.66624,
+      "grad_norm": 0.26670478343946696,
+      "learning_rate": 8.273797318442608e-06,
+      "loss": 0.216,
+      "step": 1041
+    },
+    {
+      "epoch": 0.66688,
+      "grad_norm": 0.2918048313261061,
+      "learning_rate": 8.24541016058892e-06,
+      "loss": 0.2311,
+      "step": 1042
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1562,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 521,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 477370166870016.0,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}