diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,7342 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.66816,
+  "eval_steps": 500,
+  "global_step": 1044,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.00064,
+      "grad_norm": 2.3657337561838188,
+      "learning_rate": 0.0,
+      "loss": 0.5764,
+      "step": 1
+    },
+    {
+      "epoch": 0.00128,
+      "grad_norm": 2.577238188562681,
+      "learning_rate": 6.329113924050633e-07,
+      "loss": 0.6154,
+      "step": 2
+    },
+    {
+      "epoch": 0.00192,
+      "grad_norm": 2.486400760684678,
+      "learning_rate": 1.2658227848101265e-06,
+      "loss": 0.5765,
+      "step": 3
+    },
+    {
+      "epoch": 0.00256,
+      "grad_norm": 2.427317708214583,
+      "learning_rate": 1.8987341772151901e-06,
+      "loss": 0.5898,
+      "step": 4
+    },
+    {
+      "epoch": 0.0032,
+      "grad_norm": 2.3689778807783104,
+      "learning_rate": 2.531645569620253e-06,
+      "loss": 0.5846,
+      "step": 5
+    },
+    {
+      "epoch": 0.00384,
+      "grad_norm": 2.533864042007203,
+      "learning_rate": 3.1645569620253167e-06,
+      "loss": 0.6002,
+      "step": 6
+    },
+    {
+      "epoch": 0.00448,
+      "grad_norm": 1.879595075845785,
+      "learning_rate": 3.7974683544303802e-06,
+      "loss": 0.5581,
+      "step": 7
+    },
+    {
+      "epoch": 0.00512,
+      "grad_norm": 2.056563531274739,
+      "learning_rate": 4.430379746835443e-06,
+      "loss": 0.5859,
+      "step": 8
+    },
+    {
+      "epoch": 0.00576,
+      "grad_norm": 1.4703684433180533,
+      "learning_rate": 5.063291139240506e-06,
+      "loss": 0.5479,
+      "step": 9
+    },
+    {
+      "epoch": 0.0064,
+      "grad_norm": 1.2614967629103893,
+      "learning_rate": 5.69620253164557e-06,
+      "loss": 0.5303,
+      "step": 10
+    },
+    {
+      "epoch": 0.00704,
+      "grad_norm": 1.1460599905324345,
+      "learning_rate": 6.329113924050633e-06,
+      "loss": 0.5568,
+      "step": 11
+    },
+    {
+      "epoch": 0.00768,
+      "grad_norm": 1.6204644872566638,
+      "learning_rate": 6.9620253164556965e-06,
+      "loss": 0.5111,
+      "step": 12
+    },
+    {
+      "epoch": 0.00832,
+      "grad_norm": 1.6361395875225335,
+      "learning_rate": 7.5949367088607605e-06,
+      "loss": 0.4678,
+      "step": 13
+    },
+    {
+      "epoch": 0.00896,
+      "grad_norm": 1.6402122162855226,
+      "learning_rate": 8.227848101265822e-06,
+      "loss": 0.4811,
+      "step": 14
+    },
+    {
+      "epoch": 0.0096,
+      "grad_norm": 1.2490639825476322,
+      "learning_rate": 8.860759493670886e-06,
+      "loss": 0.442,
+      "step": 15
+    },
+    {
+      "epoch": 0.01024,
+      "grad_norm": 1.0588915477376701,
+      "learning_rate": 9.49367088607595e-06,
+      "loss": 0.4263,
+      "step": 16
+    },
+    {
+      "epoch": 0.01088,
+      "grad_norm": 1.40604786962106,
+      "learning_rate": 1.0126582278481012e-05,
+      "loss": 0.4549,
+      "step": 17
+    },
+    {
+      "epoch": 0.01152,
+      "grad_norm": 1.242365760774633,
+      "learning_rate": 1.0759493670886076e-05,
+      "loss": 0.4531,
+      "step": 18
+    },
+    {
+      "epoch": 0.01216,
+      "grad_norm": 0.9396101260126887,
+      "learning_rate": 1.139240506329114e-05,
+      "loss": 0.415,
+      "step": 19
+    },
+    {
+      "epoch": 0.0128,
+      "grad_norm": 0.7578271337493023,
+      "learning_rate": 1.2025316455696203e-05,
+      "loss": 0.4268,
+      "step": 20
+    },
+    {
+      "epoch": 0.01344,
+      "grad_norm": 0.7671817679988738,
+      "learning_rate": 1.2658227848101267e-05,
+      "loss": 0.4314,
+      "step": 21
+    },
+    {
+      "epoch": 0.01408,
+      "grad_norm": 0.6494114557426621,
+      "learning_rate": 1.3291139240506329e-05,
+      "loss": 0.3666,
+      "step": 22
+    },
+    {
+      "epoch": 0.01472,
+      "grad_norm": 0.6806456497077759,
+      "learning_rate": 1.3924050632911393e-05,
+      "loss": 0.3823,
+      "step": 23
+    },
+    {
+      "epoch": 0.01536,
+      "grad_norm": 0.6595430154008608,
+      "learning_rate": 1.4556962025316457e-05,
+      "loss": 0.3529,
+      "step": 24
+    },
+    {
+      "epoch": 0.016,
+      "grad_norm": 0.7689071362735744,
+      "learning_rate": 1.5189873417721521e-05,
+      "loss": 0.3932,
+      "step": 25
+    },
+    {
+      "epoch": 0.01664,
+      "grad_norm": 0.6854685982052705,
+      "learning_rate": 1.5822784810126583e-05,
+      "loss": 0.3738,
+      "step": 26
+    },
+    {
+      "epoch": 0.01728,
+      "grad_norm": 0.558852717411362,
+      "learning_rate": 1.6455696202531644e-05,
+      "loss": 0.3708,
+      "step": 27
+    },
+    {
+      "epoch": 0.01792,
+      "grad_norm": 0.49876042518765007,
+      "learning_rate": 1.7088607594936708e-05,
+      "loss": 0.3696,
+      "step": 28
+    },
+    {
+      "epoch": 0.01856,
+      "grad_norm": 0.5271567613623678,
+      "learning_rate": 1.7721518987341772e-05,
+      "loss": 0.3558,
+      "step": 29
+    },
+    {
+      "epoch": 0.0192,
+      "grad_norm": 0.5978920045933569,
+      "learning_rate": 1.8354430379746836e-05,
+      "loss": 0.3827,
+      "step": 30
+    },
+    {
+      "epoch": 0.01984,
+      "grad_norm": 0.4975313988318536,
+      "learning_rate": 1.89873417721519e-05,
+      "loss": 0.3772,
+      "step": 31
+    },
+    {
+      "epoch": 0.02048,
+      "grad_norm": 0.5766330443378189,
+      "learning_rate": 1.962025316455696e-05,
+      "loss": 0.3832,
+      "step": 32
+    },
+    {
+      "epoch": 0.02112,
+      "grad_norm": 0.5567055560538855,
+      "learning_rate": 2.0253164556962025e-05,
+      "loss": 0.3477,
+      "step": 33
+    },
+    {
+      "epoch": 0.02176,
+      "grad_norm": 0.46871971492473,
+      "learning_rate": 2.088607594936709e-05,
+      "loss": 0.3604,
+      "step": 34
+    },
+    {
+      "epoch": 0.0224,
+      "grad_norm": 0.43316810660778116,
+      "learning_rate": 2.1518987341772153e-05,
+      "loss": 0.3592,
+      "step": 35
+    },
+    {
+      "epoch": 0.02304,
+      "grad_norm": 0.4312186195231089,
+      "learning_rate": 2.2151898734177217e-05,
+      "loss": 0.3711,
+      "step": 36
+    },
+    {
+      "epoch": 0.02368,
+      "grad_norm": 0.46951838290882675,
+      "learning_rate": 2.278481012658228e-05,
+      "loss": 0.3618,
+      "step": 37
+    },
+    {
+      "epoch": 0.02432,
+      "grad_norm": 0.468516818608629,
+      "learning_rate": 2.341772151898734e-05,
+      "loss": 0.3578,
+      "step": 38
+    },
+    {
+      "epoch": 0.02496,
+      "grad_norm": 0.45909911923620816,
+      "learning_rate": 2.4050632911392405e-05,
+      "loss": 0.3521,
+      "step": 39
+    },
+    {
+      "epoch": 0.0256,
+      "grad_norm": 0.36223572387755915,
+      "learning_rate": 2.468354430379747e-05,
+      "loss": 0.3439,
+      "step": 40
+    },
+    {
+      "epoch": 0.02624,
+      "grad_norm": 0.41920026035277985,
+      "learning_rate": 2.5316455696202533e-05,
+      "loss": 0.3052,
+      "step": 41
+    },
+    {
+      "epoch": 0.02688,
+      "grad_norm": 0.46410873807902825,
+      "learning_rate": 2.5949367088607597e-05,
+      "loss": 0.3324,
+      "step": 42
+    },
+    {
+      "epoch": 0.02752,
+      "grad_norm": 0.44804958203514283,
+      "learning_rate": 2.6582278481012658e-05,
+      "loss": 0.322,
+      "step": 43
+    },
+    {
+      "epoch": 0.02816,
+      "grad_norm": 0.4408934955100543,
+      "learning_rate": 2.7215189873417722e-05,
+      "loss": 0.3399,
+      "step": 44
+    },
+    {
+      "epoch": 0.0288,
+      "grad_norm": 0.4536362310706478,
+      "learning_rate": 2.7848101265822786e-05,
+      "loss": 0.3508,
+      "step": 45
+    },
+    {
+      "epoch": 0.02944,
+      "grad_norm": 0.4735438789842479,
+      "learning_rate": 2.848101265822785e-05,
+      "loss": 0.3431,
+      "step": 46
+    },
+    {
+      "epoch": 0.03008,
+      "grad_norm": 0.3960360490220664,
+      "learning_rate": 2.9113924050632914e-05,
+      "loss": 0.3358,
+      "step": 47
+    },
+    {
+      "epoch": 0.03072,
+      "grad_norm": 0.4449245975582831,
+      "learning_rate": 2.9746835443037974e-05,
+      "loss": 0.3472,
+      "step": 48
+    },
+    {
+      "epoch": 0.03136,
+      "grad_norm": 0.41598379019590886,
+      "learning_rate": 3.0379746835443042e-05,
+      "loss": 0.3295,
+      "step": 49
+    },
+    {
+      "epoch": 0.032,
+      "grad_norm": 0.39770652562313213,
+      "learning_rate": 3.10126582278481e-05,
+      "loss": 0.339,
+      "step": 50
+    },
+    {
+      "epoch": 0.03264,
+      "grad_norm": 0.4036450937079526,
+      "learning_rate": 3.1645569620253167e-05,
+      "loss": 0.326,
+      "step": 51
+    },
+    {
+      "epoch": 0.03328,
+      "grad_norm": 0.43082778666471144,
+      "learning_rate": 3.227848101265823e-05,
+      "loss": 0.3258,
+      "step": 52
+    },
+    {
+      "epoch": 0.03392,
+      "grad_norm": 0.402275292978643,
+      "learning_rate": 3.291139240506329e-05,
+      "loss": 0.2905,
+      "step": 53
+    },
+    {
+      "epoch": 0.03456,
+      "grad_norm": 0.43612618786182106,
+      "learning_rate": 3.354430379746836e-05,
+      "loss": 0.334,
+      "step": 54
+    },
+    {
+      "epoch": 0.0352,
+      "grad_norm": 0.45984837714015897,
+      "learning_rate": 3.4177215189873416e-05,
+      "loss": 0.3283,
+      "step": 55
+    },
+    {
+      "epoch": 0.03584,
+      "grad_norm": 0.4484461315187549,
+      "learning_rate": 3.4810126582278487e-05,
+      "loss": 0.3277,
+      "step": 56
+    },
+    {
+      "epoch": 0.03648,
+      "grad_norm": 0.4052074034352622,
+      "learning_rate": 3.5443037974683544e-05,
+      "loss": 0.3271,
+      "step": 57
+    },
+    {
+      "epoch": 0.03712,
+      "grad_norm": 0.46620186767863914,
+      "learning_rate": 3.607594936708861e-05,
+      "loss": 0.3464,
+      "step": 58
+    },
+    {
+      "epoch": 0.03776,
+      "grad_norm": 0.42593292833096424,
+      "learning_rate": 3.670886075949367e-05,
+      "loss": 0.3331,
+      "step": 59
+    },
+    {
+      "epoch": 0.0384,
+      "grad_norm": 0.4815687673225032,
+      "learning_rate": 3.7341772151898736e-05,
+      "loss": 0.3269,
+      "step": 60
+    },
+    {
+      "epoch": 0.03904,
+      "grad_norm": 0.455342196474994,
+      "learning_rate": 3.79746835443038e-05,
+      "loss": 0.3343,
+      "step": 61
+    },
+    {
+      "epoch": 0.03968,
+      "grad_norm": 0.4316078425366754,
+      "learning_rate": 3.8607594936708864e-05,
+      "loss": 0.3145,
+      "step": 62
+    },
+    {
+      "epoch": 0.04032,
+      "grad_norm": 0.4604468007808017,
+      "learning_rate": 3.924050632911392e-05,
+      "loss": 0.3206,
+      "step": 63
+    },
+    {
+      "epoch": 0.04096,
+      "grad_norm": 0.516259514156323,
+      "learning_rate": 3.987341772151899e-05,
+      "loss": 0.3538,
+      "step": 64
+    },
+    {
+      "epoch": 0.0416,
+      "grad_norm": 0.3911206186305256,
+      "learning_rate": 4.050632911392405e-05,
+      "loss": 0.3116,
+      "step": 65
+    },
+    {
+      "epoch": 0.04224,
+      "grad_norm": 0.5059680673369054,
+      "learning_rate": 4.113924050632912e-05,
+      "loss": 0.3461,
+      "step": 66
+    },
+    {
+      "epoch": 0.04288,
+      "grad_norm": 0.4477077725372915,
+      "learning_rate": 4.177215189873418e-05,
+      "loss": 0.3142,
+      "step": 67
+    },
+    {
+      "epoch": 0.04352,
+      "grad_norm": 0.4568181894979607,
+      "learning_rate": 4.240506329113924e-05,
+      "loss": 0.3258,
+      "step": 68
+    },
+    {
+      "epoch": 0.04416,
+      "grad_norm": 0.436426112682463,
+      "learning_rate": 4.3037974683544305e-05,
+      "loss": 0.2942,
+      "step": 69
+    },
+    {
+      "epoch": 0.0448,
+      "grad_norm": 0.4130845172287253,
+      "learning_rate": 4.367088607594937e-05,
+      "loss": 0.3293,
+      "step": 70
+    },
+    {
+      "epoch": 0.04544,
+      "grad_norm": 0.4907470169969436,
+      "learning_rate": 4.430379746835443e-05,
+      "loss": 0.3152,
+      "step": 71
+    },
+    {
+      "epoch": 0.04608,
+      "grad_norm": 0.44557023947548996,
+      "learning_rate": 4.49367088607595e-05,
+      "loss": 0.323,
+      "step": 72
+    },
+    {
+      "epoch": 0.04672,
+      "grad_norm": 0.43347032964452825,
+      "learning_rate": 4.556962025316456e-05,
+      "loss": 0.3132,
+      "step": 73
+    },
+    {
+      "epoch": 0.04736,
+      "grad_norm": 0.41315778348357357,
+      "learning_rate": 4.6202531645569625e-05,
+      "loss": 0.3076,
+      "step": 74
+    },
+    {
+      "epoch": 0.048,
+      "grad_norm": 0.38689299616228223,
+      "learning_rate": 4.683544303797468e-05,
+      "loss": 0.2971,
+      "step": 75
+    },
+    {
+      "epoch": 0.04864,
+      "grad_norm": 0.6649393785343345,
+      "learning_rate": 4.7468354430379746e-05,
+      "loss": 0.3196,
+      "step": 76
+    },
+    {
+      "epoch": 0.04928,
+      "grad_norm": 0.4741860571119842,
+      "learning_rate": 4.810126582278481e-05,
+      "loss": 0.3261,
+      "step": 77
+    },
+    {
+      "epoch": 0.04992,
+      "grad_norm": 0.39850926668653464,
+      "learning_rate": 4.8734177215189874e-05,
+      "loss": 0.2909,
+      "step": 78
+    },
+    {
+      "epoch": 0.05056,
+      "grad_norm": 0.45993342960946926,
+      "learning_rate": 4.936708860759494e-05,
+      "loss": 0.3232,
+      "step": 79
+    },
+    {
+      "epoch": 0.0512,
+      "grad_norm": 0.41758744126936076,
+      "learning_rate": 5e-05,
+      "loss": 0.3138,
+      "step": 80
+    },
+    {
+      "epoch": 0.05184,
+      "grad_norm": 0.4190137054884969,
+      "learning_rate": 4.999994390459423e-05,
+      "loss": 0.324,
+      "step": 81
+    },
+    {
+      "epoch": 0.05248,
+      "grad_norm": 0.38007532443802017,
+      "learning_rate": 4.999977561862863e-05,
+      "loss": 0.3167,
+      "step": 82
+    },
+    {
+      "epoch": 0.05312,
+      "grad_norm": 0.3894813419979866,
+      "learning_rate": 4.999949514285842e-05,
+      "loss": 0.3169,
+      "step": 83
+    },
+    {
+      "epoch": 0.05376,
+      "grad_norm": 0.4646671073219221,
+      "learning_rate": 4.999910247854227e-05,
+      "loss": 0.3295,
+      "step": 84
+    },
+    {
+      "epoch": 0.0544,
+      "grad_norm": 0.46440352293136283,
+      "learning_rate": 4.99985976274423e-05,
+      "loss": 0.3259,
+      "step": 85
+    },
+    {
+      "epoch": 0.05504,
+      "grad_norm": 0.4754742191954697,
+      "learning_rate": 4.9997980591824126e-05,
+      "loss": 0.3073,
+      "step": 86
+    },
+    {
+      "epoch": 0.05568,
+      "grad_norm": 0.4102331568111904,
+      "learning_rate": 4.999725137445674e-05,
+      "loss": 0.3179,
+      "step": 87
+    },
+    {
+      "epoch": 0.05632,
+      "grad_norm": 0.49865407563428993,
+      "learning_rate": 4.999640997861264e-05,
+      "loss": 0.3198,
+      "step": 88
+    },
+    {
+      "epoch": 0.05696,
+      "grad_norm": 0.450995150648752,
+      "learning_rate": 4.999545640806766e-05,
+      "loss": 0.3215,
+      "step": 89
+    },
+    {
+      "epoch": 0.0576,
+      "grad_norm": 0.42715082345561,
+      "learning_rate": 4.999439066710111e-05,
+      "loss": 0.3406,
+      "step": 90
+    },
+    {
+      "epoch": 0.05824,
+      "grad_norm": 0.47007665924261693,
+      "learning_rate": 4.999321276049563e-05,
+      "loss": 0.3169,
+      "step": 91
+    },
+    {
+      "epoch": 0.05888,
+      "grad_norm": 0.37692477648671757,
+      "learning_rate": 4.9991922693537226e-05,
+      "loss": 0.3052,
+      "step": 92
+    },
+    {
+      "epoch": 0.05952,
+      "grad_norm": 0.5254380682221013,
+      "learning_rate": 4.9990520472015246e-05,
+      "loss": 0.2928,
+      "step": 93
+    },
+    {
+      "epoch": 0.06016,
+      "grad_norm": 0.4524991571079602,
+      "learning_rate": 4.998900610222236e-05,
+      "loss": 0.3035,
+      "step": 94
+    },
+    {
+      "epoch": 0.0608,
+      "grad_norm": 0.39419071585883453,
+      "learning_rate": 4.998737959095449e-05,
+      "loss": 0.2764,
+      "step": 95
+    },
+    {
+      "epoch": 0.06144,
+      "grad_norm": 0.44021580096048935,
+      "learning_rate": 4.998564094551081e-05,
+      "loss": 0.3137,
+      "step": 96
+    },
+    {
+      "epoch": 0.06208,
+      "grad_norm": 0.4397078772822271,
+      "learning_rate": 4.998379017369375e-05,
+      "loss": 0.3092,
+      "step": 97
+    },
+    {
+      "epoch": 0.06272,
+      "grad_norm": 0.4612518337175716,
+      "learning_rate": 4.9981827283808855e-05,
+      "loss": 0.3065,
+      "step": 98
+    },
+    {
+      "epoch": 0.06336,
+      "grad_norm": 0.4491229684653282,
+      "learning_rate": 4.997975228466489e-05,
+      "loss": 0.3327,
+      "step": 99
+    },
+    {
+      "epoch": 0.064,
+      "grad_norm": 0.40909998580765855,
+      "learning_rate": 4.997756518557368e-05,
+      "loss": 0.2787,
+      "step": 100
+    },
+    {
+      "epoch": 0.06464,
+      "grad_norm": 0.3785712105730604,
+      "learning_rate": 4.997526599635011e-05,
+      "loss": 0.2837,
+      "step": 101
+    },
+    {
+      "epoch": 0.06528,
+      "grad_norm": 0.5009245567404541,
+      "learning_rate": 4.997285472731211e-05,
+      "loss": 0.3325,
+      "step": 102
+    },
+    {
+      "epoch": 0.06592,
+      "grad_norm": 0.41940515716785753,
+      "learning_rate": 4.997033138928056e-05,
+      "loss": 0.3126,
+      "step": 103
+    },
+    {
+      "epoch": 0.06656,
+      "grad_norm": 0.4356254674524969,
+      "learning_rate": 4.996769599357927e-05,
+      "loss": 0.3079,
+      "step": 104
+    },
+    {
+      "epoch": 0.0672,
+      "grad_norm": 0.4485794513852225,
+      "learning_rate": 4.996494855203493e-05,
+      "loss": 0.3227,
+      "step": 105
+    },
+    {
+      "epoch": 0.06784,
+      "grad_norm": 0.5339271359603054,
+      "learning_rate": 4.996208907697706e-05,
+      "loss": 0.3197,
+      "step": 106
+    },
+    {
+      "epoch": 0.06848,
+      "grad_norm": 0.47913200680485823,
+      "learning_rate": 4.995911758123791e-05,
+      "loss": 0.2829,
+      "step": 107
+    },
+    {
+      "epoch": 0.06912,
+      "grad_norm": 0.48619705540043057,
+      "learning_rate": 4.995603407815249e-05,
+      "loss": 0.295,
+      "step": 108
+    },
+    {
+      "epoch": 0.06976,
+      "grad_norm": 0.44791381540090275,
+      "learning_rate": 4.9952838581558406e-05,
+      "loss": 0.3009,
+      "step": 109
+    },
+    {
+      "epoch": 0.0704,
+      "grad_norm": 0.4217427510368746,
+      "learning_rate": 4.994953110579588e-05,
+      "loss": 0.3124,
+      "step": 110
+    },
+    {
+      "epoch": 0.07104,
+      "grad_norm": 0.44605995933104997,
+      "learning_rate": 4.9946111665707645e-05,
+      "loss": 0.3232,
+      "step": 111
+    },
+    {
+      "epoch": 0.07168,
+      "grad_norm": 0.3580173680896881,
+      "learning_rate": 4.99425802766389e-05,
+      "loss": 0.288,
+      "step": 112
+    },
+    {
+      "epoch": 0.07232,
+      "grad_norm": 0.4281198427243427,
+      "learning_rate": 4.993893695443721e-05,
+      "loss": 0.3008,
+      "step": 113
+    },
+    {
+      "epoch": 0.07296,
+      "grad_norm": 0.3748012845273175,
+      "learning_rate": 4.993518171545247e-05,
+      "loss": 0.2898,
+      "step": 114
+    },
+    {
+      "epoch": 0.0736,
+      "grad_norm": 0.41133247941906176,
+      "learning_rate": 4.993131457653682e-05,
+      "loss": 0.3007,
+      "step": 115
+    },
+    {
+      "epoch": 0.07424,
+      "grad_norm": 0.4049279526177947,
+      "learning_rate": 4.992733555504454e-05,
+      "loss": 0.3156,
+      "step": 116
+    },
+    {
+      "epoch": 0.07488,
+      "grad_norm": 0.3907854290589919,
+      "learning_rate": 4.9923244668832025e-05,
+      "loss": 0.305,
+      "step": 117
+    },
+    {
+      "epoch": 0.07552,
+      "grad_norm": 3.3672297158831523,
+      "learning_rate": 4.991904193625768e-05,
+      "loss": 0.3099,
+      "step": 118
+    },
+    {
+      "epoch": 0.07616,
+      "grad_norm": 0.45896342700641674,
+      "learning_rate": 4.991472737618181e-05,
+      "loss": 0.2766,
+      "step": 119
+    },
+    {
+      "epoch": 0.0768,
+      "grad_norm": 0.41249483633287526,
+      "learning_rate": 4.991030100796659e-05,
+      "loss": 0.3026,
+      "step": 120
+    },
+    {
+      "epoch": 0.07744,
+      "grad_norm": 0.39435147330028136,
+      "learning_rate": 4.9905762851475915e-05,
+      "loss": 0.3151,
+      "step": 121
+    },
+    {
+      "epoch": 0.07808,
+      "grad_norm": 0.38174217860012233,
+      "learning_rate": 4.990111292707537e-05,
+      "loss": 0.3132,
+      "step": 122
+    },
+    {
+      "epoch": 0.07872,
+      "grad_norm": 0.4106762400890441,
+      "learning_rate": 4.989635125563211e-05,
+      "loss": 0.2883,
+      "step": 123
+    },
+    {
+      "epoch": 0.07936,
+      "grad_norm": 0.3777494709551966,
+      "learning_rate": 4.989147785851477e-05,
+      "loss": 0.3075,
+      "step": 124
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.3812281839176885,
+      "learning_rate": 4.988649275759335e-05,
+      "loss": 0.3173,
+      "step": 125
+    },
+    {
+      "epoch": 0.08064,
+      "grad_norm": 0.3800442807387135,
+      "learning_rate": 4.988139597523916e-05,
+      "loss": 0.2897,
+      "step": 126
+    },
+    {
+      "epoch": 0.08128,
+      "grad_norm": 0.4101363073522896,
+      "learning_rate": 4.98761875343247e-05,
+      "loss": 0.2822,
+      "step": 127
+    },
+    {
+      "epoch": 0.08192,
+      "grad_norm": 0.3515414786001439,
+      "learning_rate": 4.987086745822352e-05,
+      "loss": 0.2831,
+      "step": 128
+    },
+    {
+      "epoch": 0.08256,
+      "grad_norm": 0.38143533758711873,
+      "learning_rate": 4.986543577081017e-05,
+      "loss": 0.3193,
+      "step": 129
+    },
+    {
+      "epoch": 0.0832,
+      "grad_norm": 0.44066201212578954,
+      "learning_rate": 4.985989249646007e-05,
+      "loss": 0.3006,
+      "step": 130
+    },
+    {
+      "epoch": 0.08384,
+      "grad_norm": 0.3699946189372801,
+      "learning_rate": 4.985423766004939e-05,
+      "loss": 0.309,
+      "step": 131
+    },
+    {
+      "epoch": 0.08448,
+      "grad_norm": 0.4732003041900607,
+      "learning_rate": 4.9848471286954965e-05,
+      "loss": 0.3253,
+      "step": 132
+    },
+    {
+      "epoch": 0.08512,
+      "grad_norm": 0.35724915771132804,
+      "learning_rate": 4.984259340305416e-05,
+      "loss": 0.2876,
+      "step": 133
+    },
+    {
+      "epoch": 0.08576,
+      "grad_norm": 0.39925687072469707,
+      "learning_rate": 4.9836604034724753e-05,
+      "loss": 0.2906,
+      "step": 134
+    },
+    {
+      "epoch": 0.0864,
+      "grad_norm": 0.43582991089095635,
+      "learning_rate": 4.9830503208844835e-05,
+      "loss": 0.2928,
+      "step": 135
+    },
+    {
+      "epoch": 0.08704,
+      "grad_norm": 0.3638802742085539,
+      "learning_rate": 4.982429095279266e-05,
+      "loss": 0.2924,
+      "step": 136
+    },
+    {
+      "epoch": 0.08768,
+      "grad_norm": 0.44414884826796713,
+      "learning_rate": 4.981796729444655e-05,
+      "loss": 0.3225,
+      "step": 137
+    },
+    {
+      "epoch": 0.08832,
+      "grad_norm": 0.4179539121719692,
+      "learning_rate": 4.981153226218477e-05,
+      "loss": 0.3067,
+      "step": 138
+    },
+    {
+      "epoch": 0.08896,
+      "grad_norm": 0.3518602463362222,
+      "learning_rate": 4.9804985884885366e-05,
+      "loss": 0.2854,
+      "step": 139
+    },
+    {
+      "epoch": 0.0896,
+      "grad_norm": 0.4013030661247676,
+      "learning_rate": 4.979832819192608e-05,
+      "loss": 0.3113,
+      "step": 140
+    },
+    {
+      "epoch": 0.09024,
+      "grad_norm": 0.3700912839771966,
+      "learning_rate": 4.97915592131842e-05,
+      "loss": 0.2873,
+      "step": 141
+    },
+    {
+      "epoch": 0.09088,
+      "grad_norm": 0.41429834446700914,
+      "learning_rate": 4.97846789790364e-05,
+      "loss": 0.3064,
+      "step": 142
+    },
+    {
+      "epoch": 0.09152,
+      "grad_norm": 0.3743350290044281,
+      "learning_rate": 4.977768752035865e-05,
+      "loss": 0.311,
+      "step": 143
+    },
+    {
+      "epoch": 0.09216,
+      "grad_norm": 0.4436631708270473,
+      "learning_rate": 4.9770584868526046e-05,
+      "loss": 0.3105,
+      "step": 144
+    },
+    {
+      "epoch": 0.0928,
+      "grad_norm": 0.4341903366251428,
+      "learning_rate": 4.9763371055412685e-05,
+      "loss": 0.307,
+      "step": 145
+    },
+    {
+      "epoch": 0.09344,
+      "grad_norm": 0.4147022361169306,
+      "learning_rate": 4.9756046113391496e-05,
+      "loss": 0.3288,
+      "step": 146
+    },
+    {
+      "epoch": 0.09408,
+      "grad_norm": 0.3884970059580215,
+      "learning_rate": 4.974861007533413e-05,
+      "loss": 0.2847,
+      "step": 147
+    },
+    {
+      "epoch": 0.09472,
+      "grad_norm": 0.4258271028958422,
+      "learning_rate": 4.97410629746108e-05,
+      "loss": 0.2908,
+      "step": 148
+    },
+    {
+      "epoch": 0.09536,
+      "grad_norm": 0.41527352386736543,
+      "learning_rate": 4.9733404845090116e-05,
+      "loss": 0.2828,
+      "step": 149
+    },
+    {
+      "epoch": 0.096,
+      "grad_norm": 0.41352693469232554,
+      "learning_rate": 4.972563572113894e-05,
+      "loss": 0.2895,
+      "step": 150
+    },
+    {
+      "epoch": 0.09664,
+      "grad_norm": 0.4391732737160089,
+      "learning_rate": 4.971775563762226e-05,
+      "loss": 0.2908,
+      "step": 151
+    },
+    {
+      "epoch": 0.09728,
+      "grad_norm": 0.42637768991672836,
+      "learning_rate": 4.970976462990299e-05,
+      "loss": 0.3103,
+      "step": 152
+    },
+    {
+      "epoch": 0.09792,
+      "grad_norm": 0.38123935717217666,
+      "learning_rate": 4.970166273384183e-05,
+      "loss": 0.2859,
+      "step": 153
+    },
+    {
+      "epoch": 0.09856,
+      "grad_norm": 0.3432072306652636,
+      "learning_rate": 4.969344998579712e-05,
+      "loss": 0.2653,
+      "step": 154
+    },
+    {
+      "epoch": 0.0992,
+      "grad_norm": 0.38378318854854493,
+      "learning_rate": 4.9685126422624644e-05,
+      "loss": 0.3135,
+      "step": 155
+    },
+    {
+      "epoch": 0.09984,
+      "grad_norm": 0.3871767364493639,
+      "learning_rate": 4.96766920816775e-05,
+      "loss": 0.3129,
+      "step": 156
+    },
+    {
+      "epoch": 0.10048,
+      "grad_norm": 0.361665333999729,
+      "learning_rate": 4.966814700080592e-05,
+      "loss": 0.3175,
+      "step": 157
+    },
+    {
+      "epoch": 0.10112,
+      "grad_norm": 0.394466639141269,
+      "learning_rate": 4.965949121835707e-05,
+      "loss": 0.333,
+      "step": 158
+    },
+    {
+      "epoch": 0.10176,
+      "grad_norm": 0.396292343881353,
+      "learning_rate": 4.965072477317493e-05,
+      "loss": 0.3084,
+      "step": 159
+    },
+    {
+      "epoch": 0.1024,
+      "grad_norm": 0.3468981265806776,
+      "learning_rate": 4.964184770460007e-05,
+      "loss": 0.2927,
+      "step": 160
+    },
+    {
+      "epoch": 0.10304,
+      "grad_norm": 0.39865168519337385,
+      "learning_rate": 4.963286005246953e-05,
+      "loss": 0.2815,
+      "step": 161
+    },
+    {
+      "epoch": 0.10368,
+      "grad_norm": 0.37451640530098157,
+      "learning_rate": 4.9623761857116594e-05,
+      "loss": 0.3188,
+      "step": 162
+    },
+    {
+      "epoch": 0.10432,
+      "grad_norm": 0.37483826123841657,
+      "learning_rate": 4.961455315937059e-05,
+      "loss": 0.2813,
+      "step": 163
+    },
+    {
+      "epoch": 0.10496,
+      "grad_norm": 0.33130533329924106,
+      "learning_rate": 4.960523400055679e-05,
+      "loss": 0.2924,
+      "step": 164
+    },
+    {
+      "epoch": 0.1056,
+      "grad_norm": 0.3364997696428208,
+      "learning_rate": 4.959580442249615e-05,
+      "loss": 0.3001,
+      "step": 165
+    },
+    {
+      "epoch": 0.10624,
+      "grad_norm": 0.3183485250122132,
+      "learning_rate": 4.958626446750515e-05,
+      "loss": 0.3195,
+      "step": 166
+    },
+    {
+      "epoch": 0.10688,
+      "grad_norm": 0.3417933668875517,
+      "learning_rate": 4.9576614178395595e-05,
+      "loss": 0.2991,
+      "step": 167
+    },
+    {
+      "epoch": 0.10752,
+      "grad_norm": 0.3487109852934047,
+      "learning_rate": 4.956685359847445e-05,
+      "loss": 0.2959,
+      "step": 168
+    },
+    {
+      "epoch": 0.10816,
+      "grad_norm": 0.31315058374358634,
+      "learning_rate": 4.95569827715436e-05,
+      "loss": 0.2799,
+      "step": 169
+    },
+    {
+      "epoch": 0.1088,
+      "grad_norm": 0.35913320358975465,
+      "learning_rate": 4.9547001741899693e-05,
+      "loss": 0.297,
+      "step": 170
+    },
+    {
+      "epoch": 0.10944,
+      "grad_norm": 0.3331478821334656,
+      "learning_rate": 4.953691055433391e-05,
+      "loss": 0.2901,
+      "step": 171
+    },
+    {
+      "epoch": 0.11008,
+      "grad_norm": 0.39224813006559756,
+      "learning_rate": 4.952670925413182e-05,
+      "loss": 0.3026,
+      "step": 172
+    },
+    {
+      "epoch": 0.11072,
+      "grad_norm": 0.3480414792318722,
+      "learning_rate": 4.951639788707307e-05,
+      "loss": 0.2845,
+      "step": 173
+    },
+    {
+      "epoch": 0.11136,
+      "grad_norm": 0.34970614984057585,
+      "learning_rate": 4.950597649943132e-05,
+      "loss": 0.2775,
+      "step": 174
+    },
+    {
+      "epoch": 0.112,
+      "grad_norm": 0.347207358588287,
+      "learning_rate": 4.9495445137973905e-05,
+      "loss": 0.2923,
+      "step": 175
+    },
+    {
+      "epoch": 0.11264,
+      "grad_norm": 0.34884090954360897,
+      "learning_rate": 4.948480384996172e-05,
+      "loss": 0.2862,
+      "step": 176
+    },
+    {
+      "epoch": 0.11328,
+      "grad_norm": 0.32979237586574495,
+      "learning_rate": 4.947405268314894e-05,
+      "loss": 0.2983,
+      "step": 177
+    },
+    {
+      "epoch": 0.11392,
+      "grad_norm": 0.43991275496703847,
+      "learning_rate": 4.9463191685782865e-05,
+      "loss": 0.3068,
+      "step": 178
+    },
+    {
+      "epoch": 0.11456,
+      "grad_norm": 0.3039302314073942,
+      "learning_rate": 4.945222090660364e-05,
+      "loss": 0.2841,
+      "step": 179
+    },
+    {
+      "epoch": 0.1152,
+      "grad_norm": 0.38589337384034356,
+      "learning_rate": 4.944114039484411e-05,
+      "loss": 0.3091,
+      "step": 180
+    },
+    {
+      "epoch": 0.11584,
+      "grad_norm": 0.34105198100381573,
+      "learning_rate": 4.942995020022954e-05,
+      "loss": 0.3228,
+      "step": 181
+    },
+    {
+      "epoch": 0.11648,
+      "grad_norm": 0.34780611263986955,
+      "learning_rate": 4.9418650372977384e-05,
+      "loss": 0.2874,
+      "step": 182
+    },
+    {
+      "epoch": 0.11712,
+      "grad_norm": 0.32541445621708004,
+      "learning_rate": 4.940724096379714e-05,
+      "loss": 0.282,
+      "step": 183
+    },
+    {
+      "epoch": 0.11776,
+      "grad_norm": 0.3082894269244735,
+      "learning_rate": 4.9395722023890035e-05,
+      "loss": 0.2799,
+      "step": 184
+    },
+    {
+      "epoch": 0.1184,
+      "grad_norm": 0.32812745331187065,
+      "learning_rate": 4.938409360494883e-05,
+      "loss": 0.2827,
+      "step": 185
+    },
+    {
+      "epoch": 0.11904,
+      "grad_norm": 0.34755898651365463,
+      "learning_rate": 4.937235575915761e-05,
+      "loss": 0.2917,
+      "step": 186
+    },
+    {
+      "epoch": 0.11968,
+      "grad_norm": 0.3413504764617461,
+      "learning_rate": 4.93605085391915e-05,
+      "loss": 0.2918,
+      "step": 187
+    },
+    {
+      "epoch": 0.12032,
+      "grad_norm": 0.3705985820501455,
+      "learning_rate": 4.934855199821647e-05,
+      "loss": 0.2826,
+      "step": 188
+    },
+    {
+      "epoch": 0.12096,
+      "grad_norm": 0.3414000814514884,
+      "learning_rate": 4.93364861898891e-05,
+      "loss": 0.276,
+      "step": 189
+    },
+    {
+      "epoch": 0.1216,
+      "grad_norm": 0.3589734824395955,
+      "learning_rate": 4.932431116835628e-05,
+      "loss": 0.2959,
+      "step": 190
+    },
+    {
+      "epoch": 0.12224,
+      "grad_norm": 0.345670488984043,
+      "learning_rate": 4.931202698825503e-05,
+      "loss": 0.281,
+      "step": 191
+    },
+    {
+      "epoch": 0.12288,
+      "grad_norm": 0.3235692312545686,
+      "learning_rate": 4.929963370471226e-05,
+      "loss": 0.2886,
+      "step": 192
+    },
+    {
+      "epoch": 0.12352,
+      "grad_norm": 0.3034427958798936,
+      "learning_rate": 4.928713137334445e-05,
+      "loss": 0.2812,
+      "step": 193
+    },
+    {
+      "epoch": 0.12416,
+      "grad_norm": 0.34712496586477004,
+      "learning_rate": 4.927452005025748e-05,
+      "loss": 0.3018,
+      "step": 194
+    },
+    {
+      "epoch": 0.1248,
+      "grad_norm": 0.3222467927855389,
+      "learning_rate": 4.926179979204633e-05,
+      "loss": 0.3023,
+      "step": 195
+    },
+    {
+      "epoch": 0.12544,
+      "grad_norm": 0.3243348951725544,
+      "learning_rate": 4.9248970655794836e-05,
+      "loss": 0.2917,
+      "step": 196
+    },
+    {
+      "epoch": 0.12608,
+      "grad_norm": 0.3096795363251828,
+      "learning_rate": 4.923603269907545e-05,
+      "loss": 0.2737,
+      "step": 197
+    },
+    {
+      "epoch": 0.12672,
+      "grad_norm": 0.31152060487499617,
+      "learning_rate": 4.922298597994897e-05,
+      "loss": 0.2828,
+      "step": 198
+    },
+    {
+      "epoch": 0.12736,
+      "grad_norm": 0.3405465489598549,
+      "learning_rate": 4.920983055696428e-05,
+      "loss": 0.2927,
+      "step": 199
+    },
+    {
+      "epoch": 0.128,
+      "grad_norm": 0.3423946912166355,
+      "learning_rate": 4.919656648915807e-05,
+      "loss": 0.2835,
+      "step": 200
+    },
+    {
+      "epoch": 0.12864,
+      "grad_norm": 0.3464727040800733,
+      "learning_rate": 4.918319383605462e-05,
+      "loss": 0.2575,
+      "step": 201
+    },
+    {
+      "epoch": 0.12928,
+      "grad_norm": 0.3620779326329196,
+      "learning_rate": 4.916971265766547e-05,
+      "loss": 0.2944,
+      "step": 202
+    },
+    {
+      "epoch": 0.12992,
+      "grad_norm": 0.3699616777622434,
+      "learning_rate": 4.915612301448919e-05,
+      "loss": 0.2942,
+      "step": 203
+    },
+    {
+      "epoch": 0.13056,
+      "grad_norm": 0.38798124559835534,
+      "learning_rate": 4.914242496751111e-05,
+      "loss": 0.2882,
+      "step": 204
+    },
+    {
+      "epoch": 0.1312,
+      "grad_norm": 0.4306348775923099,
+      "learning_rate": 4.912861857820303e-05,
+      "loss": 0.3026,
+      "step": 205
+    },
+    {
+      "epoch": 0.13184,
+      "grad_norm": 0.37275998073469474,
+      "learning_rate": 4.911470390852295e-05,
+      "loss": 0.3008,
+      "step": 206
+    },
+    {
+      "epoch": 0.13248,
+      "grad_norm": 0.33076859723493407,
+      "learning_rate": 4.91006810209148e-05,
+      "loss": 0.2881,
+      "step": 207
+    },
+    {
+      "epoch": 0.13312,
+      "grad_norm": 0.38285708110841127,
+      "learning_rate": 4.9086549978308125e-05,
+      "loss": 0.299,
+      "step": 208
+    },
+    {
+      "epoch": 0.13376,
+      "grad_norm": 0.3130925296009032,
+      "learning_rate": 4.9072310844117875e-05,
+      "loss": 0.2714,
+      "step": 209
+    },
+    {
+      "epoch": 0.1344,
+      "grad_norm": 0.3556404363171034,
+      "learning_rate": 4.9057963682244045e-05,
+      "loss": 0.291,
+      "step": 210
+    },
+    {
+      "epoch": 0.13504,
+      "grad_norm": 0.33759319834147206,
+      "learning_rate": 4.904350855707142e-05,
+      "loss": 0.2868,
+      "step": 211
+    },
+    {
+      "epoch": 0.13568,
+      "grad_norm": 0.3171722280640699,
+      "learning_rate": 4.9028945533469276e-05,
+      "loss": 0.2646,
+      "step": 212
+    },
+    {
+      "epoch": 0.13632,
+      "grad_norm": 0.39240914087101075,
+      "learning_rate": 4.9014274676791134e-05,
+      "loss": 0.302,
+      "step": 213
+    },
+    {
+      "epoch": 0.13696,
+      "grad_norm": 0.34184628116877047,
+      "learning_rate": 4.899949605287439e-05,
+      "loss": 0.2792,
+      "step": 214
+    },
+    {
+      "epoch": 0.1376,
+      "grad_norm": 0.3700664187254164,
+      "learning_rate": 4.898460972804009e-05,
+      "loss": 0.3088,
+      "step": 215
+    },
+    {
+      "epoch": 0.13824,
+      "grad_norm": 0.36421921549881453,
+      "learning_rate": 4.896961576909258e-05,
+      "loss": 0.2808,
+      "step": 216
+    },
+    {
+      "epoch": 0.13888,
+      "grad_norm": 0.3565240796039332,
+      "learning_rate": 4.895451424331924e-05,
+      "loss": 0.2714,
+      "step": 217
+    },
+    {
+      "epoch": 0.13952,
+      "grad_norm": 0.3396297883795516,
+      "learning_rate": 4.893930521849016e-05,
+      "loss": 0.267,
+      "step": 218
+    },
+    {
+      "epoch": 0.14016,
+      "grad_norm": 0.39278607376772506,
+      "learning_rate": 4.892398876285785e-05,
+      "loss": 0.2896,
+      "step": 219
+    },
+    {
+      "epoch": 0.1408,
+      "grad_norm": 0.38437399542076783,
+      "learning_rate": 4.890856494515696e-05,
+      "loss": 0.2875,
+      "step": 220
+    },
+    {
+      "epoch": 0.14144,
+      "grad_norm": 0.37920192384172824,
+      "learning_rate": 4.889303383460389e-05,
+      "loss": 0.2806,
+      "step": 221
+    },
+    {
+      "epoch": 0.14208,
+      "grad_norm": 0.35504283495161315,
+      "learning_rate": 4.887739550089656e-05,
+      "loss": 0.2781,
+      "step": 222
+    },
+    {
+      "epoch": 0.14272,
+      "grad_norm": 0.3866107761723795,
+      "learning_rate": 4.886165001421408e-05,
+      "loss": 0.3091,
+      "step": 223
+    },
+    {
+      "epoch": 0.14336,
+      "grad_norm": 0.30439690273605985,
+      "learning_rate": 4.884579744521638e-05,
+      "loss": 0.2842,
+      "step": 224
+    },
+    {
+      "epoch": 0.144,
+      "grad_norm": 0.3551417352689114,
+      "learning_rate": 4.8829837865043984e-05,
+      "loss": 0.2909,
+      "step": 225
+    },
+    {
+      "epoch": 0.14464,
+      "grad_norm": 0.309111612445454,
+      "learning_rate": 4.881377134531762e-05,
+      "loss": 0.268,
+      "step": 226
+    },
+    {
+      "epoch": 0.14528,
+      "grad_norm": 0.3282993602648903,
+      "learning_rate": 4.8797597958137914e-05,
+      "loss": 0.28,
+      "step": 227
+    },
+    {
+      "epoch": 0.14592,
+      "grad_norm": 0.3581890513495606,
+      "learning_rate": 4.878131777608509e-05,
+      "loss": 0.2842,
+      "step": 228
+    },
+    {
+      "epoch": 0.14656,
+      "grad_norm": 0.32315603560305584,
+      "learning_rate": 4.876493087221862e-05,
+      "loss": 0.2961,
+      "step": 229
+    },
+    {
+      "epoch": 0.1472,
+      "grad_norm": 0.33067996529729515,
+      "learning_rate": 4.87484373200769e-05,
+      "loss": 0.2891,
+      "step": 230
+    },
+    {
+      "epoch": 0.14784,
+      "grad_norm": 0.3178546016599421,
+      "learning_rate": 4.8731837193676944e-05,
+      "loss": 0.2781,
+      "step": 231
+    },
+    {
+      "epoch": 0.14848,
+      "grad_norm": 0.3414289060571217,
+      "learning_rate": 4.871513056751401e-05,
+      "loss": 0.2983,
+      "step": 232
+    },
+    {
+      "epoch": 0.14912,
+      "grad_norm": 0.3420355616767074,
+      "learning_rate": 4.8698317516561295e-05,
+      "loss": 0.2987,
+      "step": 233
+    },
+    {
+      "epoch": 0.14976,
+      "grad_norm": 0.347117902288678,
+      "learning_rate": 4.8681398116269596e-05,
+      "loss": 0.2857,
+      "step": 234
+    },
+    {
+      "epoch": 0.1504,
+      "grad_norm": 0.31648517023684664,
+      "learning_rate": 4.8664372442566954e-05,
+      "loss": 0.2921,
+      "step": 235
+    },
+    {
+      "epoch": 0.15104,
+      "grad_norm": 0.3308519821164363,
+      "learning_rate": 4.864724057185834e-05,
+      "loss": 0.2672,
+      "step": 236
+    },
+    {
+      "epoch": 0.15168,
+      "grad_norm": 0.31208238851833453,
+      "learning_rate": 4.86300025810253e-05,
+      "loss": 0.2798,
+      "step": 237
+    },
+    {
+      "epoch": 0.15232,
+      "grad_norm": 0.2945629591271816,
+      "learning_rate": 4.861265854742559e-05,
+      "loss": 0.2732,
+      "step": 238
+    },
+    {
+      "epoch": 0.15296,
+      "grad_norm": 0.3383764371143183,
+      "learning_rate": 4.859520854889287e-05,
+      "loss": 0.2923,
+      "step": 239
+    },
+    {
+      "epoch": 0.1536,
+      "grad_norm": 0.2935231514068793,
+      "learning_rate": 4.857765266373631e-05,
+      "loss": 0.2945,
+      "step": 240
+    },
+    {
+      "epoch": 0.15424,
+      "grad_norm": 0.3265200869108488,
+      "learning_rate": 4.855999097074027e-05,
+      "loss": 0.2925,
+      "step": 241
+    },
+    {
+      "epoch": 0.15488,
+      "grad_norm": 0.301565323998506,
+      "learning_rate": 4.8542223549163946e-05,
+      "loss": 0.271,
+      "step": 242
+    },
+    {
+      "epoch": 0.15552,
+      "grad_norm": 0.3121729061732836,
+      "learning_rate": 4.8524350478740985e-05,
+      "loss": 0.266,
+      "step": 243
+    },
+    {
+      "epoch": 0.15616,
+      "grad_norm": 0.3612945591439241,
+      "learning_rate": 4.850637183967917e-05,
+      "loss": 0.2825,
+      "step": 244
+    },
+    {
+      "epoch": 0.1568,
+      "grad_norm": 0.32526112140366076,
+      "learning_rate": 4.848828771266002e-05,
+      "loss": 0.2875,
+      "step": 245
+    },
+    {
+      "epoch": 0.15744,
+      "grad_norm": 0.3724034080949958,
+      "learning_rate": 4.847009817883844e-05,
+      "loss": 0.2829,
+      "step": 246
+    },
+    {
+      "epoch": 0.15808,
+      "grad_norm": 0.31233652215083263,
+      "learning_rate": 4.845180331984239e-05,
+      "loss": 0.2713,
+      "step": 247
+    },
+    {
+      "epoch": 0.15872,
+      "grad_norm": 0.3480570732169317,
+      "learning_rate": 4.843340321777246e-05,
+      "loss": 0.2641,
+      "step": 248
+    },
+    {
+      "epoch": 0.15936,
+      "grad_norm": 0.320447396199795,
+      "learning_rate": 4.8414897955201556e-05,
+      "loss": 0.261,
+      "step": 249
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.30011124568595027,
+      "learning_rate": 4.8396287615174484e-05,
+      "loss": 0.2566,
+      "step": 250
+    },
+    {
+      "epoch": 0.16064,
+      "grad_norm": 0.3416591207698499,
+      "learning_rate": 4.837757228120762e-05,
+      "loss": 0.2886,
+      "step": 251
+    },
+    {
+      "epoch": 0.16128,
+      "grad_norm": 0.3347844484474534,
+      "learning_rate": 4.835875203728849e-05,
+      "loss": 0.2546,
+      "step": 252
+    },
+    {
+      "epoch": 0.16192,
+      "grad_norm": 0.3805038941147735,
+      "learning_rate": 4.8339826967875443e-05,
+      "loss": 0.2867,
+      "step": 253
+    },
+    {
+      "epoch": 0.16256,
+      "grad_norm": 0.36899369410322225,
+      "learning_rate": 4.832079715789724e-05,
+      "loss": 0.2969,
+      "step": 254
+    },
+    {
+      "epoch": 0.1632,
+      "grad_norm": 0.3671046083988311,
+      "learning_rate": 4.830166269275267e-05,
+      "loss": 0.2903,
+      "step": 255
+    },
+    {
+      "epoch": 0.16384,
+      "grad_norm": 0.3268166037896424,
+      "learning_rate": 4.828242365831016e-05,
+      "loss": 0.2744,
+      "step": 256
+    },
+    {
+      "epoch": 0.16448,
+      "grad_norm": 0.38930768380996184,
+      "learning_rate": 4.8263080140907455e-05,
+      "loss": 0.2898,
+      "step": 257
+    },
+    {
+      "epoch": 0.16512,
+      "grad_norm": 0.3527164726419632,
+      "learning_rate": 4.824363222735114e-05,
+      "loss": 0.2751,
+      "step": 258
+    },
+    {
+      "epoch": 0.16576,
+      "grad_norm": 0.3831694490194351,
+      "learning_rate": 4.82240800049163e-05,
+      "loss": 0.2797,
+      "step": 259
+    },
+    {
+      "epoch": 0.1664,
+      "grad_norm": 0.39066664017067054,
+      "learning_rate": 4.820442356134612e-05,
+      "loss": 0.2716,
+      "step": 260
+    },
+    {
+      "epoch": 0.16704,
+      "grad_norm": 0.36528936400377576,
+      "learning_rate": 4.8184662984851505e-05,
+      "loss": 0.3001,
+      "step": 261
+    },
+    {
+      "epoch": 0.16768,
+      "grad_norm": 0.3720797744652676,
+      "learning_rate": 4.816479836411065e-05,
+      "loss": 0.2819,
+      "step": 262
+    },
+    {
+      "epoch": 0.16832,
+      "grad_norm": 0.4121620669424255,
+      "learning_rate": 4.8144829788268674e-05,
+      "loss": 0.2951,
+      "step": 263
+    },
+    {
+      "epoch": 0.16896,
+      "grad_norm": 0.3687003642746144,
+      "learning_rate": 4.8124757346937214e-05,
+      "loss": 0.2637,
+      "step": 264
+    },
+    {
+      "epoch": 0.1696,
+      "grad_norm": 0.3380727248902014,
+      "learning_rate": 4.810458113019401e-05,
+      "loss": 0.2749,
+      "step": 265
+    },
+    {
+      "epoch": 0.17024,
+      "grad_norm": 0.3295868074806808,
+      "learning_rate": 4.8084301228582485e-05,
+      "loss": 0.2579,
+      "step": 266
+    },
+    {
+      "epoch": 0.17088,
+      "grad_norm": 0.37764930926091167,
+      "learning_rate": 4.8063917733111406e-05,
+      "loss": 0.3095,
+      "step": 267
+    },
+    {
+      "epoch": 0.17152,
+      "grad_norm": 0.3403308904612826,
+      "learning_rate": 4.8043430735254404e-05,
+      "loss": 0.2806,
+      "step": 268
+    },
+    {
+      "epoch": 0.17216,
+      "grad_norm": 0.36741715154564875,
+      "learning_rate": 4.8022840326949595e-05,
+      "loss": 0.2806,
+      "step": 269
+    },
+    {
+      "epoch": 0.1728,
+      "grad_norm": 0.3915206701251456,
+      "learning_rate": 4.800214660059916e-05,
+      "loss": 0.2789,
+      "step": 270
+    },
+    {
+      "epoch": 0.17344,
+      "grad_norm": 0.3396593508328933,
+      "learning_rate": 4.798134964906894e-05,
+      "loss": 0.2865,
+      "step": 271
+    },
+    {
+      "epoch": 0.17408,
+      "grad_norm": 0.3711393932873003,
+      "learning_rate": 4.7960449565688e-05,
+      "loss": 0.263,
+      "step": 272
+    },
+    {
+      "epoch": 0.17472,
+      "grad_norm": 0.3481318617474482,
+      "learning_rate": 4.793944644424826e-05,
+      "loss": 0.285,
+      "step": 273
+    },
+    {
+      "epoch": 0.17536,
+      "grad_norm": 0.36766456982327766,
+      "learning_rate": 4.7918340379003985e-05,
+      "loss": 0.2872,
+      "step": 274
+    },
+    {
+      "epoch": 0.176,
+      "grad_norm": 0.3322741327391994,
+      "learning_rate": 4.7897131464671444e-05,
+      "loss": 0.2997,
+      "step": 275
+    },
+    {
+      "epoch": 0.17664,
+      "grad_norm": 0.3658059319194032,
+      "learning_rate": 4.787581979642845e-05,
+      "loss": 0.2557,
+      "step": 276
+    },
+    {
+      "epoch": 0.17728,
+      "grad_norm": 0.3273286462475089,
+      "learning_rate": 4.7854405469913944e-05,
+      "loss": 0.273,
+      "step": 277
+    },
+    {
+      "epoch": 0.17792,
+      "grad_norm": 0.3111759107193226,
+      "learning_rate": 4.7832888581227545e-05,
+      "loss": 0.2756,
+      "step": 278
+    },
+    {
+      "epoch": 0.17856,
+      "grad_norm": 0.3412050823331732,
+      "learning_rate": 4.7811269226929135e-05,
+      "loss": 0.2788,
+      "step": 279
+    },
+    {
+      "epoch": 0.1792,
+      "grad_norm": 0.34902342482932425,
+      "learning_rate": 4.7789547504038443e-05,
+      "loss": 0.269,
+      "step": 280
+    },
+    {
+      "epoch": 0.17984,
+      "grad_norm": 0.33701110337193746,
+      "learning_rate": 4.776772351003458e-05,
+      "loss": 0.2801,
+      "step": 281
+    },
+    {
+      "epoch": 0.18048,
+      "grad_norm": 0.34838767763355505,
+      "learning_rate": 4.774579734285559e-05,
+      "loss": 0.2872,
+      "step": 282
+    },
+    {
+      "epoch": 0.18112,
+      "grad_norm": 0.3268119544044272,
+      "learning_rate": 4.772376910089808e-05,
+      "loss": 0.2781,
+      "step": 283
+    },
+    {
+      "epoch": 0.18176,
+      "grad_norm": 0.3260714138226823,
+      "learning_rate": 4.770163888301668e-05,
+      "loss": 0.2696,
+      "step": 284
+    },
+    {
+      "epoch": 0.1824,
+      "grad_norm": 0.3263806440232551,
+      "learning_rate": 4.7679406788523685e-05,
+      "loss": 0.277,
+      "step": 285
+    },
+    {
+      "epoch": 0.18304,
+      "grad_norm": 0.3048596244321711,
+      "learning_rate": 4.765707291718856e-05,
+      "loss": 0.2748,
+      "step": 286
+    },
+    {
+      "epoch": 0.18368,
+      "grad_norm": 0.33173051881106563,
+      "learning_rate": 4.7634637369237526e-05,
+      "loss": 0.2765,
+      "step": 287
+    },
+    {
+      "epoch": 0.18432,
+      "grad_norm": 0.34287198001014474,
+      "learning_rate": 4.7612100245353045e-05,
+      "loss": 0.2865,
+      "step": 288
+    },
+    {
+      "epoch": 0.18496,
+      "grad_norm": 0.3273490271016689,
+      "learning_rate": 4.758946164667347e-05,
+      "loss": 0.2905,
+      "step": 289
+    },
+    {
+      "epoch": 0.1856,
+      "grad_norm": 0.3587640664765252,
+      "learning_rate": 4.7566721674792515e-05,
+      "loss": 0.2776,
+      "step": 290
+    },
+    {
+      "epoch": 0.18624,
+      "grad_norm": 0.31221104587356835,
+      "learning_rate": 4.7543880431758805e-05,
+      "loss": 0.2779,
+      "step": 291
+    },
+    {
+      "epoch": 0.18688,
+      "grad_norm": 0.30655584340609615,
+      "learning_rate": 4.752093802007543e-05,
+      "loss": 0.3008,
+      "step": 292
+    },
+    {
+      "epoch": 0.18752,
+      "grad_norm": 0.3017825263683308,
+      "learning_rate": 4.7497894542699534e-05,
+      "loss": 0.2607,
+      "step": 293
+    },
+    {
+      "epoch": 0.18816,
+      "grad_norm": 0.3262092930986735,
+      "learning_rate": 4.747475010304175e-05,
+      "loss": 0.2792,
+      "step": 294
+    },
+    {
+      "epoch": 0.1888,
+      "grad_norm": 0.2931602837859308,
+      "learning_rate": 4.745150480496583e-05,
+      "loss": 0.2965,
+      "step": 295
+    },
+    {
+      "epoch": 0.18944,
+      "grad_norm": 0.31963993149008735,
+      "learning_rate": 4.742815875278811e-05,
+      "loss": 0.2955,
+      "step": 296
+    },
+    {
+      "epoch": 0.19008,
+      "grad_norm": 0.3325849928268489,
+      "learning_rate": 4.7404712051277115e-05,
+      "loss": 0.2552,
+      "step": 297
+    },
+    {
+      "epoch": 0.19072,
+      "grad_norm": 0.34547053393459853,
+      "learning_rate": 4.738116480565301e-05,
+      "loss": 0.2944,
+      "step": 298
+    },
+    {
+      "epoch": 0.19136,
+      "grad_norm": 0.3085593082553946,
+      "learning_rate": 4.7357517121587183e-05,
+      "loss": 0.2778,
+      "step": 299
+    },
+    {
+      "epoch": 0.192,
+      "grad_norm": 0.31353055101488136,
+      "learning_rate": 4.7333769105201735e-05,
+      "loss": 0.2429,
+      "step": 300
+    },
+    {
+      "epoch": 0.19264,
+      "grad_norm": 0.31858274733561903,
+      "learning_rate": 4.730992086306906e-05,
+      "loss": 0.2829,
+      "step": 301
+    },
+    {
+      "epoch": 0.19328,
+      "grad_norm": 0.30709940331794794,
+      "learning_rate": 4.728597250221129e-05,
+      "loss": 0.2829,
+      "step": 302
+    },
+    {
+      "epoch": 0.19392,
+      "grad_norm": 0.32244914176650474,
+      "learning_rate": 4.726192413009986e-05,
+      "loss": 0.2951,
+      "step": 303
+    },
+    {
+      "epoch": 0.19456,
+      "grad_norm": 0.3227593884616251,
+      "learning_rate": 4.723777585465503e-05,
+      "loss": 0.2926,
+      "step": 304
+    },
+    {
+      "epoch": 0.1952,
+      "grad_norm": 0.3283415364290106,
+      "learning_rate": 4.7213527784245395e-05,
+      "loss": 0.2761,
+      "step": 305
+    },
+    {
+      "epoch": 0.19584,
+      "grad_norm": 0.3015938328335281,
+      "learning_rate": 4.718918002768738e-05,
+      "loss": 0.2762,
+      "step": 306
+    },
+    {
+      "epoch": 0.19648,
+      "grad_norm": 0.29949328240066386,
+      "learning_rate": 4.716473269424476e-05,
+      "loss": 0.2832,
+      "step": 307
+    },
+    {
+      "epoch": 0.19712,
+      "grad_norm": 0.3035478548839789,
+      "learning_rate": 4.714018589362818e-05,
+      "loss": 0.3102,
+      "step": 308
+    },
+    {
+      "epoch": 0.19776,
+      "grad_norm": 0.27507230052380166,
+      "learning_rate": 4.7115539735994666e-05,
+      "loss": 0.2732,
+      "step": 309
+    },
+    {
+      "epoch": 0.1984,
+      "grad_norm": 0.3311299038598881,
+      "learning_rate": 4.709079433194712e-05,
+      "loss": 0.2766,
+      "step": 310
+    },
+    {
+      "epoch": 0.19904,
+      "grad_norm": 0.37577322279526276,
+      "learning_rate": 4.7065949792533804e-05,
+      "loss": 0.2562,
+      "step": 311
+    },
+    {
+      "epoch": 0.19968,
+      "grad_norm": 0.38289115818879554,
+      "learning_rate": 4.70410062292479e-05,
+      "loss": 0.2932,
+      "step": 312
+    },
+    {
+      "epoch": 0.20032,
+      "grad_norm": 0.30727362022005966,
+      "learning_rate": 4.701596375402694e-05,
+      "loss": 0.2918,
+      "step": 313
+    },
+    {
+      "epoch": 0.20096,
+      "grad_norm": 0.5661606421554864,
+      "learning_rate": 4.699082247925234e-05,
+      "loss": 0.3,
+      "step": 314
+    },
+    {
+      "epoch": 0.2016,
+      "grad_norm": 0.3276893965611005,
+      "learning_rate": 4.6965582517748923e-05,
+      "loss": 0.2932,
+      "step": 315
+    },
+    {
+      "epoch": 0.20224,
+      "grad_norm": 0.3022909698773883,
+      "learning_rate": 4.694024398278434e-05,
+      "loss": 0.2889,
+      "step": 316
+    },
+    {
+      "epoch": 0.20288,
+      "grad_norm": 0.33933254250674033,
+      "learning_rate": 4.691480698806864e-05,
+      "loss": 0.3001,
+      "step": 317
+    },
+    {
+      "epoch": 0.20352,
+      "grad_norm": 0.2824673534404734,
+      "learning_rate": 4.6889271647753694e-05,
+      "loss": 0.2638,
+      "step": 318
+    },
+    {
+      "epoch": 0.20416,
+      "grad_norm": 0.3226521796493832,
+      "learning_rate": 4.686363807643272e-05,
+      "loss": 0.2565,
+      "step": 319
+    },
+    {
+      "epoch": 0.2048,
+      "grad_norm": 0.32498560490845607,
+      "learning_rate": 4.683790638913979e-05,
+      "loss": 0.2999,
+      "step": 320
+    },
+    {
+      "epoch": 0.20544,
+      "grad_norm": 0.3014466505484606,
+      "learning_rate": 4.681207670134923e-05,
+      "loss": 0.2852,
+      "step": 321
+    },
+    {
+      "epoch": 0.20608,
+      "grad_norm": 0.29666058601485484,
+      "learning_rate": 4.6786149128975196e-05,
+      "loss": 0.2779,
+      "step": 322
+    },
+    {
+      "epoch": 0.20672,
+      "grad_norm": 0.30132965740306905,
+      "learning_rate": 4.6760123788371114e-05,
+      "loss": 0.2824,
+      "step": 323
+    },
+    {
+      "epoch": 0.20736,
+      "grad_norm": 0.31127978685588065,
+      "learning_rate": 4.6734000796329125e-05,
+      "loss": 0.2847,
+      "step": 324
+    },
+    {
+      "epoch": 0.208,
+      "grad_norm": 0.29058258855782604,
+      "learning_rate": 4.670778027007964e-05,
+      "loss": 0.2971,
+      "step": 325
+    },
+    {
+      "epoch": 0.20864,
+      "grad_norm": 0.36291904279248527,
+      "learning_rate": 4.6681462327290725e-05,
+      "loss": 0.2675,
+      "step": 326
+    },
+    {
+      "epoch": 0.20928,
+      "grad_norm": 0.3079346495438994,
+      "learning_rate": 4.6655047086067636e-05,
+      "loss": 0.2788,
+      "step": 327
+    },
+    {
+      "epoch": 0.20992,
+      "grad_norm": 0.31244911680380855,
+      "learning_rate": 4.662853466495229e-05,
+      "loss": 0.271,
+      "step": 328
+    },
+    {
+      "epoch": 0.21056,
+      "grad_norm": 0.2892209625632523,
+      "learning_rate": 4.660192518292266e-05,
+      "loss": 0.27,
+      "step": 329
+    },
+    {
+      "epoch": 0.2112,
+      "grad_norm": 0.31939496738148104,
+      "learning_rate": 4.6575218759392336e-05,
+      "loss": 0.2698,
+      "step": 330
+    },
+    {
+      "epoch": 0.21184,
+      "grad_norm": 0.2982207659870602,
+      "learning_rate": 4.654841551420993e-05,
+      "loss": 0.2588,
+      "step": 331
+    },
+    {
+      "epoch": 0.21248,
+      "grad_norm": 0.3016620161751348,
+      "learning_rate": 4.6521515567658557e-05,
+      "loss": 0.2674,
+      "step": 332
+    },
+    {
+      "epoch": 0.21312,
+      "grad_norm": 0.33353697727099657,
+      "learning_rate": 4.6494519040455275e-05,
+      "loss": 0.2811,
+      "step": 333
+    },
+    {
+      "epoch": 0.21376,
+      "grad_norm": 0.2950010715216195,
+      "learning_rate": 4.6467426053750596e-05,
+      "loss": 0.2629,
+      "step": 334
+    },
+    {
+      "epoch": 0.2144,
+      "grad_norm": 0.3033055413044891,
+      "learning_rate": 4.644023672912788e-05,
+      "loss": 0.2774,
+      "step": 335
+    },
+    {
+      "epoch": 0.21504,
+      "grad_norm": 0.3008702820111445,
+      "learning_rate": 4.641295118860281e-05,
+      "loss": 0.2864,
+      "step": 336
+    },
+    {
+      "epoch": 0.21568,
+      "grad_norm": 0.26879685510955026,
+      "learning_rate": 4.638556955462289e-05,
+      "loss": 0.2597,
+      "step": 337
+    },
+    {
+      "epoch": 0.21632,
+      "grad_norm": 0.3532124447567259,
+      "learning_rate": 4.635809195006681e-05,
+      "loss": 0.2897,
+      "step": 338
+    },
+    {
+      "epoch": 0.21696,
+      "grad_norm": 0.31251314292012317,
+      "learning_rate": 4.633051849824396e-05,
+      "loss": 0.2571,
+      "step": 339
+    },
+    {
+      "epoch": 0.2176,
+      "grad_norm": 0.34552541821349914,
+      "learning_rate": 4.630284932289386e-05,
+      "loss": 0.2835,
+      "step": 340
+    },
+    {
+      "epoch": 0.21824,
+      "grad_norm": 0.32731628909864535,
+      "learning_rate": 4.6275084548185605e-05,
+      "loss": 0.2939,
+      "step": 341
+    },
+    {
+      "epoch": 0.21888,
+      "grad_norm": 0.32841586223389657,
+      "learning_rate": 4.6247224298717294e-05,
+      "loss": 0.2547,
+      "step": 342
+    },
+    {
+      "epoch": 0.21952,
+      "grad_norm": 0.30505092898032665,
+      "learning_rate": 4.621926869951549e-05,
+      "loss": 0.2452,
+      "step": 343
+    },
+    {
+      "epoch": 0.22016,
+      "grad_norm": 0.410868336628806,
+      "learning_rate": 4.619121787603464e-05,
+      "loss": 0.2757,
+      "step": 344
+    },
+    {
+      "epoch": 0.2208,
+      "grad_norm": 0.30578794712173124,
+      "learning_rate": 4.6163071954156546e-05,
+      "loss": 0.2756,
+      "step": 345
+    },
+    {
+      "epoch": 0.22144,
+      "grad_norm": 0.3180588945483863,
+      "learning_rate": 4.6134831060189743e-05,
+      "loss": 0.2817,
+      "step": 346
+    },
+    {
+      "epoch": 0.22208,
+      "grad_norm": 0.306137612796177,
+      "learning_rate": 4.610649532086899e-05,
+      "loss": 0.2727,
+      "step": 347
+    },
+    {
+      "epoch": 0.22272,
+      "grad_norm": 0.33496927815217786,
+      "learning_rate": 4.607806486335468e-05,
+      "loss": 0.2885,
+      "step": 348
+    },
+    {
+      "epoch": 0.22336,
+      "grad_norm": 0.3034108440418061,
+      "learning_rate": 4.604953981523225e-05,
+      "loss": 0.2875,
+      "step": 349
+    },
+    {
+      "epoch": 0.224,
+      "grad_norm": 0.30668021026111714,
+      "learning_rate": 4.602092030451162e-05,
+      "loss": 0.2695,
+      "step": 350
+    },
+    {
+      "epoch": 0.22464,
+      "grad_norm": 0.33000274101961585,
+      "learning_rate": 4.599220645962665e-05,
+      "loss": 0.285,
+      "step": 351
+    },
+    {
+      "epoch": 0.22528,
+      "grad_norm": 0.29230078568083745,
+      "learning_rate": 4.596339840943452e-05,
+      "loss": 0.2678,
+      "step": 352
+    },
+    {
+      "epoch": 0.22592,
+      "grad_norm": 0.29979922728628944,
+      "learning_rate": 4.5934496283215164e-05,
+      "loss": 0.2669,
+      "step": 353
+    },
+    {
+      "epoch": 0.22656,
+      "grad_norm": 0.2754120967287155,
+      "learning_rate": 4.590550021067072e-05,
+      "loss": 0.2573,
+      "step": 354
+    },
+    {
+      "epoch": 0.2272,
+      "grad_norm": 0.31544489465932246,
+      "learning_rate": 4.587641032192488e-05,
+      "loss": 0.2661,
+      "step": 355
+    },
+    {
+      "epoch": 0.22784,
+      "grad_norm": 0.3153056590013689,
+      "learning_rate": 4.58472267475224e-05,
+      "loss": 0.2808,
+      "step": 356
+    },
+    {
+      "epoch": 0.22848,
+      "grad_norm": 0.3283844022151439,
+      "learning_rate": 4.5817949618428405e-05,
+      "loss": 0.2674,
+      "step": 357
+    },
+    {
+      "epoch": 0.22912,
+      "grad_norm": 0.2988840679235592,
+      "learning_rate": 4.578857906602791e-05,
+      "loss": 0.267,
+      "step": 358
+    },
+    {
+      "epoch": 0.22976,
+      "grad_norm": 0.3081169219238215,
+      "learning_rate": 4.575911522212516e-05,
+      "loss": 0.2842,
+      "step": 359
+    },
+    {
+      "epoch": 0.2304,
+      "grad_norm": 0.26927718261630545,
+      "learning_rate": 4.572955821894305e-05,
+      "loss": 0.2701,
+      "step": 360
+    },
+    {
+      "epoch": 0.23104,
+      "grad_norm": 0.28195917447085084,
+      "learning_rate": 4.5699908189122556e-05,
+      "loss": 0.2356,
+      "step": 361
+    },
+    {
+      "epoch": 0.23168,
+      "grad_norm": 0.31290690820988126,
+      "learning_rate": 4.5670165265722096e-05,
+      "loss": 0.2551,
+      "step": 362
+    },
+    {
+      "epoch": 0.23232,
+      "grad_norm": 0.333293969770491,
+      "learning_rate": 4.5640329582217004e-05,
+      "loss": 0.2838,
+      "step": 363
+    },
+    {
+      "epoch": 0.23296,
+      "grad_norm": 0.322941638451104,
+      "learning_rate": 4.5610401272498836e-05,
+      "loss": 0.2544,
+      "step": 364
+    },
+    {
+      "epoch": 0.2336,
+      "grad_norm": 0.28399822125645957,
+      "learning_rate": 4.558038047087486e-05,
+      "loss": 0.2635,
+      "step": 365
+    },
+    {
+      "epoch": 0.23424,
+      "grad_norm": 0.295506533919543,
+      "learning_rate": 4.555026731206741e-05,
+      "loss": 0.2482,
+      "step": 366
+    },
+    {
+      "epoch": 0.23488,
+      "grad_norm": 0.28684742571759847,
+      "learning_rate": 4.5520061931213255e-05,
+      "loss": 0.2789,
+      "step": 367
+    },
+    {
+      "epoch": 0.23552,
+      "grad_norm": 0.2736033024469486,
+      "learning_rate": 4.548976446386305e-05,
+      "loss": 0.2725,
+      "step": 368
+    },
+    {
+      "epoch": 0.23616,
+      "grad_norm": 0.2893642157562684,
+      "learning_rate": 4.545937504598069e-05,
+      "loss": 0.2716,
+      "step": 369
+    },
+    {
+      "epoch": 0.2368,
+      "grad_norm": 0.27790560741339493,
+      "learning_rate": 4.5428893813942727e-05,
+      "loss": 0.2683,
+      "step": 370
+    },
+    {
+      "epoch": 0.23744,
+      "grad_norm": 0.26374492858298765,
+      "learning_rate": 4.539832090453771e-05,
+      "loss": 0.2652,
+      "step": 371
+    },
+    {
+      "epoch": 0.23808,
+      "grad_norm": 0.317165806439944,
+      "learning_rate": 4.536765645496564e-05,
+      "loss": 0.2614,
+      "step": 372
+    },
+    {
+      "epoch": 0.23872,
+      "grad_norm": 0.2832889981603846,
+      "learning_rate": 4.533690060283727e-05,
+      "loss": 0.2655,
+      "step": 373
+    },
+    {
+      "epoch": 0.23936,
+      "grad_norm": 0.2835605762763791,
+      "learning_rate": 4.530605348617357e-05,
+      "loss": 0.2742,
+      "step": 374
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.30420408773631924,
+      "learning_rate": 4.527511524340508e-05,
+      "loss": 0.2644,
+      "step": 375
+    },
+    {
+      "epoch": 0.24064,
+      "grad_norm": 0.2969985444201941,
+      "learning_rate": 4.5244086013371246e-05,
+      "loss": 0.2638,
+      "step": 376
+    },
+    {
+      "epoch": 0.24128,
+      "grad_norm": 0.29308018866051944,
+      "learning_rate": 4.521296593531985e-05,
+      "loss": 0.2611,
+      "step": 377
+    },
+    {
+      "epoch": 0.24192,
+      "grad_norm": 0.2643945857889153,
+      "learning_rate": 4.518175514890636e-05,
+      "loss": 0.265,
+      "step": 378
+    },
+    {
+      "epoch": 0.24256,
+      "grad_norm": 0.32678388186327273,
+      "learning_rate": 4.515045379419332e-05,
+      "loss": 0.279,
+      "step": 379
+    },
+    {
+      "epoch": 0.2432,
+      "grad_norm": 0.2804538570907157,
+      "learning_rate": 4.5119062011649714e-05,
+      "loss": 0.2734,
+      "step": 380
+    },
+    {
+      "epoch": 0.24384,
+      "grad_norm": 0.3334812457840943,
+      "learning_rate": 4.508757994215031e-05,
+      "loss": 0.2811,
+      "step": 381
+    },
+    {
+      "epoch": 0.24448,
+      "grad_norm": 0.34946434978677615,
+      "learning_rate": 4.505600772697507e-05,
+      "loss": 0.2775,
+      "step": 382
+    },
+    {
+      "epoch": 0.24512,
+      "grad_norm": 0.3199621771747357,
+      "learning_rate": 4.5024345507808495e-05,
+      "loss": 0.2654,
+      "step": 383
+    },
+    {
+      "epoch": 0.24576,
+      "grad_norm": 0.31901998740646675,
+      "learning_rate": 4.499259342673898e-05,
+      "loss": 0.2764,
+      "step": 384
+    },
+    {
+      "epoch": 0.2464,
+      "grad_norm": 0.29597971799372186,
+      "learning_rate": 4.4960751626258205e-05,
+      "loss": 0.263,
+      "step": 385
+    },
+    {
+      "epoch": 0.24704,
+      "grad_norm": 0.3404264921907249,
+      "learning_rate": 4.492882024926046e-05,
+      "loss": 0.2724,
+      "step": 386
+    },
+    {
+      "epoch": 0.24768,
+      "grad_norm": 0.3139861943992207,
+      "learning_rate": 4.489679943904203e-05,
+      "loss": 0.2926,
+      "step": 387
+    },
+    {
+      "epoch": 0.24832,
+      "grad_norm": 0.2916841972443679,
+      "learning_rate": 4.486468933930055e-05,
+      "loss": 0.2799,
+      "step": 388
+    },
+    {
+      "epoch": 0.24896,
+      "grad_norm": 0.31153621464751174,
+      "learning_rate": 4.483249009413433e-05,
+      "loss": 0.2586,
+      "step": 389
+    },
+    {
+      "epoch": 0.2496,
+      "grad_norm": 0.3093059584707331,
+      "learning_rate": 4.4800201848041765e-05,
+      "loss": 0.2882,
+      "step": 390
+    },
+    {
+      "epoch": 0.25024,
+      "grad_norm": 0.3136725266830641,
+      "learning_rate": 4.476782474592062e-05,
+      "loss": 0.2624,
+      "step": 391
+    },
+    {
+      "epoch": 0.25088,
+      "grad_norm": 0.29406434454627556,
+      "learning_rate": 4.473535893306744e-05,
+      "loss": 0.2688,
+      "step": 392
+    },
+    {
+      "epoch": 0.25152,
+      "grad_norm": 0.2875071393783711,
+      "learning_rate": 4.470280455517687e-05,
+      "loss": 0.2611,
+      "step": 393
+    },
+    {
+      "epoch": 0.25216,
+      "grad_norm": 0.32671818220693416,
+      "learning_rate": 4.4670161758340966e-05,
+      "loss": 0.2726,
+      "step": 394
+    },
+    {
+      "epoch": 0.2528,
+      "grad_norm": 0.325381652042734,
+      "learning_rate": 4.4637430689048634e-05,
+      "loss": 0.2594,
+      "step": 395
+    },
+    {
+      "epoch": 0.25344,
+      "grad_norm": 0.3191566413040394,
+      "learning_rate": 4.460461149418486e-05,
+      "loss": 0.2553,
+      "step": 396
+    },
+    {
+      "epoch": 0.25408,
+      "grad_norm": 0.33203616250805296,
+      "learning_rate": 4.4571704321030125e-05,
+      "loss": 0.2898,
+      "step": 397
+    },
+    {
+      "epoch": 0.25472,
+      "grad_norm": 0.32044203951382094,
+      "learning_rate": 4.4538709317259754e-05,
+      "loss": 0.2471,
+      "step": 398
+    },
+    {
+      "epoch": 0.25536,
+      "grad_norm": 0.5990730119683106,
+      "learning_rate": 4.4505626630943174e-05,
+      "loss": 0.2879,
+      "step": 399
+    },
+    {
+      "epoch": 0.256,
+      "grad_norm": 0.3623407001600211,
+      "learning_rate": 4.447245641054334e-05,
+      "loss": 0.2729,
+      "step": 400
+    },
+    {
+      "epoch": 0.25664,
+      "grad_norm": 0.3038880359459755,
+      "learning_rate": 4.443919880491598e-05,
+      "loss": 0.2531,
+      "step": 401
+    },
+    {
+      "epoch": 0.25728,
+      "grad_norm": 0.2900829076508625,
+      "learning_rate": 4.440585396330904e-05,
+      "loss": 0.268,
+      "step": 402
+    },
+    {
+      "epoch": 0.25792,
+      "grad_norm": 0.31975046352986625,
+      "learning_rate": 4.437242203536189e-05,
+      "loss": 0.2714,
+      "step": 403
+    },
+    {
+      "epoch": 0.25856,
+      "grad_norm": 0.31641840526539994,
+      "learning_rate": 4.433890317110475e-05,
+      "loss": 0.2834,
+      "step": 404
+    },
+    {
+      "epoch": 0.2592,
+      "grad_norm": 0.2921767184505357,
+      "learning_rate": 4.430529752095795e-05,
+      "loss": 0.2701,
+      "step": 405
+    },
+    {
+      "epoch": 0.25984,
+      "grad_norm": 0.2808718335043851,
+      "learning_rate": 4.427160523573131e-05,
+      "loss": 0.279,
+      "step": 406
+    },
+    {
+      "epoch": 0.26048,
+      "grad_norm": 0.3016336985564651,
+      "learning_rate": 4.4237826466623406e-05,
+      "loss": 0.2873,
+      "step": 407
+    },
+    {
+      "epoch": 0.26112,
+      "grad_norm": 0.31431578482180944,
+      "learning_rate": 4.420396136522095e-05,
+      "loss": 0.2727,
+      "step": 408
+    },
+    {
+      "epoch": 0.26176,
+      "grad_norm": 0.27780637237792005,
+      "learning_rate": 4.417001008349807e-05,
+      "loss": 0.2579,
+      "step": 409
+    },
+    {
+      "epoch": 0.2624,
+      "grad_norm": 0.32993357310014576,
+      "learning_rate": 4.413597277381564e-05,
+      "loss": 0.2698,
+      "step": 410
+    },
+    {
+      "epoch": 0.26304,
+      "grad_norm": 0.2801299554162777,
+      "learning_rate": 4.410184958892059e-05,
+      "loss": 0.2651,
+      "step": 411
+    },
+    {
+      "epoch": 0.26368,
+      "grad_norm": 0.29257626461413505,
+      "learning_rate": 4.4067640681945244e-05,
+      "loss": 0.2626,
+      "step": 412
+    },
+    {
+      "epoch": 0.26432,
+      "grad_norm": 0.3357411211678043,
+      "learning_rate": 4.403334620640659e-05,
+      "loss": 0.2605,
+      "step": 413
+    },
+    {
+      "epoch": 0.26496,
+      "grad_norm": 0.30154707776244105,
+      "learning_rate": 4.3998966316205636e-05,
+      "loss": 0.2534,
+      "step": 414
+    },
+    {
+      "epoch": 0.2656,
+      "grad_norm": 0.25345682609356646,
+      "learning_rate": 4.396450116562669e-05,
+      "loss": 0.2444,
+      "step": 415
+    },
+    {
+      "epoch": 0.26624,
+      "grad_norm": 0.2721282547677732,
+      "learning_rate": 4.392995090933669e-05,
+      "loss": 0.2517,
+      "step": 416
+    },
+    {
+      "epoch": 0.26688,
+      "grad_norm": 0.28639350362828037,
+      "learning_rate": 4.389531570238448e-05,
+      "loss": 0.2416,
+      "step": 417
+    },
+    {
+      "epoch": 0.26752,
+      "grad_norm": 0.3414448358787958,
+      "learning_rate": 4.386059570020015e-05,
+      "loss": 0.277,
+      "step": 418
+    },
+    {
+      "epoch": 0.26816,
+      "grad_norm": 0.26641259434816816,
+      "learning_rate": 4.382579105859428e-05,
+      "loss": 0.2737,
+      "step": 419
+    },
+    {
+      "epoch": 0.2688,
+      "grad_norm": 0.3117899630025732,
+      "learning_rate": 4.379090193375735e-05,
+      "loss": 0.2684,
+      "step": 420
+    },
+    {
+      "epoch": 0.26944,
+      "grad_norm": 0.29638243573776074,
+      "learning_rate": 4.375592848225889e-05,
+      "loss": 0.2499,
+      "step": 421
+    },
+    {
+      "epoch": 0.27008,
+      "grad_norm": 0.28547284987584026,
+      "learning_rate": 4.372087086104692e-05,
+      "loss": 0.2767,
+      "step": 422
+    },
+    {
+      "epoch": 0.27072,
+      "grad_norm": 0.2997096365805738,
+      "learning_rate": 4.368572922744715e-05,
+      "loss": 0.2703,
+      "step": 423
+    },
+    {
+      "epoch": 0.27136,
+      "grad_norm": 0.30017436052612945,
+      "learning_rate": 4.365050373916233e-05,
+      "loss": 0.2753,
+      "step": 424
+    },
+    {
+      "epoch": 0.272,
+      "grad_norm": 0.29858890262572624,
+      "learning_rate": 4.3615194554271485e-05,
+      "loss": 0.2337,
+      "step": 425
+    },
+    {
+      "epoch": 0.27264,
+      "grad_norm": 0.29100268820291153,
+      "learning_rate": 4.357980183122927e-05,
+      "loss": 0.2722,
+      "step": 426
+    },
+    {
+      "epoch": 0.27328,
+      "grad_norm": 0.3031136982843706,
+      "learning_rate": 4.354432572886522e-05,
+      "loss": 0.2705,
+      "step": 427
+    },
+    {
+      "epoch": 0.27392,
+      "grad_norm": 0.27735280662106965,
+      "learning_rate": 4.3508766406383036e-05,
+      "loss": 0.2773,
+      "step": 428
+    },
+    {
+      "epoch": 0.27456,
+      "grad_norm": 0.287598645255541,
+      "learning_rate": 4.347312402335989e-05,
+      "loss": 0.2683,
+      "step": 429
+    },
+    {
+      "epoch": 0.2752,
+      "grad_norm": 0.29025080310739393,
+      "learning_rate": 4.34373987397457e-05,
+      "loss": 0.2775,
+      "step": 430
+    },
+    {
+      "epoch": 0.27584,
+      "grad_norm": 0.27374139232520084,
+      "learning_rate": 4.340159071586241e-05,
+      "loss": 0.2447,
+      "step": 431
+    },
+    {
+      "epoch": 0.27648,
+      "grad_norm": 0.30499106877780574,
+      "learning_rate": 4.336570011240326e-05,
+      "loss": 0.245,
+      "step": 432
+    },
+    {
+      "epoch": 0.27712,
+      "grad_norm": 0.2789830538820019,
+      "learning_rate": 4.33297270904321e-05,
+      "loss": 0.2667,
+      "step": 433
+    },
+    {
+      "epoch": 0.27776,
+      "grad_norm": 0.27918341247404516,
+      "learning_rate": 4.329367181138262e-05,
+      "loss": 0.2756,
+      "step": 434
+    },
+    {
+      "epoch": 0.2784,
+      "grad_norm": 0.29003519846284037,
+      "learning_rate": 4.3257534437057674e-05,
+      "loss": 0.2567,
+      "step": 435
+    },
+    {
+      "epoch": 0.27904,
+      "grad_norm": 0.28216675485922665,
+      "learning_rate": 4.3221315129628506e-05,
+      "loss": 0.2481,
+      "step": 436
+    },
+    {
+      "epoch": 0.27968,
+      "grad_norm": 0.30220665750339526,
+      "learning_rate": 4.318501405163406e-05,
+      "loss": 0.2616,
+      "step": 437
+    },
+    {
+      "epoch": 0.28032,
+      "grad_norm": 0.2634822858500782,
+      "learning_rate": 4.314863136598022e-05,
+      "loss": 0.2462,
+      "step": 438
+    },
+    {
+      "epoch": 0.28096,
+      "grad_norm": 0.3161113446906667,
+      "learning_rate": 4.311216723593913e-05,
+      "loss": 0.286,
+      "step": 439
+    },
+    {
+      "epoch": 0.2816,
+      "grad_norm": 0.3222940745311678,
+      "learning_rate": 4.307562182514838e-05,
+      "loss": 0.2609,
+      "step": 440
+    },
+    {
+      "epoch": 0.28224,
+      "grad_norm": 0.3187572190375552,
+      "learning_rate": 4.3038995297610364e-05,
+      "loss": 0.2478,
+      "step": 441
+    },
+    {
+      "epoch": 0.28288,
+      "grad_norm": 0.2928301432408682,
+      "learning_rate": 4.3002287817691466e-05,
+      "loss": 0.2855,
+      "step": 442
+    },
+    {
+      "epoch": 0.28352,
+      "grad_norm": 0.32965679351152233,
+      "learning_rate": 4.2965499550121354e-05,
+      "loss": 0.2682,
+      "step": 443
+    },
+    {
+      "epoch": 0.28416,
+      "grad_norm": 0.3349054379926969,
+      "learning_rate": 4.2928630659992267e-05,
+      "loss": 0.2708,
+      "step": 444
+    },
+    {
+      "epoch": 0.2848,
+      "grad_norm": 0.3409906128958908,
+      "learning_rate": 4.2891681312758225e-05,
+      "loss": 0.2589,
+      "step": 445
+    },
+    {
+      "epoch": 0.28544,
+      "grad_norm": 0.29301815221408106,
+      "learning_rate": 4.285465167423432e-05,
+      "loss": 0.2712,
+      "step": 446
+    },
+    {
+      "epoch": 0.28608,
+      "grad_norm": 0.3151776631679768,
+      "learning_rate": 4.281754191059596e-05,
+      "loss": 0.2609,
+      "step": 447
+    },
+    {
+      "epoch": 0.28672,
+      "grad_norm": 0.3210236670212237,
+      "learning_rate": 4.278035218837813e-05,
+      "loss": 0.2877,
+      "step": 448
+    },
+    {
+      "epoch": 0.28736,
+      "grad_norm": 0.32717066085861096,
+      "learning_rate": 4.274308267447463e-05,
+      "loss": 0.2511,
+      "step": 449
+    },
+    {
+      "epoch": 0.288,
+      "grad_norm": 0.30041723762252004,
+      "learning_rate": 4.270573353613734e-05,
+      "loss": 0.2697,
+      "step": 450
+    },
+    {
+      "epoch": 0.28864,
+      "grad_norm": 0.2806245348619511,
+      "learning_rate": 4.266830494097546e-05,
+      "loss": 0.2535,
+      "step": 451
+    },
+    {
+      "epoch": 0.28928,
+      "grad_norm": 0.30866632722556286,
+      "learning_rate": 4.263079705695478e-05,
+      "loss": 0.2627,
+      "step": 452
+    },
+    {
+      "epoch": 0.28992,
+      "grad_norm": 0.3481734862540506,
+      "learning_rate": 4.2593210052396894e-05,
+      "loss": 0.2529,
+      "step": 453
+    },
+    {
+      "epoch": 0.29056,
+      "grad_norm": 0.3002149800768276,
+      "learning_rate": 4.255554409597846e-05,
+      "loss": 0.2446,
+      "step": 454
+    },
+    {
+      "epoch": 0.2912,
+      "grad_norm": 0.33721972074704143,
+      "learning_rate": 4.2517799356730445e-05,
+      "loss": 0.2682,
+      "step": 455
+    },
+    {
+      "epoch": 0.29184,
+      "grad_norm": 0.29436884923501094,
+      "learning_rate": 4.247997600403737e-05,
+      "loss": 0.2582,
+      "step": 456
+    },
+    {
+      "epoch": 0.29248,
+      "grad_norm": 0.2967220259575209,
+      "learning_rate": 4.244207420763655e-05,
+      "loss": 0.2494,
+      "step": 457
+    },
+    {
+      "epoch": 0.29312,
+      "grad_norm": 0.2705548005196565,
+      "learning_rate": 4.240409413761731e-05,
+      "loss": 0.2545,
+      "step": 458
+    },
+    {
+      "epoch": 0.29376,
+      "grad_norm": 0.3046845043984032,
+      "learning_rate": 4.236603596442023e-05,
+      "loss": 0.2623,
+      "step": 459
+    },
+    {
+      "epoch": 0.2944,
+      "grad_norm": 0.293075805394358,
+      "learning_rate": 4.232789985883642e-05,
+      "loss": 0.2692,
+      "step": 460
+    },
+    {
+      "epoch": 0.29504,
+      "grad_norm": 0.2934980535483767,
+      "learning_rate": 4.2289685992006704e-05,
+      "loss": 0.2768,
+      "step": 461
+    },
+    {
+      "epoch": 0.29568,
+      "grad_norm": 0.3000713141045476,
+      "learning_rate": 4.225139453542087e-05,
+      "loss": 0.2605,
+      "step": 462
+    },
+    {
+      "epoch": 0.29632,
+      "grad_norm": 0.2930768123609869,
+      "learning_rate": 4.2213025660916904e-05,
+      "loss": 0.2661,
+      "step": 463
+    },
+    {
+      "epoch": 0.29696,
+      "grad_norm": 0.3178121108720456,
+      "learning_rate": 4.21745795406802e-05,
+      "loss": 0.2461,
+      "step": 464
+    },
+    {
+      "epoch": 0.2976,
+      "grad_norm": 0.26800811778086625,
+      "learning_rate": 4.2136056347242835e-05,
+      "loss": 0.2494,
+      "step": 465
+    },
+    {
+      "epoch": 0.29824,
+      "grad_norm": 0.3137325849551812,
+      "learning_rate": 4.209745625348273e-05,
+      "loss": 0.2839,
+      "step": 466
+    },
+    {
+      "epoch": 0.29888,
+      "grad_norm": 0.32102282561988843,
+      "learning_rate": 4.205877943262291e-05,
+      "loss": 0.2452,
+      "step": 467
+    },
+    {
+      "epoch": 0.29952,
+      "grad_norm": 0.26576307125167287,
+      "learning_rate": 4.2020026058230756e-05,
+      "loss": 0.2675,
+      "step": 468
+    },
+    {
+      "epoch": 0.30016,
+      "grad_norm": 0.3003922078840633,
+      "learning_rate": 4.1981196304217154e-05,
+      "loss": 0.2625,
+      "step": 469
+    },
+    {
+      "epoch": 0.3008,
+      "grad_norm": 0.2929467586131004,
+      "learning_rate": 4.1942290344835765e-05,
+      "loss": 0.2635,
+      "step": 470
+    },
+    {
+      "epoch": 0.30144,
+      "grad_norm": 0.28946698136298926,
+      "learning_rate": 4.190330835468224e-05,
+      "loss": 0.2666,
+      "step": 471
+    },
+    {
+      "epoch": 0.30208,
+      "grad_norm": 0.3054605229887845,
+      "learning_rate": 4.1864250508693424e-05,
+      "loss": 0.2782,
+      "step": 472
+    },
+    {
+      "epoch": 0.30272,
+      "grad_norm": 0.30853302514826486,
+      "learning_rate": 4.182511698214657e-05,
+      "loss": 0.271,
+      "step": 473
+    },
+    {
+      "epoch": 0.30336,
+      "grad_norm": 0.28096000312894664,
+      "learning_rate": 4.1785907950658573e-05,
+      "loss": 0.2585,
+      "step": 474
+    },
+    {
+      "epoch": 0.304,
+      "grad_norm": 0.3125936908697687,
+      "learning_rate": 4.1746623590185155e-05,
+      "loss": 0.2524,
+      "step": 475
+    },
+    {
+      "epoch": 0.30464,
+      "grad_norm": 0.29712867964294787,
+      "learning_rate": 4.170726407702008e-05,
+      "loss": 0.2461,
+      "step": 476
+    },
+    {
+      "epoch": 0.30528,
+      "grad_norm": 0.32449012009401823,
+      "learning_rate": 4.1667829587794374e-05,
+      "loss": 0.2822,
+      "step": 477
+    },
+    {
+      "epoch": 0.30592,
+      "grad_norm": 0.3218131224973865,
+      "learning_rate": 4.162832029947553e-05,
+      "loss": 0.2725,
+      "step": 478
+    },
+    {
+      "epoch": 0.30656,
+      "grad_norm": 0.2804231247197045,
+      "learning_rate": 4.158873638936672e-05,
+      "loss": 0.2471,
+      "step": 479
+    },
+    {
+      "epoch": 0.3072,
+      "grad_norm": 0.31284785097448237,
+      "learning_rate": 4.154907803510599e-05,
+      "loss": 0.252,
+      "step": 480
+    },
+    {
+      "epoch": 0.30784,
+      "grad_norm": 0.2994921809896846,
+      "learning_rate": 4.150934541466545e-05,
+      "loss": 0.2614,
+      "step": 481
+    },
+    {
+      "epoch": 0.30848,
+      "grad_norm": 0.296657249343442,
+      "learning_rate": 4.146953870635049e-05,
+      "loss": 0.2722,
+      "step": 482
+    },
+    {
+      "epoch": 0.30912,
+      "grad_norm": 0.28293479502038055,
+      "learning_rate": 4.1429658088798996e-05,
+      "loss": 0.2634,
+      "step": 483
+    },
+    {
+      "epoch": 0.30976,
+      "grad_norm": 0.2910484344542188,
+      "learning_rate": 4.138970374098052e-05,
+      "loss": 0.2486,
+      "step": 484
+    },
+    {
+      "epoch": 0.3104,
+      "grad_norm": 0.3146840001735881,
+      "learning_rate": 4.1349675842195494e-05,
+      "loss": 0.2661,
+      "step": 485
+    },
+    {
+      "epoch": 0.31104,
+      "grad_norm": 0.2746027576599675,
+      "learning_rate": 4.130957457207441e-05,
+      "loss": 0.2501,
+      "step": 486
+    },
+    {
+      "epoch": 0.31168,
+      "grad_norm": 0.3051892164851984,
+      "learning_rate": 4.126940011057703e-05,
+      "loss": 0.2404,
+      "step": 487
+    },
+    {
+      "epoch": 0.31232,
+      "grad_norm": 0.30592656911462335,
+      "learning_rate": 4.1229152637991576e-05,
+      "loss": 0.2673,
+      "step": 488
+    },
+    {
+      "epoch": 0.31296,
+      "grad_norm": 0.31633018710097294,
+      "learning_rate": 4.1188832334933906e-05,
+      "loss": 0.2475,
+      "step": 489
+    },
+    {
+      "epoch": 0.3136,
+      "grad_norm": 0.274585788639711,
+      "learning_rate": 4.114843938234673e-05,
+      "loss": 0.2753,
+      "step": 490
+    },
+    {
+      "epoch": 0.31424,
+      "grad_norm": 0.3287109219068746,
+      "learning_rate": 4.1107973961498763e-05,
+      "loss": 0.2582,
+      "step": 491
+    },
+    {
+      "epoch": 0.31488,
+      "grad_norm": 0.3167546562615222,
+      "learning_rate": 4.106743625398395e-05,
+      "loss": 0.2662,
+      "step": 492
+    },
+    {
+      "epoch": 0.31552,
+      "grad_norm": 0.32410088907898493,
+      "learning_rate": 4.102682644172061e-05,
+      "loss": 0.2869,
+      "step": 493
+    },
+    {
+      "epoch": 0.31616,
+      "grad_norm": 0.3355149408031651,
+      "learning_rate": 4.0986144706950677e-05,
+      "loss": 0.2746,
+      "step": 494
+    },
+    {
+      "epoch": 0.3168,
+      "grad_norm": 0.3074689123409369,
+      "learning_rate": 4.094539123223881e-05,
+      "loss": 0.2433,
+      "step": 495
+    },
+    {
+      "epoch": 0.31744,
+      "grad_norm": 0.29842413604098983,
+      "learning_rate": 4.090456620047162e-05,
+      "loss": 0.2435,
+      "step": 496
+    },
+    {
+      "epoch": 0.31808,
+      "grad_norm": 0.3414546216468358,
+      "learning_rate": 4.086366979485685e-05,
+      "loss": 0.2796,
+      "step": 497
+    },
+    {
+      "epoch": 0.31872,
+      "grad_norm": 0.3150858918248773,
+      "learning_rate": 4.082270219892254e-05,
+      "loss": 0.2633,
+      "step": 498
+    },
+    {
+      "epoch": 0.31936,
+      "grad_norm": 0.2786327977030963,
+      "learning_rate": 4.0781663596516206e-05,
+      "loss": 0.2419,
+      "step": 499
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.29092976216281774,
+      "learning_rate": 4.074055417180401e-05,
+      "loss": 0.2697,
+      "step": 500
+    },
+    {
+      "epoch": 0.32064,
+      "grad_norm": 0.32610697106570463,
+      "learning_rate": 4.069937410926994e-05,
+      "loss": 0.2537,
+      "step": 501
+    },
+    {
+      "epoch": 0.32128,
+      "grad_norm": 0.30608850743455823,
+      "learning_rate": 4.0658123593714984e-05,
+      "loss": 0.264,
+      "step": 502
+    },
+    {
+      "epoch": 0.32192,
+      "grad_norm": 0.2824318843825735,
+      "learning_rate": 4.061680281025628e-05,
+      "loss": 0.2474,
+      "step": 503
+    },
+    {
+      "epoch": 0.32256,
+      "grad_norm": 0.29399230400886783,
+      "learning_rate": 4.057541194432634e-05,
+      "loss": 0.2413,
+      "step": 504
+    },
+    {
+      "epoch": 0.3232,
+      "grad_norm": 0.6605989874825366,
+      "learning_rate": 4.053395118167214e-05,
+      "loss": 0.2581,
+      "step": 505
+    },
+    {
+      "epoch": 0.32384,
+      "grad_norm": 0.30398550676295183,
+      "learning_rate": 4.049242070835435e-05,
+      "loss": 0.2864,
+      "step": 506
+    },
+    {
+      "epoch": 0.32448,
+      "grad_norm": 0.32844118327083244,
+      "learning_rate": 4.0450820710746475e-05,
+      "loss": 0.2595,
+      "step": 507
+    },
+    {
+      "epoch": 0.32512,
+      "grad_norm": 0.29753850766246936,
+      "learning_rate": 4.0409151375534014e-05,
+      "loss": 0.2627,
+      "step": 508
+    },
+    {
+      "epoch": 0.32576,
+      "grad_norm": 0.32997195633690085,
+      "learning_rate": 4.036741288971362e-05,
+      "loss": 0.2552,
+      "step": 509
+    },
+    {
+      "epoch": 0.3264,
+      "grad_norm": 0.31989725360404464,
+      "learning_rate": 4.032560544059228e-05,
+      "loss": 0.2543,
+      "step": 510
+    },
+    {
+      "epoch": 0.32704,
+      "grad_norm": 0.2734053342728525,
+      "learning_rate": 4.028372921578647e-05,
+      "loss": 0.2636,
+      "step": 511
+    },
+    {
+      "epoch": 0.32768,
+      "grad_norm": 0.3076390699096864,
+      "learning_rate": 4.024178440322128e-05,
+      "loss": 0.251,
+      "step": 512
+    },
+    {
+      "epoch": 0.32832,
+      "grad_norm": 0.27589462055645664,
+      "learning_rate": 4.019977119112962e-05,
+      "loss": 0.2599,
+      "step": 513
+    },
+    {
+      "epoch": 0.32896,
+      "grad_norm": 0.3152240440071727,
+      "learning_rate": 4.015768976805135e-05,
+      "loss": 0.303,
+      "step": 514
+    },
+    {
+      "epoch": 0.3296,
+      "grad_norm": 0.30486982512783595,
+      "learning_rate": 4.011554032283242e-05,
+      "loss": 0.2339,
+      "step": 515
+    },
+    {
+      "epoch": 0.33024,
+      "grad_norm": 0.334048138143333,
+      "learning_rate": 4.007332304462406e-05,
+      "loss": 0.2494,
+      "step": 516
+    },
+    {
+      "epoch": 0.33088,
+      "grad_norm": 0.2995442580330488,
+      "learning_rate": 4.003103812288189e-05,
+      "loss": 0.267,
+      "step": 517
+    },
+    {
+      "epoch": 0.33152,
+      "grad_norm": 0.2706177023809212,
+      "learning_rate": 3.99886857473651e-05,
+      "loss": 0.2523,
+      "step": 518
+    },
+    {
+      "epoch": 0.33216,
+      "grad_norm": 0.3532067575064781,
+      "learning_rate": 3.9946266108135575e-05,
+      "loss": 0.2767,
+      "step": 519
+    },
+    {
+      "epoch": 0.3328,
+      "grad_norm": 0.3067722195105639,
+      "learning_rate": 3.9903779395557087e-05,
+      "loss": 0.2606,
+      "step": 520
+    },
+    {
+      "epoch": 0.33344,
+      "grad_norm": 0.26860794555811235,
+      "learning_rate": 3.9861225800294366e-05,
+      "loss": 0.2295,
+      "step": 521
+    },
+    {
+      "epoch": 0.33408,
+      "grad_norm": 0.34447335382371475,
+      "learning_rate": 3.981860551331232e-05,
+      "loss": 0.2665,
+      "step": 522
+    },
+    {
+      "epoch": 0.33472,
+      "grad_norm": 0.302941110641683,
+      "learning_rate": 3.977591872587512e-05,
+      "loss": 0.2555,
+      "step": 523
+    },
+    {
+      "epoch": 0.33536,
+      "grad_norm": 0.26824391744215076,
+      "learning_rate": 3.973316562954539e-05,
+      "loss": 0.2514,
+      "step": 524
+    },
+    {
+      "epoch": 0.336,
+      "grad_norm": 0.29697669498197016,
+      "learning_rate": 3.969034641618332e-05,
+      "loss": 0.2841,
+      "step": 525
+    },
+    {
+      "epoch": 0.33664,
+      "grad_norm": 0.3140181304725373,
+      "learning_rate": 3.964746127794579e-05,
+      "loss": 0.2624,
+      "step": 526
+    },
+    {
+      "epoch": 0.33728,
+      "grad_norm": 0.2894158324939913,
+      "learning_rate": 3.9604510407285524e-05,
+      "loss": 0.2502,
+      "step": 527
+    },
+    {
+      "epoch": 0.33792,
+      "grad_norm": 0.3099217657298054,
+      "learning_rate": 3.956149399695027e-05,
+      "loss": 0.2518,
+      "step": 528
+    },
+    {
+      "epoch": 0.33856,
+      "grad_norm": 0.32382883847833865,
+      "learning_rate": 3.951841223998186e-05,
+      "loss": 0.2797,
+      "step": 529
+    },
+    {
+      "epoch": 0.3392,
+      "grad_norm": 0.284045608691047,
+      "learning_rate": 3.947526532971538e-05,
+      "loss": 0.2447,
+      "step": 530
+    },
+    {
+      "epoch": 0.33984,
+      "grad_norm": 0.2798577999403126,
+      "learning_rate": 3.943205345977831e-05,
+      "loss": 0.2588,
+      "step": 531
+    },
+    {
+      "epoch": 0.34048,
+      "grad_norm": 0.3302444926457134,
+      "learning_rate": 3.9388776824089636e-05,
+      "loss": 0.257,
+      "step": 532
+    },
+    {
+      "epoch": 0.34112,
+      "grad_norm": 0.2668807922005792,
+      "learning_rate": 3.9345435616858995e-05,
+      "loss": 0.2669,
+      "step": 533
+    },
+    {
+      "epoch": 0.34176,
+      "grad_norm": 0.28784730348838994,
+      "learning_rate": 3.930203003258579e-05,
+      "loss": 0.256,
+      "step": 534
+    },
+    {
+      "epoch": 0.3424,
+      "grad_norm": 0.27655472424226957,
+      "learning_rate": 3.9258560266058335e-05,
+      "loss": 0.2688,
+      "step": 535
+    },
+    {
+      "epoch": 0.34304,
+      "grad_norm": 0.27155813960082525,
+      "learning_rate": 3.921502651235297e-05,
+      "loss": 0.2533,
+      "step": 536
+    },
+    {
+      "epoch": 0.34368,
+      "grad_norm": 0.2879263227468744,
+      "learning_rate": 3.917142896683318e-05,
+      "loss": 0.2713,
+      "step": 537
+    },
+    {
+      "epoch": 0.34432,
+      "grad_norm": 0.3003636295000633,
+      "learning_rate": 3.912776782514872e-05,
+      "loss": 0.2543,
+      "step": 538
+    },
+    {
+      "epoch": 0.34496,
+      "grad_norm": 0.29322399922295483,
+      "learning_rate": 3.908404328323475e-05,
+      "loss": 0.2727,
+      "step": 539
+    },
+    {
+      "epoch": 0.3456,
+      "grad_norm": 0.3087123123708106,
+      "learning_rate": 3.904025553731094e-05,
+      "loss": 0.2425,
+      "step": 540
+    },
+    {
+      "epoch": 0.34624,
+      "grad_norm": 0.2719537953443515,
+      "learning_rate": 3.8996404783880606e-05,
+      "loss": 0.2431,
+      "step": 541
+    },
+    {
+      "epoch": 0.34688,
+      "grad_norm": 0.27066398885499027,
+      "learning_rate": 3.895249121972981e-05,
+      "loss": 0.2424,
+      "step": 542
+    },
+    {
+      "epoch": 0.34752,
+      "grad_norm": 0.29396800013680485,
+      "learning_rate": 3.890851504192649e-05,
+      "loss": 0.2693,
+      "step": 543
+    },
+    {
+      "epoch": 0.34816,
+      "grad_norm": 0.30739760672085653,
+      "learning_rate": 3.8864476447819565e-05,
+      "loss": 0.2462,
+      "step": 544
+    },
+    {
+      "epoch": 0.3488,
+      "grad_norm": 0.27365055553703826,
+      "learning_rate": 3.8820375635038067e-05,
+      "loss": 0.2566,
+      "step": 545
+    },
+    {
+      "epoch": 0.34944,
+      "grad_norm": 0.32111512888277294,
+      "learning_rate": 3.877621280149022e-05,
+      "loss": 0.2666,
+      "step": 546
+    },
+    {
+      "epoch": 0.35008,
+      "grad_norm": 0.26661162881748623,
+      "learning_rate": 3.87319881453626e-05,
+      "loss": 0.2344,
+      "step": 547
+    },
+    {
+      "epoch": 0.35072,
+      "grad_norm": 0.2776431714793595,
+      "learning_rate": 3.868770186511921e-05,
+      "loss": 0.2616,
+      "step": 548
+    },
+    {
+      "epoch": 0.35136,
+      "grad_norm": 0.3142251725356459,
+      "learning_rate": 3.864335415950059e-05,
+      "loss": 0.2426,
+      "step": 549
+    },
+    {
+      "epoch": 0.352,
+      "grad_norm": 0.2672867397650059,
+      "learning_rate": 3.859894522752297e-05,
+      "loss": 0.2491,
+      "step": 550
+    },
+    {
+      "epoch": 0.35264,
+      "grad_norm": 0.31108447060901917,
+      "learning_rate": 3.8554475268477274e-05,
+      "loss": 0.2562,
+      "step": 551
+    },
+    {
+      "epoch": 0.35328,
+      "grad_norm": 0.27997496171373765,
+      "learning_rate": 3.850994448192836e-05,
+      "loss": 0.2596,
+      "step": 552
+    },
+    {
+      "epoch": 0.35392,
+      "grad_norm": 0.2520410047003102,
+      "learning_rate": 3.846535306771404e-05,
+      "loss": 0.2489,
+      "step": 553
+    },
+    {
+      "epoch": 0.35456,
+      "grad_norm": 0.3199503409429983,
+      "learning_rate": 3.842070122594417e-05,
+      "loss": 0.2639,
+      "step": 554
+    },
+    {
+      "epoch": 0.3552,
+      "grad_norm": 0.2760069756283184,
+      "learning_rate": 3.837598915699981e-05,
+      "loss": 0.2452,
+      "step": 555
+    },
+    {
+      "epoch": 0.35584,
+      "grad_norm": 0.2993695077500272,
+      "learning_rate": 3.83312170615323e-05,
+      "loss": 0.238,
+      "step": 556
+    },
+    {
+      "epoch": 0.35648,
+      "grad_norm": 0.2902731977925139,
+      "learning_rate": 3.8286385140462334e-05,
+      "loss": 0.2459,
+      "step": 557
+    },
+    {
+      "epoch": 0.35712,
+      "grad_norm": 0.31143517844282426,
+      "learning_rate": 3.8241493594979114e-05,
+      "loss": 0.2639,
+      "step": 558
+    },
+    {
+      "epoch": 0.35776,
+      "grad_norm": 0.30084283840717546,
+      "learning_rate": 3.8196542626539396e-05,
+      "loss": 0.2632,
+      "step": 559
+    },
+    {
+      "epoch": 0.3584,
+      "grad_norm": 0.3101846376917541,
+      "learning_rate": 3.815153243686658e-05,
+      "loss": 0.2643,
+      "step": 560
+    },
+    {
+      "epoch": 0.35904,
+      "grad_norm": 0.2997444458678429,
+      "learning_rate": 3.8106463227949885e-05,
+      "loss": 0.2639,
+      "step": 561
+    },
+    {
+      "epoch": 0.35968,
+      "grad_norm": 0.2858888950818185,
+      "learning_rate": 3.806133520204335e-05,
+      "loss": 0.2356,
+      "step": 562
+    },
+    {
+      "epoch": 0.36032,
+      "grad_norm": 0.2746834358811253,
+      "learning_rate": 3.801614856166495e-05,
+      "loss": 0.2438,
+      "step": 563
+    },
+    {
+      "epoch": 0.36096,
+      "grad_norm": 0.28929323567876075,
+      "learning_rate": 3.797090350959574e-05,
+      "loss": 0.247,
+      "step": 564
+    },
+    {
+      "epoch": 0.3616,
+      "grad_norm": 0.292384662428972,
+      "learning_rate": 3.792560024887887e-05,
+      "loss": 0.2527,
+      "step": 565
+    },
+    {
+      "epoch": 0.36224,
+      "grad_norm": 0.2940770397336415,
+      "learning_rate": 3.788023898281873e-05,
+      "loss": 0.2522,
+      "step": 566
+    },
+    {
+      "epoch": 0.36288,
+      "grad_norm": 0.29918580811759726,
+      "learning_rate": 3.783481991498001e-05,
+      "loss": 0.2568,
+      "step": 567
+    },
+    {
+      "epoch": 0.36352,
+      "grad_norm": 0.27833302819764766,
+      "learning_rate": 3.7789343249186805e-05,
+      "loss": 0.2646,
+      "step": 568
+    },
+    {
+      "epoch": 0.36416,
+      "grad_norm": 0.302337124189296,
+      "learning_rate": 3.774380918952165e-05,
+      "loss": 0.2733,
+      "step": 569
+    },
+    {
+      "epoch": 0.3648,
+      "grad_norm": 0.2888293609928954,
+      "learning_rate": 3.76982179403247e-05,
+      "loss": 0.2368,
+      "step": 570
+    },
+    {
+      "epoch": 0.36544,
+      "grad_norm": 0.2754590853626415,
+      "learning_rate": 3.7652569706192694e-05,
+      "loss": 0.2673,
+      "step": 571
+    },
+    {
+      "epoch": 0.36608,
+      "grad_norm": 0.2622654614217577,
+      "learning_rate": 3.760686469197815e-05,
+      "loss": 0.2483,
+      "step": 572
+    },
+    {
+      "epoch": 0.36672,
+      "grad_norm": 0.32199523182319184,
+      "learning_rate": 3.756110310278835e-05,
+      "loss": 0.2483,
+      "step": 573
+    },
+    {
+      "epoch": 0.36736,
+      "grad_norm": 0.2425311343256951,
+      "learning_rate": 3.7515285143984524e-05,
+      "loss": 0.2421,
+      "step": 574
+    },
+    {
+      "epoch": 0.368,
+      "grad_norm": 0.28404037478829497,
+      "learning_rate": 3.746941102118082e-05,
+      "loss": 0.2499,
+      "step": 575
+    },
+    {
+      "epoch": 0.36864,
+      "grad_norm": 0.2493449366928021,
+      "learning_rate": 3.742348094024342e-05,
+      "loss": 0.2715,
+      "step": 576
+    },
+    {
+      "epoch": 0.36928,
+      "grad_norm": 0.29364367244129747,
+      "learning_rate": 3.7377495107289664e-05,
+      "loss": 0.2281,
+      "step": 577
+    },
+    {
+      "epoch": 0.36992,
+      "grad_norm": 0.2893040965925678,
+      "learning_rate": 3.7331453728687054e-05,
+      "loss": 0.2518,
+      "step": 578
+    },
+    {
+      "epoch": 0.37056,
+      "grad_norm": 0.32004880336534,
+      "learning_rate": 3.7285357011052394e-05,
+      "loss": 0.2386,
+      "step": 579
+    },
+    {
+      "epoch": 0.3712,
+      "grad_norm": 0.2970709857173869,
+      "learning_rate": 3.72392051612508e-05,
+      "loss": 0.2681,
+      "step": 580
+    },
+    {
+      "epoch": 0.37184,
+      "grad_norm": 0.28346007851052274,
+      "learning_rate": 3.719299838639481e-05,
+      "loss": 0.2314,
+      "step": 581
+    },
+    {
+      "epoch": 0.37248,
+      "grad_norm": 0.2952808925416027,
+      "learning_rate": 3.714673689384346e-05,
+      "loss": 0.2721,
+      "step": 582
+    },
+    {
+      "epoch": 0.37312,
+      "grad_norm": 0.2917636244004094,
+      "learning_rate": 3.710042089120131e-05,
+      "loss": 0.2471,
+      "step": 583
+    },
+    {
+      "epoch": 0.37376,
+      "grad_norm": 0.2788339304900239,
+      "learning_rate": 3.705405058631756e-05,
+      "loss": 0.262,
+      "step": 584
+    },
+    {
+      "epoch": 0.3744,
+      "grad_norm": 0.27643980002724095,
+      "learning_rate": 3.700762618728508e-05,
+      "loss": 0.256,
+      "step": 585
+    },
+    {
+      "epoch": 0.37504,
+      "grad_norm": 0.3126161432024306,
+      "learning_rate": 3.696114790243955e-05,
+      "loss": 0.2632,
+      "step": 586
+    },
+    {
+      "epoch": 0.37568,
+      "grad_norm": 0.28522093731775944,
+      "learning_rate": 3.691461594035841e-05,
+      "loss": 0.2781,
+      "step": 587
+    },
+    {
+      "epoch": 0.37632,
+      "grad_norm": 0.2601633342904399,
+      "learning_rate": 3.686803050986e-05,
+      "loss": 0.2464,
+      "step": 588
+    },
+    {
+      "epoch": 0.37696,
+      "grad_norm": 0.34631359473350637,
+      "learning_rate": 3.6821391820002596e-05,
+      "loss": 0.2431,
+      "step": 589
+    },
+    {
+      "epoch": 0.3776,
+      "grad_norm": 0.25128016854145163,
+      "learning_rate": 3.677470008008352e-05,
+      "loss": 0.245,
+      "step": 590
+    },
+    {
+      "epoch": 0.37824,
+      "grad_norm": 0.2835575049175339,
+      "learning_rate": 3.672795549963814e-05,
+      "loss": 0.2439,
+      "step": 591
+    },
+    {
+      "epoch": 0.37888,
+      "grad_norm": 0.2946996953611722,
+      "learning_rate": 3.668115828843894e-05,
+      "loss": 0.2495,
+      "step": 592
+    },
+    {
+      "epoch": 0.37952,
+      "grad_norm": 0.28051307799779646,
+      "learning_rate": 3.6634308656494605e-05,
+      "loss": 0.2502,
+      "step": 593
+    },
+    {
+      "epoch": 0.38016,
+      "grad_norm": 0.28609093005292136,
+      "learning_rate": 3.6587406814049075e-05,
+      "loss": 0.2399,
+      "step": 594
+    },
+    {
+      "epoch": 0.3808,
+      "grad_norm": 0.2675711957042453,
+      "learning_rate": 3.6540452971580566e-05,
+      "loss": 0.2581,
+      "step": 595
+    },
+    {
+      "epoch": 0.38144,
+      "grad_norm": 0.3064348974320462,
+      "learning_rate": 3.649344733980068e-05,
+      "loss": 0.2536,
+      "step": 596
+    },
+    {
+      "epoch": 0.38208,
+      "grad_norm": 0.29746364649694573,
+      "learning_rate": 3.6446390129653405e-05,
+      "loss": 0.2491,
+      "step": 597
+    },
+    {
+      "epoch": 0.38272,
+      "grad_norm": 0.2593866350853291,
+      "learning_rate": 3.63992815523142e-05,
+      "loss": 0.2399,
+      "step": 598
+    },
+    {
+      "epoch": 0.38336,
+      "grad_norm": 0.30890519652414655,
+      "learning_rate": 3.6352121819189075e-05,
+      "loss": 0.2416,
+      "step": 599
+    },
+    {
+      "epoch": 0.384,
+      "grad_norm": 0.31924495631951694,
+      "learning_rate": 3.630491114191354e-05,
+      "loss": 0.2399,
+      "step": 600
+    },
+    {
+      "epoch": 0.38464,
+      "grad_norm": 0.30411416568115157,
+      "learning_rate": 3.62576497323518e-05,
+      "loss": 0.2784,
+      "step": 601
+    },
+    {
+      "epoch": 0.38528,
+      "grad_norm": 0.2814515020258587,
+      "learning_rate": 3.621033780259566e-05,
+      "loss": 0.2382,
+      "step": 602
+    },
+    {
+      "epoch": 0.38592,
+      "grad_norm": 0.3109956267991923,
+      "learning_rate": 3.616297556496369e-05,
+      "loss": 0.2374,
+      "step": 603
+    },
+    {
+      "epoch": 0.38656,
+      "grad_norm": 0.2613978553597638,
+      "learning_rate": 3.6115563232000206e-05,
+      "loss": 0.2341,
+      "step": 604
+    },
+    {
+      "epoch": 0.3872,
+      "grad_norm": 0.27418773553455345,
+      "learning_rate": 3.6068101016474314e-05,
+      "loss": 0.2364,
+      "step": 605
+    },
+    {
+      "epoch": 0.38784,
+      "grad_norm": 0.2979316952053877,
+      "learning_rate": 3.602058913137902e-05,
+      "loss": 0.2531,
+      "step": 606
+    },
+    {
+      "epoch": 0.38848,
+      "grad_norm": 0.2636629014471582,
+      "learning_rate": 3.597302778993018e-05,
+      "loss": 0.2526,
+      "step": 607
+    },
+    {
+      "epoch": 0.38912,
+      "grad_norm": 0.3085562297513792,
+      "learning_rate": 3.592541720556561e-05,
+      "loss": 0.2463,
+      "step": 608
+    },
+    {
+      "epoch": 0.38976,
+      "grad_norm": 0.28759918324320494,
+      "learning_rate": 3.5877757591944137e-05,
+      "loss": 0.2529,
+      "step": 609
+    },
+    {
+      "epoch": 0.3904,
+      "grad_norm": 0.28944413077444653,
+      "learning_rate": 3.583004916294458e-05,
+      "loss": 0.257,
+      "step": 610
+    },
+    {
+      "epoch": 0.39104,
+      "grad_norm": 0.2854970266465452,
+      "learning_rate": 3.578229213266483e-05,
+      "loss": 0.2511,
+      "step": 611
+    },
+    {
+      "epoch": 0.39168,
+      "grad_norm": 0.2898989971613222,
+      "learning_rate": 3.573448671542088e-05,
+      "loss": 0.25,
+      "step": 612
+    },
+    {
+      "epoch": 0.39232,
+      "grad_norm": 0.3053689622027871,
+      "learning_rate": 3.568663312574589e-05,
+      "loss": 0.2347,
+      "step": 613
+    },
+    {
+      "epoch": 0.39296,
+      "grad_norm": 0.2772291824702207,
+      "learning_rate": 3.563873157838917e-05,
+      "loss": 0.2575,
+      "step": 614
+    },
+    {
+      "epoch": 0.3936,
+      "grad_norm": 0.2762627572910465,
+      "learning_rate": 3.5590782288315266e-05,
+      "loss": 0.2692,
+      "step": 615
+    },
+    {
+      "epoch": 0.39424,
+      "grad_norm": 0.3077446987122632,
+      "learning_rate": 3.5542785470702965e-05,
+      "loss": 0.2571,
+      "step": 616
+    },
+    {
+      "epoch": 0.39488,
+      "grad_norm": 0.26697384631741783,
+      "learning_rate": 3.549474134094434e-05,
+      "loss": 0.259,
+      "step": 617
+    },
+    {
+      "epoch": 0.39552,
+      "grad_norm": 0.2839430840050421,
+      "learning_rate": 3.54466501146438e-05,
+      "loss": 0.2412,
+      "step": 618
+    },
+    {
+      "epoch": 0.39616,
+      "grad_norm": 0.3102851314591803,
+      "learning_rate": 3.539851200761707e-05,
+      "loss": 0.2822,
+      "step": 619
+    },
+    {
+      "epoch": 0.3968,
+      "grad_norm": 0.2747626259070342,
+      "learning_rate": 3.535032723589031e-05,
+      "loss": 0.2377,
+      "step": 620
+    },
+    {
+      "epoch": 0.39744,
+      "grad_norm": 0.31338133017581865,
+      "learning_rate": 3.5302096015699035e-05,
+      "loss": 0.2523,
+      "step": 621
+    },
+    {
+      "epoch": 0.39808,
+      "grad_norm": 0.28920982747919183,
+      "learning_rate": 3.5253818563487266e-05,
+      "loss": 0.2484,
+      "step": 622
+    },
+    {
+      "epoch": 0.39872,
+      "grad_norm": 0.287237820672606,
+      "learning_rate": 3.5205495095906444e-05,
+      "loss": 0.2497,
+      "step": 623
+    },
+    {
+      "epoch": 0.39936,
+      "grad_norm": 0.3046383103012712,
+      "learning_rate": 3.5157125829814534e-05,
+      "loss": 0.242,
+      "step": 624
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.259652296507035,
+      "learning_rate": 3.510871098227503e-05,
+      "loss": 0.2305,
+      "step": 625
+    },
+    {
+      "epoch": 0.40064,
+      "grad_norm": 0.29531410801426594,
+      "learning_rate": 3.506025077055597e-05,
+      "loss": 0.29,
+      "step": 626
+    },
+    {
+      "epoch": 0.40128,
+      "grad_norm": 0.2794704809265177,
+      "learning_rate": 3.501174541212898e-05,
+      "loss": 0.2411,
+      "step": 627
+    },
+    {
+      "epoch": 0.40192,
+      "grad_norm": 0.37622300178142065,
+      "learning_rate": 3.496319512466827e-05,
+      "loss": 0.2597,
+      "step": 628
+    },
+    {
+      "epoch": 0.40256,
+      "grad_norm": 0.27783046503482084,
+      "learning_rate": 3.4914600126049696e-05,
+      "loss": 0.2456,
+      "step": 629
+    },
+    {
+      "epoch": 0.4032,
+      "grad_norm": 0.2689454947695003,
+      "learning_rate": 3.486596063434975e-05,
+      "loss": 0.242,
+      "step": 630
+    },
+    {
+      "epoch": 0.40384,
+      "grad_norm": 0.2926461409968296,
+      "learning_rate": 3.481727686784458e-05,
+      "loss": 0.2428,
+      "step": 631
+    },
+    {
+      "epoch": 0.40448,
+      "grad_norm": 0.27170443577638215,
+      "learning_rate": 3.476854904500905e-05,
+      "loss": 0.2458,
+      "step": 632
+    },
+    {
+      "epoch": 0.40512,
+      "grad_norm": 0.281236328115951,
+      "learning_rate": 3.471977738451572e-05,
+      "loss": 0.2526,
+      "step": 633
+    },
+    {
+      "epoch": 0.40576,
+      "grad_norm": 0.29601597383700123,
+      "learning_rate": 3.467096210523387e-05,
+      "loss": 0.2516,
+      "step": 634
+    },
+    {
+      "epoch": 0.4064,
+      "grad_norm": 0.24728339864432045,
+      "learning_rate": 3.4622103426228534e-05,
+      "loss": 0.2611,
+      "step": 635
+    },
+    {
+      "epoch": 0.40704,
+      "grad_norm": 0.2812719398754191,
+      "learning_rate": 3.4573201566759515e-05,
+      "loss": 0.2448,
+      "step": 636
+    },
+    {
+      "epoch": 0.40768,
+      "grad_norm": 0.28112539299305617,
+      "learning_rate": 3.4524256746280374e-05,
+      "loss": 0.2547,
+      "step": 637
+    },
+    {
+      "epoch": 0.40832,
+      "grad_norm": 0.27735065039600865,
+      "learning_rate": 3.447526918443748e-05,
+      "loss": 0.2487,
+      "step": 638
+    },
+    {
+      "epoch": 0.40896,
+      "grad_norm": 0.2676990809540322,
+      "learning_rate": 3.442623910106901e-05,
+      "loss": 0.229,
+      "step": 639
+    },
+    {
+      "epoch": 0.4096,
+      "grad_norm": 0.2557155562658313,
+      "learning_rate": 3.4377166716203945e-05,
+      "loss": 0.2389,
+      "step": 640
+    },
+    {
+      "epoch": 0.41024,
+      "grad_norm": 0.3009667435680307,
+      "learning_rate": 3.4328052250061124e-05,
+      "loss": 0.2451,
+      "step": 641
+    },
+    {
+      "epoch": 0.41088,
+      "grad_norm": 0.29282059904319135,
+      "learning_rate": 3.4278895923048226e-05,
+      "loss": 0.2543,
+      "step": 642
+    },
+    {
+      "epoch": 0.41152,
+      "grad_norm": 0.2657135865731784,
+      "learning_rate": 3.4229697955760767e-05,
+      "loss": 0.2282,
+      "step": 643
+    },
+    {
+      "epoch": 0.41216,
+      "grad_norm": 0.2964506837049485,
+      "learning_rate": 3.4180458568981136e-05,
+      "loss": 0.2782,
+      "step": 644
+    },
+    {
+      "epoch": 0.4128,
+      "grad_norm": 0.25293381247628216,
+      "learning_rate": 3.4131177983677624e-05,
+      "loss": 0.2258,
+      "step": 645
+    },
+    {
+      "epoch": 0.41344,
+      "grad_norm": 0.2732767109062537,
+      "learning_rate": 3.4081856421003364e-05,
+      "loss": 0.2358,
+      "step": 646
+    },
+    {
+      "epoch": 0.41408,
+      "grad_norm": 0.2848748367200754,
+      "learning_rate": 3.403249410229542e-05,
+      "loss": 0.2458,
+      "step": 647
+    },
+    {
+      "epoch": 0.41472,
+      "grad_norm": 0.3027490084677138,
+      "learning_rate": 3.398309124907372e-05,
+      "loss": 0.2558,
+      "step": 648
+    },
+    {
+      "epoch": 0.41536,
+      "grad_norm": 0.24880656293330494,
+      "learning_rate": 3.393364808304012e-05,
+      "loss": 0.2711,
+      "step": 649
+    },
+    {
+      "epoch": 0.416,
+      "grad_norm": 0.29587387204867266,
+      "learning_rate": 3.3884164826077384e-05,
+      "loss": 0.2444,
+      "step": 650
+    },
+    {
+      "epoch": 0.41664,
+      "grad_norm": 0.28673436301400285,
+      "learning_rate": 3.383464170024817e-05,
+      "loss": 0.2433,
+      "step": 651
+    },
+    {
+      "epoch": 0.41728,
+      "grad_norm": 0.27821332991459735,
+      "learning_rate": 3.378507892779408e-05,
+      "loss": 0.2694,
+      "step": 652
+    },
+    {
+      "epoch": 0.41792,
+      "grad_norm": 0.3032731123141705,
+      "learning_rate": 3.37354767311346e-05,
+      "loss": 0.2526,
+      "step": 653
+    },
+    {
+      "epoch": 0.41856,
+      "grad_norm": 0.2823653961450429,
+      "learning_rate": 3.368583533286617e-05,
+      "loss": 0.2344,
+      "step": 654
+    },
+    {
+      "epoch": 0.4192,
+      "grad_norm": 0.2785547583878776,
+      "learning_rate": 3.363615495576115e-05,
+      "loss": 0.2556,
+      "step": 655
+    },
+    {
+      "epoch": 0.41984,
+      "grad_norm": 0.2740909584217494,
+      "learning_rate": 3.3586435822766785e-05,
+      "loss": 0.2556,
+      "step": 656
+    },
+    {
+      "epoch": 0.42048,
+      "grad_norm": 0.27618045966687005,
+      "learning_rate": 3.35366781570043e-05,
+      "loss": 0.2347,
+      "step": 657
+    },
+    {
+      "epoch": 0.42112,
+      "grad_norm": 0.28265761471002915,
+      "learning_rate": 3.3486882181767785e-05,
+      "loss": 0.2604,
+      "step": 658
+    },
+    {
+      "epoch": 0.42176,
+      "grad_norm": 0.32741984857383455,
+      "learning_rate": 3.34370481205233e-05,
+      "loss": 0.2607,
+      "step": 659
+    },
+    {
+      "epoch": 0.4224,
+      "grad_norm": 0.40094463121311374,
+      "learning_rate": 3.3387176196907776e-05,
+      "loss": 0.26,
+      "step": 660
+    },
+    {
+      "epoch": 0.42304,
+      "grad_norm": 0.2471679204400455,
+      "learning_rate": 3.333726663472809e-05,
+      "loss": 0.2405,
+      "step": 661
+    },
+    {
+      "epoch": 0.42368,
+      "grad_norm": 0.27705409907769524,
+      "learning_rate": 3.328731965796e-05,
+      "loss": 0.2437,
+      "step": 662
+    },
+    {
+      "epoch": 0.42432,
+      "grad_norm": 0.28284008142695716,
+      "learning_rate": 3.32373354907472e-05,
+      "loss": 0.2334,
+      "step": 663
+    },
+    {
+      "epoch": 0.42496,
+      "grad_norm": 0.31534757829717763,
+      "learning_rate": 3.3187314357400244e-05,
+      "loss": 0.2334,
+      "step": 664
+    },
+    {
+      "epoch": 0.4256,
+      "grad_norm": 0.255747984805253,
+      "learning_rate": 3.31372564823956e-05,
+      "loss": 0.2636,
+      "step": 665
+    },
+    {
+      "epoch": 0.42624,
+      "grad_norm": 0.3016619691846896,
+      "learning_rate": 3.308716209037462e-05,
+      "loss": 0.2528,
+      "step": 666
+    },
+    {
+      "epoch": 0.42688,
+      "grad_norm": 0.2657703464402086,
+      "learning_rate": 3.303703140614251e-05,
+      "loss": 0.2524,
+      "step": 667
+    },
+    {
+      "epoch": 0.42752,
+      "grad_norm": 0.2578866002212131,
+      "learning_rate": 3.2986864654667356e-05,
+      "loss": 0.2396,
+      "step": 668
+    },
+    {
+      "epoch": 0.42816,
+      "grad_norm": 0.26977079057158276,
+      "learning_rate": 3.293666206107911e-05,
+      "loss": 0.25,
+      "step": 669
+    },
+    {
+      "epoch": 0.4288,
+      "grad_norm": 0.29855679310404576,
+      "learning_rate": 3.2886423850668555e-05,
+      "loss": 0.248,
+      "step": 670
+    },
+    {
+      "epoch": 0.42944,
+      "grad_norm": 0.2848289530467174,
+      "learning_rate": 3.283615024888632e-05,
+      "loss": 0.2398,
+      "step": 671
+    },
+    {
+      "epoch": 0.43008,
+      "grad_norm": 0.3095665068158486,
+      "learning_rate": 3.278584148134185e-05,
+      "loss": 0.2668,
+      "step": 672
+    },
+    {
+      "epoch": 0.43072,
+      "grad_norm": 0.3079101456972446,
+      "learning_rate": 3.2735497773802406e-05,
+      "loss": 0.2676,
+      "step": 673
+    },
+    {
+      "epoch": 0.43136,
+      "grad_norm": 0.2928419030570455,
+      "learning_rate": 3.268511935219204e-05,
+      "loss": 0.2406,
+      "step": 674
+    },
+    {
+      "epoch": 0.432,
+      "grad_norm": 0.3049360484070473,
+      "learning_rate": 3.263470644259059e-05,
+      "loss": 0.2322,
+      "step": 675
+    },
+    {
+      "epoch": 0.43264,
+      "grad_norm": 0.30827526555844936,
+      "learning_rate": 3.258425927123266e-05,
+      "loss": 0.2654,
+      "step": 676
+    },
+    {
+      "epoch": 0.43328,
+      "grad_norm": 0.28532564145638917,
+      "learning_rate": 3.253377806450664e-05,
+      "loss": 0.2489,
+      "step": 677
+    },
+    {
+      "epoch": 0.43392,
+      "grad_norm": 0.26453578345066836,
+      "learning_rate": 3.248326304895361e-05,
+      "loss": 0.225,
+      "step": 678
+    },
+    {
+      "epoch": 0.43456,
+      "grad_norm": 1.8850807056474126,
+      "learning_rate": 3.2432714451266395e-05,
+      "loss": 0.2606,
+      "step": 679
+    },
+    {
+      "epoch": 0.4352,
+      "grad_norm": 0.2996386468952929,
+      "learning_rate": 3.238213249828853e-05,
+      "loss": 0.2585,
+      "step": 680
+    },
+    {
+      "epoch": 0.43584,
+      "grad_norm": 0.2709794240898513,
+      "learning_rate": 3.233151741701322e-05,
+      "loss": 0.25,
+      "step": 681
+    },
+    {
+      "epoch": 0.43648,
+      "grad_norm": 0.25999498342939004,
+      "learning_rate": 3.228086943458235e-05,
+      "loss": 0.2559,
+      "step": 682
+    },
+    {
+      "epoch": 0.43712,
+      "grad_norm": 0.27235652486613204,
+      "learning_rate": 3.223018877828546e-05,
+      "loss": 0.2218,
+      "step": 683
+    },
+    {
+      "epoch": 0.43776,
+      "grad_norm": 0.28214897499940533,
+      "learning_rate": 3.2179475675558694e-05,
+      "loss": 0.2453,
+      "step": 684
+    },
+    {
+      "epoch": 0.4384,
+      "grad_norm": 0.2537492618938416,
+      "learning_rate": 3.212873035398383e-05,
+      "loss": 0.2496,
+      "step": 685
+    },
+    {
+      "epoch": 0.43904,
+      "grad_norm": 0.26958757013976903,
+      "learning_rate": 3.207795304128721e-05,
+      "loss": 0.2296,
+      "step": 686
+    },
+    {
+      "epoch": 0.43968,
+      "grad_norm": 0.26472239872638276,
+      "learning_rate": 3.2027143965338765e-05,
+      "loss": 0.2486,
+      "step": 687
+    },
+    {
+      "epoch": 0.44032,
+      "grad_norm": 0.2834861966859676,
+      "learning_rate": 3.1976303354150935e-05,
+      "loss": 0.2247,
+      "step": 688
+    },
+    {
+      "epoch": 0.44096,
+      "grad_norm": 0.24678521558396363,
+      "learning_rate": 3.192543143587771e-05,
+      "loss": 0.234,
+      "step": 689
+    },
+    {
+      "epoch": 0.4416,
+      "grad_norm": 0.2542040444114577,
+      "learning_rate": 3.187452843881357e-05,
+      "loss": 0.2486,
+      "step": 690
+    },
+    {
+      "epoch": 0.44224,
+      "grad_norm": 0.25975799605879535,
+      "learning_rate": 3.182359459139243e-05,
+      "loss": 0.227,
+      "step": 691
+    },
+    {
+      "epoch": 0.44288,
+      "grad_norm": 0.27269297119593566,
+      "learning_rate": 3.1772630122186695e-05,
+      "loss": 0.2493,
+      "step": 692
+    },
+    {
+      "epoch": 0.44352,
+      "grad_norm": 0.2645382544456425,
+      "learning_rate": 3.172163525990617e-05,
+      "loss": 0.2641,
+      "step": 693
+    },
+    {
+      "epoch": 0.44416,
+      "grad_norm": 0.23256544658247483,
+      "learning_rate": 3.167061023339706e-05,
+      "loss": 0.2494,
+      "step": 694
+    },
+    {
+      "epoch": 0.4448,
+      "grad_norm": 0.2528588904829678,
+      "learning_rate": 3.161955527164092e-05,
+      "loss": 0.2508,
+      "step": 695
+    },
+    {
+      "epoch": 0.44544,
+      "grad_norm": 0.24223213746607963,
+      "learning_rate": 3.1568470603753666e-05,
+      "loss": 0.2373,
+      "step": 696
+    },
+    {
+      "epoch": 0.44608,
+      "grad_norm": 0.2706257032622334,
+      "learning_rate": 3.1517356458984485e-05,
+      "loss": 0.2484,
+      "step": 697
+    },
+    {
+      "epoch": 0.44672,
+      "grad_norm": 0.2586546974249947,
+      "learning_rate": 3.14662130667149e-05,
+      "loss": 0.2221,
+      "step": 698
+    },
+    {
+      "epoch": 0.44736,
+      "grad_norm": 0.26931097290828954,
+      "learning_rate": 3.141504065645765e-05,
+      "loss": 0.2585,
+      "step": 699
+    },
+    {
+      "epoch": 0.448,
+      "grad_norm": 0.24003468731568395,
+      "learning_rate": 3.1363839457855704e-05,
+      "loss": 0.2479,
+      "step": 700
+    },
+    {
+      "epoch": 0.44864,
+      "grad_norm": 0.2638393403519141,
+      "learning_rate": 3.131260970068122e-05,
+      "loss": 0.2476,
+      "step": 701
+    },
+    {
+      "epoch": 0.44928,
+      "grad_norm": 0.2578002678508762,
+      "learning_rate": 3.1261351614834536e-05,
+      "loss": 0.238,
+      "step": 702
+    },
+    {
+      "epoch": 0.44992,
+      "grad_norm": 0.22187445459367192,
+      "learning_rate": 3.121006543034307e-05,
+      "loss": 0.2602,
+      "step": 703
+    },
+    {
+      "epoch": 0.45056,
+      "grad_norm": 0.23788330927817766,
+      "learning_rate": 3.11587513773604e-05,
+      "loss": 0.2225,
+      "step": 704
+    },
+    {
+      "epoch": 0.4512,
+      "grad_norm": 0.3135818451008173,
+      "learning_rate": 3.1107409686165105e-05,
+      "loss": 0.2508,
+      "step": 705
+    },
+    {
+      "epoch": 0.45184,
+      "grad_norm": 0.24678268805653034,
+      "learning_rate": 3.1056040587159856e-05,
+      "loss": 0.2271,
+      "step": 706
+    },
+    {
+      "epoch": 0.45248,
+      "grad_norm": 0.24805614993691766,
+      "learning_rate": 3.100464431087027e-05,
+      "loss": 0.2353,
+      "step": 707
+    },
+    {
+      "epoch": 0.45312,
+      "grad_norm": 0.2605380120463478,
+      "learning_rate": 3.095322108794396e-05,
+      "loss": 0.2351,
+      "step": 708
+    },
+    {
+      "epoch": 0.45376,
+      "grad_norm": 0.26629710201642864,
+      "learning_rate": 3.0901771149149434e-05,
+      "loss": 0.2272,
+      "step": 709
+    },
+    {
+      "epoch": 0.4544,
+      "grad_norm": 0.25774154383126474,
+      "learning_rate": 3.0850294725375105e-05,
+      "loss": 0.2451,
+      "step": 710
+    },
+    {
+      "epoch": 0.45504,
+      "grad_norm": 0.24404499406535052,
+      "learning_rate": 3.079879204762827e-05,
+      "loss": 0.2335,
+      "step": 711
+    },
+    {
+      "epoch": 0.45568,
+      "grad_norm": 0.2543889781995685,
+      "learning_rate": 3.074726334703399e-05,
+      "loss": 0.2371,
+      "step": 712
+    },
+    {
+      "epoch": 0.45632,
+      "grad_norm": 0.25040121918120445,
+      "learning_rate": 3.069570885483414e-05,
+      "loss": 0.2497,
+      "step": 713
+    },
+    {
+      "epoch": 0.45696,
+      "grad_norm": 0.2673283982986906,
+      "learning_rate": 3.064412880238636e-05,
+      "loss": 0.2494,
+      "step": 714
+    },
+    {
+      "epoch": 0.4576,
+      "grad_norm": 0.26875525796578803,
+      "learning_rate": 3.059252342116293e-05,
+      "loss": 0.2485,
+      "step": 715
+    },
+    {
+      "epoch": 0.45824,
+      "grad_norm": 0.2535821220319592,
+      "learning_rate": 3.054089294274985e-05,
+      "loss": 0.2468,
+      "step": 716
+    },
+    {
+      "epoch": 0.45888,
+      "grad_norm": 0.2823840280098501,
+      "learning_rate": 3.048923759884573e-05,
+      "loss": 0.2361,
+      "step": 717
+    },
+    {
+      "epoch": 0.45952,
+      "grad_norm": 0.2912253075635983,
+      "learning_rate": 3.043755762126077e-05,
+      "loss": 0.2614,
+      "step": 718
+    },
+    {
+      "epoch": 0.46016,
+      "grad_norm": 0.26512542276419154,
+      "learning_rate": 3.038585324191571e-05,
+      "loss": 0.2526,
+      "step": 719
+    },
+    {
+      "epoch": 0.4608,
+      "grad_norm": 0.2632404903766759,
+      "learning_rate": 3.0334124692840815e-05,
+      "loss": 0.2555,
+      "step": 720
+    },
+    {
+      "epoch": 0.46144,
+      "grad_norm": 0.2789242686898753,
+      "learning_rate": 3.0282372206174785e-05,
+      "loss": 0.2488,
+      "step": 721
+    },
+    {
+      "epoch": 0.46208,
+      "grad_norm": 0.28018710037790234,
+      "learning_rate": 3.023059601416376e-05,
+      "loss": 0.2492,
+      "step": 722
+    },
+    {
+      "epoch": 0.46272,
+      "grad_norm": 0.23569066414193,
+      "learning_rate": 3.017879634916027e-05,
+      "loss": 0.2396,
+      "step": 723
+    },
+    {
+      "epoch": 0.46336,
+      "grad_norm": 0.23835227519804747,
+      "learning_rate": 3.012697344362217e-05,
+      "loss": 0.2345,
+      "step": 724
+    },
+    {
+      "epoch": 0.464,
+      "grad_norm": 0.27479632185865055,
+      "learning_rate": 3.0075127530111606e-05,
+      "loss": 0.2244,
+      "step": 725
+    },
+    {
+      "epoch": 0.46464,
+      "grad_norm": 0.2941303773947425,
+      "learning_rate": 3.0023258841294e-05,
+      "loss": 0.2426,
+      "step": 726
+    },
+    {
+      "epoch": 0.46528,
+      "grad_norm": 0.2619818338222505,
+      "learning_rate": 2.9971367609936945e-05,
+      "loss": 0.2525,
+      "step": 727
+    },
+    {
+      "epoch": 0.46592,
+      "grad_norm": 0.27764627835061095,
+      "learning_rate": 2.9919454068909218e-05,
+      "loss": 0.2253,
+      "step": 728
+    },
+    {
+      "epoch": 0.46656,
+      "grad_norm": 0.3095419504402447,
+      "learning_rate": 2.986751845117972e-05,
+      "loss": 0.2561,
+      "step": 729
+    },
+    {
+      "epoch": 0.4672,
+      "grad_norm": 0.26171309023552125,
+      "learning_rate": 2.981556098981641e-05,
+      "loss": 0.2803,
+      "step": 730
+    },
+    {
+      "epoch": 0.46784,
+      "grad_norm": 0.31898245088941324,
+      "learning_rate": 2.9763581917985274e-05,
+      "loss": 0.2636,
+      "step": 731
+    },
+    {
+      "epoch": 0.46848,
+      "grad_norm": 0.28172524243454883,
+      "learning_rate": 2.9711581468949295e-05,
+      "loss": 0.2575,
+      "step": 732
+    },
+    {
+      "epoch": 0.46912,
+      "grad_norm": 0.3060716712328367,
+      "learning_rate": 2.965955987606736e-05,
+      "loss": 0.2706,
+      "step": 733
+    },
+    {
+      "epoch": 0.46976,
+      "grad_norm": 0.2891706490642625,
+      "learning_rate": 2.9607517372793268e-05,
+      "loss": 0.25,
+      "step": 734
+    },
+    {
+      "epoch": 0.4704,
+      "grad_norm": 0.3014353153093821,
+      "learning_rate": 2.9555454192674638e-05,
+      "loss": 0.2501,
+      "step": 735
+    },
+    {
+      "epoch": 0.47104,
+      "grad_norm": 0.29748044849275973,
+      "learning_rate": 2.9503370569351895e-05,
+      "loss": 0.2604,
+      "step": 736
+    },
+    {
+      "epoch": 0.47168,
+      "grad_norm": 0.2710433658776984,
+      "learning_rate": 2.9451266736557202e-05,
+      "loss": 0.2575,
+      "step": 737
+    },
+    {
+      "epoch": 0.47232,
+      "grad_norm": 0.26561072840993083,
+      "learning_rate": 2.9399142928113403e-05,
+      "loss": 0.2193,
+      "step": 738
+    },
+    {
+      "epoch": 0.47296,
+      "grad_norm": 0.2656840901800972,
+      "learning_rate": 2.9346999377932992e-05,
+      "loss": 0.2625,
+      "step": 739
+    },
+    {
+      "epoch": 0.4736,
+      "grad_norm": 0.2589355572460451,
+      "learning_rate": 2.929483632001706e-05,
+      "loss": 0.2406,
+      "step": 740
+    },
+    {
+      "epoch": 0.47424,
+      "grad_norm": 0.28626147498332316,
+      "learning_rate": 2.9242653988454237e-05,
+      "loss": 0.2438,
+      "step": 741
+    },
+    {
+      "epoch": 0.47488,
+      "grad_norm": 0.2875213940525657,
+      "learning_rate": 2.919045261741965e-05,
+      "loss": 0.238,
+      "step": 742
+    },
+    {
+      "epoch": 0.47552,
+      "grad_norm": 0.2653417159784572,
+      "learning_rate": 2.9138232441173867e-05,
+      "loss": 0.2382,
+      "step": 743
+    },
+    {
+      "epoch": 0.47616,
+      "grad_norm": 0.2979013673466959,
+      "learning_rate": 2.9085993694061853e-05,
+      "loss": 0.2419,
+      "step": 744
+    },
+    {
+      "epoch": 0.4768,
+      "grad_norm": 0.2885262466043221,
+      "learning_rate": 2.9033736610511887e-05,
+      "loss": 0.2613,
+      "step": 745
+    },
+    {
+      "epoch": 0.47744,
+      "grad_norm": 0.2866061321732059,
+      "learning_rate": 2.898146142503456e-05,
+      "loss": 0.2631,
+      "step": 746
+    },
+    {
+      "epoch": 0.47808,
+      "grad_norm": 0.26544609876731556,
+      "learning_rate": 2.8929168372221693e-05,
+      "loss": 0.2491,
+      "step": 747
+    },
+    {
+      "epoch": 0.47872,
+      "grad_norm": 0.3034821126240335,
+      "learning_rate": 2.88768576867453e-05,
+      "loss": 0.2555,
+      "step": 748
+    },
+    {
+      "epoch": 0.47936,
+      "grad_norm": 0.24983636855605562,
+      "learning_rate": 2.8824529603356497e-05,
+      "loss": 0.2225,
+      "step": 749
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.30066227757911895,
+      "learning_rate": 2.8772184356884496e-05,
+      "loss": 0.2592,
+      "step": 750
+    },
+    {
+      "epoch": 0.48064,
+      "grad_norm": 0.239130352583496,
+      "learning_rate": 2.8719822182235516e-05,
+      "loss": 0.255,
+      "step": 751
+    },
+    {
+      "epoch": 0.48128,
+      "grad_norm": 0.27667965605507494,
+      "learning_rate": 2.8667443314391762e-05,
+      "loss": 0.2289,
+      "step": 752
+    },
+    {
+      "epoch": 0.48192,
+      "grad_norm": 0.2810394687988481,
+      "learning_rate": 2.8615047988410337e-05,
+      "loss": 0.2425,
+      "step": 753
+    },
+    {
+      "epoch": 0.48256,
+      "grad_norm": 0.2518451996136543,
+      "learning_rate": 2.8562636439422202e-05,
+      "loss": 0.2424,
+      "step": 754
+    },
+    {
+      "epoch": 0.4832,
+      "grad_norm": 0.2748846728182042,
+      "learning_rate": 2.8510208902631135e-05,
+      "loss": 0.2504,
+      "step": 755
+    },
+    {
+      "epoch": 0.48384,
+      "grad_norm": 0.2575000327748514,
+      "learning_rate": 2.845776561331265e-05,
+      "loss": 0.252,
+      "step": 756
+    },
+    {
+      "epoch": 0.48448,
+      "grad_norm": 0.2864089085188715,
+      "learning_rate": 2.8405306806812947e-05,
+      "loss": 0.2237,
+      "step": 757
+    },
+    {
+      "epoch": 0.48512,
+      "grad_norm": 0.27755553898956253,
+      "learning_rate": 2.835283271854788e-05,
+      "loss": 0.2614,
+      "step": 758
+    },
+    {
+      "epoch": 0.48576,
+      "grad_norm": 0.25491803925314754,
+      "learning_rate": 2.8300343584001853e-05,
+      "loss": 0.2451,
+      "step": 759
+    },
+    {
+      "epoch": 0.4864,
+      "grad_norm": 0.2575327130038747,
+      "learning_rate": 2.8247839638726825e-05,
+      "loss": 0.2587,
+      "step": 760
+    },
+    {
+      "epoch": 0.48704,
+      "grad_norm": 0.261721402424143,
+      "learning_rate": 2.8195321118341204e-05,
+      "loss": 0.2397,
+      "step": 761
+    },
+    {
+      "epoch": 0.48768,
+      "grad_norm": 0.2604152725714389,
+      "learning_rate": 2.8142788258528813e-05,
+      "loss": 0.2567,
+      "step": 762
+    },
+    {
+      "epoch": 0.48832,
+      "grad_norm": 0.30571357627918333,
+      "learning_rate": 2.8090241295037804e-05,
+      "loss": 0.2527,
+      "step": 763
+    },
+    {
+      "epoch": 0.48896,
+      "grad_norm": 0.267067229766143,
+      "learning_rate": 2.8037680463679648e-05,
+      "loss": 0.2466,
+      "step": 764
+    },
+    {
+      "epoch": 0.4896,
+      "grad_norm": 0.25538920312770114,
+      "learning_rate": 2.7985106000328033e-05,
+      "loss": 0.2398,
+      "step": 765
+    },
+    {
+      "epoch": 0.49024,
+      "grad_norm": 0.275440862373655,
+      "learning_rate": 2.793251814091782e-05,
+      "loss": 0.2366,
+      "step": 766
+    },
+    {
+      "epoch": 0.49088,
+      "grad_norm": 0.28328615128607115,
+      "learning_rate": 2.787991712144401e-05,
+      "loss": 0.2588,
+      "step": 767
+    },
+    {
+      "epoch": 0.49152,
+      "grad_norm": 0.26254008936662954,
+      "learning_rate": 2.7827303177960638e-05,
+      "loss": 0.2653,
+      "step": 768
+    },
+    {
+      "epoch": 0.49216,
+      "grad_norm": 0.2615112810927122,
+      "learning_rate": 2.7774676546579737e-05,
+      "loss": 0.2494,
+      "step": 769
+    },
+    {
+      "epoch": 0.4928,
+      "grad_norm": 0.2587877680106713,
+      "learning_rate": 2.7722037463470297e-05,
+      "loss": 0.2304,
+      "step": 770
+    },
+    {
+      "epoch": 0.49344,
+      "grad_norm": 0.24897684478475532,
+      "learning_rate": 2.7669386164857174e-05,
+      "loss": 0.2157,
+      "step": 771
+    },
+    {
+      "epoch": 0.49408,
+      "grad_norm": 0.25826790808429917,
+      "learning_rate": 2.7616722887020034e-05,
+      "loss": 0.2308,
+      "step": 772
+    },
+    {
+      "epoch": 0.49472,
+      "grad_norm": 0.29862067935293035,
+      "learning_rate": 2.7564047866292336e-05,
+      "loss": 0.234,
+      "step": 773
+    },
+    {
+      "epoch": 0.49536,
+      "grad_norm": 0.25612992651373906,
+      "learning_rate": 2.7511361339060193e-05,
+      "loss": 0.2569,
+      "step": 774
+    },
+    {
+      "epoch": 0.496,
+      "grad_norm": 0.24860228212554142,
+      "learning_rate": 2.7458663541761376e-05,
+      "loss": 0.2531,
+      "step": 775
+    },
+    {
+      "epoch": 0.49664,
+      "grad_norm": 0.25357670328115006,
+      "learning_rate": 2.740595471088424e-05,
+      "loss": 0.2326,
+      "step": 776
+    },
+    {
+      "epoch": 0.49728,
+      "grad_norm": 0.279379640684078,
+      "learning_rate": 2.735323508296664e-05,
+      "loss": 0.2537,
+      "step": 777
+    },
+    {
+      "epoch": 0.49792,
+      "grad_norm": 0.26060823723229753,
+      "learning_rate": 2.7300504894594892e-05,
+      "loss": 0.2413,
+      "step": 778
+    },
+    {
+      "epoch": 0.49856,
+      "grad_norm": 0.24657503031180159,
+      "learning_rate": 2.7247764382402696e-05,
+      "loss": 0.2687,
+      "step": 779
+    },
+    {
+      "epoch": 0.4992,
+      "grad_norm": 0.24758999823605377,
+      "learning_rate": 2.7195013783070093e-05,
+      "loss": 0.262,
+      "step": 780
+    },
+    {
+      "epoch": 0.49984,
+      "grad_norm": 0.2637048834360739,
+      "learning_rate": 2.7142253333322387e-05,
+      "loss": 0.2477,
+      "step": 781
+    },
+    {
+      "epoch": 0.50048,
+      "grad_norm": 0.2430328229153265,
+      "learning_rate": 2.7089483269929074e-05,
+      "loss": 0.2502,
+      "step": 782
+    },
+    {
+      "epoch": 0.50112,
+      "grad_norm": 0.2387247282546657,
+      "learning_rate": 2.703670382970281e-05,
+      "loss": 0.2342,
+      "step": 783
+    },
+    {
+      "epoch": 0.50176,
+      "grad_norm": 0.24498378810652757,
+      "learning_rate": 2.698391524949832e-05,
+      "loss": 0.239,
+      "step": 784
+    },
+    {
+      "epoch": 0.5024,
+      "grad_norm": 0.2576589386419526,
+      "learning_rate": 2.693111776621136e-05,
+      "loss": 0.2391,
+      "step": 785
+    },
+    {
+      "epoch": 0.50304,
+      "grad_norm": 0.238017107988676,
+      "learning_rate": 2.6878311616777618e-05,
+      "loss": 0.2453,
+      "step": 786
+    },
+    {
+      "epoch": 0.50368,
+      "grad_norm": 0.2686878709228769,
+      "learning_rate": 2.68254970381717e-05,
+      "loss": 0.2433,
+      "step": 787
+    },
+    {
+      "epoch": 0.50432,
+      "grad_norm": 0.24828872753485792,
+      "learning_rate": 2.6772674267406005e-05,
+      "loss": 0.2617,
+      "step": 788
+    },
+    {
+      "epoch": 0.50496,
+      "grad_norm": 0.24365345279692238,
+      "learning_rate": 2.6719843541529727e-05,
+      "loss": 0.2533,
+      "step": 789
+    },
+    {
+      "epoch": 0.5056,
+      "grad_norm": 0.2738955694271426,
+      "learning_rate": 2.666700509762774e-05,
+      "loss": 0.2335,
+      "step": 790
+    },
+    {
+      "epoch": 0.50624,
+      "grad_norm": 0.2691694035690393,
+      "learning_rate": 2.6614159172819564e-05,
+      "loss": 0.2501,
+      "step": 791
+    },
+    {
+      "epoch": 0.50688,
+      "grad_norm": 0.2659614704915682,
+      "learning_rate": 2.6561306004258284e-05,
+      "loss": 0.2533,
+      "step": 792
+    },
+    {
+      "epoch": 0.50752,
+      "grad_norm": 0.25738827683610904,
+      "learning_rate": 2.6508445829129498e-05,
+      "loss": 0.2549,
+      "step": 793
+    },
+    {
+      "epoch": 0.50816,
+      "grad_norm": 0.2522328385594724,
+      "learning_rate": 2.645557888465024e-05,
+      "loss": 0.2258,
+      "step": 794
+    },
+    {
+      "epoch": 0.5088,
+      "grad_norm": 0.2788448307045321,
+      "learning_rate": 2.640270540806793e-05,
+      "loss": 0.2311,
+      "step": 795
+    },
+    {
+      "epoch": 0.50944,
+      "grad_norm": 0.28362139103692846,
+      "learning_rate": 2.6349825636659292e-05,
+      "loss": 0.2737,
+      "step": 796
+    },
+    {
+      "epoch": 0.51008,
+      "grad_norm": 0.26044366510217265,
+      "learning_rate": 2.629693980772931e-05,
+      "loss": 0.2383,
+      "step": 797
+    },
+    {
+      "epoch": 0.51072,
+      "grad_norm": 0.2535298378133926,
+      "learning_rate": 2.6244048158610148e-05,
+      "loss": 0.2509,
+      "step": 798
+    },
+    {
+      "epoch": 0.51136,
+      "grad_norm": 0.27174467067095714,
+      "learning_rate": 2.619115092666008e-05,
+      "loss": 0.2311,
+      "step": 799
+    },
+    {
+      "epoch": 0.512,
+      "grad_norm": 0.28573827812151115,
+      "learning_rate": 2.6138248349262446e-05,
+      "loss": 0.2463,
+      "step": 800
+    },
+    {
+      "epoch": 0.51264,
+      "grad_norm": 0.2524514106134234,
+      "learning_rate": 2.608534066382457e-05,
+      "loss": 0.2574,
+      "step": 801
+    },
+    {
+      "epoch": 0.51328,
+      "grad_norm": 0.23359202806831134,
+      "learning_rate": 2.60324281077767e-05,
+      "loss": 0.2291,
+      "step": 802
+    },
+    {
+      "epoch": 0.51392,
+      "grad_norm": 0.2632376849295831,
+      "learning_rate": 2.5979510918570943e-05,
+      "loss": 0.2273,
+      "step": 803
+    },
+    {
+      "epoch": 0.51456,
+      "grad_norm": 0.27408167792055305,
+      "learning_rate": 2.592658933368019e-05,
+      "loss": 0.238,
+      "step": 804
+    },
+    {
+      "epoch": 0.5152,
+      "grad_norm": 0.26117111640944357,
+      "learning_rate": 2.5873663590597065e-05,
+      "loss": 0.2212,
+      "step": 805
+    },
+    {
+      "epoch": 0.51584,
+      "grad_norm": 0.23066506444669752,
+      "learning_rate": 2.582073392683284e-05,
+      "loss": 0.2337,
+      "step": 806
+    },
+    {
+      "epoch": 0.51648,
+      "grad_norm": 0.26350559927947886,
+      "learning_rate": 2.5767800579916412e-05,
+      "loss": 0.2318,
+      "step": 807
+    },
+    {
+      "epoch": 0.51712,
+      "grad_norm": 0.2829786329932492,
+      "learning_rate": 2.5714863787393173e-05,
+      "loss": 0.2403,
+      "step": 808
+    },
+    {
+      "epoch": 0.51776,
+      "grad_norm": 0.2585327918775559,
+      "learning_rate": 2.5661923786823994e-05,
+      "loss": 0.2163,
+      "step": 809
+    },
+    {
+      "epoch": 0.5184,
+      "grad_norm": 0.2603425486862888,
+      "learning_rate": 2.5608980815784144e-05,
+      "loss": 0.2473,
+      "step": 810
+    },
+    {
+      "epoch": 0.51904,
+      "grad_norm": 0.23823317474679367,
+      "learning_rate": 2.5556035111862213e-05,
+      "loss": 0.2439,
+      "step": 811
+    },
+    {
+      "epoch": 0.51968,
+      "grad_norm": 0.25473347064625185,
+      "learning_rate": 2.550308691265907e-05,
+      "loss": 0.2547,
+      "step": 812
+    },
+    {
+      "epoch": 0.52032,
+      "grad_norm": 0.2865834038258944,
+      "learning_rate": 2.545013645578676e-05,
+      "loss": 0.2576,
+      "step": 813
+    },
+    {
+      "epoch": 0.52096,
+      "grad_norm": 0.256898434573177,
+      "learning_rate": 2.5397183978867484e-05,
+      "loss": 0.2379,
+      "step": 814
+    },
+    {
+      "epoch": 0.5216,
+      "grad_norm": 0.2472713373039885,
+      "learning_rate": 2.5344229719532486e-05,
+      "loss": 0.24,
+      "step": 815
+    },
+    {
+      "epoch": 0.52224,
+      "grad_norm": 0.2665434723354754,
+      "learning_rate": 2.529127391542103e-05,
+      "loss": 0.2244,
+      "step": 816
+    },
+    {
+      "epoch": 0.52288,
+      "grad_norm": 0.28635358929082144,
+      "learning_rate": 2.5238316804179296e-05,
+      "loss": 0.2441,
+      "step": 817
+    },
+    {
+      "epoch": 0.52352,
+      "grad_norm": 0.2928363177711813,
+      "learning_rate": 2.5185358623459332e-05,
+      "loss": 0.2346,
+      "step": 818
+    },
+    {
+      "epoch": 0.52416,
+      "grad_norm": 0.2510140453596736,
+      "learning_rate": 2.513239961091799e-05,
+      "loss": 0.2161,
+      "step": 819
+    },
+    {
+      "epoch": 0.5248,
+      "grad_norm": 0.2505869007452204,
+      "learning_rate": 2.507944000421586e-05,
+      "loss": 0.2252,
+      "step": 820
+    },
+    {
+      "epoch": 0.52544,
+      "grad_norm": 0.2412865110838159,
+      "learning_rate": 2.502648004101619e-05,
+      "loss": 0.2342,
+      "step": 821
+    },
+    {
+      "epoch": 0.52608,
+      "grad_norm": 0.2759517968691516,
+      "learning_rate": 2.497351995898382e-05,
+      "loss": 0.2383,
+      "step": 822
+    },
+    {
+      "epoch": 0.52672,
+      "grad_norm": 0.24618299352546902,
+      "learning_rate": 2.492055999578415e-05,
+      "loss": 0.245,
+      "step": 823
+    },
+    {
+      "epoch": 0.52736,
+      "grad_norm": 0.2520594945771149,
+      "learning_rate": 2.486760038908201e-05,
+      "loss": 0.2348,
+      "step": 824
+    },
+    {
+      "epoch": 0.528,
+      "grad_norm": 0.24673054113925869,
+      "learning_rate": 2.481464137654068e-05,
+      "loss": 0.256,
+      "step": 825
+    },
+    {
+      "epoch": 0.52864,
+      "grad_norm": 0.24198537637736509,
+      "learning_rate": 2.476168319582071e-05,
+      "loss": 0.2262,
+      "step": 826
+    },
+    {
+      "epoch": 0.52928,
+      "grad_norm": 0.2486757115879481,
+      "learning_rate": 2.4708726084578978e-05,
+      "loss": 0.2343,
+      "step": 827
+    },
+    {
+      "epoch": 0.52992,
+      "grad_norm": 0.24831100008566848,
+      "learning_rate": 2.465577028046752e-05,
+      "loss": 0.2422,
+      "step": 828
+    },
+    {
+      "epoch": 0.53056,
+      "grad_norm": 0.2511420700055537,
+      "learning_rate": 2.4602816021132515e-05,
+      "loss": 0.2281,
+      "step": 829
+    },
+    {
+      "epoch": 0.5312,
+      "grad_norm": 0.2506359764390447,
+      "learning_rate": 2.4549863544213243e-05,
+      "loss": 0.2399,
+      "step": 830
+    },
+    {
+      "epoch": 0.53184,
+      "grad_norm": 0.2739608567499247,
+      "learning_rate": 2.449691308734093e-05,
+      "loss": 0.2708,
+      "step": 831
+    },
+    {
+      "epoch": 0.53248,
+      "grad_norm": 0.25435131352533596,
+      "learning_rate": 2.444396488813779e-05,
+      "loss": 0.2453,
+      "step": 832
+    },
+    {
+      "epoch": 0.53312,
+      "grad_norm": 0.2506814737038616,
+      "learning_rate": 2.4391019184215862e-05,
+      "loss": 0.2858,
+      "step": 833
+    },
+    {
+      "epoch": 0.53376,
+      "grad_norm": 0.2622416583953065,
+      "learning_rate": 2.4338076213176012e-05,
+      "loss": 0.2375,
+      "step": 834
+    },
+    {
+      "epoch": 0.5344,
+      "grad_norm": 0.2646591197170153,
+      "learning_rate": 2.4285136212606836e-05,
+      "loss": 0.2235,
+      "step": 835
+    },
+    {
+      "epoch": 0.53504,
+      "grad_norm": 0.2464672456549471,
+      "learning_rate": 2.42321994200836e-05,
+      "loss": 0.2299,
+      "step": 836
+    },
+    {
+      "epoch": 0.53568,
+      "grad_norm": 0.26837339676461225,
+      "learning_rate": 2.417926607316716e-05,
+      "loss": 0.244,
+      "step": 837
+    },
+    {
+      "epoch": 0.53632,
+      "grad_norm": 0.25488064569499774,
+      "learning_rate": 2.412633640940295e-05,
+      "loss": 0.224,
+      "step": 838
+    },
+    {
+      "epoch": 0.53696,
+      "grad_norm": 0.24716442364401292,
+      "learning_rate": 2.4073410666319816e-05,
+      "loss": 0.2469,
+      "step": 839
+    },
+    {
+      "epoch": 0.5376,
+      "grad_norm": 0.2461377335096662,
+      "learning_rate": 2.4020489081429063e-05,
+      "loss": 0.2547,
+      "step": 840
+    },
+    {
+      "epoch": 0.53824,
+      "grad_norm": 0.24784331913754212,
+      "learning_rate": 2.3967571892223305e-05,
+      "loss": 0.2227,
+      "step": 841
+    },
+    {
+      "epoch": 0.53888,
+      "grad_norm": 0.23224692543862904,
+      "learning_rate": 2.3914659336175428e-05,
+      "loss": 0.2279,
+      "step": 842
+    },
+    {
+      "epoch": 0.53952,
+      "grad_norm": 0.26257276128421314,
+      "learning_rate": 2.386175165073756e-05,
+      "loss": 0.229,
+      "step": 843
+    },
+    {
+      "epoch": 0.54016,
+      "grad_norm": 0.2516791452407145,
+      "learning_rate": 2.3808849073339923e-05,
+      "loss": 0.2359,
+      "step": 844
+    },
+    {
+      "epoch": 0.5408,
+      "grad_norm": 0.2846303223760566,
+      "learning_rate": 2.375595184138986e-05,
+      "loss": 0.2402,
+      "step": 845
+    },
+    {
+      "epoch": 0.54144,
+      "grad_norm": 0.21738836142783174,
+      "learning_rate": 2.3703060192270696e-05,
+      "loss": 0.2314,
+      "step": 846
+    },
+    {
+      "epoch": 0.54208,
+      "grad_norm": 0.24056602524848011,
+      "learning_rate": 2.365017436334072e-05,
+      "loss": 0.2301,
+      "step": 847
+    },
+    {
+      "epoch": 0.54272,
+      "grad_norm": 0.2548013163769387,
+      "learning_rate": 2.3597294591932077e-05,
+      "loss": 0.2309,
+      "step": 848
+    },
+    {
+      "epoch": 0.54336,
+      "grad_norm": 0.250069732887463,
+      "learning_rate": 2.3544421115349772e-05,
+      "loss": 0.23,
+      "step": 849
+    },
+    {
+      "epoch": 0.544,
+      "grad_norm": 0.24781490415728508,
+      "learning_rate": 2.3491554170870504e-05,
+      "loss": 0.2133,
+      "step": 850
+    },
+    {
+      "epoch": 0.54464,
+      "grad_norm": 0.2730547103955746,
+      "learning_rate": 2.343869399574172e-05,
+      "loss": 0.2326,
+      "step": 851
+    },
+    {
+      "epoch": 0.54528,
+      "grad_norm": 0.2633593973225614,
+      "learning_rate": 2.3385840827180445e-05,
+      "loss": 0.2395,
+      "step": 852
+    },
+    {
+      "epoch": 0.54592,
+      "grad_norm": 0.29374562628079703,
+      "learning_rate": 2.333299490237226e-05,
+      "loss": 0.2502,
+      "step": 853
+    },
+    {
+      "epoch": 0.54656,
+      "grad_norm": 0.26333185529562214,
+      "learning_rate": 2.3280156458470282e-05,
+      "loss": 0.2384,
+      "step": 854
+    },
+    {
+      "epoch": 0.5472,
+      "grad_norm": 0.2568974141440104,
+      "learning_rate": 2.3227325732593997e-05,
+      "loss": 0.2337,
+      "step": 855
+    },
+    {
+      "epoch": 0.54784,
+      "grad_norm": 0.26984762378876725,
+      "learning_rate": 2.317450296182831e-05,
+      "loss": 0.2517,
+      "step": 856
+    },
+    {
+      "epoch": 0.54848,
+      "grad_norm": 0.27656020548021637,
+      "learning_rate": 2.3121688383222384e-05,
+      "loss": 0.2259,
+      "step": 857
+    },
+    {
+      "epoch": 0.54912,
+      "grad_norm": 0.2565724185389461,
+      "learning_rate": 2.306888223378865e-05,
+      "loss": 0.2451,
+      "step": 858
+    },
+    {
+      "epoch": 0.54976,
+      "grad_norm": 0.2506884090771504,
+      "learning_rate": 2.3016084750501685e-05,
+      "loss": 0.2493,
+      "step": 859
+    },
+    {
+      "epoch": 0.5504,
+      "grad_norm": 0.25851345241711754,
+      "learning_rate": 2.2963296170297204e-05,
+      "loss": 0.242,
+      "step": 860
+    },
+    {
+      "epoch": 0.55104,
+      "grad_norm": 0.2547749959744443,
+      "learning_rate": 2.2910516730070932e-05,
+      "loss": 0.25,
+      "step": 861
+    },
+    {
+      "epoch": 0.55168,
+      "grad_norm": 0.25007636920036586,
+      "learning_rate": 2.2857746666677626e-05,
+      "loss": 0.2113,
+      "step": 862
+    },
+    {
+      "epoch": 0.55232,
+      "grad_norm": 0.24092893007681362,
+      "learning_rate": 2.280498621692991e-05,
+      "loss": 0.2331,
+      "step": 863
+    },
+    {
+      "epoch": 0.55296,
+      "grad_norm": 0.23879978749748068,
+      "learning_rate": 2.275223561759731e-05,
+      "loss": 0.2455,
+      "step": 864
+    },
+    {
+      "epoch": 0.5536,
+      "grad_norm": 0.26997174021611564,
+      "learning_rate": 2.2699495105405117e-05,
+      "loss": 0.2408,
+      "step": 865
+    },
+    {
+      "epoch": 0.55424,
+      "grad_norm": 0.26533879574969516,
+      "learning_rate": 2.264676491703336e-05,
+      "loss": 0.2272,
+      "step": 866
+    },
+    {
+      "epoch": 0.55488,
+      "grad_norm": 0.26946982936175096,
+      "learning_rate": 2.2594045289115764e-05,
+      "loss": 0.225,
+      "step": 867
+    },
+    {
+      "epoch": 0.55552,
+      "grad_norm": 0.2475422118280942,
+      "learning_rate": 2.254133645823862e-05,
+      "loss": 0.2313,
+      "step": 868
+    },
+    {
+      "epoch": 0.55616,
+      "grad_norm": 0.2521763277366927,
+      "learning_rate": 2.2488638660939816e-05,
+      "loss": 0.2488,
+      "step": 869
+    },
+    {
+      "epoch": 0.5568,
+      "grad_norm": 0.2586514287872215,
+      "learning_rate": 2.243595213370767e-05,
+      "loss": 0.2338,
+      "step": 870
+    },
+    {
+      "epoch": 0.55744,
+      "grad_norm": 0.2569681167480472,
+      "learning_rate": 2.238327711297997e-05,
+      "loss": 0.2456,
+      "step": 871
+    },
+    {
+      "epoch": 0.55808,
+      "grad_norm": 0.23822134052488086,
+      "learning_rate": 2.2330613835142832e-05,
+      "loss": 0.239,
+      "step": 872
+    },
+    {
+      "epoch": 0.55872,
+      "grad_norm": 0.2676596591540124,
+      "learning_rate": 2.2277962536529712e-05,
+      "loss": 0.2142,
+      "step": 873
+    },
+    {
+      "epoch": 0.55936,
+      "grad_norm": 0.2651735266943684,
+      "learning_rate": 2.2225323453420265e-05,
+      "loss": 0.2206,
+      "step": 874
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.254593698215357,
+      "learning_rate": 2.2172696822039374e-05,
+      "loss": 0.239,
+      "step": 875
+    },
+    {
+      "epoch": 0.56064,
+      "grad_norm": 0.23065667281364877,
+      "learning_rate": 2.2120082878556e-05,
+      "loss": 0.22,
+      "step": 876
+    },
+    {
+      "epoch": 0.56128,
+      "grad_norm": 0.26936624455865477,
+      "learning_rate": 2.206748185908219e-05,
+      "loss": 0.2172,
+      "step": 877
+    },
+    {
+      "epoch": 0.56192,
+      "grad_norm": 0.2660000172403738,
+      "learning_rate": 2.2014893999671976e-05,
+      "loss": 0.2522,
+      "step": 878
+    },
+    {
+      "epoch": 0.56256,
+      "grad_norm": 0.2686105246494615,
+      "learning_rate": 2.1962319536320354e-05,
+      "loss": 0.2434,
+      "step": 879
+    },
+    {
+      "epoch": 0.5632,
+      "grad_norm": 0.2699637501013043,
+      "learning_rate": 2.1909758704962198e-05,
+      "loss": 0.215,
+      "step": 880
+    },
+    {
+      "epoch": 0.56384,
+      "grad_norm": 0.24387692811654585,
+      "learning_rate": 2.185721174147119e-05,
+      "loss": 0.227,
+      "step": 881
+    },
+    {
+      "epoch": 0.56448,
+      "grad_norm": 0.25969535364312796,
+      "learning_rate": 2.1804678881658798e-05,
+      "loss": 0.2298,
+      "step": 882
+    },
+    {
+      "epoch": 0.56512,
+      "grad_norm": 0.24997828503698874,
+      "learning_rate": 2.1752160361273174e-05,
+      "loss": 0.2285,
+      "step": 883
+    },
+    {
+      "epoch": 0.56576,
+      "grad_norm": 0.25925241972788576,
+      "learning_rate": 2.1699656415998153e-05,
+      "loss": 0.2257,
+      "step": 884
+    },
+    {
+      "epoch": 0.5664,
+      "grad_norm": 0.29355764733158524,
+      "learning_rate": 2.164716728145213e-05,
+      "loss": 0.2564,
+      "step": 885
+    },
+    {
+      "epoch": 0.56704,
+      "grad_norm": 0.2446641321481941,
+      "learning_rate": 2.1594693193187062e-05,
+      "loss": 0.2338,
+      "step": 886
+    },
+    {
+      "epoch": 0.56768,
+      "grad_norm": 0.24666251680979395,
+      "learning_rate": 2.1542234386687354e-05,
+      "loss": 0.243,
+      "step": 887
+    },
+    {
+      "epoch": 0.56832,
+      "grad_norm": 0.2500831302456269,
+      "learning_rate": 2.148979109736887e-05,
+      "loss": 0.2552,
+      "step": 888
+    },
+    {
+      "epoch": 0.56896,
+      "grad_norm": 0.24824818134295293,
+      "learning_rate": 2.1437363560577803e-05,
+      "loss": 0.2376,
+      "step": 889
+    },
+    {
+      "epoch": 0.5696,
+      "grad_norm": 0.2405968939613267,
+      "learning_rate": 2.1384952011589675e-05,
+      "loss": 0.2514,
+      "step": 890
+    },
+    {
+      "epoch": 0.57024,
+      "grad_norm": 0.25020717472381443,
+      "learning_rate": 2.1332556685608243e-05,
+      "loss": 0.2507,
+      "step": 891
+    },
+    {
+      "epoch": 0.57088,
+      "grad_norm": 0.23931013256902683,
+      "learning_rate": 2.128017781776448e-05,
+      "loss": 0.2309,
+      "step": 892
+    },
+    {
+      "epoch": 0.57152,
+      "grad_norm": 0.23336680019782421,
+      "learning_rate": 2.122781564311551e-05,
+      "loss": 0.2274,
+      "step": 893
+    },
+    {
+      "epoch": 0.57216,
+      "grad_norm": 0.24586358262613148,
+      "learning_rate": 2.1175470396643505e-05,
+      "loss": 0.2351,
+      "step": 894
+    },
+    {
+      "epoch": 0.5728,
+      "grad_norm": 0.26137793431193335,
+      "learning_rate": 2.1123142313254708e-05,
+      "loss": 0.2389,
+      "step": 895
+    },
+    {
+      "epoch": 0.57344,
+      "grad_norm": 0.2514533566692427,
+      "learning_rate": 2.1070831627778303e-05,
+      "loss": 0.2502,
+      "step": 896
+    },
+    {
+      "epoch": 0.57408,
+      "grad_norm": 0.267858915846267,
+      "learning_rate": 2.101853857496545e-05,
+      "loss": 0.2414,
+      "step": 897
+    },
+    {
+      "epoch": 0.57472,
+      "grad_norm": 0.24954310970649327,
+      "learning_rate": 2.0966263389488123e-05,
+      "loss": 0.2327,
+      "step": 898
+    },
+    {
+      "epoch": 0.57536,
+      "grad_norm": 0.30833194653356444,
+      "learning_rate": 2.091400630593816e-05,
+      "loss": 0.2105,
+      "step": 899
+    },
+    {
+      "epoch": 0.576,
+      "grad_norm": 0.2567906619127888,
+      "learning_rate": 2.086176755882614e-05,
+      "loss": 0.2044,
+      "step": 900
+    },
+    {
+      "epoch": 0.57664,
+      "grad_norm": 0.3040689476202613,
+      "learning_rate": 2.0809547382580357e-05,
+      "loss": 0.2488,
+      "step": 901
+    },
+    {
+      "epoch": 0.57728,
+      "grad_norm": 0.6702580409718877,
+      "learning_rate": 2.075734601154577e-05,
+      "loss": 0.2289,
+      "step": 902
+    },
+    {
+      "epoch": 0.57792,
+      "grad_norm": 0.27548502053540713,
+      "learning_rate": 2.070516367998294e-05,
+      "loss": 0.2196,
+      "step": 903
+    },
+    {
+      "epoch": 0.57856,
+      "grad_norm": 0.28339543270290896,
+      "learning_rate": 2.065300062206701e-05,
+      "loss": 0.2394,
+      "step": 904
+    },
+    {
+      "epoch": 0.5792,
+      "grad_norm": 0.2459179838809575,
+      "learning_rate": 2.06008570718866e-05,
+      "loss": 0.2262,
+      "step": 905
+    },
+    {
+      "epoch": 0.57984,
+      "grad_norm": 0.25744829749446557,
+      "learning_rate": 2.0548733263442804e-05,
+      "loss": 0.2312,
+      "step": 906
+    },
+    {
+      "epoch": 0.58048,
+      "grad_norm": 0.2515079802068137,
+      "learning_rate": 2.04966294306481e-05,
+      "loss": 0.2419,
+      "step": 907
+    },
+    {
+      "epoch": 0.58112,
+      "grad_norm": 0.2342577862187012,
+      "learning_rate": 2.0444545807325364e-05,
+      "loss": 0.2372,
+      "step": 908
+    },
+    {
+      "epoch": 0.58176,
+      "grad_norm": 0.2599855250321663,
+      "learning_rate": 2.0392482627206734e-05,
+      "loss": 0.2456,
+      "step": 909
+    },
+    {
+      "epoch": 0.5824,
+      "grad_norm": 0.2320933313979261,
+      "learning_rate": 2.034044012393265e-05,
+      "loss": 0.2276,
+      "step": 910
+    },
+    {
+      "epoch": 0.58304,
+      "grad_norm": 0.23465739358860568,
+      "learning_rate": 2.028841853105071e-05,
+      "loss": 0.217,
+      "step": 911
+    },
+    {
+      "epoch": 0.58368,
+      "grad_norm": 0.23450613382192725,
+      "learning_rate": 2.0236418082014728e-05,
+      "loss": 0.2314,
+      "step": 912
+    },
+    {
+      "epoch": 0.58432,
+      "grad_norm": 0.27114234409088317,
+      "learning_rate": 2.01844390101836e-05,
+      "loss": 0.2425,
+      "step": 913
+    },
+    {
+      "epoch": 0.58496,
+      "grad_norm": 0.2557498830478621,
+      "learning_rate": 2.013248154882029e-05,
+      "loss": 0.2366,
+      "step": 914
+    },
+    {
+      "epoch": 0.5856,
+      "grad_norm": 0.2835512186078032,
+      "learning_rate": 2.0080545931090784e-05,
+      "loss": 0.2497,
+      "step": 915
+    },
+    {
+      "epoch": 0.58624,
+      "grad_norm": 0.2696371125362169,
+      "learning_rate": 2.0028632390063057e-05,
+      "loss": 0.2562,
+      "step": 916
+    },
+    {
+      "epoch": 0.58688,
+      "grad_norm": 0.24460712055660752,
+      "learning_rate": 1.9976741158706002e-05,
+      "loss": 0.2231,
+      "step": 917
+    },
+    {
+      "epoch": 0.58752,
+      "grad_norm": 0.2535820357018551,
+      "learning_rate": 1.992487246988839e-05,
+      "loss": 0.2416,
+      "step": 918
+    },
+    {
+      "epoch": 0.58816,
+      "grad_norm": 0.2698907239185565,
+      "learning_rate": 1.9873026556377835e-05,
+      "loss": 0.2545,
+      "step": 919
+    },
+    {
+      "epoch": 0.5888,
+      "grad_norm": 0.26029626231043246,
+      "learning_rate": 1.9821203650839728e-05,
+      "loss": 0.2498,
+      "step": 920
+    },
+    {
+      "epoch": 0.58944,
+      "grad_norm": 0.25629777791200037,
+      "learning_rate": 1.9769403985836242e-05,
+      "loss": 0.2192,
+      "step": 921
+    },
+    {
+      "epoch": 0.59008,
+      "grad_norm": 0.2183385801531958,
+      "learning_rate": 1.971762779382522e-05,
+      "loss": 0.2431,
+      "step": 922
+    },
+    {
+      "epoch": 0.59072,
+      "grad_norm": 0.23240675689873014,
+      "learning_rate": 1.9665875307159194e-05,
+      "loss": 0.2368,
+      "step": 923
+    },
+    {
+      "epoch": 0.59136,
+      "grad_norm": 0.27990439485520124,
+      "learning_rate": 1.961414675808429e-05,
+      "loss": 0.2159,
+      "step": 924
+    },
+    {
+      "epoch": 0.592,
+      "grad_norm": 0.24437667337907865,
+      "learning_rate": 1.956244237873924e-05,
+      "loss": 0.2326,
+      "step": 925
+    },
+    {
+      "epoch": 0.59264,
+      "grad_norm": 0.2672743139475555,
+      "learning_rate": 1.9510762401154274e-05,
+      "loss": 0.2261,
+      "step": 926
+    },
+    {
+      "epoch": 0.59328,
+      "grad_norm": 0.2596265830481029,
+      "learning_rate": 1.945910705725016e-05,
+      "loss": 0.2084,
+      "step": 927
+    },
+    {
+      "epoch": 0.59392,
+      "grad_norm": 0.23710690609748947,
+      "learning_rate": 1.940747657883708e-05,
+      "loss": 0.2088,
+      "step": 928
+    },
+    {
+      "epoch": 0.59456,
+      "grad_norm": 0.23434142099104566,
+      "learning_rate": 1.9355871197613647e-05,
+      "loss": 0.2263,
+      "step": 929
+    },
+    {
+      "epoch": 0.5952,
+      "grad_norm": 0.25262015553228806,
+      "learning_rate": 1.9304291145165858e-05,
+      "loss": 0.2408,
+      "step": 930
+    },
+    {
+      "epoch": 0.59584,
+      "grad_norm": 0.2390227728056176,
+      "learning_rate": 1.9252736652966015e-05,
+      "loss": 0.2095,
+      "step": 931
+    },
+    {
+      "epoch": 0.59648,
+      "grad_norm": 0.2619727057545716,
+      "learning_rate": 1.920120795237174e-05,
+      "loss": 0.244,
+      "step": 932
+    },
+    {
+      "epoch": 0.59712,
+      "grad_norm": 0.26307518776309047,
+      "learning_rate": 1.914970527462489e-05,
+      "loss": 0.2181,
+      "step": 933
+    },
+    {
+      "epoch": 0.59776,
+      "grad_norm": 0.23395354616480507,
+      "learning_rate": 1.909822885085058e-05,
+      "loss": 0.2384,
+      "step": 934
+    },
+    {
+      "epoch": 0.5984,
+      "grad_norm": 0.2493641757930232,
+      "learning_rate": 1.9046778912056047e-05,
+      "loss": 0.216,
+      "step": 935
+    },
+    {
+      "epoch": 0.59904,
+      "grad_norm": 0.25736998895213725,
+      "learning_rate": 1.8995355689129734e-05,
+      "loss": 0.2393,
+      "step": 936
+    },
+    {
+      "epoch": 0.59968,
+      "grad_norm": 0.2478616201093127,
+      "learning_rate": 1.894395941284015e-05,
+      "loss": 0.2396,
+      "step": 937
+    },
+    {
+      "epoch": 0.60032,
+      "grad_norm": 0.30086040520344987,
+      "learning_rate": 1.88925903138349e-05,
+      "loss": 0.2257,
+      "step": 938
+    },
+    {
+      "epoch": 0.60096,
+      "grad_norm": 0.2630367936880907,
+      "learning_rate": 1.884124862263961e-05,
+      "loss": 0.244,
+      "step": 939
+    },
+    {
+      "epoch": 0.6016,
+      "grad_norm": 0.2763146117742531,
+      "learning_rate": 1.8789934569656932e-05,
+      "loss": 0.2354,
+      "step": 940
+    },
+    {
+      "epoch": 0.60224,
+      "grad_norm": 0.2625253600089938,
+      "learning_rate": 1.8738648385165476e-05,
+      "loss": 0.2505,
+      "step": 941
+    },
+    {
+      "epoch": 0.60288,
+      "grad_norm": 0.24931986985590165,
+      "learning_rate": 1.868739029931878e-05,
+      "loss": 0.2268,
+      "step": 942
+    },
+    {
+      "epoch": 0.60352,
+      "grad_norm": 0.251516880356145,
+      "learning_rate": 1.8636160542144305e-05,
+      "loss": 0.2454,
+      "step": 943
+    },
+    {
+      "epoch": 0.60416,
+      "grad_norm": 0.2471177944809124,
+      "learning_rate": 1.858495934354235e-05,
+      "loss": 0.2511,
+      "step": 944
+    },
+    {
+      "epoch": 0.6048,
+      "grad_norm": 0.24831641722260756,
+      "learning_rate": 1.8533786933285106e-05,
+      "loss": 0.2217,
+      "step": 945
+    },
+    {
+      "epoch": 0.60544,
+      "grad_norm": 0.2721350067876185,
+      "learning_rate": 1.8482643541015514e-05,
+      "loss": 0.238,
+      "step": 946
+    },
+    {
+      "epoch": 0.60608,
+      "grad_norm": 0.26238128599837773,
+      "learning_rate": 1.843152939624635e-05,
+      "loss": 0.2309,
+      "step": 947
+    },
+    {
+      "epoch": 0.60672,
+      "grad_norm": 0.23546747581231484,
+      "learning_rate": 1.8380444728359086e-05,
+      "loss": 0.2271,
+      "step": 948
+    },
+    {
+      "epoch": 0.60736,
+      "grad_norm": 0.2670994139739104,
+      "learning_rate": 1.832938976660295e-05,
+      "loss": 0.2258,
+      "step": 949
+    },
+    {
+      "epoch": 0.608,
+      "grad_norm": 0.25354918913617136,
+      "learning_rate": 1.827836474009383e-05,
+      "loss": 0.2229,
+      "step": 950
+    },
+    {
+      "epoch": 0.60864,
+      "grad_norm": 0.22168176356775987,
+      "learning_rate": 1.8227369877813314e-05,
+      "loss": 0.2365,
+      "step": 951
+    },
+    {
+      "epoch": 0.60928,
+      "grad_norm": 0.24136704398108372,
+      "learning_rate": 1.8176405408607574e-05,
+      "loss": 0.2261,
+      "step": 952
+    },
+    {
+      "epoch": 0.60992,
+      "grad_norm": 0.25676237350947245,
+      "learning_rate": 1.812547156118644e-05,
+      "loss": 0.2395,
+      "step": 953
+    },
+    {
+      "epoch": 0.61056,
+      "grad_norm": 0.25126035375579264,
+      "learning_rate": 1.8074568564122293e-05,
+      "loss": 0.2285,
+      "step": 954
+    },
+    {
+      "epoch": 0.6112,
+      "grad_norm": 0.23146469335060782,
+      "learning_rate": 1.8023696645849064e-05,
+      "loss": 0.2064,
+      "step": 955
+    },
+    {
+      "epoch": 0.61184,
+      "grad_norm": 0.24116583091871857,
+      "learning_rate": 1.7972856034661244e-05,
+      "loss": 0.2372,
+      "step": 956
+    },
+    {
+      "epoch": 0.61248,
+      "grad_norm": 0.24677114493679053,
+      "learning_rate": 1.7922046958712786e-05,
+      "loss": 0.2294,
+      "step": 957
+    },
+    {
+      "epoch": 0.61312,
+      "grad_norm": 0.231763728539567,
+      "learning_rate": 1.7871269646016177e-05,
+      "loss": 0.2325,
+      "step": 958
+    },
+    {
+      "epoch": 0.61376,
+      "grad_norm": 0.2433237429417326,
+      "learning_rate": 1.782052432444131e-05,
+      "loss": 0.229,
+      "step": 959
+    },
+    {
+      "epoch": 0.6144,
+      "grad_norm": 0.2509056865213579,
+      "learning_rate": 1.7769811221714544e-05,
+      "loss": 0.2192,
+      "step": 960
+    },
+    {
+      "epoch": 0.61504,
+      "grad_norm": 0.23174059339598596,
+      "learning_rate": 1.7719130565417653e-05,
+      "loss": 0.2475,
+      "step": 961
+    },
+    {
+      "epoch": 0.61568,
+      "grad_norm": 0.24237243702263703,
+      "learning_rate": 1.7668482582986793e-05,
+      "loss": 0.2213,
+      "step": 962
+    },
+    {
+      "epoch": 0.61632,
+      "grad_norm": 0.22882041502884504,
+      "learning_rate": 1.7617867501711478e-05,
+      "loss": 0.212,
+      "step": 963
+    },
+    {
+      "epoch": 0.61696,
+      "grad_norm": 0.23006173457910242,
+      "learning_rate": 1.7567285548733614e-05,
+      "loss": 0.2361,
+      "step": 964
+    },
+    {
+      "epoch": 0.6176,
+      "grad_norm": 0.2194708408727919,
+      "learning_rate": 1.7516736951046394e-05,
+      "loss": 0.2117,
+      "step": 965
+    },
+    {
+      "epoch": 0.61824,
+      "grad_norm": 0.24727331914959877,
+      "learning_rate": 1.746622193549336e-05,
+      "loss": 0.2337,
+      "step": 966
+    },
+    {
+      "epoch": 0.61888,
+      "grad_norm": 0.23884495075665296,
+      "learning_rate": 1.7415740728767337e-05,
+      "loss": 0.218,
+      "step": 967
+    },
+    {
+      "epoch": 0.61952,
+      "grad_norm": 0.24787777171744685,
+      "learning_rate": 1.7365293557409413e-05,
+      "loss": 0.2453,
+      "step": 968
+    },
+    {
+      "epoch": 0.62016,
+      "grad_norm": 0.22521812004926223,
+      "learning_rate": 1.7314880647807967e-05,
+      "loss": 0.2267,
+      "step": 969
+    },
+    {
+      "epoch": 0.6208,
+      "grad_norm": 0.2568864190939131,
+      "learning_rate": 1.7264502226197593e-05,
+      "loss": 0.2126,
+      "step": 970
+    },
+    {
+      "epoch": 0.62144,
+      "grad_norm": 0.24453133217677783,
+      "learning_rate": 1.7214158518658152e-05,
+      "loss": 0.2159,
+      "step": 971
+    },
+    {
+      "epoch": 0.62208,
+      "grad_norm": 0.2620763915586535,
+      "learning_rate": 1.716384975111368e-05,
+      "loss": 0.2245,
+      "step": 972
+    },
+    {
+      "epoch": 0.62272,
+      "grad_norm": 0.2642374662977824,
+      "learning_rate": 1.711357614933145e-05,
+      "loss": 0.2008,
+      "step": 973
+    },
+    {
+      "epoch": 0.62336,
+      "grad_norm": 0.2501059818378354,
+      "learning_rate": 1.7063337938920895e-05,
+      "loss": 0.2262,
+      "step": 974
+    },
+    {
+      "epoch": 0.624,
+      "grad_norm": 0.2436947364589447,
+      "learning_rate": 1.7013135345332653e-05,
+      "loss": 0.2238,
+      "step": 975
+    },
+    {
+      "epoch": 0.62464,
+      "grad_norm": 0.23586750109628674,
+      "learning_rate": 1.6962968593857502e-05,
+      "loss": 0.2328,
+      "step": 976
+    },
+    {
+      "epoch": 0.62528,
+      "grad_norm": 0.2531876348648337,
+      "learning_rate": 1.6912837909625395e-05,
+      "loss": 0.2516,
+      "step": 977
+    },
+    {
+      "epoch": 0.62592,
+      "grad_norm": 0.24116459963353373,
+      "learning_rate": 1.6862743517604407e-05,
+      "loss": 0.2453,
+      "step": 978
+    },
+    {
+      "epoch": 0.62656,
+      "grad_norm": 0.26038744671626085,
+      "learning_rate": 1.6812685642599758e-05,
+      "loss": 0.2346,
+      "step": 979
+    },
+    {
+      "epoch": 0.6272,
+      "grad_norm": 0.25874895626816735,
+      "learning_rate": 1.6762664509252806e-05,
+      "loss": 0.2438,
+      "step": 980
+    },
+    {
+      "epoch": 0.62784,
+      "grad_norm": 0.25081898233180916,
+      "learning_rate": 1.6712680342039994e-05,
+      "loss": 0.2113,
+      "step": 981
+    },
+    {
+      "epoch": 0.62848,
+      "grad_norm": 0.2549406470192005,
+      "learning_rate": 1.6662733365271916e-05,
+      "loss": 0.2322,
+      "step": 982
+    },
+    {
+      "epoch": 0.62912,
+      "grad_norm": 0.24131591470533428,
+      "learning_rate": 1.661282380309223e-05,
+      "loss": 0.2398,
+      "step": 983
+    },
+    {
+      "epoch": 0.62976,
+      "grad_norm": 0.24129834158887983,
+      "learning_rate": 1.6562951879476708e-05,
+      "loss": 0.2489,
+      "step": 984
+    },
+    {
+      "epoch": 0.6304,
+      "grad_norm": 0.24301150056355691,
+      "learning_rate": 1.6513117818232217e-05,
+      "loss": 0.2209,
+      "step": 985
+    },
+    {
+      "epoch": 0.63104,
+      "grad_norm": 0.2694532610607709,
+      "learning_rate": 1.6463321842995717e-05,
+      "loss": 0.2047,
+      "step": 986
+    },
+    {
+      "epoch": 0.63168,
+      "grad_norm": 0.23806358029121397,
+      "learning_rate": 1.641356417723322e-05,
+      "loss": 0.2189,
+      "step": 987
+    },
+    {
+      "epoch": 0.63232,
+      "grad_norm": 0.25277157714470466,
+      "learning_rate": 1.6363845044238868e-05,
+      "loss": 0.2283,
+      "step": 988
+    },
+    {
+      "epoch": 0.63296,
+      "grad_norm": 0.2514369322275561,
+      "learning_rate": 1.631416466713383e-05,
+      "loss": 0.2335,
+      "step": 989
+    },
+    {
+      "epoch": 0.6336,
+      "grad_norm": 0.25806045074103723,
+      "learning_rate": 1.62645232688654e-05,
+      "loss": 0.2114,
+      "step": 990
+    },
+    {
+      "epoch": 0.63424,
+      "grad_norm": 0.2430444345626924,
+      "learning_rate": 1.621492107220593e-05,
+      "loss": 0.2246,
+      "step": 991
+    },
+    {
+      "epoch": 0.63488,
+      "grad_norm": 0.2505219244759305,
+      "learning_rate": 1.6165358299751825e-05,
+      "loss": 0.2406,
+      "step": 992
+    },
+    {
+      "epoch": 0.63552,
+      "grad_norm": 0.2519470143187115,
+      "learning_rate": 1.6115835173922618e-05,
+      "loss": 0.2273,
+      "step": 993
+    },
+    {
+      "epoch": 0.63616,
+      "grad_norm": 0.24750380002574845,
+      "learning_rate": 1.6066351916959876e-05,
+      "loss": 0.2378,
+      "step": 994
+    },
+    {
+      "epoch": 0.6368,
+      "grad_norm": 0.2908287661105402,
+      "learning_rate": 1.6016908750926286e-05,
+      "loss": 0.2264,
+      "step": 995
+    },
+    {
+      "epoch": 0.63744,
+      "grad_norm": 0.245527026321788,
+      "learning_rate": 1.596750589770459e-05,
+      "loss": 0.2498,
+      "step": 996
+    },
+    {
+      "epoch": 0.63808,
+      "grad_norm": 0.22986996404259472,
+      "learning_rate": 1.5918143578996646e-05,
+      "loss": 0.228,
+      "step": 997
+    },
+    {
+      "epoch": 0.63872,
+      "grad_norm": 0.2381296455217305,
+      "learning_rate": 1.5868822016322386e-05,
+      "loss": 0.2284,
+      "step": 998
+    },
+    {
+      "epoch": 0.63936,
+      "grad_norm": 0.26770228468675383,
+      "learning_rate": 1.5819541431018874e-05,
+      "loss": 0.2462,
+      "step": 999
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.2718312488412604,
+      "learning_rate": 1.577030204423924e-05,
+      "loss": 0.231,
+      "step": 1000
+    },
+    {
+      "epoch": 0.64064,
+      "grad_norm": 0.2679264270683539,
+      "learning_rate": 1.5721104076951783e-05,
+      "loss": 0.2186,
+      "step": 1001
+    },
+    {
+      "epoch": 0.64128,
+      "grad_norm": 0.2782932295742096,
+      "learning_rate": 1.5671947749938875e-05,
+      "loss": 0.24,
+      "step": 1002
+    },
+    {
+      "epoch": 0.64192,
+      "grad_norm": 0.25226031382789144,
+      "learning_rate": 1.562283328379605e-05,
+      "loss": 0.2337,
+      "step": 1003
+    },
+    {
+      "epoch": 0.64256,
+      "grad_norm": 0.2701550570632097,
+      "learning_rate": 1.5573760898931e-05,
+      "loss": 0.2165,
+      "step": 1004
+    },
+    {
+      "epoch": 0.6432,
+      "grad_norm": 0.2523314356080161,
+      "learning_rate": 1.5524730815562518e-05,
+      "loss": 0.2303,
+      "step": 1005
+    },
+    {
+      "epoch": 0.64384,
+      "grad_norm": 0.2635832928427503,
+      "learning_rate": 1.5475743253719628e-05,
+      "loss": 0.2165,
+      "step": 1006
+    },
+    {
+      "epoch": 0.64448,
+      "grad_norm": 0.2663059645383403,
+      "learning_rate": 1.542679843324049e-05,
+      "loss": 0.2331,
+      "step": 1007
+    },
+    {
+      "epoch": 0.64512,
+      "grad_norm": 0.23852804442441813,
+      "learning_rate": 1.5377896573771465e-05,
+      "loss": 0.2393,
+      "step": 1008
+    },
+    {
+      "epoch": 0.64576,
+      "grad_norm": 0.24933816278004978,
+      "learning_rate": 1.5329037894766133e-05,
+      "loss": 0.2492,
+      "step": 1009
+    },
+    {
+      "epoch": 0.6464,
+      "grad_norm": 0.2803377336569009,
+      "learning_rate": 1.528022261548429e-05,
+      "loss": 0.2344,
+      "step": 1010
+    },
+    {
+      "epoch": 0.64704,
+      "grad_norm": 0.24403973650102356,
+      "learning_rate": 1.5231450954990955e-05,
+      "loss": 0.2414,
+      "step": 1011
+    },
+    {
+      "epoch": 0.64768,
+      "grad_norm": 0.23988369345335178,
+      "learning_rate": 1.518272313215543e-05,
+      "loss": 0.2313,
+      "step": 1012
+    },
+    {
+      "epoch": 0.64832,
+      "grad_norm": 0.2656555508424385,
+      "learning_rate": 1.513403936565026e-05,
+      "loss": 0.2219,
+      "step": 1013
+    },
+    {
+      "epoch": 0.64896,
+      "grad_norm": 0.2161101202834104,
+      "learning_rate": 1.5085399873950313e-05,
+      "loss": 0.2124,
+      "step": 1014
+    },
+    {
+      "epoch": 0.6496,
+      "grad_norm": 0.25972265909330855,
+      "learning_rate": 1.5036804875331734e-05,
+      "loss": 0.2663,
+      "step": 1015
+    },
+    {
+      "epoch": 0.65024,
+      "grad_norm": 0.2178985756765881,
+      "learning_rate": 1.498825458787102e-05,
+      "loss": 0.2353,
+      "step": 1016
+    },
+    {
+      "epoch": 0.65088,
+      "grad_norm": 0.24646566613323023,
+      "learning_rate": 1.4939749229444028e-05,
+      "loss": 0.2268,
+      "step": 1017
+    },
+    {
+      "epoch": 0.65152,
+      "grad_norm": 0.2583375090956272,
+      "learning_rate": 1.4891289017724967e-05,
+      "loss": 0.2257,
+      "step": 1018
+    },
+    {
+      "epoch": 0.65216,
+      "grad_norm": 0.21985994545469698,
+      "learning_rate": 1.4842874170185467e-05,
+      "loss": 0.2354,
+      "step": 1019
+    },
+    {
+      "epoch": 0.6528,
+      "grad_norm": 0.2311075776980603,
+      "learning_rate": 1.4794504904093559e-05,
+      "loss": 0.2208,
+      "step": 1020
+    },
+    {
+      "epoch": 0.65344,
+      "grad_norm": 0.24121863480190284,
+      "learning_rate": 1.4746181436512747e-05,
+      "loss": 0.2489,
+      "step": 1021
+    },
+    {
+      "epoch": 0.65408,
+      "grad_norm": 0.2568450202104899,
+      "learning_rate": 1.4697903984300964e-05,
+      "loss": 0.2077,
+      "step": 1022
+    },
+    {
+      "epoch": 0.65472,
+      "grad_norm": 0.2528324335224712,
+      "learning_rate": 1.4649672764109704e-05,
+      "loss": 0.234,
+      "step": 1023
+    },
+    {
+      "epoch": 0.65536,
+      "grad_norm": 0.27109413397643667,
+      "learning_rate": 1.460148799238293e-05,
+      "loss": 0.2269,
+      "step": 1024
+    },
+    {
+      "epoch": 0.656,
+      "grad_norm": 0.2455738528432716,
+      "learning_rate": 1.4553349885356213e-05,
+      "loss": 0.2346,
+      "step": 1025
+    },
+    {
+      "epoch": 0.65664,
+      "grad_norm": 0.23151523833824506,
+      "learning_rate": 1.4505258659055665e-05,
+      "loss": 0.2119,
+      "step": 1026
+    },
+    {
+      "epoch": 0.65728,
+      "grad_norm": 0.26474620316813297,
+      "learning_rate": 1.4457214529297043e-05,
+      "loss": 0.2516,
+      "step": 1027
+    },
+    {
+      "epoch": 0.65792,
+      "grad_norm": 0.24672776250362344,
+      "learning_rate": 1.4409217711684737e-05,
+      "loss": 0.2259,
+      "step": 1028
+    },
+    {
+      "epoch": 0.65856,
+      "grad_norm": 0.2416711950266535,
+      "learning_rate": 1.4361268421610829e-05,
+      "loss": 0.2078,
+      "step": 1029
+    },
+    {
+      "epoch": 0.6592,
+      "grad_norm": 0.24307662108665554,
+      "learning_rate": 1.431336687425411e-05,
+      "loss": 0.2413,
+      "step": 1030
+    },
+    {
+      "epoch": 0.65984,
+      "grad_norm": 0.22268074931915147,
+      "learning_rate": 1.4265513284579118e-05,
+      "loss": 0.2192,
+      "step": 1031
+    },
+    {
+      "epoch": 0.66048,
+      "grad_norm": 0.23053431215637107,
+      "learning_rate": 1.4217707867335179e-05,
+      "loss": 0.2136,
+      "step": 1032
+    },
+    {
+      "epoch": 0.66112,
+      "grad_norm": 0.22229963082407714,
+      "learning_rate": 1.4169950837055424e-05,
+      "loss": 0.2328,
+      "step": 1033
+    },
+    {
+      "epoch": 0.66176,
+      "grad_norm": 0.2530252244156255,
+      "learning_rate": 1.4122242408055864e-05,
+      "loss": 0.229,
+      "step": 1034
+    },
+    {
+      "epoch": 0.6624,
+      "grad_norm": 0.2588958847924931,
+      "learning_rate": 1.4074582794434388e-05,
+      "loss": 0.2394,
+      "step": 1035
+    },
+    {
+      "epoch": 0.66304,
+      "grad_norm": 0.24876729811089132,
+      "learning_rate": 1.4026972210069839e-05,
+      "loss": 0.2219,
+      "step": 1036
+    },
+    {
+      "epoch": 0.66368,
+      "grad_norm": 0.22093002070585888,
+      "learning_rate": 1.3979410868620987e-05,
+      "loss": 0.2327,
+      "step": 1037
+    },
+    {
+      "epoch": 0.66432,
+      "grad_norm": 0.24123324923982054,
+      "learning_rate": 1.3931898983525694e-05,
+      "loss": 0.2445,
+      "step": 1038
+    },
+    {
+      "epoch": 0.66496,
+      "grad_norm": 0.23897146058805963,
+      "learning_rate": 1.3884436767999798e-05,
+      "loss": 0.2432,
+      "step": 1039
+    },
+    {
+      "epoch": 0.6656,
+      "grad_norm": 0.22257963570399686,
+      "learning_rate": 1.3837024435036309e-05,
+      "loss": 0.2098,
+      "step": 1040
+    },
+    {
+      "epoch": 0.66624,
+      "grad_norm": 0.2431094631896802,
+      "learning_rate": 1.3789662197404346e-05,
+      "loss": 0.2138,
+      "step": 1041
+    },
+    {
+      "epoch": 0.66688,
+      "grad_norm": 0.26708581426404315,
+      "learning_rate": 1.3742350267648202e-05,
+      "loss": 0.2282,
+      "step": 1042
+    },
+    {
+      "epoch": 0.66752,
+      "grad_norm": 0.22594464120578142,
+      "learning_rate": 1.3695088858086464e-05,
+      "loss": 0.2351,
+      "step": 1043
+    },
+    {
+      "epoch": 0.66816,
+      "grad_norm": 0.2271257487159897,
+      "learning_rate": 1.3647878180810924e-05,
+      "loss": 0.2157,
+      "step": 1044
+    }
+  ],
+  "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": 478374861078528.0,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}