{
  "best_metric": 0.2655787765979767,
  "best_model_checkpoint": "/scratch/czm5kz/llama2-7b_8_100_0.0003_sg_finetuned_combined/checkpoint-1680",
  "epoch": 100.0,
  "eval_steps": 20,
  "global_step": 1700,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.29,
      "grad_norm": 2.5130302906036377,
      "learning_rate": 0.0002994705882352941,
      "loss": 5.7234,
      "step": 5
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.1954381465911865,
      "learning_rate": 0.0002985882352941176,
      "loss": 4.7563,
      "step": 10
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.519700050354004,
      "learning_rate": 0.0002978823529411764,
      "loss": 4.2519,
      "step": 15
    },
    {
      "epoch": 1.18,
      "grad_norm": 2.1008927822113037,
      "learning_rate": 0.00029699999999999996,
      "loss": 3.8694,
      "step": 20
    },
    {
      "epoch": 1.18,
      "eval_loss": 3.4059982299804688,
      "eval_runtime": 1.9717,
      "eval_samples_per_second": 67.962,
      "eval_steps_per_second": 8.622,
      "step": 20
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.9939683675765991,
      "learning_rate": 0.0002961176470588235,
      "loss": 3.0042,
      "step": 25
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.9956856966018677,
      "learning_rate": 0.00029523529411764704,
      "loss": 2.7303,
      "step": 30
    },
    {
      "epoch": 2.06,
      "grad_norm": 2.552236318588257,
      "learning_rate": 0.0002943529411764706,
      "loss": 2.7002,
      "step": 35
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.5456089973449707,
      "learning_rate": 0.00029347058823529413,
      "loss": 2.3598,
      "step": 40
    },
    {
      "epoch": 2.35,
      "eval_loss": 2.0730819702148438,
      "eval_runtime": 1.9745,
      "eval_samples_per_second": 67.864,
      "eval_steps_per_second": 8.61,
      "step": 40
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.949130058288574,
      "learning_rate": 0.0002925882352941176,
      "loss": 1.9988,
      "step": 45
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.44789457321167,
      "learning_rate": 0.00029170588235294116,
      "loss": 1.9843,
      "step": 50
    },
    {
      "epoch": 3.24,
      "grad_norm": 5.215721130371094,
      "learning_rate": 0.00029082352941176465,
      "loss": 1.5394,
      "step": 55
    },
    {
      "epoch": 3.53,
      "grad_norm": 3.831580638885498,
      "learning_rate": 0.0002899411764705882,
      "loss": 1.5503,
      "step": 60
    },
    {
      "epoch": 3.53,
      "eval_loss": 1.2807282209396362,
      "eval_runtime": 1.9725,
      "eval_samples_per_second": 67.933,
      "eval_steps_per_second": 8.618,
      "step": 60
    },
    {
      "epoch": 3.82,
      "grad_norm": 3.5273172855377197,
      "learning_rate": 0.00028905882352941173,
      "loss": 1.4236,
      "step": 65
    },
    {
      "epoch": 4.12,
      "grad_norm": 4.674720287322998,
      "learning_rate": 0.0002881764705882353,
      "loss": 1.1729,
      "step": 70
    },
    {
      "epoch": 4.41,
      "grad_norm": 5.43109655380249,
      "learning_rate": 0.0002872941176470588,
      "loss": 0.9213,
      "step": 75
    },
    {
      "epoch": 4.71,
      "grad_norm": 6.3855814933776855,
      "learning_rate": 0.0002864117647058823,
      "loss": 0.9106,
      "step": 80
    },
    {
      "epoch": 4.71,
      "eval_loss": 0.7538219094276428,
      "eval_runtime": 1.9756,
      "eval_samples_per_second": 67.827,
      "eval_steps_per_second": 8.605,
      "step": 80
    },
    {
      "epoch": 5.0,
      "grad_norm": 5.1275529861450195,
      "learning_rate": 0.00028552941176470585,
      "loss": 1.0219,
      "step": 85
    },
    {
      "epoch": 5.29,
      "grad_norm": 3.2476627826690674,
      "learning_rate": 0.0002846470588235294,
      "loss": 0.6821,
      "step": 90
    },
    {
      "epoch": 5.59,
      "grad_norm": 3.561427116394043,
      "learning_rate": 0.00028376470588235294,
      "loss": 0.6022,
      "step": 95
    },
    {
      "epoch": 5.88,
      "grad_norm": 4.358293056488037,
      "learning_rate": 0.0002828823529411765,
      "loss": 0.6565,
      "step": 100
    },
    {
      "epoch": 5.88,
      "eval_loss": 0.5009134411811829,
      "eval_runtime": 1.9816,
      "eval_samples_per_second": 67.622,
      "eval_steps_per_second": 8.579,
      "step": 100
    },
    {
      "epoch": 6.18,
      "grad_norm": 3.2790002822875977,
      "learning_rate": 0.00028199999999999997,
      "loss": 0.6397,
      "step": 105
    },
    {
      "epoch": 6.47,
      "grad_norm": 2.5752627849578857,
      "learning_rate": 0.0002811176470588235,
      "loss": 0.4518,
      "step": 110
    },
    {
      "epoch": 6.76,
      "grad_norm": 3.419771909713745,
      "learning_rate": 0.000280235294117647,
      "loss": 0.4741,
      "step": 115
    },
    {
      "epoch": 7.06,
      "grad_norm": 2.2088027000427246,
      "learning_rate": 0.00027935294117647054,
      "loss": 0.5214,
      "step": 120
    },
    {
      "epoch": 7.06,
      "eval_loss": 0.40768566727638245,
      "eval_runtime": 1.9831,
      "eval_samples_per_second": 67.57,
      "eval_steps_per_second": 8.572,
      "step": 120
    },
    {
      "epoch": 7.35,
      "grad_norm": 2.9267308712005615,
      "learning_rate": 0.0002784705882352941,
      "loss": 0.4434,
      "step": 125
    },
    {
      "epoch": 7.65,
      "grad_norm": 3.3790996074676514,
      "learning_rate": 0.00027758823529411763,
      "loss": 0.4631,
      "step": 130
    },
    {
      "epoch": 7.94,
      "grad_norm": 2.3158915042877197,
      "learning_rate": 0.0002767058823529412,
      "loss": 0.4464,
      "step": 135
    },
    {
      "epoch": 8.24,
      "grad_norm": 2.231661081314087,
      "learning_rate": 0.00027582352941176466,
      "loss": 0.4275,
      "step": 140
    },
    {
      "epoch": 8.24,
      "eval_loss": 0.3691316843032837,
      "eval_runtime": 1.9948,
      "eval_samples_per_second": 67.174,
      "eval_steps_per_second": 8.522,
      "step": 140
    },
    {
      "epoch": 8.53,
      "grad_norm": 2.341855525970459,
      "learning_rate": 0.0002749411764705882,
      "loss": 0.3938,
      "step": 145
    },
    {
      "epoch": 8.82,
      "grad_norm": 1.6353213787078857,
      "learning_rate": 0.00027405882352941175,
      "loss": 0.4794,
      "step": 150
    },
    {
      "epoch": 9.12,
      "grad_norm": 1.7900513410568237,
      "learning_rate": 0.0002731764705882353,
      "loss": 0.4073,
      "step": 155
    },
    {
      "epoch": 9.41,
      "grad_norm": 2.082562208175659,
      "learning_rate": 0.0002722941176470588,
      "loss": 0.3793,
      "step": 160
    },
    {
      "epoch": 9.41,
      "eval_loss": 0.3403892517089844,
      "eval_runtime": 1.988,
      "eval_samples_per_second": 67.405,
      "eval_steps_per_second": 8.551,
      "step": 160
    },
    {
      "epoch": 9.71,
      "grad_norm": 1.9803985357284546,
      "learning_rate": 0.0002714117647058823,
      "loss": 0.3943,
      "step": 165
    },
    {
      "epoch": 10.0,
      "grad_norm": 1.6986087560653687,
      "learning_rate": 0.00027052941176470587,
      "loss": 0.4254,
      "step": 170
    },
    {
      "epoch": 10.29,
      "grad_norm": 1.120859146118164,
      "learning_rate": 0.0002696470588235294,
      "loss": 0.3333,
      "step": 175
    },
    {
      "epoch": 10.59,
      "grad_norm": 2.3081905841827393,
      "learning_rate": 0.0002687647058823529,
      "loss": 0.3671,
      "step": 180
    },
    {
      "epoch": 10.59,
      "eval_loss": 0.32923150062561035,
      "eval_runtime": 1.9871,
      "eval_samples_per_second": 67.435,
      "eval_steps_per_second": 8.555,
      "step": 180
    },
    {
      "epoch": 10.88,
      "grad_norm": 3.328174352645874,
      "learning_rate": 0.00026788235294117644,
      "loss": 0.3888,
      "step": 185
    },
    {
      "epoch": 11.18,
      "grad_norm": 2.408024549484253,
      "learning_rate": 0.000267,
      "loss": 0.3964,
      "step": 190
    },
    {
      "epoch": 11.47,
      "grad_norm": 1.6264761686325073,
      "learning_rate": 0.00026611764705882353,
      "loss": 0.3579,
      "step": 195
    },
    {
      "epoch": 11.76,
      "grad_norm": 1.857405662536621,
      "learning_rate": 0.00026523529411764707,
      "loss": 0.375,
      "step": 200
    },
    {
      "epoch": 11.76,
      "eval_loss": 0.3141331374645233,
      "eval_runtime": 1.9899,
      "eval_samples_per_second": 67.341,
      "eval_steps_per_second": 8.543,
      "step": 200
    },
    {
      "epoch": 12.06,
      "grad_norm": 1.0782090425491333,
      "learning_rate": 0.00026435294117647056,
      "loss": 0.3808,
      "step": 205
    },
    {
      "epoch": 12.35,
      "grad_norm": 1.4318054914474487,
      "learning_rate": 0.0002634705882352941,
      "loss": 0.3148,
      "step": 210
    },
    {
      "epoch": 12.65,
      "grad_norm": 1.6774518489837646,
      "learning_rate": 0.0002625882352941176,
      "loss": 0.3729,
      "step": 215
    },
    {
      "epoch": 12.94,
      "grad_norm": 1.2252236604690552,
      "learning_rate": 0.00026170588235294113,
      "loss": 0.379,
      "step": 220
    },
    {
      "epoch": 12.94,
      "eval_loss": 0.3084084689617157,
      "eval_runtime": 1.9898,
      "eval_samples_per_second": 67.342,
      "eval_steps_per_second": 8.543,
      "step": 220
    },
    {
      "epoch": 13.24,
      "grad_norm": 0.9840493202209473,
      "learning_rate": 0.0002608235294117647,
      "loss": 0.3292,
      "step": 225
    },
    {
      "epoch": 13.53,
      "grad_norm": 2.3673970699310303,
      "learning_rate": 0.0002599411764705882,
      "loss": 0.3803,
      "step": 230
    },
    {
      "epoch": 13.82,
      "grad_norm": 1.2279037237167358,
      "learning_rate": 0.00025905882352941176,
      "loss": 0.3444,
      "step": 235
    },
    {
      "epoch": 14.12,
      "grad_norm": 0.928492546081543,
      "learning_rate": 0.00025817647058823525,
      "loss": 0.334,
      "step": 240
    },
    {
      "epoch": 14.12,
      "eval_loss": 0.30614903569221497,
      "eval_runtime": 1.9906,
      "eval_samples_per_second": 67.317,
      "eval_steps_per_second": 8.54,
      "step": 240
    },
    {
      "epoch": 14.41,
      "grad_norm": 1.6385815143585205,
      "learning_rate": 0.0002572941176470588,
      "loss": 0.309,
      "step": 245
    },
    {
      "epoch": 14.71,
      "grad_norm": 0.9644414782524109,
      "learning_rate": 0.00025641176470588234,
      "loss": 0.3318,
      "step": 250
    },
    {
      "epoch": 15.0,
      "grad_norm": 1.3966537714004517,
      "learning_rate": 0.0002555294117647059,
      "loss": 0.3974,
      "step": 255
    },
    {
      "epoch": 15.29,
      "grad_norm": 1.0618793964385986,
      "learning_rate": 0.0002546470588235294,
      "loss": 0.3027,
      "step": 260
    },
    {
      "epoch": 15.29,
      "eval_loss": 0.29958730936050415,
      "eval_runtime": 1.9894,
      "eval_samples_per_second": 67.357,
      "eval_steps_per_second": 8.545,
      "step": 260
    },
    {
      "epoch": 15.59,
      "grad_norm": 1.31105637550354,
      "learning_rate": 0.0002537647058823529,
      "loss": 0.3196,
      "step": 265
    },
    {
      "epoch": 15.88,
      "grad_norm": 1.4833605289459229,
      "learning_rate": 0.00025288235294117646,
      "loss": 0.3485,
      "step": 270
    },
    {
      "epoch": 16.18,
      "grad_norm": 0.8568434119224548,
      "learning_rate": 0.00025199999999999995,
      "loss": 0.3391,
      "step": 275
    },
    {
      "epoch": 16.47,
      "grad_norm": 1.2687785625457764,
      "learning_rate": 0.0002511176470588235,
      "loss": 0.3179,
      "step": 280
    },
    {
      "epoch": 16.47,
      "eval_loss": 0.29810982942581177,
      "eval_runtime": 1.9921,
      "eval_samples_per_second": 67.267,
      "eval_steps_per_second": 8.534,
      "step": 280
    },
    {
      "epoch": 16.76,
      "grad_norm": 1.125921368598938,
      "learning_rate": 0.00025023529411764703,
      "loss": 0.3231,
      "step": 285
    },
    {
      "epoch": 17.06,
      "grad_norm": 0.751727819442749,
      "learning_rate": 0.0002493529411764706,
      "loss": 0.349,
      "step": 290
    },
    {
      "epoch": 17.35,
      "grad_norm": 1.0114847421646118,
      "learning_rate": 0.0002484705882352941,
      "loss": 0.3027,
      "step": 295
    },
    {
      "epoch": 17.65,
      "grad_norm": 2.5540404319763184,
      "learning_rate": 0.0002475882352941176,
      "loss": 0.3021,
      "step": 300
    },
    {
      "epoch": 17.65,
      "eval_loss": 0.29670611023902893,
      "eval_runtime": 2.0093,
      "eval_samples_per_second": 66.689,
      "eval_steps_per_second": 8.461,
      "step": 300
    },
    {
      "epoch": 17.94,
      "grad_norm": 0.9592204689979553,
      "learning_rate": 0.00024670588235294115,
      "loss": 0.3553,
      "step": 305
    },
    {
      "epoch": 18.24,
      "grad_norm": 0.6442326307296753,
      "learning_rate": 0.0002458235294117647,
      "loss": 0.3101,
      "step": 310
    },
    {
      "epoch": 18.53,
      "grad_norm": 1.0425697565078735,
      "learning_rate": 0.00024494117647058824,
      "loss": 0.3155,
      "step": 315
    },
    {
      "epoch": 18.82,
      "grad_norm": 3.3176205158233643,
      "learning_rate": 0.00024405882352941172,
      "loss": 0.3634,
      "step": 320
    },
    {
      "epoch": 18.82,
      "eval_loss": 0.2934125065803528,
      "eval_runtime": 1.9921,
      "eval_samples_per_second": 67.267,
      "eval_steps_per_second": 8.534,
      "step": 320
    },
    {
      "epoch": 19.12,
      "grad_norm": 0.5256841778755188,
      "learning_rate": 0.00024317647058823527,
      "loss": 0.3197,
      "step": 325
    },
    {
      "epoch": 19.41,
      "grad_norm": 0.7443175315856934,
      "learning_rate": 0.00024229411764705878,
      "loss": 0.3117,
      "step": 330
    },
    {
      "epoch": 19.71,
      "grad_norm": 0.8534731864929199,
      "learning_rate": 0.00024141176470588233,
      "loss": 0.322,
      "step": 335
    },
    {
      "epoch": 20.0,
      "grad_norm": 1.1688783168792725,
      "learning_rate": 0.00024052941176470587,
      "loss": 0.3293,
      "step": 340
    },
    {
      "epoch": 20.0,
      "eval_loss": 0.2844869792461395,
      "eval_runtime": 1.9933,
      "eval_samples_per_second": 67.224,
      "eval_steps_per_second": 8.528,
      "step": 340
    },
    {
      "epoch": 20.29,
      "grad_norm": 0.8509626388549805,
      "learning_rate": 0.00023964705882352939,
      "loss": 0.2899,
      "step": 345
    },
    {
      "epoch": 20.59,
      "grad_norm": 1.0683010816574097,
      "learning_rate": 0.00023876470588235293,
      "loss": 0.3368,
      "step": 350
    },
    {
      "epoch": 20.88,
      "grad_norm": 0.8359882235527039,
      "learning_rate": 0.00023788235294117644,
      "loss": 0.3189,
      "step": 355
    },
    {
      "epoch": 21.18,
      "grad_norm": 0.746294379234314,
      "learning_rate": 0.000237,
      "loss": 0.2907,
      "step": 360
    },
    {
      "epoch": 21.18,
      "eval_loss": 0.28352469205856323,
      "eval_runtime": 1.9943,
      "eval_samples_per_second": 67.191,
      "eval_steps_per_second": 8.524,
      "step": 360
    },
    {
      "epoch": 21.47,
      "grad_norm": 0.6525380611419678,
      "learning_rate": 0.00023611764705882353,
      "loss": 0.3088,
      "step": 365
    },
    {
      "epoch": 21.76,
      "grad_norm": 0.7996423840522766,
      "learning_rate": 0.00023523529411764702,
      "loss": 0.3241,
      "step": 370
    },
    {
      "epoch": 22.06,
      "grad_norm": 0.6835091710090637,
      "learning_rate": 0.00023435294117647056,
      "loss": 0.3078,
      "step": 375
    },
    {
      "epoch": 22.35,
      "grad_norm": 0.7048420310020447,
      "learning_rate": 0.00023347058823529408,
      "loss": 0.3098,
      "step": 380
    },
    {
      "epoch": 22.35,
      "eval_loss": 0.28362563252449036,
      "eval_runtime": 1.9954,
      "eval_samples_per_second": 67.156,
      "eval_steps_per_second": 8.52,
      "step": 380
    },
    {
      "epoch": 22.65,
      "grad_norm": 0.7149125337600708,
      "learning_rate": 0.00023258823529411762,
      "loss": 0.2859,
      "step": 385
    },
    {
      "epoch": 22.94,
      "grad_norm": 0.8284432291984558,
      "learning_rate": 0.00023170588235294116,
      "loss": 0.3193,
      "step": 390
    },
    {
      "epoch": 23.24,
      "grad_norm": 0.7711451053619385,
      "learning_rate": 0.00023082352941176468,
      "loss": 0.2894,
      "step": 395
    },
    {
      "epoch": 23.53,
      "grad_norm": 0.7395182847976685,
      "learning_rate": 0.00022994117647058822,
      "loss": 0.2911,
      "step": 400
    },
    {
      "epoch": 23.53,
      "eval_loss": 0.28144964575767517,
      "eval_runtime": 1.9967,
      "eval_samples_per_second": 67.11,
      "eval_steps_per_second": 8.514,
      "step": 400
    },
    {
      "epoch": 23.82,
      "grad_norm": 0.694525957107544,
      "learning_rate": 0.00022905882352941174,
      "loss": 0.3048,
      "step": 405
    },
    {
      "epoch": 24.12,
      "grad_norm": 0.6153783798217773,
      "learning_rate": 0.00022817647058823528,
      "loss": 0.2975,
      "step": 410
    },
    {
      "epoch": 24.41,
      "grad_norm": 0.7303423285484314,
      "learning_rate": 0.00022729411764705883,
      "loss": 0.2838,
      "step": 415
    },
    {
      "epoch": 24.71,
      "grad_norm": 0.6618799567222595,
      "learning_rate": 0.00022641176470588234,
      "loss": 0.326,
      "step": 420
    },
    {
      "epoch": 24.71,
      "eval_loss": 0.28115326166152954,
      "eval_runtime": 1.9929,
      "eval_samples_per_second": 67.238,
      "eval_steps_per_second": 8.53,
      "step": 420
    },
    {
      "epoch": 25.0,
      "grad_norm": 0.8111241459846497,
      "learning_rate": 0.00022552941176470586,
      "loss": 0.3084,
      "step": 425
    },
    {
      "epoch": 25.29,
      "grad_norm": 0.6205019354820251,
      "learning_rate": 0.00022464705882352937,
      "loss": 0.2873,
      "step": 430
    },
    {
      "epoch": 25.59,
      "grad_norm": 0.8169652223587036,
      "learning_rate": 0.00022376470588235292,
      "loss": 0.2965,
      "step": 435
    },
    {
      "epoch": 25.88,
      "grad_norm": 0.9366277456283569,
      "learning_rate": 0.00022288235294117643,
      "loss": 0.3002,
      "step": 440
    },
    {
      "epoch": 25.88,
      "eval_loss": 0.27707499265670776,
      "eval_runtime": 1.9949,
      "eval_samples_per_second": 67.171,
      "eval_steps_per_second": 8.522,
      "step": 440
    },
    {
      "epoch": 26.18,
      "grad_norm": 0.7006655931472778,
      "learning_rate": 0.00022199999999999998,
      "loss": 0.3185,
      "step": 445
    },
    {
      "epoch": 26.47,
      "grad_norm": 0.6538874506950378,
      "learning_rate": 0.00022111764705882352,
      "loss": 0.2813,
      "step": 450
    },
    {
      "epoch": 26.76,
      "grad_norm": 0.8294863104820251,
      "learning_rate": 0.00022023529411764703,
      "loss": 0.2989,
      "step": 455
    },
    {
      "epoch": 27.06,
      "grad_norm": 0.6035735011100769,
      "learning_rate": 0.00021935294117647058,
      "loss": 0.2996,
      "step": 460
    },
    {
      "epoch": 27.06,
      "eval_loss": 0.27492547035217285,
      "eval_runtime": 1.9946,
      "eval_samples_per_second": 67.183,
      "eval_steps_per_second": 8.523,
      "step": 460
    },
    {
      "epoch": 27.35,
      "grad_norm": 0.44762280583381653,
      "learning_rate": 0.00021847058823529412,
      "loss": 0.2766,
      "step": 465
    },
    {
      "epoch": 27.65,
      "grad_norm": 0.6129966974258423,
      "learning_rate": 0.00021758823529411764,
      "loss": 0.2849,
      "step": 470
    },
    {
      "epoch": 27.94,
      "grad_norm": 1.0713130235671997,
      "learning_rate": 0.00021670588235294118,
      "loss": 0.3428,
      "step": 475
    },
    {
      "epoch": 28.24,
      "grad_norm": 0.5866985321044922,
      "learning_rate": 0.00021582352941176467,
      "loss": 0.2759,
      "step": 480
    },
    {
      "epoch": 28.24,
      "eval_loss": 0.27579402923583984,
      "eval_runtime": 1.9947,
      "eval_samples_per_second": 67.176,
      "eval_steps_per_second": 8.522,
      "step": 480
    },
    {
      "epoch": 28.53,
      "grad_norm": 0.5810605883598328,
      "learning_rate": 0.0002149411764705882,
      "loss": 0.2954,
      "step": 485
    },
    {
      "epoch": 28.82,
      "grad_norm": 0.6972721815109253,
      "learning_rate": 0.00021405882352941173,
      "loss": 0.3118,
      "step": 490
    },
    {
      "epoch": 29.12,
      "grad_norm": 0.5709840059280396,
      "learning_rate": 0.00021317647058823527,
      "loss": 0.2901,
      "step": 495
    },
    {
      "epoch": 29.41,
      "grad_norm": 0.5941194295883179,
      "learning_rate": 0.0002122941176470588,
      "loss": 0.3009,
      "step": 500
    },
    {
      "epoch": 29.41,
      "eval_loss": 0.27494505047798157,
      "eval_runtime": 1.9954,
      "eval_samples_per_second": 67.154,
      "eval_steps_per_second": 8.52,
      "step": 500
    },
    {
      "epoch": 29.71,
      "grad_norm": 0.5286316275596619,
      "learning_rate": 0.00021141176470588233,
      "loss": 0.2753,
      "step": 505
    },
    {
      "epoch": 30.0,
      "grad_norm": 0.6919939517974854,
      "learning_rate": 0.00021052941176470587,
      "loss": 0.3041,
      "step": 510
    },
    {
      "epoch": 30.29,
      "grad_norm": 0.7751806974411011,
      "learning_rate": 0.0002096470588235294,
      "loss": 0.2994,
      "step": 515
    },
    {
      "epoch": 30.59,
      "grad_norm": 0.6728924512863159,
      "learning_rate": 0.00020876470588235293,
      "loss": 0.2758,
      "step": 520
    },
    {
      "epoch": 30.59,
      "eval_loss": 0.2755880057811737,
      "eval_runtime": 1.9941,
      "eval_samples_per_second": 67.197,
      "eval_steps_per_second": 8.525,
      "step": 520
    },
    {
      "epoch": 30.88,
      "grad_norm": 0.6528185606002808,
      "learning_rate": 0.00020788235294117647,
      "loss": 0.3003,
      "step": 525
    },
    {
      "epoch": 31.18,
      "grad_norm": 0.6421308517456055,
      "learning_rate": 0.00020699999999999996,
      "loss": 0.2887,
      "step": 530
    },
    {
      "epoch": 31.47,
      "grad_norm": 0.8102923631668091,
      "learning_rate": 0.0002061176470588235,
      "loss": 0.2871,
      "step": 535
    },
    {
      "epoch": 31.76,
      "grad_norm": 0.6277897357940674,
      "learning_rate": 0.00020523529411764702,
      "loss": 0.2909,
      "step": 540
    },
    {
      "epoch": 31.76,
      "eval_loss": 0.27382102608680725,
      "eval_runtime": 1.9939,
      "eval_samples_per_second": 67.207,
      "eval_steps_per_second": 8.526,
      "step": 540
    },
    {
      "epoch": 32.06,
      "grad_norm": 0.6897189021110535,
      "learning_rate": 0.00020435294117647056,
      "loss": 0.3024,
      "step": 545
    },
    {
      "epoch": 32.35,
      "grad_norm": 0.4933074712753296,
      "learning_rate": 0.0002034705882352941,
      "loss": 0.2617,
      "step": 550
    },
    {
      "epoch": 32.65,
      "grad_norm": 0.6854521632194519,
      "learning_rate": 0.00020258823529411762,
      "loss": 0.2882,
      "step": 555
    },
    {
      "epoch": 32.94,
      "grad_norm": 0.6881256103515625,
      "learning_rate": 0.00020170588235294117,
      "loss": 0.3115,
      "step": 560
    },
    {
      "epoch": 32.94,
      "eval_loss": 0.274933785200119,
      "eval_runtime": 1.9951,
      "eval_samples_per_second": 67.166,
      "eval_steps_per_second": 8.521,
      "step": 560
    },
    {
      "epoch": 33.24,
      "grad_norm": 0.6826930642127991,
      "learning_rate": 0.00020082352941176468,
      "loss": 0.3037,
      "step": 565
    },
    {
      "epoch": 33.53,
      "grad_norm": 0.624535322189331,
      "learning_rate": 0.00019994117647058823,
      "loss": 0.2821,
      "step": 570
    },
    {
      "epoch": 33.82,
      "grad_norm": 0.6433701515197754,
      "learning_rate": 0.00019905882352941177,
      "loss": 0.2946,
      "step": 575
    },
    {
      "epoch": 34.12,
      "grad_norm": 0.5257152318954468,
      "learning_rate": 0.00019817647058823528,
      "loss": 0.2963,
      "step": 580
    },
    {
      "epoch": 34.12,
      "eval_loss": 0.2734356224536896,
      "eval_runtime": 1.9971,
      "eval_samples_per_second": 67.096,
      "eval_steps_per_second": 8.512,
      "step": 580
    },
    {
      "epoch": 34.41,
      "grad_norm": 0.5631464719772339,
      "learning_rate": 0.0001972941176470588,
      "loss": 0.2992,
      "step": 585
    },
    {
      "epoch": 34.71,
      "grad_norm": 0.6825907230377197,
      "learning_rate": 0.00019641176470588232,
      "loss": 0.2819,
      "step": 590
    },
    {
      "epoch": 35.0,
      "grad_norm": 0.7912545204162598,
      "learning_rate": 0.00019552941176470586,
      "loss": 0.3009,
      "step": 595
    },
    {
      "epoch": 35.29,
      "grad_norm": 0.6068854928016663,
      "learning_rate": 0.0001946470588235294,
      "loss": 0.2938,
      "step": 600
    },
    {
      "epoch": 35.29,
      "eval_loss": 0.2715306282043457,
      "eval_runtime": 1.997,
      "eval_samples_per_second": 67.101,
      "eval_steps_per_second": 8.513,
      "step": 600
    },
    {
      "epoch": 35.59,
      "grad_norm": 0.6175413131713867,
      "learning_rate": 0.00019376470588235292,
      "loss": 0.2853,
      "step": 605
    },
    {
      "epoch": 35.88,
      "grad_norm": 0.6325289011001587,
      "learning_rate": 0.00019288235294117646,
      "loss": 0.2825,
      "step": 610
    },
    {
      "epoch": 36.18,
      "grad_norm": 0.5814202427864075,
      "learning_rate": 0.00019199999999999998,
      "loss": 0.2785,
      "step": 615
    },
    {
      "epoch": 36.47,
      "grad_norm": 0.6043010950088501,
      "learning_rate": 0.00019111764705882352,
      "loss": 0.2707,
      "step": 620
    },
    {
      "epoch": 36.47,
      "eval_loss": 0.27390608191490173,
      "eval_runtime": 1.9949,
      "eval_samples_per_second": 67.173,
      "eval_steps_per_second": 8.522,
      "step": 620
    },
    {
      "epoch": 36.76,
      "grad_norm": 0.7618833780288696,
      "learning_rate": 0.00019023529411764706,
      "loss": 0.3051,
      "step": 625
    },
    {
      "epoch": 37.06,
      "grad_norm": 0.49574750661849976,
      "learning_rate": 0.00018935294117647058,
      "loss": 0.3009,
      "step": 630
    },
    {
      "epoch": 37.35,
      "grad_norm": 0.6526892185211182,
      "learning_rate": 0.00018847058823529412,
      "loss": 0.2764,
      "step": 635
    },
    {
      "epoch": 37.65,
      "grad_norm": 0.674574613571167,
      "learning_rate": 0.0001875882352941176,
      "loss": 0.2903,
      "step": 640
    },
    {
      "epoch": 37.65,
      "eval_loss": 0.27166232466697693,
      "eval_runtime": 1.9963,
      "eval_samples_per_second": 67.125,
      "eval_steps_per_second": 8.516,
      "step": 640
    },
    {
      "epoch": 37.94,
      "grad_norm": 0.7739295363426208,
      "learning_rate": 0.00018670588235294115,
      "loss": 0.295,
      "step": 645
    },
    {
      "epoch": 38.24,
      "grad_norm": 0.5252580642700195,
      "learning_rate": 0.00018582352941176467,
      "loss": 0.2616,
      "step": 650
    },
    {
      "epoch": 38.53,
      "grad_norm": 0.8815963268280029,
      "learning_rate": 0.00018494117647058821,
      "loss": 0.3151,
      "step": 655
    },
    {
      "epoch": 38.82,
      "grad_norm": 0.6178475618362427,
      "learning_rate": 0.00018405882352941176,
      "loss": 0.2902,
      "step": 660
    },
    {
      "epoch": 38.82,
      "eval_loss": 0.27082693576812744,
      "eval_runtime": 1.995,
      "eval_samples_per_second": 67.167,
      "eval_steps_per_second": 8.521,
      "step": 660
    },
    {
      "epoch": 39.12,
      "grad_norm": 0.6863218545913696,
      "learning_rate": 0.00018317647058823527,
      "loss": 0.2846,
      "step": 665
    },
    {
      "epoch": 39.41,
      "grad_norm": 0.5531587600708008,
      "learning_rate": 0.00018229411764705882,
      "loss": 0.2857,
      "step": 670
    },
    {
      "epoch": 39.71,
      "grad_norm": 0.580801248550415,
      "learning_rate": 0.00018141176470588236,
      "loss": 0.2876,
      "step": 675
    },
    {
      "epoch": 40.0,
      "grad_norm": 0.559680163860321,
      "learning_rate": 0.00018052941176470587,
      "loss": 0.2825,
      "step": 680
    },
    {
      "epoch": 40.0,
      "eval_loss": 0.27086344361305237,
      "eval_runtime": 2.0106,
      "eval_samples_per_second": 66.647,
      "eval_steps_per_second": 8.455,
      "step": 680
    },
    {
      "epoch": 40.29,
      "grad_norm": 0.4894808232784271,
      "learning_rate": 0.00017964705882352942,
      "loss": 0.2762,
      "step": 685
    },
    {
      "epoch": 40.59,
      "grad_norm": 0.4080250859260559,
      "learning_rate": 0.0001787647058823529,
      "loss": 0.2552,
      "step": 690
    },
    {
      "epoch": 40.88,
      "grad_norm": 0.7825730443000793,
      "learning_rate": 0.00017788235294117645,
      "loss": 0.3105,
      "step": 695
    },
    {
      "epoch": 41.18,
      "grad_norm": 0.6776964068412781,
      "learning_rate": 0.00017699999999999997,
      "loss": 0.2815,
      "step": 700
    },
    {
      "epoch": 41.18,
      "eval_loss": 0.2690856158733368,
      "eval_runtime": 1.9978,
      "eval_samples_per_second": 67.073,
      "eval_steps_per_second": 8.509,
      "step": 700
    },
    {
      "epoch": 41.47,
      "grad_norm": 0.7837787866592407,
      "learning_rate": 0.0001761176470588235,
      "loss": 0.2957,
      "step": 705
    },
    {
      "epoch": 41.76,
      "grad_norm": 0.63068687915802,
      "learning_rate": 0.00017523529411764705,
      "loss": 0.2904,
      "step": 710
    },
    {
      "epoch": 42.06,
      "grad_norm": 0.6573776602745056,
      "learning_rate": 0.00017435294117647057,
      "loss": 0.2979,
      "step": 715
    },
    {
      "epoch": 42.35,
      "grad_norm": 0.6143296360969543,
      "learning_rate": 0.0001734705882352941,
      "loss": 0.2618,
      "step": 720
    },
    {
      "epoch": 42.35,
      "eval_loss": 0.2710973918437958,
      "eval_runtime": 1.995,
      "eval_samples_per_second": 67.167,
      "eval_steps_per_second": 8.521,
      "step": 720
    },
    {
      "epoch": 42.65,
      "grad_norm": 0.5681264400482178,
      "learning_rate": 0.00017258823529411763,
      "loss": 0.2918,
      "step": 725
    },
    {
      "epoch": 42.94,
      "grad_norm": 0.8199780583381653,
      "learning_rate": 0.00017170588235294117,
      "loss": 0.291,
      "step": 730
    },
    {
      "epoch": 43.24,
      "grad_norm": 0.5809566378593445,
      "learning_rate": 0.0001708235294117647,
      "loss": 0.287,
      "step": 735
    },
    {
      "epoch": 43.53,
      "grad_norm": 0.7246254682540894,
      "learning_rate": 0.00016994117647058823,
      "loss": 0.2792,
      "step": 740
    },
    {
      "epoch": 43.53,
      "eval_loss": 0.2726564109325409,
      "eval_runtime": 1.9952,
      "eval_samples_per_second": 67.162,
      "eval_steps_per_second": 8.521,
      "step": 740
    },
    {
      "epoch": 43.82,
      "grad_norm": 0.6743197441101074,
      "learning_rate": 0.00016905882352941174,
      "loss": 0.2871,
      "step": 745
    },
    {
      "epoch": 44.12,
      "grad_norm": 0.6329538822174072,
      "learning_rate": 0.00016817647058823526,
      "loss": 0.269,
      "step": 750
    },
    {
      "epoch": 44.41,
      "grad_norm": 0.6352816820144653,
      "learning_rate": 0.0001672941176470588,
      "loss": 0.2671,
      "step": 755
    },
    {
      "epoch": 44.71,
      "grad_norm": 0.6283583045005798,
      "learning_rate": 0.00016641176470588235,
      "loss": 0.2907,
      "step": 760
    },
    {
      "epoch": 44.71,
      "eval_loss": 0.27082738280296326,
      "eval_runtime": 1.9938,
      "eval_samples_per_second": 67.208,
      "eval_steps_per_second": 8.526,
      "step": 760
    },
    {
      "epoch": 45.0,
      "grad_norm": 0.658597469329834,
      "learning_rate": 0.00016552941176470586,
      "loss": 0.3013,
      "step": 765
    },
    {
      "epoch": 45.29,
      "grad_norm": 0.4039109945297241,
      "learning_rate": 0.0001646470588235294,
      "loss": 0.2666,
      "step": 770
    },
    {
      "epoch": 45.59,
      "grad_norm": 0.5540700554847717,
      "learning_rate": 0.00016376470588235292,
      "loss": 0.3015,
      "step": 775
    },
    {
      "epoch": 45.88,
      "grad_norm": 0.7095409035682678,
      "learning_rate": 0.00016288235294117646,
      "loss": 0.2822,
      "step": 780
    },
    {
      "epoch": 45.88,
      "eval_loss": 0.26980897784233093,
      "eval_runtime": 1.9956,
      "eval_samples_per_second": 67.148,
      "eval_steps_per_second": 8.519,
      "step": 780
    },
    {
      "epoch": 46.18,
      "grad_norm": 0.5550056099891663,
      "learning_rate": 0.000162,
      "loss": 0.2906,
      "step": 785
    },
    {
      "epoch": 46.47,
      "grad_norm": 0.5746665000915527,
      "learning_rate": 0.00016111764705882352,
      "loss": 0.2841,
      "step": 790
    },
    {
      "epoch": 46.76,
      "grad_norm": 0.6271633505821228,
      "learning_rate": 0.00016023529411764704,
      "loss": 0.2783,
      "step": 795
    },
    {
      "epoch": 47.06,
      "grad_norm": 0.6846187710762024,
      "learning_rate": 0.00015935294117647056,
      "loss": 0.2969,
      "step": 800
    },
    {
      "epoch": 47.06,
      "eval_loss": 0.268819659948349,
      "eval_runtime": 1.9933,
      "eval_samples_per_second": 67.226,
      "eval_steps_per_second": 8.529,
      "step": 800
    },
    {
      "epoch": 47.35,
      "grad_norm": 0.4498227834701538,
      "learning_rate": 0.0001584705882352941,
      "loss": 0.2682,
      "step": 805
    },
    {
      "epoch": 47.65,
      "grad_norm": 0.4453158676624298,
      "learning_rate": 0.00015758823529411761,
      "loss": 0.2835,
      "step": 810
    },
    {
      "epoch": 47.94,
      "grad_norm": 0.5814605355262756,
      "learning_rate": 0.00015670588235294116,
      "loss": 0.2936,
      "step": 815
    },
    {
      "epoch": 48.24,
      "grad_norm": 0.5212646722793579,
      "learning_rate": 0.0001558235294117647,
      "loss": 0.2578,
      "step": 820
    },
    {
      "epoch": 48.24,
      "eval_loss": 0.26919475197792053,
      "eval_runtime": 1.9948,
      "eval_samples_per_second": 67.174,
      "eval_steps_per_second": 8.522,
      "step": 820
    },
    {
      "epoch": 48.53,
      "grad_norm": 0.5765751004219055,
      "learning_rate": 0.00015494117647058822,
      "loss": 0.2831,
      "step": 825
    },
    {
      "epoch": 48.82,
      "grad_norm": 0.719187319278717,
      "learning_rate": 0.00015405882352941176,
      "loss": 0.3037,
      "step": 830
    },
    {
      "epoch": 49.12,
      "grad_norm": 0.5635544061660767,
      "learning_rate": 0.0001531764705882353,
      "loss": 0.2855,
      "step": 835
    },
    {
      "epoch": 49.41,
      "grad_norm": 0.6977357864379883,
      "learning_rate": 0.00015229411764705882,
      "loss": 0.2715,
      "step": 840
    },
    {
      "epoch": 49.41,
      "eval_loss": 0.2697232961654663,
      "eval_runtime": 1.9956,
      "eval_samples_per_second": 67.147,
      "eval_steps_per_second": 8.519,
      "step": 840
    },
    {
      "epoch": 49.71,
      "grad_norm": 0.5384779572486877,
      "learning_rate": 0.00015141176470588236,
      "loss": 0.2872,
      "step": 845
    },
    {
      "epoch": 50.0,
      "grad_norm": 0.820724606513977,
      "learning_rate": 0.00015052941176470585,
      "loss": 0.2855,
      "step": 850
    },
    {
      "epoch": 50.29,
      "grad_norm": 0.6426348686218262,
      "learning_rate": 0.0001496470588235294,
      "loss": 0.2732,
      "step": 855
    },
    {
      "epoch": 50.59,
      "grad_norm": 0.49072563648223877,
      "learning_rate": 0.00014876470588235294,
      "loss": 0.2791,
      "step": 860
    },
    {
      "epoch": 50.59,
      "eval_loss": 0.27140939235687256,
      "eval_runtime": 1.9964,
      "eval_samples_per_second": 67.122,
      "eval_steps_per_second": 8.516,
      "step": 860
    },
    {
      "epoch": 50.88,
      "grad_norm": 0.6572851538658142,
      "learning_rate": 0.00014788235294117645,
      "loss": 0.2788,
      "step": 865
    },
    {
      "epoch": 51.18,
      "grad_norm": 0.5061244368553162,
      "learning_rate": 0.000147,
      "loss": 0.2984,
      "step": 870
    },
    {
      "epoch": 51.47,
      "grad_norm": 0.6135215163230896,
      "learning_rate": 0.0001461176470588235,
      "loss": 0.2914,
      "step": 875
    },
    {
      "epoch": 51.76,
      "grad_norm": 0.556154191493988,
      "learning_rate": 0.00014523529411764705,
      "loss": 0.2746,
      "step": 880
    },
    {
      "epoch": 51.76,
      "eval_loss": 0.2693510055541992,
      "eval_runtime": 1.9947,
      "eval_samples_per_second": 67.177,
      "eval_steps_per_second": 8.522,
      "step": 880
    },
    {
      "epoch": 52.06,
      "grad_norm": 0.4937208592891693,
      "learning_rate": 0.00014435294117647057,
      "loss": 0.2796,
      "step": 885
    },
    {
      "epoch": 52.35,
      "grad_norm": 0.5362540483474731,
      "learning_rate": 0.00014347058823529409,
      "loss": 0.2741,
      "step": 890
    },
    {
      "epoch": 52.65,
      "grad_norm": 0.4936700761318207,
      "learning_rate": 0.00014258823529411763,
      "loss": 0.2876,
      "step": 895
    },
    {
      "epoch": 52.94,
      "grad_norm": 0.6930960416793823,
      "learning_rate": 0.00014170588235294117,
      "loss": 0.2916,
      "step": 900
    },
    {
      "epoch": 52.94,
      "eval_loss": 0.2691669166088104,
      "eval_runtime": 1.9969,
      "eval_samples_per_second": 67.104,
      "eval_steps_per_second": 8.513,
      "step": 900
    },
    {
      "epoch": 53.24,
      "grad_norm": 0.5722822546958923,
      "learning_rate": 0.0001408235294117647,
      "loss": 0.2753,
      "step": 905
    },
    {
      "epoch": 53.53,
      "grad_norm": 0.4354608952999115,
      "learning_rate": 0.00013994117647058823,
      "loss": 0.2508,
      "step": 910
    },
    {
      "epoch": 53.82,
      "grad_norm": 0.7151679396629333,
      "learning_rate": 0.00013905882352941175,
      "loss": 0.3153,
      "step": 915
    },
    {
      "epoch": 54.12,
      "grad_norm": 0.5432928204536438,
      "learning_rate": 0.0001381764705882353,
      "loss": 0.295,
      "step": 920
    },
    {
      "epoch": 54.12,
      "eval_loss": 0.26740530133247375,
      "eval_runtime": 1.9964,
      "eval_samples_per_second": 67.121,
      "eval_steps_per_second": 8.515,
      "step": 920
    },
    {
      "epoch": 54.41,
      "grad_norm": 0.50511234998703,
      "learning_rate": 0.0001372941176470588,
      "loss": 0.2678,
      "step": 925
    },
    {
      "epoch": 54.71,
      "grad_norm": 0.7931966781616211,
      "learning_rate": 0.00013641176470588235,
      "loss": 0.2921,
      "step": 930
    },
    {
      "epoch": 55.0,
      "grad_norm": 1.1295318603515625,
      "learning_rate": 0.00013552941176470587,
      "loss": 0.2874,
      "step": 935
    },
    {
      "epoch": 55.29,
      "grad_norm": 0.5579237341880798,
      "learning_rate": 0.0001346470588235294,
      "loss": 0.2781,
      "step": 940
    },
    {
      "epoch": 55.29,
      "eval_loss": 0.2684433162212372,
      "eval_runtime": 1.9968,
      "eval_samples_per_second": 67.108,
      "eval_steps_per_second": 8.514,
      "step": 940
    },
    {
      "epoch": 55.59,
      "grad_norm": 0.516738772392273,
      "learning_rate": 0.00013376470588235292,
      "loss": 0.2622,
      "step": 945
    },
    {
      "epoch": 55.88,
      "grad_norm": 0.6151651740074158,
      "learning_rate": 0.00013288235294117647,
      "loss": 0.2957,
      "step": 950
    },
    {
      "epoch": 56.18,
      "grad_norm": 0.7047178149223328,
      "learning_rate": 0.00013199999999999998,
      "loss": 0.2893,
      "step": 955
    },
    {
      "epoch": 56.47,
      "grad_norm": 0.5808413624763489,
      "learning_rate": 0.00013111764705882353,
      "loss": 0.2727,
      "step": 960
    },
    {
      "epoch": 56.47,
      "eval_loss": 0.2685576379299164,
      "eval_runtime": 1.994,
      "eval_samples_per_second": 67.203,
      "eval_steps_per_second": 8.526,
      "step": 960
    },
    {
      "epoch": 56.76,
      "grad_norm": 0.6171568632125854,
      "learning_rate": 0.00013023529411764704,
      "loss": 0.2781,
      "step": 965
    },
    {
      "epoch": 57.06,
      "grad_norm": 0.5795710682868958,
      "learning_rate": 0.00012935294117647056,
      "loss": 0.2989,
      "step": 970
    },
    {
      "epoch": 57.35,
      "grad_norm": 0.4831099510192871,
      "learning_rate": 0.0001284705882352941,
      "loss": 0.2719,
      "step": 975
    },
    {
      "epoch": 57.65,
      "grad_norm": 0.6539974808692932,
      "learning_rate": 0.00012758823529411764,
      "loss": 0.2783,
      "step": 980
    },
    {
      "epoch": 57.65,
      "eval_loss": 0.2687915861606598,
      "eval_runtime": 1.9961,
      "eval_samples_per_second": 67.13,
      "eval_steps_per_second": 8.516,
      "step": 980
    },
    {
      "epoch": 57.94,
      "grad_norm": 0.6757897734642029,
      "learning_rate": 0.00012670588235294116,
      "loss": 0.2885,
      "step": 985
    },
    {
      "epoch": 58.24,
      "grad_norm": 0.5477202534675598,
      "learning_rate": 0.0001258235294117647,
      "loss": 0.2931,
      "step": 990
    },
    {
      "epoch": 58.53,
      "grad_norm": 0.6801700592041016,
      "learning_rate": 0.00012494117647058822,
      "loss": 0.2692,
      "step": 995
    },
    {
      "epoch": 58.82,
      "grad_norm": 0.5633025765419006,
      "learning_rate": 0.00012405882352941176,
      "loss": 0.2747,
      "step": 1000
    },
    {
      "epoch": 58.82,
      "eval_loss": 0.269478976726532,
      "eval_runtime": 1.9946,
      "eval_samples_per_second": 67.181,
      "eval_steps_per_second": 8.523,
      "step": 1000
    },
    {
      "epoch": 59.12,
      "grad_norm": 0.6582775115966797,
      "learning_rate": 0.00012317647058823528,
      "loss": 0.2863,
      "step": 1005
    },
    {
      "epoch": 59.41,
      "grad_norm": 0.6495881676673889,
      "learning_rate": 0.00012229411764705882,
      "loss": 0.2797,
      "step": 1010
    },
    {
      "epoch": 59.71,
      "grad_norm": 0.5010665655136108,
      "learning_rate": 0.00012141176470588235,
      "loss": 0.2832,
      "step": 1015
    },
    {
      "epoch": 60.0,
      "grad_norm": 0.5570430755615234,
      "learning_rate": 0.00012052941176470588,
      "loss": 0.2755,
      "step": 1020
    },
    {
      "epoch": 60.0,
      "eval_loss": 0.26759082078933716,
      "eval_runtime": 1.9963,
      "eval_samples_per_second": 67.125,
      "eval_steps_per_second": 8.516,
      "step": 1020
    },
    {
      "epoch": 60.29,
      "grad_norm": 0.5243686437606812,
      "learning_rate": 0.0001196470588235294,
      "loss": 0.2639,
      "step": 1025
    },
    {
      "epoch": 60.59,
      "grad_norm": 0.598508358001709,
      "learning_rate": 0.00011876470588235293,
      "loss": 0.2876,
      "step": 1030
    },
    {
      "epoch": 60.88,
      "grad_norm": 0.595809280872345,
      "learning_rate": 0.00011788235294117645,
      "loss": 0.2703,
      "step": 1035
    },
    {
      "epoch": 61.18,
      "grad_norm": 0.6807280778884888,
      "learning_rate": 0.000117,
      "loss": 0.2889,
      "step": 1040
    },
    {
      "epoch": 61.18,
      "eval_loss": 0.267448753118515,
      "eval_runtime": 1.997,
      "eval_samples_per_second": 67.102,
      "eval_steps_per_second": 8.513,
      "step": 1040
    },
    {
      "epoch": 61.47,
      "grad_norm": 0.4826294183731079,
      "learning_rate": 0.00011611764705882353,
      "loss": 0.2689,
      "step": 1045
    },
    {
      "epoch": 61.76,
      "grad_norm": 0.5328644514083862,
      "learning_rate": 0.00011523529411764704,
      "loss": 0.2928,
      "step": 1050
    },
    {
      "epoch": 62.06,
      "grad_norm": 0.6303868889808655,
      "learning_rate": 0.00011435294117647057,
      "loss": 0.2778,
      "step": 1055
    },
    {
      "epoch": 62.35,
      "grad_norm": 0.6396406292915344,
      "learning_rate": 0.0001134705882352941,
      "loss": 0.2898,
      "step": 1060
    },
    {
      "epoch": 62.35,
      "eval_loss": 0.26878321170806885,
      "eval_runtime": 2.0027,
      "eval_samples_per_second": 66.908,
      "eval_steps_per_second": 8.488,
      "step": 1060
    },
    {
      "epoch": 62.65,
      "grad_norm": 0.6628149151802063,
      "learning_rate": 0.00011258823529411765,
      "loss": 0.2777,
      "step": 1065
    },
    {
      "epoch": 62.94,
      "grad_norm": 0.5486224293708801,
      "learning_rate": 0.00011170588235294117,
      "loss": 0.2738,
      "step": 1070
    },
    {
      "epoch": 63.24,
      "grad_norm": 0.5262264609336853,
      "learning_rate": 0.00011082352941176469,
      "loss": 0.2782,
      "step": 1075
    },
    {
      "epoch": 63.53,
      "grad_norm": 0.5544824600219727,
      "learning_rate": 0.00010994117647058822,
      "loss": 0.2603,
      "step": 1080
    },
    {
      "epoch": 63.53,
      "eval_loss": 0.268815815448761,
      "eval_runtime": 1.9965,
      "eval_samples_per_second": 67.116,
      "eval_steps_per_second": 8.515,
      "step": 1080
    },
    {
      "epoch": 63.82,
      "grad_norm": 0.5544533729553223,
      "learning_rate": 0.00010905882352941175,
      "loss": 0.2955,
      "step": 1085
    },
    {
      "epoch": 64.12,
      "grad_norm": 0.5809686183929443,
      "learning_rate": 0.00010817647058823529,
      "loss": 0.2811,
      "step": 1090
    },
    {
      "epoch": 64.41,
      "grad_norm": 0.4934030771255493,
      "learning_rate": 0.00010729411764705882,
      "loss": 0.2832,
      "step": 1095
    },
    {
      "epoch": 64.71,
      "grad_norm": 0.6920121312141418,
      "learning_rate": 0.00010641176470588235,
      "loss": 0.2655,
      "step": 1100
    },
    {
      "epoch": 64.71,
      "eval_loss": 0.2684324383735657,
      "eval_runtime": 1.996,
      "eval_samples_per_second": 67.134,
      "eval_steps_per_second": 8.517,
      "step": 1100
    },
    {
      "epoch": 65.0,
      "grad_norm": 0.8359130620956421,
      "learning_rate": 0.00010552941176470587,
      "loss": 0.2858,
      "step": 1105
    },
    {
      "epoch": 65.29,
      "grad_norm": 0.47680842876434326,
      "learning_rate": 0.0001046470588235294,
      "loss": 0.256,
      "step": 1110
    },
    {
      "epoch": 65.59,
      "grad_norm": 0.5028588771820068,
      "learning_rate": 0.00010376470588235293,
      "loss": 0.2834,
      "step": 1115
    },
    {
      "epoch": 65.88,
      "grad_norm": 0.7377525568008423,
      "learning_rate": 0.00010288235294117647,
      "loss": 0.3062,
      "step": 1120
    },
    {
      "epoch": 65.88,
      "eval_loss": 0.2679174244403839,
      "eval_runtime": 1.996,
      "eval_samples_per_second": 67.134,
      "eval_steps_per_second": 8.517,
      "step": 1120
    },
    {
      "epoch": 66.18,
      "grad_norm": 0.5834652185440063,
      "learning_rate": 0.000102,
      "loss": 0.2754,
      "step": 1125
    },
    {
      "epoch": 66.47,
      "grad_norm": 0.5451943874359131,
      "learning_rate": 0.00010111764705882352,
      "loss": 0.2775,
      "step": 1130
    },
    {
      "epoch": 66.76,
      "grad_norm": 0.7877181768417358,
      "learning_rate": 0.00010023529411764704,
      "loss": 0.2968,
      "step": 1135
    },
    {
      "epoch": 67.06,
      "grad_norm": 0.706478476524353,
      "learning_rate": 9.935294117647057e-05,
      "loss": 0.2679,
      "step": 1140
    },
    {
      "epoch": 67.06,
      "eval_loss": 0.26736557483673096,
      "eval_runtime": 1.9963,
      "eval_samples_per_second": 67.124,
      "eval_steps_per_second": 8.516,
      "step": 1140
    },
    {
      "epoch": 67.35,
      "grad_norm": 0.5112646818161011,
      "learning_rate": 9.847058823529412e-05,
      "loss": 0.2893,
      "step": 1145
    },
    {
      "epoch": 67.65,
      "grad_norm": 0.5195066332817078,
      "learning_rate": 9.758823529411765e-05,
      "loss": 0.26,
      "step": 1150
    },
    {
      "epoch": 67.94,
      "grad_norm": 0.7363670468330383,
      "learning_rate": 9.670588235294116e-05,
      "loss": 0.2879,
      "step": 1155
    },
    {
      "epoch": 68.24,
      "grad_norm": 0.5193796753883362,
      "learning_rate": 9.582352941176469e-05,
      "loss": 0.2792,
      "step": 1160
    },
    {
      "epoch": 68.24,
      "eval_loss": 0.26693105697631836,
      "eval_runtime": 1.9965,
      "eval_samples_per_second": 67.117,
      "eval_steps_per_second": 8.515,
      "step": 1160
    },
    {
      "epoch": 68.53,
      "grad_norm": 0.5236190557479858,
      "learning_rate": 9.494117647058822e-05,
      "loss": 0.2693,
      "step": 1165
    },
    {
      "epoch": 68.82,
      "grad_norm": 0.5888519287109375,
      "learning_rate": 9.405882352941176e-05,
      "loss": 0.3004,
      "step": 1170
    },
    {
      "epoch": 69.12,
      "grad_norm": 0.45090892910957336,
      "learning_rate": 9.31764705882353e-05,
      "loss": 0.2667,
      "step": 1175
    },
    {
      "epoch": 69.41,
      "grad_norm": 0.553806483745575,
      "learning_rate": 9.229411764705881e-05,
      "loss": 0.2606,
      "step": 1180
    },
    {
      "epoch": 69.41,
      "eval_loss": 0.26794832944869995,
      "eval_runtime": 1.9956,
      "eval_samples_per_second": 67.149,
      "eval_steps_per_second": 8.519,
      "step": 1180
    },
    {
      "epoch": 69.71,
      "grad_norm": 0.5653892755508423,
      "learning_rate": 9.141176470588234e-05,
      "loss": 0.2884,
      "step": 1185
    },
    {
      "epoch": 70.0,
      "grad_norm": 0.6985815167427063,
      "learning_rate": 9.052941176470587e-05,
      "loss": 0.2787,
      "step": 1190
    },
    {
      "epoch": 70.29,
      "grad_norm": 0.6280454397201538,
      "learning_rate": 8.96470588235294e-05,
      "loss": 0.2744,
      "step": 1195
    },
    {
      "epoch": 70.59,
      "grad_norm": 0.669928252696991,
      "learning_rate": 8.876470588235294e-05,
      "loss": 0.2762,
      "step": 1200
    },
    {
      "epoch": 70.59,
      "eval_loss": 0.26804956793785095,
      "eval_runtime": 1.9958,
      "eval_samples_per_second": 67.141,
      "eval_steps_per_second": 8.518,
      "step": 1200
    },
    {
      "epoch": 70.88,
      "grad_norm": 0.5315139293670654,
      "learning_rate": 8.788235294117647e-05,
      "loss": 0.2856,
      "step": 1205
    },
    {
      "epoch": 71.18,
      "grad_norm": 0.5895591378211975,
      "learning_rate": 8.699999999999999e-05,
      "loss": 0.2916,
      "step": 1210
    },
    {
      "epoch": 71.47,
      "grad_norm": 0.484454870223999,
      "learning_rate": 8.611764705882352e-05,
      "loss": 0.255,
      "step": 1215
    },
    {
      "epoch": 71.76,
      "grad_norm": 0.624661922454834,
      "learning_rate": 8.523529411764705e-05,
      "loss": 0.3034,
      "step": 1220
    },
    {
      "epoch": 71.76,
      "eval_loss": 0.2677695155143738,
      "eval_runtime": 1.9953,
      "eval_samples_per_second": 67.159,
      "eval_steps_per_second": 8.52,
      "step": 1220
    },
    {
      "epoch": 72.06,
      "grad_norm": 0.46470680832862854,
      "learning_rate": 8.435294117647059e-05,
      "loss": 0.2518,
      "step": 1225
    },
    {
      "epoch": 72.35,
      "grad_norm": 0.5925688147544861,
      "learning_rate": 8.347058823529412e-05,
      "loss": 0.2665,
      "step": 1230
    },
    {
      "epoch": 72.65,
      "grad_norm": 0.5845850706100464,
      "learning_rate": 8.258823529411763e-05,
      "loss": 0.2842,
      "step": 1235
    },
    {
      "epoch": 72.94,
      "grad_norm": 0.7143939137458801,
      "learning_rate": 8.170588235294116e-05,
      "loss": 0.2874,
      "step": 1240
    },
    {
      "epoch": 72.94,
      "eval_loss": 0.26679080724716187,
      "eval_runtime": 1.996,
      "eval_samples_per_second": 67.134,
      "eval_steps_per_second": 8.517,
      "step": 1240
    },
    {
      "epoch": 73.24,
      "grad_norm": 0.7805367708206177,
      "learning_rate": 8.08235294117647e-05,
      "loss": 0.2572,
      "step": 1245
    },
    {
      "epoch": 73.53,
      "grad_norm": 0.6749709844589233,
      "learning_rate": 7.994117647058824e-05,
      "loss": 0.2832,
      "step": 1250
    },
    {
      "epoch": 73.82,
      "grad_norm": 0.618565559387207,
      "learning_rate": 7.905882352941177e-05,
      "loss": 0.2604,
      "step": 1255
    },
    {
      "epoch": 74.12,
      "grad_norm": 0.5196065306663513,
      "learning_rate": 7.817647058823528e-05,
      "loss": 0.3091,
      "step": 1260
    },
    {
      "epoch": 74.12,
      "eval_loss": 0.2668515741825104,
      "eval_runtime": 1.996,
      "eval_samples_per_second": 67.133,
      "eval_steps_per_second": 8.517,
      "step": 1260
    },
    {
      "epoch": 74.41,
      "grad_norm": 0.6182024478912354,
      "learning_rate": 7.729411764705881e-05,
      "loss": 0.2786,
      "step": 1265
    },
    {
      "epoch": 74.71,
      "grad_norm": 0.7030547857284546,
      "learning_rate": 7.641176470588234e-05,
      "loss": 0.2864,
      "step": 1270
    },
    {
      "epoch": 75.0,
      "grad_norm": 0.48524388670921326,
      "learning_rate": 7.552941176470588e-05,
      "loss": 0.2613,
      "step": 1275
    },
    {
      "epoch": 75.29,
      "grad_norm": 0.5933300256729126,
      "learning_rate": 7.46470588235294e-05,
      "loss": 0.2954,
      "step": 1280
    },
    {
      "epoch": 75.29,
      "eval_loss": 0.2671739161014557,
      "eval_runtime": 1.9942,
      "eval_samples_per_second": 67.196,
      "eval_steps_per_second": 8.525,
      "step": 1280
    },
    {
      "epoch": 75.59,
      "grad_norm": 0.6050323843955994,
      "learning_rate": 7.376470588235293e-05,
      "loss": 0.2572,
      "step": 1285
    },
    {
      "epoch": 75.88,
      "grad_norm": 0.5846253633499146,
      "learning_rate": 7.288235294117647e-05,
      "loss": 0.268,
      "step": 1290
    },
    {
      "epoch": 76.18,
      "grad_norm": 0.5491711497306824,
      "learning_rate": 7.199999999999999e-05,
      "loss": 0.2706,
      "step": 1295
    },
    {
      "epoch": 76.47,
      "grad_norm": 0.5089874863624573,
      "learning_rate": 7.111764705882352e-05,
      "loss": 0.2591,
      "step": 1300
    },
    {
      "epoch": 76.47,
      "eval_loss": 0.2666688561439514,
      "eval_runtime": 1.9966,
      "eval_samples_per_second": 67.114,
      "eval_steps_per_second": 8.514,
      "step": 1300
    },
    {
      "epoch": 76.76,
      "grad_norm": 0.7518558502197266,
      "learning_rate": 7.023529411764706e-05,
      "loss": 0.282,
      "step": 1305
    },
    {
      "epoch": 77.06,
      "grad_norm": 0.624338686466217,
      "learning_rate": 6.935294117647058e-05,
      "loss": 0.2991,
      "step": 1310
    },
    {
      "epoch": 77.35,
      "grad_norm": 0.5179735422134399,
      "learning_rate": 6.847058823529412e-05,
      "loss": 0.2667,
      "step": 1315
    },
    {
      "epoch": 77.65,
      "grad_norm": 0.6224756240844727,
      "learning_rate": 6.758823529411764e-05,
      "loss": 0.28,
      "step": 1320
    },
    {
      "epoch": 77.65,
      "eval_loss": 0.26699915528297424,
      "eval_runtime": 1.9958,
      "eval_samples_per_second": 67.139,
      "eval_steps_per_second": 8.518,
      "step": 1320
    },
    {
      "epoch": 77.94,
      "grad_norm": 0.6646649837493896,
      "learning_rate": 6.670588235294117e-05,
      "loss": 0.2726,
      "step": 1325
    },
    {
      "epoch": 78.24,
      "grad_norm": 0.4936932325363159,
      "learning_rate": 6.582352941176471e-05,
      "loss": 0.2663,
      "step": 1330
    },
    {
      "epoch": 78.53,
      "grad_norm": 0.5740850567817688,
      "learning_rate": 6.494117647058822e-05,
      "loss": 0.2642,
      "step": 1335
    },
    {
      "epoch": 78.82,
      "grad_norm": 0.5269001722335815,
      "learning_rate": 6.405882352941175e-05,
      "loss": 0.2818,
      "step": 1340
    },
    {
      "epoch": 78.82,
      "eval_loss": 0.2666696310043335,
      "eval_runtime": 1.9974,
      "eval_samples_per_second": 67.088,
      "eval_steps_per_second": 8.511,
      "step": 1340
    },
    {
      "epoch": 79.12,
      "grad_norm": 0.6438319087028503,
      "learning_rate": 6.31764705882353e-05,
      "loss": 0.3015,
      "step": 1345
    },
    {
      "epoch": 79.41,
      "grad_norm": 0.5811217427253723,
      "learning_rate": 6.229411764705881e-05,
      "loss": 0.2438,
      "step": 1350
    },
    {
      "epoch": 79.71,
      "grad_norm": 0.5257366895675659,
      "learning_rate": 6.141176470588236e-05,
      "loss": 0.2775,
      "step": 1355
    },
    {
      "epoch": 80.0,
      "grad_norm": 0.67855304479599,
      "learning_rate": 6.052941176470587e-05,
      "loss": 0.2922,
      "step": 1360
    },
    {
      "epoch": 80.0,
      "eval_loss": 0.2664375901222229,
      "eval_runtime": 1.9959,
      "eval_samples_per_second": 67.136,
      "eval_steps_per_second": 8.517,
      "step": 1360
    },
    {
      "epoch": 80.29,
      "grad_norm": 0.6666358113288879,
      "learning_rate": 5.964705882352941e-05,
      "loss": 0.2827,
      "step": 1365
    },
    {
      "epoch": 80.59,
      "grad_norm": 0.5569036602973938,
      "learning_rate": 5.876470588235294e-05,
      "loss": 0.2741,
      "step": 1370
    },
    {
      "epoch": 80.88,
      "grad_norm": 0.5340925455093384,
      "learning_rate": 5.788235294117646e-05,
      "loss": 0.2587,
      "step": 1375
    },
    {
      "epoch": 81.18,
      "grad_norm": 0.6019856929779053,
      "learning_rate": 5.6999999999999996e-05,
      "loss": 0.2883,
      "step": 1380
    },
    {
      "epoch": 81.18,
      "eval_loss": 0.266279011964798,
      "eval_runtime": 1.9934,
      "eval_samples_per_second": 67.222,
      "eval_steps_per_second": 8.528,
      "step": 1380
    },
    {
      "epoch": 81.47,
      "grad_norm": 0.4570218026638031,
      "learning_rate": 5.6117647058823526e-05,
      "loss": 0.2521,
      "step": 1385
    },
    {
      "epoch": 81.76,
      "grad_norm": 0.6152669191360474,
      "learning_rate": 5.5235294117647055e-05,
      "loss": 0.2826,
      "step": 1390
    },
    {
      "epoch": 82.06,
      "grad_norm": 0.537754237651825,
      "learning_rate": 5.4352941176470585e-05,
      "loss": 0.2782,
      "step": 1395
    },
    {
      "epoch": 82.35,
      "grad_norm": 0.7443511486053467,
      "learning_rate": 5.347058823529411e-05,
      "loss": 0.2717,
      "step": 1400
    },
    {
      "epoch": 82.35,
      "eval_loss": 0.2660791873931885,
      "eval_runtime": 1.9952,
      "eval_samples_per_second": 67.162,
      "eval_steps_per_second": 8.521,
      "step": 1400
    },
    {
      "epoch": 82.65,
      "grad_norm": 0.45767509937286377,
      "learning_rate": 5.2588235294117644e-05,
      "loss": 0.2595,
      "step": 1405
    },
    {
      "epoch": 82.94,
      "grad_norm": 0.7017338871955872,
      "learning_rate": 5.170588235294117e-05,
      "loss": 0.2865,
      "step": 1410
    },
    {
      "epoch": 83.24,
      "grad_norm": 0.4680289030075073,
      "learning_rate": 5.0823529411764696e-05,
      "loss": 0.2761,
      "step": 1415
    },
    {
      "epoch": 83.53,
      "grad_norm": 0.5954050421714783,
      "learning_rate": 4.994117647058823e-05,
      "loss": 0.2874,
      "step": 1420
    },
    {
      "epoch": 83.53,
      "eval_loss": 0.2670120894908905,
      "eval_runtime": 1.9957,
      "eval_samples_per_second": 67.146,
      "eval_steps_per_second": 8.519,
      "step": 1420
    },
    {
      "epoch": 83.82,
      "grad_norm": 0.5585493445396423,
      "learning_rate": 4.905882352941176e-05,
      "loss": 0.274,
      "step": 1425
    },
    {
      "epoch": 84.12,
      "grad_norm": 0.6001076698303223,
      "learning_rate": 4.817647058823529e-05,
      "loss": 0.2579,
      "step": 1430
    },
    {
      "epoch": 84.41,
      "grad_norm": 0.5703680515289307,
      "learning_rate": 4.729411764705882e-05,
      "loss": 0.2602,
      "step": 1435
    },
    {
      "epoch": 84.71,
      "grad_norm": 0.5915176868438721,
      "learning_rate": 4.6411764705882343e-05,
      "loss": 0.2676,
      "step": 1440
    },
    {
      "epoch": 84.71,
      "eval_loss": 0.2662909924983978,
      "eval_runtime": 2.007,
      "eval_samples_per_second": 66.766,
      "eval_steps_per_second": 8.47,
      "step": 1440
    },
    {
      "epoch": 85.0,
      "grad_norm": 0.659074068069458,
      "learning_rate": 4.552941176470588e-05,
      "loss": 0.2864,
      "step": 1445
    },
    {
      "epoch": 85.29,
      "grad_norm": 0.6976066827774048,
      "learning_rate": 4.464705882352941e-05,
      "loss": 0.277,
      "step": 1450
    },
    {
      "epoch": 85.59,
      "grad_norm": 0.6575342416763306,
      "learning_rate": 4.376470588235293e-05,
      "loss": 0.2886,
      "step": 1455
    },
    {
      "epoch": 85.88,
      "grad_norm": 0.4784369170665741,
      "learning_rate": 4.288235294117647e-05,
      "loss": 0.2637,
      "step": 1460
    },
    {
      "epoch": 85.88,
      "eval_loss": 0.2663717269897461,
      "eval_runtime": 1.9942,
      "eval_samples_per_second": 67.195,
      "eval_steps_per_second": 8.525,
      "step": 1460
    },
    {
      "epoch": 86.18,
      "grad_norm": 0.5686805844306946,
      "learning_rate": 4.2e-05,
      "loss": 0.2677,
      "step": 1465
    },
    {
      "epoch": 86.47,
      "grad_norm": 0.6093593239784241,
      "learning_rate": 4.111764705882353e-05,
      "loss": 0.2594,
      "step": 1470
    },
    {
      "epoch": 86.76,
      "grad_norm": 0.7324638366699219,
      "learning_rate": 4.023529411764706e-05,
      "loss": 0.2916,
      "step": 1475
    },
    {
      "epoch": 87.06,
      "grad_norm": 0.6090161204338074,
      "learning_rate": 3.935294117647058e-05,
      "loss": 0.2539,
      "step": 1480
    },
    {
      "epoch": 87.06,
      "eval_loss": 0.26603055000305176,
      "eval_runtime": 1.9935,
      "eval_samples_per_second": 67.218,
      "eval_steps_per_second": 8.528,
      "step": 1480
    },
    {
      "epoch": 87.35,
      "grad_norm": 0.7536705136299133,
      "learning_rate": 3.8470588235294116e-05,
      "loss": 0.2886,
      "step": 1485
    },
    {
      "epoch": 87.65,
      "grad_norm": 0.5662975311279297,
      "learning_rate": 3.7588235294117645e-05,
      "loss": 0.2741,
      "step": 1490
    },
    {
      "epoch": 87.94,
      "grad_norm": 0.5700023174285889,
      "learning_rate": 3.6705882352941175e-05,
      "loss": 0.2767,
      "step": 1495
    },
    {
      "epoch": 88.24,
      "grad_norm": 0.5497795343399048,
      "learning_rate": 3.5823529411764704e-05,
      "loss": 0.2648,
      "step": 1500
    },
    {
      "epoch": 88.24,
      "eval_loss": 0.2659338712692261,
      "eval_runtime": 1.9949,
      "eval_samples_per_second": 67.171,
      "eval_steps_per_second": 8.522,
      "step": 1500
    },
    {
      "epoch": 88.53,
      "grad_norm": 0.5747610926628113,
      "learning_rate": 3.4941176470588234e-05,
      "loss": 0.2627,
      "step": 1505
    },
    {
      "epoch": 88.82,
      "grad_norm": 0.4665067195892334,
      "learning_rate": 3.405882352941176e-05,
      "loss": 0.2847,
      "step": 1510
    },
    {
      "epoch": 89.12,
      "grad_norm": 0.7121643424034119,
      "learning_rate": 3.317647058823529e-05,
      "loss": 0.2676,
      "step": 1515
    },
    {
      "epoch": 89.41,
      "grad_norm": 0.5986043214797974,
      "learning_rate": 3.229411764705882e-05,
      "loss": 0.2782,
      "step": 1520
    },
    {
      "epoch": 89.41,
      "eval_loss": 0.2659454941749573,
      "eval_runtime": 1.9957,
      "eval_samples_per_second": 67.143,
      "eval_steps_per_second": 8.518,
      "step": 1520
    },
    {
      "epoch": 89.71,
      "grad_norm": 0.5361090898513794,
      "learning_rate": 3.141176470588235e-05,
      "loss": 0.2644,
      "step": 1525
    },
    {
      "epoch": 90.0,
      "grad_norm": 0.7882270812988281,
      "learning_rate": 3.052941176470588e-05,
      "loss": 0.2778,
      "step": 1530
    },
    {
      "epoch": 90.29,
      "grad_norm": 0.6407003402709961,
      "learning_rate": 2.9647058823529407e-05,
      "loss": 0.2635,
      "step": 1535
    },
    {
      "epoch": 90.59,
      "grad_norm": 0.6531354188919067,
      "learning_rate": 2.876470588235294e-05,
      "loss": 0.275,
      "step": 1540
    },
    {
      "epoch": 90.59,
      "eval_loss": 0.2659270465373993,
      "eval_runtime": 1.9941,
      "eval_samples_per_second": 67.198,
      "eval_steps_per_second": 8.525,
      "step": 1540
    },
    {
      "epoch": 90.88,
      "grad_norm": 0.6199626326560974,
      "learning_rate": 2.788235294117647e-05,
      "loss": 0.2681,
      "step": 1545
    },
    {
      "epoch": 91.18,
      "grad_norm": 0.6273800134658813,
      "learning_rate": 2.6999999999999996e-05,
      "loss": 0.287,
      "step": 1550
    },
    {
      "epoch": 91.47,
      "grad_norm": 0.6072767972946167,
      "learning_rate": 2.6117647058823525e-05,
      "loss": 0.2583,
      "step": 1555
    },
    {
      "epoch": 91.76,
      "grad_norm": 0.5658175945281982,
      "learning_rate": 2.5235294117647058e-05,
      "loss": 0.2717,
      "step": 1560
    },
    {
      "epoch": 91.76,
      "eval_loss": 0.26581788063049316,
      "eval_runtime": 1.994,
      "eval_samples_per_second": 67.201,
      "eval_steps_per_second": 8.525,
      "step": 1560
    },
    {
      "epoch": 92.06,
      "grad_norm": 0.6367130279541016,
      "learning_rate": 2.4352941176470587e-05,
      "loss": 0.2819,
      "step": 1565
    },
    {
      "epoch": 92.35,
      "grad_norm": 0.6430366039276123,
      "learning_rate": 2.3470588235294114e-05,
      "loss": 0.2759,
      "step": 1570
    },
    {
      "epoch": 92.65,
      "grad_norm": 0.6165040731430054,
      "learning_rate": 2.2588235294117643e-05,
      "loss": 0.2715,
      "step": 1575
    },
    {
      "epoch": 92.94,
      "grad_norm": 0.5176894664764404,
      "learning_rate": 2.1705882352941176e-05,
      "loss": 0.2646,
      "step": 1580
    },
    {
      "epoch": 92.94,
      "eval_loss": 0.2657198905944824,
      "eval_runtime": 1.9962,
      "eval_samples_per_second": 67.126,
      "eval_steps_per_second": 8.516,
      "step": 1580
    },
    {
      "epoch": 93.24,
      "grad_norm": 0.7515372037887573,
      "learning_rate": 2.0823529411764705e-05,
      "loss": 0.2631,
      "step": 1585
    },
    {
      "epoch": 93.53,
      "grad_norm": 0.4853091835975647,
      "learning_rate": 1.994117647058823e-05,
      "loss": 0.2573,
      "step": 1590
    },
    {
      "epoch": 93.82,
      "grad_norm": 0.4851018190383911,
      "learning_rate": 1.905882352941176e-05,
      "loss": 0.271,
      "step": 1595
    },
    {
      "epoch": 94.12,
      "grad_norm": 0.503040075302124,
      "learning_rate": 1.817647058823529e-05,
      "loss": 0.2863,
      "step": 1600
    },
    {
      "epoch": 94.12,
      "eval_loss": 0.26566752791404724,
      "eval_runtime": 1.9956,
      "eval_samples_per_second": 67.147,
      "eval_steps_per_second": 8.519,
      "step": 1600
    },
    {
      "epoch": 94.41,
      "grad_norm": 0.560333788394928,
      "learning_rate": 1.7294117647058823e-05,
      "loss": 0.2546,
      "step": 1605
    },
    {
      "epoch": 94.71,
      "grad_norm": 0.5879210233688354,
      "learning_rate": 1.641176470588235e-05,
      "loss": 0.2733,
      "step": 1610
    },
    {
      "epoch": 95.0,
      "grad_norm": 0.6487263441085815,
      "learning_rate": 1.5529411764705882e-05,
      "loss": 0.2894,
      "step": 1615
    },
    {
      "epoch": 95.29,
      "grad_norm": 0.6337783336639404,
      "learning_rate": 1.464705882352941e-05,
      "loss": 0.2731,
      "step": 1620
    },
    {
      "epoch": 95.29,
      "eval_loss": 0.2656712234020233,
      "eval_runtime": 1.9955,
      "eval_samples_per_second": 67.153,
      "eval_steps_per_second": 8.519,
      "step": 1620
    },
    {
      "epoch": 95.59,
      "grad_norm": 0.5507830381393433,
      "learning_rate": 1.376470588235294e-05,
      "loss": 0.2801,
      "step": 1625
    },
    {
      "epoch": 95.88,
      "grad_norm": 0.7097357511520386,
      "learning_rate": 1.2882352941176469e-05,
      "loss": 0.274,
      "step": 1630
    },
    {
      "epoch": 96.18,
      "grad_norm": 0.5612406730651855,
      "learning_rate": 1.1999999999999999e-05,
      "loss": 0.2571,
      "step": 1635
    },
    {
      "epoch": 96.47,
      "grad_norm": 0.7559302449226379,
      "learning_rate": 1.1117647058823528e-05,
      "loss": 0.2795,
      "step": 1640
    },
    {
      "epoch": 96.47,
      "eval_loss": 0.2656753361225128,
      "eval_runtime": 1.9935,
      "eval_samples_per_second": 67.217,
      "eval_steps_per_second": 8.528,
      "step": 1640
    },
    {
      "epoch": 96.76,
      "grad_norm": 0.542195737361908,
      "learning_rate": 1.0235294117647058e-05,
      "loss": 0.2549,
      "step": 1645
    },
    {
      "epoch": 97.06,
      "grad_norm": 0.6926820874214172,
      "learning_rate": 9.352941176470587e-06,
      "loss": 0.2598,
      "step": 1650
    },
    {
      "epoch": 97.35,
      "grad_norm": 0.5095303654670715,
      "learning_rate": 8.470588235294117e-06,
      "loss": 0.2753,
      "step": 1655
    },
    {
      "epoch": 97.65,
      "grad_norm": 0.5196524262428284,
      "learning_rate": 7.588235294117647e-06,
      "loss": 0.2501,
      "step": 1660
    },
    {
      "epoch": 97.65,
      "eval_loss": 0.26558738946914673,
      "eval_runtime": 1.9934,
      "eval_samples_per_second": 67.221,
      "eval_steps_per_second": 8.528,
      "step": 1660
    },
    {
      "epoch": 97.94,
      "grad_norm": 0.6086786985397339,
      "learning_rate": 6.705882352941176e-06,
      "loss": 0.2955,
      "step": 1665
    },
    {
      "epoch": 98.24,
      "grad_norm": 0.41819947957992554,
      "learning_rate": 5.823529411764706e-06,
      "loss": 0.2543,
      "step": 1670
    },
    {
      "epoch": 98.53,
      "grad_norm": 0.6898114681243896,
      "learning_rate": 4.941176470588235e-06,
      "loss": 0.288,
      "step": 1675
    },
    {
      "epoch": 98.82,
      "grad_norm": 0.5956689119338989,
      "learning_rate": 4.058823529411764e-06,
      "loss": 0.2623,
      "step": 1680
    },
    {
      "epoch": 98.82,
      "eval_loss": 0.2655787765979767,
      "eval_runtime": 1.9955,
      "eval_samples_per_second": 67.152,
      "eval_steps_per_second": 8.519,
      "step": 1680
    },
    {
      "epoch": 99.12,
      "grad_norm": 0.5556439757347107,
      "learning_rate": 3.1764705882352935e-06,
      "loss": 0.2614,
      "step": 1685
    },
    {
      "epoch": 99.41,
      "grad_norm": 0.46832630038261414,
      "learning_rate": 2.2941176470588234e-06,
      "loss": 0.2647,
      "step": 1690
    },
    {
      "epoch": 99.71,
      "grad_norm": 0.49263256788253784,
      "learning_rate": 1.4117647058823527e-06,
      "loss": 0.273,
      "step": 1695
    },
    {
      "epoch": 100.0,
      "grad_norm": 0.6977939605712891,
      "learning_rate": 5.294117647058823e-07,
      "loss": 0.2655,
      "step": 1700
    },
    {
      "epoch": 100.0,
      "eval_loss": 0.26560500264167786,
      "eval_runtime": 1.9947,
      "eval_samples_per_second": 67.178,
      "eval_steps_per_second": 8.523,
      "step": 1700
    }
  ],
  "logging_steps": 5,
  "max_steps": 1700,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 100,
  "save_steps": 20,
  "total_flos": 1.523488749060096e+16,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}