diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,5515 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.50112,
+  "eval_steps": 500,
+  "global_step": 783,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.00064,
+      "grad_norm": 2.618795845193889,
+      "learning_rate": 0.0,
+      "loss": 0.5764,
+      "step": 1
+    },
+    {
+      "epoch": 0.00128,
+      "grad_norm": 2.6347789365042167,
+      "learning_rate": 6.329113924050633e-07,
+      "loss": 0.6154,
+      "step": 2
+    },
+    {
+      "epoch": 0.00192,
+      "grad_norm": 2.7588406940768695,
+      "learning_rate": 1.2658227848101265e-06,
+      "loss": 0.5763,
+      "step": 3
+    },
+    {
+      "epoch": 0.00256,
+      "grad_norm": 3.0430509991625927,
+      "learning_rate": 1.8987341772151901e-06,
+      "loss": 0.5897,
+      "step": 4
+    },
+    {
+      "epoch": 0.0032,
+      "grad_norm": 3.0181149898389754,
+      "learning_rate": 2.531645569620253e-06,
+      "loss": 0.5849,
+      "step": 5
+    },
+    {
+      "epoch": 0.00384,
+      "grad_norm": 2.708156370622355,
+      "learning_rate": 3.1645569620253167e-06,
+      "loss": 0.6007,
+      "step": 6
+    },
+    {
+      "epoch": 0.00448,
+      "grad_norm": 2.2121700715533077,
+      "learning_rate": 3.7974683544303802e-06,
+      "loss": 0.5589,
+      "step": 7
+    },
+    {
+      "epoch": 0.00512,
+      "grad_norm": 2.5547904662956786,
+      "learning_rate": 4.430379746835443e-06,
+      "loss": 0.5868,
+      "step": 8
+    },
+    {
+      "epoch": 0.00576,
+      "grad_norm": 15.741913008515068,
+      "learning_rate": 5.063291139240506e-06,
+      "loss": 0.5496,
+      "step": 9
+    },
+    {
+      "epoch": 0.0064,
+      "grad_norm": 1.4025681078646888,
+      "learning_rate": 5.69620253164557e-06,
+      "loss": 0.5335,
+      "step": 10
+    },
+    {
+      "epoch": 0.00704,
+      "grad_norm": 2.296825978922617,
+      "learning_rate": 6.329113924050633e-06,
+      "loss": 0.5614,
+      "step": 11
+    },
+    {
+      "epoch": 0.00768,
+      "grad_norm": 1.3790308520522885,
+      "learning_rate": 6.9620253164556965e-06,
+      "loss": 0.5155,
+      "step": 12
+    },
+    {
+      "epoch": 0.00832,
+      "grad_norm": 1.5578214066440281,
+      "learning_rate": 7.5949367088607605e-06,
+      "loss": 0.4714,
+      "step": 13
+    },
+    {
+      "epoch": 0.00896,
+      "grad_norm": 1.6513983171037458,
+      "learning_rate": 8.227848101265822e-06,
+      "loss": 0.4844,
+      "step": 14
+    },
+    {
+      "epoch": 0.0096,
+      "grad_norm": 1.0659029951583614,
+      "learning_rate": 8.860759493670886e-06,
+      "loss": 0.4445,
+      "step": 15
+    },
+    {
+      "epoch": 0.01024,
+      "grad_norm": 1.5553885236728338,
+      "learning_rate": 9.49367088607595e-06,
+      "loss": 0.4296,
+      "step": 16
+    },
+    {
+      "epoch": 0.01088,
+      "grad_norm": 1.6423185898553931,
+      "learning_rate": 1.0126582278481012e-05,
+      "loss": 0.4558,
+      "step": 17
+    },
+    {
+      "epoch": 0.01152,
+      "grad_norm": 1.337112701272778,
+      "learning_rate": 1.0759493670886076e-05,
+      "loss": 0.4572,
+      "step": 18
+    },
+    {
+      "epoch": 0.01216,
+      "grad_norm": 1.091077579656484,
+      "learning_rate": 1.139240506329114e-05,
+      "loss": 0.4194,
+      "step": 19
+    },
+    {
+      "epoch": 0.0128,
+      "grad_norm": 0.8415436037519902,
+      "learning_rate": 1.2025316455696203e-05,
+      "loss": 0.43,
+      "step": 20
+    },
+    {
+      "epoch": 0.01344,
+      "grad_norm": 1.0942576082541409,
+      "learning_rate": 1.2658227848101267e-05,
+      "loss": 0.4342,
+      "step": 21
+    },
+    {
+      "epoch": 0.01408,
+      "grad_norm": 0.9266622560738933,
+      "learning_rate": 1.3291139240506329e-05,
+      "loss": 0.3692,
+      "step": 22
+    },
+    {
+      "epoch": 0.01472,
+      "grad_norm": 0.730124403683561,
+      "learning_rate": 1.3924050632911393e-05,
+      "loss": 0.3842,
+      "step": 23
+    },
+    {
+      "epoch": 0.01536,
+      "grad_norm": 1.0621346265160687,
+      "learning_rate": 1.4556962025316457e-05,
+      "loss": 0.3548,
+      "step": 24
+    },
+    {
+      "epoch": 0.016,
+      "grad_norm": 0.9227953803036282,
+      "learning_rate": 1.5189873417721521e-05,
+      "loss": 0.3948,
+      "step": 25
+    },
+    {
+      "epoch": 0.01664,
+      "grad_norm": 0.736727893263626,
+      "learning_rate": 1.5822784810126583e-05,
+      "loss": 0.375,
+      "step": 26
+    },
+    {
+      "epoch": 0.01728,
+      "grad_norm": 2.053133845104068,
+      "learning_rate": 1.6455696202531644e-05,
+      "loss": 0.3719,
+      "step": 27
+    },
+    {
+      "epoch": 0.01792,
+      "grad_norm": 0.6875688796188869,
+      "learning_rate": 1.7088607594936708e-05,
+      "loss": 0.3723,
+      "step": 28
+    },
+    {
+      "epoch": 0.01856,
+      "grad_norm": 0.6623745034402684,
+      "learning_rate": 1.7721518987341772e-05,
+      "loss": 0.3588,
+      "step": 29
+    },
+    {
+      "epoch": 0.0192,
+      "grad_norm": 0.7263864736482054,
+      "learning_rate": 1.8354430379746836e-05,
+      "loss": 0.3851,
+      "step": 30
+    },
+    {
+      "epoch": 0.01984,
+      "grad_norm": 0.5571337797316878,
+      "learning_rate": 1.89873417721519e-05,
+      "loss": 0.3792,
+      "step": 31
+    },
+    {
+      "epoch": 0.02048,
+      "grad_norm": 0.9258693296754104,
+      "learning_rate": 1.962025316455696e-05,
+      "loss": 0.3852,
+      "step": 32
+    },
+    {
+      "epoch": 0.02112,
+      "grad_norm": 0.8990784229406635,
+      "learning_rate": 2.0253164556962025e-05,
+      "loss": 0.3505,
+      "step": 33
+    },
+    {
+      "epoch": 0.02176,
+      "grad_norm": 2.3509423854017837,
+      "learning_rate": 2.088607594936709e-05,
+      "loss": 0.364,
+      "step": 34
+    },
+    {
+      "epoch": 0.0224,
+      "grad_norm": 2.6014985443321215,
+      "learning_rate": 2.1518987341772153e-05,
+      "loss": 0.3637,
+      "step": 35
+    },
+    {
+      "epoch": 0.02304,
+      "grad_norm": 0.7225091993923429,
+      "learning_rate": 2.2151898734177217e-05,
+      "loss": 0.3794,
+      "step": 36
+    },
+    {
+      "epoch": 0.02368,
+      "grad_norm": 0.6641449499251059,
+      "learning_rate": 2.278481012658228e-05,
+      "loss": 0.3699,
+      "step": 37
+    },
+    {
+      "epoch": 0.02432,
+      "grad_norm": 0.8028481090301588,
+      "learning_rate": 2.341772151898734e-05,
+      "loss": 0.3647,
+      "step": 38
+    },
+    {
+      "epoch": 0.02496,
+      "grad_norm": 0.6565265160500543,
+      "learning_rate": 2.4050632911392405e-05,
+      "loss": 0.3579,
+      "step": 39
+    },
+    {
+      "epoch": 0.0256,
+      "grad_norm": 0.7322064983471163,
+      "learning_rate": 2.468354430379747e-05,
+      "loss": 0.3484,
+      "step": 40
+    },
+    {
+      "epoch": 0.02624,
+      "grad_norm": 0.5039022880384163,
+      "learning_rate": 2.5316455696202533e-05,
+      "loss": 0.3096,
+      "step": 41
+    },
+    {
+      "epoch": 0.02688,
+      "grad_norm": 0.5483302917388222,
+      "learning_rate": 2.5949367088607597e-05,
+      "loss": 0.3371,
+      "step": 42
+    },
+    {
+      "epoch": 0.02752,
+      "grad_norm": 1.3783780747410168,
+      "learning_rate": 2.6582278481012658e-05,
+      "loss": 0.3263,
+      "step": 43
+    },
+    {
+      "epoch": 0.02816,
+      "grad_norm": 0.5158426665517652,
+      "learning_rate": 2.7215189873417722e-05,
+      "loss": 0.3449,
+      "step": 44
+    },
+    {
+      "epoch": 0.0288,
+      "grad_norm": 0.5751703137103499,
+      "learning_rate": 2.7848101265822786e-05,
+      "loss": 0.3561,
+      "step": 45
+    },
+    {
+      "epoch": 0.02944,
+      "grad_norm": 0.6590394546695514,
+      "learning_rate": 2.848101265822785e-05,
+      "loss": 0.3478,
+      "step": 46
+    },
+    {
+      "epoch": 0.03008,
+      "grad_norm": 0.5214718720462116,
+      "learning_rate": 2.9113924050632914e-05,
+      "loss": 0.3408,
+      "step": 47
+    },
+    {
+      "epoch": 0.03072,
+      "grad_norm": 0.9091430743600112,
+      "learning_rate": 2.9746835443037974e-05,
+      "loss": 0.3526,
+      "step": 48
+    },
+    {
+      "epoch": 0.03136,
+      "grad_norm": 1.1052640742837447,
+      "learning_rate": 3.0379746835443042e-05,
+      "loss": 0.3346,
+      "step": 49
+    },
+    {
+      "epoch": 0.032,
+      "grad_norm": 0.48507046472864723,
+      "learning_rate": 3.10126582278481e-05,
+      "loss": 0.3443,
+      "step": 50
+    },
+    {
+      "epoch": 0.03264,
+      "grad_norm": 0.5388534097509862,
+      "learning_rate": 3.1645569620253167e-05,
+      "loss": 0.3312,
+      "step": 51
+    },
+    {
+      "epoch": 0.03328,
+      "grad_norm": 0.5894091981022138,
+      "learning_rate": 3.227848101265823e-05,
+      "loss": 0.3317,
+      "step": 52
+    },
+    {
+      "epoch": 0.03392,
+      "grad_norm": 0.4751863643628404,
+      "learning_rate": 3.291139240506329e-05,
+      "loss": 0.2952,
+      "step": 53
+    },
+    {
+      "epoch": 0.03456,
+      "grad_norm": 0.5790137817312099,
+      "learning_rate": 3.354430379746836e-05,
+      "loss": 0.3392,
+      "step": 54
+    },
+    {
+      "epoch": 0.0352,
+      "grad_norm": 1.5087120760365582,
+      "learning_rate": 3.4177215189873416e-05,
+      "loss": 0.3339,
+      "step": 55
+    },
+    {
+      "epoch": 0.03584,
+      "grad_norm": 0.5745952365097137,
+      "learning_rate": 3.4810126582278487e-05,
+      "loss": 0.3347,
+      "step": 56
+    },
+    {
+      "epoch": 0.03648,
+      "grad_norm": 0.7517935525029015,
+      "learning_rate": 3.5443037974683544e-05,
+      "loss": 0.3324,
+      "step": 57
+    },
+    {
+      "epoch": 0.03712,
+      "grad_norm": 0.5324903887679082,
+      "learning_rate": 3.607594936708861e-05,
+      "loss": 0.3522,
+      "step": 58
+    },
+    {
+      "epoch": 0.03776,
+      "grad_norm": 0.5095112678612309,
+      "learning_rate": 3.670886075949367e-05,
+      "loss": 0.3394,
+      "step": 59
+    },
+    {
+      "epoch": 0.0384,
+      "grad_norm": 0.5442974811657504,
+      "learning_rate": 3.7341772151898736e-05,
+      "loss": 0.3322,
+      "step": 60
+    },
+    {
+      "epoch": 0.03904,
+      "grad_norm": 0.5228282028496956,
+      "learning_rate": 3.79746835443038e-05,
+      "loss": 0.34,
+      "step": 61
+    },
+    {
+      "epoch": 0.03968,
+      "grad_norm": 0.5234005807075949,
+      "learning_rate": 3.8607594936708864e-05,
+      "loss": 0.3205,
+      "step": 62
+    },
+    {
+      "epoch": 0.04032,
+      "grad_norm": 0.5456894801995082,
+      "learning_rate": 3.924050632911392e-05,
+      "loss": 0.3264,
+      "step": 63
+    },
+    {
+      "epoch": 0.04096,
+      "grad_norm": 0.6064859304760205,
+      "learning_rate": 3.987341772151899e-05,
+      "loss": 0.3603,
+      "step": 64
+    },
+    {
+      "epoch": 0.0416,
+      "grad_norm": 0.4817896106905655,
+      "learning_rate": 4.050632911392405e-05,
+      "loss": 0.3159,
+      "step": 65
+    },
+    {
+      "epoch": 0.04224,
+      "grad_norm": 0.5788597012081875,
+      "learning_rate": 4.113924050632912e-05,
+      "loss": 0.3509,
+      "step": 66
+    },
+    {
+      "epoch": 0.04288,
+      "grad_norm": 0.5412218771204261,
+      "learning_rate": 4.177215189873418e-05,
+      "loss": 0.3208,
+      "step": 67
+    },
+    {
+      "epoch": 0.04352,
+      "grad_norm": 1.3974653981931218,
+      "learning_rate": 4.240506329113924e-05,
+      "loss": 0.3319,
+      "step": 68
+    },
+    {
+      "epoch": 0.04416,
+      "grad_norm": 0.5446574915363599,
+      "learning_rate": 4.3037974683544305e-05,
+      "loss": 0.2997,
+      "step": 69
+    },
+    {
+      "epoch": 0.0448,
+      "grad_norm": 0.4568551671818489,
+      "learning_rate": 4.367088607594937e-05,
+      "loss": 0.3361,
+      "step": 70
+    },
+    {
+      "epoch": 0.04544,
+      "grad_norm": 0.5949709296262449,
+      "learning_rate": 4.430379746835443e-05,
+      "loss": 0.3211,
+      "step": 71
+    },
+    {
+      "epoch": 0.04608,
+      "grad_norm": 0.5482381035251598,
+      "learning_rate": 4.49367088607595e-05,
+      "loss": 0.3292,
+      "step": 72
+    },
+    {
+      "epoch": 0.04672,
+      "grad_norm": 0.4997143028932898,
+      "learning_rate": 4.556962025316456e-05,
+      "loss": 0.318,
+      "step": 73
+    },
+    {
+      "epoch": 0.04736,
+      "grad_norm": 0.5841404500165603,
+      "learning_rate": 4.6202531645569625e-05,
+      "loss": 0.3138,
+      "step": 74
+    },
+    {
+      "epoch": 0.048,
+      "grad_norm": 0.5028458152612025,
+      "learning_rate": 4.683544303797468e-05,
+      "loss": 0.3029,
+      "step": 75
+    },
+    {
+      "epoch": 0.04864,
+      "grad_norm": 0.48112579785481197,
+      "learning_rate": 4.7468354430379746e-05,
+      "loss": 0.3243,
+      "step": 76
+    },
+    {
+      "epoch": 0.04928,
+      "grad_norm": 1.0013986736975447,
+      "learning_rate": 4.810126582278481e-05,
+      "loss": 0.3308,
+      "step": 77
+    },
+    {
+      "epoch": 0.04992,
+      "grad_norm": 0.4247544379828167,
+      "learning_rate": 4.8734177215189874e-05,
+      "loss": 0.2958,
+      "step": 78
+    },
+    {
+      "epoch": 0.05056,
+      "grad_norm": 0.5421182493231368,
+      "learning_rate": 4.936708860759494e-05,
+      "loss": 0.3297,
+      "step": 79
+    },
+    {
+      "epoch": 0.0512,
+      "grad_norm": 1.2844161669496286,
+      "learning_rate": 5e-05,
+      "loss": 0.3198,
+      "step": 80
+    },
+    {
+      "epoch": 0.05184,
+      "grad_norm": 0.5627454560158847,
+      "learning_rate": 4.999994390459423e-05,
+      "loss": 0.3311,
+      "step": 81
+    },
+    {
+      "epoch": 0.05248,
+      "grad_norm": 0.4505373699775162,
+      "learning_rate": 4.999977561862863e-05,
+      "loss": 0.3227,
+      "step": 82
+    },
+    {
+      "epoch": 0.05312,
+      "grad_norm": 0.5349576256311448,
+      "learning_rate": 4.999949514285842e-05,
+      "loss": 0.323,
+      "step": 83
+    },
+    {
+      "epoch": 0.05376,
+      "grad_norm": 0.5591650619223139,
+      "learning_rate": 4.999910247854227e-05,
+      "loss": 0.336,
+      "step": 84
+    },
+    {
+      "epoch": 0.0544,
+      "grad_norm": 0.6827279250952998,
+      "learning_rate": 4.99985976274423e-05,
+      "loss": 0.3319,
+      "step": 85
+    },
+    {
+      "epoch": 0.05504,
+      "grad_norm": 0.7102827110283111,
+      "learning_rate": 4.9997980591824126e-05,
+      "loss": 0.3137,
+      "step": 86
+    },
+    {
+      "epoch": 0.05568,
+      "grad_norm": 0.5941384980560921,
+      "learning_rate": 4.999725137445674e-05,
+      "loss": 0.3242,
+      "step": 87
+    },
+    {
+      "epoch": 0.05632,
+      "grad_norm": 0.5293400414276046,
+      "learning_rate": 4.999640997861264e-05,
+      "loss": 0.3266,
+      "step": 88
+    },
+    {
+      "epoch": 0.05696,
+      "grad_norm": 0.6396725464059344,
+      "learning_rate": 4.999545640806766e-05,
+      "loss": 0.3297,
+      "step": 89
+    },
+    {
+      "epoch": 0.0576,
+      "grad_norm": 0.47264233800825245,
+      "learning_rate": 4.999439066710111e-05,
+      "loss": 0.347,
+      "step": 90
+    },
+    {
+      "epoch": 0.05824,
+      "grad_norm": 0.5476181291945234,
+      "learning_rate": 4.999321276049563e-05,
+      "loss": 0.3222,
+      "step": 91
+    },
+    {
+      "epoch": 0.05888,
+      "grad_norm": 0.5706613009258329,
+      "learning_rate": 4.9991922693537226e-05,
+      "loss": 0.3122,
+      "step": 92
+    },
+    {
+      "epoch": 0.05952,
+      "grad_norm": 0.48706789610757445,
+      "learning_rate": 4.9990520472015246e-05,
+      "loss": 0.2993,
+      "step": 93
+    },
+    {
+      "epoch": 0.06016,
+      "grad_norm": 0.4115404611758198,
+      "learning_rate": 4.998900610222236e-05,
+      "loss": 0.3099,
+      "step": 94
+    },
+    {
+      "epoch": 0.0608,
+      "grad_norm": 0.532905037021598,
+      "learning_rate": 4.998737959095449e-05,
+      "loss": 0.2827,
+      "step": 95
+    },
+    {
+      "epoch": 0.06144,
+      "grad_norm": 0.4643257281999667,
+      "learning_rate": 4.998564094551081e-05,
+      "loss": 0.3205,
+      "step": 96
+    },
+    {
+      "epoch": 0.06208,
+      "grad_norm": 0.5191147113172441,
+      "learning_rate": 4.998379017369375e-05,
+      "loss": 0.3144,
+      "step": 97
+    },
+    {
+      "epoch": 0.06272,
+      "grad_norm": 0.5348016711569842,
+      "learning_rate": 4.9981827283808855e-05,
+      "loss": 0.3128,
+      "step": 98
+    },
+    {
+      "epoch": 0.06336,
+      "grad_norm": 0.5010085634276595,
+      "learning_rate": 4.997975228466489e-05,
+      "loss": 0.3404,
+      "step": 99
+    },
+    {
+      "epoch": 0.064,
+      "grad_norm": 0.4889491939114262,
+      "learning_rate": 4.997756518557368e-05,
+      "loss": 0.2828,
+      "step": 100
+    },
+    {
+      "epoch": 0.06464,
+      "grad_norm": 0.46482892338474413,
+      "learning_rate": 4.997526599635011e-05,
+      "loss": 0.2892,
+      "step": 101
+    },
+    {
+      "epoch": 0.06528,
+      "grad_norm": 0.5287297970963423,
+      "learning_rate": 4.997285472731211e-05,
+      "loss": 0.3397,
+      "step": 102
+    },
+    {
+      "epoch": 0.06592,
+      "grad_norm": 0.46498628733863295,
+      "learning_rate": 4.997033138928056e-05,
+      "loss": 0.3187,
+      "step": 103
+    },
+    {
+      "epoch": 0.06656,
+      "grad_norm": 0.5394842738162251,
+      "learning_rate": 4.996769599357927e-05,
+      "loss": 0.3136,
+      "step": 104
+    },
+    {
+      "epoch": 0.0672,
+      "grad_norm": 0.5304962827503229,
+      "learning_rate": 4.996494855203493e-05,
+      "loss": 0.3286,
+      "step": 105
+    },
+    {
+      "epoch": 0.06784,
+      "grad_norm": 0.5626109543739455,
+      "learning_rate": 4.996208907697706e-05,
+      "loss": 0.3256,
+      "step": 106
+    },
+    {
+      "epoch": 0.06848,
+      "grad_norm": 0.47113589346903195,
+      "learning_rate": 4.995911758123791e-05,
+      "loss": 0.287,
+      "step": 107
+    },
+    {
+      "epoch": 0.06912,
+      "grad_norm": 0.5351822398424532,
+      "learning_rate": 4.995603407815249e-05,
+      "loss": 0.2996,
+      "step": 108
+    },
+    {
+      "epoch": 0.06976,
+      "grad_norm": 0.46338886806198026,
+      "learning_rate": 4.9952838581558406e-05,
+      "loss": 0.3056,
+      "step": 109
+    },
+    {
+      "epoch": 0.0704,
+      "grad_norm": 0.5172304018785423,
+      "learning_rate": 4.994953110579588e-05,
+      "loss": 0.3192,
+      "step": 110
+    },
+    {
+      "epoch": 0.07104,
+      "grad_norm": 0.510154274412123,
+      "learning_rate": 4.9946111665707645e-05,
+      "loss": 0.3307,
+      "step": 111
+    },
+    {
+      "epoch": 0.07168,
+      "grad_norm": 0.3838546254703525,
+      "learning_rate": 4.99425802766389e-05,
+      "loss": 0.2929,
+      "step": 112
+    },
+    {
+      "epoch": 0.07232,
+      "grad_norm": 0.5449189029345786,
+      "learning_rate": 4.993893695443721e-05,
+      "loss": 0.3069,
+      "step": 113
+    },
+    {
+      "epoch": 0.07296,
+      "grad_norm": 0.44577057887896443,
+      "learning_rate": 4.993518171545247e-05,
+      "loss": 0.2966,
+      "step": 114
+    },
+    {
+      "epoch": 0.0736,
+      "grad_norm": 0.4515440453670323,
+      "learning_rate": 4.993131457653682e-05,
+      "loss": 0.3061,
+      "step": 115
+    },
+    {
+      "epoch": 0.07424,
+      "grad_norm": 0.44391881310796194,
+      "learning_rate": 4.992733555504454e-05,
+      "loss": 0.3215,
+      "step": 116
+    },
+    {
+      "epoch": 0.07488,
+      "grad_norm": 0.42638885395980747,
+      "learning_rate": 4.9923244668832025e-05,
+      "loss": 0.3111,
+      "step": 117
+    },
+    {
+      "epoch": 0.07552,
+      "grad_norm": 0.44150879225613077,
+      "learning_rate": 4.991904193625768e-05,
+      "loss": 0.316,
+      "step": 118
+    },
+    {
+      "epoch": 0.07616,
+      "grad_norm": 0.4431237901584868,
+      "learning_rate": 4.991472737618181e-05,
+      "loss": 0.2776,
+      "step": 119
+    },
+    {
+      "epoch": 0.0768,
+      "grad_norm": 0.42121400061187486,
+      "learning_rate": 4.991030100796659e-05,
+      "loss": 0.3064,
+      "step": 120
+    },
+    {
+      "epoch": 0.07744,
+      "grad_norm": 0.392309441081599,
+      "learning_rate": 4.9905762851475915e-05,
+      "loss": 0.3209,
+      "step": 121
+    },
+    {
+      "epoch": 0.07808,
+      "grad_norm": 0.45305290693167705,
+      "learning_rate": 4.990111292707537e-05,
+      "loss": 0.3183,
+      "step": 122
+    },
+    {
+      "epoch": 0.07872,
+      "grad_norm": 0.4293788826264019,
+      "learning_rate": 4.989635125563211e-05,
+      "loss": 0.2924,
+      "step": 123
+    },
+    {
+      "epoch": 0.07936,
+      "grad_norm": 0.4272809824845029,
+      "learning_rate": 4.989147785851477e-05,
+      "loss": 0.3137,
+      "step": 124
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.417177718336002,
+      "learning_rate": 4.988649275759335e-05,
+      "loss": 0.323,
+      "step": 125
+    },
+    {
+      "epoch": 0.08064,
+      "grad_norm": 0.6352795142091163,
+      "learning_rate": 4.988139597523916e-05,
+      "loss": 0.2968,
+      "step": 126
+    },
+    {
+      "epoch": 0.08128,
+      "grad_norm": 0.4311613332706907,
+      "learning_rate": 4.98761875343247e-05,
+      "loss": 0.2864,
+      "step": 127
+    },
+    {
+      "epoch": 0.08192,
+      "grad_norm": 0.407677725063968,
+      "learning_rate": 4.987086745822352e-05,
+      "loss": 0.2887,
+      "step": 128
+    },
+    {
+      "epoch": 0.08256,
+      "grad_norm": 0.43620593917226175,
+      "learning_rate": 4.986543577081017e-05,
+      "loss": 0.326,
+      "step": 129
+    },
+    {
+      "epoch": 0.0832,
+      "grad_norm": 0.3577110758670567,
+      "learning_rate": 4.985989249646007e-05,
+      "loss": 0.3049,
+      "step": 130
+    },
+    {
+      "epoch": 0.08384,
+      "grad_norm": 0.40530215341984516,
+      "learning_rate": 4.985423766004939e-05,
+      "loss": 0.3139,
+      "step": 131
+    },
+    {
+      "epoch": 0.08448,
+      "grad_norm": 0.46328257103904175,
+      "learning_rate": 4.9848471286954965e-05,
+      "loss": 0.3306,
+      "step": 132
+    },
+    {
+      "epoch": 0.08512,
+      "grad_norm": 0.38436097877243486,
+      "learning_rate": 4.984259340305416e-05,
+      "loss": 0.2931,
+      "step": 133
+    },
+    {
+      "epoch": 0.08576,
+      "grad_norm": 0.6093078034059375,
+      "learning_rate": 4.9836604034724753e-05,
+      "loss": 0.2963,
+      "step": 134
+    },
+    {
+      "epoch": 0.0864,
+      "grad_norm": 0.46772762171913185,
+      "learning_rate": 4.9830503208844835e-05,
+      "loss": 0.2979,
+      "step": 135
+    },
+    {
+      "epoch": 0.08704,
+      "grad_norm": 0.48325714173598944,
+      "learning_rate": 4.982429095279266e-05,
+      "loss": 0.2984,
+      "step": 136
+    },
+    {
+      "epoch": 0.08768,
+      "grad_norm": 0.5477120347200936,
+      "learning_rate": 4.981796729444655e-05,
+      "loss": 0.3302,
+      "step": 137
+    },
+    {
+      "epoch": 0.08832,
+      "grad_norm": 0.534004632135256,
+      "learning_rate": 4.981153226218477e-05,
+      "loss": 0.3136,
+      "step": 138
+    },
+    {
+      "epoch": 0.08896,
+      "grad_norm": 0.4441445163984143,
+      "learning_rate": 4.9804985884885366e-05,
+      "loss": 0.2911,
+      "step": 139
+    },
+    {
+      "epoch": 0.0896,
+      "grad_norm": 0.5818680485265652,
+      "learning_rate": 4.979832819192608e-05,
+      "loss": 0.3178,
+      "step": 140
+    },
+    {
+      "epoch": 0.09024,
+      "grad_norm": 0.5596729609258517,
+      "learning_rate": 4.97915592131842e-05,
+      "loss": 0.2946,
+      "step": 141
+    },
+    {
+      "epoch": 0.09088,
+      "grad_norm": 0.4173115723963595,
+      "learning_rate": 4.97846789790364e-05,
+      "loss": 0.312,
+      "step": 142
+    },
+    {
+      "epoch": 0.09152,
+      "grad_norm": 0.5797089756431767,
+      "learning_rate": 4.977768752035865e-05,
+      "loss": 0.317,
+      "step": 143
+    },
+    {
+      "epoch": 0.09216,
+      "grad_norm": 0.599992651207998,
+      "learning_rate": 4.9770584868526046e-05,
+      "loss": 0.3162,
+      "step": 144
+    },
+    {
+      "epoch": 0.0928,
+      "grad_norm": 0.5107814027610112,
+      "learning_rate": 4.9763371055412685e-05,
+      "loss": 0.313,
+      "step": 145
+    },
+    {
+      "epoch": 0.09344,
+      "grad_norm": 0.44298297975068884,
+      "learning_rate": 4.9756046113391496e-05,
+      "loss": 0.3351,
+      "step": 146
+    },
+    {
+      "epoch": 0.09408,
+      "grad_norm": 0.5327895632430579,
+      "learning_rate": 4.974861007533413e-05,
+      "loss": 0.2917,
+      "step": 147
+    },
+    {
+      "epoch": 0.09472,
+      "grad_norm": 0.43312248711811524,
+      "learning_rate": 4.97410629746108e-05,
+      "loss": 0.2964,
+      "step": 148
+    },
+    {
+      "epoch": 0.09536,
+      "grad_norm": 0.5304322484911422,
+      "learning_rate": 4.9733404845090116e-05,
+      "loss": 0.2898,
+      "step": 149
+    },
+    {
+      "epoch": 0.096,
+      "grad_norm": 0.3997434435275148,
+      "learning_rate": 4.972563572113894e-05,
+      "loss": 0.2941,
+      "step": 150
+    },
+    {
+      "epoch": 0.09664,
+      "grad_norm": 0.6731310579298981,
+      "learning_rate": 4.971775563762226e-05,
+      "loss": 0.2979,
+      "step": 151
+    },
+    {
+      "epoch": 0.09728,
+      "grad_norm": 0.44982760512871145,
+      "learning_rate": 4.970976462990299e-05,
+      "loss": 0.3181,
+      "step": 152
+    },
+    {
+      "epoch": 0.09792,
+      "grad_norm": 0.582429222704454,
+      "learning_rate": 4.970166273384183e-05,
+      "loss": 0.2944,
+      "step": 153
+    },
+    {
+      "epoch": 0.09856,
+      "grad_norm": 0.37285441290932575,
+      "learning_rate": 4.969344998579712e-05,
+      "loss": 0.2708,
+      "step": 154
+    },
+    {
+      "epoch": 0.0992,
+      "grad_norm": 0.5242244051320633,
+      "learning_rate": 4.9685126422624644e-05,
+      "loss": 0.322,
+      "step": 155
+    },
+    {
+      "epoch": 0.09984,
+      "grad_norm": 0.42399899648411177,
+      "learning_rate": 4.96766920816775e-05,
+      "loss": 0.3194,
+      "step": 156
+    },
+    {
+      "epoch": 0.10048,
+      "grad_norm": 0.4451557052439993,
+      "learning_rate": 4.966814700080592e-05,
+      "loss": 0.3242,
+      "step": 157
+    },
+    {
+      "epoch": 0.10112,
+      "grad_norm": 0.48601797793339585,
+      "learning_rate": 4.965949121835707e-05,
+      "loss": 0.3397,
+      "step": 158
+    },
+    {
+      "epoch": 0.10176,
+      "grad_norm": 0.44145425649965603,
+      "learning_rate": 4.965072477317493e-05,
+      "loss": 0.3146,
+      "step": 159
+    },
+    {
+      "epoch": 0.1024,
+      "grad_norm": 0.4370780933539978,
+      "learning_rate": 4.964184770460007e-05,
+      "loss": 0.2978,
+      "step": 160
+    },
+    {
+      "epoch": 0.10304,
+      "grad_norm": 0.48836377923781304,
+      "learning_rate": 4.963286005246953e-05,
+      "loss": 0.2884,
+      "step": 161
+    },
+    {
+      "epoch": 0.10368,
+      "grad_norm": 0.41359620894086957,
+      "learning_rate": 4.9623761857116594e-05,
+      "loss": 0.3253,
+      "step": 162
+    },
+    {
+      "epoch": 0.10432,
+      "grad_norm": 0.44338475984615433,
+      "learning_rate": 4.961455315937059e-05,
+      "loss": 0.2871,
+      "step": 163
+    },
+    {
+      "epoch": 0.10496,
+      "grad_norm": 0.5079738340888645,
+      "learning_rate": 4.960523400055679e-05,
+      "loss": 0.3002,
+      "step": 164
+    },
+    {
+      "epoch": 0.1056,
+      "grad_norm": 0.427097830248176,
+      "learning_rate": 4.959580442249615e-05,
+      "loss": 0.3068,
+      "step": 165
+    },
+    {
+      "epoch": 0.10624,
+      "grad_norm": 0.3684291109883667,
+      "learning_rate": 4.958626446750515e-05,
+      "loss": 0.3256,
+      "step": 166
+    },
+    {
+      "epoch": 0.10688,
+      "grad_norm": 0.4683572393673182,
+      "learning_rate": 4.9576614178395595e-05,
+      "loss": 0.3054,
+      "step": 167
+    },
+    {
+      "epoch": 0.10752,
+      "grad_norm": 0.38807804876427243,
+      "learning_rate": 4.956685359847445e-05,
+      "loss": 0.3018,
+      "step": 168
+    },
+    {
+      "epoch": 0.10816,
+      "grad_norm": 0.387073605411047,
+      "learning_rate": 4.95569827715436e-05,
+      "loss": 0.2869,
+      "step": 169
+    },
+    {
+      "epoch": 0.1088,
+      "grad_norm": 0.4547710232799118,
+      "learning_rate": 4.9547001741899693e-05,
+      "loss": 0.3036,
+      "step": 170
+    },
+    {
+      "epoch": 0.10944,
+      "grad_norm": 0.39381908155851314,
+      "learning_rate": 4.953691055433391e-05,
+      "loss": 0.2966,
+      "step": 171
+    },
+    {
+      "epoch": 0.11008,
+      "grad_norm": 0.48725912333872035,
+      "learning_rate": 4.952670925413182e-05,
+      "loss": 0.3093,
+      "step": 172
+    },
+    {
+      "epoch": 0.11072,
+      "grad_norm": 0.3989019148895901,
+      "learning_rate": 4.951639788707307e-05,
+      "loss": 0.289,
+      "step": 173
+    },
+    {
+      "epoch": 0.11136,
+      "grad_norm": 0.3836802824526995,
+      "learning_rate": 4.950597649943132e-05,
+      "loss": 0.284,
+      "step": 174
+    },
+    {
+      "epoch": 0.112,
+      "grad_norm": 0.43375030098630124,
+      "learning_rate": 4.9495445137973905e-05,
+      "loss": 0.2993,
+      "step": 175
+    },
+    {
+      "epoch": 0.11264,
+      "grad_norm": 0.38353724590203925,
+      "learning_rate": 4.948480384996172e-05,
+      "loss": 0.2928,
+      "step": 176
+    },
+    {
+      "epoch": 0.11328,
+      "grad_norm": 0.41856497407127785,
+      "learning_rate": 4.947405268314894e-05,
+      "loss": 0.3058,
+      "step": 177
+    },
+    {
+      "epoch": 0.11392,
+      "grad_norm": 1.22965703788482,
+      "learning_rate": 4.9463191685782865e-05,
+      "loss": 0.3135,
+      "step": 178
+    },
+    {
+      "epoch": 0.11456,
+      "grad_norm": 0.3607079092136828,
+      "learning_rate": 4.945222090660364e-05,
+      "loss": 0.2911,
+      "step": 179
+    },
+    {
+      "epoch": 0.1152,
+      "grad_norm": 0.42652787065228764,
+      "learning_rate": 4.944114039484411e-05,
+      "loss": 0.3105,
+      "step": 180
+    },
+    {
+      "epoch": 0.11584,
+      "grad_norm": 0.39223235556161457,
+      "learning_rate": 4.942995020022954e-05,
+      "loss": 0.3308,
+      "step": 181
+    },
+    {
+      "epoch": 0.11648,
+      "grad_norm": 0.4434271422328715,
+      "learning_rate": 4.9418650372977384e-05,
+      "loss": 0.2946,
+      "step": 182
+    },
+    {
+      "epoch": 0.11712,
+      "grad_norm": 0.3871233188728082,
+      "learning_rate": 4.940724096379714e-05,
+      "loss": 0.2908,
+      "step": 183
+    },
+    {
+      "epoch": 0.11776,
+      "grad_norm": 0.3504967706065276,
+      "learning_rate": 4.9395722023890035e-05,
+      "loss": 0.2869,
+      "step": 184
+    },
+    {
+      "epoch": 0.1184,
+      "grad_norm": 0.3800847472179798,
+      "learning_rate": 4.938409360494883e-05,
+      "loss": 0.2916,
+      "step": 185
+    },
+    {
+      "epoch": 0.11904,
+      "grad_norm": 0.35973303790543976,
+      "learning_rate": 4.937235575915761e-05,
+      "loss": 0.2994,
+      "step": 186
+    },
+    {
+      "epoch": 0.11968,
+      "grad_norm": 0.39040484393002967,
+      "learning_rate": 4.93605085391915e-05,
+      "loss": 0.2994,
+      "step": 187
+    },
+    {
+      "epoch": 0.12032,
+      "grad_norm": 0.45370992844530267,
+      "learning_rate": 4.934855199821647e-05,
+      "loss": 0.2888,
+      "step": 188
+    },
+    {
+      "epoch": 0.12096,
+      "grad_norm": 0.41032353843291086,
+      "learning_rate": 4.93364861898891e-05,
+      "loss": 0.2845,
+      "step": 189
+    },
+    {
+      "epoch": 0.1216,
+      "grad_norm": 0.4036630852460455,
+      "learning_rate": 4.932431116835628e-05,
+      "loss": 0.3052,
+      "step": 190
+    },
+    {
+      "epoch": 0.12224,
+      "grad_norm": 0.3484768322428675,
+      "learning_rate": 4.931202698825503e-05,
+      "loss": 0.2896,
+      "step": 191
+    },
+    {
+      "epoch": 0.12288,
+      "grad_norm": 0.34569927077349094,
+      "learning_rate": 4.929963370471226e-05,
+      "loss": 0.2953,
+      "step": 192
+    },
+    {
+      "epoch": 0.12352,
+      "grad_norm": 0.3321813143876421,
+      "learning_rate": 4.928713137334445e-05,
+      "loss": 0.2891,
+      "step": 193
+    },
+    {
+      "epoch": 0.12416,
+      "grad_norm": 0.3696723739246585,
+      "learning_rate": 4.927452005025748e-05,
+      "loss": 0.3091,
+      "step": 194
+    },
+    {
+      "epoch": 0.1248,
+      "grad_norm": 0.3525045846275956,
+      "learning_rate": 4.926179979204633e-05,
+      "loss": 0.3089,
+      "step": 195
+    },
+    {
+      "epoch": 0.12544,
+      "grad_norm": 0.4050534156570688,
+      "learning_rate": 4.9248970655794836e-05,
+      "loss": 0.2991,
+      "step": 196
+    },
+    {
+      "epoch": 0.12608,
+      "grad_norm": 0.37675234071249,
+      "learning_rate": 4.923603269907545e-05,
+      "loss": 0.2798,
+      "step": 197
+    },
+    {
+      "epoch": 0.12672,
+      "grad_norm": 0.34914495162279696,
+      "learning_rate": 4.922298597994897e-05,
+      "loss": 0.289,
+      "step": 198
+    },
+    {
+      "epoch": 0.12736,
+      "grad_norm": 0.3865497720329682,
+      "learning_rate": 4.920983055696428e-05,
+      "loss": 0.3002,
+      "step": 199
+    },
+    {
+      "epoch": 0.128,
+      "grad_norm": 0.446766451366359,
+      "learning_rate": 4.919656648915807e-05,
+      "loss": 0.2918,
+      "step": 200
+    },
+    {
+      "epoch": 0.12864,
+      "grad_norm": 0.473700352021563,
+      "learning_rate": 4.918319383605462e-05,
+      "loss": 0.2639,
+      "step": 201
+    },
+    {
+      "epoch": 0.12928,
+      "grad_norm": 0.4098673261200793,
+      "learning_rate": 4.916971265766547e-05,
+      "loss": 0.3025,
+      "step": 202
+    },
+    {
+      "epoch": 0.12992,
+      "grad_norm": 0.47472542511654414,
+      "learning_rate": 4.915612301448919e-05,
+      "loss": 0.3039,
+      "step": 203
+    },
+    {
+      "epoch": 0.13056,
+      "grad_norm": 0.4474522945723034,
+      "learning_rate": 4.914242496751111e-05,
+      "loss": 0.2962,
+      "step": 204
+    },
+    {
+      "epoch": 0.1312,
+      "grad_norm": 0.5376434925769887,
+      "learning_rate": 4.912861857820303e-05,
+      "loss": 0.3125,
+      "step": 205
+    },
+    {
+      "epoch": 0.13184,
+      "grad_norm": 0.415507945261084,
+      "learning_rate": 4.911470390852295e-05,
+      "loss": 0.3086,
+      "step": 206
+    },
+    {
+      "epoch": 0.13248,
+      "grad_norm": 0.4284511552336465,
+      "learning_rate": 4.91006810209148e-05,
+      "loss": 0.2952,
+      "step": 207
+    },
+    {
+      "epoch": 0.13312,
+      "grad_norm": 0.40052269427223414,
+      "learning_rate": 4.9086549978308125e-05,
+      "loss": 0.3062,
+      "step": 208
+    },
+    {
+      "epoch": 0.13376,
+      "grad_norm": 0.43743782442362256,
+      "learning_rate": 4.9072310844117875e-05,
+      "loss": 0.2794,
+      "step": 209
+    },
+    {
+      "epoch": 0.1344,
+      "grad_norm": 0.4420765691985995,
+      "learning_rate": 4.9057963682244045e-05,
+      "loss": 0.298,
+      "step": 210
+    },
+    {
+      "epoch": 0.13504,
+      "grad_norm": 0.41724801069829737,
+      "learning_rate": 4.904350855707142e-05,
+      "loss": 0.2936,
+      "step": 211
+    },
+    {
+      "epoch": 0.13568,
+      "grad_norm": 0.3775542589408398,
+      "learning_rate": 4.9028945533469276e-05,
+      "loss": 0.2714,
+      "step": 212
+    },
+    {
+      "epoch": 0.13632,
+      "grad_norm": 0.4493944075287162,
+      "learning_rate": 4.9014274676791134e-05,
+      "loss": 0.3094,
+      "step": 213
+    },
+    {
+      "epoch": 0.13696,
+      "grad_norm": 0.4671949905514448,
+      "learning_rate": 4.899949605287439e-05,
+      "loss": 0.2883,
+      "step": 214
+    },
+    {
+      "epoch": 0.1376,
+      "grad_norm": 0.39187019356209574,
+      "learning_rate": 4.898460972804009e-05,
+      "loss": 0.3155,
+      "step": 215
+    },
+    {
+      "epoch": 0.13824,
+      "grad_norm": 0.3832421971309516,
+      "learning_rate": 4.896961576909258e-05,
+      "loss": 0.2891,
+      "step": 216
+    },
+    {
+      "epoch": 0.13888,
+      "grad_norm": 0.4701725265012741,
+      "learning_rate": 4.895451424331924e-05,
+      "loss": 0.2791,
+      "step": 217
+    },
+    {
+      "epoch": 0.13952,
+      "grad_norm": 0.3906675874850412,
+      "learning_rate": 4.893930521849016e-05,
+      "loss": 0.2739,
+      "step": 218
+    },
+    {
+      "epoch": 0.14016,
+      "grad_norm": 0.5048426982814626,
+      "learning_rate": 4.892398876285785e-05,
+      "loss": 0.2956,
+      "step": 219
+    },
+    {
+      "epoch": 0.1408,
+      "grad_norm": 0.5550796911808695,
+      "learning_rate": 4.890856494515696e-05,
+      "loss": 0.2942,
+      "step": 220
+    },
+    {
+      "epoch": 0.14144,
+      "grad_norm": 0.4446478533142643,
+      "learning_rate": 4.889303383460389e-05,
+      "loss": 0.2876,
+      "step": 221
+    },
+    {
+      "epoch": 0.14208,
+      "grad_norm": 0.5139145715219713,
+      "learning_rate": 4.887739550089656e-05,
+      "loss": 0.2871,
+      "step": 222
+    },
+    {
+      "epoch": 0.14272,
+      "grad_norm": 0.433835044119607,
+      "learning_rate": 4.886165001421408e-05,
+      "loss": 0.3174,
+      "step": 223
+    },
+    {
+      "epoch": 0.14336,
+      "grad_norm": 0.4118433589806149,
+      "learning_rate": 4.884579744521638e-05,
+      "loss": 0.2915,
+      "step": 224
+    },
+    {
+      "epoch": 0.144,
+      "grad_norm": 0.36660272368355734,
+      "learning_rate": 4.8829837865043984e-05,
+      "loss": 0.2984,
+      "step": 225
+    },
+    {
+      "epoch": 0.14464,
+      "grad_norm": 0.37584693808307384,
+      "learning_rate": 4.881377134531762e-05,
+      "loss": 0.2738,
+      "step": 226
+    },
+    {
+      "epoch": 0.14528,
+      "grad_norm": 0.39658204298189026,
+      "learning_rate": 4.8797597958137914e-05,
+      "loss": 0.2868,
+      "step": 227
+    },
+    {
+      "epoch": 0.14592,
+      "grad_norm": 0.42820275509861444,
+      "learning_rate": 4.878131777608509e-05,
+      "loss": 0.2921,
+      "step": 228
+    },
+    {
+      "epoch": 0.14656,
+      "grad_norm": 0.4162784808158851,
+      "learning_rate": 4.876493087221862e-05,
+      "loss": 0.3034,
+      "step": 229
+    },
+    {
+      "epoch": 0.1472,
+      "grad_norm": 0.4050308180549546,
+      "learning_rate": 4.87484373200769e-05,
+      "loss": 0.2964,
+      "step": 230
+    },
+    {
+      "epoch": 0.14784,
+      "grad_norm": 0.396192734792577,
+      "learning_rate": 4.8731837193676944e-05,
+      "loss": 0.2879,
+      "step": 231
+    },
+    {
+      "epoch": 0.14848,
+      "grad_norm": 0.4022225227959102,
+      "learning_rate": 4.871513056751401e-05,
+      "loss": 0.3058,
+      "step": 232
+    },
+    {
+      "epoch": 0.14912,
+      "grad_norm": 0.4712047740803004,
+      "learning_rate": 4.8698317516561295e-05,
+      "loss": 0.307,
+      "step": 233
+    },
+    {
+      "epoch": 0.14976,
+      "grad_norm": 0.5020568589020615,
+      "learning_rate": 4.8681398116269596e-05,
+      "loss": 0.2937,
+      "step": 234
+    },
+    {
+      "epoch": 0.1504,
+      "grad_norm": 0.5384324202189518,
+      "learning_rate": 4.8664372442566954e-05,
+      "loss": 0.3024,
+      "step": 235
+    },
+    {
+      "epoch": 0.15104,
+      "grad_norm": 0.39516572600823124,
+      "learning_rate": 4.864724057185834e-05,
+      "loss": 0.2743,
+      "step": 236
+    },
+    {
+      "epoch": 0.15168,
+      "grad_norm": 0.4548277706908328,
+      "learning_rate": 4.86300025810253e-05,
+      "loss": 0.2877,
+      "step": 237
+    },
+    {
+      "epoch": 0.15232,
+      "grad_norm": 0.3720089945799359,
+      "learning_rate": 4.861265854742559e-05,
+      "loss": 0.2819,
+      "step": 238
+    },
+    {
+      "epoch": 0.15296,
+      "grad_norm": 0.47047832876789597,
+      "learning_rate": 4.859520854889287e-05,
+      "loss": 0.3021,
+      "step": 239
+    },
+    {
+      "epoch": 0.1536,
+      "grad_norm": 0.4554923160199729,
+      "learning_rate": 4.857765266373631e-05,
+      "loss": 0.3033,
+      "step": 240
+    },
+    {
+      "epoch": 0.15424,
+      "grad_norm": 0.5479982370735748,
+      "learning_rate": 4.855999097074027e-05,
+      "loss": 0.3016,
+      "step": 241
+    },
+    {
+      "epoch": 0.15488,
+      "grad_norm": 0.38490547613491893,
+      "learning_rate": 4.8542223549163946e-05,
+      "loss": 0.2791,
+      "step": 242
+    },
+    {
+      "epoch": 0.15552,
+      "grad_norm": 0.5010326131588188,
+      "learning_rate": 4.8524350478740985e-05,
+      "loss": 0.2758,
+      "step": 243
+    },
+    {
+      "epoch": 0.15616,
+      "grad_norm": 0.408628306302986,
+      "learning_rate": 4.850637183967917e-05,
+      "loss": 0.2914,
+      "step": 244
+    },
+    {
+      "epoch": 0.1568,
+      "grad_norm": 0.4342781570295902,
+      "learning_rate": 4.848828771266002e-05,
+      "loss": 0.2977,
+      "step": 245
+    },
+    {
+      "epoch": 0.15744,
+      "grad_norm": 0.43152902680446176,
+      "learning_rate": 4.847009817883844e-05,
+      "loss": 0.2909,
+      "step": 246
+    },
+    {
+      "epoch": 0.15808,
+      "grad_norm": 0.3551063718621289,
+      "learning_rate": 4.845180331984239e-05,
+      "loss": 0.2788,
+      "step": 247
+    },
+    {
+      "epoch": 0.15872,
+      "grad_norm": 0.4023732387829879,
+      "learning_rate": 4.843340321777246e-05,
+      "loss": 0.2727,
+      "step": 248
+    },
+    {
+      "epoch": 0.15936,
+      "grad_norm": 0.3631153670647941,
+      "learning_rate": 4.8414897955201556e-05,
+      "loss": 0.2674,
+      "step": 249
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.39442642636399,
+      "learning_rate": 4.8396287615174484e-05,
+      "loss": 0.2653,
+      "step": 250
+    },
+    {
+      "epoch": 0.16064,
+      "grad_norm": 0.3999569444426015,
+      "learning_rate": 4.837757228120762e-05,
+      "loss": 0.297,
+      "step": 251
+    },
+    {
+      "epoch": 0.16128,
+      "grad_norm": 0.3348408050068023,
+      "learning_rate": 4.835875203728849e-05,
+      "loss": 0.2608,
+      "step": 252
+    },
+    {
+      "epoch": 0.16192,
+      "grad_norm": 0.4012425485757949,
+      "learning_rate": 4.8339826967875443e-05,
+      "loss": 0.2966,
+      "step": 253
+    },
+    {
+      "epoch": 0.16256,
+      "grad_norm": 0.39410881558128635,
+      "learning_rate": 4.832079715789724e-05,
+      "loss": 0.3064,
+      "step": 254
+    },
+    {
+      "epoch": 0.1632,
+      "grad_norm": 0.43770904619065176,
+      "learning_rate": 4.830166269275267e-05,
+      "loss": 0.2988,
+      "step": 255
+    },
+    {
+      "epoch": 0.16384,
+      "grad_norm": 0.34230150447086477,
+      "learning_rate": 4.828242365831016e-05,
+      "loss": 0.283,
+      "step": 256
+    },
+    {
+      "epoch": 0.16448,
+      "grad_norm": 0.4037086139344497,
+      "learning_rate": 4.8263080140907455e-05,
+      "loss": 0.2982,
+      "step": 257
+    },
+    {
+      "epoch": 0.16512,
+      "grad_norm": 0.44544678937787124,
+      "learning_rate": 4.824363222735114e-05,
+      "loss": 0.2821,
+      "step": 258
+    },
+    {
+      "epoch": 0.16576,
+      "grad_norm": 0.4065131032313394,
+      "learning_rate": 4.82240800049163e-05,
+      "loss": 0.2855,
+      "step": 259
+    },
+    {
+      "epoch": 0.1664,
+      "grad_norm": 0.5004502107420943,
+      "learning_rate": 4.820442356134612e-05,
+      "loss": 0.2787,
+      "step": 260
+    },
+    {
+      "epoch": 0.16704,
+      "grad_norm": 0.4602519114887701,
+      "learning_rate": 4.8184662984851505e-05,
+      "loss": 0.3095,
+      "step": 261
+    },
+    {
+      "epoch": 0.16768,
+      "grad_norm": 0.39727473565215526,
+      "learning_rate": 4.816479836411065e-05,
+      "loss": 0.289,
+      "step": 262
+    },
+    {
+      "epoch": 0.16832,
+      "grad_norm": 0.4929832224816534,
+      "learning_rate": 4.8144829788268674e-05,
+      "loss": 0.3009,
+      "step": 263
+    },
+    {
+      "epoch": 0.16896,
+      "grad_norm": 0.4236129546828753,
+      "learning_rate": 4.8124757346937214e-05,
+      "loss": 0.2686,
+      "step": 264
+    },
+    {
+      "epoch": 0.1696,
+      "grad_norm": 0.41378802352106486,
+      "learning_rate": 4.810458113019401e-05,
+      "loss": 0.2809,
+      "step": 265
+    },
+    {
+      "epoch": 0.17024,
+      "grad_norm": 0.37964540828378823,
+      "learning_rate": 4.8084301228582485e-05,
+      "loss": 0.2647,
+      "step": 266
+    },
+    {
+      "epoch": 0.17088,
+      "grad_norm": 0.4325588860436682,
+      "learning_rate": 4.8063917733111406e-05,
+      "loss": 0.3186,
+      "step": 267
+    },
+    {
+      "epoch": 0.17152,
+      "grad_norm": 0.388936215368637,
+      "learning_rate": 4.8043430735254404e-05,
+      "loss": 0.2893,
+      "step": 268
+    },
+    {
+      "epoch": 0.17216,
+      "grad_norm": 0.44527024310059876,
+      "learning_rate": 4.8022840326949595e-05,
+      "loss": 0.284,
+      "step": 269
+    },
+    {
+      "epoch": 0.1728,
+      "grad_norm": 0.43784541980194747,
+      "learning_rate": 4.800214660059916e-05,
+      "loss": 0.2871,
+      "step": 270
+    },
+    {
+      "epoch": 0.17344,
+      "grad_norm": 0.4555897041667051,
+      "learning_rate": 4.798134964906894e-05,
+      "loss": 0.2953,
+      "step": 271
+    },
+    {
+      "epoch": 0.17408,
+      "grad_norm": 0.43464658407728873,
+      "learning_rate": 4.7960449565688e-05,
+      "loss": 0.27,
+      "step": 272
+    },
+    {
+      "epoch": 0.17472,
+      "grad_norm": 0.4477882720096214,
+      "learning_rate": 4.793944644424826e-05,
+      "loss": 0.2933,
+      "step": 273
+    },
+    {
+      "epoch": 0.17536,
+      "grad_norm": 0.4065498956535354,
+      "learning_rate": 4.7918340379003985e-05,
+      "loss": 0.2937,
+      "step": 274
+    },
+    {
+      "epoch": 0.176,
+      "grad_norm": 0.4389073843597932,
+      "learning_rate": 4.7897131464671444e-05,
+      "loss": 0.3085,
+      "step": 275
+    },
+    {
+      "epoch": 0.17664,
+      "grad_norm": 0.39967523914970293,
+      "learning_rate": 4.787581979642845e-05,
+      "loss": 0.2629,
+      "step": 276
+    },
+    {
+      "epoch": 0.17728,
+      "grad_norm": 0.4175663492402688,
+      "learning_rate": 4.7854405469913944e-05,
+      "loss": 0.2816,
+      "step": 277
+    },
+    {
+      "epoch": 0.17792,
+      "grad_norm": 0.3699316200983033,
+      "learning_rate": 4.7832888581227545e-05,
+      "loss": 0.2847,
+      "step": 278
+    },
+    {
+      "epoch": 0.17856,
+      "grad_norm": 0.41690698282522476,
+      "learning_rate": 4.7811269226929135e-05,
+      "loss": 0.2878,
+      "step": 279
+    },
+    {
+      "epoch": 0.1792,
+      "grad_norm": 0.44308504497056483,
+      "learning_rate": 4.7789547504038443e-05,
+      "loss": 0.2769,
+      "step": 280
+    },
+    {
+      "epoch": 0.17984,
+      "grad_norm": 0.37910073417357737,
+      "learning_rate": 4.776772351003458e-05,
+      "loss": 0.2885,
+      "step": 281
+    },
+    {
+      "epoch": 0.18048,
+      "grad_norm": 0.43228700146859433,
+      "learning_rate": 4.774579734285559e-05,
+      "loss": 0.2937,
+      "step": 282
+    },
+    {
+      "epoch": 0.18112,
+      "grad_norm": 0.36863085314861516,
+      "learning_rate": 4.772376910089808e-05,
+      "loss": 0.286,
+      "step": 283
+    },
+    {
+      "epoch": 0.18176,
+      "grad_norm": 0.36574183780742076,
+      "learning_rate": 4.770163888301668e-05,
+      "loss": 0.2791,
+      "step": 284
+    },
+    {
+      "epoch": 0.1824,
+      "grad_norm": 0.3528636724414157,
+      "learning_rate": 4.7679406788523685e-05,
+      "loss": 0.2852,
+      "step": 285
+    },
+    {
+      "epoch": 0.18304,
+      "grad_norm": 0.3544995299967653,
+      "learning_rate": 4.765707291718856e-05,
+      "loss": 0.2817,
+      "step": 286
+    },
+    {
+      "epoch": 0.18368,
+      "grad_norm": 0.38798119556391136,
+      "learning_rate": 4.7634637369237526e-05,
+      "loss": 0.2846,
+      "step": 287
+    },
+    {
+      "epoch": 0.18432,
+      "grad_norm": 0.4799276501329986,
+      "learning_rate": 4.7612100245353045e-05,
+      "loss": 0.2949,
+      "step": 288
+    },
+    {
+      "epoch": 0.18496,
+      "grad_norm": 0.37932495954229145,
+      "learning_rate": 4.758946164667347e-05,
+      "loss": 0.2995,
+      "step": 289
+    },
+    {
+      "epoch": 0.1856,
+      "grad_norm": 0.42382042791552227,
+      "learning_rate": 4.7566721674792515e-05,
+      "loss": 0.2858,
+      "step": 290
+    },
+    {
+      "epoch": 0.18624,
+      "grad_norm": 0.33676351501040813,
+      "learning_rate": 4.7543880431758805e-05,
+      "loss": 0.2851,
+      "step": 291
+    },
+    {
+      "epoch": 0.18688,
+      "grad_norm": 0.44888140015841044,
+      "learning_rate": 4.752093802007543e-05,
+      "loss": 0.3109,
+      "step": 292
+    },
+    {
+      "epoch": 0.18752,
+      "grad_norm": 0.4251956966873805,
+      "learning_rate": 4.7497894542699534e-05,
+      "loss": 0.2686,
+      "step": 293
+    },
+    {
+      "epoch": 0.18816,
+      "grad_norm": 0.3403024579052465,
+      "learning_rate": 4.747475010304175e-05,
+      "loss": 0.288,
+      "step": 294
+    },
+    {
+      "epoch": 0.1888,
+      "grad_norm": 0.42640118708732144,
+      "learning_rate": 4.745150480496583e-05,
+      "loss": 0.3064,
+      "step": 295
+    },
+    {
+      "epoch": 0.18944,
+      "grad_norm": 0.3633343500965691,
+      "learning_rate": 4.742815875278811e-05,
+      "loss": 0.3049,
+      "step": 296
+    },
+    {
+      "epoch": 0.19008,
+      "grad_norm": 0.4231871364046764,
+      "learning_rate": 4.7404712051277115e-05,
+      "loss": 0.264,
+      "step": 297
+    },
+    {
+      "epoch": 0.19072,
+      "grad_norm": 0.3888065573821839,
+      "learning_rate": 4.738116480565301e-05,
+      "loss": 0.304,
+      "step": 298
+    },
+    {
+      "epoch": 0.19136,
+      "grad_norm": 0.43514328391884294,
+      "learning_rate": 4.7357517121587183e-05,
+      "loss": 0.2875,
+      "step": 299
+    },
+    {
+      "epoch": 0.192,
+      "grad_norm": 0.3408410266602059,
+      "learning_rate": 4.7333769105201735e-05,
+      "loss": 0.2501,
+      "step": 300
+    },
+    {
+      "epoch": 0.19264,
+      "grad_norm": 0.39894971801096046,
+      "learning_rate": 4.730992086306906e-05,
+      "loss": 0.2909,
+      "step": 301
+    },
+    {
+      "epoch": 0.19328,
+      "grad_norm": 0.38556731026579016,
+      "learning_rate": 4.728597250221129e-05,
+      "loss": 0.2915,
+      "step": 302
+    },
+    {
+      "epoch": 0.19392,
+      "grad_norm": 0.37408950553128606,
+      "learning_rate": 4.726192413009986e-05,
+      "loss": 0.3029,
+      "step": 303
+    },
+    {
+      "epoch": 0.19456,
+      "grad_norm": 0.39849511897928547,
+      "learning_rate": 4.723777585465503e-05,
+      "loss": 0.3017,
+      "step": 304
+    },
+    {
+      "epoch": 0.1952,
+      "grad_norm": 0.5587028139056558,
+      "learning_rate": 4.7213527784245395e-05,
+      "loss": 0.2842,
+      "step": 305
+    },
+    {
+      "epoch": 0.19584,
+      "grad_norm": 0.3364288910892129,
+      "learning_rate": 4.718918002768738e-05,
+      "loss": 0.2854,
+      "step": 306
+    },
+    {
+      "epoch": 0.19648,
+      "grad_norm": 0.38411867689380863,
+      "learning_rate": 4.716473269424476e-05,
+      "loss": 0.2914,
+      "step": 307
+    },
+    {
+      "epoch": 0.19712,
+      "grad_norm": 0.4223139892696591,
+      "learning_rate": 4.714018589362818e-05,
+      "loss": 0.321,
+      "step": 308
+    },
+    {
+      "epoch": 0.19776,
+      "grad_norm": 0.30488498967738265,
+      "learning_rate": 4.7115539735994666e-05,
+      "loss": 0.2809,
+      "step": 309
+    },
+    {
+      "epoch": 0.1984,
+      "grad_norm": 0.37213754218925743,
+      "learning_rate": 4.709079433194712e-05,
+      "loss": 0.2844,
+      "step": 310
+    },
+    {
+      "epoch": 0.19904,
+      "grad_norm": 0.3324865603595645,
+      "learning_rate": 4.7065949792533804e-05,
+      "loss": 0.2652,
+      "step": 311
+    },
+    {
+      "epoch": 0.19968,
+      "grad_norm": 0.4270950226690428,
+      "learning_rate": 4.70410062292479e-05,
+      "loss": 0.3034,
+      "step": 312
+    },
+    {
+      "epoch": 0.20032,
+      "grad_norm": 0.3325840406105098,
+      "learning_rate": 4.701596375402694e-05,
+      "loss": 0.3022,
+      "step": 313
+    },
+    {
+      "epoch": 0.20096,
+      "grad_norm": 0.39532309924298714,
+      "learning_rate": 4.699082247925234e-05,
+      "loss": 0.3098,
+      "step": 314
+    },
+    {
+      "epoch": 0.2016,
+      "grad_norm": 0.37138707474717225,
+      "learning_rate": 4.6965582517748923e-05,
+      "loss": 0.303,
+      "step": 315
+    },
+    {
+      "epoch": 0.20224,
+      "grad_norm": 0.37803236553918484,
+      "learning_rate": 4.694024398278434e-05,
+      "loss": 0.2987,
+      "step": 316
+    },
+    {
+      "epoch": 0.20288,
+      "grad_norm": 0.4138163110799673,
+      "learning_rate": 4.691480698806864e-05,
+      "loss": 0.3088,
+      "step": 317
+    },
+    {
+      "epoch": 0.20352,
+      "grad_norm": 0.4020730510714261,
+      "learning_rate": 4.6889271647753694e-05,
+      "loss": 0.2716,
+      "step": 318
+    },
+    {
+      "epoch": 0.20416,
+      "grad_norm": 0.4258264220033009,
+      "learning_rate": 4.686363807643272e-05,
+      "loss": 0.2631,
+      "step": 319
+    },
+    {
+      "epoch": 0.2048,
+      "grad_norm": 0.390891426601791,
+      "learning_rate": 4.683790638913979e-05,
+      "loss": 0.3111,
+      "step": 320
+    },
+    {
+      "epoch": 0.20544,
+      "grad_norm": 0.3778128495869106,
+      "learning_rate": 4.681207670134923e-05,
+      "loss": 0.2946,
+      "step": 321
+    },
+    {
+      "epoch": 0.20608,
+      "grad_norm": 0.441422060674416,
+      "learning_rate": 4.6786149128975196e-05,
+      "loss": 0.2861,
+      "step": 322
+    },
+    {
+      "epoch": 0.20672,
+      "grad_norm": 0.40894682758274087,
+      "learning_rate": 4.6760123788371114e-05,
+      "loss": 0.2911,
+      "step": 323
+    },
+    {
+      "epoch": 0.20736,
+      "grad_norm": 0.4107398698273085,
+      "learning_rate": 4.6734000796329125e-05,
+      "loss": 0.293,
+      "step": 324
+    },
+    {
+      "epoch": 0.208,
+      "grad_norm": 0.37097238778323643,
+      "learning_rate": 4.670778027007964e-05,
+      "loss": 0.3073,
+      "step": 325
+    },
+    {
+      "epoch": 0.20864,
+      "grad_norm": 0.45855338261074946,
+      "learning_rate": 4.6681462327290725e-05,
+      "loss": 0.2768,
+      "step": 326
+    },
+    {
+      "epoch": 0.20928,
+      "grad_norm": 0.39179575642499337,
+      "learning_rate": 4.6655047086067636e-05,
+      "loss": 0.2872,
+      "step": 327
+    },
+    {
+      "epoch": 0.20992,
+      "grad_norm": 0.42973906362772835,
+      "learning_rate": 4.662853466495229e-05,
+      "loss": 0.2812,
+      "step": 328
+    },
+    {
+      "epoch": 0.21056,
+      "grad_norm": 0.3311875194641167,
+      "learning_rate": 4.660192518292266e-05,
+      "loss": 0.2787,
+      "step": 329
+    },
+    {
+      "epoch": 0.2112,
+      "grad_norm": 0.43519079330923655,
+      "learning_rate": 4.6575218759392336e-05,
+      "loss": 0.2791,
+      "step": 330
+    },
+    {
+      "epoch": 0.21184,
+      "grad_norm": 0.3880493441804471,
+      "learning_rate": 4.654841551420993e-05,
+      "loss": 0.2679,
+      "step": 331
+    },
+    {
+      "epoch": 0.21248,
+      "grad_norm": 0.40128212466880625,
+      "learning_rate": 4.6521515567658557e-05,
+      "loss": 0.2762,
+      "step": 332
+    },
+    {
+      "epoch": 0.21312,
+      "grad_norm": 0.405636539632707,
+      "learning_rate": 4.6494519040455275e-05,
+      "loss": 0.2871,
+      "step": 333
+    },
+    {
+      "epoch": 0.21376,
+      "grad_norm": 0.3681068098396899,
+      "learning_rate": 4.6467426053750596e-05,
+      "loss": 0.2723,
+      "step": 334
+    },
+    {
+      "epoch": 0.2144,
+      "grad_norm": 0.3677215015157118,
+      "learning_rate": 4.644023672912788e-05,
+      "loss": 0.2865,
+      "step": 335
+    },
+    {
+      "epoch": 0.21504,
+      "grad_norm": 0.3599476082852491,
+      "learning_rate": 4.641295118860281e-05,
+      "loss": 0.2958,
+      "step": 336
+    },
+    {
+      "epoch": 0.21568,
+      "grad_norm": 0.3420976798828047,
+      "learning_rate": 4.638556955462289e-05,
+      "loss": 0.268,
+      "step": 337
+    },
+    {
+      "epoch": 0.21632,
+      "grad_norm": 0.42797837252993226,
+      "learning_rate": 4.635809195006681e-05,
+      "loss": 0.3001,
+      "step": 338
+    },
+    {
+      "epoch": 0.21696,
+      "grad_norm": 0.3880188415117098,
+      "learning_rate": 4.633051849824396e-05,
+      "loss": 0.2658,
+      "step": 339
+    },
+    {
+      "epoch": 0.2176,
+      "grad_norm": 0.4408734151287498,
+      "learning_rate": 4.630284932289386e-05,
+      "loss": 0.2953,
+      "step": 340
+    },
+    {
+      "epoch": 0.21824,
+      "grad_norm": 0.4501384274921474,
+      "learning_rate": 4.6275084548185605e-05,
+      "loss": 0.3047,
+      "step": 341
+    },
+    {
+      "epoch": 0.21888,
+      "grad_norm": 0.40970680859700936,
+      "learning_rate": 4.6247224298717294e-05,
+      "loss": 0.2631,
+      "step": 342
+    },
+    {
+      "epoch": 0.21952,
+      "grad_norm": 0.3768928237473631,
+      "learning_rate": 4.621926869951549e-05,
+      "loss": 0.2532,
+      "step": 343
+    },
+    {
+      "epoch": 0.22016,
+      "grad_norm": 0.524132122914001,
+      "learning_rate": 4.619121787603464e-05,
+      "loss": 0.286,
+      "step": 344
+    },
+    {
+      "epoch": 0.2208,
+      "grad_norm": 0.37447678012305663,
+      "learning_rate": 4.6163071954156546e-05,
+      "loss": 0.2849,
+      "step": 345
+    },
+    {
+      "epoch": 0.22144,
+      "grad_norm": 0.4204861065550955,
+      "learning_rate": 4.6134831060189743e-05,
+      "loss": 0.2921,
+      "step": 346
+    },
+    {
+      "epoch": 0.22208,
+      "grad_norm": 0.39411381176913984,
+      "learning_rate": 4.610649532086899e-05,
+      "loss": 0.2816,
+      "step": 347
+    },
+    {
+      "epoch": 0.22272,
+      "grad_norm": 0.38236557437381496,
+      "learning_rate": 4.607806486335468e-05,
+      "loss": 0.2977,
+      "step": 348
+    },
+    {
+      "epoch": 0.22336,
+      "grad_norm": 0.3437602495312351,
+      "learning_rate": 4.604953981523225e-05,
+      "loss": 0.2973,
+      "step": 349
+    },
+    {
+      "epoch": 0.224,
+      "grad_norm": 0.44552885118211055,
+      "learning_rate": 4.602092030451162e-05,
+      "loss": 0.2792,
+      "step": 350
+    },
+    {
+      "epoch": 0.22464,
+      "grad_norm": 0.3920461628058572,
+      "learning_rate": 4.599220645962665e-05,
+      "loss": 0.2942,
+      "step": 351
+    },
+    {
+      "epoch": 0.22528,
+      "grad_norm": 0.3469818488664516,
+      "learning_rate": 4.596339840943452e-05,
+      "loss": 0.2778,
+      "step": 352
+    },
+    {
+      "epoch": 0.22592,
+      "grad_norm": 0.38662651851766466,
+      "learning_rate": 4.5934496283215164e-05,
+      "loss": 0.2772,
+      "step": 353
+    },
+    {
+      "epoch": 0.22656,
+      "grad_norm": 0.36398054897828075,
+      "learning_rate": 4.590550021067072e-05,
+      "loss": 0.2663,
+      "step": 354
+    },
+    {
+      "epoch": 0.2272,
+      "grad_norm": 0.3681746289020703,
+      "learning_rate": 4.587641032192488e-05,
+      "loss": 0.2746,
+      "step": 355
+    },
+    {
+      "epoch": 0.22784,
+      "grad_norm": 0.36588326542323685,
+      "learning_rate": 4.58472267475224e-05,
+      "loss": 0.2899,
+      "step": 356
+    },
+    {
+      "epoch": 0.22848,
+      "grad_norm": 0.41949504709512825,
+      "learning_rate": 4.5817949618428405e-05,
+      "loss": 0.2756,
+      "step": 357
+    },
+    {
+      "epoch": 0.22912,
+      "grad_norm": 0.3678906522219436,
+      "learning_rate": 4.578857906602791e-05,
+      "loss": 0.2767,
+      "step": 358
+    },
+    {
+      "epoch": 0.22976,
+      "grad_norm": 0.39849990301878413,
+      "learning_rate": 4.575911522212516e-05,
+      "loss": 0.2938,
+      "step": 359
+    },
+    {
+      "epoch": 0.2304,
+      "grad_norm": 0.3378349644319916,
+      "learning_rate": 4.572955821894305e-05,
+      "loss": 0.2794,
+      "step": 360
+    },
+    {
+      "epoch": 0.23104,
+      "grad_norm": 0.34264859841563705,
+      "learning_rate": 4.5699908189122556e-05,
+      "loss": 0.2433,
+      "step": 361
+    },
+    {
+      "epoch": 0.23168,
+      "grad_norm": 0.34023542845416116,
+      "learning_rate": 4.5670165265722096e-05,
+      "loss": 0.2641,
+      "step": 362
+    },
+    {
+      "epoch": 0.23232,
+      "grad_norm": 0.380184434585454,
+      "learning_rate": 4.5640329582217004e-05,
+      "loss": 0.2939,
+      "step": 363
+    },
+    {
+      "epoch": 0.23296,
+      "grad_norm": 0.36627198774624553,
+      "learning_rate": 4.5610401272498836e-05,
+      "loss": 0.2628,
+      "step": 364
+    },
+    {
+      "epoch": 0.2336,
+      "grad_norm": 0.3156535101635184,
+      "learning_rate": 4.558038047087486e-05,
+      "loss": 0.2725,
+      "step": 365
+    },
+    {
+      "epoch": 0.23424,
+      "grad_norm": 0.3552740428094254,
+      "learning_rate": 4.555026731206741e-05,
+      "loss": 0.2569,
+      "step": 366
+    },
+    {
+      "epoch": 0.23488,
+      "grad_norm": 0.34583005507723713,
+      "learning_rate": 4.5520061931213255e-05,
+      "loss": 0.2896,
+      "step": 367
+    },
+    {
+      "epoch": 0.23552,
+      "grad_norm": 0.32720433735825405,
+      "learning_rate": 4.548976446386305e-05,
+      "loss": 0.2836,
+      "step": 368
+    },
+    {
+      "epoch": 0.23616,
+      "grad_norm": 0.3472456803468133,
+      "learning_rate": 4.545937504598069e-05,
+      "loss": 0.2813,
+      "step": 369
+    },
+    {
+      "epoch": 0.2368,
+      "grad_norm": 0.3129024609409033,
+      "learning_rate": 4.5428893813942727e-05,
+      "loss": 0.2785,
+      "step": 370
+    },
+    {
+      "epoch": 0.23744,
+      "grad_norm": 0.3212608103449912,
+      "learning_rate": 4.539832090453771e-05,
+      "loss": 0.2745,
+      "step": 371
+    },
+    {
+      "epoch": 0.23808,
+      "grad_norm": 0.4051093522420126,
+      "learning_rate": 4.536765645496564e-05,
+      "loss": 0.2714,
+      "step": 372
+    },
+    {
+      "epoch": 0.23872,
+      "grad_norm": 0.33089670808030447,
+      "learning_rate": 4.533690060283727e-05,
+      "loss": 0.2746,
+      "step": 373
+    },
+    {
+      "epoch": 0.23936,
+      "grad_norm": 0.34412483483382633,
+      "learning_rate": 4.530605348617357e-05,
+      "loss": 0.284,
+      "step": 374
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.3538870569896249,
+      "learning_rate": 4.527511524340508e-05,
+      "loss": 0.2728,
+      "step": 375
+    },
+    {
+      "epoch": 0.24064,
+      "grad_norm": 0.34422528829878274,
+      "learning_rate": 4.5244086013371246e-05,
+      "loss": 0.2724,
+      "step": 376
+    },
+    {
+      "epoch": 0.24128,
+      "grad_norm": 0.3438880553189361,
+      "learning_rate": 4.521296593531985e-05,
+      "loss": 0.27,
+      "step": 377
+    },
+    {
+      "epoch": 0.24192,
+      "grad_norm": 0.3729480830628615,
+      "learning_rate": 4.518175514890636e-05,
+      "loss": 0.2741,
+      "step": 378
+    },
+    {
+      "epoch": 0.24256,
+      "grad_norm": 0.364266352426804,
+      "learning_rate": 4.515045379419332e-05,
+      "loss": 0.2876,
+      "step": 379
+    },
+    {
+      "epoch": 0.2432,
+      "grad_norm": 0.4244068645937319,
+      "learning_rate": 4.5119062011649714e-05,
+      "loss": 0.285,
+      "step": 380
+    },
+    {
+      "epoch": 0.24384,
+      "grad_norm": 0.4239270721214032,
+      "learning_rate": 4.508757994215031e-05,
+      "loss": 0.2885,
+      "step": 381
+    },
+    {
+      "epoch": 0.24448,
+      "grad_norm": 0.518496867004739,
+      "learning_rate": 4.505600772697507e-05,
+      "loss": 0.2897,
+      "step": 382
+    },
+    {
+      "epoch": 0.24512,
+      "grad_norm": 0.3569247474591793,
+      "learning_rate": 4.5024345507808495e-05,
+      "loss": 0.275,
+      "step": 383
+    },
+    {
+      "epoch": 0.24576,
+      "grad_norm": 0.47273370567346207,
+      "learning_rate": 4.499259342673898e-05,
+      "loss": 0.2853,
+      "step": 384
+    },
+    {
+      "epoch": 0.2464,
+      "grad_norm": 0.3309565654434764,
+      "learning_rate": 4.4960751626258205e-05,
+      "loss": 0.2738,
+      "step": 385
+    },
+    {
+      "epoch": 0.24704,
+      "grad_norm": 0.5804310609153428,
+      "learning_rate": 4.492882024926046e-05,
+      "loss": 0.2837,
+      "step": 386
+    },
+    {
+      "epoch": 0.24768,
+      "grad_norm": 0.37371497539840526,
+      "learning_rate": 4.489679943904203e-05,
+      "loss": 0.3041,
+      "step": 387
+    },
+    {
+      "epoch": 0.24832,
+      "grad_norm": 0.5041157772845973,
+      "learning_rate": 4.486468933930055e-05,
+      "loss": 0.2902,
+      "step": 388
+    },
+    {
+      "epoch": 0.24896,
+      "grad_norm": 0.3367298037962059,
+      "learning_rate": 4.483249009413433e-05,
+      "loss": 0.2671,
+      "step": 389
+    },
+    {
+      "epoch": 0.2496,
+      "grad_norm": 0.42359488925244737,
+      "learning_rate": 4.4800201848041765e-05,
+      "loss": 0.3,
+      "step": 390
+    },
+    {
+      "epoch": 0.25024,
+      "grad_norm": 0.33702260243515686,
+      "learning_rate": 4.476782474592062e-05,
+      "loss": 0.2725,
+      "step": 391
+    },
+    {
+      "epoch": 0.25088,
+      "grad_norm": 0.40124916013223433,
+      "learning_rate": 4.473535893306744e-05,
+      "loss": 0.2786,
+      "step": 392
+    },
+    {
+      "epoch": 0.25152,
+      "grad_norm": 0.32948759650087367,
+      "learning_rate": 4.470280455517687e-05,
+      "loss": 0.27,
+      "step": 393
+    },
+    {
+      "epoch": 0.25216,
+      "grad_norm": 0.3872660669166616,
+      "learning_rate": 4.4670161758340966e-05,
+      "loss": 0.2828,
+      "step": 394
+    },
+    {
+      "epoch": 0.2528,
+      "grad_norm": 0.37866556493954856,
+      "learning_rate": 4.4637430689048634e-05,
+      "loss": 0.2682,
+      "step": 395
+    },
+    {
+      "epoch": 0.25344,
+      "grad_norm": 0.8077458122891941,
+      "learning_rate": 4.460461149418486e-05,
+      "loss": 0.2643,
+      "step": 396
+    },
+    {
+      "epoch": 0.25408,
+      "grad_norm": 0.4159988370166188,
+      "learning_rate": 4.4571704321030125e-05,
+      "loss": 0.2975,
+      "step": 397
+    },
+    {
+      "epoch": 0.25472,
+      "grad_norm": 0.3145955100370753,
+      "learning_rate": 4.4538709317259754e-05,
+      "loss": 0.2542,
+      "step": 398
+    },
+    {
+      "epoch": 0.25536,
+      "grad_norm": 0.33316688174281767,
+      "learning_rate": 4.4505626630943174e-05,
+      "loss": 0.2707,
+      "step": 399
+    },
+    {
+      "epoch": 0.256,
+      "grad_norm": 0.3367770920983804,
+      "learning_rate": 4.447245641054334e-05,
+      "loss": 0.2834,
+      "step": 400
+    },
+    {
+      "epoch": 0.25664,
+      "grad_norm": 0.37345351718532394,
+      "learning_rate": 4.443919880491598e-05,
+      "loss": 0.2629,
+      "step": 401
+    },
+    {
+      "epoch": 0.25728,
+      "grad_norm": 0.3771657138606688,
+      "learning_rate": 4.440585396330904e-05,
+      "loss": 0.279,
+      "step": 402
+    },
+    {
+      "epoch": 0.25792,
+      "grad_norm": 0.3638878533690723,
+      "learning_rate": 4.437242203536189e-05,
+      "loss": 0.2813,
+      "step": 403
+    },
+    {
+      "epoch": 0.25856,
+      "grad_norm": 0.4085861931770922,
+      "learning_rate": 4.433890317110475e-05,
+      "loss": 0.2934,
+      "step": 404
+    },
+    {
+      "epoch": 0.2592,
+      "grad_norm": 0.3447902761950793,
+      "learning_rate": 4.430529752095795e-05,
+      "loss": 0.2811,
+      "step": 405
+    },
+    {
+      "epoch": 0.25984,
+      "grad_norm": 0.3280360576186305,
+      "learning_rate": 4.427160523573131e-05,
+      "loss": 0.2901,
+      "step": 406
+    },
+    {
+      "epoch": 0.26048,
+      "grad_norm": 0.4223470174065642,
+      "learning_rate": 4.4237826466623406e-05,
+      "loss": 0.2984,
+      "step": 407
+    },
+    {
+      "epoch": 0.26112,
+      "grad_norm": 0.33753085005422373,
+      "learning_rate": 4.420396136522095e-05,
+      "loss": 0.2826,
+      "step": 408
+    },
+    {
+      "epoch": 0.26176,
+      "grad_norm": 0.3975587234792085,
+      "learning_rate": 4.417001008349807e-05,
+      "loss": 0.2663,
+      "step": 409
+    },
+    {
+      "epoch": 0.2624,
+      "grad_norm": 0.40122264906135213,
+      "learning_rate": 4.413597277381564e-05,
+      "loss": 0.281,
+      "step": 410
+    },
+    {
+      "epoch": 0.26304,
+      "grad_norm": 0.38060503437157295,
+      "learning_rate": 4.410184958892059e-05,
+      "loss": 0.2767,
+      "step": 411
+    },
+    {
+      "epoch": 0.26368,
+      "grad_norm": 0.3937592373164496,
+      "learning_rate": 4.4067640681945244e-05,
+      "loss": 0.2731,
+      "step": 412
+    },
+    {
+      "epoch": 0.26432,
+      "grad_norm": 0.4605691683979008,
+      "learning_rate": 4.403334620640659e-05,
+      "loss": 0.271,
+      "step": 413
+    },
+    {
+      "epoch": 0.26496,
+      "grad_norm": 0.6048600304083459,
+      "learning_rate": 4.3998966316205636e-05,
+      "loss": 0.2636,
+      "step": 414
+    },
+    {
+      "epoch": 0.2656,
+      "grad_norm": 0.30782074187889363,
+      "learning_rate": 4.396450116562669e-05,
+      "loss": 0.2546,
+      "step": 415
+    },
+    {
+      "epoch": 0.26624,
+      "grad_norm": 0.32851004763201513,
+      "learning_rate": 4.392995090933669e-05,
+      "loss": 0.2618,
+      "step": 416
+    },
+    {
+      "epoch": 0.26688,
+      "grad_norm": 0.3657256553775175,
+      "learning_rate": 4.389531570238448e-05,
+      "loss": 0.2516,
+      "step": 417
+    },
+    {
+      "epoch": 0.26752,
+      "grad_norm": 0.4007877515742878,
+      "learning_rate": 4.386059570020015e-05,
+      "loss": 0.2875,
+      "step": 418
+    },
+    {
+      "epoch": 0.26816,
+      "grad_norm": 0.3916229190743489,
+      "learning_rate": 4.382579105859428e-05,
+      "loss": 0.2852,
+      "step": 419
+    },
+    {
+      "epoch": 0.2688,
+      "grad_norm": 0.3789835080039222,
+      "learning_rate": 4.379090193375735e-05,
+      "loss": 0.2786,
+      "step": 420
+    },
+    {
+      "epoch": 0.26944,
+      "grad_norm": 0.3713658334959446,
+      "learning_rate": 4.375592848225889e-05,
+      "loss": 0.2601,
+      "step": 421
+    },
+    {
+      "epoch": 0.27008,
+      "grad_norm": 0.38182410459365296,
+      "learning_rate": 4.372087086104692e-05,
+      "loss": 0.2884,
+      "step": 422
+    },
+    {
+      "epoch": 0.27072,
+      "grad_norm": 0.3269413529022408,
+      "learning_rate": 4.368572922744715e-05,
+      "loss": 0.2801,
+      "step": 423
+    },
+    {
+      "epoch": 0.27136,
+      "grad_norm": 0.4184922639041488,
+      "learning_rate": 4.365050373916233e-05,
+      "loss": 0.2864,
+      "step": 424
+    },
+    {
+      "epoch": 0.272,
+      "grad_norm": 0.37287235352305415,
+      "learning_rate": 4.3615194554271485e-05,
+      "loss": 0.2416,
+      "step": 425
+    },
+    {
+      "epoch": 0.27264,
+      "grad_norm": 0.403184875756434,
+      "learning_rate": 4.357980183122927e-05,
+      "loss": 0.2822,
+      "step": 426
+    },
+    {
+      "epoch": 0.27328,
+      "grad_norm": 0.47936480297091477,
+      "learning_rate": 4.354432572886522e-05,
+      "loss": 0.2822,
+      "step": 427
+    },
+    {
+      "epoch": 0.27392,
+      "grad_norm": 0.3323032086312353,
+      "learning_rate": 4.3508766406383036e-05,
+      "loss": 0.2901,
+      "step": 428
+    },
+    {
+      "epoch": 0.27456,
+      "grad_norm": 0.32982548305517356,
+      "learning_rate": 4.347312402335989e-05,
+      "loss": 0.2781,
+      "step": 429
+    },
+    {
+      "epoch": 0.2752,
+      "grad_norm": 0.33719802987074643,
+      "learning_rate": 4.34373987397457e-05,
+      "loss": 0.2892,
+      "step": 430
+    },
+    {
+      "epoch": 0.27584,
+      "grad_norm": 0.3742589239968179,
+      "learning_rate": 4.340159071586241e-05,
+      "loss": 0.2547,
+      "step": 431
+    },
+    {
+      "epoch": 0.27648,
+      "grad_norm": 0.33344944387070863,
+      "learning_rate": 4.336570011240326e-05,
+      "loss": 0.2553,
+      "step": 432
+    },
+    {
+      "epoch": 0.27712,
+      "grad_norm": 0.3726009633442184,
+      "learning_rate": 4.33297270904321e-05,
+      "loss": 0.279,
+      "step": 433
+    },
+    {
+      "epoch": 0.27776,
+      "grad_norm": 0.3325692898911964,
+      "learning_rate": 4.329367181138262e-05,
+      "loss": 0.2857,
+      "step": 434
+    },
+    {
+      "epoch": 0.2784,
+      "grad_norm": 0.3537978742787183,
+      "learning_rate": 4.3257534437057674e-05,
+      "loss": 0.2664,
+      "step": 435
+    },
+    {
+      "epoch": 0.27904,
+      "grad_norm": 0.33720807342566356,
+      "learning_rate": 4.3221315129628506e-05,
+      "loss": 0.2586,
+      "step": 436
+    },
+    {
+      "epoch": 0.27968,
+      "grad_norm": 0.3289955180606617,
+      "learning_rate": 4.318501405163406e-05,
+      "loss": 0.2717,
+      "step": 437
+    },
+    {
+      "epoch": 0.28032,
+      "grad_norm": 0.33540693261128296,
+      "learning_rate": 4.314863136598022e-05,
+      "loss": 0.2555,
+      "step": 438
+    },
+    {
+      "epoch": 0.28096,
+      "grad_norm": 0.39806190155501686,
+      "learning_rate": 4.311216723593913e-05,
+      "loss": 0.2975,
+      "step": 439
+    },
+    {
+      "epoch": 0.2816,
+      "grad_norm": 0.36978118892520295,
+      "learning_rate": 4.307562182514838e-05,
+      "loss": 0.2713,
+      "step": 440
+    },
+    {
+      "epoch": 0.28224,
+      "grad_norm": 0.440133126057951,
+      "learning_rate": 4.3038995297610364e-05,
+      "loss": 0.2581,
+      "step": 441
+    },
+    {
+      "epoch": 0.28288,
+      "grad_norm": 0.37987673032238434,
+      "learning_rate": 4.3002287817691466e-05,
+      "loss": 0.2977,
+      "step": 442
+    },
+    {
+      "epoch": 0.28352,
+      "grad_norm": 0.4046596131382216,
+      "learning_rate": 4.2965499550121354e-05,
+      "loss": 0.279,
+      "step": 443
+    },
+    {
+      "epoch": 0.28416,
+      "grad_norm": 0.3708521015749624,
+      "learning_rate": 4.2928630659992267e-05,
+      "loss": 0.2812,
+      "step": 444
+    },
+    {
+      "epoch": 0.2848,
+      "grad_norm": 0.4037004636721946,
+      "learning_rate": 4.2891681312758225e-05,
+      "loss": 0.2703,
+      "step": 445
+    },
+    {
+      "epoch": 0.28544,
+      "grad_norm": 0.33703660226197113,
+      "learning_rate": 4.285465167423432e-05,
+      "loss": 0.2828,
+      "step": 446
+    },
+    {
+      "epoch": 0.28608,
+      "grad_norm": 0.3746034323131707,
+      "learning_rate": 4.281754191059596e-05,
+      "loss": 0.2713,
+      "step": 447
+    },
+    {
+      "epoch": 0.28672,
+      "grad_norm": 0.38277330266368087,
+      "learning_rate": 4.278035218837813e-05,
+      "loss": 0.2982,
+      "step": 448
+    },
+    {
+      "epoch": 0.28736,
+      "grad_norm": 0.36593523245520027,
+      "learning_rate": 4.274308267447463e-05,
+      "loss": 0.2621,
+      "step": 449
+    },
+    {
+      "epoch": 0.288,
+      "grad_norm": 0.37234265656844234,
+      "learning_rate": 4.270573353613734e-05,
+      "loss": 0.2824,
+      "step": 450
+    },
+    {
+      "epoch": 0.28864,
+      "grad_norm": 0.3027373339129422,
+      "learning_rate": 4.266830494097546e-05,
+      "loss": 0.2627,
+      "step": 451
+    },
+    {
+      "epoch": 0.28928,
+      "grad_norm": 0.38089507553976193,
+      "learning_rate": 4.263079705695478e-05,
+      "loss": 0.2728,
+      "step": 452
+    },
+    {
+      "epoch": 0.28992,
+      "grad_norm": 0.40394637667004296,
+      "learning_rate": 4.2593210052396894e-05,
+      "loss": 0.2636,
+      "step": 453
+    },
+    {
+      "epoch": 0.29056,
+      "grad_norm": 0.35334153721363115,
+      "learning_rate": 4.255554409597846e-05,
+      "loss": 0.2503,
+      "step": 454
+    },
+    {
+      "epoch": 0.2912,
+      "grad_norm": 0.40619095792961374,
+      "learning_rate": 4.2517799356730445e-05,
+      "loss": 0.2795,
+      "step": 455
+    },
+    {
+      "epoch": 0.29184,
+      "grad_norm": 0.3957489109427142,
+      "learning_rate": 4.247997600403737e-05,
+      "loss": 0.2683,
+      "step": 456
+    },
+    {
+      "epoch": 0.29248,
+      "grad_norm": 0.37161859948412357,
+      "learning_rate": 4.244207420763655e-05,
+      "loss": 0.2579,
+      "step": 457
+    },
+    {
+      "epoch": 0.29312,
+      "grad_norm": 0.35367469754869557,
+      "learning_rate": 4.240409413761731e-05,
+      "loss": 0.2643,
+      "step": 458
+    },
+    {
+      "epoch": 0.29376,
+      "grad_norm": 0.402949661267483,
+      "learning_rate": 4.236603596442023e-05,
+      "loss": 0.2739,
+      "step": 459
+    },
+    {
+      "epoch": 0.2944,
+      "grad_norm": 0.3164385404746002,
+      "learning_rate": 4.232789985883642e-05,
+      "loss": 0.2787,
+      "step": 460
+    },
+    {
+      "epoch": 0.29504,
+      "grad_norm": 0.3611693162845143,
+      "learning_rate": 4.2289685992006704e-05,
+      "loss": 0.2875,
+      "step": 461
+    },
+    {
+      "epoch": 0.29568,
+      "grad_norm": 0.3816626308049793,
+      "learning_rate": 4.225139453542087e-05,
+      "loss": 0.2716,
+      "step": 462
+    },
+    {
+      "epoch": 0.29632,
+      "grad_norm": 0.38875072047102993,
+      "learning_rate": 4.2213025660916904e-05,
+      "loss": 0.2781,
+      "step": 463
+    },
+    {
+      "epoch": 0.29696,
+      "grad_norm": 0.4310708243108071,
+      "learning_rate": 4.21745795406802e-05,
+      "loss": 0.2563,
+      "step": 464
+    },
+    {
+      "epoch": 0.2976,
+      "grad_norm": 0.3549880460473045,
+      "learning_rate": 4.2136056347242835e-05,
+      "loss": 0.2601,
+      "step": 465
+    },
+    {
+      "epoch": 0.29824,
+      "grad_norm": 0.4956629293601871,
+      "learning_rate": 4.209745625348273e-05,
+      "loss": 0.2958,
+      "step": 466
+    },
+    {
+      "epoch": 0.29888,
+      "grad_norm": 0.38176123190536215,
+      "learning_rate": 4.205877943262291e-05,
+      "loss": 0.2527,
+      "step": 467
+    },
+    {
+      "epoch": 0.29952,
+      "grad_norm": 0.36213560022090674,
+      "learning_rate": 4.2020026058230756e-05,
+      "loss": 0.2794,
+      "step": 468
+    },
+    {
+      "epoch": 0.30016,
+      "grad_norm": 0.36552645561780356,
+      "learning_rate": 4.1981196304217154e-05,
+      "loss": 0.273,
+      "step": 469
+    },
+    {
+      "epoch": 0.3008,
+      "grad_norm": 0.32609323253400335,
+      "learning_rate": 4.1942290344835765e-05,
+      "loss": 0.2732,
+      "step": 470
+    },
+    {
+      "epoch": 0.30144,
+      "grad_norm": 0.4226814198890639,
+      "learning_rate": 4.190330835468224e-05,
+      "loss": 0.2772,
+      "step": 471
+    },
+    {
+      "epoch": 0.30208,
+      "grad_norm": 0.39067414904241504,
+      "learning_rate": 4.1864250508693424e-05,
+      "loss": 0.2894,
+      "step": 472
+    },
+    {
+      "epoch": 0.30272,
+      "grad_norm": 0.45775201623754663,
+      "learning_rate": 4.182511698214657e-05,
+      "loss": 0.2831,
+      "step": 473
+    },
+    {
+      "epoch": 0.30336,
+      "grad_norm": 0.336688375790536,
+      "learning_rate": 4.1785907950658573e-05,
+      "loss": 0.2703,
+      "step": 474
+    },
+    {
+      "epoch": 0.304,
+      "grad_norm": 0.4282219754460838,
+      "learning_rate": 4.1746623590185155e-05,
+      "loss": 0.264,
+      "step": 475
+    },
+    {
+      "epoch": 0.30464,
+      "grad_norm": 0.40314714831857074,
+      "learning_rate": 4.170726407702008e-05,
+      "loss": 0.2577,
+      "step": 476
+    },
+    {
+      "epoch": 0.30528,
+      "grad_norm": 0.4137806913884408,
+      "learning_rate": 4.1667829587794374e-05,
+      "loss": 0.294,
+      "step": 477
+    },
+    {
+      "epoch": 0.30592,
+      "grad_norm": 0.45659145697708425,
+      "learning_rate": 4.162832029947553e-05,
+      "loss": 0.2836,
+      "step": 478
+    },
+    {
+      "epoch": 0.30656,
+      "grad_norm": 0.35567263968733753,
+      "learning_rate": 4.158873638936672e-05,
+      "loss": 0.2573,
+      "step": 479
+    },
+    {
+      "epoch": 0.3072,
+      "grad_norm": 0.4119943441685636,
+      "learning_rate": 4.154907803510599e-05,
+      "loss": 0.2626,
+      "step": 480
+    },
+    {
+      "epoch": 0.30784,
+      "grad_norm": 0.3598946097878258,
+      "learning_rate": 4.150934541466545e-05,
+      "loss": 0.2721,
+      "step": 481
+    },
+    {
+      "epoch": 0.30848,
+      "grad_norm": 0.4034273306291238,
+      "learning_rate": 4.146953870635049e-05,
+      "loss": 0.2835,
+      "step": 482
+    },
+    {
+      "epoch": 0.30912,
+      "grad_norm": 0.34858590847101595,
+      "learning_rate": 4.1429658088798996e-05,
+      "loss": 0.2737,
+      "step": 483
+    },
+    {
+      "epoch": 0.30976,
+      "grad_norm": 0.37552704992362285,
+      "learning_rate": 4.138970374098052e-05,
+      "loss": 0.2591,
+      "step": 484
+    },
+    {
+      "epoch": 0.3104,
+      "grad_norm": 0.4004568314634453,
+      "learning_rate": 4.1349675842195494e-05,
+      "loss": 0.2778,
+      "step": 485
+    },
+    {
+      "epoch": 0.31104,
+      "grad_norm": 0.3746365763341882,
+      "learning_rate": 4.130957457207441e-05,
+      "loss": 0.2617,
+      "step": 486
+    },
+    {
+      "epoch": 0.31168,
+      "grad_norm": 0.36274356671175684,
+      "learning_rate": 4.126940011057703e-05,
+      "loss": 0.2511,
+      "step": 487
+    },
+    {
+      "epoch": 0.31232,
+      "grad_norm": 0.6225737712454078,
+      "learning_rate": 4.1229152637991576e-05,
+      "loss": 0.2798,
+      "step": 488
+    },
+    {
+      "epoch": 0.31296,
+      "grad_norm": 0.37400409273643975,
+      "learning_rate": 4.1188832334933906e-05,
+      "loss": 0.2574,
+      "step": 489
+    },
+    {
+      "epoch": 0.3136,
+      "grad_norm": 0.35566534633613417,
+      "learning_rate": 4.114843938234673e-05,
+      "loss": 0.2857,
+      "step": 490
+    },
+    {
+      "epoch": 0.31424,
+      "grad_norm": 0.4108005259388909,
+      "learning_rate": 4.1107973961498763e-05,
+      "loss": 0.2693,
+      "step": 491
+    },
+    {
+      "epoch": 0.31488,
+      "grad_norm": 0.36092465432455656,
+      "learning_rate": 4.106743625398395e-05,
+      "loss": 0.2784,
+      "step": 492
+    },
+    {
+      "epoch": 0.31552,
+      "grad_norm": 0.45843265748863465,
+      "learning_rate": 4.102682644172061e-05,
+      "loss": 0.2998,
+      "step": 493
+    },
+    {
+      "epoch": 0.31616,
+      "grad_norm": 0.3836584600353168,
+      "learning_rate": 4.0986144706950677e-05,
+      "loss": 0.2851,
+      "step": 494
+    },
+    {
+      "epoch": 0.3168,
+      "grad_norm": 0.36209166782757896,
+      "learning_rate": 4.094539123223881e-05,
+      "loss": 0.2534,
+      "step": 495
+    },
+    {
+      "epoch": 0.31744,
+      "grad_norm": 0.363289922269712,
+      "learning_rate": 4.090456620047162e-05,
+      "loss": 0.2537,
+      "step": 496
+    },
+    {
+      "epoch": 0.31808,
+      "grad_norm": 0.40363410063658856,
+      "learning_rate": 4.086366979485685e-05,
+      "loss": 0.2909,
+      "step": 497
+    },
+    {
+      "epoch": 0.31872,
+      "grad_norm": 0.3675317192463888,
+      "learning_rate": 4.082270219892254e-05,
+      "loss": 0.2749,
+      "step": 498
+    },
+    {
+      "epoch": 0.31936,
+      "grad_norm": 0.35397211220950364,
+      "learning_rate": 4.0781663596516206e-05,
+      "loss": 0.2534,
+      "step": 499
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.37572349613669165,
+      "learning_rate": 4.074055417180401e-05,
+      "loss": 0.281,
+      "step": 500
+    },
+    {
+      "epoch": 0.32064,
+      "grad_norm": 0.354329683029619,
+      "learning_rate": 4.069937410926994e-05,
+      "loss": 0.2655,
+      "step": 501
+    },
+    {
+      "epoch": 0.32128,
+      "grad_norm": 0.37756930001066114,
+      "learning_rate": 4.0658123593714984e-05,
+      "loss": 0.2755,
+      "step": 502
+    },
+    {
+      "epoch": 0.32192,
+      "grad_norm": 0.3367253572596431,
+      "learning_rate": 4.061680281025628e-05,
+      "loss": 0.2583,
+      "step": 503
+    },
+    {
+      "epoch": 0.32256,
+      "grad_norm": 0.3613298986999919,
+      "learning_rate": 4.057541194432634e-05,
+      "loss": 0.2513,
+      "step": 504
+    },
+    {
+      "epoch": 0.3232,
+      "grad_norm": 0.3851876530622504,
+      "learning_rate": 4.053395118167214e-05,
+      "loss": 0.2702,
+      "step": 505
+    },
+    {
+      "epoch": 0.32384,
+      "grad_norm": 0.33176376757169096,
+      "learning_rate": 4.049242070835435e-05,
+      "loss": 0.3008,
+      "step": 506
+    },
+    {
+      "epoch": 0.32448,
+      "grad_norm": 0.43271105984804215,
+      "learning_rate": 4.0450820710746475e-05,
+      "loss": 0.2702,
+      "step": 507
+    },
+    {
+      "epoch": 0.32512,
+      "grad_norm": 0.41884794686476173,
+      "learning_rate": 4.0409151375534014e-05,
+      "loss": 0.2746,
+      "step": 508
+    },
+    {
+      "epoch": 0.32576,
+      "grad_norm": 0.35789726048814424,
+      "learning_rate": 4.036741288971362e-05,
+      "loss": 0.2665,
+      "step": 509
+    },
+    {
+      "epoch": 0.3264,
+      "grad_norm": 0.4224006865959535,
+      "learning_rate": 4.032560544059228e-05,
+      "loss": 0.2641,
+      "step": 510
+    },
+    {
+      "epoch": 0.32704,
+      "grad_norm": 0.3410319732577578,
+      "learning_rate": 4.028372921578647e-05,
+      "loss": 0.2752,
+      "step": 511
+    },
+    {
+      "epoch": 0.32768,
+      "grad_norm": 0.32165719002887183,
+      "learning_rate": 4.024178440322128e-05,
+      "loss": 0.2622,
+      "step": 512
+    },
+    {
+      "epoch": 0.32832,
+      "grad_norm": 0.3190470917399596,
+      "learning_rate": 4.019977119112962e-05,
+      "loss": 0.271,
+      "step": 513
+    },
+    {
+      "epoch": 0.32896,
+      "grad_norm": 0.40388087068190615,
+      "learning_rate": 4.015768976805135e-05,
+      "loss": 0.3163,
+      "step": 514
+    },
+    {
+      "epoch": 0.3296,
+      "grad_norm": 0.3301304322280843,
+      "learning_rate": 4.011554032283242e-05,
+      "loss": 0.242,
+      "step": 515
+    },
+    {
+      "epoch": 0.33024,
+      "grad_norm": 0.3784972626675363,
+      "learning_rate": 4.007332304462406e-05,
+      "loss": 0.2603,
+      "step": 516
+    },
+    {
+      "epoch": 0.33088,
+      "grad_norm": 0.34576256994551374,
+      "learning_rate": 4.003103812288189e-05,
+      "loss": 0.2785,
+      "step": 517
+    },
+    {
+      "epoch": 0.33152,
+      "grad_norm": 0.34457418246401467,
+      "learning_rate": 3.99886857473651e-05,
+      "loss": 0.2619,
+      "step": 518
+    },
+    {
+      "epoch": 0.33216,
+      "grad_norm": 0.4013301994744993,
+      "learning_rate": 3.9946266108135575e-05,
+      "loss": 0.2879,
+      "step": 519
+    },
+    {
+      "epoch": 0.3328,
+      "grad_norm": 0.338916338513705,
+      "learning_rate": 3.9903779395557087e-05,
+      "loss": 0.272,
+      "step": 520
+    },
+    {
+      "epoch": 0.33344,
+      "grad_norm": 0.3572729876908719,
+      "learning_rate": 3.9861225800294366e-05,
+      "loss": 0.2391,
+      "step": 521
+    },
+    {
+      "epoch": 0.33408,
+      "grad_norm": 0.4471657467826066,
+      "learning_rate": 3.981860551331232e-05,
+      "loss": 0.2783,
+      "step": 522
+    },
+    {
+      "epoch": 0.33472,
+      "grad_norm": 0.3659670843167768,
+      "learning_rate": 3.977591872587512e-05,
+      "loss": 0.2649,
+      "step": 523
+    },
+    {
+      "epoch": 0.33536,
+      "grad_norm": 0.3571631732614458,
+      "learning_rate": 3.973316562954539e-05,
+      "loss": 0.2624,
+      "step": 524
+    },
+    {
+      "epoch": 0.336,
+      "grad_norm": 0.37662157327541707,
+      "learning_rate": 3.969034641618332e-05,
+      "loss": 0.2973,
+      "step": 525
+    },
+    {
+      "epoch": 0.33664,
+      "grad_norm": 0.3917876982425241,
+      "learning_rate": 3.964746127794579e-05,
+      "loss": 0.2737,
+      "step": 526
+    },
+    {
+      "epoch": 0.33728,
+      "grad_norm": 0.3864888517871659,
+      "learning_rate": 3.9604510407285524e-05,
+      "loss": 0.2603,
+      "step": 527
+    },
+    {
+      "epoch": 0.33792,
+      "grad_norm": 0.38278130064111704,
+      "learning_rate": 3.956149399695027e-05,
+      "loss": 0.2628,
+      "step": 528
+    },
+    {
+      "epoch": 0.33856,
+      "grad_norm": 0.45847376635991144,
+      "learning_rate": 3.951841223998186e-05,
+      "loss": 0.2929,
+      "step": 529
+    },
+    {
+      "epoch": 0.3392,
+      "grad_norm": 0.36857429554569987,
+      "learning_rate": 3.947526532971538e-05,
+      "loss": 0.2544,
+      "step": 530
+    },
+    {
+      "epoch": 0.33984,
+      "grad_norm": 0.3433461223245656,
+      "learning_rate": 3.943205345977831e-05,
+      "loss": 0.2703,
+      "step": 531
+    },
+    {
+      "epoch": 0.34048,
+      "grad_norm": 0.3735053112541835,
+      "learning_rate": 3.9388776824089636e-05,
+      "loss": 0.2651,
+      "step": 532
+    },
+    {
+      "epoch": 0.34112,
+      "grad_norm": 0.3481943182819895,
+      "learning_rate": 3.9345435616858995e-05,
+      "loss": 0.2779,
+      "step": 533
+    },
+    {
+      "epoch": 0.34176,
+      "grad_norm": 0.3782402031128072,
+      "learning_rate": 3.930203003258579e-05,
+      "loss": 0.2673,
+      "step": 534
+    },
+    {
+      "epoch": 0.3424,
+      "grad_norm": 0.34788121494226615,
+      "learning_rate": 3.9258560266058335e-05,
+      "loss": 0.2804,
+      "step": 535
+    },
+    {
+      "epoch": 0.34304,
+      "grad_norm": 0.38038146851783505,
+      "learning_rate": 3.921502651235297e-05,
+      "loss": 0.2646,
+      "step": 536
+    },
+    {
+      "epoch": 0.34368,
+      "grad_norm": 0.36667667822950395,
+      "learning_rate": 3.917142896683318e-05,
+      "loss": 0.2827,
+      "step": 537
+    },
+    {
+      "epoch": 0.34432,
+      "grad_norm": 0.35582671962828055,
+      "learning_rate": 3.912776782514872e-05,
+      "loss": 0.2644,
+      "step": 538
+    },
+    {
+      "epoch": 0.34496,
+      "grad_norm": 0.3896719980448358,
+      "learning_rate": 3.908404328323475e-05,
+      "loss": 0.2857,
+      "step": 539
+    },
+    {
+      "epoch": 0.3456,
+      "grad_norm": 0.36566794820876186,
+      "learning_rate": 3.904025553731094e-05,
+      "loss": 0.2524,
+      "step": 540
+    },
+    {
+      "epoch": 0.34624,
+      "grad_norm": 0.35530792640997244,
+      "learning_rate": 3.8996404783880606e-05,
+      "loss": 0.2535,
+      "step": 541
+    },
+    {
+      "epoch": 0.34688,
+      "grad_norm": 0.36391659810864585,
+      "learning_rate": 3.895249121972981e-05,
+      "loss": 0.2539,
+      "step": 542
+    },
+    {
+      "epoch": 0.34752,
+      "grad_norm": 0.34311083691607236,
+      "learning_rate": 3.890851504192649e-05,
+      "loss": 0.2801,
+      "step": 543
+    },
+    {
+      "epoch": 0.34816,
+      "grad_norm": 0.38066224559392836,
+      "learning_rate": 3.8864476447819565e-05,
+      "loss": 0.2576,
+      "step": 544
+    },
+    {
+      "epoch": 0.3488,
+      "grad_norm": 0.3743303492757066,
+      "learning_rate": 3.8820375635038067e-05,
+      "loss": 0.268,
+      "step": 545
+    },
+    {
+      "epoch": 0.34944,
+      "grad_norm": 0.44997812051621155,
+      "learning_rate": 3.877621280149022e-05,
+      "loss": 0.2789,
+      "step": 546
+    },
+    {
+      "epoch": 0.35008,
+      "grad_norm": 0.3103607872894965,
+      "learning_rate": 3.87319881453626e-05,
+      "loss": 0.2445,
+      "step": 547
+    },
+    {
+      "epoch": 0.35072,
+      "grad_norm": 0.5459536748422096,
+      "learning_rate": 3.868770186511921e-05,
+      "loss": 0.2743,
+      "step": 548
+    },
+    {
+      "epoch": 0.35136,
+      "grad_norm": 0.37701888899445407,
+      "learning_rate": 3.864335415950059e-05,
+      "loss": 0.2536,
+      "step": 549
+    },
+    {
+      "epoch": 0.352,
+      "grad_norm": 0.32997499891684023,
+      "learning_rate": 3.859894522752297e-05,
+      "loss": 0.2605,
+      "step": 550
+    },
+    {
+      "epoch": 0.35264,
+      "grad_norm": 0.4489525001004636,
+      "learning_rate": 3.8554475268477274e-05,
+      "loss": 0.2685,
+      "step": 551
+    },
+    {
+      "epoch": 0.35328,
+      "grad_norm": 0.3719428634191375,
+      "learning_rate": 3.850994448192836e-05,
+      "loss": 0.2709,
+      "step": 552
+    },
+    {
+      "epoch": 0.35392,
+      "grad_norm": 0.3446891353423413,
+      "learning_rate": 3.846535306771404e-05,
+      "loss": 0.2611,
+      "step": 553
+    },
+    {
+      "epoch": 0.35456,
+      "grad_norm": 0.43618706390392636,
+      "learning_rate": 3.842070122594417e-05,
+      "loss": 0.276,
+      "step": 554
+    },
+    {
+      "epoch": 0.3552,
+      "grad_norm": 0.382064177861162,
+      "learning_rate": 3.837598915699981e-05,
+      "loss": 0.2552,
+      "step": 555
+    },
+    {
+      "epoch": 0.35584,
+      "grad_norm": 0.39162284786707907,
+      "learning_rate": 3.83312170615323e-05,
+      "loss": 0.2489,
+      "step": 556
+    },
+    {
+      "epoch": 0.35648,
+      "grad_norm": 0.42221136906147266,
+      "learning_rate": 3.8286385140462334e-05,
+      "loss": 0.2597,
+      "step": 557
+    },
+    {
+      "epoch": 0.35712,
+      "grad_norm": 0.44919209097377344,
+      "learning_rate": 3.8241493594979114e-05,
+      "loss": 0.2758,
+      "step": 558
+    },
+    {
+      "epoch": 0.35776,
+      "grad_norm": 0.39994455826831043,
+      "learning_rate": 3.8196542626539396e-05,
+      "loss": 0.275,
+      "step": 559
+    },
+    {
+      "epoch": 0.3584,
+      "grad_norm": 0.4015098875627428,
+      "learning_rate": 3.815153243686658e-05,
+      "loss": 0.2779,
+      "step": 560
+    },
+    {
+      "epoch": 0.35904,
+      "grad_norm": 0.4010038099709535,
+      "learning_rate": 3.8106463227949885e-05,
+      "loss": 0.2757,
+      "step": 561
+    },
+    {
+      "epoch": 0.35968,
+      "grad_norm": 0.4271974019014666,
+      "learning_rate": 3.806133520204335e-05,
+      "loss": 0.248,
+      "step": 562
+    },
+    {
+      "epoch": 0.36032,
+      "grad_norm": 0.35793331554825897,
+      "learning_rate": 3.801614856166495e-05,
+      "loss": 0.2561,
+      "step": 563
+    },
+    {
+      "epoch": 0.36096,
+      "grad_norm": 0.35924812032427766,
+      "learning_rate": 3.797090350959574e-05,
+      "loss": 0.2578,
+      "step": 564
+    },
+    {
+      "epoch": 0.3616,
+      "grad_norm": 0.3394416638286728,
+      "learning_rate": 3.792560024887887e-05,
+      "loss": 0.2645,
+      "step": 565
+    },
+    {
+      "epoch": 0.36224,
+      "grad_norm": 0.388540218562409,
+      "learning_rate": 3.788023898281873e-05,
+      "loss": 0.2633,
+      "step": 566
+    },
+    {
+      "epoch": 0.36288,
+      "grad_norm": 0.5109417791849185,
+      "learning_rate": 3.783481991498001e-05,
+      "loss": 0.2687,
+      "step": 567
+    },
+    {
+      "epoch": 0.36352,
+      "grad_norm": 0.3241009584346598,
+      "learning_rate": 3.7789343249186805e-05,
+      "loss": 0.274,
+      "step": 568
+    },
+    {
+      "epoch": 0.36416,
+      "grad_norm": 0.42943520321323175,
+      "learning_rate": 3.774380918952165e-05,
+      "loss": 0.286,
+      "step": 569
+    },
+    {
+      "epoch": 0.3648,
+      "grad_norm": 0.38392512011379704,
+      "learning_rate": 3.76982179403247e-05,
+      "loss": 0.2472,
+      "step": 570
+    },
+    {
+      "epoch": 0.36544,
+      "grad_norm": 0.37469396336292704,
+      "learning_rate": 3.7652569706192694e-05,
+      "loss": 0.2784,
+      "step": 571
+    },
+    {
+      "epoch": 0.36608,
+      "grad_norm": 0.315430957897709,
+      "learning_rate": 3.760686469197815e-05,
+      "loss": 0.2591,
+      "step": 572
+    },
+    {
+      "epoch": 0.36672,
+      "grad_norm": 0.3948730757632361,
+      "learning_rate": 3.756110310278835e-05,
+      "loss": 0.2592,
+      "step": 573
+    },
+    {
+      "epoch": 0.36736,
+      "grad_norm": 0.2922718271346419,
+      "learning_rate": 3.7515285143984524e-05,
+      "loss": 0.253,
+      "step": 574
+    },
+    {
+      "epoch": 0.368,
+      "grad_norm": 0.32249504875735685,
+      "learning_rate": 3.746941102118082e-05,
+      "loss": 0.2604,
+      "step": 575
+    },
+    {
+      "epoch": 0.36864,
+      "grad_norm": 0.3733899533072396,
+      "learning_rate": 3.742348094024342e-05,
+      "loss": 0.2845,
+      "step": 576
+    },
+    {
+      "epoch": 0.36928,
+      "grad_norm": 0.6846199585312611,
+      "learning_rate": 3.7377495107289664e-05,
+      "loss": 0.2375,
+      "step": 577
+    },
+    {
+      "epoch": 0.36992,
+      "grad_norm": 0.31965772500499295,
+      "learning_rate": 3.7331453728687054e-05,
+      "loss": 0.2639,
+      "step": 578
+    },
+    {
+      "epoch": 0.37056,
+      "grad_norm": 0.39643150388340115,
+      "learning_rate": 3.7285357011052394e-05,
+      "loss": 0.2496,
+      "step": 579
+    },
+    {
+      "epoch": 0.3712,
+      "grad_norm": 0.35821403759138415,
+      "learning_rate": 3.72392051612508e-05,
+      "loss": 0.2816,
+      "step": 580
+    },
+    {
+      "epoch": 0.37184,
+      "grad_norm": 0.36596844709726617,
+      "learning_rate": 3.719299838639481e-05,
+      "loss": 0.242,
+      "step": 581
+    },
+    {
+      "epoch": 0.37248,
+      "grad_norm": 0.34775224585235587,
+      "learning_rate": 3.714673689384346e-05,
+      "loss": 0.285,
+      "step": 582
+    },
+    {
+      "epoch": 0.37312,
+      "grad_norm": 0.3687742245411094,
+      "learning_rate": 3.710042089120131e-05,
+      "loss": 0.2589,
+      "step": 583
+    },
+    {
+      "epoch": 0.37376,
+      "grad_norm": 0.34846549220601303,
+      "learning_rate": 3.705405058631756e-05,
+      "loss": 0.275,
+      "step": 584
+    },
+    {
+      "epoch": 0.3744,
+      "grad_norm": 0.5129589237563266,
+      "learning_rate": 3.700762618728508e-05,
+      "loss": 0.2707,
+      "step": 585
+    },
+    {
+      "epoch": 0.37504,
+      "grad_norm": 0.3748604855678063,
+      "learning_rate": 3.696114790243955e-05,
+      "loss": 0.2756,
+      "step": 586
+    },
+    {
+      "epoch": 0.37568,
+      "grad_norm": 0.32727644759108254,
+      "learning_rate": 3.691461594035841e-05,
+      "loss": 0.2923,
+      "step": 587
+    },
+    {
+      "epoch": 0.37632,
+      "grad_norm": 0.33424977466935957,
+      "learning_rate": 3.686803050986e-05,
+      "loss": 0.2588,
+      "step": 588
+    },
+    {
+      "epoch": 0.37696,
+      "grad_norm": 0.40400950364676924,
+      "learning_rate": 3.6821391820002596e-05,
+      "loss": 0.254,
+      "step": 589
+    },
+    {
+      "epoch": 0.3776,
+      "grad_norm": 0.31623711185199815,
+      "learning_rate": 3.677470008008352e-05,
+      "loss": 0.2563,
+      "step": 590
+    },
+    {
+      "epoch": 0.37824,
+      "grad_norm": 0.36249142907953896,
+      "learning_rate": 3.672795549963814e-05,
+      "loss": 0.2565,
+      "step": 591
+    },
+    {
+      "epoch": 0.37888,
+      "grad_norm": 0.3631221100285136,
+      "learning_rate": 3.668115828843894e-05,
+      "loss": 0.2611,
+      "step": 592
+    },
+    {
+      "epoch": 0.37952,
+      "grad_norm": 0.3346078886412221,
+      "learning_rate": 3.6634308656494605e-05,
+      "loss": 0.2617,
+      "step": 593
+    },
+    {
+      "epoch": 0.38016,
+      "grad_norm": 0.3570393467254589,
+      "learning_rate": 3.6587406814049075e-05,
+      "loss": 0.2507,
+      "step": 594
+    },
+    {
+      "epoch": 0.3808,
+      "grad_norm": 0.3280741391605246,
+      "learning_rate": 3.6540452971580566e-05,
+      "loss": 0.2694,
+      "step": 595
+    },
+    {
+      "epoch": 0.38144,
+      "grad_norm": 0.3753506411790363,
+      "learning_rate": 3.649344733980068e-05,
+      "loss": 0.2647,
+      "step": 596
+    },
+    {
+      "epoch": 0.38208,
+      "grad_norm": 0.3590307898042495,
+      "learning_rate": 3.6446390129653405e-05,
+      "loss": 0.2603,
+      "step": 597
+    },
+    {
+      "epoch": 0.38272,
+      "grad_norm": 0.35086405624134986,
+      "learning_rate": 3.63992815523142e-05,
+      "loss": 0.2518,
+      "step": 598
+    },
+    {
+      "epoch": 0.38336,
+      "grad_norm": 0.36113674021952225,
+      "learning_rate": 3.6352121819189075e-05,
+      "loss": 0.2521,
+      "step": 599
+    },
+    {
+      "epoch": 0.384,
+      "grad_norm": 0.3501211246872176,
+      "learning_rate": 3.630491114191354e-05,
+      "loss": 0.2509,
+      "step": 600
+    },
+    {
+      "epoch": 0.38464,
+      "grad_norm": 0.3538362040001931,
+      "learning_rate": 3.62576497323518e-05,
+      "loss": 0.2916,
+      "step": 601
+    },
+    {
+      "epoch": 0.38528,
+      "grad_norm": 0.3458262257611667,
+      "learning_rate": 3.621033780259566e-05,
+      "loss": 0.251,
+      "step": 602
+    },
+    {
+      "epoch": 0.38592,
+      "grad_norm": 0.3439429816589424,
+      "learning_rate": 3.616297556496369e-05,
+      "loss": 0.2491,
+      "step": 603
+    },
+    {
+      "epoch": 0.38656,
+      "grad_norm": 0.33029933576957676,
+      "learning_rate": 3.6115563232000206e-05,
+      "loss": 0.247,
+      "step": 604
+    },
+    {
+      "epoch": 0.3872,
+      "grad_norm": 0.34999241376454604,
+      "learning_rate": 3.6068101016474314e-05,
+      "loss": 0.2509,
+      "step": 605
+    },
+    {
+      "epoch": 0.38784,
+      "grad_norm": 0.36517743046091355,
+      "learning_rate": 3.602058913137902e-05,
+      "loss": 0.2652,
+      "step": 606
+    },
+    {
+      "epoch": 0.38848,
+      "grad_norm": 0.30565774183933747,
+      "learning_rate": 3.597302778993018e-05,
+      "loss": 0.2657,
+      "step": 607
+    },
+    {
+      "epoch": 0.38912,
+      "grad_norm": 0.3786174413033551,
+      "learning_rate": 3.592541720556561e-05,
+      "loss": 0.2595,
+      "step": 608
+    },
+    {
+      "epoch": 0.38976,
+      "grad_norm": 0.33517197230276674,
+      "learning_rate": 3.5877757591944137e-05,
+      "loss": 0.2661,
+      "step": 609
+    },
+    {
+      "epoch": 0.3904,
+      "grad_norm": 0.3870662595735502,
+      "learning_rate": 3.583004916294458e-05,
+      "loss": 0.2722,
+      "step": 610
+    },
+    {
+      "epoch": 0.39104,
+      "grad_norm": 0.33071773275046074,
+      "learning_rate": 3.578229213266483e-05,
+      "loss": 0.2631,
+      "step": 611
+    },
+    {
+      "epoch": 0.39168,
+      "grad_norm": 0.3766992745350915,
+      "learning_rate": 3.573448671542088e-05,
+      "loss": 0.2623,
+      "step": 612
+    },
+    {
+      "epoch": 0.39232,
+      "grad_norm": 0.3571850708801622,
+      "learning_rate": 3.568663312574589e-05,
+      "loss": 0.2461,
+      "step": 613
+    },
+    {
+      "epoch": 0.39296,
+      "grad_norm": 0.3560726009033962,
+      "learning_rate": 3.563873157838917e-05,
+      "loss": 0.2695,
+      "step": 614
+    },
+    {
+      "epoch": 0.3936,
+      "grad_norm": 0.3700053868442448,
+      "learning_rate": 3.5590782288315266e-05,
+      "loss": 0.2826,
+      "step": 615
+    },
+    {
+      "epoch": 0.39424,
+      "grad_norm": 0.3983573086562803,
+      "learning_rate": 3.5542785470702965e-05,
+      "loss": 0.2698,
+      "step": 616
+    },
+    {
+      "epoch": 0.39488,
+      "grad_norm": 0.3449289599008837,
+      "learning_rate": 3.549474134094434e-05,
+      "loss": 0.2712,
+      "step": 617
+    },
+    {
+      "epoch": 0.39552,
+      "grad_norm": 0.39295042631291166,
+      "learning_rate": 3.54466501146438e-05,
+      "loss": 0.2528,
+      "step": 618
+    },
+    {
+      "epoch": 0.39616,
+      "grad_norm": 0.3491046524915478,
+      "learning_rate": 3.539851200761707e-05,
+      "loss": 0.2963,
+      "step": 619
+    },
+    {
+      "epoch": 0.3968,
+      "grad_norm": 0.3828147045099311,
+      "learning_rate": 3.535032723589031e-05,
+      "loss": 0.2504,
+      "step": 620
+    },
+    {
+      "epoch": 0.39744,
+      "grad_norm": 0.3804034417955643,
+      "learning_rate": 3.5302096015699035e-05,
+      "loss": 0.2647,
+      "step": 621
+    },
+    {
+      "epoch": 0.39808,
+      "grad_norm": 0.3508627973390516,
+      "learning_rate": 3.5253818563487266e-05,
+      "loss": 0.2622,
+      "step": 622
+    },
+    {
+      "epoch": 0.39872,
+      "grad_norm": 0.3722853571624787,
+      "learning_rate": 3.5205495095906444e-05,
+      "loss": 0.2625,
+      "step": 623
+    },
+    {
+      "epoch": 0.39936,
+      "grad_norm": 0.31703668399664725,
+      "learning_rate": 3.5157125829814534e-05,
+      "loss": 0.2529,
+      "step": 624
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.339874815690016,
+      "learning_rate": 3.510871098227503e-05,
+      "loss": 0.2407,
+      "step": 625
+    },
+    {
+      "epoch": 0.40064,
+      "grad_norm": 0.3662315328696493,
+      "learning_rate": 3.506025077055597e-05,
+      "loss": 0.3043,
+      "step": 626
+    },
+    {
+      "epoch": 0.40128,
+      "grad_norm": 0.3279928415081358,
+      "learning_rate": 3.501174541212898e-05,
+      "loss": 0.2535,
+      "step": 627
+    },
+    {
+      "epoch": 0.40192,
+      "grad_norm": 0.350715167992945,
+      "learning_rate": 3.496319512466827e-05,
+      "loss": 0.2669,
+      "step": 628
+    },
+    {
+      "epoch": 0.40256,
+      "grad_norm": 0.32602884803608445,
+      "learning_rate": 3.4914600126049696e-05,
+      "loss": 0.2566,
+      "step": 629
+    },
+    {
+      "epoch": 0.4032,
+      "grad_norm": 0.31088258016970977,
+      "learning_rate": 3.486596063434975e-05,
+      "loss": 0.2552,
+      "step": 630
+    },
+    {
+      "epoch": 0.40384,
+      "grad_norm": 0.30180827046053793,
+      "learning_rate": 3.481727686784458e-05,
+      "loss": 0.2547,
+      "step": 631
+    },
+    {
+      "epoch": 0.40448,
+      "grad_norm": 0.33732860100171874,
+      "learning_rate": 3.476854904500905e-05,
+      "loss": 0.2567,
+      "step": 632
+    },
+    {
+      "epoch": 0.40512,
+      "grad_norm": 0.40023070112905895,
+      "learning_rate": 3.471977738451572e-05,
+      "loss": 0.2641,
+      "step": 633
+    },
+    {
+      "epoch": 0.40576,
+      "grad_norm": 0.3468414197436556,
+      "learning_rate": 3.467096210523387e-05,
+      "loss": 0.264,
+      "step": 634
+    },
+    {
+      "epoch": 0.4064,
+      "grad_norm": 0.41069797654026435,
+      "learning_rate": 3.4622103426228534e-05,
+      "loss": 0.2759,
+      "step": 635
+    },
+    {
+      "epoch": 0.40704,
+      "grad_norm": 0.3548677977523007,
+      "learning_rate": 3.4573201566759515e-05,
+      "loss": 0.2586,
+      "step": 636
+    },
+    {
+      "epoch": 0.40768,
+      "grad_norm": 0.34730057827483873,
+      "learning_rate": 3.4524256746280374e-05,
+      "loss": 0.2654,
+      "step": 637
+    },
+    {
+      "epoch": 0.40832,
+      "grad_norm": 0.3469232464007306,
+      "learning_rate": 3.447526918443748e-05,
+      "loss": 0.2619,
+      "step": 638
+    },
+    {
+      "epoch": 0.40896,
+      "grad_norm": 0.38686628517651334,
+      "learning_rate": 3.442623910106901e-05,
+      "loss": 0.2404,
+      "step": 639
+    },
+    {
+      "epoch": 0.4096,
+      "grad_norm": 0.3119834776491984,
+      "learning_rate": 3.4377166716203945e-05,
+      "loss": 0.2511,
+      "step": 640
+    },
+    {
+      "epoch": 0.41024,
+      "grad_norm": 0.37217431767953546,
+      "learning_rate": 3.4328052250061124e-05,
+      "loss": 0.2573,
+      "step": 641
+    },
+    {
+      "epoch": 0.41088,
+      "grad_norm": 0.3660176951924303,
+      "learning_rate": 3.4278895923048226e-05,
+      "loss": 0.2656,
+      "step": 642
+    },
+    {
+      "epoch": 0.41152,
+      "grad_norm": 0.3204222601247596,
+      "learning_rate": 3.4229697955760767e-05,
+      "loss": 0.2371,
+      "step": 643
+    },
+    {
+      "epoch": 0.41216,
+      "grad_norm": 0.3700283904944704,
+      "learning_rate": 3.4180458568981136e-05,
+      "loss": 0.2923,
+      "step": 644
+    },
+    {
+      "epoch": 0.4128,
+      "grad_norm": 0.3121216825663056,
+      "learning_rate": 3.4131177983677624e-05,
+      "loss": 0.2368,
+      "step": 645
+    },
+    {
+      "epoch": 0.41344,
+      "grad_norm": 0.33775692842862853,
+      "learning_rate": 3.4081856421003364e-05,
+      "loss": 0.2478,
+      "step": 646
+    },
+    {
+      "epoch": 0.41408,
+      "grad_norm": 0.325764436348029,
+      "learning_rate": 3.403249410229542e-05,
+      "loss": 0.2593,
+      "step": 647
+    },
+    {
+      "epoch": 0.41472,
+      "grad_norm": 0.3707720956517282,
+      "learning_rate": 3.398309124907372e-05,
+      "loss": 0.2677,
+      "step": 648
+    },
+    {
+      "epoch": 0.41536,
+      "grad_norm": 0.3166532591304335,
+      "learning_rate": 3.393364808304012e-05,
+      "loss": 0.2859,
+      "step": 649
+    },
+    {
+      "epoch": 0.416,
+      "grad_norm": 0.34342576498220806,
+      "learning_rate": 3.3884164826077384e-05,
+      "loss": 0.255,
+      "step": 650
+    },
+    {
+      "epoch": 0.41664,
+      "grad_norm": 0.36895211989037313,
+      "learning_rate": 3.383464170024817e-05,
+      "loss": 0.2549,
+      "step": 651
+    },
+    {
+      "epoch": 0.41728,
+      "grad_norm": 0.35757524771092064,
+      "learning_rate": 3.378507892779408e-05,
+      "loss": 0.2839,
+      "step": 652
+    },
+    {
+      "epoch": 0.41792,
+      "grad_norm": 0.3420627089987508,
+      "learning_rate": 3.37354767311346e-05,
+      "loss": 0.265,
+      "step": 653
+    },
+    {
+      "epoch": 0.41856,
+      "grad_norm": 0.3761431521348441,
+      "learning_rate": 3.368583533286617e-05,
+      "loss": 0.2459,
+      "step": 654
+    },
+    {
+      "epoch": 0.4192,
+      "grad_norm": 0.32600703319937346,
+      "learning_rate": 3.363615495576115e-05,
+      "loss": 0.2675,
+      "step": 655
+    },
+    {
+      "epoch": 0.41984,
+      "grad_norm": 0.3269476142336683,
+      "learning_rate": 3.3586435822766785e-05,
+      "loss": 0.2688,
+      "step": 656
+    },
+    {
+      "epoch": 0.42048,
+      "grad_norm": 0.3350921803295603,
+      "learning_rate": 3.35366781570043e-05,
+      "loss": 0.2471,
+      "step": 657
+    },
+    {
+      "epoch": 0.42112,
+      "grad_norm": 0.3066280687711608,
+      "learning_rate": 3.3486882181767785e-05,
+      "loss": 0.2643,
+      "step": 658
+    },
+    {
+      "epoch": 0.42176,
+      "grad_norm": 0.3734013097320066,
+      "learning_rate": 3.34370481205233e-05,
+      "loss": 0.2729,
+      "step": 659
+    },
+    {
+      "epoch": 0.4224,
+      "grad_norm": 0.3159497563462825,
+      "learning_rate": 3.3387176196907776e-05,
+      "loss": 0.2715,
+      "step": 660
+    },
+    {
+      "epoch": 0.42304,
+      "grad_norm": 0.3365403314173876,
+      "learning_rate": 3.333726663472809e-05,
+      "loss": 0.2523,
+      "step": 661
+    },
+    {
+      "epoch": 0.42368,
+      "grad_norm": 0.35481058798184967,
+      "learning_rate": 3.328731965796e-05,
+      "loss": 0.2553,
+      "step": 662
+    },
+    {
+      "epoch": 0.42432,
+      "grad_norm": 0.32684593186658334,
+      "learning_rate": 3.32373354907472e-05,
+      "loss": 0.2439,
+      "step": 663
+    },
+    {
+      "epoch": 0.42496,
+      "grad_norm": 0.3433863164703469,
+      "learning_rate": 3.3187314357400244e-05,
+      "loss": 0.2457,
+      "step": 664
+    },
+    {
+      "epoch": 0.4256,
+      "grad_norm": 0.33223948828219063,
+      "learning_rate": 3.31372564823956e-05,
+      "loss": 0.2774,
+      "step": 665
+    },
+    {
+      "epoch": 0.42624,
+      "grad_norm": 0.3889022607769184,
+      "learning_rate": 3.308716209037462e-05,
+      "loss": 0.2652,
+      "step": 666
+    },
+    {
+      "epoch": 0.42688,
+      "grad_norm": 0.3295787331276891,
+      "learning_rate": 3.303703140614251e-05,
+      "loss": 0.2652,
+      "step": 667
+    },
+    {
+      "epoch": 0.42752,
+      "grad_norm": 0.3553030158930773,
+      "learning_rate": 3.2986864654667356e-05,
+      "loss": 0.2527,
+      "step": 668
+    },
+    {
+      "epoch": 0.42816,
+      "grad_norm": 0.31803094138145627,
+      "learning_rate": 3.293666206107911e-05,
+      "loss": 0.2619,
+      "step": 669
+    },
+    {
+      "epoch": 0.4288,
+      "grad_norm": 0.33392388467110345,
+      "learning_rate": 3.2886423850668555e-05,
+      "loss": 0.2604,
+      "step": 670
+    },
+    {
+      "epoch": 0.42944,
+      "grad_norm": 0.5718791983779791,
+      "learning_rate": 3.283615024888632e-05,
+      "loss": 0.2523,
+      "step": 671
+    },
+    {
+      "epoch": 0.43008,
+      "grad_norm": 0.35719182119680337,
+      "learning_rate": 3.278584148134185e-05,
+      "loss": 0.2809,
+      "step": 672
+    },
+    {
+      "epoch": 0.43072,
+      "grad_norm": 0.3726537342046652,
+      "learning_rate": 3.2735497773802406e-05,
+      "loss": 0.2827,
+      "step": 673
+    },
+    {
+      "epoch": 0.43136,
+      "grad_norm": 0.3370761913664551,
+      "learning_rate": 3.268511935219204e-05,
+      "loss": 0.2532,
+      "step": 674
+    },
+    {
+      "epoch": 0.432,
+      "grad_norm": 0.3276740016657337,
+      "learning_rate": 3.263470644259059e-05,
+      "loss": 0.2324,
+      "step": 675
+    },
+    {
+      "epoch": 0.43264,
+      "grad_norm": 0.3997607030788146,
+      "learning_rate": 3.258425927123266e-05,
+      "loss": 0.2799,
+      "step": 676
+    },
+    {
+      "epoch": 0.43328,
+      "grad_norm": 0.3142150656630589,
+      "learning_rate": 3.253377806450664e-05,
+      "loss": 0.2605,
+      "step": 677
+    },
+    {
+      "epoch": 0.43392,
+      "grad_norm": 0.35815771086322507,
+      "learning_rate": 3.248326304895361e-05,
+      "loss": 0.2361,
+      "step": 678
+    },
+    {
+      "epoch": 0.43456,
+      "grad_norm": 0.4093557089026302,
+      "learning_rate": 3.2432714451266395e-05,
+      "loss": 0.2735,
+      "step": 679
+    },
+    {
+      "epoch": 0.4352,
+      "grad_norm": 0.350660395761813,
+      "learning_rate": 3.238213249828853e-05,
+      "loss": 0.2696,
+      "step": 680
+    },
+    {
+      "epoch": 0.43584,
+      "grad_norm": 0.3946284387605494,
+      "learning_rate": 3.233151741701322e-05,
+      "loss": 0.2636,
+      "step": 681
+    },
+    {
+      "epoch": 0.43648,
+      "grad_norm": 0.3218343032055154,
+      "learning_rate": 3.228086943458235e-05,
+      "loss": 0.2694,
+      "step": 682
+    },
+    {
+      "epoch": 0.43712,
+      "grad_norm": 0.38999094622070707,
+      "learning_rate": 3.223018877828546e-05,
+      "loss": 0.2332,
+      "step": 683
+    },
+    {
+      "epoch": 0.43776,
+      "grad_norm": 0.36567793954540373,
+      "learning_rate": 3.2179475675558694e-05,
+      "loss": 0.2563,
+      "step": 684
+    },
+    {
+      "epoch": 0.4384,
+      "grad_norm": 0.3908225727082946,
+      "learning_rate": 3.212873035398383e-05,
+      "loss": 0.2625,
+      "step": 685
+    },
+    {
+      "epoch": 0.43904,
+      "grad_norm": 0.4048080809061653,
+      "learning_rate": 3.207795304128721e-05,
+      "loss": 0.2417,
+      "step": 686
+    },
+    {
+      "epoch": 0.43968,
+      "grad_norm": 0.3819495758094229,
+      "learning_rate": 3.2027143965338765e-05,
+      "loss": 0.2608,
+      "step": 687
+    },
+    {
+      "epoch": 0.44032,
+      "grad_norm": 0.36450745812493257,
+      "learning_rate": 3.1976303354150935e-05,
+      "loss": 0.2343,
+      "step": 688
+    },
+    {
+      "epoch": 0.44096,
+      "grad_norm": 0.32981854193809773,
+      "learning_rate": 3.192543143587771e-05,
+      "loss": 0.2452,
+      "step": 689
+    },
+    {
+      "epoch": 0.4416,
+      "grad_norm": 0.33458302589060956,
+      "learning_rate": 3.187452843881357e-05,
+      "loss": 0.26,
+      "step": 690
+    },
+    {
+      "epoch": 0.44224,
+      "grad_norm": 0.29087263740153546,
+      "learning_rate": 3.182359459139243e-05,
+      "loss": 0.2374,
+      "step": 691
+    },
+    {
+      "epoch": 0.44288,
+      "grad_norm": 0.41531050688687965,
+      "learning_rate": 3.1772630122186695e-05,
+      "loss": 0.261,
+      "step": 692
+    },
+    {
+      "epoch": 0.44352,
+      "grad_norm": 0.30548967146916095,
+      "learning_rate": 3.172163525990617e-05,
+      "loss": 0.2771,
+      "step": 693
+    },
+    {
+      "epoch": 0.44416,
+      "grad_norm": 0.34591377600837825,
+      "learning_rate": 3.167061023339706e-05,
+      "loss": 0.2623,
+      "step": 694
+    },
+    {
+      "epoch": 0.4448,
+      "grad_norm": 0.3335643857725579,
+      "learning_rate": 3.161955527164092e-05,
+      "loss": 0.2636,
+      "step": 695
+    },
+    {
+      "epoch": 0.44544,
+      "grad_norm": 0.3144634017643318,
+      "learning_rate": 3.1568470603753666e-05,
+      "loss": 0.2466,
+      "step": 696
+    },
+    {
+      "epoch": 0.44608,
+      "grad_norm": 0.39048151821741967,
+      "learning_rate": 3.1517356458984485e-05,
+      "loss": 0.2606,
+      "step": 697
+    },
+    {
+      "epoch": 0.44672,
+      "grad_norm": 0.29674666006523004,
+      "learning_rate": 3.14662130667149e-05,
+      "loss": 0.2313,
+      "step": 698
+    },
+    {
+      "epoch": 0.44736,
+      "grad_norm": 0.3710270379316393,
+      "learning_rate": 3.141504065645765e-05,
+      "loss": 0.2718,
+      "step": 699
+    },
+    {
+      "epoch": 0.448,
+      "grad_norm": 0.3189113410568584,
+      "learning_rate": 3.1363839457855704e-05,
+      "loss": 0.2599,
+      "step": 700
+    },
+    {
+      "epoch": 0.44864,
+      "grad_norm": 0.33433728058905765,
+      "learning_rate": 3.131260970068122e-05,
+      "loss": 0.2598,
+      "step": 701
+    },
+    {
+      "epoch": 0.44928,
+      "grad_norm": 0.3479366818185208,
+      "learning_rate": 3.1261351614834536e-05,
+      "loss": 0.2503,
+      "step": 702
+    },
+    {
+      "epoch": 0.44992,
+      "grad_norm": 0.29343107056101025,
+      "learning_rate": 3.121006543034307e-05,
+      "loss": 0.2737,
+      "step": 703
+    },
+    {
+      "epoch": 0.45056,
+      "grad_norm": 0.31093246539572245,
+      "learning_rate": 3.11587513773604e-05,
+      "loss": 0.2319,
+      "step": 704
+    },
+    {
+      "epoch": 0.4512,
+      "grad_norm": 0.3020599725750793,
+      "learning_rate": 3.1107409686165105e-05,
+      "loss": 0.263,
+      "step": 705
+    },
+    {
+      "epoch": 0.45184,
+      "grad_norm": 0.32507011230940397,
+      "learning_rate": 3.1056040587159856e-05,
+      "loss": 0.2384,
+      "step": 706
+    },
+    {
+      "epoch": 0.45248,
+      "grad_norm": 0.3158222849241388,
+      "learning_rate": 3.100464431087027e-05,
+      "loss": 0.2478,
+      "step": 707
+    },
+    {
+      "epoch": 0.45312,
+      "grad_norm": 0.30516321261828705,
+      "learning_rate": 3.095322108794396e-05,
+      "loss": 0.2454,
+      "step": 708
+    },
+    {
+      "epoch": 0.45376,
+      "grad_norm": 0.34842177058610063,
+      "learning_rate": 3.0901771149149434e-05,
+      "loss": 0.2379,
+      "step": 709
+    },
+    {
+      "epoch": 0.4544,
+      "grad_norm": 0.3181958205942106,
+      "learning_rate": 3.0850294725375105e-05,
+      "loss": 0.2584,
+      "step": 710
+    },
+    {
+      "epoch": 0.45504,
+      "grad_norm": 0.30214951305771537,
+      "learning_rate": 3.079879204762827e-05,
+      "loss": 0.2449,
+      "step": 711
+    },
+    {
+      "epoch": 0.45568,
+      "grad_norm": 0.39934101616880957,
+      "learning_rate": 3.074726334703399e-05,
+      "loss": 0.2491,
+      "step": 712
+    },
+    {
+      "epoch": 0.45632,
+      "grad_norm": 0.30199952276673303,
+      "learning_rate": 3.069570885483414e-05,
+      "loss": 0.2617,
+      "step": 713
+    },
+    {
+      "epoch": 0.45696,
+      "grad_norm": 0.3728506804152992,
+      "learning_rate": 3.064412880238636e-05,
+      "loss": 0.262,
+      "step": 714
+    },
+    {
+      "epoch": 0.4576,
+      "grad_norm": 0.3288139275419489,
+      "learning_rate": 3.059252342116293e-05,
+      "loss": 0.2597,
+      "step": 715
+    },
+    {
+      "epoch": 0.45824,
+      "grad_norm": 0.3000964979336861,
+      "learning_rate": 3.054089294274985e-05,
+      "loss": 0.2584,
+      "step": 716
+    },
+    {
+      "epoch": 0.45888,
+      "grad_norm": 0.35720062260151675,
+      "learning_rate": 3.048923759884573e-05,
+      "loss": 0.2483,
+      "step": 717
+    },
+    {
+      "epoch": 0.45952,
+      "grad_norm": 0.33170326945729556,
+      "learning_rate": 3.043755762126077e-05,
+      "loss": 0.2736,
+      "step": 718
+    },
+    {
+      "epoch": 0.46016,
+      "grad_norm": 0.36999753308441113,
+      "learning_rate": 3.038585324191571e-05,
+      "loss": 0.2667,
+      "step": 719
+    },
+    {
+      "epoch": 0.4608,
+      "grad_norm": 0.29794350528726254,
+      "learning_rate": 3.0334124692840815e-05,
+      "loss": 0.2687,
+      "step": 720
+    },
+    {
+      "epoch": 0.46144,
+      "grad_norm": 0.3742387734873763,
+      "learning_rate": 3.0282372206174785e-05,
+      "loss": 0.2637,
+      "step": 721
+    },
+    {
+      "epoch": 0.46208,
+      "grad_norm": 0.34276298414010076,
+      "learning_rate": 3.023059601416376e-05,
+      "loss": 0.2611,
+      "step": 722
+    },
+    {
+      "epoch": 0.46272,
+      "grad_norm": 0.30381902681771245,
+      "learning_rate": 3.017879634916027e-05,
+      "loss": 0.2527,
+      "step": 723
+    },
+    {
+      "epoch": 0.46336,
+      "grad_norm": 0.3047600134376854,
+      "learning_rate": 3.012697344362217e-05,
+      "loss": 0.2474,
+      "step": 724
+    },
+    {
+      "epoch": 0.464,
+      "grad_norm": 0.35393595518675103,
+      "learning_rate": 3.0075127530111606e-05,
+      "loss": 0.236,
+      "step": 725
+    },
+    {
+      "epoch": 0.46464,
+      "grad_norm": 0.3537780397489722,
+      "learning_rate": 3.0023258841294e-05,
+      "loss": 0.2563,
+      "step": 726
+    },
+    {
+      "epoch": 0.46528,
+      "grad_norm": 0.30556119496616985,
+      "learning_rate": 2.9971367609936945e-05,
+      "loss": 0.2649,
+      "step": 727
+    },
+    {
+      "epoch": 0.46592,
+      "grad_norm": 0.3703739862165785,
+      "learning_rate": 2.9919454068909218e-05,
+      "loss": 0.2361,
+      "step": 728
+    },
+    {
+      "epoch": 0.46656,
+      "grad_norm": 0.37165831132629096,
+      "learning_rate": 2.986751845117972e-05,
+      "loss": 0.2695,
+      "step": 729
+    },
+    {
+      "epoch": 0.4672,
+      "grad_norm": 0.33299604678686345,
+      "learning_rate": 2.981556098981641e-05,
+      "loss": 0.2967,
+      "step": 730
+    },
+    {
+      "epoch": 0.46784,
+      "grad_norm": 0.41951825830652595,
+      "learning_rate": 2.9763581917985274e-05,
+      "loss": 0.2769,
+      "step": 731
+    },
+    {
+      "epoch": 0.46848,
+      "grad_norm": 0.3237274558391439,
+      "learning_rate": 2.9711581468949295e-05,
+      "loss": 0.2707,
+      "step": 732
+    },
+    {
+      "epoch": 0.46912,
+      "grad_norm": 0.37367100802397996,
+      "learning_rate": 2.965955987606736e-05,
+      "loss": 0.286,
+      "step": 733
+    },
+    {
+      "epoch": 0.46976,
+      "grad_norm": 0.35637283945513776,
+      "learning_rate": 2.9607517372793268e-05,
+      "loss": 0.2622,
+      "step": 734
+    },
+    {
+      "epoch": 0.4704,
+      "grad_norm": 0.34632657954705776,
+      "learning_rate": 2.9555454192674638e-05,
+      "loss": 0.2632,
+      "step": 735
+    },
+    {
+      "epoch": 0.47104,
+      "grad_norm": 0.3497985509616476,
+      "learning_rate": 2.9503370569351895e-05,
+      "loss": 0.2737,
+      "step": 736
+    },
+    {
+      "epoch": 0.47168,
+      "grad_norm": 0.3421187398181242,
+      "learning_rate": 2.9451266736557202e-05,
+      "loss": 0.2718,
+      "step": 737
+    },
+    {
+      "epoch": 0.47232,
+      "grad_norm": 0.3151134822221513,
+      "learning_rate": 2.9399142928113403e-05,
+      "loss": 0.231,
+      "step": 738
+    },
+    {
+      "epoch": 0.47296,
+      "grad_norm": 0.362849073824411,
+      "learning_rate": 2.9346999377932992e-05,
+      "loss": 0.2772,
+      "step": 739
+    },
+    {
+      "epoch": 0.4736,
+      "grad_norm": 0.30664390636690253,
+      "learning_rate": 2.929483632001706e-05,
+      "loss": 0.2547,
+      "step": 740
+    },
+    {
+      "epoch": 0.47424,
+      "grad_norm": 0.3312210124408892,
+      "learning_rate": 2.9242653988454237e-05,
+      "loss": 0.2583,
+      "step": 741
+    },
+    {
+      "epoch": 0.47488,
+      "grad_norm": 0.35348522270732086,
+      "learning_rate": 2.919045261741965e-05,
+      "loss": 0.2509,
+      "step": 742
+    },
+    {
+      "epoch": 0.47552,
+      "grad_norm": 0.33437704916345523,
+      "learning_rate": 2.9138232441173867e-05,
+      "loss": 0.2513,
+      "step": 743
+    },
+    {
+      "epoch": 0.47616,
+      "grad_norm": 0.36971352614773284,
+      "learning_rate": 2.9085993694061853e-05,
+      "loss": 0.2564,
+      "step": 744
+    },
+    {
+      "epoch": 0.4768,
+      "grad_norm": 0.3459833141259172,
+      "learning_rate": 2.9033736610511887e-05,
+      "loss": 0.2754,
+      "step": 745
+    },
+    {
+      "epoch": 0.47744,
+      "grad_norm": 0.35066848228018166,
+      "learning_rate": 2.898146142503456e-05,
+      "loss": 0.2767,
+      "step": 746
+    },
+    {
+      "epoch": 0.47808,
+      "grad_norm": 0.3338800299797348,
+      "learning_rate": 2.8929168372221693e-05,
+      "loss": 0.2616,
+      "step": 747
+    },
+    {
+      "epoch": 0.47872,
+      "grad_norm": 0.3666614088595184,
+      "learning_rate": 2.88768576867453e-05,
+      "loss": 0.2697,
+      "step": 748
+    },
+    {
+      "epoch": 0.47936,
+      "grad_norm": 0.304245461090475,
+      "learning_rate": 2.8824529603356497e-05,
+      "loss": 0.2339,
+      "step": 749
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.36109036523976124,
+      "learning_rate": 2.8772184356884496e-05,
+      "loss": 0.273,
+      "step": 750
+    },
+    {
+      "epoch": 0.48064,
+      "grad_norm": 0.3130703716151443,
+      "learning_rate": 2.8719822182235516e-05,
+      "loss": 0.2682,
+      "step": 751
+    },
+    {
+      "epoch": 0.48128,
+      "grad_norm": 0.31968525350438826,
+      "learning_rate": 2.8667443314391762e-05,
+      "loss": 0.2401,
+      "step": 752
+    },
+    {
+      "epoch": 0.48192,
+      "grad_norm": 0.32458089104819654,
+      "learning_rate": 2.8615047988410337e-05,
+      "loss": 0.2547,
+      "step": 753
+    },
+    {
+      "epoch": 0.48256,
+      "grad_norm": 0.31059336468395815,
+      "learning_rate": 2.8562636439422202e-05,
+      "loss": 0.2556,
+      "step": 754
+    },
+    {
+      "epoch": 0.4832,
+      "grad_norm": 0.315018430853338,
+      "learning_rate": 2.8510208902631135e-05,
+      "loss": 0.265,
+      "step": 755
+    },
+    {
+      "epoch": 0.48384,
+      "grad_norm": 0.30233481647572924,
+      "learning_rate": 2.845776561331265e-05,
+      "loss": 0.2657,
+      "step": 756
+    },
+    {
+      "epoch": 0.48448,
+      "grad_norm": 0.335288114048365,
+      "learning_rate": 2.8405306806812947e-05,
+      "loss": 0.236,
+      "step": 757
+    },
+    {
+      "epoch": 0.48512,
+      "grad_norm": 0.34976033182981764,
+      "learning_rate": 2.835283271854788e-05,
+      "loss": 0.2771,
+      "step": 758
+    },
+    {
+      "epoch": 0.48576,
+      "grad_norm": 0.309720682513143,
+      "learning_rate": 2.8300343584001853e-05,
+      "loss": 0.2591,
+      "step": 759
+    },
+    {
+      "epoch": 0.4864,
+      "grad_norm": 0.3235847193221405,
+      "learning_rate": 2.8247839638726825e-05,
+      "loss": 0.2737,
+      "step": 760
+    },
+    {
+      "epoch": 0.48704,
+      "grad_norm": 0.3417557800653281,
+      "learning_rate": 2.8195321118341204e-05,
+      "loss": 0.2526,
+      "step": 761
+    },
+    {
+      "epoch": 0.48768,
+      "grad_norm": 0.31503431864321996,
+      "learning_rate": 2.8142788258528813e-05,
+      "loss": 0.2699,
+      "step": 762
+    },
+    {
+      "epoch": 0.48832,
+      "grad_norm": 0.316408002179215,
+      "learning_rate": 2.8090241295037804e-05,
+      "loss": 0.2683,
+      "step": 763
+    },
+    {
+      "epoch": 0.48896,
+      "grad_norm": 0.3117749940265806,
+      "learning_rate": 2.8037680463679648e-05,
+      "loss": 0.2598,
+      "step": 764
+    },
+    {
+      "epoch": 0.4896,
+      "grad_norm": 0.3316515216976238,
+      "learning_rate": 2.7985106000328033e-05,
+      "loss": 0.2539,
+      "step": 765
+    },
+    {
+      "epoch": 0.49024,
+      "grad_norm": 0.36924414386027793,
+      "learning_rate": 2.793251814091782e-05,
+      "loss": 0.2499,
+      "step": 766
+    },
+    {
+      "epoch": 0.49088,
+      "grad_norm": 0.3243715724846237,
+      "learning_rate": 2.787991712144401e-05,
+      "loss": 0.2734,
+      "step": 767
+    },
+    {
+      "epoch": 0.49152,
+      "grad_norm": 0.3528660186509523,
+      "learning_rate": 2.7827303177960638e-05,
+      "loss": 0.2801,
+      "step": 768
+    },
+    {
+      "epoch": 0.49216,
+      "grad_norm": 0.36076837190278555,
+      "learning_rate": 2.7774676546579737e-05,
+      "loss": 0.2633,
+      "step": 769
+    },
+    {
+      "epoch": 0.4928,
+      "grad_norm": 0.3395800283585683,
+      "learning_rate": 2.7722037463470297e-05,
+      "loss": 0.2434,
+      "step": 770
+    },
+    {
+      "epoch": 0.49344,
+      "grad_norm": 0.322259870407489,
+      "learning_rate": 2.7669386164857174e-05,
+      "loss": 0.2277,
+      "step": 771
+    },
+    {
+      "epoch": 0.49408,
+      "grad_norm": 0.34463620040148174,
+      "learning_rate": 2.7616722887020034e-05,
+      "loss": 0.242,
+      "step": 772
+    },
+    {
+      "epoch": 0.49472,
+      "grad_norm": 0.3520464282182829,
+      "learning_rate": 2.7564047866292336e-05,
+      "loss": 0.2466,
+      "step": 773
+    },
+    {
+      "epoch": 0.49536,
+      "grad_norm": 0.29511575157405057,
+      "learning_rate": 2.7511361339060193e-05,
+      "loss": 0.2724,
+      "step": 774
+    },
+    {
+      "epoch": 0.496,
+      "grad_norm": 0.3293691686223478,
+      "learning_rate": 2.7458663541761376e-05,
+      "loss": 0.2683,
+      "step": 775
+    },
+    {
+      "epoch": 0.49664,
+      "grad_norm": 0.34176113906772404,
+      "learning_rate": 2.740595471088424e-05,
+      "loss": 0.2459,
+      "step": 776
+    },
+    {
+      "epoch": 0.49728,
+      "grad_norm": 0.3207990032321656,
+      "learning_rate": 2.735323508296664e-05,
+      "loss": 0.2687,
+      "step": 777
+    },
+    {
+      "epoch": 0.49792,
+      "grad_norm": 0.36188064282747595,
+      "learning_rate": 2.7300504894594892e-05,
+      "loss": 0.2559,
+      "step": 778
+    },
+    {
+      "epoch": 0.49856,
+      "grad_norm": 0.30564878423967734,
+      "learning_rate": 2.7247764382402696e-05,
+      "loss": 0.2848,
+      "step": 779
+    },
+    {
+      "epoch": 0.4992,
+      "grad_norm": 0.32996121552702573,
+      "learning_rate": 2.7195013783070093e-05,
+      "loss": 0.2778,
+      "step": 780
+    },
+    {
+      "epoch": 0.49984,
+      "grad_norm": 0.3098476044334979,
+      "learning_rate": 2.7142253333322387e-05,
+      "loss": 0.262,
+      "step": 781
+    },
+    {
+      "epoch": 0.50048,
+      "grad_norm": 0.31033608411268765,
+      "learning_rate": 2.7089483269929074e-05,
+      "loss": 0.2656,
+      "step": 782
+    },
+    {
+      "epoch": 0.50112,
+      "grad_norm": 0.29147796052179703,
+      "learning_rate": 2.703670382970281e-05,
+      "loss": 0.2473,
+      "step": 783
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1562,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 261,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 358001810669568.0,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}